Structural Abnormalities of the Cornea and Lid Resulting from Collagen V Mutations

Abstract

purpose. Type V collagen forms heterotypic fibrils with type I collagen and accounts for 10% to 20% of corneal collagen. The purpose of this study was to define the ocular phenotype resulting from mutations in the type V collagen genes COL5A1 and COL5A2 and to study the pathogenesis of anomalies in a Col5a1-deficient mouse.

methods. Seven patients with classic Ehlers-Danlos syndrome (EDS) due to COL5A1 haploinsufficiency and one with an exon-skipping mutation in COL5A2 underwent an ocular examination, corneal topography, pachymetry, and specular microscopy. A Col5a1-haploinsufficient mouse model of classic EDS was used for biochemical and immunochemical analyses of corneas. Light and electron microscopy were used to quantify stromal thickness, fibril density, fibril structure, and diameter.

results. Five males and three females (mean age, 26 ± 13.57 years; range, 11–52) were studied. All patients had “floppy eyelids.” The corneas of all eyes were thinner (mean corneal thickness: 435.75 ± 12.51 μm) when compared with control corneas (568.89 ± 28.46 μm; P < 0.0001). In the Col5a1 ^+/− mouse cornea, type V collagen content was reduced by ∼49%, and stromal thickness was reduced by ∼26%. Total collagen deposition in Col5a1 ^+/− corneas also was reduced. Collagen fibril diameters were increased, but fibril density was decreased throughout the stroma at all developmental stages.

conclusions. In the eye, COL5A1 and COL5A2 mutations manifest as abnormally thin and steep corneas with floppy eyelids. Mechanisms involved in producing the latter anomalies probably involve altered regulation of collagen fibrillogenesis due to abnormalities in heterotypic type I/V collagen interactions similar to those observed in the Col5a1 ^+/− mouse cornea.

Ehlers-Danlos syndrome (EDS) is a heritable connective tissue disorder characterized by laxity and fragility of soft connective tissues such as dermis, sclera, cornea, ligaments, tendons, and blood vessels. ¹ EDS is estimated to affect 1 in 5000 births. ^{1 2 3 4} In 1997, the nosology of EDS was revised to include six distinct types, which are based on clinical, genetic, and biochemical abnormalities. ^{1 5} The present paper focuses on the ocular phenotype of classic EDS (previously, EDS types I and II). This autosomal dominant disorder is characterized by generalized joint hypermobility and soft, velvety, fragile, and hyperextensible skin that heals poorly with broad atrophic (cigarette paper-like) scars. ^{1 4 5 6} Approximately, 50% of families with classic EDS have mutations in the COL5A1 and COL5A2 genes that encode the α1(V) and α2(V) chains of type V collagen, respectively. ^{6 7 8 9 10 11 12}  

Type V collagen is a quantitatively minor fibril-forming collagen that is present in type I collagen-rich connective tissues such as cornea, sclera, dermis, tendon, ligament, and bone. ^{13 14} Type V collagen molecules may contain α1(V), α2(V), α3(V), and α4(V) chains. ^{13 15 16} The most common isoform with the broadest distribution is the α1(V)₂ α2(V) isoform in the corneal stroma. ^{14 17} In the cornea, type V collagen assembles with type I collagen as heterotypic type I/V fibrils. ^{13 17} The α1(V) and α2(V) chains of type V collagen retain noncollagenous, aminopropeptide domains, which regulate collagen fibrillogenesis and are present on the fibril surface. ^{18 19} Genetic disruption of type V collagen synthesis by fibroblasts demonstrated that heterotypic type I/V collagen interactions are involved normally in the regulation of collagen fibril diameter and fibril number in vitro. ^{20 21} The latter observations were confirmed by in vivo studies of mice with targeted mutations in Col5a1 that prevented the synthesis of α1(V) chains and the assembly of type V collagen molecules. Morphologic analysis of collagen fibrils in the mutant mouse dermis showed that the collagen fibril numbers varied with Col5a1 dosage. Homozygous mouse embryos (Col5a1 ^{− / −}) had virtually no collagen fibrils and did not survive past the early stages of organogenesis. ²² Heterozygous mice (Col5a1 ^{+/ −}) had approximately one half the number of collagen fibrils as did wild-type littermates. The heterozygous mice survived normally, but with a genotype and phenotype that make this mouse a definitive model of classic EDS in humans. ^{6 22}  

Classic EDS can also result from COL5A2 mutations in humans and col5a2 in mice. ^{11 12 23} Mutations resulting in the lack of expression of one or both COL5A2 alleles have not been described in humans. Similarly, there are no reported naturally occurring or engineered Col5a2 mutations in mice that block the synthesis of α2(V) protein chains. To date, only three human COL5A2 mutations have been described, and each one produces a classic EDS phenotype. ^{11 12} The patients were heterozygous for different exon skipping mutations in COL5A2. It is likely that these mutations that produced in-frame deletions within the main triple helical domain of α2(V) chains exerted a dominant negative effect on type V collagen metabolism. ¹¹ However, ultrastructural studies of dermis showed abnormalities of heterotypic type I collagen fibrillogenesis similar to those observed in patients who had COL5A1 haploinsufficiency. ^{11 12} A Col5a2 mouse model was engineered to express a dominant negative mutation that resulted from a loss of exon 6. This exon encodes the aminotelopeptide and the pro-α2(V) chain hinge region. ²³ A change in the conformation of the aminoterminal region of the chain, that normally protrudes from the surface of the collagen fibril, was expected to impair collagen fibrillogenesis and produce an EDS phenotype. In homozygous mice, the observed phenotype resembled classic EDS in humans. Disorganized type I collagen fibrils were observed in the dermis, and the corneal stroma was thinner than in control subjects. There was evidence of a gene dosage effect on collagen fibrillogenesis of the corneal stroma. The heterozygotes and homozygotes had abnormally thick collagen fibrils, with the thickest fibrils being in the homozygotes. More recent work with this model indicates that this is a complex mutation with dominant negative effects reducing secreted heterotrimeric type V collagen and an associated abnormal expression of the alpha 1(V) homotrimer. ²⁴  

The normal corneal stroma is composed of heterotypic type I/V collagen fibrils organized as orthogonal lamellae. ²⁰ The fibrils have small, homogeneous diameters and regular packing. The transparency and strength of the cornea requires the maintenance of this structural organization as well as, precise regulation of fibril and matrix assembly. Type V collagen is present in small amounts (2%–5%) in most type I collagen-containing connective tissues, but it accounts for 10% to 20% of the total corneal collagen. ²⁰ Type V collagen is also present in small amounts in the sclera and other dense connective tissue components of the eye. The α2(V) chain is also present with α1(XI) chains in hybrid type V/XI collagen of the vitreous humor. To date, descriptions of ocular manifestations in type V collagen mutations are limited to the corneal fibril abnormalities reported in the Col5a2 mouse model of classic EDS. ²³ The present study describes the clinical ocular manifestations in patients with classic EDS who were heterozygous for mutations in COL5A1 or COL5A2. Their clinical corneal abnormalities were compared with the abnormalities determined by direct analysis of corneas from the Col5a1 haploinsufficient mouse model of classic EDS. 

Methods

Patients

This project was approved by the Research Ethics Board of the Hospital for Sick Children and adhered to the tenets of the Declaration of Helsinki. The study group consisted of eight patients (16 eyes) from three unrelated families with classic EDS who were heterozygous for mutations in either COL5A1 or COL5A2 (Table 1 , Fig. 1 ). The mutations of two probands (A-II-1 and C-II-1) have been published. ^{6 11} Mutational analysis of proband B-I-1 and the family was undertaken by using previously published methods. ⁶ Eight age-matched healthy control subjects, five females and three males, were also recruited. Informed consent was obtained from each subject after the nature and consequences of the study were explained. 

All patients fulfilled the clinical criteria for classic EDS according to the 1997 Villefranche revised EDS classification. ¹ The clinical features included hyperextensible skin with a smooth, velvety texture and numerous wide, atrophic scars, easy bruising, and generalized joint hypermobility. Some patients also had visceral manifestations such as pneumothorax, mitral valve prolapse, and inguinal hernias. A comprehensive ocular examination included best corrected visual acuity (BCVA), slit lamp examination, corneal diameter measurement, funduscopy, retinoscopy, refraction, corneal topography using the topographic modeling system, TMS-2 (Tomey Technology, Cambridge, MA), central pachymetry, and specular microscopy. The corneal topographic indices included central corneal power (keratometric diopters), keratometric astigmatism (KA), surface regularity index (SRI), which evaluates surface regularity within the central area of the cornea, and surface asymmetry index (SAI), which is a measure of central corneal asymmetry. ²⁵ Central corneal thickness was measured by taking an average of readings by confocal microscopy. Patients and control subjects were tested with the same protocol. 

Col5a1 ^{+/ −} Mice

The Col5a1 ^{− / −} mice were embryonic lethal and the Col5a1 ^{+/ −} mice were haploinsufficient containing half the mRNA and type V collagen as the wild-type littermates. ²² All experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Measurement of Total Collagen Deposited in the Mouse Cornea

Corneas were dissected from 5-month-old animals and washed with PBS. This was followed by hydrolysis in 6 N HCl at 100°C for 18 hours. Colorimetric analysis of the hydroxyproline content of individual corneas was performed after acid hydrolysis. ²⁶ The mean collagen content from each cornea was calculated by using a conversion ratio of 0.12:1.0 to convert micrograms of hydroxyproline to total collagen. 

Immunochemical Analyses of Type V Collagen in the Mouse Cornea

Immunofluorescence microscopy of type V collagen was performed as previously described. ²⁷ Eyes of 6-week-old mice were fixed briefly in 4% paraformaldehyde in PBS (pH 7.2) for 30 minutes on ice, cryoprotected, and sectioned (5 μm). The sections were blocked with 0.05% sodium borohydride followed by 5% BSA in PBS. Incubation with the primary antibody, COL51NC3-1, ²² was at an antibody concentration of 5 μg/mL in blocking buffer. After washing in PBS-Tween-20 (0.05%), the sections were incubated with a secondary antibody, goat anti-rabbit IgG-Alexa Fluor 568 (Molecular Probes, Inc., Eugene, OR), at 1:200 dilutions in blocking buffer. The sections were washed and mounted in Vectashield containing 4′,6′-diamino-2-phenylindole (DAPI) as a nuclear stain (Vector Laboratories, Burlingame, CA). Images were captured using a digital camera (Optronics, Goleta, CA) and an image-analysis system (BioQuant, Nashville, TN), using set integration times and identical conditions to facilitate comparisons between samples. 

Western Blot Analyses of the Type V Collagen Content of Mouse Cornea

Corneas were dissected and extracted in 100 mM Tris-HCl (pH 6.8) containing 2% SDS. The extracts were diluted in a sample buffer, electrophoresed on 5% polyacrylamide gels (Bio-Rad, Hercules, CA) and transferred to nitrocellulose membranes (Hybond ECL; GE Healthcare, Piscataway, NJ). The α1(V) chain was detected using the COL51NC3-1 antibody ²² at a concentration of 5 μg/mL, followed by goat anti-rabbit IgG-HRP (1:5000). Reactivity was detected using a chemoluminescence kit (ECL; GE Healthcare). Two separate extractions of corneas from 5-month-old mice were done. A total of six independent Western blot analyses were run, and density values (intensity/mm²) were measured (Gel Doc 2000 System; Bio-Rad, with Quantity One 4.0.5 software). Three independent measurements were performed and expressed as the mean. The extracts were diluted initially to a constant volume per cornea, and serial dilutions were analyzed. Measurements were taken from the nonsaturated linear region. 

Transmission Electron Microscopy of the Mouse Cornea

Corneas from Col5a1 ^+/+ and Col5a1 ^{+/ −} mice at postnatal day 10 (P10), 6 weeks and 12 weeks were used in these experiments. Tissues were prepared for transmission electron microscopy, as described previously. ^{28 29} Briefly, fixation was in 4% paraformaldehyde, 2.5% glutaraldehyde, 0.1 M sodium cacodylate (pH 7.4) with 8.0 mM CaCl₂ for 2 hours on ice followed by postfixation with 1% osmium tetroxide. After dehydration in a graded ethanol series followed by propylene oxide, the tissues were infiltrated and embedded in a mixture of resin (Embed 812), nadic methyl anhydride, dodecenylsuccinic anhydride, and DMP-30 (EMS Sciences, Fort Washington, PA). Thick sections (1 μm) were cut and stained with methylene blue-azure B for light microscopy. Thin sections were prepared with an ultramicrotome and a diamond knife (UCT; Reichert Jung, Vienna, Austria). Staining was performed with 2% aqueous uranyl acetate followed by 1% phosphotungstic acid (pH 3.2). Sections were examined and photographed at 75 kV, using a transmission electron microscope (model 7000; Hitachi, Tokyo, Japan). The microscope was calibrated with a line grating. 

For measurement of collagen fibril diameter and density, each cornea was analyzed for both the anterior and posterior corneal stroma. The anterior stroma was defined as the anterior third, subjacent to the epithelium and the posterior stroma as the posterior third adjacent to the endothelium. Both regions were photographed in the central portion. Micrographs were taken at 52,524×. Calibrated micrographs from each region were chosen randomly in a masked manner and digitized, and all diameters were measured within a 491 × 467-nm (0.23-μm²) mask. The mask was placed based on fibril orientation (i.e., fields with fibrils in cross-section and an absence of cells). Diameters were measured along the minor axis of the cross-sections (RM Biometrics-Bioquant Image Analysis System; Nashville, TN). Fibril density was determined as the number of fibrils within a 0.23 μm² mask, normalized to 1 μm². The analyses of fibril diameter and density used 13 mutants (4 at 10 days, 4 at 6 weeks, and 5 at 12 weeks) and 11 wild-type animals (2 at 10 days, 4 at 6 weeks, and 5 at 12 weeks). The anterior cornea data were collected from 63 negatives (32 from mutants and 31 from wild-type) with two to five negatives per animal. The posterior cornea data were collected from 64 negatives (32 from both mutants and wild-type) with 2 to 5 negatives per animal. A nearest-neighbor analysis of fibril packing was performed in the mature, compacted (12-week-old) corneas. Data were analyzed from the anterior and posterior stromas of both genotypes (five animals, 12 fields each). 

Statistical Analysis

The EDS patient data are presented as the mean ± SD. Comparisons between EDS patients versus control subjects were performed with an unpaired two-tailed Student’s t-test. 

In the animal studies, the number of fibrils per field was divided by the area of the mask (0.23 μm²), and negative-specific fibril counts per square micrometer (fibril density) were analyzed in a linear mixed-effects model ³⁰ incorporating animal-to-animal and negative-to-negative variability. The counts were log transformed to satisfy the normality assumptions of the model. The thickness of the corneal stroma was analyzed in a linear mixed-effects model, ³⁰ accounting for the correlation between repeated measurements from the same animal. For most of the negative-specific distributions of fibril diameters, a normal distribution assumption was reasonable, although some negatives contained outliers. To identify outliers in each negative, the main resistant outlier detection rule was used. ³¹ This rule labels an outlier as any observation that falls below Q ₁ − 1.5(Q ₃ − Q ₁) or above Q ₃ + 1.5(Q ₃ − Q ₁), where Q ₁ and Q ₃ are the first and third sample quartiles, respectively. This rule is implemented in the usual box-plot data presentation. After removal of the outliers, a linear mixed-effects model ³⁰ was fit to the diameter measurements, incorporating animal-to-animal and negative-to-negative variability. Examination of the residuals from the fitted linear mixed-effects model did not indicate violations of the model assumptions. 

Results

Human Ocular Phenotype

Patients from three unrelated white families with classic EDS were studied (Fig. 1 , Table 1 ). Seven individuals had COL5A1 haploinsufficiency (Cole W, Young F, unpublished data), ⁶ and one was heterozygous for an exon-skipping mutation in COL5A2. ¹¹ The patient group included five males and three females (mean age, 26 ± 13.57 years; range, 11–52 years). The consistent ocular manifestations in the seven patients with COL5A1 haploinsufficiency included thin and steep but transparent corneas and floppy eyelids. The severity of these changes did not vary with age. The lid laxity was asymptomatic and reflected generalized skin laxity with an apparent thinning of the tarsal plate. Similar corneal and eyelid anomalies were observed in the one patient with a dominant negative mutation of COL5A2. There was no clinical evidence of abnormalities in the other type V collagen containing tissues of the eye, such as the sclera and vitreous humor. 

Ocular findings of the affected patients are detailed in Table 2 . Best-corrected visual acuity was 20/25 or better in 93% of eyes. One eye was 20/400 (patient C-II-1) due to deep amblyopia secondary to anisometropic myopia. All patients had floppy eyelids defined by excessive laxity of the upper tarsal plate that everts easily with minimal upward force applied on the upper eyelid (Fig. 2) . However, none of the participants had spontaneous eversion of the upper eyelid. Blue sclera was not observed. 

The affected corneas were clear, but diffusely thinned with a mean corneal thickness of 435.75 ± 12.51 μm (range, 415–448 μm) when compared with 568.89 ± 28.46 μm (range, 521–605 μm) in the control subjects (P < 0.001; Fig. 3 ). There was no apparent change associated with the age range studied. The affected corneas were steeper, with mean keratometric readings of 46.30 ± 1.36 D (range, 43.71–49.39) when compared with 44.11 ± 1.21 D in the control subjects (P < 0.0001). However, there was no suggestion of keratoconus. Two other topographic indices, SRI and SAI, were significantly higher in the EDS patient group than in the control subjects (P < 0.05). Mean SRI in the EDS group was 0.61 ± 0.31 (range, 0.20–1.34 D) and 0.35 ± 0.24 (range, 0.06–0.68 D) in the control subjects. Mean SAI in the EDS group was 1.49 ± 1.35 D (range, 0.37–4.29 D) when compared with 0.49 ± 0.25 D (range, 0.22–1.09 D) in control subjects. Endothelial cell counts were normal for age. Specular microscopy findings were normal, except for one patient (B-I-1; Table 1 ), who showed small multifocal acellular dark areas. The slit lamp examination revealed fine honeycomb-like changes in the central deep stroma and fine cortical opacities of the lenses. Otherwise, the anterior and posterior segments (without angioid streaks) were normal (B-I-1; Table 1 ). 

Mouse Col5a1-Haploinsufficient Model of Classic EDS

Corneal Structure in Col5a1 ^+/− Mice.

The mature corneas from Col5a1-haploinsufficient mice appeared grossly normal with no apparent opacity when compared with wild-type littermates. The eyelids also appeared normal on gross examination. However, corneas from Col5a1 ^+/− mice were thinner than the wild-type corneas. Light microscopic analysis of corneas from 12-week-old mice demonstrated that the corneal stroma of the Col5a1 ^+/− mice were approximately 74% of the thickness in wild-type stroma, 126 ± 15 μm versus 93 ± 14 μm, respectively (Fig. 4) . This phenotype difference was significant (P = 0.008). There were no apparent differences in the corneal epithelium or endothelium. 

Type V Collagen Expression in Col5a1 ^{+/ −} Mice.

There was a dramatic reduction in stromal type V collagen immunoreactivity in the Col5a1 ^{+/ −} mice relative to the wild-type mice. Immunofluorescence microscopy with an antibody against the aminoterminal noncollagenous domain of type V collagen demonstrated a comparable distribution throughout the corneal stroma in both wild-type and mutant mice. However, the Col5a1 ^{+/ −} mice demonstrated a reduction in signal consistent with the haploinsufficient phenotype (Fig. 5A) . Control subjects without primary antibody were negative (data not shown). This was further analyzed with a semiquantitative Western blot analysis of samples representing comparable fractions of the cornea. The results demonstrate a 49% decrease (range, 34%–79%) in type V collagen in the Col5a1 ^{+/ −} stroma (Fig. 5B) . Again, this is consistent with a Col5a1 haploinsufficient phenotype. 

Collagen Deposition in Col5a1 ^{+/ −} versus Wild-Type Corneas.

The thinner corneal stroma in the Col5a1 ^{+/ −} mice suggested an alteration in collagen content and/or fibril number. Therefore, the collagen contents of 5-month-old wild-type and haploinsufficient corneas were analyzed. Quantitation of collagen as hydroxyproline showed an approximate 14% reduction in total collagen deposition in Col5a1 ^{+/ −} corneas when compared with wild-type corneas (22.1 ± 5.3 μg vs. 19.0 ± 3.2 μg). This reduction was observed consistently in 3.5- to 5-month-old corneas (data not shown). The difference between the two genotypes cannot be attributed solely to Col5a1 haploinsufficiency, which accounts for a reduction in the 5% to 10% range. 

Fibril Structure and Diameter in the Stroma of Col5a1 ^{+/ −} Mice.

Mature (12 week) Col5a1-haploinsufficient mice had stromal collagen fibrils with larger diameters than those seen in the wild-type littermates (Fig. 6A) . The fibrils in both genotypes were cylindrical in shape and were regularly packed within organized lamellae. All fibrils throughout the stroma showed an increase in diameter in the Col5a1 ^{+/ −} corneas relative to the wild-type corneas, without any regional differences. Fibril diameter distributions were analyzed in the mature Col5a1 ^{+/ −} corneas and were significantly larger (P < 0.001) than the fibrils from wild-type corneas (Fig. 6B) . Increased mean fibril diameter was observed at all ages studied, from postnatal day (P)10 to 12 weeks (Table 3) . The anterior and posterior stroma both had a normal distribution of fibril diameters with a comparable fibril phenotype. At 12 weeks, in the anterior stroma of Col5a1 ^+/− mice, the fibrils were 7.7 nm larger (95% confidence limits [CL]: 5.9, 9.5), whereas in the posterior stroma, the fibrils were 6.3 nm larger (95% CL: 4.6, 8.1) when compared with the wild-type corneas. The larger-diameter fibrils were present in the Col5a1 ^+/− stroma at all developmental stages studies (i.e., P10 and 6 weeks and 12 weeks; Fig. 6B ). In the anterior stroma of Col5a1 ^+/− mice the fibrils were 5.7 nm larger (95% CL: 4.8, 6.6), whereas in the posterior stroma the fibrils were 5.4 nm larger (95% CL: 4.5, 6.3) when compared with the wild-type corneas across all ages studied. The developmental stability indicates that increased fibril diameter is not an acquired phenotype, but rather is due to defects in regulation at the earliest stages of fibrillogenesis. 

Fibril Density in Col5a1 ^+/− Stromas.

In addition to the larger-diameter fibrils, the haploinsufficient animals were characterized by a decreased number of collagen fibrils present in the stroma at all postnatal stages examined and in both the anterior and posterior stroma (Fig. 7A) . The differences in fibril density (fibrils per micrometer) associated with genotype (mutant versus wild-type, P < 0.001) and location (anterior versus posterior, P < 0.001) were significant. Across all ages, the density of fibrils in mutants was lower by 25% (95% CL: 16, 33). The anterior cornea was reduced by 29% (95% CL: 20, 37), and the posterior cornea was reduced by an average of 20% (95% CL: 10, 29; Fig. 7B ). Fibril packing was comparable in the Col5a1 ^+/+ and Col5a1 ^+/− stromas. A nearest-neighbor analysis was performed in the mature, compacted 12-week-old corneas. The packing in both genotypes was comparable. The mean fibril spacing in the Col5a1 ^+/+ and Col5a1 ^+/− anterior and posterior stromas was 16.6 ± 2.5 nm versus 17.7 ± 2.6 nm and 22.8 ± 4.4 nm versus 22.7 ± 4.4 nm, respectively. 

Discussion

This is the first study to define the human ocular phenotype of type V collagen mutations. The patients examined had thin, steep and transparent corneas as well as floppy eyelids. The clinical description of classic EDS (types I, II) in the revised Villefranche classification does not mention any ocular involvement. ¹ In other types of EDS, such as the kyphoscoliosis EDS (type VI), vascular EDS (type IV), and dermatosparaxis EDS (type VIIc) characteristic ocular manifestations include laxity of lids, ptosis, scleral fragility, blue sclera, strabismus, myopia, microcornea, cornea plana, keratoconus, keratoglobus, limbus-to-limbus corneal thinning, cataract, ectopia lentis retinal detachment, and angioid streaks. ^{1 32 33 34 35 36}  

Adult corneal thickness is reached by ∼3 years of age. ³⁷ A meta-analysis performed on 300 data sets from human eyes generated a mean normal central corneal thickness in pachymetry studies that was close to 535 μm. ³⁸ The consistent corneal thinning observed in the patients with EDS was consistent with that observed in the Col5a1 ^+/− mouse. In the mature Col5a1-haploinsufficient mouse the corneal stroma was ∼26% thinner than the wild-type stroma, compared with the 24% decrease in corneal thickness in the patients with EDS. The thinning of the corneal stroma in the Col5a1 ^+/− mouse model was associated with a ∼14% decrease in collagen content and a 25% reduction in the number of stromal fibrils. The significant reduction in fibril number is likely to be responsible for the thinning of the stroma. The data from in vitro studies of dermal fibroblasts of patients with classic EDS due to COL5A1 haploinsufficiency and the Col5a1 ^+/− mouse indicate that heterotypic type I/V collagen interactions are involved in the efficient initiation of fibril assembly. ^{21 22 39} The reduction in type V collagen in the haploinsufficient mice results in fewer nucleation sites. If fewer sites are used to initiate fibril assembly with a constant type I collagen pool, fewer larger-diameter fibrils are expected. This is the case in the Col5a1 ^+/− corneal stroma. 

There was a significant increase in mean fibril diameter in the stroma of Col5a1 ^+/− mice, generating a homogeneous population of larger-diameter fibrils compared with the wild-type corneas. Presumably, this increase was to accommodate more collagen per fibril nucleation site (i.e., type V collagen molecule). However, the fibril phenotype was not as severe as that observed in the dermis. The dermal phenotype was virtually identical with that observed in patients with the classic form of EDS, with respect to fibril morphology, skin fragility, and wounding (Wenstrup B, Florer J, Brunskill E, et al., manuscript submitted). ²² In the cornea, a single population of larger fibrils was assembled. In contrast, in the dermis of Col5a1 ^+/− mice a subpopulation of larger fibrils with circular profiles was formed; however, a second subpopulation of very large, structurally aberrant fibrils was also assembled, comparable to the fibril phenotype in patients with classic EDS. This difference is presumably due to higher type V collagen concentration in the cornea (10%–20%) when compared with the dermis (2%–5%). Because of the high type V collagen content of the cornea, enough remaining nucleation sites were available for fibril assembly that a second structurally aberrant fibril population did not form in an unregulated manner as is the case with dermis, in which the low type V collagen content is limiting (Wenstrup B, Florer J, Brunskill E, et al., manuscript submitted). ²² The finding of larger-diameter fibrils and decreased fibril density and number at the early developmental stages indicates that this structural phenotype reflects deficiencies in initial fibril assembly and not merely secondary effects related to fibril growth, stability, or turnover. 

The consistent thinning of the corneal stroma in both EDS patients and the mouse model primarily results from the observed nucleation of fewer collagen fibrils throughout corneal development and growth. The analyses of fibril packing in the mature, compacted stroma indicate that the reduction in type V collagen did not affect fibril packing, and therefore altered packing was not involved in the observed thinning. The 14% reduction in corneal collagen content can be partially (5%–10%) explained by the reduction in type V collagen. The remaining reduction may be a result of less efficient utilization of type I collagen during fibril assembly due to fewer nucleation events and the resultant clearance of unutilized collagen. A decreased collagen utilization may make a minor contribution to the thinner stromas. 

The structural basis of corneal transparency requires a homogeneous distribution of small diameter fibrils, regular fibril packing, and arrangement as orthogonal lamellae. ²⁰ The Col5a1 ^+/− stromas have significantly larger-diameter fibrils than do wild-type stromas. Both patients with EDS and Col5a1 ^+/− mice had no measurable alterations in corneal transparency. As a result of the latter finding, it is likely that the possible adverse effect of increased fibril diameters on transparency was offset by the persistence of cylindrical fibrils with a narrow distribution of diameters and regular packing. Similar increases in fibril diameter and retention of narrow diameter distribution and regular packing were observed in mucopolysaccharidosis III with no alteration in corneal transparency. ³⁹  

The association between corneal thickness and refraction has been analyzed by several authors but remains controversial. ^{40 41 42 43} Price et al., ⁴¹ did not observe a correlation between the central corneal thickness and the spherical equivalent. Other studies revealed significant differences in the central corneal thickness among myopic, hyperopic, and emmetropic individuals, whereas the latter had the thinnest corneas. ⁴² Recently, Sanchis-Gimeno et al., ⁴³ documented normal values in a study in 1000 young emmetropic subjects measuring the corneal thickness (Orbscan Topography System II). In the present study, of the eight eyes in this study with type V collagen mutations, one eye had mild myopia, one eye had moderate myopia, and six eyes were emmetropic, whereas four of the healthy control eyes (eight eyes) had bilateral mild myopia. The latter findings support the lack of correlation between corneal thickness and refractive error. 

In summary, the hallmark of the human ocular phenotype of COL5A1 and COL5A2 mutations was thinning of the cornea, which did not appear to change significantly with age. The ocular phenotype in the Col5a1 ^+/− mouse model also included thinning of the corneal stroma which contained a reduced density of collagen fibrils that were thicker than normal, but narrow in their distribution of diameters, with normal cylindrical shape and regular packing. The corneal abnormalities did not affect corneal transparency and the corneal thinning in the humans with type V collagen mutations did not impair vision. However, considering that corneal thickness influences the interpretation of intraocular pressure measurement, this may be important in the screening for glaucoma. ^{44 45 46} In addition, refractive surgery is likely to be contraindicated because of anticipated poor corneal wound healing and the risk of iatrogenic corneal ectasia. 

The consistent phenotypic findings in the seven individuals with COL5A1 haploinsufficiency are likely to apply to other patients with similar mutations. However, additional patients with other types of COL5A1 mutations as well as more patients with COL5A2 mutations should be studied to obtain more phenotypic information. Additional studies of the Col5a1 ^+/− mouse eye, particularly of the noncorneal components that contain type V collagen, are also needed. 

 Supported in part by the Canadian Genetic Disease Network (EH), the Canadian Institutes of Health (WGC), the Canadian Arthritis Network (WGC) and NIH Grants EY05129 (DEB) and AR47054 (RJW).

 Submitted for publication June 19, 2005; revised August 6, 2005; accepted December 7, 2005.

 Disclosure: F. Segev, None; E. Héon, None; W.G. Cole, None; R.J. Wenstrup, None; F. Young, None; A.R. Slomovic, None; D.S. Rootman, None; D. Whitaker-Menezes, None; I. Chervoneva, 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.

 

Table 1.

 

Mutations of Patients with EDS

Proband	Gene	Mutation*	Effect
A-II-I	COL5A1	3957 del GA 6	Frameshift + Stop: X1265, †
B-I-I	COL5A1	4692-4693 ins C	Frameshift + Stop: X1623, †
C-II-I	COL5A2	IVS-26-1 del AG + 1924-1928 del GGAGC 11	Del G642-P659 in frame, ‡

 Mutation 4692-4693 ins C is not published.

*  All probands were heterozygous for their mutations. DNA numbering is based on the cDNA sequences GenBank NM_000093 (COL5A1) and NM_000093 (COL5A2) with +1 as A of the ATG start codon. Amino acid numbering is based on the deduced amino acid sequence with the initiator methionine numbered as +1.

†  These mutations resulted in mRNA degradation and haploinsufficiency in type V collagen.

‡  Peptide deletion due to skipping of the 54-bp exon 28, resulting in a mutated type V collagen molecule with potential dominant negative effects.

Figure 1.

 

Pedigrees of EDS families studied. Circles: females; squares: males. Filled symbols: affected; open symbol: unaffected. The identifier number refers to the age at which the last examination took place. Individuals represented by symbols without an identifier number were not examined. Arrow: index case, proband.

Table 2.

 

Ocular Findings in the Patients with EDS

Case*	BCVA			Mean K Reading, † (D)			SRI, † (D)			SAI, † (D)			Corneal Thickness (μm)
		OD	OS	OD	OS	OD	OS	OD	OS	OD	OS
A-I-1	20/25	20/20	45.81	46.26	0.70	0.51	0.44	0.90	442	419
A-II-1	20/20	20/20	46.80	46.61	0.85	0.34	3.23	0.44	430	422
A-II-2	20/20	20/25	46.28	47.80	0.32	0.87	0.99	0.60	415	415
A-II-3	20/25	20/20-3	46.44	46.63	0.42	0.63	0.37	0.77	445	445
B-I-1	20/20	20/20	48.09	46.39	0.72	0.23	4.29	0.52	448	442
B-II-1	20/25	20/20	44.64	44.86	0.29	0.20	0.85	0.40	447	425
B-II-2	20/20	20/25+2	44.37	45.64	0.57	1.34	2.00	3.40	438	447
C-II-1	20/400	20/20	47.26	47.06	0.94	0.82	3.60	1.06	447	445

 BCVA, best corrected visual acuity.

*  The mean values ± standard deviation for eight age-matched control subjects are: mean corneal thickness = 568.5 ± 30.25 μm; mean K readings = 44.31 ± 1.10 D; mean SRI = 0.35 ± 0.24 D; and mean SAI = 0.49 ± 0.25 D.

†  These K, SRI and SAI values define the regularity of the corneal surface and the symmetry of the corneal conical shape, which determine the optical potential. Abnormal values suggest disease.

Figure 2.

 

Example of a floppy eyelid. In this case, spontaneous eversion of the left upper lid.

Figure 3.

 

Top: Corneal thickness in EDS patients studied (n = 8). Bottom: mean value of keratometric readings with SD in EDS patients studied (n = 8).

Figure 4.

 

Corneal stromal thickness is decreased in mature Col5a1 ^+/− mice. (A) Representative light micrographs of 12-week corneas from wild-type (+/+) and Col5a1-haploinsufficient (+/−) mice. (B) Bar graph of mean stromal thickness (±SD) for Col5a1 ^+/+ (n = 6) and Col5a1 ^+/− (n = 5) mice. Ep, epithelium; S, stroma; En, endothelium.

Figure 5.

 

(A) Type V collagen was localized in wild-type (+/+) and haploinsufficient (+/−) corneas using immunofluorescence microscopy. The haploinsufficient stromas have decreased reactivity for type V collagen relative to the wild-type control subject. Right: exposures of the same sections to DAPI, a nuclear counterstain. The controls without primary antibody were negative (not shown). Ep, epithelium; S, stroma. (B) Type V collagen expression in wild-type (+/+) and haploinsufficient (+/−) corneas was analyzed using semiquantitative Western analyses. The haploinsufficient corneas had a 49% (34%–79%) reduction in type V collagen relative to the wild-type control subjects. Corneas from 5-month-old mice were extracted in equal volumes and serial dilutions, representing constant fractions of the corneas, were blotted with the anti-α1(V) antibody. Numbers represent the corneal fraction loaded per well.

Figure 6.

 

Stromal collagen fibrils in Col5a1 ^+/− mice have larger diameters than in wild-type mice. (A) Transmission electron micrographs of collagen fibril architecture in the corneal stroma of 12-week-old Col5a1 ^+/− and Col5a1 ^+/+ mice. In the wild-type stroma, there was a large number of small-diameter collagen fibrils. The Col5a1 ^+/− corneal stromas contain fewer, larger-diameter fibrils. Bar, 100 nm. (B) Collagen fibril diameter distributions show a shift to a larger-diameter fibril population in Col5a1 ^+/− corneas. Fibril diameter distributions were analyzed in the anterior and posterior stroma of 12-week-old Col5a1 ^+/+ and Col5a1 ^+/− corneas. The fibril diameters in the Col5a1 ^+/− corneas were significantly larger (P < 0.001) than the fibrils from wild-type corneas. In the anterior stroma of Col5a1 ^+/− mice, the fibrils were 7.7 nm larger (95% CL: 5.9, 9.5) than in the wild-type corneas. In the posterior stroma of Col5a1 ^+/− mice, the fibrils were 6.3 nm larger (95% CL: 4.6, 8.1).

Table 3.

 

Murine Mean Collagen Fibril Diameters by Genotype, Age, and Location

Col5a1 Genotype	Stroma Location	Postnatal Age	Diameter Difference (nm)	Standard Error	P	95% Confidence Limits
												Lower	Upper
+/− vs. +/+	Anterior	10 Days	4.85	0.48	<0.001	3.55	6.14
+/− vs. +/+	Anterior	6 Weeks	4.64	0.89	0.002	2.48	6.80
+/− vs.+/+	Anterior	12 Weeks	7.67	0.77	<0.001	5.90	9.45
+/− vs. +/+	Posterior	10 Days	5.74	0.49	<0.001	4.45	7.03
+/− vs.+/+	Posterior	6 Weeks	4.11	0.89	0.003	1.95	6.27
+/− vs. +/+	Posterior	12 Weeks	6.32	0.77	<0.001	4.55	8.09

Figure 7.

 

Decreased fibril density in the Col5a1 ^+/− corneal stroma. (A) Fewer, larger-diameter fibrils are present regardless of age in Col5a1 ^+/− corneas. Transmission electron micrographs of collagen fibrils in the corneal stroma of Col5a1 ^+/+ and Col5a1 ^+/− mice at postnatal day (P)10, 6 weeks and 12 weeks. At all ages the Col5a1 ^+/− corneas have cylindrical, larger-diameter fibrils than the wild-type controls. Bar, 100 nm. (B) Col5a1 ^+/− corneas had a significant reduction (P < 0.001) in fibril density relative to wild-type corneas. At P10, 6 weeks, and 12 weeks, the Col5a1 ^+/− anterior stromas had 29% fewer fibrils/unit area than the wild-type controls (P < 0.001). In the anterior stroma of Col5a1 ^+/− versus Col5a1 ^+/+ mice at P10, 6 weeks, and 12 weeks: means of 343 fibrils/μm² (95% CL: 294, 399) versus 458 fibrils/μm² (95% CL: 363, 577); 397 fibrils/μm² (95% CL: 351, 449) versus 573 fibrils/μm² (95% CL: 493, 667); 374 fibrils/μm² (95% CL: 323, 432) vs. 538 fibrils/μm² (95% CL: 462, 625); were observed respectively. Bars, means ± SD.

The authors thank the participants for their enthusiastic contribution, Catherine Barclay for coordination of the patients, and Jane Florer and Biao Zuo for help with the mouse studies.