Murine Model of the Ehlers-Danlos Syndrome

The most commonly identified mutations causing Ehlers-Danlos syndrome (EDS) classic type result in haploinsufficiency of proα1(V) chains of type V collagen, a quantitatively minor collagen that co-assembles with type I collagen as heterotypic fibrils. To determine the role(s) of type I/V collagen interactions in fibrillogenesis and elucidate the mechanism whereby half-reduction of type V collagen causes abnormal connective tissue biogenesis observed in EDS, we analyzed mice heterozygous for a targeted inactivating mutation in col5a1 that caused 50% reduction in col5a1 mRNA and collagen V. Comparable with EDS patients, they had decreased aortic stiffness and tensile strength and hyperextensible skin with decreased tensile strength of both normal and wounded skin. In dermis, 50% fewer fibrils were assembled with two subpopulations: relatively normal fibrils with periodic immunoreactivity for collagen V where type I/V interactions regulate nucleation of fibril assembly and abnormal fibrils, lacking collagen V, generated by unregulated sequestration of type I collagen. The presence of the aberrant fibril subpopulation disrupts the normal linear and lateral growth mediated by fibril fusion. Therefore, abnormal fibril nucleation and dysfunctional fibril growth with potential disruption of cell-directed fibril organization leads to the connective tissue dysfunction associated with EDS.

Type V collagen is a quantitatively minor fibril-forming collagen. Several isoforms of type V collagen exist, differing in the type and ratios of constituent ␣ chains, including heterotrimeric molecules containing type XI collagen chains. The pro␣1(V) chain, encoded by COL5A1 at human chromosomal locus 9q34, is the rate-limiting component of type V collagen trimer assembly by virtue of the eight-cysteine motif in the NC1 domain (19). The most abundant and most widely distributed isoform of type V collagen is the [␣1(V)] 2 ␣2(V)] heterotrimer that co-assembles with type I collagen as heterotypic fibrils (20). This isoform of type V collagen retains a non-collagenous, N-terminal domain that is present on the fibril surface, and this domain has been demonstrated to have regulatory functions (21)(22)(23). Disruption of [␣1(V)] 2 [␣2(V)] heterotrimer synthesis using a dominant negative approach (24) or utilizing fibroblasts from EDS patients with characterized mutations in COL5A1 have demonstrated that heterotypic collagen I/V interactions are involved in regulation of fibril diameter and fibril number in vitro (25). In the total absence of the ␣1(V) chain, collagen fibrils were virtually absent though embryonal fibroblasts from col5a1Ϫ/Ϫ mice synthesize and secrete normal amounts of type I collagen; mice died at the onset of organogenesis (26). In addition, severe reduction of type V heterotrimers in favor of homotrimers leads to deposition of an abnormal dermal matrix (27). Evidence indicates that the [␣1(V)] 3 homotrimer does not associate with type I collagen in fibrils (20). These data support a critical role for the [␣1(V)] 2 [␣2(V)] heterotrimer in the nucleation of fibril assembly, consistent with its early evolutionary appearance.
Mice heterozygous for a col5a1 mutation show a 50% reduction in type V collagen and recapitulate many of the clinical, biomechanical, morphologic, and biochemical features of the Ehlers-Danlos syndrome, classic type and are thus excellent models for the classic form of EDS for use in further studies in the regulation of collagen biogenesis and for potential therapeutic interventions. Analysis of heterotypic, structurally abnormal fibrils from dermis of col5a1-haploinsufficient mice provides insights into the mechanisms by which collagen fibrillogenesis is regulated and the mechanism of abnormal connective biosynthesis underlying the Ehlers-Danlos syndrome.

MATERIALS AND METHODS
Generation of col5a1-deficient Mice-The construction of the targeting vector was previously described in detail (25). The col5a1-targeting vector was generated using the col5a1 sequence obtained from the Celera mouse genomic data base. Gene-specific primers amplified a 5Ј-targeting arm from a region that included part of exon 3 and a 3Ј-targeting arm from a region that included part of exon 4 from KG-1 embryonic stem cell DNA (see Fig. 1D for location of the recombination site relative to functional domains within col5a1). Germ line transmission was obtained by breeding chimeric animals to C57BL/6. Quantitation of mRNA-Quantitation of col5a1, col5a2, and col1a1 mRNA in total RNA from mice tails was examined by ribonuclease protection assay (RPA III; Ambion). The probes were obtained using PCR to attach the T7 promoter sequence to the 5Ј-end of the antisense sequence for transcription of an antisense RNA probe.
The col5a1 probe used for RPA was complementary to sequences in exon 64, downstream from the recombination-mediated deletions of exons 3-4 (Fig. 1D). RPA with this probe was designed to exclude the possibility that a transcript encoding a truncated peptide containing a reconstituted C-propeptide could have a dominant negative effect by co-assembling with wild-type pro␣1(V) or pro␣2(V) chains.
Amplified DNA was used directly in the transcription reaction; 10 ng was transcribed by T7 RNA polymerase in the presence of [ 32 P]UTP using reagents for in vitro transcription from Ambion (MaxiScript). The labeled probes were gel purified following the manufacturer's suggestions. Total cellular RNA (2.5-10 g) and at least 2 fmol of each labeled probe were used for each assay following the manufacturer's instructions. Hybridization was carried out at 68°C for 18 h followed by RNase digestion at 37°C for 30 min using 2.5 units/ml of RNase A and 100 units/ml of RNase T1. Protected fragments were separated on a 5% acrylamide/8 M urea gel and detected by exposure to X-OMAT (Kodak) for 18 h at Ϫ80°C with an intensifying screen. Films were scanned and evaluated using Pharmacia GSXL system.
Measurement of Total Collagen Deposited in the Skin-Skin was harvested from animals sacrificed at postnatal days P0 and P10, and at 4, 6, 8, 12, and 20 weeks. Two punch biopsies were taken from each animal, to the right and left of midline on the upper back. Samples were hydrolyzed in 6 N HCl at 100°C for 18 h. Colorimetric analysis of the hydroxyproline content of each skin sample was adapted from the method of Berg (25,28). The mean collagen content from the two samples per animal was calculated and normalized to skin area in millimeters squared. The conversion ratio of 0.12:1.0 was used to convert micrograms of hydroxyproline to total collagen. Semiquantitative measurement of type V content was performed after dermis was directly extracted into SDS buffer and analyzed by Western blot, using an antisera to a peptide sequence encoded by exon 6 in the NC3 domain (25), downstream from the exon 3-4 region targeted by homologous recombination DNA (see Fig. 1D for location of the peptide epitope relative to the recombination site within col5a1).
Biomechanical Analysis-The thoracic aortas were dissected, separated into ascending and descending portions, threaded with stainless steel hooks, and circumferential load-extension curves were obtained using a TA-XT2 texture analyzer (Stable MicroSystems) as described previously (29). Load-extension curves were analyzed for thoracic aortic maximal breaking strength (F max ) and incremental elastic modulus as described previously (29). Tensile testing of wounded and unwounded skin was performed as previously described (30).
Transmission Electron Microscopy-Wild-type and col5a1ϩ/Ϫ postnatal day 10 (P10) as well as 6-, 12-, and 20-week male mice were used in these experiments. The subscapular dermis was analyzed for wild-type and haploinsufficient postnatal animals. Tissues were prepared for transmission electron microscopy as previously described (17,31). Briefly, fixation was in 4% paraformaldehyde/2.5% glutaraldehyde/0.1 M sodium cacody-late, pH 7.4 with 8.0 mM CaCl 2 for 2 h on ice, followed by post-fixation with 1% osmium tetroxide for 1 h. After dehydration in a graded ethanol series followed by propylene oxide, the tissues were infiltrated and embedded in a mixture of Embed 812, nadic methyl anhydride, dodecenylsuccinic anhydride, and DMP-30 (EM Sciences, Hatfield, PA). Thick sections (1 m) were cut and stained with methylene blue-azure B for light microscopy and selection of specific regions for further analysis. Thin sections were prepared using a Reichert UCT ultramicrotome and a diamond knife. Staining was with 2% aqueous uranyl acetate followed by 1% phosphotungstic acid, pH 3.2. Sections were examined and photographed at 75 kV using a Hitachi 7000 transmission electron microscope. The microscope was calibrated using a line grating. For measurement of collagen fibril density, the dermis was divided into four equal regions. Analysis was done for both the superficial and deep dermis. The superficial dermis was defined as the region subjacent to the epidermis, and the deep dermis was the lower quarter. Both regions were photographed in the central portion. Micrographs were taken at ϫ31,680. Calibrated micrographs from each region were randomly chosen in a masked manner from the different regions. The micrographs were digitized, and all diameters were measured within a 1.6-m mask. The mask was placed based on fibril orientation, i.e. cross-section and absence of cells. Diameters were measured along the minor axis of cross-sections using a RM Biometrics-Bioquant Image Analysis System (Memphis, TN). Quantitation of fibrils/m 2 was measured from transmission electron micrographs of cross-sections from the deep dermis of the subscapular region of 10-day, 6-week, and 20-week wild-type and heterozygous mice.
Immunoelectron Microscopy-The subscapular dermis of col5a1ϩ/Ϫ and wild-type mice (15 weeks) was fixed in 4% paraformaldehyde in PBS for 30 min at 4°C. The tissues were washed in PBS, free aldehydes were reduced by incubation in 0.05% sodium borohydride in PBS, and then tissues were rinsed and blocked with 50 mM glycine in PBS. The samples were dehydrated to 70% ethanol and then infiltrated and embedded in LR White resin (EMS, Hatfield, PA). Thin sections were cut and picked up onto formvar-coated nickel grids. For antibody labeling, sections were blocked with 5% normal goat serum in PBS for 60 min at room temperature and then either anti-collagen V antibody (25 g/ml) diluted in PBS containing 0.1% bovine serum albumin and 0.05% Tween 20 or buffer alone overnight at 4°C. After four washes, the sections were incubated with goat anti-rabbit secondary antibody conjugated to 10 nm colloidal gold (Ted Pella, Redding, CA) for 1 h at room temperature, washed four times, rinsed with deionized water, and post-stained with 2% aqueous uranyl acetate. Sections were examined using a Tecnai 12 transmission electron microscope with a Gatan Multiscan 1000 camera.
Statistical Methods-For each micrograph, the number of fibrils within a 1.6-m 2 mask was divided by the area of the box to compute the fibril density. For the fibril diameters, it was not adequate to assume normal distribution within micrographs, even for the diameters trimmed of outliers (mostly abnormally large fibrils). In each micrograph, the median, which is robust (minimally affected by outliers), was used as a measure of the central location of the fibril diameter distribution. The micrograph-specific fibril densities and medians were analyzed separately by fitting a linear mixed effects model (33) incorporating animal-to-animal and micrograph-to-micrograph variability. The fibril densities were log transformed to satisfy the normality assumptions of the model (33). Based on examination of residuals from the fitted models, the model assumptions were adequate.

RESULTS
Targeted Disruption of col5a1 Causes Haploinsufficiency-The most common mutation resulting in classic EDS is the functional loss of one COL5A1 allele. We created a mouse model to study the molecular col5a1 Haploinsufficiency Disrupts Collagen Fibril Formation MAY 5, 2006 • VOLUME 281 • NUMBER 18 mechanism generating the clinical phenotype by targeted disruption of col5a1. RPA of mRNA isolated from col5a1ϩ/ϩ and col5a1ϩ/Ϫ animals showed half-reduction of col5a1-derived mRNA compared with col5a2-dervived mRNA by densitometric analysis of the autoradiograms, 0.89 Ϯ 0.14 (n ϭ 8) versus 0.36 Ϯ 0.03 (n ϭ 10), respectively. There was no difference in the steady state levels of col1a1 mRNA expression levels between col5a1ϩ/Ϫ and wild-type mice (Fig. 1, A and B). Densitometric scanning of Western blots of dermal proteins from P10 animals indicated that there was a 50% decrease in the type V collagen content in skin from the col5a1ϩ/Ϫ animals compared with wild-type littermates, demonstrating a direct relationship between mRNA and protein content. Corneal tissue, which has substantially higher type V collagen content as a proportion of total collagen than other connective tissues, was also analyzed. The type V collagen content of col5a1ϩ/Ϫ mouse corneas was only 55% that of wild-type littermates by densitometric scanning of Western blots (data not shown).
Vascular Phenotype in col5a1ϩ/Ϫ Animals-Ehlers-Danlos syndrome classic type is associated with high prevalence of aortic root dilation (34) and more rarely with rupture or dissection. Easy bruisability is a hallmark of the condition and is evidence of capillary fragility. The col5a1Ϫ/Ϫ mouse embryos and yolk sacs demonstrated blood pooling and reduced circulation at E10 before cessation of the heart's rhythmic contractions, indicating that cardiovascular insufficiency was a factor in embryonic demise (25). To determine whether cardiovascular tissue was measurably compromised in col5a1-haploinsufficient animals, biomechanical analyses were performed on ascending and descending aortas from 12-week-old animals. The results indicated that col5a1 haploinsufficiency results in decreased aortic stiffness and breaking strength (Fig. 2) (8); p Ͻ0.0001) decrease in incremental elastic modulus for the descending aorta. The difference between col5a1ϩ/Ϫ and ϩ/ϩ animals was greater in the descending aorta than in the ascending aorta, which is consistent with the current understanding that collagen and elastin bear the majority of the wall stress and determine the stiffness (compliance) of the aorta (35). As the distance from the heart increases, the elastin content is known to decrease and the collagen content to increase, with a net result of higher collagen to elastin ratios in the descending aorta (36,37).
Dermal Phenotype in col5a1ϩ/Ϫ Animals-Dermal features of hyperelasticity, friability, and poor wound healing are unique and characteristic features of the Ehlers-Danlos syndrome, classic type. Dermal abnormalities were immediately apparent in col5a1ϩ/Ϫ animals; the skin was hyperextensible in the col5a1ϩ/Ϫ mice relative to wild-type controls, but there was no skin redundancy (Fig. 3A). The tensile strength of normal and wounded skin was reduced in the col5a1ϩ/Ϫ mice relative to wild-type controls. Biomechanical testing of tensile (breaking) strength was performed on animals at ϳ12 weeks of age. Unwounded skin of col5a1ϩ/Ϫ animals failed at 27.05 Ϯ 3.53 kilogram-feet tensile stress/centimeter squared (Kgf/cm 2 ) compared with 51.99 Ϯ 7.20 (t-test ϭ 0.029625) for wild-type littermates (Fig. 3B). In addition, the tensile strength of incisional wounds on dorsal skin 8 days after wounding was analyzed. There was a significant reduction in wound strength in the col5a1ϩ/Ϫ animals relative to wild-type controls, 5.57 Ϯ 0.48 v. 9.50 Ϯ 2.57 Kgf/cm 2 (t-test ϭ 0.006071), respectively. All parameters were compatible with those seen in patients with the classic form of EDS.
Dermal thickness increases significantly in young adult C57BL/6 wildtype mice and is correlated with an increase in collagen content as determined by measurement of hydroxyproline. The col5a1ϩ/Ϫ mice demonstrated a significant delay in dermal collagen accumulation between 4 and 8 weeks of age, but by 12 weeks the total quantity of dermal collagen of col5a1ϩ/Ϫ mice approximates that of wild-type littermates (Fig. 4). The finding of quantitatively normal dermal collagen content in 12-weekϩ/Ϫ animals was surprising in light of the observed reduction in biomechanical

col5a1 Haploinsufficiency Disrupts Collagen Fibril Formation
properties of the skin and suggested the possibility of qualitative defects in dermal architecture. Light microscopy of dermis from ϩ/Ϫ animals at 12 weeks demonstrated that there was collagen fiber disarray and a general appearance of reduced density of dermal connective tissue compared with wild-type littermates (Fig. 5).
Decreased Fibril Density in the col5a1ϩ/Ϫ Dermis-Analysis of dermal collagen fibrils by transmission electron microscopy demonstrated that the fibril density (number of fibrils/m 2 ) was reduced by 38% to 46% relative to the wild-type controls at all developmental stages analyzed between postnatal day 10 and week 20 in the deep, subscapular dermis (Fig. 6). During this period of development and maturation of the dermis, the fibril density decreases as fibril diameter increases in the wild-type dermis (see also Fig. 7). The density of fibrils was on average 42% lower (95% CI: 34%, 50%; p Ͻ0.001) in mutant mice across all ages. Similar significant differences were observed at P10, 6 weeks, and 20 weeks. At P10 there were 185 fibrils/m 2 (95% CI: 154, 223) in the wild-type dermis and 114 fibrils/m 2 (95% CI: 95, 137) in the col5a1ϩ/Ϫ mice. At 6 weeks the values were 89 fibrils/m 2 (95% CI: 76, 104) and 48 fibrils/m 2 (95% CI: 41, 57), and at 20 weeks there were 52 fibrils/m 2 (95% CI: 44, 60) and 29 fibrils/m 2 (95% CI: 25, 35) in wildtype and col5a1ϩ/Ϫ mice, respectively (Fig. 6). This decrease was observed in papillary dermis as well as in dermis from the axillary region (data not shown). These data indicate a numerical reduction in fibril formation events in the col5a1-haploinsufficient dermis independent of developmental stage or tissue site.
Analyses of the fibril diameter distributions indicated two fibril subpopulations in the col5a1-haploinsufficient dermis at all stages analyzed between P10 and 20 weeks (Fig. 7). One was a relatively symmetrical subpopulation comparable with the wild-type distribution, only broader and shifted to larger diameters. The second was a subpopulation of very large diameter fibrils, present as a right shoulder on the diameter distributions. These correspond to the large, structurally aberrant fibrils seen in the electron micrographs in Fig.  8. Across the ages studied, the medians of fibril diameter distributions from the mutants were on average 17.6 nm larger (95% CI: 9.7, 25.5; p Ͻ0.001) than the medians in the wild type. The genotype differences generally increased with age. The average difference between the medians for the col5a1ϩ/Ϫ and normal mice at P10 was 8.1 nm (CI 95%: Ϫ7.2, 23.4; p ϭ 0.224). At 6 and 20 weeks the differences were 19.3 nm (CI 95%: 6.0, 32.6; p ϭ 0.009) and 25.3 nm (CI 95%: 11.8, 38.8; p ϭ 0.001), respectively.   MAY 5, 2006 • VOLUME 281 • NUMBER 18

col5a1 Haploinsufficiency Disrupts Collagen Fibril Formation
Abnormal Fibril Subpopulation in the Haploinsufficient P10, 6W, and 20W Dermis-In addition to the population of fibrils with circular profiles larger but otherwise comparable with those seen in the wild-type dermis, col5a1ϩ/Ϫ dermis contained a second subpopulation of very large, structurally aberrant fibrils (Fig. 8). This dermal phenotype was present in developing (P10 and 6 week) and mature (20 week) dermis. Like the normal fibrils in wild-type animals and the regular, cylindrical fibril subpopulation in haploinsufficient animals, the abnormal subpopulation also increased in size from P10 to 20 weeks.
Type V Collagen Is Absent in the Structurally Abnormal Subpopulation of Fibrils-The normal cylindrical fibrils in wild-type dermis demonstrate a periodic reactivity with the anti-collagen V antibody (Fig. 9) in both longitudinal and cross-sections. In contrast, in the haploinsufficient dermis the large, structurally aberrant fibrils demonstrated virtually no reactivity for type V collagen. However, the smaller, cylindrical fibrils in dermis of col5a1ϩ/Ϫ animals demonstrated a periodic reactivity with the collagen V antibody similar to wild-type dermis. This is consistent with our hypothesis that the abnormal fibrils, formed where type V collagen is limiting, represent unregulated assembly of type I collagen.
Interactions of Fibril Subpopulations Increase with Developmental Stage in the col5a1-haploinsufficient Dermis-Lateral interactions between the cylindrical fibril subpopulation and the structurally abnor-

col5a1 Haploinsufficiency Disrupts Collagen Fibril Formation
mal subpopulation were observed (Fig. 10). These observations were more common in the later developmental stages, i.e. 6 -20 weeks. At P10, a stage where the fibril growth is beginning, there are fewer interactions than at the 20-week time point where normal fibril growth is complete. The extensive interactions between the two subpopulations in the col5a1-haploinsufficient dermis at 20 weeks are evidence of dis-ruption of fibrillogenesis at two distinct stages. The type V collagendeficient dermis has altered fibril initiation, and therefore an aberrant fibril subpopulation is generated. Second, the linear and lateral growth of mature fibrils mediated by fusion of early immature fibril intermediates is disrupted through the interactions of relatively normal and structurally abnormal subpopulations (Fig. 11).

DISCUSSION
A mouse model with a targeted mutation in col5a1 was utilized to determine the mechanisms of collagen fibril abnormalities associated with the  . Collagen V reactivity is associated with cylindrical fibrils, but not structurally aberrant fibrils. Ultrastructural immunolocalization of type V collagen in col5a1ϩ/ϩ and ϩ/Ϫ dermis. The normal cylindrical fibrils in wild-type dermis (A-D) and the normal subset in the col5a1ϩ/Ϫ dermis (E-G) demonstrate a periodic reactivity with the anti-collagen V antibody. This is observed in both longitudinal (A-C) and cross-sections (D) of fibrils. In the haploinsufficient (ϩ/Ϫ) dermis the large, structurally aberrant fibrils (arrows) demonstrate virtually no reactivity for type V collagen (E-G). Post embedding immunolocalization was done using dermis from 15-week mice embedded in LR White. Controls were negative (not shown). clinical features seen in patients with Ehlers-Danlos syndrome classic type. Characteristic of the EDS phenotype, the col5a1ϩ/Ϫ mice have decreased compliance and tensile strength of the aorta and hyperextensible skin with decreased tensile strength of both normal and wounded skin. Joint laxity, a prominent feature of human EDS, appears to be absent in col5a1ϩ/Ϫ mice; the most likely explanation for this surprising observation is that type XI collagen functions in dosage compensation in tendon. 3 Heterozygous animals are haploinsufficient for col5a1 with a 50% reduction in the col5a1 mRNA and tissue type V collagen compared with wild-type animals. The col5a1ϩ/Ϫ mice demonstrate grossly defective collagen fibril formation. Fibril number is reduced by approximately one-half at all developmental stages, supporting the hypothesis that type V collagen is critical for fibril nucleation. We have recently shown that pro␣1(V)-containing type V collagen molecules are absolutely required for mesenchymal collagen fibril formation (25). At the time of fetal demise at E10.5, col5a1Ϫ/Ϫ mice contain virtually no collagen fibrils in the predermal mesenchyme; only a very small number of abnormal collagen-containing aggregates appear at the site of the future dermal-epidermal junction. In contrast, the col5a1ϩ/Ϫ described in this report assemble one subpopulation of fibrils that is relatively normal with cylindrical profiles and periodic immunoreactivity for collagen V, though the fibril diameters are slightly larger than fibrils in wild-type animals. A second fibril subpopulation consists of very large structurally aberrant fibrils with highly irregular cross-sectional shapes and a virtual lack of type V collagen reactivity.
Collagen fibril formation is initiated in close association with the fibroblast cell surface, in channels at the cell surface of connective tissue fibroblasts from tendon or dermis (31,47) or possibly in Golgi to plasma membrane carriers during embryogenesis (38). Mature collagen fibrils form via an intermediate stage. This intermediate is formed by nucleation and unilateral elongation of collagen fibrils forming short, small diameter, immature fibril intermediates. The assembly of a mature extracellular matrix from immature intermediates involves linear and lateral growth steps (Fig. 11). Mature, longer and larger diameter fibrils form via linear and lateral fusion of fibril intermediates, resulting in diameter enlargement and bidirectional growth (39 -41). Fibril-associated macromolecules such as small leucine-rich repeat proteoglycans have been implicated in the regulation of these later fibril growth steps. Defects that disrupt these later stages in fibrillogenesis would generate altered tissue structure and function. Deficiencies of decorin, lumican, biglycan, and fibromodulin have all been shown to alter fibril morphology at this stage (reviewed by Iozzo (42) and Ameye and Young (43)). Thus, 50% reduction in type V collagen appears to have both quantitative and qualitative effects in murine dermal connective tissue, depending on the stage of development. Consistent with our hypothesis that type V collagen is a rate-limiting fibril nucleator, col5a1ϩ/Ϫ mice have significant reductions in total collagen deposition at the end of the growth phase at 6 weeks. Fiber morphology abnormalities increase over time due to additive fusion events between normal (type V collagen initiated) and abnormal fibrils that appear to lack type V collagen. As a result of the abnormal fusion events, virtually every type I collagen fiber in adult col5a1ϩ/Ϫ mice may contain normal collagen fibril intermediates that are disrupted by interaction with structurally aberrant fibrilintermediate-like structures. Such fusion-mediated architectural dis-3 D. E. Birk, J. B. Florer, and R. J. Wenstrup, manuscript in preparation. FIGURE 11. Reduction in type V collagen disrupts nucleation of collagen fibril assembly and fibril growth steps. Top, schematic diagram of collagen fibril formation in col5a1ϩ/ϩ (A) and col5a1ϩ/Ϫ (B) mice. A, in the wild-type mice, type V collagen nucleates fibril assembly at the fibroblast cell surface, generating immature, small diameter, short fibril intermediates. B, when type V collagen is limiting as in col5a1ϩ/Ϫ mice, a reduced number of relatively normal fibril intermediates are nucleated. Excess type I collagen molecules are sequestered as morphologically abnormal aggregates through unregulated self-assembly, so that both normal and abnormal populations of fibril intermediates are present. Bottom, schematic diagram of fusion of fibril intermediates in col5a1ϩ/ϩ (C) and col5a1ϩ/Ϫ (D) mice. In the wild-type animals a homogeneous population of fibril intermediates grows linearly and laterally by fusion. In the col5a1ϩ/Ϫ mice (D) the interaction of the normal fibril intermediates with the structurally aberrant fibril-like structures further disrupts matrix structure.

col5a1 Haploinsufficiency Disrupts Collagen Fibril Formation
ruptions can be seen as a late stage dominant negative effect (Fig. 11D) and likely account for the fact that dermal tensile strength in haploinsufficient animals is still reduced at 12 weeks (Fig. 3), a time when total dermal collagen content is comparable with wild-type littermates (Fig.  4). Interestingly, in the cornea where the relative ratio of type V collagen to type I collagen is normally 5-to 10-fold higher than in dermis, col5a1 haploinsufficiency does not result in abnormal fibril morphology, though there is a decrease in fibril density as observed in dermis and a ϳ25% reduction in stromal thickness comparable with that seen in classic type EDS patients (46). Our interpretation is that the 5-to 10-fold increase in type V collagen to type I collagen relative to the dermis allows fibril initiation to progress normally and that the sites are able to accommodate the excess type I collagen pool by increasing diameter without sequestration into irregular fibrils. In contrast, the inability of the less abundant type V collagen-nucleating molecules in the dermis to accommodate the type I collagen pool by simply increasing diameter suggests that there is a physical limit to the maximum diameter for an immature fibril intermediate.
The precise interactions and the order of assembly of collagen fibrils in the construction of a fibrillar extracellular matrix is unknown but probably involves at least three processes: (i) interaction of a fibril-initiating collagen molecule with a membrane-bound molecule or receptor, perhaps also mediated by other extracellular matrix macromolecules known to interact with the cell surface such as tenascin X, fibronectin (44), or thrombospondin 2 (45); (ii) assembly of nascent fibril intermediates with type V collagen, likely serving the function of nucleation of these intermediates; and (iii) surface modification of growing fibril intermediates by small leucine-rich polypeptides, large proteoglycans, and fibril-associated collagens with interrupted triple helices (FACIT) collagens. It is likely that further delineation of the order of assembly of a competent collagenous extracellular matrix will require a systems approach to complement the reductionistic analyses currently employed in singly and multiply deficient mice such as has been undertaken in the characterization of the order of assembly of human complex I (NADH; ubiquinone oxidoreductase) on the mitochondrial membrane (10,32). The above approach may be applicable to early collagen fibril formation because it appears to initiate as a membranebound process in cellular subfractions, many collagen fibril-associated proteins have been identified, and fibroblastic cells from mice deficient in many of the early components may be available or can be created through knockdown technologies.
Thus, col5a1ϩ/Ϫ mice, which recapitulate the most common molecular cause of human EDS, classic type, also reproduce several of its clinical and pathological features with high fidelity. The abnormal collagen fibril morphology identified in col5a1ϩ/Ϫ mice is similar to that found in dermis of humans with EDS by Vogel et al. (4). Abnormal shaped "cauliflower" fibrils also were observed in the dermis of EDS patients by Hausser and Anton-Lamprecht (5). These animals will be highly useful models for studying characteristic wound-healing defects in EDS and for testing novel models of therapy for wound-healing disorders.