THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1985 by The American Society of Biological Chemists, Inc.

Vol. 260, No . 20, Issue of ’ September 15, PP. 11322-11329,1985 Printed in U. S. A .

A Heterozygous Collagen Defect in a Variant of the Ehlers-Danlos Syndrome Type VI1 EVIDENCE FOR A DELETED AMINO-TELOPEPTIDE DOMAIN IN THE PRO-a2(I) CHAIN*

(Received for publication, February 26,1985)

David R. EyreSSll, Frederic D. Shapirog, and John F. Aldridgell** From the Departments of $Biological Chemistry, §Orthopaedic Surgery, and 11 Pediatrics, Harvard Medical School and the $§laboratory for Skeletal Disorders and 11 Division of Genetics, The Children’s Hospital, Boston, Massachusetts 02115

A structural defect in the a2(I) chain of type I colla- gen was characterized in a new case of the Ehlers- Danlos syndrome type VII. The patient’s skin, fascia, and bone collagens all showed an abnormal additional chain, pN-a2(1)’, running slower than the a2(I) chain on electrophoresis. The extension was shown to be on the amino-terminal fragment of pN-a(I)” by cleavage with human collagenase, but pepsin was unable to con- vert pN-a2(1)’ to a2(I). Skin collagen was 4-fold more extractable and contained fewer &dimers and a lower concentration of cross-linking amino acids than control skin collagen. Electron micrographs of both dermis and bone showed markedly irregular ragged outlines of the collagen fibrils in cross-section, although the patient had no clinical signs of bone disease. Procollagen se- creted by her skin fibroblasts in culture showed equal amounts of the normal and abnormal a2(I) chains on pepsin digestion. Before pepsin, the pN-a2(I) compo- nent ran as a doublet on electrophoresis; pepsin re- moved only the normal slower chain. The suspected deletion in pN-a2(1)’ was traced by CNBr peptide anal- ysis to the N-propeptide fragment, which behaved on electrophoresis about 15-20 residues smaller than that from the normal pN-aB(1) chain. The simplest genetic explanation is a spontaneous heterozygote in which one normal and one abnormal allele for the pro-a2(1) gene are expressed, the protein defect being a deletion of the junction domain that spans the N-propeptidase cleavage site and the N-telopeptide cross-linking se- quence.

Defects of collagen structure and expression are rapidly being identified among the heritable disorders of connective tissue (1-3). Osteogenesis imperfecta is revealing a heteroge-

* This work was supported in part by grants from the United States Public Health Service (AM 34075, AM 34076, AM 34078, and HD 18658) and the New England Peabody Home for Crippled Children. Preliminary studies were reported in 1984 at the 30th annual meeting of the American Orthopaedic Research Society in Atlanta, Georgia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

This paper is dedicated to the memory of John F. Aldridge. ll Present address where correspondence and reprint requests

should be sent: Orthopaedic Department, RK10, The University of Washington, School of Medicine, Seattle, WA 98195.

** Deceased.

neous assortment of mutations in the genes for pro-al(1)’ and pro-a2(1) collagen chains that interfere with the normal as- sembly and cellular export of type I procollagen molecules (3, 4). A common problem seems to be a failure to make sufficient extracellular type I collagen for normal tissue strength that for some reason has its most dramatic clinical effects on bone. Inborn collagen defects also lie behind a t least three subtypes of the Ehlers-Danlos syndrome (EDS), a heterogeneous col- lection of disorders that affect the material properties of skin, tendons, ligaments, the vasculature, and other soft connective tissues (2, 5 ) . In EDS IV, several subvariants have emerged that all exhibit a defect in type I11 collagen production. They include a decreased synthesis of pro-al(II1) chains, a failure to export pro-type I11 molecules from the cell, and a structural defect in secreted pro-type I11 molecules (6-8). In EDS V, undercross-linking due to a defective lysyl oxidase enzyme was suggested for some cases (9) but not confirmed in others (10). In EDS VI, an underactivity of lysyl hydroxylase results in varying degrees of hydroxylysine-deficient collagen (11,12) and abnormal cross-linking (13). One variety of EDS VI1 was reported to be based on an underactivity of the collagen N- propeptidase which prevented full cleavage of the amino- terminal propeptides from type I procollagen (14). Thus, pN- al(1) and pN-aB(1) chains could be extracted from the struc- turally abnormal collagen fibrils, as in the fragile-skin disease dermatosparaxis, identified in cattle, sheep, and other animals (1, 15). Later, one of the patients identified with this form of EDS VI1 was restudied and found not to have a propeptidase defect but a structural mutation of the pro-a2(1) chain which prevented natural cleavage of the N-propeptide (16). All the pro-al(1) chains and half the pro-a2(1) chains were processed normally, and a new mutation producing a spontaneous het- erozygote was the simplest genetic explanation. It is not yet clear why deletion mutations near the amino terminus of the a2(1) chain in some patients cause osteogenesis imperfecta (17, 18) while in others they cause Ehlers-Danlos syndrome type VI1 (16).

By screening for collagen defects in surgical biopsies from patients with signs of inborn connective tissue disease, we

The abbreviations used are: pro-al(1) and pro-a2(1), type I pro- collagen chains bearing amino- and carboxyl-terminal extension pep- tides; EDS, Ehlers-Danlos syndrome; EDS I, 11, etc., variants of the syndrome; pN-al(1) and pN-aB(I), partially processed type I procol- lagen chains that have lost their carboxyl-terminal extension pep- tides; al(1) and a2(I), fully processed type I collagen polypeptide chains; pN-a2(1)’, the abnormal pN-aB(1) chain from the patient’s tissues and cell culture medium; N-telopeptide, nontriple-helical do- main of 10-20 residues at the amino terminus of the a chain in fully processed collagen molecules; SDS, sodium dodecyl sulfate; CNBr, cyanogen bromide; CB4, CB4,2, etc., names of CNBr-derivedpeptides and partially cleaved peptides of collagen.

11322

Collagen cu2(I,) Chain Mutation 11323

have identified another case of Ehlers-Danlos syndrome type VI1 with what appears to be a new structural defect of the pro-a2(1) chain. We report the biochemical and ultrastruc- tural features of collagen in the patient's tissues and the nature of the defect in the procollagen chains synthesized by her skin fibroblasts in culture.

EXPERIMENTAL PROCEDURES

The Patient-A 2%-year-old girl, with normal parents and no family history of connective tissue disease, had bilateral hip disloca- tions and overt generalized hyperlaxity of her joints noted a t birth. She had severe flat feet associated with the laxity of ligaments and other connective tissues. Her skin showed no signs of easy bruisability but was noted to be of unusual texture. The condition was typical of previously described cases of EDS VII.

Tissue Preparation-Skin, fascia, and cortical bone were obtained on two occasions 8 months apart during surgery to correct the patient's hip deformities. A sample of skin was stored in liquid NZ for fibroblast culture. Small samples of each tissue were fixed immedi- ately after surgery for electron microscopy. Remaining tissue was washed in 0.15 M NaCl containing the protease inhibitors 2 mM phenylmethanesulfonyl fluoride, 10 mM N-ethylmaleimide, 5 mM benzamidine HC1, 2 mM EDTA at pH 7.2 and frozen a t -20 "C prior to biochemical analyses.

Collagen Extractions-Dermis was dissected free of fat and washed overnight in chloroform/methanol (3:1, v/v), then water. A portion (800 mg, wet weight) was homogenized in 1 M NaCI, 0.05 M Tris/ HCl, pH 7.5, and stirred at 4 "C for 24 h. After centrifugation, insoluble material was extracted for 24 h in 3% acetic acid and the subsequent residue for 24 h in 4 M guanidine HCI, 0.05 M Tris/HCl, pH 7.5, both a t 4 "C. In a second experiment, another portion of dermis was sequentially extracted in 1 M NaCl and then 3% acetic acid only. All extracts were dialyzed exhaustively a t 4 "C against water and then 0.1 M acetic acid and freeze-dried. Another small sample of dermis was extracted directly in 4 M guanidine HCl, 0.05 M Tris/HCl, pH 7.5, a t 4 "C for 24 h. Specimens of young control human dermis (9-year-old male) were processed similarly.

Bone pieces were dissected free of soft tissue, cut to -1 mm', and decalcified with several changes of 0.5 M EDTA, 0.05 M Tris/HCl, pH 7.5, for 8 days a t 4 "C. Small samples of skin, fascia, and decal- cified bone were homogenized in 3% acetic acid and digested with pepsin (l:lO, w/w, of dry tissue) for 24 h at 4 "C. Solubilized material was freeze-dried for electrophoretic analysis. A sample of decalcified bone was also homogenized and extracted directly in 4 M guanidine HCl, 0.05 M Tris/HCl, pH 7.5. The extract was dialyzed against water and freeze-dried for electrophoretic analysis of solubilized collagen.

Fibroblast Cultures-Cells were grown out from the patient's skin biopsy and from that of a 4-year-old male control using standard techniques in Falcon flasks (75 cm') and Dulbecco's minimal essential medium containing 10% fetal calf serum, 100 units/ml penicillin, and 100 pg/ml streptomycin. At confluency, cells were incubated for 24 h in serum-free medium containing 100 pg/ml ascorbic acid, then changed for a further 24 h in fresh serum-free medium containing 100 pg/ml ascorbic acid and 10 pCi/ml [5-3H]proline (New England Nuclear). Medium was removed and mixed with a fresh solution of protease inhibitors to give a concentration of 2 mM phenylmethylsul- fonyl fluoride, 10 mM N-ethylmaleimide, 2 mM benzamidine HC1, and 2 mM EDTA. Half the medium was dialyzed against 3% acetic acid for 24 h a t 4 "C and then treated with pepsin (0.1 mg) for 24 h at 4 "C and freeze-dried. The cell layer was collected with a rubber "policeman," suspended in a solution of protease inhibitors, and stored frozen. For electrophoretic analysis of cell proteins, the cell layer was pelleted, homogenized in SDS electrophoresis buffer, and heated a t 100 "C for 2 min.

Digestion with Pepsin and Human Collagenase-Neutral salt-ex- tracted skin collagen (1 mg/ml) was treated with pepsin (1:lO by dry weight) in 3% acetic acid a t 4 "C, taking portions a t 4 and 16 h for freeze-drying and electrophoretic analysis. Gastric mucosal collagen- ase (19) was the very generous gift of Dr. David Woolley, University Hospital of South Manchester, England. Neutral salt-extracted skin collagen and pepsin-solubilized skin collagen were dissolved at 3 mg/ ml in 0.4 M NaC1, 0.05 M Tris/HCl, 50 mM arginine, 10 mM CaC12, pH 8.0, and half a volume of enzyme in 0.17 M NaCI, 0.05 M Tris/ HC1, 10 mM CaC12, pH 8.0, was added. The mixture was incubated a t 35 "C for 24 h. Aliquots of the digests were mixed with 5% (v/v) of 100 mM 1,lO-phenanthroline, 250 mM EDTA, pH 8.0, to inactivate

the enzyme, then mixed with an equal volume of SDS electrophoresis buffer and run directly.

Electrophoresis of Collagen, Procollagen, and Derived Peptides- Tissue extracts were examined by electrophoresis in SDS-5% poly- acrylamide slab gels by the method of Laemmli (ZO), staining with Coomassie Blue R-250. Procollagen from the fibroblast culture me- dium was also run on 5% slab gels with and without disulfide cleavage by dithiothreitol, and the tritium-labeled proteins were detected by fluorography (21). For analysis of CNBr-derived peptides of the procollagen chains by a modification of the method of Byers et al. (22), samples were run in wide sample slots, and the gel was stained with Coomassie Blue to detect the various pro-a, pN-a, and a chains. Individual chains of pro-al(III), pro-nl(I), pro-a2(I), pN-al(I), and pN-a2(I) and the a l ( I ) , ~N-a2(1)~, and a2(I) chains derived by pepsin treatment of the procollagen were excised from the gel. These gel strips were washed in 50% methanol to remove the Coomassie Blue dye, equilibrated with 70% formic acid, and then 50 mg of CNBr in 1 ml of 70% formic acid was added to each strip and the sealed tube was kept on a finger-shaker for 4 h at 25 "C. After diluting 10-fold with water and rinsing the gel strip, the supernatant was freeze-dried. Each gel strip was rinsed with 0.1 M Tris/HCl, 30% glycerol, pH 6.8, until a t neutral pH, then inserted into a sample well of an SDS-15% polyacrylamide slab gel and the peptides were electrophoresed. The low molecular weight peptides that had preferentially leached out with the formic acid during digestion were run on a separate SDS- 15% polyacrylamide gel. The tritiated peptides were revealed by fluorography (21).

Reverse-phase Column Chromatography-Collagen chains in the salt-soluble extracts of skin (200 pg) were fractionated by elution from a Brownlee RP300 column using an Altex high performance liquid chromatography system and a gradient of acetonitri1e:l-pro- panol (3:1, v/v) in aqueous 0.1% trifluoroacetic acid (23). Fractions (2 ml) were collected, freeze-dried, and analyzed directly by electro- phoresis on a 20-well SDS-5% polyacrylamide slab gel.

Analysis of Collagen Cross-linking Amino Acids-A diced sample of defatted dermis (150 mg, wet weight) was reacted with 10 mg of NaBHl in 50 ml of 0.1 M Na phosphate, pH 7.4, for 1 h a t 25 "C adding an additional 10 mg of NaBH, after 30 min. After acidifying with acetic acid to pH 3, the tissue was washed in water, freeze-dried, then hydrolyzed in 6 M HCl a t 110 "C for 24 h. The dried hydrolysate was eluted from a calibrated Bio-Gel P-2 column (100-200 mesh, 100 X 2.5 cm) in 10% acetic acid, pooling fractions that spanned the elution position of cross-linking amino acids but avoided the bulk of the common amino acids (24, 25). The borohydride-reduced cross- linking amino acids that form on the lysine-aldehyde pathway of cross-linking in skin collagen were resolved and quantified on a Glenco custom amino acid analyzer using a single buffer eluant, 0.35 N Na citrate, pH 5.28, a column a t 65 "C, and ninhydrin detection. Mature cross-linking amino acids on the hydroxylysine-aldehyde cross-linking pathway in fascia and bone collagens were measured directly in acid hydrolysates of the tissues using reverse-phase high performance liquid chromatography and a fluorescence detector (26). Insufficient bone and fascia were available to quantify their borohy- dride-reducible fraction of collagen cross-links. Hydroxyproline, as a measure of collagen, was assayed in each tissue hydrolysate by a colorimetric procedure (27) adapted to the Technicon autoanalyzer 11. The content of cross-linking residues was expressed relative to collagen. The hydrolysate of bone was also assayed for Ca by atomic absorption spectroscopy.

Electron Microscopy-Skin, fascia, and bone were examined by standard methods for transmission electron microscopy. All tissues were fixed for 3 h in Karnovsky's reagent (2.5% glutaraldehyde, 1% paraformaldehyde in 0.1 M cacodylate buffer, pH 7.4). Bone was then decalcified in 7.5% EDTA, 2.5% glutaraldehyde at 4 "C for 2 weeks, changing the solution every 2 days. All tissues were post-fixed in 1% osmium tetroxide and infiltrated and embedded with Spurr's medium within 24 h. Age-matched human control tissues were prepared sim- ilarly. Electron micrographs were taken on a JEOL lOOC transmission electron microscope.

RESULTS

Extractability of Tissue Collagen-About 3-4-fold more col- lagen could be solubilized from the patient's skin than from a control (Table I). An increased solubility was seen for all three sequential extractants: neutral 1 M NaC1, 3% acetic acid, and 4 M guanidine HC1. All extracts showed an abnormal

11324 Collagen (r2(I. Chain Mutation

TABLE I Collagen extractability from the patient's skin compared with

control skin Results of three extraction experiments are reported. For the three-

step (column 1) and the two-step (column 2) sequential extractions a t 4 "C, samples of defatted dermis were homogenized in the initial extractant, 1 M NaCl, 0.05 M Tris/HCl, pH 7.5. Small pieces of dermis (not homogenized) were used for the direct extraction in 4 M guani- dine HC1, 0.05 M Tris/HCl, pH 7.5 (column 3).

1

Proportion of tissue solubilized

Patient Control

(11 (21 (31 (11 (21 (3)

'% of dry weight 1 M NaCl 12.0 11.2 5.2 4.4

4 M guanidine HC1 23.7 19.2 7.2 4.8 Total 45.0 18.8 19.2 13.5 6.5 4.8

3% CH,COOH 9.3 7.6 1.1 2.1

pNa2(&

E DS CON FIG. 1. SDS-polyacrylamide electrophoresis of sequential

extracts of skin collagen for the patient (EDS) and a control (CON). Homogenized defatted dermis was extracted sequentially a t 4 "C for 24 h each in neutral 1 M NaC1, 3% acetic acid and neutral 4 M guanidine HCl. Another sample of dermis was digested with pepsin (1: lO by dry weight) for 24 h at 4 "C. Extracts were clarified by centrifugation, dialyzed against water, and freeze-dried. Lanes 1-4, patient's collagen; lanes 5-8, control skin collagen. Lanes I and 5, 1 M NaCl extract; lunes 2 and 6, 3% acetic acid extract; lunes 3 and 7, 4 M guanidine HC1 extract; lanes 4 and 8, direct pepsin extract of defatted dermis. All samples were run without disulfide cleavage.

collagen a chain on electrophoresis running above a2(I) near the position of pN-a2(I) (Fig. 1) and very similar in mobility to a component reported for a previous but unrelated case of EDS VI1 (16). On reverse-phase column chromatography, the &(I) chain and the new component co-eluted, well resolved from al(1) chains (Fig. 2). This property supported the impression that the new component was an abnormal form of a2(I) and from now on will be referred to as ~ N - a 2 ( 1 ) ~ . Another abnormality in the electrophoretic profiles of the patient's collagen was the marked deficiency of (312 dimer chains in all extracts of the patient's skin compared with the controls (Fig. 1). This appeared to be due to the ~ N - a 2 ( 1 ) ~ chain being unable to form p dimers (PI2 or p22). If present, these ought to have been resolved as bands running just above those of the normal Dl2 and p22 chains. However, the pepsin extract did faintly show such additional ,f3 bands running slightly slower than the normal pl2 and p22 bands (Fig. 1, lane 4, but best seen in the pepsin digest of bone arrowed in Fig. 4). Measurements of the relative mobilities of the normal and abnormal p chains compared with pN-a2(I), a2(I), and al(1) strongly indicated that the abnormal p22 had the com-

5 10 15

1 5 10 15 Fraction Number

FIG. 2. Reverse-phase high performance liquid chromatog- raphy of neutral salt-soluble collagen from the patient's skin. The sample (200 pg) was eluted from a Brownlee RP300 column (C8, 25 X 0.4 cm) with a complex gradient of 24-38% acetonitri1e:l- propanol (3:1, v/v) in aqueous 0.1% trifluoroacetic acid. Fractions (2 ml) were dried and run in individual lanes of an SDS-5% polyacryl- amide slab gel (inset). The two forms of a2(I) were co-eluted from the column well resolved from al(1) chains. The two resolved forms of al(1) (peaking a t fractions 4 and 8) differ in that the former lacks and the latter retains its carboxyl-telopeptide sequence. This heter- ogeneity is a normal finding for skin extracts, which probably reflects the effects of endogenous proteolysis in the tissue in uiuo or during extraction (D. R. Eyre, unpublished data).

position pN-a2(I)S-a2(I), the heterodimer of a normal and an abnormal a2(I) chain, rather than [ ~ N - a 2 ( 1 ) ~ ] ~ . Altogether, these observations indicated that in the patient's dermis the pN-a2(1)' chain was unable to cross-link intramolecularly to al(1) chains to form /3 dimers but could cross-link intermo- lecularly to al(1) or to normal cu2(I) chains.

Treatment of the 1 M NaCl collagen extract of skin with pepsin for up to 16 h a t 4 "C had no effect on the amount of pN-aZ(1)' chain seen on gel electrophoresis (not shown). Human collagenase cleaved the collagen normally, however, and indicated that the additional length of polypeptide on p N - ( ~ 2 ( 1 ) ~ was on the three-quarter amino-terminal fragment (Fig. 3).

Collagen extracted from further specimens of the patient's skin, fascia, and decalcified bone obtained during surgery 8 months later was also examined by electrophoresis (Fig. 4). All three tissues showed a prominent pN-a2(1)' band in the pepsin extract. The extracts of bone contained significantly more ~ N - a 2 ( 1 ) ~ t h a n normal a2(I); this was especially evident for the 4 M guanidine HC1 denaturant extract (Fig. 4, lane 4). This solubility difference between bone and the soft tissues probably results from basic differences in the chemistry and molecular sites of their collagen cross-links (28), rather than variability in the relative contents of pN-a2(1)' and ot2(I) chains in the different tissues. The abnormal 0-components that were especially prominent in the pepsin extract of bone (Fig. 4, lune 3) are probably the products of intermolecular cross-linking in the tissue. These could form, for example, by the reaction of a hydroxylysine-aldehyde on a telopeptide of an al(1) or a normal a2(I) chain with a hydroxylysine a t a helical cross-linking site in pN-a2(1)' (28).

Collagen a2(I) Chain Mutation 11325

1 2 3 4 -

C E FIG. 3. SDS-polyacrylamide electrophoresis of the cleavage

products of the patient's collagen with human collagenase. A 1 M NaCl extract of skin collagen was treated with gastric mucosal collagenase for 24 h a t 35 "C (lane E) . An aliquot was mixed with 5% (v/v) of 100 mM 1,lO-phenanthroline, 250 mM EDTA, pH 8, then an equal volume of SDS-sample buffer and run on a 7.5% slab gel. Substrate incubated in buffer alone was run as the control lane (lane C). The pN-t~2(1)~ chain was fully cleaved to a ?4 pN-a2(1)* fragment that retained the extension sequence (band indicated but unlabeled in lane E ) and a normal 1h a2(I)B fragment that presumably co- migrated with the ~ r 2 ( 1 ) ~ from the normal a2(I) chain.

Collagen Cross-linking-To assess directly if the degree of intermolecular cross-linking was affected in the tissue colla- gens as indicated by the increased collagen solubility and paucity of extracted p chains, skin, fascia and bone samples were assayed for collagen-specific cross-linking amino acids. The results (Table 11) showed a quarter of the control content of histidinohydroxymerodesmosine, the major residue in skin collagen on borohydride reduction, but a similar content to controls of the mature cross-links, hydroxylysyl pyridinoline and lysyl pyridinoline, in fascia and bone collagens.

Procollagen Synthesis by Skin Fibroblasts-Skin fibroblasts grew normally and secreted into the medium a similar amount of procollagen to control human skin fibroblasts. Insignificant amounts of procollagen chains were detected on electropho- resis of the disrupted cell layer, indicating no undue accu- mulation of procollagen in the cells (not shown). A normal 2 to 1 ratio of pro-d(1) to pro-a2(1) chains was evident on electrophoresis and fluorography after disulfide bond cleavage of the medium proteins (Fig. 5 ) . The partially processed chains pN-al(1) and pN-a2(1) were also evident both before and after disulfide bond cleavage (Fig. 5 ) . Unlike control patterns, however, the pN-a2(I) band ran as a doublet con- sisting of normal pN-a2(1) and a slightly faster component having a mobility indicating that it was about 20 residues shorter. This latter, abnormal pN-a2(1), was not converted to & ? ( I ) by pepsin digestion, and its mobility relative to or2(I) in the cell medium was indistinguishable from that of the pN- a2(1)' chain relative to a2(I) in the tissue extracts. Pepsin treatment of the medium proteins completely removed the

FIG. 4. SDS-polyacrylamide electrophoresis of collagens extracted by pepsin from the patient's skin (lane I ) , fascia (lane 2), and decalcified bone (lane 3). The results are consistent with all three tissues containing equal amounts of abnormal pN- a2(1)' and a2(I) chains. The denaturant (4 M guanidine HC1) extract of decalcified bone is selectively enriched in the ~ N - a 2 ( 1 ) ~ chain over the normal a2(I) chain (lune 4 ) . The additional band seen just above pN-a2(1)' in lane 3 is a pepsin overdegradation product of al(1) often present in pepsin digests of normal bone collagen. The arrowed p components running just above ,811 and pl2 derive from the pN- a2(1)' chain (see text).

slower normal pN-a2(1) chain and converted it to a2(I) as expected. The pro-a2(I) chain from the patient's cell medium also seemed to be an abnormally broad band on electropho- resis. However, since pN-al(1) was poorly resolved from it further characterizing analyses were directed at pN-aB(1).

The various proa chains, pNa chains, and pepsin-derived products were characterized by excising them as stained bands from the gel, digesting the protein with CNBr for 4 h, and running the resulting peptides in a polyacrylamide gel of higher concentration. The results confirmed the identities of the pro-al(II1) and pro-d(1) bands (not shown) and revealed differences between pN-a2(1)' and normal pN-a2(1) (Fig. 6). The peptides produced by CNBr were run as two fractions: (a) large peptides that remained in the gel strip on digestion (Fig. 6a), and (b) small peptides that diffused from the gel strip during CNBr digestion and subsequent washing (Fig. 6b). The higher molecular weight fraction from ~ N - a 2 ( 1 ) ~ included prominent peptide bands running above a2(I)CB4 not given by the normal 012(I) chain (Fig. 6a). These were identified by their relative mobilities and the fact that they disappeared on prolonged digestion with CNBr as the partial cleavage products CBpN1,4,2, CBpN1,4, and CB4,2. No ab- normalities were seen in the mobility of peptides CB4 or

11326 Collagen a2(1) Chain Mutation

CB3,5 derived from pN-a2(1)' chains isolated from fibroblast culture medium (Fig. 6 , a and b) or directly from the patient's skin collagen (not shown).

The fraction of CNBr peptides of low molecular weight from pN-~t2(1)~ revealed a band having a mobility that indi- cated a size of 50-60 amino acid residues (Fig. 6b). This had to be the N-propeptide, since the fully processed human a2(I) chain produces no peptides of this size (Fig. 6b and see chain map in Fig. 8). The same peptide was evident in the CNBr digest of the combined pN-a2(1) and ~ N - a 2 ( 1 ) ~ chains (i.e. doublet band in Fig. 5, lane 2) together with a more intense peptide band of slightly slower mobility that was assumed to be the N-propeptide derived from the normal pN-a2(I) chain (Fig. 6b, lane 4) . Semilogarithmic plots of mobility versus molecular size indicated a difference of about 15 residues

TABLE I1 Concentration of cross-linking amino ucids in collagen of the patient's

connective tissues compared with controls Defatted dermis was treated with sodium borohydride for 1 h in

0.1 M Na phosphate, pH 7.5. After acid hydrolysis, the borohydride- reduced collagen cross-linking residues were partially purified by molecular sieve chromatography (24,25) then quantified by ninhydrin reaction on the amino acid analyzer. The mature cross-linking resi- dues in fascia and decalcified bone were assayed directly on acid hydrolysates by a procedure involving high performance liquid chro- matography and fluorescence detection (24). fSD(n) where indicated. A zero means not detected; a dash means not measured. The control fascia value is from young adults, and the control bone value is the mean for a 5-month-old, 4-year-old, and 6-year-old. HHMD, histidi- nohydroxymerodesmosine; HLN, hydroxylysinonorleucine; HP, hy- droxylysyl pyridinoline; LP, lysyl pyridinoline.

Skin Fascia Bone

mollmol of collagen Patient HHMD" 0.05

HLN" 0.01 HP 0 0.30 0.22 LP 0 0 0.05

Control HHMD" 0.20 - - HLN" 0 HP 0 0.34 f 0.04 (2) 0.15 k 0.02 (3)

- - - -

- -

LP 0 0 0.04 f 0.01 (3) Leucine equivalents by ninhydrin reaction.

1 3

proa2( I ) -

+dtt -dtt

E D S

3

+pepsin

between the two peptides, pN-a2(I)TBl and normal pN- aP(1)CBl.

The same abnormal peptide, pN-a2(I)'CBl, was subse- quently isolated directly from a CNBr digest of the patient's skin collagen, but not from control skin, and its identity was confirmed by analysis of its amino acid composition and tryptic peptides (not shown).

Electron Microscopy of Tissue Collagens-Electron micros- copy showed a grossly abnormal organization of collagen fibrils in the patient's dermis (Fig. 7 , a and b). Individual fibrils had irregular ragged outlines in cross-section compared with their solid uniformity in normal skin; the fibrils also appeared to be more loosely and randomly organized into fibers. The average diameter was only slightly less at 90 nm with a range of 55-110 nm (measured on 100 fibrils), compared with 110 nm (range 95-125 nm) for age-matched control skin. A similar uneven outline of individual collagen fibrils was also evident throughout sections of fascia (not shown) and decal- cified bone (Fig. 7c) from the patient. In addition, bone collagen showed what appeared to be a lateral disruption of individual fibrils into subfilaments, most noticeable in areas that were judged by their distance from osteoblasts to have been mineralized in vivo (Fig. 7c) . This gave the impression that mineral deposition may have mechanically disrupted the lateral packing of collagen molecules within fibrils.

Bone Mineralization-The collagen and calcium contents of the patient's bone were within the normal range, with a Ca/collagen ratio by weight of 1.08. Thus, no major deficiency in mineralization of the bone collagen was evident despite its abnormal molecular structure and ultrastructure.

DISCUSSION

The results are consistent with a structural mutation in one allele of the gene that codes for the pro-a2(1) chain of type I collagen, causing half the expressed protein molecules to be structurally abnormal. In summary, and as will be reasoned below, the various findings indicate that the protein defect is a deletion of a 15-20-amino acid sequence that represents the N-telopeptide junction domain in the pro-a2(1) chain (Fig. 8). The results also suggest that a key consequence of the basic biochemical lesion in producing the clinical symp-

4 5 6

,proal(lll) -Lproal(l)

- pNal(l)

- al(l)

- a2(1) pNa2(1)

tdtt -dtt +pepsin

CON FIG. 5. SDS-polyacrylamide electrophoresis and fluorography of radiolabeled procollagen and par-

tially processed PC- and pN-collagens in the medium of fibroblast cultures. Cells were incubated with [5- 3H]proline for 24 h. Medium proteins were divided into two fractions; one half was run with (lunes I and 4 ) and without (lunes 2 and 5 ) disulfide cleavage by dithiothreitol; the other half was treated with pepsin at pH 3 for 24 h at 4 "C before running (lunes 3 and 6 ) . A 5% polyacrylamide slab gel was used. Patient cells, lunes 1-3; control cells, lunes 4-6. The medium proteins unreduced (lunes 2 and 5) were run at a higher load than the reduced samples (lunes I and 4 ) to bring out the detail in resolution of the pNo chains. Two forms of pN-a2(1) were resolved as a tight doublet of bands from the patient's medium proteins.

Collagen a2(I) Chain Mutation

Q 1 2

- 3,s

J- PN 1,492 7- - PN1,4 8- 6- 3-

b 1 2

11327

3 4 ~~ .

~ - 4 ~ 2 - 4

-pN 1 -pN 1’

FIG. 6. SDS-polyacrylamide electrophoresis and fluorography of CNBr peptides derived from the individual pNa chains and LY chains isolated from the culture medium of the patient’s fibroblasts. The collagen chains were resolved by electrophoresis (Fig. 5), excised as bands from the gel, and treated in the gel strips with CNBr (see “Experimental Procedures”). Each digest was run on a 15% slab gel as two fractions: a, large peptides remaining in the gel strip; 6, small peptides that washed out from the strip during digestion. a; lane I , a2(I) normal; lane 2, pN-a2(1)‘. b; lane I , al(1); lane 2, a2(I); lane 3, ~ N - a 2 ( 1 ) ~ ; lane 4 , combined ~ N - a 2 ( 1 ) ~ and pN-a2(I) normal.

toms is a disruption of normal collagen cross-linking. Several findings indicate an undercross-linking of the pa-

tient’s skin collagen. The increased extractability, lower than normal proportion of @ chains, and decreased concentration of borohydride-reducible cross-linking residues provide direct evidence. The pN-a2(1)’ chains which accumulate in the patient’s connective tissues appear unable to form intramo- lecular @-dimers, suggesting that the N-telopeptide aldehyde responsible for the aldol cross-links in such dimers (29) is missing from pN-(~2(1)~ or, a t least, is unable to react to form cross-links. The consequences of a missing NH-terminal al- dehyde may be particularly severe for the a2(I) chain, since, unlike al(I ) , it normally lacks also a COOH-terminal alde- hyde-forming lysine (30).

In collagens of normal bone and fascia the hydroxylysine- aldehyde pathway of cross-linking predominates, and the intramolecular aldol cross-links formed between two lysine aldehydes that are prominent in skin collagen are sparse or absent from bone (28, 31). The finding of a normal content of 3-hydroxypyridinium cross-linking residues in the patient’s bone and fascia collagens but evidence of a decreased level of cross-linking in skin collagen presumably reflects this tissue- specific difference in cross-linking chemistry. Although the content of cross-linking residues seemed normal in bone and fascia, their distribution among intermolecular sites in fibrils may not be normal. This is suggested by the disproportion- ately high extractability of the ~ N - a 2 ( 1 ) ~ chain in bone, the impression from electrophoresis of the denaturant extract that the bone collagen was more soluble than normal, and the disordered structure of fibrils in fascia and bone seen on electron microscopy. Nevertheless, the results suggest that cross-linking on the lysyl aldehyde pathway, which predomi- nates in skin and also occurs in many other soft connective tissues, is more affected than cross-linking on the hydroxy- lysyl aldehyde pathway (28).

The failure of pepsin to convert ~ N - a 2 ( 1 ) ~ into a2(I) chains

in the native collagen suggests that the pepsin-susceptible sites, normally in the N-telopeptide region, are missing, or completely inaccessible. Likewise, the nonconversion of pN- a2(1)’ into a2(I) naturally in the patient’s tissues by the N- propeptidase, despite complete processing of the al(1) chains, suggests that the sequence containing the N-propeptidase cleavage site (32, 33) is missing or cannot be cleaved due to an altered conformation.

The electrophoretic properties of the pN-collagen chains made by skin fibroblasts in culture, and of their CNBr-derived peptides, indicate that the ~ N - a 2 ( 1 ) ~ chain is 10-20 amino acid residues shorter than the normal pN-a2(1) chain, with the defect in the N-propeptide/telopeptide region. The results suggest a deletion mutation in 50% of the p r o d ( 1 ) chains. The most probable site that accounts best for all the biochem- ical findings is a deletion that spans the N-telopeptide se- quence and its adjoining N-propeptidase cleavage site (Fig. 8). In the chicken pro-a2(1) gene, one exon codes for this 19- amino acid domain (33). Based on the chicken gene nucleotide sequence and the properties of the ovine pN-a2(I) chain (34), CNBr should release as one fragment the intact N-propeptide together with the N-telopeptide sequence up to the first methionine which occurs at residue 3 of the main triple- helical domain. We interpret the abnormal abundance of the partial cleavage product identified as CBpN1,4 in the CNBr digest of ~ N - a 2 ( 1 ) ~ (Fig. 6a) as evidence for only one methi- onine, not two, being present between the N-propeptide and CB4 sequences (Fig. 8). In the normal &(I) chain (chick and human) two methionines occur close together at the start of the main helix, defining the tripeptide, a2(I)CBO (Fig. 8), and greatly reducing the probability of seeing a partial-cleav- age product containing the N-propeptide and CB4 in a CNBr digest of the normal chain. In the chick pro-a2(1) gene, one exon codes the domain containing the N-propeptide cleavage site and the N-telopeptide sequence. It ends with the codon

11328 Collagen a2(I,, Chain Mutation

FIG. 7. Transmission electron micrographs of collagen fi- brils in skin and bone from the patient (EDS) and a control

I- n ‘ I I

I

N - pept idase 0 4 ”& I

he1 ical N-telo- domain peptide

FIG. 8. Diagram of the human a2(1) chain showing the dis- tribution of methionine residues (vertical bars) and the num- bering scheme of resulting CNBr peptides. The suspected dele- tion in pN-a2(1)’ encompasses the N-propeptidase cleavage site (solid arrow), the cross-linking lysine (double arrow), the pepsin-sensitive site (open arrow), and the first methionine at the junction of CB1 and CBO.

for the first of the two methionines that define peptide CBO (33).

The electrophoretic analyses of collagen extracted from all three tissues, skin, fascia, and bone, were consistent with 50% of the expressed a2(I) chains in each tissue having the struc- tural defect. This supports the conclusion that there is only one copy of the pro-a2(1) gene in the human genome (35) and that both of its alleles are equally expressed in skin, bone, fascia, and presumably in all other connective tissues that make type I collagen.

This patient had clinical symptoms typical of EDS VI1 and no signs of bone disease. She is the third unrelated patient with symptoms of EDS VI1 to be identified with a structural defect near the amino terminus of the a2(I) collagen chain. All three cases showed an abnormal a2(I) chain running with a similar slow mobility on electrophoresis of their extracted tissue collagens (16, 36). In the most recent case, an abnor- mality in the a2(I)CB4 peptide was proposed (36). I t is puzzling why some deletions near the amino terminus of a2(I) cause osteogenesis imperfecta (17, 18), whereas in the latter (16, 36) and present cases, EDS VI1 results. The critical distinction may lie in the specific functional domain(s) that are deleted or affected. Thus, in the present case of EDS VII, and perhaps in the two others described by Steinmann et al. (16, 36), a specific deletion of the telopeptide cross-linking site in a2(I) may be the key factor. When deletions are further into the a2(I) chain, even though still close to the amino terminus (17, 18), the overriding biochemical effect might become an underproduction of extracellular collagen with resulting clinical symptoms of osteogenesis imperfecta (1). The present patient had no bone fractures or other clinical signs of osteogenesis imperfecta, despite her bone type I collagen being biochemically and ultrastructurally abnormal. The calcium content of the bone, however, suggested a normal degree of mineralization of the collagen. It would appear that deposition of sufficient bone collagen, despite the abnormal

(C). a and b (dermis), the patient’s fibrils in cross-section appear irregular and ragged in outline. a, bar = 290 nm; b, bar = 140 nm. c (bone), the patient’s bone fibrils (EDS) also appear more irregular in outline that control bone fibrils (C), and in the region a t the left of the micrograph (EDS), judged to have been mineralized in uiuo, the patient’s fibrils show an unusual appearance suggesting a lateral disruption into subfilaments. Bar = 290 nm.

Collagen a2(I) Chain Mutation 11329

molecular structure and evidence of disordered molecular packing in fibrils, was adequate for normal bone function. In manifested osteogenesis imperfecta cases, on the other hand, the critical factor underlying the clinical symptoms may be an interference with procollagen assembly that decreases the cellular output of collagen.

Acknowledgments-We are indebted to Dr. Michael Millis of Chil- dren's Hospital for supplying tissue biopsies from this patient at surgery and for his continued interest. We are most grateful to Dr. David Woolley, University Hospital of South Manchester, England for his gift of purified human gastric mucosal collagenase. We also thank Linda Lewi, Lena Ellezian, Michael Burley, and Mary Ann Hsu for excellent technical assistance, Irene Grubliauskas for the Ca assay, and Catherine Marsland for preparing t,he manuscript.

REFERENCES

1. Prockop, D. J., and Kivirikko, K. I. (1984) N. Engl. J. Med. 311, 376-386

2, Krane, S. M. (1984) in Extracellular Matrix Biochemistry (Piez, K. A,, and Reddi, A. H., eds) pp. 413-463, Elsevier, New York

3. Byers, P. H., and Bonadio, J. F. (1985) Butterworths International Medical Reuiews, Pediatrics Vol. 5: Genetic and Metabolic Disease (Scriver, C. R., and Lloyd, J., eds) Butterworths, Lon- don, in press

4. Prockop, D. J. (1984) Am. J. Hum. Genet. 36, 499-505 5. Hollister, D. W., Byers, P. H., and Holbrook, K. A. (1982) Adu.

Hum. Genet. 12, 1-87 6. Pope, F. M., Martin, G. R., Lichtenstein, J. R., Penttinen, R.,

Gerson, B., Rowe, D. W., and McKusick, V. A. (1975) Proc. Natl. Acad. Sci. U. S. A. 72, 1314-1316

7. Byers, P. H., Holbrook, K. A,, Barsh, G. S., Smith, L. T., and Bornstein, P. (1981) Lab. Inuest. 44, 336-341

8. Stolle. C. A,. Mvers, J. C.. and Pveritz, R. E. (1983) Am. J. Hum. Genet. 35; 5 4 ~

9. DiFerrante. N.. Leachman. R. D.. Aneelini. P.. Donnellv. P. V.. Francis, G., and Almazan, A. (197;) Connect. Tissue" Res. 3; 49-53

10. Siegel, R. C., Black, C. M., and Bailey, A. J. (1979) Biochem. Biophys. Res. Commun. 88,281-287

11. Pinnell, S. R., Krane, S. M., Kenzora, J. E., and Glimcher, M. J. (1972) N. Engl. J . Med. 286, 1013-1020

12. Krane, S. M., Pinnell, S. R., and Erbe, R. W. (1972) Proc. Natl.

Acad. Sci. U. S. A. 69, 2899-2903 13. Eyre, D. R., and Glimcher, M. J. (1972) Proc. Natl. Acad. Sci.

U. S. A. 69, 2594-2598 14. Lichtenstein, J. R., Martin, G. R., Kohn, L. D., Byers, P. H., and

McKusick, V. A. (1973) Science 182, 298-300 15. Lenaers, A., Ansaz, M., Nusgens, B. V., and Lapiere, C. M. (1971)

Eur. J. Biochem. 23,533-543 16. Steinmann, B., Tuderman, L., Peltonen, L., Martin, G. R.,

McCusick, V. A., and Prockop, D. J. (1980) J. Biol. Chem. 255,

17. Byers, P. H., Shapiro, J. R., Rowe, D. W., David, K. E., and

18. Sippola, M., Kaffe, S., and Prockop, D. J. (1984) J. Biol. Chem.

19. Woolley, D. E., Tucker, J. S., Green, G., and Evanson, J. M.

20. Laemmli, U. K. (1970) Nature 227, 680-685 21. Laskey, R. A., and Mills, A. D. (1975) Eur. J. Biochem. 56,335-

341 22. Byers, P. H., Siegel, R. C., Peterson, K. E., Rowe, D. W., Hol-

brook, K. A., Smith, L. T., Chang, Y-H., and Fu, J. C. C. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 7745-7749

8887-8893

Holbrook, K. A. (1983) J. Clin. Invest. 71, 689-697

259, 14094-14100

(1976) Biochem. J. 153, 119-126

23. Eyre, D. R., and Wu, J. J. (1983) FEBS Lett. 158, 265-270 24. Eyre, D. R., and Oguchi, H. (1980) Biochem. Biophys. Res. Com-

25. Eyre, D. R. (1981) Deu. Biochm. 22, 51-55 26. Eyre, D. R., Koob, T. J., and Van Ness, K. P. (1984) Anal.

27. Stegemann, H. (1958) 2. Physiol. Chem. 311, 41 28. Eyre, D. R., Paz, M. A., and Gallop, P. M. (1984) Annu. Rev.

29. Bornstein, P., and Piez, K. A. (1966) Biochemistry 5, 3460-3473 30. Bernard, M. P., Myers, J. C., Chu, M.-L., Ramirez, F., Eikenberry,

31. Evre. D. R., and Glimcher, M. J. (1973) Biochim. Bio~hvs. Acta

mun. 92,403-410

Biochem. 137,380-388

Biochem. 53, 717-748

E. F., and Prockop, D. J. (1983) Biochemistry 22,1139-1145

"295, 301-307 . "

32. Horlein. D.. Fietzek. P. P.. Wachter. E.. Laniere. C. M.. and Kuhn; K. '(1979) Eur. J. Biochem. 99, 31-38' ~ '

Nucleic Acids Res. 11, 91-104

200

(1982) J. Biol. Chem. 257, 13816-13822

40,652

33. Tate, V. E., Finer, M. J., Boedtker, H., and Doty, P. (1983)

34. Becker, V., Helle, O., and Timpl, R. (1977) FEBS Lett. 73, 197-

35. Dalgleish, R., Trapnell, B. C., Crystal, R. G., and Tolstoshev, P.

36. Steinmann, B., Rao, V. H., and Gitzelmann, R. (1984) Experientia