IGC

ww

Archives of Biochemistry and BiophysicsVol. 384, No. 1, December 1, pp. 37–46, 2000doi:10.1006/abbi.2000.2099, available online at http://www.idealibrary.com on

Sequence of Normal Canine COL1A1 cDNA1 anddentification of a Heterozygous a1(I) Collagen

ly208AIa Mutation in a Severe Case ofanine Osteogenesis Imperfecta

Bonnie G. Campbell,*,2 Joyce A. M. Wootton,* James N. MacLeod,*,† and Ronald R. Minor*,3

*Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853; and†James A. Baker Institute for Animal Health, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853

Received February 1, 2000, and in revised form August 1, 2000

ma

The sequence of canine COL1A1 cDNA was deter-

mined from four overlapping COL1A1 RT-PCR productsgenerated from canine fibroblast RNA. In the translatedregion, nucleotide identity between canine and humanCOL1A1 cDNA was 93.2%, although the canine sequencelacked nucleotides 204 to 215 in the region coding for theN-propeptide. Amino acid identity was 97.7%. Total RNAand type I collagen were collected from cultured skinfibroblasts of a 12-week-old male golden retriever withpathologic fractures suggestive of osteogenesis imper-fecta (OI) and dentinogenesis imperfecta. Sequential,overlapping ;1000-bp fragments of COL1A1 andCOL1A2 cDNA were each amplified by RT-PCR usingprimers containing 5* T7 polymerase sites. These PCRproducts were transcribed with T7 RNA polymerase, hy-bridized into RNA duplexes, and cleaved at mismatchsites with RNase. The proband had an unique cleavagepattern for the fragment of COL1A1 mRNA spanningnucleotides 709 to 1531. Sequence analysis identified a Gto C point mutation for nucleotide 1276, predicting acodon change from glycine (GGA) to alanine (GCA) foramino acid 208. This change disrupts the normal Gly-X-Ypattern of the collagen triple helix. Restriction en-zyme digestion of the RT-PCR product was consistentwith a heterozygous COL1A1 mutation. Type I colla-gen was labeled with 3H-proline, salt precipitated, andanalyzed by SDS–PAGE. Pepsin digested a chains

ere over-hydroxylated, and procollagen processingas delayed. Thus, canine and human OI appear ho-

1 GenBank Accession No. AF153062.2 Current address: Washington State University, Department of

Veterinary Clinical Sciences, Pullman, WA 99164.3

To whom correspondence and reprint requests should be ad-

dressed. Fax: (607) 253-3317. E-mail: rrm1@cornell.edu.

0003-9861/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

ologous in terms of clinical presentation, etiology,nd pathogenesis. © 2000 Academic Press

Key Words: canine; osteogenesis imperfecta; type Icollagen; COL1A1 mutation.

Most cases of osteogenesis imperfecta (OI) arecaused by a mutation in one of the two genes encodingtype I collagen. The OI phenotype ranges from mildbone fragility to perinatal death due to innumerablefractures and severe deformity. Despite the identifica-tion of over 200 mutations in the COL1A1 or COL1A2genes of human OI patients (1, 2), there is still muchthat is not understood about the relationship betweengenotype and disease severity. Comparisons of the pri-mary COL1 structure between human and other spe-cies (bovine, canine, chicken, and murine) reveal a highdegree of amino acid homology (3–8). The highly con-served nature of type I collagen arises from the precisestructural specifications required by the self-assemblyprocesses of collagen triple helix formation and fibril-logenesis (9). Preservation of type I collagen structureamong different species suggests that animals withCOL1 mutations can serve as models of human OI.

Murine models used in the study of OI include theoim mouse, which carries a naturally occurring reces-sive COL1A2 frameshift mutation (10, 11), and theMov13 mouse, in which a viral insertion prevents tran-scription of COL1A1 (12). However, neither of thesemice simulate the naturally occurring, usually domi-nant, point mutations in COL1 glycine codons thatcause the majority of human OI cases.

The proband of the current study was a 3-month-old

golden retriever puppy with dentinogenesis imperfecta

37

l

P

SASASAS3

cDRo

38 CAMPBELL ET AL.

and multiple pathologic fractures. Type I collagen har-vested from cultured skin fibroblasts of this dog wasposttranslationally over-modified, and processing ofprocollagen to collagen was delayed (13). It was hy-pothesized that a spontaneous COL1A1 or COL1A2mutation was the cause of the clinical and proteinabnormalities observed in this proband. To test thehypothesis, two objectives were addressed. First, thesequences of normal canine COL1A1 cDNA (reportedhere) and COL1A2 cDNA (8) were determined. Second,this sequence information was used to design primersfor a nonisotopic RNase cleavage assay which identi-fied the location of a COL1 mutation in the proband.This report is the first to identify a spontaneous COL1point mutation associated with clinical signs of osteo-genesis imperfecta in a nonhuman species. Dogs with aspontaneous mutation in a type I collagen gene are anew resource for studies of genotype/phenotype rela-tionships in OI, as well as for studies of structure/function relationships in type I collagen.

MATERIALS AND METHODS

The proband (dog CU1) was a 12-week-old male golden retrieverpuppy of small stature that had multiple fractures in various stagesof healing affecting the ribs and nearly every long bone (13). Denti-nogenesis imperfecta was also present (13).

RT-PCR and sequencing of canine COL1A1 cDNA. Total RNA,harvested from confluent skin fibroblasts from a clinically normalmixed breed dog and from the proband, was reverse transcribed tocDNA as previously described (8). Primers for PCR amplificationwere derived from the published human COL1A1 sequence (EMBLdatabase Accession No. Z74615) or from the canine COL1A1 se-quence (as determined during this study). Primer synthesis andautomated fluorescent dideoxy chain-termination sequencing wereperformed by Cornell Biotechnology Service (Cornell University,Ithaca, NY). 39 RACE (rapid amplification of cDNA ends, GIBCOBRL, Rockville, MD) was used to amplify the 39 end of dog COL1A1cDNA as previously described (14). Multiple attempts at 59 RACEwere unsuccessful.

Four overlapping PCR fragments were used to cover the entire

TA

Primer Pairs Used in Polymerase Chain Re

rimeraMatch human

COL1A1 cDNA?Match canine

COL1A1 cDNA?

1 11043 2 1799 2 12113 1 11766 1 12335 1 12042 1 19AUAPb

a S, sense primer; A, antisense primer; number refers to COL1A1b Designed for 39 rapid amplification of cDNA ends (GIBCO BRL,

translated region of canine COL1A1 cDNA (Table I). These PCRfragments were gel-purified (QIAquick Gel Extraction Kit, QIAGEN,

Valencia, CA), and both the sense and the antisense strands weresequenced.

Nonisotopic RNase cleavage assay (NIRCA) and mutation analy-sis. Total RNA was reverse transcribed to cDNA using primer 39 AP(GIBCO BRL). The translated region of canine COL1A1 cDNA wasamplified in two overlapping fragments using PCR primer pair S1(59-AGCAGACGGAGTTTCTC-39) and A2888 (59-ACCGACTTCAC-CGGGACGTC-39) and primer pair S2042 (59-ATTCCAGGGTCTC-CCT-39) and 39AUAP (GIBCO BRL). The translated region of canineCOL1A2 cDNA was also amplified in two overlapping fragmentsusing PCR primer pair S20 (59-GGTTCAGCTAAGTTGGAGGTACT-39) and A2999 (59CCAATGTTGCCAGGGTAAC 39) and primer pairS2530 (59-GGACCCTCTGGTATCACTG-39) and 39AUAP. NestedPCR was then performed using six overlapping primer pairs forcanine COL1A1 cDNA (Table II) and five overlapping primer pairsfor canine COL1A2 cDNA (Table III), with each primer containing aT7-RNA polymerase site on the 59 end. NIRCA was performed usinga commercial kit (MutationScreener, Ambion, Inc., Austin, TX).Briefly, the nested PCR products were transcribed to RNA withT7-RNA polymerase, hybridized to form RNA duplexes, cleaved atmismatched sites with RNase, resolved by gel electrophoresis on a2.5% high resolution agarose gel, and visualized with ethidium bro-mide staining. In cases where the NIRCA was positive (as indicatedby the presence of RNase cleavage products), the nested PCR frag-ment used for that assay was gel-purified and sequenced.

After NIRCA and preliminary sequencing identified a mutation inthe region of COL1A1 cDNA amplified by the primer pair T7-S704and T7-A1531, a smaller PCR fragment containing the putativemutated region was amplified with sense primer S1192 (59-AAG-GAGCCCGTGGCTCTGAA-39) and antisense primer A1607 (59-AGGGAAACCACGGCTACCA-39). This {S11921A1607} fragmentwas digested with AciI (New England BioLabs, Inc., Beverly, MA),according to manufacturer’s directions, and analyzed with ethidiumbromide staining after polyacrylamide gel electrophoresis.

Protein analyses. Postconfluent cultures of skin fibroblasts wereused in pulse chase experiments to analyze procollagen processing asdescribed previously (13, 15). Briefly, at 4 days postconfluence der-mal fibroblasts were fed Dulbecco’s modified Eagle’s media (GIBCOBRL No. 3160-034, Grand Island, NY) containing 20% (v/v) FBS and25 mg/ml ascorbate for 2 days, incubated 1 h in medium with 1% FBS,abeled for 1 h with 3H-proline (50 mCi/ml), and chased for 0, 1, 2, 4,

or 8 h without label. For pepsin digestion, cells were labeled for 4 hwith 3H-proline (20 mCi/ml). The culture medium was collected anda protease inhibitor mix was added to give a final concentration of 30mM EDTA, 10 mM N-ethylmaleimide, 1 mM benzamidine, 1 mM

I

ion Amplification of Canine COL1A1 cDNA

Nucleotideposition Primer sequence (59 to 39)

1–18 AGCAGACGGAGTTTCTC1058–1043 AGGGCGACCTCTTTCA799–815 GCAAGAACGGAGATGAC

2113–2098 CCAGGAACACCCTGTT1766–1780 GATGGCAAAACTGG2335–2321 GGAAGACCAGCTGCA2042–2057 ATTCCAGGGTCTCCCT

GGCCACGCGTCGACTAGTAC

NA nucleotide with which 59 end of primer aligns.ckville, MD).

BLE

act

phenylmethylsulfonyl fluoride; final pH was 7.4. Proteins were pre-cipitated with 30% saturated (NH4)2SO4 (176 mg/ml culture me-

9

AT

ATC

AG

c

TCC

c

dnrr

39COL1A1 MUTATION IN CANINE OSTEOGENESIS IMPERFECTA

dium) with stirring overnight at 4°C. The precipitate was centrifugedat 4°C for 1 h at 43,000g. The pellet was solubilized in 300 mlextraction buffer (0.2 M NaCl, 0.05 M Tris, 1 mM EDTA; pH 7.4) bystirring 30 min at 4°C and then stored at 220°C. The cell layer waswashed thrice with 13 Dulbecco’s phosphate-buffered saline (D-PBS;pH 7.1) (GIBCO BRL). Cells were scraped into 320 ml homogenizingbuffer (0.05 M Tris, 0.2 M NaCl, 30 mM EDTA, protease inhibitor(see above), 0.1% Triton-X-100; pH 7.4) and homogenized on ice (15).For analysis of procollagens, each sample was heated for 3 min at100°C in 2% (w/v) sodium-dodecyl-sulfate (SDS) and sample buffer(0.0625M Tris–HCl (pH 6.8), 2% SDS, 10% glycerol, 5% 2-mercapto-ethanol, 0.001% bromophenol blue) (16), dialyzed overnight againstsample buffer alone, separated on 5.5% SDS–polyacrylamide gelelectrophoresis (SDS–PAGE), and analyzed by autoradiography. Forpepsin digestion, an equal volume of ice cold 1 N acetic acid wasadded to the cell layer homogenate and the solubilized precipitate ofmedium and the pH was adjusted to 2.6 with HCl. Pepsin digestionand analysis of collagens by SDS–PAGE were as described previ-ously (8).

SDS–PAGE was used to compare collagens extracted from skinwith collagens produced by cultured skin fibroblasts. Frozen groundskin of a 1-year-old clinically normal mixed breed dog and probandskin were extracted with 0.5 M acetic acid, and the residue wasdigested with pepsin as described previously (8). The solubilized

TABLE II

Canine COL1A1 cDNA Primer Pairs (Containing T7-RNAPolymerase Sites) Used in the Nonisotopic RNase

Cleavage Assay

Primerpairsa 59 to 39 Primer sequenceb

T7-S84 TAATACGACTCACTATAGGACCAGCCGCAAAGAGTCTACT7-A821 TAATACGACTCACTATAGGGTTCACCGTCATCTCCGTTC

T7-S704 TAATACGACTCACTATAGGACCTGGTCCTCAAGGTTTCCAT7-A1531 TAATACGACTCACTATAGGGAAGCTCCTCGCTTTCCTTCT

T7-S1397 TAATACGACTCACTATAGGATCCTGGTCCCAAGGGTAACT7-A2316 TAATACGACTCACTATAGGGTTCACCAGGCATTCCCTGAA

T7-S2101 TAATACGACTCACTATAGGGTGTTCCTGGCGACCTTGT7-A3107 TAATACGACTCACTATAGGGACCACGTTCACCACTTGT

T7-S2921 TAATACGACTCACTATAGGATCCCCTGGTGCTGATGT7-A3752 TAATACGACTCACTATAGGGATCGTGAGCCTTCTCCT

T7-S3635 TAATACGACTCACTATAGGGTGATGCCGGTCCTGTTT7-A4635 TAATACGACTCACTATAGGGCCATGTGAAATTGTCTCCCA

a S, sense primer; A, antisense primer; number refers to COL1A1DNA nucleotide with which 59 end of primer aligns.

b Bold type indicates region of primer that matches the canineCOL1A1 cDNA sequence. Nonbold contains the T7 polymerase site.

FIG. 1. Sequence of canine COL1A1 cDNA. The nucleotide sequenideoxy chain-termination sequencing. The predicted translationalucleotide sequence. Nucleotides and amino acids in the human sequight hand column correspond to the last nucleotide or amino acid inegion, while amino acids are only numbered in the triple helical re

sequence (4, 17, 18). Dashes indicate missing residues. The start of eacid sequence, lysine residues involved in cross-link formation (✖)

^

indicated. The stop codon is represented ( ). Enzyme cleavage sites forand procollagen C-proteinase are indicated above the human amino ac

collagens were sequentially precipitated with 0.5, 1.5, and 4.0 MNaCl, and the precipitated collagens were analyzed with SDS–PAGEas described previously (8).

RESULTS

Sequence of normal canine COL1A1 cDNA. The se-quence of canine COL1A1 cDNA, including the com-plete translated region, is shown in Fig. 1. The trans-lated region was 12 nucleotides shorter in the dog(4383 nt) than in the human (4395 nt) (4, 17, 18) due tothe absence of nucleotides 204 to 215 in the N-propep-tide coding region of canine COL1A1 cDNA. Nucleotidesequence identity between canine and human COL1A1cDNAs was 93.2% in the translated region. Of the 300nucleotides in the translated region of the canine se-quence that differed from the human, 254 (84.7%) weretranslationally silent. There were 34 amino acidchanges predicted in pro-a1(I), with 17 of thesechanges in a1(I), giving amino acid identities betweenthe canine and human of 97.7% for pre-pro-a1(I) and8.4% for a1(I). The number of amino acids that dif-

fered between canine and human pro-a1(I) in eachregion of the protein are shown in Table IV.

TABLE III

Canine COL1A2 cDNA Primer Pairs (Containing T7-RNAPolymerase Sites) Used in the Nonisotopic RNase

Cleavage Assay

Primerpairsa 59 to 39 Primer sequenceb

T7-S68 TAATACGACTCACTATAGGACCAGGTGATACCTCCGCT7-A953 TAATACGACTCACTATAGGGTTACCAACAGGTCCGATT

T7-S836 TAATACGACTCACTATAGGGTGGAAGTGATGGAAGTGTGT7-A1839 TAATACGACTCACTATAGGGCCAACTTCACCAGCTGTA

T7-S1694 TAATACGACTCACTATAGGGTGCTCCAGGTCCTGATT7-A2681 TAATACGACTCACTATAGGGACCCTTCTCACCAGTAA

T7-S2572 TAATACGACTCACTATAGGGCTGGTAAAGAAGGACTCCGT7-A3487 TAATACGACTCACTATAGGGTGTAGAAGTCCCCTTCATAA

T7-S3341 TAATACGACTCACTATAGGATGGGCAACCTGGTACAGTT7-A4332 TAATACGACTCACTATAGGGAAGGACTCAACTTTCAGTTA

a S, sense primer; A, antisense primer; number refers to COL1A2DNA nucleotide with which 59 end of primer aligns.

b Bold type indicates region of primer that matches the canineCOL1A2 cDNA sequence. Nonbold contains the T7 polymerase site.

of canine COL1A1 cDNA was determined by automated fluorescentduct is shown by single-letter amino acid abbreviations below the

ce (4, 17, 18) that differ from the dog are shown. The numbers in thee row. Nucleotides are numbered from the start of the untranslatedn of a1(I). Nucleotide numbers are kept consistent with the human

cyanogen bromide peptide (CB) is shown (w). Below the dog aminod cysteine residues involved in interchain disulfide bonds (}) are

ceproenth

gioachan

signal peptidase, procollagen N-proteinase, vertebrate collagenase,id sequence (�).

40 CAMPBELL ET AL.

41COL1A1 MUTATION IN CANINE OSTEOGENESIS IMPERFECTA

FIG. 1—Continued

Co

42 CAMPBELL ET AL.

The deduced sequence of the canine pro-a1(I) indi-cated that the greatest number of amino acid changeswere in the N-terminus of this chain. The cleavage sitefor signal peptidase was conserved, and there was com-

FIG. 1—

TABLE IV

Amino Acid Differences Between Human andCanine Pre-pro-a1(I)

Region ofhuman

pre-pro-a1(I)Nucleotide

range

Numberof amino

acids

Number ofamino acidsthat differ inhomologous

dog sequence

Percentageamino acid

identitybetween dogand human

Signal sequence 120–185 22 0 100N-propeptide 186–602 135 9 93.3N-telopeptide 603–653 17 1 94.1CB1 654–662 3 0 100CB2 663–770 36 0 100CB4 771–911 47 0 100CB5 912–1022 37 0 100CB8 1,023–1859 279 5 98.2CB3 1,860–2306 149 0 100CB7 2,307–3119 271 5 98.2CB6 3,120–3695 192 5 97.4C-telopeptide 3,696–3773 26 1 96.2C-propeptide 3,774–4514 247 8 96.8

Total 120–4514 1461 34 97.7

plete amino acid identity in the Gly-X-Y region of theN-propeptide. However, the remainder of the canineN-propeptide had 13 amino acid differences from thehuman (four missing and nine changed residues in thedog). The proline and glutamine residues that markthe N-proteinase cleavage site were conserved, but thenext amino acid downstream from this cleavage sitewas predicted to be methionine in the dog as comparedwith leucine in human pro-a1(I) (Fig. 1). This was theonly amino acid difference between the canine andhuman a1(I) N-telopeptides. In the triple-helical regionof the canine molecule, a Gly-X-Y codon pattern wasmaintained from nucleotide 654 to 3695 (amino acids 1to 1014) just as in the human sequence. The totalnumber of proline residues (253), and their distributionbetween the X (128) and Y (125) positions, were thesame in canine and human pro-a1(I). Furthermore, allof the Y position prolines were in identical locations inthe two species. The asymmetric distribution of the 36lysine codons in canine a1(I) (13 in the X position, 23 inthe Y position) was identical to that in human a1(I).Similarly, there were no differences in methioninecodons in the triple helical region of this chain. Conse-quently, the cyanogen bromide digestion pattern wasthe same for canine and human a1(I) (data not shown).

C-telopeptides of dog and human differed by 1 of 26

ntinued

amino acids, and the C-proteinase cleavage site was

m

43COL1A1 MUTATION IN CANINE OSTEOGENESIS IMPERFECTA

conserved. In the C-propeptide, there were eight aminoacid differences, but all eight cysteine residues and thelocation of the stop codon were conserved.

Identification of a mutation in COL1A1 cDNA.NIRCA was used to screen the COL1A1 and COL1A2cDNAs of one normal dog and six dogs with suspectedOI for mutations. In dog CU1, a 12-week-old goldenretriever puppy with severe OI and dentinogenesis im-perfecta (13), RNase cleaved the RNA duplex spanningnucleotides 709 to 1531 of COL1A1 cDNA (Fig. 2). Thecleavage products were calculated to be 591 and 275bp, predicting a mismatch in the RNA duplex at eithernucleotide 1276 or 960 (after taking into account theT7 sites contributing to fragment length). Sequencingof the corresponding PCR product identified a G1276 toC mutation, and translation predicted an amino acidchange of glycine 208 to alanine (Fig. 3). This glycineresidue is located in CB8 (Fig. 1), and, if the canineexon pattern mirrors that of human, Gly208 is the firstamino acid encoded by exon 18 (1).

The restriction enzyme Aci I was used to differen-tially cleave sequences containing either G or C atresidue 1276 in PCR fragments spanning nucleotides1192 to 1607, and the products were examined byPAGE (Fig. 4). AciI was predicted to produce fragmentsof 199, 190, and 27 bp if nucleotide 1276 is G, andfragments of sizes 190, 117, 82, and 27 bp if nucleotide1276 is C. As shown, the results were consistent withdog CU1 being heterozygous for G/C at nucleotide 1276of COL1A1 cDNA (Fig. 4).

Protein analysis. In a pulse–chase experiment,

FIG. 2. Nonisotopic RNase cleavage assay of COL1A1 cDNA.COL1A1 cDNA from nucleotide 709 to 1531 of seven dogs (one normal(81–136), 6 with suspected OI) was amplified with PCR using primerscontaining T7 sites, transcribed to RNA with T7-polymerase, hybrid-ized into RNA duplexes, cleaved at mismatch sites with RNase, andanalyzed on a 2.5% high resolution agarose gel. Dog OI-CU1 had aunique cleavage pattern with fragments at 591 and 275 bp.

processing of procollagen to collagen was normal in the

cell layer, but was delayed in the culture medium, withless than 50% of procollagen being converted to maturea(I) chains by 8 h (Fig. 5). As reported previously, the

obility on SDS–PAGE of pepsin-digested a(I) chainsfrom dog CU1 was decreased (13). This slowing inmobility of both a1 and a2 chains was similar in pepsindigested collagens extracted from skin and those pro-duced by cultured skin fibroblasts (Fig. 6). In contrast,when hydroxylation was prevented by adding a-a9-dipyridyl to the fibroblast culture media, the pepsindigested a(I) chains migrated normally (data notshown), indicating that the decreased electrophoretic

FIG. 3. Sequence and translation of the normal and mutant allelesof OI-CU1 COL1A1 cDNA from nucleotide 1257 to 1310 (amino acids202–219). The sequence of a purified PCR product spanning nucleo-tides 1192 to 1607 of OI-CU1 COL1A1 cDNA was determined byautomated dideoxy fluorescent sequencing, and the result for nucle-otides 1257 to 1310 (amino acids 202–219) is shown. Overlapping Gand C peaks were consistently seen for nucleotide 1276 (underlined),suggesting a heterozygous G1276 to C mutation. Translation pre-dicts that this mutation will change glycine 208 (bold) to alanine,thus disrupting the Gly-X-Y pattern of the triple helix of type Icollagen.

FIG. 4. Electrophoretic separation of canine COL1A1 cDNAs am-plified from nucleotide 1192 to 1607 before and after AciI digestion.A 416-bp PCR fragment was amplified from COL1A1 cDNA of six OIdogs and one control dog (81–136) using primers S1192 and A1607.The PCR fragments were analyzed on a 5% polyacrylamide gel beforeand after digestion with restriction enzyme AciI. As predicted by thesequence data, AciI digestion produced fragments of 199, 190, and 27bp from the normal allele product and fragments of 190, 117, 82, and27 bp from the abnormal allele product. (The 27-bp fragment was runoff this gel). The presence of cleavage products corresponding to both

the normal and mutant allele products in dog OI-CU1 was consistentwith a heterozygous G to C mutation at nucleotide 1276 of COL1A1.

topPte

di

c(a

44 CAMPBELL ET AL.

mobility was due to over-hydroxylation (19). As mea-sured with densitometry, dog CU1 maintained a nor-mal 2:1 ratio for a1(I):a2(I).

In SDS–PAGE gels of both the cell layer and culturemedium the pro-a1(I) and pro-a2(I) chains migrated asa doublet. In the cell layer both bands of the doubletwere processed to a chains but in the culture mediumhe density of the lower band decreased more than thatf the upper band in the doublet. To characterize theseroteins reversed phase liquid chromatography andico Tag amino acid analysis are currently being usedo separate and characterize the collagens from differ-nt tissues of dogs with OI.

DISCUSSION

Structure and function are highly conserved in typeI collagen, as evidenced by the degree of sequenceidentity between species. The 93% identity found be-tween canine and human COL1A1 cDNA is compara-ble to the 91% previously reported for canine and hu-man COL1A2 cDNA (8). Amino acid identity was evenhigher, with 98% of amino acids conserved in pro-a1(I),compared with 94% in pro-a2(I) (8).

The greatest interspecies variability in the structureof type I collagen is in the amino-terminus of the N-propeptide. While the region coding for the N-propep-tide in human (4) and bovine (4) COL1A1 cDNA is 417bp long and spans from nucleotide 186 to 602, otherspecies lack the following nucleotides: 204–215 are

FIG. 5. Autoradiographs of 5% SDS–PAGE gels showing productsof procollagen processing from the media of cultured skin fibroblasts.Each lane contains 3H-proline-labeled proteins from the culture me-

ium collected from one dish of dog fibroblasts at the postlabelingncubation time indicated. (A) Normal dog: 80% of pro-a chains were

processed to pC-a, pN-a or a chains. In gels of culture mediumproteins, pro-a1(I) chains make up the lower band of a doublet. Theomposition of the upper band in this doublet is under investigation.B) OI dog CU1: processing to mature a chains was minimal evenfter 8 h of incubation.

absent in the canine (Fig. 1), 192–218 in the mousedfi

(18), and 195–215 in the chick (18). In each case, thenumber of absent bases is divisible by three, ensuringpreservation of the proper reading frame.

As in a2(I) (8), the location of all 36 lysine residues incanine a1(I), and the two lysines of the N- and C-propeptide involved in cross-link formation, were iden-tical to those in human a1(I). Such strict conservationsuggest that these lysine residues play very importantstructural and functional roles in type I collagen mol-ecules in all species. The apparent importance of theselysine residues is also consistent with the fact thatover-hydroxylation of lysine is associated with OI invariants in which there is a slowing in winding of thetriple helix of type I collagen (19–23).

Genbank contains one other report of canine COLA1cDNA sequence (Accession No. AF056303), which cov-ers nucleotides 91 to 662 of mRNA isolated from canineosteoblasts (24). This region codes for the signal se-quence through the first three amino acids of the triplehelical region of pre-pro-a1(I). When compared to thenormal dog sequence of this report, there were 17 mis-matched nucleotides, and five mismatched amino ac-ids. Suprisingly, three Y-position codons for the Gly-X-Y region of the N-propeptide, which were found to beproline in this study, were reported to be arginine,serine, and leucine in the osteoblast mRNA (24). Theother two disagreements in amino acid sequence in-volved the last two residues of the N-propeptide, whichFig. 1 shows as Ala-Pro, and the osteoblast sequence

FIG. 6. SDS–PAGE gels comparing pepsin-resistant collagens pro-duced by cultured fibroblasts with pepsin-resistant collagens ex-tracted from control (C) and proband (OI) dog skin. (A) Autoradio-graph of (3H) collagens separated on a 5.75% gel; (B) A 5.25% gelstained with Coomassie blue. The slowing in the electrophoreticmobility of both a1 and a2 chains extracted from skin of the affected

og is similar to that of the chains produced by his cultured skinbroblasts.

wn

1

1

1

1

45COL1A1 MUTATION IN CANINE OSTEOGENESIS IMPERFECTA

reported as Val-Ser. The latter would interrupt thePro-Gln sequence cleaved by N-proteinase. Additionalsequence data for the region in question from two con-trol dogs and four dogs with OI were all consistent withFig. 1 (data not shown).

The Gly208Ala mutation identified in the probanddisrupts the Gly-X-Y pattern essential for triple helixformation. Parallel studies with the NIRCA protocoldescribed here have shown that dog CU3 has a muta-tion in COL1A2, involving a deletion of four nucleo-tides and an insertion of five nucleotides, resulting in aframe-shift and premature termination of translationof the pro-a2(I) chain (9 and manuscript submitted).Similar studies on the other affected dogs in Fig. 2 arein progress.

Six cases of human OI caused by substitution ofalanine for glycine have been reported to date (25–29).For both COL1 genes, the more severe phenotypestended to be caused by glycine to alanine mutationscloser to the C-terminus of the triple helix. Neverthe-less, disease caused by alanine substitutions near the59 end of pro-a1(I) at Gly154 (28) and at Gly220 (29)

as still of moderate severity. The finding of an ala-ine substitution for Gly208 in pro-a1(I) in a dog with

OI type III is consistent with the correlation of geno-type and phenotype seen in different variants of hu-man OI.

Bachinger proposed a model in which a glycine mu-tation stops propagation of triple helix formation, anddisease severity depends on how readily renucleationoccurs amino-terminal to the mutation site (30). Inturn, the ability to renucleate depends on the stabilityof the amino acids in the involved region (30). On a plotof relative stability versus tripeptide unit number ofhuman type I procollagen, glycine 208 falls in an areaof low relative stability (31). This residue is not locatedin one of the hydrophobic regions where the most com-mon, and most stable, “collagen-typical” tripeptides ofGly-Pro-Hyp, Gly-Pro-Ala, and Gly-Ala-Hyp tend to begrouped (32). Since the amino acid sequences of caninea1(I) and a2(I) (8) were both identical to the humansequence from residue 201 to 246 of each chain, it isreasonable to assume that the stability of the caninetriple helix mirrors that of the human triple helix inthis region. Therefore, the severe phenotype associatedwith the substitution of glycine 208 to alanine in thecanine proband may be in part due to its location in anarea of low triple helical stability which is inefficient orineffective in the renucleation process. Furthermore, ithas been suggested that nonhydrophobic regions arenucleation sites for mineralization (33). The severity ofdisease seen with substitution of Gly208 may also berelated to a disruption of mineralization due to thisresidue’s location in a nonhydrophobic region withinthe gap zone (32, 34). Current structural studies of

bone collagen from the proband will show whether

1

mutant a1(I) chains accumulate in the brittle bones,and whether the presence of these mutant chainscauses secondary abnormalities in a2(I) chains, ora1(I) chains that do not contain the mutation.

Recent studies by Forlino et al. (35) have developed anonlethal knock-in murine model for studies of OI. Itwill now be important to compare the primary andsecondary biochemical and morphological lesions indifferent tissues of these mice to those in the tissues ofdogs with different spontaneous mutations in COL1A1and COL1A2.

In conclusion, this study demonstrates that canineCOL1A1 mutations can produce clinical and biochem-ical changes that are analogous to those in human OIpatients. Thus, spontaneous collagen gene mutationsin the dog are generating excellent new experimentalmodels for studies of structure–function relationshipsin type I collagen, as well as potential models for stud-ies of therapeutic regimens for the treatment of differ-ent variants of OI.

ACKNOWLEDGMENTS

This work was supported by Research Grants AR08428 (B.G.C.)and AR20793 (R.R.M.) from the National Institute of Health, andgrants from The IAMS Company, Lewisburg, Ohio. The authorsthank Dr. Da-Nian Gu and Ms. Barbara Hover for their contribu-tions to this study.

REFERENCES

1. Dalgleish, R. (1997) Nucleic Acids Res. 25, 181–187.2. Dalgleish, R. (1998) Nucleic Acids Res. 26, 253–255.3. Bernard, M. P., Myers, J. C., Chu, M. L., Ramirez, F., Eiken-

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

4. Tromp, G., Kuivaniemi, H., Stacey, A., Shikata, H., Baldwin,C. T., Jaenisch, R., and Prockop, D. J. (1988) Biochem. J. 253,919–922.

5. Kuivaniemi, H., Tromp, G., Chu, M. L., and Prockop, D. J. (1988)Biochem. J. 252, 633–640.

6. Phillips, C. L., Lever, L. W., Pinnell, S. R., Quarles, L. D., andWenstrup, R. J. (1991) J. Invest. Dermatol. 97, 980–984.

7. Phillips, C. L., Morgan, A. L., Lever, L. W., and Wenstrup, R. J.(1992) Genomics 13, 1345–1346.

8. Campbell, B. G., Wootton, J. A., MacLeod, J. N., and Minor, R. R.(1998) Arch. Biochem. Biophys. 357, 67–75.

9. Campbell, B. G. (1999) Ph.D. Dissertation, Cornell University,Ithaca, pp. 190.

0. Chipman, S. D., Sweet, H. O., McBride, D. J., Jr., Davisson,M. T., Marks, S. C., Jr., Shuldiner, A. R., Wenstrup, R. J., Rowe,D. W., and Shapiro, J. R. (1993) Proc. Natl. Acad. Sci. USA 90,1701–1705.

1. McBride, D. J., Jr., and Shapiro, J. R. (1994) Genomics 20,135–137.

2. Dziadek, M., Timpl, R., and Jaenisch, R. (1987) Biochem. J. 244,375–379.

3. Campbell, B. G., Wootton, J. A., Krook, L., DeMarco, J., andMinor, R. R. (1997) J. Am. Vet. Med. Assoc. 211, 183–187.

4. Frohman, M. A., Dush, M. K., and Martin, G. R. (1988) Proc.Natl. Acad. Sci. USA 85, 8998–9002.

2

2

2

2

22

2

3

3

3

3

3

46 CAMPBELL ET AL.

15. Minor, R. R., Sippola-Thiele, M., McKeon, J., Berger, J., andProckop, D. J. (1986) J. Biol. Chem. 261, 10006–10014.

16. Laemmli, U. K. (1970) Nature 227, 680–685.17. Bernard, M. P., Chu, M. L., Myers, J. C., Ramirez, F., Eikenberry,

E. F., and Prockop, D. J. (1983) Biochemistry 22, 5213–5223.18. Chu, M. L., de Wet, W., Bernard, M., Ding, J. F., Morabito, M.,

Myers, J., Williams, C., and Ramirez, F. (1984) Nature 310,337–340.

19. Vogel, B. E., Minor, R. R., Freund, M., and Prockop, D. J. (1987)J. Biol. Chem. 262, 14737–14744.

20. Trelstad, R. L., Rubin, D., and Gross, J. (1977) Lab. Invest. 36,501–508.

21. Brenner, R. E., Vetter, U., Nerlich, A., Worsdorfer, O., Teller,W. M., and Muller, P. K. (1989) Eur. J. Clin. Invest. 19, 159–166.

22. Tenni, R., Valli, M., Rossi, A., and Cetta, G. (1993) Am. J. Med.Genet. 45, 252–256.

3. Lehmann, H. W., Rimek, D., Bodo, M., Brenner, R. E., Vetter, U.,Worsdorfer, O., Karbowski, A., and Muller, P. K. (1995) Eur.J. Clin. Invest. 25, 306–310.

4. Wezeman, F. H., and Moskal, S. F. (1998) Canis familiarisPrepro-Alpha 1 Type I Collagen mRNA, Partial cds.1. Loyola

Univ. Medical Center, Maywood, IL, Genbank Accession No.AF056303.

3

5. Lamande, S. R., Dahl, H. H., Cole, W. G., and Bateman, J. F.(1989) J. Biol. Chem. 264, 15809–15812.

6. Valli, M., Sangalli, A., Rossi, A., Mottes, M., Forlino, A., Tenni,R., Pignatti, P. F., and Cetta, G. (1993) Eur. J. Biochem. 211,415–419.

7. Lu, J., Costa, T., and Cole, W. G. (1995) Hum. Mut. 5, 175–178.8. Krakow, D., Wilcox, W. R., Lachman, R. S., King, L. M., and

Cohn, D. H. (1996) Am. J. Hum. Genet. 59S, A266.9. Sarafova, A. P., Choi, H., Forlino, A., Gajko, A., Cabral, W. A.,

Tosi, L., Reing, C. M., and Marini, J. C. (1998) Hum. Mutat. 11,395–403.

0. Bachinger, H. P., Morris, N. P., and Davis, J. M. (1993) Am. J.Med. Genet. 45, 152–162.

1. Bachinger, H. P., and Davis, J. M. (1991) Int. J. Biol. Macromol.13, 152–156.

2. Dolz, R., and Heidemann, E. (1986) Biopolymers 25, 1069 –1080.

3. Maitland, M. E., and Arsenault, A. L. (1991) Calcif. Tissue Int.48, 341–352.

4. Chapman, J. A. (1974) Connect. Tissue Res. 2, 137–150.

5. Forlino, A., Porter, F. D., Lee, E. J., Westphhal, H., and Marini,

J. C. (1999) J. Biol. Chem. 274, 37923–37931.