Proc. Natl. Acad. Sci. USA Vol. 88, pp. 5423-5427, June 1991 Medical Sciences

Mutation in a gene for type I procollagen (COL1A2) in a woman with postmenopausal osteoporosis: Evidence for phenotypic and genotypic overlap with mild osteogenesis imperfecta (Type I collagen/posttranslational overmodifications/glycine substitutions/detection of mutations/direct DNA sequencing)

LORETTA D. SPOTILA, CONSTANTINOS D. CONSTANTINOU*, LARISA SEREDA, ARUPA GANGULY, B. LAWRENCE RIGGSt, AND DARWIN J. PROCKOPf Department of Biochemistry and Molecular Biology, Jefferson Institute of Molecular Medicine, Jefferson Medical College, Philadelphia, PA 19107-6799

Communicated by Gerald D. Aurbach, February 28, 1991 (received for review November 12, 1990)

ABSTRACT Mutations in the two genes for type I collagen (COLIAJ or COLIA2) cause osteogenesis imperfecta (01), a heritable disease characterized by moderate to extreme brittleness of bone early in life. Here we show that a 52-year-old postmenopausal woman with severe osteopenia and a compression fracture of a thoracic vertebra had a mutation in the gene for the a2(I) chain of type I collagen (COLIA2) similar to mutations that cause 01. cDNA was prepared from the woman's skin fibroblast RNA and assayed for the presence of a mutation by treating DNA heteroduplexes with carbodiimide. The results indicated a sequence variation in the region encoding amino acid residues 660-667 of the a2(1) chain. Further analysis demonstrated a single-base mutation that caused a serine-for-glycine substitution at position 661 of the a2(I) triple-helical domain. The substitution produced posttranslational overmodification of the collagen triple helix, as is seen with most glycine substitutions that cause O0. The patient had a history of five previous fractures, slightly blue sclerae, and slight hearing loss. Therefore, the results suggest that there may be phenotypic and genotypic overlap between mild osteogenesis imperfecta and postmenopausal osteoporosis, and that a subset of women with postmenopausal osteoporosis may have mutations in the genes for type I procollagen.

Involutional osteoporosis (1, 2) is a severe health problem that causes over 1.2 million fractures per year in the United States (1). Two forms of the disease have been described: postmenopausal or type I osteoporosis that primarily affects women within 15 to 20 years after menopause, and agerelated or type II osteoporosis that occurs in both men and women over the age of 70. From 15% to 20% of women develop vertebral fractures from postmenopausal osteoporosis, and more than half of women over the age of 70 years have fractures related to osteoporosis (1, 3). Both types of osteoporosis are probably determined, in part, by insufficient accumulation of skeletal mass in young adulthood (1, 2). Normally, bone mass increases until age 30 and then declines. When bone density drops below a threshold level, fractures occur following minor trauma. Type I osteoporosis is apparently triggered by the transient acceleration of bone loss that occurs with the fall in estrogen levels in women at menopause. Several reports suggest that the disease is familial (1, 2, 4). The bone fragility of osteoporosis resembles that seen in mild forms of osteogenesis imperfecta (01), a heritable disease characterized by brittleness of bone early in life (5-7). In 01 the decrease in bone mass and the consequent bone fragility vary from mild to severe. The most severe variants of 01 are lethal, but many are milder. Some probands with The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

5423

milder variants have multiple fractures as children, are relatively fracture-free for many years after puberty, but then experience increased numbers of fractures later in life. The increase in fractures can be particularly prominent in women following menopause. Mild 01 is usually diagnosed on the basis of frequent fractures from minor trauma in early childhood together with associated features that include blue sclerae, abnormal teeth (dentinogenesis imperfecta), thin skin, and progressive hearing loss. Blue sclerae and the other associated features, however, are frequently absent in many probands and families with very fragile bones in childhood and, therefore, there may well be phenotypic overlap between mild variants of 01 and osteoporosis (5). Recent reports (6-12) established that 90% or more of probands with 01 have mutations in one of the two genes for type I collagen (COLlAl and COL1A2), the fibrous protein that provides a major part of the tensile strength of bone and several other tissues. The most frequent mutations causing OI are single-base changes that substitute amino acids with bulkier side chains for one of the glycine residues in the repeating -Gly-Xaa-Yaa- sequences of the triple-helical domain of collagen (Xaa and Yaa stand for the amino acids following and preceding glycine in the tripeptide repeats). The presence of the bulkier amino acid delays folding of the triple helix and thereby causes posttranslational overmodification of the protein. Many of the glycine substitutions also lower the thermal stability of the triple helix, and one glycine substitution was shown to produce a flexible kink in the triple helix (9) and to alter the self-assembly of the collagen into fibrils (10). Because of the frequency of mutations in type I procollagen in 01 and because of the similar bone fragility seen in mild 01 and osteoporosis (5), we and others have tested the hypothesis that some individuals with osteoporosis may have mutations in type I procollagen genes. Previously, the heterozygous parents of a child who was homozygous for a mutation causing 01 were found to have radiologic evidence of reduced bone density while still in their thirties (13, 14). Also, fibroblasts from a woman with osteopenia and ankylosing spondylitis (15) were shown to synthesize a type I procollagen similar to type I procollagens synthesized by fibroblasts from many probands with Ol in that the protein was posttranslationally overmodified and thermally unstable. In addition, fibroblasts from a family with osteoporosis and idiopathic scoliosis (16) were shown to synthesize type I procollagen with a truncated proa2(I) chain. Here we report Abbreviations: 01, osteogenesis imperfecta; Xaa and Yaa, amino acids following and preceding glycine in the repeating tripeptide sequence of collagen; CDI, the water-soluble carbodiimide used here; PCR, polymerase chain reaction. *Present address: Department of Medicine, Duke University Medical Center, Durham, NC 27710. tThe Mayo Clinic, Rochester, MN 55905. tTo whom reprint requests should be addressed.

5424

Proc. Natl. Acad. Sci. USA 88 (1991)

Medical Sciences: Spotila et al.

that a 52-year-old woman who presented with postmenopausal osteoporosis has a single-base mutation that changes a glycine codon to a serine codon in the gene for the a2(I) chain of type I collagen (COLIA2).

MATERIALS AND METHODS Tissue Culture and Protein Analysis. Dermal fibroblasts from the patient (JIMM-311, Jefferson Institute of Molecular Medicine, Philadelphia) and a normal 42-year-old female control (JIMM-86) were grown under standard conditions as described (17) and then incubated at confluency for 4 hr in fresh medium containing 50 tkg of ascorbate and 50 p.Ci (1 uCi = 37 kBq) of L-[2,3,4,5-3H]proline (100 Ci/mmol, Amersham) per ml. Proteins were precipitated from the media with ammonium sulfate, digested with a mixture of trypsin and chymotrypsin, and separated by gel electrophoresis in 7% polyacrylamide/SDS (17). Procollagen was digested with vertebrate collagenase (ref. 17; a gift from John J. Jeffrey, Department of Medicine, Albany Medical College, Albany, NY). The electrophoretic mobility of the A and B fragments was assayed after trypsin/ chymotrypsin digestion in a SDS/6-15% polyacrylamide gel. Thermal stability of type I procollagen was assayed by brief digestion with trypsin and chymotrypsin and electrophoresis in a SDS/7% polyacrylamide gel (see refs. 9 and 17, and references cited). Mutation Detection. Total RNA was extracted from cultured fibroblasts and used to synthesize first-strand cDNA with Moloney murine leukemia virus reverse transcriptase (BRL) (18, 19). The RNA from the RNA'cDNA hybrids was hydrolyzed with 50 mM NaOH for 2 hr at 56°C. The samples were neutralized and passed through a Sephacryl S-200 (Pharmacia) column. cDNA was then used as template for the polymerase chain reaction (PCR) (20). Oligonucleotide primers were designed based on published sequences (21) that allowed amplification of the triple helical domains of the al(I) chain in five overlapping segments and the a2(I) chain in four overlapping segments (Table 1). Each PCR product was then denatured and renatured to generate heteroduplexes of coding sequences from both alleles, and the heteroduplexes were examined for the presence of unpaired bases with a procedure (22) in which unpaired Gs and Ts react with a water-soluble carbodiimide (CDI). CDI bound to DNA was then detected by primer extension with Thermus aquaticus (Taq) DNA polymerase (Perkin-Elmer/Cetus) under conditions in which extension stops at the site of a CDI-modified base. Direct Sequencing of the PCR Product. cDNA was used as template for PCR (20) with primers LS3 and LS5 (Table 1) at 94°C for 1.5 min, 56°C or 58°C for 1 min, and 75°C for 1.5 min. Table 1. Oligonucleotides used for amplification of the a2(I) triple-helical domain coding sequences Oligonucleotide Location* Sequence LS1t 266-273 5'-GCTCACCCTTGTTACCGCTC-3' LS20t -15 through -9 5'-ACTIIGCTGCTCAGTATGAT-3' LS2t 213-219 5'-ACGGCCTTACTGGTGCCAAG-3' LS4O 549-555 5'-CTCTCTCCTGGGAGTCCACT-3' LS3t 516-522 5'-ATTGGAAGCCGAGGTCCTTC-3' LS5t 797-803 5'-TCACCCACAGCACCAGCAAC-3'

CDC6t +7-+13 5'-TAGAAGTCTCCATCGTAACC-3' CDC11t 757-764 5'-TCCCTCTGGAGAGGCTGGTACT-3' *Amino acid codon within which the oligonucleotide begins or ends. Codon number 1 is the first codon of the triple helical domain. tPrimes the synthesis of the antisense strand. *Primes the synthesis of the sense strand.

The picomole ratio of the forward primer to the reverse primer was 20:4, and 30 cycles of amplification were performed. In a second PCR, 1/100th of the first PCR product was used as template for 20 cycles in which the picomole ratio offorward to reverse primers was 50:1. The final product was purified, and the volume was reduced by using Ultrafree MC filtration units (Millipore no. UFC3TTKOO). The purified DNA that consisted of single-stranded sense strand in excess was then sequenced by the dideoxynucleotide chaintermination method (23) using internal primers (Table 2) and a sequencing kit (Sequenase; United States Biochemical). Allele-Specific Oligonucleotide Hybridization. Genomic DNA was extracted from fibroblasts (18) and used as template for PCR with two primers located within exon 37 of the COLIA2 gene (forward primer, 5'-AACCGGATCCTAAAGGAGAAAGAGGAGGCA-3'; reverse primer, 5'-GCTCCAACGGGGCCTGT-3'). The products were applied to Zetabind (24) and hybridized with oligonucleotides specific for the two alleles. The sequence for the normal allele was 5'-CAACAACACCGTTlTlTCACC-3' and the sequence for the mutated allele was 5'-GGTGAAAACAGTGTTGTTG-3'. The probes were labeled with [y-32P]ATP and phage T4 polynucleotide kinase. Excess probe was removed by washing at the melting temperature of each oligonucleotide.

RESULTS The Patient. The patient was a 52-year-old Caucasian woman who was evaluated at the Mayo Clinic after she developed acute mid-thoracic back pain following a severe jolt while driving a truck. X-ray examination of the spine showed an anterior compression fracture ofthe ninth thoracic vertebra and generalized demineralization of the spinal column consistent with osteoporosis (Fig. 1). Bone densitometry of the lumbar spine assayed by dual-energy x-ray absorptiometry (Hologic QDR1000) was 0.75 g/cm2, a value that was in the lowest second percentile for the same age and sex (mean normal value 1.13 g/cm2). On physical examination, the sclerae had a slightly bluish cast, the skin was not abnormally thin, and there was no hyperextensibility of joints. Routine laboratory tests were normal, including serum protein electrophoresis. The patient had a normal menopause 7 years earlier. There was no history of any disease or use of drugs known to be associated with osteoporosis (1, 2). She had tinnitus with slight hearing loss in both ears, but did not wear a hearing aid. Evaluation of her hearing several years prior to the vertebral fracture indicated that she had a mild high-frequency sensory-neural deficit compatible with presbyacusis but without the prominent conductive loss typical of 01. Dentition was normal. Her height of 5 feet 2 inches (157.5 cm) prior to the vertebral fracture was normal considering the height of her parents (father, 5 feet 4 inches; mother, 5 feet 2 inches). She had five previous fractures. At age 7 she had a fracture of the left midfemur in a fall down an embankment in which a heavier child fell on her. At age 8 she fractured her left radius after a fall while ice skating, and she refractured the left radius 1 year later when she fell off a barrel. At age 30 she Table 2. Oligonucleotides used for direct sequence analysis of the PCR product of LS3-LS5 amplification Oligo-

Locationt Sequence 776-771 5'-CTGGAGTGCCAGGAGGT-3' S-1 725-730 5'-TCCACGAAGCCCTTCTT-3' S-2 S-3 666-671 5'-GCTCCAACGGGGCCTGT-3' 599-604 S-4 5'-CCCCGGTCACCTGTGGC-3' *Primes the synthesis of the anti-sense strand. tAmino acid codons spanned by the oligonucleotide.

nucleotide*

Medical Sciences:

Proc. Natl. Acad. Sci. USA 88 (1991)

Spotila et al.

5425

since incubation of fibroblasts with 0.3 mM a,a'-dipyridyl to inhibit posttranslational modifications of the protein by prolyl hydroxylase and lysyl hydroxylase abolished the difference of migration (data not shown). Similar posttranslational overmodification was apparent in proa chains from intact type I procollagen (not shown). The overmodification was not apparent in type I procollagen or collagen from the 42-year-old female control. In other experiments posttranslational overmodification was not seen in type I collagen synthesized by skin fibroblasts from a 73-year-old woman without osteoporosis and from two other unrelated women with post-

FIG. 1. X-ray of patient's dorsal spine. There is a compression fracture of the ninth thoracic vertebra and an apparent decrease in bone density.

fractured her coccyx when she fell on a slippery floor. At age 35 she fractured a right phalanx when her finger was caught in a meat grinder. Analysis of Type I Collagen.- Type I procollagen synthesized by the patient's dermal fibroblasts was examined by polyacrylamide gel electrophoresis in SDS. There was delayed migration of both the al(I) and a2(I) chains derived from the secreted type I procollagen (Fig. 2 Left). The difference in migration was more apparent when the vertebrate collagenase A and B fragments were examined (Fig. 2 Right). There was delayed migration of the vertebrate collagenase A fragments (amino acid residues 1-775) but not of the B fragments (residues 776-1014). The delayed migration was explained by posttranslational overmodification of the chains (6, 7, 9),

c)

i:

EL

C)

al

rn0 a o 0

0

a2

-\

(I)

(-) _l ( ()A

a2

(I)A

[cx1~~~~~~~~~~~~~~~~a (I1))3 a2(I)B

CY2 (1)

FIG. 2. (Left) Polyacrylamide gel electrophoresis of type I collagen of patient and control fibroblasts. The samples were not subjected to reduction prior to electrophoresis. Therefore, al(III) chains migrate as a trimer. (Right) Polyacrylamide gel electrophoresis of type I collagen after cleavage with vertebrate collagenase into the A fragment (775 amino acids) and the B fragment (239 amino acids) followed by trypsin/chymotrypsin digestion.

menopausal osteoporosis. The thermal stability of type I collagen was assayed by incubating the procollagen with proteases under conditions in which triple-helical collagen resists digestion, but the unfolded protein is digested to small fragments (see refs. 9, 17). As discussed elsewhere (6, 7,.9, 12), reduced thermal stability is indicative of a change in primary structure of the molecule, but normal thermal stability does not eliminate the possibility of a structural defect. Assay of the thermal stability of the patient's type I collagen did not reveal any consistent difference from the control (not shown). Mutation Analysis. Because the posttranslational overmodification of the a(I) chains suggested a structural defect (6, 7, 9, 12), attempts were made to find a mutation that altered coding sequences of either the al(I) or a2(I) chain. Total RNA was prepared from the patient's and control's fibroblasts and used to synthesize first-strand cDNA (19). The single-stranded cDNA was then used as a template for nine PCRs (20) using primers that amplified all 3052 base pairs (bp) of coding sequence for the triple-helical domains of the proal(I) and proa2(I) chains. Each PCR product was analyzed by a procedure in which heterozygous single-base mutations can be detected by denaturing and renaturing the products to form heteroduplexes and then treating the heteroduplexes with a water-soluble CDI (22). Analysis of the nine PCR products by primer extension suggested that only one (Fig. 3) contained a sequence variation. The region of interest spanned nucleotides 1951-2813 (amino acids 516-803 of the triple-helical domain) of the a2(I) coding sequence (21). In addition to the full-length fragment of 862 nucleotides, a new fragment of about 460 nucleotides was obtained when the CDI-treated heteroduplex was used for primer extension with primer LS3. Extension in the reverse direction with primer LS5 on the same CDI-modified heteroduplex gave a fragment of about 436 nucleotides (data not shown). The results suggested that a sequence variation was present in the region encoding amino acid residues 660-667. Analysis of RNA-derived cDNA from nine other cell lines from individuals with osteoporosis gave negative results with the same set of primers (Fig. 3 and not shown). Sequence Analysis. To characterize the sequence variation that gave rise to the mismatch observed with CDI, direct sequencing was performed on the relevant PCR product as well as on the three PCR products that spanned the remainder of the coding sequences for the a2(I) triple-helical domain from the patient. The results showed a single-base mutation that changed the codon -GGT- for glycine-661 of the a2(I) chain to -AGT-, a codon for serine. The PCR products contained both G and A (Fig. 4), indicating that the patient was heterozygous at this position. Dideoxynucleotide sequencing of three PCR products spanning the remaining coding sequences for the a2(I) triple-helical chain did not reveal any other difference that would alter an amino acid (21). Allele-Specific Oligonucleotide Hybridization. To confirm the mutation, the patient's genomic DNA was used as a template for the PCR. The PCR products were then hybridized with allele-specific oligonucleotides for the normal coding sequence, or for the normal coding sequence with a

Medical Sciences: Spotila et al.

5426

Proc. Natl. Acad. Sci. USA 88

(1991)

CONTROL

L S3

LS5

4

NORMAL CDI

+

A

-

4-.

-

-

PATIENT

B

C

ASO- Gly0-1 ASO Ser Oi

+

+-

-

ASO Gly ASO- Ser

C

-

-

878

460

nt

FIG. 5. Allele-specific oligonucleotide (ASO) hybridization to amplified genomic DNA. Three concentrations of control and patient PCR product were hybridized with specific oligonucleotides for the normal allele (ASO-Gly-661) or mutated allele (ASO-Ser-661).

rt

However, the mutation was found by allele-specific oligonucleotide hybridization to PCR products of genomic DNA from the patient's three sons, ages 24, 29, and 31. The sons had suffered one to four fractures each following trauma as adolescents. In further experiments, allele-specific oligonucleotide hybridization was carried out with PCR products prepared with genomic DNA from 54 unrelated individuals. None of the 108 chromosomes from the 54 individuals contained the mutation that converted the codon for glycine-661 of the a2(I) chain to the codon for serine (not shown). cancer.

FIG. 3. Detection of mismatched nucleotides (nt) in heteroduplexes of PCR products (22) from control (JIMM-86), patient A and patient B (two individuals with osteoporosis), and patient C (JIMM311, the individual in this study). CDI + indicates that the heteroduplex was treated with CDI prior to extension; CDI - indicates the untreated controls. The PCR product was synthesized by using the forward primer LS3 and the reverse primer LS5 (Table 1). The band that is smaller than 460 nt was considered as background, since it was observed in all samples not CDI-treated. Size of fragments was determined by comparison to the Hae III-digested 4X174 molecular weight standards (BRL).

single-base substitution that converted the codon for glycine661 of the a2(I) chain to the codon for serine. Both oligonucleotides hybridized with PCR products prepared from the patient's genomic DNA (Fig. 5). As expected, the oligonucleotide with the serine codon did not hybridize with PCR products from a control sample of genomic DNA. Allele-specific oligonucleotide hybridization to PCR products demonstrated that the mutation was not present in genomic DNA from the patient's 89-year-old mother, although she had severe thoracic kyphosis and radiographic evidence (not shown) of age-related or type II osteoporosis. DNA was not available from the patient's father who had no history of fractures and who died at the age of 72 of colon NORMAL

PATIENT A Asn

'GIy. Ser

A A

GA

C

G

T

A

C

G

T

;-__.._

A

Asn

A

G

Gly

G

Val

G

T= Val

G T

_

T T

FIG. 4. Direct sequencing of PCR product from the patient and normal control in the region of mismatch. The PCR product of the amplification of cDNA with primers LS3 and LS5 (Table 1) was sequenced with four internal primers (Table 2). The sequence shown was obtained with primer S-2. Sequencing primers were in the reverse orientation, so the sequence derived was antisense. For presentation, the x-ray film was reversed, and the sense sequence is presented. a

DISCUSSION The woman studied here was diagnosed by one of us (B.L.R.) as having postmenopausal osteoporosis because she had a compression fracture of the ninth thoracic vertebra, was 7 years postmenopausal, and had severe osteopenia. The patient had a history of five previous fractures, she had slightly blue sclerae and a slight hearing loss. Her three sons, who inherited the mutation characterized here, had fractures as adolescents. Therefore, the diagnosis of mild 01 cannot be fully excluded on the basis of the clinical criteria that are currently the only means of distinguishing osteoporosis from mild osteogenesis imperfecta (1, 2, 5). The patient, therefore, probably represents phenotypic overlap between the two conditions. The serine substitution at position 661 of the a2(I) chain of type I collagen found in the woman here is similar to >70 glycine substitutions that have been shown to cause OI (refs. 5-7 and Table 3). Any mutation that changes one of the 338 codons encoding glycine in the repeating -Gly-Xaa-Yaaamino acid sequence of collagen introduces an amino acid with a bulkier side chain into the center of the triple helix where only glycine, the smallest amino acid, can be accommodated (see refs. 11, 12, 25). As a result, the replacement of glycine either prevents folding of the three a chains of the protein into a triple helix or distorts the rod-like conformation of the molecule and thereby interferes with the normal self-assembly of collagen into fibrils (8, 10-12). The phenotypes produced by the glycine replacements vary from mild to lethal forms of 01, apparently because the effects on folding of the triple helix and on the processing of procollagen to collagen fibrils are highly dependent on the nature of the amino acid substituted for glycine and the position of the substitution (5-12). The large variation in phenotypes produced by glycine replacements and other mutations in the type I procollagen genes may also reflect differences in genetic background and other still-undefined factors (5-7, 11, 12). The serine substitution defined here is similar to two other serine substitutions in that it did not decrease the thermal stability of the procollagen but it produced posttranslational overmodification (Table 3). The posttranslational modifications of procollagen consist of a series of hydroxylations and glycosylations that are normally terminated by

Medical Sciences:

Spotila et al.

Table 3. List of serine-for-glycine substitutions in type I procollagen Decrease Overin thermal Glycine modifiChain replaced cationt Phenotype stability* + Lethal 01 >200C Proal(I) Gly-598 Lethal 01 Normal + Gly-631 + Severe 01 10C Gly-832 Normal Severe 01 + Gly-844 ND Lethal 01 + Gly-913 ND Lethal 01 ND Gly-964 + Lethal 01 ND Gly-1003 Severe 01 ND ND Gly-1009 + Proa2(I) Normal Gly-661 Osteoporosis Lethal 01 ND ND Gly-865 References for the mutations cited here can be found in ref. 7. ND, not determined. *Thermal stability in each case was determined by brief protease digestion (see refs. 9 and 17) of the patient's and control's type I collagen. Numbers indicate the degrees below the control at which the patient's collagen was cleaved into one or more fragments. tOvermodification indicates that the al(I) and a2(I) chains from the patient were more slowly migrating than those from the control in polyacrylamide/SDS gels.

folding of the protein into a triple-helical conformation (5, 11, 12). In 01, overmodification occurs because a change in the primary structure of either the proal(I) or proa2(I) chain delays folding of the triple helix. Recent studies (26) indicate that posttranslational overmodification can in itself slow the processing of procollagen to collagen and increase the critical concentration for self-assembly of collagen into tibrils. Therefore, mutations such as the serine substitution for glycine-661 in the a2(I) chain seen here can alter several different steps in the synthesis and assembly pathways for type I collagen. Genetic factors in involutional osteoporosis are difficult to define (1, 2) because of the late onset of the disease, and a number of independent factors may influence bone mass. However, genetic factors have previously been implicated in postmenopausal osteoporosis by comparisons of different racial groups, twin studies, and evaluations of daughters of women with the disease (1-4). The woman studied here probably represents a subset of the postmenopausal osteoporosis population. The mutation substituting serine for glycine-661 in the a2(I) chain was not found in nine other patients with osteoporosis. However, >70 different mutations in the COLJAJ and COLIA2 genes have been found in probands with 01, and only two or three unrelated individuals were found to have the same mutation (5-7). Therefore, defining the frequency of such mutations in postmenopausal osteoporosis will require extensive analysis of the alleles from different probands. Searching for such mutations is still technically difficult, but defining the mutations may be of considerable benefit to patients. Numerous studies (1) suggest that, if patients who are at risk for osteoporosis can be identified early in life, estrogen and other

Proc. Natl. Acad. Sci. USA 88 (1991)

5427

therapies can be used to diminish bone loss before it reaches a critical stage. We are grateful to Ms. Gi-Chung Chen, Nasrin Rafi, and Kanger Wu for expert assistance in cell culture. Also, we are grateful to Clinton T. Baldwin, Gerard C. Tromp, Helena Kuivaniemi, Robert G. Knowlton, James R. Spotila, and Leena Ala-Kokko for helpful criticisms and suggestions. The work was supported in part by National Institutes of Health Grant AR 38188 and by a grant from the Lucille P. Markey Charitable Trust.

1. Riggs, B. L. & Melton, L. J. (1986) N. Engl. J. Med. 314, 1676-1686. 2. Hahn, B. H. (1989) in Arthritis and Allied Conditions, ed. McCarty, D. J. (Lea & Febiger, Philadelphia), 11th Ed., pp. 1812-1843. 3. Melton, L. J. & Riggs, B. L. (1987) in The Osteoporotic Syndrome: Detection, Prevention and Treatment, ed. Avioli, L. V. (Grune & Stratton, New York), pp. 1-30. 4. Seeman, E., Hopper, J. L., Bach, L. A., Cooper, M. E., Parkinson, E., McKay, J. & Jerums, G. (1989) N. Engl. J. Med. 320, 554-558. 5. Prockop, D. J. (1987) in Harrison's Principles of Internal Medicine, eds. Braunwald, E., Isselbacher, K. J., Petersdorf, R. G., Wilson, J. D., Martin, J. B. & Fauci, A. S. (McGrawHill, New York), 11th Ed., pp. 1680-1688. 6. Byers, P. H. (1990) Trends Genet. 6, 293-300. 7. Kuivaniemi, H., Tromp, G. & Prockop, D. J. (1991) FASEB J. 5, 2052-2060. 8. Kadler, K. E., Hojima, Y. & Prockop, D. J. (1990) Biochem. J. 268, 339-343. 9. Vogel, B. E., Doelz, R., Kadler, K. E., Hojima, Y., Engel, J. & Prockop, D. J. (1988) J. Biol. Chem. 263, 19249-19255. 10. Kadler, K. E., Torre-Blanco, A., Adachi, E., Vogel, B. E., Hojima, Y. & Prockop, D. J. (1991) Biochemistry, in press. 11. Engel, J. & Prockop, D. J. (1991) Annu. Rev. Biophys. Biophys. Chem., in press. 12. Prockop, D. J. (1990) J. Biol. Chem. 265, 15349-15352. 13. Dickson, L. A., Pihlajaniemi, T., Deak, S., Pope, F. M., Nicholls, A., Prockop, D. J. & Meyers, J. C. (1984) Proc. Natl. Acad. Sci. USA 81, 4524-4528. 14. Prockop, D. J. (1988) Arthritis Rheum. 31, 1-8. 15. Constantinou, C. D., Pack, M. & Prockop, D. J. (1989) Cytogenet. Cell Genet. 51, 979. 16. Shapiro, J. R., Burn, V. E., Chipman, S. D., Velis, K. P. & Bansal, M. (1989) Connect. Tissue Res. 21, 117-124. 17. Pack, M., Constantinou, C. D., Kalia, K., Nielsen, K. B. & Prockop, D. J. (1989) J. Biol. Chem. 264, 19694-19699. 18. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning. A Laboratory Manual (Cold Spring Harbor Lab., Cold Spring Harbor, NY), 1st Ed., p. 194. 19. Gubler, U. & Hoffman, B. J. (1983) Gene 25, 263-269. 20. Saiki, R. K., Scharf, S., Faloona, F., Mullis, K. B., Horn, G. T., Erlich, H. A. & Arnheim, N. (1985) Science 230, 13501354. 21. Kuivaniemi, H., Tromp, G., Chu, M.-L. & Prockop, D. J. (1988) Biochem. J. 252, 633-640. 22. Ganguly, A. & Prockop, D. J. (1990) Nucleic Acids Res. 18, 3933-3939. 23. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 24. Potashkin, J. & Huberman, J. (1986) Exp. Cell Res. 165, 29-36. 25. Piez, K. A. (1984) in Extracellular Matrix Biochemistry, eds. Piez, K. A. & Reddi, A. (Elsevier, New York), pp. 1-40. 26. Torre-Blanco, A., Hojima, Y. & Prockop, D. J. (1990) Fed. Proc. Fed. Am. Soc. Exp. Biol. 26, 3421.