Defective splicing of mRNA from one COL1A1
allele of type I collagen in nondeforming (type I)
osteogenesis imperfecta.
M L Stover, … , M B McKinstry, D W Rowe
J Clin Invest.
1993;
92(4)
:1994-2002.
https://doi.org/10.1172/JCI116794
.
Osteogenesis imperfecta (OI) type I is the mildest form of heritable bone fragility resulting
from mutations within the COL1A1 gene. We studied fibroblasts established from a child
with OI type I and demonstrated underproduction of alpha 1 (I) collagen chains and alpha 1
(I) mRNA. Indirect RNase protection suggested two species of alpha 1 (I) mRNA, one of
which was not collinear with fully spliced alpha 1 (I) mRNA. The noncollinear population
was confined to the nuclear compartment of the cell, and contained the entire sequence of
intron 26 and a G-->A transition in the first position of the intron donor site. The G-->A
transition was also identified in the genomic DNA. The retained intron contained an in-frame
stop codon and introduced an out-of-frame insertion within the collagen mRNA producing
stop codons downstream of the insertion. These changes probably account for the failure of
the mutant RNA to appear in the cytoplasm. Unlike other splice site mutations within
collagen mRNA that resulted in exon skipping and a truncated but inframe RNA transcript,
this mutation did not result in production of a defective collagen pro alpha 1 (I) chain.
Instead, the mild nature of the disease in this case reflects failure to process the defective
mRNA and thus the absence of a protein product from the mutant allele.
Research Article
Find the latest version:
http://jci.me/116794/pdf
Defective Splicing
of mRNA from One COLlAl Allele of
Type
I
Collagen in
Nondeforming (Type I) Osteogenesis Imperfecta
M. L.Stover,D.Primorac, S. C.Liu,* M. B.McKinstry,andD. W. Rowe
Fromthe Departments ofPediatrics and *BiostructureandFunction, University ofConnecticut HealthCenter, Farmington, Connecticut 06030
Abstract
Osteogenesis imperfecta (OI) type I is the mildest form of heritable bone fragility resulting from mutations within the COLlAlgene.We studied fibroblasts established fromachild
with01typeI and demonstrated underproduction ofal(I)
col-lagenchains andal(I) mRNA. Indirect RNase protection
sug-gestedtwospecies ofal(I) mRNA,oneof whichwasnot col-linear with fully spliced al (I) mRNA. The noncolcol-linear
popula-tionwas confinedtothe nuclearcompartmentof the cell, and
contained the entiresequenceof intron 26 andaG-* A
transi-tion in the first positransi-tion of the intron donor site. The G-- A
transition was also identified in the genomic DNA. The
re-tained intron conre-tainedanin-frame stopcodon andintroduced anout-of-frame insertion within the collagen mRNA producing
stopcodons downstream of the insertion. These changes proba-blyaccountfor the failure of themutantRNAtoappearin the
cytoplasm. Unlike other splice site mutations within collagen mRNA that resulted inexonskipping andatruncated but
in-frame RNA transcript, this mutation didnotresultin produc-tion ofadefective collagen proal (I) chain. Instead, the mild
natureof the disease in thiscasereflects failuretoprocessthe defective mRNA and thus the absence ofaprotein product from
themutantallele.(J. Clin. Invest. 1993. 92:1994-2002.) Key words:osteogenesis imperfecta*RNA splicing-nuclearRNA
transport-RNase protection*premature stopcodon
Introduction
Osteogenesis imperfecta (01)1 isaheritabledisorder that
re-sults in bone fragility. Identification of mutations within the COLlA1 andCOL1A2 genesthat encode the chains oftypeI
collagen has beguntoprovideadetailed understanding ofthe
structural requirements for thetypeIcollagen molecule( 1,2).
Mutations that disrupt thetriple-helicalconfiguration of
colla-This workwasaposterpresentationatthe Americal Society ofBone and MineralResearch, Minneapolis, MN, September 30-October4, 1992.
Address correspondencetoDr. DavidW. Rowe, Department of
Pediatrics, School of Medicine, University ofConnecticut Health
Center,263Farmington Avenue, Farminton, CT 06030-1515. D. Primorac's permanent address is the University of Zagreb,
SchoolofMedicineatSplit, 58000 Split, Croatia.
Receivedfor publication2February1993andinrevisedform 30
April1993.
1.Abbreviations usedin thispaper: nt, nucleotide; 01, osteogenesis
imperfecta.
gen, including substitutions for glycine in the gly-x-y triplet
(3), partial
gene deletions (4), and exon skipping(5),
pro-duce, dependingonthe locationof the mutation, arange of disease severity that extends from lethal(01
type II) toseverely deforming(01typeIII)tomildly deforming(01
typeIV) (6). In01 type I fractures are usually limited to the prepubertal years and are not associated with bone deformity. At times the fracture frequencyis not recognized as a distinctabnormality and thedisorder only comes to medical attention when early-onset osteoporosis develops in adult family members. Thus, mild forms of 01 and "familialosteoporosis"often coexist, in part because theunderlyingpathogenesis may be similar. For mutations in theaI(I) chain (COLlA 1), there appears to be a positional effect in which those at the COOH terminus aremorelikelytoresult in severe disease than those at theNH2 terminus (7). In fact, substitutions for glycine have been iden-tified near theNH2-terminalend ofthetriplehelix that result in nondeforming bone disease (type I01)and heritable osteoporo-sis(8). Mutations within the a2 (I) chain (COLIA2 gene) that produce mild 01 orosteoporosis (9)do notalwaysoccur atthe NH2-terminal end of the chain (10).
The most common mechanism for 01 type I appears to be decreased synthesis of normal type I collagen molecules. The synthesis of proa 1 ( 1 ) collagen chains ( 11 ) and the steady-state level ofa1(I) mRNA appears to be reduced by - 50% (12,
13), suggesting that one of the two COLlA 1 alleles is "null." Wepreviouslyshowed that dermalfibroblasts fromsome indi-viduals withtype I 01containnormal or elevated amounts of
aI(I)mRNA inthenucleus( 14). Furthermore,anovel spe-cies ofa1(I)collagenmRNA present in the nuclear compart-ment of cells from one such child was not collinear with a cDNA probe (15). In this paper we show that a mutation withinasplicedonorsiteresultsininclusion oftheentire suc-ceedingintron in the mature mRNA that accumulates in the nuclear compartment. Apparently because no abnormal proaI(I) chains are synthesized from the mutant allele,the clinicalphenotypeof thischild is mild.
Methods
Description ofpatient. CF was a 13-yr-old female at the time of the dermalbiopsy (cell line 053). She hadbeenfollowed at the Newington Childrens Hospital (NCH) between 10 and 15 yr of age. She had
sus-tainedsevenfracturesbefore her first NCHvisit,allof which healed withoutdeformity. Subsequently, the fracture rate diminished and her primary problems were ligament strains and knee discomfort second-arytopatellar subluxation. Herheightwas atthe 50-75percentile. She hadblue sclera,ligamentous laxity,amildthoracicscoliosis (<10%), and normaldentition. Hercurrentage is28 yr, and she and her family, includinganaffected brother, have been lost to follow-up.
Cell culture. A fibroblast strain wasderived from askin explant
takenfromtheinner aspect of the upperarm.Fibroblastswere main-tainedaspreviouslydescribed ( 12). Confluent cultureswereharvested forprotein or RNAafter 48 hinfresh culture media supplemented
dailywith 25,ug/mlascorbic acid. 1994 M. L.Stover, D.Primorac, S.C.Liu,M.B.McKinstry, and D. W.Rowe
J.Clin.Invest.
© The American Society for Clinical Investigation, Inc.
0021-9738/93/10/1994/09 $2.00
Collagen analysis. Forexperiments analyzing collagen synthesis, the cultured cells wereradiolabeled with [2,3-3H]-L-proline (DuPont
Co., NEN Research Products, Boston, MA) as previously described ( 12 ). Collagen synthesis was determined from collagenase susceptible
countsin analiquot of the combined medium and cell extract (16). Therelative proportion of collagen al (I), a2(I), and al(III) chains
wasdetermined by interrupted SDS-PAGE electrophoresis of pepsin-treatedprotein extracts( 17).
Cellfractionationandnucleicacid extraction. Total RNA was ex-tractedfrom theconfluent fibroblasts cell layer using the SDS
protein-ase Kmethod(18).The nuclear and cytoplasmic compartments of the cellswereisolatedaspreviouslydescribed ( 14). Briefly, confluent fibro-blast cultures ( 12-100-mm petri plates) were scraped from the plate in 10mlof coldPBS, centrifuged,and resuspended in 15 ml of lysis buffer
(10mMTris,pH 7.5, 10 mM KCI, and 1.5mM MgCl2)containing 0.25%TritonX-100. The pelletwasdisrupted by dounce homogeniza-tion withatight-fitting pestle. Thesuspension was centrifuged for 10 minat2,200g and the supernatant was transferred to a 30-ml corex tube(cytoplasmicextract) containing 2.0mlof a lOXextraction buffer
(10% SDS,0.10 M Tris,pH 7.5, 0.05 M EDTA, and 500Mg/ml of
proteinaseK). The nuclear pellet was reextractedasecond time in5ml oflysisbuffer with Douncehomogenization. Aftercentrifugation, the supernatantwascombined with the first extract, producing a final con-centration of the extraction buffer of lx. The nuclear pellet was ex-tracted in 15 ml of lx extraction buffer. Subsequent steps of RNA isolationfollowed the SDSproteinaseKprotocol ( 18).
DNA was harvested from cultured fibroblasts in 1X extraction buffercontaining 100
Mg/ml
of proteinase K. The digest was extractedsequentiallyin phenol/chloroform and chloroform and adjusted to 0.3
MNaOAc. Twotothree volumes of ethanol was layered over the
ex-tract,and the DNA was spooled from the interface, washed in 70%
ethanol, dried,and resuspended in water.
RNAanalysis.Therelative content ofal (I)and a2(I)mRNA in thenuclear and cytoplasmic compartments was assessed by dot hybrid-izationaspreviouslydescribed ( 14). The filters were hybridized with
al(I)-anda22(I)-specificcDNA probes (HF404[19]and HF32 [20]) afterlabeling bythe randomprimer method (21 ). The relative
inten-sity ofhybridization was quantitated in a Betascope G03 (Betagen, Waltham, MA) from the slope of the dilution curve of patient and control RNA.
The methodforindirectRNaseprotectionof nuclear and cytoplas-micRNAusing single-stranded (SS)antisense cDNA toa I( 1 ) mRNA has been previously described (15). Briefly, hybridization of the sscDNA toRNA wascarried out at 500Cin20 Mlofhybridization
buffer(0.75MNaCl,1 mMEDTA,50mMHepes)(N-
[2-hydroethyl]-piperazine-AV-[2-ethanesulfonic
acid])for 3 h. The sample was dilutedto300Mlin RNasebuffer (0.2MNaCl,20 mMHepes,pH 7.5, and 2 mMEDTA)andincubated with 1
Mg
RNaseAand10Mg RNaseTlfor 30 minat37°C. TheRNase-resistanthybridswereextracted in SDSproteinase K,ethanol precipitated, and reappliedto 1% agarose gel made up in 2.2Mformaldehyde.Theelectrophoresiswascarriedout at
roomtemperature ina25 mMimidazole,25 mM 3-N-morpholino]-propanesulfonicacid(MOPS),0.5 mM Na-EDTAbuffer,pH 7.0, for 3hat100V.TheRNAfragmentsresistanttotheRNasedigestionwere transferredto anitrocellulose filter withaPosiblot apparatus
(Strata-gene,LaJolla, CA)andidentifiedby hybridizationwith 2.5 x 106 cpm of[
32P04]UTP-labeled
cRNAprobeof HF404 transcribed fromasp64vector(22).
The direct RNase protection wasperformed with a cRNA ofa
PCR-derived DNAfragmentthathad been clonedintopBSII-SK+
vec-tor(Stratagene). Nuclear and cytoplasmicRNA(7
Mg)
were hybrid-ized with 104cpmof thecRNAovernightat50°C in 50Mlofbuffercontaining0.4MNaCl, 10mMHepes, pH 7.3,and 1 mM EDTA. The
next day the solution was diluted to 300
Ml
of T2 buffer (50 mM NaOAc, pH 4.6,2 mM EDTA, and0.1 MNaCl) containing 5M1
of crude T2 RNaseaspreviouslydescribed(23). Thefragmentsof theprotected probewereidentified byelectrophoresisina6.0%denaturing polyacrylamide gel. An RNA ladder of 0.16-1.77 kb (no. 56235A;
GIBCO BRL, Gaithersburg, MD) was end labeled with'-[32PO4]ATP
accordingtothe supplier's directions.
PCRofRNAand genomicDNA. 5.0
,g
of nuclear or cytoplasmic RNAwasreverse transcribed to cDNA using an al(1)-specific primer directedtosequences within the COOH-terminal propeptide as previ-ously described (8). The reaction was carried out in 50 mM Tris, pH8.3,7mMKCl,3mM
MgCl2,
10mM dithiothreitol,and2.5,gof the primer. The mixture was heated at650Cfor 3min,cooled, adjusted to contain500MgM
dNTPs, and 500Uof Moloney murine leukemia virus (M-MLV) reverse transcriptase. The incubation lasted for 1 h at 370C and was followed by phenol/chloroform extraction and ethanol precip-itation. The sample was dissolved in 20,lwater and 1 Ml was used for PCRamplification. In the case of genomic DNA, 1 Mg of DNA was used in the PCR reaction. The oligonucleotide pairs for each set of amplificationsaregiven in Table I.ThePCR mixture contained 50 mMKCl, 10mM Tris, pH 8.3, 1.5 mMMgCl2,0.0 1%gelatin,and 2.5 UAmpliTaq (PerkinElmerCorp., Norwalk, CT) inafinal reaction volume of 84 Ml. The mixture was heated to 940Cfor5min,after which the 16 Ml containing 200 gM
dNTPs(hotstart[24])wasadded and thesamplewascycled25 times withdenaturationat940Cfor 1 min,hybridizationat620C for 1 min,
andpolymerizationat720C for 1 min inaDNAthermalcycler(Perkin
Elmer-Corp.). Theproductswereextracted inphenol/chloroformand
1:10of the original reaction volumewasappliedto a5%
polyacryl-amide TBEmini-gel.
For directDNAsequencingof PCRproducts,weused the dsDNA CycleSequencing System (GIBCOBRL). PCR products were cloned either withG-tailinginto C-tailedpUC9vector(Pharmacia Fine Chem-icals, Piscataway, NJ)orbyblunt endcloning into Sma site pBSII-SK+. Forsequencingof the clonedDNA,weused theSequenase Re-agent Kit(no.70750;United States Biochem.Corp.,Cleveland, OH).
Results
Total collagen synthesis by the fibroblasts derived from the patient (Table II)was4.0%oftotal protein synthesis. For com-parison, three cell strains derived from patients with severe forms ofthediseaseare shown. We have observed significant variation in the rate and percentage of collagen synthesis in partdependentonthe ageofthe donor. The range of synthesis innon-OI controlcells is 5-7% (5). In most of the cell strains obtained frompatientswith 01 type I, a reduced rate of total collagen synthesis isobserved.
Therelative proportion oftype I and type III collagen was determinedonpepsin-treated radiolabeled proteinextracts of themedium and celllayer(Fig. 1). The ratio was elevated to 0.38and 0.45intwoexperiments(controlpatientsrangefrom 0.15 to 0.25). Because an equal amountoftotal collagen is loadedintoeachelectrophoresislane, the elevation in the colla-gen type III-to-Iratioreflects underproduction of type I colla-gen. The electrophoretic mobilities ofthe a I(I) and a2(I) chains from thepatient samplewere identical to those of the controlsample.
Therelativecontentofa 1(I) and a2 (I) mRNA was deter-mined bydotblothybridizationandcomparedwith the ratio of control cells(Fig.2). The slopeofthelineofdotdensitywhen thefilterwashybridizedto an a 1(I)or a2(I)cDNAprobewas
compared withtheslopeofacontrol RNAappliedtothe same
filter.Inthisassay, the
alI
(I)/a2(I)mRNAratio ofthecontrol is normalized to 1. Whenthe slope ofthe al(I) and a2(I) mRNA derived from cell strain 053 was compared with the standard (std),the al(I)/stdwas0.30 and the a2(I)/stdwas1.19, makingthealI (I)/a2(I)ratioofthepatient0.25.Thus,
the amountofa 1(I)mRNA presentinthetotal cellularRNA
Table I. Oligonucleotide Pairs Used to Amplify al(I)mRNAEncoded by HF404
NT* 3-Oligonucleotide 5'
Oligonucleotide pair/exonsamplified 5'Oligonucleotide-3' NT*
1:35-39 2541 AGCAGGACCATCAGCACCAGGGGAT CCCGGCCCTGCTGGCTTTGCTGGCC 2948
2:30-36 2113 TTTCGGCCACGACTAACCAGG AGACCTTGGCGCCCCTGGCCCCTCT 2621
3:25-31 1800 TTGCACACCACGCTCGCCAGGGAAA TCAAGATGGTCGCCCCGGA 2159
4:24-28 1746 AGCAGGGCCTTGTTCACCTCTCTCG AGCCCTGGCAGCCCTGGTCCTGATG 2005
5:19-23 1330 GCTTCACCGGGACGACCAGTC GTATTGCTGGTGCTCCTGG 1708
*Thenucleotide number is determined from the a
1(I) collagen
mRNAstartsite.cells have a 2:1 ratio of al(I)/a2(I), the amount of alI (I) of thepatient sample is reduced to< 1:1.
Analysis of this patient and other cell lines with 01 type1
indicatedthat theaI(I)/ a2 (I) collagen mRNA ratio extracted from isolated nucleiapproximated or exceeded 2:1 in contrast to the low ratio from cytoplasmic compartment ( 14). This observationsuggestedadefect in a 1(I) mRNA transport simi-lar to that observedinsplicing defects associated with
f3-thalas-semia (25, 26).AnindirectRNaseprotectionwasdevelopedto detect insertions withincollagenmRNA( 15).Itrelieson pro-tection of the mature mRNA with a single-stranded antisense cDNA.TheprotectedorcleavedRNAis detected byNorthern hybridization ofthe RNase-digested products. Nuclear and cy-toplasmicRNA wasprotected with a 1.8-kb cDNA fragment derived from HF404 that encodes exons 19-43 (Fig. 3). In
RNA from cell strain 053, the expected 1.8-nucleotide (nt) band andtwo smaller bands(1.4and 0.4 nt)were visible in
nuclear RNA.Onlythefully protectedband was present in the cytoplasmic RNAand wassimilar in sizeto RNA extracted fromnormal cells.
TocharacterizetheabnormalRNAspecies, nuclear RNA was reversetranscribed and then PCRamplified insegments to spantheregion encoded by HF404 (Fig. 4). Of the five PCR primersets used(Table I),two(sets 3 and4) produceda sec-ondhighermolecularweight ( - 150bp larger) band from the
patientbut notthe control RNA. Athirdband well above the 1.3-kb marker was also unique to these oligonucleotide sets and probably represents a heterodimer produced when two complementary strands of differentlengths hybridized during
TableII. Production of Total Collagen by Dermal Fibroblasts
DerivedfromPatients with Various Forms of OI
Collagen Total protein Percent Cellline synthesis synthesis collagen
dpm/mgDNA
053(This case) 372 1,729 3.98 118 (Lethal01) 1,385 4,042 6.34
144(TypeIII01,age 5) 818 2,264 6.69
147(TypeIII01,age 46) 693 2,292 5.59
the laterounds ofamplification. Both primersets amplifya
cDNAsegment that hasintron 26 in common.
The PCRproducts ofthe normalandmutant RNAwere
cloned. DNAsequencingrevealed thepresenceofthe intron 26
(Fig.5A)in thelarger species. Furthermore,theexpectedfirst
positiondonor site Gwasreplaced byanA. This mutation was
confirmed when the corresponding region of the genomic DNA wasisolated by PCRand foundtobe presentin - 50% of
the clonessequenced (Fig.5 B).
Thecloned reverse transcriptase PCR fragments
encom-passingexons25-28 derivedfrom oligonucleotideset 3 were
used to generatearadiolabeledantisense cRNAprobes comple-mentary to the retained intron andthenormally spliced
tran-script. Theseprobeswerehybridizedtonuclear and cytoplas-micRNAderived from thepatientand control cell strain(Fig.
Delay Unreduced Reduced
1 2 3 4
-Y1(I111)unreduced
_pm% -(x1(II1)reduced
mm
_mI
U..~--c1(1)
oif wow':2*_1 x_-a2(I)
Figure1.SDS-PAGEof radiolabeled collagen from patient and con-trolfibroblastcultures.Thecells werelabeled for 24hand the cell layer and media combined for analysis. The sample was treated with
pepsintodestroy the noncollagen proteins before the gel electropho-resis. The typeIIIcollagen is separatedfrom type I based on the
unique intrahelicalcysteinedisulfide bonds within typeIIIcollagen. Therelativeratio of theaI
(111)/al
(I)wasobtained by quantitativedensitometryof the pattern inlanes3 and4. Lanes1 and 3, patient; lanes 2 and4, control.
al(I)mRNA
6-=
5-o 4.-I
3-
2-0I
I I I I I I I II-I
0 1 2
glTotalRNA Ratio PatientlStd=0.3
1
2
a2(1) mRNA
I I I I I I I II I
0 1 2
pgTotal RNA
RatioPatientlStd=1.19 Patienta((1) / ca2(1) mRNA=0.25
Figure 2. Plotof dot hybridization densityandtotal RNAobtained
from patient- and control-cultured fibroblasts.TheslopefortheaI (I)
mRNAofthepatient islessthanthecontrolwhile thatofa2(I)is similartocontrol.Thisfinding is consistent withareduction ofthe
a1(I)mRNAthat accumulateswithinthefibroblasts ofthepatient.
6). The probe containing the included intron protected the
predicted 442-nt fragmentinthe nuclear (Fig. 6 A, lanes 4 and 6)butnotthecytoplasmiccompartment(Fig.6A, lanes 3and
5) ofthepatient RNA. Noevidence of the 442-ntbandwas presentinthe controlRNA(Fig. 6 A, lanes 7 and 8). The RNA encoded by the normal COLlA1 allele from the patientaswell asallthe RNA from thecontrol cell linedidnot protectthe
intron 26-containing probe. Instead the probe was cleaved
producingtwofragments (209 and 90 nt) collinear with the flankingexons. The cRNAprobe derivedfrom the normally spliced allele was protected equally well by total fibroblast
RNAfrom the patient (Fig. 6 B, lane 6) and control (Fig. 6 B, lane 5), indicating thatexonskippingwasnotoccurringtoa significant degreeas aresult of the donorsitemutation(27). The intensity of the normally spliced transcriptinthe patient
total RNAextractis significantly lessthan the control, con-firming the lowamountof normala1(I) mRNA that ispresent
inthiscellstrain. Discussion
The fibroblasts derived from this patient hadaG-- A
transi-tionatposition + 1 of intron 26 in oneof thetwoCOLlA
1
2
3
4
5A B
A
B C
A B
C
A
B
A
BC
Figure 3. Indirect RNase protection of nuclear and cytoplasmical (I)
mRNAextractedfrom patient's cul-turedfibroblasts. The fully protected ( 1.8-nt) bandpresentin both
com-partmentsisidenticaltothe band foundin normal fibroblasts. However, thelower bands ( 1.4 and 0.4 nt) that
werepresentinthe nuclear RNA (lane
1) but absent from the cytoplasmic RNA(lane 2)wereuniquetothis
pa-tient.
allelessothatthe donorsplicesitewasinactivated. Asa
conse-quenceofthis mutation, all of intron 26wasretained and the frame-shifted mRNA transcript contained severalstopcodons within and downstream of the intron insertion. Analysis of RNAtranscriptsfrom other multiexongenessuchas
dihydro-folate reductase(28)and triosephosphateisomerase(29)have shown thattransportof the abnormaltranscriptinto the
cyto-plasmwasreduced when thestopcodonwaslocatedin allbut
the terminalexon.The behavior of themutantal(I) mRNA
in the 053cells isconsistent with the model thatastopcodon
before thepenultimateexonprohibitstransportfrom the nu-cleus,since therewas noevidence of abnormal transcript
accu-mulation inthecytoplasm.
M
1.30 1.00
0.80
0.60
Figure4. PCR of reverse-transcribed cDNAusing
nuclearRNA extracted from control- and
patient-cultured fibroblasts.(Lanes A)PatientRNA; (lanes B) control RNA; (lanes C)clonedal(I)cDNA. The
0.31 newband ispresentin lanes 3 A and 4 A, and is
0.27 - 150bp
larger
that theexpectedband. Thehetero-dimerband is also present in these lanes. Theprimer 0.1 9 pairsaregiven: (lanes 1)exons35-39; (lanes2)exons
-0' 12 30-36; (lanes 3)exons25-31; (lanes 4)exons24-28; (lanes 5)exons19-23.(Lane M)PhiX HaeIII DNA
markers.
DefectiveSplicing ofal(I)mRNA in OsteogenesisImperfectaType 1997
6-c 5-*
4-'r
3-._ d
2-01
1
-8..
-Patient Control
GAT C GAT C
ftm.
W. _Wh v"
A.
-0
x
wU
a 0
x
Iw
... GOT OTT CCC GOR CCC CCT GOC OCT OTC
... Gly Ual Hyp Gly Pro Hyp
Gly
Ala UalIntron 26 a
gtaagtatct cctttccatc cctacctcct tccgattgct
gccccggcac tttcgtctcc ctgcaggagg ggtgcta...
OAT GAR ORO OCT ...
Asp Gly Glu Ala ...
Muta nt All
IelIe
N ormalI
AIelIe
G A T C G A T C
t~~~~~~~~~~~r ~ ~ ~ ~. L
0
_' . o~~~~~~~~~~~~C
dGGTTaagac
GGCCTTCstagttN
Figure 5. DNA sequencingofthemutation within intron 26.(A)
DNAsequencingof cloned PCRfragmentsfromthepatientand controlnuclearRNA. Theabnormalpatientalleledemonstrated the
presenceofacomplete intron 26 withaG->Atransition in the+ 1
donorposition. The patientalso hadsequencedemonstratingnormal
splicing of intron26. Thepartialsequenceofexons26 and 27 and intron26 isgiven below the figure. (B)DNAsequencingof cloned
alleles ofpatient053. The G-. Atransition thatwasobserved in the
nuclearRNAwasconfirmedinthegenomicDNA. Thepartial se-quenceofexonandintron26 isgivenbelow thefigure.
Aframeshiftorpremature stopcodon that results froma
mutation within an exon would be expected to produce the sameeffect as aretainedintron, althoughanexample of this
type of mutation has not been
reported.
All ofthe defined mutations ofCOL I A 1 orCOLI A2that inducedaframeshift did notresult inapremature stop codon before the terminal exon(30-33).Lowlevelsofal(I)mRNA(basedonreversed ratio ofa1 / a2 mRNAratio of1:2)wereobserved withaframe shift inexon 51that inactivatedthe authentic translation termi-nation codonandextended the translated producttothe first polyadenylation signal (34). Apotential
mechanism for the lowaI(I)mRNA seen in this case isanindirect effectonsplic-ing secondary to the altered
transcriptional
terminationsig-nal(35).
Infibrillarcollagen genesall theexonsthat encodethe tri-ple-helical portion of theachainarein-framecassettes of the Gly X-Y
triplet starting
with the first G of theglycine
codon and ending withthelast nucleotide of theYposition
amino acid.Thecollagen mutationsthat causeabnormalities ofsplic-ingwereinitially discovered because of
synthesis
ofshortenedachains.Theshortened chain resulted fromexon
skipping
due to amutation in eitheranintron donororacceptor site(Table
III). These mutations canresult in more severe forms of01 becausetheshortenedRNA
transcripts
are inframe,
are trans-portedto thecytoplasm, andaretranslated into afunctional although truncatedachain. The shortenedachain
isincorpo-rated into a trimeric moleculethat interferes with
secretion,
proteolytic processing, and
fibrillogenesis,
thusaccounting
for the strongdominantnatureofmutations of structural macro-molecules(36).Anexplanation for whyasimilar mutation inadonorsite results in exon skipping in the casescited above and intron retention inour case comesfrom themodelofexondefinition developed by Robbersonetal. (37) and Talerico and Berget (38).Inthis model,thefirststepisthebinding of U2RNA to a potential intronacceptor site followed by thesecond step, in which U I RNA andspecific proteins identifyand bind to the strongest
adjacent
downstream donor site(39). Thisprocess definesthe exonsand initiatesthesplicing
machinerytoexcise theintervening intronsequence.IfU1RNA identifiesan ade-quate donorwithin 300 bases (theupperlimit in size for inter-nal exons [40]), then an exon isdefined (Fig. 7 a) andthe intervening intronsarespliced. Withamutantdonorsite, ifthe intron islarge and U 1 RNA does not detect an adequate down-streamdonorsite within 300bases, no acceptable exonis de-fined, so thatit and itsflankingintrons are spliced out of the maturemessage (exonskipping; Fig. 7b).However, if the in-tron issmall, then U 1RNA willidentifytheauthentic donor site ofthesubsequent downstream intron. This definesthetwo exonsand theincluded intronas anacceptable"exon."Thus, theentire intron is retained in themature RNA(intronreten-tion; Fig.
7 c). Thismodel accountsforwhy a mutant donor sitecanresult ineitherexonskippingorintron retentionand alsopredicts
that allintronacceptorsite mutations willresult inexonskipping,because thedefiningprocess to identify an exonisinitiated fromthe acceptorsite.Approximately halfof the introns withintheaI(I) geneare
< 150bp, makingtherisk fora donorsite mutation leading to a retained intronacommonpossibility. There are 25
exon/in-tron/exon
combinations
thatare <300bp, thesizelimit for an acceptableexon if the donorsite oftheincluded intron was inactivated. 17ofthecandidate siteswouldinduce a frame shiftmutation,
while5 of the in-frame introns containanin-frame stopcodon. Thephenotype of
apatient
harboringamutation that introduces aframe shift-producing or stopcodon-con-1998 M. L. Stover, D. Primorac, S. C. Liu, M. B. McKinstry, and D. W. Rowe
A
(D Cs c 0
Cs
Exon 26
Exon27
GOT CCT OCT GOC ARA
Gly Pro Ala Gly Lys
A
M 1 2 3 4 5 6 7 8 9 M-1.38
-0.78
-0.40
I I
o
--*
0- x.. * i..209-1 ...- -3 9. 0
-- 209 1~l43 ~90
H--
3-_I
3
M 1 2 3 4 5 6 M
I I I
1
I,- 7JJ~-O
-. 40
-1.38
-0.78
-0.53
-0.404 I I 1 1
g .l [rI -I. I I IJ
I I l I _ o
go -- -0.16
1241 25 1 Z6 27 I28
-*-- 379
I
taining retained intron would beexpectedtobe similar regard-less of the site of the intron. However, variability might be observed if theexon/intron/exon size approachesorexceeds 300bases. At thatpointexon definitionwould not be
deter-mined invariably and the entire domain would be either
re-tained orskipped.Diseaseseveritywould reflect theproportion of mRNA that contained theskippedexon.
Review of published data ofcollagen splicing mutations is consistent with the exon definition model, with one possible exception (Table III). All acceptor mutations resulted inexon
skipping (49, 50),asdid donor mutations where the
exon/
in-tron/exon size(scanningdomain)was >300 nt(5, 45-48). In a number ofthesecases(49)the mRNAproductofthe mutantallelewas lower than predicted bythe model. It seemslikely thatcryptic splice sites mayhave been used andproducedsome
transcripts containing frame shift mutations that were not
transported to the cytoplasm. Since nuclear RNA was not
ex-aminedinthese cases, thispossibilityneeds further
examina-tion.
The exon definition model may require modification to accommodate an exon 14 donor site mutation in COLlAl (50). Here intron retention is predicted that would insert three in-frame stop codons and produce an OI type I phenotype. Instead, exon 14 was skipped and resulted in a lethal phenotype (Table III). Apotential explanation for thisexception isthe natureofthe mutation. This donor site mutationwaslocatedat position + 5, which is a lesshighly conserved consensus base andproducedeither normalsplicingor exonskipping.Normal splicingwasdecreased at elevated temperatures. This suggests thattheexon-definingmechanism canbedisruptedifthere is weakbinding ofU1 RNA atthe mutantdonorsite. Inthose transcriptswhere U 1 binds the mutant site but does not result in cleavage,there isinterference withsplicing of theupstream intron such that theupstream intronforms alariat with the downstream branchpointresultingin exonskipping. Coopera-tionbetweenbranchpointsand donorsitesindeterminingthe extentofexonskippinghas been observedinadifferentcontext (39, 51). Other factors suchasstrength ofcompeting donor and acceptor splice sitesand exon size also influence which splicing alternative will predominate,and may accountfor the heterogeneity of diseaseseverity observed in mildto moder-ately deforming01 that result from mutations that affect splic-ingofcollagen mRNA. Theexception observed in this case suggests that anincluded intronwillonlybe observedwhen the donorsite iscompletelyinactivatedand thescanning mecha-nismprogresses to the nextdownstream strong donorsite.
Analysis of patients with a null collagen allele has been difficultbecauseprotein datapointingtothesite ofapotential
*1i
Figure6.DirectRNase protection ofRNAfrompatient and control fibroblast strains.(A)RNAwashybridized with the intron
26-con-tainingcRNAprobederived fromcloning and sequencingexperiment discussedinFig.5 b. Twoseparatepreparations ofnuclear and
cyto-plasmicRNAfromthepatient's fibroblast (lanes 3-6)wereanalyzed.
Theprobeis 522nt(lanesIand 9) and is incompletely digestedby the RNasetreatment(lane 2). Itprotectsa442-nt fragment ofRNA from thenuclearcompartmentfromthepatient fibroblasts(lanes4
and6),whichisnotfoundinthe cytoplasmiccompartmentfromthe
samecellpreparations (lanes 3and5).RNAfromacontrolcellstrain
didnotshowthis band(lane 7, cytoplasmic; lane 8,nuclear).The
smallersize results from RNasecleavageof thepolylinker portionof thecloningvector.Theprobeisnotfully protected by normally splicedRNAand is cleaved intotwoexon-containing fragments
con-tainingexons24-26(209 nt)andexons27-28(90 nt).Thetwo
fragmentswerepresentinall lanescontaininga1(I) collagenmRNA.
Kinased RNAmarkersareinthe lanes M. Theprobeand the
ex-pectedband sizesarediagrammedbelow thefigure. (B)RNase pro-tection with the introncontaining(lanes 1-3)andfully spliced (lanes 4-6)cRNAprobes.Lane 1,introncontaining probealone;lane2, patient cytoplasmic RNA;lane3, patientnuclearRNA;lane4, spliced probe alone;lane5,totalRNAfromcontrolcells;lane 6 total
RNAfrompatientcells. Kinased RNA markersareinlanesM.A
diagramof thefully splicedcRNAisgivenbeneath thefigure.
Defective Splicing ofal(I) mRNA inOsteogenesis ImperfectaType I 1999
B
14-Table III. OutcomeofReported Splicing Mutations
Outcome Scan
Intron No. Intronmutation domain Predicted Observed Clinicalphenotype
al(I)mRNA
6(45) Donor(-1) 345 Exon6skip Exon 6skip EDS VII
14(50) Donor(+5) 213 Intron 14inclusion (+) Exon14partial skip* Lethal 01 16 (41) Acceptor Exon 17skip Exon17skip Moderate 01 21(42) Acceptor Exon 22skip Exon 22skip Severe nonlethal OI 26(This case) Donor(+ 1) 251 Intron26inclusion (+) Intron 26inclusion (+) Mild(typeI)OI
47(46) Donor (+ 1) 482 Exon 47skip Exon47skip Lethal01
a2(I) mRNA
6(47) Donor(+ 1) 749 Exon6skip Exon 6skip EDS VII 6(48) Donor(-1) 749 Exon6skip Exon 6skip EDS VII 10 (43) Acceptor Exon11 skip Exon 11 skip EDSVII/Mild01 12 (5) Donor(+2) 1399 Exon 12skip Exon 12skip Moderate OI 27(44) Acceptor Exon28skip Exon28skip Lethal01
Theintron containing an acceptor or donor site mutation is designated in columns 1 and 2. Thescanningdomain(exon/intron/exonsize)isonly
given for donor sitemutations because it is this type of mutation in which thescanningdomain appearstobeimportant.Thepredictedand actual outcomesof the acceptorordonormutationareindicated. Whetheraretained intron would containastop codonorinduceonebyaframe shift within the mature RNA isgiveas + or -.Thediseaseseverity observedas aconsequence of the mutation isgiven.EDSVIIis Ehlers-Danlos syndrome, type VII. *Thisexception is discussed in thetext.
'acceptor 5' donor 3'acceptor 5 donor 3' acceptor 5' donor a
Normal
Splicing
3' acceptor 5' cnor 3'
acceptor
qdonor 3'apetor
5'donor%% I I e0
% I I
.
a
a
c IntronInclusion
Exon defined because scan domainc300bp
lI
3' acceptor 5'cdbpor 3'acceptor 3'acceptor 5'donorb Exon skipping /
I
4 1
4% I
4 I
4.
Exon undefined because scan domain>300bp
3'acceptor5'dohr 3' acceptor 3'adceptor 5'donor
4 .
USE BLACK INK LANCASTER PRESS
Figure7.Outcome ofsplicing with a mutation that inactivates a donor site. U2 RNA (open circle) identifies the3'acceptor site and subsequently U 1 RNA(filled circle),andassociated proteins identify the downstream donor site. (a) Normal splicingoccurswhen exons areidentified and theinterveningintronsareremoved.(b)Exonskippingoccurswithadamageddonorsite. If thescanningmechanism doesnotidentifyan ade-quate donor within 300bpof the acceptorsite,theregionisnotconsideredan exonand theregionisspliced.Theoutcomeis that theexon
ad-jacenttothemutantdonorsite isskipped.(c)Introninclusionalso can occurwithadamaged donor site. If thescanningmechanism does
identifyastrongdonorsite within 300 bp of the acceptor site, then theexonis defined by thepositionof the donor site. If thenewdonorsite is thenextdownstream exon, then the entire intron is includedaspart of theenlargedexon.Ifanadequatedonor isidentified within theintron,
thencryptic splicingandinclusion ofapartialintronareobserved.
mutation are lacking. Initialcharacterizationofnuclear RNA not only suggestspotentialmechanisms to account for the un-derproduction ofmature mRNAfromthe mutant allele, it pro-videslocalizationof an unprocessed intron that might not be obtained from analysis oftotalRNA because ofthe enrichment oftheabnormal transcript in the nuclear compartment. The difficulty of detecting null-producing mutations in cDNA de-rived fromtotal RNA is illustrated by the presence of 2 abnor-malof 19 sequenced cDNAtranscriptsofthe 2.5-kb ferroche-lastase mRNA (52), < 2% of the expected cystic fibrosis trans-membrane conductance regulator transcript in bronchial epithelial cells detected by mutation specific oligonucleotide hybridization (53), and < 5% of a frame shift-producing exon skip mutation ofthe 5.3-kb lowdensity lipoprotein receptor mRNA (54). Itis likelythatframe shift mutationsthat result from insertionsordeletionswithin exons of COLlA 1 will also accountfortype I 01 but these will be hard to localize by the RNaseprotection techniques used in thiscase. PCRof frag-mentsofnuclear mRNAenriched forthepresenceofthe mu-tanttranscript followed byananalysis usingtechniques capa-bleof detectinga l-bp changewillberequiredtoidentify such cases.Discovery of mutations withintheregulatory domains of theCOLIA 1gene thatleadtoreducedtranscriptionalactivity will be an even greater challenge because the regions essential for theactivity of thisgene in collagen-producingtissues are not yetclearlydefined(55). Thesewillbeimportant abnormal-itiestoidentifybecause it is likely that subtle mutations that result in apartially null allele of theCOLlA1 gene may be a commoncauseoftype I 01 andheritable forms ofosteoporosis.
Acknowledgements
This workwassupported by National Institutes of Health grant
AR-30426. This workwasdone in partas fulfillment ofD.Primorac's
Ph.D.requirement at the University of Zagreb.
References
1.Rowe, D. W. 1991.Osteogenesis imperfecta.In Boneand Mineral
Re-search.Vol7.Ed.Johan,N. M.Heersche,andJohnA.Kanis,editors. Elsevier Science PublishersB.V., Amsterdam. 209-241.
2.Byers,P.H., andR. D.Steiner. 1992.Osteogenesis imperfecta.Annu. Rev. Med.43:269-282.
3.Steinmann, B.,V. H.Rao,A.Vogel,P.Bruckner,R.Gitzelmann,andP. H.
Byers. 1984.Cysteinein thetriple-helicaldomain ofoneallelicproductof the
aI(I)geneoftype Icollagen producesalethal form ofosteogenesis imperfecta.J. Biol.Chem.259:11129-11138.
4.Chu,M.L.,C.J.Willams,G.Pepe,J. L.Hirsch,D. J.Prockop,and F.
Ramirez. 1984. Internaldeletioninacollagengene inaperinatallethal form of
osteogenesis imperfecta.Nature(Lond.).304:78-80.
5.Chipman,S. D., J.R.Shapiro,M.B.McKinstry,M.L.Stover,P.Branson,
andD. W.Rowe. 1992.Expressionofmutanttype Icollageninosteoblastand
fibroblast culturesfromaprobandwithosteogenesis imperfectatypeIVJ.Bone Miner.Res.7:793-805.
6.Kuivaniemi,H., G. Tromp, andD.J.Prockop.1991.Mutations incollagen
genes: Causes ofrare andsomecommondiseases inhumans.FASEB(Fed.Am.
Soc. Exp.Biol.)J. 5:2052-2060.
7.Bateman,J.F.,I.Moeller,M.Hannagan,D.Chan,and W.G. Cole. 1992. Characterization of threeosteogenesis imperfecta collagen alpha 1(I)glycineto
serine mutations demonstratingaposition-dependent gradient of phenotypic
se-verity. Biochem.J.288:131-135.
8.Shapiro, J.,M. L.Stover,V.Burn,M.McKinstry,A.Burshell, S.Chipman,
andD.Rowe. 1992.Anosteopenic nonfracturesyndromewithfeatures of mild
osteogenesis imperfectaassociated with the substitution ofacysteineforglycine
attriple helixposition43 in the proaI(I)chain oftype Icollagen.J. Clin.Invest.
89:567-573.
9.Spotila,L. D., C. D.Constantinou,L. Sereda, A.Ganguly,B. L. Riggs,and
D. J. Prockop. 1991.Mutationin a gene for type Iprocollagen(COLIA2)in a
womanwithpostmenopausal osteoporosis: evidenceforphenotypicand geno-typic overlap with mild osteogenesis imperfecta. Proc. Nail. Acad. Sci. USA. 88:5423-5427.
10.Wenstrup,R. J., A. W.Shrago-Howe, L. W. Lever, C. L.Phillips,P. H. Byers, and D. H. Cohn. 1991. The effects ofdifferentcysteinefor glycine substitu-tions within alpha 2 (I) chains. Evidence of distinctstructuraldomains within the type I collagen triple helix.J. Biol. Chem. 266:2590-2594.
11. Barsh, G. S., K. E. David,and P. H. Byers. 1982. Type Iosteogenesis imperfecta:anonfunctionalallele for proa I (I)chainsof type Iprocollagen. Proc. Natl. Acad. Sci. USA. 79:3838-3842.
12.Rowe, D. W., J. R.Shapiro,M.Poirier,and S.Schlesinger. 1985.
Dimin-ished type Icollagen synthesis and reducedaI(I)collagen mRNA in cultured
fibroblasts from patients with dominantly inherited(type I)osteogenesis imper-fecta.J. Clin. Invest. 71:689-697.
13. Willing,M. C., C. J.Pruchno,M.Atkinson,and P. H. Byers.1992. Osteo-genesis imperfectatype Iis commonly dueto aCOLlAl null allele of type I
collagen. Am.J.Hum. Genet. 51:508-515.
14. Genovese, C.,and D. W.Rowe. 1987. Analysis of cytoplasmic and nuclear
mRNAin fibroblasts frompatients withtypeIosteogenesis imperfecta.Methods
Enzymol. 145:223-235.
15.Genovese,C., A.Brufsky,J. R.Shapiro,and D. W.Rowe.1989. Detection
of mutations inhuman type Icollagen RNAbyindirect RNase protection.J.
Biol. Chem. 264:9632-9637.
16. Peterkofsky,B.,andB.Dieglemann.1971.Useofamixtureof proteinase-free collagneases forthespecific assay of radioactivecollagen in thepresenceof
other proteins. Biochemistry. 6:988-993.
17. Sykes,B.,B.Puddle,M.Francis, andR.Smith.1976. Theestimationof two collagens from human dermis interrupted gel electrophoresis. Biochem.
Biophys. Res. Commun. 72:1472-1480.
18. Genovese,C., D. W.Rowe,and B. Kream. 1984.Constructionof DNA sequencescomplementaryto rata, and a2 collagen mRNA andtheiruse in
studyingtheregulation oftype Icollagen synthesisby 1,25dihydroxy vitaminD.
Biochemistry. 23:6210-6216.
19. Chu,M-L.,J. C. Myers, M. P.Bernard,J. F. Ding, and F.Ramie-z. 1982.
Cloning and characterization of five overlapping cDNAs specific for thehuman
proaI(I) collagen chain. Nucleic Acids Res. 10:5925-5933.
20.Myers, J. C., M. L.Chu, S.H. Faro, W. J. Clark, D. J.Prockop,and F.
Ramirez.1981.CloningacDNAforthepro-a2chainofhumantype I collagen.
Proc. Natl.Acad.Sci.USA. 78:3516-3520.
21. Feinberg,A. P.,andB.Vogelstein. 1983. A techniquefor radiolabeling
DNArestrictionendonucleasefragmentsto highspecific activity.Anal. Biochem. 13:6-13.
22.Myers, R. M., Z.Larin,and T.Maniatis. 1985. Detection of singlebase
substitutions by ribonuclease cleavage of mismatches in RNA:DNAduplexes.
Science (Wash. DC). 230:1242-1246.
23. Lichtler,A., N. L. Barrett,and G.G. Carmichael. 1992. Simple, inexpen-sive preparation of T1 /T2 ribonuclease suitable forusein RNaseprotection experiments. Biotechniques. 12:231-232.
24. Chow,Q.,M.Russell,D. E.Buck,J.Raymond,and W. Block. 1992. Prevention of pre-PCR mis-primingandprimer dimerization improveslow-copy
number amplification. Nucleic Acids Res. 20:1717-1723.
25.Maquat, L. E., and A. J.Kinniburgh. 1985.Af,' thalassemicfl-globin RNAthat is labile inbone marrow cellsis relatively stablein HeLa cells.Nucleic Acids Res. 13:285-2867.
26.Baserga,S.J.,andE. J.Benz,Jr.1988. Nonsense mutations inthe human
,B-globin gene affect mRNA metabolism. Proc. Natl.Acad.Sci. USA. 85:2056-2060.
27.Kuivaniemi,H.,S. Kontusaari, G.Tromp, M. Zhao, C. Sabol, and D. J.
Prockop. 1990. Identical G+ 1 to Amutations in three different introns of the
type IIIprocollagengene(COL3A1)produce differentpatternsofRNAsplicing in three variants of Ehlers-Danlos syndromeIV. Anexplanation for exon skip-ping withsomemutations andnotothers.J.Biol.Chem.265:12067-12074.
28. Urlaub, G.,P. J.Mitchell, C.J.Ciudad, andL. A.Chasin. 1989. Nonsense mutations inthedihydrofolate reductase gene affectRNAprocessing. Mol. Cell. Biol. 9:2868-2880.
29. Cheng, J., M.Fogel-Petrovic,and L. E.Maquat.1990.Translationto near thedistal end ofthepenultimateexonisrequired fornormal levelsofspliced triosephosphate isomerasemRNA.Mol. Cell. Biol. 10:5215-5225.
30.Bateman,J.F., S.R.Lamande,H. M.Dahl,D.Chan,T.Mascara,and W.G.Cole. 1989.Aframeshift mutationresultsinatruncated nonfunctional
carboxyl-terminal proalpha I(1) propeptideoftype Icollageninosteogenesis
imperfecta.J.Biol.Chem.264:10960-10964.
31.Pihlajaniem, T., L. A.Dickson, F. M.Pope, V. V. R. Korhonen,A.
Nicholls,D. J.Prockop,and J.C.Myers. 1984.OsteogenesisImperfecta: cloning
ofa pro-a2(I) collagen gene with a frameshift mutation. J. Biol. Chem. 259: 12941-12944.
with osteogenesisimperfecta whose type I collagen does not containa2( I)chains. Collagen Relat. Res. 4:389-394.
33.Chipman, S. D., H. 0. Sweet, D. J. McBride, M. Davisson, S. C. Marks, A. R.Shuldiner, R. J.Wenstrup, D. W. Rowe, and J. R. Shapiro. 1993. Defective proa2(I) collagen synthesis in a recessive mutation in mice: a potential model of humanosteogenesisimperfecta. Proc. Natl. Acad. Sci. USA. 90:1701-1705.
34. Willing, M. C., D. H. Cohn, and P. H. Byers. 1990. Frameshift mutation near the 3' endof theCOLlA 1 gene of type I collagen predicts an elongated proalpha 1(I) chain and results in osteogenesis imperfecta type I. J. Clin. Invest. 85:282-290.
35.Niwa, M., and S. M. Berget. 1991.Mutation of the AAUAAA polyadeny-lation signal depresses in vitro splicing ofproximal but not distal introns. Genes & Dev5:2086-2095.
36. Kadler, K. E.,A.Torre-Blanco, E. Adachi, B. E. Vogel, Y.Hojima, and D.J.Prockop. 1991. A typeIcollagenwith substitutionof a cysteine for glycine-748 in thea1(I) chaincopolymerizes with normal type I collagen and can
gener-atefractallikestructures.Biochemistry.30:5081-5088.
37.Robberson, B. L., G. J. Cote, and S. M. Berget. 1990. Exon definition may facilitatesplice site selection in RNAs with multiple exons. Mol. Cell. Biol. 10:84-94.
38.Talerico, M.,andS.M.Berget. 1990.Effect of 5' splice site mutations on splicing of the preceding intron.Mol. Cell. Biol. 10:6299-6305.
39.Stolow, D. T., and S.M.Berget.1991.Identificationof nuclear proteins thatspecifically being to RNAscontaining 5' splice sites. Proc. Natl.Acad.Sci. USA. 88:320-324.
40.Hawkins,J.D.1988.Asurveyofintron and exon lengths. Nucleic Acids Res. 16:9893-9908.
41.Pruchno, C. J., G.A.Wallis,B. J.Starman, A. Ayleswort, and P. H. Byers. 1989.Inefficientsplicing and production of both a normal and shortened mRNA from the same COLIA1 alleleoftype Icollagen in afather and son with osteogen-esisimperfecta(OI).Am.J.Med.Genet. 45:213a. (Abstr.)
42.Wallis, G. A.,B.J.Starman and P. H. Byers. 1989.Clinical heterogeneity of OIexplainedby molecularheterogeneityandsomaticmosaicism. Am. J. Hum. Gen. 45:228a. (Abstr.)
43. Kuivaniemi, H., C. Sabol, G. Tromp, M. Sippola-Thiele, and D. J. Prockop. 1988.A19-base pair deletion in the pro-alpha2(I)geneoftype I procol-lagen that causesin-frame RNA splicing from exon 10 to exon 12 in a proband withatypical osteogenesis imperfecta and in his asymptomatic mother.J.Biol. Chem. 263:11407-1 1413.
44.Tromp, G., and D. J. Prockop.1988. Single basemutation in the proalpha 2(I) collagen gene that causes efficient splicingof the RNA from exon 27 to exon
29 and synthesis of a shortened but in-frame proalpha 2(I) chain. Proc. Natl. Acad. Sci. USA. 85:5254-5258.
45. Weil, D., M. D'Alessio, F. Ramirez, W. DeWet, W. G. Cole, D. Chan, ana J. F. Bateman. 1989. A base substitution in the exon of a collagen gene causes alternative splicing and generates a structurally abnormal polypeptide in a patient with Ehlers-Danlos syndrome typeVII. EMBO (Eur. Mol. Biol. Organ) J. 8:1705-1710.
46. Byers, P. H. 1990. Brittle bones:fragilemolecules;Disorders of collagen genestructureandexpression. Trends Genet. 6:293-300.
47.Weil, D., M.D'Alessio, F.Ramirez, B. Steinmann, M. K.Wirtz,R. Glan-ville, and D. W.Hollister. 1989.Temperature-dependent expressionofacollagen splicing defect in the fibroblasts ofapatient with Ehlers-DanlossyndromeType VII. J. Biol.Chem.264:16804-16809.
48.Weil, D., M.D'Alessio, F. Ramirez, and D. R. Eyre. 1990. Structural and functionalcharacterizationofasplicingmutation in thepro-alpha 2(I) collagen
gene ofanEhlers-Danlos type VIIpatient.J.Biol. Chem. 265:16007-16011. 49. Byers, P. H., G. A.Wallis, and M. C. Willig. 1991.Osteogenesis imper-fecta: translation of mutationtophenotype.J. Med. Genet. 28:433-442.
50.Bonadio,J., F.Ramirez,and M. Barr. 1990. An intron mutation in the humanalphaI(I) collagen gene alters theefficiency of pre-mRNAsplicingand is
associatedwithosteogenesisimperfectatype II. J. Biol. Chem. 265:2262-2268. 51.Dominski,Z., and R. E. Kole. 1992.Cooperationofpre-mRNAsequence elements insplicesite selection. Mol. Cell. Biol. 12:2108-2114.
52.Nakahashi, Y., H.Fujita, S.Taketani,N.Ishida,A.Kappas,andS. Sassa. 1992. The molecular defect of ferrochelatase in apatientwitherythropoietic
protoporphyria. Proc. NatI. Acad. Sci. USA. 89:281-285.
53. Hamosh, A., B. C.Trapnell,P. L.Zeitlin, C.Montronse-Rafizadeh,B.J. Rosenstein, R. C. Crystal, and G. R.Cutting. 1991. Severedeficiencyofcystic
fibrosis transmembrane conductance regulator messengerRNAcarrying non-sensemutationR553XandW1316X inrespiratory epithelialcells ofpatients
with cystic fibrosis. J. Clin. Invest.88:1880-1885.
54.Koivisto, L-M., H.Turtola,K.Aaltoj-Setala,B.Top,R. R.Frants,P.T. Kovanen,A.C. Syvanen, andK.Kontula. 1992. The familial
hypercholesterol-emia(FH)-North Karelia mutation of the lowdensitylipoproteinreceptor gene deletessevennucleotidesofexon6 and isacommon causeof FH in Finland. J. Clin.Invest.90:219-228.
55.Pavlin, D., A. C. Lichtler, A. Bedalov, B. E. Kream, J. R. Harrison, H. F. Thomas, G. A. Gronowicz, S.H.Clark, C.0.Woody,and D. W. Rowe. 1992. Differential utilization of regulatory domains within the a 1 (I) collagen promoter in osseous andfibroblastic cells. J. Cell Biol. 116:227-236.