Variable Bone Fragility Associated With an Amish COL1A2 Variant and a Knock-in Mouse Model

15 Pages·2010·0.42 MB·English

JJBMR
ORIGINAL ARTICLE
Variable Bone Fragility Associated With an Amish
COL1A2 Variant and a Knock-in Mouse Model
Ethan Daley,1 Elizabeth A Streeten,2 John D Sorkin,3 Natalia Kuznetsova,4 Sue A Shapses,5
Stephanie M Carleton,6 Alan R Shuldiner,2,3 Joan C Marini,4 Charlotte L Phillips,6 Steven A Goldstein,1
Sergey Leikin,4 and Daniel J McBride Jr2
1OrthopaedicResearchLaboratories,DepartmentofOrthopaedicSurgery,UniversityofMichigan,AnnArbor,MI,USA
2DivisionofEndocrinology,Diabetes&Nutrition,UniversityofMarylandBaltimore,Baltimore,MD,USA
3DivisionofGerontology,UniversityofMarylandBaltimoreandtheBaltimoreVAMedicalCenter,GeriatricResearch,Educationand
ClinicalCenter(GRECC),Baltimore,MD,USA
4NationalInstituteofChildHealthandHumanDevelopment,NationalInstitutesofHealth,Bethesda,MD,USA
5DepartmentofNutritionalSciences,RutgersUniversity,NewBrunswick,NJ,USA
6DepartmentofBiochemistry,UniversityofMissouri–Columbia,Columbia,MO,USA
ABSTRACT
Osteogenesis imperfecta (OI) is a heritable form of bone fragility typically associated with a dominant COL1A1 or COL1A2 mutation.
VariablephenotypeforOIpatientswithidenticalcollagenmutationsiswellestablished,butphenotypevariabilityisdescribedusingthe
qualitativeSillenceclassification.PatterninganewOImousemodelonaspecificcollagenmutationthereforehasbeenhinderedbythe
absenceofanappropriatekindredwithextensivequantitativephenotypedata.WebenefitedfromthelargesibshipsoftheOldOrder
Amish(OOA)todefineawiderangeofOIphenotypesin64individualswiththeidenticalCOL1A2mutation.Stratificationofcarrierspine
(L1–4)arealbonemineraldensity(aBMD)Z-scoresdemonstratedthat73%hadmoderatetoseveredisease(lessthan(cid:1)2),23%hadmild
disease((cid:1)1to(cid:1)2),and4%wereintheunaffectedrange(greaterthan(cid:1)1).Alineofknock-inmicewaspatternedontheOOAmutation.
Bone phenotype was evaluated in four F lines of knock-in mice that each shared approximately 50% of their genetic background.
1
Consistentwiththehumanpedigree,thesemicehadreducedbodymass,aBMD,andbonestrength.Whole-bonefracturesusceptibility
wasinfluencedbyindividualgenomicfactorsthatwerereflectedinsize,shape,andpossiblybonemetabolicregulation.Theresults
indicatethattheG610COI(Amish)knock-inmouseisanoveltranslationalmodeltoidentifymodifyinggenesthatinfluencephenotype
andfor testingpotential therapies forOI. (cid:1) 2010AmericanSociety for Boneand MineralResearch.
KEYWORDS: OSTEOGENESISIMPERFECTA;BONE;COLLAGEN;KNOCK-IN;RODENT
Introduction COL1A2mutationshavebeenidentifiedinOIpatients,withmany
examples of recurrent mutations in unrelated individuals.(3)
Osteogenesis imperfecta (OI) is a heritable form of bone However, there are no large pedigrees with quantitative
fragility with an overall spectrum of disease severity that phenotype datafor OIpatients with theidenticalmutation.
ranges from a perinatal lethal form to mild disease that can The predominant type I collagen isotype in the extracellular
remain clinically silent.(1–3) Multiple fractures is the principal matrix(ECM)isaheterotrimericmoleculecomprisedoftwoa1(I)
clinicalpresentationthatbringsOIpatientstomedicalattention. chains and one a2(I) chain. A homotrimeric isotype of type I
Additional phenotype traits can include bone deformity (e.g., collagen,comprisedofthreea1(I)chains,isalsofoundintissues
long bone bowing), short stature, blue sclerae, dentinogenesis insmallamountsinthenonpathologicstate.Thehomotrimeric
imperfecta(DI),andhearingloss.SillenceoriginallyclassifiedOI isotypeisassociatedwithautosomalrecessiveconnectivetissue
into types I to IV on the basis of clinical presentation, disordersinhumanswhennoa2(I)chainsoronlynonfunctional
radiographic features, and mode of inheritance.(4,5) OI types I a2(I) chains are synthesized.(6,7) The amino acid sequences of
toIVtypicallyareassociatedwithheterozygousmutationsinthe both a1(I) and a2(I) chains are homologous and have a
genes(COL1A1andCOL1A2)thatencodetheachainsoftypeI characteristic [Gly-X-Y]338 repeat within their respective triple-
procollagen molecules. Currently, more than 800 COL1A1 and helical domains. The small side chains of glycine residues
ReceivedinoriginalformMarch18,2009;revisedformMay20,2009;acceptedJuly6,2009.PublishedonlineJuly13,2009.
Addresscorrespondenceto:DanielJMcBride,Jr.,642MontgomeryWoodsDrive,Hockessin,DE19707,USA.E-mail:[email protected]
JournalofBoneandMineralResearch,Vol. 25,No.2,February2010,pp247–261
DOI:10.1359/jbmr.090720
(cid:1)2010AmericanSocietyforBoneandMineralResearch
247
positioned at every third position of the repeat are sterically (GGT)toacysteine(TGT)codon.(24)Wealsoreportthecreationof
required for helix formation, and mutations in these glycine the first glycine-substitution knock-in Col1a2 mouse that
residuesaccountfor80%oftheknowntriple-helicalmutations replicates the gene and protein phenotypes observed in the
thatcause moderatelysevere tolethal formsof OI. OOAkindred.Theknock-inG610COI(Amish)mouserepresentsa
Over the past two decades, several laboratories have uniquetranslationalmodelforevaluatingthemolecularbasisof
developedmurineOImodelsbasedonmutationsintheproa1(I) phenotypevariabilityandtotestpotentialtherapeuticstrategies
chain of type I collagen patterned after those described in forOI.
humanOItoextendourunderstandingofOIpathobiology.All
the existing mouse models have a moderate-to-severe bone
phenotypewithimpairedviability.Thetwoearlymodelsofnote Materials and Methods
are the Col1a1 null allele Mov-13 mice(8–10) and the ‘‘protein
suicide’’ or collagen minigene model.(11–13) The homozygous Ethical considerations
Mov-13 mouse produces no type I collagen, and this is lethal.
Allhumansubjectresearchwasreviewedandapprovedbythe
HeterozygousMov-13micedepositreducedamountsofnormal
Institutional Review Board of the University of Maryland
typeIcollagenintheECMthatisassociatedwithanosteopenic
Baltimore (UMB). Informed consent was obtained for all study
phenotypeandabnormalskeletalbiomechanics.Theminigene
participants.Allanimalprocedureswerereviewedandapproved
modelconstructexpressesaversionofthehumanCOL1A1gene
by theUMBInstitutional Animal Careand UseCommittee.
that is missing the central 41 exons that encode a significant
portion of the triple-helical domain. The shortened human
proa1(I) chains associate with normal murine proa1(I) and
Recruitment of family members and phenotype
proa2(I)chains,depletingtheamountofnormaltypeIcollagen
assessment
and altering ECM organization. Most recently, the Cre/lox
recombination system was used to develop a Col1a1 G349C The study had three stages of informed consent: (1) mutation
substitution to reproduce a human OI type IV (moderately screening, (2) clinical assessment, and (3) skin-biopsy harvest.
severe) phenotype termed BrtlIV.(14–17) The BrtlIV mouse The screening enrollment age requirement of 6 years or older
represented a major advance in OI mouse models because it was waived if one parent was mutation-positive by sequence
istheonlyCol1a1 modelwith atriple-helicalglycinemutation. analysis. After obtaining informed consent (or assent from the
The OI phenotype pattern of proa2(I) glycine substitutions children), buccal cells were harvested as a source of genomic
appears to be different from proa1(I) substitution.(3) Generally, DNA (gDNA) for genotyping. All gDNA- and cDNA-derived
proa2(I) mutations are less severe. They also may represent sequence data (see below) were analyzed, using Sequencher
an easier target for treatment because only 50% of collagen software (GeneCodes, AnnArbor, MI,USA).
moleculesareaffectedinCOL1A2heterozygotesversus75%in All volunteers verifiedbygDNAsequence analysistocarry a
COL1A1 heterozygotes. However, the only reported mouse singlecopyoftheTallele(OIgroup)andtheircontrol(Gallele)
modelforOIwithaCol1a2mutationisosteogenesisimperfecta- nuclear family members (age(cid:2)6 years) were invited to
murine(oim).Homozygousoimmiceexhibitskeletaldiseasewith participate in the clinical assessment and skin-biopsy phases.
clinical and biochemical features of Sillence type III (severe The phenotype assessment consisted of a medical and family
progressive) OI. The oim mutation and its biologic conse- history, physical examination, blood draws for serum procolla-
quences(18–21)arestrikinglysimilartothosefoundinaGerman gen type I C-terminal propeptide (PICP) and CrossLaps
child with type III OI.(6,22,23) In both instances, a frame-shift measurements (markers of bone formation and resorption),
mutation in the region of the COL1A2 gene that encodes the andmeasurementofarealbonemineraldensity(aBMD).Lumbar
terminal portion of the proa2(I) C-propeptide predicts the spine (L1–4) and proximal femur (total and subregions) aBMD
synthesis of nonfunctional proa2(I) chains. Despite being the wasmeasuredbydual-energyX-rayabsorptiometry(DXA)ona
onlynaturallyoccurringOImousemodel,theoimmousehasa QDR 4500W instrument (Hologic, Bedford, MA, USA), and the
rarely observed type of mutation and an atypical biochemical results reviewedby a singleclinician (EAS) and another author
phenotypebecauseonlyproa1(I) homotrimericmoleculesare (DJM). The in vivo coefficient of variation for replicate aBMD
3
formed. measurementswas0.8%forthelumbarspine,0.9%fortotalhip,
The absence of a COL1A2 murine OI model with autosomal and1.5%forthefemoralneck.TheHologicPediatricReference
dominantinheritance,mildtomoderatedisease,andphenotype Curve Option (hip: 5 to 20 years; spine: 3 to 20 years) and
variation represents a major gap in our tools to understand OI Caucasian reference (age>20 years) were used to obtain age-
pathobiology and to develop effective treatment. Defining the specific Z-scores. Since OI is frequently associated with short
molecularbasisofphenotypevariationalsohasbeenlimitedby stature and bone mineral content (BMC) and aBMD are
the absence of large numbers of OI patients carrying the influenced by bone size, volumetric bone mass [bone mineral
identicalcausativemutationwithquantitativephenotypedatato apparent density (BMAD)] also was estimated to minimize the
useasaprototypeforanewOImousemodel.Herewereportthe influence of small bone size.(25,26) A subset of the phenotype
clinical and biochemical phenotype of a COL1A2 gene variant assessment volunteers also provided 3mm dermal punch
found among a Lancaster County, Pennsylvania, Old Order biopsies from the posterior upper arm from which fibroblasts
Amish (OOA) kindred. This COL1A2 variant is a G-to-T werederivedandstoredatUMBand/orsubmittedtotheCoriell
transversion at nucleotide 2098 that alters the gly-610 codon Institute (OIfamilies 1997,1998, 2187, and 2229).
248 JournalofBoneandMineralResearch DALEYETAL.
Generation of knock-in mice glycerol buffer (pH 7.5) were recorded in a Nano II differential
scanning calorimeter (Calorimetry Sciences Corp. Lindon, UT,
Knock-inmicewerecreatedusingembryonicstemcellsandCre/
USA) at several different scanning rates. The apparent melting
loxPtechnology.A13kbgenomicclonecontainingtherelevant
temperature T was defined at the maximum of the melting
segmentofthemurineCol1a2genewasisolatedfromalphage m
peak after baseline subtraction.
library constructed with gDNA of the 129Sv/Ev Taconic mouse
Total RNA was extracted from fibroblast cultures, and cDNA
strain. PCR site-directed mutagenesis was used to change the
was synthesized using RACE1 [50-GAT GGA TCC TGC AGA AGC
targeted codon from GGT to TGT. The Col1a2 G610C-positive
(T)17-30]. An aliquot of cDNA and forward exon primer and
embryonicstemcellswereinjectedintoblastocystsofC57BL/6J
downstreamreverseexonprimer(selectedtocrossintron-exon)
(B6)mice.Foundermicethatretaintheneotargetingvectorare
boundaries were used for second-strand synthesis and PCR
termedneoþorG610CNeomice.Progenyobtainedbybreeding
amplification.PCRproductsoftheexpectedsizewereconfirmed
a founder male with a female that expressed Cre recombinase
by agarose gel size separation prior to purification for direct
(JacksonLaboratory,BarHarbor,ME,USA,StockNumber003724)
nucleotide sequencing.
aretermedneo–orG610COImice.TheG610COImouselineis
available to the research community through the Jackson
Phenotype assessment of F G610C OI Mice
Laboratory Mouse Repository (JaxStock Number:007248). 1
FourF strainsofG610COImicewereusedtodeterminethe aBMD measurements
1
effectofgeneticbackgroundonphenotypein2-month-oldmale
The aBMDs of right femurs were measured by DXA (GE-Lunar
mice. Incipient congenic G610C OI B6 ((cid:3)98% B6 genetic
densitometer, PIXImus; software version; 2.10, GE Medical
background)malebreedersweregeneratedbysixgenerations
Systems, WI, USA). Femur measurements were obtained by
of backcrosses to Jackson Laboratory B6 mice (Stock Number
placing an excised femur on a Delrin block. The percent
000664). Experimental heterozygous B6 male breeders were
coefficientofvariation(CV)was1.3%forBMCand1.4%forBMD.
crossed with A/J (Stock Number 000646), BALB/cByJ (Stock
Number001026),C3H/HeJ(StockNumber000659),andFVB/NJ
Confocal Raman microspectroscopy
(Stock Number 001806) females purchased from the Jackson
Laboratory. Progeny of these crosses are designated, respec- Segments of cortical bone from the femur diaphysis were
tively,asA.B6,Cby.B6,C3.B6,andFVB.B6.Experimentalmicewere embeddedinCryo-gel(Instrumedics,Inc.,St.Louis,MO,OSA)and
housedatUMBinasinglespecific-pathogen-freeroomandwere cutlongitudinallynormaltothebonesurfaceonaCryo-CutOne
exposedtoidenticalenvironmentalconditionsconsistingofa12 cryostat(Vibratome,Bannockburn,IL,USA).The30mmsections
hour light/dark cycle, an ambient temperature of 238C, and were examined with a Senterra confocal Raman microscope
adlibitumaccesstowaterandlaboratorymousechow.Genotype (Bruker Optics, Billerica, MA, USA). Raman spectra from 40 to
wasassignedusingaPCRassaythatcandiscriminatethethree 4500cm(cid:1)1werecollectedwith9to18cm(cid:1)1resolutionusinga
possible G610C OI mouse genotypes. The forward primer (TCC circularly polarized 20mW (9mW at the sample), 532nm laser
CTGCTTGCCCTAGTCCCAAAGATCCTT)andthereverseprimer beam, X40/0.95 NA objective, and a 50mm confocal pinhole.
(AAGGTATAGATCAGACAGCTGGCACATCCA)willgeneratea Thespectrawereacquiredat12pointsvisuallyselectedoutside
165bp (wild type) or a 337 and a 165bp (heterozygous) or a osteocytesalongtheperiostealtoendostealaxisinthemidplane
337bp(homozygous)PCRproductusingG610COImicegDNA. of the sections. The accumulation time for each point was
All animalswere euthanizedby CO asphyxiation. 1minute.Themineralcontentofbonematrix(PO :CH),collagen
2 4
contentofbonematrix(amideIII:CH),andmineral:collagenratio
Human and mouse collagen analysis (PO :amide III) were determined from the areas under n -PO
4 1 4
(integratedfrom901to987cm(cid:1)1),amideIII(1214to1305cm(cid:1)1),
Pepsin-soluble collagen was purified from mouse tail tendons
and CH (2824 to 3035cm(cid:1)1) Raman peaks. Relative collagen
andhumanfibroblastculturemediumforanalysisbySDS-PAGE
contentfromPro:CHratio(833to867cm(cid:1)1)wasconsistentwith
and differential scanning calorimetry (DSC). Purified collagen
thatevaluatedfromtheamideIII:CHratiobutlessaccurateowing
samples were labeled by fluorescent monoreactive Cy5 NHS
to low intensityof thePro peak.
ester(GEHealthcareBio-SciencesCorp.,Piscataway,NJ,USA)or
double-labeledbyCy5andfluorescentBODIPY-L-Cys(Molecular
Femur mCT
Probes,Invitrogen,Inc.,Carlsbad,CA,USA).(27)Labeledproteins
weresize-separatedbySDS/PAGEonprecast3%to8%gradient Micro-computed tomography (mCT) was used to quantify
Tris-acetateminigels(Invitrogen).Followingelectrophoresis,the femoralgeometry,morphology,andmineralization(GEMedical
gelswerebrieflyrinsedindistilledwaterandscannedonaFuji Systems,London,ON,Canada).FemursfromF male2-month-
1
FLA5000fluorescencescanner(FujiMedicalSystems,Stamford, old mice were scanned over 200 degrees of rotation and
CT,USA)usinga473nmexcitationlaserforBODIPY-L-Cysanda reconstructedon18mmvoxels.Corticalandtrabecularanalyses
635nm laser for Cy5. For DSC, collagen was redissolved (1 to includedmeasuresofbonemineralcontentanddensity,cortical
2mg/mLfinalcollagenconcentration)in2mMHCl(pH2.7)orin andtrabecularthickness,andtrabecularspacing.Corticalregions
0.2M sodium phosphate/0.5M glycerol (pH 7.5) and then of interest (ROIs) consisted of cylindrical segments (radius
dialyzed extensively against the same buffer to remove excess 1.35mm,height3mm)ofthemiddiaphyses.TrabecularROIsalso
salt.DSCthermograms(28)ofapproximately0.1mg/mLofpepsin- consistedofcylindricalsegments(radius0.57mm,heightequal
treated collagen in 2mM HCl (pH 2.7) and in the phosphate/ to10%oftotalfemurlength)ofthedistalfemoralmetaphyses,
AMISHOIMUTATION JournalofBoneandMineralResearch 249
proximaltothegrowthplates.Separatethresholdswereapplied
to the cortical and trabecular ROIs (2000 and 1200 HU,
respectively).
Ex vivo mechanical testing
Femursweremechanicallytestedatroomtemperatureusinga
four-point bending apparatus (858 Mini-Bionix, MTS, Eden
Prairie,MN,USA)inamannerdescribedpreviouslybyJepsen.(10)
With posterior femoral surfaces kept in tension, displacement
wasappliedat0.05mm/s.Theupperandlowertinesofthefour-
pointsupportwereseparatedby6.35and2.2mm,respectively.
Custom software was used to determine stiffness, yield, maxi-
mum load, and pre- and postyield fracture energy. Standard
beam theory was used to calculate yield stress and Young’s
modulus based on the mechanical testing data and femoral
geometryasmeasured by mCT.
Fig.1.OOA COL1A2 G610C putative founder couple pedigree. The
Statistical analysis
COL1A2 variant is a G-to-T transversion at nucleotide 2098 that alters
Statistical analysis was performed using InStat and Prism thegly-610codon(GGT)toacysteine(TGT)codon.ThemutantTallele
statistical software packages (Graphpad, La Jolla, CA, USA) or has been mapped in 64 heterozygous descendants of the putative
SASstatisticalsoftware(SASInstitute,Cary,NC,USA).Atwo-tailed foundercouplebornapproximately150yearsago.Fourofthefounder
couple’soffspringhaveanestimated800livingdescendants,whereas
p value of .05 or less for any single analysis was considered
statistically significant. Outliers (defined as values exceeding(cid:4) theotherfiveoffspringproducednoprogeny.(Controlgenotype¼GG;
OIgenotype¼GT.)
2sfromthegroupmean)wereremovedfromthemCTandfour-
pointbending datasets.
Results an average of approximately 6 individuals. The overall sibship
genotype distribution exhibited a normal Mendelian 1:1 ratio
OOA kindred genotyping/pedigree structure expectedforoffspringfromonecarrierparentandonewild-type
parent. No homozygous individuals and no carrier(cid:5)carrier
Putative founder couple identification
coupleswere identified.
A54-year-oldwomanwithahistoryoffracture,lowhipandspine
aBMD,andadifferentialdiagnosisofidiopathicosteoporosisor OOA kindred phenotype
OI was identified in the Amish Family Osteoporosis Study
(AFOS).(29,30) A brother and a niece of the woman also were Standing height
identified with a history of fracture and low aBMD. gDNA was
Standingheightasafunctionofage(Fig.2)suggestedthatOI
extracted from blood of the proband’s brother for analysis of
family members tended to be shorter than their unaffected
COL1A1 and COL1A2 genes.(31–33) Nucleotide sequencing
familymembers.StandingheightwasconvertedtoZ-scoresto
identified a G-to-T substitution that converts the triple-helical
better assess the effect of the T allele across the wide age
codonforglycine-610(GGT)tocysteine(TGT).Sequenceanalysis range of study participants. OOA-specific standing height
confirmedthattheproband,brother,andniececarriedasingle Z-scores were generated separately for women and men from
copyofthemutantTallele.ThreeadditionalAFOSDNAsamples
height data obtained from 1687 females and 1416 males
(a grandmother, mother, and granddaughter) were found (age>20 years) enrolled in non-OI research studies (unpub-
subsequentlyto carrysingle copies ofthevariant Tallele.
lishedUMBdata).SinceOOA-specificdatawereunavailablefor
The Anabaptist Genealogy Database (AGDB4)(34,35) was
studyparticipants6to20yearsofage,standingheightZ-scores
queriedusingPedHunter(36)software tolinkthesixindividuals
were generated using the National Health and Nutrition
withthemutantTalleletoaputativefoundercouple(Fig.1)born
Examination Survey (NHAHES) data (https://web.emmes.com/
approximately 150 years ago. AGDB genealogic records listed study/ped/resources/htwtcalc.htm).MeanstandingheightZ-score
approximately 1200 descendents from four of the putative was significantly less (p<.001) for carriers of the T allele as
founder couple’s children, of which approximately 800 are
compared with control family members but with appreciable
estimated to be alive today. Genotype status was obtained for
overlap in Z-scores between control and OI study participants
195 study participants; 149 were direct descendants of the
(see Fig.2).
putativefoundercouple,19weredirect-descendantmarried-in
spouses, and 27 were descendants of the putative founder
Biomarker measurements
couple’ssiblingsormoredistantrelatives.ThevariantTallelewas
identifiedin64descendantsin22sibshipsranginginagefrom PICPvaluesforOIfamilymemberswererelativelyconstantasa
3weeksto86years.Sibshipsrangedfrom2to11individualswith function of age (see Fig. 2). However, PICP for control family
250 JournalofBoneandMineralResearch DALEYETAL.
Fig.2.OOAkindredstanding-heightandserumbiomarkermeasurements.Height(leftfourpanels):Male(left)andfemale(right)carriers(GT)ofthe
COL1A2G610Callele(OI)tendtohavereducedstandingheightcomparedwithcontrol(GG)familymembers.StandingheightZ-scoresforindividuals
youngerthan20yearsofagewerecalculatedusingNHANESnormativedata,whereasanOOA-specificZ-scorewascalculatedforindividualsolderthan
20yearsofage.ThemeanOIZ-scoreforstandingheightwassignificantlyreduced(p<.0001)comparedwithfamilycontrolsforbothagegroups.
Biomarkers (right four panels): PICP (left) and CrossLaps (right) are serum biomarkers associated with collagen metabolism. PICP reflects collagen
biosynthesisinbone.PICPvaluesexhibitedvariationswithageanddiseasestatus.TheparticipantsheterozygousfortheG610CCOL1A2allele(OI)who
wereyoungerthan20yearsofageexhibitedsignificantlyreducedPCIPascomparedwithnormalcontrolfamilymembers.CrossLapslevelsreflectthe
degradationofbonecollagenandcanbeconsideredasurrogateforboneresorption.Noobviousassociationwithageordiseasestatuswasobservedfor
theCrossLapsdata.
members younger than 20 years of age were markedly higher Genome Informatics (MGI) ID 3805459] contains 2712 nucleo-
compared with OI family members. Genotype (p<.0001) and tides, whereas the wild-type B6 allele contains 730
age(p<.0086)werefoundtomakesignificantcontributionsina nucleotides.F foundermicecDNAsequenceanalysisconfirmed
1
multiple regression model using age, sex, and genotype as expression of mutant and wild-type mRNA. Offspring carrying
variablesforPICP.Incontrast,onlyage(p<.0001)wasfoundto one or two copies of the knock-in Col1a2tm1Mcbr allele from F
1
make asignificant contribution to serum CrossLaps. founder breeding survived without any detectable lethality at
birth orbeyond weaningand werefertile.
BMD measurements GenomicDNAsequenceanalysisconfirmedtheCre-mediated
excisionoftheFloxedneocassetteinanF malemousethatwas
2
subsequently used to sire G610C OI mice progeny on a B6
Lumbarspine(L1–4)andfemoralneck(FN)BMADbyageshown
background. Intron 33 in G610C OI mice is comprised of 852
formalesandfemalessuggestedthattheOIgrouphasreduced
nucleotides,anditsMGIalleledesignationisCol1a2tm1.1Mcbr(MGI
BMAD compared with family controls. Age, sex,genotype, and
AccessionIdentifier3711122).Itdiffersinsizefromthewild-type
weight were used as predictor variables in multiple regression
models for lumbar spine and FN BMAD. Sex (p<.03) and B6 sequence by the addition of 144bp of residual targeting
genotype (p<.0001) made significant contributions to lumbar vectorsequencethatflankedtheloxPsitesandthelossof22bp
spine and FN BMAD. Weight (p<.0001) also contributed to of B6 sequence (GenBank Accession Identifier DQ377844).
lumbar spine BMAD, and age (p<.0001) contributed to FN Heterozygous(cid:5)wild-type breeding produced litters with both
expectedgenotypes.However,noviablehomozygousoffspring
BMAD.BMDassessedasZ-scoresforthespineandhipareshown
wereobtainedfromheterozygousmatings.Genotypeanalysisof
inFig.3.MeanZ-scoredifferencesbetweentheOIandcontrol
deadpupsrecoveredwithin24hoursofbirthdiddemonstrate
groups at the lumbar spine and the FN were significant
(p<.0001). The mean Z-score at the lumbar spine was thepresenceofpupshomozygousfortheCol1a2tm1.1Mcbrallelein
(cid:1)2.63(cid:4)0.99fortheOIgroupand(cid:1)0.24(cid:4)0.82forthecontrol the litters.
group.MeanFNZ-scoreswere(cid:1)1.18(cid:4)0.91fortheOIgroupand
0.26(cid:4)0.78for thecontrol group. Human and mouse SDS-PAGE analysis
DoublelabelingwithCy5andBodipy-L-CysrevealedintenseCys
Production of the knock-in mice
staining of a2(I) chains from all mutant animals and proband
Nucleotide gDNA sequence (GenBank Accession Identifier collagens(Fig.4A,B),indicatingthepresenceofexposedreactive
DQ377843) analysis demonstrated that F founder mice Cys-SH residues ina2(I)chains oftype Icollagentriplehelices.
1
(male¼1, female¼2) were heterozygous for the knock-in Interestingly,SDS-PAGEalsorevealedthepresenceofreducible
mutationinCola2exon33(equivalenttoCOL1A2exon35).Intron a2-Cys-S-S-Cys-a2dimersintendonsofallmutantanimals(see
32 for the F founder knock-in allele [Col1a2tm1Mcbr; Mouse Fig.4C,D).SincetypeIcollagentriplehelixcontainsonlyonea2
1
AMISHOIMUTATION JournalofBoneandMineralResearch 251
Fig.3.OOAkindredbonemineraldensity(DXA)measurements.TheprincipalquantitativemeasureofphenotypeseveritywasaBMDusingDXA.BMAD
wascalculatedfromL1–4spine(A,B)andfemoralneck(C,D)DXAaBMDmeasurementsasanestimateofvolumetricbonedensity.Bothmale(left)and
female(center)Talleleindividuals(OI)tendedtohavereducedBMADcomparedwithfamilycontrolsasafunctionofage.MeanZ-scoresusingHologic
normativedatearereducedinOIfamilymembersascomparedwithcontrolsinboththespine(E)andhip(F).
chain,suchdimersmustformbetweentwodifferentmolecules. thatofnormalheterotrimers.(28,37)Thus,totestwhetherthehigh
The intermolecular dimers likely form prior to fibrillogenesis a1(I):a2(I) ratio was consistent with homotrimer formation, we
becauseinthefibrilstheCysresiduesofadjacenttriplehelices performed DSC scans of collagen from all mouse tendons and
areaxiallyseparatedbyatleastoneDperiod.Dimerformation humanfibroblastculturesin2mMHCl(pH2.7).Inadditiontothe
likely occurs in the Golgi stack by nonspecific side-by-side denaturation peak of normal type I heterotrimers, the DSC
interactionsofprocollagenmolecules.Surprisingly,theaberrant thermograms of collagen from neoþ animals (see Fig. 4E) did
intermolecular dimers are incorporated into the tissues of reveal a low-temperature peak consistent with mutant hetero-
mutantanimals.Fromtheintensitiesofthebandscontainingthe trimers and a high-temperature peak. The high-temperature
S-S dimers and corresponding bands without the dimers, the peakwasidenticaltooimandartificialhomotrimersmeasuredat
estimatedfractionofmutantchainsinvolvedinthedimerswas thesameconditions.(28,38)Thispeakwasabsentincollagenfrom
approximately1%inheterozygousneoþ,approximately1.5%in neo–animals(seeFig.4F)andfromhumanprobandfibroblasts
heterozygousneo–,andapproximately9%inhomozygousneoþ (seeFig.4G).DeconvolutionoftheDSCthermogramssuggested
animals(Table1). approximately 24% and approximately 62% content of homo-
SDS-PAGE analysis also demonstrated an abnormally high trimers in heterozygous and homozygous neoþ animals,
ratioofa1(I):a2(I)fluorescenceintensitiesinneoþ(Col1a2tm1Mcbr respectively, inagreementwith SDS-PAGE.
allele) animals, approximately 3:1 in heterozygous neoþ and To evaluate the effect of the G610C substitution on the
approximately 8:1 in homozygous neoþ versus approximately thermal stability of mutant collagen, we measured similar DSC
2:1 in wild-type and heterozygous neo– (Col1a2tm1.1Mcbr allele) thermogramsin0.2Msodiumphosphateand0.5Mglycerol(pH
animals.Mostlikelythiswascausedbyinsufficientsynthesisof 7.5), which can be used to evaluate the T under physiologic
m
mutant a2(I) chains and formation of a1(I) homotrimers [e.g., conditions by subtraction of 1.78C. From the buffer-corrected
owing to partial degradation and/or slower transcription, thermograms(seeFig.4H–J),weestimatethattheincorporation
splicing, and/or translation of the neoþ (Col1a2tm1Mcbr) mRNA ofthemutanta2(I)chainresultsinaT reductionby2.38Cinmouse
m
containing the intron 33 targeting vector]. Based on this andapproximately18Cinhumancollagenscorrespondingly.
assumption,weestimatedthehomotrimer contentasapproxi-
mately23%andapproximately67%intendonsofheterozygous F G610C OI mouse phenotype
1
andhomozygous neoþ animalscorrespondingly (see Table1).
Breeding
Atotalof301experimentalmicewereobtainedfrom48litters.
Human and mouse DSC analysis
Genotypewasdeterminedforthemice,exceptforoneB6.Cby
Previousdifferentailscanningcalorimetry(DSC)studiesshowed femalemouse.MeanlittersizesdifferedamongthefourF stains
1
that the denaturation temperature T of type I collagen (one-wayANOVAp¼.0039).Themeanlittersizeswere7.9(cid:4)1.9
m
homotrimers at acidic pH is approximately 2.58C higher than (FVB.B6),6.5(cid:4)2.5(Cby.B6),5.0(cid:4)2.2(C3.B6),and5.0(cid:4)1.5(A.B6).
252 JournalofBoneandMineralResearch DALEYETAL.
Fig.4.TypeIcollagenanalysis.TypeIcollagenfromhumanfibroblastculturesandfrommousetailtendonswasanalyzedbySDS-PAGE(A–D)andDSC(E–
J).G610CgenotypestatusisrepresentedasGG(control),GT(heterozygous),andTT(homozygous).NeoþrepresentstheCol1a2tm1McbralleleinG610C
mice, and neo– represents the G610C OI mouse Col1a2tm1.1Mcbr allele. The gels labeled by Cy5 and Bodipy-Cys are shown only in the Bodipy-Cys
fluorescence(A,B).TheCy5fluorescenceofthesegelswasusedforidentificationofthebandsandcorrectionfornonspecificBodipy-Cyslabeling(the
residualnonspecificlabelingstillcanbeobservedinthedimerbands).DTTwasincludedinthesampleloadingbufferinpanel(C)andexcludedfromthe
samplebufferinpanels(A),(B),and(D).NormalizedDSCthermogramsofpurifiedpepsin-treatedcollagenfromG610C(neoþ)andG610COI(neo–)mouse
tailtendonsandhumanfibroblastsweremeasuredatpH2.7(2mMHCl;E–G)andpH7.5(measuredin0.2Msodiumphosphateand0.5Mglyceroland
correctedby1.78Ctorepresentphysiologicalconditions;H–J).EachpeakrepresentstheheatofdenaturationofadistinctmolecularformoftypeI
collagen.Relativeareasunderthepeaksoneachthermogramrepresentthefractionofthecorrespondingmoleculesinthemixture.Thesuperscriptm
identifiesmutanta2(I)chains.
Goodness-of-fit testing indicated a significant deviation from squaredbothp<.0001).Wild-typeanimalswereheavierthanOI
expected genotype ratios looking at data from all litters animals (difference 1.77g,SE 0.3279, p<.0001).The A.B6 mice
(p¼.008). However, no individual strains had a significant had the lowest and the FVB.B6 mice had the greatest body
deviation fromexpected ratios. weight. FVB.B6 mice were significantly heavier than A.B6 mice
(difference 2.07g, p<.0001), FVB.B6 mice were significantly
heavier than Cby.B6 mive (difference 1.37g, p<.003 (both p
Body weight
values adjusted for multiple comparisons using Tukey-Kramer).
Therewasacurvilinearrelationbetweenbodyweight(measured There was no evidence of a strain-by-genotype interaction
longitudinally on days 21 to 60) and age (Fig. 5; age and age (strain(cid:5)genotypep<.8).
AMISHOIMUTATION JournalofBoneandMineralResearch 253
Table1. Differential ScanningCalorimetry (DSC) and SDS-PAGE Analysis ofType ICollagen
DSC SDS-PAGE
Sample Genotype a1 a2, % a1 a2m,%a a1 ,% a1 ,% ReactiveCys-SH a2m–S–S–a2m,%a
2 2 3 3
Mouse GG 100 0 0 0 0 0
Mouseneoþ GT 59(cid:4)5b 17(cid:4)1b 24(cid:4)5b 23(cid:4)5 Yes 1(cid:4)0.3
Mouseneoþ TT 0 38(cid:4)1b 62(cid:4)1b 67(cid:4)5 Yes 9(cid:4)3
MouseNeo– GT 51(cid:4)1b 49(cid:4)1b 0 0 Yes 1.5(cid:4)0.3
Human GG 100 0 0 0 No —
Human GT 50(cid:4)10b 50(cid:4)10b 0 0 Yes —
aThesuperscriptmidentifiesmutantchains.
bEstimateddeconvolutionerror.
aBMD of G610C OI mouse femurs mineral content and bone location (p¼.0065). The miner-
al:collagen ratio (PO :amideIII) also was found to have a
G610COImicehadlowerBMC(D¼0.005g,p<.001)andBMD 4
significantgenotype(p<.0001)effectandacurvilinearrelation-
(D¼0.005g/cm2,p¼.001).MeanaBMDrankorderoftheG610C
ship between mineral content and location (p¼.047) but no
OImicebymaternalbackgroundstrainforwasA.B6<Cby.B6<
evidenceofastraineffect(p¼.36).Thecurvilinearrelationships
FVB.B6<C3.B6.PairwisecomparisonsoftheG610COImiceusing
forPO :CHandPO :amideIIIhadpeakvalueslocatedaroundthe
the Scheffe multiple-comparisons adjustment procedure 4 4
middleofthecorticalbone.Onlygenotypewasfoundtohavea
resulted in a statistically significantly difference between
significant (p<.0001) effect for the collagen content of bone
A.B6 OI and C3.B6 OI (D¼0.007g/cm2, p¼.0061). A strain-by-
matrix(amideIII:CH).Mousestrain-by-genotypeinteractionwas
genotypeinteractionwasnotsignificantforeitherBMCorBMD
notsignificant for any oftheRaman data.
(p(cid:6).9).
Confocal Raman microspectroscopy mCT
Mineralcontentofbonematrix(PO :CH)(Fig.6)wasfoundtobe Table 2 shows mean values of selected trabecular and cortical
4
affectedbygenotype(p<.0001)andmousestrain(p¼.009).A boneparametersobtainedfromisolatedfemora.Corticalshape
significant curvilinear relationship was observed between and femur length were qualitatively very similar for each
Fig.5.F G610COImousegrowthcurves.Body-weightcurvesdemonstratedacurvilinearrelationbetweenweightandageforallgroupsofmice.Strain
1
andgenotypewereindependentpredictorsofweight.Thewild-type(control;blacksquares)micewere1.77gheavierthantheOImice(heterozygousfor
theCol1a2tm1.1Mcbrallele;graydiamonds),withthegreatestdifferenceinbodyweightbetweentheFVB.B6andtheA.B6groups.
254 JournalofBoneandMineralResearch DALEYETAL.
Fig.6.F G610COImouseRamanmicrospectroscopy.ThecollagencontentoftheOI(heterozygousfortheCol1a2tm1.1Mcbrallele)femurswasreduced
1
relativetowild-type(WT)controlfemurs,andthecollagenwashypermineralized.ArepresentativeRamanspectrumillustratesthepeaksusedfordata
analysis(upperleft),andcomparisonofchemical-specificratiosinWT(blackbars)andG610COI(graybars)femursdeterminedfromtheRamanspectraare
showninthebalanceofthefigure.
backgroundstraincontrol-OIpair(datanotshown).Overall,the ment,andenergytofailure.ThepresenceoftheCol1a2tm1.1Mcbr
mCT trabecular and cortical data show that all G610C OI mice alleleresultedinastrikingreductioninfailureloadforallF OI
1
carryingonecopyoftheCol1a2tm1.1Mcbrallelehadlessboneas mice compared with their strain-specific controls. The mean
measured by bone volume fraction. The trabecular bone was failure load range was greater for control mice (24N for A.B6
distinguished by fewer and more widely spaced trabeculae. mice to 35N for C3.B6) compared with OI mice (15N for A.B6
ExceptfortheFVB.B6mice,themeancorticalthicknessandarea mice to18N forC3.B6mice).TheimpactoftheCol1a2tm1.1Mcbr
werereducedbythepresenceoftheCol1a2tm1.1Mcbrallele.The alleleonthemagnitudeoftheintrastraincontrol-OIdifferencein
FVB.B6controlandOIgroupshadsimilarmeanvaluesforcortical stiffness was variable. The greatest intrastrain OI-control mean
thickness and area. All the OI femurs had reduced moment of difference was found in the C3.B6 group, where the
inertia (I ) compared with age-matched genetic background Col1a2tm1.1Mcbr allele was associated with a 71% reduction in
yy
controls. The cortical volumetric tissue mineral content (vTMC) meanstiffness.TheOImiceintheotherthreegroupshadmean
was lower in all OI mice relative to their genetic background stiffnessvaluesof82%(A.B6),91%(Cby.B6),and98%(FVb.B6)of
controls,exceptfortheFVB.B6OIgroup,whichwasthesameas their intrastrain controls. Mean values for postyield ultimate
theFVB.B6controlgroup.Thecorticalvolumetrictissuemineral displacement forOI micewere 45%(Cby.B6), 52%(C3.B6), and
density(vTMD)waselevatedinallmicecarryingCol1a2tm1.1Mcbr 59% (FVB.B6) of strain-matched control values. However, the
allele relative to their genetic background controls. The meanvalueforA.B6OImicewas95%ofthecontrolvalue.Failure
trabecular vTMD was statistically the same for each genetic energyrangedfrom0.78(Cby.B6)to1.14N-mm(C3.B6)fortheOI
background pair. mice and from3.11(A.B6) to 6.3N-mm (FVB.B6) forcontrols.
Ultimate stress and Young’s modulus had statistically
Four-point bending biomechanics significant genotype and strain differences, but the genotype-
by-genetic-background interaction was not significant (see
Isolated femurs were loaded to failure in a four-point bend- Table 3). Mean ultimate stress was reduced for all OI groups
testing apparatus to assess the impact of a single copy of the relative to their genetic background controls and ranged from
Col1a2tm1.1Mcbralleleandtheroleofgeneticbackgroundonthe 67%(Cby.B6OImice)to81%(C3.B6OImice).Young’smodulus
structural phenotype (Table 3). Specifically, femurs from mice was increased across all OI groups carrying the Col1a2tm1.1Mcbr
withonecopyofthevariantalleleweresignificantlyweakerand allele relative to their genetic background controls. The
more brittle than femurs from control mice. In addition, a magnitude of the mean Young’s modulus was not different
genotype-by-genetic-background (strain) interaction was sig- between OI-control pairs but varied by genetic background of
nificant for failure load, stiffness, postyield ultimate displace- the strain.
AMISHOIMUTATION JournalofBoneandMineralResearch 255
n 01 01 01 01
ai 0 0 0 0
Str <.0 <.0 <.0 <.0
e
alu pe 1 1 1
v y 0 0 0
VAp Genot <.00 <.00 <.00
O
C
N
A
n
o
cti 09 94 54 24 52
a 3 0 1 1 0
nter .0 .0 .0 .0 .0
I
16 73 00 4 4 7 31
92 41 09 31 01 27 82 90 37
00 27 76 10 20 70 00 23 82
OI 0.0. 2.0. 4.8. 0.0. 0.0. 0.0. 0.0. 2.0. 6.6.
33 52
5 0
1
6
B
3.
C
28 17 00 6 5 46 2
76 59 05 61 31 08 32 49 11
WT 0.10.0 4.01.1 3.38.5 0.10.0 0.20.0 0.90.0 0.10.0 2.80.2 8.24.2
31 41
5 0
1
34 13 00 1 1 1 35
81 67 02 31 91 46 01 60 36
00 54 20 10 10 70 10 22 34
OI 0.0. 2.0. 3.2. 0.0. 0.0. 0.0. 0.0. 2.0. 5.2.
72 22
4 0
6 1
B
B.
V
F 3 7
80 14 06 8 6 7 4
WT 0.140.02 4.500.43 9.907.09 0.140.00 0.180.00 0.760.04 0.120.01 2.280.15 9.323.26
5 61
4 9
66 59 00 4 6 9 66
04 45 08 21 81 87 81 38 7
10 88 79 10 10 60 00 12 70
OI 0.0. 2.0. 3.1. 0.0. 0.0. 0.0. 0.0. 2.0. 1.1.
14 42
5 0
6 1
B
y.
b
C 06 34 00 9 9 2 7
07 91 03 51 01 10 32 58 03
WT 0.20.0 4.71.1 6.03.3 0.10.0 0.20.0 0.80.1 0.10.0 2.50.3 9.22.8
23 12
5 0
1
T
C
m
m 6 9
dFro OI 0.0590.026 1.7980.491 8.5005.650 0.100.003 0.170.007 0.580.018 0.0610.004 1.800.07 1.563.20
Derive A.B6 483 1031
s
arameter WT 0.0820.026 2.9230.781 469.40010.860 0.120.0076 0.180.0084 0.660.044 0.0820.0090 2.000.14 999.489.51
P
al
c
Corti (%)
n
d o
2.Trabecularan cularBV/TV(%) cularnumber cularvBMD albonevol.fracti althickness(mm) 2alarea(mm) 4alI(mm)yy alvTMC(mg) alvTMD(mg/mL)
e n e e e c c c c c c
Tabl MeaSD Trab Trab Trab Corti Corti Corti Corti Corti Corti
256 JournalofBoneandMineralResearch DALEYETAL.