1
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
www.na u e.com/scien i ic epo s
SNPs in bone- ela ed miRNAs a e
associa ed wi h he os eopo o ic
pheno ype
Lau a De-Uga e1, En ique Ca o-Molina1, Ma ia Rod íguez-Sanz1, Miguel Angel Ga cía-Pé ez2,
José M. Olmos3, Manuel Sosa-Hen íquez4, Ramón Pé ez-Cano5, Ca los Gómez-Alonso6, Luis
Del Rio7, Jesús Ma eo-Agudo8, José An onio Blázquez-Cab e a9, Jesús González-Macías10,
Ja ie del Pino-Mon es11, Manuel Muñoz-To es12, Manuel Diaz-Cu iel13, Jo ge Malou 14,
An onio Cano15, José Luis Pé ez-Cas illon16, Xa ie Nogues 1, Na alia Ga cia-Gi al 1 &
Adol o Diez-Pe ez1
Biogenesis and unc ion o mic oRNAs can be in luenced by gene ic a ian s in he p i-miRNA sequences
leading o pheno ypic a iabili y. This s udy aims o iden i y single nucleo ide polymo phisms (SNPs)
a ec ing he exp ession le els o bone- ela ed ma u e mic oRNAs and hus, igge ing an os eopo o ic
pheno ype. An associa ion analysis o SNPs loca ed in p i-miRNA sequences wi h bone mine al densi y
(BMD) was pe o med in he OSTEOMED2 coho (n = 2183). Func ional s udies we e pe o med o
assessing he ole o BMD-associa ed miRNAs in bone cells. Two SNPs, s6430498 in he miR-3679
and s12512664 in he miR-4274, we e signi ican ly associa ed wi h emo al neck BMD. Fu he ,
we measu ed hese BMD-associa ed mic oRNAs in abecula bone om os eopo o ic hip ac u es
compa ing o non-os eopo o ic bone by qPCR. Bo h mic oRNAs we e ound o e exp essed in ac u ed
bone. Inc eased ma ix mine aliza ion was obse ed a e miR-3679-3p inhibi ion in human os eoblas ic
cells. Finally, geno ypes o s6430498 and s12512664 we e co ela ed wi h exp ession le els o
miR-3679 and miR-4274, espec i ely, in os eoblas s. In bo h cases, he allele ha gene a ed highe
mic oRNA exp ession le els was associa ed wi h lowe BMD alues. In conclusion, wo os eoblas -
exp essed mic oRNAs, miR-3679 and miR-4274, we e associa ed wi h BMD; hei o e exp ession
could con ibu e o he os eopo o ic pheno ype. These indings open new a eas o he s udy o bone
diso de s.
Mic oRNAs (miRNAs) ha e opened a new ield o esea ch o complex diseases wi h a gene ic basis. These small
non-coding RNAs inhibi he exp ession o a ge mRNAs by binding o hei 3′-un ansla ed egions (3′UTRs).
These molecules ha e added a new s ep o complexi y in gene egula ion, bu may also help o inc ease ou unde -
s anding o many mul i ac o ial diseases ha ha e been a mys e y up o now.
1IMIM (Hospi al del Ma Medical Resea ch Ins i u e), Uni e si a Au ònoma de Ba celona, CIBERFES, RETICEF
(ISCIII), Ba celona, Spain. 2Depa men o Gene ics and Ins i u e o Heal h Resea ch INCLIVA, Uni e si y o Valencia,
Valencia, Spain. 3Depa men o In e nal Medicine, Hospi al Uni e si a io Ma qués de Valdecilla-IDIVAL/Hospi al de
To ela ega, Uni e sidad de Can ab ia. RETICEF, San ande , Spain. 4Unidad Me abólica Ósea, Hospi al Uni e si a io
Insula , Uni e sidad de Las Palmas de G an Cana ia, Cana ias, Spain. 5Depa amen o de Medicina (USE), UGC
Medicina In e na, Hospi al Uni e si a io Vi gen Maca ena, Se ille, Spain. 6Se icio de Me abolismo Óseo y Mine al,
Hospi al Uni e si a io Cen al de As u ias, O iedo, Spain. 7CETIR G up Mèdic, RETICEF, Ba celona, Spain. 8Se icio
COT, Hospi al Uni e si a io Miguel Se e , Za agoza, Spain. 9Se icio de Medicina In e na, Complejo Hospi ala io
Uni e si a io de Albace e, Albace e, Spain. 10Depa amen o de Medicina In e na, H. Ma qués de Valdecilla,
Uni e sidad de Can ab ia, IDIVAL, RETICEF, San ande , Spain. 11Se icio de Reuma ología. Hospi al Uni e si a io
de Salamanca, RETICEF (ISCIII), IBSAL (Biomedical Resea ch Ins i u e o Salamanca), Salamanca, Spain. 12UGC
Endoc inologia y Nu ición. Hospi al Uni e si a io San Cecilio. G anada, RETICEF, Ibs, G anada, Spain. 13Unidad de
En e medades Me abólicas Óseas. Se icio de Medicina In e na. Fundacion Jimenez Diaz, Mad id, Spain. 14Hospi al
de la San a C eu I San Pau. Ins i u d’In es igació Biomèdica San Pau, Ba celona, Spain. 15Depa men o Pedia ics,
Obs e ics and Gynecology, Uni e si y o Valencia, INCLIVA, Valencia, Spain. 16Hospi al Uni e si a io Río Ho ega,
Valladolid, Spain. Co espondence and eques s o ma e ials should be add essed o N.G. (email: [email p o ec ed])
Recei ed: 13 July 2016
Accep ed: 7 Ma ch 2017
Published: xx xx xxxx
OPEN
www.na u e.com/scien i ic epo s/
2
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
miRNAs ha e been ex ensi ely s udied in bone esea ch, pa icula ly hei ela ionship o os eopo osis1–3.
These s udies clea ly showed al e ed miRNAs p o iling in se um om pa ien s wi h os eopo osis, as well as in
bone issue a e os eopo o ic (OP) ac u e. Howe e , hese miRNA exp ession signa u es obse ed in pa ien s
wi h os eopo osis do no p o ide e idence o causali y because he al e ed pa e n could be a consequence o he
disease o e en un ela ed o he pa hogenesis.
Ano he app oach in miRNAs s udies is he associa ion analysis be ween one SNP wi hin a candida e miRNA
(miR-SNP) o in a miRNA a ge si e and one disease ela ed-ou come. In his case, he associa ed a ian is
likely in ol ed in he pa hophysiology o con e s suscep ibili y o de elop he disease. Mo eo e , many e idences
sugges ha he gene ics o complex ai s a e a ibu able o gene ic a ia ions ha modula e gene exp ession,
a he han he a ia ions esul ing in p o ein s uc u e changes4. Howe e , unc ional assays a e needed in o de
o elucida e he ole o he associa ed a ian s in he pa hophysiology o he disease since he SNP could be in
linkage disequilib ium wi h he ue unc ional a ian .
The aim o his s udy was o iden i y SNPs wi hin candida e miRNAs in o de o pe o m an associa ion s udy
be ween hose SNPs and bone mine al densi y (BMD), he main ou come used o de ine os eopo osis. Fi s ,
we sea ched o miR-SNPs in he p ima y miRNA ansc ip (p i-miRNA), which has a hai pin s uc u e wi h
a e minal loop and wo single-s anded lanking egions. The p i-miRNAs a e ecognized and clea ed by he
D osha and DCGR8 complex, esul ing in a sho e s uc u e called p e-miRNA5. In his s ep, he sequences a
he unpai ed lanking a ms and wi hin he hai pin double-s anded s em s uc u e a e c ucial o co ec bind-
ing and clea age by he D osha-DCGR8 complex. Thus, he exis ence o gene ic a ian s wi hin he p i-miRNA
sequences could lead o an al e a ion o he hai pin s uc u e, a ec ing molecula p ocessing and he unde lying
miRNA ma u a ion6. Changes in miRNA ma u a ion would igge changes in miRNA abundancy, and conse-
quen ly a de egula ion o he exp ession le els o a ge genes. Suppo ing his idea, la ge-scale in silico analyses o
SNPs in human miRNA genes ha e demons a ed lowe SNP densi ies in he miRNA sequence han hei lanking
egions o he human genome7. Hence, ou s udy was based on he de ec ion and subsequen gene ic associa ion
analysis o pu a i e unc ional miR-SNPs. Fu he mo e, associa ed miR-SNPs we e explo ed in bone cells in o de
o alida e he associa ion wi h he OP pheno ype.
Resul s
An o e all o e iew o me hodology and esul s is schema ized in Fig.1.
Associa ion analysis wi h BMD. The i s app oach used in ou s udy was o iden i y unc ional a ian s
wi hin mic oRNAs in ol ed in bone me abolism. The mino allele equency (MAF) o many o he miR-SNPs
ound in da abases has no been assessed in Eu opean popula ion; he e o e, a alida ion s ep was necessa y. In
his ega d, SNPs in mic oRNA sequences we e i s ly alida ed as polymo phic in pa ien s om BARCOS subco-
ho 8. O he 53 miR-SNPs es ed, 14 we e alida ed in his coho . Howe e , only 5 SNPs had a MAF > 0.01 and
we e inally geno yped in he OSTEOMED2 coho o hei associa ion analysis wi h LS (lumba spine) BMD
and FN ( emo al neck) BMD (Table1).
SNP s6430498 in he miR-3679 was signi ican ly associa ed wi h FN BMD in a log-addi i e (p alue = 0.036,
Be a coe icien [95% CI] = −0.008 [−0.015 o −0.001] and ecessi e model (p alue = 0.021, Be a coe icien
[95% CI] = −0.017 [−0.032 o −0.003]); and he SNP s12512664 in he miR-4274 was also signi ican ly asso-
cia ed wi h FN BMD in a log-addi i e (p alue = 0.016, Be a coe icien [95% CI] = 0.008 [0.002 o 0.015] and
ecessi e model (p alue = 0.01, Be a coe icien [95% CI] = 0.015 [0.004 o 0.027). Fo bo h SNPs, he geno yping
e iciency was app oxima ely 98%. The A alleles o s6430498 (mino i y allele) and s12512664 (majo i y allele)
we e ound o be associa ed wi h lowe BMD alues.
Quan i ica ion o miRNA exp ession le els in o al FN bone samples. The an h opome ic ea u es
o he OP and Con ol g oups we e shown in Table2. Using he Mann–Whi ney U es , no s a is ical di e ences
in any ea u es we e obse ed be ween bo h g oups.
Figu e 1. Schema ic o e iew o he whole p ocedu es, samples and esul s o he s udy.
www.na u e.com/scien i ic epo s/
3
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
TARGET
GENE miRNA SNP MAFB Associa ion wi h BMD
ESR1
miR-106b s72631827 Monomo phic
miR-130b s72631822 Monomo phic
miR-148b s74878365 Monomo phic
miR-18a s41275866 Monomo phic
miR-222 s72631825 Monomo phic
miR-373 s80338016 Monomo phic
miR-520c s7255628 Monomo phic
miR-93 s72631824 Monomo phic
miR-96 s41274239 0.0033 MAF < 0.01
s73159662 0.0058
TGFB2
miR-141 s34385807 Monomo phic
miR-149 s71428439 Monomo phic
miR-182
s77586312 Monomo phic
s75953509 Monomo phic
s80041074 0.0033 MAF < 0.01
miR-199b s72631835 Monomo phic
miR-193a s60406007 Monomo phic
miR-200b s72563729 Monomo phic
miR-33a s77809319 Monomo phic
miR-431 s76090066 0.00083 MAF < 0.01
s128840′05 Monomo phic
miR-590 s6971711 Monomo phic
miR-7-1 s76662330 Monomo phic
miR-7-2 s41276930 0.005 MAF < 0.01
s75737367 Monomo phic
PTH1R miR-339
s13232101 Monomo phic
s72631820 Monomo phic
s72631831 Monomo phic
RUNX2 miR-122 s41292412 0.0033 MAF < 0.01
miR-154 s41286570 0.0004
IL6R
miR-124-2 s72631829 Monomo phic
miR-124-3 s34059726 Monomo phic
miR-125a s12975333 Monomo phic
miR-140 s7205289 Monomo phic
miR-320d-1 s74826059 Monomo phic
miR-499 s3746444 0.21 No associa ed
s7267163 0.0025 MAF < 0.01
LRP5 miR-27a s11671784 0.0162 No associa ed
IL6
miR-146a s2910164 0.26 No associa ed
miR-146b s76149940 Monomo phic
miR-202 s12355840 Monomo phic
miR-365-2 s35143473 Monomo phic
VDR miR-10a s72631828 Monomo phic
miR-223 s34952329 Monomo phic
CYP24A1
miR-30b s111424617 Monomo phic
miR-30e s112439044 Monomo phic
miR-183 s72631833 Monomo phic
s41281222 Monomo phic
miR-101-2 s78851134 0.0004 MAF < 0.01
Highly
exp essed in
HObs
miR-1282 s11269 Monomo phic
miR-3679 s6430498 0.35 Associa ed wi h FN BMD
s10175383 Monomo phic
miR-4274 s12512664 0.47 Associa ed wi h FN BMD
Table 1. Valida ion o miR-SNPs o he BMD associa ion analysis. MAFB; Mino allele equency in BARCOS
coho ; SNPs wi h MAF<0.01 we e excluded o geno yping. In bold; Valida ed SNPs o geno yping in o al
OSTEOMED2 coho .
www.na u e.com/scien i ic epo s/
4
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
MiR-3679 and miR-4274, which ha bo ed BMD-associa ed SNPs in hei p e-miRNA sequence (Fig.2), we e
quan i ied by qPCR in o de o compa e he exp ession le els be ween OP and non-OP bone samples.
MiR-3679-5p was no de ec ed in o al bone samples o in cul u ed human os eoblas cells, bu miR-3679-3p
was p esen in bo h sample ypes. The e o e, only ma u e miR-3679-3p was assayed in he qPCR. Bo h miRNAs
miR-3679-3p and miR-4274 we e signi ican ly o e exp essed in he OP samples (Table3).
O he bone miRNAs, miR-631 and miR-574-5p1, 9, we e also es ed in o de o p eclude a non-speci ic phe-
nomenon ela ed o a gene al up egula ion o gene exp ession. No di e ences we e ound be ween os eopo o ic
and non-os eopo o ic bone samples (p = 0.626 and p = 0.183 espec i ely).
Associa ion analysis o miRNA exp ession le els wi h geno ypes o BMD-associa ed
SNPs. Human p ima y os eoblas s (n = 38) we e cul u ed o DNA and RNA ex ac ion and so ed by
Bicological g oups nAge (Mean ± SD) BMI (kg/m2)
(Mean ± SD) BMD (g/cm2)
(Mean ± SD)
Os eopo o ic 10 75.6 ± 6.38 27.11 ± 2.94 F agili y ac u e
Non-os eopo o ic 10 71.7 ± 7.36 27.42 ± 3.15 0.882 ± 0.137
Table 2. Pa ien cha ac e is ics o os eopo o ic ac u e and non-os eopo o ic g oups. Abb e ia ions: SD:
S anda d De ia ion; BMI: Body Mass Index; BMD: Bone Mine al Densi y.
Figu e 2. P edic ed changes in seconda y s uc u e p o oked by he BMD-associa ed SNPs acco ding o
RNAs uc u e web se e . Black a ows ag he co esponding allele o s6430498 in he miR-3679 and
s12512664 in he miR-4274.
miRNA Biological
G oup RQ
(Median) IQR P alue
miR-3679 Os eopo o ic 89.601 220.636 0.001
Con ol 1.423 0.964
miR-4274 Os eopo o ic 144.268 318.409 0.001
Con ol 1.197 2.154
Table 3. miRNA exp ession le els, compa ison be ween os eopo o ic and non-os eopo o ic bone samples.
Abb e ia ions: RQ = Rela i e quan i ica ion; IQR = In e qua ile Range.
www.na u e.com/scien i ic epo s/
5
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
geno ype o bo h s6430498 and s12512664. Associa ion be ween exp ession le els o ma u e miRNA miR-
3679-3p and miR-4274 wi h i s co esponding own geno ype was analyzed. A signi ican co ela ion was obse ed
be ween miRNA le els and he gene ic a ian (Fig.3). The A allele o bo h SNPs was associa ed wi h highe
exp ession o each co esponding miRNA (miR-3674; log-addi i e model; p alue = 0.015; ecessi e model; p
alue = 0.03, and miR-4274; ecessi e model; p alue = 0.013).
Addi ionally, in o de o co obo a e ha he di e ences among exp ession le els a e due o geno ypes “pe se”
and no o o he cellula ci cums ances, o he bone- ela ed miRNAs1, 9 we e checked in hese cells. Exp ession
le els o miR-320a, and miR-22-3p we e measu ed and no di e ences we e ound i espec i e o geno ypes.
Mo eo e , as a sensi i i y es , we analyzed he miR-3679-3p and miR-4274 exp ession le els sepa a ed acco d-
ing o he exchanged geno ype amongs bo h miRNAs ( s12512664 and s6430498 espec i ely). Once again,
no di e ences we e ound in bo h miRNA analyzed (miR-3679-3p; s12512664: p- alue = 0.713 and miR-4274;
s6430498: p- alue = 0.645).
Bioin o ma ic analysis. A sub le e ec o he BMD-associa ed SNPs on seconda y p e-miRNA s uc u e
was de ec ed. The gene ic a ian s s6430498 and s12512664 did no p o oke e iden changes in he loop o ma-
ion on he RNAs uc u e web se e (Fig.2).
Among p edic ed a ge s o miR-3679-3p a e WHSC1/WHSC1L1, THRAP2, SMAD2, LRP6, SOX4, WNT7A,
and MGAT5B. The miR-3679 is encoded wi hin MGAT5 gene. The p o ein encoded by his gene, he manno-
syl (alpha-1,6-)-glycop o ein be a-1,6-N-ace yl-glucosaminyl ans e ase, is one o he mos impo an enzymes
in ol ed in he egula ion o he biosyn hesis o glycop o ein oligosaccha ides. The mos signi ican pa hways
egula ed by his miRNA is he glyce ophospholipid me abolism (p- alue = 0.02) and MAPK signaling pa hway
(p- alue = 0.049).
P edic ed a ge s o miR-4274 a e BTF3, LRRC1, SOCS5, RAB10, FGR, RABGEF2 and ESRRG genes. E bB
(p- alue = 0.002), p o eoglycan in cance (p- alue = 0.006) and mTOR (p- alue = 0.01) signaling pa hways a e
he op pa hways in ol ing gene a ge s o his miRNA.
In- i o assessmen o os eoblas ac i i y and ma ix mine aliza ion. hOBs we e ea ed wi h
miR-3679-3p and miR-4274 mimics o inhibi o s and hei espec i e con ols o in es iga ing hei ole in
os eoblas unc ion (Supplemen al Fig.1).
Inhibi ion o miR-3679-3p signi ican ly enhanced 10% ma ix mine aliza ion le els a 28-day cul u e
(Inhibi o : 1.64 ± 0.021; con ol: 1.49 ± 0.056; p alue = 0.004). On he o he hand, o e exp ession o miR-
3679-3p in hOBs weakly educes ma ix mine aliza ion al hough di e ences we e no signi ican (Mimic:
1.53 ± 0.085; con ol: 1.68 ± 0.256; p alue = 0.589). No signi ican di e ences in alkaline phospha ase (ALP)
ac i i y and cell p oli e a ion we e obse ed a e miR-3679-3p ans ec ion.
Fo miR-4274 ans ec ions, no changes in any pa ame e e alua ed we e ound.
Discussion
Two os eoblas - ela ed mic oRNAs, miR-3679 and miR-4274, we e ound o be associa ed wi h emo al neck
BMD and o e exp essed in OP ac u ed bone. These miRNAs ha bo gene ic a ian s in hei p e-miRNA
sequences ha we e co ela ed wi h he miRNA exp ession le els in human os eoblas s. Finally, inhibi ion o
Figu e 3. Co ela ion be ween miRNA exp ession le els and geno ypes o miR-3679 and miR-4274. MiRNA
exp ession le els a e ep esen ed as a mean ± SD o he ela i e exp ession in Real-Time PCR. U6 was used
o no maliza ion. Samples o 38 human p ima y os eoblas s we e used o expe imen s. (n) is he numbe o
samples o each geno ype g oup.

www.na u e.com/scien i ic epo s/
6
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
miR-3679-3p in human os eoblas ic cells inc eased ma ix mine aliza ion. The lowe densi y o gene ic a ian s in
miRNA genes compa ed o o he non-coding genomic egions sugges s ha SNPs can ha e a ema kable biolog-
ical ole in p i-miRNAs by di e en ial egula ion o hei a ge genes10. Hence, we chose gene ic a ian s wi hin
he p i-miRNA sequences o hei pu a i e unc ionali y. Mo eo e , he BMD-associa ed miR-SNPs in ou s udy
we e loca ed in he p e-miRNA molecules, which ha e been desc ibed o exhibi s ong selec i e cons ain s11, 12.
SNPs wi hin he p i-miRNA sequence can a ec he miRNA ma u a ion in mul iple aspec s, modi ying
he hai pin s uc u e o changing he binding a ini y o biogenesis enzymes o he miRNA hai pin. Mo eo e ,
a ian s can a ec al e na i e clea age si es o biogenesis enzymes, p oducing no el isomiRs o changing hei
equency. These changes in ma u a ion can esul in al e ed exp ession o he unc ional miRNA, esul ing in
de egula ion o a ge genes. Hence, se e al s udies ha e sugges ed ha unc ional SNPs in p i-miRNAs a ec he
p ocessing and exp ession le els o ma u e miRNAs13.
Two mi -SNPs, s6430498 and s12512664, wi hin he p e-miRNA sequence o miR-3679 and miR-4234,
espec i ely, we e associa ed wi h FN BMD in he OSTEOMED2 coho . Bioin o ma ic analysis o he hai -
pin s uc u e o hese BMD-associa ed miRNAs showed no s iking e ec on i s seconda y s uc u e caused
by allelic changes bu we canno ule ou ha he exp ession was al e ed by a ec ing he a ini y binding o he
p ocessing enzymes o hei accesso y p o eins. In his ega d, o he s udies ha e demons a ed ha miR-SNPs
in p e-miRNA sequences lead o al e ed le els o ma u e miRNA in i o e en hough changes in i s seconda y
s uc u e a e no p edic ed by RNA old p og am14.
In ou s udy he AA geno ype o bo h s6430498 and s12512664 showed highe exp ession le els o miR-
3679-3p and miR-4274 in hOBs, espec i ely. Mo eo e , hese miRNAs we e o e exp essed in OP bone samples.
Consis en ly wi h miR-3679-3p o e exp ession in OP ac u ed bone, ans ec ion esul s sugges ha miR-
3679-3p nega i ely egula es ma ix mine aliza ion by os eoblas s which could lead o an OP pheno ype. Fu he
suppo ing hese esul s, he A allele o bo h SNPs was associa ed wi h lowe FN BMD le els.
In o e all, p edic ed a ge s and pa hways in ol ing miR-3679-3p belong o cell cycle, di e en ia ion and is-
sue de elopmen , sugges ing a dys unc ion in bone emodeling in he os eopo o ic bone. In ag eemen wi h hese
da a, an impai ed mine aliza ion was obse ed when miR-3679-3p was o e exp essed in os eoblas s. In con as ,
analysis o a ge ed genes and pa hways o miR-4274 sugges s ha his miRNA could be in ol ed, b oadly, in
esicula and memb ane a icking, cell mig a ion and adhesion, and cell me abolism. Tha would explain why
changes in he assayed pa ame e s we e no de ec ed a e miR-4274 ans ec ion in os eoblas s.
Al oge he , hese indings sugges ha hese a ian s can play a ole in bone me abolism. Since in o ma ion
abou hese wo miRNAs is e y sca ce, his s udy could be conside ed as a p elimina y app oach p esen ing miR-
3679-3p and miR-4274 wi hin he complex ne wo k o bone egula ion.
Ou s udy is s eng hened by he ex emely ca e ul in con ol o po en ially con ounding cha ac e is ics
among bone samples in e ms o age, sex, BMI and me abolic diseases highly p e alen in elde ly popula ion15, 16.
Mo eo e , we excluded om he s udy samples om pa ien s ea ed wi h ho mones and o he an i-os eopo o ic
d ugs ha could al e he miRNA exp ession in bone cells17. These s ic inclusion c i e ia suppo ha miRNAs
de ec ed in ou s udy can be mo e eliably conside ed in ol ed in he os eopo o ic pheno ype.
One limi a ion is he use o os eoa h i ic samples as con ol g oup. Due o ob ious e hical easons he col-
lec ion o bone om heal hy indi iduals is no allowed. Howe e , in an a emp o minimize his po en ial limi-
a ion, we ob ained he bone samples om a loca ion dis an om he in e ace be ween bone and ca ilage and,
he e o e, as a away as possible om he os eoa h i ic lesion. Mo eo e , al hough bo h miRNAs ha e ound
o e exp essed in OP bone issue, i canno be a ibu ed o he SNP egula ion in hese samples since his could
no be analyzed due o lack o bone sample’s a ailabili y. O he limi a ion o he s udy is he need o eplica e he
associa ion esul s in o he coho s because hese miR-SNPs lacked linkage in o ma ion wi h o he ag-SNPs
ela ed o bone pheno ypes. Howe e , his is he i s ime ha wo gene ic a ian s in human os eoblas - ela ed
miRNAs ha e been associa ed wi h FN BMD in an ex ensi e coho o pos menopausal women, and his associ-
a ion was unc ionally demons a ed in bone samples.
In conclusion, wo pu a i e unc ional SNPs in he p e-miRNA sequence o miR-3679 and miR-4274 we e
associa ed wi h emo al neck BMD. In bo h cases, he allele associa ed wi h lowe BMD was co ela ed wi h
highe exp ession le els o ma u e miRNA in human os eoblas ic cells. Mo eo e , bo h miRNAs we e o e ex-
p essed in OP ac u ed bone. Ou esul s open new explo a o y a enues o u u e s udies in he bone ield.
Me hods
E hics S a emen . The s udy p o ocols o ob aining DNA om blood samples, bone issue samples and
p ima y os eoblas s we e app o ed by he CEIC-Pa c de Salu MAR, he coo dina ing cen e (Regis y numbe
2010/3882/I). Me hods we e ca ied ou in acco dance wi h he app o ed guidelines by he CEIC-Pa c de Salu
MAR. The app o ed p o ocols we e explained o po en ial s udy pa icipan s, who p o ided w i en in o med
consen be o e being included in he s udy. All s udies we e ca ied ou in acco dance wi h he p inciples o he
Decla a ion o Helsinki as e ised in 2008.
S udy subjec s. Gene ic associa ion analysis was pe o med in an obse a ional, clinical coho s udy
(OSTEOMED2) o pa ien s ec ui ed om 14 medical cen e s in se e al egions o Spain. All pa ien s we e
consecu i e, unselec ed, pos menopausal women a ended in an ou pa ien clinic. Pa ien s we e p ospec i ely
ec ui ed ega dless o BMD alue (Table4). Exclusion c i e ia we e any his o y o me abolic o endoc ine dis-
ease, ch onic enal ailu e, ch onic li e disease, malignancy (excep supe icial skin cance ), Page ’s disease o
bone, malabso p ion synd ome, and any an i-OP o bone-a ec ing ea men . In addi ion, women wi h ea ly
menopause (be o e he age o 40) we e excluded om analysis.
BMD (g/cm2) was measu ed a he lumba spine (LS) L2-L4 and a he non-dominan emo al neck (FN) using
he dual-ene gy X- ay densi ome e s a ailable in each pa icipa ing cen e (Supplemen al Table1).
www.na u e.com/scien i ic epo s/
7
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
Whole bone samples o qPCR miRNA quan i ica ion we e ob ained om he ansce ical egion o he
emo al neck o pos menopausal women unde going hip eplacemen due o ei he OP ac u e (n = 10) o os e-
oa h i is in he absence o os eopo osis (n = 10) (T-sco e measu emen s [mean ± SD]: 0.616 ± 0.523) (Table2).
Bone was aken om a loca ion dis an om he in e ace be ween bone and ca ilage and, he e o e, as a away
as possible om he os eoa h i ic lesion. Again, none o he pa icipan s had a his o y o me abolic o endoc ine
disease, ch onic enal ailu e, ch onic li e disease, malignancy, Page ’s disease o bone, malabso p ion synd ome,
ho mone eplacemen he apy, an i- eso p i e o anabolic agen s, o al co icos e oids, an i-epilep ic d ugs, o
ea men wi h li hium, hepa in, o wa a in.
SNPs selec ion in p i-miRNA sequences. In he i s s ep, 9 candida e genes we e selec ed o
hei well-known unc ion in bone egula ion: ESR1, TGFB2, PTH1R, RUNX2, IL6R, LRP5, IL6, VDR, and
CYP24A118–23 (Table1). Then, miRNAs ha bind o he mRNA 3′UTR o hese genes we e iden i ied using he
mic oRNA.o g websi e24 (h p://www.mic o na.o g), Ta ge Scan Human (h p://www. a ge scan.o g) and PicTa
(h p://pic a .mdc-be lin.de). The algo i hm used is based on sequence complemen a i y, binding ene gy o he
miRNA– a ge duplex, e olu iona y conse a ion o he a ge si e sequence and a ge posi ion in aligned UTRs
o homologous genes. Some lis ed miRNAs a ge mo e han one bone- ela ed gene, as o example, miR-130b,
miR106b, miR148b and miR-93 ha a ge bo h ESR1 and TGFB2 genes. These miRNAs we e loca ed in ESR1 a -
ge miRNAs in o de o simpli y he Table1. Nex , miRNAs wi h he highes exp ession le els in p ima y human
os eoblas s (hOB) we e selec ed (Table1) acco ding o esul s ob ained in a p e ious miRNA a ay pe o med
in ou g oup (GEGSE74211; h p://www.ncbi.nlm.nih.go /geo/que y/acc.cgi? acc = GSE74211)1. Finally, SNPs
in he p i-miRNA sequences we e sea ched using he Ensembl (www.ensembl.o g), UCSC Genome B owse
on Human Dec. 2013 (GRCh38/hg38) Assembly25 (h p://genome.ucsc.edu/), and HapMap (www.hapmap.o g)
da abases.
Only hose SNPs wi h a mino allele equency (MAF) in U ah esiden s wi h ances y om no he n and
wes e n Eu ope (CEU) > 0.01 we e included in he s udy; SNPs lacking published MAF we e alida ed in pa -
icipan s om he BARCOS coho 8 (included in OSTEOMED2) by means o PCR-RFLP, geno yping o Sange
sequencing.
Polymo phism geno yping. DNA was ob ained om pe iphe al blood collec ed in EDTA ubes and geno-
yping was pe o med in LGC Genomics pla o m using KASPa 4.0 geno yping sys ems. To ensu e geno yping
quali y, a andom sample (5% o he o al numbe o samples) was also geno yped in a sepa a e con ol pla e.
The e was 100% conco dance be ween hese esul s.
Human os eoblas s cul u e. Human p ima y os eoblas s (hOB) we e ob ained om abecula bone o
pa ien s who unde wen knee o hip eplacemen due o os eoa h i is. Bony issue was cu up in o small pieces,
washed in phospha e bu e ed solu ion (PBS, Gibco by Li e Technologies; Paisley, UK) o emo e non-adhe en
cells, and placed on a 140 mm cul u e pla e. Samples we e incuba ed in hOB medium: Dulbecco’s Modi ied Eagle
Medium (DMEM; Gibco; In i ogen, Paisley, UK), supplemen ed wi h 10% e al cal se um (FBS; Sigma-Ald ich;
S . Louis, USA), 100 U/ml penicillin/s ep omycin (Sigma-Ald ich; S . Louis, USA), 0.4% ungizone (Gibco by
Li e Technologies; Paisley, UK), and 100 ug/ml asco bic acid (Sigma; S einheim, Ge many). Alkaline Phospha ase
(ALP) ac i i y and os eocalcin gene exp ession we e measu ed in o de o con i m he os eoblas ic pheno ype
(da a no shown). All expe imen s we e pe o med a maximum passage 2.
Thi y eigh samples we e g own in pa allel: one o DNA ex ac ion and he o he o RNA ex ac ion.
Addi ionally, 11 hOB samples we e used o Aliza in Red, ALP and MTS assays.
Human os eoblas DNA ex ac ion and sequencing. DNA was ex ac ed om cul u ed hOBs
using he Wiza d® Genomic DNA Pu i ica ion Ki (P omega) acco ding o manu ac u e 's ins uc-
ions. Geno ypes o s6430498 and s12512664 we e assessed by Sange sequencing using he BigDye®
Te mina o 3.1 (Applied Biosys ems) in he genomic se ices o Uni e si a Pompeu Fab a. P ime s
(Sigma-Ald ich) we e designed using he P ime 3 impu ( . 0.4.0) and he UCSC websi e: o s6430498 in
miR-3679, F: 5′-CGGTGAGGAGTTTTCTGAATG-3′ and R: 5′-CACCAAGCATAATAGCTAAAAATCAA-3′
( agmen size: 400 bp) and o s12512664 in miR-4274, F: 5′-CATCCACTTTGGGGAGAAGT-3′ and R:
5′-CCAAGGTACCACTGCCTCAT-3′ ( agmen size: 392 bp).
RNA ex ac ion o os eoblas s and o al bone. miRNA-en iched ac ion om hOB cul u es
was ex ac ed using MiRNeasy mini ki and RNeasy MinElu e Cleanup (Qiagen) acco ding o manu ac u e
ins uc ions.
Pa ien cha ac e is ic
Mean ± SD
LS BMD n = 2183 FN BMD n = 2015
Age (yea s) 57.61 ± 9.26 58.80 ± 8.99
Age o menopause
(yea s) 48.7 ± 3.94 48.7 ± 3.92
BMI (kg/m2) 26.56 ± 4.18 26.48 ± 4.13
BMD (g/cm2) 0.870 ± 0.16 0.707 ± 0.14
Table 4. Baseline cha ac e is ics o he OSTEOMED2 coho . Abb e ia ions: BMI = body mass index;
BMD = bone mine al densi y; LS = lumba spine; FN = emo al neck.
www.na u e.com/scien i ic epo s/
8
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
Fo RNA ex ac ion o o al bone, a piece o issue was cu ou and added o QIAzol Lysis Reagen (Qiagen),
hen homogenized o 5 min using he TissueLyse sys em. Chlo o o m was added o each sample, ollowed by
cen i uga ion o 15 min (12000 g). The uppe wa e phase was collec ed and he ex ac ion con inued acco ding
o manu ac u e ’s ins uc ions. The quali y o he o al RNA was e i ied by an Agilen 2100 Bioanalyze p o-
ile. The concen a ion o he pu i ied RNA was analyzed on a spec opho ome e (Nanod op, The mo Fishe
Scien i ic Inc).
miRNA quan i ica ion by qPCR. Using he miSc ip II RT ki (Qiagen), 1 µg o o al RNA was
e e se- ansc ibed in 20 μl eac ions. cDNA om o al bone was dilu ed x8 and cDNA om hOBs was dilu ed
x16; in bo h cases, 2 µl we e assayed in 10 µl PCR eac ions in 384-well pla es using MiSc ip Sybe G een PCR
ki acco ding o he p o ocol. The sequence o he ma u e miRNAs selec ed, acco ding o he mi Base web si e,
was used as a o wa d p ime and he Uni e sal p ime as a e e se. U6 snRNA was used as he e e ence gene o
no maliza ion. Ampli ica ion was pe o med in a Quan S udio 12 K Flex Real-Time PCR (Applied Biosys ems),
and he Exp ession Sui e so wa e was used bo h o de e mina ion o ela i e quan i ica ion (RQ) (by 2−ΔΔC
me hod) and o mel ing cu e analysis.
Bioin o ma ics analyses o BMD-associa ed miRNAs. Gene a ge p edic ion was assessed using
he ollowing six p og ams: PicTa (h p://pic a .mdc-be lin.de), Ta ge Scan Human (h p://www. a ge scan.
o g), miRDB (h p://mi db.o g), MiRanda (h p://www.mic o na.o g), DIANA-Ta Base (h p://diana.imis.
a hena-inno a ion.g ), and miRTa Base (h p://mi a base.mbc.nc u.edu. w). The DIANA-mi Pa h web-based
compu a ional ool26 was used o iden i y molecula pa hways po en ially al e ed by he in e sec ion o miRNAs
di e en ially exp essed in ac u ed bone.
miRNASNP da abase10, RNAs uc u e (h p:// na.u mc. oches e .edu/RNAs uc u eWeb) and RNA old
so wa e (h p:// na. bi.uni ie.ac.a /cgi-bin/RNA old.cgi) we e used o p edic he e ec o he a ian s on he
miRNA seconda y s uc u e. These p edic ions a e based on he assump ion ha a ian s can des abilize he
hai pin and he e o e educe he ma u e miRNA le els.
Cell ans ec ion. P ima y hOB cells we e seeded a he ollowing condi ions: 96-well pla e a 12.000 cells/
well o MTS and ALP ac i i y assays and 24-well pla e a 45.000 cells/well o Aliza in Red quan i ica ion.
Once cells eached 60–70% o con luence, ansien ans ec ions we e pe o med using mi Vana mimics and
inhibi o s o bo h hsa-miR-3679-3p and hsa-miR-4274. Mi Vana™ miRNA Mimic Nega i e Con ol #1 and mi -
Vana™ miRNA Inhibi o Nega i e Con ol #1 we e used as con ols. All p oduc s we e pu chased om Ambion®
Li e Technologies. Mimics and con ols mimics we e used a 100 nM and inhibi o s and con ol inhibi o s a
400 nM. In o de o moni o ans ec ion e iciency, miRIDIAN mic oRNA Mimic T ans ec ion Con ol wi h Dy547
(Dha macon) was ans ec ed in o he cells a he same condi ions as he miRNAs es ed. T ans ec ion o miRNAs was
ca ied ou using Lipo ec amine RNAiMAX (In i ogen; Ca lsbad, USA) acco ding o he manu ac u e ’s ins uc ion.
Aliza in Red quan i ica ion. To induce os eoblas ic mine aliza ion, hOBs (n = 4) we e cul u ed du ing
28 days wi h hOB medium supplemen ed wi h 5 mM β-glice ophospha e (Sigma-Ald ich, S Louis, MO, USA).
Cells we e ans ec ed wi h bo h mimics and inhibi o s o miR-3679-3p and miR-4274, and hei co esponding
con ols a day 1, and day 14 a e seeding. Du ing he cell cul u e ime pe iod, he medium was changed once
e e y 3 days.
A day 28, cells we e s ained wi h aliza in ed o quan i y mine alized nodules. A his ime poin , media was
emo ed om he cell monolaye and gen ly washed 3 imes wi h PBS. The cells we e hen ixed in 10% bu e ed
o malin o 10 minu es a oom empe a u e. Fixa i e was emo ed and cul u es we e washed in PBS. The cell
laye was s ained wi h 2% Aliza in-S (Sigma-Ald ich, S Louis, MO, USA) a ~pH 4.2 o 20 minu es. Cell p epa-
a ions we e washed wi h PBS o elimina e nonspeci ic s aining. To quan i y calcium deposi ion, he dye was
leached by he addi ion o 10% ce ylpy idinium chlo ide un il all o dye had been d awn. Op ical densi y was hen
quan i ied by spec opho ome y a 550 nm, using 10% ce ylpy idinium chlo ide as a blank e e ence.
ALP ac i i y assay. ALP ac i i y was measu ed 48 hou s a e miRNA ans ec ion o hOBs (n = 4) using
he Alkaline Phospha ase Assay Ki (Colo ime ic) (Abcam; Camb idge, UK) acco ding o he manu ac u e ’s
ins uc ions.
Cell iabili y assay (MTS). Viable cells we e de e mined 48 hou s pos - ans ec ion o hOBs (n = 3) using
he CellTi e 96® AQueous One Solu ion Cell P oli e a ion Assay (P omega; WI, USA), acco ding o he manu-
ac u e ’s ins uc ions.
S a is ical me hods. Ha dy-Weinbe g equilib ium (HWE) was calcula ed using he chi-squa e es .
Mul iple linea eg ession models we e i ed o assess he associa ion be ween geno yped SNPs and BMD.
Log-addi i e, dominan and ecessi e models we e es ed o each SNP associa ion analysis. Po en ial con ound-
e s conside ed o adjus men we e densi ome e de ices, body mass index (BMI) and age. Associa ion analyses
we e pe o med using R so wa e e sion 2.13.2 wi h he SNPassoc, o eign, gda a and mul es packages.
Mann-Whi ney U es was pe o med o OP and non-OP g oup compa isons in he SPSS .12.0 o Windows.
This es was also used o ea men compa isons be ween miRNA mimics o inhibi o s and hei espec i e
con ols.
Linea eg ession was used o analyze he associa ion be ween miRNA exp ession le els and sample geno-
ypes. Log-addi i e, dominan and ecessi e models we e es ed o each miRNA analysis. Co ela ion analyses
we e pe o med using R so wa e e sion 2.13.2 wi h he SNPassoc and o eign packages.
All analyses we e wo- ailed, and p- alues < 0.05 we e conside ed signi ican .
www.na u e.com/scien i ic epo s/
9
Scien i ic RepoR s | 7: 516 | DOI:10.1038/s41598-017-00641-7
Re e ences
1. De-Uga e, L. e al. MiRNA p o iling o whole abecula bone: iden i ica ion o os eopo osis- ela ed changes in MiRNAs in human
hip bones. BMC Med Genomics 8, 75 (2015).
2. Panach, L., Mi su , D., Ta in, J. J., Cano, A. & Ga cia-Pe ez, M. A. Se um Ci cula ing Mic oRNAs as Bioma ke s o Os eopo o ic
F ac u e. Calci Tissue In 97, 495–505 (2015).
3. Seelige , C. e al. Fi e F eely Ci cula ing miRNAs and Bone Tissue miRNAs a e Associa ed wi h Os eopo o ic F ac u es. J Bone
Mine Res, doi:10.1002/jbm .2175 (2014).
4. Es ada, K. e al. Genome-wide me a-analysis iden i ies 56 bone mine al densi y loci and e eals 14 loci associa ed wi h isk o
ac u e. Na Gene 44, 491–501 (2012).
5. Wahid, F., Shehzad, A., Khan, T. & Kim, Y. Y. Mic oRNAs: syn hesis, mechanism, unc ion, and ecen clinical ials. Biochim Biophys
Ac a 1803, 1231–1243 (2010).
6. Cammae s, S., S azisa , M., De Rijk, P. & Del Fa e o, J. Gene ic a ian s in mic oRNA genes: impac on mic oRNA exp ession,
unc ion, and disease. F on Gene 6, 186 (2015).
7. Han, M. & Zheng, Y. Comp ehensi e analysis o single nucleo ide polymo phisms in human mic oRNAs. PLoS One 8, e78028
(2013).
8. Yosko i z, G. e al. Analyses o RANK and RANKL in he pos -GWAS con ex : unc ional e idence o i amin D s imula ion
h ough a RANKL dis al egion. J Bone Mine Res 28, 2550–2560 (2013).
9. Ga cía-Gi al , N., Yosko i z, D.-U. L., Güe i, G., G inbe g, R., Nogués, D., Mellibo sky, X., Balcells, L., Díez-Pé ez, S. & Es udio, A.
del pa ón de exp esión de mic oRNAs en el hueso os eopo ó ico. Re Os eopo os Me ab Mine 8, 5–14 (2016).
10. Go ng, J. e al. Genome-wide iden i ica ion o SNPs in mic oRNA genes and he SNP e ec s on mic oRNA a ge binding and
biogenesis. Hum Mu a 33, 254–263 (2012).
11. Saunde s, M. A., Liang, H. & Li, W. H. Human polymo phism a mic oRNAs and mic oRNA a ge si es. P oc Na l Acad Sci USA
104, 3300–3305 (2007).
12. Iwai, N. & Na aba, H. Polymo phisms in human p e-miRNAs. Biochem Biophys Res Commun 331, 1439–1444 (2005).
13. Ryan, B. M., Robles, A. I. & Ha is, C. C. Gene ic a ia ion in mic oRNA ne wo ks: he implica ions o cance esea ch. Na Re
Cance 10, 389–402 (2010).
14. Wu, M. e al. Gene ic a ia ions o mic oRNAs in human cance and hei e ec s on he exp ession o miRNAs. Ca cinogenesis 29,
1710–1716 (2008).
15. Kocijan , R. e al. Ci cula ing mic oRNA Signa u es in Pa ien s Wi h Idiopa hic and Pos menopausal Os eopo osis and F agili y
F ac u es. J Clin Endoc inol Me ab 101, 4125–4134 (2016).
16. Hei lmeie , U. e al. Se um miRNA Signa u es A e Indica i e o Skele al F ac u es in Pos menopausal Women Wi h and Wi hou
Type 2 Diabe es and In luence Os eogenic and Adipogenic Di e en ia ion o Adipose Tissue-De i ed Mesenchymal S em Cells In
Vi o. J Bone Mine Res 31, 2173–2192 (2016).
17. L axman, N. e al. Second gene a ion sequencing o mic oRNA in Human Bone Cells ea ed wi h Pa a hy oid Ho mone o
Dexame hasone. Bone 84, 181–188 (2016).
18. Sa ion, P. e al. Gene ic analysis o high bone mass cases om he BARCOS coho o Spanish pos menopausal women. PLoS One
9, e94607 (2014).
19. Bus aman e, M. e al. Polymo phisms in he in e leukin-6 ecep o gene a e associa ed wi h bone mine al densi y and body mass
index in Spanish pos menopausal women. Eu J Endoc inol 157, 677–684 (2007).
20. Yu, B. e al. Pa a hy oid ho mone induces di e en ia ion o mesenchymal s omal/s em cells by enhancing bone mo phogene ic
p o ein signaling. J Bone Mine Res 27, 2001–2014 (2012).
21. Ba y, E. L. e al. Gene ic a ian s in CYP2R1, CYP24A1, and VDR modi y he e icacy o i amin D3 supplemen a ion o inc easing
se um 25-hyd oxy i amin D le els in a andomized con olled ial. J Clin Endoc inol Me ab 99, E2133–2137 (2014).
22. Chen, G., Deng, C. & Li, Y. P. TGF-be a and BMP signaling in os eoblas di e en ia ion and bone o ma ion. In J Biol Sci 8, 272–288
(2012).
23. Richa ds, J. B. e al. Collabo a i e me a-analysis: associa ions o 150 candida e genes wi h os eopo osis and os eopo o ic ac u e.
Ann In e n Med 151, 528–537 (2009).
24. Be el, D., Wilson, M., Gabow, A., Ma ks, D. S. & Sande , C. The mic oRNA.o g esou ce: a ge s and exp ession. Nucleic acids
esea ch 36, D149–153 (2008).
25. Meye , L. R. e al. The UCSC Genome B owse da abase: ex ensions and upda es 2013. Nucleic acids esea ch 41, D64–69 (2013).
26. Vlachos, I. S. e al. DIANA miRPa h .2.0: in es iga ing he combina o ial e ec o mic oRNAs in pa hways. Nucleic Acids Res 40,
W498–504 (2012).
Acknowledgemen s
We would like o hank Na a o-Casado L. (lead esea che ) and Na a o-Ma ínez A. (collabo a o esea che )
o Albace e coho , C. Hospi ala io Uni e si a io de Albace e; Rica do Usa egui-Ma ín and Ta iana Ca anco-
Medina o hei suppo ing in sample collec ion om Hospi al Uni e si a io de Salamanca and o Layla Panach
(B.Sc.) who is a p edoc o al ellow (F.P.U. P og am) in INCLIVA (Valencia) o he echnical assis ance. We hank
o Ga cia Fon ana B, Luque Fe nandez I om UGC Endoc inologia y Nu ición, Hospi al Uni e si a io San
Cecilio, o hei collabo a ion in sample collec ing. The au ho s hank Elaine M. Lilly, PhD, o help ul ad ice
and c i ical eading o he manusc ip . This wo k was suppo ed by RETICEF (RD12/0043/0022), CIBERFES
(CB16/10/00245) and he G an FIS PI10/01537, PI12/02775, PI12/02582, and PI13/00116 (Ca los III Heal h
Ins i u e, Science and Inno a ion Minis y); FIS and G an s om he Gene ali a de Ca alunya (DIUE 2014 SGR
775), “664367/FOCUS” p ojec (EU Heal h P og amme; 2014–2020) and FEDER unds ha e suppo ed his
s udy.
Au ho Con ibu ions
L.D.U., N.G.G. and A.D.P. concei ed he concep o his pape . M.A.G.P., J.M.O., M.S.H., R.P.C., C.G.A.,
L.D.L. J.M.A., J.A.B.C., J.G.M., J.P.M., M.M.T., M.D.C., J.M., A.C., J.L.P.C. and X.N. pe o med pa ien and da a
collec ion. J.A.B.C., A.D.P. and N.G.G. designed he OSTEOMED2 ne wo k. M.R. managed he pa ien da abase
and pe o med s a is ical analyses. L.D.U., E.C.M. and N.G.G. se up he expe imen s and eco ded da a. All
au ho s pa icipa e in he analyses o da a, and discussed he esul s and con ex o he manusc ip . All au ho s
ead and app o ed he inal manusc ip .
Addi ional In o ma ion
Supplemen a y in o ma ion accompanies his pape a doi:10.1038/s41598-017-00641-7