Meta-analysis of gene-environment-wide association scans accounting for education level identifies additional loci for refractive error

ARTICLE Recei ed 9 Feb 2015 |Accep ed 10 Feb 2016 |Published 6 Ap 2016 Me a-analysis o gene–en i onmen -wide associa ion scans accoun ing o educa ion le el iden ifies addi ional loci o e ac i e e o Qiao Fan e al.# Myopia is he mos common human eye diso de and i esul s om complex gene ic and en i onmen al causes. The apidly inc easing p e alence o myopia poses a majo public heal h challenge. He e, he CREAM conso ium pe o ms a join me a-analysis o es single- nucleo ide polymo phism (SNP) main e ec s and SNP educa ion in e ac ion e ec s on e ac i e e o in 40,036 adul s om 25 s udies o Eu opean ances y and 10,315 adul s om 9 s udies o Asian ances y. In Eu opean ances y indi iduals, we iden i y six no el loci (FAM150B-ACP1,LINC00340,FBN1,DIS3L-MAP2K1,ARID2-SNAT1 and SLC14A2) associa ed wi h e ac i e e o . In Asian popula ions, h ee genome-wide significan loci AREG,GABRR1 and PDE10A also exhibi s ong in e ac ions wi h educa ion (Po8.5 105), whe eas he in e ac ions a e less e iden in Eu opeans. The disco e y o hese loci ep esen s an impo an ad ance in unde s anding how gene and en i onmen in e ac ions con ibu e o he he e ogenei y o myopia. Co espondence and eques s o ma e ials should be add essed o C.C.W.K (email: c.c.w[email p o ec ed]) o o S.-M.S. (email: [email protected]). #A ull lis o au ho s and hei a filia ions appea s a he end o he pape . DOI: 10.1038/ncomms11008 OPEN NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 1 Myopia, o nea sigh edness, has apidly eme ged as a global heal h conce n in he las h ee decades1.I is one o he leading causes o isual impai men and is associa ed wi h po en ially blinding ocula complica ions including myopic maculopa hy, glaucoma, ca a ac and e inal de achmen 2. E idence om amily and win s udies s ongly suppo s he he i abili y o myopia3. Es ima es o he he i abili y o he quan i a i e ai e ac i e e o ha e been epo ed o be as high as 90% ( e . 4). On he o he hand, he apid upsu ge o myopia in he las ew decades in many pa s o he wo ld is likely o be a consequence o li es yle changes such as he inc easing educa ional in ensi y, in pa icula in u ban Eas Asia5,6, and po en ially gene and en i onmen (G E) in e ac ions. Majo a emp s unde aken in genome-wide associa ion s udies (GWAS) o elucida e he gene ic de e mina ion o myopia and e ac i e e o ha e ecen ly led o he disco e y o 430 dis inc suscep ibili y loci7,8. Ne e heless, collec i ely hese gene ic a ian s a e es ima ed o explain o12% o pheno ypic a iance in e ac i e e o 7,8. As myopia is a esul o he combina ion o gene ic and en i onmen al ac o s, in e play be ween genes and en i onmen may accoun o a subs an ial p opo ion o he pheno ypic a iance. In ecen imes, we showed in e ac ions be ween educa ion and gene ic isk sco e o myopia de i ed om 26 known GWAS single-nucleo ide polymo phisms (SNPs) in he Ro e dam S udy9; he combined e ec o gene ic p edisposi ion and educa ion on he isk o myopia was subs an ially g ea e han an icipa ed om a simple sum o hese wo ac o s. A he gene le el, some genes such as SHISA6-DNAH9 ha e been shown o in e ac wi h educa ion le el and exhibi s ong gene ic e ec s o myopia among Asians wi h a leas highe seconda y educa ion10. In he cu en s udy, we demons a e ha new gene ic e ec s implica ed in myopia de elopmen could be unco e ed by s udying in e ac ions be ween gene ic a ian s and educa ion le el. In he con ex o he ae iology o e ac i e e o s, educa ion a ainmen is gene ally conside ed a su oga e measu e o accumula ed nea wo k ac i i y1. When iewing nea objec s, he eye gene a es ex a op ical powe h ough he p ocess o accommoda ion o ocus he image on he e inal plane, o main ain clea ision11. The e is an accommoda i e lag (less accommoda ion p oduced han needed) in many myopes, esul ing in a hype opic de ocus on he e ina o nea wo k, which has long been p oposed o p omo e eye g ow h1,12, bu whe he his occu s be o e o a e he onse o myopia in humans is less clea . The e ina has a cen al ole in he mechanism linking such isual inpu wi h eye g ow h and e ac i e de elopmen 13.Se e al neu o ansmi e s o molecules ha e been implica ed in his p ocess by animal s udies including dopamine, ace ylcholine, asoac i e in es inal pep ide, GABA (g-aminobu y ic acid) and glucagon14,15. Howe e , an o ganized amewo k o he e inal signalling mechanisms unde lying e ac i e e o de elopmen unde a ious en i onmen al condi ions emains o be elucida ed. Fac o ing in en i onmen al exposu es may enhance powe o he de ec ion o genes, especially in ci cums ances whe e a gene ic locus has a di e en ial e ec condi ional on specific en i onmen exposu es16. Gene–en i onmen -wide in e ac ion s udies (GEWIS) using a join me a-analysis (JMA) app oach on SNP main e ec s and SNP en i onmen in e ac ions ha e ecen ly been desc ibed17,18. This app oach has success ully iden ified six no el loci associa ed wi h as ing insulin and glucose accoun ing o in e ac ions wi h body mass index18. I also led o he iden ifica ion o wo no el loci o pulmona y unc ion ha did no eme ge om analyses based on he gene ic main e ec s alone19. The well- documen ed e ec s o educa ional a ainmen on myopia and e ac i e e o make he p oposed in e ac ion an excellen analy ical candida e o he GEWIS. The a ailabili y o la ge-scale GWAS sphe ical equi alen da a se s om he Conso ium o Re ac i e E o And Myopia (CREAM) makes G E in e ac ion analyses easible. To iden i y addi ional gene ic a ian s o e ac i e e o , we pe o med GEWIS-based analyses on 40,036 adul s o Eu opean ances y om 25 s udies and 10,315 adul s o Asian ances y om 9 s udies. We iden ified nine new loci using he JMA app oach, whe e h ee loci exhibi ed G E in e ac ion on e ac i e e o in Asians, including he GABA C ecep o subuni 1 gene GABRR1. Resul s Educa ional le el and i s main e ec s on sphe ical equi alen . The baseline cha ac e is ics o 50,351 pa icipan s om 34 s udies in ou me a-analysis a e shown in Table 1. A o al o 40,036 pa icipan s we e o Eu opean descen and 10,315 we e o Asian descen ; he age o he pa icipan s anged om 20 o 99 yea s. We g ouped indi iduals in o wo educa ional ca ego ies: a highe educa ion g oup ha included indi iduals who comple ed highe seconda y o uni e si y educa ion and a lowe educa ion g oup comp ising hose wi h lowe seconda y educa ion o below (see Me hods). Among Eu opeans, he p opo ions o pa icipan s in he highe educa ion g oup anged om 16.5% (FITSA20 and OGP Talana21) o 94.4% (AREDS22) wi h an a e age o 50.7% (Supplemen a y Table 1). In Asians, he p opo ions o indi iduals in he highe educa ion g oup anged om 6.7% (SiMES23) o 75.9% (Nagahama24) wi h an a e age o 30.0%. Ac oss all s udies, indi iduals in he highe educa ion g oup had a sphe ical equi alen e ac i e e o ha was on a e age 0.59 diop es (D) mo e myopic, o less hype opic, compa ed wi h hose in he lowe educa ion g oup (b¼0.59 D; 95% confidence in e al (CI): 0.64 o 0.55 D). High educa ion le el was associa ed wi h a wo old mo e myopic sphe ical equi alen in indi iduals o Asian as compa ed wi h Eu opean ances y (Asians: b¼1.09 D, 95% CI: 1.20 o 0.98 D; Eu opeans: b¼0.49 D, 95% CI: 0.54 o 0.44 D; Fig. 1). Among Asian s udies, we also obse ed he e ogenei y o educa ion e ec s o e ac i e e o . The educa ion e ec s on sphe ical equi alen in Singapo e Chinese we e significan ly la ge han ha in o he Asian s udies (Singapo e Chinese: b¼1.75 D, 95% CI: 1.92 o 1.58 D; o he Asian coho s: b¼0.60 D, 95% CI: 0.75 o 0.46 D). GEWIS in Eu opeans. A e s ingen quali y con ol (QC) fil e ing, B6 million SNPs in each s udy we e eligible o he genome-wide JMA es (Supplemen a y Table 2). The JMA o SNP main e ec s and SNP educa ion in e ac ions in 40,036 Eu opean Ances y indi iduals showed an associa ion wi h sphe ical equi alen a 12 p e iously implica ed loci (Fig. 2a, Supplemen a y Table 3 and Supplemen a y Fig 1). We also iden ified ou p e iously un epo ed loci associa ed wi h sphe ical equi alen achie ing genome-wide significance (P JMA o5.0 108;P he Z0.086; Table 2): FAM150B-ACP1, LINC00340,FBN1 and DIS3L-MAP2K1. The significan associa ion o JMA es ing a hese loci in Eu opeans was p ima ily d i en by SNP e ec s in bo h he lowe and highe educa ion s a a (4.40 108 P main 1.35 106and 7.61 1011 P main 1.75 106, espec i ely). SNP  educa ion in e ac ion was no significan (P in Z0.208). The es i- ma ed e ec sizes o SNP e ec s on sphe ical equi alen we e highly simila ac oss educa ion s a a. GEWIS in Asians. The JMA o sphe ical equi alen in 10,315 indi iduals om he Asians coho s iden ified genome-wide significan associa ion o h ee genes: AREG,GABRR1 and PDE10A (P JMA o5.0 108; Table 3 and Fig. 2b). ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 2NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions SNP educa ion in e ac ion e ec s associa ed wi h sphe ical equi alen we e obse ed a all h ee loci, wi h gene ic e ec s significan ly la ge wi hin pa icipan s who had a highe le el o educa ion compa ed wi h hose wi h a lowe educa ion le el: AREG ( s12511037, b in ¼0.89±0.14 D, P in ¼6.87 1011), GABRR1 ( s13215566, b in ¼0.56±0.14 D, P in ¼8.48 105) and PDE10A ( s12206610, b in ¼0.72±0.13 D, P in ¼2.32 108). The geno ype and pheno ype associa ions we e highly significan in he highe educa ion s a um (main gene ic e ec s, 1.97 1010 P main 8.16 108) bu we e conside ably weake in he lowe educa ion s a um (0.008 P main 0.243). The e was no e idence o in e -s udy he e ogenei y a index SNPs wi hin AREG,GABRR1 o PDE10A loci (Q- es : P he Z0.122). GABRR1 and PDE10A index SNPs we e no associa ed wi h sphe ical equi alen in Eu opean samples, o ei he he JMA es , SNP main e ec o SNP educa ion in e ac ion (Table 3). AREG SNP s12511037 was excluded in he me a-analysis o Eu opean s udies a e QC fil e ing; hence, a p oxy SNP, s1246413, in linkage disequilib ium (LD) wi h s12511037 in Asians ( 2¼0.97) was es ed bu no associa ed wi h sphe ical equi alen (P JMA ¼0.527; P in ¼0.176). The me a- eg ession including s udy-le el cha ac e is ics as co a ia es in he model confi med he he e ogenei y be ween popula ions o Eu opean and Asian ances y (GABRR1:P¼0.006; PDE10A:P¼0.0419; Supplemen a y Table 4). Fo PDE10A, besides e hnici y, a e age sphe ical equi alen o each s udy also explained he in e -s udy he e ogenei y o he in e ac ion e ec s (P¼0.025). We examined whe he he unde lying assump ion o G E independence held a hese h ee G E in e ac ion loci. We pe o med a me a-analysis o logis ic eg ession analysis o educa ion le el on AREG SNP s12511037, GABRR1 SNP s13215566 and PDE10A SNP s12296610, adjus ing o age, gende and popula ion s a ifica ion in Asian coho s (n¼10,315). Ou analysis did no e eal any significan associa ions be ween hese loci and educa ion le el (PZ0.102, P he Z0.170; Supplemen a y Table 5). Fu he mo e, he h ee loci we e no associa ed wi h educa ional a ainmen in a la ge me a-analysis o GWAS ecen ly conduc ed in Eu opean coho s25. Thus, ou G E esul s a e unlikely o be biased due o dependence be ween gene and educa ion. Table 1 | Cha ac e is ics o s udy pa icipan s. S udy NS udy yea Age (s.d.) Age ange Male (%) Sphe ical equi alen Eu opeans (n¼40,036) ALIENOR 509 2006–2008 79.2 (4.1) 73–93 43.2 0.98 (1.98) ALSPAC 1,865 1999–2000 45.9 (4.5) 32–59 0 0.76 (2.16) AREDS 1,842 1992 68.1 (4.7) 55–81 41.0 0.54 (2.15) BATS 383 1992–2013 24.8 (7.8) 20–67 41.3 0.67 (1.58) BMES 1,896 1992–2009 66.8 (8.9) 49–94 43.8 0.58 (1.94) CROATIA-Ko cula 807 2007–2008 56.2 (13.3) 25–94 34.9 0.13 (1.59) CROATIA-Spli 787 2008–2009 51.9 (13.0) 25–80 38.6 1.27 (1.59) DCCT 1,057 1982–1993 35.4 (5.8) 25–49 54.1 1.47 (1.80) EGCUT 904 2002–2013 56 (17.0) 25–99 38.8 0.33 (3.36) EPIC 1,083 2004–2011 68.8 (7.5) 50–88 43.8 0.34 (2.27) ERF 2,604 2002–2005 48.9 (14.4) 25–87 45.0 0.12 (2.03) FES 2,479 1973–1975/ 1989–1991 54.8 (9.3) 28–84 55.3 0.27 (2.37) FITSA 188 2000–2001 68.5 (3.3) 63–76 0 1.44 (2.08) GHS1 3,178 2007–2008 55.3 (10.9) 35–74 50.4 0.38 (2.47) GHS2 1,354 2008 54.6 (10.8) 36–74 49.6 0.39 (2.51) KORA 2326 2004–2006 55.1 (11.8) 35–84 49.4 0.26 (2.18) OGP Talana 456 2002 52.6 (16.3) 25–89 57.3 0.20 (0.24) ORCADES 1,124 2009 56.5 (13.2) 29–92 39.1 0.10 (2.07) RAINE 348 2010–2012 20.4 (0.34) 20–22 49.1 0.03 (1.29) RS1 5,702 1991–1993 68.7 (8.7) 55–99 41.0 0.83 (2.55) RS2 2,021 2000–2002 64.3 (7.9) 55–95 46.0 0.48 (2.51) RS3 2,918 2006–2009 56.9 (6.6) 45–86 44.0 0.28 (2.60) TwinsUK 2,154 1998–2010 53.8 (11.4) 25–84 8.4 0.96 (2.78) WESDR 561 1979–2007 31.7 (7.0) 25–65 50.3 1.65 (2.07) YFS 1,490 2011 41.9 (5.0) 34–49 44.6 1.09 (2.16) Asians (n¼10,315) BES 589 2006–2011 62.1 (8.5) 50–90 34.0 0.06 (1.86) Nagahama 723 2008–2010 49.2 (15.2) 30–74 33.6 1.93 (2.46) SCES-610K 1,710 2009–2011 57.5 (7.0) 44–84 51.6 0.72 (2.69) SCES-OmniE 543 2011–2012 59.3 (8.9) 46–83 51.2 0.89 (2.74) SiMES 2,256 2004–2006 46.8 (10.2) 40–80 49.1 0.03 (1.81) SINDI 2,088 2007–2009 55.8 (8.8) 43–84 51.5 0.04 (2.07) SP2-1M 811 1992–1998 46.8 (10.2) 25–80 62.3 1.80 (2.84) SP2-610 854 1992–1998 48.4(11.3) 25–82 19.6 1.44 (2.89) STARS 741 2007–2009 38.5 (5.2) 26–58 52.4 2.80 (2.85) ALIENOR, an ioxydan s, lipids essen iels, nu i ion e maladies oculaiRes; ALSPAC, a on longi udinal s udy o pa en s and child en; AREDS, age- ela ed eye disease s udy; BATS, B isbane adolescen wins s udy; BMES, blue moun ains eye s udy; DCCT, diabe es con ol and complica ions ial; EGCUT, es onian genome cen e o he uni e si y o Ta u; EPIC, EPIC-No olk eye s udy; ERF, e asmus ucphen amily s udy; FES, F amingham eye s udy; FITSA, finnish win s udy on aging; GHS, Gu enbe g heal h s udy; KORA, coope a i e heal h esea ch in he egion o Augsbu g; OGP Talana, oglias a gene ic pa k, alana s udy; ORCADES, o kney complex disease s udy; RAINE, RAINE eye heal h s udy; RS, Ro e dam s udy; TwinsUK, Twins UK s udy; WESDR, Wisconsin epidemiologic s udy o diabe ic e inopa hy; YFS, young finns s udy; BES, Beijing eye s udy; SCES, Singapo e Chinese eye s udy; SiMES, Singapo e Malay eye s udy; SINDI, Singapo e Indian eye s udy; SP2, Singapo e p ospec i e s udy p og am; STARS, s abismus, amblyopia and e ac i e e o s udy in p eschool singapo ean child en. s.d., s anda d de ia ion; age in yea s; sphe ical equi alen in diop es. De ails o each s udy coho a e desc ibed in Supplemen a y No e 1. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 ARTICLE NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 3 We also e alua ed he associa ion o sphe ical equi alen in Asian coho s o ou loci iden ified om Eu opean popula ions. Two o hem showed significan associa ions in he JMA es : FAM150B-ACP1 (P JMA ¼0.031) and DIS3L-MAP2K1 (P JMA ¼0.0042; Table 2). The SNP e ec sizes in lowe and highe educa ion s a a in Asians we e simila a FAM150B- ACP1. The signal a he DIS3L-MAP2K1 locus was mainly d i en by SNP educa ion in e ac ion in Asians (P in ¼7.95 104), whe eas he in e ac ion e ec was no s a is ically significan in Eu opeans (P in ¼0.208). Combined GEWIS o all coho s. We subsequen ly conduc ed a JMA in he combined da a including bo h he Eu opean and Asian pa icipan s o all 34 s udies. This analysis e ealed wo addi ional SNPs: ARID2-SNAT1 (P JMA ¼4.38 108) and SLC14A2 (P JMA ¼2.54 108). Bo h loci showed sugges i e associa ion wi h sphe ical equi alen in Eu opean coho s, wi h he same di ec ion o e ec and simila e ec sizes in Asian coho s (Table 2). We also de ec ed genome-wide significan associa ions wi h sphe ical equi alen o 17 known loci8 iden ified in ou p e ious CREAM GWAS (Supplemen a y Table 3). The egional plo s o he iden ified no el loci a e p esen ed in Supplemen a y Fig 2. Gene and educa ion in e ac ions o GWAS known loci.We also e alua ed he in e ac ions be ween educa ion and p e iously epo ed gene ic associa ion wi h sphe ical equi alen a 39 loci iden ified om ecen wo la ge GWAS s udies7,8. Two SNP educa ion in e ac ions we e nominally significan (Supplemen a y Table 6): TJP2 in Eu opeans ( s11145488, P in ¼6.91 103) and SHISA6-DNAH9 in Asians ( s2969180, P in ¼4.02 103). In gene al, he index SNPs es ed a 39 loci had la ge SNP educa ion in e ac ion e ec on sphe ical equi alen in Asians e sus Eu opeans (me a- eg ession P o old changeso0.001; Supplemen a y Fig. 3). Fo 20 SNPs wi h he same di ec ion o he in e ac ion e ec , he magni udes o in e ac ion e ec s we e ou old la ge on a e age in Asians han in Eu opeans (P¼0.003). Gene and nea wo k in e ac ions o h ee iden ified loci. High educa ion le els may eflec an es ima o o he g ea e accumula i e e ec o nea wo k26,27. We hus examined whe he he e was e idence o SNP nea wo k in e ac ions associa ed wi h sphe ical equi alen a he h ee loci (AREG,GABRR1 and PDE10A) in paedia ic coho s (SCORM28, Guangzhou Twins29 and ALSPAC30; combined n¼5,835; Supplemen a y Table 7). Ten a i e suppo o a SNP nea wo k in e ac ion was obse ed o PDE10A ( s12206610, P in ¼0.032, P he ¼0.658), wi h he s onge gene ic e ec in child en spending mo e hou s on eading, w i ing o compu e use. Weake suppo o an in e ac ion was no ed a GABRR1 ( s13215566, P in ¼0.309, P he ¼0.655), al hough he di ec ion o me a-analysed in e ac ion e ec was la gely consis en ac oss paedia ic s udies wi h ha obse ed in adul s. We did no obse e he in e ac ion a AREG ( s12511037, P in ¼0.795, P he ¼0.062). Gene exp ession in human issues. Using he Ocula Tissue Da abase31, we examined he exp ession o he associa ed genes in 20 no mal human dono eyes. The majo i y o genes iden ified we e exp essed in human e ina, scle a, cho oid o e inal pigmen epi helium (RPE) (Supplemen a y Table 8). Among hese genes, GABRR1,ACP1 and SNAT1 had he highes exp ession in he e ina. The P obe Loga i hmic In ensi y E o -no malized messenge RNA exp ession le els in he e ina anged om 121.66 o 236.69. O no e, MAP2K1 was widely exp essed in he e ina, scle a and cho oid/RPE.  (95% CI) P All: –0.59 (–0.64, –0.55) <0.0001 Eu opeans: –0.49 (–0.54, –0.44) <0.0001 Asians: –1.09 (–1.20, –0.98) <0.0001 Eu opeans Asians -coe icien –3.0 –2.5 –2.0 –1.5 –1.0 –0.5 0.0 0.5 DCCT OGP Talana BMES CROATIA-Spli CROATIA-Ko cula EPIC BATS EGCUT WESDR ORCADES RAINE ALIENOR FITSA AREDS KORA GHS2 GHS1 RS1 RS2 RS3 Nagahama BES SiMES SlNDI SP2-610 SP2-1M STARS SCES-610K SCES-OmniE ERF YFS FES TwinsUK ALSPAC Figu e 1 | Fo es plo o educa ion main e ec s on sphe ical equi alen ac oss s udies. The b-coe ficien ep esen s he di e ences o diop es in e ac i e e o compa ing indi iduals in highe educa ion g oup e sus lowe educa ion g oup in Eu opeans (n¼40,036), Asians ( n¼10,315) and all s udies (n¼50,351). The s udies a e so ed by e ec size o educa ion on sphe ical equi alen wi hin Eu opeans and Asians s udies, espec i ely. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 4NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions Discussion This s udy ep esen s he mos comp ehensi e genome-wide scan o gene and educa ion in e ac ions o da e, o e ac i e e o . He e we iden ified no el gene ic loci associa ed wi h e ac i e e o by es ing he join con ibu ion o SNP main e ec s and SNP educa ion e ec s in la ge mul i-e hnic popula ions. Th ee loci (AREG,GABRR1 and PDE10A) showed s ong in e ac ions wi h educa ion in popula ions o Asian descen , wi h la ge gene ic e ec s wi hin pa icipan s who had a highe le el o educa ion compa ed wi h hose wi h a lowe educa ion le el; no in e ac ions achie ed s a is ical significance in Eu opeans o op JMA associa ions o known myopic loci. Apa om confi ming known associa ions a 17 p e ious published loci, we iden ified six new loci (FAM150B-ACP1,LINC00340,FBN1,DIS3L-MAP2K1, ARID2-SNAT1 and SLC14A2) significan ly associa ed wi h sphe ical equi alen using he combined mul i- acial coho . A ecen me a-analysis o GWAS in mul i-e hnic popula ions comp ises 32 s udies (n¼45,756) om CREAM and a la ge GWAS in Eu opeans (n¼45,771) ha e epo ed a o al o 39 gene ic loci associa ed wi h e ac i e pheno ypes7,8. The cu en genome-wide me a-analysis included B5,000 mo e subjec s han he p e ious GWAS o main e ec s. We iden ified nine addi ional no el loci using he JMA app oach. These loci can be placed wi hin he biological con ex o he isually e oked signalling cascade ha begins in he e ina and media es scle a emodelling32. The newly iden ified genes a e in ol ed in e inal neu o ansmission (GABRR1 and SNAT1), ex acellula ma ix emodelling (FBN1,MAP2K1 and AREG), ci cadian hy hm (PDE10A) and pla ele -de i ed g ow h ac o ecep o signalling (ACP1) (Supplemen a y Table 9). Ne wo k analysis e ealed ha mos o he no el genes may end o be co-exp essed and co-localized wi h he known myopia suscep ibili y genes h ough mul iple biological ne wo ks such as LAMA2,GJD2, RASGRF1,BMP3,RDH5,ZMAT4,RBFOX1,RDH5 and so on (Supplemen a y Fig. 4). Ou da a, in line wi h p e ious findings, subs an ia e he assump ion o he e ogenei y in he molecula mechanisms in ol ed in e ac i e e o and myopia. Among he no el loci, GABRR1 on ch omosome 6q15 (53 kb), encoding GABA C ecep o subuni 1, is an in e es ing unc ional candida e sugges i e o a ole in myopia de elopmen . Modula ion o synap ic plas ici y ia GABA-media ed inhibi ion would be well placed o al e he ‘gain’ o he isually guided CD55 FAM150B-ACP1 CHRNG LINC00340 KCNQ5 LAMA2 TOX TJP2 RDH5 SIX6 GJD2 ARID2 -SNAT1 FBN1 DIS3L-MAP2K1 A2BP1 DNAH9 KCNJ2 SLC14A2 Eu opean ances y popula ions 1 2 3 4 5 6 7 8 9 10 11 12 13 15 16 18 20 221714 20 15 10 5 0 AREG GABRR1 PDE10A Asian popula ions Ch omosome 10 8 6 4 2 0 1234567891011121315161820221714 a b –Log10 (P) –Log10 (P) Figu e 2 | Manha an plo s o log 10 (P) o he JMA on SNP main e ec s and SNP educa ion e ec s on sphe ical equi alen in (a) Eu opean ances y popula ions and (b) Asian popula ions. The ho izon al ed line indica es he genome-wide significance le el o P JMA o5108. The ho izon al blue line indica es he sugges i e significance le el o P JMA o1105. No el loci eaching genome-wide significance a e labelled in ed and known loci a e in g ey. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 ARTICLE NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 5 eedback sys em con olling e ac i e de elopmen 33. The lead SNP s13215566 in GABRR1, oge he wi h se en SNPs wi hin he LD block ( 2Z0.8), a e in onic, po en ially a ec ing egula o y mo i s (such as z p128 and gcm1) ha may influence ansc ip ional egula ion (Supplemen a y Table 10). The a ian s13215029, in LD ( 2¼1) wi h s13215566, is associa ed wi h cis-ac ing exp ession o GABRR1 (P¼2.3 104)inskin issues (Supplemen a y Table 9)34. A ecen pha macological s udy p o ided e idence ha e inal dopamine gic and GABAe gic neu o ansmi e s play a subs an ial ole in he modula ion o e ac i e de elopmen in o m-dep i a ion myopia35–37.S one e al.35 ha e epo ed ha an agonis s o GABA-A, -B and -C ecep o s inhibi ed o m-dep i a ion myopia in chicks, wi h g ea es e ec in he equa o ial dimension. GABA ecep o s, exp essed in bipola and ganglion neu on cells, also in e ac wi h dopamine pa hways in he e ina36. A ecen p o eomics s udy de e mined ha le els o GABA anspo e -1 we e significan ly educed in myopic mu ine e ina a e a opine ea men , implying ha GABA signalling is in ol ed in he an i-myopic e ec s o a opine37. Al oge he , hese da a sugges ha GABA C ecep o 1 may egula e he de elopmen o myopia h ough unc ional eedback om RPE o neu on cells in he e ina. Fu he s udies a e needed o in es iga e he e ec o gene ic dele ion o GABRR1 on e ac i e eye de elopmen and he ole o he GABAe gic pa hway in myopia de elopmen using gene knockou mice. The e o e, ou esul in humans is in line wi h animal expe imen s, suppo ing he no ion ha he GABAe gic neu o ansmi e signalling pa hway in he e ina could be a po en ial ac o in he p og ession o myopia. SNP s10889855 on ch omosome 6 is an in onic a ian wi hin he ARID2 gene (AT- ich in e ac i e domain 2) and 500 kb downs eam o SNAT1 (solu e ca ie amily 38, membe , alias SLC38A1). SNAT1 is a anspo e o glu amine, a p ecu so o GABA38. I is also highly exp essed in human e ina. In ou p e ious me a-analysis in CREAM8, we iden ified a ian s in ano he glu ama e ecep o gene GRIA4 (encoding glu ama e ecep o , iono opic); al oge he , cu en e idence suppo s he no ion ha e inal neu o ansmi e s GABA and glu amine may be in ol ed in he e ac i e de elopmen . The s onges associa ion signal o gene and en i onmen in e ac ions was om s12511037, loca ed 14 kb downs eam he AREG gene (amphi egulin). AREG is a ligand o he epide mal g ow h ac o ecep o p omo ing he g ow h o no mal epi helial cells, which is c i ical o cell di e en ia ion and p oli e a ion such as eg ow h o he wounded co nea39. A link has been ound be ween he musca inic ace ylcholine ecep o s, dominan in myopia p og ession, and he epide mal g ow h ac o ecep o in musca inic sys em40,41. Ano he no el associa ion, s16949788 on ch omosome 15, de i es om a egion ha spans DIS3L and MAP2K1.MAP2K1 encodes mi ogen-ac i a ed p o ein kinase 1, which binds o musca inic ecep o s du ing p oli e a ion42 and inhibi s he p oli e a ion o human scle al fib oblas s exposed o all- ans e inoic acid43. All- ans e inoic acid is a modula o o ocula g ow h, inhibi ing he p oli e a ion o human scle al fib oblas s44. FBN1 (Fib illin 1), a membe o he fib illin amily, encodes a la ge ex acellula ma ix glycop o ein. Mu a ions in FBN1 cause Table 2 | Six gene ic loci associa ed wi h sphe ical equi alen om he JMA in he Eu opean popula ions and combined analyses. SNP (Ch :BP) Gene Allele FREQ Subg oup Eu opeans (n¼40,036) Asians (n¼10,315) All (n¼50,351) bP- alue P he bP- alue P he bP- alue P he s60843830 (2:286756) FAM150B- ACP1 C/G 0.66/0.74 JMA 3.71 10 80.086 0.031 0.980 1.27 1090.395 Lowe educa ion 0.11 4.73 1080.09 0.010 0.10 1.65 109 Highe educa ion 0.09 1.75 10 60.06 0.509 0.09 9.83 107 s10946507 (6:22100367) LINC00340 (6p22.3) A/G 0.47/0.16 JMA 3.07 1080.213 0.433 0.396 2.24 10 80.249 Lowe educa ion 0.08 7.08 10 70.04 0.313 0.08 6.13 10 7 Highe educa ion 0.09 1.19 1080.08 0.450 0.09 1.20 10 8 s8023401 (15:48703823) FBN1 G/A 0.87/0.95 JMA 1.66 10 90.180 0.572 0.979 2.85 10 90.495 Lowe educa ion 0.15 4.40 10 80.06 0.304 0.13 8.17 10 8 Highe educa ion 0.16 7.61 10 11 0.03 0.828 0.14 2.02 10 9 s16949788 (15:66590037) DIS3L- MAP2K1 T/C 0.91/0.94 JMA 1.34 1080.721 0.0042 0.219 2.19 10 80.245 Lowe educa ion 0.15 1.35 1060.21 0.103 0.13 4.88 10 6 Highe educa ion 0.17 1.89 10 90.59 0.014 0.16 3.90 10 9 s10880855 (12:46144855) ARID2-SNAT1 T/C 0.51/0.43 JMA 7.83 1070.790 0.019 0.779 4.38 10 80.867 Lowe educa ion 0.09 1.26 10 70.06 0.067 0.09 8.42 109 Highe educa ion 0.07 1.60 10 50.16 0.033 0.07 3.55 10 6 s10853531 (18:42824449) SLC14A2 G/A 0.80/0.83 JMA 7.82 10 60.052 0.0023 0.812 2.54 1080.111 Lowe educa ion 0.11 1.27 10 60.15 9.01 10 40.11 3.38 10 9 Highe educa ion 0.08 2.12 10 60.11 0.288 0.09 7.14 10 6 b,b-coe ficien co esponds o he e ec in sphe ical equi alen (diop es) o 1 addi ional copy o he isk allele in he highe o lowe educa ion g oup. FREQ, allele equency o he isk allele in Eu opean/Asian coho s; JMA, join me a-analysis on SNP e ec and SNP educa ion in e ac ion e ec on sphe ical equi alen ; P he ,P- alue o he es o he e ogenei y a each SNP; SNP, single-nucleo ide polymo phism. Allele, isk allele/o he allele. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 6NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions Ma an’s synd ome, a diso de o connec i e issue a ec ing he ocula , skele al and ca dio ascula sys ems45. As a candida e gene o myopia, a emp s o s udy i s associa ion wi h myopia p e iously p oduced inconclusi e esul s46,47, p obably owing, in pa , o unde powe ed s udies wi h insu ficien sample sizes. Using da a om a la ge mul i-e hnic popula ion, ou esul s suppo he ole o FBN1 in myopia de elopmen . The genome-wide significan SNPs om he JMA app oach did no exhibi any in e ac ions wi h educa ion in Eu opeans, in con as o he significan in e ac i e e ec among Asians. In pa icula , he G E in e ac ions a AREG,GABRR1 and PDE10A we e only e iden in Asian popula ions. The e a e a numbe o possible easons o he obse ed di e ences. Fi s , he a ia ion o LD pa e ns and join e ec s o gene ic a ian s migh a ec he ans e abili y o G E signals ac oss popula ions. Simila LD pa e ns we e seen a GABRR1 and PDE10A egions ac oss popula ions whe eas a long s e ch o LD flanking AREG was p esen only in Asian popula ions (Supplemen a y Fig. 5). As he ue causal a ian s ans e able ac oss popula ions a e p obably no implica ed in ou s udy, he iden ified no el myopia isk loci p o ide a much-needed s a ing poin o ollow-up and unc ional downs eam analyses. Second, we used educa ion le el in adul s as a su oga e measu e o he unde lying isk ac o s o influencing e ac i e de elopmen . Ideally, nea wo k in ensi y would be measu ed p ospec i ely in child en be o e he onse o myopia. S udies included in ou analyses do no ha e addi ional da a on ele an childhood exposu es. Thus, he bes a ailable su oga e measu e o cumula i e nea wo k exposu e in adul coho s is educa ional le el. We ne e heless belie e ha educa ion is a highly eliable p oxy o he ele an exposu es unde lying e ac i e de elopmen and is uni e sally associa ed wi h e ac i e e o in ou s udy. Thi d, he di e ences o G E in e ac ions in Asian e sus Eu opeans may eflec quan i a i e di e ences in nea wo k in ensi y du ing childhood. Fo example, 6- and 7- yea -old child en in England and Aus alia epo ed less nea wo k ac i i y ou side o school (1.0–2.3 h pe day)48,49 compa ed wi h child en in Singapo e and China (2.7–3.5 h pe day)50,51. A simila end was obse ed in olde child en48,52–55. We hus specula e ha he o al exposu e o nea wo k ac i i y may be g ea e in Eas Asians compa ed wi h Eu opean-de i ed popula ions wi h he same le els o educa ion; hence, G E in e ac ion es ima es would end o be infla ed in Asian popula ions compa ed wi h Eu opean g oups. Fou h, o he en i onmen al ac o s such as ou doo ac i i ies could also in e ac wi h genes. Eas Asian child en end o ha e less exposu e o ou doo ac i i ies compa ed wi h hei Eu opean pee s56. Howe e , he majo i y o adul coho s did no epo ime ou doo s and hus could no be accoun ed o in he cu en s udy. Finally, he popula ion mean o e ac i e e o is less myopic in Eu opeans (0.10 D) e sus Asians ( 0.60 D). O no e, o he p e iously known myopia loci, he magni udes o in e ac ion e ec s we e ou old la ge on a e age in Asians han in Eu opeans (Supplemen a y Fig. 2). The impac o G E in e ac ions may be seen a ce ain se e i y le els o myopia. The isk alleles o s12511037 in AREG, s1321556 in GABRR1 and s12206610 in PDE10A had no o weak influence on myopic shi in he lowe educa ion g oup compa ed wi h he highe educa ion g oup. This sugges s ha he he edi a y p edisposi ion o myopia could be la en o he isk allele ca ie s, i hey a e less exposed o he myopiagenic en i onmen associa ed wi h high- le el educa ion. A lack o s ong SNP nea wo k associa ions in child en migh be due o he inadequa e s a is ical powe in paedia ic coho s o ela i ely small sample sizes, o he possibili y ha en i onmen al isk exposu es o he han nea wo k migh unde lie he SNP educa ion in e ac ion seen in he adul Asian samples. In summa y, we iden ified nine no el loci associa ed wi h e ac i e e o in a la ge mul i-e hnic coho s udy by GEWIS app oach. Ou da a p o ide e idence ha specific gene ic a ian s in e ac wi h educa ion, o influence e ac i e de elopmen , and u he suppo a ole o GABA neu o ansmi e signalling in myopia de elopmen . These findings p o ide p omising Table 3 | Th ee gene ic loci associa ed wi h sphe ical equi alen wi h a significan SNP educa ion in e ac ion in Asians and esul s in Eu opean popula ions. SNP (Ch :BP) Gene Allele FREQ Subg oup Asians (n¼10,315) Eu opeans (n¼40,036) bP- alue P he bP- alue P he s12511037* (4:75334864) AREG C/T 0.91/0.95 Lowe educa ion 0.07 0.243 0.05 0.323 Highe educa ion 0.70 1.97 10 10 0.03 0.579 SNP educa ion 0.89 6.87 10 11 0.704 0.02 0.176 0.284 JMA 5.55 1010 0.405 0.527 0.186 s13215566 (6:89918638) GABRR1 C/G 0.94/0.84 Lowe educa ion 0.13 0.030 0.03 0.258 Highe educa ion 0.68 1.46 10 80.01 0.817 SNP educa ion 0.56 8.48 1050.134 0.02 0.459 0.457 JMA 3.81 1080.122 0.502 0.630 s12206610 6:166016800 PDE10A C/T 0.90/0.87 Lowe educa ion 0.16 0.008 0.01 0.759 Highe educa ion 0.59 8.16 1080.01 0.810 SNP educa ion 0.72 2.32 1080.920 0.002 0.421 0.111 JMA 9.21 1090.902 0.954 0.305 b(highe educa ion/lowe educa ion), b-coe ficien co esponds o he e ec in sphe ical equi alen (diop es) o 1 addi ional copy o he e ec allele in he highe /lowe educa ion g oup; b (SNP educa ion), b-coe ficien co esponds o he di e ence in sphe ical equi alen (diop es) o 1 addi ional copy o he e ec allele in he highe e sus lowe educa ion g oup; FREQ, allele equency o he e ec allele in Asian/Eu opean coho s; JMA, join me a-analysis on SNP e ec and SNP educa ion in e ac ion e ec on sphe ical equi alen ; LD, linkage disequilib ium; P he ,P- alue o he es o he e ogenei y; SNP, single-nucleo ide polymo phism. band P- alues o SNP educa ion in e ac ion we e calcula ed by he me a-analysis o conduc ing a 1d Wald’s es o single in e ac ion pa ame e . Allele is lis ed as e ec allele/o he allele. *SNP s12511037 was no p esen in Eu opean s udies a e quali y con ol. He e we p esen he esul s o a p oxy SNP s1246413 (T/G, equency o isk allele T¼0.95) in LD wi h s12511037 ( 2¼0.97). NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 ARTICLE NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 7 candida e genes o ollow-up wo k and may lead o new gene ic a ge s o he apeu ic in e en ions on myopia. Me hods S udy pa icipan s.Thi y- ou s udies om membe s o CREAM, comp ising 40,036 indi iduals o Eu opean ances y om 25 s udies and 10,315 indi iduals o Asian ances y om 9 s udies, we e made a ailable o his analysis (Table 1 and Supplemen a y No e 1). Indi iduals aged o20 yea s we e excluded and so we e hose who had unde gone ca a ac su ge y, lase o o he in a-ocula p ocedu es ha could al e e ac ion. Many o hese s udies we e also included in he p e ious CREAM GWAS on sphe ical equi alen 8. All s udies adhe ed o he ene s o he Decla a ion o Helsinki and we e app o ed by hei local esea ch e hics commi ees. The exac names o he Ins i u ional Resea ch Boa d commi ees can be ound unde Supplemen a y No e 2. All pa icipan s p o ided a signed consen o m be o e he s a o he s udy. Pheno yping and educa ion le els.All pa icipan s unde wen oph halmological examina ions (Supplemen a y Table 1). Non-cycloplegic e ac ion was measu ed by au o e ac ion and/o subjec i e e ac ion. Sphe ical equi alen was calcula ed as he sphe e powe plus hal o he cylinde powe o each eye. The mean sphe ical equi alen o he igh and le eyes was used as a quan i a i e ou come. When da a om only one eye was a ailable, he sphe ical equi alen o ha eye was used. Fo each s udy, he pa icipan s epo ed he highes le el o educa ion achie ed o he yea s o schooling h ough a sel - epo ed ques ionnai e, o in an in e iew. We dicho omized educa ion o all pa icipan s in each s udy. The highe educa ion g oup consis ed o hose who had achie ed he highes educa ional le el o A-le els, high school (highe seconda y educa ion), oca ional aining ( o example, diploma), uni e si y deg ee o hose wi h Z12 yea s spen in o mal educa ion (beginning om fi s g ade). Those who had achie ed he highes educa ional le el o O-le el, middle school (lowe seconda y educa ion) o hose wi h o12 yea s o o mal educa ion we e classified in o he lowe educa ion g oup. I bo h numbe o o mal s udy yea s and educa ion le els we e a ailable in he coho , we classified pa icipan s based on yea s o o mal educa ion. Fo he ou coho s o ela i ely young Eu opean pa icipan s (BATS, DCCT, RAINE and WESDR; o al n¼2,349), almos all o hem had comple ed 12 o mo e yea s o schooling. We hus chose o ca ego ize indi iduals wi h e ia y o uni e si y educa ion as he highe educa ion g oup in hese s udies. Sensi i i y analysis excluding hese ou coho s did no app eciably change ou me a-analysis esul s. Geno yping and impu a ion.De ailed in o ma ion on he geno yping pla o ms and QC p ocedu es o each s udy is p o ided in Supplemen a y Table 2 and Supplemen a y No e 1. Each s udy applied s ingen QC fil e s o GWAS. In gene al, duplica e DNA samples, indi iduals wi h low call a e (o95%), gende misma ch o e hnic ou lie s we e excluded. SNPs we e excluded i low geno yping call a e (45% missingness), monomo phic SNPs, wi h mino allele equency (MAF) o1% o in Ha dy–Weinbe g disequilib ium (Po106). A e QC fil e ing, he a ay geno ypes o each s udy we e impu ed using he 1000 Genomes P ojec da a as e e ence panels (build 37, phase 1 elease, Ma ch 2012) wi h he so wa e Minimac57 o IMPUTE2 ( e . 58). App oxima ely six million SNPs ha passed impu a ion quali y h esholds (MACH: 240.5 o IMPUTE in o sco e 40.5) and wi h MAF Z5% we e eligible o he me a-analysis (Supplemen a y Table 2). S a is ical models.Fo each s udy, a linea eg ession model o each geno yped o impu ed SNP was cons uc ed wi h he mean sphe ical equi alen as he ou - come. We assumed an addi i e gene ic model whe e he numbe o isk alleles is an o dinal a iable (0, 1 and 2) o di ec ly geno yped SNPs o a con inuous a iable o allele dosage p obabili y anging om 0 o 2 o impu ed SNPs. The p ima y analy ic model included SNP, educa ion and SNP educa ion in e ac ion e m, as well as age and sex as co a ia es. Addi ional adjus men s o he op p incipal componen s o genomic ma ke a ia ions we e pe o med in indi idual s udies when applicable ( ha is, when he e was e idence o popula ion s a ifica ion). We used he ollowing addi i e gene ic model o es o a join e ec o SNP (b SNP ) and SNP educa ion in e ac ion (b SNP educa ion ) on mean sphe ical equi alen : Y¼b0þbSNPSNP þbeduca ioneduca ion þbSNPeduca ionSNPeduca ion þbCco þe ð1Þ whe e Yis he mean sphe ical equi alen and educa ion is a dicho omous a iable (0 ¼lowe educa ion g oup and 1 ¼highe educa ion g oup); co is a se o co a ia es such as age, sex and fi s op fi e p incipal componen s when applicable. Fo amily-based s udies, he kinship ma ix was es ima ed empi ically om he SNP da a and included as a andom e ec in he gene alized mixed model59.To es an e ec o SNP educa ion in e ac ion, we assessed b SNP educa ion om equa ion (1). The linea eg ession analyses in each s udy we e conduc ed wi h Quickes o P obABEL o he un ela ed samples and MixABEL o amily-based da a. The command ‘ obus ’ was used in he abo e so wa e o calcula e he obus (‘sandwich’, Hube -Whi e) s.e. o b SNP and b SNP educa ion , and e o co a iance o b, o co ec he po en ial infla ion o alse posi i e a e o he in e ac ion P- alue60. In addi ion, each s udy also es ed he main e ec o educa ion on sphe ical equi alen by adjus ing o age and gende using he linea eg ession model: Y¼b0þbeduca ioneduca ion þbCco þeð2Þ whe e he defini ion o each a iable is he same as in equa ion (1). We pe o med me a-analysis o he educa ion e ec s on mean sphe ical equi alen in Eu opeans, Asians (Singapo e Chinese e sus o he s) and combined da a using a fixed-e ec model wi h in e se- a iance weigh ing (R package ‘me a’). GEWIS join me a-analyses.We adop ed he JMA app oach17,61, o simul aneously es bo h SNP main e ec s and SNP educa ion in e ac ions o sphe ical equi alen wi h a fixed-e ec model, using SNP and SNP educa ion eg ession coe ficien s (b SNP and b SNP educa ion , espec i ely) and a b’s co a iance ma ix om each s udy. A Wald’s s a is ic, ollowing a w2-dis ibu ion wi h wo deg ees o eedom, was used o es he join significance o he b SNP and b SNP educa ion . The JMA was pe o med wi h METAL62, as p e iously desc ibed by Manning e al.61. A Coch an’s Q- es was used o assess he e ogenei y o he b-coe ficien s ac oss s udies o he SNP and in e ac ion e ec s. To es o in e ac ion be ween he SNP and educa ion, we conduc ed a seconda y me a- analysis o he SNP educa ion in e ac ion e ec s o sphe ical equi alen (b SNP educa ion , one deg ee o eedom), wi h a fixed-e ec s model using in e se- a iance weigh ing in METAL; his is a adi ional me a-analysis o in es iga e SNP educa ion in e ac ions pe se. E ec s and s.e. o he SNP e ec on sphe ical equi alen in he lowe educa ion g oup (b SNP ) and highe educa ion g oup (b SNP þb SNP educa ion ) we e de i ed om he JMA ou pu 61. We used he P- alue o 5 10 8as a significan h eshold o JMA es . Fo he SNP and SNP educa ion e ec s o he iden ified op loci, he P- alue h eshold o significance was se a 0.0055 ¼0.05/9 (9 index SNPs unde lying analyses). We pe o med a me a- eg ession o explo e sou ces o he e ogenei y in ou me a-analysis o h ee loci showing G E in e ac ions (R package ‘me a o ’). Me a- eg ession included he ollowing s udy-specific a iables as co a ia es: s udy sample size, p opo ion o indi iduals in he highe educa ion g oup, a e age sphe ical equi alen , educa ion main e ec s on sphe ical equi alen (highe educa ion le el e sus lowe ), e hnici y (Asian e sus Eu opean), s udy design (independen samples e sus amily-based s udies), s udy yea and a e age age. Me a- eg ession was also conduc ed o es he old changes o he in e ac ion b-coe ficien s in Asians e sus Eu opeans o he 39 known myopia loci. The s udy-specific genomic con ol infla ion ac o s l gc o he join es o SNP and in e ac ion e ms anged om 1.009 o 1.125 wi h an a e age o 1.019 (Supplemen a y Table 2), calcula ed by he a io o he obse ed median w2di ided by he expec ed median o he 2d w2-dis ibu ion (1.382). Genomic con ol co ec ion was applied o each indi idual s udy63. Fo s udies o small sample sizes (no500) wi h l gc 41.05, we u he , be o e s a ing he me a-analysis, excluded SNPs showing significan join P- alue o1105bu nei he he SNP no SNP educa ion e ec s suppo ed such an associa ion. Quan ile–quan ile plo s showed only modes infla ion o he es s a is ics in he JMA es (Eu opeans: l gc ¼1.081; Asians: l gc ¼1.053; Combined: l gc ¼1.092; Supplemen a y Fig. 1), simila o p e ious GEWIS s udies wi h compa able sample sizes18,19.Weexcludeda small numbe o ma ke s in he me a-analysis wi h P he o0.0001. The l gc o he SNP educa ion in e ac ion e m in he indi idual s udies anged om 1.01 o 1.08, indica ing li le e idence o es s a is ic infla ion on SNPeduca ion e ec o each s udy. Anno a ion o gene ic a ian s and gene exp ession in humans.The coo dina es and a ian iden ifie s a e epo ed on he NCBI B37 (hg19) genome build and anno a ed using UCSC Genome B owse 64. We iden ified a ian s wi hin each o he LD blocks ( 2Z0.8) in Eu opean and Asian popula ions o he 1000 Genomes P ojec (100 kb flanking he op SNP a each locus), o apply unc ional anno a ions o ansc ip ion egula ion using HaploReg65 and Encyclopedia o DNA Elemen s66 da a. We also gene a ed unca ional associa ion and co- exp ession ne wo k using GeneMANIA67, o de e mine whe he he disease- ela ed genes iden ified in his s udy and p e ious GWAS7,8 a e unc ionally connec ed. To assess gene exp ession in human issues, we examined he Ocula Tissue Da abase and he EyeSAGE da abase31,68. The es ima ed gene and exome-le el abundances a e a ailable online. No maliza ion o gene exp ession used he P obe Loga i hmic In ensi y E o me hod wi h genomic con ol-backg ound co ec ion31. Rela ionships be ween geno ype and cis egula ion o gene exp ession le els (Supplemen a y Table 9) we e assessed using exp ession quan i a i e ai locus associa ions in mul iple human issues om UK samples34, as well as gene exp ession p ofiles ob ained om GTExPo al da abase69. Re e ences 1. Mo gan, I. G., Ohno-Ma sui, K. & Saw, S. M. Myopia. Lance 379, 1739–1748 (2012). 2. Saw, S. M., Gazza d, G., Shih-Yen, E. C. & Chua, W. H. Myopia and associa ed pa hological complica ions. Oph halmic Physiol. Op . 25, 381–391 (2005). 3. Wojciechowski, R. Na u e and nu u e: he complex gene ics o myopia and e ac i e e o . Clin. Gene . 79, 301–320 (2011). ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 8NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 4. Sanfilippo, P. G., Hewi , A. W., Hammond, C. J. & Mackey, D. A. The he i abili y o ocula ai s. Su . Oph halmol. 55, 561–583 (2010). 5. Wong, T. Y. e al. P e alence and isk ac o s o e ac i e e o s in adul Chinese in Singapo e. In es . Oph halmol. Vis. Sci. 41, 2486–2494 (2000). 6. Kempen, J. H. e al. The p e alence o e ac i e e o s among adul s in he Uni ed S a es, Wes e n Eu ope, and Aus alia. A ch. Oph halmol. 122, 495–505 (2004). 7. Kie e , A. K. e al. Genome-wide analysis poin s o oles o ex acellula ma ix emodeling, he isual cycle, and neu onal de elopmen in myopia. PLoS Gene . 9, e1003299 (2013). 8. Ve hoe en, V. J. e al. Genome-wide me a-analyses o mul iances y coho s iden i y mul iple new suscep ibili y loci o e ac i e e o and myopia. Na . Gene . 45, 314–318 (2013). 9. Ve hoe en, V. J. e al. Educa ion influences he ole o gene ics in myopia. Eu . J. Epidemiol. 28, 973–980 (2013). 10. Fan, Q. e al. Educa ion influences he associa ion be ween gene ic a ian s and e ac i e e o : a me a-analysis o fi e Singapo e s udies. Hum. Mol. Gene . 23, 546–554 (2014). 11. Goss, D. A. Nea wo k and myopia. Lance 356, 1456–1457 (2000). 12. Gwiazda, J., Tho n, F. & Held, R. Accommoda ion, accommoda i e con e gence, and esponse AC/A a ios be o e and a he onse o myopia in child en. Op om. Vis. Sci. 82, 273–278 (2005). 13. S auss, O. The e inal pigmen epi helium in isual unc ion. Physiol. Re . 85, 845–881 (2005). 14. Ryme , J. & Wildsoe , C. F. The ole o he e inal pigmen epi helium in eye g ow h egula ion and myopia: a e iew. Vis. Neu osci. 22, 251–261 (2005). 15. Co iall, C. L., B ew, J., Vessey, K. A. & McB ien, N. A. Diisop opylfluo ophospha e al e s e inal neu o ansmi e le els and educes expe imen ally-induced myopia. Naunyn Schmiedebe gs A ch. Pha macol. 364, 372–382 (2001). 16. Thomas, D. Gene--en i onmen -wide associa ion s udies: eme ging app oaches. Na . Re . Gene . 11, 259–272 (2010). 17. Ascha d, H., Hancock, D. B., London, S. J. & K a , P. Genome-wide me a- analysis o join es s o gene ic and gene-en i onmen in e ac ion e ec s. Hum. He ed. 70, 292–300 (2010). 18. Manning, A. K. e al. A genome-wide app oach accoun ing o body mass index iden ifies gene ic a ian s influencing as ing glycemic ai s and insulin esis ance. Na . Gene . 44, 659–669 (2012). 19. Hancock, D. B. e al. Genome-wide join me a-analysis o SNP and SNP-by- smoking in e ac ion iden ifies no el loci o pulmona y unc ion. PLoS Gene . 8, e1003098 (2012). 20. Pa ssinen, O. e al. He i abili y o sphe ical equi alen : a popula ion-based win s udy among 63- o 76-yea -old emale wins. Oph halmology 117, 1908–1911 (2010). 21. Biino, G. e al. Ocula e ac ion: he i abili y and genome-wide sea ch o eye mo phome y ai s in an isola ed Sa dinian popula ion. Hum. Gene . 116, 152–159 (2005). 22. Age-Rela ed Eye Disease S udy Resea ch, G. A andomized, placebo-con olled, clinical ial o high-dose supplemen a ion wi h i amins C and E, be a ca o ene, and zinc o age- ela ed macula degene a ion and ision loss: AREDS epo no. 8. A ch. Oph halmol. 119, 1417–1436 (2001). 23. Foong, A. W. e al. Ra ionale and me hodology o a popula ion-based s udy o eye diseases in Malay people: The Singapo e Malay eye s udy (SiMES). Oph halmic Epidemiol. 14, 25–35 (2007). 24. Naka a, I. e al. P e alence and cha ac e is ics o age- ela ed macula degene a ion in he Japanese popula ion: he Nagahama s udy. Am. J. Oph halmol. 156, 1002–1009 e2 (2013). 25. Rie eld, C. A. e al. GWAS o 126,559 indi iduals iden ifies gene ic a ian s associa ed wi h educa ional a ainmen . Science 340, 1467–1471 (2013). 26. Lee, Y. Y., Lo, C. T., Sheu, S. J. & Lin, J. L. Wha ac o s a e associa ed wi h myopia in young adul s? A su ey s udy in Taiwan Mili a y Consc ip s. In es . Oph halmol. Vis. Sci. 54, 1026–1033 (2013). 27. Wong, L., Coggon, D., C uddas, M. & Hwang, C. H. Educa ion, eading, and amilial endency as isk ac o s o myopia in Hong Kong fishe men. J. Epidemiol. Communi y Heal h 47, 50–53 (1993). 28. Saw, S. M., Hong, C. Y., Chia, K. S., S one, R. A. & Tan, D. Nea wo k and myopia in young child en. Lance 357, 390 (2001). 29. Zheng, Y., Ding, X., Chen, Y. & He, M. The Guangzhou Twin P ojec : an upda e. Twin Res. Hum. Gene . 16, 73–78 (2013). 30. F ase , A. e al. Coho P ofile: he A on Longi udinal S udy o Pa en s and Child en: ALSPAC mo he s coho . In . J. Epidemiol. 42, 97–110 (2013). 31. Wagne , A. H. e al. Exon-le el exp ession p ofiling o ocula issues. Exp. Eye Res. 111, 105–111 (2013). 32. Young, T. L., Me lapally, R. & Shay, A. E. Complex ai gene ics o e ac i e e o . A ch. Oph halmol. 125, 38–48 (2007). 33. Chen, Y. P. e al. Selec i e b eeding o suscep ibili y o myopia e eals a gene-en i onmen in e ac ion. In es . Oph halmol. Vis. Sci. 52, 4003–4011 (2011). 34. G undbe g, E. e al. Mapping cis- and ans- egula o y e ec s ac oss mul iple issues in wins. Na . Gene . 44, 1084–1089 (2012). 35. S one, R. A. e al. GABA, expe imen al myopia, and ocula g ow h in chick. In es . Oph halmol. Vis. Sci. 44, 3933–3946 (2003). 36. Schmid, K. L., S asbe g, G., Rayne , C. L. & Ha field, P. J. The e ec s and in e ac ions o GABAe gic and dopamine gic agen s in he p e en ion o o m dep i a ion myopia by b ie pe iods o no mal ision. Exp. Eye Res. 110, 88–95 (2013). 37. Ba a hi, V. A. e al. In ol emen o GABA anspo e s in a opine- ea ed myopic e ina as e ealed by iTRAQ quan i a i e p o eomics. J. P o eome Res. (2014). 38. Alb ech , J., Sido yk-Weg zynowicz, M., Zielinska, M. & Aschne , M. Roles o glu amine in neu o ansmission. Neu on Glia. Biol. 6, 263–276 (2010). 39. Wang, K., Yamamo o, H., Chin, J. R., We b, Z. & Vu, T. H. Epide mal g ow h ac o ecep o -deficien mice ha e delayed p ima y endochond al ossifica ion because o de ec i e os eoclas ec ui men . J. Biol. Chem. 279, 53848–53856 (2004). 40. Tsai, W., Mo ielli, A. D. & Pe al a, E. G. The m1 musca inic ace ylcholine ecep o ansac i a es he EGF ecep o o modula e ion channel ac i i y. EMBO J. 16, 4597–4605 (1997). 41. Duncan, G. & Collison, D. J. Role o he non-neu onal choline gic sys em in he eye: a e iew. Li e Sci. 72, 2013–2019 (2003). 42. Wo a, D. R., Wa enbe g, E. V., Langason, R. B. & el-Fakahany, E. E. M1, M3 and M5 musca inic ecep o s s imula e mi ogen-ac i a ed p o ein kinase. Pha macology 56, 175–186 (1998). 43. Huo, L. e al. All- ans e inoic acid modula es mi ogen-ac i a ed p o ein kinase pa hway ac i a ion in human scle al fib oblas s h ough e inoic acid ecep o be a. Mol. Vis. 19, 1795–1803 (2013). 44. A umugam, B. & McB ien, N. A. Musca inic an agonis con ol o myopia: e idence o M4 and M1 ecep o -based pa hways in he inhibi ion o expe imen ally-induced axial myopia in he ee sh ew. In es . Oph halmol. Vis. Sci. 53, 5827–5837 (2012). 45. Die z, H. C. & Pye i z, R. E. Mu a ions in he human gene o fib illin-1 (FBN1) in he Ma an synd ome and ela ed diso de s. Hum. Mol. Gene . 1799–1809 (1995). 46. Fa b o he , J. E. e al. Linkage analysis o he gene ic loci o high myopia on 18p, 12q, and 17q in 51 U.K. amilies. In es . Oph halmol. Vis. Sci. 45, 2879–2885 (2004). 47. Yip, S. P. e al. A DNA pooling-based case-con ol s udy o myopia candida e genes COL11A1, COL18A1, FBN1, and PLOD1 in a Chinese popula ion. Mol. Vis. 17, 810–821 (2011). 48. F ench, A. N., Mo gan, I. G., Mi chell, P. & Rose, K. A. Risk ac o s o inciden myopia in Aus alian schoolchild en: he Sydney adolescen ascula and eye s udy. Oph halmology 120, 2100–2108 (2013). 49. Guggenheim, J. A. e al. Time ou doo s and physical ac i i y as p edic o s o inciden myopia in childhood: a p ospec i e coho s udy. In es . Oph halmol. Vis. Sci. 53, 2856–2865 (2012). 50. Saw, S. M. e al. Nea wo k in ea ly-onse myopia. In es . Oph halmol. Vis. Sci. 43, 332–339 (2002). 51. Guo, Y. e al. Myopic shi and ou doo ac i i y among p ima y school child en: one-yea ollow-up s udy in Beijing. PLoS ONE 8, e75260 (2013). 52. Mu i, D. O., Mi chell, G. L., Moeschbe ge , M. L., Jones, L. A. & Zadnik, K. Pa en al myopia, nea wo k, school achie emen , and child en’s e ac i e e o . In es . Oph halmol. Vis. Sci. 43, 3633–3640 (2002). 53. Hawo h, C. M., Da is, O. S. & Plomin, R. Twins Ea ly De elopmen S udy (TEDS): a gene ically sensi i e in es iga ion o cogni i e and beha io al de elopmen om childhood o young adul hood. Twin Res. Hum. Gene . 16, 117–125 (2013). 54. Lin, Z. e al. Nea wo k, ou doo ac i i y, and hei associa ion wi h e ac i e e o . Op om. Vis. Sci. 91, 376–382 (2014). 55. Lu, B. e al. Associa ions be ween nea wo k, ou doo ac i i y, and myopia among adolescen s uden s in u al China: he Xichang Pedia ic Re ac i e E o S udy epo no. 2. A ch. Oph halmol. 127, 769–775 (2009). 56. F ench, A. N., Ashby, R. S., Mo gan, I. G. & Rose, K. A. Time ou doo s and he p e en ion o myopia. Exp. Eye Res. 114, 58–68 (2013). 57. Howie, B., Fuchsbe ge , C., S ephens, M., Ma chini, J. & Abecasis, G. R. Fas and accu a e geno ype impu a ion in genome-wide associa ion s udies h ough p e-phasing. Na . Gene . 44, 955–959 (2012). 58. Howie, B. N., Donnelly, P. & Ma chini, J. A flexible and accu a e geno ype impu a ion me hod o he nex gene a ion o genome-wide associa ion s udies. PLoS Gene . 5, e1000529 (2009). 59. Aulchenko, Y. S., Ripke, S., Isaacs, A. & an Duijn, C. M. GenABEL: an R lib a y o genome-wide associa ion analysis. Bioin o ma ics 23, 1294–1296 (2007). 60. Voo man, A., Lumley, T., McKnigh , B. & Rice, K. Beha io o QQ-plo s and genomic con ol in s udies o gene-en i onmen in e ac ion. PLoS ONE 6, e19416 (2011). NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11008 ARTICLE NATURE COMMUNICATIONS | 7:11008 | DOI: 10.1038/ncomms11008 | www.na u e.com/na u ecommunica ions 9