The CpG-sites of the CBX3 ubiquitous chromatin opening element are critical structural determinants for the anti-silencing function.

1 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 www.na u e.com/scien i ic epo s The CpG-si es o he CBX3 ubiqui ous ch oma in opening elemen a e c i ical s uc u al de e minan s o he an i-silencing unc ion Jessica Kunkiel1,2, Na ascha Gödecke3, Mania Acke mann2,4, Di k Ho mann2, Axel Schambach2,5, Nico Lachmann2,4, Dagma Wi h2,3 & Thomas Mo i z1,2 Supp ession o he apeu ic ansgene exp ession om e o i al gene he apy ec o s by epigene ic de ence mechanisms ep esen s a p oblem ha is pa icula ly encoun e ed in plu ipo en s em cells (PSCs) and hei di e en ia ed p ogeny. T ansgene exp ession in hese cells, howe e , can be s abilised by CpG- ich ubiqui ous ch oma in opening elemen s (UCOEs). In his con ex we ecen ly demons a ed p o ound an i-silencing p ope ies o he small (679 bp) CBX3-UCO elemen and we now con i med his obse a ion in he con ex o he de ined mu ine ch omosomal loci ROSA26 and TIGRE. Mo eo e , since he s uc u al basis o he an i-silencing ac i i y o UCOEs has emained poo ly de ined, we in e oga ed a ious CBX3 sub agmen s in he con ex o len i i al ec o s and mu ine PSCs. We demons a ed ma ked hough dis inc an i-silencing ac i i y in he plu ipo en s a e and du ing PSC- di e en ia ion o se e al o he CBX3 sub agmen s. This ac i i y was signi ican ly co ela ed wi h CpG con en as well as endogenous ansc ip ional ac i i y. In e es ingly, also a sc ambled CBX3 e sion wi h p ese ed CpG-si es e ained he an i-silencing ac i i y despi e he lack o endogenous p omo e ac i i y. Ou da a he e o e highligh he impo ance o CpG-si es and ansc ip ional ac i i y o UCOE unc ionali y and sugges con ibu ions om di e en mechanisms o he o e all an i-silencing unc ion o he CBX3 elemen . Al hough he design o len i i al ec o s has imp o ed subs an ially o e he las decade, e en mode n SIN-len i i al ec o s emain subjec o posi ional e ec s ha can lead o silenced o a iega ed ansgene exp es- sion1, 2. In pa icula plu ipo en s em cells a e p oblema ic in his espec , as hese cells ha bou s ong in insic de ence mechanisms agains o eign and especially i al DNA3–7. Mo eo e , he di e en ia ion o plu ipo en s em cells owa ds mo e ma u e cell ypes in ol es genome wide epigene ic emodelling wi h a po en ially de i- men al impac on ansgene exp ession. To e ade his ansc ip ional ep ession so called ubiqui ous ch oma in opening elemen s (UCOEs) ha e been applied8. In p e ious s udies, we and o he s ha e demons a ed he abili y o UCOEs o es ablish s able exp ession o len i i al ansgenes d i en by bo h i al and physiological p omo e s in mu ine and human plu ipo en as well as mul ipo en a ge cells and he eo di e en ia ed p ogeny9–13, hus demons a ing b oad applicabili y o UCOEs. The mos commonly used UCOE is de i ed om he human HNRPA2B1-CBX3 housekeeping gene locus. This elemen con ains wo di e gen ly ansc ibed p omo e s, he HNRPA2B1 p omo e and he CBX3 p omo e which a e bo h spanned by a la ge s e ch o non-me hyla ed CpG-islands (CGI)11, 13. A cen al 1.5 kb sequence 1Rep og amming and Gene The apy G oup, REBIRTH Clus e o Excellence, Hanno e Medical School, 30625, Hanno e , Ge many. 2Ins i u e o Expe imen al Hema ology, Hanno e Medical School, 30625, Hanno e , Ge many. 3Model Sys ems o In ec ion and Immuni y G oup, Helmhol z Cen e o In ec ion Resea ch, 38124, B aunschweig, Ge many. 4T ansla ional Hema ology o Congeni al Diseases Junio -G oup, REBIRTH Clus e o Excellence, Hanno e Medical School, 30625, Hanno e , Ge many. 5Di ision o Hema ology/Oncology, Bos on Child en’s Hospi al, Bos on, MA, USA. Co espondence and eques s o ma e ials should be add essed o T.M. (email: mo i z. homas@mh- hanno e .de) Recei ed: 10 Feb ua y 2017 Accep ed: 10 May 2017 Published: xx xx xxxx OPEN www.na u e.com/scien i ic epo s/ 2 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 con aining he an i-silencing ac i i y (A2UCOE) has been iden i ied, which is associa ed wi h a pe missi e ch o- ma in s uc u e ma ked by hype ace yla ion o his ones H3 and H4, ime hyla ion o lysine 4 on his one H3 (H3K4me3), and he lack o H114. We ha e ecen ly shown ha a much smalle sequence comp ising only he 0.7 kb CBX3 moie y o he A2UCOE ( e e ed o as CBX3) can e ec i ely p o ec SIN-len i i al ansgene exp es- sion in mu ine and human plu ipo en cells as well as hei hema opoie ic de i a i es15. The CBX3 p e en ed CpG me hyla ion o a spleen ocus o ming i us (SFFV) p omo e and educed he p esence o ep essi e ch oma in ma ks (e.g. H3K9me3 and H3K27me3) while in oducing ac i e ma ks such as H3K4me3 and PhosPol2. While mechanis ically he an i-silencing unc ion o he UCOEs has been a ibu ed o he p esence o non-me hyla ed CGIs, a ea u e sha ed by all known UCOEs, up o now he an i-silencing p ope ies o he UCOE ha e no been linked o speci ic s uc u al ea u es. To po en ially iden i y such ea u es wi hin he CBX3 s uc u e we ha e gene a ed a numbe o CBX3-de i ed sub agmen s and analysed he unc ionali y o hese sub agmen s in he con ex o len i i al ec o s and a b oad ange o in eg a ion si es wi hin mu ine ESCs and iPSCs as well as hei di e en ia ed p ogeny. In addi ion, we ha e in es iga ed he ele ance o he CpG-islands o he an i-silencing p ope ies by employing wo syn he ic sc ambled-CBX3 e sions wi h ei he main ained o des oyed CpG-si es. Mo eo e , o explo e he ele ance o ansc ip ional ac i i y o he an i-silencing unc ion we examined he p omo e ac i i y o ou CBX3 sub agmen s. So a all in es iga ions assessing UCOE unc ion ha e been pe o med in bulk cell popula ions wi h a ying in eg a ion si es while a di ec compa ison o exp es- sion casse es wi h and wi hou UCOE in a de ined in eg a ion si e and ch omosomal a chi ec u e has no ye been pe o med. Acco dingly, i is no known i UCOEs can p o ec om silencing i espec i e o he na u e o he ch omosomal in eg a ion si e. The e o e, we now ha e examined he an i-silencing abili y o he CBX3 a wo de ined ch omosomal loci, he ROSA2616 and he TIGRE locus T117, in single-cell de i ed mESC clones u ilising ecombinase media ed casse e exchange (RMCE). Resul s CBX3 sub agmen s and hei ac i i y in plu ipo en s em cells. To iden i y he s uc u al a ib- u es esponsible o he an i-silencing unc ion o he CBX3, we gene a ed a o al o six pa ially o e lapping sub agmen s o he CBX3 elemen (Table1, Fig.1A). To gene a e cell popula ions ep esen ing a b oad ange o di e en in eg a ion si es, we employed len i i al ansduc ion. To his end, he sub agmen s we e in oduced in o a hi d-gene a ion sel -inac i a ing (SIN)-len i i al ec o ups eam o he i al SFFV p omo e d i ing he exp ession o he eGFP epo e . Vec o s employing he ull leng h CBX3 (C-SG) and he 1.5 kb A2UCOE (A2- SG) se ed as posi i e con ols, whe eas he SFFV p omo e alone (SG) was used as a nega i e con ol (Fig.1B). A mu ine ESC as well as an induced plu ipo en s em cell (miPSC) line we e ansduced wi h an MOI o 10 and 30, espec i ely. By his p ocedu e, simila ec o copy numbe s (VCN) o app oxima ely 2–3 o mESCs and 1–2 o miPSCs we e achie ed o all ec o s (Supplemen a y TableS1). Exp ession o eGFP in plu ipo en cells was moni o ed o e a pe iod o 31 days, i.e. 9 passages. T ansduc ion e iciencies we e assessed by low cy ome y on day 3 (Supplemen a y Fig.S1A) and se ed as a e e ence poin o de e mine he ela i e numbe o eGFP-exp essing cells o e ime. Fo all ec o s and in bo h he mESC as well as he miPSC line, loss o ansgene exp ession was p ima ily obse ed du ing he i s 7 days (passage 2). This d op in ansgene exp ession was mo e p ominen o SG- han C-SG- o A2-SG- ansduced con ol cells, indica ing he ansgene s abilizing unc ion o he UCOEs. Cells ansduced wi h he ec o s C(1-508)-SG, C(85-508)-SG, C(170-508)-SG, C(340-508)-SG, o C(503-679)-SG main ained ansgene exp ession a le els compa able o he posi i e con ols anging om 71% o 80% in mESCs and 61% o 68% in miPSCs (Fig.2A,B). Only he C(1-339) agmen ailed o display an i-silencing p ope ies. Following his d op, he ansgene exp ession s abilized and s able eGFP-exp ession was obse ed o a leas 9 passages i.e. un il day 31 (Figs2C,D, S1B,C). Taken oge he , hese esul s indica e pa icula ly he CpG- ich cen al and 3′ egion o he CBX3 egion o be in ol ed in UCOE-media ed ansgene s abiliza ion in plu ipo en cells. Biologic ac i i y o he a ious agmen s was also e lec ed by he median luo escence in ensi y (MFI) o eGFP exp ession in ansduced cell popula ions (Table2), indica ing he con ibu ion o he CBX3-de i ed ele- men s o he o e all p omo e ac i i y. While he lowes MFI was obse ed o SG- ansduced cells, a leas in mESCs, C(1-339)-SG ansduced cells displayed simila exp ession in ensi ies as C(1-508)-SG, C(85-508)-SG, and C(170-508)-SG ansduced cells. Lowe exp ession in ensi ies we e obse ed pa icula ly o C(340-508)-SG ansduced cells in bo h mESCs and miPSCs. UCO elemen s Leng h (bp) CpG-si es (#) G+C (%) CpG densi y (CpG/100 bp) A2UCOE 1550 142 64.0 9.2 CBX3 679 63 67.9 9.3 CBX3(1-339) 339 18 57.6 5.3 CBX3(1-508) 508 35 62.2 6.9 CBX3(85-508) 424 31 63.9 7.3 CBX3(170-508) 339 26 66.7 7.7 CBX3(340-508) 168 17 71.4 10.1 CBX3(503-679) 177 28 83.1 15.8 Table 1. S uc u al a ibu es o he A2UCOE, CBX3 and CBX3 sub agmen s. bp, base pai s; #, numbe o www.na u e.com/scien i ic epo s/ 3 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 An i-silencing unc ion o CBX3 sub agmen s in di e en ia ed mESCs/iPSCs: co ela ion wi h CpG-si es. Du ing di e en ia ion s em cells unde go ex ensi e epigene ic emodelling and euch oma ic egions ha se ed as a ou able in eg a ion si es o len i i al ec o s may become he e och oma ic, he eby posing a hind ance o len i i al ansgene exp ession1, 2. The in i o di e en ia ion o mESCs and miPSCs he e- o e ep esen s a much mo e s ingen sys em o analyse he po ency o an i-silencing elemen s such as ou CBX3 sub agmen s. To his end, he cells we e subjec ed o undi ec ed di e en ia ion and he exp ession o eGFP in di e en ia ed, SSEA-1 nega i e cells was assessed on day 8 (Fig.3A,B). Whe eas SG- ansduced con ol mESCs (7%) as well as miPSCs (4%) showed a nea ly comple e loss o ansgene exp ession upon di e en ia ion, inco - po a ion o CBX3 o A2UCOE in o he ec o ma kedly inc eased he popula ion o ansgene posi i e cells (CBX3: 63% (mESC), 81% (miPSC); A2UCOE: 64% (mESC), 72% (miPSC)). Also, all cons uc s wi h CBX3 sub- agmen s s abilized ansgene exp ession o some ex en in di e en ia ed mESC- and miPSC-de i ed cells when compa ed o SG con ols, al hough hese di e ences we e signi ican only o CBX3(1-508) and CBX3(85-508). Wi h eGFP-exp essing cells anging om 34% o 41%, howe e , nei he o hese sub agmen s showed ull CBX3 ac i i y. On he o he hand, CBX3(1-339) and CBX3(340-508) clea ly showed he lowes an i-silencing ac i i y in bo h cell ypes. No majo di e ences we e obse ed be ween C(1-508)-SG, C(85-508)-SG, C(170-508)-SG and o some deg ee also C(503-679)-SG ansduced cells. These da a indica e ha in addi ion o a cen al egion common o he h ee sub agmen s wi h highe ac i i y, also he 3′-end egion o he CBX3 (bp 503 o bp 679) is in ol ed in an i-silencing. In e es ingly, hese ou CBX3 sub agmen s all con ain a high numbe o CpG-si es and he amoun o CpG-si es pe agmen was ound o posi i ely co ela e wi h i s an i-silencing capaci y (measu ed as he pe - cen age o eGFP-posi i e cells ollowing di e en ia ion) based on da a o mESCs (R2 = 0.6811, p < 0.0001) as well as miPSCs (R2 = 0.6248, p < 0.0001) (Fig.3C,D). While a simila co ela ion was obse ed o UCOE leng h (R2 = 0.6299, p < 0.0001 in mESCs; R2 = 0.5848, p < 0.0001 in miPSCs), no signi ican co ela ion was de ec ed o CpG-si e densi y (CpGs/100 bp) o GC-con en (da a no shown). Func ionali y o a sc ambled CBX3-UCOE in plu ipo en and di e en ia ed mESCs/miP- SCs. To u he assess he ele ance o he CpG-si es wi hin he CBX3 elemen , wo sc ambled e sions o he CBX3 we e designed. The i s elemen was gene a ed by andom shu ling o he DNA sequence while e aining he CpG-si es ( oughly 20% o o al sequence) a hei o iginal posi ion wi hou changing he o e all nucleo ide composi ion (Csc CpG). The second elemen was designed by andomly shu ling he comple e DNA sequence he eby des oying he CpG-si es (Csc ; alignmen o sequences gi en in Supplemen a y Fig.S2). Again, he elemen s we e inse ed in o he SG len i i al ec o hus gi ing ise o he cons uc s Csc CpG-SG and Csc -SG (Fig.4A). Figu e 1. CBX3 sub agmen s and len i i al ec o s. (A) Schema ic ep esen a ion o he CBX3 elemen wi h i s CpG-si es (black ba s) and dele ed splice si es ( ed ba s) as well as he CBX3 sub agmen s ob ained by PCR (CBX3(1-339) - CBX3(340-508)) and he CBX3(503-679) sub agmen gene a ed by es ic ion diges ion wi h XhoI and ApaI. (B) Thi d-gene a ion sel -inac i a ing (SIN) len i i al cons uc s exp essing an eGFP cDNA om he spleen ocus o ming i us (SFFV) p omo e in he absence o p esence o ei he he 1.5 kb A2UCOE, he 679 bp CBX3 o any o he CBX3 sub agmen s. Abb e ia ions: ΔLTR: Long e minal epea ha bo ing he SIN mu a ion in he U3 egion o he LTR; ψ: ex ended encapsida ion signal; RRE: Re - esponse elemen ; cPPT: cen al polypu ine ac ; GFP: (enhanced) g een luo escen p o ein; wPRE: woodchuck hepa i is i us pos - ansc ip ional elemen . www.na u e.com/scien i ic epo s/ 4 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 Again, mESCs and miPSCs we e ansduced while aiming o compa able ansduc ion e iciencies and VCNs (Supplemen a y TableS3). The eGFP exp ession in plu ipo en cells was moni o ed by low cy ome y o a pe iod o 24 days (Fig.4B,C). In ag eemen wi h ou p e ious da a, loss o eGFP exp ession was p ima ily obse ed up o day 7 (passage 2) wi h ela i ely cons an numbe s o eGFP-posi i e cells he ea e . A e 24 Figu e 2. Ac i i y o CBX3 sub agmen s in plu ipo en s em cells. Mu ine ESCs and iPSCs we e ansduced wi h len i i al ec o s wi h o wi hou he A2UCOE, he CBX3 o he CBX3 sub agmen s. The eGFP exp ession was moni o ed o 31 days and he ela i e exp ession (d3 = 100%) is shown o day 7 in (A) mESCs and (B) miPSCs. The ime-cou se o ela i e eGFP exp ession o 31 days is depic ed o (C) mESCs and (D) miPSCs. Ac i i y o he sub agmen s was compa ed o he SG nega i e con ol. Da a ep esen independen expe imen s n = 3; mean ± SD; *p < 0.05, **p < 0.01, ****p < 0.0001 as de e mined by one-way ANOVA. Median luo escence in ensi y (MFI) Vec o mESC miPSC d3 d31 d3 d31 SG 54 ± 6 28 ± 4 26 ± 9 33 ± 1 A2-SG 201 ± 51*** 142 ± 21**** - - C-SG 223 ± 25**** 121 ± 17**** 122 ± 61** 106 ± 38* C(1-339)-SG 119 ± 27 104 ± 22*** 37 ± 17 57 ± 6 C(1-508)-SG 148 ± 10*128 ± 13**** 75 ± 22 89 ± 12 C(85-508)-SG 142 ± 37*135 ± 23**** 89 ± 24 97 ± 39* C(170-508)-SG 125 ± 37 113 ± 13*** 75 ± 14 97 ± 16* C(340-508)-SG 53 ± 17 43 ± 11 48 ± 11 48 ± 10 C(503-679)-SG 87 ± 4 56 ± 7 67 ± 6 81 ± 22 Table 2. Exp ession in ensi y o CBX3 sub agmen s in plu ipo en s em cells. eGFP luo escence media ed by he UCO elemen s was compa ed o he SG nega i e con ol. Da a ep esen independen expe imen s n = 3; mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as de e mined by one-way ANOVA. www.na u e.com/scien i ic epo s/ 5 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 days, only he Csc CpG-SG ansduced mESCs and miPSCs displayed signi ican ly mo e eGFP-posi i e cells (72% and 44%) han SG- ansduced nega i e con ols (43% and 26%). In con as , no signi ican di e ences o he nega i e con ol we e iden i ied in bo h Csc -SG ansduced mESCs (51%) and miPSCs (29%) a e 24 days (Supplemen a y Fig.S3A,B). O no e, only C-SG ansduced cells demons a ed an inc ease in MFI abo e SG le els (Supplemen a y TableS3), whe eas nei he he inco po a ion o Csc CpG no Csc could achie e his. To assess he an i-silencing ac i i y o he sc ambled CBX3 elemen s du ing di e en ia ion, we de e mined he ela i e eGFP exp ession in SSEA-1 nega i e cells ollowing eigh days o undi ec ed di e en ia ion using day 0 as e e ence (Fig.4D–F). Con i ming ou p e ious da a, SG ansduced mESCs and miPSCs los mos o hei exp ession upon di e en ia ion (down o 8% and 3%, espec i ely), whe eas C-SG ansduced mESCs and miP- SCs demons a ed signi ican ly highe le els o eGFP exp essing cells (48% and 68%, espec i ely). In e es ingly, also Csc CpG-SG cells showed signi ican ly ele a ed le els o ansgene exp ession upon di e en ia ion o bo h mESCs (33%) and miPSCs (47%). Csc -SG ansduced cells, on he o he hand, ailed o show signi ican an i-silencing ac i i y in his assay wi h ansgene exp ession in mESC- o miPSC-de i ed cells o only 17% o 4%, espec i ely. These esul s u he highligh ed he impo ance o he CpG-si es o he an i-silencing unc ion o he CBX3. P omo e ac i i y o CBX3-UCOE agmen s in plu ipo en and di e en ia ed mESCs/miP- SCs. The da a gi en abo e on MFI o eGFP exp ession sugges ha bo h he A2UCOE and he CBX3 s ill ha bou in insic ansc ip ional ac i i y. The dual p omo e ac i i y o he A2UCOE, howe e , seems o be dis- pensable o i s an i-silencing ac i i y, since he CBX3 wi h i s single p omo e is s ill highly e icien in his espec . On he o he hand, ano he 0.7 kb UCOE de i ed om he i s in on o CBX3 bu lacking in insic ansc ip ional ac i i y was shown o media e only pa ial p o ec ion om silencing o he SFFV p omo e in i o18. Taken oge he wi h he obse a ion ha he A2UCOE showed an o ien a ion-dependen an i-silencing Figu e 3. Ac i i y o he CBX3 sub agmen s in di e en ia ed cells. The ansduced cells we e subjec ed o an undi ec ed EB-based di e en ia ion and he eGFP exp ession was de e mined on day 0 as well as in SSEA-1 nega i e cells on day 8. The ela i e exp ession o eGFP was de e mined (d0 = 100%) o (A) mESCs and (B) miPSCs. (C,D) The co ela ion o ela i e eGFP exp ession in di e en ia ed cells and he numbe o CpG-si es con ained wi hin each elemen is shown o (C) mESCs and (D) miPSCs (p < 0.0001). Ac i i y o he sub agmen s was compa ed o he SG nega i e con ol. Da a ep esen independen expe imen s n = 3; mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as de e mined by one-way ANOVA. www.na u e.com/scien i ic epo s/ 6 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 e ec wi h imp o ed unc ion only when he CBX3 p omo e was acing he jux aposed he e ologous p omo e 11, his poses he ques ion whe he in insic p omo e ac i i y in he di ec ion o he ansgene has a ole in he an i-silencing abili y o he UCOEs. Since he CBX3 is a CpG-island p omo e and hus does no con ain known de ined mo i s, such as a TATA box, ha could be mu a ed o e ase he p omo e unc ion, we in es iga ed he p omo e ac i i y o he CBX3 sub agmen s ins ead and compa ed his o hei an i-silencing abili y. Fo hese s udies, he SFFV p omo e was excised om he SG len i i al ec o s, gi ing ise o he ec o cons uc s depic ed in Fig.5A. Th ee days ollowing he ansduc ion o mESCs and miPSCs, he eGFP exp ession was assessed by low cy ome y (Fig.5B). Since o some cons uc s i was di icul o de e mine he i e o he len i i al supe na an s due o low o no eGFP exp ession, some a iabili y o VCN could no be a oided (Supplemen a y TableS4). As expec ed, igh a e ansduc ion (day 3) he eGFP con ol cons uc (G) showed only a minimal backg ound MFI o mESCs as well as miPSCs. Minimal o no p omo e ac i i y was also obse ed o he sub agmen s C(1- 339), C(340-508) as well as he syn he ic elemen Csc CpG. In con as , C(1-508), C(85-508), C(170-508), and Figu e 4. An i-silencing ac i i y o syn he ic sc ambled CBX3 elemen s. The DNA sequence o he CBX3 was shu led andomly o c ea e wo syn he ic elemen s ha ha bo he same nucleo ide composi ion in a andom o de wi h one sequence main aining he CpG-si es a hei o iginal posi ion (Csc CpG) and he o he comple ely des oying hese si es (Csc ). (A) The elemen s we e in oduced in o he SFFV.eGFP (SG) hi d gene a ion SIN-len i i al ec o ups eam o he p omo e . (B,C) Following ansduc ion, he eGFP exp ession was moni o ed by low cy ome y o 24 days in SSEA-1 posi i e (B) mESCs and (C) miPSCs. (D–F) The ansduced cells we e subjec ed o an EB-based undi ec ed di e en ia ion wi h ansgene exp ession measu ed on day 0 and in SSEA-1 nega i e cells on day 8. (D) Rep esen a i e low cy ome ic analysis o mESCs and miPSCs p io o (d0) and a e di e en ia ion (d8). Rela i e eGFP exp ession was de e mined in di e en ia ed (SSEA-1 nega i e) (E) mESCs and (F) miPSCs. Ac i i y o he syn he ic elemen s was compa ed o he SG nega i e con ol. Da a ep esen independen expe imen s n = 3; mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as de e mined by one-way ANOVA. www.na u e.com/scien i ic epo s/ 7 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 C(503-679) clea ly demons a ed endogenous p omo e ac i i y, which o some cons uc s was simila o he ac i i y o CBX3 ha was used as he posi i e con ol. To acili a e he ead-ou , o u he analysis he backg ound MFI o G- ansduced cells was deduc ed om all MFIs measu ed in he same expe imen al se ing yielding MFIco . Mo eo e , o accoun o he di e ences in VCN, he MFIco pe VCN (MFIco /VCN) was calcula ed and used o u he analysis. When MFIco /VCN was moni o ed o e a 28-day pe iod, ela i ely s able eGFP exp ession le els we e obse ed o he indi idual cons uc s (Supplemen a y Fig.S4A,B) and e en on day 28 signi ican p omo e ac i i y was de ec ed o he C(1-508) sub agmen in bo h mESCs and miPSCs as well as o he C(85-508) in mESCs and he C(503-679) in miPSCs. Al hough no signi ican , C(170-508) also showed p omo e ac i i y in bo h mESCs and miPSCs (Fig.5C,D). O no e, despi e i s an i-silencing ac i i y in he con ex o SG len i i al ec o s, he Csc CpG showed Figu e 5. P omo e ac i i y o CBX3 de i ed agmen s. (A) Thi d gene a ion SIN-len i i al ec o s ha bo ing an eGFP cDNA p eceded by CBX3-de i ed elemen s. (B) Rep esen a i e low cy ome ic analysis o eGFP exp ession in mESCs and miPSCs h ee days a e ansduc ion. (C-D) Median luo escence in ensi y (MFI) o eGFP-exp essing cells was measu ed by low cy ome y, co ec ed o backg ound eGFP exp ession (MFIco ; G = 0) and de e mined pe ec o copy numbe (VCN) in plu ipo en (SSEA-1 posi i e) (C) mESCs and (D) miPSCs 28 days a e ansduc ion. Ac i i y o he sub agmen s was compa ed o he eGFP nega i e con ol. Da a ep esen independen expe imen s n = 3; mean ± SD; *p < 0.05, ***p < 0.001, ****p < 0.0001 as de e mined by one-way ANOVA. www.na u e.com/scien i ic epo s/ 8 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 almos no eGFP exp ession in hese expe imen s, indica ing ha he media ed an i-silencing e ec was independ- en o ansc ip ional ac i i y. In con as , he o he agmen s ha p e iously demons a ed good an i-silencing abili y in he con ex o SG-based len i i al ec o s, i.e. C(1-508), C(85-508), C(170-508), and C(503-679), exhib- i ed no able p omo e ac i i y in mESCs as well as miPSCs. Acco ding o hese da a, o he CBX3 sub agmen s highe endogenous p omo e ac i i y ended o co ela e posi i ely wi h hei abili y o p o ec SG-based ec o s om silencing in mESCs (p < 0.0001) o miPSCs (p = 0.0007) (Supplemen a y Fig.S4C,D). Li le endogenous p omo e ac i i y was, howe e , obse ed o all cons uc s excep o he comple e CBX3 elemen ollowing eigh days o di ec ed di e en ia ion (Supplemen a y Fig.S4E,F). CBX3 unc ion a de ined ch omosomal loci. Len i i al ansduc ion o s em and p ogeni o cells is accompanied wi h genome-wide in eg a ion o ec o s. This was desc ibed o esul in he e ogeneous ansgene exp ession in indi idual cell clones o cell popula ions9–12, 19. I was shown ha ansduc ion wi h UCOE-con aining ec o s esul ed in signi ican ly less he e ogenei y han popula ions o cells wi h UCOE- ee ec o s9, 10, 15. Howe e , i was no clea whe he he UCOE-media ed p o ec ion om epigene ic modula ions was exhibi ed equally a all in eg a ion si es o only as a consequence o in eg a ion in o ch omosomal si es wi h a ou able epigene ic ea u es. To add ess his no ion, we used Flp ecombinase media ed casse e exchange (RMCE) o in es iga e he impac o a UCOE a wo de ined ch omosomal si es. In pa icula , we in eg a ed a casse e comp ising he silencing p one i al SFFV p omo e d i ing eGFP wi h o wi hou he CBX3 UCOE a wo de ined ch omosomal loci in mouse emb yonic s em cells (mESCs); he ROSA26 and he TIGRE locus (Fig.6A). The inco po a ion o he eGFP epo e allowed single cell analyses o isualise he he e ogeneous e ec o ansgene silencing. In he ROSA26 locus o plu ipo en mESCs he pe cen age o ansgene exp essing cells g adually dec eased o 60-80% o all isogenic SFFV.eGFP (SG) as well as all CBX3.SFFV.eGFP (C-SG) clones (Supplemen a y Fig.S5A), whils he exp ession o eGFP emained ai ly s able o e 6–10 consecu i e passages a he TIGRE locus o bo h SG and C-SG clones (Supplemen a y Fig.S5B). The compa able loss o eGFP exp es- sion a ROSA26- a ge ed isogenic subclones indica es simila epigene ic silencing kine ics upon RMCE, which sugges s he p esence o a highly ep oducible modi ica ion. To assess he CBX3 unc ion du ing di e en ia ion, he ansgene exp ession was de e mined o bo h ROSA26 and TIGRE a ge ed cells (Fig.6B,C) on day 0 and day 8 o an undi ec ed EB-based di e en ia ion. He e, day 0 se ed as e e ence poin o de e mine he ela i e numbe o eGFP-exp essing cells a e eigh days o di e en ia ion. The SG clones o ROSA26 a ge ed cells showed nea ly a comple e loss o ansgene exp es- sion wi h only 1–5% eGFP-posi i e cells emaining. Inco po a ion o he CBX3 inc eased eGFP-posi i e cells o 7–15% (Fig.6B). A he TIGRE locus, undi ec ed di e en ia ion esul ed in 31–42% eGFP-exp essing cells in SG Figu e 6. An i-silencing unc ion o he CBX3 a wo de ined ch omosomal loci. (A) Mu ine ESCs we e a ge ed a ei he he ROSA26 o he TIGRE ch omosomal locus by using FLP-media ed casse e exchange o in oduce ei he an SFFV.eGFP (SG) o CBX3.SFFV.eGFP (CSG) casse e. Subclones o (B) ROSA26- (SG: 3 clones; CSG: 4 clones) and (C) TIGRE- (SG: 3 clones; CSG: 3 clones) a ge ed cells we e subjec ed o EB- based undi ec ed di e en ia ion and he ela i e pe cen age o eGFP-exp essing cells was de e mined by low cy ome y on day 8 in SSEA-1 nega i e cells (d0 = 100%). (D) Rep esen a i e low cy ome ic analysis o a ge ed cells on day 0 and day 8. Ba cha s ep esen mean ± SD; n = 3 pe clone. S uden ’s - es was pe o med on pooled SG-clones (ROSA26, n = 9; TIGRE, n = 9) and C-SG clones (ROSA26, n = 12; TIGRE, n = 9); **p < 0.01, ****p < 0.0001. www.na u e.com/scien i ic epo s/ 9 SCIEnTI IC RepoRTS | 7: 7919 | DOI:10.1038/s41598-017-04212-8 clones, which was inc eased o 71–89% eGFP-posi i e cells in C-SG clones (Fig.6C). Rep esen a i e low cy o- me ic analyses a e shown in Fig.6d. Taken oge he , while di e en le els o epigene ic ep ession we e obse ed in hese wo de ined ch omosomal si es, he CBX3 was able o signi ican ly s abilise ansgene exp ession a bo h loci. This p o es ha UCOEs can o e come epigene ic es ic ions in de ined ch omosomal si es, he eby jus i y- ing he use o bulk popula ions o in es iga e UCOE unc ions. Discussion UCOEs such as he A2UCOE ha e been demons a ed o e icien ly nega e epigene ic ep ession o ansgene exp ession om e o i al ec o s in a a ie y o cell lines and p ima y cells including hema opoie ic s em cells as well as mu ine and human plu ipo en ESCs and iPSCs and hei di e en ia ed p ogeny9–11, 20. This p o ec- i e ac i i y has been linked wi h educed p omo e CpG me hyla ion as well as dec eased le els o ep essi e and inc eased le els o ac i e his one ma ks in he ch omosomal neighbou hood o UCOEs. While adi ionally bidi ec ional p omo e ac i i y has been deemed a necessa y ai o he an i-silencing unc ion o UCOEs, we ecen ly desc ibed he 0.7 kb CBX3 elemen ha lacks he HNRPA2B1 p omo e moie y o he A2UCOE bu s ill po en ly s abilizes len i i al ansgene exp ession in mul ipo en and plu ipo en s em cells15. Thus, he CBX3 ep esen s a unique single-p omo e UCOE ha s ill shows mos i no all o he an i-silencing ac i i y o he A2UCOE and due o i s small size i does no ad e sely a ec i al i es15. Al hough he CBX3 has been shown o p o ec ansgene exp ession om i al, housekeeping and issue-speci ic p omo e elemen s, a link o his ac i i y o dis inc s uc u al ea u es o he CBX3 is s ill missing. To add ess his poin , we in es iga ed he an i-silencing abili y o a ious CBX3 sub agmen s. While all sub ag- men s excep o he 5′ CBX3(1-339) agmen showed subs an ial an i-silencing ac i i y in he plu ipo en s a e, in di e en ia ed cells wo sub agmen s, CBX3(1-339) and CBX3(340-508), ailed o show an i-silencing ac i - i y. The emaining sub agmen s CBX3(1-508), CBX3(85-508), CBX3(170-508) and CBX3(503-679) all showed ac i i y in plu ipo en as well as di e en ia ed cells. These da a indica e di e en mechanisms o be esponsible o ansgene silencing in he plu ipo en s a e o du ing cellula di e en ia ion. While ansgene silencing in he plu ipo en s a e is well documen ed1, 9, 10, 21, also he ex ensi e ch oma in emodelling associa ed wi h he exi om plu ipo ency has been ad oca ed as a ac o ha impai s e o i al ansgene exp ession. In his con- ex , loss o ansgene exp ession in di e en ia ed cells has been desc ibed by a numbe o g oups6, 22, and his no ion is u he suppo ed by p e ious da a om ou g oup desc ibing p o ound ansgene silencing as well as an i-silencing e icacy o he A2UCOE pa icula ly du ing he ea ly di e en ia ion s a e o ESCs and iPSCs when cells lea e plu ipo ency9, 10, 15. O no e, ou da a imply ha he an i-silencing unc ion o he CBX3 elemen canno be mapped o a single egion and hus emphasize he CBX3 elemen as a minimal UCOE e sion wi h almos comple e unc ionali y o he o iginal 1.5 kb A2UCOE. While all sub agmen s displayed a leas some le el o ac i i y, none o he agmen s could comple ely subs i u e o he CBX3 in ou expe imen al models. Ne e heless, he conside able ac i i y o he CBX3(503-679) and he h ee sub agmen s con aining he 170-508 bp sequence indica e he ele- ance o he cen al as well as he 3′ po ion o he CBX3 in his con ex . In con as , he a he low ac i i y o he CBX3(1-339) sub agmen and he simila p ope ies o he CBX3(1-508), CBX3(85-508), and CBX3(170-508) sub agmen s sugges only mode a e con ibu ion, i any a all, om he 5′ end o he CBX3. We ha e no been able, howe e , o gene a e CBX3 sub agmen s lacking only he 5′ 170 o 339 bp o mo e o mally es his hypo h- esis. Impo an ly, we demons a ed a signi ican co ela ion o agmen leng h and, e en mo e impo an , o e all CpG con en wi h an i-silencing ac i i y, u he suppo ing he no ion ha he highly CpG-dense cen al and 3′ egions may be c ucial de e minan s o CBX3 unc ion. The CpG-islands wi hin he A2UCOE ha e long been conside ed a cen al componen o he an i-silencing unc ion. Indeed, we now show ha a sc ambled CBX3 elemen (CBX3sc CpG) ha bea s no esemblance o he CBX3 (o any o he UCOE) besides main ained CpG-si es, has he abili y o p o ec len i i al ansgene exp es- sion in plu ipo en cells and he eo di e en ia ed p ogeny. A comple ely sc ambled CBX3 a ian (CBX3sc ) lacking he CpG-si es displayed minimal o no unc ionali y. Al hough no comple ely equi alen in unc ionali y o he CBX3 elemen , his abili y o he CBX3sc CpG is qui e ema kable, and s ongly indica es he CpG-si es o he CBX3 as impo an s uc u al de e minan s o i s an i-silencing unc ion. CpG-islands a e de ec ed in he p oximi y o 60-70% o gene p omo e s23, 24, and hei non-me hyla ed CpG-si es ha e been iden i ied as speci ic a ge s o a ious his one modi ying p o eins as well as DNA deme hyla ing enzymes ( e iewed in e . 25). All o hese p o eins including amongs o he s C p1 (CxxC inge p o ein 1), Mll (mixed lineage leukemia p o ein), o Kdm2a (H3K36 deme hylase) sha e a common ea u e, a zinc inge CxxC binding domain ha speci ically ecog- nizes and binds non-me hyla ed CpG-si es26. Speci ically C p1 has been shown o associa e wi h non-me hyla ed CpGs bo h in i o and in i o27 and o in e ac wi h he H3K4 me hyl ans e ase Se d128, 29, he eby leading o he deposi ion o he ac i e ch oma in ma k H3K4me3. In e es ingly, a i icial, p omo e -less, non-me hyla ed CpG-s e ches we e able o ec ui C p1 and c ea e new H3K4me3 ma ks when in eg a ed in o he genome o mouse ESCs27. Thus, u u e s udies ha in es iga e he physical in e ac ion o CBX3 wi h such CxxC-domain con aining p o eins may help o shed ligh on he molecula mechanism by which he CpG-islands o he CBX3 media e i s an i-silencing unc ion. Mo eo e , egions highly en iched in non-me hyla ed CpG-dinucleo ides (CpG islands) a e associa ed wi h an open ch oma in s uc u e ha pa icula ly in plu ipo en cells is ma ked by a bi alen ch oma in con ain- ing bo h ac i e as well as ep essi e ch oma in ma ks30. Syn he ic CpG-islands ha e been shown o induce his bi alen s a e in ESCs i he syn he ic sequence in addi ion o he CpG-dinucleo ides is also en iched o G and C nucleo ides31. This may imply ha in addi ion o he CpG-si es also he o e all GC-densi y o he CBX3 and he eo de i ed sub agmen s suppo s he o e all unc ion o hese elemen s. Conside ing he impac o ansc ip ional ac i i y on he ch oma in s a us and he ec ui men o bind- ing ac o s, such as C p1, in he icini y o CGIs, i becomes e iden ha he p omo e ac i i y o CBX3 may