Histone H2AX is integral to hypoxia-driven neovascularization

Transcription

1 Histone H2AX is integrl to hypoxi-driven neovsculriztion Mtin Economopoulou 1,5, Hrld F Lnger 1,5, Arkdy Celeste 1, Vleri V Orlov 1, Eun Young Choi 1, Mingcho M 2, Athnssios Vssilopoulos 3, Els Cllen 1, Chuxi Deng 3, Crig H Bssing 4, Mnfred Boehm 2, Andre Nussenzweig 1 & Trintfyllos Chvkis 1 29 Nture Americ, Inc. All rights reserved. H2A histone fmily memer X (H2AX, encoded y H2AFX )nd its C-terminl phosphoryltion (c-h2ax) prticiptes in the DNA dmge response nd medites DNA repir 1 6.Hypoxiis physiologicl stress tht induces repliction-ssocited DNA dmge response 7. Moreover, hypoxi is the mjor driving force for neovsculriztion 8, s the hypoxi-medited induction of vsculr growth fctors triggers endothelil cell prolifertion 8. Here we studied the role of the hypoxi-induced DNA dmge response in endothelil cell function nd in hypoxi-driven neovsculriztion in vivo. Hypoxi induced replictionssocited genertion of c-h2ax in endothelil cells in vitro nd in mice. Both in cultured cells nd in mice, endothelil cell prolifertion under hypoxic conditions ws reduced y H2AX deficiency. Wheres developmentl ngiogenesis ws not ffected in H2fx / mice, hypoxi-induced neovsculriztion during pthologic prolifertive retinopthy, in response to hind lim ischemi or during tumor ngiogenesis ws sustntilly lower in H2fx / mice. Moreover, endothelil-specific H2fx deletion resulted in reduced hypoxi-driven retin neovsculriztion nd tumor neovsculriztion. Our findings estlish tht H2AX, nd hence ctivtion of the DNA repir response, is needed for endothelil cells to mintin their prolifertion under hypoxic conditions nd is crucil for hypoxi-driven neovsculriztion. Upon DNA dmge induction in cells, histone H2AX is rpidly phosphorylted t conserved serine-glutmine motif in its croxy terminus (mino cid residues 139 nd 14) 3. H2AX cn medite DNA repir y mintining repir nd signling fctors in proximity to the DNA dmge site 3 6. The kinses orchestrting the DNA dmge response include txi telengiectsi mutted kinse (ATM), ATM- nd Rd3-relted kinse (ATR) nd DNA-dependent protein kinse 2,3,9,1. These signling pthwys cn e ctivted y ionizing rdition or y repliction-ssocited stresses, such s hydroxyure or phidicolin 7,11,12. Hypoxi lso induces type of repliction stress. Previous studies hve suggested tht hypoxi, lthough not necessrily inducing DNA dmge to cells, cn induce DNA dmge response, including the ctivtion of the ATR pthwy, the ATM pthwy or oth, s well s the phosphoryltion of H2AX Hypoxi is the mjor trigger of neovsculriztion. Upon hypoxi, stiliztion of the hypoxi-inducile trnscription fctor results in the upregultion of lrge numer of genes, including vsculr endothelil growth fctor, therey stimulting neovsculriztionrelted endothelil cell functions such s prolifertion nd sprouting 8. Although vessel formtion is essentil for norml development, hypoxi-induced neovsculriztion in dults cn lso e triggered y pthologicl conditions 8. This neovsculriztion response my e insufficient, s in ischemic disese, or excessive, s in prolifertive retinopthies, including the retinopthy of premturity (ROP) 8,16.In ROP, the retinl hypoxi ssocited with reduced vsculriztion fter premture irth induces exoritnt norml vessel growth tht my result in complictions such s hemorrhges nd even lindness 17. During the initil steps of hypoxi-induced neovsculriztion, endothelil cells re suject to hypoxi 8,18. Hypoxi cn induce repliction rrest in tumor cells nd other cell types 7, yet endothelil cells proliferte in response to hypoxi. Thus, we resoned tht n efficient DNA repir response in endothelil cells my e prerequisite for their prolifertion nd for neovsculriztion under hypoxic conditions. To exmine the DNA repir response to hypoxi in endothelil cells, we exploited H2AX, s g-phosphoryltion of H2AX t Ser139 is very sensitive mrker of the DNA dmge response 3 6,ndpreviousstudy hs shown g-h2ax induction in tumor cells y hypoxi 13. Moreover, H2AX cn medite DNA repir 3 5 ; thus, H2AX-deficient cells nd mice 4,19 provide excellent tools to functionlly ddress the role of DNA repir pthwys in hypoxi-induced ngiogenesis. Hypoxi tretment induced distinct g-h2ax foci in primry cultures of humn umilicl vein endothelil cells (HUVECs; Fig. 1). For comprison, we lso sujected endothelil cells to lowgrde irrdition (2 Gy), which cuses doule-strnded DNA reks, or treted them with hydroxyure, which induces replictionssocited DNA dmge (Fig. 1). Wheres irrdition induced 1 Experimentl Immunology Brnch, Ntionl Cncer Institute, 2 Trnsltionl Medicine Brnch, Ntionl Hert, Lung nd Blood Institute nd 3 Genetics of Development nd Disese Brnch, Ntionl Institute for Dietes nd Digestive nd Kidney Diseses, US Ntionl Institutes of Helth, Bethesd, Mrylnd, USA. 4 Deprtment of Pthology nd Lortory Medicine, Center for Childhood Cncer Reserch, Children s Hospitl of Phildelphi, Armson Fmily Cncer Reserch Institute, University of Pennsylvni School of Medicine, Phildelphi, Pennsylvni, USA. 5 These uthors contriuted eqully to this work. Correspondence should e ddressed to T.C. (chvkist@mil.nih.gov). Received 2 Octoer 28; ccepted 27 Ferury 29; pulished online 19 April 29; doi:1.138/nm.1947 NATURE MEDICINE VOLUME 15 [ NUMBER 5 [ MAY

2 29 Nture Americ, Inc. All rights reserved. Figure 1 Genertion of g-h2ax in response to hypoxi in endothelil cells nd the importnce of H2AX for endothelil cell prolifertion under hypoxi. () Immunofluorescence stining for g-h2ax (green) followed y confocl microscopy. HUVECs were kept in normoxi, irrdited with 2 Gy, treted with hydroxyure (HU) or exposed to hypoxi for 1 h or 6 h, s indicted. Nuclei re shown y DAPI stining (lue). Representtive imges re depicted (n ¼ 3). Scle r, 1 mm. () Western lot nlysis of the expression of ATR (left) or ATM (right) in HUVECs trnsfected with sirna ginst ATR (ATR sirna) or ATM (ATM sirna) or control (Con) sirna. (c) Western lot nlysis of the expression of g-h2ax in HUVECs trnsfected with sirna trgeting ATR, sirna trgeting ATM or control sirna (n ¼ 3). HUVECs were kept in normoxi or exposed to hypoxi for 6 h. Actin ws used s loding control. (d) Western lot nlysis of totl H2AX in HUVECs trnsfected with sirna trgeting H2AX or control sirna. Actin ws used s loding control. (e) Prolifertion of HUVECs trnsfected with sirna trgeting H2AX or control sirna, s ssessed under normoxic conditions or hypoxic conditions in the presence of 1% FCS or FGF-2 (5 ng ml 1 ). Bseline prolifertion under ech condition in the sence of stimulus ws defined s the 1% control. Dt re mens ± s.e.m. (n ¼ 4). P o.5. (f) Prolifertion of nd H2fx / primry mouse lung endothelil cells, s ssessed under normoxic conditions or hypoxic conditions in the presence of 1% FBS (seline conditions) or in the presence of 2% FBS nd 1% endothelil cell growth serum (ECGS) (stimulted conditions). Prolifertion ws ssessed y cell counting nd is expressed s the solute cell numer. Dt re mens ± s.e.m. (n ¼ 3). P o.5. (g) Immunofluorescence nlysis in P15 retins for lectin (red) indicting lood vessels, g-h2ax (green) nd DAPI (lue) in mice sujected to ROP protocol or mice kept in room ir (non-rop). Representtive imges re shown (n ¼ 4 mice per group). Scle r, 1 mm. g-h2ax in ll endothelil cells, hypoxi tretment induced phenotype similr to tht induced y hydroxyure, in tht it resulted in g-h2ax formtion only in supopultion of cells (Fig. 1). The percentges of cells showing g-h2ax foci were 3.1 ±.5%, 99.1 ±.4%, 19.1 ± 1.2%, 1.9 ±.7% nd 17.1 ± 1.2% for normoxic, irrdited, hydroxyure-treted, 1-h hypoxi-treted or 6-h hypoxi-treted cells, respectively (mens ± s.e.m.). Western lot nlysis confirmed the genertion of g-h2ax in HUVECs upon hypoxi tretment (Supplementry Fig. 1 online). Hypoxiinduced g-h2ax foci were predominntly present in proliferting endothelil cells tht stined positive for prolifertive cell nucler ntigen 2 (Supplementry Fig. 2 online) nd coloclized with repliction protein A 21 (Supplementry Fig. 3 online), suggesting repliction stress ssocited genertion of g-h2ax y hypoxi. We then determined which of the two mjor DNA repir signling pthwys, the ATM pthwy or the ATR pthwy, ws involved in the genertion of g-h2ax in endothelil cells during hypoxi. Smll interfering RNA (sirna)-medited knockdown of ATR, ut not of ATM, prevented hypoxi-dependent g-h2ax production (Fig. 1,c). We otined similr results when we used other ATR-specific sirnas (dt not shown). Tht ATM is not involved in the hypoxi-induced response in HUVECs ws supported y the sence of ATM phosphoryltion t Ser1981, mrker of ATM ctivtion 9, fter hypoxi tretment (dt not shown). These dt suggest tht hypoxi induces repliction stress ssocited g-h2ax genertion in endothelil cells in n ATR-dependent mnner. We next ddressed the functionl importnce of the DNA repir response for endothelil cell prolifertion nd hypoxi-induced neovsculriztion. We studied the effect of loss of H2AX on endothelil cell prolifertion, either in HUVECs y compring the effects of H2AX-trgeted nd control sirna (Fig. 1d) or in primry mouse endothelil cells y compring cells isolted from H2fx +/+ nd H2fx / mice. In HUVECs, H2AX downregultion induced f Normoxi Irrdition HU Hypoxi 1 h Hypoxi 6 h e Cell prolifertion Con ATR sirna sirna 6 Con sirna ATM sirna ATM Con sirna 5 H2AX sirna c Normoxi Hypoxi d H2AX Con sirna: Con ATR ATM Con ATR ATM sirna sirna 4 γ-h2ax H2AX 3 2 Actin Actin 1 FGF-2 FCS FGF-2 FCS Normoxi Hypoxi + 1% FBS H2fx g / + 1% FBS Lectin γ-h2ax DAPI Merge Cell prolifertion (cell numer 1 3 ) % FBS + ECGS Normoxi + 2% FBS + ECGS Hypoxi ROP Non- ROP γ-h2ax DAPI Tuulin mrginl reduction in firolst growth fctor-2 (FGF-2)- or fetl clf serum (FCS)-induced cell prolifertion under normoxi (Fig. 1e). Notly, H2AX downregultion significntly decresed growth fctor induced nd FCS-induced HUVEC prolifertion under hypoxic conditions (Fig. 1e). We oserved similr results using H2AX-deficient primry mouse endothelil cell cultures (Fig. 1f). In contrst, other prongiogenic functions of endothelil cells, such s extrcellulr mtrix dependent tue formtion, were not ffected y the sence of H2AX (Supplementry Fig. 4 online). These dt indicte crucil role for H2AX in mintining endothelil cell prolifertion under hypoxic conditions. These findings point to role of H2AX in hypoxi-driven neovsculriztion. We investigted this hypothesis in vivo using mouse model of ROP in which hypoxi induces retinl neovsculriztion 17,22,23. In the ROP model, postntl dy 7 (P7) pups re incuted in 75% oxygen for 5 d, which induces regression nd olitertion in the developing retin vsculture. This results in mssive retinl hypoxi fter P12, when pups re trnsferred to norml oxygen, leding to incresed endothelil cell prolifertion nd pthologicl exuernt neovsculriztion (prolifertive retinopthy) 17,22,23. In pups sujected to the ROP protocol, we found incresed mounts of g-h2ax during the hypoxi phse (P13 P15) compred to control mice kept under normoxic conditions (Supplementry Fig. 5 online). Immunostining indicted tht this increse ws, t lest in prt, due to g-h2ax genertion in endothelil cells, with little g-h2ax stining ssocited with vsculr nuclei in the developing vsculture of control pups tht were kept under normoxic conditions (Fig. 1g). Induction of g-h2ax in retins of mice sujected to the ROP protocol ws evident predominntly in the nuclei of proliferting retinl vsculr endothelil cells, including endothelil nuclei in pthologicl neovsculr tufts (Supplementry Fig. 5). Thus, hypoxi induces genertion of g-h2ax in the retinl vsculture in vivo. ATR Tuulin 554 VOLUME 15 [ NUMBER 5 [ MAY 29 NATURE MEDICINE

3 Retinl neovsculriztion (neovsculr nuclei per section) c d Lectin BrdU Merge e f g Prolifertion Apoptosis Retinl neovsculriztion (neovsculr nuclei per section) Cre Cre + H2fx flox/ H2fx flox/ 29 Nture Americ, Inc. All rights reserved. Figure 2 A role for H2AX in retinl neovsculriztion under hypoxi. () Quntifiction of retinl neovsculriztion on P17 in nd H2fx / mice sujected to the ROP protocol. Dt re mens ± s.e.m. (n ¼ 3 or 5 mice per group). P o.5. () Periodic cid Schiff nd hemtoxylin stining of 6-mm prffin-emedded xil sections of the retin. Sections from the retins of nd H2fx / mouse sujected to the ROP protocol re shown. The rrows indicte newly formed vessels nterior to the internl limiting memrne (n ¼ 3 or 5 mice per group). Scle r, 1 mm. (c) Lectin stining of retin whole mounts t P17 from or H2fx / mouse sujected to the ROP protocol (n ¼ 3 mice per group). For ech imge, multiple individul imges were stitched together. Scle r, 5 mm. (d) Lectin (green) nd BrdU (red) stining of retin whole mounts t P17 from or H2fx / mouse sujected to the ROP protocol. Arrows indicte the neovsculr tufts. Representtive imges re shown (n ¼ 4 or 5 mice per group). Scle r, 1 mm. (e) Endothelil cell prolifertion t dy 15 in the retins of mice sujected to the ROP protocol, s ssessed y BrdU incorportion. Endothelil cell prolifertion in mice represents the 1% control. Dt re mens ± s.e.m. (n ¼ 3 or 4 mice per group). P o.5. (f) Apoptosis t dy 15 in retinl vessels of mice sujected to the ROP protocol, s ssessed y TUNEL stining. Endothelil cell poptosis in mice represents the 1% control. Dt re mens ± s.e.m. (n ¼ 3 mice per group). P o.5. (g) Retinl neovsculriztion quntified on P17 in H2AX-sufficient (VEC-Cre H2fx flox/ ) nd endothelil-specific H2AXdeficient (VEC-Cre + H2fx flox/ ) mice sujected to the ROP protocol. Dt re mens ± s.e.m. (n ¼ 5 or 6 mice per group). P o.5. We next studied the effect of H2AX deficiency on norml development of the retinl vsculture (etween dys P1 nd P2) nd on hypoxi-induced neovsculriztion in the ROP model. Consistent with low levels of g-h2ax in the developing retinl vsculture (Fig. 1g), we oserved no difference etween wild-type () nd H2fx / mice when we nlyzed retinl vsculr development t dys P5, P1 nd P15 (Supplementry Fig. 6 online). We then compred hypoxi-induced retinl neovsculriztion t P17 in nd H2fx / mice sujected to the ROP protocol. We oserved significnt (P o.5) decrese in hypoxi-driven retinl neovsculriztion in H2fx / mice (Fig. 2). Immunohistochemicl nlysis of retinl sections reveled 5% reduction in the extent of pthologicl neovsculriztion in H2fx / mice (Fig. 2,). We oserved fewer pthologicl neovsculr tufts nd more extensive physiologicl vsculture in retin whole mounts from H2fx / mice s compred to mice (Fig. 2c,d). This decrese in pthologicl retinl neovsculriztion ws ccompnied y decrese in endothelil cell prolifertion nd n increse in endothelil cell poptosis (Fig. 2d f). To test the role of endothelil H2AX in hypoxi-driven pthologicl ngiogenesis nd exclude the possiility tht systemic H2AX deficiency could ccount for the oserved phenotype, we generted endothelilspecific H2AX-deficient mice using the vsculr-endothelil Figure 3 H2AX deficiency decreses neovsculriztion nd lood flow fter hind lim ischemi. () Blood flow, s ssessed y Doppler scnning in nd H2fx / mice sujected to the hind lim ischemi protocol. Dt re mens ± s.e.m. (n ¼ 5 mice per group). () Endothelil cell prolifertion in the ischemic muscle in H2fx / nd mice. Dt re mens ± s.e.m. (n ¼ 3 or 5 mice per group). (c) Representtive imges of BrdU (red) nd CD31 (green) doule immunostining in ischemic muscles of nd H2fx / mice (n ¼ 3 or 5 mice per group). Blue, DAPI stining to identify nuclei. Scle rs, 1 mm. P o.5; P o.1. (VE)-cdherin Cre system 24 nd H2fx-floxed mice 25.Inprticulr,we crossed H2fx flox/flox mice with H2fx +/ mice tht expressed the Cre recominse under the control of the VE-cdherin promoter (VEC- Cre + ) to generte VEC-Cre + H2fx flox/ mice (endothelil-specific H2AX-deficient) nd VEC-Cre H2fx flox/ mice (H2AX-sufficient). Endothelil-specific H2AX-deficient mice showed significntly less neovsculriztion s compred to H2AX-sufficient mice (Fig. 2g nd Supplementry Fig. 7 online). These findings indicte tht endothelil H2AX is dispensle during norml vsculr development ut is functionl in the context of hypoxi-driven retinl ngiogenesis. We next studied the role of H2AX in the hind lim ischemi model, in which neovsculriztion is induced y cute disruption of lood supply to the hind lim nd consequent tissue ischemi nd hypoxi 26. Consistent with the findings from the ROP model, H2fx / compred to mice showed impired neovsculriztion Reltive lood flow CD31 + /BrdU + rtio of totl BrdU + cells (%) Time (d) c NATURE MEDICINE VOLUME 15 [ NUMBER 5 [ MAY

4 29 Nture Americ, Inc. All rights reserved. Vessels per field f Tumor volume (cm 3 ) Dy 21 Dy Time (d) in the hind lim ischemi model (Fig. 3). Wheres the group recovered to 96.6% nd 7.7% of norml lood flow t dys 14 nd 21, respectively, the H2fx / group recovered to only 48% nd 42.9% on these dys (Fig. 3). In ddition, we found reduction in endothelil cell prolifertion in the ischemic muscles of H2fx / mice s compred to mice (Fig. 3,c). To provide further evidence for the specific function of H2AX in hypoxi-driven neovsculriztion s opposed to neovsculriztion under normoxic conditions, we studied hypoxi-driven tumor neovsculriztion nd growth fctor induced neovsculriztion. After injecting Lewis lung crcinom cells sucutneously in nd H2fx / mice, we ssessed tumor ngiogenesis y stining tumor sections for the endothelil-specific mrker CD31. We oserved significnt decrese in vsculr density t two different time points (dys 21 nd 25 fter tumor implnttion) in tumors implnted into H2fx / mice (Fig. 4,). In ddition, we found decresed endothelil cell prolifertion nd incresed endothelil cell poptosis in tumors implnted into H2fx / mice (Fig. 4c,d). Pericyte coverge (NG-2 positive cells) of vessels ws similr in tumor vessels of nd H2fx / mice (Fig. 4e), suggesting tht H2AX deficiency leds to reduced tumor ngiogenesis predominntly y ffecting endothelil cell prolifertion rther thn vessel mturtion. Associted with impired tumor ngiogenesis, oth the growth nd the finl weight of tumors were significntly reduced in H2fx / mice (Fig. 4f,g). To investigte the role of endothelil H2AX in tumor ngiogenesis, we implnted tumors into endothelil-specific H2AX-deficient mice nd littermte H2AX-sufficient mice. Tumors implnted into endothelilspecific H2AX-deficient mice showed significntly (P o.1) less neovsculriztion s compred to tumors implnted into H2AXsufficient mice (Supplementry Fig. 8 online). In ddition, tumor c Endothelil cell prolifertion d Endothelil cell poptosis e g Pericyte coverge (%) Tumor weight (g) Figure 4 Tumor ngiogenesis nd tumor growth re reduced due to H2AX deficiency. () Tumor ngiogenesis in or H2fx / mice. Tumors implnted sucutneously were excised nd stined for CD31 to visulize lood vessels. The numer of vessels per field t dys 21 nd 25 were quntified. The dt re expressed s percentge of control, set s the numer of vessels per field t dy 21 or dy 25 in mice. Dt re mens ± s.e.m. (n ¼ 6 1 tumors per group). P o.5; P o.1. () Representtive imges of CD31 stining in tumors implnted into or H2fx / mice (n ¼ 6 1 tumors per group). Scle r, 1 mm. (c) Endothelil cell prolifertion in tumors excised from nd H2fx / mice, s ssessed y evluting the numer of BrdU-positive endothelil cell nuclei per field. Endothelil cell prolifertion in mice represents the 1% control. Dt re mens ± s.e.m. (n ¼ 6 9 tumors per group). P o.5. (d) Endothelil cell poptosis in tumors excised from nd H2fx / mice, s ssessed y evluting the numer of cspse 3-positive endothelil cells per field. Endothelil cell poptosis in mice represents the 1% control. Dt re mens ± s.e.m. (n ¼ 4 tumors per group). P o.5. (e) Pericyte coverge of vessels in tumors of nd H2fx / mice, s ssessed y NG-2 stining. Pericyte coverge is shown s the percentge of vessels covered y NG-2 positive cells. Dt re mens ± s.e.m. (n ¼ 8 tumors per group). (f) Tumor volume ws followed from 6 d to 25 d fter the tumors were implnted into nd H2fx / mice. Dt re mens ± s.e.m. (n ¼ 8 1 tumors per group). P o.1 s compred to the volume of tumors implnted in mice on the sme dy. (g) The weight of tumors excised from or H2fx / mice t dy 25. Dt re mens ± s.e.m. (n ¼ 8 1 tumors per group). P o.5. growth ws significntly (P o.5) decresed in endothelil-specific H2AX-deficient mice (Supplementry Fig. 8). These dt indicte tht endothelil H2AX hs role in hypoxi-driven tumor ngiogenesis. In contrst to tumor ngiogenesis, ngiogenesis induced y FGF-2 in Mtrigel plugs injected into or H2fx / mice ws not ffected y H2AX deficiency (Supplementry Fig. 9 online). Here we hve found tht hypoxi-triggered neovsculriztion requires the function of H2AX in endothelil cells. Phosphoryltion of H2AX ws induced y hypoxi in proliferting endothelil cells in vitro nd in the mouse retin in vivo, in ccordnce with previous oservtions tht hypoxi induces repliction stress ssocited DNA repir response in other cell types 7,13,27. The hypoxi-induced DNA repir response my not e the result of induction of DNA dmge y hypoxi itself. Low levels of DNA dmge occur during repliction, nd hypoxi-induced repliction stress my result in ccumultion of this dmge 28. Alterntively, hypoxi hs een proposed s direct inducer of DNA dmge y repressing genes crucil to DNA repir 29,3. Irrespective of the mechnism responsile, we showed tht hypoxi-induced ctivtion of g-h2ax nd n intct DNA repir response in endothelil cells re of prticulr importnce during hypoxi-driven pthologicl neovsculriztion. Loss of H2AX in endothelil cells mrkedly compromises endothelil cell prolifertion under hypoxi ut only mrginlly under normoxi. Moreover, hypoxi-induced neovsculriztion during pthologic prolifertive retinopthy, in response to hind lim ischemi or during tumor growth ws significntly reduced y H2AX deficiency, wheres physiologicl developmentl vsculriztion ws not ffected. Our findings point to the possiility tht vsculr-trgeted inhiitors of H2AX my provide new strtegy for preventing hypoxi-induced pthologicl neovsculriztion in conditions such s in tumor growth, ROP or dietic retinopthy 16,17. METHODS Mice nd hypoxi-induced retinl vsculriztion. Mouse studies were pproved y the US Ntionl Cncer Institute Animl Cre nd Use Committee. H2fx / mice hve een previously descried 19. Littermte H2fx / or H2fx +/+ mice were used in the ROP model tht ws induced s 556 VOLUME 15 [ NUMBER 5 [ MAY 29 NATURE MEDICINE

5 29 Nture Americ, Inc. All rights reserved. previously descried y us nd others 22,23,31. At dy P17, we killed the mice nd processed their eyes for quntifiction of preretinl neovsculr nuclei, s previously descried 22,23. Briefly, we stined 6-mm prffin-emedded seril sections with periodic cid Schiff nd evluted ten intct sections of equl length, ech 18 mm prt. We counted ll retinl vsculr cell nuclei nterior to the internl limiting memrne in ech section. We clculted the men numer of the neovsculr nuclei from ten counted sections. We did not oserve ny neovsculr cell nuclei nterior to the internl limiting memrne in norml mice kept under normoxic conditions. Moreover, we otined C57BL/6 mice from Chrles River nd sujected them to the ROP protocol or kept them under normoxic conditions, nd we used their retins for nlysis of H2AX phosphoryltion, s descried in the Supplementry Methods online. In ddition, we sujected endothelil-specific H2AX-deficient nd H2AXsufficient mice to the ROP protocol. H2fx flox/flox mice hve een previously descried 25. To induce deletion of H2fx in endothelil cells, we crossed H2fx flox/flox mice (kindly provided y F. Alt) with H2fx +/ mice tht lso crried gene construct in which expression of the Cre recominse is under the control of the promoter of the gene encoding VE-cdherin 24 (VEC-Cre + )to generte VEC-Cre + H2fx flox/ mice (endothelil-specific H2AX-deficient) nd VEC-Cre H2fx flox/ mice (H2AX-sufficient, used s littermte controls). We purchsed VEC-Cre + mice from the Jckson Lortory. We ssessed the efficiency of Cre-medited excision in vsculr endothelil cells y immunohistochemistry nlysis (Supplementry Fig. 7). Briefly, we froze nd sectioned eyes from endothelil-specific H2AX-deficient or H2AX-sufficient mice. We fixed the sections in ice-cold cetone for 1 min, wshed them with PBS, locked them with PBS contining 1% BSA, 5% got serum (Sigm-Aldrich) nd.1% Triton-X 1 (Fluk), nd incuted them with rit ntiody specific for totl H2AX (1 in 1, Novus) t 4 1C overnight. After wshing the sections with PBS, we further incuted them with n AlexFluor 568 conjugted got ntiody to rit IgG (Invitrogen), with FITC-conjugted Bndeire simplicifoli isolectin B4 (Sigm-Aldrich) to visulize vessels, nd with DAPI. We otined the imges with Zeiss lser-scnning confocl microscope. Retin whole mounts. We used previously pulished protocol to prepre retin whole mounts 22,32. Briefly, fter sujecting mice to the ROP protocol, we injected them intrperitonelly t dy P16 with BrdU nd killed them t dy P17. We enucleted the eyes nd fixed them t 4 1C overnight. After extrction nd wshing of the retins, we permeilized them t 4 1C overnight nd stined them with FITC-conjugted B. Simplicifoli isolectin B4. We further stined some retins for BrdU. We sujected these retins to ntigen retrievl with 6N HCl nd.1% Triton-X 1, followed y locking with PBS contining 1% BSA, 5% got serum nd incution with iotinylted BrdU-specific ntiody (1 in 1, BD Biosciences). We detected BrdU-specific ntiody with AlexFluor 568 conjugted streptvidin t 4 1C for 6 h. Finlly, we fltmounted the retins nd pplied cover slip. We otined imges with Zeiss lser-scnning confocl microscope nd Zeiss Axiovert 2M with Axiovision 4.6 softwre. Tumor ngiogenesis. We followed previously descried protocol for tumor growth nd ngiogenesis 33. We sucutneously injected Lewis lung crcinom cells (1 1 6 ) t two sites in the cks of littermte H2fx / or mice or into the cks of endothelil-specific H2AX-deficient or H2AX-sufficient mice. We determined tumor volume nd tumor weight fter tumor excision s previously descried 33. At ech time point, we excised nd fixed the implnted tumors in 4% prformldehyde solution for 1 h t 4 1Cndincutedthemin 15% nd 3% sucrose for 16 nd 24 h, respectively. We sectioned the tissues (8-mm thick sections), locked the slides with 5% got serum in 1% BSA nd.1% Triton-X 1 in PBS for 1 h t 22 1C nd stined them with rt CD31- specific primry ntiody (BD Biosciences) overnight t 4 1C. After wshing the sections with PBS, we proed them with n AlexFluor 568 conjugted got ntiody to rt IgG (Invitrogen). We ssessed tumor vsculrity y counting CD31-positive vessels in six to ten sections per tumor. To ssess vessel prolifertion, we stined dy 25 tumor sections for BrdU using the in situ BrdU kit (BD Biosciences) s descried in the Supplementry Methods, nd to visulize vessels we stined for CD31. We counted BrdUpositive vsculr nuclei in two rndom fields per section from six different sections per tumor. We ssessed poptosis in the vessels of tumors implnted in or H2fx / mice y stining for CD31 nd cleved cspse-3: we processed tumors s descried in the previous prgrph nd incuted sections with rit ntiody ginst mouse cleved cspse-3 (1 in 3, Cell Signling Technology) nd with rt CD31-specific ntiody overnight t 4 1C. We detected primry ntiodies using fluorescently leled secondry ntiodies specific for rit nd rt IgG (Invitrogen). We counted the cleved cspse-3 positive endothelil cells in three rndom fields per section from six different sections per tumor. To visulize pericyte coverge of the tumor vessels in tumors implnted in H2fx / nd H2fx +/+ mice, we used doule stining of CD31 nd NG2. We stined 1-mm sections for CD31 s descried ove, nd we lso incuted them with rit ntiody to NG2 (1 in 1, Millipore) t 4 1C overnight. We detected primry ntiodies using fluorescently leled secondry ntiodies specific for rit nd rt IgG. We counted NG-2 positive cells covering lood vessels in four different sections per tumor. Additionl methodology. Detiled methods for endothelil cell culture nd trnsfections, in vitro immunofluorescence, western lotting, in vitro endothelil cell prolifertion ssy, mouse hind lim ischemic surgery, endothelil cell prolifertion nd poptosis in the retin in vivo, retin digestion nd immunofluorescence, physiologicl developmentl retinl ngiogenesis, in vitro Mtrigel tue formtion ssy nd in vivo Mtrigel ssy re descried in the Supplementry Methods. Note: Supplementry informtion is ville on the Nture Medicine wesite. ACKNOWLEDGMENTS This reserch ws supported y the Intrmurl Reserch Progrm of the US Ntionl Institutes of Helth Ntionl Cncer Institute. H.F.L. ws supported y the Germn Acdemy of Sciences (Leopoldin). We would like to cknowledge M.E. Kruhlk for the help with microscopy, D. Winkler nd S. Kul for help with genotyping, G. Tosto for help with Mtrigel experiments, F. Alt (Hrvrd University) for providing the H2fx floxed mice nd A. Singer nd D.S. Singer for criticlly reding the mnuscript. AUTHOR CONTRIBUTIONS M.E. designed nd conducted experiments nd wrote the mnuscript, H.F.L. conducted experiments, A.C. designed experiments, V.V.O. conducted experiments, E.Y.C., M.M., A.V. nd E.C. conducted experiments, C.D. supervised experiments, C.H.B. generted H2fx floxed mice, M.B. designed, conducted nd supervised experiments, A.N. supervised nd designed experiments, nd T.C. designed, conducted nd supervised experiments nd wrote the mnuscript. Pulished online t Reprints nd permissions informtion is ville online t reprintsndpermissions/ 1. Rouse, J. & Jckson, S.P. Interfces etween the detection, signling nd repir of DNA dmge. Science 297, (22). 2. McGown, C.H. & Russell, P. The DNA dmge response: sensing nd signling. Curr. Opin. Cell Biol. 16, (24). 3. Fernndez-Cpetillo, O., Lee, A., Nussenzweig, M. & Nussenzweig, A. H2AX: the histone gurdin of the genome. DNA Repir (Amst.) 3, (24). 4. Celeste, A. et l. Histone H2AX phosphoryltion is dispensle for the initil recognition of DNA reks. Nt. Cell Biol. 5, (23). 5. Bssing, C.H. & Alt, F.W. H2AX my function s n nchor to hold roken chromosoml DNA ends in close proximity. Cell Cycle 3, (24). 6. Sedelnikov, O.A., Pilch, D.R., Redon, C. & Bonner, W.M. Histone H2AX in DNA dmge nd repir. Cncer Biol. Ther. 2, (23). 7. Hmmond, E.M. & Gicci, A.J. The role of ATM nd ATR in the cellulr response to hypoxi nd re-oxygention. DNA Repir (Amst.) 3, (24). 8. Crmeliet, P. Angiogenesis in life, disese nd medicine. Nture 438, (25). 9. Hurley, P.J. & Bunz, F. ATM nd ATR: components of n integrted circuit. Cell Cycle 6, (27). 1. Zhng, Y.W., Hunter, T. & Arhm, R.T. Turning the repliction checkpoint on nd off. Cell Cycle 5, (26). 11. Shiloh, Y. ATM nd ATR: networking cellulr responses to DNA dmge. Curr. Opin. Genet. Dev. 11, (21). 12. Lopes, M. et l. The DNA repliction checkpoint response stilizes stlled repliction forks. Nture 412, (21). NATURE MEDICINE VOLUME 15 [ NUMBER 5 [ MAY

6 29 Nture Americ, Inc. All rights reserved. 13. Hmmond, E.M., Dorie, M.J. & Gicci, A.J. ATR/ATM trgets re phosphorylted y ATR in response to hypoxi nd ATM in response to reoxygention. J. Biol. Chem. 278, (23). 14. Hmmond, E.M., Dorie, M.J. & Gicci, A.J. Inhiition of ATR leds to incresed sensitivity to hypoxi/reoxygention. Cncer Res. 64, (24). 15. Gison, S.L., Bindr, R.S. & Glzer, P.M. Hypoxi-induced phosphoryltion of Chk2 in n txi telngiectsi mutted-dependent mnner. Cncer Res. 65, (25). 16. Arjm, O. & Nikinm, M. Oxygen-dependent diseses in the retin: role of hypoxiinducile fctors. Exp. Eye Res. 83, (26). 17. Cmpochiro, P.A. & Hckett, S.F. Oculr neovsculriztion: vlule model system. Oncogene 22, (23). 18. Noguer-Troise, I. et l. Blockde of Dll4 inhiits tumour growth y promoting nonproductive ngiogenesis. Nture 444, (26). 19. Celeste, A. et l. Genomic instility in mice lcking histone H2AX. Science 296, (22). 2. Celis, J.E. & Celis, A. Cell cycle-dependent vritions in the distriution of the nucler protein cyclin proliferting cell nucler ntigen in cultured cells: sudivision of S phse. Proc. Ntl. Acd. Sci. USA 82, (1985). 21. Zou, L. & Elledge, S.J. Sensing DNA dmge through ATRIP recognition of RPA-ssDNA complexes. Science 3, (23). 22. Economopoulou, M. et l. Inhiition of pthologic retinl neovsculriztion y lphdefensins. Blood 16, (25). 23. Smith, L.E. et l. Oxygen-induced retinopthy in the mouse. Invest. Ophthlmol. Vis. Sci. 35, (1994). 24. Alv, J.A. et l. VE-Cdherin-Cre-recominse trnsgenic mouse: tool for linege nlysis nd gene deletionin endothelil cells. Dev. Dyn. 235, (26). 25.Bssing,C.H.et l. Incresed ionizing rdition sensitivity nd genomic instility in the sence of histone H2AX. Proc. Ntl. Acd. Sci. USA 99, (22). 26. Luttun, A. et l. Revsculriztion of ischemic tissues y PlGF tretment, nd inhiition of tumor ngiogenesis, rthritis nd therosclerosis y nti-flt1. Nt. Med. 8, (22). 27. Hmmond, E.M., Denko, N.C., Dorie, M.J., Arhm, R.T. & Gicci, A.J. Hypoxi links ATR nd p53 through repliction rrest. Mol. Cell. Biol. 22, (22). 28. Hmmond, E.M., Kufmnn, M.R. & Gicci, A.J. Oxygen-sending nd the DNA dmge response. Curr. Opin. Cell Biol. 19, (27). 29. Hung, L.E., Bindr, R.S., Glzer, P.M. & Hrris, A.L. Hypoxi-induced genetic instility clculted mechnism underlying tumor progression. J. Mol. Med. 85, (27). 3. To, K.K.W., Sedelnikov, O.A., Smons, M., Bonner, W.M. & Hung, L.E. The phosphoryltion sttus of PAS-B distinguishes HIF-1 from HIF-2 in NBS1 repression. EMBO J. 25, (26). 31. Orlov, V.V., Economopoulou, M., Lupu, F., Sntoso, S. & Chvkis, T. Junctionl dhesion molecule-c regultes vsculr endothelil permeility y modulting VEcdherin medited cell-cell contcts. J. Exp. Med. 23, (26). 32. Gerhrdt, H. et l. VEGF guides ngiogenic sprouting utilizing endothelil tip cell filopodi. J. Cell Biol. 161, (23). 33. Lin, M.I., Yu, J., Murt, T. & Sess, W.C. Cveolin-1 deficient mice hve incresed tumor microvsculr permeility, ngiogenesis nd growth. Cncer Res. 67, (27). 558 VOLUME 15 [ NUMBER 5 [ MAY 29 NATURE MEDICINE