Competing waves of oligodendrocytes in the forebrain and postnatal elimination of an embryonic lineage

Transcription

1 Compting wavs of oligodndrocyts in th forbrain and postnatal limination of an mbryonic linag Nicoltta Kssaris 1, Matthw Fogarty 1, Palma Iannarlli 1, Matthw Grist 1, Michal Wgnr 2 & William D Richardson 1 Th dvlopmntal origin of oligodndrocyt prognitors (OLPs) in th forbrain has bn controvrsial. W now show, by Cr-lox fat mapping in transgnic mic, that th first OLPs originat in th mdial ganglionic minnc (MGE) and antrior ntopduncular ara (AEP) in th vntral forbrain. From thr, thy populat th ntir mbryonic tlncphalon including th crbral cortx bfor bing joind by a scond wav of OLPs from th latral and/or caudal ganglionic minncs (LGE and CGE). Finally, a third wav ariss within th postnatal cortx. Whn any on population is dstroyd at sourc by th targtd xprssion of diphthria toxin, th rmaining clls tak ovr and th mic surviv and bhav normally, with a normal complmnt of oligodndrocyts and mylin. Thus, functionally rdundant populations of OLPs compt for spac in th dvloping brain. Notably, th mbryonic MGE- and AEP-drivd population is liminatd during postnatal lif, raising qustions about th natur and purpos of th comptition. Diffrnt subclasss of nurons and glia ar gnratd squntially in diffrnt parts of th vntricular zons of th dvloping spinal cord and brain. For xampl, spinal motor nurons and oligodndrocyts (OLs, th mylin-forming clls) ar drivd from prcursors that rsid in a spcializd domain of th vntral vntricular zon calld motor nuron prcursors (pmn), dfind by xprssion of th transcription factor Olig2 (rfs. 1 4). From thr, OLPs migrat all through th spinal cord bfor diffrntiating into mylin-forming OLs. Latr, at last on additional sourc of OLPs ariss in th dorsal spinal cord, contributing 1 15% of th final OL population 5 7. OLP gnration in th pmn dpnds on th signaling molcul Sonic hdghog (Shh), whras th dorsal sourc might b Shh indpndnt 5 9. OL gnration in mor antrior parts of th nural tub particularly th forbrain is not as wll undrstood. Th nuropithlium of th MGE and th AEP in th vntral forbrain xprsss Shh and its rcptor Patchd (Ptc) as wll as Olig2, suggsting that OLPs might b formd primarily in ths rgions Indd, migratory OLPs, dfind by th xprssion of platlt-drivd growth factor rcptor-a (Pdgfra), can b sn straming away from th MGE and AEP in all dirctions aftr mbryonic day 12 (E12) in th mous, apparntly ntring th dvloping crbral cortx (dorsal tlncphalon) around E16 (rfs. 11,13). Th numbr of OLPs in th cortx incrass markdly btwn E16 and birth (BE18), but it is not known whthr this rflcts continuing inward migration and prolifration of vntrally drivd OLPs or a nw sourc of OLPs within th cortx. Thr is som vidnc on both sids. For xampl, som OLPs in th cortx xprss th transcription factors Dlx1 and Dlx2, suggsting that thy originat in th vntral tlncphalon whr Dlx1 and Dlx2 ar first activatd 14,15. Morovr, in chick mbryos, th xprssion of th oligodndrocyt-linag markrs mylin protolipid protin (Plp-Dm2) and th O4 antign is initially rstrictd to th AEP 16, and chick-quail grafting xprimnts indicat that all cortical OLs ar vntrally drivd in birds 17. In contrast, cll fat mapping using Emx1-Cr transgnic mic suggsts that many OLs in th cortx ar gnratd locally 18. Furthrmor, rtroviral fat mapping dmonstrats that cortical OLs continu to b formd in th postnatal priod from prcursors in th subvntricular zons (SVZ) locatd at th tips of th latral vntricls at th corticostriatal boundary 19,2.Itis important to rsolv this uncrtainty bcaus OLs with diffrnt dvlopmntal origins, prhaps spcifid via diffrnt signaling pathways, might hav distinct proprtis and functions in th matur brain. To rsolv th qustion of OL origins, w usd th Cr-lox approach in transgnic mic to follow th dvlopmnt of distinct vntral or dorsal prcursor populations in th tlncphalon. W found that th dvlopmnt of OL linag clls in th forbrain is unxpctdly complx and dynamic. Thr was an arly wav of OLP gnration from Nkx2.1-xprssing prcursors in th vntral forbrain; ths first appard in th vntricular zon of th vntral MGE and AEP around E12.5 and subsquntly migratd widly into all parts of th tlncphalon, ntring th crbral cortx aftr E16. At first thy wr th only OLPs in th forbrain, but soon thir contribution to th total OLP population dclind. By postnatal day 1 (P1), thr wr vry fw Nkx2.1-drivd OLPs or OLs in th cortx, although thy wr still th major population in th vntral (subpallial) rgion. Instad, thy wr ovrtakn in th cortx by othr populations of OLPs drivd first from Gsh2-positiv prcursors in th LGE and CGE and latr from ndognous Emx1-positiv cortical prcursors. Evn in th vntral 1 Wolfson Institut for Biomdical Rsarch and Dpartmnt of Biology, Univrsity Collg London, Gowr Strt, London WC1E 6BT, UK. 2 Institut für Biochmi,Emil- Fischr Zntrum, Fahrstrass 17, 9154 Erlangn, Grmany. Corrspondnc should b addrssd to B.R. (w.richardson@ucl.ac.uk) or N.K. (n.tkki-kssaris@ucl.ac.uk). Rcivd 7 Sptmbr; accptd 3 Novmbr; publishd onlin Dcmbr ; doi:1.138/nn162 NATURE NEUROSCIENCE VOLUME 9 [ NUMBER 2 [ FEBRUARY

2 Nkx2.1-Cr /R26R Nkx2.1-Cr/ /R26R Figur 1 Thr succssiv wavs of OLs gnratd from distinct prcursor populations at diffrnt tims during forbrain dvlopmnt. (a,b) Gnration of th first wav of Pdgfra + OLPs bgan at E12.5 from Nkx2.1-xprssing prcursors in th MGE-AEP. Ths OLPs xprssd (grn) in Nkx2.1-Cr/Rosa26R- mbryos (arrows in b). (c,d) ByE14.5,Pdgfra + + clls bgan to appar at th corticostriatal boundary in Nkx2.1- Cr/Rosa26R- mbryos (arrows in d), dmonstrating migration of MGE-AEP drivd OLPs into th dvloping cortx. (,f) By E16.5, a nw wav of (Pdgfra +, ) OLPs was obsrvd in Nkx2.1-Cr/R26R- cortx; ths must hav bn gnratd from Nkx2.1 prcursors (arrowhads in f). (g,h) All Pdgfra + clls in th tlncphalon coxprssd in Nkx2.1-Cr/ /Rosa26R- tripl transgnic mbryos (arrows in h), indicating that all OLPs wr drivd from th vntral forbrain (MGE-AEP and LGE- CGE) at this stag. (i,j) Anothr wav of (Pdgfra +, ) OLPs appard in th tlncphalon of arly postnatal Nkx2.1-Cr//Rosa26R- mic (arrowhads in j). (k,l) Emx1-xprssing cortical prcursors bgan to gnrat OLPs and OLs aftr birth (arrows in l). (m,n) Tamoxifn administrd onc to prgnant Emx1-CrER T2 fmals at E9.5, bfor invasion of th cortx by vntrally drivd OLPs, activatd Cr rcombination in th mbryos and rsultd in th xprssion of th rportr in a subst of + OLPs (arrows in n) which must thrfor hav bn gnratd from ndognous cortical prcursors. (o) Modl illustrating our conclusion that thr squntial wavs of OLPs wr gnratd from diffrnt parts of th tlncphalic vntricular zon: (i) from Nkx2.1-xprssing prcursors starting at E12.5; (ii) from Gsh2-xprssing LGE-CGE prcursors starting at E15.5; and (iii) from Emx1-xprssing cortical prcursors starting around birth (P). Scal bars: (a,c,,g,i,k), 5 mm; (b,d,f,h,j,l,n), 6 mm. forbrain, clos to thir original sourc, th Nkx2.1-drivd OLPs and OLs wr gradually lost and rplacd by othr populations and wr almost compltly liminatd from th adult forbrain. Ths findings hlp to intgrat th plthora of prvious studis that rport ithr vntral, dorsal or multipl OLP origins Our rsults rais th qustion of whthr th diffrnt OL linags from MGE-AEP, LGE-CGE and cortx ar functionally spcializd. To addrss this, w dvisd a gntic ablation stratgy to liminat ach of th thr OLP populations sparatly with diphthria toxin A fragmnt (DTA; rf. ). W found that whn w killd any on of th populations, th ffct was bnign: adjacnt populations sprad into th vacant trritory, a normal distribution of OLPs was rstord and th mic dvlopd and survivd normally. Thrfor, w hav not, so far, found vidnc of functional htrognity among diffrnt OL linags in th forbrain. Our data dmonstrat that diffrnt rgional subpopulations of OLPs compt for spac in th dvloping brain. This might contribut to th ultimat dmis of th mbryonic MGE- AEP (Nkx2.1) drivd OLP population. RESULTS Th first OLPs ar drivd from th MGE-AEP Prvious immunohistochmical and transplant studis suggstd that thr is at last on sourc of OL linag clls in th vntral forbrain (MGE and AEP) 11,13 17,21,22. To dirctly tst whthr OLPs ar gnratd in th vntral tlncphalon, w gnratd a P1-drivd artificial chromosom (PAC) transgnic mous that xprssd Cr rcombinas undr th transcriptional control of Nkx2.1, which ncods a a b i j c d k g E12.5 E14.5 E16.5 E16.5 Pdgfra/ f h m o 2 homodomain transcription factor xprssd in th MGE, sptum, AEP, proptic ara and othr mor postrior vntral forbrain rgions. Exprssion of th Cr transgn faithfully rcapitulatd th ndognous Nkx2.1 pattrn (Supplmntary Fig. 1 onlin). W crossd th Nkx2.1-Cr mic to th Cr-dpndnt Rosa26R- rportr lin and xamind th mbryonic offspring for th prsnc of Pdgfra-positiv (Pdgfra + ) OLPs that also xprssd grn fluorscnt protin (), indicating that thy must hav originatd from Nkx2.1-xprssing prcursors. In most parts of th tlncphalon, th + Pdgfra + OLP population was only a small fraction of all + clls. In th cortx, th majority of + clls rprsnt GABArgic intrnurons that ar known to originat in th MGE (rfs ). Th first + Pdgfra + OLPs appard in th vntral MGE and AEP at E11.5 E12.5 (Fig. 1a,b) and gradually sprad throughout th tlncphalon in a vntral-to-dorsal mannr. Until E14.5, th grat majority of Pdgfra + OLPs in th vntral tlncphalon as wll as in th crbral cortx coxprssd, indicating that th main sourc of OLs in th mbryonic forbrain lay within th Nkx2.1-xprssing nuropithlium (Fig. 1c,d). Howvr, by E16.5, a nw population of -ngativ ( )Pdgfra + OLPs appard, mainly within th cortical intrmdiat zon (arrowhads, Fig. 1,f). Ths OLPs must hav originatd outsid th MGE-AEP. Subsqunt wavs of OLPs from th LGE-CGE and cortx Anothr rgion of th tlncphalon that xprsss Olig2 and sms likly to gnrat OLs is th LGE. To fat map th LGE, w gnratd PAC transgnics that xprssd Cr undr Gsh2 control. Gsh2 was P2 P P1 Pdgfra/ l n 1 3 Nkx2.1-Cr/ /R26R Emx1-Cr /R26R Emx1-Cr-ER T2 /R26R Emx1 Gsh2 Nkx VOLUME 9 [ NUMBER 2 [ FEBRUARY 26 NATURE NEUROSCIENCE

3 Nkx2.1-Cr/R26R- a b c d P1 Emx1 f g MC CC AC POA LOT POA h i 1 j 1 5 P AC Motor cortx Prcntag contribution 1 5 xprssd strongly in th LGE-CGE but partially ovrlappd with Nkx2.1 in th MGE (Supplmntary Fig. 1). To map th ovrall contribution of th vntral forbrain, w gnratd tripl-transgnic /Nkx2.1-Cr/Rosa26R- mic. In E16.5 tripl transgnics, practically all Pdgfra + OLPS in th tlncphalon, including th latral cortx, wr also + (Fig. 1g,h). This shows that bfor birth, all OLPs in th cortx wr vntrally drivd, migrating first from th MGE-AEP and subsquntly from th LGE, CGE or both. By th day of birth, howvr, OLPs had onc again startd to accumulat in th cortx of th tripl transgnics (arrowhad, Fig. 1i,j), indicating th prsnc of at last on additional sourc of OLPs outsid ithr th MGE-AEP or th LGE-CGE possibly within th cortx itslf. To xamin th fats of ndognous cortical prcursors, w gnratd Emx1-Cr transgnics (Supplmntary Fig. 1). In ths mic, Cr was xprssd strongly in cortical prcursors, and activation of th rportr in Emx1-Cr/Rosa26R- mic was widsprad in th postnatal cortx (Supplmntary Fig. 1). was prsnt throughout th cytoplasm of th clls and vn along axons, so that fibr tracts projcting from th cortx to th vntral forbrain wr labld (arrow, Fig. 1k). Emx1-drivd ( + Pdgfra + ) OLPs first appard in th cortx around birth (Fig. 1k,l) and rapidly incrasd in numbr thraftr. Emx1-drivd OLPs wr nvr obsrvd in th vntral tlncphalon. Our data suggstd that cortical OLPs ar composd of two immigrant populations from th MGE-AEP and th LGE-CGE along with a rsidnt cortical population. To xclud th possibility that th Emx1-drivd OLPs ar vntrally drivd clls that bgin to xprss Emx1 d novo aftr thy migrat into th cortical fild, w gnratd anothr transgnic lin that xprssd th tamoxifn-inducibl form of Cr rcombinas (CrER T2 ) undr Emx1 control (Supplmntary Fig. 1). W inducd Cr activity transintly by administring tamoxifn onc at E9.5, bfor th onst of gliognsis. Bcaus tamoxifn was clard within 48 h, Cr could not hav bn activatd in OLPs that migratd into th cortx from lswhr aftr BE11.5. W analyzd th tamoxifn-tratd mic at P9. Pdgfra xprssion in th CNS is rstrictd to arly OLPs and is downrgulatd as soon as th clls xit th cll cycl and bgin to diffrntiat into OLs 29,3. W thrfor usd an antibody 5 P P1 P3 Adult P P1 P3 Adult LOT Gsh2 Nkx2.1 Sptum P8 P1 P P1 P3 Adult P MC P P1 P3 Adult CC P1 P3 Adult / Figur 2 Th mbryonic Nkx2.1-drivd OL linag is rapidly liminatd during postnatal lif. (a,b) InNkx2.1-Cr/Rosa26R- mic, + (grn), + (rd) doubl-positiv OLPs wr alrady a minority in th P1 cortx (arrows in a), and thir contribution droppd to zro by P8. (c,d) At P1, practically all OLPs in vntral rgions such as th sptum wr drivd from Nkx2.1-xprssing (MGE-AEP drivd) prcursors (, doubl-positiv; yllow nucli in c), but vn hr thy wr rplacd by othr populations ( ) by P8. ( j) Th proportional contributions of th thr diffrnt populations of + OL linag clls ar prsntd as prcntag of th total numbr of + clls in ach rgion (avrag ± s.d.). OLs and OLPs drivd from Nkx2.1-xprssing prcursors wr largly liminatd btwn birth and adulthood from most of th rgions xamind. OLs and OLPs gnratd from cortical prcursors (Emx1 drivd) rmaind within th cortx at all stags. Gsh2- xprssing prcursors gav ris to OLs and OLPs that sprad throughout th tlncphalon. POA, proptic ara; MC, motor cortx; AC, antrior commissur; LOT, latral olfactory tract; CC, corpus callosum. Scal bar, 1 mm. to, which bgins to b xprssd in OLPs slightly latr than Pdgfra but prsists in matur OLs wll aftr Pdgfra is downrgulatd (rf. 31). CrER T2 -mdiatd rcombination was rlativly infficint compard to rgular Cr, so that wll-sparatd -labld clls or clustrs of clls wr obsrvd throughout th cortx. Ths includd + OLs and OLPs, as wll as clls with th morphology of nurons and astrocyts (Fig. 1m,n). Th + + clls comprisd B2% of all + clls and wr obsrvd in all parts of th cortx (arrows, Fig. 1n). W concludd that thy wr formd from ndognous cortical prcursors and not from inwardly migrating clls. Postnatal radication of th Nkx2.1-drivd linag Our data dmonstratd that thr wr thr succssiv wavs of OLP gnration and migration in th mbryonic tlncphalon from th MGE-AEP, th LGE-CGE and cortx, rspctivly (Fig. 1o). To dtrmin how ths thr OLP populations contribut to th postnatal and adult tlncphalon and th xtnt to which thy intrmingl, w analyzd Nkx2.1-Cr/Rosa26R-, /Rosa26R- and Emx1- Cr/Rosa26R- mic from nwborn (P) to young adult (P8) (Fig. 2). Th contribution of Nkx2.1-drivd + OLs and OLPs throughout th forbrain was high at birth in dorsal rgions such as th corpus callosum as wll as in vntral rgions such as th sptum and proptic ara. W wr surprisd to find that by P1, th proportion of Nkx2.1-drivd OLs and OLPs had plummtd to a vry low lvl in th cortx (Fig. 2a,b,g,j). Most rmarkably, th Nkx2.1-drivd population dclind to a vry small fraction of all OL linag clls in most parts of th adult forbrain, vn in vntral rgions clos to thir original sourc (Fig. 2c,d,f,h,i). Thus, th Nkx2.1-xprssing rgion of th vntral forbrain gav ris to migratory OLPs that populatd th mbryonic forbrain including th cortx, but most of ths wr liminatd aftr birth. Th majority of OLPs in most rgions of th postnatal tlncphalon wr gnratd by Gsh2-xprssing prcursors (Fig. 2 j). Th Emx1 prcursor contribution was almost ntirly rstrictd to th cortx throughout lif, although in th adult, a fw Emx1-drivd OLs wr found in th sptum. From P1 to P8, th cortx was populatd by similar proportions of Emx1- and Gsh2-drivd OLs and OLPs (Emx1 contribution dclind from B7% to B3%). NATURE NEUROSCIENCE VOLUME 9 [ NUMBER 2 [ FEBRUARY 26 1

4 a b c d f -DTA -DTA I g h j k -DTA I II II III IV V Figur 3 Dsign and activity of a Cr-inducibl Dta transgn undr transcriptional control. (a) Intron and xon structur of th locus (top) 4xpA DTA 4xpA DTA V I II DTA V -DTA / Th cortical Emx-drivd and vntral tlncphalic Nkx2.1- and Gsh2-drivd OLs and OLPs accountd for narly all OL linag clls in th arly postnatal tlncphalon. Howvr, by adult stags, th total contribution of tlncphalic OLs and OLPs droppd to 6 %, spcially in vntral trritoris such as th proptic ara (Fig. 2f) and in fibr tracts such as th corpus callosum, latral olfactory tract and antrior commissur (Fig. 2h,i,j). Although w cannot xclud th possibility that this might hav bn du to th slctiv downrgulation of th Rosa26 promotr in som matur clls, w bliv that it is mor likly that this rflcts a postrior to antrior migration of clls along fibr tracts. W obsrvd such an influx of clls into th tlncphalon from th dincphalon along axons of th intrnal capsul (data not shown). Functional quivalnc of OLs with diffrnt origins Th prsnc of distinct OL populations within th tlncphalon ld us to qustion whthr th diffrnt OL linags wr functionally spcializd. To addrss this, w dvisd a transgnic approach to liminat OL linag clls according to thir sit of origin, by th targtd xprssion of DTA (rf. 32). A transgnic mous lin was producd using a PAC-basd transgn with th structur -lox- -poly(a)-lox-dta (Fig. 3a). Th transgn was faithfully rgulatd bcaus and wr coxprssd in th vast majority of OL linag clls throughout th CNS (498% coxprssion in th P3 cortx) (Fig. 3b ). Aftr Cr rcombination, should hav bn xcisd along with th poly(a) signal, thus activating DTA and killing th clls. Hncforth, w rfr to this DTA lin simply as -DTA. To tst our cll ablation stratgy, w crossd -DTA to an Olig2- Cr lin that xprssd Cr rcombinas in all OL linag clls, rgardlss of thir mbryonic origin (N.K., D. Rowitch and W. R., unpublishd data). Olig2-Cr/-DTA nonats had nithr + nor + clls anywhr in th brain or spinal cord, confirming fficint xcision i / Olig2-Cr/ -DTA Cm r / and th -DTA transgn bfor (middl) and aftr (bottom) xcision of th -poly(a) casstt. (b ) Coronal sction through th tlncphalon of a -Dta transgnic mous at P3 and high-magnification imags of th cortx from th sam sction. All + OLPs and OLs in th -DTA mous coxprssd, dmonstrating th vracity of transgn xprssion. (f k) Th Dta transgn was activatd by crossing -DTA to an Olig2-Cr mous, xcising th casstt in all Olig2-xprssing prcursors and thir dscndants, hnc prmitting th xprssion of DTA and th highly spcific killing of all + OL linag clls. Th ffctivnss of this tool is dmonstratd by th loss of all + + OL linag clls in th doubl transgnics (compar f h with i k). Th Olig2-Cr/-DTA mic did shortly aftr birth, apparntly from a motor dficit that prvntd thm from fding. Scal bars: (b) 1 mm;(c ) 5 mm; (f,i) 5 mm; (g h, j k) 8 mm. of th coding squncs and th activation of DTA (Fig. 3f k). Ths mbryos appard normal at birth but did within h. W do not yt know th caus of dath but suspct a motor dficit. To ask whthr th rgion-spcific tlncphalic OL populations ar functionally distinct, w gnratd a sris of transgnics that combind -DTA with ithr Emx1-Cr or (Fig. 4). In th prsnc of Rosa26R-lacZ as an indpndnt rportr for Cr activity, clls from th rgion undr study should hav bn positiv for b-galactosidas (b-gal) (that is, labld blu). Thrfor, if rgion-spcific ablation had bn succssful, thr should hav bn an absnc of blu OLs or OLPs whras b-gal + OLs and OLPs that wr gnratd outsid th rgion of intrst should hav survivd. This was illustratd most markdly in whit mattr tracts such as th corpus callosum and th antrior commissur, whr th majority of cll bodis rprsntd OLs and OLPs or othr glia (Fig. 4a d). For xampl, at P12 thr wr many b-gal + (blu) clls in th corpus callosum and antrior commissur of doubl-transgnic /Rosa26R-lacZ mic (Fig. 4a,c) but practically non in /Rosa26R-lacZ/-DTA tripl transgnics (Fig. 4b,d). Nvrthlss, thr smd to b just as many + + clls in th corpus callosum and antrior commissur of ablatd mic (/-DTA) as in thos of nonablatd mic (- DTA) (Fig. 4 h). For xampl, w countd (9. ± 1.5) clls pr mm 3 in th corpus callosum of nonablatd mic (Fig. 4i) compard to (9.1 ± 1.2) 1 4 such clls in /-DTA mic (Fig. 4j). Similarly, w countd (9. ± 1.4) clls pr mm 3 in th antrior commissur of control mic (Fig. 4k)compard to(9. ±.8) 1 4 such clls in /-DTA mic (Fig. 4l). (Data ar prsntd as man ± s.d.) W concludd that th loss of Gsh2-drivd OLs and OLPs was compnsatd for by on or mor populations in th corpus callosum prsumably by Emx1-drivd (cortical) clls, and in th antrior commissur prsumably by Nkx2.1-drivd clls, although thr could also hav bn invasion from furthr afild. Compnsation was also sn in th corpus callosum whn w ablatd th Emx1-drivd OL and OLP population (data not shown). In non of ths xampls was thr any obvious altration in th amount of mylin in adult mic, as visualizd by Sudan black histochmistry (Fig. 4m p). Although th numbr and distribution of + + OLs and OLPs in Emx1-Cr/-DTA, /-DTA and Nkx2.1-Cr/ -DTA mic wr ultimatly indistinguishabl from thos in normal controls, th accumulation of OLPs was noticably dlayd in som rgions (Fig. 5). For xampl, in -DTA/ mic, thr was a slight dlay in th arrival of OLPs in th outr layrs of th cortx (compar Fig. 5a,c), consistnt with our obsrvation that Emx1- drivd OLPs bgan to b gnratd slightly latr than did th Gsh2- drivd OLPs. Dspit th loss of Gsh2-drivd OLPs, th mic rcovrd fully by P1, by which tim thr was a normal numbr 176 VOLUME 9 [ NUMBER 2 [ FEBRUARY 26 NATURE NEUROSCIENCE

5 Figur 4 Gntic ablation of rgion-spcific OL populations rvals functional rdundancy and compnsation among th diffrnt linags. (a d) Gsh2-drivd whit mattr glia (mainly OL-linag clls) wr visualizd in th corpus callosum (a,b) or antrior commissur (c,d) of P12 /Rosa26R-lacZ (a,c) and/ -DTA/Rosa26-R-lacZ (b,d) mic (coronal sctions, arrows). Many blu clls ar visibl in ths tracts in /Rosa26R-lacZ mic but ar missing from th quivalnt structurs of /-DTA/Rosa26-R-lacZ mic, showing that Gsh2-drivd OLs hav bn ffctivly ablatd in th lattr through th action of DTA. ( l) Dspit th spcific loss of Gsh2- drivd OLs, thr was no apparnt rduction in th total complmnt of OLs in ths whit mattr tracts as visualizd by staining (compar with f,and g with h) or staining (compar i with j, andk with l). (m p) Thr was no apparnt rduction in th amount of mylin in adult whit mattr tracts of ablatd /-DTA mic as visualizd by Sudan black histochmistry (compar m with n, o with p). Thus it sms that th loss of Gsh2-drivd OLs was compnsatd for by th invasion of altrnativ OL linags. Similar compnsation was obsrvd whn th Emx1-drivd OL linag was ablatd in Emx1-Cr/-DTA mic (data availabl on rqust). Th mic rsulting from ths crosss survivd, bhavd and rproducd normally until at last P6. Ths xprimnts dmonstrat that th diffrnt rgional OL linags comptd for trritory in th normal dvloping CNS and furthr suggst that th diffrnt OL populations wr functionally quivalnt. Scal bar, 4 mm. and distribution of OLPs in th cortx (Fig. 5b,d). Evn in th olfactory tracts in th vntral forbrain normally populatd solly by Gsh2- drivd OLs thr was complt rplnishmnt of OLs from othr sourcs. Th mic all survivd and rproducd normally, showing no obvious nurological symptoms. Ths and othr data imply that th Gsh2-, Emx1- andnkx2.1-drivd OL linags ar functionally quivalnt, in contrast to th diffrnt nuronal populations that ar gnratd from th sam grminal filds. P3 P15 a b -DTA c d -DTA -DTA -DTA a b c d P12 β-galactosidas Figur 5 Transint dlay in accumulation of OLPs aftr ablation of th LGE- CGE drivd population in /-DTA mic. Coronal forbrain sctions of P3 or P15 mic wr immunolabld for (grn) and (data not shown). + clls wr -xprssing OLPs that had not undrgon Cr rcombination to xcis and activat DTA. (a,b) Postrior, dorsomdial cortx of nonablatd mic. + clls wr mor or lss vnly distributd through th outr cortx of control mic, both at P3 and P15. (c,d) Ablatd mic. Accumulation of OLPs was dlayd in th outr cortx aftr ablation of th LGE-CGE drivd population (compar a and c). Th distribution was normalizd by P15 (compar b and d). Bcaus LGE-CGE drivd OLPs wr absnt from th ablatd mic (c,d), nighboring populations had to mak up th diffrnc. () Orintation of sctions. Scal bars, 3 mm. -DTA -DTA -DTA -DTA f g h P12 i j k l P12 DISCUSSION W xamind th dvlopmntal origins of OLPs in th forbrain by Cr-lox fat mapping in transgnic mic. W discovrd that OLPs ar gnratd in most parts of th tlncphalic vntricular zon including th MGE-AEP, th LGE-CGE and cortx. Thr was a tmporal vntralto-dorsal gradint in production OLPs wr gnratd first from th MGE-AEP at E11.5, thn from th LGE-CGE around E15 and finally from th cortx aftr birth. Th arlist-formd OLPs, from th MGE- AEP, wr latr liminatd during postnatal dvlopmnt. Ths findings hlp rsolv th confusion surrounding prvious rports that mphasizd vntral, dorsal or multipl origins, dpnding on th dvlopmntal stag xamind, th xprimntal approach usd or both Although w hav dfind thr rgion-spcific OLP populations basd on th xprssion of thr rgionally rstrictd transcription factors (Nkx2.1, Gsh2, Emx1), it sms possibl that OLP gnration is not itslf rgionally rstrictd but procds in a Mxican wav from vntral to dorsal aras. This harks back to th arly ida that th ntir nuropithlium undrgos a switch from nurognsis to gliognsis as part of a constitutiv maturation program 33. Howvr, th pionring arly studis wr ncssarily spculativ to a dgr bcaus, with th mthods availabl at that tim, it was difficult to idntify glial clls unambiguously and it was not possibl to dtct long-rang OLP migration or othr dynamic cll bhavior. Our data sm to contradict prvious fat mapping studis in chickquail and chick-mous chimras, which pointd to a purly vntral sourc of cortical OLPs 17. It sms possibl that thr is a ral diffrnc btwn birds and rodnts in this rspct prhaps additional, mor dorsal sourcs of OLPs volvd in mammals to cop with thir incrasd cortical volum. Altrnativly, lat (dorsal) wavs of OLP gnration might hav bn missd in th avian transplant studis. W wr surprisd to find that th arlist-forming population of OL linag clls from Nkx2.1-xprssing prcursors in th MGE-AEP was almost compltly radicatd by adulthood in most parts of th forbrain. This might b partly xplaind by a dilution ffct causd by m n o p Adult mylin Corpus callosum Antrior commissur NATURE NEUROSCIENCE VOLUME 9 [ NUMBER 2 [ FEBRUARY

6 th B2-fold incras in crbral cortical volum btwn E16 and P1 (N.K., unpublishd data). Howvr, this cannot xplain th rmoval of MGE-AEP drivd OLs from th vntral tlncphalon. It is possibl that MGE-AEP drivd OLPs and OLs compt lss ffctivly for survival factors in th postnatal CNS than do thir countrparts, or that thy hav som othr inhrnt disadvantag. Altrnativly, thr might b a stady rat of turnovr and rplacmnt of all OLs during postnatal lif, and th Nkx2.1-drivd OLs might b gradually substitutd by othr linags. This is a plausibl xplanation: w obsrvd in th cours of our fat mapping xprimnts that th postnatal SVZ at th tips of th latral vntricls known to b a sourc of nurons and glia aftr birth 19,2 was drivd from th mbryonic LGE-CGE and cortx, but not from th MGE-AEP (data not shown). Not that not all Nkx2.1-drivd clls wr liminatd in th adult; cortical intrnurons and basal ganglia nurons prsistd long trm. Spcification of OLPs in th vntral nural tub rlis on morphogntic signals, including Sonic hdghog (Shh), from local organizing cntrs. For xampl, Shh is scrtd from th notochord and floor plat at th vntral midlin of th spinal cord 34,35. In th tlncphalon, it is synthsizd by vntral nuropithlial clls in th proptic ara, AEP and MGE 11,36. In both th spinal cord and forbrain, Shh xprssion is rquird for th production of vntrally drivd nurons and OLPs 1,11,22,36 4. Som OLPs still form in th spinal cord and forbrain of Shh / mic 5,22 and thr is clar in vitro vidnc for th xistnc of a Hdghog (Hh)-indpndnt rout to OLP production 8,9. Fibroblast growth factor (FGF) spcifis OLPs in culturs of mbryonic dorsal spinal cord or crbral cortx in th prsnc of cyclopamin, which blocks all hdghog signaling by binding to th Hh corcptor Smoothnd (Smo) 8,9. Th xistnc of an Hh-indpndnt pathway has also bn dmonstratd by th FGF-mdiatd induction of OL linag markrs in culturd Smo / mbryonic stm clls 5.Itis possibl that th postnatal wav of OLPs that w obsrvd within th crbral cortx was triggrd by this Hh-indpndnt, FGF-dpndnt pathway, but it is also possibl that lat activation of Shh or anothr Hh family mmbr within th cortx might hav bn rsponsibl. Diffrnt OLP linags hav bn dscribd in th forbrain that ithr do or do not xprss Pdgfra 1,21. Howvr, thr is currntly no vidnc for a -indpndnt OL linag, so our quantitativ work, which is basd on staining, dscribs th origin of all OLs and OLPs rgardlss of thir linag as dfind by th xprssion of othr markrs. Additional gntic studis will b rquird to rsolv th issu of a Pdgfra-indpndnt linag bcaus Pdgfra OLPs might rprsnt clls that hav bgun to diffrntiat into OLs and hav consquntly downrgulatd Pdgfra, rathr than a distinct typ of prognitor cll pr s. Morphological and biochmical subtyps of mylinating OLs hav also bn dscribd 41 43, but w do not yt know whthr ths ar gnratd qually from all parts of th tlncphalon. In any cas, it sms that th diffrnt populations of OLPs in th forbrain ar functionally quivalnt bcaus whn on population was substitutd by anothr, th mic survivd and bhavd normally. W did not subjct th mic to dtaild bhavioral analysis, so it rmains possibl that thr might b subtl prformanc diffrncs among th various OL populations or that diffrncs will mrg as th mic ag. It is notabl that whn any on OLP population was ablatd, nighboring populations quickly xpandd to fill th availabl spac dmonstrating that OLPs normally compt with on anothr for trritory in th dvloping CNS. What is th natur of this comptition? How do OLPs sns th absnc of thir nighbors and invad th mpty spac? W prviously prsntd vidnc that OLPs in th spinal cord compt for limiting quantitis of PDGF ligand and prolifrat until thir rat of PDGF consumption (by rcptor binding and intrnalization) xactly balancs th rat of supply from othr clls (nurons and astrocyts) 44,45. Bcaus th rat and pattrn of supply is fixd, so is th final numbr and distribution of OLPs, no mattr whr thy originat. It sms likly that similar principls apply in th dvloping forbrain. Th xistnc of both vntral and dorsal OLP origins in th tlncphalon mirrors th situation in th spinal cord. In th cord, thr is a major sourc of OLPs in th pmn domain of th vntral vntricular zon (rfs. 1 4). Ths OLPs bgin to b producd around E12.5 in th mous. Howvr, a minority (B1 15%) of spinal cord OLPs ar gnratd from mor dorsal trritoris that com into play latr, around E16.5 (rfs. 5 7). Th dorsally drivd OLPs ar normally hld in chck by thir vntrally drivd countrparts, bcaus whn th vntral sourc is supprssd in Nkx6.1/Nkx6.2 doubl knockout mic, th dorsal OLPs prolifrat mor than thy usually do and can gnrat up to 3% of th normal numbr of OLPs bfor birth 7. It is concivabl that dorsally drivd OLPs might vn rpopulat th cord compltly, givn nough tim, but th Nkx6-null mic di at birth. Thus, dorsally and vntrally drivd OLPs compt for spac in th spinal cord just as thy do in th forbrain. Undr normal circumstancs, th vntrally drivd population prdominats in th spinal cord, whras th mor dorsally drivd populations prvail in th forbrain. This compnsatory rdundancy might srv to nsur rapid, fail-saf mylination of th complx mammalian brain, a flxibl rspons to dmylinating damag or disas, or both. METHODS Gnration of transgnic mic. PAC transgnic mic xprssing Cr undr control of Nkx2.1, Gsh2 or Emx1 wr gnratd as prviously dscribd 6.Th gnomic DNA fragmnts usd to gnrat ach of th thr mic spannd approximatly 18 kilobas (kb) of gnomic DNA for Nkx2.1 and Emx1, and 11 kb for Gsh2. Dtails of th gnomic PACs ar availabl on rqust. In all thr cass, th codon-improvd Cr rcombinas (icr) with a nuclar localization signal 46 was fusd to th translation initiation codon using a polymras chain raction (PCR) basd approach. This was followd by an SV4 polyadnylation signal. Th coding rgion of Nkx2.9 was rmovd from th gnomic Nkx2.1 PAC by homologous rcombination. To gnrat th -lox--stop-lox-dta mous, w usd a 12-kb NotI fragmnt from a gnomic PAC. This rgion includd 6 kb upstram and 5 kb downstram of. W rplacd th ntir opn rading fram (ORF) with a loxp-flankd ( floxd ) -poly(a) casstt, followd by th DTA fragmnt coding squncs, which ncod an attnuatd G383A vrsion of DTA (gift from Ian Maxwll, Univrsity of Colorado, Dnvr) 47. Th gnomic rgion spanning xons 3 5 was rplacd with th casstt lox--polya 4 -lox-dtafrt-cm r -frt by homologous rcombination in bactria. Th Cm r casstt was thn rmovd by transint activation of Flp rcombinas in bactria, producing alatntdta transgn that xprsss undr transcriptional control (-DTA). In th prsnc of Cr rcombinas, th -poly(a) casstt is rmovd, thus activating th DTA casstt and killing th xprssing clls (furthr dtails availabl on rqust). All PACs wr purchasd from th UK Human Gnom Mapping Projct Rsourc cntr. PAC modification was carrid out in a bactrial systm 48.Cr rportr mic usd in this work wr Rosa26R- (rf. 49) for immunohistochmistry and Rosa26R-lacZ (rf. 5) for nzymatic dtction of b-galactosidas activity. All animal xprimnts wr approvd by th Univrsity Collg London local thical committ and conformd to th UK Animals (Scintific Procdurs) Act In situ hybridization and immunohistochmistry. Tissus wr prpard for in situ hybridization and immunohistochmistry as dscribd 6. Th prob usd to dtct xprssion of th Cr transgn was a 9-bp icr fragmnt spanning th ntir icr ORF. Antibodis usd in this study includ rabbit antibody to at 1:8, dilution (AbCam); guina pig antibody to at 1:2, and rat antibody to Pdgfra at 1:5 dilution (BD Bioscincs). Mylin was staind with Sudan black according to standard protocols. 178 VOLUME 9 [ NUMBER 2 [ FEBRUARY 26 NATURE NEUROSCIENCE

7 Cll counts. For quantification of OLs and OLPs drivd from diffrnt tlncphalic rgions (Fig. 2), w countd and +, + OLs and OLPs in confocal microscop imags takn with a 2 objctiv (at last thr filds from thr or mor 2-mm sctions takn from on or two mic). Th avrag numbr of + clls pr fild rangd from B5 (for xampl, in th proptic ara at P) to ovr 45 (for xampl, in th adult antrior commissur). For ach xprimnt, th data ar prsntd as th proportion of total + clls that wr also +. To quantify th total dnsity of OLs and OLPs in control mic rlativ to mic in which on tlncphalic OLP population had bn ablatd (for xampl, in /-DTA mic) (Fig. 5), w countd + + clls in whit mattr tracts (corpus callosum and antrior commissur) in micrographs takn with a 5 objctiv (fiv or six filds in fiv or six sctions from ach of two mic). Th numbrs wr normalizd and ar quotd in th txt as + clls pr mm 3. Induction of CrER T2 by tamoxifn. Tamoxifn was dissolvd in corn oil by sonication for 3 min. A singl dos of 4 mg (1 ml at4mgml 1 )was administrd to prgnant fmals by oral gavag. At this dos, spontanous abortion was usually not a problm. All pups wr dlivrd by csaran sction at E18.5, rard by a fostr mothr and xamind at P9. Not: Supplmntary information is availabl on th Natur Nuroscinc wbsit. ACKNOWLEDGMENTS W thank our collagus at th Wolfson Institut for Biomdical Rsarch and lswhr for discussion, tchnical hlp and advic spcially M. Fruttigr, U. Dnnhy and R. Tavira-Marqus. W thank I. Maxwll for supplying th DTA plasmid. This work was fundd by th UK Mdical Rsarch Council, th Wllcom Trust Functional Gnomics Initiativ and a Wllcom Trust Priz Studntship (M.F.). COMPETING INTERESTS STATEMENT Th authors dclar that thy hav no compting financial intrsts. Publishd onlin at Rprints and prmissions information is availabl onlin at rprintsandprmissions/ 1. Sun, T., Pringl, N.P., Hardy, A.P., Richardson, W.D. & Smith, H.K. Pax6 influncs th tim and sit of origin of glial prcursors in th vntral nural tub. Mol. Cll. Nurosci. 12, (1998). 2. Lu, Q.R. t al. Common dvlopmntal rquirmnt for Olig function indicats a motor nuron/oligodndrocyt linag connction. Cll 19, 86 (22). 3. Takbayashi, H. t al. Th basic hlix-loop-hlix factor Olig2 is ssntial for th dvlopmnt of motonuron and oligodndrocyt linags. Curr. Biol. 12, (22). 4. Zhou, Q. & Andrson, D.J. Th bhlh transcription factors OLIG2 and OLIG1 coupl nuronal and glial subtyp spcification. Cll 19, (22). 5. Cai, J. t al. Gnration of oligodndrocyt prcursor clls from mous dorsal spinal cord indpndnt of Nkx6 rgulation and Shh signaling. Nuron 45, (). 6. Fogarty, M., Richardson, W.D. & Kssaris, N. A subst of oligodndrocyts gnratd from radial glia in th dorsal spinal cord. Dvlopmnt 132, (). 7. Vallstdt, A., Klos, J.M. & Ericson, J. Multipl dorsovntral origins of oligodndrocyt gnration in th spinal cord and hindbrain. Nuron 45, (). 8. Chandran, S. t al. FGF-dpndnt gnration of oligodndrocyts by a hdghogindpndnt pathway. Dvlopmnt 13, (24). 9. Kssaris, N., Jamn, F., Rubin, L. & Richardson, W.D. Coopration btwn sonic hdghog and fibroblast growth factor/mapk signalling pathways in nocortical prcursors. Dvlopmnt 131, (24). 1. Spassky, N. t al. Sonic hdghog-dpndnt mrgnc of oligodndrocyts in th tlncphalon: vidnc for a sourc of oligodndrocyts in th olfactory bulb that is indpndnt of PDGFR alpha signaling. Dvlopmnt 128, (21). 11. Tkki-Kssaris, N. t al. Hdghog-dpndnt oligodndrocyt linag spcification in th tlncphalon. Dvlopmnt 128, (21). 12. Fuccillo, M., Rallu, M., McMahon, A.P. & Fishll, G. Tmporal rquirmnt for hdghog signaling in vntral tlncphalic pattrning. Dvlopmnt 131, (24). 13. Pringl, N.P. & Richardson, W.D. A singularity of PDGF alpha-rcptor xprssion in th dorsovntral axis of th nural tub may dfin th origin of th oligodndrocyt linag. Dvlopmnt 117, (1993). 14. H, W., Ingraham, C., Rising, L., Godri, S. & Tmpl, S. Multipotnt stm clls from th mous basal forbrain contribut GABArgic nurons and oligodndrocyts to th crbral cortx during mbryognsis. J. Nurosci. 21, (21). 15. Marshall, C.A. & Goldman, J.E. Subpallial Dlx2-xprssing clls giv ris to astrocyts and oligodndrocyts in th crbral cortx and whit mattr. J. Nurosci. 22, (22). 16. Prz Villgas, E.M. t al. Early spcification of oligodndrocyts in th chick mbryonic brain. Dv. Biol. 216, (1999). 17. Olivir, C. t al. Monofocal origin of tlncphalic oligodndrocyts in th chick mbryo: th ntopduncular ara. Dvlopmnt 128, (2). 18. Gorski, J.A. t al. Cortical xcitatory nurons and glia, but not GABArgic nurons, ar producd in th Emx1-xprssing linag. J. Nurosci. 22, (22). 19. Lvison, S.W. & Goldman, J.E. Both oligodndrocyts and astrocyts dvlop from prognitors in th subvntricular zon of postnatal rat forbrain. Nuron 1, (1993). 2. Luskin, M.B. & McDrmott, K. Divrgnt linags for oligodndrocyts and astrocyts originating in th nonatal forbrain subvntricular zon. Glia 11, (1994). 21. Spassky, N. t al. Multipl rstrictd origin of oligodndrocyts. J. Nurosci. 18, (1998). 22. Nry, S., Wichtrl, H. & Fishll, G. Sonic hdghog contributs to oligodndrocyt spcification in th mammalian forbrain. Dvlopmnt 128, (21). 23. Wichtrl, H., Turnbull, D.H., Nry, S., Fishll, G. & Alvarz-Buylla, A. In utro fat mapping rvals distinct migratory pathways and fats of nurons born in th mammalian basal forbrain. Dvlopmnt 128, (21). 24. Ivanova, A. t al. Evidnc for a scond wav of oligodndrognsis in th postnatal crbral cortx of th mous. J. Nurosci. Rs. 73, (23).. Palmitr, R.D. t al. Cll linag ablation in transgnic mic by cll-spcific xprssion of a toxin gn. Cll 5, (1987). 26. Corbin, J.G., Nry, S. & Fishll, G. Tlncphalic clls tak a tangnt: non-radial migration in th mammalian forbrain. Nat. Nurosci. 4, (21). 27. Parnavlas, J.G. Th origin and migration of cortical nurons: nw vistas. Trnds Nurosci. 23, (2). 28. Marin, O. & Rubnstin, J.L. A long, rmarkabl journy: tangntial migration in th tlncphalon. Nat. Rv. Nurosci. 2, (21). 29. Hart, I.K., Richardson, W.D., Hldin, C.-H., Wstrmark, B. & Raff, M.C. PDGF rcptors on clls of th oligodndrocyt-typ-2 astrocyt (O-2A) cll linag. Dvlopmnt 15, (1989). 3. Hall, A., Gis, N.A. & Richardson, W.D. Spinal cord oligodndrocyts dvlop from vntrally drivd prognitor clls that xprss PDGF alpha-rcptors. Dvlopmnt 122, (1996). 31. Stolt, C.C. t al. Trminal diffrntiation of mylin-forming oligodndrocyts dpnds on th transcription factor. Gns Dv. 16, (22). 32. Britman, M.L. t al. Gntic ablation: targtd xprssion of a toxin gn causs microphthalmia in transgnic mic. Scinc 238, (1987). 33. Altman, J. Prolifration and migration of undiffrntiatd prcursor clls in th rat during postnatal gliognsis. Exp. Nurol. 16, (1966). 34. Rolink, H. t al. Floor plat and motor nuron induction by vhh-1, a vrtbrat homolog of hdghog xprssd by th notochord. Cll 76, (1994). 35. Echlard, Y. t al. Sonic hdghog, a mmbr of a family of putativ signaling molculs, is implicatd in th rgulation of CNS polarity. Cll, (1993). 36. Ericson, J. t al. Sonic hdghog inducs th diffrntiation of vntral forbrain nurons: a common signal for vntral pattrning within th nural tub. Cll 81, (1995). 37. Ponct, C. t al. Induction of oligodndrocyt prcursors in th trunk nural tub by vntralizing signals: ffcts of notochord and floor plat grafts, and of sonic hdghog. Mch. Dv. 6, (1996). 38.Pringl, N.P. t al. Dtrmination of nuropithlial cll fat: induction of th oligodndrocyt linag by vntral midlin clls and Sonic hdghog. Dv. Biol. 177, 3 42 (1996). 39. Ericson, J. t al. Pax6 controls prognitor cll idntity and nuronal fat in rspons to gradd Shh signaling. Cll 9, (1997). 4. Orntas, D.M., Hays, J.E., Dyr, K.L. & Millr, R.H. Sonic hdghog signaling is rquird during th apparanc of spinal cord oligodndrocyt prcursors. Dvlopmnt 126, (1999). 41. Bjartmar, C., Hildbrand, C. & Loindr, K. Morphological htrognity of rat oligodndrocyts: lctron microscopic studis on srial sctions. Glia 11, (1994). 42. Butt, A.M., Ibrahim, M., Rug, F.M. & Brry, M. Biochmical subtyps of oligodndrocyt in th antrior mdullary vlum of th rat as rvald by th monoclonal antibody Rip. Glia 14, (1995). 43. Butt, A.M., Ibrahim, M. & Brry, M. Th rlationship btwn dvloping oligodndrocyt units and maturing axons during mylinognsis in th antrior mdullary vlum of nonatal rats. J. Nurocytol. 26, (1997). 44. Calvr, A.R. t al. Oligodndrocyt population dynamics and th rol of PDGF in vivo. Nuron 2, (1998). 45. van Hyningn, P., Calvr, A.R. & Richardson, W.D. Control of prognitor cll numbr by mitogn supply and dmand. Curr. Biol. 11, (21). 46. Shimshk, D.R. t al. Codon-improvd Cr rcombinas (icr) xprssion in th mous. Gnsis 32, (22). 47. Maxwll, F., Maxwll, I.H. & Glod, L.M. Cloning, squnc dtrmination, and xprssion in transfctd clls of th coding squnc for th tox 176 attnuatd diphthria toxin A chain. Mol. Cll. Biol. 7, (1987). 48. L, E.C. t al. AhighlyfficintEschrichia coli-basd chromosom nginring systm adaptd for rcombinognic targting and subcloning of BAC DNA. Gnomics 73, 56 (21). 49. Mao, X., Fujiwara, Y., Chapdlain, A., Yang, H. & Orkin, S.H. Activation of E xprssion by Cr-mdiatd xcision in a nw ROSA26 rportr mous strain. Blood 97, (21). 5. Soriano, P. Gnralizd lacz xprssion with th ROSA26 Cr rportr strain. Nat. Gnt. 21, 7 71 (1999). NATURE NEUROSCIENCE VOLUME 9 [ NUMBER 2 [ FEBRUARY