Terminal osteoblast differentiation, mediated by runx2 and p27 KIP1, is disrupted in osteosarcoma

THE JOURNAL OF CELL BIOLOGY © The Rocke elle Uni e si y P ess $8.00 The Jou nal o Cell Biology, Vol. 167, No. 5, Decembe 6, 2004 925–934 h p://www.jcb.o g/cgi/doi/10.1083/jcb.200409187 JCB: ARTICLE JCB 925 Te minal os eoblas di e en ia ion, media ed by unx2 and p27 KIP1 , is dis up ed in os eosa coma Da id M. Thomas, 1 Sand a A. Johnson, 4 Na alie A. Sims, 4 Melanie K. T i e , 1 John L. Sla in, 1 B ian P. Rubin, 6 Paul Wa ing, 1 G an A. McA hu , 1 Ca l R. Walkley, 1 And ew J. Holloway, 1 Dileepa Diyagama, 1 Jona hon E. G im, 5 B uce E. Clu man, 5 Da id D.L. Bow ell, 1 Jong-Seo Lee, 2 Gab iel M. Gu ie ez, 2 Denise M. Piscopo, 2 Shannon A. Ca y, 3 and Philip W. Hinds 2 1 Ian Po e Founda ion Cen e o Cance Genomics and P edic i e Medicine, and Si Donald and Lady T escow hick Labo a o ies, Pe e MacCallum Cance Cen e , Vic o ia 3002, Melbou ne, Aus alia 2 Depa men o Pa hology, Ha a d Medical School, Bos on, MA 02115 3 Albe Eins ein College o Medicine o Yeshi a Uni e si y, B onx, NY 10461 4 Depa men o Medicine, The Uni e si y o Melbou ne, S . Vincen ’s Hospi al, Vic o ia 3065, Melbou ne, Aus alia 5 Di isions o Clinical Resea ch and Human Biology, F ed Hu chinson Cance Resea ch Cen e , Sea le, WA 98109 6 Depa men o Pa hology, Uni e si y o Washing on, Sea le, WA 98195 he molecula basis o he in e se ela ionship be- ween di e en ia ion and umo igenesis is unknown. The unc ion o unx2, a mas e egula o o os eo- blas di e en ia ion belonging o he un amily o umo supp esso genes, is consis en ly dis up ed in os eosa coma cell lines. Ec opic exp ession o unx2 induces p27 KIP1 , he eby inhibi ing he ac i i y o S-phase cyclin complexes and leading o he dephospho yla ion o he e inoblas- oma umo supp esso p o ein (pRb) and a G1 cell cy- cle a es . Runx2 physically in e ac s wi h he hypophos- pho yla ed o m o pRb, a known coac i a o o unx2, T he eby comple ing a eed- o wa d loop in which p o- g essi e cell cycle exi p omo es inc eased exp ession o he os eoblas pheno ype. Loss o p27 KIP1 pe u bs ansien and e minal cell cycle exi in os eoblas s. Consis en wi h he incompa ibili y o malignan ans o ma ion and pe - manen cell cycle exi , loss o p27 KIP1 exp ession co ela es wi h dedi e en ia ion in high-g ade human os eosa comas. Physiologic coupling o os eoblas di e en ia ion o cell cycle wi hd awal is media ed h ough unx2 and p27 KIP1 , and hese p ocesses a e dis up ed in os eosa coma. In oduc ion Os eosa coma is he mos common bone cance in young adul s, and a leading cause o cance dea h in his age g oup. Loss o di e en ia ion has powe ul p ognos ic signi icance in os eosa coma, wi h well-di e en ia ed umo s being classi ied as low-g ade and dedi e en ia ed umo s usually alling in o he high-g ade ca ego y. In one la ge se ies, 81% o os eosa - comas we e ei he poo ly di e en ia ed o undi e en ia ed (Dahlin, 1957), and a la e ma ke o os eogenic di e en ia- ion, os eocalcin, was unde ec able in  75% o os eosa comas (Hopyan e al., 1999). In i o da a suppo he iew ha os eo- blas di e en ia ion is an agonis ic o oncogenic p ocesses. Cellula immo aliza ion a enua es exp ession o os eoblas pheno ype (Bodine e al., 1996; Feue bach e al., 1997), whe eas os eoblas di e en ia ion is associa ed wi h p og es- si e loss o p oli e a i e capaci y, culmina ing in e minal cell cycle exi (S ein e al., 1996; Aubin, 1998). Al hough i is clea ha di e en ia ion is an i he ical o umo de elopmen in os eosa coma, he molecula basis o he igh ly coupled ela- ionship be ween malignan ans o ma ion and loss o di e - en ia ion emains poo ly unde s ood. The biology o os eoblas di e en ia ion has ecen ly been mapped in conside able de ail. Runx2 ( un - ela ed ansc ip ion ac o 2) is a key ansc ip ional egula o o os eogenesis (Ogawa e al., 1993; Le anon e al., 1994; Ducy e al., 1997, 1999). Mice nullizygous o RUNX2 exhibi a comple e lack o ossi ica ion (Komo i e al., 1997; O o e al., 1997), whe eas he e ozygo es exhibi skele al abno mali ies compa able o cleidoc anial dysplasia (Mundlos e al., 1997). The un amily, o which unx2 belongs, is s ongly linked o human cance (Lund and an Lohuizen, 2002). RUNX1 (AML1) is mu a ed in human leukemia, and mice exp essing loss-o - unc ion unx1 mu an s a e p one o leukemia (Pe y e al., 2002). RUNX3 is subjec o inac i a ing mu a ions o p omo e hype me hyla ion Co espondence o Da id Thomas: [email p o ec ed] G. Gu ie ez’s and P.W. Hinds’s p esen add ess is Dep . o Radia ion Oncol- ogy and Molecula Oncology Resea ch Ins i u e, Tu s-New England Medical Cen e , Bos on, MA 02111. Abb e ia ions used in his pape : BMP, bone mo phogene ic p o ein; MEF, mu ine emb yonic ib oblas ; PCNA, p oli e a ing cell nuclea an igen. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 JCB • VOLUME 167 • NUMBER 5 • 2004926 in gas ic cance s (Li e al., 2002). Runx2 physically in e - ac s wi h he e inoblas oma umo supp esso p o ein (pRb; Thomas e al., 2001), which is mu a ed in up o 60% o os eo- sa comas (Toguchida e al., 1989). Finally, unx2 exp ession a ies wi h cell cycle s a us and may egula e os eoblas p oli - e a ion by unknown mechanisms (P a ap e al., 2003). In his s udy, we in es iga ed he e ec s o os eogenic di e en ia ion on p oli e a ion and g ow h a es , and hei dis- up ion in os eosa comas. Consis en wi h a ole in supp ession o p oli e a ion, unx2 p o ein was absen o non unc ional in six ou o se en os eosa coma cell lines. Bo h spon aneous and induced os eoblas di e en ia ion a e associa ed wi h inc eased p27 KIP1 mRNA and p o ein exp ession. Ec opic exp ession o unx2 induced an Rb- and p27 KIP1 -dependen g ow h a es . This was due in pa o inc eased exp ession o p27 KIP1 p o ein, which inhibi ed S-phase Cdk complexes and he dephospho y- la ion o pRb. In e es ingly, unx2 is shown o in e ac p e e - en ially wi h he hypophospho yla ed o m o pRb, a known coac i a o o unx2. Al hough p27 KIP1 exp ession is associa ed wi h os eoblas di e en ia ion, loss o p27 KIP1 had only a mino e ec on os eoblas di e en ia ion in i o and in i o. No a- bly, he i e e sibili y o bo h he os eogenic pheno ype and e minal cell cycle exi in i o is dependen on exp ession o p27 KIP1 . Immunohis ochemical analysis o human os eosa co- mas con i med ha exp ession o p27 KIP1 was los as umo s los e idence o os eogenic di e en ia ion. Toge he , hese da a sugges ha unx2 es ablishes a e minally di e en ia ed s a e h ough Rb- and p27 KIP1 -dependen mechanisms, and ha hese p ocesses a e dis up ed in os eosa comas. Resul s Cha ac e iza ion o he os eoblas pheno ype in a panel o os eosa coma cell lines We i s used ansc ip ional p o iling o objec i ely cha ac e - ize he di e en ia ion s a e o a panel o os eosa coma cell lines (SAOS2, MG63, B143, HOS, SJSA, and G292) ela i e o an os eoblas -like e e ence. The e e ence consis ed o p ima y s omal s em cells in which exp ession o ma ke s o he ma- u e os eoblas pheno ype was induced by cul u e in he p es- ence o asco bic acid, dexame hasone, and ino ganic phospha e (G on hos e al., 2003). These ma ke s include unx2, os e ix, os eocalcin, and he abili y o mine alize in i o. Consis en wi h a ans o med s a e, se e al pu a i e oncogenes, including FOS and cyclins A1, B2, E1, and D1 (Fig. 1 A), we e com- monly o e exp essed in os eosa coma cell lines, whe eas he Cdk inhibi o s p16 INK4A and p57 KIP2 we e ela i ely unde ex- p essed. Shown in Fig. 1 A is he exp ession pa e n o 16 bone- ela ed genes selec ed om a p e iously published mi- c oa ay s udy on he os eoblas pheno ype (Balin e al., 2003). Twel e genes, including he key os eoblas ansc ip ion ac o , RUNX2 , we e unde exp essed ac oss all os eosa coma cell lines ela i e o ou os eoblas ic e e ence. Os eocalcin, a la e and speci ic os eoblas ma ke no ep esen ed on ou a ays, was unde ec able by RT-PCR in he os eosa coma lines (no depic ed). O e all, he a e age median exp ession o 16 os eo- blas - ela ed genes in he os eosa coma cell lines was educed o 38  8% o he os eoblas ic e e ence. The pa e n o gene exp ession obse ed in os eosa co- mas sugges s a loss o educ ion in ac i i y o unx2, a key an- sc ip ional egula o o os eoblas di e en ia ion. To assess ac- i i y o he unx2 pa hway in he os eosa coma cell lines, we measu ed endogenous unx2 ansc ip ional ac i i y using an os eoblas -speci ic p omo e –luci e ase cons uc , 6ose2-luc (Ducy and Ka sen y, 1995), consis ing o six andem epea s o he os eocalcin unx2 binding si e. A mu an con ol con aining a dinucleo ide subs i u ion in he unx2 binding si es was used, abolishing unx2 binding and ac i i y (Ducy and Ka sen y, 1995). The a io o wild- ype o mu an epo e ac i i y he e- o e e lec s endogenous unx2 unc ion, and i was ma kedly educed in all se en os eosa coma cell lines, o le els compa a- ble o ha obse ed in ib oblas s (CCL-7625; Fig. 1 B). A pos- i i e con ol de i ed om an os eogenic os eoma (CCL-7672) demons a ed unx2 ac i i y 80- old highe han in CCL-7625 cells. Simila esul s we e obse ed wi h na i e os eopon in and os eocalcin p omo e –luci e ase cons uc s (unpublished da a). The e ec o ec opically exp essing he os eoblas -spe- ci ic MASN iso o m o unx2 (he ea e unx2) on epo e ac- i i y was s udied (Fig. 1 C). Ec opic exp ession o unx2 in- c eased epo e ac i i y 10–60- old in some os eosa coma cell lines, sugges ing ha endogenous le els o unx2 we e limi ing o ansc ip ional ac i i y in hese cell lines. Lack o co ela ion be ween unx2 p o ein le els and ansc ip ional ac i i y The e was li le o no co ela ion be ween unx2 p o ein le els and ansc ip ional ac i i y. Immunoblo analysis o endoge- nous unx2 p o ein demons a ed low le els in i e ou o se en cell lines (Fig. 1 D, op and middle). The uppe band ep esen s he MRIPV iso o m o unx2, whe eas he as e mig a ing band in SAOS2 ep esen s he os eoblas -speci ic MASN iso- o m (unpublished da a). The MRIPV iso o m s ongly induces AP ac i i y bu no os eocalcin, whe eas he MASN iso o m mo e po en ly ac i a es os eocalcin bu no AP (Ha ada e al., 1999). Consis en wi h he esul s o he ansc ip ional assays, no unx2 p o ein was de ec able by Wes e n blo in G292 cells, which showed he g ea es induc ion o ansc ip ion by ec opic unx2 (Fig. 1 D). In con as , abundan endogenous unx2 p o- ein was p esen in SAOS2 cells. The s iking con as be ween p o ein le els and in insic ac i i y o unx2 in SAOS2 cells is highligh ed by compa ison wi h p o ein le els in he os eoblas- ic con ol, al hough os eoblas gene exp ession is 5.6  2.9– old highe in he e e ence han in SAOS2 cells. We ha e con- i med ha no mu a ions exis in he genomic sequences o unx2 in SAOS2 cells o any o he cell line in his s udy. Thus, some os eosa comas, such as G292, appea o lack unc ional unx2, whe eas o he s, such as SAOS2, appea unable o ac i- a e ansc ip ion o unx2-dependen genes e en when ec opic unx2 is supplied. In e es ingly, we obse ed ha all lines e- sponsi e o ec opic unx2 exp ess pRb, whe eas hose ha ail o espond exp ess no o low le els o his known unx2 coac i- a o . Fu he mo e, some pRb-posi i e cell lines show only limi ed ac i a ion in esponse o unx2 exp ession, sugges ing on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 DISRUPTION OF DIFFERENTIATION IN OSTEOSARCOMA • THOMAS ET AL. 927 ha addi ional ac o s in luence he ac i i y o unx2 in os eo- sa coma cell lines (Fig. 1 D). Toge he , hese da a con i m ga he ing e idence ha unx2 ac i i y is c i ically dependen on co ac o s o pos ansla ional modi ica ions, and ha oncogenic ans o ma ion esul s in consis en dys egula ion o unx2 ac- i i y by mul iple mechanisms. Runx2 inhibi s cell g ow h h ough p27 KIP1 and pRb I appea s ha no mal unx2 unc ion is incompa ible wi h ma- lignan ans o ma ion o os eoblas ic cells. To de e mine why, we examined he e ec o eexp ession o unx2 in G292 cells. As shown in Fig. 1 (B and C), in G292 cells unx2 le els ap- pea ed o be a e limi ing o ansc ip ional ac i i y. This sug- ges s ha he molecula appa a us o ull unx2 ac i i y, includ- ing any po en ial umo supp esso unc ions, exis s in G292 cells. Consis en wi h his idea, ec opic exp ession o unx2 sup- p essed cell g ow h (Fig. 2 A). In con as , his e ec was no seen in SAOS2 cells, in which o ced exp ession o unx2 had li le e ec on ansc ip ional ac i i y. Because SAOS2 lacks unc ional pRb, whe eas G292 has wild- ype pRb, we hypo he- sized ha he lack o pRb may accoun o he inabili y o unx2 o educe he p oli e a i e capaci y o hese cells. Indeed, when o e exp essed in wild- ype and RB  /  3T3 cell lines, unx2 sup- p essed colony numbe s o 3T3 cells by 60–90%, an e ec de- penden on pRb (unpublished da a). To s udy his e ec u he , we used wo unx2 cons uc s con aining ansac i a ion do- main mu a ions. One o hese (27ala) lacks ansc ip ional ac i - i y, whe eas he o he (3ala) possesses wild- ype ac i i y (Thi una ukka asu e al., 1998). In oduc ion o hese con- s uc s in o 3T3 cells showed ha he colony supp ession ac i - i y o unx2 is due o a G1 cell cycle a es dependen on an- sc ip ional ac i i y and pRb (Fig. 2 B). Runx2 was ec opically exp essed in p ima y human ib oblas s (CCL-211 and IMR90) and os eosa coma cell lines (U2OS and SAOS2), using adeno- i al ec o s. As expec ed, unx2 inhibi ed he S-phase ac ion o ib oblas ic bu no os eosa coma cells. Ini ial s udies o cell cycle p o ein exp ession in hese ans ec ed cells e ealed a speci ic induc ion o p27 KIP1 p o ein bu no e ec on p21 CIP1 (Fig. 2 C). Coimmunop ecipi a ion om CCL-211 cells showed ha p27 KIP1 was s ongly associa ed wi h cyclin A and Cdk2 (Fig. 2 D), supp essed in i o kinase ac i i y o cyclin A–Cdk2 complexes (Fig. 2 E), and was accompanied by dephospho yla- ion o endogenous pRb (Fig. 2 E). We ha e shown p e iously ha pRb binds and coac i a es unx2 (Thomas e al., 2001). We hypo hesized ha a eed- o wa d loop, in eg a ing p og essi e cell cycle wi hd awal and di e en ia ion, would be comple ed i unx2 speci ically in e ac ed wi h he hypophospho yla ed o m o pRb. This is he case (Fig. 2 F). Collec i ely, hese da a a e consis en wi h he ansc ip ional induc ion o g ow h a es by unx2 h ough an Rb- and p27 KIP1 -dependen mechanism ha is ein o ced by coac i a ion o unx2 by di ec in e ac ions wi h he hypophospho yla ed o m o pRb. A ole o p27 KIP1 in os eogenic di e en ia ion The da a desc ibed abo e sugges a ole o p27 KIP1 and pRb in media ing unx2-dependen p oli e a ion a es . We ha e p e i- ously epo ed a ole o pRb in unx2-dependen exp ession o di e en ia ion- ela ed genes, and so wished o de e mine he physiologic signi icance o p27 KIP1 in os eoblas di e en ia ion. As epo ed p e iously (D issi e al., 1999), os eoblas di e en- ia ion in i o is associa ed wi h inc eased exp ession o p27 KIP1 (Fig. 3 A). Bone mo phogene ic p o eins (BMPs) a e powe ul os eoinduc i e agen s whose e ec s a e media ed by unx2 (Tsuji e al., 1998; Go i e al., 1999). T ea men o mu- ine emb yonic ib oblas s (MEFs) wi h a syn he ic BMP4/7 usion p o ein induced os eoblas di e en ia ion (and cell cycle Figu e 1. Runx2-dependen os eogenic di e - en ia ion is dis up ed in os eosa coma cell lines. (A) Gene exp ession a ays we e pe o med using RNA ex ac ed om con luen cul u es o SAOS2, MG63, HOS, B143, SJSA, and G292 cell lines, no malized o e e ence RNA as desc ibed in Ma e ials and me hods sec ion. Da a p esen ed a e he median log- ans o med da a o six cell lines. (B) Cells we e ans ec ed wi h 6ose2-luc (o wi h 6ose2mu -luc) and cy- omegalo i us (CMV)–gal plasmids. A e 24 h, luci e ase ac i i y was measu ed and no - malized o -galac osidase. The a io o he ac i i y o 6ose2-luc o 6ose2mu -luc ac i i y is shown. Os , os eogenic os eoma cell line CCL- 7672. Da a shown a e means  SEM. (C) Cells we e ans ec ed wi h 6ose2-luc and CMV-gal plasmids, wi h o wi hou a unx2 exp ession ec o . A e 24 h, luci e ase ac i i y was measu ed and no malized o ans ec ion e iciency wi h -galac osidase. The a io o luci e ase ac i i y in he p esence o absence o unx2 is shown. Da a shown a e means  SEM. (D) Wes e n blo o unx2 and pRb in nuclea ex ac s om he indica ed cell lines. HOb, human p ima y os eoblas . on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 JCB • VOLUME 167 • NUMBER 5 • 2004928 a es ), concomi an wi h exp ession o unx2 and p27 KIP1 mRNA (Fig. 3 B). Simila ly, BMP2 induced a G1 cell cycle a - es in MEFs ha was dependen in pa on he p esence o bo h pRb and p27 KIP1 (Fig. 4 A). This is consis en wi h a ole o unx2 in inhibi ion o cell p oli e a ion and induc ion o di e - en ia ion by BMPs. To con i m hese obse a ions in a p e- os eoblas cell model, we used siRNA o knockdown p27 KIP1 in MC3T3E1 cells (Fig. 4 B). The le el o knockdown achie ed was  75%, which we conside signi ican because p27 KIP1 is belie ed o ac as a haploinsu icien umo supp esso gene (Fe o e al., 1998). Consis en wi h obse a ions ha unx2-null os eoblas s ha e inc eased a es o p oli e a ion (P a ap e al., 2003), we obse ed an inc eased a e o p oli e a ion in MC3T3E1 cells in which p27 KIP1 was educed. BMP2 ea - men o MC3T3E1 cells exp essing a con ol siRNA ec o e- sul ed in a 42% educ ion in S-phase cells, compa ed wi h a 17% educ ion in cells exp essing he p27 KIP1 siRNA (Fig. 4 C). Fu he mo e, as obse ed in G292 cells and 3T3 ib oblas s, ec opic exp ession o unx2 supp essed g ow h o MC3T3E1 cells, and his e ec was abolished in cells exp essing siRNA o p27 KIP1 (Fig. 4 D; Chi squa e P  0.001). These da a sug- ges ha p27 KIP1 , like pRb, plays a ole in egula ing basal a es o p oli e a ion in p eos eoblas s and con ibu es o he g ow h a es associa ed wi h os eoblas ic di e en ia ion. Nex , we wished o de e mine whe he p27 KIP1 is equi ed o he os eoblas pheno ype. Loss o p27 KIP1 pa ially a enu- a ed BMP2-induced AP ac i i y (Fig. 5 A), and bo h basal and BMP2-induced exp ession o os eocalcin, os eopon in, and ype I collagen mRNA we e educed (bu no abolished by) he absence o p27 KIP1 (Fig. 5 B). To de e mine he ne e ec o loss o p27 KIP1 in i o, he long bones o mu ine wild- ype and p27 KIP1  li e ma es we e analyzed by his omo phome y. The e was a mino e ec o loss o p27 KIP1 on he o ma ion o Figu e 2. Runx2 induces a g ow h a es h ough induc ion o p27KIP1. (A) E ec o ec opic exp ession o unx2 on colony supp ession assay. (B) Re o i al ec o s exp essing unx2 27ala o unx2 3ala we e used o in ec 3T3 o RB  /  3T3 cells, ollowed by selec ion o 3 d in 2 g/ml pu omycin. Cell cycle p o ile was de e mined by low cy ome y. (C) IMR90, CCL-211, SAOS2, and U2OS cells we e in ec ed wi h adeno i al cons uc s exp essing FLAG- agged unx2. Wes e n blo o unx2, p27KIP1, and p21CIP1. The pe - cen age o cells in S-phase, de i ed om pa allel cul u es subjec ed o DNA analysis by low cy ome y, is indica ed (bo om). (D) CCL-211 cells we e in ec ed wi h adeno i al cons uc s exp essing unx2. Cyclin A immunop e- cipi a es we e subjec ed o Wes e n blo o p27KIP1, Cdk2, and cyclin A. The bo om panel is di ec ly blo ed o p27KIP1. As e isks indica e Ig hea y chain bands. (E) Cyclin A immunop ecipi a es assayed o kinase ac i i y in he p esence o unx2. CCL-211 cells we e ea ed as in C. Di ec Wes e n blo o Rb, cyclin A, cyclin E, Cdk2, p27KIP1, p21CIP1, unx2, and GFP o demon- s a e equal i e s o i us in each cul u e. (F) Runx2 binds he hypophospho - y la ed o m o pRb. COS-7 cells we e ans ec ed wi h ec o , pRb, and HA- agged pRb. A pulldown was pe o med using GST- unx2. Inpu is 5% o ha used o he pulldown. ppRb, phospho yla ed species o pRb. Black lines indica e ha in e ening lanes ha e been spliced ou . Figu e 3. Os eogenic di e en ia ion in i o is associa ed wi h induc ion o p27KIP1 mRNA and p o ein. (A) MC3T3E1 cells we e cul u ed in di e en ia ion media con aining 50 g/ml asco bic acid and 2 mM -glyce ophospha e. ( op) Wes e n blo o p27KIP1 p o ein a e 2–8 d in di e en ia ion media. (bo om) Aliza in ed s aining o mine alized cul u es o e 3–12 d unde iden ical condi ions. (B) Mu ine emb yonic ib oblas s (MEFs) cul u ed in he p esence o a BMP4/7 usion p o ein o 14 d. ( op) Induc ion o AP ac i i y. Da a shown a e means  SEM. (bo om le ) Aliza in ed s aining o mine - aliza ion. (bo om igh ) RT-PCR o unx2, p27KIP1, p21CIP1, and glyce alde- hyde phospha e dehyd ogenase (GAPDH). Fold changes in gene exp ession ela i e o GAPDH a e gi en below each panel. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 DISRUPTION OF DIFFERENTIATION IN OSTEOSARCOMA • THOMAS ET AL. 929 unmine alized bone (os eoid), bo h as a unc ion o o al bone olume and as measu ed by os eoid hickness (Fig. 5 C). This e ec was no a esul o accele a ed mine aliza ion caused by loss o p27 KIP1 , because he e was no e idence o al e ed min- e al apposi ion a e as measu ed di ec ly by dual calcein la- beling. Os eoblas and os eoclas numbe s we e unchanged (unpublished da a). Because os eoclas s do no eso b unmine - alized os eoid (Chambe s e al., 1985), any de ec in os eoclas unc ion is unlikely o accoun o he educ ion in os eoid. These da a a e consis en wi h a e y mino e ec o loss o p27 KIP1 on os eoblas di e en ia ion and unc ion. p27 KIP1 is needed o e minal cell cycle eg ess p27 KIP1 plays a key ole in e minal cell cycle exi in os eosa - coma cells (Alexande and Hinds, 2001). Because e minal di e en ia ion is associa ed wi h pe manen cell cycle wi h- d awal, we es ed whe he MEF cul u es lacking p27 KIP1 could een e he cell cycle a e cul u e unde p olonged di e en i- a ing condi ions. The e was no p io di e ence in cell cycle p o ile o li e ma e wild- ype o p27 KIP1  /  p econ luen ea ly passage MEFs (unpublished da a). Howe e , al hough wild- ype MEFs pos di e en ia ion g ew poo ly a e passage, MEFs lacking p27 KIP1 p oli e a ed obus ly (Fig. 6 A). This sugges s a ole o p27 KIP1 in media ing he i e e sibili y o he pos con luen s a e. Howe e , wild- ype undi e en ia ed MEFs also ailed o een e he cell cycle, p obably because p27 KIP1 also accumula ed in un ea ed cul u es upon con lu- ence (Hi ano e al., 2001; and unpublished da a). Consis en wi h an addi ional e ec o BMP2 upon p27KIP1, he deg ee o g ow h inhibi ion a e passage was g ea e in cul u es ea ed wi h BMP2 (Fig. 6 A). Mo eo e , e en pos di e en ia ed MEFs lacking p27KIP1 demons a ed a educ ion in cell g ow h when passaged a e BMP2 ea men , indica ing ha p27KIP1 con ibu es o (bu is no solely esponsible o ) e minal g ow h a es . We ha e no obse ed compensa o y inc eases in p21CIP1 o p57KIP2 in p27KIP1-null cul u es (unpublished da a), and he mechanisms accoun ing o he esidual e ec s a e no known. Reen y in o he cell cycle o BMP2- ea ed cul u es was associa ed wi h loss o di e en ia ion. Compa ed wi h wild- ype cul u es, BMP2- ea ed p27KIP1-null cul u es apidly los exp ession o os eocalcin wi hin 2 d o passage (Fig. 6 B). Addi ionally, he numbe o AP-posi i e cells was 20- old g ea e in wild- ype compa ed wi h p27KIP1  /  cul u es (4.8  1.8% [n  263, 10 high-powe ields], compa ed wi h 0.2  0.1% AP-posi i e cells [n  2039]). This sugges s ha e mi- nal cell cycle exi , in his case dependen on p27KIP1, is equi ed o main enance o he di e en ia ed s a e. Pe manen cell cycle wi hd awal is a ea u e o bo h se- nescence and e minal di e en ia ion (Golds ein, 1990; Selle s e al., 1998). Pos di e en ia ed wild- ype MEFs assumed a bi- nuclea ed, enla ged, la ened mo phology eminiscen o epli- ca i e senescence. Whe he di e en ia ed o no , signi ican numbe s o pos con luen wild- ype MEFs s ained o senes- cence-associa ed -galac osidase ac i i y (Dim i e al., 1995). This e ec was g ea e in cul u es ha had been ea ed wi h BMP2 (Fig. 6 D). In con as , MEFs lacking p27KIP1 did no s ain o senescence-associa ed -galac osidase and mo pho- logically esembled undi e en ia ed cul u es. Al hough he e - ec s o bo h BMP2 and p27KIP1 we e independen ly s a is i- cally signi ican , he in e ac ion be ween hese ac o s ailed o each signi icance, pe haps because o he powe ul e ec o con luence on accumula ion o p27KIP1 in bo h ea ed and con- ol cul u es. Toge he , hese da a sugges ha cul u e condi- ions equi ed o di e en ia ion o MEFs (including bo h BMP2 ea men and p olonged con luence), as well as exp es- sion o p27KIP1, con ibu e o he en y o MEFs in o a senes- cence-like s a e. Figu e 4. G ow h a es due o exp ession o unx2 o ea men wi h BMP2 is educed by knockdown o p27KIP1. (A) P ima y MEFs o he indi- ca ed geno ypes we e ea ed wi h 100 ng/ml BMP2 o 48 h ollowed by low cy ome ic analysis o DNA con en . Da a shown a e he change in G1 ac ion due o ea men . This expe imen was epea ed wice wi h simila esul s. (B) Wes e n blo o p27KIP1 in MC3T3E1 cells in ec ed wi h a e o i us exp essing ei he con ol siRNA o siRNA o p27KIP1. (C) DNA analysis using low cy ome y o cul u es o cells de i ed as desc ibed in B. (D) Colony supp ession assay o cells as desc ibed ans ec ed wi h emp y ec o o unx2. Each expe imen was pe o med in iplica e. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 JCB • VOLUME 167 • NUMBER 5 • 2004930 Exp ession o p27 KIP1 is los in os eosa coma We inally wished o de e mine whe he hese obse a ions ha e ele ance o human os eosa coma. p27KIP1 exp ession appea s o be key o cell cycle wi hd awal and e minal di e en ia ion in os eoblas s, and in eg a es he unc ions o BMPs, pRb, and unx2 in hese p ocesses. Rega dless o he na u e o he de ec in he pRb– unx2 pa hway in os eosa coma cells, he ne e ec will be loss o g ow h es ain due o diminished exp ession o p27KIP1. Consis en wi h his, we ound negligible exp ession o p27KIP1 p o ein in high-g ade os eosa coma cells, al hough p27KIP1 was clea ly seen in os eoclas s as epo ed p e iously (Okahashi e al., 2001; Fig. 7, A and D), co ela ing in e sely wi h exp ession o p oli e a ing cell nuclea an igen (PCNA; Fig. 7, D and F, a ows). These high-g ade os eosa comas dem- ons a ed equen mi o ic igu es and li le di e en ia ion, as e idenced by os eoid p oduc ion and os eocalcin exp ession (Fig. 7, B and E). High-g ade umo cells exp essed high le els o PCNA, consis en wi h a high S-phase ac ion (Fig. 7, C and F). In con as , in lowe g ade umo s wi h mine alizing os eoid and lowe cellula i y wi h mo e no mal os eoblas ic mo phol- ogy, exp ession o bo h p27KIP1 and os eocalcin is e iden , espe- cially in e minally di e en ia ed (PCNA nega i e) os eocy es embedded wi hin bone (Fig. 7, G and H, a ows). C i ically, he e was a signi ican ela ionship be ween exp ession o p27KIP1 p o ein and os eoblas di e en ia ion sco ed by os eoid p oduc ion in a panel o 100 os eosa comas (Fig. 7 J, P  0.05). This e ec was independen o p oli e a i e a e, because he e was no signi ican ela ionship be ween PCNA exp ession and p27KIP1. These da a suppo he iew ha he loss o di e en ia- ion o os eosa comas, which con eys ad e se p ognos ic signi - icance, is associa ed wi h loss o exp ession o p27KIP1. Figu e 5. Role o p27 KIP1 in os eoblas unc ion. (A) MEFs o he indica ed geno ypes we e di - e en ia ed in he p esence o 100 ng/ml BMP2, asco bic acid, and -glyce ophos- pha e o 7 d and analyzed o AP ac i i y. Da a shown a e means  SEM o old change ela i e o un ea ed wild- ype con ols (n  4 expe imen s using independen ly de i ed li e - ma e-ma ched cul u es, each in iplica e). Ge- no ype e ec signi ican by analysis o a i- ance (ANOVA; P  0.01). (B) RT-qPCR analysis o gene exp ession in MEFs o he in- dica ed geno ypes, no malized o ARPPo. Da a shown a e means  SEM. Geno ype e ec signi ican o BMP2 induc ion o os eocalcin (P  0.01) and ype I collagen (P  0.05, ANOVA), bu no os eopon in. (C) His omo pho- me ic analysis o long bones in mice be ween 8 and 12 wk old o he indica ed geno ypes. Da a shown a e means  SEM. Bold- aced da a a e signi ican ly di e en om wild- ype li e ma es (P  0.05). Figu e 6. p27KIP1 is equi ed o e minal g ow h a es in i o. MEFs we e cul u ed in he p esence o absence o 100 ng/ml BMP2, asco bic acid, and -glyce ophospha e o 14 d a e con luence. Cells we e hen passaged. (A) 5  104 cells we e g own unde s anda d cul u e condi ions o 3 d, and hen coun ed. Da a shown a e means  SEM. (B) RT-qPCR analysis o os eocalcin gene exp ession in MEFs o he indica ed geno ypes 2 d a e passage. Da a shown a e means  SEM o cycle numbe (CT) no malized o ARPPo and exp ession be o e passage. The in e ac ion be ween BMP2 and geno ype was signi ican (P  0.01, ANOVA). (C) Pho omic og aph o senescence-associa ed -galac osidase–s ained cul u es. (i) Wild- ype un ea ed cul u es; (ii) p27KIP1  /  un ea ed cul u es; (iii) wild- ype di - e en ia ed cul u es; and (i ) p27KIP1  /  di e - en ia ed cul u es. (D) Cells we e s ained o senescence-associa ed (SA) -galac osidase ac i i y a e passage and coun ed (200 cells in iplica e cul u es). Da a shown a e means  SEM. The e ec o BMP2 and geno ype was signi ican (P  0.01), al hough he e was no in e ac ion by ANOVA. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 DISRUPTION OF DIFFERENTIATION IN OSTEOSARCOMA • THOMAS ET AL. 931 Discussion The in e se ela ionship be ween p oli e a ion and di e en ia- ion in os eoblas s has been ca e ully documen ed o many yea s, al hough he mechanisms ha e no been delinea ed. These obse a ions ha e led o he p oposi ion ha ull exp es- sion o he os eoblas pheno ype is necessa ily associa ed wi h e minal cell cycle exi (S ein e al., 1996; Aubin, 1998). In ac- co d wi h hese obse a ions, os eoblas s lacking unc ional unx2 appea o lose a g ow h es ain (P a ap e al., 2003). Ou da a p o ide a mechanis ic basis o hese obse a ions (Fig. 8). Os eogenic di e en ia ion in cul u e imposes a g ow h es ain and, e en ually, a e minal cell cycle exi esembling senescence, h ough unx2-dependen induc ion o p27KIP1. We and o he s show ha os eogenic di e en ia ion in i o is asso- cia ed wi h inc eased exp ession o p27KIP1 (D issi e al., 1999). This leads o a pRb-dependen g ow h a es h ough inhibi ion o S-phase cyclin complexes. We ha e shown p e iously ha in e ac ions be ween unx2 and pRb enhance unx2-dependen ansc ip ional ac i i y (Thomas e al., 2001). Because i is he hypophospho yla ed o m o pRb ha binds unx2, he induc- ion o p27KIP1 will enhance he ansac i a ion o unx2 by pRb, leading o p og essi e g ow h a es and exp ession o he ma u e os eoblas pheno ype (Fig. 8). I is likely ha loss o unc ion o any componen o his eed- o wa d loop will dis- up bo h di e en ia ion and a es ain on cell g ow h. Al- hough mu a ions ha e been documen ed in pRb in os eosa - coma, he molecula e en s ha a ec unx2 unc ion in cell lines in which pRb is no a ec ed emain unknown. We obse ed a e y mino de ec in os eoid syn hesis in p27KIP1  /  mice, accompanied by de ec s in BMP2-induced di - e en ia ion in i o, consis en wi h appa en ly no mal skele al pa e ning in mice nullizygous o p27KIP1 (Fe o e al., 1996; Kiyokawa e al., 1996; Nakayama e al., 1996). I is possible ha unc ional edundancy allows compensa ion o loss o p27KIP1, and consis en wi h his hypo hesis, BMP2 induced an a enua ed g ow h a es in p27KIP1-null ib oblas s. Al hough he mecha- nisms by which BMP2 induces a g ow h a es in he absence o p27KIP1 a e no known, BMP2 and unx2 ha e bo h o e lapping and dis inc e ec s o p omo e os eoblas di e en ia ion. In addi- ion o inducing p27KIP1, BMP2 has ecen ly been epo ed o in- hibi Cdk6 le els by a SMAD-dependen mechanism (Ogasa- wa a e al., 2004). The induc ion o p27KIP1 leads o mo e han a simple p o- li e a i e a es . Al hough he p oli e a ion o ea ly passage MEFs was no s ikingly a ec ed by loss o p27KIP1 (Fe o e al., 1996; Kiyokawa e al., 1996; Nakayama e al., 1996; and un- published da a), we ound ha p27KIP1 was equi ed o main ain a g ow h-a es ed s a e in di e en ia ed cells. p27KIP1 has been sugges ed p e iously o be a pa o he no mal ime ha de e - mines he cessa ion o p oli e a ion and commi men o di e - en ia ion o oligodend ocy e p ecu so s (Casaccia-Bonne il e al., 1997; Du and e al., 1998). The D osophila melanogas e p27KIP1 homologue, dacapo, ini ia es e minal cessa ion o cell di ision and di e en ia ion, an e ec ha in e ac s gene ically wi h pRb (de Nooij e al., 1996; Lane e al., 1996). These da a sugges ha p27KIP1 may ac as a “ a e swi ch” ha , once ex- p essed a su icien le els, commi s he os eoblas o a pos mi- o ic s a e. I e e sible cell cycle exi is a ea u e o senescence, in which p27KIP1 plays a ole and which may ep esen a de- ense o oncogenic ans o ma ion (Se ano e al., 1997; Selle s Figu e 7. Exp ession o p27 KIP1 , os eocalcin, and p oli e a ing cell nuclea an igen (PCNA) in human os eosa coma samples. (A–I) High- powe pho omic og aphs o pa allel sec ions om wo high-g ade (A–C and G–I) and one low-g ade (G–I) human os eosa comas we e s ained o p27KIP1 (A, D, and G), os eocalcin (B, E, and H), and PCNA (C, F, and I). A ows in D–F indica e mul inuclea ed os eoclas ; a ows in G–I indica e os eocy es. Ba , 50 m. (J) Blinded quan i a ion o s aining o p27KIP1 and PCNA in umo s wi h e idence o os eo- blas di e en ia ion (os eoid p oduc ion) com- pa ed wi h dedi e en ia ed umo s. E o ba s ep esen SEM. *, P  0.05. Figu e 8. A model o in e ac ions be ween cell cycle p o eins and unx2 in os eoblas s. The in e ac ion o hypophospho yla ed pRB wi h unx2 comple es a posi i e eedback loop, p omo ing cell cycle wi hd awal and exp ession o he os eoblas pheno ype. See Discussion sec ion. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 JCB • VOLUME 167 • NUMBER 5 • 2004932 e al., 1998; Alexande and Hinds, 2001; Thomas e al., 2001). Clea ly, e minal cell cycle exi , whe he in esponse o di e - en ia ion o o oncogenic e en s, is undamen ally inconsis en wi h oncogenic ans o ma ion. Does p27KIP1 ac as a umo supp esso in bone? In animal models, loss o p27KIP1 is associa ed wi h in equen spon ane- ous pi ui a y umo s and in es inal adenomas bu accele a es he a e o umo o ma ion when combined wi h ca cinogen exposu e (Fe o e al., 1998) o mu a ions o TP53 (Philipp-S a- heli e al., 2004). In e es ingly, os eosa comas we e obse ed in his la e s udy, albei a low equency. Consis en wi h a ole o p27KIP1 in os eosa coma, he p o ooncogene c-Fos, which causes os eosa coma in mice (G igo iadis e al., 1993), induces cyclin A–Cdk2 ac i i y and ep esses p27KIP1 in os eo- blas s (Sun e s e al., 2004). Unusually, p27KIP1 appea s o ac as a haploinsu icien umo supp esso in he mouse (Fe o e al., 1998), and, whe e human umo s ha e unde gone a loss-o - he e ozygosi y e en , silencing o he emaining allele is a e (Kawama a e al., 1995; Ponce-Cas aneda e al., 1995). This may be due o he dual ole o p27KIP1 as an assembly ac o o G1-phase cyclin complexes as well as a s oichiome ic inhibi- o o S-phase cyclin complexes (She and Robe s, 1999). Im- po an ly, dec eases in p27KIP1 p o ein exp ession ha e been ound in 60% o human ca cinomas (Slinge land and Pagano, 2000), and a e associa ed in b eas cance wi h poo p ognosis (F ede sdo e al., 1997). Ou clinical s udies e eal an associ- a ion be ween p27KIP1 exp ession and loss o di e en ia ion in human os eosa comas, independen o a es o p oli e a ion pe se (Fig. 7). Loss o di e en ia ion in sa comas in gene al is a ma ke o high-g ade s a us, which in u n is associa ed wi h wo se p ognosis. Dis up ion o unx2-dependen ansc ip ional ac i i y is common in os eosa coma cell lines and leads in a clinically measu able ashion o loss o bo h di e en ia ion and exp es- sion o p27KIP1 in human os eosa comas. Al hough he speci ic mechanisms con ibu ing o he loss o unc ion o unx2 a e no unde s ood, global deme hyla ion o os eosa coma cell lines esul s in eac i a ion o di e en ia ion concomi an wi h e e sion o ans o ma ion (unpublished da a). Me hyla ion is a well-desc ibed, common me hod o silencing umo supp es- so pa hways (Baylin and He man, 2000), and he es o a ion o di e en ia ion by deme hyla ion u he suppo s he no ion ha umo s gain a selec i e su i al ad an age by silencing di e en ia ion- ela ed g ow h inhibi o y p ocesses. We hope ha iden i ying a ge s o me hyla ion-induced silencing will shed addi ional ligh on he in e ac ions be ween di e en ia ion and cell cycle exi . Ma e ials and me hods Cell lines and eagen s The os eosa coma cell lines we e main ained in DME (GIBCO BRL) con- aining 15% FCS. CCL-7625 and CCL-7672 cells we e ob ained om Ame ican Type Cul u e Collec ion. MC3T3E1 cells we e main ained in MEM supplemen ed wi h 10% FCS. RB  /  3T3 and NIH3T3 cells we e in- ec ed wi h pBABEpu o e o i us (Mo gens e n and Land, 1990) exp ess- ing mu an cons uc s o unx2 (Thi una ukka asu e al., 1998), and wi h pooled clones selec ed wi h pu omycin. S able exp ession o cons uc s was con i med by immunoblo (unpublished da a). Wild- ype and RB  /  MEFs wi h he desc ibed geno ypes (wild- ype and RB  /  ) we e de i ed om ma ched li e ma es (gi s o T. Jacks, Massachuse s Ins i u e o Tech- nology, Bos on, MA), and p27KIP1  /  mice we e ob ained om J. Robe s (F ed Hu chinson Cance Resea ch Cen e , Sea le, WA) (Fe o e al., 1996). Mine aliza ion was induced by cul u e in he p esence o 5 mM -glyce ophospha e and 50 g/ml asco bic acid o 3 wk a e con lu- ence (Thomas e al., 2001). Fo colony supp ession assays, cells (105/10-cm pla e) we e ans ec ed wi h cons uc s as indica ed in Figs. 2 A and 4 D. A e 24 h, cells we e selec ed in he p esence o an ibio ic (2 g/ml pu o- mycin o 1–5 g/ml neomycin) o 14–21 d. Colonies we e isualized wi h c ys al iole . Doses o 5-aza-2-deoxycy idine we e i a ed in p elimi- na y s udies o each cell line o achie e g ow h a es wi hou signi ican cell dea h, usually be ween 2 and 5 M. SV-Rb and SV-HARb we e used o he exp ession o ull-leng h pRb (Hinds e al., 1992). Exp ession cons uc s o unx2 we e cloned in o pBABEpu o (Mo gens e n and Land, 1990). The e o i us was ampli ied and pu i ied in ampho opic packaging cell line Phoenix 293 (cou esy o G. Nolan, S an o d Uni e si y, Palo Al o, CA), acco ding o he me hod o Pea e al. (1993). Plasmids we e ans ec ed in o cells wi h he use o Fu- gene, acco ding o he manu ac u e ’s ins uc ions (Roche Pha maceu i- cals). Adeno i al cons uc s exp essing pRb and unx2-FLAG we e gene - a ed as epo ed p e iously (Thomas e al., 2001). Cell-based assays Luci e ase assays we e pe o med acco ding o manu ac u e ’s ins uc ions (P omega). Whe e indica ed (Fig. 1, B and C), esul s we e no malized o ans ec ion e iciency wi h -galac osidase ac i i y o p o ein con en . AP ac i i y was assayed as desc ibed p e iously (Selle s e al., 1998). Assays o mine aliza ion we e pe o med as desc ibed p e iously (Thomas e al., 2001). Quan i a ion was pe o med by dissol ing s ained mine alized cul- u es in 10% ce ylpy idinium chlo ide, ollowed by spec opho ome ic analysis a 540 nm. Bo h AP ac i i y and mine aliza ion we e no malized o p o ein con en (Bio-Rad Labo a o ies). Flow cy ome y o DNA con en was pe o med as desc ibed p e iously (Thomas e al., 2001). RT-PCR analysis o gene exp ession RNA was ex ac ed wi h he use o TRIzol (In i ogen), acco ding o he manu ac u e ’s ins uc ions. cDNA was p oduced om 1 g o o al RNA wi h he use o a comme cially a ailable ki (SUPERSCRIPT Choice sys em o cDNA syn hesis; GIBCO BRL). Semiquan i a i e PCR analysis was pe - o med a e op imiza ion. P ime sequences and PCR condi ions a e a ail- able on eques . Fo quan i a i e RT-PCR, exp ession o each a ge gene was no malized o exp ession o ARPP0 wi h he use o an ABI-P ism 7700 Ligh Cycle and SYBR G een. Op imal PCR condi ions we e es ablished o each gene in p elimina y expe imen s. Analysis o p o ein exp ession and kinase assays Nuclea ex ac s and immunoblo analyses we e pe o med as desc ibed p e iously (Thomas e al., 2001). The ollowing an ibodies we e used: an i-FLAG an ibody (M2; Sigma-Ald ich); human pRb: monoclonal an i- body 245 (BD Biosciences); unx2: M-70 (San a C uz Bio echnology, Inc.); p27KIP1: K25020 (T ansduc ion Labo a o ies); and cyclin E: HE12, cyclin A: H432, and Cdk2: M2 (San a C uz Bio echnology, Inc.). Ho se- adish pe oxidase–conjuga ed seconda y an ibodies we e used (Jackson ImmunoResea ch Labo a o ies) and signal was de ec ed by ECL (NEN Li e Science P oduc s). The GST- unx2 pulldown s udies shown in Fig. 4 D we e pe o med as desc ibed p e iously (Thomas e al., 2001). Kinase as- says we e pe o med as desc ibed p e iously (Alexande and Hinds, 2001). Cyclin A was immunop ecipi a ed using aga ose-conjuga ed an i- body BF683 (Ups a e Bio echnology). Immunohis ochemis y 25 pa a in-embedded os eosa coma samples we e ob ained om he pa- hology a chi es a S . Vincen ’s Hospi al Melbou ne, wi h app o al om he Human Resea ch E hics Commi ee. 2-mm co es we e punched and hen assembled in o a issue mic oa ay. Sec ions we e cu a 3 m and moun ed on o Supe os Plus slides. P ima y an ibodies we e incuba ed o 30 min. Fo p27KIP1, clone SX53G8 (DakoCy oma ion) was used a 1:50. A p edilu e monoclonal an ibody o human os eocalcin (OC-1; Biogenex) was used a 1:4. Fo PCNA, clone PC10 (DakoCy oma ion) was used a 1:400. The p ima y an ibody was de ec ed wi h he mouse En ision sys- em (DakoCy oma ion). Immuno eac i i y was isualized wi h AEC ch o- mogen (DakoCy oma ion), using hema oxylin as a coun e s ain. Samples o p27KIP1 and os eocalcin we e incuba ed in 10 mM boiling sodium ci- a e bu e , pH 6.0, o 2 min be o e s aining. on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004 DISRUPTION OF DIFFERENTIATION IN OSTEOSARCOMA • THOMAS ET AL. 933 Slides we e imaged using a mic oscope (Axioskop 2; Ca l Zeiss Mic oImaging, Inc.) wi h a Plan-Neo lua objec i e (40, 0.75 NA), and a cooled colo digi al came a (RT Slide SPOT; Diagnos ic Ins umen s) and so wa e (SPOT V4.0.2 o Windows). Subsequen p ocessing o TIFF iles was unde aken in Adobe Pho oshop (V7.0.1). Images we e c opped, labeled as indica ed in Fig. 7, and assembled in o composi es o igu es a e mino adjus men s o con as and colo balance we e ap- plied o all pa s o each image. Mic oa ay analysis To al RNA om os eosa coma and common e e ence cell lines was iso- la ed using phenol-chlo o o m ex ac ion (TRIzol; In i ogen) and pu i ied by column ch oma og aphy (RNeasy; QIAGEN). The common e e ence RNA, con aining pooled RNA om 11 human umo cell lines, was p e- pa ed as desc ibed p e iously (Pollack, 2002). To al RNA (40–50 g) was e e se ansc ibed wi h Moloney Mu ine Leukemia Vi us Re e se ansc ip ase (P omega), in he p esence o amino-allyl (AA)–modi ied dUTP (Sigma-Ald ich). AA-dUTP cDNA was labeled by coupling o Cy3 and Cy5 ( e e ence and sample, espec i ely) mono eac i e dyes (Ame - sham Biosciences). cDNA a ays con aining 10.5 K elemen s ep esen - ing 9,381 unique cDNA (Unigene build 172) we e p oduced a Pe e MacCallum Cance Cen e Mic oa ay Co e acili y on supe amine slides (Telechem), wi h a obo ic a aye (Vi ek/Bio-Rad Labo a o ies). Labeled p obe was hyb idized o he a ay in 3.1  SSC and 50% o mamide a 42C o 14–16 h in a humidi ied and empe a u e-con olled chambe (HyP o20; The mo Hybaid). Slides we e washed a oom empe a u e wi h 0.5  SSC/0.01% SDS ( o 1 min), hen wi h 0.5  SSC ( o 3 min), and inally wi h 0.06  SSC ( o 3 min). Scanning was pe o med wi h an Ag- ilen G2565AA Mic oa ay Scanne and da a was ex ac ed wi h Gene- Pix P o 4.1 so wa e (Axon Ins umen s, Inc.). All a ay expe imen s, in- cluding cell cul u e, we e pe o med independen ly wice. A comple e lis o genes is a ailable om he au ho s on eques . Da a om each inde- penden expe imen we e a e aged, and hen he median ob ained om all six cell lines was used in he da a shown in Fig. 1. Da a we e analyzed wi h GeneSp ing so wa e (Silicon Gene ics), and samples we e no mal- ized wi h LOWESS (Yang e al., 2002). His omo phome y Tibiae we e collec ed om male p27KIP1  /  and wild- ype li e ma es a 16 wk o age, ixed in cold 4% pa a o maldehyde in PBS o e nigh , and embedded in me hylme hac yla e (Sims e al., 2000). Double luo och ome labeling o quan i a e mine al apposi ional a es was pe o med as de- sc ibed p e iously (Sims e al., 2000). 5-m sec ions we e s ained wi h oluidine blue o analyzed uns ained o luo och ome labels acco ding o s anda d p ocedu es in he p oximal ibia using he Os eomeasu e sys em (Os eome ics, Inc.). Tibial co ical hickness and pe ios eal mine al apposi- ional a es we e measu ed as desc ibed p e iously (Sims e al., 2000). The au ho s would like o hank membe s o he Thomas lab and Phil Da cy o help ul discussions. D.M. Thomas is he ecipien o a Na ional Heal h and Medical Re- sea ch Council R.D. W igh ellowship (Regkey 251752) and is suppo ed by he Cance Council o Vic o ia. P.W. Hinds and G. Gu ie ez a e suppo ed by Na ional Ins i u es o Heal h g an AG20208. Submi ed: 30 Sep embe 2004 Accep ed: 28 Oc obe 2004 Re e ences Alexande , K., and P.W. Hinds. 2001. Requi emen o p27(KIP1) in e inoblas- oma p o ein-media ed senescence. Mol. Cell. Biol. 21:3616–3631. Aubin, J.E. 1998. Bone s em cells. J. Cell. Biochem. Suppl. 30-31:73–82. Balin , E., D. Lapoin e, H. D issi, C. Van De Meijden, D.W. Young, A.J. Van Wijnen, J.L. S ein, G.S. S ein, and J.B. Lian. 2003. Pheno ype disco e y by gene exp ession p o iling: mapping o biological p ocesses linked o BMP-2-media ed os eoblas di e en ia ion. J. Cell. Biochem. 89:401–426. Baylin, S.B., and J.G. He man. 2000. DNA hype me hyla ion in umo igenesis: epigene ics joins gene ics. T ends Gene . 16:168–174. Bodine, P.V., M. T ailsmi h, and B.S. Komm. 1996. De elopmen and cha ac- e iza ion o a condi ionally ans o med adul human os eoblas ic cell line. J. Bone Mine . Res. 11:806–819. Casaccia-Bonne il, P., R. Tikoo, H. Kiyokawa, V. F ied ich J ., M.V. Chao, and A. Ko . 1997. Oligodend ocy e p ecu so di e en ia ion is pe u bed in he absence o he cyclin-dependen kinase inhibi o p27KIP1. Genes De . 11:2335–2346. Chambe s, T.J., J.A. Da by, and K. Fulle . 1985. Mammalian collagenase p e- disposes bone su aces o os eoclas ic eso p ion. Cell Tissue Res. 241: 671–675. Dahlin, D.C. 1957. Bone Tumo s. Cha les C. Thomas, Sp ing ield, IL. 224 pp. de Nooij, J.C., M.A. Le end e, and I.K. Ha iha an. 1996. A cyclin-dependen ki- nase inhibi o , Dacapo, is necessa y o imely exi om he cell cycle du ing D osophila emb yogenesis. Cell. 87:1237–1247. Dim i, G.P., X. Lee, G. Basile, M. Acos a, G. Sco , C. Roskelley, E.E. Me- d ano, M. Linskens, I. Rubelj, and O. Pe ei a-Smi h. 1995. A bioma ke ha iden i ies senescen human cells in cul u e and in aging skin in i o. P oc. Na l. Acad. Sci. USA. 92:9363–9367. D issi, H., D. Hushka, F. Aslam, Q. Nguyen, E. Bu one, A. Ko , A.J. an Wijnen, J.B. Lian, J.L. S ein, and G.S. S ein. 1999. The cell cycle egula- o p27(KIP1) con ibu es o g ow h and di e en ia ion o os eoblas s. Cance Res. 59:3705–3711. Ducy, P., and G. Ka sen y. 1995. Two dis inc os eoblas -speci ic cis-ac ing ele- men s con ol exp ession o a mouse os eocalcin gene. Mol. Cell. Biol. 15:1858–1869. Ducy, P., R. Zhang, V. Geo oy, A.L. Ridall, and G. Ka sen y. 1997. Os 2/ Cb a1: a ansc ip ional ac i a o o os eoblas di e en ia ion. Cell. 89: 747–754. Ducy, P., M. S a buck, M. P iemel, J. Shen, G. Pine o, V. Geo oy, M. Amling, and G. Ka sen y. 1999. A Cb a1-dependen gene ic pa hway con ols bone o ma ion beyond emb yonic de elopmen . Genes De . 13:1025–1036. Du and, B., M.L. Fe o, J.M. Robe s, and M.C. Ra . 1998. p27KIP1 al e s he e- sponse o cells o mi ogen and is pa o a cell-in insic ime ha a es s he cell cycle and ini ia es di e en ia ion. Cu . Biol. 8:431–440. Fe o, M.L., M. Ri kin, M. Tasch, P. Po e , C.E. Ca ow, E. Fi po, K. Polyak, L.H. Tsai, V. B oudy, R.M. Pe lmu e , e al. 1996. A synd ome o mul- io gan hype plasia wi h ea u es o gigan ism, umo igenesis, and e- male s e ili y in p27(KIP1)-de icien mice. Cell. 85:733–744. Fe o, M.L., E. Randel, K.E. Gu ley, J.M. Robe s, and C.J. Kemp. 1998. The mu ine gene p27Kip1 is haplo-insu icien o umou supp ession. Na- u e. 396:177–180. Feue bach, D., E. Loe sche , K. Bue ki, T.K. Sampa h, and J.H. Feyen. 1997. Es ablishmen and cha ac e iza ion o condi ionally immo alized s o- mal cell lines om a empe a u e-sensi i e T-Ag ansgenic mouse. J. Bone. Mine . Res. 12:179–190. F ede sdo , S., J. Bu ns, A.M. Milne, G. Packham, L. Fallis, C.E. Gille , J.A. Royds, D. Pes on, P.A. Hall, A.M. Hanby, e al. 1997. High le el exp es- sion o p27(kip1) and cyclin D1 in some human b eas cance cells: in- e se co ela ion be ween he exp ession o p27(kip1) and deg ee o ma- lignancy in human b eas and colo ec al cance s. P oc. Na l. Acad. Sci. USA. 94:6380–6385. Golds ein, S. 1990. Replica i e senescence: he human ib oblas comes o age. Science. 249:1129–1133. Go i, F., T. Thomas, K.C. Hicok, T.C. Spelsbe g, and R.L. Riggs. 1999. Di e - en ia ion o human ma ow s omal p ecu so cells: bone mo phogene ic p o ein-2 inc eases OSF2/CBFA1, enhances os eoblas commi men , and inhibi s la e adipocy e ma u a ion. J. Bone Mine . Res. 14:1522–1535. G igo iadis, A.E., K. Schellande , Z.Q. Wang, and E.F. Wagne . 1993. Os eo- blas s a e a ge cells o ans o ma ion in c- os ansgenic mice. J. Cell Biol. 122:685–701. G on hos, S., A.C. Zanne ino, S.J. Hay, S. Shi, S.E. G a es, A. Ko esidis, and P.J. Simmons. 2003. Molecula and cellula cha ac e isa ion o highly pu i ied s omal s em cells de i ed om human bone ma ow. J. Cell Sci. 116:1827–1835. Ha ada, H., S. Tagashi a, M. Fujiwa a, S. Ogawa, T. Ka suma a, A. Yamaguchi, T. Komo i, and M. Naka suka. 1999. Cb a1 iso o ms exe unc ional di e ences in os eoblas di e en ia ion. J. Biol. Chem. 274:6972–6978. Hinds, P.W., S. Mi nach , V. Dulic, A. A nold, S.I. Reed, and R.A. Weinbe g. 1992. Regula ion o e inoblas oma p o ein unc ions by ec opic exp es- sion o human cyclins. Cell. 70:993–1006. Hi ano, M., K. Hi ano, J. Nishimu a, and H. Kanaide. 2001. T ansc ip ional up- egula ion o p27(KIP1) du ing con ac -induced g ow h a es in ascula endo helial cells. Exp. Cell Res. 271:356–367. Hopyan, S., N. Gokgoz, R.S. Bell, I.L. And ulis, B.A. Alman, and J.S. Wunde . 1999. Exp ession o os eocalcin and i s ansc ip ional egula o s co e- binding ac o alpha 1 and MSX2 in os eoid- o ming umo s. J. O hop. Res. 17:633–638. Kawama a, N., R. Mo ose i, C.W. Mille , D. Pa k, K.S. Spi in, T. Nakamaki, S. Takeuchi, Y. Ha a, J. Simpson, S. Wilcyznski, e al. 1995. Molecula analysis o he cyclin-dependen kinase inhibi o gene p27/Kip1 in hu- man malignancies. Cance Res. 55:2266–2269. Kiyokawa, H., R.D. Kineman, K.O. Mano a-Todo o a, V.C. Soa es, E.S. Ho - man, M. Ono, D. Khanam, A.C. Hayday, L.A. F ohman, and A. Ko . on Feb ua y 25, 2016jcb. up ess.o gDownloaded om Published Decembe 6, 2004