ARTICLE
A FoxA2+long- e m s em cell popula ion
is necessa y o g ow h pla e ca ilage
egene a ion a e inju y
Shanmugam Mu uganandan 1, Rachel Pie ce1, Dian As a i Teguh 2, Rocio Fuen e Pe ez3, Nicole Bell4,
B andon Nguyen5, Ka he ine Hohl 2,6, B ian D. Snyde 7, Ma k W. G ins a 6, Hannah Albe ico1,
Do i Woods1, Yiwei Kong1, Co neliu Sima 8, Sanke Bhaga 9, Kailing Ho10, Vicki Rosen 10,
Lau a Game 10 & And eia M. Ionescu 1✉
Longi udinal bone g ow h, achie ed h ough endochond al ossifica ion, is accomplished by a
ca ilaginous s uc u e, he physis o g ow h pla e, comp ised o mo phologically dis inc
zones ela ed o chond ocy e unc ion: es ing, p oli e a ing and hype ophic zones. The
es ing zone is a s em cell- ich egion ha gi es ise o he g ow h pla e, and exhibi s
egene a i e capabili ies in esponse o inju y. We disco e ed a FoxA2+g oup o long- e m
skele al s em cells, si ua ed a he op o es ing zone, adjacen he seconda y ossifica ion
cen e , dis inc om he p e iously cha ac e ized PTH P+s em cells. Compa ed o PTH P+
cells, FoxA2+cells exhibi highe clonogenici y and longe i y. FoxA2+cells exhibi dual
os eo-chond o-p ogeni o ac i i y du ing ea ly pos na al de elopmen (P0-P28) and chon-
d ogenic po en ial beyond P28. When he g ow h pla e is inju ed, FoxA2+cells expand in
esponse o auma, and p oduce physeal ca ilage o g ow h pla e issue egene a ion.
h ps://doi.o g/10.1038/s41467-022-30247-1 OPEN
1Depa men o Biology, 134 Muga Li e Sciences Building, No heas e n Uni e si y, 360 Hun ing on A e, Bos on, MA 02115, USA. 2Cen e o Ad anced
O hopedic S udies, Be h Is ael Deaconess Medical Cen e , 330 B ookline A enue, Bos on, MA 02215, USA. 3Di ision o Pedia ics, Uni e si y o O iedo,
O iedo 33206, Spain. 4New Yo k Uni e si y College o Den is y, 345 E.24 h S , New Yo k, NY 10010, USA. 5Mode na The apeu ics, One Upland Rd,
No wood, Ohio, MA 02062, USA. 6Depa men s o Biomedical Enginee ing, Chemis y, and Medicine, Bos on Uni e si y, 590 Commonweal h A e, SCI 518,
Bos on, MA 02215, USA. 7Depa men o O hopedic Su ge y, Bos on Child en’s Hospi al, 300 Longwood A e, Bos on, MA 02115, USA. 8Depa men o
O al Medicine, In ec ion, and Immuni y, Ha a d School o Den al Medicine, 188 Longwood A enue, Bos on, MA 02115, USA. 9Ul agenyx Pha maceu ical,
840 Memo ial D i e, Camb idge, MA 02139, USA. 10 Depa men o De elopmen al Biology, Ha a d School o Den al Medicine, 188 Longwood A enue,
Bos on, MA 02115, USA. ✉email: a.ionescu@no heas e n.edu
NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions 1
1234567890():,;
Longi udinal bone g ow h is achie ed by endochond al ossi-
fica ion, implemen ed by he physis o g ow h pla e (GP), a
ca ilaginous s uc u e loca ed be ween he epiphysis and
me aphysis a he ends o long bones o ming he appendicula
skele on. Human GP is comp ised o fi e mo phologically dis inc
zones: es ing (RZ), p oli e a ion (PZ), hype ophy (HZ), calci-
fica ion (CZ), and ossifica ion (OZ)1. Physeal inju ies a e a majo
cause o skele al mo bidi y in g owing child en, mani es ing as
angula o sho ened limb de o mi ies, due o pa ial o comple e
a es o GP unc ion2. Clinically, GP inju ies a e classified
acco ding o he Sal e –Ha is (SH) sys em, p edica ed on ac-
u e mo phology in ol ing he physis ± epiphysis and/o me a-
physis: SH1) physeal shea sepa a ion; SH2) physis +me aphysis;
SH3) physis +epiphysis; SH 4) physis +epiphysis +me aphysis;
SH 5) c ush inju y o physis. SH1 ac u es ha e he bes p og-
nosis, healing wi hou dis up ed g ow h. The clea age plane is
p ima ily h ough he HZ, s uc u ally he weakes zone com-
p ising he GP, owing o lack o calcifica ion and limi ed col-
lagenous ma ix. S i ened by a collagen ma ix, he adjacen RZ
and PZ a e una ec ed3,4. In con as , SH 3–5 inju ies dis u b all
zones, esul ing in dis u bed bone g ow h3,4. These clinical sce-
na ios sugges ha in ol emen o he RZ a ec s he abili y o he
GP o egene a e in esponse o inju y. In abbi s, e en when he
PZ and HZ om ulna GP we e excised, he es o a ion o he GP
was s ill possible, as long as he RZ emained in ac 5,6. As he
name implies, he “ es ing zone”houses a quiescen p ogeni o
cell popula ion1. Se e al s udies ha e iden ified in mouse7, a 8,
and abbi 9slow-cycling cells p esen in he RZ immedia ely
adjacen o he seconda y ossifica ion cen e (SOC). In pulse-
chase expe imen s, animals we e adminis e ed [3H] hymidine o
5-b omo-2-deoxyu idine (B dU) o a p olonged pe iod su ficien
o allow slow-cycling s em cells o comple e one cycle. A he end
o he chase, he label diminishes in p oli e a ing cells, bu
emains undilu ed in slow-cycling cells. While hese epo s
demons a e he p esence o slow-cycling, po en ial p ogeni o
cells a he RZ, hei gene ic signa u e emains undefined, as he e
is no bioma ke o isola e and cha ac e ize hese cells.
Recen ly New on e al., using clonal gene ic acing wi h a
col2CRE;ERT;R26-Con e i mouse line and unc ional pe u ba ions,
es ablished ha o ma ion o he SOC igge ed GP chon-
d op ogeni o s o unde go a adical shi in clonali y and o
acqui e sel - enewal capabili ies10. Howe e , since “Con e i”
labeling is based on collagen ype II exp ession, an ube -ma ke
o all GP cells, he e was no unique bioma ke o isola e and/o
cha ac e ize indi idual s em cell subg oups. Nex , Mizuhashi e al.
employed a PTH PC e.ERT;Toma o /+mouse s ain o e eal ha
PTH P ma ked a subse o s em cells loca ed a he bo om o he
RZ11. Labeling, acing and isola ing RZ cells based on hei
PTH P exp ession bo h in i o and in i o, demons a ed ha a
subg oup PTH P+(posi i e) s em cell e ol ed p edominan ly
in o columna chond ocy es11. Toge he wi h ano he ye uni-
den ified PTH P−(nega i e) subg oup loca ed a he op o
he RZ, adjacen o he SOC, PTH P+cells, loca ed p ima ily a
he bo om o he RZ, con ibu ed o long- e m GP g ow h11.
In he p esen wo k, we p o e ha a popula ion o Fox-
A2+col10−cells, loca ed a he op o he RZ, a e he PTH P-
(nega i e) s em cell popula ion, and ep esen a subg oup o long-
e m skele al s em cells (LTSSC) capable o dual os eo-chond o-
p ogeni o ac i i y du ing ea ly pos na al de elopmen (P0-P28)
and chond ogenic po en ial beyond P28, undamen al o GP
u no e and egene a ion ollowing inju y.
Resul s
FoxA2 exp ession p efigu es SOC o ma ion. We p e iously
disco e ed ha FoxA ansc ip ion ac o s a e key egula o s o
chond ocy e hype ophy12. FoxA1-3 a e highly exp essed in he
hype ophic zone (HZ) o newbo n mice GP (Fig. 1a, , k).
Howe e , unlike FoxA3, which is exp essed in a b oade domain
h oughou he GP (Fig. 1 ), o FoxA1, which is highly exp essed
in he HZ bu e y li le elsewhe e (Fig. 1k), FoxA2 is exp essed in
wo sepa a e domains: he HZ and a disc e e pe ia icula domain
loca ed a he ends o he long bones (Fig. 1a). Th oughou
pos na al de elopmen (P0-P14), his dis inc FoxA2+popula ion
is con inuously enla ging, possibly p efigu ing he o ma ion o
he SOC (Fig. 1a–e). FoxA1 and FoxA3 also co ela e wi h SOC
de elopmen , bu eme ge in he epiphyseal ca ilage a ound P5
(Fig. 1 –o), much la e han FoxA2, which is exp essed om bi h
(P0) (Fig. 1a–e). In he p esump i e SOC, he fi s col10+cells
appea by P7, as shown in Tg.col10mChe y mice, bu FoxA2+cells
appea as ea ly as P0 (Fig. 1p– ). Compa ison be ween FoxA2+
cells in he pe ia icula egion wi h he columna GP cells, did
no e eal any his ological o mo phological di e ences and no
no iceable change in Ki67 exp ession and p oli e a ion, p io o
P5 (Fig. 1u–y). Howe e , a e P5 FoxA2+cells down egula e he
exp ession o Ki67, in p epa a ion o hype ophic di e en ia-
ion (Fig. 1w, x). Al oge he , hese expe imen s sugges ha
FoxA2 exp ession ma ks he epiphyseal ca ilage, mos o which
will o m he seconda y ossifica ion cen e , a soone han col.10
exp ession.
FoxA2+cells gi e ise o he GP RZ op compa men .To
de e mine whe he FoxA2 ma ks he a e o he cells des ined o
become SOC, we labeled and aced FoxA2+cells, using a Fox-
A2C eERT/+mouse line consis ing o a amoxi en-inducible
C eERT2 d i e knocked-in he FoxA2 3’UTR13. Two amox-
i en pulses, adminis e ed P3-4 o P7-8, o FoxA2C eERT/+;
ZsG eenfl/+;Tg.col10mche y mice ga e ise, by P18, o ZsG een+
cells p esen p ima ily in he SOC, wi h a ew ZsG een+cells
p esen a he op o he GP RZ (Fig. 2A(a, b, e)). I was p e-
iously published ha hype ophic chond ocy es become bone
cells and con ibu e o he os eogenic lineage14–16. FoxA2 is no
exp essed in he epiphyseal bone o he SOC, bu ins ead i is
Ki67
P0 P3 P5 P7
FoxA2
Tg.Col10
mche y
abcd
p s
u wx y
s
P14
e
FoxA3
FoxA1
gij
Immunohis ochemis y FoxA2 / Hoesch
h
klmn o
q
Fig. 1 FoxA1-3 exp ession domain p edic s he o ma ion o he SOC.
Immunohis ochemis y o FoxA2 (a–e), FoxA3 ( –j), FoxA1 (k–o), Ki67
(u–y), and fluo escence mic oscopy on Tg.col10a1mche y mice (p– )on
pos na al day P0, P3, P5, P7, P14 ibia sec ions. FoxA1-3 (yellow), Ki67
(whi e), Hoechs (blue), mche y fluo escence ( ed). Scale ba s, 100 µm.
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1
2NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions
es ic ed o he GP hype ophic ca ilage and a small disc e e
domain a he end o he long bones, as shown by immunohis-
ochemis y o FoxA2 (Fig. 1a–e). Thus, he p esence o Fox-
A2+p ogeny in he epiphyseal SOC i is likely he esul o
FoxA2+hype ophic chond ocy es con ibu ing o he os eo-
genic lineage. To confi m his, we adminis e ed 2x amoxi en
injec ions o FoxA2C e.ERT/+;Toma o / ;Tg.col1GFP mice a P3-4,
P7-8, o P13-14 and we ollowed hem 3 mon hs la e , demon-
s a ing ha Toma o+p ogeny (de i ed om FoxA2+cells) a e
col.1+exp essing bone cells (Fig. 2B(a–c)).
By P13, du ing SOC enla gemen , we see a significan build-up o
ZsG een+cells in he GP ca ilage. In FoxA2C eERT2/+;ZsG eenfl/+;
Tg.col10mche y mice injec ed wi h amoxi en P13-14 and ha es ed a
P18, we coun ed 126 ± 40 cells/hindlimb. Fi e amoxi en injec ions,
adminis e ed P13-17 o FoxA2C eERT2/+;ZsG eenfl/+;Tg.col10
mche y
mice, led o a u he inc ease in ZsG een+cells (387 ± 109 cells/
hindllimb) (Fig. 2A(c, d, e)).
As FoxA2+cells a e loca ed on he bo de be ween GP and SOC,
we sough o delinea e he ca ilage/bone in e ace by coun e -
s aining wi h 5-DTAF(4,6-Dichlo o iazinyl Aminofluo escein) dye,
which labels e ficien ly he bone ma ix bu no he ca ilage ma ix
in Agg ecanC eERT/+;Toma ofl/+mice (Supplemen a y Fig. 1A(a)).
A P13-17 amoxi en pulse adminis e ed o FoxA2C eERT2/+;
Toma ofl/+mice, e ealed ha 75% o FoxA2+cells a e in he
GP ca ilage, ou side he 5-DTAF labeled bone domain, and 25%
a e in he g een labeled SOC (Supplemen a y Fig. 1A(b, c)). This
FoxA2+popula ion loca ed a he op o he GP RZ, a he
ca ilage/bone in e ace, i is no only p esen in mice, bu in abbi s
as well (Supplemen a y Fig. 1B).
To in es iga e whe he he FoxA2+cells p esen in he RZ in
ea ly pos na al de elopmen (P0-P14) s ill pe sis in he GP
ca ilage 3 mon hs la e , we pe o med a long- e m pulse chase o
he FoxA2C eERT/+;Toma ofl/+mice. Tamoxi en pulses a P3-4
(du ing SOC p efigu a ion), a P7-8 (SOC hype ophy), and
P13-14 (SOC enla gemen ) o FoxA2C eERT/+;Toma ofl/+mice
ga e ise o Toma o+labeled cells which, 3 mon hs la e , we e
s ill p esen in he RZ ei he as single cells, clus e s o columns o
p ogeny (Fig. 2B(d– )), o became col.1+bone cells in he SOC
(Fig. 2B(a–c)). Al oge he , hese expe imen s sugges ha a
FoxA2+popula ion, le behind om epiphyseal ca ilage
A
a***
***
Tam P3-4 P7-8 P13-14 P13-17
Quan i ica ion o GP RZ FoxA2+ cells ( ) . SOC FoxA2+ cells ( ) using
FoxA2C e.ERT/+; ZsG een /+Tg.col10mche y mice
a1
a1 b1
b1 c1
c1 d1
d1
FoxA2+ cells om GP
FoxA2+ cells om SOC
bcd
e
*
Tam Ha es
P3-4 P18
Tam Ha es
P7-8 P18
Tam Ha es
P13-14 P18
Tam Ha es
P13-17 P18
# FoxA2+ cells
Tam Ha es
P3-4 P90
Tam
P7-8 P90
Ha es
P13-14 P90
BDual con ibu ion o FoxA2+ cells o os eogenic and chond ogenic lineages
FoxA2C e.ERT/+Toma o / Tg.Col1GFP mice FoxA2C e.ERT/+Toma o / 5-DTAF mice
Ha es
P3-4 P90
Ha es
P7-8 P90
Ha es
P13-14 P90
abcde
1
a1
2
a2 b1 b2
211
2
c1 c2 d1 d2
2
1
e1 e2
1
2
1 2
2
1
0
200
400
600
800
1000
***
***
Tam Tam Tam Tam
Ha es
Fig. 2 A popula ion o FoxA2+col10−cells p efigu es SOC o ma ion and gi es ise o he op compa men o he GP RZ. A Tibia sec ions ha es ed
om FoxA2C eERT2/+;ZsG eenfl/+; Tg.Col10mche y mice ea ed wi h 2x amoxi en (Tam) injec ions a pos na al days P3-4 (a), P7-8 (b), P13-14 (c), o 5x
amoxi en injec ions P13-P17(d), and ha es ed a P18. Hoechs (blue), ZsG een fluo escence mic oscopy (g een), mche y fluo escence mic oscopy ( ed).
Scale ba s, 100 µm. Rep esen a i e de ails, om he in e ace be ween GP and SOC, a e shown in numbe ed inse s (a1–d1). Quan ifica ion o he numbe o
FoxA2+(ZsG een+) cells in GP is achie ed by coun ing ZsG een+cells loca ed in GP (ex ending 100 µm away om he GP/SOC in e ace, owa ds he
GP). The numbe o ZsG een+cells ep esen s a sum o n=8 sec ions pe mouse hindlimb. Simila quan ifica ion is pe o med o FoxA2+cells in SOC.
Da a p esen ed as mean ± SD, n=3 mice. The as e isks indica e significan di e ence: One-way ANOVA, Tukey es ; ***p=0.000003 (P3-4 GP s P3-4
SOC), 0.000005 (P7-8 GP s P7-8 SOC), 0.000569 (P3-4 GP s P13-17 GP), 0.000828 (P7-8 GP s P13-17 GP), and *p=0.011031(P13-17 GP s P13-17
SOC). Comple e s a is ical in o ma ion is p o ided in Supplemen a y Table 1 (e). BTibia sec ions om FoxA2C eERT2/+;Toma ofl/+;Tg.Col1GFP mice ea ed
wi h 2x amoxi en injec ions P3-P4 (a), P7-P8 (b), P13-P14 (c), and ha es ed a P90. Tibia sec ions om FoxA2C eERT2/+;Toma ofl/+mice ea ed wi h 2x
amoxi en injec ions P3-P4 (d), P7-P8 (e), P13-P14 ( ), ha es ed a P90. Toma o ( ed), Hoechs (blue), GFP (g een) (a–c), 5-DTAF (g een) (d– ). Scale
ba s, 100 µm. Rep esen a i e de ails a e shown in numbe ed inse s (a1– 1).
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1 ARTICLE
NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions 3
pos na al de elopmen (du ing SOC p efigu a ion, hype ophy
and enla gemen ), con ibu es o he GP RZ op compa men .
FoxA2+long- e m s em cells . PTH P+sho - e m s em cells.
Gi en ha FoxA2+cells eside in he RZ, a known loca ion o
s em cells5,11, and ha FoxA2+cells a e loca ed in close p oxi-
mi y o he SOC, which is iewed as a signaling cen e o RZ
s emness10, we asked whe he FoxA2+cells a e di e en han he
p e iously cha ac e ized PTH P+s em cell subse 11. Using a
PTH Pmche y mouse line, Mizuhashi e al. ha e ecen ly shown
ha PTH P+cells a e es ic ed o he bo om o he RZ, and
hey a e dedica ed, a leas o some deg ee, o gi ing ise o col-
umns o p ogeny11. He e we show ha FoxA2+cells a e loca ed
a he op o he RZ and hey a e geog aphically sepa a ed om
he p e iously cha ac e ized PTH P+cells11, in bo h ibia
and ulna o FoxA2C eERT/+;ZsG eenfl/+;PTH Pmche y mice
(Fig. 3A(a, b)). To confi m ou his ological findings, we pe -
o med FACS analysis on cells isola ed om GP issue o
FoxA2C eERT/+;ZsG eenfl/+;PTH Pmche y mice injec ed wi h
amoxi en P13 o P17, and sac ificed a P18. We so ed he
FoxA2+(ZsG een+) cells, he PTH P+(mChe y+) cells and he
double posi i e (DP) cells. Consis en wi h he his ological da a,
we obse ed no o e lap be ween FoxA2+cells and PTH P+cells
(Fig. 3A(d)). The pe cen age o double posi i e (DP) cells
(0.017 ± 0.004%) is he same as in insic (backg ound) fluo es-
cence o double nega i e (DN) con ol cells (0.014%).
To in es iga e whe he FoxA2+cells can become PTH P+cells
o e ime, we pulse-chased FoxA2C eERT/+;ZsG eenfl/+;PTH Pmche y
mice, by injec ing hem wi h amoxi en P13 o P17, and sac ificing
hem a P40 (Fig. 3A(c, e)). His ological da a e eals he p esence o
yellow FoxA2+PTH P+cells, while FACS analysis confi ms an
0.1% o double posi i e (DP) cells (Fig. 3A(c, e)), confi ming ha
FoxA2+cells can become PTH P+cells o e ime.
To inqui e abou FoxA2+cells sel - enewabili y and mul ipo ency,
we used iple ansgenic mice FoxA2C eERT2/+;ZsG eenfl/+;
Tg.col10mChe y, ea ed wi h amoxi en P13-P17 and ha es ed a
P18, ha allowed ei he g een labeling o RZ FoxA2+col10−cells, o
yellow labeling o HZ FoxA2+col10+cells (Supplemen a y Fig. 2A).
To compa e FoxA2+cells wi h PTH P+cells, we isola ed PTH P+
cells om P18 GP issue o PTH Pmche y mice. Bo h FoxA2+cells
and PTH P+cells isola ed om GP ca ilage ha e high le els o
ca ilage specificgenes(collagen ype 2,agg ecan) and low le els o
bone ma ke s (collagen ype 1, alkaline phospha ase) when compa ed
wi h bone ma ow mesenchymal cells (BMSC) (Supplemen a y
Fig. 2B).
In a c u (colony o ming uni ) assay, RZ FoxA2+col.10-(g een)
cells o med dis inc colonies (Supplemen a y Fig. 3a–c), while
HZ FoxA2+col.10+(yellow) cells ailed (Supplemen a y
Fig. 3d– ). This indica es ha FoxA2+cells, om RZ no HZ,
ha e he capaci y o o m clones when cul u ed ex i o. Nex we
compa ed he sel - enewabili y o FoxA2+cells (isola ed om
FoxA2C eERT2/+;ZsG eenfl/+;Tg.col10mChe y mice), wi h ha o
PTH P+cells (isola ed om PTH Pmche y mice) (Fig. 3B,
Supplemen a y Fig. 4A). A pla ing, PTH P+cells ga e ise o
40% mo e colonies han FoxA2+cells (51 s. 37), bu only 11%
(6/51) PTH P+p ima y colonies could o m seconda y colonies,
whe eas 38% (14/37) FoxA2+p ima y colonies could be u he
passaged (Fig. 3B, Supplemen a y Fig. 4B). FoxA2+colonies ha e
highe clonogenici y and longe i y han colonies es ablished om
PTH P+cells. Abou 9% (10/112) FoxA2+colonies each
Passage 9 and beyond, whe eas only 1.4% (2/143) PTH P+
colonies could each Passage 5 (Fig. 3B, Supplemen a y Fig. 4B).
To inqui e whe he FoxA2+clones a e mul ipo en , indi idual
FoxA2+clones, om Passage 1 and 9, we e u he expanded
in i o, and gene a ed Alcian-blue ma ix, Aliza in ed
mine alized ma ix and Oil Red d ople s du ing cul u e
in chond ogenic, os eogenic, o adipogenic di e en ia ion media
(4/4 clones) (Fig. 3C). Al oge he , hese expe imen s demons a e
ha , in i o, FoxA2+cells ha e mul ipo ency and highe
sel - enewabili y and longe i y han PTH P+cells.
Dual os eo-chond o-po en ial o FoxA2+cells p io o P28.
To in es iga e he dynamics o FoxA2+cells clonali y in i o, we
aimed o label su ficien FoxA2+cells o demons a e significan
con ibu ion o he GP issue, bu spa se enough o allow
assessmen o single clones (cells, clus e s and columns o p o-
geny) o e ime. Two amoxi en injec ions, adminis e ed P14-P15
o FoxA2C eERT/+;Toma ofl/+mice, labeled 7.92 ± 0.9 FoxA2+
cells, which is 8% om he o al numbe o FoxA2+cells de ec ed
by immunohis ochemis y a P14 (97.67 ± 9 cells) (Supplemen-
a y Fig. 5A(a, b, e)). Fi e amoxi en injec ions (adminis e ed
P14-P18 o FoxA2C eERT/+;Toma ofl/+mice) labeled 22.6 ± 4
FoxA2+cells, inc easing he labeling e ficiency ela i e o he
numbe o FoxA2+cells iden ified ia immunohis ochemis y,
om 8 o 23% (Supplemen a y Fig. 5A(a, c, e)). Fi e amoxi en
injec ions (adminis e ed P14-P18 o FoxA2C eERT2/+;Toma oflfl
mice), labeled 36.9 ± 2.3 FoxA2+cells, inc easing he e ficiency o
labeling om 23% (one Toma o floxed allele) o 38% ( wo
Toma o floxed alleles) (Supplemen a y Fig. 5A(a, d, e)). This
inc ease in labeling e ficiency was eflec ed in a 2- old inc ease in
he numbe o columns in FoxA2C eERT/+;Toma oflfl mice as
compa ed wi h FoxA2C eERT/+;Toma ofl/+mice ha es ed
9 mon hs a e he las amoxi en injec ion (Supplemen a y
Fig. 5B). O e all, inc eased labeling o he FoxA2+cells
demons a ed inc eased con ibu ion o he FoxA2+cells o he
GP ca ilage, bu i also inc eased densi y o he Toma o+cells
hinde ing single clone analysis. As such, in lineage acing
expe imen s, we labeled a mode a e amoun o FoxA2+cells by
gi ing 5x amoxi en injec ions (P14 o P18) o FoxA2C eERT/+;
Toma ofl/+mice (wi h one Toma o floxed allele) (Fig. 4A).
When FoxA2C eERT2/+;Toma ofl/+mice we e pulsed wi h
amoxi en P14-P18 and chased o 1 day o 1, 3, 9 mon hs we
obse ed inc easing numbe s o double s, clus e s, and columns
o p ogeny (Fig. 4A). Al hough he ini ial pe cen age o columns
is 1% a 1 mon h, he e was subs an ial inc ease in column
o ma ion o e ime, and by 9 mon hs, he columns o p ogeny
accoun o 26% o he o al combined s uc u es (single cells,
clus e s, and columns) (Fig. 4C1, Supplemen a y Fig. 6C). This is
in con as wi h PTH P+cells, which ga e ise o a la ge numbe
o columns ini ially, bu he numbe dec eased significan ly o e
ime11. FoxA2+cells beha e as LTSSC, fi s gi ing ise o
p ogeny, which hen p oduce columns o chond ocy es. Conse-
quen ly, he pe cen age o FoxA2+clus e s peaks ea ly (43% by
1 mon h), bu d ops o 31% by 9 mon hs, when FoxA2+column
o ma ion is high (Fig. 4C1). Las ly, despi e o ming clus e s and
columns, a significan numbe o FoxA2+cells emains as single
cells. The pe cen age o single cells dec eases apidly o e ime,
om 82% (1 day), o 55% (1 mon h), bu he e is s ill a high
ac ion o single cells as la e as 9 mon hs a e labeling (Fig. 4C1).
Al oge he , hese findings sugges ha FoxA2+cells cons i u e
a mo e quiescen , LTSSC popula ion loca ed nea he SOC,
which gi es ise o an inc easing numbe o columns o p ogeny
o e ime.
Nex , we compa ed he clonali y o FoxA2+cells, labeled
du ing SOC enla gemen (P14-P18), wi h he clonali y o
FoxA2+cells, labeled du ing SOC os eogenic ma u a ion (P28-
P37). Quan ifica ion o he pe cen age o FoxA2+clones (single
cells, clus e s o columns) o he o al numbe o Toma o+uni s,
shows a simila dynamic. Bo h FoxA2+cells labeled a a younge
age ( amoxi en P14-18), o a an olde age ( amoxi en P28-37),
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1
4NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions
exhibi he same LTSSC cha ac e is ics in i o, as shown by an
inc eased pe cen age o FoxA2+columns o e ime (Fig. 4A–C).
Howe e , quan ifica ion o he numbe o FoxA2+clones high-
ligh s highe he e ogenei y in he mice labeled a a younge age.
A high numbe (138 FoxA2+clones/hindlimb) a e ini ially
labeled in FoxA2C eERT/+;Toma ofl/+mice ( ea ed wi h amoxi en
P14-P18), bu his numbe dec eases significan ly ( o 41
FoxA2+clones/hindlimb) by 3 mon hs a e he ini ial labeling
(Fig. 4(C3), Supplemen a y Fig. 6C). In con as , in FoxA2C eERT/+;
Toma ofl/+mice ( ea ed wi h amoxi en P28-P37) he numbe o
FoxA2+clones is no significan ly changed be ween 1-day
(55 FoxA2+clones/hindlimb) and 3 mon hs (57 FoxA2+clones/
hindlimb) a e he ini ial labeling (Fig. 4C4, Supplemen a y
Fig. 6C). These esul s aise he possibili y ha FoxA2+cells
labeled a e P28 emain mos ly in GP, and hei con ibu ion o
he SOC slows down significan ly o e ime.
0
20
40
60
Colonies om FoxA2+ (Zsg een+) s em cells
Colonies om PTHRP+ (mche y+) s em cells
c u assay o FoxA2+ cells s PTH P+ cells
B
FoxA2 C e.ERT/+ ZsG een /+ PTH P mche y mice
PTH P+
FoxA2+
a
P oximal ibia a
Dis al ulna
PTH P+
FoxA2+
ab
a1 b1
a1 b1
A
P18
Tam Ha es
P13-17
a
Dis al ulna
P40
Tam Ha es
P13-17
PTH P+
FoxA2+
c1
c1
FoxA2+
PTH P+
Passage 1
Passage 9
G ow h
media
Adipogenic
media
Os eogenic
media
Chond ogenic
media
T ilineage di e en ia ion assay o FoxA2+ cells
abc d
e gh
C
c
d. Ha es a P18
P1 P2 P3 P4 P5 P6 P7 P8 P9
Passage numbe
Colonies / 1000 cells
10
102 103104 105 106
102 103104 105 106
ZsG een
mChe y
102 103104 105 106
mChe y
10
10
10
102 103104 105 106
ZsG een
e. Ha es a P40
1.579% 0.015%
0.681%
2.445% 0.108%
0.396%
Fig. 3 FoxA2+cells a e long- e m s em cells, and hey a e dis inc om he sho - e m PTH P+s em cells. A Tibia (a) and ulna (b) omFoxA2C eERT2/+;
ZsG eenfl/+;PTH Pmche y mice ea ed wi h amoxi en (Tam) P13-P17 and ha es ed a P18, and om FoxA2C eERT2/+;ZsG eenfl/+;PTH Pmche y mice ea ed wi h
amoxi en P13-P17 and ha es ed a P40 (c). Flo escence mic oscopy o ZsG een (g een), mche y ( ed), Hoechs (blue). Scale ba s, 50 µm. Sepa a ion o
FoxA2+(g een) cells om PTH P+( ed) cells and double posi i e FoxA2+PTH P+(yellow) cells by Fluo escence-ac i a ed cell so ing (FACS) o GP cells
isola ed om FoxA2C eERT2/+;ZsG eenfl/+;PTH P
mche y mice, injec ed wi h amoxi en om P13 o P17, and ha es ed a P18 (d)o P40(e). BColony o ming uni
(c u) assay o FoxA2+(ZsG een+) cells ( om FoxA2C eERT2/+;ZsG eenfl/+;col.10mche y mice ea ed wi h amoxi en P13-P17 and ha es ed a P18) and PTH P+
(mche y+)cells ( om PTH Pmche y mice ha es ed a P18). Quan ifica ion o he numbe o colonies (pe 103cells) p esen ed as mean ± SD om n=3
expe imen s. CT ilineage di e en ia ion o FoxA2+(ZsG een+) clones, om passage P1 (a–d) and passage P9 (e–h). Fou independen clones, o each
passage, we e es ed in ei he g ow h media (a,e), o adipogenic (b, ), os eogenic (c,g) and chond ogenic (d,h) media. Scale ba s, 100 µm.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1 ARTICLE
NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions 5

Indeed, o FoxA2C eERT2/+;Toma ofl/+mice ea ed wi h
amoxi en P14 –P18, and ha es ed a P19, a 1-day chase e eals
ha 137 cells o he FoxA2+cells a e in he GP and 43 in he
SOC (Supplemen a y Fig. 6A1). A 1-mon h chase inc eases he
SOC popula ion, om 43 o 253, sugges ing ha FoxA2+cells
can con ibu e o he epiphyseal bone, du ing SOC os eogenic
ma u a ion (Supplemen a y Fig. 6A1). This con ibu ion is no
longe endochond al, as RZ FoxA2+cells no longe exp ess
col.10 by P21 (Supplemen a y Fig. 6B). Ins ead, FoxA2+cells a e
os eo-p ogeni o s ha suppo di ec ly he os eogenic lineage.
This is consis en wi h ou p e ious findings demons a ing ha
FoxA2+cells labeled du ing ea ly pos na al de elopmen ga e
ise o col.1+p ogeny in he SOC (Fig. 2B). In addi ion, as
p e iously shown, he FoxA2+cells labeled P14 o P18 gi e ise
o clus e s and inc easing columns o p ogeny o e ime,
sugges ing chond o-p ogeni o quali y (Fig. 4A). Al oge he ,
hese findings highligh dual os eo-chond o-p ogeni o a e o
FoxA2+cells labeled p io o P28.
In con as , FoxA2+cells labeled a e P28 emain mos ly in
GP, and hei con ibu ion o he SOC slows down (Supplemen-
a y Fig. 6A2). Fo olde mice, 1-day chase e eals ha 55
FoxA2+cells a e in he GP and 10 FoxA2+cells in he SOC
abo e (Supplemen a y Fig. 6A2). A 1-mon h chase shows a
simila numbe o FoxA2+cells (23) in he SOC, sugges ing ha
Ha es : 1day
ab dgjc e hi kl
Lineage acing FoxA2C e.ERT/+Toma o /+ mice, Tamoxi en s a ing P28
Quan i ica ion Toma o+ cells, clus e s, columns
Ha es : 1 mon h Ha es : 3 mon hs Ha es : 9 mon hs
j
g
d
ba
B1 B2 B3 B4
B
k
ab
b
a
c
ce
d
ig
gi
ed
h
A1 A2 A3
jkl
lj
k
A4
Ha es : 1day
Lineage acing FoxA2C e.ERT/+ Toma o /+ mice, Tamoxi en a P14
Ha es : 1 mon h Ha es : 3 mon hs Ha es : 9 mon hs
A
C
FoxA2C e.ERT2; Toma o /+
Tamoxi en a P14
FoxA2C e.ERT2; Toma o /+
Tamoxi en a P28
# FoxA2+clones / HL
# FoxA2+clones / HL
C3
Numbe o FoxA2+ clones
single cells clus e s columns
0
40
80
120
160
1d 1mo 3mo 9mo
C4
Pe cen age (%) o FoxA2+ clones
FoxA2C e.ERT2; Toma o /+
Tamoxi en a P14
FoxA2C e.ERT2; Toma o /+
Tamoxi en a P28
% FoxA2+clones / HL
% FoxA2+clones / HL
C1
single cells clus e s columns
C2
0
40
80
120
1d 1mo 3mo 9mo
0
40
80
120
1d 1mo 3mo 9mo
0
40
80
120
160
1d 1mo 3mo 9mo
Tam Ha es
P14-18 P19
Tam Ha es
P14-18 P48
Tam Ha es
P14-18 P108
Tam Ha es
P14-18 P288
Tam Ha es
P28-37 P38
Tam Ha es
P28-37 P67
Tam Ha es
P28-37 P127
Tam Ha es
P28-37 P307
Fig. 4 Dual os eo-chond o-po en ial o FoxA2+cells in he ea ly weeks o pos na al de elopmen (P14) and s onge chond ogenic po en ial beyond
P28. A Tibia sec ions ha es ed om FoxA2C eERT2/+;Toma ofl/+mice ea ed wi h 5x amoxi en (Tam) injec ions s a ing a P14, and ha es ed a 1 day
(A1), 1 mon h (A2), 3 mon hs (A3), and 9 mon hs (A4) a e he las injec ion. Hoechs (g ay), 5-DTAF (g een). Ba s, 100 µm. Rep esen a i e de ails om
GP sec ions, in numbe ed inse s (a–l). BTibia sec ions ha es ed om FoxA2C eERT2/+;Toma ofl/+mice ea ed wi h 10x amoxi en injec ions s a ing a
P28, and ha es ed a 1 day (B1), 1 mon h (B2), 3 mon hs (B3), and 9 mon hs (B4) a e he las injec ion. Hoechs (g ay), 5-DTAF (g een). Ba s, 100 µm.
Rep esen a i e de ails om GP sec ions, in numbe ed inse s (a–l). CQuan ifica ion o he numbe (C3,C4) and he pe cen age (C1,C2) o single cells,
clus e s, columns o med a 1 day, and 1, 3, 9 mon hs a e he las injec ion. Pe cen age (%) single cells is calcula ed by he numbe o single Toma o+
cells in he GP (ou side he g een 5-DTAF domain) o he o al numbe o Toma o+uni s (single cells, clus e s and columns). Each poin ep esen s he
sum o n=8 sec ions pe mouse hindlimb. Simila quan ifica ion o pe cen age (%) clus e s and columns. One Toma o+uni , is defined by 1 single cell, o
1 clus e comp ised o wo o mo e cells g ouped oge he , o by 1 column o cells.
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1
6NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions
FoxA2+cells con ibu ion o he epiphyseal bone slows down
significan ly a e P28 (Supplemen a y Fig. 6A2).
Sel - enewabili y o FoxA2+cells in a econs i u ion assay.To
demons a e sel - enewal and mul ipo en capabili ies o FoxA2+
cells in i o, we pe o med a se ial ansplan a ion assay. We
isola ed FoxA2+(ZsG een+) cells om FoxA2C eERT2/+;
ZsG eenfl/+;Tg.col10mChe y mice, as p e iously desc ibed, labeled
hem wi h DiD fluo escen dye, and ansplan ed hem
subcu aneously in o ZsG een- li e ma es (Fig. 5A, Supplemen-
a y Fig. 7A). A e 1 mon h, FoxA2+(ZsG een+) cells om he
g a s exhibi ed sel - enewabili y and e ained colony- o ming
abili y (Fig. 5B(a)). In addi ion, he g a ed cells ha e unde gone
mul iple ounds o cell di ision shown by a log- old educed
fluo escence, obse ed as a shi in he in ensi y his og am plo s
(Fig. 5B(c)). Peak DiD in ensi y a day 0 (D0), immedia ely be o e
ansplan a ion, was 1.2 × 104, and i was educed o 1.1 × 103on
day 30 (D30) pos - ansplan ion (Fig. 5B(c)). The pe cen age o
ZsG een+cells wi h high in ensi y DiD fluo escence (DiDhigh)
Se ial ansplan a ion o FoxA2+ cells in i o
A
P ima y T ansplan
a1 a2
a3
1
2
3
1
23
b2
b3
b1
Isola ion and assessmen o sel - enewabili y po en ial o FoxA2+ cells
B
ab
Seconda y T ansplan
FACS
sepa a ion
Label wi h DiD Fi s
T ansplan
Explan
a e
1mo.
In i o colony
o ma ion
and expansion
Label again wi h
DiD dye
Second
T ansplan
In i o ZsG +colony
o ma ion (a e i s
ansplan )
Chond ocy e
isola ion
Take cells o
DiD analysis
by FACS
(Day 30 i s
ansplan )
Take cells o
DiD analysis by
FACS (Day 0
second
ansplan )
Explan
a e
1 mo.
Take cells o
DiD analysis by
FACS (Day 30
second
ansplan )
In i o ZsG +colony
o ma ion (a e
second ansplan )
T ansplan a ion
o p imed
p ogeni o s
105 cells
Take cells o
DiD analysis
by FACS
(Day 0 i s
ansplan )
ZsG –li e ma e
Seconda y ecipien
ZsG –li e ma e
Te ia y ecipien
FoxA2+Col10-
His ological
analysis o
explan s a
1mo.
Lineage
p iming
FoxA2C e.ERT/+ ZsG een /+
Tg.col10mche y mice
P ima y hos
Chond ogenic
Adipogenic
Os eogenic
***
0
20
40
60
80
100
***
DiDhigh
D0 D30
% ZsG een+ cells
0
20
40
60
80
100
DiDlow
D0 D30
% ZsG een+ cells
0
20
40
60
80
100
0
20
40
60
80
100
***
***
D0 D30
% ZsG een+ cells
D0 D30
ZsG een+ cells coun
DiDhigh DiDlow
de g
h
ZsG een+ cells coun
c
100
50
0
100
50
0
100
50
0
100
50
0
DiD in ensi y
101 103105107
DiD in ensi y
101 103105107
100
50
0
% ZsG een+ cells
100
50
0
Fig. 5 Se ial ansplan a ion o FoxA2+cells in i o. A Expe imen al s a egy o he isola ion and assessmen o sel - enewabili y and di e en ia ion
po en ial o FoxA2+cells in i o. BRep esen a i e images o colonies de i ed om FoxA2+cells isola ed om p ima y (a) and seconda y ansplan s (b).
Scale ba s, 1 cm. DiD fluo escence in ensi y his og am p ofiles a Day 0 ( ed), Day 30 (blue), and nega i e con ol (black) om p ima y (c) and seconda y
( ) ansplan s. Pe cen age o DiDhigh cells in he FoxA2+cells om p ima y (d) and seconda y (g) ansplan s. Pe cen age o DiDlow cells in he FoxA2+
cells om p ima y (e) and seconda y (h) ansplan s. Da a a e p esen ed as mean ± SD o each g oup, n=3 mice. The as e isks indica e significan
di e ence: wo- ailed S uden ’s unpai ed samples es ; ***p=7.0148E−07 (DiDhigh p ima y ansplan D0 s D30), 7.84512E−06 (DiDlow p ima y
ansplan D0 s D30), 3.32176E−06 (DiDhigh seconda y ansplan D0 s D30), and 3.46695E−06 (DiDlow seconda y ansplan D0 s D30). Comple e
s a is ical in o ma ion is p o ided in Supplemen a y Table 1.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1 ARTICLE
NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions 7
was 75 ± 1% p io o ansplan a ion (D0), bu i was educed o
27.3 ± 1.1% in he cells de i ed om g a s a D30 pos -
ansplan a ion (Fig. 5B(d)). Simul aneously, he pe cen age o
cells exhibi ing low in ensi y DiD fluo escence (DiDlow) was
significan ly inc eased om 20.3 ± 0.6% a D0 o 42.3 ± 1.1% a
D30 (Fig. 5B(e)). This sugges ha FoxA2+(ZsG een+) cells
p oli e a e in he p ima y ansplan s and dilu e he DiD fluo -
escen dye.
Nex , FoxA2+(ZsG een+)cells om he p ima y g a we e
s ained wi h DiD dye and ansplan ed in o he e ia y ecipien
mice (Fig. 5A, Supplemen a y Fig. 7A). A simila educ ion in
DiDhigh cell popula ion ( om 75.3 ± 1.5% on D0 o 29.7 ± 1.5%
on D30) and a p opo iona e inc ease in DiDlow cell popula ion
( om 19.7 ± 0.6% on D0 o 46.7 ± 1.1% on D30) we e consis en
wi h he p ope ies obse ed wi h cells de i ed om p ima y
g a s (Fig. 5B( , g, h)). In addi ion, FoxA2+(ZsG een+) cells
om he seconda y g a s e ain colony o ming abili y
(Fig. 5B(b)).
The cells collec ed a he end o se ial ansplan a ion exhibi ed
mul i-lineage di e en ia ion capabili ies, as indica ed by posi i e
Alcian blue, LipidTox o Aliza in Red-S s aining in c oss-sec ions
o e ia y g a s (Supplemen a y Fig. 7B(a–c)). These e ia y g a s
we e p oduced by ansplan a ion o lineage-p imed (chond o-
genic, adipogenic, os eogenic) second g a -de i ed FoxA2+(Zs-
G een+) cells (Fig. 5A, Supplemen a y Fig. 7B(a–c)). In con as ,
second g a -de i ed FoxA2+(ZsG een+) cells, wi hou lineage
p iming, p oduced e ia y g a s ha we e nega i e o cell
di e en ia ion (Fig. 5A, Supplemen a y Fig. 7B(d– )). Al oge he ,
hese expe imen s indica e ha FoxA2+cells p ese e bo h hei
sel - enewing abili y and hei mul ipo ency du ing in i o se ial
ansplan a ion.
GP ca ilage egene a ion a e SH1-like inju y. To in es iga e
whe he FoxA2+cells expand and epai he GP ca ilage in
esponse o auma, we de eloped a mu ine model o GP inju y.
Sal e –Ha is ype1 ac u es (SH1), which in ol e sepa a ion ac oss
he en i e physis, h ough he HZ, ha e he bes p ognosis o epai
and a e he leas likely o cause g ow h a es 3,4. We ep oduced his
clinical obse a ion and de eloped a SH1-like su gical model, o es
i FoxA2+cells could epai he GP ca ilage. A 30 g needle is
passed in he ans e se plane, h ough he HZ o he p oximal ibial
physis, c ea ing a disc e e de ec , isualized by Con as -Enhanced
Compu ed Tomog aphy (CECT)17,18 (Fig. 6A, B). Quan i a i e
analysis o sequen ial sagi al planeimageses ima es ha hede ec
acks, on a e age, 2216 ± 195.6 μm medial o la e al, 280 ± 60 μm
an e io o pos e io , 165.6 ± 36.9 μm heigh , and 28.6° ± 5.5 angle
ela i e o mid-co onal plane (Supplemen a y Fig. 8A, B). Colo
maps o ep esen a i e CECT images o con ol and ope a ed ibias
e eal ha inju y si e heals wi hin 7 days, i espec i e o mouse
gende (Fig. 6C). The inju y gap olume dec eases significan ly om
0.29 ± 0.07 mm3a 1 day, o 0.04 ± 0.04 mm3, close o backg ound
le el, a 7-day pos -op (Fig.6D).O e alllongi udinalg ow hwasno
a ec ed a 7- and 21-day pos -op (Supplemen a y Fig. 8C).
A 1-day pos -op, he inju ed GP displays significan damage in
he HZ (Fig. 6E(a, b)). In addi ion, gi en ha he cen al egion is
mo e cu ilinea , wi h a sho e GP in he middle, he inju y
causes mo e damage in his a ea, des oying he HZ, he PZ, and
mos o he RZ. A 1-day pos -op, he heal hy ca ilage a ea is
42% o he o al measu ed a ea (heal hy plus inju ed) (Fig. 6F).
Wi hin days, he de ec shows healing and he ca ilage is
egene a ing. By he 3 d day, he cen al gap is closing in, and he
heal hy ca ilage a ea is up o 63% ((Fig. 6E(c, d), F). By he 7 h
day, he GP is 96% egene a ed, displaying all h ee laye s (RZ,
PZ, HZ) ((Fig. 6E(e, ), F). The inju ed GP healed wi h physeal
ca ilage, no fib oca ilage o bone.
FoxA2+cells expand and egene a e he GP a e SH1-like
inju y. A e SH1-like su ge y on FoxA2C eERT/+;Toma ofl/+
mice, he e was subs an ial expansion o Toma o+cells in he
inju ed GP (Fig. 7A, B, Supplemen a y Fig. 9). We quan ified he
numbe o FoxA2+cells in GP ca ilage as well as in neighbo ing
me aphyseal bone (MB) and epiphyseal bone (SOC) (Fig. 7B). A
1-day pos -op, he e was no di e ence be ween GP FoxA2+cells
in con ol e sus ope a ed mice (Fig. 7A(a, d), Fig. 7B(b1)).
Howe e , he e is a significan loss (94%) o FoxA2+cells a he
HZ/MB bo de , which con inues o be sus ained 3-day pos -op
(Fig. 7B(b3)). This indica es ha he e is no expansion o he MB
Toma o+cells in esponse o inju y. Simila ly, he e is no
expansion o he Toma o+cells in he SOC a 1 day o 3 days
a e SH1-like su ge y (Fig. 7B(b2)). Howe e , he e is a 2.7- old
expansion o Toma o+cells in he GP 3 days a e he SH1-like
su ge y (Fig. 7B(b1)).
These findings a e suppo ed by a p oli e a ion assay, which
demons a es ha Toma o+cells a e B dU+(Supplemen a y
Fig. 10A). This expansion o FoxA2+cells a 3 days pos -op is he
esul o FoxA2+cells labeled p io o su ge y who a e
mul iplying in esponse o inju y, and no esh labeling o
FoxA2+cells due o de no o exp ession o FoxA2 in he inju ed
ca ilage emnan s. Immunohis ochemis y o FoxA2 in he
inju ed GP ca ilage shows he exp ession o FoxA2 in i s
es ablished RZ domain, bu no in he ca ilage emnan s a d1,
d2 o d3 pos -op (a–c) (Supplemen a y Fig. 10B).
By 7-day pos -op, FoxA2+(Toma o+) cells a e significan ly
educed in he GP issue, and in pa allel, we see a su ge in
Toma o+cells in he MB (3- old inc ease) (Fig. 7B(b3)). This
sugges s ha he Toma o+cells, ha ha e expanded in he GP
issue a 3-day pos -op, ansi ed below in he bone spongiosa by
7-day pos -op (Fig. 7A( 2)). In he cen e , whe e he wide gap was
now educed o a hin sli , he las Toma o+cells a e mig a ing
down, aligned in sho (4–8 cell) s acks o clones (Fig. 7A( 1)).
FoxA2+cells a e needed o ca ilage egene a ion. To de e -
mine whe he FoxA2+cells pa icipa e in egene a ing he inju ed
GP issue ia mul i-lineage di e en ia ion, we asked i FoxA2+cells
di e en ia e in o col.10+cells (chond ogenic lineage) o col.1+cells
(os eogenic lineage). SH1-like su ge y on FoxA2C e.ERT/+;ZsG een /+;
Tg.col10mche y mice demons a es ha FoxA2+(ZsG een+)cells
gene a e yellow hype ophic cells (col.10mche y+)(Fig.8A(b)),
while SH1-like su ge y on FoxA2C e.ERT/+;Toma o / ;Tg.col1GFP mice
shows ha FoxA2+(Toma o+) cells ansi in he me aphyseal bone
o become yellow bone cells (col.1GFP+)(Fig.8A(a)).
To inqui e whe he FoxA2+cells a e needed o ca ilage
egene a ion, we gene a ed FoxA2C eERT/+;DTA / mice, in which
FoxA2+cells we e abla ed by p oduc ion o diph he ia oxin
(DTA). In FoxA2C eERT2/+;DTA /+;Toma ofl/+mice he e was a
36% dec ease in FoxA2+(Toma o+) cells compa ed wi h
FoxA2C eERT2/+;Toma ofl/+li e ma es (Fig. 8B). Loss o FoxA2+
cells lead o a delay in ca ilage closu e in FoxA2C eERT2/+;DTA /
mice 7-day pos -op (Fig. 8C(b1, d1)). Ca ilage egene a ion in
FoxA2C eERT2/+;DTA / mice was significan ly educed (down o
72%) compa ed wi h con ol mice (Fig. 8C, D), sugges ing ha
FoxA2+cells a e necessa y o GP epai .
Summa y
In summa y, we demons a e ha a popula ion o FoxA2+col10-
cells, esidual om epiphyseal ca ilage pos na al de elopmen ,
con ibu es o he op sub-compa men o he RZ (Fig. 9). Based on
hei loca ion, FoxA2+cells a e physically sepa a ed om he
PTH P+cells, which a e loca ed a he bo om o he RZ, and
exhibi highe clonogenici y and longe i y han PTH P+cells.
Sho - e m pulse-chase o he FoxA2+cells gene a es dyads/clus e s
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1
8NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions
o cells, whe eas long- e m pulse-chase gene a es inc easing col-
umns o p oli e a ing p ogeny, sugges ing ha FoxA2+cells may
fi s gi e ise o p ogeny ha subsequen ly di e en ia e in o col-
umns o chond ocy es. P io o P28, FoxA2+cells ha e a dual
os eo-chond o-p ogeni o a e, and a e P28, FoxA2+LTSSC
emain mos ly in GP, while hei con ibu ion o he SOC winds
down significan ly. In a mu ine model o GP inju y, FoxA2+cells
expand in esponse o auma and pa icipa e in p oduc ion o
physeal ca ilage allowing o success ul egene a ion. In conclusion,
FoxA2+cells a e a LTSSC popula ion necessa y o bo h GP
u no e and ca ilage egene a ion ollowing inju y.
Discussion
While POC (p ima y ossifica ion cen e s) o he long bones o m
a E15.5, du ing mouse emb yonic de elopmen , he SOCs appea
some ime la e , p ima ily a e bi h, a ound pos na al day
P719,20. Chond ocy e hype ophy (P7-P8) is ollowed by
RZ
PZ
HZ
SH1 – like GP inju y model CECT imaging o he inju y wi h CA4+
AB
Con ol + CA4+Su ge y + CA4+
Con ol - CA4+
1a 2a 3a
1b 2b 3b
CECT imaging o GPC egene a ion
.
C
**
Inju y gap olume (mm3)
P=0.055
d1 d3 d7
d1
d3
d7
sopsyad-op
Su ge y
Con ol
Quan i ica ion ca ilage olume
D
0.0
0.1
0.2
0.3
0.4
0.5
Con ol Su ge y
A
B
d1 d3 d7
d1
d3
d7
1
2
3
ab
d
c
e
aa.
SO/FG in con ol s. ope a ed mice
sopsyad-op
% GPC egene a ion
% SO+ a ea (A) s
o al a ea (A+B)
1
2
3
E
*
Image J quan i ica ion
F
**
***
0
20
40
60
80
100
-1000 Hounds ield Uni s (HU) 5000
100
80
60
40
20
0
Fig. 6 A mu ine model o GP inju y. A Schema ics o Sal e –Ha is ype 1 (SH1)-like inju y model. BCECT (con as -enhanced compu ed omog aphy)
imaging o he SH1-like inju y, 1-day a e su ge y. Rep esen a i e sagi al (1a,2a,3a) and ans e se (1b,2b,3b)μCT slices o he p oximal ibia, a 1-day
pos -su ge y, be o e (1a,1b) and a e (2a,2b,3a,3b) exposu e o CA4+. Scale ba s, 500 µm. CColo maps o ep esen a i e con as -enhanced μCT
images o con ol and ope a ed ibias a 1, 3, and 7 days pos -ope a ion, iewed om he ans e se plane. The colo scale co esponds o Hounsfield Uni s
(HU). Fo e e ence, ai a enua es a −1000 H and wa e a 0 HU. Scale ba s, 500 µm. DQuan ifica ion o he inju y gap olume using CECT imaging.
Da a a e p esen ed as mean ± SD, n=5 mice o each ime poin (d1,d3,d7)). The as e isks indica e significan di e ence: Two- ailed S uden ’s unpai ed
samples es ; **p=0.0079 (d1 s d7). ESa aninO/Fas G een s aining o con ol and ope a ed ibias, 1 day (a,b), 3 days (c,d), and 7 days (e, ) a e
SH1-like su ge y. Inse , cen al a ea o he de ec (1–3). Scale ba s, 100 µm. FImageJ quan ifica ion o he inju y. A is heal hy ca ilage a ea (Sa aninO+), B
is inju ed a ea (Sa aninO−), and A+B is he o al a ea. Pe cen age (%) ca ilage egene a ion is A/A+B. Da a a e p esen ed as mean ± SD, n=3 mice.
The as e isks indica e significan di e ence: Two- ailed S uden ’s unpai ed samples es ; ***p=0.000839 (d1 s d7), **p=0.007121 (d3 s d7),
*p=0.037861 (d1 s d3). Comple e s a is ical in o ma ion is p o ided in Supplemen a y Table 1.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-30247-1 ARTICLE
NATURE COMMUNICATIONS | (2022) 13:2515 | h ps://doi.o g/10.1038/s41467-022-30247-1 | www.na u e.com/na u ecommunica ions 9