ARTICLE
Recei ed 8 Sep 2015 |Accep ed 15 Jun 2016 |Published 4 Aug 2016
No mal s oma supp esses cance cell
p oli e a ion ia mechanosensi i e egula ion
o JMJD1a-media ed ansc ip ion
Riina Kaukonen1, Anja Mai1, Ma ia Geo giadou1, Ma kku Saa i1, Nicola De F anceschi1, Timo Be z2, Ha i Sih o3,
Sami Ven ela
¨1,4, Lau a Elo1,5, Eija Joki alo6, Jukka Wes e ma ck1, Pi kko-Liisa Kellokumpu-Leh inen7,
Heikki Joensuu3,8, Reida G enman4& Johanna I aska1,9
Tissue homeos asis is dependen on he con olled localiza ion o specific cell ypes and he
co ec composi ion o he ex acellula s oma. While he ole o he cance s oma in
umou p og ession has been well cha ac e ized, he specific con ibu ion o he ma ix i sel
is unknown. Fu he mo e, he mechanisms enabling no mal—no cance —s oma o p o ide
umou -supp essi e signals and ac as an an i umo igenic ba ie a e poo ly unde s ood. He e
we show ha ex acellula ma ix (ECM) gene a ed by no mal fib oblas s (NFs) is so e han
he CAF ma ix, and i s physical and s uc u al ea u es egula e cance cell p oli e a ion.
We find ha no mal ECM igge s down egula ion and nuclea exi o he his one
deme hylase JMJD1a esul ing in he epigene ic g ow h es ic ion o ca cinoma cells.
In e es ingly, JMJD1a posi i ely egula es ansc ip ion o many a ge genes, including
YAP/TAZ (WWTR1), and he e o e gene exp ession in a s i ness-dependen manne . Thus,
no mal s omal es ic s cance cell p oli e a ion h ough JMJD1a-dependen modula ion o
gene exp ession.
DOI: 10.1038/ncomms12237 OPEN
1Cen e o Bio echnology, Uni e si y o Tu ku, 20520 Tu ku, Finland. 2Ins i u e o Cell Biology, Cen e o Molecula Biology o Inflamma ion, 48149
Muens e , Ge many. 3Labo a o y o Molecula Oncology, T ansla ional Cance Biology P og am, Uni e si y o Helsinki, 00290 Helsinki, Finland. 4Depa men
o O o hinola yngology, Head and Neck Su ge y, Tu ku Uni e si y and Tu ku Uni e si y Hospi al, 20521 Tu ku, Finland. 5Depa men o Ma hema ics and
S a is ics, Uni e si y o Tu ku, 20520, Tu ku, Finland. 6Ins i u e o Bio echnology, Elec on Mic oscopy Uni Uni e si y o Helsinki, 00014 Helsinki, Finland.
7Depa men o Oncology, Tampe e Uni e si y Hospi al, 36280 Tampe e, Finland. 8Depa men o Oncology, Helsinki Uni e si y Cen al Hospi al, 00029
Helsinki, Finland. 9Depa men o Biochemis y and Food Chemis y, Uni e si y o Tu ku, 20520 Tu ku, Finland. Co espondence and eques s o ma e ials
should be add essed o J.I. (email: johanna.i aska@u u.fi).
NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions 1
In no mal issue, di e en cell ypes a e spa ially confined.
Fo example, he no mal epi helium is sepa a ed om
he unde lying s omal ex acellula ma ix (ECM) by he
basemen memb ane, and hus hey a e no in a di ec con ac
wi h each o he . The s omal ECM is p oduced by he fib oblas s
and se es as an impo an egula o o issue homeos asis1.
In con as , in asi e ca cinomas con ain a complex mix u e
o umou cells and s omal componen s whe e he cance
cells in e ac wi h he al e ed ECM and a e also embedded in i .
Changes in he ma ix s i ness as well as ansi ion be ween wo-
dimensional and h ee-dimensional ma ix con ac s influence cell
p oli e a ion p o oundly2–5. Thus, cance p og ession in ol es
no only gene ic al e a ions o cance cells bu also changes in he
umou mic oen i onmen 6. The ole o he s oma as a po en
an i umo igenic ba ie has al eady been elucida ed o e wo
decades ago7,8. The exposu e o ca cinoma cells o no mal
basemen memb ane-like componen s has he capaci y o e e
b eas cance cells o a nea -no mal pheno ype9. Fu he mo e,
in cocul u e sys ems, no mal fib oblas s (NFs) can inhibi
he g ow h o ce ain cance cells10,11, while a popula ion o
ans o med cance -associa ed fib oblas s (CAFs) can e e se he
g ow h-inhibi ing e ec o NFs12. Thus, c oss- alk be ween
epi helial cells and fib oblas s is a c i ical ea u e o cance
p og ession. Howe e , he specific con ibu ion o he ma ix
i sel has no been add essed in de ail.
Tumou s oma a a e cha ac e ized by inc eased issue s i ness
and al e ed ma ix a chi ec u e ha a ou p oli e a ion, me as asis
and d ug esis ance ia abe an mechanosignalling13–15. S i ness
also ac i a es he main media o s o mechano ansduc ion:
ansc ip ion ac o s YAP (Yes-associa ed p o ein) and TAZ
( ansc ip ional coac i a o wi h PDZ-binding mo i )16.
Epigene ic egula ion is also implica ed in cance p og ession as
i p o oundly egula es he ansc ip ion p ofile and pheno ype
o cells. Many lines o e idence sugges ha deme hyla ion o
ep essi e his one me hyla ion ma ks, such as his one H3 lysine 9
(H3K9), om he gene p omo e s by his one deme hylases is a
p e equisi e o ansc ip ional ac i a ion17–19. JMJD1A (KDM3A)
deme hyla es monome hyl and dime hyl his one H3K9 in i o
and in i o and has been implica ed as a posi i e egula o o
ansc ip ion o se e al g ow h-p omo ing genes17–19.
While he mechanisms whe eby cance s oma and CAFs
con ibu e o umou p og ession a e being ac i ely in es iga ed,
much less is known abou how he no mal s oma exe s umou -
supp essi e signals o con ol issue homeos asis. Fu he mo e, he
ole o epigene ic egula o s in he abili y o he cells o espond o
he s i ness o he umou mic oen i onmen is no known. To
in es iga e his, we compa ed he abili y o ma ices gene a ed by
NFs o CAFs om he same pa ien as well as ma ices om
immo alized NFs o influence cance cell p oli e a ion and gene
exp ession. We find ha he ma ix gene a ed by NFs, bu no CAF
ma ix, p o oundly inhibi s cance cell p oli e a ion h ough
mechanosensi i e down egula ion o he his one deme hylase
enzyme JMJD1a.
Resul s
No mal ma ix inhibi s cance cell p oli e a ion. Fib oblas s
p oduce in i o a obus cell-de i ed ma ix (CDM) ha
ecapi ula es many ea u es o he a chi ec u e and composi ion
o in i o ECM20. To in es iga e he po en ial e ec o ma ix
on cance cell p oli e a ion, we gene a ed ma ices om
elome ase-immo alized NFs (TIFFs) (Fig. 1a) and es ed hei
e ec i eness on wo highly p oli e a i e and widely s udied
cance cell lines, namely ce ical cance HeLa and b eas cance
MDA-MB-231 cells. Rema kably, bo h o hese cell lines we e
significan ly g ow h-inhibi ed (Fig. 1b) by he no mal ma ix
compa ed wi h s anda d g ow h condi ions on plas ic. The
g ow h- es ic i e p ope ies o CDM we e only obse ed wi h
he in ac CDM as ma ix p o eins such as fib onec in o
collagen I, o solubilized and e-pla ed CDM did no inhibi
MDA-MB-231 cell p oli e a ion (Supplemen a y Fig. 1a,b).
Soluble ac o s we e no implica ed ei he because cul u ing
he MDA-MB-231 cells in condi ioned TIFF medium did
no influence p oli e a ion (Supplemen a y Fig. 1c). Thus, only
he a chi ec u ally in ac CDM possessed g ow h-inhibi o y
p ope ies.
When in es iga ing ma ix-induced e ec s in mo e de ail we,
a he unexpec edly, obse ed ha g ow h inhibi ion induced by
TIFF CDM was main ained in cance cells ollowing de achmen
om he ma ix by ypsiniza ion and epla ing on plas ic
(Fig. 1c). E en hough he ma ix-exposed cells we e e u ned o
plas ic in ull se um-con aining medium, bo h MDA-MB-231 and
HeLa cells con inued o p oli e a e significan ly slowe han he
same cance cell lines cul u ed con inuously on plas ic.
Thus, exposu e o cance cells o CDM om NFs is no only
g ow h-inhibi o y bu has he po en ial o e e malignan
cance cell p oli e a ion in a sus ained manne .
We obse ed ha cance cells g own on no mal TIFF-de i ed
CDM had significan ly al e ed cell mo phology compa ed wi h
cells on plas ic (Fig. 1d and Supplemen a y Fig. 1d). This was
in e es ing as mechanical cues and en i onmen al s i ness a e
known o a ec he cy oskele on and nuclea unc ions including
ch oma in condensa ion and global epigene ic s a us o a cell21–25,
and he e o e changes in cell mo phology and gene exp ession
could explain he ma ix-dependen e e sion in he cance cell
pheno ype.
Ma ix induces gene exp ession changes. We hypo hesized ha
exposu e o CDM could induce changes in epigene ic modifie s,
hence supp essing cance cell g ow h in a sus ained manne . To
in es iga e his possibili y, we pe o med Illumina whole-genome
ansc ip ion analysis in MDA-MB-231 and HeLa cells ha es ed
di ec ly om TIFF CDMs a e 6 days (CDM), ollowing
de achmen om CDMs and epla ing on plas ic o 5 days
(CDM o plas ic) o in cells g own con inuously on plas ic
(Fig. 1e). As CDM induced sus ained g ow h inhibi ion in bo h
cell lines, we ocused ou a en ion on common ansc ip ional
al e a ions o epigene ic enzymes. A single well-cha ac e ized
his one deme hylase JMJD1a was significan ly down egula ed in
bo h cell lines, sugges ing ha his gene migh be linked o he
sus ained pheno ypic al e a ions igge ed by he CDM in bo h
cance cell lines. Among he o he significan ly al e ed genes
we e signalling p o eins SORBS2 and PDE7B, which we e
down egula ed in bo h cell lines and in bo h condi ions (CDM
and CDM o plas ic; Fig. 1 and Supplemen a y Da a 1). In
addi ion, exp ession o 27 genes on CDM and 150 genes
ollowing CDM de achmen we e al e ed significan ly in bo h cell
lines. Se e al ma ix-modi ying p o eins (MMP3, PLAUR1 and
COL1A2) we e up egula ed on he ma ix in bo h cell lines, while
many genes in ol ed in egula ing choles e ol syn hesis
(IDI1,ACAT1 and HMGCS1) we e down egula ed on he ma ix.
Following ma ix de achmen , se e al hypoxia- ela ed genes we e
down egula ed (JMJD1A (KDM3A),ALDOC,DDIT4,GDF15,
ANG and MUC1) e en hough he a he hin CDM laye is
unlikely o es ic oxygen di usion in he in i o cul u es26.In
addi ion, MYC- a ge genes (FAPB5 and NME1) we e
up egula ed ollowing CDM de achmen and epla ing on
plas ic. Since he mechanism o sus ained g ow h inhibi ion was
o p ima y in e es , we ocused specifically on he epigene ic
modifie enzyme JMJD1a ha was down egula ed ollowing
ma ix de achmen in bo h cell lines (Fig. 1 ).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237
2NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions
No mal ma ix es ains JMJD1a exp ession. JMJD1a is a
H3K9-specific deme hylase ha has been linked o se e al
biological p ocesses including g ow h, de elopmen and
ep og amming17,27–29. Loss o JMJD1a dec eases gene
exp ession and inc eases he inhibi o y H3K9me2 modifica ions
in he p omo e egions o mul iple genes19. We alida ed ha
JMJD1a was down egula ed in cance cells ollowing de achmen
om CDM bo h on he p o ein (Fig. 2a,b) and on mRNA
le els (Fig. 2c). In e es ingly, JMJD1A was down egula ed al eady
in cells g own on CDM a he p o ein le el (Fig. 2d,e), while
he mRNA le els o JMJD1a we e no changed (Fig. 2 ).
This co ela ed wi h educed s abili y o he JMJD1a p o ein on
CDM when compa ed wi h plas ic (Fig. 2g), indica ing ha
CDM-induced ep ession o JMJD1a fi s on he p o ein le el and
subsequen ly on mRNA le el co ela es wi h he CDM-induced
g ow h inhibi ion o cance cells when compa ed wi h cells g own
on plas ic (Fig. 1b,c). Silencing o JMJD1a ecapi ula ed he
g ow h inhibi ion in plas ic-cul u ed cance cells and induced a
fla quiescen -looking mo phology (Fig. 2h and Supplemen a y
Fig. 1e– ). Con e sely, p oli e a ion was enhanced in cells
wi h high JMJD1a-GFP le els when compa ed wi h high g een
fluo escen p o ein (GFP)-exp essing cells on plas ic (Fig. 2i and
Supplemen a y Fig. 1g). Impo an ly, o ced exp ession o
JMJD1a was su ficien o escue he CDM-induced g ow h
inhibi ion on TIFF-de i ed CDM (Fig. 2j). Fu he mo e, he
g ow h o MDA-MB-231 (b eas cance ) cells in an in i o
chicken emb yo cho ioallan oic memb ane (CAM) assay as well
as o ho opic umou g ow h in mice we e significan ly educed
Fib onec in
TIFF CDM
d
eCDM CDM o plas ic
MMP3
COL1A2
ACAT1
IDI1
HMGCS1
PLAUR1
JMJD1a
DDIT4
ANG
ALDOC
MUC1
GDF15
FABP5
NME1
FBXO32
SORBS2
PDE7B
15027 2
Numbe o
common genes:
CDM
CDM and
CDM o pl. CDM o pl.
CDM CDM o plas ic
T ypsiniza ion
Plas ic Plas ic
6 Days 5 Days
a
TIFF CDM Plas ic
DAPI
Ac in
β1-in eg in
bMDA-MB-231
1 2 3 4
0
20
40
60
80
100
Time (days)
P oli e a ion (% o max)
HeLa
P< 0.0001
CDM=35.2±0.6
Plas ic=26.5±0.6
Doubling ime:
1 2 3 4
0
20
40
60
80
100
P= 0.0001
CDM=29.81±0.8
Plas ic=23.6±1.0
Time (days)
Doubling ime:
Collagen I
c
CDM o pl.=41.5±2.2
Plas ic=35.9±0.86
Doubling ime:
P= 0.038
Time (days)
0
20
40
60
80
100
12 453
MDA-MB-231
P oli e a ion (% o max)
CDM o pl.=42.9±4.3
Plas ic=30.6±2.0
Doubling ime:
P= 0.014
0
20
40
60
80
100
12 453
Time (days)
HeLa
Figu e 1 | Fib oblas -de i ed CDM induces sus ained g ow h inhibi ion o cance cells. (a) Collagen I and fib onec in s aining o CDM gene a ed by NFs
(TIFFs). Scale ba , 20 mm. (b) P oli e a ion o MDA-MB-231 and HeLa cells pla ed on TIFF CDM o on plas ic in ull medium o he indica ed imes.
n(HeLa) ¼10, n(MDA-MB-231) ¼8. (c) P oli e a ion o MDA-MB-231 and HeLa cells a e de achmen om TIFF ma ices (6 days on ma ix be o e
de achmen ) and epla ing on plas ic in ull medium o he indica ed imes n(HeLa) ¼7, n(MDA-MB-231) ¼10. (d) Rep esen a i e images o
MDA-MB-231 cell mo phology on CDM and plas ic. Shown a e maximum in ensi y p ojec ions o con ocal images. Scale ba , 10 mm. (e) Schema ic
ep esen a ion o he expe imen al se -up. Red a ows indica e he ime poin s o sample collec ion o Illumina gene exp ession analysis. ( ) Common gene
exp ession changes in MDA-MB-231 and HeLa cells on CDM and 5 days a e CDM de achmen (CDM o plas ic (pl.)). The numbe s o commonly
egula ed genes (up- o down egula ed) in bo h cell lines a e indica ed in he able. Up egula ed genes a e ma ked wi h ed and down egula ed genes wi h
blue. All da a a e mean±s.e.m. Unpai ed - es was used o s a is ical analyses.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237 ARTICLE
NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions 3
ed
a
JMJD1a
Ac in
MDA-MB-231
CDM o
plas ic Plas ic
HeLa
CDM o
plas ic Plas ic
JMJD1a
Tubulin
CDM Plas ic
JMJD1a
GAPDH
MDA-MB-231
CDM Plas ic
JMJD1a
Tubulin
HeLa
MDA-MB-231
JMJD1a p o ein le els
CDM o
plas ic
Plas ic
MDA-MB-231
1.5
0
0.5
1.0
n=7
P= 0.007
JMJD1a mRNA le els
CDM o
plas ic
Plas ic
n=5
P= 0.0072
1.5
0
0.5
1.0
bc
Time (h)
HeLa
CDM o
plas ic
Plas ic
1.5
0
0.5
1.0
n=3
P= 0.0365
JMJD1a mRNA le els
h
JMJD1a p o ein le els
CDM Plas ic
1.5
0
0.5
1.0
n=5
P= 0.0075
MDA-MB-231
JMJD1a mRNA le els
2.0
1.5
0
0.5
1.0
MDA-MB-231
n=3
P=ns
CDM Plas ic
iP oli e a ion on CDMP oli e a ion on plas ic
0.0
0.1
0.2
0.3
0.4
siCon ol siJMJD1a
P= 0.0014
MDA-MB-231
j
k
0
10
20
30
siCon ol siJMJD1a
Tumou a ea (mm
2
)
Tumou a ea (mm
2
)
P< 0.001
MDA-MB-231
l
P oli e a ion (% o max.)
JMJD1a
Tubulin
siCon ol
siJMJD1a
0 h
1 h
3 h
5 h
0 h
1 h
3 h
5 h
0.0
0.5
1.0
1.5
Plas ic
CDM
P= 0.031
P=ns
JMJD1a p o ein le els
Time o CHX
g
0 20 40 60 80
0
50
100 Low JMJD1a=19.8±2.1
High JMJD1a=16.6±0.6
Doubling ime
P= 0.01
siCon ol=24.6±1.4
Doubling ime
P= 0.017 GFP=20.6±0.8
siJMJD1a=54.3±7.5
0
20
40
60
80
100
0 20 40 60 80 100
P oli e a ion (% o max.)
Time (h)
MDA-MB-231
siCon ol
siJMJD1a
Day 0
JMJD1a
Tubulin
Day 8
020 40 60 80
0
20
40
60
80
100 GFP
JMJD1a
Time (h)
P< 0.001
P oli e a ion (% o max.)
50
M (K)
150
37
50
M (K)
150
M (K)
150
37
M (K)
150
50
150
50
M (K)
150
50
Figu e 2 | No mal ECM es ains JMJD1a exp ession, cell p oli e a ion and umou g ow h. (a,b) Wes e n blo (a) and quan ifica ion (b) o JMJD1a
p o ein le els no malized o loading con ol. (c) JMJD1a mRNA exp ession (qRT–PCR) ela i e o GAPDH mRNA in MDA-MB-231 and HeLa cells a e
ma ix de achmen (6 days on TIFF CDM and 5 days on plas ic; CDM o plas ic) and on plas ic. (d,e) Wes e n blo (d) and quan ifica ion o JMJD1a p o ein
le els (e) no malized o loading con ol. ( ) JMJD1a mRNA exp ession (qRT–PCR) ela i e o GAPDH mRNA ( ) in he indica ed cells pla ed on ei he TIFF
CDM o on plas ic o 4 days. (g) Wes e n blo quan ifica ion showing JMJD1a s abili y on CDM and plas ic 24 h a e pla ing. Time o he cycloheximide
(CHX) ea men is indica ed and P alues a e calcula ed be ween 0 and 5 h. Pai ed - es was used o s a is ical analysis, n¼3. (h) P oli e a ion o
MDA-MB-231 cells upon JMJD1a silencing on plas ic, n¼3. (i) P oli e a ion o JMJD1a-GFP o GFP-o e exp essing MDA-MB-231 cells on plas ic. Cells
we e so ed by FACS (JMJD1a: high and low; GFP: high), n¼3. (j) P oli e a ion o GFP con ol and JMJD1a-GFP-o e exp essing MDA-MB-231 cells on TIFF-
de i ed CDM. n(GFP) ¼11 CDMs and n(JMJD1a-GFP) ¼12 CDMs. Two-way analysis o a iance (ANOVA) was used o calcula e he P alue. Da a a e
mean±s.e.m. (k) Con ol o JMJD1a siRNA- ans ec ed MDA-MB-231 cells (1 106) we e implan ed on CAM memb anes inside a plas ic ing o analyse
umou g ow h in i o o 3 days. Shown a e quan ified umou a eas om h ee indi idual expe imen s n(siCon ol) ¼25, n(siJMJD1a) ¼23 eggs.
(l) O ho opic umou g ow h assay. Con ol o JMJD1a siRNA- ans ec ed MDA-MB-231 cells (2 106) we e injec ed in o he a pad o nude mice
(n¼19) and umou g ow h was measu ed 8 days a e injec ion. Wes e n blo in showing he silencing e ficacy o JMJD1a siRNA on he day o he
injec ion (Day 0) and a he end o he expe imen (Day 8). Shown a e mean±s.d. and (g–i) mean±s.e.m. Pai ed - es was used o s a is ical analyses in
b–hand non-pai ed - es in j,k.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237
4NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions
upon silencing o JMJD1a (Fig. 2k,l). Thus, JMJD1a is
down egula ed ollowing CDM exposu e o cells and is a po en
egula o o cance cell p oli e a ion in i o and in i o.
NFs and CAFs gene a e a chi ec u ally dis inc CDM. To es
he abili y o s omal ECM o influence cance cells in a mo e
clinically ele an model, we isola ed CAFs om he umou
s oma o h ee head and neck squamous cell ca cinoma
(HNSCC) pa ien s (Supplemen a y Fig. 2a) and NFs om an
una ec ed a ea o he same pa ien s. CAFs we e iden ified by he
high exp ession o CAF ma ke smoo h muscle ac in-a(a-SMA),
which was low o absen in he NFs and he TIFFs (Fig. 3a and
Supplemen a y Fig. 2b). RNA-sequencing demons a ed ha all
h ee CAFs and NFs clus e ed oge he (Supplemen a y Fig. 2c,d),
which indica ed ha CAFs de i ed om di e en pa ien s we e
mo e simila o o he CAFs han o hei co esponding NFs and
all he NFs esembled each o he (Supplemen a y Fig. 2c,d).
Howe e , all NF and CAF cell lines exp essed genes, which a e
ypically highly exp essed by fib oblas s, such as Vimen in,
Fib onec in 1 and se e al di e en collagens alida ing ha hey
a e fib oblas s (Supplemen a y Fig. 2e). CAFs had ele a ed le els
o YAP and TAZ (Fig. 3b–d) compa ed wi h NFs, and hey we e
also mo e able o con ac collagen gels (Supplemen a y Fig. 2 ),
in line wi h a p e ious epo on he ole o YAP and con ac ili y
in he CAF pheno ype30. In addi ion, we obse ed a significan
up egula ion o b1-in eg in, which has also been connec ed o
con ac ili y and mechanosignalling (Supplemen a y Fig. 2g,h).
We analysed he CDM p oduced by he di e en fib oblas s
using immunofluo escence and scanning elec on mic oscopy
(SEM). Immunofluo escence s aining e ealed ha simila ly o
TIFF CDM (Fig. 1a) he ma ix bundles in NF CDM we e mo e
spa se compa ed wi h he co esponding CAF ma ix, which
displayed a dense and mo e uni o m collagen and fib onec in
s aining (Fig. 3e and Supplemen a y Fig. 3a,b). Fu he mo e, based
on SEM analysis, he NF and TIFF CDMs had mo e uni o m and
aligned s uc u es compa ed wi h CAF-de i ed ma ices (Fig. 3 ).
Acco ding o he RNA-sequencing da a, he mRNA exp ession o
di e en ypes o collagen o fib onec in was no changed,
sugges ing ha ma ix assembly and/o u no e , a he han
p oduc ion, esul s in dis inc CDM a chi ec u e be ween NF and
CAF CDM. Howe e , e en hough we could no find significan
di e ences on mRNA le els o ma ix- ela ed genes o in cell
adhesion o solubilized and e-pla ed NF and CAF CDM
(Supplemen a y Fig. 3c), he NF and CAF CDMs may di e in
hei p o ein composi ion as sugges ed ea lie by o he s31.
These da a show ha CDMs gene a ed by NFs and CAFs di e
and ha CAF-de i ed ma ices om di e en pa ien s sha e
simila ea u es ha a e dis inc om ma ices made by NFs om
he same indi idual.
CAF ma ix lacks g ow h-inhibi o y p ope ies. To es he
abili y o he pa ien -de i ed s omal ECM o influence he
p oli e a ion o cance cells, we cul u ed MDA-MB-231 and HeLa
cells on CDMs de i ed om ei he NFs o CAFs. In e es ingly,
NF CDM was significan ly g ow h-inhibi o y compa ed wi h
CDM gene a ed by CAFs om he same pa ien (Fig. 3g), and he
same was obse ed when compa ing TIFF and CAF CDM
(Supplemen a y Fig. 3d). Impo an ly, also he g ow h o he
pa ien -de i ed p ima y squamous cell ca cinoma (SCC) cells
was inhibi ed by he NF CDM ma ix compa ed wi h he CAF
CDM (Supplemen a y Fig. 3e), demons a ing ha he g ow h-
es ic i e abili y o NF CDM is widely applicable o di e en
ca cinomas. Simila ly o he TIFF CDM, he g ow h es ic ion
was specifically due o he ma ix and no soluble ac o s as
cocul u e o SCC o MDA-MB-231 cells wi h NFs o CAFs
sepa a ed by a fil e o condi ioned medium om NFs o CAFs
had no e ec on p oli e a ion (Supplemen a y Fig. 3 –h). In line
wi h esul s ob ained in b eas cance cells, JMJD1a silencing was
su ficien o inhibi he p oli e a ion o pa ien -de i ed SCC
cells (Supplemen a y Fig. 3i). Concu en wi h he abili y o
NF-de i ed CDM o down egula e JMJD1a le els, cance cells on
plas ic and CAF CDM exp essed high le els o JMJD1a, whe eas
JMJD1a was down egula ed on no mal CDMs (TIFF and NF;
Fig. 3h), u he alida ing he abili y o no mal CDM o es ic
p oli e a ion by JMJD1a down egula ion.
JMJD1a le els co ela e wi h ac i a ed s oma wi hin umou s.
The da a abo e demons a e ha NF CDM down egula es and
CAF CDM suppo s le els o JMJD1a. We ound his ologic
ea u es in human b eas cance and HNSCC umou s ha a e
compa ible wi h ou expe imen al model, sugges ing ha he CAF
ma ix suppo s JMJD1a exp ession. We obse ed ha JMJD1a
exp ession coincides wi h he p esence o a-SMA-posi i e
s omal cells, which a e cha ac e is ic o he umou
mic oen i onmen 32,33. S aining o 10 no mal b eas issue
samples, 28 p ima y b eas cance sec ions and 7 lymph node
me as asis e ealed ha in no mal b eas issues JMJD1a
exp ession was low o absen and a-SMA was es ic ed o he
basal mamma y epi helial cells (Fig. 4a). In con as , 27/28 o he
b eas cance s we e JMJD1a-posi i e (low: 57%; in e media e:
29%; high: 11%) and 27/28 o he umou s had a-SMA-posi i e
s oma (Fig. 4a). In addi ion, all me as ases we e posi i ely
s ained o JMJD1a and a-SMA. In HNSCC pa ien samples,
10/14 o he cance s we e JMJD1a- and a-SMA-posi i e (JMJD1a
low: 50%; JMJD1a high: 21%). All HNSCC samples ha we e
highly JMJD1a-posi i e also exhibi ed in ense a-SMA exp ession
in he s oma (Fig. 4b). These analyses sugges ha JMJD1a is
exp essed in b eas and HNSCC ca cinomas and co ela es
wi h he p esence o a-SMA-posi i e s oma in pa ien s.
Since inc eased ma ix s i ness has been linked o cance -
associa ed s omal al e a ions and cell p oli e a ion, we measu ed
he s i ness o he pa ien -de i ed fib oblas -gene a ed CDMs
and TIFF CDM using a omic o ce mic oscopy (AFM). High
inden a ion o ces o up o 30 nN we e applied. Pai wise
compa ison o ma ices gene a ed om fib oblas s om he
same pa ien demons a ed ha CAF-de i ed ma ices we e
significan ly s i e han he no mal ma ix (Fig. 4c). In addi ion,
s i ness o he TIFF CDM was simila o he NF CDMs, in
line wi h he simila g ow h-inhibi o y p ope ies o he
ma ices (Fig. 4d). These s i ness alues o NF-gene a ed
CDMs and he highe ange o s i ness in he CAF CDM a e
highly consis en wi h ea lie measu emen s on CAF-con ac ed
collagen gels30 and di e ences in issue s i ness obse ed in
no mal b eas issue and cance 34.
JMJD1a le els and localiza ion a e egula ed by s i ness.The
equi emen o in ac ECM and he ac ha no mal CDM was
less s i compa ed wi h he CAF CDM sugges ed ha ma ix
s i ness, in addi ion o ma ix a chi ec u e and possibly
composi ion, could be in ol ed in he abili y o he no mal ma ix
o inhibi cance cell p oli e a ion. To es his in a con olled
manne , cance cells we e g own on collagen I-coa ed hyd ogels o
a ying s i ness. MDA-MB-231, HeLa and pa ien -de i ed SCC
cells p oli e a ed significan ly mo e on s i e suppo s
(Supplemen a y Fig. 4a,b), sugges ing ha he lowe s i ness o NF
CDM is likely o con ibu e o i s g ow h- es ic i e p ope ies.
JMJD1a has hus a been epo ed o localize o he
nucleus19,27,35 in line wi h i s unc ion as a his one
deme hylase. Howe e , nuclea ac iona ion analyses e ealed
ha JMJD1a localizes bo h o he nucleus and o he cy oplasm in
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237 ARTICLE
NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions 5

MDA-MB-231 cells (Supplemen a y Fig. 4c), sugges ing ha i
migh be shu ling be ween he wo compa men s. YAP/TAZ
ansc ip ion ac o s a e well-es ablished mechanosensi i e
egula o s o cell p oli e a ion, such ha s i ma ix and cell
sp eading suppo YAP/TAZ p o ein s abili y and nuclea
localiza ion36. To s udy he po en ial mechanosensi i i y o
JMJD1a, we in es iga ed JMJD1a localiza ion and le els unde
condi ions known o egula e YAP/TAZ36. YAP/TAZ p o ein
YAP p o ein le els (RQ)
Pa ien 3
NF CAF
P=ns
n=3
0.5
1.0
1.5
2.0
0
Pa ien 2
NF CAF
P= 0.008
n=5
Pa ien 1
NF CAF
3.0
2.0
1.0
0
P= 0.0211
n=4
3.0
2.0
1.0
0
×10,000
NF CAF
Pa ien 2
P= 0.003
n=6
6.0
4.0
2.0
0
TAZ p o ein le els (RQ)
Pa ien 1
NF CAF
4.0
2.0
0
1.0
3.0
P= 0.02
n=4
Pa ien 3
5
10
20
0
NF CAF
15
P= 0.043
n=3
α-SMA
Tubulin
NF CAF NF CAF
Pa ien 2Pa ien 1 Pa ien 3
NF CAF
CAF CDM
NF CDM
TIFF CDM
Pa ien 1Pa ien 3
Fib onec inCollagen
NF CDM CAF CDM NF CDM CAF CDM
0 2 4 6
0
20
40
60
80
100
Time (days)
P oli e a ion (% o max.)
HeLa
Doubling ime:
NF CDM: 116.3±23.8
CAF CDM: 61.4±6.8
P= 0.044
13 5
0 2 4
0
20
40
60
80
100
Time (days)
Doubling ime:
NF CDM: 35.8±1.6
CAF CDM: 29.3±1.0
P= 0.002
MDA-MB-231
P oli e a ion (% o max.)
135
M (K)
50
50
ab
cd
e
g
h
Pa ien 1 Pa ien 2
Tubulin
YAP
TAZ
NF CAF NF CAF
Ac in
YAP
TAZ
Pa ien 3
Tubulin
YAP
TAZ
NF CAF
M (K)
75
50
50
50
37
NF
CDM
CAF
CDM Plas ic
JMJD1a
Ac in
0.1 1.05 1.0
TIFF
CDM
CAF
CDM Plas ic
JMJD1a
Ac in
0.65 1.02 1.0
M (K) M (K)
150 150
50
37
50
37
Figu e 3 | Pa ien -de i ed CAF and NF CDMs a e a chi ec u ally and unc ionally dis inc . (a,b) Rep esen a i e wes e n blo s showing SMA-a(a) and
YAP/TAZ exp ession (b) in NFs and CAFs isola ed om h ee SCC pa ien s. (c,d) Quan ifica ion o YAP (c) and TAZ (d) exp ession in NFs and CAFs
no malized o loading con ol. Da a a e mean±s.d. (e) Collagen I ( ed) and fib onec in (blue) s aining o pa ien #1 and #3 NF and CAF CDM. Scale ba ,
20 mm. ( ) Rep esen a i e SEM images o TIFF, Pa ien #1 NF and CAF CDM. Scale ba , 5 mm. (g) P oli e a ion o MDA-MB-231 and HeLa cells on Pa ien #1
NF and CAF CDM. n(MDA-MB-231) ¼10–11 and n(HeLa) ¼19 om h ee independen expe imen s. Da a a e mean±s.e.m. and P alues a e calcula ed
om he doubling imes. (h) JMJD1a exp ession in MDA-MB-231 cells cul u ed on TIFF, NF and CAF CDMs and on plas ic. Quan ifica ion shows ela i e
JMJD1a amoun no malized o loading con ol. (c,d) Non-pa ame ic Mann–Whi ney es was used o s a is ical analyses in c,d and non-pai ed - es in g.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237
6NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions
le els a e known o be down egula ed on so suppo s.
In e es ingly, we ound ha also JMJD1a s abili y and he e o e
he p o ein le els we e educed when cells we e cul u ed on so
hyd ogels (Fig. 5a,b and Supplemen a y Fig. 4d).
In addi ion o ma ix s i ness, RhoA-signalling, ac omyosin
con ac ili y and cell sp eading a e known egula o s o
YAP/TAZ localiza ion and p o ein s abili y16. To es whe he
hese ques egula e JMJD1a as well, we pla ed MDA-MB-231 cells
on mic opa e ns wi h equal o al adhesi e su ace dis ibu ed
o e a a iable sp eading a ea. In cells sp eading on 800-mm2
fib onec in-coa ed mic opa e ns (la ge), bo h JMJD1a and
YAP/TAZ we e p edominan ly nuclea compa ed wi h cells
sp eading on 400-mm2(small) fib onec in-coa ed mic opa e ns,
whe e bo h p o eins we e p edominan ly cy oplasmic (Fig. 5c,d).
Acco dingly, we obse ed ha bo h JMJD1a and YAP/TAZ
became cy osolic on lowe s i ness hyd ogels (0.5 kPa) as soon as
a
No mal b eas B eas cance Me as asis
B eas cance Me as asis
None
Low (+1)
Medium (+2)
High (+3)
α-SMA
No mal b eas B eas ca cinoma
JMJD1a
Nega i e s oma Low Medium
α-SMA
None Low Medium
c
Low
Low
Adjacen issue slices
Medium
Medium
×10 ×20
Adjacen issue slices
B eas ca cinoma
S
LB
S
S
TT
SS
α-SMA JMJD1a
HNSCC
b
None
Low
High
JMJD1a and α-SMA
JMJD1a
×10 ×20
d
0
100
200
300
400
S i ness (Pa)
NF ma ix CAF ma ix
Pa ien 1 ma ices
0
100
200
300
400
NF ma ix CAF ma ix
Pa ien 3 ma ices
S i ness (Pa)
0
100
200
300
400
S i ness (Pa)
TIFF ma ix
P< 0.0001 P< 0.0001
Figu e 4 | JMJD1a le els co ela e wi h a-SMA-posi i e s oma in human umou s. (a,b) JMJD1a and a-SMA s aining om sec ions o he same no mal
b eas issue and p ima y b eas ca cinomas (a) o HNSCC (b) issue. Fo b eas ca cinomas, examples o he immunos aining o di e en exp ession le els
and wo di e en magnifica ions o he same issue a e shown. Scale ba , 200 mm. S, s oma; T, umou . Inse in ahighligh s he a-SMA localiza ion in
basal cells in he no mal mamma y gland (B, basal; L, luminal). Quan ifica ion o he incidence o JMJD1a o a-SMA posi i i y in he analysed samples is
shown. (c,d) S i ness o he pa ien -de i ed NF o CAF (c) and TIFF (d) CDMs is exp essed by he Young’s modulus, which was measu ed using
AFM inden a ion. Each g ey do ep esen s indi idual measu emen and black line indica es he mean. Non-pai ed - es was used o s a is ical analysis.
Non-pai ed - es was used o s a is ical analysis.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237 ARTICLE
NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions 7
he cells had ully adhe ed (3h) and we e inc easingly nuclea on
s i e suppo s bo h in MDA-MB-231 and pa ien SCC cells
(Fig. 5e–g and Supplemen a y Fig. 5a–d), whe eas ano he
nuclea p o ein Son was no mechanosensi i e on hyd ogels
(Supplemen a y Fig. 5e). Impo an ly, bo h JMJD1a and
YAP/TAZ we e p edominan ly cy oplasmic in MDA-MB-231
cells g own on so , no mal CDM compa ed wi h hei nuclea
localiza ion on plas ic o collagen- and fib onec in-coa ed
plas ic (Fig. 5h,i and Supplemen a y Fig. 5 ), and e y simila
egula ion was obse ed also in he pa ien -de i ed SCC cells
(Supplemen a y Fig. 5g). Co espondingly, JMJD1a was
p ominen ly cy oplasmic on NF CDM compa ed wi h i s
nuclea localiza ion on CAF CDMs (Fig. 5j), indica ing
ha CDM-induced egula ion o JMJD1 localiza ion and le els
is simila o he p e iously es ablished egula ion o YAP/TAZ
(Supplemen a y Fig. 5h).
Howe e , unlike YAP/TAZ, JMJD1a localiza ion was no
dependen on an in ac ac in cy oskele on o Rho-signalling as
g
Nuclea JMJD1a/ o al
CDM
Plas ic
Collagen
+ FN
1.0
0.5
1.5
0
P< 0.0001
P< 0.0001 ns
i
a
0
1.0
0.5
1.5
2.0 P= 0.01
P= 0.029
0.5 50 PL
Rela i e JMJD1a le els
S i ness
(kPa)
0.5 kPa 50 kPa
NNCC
Lamin A/C
GAPDH
JMJD1a
YAP
TAZ
0.5 kPa4 kPaPlas ic
JMJD1a YAP/TAZ DAPI Me ge
CDMPlas ic
Me ge ROIAc in DAPIJMJD1a
Cy o
Nuc+ Cy o
Nuc
Cell numbe
0.5 kPa
4 kPa
50 kPa
Plas ic
0
50
100
150
Cell numbe
JMJD1a
5
15
25
5
15
25 YAP/TAZ
La geSmall
JMJD1aJMJD1aYAP/TAZ YAP/TAZ
DIC DIC
Pa e n Pa e n
0.5 50 PL
JMJD1a
Tubulin
kPa
kl
b
e
Cy o
Nuc+ Cy o
Nuc
c
Rho
ROCK
ECM
Ac omyosin
Nucleus
JMJD1a
YAP/TAZ
SRC
?
ECM
YAP/TAZ
Nucleus
JMJD1a SRC
JMJD1a
STIFF
SOFT
CA-S c exp ession on 0.5 kPa
JMJD1a YAP/TAZ
JMJD1a DAPI
YAP/TAZ DAPI
JMJD1a YAP/TAZ
JMJD1a DAPI
YAP/TAZ DAPI
CA-S c exp ession on plas ic
La ge
Small
NF CDM CAF CDM
JMJD1a DAPI
j
JMJD1a
d
Cell numbe
h
La ge
Small
Cy o
Nuc+ Cy o
Nuc
50
M (K)
150
75
37
75
50
150
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237
8NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions
Blebbis a in, Cy ochalasin D o ROCK inhibi o Y27632 ea men
did no change JMJD1a localiza ion, while YAP/TAZ became
p edominan ly cy oplasmic wi h all hese ea men s
(Supplemen a y Fig. 6a). In eg in signalling is known o be
impo an o mechanosignalling37–40, albei he specific
equi emen o in eg in–ECM in e ac ion in YAP/TAZ
egula ion emains con o e sial41–43. In o de o in es iga e
whe he in eg in b1 ac i i y could egula e JMJD1a, we pla ed
cells on in eg in b1 inac i a ing (4B4) o ac i a ing monoclonal
an ibodies (12G10) and analysed JMJD1a localiza ion in se um-
ee condi ions. We ound ha locking in eg ins in o ac i e o
inac i e con o ma ion was no su ficien o al e JMJD1a
localiza ion on so o s i (Supplemen a y Fig. 6b). Fu he mo e,
silencing o in eg in b1 om MDA-MB-231 cells had no e ec on
he nuclea localiza ion o JMJD1a, sugges ing ha JMJD1a nuclea
localiza ion is unlikely o be dependen on a specific in eg in
he e odime (Supplemen a y Fig. 6c). Nex , we es ed whe he
known in eg in downs eam e ec o s, Focal adhesion kinase o
Rous sa coma oncogene cellula homolog (SRC)-kinase, could
egula e JMJD1a localiza ion. We ound ha pha macological
inhibi ion o Focal adhesion kinase and SRC alone o in
combina ion did no al e JMJD1a localiza ion on plas ic bu
SRC inhibi ion induced YAP/TAZ ansloca ion o he cy oplasm
(Supplemen a y Fig. 6d). Howe e , exp ession o cons i u i ely
ac i e SRC (CA-SRC) inc eased y osine phospho yla ion o GFP-
JMJD1a (Supplemen a y Fig. 6e) and was su ficien o induce
nuclea JMJD1a on so 0.5kPa hyd ogels (Fig. 5k), indica ing ha
SRC ac i a ion is su ficien o suppo JMJD1a nuclea localiza ion
on so . In e es ingly, o ced exp ession o CA-SRC did no al e
YAP/TAZ localiza ion on so (Fig. 5k), demons a ing ha
s i ness-media ed egula o y pa hways o YAP/TAZ and JMJD1a
localiza ion a e dis inc (Fig. 5i).
JMJD1a egula es YAP/TAZ exp ession. JMJD1a and YAP/TAZ
le els we e significan ly educed on TIFF CDM (Fig. 6a) in line
wi h hei cy oplasmic ansloca ion (Supplemen a y Fig. 5 ), and
his co ela ed wi h educed ansc ip ion o he well-known
YAP/TAZ a ge genes Connec i e issue g ow h ac o (CTGF)
and Th ombospondin 1 (THBS1) on TIFF CDM compa ed wi h
plas ic (Fig. 6b). Fu he mo e, we obse ed ha le els o
YAP/TAZ and JMJD1a in indi idual MDA-MB-231 cells
co ela ed significan ly (Fig. 6c). Thus, we we e in e es ed o
in es iga e he po en ial link be ween JMJD1a and YAP/TAZ.
Ch oma in immunop ecipi a ion (ChIP) assays e ealed ha
JMJD1a is ec ui ed o TAZ p omo e (Fig. 6d, Supplemen a y
Fig. 7a) and, in line wi h he deme hylase ac i i y o JMJD1a,
ansien JMJD1a silencing inc eased H3K9me2 me hyla ion
on he TAZ p omo e (Fig. 6e and Supplemen a y Fig. 7a).
Fu he mo e, JMJD1a silencing wi h wo independen sho
in e e ing RNAs (siRNAs) educed YAP/TAZ p o ein and mRNA
le els in MDA-MB-231 and pa ien -de i ed SCC cells (Fig. 6 –h
and Supplemen a y Fig. 7b–e) as well as exp ession o YAP/TAZ
a ge genes in MDA-MB-231 (Fig. 6i and Supplemen a y Fig. 7a).
Con e sely o JMJD1a silencing, o e exp ession o JMJD1a-GFP
(which localizes o he nucleus simila ly o endogenous JMJD1a on
plas ic, Supplemen a y Fig. 7 ) inc eased YAP/TAZ le els
(Fig. 6j,k) as well as THBS1 and CTGF gene exp ession (Fig. 6l).
Impo an ly, o e exp ession o JMJD1a was su ficien o inc ease
YAP/TAZ le els e en on TIFF-de i ed CDMs (Fig. 6m) and 4 kPa
hyd ogels (Fig. 6n), whe e YAP/TAZ p o ein-le el s abili y is
comp omised because o inc eased cy oplasmic localiza ion and
deg ada ion36. Howe e , o e exp ession o wild- ype o ac i e
YAP mu an (YAP-5SA ( e . 44)) alone was no su ficien o escue
p oli e a ion on MDA-MB-231 cells g own on TIFF-de i ed
CDMs o in JMJD1a-silenced cells, sugges ing ha addi ional
JMJD1a a ge genes con ibu e o he CDM-induced g ow h
inhibi ion (Supplemen a y Fig. 7g,h).We also ound ha silencing
o JMJD1a had no e ec on ei he YAP/TAZ nuclea localiza ion
(Supplemen a y Fig. 8a) o phospho yla ion o LATS1/2
(Supplemen a y Fig. 8b), which is a nega i e egula o o
YAP/TAZ p o ein s abili y. Fu he mo e, we could no de ec
inc eased YAP (S-127) phospho yla ion, which is associa ed wi h
educed s abili y o YAP, upon JMJD1a silencing (Supplemen a y
Fig. 8c) and p o ein s abili y o YAP/TAZ was no educed
upon JMJD1a silencing, sugges ing ha JMJD1a egula es
YAP/TAZ on he ansc ip ional le el (Supplemen a y Fig. 8d).
This u he demons a es ha JMJD1a is a p e iously undesc ibed
ansc ip ional ac i a o o YAP/TAZ exp ession, and ha
o ced exp ession o JMJD1a can suppo YAP/TAZ le els e en
on so subs a es because o i s abili y o inc ease he ansc ip ion
o YAP/TAZ.
JMJD1a and YAP/TAZ exp ession co ela es in human cance .
We ound ha JMJD1a and YAP/TAZ exp ession co ela ed also
in human ca cinomas. In a la ge coho o p ima y b eas
umou s45, JMJD1a and YAP/TAZ ( he YAP an ibody ecognizes
bo h ansc ip ion ac o s) we e s ongly associa ed wi h se e al
commonly assessed clinicopa hological p ognos ic ac o s
(Supplemen a y Tables 1–4 and Supplemen a y No e 1). In all,
689 (94.3%) and 262 (35.8%) ou o he 731 umou s a ailable o
JMJD1a s aining had posi i e cy oplasmic and nuclea s aining,
espec i ely, and 645 (86.5%) and 514 (68.9%) ou o he 746
cance s a ailable o YAP/TAZ s aining had posi i e cy oplasmic
and nuclea YAP/TAZ exp ession.
Fully in line wi h hei co ela ed exp essions in i o, we ound
ha JMJD1a and YAP/TAZ le els significan ly co ela ed in
Figu e 5 | Mechanosensi i e egula ion o JMJD1a on so and s i ECM and CDM. (a,b) Rep esen a i e wes e n blo (a) and quan ifica ion (b) showing
JMJD1a exp ession in cells pla ed on 0.5 and 50 kPa hyd ogels and on plas ic (PL). Tubulin was used as loading con ol, n¼4 (mean±s.d.). (c) JMJD1a
( ed) and YAP/TAZ (g een) and DAPI (blue) s aining in MDA-MB-231 cells on la ge (800 mm2) and small (400 mm2) sp eading a ea mic opa e ns
(adhesi e a ea is he same). Cell mo phology is shown as DIC. (d) Quan ifica ion o cy oplasmic and nuclea JMJD1a and YAP/TAZ localiza ion on small
and la ge mic opa e ns. n(cells) ¼20 pe pa e n size. (e, ) Immunofluo escence s aining showing (e) and quan i ying ( ) YAP/TAZ and JMJD1a
localiza ion on collagen I and fib onec in-coa ed hyd ogels o a ying s i ness (0.5, 4 and 50 kPa) and on plas ic. Scale ba , 10 mm. (g) Rep esen a i e
wes e n blo showing JMJD1a and YAP/TAZ nuclea (N) and cy oplasmic (CP) localiza ion in cells pla ed on 0.5 and 50 kPa hyd ogels. Lamin A/C and
GAPDH we e used as ac iona ion con ols. (h,i) Immunofluo escence s aining quan ifica ion (i) and ep esen a i e images (h) o JMJD1a localiza ion in
cells pla ed on TIFF CDM, plas ic o collagen and fib onec in ligands (2.5 mgml1collagen and 2.5 mgml1fib onec in). Nuclea localiza ion o JMJD1a was
quan ified wi h he CellP ofile so wa e. n(cells): CDM ¼74 cells; plas ic ¼33 cells and collagen þFN ¼61. Scale ba , 10 mm. (j) Rep esen a i e
immunofluo escence images showing JMJD1a localiza ion in MDA-MB-231 cells g owing on NF and CAF CDMs o 3 days. Scale ba , 10 mm. (k) YAP/TAZ
and JMJD1a localiza ion in CA-SRC-exp essing MDA-MB-231 cells g owing on so 0.5 kPa hyd ogels o on plas ic. Rep esen a i e images om h ee
independen expe imen s. Scale ba , 10 mm. (l) Model o dis inc mechano ansduc ional egula ion o YAP/TAZ and JMJD1a on so and s i . Red a ows
indica e he pa hway, which we and o he s ha e shown o egula e YAP/TAZ nuclea localiza ion. Blue a ow indica es he SRC kinase-media ed and
s i ness-dependen egula ion o JMJD1a and YAP/TAZ.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12237 ARTICLE
NATURE COMMUNICATIONS | 7:12237 | DOI: 10.1038/ncomms12237 | www.na u e.com/na u ecommunica ions 9