CD24
C
D
E
LacZ Snail1 Zeb2 Twist1
CD
44
Snail Twist1 Zeb2 LacZ
0.2 95.9 90.9 84.7
Twist1 Snail Zeb2 LacZ
** **
**
Snail Twist1 Zeb2 LacZ
Supplementary Figure S1. EMT promoting transcription factors induce stemness characteristics.
A, Overexpression of Foxq1 significantly induced EMT in HMLE cells. Left Panel: Morphological change.
Right panel: expression of EMT markers. B, Overexpression of Snail, Twist1 and Zeb2 significantly
induced EMT in HMLE cells. C, Overexpression of Snail, Twist1 and Zeb2 significantly increased the
CD44high/CD24low cell population in HMLE cells. D, Summary of the mammosphere formation assay for
Snail1, Twist1 and Zeb2 overexpressing HMLE cells (**P<0.01). E, Representative figures for the
mammosphere formation assay of Snail1, Twist1 and Zeb2 overexpressed HMLE cells. HMLE cells
transfected with LacZ was used as a control.
B
A
LacZ Foxq1
0.01
0.1
1
10
100
E-cadherin N-cadherin Vimentin Fibronection
LacZFoxq1
Rela
tive e
xpre
ssio
n
Num
ber
of
mam
mosphere
s
(per1
,000 c
ells
)
0
10
20
30
40
50
0 100 nM 2 nM 4 nM 50 nM Paclitaxel Doxorubicin
LacZ
F
oxq1
Cell types Numbers of implanted cells
20,000 9,000 3,000
4T1/NT 5/5 (100%) 5/5 (100%) 4/5 (80%)
4T1/sh3 5/5 (100%) 4/5 (80%) 1/5 (20%)
A B
C
Nu
mb
er
of
ma
mm
osp
he
res
(pe
r1,0
00
ce
lls)
4T1/NT 4T1/sh3 05
101520253035 **
Supplementary Figure S2. The role of Foxq1 in drug resistance, mammosphere formation and
tumor initiation. A, Representative images of the clonogenic assays of HMLE cells with or without
Foxq1 overexpression treated with different doses of doxorubicin (dox) or paclitaxel (pac). B, Knockdown
of Foxq1 significantly decrease mammosphere formation ability of 4T1 cells. Top panel: summary of
mammosphere formation results. Bottom panels: representative image of mammosphere formation. C,
Summary of tumor initiation capability of Foxq1 in 4T1 cells. Five mice were used for each group.
Percentiles represent tumor formation in specific groups. D, The effect of Foxq1 in apoptosis. Tumors
were collected from the 4T1 implanted BALB/C mice, and the prepared tissue slides were analyzed with
IHC using anti-BAX, p27 and Cleaved Caspase 8 antibodies, as well as H&E staining. The origins of all
four tumor samples are indicated at the top of the panels. NT: tumor from 4T1/NT cells bearing mouse.
SH3: tumor from 4T1/SH3 Foxq1 knockdown cell bearing mouse. NT+Pac: tumor from 4T1/NT cells
bearing mice treated with Paclitaxel. SH3+Pac: tumor from 4T1/Sh3 cells bearing mice treated with
Paclitaxel. Scale bar, 20 µm.
D NT SH3 NT+Pac SH3+Pac
p27
Cle
aved
Caspase
8
HE
B
AX
E
C
D
B A
***
**
Rela
tive e
xpre
ssio
n
***
Rela
tive e
xpre
ssio
n
**
**
0
4
6
8
2
10
0
4
6
8
2
10
Vector (μg)
Twist1(μg)
0.6 0 0.3
0 0.6 0.3
Rela
tive L
uc. activity
1.0
1.5
2.0
0.5
2.5
0
T-Pα2
**
**
T-Pβ1
***
***
0.6 0 0.3
0 0.6 0.3
Rela
tive L
uc. activity
2
0
4
3
1
Twist1 LacZ Zeb2 LacZ
F
Supplementary Figure S3. Twist1 and Zeb2 regulate PDGFRs. A and B, The relative expression level
of the Foxq1, Zeb2/Twist1 and PDGFRα and β genes in Twist1 (A) or Zeb2 (B) overexpressing HMLE cells was
measured by real-time RT-PCR assay (*P<0.05, **P<0.01 and ***P<0.001). C, Potential binding sites of Twist1
and Zeb2 in the promoter regions of PDGFRα and β genes were identified by an in silico analysis. Top and
middle panels show Twist1 binding sites in PDGFRα and β promoter region. Low panel shows Zeb2 binding site
in PDGFRβ promoter region. The conserved binding sites are highlighted with red. D, ChIP-qPCR analysis
shows that enrichment of one Twist1 binding site DNA from the PDGFRα and β promoter region, respectively
(**P<0.01). Dotted line represents 2-fold enrichment. However, ChIP-qPCR assay did not show enrichment of
DNA for Zeb2 binding in PDGFRβ promoter. E, Luciferase assay shows Twist1 activated the PDGFRα (left
panel) and β (right panel) gene promoter in a dose-dependent manner (**P<0.01, ***P<0.001). F, The binding of
Twist1 to the PDGFRα and β promoter region was confirmed by luciferase assay. The Twist1 conservative
binding sequence (WT) and mutant sequence (MT) for PDGFRα and β promoters was shown on the top of the
panels. The mutation of the binding sequence diminished the activation of both gene promoters by Twist1
(**P<0.01, ***P<0.001).
00.5
11.5
22.5
3
WT MT
Rela
tive L
uc. activity
CGCCACCTGCTG WT MT
T-Pβ1
CGCATCCCACTG
-6,000
PDGFR +1 -500 -1,000 -1,500 -2,000
T-Pα
1
T-Pα
2
T-Pα
3
T-Pα
4
T-Pα
5
PDGFRβ +1 -2,000
T-Pβ
1
T-Pβ
2
T-Pβ
5
T-Pβ
6
T-Pβ
7
T-Pβ
8
T-Pβ
9
-4,000
T-Pβ
10
T-Pβ
3
T-Pβ
4
PDGFRβ +1
Z-Pβ
1
Z-Pβ
5
Z-Pβ
6
Z-Pβ
2
Z-Pβ
3
Z-Pβ
4
+2,000 -2,000 -4,000
00.5
11.5
22.5
33.5
4
Twist1
LacZ
TCACACATGGAA WT MT TCAATCACAGAA
T-Pα2
WT MT
Rela
tive L
uc. activity
** ***
Twist1
LacZ
-Foxq1-/Twist1-/PDGFRs-
-Foxq1+/Twist1+/PDGFRs+
*
Supplementary Figure S4. Correlation and survival analysis of Foxq1 and Twist1 with PDGFRα
and PDGFRβ in tumor samples. A, The expression correlation of Foxq1 or Twist1 with PDGFRα and
β in breast cancer. Level 3 gene expression (RNAseV2) data of breast tumors from The Cancer
Genome Atlas (TCGA) database was used for the analysis. In the table, the upper panels (highlighted
in green) show correlation coefficients and the lower panels (highlighted in orange) include the
corresponding correlation p-values. B, The Kaplan-meier plot shows that overexpression of Foxq1 and
Twist1 with PDGFRs predicts poor survival of breast cancer patients in TCGA dataset (*P<0.05). C,
The expression correlation of Foxq1 or Twist1 with PDGFRα and β in Uterine corpus endometrial
carcinoma (UCEC). Level 3 gene expression (RNAseV2) data in UCEC from The Cancer Genome
Atlas (TCGA) database was used for the analysis. The table setting is same as in panel A.
Pvalue/correlation FOXQ1 Twist1 PDGFR-A PDGFR-B
FOXQ1 0.29 0.21 0.16
Twist 4E-14 0.45 0.47
PDGFR-A 4.6E-11 <2.2E-16 0.67
PDGFR-B 3.2E-4 <2.2E-16 <2.2E-16
Breast cancer of TCGA database A
B
Pvalue/Correlation FOXQ1 TWIST1 PDGFRA PDGFRB
FOXQ1 0.164 0.16 0.143
TWIST1 0.00227 0.375 0.32
PDGFRA 0.0431 2.33E-11 0.709
PDGFRB 0.0107 3.36E-10 0
C UCEC of TCGA database
Supplementary Figure S5. Effect of PDGFRα and PDGFRβ on cell migration, invasion and stem
cell characteristics. A-B, Knockdown of PDGFRα (A) and PDGFRβ (B) expression in HMLE /Foxq1
cells. Five different shRNAs specifically targeting PDGFRα or β were tested. β-actin was used as a
protein loading control. C-D, The effect of PDGFRs silencing on cell migration and invasion of
HMLE/Foxq1 cells. All stable cell lines with significant inhibition of PDGFRα and β expression showed
significant decrease in cell migration (C) (*P<0.05) and invasion (D) ( **P<0.01). E Effect of PDGFRs on
Foxq1 induced CD44high/CD24low cell population. Two PDGFRα knockdown cells showed minor cell
population changes, while two PDGFRβ knockdown cell models showed marked decrease of
CD44high/CD24low cell population. F, Summary of the effect of PDGFRα and PDGFRβ knockdown on
mammosphere formation of HMLE/Foxq1 cells (*P<0.05, **P<0.01). G, Representative figures of
mammosphere formation of HMLE/Foxq1 cells with PDGFRα and PDGFRβ knockdown. H,
Immunofluoresence assay showed no expression change of E-cadherin and Vimentin in HMLE/Foxq1
cells with individual and double knockdown of PDGFRα and β.
HMLE/Foxq1
α1 α5 β3 NT β4 HMLE/LacZ
H
E-c
adherin
V
imentin
HMLE/ LacZ NT 1 β4 1β4
HMLE/Foxq1
F
*
**
HMLE/Foxq1
α1 α5 β3 NT β4
*
HMLE/LacZ
0
40
60
20
80
Nu
mb
er
of
ma
mm
osp
he
res
(pe
r1,0
00
ce
lls)
-PDGFRα
α1 α2 α3 α4 α5 NT
-β-actin
-PDGFRβ
-β-actin
β1 β2 β3 β4 β5 NT A B
C D
Rela
tive
Invaded C
ells
β3 β4 NT α1 α5
** ** ** **
G
Rela
tive
Mig
rate
d C
ells
** *
β3 β4 NT α1 α5
** **
CD24
CD
44
HMLE/Foxq1
α1 α5 β3 NT β4
HMLE/LacZ
E
68.0 83.0 84.5 88.2 65.7 0.2
Supplementary Figure S6. Imatinib treatment on PDGFRs phosphorylation, mammosphere
formation and EMT. A, The expression of c-abl, kit and PDGFRs in HMLE cells with Foxq1 or
control LacZ overexpression was detected by real time RT-PCR (*P<0.05, **P<0.01). B,
Overexpression of Foxq1 leads to phosphorylation of PDGFRα and β, but not the c-ABL and c-kit. C,
imatinib (Ima) treatment inhibit phosphorylation of PDGFRs. Western blot analysis was performed
using total and phosphor-PDGFRα and β antibodies. D, Representative figures for mammosphere
formation in HMLE/Foxq1 cells after treatment of different doses of imatinib. E, No expression
changes for epithelial and mesenchymal markers in HMLE/Foxq1 cells with different doses of imatinib
treatment were detected by western blotting assay. F, Immunofluoresence assay shows no
expression change of E-cadherin and Vimentin proteins in HMLE/Foxq1 cells with different doses of
imatinib treatment.
B A
Rela
tive e
xpre
ssio
n
0 2.5 5
-β-actin
-PDGFRβ
-PDGFRα
-p-PDGFRβ
-p-PDGFRα
Ima (µM)
*
**
PDGFRα PDGFRβ c-ABL c-Kit 0
2
3
4
1
5
6
7 Foxq1 LacZ
C
-p-ABL
-p-Kit
-β-actin
-p-PDGFRβ
-p-PDGFRα
LacZ Foxq1
0 2.5 5
HMLE/Foxq1, Ima (µM) HMLE/ LacZ
0 2.5 5
HMLE/Foxq1, Imatinib (µM)
10
HMLE/LacZ
E-c
adherin
V
ime
ntin
D
E F
HMLE/Foxq1
Dox (nM)
C
Rela
tive s
urv
ived c
ells
Rela
tive s
urv
ived c
ells
***
B
** ***
** **
**
**
α1 β4 α1β4
NT
Pac (nM) 0 2 4
Rela
tive s
urv
ived c
ells
α1 β4 α1β4
NT
*
***
**
* **
***
Dox (nM) 0 40 80
NMuMG NMuMG D
Dox (nM) Pac (nM)
Rela
tive s
urv
ived c
ells
Rela
tive s
urv
ived c
ells
HMLER/Foxq1 HMLER/Foxq1
Ima 0 µM
Ima 5 µM
***
HMLE/Foxq1
Ima 0 µM
Ima 5 µM
Ima 0 µM
Ima 5 µM Ima 0 µM
Ima 5 µM
Pac (nM)
0
0.05
0.10
0.15
0.20 1.00
1.05
Rela
tive s
urv
ived c
ells
0
0.05 0.10 0.15 0.20
1.00 1.05
0.25 0.30 0.35 0.40
Supplementary Figure S7. The Effect of PDGFRs on chemoresistance. A and B, Chemoresistance
of HMLE/Foxq1 (A) and HMLER/Foxq1 (B) cells with individual or double PDGFRα and β knockdown
was analyzed by an MTT assay after treatment with various doses of doxorubicin left panel) and
paclitaxel (right panel). Results are presented as relative cell survival compared to the non-treatment
control (*P<0.05, **P<0.01 and ***P<0.001). C, Increased sensitivity of HMLE/Foxq1 cells to doxorubicin
(left panel) or paclitaxel (right panel) in the presence of imatinib (***P<0.001). D, NMuMG cells shows
no increased sensitivity to doxorubicin (left panel) and paclitaxel (right panel) in the presence of imatinib
(P>0.05).
HMLE/Foxq1
0 1 2
0.2
0.4
0.6
0.8
1.2
0
1.0
A
0 0.05 0.10 0.15 0.20
1.00 1.05
0.25 0.30
1.10 HMLE/Foxq1 R
ela
tive s
urv
ived c
ells
Dox (nM) 0 50 100
α1 β4 α1β4
NT
***
** **
***
*** ***
Rela
tive s
urv
ived c
ells
**
α1 β4 α1β4
NT
** ** **
** * **
Pac (nM)
Supplementary Figure S8. Immunohistochemistry (IHC) assays using tumor samples and our
working model. A, Tumors were collected from the HMLER/Foxq1 implanted NCR nu/nu mice, and
the prepared tissue slides were analyzed with IHC using anti-BAX, p27 and Cleaved Caspase 8
antibodies, as well as H&E staining. The origins of all four tumor samples are indicated at the top of
the panels. Scale bar, 50 µm. B, Tumors were collected from 4T1 implanted BALB/c mice. The other
settings are the same as panel A. C. Schematic model of the molecular mechanism underlying
Foxq1/PDGFR-driven breast cancer oncogenesis and chemoresistance. Foxq1, an EMT promoting
transcription factor, simultaneously regulates PDGFRα and β genes through direct or indirect
mechanisms. Twist1 and Zeb2 are the mediators of Foxq1’s indirect regulatory mechanism. This study
reveals the central regulatory role of Foxq1 in the TGFβ and PDGF signaling transition. Moreover, the
results of this study highlight PDGFR as important mediator of Foxq1/Twist1 promoted oncogenesis
and chemoresistance, which suggests an implication of designing novel combinational therapy for
breast cancer treatment (the dashed line indicates a relationship need further validation of in vivo
study).
Proliferation
Cell motility
Stemness
Chemoresistance
Oncogenesis
FOXQ1
ZEB2
PDGFRα
PDGFRβ
PD
GF
Rα
P
DG
FRβ
TG
F
β1
TG
Fβ
Rs
Nucle
us
Cyto
pla
sm
TWIST1
EMT
Control Ima Dox Ima+Dox
HMLER/Foxq1
p2
7
Cle
ave
d
Casp
ase
8
HE
B
AX
Control Ima Pac Ima+Pac
4T1
p2
7
Cle
ave
d
Casp
ase
8
HE
B
AX
A B
C