Soluble factors derived from tumor mammary cell lines induce a stromal mammary adipose reversion in human and mice adipose cells. Possible role of TGF-beta1 and TNF-alpha.

Abstract

In carcinomas such as those of breast, pancreas, stomach, and colon, cancer cells support the expansion of molecular and cellular stroma in a phenomenon termed desmoplasia, which is characterized by a strong fibrotic response. In the case of breast tissue, in which stroma is mainly a fatty tissue, this response presumably occurs at the expense of the adipose cells, the most abundant stromal phenotype, generating a tumoral fibrous structure rich in fibroblast-like cells. In this study, we aimed to determine the cellular mechanisms by which factors present in the media conditioned by MDA-MB-231 and MCF-7 human breast cancer cell lines induce a reversion of adipose cells to a fibroblastic phenotype. We demonstrated that soluble factors generated by these cell lines stimulated the reversion of mammary adipose phenotype evaluated as intracellular lipid content and expression of C/EBP alpha and PPAR gamma. We also demonstrated that exogenous TGF-beta 1 and TNF-alpha exerts a similar function. The participation of both growth factors, components of media conditioned by tumoral mammary cells, on the expression and nuclear translocation of C/EBP alpha and PPAR gamma was tested in 3T3-L1 cells by interfering with the inhibitory effects of media with agents that block the TGF-beta 1 and TNF-alpha activity. These results allow us to postulate that TGF-beta 1 and TNF-alpha present in this media are in part responsible for this phenotypic reversion.

Full text

Page 1

LETTER TO THE EDITOR
Soluble factors derived from tumor mammary cell lines induce
a stromal mammary adipose reversion in human and mice
adipose cells. Possible role of TGF-b1 and TNF-a
Javier Guerrero Æ Æ Nicola ´s Tobar Æ Æ Mo ´nica Ca ´ceres Æ Æ
Lorena Espinoza Æ Æ Paula Escobar Æ Æ Javier Dotor Æ Æ
Patricio C. Smith Æ Æ Jorge Martı ´nez
Published online: 1 August 2009
? Springer Science+Business Media, LLC. 2009
Abstract
stomach, and colon, cancer cells support the expansion of
molecular and cellular stroma in a phenomenon termed
desmoplasia, which is characterized by a strong fibrotic
response. In the case of breast tissue, in which stroma is
mainly a fatty tissue, this response presumably occurs at
the expense of the adipose cells, the most abundant stromal
phenotype, generating a tumoral fibrous structure rich in
fibroblast-like cells. In this study, we aimed to determine
the cellular mechanisms by which factors present in the
media conditioned by MDA-MB-231 and MCF-7 human
breast cancer cell lines induce a reversion of adipose cells
to a fibroblastic phenotype. We demonstrated that soluble
factors generated by these cell lines stimulated the rever-
sion of mammary adipose phenotype evaluated as intra-
cellular lipid content and expression of C/EBPa and
PPARc. We also demonstrated that exogenous TGF-b1 and
TNF-a exerts a similar function. The participation of both
growth factors, components of media conditioned by
In carcinomas such as those of breast, pancreas,
tumoral mammary cells, on the expression and nuclear
translocation of C/EBPa and PPARc was tested in 3T3-L1
cells by interfering with the inhibitory effects of media
with agents that block the TGF-b1 and TNF-a activity.
These results allow us to postulate that TGF-b1 and TNF-a
present in this media are in part responsible for this phe-
notypic reversion.
Keywords
factors ? TGF-b1 ? TNF-a
Desmoplasia ? Adipogenic transcription
Introduction
It is currently accepted that tumoral progression not only
depends on the intrinsic malignancy of tumor cells but also
on the surrounding microenvironment composed of stromal
cells and the extracellular matrix (ECM) [1]. Moreover, it
has been described that tumor cells are able to modify the
adjacent stroma creating a functional structure that favors
tumor progression [2].
Breast tumors belong to a group of neoplastic lesions
which, under the influence of tumoral cell products, orig-
inate a fibrous structure responsible for the dense and hard
consistency of the tumoral mass. This trait also constitutes
a factor that increases the relative risk of tumor recurrence
[3]. Either in the functional origin or in the cellular com-
position of this fibrotic structure, myofibroblasts are the
predominant stromal phenotype which secretes abundant
amount of collagen and other extracellular matrix (ECM)
proteins through an active phenomenon known as ‘‘Des-
moplastic Response’’ [4]. Some of the changes in the
composition of ECM, derived from the desmoplastic
response, induce changes in breast density that has been
identified as a predisposing factor for breast cancer that
J. Guerrero ? N. Tobar ? M. Ca ´ceres ? J. Martı ´nez (&)
Laboratory of Cellular Biology, INTA, University of Chile,
Casilla 138, Santiago 11, Chile
e-mail: jmartine@inta.cl
L. Espinoza ? P. Escobar
East Metropolitan Health Service, Dr. Luis Tisne ´, Santiago,
Chile
J. Dotor
Digna Biotech, Madrid, Spain
P. C. Smith
Laboratory of Periodontal Physiology, Faculty of Medicine,
Pontifical Catholic University of Chile, Santiago, Chile
123
Breast Cancer Res Treat (2010) 119:497–508
DOI 10.1007/s10549-009-0491-1

Page 2

confers a higher risk compared with women with fatty
breast [5].
The desmoplastic reaction has been attributed to fibro-
blast-like cells which may derive from adipose tissue in
mammary tumors [6]. Fibroblast accumulation can occur
due to the activity of specific growth factors produced by
tumoral cells. Using 3T3-L1 cells and human adipose
fibroblasts, it has been demonstrated that TNF-a and IL-11
derived from human tumor mammary cell lines inhibit
adipocyte differentiation, a previous step to expansion of
fibroblastic compartment without modifying fibroblast
proliferation [6].
On the other hand, it has been also proposed that
Platelet-derived Growth Factor (PDGF) is a paracrine
factor produced by mammary cells whose primary target
are stromal cells and which is considered the initiator of
tumor desmoplasia [4]. Finally, it has been suggested that
TGF-b1 may be one of the main factors involved in des-
moplastic response. In breast cancer TGF-b1 is highly
expressed, preferentially at the advancing edges of primary
tumors and in lymph node metastasis, suggesting that it
plays a role in interacting with neighboring stromal cells
[7].
Epithelial control of mammary adipose cells is observed
under physiologic conditions as well. During pregnancy
and lactation, reproductive hormones induce the expansion
and terminal differentiation of the mammary epithelium
into secretory, milk-producing, lobular alveoli in a process
that also includes the dedifferentiation of adipocytes into
tiny preadipocytes [8].
CCAAT/enhancer binding proteins (C/EBPs) and per-
oxisome proliferator-activated receptor-gamma (PPARs)
are two families of transcription factors that play a critical
role in either the onset of adipocyte differentiation or the
maintenance of the fully differentiated adipocyte pheno-
type by transactivating adipocyte specific genes [9].
Inhibiting PPARc activity, either in 3T3-L1 adipocytes or
in the adipose tissue of mice was shown to lead to dedif-
ferentiation or adipocyte death [10]. It has also been
demonstrated that C/EBPa is a well-characterized factor
that mediates the expression of genes characteristic of the
terminally differentiated state and also acts as inhibitor of
mitotic growth in most cell lines tested [11].
In the present work, we attempt to unravel some of the
cellular mechanisms of mammary desmoplastic response
by evaluating whether cell-secreted soluble factors pro-
duced by human mammary cell lines (MCF-7 and MDA-
MB-231) are able to induce reversion of mature mammary
adipocytes of human and mouse origin to the fibroblastic
phenotype, in addition to the well-known inhibitory role on
adipose differentiation. To accomplish this, we used human
mammary-derived adipocyte and fully differentiated mur-
ine 3T3-L1 cells to identify the main factors secreted by
the cell lines responsible for the reversion process and the
impact of these factors on the expression and abundance of
PPARc and C/EBPa. Our results, obtained from cell culture
studies and immunohistochemical analysis of human tumor
samples, demonstrated that TGF-b1 and TNF-a of epithe-
lial origin are in part responsible for the adipocyte rever-
sion and that PPARc and C/EBPa expression seem to be a
molecular target of these tumoral factors.
Materials and methods
Antibodies and reagents
Antibodies anti-PPARc and C/EBPa were obtained from
Santa Cruz Biotechnology, (Santa Cruz, CA). Anti-b-actin
was supplied by Sigma (St. Louis, MO), anti-TGF-b was
from Chemicon International, Inc. (Temecula, CA), anti-
human TNF-a (neutralizing of TNF-a) was obtained from
R&D System, Inc. (Minneapolis, MN). The P144 peptide
(TSLDASIIWAMMQN, encompassing aminoacids 730–
743) interfering with TGF-b1 binding to its cellular
receptors [12] was produced by Digna Biotech (Madrid,
Spain). Recombinant TNF-a and TGF-b were purchased
from R&D systems and Calbiochem (La Jolla, CA),
respectively. Insulin, Dexamethasone, isobutylmethylxan-
thine (IBMX), Indomethacin, and Type I collagenase were
purchased from Sigma Chemical Co. Dulbecco’s Modified
Eagle’s Medium (DMEM), fetal bovine serum (FBS), and
bovine serum (BS) were purchase from GibcoTM(Grand
Island, NY).
Patients and normal tissue acquisition
Eight patients undergoing surgical mammary reduction at
the East Metropolitan Health Service (Santiago, Chile)
were recruited for the study. The experimental protocols
were previously approved by institution ethics committee
and all patients gave their informed consent for the pro-
cedure. All subjects were healthy and had no evidence of
diabetes according to routine laboratory tests. Mammary
adipose tissue biopsies were obtained at the time of surgery
and immediately transported to the laboratory in sterile
DMEM for processing.
Adipocyte isolation and three-dimensional
collagen gel culture system
Samples of mammary adipose tissue were minced finely
and digested in a collagenase solution (0.25 mg/ml Type I
collagenase, 3% bovine serum albumin in DMEM) with
shaking at 37?C for 1 h. The resultant digested material
498 Breast Cancer Res Treat (2010) 119:497–508
123

Page 3

was then centrifuged at 1,500g for 5 min and the super-
natant was washed three times with 3% BSA in DMEM.
Under these experimental conditions, a floating layer that
contains only very low density cells (that we identify as
unilocular mature adipocytes) was generated. This cellular
fraction was extensively washed three times with HBSS.
The remaining more dense phenotypes present in the breast
sample, sediment in the cellular pellet.
A collagen gel solution was prepared as previously
described [13]. Briefly, 8 volumes of type I rat tail collagen
was mixed with 1 volume of tenfold concentrated DMEM
and 1 volume of reconstruction buffer (2.2 g of NaHCO3
and 4.77 g of HEPES in 100 ml of 0.05 N NaOH at 4?C) to
obtain a homogeneous solution. The collagen solution was
mixed with the purified mature adipocytes at room tem-
perature and 1 ml of collagen gel solution containing
nearly 105cells was placed in a 35 mm culture dish. The
culture dishes were immediately warmed to 37?C to allow
the gel to form and were covered with 2 ml of DMEM
overnight. Afterward, cultures were treated with increasing
concentrations of conditioned media (CM) from mammary
tumoral cells lines MCF-7 and MDA-MB-231 for 10 days.
Collagen was obtained from rat tail tendon as previously
described [14].
Mammary cell lines and preparation
of conditioned media
Mammary tumoral cells lines MCF-7 and MDA-MB-231
cells were purchased from ATCC (Manassas, VA) and
were grown in a phenol red-free DMEM/F12 enriched with
10% FCS. To prepare conditioned media, approximately
105cells/cm2of MCF-7 and MDA-MB-231 were cultured
in standard conditions and further incubated for 48 h in
serum-free DMEM. After this, media were collected and
centrifuged for 5 min at 2,000 rpm to clarify.
Oil red O staining
The intracellular lipid content was evaluated with the
lipophilic dye Oil Red O [15]. To do so, media was
removed and cells in culture were fixed and dehydrated for
5 min with 2 ml of 100% isopropanol. Samples were then
stained for 1 h with 2 ml of a saturated solution of Oil Red
O in 60% (v/v) isopropanol and then washed twice with
distilled water for 15 min.
ELISA assay
To determine TGF-b and TNF-a concentrations in mam-
mary tumoral cell conditioned media, TGF-b (559119) BD
Bioscience (San Diego, CA) and TNF-a (T916008) US
Biological (Swampscott, MA) ELISA kits were used. The
samples (in quadruplicate) were processed strictly follow-
ing the supplier’s instructions. At the end of the procedure,
absorbance was immediately determined at a wavelength
of 450 nm for TGF-b and 405 nm for TNF-a, with cor-
rections at 570 and 650 nm, respectively. TGF-b and TNF-
a concentrations were calculated using standard curves
prepared for each individual determination.
Immunohistochemistry
Immunostaining for TGF-b, TNF-a, PPARc, and C/EBPa
was performed on 5 lm sections of formalin fixed, paraf-
fin-embedded human mammary tissue biopsies. Tissue
sections were deparaffinized in xylene and hydrated in a
graded sequence of ethanol solutions. For antigen retrieval,
sections were pretreated in a microwave in citrate buffer
(pH 6.0). After cooling, nonspecific binding was blocked
with diluted serum (4% normal goat serum) followed by
incubation with antibodies against TGF-b (1:300), TNF-a
(1:150), PPARc (1:50), and C/EBPa (1:2,000) at room
temperature in a humidified chamber. Negative controls
were analyzed on adjacent sections incubated without pri-
mary antibody. After incubation with the primary antibody,
sections were washed with PBS and subsequently treated
using the corresponding biotinylated secondary antibody
from Ultravision ONE kit, Thermo Fisher Scientific (TL-
015-HDJ, Fremont, CA) according to the manufacturer’s
protocol. Peroxidase activity was visualized using the 3,30-
diaminobenzidine and sections were counterstained with
haematoxylin. Assessment of staining intensity and distri-
bution was made using the semiquantitative histologic score
(HSCORE) system as described by Budwit-Novotny et al.
[16].Thisscorewascalculatedinfourcontrolandninetumor
samples using the following equation: HSCORE = RPi
(i ? 1), where i = intensity of staining with a value of 1, 2,
or 3, (weak, moderate or strong, respectively) and Pi the
percentage of stained mammary adipose cells for each
intensity, varying from 0 to 100%. This HSCORE has been
previously reported [17, 18].
Adipocyte 3T3-L1 differentiation
3T3-L1 cells were purchased from ATCC (Manassas,
VA) and cultured in DMEM media supplemented with
10% BS, 100 lg/ml streptomycin and 100 units/ml of
penicillin in a humidified atmosphere with 5% CO2 at
37?C. Two days after reaching confluence, the induction
of adipocyte differentiation was started by culturing
for 2 days with adipogenic medium containing 1 lM
Dexamethasone, 0.5 mM IBMX, and 400 nM insulin
Breast Cancer Res Treat (2010) 119:497–508499
123

Page 4

solution in DMEM 10% FBS. For the following 8 days,
cells were incubated with DMEM 10% FBS containing
400 nM insulin alone. During this period, media was
replaced every 2 days.
Western blots
Mammary mature adipocyte cells or differentiated adipo-
cyte 3T3-L1 cells were lysed with a lysis buffer (50 mM
Hepes, pH 7.4, 150 mM NaCl, 2 mM MgCl2, 2 mM
EGTA, 1% Triton X-100, 10% glycerol, 2 mM PMSF,
2 lg/ml pepstatin, 2 lg/ml leupeptin and 1 mM sodium
orthovanadate) at 4?C. Equal amounts of proteins from
different treatments were resolved by SDS–PAGE and
analyzed by immunoblotting with antibodies anti-PPARc,
C/EBPa using the ECL chemiluminescence detection kit
(Amersham, Arlington Heights, FL).
Immunofluorescence
3T3-L1 cells were plated on cover slips in 24-well plates
and differentiated as described above. After the treat-
ment with mammary cell CMs and inhibitors (neutral-
izing Anti-TNF-a and P144), cells were washed in PBS
with 1 mM Ca?2, fixed with 4% paraformaldehyde for
20 min and permeabilized with 0.25% Triton X-100 for
10 min at RT. Samples were blocked with 10% goat
serum in PBS for 1 h and then co-incubated at 4?C
overnight with the primary antibodies in blocking solu-
tion: monoclonal anti-PPARc (1:50) and polyclonal anti-
C/EBPa (1:100). The cover slips were washed in PBS
and incubated with secondary antibodies (all from
Molecular Probes) diluted 1:100 in blocking solution:
Alexa 488-anti mouse, Alexa 546-anti rabbit at room
temperature for 2 h. Cover slips were washed in PBS
and water. Finally, samples were mounted in fluores-
cence mounting media (Dako).
Adipocyte size image analysis
Sections of 5 lm of paraffin-embedded human mammary
tissue obtained from adjacent regions of both normal and
tumoral samples were stained with haematoxylin and eosin
to evaluate adipocyte size as described by Chen et al. [19],
with some modifications. Briefly, images of the histology
sections were obtained in a jpg format at 109 magnifica-
tion and were converted into a binary format (8-bit) with
Adobe PhotoShop 5.0 (Adobe Systems, San Jose, CA) and
Image J (National Institutes of Health, http://rsbweb.nih.
gov/ij/). The binary images were compared with the
original images to ensure an accurate conversion. For the
determination of the total number and cross-sectional areas
of adipocytes, the commands Image/Adjust/Threshold and
then Analyze/Analyze Particules were used. Adipocyte
cross-sectional areas were expressed as pixel number, and
the results saved as an Excel document for analysis.
Results
Media conditioned by human mammary tumoral
cells revert to mammary adipose phenotype
To test whether soluble factors produced by human tumoral
mammary cells affect the lipid content of mammary adi-
pose tissue, we seeded recently dispersed human mammary
fat cells into a semisolid collagen solution and cultured
them in the presence or absence of media conditioned
(MC) by the weakly invasive MCF-7 and the strongly
invasive MDA-MB-231 mammary cells.
After 10 days, cells were stained with Red Oil O and the
proportion of fibroblast-like cells was scored. Figure 1A
shows a representative image of human adipose mammary
cells exposed to 50% MC by MDA-MB-231. As displayed
in this image, a high proportion of mammary fatty cells lost
their lipid content and acquired a fibroblast shape. Fig-
ure 1B shows the quantitative analysis of these experi-
ments using a range of concentrations of MC derived from
MDA-MB-231 and MCF-7 cells. This data indicated that
MC derived from MDA-MB-231 cells was more effective
in reverting the adipose phenotype to spindle-shaped cells
(Fig. 1B).
It has been proposed that TGF-b1 and TNF-a, two sol-
uble factors produced in a high proportion by tumoral cells,
also play a role in the control of cellular lipid content
[6, 20]. To test whether these factors induce lipid loss in
our system, we assessed their effectiveness in the reversion
of adipose phenotype and found that both factors are
capable of generating a conversion of adipose cells into
fibroblasts (Fig. 1C).
TGF-b1 and TNF-a are differentially expressed
in human mammary cell lines
Next, we analyzed the presence of TGF-b1 and TNF-a in
MC by MCF-7 and MDA-MB-231 cells using an ELISA
assay. Figure 2 shows that while both cell lines produced a
fairly similar amount of TNF-a, the more invasive MDA-
MB-231 cells produced five times more TGF-b1 than the
less invasive MCF-7 line.
500 Breast Cancer Res Treat (2010) 119:497–508
123

Page 5

TGF-b1 and TNF-a are expressed in human
mammary tumors
To evaluate the cellular origin and distribution of TGF-b1
and TNF-a in mammary tumor samples, we analyzed by
immunohistotochemistry the presence of these factors in
samples of human ductal infiltrant mammary tumors. As
Fig. 3 shows, both factors were expressed only by tumoral
cells located at the invasive front without a relevant
expression in the stromal adipose cells.
TGF-b1 and TNF-a inhibit the expression
of transcriptional factors involved in the maintenance
of adipose phenotype
The maintenance of the adipose phenotype largely depends
on the expression of specific transcription factors such as C/
EBPaandPPARc,which are involved inthe maintenanceof
adipose homeostasis [21]. To assess whether the expression
of these transcriptional factors are sensitive to TGF-b1 and
TNF-a, we analyzed by western blotting, the expression of
0
10
20
30
40
50
%MC MCF-7%MC MDA-231
Control
25
50
25
50
**
**
Control
TGF-β
TNF-α
% Fibroblasts-like cells
0
10
20
30
40
50
% Fibroblasts-like cells
a
a
a
I
b
b
b
b
II
(A)
(B)(C)
Fig. 1 Adipocyte reversion in three-dimensional collagen gel culture.
A 105human mature mammary adipocytes were cultured in 1.5 ml
semisolid collagen gel for 10 days in the absence (I) or presence (II)
of 50% medium conditioned by MDA-MB-231 cells. Cells were
stained with oil red O that identified lipid content in mature spherical
adipocytes (a) and elongated cells with a fibroblast phenotype (b). B
Quantification of fibroblast-like cells in cultures treated with
increasing proportions of media conditioned by mammary tumoral
cells MCF-7 and MDA-MB-231. C Quantification of fibroblast-like
cells in collagen matrix cultures treated for 10 days with 10 ng/ml of
TGF-b and TNF-a. Bars represent mean ± SE from five independent
experiments. A Kruskal–Wallis one-way ANOVA followed by a
Dunn’s post hoc analysis was used to determine significant differ-
ences from control. * P\0.05; ** P\0.001
0
50
100
150
200
250
300
350
400
450
MCF-7MDA-MB-231
(pg*ml-1 CM/106 cells/48h)
0
5
10
15
20
25
30
35
40
45
50
(pg*ml-1 CM/106 cells/48h)
MCF-7MDA-MB-231
TGF–β1
TNF–α
Fig. 2 Quantification of TGF-b and TNF-a in mammary cancer cell
conditioned media. Samples of media conditioned by mammary
tumoral cell lines MCF-7 and MD-MDA-231 were collected and
immediately quantified by an ELISA immunoassay for TGF-b and
TNF-a as described under ‘‘Materials and methods’’. Data represent
the mean ± SE
Breast Cancer Res Treat (2010) 119:497–508501
123