Carcinogenesis vol.29 no.12 pp.2243–2251, 2008
Advance Access publication August 19, 2008
Fibulin-5 initiates epithelial–mesenchymal transition (EMT) and enhances EMT
induced by TGF-b in mammary epithelial cells via a MMP-dependent mechanism
Yong-Hun Leey, Allan R.Albig2,y, MaryAnn Regner,
Barbara J.Schiemann1and William P.Schiemann?
Department of Pharmacology and1Department of Obstetrics and Gynecology,
University of Colorado Health Sciences Center, Aurora, CO 80045, USA
2Present address: Department of Life Sciences, 600 Chestnut Street, Indiana
State University, Terre Haute, IN 47809, USA
?To whom correspondence should be addressed. Department of Pharmacology,
University of Colorado Health Sciences Center, RC1 South Tower, Room
L18-6110, 12801 East 17th Avenue, PO Box 6511, Aurora, CO 80045, USA.
Tel: þ1 303 724 1541; Fax: þ1 303 724 3663;
Epithelial–mesenchymal transition (EMT) is a normal physiolog-
ical process that regulates tissue development, remodeling and
repair; however, aberrant EMT also elicits disease development
in humans, including lung fibrosis, rheumatoid arthritis and can-
cer cell metastasis. Transforming growth factor-b (TGF-b) is
a master regulator of EMT in normal mammary epithelial cells
(MECs), wherein this pleiotropic cytokine also functions as a po-
tent suppressor of mammary tumorigenesis. In contrast, malig-
nant MECs typically evolve resistance to TGF-b-mediated
cytostasis and develop the ability to proliferate, invade and me-
tastasize when stimulated by TGF-b. It therefore stands to reason
that establishing how TGF-b promotes EMT may offer new in-
sights into targeting the oncogenic activities of TGF-b in human
breast cancers. By monitoring alterations in the actin cytoskeleton
and various markers of EMT, we show here that the TGF-b gene
target, fibulin-5 (FBLN5), initiates EMT and enhances that in-
duced by TGF-b. Whereas normal MECs contain few FBLN5
transcripts, those induced to undergo EMT by TGF-b show sig-
nificant upregulation of FBLN5 messenger RNA, suggesting that
EMT and the dedifferentiation of MECs override the repression
of FBLN5 expression in polarized MECs. We also show that
FBLN5 stimulated matrix metalloproteinase expression and ac-
tivity, leading to MEC invasion and EMT, to elevated Twist ex-
pression and to reduced E-cadherin expression. Finally, FBLN5
promoted anchorage-independent growth in normal and malig-
nant MECs, as well as enhanced the growth of 4T1 tumors in
mice. Taken together, these findings identify a novel EMT and
tumor-promoting function for FBLN5 in developing and pro-
gressing breast cancers.
Epithelial–mesenchymal transition (EMT) is a normal physiological
process that occurs when polarized epithelial cells acquire a mesen-
chymal phenotype that resembles those of fully differentiated fibro-
blasts or myofibroblasts (1–3). Whereas physiological EMT is
essential for embryonic development, for wound healing and for tis-
sue homeostasis, remodeling and repair, pathological reactivation of
EMT in adult tissues can engender disease development in humans,
including chronic inflammation, lung fibrosis, rheumatoid arthritis
and cancer cell metastasis (1–3). Transforming growth factor-b
(TGF-b) was first described as inducer of EMT in normal mammary
epithelial cells (MECs) (4) and now is recognized as a master regu-
lator of EMT in a variety of cell types and tissues (5). Phenotypically,
EMT stimulated by TGF-b proceeds through a highly coordinated
spatiotemporal sequence of events that includes the (i) disassembly
of cell–cell junctions; (ii) reorganization of the actin cytoskeletal; (iii)
loss of epithelial polarity and (iv) remodeling of cell–matrix adhe-
sions (5,6). Malignant cells also undergo EMT, but in doing so, they
acquire additional characteristics that facilitate their ability to un-
dergo intravasation and extravasation and to sustain metastatic growth
in distant locales (1). Thus, oncogenic EMT usually manifests in
a genetic and cellular context that is highly abnormal and distinct
from that observed during physiologic EMT. Moreover, oncogenic
EMT enables cancer cells to acquire invasive and metastatic pheno-
types, and consequently, to promote the dissemination of cancer cells
beyond their tissue of origin, which represents the most lethal and
deadly aspect of cancer (6).
of mammary gland development, during which it potently suppresses
mammary tumorigenesis (6–8). In contrast, developing and progress-
ing breast cancers frequently inactivate the tumor suppressing activ-
ities of TGF-b, an event that facilitates the growth and metastatic
spread of malignant MECs stimulated by TGF-b. Although the mo-
lecular mechanisms that enable mammary tumorigenesis to convert
the cellular response of MECs to TGF-b remain to be fully elucidated,
it is tempting to speculate that this switch in TGF-b function may
reflect its ability to initiate and stabilize EMT in malignant MECs.
endothelial cells (9,10) and to establish FBLN5 as a multifunctional
signaling molecule that (i) regulates the proliferation, motility and in-
vasion of normal and malignant cells both in vitro and in vivo (9–11);
(ii) antagonizes endothelial cell activities coupled to angiogenesis both
invitro and invivo(10,11) and (iii) inhibits thegrowth of fibrosarcomas
in mice (11). We also observed tumorigenesis to significantly repress
the synthesis of FBLN5 transcripts in a variety of human cancers, in-
cluding those of the breast (9). Unfortunately, the identity of the
FBLN5-expressing cell types targeted by mammary tumorigenesis re-
main unknown, as do the direct effects of FBLN5 on the behaviors of
normal and malignant MECs. Interestingly, recent evidence indicates
that Fibulin family members, including FBLN5, are expressed in a de-
velopmentally regulated manner to regions of EMT during arterial,
endocardial cushion tissue, neural crest and mesenchymal tissue de-
velopment (12–14). Thus, FBLN5 may be an important and novel
regulator of normal EMT during embryonic development, as well as
an inducer of oncogenic EMT during the development and progression
of human breast cancers. The aim of the present study was to establish
the function of FBLN5 in regulating EMT in normal and malignant
MECs and in regulating the growth of mammary tumors in mice.
Materials and methods
Normal and malignant human mammary tissue microarrays were purchased
from US Biomax (catalog # BR241; Rockville, MD), Millipore (catalog #
TMA1201 and TMA1010; Temecula, CA) and Biomeda (catalog#M90; Foster
City, CA) and were deparaffinized in xylene and rehydrated through a graded
series of alcohols prior to inactivating endogenous peroxidase activity by in-
cubation in 3% hydrogen peroxide for 5 min at room temperature. Antigen
retrieval was performed by pressure cooking the sections in 10 mM sodium
citrate/0.5% Tween 20 (pH 6.0) for 10 min at 120?C, at which point non-
specific binding sites were blocked by incubation of the sections in phos-
phate-buffered saline containing 0.1% Tween-20/1% fetal bovine serum for
1 h at room temperature. Afterward, anti-FBLN5 antibodies [Santa Cruz Bio-
technology, Santa Cruz, CA; at a 1:100 dilution or rabbit polyclonal at 1:100 as
Abbreviations: ECM, extracellular matrix; EMT, epithelial–mesenchymal
transition; FBLN5, fibulin-5; GFP, green fluorescent protein; LOX, lysyl
oxidase; MEC, mammary epithelial cell; MMP, matrix metalloproteinase;
PCR, polymerase chain reaction; TGF-b, transforming growth factor-b.
yThese authors contributed equally to this work.
? The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: firstname.lastname@example.org 2243
described (15)] were applied to the tissue sections for 1 h at room temperature,
followed by additional 1 h incubation with biotin-conjugated anti-rabbit sec-
ondary antibodies (1:200; Jackson Immuno Research, West Grove, PA). The
resulting immunocomplexes were visualized using the VectaStain ABC Kit
(Vector Laboratories, Burlingame, CA) and 3,3#-diaminobenzidine (Millipore)
and subsequently were counterstained with hematoxylin prior to tissue section
dehydration and mounting with Permount Mounting Media (Fisher Scientific,
Pittsburgh, PA). Negative staining controls for these analyses comprised the
use of adjacent tissue sections that were processed in parallel in the absence of
primary antibody. The immunoreactivity against FBLN5 in normal (n 5 36)
and malignant (n 5 79) mammary tissues was scored semiquantitatively in
a blinded manner by five individuals who used the following scale: (i) 0–1
represented no-to-low staining; (ii) 1–2 represented moderate staining and (iii)
2–3 represented high staining. Data are presented as the % of normal or
malignant mammary specimens exhibiting low (average staining intensity
, 1), moderate (1 , average staining intensity , 2) or high (average staining
intensity . 2) FBLN5 immunoreactivity.
Semiquantitative real-time polymerase chain reaction assays
Total RNA from control and FBLN5-expressing NMuMG and 4T1 cells was
purified using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA) according to
the manufacturer’s recommendations. In some experiments, green fluorescent
protein (GFP)- or FBLN5-expressing NMuMG and 4T1 cells were stimulated
with TGF-b1 (5 ng/ml) in the absence or presence (10 lM) of the type I MMP-
2/3 (catalog # 444239; Calbiochem, San Diego, CA) or MMP-2/9 (catalog #
444241; Calbiochem) inhibitors. Afterward, complementary DNAs were syn-
thesized by iScript reverse transcription (Bio-Rad, Hercules, CA), which then
were diluted 10-fold in H2O and employed in semiquantitative real-time poly-
merase chain reaction (PCR) reactions (25 ll) that used the SYBR Green
system (Bio-Rad) supplemented with 5 ll of diluted complementary DNA and
concentrations were controlled by normalizing individual gene signals to their
corresponding glyceraldehyde 3-phosphate dehydrogenase RNA signals. The
oligonucleotide primer pairs used were as follows: (i) matrix metalloproteinase
(MMP)-2, forward: 5#-TAACCTGGATGCTGTCGTGGA-3# and reverse: 5#-
GCCCAGCCAGTCTGATTTGAT-3#; (ii) MMP-3, forward: 5#-TGTTCCT-
GATGTTGGTGGCTT-3# and reverse: 5#-TGTCTTGGCAAATCCGGTG-3#;
(iii) tissue inhibitor of metalloproteinases-1, forward: 5#-AAGCCTCTGT-
GGATATGCCCA-3# and reverse: 5#-AACCAAGAAGCTGCAGGCACT-3#;
(iv) tissue inhibitor of metalloproteinases-2, forward: 5#-GTCCCATGATC-
(v) tissue inhibitor of metalloproteinases-3, forward: 5#-CCCTGGCTATCAGTC-
CAAACA-3# and reverse: 5#-TGGCGTTGCTGATGCTCTT-3#; (vi) thrombo-
spondin-1, forward: 5#-GAACTCATTGGAGGTGCACGA-3# and reverse:
forward: 5#-CCCAGATAAGCCCAAAAACCCCG-3# and reverse: 5#-CCG-
CCAGAACCATCGCTCCTTG-3#; (viii) MT2-MMP, forward: 5#-CGGGTGA-
CAAGGGGAC-3#; (ix) MT3-MMP, forward: 5#-TTTGATGGGGAGGGAG-
GATTTTTGG-3# and reverse: 5#-GCGGTGGGGTCGTTGGAATGTTC-3#; (x)
E-cadherin, forward: 5#-CCCTACATACACTCTGGTGGTTCA-3# and reverse:
TCGCTATGGTTACTGCCAGCA-3# and reverse: 5#-TTGGCAAGACCTTC-
CATCGTC-3#; (xii) Twist, forward: 5#-CGGGTCATGGCTAACCTG-3# and re-
verse: 5#-CAGCTTGCCATCTTGGAGTC-3# and (xiii) glyceraldehyde 3-
GCTC-3# and reverse: 5#-GCAGGGATGATGTTCTGGGCAGC-3#.
Retroviral plasmids, transgene expression and recombinantFBLN5 production
A bicistronic retroviral vector (pMSCV-IRES-GFP) encoding for murine
FBLN5 was described previously (9) and used to infect normal murine
NMuMG and malignant, metastatic murine 4T1 MECs as described previously
(9). Cells expressing GFP were isolated and collected 48 h later on a MoFlo
Cell Sorter (Beckman Coulter, Fullerton, CA) and subsequently were ex-
panded to yield stable polyclonal populations of control (i.e. GFP) and
The synthesis of a bacterial expression vector encoding FBLN5 fused to the
C-terminus of glutathione S-transferase, as well as the purification of recombi-
nant FBLN5 from transformed Escherichia coli was described previously (11).
Cell biological assays
The effect of FBLN5 on the behaviors of normal and malignant MECs and on
their responses to TGF-b were determined as follows: (i) cell proliferation
using10 000 cellsperwell in a [3H]thymidineincorporationassayasdescribed
previously (9–11) and (ii) EMTinduced by TGF-b1 (5 ng/ml) in the absence or
presence (10 lM) of the type I MMP-2/3 or MMP-2/9 inhibitors as described
previously (16). All images were captured on a Nikon Diaphot microscope.
The ability of FBLN5 and TGF-b to alter the anchorage-independent growth
of normal and malignant MECs was performed as described previously (16).
Briefly, duplicate cultures of control (i.e. GFP) or FBLN5-expressing NMuMG
or 4T1 cells (10 000 cells per plate) were grown in 0.3% agar on a cushion of
0.6% agar in 60 mm plates. Normal and malignant MEC growth in the absence
or presence of TGF-b1 (5 ng/ml) was allowed to proceed for 14 days, where-
upon the number of colonies formed was quantified under a light microscope.
Alterations in MEC character also were monitored by immunoblotting qui-
escent and TGF-b-stimulated NMuMG cell lysates with antibodies against
various EMT markers. In doing so, clarified whole-cell extracts were resolved
on 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels,
transferred electrophoretically onto nitrocellulose membranes and blocked in
5% milk before incubation with the following primary antibodies (dilutions):
(i) anti-E-cadherin (1:1000; BD Biosciences, San Jose, CA); (ii) anti-N-cad-
herin (1:250;Cell Signaling,Danvers, MA); (iii) anti-COX-2 (1:2000;Cayman
Chemical Company, Ann Arbor, MI) and (iv) anti-b3 integrin (1:1000; Santa
Cruz Biotechnology). The resulting immunocomplexes were visualized by
enhanced chemiluminescence, and differences in protein loading were moni-
tored by reprobing stripped membranes with anti-b-actin antibodies (1:1000;
Rockland Immunology, Gilbertsville, PA).
Altered MMP activity in NMuMG and 4T1 cells was determined by mixing
control and FBLN5-expressing NMuMG and 4T1 cells (1.2 ? 106cells per
well) in 0.5 ml collagen, which subsequently was solidified in 24-well plates.
Twenty-four hours later, the medium was discarded and the collagen matrices
were removed and pelleted by microcentrifugation before fractionating super-
natants (20–80 ll/lane) through 10% sodium dodecyl sulfate–polyacrylamide
gel electrophoresis supplemented with 0.1% gelatin (Sigma, St Louis, MO).
Afterward, zymogram renaturing and developing were preformed using Novex
Zymogram buffer system according to the instructions of the manufacturer
(Invitrogen, Carlsbad, CA).
Tumor growth study
Control (GFP) and FBLN5-expressing 4T1 cells were resuspended in sterile
phosphate-buffered saline and injected orthotopically at a density of either
50 000 or 12 500 cells per injection into the mammary fat pad of 6-week-old
female Balb/C and Nude mice (four mice per condition; Jackson Laboratories,
Bar Harbor, ME). Mice were monitored daily and primary tumors were mea-
sured with digital calipers (Fisher Scientific) between days 10 and 32. Tumor
volumes were calculated using the following equation: tumor volume 5
0.5 ? x2? y, where ‘x’ is the tumor width and ‘y’ is the tumor length.
Thirty-two days post-inoculation, the mice were killed and their primary tu-
mors were excised and weighed. Animal studies were performed in accordance
with the animal protocol procedures approved by the Institutional Animal Care
and Use Committee of University of Colorado.
Mammary tumorigenesis and EMT induces FBLN5 expression
We reported previously that tumorigenesis downregulates FBLN5
messenger RNA expression in a variety of human malignancies, in-
cluding metastatic breast cancers (9). Although the identity of the
FBLN5-expressing cells targeted by mammary tumorigenesis remains
to be determined definitively, these findings pointed toward FBLN5
functioning as a tumor suppressor in mammary tissues. Along these
lines, Oncomine analyses comparing the gene expression profiles of
40 breast carcinomas versus seven normal breast tissues also found
FBLN5 messenger RNA expression to be reduced in mammary tu-
mors as compared with their normal counterparts (Figure 1A). In stark
contrast, we observed human breast cancers to contain abundant lev-
els of FBLN5 protein, and in fact, to express more FBLN5 protein
than did their normal counterparts (Figure 1B and C). This finding
was unexpected and surprising because Fibulins, including FBLN5,
typically are not expressed in epithelial cells; however, these mole-
cules are readily expressed in a variety of mesenchymal and stromal
components (12–14). Thus, the dramatic increase in FBLN5 expres-
sion by breast cancer cells could suggest that a loss in the epithelial
phenotype that occurs during MEC dedifferentiation may override the
normal mechanisms operant in repressing FBLN5 production in nor-
mal, polarized MECs. A corollary states that measures capable of
Y.-H.Lee et al.
inducing EMT in MECs may enable the reactivation of FBLN5 ex-
pression. We tested this hypothesis by administering TGF-b to normal
murine NMuMG MECs, which readily undergo EMT when treated
with TGF-b (16). In accordance with our hypothesis, TGF-b stimu-
lation of EMTin NMuMG cells significantly induced their expression
of FBLN5 (Figure 1D), suggesting that FBLN5 may participate in
Fig. 1. Mammary tumorigenesis and EMT induces FBLN5 expression. (A) The expression of FBLN5 messenger RNA (mRNA) in normal and malignant
mammary tissues was accessed using the Oncomine Research Database (http://www.oncomine.org) and are plotted on a log scale. The data set included patient
samples from seven normal mammary tissues (normal) and 40 breast carcinoma tissues (tumor) and were plotted on a log scale. Lines within the boxes, median
expression value for each group. Top and bottom edges of the boxes represent the 75th and 25th percentiles of the gene expression distribution, respectively
(?P , 0.001; Student’s t-test). (B) Normal and malignant human breast cancer tissue arrays (top panels, US Biomax; bottom panels, Millipore) were stained with
anti-FBLN5 antibodies [top panels, Santa Cruz Biotechnology; bottom panels, rabbit polyclonal (15)]. FBLN5 protein expression was upregulated dramatically in
human breast cancers as compared with normal controls. IDC, invasive ductal carcinoma; IVTS, intravasucular tumor spread. (C) Summary of average signal
intensities (ASIs) for FBLN5 immunohistochemistry in normal (n 5 36) and malignant (n 5 79) mammary tissues presented as the % of mammary specimens
examined. (D) NMuMG cells were incubated for 36 h in the absence or presence of TGF-b1 (5 ng/ml) to induce EMT. Afterward, total RNA was isolated and
subjected to semiquantitative real-time PCR to monitor the expression of FBLN5 transcripts, which were normalized to their corresponding GAPDH signals. Data
are the mean (±SE; n 5 3) fold-changes in FBLN5 expression relative to that observed in resting NMuMG cells (?P , 0.05; Student’s t-test).
FBLN5 promotes EMT and mammary tumorigenesis
EMT stimulated by TGF-b. Collectively, these findings indicate that
the role played by FBLN5 during mammary tumorigenesis is more
complicated than originally hypothesized (9), and suggest that
FBLN5 expression may function as an inducer of EMT in normal
and malignant MECs.
FBLN5 initiates EMT and enhances EMT induced by TGF-b in
normal and malignant MECs
We tested the aforementioned hypothesis by stably expressing FBLN5
in NMuMG cells and in malignant, metastatic 4T1 breast cells and
subsequently monitored their ability to undergo EMT in the absence
or presence of TGF-b. Figure 2A shows that stable expression of
FBLN5 in NMuMG cells induced a partial EMT, as evidenced by
the altered cortical actin architectures observed in resting cells. In-
deed, actin cytoskeletal rearrangements (Figure 2B, red channel) were
readily apparent in NMuMG that expressed FBLN5 (Figure 2B,
green channel), which contrasted sharply with the normal cuboidal
morphologies and cortical actin architectures exhibited by adjacent
NMuMG cells that failed to express the FBLN5 transgene
(Figure 2B). Moreover, FBLN5 expression greatly potentiated the
ability of TGF-b to induce EMT in NMuMG cells (Figure 2A).
Similarly, FBLN5 expression also enhanced the resting and
Fig. 2. FBLN5 initiates EMTand enhances EMTinduced by TGF-b in normal and malignant MECs. Parental (i.e. GFP) and FBLN5-expressing NMuMG (A and
B) and 4T1 (C) cells were incubated for 36 h in the absence or presence of TGF-b1 (5 ng/ml) as indicated. Altered actin cytoskeletal architectures were identified
by rhodamine–phalloidin immunofluorescence as indicated. Three color composite of quiescent FBLN5-expressing NMuMG cells showed that FBLN5-
expressing cells exhibited a partial EMT, whereas adjacent cells lacking FBLN5 expression exhibited normal MEC morphology (B). Channel colors: red, actin
cytoskeleton; green, GFP as an indirect measure of FBLN5 transgene expression; blue, 4#,6-diamidino-2-phenylindole (DAPI) staining. Color channels also are
presented individually in gray scale. Shown are representativeimages from a single experiment that was repeated twicewith similar results. (D) Parental (i.e. GFP)
and FBLN5-expressing NMuMG cells were stimulated with TGF-b1 as above. Afterward, changes in the expression of E-cadherin (E-cad), N-cadherin (N-cad),
b3 integrin (b3-Int) and Cox-2 were monitored by immunoblotting detergent-solubilized whole-cell extracts with antibodies against these proteins as indicated.
Differences in protein loading were monitored by reprobing stripped membranes with b-actin antibodies. Data are from a representative experiment that was
repeated twice with similar results. (E and F) Control (i.e. GFP) or FBLN5-expressing NMuMG or 4T1 cells were incubated for 36 h in the absence or presence of
TGF-b1 (5 ng/ml)to induceEMTindicated.Afterward,total RNAwasisolated and subjectedto semiquantitativereal-time PCR to monitor the expression of Twist
transcripts, which were normalized to their corresponding GAPDH signals. Data are the mean (±SE; n 5 3) fold-changes in Twist expression relative to that
observed in resting NMuMG cells (?P , 0.05; Student’s t-test).
Y.-H.Lee et al.
TGF-b-stimulated EMT in 4T1 breast cancer cells (Figure 2C), sug-
gesting that the oncogenic activities of FBLN5 are not restricted to
NMuMG cells. Besides its effects on actin stress fiber formation,
FBLN5 expression also enhanced the ability of TGF-b to suppress
E-cadherin expression and to induce the expression of N-cadherin, b3
integrin and Cox-2 in NMuMG cells (Figure 2D). Moreover,we found
that EMT stimulated by TGF-b (Figure 2E) or the process of mam-
mary tumorigenesis (Figure 2F) both induced Twist expression, and
more importantly, that FBLN5 significantly enhanced resting and
TGF-b-stimulated Twist expression in normal and malignant MECs
(Figure 2E and F). Collectively, these findings show that TGF-b in-
duces FBLN5 expression in MECs in a manner reflecting its ability to
stimulate EMT, and that FBLN5 expression promotes the acquisition
of EMT in normal and malignant MECs.
FBLN5 promotes EMT in normal and malignant MECs in a
A hallmark of EMT phenotypes is the acquisition of highly motile and
invasive phenotypes (2,5), which often corresponds to elevated ex-
pression and activity of MMPs (17). Semiquantitative real-time PCR
analyses showed a clear linkage between FBLN5 expression and that
of MMPs (Figure 3). To more thoroughly explore this potential con-
nection, we performed gelatin zymography on conditioned media
collected from GFP- and FBLN5-expressing NMuMG cells before
and after their stimulation with TGF-b. As shown in Figure 4A,
FBLN5 expression significantly augmented resting and TGF-b-stim-
ulatedMMP-2 and -9 proteolytic activity. Moreover,administration of
recombinant FBLN5 to NMuMG cells also enhanced their resting and
TGF-b-stimulated MMP-9 protease activity (Figure 4B). Similar
stimulatory effects of FBLN5 expression were observed in resting
and TGF-b-stimulated 4T1 cells (Figure 4C). Functionally, FBLN5
enhanced the EMT phenotype in normal (Figure 5A) and malignant
(Figure 5B) MECs, as well as significantly augmented their tonic and
TGF-b-stimulated invasion through synthetic basement membranes
(Figure 5C; data not shown). Importantly, pharmacological treatment
of the cells with type I MMP-2/3 or MMP-2/9 inhibitors impaired the
ability of FBLN5 and TGF-b to promote these events in normal (Fig-
ure 5A and C) and malignant (Figure 5B; data not shown) cells.
Accordingly, semiquantitative real-time PCR analyses demonstrated
that FBLN5 significantly enhanced the ability of TGF-b to repress the
transcription of E-cadherin (Figure 5D), while simultaneously induc-
ing that of Twist (Figure 5E). As above, both transcriptional responses
mediated by FBLN5 were reversed by treating the cells with the type I
MMP-2/3 or MMP-2/9 inhibitors (Figure 5D and E). Collectively,
these findings identify an important mechanism that links FBLN5 to
Fig. 4. FBLN5 enhances resting and TGF-b-stimulated MMP activation in
normal and malignant MECs. Gelatin zymography was carried out on
conditioned media harvested from parental (i.e. GFP) or FBLN5-expressing
cells that were incubated in the absence or presence of TGF-b1 (5 ng/ml) for
24 h. (A) MMP-2 and MMP-9 activation in GFP- and FBLN5-expressing
NMuMG cells, (B) MMP-9 activation NMuMG cells stimulated with
recombinant FBLN5 or(C) MMP-9activationin GFP-orFBLN5-expressing
4T1 cells. Images are from representative experiments that were repeated at
least once with similar results.
Fig. 3. FBLN5 alters the protease expression profiles in NMuMG cells.
Parental (i.e. GFP) or FBLN5-expressing NMuMG cells were induced to
undergo EMT by TGF-b1 as above. Afterward, total RNA was isolated and
used to monitor changes in the expression of MMP-2, MMP-3, plasminogen
activator inhibitor-1, TIMPs 1–3, TSP-1 and MT1–3-MMPs by
semiquantitative real-time PCR as indicated.Individualtargetgene transcript
signals were normalized to that observed for GAPDH. Data are the mean
fold-changes (n 5 3) in target gene expression relative to that observed in
resting NMuMG cells.
FBLN5 promotes EMT and mammary tumorigenesis
the increased expression and activation of MMPs 2 and 9 that function
to enhance TGF-b stimulation of EMT and invasion in MECs.
FBLN5 enhances the growth of mammary tumors in mice
Our findings thus far implicate FBLN5 as a promoter of mammary
tumorigenesis, during which normal MECs lose their polarity and
acquire the ability to grow in an anchorage-independent manner. In-
deed, when compared with their control counterparts, we observed
FBLN5 expression to significantly increase the ability of NMuMG
and 4T1 cells to grow in soft agar, a response that was further en-
hanced byinclusionofTGF-b to theseMEC cultures(Figure6Aand B).
These findings suggest that FBLN5 may promote the growth of mam-
mary tumors in mice. We tested this hypothesis by monitoring the
growth of GFP- and FBLN5-expressing 4T1 cells following their
orthotopic injection into the mammary fat pads of syngeneic Balb/C
mice. In keeping with our hypothesis, we found the growth of 4T1
tumors to be increased significantly by the expression of FBLN5
Fig. 5. FBLN5 promotes EMT in normal and malignant MECs in a MMP-dependent manner. Parental (i.e. GFP) or FBLN5-expressing NMuMG (A) or 4T1 (B)
cells were induced to undergo EMT by addition of TGF-b1 (5 ng/ml) in the absence or presence of the type I MMP-2/3 (10 lM) or MMP-2/9 (10 lM) inhibitors as
indicated. Afterward, altered architecture in the actin skeleton was monitored by rhodamine–phalloidin immunofluorescence. Shown are representative images
froma singleexperimentthatwasrepeatedtwicewithsimilarresults.(C) Parental(i.e. GFP)or FBLN5-expressingNMuMGcells were incubatedinthe absenceor
presence of either diluent (Dil), type I MMP-2/3 (10 lM; M2/3) or MMP-2/9 (10 lM; M2/9) inhibitors while undergoing invasion through synthetic basement
membranes in response to TGF-b1 (5 ng/ml). Data are the mean (±SE; n 5 3) percent-change in cell invasion relative to that observed in resting GFP-expressing
NMuMG cells. (D and E) Parental (i.e. GFP) and FBLN5-expressing NMuMG were induced to undergo EMT by TGF-b1 (5 ng/ml) in the absence or presence of
the type I MMP inhibitors as above. Afterward, total RNA was isolated and employed in semiquantitative real-time PCR reactions to monitor changes in the
expression of E-cadherin (D) or Twist (E), whose signals were normalized to those observed for GAPDH. Data are the mean (±SE; n 5 3) fold-changes relative to
that observed in resting GFP-expressing NMuMG cells. Dil, diluent; M2/3, MMP-2/3 type I inhibitor; M2/9, MMP-2/9 type inhibitor.
Y.-H.Lee et al.
(Figure 6C and D). Interestingly, although FBLN5 did elicit minor
alterations in the resting rates of DNA synthesis in normal and ma-
lignant MECs, the expression of this extracellular matrix (ECM) pro-
tein failed to effect the overall proliferative response of these same
cells when stimulated by TGF-b (Figure 6E and F). Thus, the ability
of FBLN5 to promote the growth of malignant MECs in soft agar and
mice does notappear to reflect intrinsic differences in the proliferation
rates of these cells to FBLN5, but instead may be linked to its ability
to enhance TGF-b stimulation of EMT, invasion and MMP expression
in developing and progressing mammary tumors.
TGF-b is a pluripotent cytokine that regulates tissue morphogenesis
and differentiation by effecting cell proliferation and survival and by
altering the production of ECM proteins within cell and tissue micro-
environments (6,18). TGF-b also is a major inducer of physiological
EMT during development and wound healing and of pathological
EMT during fibrosis and tumorigenesis (5,6,19), during which onco-
genic EMTis considered to bean important and essential evolutionary
step in the development of metastatic disease (1,3). The ability of
TGF-b to induce EMT was first observed to occur in normal MECs
(4), but now is recognized to take place in a number of epithelial cell
Given the associations of EMT to cancer development and progres-
sion, together with the ability to TGF-b to promote both phenomena
in malignant MECs, it stands to reason that increasing our knowledge
of how TGF-b regulates physiological and pathological EMT will
enhance the ability of science and medicine to chemotherapeutically
target the oncogenic activities of TGF-b in developing and
Fig. 6. FBLN5 enhances anchorage-independent growth of MECs in vitro and the growth of 4T1 tumors in mice. Parental (i.e. GFP) and FBLN5-expressing
NMuMG (A) or 4T1 (B) cells were cultured in soft agar in the absence or presence of TGF-b1 (5 ng/ml) for 14 days, whereupon colony formation was quantified
by light microscopy. Data are the mean (±SE; n 5 3) colonies formed per microscope field (?P , 0.05; Student’s t-test). (C) Parental (i.e. GFP) or FBLN5-
expressing 4T1cells were injectedorthotopicallyinto the mammaryfat pads of syngeneicBalb/C.Tumorvolumes were measured as indicated beginning at day 11
and continued through day 32. Data are the mean (±SE; n 5 3) tumor volumes observed over 31 days (?P , 0.05; Student’s t-test). (D) Mean (±SE) tumor
volumes measured at the time of necropsy in 4T1 tumors that expressed GFP (n 5 13) or FBLN5 (n 5 14) (?P , 0.05; Student’s t-test). Parental (i.e. GFP) and
FBLN5-expressingNMuMG(E) or4T1(F) cells werestimulatedwithTGF-b1(5 ng/ml)for48 h asindicated.CellularDNAwasradiolabeledwith[3H]thymidine
and quantified by scintillation counting. Data are the mean (±SE; n 5 3) [3H]thymidine incorporation normalized to unstimulated GFP-expressing cells
(?P , 0.05; Student’s t-test).
FBLN5 promotes EMT and mammary tumorigenesis
progressing breast cancers. To this end, we show for the first time that
TGF-b significantly induces FBLN5 expression in MECs undergoing
EMT (Figure 1C), and more importantly, that FBLN5 expression
initiates EMT and enhances that stimulated by TGF-b via a MMP-
dependent mechanism in normal and malignant MECs (Figures 2–5).
Finally, our finding that human breast cancers significantly upregulate
their expression of FBLN5 suggests that the inappropriate expression
of this ECM protein actually may enhance the growth of developing
mammary neoplasms. Accordingly, we observed the growth of 4T1
tumors in genetically normal mice to be enhanced significantly by
FBLN5 (Figure 6), thereby associating a novel tumor-promoting
function to this TGF-b gene target.
FBLN5 is a member of the Fibulin family of ECM proteins (12–14)
and functions in regulating epithelial, endothelial and fibroblast ad-
hesion, proliferation and motility in a context-specific manner (9–12).
FBLN5 also is a gene target for TGF-b in fibroblasts and endothelial
cells (9,10), and its expression is (i) developmentally regulated and
widespread in many adult tissues, including heart, spleen, kidney,
lung, colon and ovary (9,12,23,24); (ii) upregulated in developing
or injured blood vessels (23,24) and (iii) capable of enhancing wound
closure in mice (25). In contrast, FBLN5 deficiency elicits lung and
vasculature abnormalities in mice that arise from aberrant elastic fiber
organization (26,27) that resembles cutis laxa syndrome, which in
humans is linked to genetic defects in FBLN5 (28,29). Thus, FBLN5
mediates cell–cell and cell–matrix signaling coupled to the regulation
of tissuedevelopment, remodeling and repair. Along these lines, TGF-
b also is widely expressed during development to regulate EMT,
particularly that occurring in the lung, kidney and mammary gland
and that occurring during wound healing and tissue remodeling
(5,6,19). Interestingly, circumstantial evidence also implicates a role
for Fibulins, including FBLN5, in mediating EMT. Indeed, FBLNs
1 and 2 are expressed prominently in active EMT regions during
embryonic development and organogenesis, particularly during for-
mation of the neural tube and crest, skeletal muscle and the epicar-
dium, endocardial cushion tissue and cardiac valves and septa
(12–14). Similarly, FBLN5 localizes to regions of EMT during arte-
rial, endocardial cushion tissue, neural crest and mesenchymal tissue
development (23,24). Thus, FBLN5 may be an important regulator of
normal EMT during embryonic development, as well as abnormal
EMT during cancer development, a supposition wholly supported
by the current study.
FBLN5 remain to be fully elucidated. However, FBLN5 does interact
physically with a growing list of ECM and secreted proteins, many of
which are implicated in promoting EMT and disease development in
humans. Indeed, known FBLN5-binding proteins include (i) the integ-
rins avb3, avb5 and a9b1 (10,23,26); (ii) apolipoprotein(a) (30); (iii)
the lysyl oxidase (LOX) family members, LOXL1, LOXL2 and
LOXL4 (31,32); (iv) tropoelastin (27); (v) the elastin-binding protein,
Emilin-1 (33); (vi) latent TGF-b-binding protein 2 (34); (vii) extra-
cellular superoxide dismutase (35) and (viii) fibrillin-1 (36). Interest-
ingly, pharmacological treatments capable of inhibiting b1 integrin
activity prevented TGF-b stimulation of EMT in normal MECs (37).
Moreover, we recently established avb3 integrin as an essential me-
diatorof oncogenic signalingbyTGF-b inmalignantMECs,including
its ability to promote their EMTand invasion in vitro, as well as their
growth and pulmonary metastasis in mice (38–40). It should be noted
that FBLN5 is unique amongst Fibulin family members by the pres-
ence of an integrin-binding RGD motif (12), and as such, future stud-
ies need to address the extent to which the binding of FBLN5 to
integrins(i)contributestoitsstimulation of EMTandmammarytumor
growth and(ii)promotes theoncogenicactivities ofTGF-b.Due to the
heterogenous nature of human breast cancers, it remains to be inves-
tigated as to whether aberrant and/or upregulated FBLN5 expression
associates preferentially with distinct subtypes of human breast can-
cers, and if so, as to how this event impacts the manner in which these
developing neoplasms sense and respond to TGF-b.
Future studies also need to address the potential role of LOXs in
mediating the oncogenic activities of FBLN5 and TGF-b, both of
which significantly induce the expression of LOX family members
in normal and malignant MECs (M.A.Taylor and W.P.Schiemann, in
preparation). LOX family members are a group of related copper-
dependent amine oxidases that function in cross-linking collagens
to elastin in the ECM, thereby increasing the tensile strength and
structural integrity of tissues during embryonic development and or-
ganogenesis and during the maintenance of normal tissue homeostasis
(41,42). Interestingly, FBLN5 interacts physically with the LOX fam-
ily members LOXL1, LOXL2 and LOXL4 (31,32), while LOXL1-
deficiency elicits elastogenic defects reminiscent of those observed in
FBLN5-deficient mice (26,27,31). Thus, LOX family members may
function coordinately with FBLN5 in mediating EMTand in promot-
ing the development and progression of human breast cancers. Ac-
cordingly, aberrantLOX activity is associated
progression, particularly the development of desmoplasia, whose abil-
ity to enhance tumor rigidity has been linked to the selection, expan-
sion and dissemination of metastatic cells (43,44). With respect
to breast cancer, elevated expression of LOX family members, par-
ticularly that of LOX, LOXL and LOXL2, correlates with increased
malignancy and the acquisition of invasive/metastatic phenotypes and
with the induction of EMT (41,42,45–48). In particular, LOX expres-
sion recently was shown to be essential for hypoxia-induced metas-
tasis of human MDA-MB-231 breast cancer cells in mice (45).
Moreover, elevated LOX expression in human breast cancers was
found most frequently in poorly differentiated, high-grade tumors
and, consequently, was found to predict for increased disease recur-
rence and decreased patient survival (45). Clearly, the molecular
connections potentially linking FBLN5 and LOX to the acquisition
of oncogenic signaling by TGF-b needs to be examined in greater
detail. Experiments designed to address these interesting issues are
CA129359); Komen Foundation (BCTR0706967 to W.P.S.); Univer-
sity of Colorado Cancer Center to Y.-H.L.
Members of the Schiemann Laboratory are thanked for critical reading of the
manuscript. The technical expertise and support from members of the Flow
Cytometry and Pathology Cores at the University of Colorado Cancer Center
also is gratefully acknowledged.
Conflict of Interest Statement: None declared.
1.Thiery,J.P. (2002) Epithelial-mesenchymal transitions in tumor progres-
sion. Nat. Rev. Cancer, 2, 442–454.
2.Thiery,J.P. (2003) Epithelial-mesenchymal transitions in development and
pathologies. Curr. Opin. Cell Biol., 15, 740–746.
3.Grunert,S. et al. (2003) Diverse cellular and molecular mechanisms con-
tribute to epithelial plasticity and metastasis. Nat. Rev. Mol. Cell Biol., 4,
4.Miettinen,P.J. et al. (1994) TGF-b induced transdifferentiation of mam-
mary epithelial cells to mesenchymal cells: involvement of type I receptors.
J. Cell Biol., 127, 2021–2036.
5.Zavadil,J. et al. (2005) TGF-b and epithelial-to-mesenchymal transitions.
Oncogene, 24, 5764–5774.
6.Galliher,A.J. et al. (2006) Role of TGF-b in cancer progression. Future
Oncol., 2, 743–763.
7.Blobe,G.C. et al. (2000) Role of TGF-b in human disease. N. Engl. J. Med.,
8.Wakefield,L.M. et al. (2002) TGF-b signaling: positive and negative effects
on tumorigenesis. Curr. Opin. Genet. Dev., 12, 22–29.
9.Schiemann,W.P. et al. (2002) Context-specific effects of fibulin-5
(DANCE/EVEC) on cell proliferation, motility, and invasion. Fibulin-5 is
induced by TGF-b and affects protein kinase cascades. J. Biol. Chem., 277,
Y.-H.Lee et al.
10.Albig,A.R. et al. (2004) Fibulin-5 antagonizes VEGF signaling and angio-
genic sprouting by endothelial cells. DNA Cell Biol., 23, 367–379.
11.Albig,A.R. et al. (2006) Fibulins 3 and 5 antagonize tumor angiogenesis
in vivo. Cancer Res., 66, 2621–2629.
12.Albig,A.R. et al. (2005) Fibulin-5 function during tumorigenesis. Future
Oncol., 1, 23–35.
13.Argraves,W.S. et al. (2003) Fibulins: physiological and disease perspec-
tives. EMBO Rep., 4, 1127–1131.
14.Chu,M.L. et al. (2004) Fibulins in development and heritable disease. Birth
Defects Res. C Embryo Today, 72, 25–36.
15.Lemaire,R. et al. (2004) Fibulin-2 and fibulin-5 alterations in Tsk mice
associated with disorganized hypodermal elastic fibers and skin tethering.
J. Invest. Dermatol., 123, 1063–1069.
16.Sokol,J.P. et al. (2005) The use of cystatin C to inhibit epithelial-mesen-
chymal transition and morphological transformation stimulated by TGF-b.
Breast Cancer Res., 7, R844–R853.
17.Fata,J.E. et al. (2004) Regulation of mammary gland branching morpho-
genesis by the extracellular matrix and its remodeling enzymes. Breast
Cancer Res., 6, 1–11.
18.Blobe,G.C. et al. (2001) Functional roles for the cytoplasmic domain of the
type III TGF-b receptor in regulating TGF-b signaling. J. Biol. Chem., 276,
19.Nawshad,A. et al. (2005) TGF-b signaling during epithelial-mesenchymal
transformation: implications for embryogenesis and tumor metastasis.
Cells Tissues Organs, 179, 11–23.
20.Fan,J.M. et al. (1999) TGF-b regulates tubular epithelial-myofibroblast
transdifferentiation in vitro. Kidney Int., 56, 1455–1467.
21.Hales,A.M. et al. (1994) TGF-b1 induces lens cells to accumulate
a-smooth muscle actin, a marker for subcapsular cataracts. Curr. Eye
Res., 13, 885–890.
22.Kasai,H. et al. (2005) TGF-b1 induces human alveolar epithelial to mes-
enchymal cell transition. Respir. Res., 6, 56.
23.Nakamura,T. et al. (1999) DANCE, a novel secreted RGD protein
expressed in developing, atherosclerotic, and balloon-injured arteries.
J. Biol. Chem., 274, 22476–22483.
24.Kowal,R.C. et al. (1999) EVEC, a novel epidermal growth factor-like
repeat-containing protein upregulated in embryonic and diseased adult vas-
culature. Circ. Res., 84, 1166–1176.
25.Lee,M.J. et al. (2004) Fibulin-5 promotes wound healing in vivo. J. Am.
Coll. Surg., 199, 403–410.
26.Nakamura,T. et al. (2002) Fibulin-5/DANCE is essential for elastogenesis
in vivo. Nature, 415, 171–175.
27.Yanagisawa,H. et al. (2002) Fibulin-5 is an elastin-binding protein essential
for elastic fiber development in vivo. Nature, 415, 168–171.
28.Loeys,B. et al. (2002) Homozygosity for a missense mutation in fibulin-5
(FBLN5) results in a severe form of cutis laxa. Hum. Mol. Genet., 11,
29.Markova,D. et al. (2003) Genetic heterogeneity of cutis laxa: a heterozy-
gous tandem duplication within the fibulin-5 (FBLN5) gene. Am. J. Hum.
Genet., 72, 998–1004.
30.Kapetanopoulos,A. et al. (2002) Direct interaction of the extracellular
matrix protein DANCE with apolipoprotein(a) mediated by the kringle
IV-type 2 domain. Mol. Genet. Genomics, 267, 440–446.
31.Liu,X. et al. (2004) Elastic fiber homeostasis requires lysyl oxidase-like 1
protein. Nat. Genet., 36, 178–182.
32.Hirai,M. et al. (2007) Fibulin-5/DANCE has an elastogenic organizer ac-
tivity that is abrogated by proteolytic cleavage in vivo. J. Cell Biol., 176,
33.Zanetti,M. et al. (2004) EMILIN-1 deficiency induces elastogenesis and
vascular cell defects. Mol. Cell. Biol., 24, 638–650.
34.Hirai,M. et al. (2007) Latent TGF-b-binding protein 2 binds to DANCE/
fibulin-5 and regulates elastic fiber assembly. EMBO J., 26, 3283–3295.
35.Nguyen,A.D. et al. (2004) Fibulin-5 is a novel binding protein for extra-
cellular superoxide dismutase. Circ. Res., 95, 1067–1074.
36.El-Hallous,E. et al. (2007) Fibrillin-1 interactions with fibulins depend on
the first hybrid domain and provide an adaptor function to tropoelastin.
J. Biol. Chem., 282, 8935–8946.
37.Bhowmick,N.A. et al. (2001) Integrin b1 signaling is necessary for TGF-b
activation of p38 MAPK and epithelial plasticity. J. Biol. Chem., 276,
38.Galliher,A.J. et al. (2006) b3 integrin and Src facilitate TGF-b mediated
induction of epithelial-mesenchymal transition in mammary epithelial
cells. Breast Cancer Res., 8, R42.
39.Galliher,A.J. et al. (2007) Src phosphorylates Tyr284 in TGF-b type II
receptor and regulates TGF-b stimulation of p38 MAPK during breast
cancer cell proliferation and invasion. Cancer Res., 67, 3752–3758.
40.Galliher-Beckley,A.J. et al. (2008) Grb2 binding to Tyr284 in TbR-II is
essential for mammary tumor growth and metastasis stimulated by TGF-b.
Carcinogenesis, 29, 244–251.
41.Lucero,H.A. et al. (2006) Lysyl oxidase: an oxidative enzyme and effector
of cell function. Cell. Mol. Life Sci., 63, 2304–2316.
42.Payne,S.L. et al. (2007) Paradoxical roles for lysyl oxidases in cancer—a
prospect. J. Cell. Biochem., 101, 1338–1354.
43.Anderson,A.R. et al. (2006) Tumor morphology and phenotypic
evolution driven by selective pressure from the microenvironment. Cell,
44.Paszek,M.J. et al. (2004) The tension mounts: mechanics meets morpho-
genesis and malignancy. J. Mammary Gland Biol. Neoplasia, 9, 325–342.
45.Erler,J.T. et al. (2006) Lysyl oxidase is essential for hypoxia-induced me-
tastasis. Nature, 440, 1222–1226.
46.Kirschmann,D.A. et al. (2002) A molecular role for lysyl oxidase in breast
cancer invasion. Cancer Res., 62, 4478–4483.
47.Payne,S.L. et al. (2005) Lysyl oxidase regulates breast cancer cell
migration and adhesion through a hydrogen peroxide-mediated mecha-
nism. Cancer Res., 65, 11429–11436.
48.Payne,S.L. et al. (2006) Lysyl oxidase regulates actin filament formation
through the p130(Cas)/Crk/DOCK180 signaling complex. J. Cell.
Biochem., 98, 827–837.
Received May 27, 2008; revised August 6, 2008; accepted August 16, 2008
FBLN5 promotes EMT and mammary tumorigenesis