Integrin-Mediated Macrophage Adhesion Promotes
Lymphovascular Dissemination in Breast Cancer
db4 integrin-expressing macrophages release TGF-b1 near
breast cancer lymphovasculature
dTGF-b1 drives b4 integrin clustering on macrophages,
enhancing macrophage adhesion
dTGF-b1 signals through RhoA to drive to lymphatic
endothelial cell contraction
dLymphatic remodeling signaling cascade facilitates breast
Rachel Evans, Fabian Flores-Borja,
Sina Nassiri, ..., Frederic Festy,
Michele De Palma, Tony Ng
Breast cancer metastasis through
lymphatic vessels is associated with poor
prognosis. Evans et al. describe b4
integrin-expressing macrophages that
regulate lymphatic vessel structure in
breast cancer. Macrophage-released
TGF-b1 drives lymphatic cell contraction
via RhoA activation, culminating in
lymphatic hyperpermeability. This study
deﬁnes a signaling cascade that could be
Evans et al., 2019, Cell Reports 27, 1967–1978
May 14, 2019 ª2019 The Authors.
Integrin-Mediated Macrophage Adhesion Promotes
Lymphovascular Dissemination in Breast Cancer
Michele De Palma,
and Tony Ng
Richard Dimbleby Department of Cancer Research, Randall Division & Division of Cancer Studies, Kings College London, London, UK
Breast Cancer Now Research Unit, King’s College London, Guy’s Hospital, London, UK
Swiss Institute for Experimental Cancer Research (ISREC), School of Life Sciences, E
´cole Polytechnique Fe
´rale de Lausanne (EPFL),
Pathology Core Facility, University College London Cancer Institute, London, UK
Institute for Mathematical and Molecular Biomedicine, King’s College London, London, UK
King’s Health Partners Cancer Biobank, King’s College London, London, UK
Research Oncology, Division of Cancer Studies, Guy’s Hospital, King’s College London, London, UK
Division of Imaging Sciences and Biomedical Engineering, King’s College London, London, UK
Department of Oncology, Cancer Research UK and Medical Research Council, Oxford Institute for Radiation Oncology,
University of Oxford, UK
Tissue Engineering and Biophotonics, King’s College London, London, UK
UCL Cancer Institute, University College London, London, UK
Present address: Cancer Institute, University College London, London, UK
Present address: Centre for Immunobiology and Regenerative Medicine, Barts and The London School of Medicine and Dentistry, Queen
Mary University of London, London, UK
Present address: Division of Infection and Immunity, Institute of Immunity and Transplantation, University College London, London, UK
*Correspondence: firstname.lastname@example.org (R.E.), email@example.com (T.N.)
Lymphatic vasculature is crucial for metastasis in tri-
ple-negative breast cancer (TNBC); however, cellular
and molecular drivers controlling lymphovascular
metastasis are poorly understood. We deﬁne a
macrophage-dependent signaling cascade that fa-
cilitates metastasis through lymphovascular remod-
eling. TNBC cells instigate mRNA changes in macro-
phages, resulting in b4 integrin-dependent adhesion
to the lymphovasculature. b4 integrin retains macro-
phages proximal to lymphatic endothelial cells
(LECs), where release of TGF-b1 drives LEC contrac-
tion via RhoA activation. Macrophages promote
gross architectural changes to lymphovasculature
by increasing dilation, hyperpermeability, and disor-
ganization. TGF-b1 drives b4 integrin clustering at
the macrophage plasma membrane, further promot-
ing macrophage adhesion and demonstrating the
dual functionality of TGF-b1 signaling in this context.
b4 integrin-expressing macrophages were identiﬁed
in human breast tumors, and a combination of
vascular-remodeling macrophage gene signature
and TGF-bsignaling scores correlates with metas-
tasis. We postulate that future clinical strategies for
patients with TNBC should target crosstalk between
b4 integrin and TGF-b1.
Tumor cells establish complex interactions with cells within their
microenvironment that determine malignancy progression (Balk-
will et al., 2012). Tumor cell dissemination can occur through
blood or lymphovasculature; however, targeting blood vascula-
ture has limited clinical efﬁcacy when lymphatic dissemination
is prevalent (Wong and Hynes, 2006).
Breast cancer is divided into subtypes based on histopatho-
logical features and gene signatures (Gazinska et al., 2013).Tri-
ple-negative breast cancer (TNBC) is characterized by a lack
of druggable targets, is highly metastatic, and is associated
with dismal prognosis (Gazinska et al., 2013; Dent et al., 2007).
The prognostic signiﬁcance of lymphangiogenesis in TNBC is
under debate. However, invasion into lymphatic vessels corre-
lates with poor prognosis, suggesting that targeting an existing
lymphatic vessel network could provide an effective treatment
strategy (Choi et al., 2005; Mohammed et al., 2007, 2011; Liu
et al., 2009).
The relationship between tumor and immune cells is often bidi-
rectional and involves both tumor-promoting and -antagonizing
mechanisms (Pollard, 2004; Quail and Joyce, 2013). Among
innate immune cells, macrophages have been implicated in the
promotion of tumor progression and, in particular, breast cancer
metastasis (Condeelis and Pollard, 2006; Kitamura et al., 2015;
Pollard, 2004; Harney et al., 2015). However, it remains unclear
how certain subsets of tumor-associated macrophages (TAMs)
inﬂuence breast cancer metastasis spatially, temporally, and at
a molecular level.
Cell Reports 27, 1967–1978, May 14, 2019 ª2019 The Authors. 1967
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
(legend on next page)
1968 Cell Reports 27, 1967–1978, May 14, 2019
The integrin family are adhesion receptors of paramount impor-
tance for immunecell adhesion and migration during inﬂammatory
processes (Evans et al., 2009). Their ability to form adhesive con-
tacts is regulated by soluble factors, as part of the chemoattrac-
tant-adhesion crosstalk that causes a combination of changes
in integrin conformation and clustering on the plasma membrane
(PM) that regulate downstream signaling (Hynes, 2002). In malig-
nancy, many integrins common in epithelial cells are also present
in solid tumors, and some, such as avb3anda5b1, are speciﬁcally
upregulated in cancer (Desgrosellier and Cheresh, 2010). Tumor-
expressed integrins affect tumor cell migration, proliferation,
survival, and anchorage to the extracellular matrix. Endothelial-
cell-expressed integrins are implicated in angiogenesis, lymphan-
giogenesis, and vascular remodeling (Avraamides et al., 2008).
While the importance of integrins with respect to maintaining a
pro-tumoral immune microenvironment in solid tumors is not
well deﬁned, in chronic lymphocytic leukemia, impaired integrin
signaling in non-leukemic T cells changes the immune microenvi-
ronment to be more immunosuppressive, which may facilitate
malignancy progression (Ramsay et al., 2013).
We seek to identify the role of TAMs in regulating existing lym-
phovasculature in TNBC, where lymphatic dissemination is not a
direct result of lymphangiogenesis.
We propose that macrophages have an important role in
controlling established tumoral lymphatic networks in TNBC
and that lymphatic dissemination of cancer cells is facilitated
by a cascade of signaling events initiated by integrin-mediated
adhesion of macrophages at the sites of lymphatic vessels.
Lymphovascular Macrophages in TNBC Mouse Models
Are Retained through Binding of b4 Integrin to Laminin-5
To identify endogenous macrophages with respect to lymphatic
vasculature in murine TNBC tumors, we scored F4/80+Tie2+
macrophages within podoplanin+ lymphovasculature across mul-
tiple ﬁelds of view (FOVs) from 4T1.2 and BLG-Cre;Brca1
TNBC models (Molyneux et al., 2010; Melchor et al., 2014;Fig-
ures 1A and 1B). The Tie2-expressing macrophage (TEM) subset
is associated with angiogenesis and lymphatic development
(De Palma et al., 2005, 2007; Gordon et al., 2010). Lymphovascu-
lar-associated macrophages expressing Tie2 have recently been
reported in a small breast cancer cohort (Bron et al., 2015).
In 4T1.2 tumors, we found a mean value of 6.3 F4/80+Tie2+
macrophages within podoplanin+ vasculature (versus 1.7 in
podoplaninregions) per FOV. In BLG-Cre;Brca1
mors, we observed 8.8 F4/80+Tie2+ macrophages in podopla-
nin+ vasculature (versus 2.0 in podoplaninregions) per FOV.
Therefore, F4/80+Tie2+ macrophages are enriched in lymphovas-
cular regions in murine TNBC models.
The b4 integrin subunit is a transmembrane glycoprotein asso-
ciating exclusively with the a6 integrin subunit. a6b4 integrin is
expressed predominantly on epithelial and endothelial cells
and binds to laminins to form adhesion complexes, hemidesmo-
somes (Stewart and O’Connor, 2015). Microarray analysis of
endogenous macrophages co-cultured with 4T1.2 tumor cells
showed a mean 1.8-fold upregulation of b4 integrin at the RNA
level, compared with non-educated endogenous macrophages,
and that the RAW264.7 macrophage cell line similarly exhibited a
mean 1.58-fold increase in b4 integrin levels, compared with
endogenous macrophages (Figure 1C; see also data published
in ArrayExpress: MTAB-4064).
4T1.2 tumors were excised and disaggregated at day 10.
Within 4T1.2 tumors, we deﬁned a population of macrophages
The inﬂuence of tumor education on macrophage adhesion to
b4 integrin ligand, laminin-5, was investigated. Tumor-educated
endogenous macrophages displayed increased adhesion to lam-
inin-5 (30.7% ±7.2% to 81.7% ±13.2% adherent cells on 0.5 mM
laminin-5; Figure 1E). As laminin-5 is reportedly localized in areas
with high blood vessel density, we investigated whether laminin-
5 was also in areas of lymphovasculature. 4T1.2 tumor tissue
analysis showed laminin-5 furnished around podoplanin+ lympho-
vasculature (Figure 1F). In addition we observed macrophages
expressing a6b4integrininlymphovascularregions(Figure 1G).
To study b4 expression in vivo, we used primary 4T1.2 tumor
sections stained with Lyve1-Cy3 and b4 integrin-Cy5. Tissues
were imaged using a protocol involving laser photobleaching
to remove autoﬂuorescence. Our methodology reveals b4 integ-
rin throughout the tumor; however, within lymphatic vessels,
there is differential distribution of b4 integrin with relative in-
creases in b4 accumulation observed in lymphovascular areas
proximal to Lyve1+ lymphatic endothelial cells (LECs) (Figure 1H,
white arrow). Additionally, there were lymphovascular areas with
an increased localized Pearson coefﬁcient, suggesting that
LECs and b4-integrin-expressing macrophages were in close
contact (Figure 1H, blue arrow) (mean colocalization coefﬁcient,
Figure 1. Lymphovascular Macrophages in TNBC Mouse Models Are Retained through Binding of b4 Integrin to Laminin-5
(A and B) Tumor sections from 4T1.2 (A) and BLG-Cre;Brca1
(B) were stained with F4/80-FITC, podoplanin-AF555, and Tie2 -Cy5-conjugated ant ibody.
F4/80+Tie2+ macrophages within podoplanin+ areas versus those in other areas were quantiﬁed per ﬁeld of view (FOV). Vessel lumen is outlined; arrow indicates
a macrophage within a podoplanin+ area. Images were acquired with a 340 air objective. Scale bars, 100 mm (main image) and 25 mm (zoomed inset).
(C) Array-derived expression proﬁle of b4-integrin (Itgb4) across samples. Barplot shows log
fold change of normalized expression value for b4 integrin (ratio of
the median value of probe in BMM samples).
(D) Day-12 4T1.2 tumors were disaggregated. Tie2 and b4 integrin FMO controls are indicated in 2 left panels. Right dot plot and histogram depict b4-integrin-
expressing macrophages from representative 4T1.2 tumor (n = 8).
(E) BMMs co-cultured alone or with 4T1.2-GFP cells plated on laminin-5. The percentage of adherent cells were quantiﬁ ed in triplicate (n = 2).
(F and G) 4T1.2 tumor sections were stained with laminin-5-Dylight488 and podoplanin-AF555 (F), and Lyve1-Cy3, F4/80-FITC, and b4 integrin-Cy5 (G); inset
shows F4/80+b4 integrin+ macrophages around lymphatic endothelium.
(H) Stained sections (Lyve1-Cy3 and b4 integrin-Cy5) were imaged using a custom-built microscope (320 air objective). Area of distinct b4 integrin and Lyve1
within lymphatic vessel (white arrow) and area of close contact between b4 integrin and Lyve1 (blue arrow) are indicated. Scale bars, 50 mm (main panels) and
25 mm (inset).
Cell Reports 27, 1967–1978, May 14, 2019 1969
(legend on next page)
1970 Cell Reports 27, 1967–1978, May 14, 2019
TAMs Drive Disorganized and Hyperpermeable
Lymphatic Architecture, and Contact between
Macrophages and LECs Results in RhoA-Dependent
We used a mammary image window (MIW) subcutaneously im-
planted over a 4T1.2-mCherry tumor (Kedrin et al., 2008;
Figure 2A). Injection of 76 kDa dextran-FITC (ﬂuorescein isothio-
cyanate) allowed visualization of lymphatic vasculature. Using
multiphoton microscopy, we observed that, within the tumor,
lymphatic vessels leaked dextran dye across the FOV (Fig-
ure 2Aii, left panel), suggesting high levels of vessel permeability;
however, in more distal regions, lymphatic vessels had a distinct
structure and 4T1.2-mCherry intra-lymphatic tumor cells could
be seen within vessels, suggesting ongoing metastasis (Fig-
ure 2Aii, middle and right panels, respectively). To understand
how increasing TAMs could phenotypically inﬂuence lymphatic
vasculature, we studied the permeability of lymphatic vessels
from 4T1.2 tumor-bearing mice given an intermittent bolus of
RAW264.7 macrophages during tumor development. Both
RAW264.7 macrophages and the 4T1.2 tumor line are derived
from a BALB/c genetic background, allowing us to investigate
the effects of elevated macrophage numbers on tumor progres-
sion in vivo using a syngeneic model of TNBC.
To quantify lymphatic vessel permeability in vivo,weadapteda
protocol previously used in angiogenesis studies (Finsterbusch
et al., 2014). Using a subcutaneous injection of Evans Blue dye,
we quantiﬁed the permeability of the tumoral lymphatics. Tumors
with elevated macrophages contained hyperpermeable lymphatic
vessels with an increase in mean optical density (OD) per gram
from 0.7812 ±0.2956 to 2.290 ±0.5160 when compared with
PBS-treated control, suggesting a facilitated pathway between
the primary tumor and lymphatic vasculature (Figure 2B).
To understand the effects of elevated macrophages on tu-
moral lymphatic vessel architecture, we stained tumor sections
from mice treated with PBS or RAW264.7 macrophages with
the lymphatic vessel markers, Lyve1 and podoplanin (Figure 2C;
Figures S1A and S1B), demonstrating that both lymphatic
markers gave a similar staining distribution. Typical sections
from PBS-treated mice showed small, well-formed vessels to-
ward the tumor periphery or within the peri-tumoral areas with
a mean diameter of 13.66 mm±1.295 mm. This was in contrast
to RAW264.7-treated mice that had larger vessels with a mean
diameter of 48.00 mm±6.065 mm, indicating increased vessel
dilation (Figure S1C).
To quantify changes in lymphatic architecture in tumors with
elevated levels of macrophages, we blindly scored lymphovas-
culature for disorganization based on the following criteria.
Smaller vessels with a clear lumen were given low scores
(0 and 1) compared with larger disorganized vessels with unclear
borders (2 and 3). PBS-treated tumors had a mean disorganiza-
tion score of 0.25 ±0.16 and 1.6 ±0.33, compared with 1.8 ±
0.29 and 2.5 ±0.17 for tumors treated with RAW264.7 macro-
phages (Figure 2C).
To further investigate whether macrophages were sufﬁcient to
induce a disorganized lymphatic phenotype, we ablated endoge-
nous macrophages using clodronate-containing liposomes post-
establishment of 4T1.2 tumors. Endogenous macrophages were
reconstituted post-clodronate treatment with non-educated
bone marrow macrophages (BMMs) or tumor-educated BMMs
for 48 h (Figure 2Di). The extent of lymphatic disorganization in
the 4T1.2 primary tumors was greater after reconstitution with
endogenous tumor-educated BMMs, compared with non-
educated BMMs (0.333 ±0.3 to 2 ±0.29; Figure 2D, ii and iii).
These results demonstrate that the presence of TAMs results in
a disorganized lymphatic vasculature around the primary tumor,
that the extent of disorganization is related to overall macrophage
levels, and that this occurs at an early time point in tumor develop-
ment (days 10–14).
To investigate how TAMs affect lymphatic endothelia, we
added endogenous macrophages to monolayers of primary
LECs isolated from BALB/c mice (Figure 2E). Primary LECs
had a mean spread area of 1,132 mm
reduced slightly to 808.6 mm
after the addition of
endogenous uneducated macrophages but dramatically
reduced to 324.1 mm
macrophages and 473.7 mm
with ex vivo TAMs
). Similar LEC contraction occurred
when the murine LEC line, SV-LEC (Ando et al., 2005), was grown
as a monolayer and endogenous macrophages (Figure S1D) or
RAW264.7 macrophages added (Figure 2Fi). SV-LEC contraction
Figure 2. TAMs Drive Dilated, Hyperpermeable, and Disorganized Lymphatic Architecture through LEC RhoA Activation
(A) (i) Mouse with mCherry-tagged 4T1.2 tumor and implanted mammary imaging window (MIW) at days 10–14. (ii) Left panel: lymphatic vessels (green) sur-
rounding tumor (red). Middle panel: lymphatic vessels (green) distal to main tumor bulk (red). Right panel: lymphatic vessel (green) with tumor cells (red) within
vessel. Scale bars, 100 mm.
(B) 4T1.2 tumor-bearing mice were treated with PBS or RAW264.7 macrophages over 3 weeks. 1% Evans Blue dye stained lymphatics in vivo. Lymphatic
permeability was calculated as optical density per gram of tumor. Data represent means ±SEM; signiﬁcance was determined using unpaired t tests (**p < 0.01).
(C) (i) Lymphatic vessels within tumors from mice treated with PBS or RAW264.7 macrophages stained with Lyve1-Cy3 or podoplanin-AF555 (red) and blindly
scored for disorganization. Scale bars, 50 mm. (ii) Four FOVs in 4 PBS-treated and 4 RAW264.7 macrophage-treated tumor samples scored blindly for disor-
ganization. Data represent means ±SD; signiﬁcance was determined using unpaired t tests (***p < 0.001).
(D) (i) Timeline depicting clodronate-containing liposome protocol. (ii) Tumor sections from clodronate-treated mice reconstituted with PBS, BMM, or BMM
stained with Lyve1-Cy3 or podoplanin-AF555 (red). Lymphatic disorganization within tumors from 6 mice was quantiﬁed from >3 FOVs per mouse from Lyve1-
stained sections. Data represent means ±SD; signiﬁcance was determined using unpaired t tests (**p < 0.01).
(E) Primary LECs were cultured alone, with BMM, eBMM, or TAM. LECs were stained with podoplanin-AF555, and macrophages were stained with F4/80-FITC.
Confocal microscopy (x40 air objective) was used to quantify the area of LECs from 3 FOVs (n = 2). Scale bar, 10 mm.
(F) (i and ii) Monolayer of SV-LECs (CellTracker Green CMFDA) with RAW264.7 macrophages (CellTracker Orange CMTMR) after 24 h. Area of SV-LECs was
measured using ImageJ software. Data represent means ±SEM; signiﬁcance was determined using unpaired t tests (**p < 0.01). Scale bars, 25 mm.
(G) (i) SV-LECs transfected with RhoA RAICHU biosensor (RAICHU R/G) or RhoA-GFP as a control. Transfected SV-LECs were cultured alone or with BMM or
eBMM for 24 h. (ii) Multiphoton microscopy was used to determine the ﬂuorescence lifetim e decay (Tau; in nanoseconds) of SV-LECs transfected with RhoA-GFP
or RhoA RAICHU biosensor. Data represent means ±SD; signiﬁcance was determined using unpaired t tests (**p < 0.01). N.S., not signiﬁcant.
Cell Reports 27, 1967–1978, May 14, 2019 1971
(legend on next page)
1972 Cell Reports 27, 1967–1978, May 14, 2019
occurred with areas reducing from 835.9 mm
and from 632.5 mm
. In addition, the area of SV-LECs was
quantiﬁed with and without contact with RAW264.7 macro-
phages. SV-LEC contraction was only observed when direct con-
tact between the 2 cell types occurred (436.4 mm
)(Figure 2Fii). Collectively, our evidence sug-
gests that direct contact between TAMs and LECs is required for
contraction events to occur.
RhoA regulates many events in blood-vessel-speciﬁc endo-
thelial cells during angiogenesis, such as motility, proliferation,
and permeability (Bryan et al., 2010). We sought to test whether
RhoA regulates contraction events observed in LECs. SV-LECs
were transiently transfected with the GFP- and monomeric red
ﬂuorescent protein (mRFP)-expressing RhoA RAICHU biosensor
(Heasman et al., 2010; Makrogianneli et al., 2009; Yoshizaki
et al., 2003), which allows measurement of the ﬂuorescent
lifetime decay (Tau) when ﬂuorescence resonance energy trans-
fer (FRET) occurs between the GFP and mRFP upon RhoA acti-
vation. After SV-LEC transfection, non-educated or tumor-
educated endogenous macrophages were added to SV-LECs
for 24 h. The ﬂuorescence lifetime of the RAICHU probe
(expressed exclusively in the SV-LECs) was measured using
multiphoton microscopy. SV-LEC co-culture with tumor-
educated macrophages led to a reduction in Tau of the
biosensor from 1.797 ns ±0.0252 ns to 1.622 ns ±0.0338 ns,
indicating an increase in FRET between the GFP- and
RFP-terminal ﬂuorophores and, consequently, an increase in
RhoA activity (Figure 2G). No change in Tau was observed
when SV-LECs were co-cultured with non-educated endoge-
nous macrophages (Figure 2Gii). These results demonstrate
that RhoA activity increases during LEC contraction and that
this only occurs in the presence of tumor-educated macro-
phages in contact with lymphatic endothelia.
LEC Contraction Is Dependent on TGF-b1 Release from
Transforming growth factor (TGF)-breceptor ligation in ﬁbro-
blasts results in RhoA activation (Fleming et al., 2009). We inves-
tigated the release of active TGF-b1 and TGF-b2 isoforms from
non-educated and tumor-educated macrophages by ELISA
(Figure 3A). TGF-b1 levels increased from 2,600 pg to 4,400 pg
in tumor-educated endogenous macrophages (increase in opti-
cal absorbance at 450 nm from 1.286 ±0.07119 to 2.585 ±
0.1077). In contrast, TGF-b2 levels were not signiﬁcantly
changed. While TGF-bis present throughout the tumor microen-
vironment, membrane-bound TGF-bcan have a potent effect on
downstream signaling through increasing the concentration
gradient of this molecule (Savage et al., 2008). Our data showed
that 4T1.2 education of endogenous macrophages signiﬁcantly
increased the levels of plasma-membrane-bound TGF-b1(Fig-
ure S2A), allowing stringent spatial control of downstream
To test the hypothesis that macrophage-released TGF-b1 was
responsible for LEC contraction, we investigated the effect of a
TGF-breceptor inhibitor, SB-431542 (Inman et al., 2002;Fig-
ure S2B). As expected, RAW264.7 macrophages alone induced
LEC contraction (950.6 mm
to 335.8 mm
); however, this did not occur in the presence of
SB-431542 or when TGF-b1orb4 integrin were transiently
knocked down in RAW264.7 macrophages, demonstrating that
the presence of b4 integrin and TGF-bin macrophages or
TGF-breceptor ligation on LECs was sufﬁcient to prevent
contraction (Figures 3B, S2C, and S2D).
The role of macrophage-released TGF-b1 on lymphovascular
disorganization was investigated in vivo. A stable knockdown of
TGF-b1 was generated in RAW264.7 macrophages using lentivi-
ral short hairpin RNA (shRNA) (Figure S2E). Similar to our previ-
ous in vivo studies, macrophages were administered intrave-
nously throughout tumor development. After 2 weeks’ growth,
tissue sections were stained for Lyve1 and podoplanin. The
extent of lymphatic disorganization in tumors with RAW264.7-
TGFb1 knockdown, compared with that in RAW264.7-NTC,
was blindly scored in Lyve1-podoplanin-stained tissues as
described earlier. Our results show that absence of TGF-b1in
RAW264.7 macrophages was sufﬁcient to signiﬁcantly decrease
the extent of lymphatic disorganization observed, compared
with that in RAW264.7-NTC macrophages (1.8 ±0.16 to 1.1 ±
0.18) (Figure 3C) and that these changes were evident at an early
To functionally associate macrophage-released TGF-b1to
structural changes in the lymphatic endothelium in vivo,we
Figure 3. Macrophage-Expressed TGF-b1 Regulates b4 Integrin Clustering on the Macrophage Plasma Membrane and Is Required for LEC
(A) BMMs cultured alone (BMM) or with 4T1.2 cells (BMM coculture). Supernatants were probed for (i) TGFb1 and (ii) TGF-b2 by ELISA. Data represent
means ±SD; signiﬁcance was determined using unpaired t tests (***p < 0.001). N.S., not signiﬁcant.
(B) SV-LECs grown as monolayers. Tumor-educated RAW264.7 macrophages (eRAW) were added plus DMSO control or 10 mM SB-431542. After 24 h, SV-LEC
areas were quantiﬁed. Data represent means ±SD; signiﬁcance was determined using unpaired t tests (****p < 0.0001). N.S., not signiﬁcant.
(C) Tumor-bearing mice were injected with RAW264.7-NTC or RAW264.7-TGFb1 knockdown until day 14. Tumor sections were stained with podoplainin-AF555
or Lyve1-Cy3 and Lyve1+ vessels blindly scored for lymphatic disorganization (*p < 0.05). Scale bars, 50 mm.
(D) Tumor-bearing mice were injected with RAW264.7-NTC or RAW264.7-TGFb1 un til day 21. Tumor sections were stained with F4/80-FITC, pMLC (and Rabbit-
Cy3 secondary antibody), and podoplanin-Cy5. F4/80+ cells within podoplanin+ regions were identiﬁed, and a 65-mm
region of interest (ROI) was identiﬁed
(white circles) where the ﬂuorescence intensity of the pMLC signal was quantiﬁed. Scale bar, 50 mm (4 FOVs from n = 2 tumors from each condition). Data
represent means ±SD; signiﬁcance was determined using unpaired t tests (**p < 0.01).
(E) RAW264.7-NTC and RAW264.7-TGFb1 macrophage areas were measured by confocal microscopy. Data represent means ±SD; signiﬁcance was deter-
mined using unpaired t tests (**p < 0.01).
(F) RAW264.7-NTC and RAW264.7-TGFb1 were stained with anti-b4 integrin-AF647 and imaged using structured illumination microscopy (Nikon 3100 oil
objective). Focal adhesion area was determined using ImageJ on thresholded image s. Data represent means ±SD; signiﬁcance was determined using unpaired
t tests (**p < 0.01). Scale bars, 10 mm (main image) and 1 mm (insets).
Cell Reports 27, 1967–1978, May 14, 2019 1973
(legend on next page)
1974 Cell Reports 27, 1967–1978, May 14, 2019
quantiﬁed levels of phospho-myosin light chain (pMLC) in LECs
adjacent to macrophages. Since RhoA activity is high in con-
tracting LECs, and since active RhoA phosphorylates MLC,
pMLC can be used as a readout of LEC contractility in cells
proximal to lymphatic-associated macrophages. We observed
that, when mice were injected with RAW264.7-TGFb1 knock-
down, compared with RAW264.7-NTC, there was a signiﬁcant
reduction in pMLC levels in lymphatic vasculature adjacent to
RAW264.7 macrophages when TGF-b1 was absent (1.97 3
±401,151 to 6.56 310
3187,133) (Figure 3D).
TGF-b1 Controls b4 Clustering at the Macrophage
We studied the effect of TGF-b1 on the phenotypic functionality of
macrophages by quantifying the spreading response of macro-
phages. There was clear reduction in cell spreading when
TGF-b1 was knocked down in RAW264.7 macrophages,
compared with the non-targeted control counterpart (235.2 mm
to 91.91 mm
)(Figure 3E). To understand
how TGF-b1 could control macrophage spreading, we investi-
gated the effect of TGF-b1onb4 expression. Since integrins can
be constitutively expressedon the cell surface, we sought to study
the plasma membrane distribution of b4 integrin using structured
illumination microscopy in RAW264.7-TGFb1 shRNA versus
RAW264.7-NTC. Our results clearly show that, while there may
be small differences in the overall amountof b4 integrin expressed
on the cell surface (Figures S3A and S3B), the size of integrin clus-
ters that can form ﬁrm adhesive contact with integrin ligand are
signiﬁcantly reduced when TGF-b1isabsent(1.97mm
to 1.559 mm
;Figure 3F, i and ii). These re-
sults collectively indicate that TGF-b1 has both a paracrine role in
controlling the lymphatic endothelium and an autocrine role in
regulating b4 activity in tumor-educated macrophages.
b4 Integrin+ Macrophages and Lymphatic Remodeling
Are Associated with TGF-bSignaling and Adverse
Outcome in TNBC Patients
To establish that human macrophages express ITGB4 RNA (b4
integrin), we performed an analysis of a compendium of data
composed of macrophages from in vitro and in vivo datasets.
We observed that ITGB4 is expressed in both human and mouse
total macrophages (Figures 4A and S4A). From the same com-
pendium, a correlation between ITGB4 expression and signaling
downstream of TGF-b1 was established (Figure 4B). Single-cell
transcriptome analysis of non-tumor cells isolated from primary
breast tumors revealed that TAMs expressed high levels of
ITGB4, compared with other non-tumor cells within the tumor
microenvironment (Figure 4C). To identify patients who may
have enrichment of macrophages capable of lymphovascular re-
modeling, we used a gene signature containing genes enriched
in TEMs (Pucci et al., 2009) in a cohort of 122 TNBC gene expres-
sion patterns (Gazinska et al., 2013). We plotted the activation
score of the TEM gene signature against the TGF-bsignaling
pathway for each tumor and observed the enrichment of patients
with distant metastasis when both of these gene signatures were
present in the primary tumor (Figure 4D). Kaplan-Meier plots also
showed a signiﬁcant reduction in distant metastasis-free survival
(DMFS) in patients classiﬁed as having a high TEM-TGF-bactiva-
tion score (Figure 4E). To investigate the presence of lymphatic-
associated macrophages in breast cancer patients, samples
from 20 patients were used. Of these patients, 10 were previ-
ously characterized as having lymphatic vessel invasion (LVI),
and the remaining 10 did not have LVI. To assess macrophage
localization with respect to lymphatic vasculature, we dual-
stained sections with an antibody against CD14 and podoplanin
(Figure 4F). The sections were scored for the presence of CD14+
macrophages within or proximal to lymphatic vasculature. In our
cohort of 20 patients, all samples exhibited some degree of
CD14 and podoplanin positivity. Six cases (30%) had macro-
phages associated with lymphatic vessels; of these, 4 were
shown to be positive for LVI. In this small study, our results sug-
gest that 67% of patients with lymphatic-associated macro-
phages also have LVI. In a separate small patient cohort
(8 patients), we demonstrated CD68+ macrophages expressing
b4 integrin (ITGB4) in close proximity to podoplanin+ vessels
using consecutive parafﬁn-embedded sections (Figures 4G
and 4H). We quantiﬁed CD68+ITGB4+ macrophages per square
millimeter and saw an association between CD68+ITGB4+
Figure 4. b4 Integrin-Expressing Macrophages and Lymphatic Remodeling Associated with TGF-bSignaling and Adverse Outcome in TNBC
(A) ITGB4 expression in human macrophages. The y axis indicates normalized expression on log
scale. Red line indicates median expression of all genes. Raw
gene counts were obtained from the ARCHS4 database.
(B) Correlation between ITGB4 expression and enrichment of TGF-bsignaling in human macrophages (Spearman rho = 0.26; p < 0.001. The x axis indicates
normalized expression on the log
scale. The y axis indicates single sample gene set enrichment analysis (ssGSEA) enrichment scores computed for the TGF-b
hallmark gene set obtained from the molecular signatures database (MSigDB). Red curve indicates loess ﬁt. Associatio n strength was quantiﬁed using Spearman
correlation coefﬁcient. Raw gene counts were obtained from the ARCHS4 database.
(C) Expression of ITGB4 in single cell RNA sequencing (scRNaseq) data of primary breast cancer (GEO: GSE75688). Data are reported as log
transcripts per million.
(D) Activation score of TEM gene signature and TGF-bsignaling. Red and green dots indicate TNBC with or without distant metastasis, respectively. Enrichment
of TNBC with distant metastasis in the top right quadrant, established by hypergeo metric testing.
(E) Kaplan-Meier survival curves showing distant metastasis-free survival in TNBC. Stratiﬁcation based on samples with high TGF-bsignaling and TEM gene
signature activation score classiﬁed as ‘‘High TEM-TGFbsignature’’ versus the remainder (‘‘Low TEM-TGFbsignature’’).
(F) Representative breast cancer section (from n = 20) stained with CD14 (red) and podoplanin (brown). Scale bars, 100 mm. Zoomed inset demonstrates CD14+
macrophages associated with podoplanin+ lymphatic vasculature (black arrows). Tissues were selected from 8 patients with or without lymph node positivity.
Consecutive sections were stained singly for podoplanin lymphovasculature or doubly using pan-macrophage marker, CD68, and anti-b4 integrin antibody.
(G) Double-stained macrophages per square millimeter shown with patient clinical details (LVI and lymph node positivity).
(H) CD68+ITGB4+ macrophages are indicated in upper right panels (red arrows). CD68 and ITGB4 stainings are indicated below as 2 single panels;
CD68+ITGB4+ macrophages are indicated with red arrows. Podoplanin+ vessels shown in upper left images (black arrows). Scale bars, 20 mm.
Cell Reports 27, 1967–1978, May 14, 2019 1975
macrophage score and lymph node positivity in individual pa-
tients (Figure S4B). Future studies will endeavor to repeat this
small study in a larger patient cohort to investigate whether
this relationship is statistically signiﬁcant. The combination of
our data suggests that b4-integrin-expressing lymphovascular
macrophages may be driving LVI and subsequent metastasis
to lymph nodes via the lymphatic remodeling signaling cascade.
This study demonstrates how crosstalk between a previously
unreported tumor-inﬁltrating myeloid subpopulation and an
existing lymphatic vasculature can promote metastasis through
quantiﬁable architectural changes in lymphatic vessels. We
identiﬁed a population of b4 integrin-expressing macrophages
that drive lymphatic remodeling through TGF-bsignaling and
are associated with adverse pathological response in TNBC
Our study uses both endogenous BMMs and the RAW264.7
macrophage cell line, which is strain-matched to the lympho-
tropic tumor cell line, 4T1.2. Through intravital imaging and
ex vivo tissue analysis, our TNBC model allowed us to probe
the relationship between the tumor lymphatic vasculature and
macrophages in vivo and directly translate these phenotypic
observations into in vitro assays for mechanistic studies. We
then directly assessed the prognostic signiﬁcance of the key
molecules in the lymphatic signaling cascade in predicting
adverse pathological outcome for a cohort of TNBC patients.
In breast cancer samples previously characterized for LVI, we
identiﬁed lymphatic-associated macrophages in approximately
a third of the samples and show that LVI was present in the
majority of these cases. We identiﬁed b4 integrin-expressing
macrophages proximal to lymphatic endothelium in breast can-
cer samples and demonstrate that, in patients with a larger a6b4-
expressing macrophage inﬁltrate, there is a trend toward sentinel
lymph node metastasis. Our data suggest that b4 integrin-ex-
pressing macrophages may drive metastasis via the lymphovas-
cular route in human breast cancer.
Our study reveals that macrophages are retained in lymphatic
endothelium in a TNBC model through the upregulation of b4
integrin on tumor-educated macrophages. While the adhesion
receptor a6b4 integrin is ubiquitously expressed in early breast
cancer (Diaz et al., 2005), transcriptome analysis of breast can-
cer patient samples revealed a correlation between expression
levels and prognosis (Lu et al., 2008). Through analysis of b4
integrin at the transcriptome and protein levels, we demonstrate
a population of endogenous macrophages that express b4
integrin and are adherent to laminin-5 in lymphovascular areas.
Collectively, our data suggest that b4 integrin acts to ensure
that tumor-inﬁltrating macrophages are in a prime location for
sustained interaction with LECs.
We have deﬁned dual functionality of TGF-b1 where it can
affect signaling within TAMs and LECs. First, we show that
TGF-b1 is required for b4 integrin clustering at the macrophage
plasma membrane. Integrin clustering can positively regulate
levels of cell adhesion rapidly in response to soluble stimuli (Hy-
nes, 2002). TGF-bhas previously been demonstrated to control
a6b1 and a6b4 integrin clustering in HER2-overexpressing
mammary tumor cells (Wang et al., 2009). Here, we describe
TGF-b1-dependent b4 integrin clustering in macrophages that
control the macrophage-spreading response necessary for
TAM adhesion at the site of lymphatic vasculature.
Second, TGF-b1 acts in a paracrine manner to activate RhoA
in LECs lining the lymphatic vessel, as demonstrated through
RAICHU-ﬂuorecent lifetime imaging microscopy (FLIM) technol-
ogy (Heasman et al., 2010; Makrogianneli et al., 2009; Vega et al.,
2011). Our study shows that signaling within LECs in contact with
TAMs drives LEC contraction, which correlates to gross archi-
tectural changes and hyperpermeability of the lymphatic vessel
network that could actively facilitate metastasis. We have previ-
ously demonstrated the activation of RhoGTPases by integrin
signaling in cis (on the immune cells that are triggered by adhe-
sion processes (Makrogianneli et al., 2009; Carlin et al., 2011;
Heasman et al., 2010; Ramsay et al., 2013). Our present study
indicates that this phenomenon can also occur in trans, i.e., acti-
vation of RhoGTPases in the endothelial cells that are contacted
by the adherent macrophages, through the expression of factors
such as TGF-b1. The role of macrophage-released TGF-b1
in vivo is shown to have an effect on the RhoA pathway in prox-
imal LECs and a concomitant role in lymphovasculature
In summary, this study identiﬁes an alternative macrophage-
mediated signaling pathway involved in the promotion of
lymphatic metastasis. Our work emphasizes the importance in
considering crosstalk between macrophages and the lymphatic
vessel network in TNBC, where aggressive tumor growth and
rapid metastasis often mean a poor outcome. We hope this
study will guide future endeavors to focus on therapeutically
targeting the lymphatic remodeling signaling cascade in TNBC
Detailed methods are provided in the online version of this paper
and include the following:
dKEY RESOURCES TABLE
dCONTACT FOR REAGENT AND RESOURCE SHARING
dEXPERIMENTAL MODEL AND SUBJECT DETAILS
BHuman breast cancer samples
BRAW264.7 macrophage treatment
BImage acquisition and analysis for colocalization
studies in tissue
BStructured Illumination Microscopy (SIM)
BMammary imaging window implantation and intravital
BLymphatic vessel permeability
BLymphatic endothelial cell contraction
1976 Cell Reports 27, 1967–1978, May 14, 2019
BTGFb1 stable knockdown in RAW264.7 macrophages
BHuman tissue staining
dQUANTIFICATION AND STATISTICAL ANALYSIS
BGene expression microarray analysis
BAnalysis of gene signatures
dDATA AND SOFTWARE AVAILABILITY
Supplemental Information can be found online at https://doi.org/10.1016/j.
We thank Cancer Research UK King’s Health Partners Cancer Centre at King’s
College London; Nikon Imaging Centre, King’s College London; Mathew
Smalley for tumor tissue from BLG-Cre;Brca1
mice; Steven Alex-
ander for the murine SV-LEC line; UCL Pathology core facility; Kalnisha Naidoo
for advice; and James Arnold, Victoria Sanz-Moreno, Hellmut Augustin, and
Anne Ridley for reviewing the manuscript. This work was funded by the Cancer
Research UK King’s Health Partner’s Centre at KCL, KCL/UCL Comprehen-
sive Cancer Imaging Centre, and Breakthrough Breast Cancer (recently
merged with Breast Cancer Campaign, forming Breast Cancer Now).
R.E. conceptualized the study; designed, performed, and analyzed experi-
ments, and wrote the manuscript; F.F.-B. performed FACS (ﬂuorescence-acti-
vated cell sorting) acquisition and analysis, assisted with in vivo experiments,
assisted with experiment analysis, and assisted with writing the manuscript;
S.N. performed gene analysis on macrophage populations and assisted with
writing the manuscript; E.M. stained, quantiﬁed, and analyzed CD68+ITGB4+
patient tissues; K.L. performed in vitro macrophage gene array analysis and
assisted with writing the manuscript; A.G. analyzed TNBC gene expression
data and assisted with writing the manuscript; J.M. assisted with in vivo exper-
iments and writing the manuscript; C.G. and J.O. selected, stained, and
analyzed breast cancer sections; P.G. assisted with in vivo experiments;
V.M. designed the lymphatic disorganization scoring and assisted with data
analysis; A.C. assisted with analysis; F.N. assisted with antibody optimization;
P.B. gave technical advice on analyzing FRET-FLIM data; R.M. and E.F.-D.
performed tumor transplantation; G.F. and B.V. gave technical advice on intra-
vital imaging; M.S. contributed reagents; A.T. contributed to clinical translation
and reviewed the manuscript; F.F. wrote the colocalization software and
analyzed colocalization data; M.D.P. contributed reagents and reviewed the
manuscript; and T.N. provided funding, contributed to clinical translation,
and assisted with writing the manuscript.
DECLARATION OF INTERESTS
The authors declare no competing interests.
Received: June 12, 2018
Revised: March 14, 2019
Accepted: April 17, 2019
Published: May 14, 2019
Ando, T., Jordan, P., Joh, T., Wang, Y., Jennings, M.H., Houghton, J., and
Alexander, J.S. (2005). Isolation and characterization of a novel mouse
lymphatic endothelial cell line: SV-LEC. Lymphat. Res. Biol. 3, 105–115.
Avraamides, C.J., Garmy-Susini, B., and Varner, J.A. (2008). Integrins in angio-
genesis and lymphangiogenesis. Nat. Rev. Cancer 8, 604–617.
Balkwill, F.R., Capasso, M., and Hagemann, T. (2012). The tumor microenvi-
ronment at a glance. J. Cell Sci. 125, 5591–5596.
Barber, P.R., Ameer-Beg, S.M., Gilbey, J., Carlin, L.M., Keppler, M., Ng, T.C.,
and Vojnovic, B. (2009). Multiphoton time-domain ﬂuorescence lifetime imag-
ing microscopy: practical application to protein-protein interactions using
global analysis. J. R. Soc. Interface 6, S93–S105.
Barber, P.R., Tullis, I.D., Pierce, G.P., Newman, R.G., Prentice, J., Rowley,
M.I., Matthews, D.R., Ameer-Beg, S.M., and Vojnovic, B. (2013). The Gray
Institute ‘open’ high-content, ﬂuorescence lifetime microscopes. J. Microsc.
Barbie, D.A., Tamayo, P., Boehm, J.S., Kim, S.Y., Moody, S.E., Dunn, I.F.,
Schinzel, A.C., Sandy, P., Meylan, E., Scholl, C., et al. (2009). Systematic
RNA interference reveals that oncogenic KRAS-driven cancers require
TBK1. Nature 462, 108–112.
Bron, S., Henry, L., Faes-Van’t Hull, E., Turrini, R., Vanhecke, D., Guex, N., If-
ticene-Treboux, A., Marina Iancu, E., Semilietof, A., Rufer, N., et al. (2015).
TIE-2-expressing monocytes are lymphangiogenic and associate speciﬁcally
with lymphatics of human breast cancer. OncoImmunology 5, e1073882.
Bryan, B.A., Dennstedt, E., Mitchell, D.C., Walshe, T.E., Noma, K., Loureiro, R.,
Saint-Geniez, M., Campaigniac, J.P., Liao, J.K., and D’Amore, P.A. (2010).
RhoA/ROCK signaling is essential for multiple aspects of VEGF-mediated
angiogenesis. FASEB J. 24, 3186–3195.
Carlin, L.M., Evans, R., Milewicz, H., Fernandes, L., Matthews, D.R., Perani,
M., Levitt, J., Keppler, M.D., Monypenny, J., Coolen, T., et al. (2011). A tar-
geted siRNA screen identiﬁes regulators of Cdc42 activity at the natural killer
cell immunological synapse. Sci. Signal. 4, ra81.
Choi, W.W., Lewis, M.M., Lawson, D., Yin-Goen, Q., Birdsong, G.G., Cotsonis,
G.A., Cohen, C., and Young, A.N. (2005). Angiogenic and lymphangiogenic
microvessel density in breast carcinoma: correlation with clinicopatho logic
parameters and VEGF-family gene expression. Mod. Pathol. 18, 143–152.
Condeelis, J., and Pollard, J.W. (2006). Macrophages: obligate partners for
tumor cell migration, invasion, and metastasis. Cell 124, 263–266.
De Palma, M., Venneri, M.A., Galli, R., Sergi Sergi, L., Politi, L.S., Sampaolesi,
M., and Naldini, L. (2005). Tie2 identiﬁes a hematopoietic lineage of proang io-
genic monocytes required for tumor vessel formation and a mesenchymal
population of pericyte progenitors. Cancer Cell 8, 211–226.
De Palma, M., Murdoch, C., Venneri, M.A., Naldini, L., and Lewis, C.E. (2007).
Tie2-expressing monocytes: regulation of tumor angiogenesis and therapeutic
implications. Trends Immunol. 28, 519–524.
Dent, R., Trudeau, M., Pritchard, K.I., Hanna, W.M., Kahn, H.K., Sawka, C.A.,
Lickley, L.A., Rawlinson, E., Sun, P., and Narod, S.A. (2007). Triple-negative
breast cancer: clinical features and patterns of recurrence. Clin. Cancer Res.
Desgrosellier, J.S., and Cheresh, D.A. (2010). Integrins in cancer: biological
implications and therapeutic opportunities. Nat. Rev. Cancer 10, 9–22.
Diaz, L.K., Cristofanilli, M., Zhou, X., Welch, K.L., Smith, T.L., Yang, Y., Sneige,
N., Sahin, A.A., and Gilcrease, M.Z. (2005). Beta4 integrin subunit gene
expression correlates with tumor size and nuclear grade in early breast cancer.
Mod. Pathol. 18, 1165–1175.
Evans, R., Patzak, I., Svensson, L., De Filippo, K., Jones, K., McDowall, A., and
Hogg, N. (2009). Integrins in immunity. J. Cell Sci. 122, 215–225.
Finsterbusch, M., Voisin, M.B., Beyrau, M., Williams, T.J., and Nourshargh, S.
(2014). Neutrophils recruited by chemoattractants in vivo induce microvas-
cular plasma protein leakage through secretion of TNF. J. Exp. Med. 211,
Fleming, Y.M., Ferguson, G.J., Spender, L.C., Larsson, J., Karlsson, S.,
Ozanne, B.W., Grosse, R., and Inman, G.J. (2009). TGF-beta-mediated activa-
tion of RhoA signalling is required for efﬁcient (V12)HaRas and (V600E)BRAF
transformation. Oncogene 28, 983–993.
Gazinska, P., Grigoriadis, A., Brown, J.P., Millis, R.R., Mera, A., Gillett, C.E.,
Holmberg, L.H., Tutt, A.N., and Pinder, S.E. (2013). Comparison of basal-like
triple-negative breast cancer deﬁned by morphology, immunohistochemistry
and transcriptional proﬁles. Mod. Pathol. 26, 955–966.
Cell Reports 27, 1967–1978, May 14, 2019 1977
Gordon, E.J., Rao, S., Pollard, J.W., Nutt, S.L., Lang, R.A., and Harvey, N.L.
(2010). Macrophages deﬁne dermal lymphatic vessel calibre during develop-
ment by regulating lymphatic endothelial cell proliferation. Development 137,
Harney, A.S., Arwert, E.N., Entenberg, D., Wang, Y., Guo, P., Qian, B.Z., Oktay,
M.H., Pollard, J.W., Jones, J.G., and Condeelis, J.S. (2015). Real-time imaging
reveals local, transient vascular permeability, and tumor cell intravasatio n
stimulated by TIE2hi macrophage-derived VEGFA. Cancer Discov. 5,
Heasman, S.J., Carlin, L.M., Cox, S., Ng, T., and Ridley, A.J. (2010).
Coordinated RhoA signaling at the leading edge and uropod is required for
T cell transendothelial migration. J. Cell Biol. 190, 553–563.
Hynes, R.O. (2002). Integrins: bidirectional, allosteric signaling machines. Cell
Inman, G.J., Nicola
´s, F.J., Callahan, J.F., Harling, J.D., Gaster, L.M., Reith,
A.D., Laping, N.J., and Hill, C.S. (2002). SB-431542 is a potent and speciﬁc in-
hibitor of transforming growth factor-beta superfamily type I activin receptor-
like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol. Pharmacol. 62, 65–74.
Kedrin, D., Gligorijevic, B., Wyckoff, J., Verkhusha, V.V., Condeelis, J., Segall,
J.E., and van Rheenen, J. (2008). Intravital imaging of metastatic behavior
through a mammary imaging window. Nat. Methods 5, 1019–1021.
Kitamura, T., Qian, B.Z., Soong, D., Cassetta, L., Noy, R., Sugano, G., Kato, Y.,
Li, J., and Pollard, J.W. (2015). CCL2-induced chemokine cascade promotes
breast cancer metastasis by enhancing retention of metastasis-associated
macrophages. J. Exp. Med. 212, 1043–1059.
Lachmann, A., Torre, D., Keenan, A.B., Jagodnik, K.M., Lee, H.J., Wang, L.,
Silverstein, M.C., and Ma’ayan, A. (2018). Massive mining of publicly available
RNA-seq data from human and mouse. Nat. Commun. 9, 1366.
Li, W.V., and Li, J.J. (2018). An accurate and robust imputation method
scImpute for single-cell RNA-seq data. Nat. Commun. 9, 997.
Lelekakis, M., Moseley, J.M., Martin, T.J., Hards, D., Williams, E., Ho, P.,
Lowen, D., Javni, J., Miller, F.R., Slavin, J., and Anderson, R.L. (1999). A novel
orthotopic model of breast cancer metastasis to bone. Clin. Exp. Metastasis
Liu, H.T., Ma, R., Yang, Q.F., Du, G., and Zhang, C.J. (2009). Lymphangiogenic
characteristics of triple negativity in node-negative breast cancer. Int. J. Surg.
Pathol. 17, 426–431.
Lu, S., Simin, K., Khan, A., and Mercurio, A.M. (2008). Analysis of integrin beta4
expression in human breast cancer: association with basal-like tumors and
prognostic signiﬁcance. Clin. Cancer Res. 14, 1050–1058.
Makrogianneli, K., Carlin, L.M., Keppler, M.D., Matthews, D.R., Ofo, E.,
Coolen, A., Ameer-Beg, S.M., Barber, P.R., Vojnovic, B., and Ng, T. (2009).
Integrating receptor signal inputs that inﬂuence small Rho GTPase activation
dynamics at the immunological synapse. Mol. Cell. Biol. 29, 2997–3006.
Melchor, L., Molyneux, G., Mackay, A., Magnay, F.A., Atienza, M., Kendrick,
H., Nava-Rodrigues, D., Lo
´a, M.A., Milanezi, F., Greenow, K., et al.
(2014). Identiﬁcation of cellular and genetic drivers of breast cancer heteroge-
neity in genetically engineered mouse tumour models. J. Pathol. 233, 124–137.
Mohammed, R.A., Martin, S.G., Gill, M.S., Green, A.R., Paish, E.C., and Ellis,
I.O. (2007). Improved methods of detection of lymphovascular invasion
demonstrate that it is the predominant method of vascular invasion in breast
cancer and has important clinical consequences. Am. J. Surg. Pathol. 31,
Mohammed, R.A., Ellis, I.O., Mahmmod, A.M., Hawkes, E.C., Green, A.R.,
Rakha, E.A., and Martin, S.G. (2011). Lymphatic and blood vessels in basal
and triple-negative breast cancers: characteristics and prognostic signiﬁ-
cance. Mod. Pathol. 24, 774–785.
Molyneux, G., Geyer, F.C., Magnay, F.A., McCarthy, A., Kendrick, H., Natrajan,
R., Mackay, A., Grigoriadis, A., Tutt, A., Ashworth, A., et al. (2010). BRCA1
basal-like breast cancers originate from luminal epithelial progenitors and
not from basal stem cells. Cell Stem Cell 7, 403–417.
Peter, M., Ameer-Beg, S.M., Hughes, M.K., Keppler, M.D., Prag, S., Marsh, M.,
Vojnovic, B., and Ng, T. (2005). Multiphoton-FLIM quantiﬁcation of the EGFP-
mRFP1 FRET pair for localization of membrane receptor-kinase interactions.
Biophys. J. 88, 1224–1237.
Pollard, J.W. (2004). Tumour-educated macrophages promote tumour pro-
gression and metastasis. Nat. Rev. Cancer 4, 71–78.
Pucci, F., Venneri, M.A., Biziato, D., Nonis, A., Moi, D., Sica, A., Di Serio, C.,
Naldini, L., and De Palma, M. (2009). A distinguishing gene signature shared
by tumor-inﬁltrating Tie2-expressing monocytes, blood ‘‘resident’’ mono-
cytes, and embryonic macrophages suggests common functions and devel-
opmental relationships. Blood 114, 901–914.
Quail, D.F., and Joyce, J.A. (2013). Microenvironmental regulation of tumor
progression and metastasis. Nat. Med. 19, 1423–1437.
Ramsay, A.G., Evans, R., Kiaii, S., Svensson, L., Hogg, N., and Gribben, J.G.
(2013). Chronic lymphocytic leukemia cells induce defective LFA-1-directed T-
cell motility by altering Rho GTPase signaling that is reversible with lenalido-
mide. Blood 121, 2704–2714.
Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K.
(2015). limma powers differential expression analyses for RNA-sequencing
and microarray studies. Nucleic Acids Res. 43, e47.
Savage, N.D., de Boer, T., Walburg, K.V., Joosten, S.A., van Meijgaarden, K.,
Geluk, A., and Ottenhoff, T.H. (2008). Human anti-inﬂammatory macrophages
induce Foxp3+ GITR+ CD25+ regulatory T cells, which suppress via mem-
brane-bound TGFbeta-1. J. Immunol. 181, 2220–2226.
Stewart, R.L., and O’Connor, K.L. (2015). Clinical signiﬁcance of the integrin
a6b4 in human malignancies. Lab. Invest. 95, 976–986.
Vega, F.M., Fruhwirth, G., Ng, T., and Ridley, A.J. (2011). RhoA and RhoC have
distinct roles in migration and invasion by acting through different targets.
J. Cell Biol. 193, 655–665.
Wang, S.E., Xiang, B., Zent, R., Quaranta, V., Pozzi, A., and Arteaga, C.L.
(2009). Transforming growth factor beta induces clustering of HER2 and integ-
rins by activating Src-focal adhesion kinase and receptor association to the
cytoskeleton. Cancer Res. 69, 475–482.
Weisser, S.B., van Rooijen, N., and Sly, L.M. (2012). Depletion and reconstitu-
tion of macrophages in mice. J. Vis. Exp. (66), 4105.
Wong, S.Y., and Hynes, R.O. (2006). Lymphatic or hematogenous dissemina-
tion: how does a metastatic tumor cell decide? Cell Cycle 5, 812–817.
Yoshizaki, H., Ohba, Y., Kurokawa, K., Itoh, R.E., Nakamura, T., Mochizuki,
N., Nagashima, K., and Matsuda, M. (2003). Activity of Rho-family GTPases
during cell division as visualized with FRET-based probes. J. Cell Biol. 162,
1978 Cell Reports 27, 1967–1978, May 14, 2019
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER
Rat monoclonal anti-Lyve1 Novus Biologicals #NB-600-1008
Rabbit polyclonal anti-Tie2 (C-20) Santa Cruz #sc-324
Rabbit polyclonal phospho-Smad2/3 (D27F4) Cell Signaling #8828
Mouse monoclonal anti-ITGB4 Abcam #ab29042
Mouse monoclonal anti-CD68 antibody Ventana Cell Marque #168M
Mouse monoclonal anti-CD14 (EPR3653) Ventana Cell Marque #114R
Mouse monoclonal anti-podoplanin (D2-40) Ventana Cell Marque #332M
Rat monoclonal anti-CD45-APC-Cy7 Biolegend #103115
Rat monoclonal Ly6G-Biotin Biolegend #127603
Streptavidin AF488 Biolegend #405235
Rat monoclonal CD11b-eFluor450 ThermoFisher Scientiﬁc #48-0112-82
Rat monoclonal Tie-2 PE Biolegend #124007
Rat monoclonal b4 integrin-BV711 BDBiosciences #744154
CD31 PerCPCy5.5 Biolegend #102419
Rat monoclonal anti-F4/80-FITC (clone BM8) Abcam #Ab60348
Rabbit polyclonal anti-laminin-5 Abcam #Ab14509
Rabbit polycloncal Anti-Phospho myson light
Cell Signaling #3671
Mouse monoclonal anti-podoplanin antibody Santa Cruz #sc-166906
Rabbit polyclonal anti-TGFb1 antibody Proteintech #11522-1-AP
Breast cancer tumor tissues (parafﬁn-embedded) King’s College London breast cancer
Team lead – Dr Cheryl Gillet
4T1.2 tumor tissues (frozen) King’s College London Dr Rachel Evans
tumor tissues (frozen) King’s College London Dr Rebecca Marlow
Chemicals, Peptides, and Recombinant Proteins
red (CMTMR) and Cell tracker
Life Technologies #C34552, C2925
Murine CSF1 Sigma #M9170
Human recombinant laminin-5 Novus Biologicals #H00003911
Clodronate and PBS liposomes Liposoma Technology #CP-005-005
acetoxymethyl ester (BCECF)
Thermo Scientiﬁc #B1170
SB-431542 Sigma #S4317
Evans Blue dye Sigma #E2129
Formamide Sigma #F9037
76kDa dextran Texas Red Sigma #R05027
76kDa dextran ﬂuorescein Santa Cruz #sc-263323
Critical Commercial Assays
Murine TGFb1 quantikine ELISA kit R&D Ltd #MB100B
Murine TGFb2 quantikine ELISA kit R&D Ltd #DB250
Experiment ArrayExpress accession Array Express ArrayExpress: E-MTAB-4064.
Breast Cancer Gene Expression data Gene Expression Omnibus GEO: GSE75688
ARCHS4 database (Lachmann et al., 2018) N/A
(Continued on next page)
Cell Reports 27, 1967–1978.e1–e4, May 14, 2019 e1
CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to the Lead Contact, Tony Ng (firstname.lastname@example.org).
For a detailed description of the experimental procedures please see Supplemental Information.
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Bone marrow macrophages
Monocytes were isolated from female BALB/c mice femurs and cultured in mCSF-1 for 5 d.
All cell lines were tested as mycoplasma negative and authenticated by IDEXX Laboratories Ltd, UK.
BALB/c immune-competent mice were 6–8 weeks of age and maintained under pathogen-free conditions. Tumors were established
by injection of 1x10
4T1.2 (Lelekakis et al, 1999) cells into the mammary fat pad.
Mammary tumor chunks (approximately 0.2cm
) dissected from BLG-Cre;Brca1
mice (Molyneux et al., 2010) were trans-
planted orthotopically into mammary fat pads of recipient 5-week old C57BL6J mice. Tumors were grown for 4-8 weeks before
mice were culled and tumor tissues harvested.
Human breast cancer samples
Parafﬁn embedded samples (n = 20) (KHP Cancer Biobank Molecular Taxonomy of Breast Cancer International Consortium
(METABRIC) dataset cohort) were used. Ten patients were previously characterized as having lymphatic vessel invasion (LVI) and
the remaining 10 did not have LVI. Please see SI for details on staining.
REAGENT or RESOURCE SOURCE IDENTIFIER
Experimental Models: Cell Lines
4T1.2 cells derived from female BALB/C mouse (Lelekakis et al., 1999) N/A
SV-LEC (derived from male ‘‘immortomouse’’) (Ando et al., 2005) Gift from Dr Steven Alexander
Primary LEC from male BALB/C mouse Generon Ltd #BALB5064L
RAW264.7 derived from male BALB/C mouse ATCC Ltd #ATCC-TIB71
HEK293T (derived from human fetus) ATCC Ltd #ATCC-CRL-11268
Experimental Models: Organisms/Strains
Female BALB/c mice Charles River N/A
Female C57Bl6J mice Charles River N/A
RAICHU RhoA biosensor construct King’s College London (Heasman et al., 2010)
GFP-RhoA construct King’s College London (Heasman et al., 2010)
Software and Algorithms
Gray Laboratories Oxford University and
University College London
Dr Paul Barber (Barber et al., 2013)
Prism Software https://www.graphpad.com/
ImageJ (Fiji) https://imagej.nih.gov/ij/ N/A
Colocalization plugin for ImageJ (within this manuscript) Dr Fred Festy
TGFb1 shRNA (GIPZ) Open Biosystems University College London library
ITGB4 shRNA (GIPZ) Open Biosystems University College London
RNA easy minikit Quiagen #74104
Live/Dead Yellow dye Invitrogen #L34959
Affymetrix Mouse Gene 1.0 ST arrays Thermo Scientiﬁc #901168
e2 Cell Reports 27, 1967–1978.e1–e4, May 14, 2019
All experiments were performed in accordance with the local ethical review panel, the UK Home Ofﬁce Animals Scientiﬁc Procedures
Act, 1986 and the UKCCCR guidelines.
RAW264.7 macrophage treatment
Tumor-bearing mice were injected with 100 mL PBS or 1x10
RAW264.7 macrophages starting on the second day after tumor inoc-
ulation and repeated every 2 days until the end of the experiment.
Endogenous macrophages were ablated using clodronate-containing liposomes (Weisser et al., 2012).
Tissue sections were ﬁxed with 4% paraformaldehyde (PFA), blocked in 5% BSA followed by staining. Hoechst-33342 (0.1 mg/ml)
was used for nuclear staining and samples mounted using Mowiol (with DABCO). Image acquisition by confocal microscopy was
performed using a Nikon Eclipse Ni-E Upright. Image acquisition was conducted using NIS Elements C software and analyzed using
Image acquisition and analysis for colocalization studies in tissue
Cy3 and AF647 dyes were imaged before and after photobleaching using (x20 0.75NA air objective, Nikon) and a cooled CCD de-
tector (Hamamatsu ORCA-03G, 1024 31024) with respective integration time of 100 ms and 1000 ms. Dyes were photobleached
using a mode-locked Titanium Sapphire Laser (Coherent, Chameleon Ultra 2) tuned at 730 nm with pulse duration of about 200
fs, a repetition rate of 80 MHz and average laser power on the sample of 30 mW. To measure the relative level of b4 integrin expres-
sion within the lymphovasculature compared with the rest of the tissue, we measured average AF647 intensity within lymphovascu-
lature areas (high Cy3 intensity) normalized by the average AF647 intensity outside lymphovasculature areas (low Cy3 intensity).
Structured Illumination Microscopy (SIM)
RAW264.7-NTC or RAW264.7-TGFb1 KD were stained with rat anti-b4 integrin antibody and anti-rat AF647 antibody. Image acqui-
sition by SIM was performed using Nikon N-SIM microscope equipped with a 640nm laser, a Andor iXon Ultra 897 EMCCD camera
and a 100x 1.49NA oil immersion objective. Images were analyzed using ImageJ software.
Mammary imaging window implantation and intravital microscopy
Mammary Imaging Window (MIW) surgery was performed 10-14 days after tumor innoculation (Kedrin et al., 2008). Images shown are
representative of a minimum of 5 independent experiments.For imaging lymphatic vasculature, mice were injected subcutaneously
at the tail base with 50 mL 76kDa dextran-ﬂuorescein or dextran-Texas red 15 min prior to imaging. Mice were imaged for a maximum
period of 4 h per day using a x20 air objective. All post hoc image processing and image reconstructions were done using ImageJ
Lymphatic vessel permeability
Tumor-bearing mice were injected subcutaneously at the tail base with 1% Evans Blue dye. After 30 min the mice were culled and the
tumors incubated in formamide overnight at 55C. Optical density of formamide was read at 620nm and quantiﬁcation of lymphatic
permeability was given as OD per g tumor.
Laminin-5 was plated onto 96 well plates overnight at 4C and non-speciﬁc interactions blocked with BSA. Macrophages (5 310
were labeled with 1 mM2
0,70-bis-(2-carboxyethyl)-5-(and-6)-carboxyﬂuorescein-acetoxymethyl ester (BCECF) for 30 min at room
temperature. 100 mL of cells (1 310
/ml) were added at 37C, plates washed, and adhering macrophages quantiﬁed using a ﬂuo-
rescence microtiter plate reader.
Lymphatic endothelial cell contraction
SV-LEC cells or primary lymphatic endothelial cells were grown as a monolayer. On day 3 LECs and macrophages were stained for
30min at 37C using 1 mg/ml CMTMR or CMFDA respectively. Macrophages were added to SV-LEC monolayers overnight. Confocal
images of the co-culture and the area around individual SV-LECs was calculated using ImageJ software.
SV-LECs were transiently transfected with the RAICHU RhoA biosensor (Yoshizaki et al., 2003). The biosensor was modiﬁed to
express GFP and mRFP (Makrogianneli et al., 2009). Multiphoton time-correlated single photon counting FLIM was performed to
Cell Reports 27, 1967–1978.e1–e4, May 14, 2019 e3
quantify RhoA biosensor FRET Fluorescence excitation was provided by a Fianium laser, which generates optical pulses with a dura-
tion of 40 ps at a repetition rate of 80 MHz. For the imaging of RAICHU-transfected SV-LECs, multi-photon excitation was employed
using a solid-state pumped (8-W Verdi; Coherent), femtosecond self-mode locked Ti:Sapphire (Mira; Coherent) laser system (Peter
et al., 2005; Barber et al., 2009). Imaging data comprised of 256 3256 pixel resolution and 256 time channels. The ﬂuorescence life-
time was calculated as described (Barber et al., 2013).
TGFb1 stable knockdown in RAW264.7 macrophages
Stable TGFb1 knockdown RAW 264.7 macrophage lines were generated by lentiviral transduction using the pGIPZ system (Open
Biosystems). Viral packaging was performed by transiently transfecting HEK293T cells with the pGIPZ shRNA transfer vector and
the accessory plasmids pCMV-dR8.91 and pMD2G. Stable cell lines were established using three different shRNA lentiviral vectors.
RAW 264.7 macrophages were cultured in puromycin (1 mg/ml) to enable the selection of successfully transduced cells and efﬁcacy
of knockdown was assessed by western blotting.
RAW264.7 cell lines (TGFb1-knockdown or NTC) were stained with a Live-Dead Yellow dye followed by staining with a primary rat
anti-b4 integrin antibody and anti-rat AF647-conjugated secondary antibody.
Tumors were disaggregated with Collagenase (Sigma UK) and DNase I (Applichem, UK) before staining with Live-Dead Yellow,
CD45-APC Cy7, Ly6G-Biotin + Streptavidin AF488, CD11b-eFluor450, Tie-2 PE b4 integrin-BV711 and CD31 PerCPCy5.5. Cells
were ﬁxed with 1% PFA and analyzed in a FACS Canto II (BD Biosciences) cytometer. Data analyzed using FlowJo software (TreeStar
Inc., Ashland, OR, USA).
Human tissue staining
Sections were stained using anti-CD14/anti-podoplanin using Ventana Benchmark Ultra and Ultra view DAB and Alkaline Phospha-
tase detection systems. Sections were assessed independently by two histopathologists and scored for CD14+ macrophages within
or proximal to lymphatic vasculature.
Alternatively, using consecutive sections the ﬁrst section was stained with anti-podoplanin and the second section stained with
anti-ITGB4 anti-CD68. All sections were stained with DAB+ substrate/chromagen. All incubations were at room temperature.
The slides were scanned in the Hamamatsu NanoZoomer S210 Digital slide scanner. The image analysis was performed on the
whole section with the color deconvolution module and the positive pixel algorithm from QuPath image analysis software.
QUANTIFICATION AND STATISTICAL ANALYSIS
Gene expression microarray analysis
RNA was extracted from macrophage cell cultures and proﬁled using Affymetrix Mouse Gene 1.0 ST arrays. Differential expression be-
tween conditions was estimated by ﬁtting a linear model and performing empirical Bayes moderated t tests using the package ‘limma’
(v3.22.4) (Ritchie et al., 2015).The expression score for a speciﬁc gene ineach sample is deﬁned as the weighted sum of gene-standard-
ized (Z-score) expression values, with weights +1/-1 according to relative increase or decrease in BMM + 4T1.2 compared with BMM.
Analysis of gene signatures
To establish ITGB4 expression and assess association between ITGB4 expression and activation of the TGFbsignaling in macro-
phages, processed gene counts were obtained from the ARCHS4 database (Lachmann et al., 2018) and further normalized for down-
stream analyses. Enrichment of TGFbsignaling was computed using the ssGSEA method (Barbie et al., 2009) as implemented in the
GSVA package from Bioconductor.
False zero expression due to dropout events in scRNA-seq data was corrected using the scImpute algorithm as previously
described (Li and Li, 2018). scRNaseq data is reported as log2(TPM+1).
Macrophage-mediated vascular remodeling pathway signature (Pucci et al., 2009) was converted to a human gene list using Bio-
mart ID conversion (Ensembl Genes 84// Mus musculus genes GRCm38.p4). TGFb(KEGG) gene signature was derived from
(MSigDB). Gene signature activity was calculated using a weighted average sum over all genes for each tumor. Pearson’s correlation
between the activation scores was reported. Hypergeometric testing was used to establish the signiﬁcance of overlap between
TNBC with distant metastasis (DM) on those of dual high activation scores. Kaplan-Meier plots were generated for each dataset
to provide a visualization of survival stratiﬁcation.
All other statistical analysis is described in the text and legends and was performed using Prism software (GraphPad). P values less
than 0.05 were considered signiﬁcant. The statistical test used is indicated in the ﬁgure legends and the signiﬁcance of ﬁndings is
indicated in the ﬁgures.
DATA AND SOFTWARE AVAILABILITY
The accession number for the microRNA experimental data reported in this paper is ArrayExpress: E-MTAB-4064.
e4 Cell Reports 27, 1967–1978.e1–e4, May 14, 2019