Development of a new method for isolation and long-term
culture of organ-specific blood vascular and lymphatic
endothelial cells of the mouse
Takashi Yamaguchi, Taeko Ichise, Osamu Iwata, Akiko Hori, Tomomi Adachi, Masaru Nakamura,
Nobuaki Yoshida and Hirotake Ichise
Laboratory of Gene Expression and Regulation, Center for Experimental Medicine, Institute of Medical Science, University of Tokyo, Japan
As an indispensable component of the vascular system,
endothelial cells (ECs) have pivotal roles in develop-
ment and in health and disease [1]. Their properties
have been studied by a combination of in vitro analy-
ses of human primary ECs and in vivo analyses of
genetically modified mice exhibiting vascular pheno-
types. Human primary ECs are well-established
resources and are suitable for studying signal transduc-
tion and cellular physiology in vitro. However, it is still
difficult to control their gene expression strictly by
current overexpression and knockdown procedures. In
addition, they are not representative of all types of
ECs at various developmental stages and in vascular
beds [2]. On the other hand, the use of genetically
Keywords
Cre ⁄loxP recombination; endothelial cell
culture; endothelial heterogeneity; SV40
tsA58 large T antigen; transgenic mouse
Correspondence
H. Ichise, Laboratory of Gene Expression
and Regulation, Center for Experimental
Medicine, Institute of Medical Science,
University of Tokyo, 4-6-1 Shirokanedai,
Minato-ku, Tokyo 108-8639, Japan
Fax: +81 3 5449 5455
Tel: +81 3 5449 5754
E-mail: h-ichise@ims.u-tokyo.ac.jp
(Received 27 November 2007, revised 13
February 2008, accepted 22 February 2008)
doi:10.1111/j.1742-4658.2008.06353.x
Endothelial cells are indispensable components of the vascular system, and
play pivotal roles during development and in health and disease. Their
properties have been studied extensively by in vivo analysis of genetically
modified mice. However, further analysis of the molecular and cellular phe-
notypes of endothelial cells and their heterogeneity at various developmen-
tal stages, in vascular beds and in various organs has often been hampered
by difficulties in culturing mouse endothelial cells. In order to overcome
these difficulties, we developed a new transgenic mouse line expressing the
SV40 tsA58 large T antigen (tsA58T Ag) under the control of a binary
expression system based on Cre ⁄loxP recombination. tsA58T Ag-positive
endothelial cells in primary cultures of a variety of organs proliferate con-
tinuously at 33 C without undergoing cell senescence. The resulting cell
population consists of blood vascular and lymphatic endothelial cells,
which could be separated by immunosorting. Even when cultured for two
months, the cells maintained endothelial cell properties, as assessed by
expression of endothelium-specific markers and intracellular signaling
through the vascular endothelial growth factor receptors VEGFR–2 and
VEGFR-3, as well as their physiological characteristics. In addition, lym-
phatic vessel endothelial hyaluronan receptor-1 (Lyve-1) expression in liver
sinusoidal endothelial cells in vivo was retained in vitro, suggesting that an
organ-specific endothelial characteristic was maintained.These results show
that our transgenic cell culture system is useful for culturing murine endo-
thelial cells, and will provide an accessible method and applications for
studying endothelial cell biology.
Abbreviations
BEC, blood vascular endothelial cell; DiI, 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate; EC, endothelial cell; ESC,
embryonic stem cell; HRP, horseradish peroxidase; LDL, low-density lipoprotein; LEC, lymphatic endothelial cell; Lyve-1, lymphatic vessel
endothelial hyaluronan receptor-1; MACS, magnetic-activated cell separation; MAPK, mitogen-activated protein kinase; PFA,
paraformaldehyde; Prox-1, prospero-related homeobox-1; SV40T Ag, SV40 large T antigen; tsA58T Ag, large T antigen of SV40 mutant strain
tsA58; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor.
1988 FEBS Journal 275 (2008) 1988–1998 ª2008 The Authors Journal compilation ª2008 FEBS
modified mice has accelerated the understanding of
genetic mechanisms of endothelial development and
functions. However, further analyses of vascular phe-
notypes in vivo have been hampered by the compli-
cated relationship between ECs and non-ECs such as
mural, hematopoietic and mesenchymal fibroblast cells,
even though a conditional genetic modification such as
endothelium-specific knockouts can provide a partial
solution to this problem. Therefore, the isolation and
maintenance of murine endothelial cells from various
developmental stages and locations is important for
dissecting molecular and cellular mechanisms of endo-
thelial development and function.
Murine primary cells, including ECs, have a more
limited growth potential than human primary cells.
Thus, ‘immortalization’ techniques have been strongly
recommended for most analyses that require a large
quantity of transcripts, proteins or cells. For immortal-
ization of ECs, viral oncogenic proteins have been
used in previous studies. The polyoma middle T anti-
gen (PyMT Ag) allows selective proliferation of ECs in
mixed-cell populations [3–5], aiding in analyses of
genetically modified ECs in vitro [6–11]. However,
PyMT Ag causes endothelioma or hemangioma in vivo
[3] and mimics activated receptor tyrosine kinases [12],
which might obscure the analysis of endogenous recep-
tor-mediated signaling. Alternatively, tsA58T Ag, a
mutated SV40T Ag leading to temperature-dependent,
cell-type-independent cell proliferation [13,14], has
been used for ‘conditional immortalization’ of ECs of
wild-type and genetically modified mice [15–22].
Despite the fact that tsA58T Ag-directed immortaliza-
tion of ECs has been demonstrated, the method has
been under-utilized due to the specialized techniques
and expertise that are required for immunological iso-
lation of ECs [23,24] to prevent proliferation of
tsA58T Ag-expressing non-ECs.
Results and Discussion
Generation of a transgenic mouse line carrying
the CAG-bgeo-tsA58T Ag transgene
In order to circumvent the problems described above,
we developed a new transgenic mouse line expressing
tsA58T Ag under the control of a binary expression
system based on Cre ⁄loxP recombination. To obtain a
transgenic mouse line with the potential to express
tsA58T Ag in a variety of tissues including ECs, we
exploited embryonic stem cell (ESC)-mediated trans-
genesis. Briefly, we constructed a transgene driven by
the CAG promoter [25] that expresses the b–geo gene
[26] in the absence of Cre recombinase, but expresses
the tsA58T Ag gene after Cre-mediated excision of
the lox P-flanked b–geo gene (Fig. 1). The plasmid
vector-free transgene was introduced into ESCs, and
G418-resistant clones were selected. We next per-
formed 5-bromo-4-chloro-3-indolyl-b-d-galactopyrano-
side (X-gal) staining of embryoid bodies derived from
each clone and screened for the expression pattern of
b–geo in the embryoid bodies. Clone T26 had the most
favorable b–geo expression pattern among the G418-
resistant clones (data not shown). tsA58T Ag expres-
sion in ESCs after Cre-mediated excision was verified
by Western blotting (data not shown). The T26 trans-
genic mouse line was obtained through germline trans-
mission from chimeric mice. They grew normally, were
fertile, and did not display any defects.
Endothelium-specific expression of tsA58T Ag
in the transgenic mouse
We next crossed female T26 transgenic mice with
male Tie2–Cre transgenic mice [27], which removed a
loxP-flanked DNA fragment in endothelial cells and
T26 Tg
T26/Tie2-Cre Tg
Tie2-Cre Tg
tsA58T Ag-expressing
endothelial cell
pA
loxP loxP
pA
tsA58T
CAG
tsA58T
CAG
Enzymatic digestion of organs
Culture at 33 °C
Serial passages every 2
–
3 days
at split ratio 1 : 3
day 20–30
day 0
βgeo
Fig. 1. An endothelial cell culture scheme based on endothelium-
specific expression of tsA58T antigen. pA, polyadenylation signal
sequence; Tg, transgenic mouse.
T. Yamaguchi et al. A new method for mouse endothelial cell culture
FEBS Journal 275 (2008) 1988–1998 ª2008 The Authors Journal compilation ª2008 FEBS 1989
hematopoietic cells (Fig. 1). The resulting T26 ⁄Tie2–
Cre double-transgenic mice were born and grew
normally, but died suddenly within 6–12 weeks after
birth. To determine whether the expression of tsA58T
Ag was induced in ECs, we performed immunohisto-
chemistry using antibodies against the pan-EC mar-
ker, CD31, the lymphatic endothelial and liver
sinusoidal endothelial marker Lyve-1 (lymphatic vessel
endothelial hyaluronan receptor-1) [28–32] and SV40T
Ag. Immunostaining revealed that tsA58T Ag was
Brain
A
C
B
Uterus HeartLung Liver
SV40T CD31 SV40T Lyve-1
EmbryoYolk sac
SV40T CD31
ThymusCardiac valve
SV40T CD31
Fig. 2. Expression pattern of tsA58T Ag in T26 ⁄Tie2–Cre double-transgenic mice. (A) tsA58T Ag (red) was expressed in CD31-positive ECs
(green) of an E9.5 T26 ⁄Tie2–Cre double-transgenic embryo and its yolk sac. (B) tsA58T Ag (red) was expressed in CD31-positive ECs (green,
left panels) and Lyve-1-positive ECs (green, right panels) of 3)6-week-old T26 ⁄Tie2–Cre double-transgenic mice. Lyve-1-positive ECs were
not detected in the brain (top right), which is known to be an LEC-free organ. (C) tsA58T Ag (red) was also expressed in non-endothelial cells
of the thymic medulla and interstitial cells of the cardiac valve. Arrowheads indicate CD31-positive ECs (green). All micrographs are shown
at the same magnification. Scale bar = 50 lm.
A new method for mouse endothelial cell culture T. Yamaguchi et al.
1990 FEBS Journal 275 (2008) 1988–1998 ª2008 The Authors Journal compilation ª2008 FEBS
expressed in CD31-positive ECs of E9.5 embryos
proper and yolk sacs (Fig. 2A). Postnatally, tsA58T
Ag was not only expressed in CD31-positive ECs in
the brain, heart, lung, liver and uterus (Fig. 2B), but
was also expressed in Lyve-1-positive lymphatic endo-
thelial cells (LECs) in the heart, lung and uterus, and
sinusoidal ECs in the liver of 3–6-week-old double-
transgenic mice (Fig. 2B), indicating that endothe-
lium-specific expression of tsA58T Ag was achieved
as expected. Despite the mortality of the young dou-
ble-transgenic mice, no gross abnormalities, such as
endothelial hyperplasia, dysplasia or bleeding, could
be found in live or dead double-transgenic mice.
However, immunostaining revealed that tsA58T Ag
was expressed in non-ECs, including a subset of thy-
mocytes and cardiac valvular cells (Fig. 2C). These
observations are comparable to those of previous
studies using the same Tie2–Cre transgenic mouse
line, which showed that recombination occurred in
hematopoietic cells as well as ECs [27], and that
cardiac valvular cells were derived from endothelial
cells through an endothelial-to-mesenchymal transi-
tion during early development [33]. The presence of
these cells may cause a dysfunctional cardiac flow
and cause the sudden death of the transgenic mice,
although it remains to be determined whether T anti-
gen-expressing cardiac valves are functionally affected.
Endothelial cell culture from organs of
T26/Tie2–Cre double-transgenic mice
Following the demonstration of endothelium-specific
expression of tsA58T Ag in vivo, we performed primary
cell culturing (Fig. 1). Several organs (the brain, heart,
lung, liver and uterus) were obtained from 3-week-old
T26 single- or T26 ⁄Tie2–Cre double-transgenic mice,
dissected, and dissociated by enzymatic digestion. Dis-
persed cell suspensions were plated onto gelatin-coated
plastic dishes and cultured at 33 C (day 0 in Fig. 1).
For the initial 2 weeks, primary cells, including both
tsA58T Ag-negative cells (primarily fibroblasts) and
tsA58T Ag-positive cells, proliferated, and ECs could
barely be morphologically distinguished. However,
tsA58T Ag-negative cells gradually stopped proliferat-
ing and underwent senescence at about 2 weeks, as
assessed by morphology (data not shown). In contrast,
the remaining cells continued to proliferate over the
2 weeks and formed colonies that were distinguishable
under light-field microscopy (data not shown). tsA58T
Ag-negative senescent cells were progressively excluded
by serial passages. At day 30, the dishes consisted
almost exclusively of viable tsA58T Ag-positive cells
(Figs 1 and 3A). Cells obtained from T26 single-trans-
genic mice did not grow beyond 2–3 weeks (data not
shown), confirming that tsA58T Ag-directed prolifera-
tion was only achieved by Cre-mediated excision.
Characterization of tsA58T Ag-expressing
endothelial cell populations
In order to examine whether the tsA58T Ag-positive
cells maintained EC properties, we first performed
immunocytochemistry for EC markers and assessed the
uptake of acetylated low-density lipoproteins (LDLs).
The cell populations derived from the brain, lung, heart,
liver and uterus stained positive for CD31 (Fig. 3B for
the brain, liver and uterus; data not shown for the lung
and heart), strongly suggesting that the tsA58T Ag-posi-
tive cells originated from ECs. A subset of the cell popu-
lations from the lung and heart (data not shown) and a
Liver Uterus
Brain Liver Uterus
Brain
DAPI
A
B
C
D
tsA58T Ag Merge
Brain Uterus Brain
Fig. 3. Endothelial cell culture from organs of T26 ⁄Tie2–Cre dou-
ble-transgenic mice. (A) Proliferating cells obtained from the brain
were immunostained for SV40T Ag. Proliferating cells without
undergoing senescence were tsA58T Ag-positive. DAPI, 4,6-diami-
dino-2-phenylindole. (B,C) Immunostaining revealed that tsA58T
Ag-positive proliferating cells obtained from each organ maintained
expression of the endothelial-specific markers CD31 (B) and Lyve-1
(C). (D) 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine per-
chlorate (DiI)-labeled acetylated LDLs are taken up by these cells.
Bar = 50 lm (D, right panel) or 200 lm (all other panels).
T. Yamaguchi et al. A new method for mouse endothelial cell culture
FEBS Journal 275 (2008) 1988–1998 ª2008 The Authors Journal compilation ª2008 FEBS 1991
substantial proportion of the cell population from the
uterus (Fig. 3C) also stained positive for Lyve-1, indi-
cating that cell populations obtained from these tissues
were a mixture of blood vascular ECs (BECs) and
LECs. DiI-labeled acetylated LDLs were taken up by
all types of cell populations (Fig. 3D for brain and uter-
ine ECs; data not shown for others), but not by non-
endothelial NIH3T3 cells (data not shown), indicating
that the cells maintained the physiological characteristic
of acetylated-LDL uptake.
Lyve-1-positive liver sinusoidal endothelial cells
in vitro
Intriguingly, almost all of the cell population from the
liver were also positive for Lyve-1 (Fig. 3C), and Wes-
tern blot analysis revealed that they were Lyve-1-posi-
tive, prospero-related homeobox-1 (Prox-1)-negative
[34] ECs (Fig. 4B), suggesting that the population rep-
resented Lyve-1-positive liver sinusoidal ECs [30–32]
and maintained the property of Lyve-1 expression
in vitro. These results also suggest that Lyve-1 expres-
sion in liver sinusoidal ECs, reported as a marker of
differentiated organ-specific ECs [32] and a potential
diagnostic marker of liver cancer and cirrhosis [30], is
regulated in a cell-autonomous manner and is irrevers-
ible in the culture conditions used in this study. These
cultured ECs might allow us to investigate more prop-
erties of liver sinusoidal ECs in health and disease.
Isolation and characterization of BECs and LECs
We next isolated LECs from the mixed cell population
by magnetic immunosorting using an antibody against
Lyve-1 (Fig. 4A). We used uterine ECs for this purpose
because they contained large numbers of Lyve-1-posi-
tive cells as assessed by immunostaining (Fig. 3C) and
further confirmed by double staining for Lyve-1 and
another lymphatic endothelial marker, Prox-1 [34]
(Fig. 4A). As shown by the immunostaining of posi-
tively sorted or depleted cells (Fig. 4A), Lyve-1-positive
ECs were enriched as expected. Western blot analysis
revealed that Prox-1 and vascular endothelial growth
factor receptor 3 (VEGFR-3), which is expressed pre-
dominantly in LECs [35,36], were also expressed in
Lyve-1-positive ECs (Fig. 4B), confirming that LECs
were obtained from the mixed EC population.
tsA58T Ag-positive BECs and LECs transduced
signals of endothelial growth factors
We further examined whether isolated ECs constitu-
tively expressing tsA58T Ag could respond to
angiogenic and lymphangiogenic growth factors.
Serum-depleted LECs were treated with vascular endo-
thelial growth factors A or C (VEGF-A or VEGF-C)
(Fig. 4C). Phosphorylation of VEGFR-2 and mitogen-
activated protein kinases (MAPKs), but not of
VEGFR-3, was induced by VEGF-A, whereas phos-
phorylation of VEGFR-2, VEGFR-3 and MAPKs was
induced by VEGF-C, as reported in a previous study
using human primary LECs [36]. These results suggest
that growth factor signals were transduced properly
via endothelium-specific receptors in these cells. Mes-
enteric BECs and LECs (Fig. 5A,B) were also obtained
by the same strategy as illustrated in Figs 1 and 4, and
were treated with VEGF-A or VEGF-C (Fig. 5C).
MAPK and Akt phosphorylation were induced in both
BECs and LECs by stimulation with VEGF-A or
VEGF-C, indicating that the cultured ECs responded
to the endothelial growth factors.
Implications for tube formation-based assays and
transfection assays of tsA58T Ag-expressing ECs
We also examined whether the cells formed tube-like
structures on collagen gel. Both uterine BECs and
LECs could form tube-like structures (Fig. 4D). In
addition, an SV40-ori-containing plasmid carrying a
GFP expression cassette could be introduced by lipo-
fection and maintained for at least 5 days after trans-
fection as assessed by GFP expression (Fig. 4E). These
Fig. 4. Isolation and characterization of uterine BECs and LECs expressing tsA58T Ag. (A) Scheme for sorting of LECs from the uterine EC
population (days 30–40). A substantial proportion of uterine ECs were positive for Lyve-1 and Prox-1 (red and green on the top panel, respec-
tively), indicating that the uterine EC population was a mixed cell population of BECs and LECs. Lyve-1-positive LECs were isolated from
mixed ECs by magnet immunosorting using anti-Lyve-1 antibody. Scale bars ¼200 nm. (B) Western blotting revealed that Lyve-1-positive
uterine ECs maintained expression of Lyve-1, Prox-1 and VEGFR-3, indicating that they represent LECs. In contrast, liver ECs were positive
for Lyve-1 but not for Prox-1, indicating that they represent liver sinusoidal ECs. (C) The uterine LECs transduced growth-factor signals via
VEGFR-2 and VEGFR-3. IP, immunoprecipitation; IB, immunoblot; P-Y, phosphotyrosine. (D) The uterine BECs and LECs formed tube-like
structures. Bars = 200 lm. (E) SV40-ori-positive plasmids bearing GFP and drug-resistance genes were maintained in the uterine BECs and
LECs for at least 5 days after transfection under drug-selection pressure. Bars = 200 lm. All cells were cultured at 33 C, and day 40–50 ECs
were used for experiments shown in B–E.
A new method for mouse endothelial cell culture T. Yamaguchi et al.
1992 FEBS Journal 275 (2008) 1988–1998 ª2008 The Authors Journal compilation ª2008 FEBS
