Hyperplasia of Lymphatic Vessels in
VEGF-C Transgenic Mice
Michael Jeltsch,* Arja Kaipainen,* Vladimir Joukov,
Xiaojuan Meng, Merja Lakso, Heikki Rauvala, Melody Swartz,
Dai Fukumura, Rakesh K. Jain, Kari Alitalo†
No growth factors specific for the lymphatic vascular system have yet been described.
Vascular endothelial growth factor ( VEGF) regulates vascular permeability and angio-
genesis, but does not promote lymphangiogenesis. Overexpression of VEGF-C, a ligand
of the VEGF receptors VEGFR-3 and VEGFR-2, in the skin of transgenic mice resulted
in lymphatic, but not vascular, endothelial proliferation and vessel enlargement. Thus,
VEGF-C induces selective hyperplasia of the lymphatic vasculature, which is involved in
the draining of interstitial fluid and in immune function, inflammation, and tumor me-
tastasis. VEGF-C may play a role in disorders involving the lymphatic system and may
be of potential use in therapeutic lymphangiogenesis.
The four known members of the VEGF
family, VEGF (1), VEGF-B (2), VEGF-C
(3, 4), and platelet-derived growth factor
(5) have different roles as regulatory factors
of endothelia (6). In order to clarify the
function of VEGF-C in vivo, its cDNA was
cloned between the human keratin 14
(K14) promoter and polyadenylation signal
for expression in transgenic mice (7). The
K14 promoter has been shown to target
gene expression to the basal cells of strati-
fied squamous epithelia (8). Of the 27 mice
analyzed at 3 weeks of age, three were trans-
gene-positive, having approximately 40 to
50, 20, and 4 to 6 copies, respectively, of
the transgene in their genome. The latter
two mice transmitted the gene to two of 40
and six of 11 pups, respectively.
The transgenic mice were small and had
slightly swollen eyelids and poorly devel-
oped fur. Histological examination showed
that the epidermis was hyperplastic and the
number of hair follicles was reduced; these
effects were considered secondary to other
phenotypic changes (9). The dermis was
atrophic, and its connective tissue was re-
placed by large, dilated vessels devoid of red
cells but lined with a thin endothelial cell
layer (Fig. 1, A and B). Such abnormal
vessels were confined to the dermis and
resembled the dysfunctional, dilated spaces
characteristic of hyperplastic lymphatic ves-
sels (10). In addition, the ultrastructural
features were reminiscent of lymphatic ves-
sels, which differ from blood vessels in that
they have overlapping endothelial junc-
tions, anchoring filaments in the vessel
wall, and a discontinuous or even partially
absent basement membrane (Fig. 1C) (11).
Antibodies to collagen types IV and XVIII
(12), and laminin gave little or no staining
of the vessels, whereas the basement mem-
brane staining of other vessels was promi-
nent (Fig. 1D) (13). The endothelium was
also characterized by positive staining with
monoclonal antibodies to desmoplakins I
and II, expressed in lymphatic, but not in
vascular endothelial cells (14). Collective-
ly, these findings suggest that the abnormal
vessels are of lymphatic origin.
Abundant VEGF-C mRNA was detect-
ed in the epidermis and hair follicles of the
transgenic mice (Fig. 2, A and B), whereas
mRNAs encoding its receptors VEGFR-3
(15) and VEGFR-2 (16), as well as the
Tie-1 endothelial receptor tyrosine kinase
(17), were expressed in endothelial cells
lining the abnormal vessels (Fig. 2, C and
E) (13). In the skin of littermate control
animals, VEGFR-3 was detected only in the
superficial subpapillary layer of lymphatic
vessels, whereas VEGFR-2 was observed in
all endothelia (Fig. 2, D and F), in agree-
ment with earlier findings (18, 19).
The lymphatic endothelium has a great
capacity to distend in order to adapt to
functional requirements. To ascertain
whether vessel dilation was due to endothe-
lial distension or proliferation, we carried
out in vitro proliferation assays. The
VEGF-C receptor interaction in the trans-
genic mice apparently transduces a mito-
genic signal, because in contrast to litter-
mate controls, the lymphatic endothelium
of the skin from young K14–VEGF-C mice
showed increased DNA synthesis, as dem-
onstrated by bromodeoxyuridine (BrdU) in-
corporation followed by staining with anti-
bodies to BrdU (Fig. 3, A and B).
Angiogenesis is a multistep process that
includes endothelial sprouting, migration,
and proliferation (20). To estimate the con-
tribution of such processes to the transgenic
phenotype, we analyzed the morphology
and function of the lymphatic vessels, using
fluorescent microlymphography (21). A
typical honeycomb-like network with simi-
lar mesh sizes was detected in both control
and transgenic mice (Fig. 3, C and D), but
the diameter of the vessels in transgenic
M. Jeltsch, A. Kaipainen, V. Joukov, K. Alitalo, Molecular/
Cancer Biology Laboratory, Haartman Institute, Post Of-
fice Box 21 (Haartmaninkatu 3), University of Helsinki,
SF-00014 Helsinki, Finland.
X. Meng, M. Lakso, H. Rauvala, Biotechnology Institute,
00014 University of Helsinki, Helsinki, Finland.
M. Swartz, D. Fukumura, R. K. Jain, Department of Ra-
diation Oncology, Massachusetts General Hospital and
Harvard Medical School, Boston, MA 02114, USA.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. E-
mail: Kari.Alitalo@Helsinki.FI
l
m
e
e
d
m
A B
C D
Fig. 1. Analysis of the skin of transgenic mice. (A) Hematoxylin-eosin stained section of the skin of a
2-month-old transgenic mouse. Hyperkeratotic epidermis (e) showed underlying vessel spaces lined
with endothelium (arrows) but devoid of red cells (compare with the dermal vein shown with an
arrowhead). The dermis (d) is atrophic compared with the control littermate skin (B) (45% versus 65% of
the dermal thickness, respectively), and the muscle layer (m) is also reduced. (C) Electron microscopy
shows the endothelial junctions of an abnormal vessel (arrow) (25); l, lumen; m, mesenchyme. In (D), the
basal lamina is stained for type XVIII collagen in veins (arrowheads) but not in the lymphatic endothelium
(arrow) (26). Scale bars for (A) and (B), 250 mm; (C), 100 nm; and (D), 25 mm.
REPORTS
www.sciencemag.org
z
SCIENCE
z
VOL. 276
z
30 MAY 1997 1423

mice was approximately twice that of con-
trols (Table 1). Some dysfunction of the
abnormal vessels was indicated by the fact
that it took longer for the dextran to com-
pletely fill the abnormal vessels. Injection
of fluorescein isothiocyanate (FITC)–dex-
tran into the tail vein, followed by fluores-
cence microscopy of the ear showed that
the blood vascular morphology was unal-
tered and leukocyte rolling and adherence
appeared normal (Fig. 3, E and F) (13).
Thus, the endothelial proliferation induced
by VEGF-C leads to hyperplasia of the su-
perficial lymphatic network, but does not
induce the sprouting of new vessels.
These effects of VEGF-C overexpres-
sion are unexpectedly specific, particularly
as VEGF-C is also capable of binding to
and activating VEGFR-2, which is the
major mitogenic receptor of blood vessel
endothelial cells (16). In culture, high
concentrations of VEGF-C stimulate the
growth and migration of bovine capillary
endothelial cells that express VEGFR-2,
but not significant amounts of VEGFR-3
(3). In addition, VEGF-C induces vascular
permeability in the Miles assay, presum-
ably by its effect on VEGFR-2 (22). In
vivo, the specific effects of VEGF-C on
lymphatic endothelial cells may reflect a
requirement for the formation of VEGFR-
3 3 VEGFR-2 heterodimers for endothe-
lial cell proliferation at physiological con-
centrations of the growth factor. Such pos-
sible heterodimers may help to explain
how three homologous VEGFs exert par-
tially redundant, yet strikingly specific,
biological effects.
In summary, VEGF-C appears to induce
specific lymphatic endothelial proliferation
and hyperplasia of the lymphatic vascula-
B
VEGF-C, sense
D
F
VEGFR-3, control
VEGFR-2, control
A
VEGF-C
C
E
VEGFR-3
VEGFR-2
Fig. 2. In situ hybridization analysis of the skin of
K14–VEGF-C transgenic mice. (A) and (B) show
hybridization of transgenic skin with VEGF-C an-
tisense and sense probes. (C) and (D) and (E) and
(F) show in situ hybridization for VEGFR-3 and
VEGFR-2, respectively, in the skin of transgenic
and littermate control animals (27 ). Arrows in (A)
indicate basal keratinocytes; in (C) through (E),
they point out the lymphatic endothelium; and in
(F), arrows show endothelial cells. Scale bar, 20
mm.
C D
lv lv
ø
v
h
E F
d
d
A B
Fig. 3. Immunohistochemical analysis of endothelial proliferation and intravital fluorescence microlym-
phography and microvessel angiography. (A) and (B) show staining of cells in the S phase of the cell
cycle, through BrdU incorporation into DNA and its immunohistochemical detection (28). In 2-week-old
transgenic mice, the nuclear staining was observed in many endothelial cells of the lymphatic vessels (lv)
as well as in keratinocytes [red arrows in (A)]. In nontransgenic littermates, mainly nuclei of keratinocytes
of epidermis and some dermal cells are stained [red arrows in (B)]; unstained nuclei were observed in
both cases (green arrows). Ø marks artefactual detachment of the epidermis during sample preparation.
(C) and (D) illustrate the lymphatic vessels of transgenic and control skin, respectively, through fluores-
cence microscopy after intradermal injection of FITC-dextran (21, 29). The measured parameters are
diameter (d) [in (C) and (D)] and horizontal (h) and vertical (v) mesh sizes [in (D) only]. Blood vessels of the
ear after injection of FITC-dextran into the tail vein of transgenic and control mice are shown in (E) and
(F), respectively (24). Scale bars for (A) and (B), 5 mm; (C) and (D), 250 mm; (E) and (F), 1 mm.
Table 1. Structural parameters of lymphatic and blood vessel networks for transgenic and control
mice. Mesh size describes vessel density. Diameters and mesh sizes are in micrometers (mean 6
SD).
Parameter Transgenic Control P value*
Lymphatic vessels†
Diameter 142.3 6 26.2 68.2 6 21.7 0.0143
Horizontal mesh size 1003 6 87.1 960.8 6 93.1 0.2207
Vertical mesh size 507.3 6 58.9 488.8 6 59.9 0.5403
Blood vessels‡
Median diameter 8.3 6 0.6 7.6 6 1.1 0.1213
Vessel density (cm/cm2) 199.2 6 6.6 216.4 6 20.0 0.3017
*Mann-Whitney test. †Number of animals for transgenic and control groups were 4 and 5, respectively. ‡Number
of animals for transgenic and control groups were 3 and 6, respectively.
SCIENCE
z
VOL. 276
z
30 MAY 1997
z
www.sciencemag.org1424