Ligand-induced Vascular Endothelial Growth Factor Receptor-3
(VEGFR-3) Heterodimerization with VEGFR-2 in
Primary Lymphatic Endothelial Cells Regulates Tyrosine
Phosphorylation Sites*
Received for publication, April 30, 2003
Published, JBC Papers in Press, July 24, 2003, DOI 10.1074/jbc.M304499200
Johan Dixelius‡, Taija Ma¨kinen§¶, Maria Wirzenius§, Marika J. Karkkainen§,
Christer Wernstedt
, Kari Alitalo§, and Lena Claesson-Welsh‡**
From the ‡Department of Genetics and Pathology, Uppsala University, Rudbeck Laboratory, Dag Hammarskjo¨lds va¨g 20,
S-751 85 Uppsala, Sweden, the §Molecular/Cancer Biology Laboratory, Biomedicum Helsinki, Post Office Box 63, 00014
University of Helsinki, Finland, and the
Ludwig Institute for Cancer Research, Uppsala Biomedical Center, Box 595,
S-751 24 Uppsala, Sweden
Vascular endothelial growth factors (VEGFs) regulate
the development and growth of the blood and lymphatic
vascular systems. Of the three VEGF receptors (VEGFR),
VEGFR-1 and -2 are expressed on blood vessels; VEGFR-2 is
found also on lymphatic vessels. VEGFR-3 is expressed
mainly on lymphatic vessels but it is also up-regulated in
tumor angiogenesis. Although VEGFR-3 is essential for
proper lymphatic development, its signal transduction
mechanisms are still incompletely understood. Trans-phos-
phorylation of activated, dimerized receptor tyrosine ki-
nases is known to be critical for the regulation of kinase
activity and for receptor interaction with signal transduc-
tion molecules. In this study, we have identified five tyrosyl
phosphorylation sites in the VEGFR-3 carboxyl-terminal
tail. These sites were used both in VEGFR-3 overexpressed
in 293 cells and when the endogenous VEGFR-3 was acti-
vated in lymphatic endothelial cells. Interestingly, VEGF-C
stimulation of lymphatic endothelial cells also induced the
formation of VEGFR-3/VEGFR-2 heterodimers, in which
VEGFR-3 was phosphorylated only at three of the five sites
while the two most carboxyl-terminal tyrosine residues ap-
peared not to be accessible for the VEGFR-2 kinase. Our
data suggest that the carboxyl-terminal tail of VEGFR-3
provides important regulatory tyrosine phosphorylation
sites with potential signal transduction capacity and that
these sites are differentially used in ligand-induced homo-
and heterodimeric receptor complexes.
The receptor tyrosine kinase vascular endothelial growth
factor receptor 3 (VEGFR-3,1 previously denoted fms-like tyro-
sine kinase-4 or Flt-4) is essential for the development of the
blood and lymphatic vasculature. Inactivation of the VEGFR-3
gene in mouse embryos leads to a disturbed vascular develop-
ment resulting in an irregular vessel pattern and a reduced
cross-sectional area of large vessels. The embryos die at em-
bryonic day 9.5 because of fluid accumulation in the pericardial
cavity and cardiovascular failure (1).
In adult tissues, VEGFR-3 is expressed primarily on lym-
phatic endothelial cells (2) and appears to exert its major func-
tions within this system. Thus, inactivating missense point
mutations in one VEGFR3 allele leads to chronic lymphedema
(3). Further, overexpression of a soluble VEGFR-3 in mice leads
to regression of lymph vessels and features characteristic of
lymphedema, without any apparent effects on the blood vascu-
lature (4). In several tumor models, overexpression of the
VEGFR-3 ligand VEGF-C increases lymphangiogenesis and
promotes spread of metastases (5–7). The same effect is
achieved by overexpression of VEGF-D, another VEGFR-3
ligand (8).
VEGFR-3 is to some extent expressed also on quiescent vas-
cular endothelial cells, primarily in fenestrated capillaries (2,
9, 10). Very low levels can be occasionally detected in the blood
vascular endothelium of wound granulation tissue and in ves-
sels stimulated with VEGFs (11, 12). Further, the endothelium
of angiogenic blood vessels of several tumors express VEGFR-3
(11, 13, 14). These results suggest that VEGFR-3 could be
involved in aspects of angiogenesis in adults.
The VEGFR-3 is similar in overall structure to the VEGFR-1
and VEGFR-2 (15); the extracellular, ligand-binding domain is
composed of seven immunoglobulin-like folds, and the intracel-
lular domain is characterized by an interrupted tyrosine kinase
domain. In contrast to the other VEGF receptors, the VEGFR-3
extracellular domain is cleaved within the fifth immunoglobu-
lin-homology domain; the resulting two polypeptides are held
together by a disulfide bridge (16). In humans, but not in mice,
a retroviral insertion between the last two exons of VEGFR-3
(17) results in two splice variants of the VEGFR-3, the shorter
form of which lacks 65 amino acids in the cytoplasmic tail
(18, 19).
Activation of the VEGFR-3 tyrosine kinase appears to follow
the consensus scheme for receptor tyrosine kinases. Ligand
binding results in receptor dimerization and sequential activa-
tion of the intrinsic kinase activity. Trans-phosphorylation be-
tween the partners in the dimer regulates kinase activity and
creates docking sites for signaling molecules with characteris-
tic domains such as Src homology (SH)-2 or phosphotyrosine
binding domains. The specificity of binding is determined by
* This study was supported by grants from the Swedish Cancer
Foundation (3820-B01-06XAC), the Novo Nordisk Foundation, the
Pharmacia Corporation, the Swedish Science Council (285-1998-697),
Finnish Cancer Organizations, and by the Finnish Cultural Founda-
tion. The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact.
¶ Present address: Dept. of Molecular Neurobiology, Max-Planck In-
stitute of Neurobiology, Am Klopferspitz 18A, D-82152 Martinsried,
Germany.
** To whom correspondence should be addressed. Fax: 46-18-55-89-
31; E-mail: Lena.Welsh@genpat.uu.se.
1 The abbreviations used are: VEGFR, vascular endothelial growth
factor receptor; PAE, porcine aorta endothelial; LEC, lymphatic endo-
thelial cell; R3-KD, VEGFR-3 kinase dead; PDGF, platelet-derived
growth factor.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 278, No. 42, Issue of October 17, pp. 40973–40979, 2003
© 2003 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org 40973

the sequence adjacent to the phosphotyrosine residue. The
activated VEGFR-3 associates with adaptor proteins Shc and
Grb2 via tyrosine 1337 (16, 20). Moreover, VEGFR-3 activation
leads to protein kinase C-dependent activation of extracellular
signal-related kinases (Erk)-1 and -2, implicated in cell prolif-
eration. Furthermore, VEGFR-3 mediates the activation of pro-
tein kinase B/Akt (21), implicated in cell survival. In accord-
ance, VEGFR-3 transduces signals resulting in proliferation,
migration, and survival of lymphatic endothelial cells (21). In
this report, we have examined the potential phosphorylation of
tyrosine residues in the long form of the activated VEGFR-3
and provide evidence for phosphorylation in five positions in
the cytoplasmic tail. In addition, we show complex formation
between VEGFR-3 and the related VEGFR-2 in primary lym-
phatic endothelial cells. In the heterodimeric configuration,
VEGFR-2 failed to phosphorylate VEGFR-3 on two of the car-
boxyl-terminal sites, Tyr-1337 and Tyr-1663. This has implica-
tions for the VEGFR-3 signal transduction properties.
EXPERIMENTAL PROCEDURES
Growth Factors and Antibodies—The growth factors used were epi-
dermal growth factor (EGF, no. 100–15; Peprotech, Rockyhill, NJ),
human VEGF (no. 100–20; Peprotech, Rockyhill, NJ), and human
VEGF-C (Thr-103-Leu-215; Ref. 22). The antibodies used were mouse
anti-VEGFR-3 (clones 9D9F9, 2E11D11; Refs. 21, 23); rabbit anti-
VEGFR-2 (RS-2; Ref. 24); anti-phosphotyrosine 4G10 05–321, Upstate
Biotechology, Lake Placid, NY; rabbit anti-V5 A190–120A, Bethyl Lab-
oratories, Montgomery, TX; and anti-actin, sc-1615, Santa Cruz Bio-
technology, Santa Cruz, CA. Rabbit anti-human podoplanin antibodies
were kindly provided by Dontscho Kerjaschki, Vienna, Austria.
Cell Lines and Cell Culture—Porcine aorta endothelial (PAE) cells
stably overexpressing VEGFR-3 or VEGFR-2 were maintained in F12/
10% fetal calf serum. Human 293T cells were used for transient expres-
sion of VEGFR-3 and were maintained in Dulbecco’s modified Eagle’s
medium/10% fetal calf serum. Primary lymphatic endothelial cells
(LECs) were separated from human dermal microvascular endothelial
cells as previously described (21) using antibodies against podoplanin.
The cells were cultured on gelatin-coated plastic in endothelial basal
medium (EBM, CC-3121; Clonetics, Walkersville, MD) supplemented
with 5% fetal calf serum, 30
g/ml endothelial growth culture supple-
ment (ECGS, E-7060; Sigma), 10 ng/ml EGF, and 10 ng/ml VEGF-C.
Generation of Mutated VEGFR-3—VEGFR-3 tyrosine mutants were
generated by the GeneEditor in vitro site-directed mutagenesis kit
(Promega) using oligonucleotides in which one nucleotide change in the
tyrosine-encoding sequence was introduced, resulting in Tyr
Phe
amino acid change in the protein sequence.
VEGFR-3 kinase dead (R3-KD; R1041P and K879G) mutants were
generated as above using oligonucleotides that introduced desired nu-
cleotide changes in the VEGFR-3 kinase domain-encoding sequence.
The R1041P represents a mutation found in lymphedema (25), whereas
in the K879G mutant the ATP-binding Lys has been changed into Gly.
Both mutant proteins were found to be kinase-inactive when expressed
in 293T cells (25) (data not shown).
Transient Transfections—Vectors (pcDNA3.1/Zeo; Invitrogen) encod-
ing wild type and mutant VEGFR-3 were transfected into human 293T
cells using the calcium phosphate method. In brief, 6  106 cells in
10-cm Petri dishes were treated with 25
M chloroquine for 1.5 h.
Vector cDNA (4–10
g) in 250 mM CaCl2 was mixed with 2 concen-
trated Hank’s balanced salt solution, incubated for 20 min, and then
added to the cells. 5–6 h later, cells were treated for 2 min with
culture medium containing 10% glycerol. The cells were harvested 48 h
post-transfection.
Immunocomplex Kinase Assay and SDS-PAGE—The cells were
starved overnight in serum-free medium supplemented with 0.1% bo-
vine serum albumin and treated for 8 min with or without VEGF or
VEGF-C using 50 ng/ml washed in Tris-buffered saline/100
M Na3VO4
on ice. The cells were lysed in ice-cold Nonidet P-40 lysis buffer (1%
Nonidet P40, 20 mM Hepes, pH 7.5, 150 mM NaCl, 10% glycerol, 2.5 mM
EDTA, 10 Kallekrein inhibitory units aprotinin/ml, 1 mM phenylmeth-
ylsulfonyl fluoride, and 100
m Na3VO4). An aliquot of the cell lysate
was saved for control blotting. Lysates were clarified by centrifugation
and incubated for 2 h on ice with in-house anti-VEGFR-3 antibodies.
The mouse and rabbit antibodies were precipitated with protein G-
Sepharose (17–0618-01; Amersham Biosciences) and protein A-Sepha-
rose (Immunosorb A; Medicago AB, Uppsala, Sweden), respectively.
The precipitate was washed three times in lysis buffer and twice in
kinase buffer (20 mM Hepes, 10 mM MgCl2, 2 mM MnCl2, 0.05% Triton
X-100). The precipitate was incubated for 10 min in kinase buffer
containing 20
Ci of [-32P]ATP (Amersham Biosciences) at 37 °C and
heated in sample buffer (8% SDS, 0.4 M Tris-HCl, pH 8.0, 1 M sucrose,
10 mM EDTA, 0.02% bromphenol blue, 4% -mercaptoethanol). The
samples were separated by SDS-PAGE, using 7% polyacrylamide gel,
transferred to a nitrocellulose membrane, and detected by a BioImager
(BioImager, BAS-1800II; Fujifilm, Tokyo, Japan) screen that was sub-
sequently scanned using the BioImager.
Immunoblotting—Samples were prepared essentially as described
above but without the kinase reaction, separated by SDS-polyacryl-
amide gel electrophoresis (SDS-PAGE), and transferred to a nitrocellu-
lose membrane. The membranes were incubated with indicated pri-
mary antibodies and subsequently with horseradish peroxidase-
conjugated secondary antibodies. Immunoreactive sites were visualized
using an enhanced chemiluminescence detection system (Amersham
Biosciences).
Phosphopeptide Mapping—The detailed procedure has been de-
scribed previously (26, 27). Briefly, the phosphorylated bands on the
membrane from the in vitro complex assay were localized using the
BioImager scan, cut out and treated for 30 min at 37 °C in 0.5%
polyvinylpyrrolidone in 100 mM acetic acid, washed four times in H2O,
and digested by 1
g of trypsin (V542A; Promega, Madison, WI) in 200

l of 50 mM NH4CO3 at 37 °C overnight. The supernatant was lyophi-
lized in a centrifugal vacuum concentrator, dissolved in 50
l of perfor-
mic acid (formic acid, 30% H2O2, 9:1), incubated for 1 h at room tem-
perature, diluted with 500
l of H2O, and frozen to 135 °C. The frozen
samples were lyophilized again and dissolved in 50 mM NH4CO3 sup-
plemented with 1
g of trypsin and incubated overnight at 37 °C. 140
l
of pH 1.9 buffer (2.2% formic acid and 7.8% acetic acid in H20) was
added, and any particles were precipitated by centrifugation. 180
l of
the supernatant was lyophilized and dissolved in 7
l of pH 1.9 buffer.
The samples were centrifuged, and 6
l of each sample was gently dried
onto a cellulose-covered thin-layer chromatography glass plate
(1.05716; Merck, Darmstadt, Germany) as a spot of5 mm in diameter,
localized 5 cm from the left side and 3 cm from the bottom of the plate.
The peptides were separated by electrophoresis (x-axis) using pH 1.9
buffer, and the plate was dried. Separation in the second dimension
(y-axis) was performed by ascending chromatography using isobutyric
acid buffer (62.5% isobutyric acid, 1.9% n-butanol, 4.8% pyridine, 2.9%
acetic acid in H2O) overnight. The plate was dried, and the peptides
with incorporated 32P were detected by the BioImager. Using the re-
sulting phosphopeptide map for localization, cellulose-containing pep-
tide spots of interest were scraped off the chromatography plate. The
peptides were extracted by pH 1.9 buffer, lyophilized, and used for
Edman degradation or phosphoamino acid analysis.
Edman Degradation and Radio Amino Acid Sequencing—For Ed-
man degradation, phosphopeptides were coupled to sequelon-amino
acid membranes (Millipore, Sundbyberg, Sweden) and sequenced using
a gas phase sequencer (Applied Biosystems, Foster City, CA). The
fractions were spotted onto thin-layer chromatography plates, detected
by BioImager, and analyzed by the BioImager software.
Phosphoamino Acid Analysis—The samples extracted from the thin-
layer plates were hydrolyzed in 6 M HCl at 110 °C and lyophilized,
dissolved in H2O, and lyophilized again. The samples were dissolved in
7
l of pH 1.9 buffer supplemented with non-radioactive phosphoryl-
ated serine, threonine, and tyrosine as markers and spotted onto a
thin-layer chromatography plate. Electrophoresis using first the pH 1.9
buffer (x-axis) and subsequently the pH 3.5 buffer (acetic acid 5% and
pyridine 0.5% in H2O; y-axis) provided two-dimensional separation of
the hydrolyzed amino acid residues. The dried plates were sprayed with
ninhydrin for detection of the marker phosphoamino acid residues. The
plate was dried, and the incorporated 32P was visualized using the
BioImager. By overlaying the resulting image with the ninhydrin pat-
tern of the marker amino acids, the identity of the amino acid residues
with incorporated 32P was determined.
RESULTS
Phosphorylation of VEGFR-3—The intracellular domain of
the long form of VEGFR-3 contains 16 tyrosine residues. Of
these, six tyrosine residues are located in the carboxyl-terminal
tail (denoted Tyr-1230, -1231, -1265, -1333, -1337, and -1363;
Fig. 1). Trypsin digestion of the VEGFR-3 intracellular domain
results in the generation of up to 17 tyrosine-containing pep-
tides, of which 4 are derived from the carboxyl-terminal tail
VEGFR-3 Tyrosine Phosphorylation Sites40974

(Table I). We wished to determine which of the carboxyl-termi-
nal tyrosine residues could serve as potential phosphorylation
sites. Stimulation of PAE cells expressing VEGFR-3 with
VEGF-C resulted in strong induction of VEGFR-3 phosphoryl-
ation, as estimated in an in vitro immunocomplex kinase assay
(Fig. 2A). Trypsin-digested peptides of the receptor were sepa-
rated by electrophoresis according to the charge/mass ratio
(x-axis) and by liquid chromatography according to hydropho-
bicity (y-axis), creating a phosphopeptide map. In this analysis,
a corresponding increase in phosphorylation was detected in an
analysis of trypsin-generated peptides from the ligand-stimu-
lated VEGFR-3 (Fig. 2B).
Position of Phosphorylated Residues—Peptides giving repro-
ducible spots in the phosphopeptide maps were extracted, hy-
drolyzed in hydrochloric acid, and separated by two-dimen-
sional electrophoresis on thin-layer plates. Unlabeled
phosphoamino acid residues served as references (indicated by
circles in the insets in Fig. 2C). Peptides displaying tyrosine
phosphorylation (circled in Fig. 2B) were subjected to Edman
degradation. Chemical identification of amino acid residues
was not feasible because of the minute amounts of protein in
the assay. Instead, the material in each cycle was spotted
individually on thin-layer chromatography plates, and the 32P
content in each fraction was quantified after exposure and
detection using a BioImager. The positions of the radioactive
peaks were the basis for tentative identification of a tryptic
peptide containing a tyrosine residue in the corresponding po-
sition. Peptides from some distinct spots displayed identical
radio sequences (spots denoted c in Fig. 2). This may be because
of serine phosphorylation of the corresponding peptide, which
contains five serine residues. Serine phosphorylation was in-
deed detected in the left-most spot (data not shown). The pre-
dicted change in the phosphopeptide map position because of
such a modification is in compliance with the observed posi-
tions. An alternative explanation for multiple spots is partial
oxidation of cysteine residues with consequences for the charge
and thereby the migration position of the peptide.
Analyses of VEGFR-3 Tyrosine to Phenylalanine Mutants—
The above results indicated that all tyrosine residues in the
carboxyl-terminal tail, with the possible exception of Tyr-1333,
were phosphorylated. To further exploit these findings, we
created six receptor variants, point-mutated in the carboxyl-
terminal tail, each with a single amino acid exchange from
tyrosine to phenylalanine. These receptors were denoted
Y1230F-R3, Y1231F-R3, Y1265F-R3, Y1333F-R3, Y1337F-R3,
and Y1363F-R3. Phosphopeptide mapping of the mutant recep-
tors transiently expressed in 293 cells confirmed the prelimi-
nary identification of the phosphopeptides (Fig. 3). In the phos-
phopeptide map of Y1265F-R3 one spot was lost, whereas the
map of Y1337F-R3 showed loss of three spots (Fig. 3). The
corresponding spots in the Y1333F-R3 map were shifted along
the y axis, indicating increased hydrophobicity, in agreement
with the expected change as a result of the exchange of tyrosine
for phenylalanine. The Tyr-1337 residue was still phosphoryl-
ated in this peptide. The peptides containing Tyr-1230/Tyr-
1231 and Tyr-1363 were not fully separated, but analysis of
this region in the phosphopeptide map (boxed in Fig. 3A) by
scanning densitometry confirmed that loss of tyrosine phospho-
rylation as a consequence of the different mutations resulted in
the expected phosphopeptide pattern (Fig. 3H).
VEGFR-2 and VEGFR-3 Heterodimerization in Primary
Cells—We wished to verify that the phosphorylation pattern of
VEGFR-3 overexpressed in PAE or 293 cells mimicked that of
the endogenously expressed VEGFR-3 in LECs. Fig. 4 shows
TABLE I
Predicted tyrosine-containing peptides
Peptide number Amino acid sequence Tyrosine positionsa Domain
1 TGY812LSIIMDPGEVPLEEQCEY830LSY833DASQWEFPR 3, 21, 24 Juxtamembrane
2 VLGY853GAFGK 4 Kinase domain 1
3 Y932GNLSNFLR 1 Kinase domain 1
4 ASPDQEAEDLWLSPLTMEDLVCY1017SFQVAR 23 Kinase domain 2b
5ac ICDFGLARDIY1063KDPDY1068VRK 11, 16 Kinase domain 2
5bc ICDFGLARDIY1063KDPDY1068VR 11, 16 Kinase domain 2
5cc ICDFGLARDIY1063K 11 Kinase domain 2
5dc DIY1063K 3 Kinase domain 2
5ec DIY1063KDPDY1068VRK 3, 8 Kinase domain 2
5f c DIY1063KDPDY1068VR 3, 8 Kinase domain 2
5g c DPDY1068VRK 4 Kinase domain 2
5hc DPDY1068VR 4 Kinase domain 2
6 VY1091TTQSDVWSFGVLLWEIFSLGASPY1115PGVQINEEFCQR 2,26 Kinase domain 2
7 Y1230Y1231NWVSFPGCLAR 1,2 Carboxyl-terminal tail
8 TFEEFPMTPTTY1265K 12 Carboxyl-terminal tail
9 GGQVFY1333NSEY1337GELSEPSEEDHCSPSAR 6, 10 Carboxyl-terminal tail
10 VTFFTDNSY1363 9 Carboxyl-terminal tail
a In tryptic peptide.
b The first 15 amino acids from kinase insert.
c Possible alternative peptides of a single amino acid stretch.
FIG. 1. Schematic representation of VEGFR-3. Positions of tyro-
sine residues are indicated by the dots in the intracellular domain and
by numbers in the carboxyl-terminal tail.
VEGFR-3 Tyrosine Phosphorylation Sites 40975

that the phosphopeptide map of LEC-derived VEGFR-3 was
essentially indistinguishable from that of the overexpressed
recombinant receptor, confirming the relevance of our ap-
proach. Minor differences in migration positions of a collection
of spots in the right margin were anticipated, because there
had been some variations in this region of the TLC plate be-
tween repeated maps of VEGFR-3 overexpressed in the 293
cells.
The immunocomplex kinase assay of VEGFR-3 immunopre-
cipitated from VEGF-C-stimulated LECs demonstrated that a
phosphorylated protein of 220 kDa was co-immunoprecipitated
by the VEGFR-3-specific antibodies. We hypothesized that this
component could correspond to VEGFR-2. This was confirmed
by phosphopeptide mapping (data not shown). To investigate
under which conditions the association between VEGFR-2 and
VEGFR-3 occurred, LECs were treated with VEGF, specific for
VEGFR-2, or VEGF-C, a ligand for both receptors. Immuno-
precipitation with antibodies specific for the respective recep-
tors, followed by immunocomplex kinase assay, demonstrated
that the association was evident only after stimulation with
VEGF-C (Fig. 5A). The specificity of the antibodies was con-
firmed by immunoprecipitation and blotting of cell lysates de-
rived from PAE cells overexpressing either VEGFR-2 or
VEGFR-3 (Fig. 5B). The different levels of kinase activity and
the different properties of the antisera used precluded a deter-
mination of the relative stoichiometry of heterodimers versus
homodimers in the VEGF-C-treated LECs.
Distinct Pattern of Phosphorylation of VEGFR-3 in the Het-
erodimeric Configuration—To examine whether VEGFR-3 was
phosphorylated similarly in the homodimeric and the het-
erodimeric configuration, a kinase-dead mutant VEGFR-3
(R1041P; here denoted R3-KD) was expressed alone or in com-
bination with VEGFR-2 or VEGFR-3 in 293 cells. The V5-
tagged R3-KD was specifically recognized by anti-V5 antibod-
FIG. 2. Phosphopeptide mapping, radioactive amino acid sequencing, and phosphoamino acid analyses of VEGFR-3. PAE cells
overexpressing VEGFR-3 were treated with or without VEGF-C for 8 min. A, immunoprecipitated VEGFR-3 was 32P-labeled by immunocomplex
kinase assay, subjected to SDS-PAGE, and transferred to nitrocellulose membrane. B, trypsin treatment of the VEGFR-3 band on the membrane
was followed by separation in two dimensions on a cellulose thin-layer plate. Circled peptides (a–d) were present in the VEGF-C-treated cells only.
One of the peptides migrated in three positions (three spots labeled c). C, radioactive amino acid sequences of peptides a–d in panel B. Numbers
of consecutive cycles in the Edman degradation and tentative alignment of tyrosine residues in tryptic VEGFR-3 peptides with the radioactive
peaks are indicated. The corresponding phosphoamino acid analyses are inserted in each panel, a–d. Positions of reference phospho-serine (S),
phospho-threonine (T), and phospho-tyrosine (Y) are circled.
VEGFR-3 Tyrosine Phosphorylation Sites40976

ies, ensuring that the analysis was focused on R3-KD
dimerized with kinase-active VEGFR-2 or VEGFR-3. In this
setup, phosphorylation of R3-KD was dependent on the co-
expressed kinase-active receptors. As shown in Fig. 6, wild type
VEGFR-3-mediated phosphorylation of R3-KD resulted in a
phosphopeptide map very similar to the previous VEGFR-3
maps (see Fig. 3). In contrast, the map of R3-KD phosphoryl-
ated by VEGFR-2 lacked spots corresponding to peptides con-
taining tyrosine residues Tyr-1337 and -1363. Repeating the
analysis using another kinase-inactive VEGFR-3 (K879G) gave
identical results (data not shown).
DISCUSSION
Tyrosine phosphorylation sites in receptor tyrosine kinases
regulate both kinase activity and interaction with signal trans-
duction molecules. Thus, identification of these sites is of fun-
damental importance in understanding the signaling of a spe-
cific receptor. We show that five (Tyr-1230, -1231, -1265, -1337,
and -1363) of the six tyrosine residues in the carboxyl-terminal
tail of VEGFR-3 are potential phosphorylation sites. The role of
most of these sites in signal transduction downstream of
VEGFR-3 remains to be determined. Tyr-1337 is required for
association of the Shc-Grb2 complex to VEGFR-3 (20), and this
FIG. 3. Phosphopeptide maps of wild type and mutated VEGFR-3. Phospo-VEGFR-3 immunoprecipitates were subjected to phosphopep-
tide mapping. A, the wild type phosphopeptide map. B–G, phosphopeptide maps of the point-mutated receptors. Peptides lost in the mutant maps
as compared with the wild type map are indicated by arrows. Peptides with shifted positions as a consequence of the mutation are indicated by
arrowheads. H, phosphopeptides that appeared to be lost in Y1230F-R3, Y1231F-R3, and Y1363F-R3 were not fully separated from the other
peptides in the region boxed in panel A. Densitometric scanning of the corresponding regions in the phosphopeptide maps of these receptors was
carried out using the BioImager software. The wild type receptor and Y1265F-R3 were used as references for the densitometric curves. Arrows
indicate major changes for each mutant VEGFR-3.
VEGFR-3 Tyrosine Phosphorylation Sites 40977

interaction has been linked to the transforming capacity of the
receptor overexpressed in fibroblasts. In accordance, the short
form of the VEGFR-3 that lacks the tyrosines Tyr-1333, -1337,
and -1363 is unable to mediate fibroblast transformation.
Our preliminary results suggest that the tyrosine residues
1063 and 1068 in the second kinase domain become phospho-
rylated upon receptor activation (data not shown). The posi-
tions of these sites correspond to those previously implicated in
positive regulation of tyrosine kinase activity, e.g. Tyr-1054
and -1059 in VEGFR-2 (28). We therefore suggest that Tyr-
1063 and -1068 in VEGFR-3 serve a positive regulatory role in
the activation of the VEGFR-3 kinase. Of the remaining tyro-
sine residues in the intracellular domain of VEGFR-3, five are
located in the first and second parts of the kinase domain, and
three are located in the juxtamembrane domain (Tyr-812, -830
and -833). Tyrosine residues at positions conserved relative to
Tyr-812 are found both in VEGFR-1 (Tyr-794) and VEGFR-2
(Tyr-801) (29). All three VEGFR-3 juxtamembrane tyrosine
residues are contained within the same tryptic peptide. Edman
degradation of this peptide has not allowed an unambiguous
conclusion on the phosphorylation of these tyrosine residues.
The kinase insert domain plays an important role in signal
transduction by the PDGF receptors. This sequence is of vary-
ing length in different receptor tyrosine kinases, and it is
usually not conserved between otherwise related receptors,
such as the PDGF - and -receptors (30), which has prompted
the suggestion that the kinase insert is important in receptor
type-specific signaling. It is therefore interesting that neither
the VEGFR-1 nor the VEGFR-3 kinase insert contains any
tyrosine residues. In contrast, the VEGFR-2 kinase insert con-
tains three tyrosine residues, of which at least one is a phos-
phorylation site (28).
Our analysis of VEGFR-3 tyrosine phosphorylation in pri-
mary lymphatic endothelial cells allowed the following conclu-
sions. 1) The phosphorylation pattern of VEGFR-3 was faith-
fully reproduced between the receptor overexpressing cells and
primary LECs. 2) In these primary cells, VEGF-C, but not
VEGF, treatment induced formation of VEGFR-2 and
VEGFR-3 heterodimers. 3) VEGFR-3 phosphorylation-site us-
age was altered in the heterodimeric configuration. Thus, the
two most carboxyl-terminal tyrosine residues in VEGFR-3 are
substrates only for the VEGFR-3. VEGFR-3 in adult blood
vessels and at least some lymphatic endothelia occurs in areas
with VEGFR-2 expression (2, 31), indicating that VEGFR-2/
VEGFR-3 heterodimers may form in vivo. Interestingly, recent
data suggest that VEGFR-3 modulates sensitivity to VEGFR-2
signaling to promote vascular integrity in blood vascular endo-
FIG. 4. Analysis of VEGFR-3 phosphorylation in primary LECs. A, LECs with or without VEGF-C treatment were subjected to in vitro
complex kinase assay. The expected three VEGFR-3 bands were detected as indicated (Fig. 2A). The arrow indicates a band of 220 kDa, which was
identified as VEGFR-2. B, VEGFR-3 bands from the VEGF-C-stimulated LECs were cut out and subjected to phosphopeptide mapping. The left
(green) and the right (red) panels indicate phosphopeptide maps created from VEGFR-3 transiently expressed in 293T cells and primary LECs,
respectively. These two phosphopeptide maps are overlaid in the middle panel.
FIG. 5. VEGFR-2 co-precipitates with VEGFR-3 after stimula-
tion with VEGF-C. A, primary LECs were treated with VEGF (binds
VEGFR-2 but not VEGFR-3) or VEGF-C (binds to VEGFR-2 and
VEGFR-3). The receptors were immunoprecipitated with receptor-spe-
cific antibodies and subjected to immunocomplex kinase assay (upper
panel). The migration rates of VEGFR-2 and the three bands corre-
sponding to VEGFR-3 are indicated. Lower panel, control blot of cell
lysates showing equal amounts of actin in the different samples. B,
demonstration of the specificity of the antibodies used for immunoprecipi-
tation in panel A, using PAE cells overexpressing VEGFR-2 or VEGFR-3,
stimulated with VEGF or VEGF-C, respectively. Superscript 1 indicates
the mixture of the two anti-VEGFR-3 antibodies 9D9F9, 2E11D11,
and Superscript 2 indicates the use of the 9D9F9 antibody alone.
VEGFR-3 Tyrosine Phosphorylation Sites40978

thelial cells co-expressing the two receptors (32, 33). It is pos-
sible that such cross-talk between the receptors is dependent
on heterodimerization.
Growth factors of the VEGF and PDGF families are dimeric
proteins in which each monomer contributes one receptor bind-
ing site. The different PDGF variants induce homo- or het-
erodimers of the PDGF receptors in a manner dictated by the
receptor-specificity of the monomers. Similarly, treatment with
VEGF, which is a common ligand for VEGFR-1 and VEGFR-2,
induces receptor heterodimerization and functional signaling
units (34). PDGF - and -receptor heterodimers have been
shown to have signal transduction properties distinct from the
respective - and - homodimeric forms, because of receptor
tyrosine phosphorylation specific for the heterodimeric recep-
tors (35). The differences in the phosphorylation-site pattern
between homo- and heterodimeric VEGFR-3 suggest that the
signal transduction properties and biological function are dis-
tinct for the heterodimerized VEGFR-3. In particular, Shc and
Grb2, which are known to bind to Tyr-1337 (16), are most likely
not substrates for heterodimeric VEGFR-3. VEGF-C is pro-
duced as a 60-kDa protein with affinity for VEGFR-3 but poor
affinity for VEGFR-2 (22). A stepwise proteolysis of VEGF-C
results in a 20-kDa fragment with high affinity for both recep-
tors. Thus, VEGF-C proteolysis provides a mechanism whereby
VEGFR-2/VEGFR-3 heterodimerization and, in turn, VEGFR-3
signaling, could be modulated.
REFERENCES
1. Dumont, D. J., Jussila, L., Taipale, J., Lymboussaki, A., Mustonen, T.,
Pajusola, K., Breitman, M., and Alitalo, K. (1998) Science 282, 946–949
2. Witmer, A. N., Dai, J., Weich, H. A., Vrensen, G. F., and Schlingemann, R. O.
(2002) J. Histochem. Cytochem. 50, 767–777
3. Karkkainen, M. J., and Alitalo, K. (2002) Semin. Cell Dev. Biol. 13, 9–18
4. Makinen, T., Jussila, L., Veikkola, T., Karpanen, T., Kettunen, M. I.,
Pulkkanen, K. J., Kauppinen, R., Jackson, D. G., Kubo, H., Nishikawa, S.,
Yla-Herttuala, S., and Alitalo, K. (2001) Nat. Med. 7, 199–205
5. Mattila, M. M., Ruohola, J. K., Karpanen, T., Jackson, D. G., Alitalo, K., and
Harkonen, P. L. (2002) Int. J. Cancer 98, 946–951
6. Karpanen, T., Egeblad, M., Karkkainen, M. J., Kubo, H., Yla-Herttuala, S.,
Jaattela, M., and Alitalo, K. (2001) Cancer Res. 61, 1786–1790
7. Skobe, M., Hawighorst, T., Jackson, D. G., Prevo, R., Janes, L., Velasco, P.,
Riccardi, L., Alitalo, K., Claffey, K., and Detmar, M. (2001) Nat. Med. 7,
192–198
8. Stacker, S. A., Caesar, C., Baldwin, M. E., Thornton, G. E., Williams, R. A.,
Prevo, R., Jackson, D. G., Nishikawa, S., Kubo, H., and Achen, M. G. (2001)
Nat. Med. 7, 186–191
9. Lymboussaki, A., Partanen, T. A., Olofsson, B., Thomas-Crusells, J., Fletcher,
C. D., de Waal, R. M., Kaipainen, A., and Alitalo, K. (1998) Am. J. Pathol.
153, 395–403
10. Partanen, T. A., Arola, J., Saaristo, A., Jussila, L., Ora, A., Miettinen, M.,
Stacker, S. A., Achen, M. G., and Alitalo, K. (2000) FASEB J. 14, 2087–2096
11. Witmer, A. N., van Blijswijk, B. C., Dai, J., Hofman, P., Partanen, T. A.,
Vrensen, G. F., and Schlingemann, R. O. (2001) J. Pathol. 195, 490–497
12. Enholm, B., Karpanen, T., Jeltsch, M., Kubo, H., Stenback, F., Prevo, R.,
Jackson, D. G., Yla-Herttuala, S., and Alitalo, K. (2001) Circ. Res. 88,
623–629
13. Partanen, T. A., Alitalo, K., and Miettinen, M. (1999) Cancer 86, 2406–2412
14. Valtola, R., Salven, P., Heikkila, P., Taipale, J., Joensuu, H., Rehn, M.,
Pihlajaniemi, T., Weich, H., deWaal, R., and Alitalo, K. (1999) Am. J.
Pathol. 154, 1381–1390
15. Pajusola, K., Aprelikova, O., Korhonen, J., Kaipainen, A., Pertovaara, L.,
Alitalo, R., and Alitalo, K. (1992) Cancer Res. 52, 5738–5743
16. Pajusola, K., Aprelikova, O., Pelicci, G., Weich, H., Claesson-Welsh, L., and
Alitalo, K. (1994) Oncogene 9, 3545–3555
17. Hughes, D. C. (2001) J. Mol. Evol. 53, 77–79
18. Galland, F., Karamysheva, A., Pebusque, M. J., Borg, J. P., Rottapel, R.,
Dubreuil, P., Rosnet, O., and Birnbaum, D. (1993) Oncogene 8, 1233–1240
19. Pajusola, K., Aprelikova, O., Armstrong, E., Morris, S., and Alitalo, K. (1993)
Oncogene 8, 2931–2937
20. Fournier, E., Dubreuil, P., Birnbaum, D., and Borg, J. P. (1995) Oncogene 11,
921–931
21. Makinen, T., Veikkola, T., Mustjoki, S., Karpanen, T., Catimel, B., Nice, E. C.,
Wise, L., Mercer, A., Kowalski, H., Kerjaschki, D., Stacker, S. A., Achen,
M. G., and Alitalo, K. (2001) EMBO J. 20, 4762–4773
22. Joukov, V., Sorsa, T., Kumar, V., Jeltsch, M., Claesson-Welsh, L., Cao, Y.,
Saksela, O., Kalkkinen, N., and Alitalo, K. (1997) EMBO J. 16, 3898–3911
23. Jussila, L., Valtola, R., Partanen, T. A., Salven, P., Heikkila, P., Matikainen,
M. T., Renkonen, R., Kaipainen, A., Detmar, M., Tschachler, E., Alitalo, R.,
and Alitalo, K. (1998) Cancer Res. 58, 1599–1604
24. Waltenberger, J., Claesson-Welsh, L., Siegbahn, A., Shibuya, M., and Heldin,
C. H. (1994) J. Biol. Chem. 269, 26988–26995
25. Karkkainen, M. J., Ferrell, R. E., Lawrence, E. C., Kimak, M. A., Levinson,
K. L., McTigue, M. A., Alitalo, K., and Finegold, D. N. (2000) Nat. Genet. 25,
153–159
26. Boyle, W. J., van der Geer, P., and Hunter, T. (1991) Methods Enzymol. 201,
110–149
27. van der Geer, P. L. K., Sefton, B., and Hunter, T. (1993) in Protein Phospho-
rylation (Hardie, D., ed) pp. 31–59, Oxford University Press, New York
28. Matsumoto, T., and Claesson-Welsh, L. (2001) Sci. STKE 2001,
http://stke.sciencemag.org/cgi/reprint/sigtrans;2001/112/re21
29. Igarashi, K., Shigeta, K., Isohara, T., Yamano, T., and Uno, I. (1998) Biochem.
Biophys. Res. Commun. 251, 77–82
30. Claesson-Welsh, L. (1994) Prog. Growth Factor Res. 5, 37–54
31. Saaristo, A., Veikkola, T., Tammela, T., Enholm, B., Karkkainen, M. J.,
Pajusola, K., Bueler, H., Yla-Herttuala, S., and Alitalo, K. (2002) J. Exp.
Med. 196, 719–730
32. Kubo, H., Fujiwara, T., Jussila, L., Hashi, H., Ogawa, M., Shimizu, K., Awane,
M., Sakai, Y., Takabayashi, A., Alitalo, K., Yamaoka, Y., and Nishikawa,
S. I. (2000) Blood 96, 546–553
33. Matsumura, K., Hirashima, M., Ogawa, M., Kubo, H., Hisatsune, H., Kondo,
N., Nishikawa, S., and Chiba, T. (2003) Blood 101, 1367–1374
34. Huang, K., Andersson, C., Roomans, G. M., Ito, N., and Claesson-Welsh, L.
(2001) Int. J. Biochem. Cell Biol. 33, 315–324
35. Ekman, S., Thuresson, E. R., Heldin, C. H., and Ronnstrand, L. (1999) Onco-
gene 18, 2481–2488
FIG. 6. VEGFR-2 heterodimerized with VEGFR-3 fails to phosphorylate VEGFR-3 on Tyr-1337 and Tyr-1363. V5 epitope-tagged
kinase-dead VEGFR-3 (R3-KD) was expressed alone or in combination with either wild type VEGFR-3 or wild type VEGFR-2 in 293T cells. Cells
stimulated with VEGF-C were lysed, and R3-KD was immunoprecipitated and used for phosphopeptide mapping. Spots representing phosphopep-
tides containing tyrosine residues that failed to become phosphorylated by VEGFR-2 are circled by the dashed lines.
VEGFR-3 Tyrosine Phosphorylation Sites 40979