[CANCER RESEARCH 64, 8630 – 8638, December 1, 2004]
Loss of Neural Cell Adhesion Molecule Induces Tumor Metastasis by Up-regulating
Lymphangiogenesis
Ivana Crnic,1 Karin Strittmatter,1 Ugo Cavallaro,1,2 Lucie Kopfstein,1 Lotta Jussila,3 Kari Alitalo,3 and
Gerhard Christofori1
1
Institute of Biochemistry and Genetics, Department of Clinical-Biological Sciences, University of Basel, Basel, Switzerland; 2Istituto Federazione Italiana Ricerca Cancro di
Oncologia Molecolare, Milano, Italy; and 3Molecular/Cancer Biology Laboratory and Ludwig Institute for Cancer Research, Biomedicum Helsinki, University of Helsinki,
Helsinki, Finland
ABSTRACT
Reduced expression of neural cell adhesion molecule (NCAM) has been
implicated in the progression to tumor malignancy in cancer patients.
Previously, we have shown that the loss of NCAM function causes the
formation of lymph node metastasis in a transgenic mouse model of
pancreatic ␤ cell carcinogenesis (Rip1Tag2). Here we show that tumors of
NCAM-deficient Rip1Tag2 transgenic mice exhibit up-regulated expression of the lymphangiogenic factors vascular endothelial growth factor
(VEGF)-C and -D (17% in wild-type versus 60% in NCAM-deficient
Rip1Tag2 mice) and, with it, increased lymphangiogenesis (0% in wildtype versus 19% in NCAM-deficient Rip1Tag2 mice). Repression of
VEGF-C and -D function by adenoviral expression of a soluble form of
their cognate receptor, VEGF receptor-3, results in reduced tumor lymphangiogenesis (56% versus 28% in control versus treated mice) and
lymph node metastasis (36% versus 8% in control versus treated mice).
The results indicate that the loss of NCAM function causes lymph node
metastasis via VEGF-C- and VEGF-D-mediated lymphangiogenesis.
These results also establish Rip1Tag2;NCAM-deficient mice as a unique
model for stochastic, endogenous tumor lymphangiogenesis and lymph
node metastasis in immunocompetent mice.
INTRODUCTION
Neural cell adhesion molecule (NCAM, CD56) is a Ca2⫹-independent cell adhesion molecule, which mediates homotypic and heterotypic cell-cell and cell-matrix adhesion (1–3). Its role in neurite
outgrowth, axon guidance, and long-term potentiation has been investigated in great detail (4). However, NCAM is also expressed in
many other cell types, including epithelial cells of various organs,
muscle, and pancreatic ␤ cells. On the basis of alternative splicing of
the COOH-terminal domains, NCAM isoforms are grouped into three
main classes: a glycosylphosphatidyl inositol-linked Mr 120,000 isoform (NCAM120) and transmembrane Mr 140,000 and Mr 180,000
isoforms (NCAM140 and NCAM180, respectively; refs. 1–3).
NCAM140 and 180 are predominantly expressed during embryonic
development, whereas NCAM120 is found in many different adult
tissues (5, 6). In various cancer types, the expression of NCAM shifts
from the Mr 120,000 isoform to the Mr 140,000 and Mr 180,000
isoforms (7). Moreover, an overall decrease in the NCAM level has
been observed in a subset of tumors. For example, in colon carcinoma,
pancreatic cancer, and astrocytoma, NCAM expression is markedly
down-regulated, and the loss of NCAM correlates with poor prognosis
Received 7/15/04; revised 9/2/04; accepted 9/23/04.
Grant support: Boehringer Ingelheim Inc., Austrian Industrial Research Promotion
Fund, Swiss National Science Foundation (I. Crnic, U. Cavallaro, and G. Christofori),
Human Frontier Science Program (R6P0231/2001-M), European Union (QLC1-CT-200101172, and QLK3-CT-2002-02059), and NIH Grant HD37243 (L. Jussila and K. Alitalo).
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.
Note: U. Cavallaro is currently in IFOM-Federazione Italiana Ricerca Cancro Institute
of Molecular Oncology, Milano, Italy. Supplementary data for this article can be found at
Cancer Research Online at http://cancerres.aacrjournals.org.
Requests for reprints: Gerhard Christofori, Institute of Biochemistry and Genetics,
Department of Clinical-Biological Sciences, University of Basel, Mattenstrasse 28, CH–
4058 Basel, Switzerland. E-mail: [email protected].
©2004 American Association for Cancer Research.
(8 –11). In contrast, in neuroblastoma and certain neuroendocrine
tumors, cancer progression correlates with increased NCAM expression and extensive polysialylation (12–16).
Although NCAM has been extensively studied as a cell-cell adhesion molecule (2, 17), recent reports highlight the role of NCAM in
signal transduction. For example, in primary neurons or PC12 cells
NCAM is able to activate fibroblast growth factor receptor, thereby
stimulating classical fibroblast growth factor receptor signal transduction pathways and neurite outgrowth (18 –20). Previously, we have
studied the functional role of NCAM during tumor progression in a
transgenic mouse model of pancreatic ␤ cell tumorigenesis
(Rip1Tag2). In Rip1Tag2 transgenic mice, SV40 T antigen (Tag) is
expressed under the control of the rat insulin promoter (Rip1), resulting in the development of ␤ cell tumors in a multistage tumor
progression pathway (21). Although Rip1Tag2 transgenic mice usually do not develop any metastases, on ablation of NCAM function,
metastases to the regional lymph nodes of the pancreas are observed
(22). We have previously shown that, similar to NCAM-mediated
outgrowth in neuronal cells, in ␤ tumor cells, NCAM associates with
fibroblast growth factor receptor and stimulates classical fibroblast
growth factor receptor signal transduction cascades, which lead to the
activation of ␤1 integrin function and cell-substrate adhesion (23).
Yet, how the loss of NCAM may cause the metastatic dissemination
of ␤ tumor cells to regional lymph nodes has remained elusive.
In the past few years, several major players in the regulation of
lymphangiogenesis, the formation of new lymphatic vessels, have
been identified. Most importantly, vascular endothelial growth factor
(VEGF)-C and -D, members of the VEGF family of growth factors,
have been characterized as the critical players in the induction of
lymphangiogenesis in physiology and disease (24 –26). VEGF-C and
VEGF-D bind preferentially to VEGF receptor (VEGFR)-3 (Flt-4),
which in the adult is predominantly expressed on lymphatic endothelial cells. Fully processed VEGF-C and -D bind with high affinity also
to VEGFR-2 (27–29). The lymphangiogenic capability of VEGF-C
and -D has been shown in many different experimental settings. For
example, forced expression of VEGF-C or -D in tumor cell lines or in
transgenic mouse models of tumorigenesis results in up-regulated
lymphangiogenesis and in the formation of lymph node metastasis
(24). Conversely, a soluble form of VEGFR-3 blocks lymphangiogenesis, by binding and sequestering VEGF-C and -D (30, 31).
Up-regulated expression of VEGF-C, and to a lesser extent of
VEGF-D, also highly correlates with lymphangiogenesis and lymph
node metastasis in cancer patients (24, 32). The recent discovery of a
number of novel markers for lymphatic endothelium such as
VEGFR-3 (Flt-4; refs. 33, 34), LYVE-1 (35, 36), podoplanin (37), and
the transcription factor Prox-1 (38) has facilitated studies on the
molecular mechanisms of lymphangiogenesis.
Here we report that the lymph node metastases, which are observed
in NCAM-deficient Rip1Tag2 transgenic mice, are caused by upregulated expression of VEGF-C and -D and, with it, increased tumor
lymphangiogenesis. Interference with lymphangiogenesis in these
mice by adenoviral expression of soluble VEGFR-3 results in a
8630
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
reduction of tumor lymphangiogenesis and of lymph node metastasis,
indicating that VEGF-C– driven tumor lymphangiogenesis is a metastatic route for NCAM-deficient ␤ tumor cells. Moreover, the results
establish NCAM-deficient Rip1Tag2 transgenic mice as a unique
animal model for stochastic tumor lymphangiogenesis.
MATERIALS AND METHODS
Recombinant Adenoviruses. For the expression of soluble VEGFR-3, a
cDNA fragment encoding the first three immunoglobulin homology domains
of the extracellular portion of mouse VEGFR-3 (Flt-4; amino acids 1–332) was
amplified by reverse transcription-PCR from RNA extracted from embryonic
day 10.5 mice with the primer pair 5⬘-CCA TCG ATG AGA ATG CAG CCG
GGC GCT GCG-3⬘/5⬘-CGC CGG ATA TCA CTG ATG AAG GGC TTT
TCG TGC AC-3⬘ and ligated in frame to a cDNA construct encoding the
mouse immunoglobulin Fc-domain (39). Recombinant E1/E3 defective adenovirus expressing soluble VEGFR-3 (AdsFlt-4) was generated with homologous recombination in E. coli as described previously (ref. 40; see supplementary information). Recombinant adenovirus expressing enhanced green
fluorescent protein (AdEGFP) was used as control (41).
Cell Culture. Murine fibroblastoid cells (L cells) were cultured in DMEM
supplemented with 10% FCS (Invitrogen, Carlsbad, CA), 2 mmol/L glutamine,
100 units/mL penicillin, and 100 ␮g/mL streptomycin sulfate. Cells were
transfected with pCDNA-sFlt– 4 with LipofectAMINE (Invitrogen), and conditioned medium (48 hours conditioning) was concentrated on Centricon
YM-3 columns (Millipore, Billerica, MA).
Immunoblotting. Forty microliters of concentrated conditioned media and
10 ␮L of mice sera were resolved by 15% and 10% SDS-PAGE, respectively,
and transferred to Immobilon-P membranes (Millipore). Membranes were
probed with 1 ␮g/mL antibody specific for mouse immunoglobulin heavy
chains (Pierce, Rockford, IL) or with 0.2 ␮g/mL antibody specific for mouse
Flt-4 (R&D Systems, Minneapolis, MN). Immunostained proteins were visualized with the enhanced chemiluminescence detection system (Amersham
Biosciences, Piscataway, NJ) according to the manufacturer’s recommendations.
Adenoviral Gene Delivery in Mice. Two hundred microliters of HEPESbuffered saline containing 5 ⫻ 109 adenoviral particles (AdsFlt-4 or control
AdEGFP) were injected into the tail vein (i.v.) of Rip1Tag2;NCAM-deficient
mice, 3 times (day 1, 10, and 20), starting at 9 to 10 weeks of age. Glucose (5%
w/v) was added to the drinking water at 10 weeks of age to counteract
hypoglycemia, which results from insulinoma development. Mice were sacrificed 4 weeks after the first injection, and tumor incidence and diameters were
determined. Tumor volumes were calculated from the tumor diameter by
assuming a spherical shape of the tumors, and total tumor volumes per mouse
were determined.
Histology and Immunohistochemistry. Pancreata were fixed overnight in
4% paraformaldehyde in PBS, dehydrated, and embedded in paraffin. Alternatively, freshly isolated tissue was snap frozen in liquid nitrogen and then
embedded in OCT (Tissue Tek, Redding, CA). Five-micron sections were cut
for immunohistochemical analysis. For BrdUrd labeling, mice were injected
i.p. with 100 ␮g BrdUrd (Sigma, St. Louis, MO) per gram of body weight 2
hours before sacrifice. For immunohistochemistry on paraffin sections, the
following antibodies were used: goat antihuman VEGF-C (C-20, Santa Cruz
Biotechnology, Santa Cruz Biotechnology, Santa Cruz, CA); rabbit antimouse
LYVE-1 (a gift from David Jackson, Oxford, United Kingdom), hamster
antimouse podoplanin (8.1.1; Developmental Studies Hybridoma Bank, Iowa
University, Iowa City, Iowa), rabbit antihuman VEGF-D (a gift from Herbert
Weich, German Research Center for Biotechnology, Braunschweig, Germany), biotinylated mouse anti-BrdUrd (Zymed, South San Francisco, CA),
and guinea pig anti-insulin (DakoCytomation, Glostrup, Denmark). Rabbit
antimouse Prox1 (42), rat antimouse CD-31 (PharMingen, Franklin Lakes,
NJ), and double-stainings with these antibodies were performed on fresh
frozen sections fixed for 10 minutes at ⫺20°C in methanol. Apoptotic cells
were visualized with the In Situ Cell Death Detection Kit, POD (Roche, Basel,
Switzerland). All biotinylated secondary antibodies for immunohistochemistry
(Vector, Burlingame, CA) were used at a 1:200 dilution, and positive staining
was visualized with the ABC horseradish peroxidase kit (Vector) and DAB
Peroxidase Substrate (Sigma) according to the manufacturers’ recommendations. For the analysis of tissue morphology, slides were slightly counterstained with hematoxylin. For immunofluorescence analysis, Alexa Fluor 568
and Alexa Fluor 488-labeled secondary antibodies diluted 1:400 were used
(Molecular Probes, Eugene, OR). The 6-diamidino-2-phenylindole (DAPI)
was used for nuclear counterstaining. To ensure specificity of the antibodies
used, pancreatic sections from normal control mice or sections from Rip1Tag2
mice without primary antibodies were used as negative controls. Pancreatic
sections from Rip1Tag2; Rip1VEGF-C transgenic mice were included as
positive controls (43).
For quantitation of blood vessel density, proliferation (BrdUrd incorporation), and apoptosis [terminal deoxynucleotidyl transferase-mediated nick end
labeling (TUNEL) reaction], the numbers of vascular profiles positive for
CD-31 or the number of nuclei staining positive for BrdUrd or the TUNEL
reaction were determined in 10 comparable fields per section at 400 ⫻ magnification, and the mean of positive cells ⫾ SD/field was calculated. Lymphangiogenesis was quantified as described in the legend to Table 1.
RESULTS
Reduced NCAM Expression Correlates with Higher VEGF-C
Expression. As we have previously reported, the loss of NCAM
function in the Rip1Tag2 transgenic mouse model of ␤ cell tumorigenesis results in a marked disaggregation of primary ␤ cell
tumors (23) and in the formation of metastasis to the regional
draining lymph nodes of the pancreas (ref. 22; supplementary Fig.
S1). A comparable induction of lymph node metastasis has been
observed on forced expression of VEGF-C in the Rip1Tag2 mouse
model (43). This intriguing similarity motivated us to investigate
whether VEGF-C and VEGF-D play a role in lymph node metastasis in NCAM-deficient Rip1Tag2 mice. NCAM is a haploinsufficient gene, manifested, for example, by the lack of significant
difference in the phenotype of heterozygous (NCAM⫹/⫺) versus
homozygous (NCAM⫺/⫺) knockout mice (17, 44, 45). Similarly,
Table 1 VEGF-C expression, tumor lymphangiogenesis, and metastasis in Rip1Tag2;NCAM-deficient mice
LVD †
Genotype
Rip1Tag2
Rip1Tag2;NCAM⫹/⫺ and Rip1Tag2;NCAM⫺/⫺
Rip1Tag2;Rip1VEGF-C
VEGF-C *
0%
⬍50%
⬎50%
Metastasis ‡
17% §
(n ⫽ 175, N ⫽ 8)
60% §
(n ⫽ 379, N ⫽ 18)
81%
(n ⫽ 153, N ⫽ 9)
75% §
(n ⫽ 154, N ⫽ 9)
51% §
(n ⫽ 349, N ⫽ 8)
5%
(n ⫽ 153, N ⫽ 9)
25%
0%
30%
19%
12%
83%
0%
(N ⫽ 7)
47%
(N ⫽ 15)
60%
(N ⫽ 5)
Abbreviations: LVD, lymphatic vessel density; n, number of tumors; N, number of mice.
* VEGF-C expression is presented as percentages of VEGF-C–positive tumors per genotype.
† Lymphatic vessels were identified by LYVE-1 immunoreactivity with light microscopy. LVD was determined by assessing the extent by which lymphatic vessels surrounded the
perimeter of a tumor. Tumors from all mice of a particular genotype were then grouped into three classes: tumors that were not in significant contact with any lymphatic vessel (0%),
tumors that were surrounded ⬍50% of the tumor perimeter (⬍50%), and tumors that were surrounded ⬎50% by lymphatic vessels (⬎50%). Results are given as percentages of tumors
in a given LVD class.
‡ Mice with detectable lymph node metastasis per total number of mice.
§ P ⬍ 0.001 (z test with Yates correction for continuity).
8631
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
NCAM⫹/⫺ and NCAM⫺/⫺ Rip1Tag2 mice also do not exhibit
any detectable differences in tumor progression and in the incidence of lymph node metastasis (22). Therefore, we used Rip1Tag2;
NCAM⫹/⫺ and Rip1Tag2;NCAM⫺/⫺ mice in the same experimental
groups and, for the purpose of clarity, refer to both of them as
Rip1Tag2;NCAM-deficient mice.
First, we determined VEGF-C expression in ␤ cell tumors of
Rip1Tag2 and Rip1Tag2;NCAM-deficient mice by immunohistochemical staining with specific antibodies against VEGF-C. These
experiments revealed a significant up-regulation of VEGF-C protein
expression in NCAM-deficient tumors (Fig. 1, B, C, and F; Table 1).
Although only a very small subset of wild-type ␤ cell tumors expressed VEGF-C, it was detected in more than half of the NCAM-
deficient tumors analyzed (Table 1). VEGF-C staining in ␤ tumor
cells seemed granular and was predominantly found in the perinuclear
area of tumor cells (Fig. 1, A–C and F; data not shown). VEGF-C
expression seemed to be independent of the tumor stage, as it was
detectable in benign tumors (adenomas) as well as in malignant
tumors (carcinomas; Fig. 1, B, F, and G; data not shown). Smaller
adenoma expressed VEGF-C with higher incidence and more uniformly (Fig. 1B) than larger tumors, where VEGF-C expression was
limited to specific areas within the tumor (Fig. 1G; and data not
shown) or single scattered cells (Fig. 1C). Simultaneous immunofluorescence staining for insulin and VEGF-C revealed that VEGF-C is
expressed exclusively by insulin-expressing ␤ tumor cells, confirming
that VEGF-C is produced by ␤ tumor cells and not by other cell types.
Fig. 1. VEGF-C and VEGF-D expression in Rip1Tag2;NCAM-deficient tumors. Immunohistochemical analysis of pancreata from Rip1Tag2;NCAM⫹/⫹ and Rip1Tag2;NCAMdeficient mice, as indicated. Immunohistochemical analysis with antibodies against VEGF-C (A–C) reveals that tumors of wild-type Rip1Tag2 mice only rarely express VEGF-C (A;
see also Table 1), whereas in small adenoma of NCAM-deficient Rip1Tag2 mice, VEGF-C (gray) is often expressed by a majority of the tumor cells within the tumor (B). In larger
tumors, VEGF-C is expressed by a subset of tumor cells scattered within the tumor (C, arrows). Immunohistochemical analysis with antibodies against VEGF-D (D and E) reveals
that VEGF-D (gray) is often expressed by a subset of tumor cells (D, arrows) or only a few cells scattered within a tumor (E, arrows). Immunohistochemical analysis with antibodies
against VEGF-D (E) and VEGF-C (F) on serial sections reveals that most of the ␤ cells of a Rip1Tag2;NCAM⫺/⫺ tumor express VEGF-C (F), whereas only a few cells of the same
tumor express VEGF-D (E). G–I, visualization of VEGF-C (G, red) and insulin (H, green) expression by immunofluorescence microscopy in a Rip1Tag2;NCAM⫺/⫺ tumor. Overlay
of the VEGF-C and insulin staining visualizes cells that coexpress VEGF-C and insulin (I, yellow) and cells that do not express VEGF-C (green), both displayed at higher magnification
in the inset. VEGF-C expressing cells (red) that do not express insulin are not observed. Light red staining marked by arrowheads originates from nonspecific staining of erythrocytes
in blood capillaries. Scale bar, 50 ␮m. (T, tumor; E, exocrine)
8632
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
Fig. 2. Tumor lymphangiogenesis in NCAMdeficient tumors of Rip1Tag2 mice. Immunohistochemical staining of histologic sections of pancreata from Rip1Tag2 and Rip1Tag2;NCAM-deficient
mice with antibodies against LYVE-1 and
VEGFR-3 (Flt-4), as indicated. Arrowheads indicate lymphatic vessels visualized by LYVE-1 or
VEGFR-3 antibodies. In Rip1Tag2 mice LYVE-1
antibodies decorate lymphatic vessels in the exocrine pancreas (A), whereas tumors of Rip1Tag2;
NCAM-deficient mice are completely or partially
surrounded by lymphatic vessels (B–F). Intratumoral lymphatic vessels of Rip1Tag2;NCAM-deficient tumors seem thin and collapsed and are usually localized at the tumor periphery (D). Clusters
of ␤ tumor cells are frequently detected in lymphatic vessels surrounding the primary tumor (E) or
in its proximity (F, arrows). Scale bar, 50 ␮m. (T,
tumor; E, exocrine pancreas)
On the other hand, not all of the insulin-expressing cells expressed
VEGF-C (Fig. 1, G–I).
Immunohistochemical analysis of VEGF-D expression in NCAMdeficient Rip1Tag2 tumors revealed an even more variable expression
pattern: whereas in some tumors a high proportion of tumor cells were
expressing VEGF-D (Fig. 1D), only single, scattered VEGF-D expressing cells were found in other tumors (Fig. 1E). A comparison of
VEGF-D and VEGF-C distribution on serial sections revealed that
VEGF-D was frequently found to be expressed by a small subset
of VEGF-C– expressing tumor cells, as exemplified in Fig. 1, panels
E and F. Tumors of NCAM-expressing Rip1Tag2 mice were all
negative for VEGF-D expression (data not shown).
Taken together, these results show that the loss of NCAM expression leads to the up-regulation of VEGF-C and -D expression in
tumors of Rip1Tag2 mice. The lack of a clear correlation between
VEGF-C and -D expression and tumor stage suggests that the expression of VEGF-C and -D may involve stochastic events that are not
directly affected by tumor progression.
Reduced NCAM Expression Correlates with Tumor Lymphangiogenesis. Because Rip1Tag2;NCAM-deficient tumors exhibited increased expression of VEGF-C and -D, we verified whether this
resulted in the induction of lymphangiogenesis. Immunohistochemical staining with antibodies against LYVE-1 and VEGFR-3 (Flt-4)
revealed a significant increase of lymphatic vessel density in
Rip1Tag2;NCAM-deficient mice (Fig. 2; Table 1). In normal control
mice, lymphatic vessels were present only in the exocrine pancreas
and were not in significant contact with islets of Langerhans (43). In
Rip1Tag2 mice, the majority of lymphatic vessels were also not in
significant contact with ␤ cell tumors (Fig. 2A). In contrast, almost
half of the NCAM-deficient tumors were either completely or partially surrounded by lymphatic vessels (Fig. 2B, C and F; Fig. 3; Table
1). In accordance with previously published results, the highest percentage of tumors surrounded by lymphatic vessels was observed in
Rip1Tag2;Rip1VEGF-C mice (Table 1), which we have used as a
reference in these experiments (ref. 43; data not shown).
A small subset of Rip1Tag2;NCAM-deficient and Rip1Tag2;
8633
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
Fig. 3. Specificity of lymphatic and blood endothelial cell markers. Immunofluorescence staining of histologic sections from pancreata of
Rip1Tag2;NCAM-deficient mice with antibodies
against podoplanin (PODO), LYVE-1, Prox-1, and
CD31, as indicated. Nuclei are stained with DAPI.
A, double-immunofluorescence staining reveals
significant colocalization of the two different lymphatic endothelial markers podoplanin (red) and
LYVE-1 (green) in lymphatic endothelium surrounding the NCAM⫺/⫺ tumors. The arrow indicates a cluster of tumor cells within a lymphatic
vessel surrounding the tumor. B, double-immunofluorescence staining with antibodies against podoplanin (red) and Prox-1 (green) reveals significant colocalization of nuclear Prox-1 (arrows) with
lymphatic endothelium in the vicinity of a NCAMdeficient Rip1Tag2 tumor. Dashed lines indicate
the tumor edge. Double-immunofluorescence staining with antibodies against LYVE-1 (green)/CD31
(red; C) and podoplanin (red)/CD31 (green; D)
reveals that both LYVE-1 and podoplanin (PODO)
antibodies specifically stain lymphatic vessels.
Blood vessels strongly express CD31 and are negative for LYVE-1 and podoplanin. Weak CD31
expression is also observed on lymphatic vessels.
Scale bar, 100 ␮m. (T, tumor; E, exocrine pancreas)
Rip1VEGF-C tumors (9% and 12%, respectively) contained intratumoral structures that reacted positively with antibodies against
LYVE-1 (Fig. 2D; data not shown) and podoplanin (data not shown).
The majority of intratumoral lymphatic vessels was “collapsed,” although noncollapsed vessels with open lumen were sometimes observed within tumors (data not shown). Notably, the occurrence of
intratumoral lymphatics did not significantly correlate with the incidence of lymph node metastasis.
In addition to LYVE-1, we used antibodies against another specific
marker for lymphatic endothelial cells, podoplanin, which confirmed
the increased lymphatic vessel density in NCAM-deficient tumors
(Fig. 3, A, B, and D). Immunofluorescence costaining for podoplanin
and LYVE-1 revealed a significant colocalization in most of the
vessels stained (Fig. 3A). However, a few minor but significant
differences in specificity could be observed. For example, colocalization of the two markers was best on vessels surrounding NCAMdeficient tumors, whereas some vessels in the exocrine pancreas were
only weakly stained by the podoplanin antibodies (Fig. 3A). Prox-1
was also specifically expressed in lymphatic vessels of Rip1Tag2;
NCAM-deficient tumors, where it colocalized with podoplanin (Fig.
2B). Antibodies against VEGFR-3 (Flt-4) decorated some intratumoral blood capillaries in addition to lymphatic vessels (data not
shown), and, therefore, they were not the antibody of choice to study
lymphangiogenesis in NCAM-deficient Rip1Tag2 mice and possibly
in other systems.
To confirm the specificity of LYVE-1 and podoplanin as markers
of lymphatic endothelium, we performed a costaining for LYVE-1
and for the blood vessel endothelial marker CD31 in tumors from
Rip1Tag2;NCAM-deficient mice. As expected, antibodies against
CD31 strongly decorated blood vessels and capillaries, whereas
LYVE-1–positive lymphatic vessels showed only very weak staining
for CD31 (Fig. 3C). Podoplanin/CD31 double fluorescence analysis in
Rip1Tag2;NCAM-deficient tumors also revealed only a weak signal
for CD31 on podoplanin-positive lymphatic vessels (Fig. 3D).
As it has been reported for Rip1Tag2;Rip1VEGF-C mice (43),
tumor cell clusters were frequently found in the lumen of LYVE-1/
podoplanin-positive lymphatic vessels in Rip1Tag2;NCAM-deficient
mice (arrows in Fig. 2, E and F, and Fig. 3A), suggesting that
lymphatic vessels provide a metastatic route for ␤ tumor cells in
Rip1Tag2;NCAM-deficient mice.
Lymphangiogenesis Is a Rate-Limiting Step in ␤ Cell Lymph
Node Metastasis. In the Rip1Tag2 tumor model, both VEGF-C overexpression and NCAM deficiency resulted in the formation of lymph
node metastasis (22, 43). To determine whether up-regulated lymphangiogenesis is a rate-limiting event in the formation of lymph node
metastasis, we crossed Rip1Tag2;NCAM-deficient mice with Rip1Tag;
Table 2 Correlation between lymphangiogenesis and lymph node metastasis
Genotype
Metastasis and
⬎50% LVD *
Metastasis and no
⬎50% LVD *
No metastasis and
⬎50% LVD *
No metastasis and no
⬎50% LVD *
Rip1Tag2 (N ⫽ 7)
Rip1Tag2;NCAM⫹/⫺and Rip1Tag2;NCAM⫺/⫺ (N ⫽ 14)
Rip1Tag2;Rip1VEGF-C (N ⫽ 5)
0%
14%
60%
0%
29%
0%
0%
36%
40%
100%
21%
0%
Abbreviations: LVD, lymphatic vessel density; N, number of mice per genotype.
* Correlation between tumor lymphangiogenesis and lymph node metastasis was monitored for each mouse by grouping the different genotype mice into four different classes: mice
with lymph node metastasis and ⬎50% tumor LVD (see legend to Table 1), mice with metastasis and without ⬎50% LVD, mice without metastasis and with ⬎50% LVD, and mice
without metastasis and without ⬎50% LVD. Results are given as percentages of the different genotype mice belonging to a particular class.
8634
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
Fig. 4. Expression of recombinant soluble VEGFR-3 (sFlt-4). A, detection of sFlt-4 in
the conditioned medium of L cells transfected with an expression construct encoding a
soluble version of VEGFR-3 fused to a mouse immunoglobulin Fc domain (sFlt-4; Lanes
1 and 3) or with empty vector (mock; Lanes 2 and 4). Soluble Flt-4 is visualized by
immunoblotting with antibodies against mouse immunoglobulin Fc (Lanes 1 and 2) and
against Flt-4 (Lanes 3 and 4). B, detection of soluble VEGFR-3 (sFlt-4) in the sera of one
AdEGFP-injected mouse (Lane 1) and four AdsFlt-4 –injected mice (Lanes 2 to 5) by
immunoblotting with antibodies against mouse Flt-4. A reproduction of the Ponceau S
staining is shown at the bottom of the panel to ensure equal loading of protein. C.
Recombinant soluble VEGFR-3 purchased from R&D Systems (rsVEGFR-3; Lanes 1 to
3, 6.7 ng, 2.2 ng, and 0.7 ng, respectively) was used as control to quantitate serum sFlt-4
levels (AdsFlt-4; Lanes 4 to 6, 10 ␮L of 1 of 100, 1 of 300 and 1 of 900 serum dilution
from a single mouse). Note that purchased soluble VEGFR-3 contains a longer extracellular portion of VEGFR-3 than the adenoviral soluble VEGFR-3 construct (745 versus
332 amino acids, respectively; see Material and Methods).
Rip1VEGF-C mice and examined the incidence of lymph node metastasis by histopathological analysis of pancreata from these mice. In
these separate breeding experiments, 20% of the Rip1Tag2;NCAMdeficient mice (n ⫽ 20), 37.5% of Rip1Tag2;Rip1VEGF-C mice
(n ⫽ 8), and 80% of the composite Rip1Tag2;Rip1VEGF-C;NCAMdeficient mice (n ⫽ 10) exhibited metastasis to the regional lymph
nodes of the pancreas. These results indicate that the induction of
lymphangiogenesis in Rip1Tag2;NCAM-deficient mice is indeed a
rate-limiting step for the metastatic dissemination of ␤ tumor cells to
the regional lymph nodes. However, because the loss of NCAM
function potentiated VEGF-C– dependent lymph node metastasis in
Rip1Tag2;Rip1VEGF-C;NCAM-deficient composite mice, an additional mechanism, that involves the loss of NCAM function but is
distinct from VEGF-C/D–mediated lymphangiogenesis, may be involved as well. This conclusion is also supported by the fact that not
all of the NCAM-deficient mice that exhibited increased lymphatic
vessel density surrounding their tumors developed lymph node metastasis (Table 2). Conversely, there were also subsets of mice displaying lymph node metastasis in the absence of lymphangiogenesis
(Table 2).
The results presented above suggest that NCAM-deficiency leads to
lymph node metastasis, at least in part, through VEGF-C– and VEGFD–mediated lymphangiogenesis. To test this hypothesis, we used a
soluble form of VEGFR-3 (sFlt-4) to interfere with the function of
VEGF-C and -D, and potentially with tumor lymphangiogenesis and
metastasis, in Rip1Tag2;NCAM-deficient mice. To systemically express high levels of sFlt-4, we generated a recombinant, replicationdefective adenovirus encoding for the ligand-binding, extracellular
domain of the receptor (the first three immunoglobulin domains of
mouse Flt-4) with its COOH-terminus fused to a mouse immunoglobulin heavy chain (Fc) under the control of the cytomegalovirus enhancer/promoter. Intact expression of the recombinant protein was
confirmed in vitro by analyzing conditioned medium of L cells
transfected with an expression construct encoding sFlt-4 (Fig. 4A).
Recombinant adenoviruses either expressing sFlt-4 (AdsFlt-4) or
EGFP as control (AdEGFP) were then injected into the tail vein of
Rip1Tag2;NCAM-deficient mice starting at 9 weeks of age, with a
bolus of 5 ⫻ 109 viral particles administered at day 1, 10, and 20.
From previous experiments, it was known that i.v. application of
recombinant adenovirus exclusively infects hepatocytes in the liver,
which in turn produce and secrete soluble proteins into the blood
stream (39). The presence of sFlt-4 in the sera of mice injected i.v.
with purified recombinant adenoviral particles was confirmed by
immunoblotting indicating that sFlt-4 was systemically present in
these mice at high levels (ranging between ⬃50 ␮g/mL and 300
␮g/mL; Fig. 4, B and C).
Immunohistochemical staining for podoplanin in adenovirusinjected Rip1Tag2;NCAM-deficient mice revealed a significant reduction in the number of tumors associated with lymphatic vessels on
systemic expression of sFlt-4 as compared with EGFP (P ⬍ 0.001,
z-test; Table 3).
Analysis of the incidence of lymph node metastasis revealed a
reduction in the number of mice with lymph node metastasis on
treatment with AdsFlt-4 as compared with AdEGFP (Table 3). Notably, the one and only lymph node metastasis that has been detected in
a AdsFlt-4-treated Rip1Tag2;NCAM-deficient mouse exhibited a
highly malignant, anaplastic phenotype, and most likely has originated from a primary tumor of similar malignancy (supplementary
Fig. S2). Because these experiments had to be performed at the very
late stages of Rip1Tag2 tumor development, and for ethical reasons,
the number of experimental mice had to be kept at a minimum. As a
Table 3 Soluble Flt-4 represses tumor lymphangiogenesis and lymph node metastasis
in Rip1Tag2;NCAM-deficient mice
Podoplanin *
AdEGFP
(n ⫽ 201, N ⫽ 11)
AdsFlt-4
(n ⫽ 232, N ⫽ 12)
0%
⬍50%
⬎50%
Metastasis †
12.44% ‡
31.34%
56.22% ‡
4/11 §
40.52% ‡
31.03%
28.45% ‡
1/12 §
Abbreviations: LVD, lymphatic vessel density; n, number of tumors; N, number of
treated mice.
* LVD was determined by podoplanin immunohistochemistry with a light microscope
and scored as described in Table 1. The results are given as percentages of tumors from
control and AdsFlt4-treated mice in the different LVD groups.
† Number of mice with lymph node metastasis of the total number of treated mice.
‡ P ⬍ 0.001 (z test with Yates correction for continuity).
§ P ⫽ 0.155 (two-tailed Fisher exact test).
8635
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
Table 4 Systemic expression of sFlt-4 does not affect general tumor growth in
Rip1Tag2;NCAM-deficient mice
Tumor incidence
(number per mouse)
Tumor volume (mm3)
Proliferation *
(cells per field)
Apoptosis †
(cells per field)
Blood vessel density ‡
(vessels per field)
AdEGFP
AdsFlt-4
11.2 ⫾ 5.5
10.2 ⫾ 6.7
39.5 ⫾ 30.0
57.0 ⫾ 29.5
56.5 ⫾ 45.5
59.6 ⫾ 30.9
10.9 ⫾ 8.1
9.9 ⫾ 6.1
27.4 ⫾ 12.5
26.7 ⫾ 10.6
NOTE. Values are mean ⫾ SD. The number of treated mice was 11 and 12 for
AdEGFP and AdsFlt-4, respectively.
* Cells per microscopic field (400⫻ magnification) in S phase as determined by
BrdUrd incorporation.
† Apoptotic cells per microscopic field (400⫻ magnification) as determined by
TUNEL staining.
‡ CD31-positive vessels per microscopic field (400⫻ magnification).
result, the number of mice in this experiment is rather low and,
although a clear trend is apparent, the reduction of lymph node
metastasis on AdsFlt-4 treatment is not statistically significant
(P ⫽ 0.115, two-tailed Fisher exact test; Table 3). Besides the changes
in the incidence of lymph node metastasis, no significant differences
were observed for tumor incidences and volumes, tumor cell proliferation (BrdUrd-incorporation), apoptosis (TUNEL assay), and tumor
angiogenesis (CD31 staining; Table 4).
Together, the results indicate that interfering with the function of
VEGF-C and -D during tumor development in Rip1Tag2;NCAMdeficient mice reduces lymphangiogenesis and, to a lesser extent,
lymph node metastasis but does not affect tumor cell proliferation,
tumor cell apoptosis, or blood vessel angiogenesis. Thus, our data
raise the possibility that, at least in the Rip1Tag2 transgenic mouse
model, the loss of NCAM induces lymph node metastasis by upregulated expression of VEGF-C and VEGF-D, which in turn induces
lymphangiogenesis.
DISCUSSION
On the basis of changes in gene expression, alternative splicing, and
differential polysialylation, the cell-cell adhesion molecule NCAM
has been functionally implicated in tumor progression in a variety of
cancer types (7). However, the cellular and molecular mechanisms by
which changes in NCAM function may affect tumor progression have
remained elusive. A first glimpse into such mechanisms has come
from experiments with Rip1Tag2 transgenic mice, where the loss of
NCAM contributes to the metastatic dissemination of ␤ tumor cells to
regional lymph nodes (22). Here, we set out to identify and characterize the molecular mechanisms connecting the loss of NCAM function with lymph node metastasis. We show that, on loss of NCAM
function, ␤ cell tumors of Rip1Tag2 transgenic mice express increased levels of the lymphangiogenic factors VEGF-C and VEGF-D.
As a result, NCAM-deficient tumors exhibit up-regulated lymphangiogenesis, which in turn provides tumor cells with a route for
metastatic dissemination to local lymph nodes.
Immunohistochemical analysis with antibodies against VEGF-C
and -D, and against a number of well-established markers for lymphatic vessels, revealed that in a significant proportion of NCAMdeficient Rip1Tag2 tumors, both the expression of VEGF-C and -D
and lymphangiogenesis are significantly up-regulated (Table 1). However, VEGF-D expression by ␤ tumor cells of NCAM-deficient
Rip1Tag2 mice was less frequent as compared with the expression of
VEGF-C (Fig. 1, D–F), implicating that VEGF-C is the major lymphangiogenic player in Rip1Tag2;NCAM-deficient mice.
Interestingly, tumors for which the diameter is partially (⬍50%)
surrounded by lymphatic vessels were equally represented in NCAMdeficient as well as NCAM-expressing Rip1Tag2 tumors (Table 1).
Such a significant interaction between ␤ cell tumors and lymphatic
vessels can be explained by the fact that the mouse pancreas contains
a rich network of lymphatic vessels, which is frequently found in the
vicinity of islets of Langerhans (46). Hence, it is conceivable that
outgrowing ␤ cell tumors may have a high chance of becoming
partially surrounded by pre-existing lymphatic vessels of the pancreas.
In contrast, during active tumor lymphangiogenesis, for example
induced by up-regulated VEGF-C and -D expression in NCAMdeficient tumors or by transgenic expression of VEGF-C in Rip1Tag2;
Rip1VEGF-C tumors, most of the tumor perimeter (⬎50%) is surrounded by lymphatic vessels (ref. 43; Table 1). Thus, Rip1Tag2;
NCAM-deficient mice represent a model for stochastic, endogenous
tumor lymphangiogenesis, a useful tool for future investigations.
Combining the lymphangiogenic effect of forced expression of
VEGF-C in Rip1VEGF-C transgenic mice with the lymphangiogenesis observed in NCAM-deficient mice in composite Rip1VEGF-C;
NCAM-deficient;Rip1Tag2 mice resulted in an increase of lymph
node metastasis that was more than additive. This result indicates that
the expression of VEGF-C and -D is a rate limiting event in the
formation of lymph node metastases, but that additional mechanisms,
triggered by the ablation of NCAM function, are also involved in this
process in Rip1Tag2 mice. Variations in the genetic background of the
different composite mouse lines may also contribute: the incidence of
lymph node metastasis observed in NCAM-deficient mice varies with
different genetic backgrounds (50% in ref. 22, 47% and 37%, respectively, when Rip1Tag2 were excusivley crossed to NCAM knockout
mice, and 20% when Rip1Tag2 mice were crossed to NCAM-deficient and Rip1VEGF-C transgenic mice). A correlation analysis between lymphangiogenesis and lymph node metastasis within single
mice revealed that in a subset of NCAM-deficient mice, lymph node
metastasis developed in the absence of active tumor lymphangiogenesis (Table 2). On the other hand, the finding that some mice exhibited
ongoing lymphangiogenesis in their tumors and did not develop
lymph node metastasis may be explained by the fact that some of the
metastases were missed by the histologic analysis or that the mice
were sacrificed before the formation of metastasis. Likewise, not all of
the Rip1VEGF-C transgenic mice that exhibited active lymphangiogenesis developed lymph node metastasis (ref. 43; Table 2).
Systemic expression of a soluble form of VEGFR-3 (sFlt-4) by
adenoviral gene delivery reduced the percentage of tumors surrounded
by lymphatic vessels. Such a repression of tumor lymphangiogenesis
by sFlt-4 is in accordance with previous reports on experimental
tumors in chicken, mouse, and rat (31, 47, 48). Here we show for the
first time that sFlt-4 is also able to reduce tumor lymphangiogenesis
caused by the up-regulation of VEGF-C and -D in endogenously
growing tumors not involving tumor transplantation. Soluble Flt-4 in
the circulation of Rip1Tag2;NCAM-deficient mice also reduced the
incidence of lymph node metastasis from 36% (4 of 11) in the control
group of mice to 8% (1 of 12; Table 3), raising the possibility that
loss-of-NCAM–mediated induction of lymphangiogenesis contributes
to the formation of lymph node metastasis.
In contrast to the significant reduction of lymphatic vessel density,
blood vessels were not affected by the adenoviral expression of sFlt4
in Rip1Tag2;NCAM-deficient mice (Table 4). This result indicates
that the treatment specifically repressed lymphangiogenesis and not
tumor angiogenesis, a selectivity that has been previously reported in
various experimental models (48, 49).
There has been much controversy about the functionality of intratumoral lymphatic vessels and their importance for metastatic tumor
dissemination (reviewed in ref. 32). Although we clearly observed
intratumoral lymphatic vessels in Rip1Tag2;NCAM-deficient mice,
8636
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
their occurrence did not correlate with the incidence of lymph node
metastasis. Moreover, the lumina of these intratumoral lymphatic
vessels were usually collapsed. Finally, sFlt-4 reduced only lymphatic
vessels surrounding tumors and not intratumoral lymphatic vessels (data
not shown). Together, these observations support the notion that intratumoral lymphatics detected in Rip1Tag2 tumors are not functional.
The correlation between the loss of NCAM function and the gain of
VEGF-C and -D expression in Rip1Tag2 tumors raises a number of
questions about the cellular and molecular mechanisms involved in
this process. First, is the loss of NCAM directly affecting VEGF-C
and -D gene expression? To address this issue, we have investigated
VEGF-C expression in ␤ tumor cell lines derived from tumors of
wild-type and NCAM-deficient Rip1Tag2 mice (23). The results
indicate that in cultured cells VEGF-C expression is not directly
affected by changes in NCAM expression (data not shown). Hence,
we speculate that up-regulated VEGF-C expression is caused by
mechanisms that depend on the tissue or tumor context in Rip1Tag2
mice in vivo.
One obvious phenotypic difference between wild-type and NCAMdeficient Rip1Tag2 tumors is the dramatic tissue disaggregation of
NCAM-deficient tumors. In previous work, we have shown that the
loss of NCAM results in a failure to activate ␤1 integrin via fibroblast
growth factor receptor signaling (23), a likely cause for tumor tissue
disaggregation. It is conceivable that such changes in cell-matrix
adhesion may be the cause for the up-regulation of VEGF-C and -D
expression, for example, by changing the interstitial fluid pressure
within the tumor or by causing major fluid leakages and increased
tissue fluid volume. In fact, many of the ␤ cell tumors of Rip1Tag2;
NCAM-deficient mice exhibit hemorrhages filled with lymphatic
fluid (23), an observation that initially motivated us to investigate
lymphangiogenesis in these tumors.
Alternatively, tissue disaggregation may lead to an inflammatory
response which in turn could affect lymphangiogenesis. Such mechanism is attractive, because human and mouse VEGF-C promoters
contain conserved binding sites for nuclear factor ␬B (50), a transcription factor which plays a central role in inflammation and which
has often been correlated with cancer (51). Notably, the proinflammatory cytokines interleukin-1 and tumor necrosis factor-␣ are able to
activate both nuclear factor ␬B transcriptional activity and VEGF-C
gene expression (52). Moreover, heregulin-␤1 increases VEGF-C
expression through nuclear factor ␬B/p38 mitogen-activated protein
kinase signaling pathways in human breast cancer cells (53). Hence,
it will be important to investigate whether NCAM-deficient tumors
exhibit certain inflammatory responses, which type of inflammatory
cells have infiltrated in these tumors, and which cytokines could
possibly be responsible for the up-regulation of VEGF-C and -D in
NCAM-deficient ␤ tumor cells. Finally, future studies will also have
to address whether changes in NCAM expression also correlate with
up-regulated lymphangiogenesis and lymph node metastasis in cancer
patients.
ACKNOWLEDGMENTS
REFERENCES
1. Walsh FS, Doherty P. Neural cell adhesion molecules of the immunoglobulin superfamily: role in axon growth and guidance. Annu Rev Cell Dev Biol 1997;13:425–56.
2. Crossin KL, Krushel LA. Cellular signaling by neural cell adhesion molecules of the
immunoglobulin superfamily. Dev Dyn 2000;218:260 –79.
3. Panicker AK, Buhusi M, Thelen K, Maness PF. Cellular signalling mechanisms of
neural cell adhesion molecules. Front Biosci 2003;8:900 –11.
4. Kiss JZ, Muller D. Contribution of the neural cell adhesion molecule to neuronal and
synaptic plasticity. Rev Neurosci 2001;12:297–310.
5. Gower HJ, Barton CH, Elsom VL, et al. Alternative splicing generates a secreted form
of N-CAM in muscle and brain. Cell 1988;55:955– 64.
6. Crossin KL, Chuong CM, Edelman GM. Expression sequences of cell adhesion
molecules. Proc Natl Acad Sci USA 1985;82:6942– 6.
7. Cavallaro U, Christofori G. Cell adhesion and signalling by cadherins and Ig-CAMs
in cancer. Nat Rev Cancer 2004;4:118 –32.
8. Fogar P, Basso D, Pasquali C, et al. Neural cell adhesion molecule (N-CAM) in
gastrointestinal neoplasias. Anticancer Res 1997;17:1227–30.
9. Sasaki H, Yoshida K, Ikeda E, et al. Expression of the neural cell adhesion molecule
in astrocytic tumors: an inverse correlation with malignancy. Cancer (Phila) 1998;82:
1921–31.
10. Tezel E, Kawase Y, Takeda S, Oshima K, Nakao A. Expression of neural cell
adhesion molecule in pancreatic cancer. Pancreas 2001;22:122–5.
11. Huerta S, Srivatsan ES, Venkatesan N, et al. Alternative mRNA splicing in colon
cancer causes loss of expression of neural cell adhesion molecule. Surgery (St Louis)
2001;130:834 – 43.
12. Gluer S, Schelp C, Madry N, et al. Serum polysialylated neural cell adhesion
molecule in childhood neuroblastoma. Br J Cancer 1998;78:106 –10.
13. Komminoth P, Roth J, Lackie PM, Bitter-Suermann D, Heitz PU. Polysialic acid of
the neural cell adhesion molecule distinguishes small cell lung carcinoma from
carcinoids. Am J Pathol 1991;139:297–304.
14. Lantuejoul S, Moro D, Michalides RJ, Brambilla C, Brambilla E. Neural cell adhesion
molecules (NCAM) and NCAM-PSA expression in neuroendocrine lung tumors.
Am J Surg Pathol 1998;22:1267–76.
15. Trouillas J, Daniel L, Guigard MP, et al. Polysialylated neural cell adhesion molecules expressed in human pituitary tumors and related to extrasellar invasion. J Neurosurg 2003;98:1084 –93.
16. Lantuejoul S, Laverriere MH, Sturm N, et al. NCAM (neural cell adhesion molecules)
expression in malignant mesotheliomas. Hum Pathol 2000;31:415–21.
17. Stork O, Welzl H, Wotjak CT, et al. Anxiety and increased 5-HT1A receptor response
in NCAM null mutant mice. J Neurobiol 1999;40:343–55.
18. Doherty P, Walsh FS. CAM-FGF Receptor Interactions: A Model for Axonal Growth.
Mol Cell Neurosci 1996;8:99 –111.
19. Leshchyns’ka I, Sytnyk V, Morrow JS, Schachner M. Neural cell adhesion molecule
(NCAM) association with PKCbeta2 via betaI spectrin is implicated in NCAMmediated neurite outgrowth. J Cell Biol 2003;161:625–39.
20. Niethammer P, Delling M, Sytnyk V, et al. Cosignaling of NCAM via lipid rafts and
the FGF receptor is required for neuritogenesis. J Cell Biol 2002;157:521–32.
21. Hanahan D. Heritable formation of pancreatic beta-cell tumours in transgenic mice
expressing recombinant insulin/simian virus 40 oncogenes. Nature (Lond) 1985;315:
115–22.
22. Perl AK, Dahl U, Wilgenbus P, et al. Reduced expression of neural cell adhesion
molecule induces metastatic dissemination of pancreatic beta tumor cells. Nat Med
1999;5:286 –91.
23. Cavallaro U, Niedermeyer J, Fuxa M, Christofori G. N-CAM modulates tumour-cell
adhesion to matrix by inducing FGF-receptor signalling. Nat Cell Biol 2001;3:650 –7.
24. Alitalo K, Carmeliet P. Molecular mechanisms of lymphangiogenesis in health and
disease. Cancer Cell 2002;1:219 –27.
25. Stacker SA, Achen MG, Jussila L, Baldwin ME, Alitalo K. Lymphangiogenesis and
cancer metastasis. Nat Rev Cancer 2002;2:573– 83.
26. Lohela M, Saaristo A, Veikkola T, Alitalo K. Lymphangiogenic growth factors,
receptors and therapies. Thromb Haemost 2003;90:167– 84.
27. Orlandini M, Marconcini L, Ferruzzi R, Oliviero S. Identification of a c-fos-induced
gene that is related to the platelet-derived growth factor/vascular endothelial growth
factor family. Proc Natl Acad Sci USA 1996;93:11675– 80.
28. Stacker SA, Stenvers K, Caesar C, et al. Biosynthesis of vascular endothelial growth
factor-D involves proteolytic processing which generates non-covalent homodimers.
J Biol Chem 1999;274:32127–36.
29. Achen MG, Jeltsch M, Kukk E, et al. Vascular endothelial growth factor D (VEGF-D)
is a ligand for the tyrosine kinases VEGF receptor 2 (Flk1) and VEGF receptor 3
(Flt4). Proc Natl Acad Sci USA 1998;95:548 –53.
30. He Y, Kozaki K, Karpanen T, et al. Suppression of tumor lymphangiogenesis and
lymph node metastasis by blocking vascular endothelial growth factor receptor 3
signaling. J Natl Cancer Inst (Bethesda) 2002;94:819 –25.
31. Krishnan J, Kirkin V, Steffen A, et al. Differential in vivo and in vitro expression of
vascular endothelial growth factor (VEGF)-C and VEGF-D in tumors and its relationship
to lymphatic metastasis in immunocompetent rats. Cancer Res 2003;63:713–22.
32. Cassella M, Skobe M. Lymphatic vessel activation in cancer. Ann N Y Acad Sci
2002;979:120 –30.
33. Finnerty H, Kelleher K, Morris GE, et al. Molecular cloning of murine FLT and
FLT4. Oncogene 1993;8:2293– 8.
34. Galland F, Karamysheva A, Pebusque MJ, et al. The FLT4 gene encodes a transmembrane tyrosine kinase related to the vascular endothelial growth factor receptor.
Oncogene 1993;8:1233– 40.
35. Banerji S, Ni J, Wang SX, et al. LYVE-1, a new homologue of the CD44 glycoprotein, is a lymph-specific receptor for hyaluronan. J Cell Biol 1999;144:789 – 801.
We are grateful to Drs. Matt Cotten, Michael Detmar, David Jackson,
Dontscho Kerjaschki, and Herbert Weich for reagents, advice, and scientific
input and Terhi Kärpänen for help with soluble receptor determinations. We
thank Joachim Niedermeyer for the generation and analysis of Rip1VEGF-C;
NCAM-deficient;Rip1Tag2 composite mice, Michaela Herzig for help with
cloning, Edelgard Kux, Ursula Schmieder, and Helena Antoniadis for excellent
technical assistance, and Miguel Cabrita and Birgit Schaffhauser for critical
comments on the manuscript. The monoclonal antibody against mouse podoplanin was obtained from the Developmental Studies Hybridoma Bank under
the auspices of the National Institute of Child Health and Human Development.
8637
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
NCAM AND LYMPHANGIOGENESIS
36. Wigle JT, Harvey N, Detmar M, et al. An essential role for Prox1 in the induction of
the lymphatic endothelial cell phenotype. EMBO J 2002;21:1505–13.
37. Breiteneder-Geleff S, Soleiman A, Kowalski H, et al. Angiosarcomas express mixed
endothelial phenotypes of blood and lymphatic capillaries: podoplanin as a specific
marker for lymphatic endothelium. Am J Pathol 1999;154:385–94.
38. Wigle JT, Oliver G. Prox1 function is required for the development of the murine
lymphatic system. Cell 1999;98:769 –78.
39. Compagni A, Wilgenbus P, Impagnatiello MA, Cotten M, Christofori G. Fibroblast
growth factors are required for efficient tumor angiogenesis. Cancer Res 2000;60:
7163–9.
40. Chartier C, Degryse E, Gantzer M, et al. Efficient generation of recombinant adenovirus
vectors by homologous recombination in Escherichia coli. J Virol 1996;70:4805–10.
41. Glotzer JB, Michou AI, Baker A, Saltik M, Cotten M. Microtubule-independent
motility and nuclear targeting of adenoviruses with fluorescently labeled genomes.
J Virol 2001;75:2421–34.
42. Wilting J, Papoutsi M, Christ B, et al. The transcription factor Prox1 is a marker for
lymphatic endothelial cells in normal and diseased human tissues. FASEB J 2002;
16:1271–3.
43. Mandriota SJ, Jussila L, Jeltsch M, et al. Vascular endothelial growth factor-Cmediated lymphangiogenesis promotes tumour metastasis. EMBO J 2001;20:672– 82.
44. Esni F, Taljedal IB, Perl AK, et al. Neural cell adhesion molecule (N-CAM) is
required for cell type segregation and normal ultrastructure in pancreatic islets. J Cell
Biol 1999;144:325–37.
45. Montag-Sallaz M, Montag D, Schachner M. Altered processing of novel information
in N-CAM-deficient mice. Neuroreport 2003;14:1343– 6.
46. Regoli M, Bertelli E, Orazioli D, Fonzi L, Bastianini A. Pancreatic lymphatic system
in rodents. Anat Rec 2001;263:155– 60.
47. Papoutsi M, Siemeister G, Weindel K, et al. Active interaction of human A375
melanoma cells with the lymphatics in vivo. Histochem Cell Biol 2000;114:373– 85.
48. Karpanen T, Egeblad M, Karkkainen MJ, et al. Vascular endothelial growth factor C
promotes tumor lymphangiogenesis and intralymphatic tumor growth. Cancer Res
2001;61:1786 –90.
49. Makinen T, Jussila L, Veikkola T, et al. Inhibition of lymphangiogenesis with
resulting lymphedema in transgenic mice expressing soluble VEGF receptor-3. Nat
Med 2001;7:199 –205.
50. Chilov D, Kukk E, Taira S, et al. Genomic organization of human and mouse genes
for vascular endothelial growth factor C. J Biol Chem 1997;272:25176 – 83.
51. Hanada T, Yoshimura A. Regulation of cytokine signaling and inflammation. Cytokine Growth Factor Rev 2002;13:413–21.
52. Ristimaki A, Narko K, Enholm B, Joukov V, Alitalo K. Proinflammatory cytokines
regulate expression of the lymphatic endothelial mitogen vascular endothelial growth
factor-C. J Biol Chem 1998;273:8413– 8.
53. Tsai PW, Shiah SG, Lin MT, Wu CW, Kuo ML. Up-regulation of vascular endothelial growth factor C in breast cancer cells by heregulin-beta 1. A critical role of
p38/nuclear factor-kappa B signaling pathway. J Biol Chem 2003;278:5750 –9.
8638
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.
Loss of Neural Cell Adhesion Molecule Induces Tumor
Metastasis by Up-regulating Lymphangiogenesis
Ivana Crnic, Karin Strittmatter, Ugo Cavallaro, et al.
Cancer Res 2004;64:8630-8638.
Updated version
Supplementary
Material
Cited articles
Citing articles
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/64/23/8630
Access the most recent supplemental material at:
http://cancerres.aacrjournals.org/content/suppl/2004/12/06/64.23.8630.DC1
This article cites 52 articles, 20 of which you can access for free at:
http://cancerres.aacrjournals.org/content/64/23/8630.full.html#ref-list-1
This article has been cited by 15 HighWire-hosted articles. Access the articles at:
/content/64/23/8630.full.html#related-urls
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 14, 2017. © 2004 American Association for Cancer
Research.