(CANCER RESEARCH 51. 394-399. Januar) l, 1991]
Endothelial Cell Membrane Vesicles in the Study of Organ Preference of
Metastasis1
Robert C. Johnson, Hellmut G. Augustin-Voss, Duzhang Zhu, and Bendicht U. Paul!2
Cancer Cell Biology Laboratories. Department of Pathology. Cornell University College of Veterinary Medicine, Ithaca, New York 14853
ABSTRACT
Many malignancies exhibit distinct patterns of metastasis that appear
to be mediated by receptor/ligand-like interactions between tumor cells
and organ-specific vascular endothelium. In order to study endothelial
cell surface molecules involved in the binding of metastatic cells, we
developed a perfusion method to isolate outside-out membrane vesicles
from the lumenal surface of rat lung microvascular endothelium. Lungs
were perfused in situ for 4 h at 37°Cwith a solution of 100 mM
formaldehyde, 2 HIMdithiothreitol in phosphate-buffered saline to induce
endothelial cell vesiculation. Radioiodinated rat lung endothelial cell
membrane vesicles bound lung-metastatic tumor cells (B16F10,
R323OAC-MET) in significantly higher numbers than their low or
nonmetastatic counterparts (HIM II. R323OAC-LR). In contrast, leg
endothelial membrane vesicle showed no binding preference for either
cell line. Neuraminidase treatment of vesicles abolished specificity of
adhesion of lung-derived vesicles to lung metastatic tumor cells. These
results demonstrate that in situ perfusion is an appropriate technique to
obtain pure endothelial cell membrane vesicles containing functionally
active adhesion molecules. The preferential binding of lung-derived endo
thelial cell membrane vesicles by lung metastatic tumor cells is evidence
of the importance of endothelial cell adhesion molecules in the formation
of métastases.
procedures involved in growing organ-specific microvascular
endothelial cells (14-16), including modest yields of primary
endothelial cell isolates, contaminations of primary cultures
with other mesenchymal cells (14-16), and loss of organ-spe
cific microvascular phenotype after prolonged culture periods
on plastic (17-23).
In this paper we describe a method that circumvents the use
of cumbersome cell isolation techniques in the identification,
isolation, and characterization of endothelial cell adhesion mol
ecules involved in organ-specific metastasis. Outside-out mem
brane vesicles representative of lumenal membranes of microvascular endothelium are obtained by perfusing organs with low
strength formaldehyde solutions (24, 25). Using endothelial
vesicles in a tumor cell adhesion assay, our data show that
membrane vesicles prepared from the microvasculature of the
lung have a significantly higher binding avidity for lung-met
astatic tumor cells than endothelial cell membrane vesicles
prepared from the microvasculature of a nonmetastasized or
gan.
MATERIALS AND METHODS
Cell Cultures. Murine B16 melanoma cells (B16FO and B16F10)
were obtained from Dr. I. J. Fidler, University of Texas System Cancer
Center, M. D. Anderson Hospital and Tumor Institute, Houston, TX
(26) and R323OAC rat mammary carcinoma cells (R323OAC-MET
and R323OAC-LR) were obtained from Dr. J. A. Kellen, Sunnybrook
Medical Center, University of Toronto, Toronto, Ontario, Canada (27).
B16-F10 and R323OAC-MET were selected in vivo for high lung
colonization (26-28). The parental B16-FO was low lung metastatic
and the (Con A' and WGA)-resistant variant R323OAC-LR was non
INTRODUCTION
Blood vessels are the most common conduits for the met
astatic spread of cancer cells. Colonization of a secondary organ
site by this route is preceded by tumor cell arrest in the microvasculature and subsequent traversal of the endothelial cell
lining. There has been much debate of how this early step in
the metastatic cascade might be accomplished. The first school
of thought argues that metastasis is a random process during
which tumor cells are mechanically trapped in the microvasculature of the first organ entered by blood-borne tumor cells (1,
2). This theory is supported by the observation that the principal
organ for metastatic involvement is the lung which, in hematogenous cancer cell spread, oftentimes is the first microvascu
lar entrapment site. The second school of thought suggests that
tumor cells recognize and bind to organ-specific endothelial cell
adhesion molecules expressed on the lumenal surface of microvascular endothelial cells of the metastasized organ (3-11),
thereby mimicking the homing of lymphocytes to parental
lymphoid tissues (12, 13). Although this hypothesis can explain
many of the distinct metastatic patterns of specific tumor types,
attempts to identify such adhesion molecules have been elusive
to date. This is largely due to complex isolation and culture
metastatic. Cells were maintained in RPMI 1640 medium supple
mented with 10% heat-inactivated fetal bovine serum (Gibco, Grand
Island, NY).
Preparation of Endothelial Cell Membrane Vesicles. Endothelial cell
membrane vesicles were prepared from the microvasculature of rat
lungs and legs. Three-month-old male Sprague-Dawley rats were given
injections i.p. of 0.5 ml of 20% sodium citrate, sacrificed with an
overdose of barbiturate, and prepared for immediate organ perfusion.
Lung perfusates entered through the pulmonary artery and exited from
the left heart atrium; leg perfusates entered through the common iliac
artery and exited via the iliac vein. Lung and leg vascular beds were
first flushed with PBS, pH 7.4, containing 1 mM CaCl2 and 0.5 mM
MgCI2 at a relatively rapid flow rate of 7 ml/min for 20 min at 37°C.
Lungs were inflated periodically through the trachea during the washing
procedure to improve removal of blood components. Flushing was
followed by perfusion of the vasculature with 100 mivi formaldehyde
and 2 mM dithiothreitol in PBS-CM at a flow rate of 0.25 ml/min for
4 h at 37°C(24, 25). Perfusates were centrifuged at 200 x g to remove
Received 7/23/90; accepted 10/22/90.
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.
1This work was supported by Public Health Service Grant CA 47668 (B. U.
P.) from the National Cancer Institute, a grant from the Cornell Biotechnology
Program (B. U. P.) and a fellowship from the Deutsche Forschungsgemeinschaft
(H. G. A-V.).
2To whom requests for reprints should be addressed, at Department of
Pathology, Cornell University, College of Veterinary Medicine. Ithaca, NY
14853.
whole cell contaminants and vesicles were collected by high speed
centrifugation (30,000 x g-, l h; 4°C).Vesicles were washed 3 times in
PBS-CM containing 0.2 mM phenylmethylsulfonyl fluoride and were
immediately used in our experiments.
3The abbreviations used are: Con A. concanavalin A; WGA, wheat germ
agglutinin: FACS. fluorescein-activated cell sorter; mAb, monoclonal antibody;
PBS, phosphate-buffered saline; PBS-CM, phosphate-buffered saline containing
1 mrn CaCl¡and 0.5 mM MgCI2.
394
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
LUNG ENDOTHELIAL
MEMBRANE VESICLES BIND LUNG METASTATIC CELLS
Lectin-induced Vesicle Agglutination. Agglutination experiments
were performed with the following lectins: Ricinus communis agglutinili
I, peanut agglutinin, WGA and Con A (Vector Laboratories Inc.,
Burlingame CA). Twenty-^1 aliquots of vesicles (2-5 fig protein) were
incubated with an equal volume of lectin at 37°Cfor 15 min at a final
lectin concentration of 100 Mg/m'- The preparations were examined by
phase contrast microscopy.
Flow Cytometric Analysis. Endothelial membrane vesicles were la
beled with the membrane-permeable, pH-sensitive, fluorescent dye
2',7'-bis (carboxyethyl)-S (and -6)carboxyfluorescein acetoxymethyl es
ter (Molecular Probes, Eugene, OR). Vesicles were incubated in PBS
containing 7.5 pM 2',7'-bis(carboxyethyl)-5 (and -6)carboxylfluorescein
acetoxymethyl ester for l h at 37°C.After washing in PBS, vesicles
membrane vesicles. Pelleted vesicles were processed in 400-/il plastic
centrifuge tubes, essentially as described above, except that the aldehyde
fixation was for only 1 h. For thin sectioning, the narrow Araldite
plastic blocks containing the vesicles were reembedded into conven
tional flat embedding molds, then cut, stained, and analyzed as dis
cussed above.
Scanning Electron Microscopy of Immunogold-labeled Vesicles. Hl ft
FIO melanoma cells were grown on Thermanox coverslips (Lux Sci
entific Corp., Thousand Oaks, CA) placed into wells of 96-well plates,
and a binding assay was performed with lung-derived endothelial cell
membrane vesicles as described above. Tumor cells with bound endo
thelial cell vesicles were incubated with a mAb that selectively recog
nized lumenal membranes of rat lung capillary endothelial cells.4 After
were counted and sized in a FACS (Coulter Epics 752, Coulter Elec
tronics, Hialeah, FL) relative to a size standard of lO-pm latex beads.
To determine whether isolated endothelial cell membrane vesicles
were representative of the microvascular endothelium of the perfused
organ, vesicles were incubated for l h at 4°Cwith a mAb that selectively
bound rat lung microvascular endothelium.4 Vesicles were washed
l h at room temperature, wells were washed 4 times with RPMI 1640,
then incubated for 30 min with goat anti-mouse IgG conjugated to 30
nm gold particles (Vector Laboratories, Burlingame, CA) diluted 1:10
in RPMI 1640 containing 10% heat-inactivated normal goat serum.
Wells were washed 3 times with medium and once with 100 mM
cacodylate buffer. Cells were fixed with 2% glutaraldehyde in 100 mM
thoroughly in PBS, containing 0.4% bovine serum albumin, then la
cacodylate buffer, pH 7.3. dehydrated in graded ethanol solutions, then
beled with fluorescein isothiocyanate-conjugated goat anti-mouse IgG.
critical point dried in a Polaron Jumbo critical point dryer (Polaron,
Washed vesicles were analyzed by FACS, determining the ratio of Watford, England) (8). Coverslips containing the critical point dried
cells were attached to metal stubs with silver paint and were sputter
labeled versus nonlabeled vesicles.
lodination of Endothelial Cell Membrane Vesicles. Vesicle-associated
coated with a thin layer of carbon in an Edwards evaporator (Manor
membrane proteins were labeled by lactoperoxidase-catalyzed iodinaRoyal, Crawley, Sussex, England). Specimens were examined in a Jeol
tion, essentially as described by Soule et al. (29). Briefly, 200 ^1 of 200
SEM 35CF scanning electron microscope, using both secondary and
mM phosphate buffer, pH 7.3, containing 0.2 mCi Na'25I, 200 p\ of backscatter electron modes. Vesicle binding was counted on 200 B16Enzymobead reagent (Bio-Rad, Richmond, CA), and 100 ¡i\of 1% ß- F10 and 100 B16-FO melanoma cells.
D-glucose were added to 500 ^1 (50 ^g protein/ml) of endothelial cell
Lung Colony Assay. Syngeneic animals were given injections i.v. of
1 x IO5 tumor cells via the lateral tail vein in 0.2 ml of PBS (10
membrane vesicles in PBS-CM. Vesicles were incubated in this mixture
for 30 min at room temperature. The reaction was stopped by removing
animals/cell line). Animals were killed 21 days later by CO2 gassing
the Enzymobeads by centrifugation (100 x g; 10 min; 4°C).Vesicles
and the number of lung colonies was counted either directly under a
were washed 4 times in PBS-CM to remove all unbound 125I.
dissecting microscope (B16-F10, B16-FO) (26) or was determined in
Tumor Cell/Vesicle Binding Assay. Variable numbers of tumor cells triplicate, complete transverse sections of the lungs (R323OAC-MET,
were seeded into wells of 96-well tissue culture plates (Becton Dickin
R323OAC-LR) (28). The number of tumor colonies formed per cell
son, Lincoln Park, NJ) so that subconfluent monolayers of equal tumor
line were expressed as the median and range.
cell numbers and surface area were present after 48 h of incubation at
37°C.Tumor cells were then washed with RPMI 1640 and nonspecific
binding sites were blocked by incubation with 0.4% bovine serum
albumin in RPMI 1640 for 30 min at 37°C.Forty p\ of vesicle suspen
RESULTS
sion in RPMI 1640 (approximately 100,000 cpm) were added to each
of three wells and the plates were centrifuged at 200 x g for 5 min at
4°C.After 30 min of incubation at 37°C,cells were washed 3 times
In Vivo Preparation of Endothelial Cell Vesicles. In situ
perfusion with 100 mM formaldehyde and 2 mivi dithiothreitol
in PBS-CM during a 4-h period at a flow rate of 250 ¡A/mm
preserved the normal ultrastructure of rat lungs and legs. The
endothelium remained continuous and the junctional com
plexes normally interconnecting neighboring endothelial cells
were intact (Fig. 1). There was no recognizable interstitial
edema, despite the absence of oncotically active components
(e.g., protein, dextran) in the perfusion fluid. Trypan blue added
to test perfusates and instilled into the tissue at the standard
flow rate of 250 ¿il/minpermeated the most peripheral tissue
zones of the infused organs, attesting to a complete organ
perfusion.
The number of lumenal, outside-out endothelial cell mem
brane vesicles harvested during a 4-h perfusion period with
formaldehyde was approximately 3 x IO6vesicles/rat lung and
1 x IO7 vesicles/rat hind leg as determined by FACS analysis.
with medium, solubilized with 1% sodium dodecyl sulfate and counted
in a gamma counter. Data are presented as relative percentages of the
total cpm in 40 M'of vesicle suspension, setting binding of lung-derived
endothelial cell vesicles to the high lung metastatic tumor variants
(B16F10, R323OAC-MET) as 100%
Adhesion assays were also performed with neuraminidase-pretreated
vesicles (0.5 unit/ml; 30 min at 37°C;Sigma Type X, affinity purified;
Sigma Chemical Co., St. Louis, MO). The neuraminidase solution used
in our experiments was shown to contain a single, major peak when
run on a reverse-phase high-pressure liquid chromatography column.
This preparation was demonstrated to be free of other glycosidases and
proteases (30). Vesicles were washed 3 times with cold PBS-CM prior
to their use in a tumor cell/vesicle adhesion assay. Binding intensity
was expressed relative to total cpm in 40 ¿¿1
of vesicle suspension.
Transmission Electron Microscopy. In order to assure that the coher
ence of the lung vasculature was preserved after the 4-h perfusion
period, lungs were dissected into 1-mm3 tissue cubes and fixed with a
mixture of 2% paraformaldehyde and 2.5% glutaraldehyde in 100 mM
cacodylate buffer, pH 7.3, for 24 h at 4°C.Tissues were postfixed with
1% OsO4, dehydrated in graded ethanol solutions, and embedded in a
LX112/Araldite mixture. One-/jm-thick sections were stained with
toluidine blue for light microscopy. Ultrathin sections were stained
with uranyl acetate and lead citrate and examined in a Phillips EM301 electron microscope.
Ultrastructural analysis was also performed on isolated endothelial
4 D. Zhu et al., manuscript in preparation.
Vesicles originated primarily from the microvascular endothe
lium which accounted for more than 90% of the lumenal surface
area of the vasculature in each of the perfused organs. Given a
total lumenal surface area of 0.34 m2 of the rat lung microvasculature (31), approximately 900 vesicles were released from 1
cm2 of endothelial cell lumenal surface. The vesicle diameter
measured 0.47 ±29 ^m (SD) (Fig. 2). Ultrastructurally, mem
brane vesicles were spherical, devoid of cellular organdÃ-es, and
delimited by characteristic plasma membrane (Fig. 3). Vesicles
were true representatives of the microvascular lumenal mem395
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
LUNG ENDOTHELIAL
MEMBRANE VESICLES BIND LUNG METASTAT1C CELLS
Fig. I. Transmission electron microscopy of perfused lung details structural
integrity of cndothclium and intact tight junclional complexes (arrow), x 11.360.
w
1000
Fig. .1. L'ltrastructurc of the large-size fraction of cndothelial cell membrane
vesicles: large vesicles (approximately 2 ^m in diameter) are interdispersed with
clusters of small vesicles (arrow). Vesicles are devoid of cellular organdÃ-es and
contain electron-dense flocculent material condensed along the inner surface of
the plasma membrane, x 5440. Inserì,
typical trilammellar structure of the plasma
membrane, x 159.600.
TÃ-
i
>
120
500
-
110
100
90
(jim)
1.05
.7Õ
Fig. 2. Size distribution of lung-dcmed cndothelial cell membrane vesicles as
determined by FACS analysis: vesicles are extremely heterogenous in size. The
average vesicle diameter measures 0.45 ±0.27 ^m (SD).
.15
O
80
I
w
S
m
60
40
30
brane as 90% of the vesicles stained positively with a mAb that
20
selectively bound to the lumenal membranes of lung capillary
10
endothelial cells.
O
R323OAC- R323OACB16FO
B16F10
The protein content of an average lung endothelial vesicle
MET
LR
preparation ranged from 90 to 140 ¿ig
and that of a leg prepa
Fig. 4. Preferential attachment of lung-derived cndothelial cell membrane
ration from 200 to 300 ng. The higher amount of protein
vesicles to lung-metastatic tumor cells: murine HId I III melanoma cells and rat
R323OAC-MET mammary carcinoma cells both selected for high lung coloni
recovered from leg perfusates reflects an increased endothelial
zation, bind significantly more lung-derived endothelial vesicles (O) than their
cell surface area as well as a minor contamination with erythlow or nonmetastatic counterparts (BI6-FO and R323OAC-LR). In contrast, none
rocytes. Vesicles from both lung and leg could be agglutinated
of the four tumor cell lines showed any binding preferences for leg-derived
with Ricinus communia agglutinin I, peanut agglutinin, VVGA, endothelial cell vesicles (D). Data are expressed as relative percentages of adhesion
with the vesicle-binding capacities of B16-F10 and R3234OAC-MET set at 100%.
and Con A (data not shown).
Columns, mean of three separate experiments; bars, SD. Vesicle binding to low
Endothelial Vesicle Binding to Tumor Cells. Tumor cells with and high metastatic tumor cells are compared by Student's t test: B16-FIO versus
B16-FO: R.12.1OAC-MET versus R323OAC-LR: />< 0.01.
different lung metastatic potential were tested for their capacity
to bind lung- and leg-derived endothelial cell membrane vesi
cles. Our data show that tumor cells selected in vivo for high bound similar numbers of leg endothelial membrane vesicles.
These numbers were slightly lower but statistically no different
lung colonization (Fig. 4) bound significantly more lung endo
thelial vesicles than their low (B16-FO) or nonmetastatic
from the number of lung endothelial membrane vesicles binding
(R323OAC-LR) counterparts. In contrast, all four cell lines to low or nonmetastatic tumor cells. Vesicle binding results
396
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
LUNG ENDOTHELIAL
MEMBRANE VESICLES BIND LUNG METASTATIC CELLS
correlated well with the ability of these tumor cell lines to
produce lung colonies after tail vein inoculation (Table 1).
Preferential binding of lung endothelial vesicles to B16-F10
cells was abolished by prior incubation of membrane vesicles
with neuraminidase (Fig. 5).
Scanning Electron Microscopy. Scanning electron microscopy
confirmed the above binding data, showing higher numbers of
B16-F10 melanoma cells with bound lung-derived endothelial
vesicles than B16-FO. Seventy-two % of B16-F10 cells bound
no vesicles, 23% bound 1-3 vesicles (Fig. 6A), and 5% bound
more than 3 vesicles. Corresponding values for B16-FO cells
were 85, 12, and 3%, respectively. Vesicles bound most fre
quently to the marginal zones of the melanoma cells where they
were sometimes associated with microvilli. Endothelial vesicles
were differentiated from B16-F10 membrane cell blebs by stain
ing with a mAb that monospecifically interacted with lumenal
membranes of rat lung capillary endothelium (Fig. 6, B and C).
ferritin hydrazide, maintain on their ectodomains glycoconjugates bearing monosaccharides such as A'-acetylneuraminic
acid, /3-jV-acetylgIucosamine, and /i-D-galactose, express 5'-nu-
cleotidase activity, and show identical electrophoretic profiles
of in situ labeled surface proteins (25). Despite these favorable
comparisons, caution must be exercised in functional studies
involving endothelial cell membrane vesicles. The formaldehyde
solution used to induce vesicle formation may cause changes in
the structure of vesicle membrane proteins due, in part, to its
well-known protein cross-linking activity (32). Since formalde
hyde had no effect on the expected tumor cell-binding charac
teristics of lung- and leg-derived endothelial membrane vesicles,
it was concluded that the presumed tumor cell-binding domain
of vesicle adhesion molecules was either unaffected or, if altered
by the aldehyde, reversed during the extensive washing of
vesicles in neutral buffer solutions. Alternatively, vesicle bind
ing may have been mediated by as yet undetermined, formal
dehyde-resistant carbohydrate moieties.
The results presented here complement several other recent
DISCUSSION
studies on the role of endothelial cells in organ preference of
metastasis. Pauli and Lee (8) showed that organotypic tumor
Progress toward understanding the molecular basis of tumor
cell-endothelial cell interactions in secondary organs has been cell adhesion molecules could be induced in large vessel endo
thelial cells (e.g., BAEC), when these cells were grown on organtroubled by problems associated with the isolation and culture
derived extracellular matrices. Using these modulated BAEC
of phenotypically representative populations of microvascular
endothelial cells (17-23). In this paper we describe a new in a tumor attachment assay, they found that tumor cells which
adhesion assay that is not contingent upon the culture of metastasized to a given organ, had a significantly higher binding
affinity for BAEC grown on extracellular matrix of the metas
microvascular endothelial cells but, instead, relies on the iso
tasized organ than they had for BAEC grown on the matrix of
lation of lumenal endothelial membrane vesicles by perfusing
any other organ not colonized by these tumor cells. Auerbach
target organs with a low strength formaldehyde solution. Vesi
et al. (4) showed selective attachment of hepatoma cells to
cles obtained by this method have been reported to be true,
sinusoidal endothelial cells, of glioma cells to brain-derived
biochemical representatives of the endothelial cell lumenal
capillary endothelium, of teratoma cells to ovary-derived micromembrane (24, 25). For example, similar to in situ aortal
endothelial cells, their vesicle surfaces stain intensely with vascular endothelium, and of mammary tumor cells to lym
ruthenium red, bind in a normal pattern cationized ferritin and phatic endothelium. Comparably, Schirrmacher (33) and Roos
et al. (34) provided evidence for the propensity of liver-metastatic tumor cells to bind to hepatic sinusoidal endothelium.
Table 1 Formation of experimental métastases21 days after tail vein injection
into syngeneic hosts
Nicolson et al. (35) postulated that differences in endothelial
of cells
of lung
cell adhesion were primarily dependent upon quantitative, and
colonies"4
injectedI
incidence6/7
less upon qualitative differences between various types of mi
x 10*
B16-FO
(0-20)
crovascular
endothelial cells, and suggested that tumor cells
1
x
10'
B16-F10
103(82->150)
7/7
might use similar molecules to those involved in cell-cell inter
1 x 10!
R3230AC-MET
204(176-231)
10/10
1 x 10*Tumor
R3230AC-LRNo.
0/10No.
actions in the microvasculature (e.g., leukocyte/endothelial cell
1Median (range).
interactions). This hypothesis was supported by recent reports
of Rice and Bevilacqua (36, 37), showing that the lymphokineinducible endothelial cell adhesion molecule, INCAM-110,
120110served as receptor for melanoma cells and ELAM-1 as receptor
100for HT-29 colon carcinoma cells.
90 Preferential binding of lung-derived endothelial vesicles to
80 B16-F10 was abolished by prior incubation of the vesicles with
70 neuraminidase. This result complements the finding that neur
60 aminidase treatment of peripheral lymph node sections as well
50 as systemic application of neuraminidase reduces binding of
40 lymphocytes to high endothelial venules (HEV) (30, 38). Fur
30 thermore it implies that sialic acid is an important component
20
of the recognition determinant or exerts a modulating effect on
10HEV receptors, e.g., changing its binding avidity (39). Con
o
B16F10versely, there is evidence that negative charges imparted by
B16FOB16F10
NEURAMIN.
NEURAMIN.
sialic acid residues influence recognition of peripheral lymph
Fig. 5. Neuraminidase treatment of vesicles abolishes preferential binding of
nodes by lymphocytes (40). These data suggest that sialic acid
lung-derived endothelial vesicles for lung metastatic B16-F10 melanoma cells.
Binding intensities of neuraminidase-treated leg (G) and lung endothelial vesicles
does confer a modulatory effect, albeit nonspecifically, on both
<M)for high and low metastatic B16 melanoma cells are statistically similar but
lymphocyte and tumor cell binding to microvascular endothelial
vary significantly from the binding of untreated lung-derived endothelial vesicles
to B16-F10 cells (P < 0.01). using Student's / test for unpaired data. Columns,
cells. Since sialic acid is a ubiquitous carbohydrate residue, it
mean of two separate experiments; bars, SE.
seems improbable that it is part of a specific recognition system
397
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
LUNG ENDOTHELIAL
MEMBRANE VESICLES BIND LUNG METASTATIC CELLS
Fig. 6. Adherence of lung-derived endothelial cell membrane vesicles to B16-F10 melanoma cells: A, three lung-derived endothelial cell membrane vesicles (f7)
adhere to the periphery of a B16-F10 melanoma cell: x 7700; B and C distinguish endothelial cell vesicles from tumor cell blebs, vesicles are stained with an
endothelial vesicle-specific antibody and visualized with secondary antibody coupled to 30 nm gold particles. A vesicle with gold particles bound (arrow) is depicted
in conventional secondary (B) and backscatter (C) electron modes, x 13,200.
despite the potential diversity imparted by linkage number and
associated residues. However, there is evidence that specific
heterotypic sugar-sugar cohesion between opposing glucosphingolipids on the surface of lymphoma and melanoma cells,
including sialosyllactosylceramide (Gsn) on the melanoma cell,
are influential in cell-cell binding (41).
The data presented here and those previously published by
our laboratory and others (4, 8, 10, 33-35) indicate that endo
thelial cell adhesion molecules are sets of molecules that are
organotypically expressed and restricted to defined segments of
the vasculature. They are recognized by specific tumor cell
types, causing their arrest in the microcirculation of selected,
secondary organs and, thus, contributing to site-specific metas
tasis.5 These molecules are down-regulated or lost as endothelial
cells are removed from their microenvironment and are cultured
on artificial substrates (8). In studies on tumor cell endothelial
cell interactions, problems associated with the culture of microvascular endothelial cells and the maintenance of organ-specific
phenotypes may be eluded by the use of endothelial cell mem
brane vesicles freshly prepared from host organs by perfusion.
ACKNOWLEDGMENTS
The authors wish to thank Dr. Andrew Yen for his advice and
assistance in the flow cytometric analysis, and Ursula Nanna, Mary
Jane /anelli, and Ed Dougherty for their expert technical support.
REFERENCES
1. Weiss, L.. and Dimitrov. D. S. Mechanical aspects of the lungs as cancer
cell-killing organs during hematogenous metastasis. J. Theor. Biol., 121:
307-322. 1986.
2. Weiss, L., Orr. F. \V.. and Honn, K. V. Interactions between cancer cells and
the microvasculature: a rate regulator for metastasis. Clin. Exp. Metastasis,
7: 127-167. 1989.
*C. F. Cheng et al., manuscript in preparation.
3. Alby, L.. and Auerbach, R. Differential adhesion of tumor cells to capillary
endothelial cells in vitro. Proc. Nati. Acad. Sci. USA, SI: 5739-5743, 1984.
4. Auerbach, R.. Lu. W. C.. Pardon, E., Gumkowski, F.. Kaminska. G-, and
Kaminska. M. Specificity of adhesion between murine tumor cells and
capillary endothelium: an in vitro correlate of preferential metastasis in vivo.
Cancer Res., 47: 1492-1496, 1987.
5. Auerbach. R. Patterns of tumor metastasis: organ selectivity in the spread of
cancer cells. Lab. Invest.. 5«:361-364. 1988.
6. Nicolson. G. L. Organ Specificity of tumor metastasis: role of preferential
adhesion, invasion and growth of malignant cells at specific secondary sites.
Cancer Metastasis Rev., 7: 143-188, 1988.
7. Nicolson, G. L. Cancer metastasis: tumor cell and host organ properties
important in metastasis to specific secondary sites. Biochim. Biophys. Acta,
948: 175-224, 1988.
8. Pauli, B.U., and Lee. C. L. Organ preference of metastasis: the role oforganspecifically modulated endothelial cells. Lab. Invest., 58: 379-387, 1988.
9. Sher, B. T.. Bargatze. R.. Holzmann. B., Gallatin. W. M., Mathews, D., Wu,
N., Picker. L.. Butcher. E. C., and Weissman, 1. L. Homing receptors and
metastasis. Adv. Cancer Res.. 51: 361-390. 1988.
10. Pauli, B. U., Johnson. R. C.. and El-Sabban. M. Organotypic endothelial
cell surface molecules mediate organ preference of metastasis. In: N. Simionescu. and M. Simionescu, (eds.), Endothelial Cell Dysfunction, New York;
Plenum Press, in press. 1990.
11. Pauli, B. U.. Augustin-Voss. H. G.. El-Sabban, M. E., Johnson, R. C. and
Hammer. D. A. Organ preference of metastasis: the role of endothelial cell
adhesion molecules. Cancer Metastasis Rev., in press, 1990.
12. Stoolman. L. M. Adhesion molecules controlling lymphocyte migration. Cell,
56:907-910. 1989.
13. Butcher, E. C. Cellular and molecular mechanisms that direct leukocyte
traffic. Am. J. Pathol., 136: 3-11, 1990.
14. Folkman. J., Haudcnschild. C. C.. and Zetter, B. R. Long-term culture of
capillary endothelial cells. Proc. Nati. Acad. Sci. USA, 76: 5217-5221, 1979.
15. Bowman, P. D.. Betz. A. L., Ar, D., Wolinsky, J. S., Penny, J. B., Shivers.
R. R., and Goldstein, G. W. Primary culture of capillary endothelial from
rat brain. In Vitro (Rockville), 17: 353-362. 1981.
16. Carson. M. P.. and Haudenschild. C. C. Microvascular endothelium and
pericytes: high yield, low passage cultures. In Vitro (Rockville), 22: 344-354,
1986.
17. Borsum. T., Hagen. T.. Henrickson, T.. and Carlander, B. Alteration in the
protein composition and cell surface structure of human endothelial cells
during growth in primary culture. Atherosclerosis. 44: 367-378. 1982.
18. McAuslan. B. R.. Hannan. G. N., and Reilly, W. Signals causing change in
morhpological phenotype growth mode, and gene expression of vascular
endothelial cells. J. Cell. Physiol.. 112: 96-106. 1982.
19. Rogers, K., and Kalnmus, V. Comparison of the cytoskeleton in aortic
endothelial cells in situ and in vitro. Lab. Invest., 49: 650-654, 1983.
20. deBono. D. P.. and Green. C. The adhesion of different cell types to vascular
398
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
LUNG ENDOTHELIAL
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
MEMBRANE VESICLES BIND LUNG METASTATIC CELLS
endothelium: affects of culture density and age. Br. J. Exp. Med., 65: 145154, 1984.
Bartlet, C. P.. Heale, G., and Levene, C. I. Some factors regulating collagen
polymorphism in cultured porcine and bovine aortic endothelium. Athero
sclerosis, 54: 301-309, 1985.
Hormia. M. Surface glycoproteins of cultured human umbilical vein endothelial cells. Cell Biol. Int. Rep., 9: 637-646, 1985.
Ik-Mom. P. N., and Nicolson, G. L. Differential expression of cell surface
glycoproteins on various organ-derived microvascular endothelia and endothelial cell cultures. J. Cell. Physiol., 136: 398-410, 1988.
Scott, R. E. Plasma membrane vesiculation: a new technique for isolation of
plasma membranes. Science (Washington DC), ¡94:743-745, 1976.
Moldovan, N. I., Radu, A. N., and Simionescu, N. Endothelial cell plasma
membrane obtained by chemically induced vesiculation. Exp. Cell Res., 170:
499-510, 1987.
Fidler, I. J. Selection of successive tumour lines for metastasis. Nature
(Lond.), 242: 148-149, 1973.
Ng, R., Kellen, J. A., and Wong, A. C. H. Plasminogen activators as markers
of tumor colonization potential. Invasion Metastasis. 3: 243-248, 1983.
Pauli. B. U., Kellen. J. A., and Ng. R. Correlation of fibrinolytic activity with
invasion and metastasis of R323OAC rat mammary carcinoma cell line.
Invasion Metastasis 7: 158-171, 1987.
Soule, H. R., Langford, R. E., and Butel, J. S. Detection of simian virus 40
surface-associated large tumor antigen by enzyme catalyzed radioiodination.
Int. J. Cancer, 29: 337-344. 1982.
Rosen, S. D.. Singer, M. S., and Yednock. T. A. Involvement of sialic acid
on endothelial cells in organ-specific lymphocyte recirculation. Science
(Washington DC), 228: 1005-1007, 1985.
Weibel, E. R. Morphometry and lung models. In: E. R. Weibel, and H. Elias,
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
(eds.). Quantitative Methods in Morphology, pp. 253-267. Berlin: SpringerVerlag, 1967.
Means, G. E., and Feeney, R. E. Chemical Modification of Proteins, pp.
125-128. San Francisco, CA: Holden-Day, Inc., 1971.
Schirrmacher. V. Cancer metastasis: experimental approaches, theoretical
concepts and impacts for treatment strategies. Adv. Cancer Res., 43; 1-73,
1985.
Roos, E., Middelkoop. O. P.. and van de Pavert, I. V. Interactions between
lymphoid tumor cells and isolated liver endothelial cells. J. Nati. Cancer
Inst.. 72: 1173-1180. 1984.
Nicolson. G. L., Belloni. P. N., Tressler. R. J.. Dulski. K., Inoue, T., and
Cavanough. P. G. Adhesive, invasive and growth properties of selected
metastatic variants of a murine large cell lymphoma. Invasion Metastasis, 9:
102-116. 1989.
Rice. G. E., Gimbrone, M. A., and Bevilacqua. M. P. Tumor cell-endothelial
cell interactions. Increased adhesion of human melanoma cells to activated
vascular endothelium. Am. J. Pathol., 133: 204-210. 1988.
Rice, G. E., and Bevilacqua, M. P. An inducible endothelial cell surface
glycoprotein mediates melanoma cell adhesion. Science (Washington DC).
246: 1303-1306. 1989.
Rosen. S. D., Chi, S., True, D. D.. Singer, M. S., and Yednock, T. A.
Intravenously injected sialidase inactivates attachment sites for lymphocytes
on high endothelial cell venules. J. Immunol.. 142: 1895-1902, 1989.
Yednock, T. A., and Rosen, S. D. Lymphocyte homing. Adv. Immunol.. 44:
313-378. 1989.
Stoolman. L. M.. and Rosen. S. D. Possible role for cell-surface carbohydratebinding molecules in lymphocyte recirculation. J. Cell Biol., 96: 722-729,
1983.
Kojima, N.. and Hakomori, S. Specific interaction between gangliotriaosylceramide (Gg3) and Sialosyllactosylccramide (GMJ) as a basis for specific
cellular recognition between lymphoma and melanoma cells. J. Biol. Chem..
264:20159-20162, 1989.
399
Downloaded from cancerres.aacrjournals.org on July 31, 2017. © 1991 American Association for Cancer Research.
Endothelial Cell Membrane Vesicles in the Study of Organ
Preference of Metastasis
Robert C. Johnson, Hellmut G. Augustin-Voss, Duzhang Zhu, et al.
Cancer Res 1991;51:394-399.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/51/1/394
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 July 31, 2017. © 1991 American Association for Cancer Research.