Zhou, Y. et al. Human cytotrophoblasts adopt a vascular phenotype as they differentiate. A strategy for successful endovascular invasion? J. Clin. Invest. 99, 2139-2151

Department of Stomatology, University of California San Francisco, 94143-0512, USA.
Journal of Clinical Investigation (Impact Factor: 13.22). 06/1997; 99(9):2139-51. DOI: 10.1172/JCI119387
Source: PubMed
ABSTRACT
Establishment of the human placenta requires that fetal cytotrophoblast stem cells in anchoring chorionic villi become invasive. These cytotrophoblasts aggregate into cell columns and invade both the uterine interstitium and vasculature, anchoring the fetus to the mother and establishing blood flow to the placenta. Cytotrophoblasts colonizing spiral arterioles replace maternal endothelium as far as the first third of the myometrium. We show here that differentiating cytotrophoblasts transform their adhesion receptor phenotype so as to resemble the endothelial cells they replace. Cytotrophoblasts in cell columns show reduced E-cadherin staining and express VE-(endothelial) cadherin, platelet-endothelial adhesion molecule-1, vascular endothelial adhesion molecule-1, and alpha-4-integrins. Cytotrophoblasts in the uterine interstitium and maternal vasculature continue to express these receptors, and, like endothelial cells during angiogenesis, also stain for alphaVbeta3. In functional studies, alphaVbeta3 and VE-cadherin enhance, while E-cadherin restrains, cytotrophoblast invasiveness. Cytotrophoblasts expressing alpha4 integrins bound immobilized VCAM-1 in vitro, suggesting that this receptor-pair could mediate cytotrophoblast-endothelium or cytotrophoblast-cytotrophoblast interactions in vivo, during endovascular invasion. In the pregnancy disorder preeclampsia, in which endovascular invasion remains superficial, cytotrophoblasts fail to express most of these endothelial markers (Zhou et al., 1997. J. Clin. Invest. 99:2152-2164.), suggesting that this adhesion phenotype switch is required for successful endovascular invasion and normal placentation.

Full-text

Available from: Mary J Janatpour
Cytotrophoblasts Mimic a Vascular Phenotype
2139
J. Clin. Invest.
© The American Society for Clinical Investigation, Inc.
0021-9738/97/05/2139/13 $2.00
Volume 99, Number 9, May 1997, 2139–2151
Human Cytotrophoblasts Adopt a Vascular Phenotype as They Differentiate
A Strategy for Successful Endovascular Invasion?
Yan Zhou,* Susan J. Fisher,*
‡§
i
Mary Janatpour,* Olga Genbacev,* Elisabetta Dejana,
Margaret Wheelock,**
and Caroline H. Damsky*
i
Department of
*
Stomatology, Department of
Obstetrics, Gynecology and Reproductive Sciences, Department of
§
Pharmaceutical
Chemistry, and Department of
i
Anatomy, University of California San Francisco, San Francisco, California 94143-0512;
Istituto di
Richerche Farmacologiche Mario Negri, Milan, Italy; and
**
Department of Biology, University of Toledo, Toledo, Ohio 43606-3390
Abstract
Establishment of the human placenta requires that fetal cy-
totrophoblast stem cells in anchoring chorionic villi become
invasive. These cytotrophoblasts aggregate into cell col-
umns and invade both the uterine interstitium and vascula-
ture, anchoring the fetus to the mother and establishing
blood flow to the placenta. Cytotrophoblasts colonizing spi-
ral arterioles replace maternal endothelium as far as the
first third of the myometrium. We show here that differenti-
ating cytotrophoblasts transform their adhesion receptor
phenotype so as to resemble the endothelial cells they re-
place. Cytotrophoblasts in cell columns show reduced E-cad-
herin staining and express VE-(endothelial) cadherin, plate-
let-endothelial adhesion molecule-1, vascular endothelial
adhesion molecule-1, and
a
4-integrins. Cytotrophoblasts in
the uterine interstitium and maternal vasculature continue
to express these receptors, and, like endothelial cells during
angiogenesis, also stain for
a
V
b
3. In functional studies,
a
V
b
3 and VE-cadherin enhance, while E-cadherin re-
strains, cytotrophoblast invasiveness. Cytotrophoblasts ex-
pressing
a
4 integrins bound immobilized VCAM-1 in vitro,
suggesting that this receptor-pair could mediate cytotro-
phoblast–endothelium or cytotrophoblast–cytotrophoblast
interactions in vivo, during endovascular invasion. In the
pregnancy disorder preeclampsia, in which endovascular
invasion remains superficial, cytotrophoblasts fail to ex-
press most of these endothelial markers (Zhou et al., 1997.
J. Clin. Invest.
99:2152–2164.), suggesting that this adhe-
sion phenotype switch is required for successful endovascu-
lar invasion and normal placentation. (
J. Clin. Invest.
1997.
99:2139–2151.) Key words: placentation
trophoblast
en-
dothelium
integrins
cadherins
Introduction
Human placental development depends critically on the differ-
entiation of the placenta’s specialized epithelial cells, cytotro-
phoblasts (CTB).
1
Two differentiation pathways exist (see Fig.
1). In one, CTB remain in the fetal compartment and fuse to
form multinucleate syncytiotrophoblasts that cover the float-
ing chorionic villi. These villi, which are in direct contact with
maternal blood in the intervillous space, perform nutrient and
gas exchange for the fetus. In the other pathway, the focus of
this paper, a subset of CTB in anchoring chorionic villi aggre-
gate into cell columns that attach to the uterine wall. From
there, CTB invade the uterine wall (interstitial invasion, see
Fig. 1
A
) and its blood vessels (endovascular invasion, see Fig.
1
B
) as far as the first third of the myometrium. Anchoring villi
thus attach the fetus to the uterus and establish the flow of ox-
ygenated maternal blood to the intervillous space (1–4, re-
viewed in 5, 6).
CTB differentiation along the invasive pathway is a com-
plex, multi-step process. CTB within cell columns lose the abil-
ity to divide (7; Genbacev and Fisher, unpublished data). As
they invade the uterine wall, CTB upregulate expression of
MMP-9, the 92-kD matrix metalloproteinase (8), HLA-G, a
trophoblast-specific HLA class I molecule that is likely to be
important in avoiding rejection of the conceptus by the mater-
nal immune system (9, 10), and specific hormones, including
human placental lactogen (hPL).
Relationships of differentiating CTB with extracellular ma-
trix (ECM) and with other cells also change. For example,
CTB intricately regulate expression of ECM ligands and their
integrin receptors. CTB villus stem cells express the
a
6
b
4 inte-
grin laminin receptor strongly. As they differentiate, CTB
downregulate
a
6
b
4 and sequentially upregulate expression of
fibronectin and the
a
5
b
1 fibronectin receptor in cell columns,
and the
a
1
b
1 laminin/collagen receptor in the uterine wall (11,
12). CTB stem cells can be isolated from placental tissue and
when plated on Matrigel, they recapitulate their differentia-
tion program. Although we do not understand the signals that
promote this differentiation program (see reference 13), we
can use function-perturbing antibodies in conjunction with this
in vitro model of invasion to determine which molecules are
responsible for mediating the changes in migration and inva-
sion. We have shown that interactions of
a
5
b
1 with fibronectin
restrain invasion, and that interactions of
a
1
b
1 with collagen
type IV and laminin promote CTB invasion in this system (13).
We hypothesize that CTB balance invasion-restraining and in-
vasion-promoting adhesion mechanisms as they differentiate,
Address correspondence to Caroline H. Damsky, HSW 604 UCSF,
513 Parnassus Ave., San Francisco, CA 94143-0512. Phone: 415-476-
8922; FAX: 415-502-7338; E-mail: damsky@cgl.ucsf.edu
Received for publication 18 September 1996 and accepted in re-
vised form 27 January 1997.
1.
Abbreviations used in this paper:
CTB, cytotrophoblasts; ECM, ex-
tracellular matrix; HUVEC, human umbilical vein endothelial cells;
MYO, myometrium; PECAM-1, platelet-endothelial cell adhesion
molecule-1; VEGF, vascular endothelial growth factor; VCAM-1,
vascular cell adhesion molecule-1.
Page 1
2140
Zhou et al.
which may help regulate the depth of CTB invasion into the
uterus (13).
The significance for normal placentation of this CTB dif-
ferentiation program is highlighted by the fact that in the preg-
nancy disorder preeclampsia, in which both interstitial and en-
dovascular invasion are abnormally shallow (14–17), CTB
show significant defects in differentiation. For example, levels
of MMP-9, HLA-G, and hPL in preeclamptic CTB are abnor-
mally low (17, 18; Lim et al.).
2
Furthermore, CTB in pre-
eclampsia increase expression of the
a
5
b
1 fibronectin recep-
tor, but fail both to upregulate
a
1
b
1 and to downregulate
a
6
b
4
(Lim et al., 19). Thus, they appear arrested in their differentia-
tion program and they express an ECM receptor phenotype
that may not be optimal for invasion to the appropriate depth.
In continuing to search for fundamental insights into CTB
differentiation during placentation, we have focused recently
on the virtually unique ability of CTB to invade maternal
blood vessels, to replace the maternal endothelium within the
spiral arterial segments as far as the myometrium (MYO), and
to remodel the muscular walls (tunica media) of these arteries.
We hypothesized that to accomplish this, differentiating CTB
stem cells must lose their epithelial phenotype and transform
their cell–cell adhesion molecule phenotype dramatically, both
to become invasive and to be able to interact with, and ulti-
mately mimic certain behaviors of, the endothelial cells they
displace. We therefore examined the phenotype of differenti-
ating CTB for loss of epithelial adhesion receptors and the on-
set of expression of adhesion receptors characteristic of endo-
thelium, or of leukocytes that are interacting with endothelium.
Our results document that CTB undergo a comprehensive
transformation of their adhesion molecule repertoire so as to
mimic that of endothelial cells. Furthermore, in functional
studies we show that the newly expressed adhesion receptors
contribute to enhanced motility and invasiveness of differenti-
ating CTB. Since CTB in preeclampsia are defective in endo-
vascular invasion and colonization (16, 17, 19) and do not exe-
cute the switch to a vascular adhesion phenotype (20), we
propose that this switch observed in normal CTB, which re-
sults in mimicry of vascular cells, is a fundamental requirement
for normal placentation in humans.
Methods
Cell culture and isolation of CTB.
JEG-3 choriocarcinoma cells were
cultured in Dulbecco’s MEM with 4.5 g/liter glucose (DME-high glu-
cose), containing 10% FBS. Cultures of human foreskin kerati-
nocytes and human umbilical vein endothelial cells (HUVEC) were
kindly provided by Dr. Randall Kramer and Dr. Kee-Hak Lim,
UCSF, San Francisco, CA, respectively.
Normal CTB were isolated from chorionic villi by established
procedures (8, 21). Briefly, placentas from normal uncomplicated
pregnancies were obtained immediately after early gestation termina-
tions (7–22 wk) or after delivery at term (34–40 wk). After initial pu-
rification of CTB on Percoll gradients, remaining leukocytes were re-
moved with an antibody to CD-45 coupled to magnetic beads.
Purified cells were used immediately or cultured on Matrigel-coated
substrates (Collaborative Biomedical Products, Bedford, MA) for
varying lengths of time in serum-free DME-high glucose, with 2%
Nutridoma (Boehringer Mannheim Biochemicals, Indianapolis, IN).
Normal CTB from 7–22 wk placentas differentiate over 24–72 h, as
shown previously by increased production of
a
1
b
1 (13, Lim et al.),
2
the MMP-9 metalloproteinase (8) and HLA-G (10), and they invade
Matrigel in an established invasion assay (8).
To assure that endothelial cells in the maternal decidua do not ex-
press epithelial antigens in common with CTB stem cells (in particu-
lar E-cadherin), cultures of primary decidual endothelium were gen-
erated as described by Gallery et al. (22). Briefly, 0.5 mm
3
pieces of
decidual tissue from placental bed biopsies were treated with trypsin
and pronase, and then squeezed with the blunt end of a scalpel blade.
The extruded cells were filtered through gauze and incubated on ice
with magnetic beads coated with the ulex europaeus-1 lectin (E.Y.
Laboratories, Inc., San Mateo, CA). Bound cells were plated on gela-
tin-coated petri dishes in DME-H21/Ham’s F-12 (1:1), with 20% FBS
supplemented with ITS (Collaborative Research, Inc., Lexington,
MA). Cells were grown to confluency and passaged three times be-
fore performing immunocytochemistry.
Antibodies.
Antibodies against adhesion receptors were ob-
tained from the following sources: Integrin
a
1, TS2/7, T Cell Sciences,
Inc., Cambridge, MA; integrin
a
4, HP1/2, Dr. R. Lobb, Biogen, Cam-
bridge, MA; integrin
a
V
b
3, complex-specific function perturbing
monoclonal antibody LM609, Dr. David Cheresh, Scripps Research
Foundation, La Jolla, CA; integrin
b
3, II344; integrin
a
V
b
5, com-
plex-specific function perturbing monoclonal antibody, P3G2, Dr.
Elizabeth Wayner, Fred Hutchinson Cancer Center, Seattle, WA; in-
tegrin
b
6, E7P6, Dr. Robert Pytela, UCSF, San Francisco, CA; E-selec-
tin, S25, Dr. M. Gimbrone, Brigham and Women’s Hospital, Boston,
MA; E-cadherin, E9, (23), for immunoblotting and staining, and poly-
clonal anti-GP-80, made against purified 80 kD fragment of E-cad-
herin (24), for function perturbation; VE-cadherin, BV6, BV9 and
TEA 1.31, Dr. E. Dejana (25, 26); P-cadherin, 6A9, Dr. M. Wheelock;
platelet-endothelial cell adhesion molecule-1 (PECAM-1), 390, Dr.
H.S. Baldwin, Univ. Pennsylvania, Philadelphia, PA; vascular cell ad-
hesion molecule-1 (VCAM-1), 11/26 for staining, Dr. T. Yednock,
(Athena Neurosciences, Inc., S. San Francisco, CA), and 4B9 for
function-perturbation, Dr. R. Lobb, Biogen; cytokeratin, 7D3 (rat
anti-human, 7D3;13) and K8.13 (mouse anti-human; Sigma Immu-
nochemicals, St. Louis, MO); human von Willebrand Factor, F/8/86
(DAKO A/S, Glostrup, Denmark).
Immunocytochemistry.
Chorionic villi with attached decidua were
dissected from placentas immediately after elective terminations or
delivery. Placental bed biopsy tissues were obtained from the site of
implantation immediately after cesarean delivery (see description of
informed consent procedures in 20). Tissues were processed for dou-
ble indirect immunocytochemistry as described previously (11, 19).
Briefly, tissues were fixed in 3% paraformaldehyde for 30 min, infil-
trated with 5–15% sucrose followed by OCT, and frozen in liquid ni-
trogen. Sections (5–7
m
m) were cut on a Hacker-Slee cryostat. Iso-
lated CTB plated on Matrigel-coated coverslips for varying times
were fixed in 3% paraformaldehyde for 10 min, and permeabilized
with cold methanol. Fixed tissue sections or cell cultures were stained
with a mixture of two primary antibodies (rat or mouse anti-human
cytokeratin to mark CTB, and an antibody of a different species ori-
gin against individual adhesion receptors of interest) for 1 h to over-
night, washed, and incubated with secondary antibodies conjugated
to fluorescein or rhodamine (Jackson ImmunoResearch Labs, Inc.,
West Grove, PA). The secondary antibodies were cross-absorbed
against nonimmune IgG of other species to eliminate cross-reactivity.
Samples were viewed with the Zeiss Axiophot Epifluorescence mi-
croscope equipped with filters to selectively view the rhodamine and
fluorescein images with no cross-contamination.
Cell extraction and immunoblotting.
Immunoblotting was carried
out to determine the steady state levels of cadherins present during
CTB differentiation. Freshly isolated CTB or CTB cultured on Matri-
gel-coated tissue culture wells were washed 2
3
with PBS and ex-
tracted with 200
m
l of lysis buffer (50 mM Tris buffer, pH 8.0, contain-
ing 0.1% SDS, 0.5% NP-40, 120 mM NaCl, 100
m
m PMSF, 20 U/ml
aprotinin, 10
m
g/ml leupeptin). Cell extracts were centrifuged at
2. Lim, K.-H., Y. Zhou, M. Janatpour, M. McMaster, K. Bass, S.-H.
Chun, and S.J. Fisher, manuscript submitted for publication.
Page 2
Cytotrophoblasts Mimic a Vascular Phenotype
2141
Figure 1. (A) Diagram of a longitudinal section of an anchoring chorionic villus (AV) at the fetal-maternal interface at z 10 wk gestational age.
The anchoring villus (AV) functions as a bridge between the fetal and maternal compartments, whereas floating villi (FV) are suspended in the
intervillus space and are bathed by maternal blood. CTB in AV (Zone I) form cell columns (Zones II & III). CTB then invade the uterine inter-
stitium (decidua and first third of the myometrium, (Zone IV) and maternal vasculature (Zone V), thereby anchoring the fetus to the mother and
accessing the maternal circulation. Zone designations mark areas in which CTB have distinct patterns of adhesion receptor expression as de-
scribed in the text and in reference 11. (B) Diagram of a spiral artery in which endovascular invasion is in progress (10–18 wk gestation). En-
dometrial and then myometrial segments of spiral arteries are modified progressively. In fully modified regions (a) the vessel diameter is large.
CTB are present in the lumen and occupy the entire surface of the vessel wall. A discrete muscular layer (tunica media) is not evident. (b) Par-
tially modified vessel segments. CTB and maternal endothelium occupy discrete regions of the vessel wall. In areas of intersection, CTB appear
to lie deep to the endothelium and in contact with the vessel wall (20). (c) Unmodified vessel segments in the myometrium. Vessel segments in
the superficial third of the myometrium will become modified when endovascular invasion reaches its fullest extent (by 22 wk), while deeper seg-
ments of the same artery will retain their normal structure.
Page 3
2142
Zhou et al.
12,000
g
for 5 min to remove insoluble material. Samples containing
equal amounts of protein were mixed with SDS sample buffer, ana-
lyzed by SDS-PAGE under reducing conditions, and transferred to
nitrocellulose. The nitrocellulose membranes were incubated with
primary antibodies (E9, rat anti-human E-cadherin; BV9, mouse anti-
human VE-cadherin, and 6A9, mouse anti-human P-cadherin) and
peroxidase-conjugated secondary antibodies by standard procedures.
Immune complexes were visualized using an enhanced chemilumi-
nescence procedure and Hyperfilm (Amersham Life Sciences-USB,
Arlington Heights, IL).
Invasion assays.
Invasion assays were conducted as described
previously (7, 8, 13). Briefly, isolated CTB (2.5
3
10
5
) were plated on
Transwell inserts (6.5 mm; Costar Corp., Cambridge, MA), contain-
ing polycarbonate filters with 8-
m
m pores, that had been coated with
Matrigel. After 48 h in the presence of 100
m
g/ml control IgG or func-
tion-perturbing antibodies against
a
V
b
3 (LM609: 50
m
g/ml), E-cad-
herin (E9: 50
m
g/ml), or VE-cadherin (BV6; 100
m
g/ml), the cultures
were fixed with paraformaldehyde, permeabilized with cold metha-
nol, and stained with the 7D3 antibody against human cytokeratin.
The filter inserts, with the CTB cultures, were excised with a scalpel
blade and mounted on polylysine-coated slides, with the underside of
the filter facing upward. The Matrigel-coated and underside surfaces
of the filters were distinguishable when viewed under the Zeiss Axio-
phot Epifluorescence microscope, and therefore cytokeratin-positive
Figure 2. CTB switch
the pattern of expres-
sion of aV-family inte-
grin receptors during
differentiation in vivo.
Sections of 2nd trimes-
ter (18–22 wk) placental
tissue were double
stained with anti-cyto-
keratin (CK, 7D3) to
mark CTB (A, C, E, and
G), and with anti-aVb5
(P3G2, B); anti-aVb6
(E7P6, D); or anti-
aVb3 (II344, F, H).
aVb5 is detected on
CTB in chorionic villi
(AV), but not in other
locations. aVb6 is de-
tected only on villus
CTB at sites of column
formation, and on the
first layer of cell col-
umn. aVb3 is detected
on column CTB near-
est the uterine wall and
on interstitial and endo-
vascular CTB (Zones
III-V). BV, maternal
blood vessel; EC, en-
dothelial cells.
Page 4
Cytotrophoblasts Mimic a Vascular Phenotype
2143
cell processes and whole cells that had penetrated the Matrigel could
be counted. In all invasion assays, the total number of processes plus
cells on the underside of each filter were counted. In the case of the
anticadherin assays, three filters were used for each condition in each
experiment. The VE-cadherin assay was repeated five times, and the
E-cadherin assay was repeated four times. Data are expressed as per-
cent control. In the anti-
a
V
b
3 invasion assays, four filters were used
for each experimental condition. The experiment was repeated two
times. The statistical significance of the data was analyzed by a Stu-
dent’s
t
test.
Adhesion assay.
A substrate of immobilized VCAM-1 was pre-
pared by incubating 8-well chamber slides overnight at 4
8
C with 200
m
l protein A (Sigma Chemical Corp., St. Louis, MO) at 1
m
g/ml in
PBS, pH 7.4. After nonspecific staining was blocked with 1% BSA in
PBS, 200
m
l/well of purified soluble chimeric VCAM-1–IgG
F
c (1.5
m
g/ml in PBS: gift of Dr. Ted Yednock, Athena Neurosciences, Inc.)
was added to the protein A-coated wells for 5 h at 24
8
C, or overnight
at 4
8C. This results in the binding of the VCAM-1 chimera to protein
A via its Fc end, leaving the adhesive domains of VCAM-1 available
for interaction with counter-receptors. CTB (10
5
) were then added to
the VCAM-1–coated wells in 250 ml serum-free DME containing
control IgG or either anti-VCAM-1 (100 mg/ml) or anti-integrin a4
(80 mg/ml) IgG (four samples of each condition/experiment), and in-
cubated at 378C for 1h. The samples were washed with PBS to re-
move nonadherent cells, fixed and adherent cells counted. The exper-
iment was repeated three times. The statistical significance of the
data was analyzed by a Student’s t test.
Isolation of RNA and Northern blotting. RNA was extracted from
freshly isolated or cultured CTB by the method of Chomczynski and
Sacchi (27), as described previously (13). Total RNA (10 mg/lane)
was separated by formaldehyde-agarose gel electrophoresis, trans-
ferred to Nytran membranes (Schleicher & Shuell, Keene, NH) and
analyzed by Northern blot hybridization as described previously (28).
Probes were synthesized by random priming of a 2.4 kb fragment of
human VE-cadherin and a 2.5 kb fragment of E-cadherin (gift of Dr.
D. Rimm, Yale University, New Haven, CT) using a[
32
P]dCTP and
the Klenow fragment of DNA polymerase-1 according to standard
methods. Final posthybridization was carried out in 0.33 SSC (45
mM NaCl, 15 mM sodium citrate) and 0.1% SDS at 658C (13). In all
experiments, gels were stained with acridine orange prior to transfer
to ensure integrity of the RNA samples, and to confirm that equal
amounts of RNA had been loaded onto each lane.
Results
There are several critical stages in the differentiation of CTB
stem cells along the invasive pathway: (a) conversion of chori-
onic villus CTB from a monolayer to a cell column; (b) their
invasion of the uterine wall and its vascular bed (Fig. 1 A); and
(c) their interaction with maternal endothelium in spiral arteri-
oles (Fig. 1 B). It is likely that significant switches in adhesion
receptor phenotype are important for enabling CTB to exe-
cute this complex morphogenetic program. In this study, we
tested the hypotheses that CTB mimic broadly the adhesion
phenotype of the endothelial cells they replace, and that the
changes in adhesion phenotype have the net effect of enhanc-
ing CTB motility and invasiveness. To test these hypotheses,
we first stained tissue sections of the fetal-maternal interface
for specific integrins, cadherins, and immunoglobulin family
adhesion receptors that are characteristic of endothelial cells
and leukocytes. We then tested the functional consequences
for CTB adhesion and invasion of expressing the particular ad-
hesion receptors that were upregulated during CTB differenti-
ation.
The
a
V
b
3 integrin is spatially regulated on differentiating
CTB and promotes their invasiveness. The distribution pat-
terns of aV integrin family members were examined because
of their regulated expression on endothelial cells during angio-
genesis (29) and their upregulation on some types of meta-
static tumor cells (e.g., melanoma; 30, 31). aV family members
displayed unique and highly specific spatial staining patterns
on CTB in anchoring villi and the placental bed (Fig. 2). The
aVb5-complex-specific antibody (Fig. 2, A and B) stained the
CTB stem cell monolayer in chorionic villi. Staining was uni-
form over the entire cell surface. At sites of column initiation
in first trimester tissue (10 wk), aVb5 staining extended to the
first 2–3 layers of column CTB. The syncytiotrophoblast layer,
and CTB in more distal layers of cell columns and in the pla-
cental bed, did not stain for aVb5. In contrast, anti-aVb6 did
not stain most villus CTB: only villus CTB that were at sites of
column formation stained (Fig. 2, C and D). The CTB layer
still in contact with basement membrane at sites of column ini-
tiation stained brightly for aVb6. In second trimester tissues
(18 wk; Fig. 2, C and D), the next layer of the CTB within the
column showed reduced staining. In first trimester tissues
(10 wk), the first 2–3 layers of CTB in cell columns also stained
strongly for b6 (not shown), as was observed for aVb5. The
rest of the CTB stem cells in chorionic villi, CTB in more distal
regions of cell columns, and CTB within placental bed and vas-
culature did not stain for aVb6, documenting a specific associ-
ation of this integrin with initiation of column formation.
In yet a different pattern, staining for anti-aVb3 was weak
or not detected on villus CTB or on CTB in the initial layers of
cell columns. However, strong staining was detected on CTB
within the uterine wall and vasculature (Fig. 2, E and H). Thus,
individual members of the aV family, like those of the b1 fam-
ily, are spatially regulated during CTB differentiation. Of par-
ticular relevance to this study is the observation that aVb3 in-
tegrin, whose expression is stimulated on endothelial cells by
angiogenic factors (29, 32), is enhanced on CTB that have in-
vaded the uterine wall and maternal vasculature.
Blocking aVb3 function suppresses endothelial migration
during angiogenesis (29). To determine whether perturbing
the function of aVb3 also affects CTB invasion in vitro, freshly
isolated first trimester CTB were plated for 48 h on Matrigel-
Figure 3. Antibody
against aVb3 strongly
inhibits invasion of iso-
lated CTB. CTB iso-
lated from first trimes-
ter placental villi were
cultured for 48 h on
Matrigel-coated Tran-
swell inserts in the pres-
ence of control mouse
IgG or complex-spe-
cific anti–aVb3 anti-
body LM609 (50 mg/ml)
as described in Meth-
ods. Cells and pro-
cesses that had invaded
the Matrigel, and were
visible on the under-
sides of the filters, were
counted. Four wells of each condition were counted in each experi-
ment. The experiment was conducted twice. Data were analyzed by
the Student’s t test. Bars indicate standard error of the mean.
Page 5
2144 Zhou et al.
coated Transwell filters in the presence of control mouse IgG
or the complex-specific anti–aVb3 IgG, LM609. CTB invasion
was evaluated by counting cells and cellular processes that had
invaded the Matrigel barrier and extended through the holes
in the Transwell filters (7, 13). LM609 reduced CTB invasion
by more than 75% in this assay (Fig. 3), indicating that this re-
ceptor, like the a1b1 integrin (13), contributes significantly to
the invasive phenotype of CTB.
Cadherin switching accompanies CTB differentiation in
vivo. In first and second trimester chorionic villi, the CTB epi-
thelial monolayer stained strongly for the ubiquitous epithe-
lial cadherin, E-cadherin, in a polarized pattern (Fig. 4). Stain-
ing was strong on the surfaces of CTB in contact with one
another and with the overlying syncytiotrophoblast layer, and
was absent at the basal surface of CTB in contact with base-
ment membrane (most evident in Fig. 4 F). In cell columns,
E-cadherin staining intensity was reduced on CTB near the
uterine wall and on CTB within the decidua (Fig. 4, A, B, E,
and F). This reduction in staining was particularly pronounced
in second trimester tissue (Fig. 4 F). At this stage, E-cadherin
staining was also very weak or undetectable on CTB that had
colonized maternal blood vessels and on CTB in the surround-
ing myometrium (Fig. 4, G and H). All locations of reduced
E-cadherin staining were areas in which invasion is active dur-
ing the first half of gestation. Interestingly, staining intensity of
E-cadherin was strong on CTB in all locations in term placen-
tas (34–40 wk, not shown), at which time CTB invasive activity
is poor (8, 13). With the exception of uterine gland epithelium,
maternal cells in these tissues, including decidual endothelium,
did not stain for E-cadherin at any stage of gestation. Taken
together, these data are consistent with the idea that CTB
transiently reduce E-cadherin function at times and places of
their greatest invasive activity.
Cadherin switching occurs frequently during embryonic de-
Figure 4. E-cadherin staining is reduced in normal differentiating 1st and 2nd trimester CTB. Sections of 1st trimester (10 wk, A–D) and 2nd tri-
mester (22 wk, E–H) placental bed tissue were stained with antibody against cytokeratin (CK, 7D3, A, C, E, and G) or E-cadherin (E9, B, D, F,
and H). E-cadherin staining was strong on anchoring villus (AV), CTB, and syncytium, and on CTB in the proximal portion of cell columns
(Zone II). Staining was sharply reduced on CTB in the distal column (Zone III) and in the uterine interstitium (Zone IV). Staining was not de-
tected on CTB within maternal vessels (Zone V). COL, column; BV, blood vessel.
Page 6
Cytotrophoblasts Mimic a Vascular Phenotype 2145
velopment when significant morphogenetic events take place
(33). We therefore stained sections of first and second trimes-
ter placental tissue with antibodies to other classical cadherins.
These tissues did not stain with antibodies against P-cadherin
(not shown). However, they did stain with three different mono-
clonal antibodies that recognize the endothelial cadherin, VE–
cadherin (cadherin-5; 26, 34, 35). In chorionic villi, antibody to
VE-cadherin did not stain villus CTB, although it stained the
endothelium of fetal blood vessels within the villus stroma.
In contrast, anti–VE-cadherin stained CTB in cell columns
and in the decidua (Fig. 5, A, B, E, and F). These are just the
areas in which E-cadherin staining was reduced. VE-cadherin
staining was stronger in second trimester tissues in these areas.
In maternal vessels that had not yet been modified by CTB,
VE-cadherin stained the endothelial layer strongly (i.e., at 10
wk gestation, Fig. 5, C and D). To confirm that maternal decid-
ual endothelial cells expressed VE-cadherin, but not E-cad-
herin, these cells were isolated from decidual tissue fragments
as described previously (22), and stained with antibodies
against E-cadherin, VE-cadherin, and the endothelial marker,
von Willebrand Factor (vWF). The decidual endothelial cul-
tures stained strongly for both VE-cadherin and vWF, but not
for E-cadherin (data not shown). Following endovascular inva-
sion, CTB lining maternal blood vessels stained strongly for
VE-cadherin (Fig. 5, G and H). They did not stain for vWF
(not shown; see companion paper, 20). Thus, CTB that invade
the uterine wall and vasculature express a cadherin character-
istic of endothelial cells.
Normal CTB upregulate VE-cadherin in vitro. To determine
the functional consequences for CTB differentiation/invasion
of expressing VE-cadherin, it was important to show that CTB
could modulate their cadherin repertoire in vitro. CTB were
Figure 5. VE-cadherin (VE-cad) staining is detected on differentiating and endovascular CTB. Sections of 1st trimester (10 wk, A–D) and 2nd
trimester (22 wk, E–H) placental bed tissue stained with antibody against cytokeratin (CK, A, C, E, and G) or VE-cadherin (BV6, B, D, F, and
H). VE-cadherin is not detected on CTB in AV(although fetal blood vessels in villus stromal core are stained). VE-cadherin is detected on col-
umn CTB (B and F) and on interstitial and endovascular CTB (G and H). VE-cadherin is also detected on maternal endothelial cells (EC) in
vessels that have not been modified by CTB (C and D). VE-cadherin staining was more intense on 2nd trimester CTB than 1st trimester column
CTB, and in the uterine wall VE-cadherin staining was more intense on endovascular than interstitial CTB.
Page 7
2146 Zhou et al.
therefore isolated from first and second trimester placentas.
At several time points after plating, extracts of isolated CTB
were separated by SDS-PAGE, transferred to nitrocellulose,
and immunoblotted with antibodies to E- and VE-cadherin
(Fig. 6 A). Anti–E-cadherin recognized an 80 kD band in
freshly isolated CTB and a band of similar density at 120 kD at
all subsequent time points (Fig. 6 A, a). These data indicate
that CTB express similar levels of total E-cadherin protein
throughout their differentiation. The 80-kD band observed at
0 time reflects the presence of a large, stable E-cadherin frag-
ment that is generated when cells are exposed to trypsin in the
presence of Ca
21
(e.g., 24), as occurs for a brief period during
the cell isolation period. The anti–VE-cadherin antibody did
not bind to anything in extracts of freshly isolated CTB. How-
ever, by 10 h, this antibody recognized a single band at 140 kD,
which increased in intensity at 36 h (Fig. 6 A, b). This band co-
migrated with a 140-kD band recognized by anti–VE-cadherin
in extracts of human umbilical vein endothelial cells (HUVEC;
Fig. 6 A, b). Both the full length VE-cadherin and a large pro-
teolytic fragment at about 100 kD, analogous to the 80 kD
E-cadherin fragment, were detected in HUVEC. However,
neither HUVEC (Fig. 6 A, a) nor cultured decidual endothe-
lial cells (not shown) expressed E-cadherin. Interestingly, both
E-cadherin and VE-cadherin were detected in extracts of the
JEG-3 choriocarcinoma cell line (Fig. 6 A, a and b). Neither
CTB, nor JEG-3 and HUVEC expressed P-cadherin, although
this cadherin was expressed prominently by keratinocytes (Fig.
6 A, c).
To determine whether cadherin modulation by differenti-
ating CTB was reflected at the mRNA level, Northern blotting
was carried out on RNA extracted from cultured first and sec-
ond trimester CTB (Fig. 6 B). Steady state levels of E-cadherin
mRNA were either somewhat reduced (as shown in Fig. 6 B)
or remained unchanged during the first 12 h of culture. In con-
trast, when the same blots were stripped and reprobed for VE-
cadherin, the VE-cadherin mRNA levels were consistently
very low in freshly isolated CTB and increased over the next
24 h. 28S ribosomal RNA was monitored by acridine orange
prior to transfer of the gel to assure uniform loading of sam-
ples (not shown). Taken together, these results indicate that
normal CTB continue to produce E-cadherin mRNA and pro-
tein as they differentiate, but upregulate expression of VE-
cadherin at both mRNA and protein levels.
E-cadherin and VE-cadherin have opposing effects on CTB
invasion in vitro. Next, we used function-perturbing anti–cad-
herin antibodies, in conjunction with the Matrigel invasion as-
say, to assess the functional consequences of cadherin modula-
tion for CTB invasiveness (Fig. 7). We plated isolated second
trimester CTB for 48 h on Matrigel-coated filters in the pres-
ence of control IgG or function-perturbing antibodies against
VE-cadherin or E-cadherin. By 48 h significant invasion was
Figure 6. VE-cadherin protein and mRNA are upregulated in CTB
differentiating in vitro. (A) CTB were isolated from 2nd trimester
chorionic villi and lysed immediately or after culture for the times in-
dicated. Lysates of CTB, human umbilical vein endothelial cells
(HUVEC), JEG-3 choriocarcinoma cells and primary human foreskin
keratinocytes were analyzed by immunoblotting using (a) anti–E-cad-
herin (E9); (b) anti–VE-cadherin (BV9); or (c) anti–P-cadherin
(6A9). (B) RNA was isolated from first and second trimester CTB
immediately after isolation from chorionic villi or after culture for the
times indicated. The RNA samples were analyzed by Northern blot-
ting using a[
32
P]dCTP-labeled cDNA probes for E-cadherin or VE-
cadherin. PBL, peripheral blood lymphocytes (negative control used
for E-cadherin); FIB, placental fibroblasts (negative control used for
VE-cadherin). Lines indicate positions of 28S and 18S RNA.
Figure 7. E-cadherin
and VE-cadherin have
opposing effects on
CTB invasion. CTB iso-
lated from 2nd TM
chorionic villi were
plated on Matrigel-
coated filters in Tran-
swell inserts and cul-
tured for 48 h in the
presence of control IgG
(100 mg/ml), anti–
VE-cadherin IgG
(BV6, 100 mg/ml) or
anti–E-cadherin IgG
(anti-GP80, 50 mg/ml).
Cells and processes
that had invaded the
Matrigel and were visible on the undersides of the filters were
counted. Triplicate wells were counted in each experiment. The ex-
periment was repeated four times for E-cadherin and five times for
VE-cadherin. Data were analyzed by the Student’s t test. Bars indi-
cate standard error of the mean.
Page 8
Cytotrophoblasts Mimic a Vascular Phenotype 2147
evident in control CTB. In cultures treated with anti–E-cad-
herin, CTB invasiveness increased more than threefold, sug-
gesting that the E-cadherin normally has a restraining effect on
CTB invasiveness. In contrast, antibody against VE-cadherin
reduced the invasion of CTB to about 60% of control (Fig. 7).
This suggests that the presence of VE-cadherin normally facili-
tates CTB invasion. Taken together, these functional data sug-
gest that as they differentiate, CTB modulate their cadherin
repertoire to one that contributes to their increased invasive-
ness.
Differentiating CTB undergo a comprehensive switch in
cell–cell adhesion molecule expression such that they mimic the
adhesion phenotype of endothelial cells. Our data thus far in-
dicate that, as they differentiate, CTB downregulate adhesion
receptors highly characteristic of epithelial cells (integrin a6b4
[11] and E-cadherin) and upregulate analogous receptors that
are expressed on endothelial cells (integrins a1b1 [11, 13] and
aVb3, and VE-cadherin). These observations support our hy-
pothesis that normal CTB undergo a comprehensive switch in
phenotype so as to resemble the endothelial cells they replace
during endovascular invasion. To investigate this possibility
further, we stained sections of first and second trimester nor-
mal placental bed biopsies with antibodies against cell–cell ad-
hesion molecules characteristic of quiescent or activated en-
dothelial cells, or against leukocyte cell adhesion molecules
that are involved in transendothelial trafficking. The data are
summarized in Table I.
E-selectin was expressed constitutively by normal first and
second trimester CTB in all compartments, including CTB
stem cells in the chorionic villi (Table I). However, chorionic
villus CTB did not stain for most of the other vascular adhe-
sion molecules examined. In contrast, CTB in cell columns
stained for VCAM-1 (Fig. 8, A–D), and the a4 subunit of its