The human villous cytotrophoblast: interactions with extracellular
matrix proteins, endocrine function and cytoplasmic differentiation
in the absence of syncytium formation
Lee-Chuan Kao*, Stephen Caltabiano+, Samuel Wu, Jerome F. Strauss III*+ and Harvey J. Kliman+
*Departments of Obstetrics and Gynecology, and +Pathology and Laboratory Medicine, University of Pennsylvania,
Philadelphia, PA 19104
Running head: Trophoblast differentiation
Key words: placenta, extracellular matrix, hCG
Supported by USPHS Grants DK-07314 (LCK), HD-06274 (JFS) and HD-00715
(HJK).
Address all correspondence to :
Harvey J. Kliman, M.D.-Ph.D.
Dept of Pathology and Lab. Medicine
6 Founders Pavilion
Hospital of the University of Pennsylvania
3400 Spruce Street
Philadelphia, PA 19104-4823
215 662-6518
Human syncytiotrophoblasts are derived from villous cytotrophoblasts by cell fusion. Coincident with this morphologic transformation, trophoblasts acquire specific endocrine functions, including elaboration of chorionic gonadotropin (hCG). We wondered if syncytia formation was a prerequisite for biochemical differentiation, or simply was one part of the differentiation program. By growing purified human cytotrophoblasts under serum-free conditions and manipulating the culture surface, we were able to disassociate morphologic from biochemical differentiation. We have shown previously (Endocrinology 118:1567, 1986) that human cytotrophoblasts grown in the presence of fetal calf serum flatten out, aggregate and form functional syncytiotrophoblasts in vitro over 24-96 h. Here we demonstrate that when grown in the absence of serum, the cells do not undergo these morphologic changes, but remain as individual spherical cells. If the culture surface was precoated with fibronectin or a variety of collagens, but not albumin, the cells regained their ability to flatten, aggregate and form syncytia. Attachment to and syncytia formation on fibronectin was blocked by the addition of the RGDS containing peptide, Gly-Arg-Gly-Asp-Ser-Pro. Attachment to and syncytia formation on type I collagen was blocked by anti-human fibronectin F(ab')2 fragments, while association with type IV collagen was not affected by this antibody, suggesting that fibronectin mediates trophoblast association with type I collagen, but not type IV. Although syncytia formation did not occur when cytotrophoblasts were cultured under serum-free conditions in the absence of ECM proteins, biochemical differentiation was not affected. These cells secreted hCG at the same rate under serum-free conditions whether they were plated on plastic only-which prevented syncytia formation-or fibronectin, laminin or type IV collagen-which allowed syncytia formation to occur. Furthermore, cytoplasmic differentiation in the absence of syncytia formation was confirmed by performing transmission electron microscopy on cytotrophoblasts grown under serum-free conditions in the presence of 8-bromo-cAMP. We conclude that syncytia formation is not a prerequisite for biochemical differentiation, but simply part of the trophoblast differentiation program.
Human villous cytotrophoblasts form syncytial trophoblasts through
a process of cell fusion (Boyd and Hamilton, 1970 and Kliman et al., 1986). There has been speculation that biochemical differentiation,
such as the ability to secrete chorionic gonadotropin (hCG) (Pierce
and Midgley, 1963, and Morrish et al., 1987) and placental lactogen (Hoshina et al., 1983, and Hoshina et al., 1985), is tied to syncytia formation. We have shown previously
that cultured human cytotrophoblasts differentiate biochemically
and morphologically in a time-dependent fashion (Kliman et al., 1986, Feinman et al., 1986, and Kliman et al., 1987). In the presence of fetal calf serum (FCS), the mononuclear
cytotrophoblasts flatten out onto the culture surface, aggregate
and form syncytia over a 24 to 96 h period. Simultaneously with
these morphological changes, the trophoblasts secrete increasing
amounts of hCG into the culture media. This in vitro system did not allow us to dissociate the morphologic and biochemical
changes. Here we demonstrate that by culturing cytotrophoblasts
under serum-free conditions and manipulating the culture surface
by precoating with a variety of extracellular matrix proteins,
morphologic and biochemical differentiation can be uncoupled.
In addition, we have gained insight into the specificity of associations
between cultured cytotrophoblasts and components of the extracellular
matrix.
Human cytotrophoblasts were purified from normal term placentae
obtained following spontaneous vaginal delivery or uncomplicated
Cesarean section using the method of Kliman et al. (1986). Briefly, 30 gm of soft, villous tissue was subjected
to three trypsin/DNase digestions. The resultant cell suspensions
were pooled, placed on a discontinuous Percoll gradient (5 to
70% Percoll in 5% steps), and centrifuged at 1200 x g for 20 min.
A greater than 95% pure population of cytotrophoblasts resulted
as a band at a density of 1.05 to 1.06 gm/ml. The cytotrophoblasts
were diluted to a final concentration of 106 cells/ml with Dulbecco's Modified Eagles' Medium containing 25
mM HEPES and 25 mM glucose (DMEM-H-G) with glutamine (4 mM) and
gentamicin (50 µg/ml) added, and, in serum-supplemented media,
1 to 20% (vol/vol) heat-inactivated fetal calf serum. 8Bromo-cAMP
(1.5 mM, Sigma Chemical Co., St. Louis, MO) was added to some
media. One ml of the cell suspension was plated into 16 mm Nunclon
multidishes (Nunc, Roskilde, Denmark) for biochemical studies,
while 0.2 to 1.0 ml was plated onto 22 mm square coverslips (#1
or 11/2) placed in 35 mm Nunclon multidishes, or into 4-chamber Lab-Tek
(Miles Laboratories, Mishawaka, Indiana) culture slides for histologic
studies. For electron microscopy, 10 ml of the cell suspension
was cultured in 80 cm2 Nunclon flasks in the absence of serum, with or without added
8-bromo-cAMP. The cells were incubated at 37_C in humidified 5%
CO2/95% air with media changes every 24 h.
Extracellular matrix proteins
Fibronectin (Collaborative Research, Bedford, MA) was dissolved
in DMEM-H-G with 50 µg/ml gentamicin to a final concentration
of 50 µg/ml. Laminin (Collaborative Research) was dissolved in
DMEM-H-G with 50 µg/ml gentamicin to a final concentration of
20 µg/ml. Collagens (Sigma Chemical Co.) were dissolved in 0.15%
NaCl (w/vol) with 50 µg/ml gentamicin to a final concentration
of 80 or 100 µg/ml. Bovine serum albumin, fraction 5 (Miles Scientific,
Naperville, IL) was dissolved in DH2O to a final concentration
of 80 µg/ml. All solutions were filtered with a 0.2 µm syringe
filter, applied to either cover glasses or plastic culture surfaces
until covered, allowed to stand overnight in a humidified incubator
at 37_C, aspirated and dried in a culture hood for 1-2 h prior
to use.
Antibodies and R-G-D-S peptide
Affinity purified F(ab')2 fragment goat anti-human fibronectin (Accurate Chemical and Scientific
Corporation, Westbury, NY) was diluted to 20 µg/ml into each well
from a 1.1 mg/ml stock at the time of cell plating. Chromatography
purified goat IgG F(ab')2 fragment (Accurate) was used as a control at 20 µg/ml as described
above for anti-fibronectin fragments. An R-G-D-S containing hexapeptide,
Gly-Arg-Gly-Asp-Ser-Pro (G-R-G-D-S-P), generously provided by
Dr. Henry T. Keutman, Endocrine Unit, Massachusetts General Hospital,
was used at a final concentration of 1 mM.
Histology and immunohistochemistry
Cultured cells were grown on coverslips, fixed, stained and photographed
as previously described (Kliman et al., 1986). The percentage of nuclei in syncytia was determined counting
500 nuclei in sequential high power fields and determining whether
they were in single isolated cells, single cells of an aggregate
(2 or more cells attached to each other or single cells attached
to syncytia), or syncytia (defined as a cell with two or more
nuclei within one obvious cytoplasmic border). The total number
of nuclei in syncytia was divided by 500.
Following 24 h of culture, the trophoblasts were washed once with
warm DMEM-H-G, and then fixed with half-strength Karnovsky's fixative
(Karnovsky, 1965) for 15 min at room temperature, followed by
incubation overnight at 4_C . The fixed cells were removed from
the flasks by adding approximately 100 pre-cleaned 3 mm glass
beads to the tissue culture flasks, agitating gently for 1 minute,
then aspirating off the cell suspension. The cells were then pelleted,
transferred to 0.1 M cacodylate buffer, pH 7.4, postfixed with
1% osmium for 1 h, dehydrated in graded ethanols and propylene
oxide, and embedded in Epon. Silver sections were stained with
uranyl acetate and lead citrate and examined at 75 kV with a Hitachi
600 electron microscope (Hitachi Corp., Tokyo, Japan).
hCG was quantitated using reagents obtained from Serono (hCG MAIA
clone; Braintree, MA). The assay was calibrated to the First International
Reference Preparation.
We next investigated whether purified extracellular matrix proteins
could support syncytia formation in serum-free medium. Precoating
glass coverslips with fibronectin or a variety of collagens permitted
cytotrophoblasts to form syncytia in serum-free media (Fig. 3).
Although culture for 96 h in the presence of FCS produced the greatest
degree of syncytial formation (83% ), approximately 50% of the
trophoblast nuclei were found in syncytia when the culture surface
was precoated with fibronectin; placental collagens types I, IV
and V; calf skin collagen type I; and rat tail collagen. Approximately
33% of the trophoblast nuclei were found in syncytia when the
culture surface was precoated with placental collagens types III
and V. In contrast, without serum or precoating, only 2.5% of
the trophoblast nuclei were found in syncytia at the end of the
96 h of culture period. Precoating with bovine serum albumin yielded
results similar to no precoating. This result suggests that fibronectin
and collagens do not support syncytia formation by, for example,
neutralization of negative charges on the coverslip, but mediate
flattening, aggregation and syncytia formation by a more specific
mechanism.
R-G-D-S specific binding of trophoblasts to fibronectin
We next investigated the specificity of trophoblast binding to
fibronectin. When human trophoblasts were cultured on fibronectin
pre-treated coverslips in serum-free media, they flattened out,
aggregated and formed syncytia (Fig. 4 a), in a fashion similar
to cells grown in the presence of FCS (Fig. 1). When the R-G-D-S
containing hexapeptide, G-R-G-D-S-P, was added at the time of
plating, virtually all adhesion to the fibronectin coated coverslip
was eliminated (Fig. 4 b). In order to verify that the inhibition
of fibronectin adhesion was specific, we performed the same experiment
using type IV collagen pre-coated coverslips. In the presence
of the hexapeptide, human trophoblasts attached, flattened, aggregated,
and formed syncytia on the type IV collagen surface in the presence
of the hexapeptide (Fig. 4 c). These results suggest that human
trophoblasts have a specific, R-G-D-S-sensitive, receptor for
fibronectin, but that they also express binding sites or molecules
which allow them to adhere to collagen.
Trophoblast binding to types I and IV collagen: involvement of
fibronectin in binding to type I collagen
Figures 3 and 4 c illustrate that human trophoblasts bind to a variety of collagens. These results raise the following question: are there specific collagen binding sites on the human trophoblast, or do these cells bind to collagens via other proteins, such as fibronectin and laminin?
Utilizing specific anti-human fibronectin F(ab')2 fragments, we attempted to determine if trophoblasts bind to types I and IV collagens directly or indirectly through fibronectin. When trophoblasts were cultured on type I collagen pre-treated coverslips, they adhered, flattened and formed aggregates within 24 h (Fig. 5 a). Normal goat serum F(ab')2 fragments added at the time of plating had no effect on the cultures (Fig. 5 b). If, on the other hand, goat anti-human fibronectin F(ab')2 fragments were added at the same concentration at the time of plating, attachment of trophoblasts was virtually eliminated. In addition, the few cells that did attach did not flatten out or aggregate, but remained spherical-similar to the results under serum-free conditions with no precoating of the culture surface (Fig. 5 c). Since no fibronectin was added to these cultures, but the anti-human fibronectin F(ab')2 fragments still prevented attachment of trophoblasts to type I collagen, we verified that this particular antibody did not cross react with type I collagen. Ouchterlony double-immunodiffusion analysis of the anti-human fibronectin F(ab')2 fragments against type I collagen revealed no detectable line of reaction while a strong line of reaction was seen against fibronectin.
Type IV collagen supported attachment, aggregation and syncytia
formation (Fig. 5 d) and the presence of normal goat serum F(ab')2 fragments had no effect on these interactions (Fig. 5 e). In
contrast to the results shown in figure 5 c for type I collagen,
the presence of goat anti-human fibronectin F(ab')2 fragments had no effect on the interaction of the cytotrophoblasts
with type IV collagen (Fig. 5 f). Thus, anti-fibronectin F(ab')2 only interfered with trophoblast binding to type I collagen,
not type IV collagen. These results suggest that trophoblasts
bind to type I collagen via cell-associated fibronectin, while
they either bind to type IV collagen directly or through some
other intermediate protein, such as laminin. These results are
consistent with emerging evidence which indicates that fibronectin
is an intermediate between cells and type I collagen (Akiyama
and Yamada, 1987).
Stimulation of hCG secretion in the absence of syncytia formation
We have demonstrated that under serum-free conditions without precoating or with type I collagen precoating in the presence of anti-fibronectin F(ab')2 that cytotrophoblasts do not flatten out, aggregate, or form syncytia. One might ask: Are these cells viable? Are they biochemically active? In order to answer these questions, we performed biochemical and electron microscopic studies on human trophoblasts cultured under serum-free conditions with and without pre-treatment of the culture surfaces.
We had shown previously that 8-bromo-cAMP stimulates hCG secretion
by cultured human trophoblasts in the presence of serum (Feinman
et al., 1986). We wondered if trophoblasts could be stimulated to secrete
hCG under serum-free conditions where 1) they could not form syncytia
or 2) they could syncytialize because they were plated on fibronectin,
laminin or type IV collagen-treated surfaces. Figure 6 illustrates
the results of several such experiments. Trophoblasts were cultured
over a 24 h period under serum-free conditions without pretreatment,
or in serum-free medium on surfaces coated with fibronectin, type
IV collagen, or laminin. In addition, half of the cultures were
treated with 1.5 mM 8-bromo-cAMP. These data establish that trophoblasts
secrete hCG in a qualitatively similar fashion under serum-free
conditions whether cultured in the absence or presence of extracellular
matrix proteins. Furthermore, these results indicate that trophoblasts
which do not flatten down on the culture surface and do not subsequently
form syncytia because they are grown under serum-free conditions
(see Fig. 1 b) still express endocrine activities associated with
the syncytial trophoblast.
We answered this question by plating trophoblasts onto fibronectin in the absence or presence of an R-G-D-S-containing hexapeptide, G-R-G-D-S-P, (Fig. 4), which is known to block fibronectin-receptor-specific cell attachment to fibronectin (Yamada et al., 1985). This experiment demonstrated virtually complete inhibition of adhesion of the trophoblasts to fibronectin in the presence of the hexapeptide. Interestingly, the few cells that did adhere in the presence of the hexapeptide remained spherical. This suggests that human trophoblasts bind to fibronectin through an R-G-D-S-specific receptor (Akiyama and Yamada, 1987), and further, that interaction with fibronectin may facilitate spreading of these cells. Further experiments with types I and IV collagen (Fig. 5) suggested that these cells bind to type I collagen via fibronectin and that in contrast, association to type IV collagen does not involve fibronectin. Cytotrophoblasts synthesize and secrete fibronectin (Ulloa-Aguirre et al., 1987) and the inability of cytotrophoblasts to flatten out and form syncytia in serum-free media in the absence of an exogenous matrix component suggests that either the fibronectin produced by the cells is insufficient in quantity or not the appropriate type to promote normal morphologic differentiation in vitro. The latter possibility is supported by biochemical studies on human placental fibronectin. Zhu and co-workers (Zhu et al., 1984, and Zhu and Laine, 1985), and Isemura et al. (1984) have determined that human placental fibronectin and serum fibronectin contain different carbohydrates. Zhu et al. (1985) have demonstrated that placental fibronectin has high molecular weight polylactosamine carbohydrate and this additional carbohydrate imparts to the placental fibronectin a decreased binding affinity to gelatin. Furthermore, these workers (Zhu and Laine, 1987) have shown that placental fibronectin acquires increasing amounts of these carbohydrates as gestation proceeds, raising the possibility that this change may play a role in placental separation from the uterus. These biochemical studies, therefore, may explain why human cytotrophoblasts cultured under serum-free conditions without exogenous matrix do not flatten in spite of the fact that they produce fibronectin.
Our experiments revealed that we could modulate morphologic differentiation-that is, progression of mononuclear cytotrophoblasts to multinuclear syncytial trophoblasts-by modifying the surface on which the cells were cultured. We wondered whether cytotrophoblasts cultured in the absence of serum without the precoating of the surface (thus constraining them to remain as single cells) could still differentiate biochemically (i.e., could still be stimulated to secrete hCG) in spite of their inability to form syncytial structures. The results presented in figure 6 clearly demonstrate that trophoblasts grown in the absence of serum can secrete hCG and that 8-bromo-cAMP augments this secretory activity whether they are placed on uncoated surfaces or surfaces coated with fibronectin, type IV collagen, or laminin. Thus, trophoblasts which remain as single mononuclear cells can express endocrine functions of the syncytial trophoblast.
To confirm this finding, we performed transmission electron microscopy on cytotrophoblasts grown without serum or precoating, with or without added 8-bromo-cAMP, to evaluate whether the cells when stimulated acquire the necessary machinery for synthesis and secretion of hormones. Cytotrophoblasts grown in serum-free media exhibited relatively simple fine structural features after 24 h of culture. They contained microvilli and coated pits, occasional lipid droplets and scattered short segments of RER-consistent with cytotrophoblast ultrastructure (Boyd and Hamilton, 1970 and Kliman et al., 1986). Cells cultured in the presence of 8-bromo-cAMP remained spherical and mononuclear, but they were larger, contained stacks of dilated RER, had well-developed golgi, and contained greater numbers of mitochondria-features typical of the syncytiotrophoblast-all confirming their increase in biosynthetic capacity. Therefore, we have demonstrated that these cells differentiated both biochemically and structurally in culture in the absence of syncytium formation.
It is generally accepted that the cytotrophoblast is the undifferentiated
stem cell of the villous trophoblast and that the syncytial trophoblast
is the fully differentiated, end-stage trophoblast (Pierce and
Midgley, 1963 and Boyd and Hamilton, 1970). Some have suggested
that biochemical differentiation can occur only after syncytial
formation (Morrish et al., 1987). Our work contradicts this notion. We believe that the
trophoblasts of the human chorionic villi form a continuum of
differentiated cells, some of which are mononuclear and others
multinuclear. In addition, our results support the hypothesis
that syncytia formation is not a prerequisite for biochemical
differentiation. Therefore, the formation of the syncytiotrophoblast
is not the trigger for biochemical differentiation, but is only
one of the consequences of the differentiation program. This scheme
parallels very closely the biology of myoblast differentiation
(Holtzer et al., 1980). The serum-free system described here, and its ability
to disassociate morphologic from biochemical differentiation in
the human trophoblast, should permit us to elucidate the mechanisms
of trophoblast regulation and maturation.
We gratefully acknowledge Justina Minda for assistance in the electron microscopy and Julia Haimowitz for the preparation of the manuscript.
Akiyama, S. K., and Yamada, K. M. (1987). Fibronectin. Adv. Enzymol. 59, 1-57.
Boyd, J. D., and Hamilton W. J. (1970). In "The human placenta," pp. 167-174. MacMillan, London.
Feinman MA, Kliman HJ, Caltabiano S and Strauss JF III. (1986).
8-Bromo-3'5'-adenosine monophosphate stimulates the endocrine
activity of human cytotrophoblasts in culture. J. Clin. Endocrinol. Metab. 63, 1211-1217.
Hoshina, M., Hussa, R., Pattillo, R., and Boime, I. (1983). Cytologic
distribution of chorionic gonadotropin subunit and placental lactogen
messenger RNA in neoplasms derived from human placenta. J. Cell
Biol. 97, 1200-1206.
Holtzer, B., Croop, J., Toyama, Y., Bennett, G. H., Fellini, S.,
and West, C. (1980). Differences in differentiation programs between
presumptive myoblasts and their daughters, the definitive myoblast
and myotubes. In "Plasticity of Muscle" (D. Pete, Ed.), p 133. de Gruyter, New
York, NY.
Hoshina, M., Boothby, M., Hussa, R., Pattillo, R., Camel, H. M.,
and Boime, I. (1985). Linkage of human chorionic gonadotrophin
and placental lactogen biosynthesis to trophoblast differentiation
and tumorigenesis. Placenta 6, 163-172.
Isemura, M., Yamaguchi, Y., Munakata, H., Aikawa, J., Kan, M.,
Yamane, I., Yosizawa, Z. (1984). Isolation and characterization
of human placenta fibronectin. J. Biochem. 96, 163-9.
Karnovsky, M. (1965). A formaldehyde-glutaraldehyde fixative of
high osmolarity for use in electron microscopy. J. Cell Biol. 27, 127A.
Kliman, H. J., Nestler, J. E., Sermasi, E., Sanger, J. M., and
Strauss, J. F. III. (1986). Purification, characterization and
in vitro differentiation of cytotrophoblasts from human term placentae.
Endocrinology 118, 1567-1582.
Kliman, H. J., Feinman, M. A., and Strauss, J. F. III. (1987).
Differentiation of human cytotrophoblasts into syncytiotrophoblasts
in culture. In "Trophoblast Research" (R. Miller and H. Thied, Eds.), Vol. 2,
pp. 407-421. Plenum Medical, New York, NY.
Morrish, D. W., Bhardwaj, D., Dabbagh, L. K., Marusyk, H., and
Siy, O. (1987). Epidermal growth factor induces differentiation
and secretion of human chorionic gonadotropin and placental lactogen
in normal human placenta. J. Clin. Endocrinol. Metab. 65, 1282-1290.
Pierce, G. B., Jr., and Midgley A. R., Jr. (1963). The origin
and function of human syncytiotrophoblastic giant cells. Amer. J. Path. 43, 153-173.
Ulloa-Aguirre, A., August, A. M., Golos, T. G., Kao, L.-C., Sukuragi,
N., Kliman, H. J., and Strauss, J. F. III. (1987). 8-Bromo-3'5'
adenosine monophosphate regulates expression of chorionic gonadotropin
and fibronectin in human cytotrophoblasts. J. Clin. End. Metab. 64, 1002-1009.
Yamada, K. M., Akiyama, S. K., Hasegawa, T., Hasegawa, E., Humphries,
M. J., Kennedy, D. W., Nagata, K., Urushihara, U., Olden, K.,
and Chen, W.-T. (1985). Recent advances in research on fibronectin
and other cell attachment proteins. J. Cell. Biochem. 28, 79-97.
Zhu, B. C.-R., Fisher, S. F., Pande, H., Calaycay, J., Shively,
J. E., and Laine, R. A. (1984). Human placental (fetal) fibronectin,
increased glycosylation and higher protease resistance than plasma
fibronectin. J. Biol. Chem. 259, 3962-3970.
Zhu, B. C., Laine, R. A. (1985). Polylactosamine glycosylation
on human fetal placental fibronectin weakens the binding affinity
of fibronectin to gelatin. J. Biol. Chem. 260, 4041-5.
Zhu, B. C., Laine, R. A. (1987). Developmental study of human fetal placental fibronectin: alterations in carbohydrates of tissue fibronectin during gestation. Arch. Biochem. Biophys. 252, 1-6.