Trophoblast 'pseudo-tumorigenesis': Significance and contributory factors

The ECM is composed of a network of secreted proteins and glycoproteins that forms the 'ground substance' seen outside cells and yet play a vital role in cellular function. It is not just an inert framework that supports or surrounds cells. It binds many growth factors and hormones and can either sequester these signals from the cells that contact it, or conversely, present the hormonal signals to these cells. Tumor cells manifest their invasive properties at several points: to enter and escape the circulation, and to penetrate into normal tissues.

Much like malignant tumor invasion of host tissue, trophoblast invasion of the maternal uterus is a multi-step process involving attachment of the trophoblast cells to the components of the ECM, degradation of the ECM and migration through the ECM [2]. Collagen type I, the major component of the interstitium ECM, has a remarkable stimulatory effect on gelatinase secretion by cytotrophoblast [3]. Studies have clearly shown that the ECM affects cell behavior [4]. Cytotrophoblast plated on Matrigel respond differently depending on the thickness of the Matrigel on which they are resting [4]. The mechanisms underlying these altered responses are, however, not clear.

A plethora of studies suggest that trophoblast invasion is not due to passive growth pressure, but due to an active biochemical process. A cell is considered to be invasive by virtue of its ability to secrete proteases capable of digesting its immediate environment, and human cytotrophoblast cells are no exceptions [2, 5‐7]. Serine proteases, cathepsins and matrix-metalloproteinases (MMPs) have all been implicated in this process [8]. This secretion begins even at the blastocyst stage [9].

MMPs are a family of zinc-dependent endopeptidases [8, 10]. They enzymatically digest ECM proteins and therefore play an important role in tissue remodeling processes [3]. Regulation of enzyme activity can occur via differential activation of the MMP, or by tissue inhibitors of metalloproteinases (TIMP). Direct evidence links the expression of MMPs, and particularly MMP-9, to the metastatic phenotype [7, 11], and the tissue inhibitor of metalloproteinases (TIMP) to the inhibition of metastatization [12]. Early studies [13] demonstrated that a proteinase cascade is required for the invasion of melanoma cells, and that the metalloproteinases play a major role in this process. Further, an inverse correlation is reported between TIMP levels and the invasive potential of murine and human cells [14].

In vitro, human cytotrophoblast invades acellular, amniotic membranes [15] as well as reconstituted basement Matrigel membranes [4, 16]. This invasiveness was also found to be independent of the surrounding uterine microenvironment [15]. They thus inherently behave like metastatic cells. This invasive behavior is clearly due to the ability of cytotrophoblast to secrete MMPs, since TIMP expression inhibits their invasiveness [17]. Furthermore, cultured cytotrophoblast cells secrete MMPs and the expression of MMP-2 and MMP-9 has been shown to mediate cytotrophoblast invasion into the Matrigel [2, 5]. Accordingly, changes in the synthesis of MMP-9 correlate with gestation-related changes in trophoblast behavior. Cytotrophoblast production and activation of MMP-9 peak during the first-trimester of pregnancy, coinciding with maximal invasive behavior in vivo [17, 18].

Transformed cells also often secrete a protease called plasminogen activator (PA), which cleaves a peptide bond in the serum protein plasminogen, converting it into the protease plasmin. Thus, the secretion of a small amount of plasminogen activator causes a large increase in protease concentration by catalytically activating the abundant plasminogen in normal serum. Normal cells treated with protease exhibit some characteristics of transformation (loss of actin microfilaments, growth stimulation etc.), suggesting that plasminogen activator secretion may help maintain the transformed state of certain cell lines [15]. This may be related to their tumor-forming capacity because the resulting increase in plasmin may help the cells penetrate the basal lamina.

Interestingly, the normally invasive extra-embryonic cells of the conceptus secrete PA while implanting in the uterine wall; this provides a compelling analogy to invasion by tumor cells. Human trophoblast cells also express urokinase-type PA receptor [19], which can bind active urokinase-type PA and localize proteolysis to the leading edge of migrating cells [20].

Proteolysis of the thrombin receptor, protease-activated receptor-1 (PAR-1), is known to enhance normal and pathological cellular invasion [21]. Recent evidence suggests that PAR-1 is the predominant thrombin receptor on invasive human extravillous trophoblast cells [21], and that PAR-1 (and potentially PAR-2 and PAR-3) may play an important role in the invasive phase of human placentation [21]. Receptor activation was shown to mediate extravillous trophoblast invasion in vitro.

ECM components are known to influence adhesion, spreading and migration of cells through specific cell-surface receptors called integrins [22]. Integrins are a large class of hetero-dimeric transmembrane glycoproteins composed of α- and β-subunits, and the ligand-binding site is composed of parts of both chains. Depending on the type of α/β combination, the integrins bind to different matrix glycoproteins and mediate different cell-cell interactions. For instance, α5β1 binds to fibronectin (Fn), α6β1 binds to laminin, α1β1 binds to collagen type IV etc. Tumor cells are known to produce elevated levels of these receptors for the basal lamina proteins, along with increased production of MMPs [10].

Interestingly, cytotrophoblast cells also have integrins, which allow these cells to recognize their immediate environment and adapt to it. Villous cytotrophoblast cells (that are destined to eventually differentiate by fusion to form syncytiotrophoblast) predominantly express the α6β4 integrin (a probable laminin receptor [7]), polarized along the basement membrane. In contrast, the highly invasive extra-villous cytotrophoblast cells (that break through the syncytial masses as multi-layered columns, to reach the decidua, the intima of the uterine blood vessels and the proximal-third of the myometrium) modulate the expression of their integrins. In the proximal region (the cytotrophoblast-columns), they express α6β4 in a non-polarized way, whereas in the most distal part (the placental bed), they express the α5β1 integrin, a Fn receptor []. Thus, as cytotrophoblast cells migrate from the villi into the decidua, they modulate their integrin repertoire from being α6β4-positive/α5β1-negative to becoming α6β4-negative/α5β1-positive. Endovascular cytotrophoblast cells express yet another integrin, α1β1, a collagen receptor [23] along with vascular cell adhesion molecule (VCAM) [10]. These changes in expression of specific cell adhesion molecules, are linked to the acquisition of an invasive phenotype, as the shift now allows the aggressive trophoblast cells to blend in with the endothelial cells as penetration of maternal vessels progresses [24]. Likewise, the wide range of altered behaviors that underlie malignancy are believed to have their basis in new or variant surface proteins made by malignant tumor cells.

Molecules made by the surrounding parenchyma, in some cases, are known to influence the potential of either cancer cells or trophoblast cells to invade. For instance, Insulin-like Growth Factor Binding Protein-1 (IGFBP-1), which is the major secretory product of the decidualized endometrium, has been shown to bind to the α5β1 integrin and stimulate cellular migration [25]. As this integrin is upregulated by the invading trophoblast [23], and is also present along with αvβ3 in metastatic malignant melanoma cells [26], such a mechanism may serve to activate integrins and promote invasive behavior during placental development. In vitro studies using human placental cells, also seem to suggest this. Both IGF-II (which is produced by the invasive extra-villous trophoblast cells throughout gestation) and IGFBP-1, were found to synergistically stimulate the invasive properties of first-trimester trophoblast cells [27].

Telomeres are complex DNA-protein structures that grant stability to chromosomes by effectively shielding against exonuclease digestion and preventing aberrant recombination [28]. Telomere shortening (owing to the 'end-replication problem') [29] has been proposed as a regulatory mechanism that controls the number of times a cell can divide before undergoing cellular senescence [30‐32]. Immortal cancer cells escape this growth-arrest signal by activating a cellular mechanism of telomere length maintenance [30]. Most often, this mechanism involves the activity of an enzyme called telomerase.

Telomerase is a specialized ribonucleoprotein enzyme complex that helps to compensate for replication-associated loss in telomeric repeats, which are essential for chromosome stability in actively dividing cells [33, 34]. Telomerase activity has been detected in most human tumor cells and tissues examined to date, as well as in human germ-line tissues, but as a sharp contrast, not in most normal somatic cells, thereby linking telomerase to unlimited cell proliferation and hence cell immortality [34]. Studies over the last several years have indicated the obligate requirement of telomerase for progression of human cancers [30, 35]. This is needed to achieve exponential growth patterns that would be otherwise unattainable.

In contrast to the majority of human somatic tissues and alike most progressive cancers, the human placenta expresses telomerase activity [36, 37]. Further, our studies suggest a functional role for telomerase in the maintenance of normal pregnancy [36].

We observed that mononuclear human cytotrophoblast cells retain their proliferative potential throughout gestation [36]. This observation has a very special significance. The presence of functional telomerase in human cytotrophoblast cells (and its retention even at term) probably ensures a capacity for high rates of proliferation, an event that is absolutely essential for normal placental growth and function.

The developing human fetus is a semi-allograft to the mother. While half the antigens of the fetus, being maternally derived, would be considered 'self', the other half, being paternally derived, would be considered 'foreign'. Therefore, one would expect that the mother would mount an immune response against the fetus. Yet, such an immune response leading to the death of the fetus is not usually seen. Although the placental association places the fetal circulation in close proximity with the maternal circulation, these two blood-streams do not mix. They are separated by tissue layers called the placental barrier. In humans, the placental barrier is essentially composed of the multinucleate syncytiotrophoblast layer alone. The outer surface of the syncytiotrophoblast layer is continuously bathed in maternal blood throughout gestation. Early in pregnancy, the fetal capillaries lie in a central position in the villus, but as pregnancy advances, the thin-walled anastomosing, endothelial fetal capillary tubes become more pronounced, come in close apposition with the inner surface of the syncytiotrophoblast layer and disperse the cytotrophoblast cells so that, the syncytiotrophoblast layer becomes the only barrier that effectively separates fetal and maternal circulations. Hence, the integral maintenance of this layer throughout gestation is of prime importance. The syncytiotrophoblast layer, however, is not immortal. It is well known that human syncytiotrophoblast has a defined life span, and is programmed to die at the end of it. Accumulated evidence points to a role for apoptosis in this renewal process. And therefore, there is the need for a mechanism to constantly replenish this crucial layer that periodically disappears.

Our studies prove the existence of mechanisms (namely, the continued presence of telomerase activity even in cells of term placenta) to aid in the maintenance of this vital layer [36]. Cytotrophoblastic stem cells are capable of division throughout gestation, while a subset of post-mitotic cells serves to replenish an outer syncytiotrophoblast layer after membrane fusion.

The trophoblastic epithelium of the placenta maintains a high concentration of steroid and protein hormones that are believed to confer immuno-protection to the growing fetus. The increased levels of progesterone at the placental/decidual border are equivalent to the in vitro levels, which inhibit lymphocyte responses to mitogens or allogeneic cells. This hormone may thus play a significant role in blunting maternal immune responses against the fetus [39]. Similarly, the activity of lymphocytes can be suppressed in vitro by a variety of substances synthesized in the course of a normal pregnancy including, human Chorionic Gonadotropin (hCG), human Placental Lactogen (hPL) [or human Chorionic Somatomammotropin (hCS)], prolactin, cortisone, estrogens and a host of other proteins and glycoproteins [40, 41]. The uterus also produces high concentrations of Transforming Growth Factor (TGF β) and prostaglandin E₂ (PGE₂), both of which are potent inhibitors of immune responses [42, 43]. Placental production of anti-inflammatory cytokines, TGF β2, Interleukin (IL)-4 and IL-10 have been postulated to counteract the deleterious effects of inflammatory cytokines IL-2, Interferon (IFN)-γ and Tumor Necrosis Factor (TNF)-α [44, 45]. Furthermore, Interferon β produced by the human trophoblast, is reported to have marked immunosuppressive effects on the mitogen-induced proliferation of human T-cells and B-cells in vitro in a dose-dependent manner, suggesting that the local cytokine network at the feto-maternal junction may also play an important role in the immunobiology of human pregnancies. Interestingly, trophoblast cells also abundantly express Fas-L (or CD95-L). Fas-L expression has been proposed to contribute to immune privilege in this unique environment, by fostering apoptosis of activated Fas-expressing lymphocytes of maternal origin [46]. An apoptotic process mediated by Fas-L may thus play a role in placental invasion during implantation, and this underscores similarities between trophoblast cells and neoplastic cells. Recent evidence suggests that hCG may be a crucial link in the development of peritrophoblastic maternal immune tolerance and may facilitate trophoblast invasion by regulating the Fas-FasL system [47].

Expression of the 'decay-accelerating factor' (or DAF) has been reported on the trophoblastic epithelium of the feto-maternal interface [49]. It was also observed that there was a quantitative increase in the expression of DAF during placental development [49]. Therefore, it was suggested that DAF might function by specifically inhibiting amplification convertases formed at the interface, either directly or indirectly, as a result of maternal complement activation [49]. More recently, the roles of two more proteins, 'membrane co-factor protein' (or MCP), which acts at the level of the C3 convertase enzymes to activate C3 to C3b, and another protein 'CD59', which regulates the formation and function of the terminal cytolytic membrane attack complex (MAC), have been implicated in the regulation of complement activity [50].

The strategically positioned cells of the placenta (or the trophoblast cells) completely encase the developing embryo and thereby prevent circulating maternal immune cells from recognizing foreign fetal antigens. Further, trophoblast cells themselves regulate the expression of their MHCs (the products of which are central to immune recognition and subsequent rejection of grafts) in a unique fashion. Unlike all other nucleated cells, trophoblast cells lack constitutive expression of the highly polymorphic Class I genes [41, 52‐54]. Instead they express a non-polymorphic gene, HLA-G, in abundant measures. The expression of this non-classical MHC class I gene on the cytotrophoblast surface, together with the total lack of expression of HLA heavy chain genes on syncytiotrophoblast, suggested that the non-polymorphic nature of HLA-G was responsible for the lack of cytotoxic immune responses in the placenta [41, 52‐54]. Interestingly, IFN β has been shown to enhance both transcription and cell surface expression of HLA-G on human trophoblast cells and amniotic cells [55].

In addition to the lack of constitutive expression of polymorphic MHC Class I genes, trophoblast cells also lack interferon γ-inducible expression of MHC Class I or II genes, suggesting that there is an active repression of cytokine-induced expression of Class I and Class II genes as well, in trophoblast cells [41, 43, 52, 53]. Major candidates for potent suppressors of MHC gene expression in human trophoblast cells include hCS, and a novel untranslated RNA or 'utron', called Trophoblast STAT Utron [41].

However, no single hypothesis has been able to convincingly explain the successful survival of the 'transplanted semi-allograft' throughout pregnancy. It is possible that many (or all) of these proposed mechanisms work in concert. In any case, acquisition of this 'transient state of tolerance' specific for paternal alloantigens suggests, that immune privilege at the feto-maternal interface is expressed systematically, is actively acquired and is necessary for survival of the fetus to term. This situation is reminiscent of that seen in most tumors. As with trophoblast cells, tumor cells often have low levels of Class I HLA, and in some instances even lack a full complement of the antigens [56].

Another common feature of placentation and cancer metastasis is the acquisition of extensive vascularization. Both the implanting embryo and the newly arrived tumor nodule require a rich blood supply in order to grow. Angiogenesis is the process of growth and development of new capillary blood vessels from pre-existing vessels. When pathological, it contributes to the development of numerous types of tumors, and the formation of metastases. In order to grow, carcinoma need new blood vessels to form, so that they can constantly feed themselves. Therefore, the concept according to which the development of cancer is angiogenesis-dependent, is well recognized. The transition from the latent phase to the invasive and metastatic phase of a cancer is linked to what is called the angiogenic-switch. It implies complex cellular and molecular interactions between cancerous cells, endothelial cells and the components of the ECM. This is made possible by specific proteins secreted by tumor cells that are able to stimulate the proliferation of capillary endothelial cells. The molecular communication that sends signals to a vessel to sprout and branch through the ECM in a tumor, is remarkably similar to the processes that occur during embryo implantation and placentation [10]. As described earlier, human cytotrophoblast cells and invasive melanoma both exhibit similar patterns of integrin expression that have been shown to adopt a vascular phenotype capable of invading maternal spiral arterioles, strikingly similar to those noted in endothelial cells as they migrate toward the tumor [57, 58]. During angiogenesis, endothelial cells also exhibit invasive and migratory behavior and utilize cell processes similar to those of both invading cytotrophoblast cells and spreading tumor cells [10].

There is evidence that human trophoblast cells release cytokines including basic Fibroblast Growth Factor (bFGF) [59]. During tumor growth, angiogenesis can be initiated by bFGF or by Vascular Endothelial Growth Factor (VEGF), which signal nearby vessels to send out new branches [60]. VEGF and bFGF are both very potent initiators of angiogenesis. During embryo implantation, VEGF mRNA and protein are localized in endometrial macrophages, as well as in the stroma and glandular epithelium [61, 62]. Interestingly, invading human cytotrophoblast cells express the VEGF receptor [63]. Once angiogenesis is initiated, events at the apex of a branching vessel are similar to those in other invading cells. Integrin-mediated migration and MMP activity directed toward the leading edge, create a path through the ECM.

The most widely known transcription factor, the AP-1 complex (Activator Protein-1), is a heterodimer of Jun and Fos, products of the proto-oncogenes c-jun and c-fos, which belong to the family of immediate response genes. There are numerous reports showing that AP-1 is involved in the MMP-1 response to IL-1 (Interleukin-1), TNF (Tumor Necrosis Factor) and TGF β [7, 24]. AP-1 is also involved in the MMP-3 regulation by PDGF and TGF β [7, 24]. MMP-1, -3 and most importantly, MMP-9, have a promoter site capable of binding Fos-Jun heterodimers. The AP-1 complex thus occupies a focal position in mediating signals that lead to increased MMP expression, and consequently, to the acquisition of an invasive phenotype.

Fos and Jun are both abundantly expressed in the early human trophoblastic cells [65]. Interestingly, the levels of c-fos expression in human amniotic and chorionic cells are close to the level of v-fos expression that results in the induction of osteosarcoma in mice and transformation of fibroblasts in vitro [66]. Further, EGF (Epidermal Growth Factor)-mediated effects on human trophoblast proliferation/ differentiation are believed to be via modulation in c-fos and c-jun expression [67]. Jun B, a member of AP-1 transcription factor family, has been shown to be essential for establishment of proper vascular connections with the maternal circulation during mammalian placentation [68]. Lack of Jun B resulted in embryonic lethality.

Another proto-oncogene of interest is c-myc, whose product is a transcription factor essential for cell cycle progression. Interestingly, several groups have shown that c-Myc activates telomerase, an effect attributed to direct interaction of c-Myc with the hTERT promoter [69, 70]. Alteration of c-myc expression is associated with a large number of tumors of diverse origin, that are characterized by uncontrolled telomerase activity. Similar to many actively dividing cancer cells, c-Myc expression was found to be critical for early trophoblast proliferation [71].

The proto-oncogene c-sis encodes the β chain of PDGF. This cytokine is expressed by the early proliferative trophoblast (both villous and extra-villous cytotrophoblast cells) and by blastocysts [72‐74]. The high transcriptional level in cytotrophoblast cells is comparable to that in human tumor cell lines that actively produce PDGF [72]. Further, the localization of c-sis transcripts in these cells, parallels that of c-myc transcription. Since PDGF stimulates the expression of c-Myc (which is a part of the post-receptor intracellular signaling pathway for the stimulation of cell proliferation by growth factors), and since these two products are co-localized in cytotrophoblastic cells, it is believed that PDGF at least partially controls trophoblast proliferation [72].

The human placenta also expresses c-erb B and c-fms, the products of which are the EGF and the M-CSF (Macrophage-Colony Stimulating Factor) receptors, respectively.

EGF receptor (EGF-R): EGF-Rs are predominantly found on villous cytotrophoblast cells and syncytiotrophoblast [75, 76]. Interestingly, the surrounding decidual cells also massively express EGF-R [24], thereby implying important functional roles for EGF-R ligands at the feto-maternal interface. There are 3 known ligands for EGF-R: EGF, TGF α and amphiregullin. All 3 ligands have been shown to enhance extra-villous trophoblast proliferation [77]. Immunoreactive EGF has been localized to uterine epithelial cells and decidual cells, as well as to both cyto- and syncytiotrophoblast of the chorionic villi [78]. Immunoreactive TGF α has been detected in decidual cells and nearly all trophoblast sub-populations of the human placenta at various gestational ages [77]. Amphiregullin has been demonstrated in the cytoplasm, as well as in the nuclei of syncytiotrophoblast cells of only early gestational placentae (upto 18 weeks) [79]. EGF and TGF α were found to stimulate normal cytotrophoblast proliferation acting through these high-affinity binding sites [80]. Studies have also shown that EGF acts as an autocrine factor in regulating early placental growth and function in synergy with thyroid hormone [80]. The TGF α-EGFR autocrine loop has been implicated in the uncontrolled proliferation of malignant trophoblast cells [81].

M-CSF receptor (M-CSF-R): CSF-1 is a homodimeric glycoprotein growth factor required for the proliferation and differentiation of mononuclear phagocytic cells. It has also been shown to have a role in placental growth and development under hormonal influence [82]. During pregnancy, the mRNA for its receptor has been localized to the trophoblast by in situ hybridization [83]. M-CSF-Rs [84, 85] are already expressed at the blastocyst stage, then later again by the invasive extra-villous cytotrophoblast cells. CSF-1 significantly increases the proliferation of first-trimester cytotrophoblast cells in an autocrine fashion. The expression of M-CSF-R has been correlated with trophoblastic invasiveness and the metastatic phenotype [24].

VEGF receptor (VEGF-R): The product of the c-flt proto-oncogene is an fms-like tyrosine kinase, which is the receptor for VEGF. This VEGF receptor is expressed by the invasive extra-villous cytotrophoblast cells [63]. As described earlier, VEGF produced by the endometrium, is a potent angiogenic factor and plays a crucial role during embryo implantation in the vascularization process. The expression of the functional receptor by the trophoblast thus favors the invasion process. VEGF has also been shown to promote proliferation of the invasive extra-villous cytotrophoblast cells [86]. The c-flt gene product (VEGFR-1) can also be bound by the trophoblast-derived Placenta Growth Factor (PlGF; a member of the VEGF family of angiogenic factors) to affect cell proliferation, invasion and/or other metabolic activities in an autocrine manner [87].

The parallel production of growth factors as well as their cognate receptors in the human placenta suggests an 'autocrine regulatory model' in which normal trophoblast cells synthesize agents capable of promoting their own proliferation. Autocrine regulation of growth is key to several tumor systems and hence this finding is interesting in the placental context.

A fundamental characteristic of transformed cells is that they continue to divide even when their normal non-malignant counter-parts cease dividing (loss of growth control). Transformed cells often dispense with hormone- or growth factor-requirement for their growth, and can thrive in initial serum concentrations that are much lower than those required by normal cells. Many transformed cells continuously produce both growth factors and their receptors, thereby providing themselves an auto-stimulatory growth impetus. Inappropriate or unregulated expression of these growth stimulators results is a loss of capacity for growth arrest.