Mechanisms of trophoblast migration, endometrial angiogenesis in preeclampsia: The role of decorin

abstract

The objective of the present review is to synthesize the information on the cellular and molecular players responsible for maintaining a homeostatic balance between a naturally invasive human placenta and the maternal uterus in pregnancy; to review the roles of decorin (DCN) as a molecular player in this homeostasis; to list the common maladies associated with a break-down in this homeostasis, resulting from a hypo-invasive or hyper-invasive placenta, and their underlying mechanisms. We show that both the fetal components of the placenta, represented primarily by the extravillous trophoblast, and the maternal component represented primarily by the decidual tissue and the endometrial arterioles, participate actively in this balance. We discuss the process of uterine angiogenesis in the context of uterine arterial changes during normal pregnancy and preeclampsia. We compare and contrast trophoblast growth and invasion with the processes involved in tumorigenesis with special emphasis on the roles of DCN and raise important questions that remain to be addressed. Decorin (DCN) is a small leucine-rich proteoglycan produced by stromal cells, including dermal fibroblasts, chondrocytes, chorionic villus mesenchymal cells and decidual cells of the pregnant endometrium. It contains a 40 kDa protein core having 10 leucine-rich repeats covalently linked with a glycosaminoglycan chain. Biological functions of DCN include: collagen assembly, myogenesis, tissue repair and regulation of cell adhesion and migration by binding to ECM molecules or antagonising multiple tyrosine kinase receptors (TKR) including EGFR, IGF-IR, HGFR and VEGFR-2. DCN restrains angiogenesis by binding to thrombospondin-1, TGFβ, VEGFR-2 and possibly IGF-IR. DCN can halt tumor growth by antagonising oncogenic TKRs and restraining angiogenesis. DCN actions at the fetal-maternal interface include restraint of trophoblast migration, invasion and uterine angiogenesis. We demonstrate that DCN overexpression in the decidua is associated with preeclampsia (PE); this may have a causal role in PE by compromising endovascular differentiation of the trophoblast and uterine angiogenesis, resulting in poor arterial remodeling. Elevated DCN level in the maternal blood is suggested as a potential biomarker in PE.

KEYWORDS:

• placenta
• decidua
• angiogenesis
• trophoblast
• extravillous trophoblast
• decorin
• adhesion
• migration
• invasion
• preeclampsia
• leucine rich proteoglycan
• tumor growth and metastasis
• tyrosine kinase receptors

Structure of the human placenta as an invasive organ

The human placenta is a highly invasive organ in which a subpopulation of trophoblast cells proliferate, migrate and invade the uterine endometrium and its arteries to nourish the fetus.^1,2 Adequate placental perfusion with maternal blood depends on highly regulated invasion and remodeling of utero-placental arteries by the trophoblast.^3,4 There are 2 distinct pathways of trophoblast differentiation from the bipotent stem cells in the cytotrophoblast layer of chorionic villi (Fig. 1): the villous pathway, in which cells proliferate and fuse, giving rise to the syncytiotrophoblast layer facing the maternal sinusoids, engaged primarily in exchange and endocrine functions; and the extravillous pathway in which cells break out of the villi as discrete cell columns which proliferate at their base,^5,6 migrate and invade the decidua and its arteries by adopting an “endovascular phenotype” (expressing certain endothelial cell markers) and replace the endothelial lining of the arteries.^7-9 Proliferation, migration and invasiveness of the extravillous trophoblast (EVT) cells are exquisitely regulated in situ,^1,2,10 and this may fail in certain pathological conditions. Poor EVT invasion and arterial remodeling underlies preeclampsia and certain forms of intrauterine growth restriction (IUGR),^4,11-13 whereas uncontrolled invasion features trophoblastic neoplasias, including invasive moles and choriocarcinomas.¹³

Figure 1. Schematic diagram of human placenta at approximately 10 weeks of gestation showing the fetal compartment including the trophoblast and maternal sinusoids, and the maternal compartment including the decidua and uterine vessels (UV). Chorionic villi (CV) consist of 2 layers of trophoblast cells, the inner cytotrophoblast (CT) and the outer syncytiotrophoblast (ST). Cytotrophoblast cells in some villi known as anchoring villi (AV) migrate out as cell columns as extravillous trophoblasts (EVT) to invade the decidua. Some EVT cells remain dispersed in the deciduas as interstitial trophoblast (IT), some fuse to become trophoblast giant cells (GC) and others form cytotrophoblastic shell (CS) bordering the decidua and the maternal sinusoids. Some EVT cells from the CS enter uterine arteries as endovascular trophoblasts. It is also suggested that some IT cells invade the arteries from the exterior. (Reproduced with permission from Graham CH and Lala PK. Biochem Cell Biology 1992; 70: 867–874.).

Uterine angiogenesis and arterial remodeling during normal pregnancy and preeclampsia

While blood vessels in the fetal compartment of the placenta arise by the process of “vasculogenesis” by recruitment of mesenchymally derived angioblasts, those in the maternal compartment represented by the decidualized endometrium arise by sprouting from pre-existing uterine vessels, or the process of “angiogenesis.” The arteries, but not the veins, in the endometrium and the inner third of the myometrium are extensively modified during normal pregnancy by the EVT cells. It has been suggested that EVT cell preference for migration toward the uterine arteries rather than veins is guided by the differential expression of chemokine receptors ephrin type-B receptor 2 (EPHB2) on arterial endothelium and ephrin type-B receptor 4 (EPHB4) on venous endothelium and their transmembrane ligand ephrin on EVT cells.¹⁴ The modification or “remodeling” of the arteries is a process characterized by the disruption, and eventual loss of the smooth muscle cells, described as “fibrinoid necrosis” of the tunica media,^15,16 loss of elastic fibers and replacement of the endothelial lining with EVT cells which have adopted an “endovascular phenotype.”^3,4,15 Endovascular phenotype or “endovascular differentiation” of the trophoblast is characterized by gestational age-dependent acquisition of endothelial markers VE-cadherin, VCAM-1, PECAM-1 and N-cadherin.^15,17 Prior to the arrival of EVT cells, the uterine arteries, possibly primed by pregnancy-associated hormones, undergo certain morphological and molecular changes in the endothelium such as basophilia, swelling and expression of the endothelial activation marker VCAM. This is regarded as the trophoblast-independent phase of arterial remodeling,¹⁸ while subsequent completion of the remodeling process is considered as trophoblast-dependent.¹⁵ There are 2 views on EVT cell entry, presumably leading to the disruption of vascular smooth muscle cells (VSMC): (i) Certain EVT cells in the cytotrophoblastic shell (Fig. 1) undergo endovascular differentiation and invade the distal ends of the arterioles within the endometrium to cause apoptosis of the endothelial cells; they crawl against the blood-flow to replace the endothelial lining as far as the distal third of the myometrium. (ii) Interstitial EVT cells located deeper in the decidua invade the arteries from outside.^19,20 While these 2 views are not mutually exclusive, presence of interstitial EVT cells has not been documented within the myometrium. In vitro studies suggest that EVT cells when confronting myometrial smooth muscle cells produce Mel-CAM (also known as MUC18 and CD146), an anti-migratory molecule.²¹ Another feature of the arterial remodeling is the loss of elastic fibers in the tunica intima and tunica adventitia. It has been suggested that trophoblasts and VSMCs release MMP-12 in a cooperative manner to degrade elastin.²² These events transform the arteries from narrow, pulsatile and high resistance tubes to wide, non-pulsatile and low resistance vessels that allow steady flow of maternal arterial blood to the placenta for fetal nourishment. A failure of or a compromised remodeling of the arteries is frequently associated with IUGR in the fetus and or preeclampsia (PE) in the mother (Fig. 2).

Figure 2. A schematic comparison of the structures of human placentas in normal and preeclamptic pregnancy, showing poor uterine spiral arterial remodeling in PE. Poor EVT invasion of the spiral arteries is believed to compromise the arterial remodeling. This causes reduced flow of maternal arterial blood to the placental sinusoids resulting in hypo-perfused placenta that may result in IUGR in the fetus and PE in the mother. (Adapted and modified from Wang A, Rana S, Karumanchi AS. Preeclampsia: The Role of Angiogenic Factors in Its Pathogenesis. Physiology. 2009 June; 24(3):147–158.).

Patho-biology of preeclampsia

PE is a serious maternal syndrome of placental origin affecting 3–5% of pregnancies, associated with significant maternal as well as neonatal morbidity resulting from premature child birth. Clinical signs appear as unexplained hypertension and proteinuria as early as at 20 weeks of gestation in an otherwise normotensive mother. The disease is typically divided into 2 phases, the placental phase characterized by compromised trophoblast invasion and uterine arterial remodeling, and the maternal phase characterized by wide-spread maternal endothelial damage and dysfunction, possibly resulting from a variety of toxic products released from a hypo-perfused, hypoxic placenta.^23,24 These products may include lipid peroxidases, syncytiotrophoblast microvillous fragments, and inflammatory chemokines.^23-25 The placental origin of the disease is supported by the fact that a complete cure lies in the delivery of the placenta. Many studies indicate that a deficiency of trophoblast migration/invasion promoting factors or an increase in trophoblast migration/invasion restraining factors in situ may be associated with this disease (reviewed in refs. 13, 23, 24). Some of them may appear as blood biomarkers.

Factors regulating EVT cell proliferation, migration and invasiveness

A large number of molecules produced at the fetal-maternal interface were shown to exquisitely regulate trophoblast growth, migration and invasion in a positive or negative manner to maintain a healthy utero-placental homeostasis. These molecules include growth factors, inflammatory cytokines, growth factor binding proteins and proteoglycans and lipid derivatives produced by the trophoblast (autocrine) or the decidua including decidual leukocytes and immune cells (paracrine). Success in propagating pure human first trimester EVT cells in vitro, having the phenotype of EVT cells in situ,^26-28 has helped to identify some of the receptor-ligand interactions and signaling pathways regulating their proliferative, invasive and migratory functions and their alteration in PE.^13,29-31 While the mechanisms for EVT cell invasion are shared with cancer cells,^32-41 in contrast to cancer cells, EVT cell proliferation, migration and invasiveness are stringently regulated in situ.^{1,2,7,13,41-45} To explore this regulation, a number of in vitro models have been utilised: (a) when villus cytotrophoblast cells are placed on matrigel, a subset (progenitor cells) differentiated along the invasive pathway,⁴⁶ a process stimulated by EGF⁴⁷ and IL-1β.⁴⁸ When grown on plastic, they differentiated into syncytiotrophoblast,⁴⁹ but become enriched for EVT cells when plated on laminin.⁵⁰ (b) Chorionic villus explants when plated on matrigel.⁵¹ led to EVT cell sprouting that was stimulated with decidua derived activin.⁵² and uterine NK cell derived IFN-γ.⁵³ (c) Our laboratory developed a method of propagating pure first trimester EVT cells from villus explants,^26-28 duplicated by others.²¹ They express all the in situ markers of EVT: cytokeratin 7, HLA class 1 framework antigen, uPA-R, IGF-II mRNA and protein, and integrin chains α₁, α₅, α_v,β₁ and vitronectin receptor α_vb₃/β₅ but not α₆ or β₄,²⁸ and HLA-G when grown on laminin or matrigel.⁵⁴ They senesce after 5–15 passages.²⁸ Subsequently, we produced an immortalised EVT cell line HTR-8/SVneo, by SV40-Tag transfection of a short-lived line HTR-8.⁵⁵ This cell line has fully retained the normal EVT cell phenotype including expression of cell surface HLA-G. Since in vitro derived cell lines can only serve as in vitro models for studies of molecular pathways, not otherwise identifiable, they need to be verified with primary isolates of trophoblast at least in limited experiments, whenever feasible. Utilizing HTR-8 and HTR-8/SVneo cells, and villus explant cultures on matrigel, our laboratory and others have established that locally-produced molecules which regulate EVT cell functions can be broadly placed into 3 functional groups.

Factors stimulating EVT proliferation

(a) Members of the EGF family EGF,⁵⁶ TGF-α,^56,57 and amphiregulin,⁵⁸ produced in situ stimulated EVT cell growth. EGF also promoted differentiation of cytotrophoblast stem cells along the invasive pathway.⁴⁸ (b) CSF-1 produced by pregnant endometrial glands,^59-62 stimulated EVT cell proliferation but not invasiveness.⁶³ (c) Decidua-derived VEGF-A,⁶⁴ and PLGF,⁶⁵ also stimulated EVT cell proliferation, only in the absence of heparin binding domains. EVT cells expressed receptors for both VEGF-A (Flt-1, KDR) and PLGF (Flt-1).^64,65

Factors stimulating EVT migration and invasiveness

Certain trophoblast-derived molecules including IGF-II and uPA, decidua-derived molecules including IGFBP-1, endothelin-1, PGE2, and chemokines CX3CL1, CCL4, CCL14 and villus mesenchymal cell derived molecules such as HGF were shown to stimulate EVT cell migration and invasiveness.

Roles of IGF-II and IGFBP1

In situ hybridization revealed that EVT cells invading the decidua expressed IGF-II mRNA, whereas decidual cells in the proximity of invading EVT cells expressed IGFBP-1 mRNA, suggesting a molecular cross talk at the fetal-maternal interface.⁶⁶ To our surprise both molecules independently stimulated EVT cell migration and invasiveness. IGF-II actions were mediated by binding IGF-IIR, involving inhibitory G proteins and activation of ERK1/2.⁶⁷ Decidua-derived IGFBP-1 also stimulated EVT cell migration independent of IGF-I or IGF-II binding. This was mediated by binding of the Arg-Gly-Asp (RGD) domain of IGFBP-1 to α5β1 integrin on the EVT cell surface, followed by activation of FAK and MAPK.⁶⁸ Patho-physiological significance of these findings was demonstrated by low IGFBP-1 levels in early gestation sera of women destined to develop PE.^69-72 and low IGFBP-1 mRNA expression by the decidua in PE.⁷³ Increased IGFBP-1 levels in maternal blood reported during overt PE.⁷⁴ appear to be non-decidual in origin.⁷³

Migration stimulation by uPA

EVT cells produced uPA,⁷⁵ expressed high affinity uPA receptors (uPAR).⁷⁶ and showed a polarized distribution of uPAR-bound uPA on cell surface at the invasion front in situ.⁷⁷ While EVT cells used the catalytic domain of uPA for matrix degradation,^1,7 the binding of its amino-terminal to the EVT cell surface uPAR stimulated EVT cell migration resulting from an elevation of [Ca²⁺]_i, and activation of PLC, PI3K and MAPK.⁷⁸ These findings may explain the reasons for the observed uPA/uPAR imbalance in sera of PE mothers.⁷⁹

Migration stimulation by endothelin (ET)-1 and PGE2

ET1is produced by vascular endothelial cells within the placenta and the decidua. ET1 action on EVT cells was shown to follow PLC-dependent elevation of [Ca²⁺]_i and activation of MAPK.⁸⁰ PGE2, a major decidual product, was shown to stimulate EVT cell migration.^81-83 by binding to PGE receptors, in particular EP1 and EP4. EP1-mediated migratory responses resulted from an increase in [Ca²⁺]_i because of release from intracellular storage sites, and an activation of the calpain family of [Ca²⁺]_i dependent proteases, required for cell detachment from the substratum.⁸¹ Migratory responses resulting from EP1 or EP4 activity were dependent on RhoA-ROCK pathway, in which ROCK played an obligatory role, whereas RhoA action could be substituted by MAPK. However, RAC1 and CDC42 played an essential role in PGE2 or EP1/EP4 mediated migratory responses.^82,83 These findings are relevant to intervention of PE, in which trophoblast hypo-invasiveness and increased peripheral vascular resistance in the pregnant mother has been ascribed to decreased levels or responses to PGE2.

Migration stimulation by chemokines and HGF

Human decidual cells and endometrial gland cells produce a plethora of chemokines including CX3CL1, CCL4 and CCL14. They were shown to promote human trophoblast migration by binding to chemokine receptors on EVT cells.⁸⁴ Chorionic villus mesenchymal cells include a multipotent subset of stromal cells which produce many soluble cytokines and growth factors including HGF.⁸⁵ HGF exibited migration and invasion stimulatory effects on EVT cells.^85-87 HGF stimulated EVT cell motility depended on both MAPK and PI3K signaling pathways and iNOS expression.⁸⁶ HGF ligation of the c-Met receptor on EVT cells stimulated MMP-9 expression and elevation of cAMP leading to protein kinase A (PKA)-dependent Rap1 and integrin β1 activation.⁸⁵ These findings are consistent with the reports of reduced HGF production by the PE placentas ⁸⁷ and reduced soluble Met (s-Met) in the blood of PE mothers.⁸⁸

Factors blocking EVT cell proliferation, migration and invasiveness

All these EVT cell functions are controlled mostly in a paracrine manner by at least 3 decidua-derived factors TGFβ, TNFα and DCN. TGFβ is produced by the syncytiotrophoblast and most abundantly by the decidua.^27,29 This molecule exerts anti-proliferative,²⁷ anti-migratory.^68,89 and anti-invasive.^7,35,90 actions on EVT cells. The anti-migratory function resulted from an up-regulation of integrins, making the cells more adhesive to the ECM.⁸⁹ The anti-invasive action was a combined result of reduced migration, a reduction of matrix-degrading ability because of down-regulation of uPA.^35,91 and upregulation of protease inhibitors TIMP-1.³⁵ and PAI-1.⁹¹ EVT cells expressed all the TGFβ receptors (RI, II, III), downstream signaling molecules SMAD proteins (receptor associated smads 2, 3, co-smad 4, and inhibitory smads 6, 7) and responded to TGFβ by phosphorylation and nuclear translocation of smad 3⁹² and smad-dependent up-regulation of PAI-1⁹³ and TIMP-1.⁹⁴ Trophoblast cells also express a TGFβ co-receptor endoglin which binds TGFβ-1 and β3 but not β2.^95,96 A stimulation of EVT cell sprouting in villus explant cultures noted after blocking endoglin.⁹⁶ is consistent with proliferation, migration and invasion-blocking roles of endogenous TGFβ.³¹ A failure of downregulation of TGFβ-3 in villi was postulated as a precursor of preeclampsia.⁹⁷ Nodal, another member of the TGFβ superfamily inhibits EVT migration by activation of ALK7, and nodal/ALK7 overexpression is implicated in PE.⁹⁸ We discovered that choriocarcinoma cells are resistant to anti-proliferative and anti-invasive actions of TGFβ,⁹⁹ primarily due to loss of smad3 expression.^92-94 TNFα, a product of decidual macrophages was shown to inhibit EVT cell migration by up-regulating PAI-1.¹⁰⁰ We observed that a TGFβ-binding and inactivating proteoglycan DCN is co-localized with TGFβ in the decidual ECM, suggesting that DCN may serve as a reservoir for TGFβ in its inactive form,²⁹ until trophoblast-derived protease cascade at the invasion front breaks DCN-TGFβ complex and activates TGFβ to prevent over-invasion. However, to our surprise, DCN on its own was found to block EVT cell proliferation, migration and invasiveness, independent of TGFβ. An upregulation of p21 (a CDK inhibitor) explained, in part, the anti-proliferative effects. Again, choriocarcinoma cells proved to be resistant to these DCN actions.³⁰

Structural basis of DCN functions

DCN is a small leucine-rich proteoglycan (SLRP) produced in a variety of connective tissues, such as the dermis, cartilages, the decidua,^29,30 and mesenchymal cells of the chorionic villi.^29,101 A diagram of mature DCN structure derived from numerous studies^102-104 is presented in Figure. 3. It consists of a 40 kDa core protein, harbouring 10 leucine-rich repeats (LRR), covalently linked with a tissue-specific glycosaminoglycan (GAG) chain (chondroitin or dermatan sulfate) at its N-terminus of domain II. The protein is encoded by chromosome 12q22 in the human. The mature protein is highly conserved across various species and responsible for many of the biological functions. It regulates matrix assembly, including collagen fibrillogenesis because of its ability to bind collagen type1. DCN null mice survive to adulthood, but show poor fertility and skin fragility due to disruption of collagen fibrils in the dermis.¹⁰⁵ Mutation of DCN gene has been associated with a rare autosomal dominant form of corneal dystrophy called “Congenital stromal corneal dystrophy (CSCD)," also named as “Witschel dystrophy.” ¹⁰⁶ Mice genetically deficient in both DCN and biglycan genes undergo preterm labor and expulsion of the fetus owing to premature rupture of the fetal membranes.¹⁰⁷ Because of its TGFβ binding ability at LRR4 and 5, DCN can store TGFβ in the ECM.¹⁰³ It can inactivate mature TGFβ in some circumstances¹⁰⁸ but not others,¹⁰⁹ apparently by interfering with TGFβ-binding to types I and III TGFβR. Whether this occurs universally in vivo is unclear, since the inactivation improves in vitro after removal of the GAG chains.¹¹⁰ DCN can also bind myostatin, another member of the TGFβ superfamily and a myogenic inhibitory protein.¹¹¹ This binding has been shown to support myogenesis by inhibition of myostatin. For example, DCN immobilized on ECM in vitro sequesters myostatin into the ECM, preventing its inhibitory action on myoblast proliferation. Furthermore, ectopically introduced DCN into myoblasts promotes their proliferation and differentiation by inactivating endogenous myostatin.¹¹² DCN can modulate cell adhesion by binding with ECM molecules fibronectin^113,114 or thrombospondin.¹¹⁵ However, as elaborated below, many DCN functions are also receptor-mediated.

DCN acts as an antagonistic ligand to numerous tyrosine kinase receptors

EGFR was the first DCN receptor identified by the Iozzo group^116-120 having a lower affinity than EGF. DCN-binding led to a transient phosphorylation of EGFR kinase, followed by a reduction of the receptor number, attributed to accelerated receptor endocytosis by the caveolar pathway. DCN binding was mapped within a narrow region of the ligand binding domain 2 of the EGFR, partially overlapping with the EGF-binding domain.¹¹⁹ A second DCN receptor IGF-IR was soon identified on endothelial cells that could bind DCN with an affinity comparable to IGF-1, showing DCN induced receptor phosphorylation, followed by receptor down-regulation.¹²¹ However DCN could also bind IGF-I, compounding the biological effects of DCN on endothelial cells in the presence of IGF-I in the micro-environment.¹²¹ Subsequently DCN was shown to be an antagonistic ligand for Met, the receptor for HGF.¹²² Since DCN core protein and its 26-amino acid residue LRR5 were shown to inhibit VEGF-induced angiogenesis in endothelial cells in vitro,¹²³ we suspected that the anti-angiogenic action of DCN may have resulted from DCN binding to VEGFR-2.

TKR mediated DCN actions of EVT cells and endothelial cells

We investigated whether EGFR, IGF-IR and VEGFR-2 played a role in negative regulatory actions of DCN in EVT cells.³¹ We noted that these TKRs are promptly phosphorylated by DCN similar to their natural ligands, and these events can be blocked with respective tyrosine kinase inhibitors. We found that DCN inhibits EVT cell outgrowth in situ from intact first trimester chorionic villous explants on matrigel. Using our EVT cell line HTR-8/SVneo, we further showed that anti-proliferative action of DCN is mediated primarily by EGFR and VEGFR-2, and anti-migratory action primarily by IGF-IR. Furthermore, VEGF-A mediated migration of EVT cells was blocked with DCN in a VEGFR-2 dependent manner.¹²⁴

Molecular localization of VEGFR-2 binding on DCN protein

Far-Western blotting and co-immunoprecipitation studies revealed that DCN binds both native (EVT or endothelial cell expressed) and recombinant VEGFR-2 and that this binding is abrogated with a VEGFR-2 blocking antibody, indicating an overlap between the ligand-binding and the DCN-binding domains of VEGFR-2¹²⁵ I-labeled VEGF-E (a virus-derived VEGFR-2 specific ligand) bound to EVT with a K_d of 566 pM, and DCN displaced this binding with a K_i of 3.93–5.78 nM, indicating a 7 to 10-fold lower affinity of DCN for VEGFR-2. DCN peptide fragments derived from its LRR- 5 domain that blocked DCN-VEGFR-2 interactions or VEGF-E binding in EVT cells also blocked VEGF-A and VEGF-E-induced EVT cell proliferation and migration, indicative of functional VEGFR-2-binding sites of DCN in a 12 AA span of LRR5 (Fig. 4). This binding inhibited VEGF-induced EVT migration by interfering with ERK1/2 activation. Subsequently we explored the mechanisms for VEGFR-2 dependent DCN antagonism of migration and endovascular differentiation of EVT cells.¹²⁵ We found that DCN inhibited VEGF-induced EVT cell migration and endothelial-like tube formation (an in vitro surrogate of endovascular phenotype). VEGF activated p38 MAPK, MEK3/6, and ERK1/2 in EVT cells, and the activation of these pathways was blocked by DCN. Employing selective MAPK inhibitors, we showed that both p38 MAPK and ERK pathways contributed independently to VEGF-induced EVT migration and tube formation. VEGF-mediated up-regulation of endothelial markers VE-cadherin and β-catenin in EVT as well as endothelial cells was blocked by DCN and MAPK inhibitors. These results revealed that DCN inhibits VEGF induced EVT migration and endovascular differentiation by interfering with p38 MAPK and ERK1/2 pathways in parallel.¹²⁵

Figure 3. A schematic diagram of the structure of human decorin. It contains backbone of a 40kDa core protein containing 4 domains. Domain-I contains a propeptide and a signal peptide (in green) which are cleaved in the matured protein. Domain-II is covalently linked with a GAG chain at its N-terminus. Domain-III contains 10 Leucine Rich Repeats (LRR) and N-linked Glycan side-chains (in orange). Domain -IV contains cysteine residues and forms a large loop.The TGFβ binding sites are located in LRR4 and LRR5. The VEGFR-2 binding site is located within LRR5. The binding site for myostatin is currently undetermined.

Figure 4. A schematic diagram of the VEGFR-2 binding site in a 12 aa span of LRR5 in decorin protein core. The VEGF-A and decorin binding sites of VEGFR-2 are overlapping. (Reproduced with permission from Khan GA, Girish GV, Lala N, Di Guglielmo GM, Lala PK. Decorin is a novel VEGFR-2-binding antagonist for the human extravillous trophoblast. Mol Endocrinol. 2011 Aug; 25(8):1431–43.).

Mechanisms of DCN actions on the trophoblast

Decidua-derived DCN appears to be a powerful negative regulator of trophoblast migration and invasion independent of TGFβ binding; this was primarily due its binding and antagonizing several TKRs including EGFR, IGF-IR, Met, and VEGFR-2 which promote these functions when stimulated by their natural ligands. The underlying mechanisms are most likely similar to those reported with other cell types. In the case of EGFR, DCN binding is followed by a rapid receptor endocytosis and degradation by the caveolar pathway and a retardation of the receptor recycling.¹²⁰ A rapid receptor internalization and receptor shedding were noted after DCN binding to Met, followed by downregulation of β-catenin and Myc.¹²² DCN antagonized IGF-IR functions in tumor cells by interfering with IGF-IR activity and attenuating downstream signaling, but no change in receptor endocytosis.¹²⁶ In the case of VEGFR-2 binding to DCN, we found no evidence of accelerated receptor endocytosis in comparison to VEGF binding on EVT cells (Lala PK and Girish GV, unpublished data). VEGFR-2 binding compromised trophoblast migration and endovascular differentiation by interfering with p38 MAPK and ERK1/2 pathways in parallel.¹²⁵ It is highly likely that other TKRs with DCN binding ability will be identified in the future on cells including trophoblast.

Role of DCN in angiogenesis

While most studies have established that DCN blocks angiogenesis, some studies suggested a pro-angiogenic role. This difference is ascribed to differences in the cellular and molecular environment in DCN action, elegantly reviewed by Jarvelainen et al.¹²⁷ Anti-angiogenic actions: Many studies including ours have identified DCN as a potent anti-angiogenic molecule due to its actions on vascular endothelial cells, mediated by multiple mechanisms: (i) In vitro studies with vascular endothelial cells revealed that DCN binding to the ECM molecule thrombospondin-1 enhances anti-angiogenic role of this molecule.¹²⁸ (ii) DCN binding to IGF-IR blocked angiogenesis in tumor cells,¹²⁶ in contrast to increased motility reported for endothelial cells grown on type 1 collagen matrix.¹²⁹ (iii) VEGFR-2, the most potent pro-angiogenic receptor on endothelial cells, antagonizes VEGF-mediated migration and tube formation both by the trophoblast and endothelial cells.^124,125 Inhibition of VEGF-induced angiogenesis by DCN or certain DCN peptides reported earlier¹²³ can be explained on the basis of VEGFR-2 binding of these molecules demonstrated by us.¹²⁴ (iv) DCN mediated down regulation of VEGF production by tumor cells was shown as another mechanism for suppression of tumor-induced angiogenesis in vitro and in vivo. This was demonstrated by addition of exogenous DCN to tumor cells or ectopic DCN expression in tumor cells.¹³⁰ (v) DCN can also act on VSMC by binding to PDGF, a growth promoter of VSMC. DCN binding to PDGF inhibited PDGF-stimulated PDGFR phosphorylation and proliferation of VSMC, required for angiogenesis.¹³¹ Pro-angiogenic actions: DCN null mice exhibited reduced corneal angiogenesis in response to injury, suggesting a pro-angiogenic role of DCN.¹²⁷ However, this may possibly be explained by upregulation of pro-angiogenic biglycans.¹³² or activation of endogenous TGFβ which can have indirect angiogenic functions.¹³³ DCN null mice were shown to up-regulate biglycans¹⁰⁷ which were reported to have angiogenic ability.¹³² Finally, targeted DCN gene therapy blocked angiogenesis in mouse cornea,¹³⁴ reinforcing its anti-angiogenic role. Other pro-angiogenic actions of DCN can possibly be explained in the context of DCN-ECM interactions, favoring the assembly of a pro-angiogenic matrix.^127,135 For example, endothelial tube formation was stimulated in DCN over-expressing cells placed in a 3D matrix which allowed DCN binding to both IGF-IR and IGF-I.¹²¹ It is also possible that the nature of microvasculature may dictate DCN action. For example, ectopic DCN over-expression in mouse cerebral endothelial cells upregulated VEGF-A to promote angiogenesis.¹³⁶

Role of DCN in tumor progression

Interestingly, many of the DCN actions in restraining placental invasion of the pregnant uterus and its vasculature are also applicable to tumor progression. Currently there is a substantial amount of literature suggesting that DCN has tumoristatic functions in certain cancers because of its dual roles: limiting angiogenesis and antagonising TKRs having oncogenic roles because of their mutational activation. Additional mechanisms have also been found. It has been suggested DCN is a natural tumor-suppressor molecule that can be harnessed for therapy. Indeed, SLRPs including DCN have been shown to play an important role in health and disease including cancer.¹³⁷ These molecules produced by stromal cells in ECM within the tumor micro-environment are believed to be important contributors in controlling tumor growth.^137,138 In support, the relative level of DCN expression in soft tissue tumors has been correlated with prognosis.¹³⁹ Some studies examined the effects of either ectopic introduction of DCN gene, or a deletion of DCN gene. For example, Grant et al.¹³⁰ demonstrated reduced angiogenic abilities of DCN transfected fibrosarcoma, osteosarcoma and carcinoma cell lines tested in vitro and in vivo. DCN transfection of breast carcinoma cells was shown to antagonise an elaborate angiogenic network. There was concurrent inhibition of Met, HIF-1α and induction of angiostatic molecules thrombospondin-1 and TIMP-3.¹⁴⁰ Deletion of DCN gene promoted colorectal carcinogenesis in a mouse model.¹⁴¹ Using a DCN knockout mouse model, these authors demonstrated that colorectal carcinogenesis resulting from DCN knockdown was associated with epithelial mesenchymal transition (EMT), as marked by down-regulation of E-cadherin and up-regulation of β-catenin.¹⁴² Goldoni et al.¹⁴³ demonstrated anti-metastatic role of exogenous DCN treatment in experimental breast cancer. They compared the effects of DCN alone or in combination with an ErbB2 kinase inhibitor AG879 on the growth of an ErbB2-over-expressing mammary carcinoma cell line. DCN was shown to inhibit ErbB2 kinase in tumor cells. When DCN and AG879 were used in combination, the inhibitory effects were synergistic or additive on tumor growth in vitro. While the combination therapy did not improve anti-tumor effects of DCN on primary tumor growth in vivo, DCN significantly reduced lung metastasis. Similarly DCN and carboplatin were shown to have synergistic growth-inhibitory effects on ovarian cancer cells.¹⁴⁴ However, caution has been raised against DCN therapy in human pancreatic cancer.¹⁴⁵ In pancreatic cancer tissues DCN overexpression was noted in stromal satellite cells possibly as a protective defense mechanism. While DCN alone had cytostatic effects on several pancreatic cancer cell lines in vitro, it attenuated cytostatic effects of carboplatin or gemcitabine in some cell lines. It is likely that DCN will be useful in combination therapies in tumors with low DCN expression in situ. Furthermore, genetic mutations may lead to DCN resistance in certain cancers. For example, in contrast to the migration and invasion restraining roles of decidua-produced DCN as well as TGFβ on the normal human trophoblast cells, as detailed earlier, we found that human choriocarcinoma cells are resistant to both molecules,^30,92-94,99 explaining their emancipation from decidua-mediated controls.

Possible role of DCN over-expression in the decidua in pathogenesis of PE

DCN mediated dual impediment of endovascular differentiation of the EVT and angiogenesis,^124,125hall marks of PE, may imply that DCN overexpression at the fetal-maternal interface may contribute to the pathogenesis of PE. We tested this contention by quantitating DCN expression in placentas from 16 healthy and 16 PE pregnancies at comparable gestational ages using 2 approaches.¹⁴⁶ DCN mRNA expression at the tissue level was quantified with real time qPCR in chorionic villus tissues. DCN mRNA at the cellular level was measured both in chorionic villi and decidua basalis employing in situ hybridization with³⁵S labeled antisense cRNA probe. Hybridization with sense cRNA probe served as controls. Using both approaches we found that DCN expression by chorionic villus mesenchymal cells was unaltered in PE. In contrast, DCN mRNA expression by decidual cells in the basal plate (where EVT cells invade) was profoundly up-regulated in PE placentas, as compared to healthy placentas of equivalent gestational ages. Immunocytochemical localization of DCN in the same tissues confirmed these findings at the protein level. We suggest that over-expression of DCN in the basal decidua contributes to the pathogenesis of PE because of a dual action of DCN on the trophoblast and endothelial cells. In the former DCN causes a hypo-invasive phenotype and poor endovascular differentiation. Anti-angiogenic action of DCN on endothelial cells in combination with poor endovascular differentiation of the trophoblast contribute to poor uterine arterial remodeling. In addition, we discovered that exposure to exogenous DCN or ectopic overexpression of DCN in EVT cells significantly upregulated the expression of sFlt-1. This anti-angiogenic molecule produced by the trophoblast in PE placentas is recognized as a blood biomarker in PE mothers.¹⁴⁷ All these events are patho-biological hall marks of placentas associated with PE. We suggest that DCN in the maternal blood is another potential biomarker in PE. This suggestion has recently been validated by longitudinal studies of maternal plasma samples in pregnant subjects during the second trimester (Siddiqui M et al. submitted for publication).

Gaps in knowledge and future directions

DCN is a multifunctional molecule, the functions mediated by a variety of binding events that regulate cell adhesion, migration and invasion. DCN can inactivate TGFβ under certain conditions, but whether this happens in situ in the decidua where they are co-localized, remains to be explored. DCN binds numerous TKRs which have been identified; it may possibly bind to other TKRs that remain to be identified. It is highly likely that some of the DCN actions on the trophoblast are receptor-independent, mediated by binding to ECM molecules that remain to be identified. We have preliminary evidence that DCN binding to EVT cells compromises many other functional events such as intercellular communications that remain to be fully characterized at the molecular level. The role of DCN on VSMC and trophoblast-VSMC interactions remain an area of investigation in the context of uterine arterial remodeling. Very little information is available as to whether DCN acts as a protective molecule in trophoblastic tumorigenesis or metastasis to the placenta from pregnancy-associated cancers in maternal organs including the breast. Recently Borbely et al.¹⁴⁸ reported immunohistochemical localization of DCN protein in trophoblast cells of invasive pregnancies including placenta acreta accreta, invasive moles and choriocarcinomas, the significance of which remains unknown. One possibility is that these hyper-invasive placental pathologies involve epithelial-mesenchymal transition of the trophoblast, allowing cells to make this mesenchymal molecule. In situ hybridization is needed to confirm the cellular source of DCN in these pathologies.

Abbreviations

BP	=	binding protein
DCN	=	decorin
ECM	=	extracellular matrix
EGF	=	epidermal growth factor
EVT	=	extravillous trophoblast
EPH	=	ephrin
ET	=	endothelin
EP	=	PGE receptor
GAG	=	glycosaminoglycan
HGF	=	hepatocyte growth factor
HIF	=	hypoxia inducible factor
HLA	=	human leukocyte antigen
IGF	=	insulin like growth factor
IUGR	=	intra-uterine growth restriction
IL	=	interleukin
IFN	=	interferon
Met	=	HGF receptor
MMP	=	matrix metalloprotease
PDGF	=	platelet derived growth factor
PE	=	preeclampsia
PECAM1	=	platelet/endothelial cell adhesion molecule 1
PGE	=	prostaglandin E
PAI	=	plasminogen activator inhibitor
R	=	receptor
SLRP	=	small leucine rich proteoglycan
TGF	=	transforming growth factor
TKR	=	tyrosine kinase receptor
TNF	=	tumor necrosis factor
TIMP	=	tissue inhibitor of metalloproteinase
uPA	=	urokinase type plasminogen activator
VEGF	=	vascular endothelial growth factor
VCAM	=	vascular cell adhesion protein
VSMC	=	vascular smooth muscle cell

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

Research collaboration from Dr. Victor Han (Department of Pediatrics at the University of Western Ontario) is gratefully acknowledged.

Funding

Studies from the authors' lab reported here were funded by grants from the Canadian Institute of Health Research (CIHR) and the Children's Health Research Institute (London, ON).