Extravillous trophoblasts invade more than uterine arteries: evidence for the invasion of uterine veins
- First Online:
- Cite this article as:
- Moser, G., Weiss, G., Sundl, M. et al. Histochem Cell Biol (2017) 147: 353. doi:10.1007/s00418-016-1509-5
During the first trimester of pregnancy, extravillous trophoblasts (EVTs) invade into the decidual interstitium to the first third of the myometrium, thereby anchoring the placenta to the uterus. They also follow the endovascular and endoglandular route of invasion; plug, line and remodel spiral arteries, thus being responsible for the establishment of hemotrophic nutrition with the beginning of the second trimester and invade and open uterine glands toward the intervillous space for a histiotrophic nutrition during the first trimester. The aim of this study was to provide proof that uterine veins are invaded by EVTs similar to uterine arteries and glands in first trimester of pregnancy. Therefore, serial sections from in situ first trimester placenta were immuno-single- and immuno-double-stained to distinguish in a first step between arteries and veins and secondly between invaded and non-invaded vessels. Subsequently, invasion of EVTs into uterine vessels was quantified. Our data show that uterine veins are significantly more invaded by EVTs than uterine arteries (29.2 ± 15.7 %) during early pregnancy. Counted vessel cross sections revealed significantly higher EVT invasion into veins (59.5 ± 7.9 %) compared to arteries (29.2 ± 15.7 %). In the lumen of veins, single EVTs were repeatedly found, beside detached glandular epithelial cells or syncytial fragments. This study allows the expansion of our hitherto postulated concept of EVT invasion during first trimester of pregnancy. We suggest that invasion of EVTs into uterine veins is responsible the draining of waste and blood plasma from the intervillous space during the first trimester of pregnancy.
KeywordsExtravillous trophoblasts Endovascular trophoblasts Endoglandular trophoblasts Uterine veins Invasion Placenta
Extravillous trophoblasts (EVTs) of the decidua basalis originate from trophoblastic cell columns of anchoring villi and invade into maternal uterine decidua finally reaching the inner third of the myometrium (Kaufmann et al. 2003). Invasion of EVTs serves to attach the placenta to the uterus (interstitial invasion) and is responsible for the accession of nutrients to the embryo within the placenta (endovascular and endoglandular invasion) (Moser et al. 2010, 2015; Pijnenborg et al. 2006, 2011).
Spontaneous decidualization is initiated in the late secretory phase (around day 22 of the menstrual cycle). Thus, decidualization typically starts prior to implantation under the influence of progesterone, and the first signs become visible on day 23 of the menstrual cycle. Following implantation, decidualization is enhanced by high levels of progesterone (de Ziegler et al. 1998). It involves the endometrial stroma and the vessel walls of the spiral arteries. Thereafter, the presence of endovascular trophoblasts is responsible for the onset of “trophoblast-associated remodeling” of the spiral arteries. During this well-described process, an amorphous fibrinoid material is deposited and replaces the original smooth muscle layer of the vasculature, together with a complete loss of the elastic lamina (Pijnenborg et al. 2011). The decidua-associated spiral artery remodeling shows swelling and vacuolation of the endothelium together with disintegration of the vascular smooth muscle layer and swelling of individual muscle cells. Uterine natural killer cells and macrophages are important triggers of this early remodeling step (Pijnenborg et al. 2011).
The structure of uterine veins differs from that of spiral arteries in some minor aspects. The muscular layer is reduced, and near the venous openings to the intervillous space, smooth muscle cells are usually completely absent. In the rhesus monkey, the endothelium in venous segments near the intervillous space has sometimes been replaced by intraluminal trophoblast cells (Blankenship et al. 1993; Frank and Kaufmann 2006). However, so far nobody has had a specific look at uterine veins in the invaded regions of the first trimester decidua basalis. Aside that, in strongly invaded regions some of the arteries are not invaded at all, while veins seemed to be invaded by extravillous trophoblasts. Thus, the aim of this study was to provide proof that uterine veins are invaded by extravillous trophoblasts in the same manner as uterine arteries and glands in first trimester of pregnancy. Therefore, serial sections from in situ first trimester placenta were immuno-single- and immuno-double-stained to distinguish in a first step between arteries and veins and secondly between invaded and non-invaded vessels. Subsequently, invasion of EVTs into uterine vessels was quantified.
Materials and methods
First trimester placentas were obtained from elective surgical terminations of pregnancies [gestational age (GA) 5–11 weeks, n = 41]. Informed consent was obtained from each woman included in the study with approval of the ethics committee of the Medical University of Graz. From every placenta, various tissue samples (villi, decidua basalis, decidua parietalis) were collected, depending on availability in the respective specimen. Invaded decidua basalis was identified in 16 (n = 16, GA 6–11 weeks) out of the 41 cases. For the preparation of formalin-fixed paraffin-embedded (FFPE) sections, tissues were fixed in 4 % neutrally buffered formalin for at least 24 h and routinely embedded in paraffin.
Preparation of sections
Serial 5-µm paraffin sections were cut and placed on Superfrost Plus slides (Menzel, Braunschweig, Germany). FFPE sections were deparaffinized in xylene and rehydrated through a graded series of alcohol. Heat-induced antigen retrieval was performed in antigen retrieval solution at pH 9 (Leica Biosystems, Nussloch, Germany) in a pressure cooker (Model DC2002, Biocare Medical, Concord, USA) for 7 min at 120 °C before immunohistochemistry.
Acris (Herford, Germany)
Rabbit IgG pc
Human desmin (M 0760)
DakoCytomation (Pleasanton, Canada)
Mouse IgG mc
BD Pharmingen (Vienna, Austria)
Mouse IgG mc
Sigma-Aldrich (St. Louis, USA)
Rabbit IgG pc
Smooth muscle actin (1A4/M0851)
DakoCytomation (Pleasanton, Canada)
Mouse IgG mc
EphB4 (D1C7 N)
Cell signaling (Danvers, USA)
Rabbit IgG mc
Mouse IgG1 (DAK-GO1)
Dako (Carpinteria, USA)
Matched to each primary antibody
Mouse IgG mc
Rabbit immunoglobulin fraction (X 0903)
Dako (Carpinteria, USA)
Matched to each primary antibody
Rabbit IgG pc
Immunohistochemical double staining
Immunohistochemical double staining was performed using the Multivision Polymer Detection system (MultiVision anti-rabbit/AP + anti-mouse/HRP polymers; Thermo scientific, Fremont, USA) according to the manufacturer’s instructions. Primary antibodies were diluted and mixed in antibody diluent (Dako, Carpinteria, USA), and this primary antibody cocktail was applied for 30 min at room temperature to the tissue sections. Table 1 lists details of all antibodies used and their respective dilutions. For this protocol, a hematoxylin counterstain is not recommended.
Quantification of trophoblast invasion
Type of vessel-criteria for classification
Type of vessel
Positive staining for vWF, clear multi-layered (artery) or single-layered (arteriole) tunica media and/or positive staining for desmin
Positive staining for vWF, tunica media not visible, often collapsed, irregular shape, no staining for desmin
Positive staining for vWF
Vessels with diameter lower than ~10 µm (capillaries)
Vessels completely surrounded by EVTs (maybe, the tunica media have already been completely replaced by EVTs)
Vessels where a clear discrimination between artery and vein was not possible
EVTs in the lumen of the vessel and/or endothelium replaced by EVTs
EVTs attached to the vessel wall from the outer side
EVTs in the surrounding interstitium, but not associated with the vessel
A subset of three out of the 16 placentas was assessed additionally with an immuno-staining against smooth muscle actin (smA) and hematoxylin and eosin (H&E) staining. The same subset was used for automated image acquisition by the VIS software and counted by one observer. Another subset of 10 out of the 16 placentas was assessed additionally with an immuno-staining against EphB4 and an immune-double staining against HLA-G and EphB4.
Data are reported as means ± standard deviations. Student’s t test was applied for the quantification of EVTs replacing epi-/endothelium and EVTs in spatial proximity between glands and vessels, after testing for normal distribution (Kolmogorov–Smirnov test). Statistical analysis was done using SPSS IBM Statistics 21. A p value <0.05 was considered significant.
Endovascular trophoblasts in arteries and veins: qualitative characteristics
In one-third of the cases, despite the application of the above-described morphological criteria and the desmin staining in the serial section, a classification between arterial or venous vessels was not possible. This applied mostly for vessels with a diameter smaller than ~10 µm (Fig. 5a–b) and for vessels completely surrounded by EVTs (Fig. 5c–f). For the latter, it could be speculated that the tunica media have already been completely replaced by EVTs. For the quantification approach, all these vessels were defined as “vessels unclassified” and were excluded from semi-quantitative analysis (33 % of all counted vessels). The detailed criteria for inclusion and exclusion to the arterial/venous system are listed in Table 2.
Endovascular trophoblasts in arteries and veins: quantitative analysis
Control 1: Comparison—manually and automated image acquisition
In a subset of three placentas, results revealed no significant differences between the two image selection methods (manually or automated systematically randomly selected). Setting the value for manual selection to 100 %, neither in invaded arteries (manual: 100 ± 23 %; automated: 170 ± 58 %), nor in invaded veins (manual: 100 ± 23 %; automated: 99 ± 17 %) a significant difference could be detected (Fig. 6b).
Control 2: Additional smA and H&E staining
As a control for the setup of quantification described above, in a subset of three placentas respective serial sections were additionally immuno-stained with an antibody against smooth muscle actin (smA). The immuno-staining against smA confirmed no or only a weak muscular layer in the vessels classified as vein (Fig. 2e, f). Serial sections from the same subset were additionally stained with hematoxylin and eosin (H&E) for a clear and precise morphology (Fig. 4d). Both smA-immuno-staining and H&E-staining confirmed the classification into artery and vein, previously performed with combination vWF/HLA-G double staining and desmin staining.
Discrimination between arteries and veins was additionally tested with immune-single and immune-double staining with the venous marker EphB4. Several Eph4-positive vessels were associated with and invaded by EVTs (Fig. 3a–d). Morphologically unambiguous arteries react usually negative with EphB4 (Fig. 3e, f), but sporadically the arterial endothelium displays positive staining for EphB4 (Fig. 3g, h).
Glandular epithelial cells in the lumen of vessels
“Uterine arteries are invaded by extravillous trophoblast, while veins are not” is the generally accepted opinion in placental research and teaching. Maybe, we need to expand our view! Our data show that uterine veins are invaded by EVTs even to a greater extent than uterine arteries in early pregnancy. Besides that, we found detached glandular epithelial cells or syncytial fragments (KRT7-positive, but HLA-G negative) in the lumen of such veins. Structural identification is considerably facilitated by application of respective immunohistochemical markers. As previously suggested, we recommend the utilization of immunohistochemical double staining for HLA-G in combination with vWF/KRT7 as specific markers for extravillous trophoblasts to avoid potential misinterpretation between EVTs and glandular epithelial cells (Moser et al. 2011).
The number of counted vessel cross sections varied considerable between the individual placentas. However, invaded arteries and veins were found in all assessed placentas. Notably, more than every second venous cross section was invaded by EVTs. Vessels completely surrounded by EVTs have been excluded from quantification, since it could be speculated that here the tunica media of an artery have already been completely replaced by EVTs. Approximately one quarter of the excluded “vessels unclassified” were vessels completely surrounded by EVTs. For this quarter, it could be speculated that all of them were completely remodeled arteries. However, even if these “arteries” would be added to the arterial invasion, there is still an approximately equal invasion toward arteries and veins. This exclusion may skew our results and needs to be mentioned as a possible limitation.
So far, the focus was on the invasion and transformation of uterine spiral arteries and it seems as if invasion of veins has been ignored to date. In Benirschke et al. (2006) uteroplacental veins of the first trimester are described as “Endothelial tubes, surrounded by few regressive medial and adventitial cells, embedded in decidua and extravillous trophoblast cells; the latter rarely invade the venous walls and never the venous lumina” (Benirschke et al. 2006). However, we repeatedly observed EVTs in the lumen of veins, and Fig. 2b provides respective evidence. Also Pijnenborg et al. (2006) stated it as fact that endovascular invasion only occurs within arteries and never into veins, but the authors did not provide any evidence for this (Pijnenborg et al. 2006).
A look into the historical literature shows that already in 1941 decidual capillaries and venules were observed to link with the trophoblast lacunae at 16 days of development (Hertig and Rock 1941; Schneider and Moser 2016). Decades later, Kam et al. (1999) described that the veins in the uterine mucosa in the absence of trophoblast showed no endothelial swelling and few actin-positive medial cells, whereas in the decidua basalis trophoblasts were present around some veins, but without modification of vessel walls (Kam et al. 1999). Craven et al. (2000) investigated tissues from 100 first trimester placentas; all of these decidual tissues had dilated veins. They described that extravillous trophoblast cells were attached to the endothelium and cells appeared to invade the underlying stroma. Moreover, they described the presence of placental villi and cell islands (described as clusters of mononuclear trophoblast cells) in the lumen of uterine veins in the decidua and deep within the myometrium. These cell islands were completely detached from the placental villi, confirmed by serial sections (Craven et al. 2000).
It would be of great interest to stain such serial sections with the respective markers to distinguish between fetal villous and extravillous trophoblast cells and maternal glandular epithelial cells. We suggest that also detached uterine glandular epithelial cells are passing uteroplacental veins and thus can be found in the lumen of such veins. These epithelial cells have been removed from their original site, uterine glands, by invading endoglandular trophoblasts (Moser et al. 2015), have passed the intervillous space and are now drained into the maternal circulation.
In their study from 2000, Craven et al. suggest that trophoblast cell islands and villi enter maternal veins, floating villi implant in veins (as part of lateral placental growth), and link this observation with the transformation of the venous wall to a new basal plate tissue (Craven et al. 2000). The same authors compared and examined in another study the late secretory endometrial biopsy with first trimester decidua (Craven and Ward 1999) and showed that the decidual veins were dilated, but there were no dilated veins in the secretory endometrium. The dilated veins in the decidua often contained syncytiotrophoblastic fragments; many of them associated with blood clots. There was no correlation of the gestational age of a case with the number of syncytiotrophoblast fragments in the uteroplacental veins (range of gestational age 7–11 weeks) (Craven and Ward 1999). In another study, Craven et al. (2002) identified fibrin deposition in decidual veins that was associated with trophoblast cell invasion. They also described that the endothelium became incomplete at sites of trophoblast attachment; this was accompanied with the deposition of fibrin along the vein wall (Craven et al. 2002). We also observed an incomplete endothelium in veins (Fig. 7). However, in all of their studies Craven et al. were indeed describing attachment of trophoblast cells from the luminal side, but they never described invasion into veins (Craven and Ward 1999; Craven et al. 2000, 2002). Besides that, we want to suggest that the fibrin described by Craven et al. 2002 represents actually fibrin-type fibrinoid according to the definition in (Kaufmann et al. 1996).
We frequently observed non-invaded arteries in invaded and also strongly invaded decidual regions (Fig. 4). Despite massive EVT invasion in the surrounding, there appears no loss of the smooth muscle layer in these arteries. This is a quite common occurrence, but—to the best of our knowledge—has not been mentioned yet in the literature. Assessment of serial sections showed that the same artery may be invaded by EVTs in other portions of the vessel (data not shown).
There is still only limited knowledge on the specific types of fetal cells circulating in maternal blood. Recently, fetal cells in maternal blood were found to express a typically extravillous gene pattern. Due to this gene expression pattern, the authors suggest that it is likely that a part of the fetal cells present in maternal blood are EVTs. These cells exhibit expression of markers with invasive capability, for endothelial cell-to-cell adhesion and at the same time retaining epithelial markers (Hatt et al. 2014). We suggest that these EVTs reach the maternal circulation via invasion of uterine arteries and veins as well as via the endoglandular pathway: EVTs invade uterine glands and are flushed together with the glandular secretion products (and detached glandular epithelial cells) toward the intervillous space. Putative syncytial fragments, KRT7-positive, but HLA-G negative, may be present in the lumen of invaded veins as well. Finally, they are drained into the maternal circulation via the opened and already dilated uterine veins. This thesis may also be an explanation why uterine veins are dilated in the decidua during very early pregnancy.
According to Smith et al. (2009) there are trophoblast-independent and trophoblast-dependent phases of remodeling; extensive disruption and remodeling of the spiral arteries occur before endovascular colonization by EVTs and are coincident with vascular infiltration by uNK and macrophages (Smith et al. 2009). Taken together, even with visualization of the smooth muscle layer, and since the majority of smooth muscle cell loss from the spiral arteries in early stages of remodeling occurs before EVT presence (Smith et al. 2009), the discrimination between a vein with little smooth muscle and a remodeled artery that has lost all its smooth muscle cells is tricky. Ephrin-B2, an Eph family transmembrane ligand, and its receptor EphB4 have been attributed to be markers for arteries and veins, respectively (Wang et al. 1998). These ligand–receptor interactions with various members of the ephrin/eph family may stimulate cytotrophoblast migration toward arteries and veins (Red-Horse et al. 2005). Another study reported a more widespread pattern of ephrin/eph expression throughout the vascular system (Adams et al. 1999). It needs to be mentioned that the studies by Wang et al. (1998) and Adams et al. (1999) have been conducted with mice and not with human specimens. Human placental venous endothelial cells did not overexpress EphB4 compared to arterial endothelial cells as demonstrated by Lang et al. (2008). We observed that EphB4 occasionally stains the endothelium of morphologically unambiguous arteries (Fig. 3g, h). Thus, EphB4 may serve as helpful marker and additional control, but one should refrain from taking it as a single criterion for discrimination between arteries and veins.
However, and viewed from the opposite side, the majority of all vWF-positive vessels in the invaded decidua basalis are associated with and invaded by EVTs. From this point of view, it seems questionable whether it can be true that the majority of all vessels are arteries and why the venous part should be so extremely underrepresented in the decidua? And even if some or even numerous of the vessels have been classified wrongly, the fact that the majority of the vessels are invaded by EVTs gives enough hint for a venous invasion by EVTs.
In conclusion, EVTs strongly invade uterine veins beside the decidual stroma, spiral arteries and uterine glands. Detached glandular epithelial cells or putative syncytial fragments (KRT7-positive, but HLA-G negative) can be detected in the lumen of such veins. All potential routes of EVTs invasion during the first trimester are schematically represented in Fig. 8. Because of the large amount of venous EVT invasion, we suggest that there is a biological relevance to invade and open veins toward the intervillous space for draining of blood plasma, glandular secretion products and debris before the establishment of the uteroplacental blood flow at the end of the first trimester. Whether or not pathologies derive from failures of venous invasion has not been addressed yet but may be the subject of further investigation.
Open access funding provided by Medical University of Graz. Thanks to Dr. Andreas Glasner, Astrid Blaschitz, Rudolf Schmied, Amin El-Heliebi and Nina Schlögl for their valuable help and expertise. This work was supported by the Austrian Science Fund (Grant P24739-B23, Granted to G.M.) and the funds of the Oesterreichische Nationalbank (Oesterreichische Nationalbank, Anniversary Fund, project number: 16513, granted to M.G.). G.M. is funded by the Post Doc Program of the Medical University of Graz and by the Austrian Science Fund (Grant P24739-B23). G.W. is funded by the Austrian Science Fund (Grant P24739-B23).
Compliance with ethical standards
Conflict of interest
No competing financial interests exist.
|Funder Name||Grant Number||Funding Note|
|Austrian Science Fund|
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.