Human Extravillous Trophoblasts Penetrate Decidual Veins and Lymphatics before Remodeling Spiral Arteries during Early Pregnancy

In humans, the defective invasion of the maternal endometrium by fetal extravillous trophoblasts (EVTs) can lead to insufficient perfusion of the placenta, resulting in pregnancy complications that can put both mother and baby at risk. To study the invasion of maternal endometrium between (W)5.5–12 weeks of gestation by EVTs, we combined fluorescence in situ hybridization, immunofluorescence and immunohistochemistry to determine the presence of (male) EVTs in the vasculature of the maternal decidua. We observed that interstitial mononuclear EVTs directly entered decidual veins and lymphatics from W5.5. This invasion of decidual veins and lymphatics occurred long before endovascular EVTs remodelled decidual spiral arteries. This unexpected early entrance of interstitial mononuclear EVTs in the maternal circulation does not seem to contribute to the materno-placental vascular connection directly, but rather to establish (and expand) the materno-fetal interface through an alternative vascular route.


Introduction
During human placental development, at the tips of the anchoring villi, cytotrophoblasts proliferate into cell columns and from there they can detach and invade the maternal decidualized endometrium (decidua) and even the myometrium [1,2]. Those invading fetal cytotrophoblasts are known as extravillous trophoblasts (EVTs).
At term, there are several subtypes of EVTs present in the maternal decidua, depending on their localization: the interstitial mononuclear (as well as multinuclear) EVTs are dispersed in the decidual mesenchyme; the endovascular EVTs and intramural EVTs are both directly associated with remodeled spiral arteries and are present in their lumen (or replacing the endothelial cells) and in their tunica media [3] respectively and those EVTs migrate-colonize the spiral a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 arteries in a retrograde fashion [4]; and a fourth category of 'epithelial' EVTs lines, together with maternal endothelial cells in a mosaic fashion, the basal plate of the maternal decidua basalis [5].
In humans, the defective invasion of the maternal decidua by fetal EVTs leads to insufficient perfusion of the placenta, resulting in pregnancy complications, such as intrauterine growth restriction, (recurrent) spontaneous abortion, (very) premature birth and preeclampsia [6,7]. The unique vascular remodeling of the maternal decidua during pregnancy, whereby both fetal and maternal cells play complementary regulatory roles, is a fundamental process for a successful pregnancy [8,9]. Endovascular EVTs are observed in spiral arteries from (W)8 weeks of gestation, 'plugging' (or blocking) the entrance of spiral arteries to the intervillous space preventing maternal blood flow until about W12 [10][11][12]. During this period (W8-W12), decidual veins become dilated in both decidua parietalis and basalis to give rise to decidual venous lakes [13]. By W12, both decidual veins and spiral arteries are in open connection with the intervillous space of the placenta allowing the maternal blood to circulate between the placental villi [10][11][12]. This materno-placental vascular connection provides efficient exchange of nutrients, waste products, hormones and gases between the maternal blood and the fetal blood and is crucial to ensure fetal growth.
After the materno-placental vascular connection is established, the constant maternal blood flow transports small tissue fragments (syncytial knots) that are shed from the outer layer of the placental villi of multinucleated (non-viable) syncytiotrophoblasts into the decidual venous where those knots become trapped and are subsequently cleared [13][14][15]. If the syncytial knots escape to enter the maternal circulation, they can cause trophoblastic embolism and even sudden death of the mother [16].
Interestingly, the presence of fetal cells [17], placental-derived particles and exosomes [18,19] in the maternal blood from W6 has been described, but in the absence of robust maternoplacental vascular connection this remains to be clarified. Therefore, we sought to systematically characterize the presence of EVTs in the endometrium in the period W5.5-W12. Our results indicate that between W5.5-W12, interstitial mononuclear EVTs efficiently enter the maternal (blood and lymph) circulation. This novel aspect of EVT invasion, enlarging materno-fetal interface early during pregnancy, may shed new light in our understanding of pregnancy complications and maternal immune tolerance.

Ethics approval and human tissue collection
The Medical Ethics Committee of the Leiden University Medical Center (protocol P08.087) approved the collection and use of material for this study. Written informed consent was obtained from all patients (N = 13). Tissue was obtained by vacuum aspiration from women undergoing voluntary pregnancy termination without medical indication and obtained anonymized. The gestational age (in weeks and days) was determined by obstetric ultrasonography and can be converted to weeks post conception by subtracting two weeks. Tissue samples (S1 Table) were collected in cold saline solution (0.9% NaCl). The decidua basalis and parietalis were identified retrospectively by the presence of scattered pKRT-positive EVTs in histological sections.
sections were deparaffinised using xylene and dehydrated by standard procedures. The sections were then treated with 0.1% Azocarmine B (Edward Gurr Ltd., London, UK) in 5% glacial acetic acid (Merck, Darmstadt, Germany) for 3 minutes, rinsed several times in water, treated with 5% phosphotungstic acid (Fluka, Sigma-Aldrich, St. Louis, USA) solution for 6 minutes, again rinsed several times in water and immersed in a 1:1 solution of water to 0.4% orange G (Merck, Darmstadt, Germany)/ 0.2% aniline blue (Nustain, Nottingham, UK) in 1% glacial acetic acid for 15 minutes, all at room temperature (RT). Sections were then rinsed in water, dehydrated through a series of ethanol solutions, xylene and mounted in Entellan (Merck, Darmstadt, Germany).

Immunofluorescence
Paraffin sections were deparaffinised using xylene and dehydrated by standard procedures and used for immunofluorescence as previously described [20]. Briefly, antigen retrieval was performed in citrate buffer, followed by treatment with blocking solution consisting of

Fluorescence in situ hybridization (FISH) for chrX and chrY
FISH with probes of human chrX and chrY was carried out to distinguish fetal (male) cells from maternal (female) cells. After selected immunostained slides were imaged, they were incubated with 0.4% pepsin (pepsin from porcine gastric mucosa, P7000-100G, Sigma-Aldrich, St Louis, MO, USA) in 0.02M HCl for 5 minutes at 37˚C. Centromere specific alphoid repeat probes used were pBamX5 and pDP97 of the human chrX and chrY, respectively [21,22]. Both probes were labeled by nick translation with FITC (chrX probe) or Cy3 (chrY probe). Probe labeling and hybridization reaction were performed using previously published protocols [23,24].

Image acquisition
Slides were either scanned on a Pannoramic 250 FLASH digital scanner (3D HISTECH Ltd., Budapest, Hungary) and representative areas selected for images using the software program 'Pannoramic viewer' (3D HISTECH Ltd., Budapest, Hungary); or photographed on a Leica DMRA fluorescence microscope (Leica, Wetzlar, Germany) equipped with a CoolSnap HQ2 camera (Photometrics, Tucson, USA). Figures were compiled using Photoshop CS6 (Adobe Systems Inc., San Jose, USA).

Quantification of veins being invaded by EVTs
The combination of Azan staining and immunofluorescence for ACTA2, PECAM1 and PDPN on serial sections from the decidua basalis enabled us to distinguish decidual veins from arteries and lymphatic vessels. Decidual veins were PECAM1-positive, ACTA2-negative and PDPN-negative vessels, with reduced amounts of extracellular matrix surrounding them. Veins were considered invaded when there were one or multiple (male) EVTs (KRT-positive, chrY-positive) penetrating the (maternal) vessel wall (chrY-negative, PECAM1-positive, ACTA2-negative and PDPN-negative). To quantify the percentage of decidual veins invaded by EVTs, we manually counted decidual veins (with and without EVTs invasion) in each paraffin section of decidua basalis of different individuals (N = 9) between W5.5-W12. The number of paraffin sections (n) counted is shown. The results are shown as mean ± standard deviation.

Results
Interstitial mononuclear EVTs invade decidual veins from W5.5-W7.5 We investigated the presence of (male) EVTs in the decidua in the period W5.5-W12 and have analyzed a total of n = 19 samples of decidua basalis and n = 20 decidua parietalis from N = 13 women (S1 Table). Due to a paucity of markers to identify specifically EVTs, we used a combined approach of cytogenetics (FISH for sex chromosomes chrX and chrY) and immunofluorescence for pan-keratin (pKRT) to distinguish maternal glandular epithelial cells from the fetal (male) EVTs in the endometrium (S1 Fig). This staining-combination also allowed us to distinguish decidua parietalis from decidua basalis (S1 Fig). Combining cytochemistry (Azan staining) and immunofluorescence for the arterial-marker ACTA2, the endothelial-marker PECAM1 and the lymphatic-marker PDPN, we were able to differentiate between arteries, veins and lymphatic vessels.
By contrast, individual interstitial mononuclear EVTs were always observed outside the arterial smooth-muscle layer of the ACTA2-positive PECAM1-positive decidual spiral arteries, even in those arteries located close to the maternal basal plate and in the vicinity of EVTs (Fig  2).

From W8-W12, mononuclear EVTs invade spiral arteries as well as decidual veins
Between W8-W12 (N = 7), interstitial mononuclear EVTs were still observed penetrating many decidual veins both close to the basal plate and more interiorly located (Fig 3A and 3B  and S2 Fig). We counted decidual veins in paraffin sections (n = 47) from different individuals between W8.1-W12 (N = 5), and observed that on average 16% (165/1038) of the decidual veins were invaded by EVTs (Fig 1E, blue bars), comparable to what was observed between W5.5-W7.5. At this gestational age, many decidual spiral arteries located at the maternal basal plate were indeed being remodeled and contained endovascular and intramural EVTs in the lumen and tunica media, respectively (Fig 3C). However, we noticed that deeper located decidual spiral arteries remained unremodelled, whereas neighboring deeper located veins were consistently invaded by individual interstitial mononuclear EVTs in all decidua basalis analyzed (Fig 3B and 3C and S2 Fig).
From the decidual samples analyzed in this study, cytogenetic analysis revealed that one individual decidua was from a W8.4 conceptus with Klinefelter syndrome (mosaic 47,XXY and 46,XY) (Fig 4A), even though additional chromosomal abnormalities were not investigated. Importantly, interstitial mononuclear XXY EVTs were observed entering decidual veins (Fig 4A), as in the other decidua of similar age, suggesting that this aspect of the establishment of the materno-fetal interface occurred normally.

Interstitial mononuclear EVTs entered decidual lymphatics from W5.5-W12
To exclude that the PECAM1-positive ACTA2-negative vessels were lymphatic vessels instead of veins, we investigated the expression of PDPN in the decidua [25][26][27]. Notably, not only PDPN-negative PECAM1-positive decidual veins were being invaded by interstitial mononuclear EVTs, but also many PDPN-positive PECAM1-positive decidual lymphatic vessels were in fact being invaded by interstitial mononuclear EVTs between W5.5-W12 (N = 11 out of 12 analyzed) (Fig 4B). We observed dilated PDPN-positive lymphatics encircling (or insulating) spiral arteries to create characteristic round structures (data not shown), confirming findings by Volchek and colleagues (2010) [27]. The biological significance of these characteristic structures remains to be investigated. We also report PDPN-positive mesenchymal stroma inside the placental villi (cyan arrow in S2 Fig, top right panel), as previously reported by Wang et al (2011) [28]. At later time points, from W7-W12 (N = 6), bona fide syncytial knots (central area of multiple nuclei and large peripheral area of cytoplasm) were often present in dilated veins in both the decidua basalis (Fig 5A and 5B) and parietalis (Fig 5B), suggestive of the gradual establishment of materno-placental vascular connection. Allo-epi-endothelium in the basal plate of first trimester decidua basalis Finally, we report that the basal plate of the maternal decidua (between the anchoring villi) was lined by both (male) pKRT-positive EVTs and (female) PECAM1-positive maternal endothelial cells in the period analyzed (W5.5-W12) (Fig 6). This allo-epi-endothelium monolayer organization of 'epithelial' EVTs and maternal endothelial cells in a patchy-mosaic fashion has been described in term decidua [5] and we describe it now in the human decidua during the first trimester as well.

Interstitial mononuclear EVTs enter the maternal circulation via veins and lymphatics
Our systematic analysis of first trimester decidua using a combination of immunofluorescence and FISH to distinguish between the types of vasculature and identify the EVTs unambiguously allowed us to reveal for the first time the direct invasion by interstitial mononuclear EVTs of decidual veins and lymphatics in the maternal decidua basalis by W5.5. This suggests an efficient entry of interstitial mononuclear EVTs into the maternal (blood and lymph) circulation in the absence of robust materno-placental vascular connection [10][11][12]. Consistent with our findings, very recently the group of Huppertz has also showed evidence for EVTs invasion of uterine veins [29], using an antibody against human leukocyte antigen (HLA)-G, often used as specific EVT marker [30]. However, caution should be taken when using only HLA-G to identify EVTs, due to the different HLA-G isoforms [31]. In addition, using HLA-G immunostaining as means to identify EVTs, the same group has also reported that EVTs (coined "endoglandular") also invaded uterine glands [29,32,33]. In our study, using a combination of immunofluorescence and FISH, we were unable to detect (male) EVTs in maternal uterine glands between W5.5 and W12.
The timing of the venous invasion process (by the EVTs) is different from the passive deportation of (non-viable) syncytial knots detaching directly from placental villi to the intervillous space and being then transported to the decidual venous lakes once the materno-placental vascular connection has been established [13][14][15]34].

Materno-fetal interface: EVTs stepping out of the uterus
The materno-fetal interface encompasses three main processes: the establishment of immune tolerance, regulation of decidual invasion by EVTs and uterine vascular remodeling [35], all taking place primarily in the uterus. Our results are in agreement with the timing of vascular remodeling as well as with the invasion of the decidua by interstitial and endovascular EVTs. Moreover, we observed the formation of the allo-epi-endothelium in the basal plate of the decidua basalis and therefore acknowledge the existence of epithelial EVTs [5].
Here, we propose that when interstitial mononuclear EVTs enter the maternal circulation they are not only substantially enlarging the materno-fetal interface, but they are also doing so earlier than previously accepted [via placental debris after materno-placental vascular connection [18]] and via an alternatively vascular route (decidual lymphatics and veins).
This novel venue of materno-fetal interface (taking place outside the uterus) may contribute significantly to trigger the maternal immune system to perhaps confer recognition to prevent rejection of the fetus while in the uterus. The EVTs have developed several strategies to ensure immune tolerance while in the maternal environment. This includes the expression of HLA-G, instead of HLA-A and HLA-B, in this way avoiding clearance by maternal natural killer cells [36]. In the uterus, the EVTs encounter and crosstalk with uterine macrophages, uterine natural killer cells and T lymphocytes, modulating locally the maternal innate and adaptive immune response [35,36]. The EVTs that enter the maternal circulation may have a primary role contributing to the adaptive and innate immune response, but through the recruitment of maternal cells to the uterus subsequently also contribute to successful vascular remodeling [4].

Materno-fetal interface and pregnancy complications
Sex chromosomal aneuploidy disorders, including Klinefelter syndrome, often produce smallfor-gestational age fetuses [37]. We describe here the decidua basalis of one W8.4 pregnancy with Klinefelter syndrome (mosaic 47,XXY and 46,XY) and show interstitial mononuclear EVTs entering the maternal circulation via decidual veins. We were unable to detect "plugged" of remodeled arteries colonized by endovascular EVTs, probably due to the young age of the fetus.
Our understanding of placental complications in humans is rather limited due to the lack of in vitro assays and suitable animal models. However it is clear that the materno-fetal interface in all its facets is indispensable for a successful pregnancy [38,39]. We speculate that pregnancy complications such as preeclampsia and intrauterine growth restriction that are closely associated to defects in vascular remodeling [3,7,40], as well as recurrent abortion, may correlate with defective presence of EVTs in the maternal circulation.