In this study preeclampsia was defined according to the guidelines of the American College of Obstetricians and Gynaecologists as already mentioned above (1, 2). The institutional ethics committee of the Medical University of Graz (29-319 ex 16/17) approved the study. Subjects included in the study signed an informed consent form before participation, the characteristics of which are shown in the Table 1. Included placentae from singleton pregnancies were used within 30 minutes of caesarean section or vaginal delivery. PE was defined as a sustained blood pressure of 140/90 mm Hg or greater (on two occasions at least 4 hours apart) occurring after 20 weeks of gestation in a woman with previously normal blood pressure, accompanied by one or more of the following new onset conditions: proteinuria, thrombocytopenia, renal insufficiency, impaired liver function, pulmonary edema, neurological complications or uteroplacental dysfunction. Onset of PE was defined as early (EO, delivered and detected before 34 weeks of gestation) or late (LO, delivered and detected after 34 weeks of gestation) (1). Placentae from normal pregnancies served as controls.

Primary HBCs were isolated according to a modified protocol as described by Tang et al. (48). To avoid contamination with decidual macrophages, the decidual membrane was removed before isolating HBCs. The villous tissue was dissected, washed in 0.4% saline solution (Fresenius, Cat #C924228), and finely minced. Between 60 and 100 g of the minced tissue was stored overnight at 4°C in 1 x phosphate buffered saline (PBS, Medicado, Cat #09-9400-100). On the next day, the tissue was digested with trypsin (0.25%, Sigma Aldrich, Cat #T4549) and DNase I (0.08 mg/ml; Roche, Cat #10104159001), followed by digestion with collagenase A (1 mg/ml; Roche, Cat #10103586001) and DNase I (0.08 mg/ml, Roche, Cat #10104159001). The cell suspension containing the HBCs was applied to a Percoll gradient (20-40%, Sigma Aldrich, Cat #P4937) and centrifuged unrestrained at 1000 g for 30 min. At this point, the HBCs appearing as bands between 30 and 35% gradient layers were aspirated and purified by negative selection using Dynabeads (Invitrogen, Cat #11033) coated with antibodies against epithelial growth factor receptor (EGFR, Santa Cruz, Cat #sc-120) and CD10 (Sigma Aldrich, Cat #SAB4700440). After immunopurification, cells were seeded in macrophage medium (MaM, ScienCell, Cat #SC1921) containing 5% FBS (ScienCell, Cat #SC1921), PenStrep (ScienCell, Cat #SC1921) and macrophage growth supplements (ScienCell, Cat #SC1921) at a cell density of 1x10⁶ cells/ml. Cells were cultured at 21% oxygen and 37°C. Quality control of the isolated HBCs was performed on the fixed cells after 6 days by immunocytochemistry for CD163 (Thermo Fischer Scientific, Cat #MA1-82342), CD90 (Dianova, Cat # DIA100), CD80 (Abcam, Cat #ab86473), CD68 (Dako, Cat #GA613), CD86 (Abcam, Cat #ab270719), CD206 (Novus, Cat #H00004360), CD209 (R&D Systems, Cat #MAB1621), and isotype control (Dako, Cat #X0931) as previously described by Schliefsteiner et al. (31). Cell culture images were obtained using brightfield microscope with a SC50 Olympus camera and CellSens software.

Tissue sections were taken from four different areas of placenta (reaching from chorionic plate to the decidual side and a central region of the placental disk) and fixed overnight in 4% neutral buffered paraformaldehyde solution. After paraffin embedding, tissue sections with a thickness of 5μm were mounted on glass slides. The paraffin was then removed with xylene and rehydrated in an ethanol dilution series. Antigen retrieval was performed using a citrate buffer (Gatt, Cat #403139070) adjusted to pH 6.0. UltraVision LP detection system (Thermo Fischer Scientific, Cat #TL125HL) was used for histochemical immunostaining. Tissue was incubated with Hydrogen Peroxide Block (Thermo Fischer Scientific, Cat #TL125HL) for 15 minutes and washed in TBE buffer (Gatt, Cat #403211370), followed by a 5-minute incubation with Ultra V protein block (Thermo Fischer Scientific, Cat #TL125HL). The primary antibody isotype control (1:200, Dako, Cat #X0931) and CD163 (1:200, Thermo Fischer Scientific, Cat #MA1-82342) were diluted in antibody diluent (Agilent, Dako, Cat #S0809) and incubated overnight at 4°C in a humidified chamber. After washing step, primary antibody enhancer (Thermo Fischer Scientific, Cat #TL125HL) was applied for 20 minutes. After another washing step samples were incubated with Large HRP Polymer (Thermo Fischer Scientific, Cat #TL125HL) solution for 30 minutes, followed by intensive washing and incubation with AEC Chromogen Solution (Abcam, Cat #64252) for 10 minutes. The tissue was counterstained with Haematoxylin (Gatt-Koller Cat #401296170) for 1 minute and mounted with embedding medium. Images were acquired using CellSens Standard software and an Olympus BX53 light microscope with an Olympus UC90 camera. Per slide, images of 5-10 different areas were taken and quantified with Qupath software (49).

FACS was performed to quantify the cell populations expressing M1 and M2 polarization markers of HBCs. On the fifth day after isolation, cells were harvested using accutase (Thermo Fischer Scientific, Cat #00-4555-56) and gentle scraping. Viability and number of cells after scraping was determined using a CASY cell counter model TT (Innovatis, Bielefeld). At least 1x10⁵ viable cells per tube were used for the experiment. Cells were resuspended in 3% FCS - HBSS solution for 10 min at room temperature to block Fc-receptors and reduce non-specific binding. For surface staining, cells were incubated with a fluorochrome-conjugated antibody in the amount indicated in the Supplementary Table 1 for 20 minutes at 4°C in the dark. Cells were washed with staining buffer [PBS containing 0.1% BSA (Sigma Aldrich, Cat #A2153) and 2mM EDTA (Thermo Fischer Scientific, Cat #15575020)], centrifuged at 300 g for 5 minutes and resuspended in 200 µL staining buffer. For detection of surface molecules, a minimum of 10000 live events per sample were counted. In order to identify expression of surface markers, cells were separated by size using forward and size scatter (FSC and SSC, respectively), followed by doublet discrimination. In the next step, cells were discriminated into live and dead cells using the 7-AAD dye (BD Biosciences, Cat #559925) by plotting it against the SSC area. In the fourth step cells were plotted for the respective marker against the SSC area. For staining of intracellular molecules, cells were fixed and permeabilized with BD Cytofix/Cytoperm kit (BD Biosciences, Cat #554714). Staining was performed according to the manufacturer’s instructions. A minimum of 10000 events per sample were counted. To investigate the expression of intracellular polarization markers cells were separated by size using forward and size scatter (FSC and SSC, respectively), followed by doublet discrimination and gating against SSC-area and respective marker. The same gating strategy was employed on CTR and PE macrophages. Surface molecules were compensated by individual staining on OneComp eBeadsTM Compensation Beads (Thermo Fischer Scientific, Cat #01-1111-42). Isotype controls corresponding to each fluorochrome in the experiment were used to detect non-specific positive signals. Antibodies used for FACS analysis and their corresponding dilutions are listed in the Supplementary Table 1. Cell sorting was performed using a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, USA) and analysed with FlowJoTM v10.8 software for gate setting and data analysis.

Inflammation 20-Plex Human ProcartaPlex™ Panel (Invitrogen, Thermo Fischer Scientific, Cat #EPX200-12185-901) was used to quantify the secretion of pro- and anti-inflammatory molecules. Human TIMP Magnetic Luminex Performance Assay 4- Plex Kit (R&D Systems, Cat #LKTM003) was used to analyse TIMP secretion. HBCs were cultured in MaM for 5-6 days before supernatants were collected and centrifuged at 4000 rpm, 4°C for 15 minutes. MaM medium processed under the same conditions as the samples served as a blank. Multiplex assays were performed according to the manufacturer’s instructions. Cytokines reaching the detection limits were normalized to total protein content in the supernatant using the Pierce BCA kit (Thermo Fischer Scientific, Cat #23225).

TGF-beta 1 Quantikine ELISA kit for human/mouse/rat/porcine/rabbit (R&D Systems, Cat #DB100B) was used for the detection of TGF-β1. Next, to quantify secretion of IL-8, human IL -8/CXCL8 Quantikine ELISA kit (R&D Systems, Cat #D8000C) was used. Both ELISA kits were performed according to the manufacturers’ instructions. To quantify the amount of secreted TGF-β1 and IL-8, cells were cultured in MaM for 5-6 days before collection of the supernatants. MaM medium processed under the same conditions as the samples, but without cells served as a blank. Cytokine levels were normalized to the total protein content in the supernatant measured with the Pierce BCA kit (Thermo Fischer Scientific, Cat #23225), in order to account for deviating volume concentrations.

HBCs were washed twice with ice cold Hanks’ salt balanced solution (HBSS, Thermo Fischer Scientific, Cat #14175-053) and harvested in 700 µL QIazol Lysis Reagent (Quiagen, Cat #79306). Total RNA content was isolated using miRNeasy Mini Kit (Quiagen, Cat #217004). Reverse transcription was performed using 1 µg of RNA and Luna Script RT SuperMix Kit (New England BioLabs, Cat #M3010). For qPCR analysis, SYBR Green Luna Universal qPCR Master Mix (New England BioLabs, Cat #M3003) and CFX-384 Touch Real time PCR detection system (Bio-Rad) were used. Expression of target genes was normalized to the following housekeeping genes (18S, RPL30 and HPRT) using 2^(^ΔΔCt) method. Primer sequences used for qPCR analysis are listed in the Supplementary Table 2.

Phagocytosis Assay Kit (Abcam, Cat #ab234053) was used according to the manufacturer’s instructions. For detection of phagocytic activity 1 x 10⁶ cells/ml were used. On the fifth day post isolation HBCs were treated with zymosan slurry and incubated for 3 hours at 21% oxygen and 37°C. After washing steps, cells were detached using accutase (Thermo Fischer Scientific, Cat #00-4555-56) and careful scraping, followed by a washing step with staining buffer containing PBS with 0.1% BSA (Sigma Aldrich, Cat #A2153) and 2mM EDTA (Thermo Fischer Scientific, Cat #15575020). Afterwards measurements were performed in the FITC channel. Untreated cells served as controls. Cells were separated by size using FSC-A and SSC-A, followed by doublet discrimination gating the area and height of FSC. Lastly, the FITC fluorescent signal was determined using histograms. Cell sorting was performed on a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, USA) using FlowJoTM v10.8software for gate setting and data analysis.

Phagocytosis Assay Kit (Abcam, Cat #ab234053) was used according to the manufacturer’s instructions. HBCs were seeded at the density of 0.5 x 10⁶ cells/ml in 24-well black/clear bottom plates. 5 µL of zymosan slurry was added to the cells and incubated at 21% oxygen and 37°C for 3 hours, followed by a washing step. Next, cells were fixed in a plate containing 4% neutral buffered paraformaldehyde solution, followed by an intensive wash step with TBE buffer containing 1x TBE and 0.1% Tween (Thermo Fischer Scientific, Cat #003005). HBCs were incubated with Protein Block (Thermo Fischer Scientific, Cat #TL125HL) for 20 minutes. Cells were then counterstained with CD163 (1:100, Thermo Fischer Scientific, Cat #MA1-82342), diluted to working concentration in Antibody diluent (Agilent, Dako, Cat #S0809) and incubated overnight at 4°C. After serial washing steps, the plate was incubated with the secondary antibody Dylight633 (goat versus mouse 1:200, Thermo Fischer Scientific, Cat #35512) for 2 h at room temperature. To stain the nuclei, the plate was counterstained with DAPI (1:1000, Sigma Aldrich, Cat #D9542) diluted in antibody diluent for 10 min. After intensive washing, 300 µL of PBS was added to each well and stored at 4°C. Image acquisition was performed using a Nikon microscope with the Zyla sCMON camera. All statistical analysis was carried out on 25 different locations per well using 20x magnification. For better visualization of phagocytosis shown images in the Figures 4C–E were taken with 40x magnification. The number of FITC labelled beads was counted within the cells positive for CD163 and DAPI staining using Nis Elements viewer version 5.20.01 software. Cell surface area was measured using the measuring tool provided within the software.

HBCs supernatants were collected on the fifth day post isolation. The supernatants were centrifuged at 4000 rpm for 15 minutes at 4°C. Total protein concentration was determined using Pierce BCA kit (Thermo Fischer Scientific, Cat #23225) according to the manufacturer’s guidelines. A total of 15 µg of protein sample was diluted with Tris-Glycine SDS sample buffer (Thermo Fisher Scientific, Cat # LC2676) and loaded onto 10% Tris-Glycine gels containing 0.1% gelatin (Thermo Fisher Scientific, Cat #ZY00105BOX) and separated for 135 min at 125 V, 35 mA. After electrophoresis, the gels were incubated in 1x Zymogram Renaturating buffer (Thermo Fisher Scientific, Cat #LC2670) at room temperature with gentle agitation. Followed by 30 minutes incubation with 1x Zymogram developing buffer (Thermo Fisher Scientific, Cat #LC2671). Fresh developing buffer was added and the gels were stored overnight at 37°C. Next day, gels were stained with Coomassie Brilliant Blue (Sigma Aldrich, Cat #1.15444) for 50 minutes and decolorized in 50% distilled water (Fresenius, Cat #C920928): 40% methanol (Sigma Aldrich, Cat #322415): 10% acetic acid (Roth, Cat #3738.5) solution for 10 minutes. Protease activity appearing as a clear band on the dark background was visualized with the ChemiDoc™Touch Imaging System (Bio-Rad). Band densitometry was determined using Image Lab Software Version 6.1 (Bio-Rad).

SPSS (IBM SPSS Statistics version 26) was used for statistical calculations. Next, graphs were generated using Graph Pad Prism 9.3.1 software (GraphPad Software Inc.). To test normal distribution Shapiro-Wilk test was used. Skewed data were transformed using natural logarithm (ln) before applied to statistical analysis and re-transformed for the graphical presentation. To assess statistical significance of the patient characteristics (Table 1) one-way ANOVA with Tukey’s post-hoc test was used. If normality testing failed, Kruskal-Wallis test with Dunn’s post-hoc test was performed. Next, to compare differences between three groups (CTR; EO-PE and LO-PE) ANCOVA with adjustment for gestational age and Sidak’s post-hoc test was used. Equal variances of variables were verified by Levene´s test. When comparing the effect of HBCs conditioned medium on fpECs without adjustment for gestational age (Supplementary Figure 5) two-way ANOVA with Sidak’s post-hoc test was used. All values are given as mean ± S.E.M. p-values ≤ 0.05 were considered statistically significant.

The characteristics of the study population are shown in Table 1. Women who developed early (EO, n=8) and late onset (LO, n=6) PE were included in this pilot study. Advanced maternal age, high pre-pregnancy BMI, nulliparity, gestational diabetes, chronic hypertension are some of the risk factors for the development of PE (1). Women in PE groups were significantly older as those in CTR group and their pre-pregnancy BMI ranged from 19.6 to 25.4 kg/m2. As expected, systolic and diastolic blood pressure levels differed significantly between the CTR and PE groups. Gestational age of EO-PE group was significantly lower than of CTR and LO-PE group. Consequently, early gestational age of the EO-PE group is directly related to placental- and fetal weight, both of which were significantly lower than in CTRs. Since development of the placenta depends on gestational age (50), we adjusted the (normally distributed) data for that respective factor. We found a significant difference in the levels of sFlt between EO- and LO-PE group, while there were no significant differences between other clinical parameters (PlGF, platelets, uric acid, AST and ALT).

To study polarization, we first examined the distribution and number of HBCs in placental tissue using immunohistochemistry approach. Since HBCs have been shown to be strongly positive for CD163, we stained 5 µm serial sections of CTR (n=5), EO-PE (n=6) and LO-PE (n=6) placental tissue, mouse IgG served as a negative control. CD163 is used as a marker used for placenta resident macrophages, and if combined with other markers (e.g. Folate receptor-β, CD206, CD209) is often associated with M2 polarization (26, 30, 31, 51). HBCs positive for CD163 were found in the villous stroma, moreover in stem, intermediate and terminal villi of CTR, EO- and LO- PE placentae (Figure 1A). Furthermore, quantification of CD163-positive cells revealed a significantly decreased number of stained (Figure 1B) cells per mm² in both PE groups, indicating reduced number of HBCs in respective groups. Next, we adjusted the number of isolated HBCs to the wet weight (grams) of placental tissue used for isolation. Consistent with immunohistochemical analysis, we found significantly reduced number of primary LO-PE HBCs. Reduction of HBCs was also observed in the EO-PE group, however, did not reach significance compared to the CTRs (Figure 1C).

In vitro, cell morphology of HBCs isolated from CTR, EO- and LO-PE placentae showed substantial differences (Supplementary Figure S1). Normally, directly after isolation, HBCs are round shaped cells with many vacuoles in the cytosol. Within 48-72 hours after isolation, cells differentiate and develop different shapes. Usually, M2 characterized macrophages exhibit an elongated, spindle-shaped morphology, whereas M1 polarized cells form a round, dendritic cell-like morphology with large filopodia (52, 53). CTR HBCs developed typical M2 features (Supplementary Figure 1A), whereas within isolations of EO- and LO-PE HBCs more of round shaped cells with larger filopodia next to M2 morphologies were found (Supplementary Figures 1B, C).

Basal expression of surface and intracellular M1 and M2 markers was determined by FACS (Tables 2 and 3) on primary isolated CTR (n=12), EO-PE (n=6) and LO-PE (n=5) HBCs. The gating strategy is shown in Supplementary Figure 2. FACS analysis revealed distinct expression of M1 and M2 markers in EO- and LO- PE HBCs (Table 2). EO-PE HBCs tend to express higher levels of pro-inflammatory CD11b, CD11c, CD40 (p=0.09) and TLR4 (p ≤ 0.05), whereas expression of listed markers was similarly distributed between LO-PE and CTR HBCs, except for CD209 (p ≤ 0.001). Furthermore, we found significant increase of the major histocompatibility class (MHC) II molecule HLA-DR within the EO-PE group compared to the CTR group (p≤ 0.001). Interestingly, the HLA-DR expression was significantly different between EO- and LO-PE HBCs (p≤ 0.01) as well. Among the M1 markers only expression of CD80 (p=0.05) and TLR1 (p ≤ 0.05) were elevated in LO-PE group. Notably, expression of CD80 and TLR1 was elevated in EO-PE group as well, but only by trend. Interestingly, surface expression of TLR2 was reduced in both EO- (p ≤ 0.05) and LO-PE group. Next, we investigated the expression levels of CD86. Since CD86 can serve as M1 or M2b marker (54, 55), the secretion profile and expression of other markers should be taken into account when interpreting its expression. We found significant induction of the expression of the respective marker in EO-PE HBCs (p ≤ 0.05) compared to the CTR (p ≤ 0.05) or LO-PE (p ≤ 0.05) group. In LO-PE HBCs the expression of CD86 did not differ from the control. In contrast to the increase of pro-inflammatory markers in EO-PE HBCs, a slight decrease in the anti-inflammatory markers CD206 and folate receptor β (FR-β) was observed. Expression pattern of CD206 and FR-β was similar in LO-PE HBCs as well. Expression of CD163 on primary isolated HBCs was evenly distributed between all investigated groups, confirming that CD163 can be used rather as a reliable tissue resident marker (Figures 1A, B) than a direct indicator of M2 phenotype. Next, M2 marker CD209 was suppressed in the LO-PE group (p ≤ 0.001). Its expression was only minimally decreased by 10% in EO-PE HBCs group. Interestingly, we found a significant difference in the expression of CD209 between EO- and LO-PE HBCs (p ≤ 0.01).

In addition to the investigation of surface polarization markers, we studied intracellular markers (Table 3). Next to the pan-macrophage marker CD68, two important regulators of TLR-Myd88 signaling (56, 57), IRF4 and IRF5 were investigated. Noteworthy, IRF4/IRF5 axis is involved in the initiation control of a specific M1/M2 polarization program. IRF5 as a positive regulator of Myd88 induces the expression of pro-inflammatory genes and establishment of M1 phenotype (58). Whereas IRF4, as a negative regulator of Myd88, leads to the activation of anti-inflammatory genes and initiation of M2 polarization (56, 59).We found the highest expression of IRF5 in EO-PE group and the lowest in CTRs. Expression of IRF4 the regulator of M2 polarization was evenly distributed between CTR and LO-PE group. Importantly, expression of IRF4 in EO-PE HBCs was reduced. Moreover, balance in favour of IRF5 together with higher expression of other surface M1 markers (Table 2) indicates possible phenotypic switch towards M1 polarization of EO-PE HBCs.

Polarized macrophages are known to secrete specific patterns of cytokines, chemokines, and growth factors, allowing us to characterize polarization states (60). Using multiplex ELISA-on-bead technology, we determined the secretion profile of cytokines and chemokines secreted by CTR (n=8), EO-PE (n=6) and LO-PE (n=4) HBCs (Figure 2, Supplementary Figure 3). Moreover, secretion of IL-8 was determined using ELISA, since its secretion excided the detection limit of the multiplex ELISA (CTR n=10, EO-PE n=5, LO-PE n= 5; Figure 2J). Notably, TGF-β1 was measured with ELISA (CTR n=12, EO-PE n=6, LO-PE n=6, Figure 2K) since the sample preparation requires acidification of the samples for binding of TGF-β epitopes.

Secretion profile of EO- and LO-PE HBCs differs from the CTR group. First, among pro-inflammatory cytokines, we discovered a trend of an increased secretion of IL-6, IL-12p70 and P-Selectin by EO-PE HBCs (Figures 2A, B; Supplementary Figure 3A). Contrary, to the EO-PE HBCs where secretion of TNF-α was unchanged, LO-PE released higher amounts of TNF-α as CTRs or EO-PE HBCs (p ≤ 0.05) (Figure 2C). LO-PE HBCs secreted higher amounts of ICAM, but only by trend (Figure 2D). Moreover, both PE groups secreted less IL-17a compared to CTRs (Figure 2E). Although, release of E-Selectin was higher in both PE groups, we did not find any differences in the production of other pro-inflammatory cytokines, such as: IL-1α, IL-1 β, CCL-3, CCL-4, IFN- α, and IFN-γ between investigated groups (Figure 2F, Supplementary Figures 3B–H). A decrease of IL-8, measured by ELISA was detected in LO-PE HBC group, whereas unchanged between CTR HBCs and EO-PE group. Interestingly, we found a significant difference in the release of IL-8 between EO- and LO-PE HBCs (p ≤ 0.05) (Figure 2J).

Next, we examined the secretion of anti-inflammatory cytokines namely, IL-4, IL-10, IL-13, and TGF-β1, which serve as important drivers of M2 polarization (53, 61). We did not find any differences in the secretion of IL-13; however, secretion of IL-4 and IL-10 was increased, but only by trend in EO-PE HBCs. LO-PE HBCs secreted significantly higher levels of IL-4 (p ≤ 0.05) and IL-13 (p ≤ 0.05) compared to CTR HBCs (Figures 2G–I). Interestingly, both PE groups released significantly higher amounts of anti-inflammatory TGF-β1, which was even more pronounced in the LO-PE group (p ≤ 0.01) (Figure 2K).

Noteworthy, some of the secreted pro-inflammatory cytokines are not reliable identifiers of a specific phenotype, since they are expressed by both M1 and M2 macrophages. E.g. IL-6 is a pro-inflammatory cytokine produced by both M1- and M2a-polarized macrophages (62). In addition, secretion of ICAM is mediated by NF-κB, but can serve as both an M1 and M2 cytokine due to its pro-angiogenic nature (63). The observed changes in the secretion profile indicate a switch in the phenotype and possible protective mechanisms of PE HBCs in an attempt to reduce the extent of inflammation by increasing the production of anti-inflammatory cytokines.

Macrophages are capable of responding to the local stimuli and acquiring different phenotypes and functions to meet changing physiological needs (64). Next, we examined transcriptional changes in HBCs that may be triggered by PE. Basal gene expression of CTR (n=12), EO- (n=5) and LO-PE (n=5) HBCs was determined on the fifth day after isolation. We analysed selected genes associated with phenotype and functionality of macrophages undergoing inflammation (Figure 3). Dynamic changes in gene expression were observed between the PE subgroups EO and LO. First, we found -as expected - an upregulation of NFKB1 in both EO- and LO-PE groups (p ≤ 0.05) (Figure 3A). Among the inflammatory pathways involved in M1 polarization, NF-κB plays an important role and regulates the expression of pro-inflammatory genes such as cytokines, adhesion molecules and growth factors (65). Next, the expression of HIF1, another M1-associated gene was upregulated in the LO-PE group (p ≤ 0.05), whereas its expression was surprisingly decreased in EO-PE group (p ≤ 0.05) (Supplementary Figure 4). In line with the increased secretion of ICAM (Figure 2H), mRNA of the respective gene was upregulated in the both PE groups (Supplementary Figure 4). In contrast to the expression of ICAM, VCAM was downregulated in EO-PE, whereas in LO-PE group remained on the level of CTRs (Supplementary Figure 4). Consistent with secretion of IL-8 (Figure 2J), higher fold change of IL8 was detected in CTR group (Supplementary Figure 4). The expression of TGFB1, which acts as important M2 inducer (66), was significantly elevated in both, EO- and LO-PE HBCs (p ≤ 0.05) (Figure 3B). Although, the differences in the secretion of CCL-4 (Supplementary Figure 3H) were not noticeable, higher expression was detected in the CTR group by RT-qPCR. In contrast to the secretion profile of IL -6 and IL -10, which was higher in the PE group (Figures 2A, H), qPCR analysis revealed downregulation of respective cytokines in the EO- and LO-PE groups (Supplementary Figure 4), possibly due to the tight post-transcriptional gene regulation of these cytokines in particular (67). As polarized macrophages metabolise L-arginine differently, M2 via arginase-1 and M1 macrophages via nitric oxide synthase (iNOS) (68), we looked at expression of ARG1 gene, encoding arginase-1, and NOS2, encoding iNOS (68). Interestingly, ARG1 expression was increased in both PE groups (Figure 3C). However, expression of NOS2 was strongly upregulated in EO-PE (p ≤ 0.05); whereas expression of the respective gene in LO-PE HBCs was even lower as in CTR HBCs (Figure 3D). In addition to the metabolism of L-arginine, regulation of reactive oxygen species (ROS), represents an important link between M1 and M2 polarization (69). M1 macrophages produce higher amounts of ROS and consequently downregulate antioxidant enzymes such as CAT encoding catalase, or SOD encoding superoxide dismutase (70), whereas, M2 macrophages are thought to produce lower levels of ROS and express higher levels of CAT or SOD (71). The expression of the genes CAT and SOD, was significantly attenuated in both, EO- and LO-PE groups (Figure 3E, F). Next in respect to observed TGF-β1 differences, we looked at the expression of genes involved in tissue remodeling and adhesion. Importantly, we found upregulation of MMP9 in both PE groups (Figure 3G). The expression of other genes involved in tissue remodeling (MMP2, MMP12, TIMP1, TIMP2), did not differ between groups (Supplementary Figure 4). Adhesion molecules, such as CDH2 has been downregulated in both EO-PE (p ≤ 0.05) and LO-PE HBCs (Figure 3H). Similarly, as CDH2 we identified reduced expression of CDH5 in both, EO- and LO- PE HBCs (Figure 3I).

In the healthy placenta, HBCs are often found in close proximity to feto-placental endothelial cells (fpEC), and M2 macrophages have the ability to regulate placental angiogenesis by secretion of pro-angiogenic factors (26). To gain insight into their role in angiogenesis, HBCs were examined for the expression of FLT, VEGFA, KDR, and EGFR; which were all downregulated in EO- and LO-PE group (Supplementary Figure 4).

Furthermore, EO-PE and CTR fpECs were treated with conditioned medium (CM) collected from CTR and EO-PE HBCs. In order to further investigate the influence of HBCs on the fpEC, metabolic activity of fpECs using the MTS assay (Supplementary Figure 5A) and the proliferation of the fpEC by incorporation of BrDU (Supplementary Figure 5B) were measured. CM of CTR HBCs increased the NAD(P)H dehydrogenase activity of CTR fpEC, whereas PE CM had no effect on the activity of PE fpEC (Supplementary Figure 5A). A similar effect was observed when the proliferation of CTR fpEC was measured (Supplementary Figure 5B).

Macrophages, as professional phagocytes eliminate pathogens and apoptotic cells. The elimination of apoptotic cells plays an important regulatory role regarding the reduction of the inflammatory burden (72). Phagocytosis was measured and visualised using two different approaches. First, it was assessed by FACS (CTR n=10, EO-PE n=5, LO-PE n=5), where median fluorescence intensity (MFI) was used to quantify the phagocytic activity. PE HBC showed significantly higher phagocytic activity (p ≤ 0.01) than CTRs (Figures 4A, B). Second, we analysed phagocytic activity using HCS (Figures 4C–E). For better visualisation cells were stained with HBCs tissue resident marker CD163. Analysis of CTR (n=7) and PE (n=5, EO n=3, LO n=2) confirmed higher (though not significant) phagocytosis of PE HBCs (Figure 4F). Furthermore, visualisation of phagocytosis allowed us to analyse morphology of the cells, calculating cell size - surface area (µM²), which was lower in EO- and LO-PE groups, when compared to the surface area (µM²) of CTRs (Supplementary Figure 6) confirming in vitro observations of smaller round cell morphologies of PE HBC (Supplementary Figures 1A–C).

In a normal placenta M2-polarized HBCs contribute to tissue remodeling and repair (73). To confirm strong upregulation of MMP9 (Figure 3G), we additionally performed gelatin zymography to assess the activity of MMP-2 and MMP-9 (Figure 5A). As shown with MMP2 mRNA expression (Supplementary Figure 4), detectable MMP-2 activity (Figure 5A) did not differ between studied groups (Figure 5B). In line with upregulation of MMP9, EO-PE HBCs displayed significantly (p ≤ 0.05) higher MMP-9 activity as CTR. LO-PE HBCs MMP-9 activity was increased, but only by trend (Figure 5C).

Expression and production of MMPs are usually tightly regulated within the complex network of their four different tissue inhibitors of metalloproteinases 1-4 (TIMP). HBCs secretion of TIMP- (1, 2, 74, 75) was assessed using a multiplex ELISA-on-bead assay (Figure 5D). Production of TIMP-1, TIMP-2, TIMP-3 was unchanged in EO-PE group, we noticed a decreased production of TIMP-4 in the respective group. Furthermore, LO-PE HBCs secretion of TIMP-1 was significantly decreased, followed by trend in the reduced production of TIMP-2. Interestingly, release of TIMP-3 in LO-PE group was similar as in CTR. LO-PE HBCs production of TIMP-4 was elevated, but only by trend. To further explore the gelatinolytic activity of HBCs, the ratio between mRNA expression of MMP2/TIMP1, MMP2/TIMP2, MMP9/TIMP1, and MMP9/TIMP2 was calculated. Normally, TIMPs regulate inhibition of MMPs by binding in 1:1 reversible complex with the MMPs (76, 77). The shift in MMP/TIMP balance in favor of MMPs reflects as increased extracellular matrix (ECM) proteolysis, or if in favor of TIMP as decreased proteolysis and protection of the ECM (78–80). In CTR, EO- and LO- PE HBCs the ratio between MMP2 and TIMP1 or TIMP2 (Figure 5E) stayed unchanged. Similar MMP9/TIMP1 ratio of investigated groups, was favoring the gelatinolytic activity, indicating that MMP-9 is the main gelatinase produced by HBCs. However, significant differences were demonstrated in MMP-9/TIMP-2 of EO- and LO-PE HBCs, confirming higher gelatinolytic activity shown with gelatin zymography (Figures 5A, C, F).