Circulating and placental FKBPL and Gal-3 are increased in preeclampsia. To determine if FKBPL and Gal-3 expression was altered in preeclampsia compared to normotensive pregnancies, placental and plasma samples were interrogated by Western blotting and ELISA, respectively. With respect to variations between case (preeclampsia) and control groups, no differences in BMI were observed, however, gestational age was significantly lower and maternal age significantly higher in the preeclampsia group (Supplementary Table 1). Placental FKBPL protein expression was over two-fold higher from women with preeclampsia compared to normotensive controls (control 1.00 ± 0.22 vs preeclampsia 2.28 ± 1.99, fold change, p = 0.02; Fig. 2A and Supplementary Fig. 1). Although no correlation was observed between placental FKBPL protein expression and preeclampsia (r = 0.370, p = 0.144), this became statistically significant after adjusting for confounders including gestational and maternal age (r = 0.519, p = 0.047, Table 1). Similarly, placental Gal-3 protein expression was increased in the preeclampsia group (control 167.4 ± 56.7 vs preeclampsia 498.2 ± 531.5, pg/ml, p = 0.004; Fig. 1B), however, no correlation between placental Gal-3 expression and preeclampsia was observed (r = 0.361, p = 0.155), even when adjusted for gestational and maternal age (r = 0.356, p = 0.193; Table 1).
Whilst there were no differences in maternal age between the normotensive and preeclampsia plasma samples, gestational age was significantly lower and body mass index (BMI) significantly higher in the preeclampsia group (supplementary Table 2). Plasma samples analysed by ELISA demonstrated a significant increase in FKBPL concentration in women with preeclampsia compared to normotensive pregnancies (control 0.88 ± 0.35 vs preeclampsia 1.41 ± 0.42, ng/ml, p < 0.0001; Fig. 1C and supplementary Fig. 1). Further, there was a significant positive correlation between plasma FKBPL and preeclampsia (r = 0.578, p < 0.001), even when adjusted for differences in gestational age and BMI (r = 0.559, p < 0.001; Table 2). Similarly, plasma Gal-3 concentration from women with preeclampsia compared to controls was also increased (control 222.2 ± 72.91 vs preeclampsia 288.8 ± 71.98, pg/ml, p = 0.004; Fig. 1E). Aligned to this, plasma Gal-3 concentration was positively correlated with preeclampsia (r = 0.389, p = 0.007), although the statistical significance was lost when adjusted for differences in gestational age and BMI as confounding factors (r = 0.281, p = 0.064; Table 2).
Inflammation regulates FKBPL and Gal-3 expression in trophoblast and endothelial cell 2D monocultures. To investigate the regulation of FKBPL and Gal-3 under inflammatory conditions, trophoblasts (ACH-3Ps) or human umbilical vein endothelial cells (HUVECs) were treated with TNF-α (10ng/ml), an inflammatory stimulus elevated in preeclampsia,[14] for 24 to 72 hours. FKBPL protein expression was significantly increased ~ 1.5-fold, in ACH-3Ps at both time points (control 1.00 ± 0.074 vs TNF-α-24h 1.65 ± 0.16 vs TNF-α-72h 1.58 ± 0.13, fold change, p = 0.021; Fig. 3A, B and supplementary Fig. 2). TNF-α exposure also stimulated Gal-3 protein expression by ~ 2.5-fold in ACH-3Ps exposed to 24 hour TNF-α however this increase was non-significant by 72 hours (control 0.751 ± 0.23 vs TNF-α-24h 2.40 ± 0.33 vs TNF-α-72h 2.00 ± 0.454, fold change, p = 0.036; Fig. 3A, C and supplementary Fig. 2).
Similarly, FKBPL protein expression was increased in HUVECs ~ 2-fold following 24 hours of TNF-α treatment and was maintained at 72 hours (control 0.84 ± 0.12 vs TNF-α-24h 1.84 ± 0.28 vs TNF-α-72h 1.76 ± 0.12, fold change, p = 0.03; Fig. 3D, E and supplementary Fig. 3). When we examined Gal-3 protein expression in HUVECs, there was a trend towards an increase in Gal-3 following 24 hour exposure to TNF-α, though this did not become significant until it increased by ~ 2-fold at 72 hours (control 0.65 ± 0.01 vs TNF-α-24h 1.10 ± 0.14 vs TNF-α-72h 1.54 ± 0.12, fold change, p = 0.006; Fig. 3D, F and supplementary Fig. 3).
Trophoblast migration is stimulated by the presence of endothelial cells or inflammatory conditions. We next used our placenta-on-a-chip model to assess the invasion and migratory ability, as well as regulation of FKBPL and Gal-3, under inflammatory conditions (typical of preeclampsia), of trophoblasts in the absence and presence of endothelial cells. ACH-3Ps were added to one of the media channels whereas the extracellular matrix (ECM) solution with and without HUVECs was placed in the middle channel (Fig. 4A). Cells were treated with 10ng/ml TNF-α for 24 or 72 hours, with untreated cells as a treatment control (Fig. 4B). ACH-3Ps were limited in their invasion through the collagen matrix without HUVECs present and in the absence of TNF-α (Fig. 4A-1 and B). TNF-α exposure stimulated a significant increase in ACH-3Ps migration at both time points (control 130.3 ± 11.23 vs TNF-α-24h 252 ± 41.67 vs TNF-α-72h 269.3 ± 43.6, p = 0.0039; Fig. 4A-1 and C).
When HUVECs were added to the central matrix channel, ACH-3Ps actively traversed the collagen matrix across the chip, and there were no differences in ACH-3Ps migration exposed to inflammatory conditions (Fig. 4A-2, B and C). Interestingly, there was a significant increase in the number of migrating trophoblasts in the presence of endothelial cells, as determined by cytokeratin 7 staining (without HUVECs vs with HUVECs; Control 130.3 ± 11.23 vs 799 ± 8.71, TNF-α-24h 252 ± 41.67 vs 761 ± 18.66, TNF-α-72h 269.3 ± 43.6 vs 838 ± 7.93, p < 0.0001; Fig. 4C).
In the 3D monoculture microfluidics ACH-3P system, FKBPL protein expression was significantly reduced with TNF-α treatment at 24 hours before it was restored and increased by 72 hours (control 1.00 ± 0.04 vs TNF-α-24h 0.35 ± 0.05 vs TNF-α-72h 1.25 ± 0.03, fold change, p < 0.0001; Fig. 4A-1, D and supplementary Fig. 4A). However, in the co-culture system, the presence of endothelial cells increased the FKBPL expression of trophoblasts following 24 hours of TNF-α treatment which was reduced by 72 hours (control 0.91 ± 0.1 vs TNF-α-24h 1.63 ± 0.09 vs TNF-α-72h 0.63 ± 0.02, fold change, p < 0.0001; Fig. 4A-2, E and supplementary Fig. 5A). A similar effect was observed for Gal-3 expression in the absence (control 1.00 ± 0.05 vs TNF-α-24h 0.41 ± 0.05 vs TNF-α-72h 1.34 ± 0.06, fold change, p < 0.0001; Fig. 4A-1,F and supplementary Fig. 4B) and presence of endothelial cells (control 1.00 ± 0.06 vs TNF-α-24h 2.10 ± 0.15 vs TNF-α-72h 0.98 ± 0.09, fold change, p p < 0.0001, Fig. 4A-2, G and supplementary Fig. 5B). However, there was no difference in Gal-3 expression between the control and 72 h TNF-α treatment (Fig. 4G).
Endothelial cells spontaneously form microvascular networks within microfluidics chips that are impacted by the presence of ACH-3P cells and inflammatory conditions. In order to investigate mechanisms of placental growth and vasculature, our next step was to examine microvascular network formation of HUVECs both in the presence and absence of ACH-3Ps, and/or inflammation (Fig. 5A1 and A2). Immunofluorescent microscopy showed intricate microvascular network formation by endothelial cells with junctions between branches (Fig. 5A and B). Within the microfluidic environment without ACH-3Ps (Fig. 5A-1), 24-hour TNF-α treatment had no effect on endothelial FKBPL protein expression, however, after 72 hours FKBPL protein expression was reduced by ~ 4-fold (control 1.00 ± 0.1 vs TNF-α-24h 1.12 ± 0.05 vs TNF-α-72h 0.28 ± 0.01, fold change, p < 0.0001; Fig. 5A-1, B and C). However, in the co-culture system (Fig. 5A-2), the presence of trophoblast cells and TNF-α treatment increased the FKBPL expression of HUVECs following 24 hours of TNF-α treatment which was reduced significantly by 72 hours (control 1.00 ± 0.08 vs TNF-α-24h 1.50 ± 0.11 vs TNF-α-72h 0.18 ± 0.01, fold change, p < 0.0001; Fig. 5A-2, D and supplementary Fig. 7A). In the 3D monoculture microfluidics HUVEC system, there was no difference in endothelial Gal-3 protein expression within 24-hour TNF-α treatment, however, after 72 hours, Gal-3 protein expression was also reduced by ~ 4-fold (control 1.00 ± 0.01 vs TNF-α-24h 0.77 ± 0.13 vs TNF-α-72h 0.17 ± 0.02, fold change, p = 0.0006; Fig. 5A-1, E and supplementary Fig. 6). Similar to FKBPL expression, when ACH-3Ps were introduced to microfluidic chips, TNF-α treatment for 24 hours led to an initial increase in Gal-3 expression of HUVECs by ~ 2.5-fold that was significantly reduced by 72 hours (control 1.00 ± 0.16 vs TNF-α-24h 2.34 ± 0.09 vs TNF-α-72h 0.41 ± 0.02, fold change, p < 0.0001; Fig. 5A-2, F and supplementary Fig. 7B).
When there were no trophoblast cells in the system, there was a significant increase in CD31 protein expression of HUVECs following 24 hours of TNF-α treatment, which was reduced significantly by 72 hours (control 1.00 ± 0.07 vs TNF-α-24h 1.25 ± 0.04 vs TNF-α-72h 0.29 ± 0.02, fold change, p < 0.0001; Fig. 5A-1, B and G). In the presence of ACH-3Ps a progressive decrease in CD31 expression following TNF-α treatment was observed (control 1.00 ± 0.07 vs TNF-α-24h 0.57 ± 0.11 vs TNF-α-72h 0.05 ± 0.004, fold change, p = 0.0004; Fig. 5A-2, H and supplementary Fig. 7A) in the system. FKBPL is known to have an anti-angiogenic function,[18] therefore an increase in FKBPL at 24 hours likely led to inhibited vasculature and hence CD31 expression at 72 hours, which initiated a downregulation of FKBPL as a result of compensatory mechanism to restore angiogenesis.
We also examined the 3D HUVECs monoculture and co-culture microvascular network structures within microfluidics chips by measuring the number of master segments, master junctions and total isolated branches using an Angiogenesis Analyzer macro.[54] There was a significant difference in 3D HUVECs monoculture and co-culture microvascular network structures and following 72 h TNF- α treatment including the number of master segments (-ACH3Ps vs + ACH-3Ps; control 636.3 ± 47.32 vs 409.8 ± 10.7, TNF-α-24h 616.7 ± 6.7 vs 320.6 ± 10.4, TNF-α-72h 327.7 ± 63.7 vs 82.3 ± 11.68, p < 0.0001; Fig. 6A and B), number of master junctions (-ACH-3Ps vs + ACH-3Ps; control 297.3 ± 16.5 vs 204.3 ± 19.4, TNF-α-24h 274.7 ± 3.3 vs 184.2 ± 5.3, TNF-α-72h 163.7 ± 24.6 vs 99.7 ± 16.60, p < 0.0001; Fig. 6C) and total isolated branches (-ACH-3Ps vs + ACH-3Ps; control 1575 ± 232.4 vs 397.0 ± 65.2, TNF-α-24h 2352 ± 219.3 vs 495.8 ± 102.4, TNF-α-72h 3223 ± 114.6 vs 4568 ± 539.9, p < 0.0001; Fig. 6D). The vast majority of these results show reduction in microvascular growth in our placenta-on-a-chip model in the presence of trophoblast cells or TNF- α (Fig. 6B and C), which is reflective of SUA remodeling during placental development. However, prolonged TNF-α treatment (72h) in addition to inducing reduced placental microvascular growth, also impaired the vasculature measured by the increase in isolated branches, which is even more pronounced in the presence of ACH-3Ps (Fig. 6D).