Human umbilical cord mesenchymal stem cell derived exosomes (HUCMSC-exos) recovery soluble fms-like tyrosine kinase-1 (sFlt-1)-induced endothelial dysfunction in preeclampsia

Human umbilical cord tissues were collected from full-term, healthy cesarean births and were donated by consenting mothers at the Department of Obstetrics of Shanghai First Maternity and Infant Hospital from 2018 to 2019. The demographic and clinical characteristics of the participants in this study were reported in our previous study [17]. The HUCMSCs from human umbilical cord Wharton’s jelly were isolated by the tissue adherence method [18]. Umbilical arteries and veins were removed, and the remaining Wharton’s jelly was transferred to sterile, ice-cold Hank's balanced salt solution (HBSS, Gibco, Thermo Fisher Scientific, Inc., Waltham, MA, USA) containing antibiotics (streptomycin/penicillin, Gibco). The Wharton’s jelly tissues were washed multiple times with HBSS to remove excess blood and were subsequently cut into small pieces. The pieces were transferred to 10 cm² dishes with 5 mL of minimum Eagle’s medium alpha (Gibco) containing 10% fetal bovine serum (FBS, Gibco) and 1% streptomycin/penicillin and were incubated at 37 °C in 5% CO₂. The explants were left undisturbed for 7–10 days to allow cell migration, and the medium was changed every three days. After 2 weeks, the HUCMSCs (passage 0) were grown to 60–70% confluence and passaged by trypsinization. Sample collection was approved by the Research and Ethics Committee of Shanghai First Maternity and Infant Hospital (No. KS2024).

MSC surface marker detection by a Fluorescence-activated cell sorting (FACS) Calibur Flow Cytometer (BD Biosciences, San Jose, CA, USA) was previously shown [17]. Briefly, 1 × 10⁶ passage 3 cells were trypsinized and suspended in 100 µL of phosphate buffer saline (PBS). Then, the cells were incubated for 30 min with the following monoclonal antibodies against HUCMSCs surface markers: CD45-PE, CD73-APC, CD90-FITC and CD105-PE-Cy7 (BD Biosciences, San Jose, CA, USA).

For osteogenic differentiation, passage 3 cells were cultured in differentiation medium (Gibco) for 3 weeks in six-well plates. After the cells were fixed with 4% paraformaldehyde (PFA) and stained with alizarin red S (Sigma–Aldrich, Merck, Darmstadt, Germany), the cells were observed by microscopy (Nikon Eclipse Ti, Tokyo, Japan).

HUCMSCs at passages 3–6 were cultured in FBS-free medium with 2% bovine serum albumin (BSA, Sigma–Aldrich) for 48 h, and then 500 mL of the supernatant was centrifuged at 3000 g for 10 min to remove cell debris and passed through a 0.22-µm filter. The cleared supernatant was ultracentrifuged at 110,000 g for 70 min and washed in PBS using the same ultracentrifugation conditions to isolate exosomes. Exosome pellets were suspended in 5 mL of PBS and stored at − 80 °C.

The size and morphology of the exosomes were examined using a transmission electron microscope (TEM) at the Laboratory of Electron Microscopy (Chinese Academy of Sciences, Shanghai, China). The size distribution of HUCMSC-exos was determined using nanoparticle tracking analysis (NTA, ZetaView, Particle Metrix, Meerbusch, Germany). These procedures were performed as previously described [12].

All mouse experiments were approved by the Department of Laboratory Animal Science, Tongji University (No. TJLAC-020-061). Adult male (n = 40) and female (n = 80) CD-1 (ICR) mice were purchased from Jackson Laboratories. Animals were housed in a temperature- and humidity-regulated environment with a 12-h light cycle. For a consistent and accurate assessment of the gestational age of mouse embryos, male and female (1:2) mice were pair-housed for one night. Embryonic day 0.5 (E0.5) was designed as the first morning at which a vaginal plug was noted. A well-characterized sFlt-1 overexpression model was established by injecting 6 × 10⁸ pfu of the murine sFlt-1-expressing adenovirus into pregnant ICR mice via the tail vein on E8.5 [19].

Pregnant mice (n = 32) were randomly divided into four groups (n = 8 per group) and injected with either Cytomegalovirus (CMV)-null adenovirus or murine sFlt-1 adenovirus on E8.5 and HUCMSC-exos (EXO, 100 µg/dam) or sterile saline (NS, 100 µL) on E6.5, E9.5, E12.5 and E15.5 via the tail vein. The groups were designated as follows: (i) control (CTL): pregnant mice administered CMV-null adenovirus and NS; (ii) EXO: pregnant mice administered CMV-null adenovirus and HUCMSC-exos; (iii) sFlt-1: pregnant mice administered sFlt-1 adenovirus and NS; and (iv) sFlt-1 + EXO: pregnant mice administered sFlt-1 adenovirus and HUCMSC-exos. In total, there were 4 treatment groups with respect to sFlt-1 overexpression and HUCMSC-exos administration. Blood pressure was noninvasively measured on E18.5 by determining the tail blood volume with a volume pressure recording sensor (CODA System, Kent Scientific, Torrington, CT, USA) that was averaged over a 10-min period. Embryos were harvested on E18.5, and serum, tissues, and urine samples were obtained before euthanasia as described elsewhere [20].

Placental and kidney tissues were fixed with 4% PFA for 48 h and processed by conventional procedures. Sections 3–5 μm in thickness were cut from the paraffin-embedded tissues, mounted on poly-L-lysine-coated slides, deparaffinized in xylene, dehydrated in alcohol and then stained with hematoxylin and eosin (H&E) or periodic acid-methenamine silver (PAS) stain.

For IHC analysis, paraffin sections were deparaffinized and incubated with citrate buffer for antigen retrieval. The slides were then incubated with rabbit anti-CD31 polyclonal antibody (1:100, Abcam, Cambridge, MA, USA) overnight at 4 °C and developed using the ImmPRESS horseradish peroxidase (HRP) anti-rabbit IgG polymer detection kit (Weiao, Shanghai, China).

Using CD31 as an endothelial marker, the densities, diameters, and areas of fetal blood vessels were analyzed as previously described [21]. Four images per tissue section were taken using a Nikon inverted microscope with a 40 × objective (vascular density) or 100 × objective (vascular diameter). For each image, the number of capillaries was counted, and the lumen area was measured using ImageJ imaging analysis software (NIH, Bethesda, MD). Five capillaries per image were randomly selected for diameter measurements by CaseViewer software (3DHISTECH, Ltd., Hungary).

ELISAs for mouse sFlt-1 and sEndoglin were performed according to the manufacturer’s instructions (R&D Systems). Urine albumin/creatinine ratio was measured using Urinary Albumin and Creatinine Assay kits (Abnova, Taipei City, Taiwan, China). Briefly, the various samples were diluted 1:2 in dilutions and were incubated in a 96-well plate pre-coated with capture antibodies. The wells were washed and incubated with a secondary antibody conjugated to horseradish peroxidase. Then substrate solution was added, and optical density was determined at 450 nm. The concentrations were calculated using a standard curve of the respective recombinant proteins.

Total protein concentration was measured using the bicinchoninic acid assay (Pierce^®, Thermo Fisher Scientific, Bonn, Germany). Western blot for expressions in exosomes was performed using CD63 (1:1000, System Bioscience, SBI, USA), CD81 (1:1000, SBI), CD9 (1:1000, SBI), sFlt-1 (1:1000, Abcam), Flag (1:2000, Sigma), eNOS (1:1000, Cell Signaling Technology, CST), Versican (1:1000, Abcam), α-Tublin (1:2000, Abmart, Shanghai, China), Flotillin-1 (1:1000, CST) and GAPDH (1:5000, Abmart) antibodies by previously described methodology.

The HUVECs were isolated from 3 individual donors by a standard collagenase enzyme digestion method and cultured steadily in Endothelial Cell Medium (ECM, ScienCell, San Diego, CA) containing 5% FBS, 1% P/S and 1% endothelial cell growth supplement (ECGS).

The human sFlt-1 (ID: NM_001159920.1) cDNA was cloned into Lenti-X Tet-One System expression vector (Clontech, Moutain View, CA) as described previously. The recombinant lentivirus tet-sFlt-1 and the negative control lentivirus (NC-lentivirus; Hanyin Co. Shanghai, China) were prepared and titered to 10⁹ TU/ml (transfection unit). HUVECs were infected with lentiviruses expressing Tet-on-sFlt-1 to obtain cells overexpressing sFlt-1 only at the time when tetracycline existing. HUVECs expressed the sFlt-1 in the cellular background with doxycycline (Dox) and the efficiency of overexpression was examined by western blot analysis.

Exosomes were labeled with the fluorescent dye 1,10-dioctadecyl-3,3,30,30-tetramethylindocarbocyanine perchlorate (Dil, red, Sigma) by addition to PBS and incubated for 15 min according to the manufacturer’s protocol [22]. The labeled exosome suspensions were filtered using a 100-kDa MWCO hollow fiber membrane (Thermo Fisher Scientific) to remove the excess dye. HUVECs were seeded in 6-well plates and incubated with the Dil-labeled exosomes (100 μg/mL) for 24 h. Then the cells were fixed with 4% paraformaldehyde for 15 min and stained with phalloidin-FITC (Sigma). Nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI, Sigma). The cellular uptake of exosomes was tested by a confocal microscopy (TCS SP8; Leica, Wetzlar, Germany).

Cell proliferation and migration ability was performed using modified systems as described previously [23]. In brief, 3 × 10³ cells/well were plated in 96-well plates, later Dox or the vehicle were added to induce overexpression of sFlt-1. After 24 h, HUCMSC-exos with different doses were added. Then cell numbers were assessed using the cell counting kit-8 (CCK-8) at 450 nm. For migration, HUVECs were plated in 6-well plates and treated with Dox or the vehicle to induce overexpression of sFlt-1. Then 3 × 10⁴ cells were seeded into the upper chambers and exosomes (100 µg/mL) or NS were added into the lower chambers. Following incubation for 16 h, fluorescent stain (calcein-AM) was added to each lower chamber. The migrated cells were counted by fluorescence analysis (Nikon, Tokyo, Japan).

We isolated HUCMSCs from human umbilical cord Wharton’s jelly using the tissue adherence method. Observation under an inverted microscope showed that the HUCMSCs (passage 0) had long fusiform shapes (Fig. 1A). After osteogenic induction, alizarin red staining showed multiple calcium nodules in the cells, which indicated that the cells had well-developed osteogenic differentiation functions (Fig. 1B). Flow cytometric analysis showed that the HUCMSCs were positive for CD73, CD90 and CD105 but lacked the hematopoietic marker CD45 (Fig. 1C), which was consistent with the characteristics of MSCs.

The characteristics of HUCMSC-exos isolated and purified using a well-established ultracentrifuge method. Our hypothesis was supported by transmission electron microscopy analysis, which showed a cup shape (Fig. 1D). NTA identified particles with diameters of 30–150 nm (Fig. 1E). Furthermore, exosomes were positive for the exosomal protein markers CD63 and CD81 (Fig. 1F). These results indicated that we successfully isolated HUCMSC-exos.

We first established the preeclampsia-like mice model by injecting the sFlt-1 adenovirus into pregnant ICR mice via the tail vein on E8.5. HUCMSC-exos were administered to examine the potential of exosomes in treating PE. Pregnant mice were injected with exosomes or diluent on E6.5, E9.5, E12.5 and E15.5 (Fig. 2A). Hypertension is the most common diagnostic sign of PE, then blood pressure response in the sFlt-1-induced preeclampsia model was examined and shown in Fig. 2C. The sFlt-1-injected mice had both increased systolic and diastolic blood pressure. No significant differences were observed among the CTL and EXO groups, while dams treated with both sFlt-1 and HUCMSC-exos remained nearly normotensive on E18.5 relative to that of dams treated only with sFlt-1. SFlt-1 injection induced the production of sFlt-1 and increased its serum concentration in mice (Fig. 2E, sFlt-1 vs CTL, P < 0.01) without affecting sEndoglin (sEng) levels. Circulating sFlt-1 concentrations were slightly decreased by exosome treatment (sFlt-1 vs EXO + sFlt-1, P = 0.13). Interestingly, dams treated with both sFlt-1 and HUCMSC-exos showed slightly reduced sEng level (sFlt-1 vs EXO + sFlt-1, P = 0.09), thus further studies need to be explored.

The effects of exosomes on PE-induced IUGR were evaluated, and the observed decrease in body weight (Fig. 2B) was most likely caused by small-sized fetuses and placentas rather than a reduction in the number of fetuses (Fig. 2F). Fetuses that survived to term had decreased weights and were small in the sFlt-1-treated group, and exosomes prevented this growth restriction in fetuses and placentas (Fig. 2F). The urine albumin/creatinine ratio (ACR) was slightly increased after sFlt-1 injection, but the difference was not significant (Fig. 2D P=0.19). These results suggest that HUCMSC-exos significantly reduce the IUGR that occurs in the fetuses of dams exposed to sFlt-1. Therefore, HUCMSC-exos have the potential to treat PE by improving sFlt-1-induced hypertension and IUGR in mice.

The sFlt-1 virus-treated mice showed retarded fetal development, which is driven by impaired placental feto-maternal exchange. Thus, we examined the histology of the placentas, including the labyrinth (La) and junctional zone (JZ) by H&E and IHC staining (Fig. 3A). The total placental area did not change among the four groups (Fig. 3B), while the thickness of the La layer was notably reduced in sFlt-1-injected mice and rescued by HUCMSC-exo (Fig. 3C, D). Then, we analyzed placental vascularization by staining for CD31, which specifically binds to endothelial cells. High magnification views showed extensive vascular damage in the La, which indicated impaired blood flow, in the sFlt-1 group. Importantly, in EXO + sFlt-1 mice, there was a partial reversal of the vascular narrowing observed in placentas from sFlt-1 mice, as indicated by increased lumen diameters and areas (Fig. 3E). The number of lumens per image in the exosome-treated group relative to the sFlt-1 group was slightly increased but was not significant (Fig. 3F). Quantification of vascular diameter after CD31 staining showed that this decreased from 22.1 ± 1.95 μm in CTL mice to 14.45 ± 1.64 μm in sFlt-1 mice (P < 0.01) and reversed to 21.6 ± 1.44 μm in EXO + sFlt-1 mice (P < 0.01) (Fig. 3G). Furthermore, the fetal vascular area (%) was reduced in sFlt-1 mice (19.49 ± 0.71 vs 25.23 ± 1.38, p < 0.001) and increased to 22.75 ± 1.4 in EXO + sFlt-1-treated mice (p = 0.06) (Fig. 3H). There was no significant difference in the vascular number, diameter, or area between EXO and CTL mice. Neither glomerular endotheliosis nor mesangial expansion, which are characteristic features of PE, was present in any group, as determined by PAS stain (Fig. 3I).

A substantial body of evidence indicates that PE in humans is a consequence of vascular endothelial damage, resulting in multiorgan dysfunction. Therefore, we chose HUVECs as the cell model for studying the effects of HUCMSC-exos on endothelial dysfunction in PE. The Tet-On-sFlt-1_Flag HUVEC model was established to simulate vascular endothelial cells in women with PE. In the absence of Dox, negative control (NC) HUVECs did not express any detectable sFlt-1 protein, and following induction with different doses of Dox, strong signals for sFlt-1 were observed in sFlt-1-overexpressing (OV-sFlt-1) cells, as measured by western blotting (Fig. 4A). HUCMSC-exos labeled with the fluorescent dye Dil were incubated with NC and OV-sFlt-1 HUVECs for 24 h, and most recipient cells were positive for Dil fluorescence and showed no difference among the groups (Fig. 4B).

A series of cellular functional analyses were performed, and we first demonstrated that HUCMSC-exos are capable of facilitating angiogenesis because they contain many proangiogenic components. OV-sFlt-1-HUVEC proliferation and migration were significantly inhibited, and these attenuated angiogenetic behaviors were reversed after incubation with HUCMSC-exos (Fig. 4C, D). It is suggested that high levels of circulating sFlt-1 lead to suppression of eNOS signaling pathway, which, in turn, enhances sensitivity to vasoconstrictors that induces maternal hypertension in PE. Therefore, we examined the protein expression of eNOS in HUVECs and as expected, found that OV-sFlt-1-HUVECs exhibited decreased eNOS protein expression, which was partially rescued in the presence of HUCMSC-exos (Fig. 4E). These findings suggest that HUCMSC-exos have the potential protective effects on sFlt-1-induced endothelial and eNOS dysfunction.

HUCMSC-exos modulated sFlt-1-induced endothelial dysfunction and improved placental angiogenesis in PE-like mice. It was, therefore, decided to investigate how exosomes enhanced vascular angiogenesis. After successfully extracting proteins from exosomes and parallel cells, proteomics analysis using tandem mass tag (TMT) technology was performed to detect protein expression profiles. Unsupervised hierarchical clustering analysis was used to generate a heat map of the differentially expressed proteins, and a total of 568 proteins were differentially expressed between HUCMSC-exos and HUCMSCs using the cutoff value of a two-fold change and a P-value < 0.05 (Fig. 5A). On the basis of gene ontology (GO) enrichment analysis, both HUCMSCs and HUCMSC-exos displayed functional enrichment in biological processes related to vesicle-mediated transport, cell communication, cell migration, and angiogenesis (Fig. 5B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis is shown in Fig. 5C. The degree of gene enrichment was represented by abscissa, the amount of gene enrichment was represented by the bubble size, and p-value was represented by the color depth. The main pathways of enrichment included tight junction, protein export, platelet activation, oxidative phosphorylation, HIF-1 signaling pathway, gap junction, focal adhesion, endocytosis, EGFR tyrosine kinase inhibitor resistance and extracellular matrix–receptor (ECM-receptor). We next analyzed the differentially expressed genes (DEGs) between these two chambers. Compared with HUCMSCs, the up-expressed proteins in HUCMSC-exos were MMP2 (Matrix metalloproteinase 2, fold change = 14.7) and VCAN (versican, fold change = 11.7). Previous study showed that MSC-exo could promote the migration of endothelium cells by delivering MMP2. Then, we verified that VCAN, which contributes to tissue development and maintenance by participating in cell adhesion, proliferation and angiogenesis via binding to ECM components, was obviously upregulated in HUCMSC-exos compared with cell lysates (Fig. 5D).