Placenta-derived Exosomes Mitigate Hypoxia-Induced Trophoblast Apoptosis and Inflammatory Progression via SASH1

Summary

SAM and SH3 domain-containing protein 1 (SASH1) is identified as a key regulator of trophoblast apoptosis and inflammation in pre-eclampsia (PE). Placenta-derived exosomes (P-EXOS) alleviate hypoxia-induced injury by suppressing SASH1 expression. These findings reveal a novel mechanism underlying PE pathogenesis and suggest SASH1 as a potential therapeutic target.

Abstract

SASH1 is a signal adaptor protein involved in cell growth, apoptosis, and immune regulation, and has been increasingly studied in tumor and immune cells. Emerging evidence suggests that SASH1 plays an important role in inflammatory responses and cellular homeostasis, processes that are closely associated with the development of PE. This study aimed to determine whether SASH1 contributes to trophoblast apoptosis and inflammatory responses in PE and whether P-EXOS exerts protective effects through SASH1 regulation. In this study, three PE-related transcriptomic datasets (GSE75010, GSE10588, and GSE60438) were analyzed to identify shared differentially expressed genes (DEGs), followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Machine learning algorithms were further applied to screen key candidate genes, and single-cell RNA sequencing data were used to characterize cellular heterogeneity in placental tissue and to determine cell type-specific expression patterns. SASH1 was identified as a consensus candidate gene and was significantly upregulated in trophoblast cells from PE samples. In vitro, a hypoxia-treated HTR-8/SVneo trophoblast cell model was established, combined with SASH1 knockdown, SASH1 overexpression, and co-culture with P-EXOS. Functional experiments showed that knockdown of SASH1 significantly suppressed hypoxia-induced trophoblast apoptosis and reduced the secretion of pro-inflammatory cytokines, including IL-6, IL-1β, and TNF-α, whereas SASH1 overexpression promoted apoptosis and inflammatory responses. In addition, P-EXOS treatment markedly reduced SASH1 expression at both mRNA and protein levels and attenuated hypoxia-induced trophoblast injury, while SASH1 overexpression largely abolished these protective effects. Taken together, these findings indicate that SASH1 plays a critical role in trophoblast apoptosis and inflammatory responses in PE. P-EXOS may alleviate hypoxia-induced trophoblastic injury by suppressing SASH1 expression, providing new insights into the molecular mechanisms and potential therapeutic targets for PE.

Introduction

PE is a pregnancy-specific multisystem disorder that typically occurs after 20 weeks of gestation and is clinically characterized by hypertension, proteinuria, and multi-organ dysfunction. It remains a leading cause of maternal and perinatal morbidity and mortality worldwide¹. Despite continuous advances in clinical management, the definitive treatment for PE is still delivery of the placenta, underscoring the incomplete understanding of its pathogenesis and the urgent need to identify novel molecular and cellular mechanisms as well as potential therapeutic targets^2,3.

Adequate trophoblast invasion and promotion of normal placental development are critical mechanisms underlying the pathogenesis of PE⁴. During normal pregnancy, trophoblast cells invade the uterine decidua and remodel spiral arteries to establish a low-resistance, high-capacity circulation. Impaired invasion and migration of trophoblast cells leads to reduced placental perfusion, resulting in a persistent hypoxic and oxidative stress environment. Hypoxia directly impacts cellular apoptosis, and aberrant release of pro-inflammatory mediators disrupts homeostasis at the maternal-fetal interface, further exacerbating placental dysfunction⁵. Inflammation is also an important driver of PE. IL-17, IL-6, and TNF-α have been found to be markedly elevated in the placenta and peripheral blood of PE patients⁶. Hypoxia and inflammation mutually reinforce each other, forming a vicious cycle that amplifies systemic inflammation, causes endothelial damage, and contributes to placental abnormalities. The underlying molecular networks governing trophoblast apoptosis and inflammatory responses remain largely unknown.

SASH1 is a member of the SLy/SASH1 family of intracellular scaffold proteins that functions as a signaling adaptor⁷. SASH1 has been primarily characterized as a tumor suppressor that inhibits epithelial-mesenchymal transition (EMT), cell migration, and invasion through interactions with signaling partners such as CRKL and the PI3K-Akt-mTOR pathway^8,9. Beyond its tumor-suppressive roles, emerging evidence suggests that SASH1 participates in inflammatory signaling and immune regulation⁷. However, the expression and function of SASH1 in the placenta, particularly in the context of trophoblast biology and PE, have not been investigated.

Exosomes have emerged as important cellular mediators and are being increasingly studied in the context of pregnancy¹⁰. Exosomes transport various bioactive cargos, including proteins, lipids, and noncoding RNAs, to modulate recipient cell function. Increasing evidence suggests that P-EXOS play a critical role in immune regulation during pregnancy, and that their activity and cargo are significantly altered in PE, potentially intensifying the inflammatory state¹¹. Exosomes have also been implicated in regulating oxidative stress¹² and angiogenesis¹³, potentially exerting beneficial effects in pregnancy-related complications. However, their molecular targets and mechanisms of action in PE remain incompletely understood.

It was hypothesized that SASH1 is dysregulated in the PE placenta and functions as a key driver of trophoblast apoptosis and inflammatory responses, and that P-EXOS may exert protective effects against trophoblast injury by modulating SASH1 expression. To test this hypothesis, multiple PE transcriptomic datasets were integrated and single-cell RNA sequencing data with multi-algorithm machine learning to identify key candidate genes. Using a hypoxia-treated HTR-8/SVneo trophoblast cell model and exosome-based intervention experiments, the functional role of SASH1 in trophoblast dysfunction and the therapeutic potential of P-EXOS in PE was investigated.

Protocol

Ethics statement

This study was approved by the Ethics Committee of Shijiazhuang Fourth Hospital (Approval No. 20200031). Written informed consent was obtained from all placental tissue donors prior to sample collection. All procedures involving human participants were conducted in accordance with the Declaration of Helsinki. A complete list of reagents, consumables, equipment, and software used in this protocol is provided in the Table of Materials.

Data collection

RNA-seq data were obtained from the GEO database. The datasets included GSE75010, which contains gene expression data from 157 PE placentas and 173 non-PE placentas (N = 330). GSE10588 contains gene expression data from 26 normal placentas and 17 severe PE placentas (N = 43). GSE60438 contains transcriptome profiling data of decidua basalis from pre-eclamptic patients and normotensive pregnancies (N = 125). Single-cell transcriptome data were obtained from the GEO dataset GSE183338. It includes single-nucleus samples from the chorionic villi/maternal-fetal interface of PE and healthy pregnancies.

DEGs analysis

DEGs associated with PE, the GSE75010, GSE10588, and GSE60438 datasets were first preprocessed and normalized. Then, differential analysis was performed using the R package “limma”¹⁴ based on the sample grouping information. Genes with p < 0.05 and |log2FC| > 0.5 were selected. Volcano plots of DEGs were generated using the R package ggplot2. Heat maps of the top 20 DEGs were drawn using the R package pheatmap. Subsequently, the intersection of the DEGs selected from the three datasets was taken, and a protein-protein interaction (PPI) network was constructed for the candidate genes using the online platform STRING, with an interaction score ≥0.15. The top 20 hub genes were further identified from this PPI network based on their degree of connectivity, ranked using Cytoscape software. The PPI network results were visualized using Cytoscape software or STRING.

Enrichment analysis

Gene enrichment analysis was performed using the ClusterProfiler and DOSE packages in combination with the Metascape website. The databases were obtained from GO and KEGG. Enrichment analysis was conducted using the “EnrichGO” function. Pathways with p < 0.05 were considered significantly enriched. The enrichment results were visualized using the “ggplot2” and “ggpubr” packages.

Machine learning

To identify robust and biologically meaningful DEGs associated with PE, a multi-model machine learning feature selection analysis was performed based on the publicly available transcriptome dataset GSE60438 (platform: GPL6884). This dataset contains expression profiling of decidua basalis samples collected from pre-eclamptic and normotensive pregnancies at Cesarean section. The pre-filtered DEGs were standardized, and the expression matrix, together with corresponding clinical grouping information, was used as input for four distinct machine learning algorithms in order to reduce model bias and enhance the stability of feature selection.

The four algorithms were applied simultaneously, without a specific order. LASSO was performed using the “glmnet” package to conduct regression analysis and select important feature genes. An L1 regularization term was added to the loss function, which shrinks the coefficients of less important features to zero, thereby achieving feature selection. SVM-RFE was implemented using the “e1071” package to construct a support vector machine with recursive feature elimination. A classifier was first trained using SVM, and the least informative features were iteratively removed based on feature weights, yielding an optimal feature subset. XGBoost was applied using the “xgboost” package to build multiple decision trees. Each tree fitted the residuals of the previous tree, and the weighted outputs were accumulated to obtain the final prediction. Boruta was performed using the “randomForest” package, generating shadow features that competed with real features in training a random forest. Features with importance values significantly higher than random noise were retained.

Single-cell transcriptome data analysis

Single-cell transcriptome data were obtained from the GEO database, and the raw count matrix was retrieved from GSE183338. The count matrix was imported using the “Read10X” function of the Seurat package and converted to a dgCMatrix format. Individual objects were merged into a single aggregate object using the “merge” function, and cell labels were made unique using “RenameCells”. Low-quality cells were filtered based on the following criteria: genes expressed in fewer than three cells were removed, and cells expressing fewer than 200 genes were excluded. Quality-controlled cells were normalized and highly variable genes were identified. Global scaling normalization was applied using “LogNormalize” (scale factor = 10,000), highly variable genes (n = 2,000) were selected using “FindVariableFeatures”, and data were scaled using “ScaleData”. Principal component analysis was performed on highly variable features, and the top 30 principal components were retained. Batch effects between samples were corrected using the Harmony method. Cells were visualized and downscaled using UMAP. Shared nearest neighbor graphs were constructed using “FindNeighbors” and “FindClusters” based on the Louvain algorithm. The resolution parameter in “FindClusters” was optimized between 0.1 and 1. The clustering tree was visualized using the “clustree” function, and a resolution of 0.9 was selected to define cell clusters. Potential doublets were removed using the Scrublet algorithm. Cell clusters were annotated by identifying differentially expressed marker genes using the “FindAllMarkers” function. The non-parametric Wilcoxon rank sum test was applied with Bonferroni correction. Cell identities were assigned based on surface markers, relevant literature, and the Cell Classification Database¹⁵.

Cell culture

The trophoblast cell line HTR-8/SVneo cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Hypoxic conditions were established by culturing cells at 1% O₂, 5% CO₂, and 94% N₂ for 24 h; normoxic controls were maintained at 20% O₂ and 5% CO₂¹⁶. All cell culture procedures should be performed in a Class II biosafety cabinet using aseptic technique. Culture media, transfection reagents, and cell waste should be disposed of in accordance with institutional biosafety guidelines.

Cell transfection

Plasmids containing sh-SASH1, sh-NC, OE-SASH1, and OE-NC were synthesized. HTR-8/SVneo cells were seeded at a density of 5 × 10⁵ cells per well in six-well plates. Cells were subsequently transfected with 2 µg sh-SASH1, sh-NC, OE-SASH1, or OE-NC plasmid per well using a transfection reagent according to the manufacturer's instructions. Briefly, plasmid DNA and P3000 Reagent were diluted in Opti-MEM, mixed with Lipofectamine 3000 diluted separately in Opti-MEM, incubated for 15 min at room temperature, and added to cells at 70–80% confluency. Forty-eight hours post-transfection, SASH1 expression was assessed by RT-qPCR and Western blot. The shRNA target sequences used for SASH1 knockdown are listed in Supplementary Table 1.

Real-time quantitative PCR

Total RNA was extracted from HTR-8/SVneo cells and reverse-transcribed into cDNA using a reverse transcription kit at 42 °C for 30 min, followed by 85 °C for 5 min. Real-time quantitative PCR (qPCR) was performed using SYBR Green master mix with the following cycling conditions: 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The relative mRNA expression was calculated using the ΔΔCt method, with β-actin as the internal reference. Primer sequences used in this experiment are listed in Supplementary Table 2.

Western blot assay

Total protein was extracted from HTR-8/SVneo cells using the lysis buffer. Cell lysates were collected, incubated on ice, and centrifuged at 12,000 × g for 30 min at 4 °C to remove insoluble debris. Protein concentration was determined using a spectrophotometer. Equal amounts of protein (50 µg) were separated by SDS-PAGE and subsequently transferred onto PVDF membranes. The membranes were blocked with 5% non-fat milk and incubated with primary antibodies overnight at 4 °C. After washing, the membranes were incubated with the corresponding secondary antibodies, and protein bands were visualized using an enhanced chemiluminescence detection system.

For protein detection, primary antibodies included anti-SASH1 and β-actin. Appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies—goat anti-rabbit and goat anti-mouse—were employed. β-actin was used as the internal loading control to ensure equal protein loading. The intensity of protein bands was measured and quantified using ImageJ software.

Isolation of P-EXOS

P-EXOS were isolated from placental villous tissue obtained from term placentas of healthy women undergoing elective cesarean section. Placental villous tissue was washed thoroughly with sterile PBS, minced into approximately 1 mm³ fragments, and cultured in RPMI-1640 medium supplemented with 10% exosome-depleted FBS at 37 °C in 5% CO₂ for 48 h. The conditioned medium was subjected to differential centrifugation as follows: 300 × g for 10 min to remove cells and tissue debris; 2,000 × g for 20 min to remove cell debris; and 10,000 × g for 30 min to remove microvesicles, all at 4 °C. The resulting supernatant was ultracentrifuged at 120,000 × g for 70 min at 4 °C to pellet exosomes. The pellet was washed once with PBS and re-ultracentrifuged at 120,000 × g for 70 min at 4 °C. The final pellet was resuspended in PBS. The isolated exosomes were characterized by Western blot analysis for exosome markers (PLAP, CD63, and TSG101, with GM130 as a negative control) and further examined by transmission electron microscopy for morphological observation.

P-EXOS Cellular Uptake Experiment

To confirm cellular internalization of P-EXOS, exosomes were fluorescently labeled with the lipophilic membrane dye PKH67 according to the manufacturer's protocol. Briefly, P-EXOS were incubated with PKH67 (4 µM) in Diluent C for 5 min at room temperature, and the reaction was quenched with an equal volume of 1% bovine serum albumin (BSA). Labeled exosomes were re-isolated by ultracentrifugation (120,000 × g, 70 min, 4 °C) to remove unbound dye. PKH67-labeled P-EXOS (50 µg/mL) were then added to HTR-8/SVneo cells and co-incubated for 24 h under normoxic or hypoxic (1% O₂) conditions. Cells were subsequently washed three times with PBS, fixed with 4% paraformaldehyde for 15 min, and the nuclei were counterstained with DAPI (1 µg/mL). Internalization of PKH67-labeled exosomes was visualized by confocal laser scanning microscopy (CLSM; excitation 490 nm, emission 502 nm). For functional co-culture experiments, HTR-8/SVneo cells were treated with P-EXOS at a concentration of 50 µg/mL (protein equivalent) in complete RPMI-1640 medium supplemented with 10% exosome-depleted FBS under hypoxic conditions (1% O₂) for 24 h.

Enzyme-linked immunosorbent assay (ELISA)

Cell culture supernatants were collected, and the levels of IL-6, IL-1β, and TNF-α were measured using IL-6 ELISA kit, IL-1β ELISA kit, and TNF-α ELISA kit, respectively, according to the manufacturers' instructions. Absorbance at 450 nm was measured using a microplate reader, and the actual concentrations were calculated from the standard curves.

TdT-mediated dUTP nick-end labeling (TUNEL)

Apoptotic cells were detected using the TUNEL assay kit according to the manufacturer's instructions. Briefly, cells were fixed with 4% paraformaldehyde for 15 min at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 5 min on ice, and incubated with TUNEL reaction mixture for 60 min at 37 °C in the dark. Nuclei were counterstained with DAPI, and TUNEL-positive cells were visualized using a fluorescence microscope and quantified by counting the percentage of TUNEL-positive cells in at least five randomly selected fields per sample.

Statistical analysis

All data were analyzed using R and GraphPad Prism. Continuous variables are presented as mean ±SD. Two-group comparisons were performed using Student’s t-test, whereas multiple-group comparisons were conducted using one-way ANOVA followed by Tukey’s post hoc test. Statistical significance for categorical variables was assessed by the Chi-square test or Fisher’s exact test. Unless otherwise stated, correlations between molecules were calculated using Spearman correlation analysis. Exosome characterization experiments were performed using P-EXOS isolated from three independent placenta donors. Cell-based experiments were performed in three independent biological replicates, representing independent experiments conducted on separate occasions using HTR-8/SVneo cells of different passages, with each replicate using P-EXOS isolated from a different placental donor. A p < 0.05 was considered statistically significant.

Results

Transcriptomic analysis of healthy and PE patients

For gene expression differences between the normal (Normal) group and PE patients, the differential expression by the R package limma in the merged datasets was analyzed. A total of 89 DEGs were identified (p < 0.05 and |log₂FC| > 0.5), 69 upregulated genes and 20 downregulated genes for GSE75010 (Figure 1A). There are 2,097 differentially expressed genes, and 1,121 are upregulated, and 976 are downregulated (Figure 1B). There are 7,633 DEGs for the GSE60438 dataset with 3,371 upregulated genes and 4,262 downregulated genes (Figure 1C). The substantial differences in DEG numbers across the three datasets likely reflect differences in sample size, tissue origin (placental villi vs. decidua basalis), and PE severity classification, which further underscores the value of taking the intersection across datasets to identify robust common DEGs.

Functional enrichment analysis of DEGs

For the systematic identification of key DEGs associated with PE, Venn analysis was done on the DEGs from the GSE75010, GSE10588, and GSE60438 PE datasets, and ultimately identified 56 common DEGs (Figure 2A). A PPI network was constructed for these 56 common DEGs using the online platform STRING (interaction score ≥0.15), and the top 20 hub genes were further identified based on their degree of connectivity within the network, ranked using Cytoscape software (Figure 2B). Based on such genes, GO enrichment was also performed to explore potential functionalities in BP, CC, and MF (Figure 2C). The DEGs were most likely enriched in hormone secretion and regulation of transmembrane receptor serine/threonine kinase signaling. In terms of cells, the genes were mainly localized to focal adhesions, cell–substrate junctions, and a variety of vesicle- and lysosome-related genes. In molecular functions, the enriched terms were prolyl hydroxylase activity, hormone activity, and transmembrane transport of lipid-related molecules. These results suggest that the DEGs may play important roles in placental cell signaling, cell-matrix interaction, and endocrine-related function maintenance. At the same time, it was found that KEGG pathway enrichment yielded the DEG most likely to be enriched in cytokine–cytokine receptor interaction, cell adhesion molecule interaction, and the TGF-β and HIF-1 signaling pathways (Figure 2D). Enrichment of certain virus-related terms (e.g., Hepatitis viruses, Ebolavirus) likely reflects non-specific overlap with shared immune-regulatory genes rather than a direct biological association with PE; the interpretive focus therefore remains on inflammation-, adhesion-, and hypoxia-related pathways. These pathways are related to inflammatory response, cell adhesive-mediated, and hypoxia-related signaling and may have an influence on PE progression by varying the immune-inflammatory status and cell-cell communication in the placental microenvironment.

Machine learning identifies SASH1 as a key gene in PE

To identify key prognostic genes, four machine learning methods were applied for feature selection. LASSO regression with cross-validation showed that the model achieved the minimum cross-validation error when log(λ.min) = -3.8075. Genes with non-zero coefficients at this λ value were selected as key features identified by LASSO (Figure 3A–B). The Boruta algorithm evaluates feature importance by comparison with shadow features. Genes with importance values higher than shadowMax were labeled as Confirmed and considered key candidate genes, whereas those with importance values lower than shadowMin were labeled as Rejected and regarded as having no significant contribution. Genes between these two thresholds were labeled as Tentative, indicating potential but uncertain importance (Figure 3C). SVM-RFE analysis demonstrated that when 40–50 features were retained, the cross-validation accuracy reached the highest or near-highest level, suggesting optimal predictive performance at this feature scale (Figure 3D). The moderate cross-validation accuracy (~0.6) indicates that no single algorithm performed optimally for this dataset; a multi-algorithm intersection strategy was therefore adopted to enhance the robustness of feature selection. XGBoost quantified the relative importance of genes using Gain values, thereby identifying features with higher contributions to model prediction (Figure 3E). Venn intersection analysis of the results from the four methods identified SASH1 as the only overlapping gene. Although SASH1 was classified as Tentative by the Boruta algorithm, its consistent selection across all three other algorithms supports its de Susignation as a consensus candidate gene (Figure 3F).

Single-cell analysis

The cellular heterogeneity and molecular processes in PE placental tissues were investigated. Cells were quality-filtered based on the number of detected genes, number of RNA counts, and percentage of mitochondrial gene expression. Low-quality cells (less than 200 detected genes or with more than 20% mitochondrial gene content) were removed to obtain a high-quality single-cell sample for downstream analysis. The quality control metrics and identification of highly variable genes are shown in Supplementary Figure 1A–B. After normalizing and identifying highly variable genes, principal component analysis (PCA) was performed using the top 2,000 highly variable genes and significant principal components using JackStraw analysis and ElbowPlot analysis (Supplementary Figure 2A–B).

Based on the principal components, uniform manifold approximation and projection (UMAP) was used to reduce dimensionality and perform unsupervised clustering, where all cells were classified into 22 clusters, highlighting the high cellular heterogeneity in placental tissue (Figure 4A). By studying cluster-specific marker gene expression and visualizing DEGs in a heatmap (Figure 4B), each cluster was annotated based on a canonical marker and projected back onto the UMAP embedding (Figure 4C). Based on this major placental cell populations, including B cells/plasma cells, endothelial cells, fibroblasts/stromal cells, macrophages/dendritic cells, natural killer cells, T cells, and trophoblast cells were identified. Comparisons between control and PE groups reveal that several types of cells present a substantial relative distribution in PE placentas, suggesting remodeling of the placental microenvironment (Figure 4D).

Tissue origin and disease status were used to track cells in the same UMAP embedding based on tissue source (decidua vs. villi) and disease grouping (control vs. PE) (Figure 4E). Cell number distribution was partially separated by tissue origin, while cells from different disease groups were highly interconnected. This suggests that PE does not substantially rearrange the entire transcriptomic structure, but may have effects on individual cell populations or on particular molecular features. SASH1 expression was mapped onto the UMAP to explore its distribution among placental cell populations (Figure 4F–G). SASH1 shows strong cell type-specific expression, with notable differences between decidual and villous cell populations. Statistical comparison of SASH1 expression between PE and control groups across cell types (Wilcoxon rank-sum test) revealed significant differences in multiple placental cell populations, with the most pronounced elevation observed in trophoblast cells from PE samples (Figure 4H). This finding provides single-cell-level evidence that SASH1 dysregulation is particularly pronounced in the trophoblast compartment, supporting the use of the HTR-8/SVneo trophoblast cell line for subsequent functional experiments.

SASH1 knockdown inhibits hypoxia-induced trophoblast apoptosis and inflammatory responses

To validate the bioinformatic predictions, a PE-related trophoblast cell model was established under hypoxic conditions (1% O₂). qPCR and Western blot analyses were performed to assess SASH1 expression (Figure 5A–B). Compared with normoxic HTR-8/SVneo cells, hypoxia-treated cells showed significantly increased SASH1 mRNA and protein expression levels, providing preliminary experimental validation of the bioinformatic results. To evaluate the effect of SASH1 knockdown on PE-related pathological processes, SASH1 was silenced in hypoxia-treated HTR-8/SVneo cells. Quantitative PCR and Western blot analyses (Figure 5C–D) showed that SASH1 expression was significantly reduced in all sh-SASH1 groups compared with the sh-NC group. Among them, sh-SASH1#3 exhibited the highest knockdown efficiency and was therefore selected for subsequent experiments. TUNEL staining and ELISA were performed under three conditions: normoxic control (Normal), hypoxia with negative control shRNA (Hypoxia + sh-NC), and hypoxia with SASH1 knockdown (Hypoxia + sh-SASH1). Compared with the Normal group, the Hypoxia + sh-NC group exhibited significantly increased apoptosis and elevated secretion of pro-inflammatory cytokines IL-1β, IL-6, and TNF-α, confirming hypoxia-induced trophoblast injury. SASH1 knockdown markedly attenuated these effects, reducing apoptosis (Figure 5E) and cytokine levels (Figure 5F) to levels approaching those of the Normal group, suggesting that SASH1 knockdown can largely rescue hypoxia-induced trophoblast apoptosis and inflammatory responses. To further establish the causal role of SASH1, gain-of-function experiments were performed. qPCR and Western blot confirmed successful SASH1 overexpression in HTR-8/SVneo cells (Figure 5G–H). Under normoxic conditions, SASH1 overexpression (OE-SASH1) significantly increased trophoblast apoptosis and pro-inflammatory cytokine secretion compared with the OE-NC group, recapitulating the hypoxia-induced phenotype (Figure 5I–J). Taken together, the loss- and gain-of-function data establish that SASH1 is both necessary and sufficient to drive trophoblast apoptosis and inflammatory responses, confirming its causal role in PE-related pathology.

P-EXOS suppresses hypoxia-induced trophoblast apoptosis and inflammatory responses by modulating SASH1

P-EXOS is considered to have potential therapeutic value in PE. In this study, bioinformatics analysis suggested that SASH1 is associated with exosome-related signaling pathways (as indicated by GO and KEGG enrichment, Figure 2C–D), leading us to hypothesize that P-EXOS may exert a protective effect in PE by regulating SASH1. Prior to functional studies, isolated P-EXOS were first characterized. Western blot analysis confirmed the presence of exosome marker proteins PLAP, CD63, and TSG101, with no detectable expression of the Golgi marker GM130, indicating high purity of the isolated vesicles. Transmission electron microscopy further revealed typical cup-shaped vesicular structures (Figure 6A–B). To verify whether P-EXOS could be efficiently internalized by target cells, immunofluorescence staining was performed to assess the cellular uptake of P-EXOS. Compared with the control group, HTR-8/SVneo cells treated with P-EXOS exhibited prominent co-localization signals in the merged images, indicating that P-EXOS were effectively internalized by HTR-8/SVneo cells (Figure 6C).

Subsequently, HTR-8/SVneo cells were co-cultured with P-EXOS under hypoxic conditions. Both quantitative PCR and Western blot assays showed a pronounced reduction in SASH1 mRNA and protein levels in the P-EXOS-treated group compared with controls (Figure 6D–E), providing initial evidence for a regulatory interaction between P-EXOS and SASH1. Consistent with these molecular changes, TUNEL staining revealed a substantial attenuation of hypoxia-induced apoptosis in HTR-8/SVneo cells following P-EXOS treatment (Figure 6F). At the same time, ELISA assays demonstrated markedly decreased secretion of pro-inflammatory cytokines, including IL-6, IL-1β, and TNF-α (Figure 6G).

To determine whether the protective effects of P-EXOS are mechanistically dependent on SASH1 suppression, a rescue experiment was performed under hypoxic conditions. Cells overexpressing SASH1 (OE-SASH1 + P-EXOS) were compared with negative control cells (OE-NC + P-EXOS). SASH1 overexpression significantly abolished the protective effects of P-EXOS, as evidenced by increased apoptosis (Figure 6H) and elevated pro-inflammatory cytokine secretion (Figure 6I) compared with the OE-NC + P-EXOS group. These results demonstrate that the protective effect of P-EXOS is mediated specifically through the suppression of SASH1, rather than through parallel pathways, underscoring the therapeutic potential of P-EXOS in PE.

DATA AVAILABILITY:

The datasets supporting the findings of this study are publicly available and were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), with accession numbers GSE75010, GSE10588, GSE60438, and GSE183338. Cell type annotation was performed with reference to the Cell Classification Database (https://ngdc.cncb.ac.cn/celltaxonomy/). Uncropped Western blot images and all other raw experimental data supporting the findings of this study are provided in the Supplementary Raw Data folder.

Figure 1: Identification of DEGs in PE placental and decidual tissues across three independent datasets. (A) Volcano plot of DEGs between 157 PE placental samples and 173 non-PE placental samples from the combined GSE75010 dataset. (B) Volcano plot of DEGs between severe PE placental samples (n = 17) and normal placental samples (n = 26) from the GSE10588 dataset. (C) Volcano plot of DEGs between PE decidual samples and normotensive control decidual samples from the GSE60438 dataset. DEGs were identified using p < 0.05 and |log₂FC| > 0.5. Please click here to view a larger version of this figure.

Figure 2: Functional enrichment and PPI analysis of common DEGs associated with pre-eclampsia. (A) Venn diagram of DEGs identified from PE-related datasets GSE75010, GSE10588, and GSE60438. (B) PPI network of the top 20 hub genes selected from the 56 common DEGs based on degree of connectivity, ranked using Cytoscape software (STRING interaction score ≥0.15). (C) GO and KEGG enrichment analyses of the common DEGs are presented as bar plots. (D) GO and KEGG enrichment analyses of the common DEGs, presented as chord diagrams. Please click here to view a larger version of this figure.

Figure 3: Machine learning-based screening of candidate hub genes associated with pre-eclampsia. (A) LASSO regression analysis showing coefficient profiles of candidate genes. (B) Selection of the optimal regularization parameter (λ) by LASSO regression using cross-validation. (C) Boruta algorithm. (D) SVM-RFE analysis. (E) XGBoost algorithm. (F) Venn diagram showing the intersection of candidate genes identified by LASSO, Boruta, SVM-RFE, and XGBoost algorithms. Abbreviations: LASSO = least absolute shrinkage and selection operator; SVM-RFE = support vector machine recursive feature elimination. Please click here to view a larger version of this figure.

Figure 4: Single-cell transcriptomic analysis reveals cell type-specific SASH1 expression patterns in pre-eclampsia. (A) UMAP clustering of the single-cell dataset. (B) Top five marker annotation analysis of the PE single-cell transcriptome dataset GSE183338. (C) Annotation of cell subpopulations. (D) Bar plot showing the distribution of cell proportions in control and PE samples. (E) UMAP plot colored by tissue origin (decidua and villi) and disease status (control and PE). (F) UMAP plot showing SASH1 expression in placental cell populations grouped by tissue origin (decidua and villi). (G) UMAP plot showing SASH1 expression in placental cell populations grouped by disease status (control and PE). (H) Violin plot showing the expression distribution of SASH1 across different placental cell types. ns p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001. Abbreviations: UMAP = uniform manifold approximation and projection; ns = not significant. Please click here to view a larger version of this figure.

Figure 5: SASH1 regulates hypoxia-induced trophoblast apoptosis and inflammation. (A) qPCR analysis of SASH1 mRNA expression and densitometric quantification of SASH1 protein levels in HTR-8/SVneo cells under normal and hypoxic conditions. (B) Representative Western blot images corresponding to the densitometric data shown in (A). (C) qPCR analysis of SASH1 mRNA expression and densitometric quantification of SASH1 protein levels in hypoxia-treated HTR-8/SVneo cells transfected with sh-NC or sh-SASH1 constructs (sh-SASH1#1, sh-SASH1#2, and sh-SASH1#3). (D) Representative Western blot images corresponding to the densitometric data shown in (C). (E) TUNEL staining of HTR-8/SVneo cells in the Normal, Hypoxia + sh-NC, and Hypoxia + sh-SASH1(#3) groups. Scale bar = 20 µm. (F) ELISA analysis of inflammatory cytokines (IL-1β, IL-6, and TNF-α) in the Normal, Hypoxia + sh-NC, and Hypoxia + sh-SASH1(#3) groups. (G) qPCR analysis of SASH1 mRNA expression and densitometric quantification of SASH1 protein levels in HTR-8/SVneo cells transfected with OE-NC or OE-SASH1 constructs. (H) Representative Western blot images corresponding to the densitometric data shown in (G). (I) TUNEL staining of HTR-8/SVneo cells in the Normal + OE-NC and Normal + OE-SASH1 groups. Scale bar = 20 µm. (J) ELISA analysis of inflammatory cytokines (IL-1β, IL-6, and TNF-α) in the Normal + OE-NC and Normal + OE-SASH1 groups. *p < 0.05; **p < 0.01; ***p < 0.001. Data are represented as mean ± SD from three independent biological replicates (n = 3). Please click here to view a larger version of this figure.

Figure 6: P-EXOS alleviates hypoxia-induced trophoblast injury via suppression of SASH1. (A) Western blot analysis of P-EXOS marker proteins PLAP, CD63, and TSG101, with GM130 serving as a negative control. (B) Transmission electron microscopy (TEM) analysis of P-EXOS morphology. Scale bar = 1.0 µm. (C) Confocal laser scanning microscopy images showing uptake of PKH67-labeled P-EXOS (green) in HTR-8/SVneo cells (red). Nuclei were counterstained with DAPI (blue). Scale bar = 20 µm. (D) qPCR analysis of SASH1 mRNA expression and densitometric quantification of SASH1 protein levels in HTR-8/SVneo cells in the Control and P-EXOS groups. (E) Representative Western blot images corresponding to the densitometric data shown in (D). (F) TUNEL staining of HTR-8/SVneo cells in the Control and P-EXOS groups. Scale bar = 20 µm. (G) ELISA analysis of inflammatory cytokines (IL-1β, IL-6, and TNF-α) in hypoxia-treated HTR-8/SVneo cells in the Control and P-EXOS groups. (H) TUNEL staining of HTR-8/SVneo cells in the OE-NC + P-EXOS and OE-SASH1 + P-EXOS groups. Scale bar = 20 µm. (I) ELISA analysis of inflammatory cytokines (IL-1β, IL-6, and TNF-α) in hypoxia-treated HTR-8/SVneo cells in the OE-NC + P-EXOS and OE-SASH1 + P-EXOS groups. Abbreviations: P-EXOS = placenta-derived exosomes; TEM = transmission electron microscopy; PLAP = placental alkaline phosphatase; TSG101 = tumor susceptibility gene 101. *p < 0.05; **p < 0.01. Data are represented as mean ±SD from three independent biological replicates (n = 3). Please click here to view a larger version of this figure.

Supplementary Figure 1: Quality control of single-cell RNA-seq data. (A) Violin plot of gene expression levels after quality control. (B) Variable feature plot showing highly variable genes identified after quality control.Please click here to download this file.

Supplementary Figure 2: Determination of statistically significant principal components for single-cell clustering. (A) JackStraw dot plot used to identify statistically significant principal components. (B) Elbow dot plot showing the variance explained by principal components.Please click here to download this file.

Supplementary Table 1: shRNA target sequences used for SASH1 knockdown. shRNA target sequences (5′–3′) for the negative control (sh-NC) and three SASH1-targeting constructs (sh-SASH1#1, sh-SASH1#2, and sh-SASH1#3) used for gene silencing.Please click here to download this file.

Supplementary Table 2: Primer sequences used for quantitative PCR. Forward and reverse primer sequences (5′–3′) for SASH1 and β-actin (internal reference control) used in qPCR analysis.Please click here to download this file.

Discussion

In this study, SASH1 was identified as a key gene dysregulated in PE through integrated transcriptomic analysis, single-cell sequencing, and multi-algorithm machine learning, an approach that is particularly relevant given that targeted therapies for PE remain unavailable¹⁷. It was demonstrated that SASH1 is significantly upregulated in hypoxic trophoblast cells and specifically elevated in the trophoblast compartment of PE placentas at single-cell resolution. Through loss- and gain-of-function experiments, it was established that SASH1 is both necessary and sufficient to drive trophoblast apoptosis and pro-inflammatory cytokine secretion. Furthermore, P-EXOS treatment attenuated hypoxia-induced trophoblast injury, and this protective effect was abolished by SASH1 overexpression, confirming that P-EXOS-mediated cytoprotection is mechanistically dependent on SASH1 suppression.

In addition, alterations in the proportions of multiple placental cell types have been observed in PE, including B cells/plasma cells, endothelial cells, fibroblasts/stromal cells, macrophages/dendritic cells, NK cells, T cells, and trophoblast cells. These findings suggest that the placental microenvironment undergoes substantial remodeling during PE. Such an inflammatory microenvironment not only exacerbates trophoblast dysfunction but may also accelerate disease progression through immune-extracellular matrix interactions¹⁸. Moreover, neutrophil extracellular traps (NETs) have been recently implicated in exacerbating placental inflammation in PE¹⁹. SASH1 is believed to play a significant role in PE development through its involvement in inflammatory signaling, cell adhesion, and hypoxia-related processes. In this study, gene expression profiles were extracted from placental samples of PE patients and normotensive pregnancies using the limma package, and multiple machine learning algorithms—LASSO regression, SVM-RFE, XGBoost, and Boruta for feature selection. Across all algorithms, SASH1 consistently emerged as a consensus candidate gene. At the single-cell level, SASH1 showed cell-type-specific expression across diverse placental cell populations and maintained elevated expression in PE samples, with the most pronounced difference observed in trophoblast cells, suggesting that it may exert distinct functional effects in different placental cell contexts.

SASH1 is a signaling adaptor protein that has been reported to participate in various biological processes, including cell migration and invasion, epithelial-mesenchymal transition, and stress-responsive signaling regulation^8,9. To investigate the functional role of SASH1 in PE-related pathology, a hypoxia-induced trophoblast model was established using HTR-8/SVneo cells. Our results demonstrated that hypoxic exposure significantly upregulated SASH1 expression, whereas SASH1 knockdown markedly attenuated hypoxia-induced apoptosis and reduced the secretion of pro-inflammatory cytokines, including IL-1β, IL-6, and TNF-α. Conversely, SASH1 overexpression under normoxic conditions was sufficient to recapitulate the hypoxia-induced phenotype, establishing the causal role of SASH1 in driving trophoblast injury. These findings suggest that SASH1 may negatively regulate trophoblast functional status by modulating inflammatory responses. Given that trophoblasts represent a major source of placental inflammatory mediators²⁰, aberrant SASH1 expression may constitute a critical molecular basis for inflammation amplification in PE. Notably, the pro-apoptotic and pro-inflammatory functions of SASH1 observed in trophoblast cells contrast with its canonical tumor-suppressive role in cancer cells, where it typically inhibits EMT and migration^8,9. This context-dependent functional duality may reflect cell type-specific differences in SASH1-interacting partners and downstream signaling networks, and warrants further investigation.

Exosomes are recognized as important mediators of intercellular communication and have been implicated in the pathogenesis of PE²¹. In particular, P-EXOS may play an important role in immune regulation during normal pregnancy, such as maintaining maternal-fetal immune tolerance, which may confer protection against PE²². In this study, bioinformatic analysis revealed an association between SASH1 and exosome-related signaling pathways, prompting further investigation into the role of P-EXOS in regulating trophoblast function. Co-culture experiments demonstrated that P-EXOS treatment significantly decreased SASH1 mRNA and protein levels in hypoxia-treated trophoblast cells, and markedly attenuated hypoxia-induced apoptosis and inflammatory responses. Importantly, SASH1 overexpression abolished these protective effects, confirming that the P-EXOS-mediated cytoprotection is mechanistically dependent on SASH1 suppression rather than acting through parallel pathways.

The molecular mechanisms by which P-EXOS suppresses SASH1 expression remain to be elucidated. P-EXOS are known to carry diverse bioactive cargos, including microRNAs, proteins, and lipids, that can modulate gene expression in recipient cells¹⁰. One possibility is that P-EXOS delivers specific miRNAs that directly or indirectly target SASH1. Alternatively, P-EXOS may modulate upstream signaling pathways—such as NF-κB or PI3K-Akt—that converge on SASH1 transcriptional regulation. Given that SASH1 has been implicated in the PI3K-Akt-mTOR pathway in cancer cells⁸, it is plausible that exosome-mediated modulation of this axis may contribute to SASH1 downregulation in trophoblast cells. Future studies employing exosomal cargo sequencing and pathway-specific inhibitors will be needed to dissect the precise molecular cascade through which P-EXOS regulates SASH1 expression.

In this study, transcriptomic profiling was integrated with single-cell RNA sequencing to investigate the molecular changes in placental tissue in PE. It was demonstrate that SASH1 plays a key role in hypoxia-induced trophoblast apoptosis and inflammatory responses, and that P-EXOS can exert a protective effect by suppressing SASH1 expression, offering new insights into the molecular mechanisms underlying PE and suggesting a potential therapeutic strategy. However, this study has several limitations. First, the functional findings were primarily derived from the HTR-8/SVneo immortalized trophoblast cell line; validation in primary trophoblast cells is warranted. Second, SASH1 protein expression was not validated in clinical PE placental tissues by immunohistochemistry, which limits the translational interpretation of the findings and should be addressed in future studies. Third, the SVM-RFE classifier showed moderate cross-validation accuracy (~0.6), reflecting the inherent heterogeneity of the dataset; larger independent cohorts will be required to develop more robust diagnostic models. Fourth, the precise molecular mechanism by which P-EXOS suppresses SASH1 expression remains unclear. Fifth, non-specific staining bands were observed in the SASH1 Western blot detection, which may be due to suboptimal antibody specificity. The antibody used in this experiment is a polyclonal antibody, which generally exhibits lower specificity in immunodetection compared with monoclonal antibodies. Nonetheless, these non-specific bands do not affect the SASH1 protein findings or the conclusions of this study. Future studies incorporating in vivo experiments, clinical sample analysis, and exosomal cargo characterization are needed to further elucidate the functional role of the P-EXOS-SASH1 axis in PE and to evaluate its translational potential.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
RPMI-1640 Medium	Gibco	11875085
Fetal Bovine Serum (FBS)	Gibco	16140071
Exosome-depleted FBS	System Biosciences	EXO-FBS-50A-1
Penicillin/Streptomycin	Thermo Fisher Scientific, CN	15140122
Lipofectamine 3000 Transfection Reagent (with P3000 Reagent)	Thermo Fisher Scientific, CN	L3000150
Opti-MEM Reduced Serum Medium	Thermo Fisher Scientific, CN	31985070
SASH1 shRNA plasmid (HuSH)	origene	TL301836
SASH1 ORF expression plasmid	origene	RC218866
Empty vector control	origene	PS100001
Reverse Transcription Kit	Thermo Fisher Scientific, CN	4368814
PowerUp SYBR Green Master Mix	Thermo Fisher Scientific, CN	A46110
SASH1 primers	Sangon Biotech	Custom synthesized
β-actin primers	Sangon Biotech	Custom synthesized
Cell Lysis Buffer	Beyotime	P0013B
NanoDrop Spectrophotometer	Thermo Fisher Scientific, CN	NanoDrop Ultra
PVDF Membrane	Thermo Fisher Scientific, CN	88518
Anti-SASH1 Antibody	ABclonal	A15248
Anti-β-actin Antibody	Abcam	ab6276
HRP-conjugated Goat Anti-Rabbit/Anti-Mouse Secondary Antibody	Beyotime	A0208
ImageJ Software	NIH (public domain)	ImageJ
PKH67 Green Fluorescent Cell Linker Kit (incl. Diluent C)	Sigma-Aldrich, St. Louis, MO, CN	MINI67
Bovine Serum Albumin (BSA)	Sigma-Aldrich, St. Louis, MO, CN	10711454001
DAPI	Beyotime	C1002
Confocal Laser Scanning Microscope (CLSM)	Leica	Leica TCS SP8
Human IL-6 Quantikine ELISA Kit	R&D Systems	D6050B
Human IL-1β/IL-1F2 DuoSet ELISA	R&D Systems	DY201
Human TNF-α Quantikine ELISA Kit	R&D Systems	DTA00D
Microplate Reader	Thermo Scientific	Multiskan FC
In Situ Cell Death Detection Kit (TUNEL Assay Kit)	Roche	11684795910
Triton X-100	Beyotime	P0096
Paraformaldehyde (4%)	Beyotime	P0099
Fluorescence Microscope	Olympus	BX53
Refrigerated Centrifuge (12,000 × g capacity)	Eppendorf	Eppendorf 5430 R
Ultracentrifuge (120,000 × g capacity)	Beckman Coulter	Optima MAX-XP
R Software	The R Foundation	v4.2.1
GraphPad Prism	GraphPad Software	v10.0
Seurat (R package)	Satija Lab (open-source)	v4.0.4
Harmony (R package, batch correction)	Open-source	v1.2.0
Scrublet (Python package, doublet detection)	Open-source	v0.2.3
clustree (R package)	Open-source	v0.5.1
glmnet (R package, LASSO)	Open-source	v4.1.8
e1071 (R package, SVM-RFE)	Open-source	v1.7-14
xgboost (R package)	Open-source	v1.7.7.1
randomForest (R package, Boruta)	Open-source	v4.7-1.1
Gene Expression Omnibus (GEO) Database	NCBI	https://www.ncbi.nlm.nih.gov/geo/
STRING Database	string-db.org	https://www.string-db.org/
Metascape	metascape.org	https://metascape.org/
Cytoscape Software	Cytoscape Consortium	v3.10.4
ggplot2 (R package)	Open-source	v3.5.1
limma (R package)	Open-source	v3.58.1
pheatmap (R package)	Open-source	v1.0.12
ClusterProfiler (R package)	Open-source	v4.8.3
DOSE (R package)	Open-source	v3.28.2
ggpubr (R package)	Open-source	v0.6.0