Prenatal diagnosis of congenital anomalies using genetic and imaging analyses struggles to accurately predict disease severity, hindering effective parental counseling and patient stratification for prenatal therapies. While level 1 evidence supports prenatal therapy for conditions like CDH, TTTS, and MMC, selecting appropriate candidates for other conditions, such as LUTO, remains challenging due to the lack of autologous models of developing human tissues. Current methods, such as using human embryonic stem cells or induced pluripotent stem (iPS) cells to derive organoids, involve extensive manipulation and lengthy differentiation protocols, reducing patient fidelity and limiting applicability for prenatal disease modeling and targeted therapy. Primary organoids derived from discarded postnatal samples offer advantages, but ethical and legal restrictions on accessing fetal tissues obtained postmortem hinder research. This necessitates the development of new patient-specific in vitro models that can be derived during pregnancy without requiring tissue samples or reprogramming. Amniotic fluid (AF), containing cells from multiple developing organs, presents a promising source for such a model. This study aims to characterize the cellular identities in human AF using single-cell analysis and to determine whether these cells can be used to derive tissue-specific primary fetal organoids suitable for prenatal disease modeling and therapeutic investigation.

The current literature highlights the limitations of existing prenatal diagnostic and therapeutic approaches due to the challenges in predicting disease severity and the lack of suitable patient-specific models. Studies demonstrating the benefits of prenatal interventions for certain conditions like CDH, TTTS, and MMC are contrasted with the need for improved patient selection criteria for others, such as LUTO. The use of iPS cell-derived organoids, while offering a potential model, suffers from limitations due to lengthy differentiation protocols and decreased patient fidelity. The ethical and logistical constraints associated with obtaining fetal tissue from postmortem samples also pose significant challenges to research efforts. Previous work on AF stem cells has focused primarily on mesenchymal and hematopoietic lineages, while the potential of AF epithelial cells has been less explored. This study bridges this gap by thoroughly investigating the heterogeneity of the AF epithelial cell population and assessing its suitability for generating primary fetal organoids.

Amniotic fluid (AF) samples were collected from 12 pregnancies (15-34 gestational age weeks) during standard clinical procedures (amniocenteses and amniodrainages) after obtaining written informed consent and ethical approval. Viable nucleated cells with intact cell membranes were isolated using fluorescence-activated cell sorting (FACS). 3' single-cell RNA sequencing (scRNA-seq) was performed to characterize the cellular identities present in the AF. Unsupervised UMAP analysis and SingleR annotation, confirmed by pan-epithelial marker gene expression, were used to identify the epithelial cell cluster. scGSEA was used to identify gastrointestinal, renal, and pulmonary signatures within this cluster. To derive organoids, viable AF cells were seeded in Matrigel droplets and cultured in a generic epithelial medium. Clonal lines were established by picking individual organoids, dissociating them into single cells, and replating. Bulk RNA sequencing and scRNA-seq were performed on the resulting organoids to determine their tissue identity. Similar methodology was applied to tracheal fluid (TF) samples obtained from fetuses with CDH during fetoscopic endoluminal tracheal occlusion (FETO). Organoid characterization involved various techniques, including immunostaining, X-ray phase-contrast computed tomography (PC-CT), microCT, and transmission electron microscopy (TEM). Functional assays such as dipeptidyl peptidase IV and disaccharidase assays were also performed on the small intestinal organoids. A kidney potassium ion channel assay and an epithelial barrier integrity assay using inulin-FITC were conducted on the kidney organoids. Ciliary beat frequency (CBF) analysis using high-speed video microscopy was done on the lung organoids. Organoid maturation assays were performed using tissue-specific culture media for each tissue type. Statistical analysis included t-tests, ANOVA, and other appropriate methods.

Single-cell analysis of AF revealed the presence of epithelial cells originating from various developing organs, including the gastrointestinal tract, kidneys, and lungs. These cells expressed tissue-specific progenitor markers. In culture, these cells formed clonal organoids that maintained their tissue identity over multiple passages. Small intestinal AFOs (SIAFOs) displayed crypt-like structures, cell proliferation at the crypt base, and expressed markers of various intestinal cell types. Maturation assays showed increased expression of goblet cell and enterocyte markers upon treatment with intestinal-specific medium. Functional assays confirmed the presence of digestive enzymes. Kidney AFOs (KAFOs) exhibited a tubuloid-like phenotype, expressing markers of different renal tubule segments, and displayed functional voltage-gated potassium channels and epithelial tight junctions. Lung AFOs (LAFOs) showed expression of pulmonary stem/progenitor cell markers and could be differentiated into proximal and distal airway lineages upon exposure to the respective media. Organoids derived from AF and TF of CDH fetuses exhibited some features of the disease, including altered expression of surfactant protein genes and an increased percentage of pulmonary neuroendocrine cells. Analysis of CDH organoids revealed upregulated pathways related to surfactant production and metabolism and downregulation of pathways related to laminin interaction and ECM.

This study demonstrates the feasibility of deriving primary fetal epithelial organoids from readily accessible fetal fluids (AF and TF) during ongoing pregnancies. The use of AF provides a non-invasive approach for obtaining tissue-specific progenitor cells, avoiding the ethical and logistical limitations associated with obtaining postmortem samples. The relatively short timeframe (4–6 weeks) required for organoid generation from fluid collection to characterization makes this approach highly relevant for prenatal intervention, counseling, and therapy. The organoids derived in this study accurately recapitulated the characteristics of the respective tissues of origin and exhibited functional capabilities, as demonstrated through various assays. The application of this method to CDH fetuses showed that the resulting organoids display certain disease features, making this technology a promising tool for personalized disease modeling and development of targeted therapies. The study expands the understanding of AF cellular composition by revealing the presence of a diverse epithelial cell population with multiple tissue origins. The unique ability to model late gestational stages, currently inaccessible through termination of pregnancy, opens up new avenues for prenatal research.

This research successfully demonstrates a minimally invasive method for deriving primary fetal epithelial organoids from amniotic and tracheal fluids during ongoing pregnancies. The resulting organoids, exhibiting tissue-specific characteristics and functional capabilities, offer a valuable tool for prenatal disease modeling and the development of personalized therapies. Future research could focus on expanding the range of tissues modeled, optimizing organoid derivation protocols, and validating the use of these organoids in larger cohorts of patients with various congenital conditions, to further explore the potential of this technology for prenatal diagnosis and treatment.

The study's limitations include the relatively small sample size, especially for specific tissue types like the small intestine. The focus on epithelial cells limits the ability to model complex conditions involving mesenchymal and vascular compartments. While the CDH model shows promise, a larger patient cohort is needed to fully validate the platform and refine its predictive capabilities. The timeframe for organoid generation, although suitable for prenatal interventions, may still be considered long for some clinical situations.