Silk-Ovarioids: establishment and characterization of a human ovarian primary cell 3D-model system.

Abstract

Study question: What is the best protocol to establish a long-term stable three-dimensional (3D) model for human primary ovarian cells?

Summary answer: We developed and characterized long-term cultured 3D models of primary ovarian somatic cells isolated from adult tissues, using Biosilk as a scaffold.

What is known already: In vitro models that mimic ovaries are crucial for elucidating the biological mechanisms underlying follicle activation and growth, hormonal activity, ovarian angiogenesis, damage in response to toxic exposures, and other biological mechanisms that enable the functionality of this complex organ. Three-dimensional systems are particularly relevant because they replicate heterogeneity and cell-cell communication among different ovarian cell types. However, complex models using human ovarian primary cells are yet to be developed.

Study design size duration: Ovarian tissue samples were collected from five patients (age 26 ± 5 years) who underwent gender-affirming surgery. The cortex and medulla were separated and dissociated into single-cell suspensions using mechanical and enzymatic methods. Three approaches were tested to establish a 3D model culture system: matrix-free ovarian spheroids (MFOS), a Matrigel-based three-layer gradient system (3LGS), and Biosilk scaffolds (Silk-Ovarioid). In parallel, paired controls from each patient and ovarian area were cultured in a standard 2D system for the same duration.

Participants/materials setting methods: The 3D culture systems were monitored every second day to detect signs of aggregation and growth. Freshly fixed tissue, as well as 2D- and 3D-cultured samples were further processed for transcriptomic profiling after 42 days of culture using RNA sequencing. The culture of the 3D system was further characterized, regarding its protein profile and steroid and cytokine production, through proteomics and liquid chromatography-tandem mass spectrometry and the Luminex platform, respectively. The key findings from the high-throughput assays were finally validated through RNA fluorescent in situ hybridization (RNA-FISH) and immunofluorescence staining.

Main results and the role of chance: The 3D model systems MFOS (n = 120) and 3LGS (n = 18) failed to form aggregates capable of long-term maintenance in culture (MFOS: maximum of 15 days for both cortex and medulla; 3LGS: maximum of 11 days for medulla only). In contrast, we successfully established ovarian cortex- and medulla-derived 3D systems using Biosilk, termed Silk-Ovarioids (n = 120). Silk-Ovarioids were maintained for up to 42 days as free-floating culture without any signs of cell death, as confirmed by the absence of TUNEL, γ-H2A.X, and cleaved caspase 3 fluorescent signals. The presence of key ovarian somatic cell types, including granulosa, stromal, endothelial, and perivascular cells, was confirmed by transcriptomics and proteomics in the majority of Silk-Ovarioids. Validation through RNA-FISH and immunostaining was performed using the following markers: AMHR2 for granulosa cells, PDGFRα for stromal cells, CLDN5 and GPIHBP1 for endothelial cells, GJA4/Cx37 and MCAM for perivascular cells. Notably, Silk-Ovarioids exhibited the formation of a pro-angiogenic hypoxic core, as evidenced by the transcriptomic and proteomic data and visualized by the expression of hypoxia markers MMP2 and PDGFRβ. This hypoxic environment led to development of vessel-like structures after 4-6 weeks of culture, which were positive for the angiogenic markers TGFBR2, BMP2, and PDGFα. The functionality of Silk-Ovarioids was further confirmed by the identification of de novo extracellular matrix secretion (Col1α1 and Lamα1), and by the detection of pro-angiogenic cytokines (e.g. IL-6, IL-8, and GM-CSF) and steroids (e.g. pregnenolone and epitestosterone) in the culture media.

Large scale data: The RNA-sequencing count matrix is deposited in Gene Expression Omnibus with accession number GSE253571. Raw data are deposited in Swedish National Data Service with the DOI https://doi.org/10.48723/h8cm-bs19. Single-cell RNA-seq data have been downloaded from the ArrayExpress database at EMBL-EBI with the accession codes 'E-MTAb - 8381'. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD048710. The code used for the analysis can be found in https://github.com/tialiv/Silk-Ovarioid_project.

Limitations reasons for caution: The ovarian samples were collected from patients undergoing androgen treatment, raising the concern that androgen exposure may alter the behavior of cells in Silk-Ovarioids compared to those derived from androgen-unstimulated patients. Furthermore, the cell culture media used in this study were supplemented with fetal bovine serum and did not contain any supplements or growth factors that could be essential for the resemblance of Silk-Ovarioids to the tissue of origin.

Wider implications of the findings: The Silk-Ovarioids exhibited low intra-batch variability and long-term culture stability, underscoring their potential as a robust step toward developing a bioengineered, patient-specific artificial ovary. In addition, Silk-Ovarioids could be utilized as the first ovarian angiogenesis in vitro model, function as biological scaffold for in vitro folliculogenesis, and be used for toxicological and pharmacological studies targeting the ovaries.

Study funding/competing interests: This study was funded by: a research grant from the Center for Innovative Medicine (CIMED) at Karolinska Insitutet; European Union's Horizon 2020 Research and Innovation Programme (project ERIN no. 952516); a Horizon Europe grant (NESTOR, grant no. 101120075) of the European Commission; the Swedish Research Council for Sustainable Development FORMAS (2018-02280, 2020-01621); StratRegen Funding from Karolinska Institute, Swedish Research Council VR (grant no. 2020-02132); Swedish Childhood Cancer Fund (Reference PR2017-0044, PR2020-0096); Estonian Research Council (grant no. PRG1076); Swedish Research Council (grant no. 2024-02530); Novo Nordisk Foundation (grant no. NNF24OC0092384); European Union's H2020 project Sinfonia (no. 857253) (INL research); and SbDToolBox, with reference NORTE-01-0145-FEDER-000047, supported by Norte Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through the European Regional Development Fund (INL research). The authors have no conflicts of interest to declare.