Morphological and Chemical Investigation of Ovarian Structures in a Bovine Model by Contrast-Enhanced X-ray Imaging and Microscopy

Abstract

An improved understanding of an ovary’s structures is highly desirable to support advances in folliculogenesis knowledge and reproductive medicine, with particular attention to fertility preservation options for prepubertal girls with malignant tumors. Although currently the golden standard for structural analysis is provided by combining histological sections, staining, and visible 2D microscopic inspection, synchrotron radiation phase-contrast microtomography is becoming a new challenge for three-dimensional studies at micrometric resolution. To this aim, the proper use of contrast agents can improve the visualization of internal structures in ovary tissues, which normally present a low radiopacity. In this study, we report a comparison of four staining protocols, based on iodine or tungsten containing agents, applied to bovine ovarian tissues fixed in Bouin’s solution. The microtomography (microCT) analyses at two synchrotron facilities under different set-ups were performed at different energies in order to maximize the image contrast. While tungsten-based agents allow large structures to be well identified, Iodine ones better highlight smaller features, especially when acquired above the K-edge energy of the specific metal. Further scans performed at lower energy where the setup was optimized for overall quality and sensitivity from phase-contrast still provided highly resolved visualization of follicular and intrafollicular structures at different maturation stages, independent of the staining protocol. The analyses were complemented by X-ray Fluorescence mapping on 2D sections, showing that the tungsten-based agent has a higher penetration in this type of tissues.

Difficulties and artefacts in cryo-fixation of ovarian tissues for X-ray fluorescence analyses

“…1, the tissue's morphology in paraffin-embedded samples is quite well preserved, showing an abundant presence of follicles at different stages of maturation. Stroma cells (white arrows) and oocyte structures, like the granulosa and theca cells of follicles (black arrows), are well visible and easily identified in the XRF elemental mapping, especially by following phosphorus (P) distribution, as already shown in previous works for both bovine 26 and human ovarian tissues. 24,25,33 The morphology highlighted by P (revealing cell nuclei) and S (maximally abundant in the extracellular matrix) distributions resemble, at lower spatial resolution, the standard histological images (Fig.…”

“…The ovarian tissue samples were obtained from young heifers at a slaughterhouse (macello Pelizzari, Loria, Treviso, Italy) accordingly to previously described procedures. 26 Samples consisted of strips (3 mm thick, 4–5 mm wide) from the cortical part of the ovaries, manually separated from the medullar portion. All samples were collected in ice-cold PBS and then processed in three ways: (a) some tissue strips were immediately chemically fixed in PFA 10%, washed and dehydrated in ethanol prior to being included in paraffin and sliced to 10 μm thickness to be deposited onto ultralene foils; (b) fresh samples were embedded in tissue-tek (Sakura Finetek USA) slowly cooled down to −80 °C in less than 10 minutes, then sliced to a thickness of 10–20 μm using a cryomicrotome, maintained at −80 °C and directly transferred to the cryo-stage; (c) some samples from (b) were freeze-dried in vacuum and sandwiched between two ultralene foils.…”

Advances in sparse dynamic scanning in spectromicroscopy through compressive sensing

“…Our ML CS strategy can be outlined as: for a given sample (e.g. Bovine Ovary tissue [21]) i) acquire a standard/full STXM and a coarse/fast XRF map ii) train our neural network on transmission (TX) data (including differential phase contrast), diagnostics, in association with the XRF data. For subsequent samples (still the same or very similar setup conditions) we use the resulting trained model to infer XRF-like data, by processing point-by-point the corresponding STXM data.…”

“…A demonstration of the ML approach described in paragraph 2.4 is reported in Fig 4, where an XRF-like map inferred by ML is compared with a real XRF one acquired on a 5 μm thick paraffined slice of bovine ovary tissue, obtained from a young heifer at a slaughterhouse (macello Pelizzari, Loria, Treviso, Italy). More details on the sample preparation are given in [21].…”