Titre:T owards a Molecular Understanding of the Sperm Protein CRISP2 in Humans.

Promoteur: Madame Elise HENNEBERT

Résumé de la dissertation

Cysteine-rich secretory proteins (CRISPs) are widely expressed across mammalian species, with variable numbers of paralogues between species. Interestingly, CRISPs show predominant expression in reproductive tissues, and several studies have implicated them in fertilization processes. However, most of the research on CRISPs has been conducted in rodent models, complicating extrapolation to humans due to significant interspecies differences. Among the three CRISPs described in humans, CRISP2 is of particular interest as, unlike CRISP1 and 3 that are added to the sperm surface during epididymal maturation, CRISP2 is of testicular origin and expressed ab initio in sperm.

The aim of this thesis was to contribute to the characterization of human CRISP2 (hCRISP2) through multiple complementary approaches, with a particular focus on investigating its receptor at the oolemma.

The first part of the thesis focused on the localization and molecular characterization of hCRISP2.  Its localization across human testicular and epididymal tissues, as well as ejaculated sperm was investigated in immunofluorescence. No expression was detected in the epididymal epithelium, but hCRISP2 was clearly present at multiple stages of spermatogenesis: it was found within the nucleus of primary spermatocytes, round spermatids, and early elongated spermatids. In more advanced stages, hCRISP2 was also detected in the cytoplasm, the flagellum, and the equatorial segment of the acrosome. The presence of aggregated material with hCRISP2 immunoreactivity in the apical pole of Sertoli cells suggested that most of the hCRISP2 involved in spermatogenesis is phagocytized by these cells during spermiation. In ejaculated sperm, hCRISP2 was consistently observed in the cytoplasmic droplet, the flagellum, and the equatorial segment, consistent with its proposed roles in sperm motility and gamete interaction. Native and denaturing electrophoresis combined with western blot analyses depicted the ability of hCRISP2 to form stable high molecular weight complexes, and mass spectrometry revealed that these complexes likely consist exclusively of hCRISP2, with only limited post-translational modifications.

The next part of the thesis aimed to assess the ability of hCRISP2 to bind to the oolemma. As a first approach, synthetic peptides corresponding to a twelve–amino acid region conserved across CRISP proteins were used. This region was originally identified in rat CRISP1 as being involved in oocyte binding. In the present study, peptides corresponding to the homologous sequence in both hCRISP1 and hCRISP2 were synthesized and tested on human oocytes, as mouse CRISP1 and CRISP2 have been shown to likely share the same oocyte receptor. These peptides produced clear oolemma staining, while control oocytes incubated with DMSO or with a peptide derived from a different region of hCRISP2 showed no staining. Interestingly, a control peptide composed of the same amino acid residues as the conserved motif, but arranged in a randomized sequence, also produced strong oolemma staining. This unexpected result raises the possibility that the observed binding may not be sequence-specific and could result from general physicochemical properties of the peptides rather than a specific receptor-mediated interaction. To clarify this, we decided to test the binding capacity of the full-length hCRISP2 protein, which may offer a more physiologically relevant context for assessing specific interactions with the oolemma. Recombinant protein was thus produced in both E. coli and mammalian Expi293F cells for subsequent binding experiments. However, due to low yields and insufficient purity, direct binding assays with recombinant proteins could not be performed.

Finaly, a third part of the thesis aimed to characterize the hCRISP2 interactome through two complementary strategies: one targeting extracellular binding partners on the oocyte surface, and the other identifying intracellular interactors relevant to sperm function. To explore potential oocyte receptors, a comprehensive list of candidate surface proteins was compiled using published proteomic datasets from four human and five mouse oocyte proteomes. First, gene names were mapped to their human orthologues, and corresponding protein sequences retrieved, yielding an initial pool of 8,135 candidates. These were then analyzed using DeepLoc2.1 to predict subcellular localization based on their primary amino acid sequence. Since the receptor must be accessible to sperm, only proteins predicted to localize to the plasma membrane or extracellular space were retained and compiled in a list of 655 proteins, representative of the oocyte surface proteomic landscape. To identify potential hCRISP2 interactors among these candidates, a functional screening strategy was developed using a Gateway-compatible plasmid library to express the candidates as GFP-tagged constructs in mammalian cells. So far, 192 coding sequences have been recovered, and 86 of them expressed in HeLa cells. These cells were incubated with the hCRISP-derived peptides to test for binding. Although no specific interactions have yet been observed, this pipeline represents a scalable and robust platform for receptor screening. Complementing this approach, AlphaFold Multimer was used to computationally predict structural interactions between hCRISP2 and the curated surface protein list. Sixty candidates were identified with a predicted interface confidence score (ipTM) ≥ 0.7, and the structural consistency of predicted interfaces across multiple candidates suggests non-random, biologically plausible interactions. Manual refinement is currently ongoing to establish a shortlist for experimental validation. In parallel, to investigate intracellular partners of hCRISP2, a yeast two-hybrid (Y2H) screen was conducted using full-length hCRISP2 as bait. After confirming the absence of toxicity or autoactivation, a screening against a normalized universal human cDNA library yielded 328 growing colonies. Functional classification of these hits revealed several main protein groups, some of which are implicated in biological processes consistent with known or hypothesized hCRISP2 functions, such as spermatogenesis, cytoskeletal organization, and ion homeostasis.