The role of spermatozoa-zona pellucida interaction in the selection of fertilization-competent spermatozoa in humans

Abstract

Spermatozoa-zona pellucida (ZP) binding interaction initiates a series of highly coordinated events leading to fertilization. Sperm selection in vivo is a dynamic and continuous process. Of the millions of ejaculated spermatozoa, only thousands of them can survive the journey and reach the oviduct. Human ZP selectively interacts with high-quality spermatozoa characterized by intact DNA, normal morphology and high fertilization potential. Intracytoplasmic sperm injection (ICSI) is universally used as an infertility treatment. Although ICSI can drastically improve the fertilization outcome, attention has been drawn to the concerns about ICSI bypassing the natural selection mechanisms, which could introduce a defective spermatozoon into the oocyte resulting in suboptimal fertilization outcome. The current sperm selection method for ICSI lacks the discriminatory power to select a single spermatozoon in terms of fertilization potential and genetic quality. We therefore hypothesize that the spermatozoa-ZP interaction plays a functional role in selecting fertilization-competent spermatozoa. The first objective was to establish the spermatozoa-ZP coincubation assays to isolate ZP-bound and -unbound spermatozoa under optimal conditions. Additional tests were performed to confirm that the experimental protocols had no detrimental effects on sperm parameters in terms of viability and motility. Time course data revealed that 2 hr was the optimal time-point to recover the maximum number of ZP-bound spermatozoa. The second objective was to examine the relationship between the ZPbinding ability and the genetic quality of spermatozoa. ZP-bound and -unbound spermatozoa were collected by the coincubation assays according to their ZPbinding ability. The ZP-binding ability of spermatozoa was closely related to their genetic quality in terms of DNA integrity, chromatin structure, protamination degree and global methylation level. Human ZP was highly selective for genetically normal spermatozoa with distinct epigenetic profiles. The third objective was to examine the relationship between the ZP- binding ability of spermatozoa and the expression levels of protein markers. Heat shock 70 kDa protein 2 (HSPA2) and sperm acrosome associated 3 (SPACA 3) were identified as the protein markers associated with the ZP-binding ability. Immunofluorescence staining revealed quantitative differences in these two markers between the ZP-bound (recovered at 30 min) and the -unbound spermatozoa. Dual immunofluorescence staining with CD46 indicated that HSPA2 and SPACA3 were detected on both acrosome-reacted and acrosome-intact spermatozoa that had bound to the ZP. The forth objective was to establish an hybrid artificial intelligence (AI) model for morphologic prediction of spermatozoa based on their ZP-binding ability. The hybrid AI model was established upon two artificial neural networks (ANNs), convolutional neural network (CNN) and sparse neural network (SNN). The combined use of CNN and SNN demonstrated success in classifying 1) ZP-bound and -unbound spermatozoa and 2) clinical samples of spermatozoa with defective and normal ZP-binding ability. The model achieved a high overall accuracy and a low cost in all sample groups. In conclusion, this thesis demonstrated selective binding of human ZP to spermatozoa in terms of fertilization potential and genetic quality. Our results validated the feasibility of incorporating ZP-binding ability to establish a novel method to identify high-quality spermatozoa by image analysis of sperm morphology.