Beyond black-box models: explainable AI for embryo ploidy prediction and patient-centric consultation.

PURPOSE: To determine if an explainable artificial intelligence (XAI) model enhances the accuracy and transparency of predicting embryo ploidy status based on embryonic characteristics and clinical data. METHODS: This retrospective study utilized a dataset of 1908 blastocyst embryos. The dataset includes ploidy status, morphokinetic features, morphology grades, and 11 clinical variables. Six machine learning (ML) models including Random Forest (RF), Linear Discriminant Analysis (LDA), Logistic Regression (LR), Support Vector Machine (SVM), AdaBoost (ADA), and Light Gradient-Boosting Machine (LGBM) were trained to predict ploidy status probabilities across three distinct datasets: high-grade embryos (HGE, n = 1107), low-grade embryos (LGE, n = 364), and all-grade embryos (AGE, n = 1471). The model's performance was interpreted using XAI, including SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) techniques. RESULTS: The mean maternal age was 38.5 ± 3.85 years. The Random Forest (RF) model exhibited superior performance compared to the other five ML models, achieving an accuracy of 0.749 and an AUC of 0.808 for AGE. In the external test set, the RF model achieved an accuracy of 0.714 and an AUC of 0.750 (95% CI, 0.702-0.796). SHAP's feature impact analysis highlighted that maternal age, paternal age, time to blastocyst (tB), and day 5 morphology grade significantly impacted the predictive model. In addition, LIME offered specific case-ploidy prediction probabilities, revealing the model's assigned values for each variable within a finite range. CONCLUSION: The model highlights the potential of using XAI algorithms to enhance ploidy prediction, optimize embryo selection as patient-centric consultation, and provides reliability and transparent insights into the decision-making process.

Abnormal mitochondrial function and impaired granulosa cell differentiation in androgen receptor knockout mice.

In the ovary, the paracrine interactions between the oocyte and surrounded granulosa cells are critical for optimal oocyte quality and embryonic development. Mice lacking the androgen receptor (AR⁻/⁻) were noted to have reduced fertility with abnormal ovarian function that might involve the promotion of preantral follicle growth and prevention of follicular atresia. However, the detailed mechanism of how AR in granulosa cells exerts its effects on oocyte quality is poorly understood. Comparing in vitro maturation rate of oocytes, we found oocytes collected from AR⁻/⁻ mice have a significantly poor maturating rate with 60% reached metaphase II and 30% remained in germinal vesicle breakdown stage, whereas 95% of wild-type AR (AR⁺/⁺) oocytes had reached metaphase II. Interestingly, we found these AR⁻/⁻ female mice also had an increased frequency of morphological alterations in the mitochondria of granulosa cells with reduced ATP generation (0.18 ± 0.02 vs. 0.29 ± 0.02 µM/mg protein; p < 0.05) and aberrant mitochondrial biogenesis. Mechanism dissection found loss of AR led to a significant decrease in the expression of peroxisome proliferator-activated receptor γ (PPARγ) co-activator 1-ß (PGC1-ß) and its sequential downstream genes, nuclear respiratory factor 1 (NRF1) and mitochondrial transcription factor A (TFAM), in controlling mitochondrial biogenesis. These results indicate that AR may contribute to maintain oocyte quality and fertility via controlling the signals of PGC1-ß-mediated mitochondrial biogenesis in granulosa cells.

Androgen receptor (AR) physiological roles in male and female reproductive systems: lessons learned from AR-knockout mice lacking AR in selective cells.

Androgens/androgen receptor (AR) signaling is involved primarily in the development of male-specific phenotypes during embryogenesis, spermatogenesis, sexual behavior, and fertility during adult life. However, this signaling has also been shown to play an important role in development of female reproductive organs and their functions, such as ovarian folliculogenesis, embryonic implantation, and uterine and breast development. The establishment of the testicular feminization (Tfm) mouse model exploiting the X-linked Tfm mutation in mice has been a good in vivo tool for studying the human complete androgen insensitivity syndrome, but this mouse may not be the perfect in vivo model. Mouse models with various cell-specific AR knockout (ARKO) might allow us to study AR roles in individual types of cells in these male and female reproductive systems, although discrepancies are found in results between labs, probably due to using various Cre mice and/or knocking out AR in different AR domains. Nevertheless, no doubt exists that the continuous development of these ARKO mouse models and careful studies will provide information useful for understanding AR roles in reproductive systems of humans and may help us to develop more effective and more specific therapeutic approaches for reproductive system-related diseases.

Premature aging with impaired oxidative stress defense in mice lacking TR4.

Early studies suggest that TR4 nuclear receptor is a key transcriptional factor regulating various biological activities, including reproduction, cerebella development, and metabolism. Here we report that mice lacking TR4 (TR4(-/-)) exhibited increasing genome instability and defective oxidative stress defense, which are associated with premature aging phenotypes. At the cellular level, we observed rapid cellular growth arrest and less resistance to oxidative stress and DNA damage in TR4(-/-) mouse embryonic fibroblasts (MEFs) in vitro. Restoring TR4 or supplying the antioxidant N-acetyl-l-cysteine (NAC) to TR4(-/-) MEFs reduced the DNA damage and slowed down cellular growth arrest. Focused qPCR array revealed alteration of gene profiles in the DNA damage response (DDR) and anti-reactive oxygen species (ROS) pathways in TR4(-/-) MEFs, which further supports the hypothesis that the premature aging in TR4(-/-) mice might stem from oxidative DNA damage caused by increased oxidative stress or compromised genome integrity. Together, our finding identifies a novel role of TR4 in mediating the interplay between oxidative stress defense and aging.

Abnormal mammary gland development and growth retardation in female mice and MCF7 breast cancer cells lacking androgen receptor.

Phenotype analysis of female mice lacking androgen receptor (AR) deficient (AR-/-) indicates that the development of mammary glands is retarded with reduced ductal branching in the prepubertal stages, and fewer Cap cells in the terminal end buds, as well as decreased lobuloalveolar development in adult females, and fewer milk-producing alveoli in the lactating glands. The defective development of AR-/- mammary glands involves the defects of insulin-like growth factor I-insulin-like growth factor I receptor and mitogen-activated protein kinase (MAPK) signals as well as estrogen receptor (ER) activity. Similar growth retardation and defects in growth factor-mediated Ras/Raf/MAPK cascade and ER signaling are also found in AR-/- MCF7 breast cancer cells. The restoration assays show that AR NH2-terminal/DNA-binding domain, but not the ligand-binding domain, is essential for normal MAPK function in MCF7 cells, and an AR mutant (R608K), found in male breast cancer, is associated with the excessive activation of MAPK. Together, our data provide the first in vivo evidence showing that AR-mediated MAPK and ER activation may play important roles for mammary gland development and MCF7 breast cancer cell proliferation.

Subfertility with defective folliculogenesis in female mice lacking testicular orphan nuclear receptor 4.

Testicular orphan nuclear receptor 4 (TR4) plays essential roles for normal spermatogenesis in male mice. However, its roles in female fertility and ovarian function remain largely unknown. Here we found female mice lacking TR4 (TR4-/-) displayed subfertility and irregular estrous cycles. TR4-/- female mice ovaries were smaller with fewer or no preovulatory follicles and corpora lutea. After superovulation, TR4-/- female mice produced fewer oocytes, preovulatory follicles, and corpora lutea. In addition, more intensive granulosa apoptosis was found in TR4-/- ovaries. Functional analyses suggest that subfertility in TR4-/- female mice can be due to an ovarian defect with impaired folliculogenesis rather than a deficiency in pituitary gonadotropins. Molecular mechanism dissection of defective folliculogenesis found TR4 might induce LH receptor (LHR) gene expression via direct binding to its 5' promoter. The consequence of reduced LHR expression in TR4-/- female mice might then result in reduced gonadal sex hormones via reduced expression of enzymes involved in steroidogenesis. Together, our results showed TR4 might play essential roles in normal folliculogenesis by influencing LHR signals. Modulation of TR4 expression and/or activation via its upstream signals or unidentified ligand(s) might allow us to develop small molecule(s) to control folliculogenesis.

Androgen receptor in sertoli cell is essential for germ cell nursery and junctional complex formation in mouse testes.

To examine the role of androgen receptor (AR) in Sertoli cells (SC), we used a SC-specific AR knockout (S-AR-/y) mouse to further evaluate the chronological changes of seminiferous tubules and the molecular mechanisms of SC androgen/AR signals on spermatogenesis. Testes morphology began changing as early as postnatal day (PD) 10.5 in wild-type (WT), but not in S-AR-/y mice. After puberty (PD 50), the SC nuclei of WT testes migrated to the basal area of the seminiferous epithelium; however, in S-AR-/y testes, SC nuclei were disarranged and dislocated. Results from electron microscopy further showed an obvious duplication of basal lamina of the seminiferous epithelium in S-AR-/y testes at PD 50 compared with WT testes. Using quantitative RT-PCR analyses, the expression of SC gene profiles were compared in PD 10.5 testes. In S-AR-/y testes, the expression levels of 1) vimentin were significantly increased and laminin alpha5 was significantly decreased in PD 10.5, which contributed to functional defects in cytoskeletons and production of the basement membrane component of SC leading to cell morphology deterioration and thus affecting the integrity of seminiferous epithelium; 2) claudin-11, occludin, and gelsolin were significantly decreased in PD 10.5, which contributed to defects in intact junctional complex formation of SC leading to impairment of the integrity of the blood-testis barrier; 3) calcium channel, voltage-dependent, P/Q-type, alpha1A subunit; tissue-type plasminogen activator; transferrin; and epidermal fatty-acid-binding protein were significantly decreased in PD 10.5, which contributed to functional defects in production and/or secretion of specific proteases, transport proteins, and paracrine factors of SC, leading to impairment of its germ cells' nursery functions.

Insulin and leptin resistance with hyperleptinemia in mice lacking androgen receptor.

Epidemiological evidence suggests that sex differences exist in type 2 diabetes. Men seem to be more susceptible than women to the consequences of obesity and sedentary lifestyle, possibly because of differences in insulin sensitivity and regional body fat deposition. Thus, lacking androgen receptor (AR) in male individuals may promote insulin resistance. To determine whether lacking AR in male individuals contributes to in vivo insulin resistance, an AR knockout model (AR(-/y)) was used to study the correlation between AR and insulin resistance. Progressive reduced insulin sensitivity and impaired glucose tolerance were seen in AR(-/y) mice with advancing age. Aging AR(-/y) mice displayed accelerated weight gain, hyperinsulinemia, and hyperglycemia, and loss of AR contributes to increased triglyceride content in skeletal muscle and liver. Leptin is higher in serum of AR(-/y) mice. Treatment with exogenous leptin fails to stimulate weight loss in AR(-/y) mice in advanced age, suggesting leptin resistance in the AR(-/y/) mice. Exogenous dihydrotestosterone replacement fails to reverse the metabolic abnormalities and insulin resistance in AR(-/y) mice. Our in vivo studies demonstrate that androgen-AR plays key roles in the development of insulin and leptin resistance, which may contribute to the development of type 2 diabetes and cardiovascular disease.

Androgen receptor roles in spermatogenesis and fertility: lessons from testicular cell-specific androgen receptor knockout mice.

Androgens are critical steroid hormones that determine the expression of the male phenotype, including the outward development of secondary sex characteristics as well as the initiation and maintenance of spermatogenesis. Their actions are mediated by the androgen receptor (AR), a member of the nuclear receptor superfamily. AR functions as a ligand-dependent transcription factor, regulating expression of an array of androgen-responsive genes. Androgen and the AR play important roles in male spermatogenesis and fertility. The recent generation and characterization of male total and conditional AR knockout mice from different laboratories demonstrated the necessity of AR signaling for both external and internal male phenotype development. As expected, the male total AR knockout mice exhibited female-typical external appearance (including a vagina with a blind end and a clitoris-like phallus), the testis was located abdominally, and germ cell development was severely disrupted, which was similar to a human complete androgen insensitivity syndrome or testicular feminization mouse. However, the process of spermatogenesis is highly dependent on autocrine and paracrine communication among testicular cell types, and the disruption of AR throughout an experimental animal cannot answer the question about how AR in each type of testicular cell can play roles in the process of spermatogenesis. In this review, we provide new insights by comparing the results of cell-specific AR knockout in germ cells, peritubular myoid cells, Leydig cells, and Sertoli cells mouse models that were generated by different laboratories to see the consequent defects in spermatogenesis due to AR loss in different testicular cell types in spermatogenesis. Briefly, this review summarizes these results as follows: 1) the impact of lacking AR in Sertoli cells mainly affects Sertoli cell functions to support and nurture germ cells, leading to spermatogenesis arrest at the diplotene primary spermatocyte stage prior to the accomplishment of first meiotic division; 2) the impact of lacking AR in Leydig cells mainly affects steroidogenic functions leading to arrest of spermatogenesis at the round spermatid stage; 3) the impact of lacking AR in the smooth muscle cells and peritubular myoid cells in mice results in similar fertility despite decreased sperm output as compared to wild-type controls; and 4) the deletion of AR gene in mouse germ cells does not affect spermatogenesis and male fertility. This review tries to clarify the useful information regarding how androgen/AR functions in individual cells of the testis. The future studies of detailed molecular mechanisms in these in vivo animals with cell-specific AR knockout could possibly lead to useful insights for improvements in the treatment of male infertility, hypogonadism, and testicular dysgenesis syndrome, and in attempts to create safe as well as effective male contraceptive methods.

Differential effects of spermatogenesis and fertility in mice lacking androgen receptor in individual testis cells.

Using a Cre-Lox conditional knockout strategy, we generated a germ cell-specific androgen receptor (AR) knockout mouse (G-AR(-/y)) with normal spermatogenesis. Sperm count and motility in epididymis from AR(-/y) mice are similar to that of WT (G-AR(+/y)) mice. Furthermore, fertility tests show there was no difference in fertility, and almost 100% of female pups sired by G-AR(-/y) males younger than 15 weeks carried the deleted AR allele, suggesting the efficient AR knockout occurred in germ cells during meiosis. Together, these data provide in vivo evidence showing male mice without AR in germ cells can still have normal spermatogenesis and fertility, suggesting the essential roles of AR during spermatogenesis might come from indirect cell-cell communication in a paracrine fashion. We then compared the consequences of AR loss in the spermatogenesis and fertility of G-AR(-/y) mice with two other testicular cell-specific AR(-/y) mice and total AR knockout male mice. The results provide clear in vivo evidence that androgen/AR signaling in Sertoli cells plays a direct important role in spermatogenesis and in Leydig cells plays an autocrine regulatory role to modulate Leydig cell steroidogenic function. Total AR knockout male mice have the most severe defects among these mice. These contrasting data with G-AR(-/y) mice suggest AR might have different roles in the various cells within testis to contribute to normal spermatogenesis and male fertility in mice.

Oligozoospermia with normal fertility in male mice lacking the androgen receptor in testis peritubular myoid cells.

Androgens and the androgen receptor (AR) play important roles in the testes. Previously we have shown that male total AR knockout (T-AR-/y) mice revealed incomplete germ cell development and lowered serum testosterone levels, which resulted in azoospermia and infertility. However, the consequences of AR loss in particular types of testicular cells remain unclear. Using a Cre-loxP conditional knockout strategy, we generated a tissue-selective knockout mouse with the AR gene deleted in testis peritubular myoid cells (PM-AR-/y). Phenotype analyses showed that PM-AR-/y mice were indistinguishable from WT AR (AR+/y) mice with the exception of smaller testes size. PM-AR-/y mice have serum testosterone concentrations comparable with AR+/y mice. PM-AR-/y mice have oligozoospermia in the epididymis; however, fertility was normal. Although normal germ cell distribution ratio was found, total germ cell number decreased in PM-AR-/y mice. Further mechanistic studies demonstrated that PM-AR-/y mice have defects in the expression of Sertoli cells' functional marker genes such as tranferrin, epidermal fatty acid-binding protein, androgen-binding protein, and other junction genes including occludin, testin, nectin, zyxin, vinculin, laminingamma3, gelsolin, connection43, and N-cadherin. Furthermore, there were defects in peritubular myoid cell contractility-related genes such as endothelin-1, endothelin receptor A and B, adrenomedullin, adrenomedullin receptor, and vasopressin receptor 1a. Together, our PM-AR-/y mice provide in vivo evidence for the requirement of functional AR in peritubular myoid cells to maintain normal Sertoli cells function and peritubular myoid cell contractility, thus ensuring normal spermatogenesis and sperm output.

Subfertility and defective folliculogenesis in female mice lacking androgen receptor.

The roles of the androgen receptor (AR) in female fertility and ovarian function remain largely unknown. Here we report on the generation of female mice lacking AR (AR(-/-)) and the resulting influences on the reproductive system. Female AR(-/-) mice appear normal but show longer estrous cycles and reduced fertility. The ovaries from sexually mature AR(-/-) females exhibited a marked reduction in the number of corpora lutea. After superovulation treatment, the AR(-/-) ovaries produced fewer oocytes and also showed fewer corpora lutea. During the periovulatory period, an intensive granulosa apoptosis event occurs in the AR(-/-) preovulatory follicles, concurrent with the down-regulation of p21 and progesterone receptor expression. Furthermore, the defective conformation of the cumulus cell-oocyte complex from the AR(-/-) females implies a lower fertilization capability of the AR(-/-) oocytes. In addition to insufficient progesterone production, the diminished endometrial growth in uteri in response to exogenous gonadotropins indicates that AR(-/-) females exhibit a luteal phase defect. Taken together, these data provide in vivo evidence showing that AR plays an important role in female reproduction.

Infertility with defective spermatogenesis and hypotestosteronemia in male mice lacking the androgen receptor in Sertoli cells.

Androgens and the androgen receptor (AR) play important roles in male fertility, although the detailed mechanisms, particularly how androgen/AR influences spermatogenesis in particular cell types, remain unclear. Using a Cre-Lox conditional knockout strategy, we generated a tissue-specific knockout mouse with the AR gene deleted only in Sertoli cells (S-AR(-/y)). Phenotype analyses show the S-AR(-/y) mice were indistinguishable from WT AR mice (B6 AR(+/y)) with the exception of testes, which were significantly atrophied. S-AR(-/y) mice were infertile, with spermatogenic arrest predominately at the diplotene premeiotic stage and almost no sperm detected in the epididymides. S-AR(-/y) mice also have lower serum testosterone concentrations and higher serum leuteinizing hormone concentrations than B6 AR(+/y) mice. Further mechanistic studies demonstrated that S-AR(-/y) mice have defects in the expression of anti-Müllerian hormone, androgen-binding protein, cyclin A1, and sperm-1, which play important roles in the control of spermatogenesis and/or steroidogenesis. Together, our Sertoli cell-specific AR knockout mice provide in vivo evidence of the need for functional AR in Sertoli cells to maintain normal spermatogenesis and testosterone production, and ensure normal male fertility.

Global analysis of differentially expressed genes in early gestational decidua and chorionic villi using a 9600 human cDNA microarray.

The global gene expression profiles of the decidua and chorionic villi of early human pregnancies were analysed by using cDNA microarray technology. Decidual and villous placental tissues were obtained from first trimester abortus and mRNA was extracted for cDNA microarray analysis. The human cDNA microarray [9600 clones, including known regulatory genes and expressed sequence tags (EST)] with colorimetric detection was used to identify differentially expressed genes between early gestational decidua and villi. According to cDNA microarray analysis, we have identified 641 genes with highly expressed mRNA in both decidua and villi, 49 genes with higher expressions in decidua, and 75 genes with higher expression in chorionic villi. These differentially expressed genes were further grouped into categories by their putative functions, including: cell growth-related factors, hormones/cytokines, cell adhesion molecules, signal transduction molecules, apoptosis-related factors, cytoskeleton/extracellular matrix proteins, and EST. Immunohistochemical stainings of cathepsin L, leukaemia inhibitory factor-receptor, and proliferative cell nuclear antigen showed results consistent with the microarray data. Identification of the differentially expressed genes between decidua and villi by microarray provide a global profiling of the gene expression pattern. This work adds to our understanding of placentation by reporting the gene expression profiles during first trimester human pregnancies using cDNA microarray.