A greater health risk has been identified for children conceived by fathers over 50 years of age, including an increased adjusted odds ratio (OR) for developing the neurodevelopmental disorders autism spectrum disorder (OR: 2.26 to 3.37), schizophrenia (OR 1.59–4.62), and bipolar disorder (OR 1.27–2.84) [33]. While the paternal age effect appears to be a multifaceted phenomenon, the association with these multifactorial disorders is robust and reproducible. Paternal factors, such as epimutations arising in the sperm as men age, may occur at genes essential for embryonic development or implicated in offspring health conditions, providing a potential mechanism for these adverse outcomes. In particular, there have been studies demonstrating DNA methylation errors at genes important for neurodevelopment in sperm upon advanced paternal age [25, 31, 32, 34]. Yet, minimal information exists on the possible consequences that age-related sperm methylation errors may have post-fertilization on embryogenesis and offspring health. In a mouse study that analyzed DNA methylation in aged sperm, similar epigenetic errors were observed in offspring embryonic brain as well as effects on offspring behavior [35]. Likewise, our group reported generational inheritance of corresponding epigenetic dysregulation in sperm and embryo during IVF treatment [25]. Here, we are the first to dissect the epigenomes of the ICM and TE cell lineages of human blastocysts in response to paternal aging. We are also the first to describe an altered transcriptome that appears to be independent of the DNA methylation changes, yet still impacting gene pathways implicated in neuronal signaling and neurodevelopmental disorders, specifically in the inner cell mass.

Advanced paternal age has the potential to lead to a detrimental effect on semen quality as well as increased DNA fragmentation, particularly in men over 50 years of age [36]. We purposefully excluded patients that were outside normal clinical semen parameter ranges [30], to eliminate any possible confounding influence from male factor infertility. Studies on clinical reproductive outcomes of older men are oftentimes difficult to interpret due to the lack of adjustment for maternal age. For our study we strategically included only donor oocyte IVF cycles (young fertile women) to control for the profound effects of female aging on fertility. In spite of our study design, the blastocyst epigenome and transcriptome were both disrupted in these embryos derived from advanced paternal age compared to young paternal age counterparts.

A significant increase in global methylation was observed in both ICM and TE lineages of APA-derived blastocysts with a corresponding high proportion of hypermethylated regions. These DMRs were clustered at significant cytobands throughout the genome; four cytobands were independently enriched in both ICM and TE tissues, with the greatest enrichment being chr19p13.3. We previously identified this subtelomeric region as being enriched in sperm and intact blastocysts derived from APA fathers [25]. Subtelomeric regions may be excluded from large-scale epigenetic reprogramming events [37,38,39], or a looser chromatin conformation may be required to access the large number of genes in this region. Both explanations support an avenue of vulnerability to epigenetic disruption.

A statistically significant overlap of DMR-associated genes was detected among ICM and TE datasets from aged fathers compared to young. With that said, very similar gene ontology enrichment was observed, particularly in the cellular compartment and molecular function categories. Many overlapping cellular component terms centered on neurotransmission activity, such as glutamatergic synapse, neuron to neuron synapse, asymmetric synapse, postsynaptic specialization, and postsynaptic density. We observed a similar association with neurotransmission, specifically opioid signaling, in our prior work on APA-derived sperm and blastocysts [25]. Molecular function overlapping terms included various activities required for embryonic development like regulation of cell proliferation and differentiation, metabolism, and cell signaling. Such terms included GTPase regulator activity, NTPase regulator activity, protein serine/threonine kinase activity, and phosphatidylinositol binding. Finally, while the biological process category was mostly unique between ICM and TE, the premise of the terms were similar and relevant to the other categories; neurotransmission and embryonic developmental activity. Reactome enrichment identified comparable pathways in the ICM of APA-derived blastocysts, including neuronal system and various signal transduction processes. For example, RHO GTPases are involved in the regulation of many fundamental cellular processes important during embryonic development, including morphogenesis, polarity, movement, and cell division. As a consequence, RHO GTPases play important roles in neuronal development, and disruption plays a role in the etiology of neurodegenerative diseases [40]. Similar signal transduction processes were identified in APA-derived TE DMRs, as well as hormone-driven regulatory pathways involved in energy metabolism.

APA-derived ICM and TE DMRs encompassed statistically significant enrichment for neurodevelopmental genes implicated in autism spectrum disorder and schizophrenia. The incidence of these neuropsychiatric conditions is known to increase progressively with increasing paternal age [33]. Evidence for abnormal DNA methylation in association with these disorders is also described [41,42,43]. Our group reported a relationship between paternal aging and epimutations at genes implicated in these disorders in sperm and intact blastocysts [25], with 108 genes overlapping our present study in ICM and TE tissues. A number of these genes were also present in pathways identified by gene ontology enrichment analysis, such as ANKS1B, CACNA1C, IGSF9B and SHANK3. Mechanistically, alterations to DNA methylation may occur in sperm over time as men age, and persist through fertilization at this group of genes which appear to be highly susceptible to epimutations.

Another group of genes that present a level of vulnerability are imprinted genes, since gametes acquire parent-of-origin specific genomic imprints that then persist through fertilization and embryonic development by evading epigenetic reprogramming [44]. Imprinted genes are known to play an essential role in brain development and contribute to some neurodevelopmental conditions [45], as well as development of the placenta. Upon paternal aging in a mouse model, methylation differences were found in brain-expressed imprinted loci, with concurrent behavioral changes [46]. Twenty-three imprinted genes were found to be altered within the ICM or TE of APA-derived blastocysts compared to young, with seven altered in both lineages. In particular, DLGAP2 is one imprinted gene that had differential methylation in both the ICM and TE, as well as in the intact blastocyst and sperm from aged fathers [25]. This gene is also found in the overlapping gene ontology terms involved in neurotransmission from ICM and TE DMRs, and has been implicated in the development of autism [43, 47, 48]. Imprinting control regions (ICRs) were also impacted within the ICM and TE of APA-derived blastocysts. Results from our targeted methylation analysis at the KCNQ1OT1 ICR validates the hypermethylation observed for this region in the global methylome data, as well as aligns with a paternal age effect representing an aberrant gain of methylation presumably from the sperm contribution. To translate these results to later stages in embryonic development, a comparable gain of methylation was observed at the Kcnq1ot1 ICR in mouse embryonic placentas derived from aged males [49], and older paternal ages have been linked to hypermethylation in human IVF placentas collected at time of delivery [50].

There was a remarkably large overlap of DMR-associated genes between the ICM and TE methylomes upon paternal aging, suggesting that many of the same genomic regions may be susceptible to methylation dysregulation. Concordant with the increased risk of impaired neurocognitive phenotypes observed in offspring, paternal age-induced epigenetic alterations occurred at genes involved in several neuronal signaling gene ontology terms, imprinted genes, and genes implicated in neurodevelopmental disorders. While this result was mostly expected in the ICM methylome, we were surprised to discover this to also be true in the TE methylome. Interestingly, a number of recent studies have discussed the placenta-brain-axis (PBA), such that abnormal regulation of certain genes in the placenta affect the fetal brain [51,52,53,54,55]. The placenta produces neurotransmitters that may circulate and influence brain development, and it has been implicated that neurobehavioral disorders such as autism spectrum disorder likely trace their origins back to placental disturbances. Due to this intimate relationship, it has been proposed that the placenta is a promising tissue for identifying DNA methylation changes at genes that also function in the fetal brain, with possible associations to autism spectrum disorder diagnoses.

Like the methylome, paternal age strongly impacted ICM and TE transcription. However, only differentially expressed genes in the ICM of APA-derived blastocysts exhibited statistical enrichment for gene ontology terms and pathways. Neurotransmission and other signaling pathways known to play a role in the brain, such as glutamergic synapse, GABAergic synapse, apelin signaling and relaxin signaling, were identified. Significantly altered genes associated with some of these pathways include SLC38A3 and ADCY5, which are critical for amino acid and amide transport vital to not only ICM proliferation, but also future brain function [56, 57]. Others include NOS2, an important signaling molecule of the central nervous system associated with neurotransmission and diverse brain disorders [58], and AGTR1, which is associated with susceptibility to brain neurodegeneration [59]. Linear regression analysis illustrated a progressive increase in gene expression as paternal age increased for eleven genes, two of which show an association with brain function; ITGA2 has been linked to axonogenesis [60] while SLC25A27 is involved metabolism in the brain, and is implicated in autism spectrum disorder as well as a possible connection to schizophrenia [61, 62]. Interestingly, all of the top ten KEGG pathways have some involvement in neurodevelopmental signaling and many are involved in the etiology of autism spectrum disorder. Numerous key gene ontology terms were identified including the regulation of hormones and peptides through transport and secretion, as well as amino acid and organic acid transport activity. Additional DEGs in the ICM of APA-derived blastocysts have been determined critical for blastocyst cell proliferation, such as IL6 involved in ICM cell numbers and expansion. Yet, despite this strong association with pathways important to brain function, it may be too early in development to impact transcription for genes implicated in neurodevelopmental disorders or imprinted genes, as none of these experienced a statistical enrichment among the differentially expressed genes in either the ICM or TE of blastocysts derived from APA fathers.

Though both the ICM and TE comparisons lead to a similar number of differentially expressed genes, by comparison we found no significant signaling pathways or gene ontology terms enriched in the trophectoderm of aged fathers. As the trophectoderm lineage is responsible for implantation into the uterus and forms the placenta, our results support clinical data showing comparable IVF outcomes of donor oocyte APA cycles to those of younger men [5, 63]. Rather, the ICM lineage that gives rise to the fetus displayed various altered pathways and gene ontology terms important in brain function, which may contribute to downstream consequences of offspring health and align with clinical reports of greater risks after birth.

Interestingly, enrichment for the glutamatergic synapse pathway was identified in both the APA-derived methylome and transcriptome datasets. Glutamatergic synapse is involved in establishing neuronal network connections during brain development and mediating the cellular processes required for neurotransmission [64]. Thus, disruptions may play a relevant role in neurocognitive disease, and presents a possible route for identifying early markers to neurodevelopmental disorders.

Mechanisms exist for aged sperm to transfer altered chromatin signatures to the embryo. However, it is unclear when the adverse effects on offspring health initially occur. The question remains, how early in embryo development epigenetic differences and disease risk are identifiable. Here, we report alterations to the methylome and transcriptome as early as the blastocyst stage, with differences observed in the first two cell lineages. However, despite these events taking place within the same embryos, there is very limited overlap and no correlation between DMR-associated genes in the methylome data and the differentially expressed genes in the transcriptome data for either the ICM or the TE lineages. This suggests that two independent mechanisms are likely at play during embryogenesis; the immediate cascade of altered gene expression occurring in the APA-derived preimplantation embryo, and the long-term alteration of epigenetic marks potentially inherited from APA sperm that will influence transcription later in development and after birth. Since altered DNA methylation does not appear to directly influence gene expression at the preimplantation stage, it may be that additional epigenetic mechanisms, such as histone modifications and miRNA expression profiles, or another paternal effect factor, is leading to the immediate and transient changes in blastocyst gene expression. The large number of synthesized mRNA transcripts expressed in the preimplantation embryo likely represent the plasticity of the embryo and its ability to adapt to its ever-changing environment. Excess transcripts present in the blastocyst may be generated only for utilization if needed, with post-transcriptional regulatory processes occurring at this time. Thus, the transcriptome represents a snapshot of the present timepoint, and is likely not a true reflection of future health and disease risk. Nevertheless, it was interesting to observe transcriptomic dysregulation involving numerous neuronal signaling pathways in the ICM only, showing potential early disruption to future brain development. Meanwhile, methylome changes are stable and inherited as cell division and differentiation occur during development, and are likely the mechanism leading to future predisposition to disease. Both ICM and TE methylomes were highly impacted at genes implicated in neurodevelopmental disorders and neuronal signaling pathways, with possible subsequent transcriptional changes occurring later in development.

It is important to highlight the strengths and limitations to this study. The use of only donor-oocyte IVF cycles controlled for the profound influence of female aging on fertility and was a strategic approach to isolating the effects of paternal age on embryos. Significant patient confounders such as paternal BMI, smoking status, semen analyses results and IVF cycle outcomes were controlled for, however, additional medical history like familial backgrounds and offspring health conditions are not collected by the IVF clinic. The small sample sizes used for genome-wide methylome and transcriptome analyses, although necessary for economic purposes, could represent a technical limitation that may have adversely impacted the statistical power of our results. Finally, the starting input material was also considerably limited, especially following the dissection of ICM and TE tissues and further isolation of DNA and RNA from the same embryo. However, the ability to examine methylome and transcriptome within the same embryos and tissues was a significant strength for the aim of our study.

Based on the divergent ages between our young and APA groups, it is difficult to ascertain if a specific paternal age threshold leads to altered methylation and transcription patterns. Epidemiological studies suggest a linear relationship between increasing paternal age and the risk for neurodevelopmental disorders in their offspring [33]. Likewise, we and others have reported a linear relationship between increasing paternal age and methylation alterations in sperm [25, 65]. However, APA is a subtle and varying effect to male reproductive potential, and inter-individual variability between males exceeds age-associated variation [65]. Similarly, not every blastocyst that is derived from an APA father will become a child that presents with a neurodevelopmental disorder, and these disorders themselves exist on a spectrum of symptom severity that may not always be clinically diagnosed. Since the dynamic nature of epigenetic modifications enables them to be influenced by various intrinsic and environmental factors, it would be unreasonable to expect an absolute effect similar to a genetic mutation. Therefore, we instead predict that a threshold for APA-induced epigenetic alterations exists, and only if surpassed, culminates in a predisposition to disease and ultimately an observed phenotype in offspring. Ongoing investigation into specific genes which consistently show altered methylation patterns will further our understanding of the role of paternal age in the etiology of these neurological conditions.

While our results revealed potential expression markers for susceptibility that can be detected as early as the first blastocyst tissue lineage division, the lack of correlation between the ICM and TE makes transcription profiling of biopsied TE cells futile. Rather, the placenta-brain-axis (PBA) relationship supports the idea that TE cells are a promising avenue for classifying DNA methylation errors at genes that also function in the fetal brain, with possible associations with future neurodevelopmental disorders. In this fashion, we may achieve a way to epigenetically rank embryos for offspring disease risk assessment using biopsied TE cells from IVF blastocysts.