Does advanced paternal age affect the embryonic mouse brain differently by sex?

A four-group design with matched embryonic stages

The paper asks whether paternal aging leaves a different transcriptional trace in male and female embryonic brains. To test this, the authors crossed 3-month-old female mice with either 3-month-old or 19-month-old males and collected the telencephalon, the embryonic forebrain region that gives rise to the cortex, at embryonic day 14.5, a stage the authors describe as critical for neurogenesis and corticogenesis.

The design yields four groups: young-father offspring (YFO) male, YFO female, aged-father offspring (AFO) male, and AFO female, each with six biological replicates, for 24 samples in total. Sex was assigned by genomic PCR targeting the Ube1 gene on tail tissue, so male and female samples were identified before sequencing rather than inferred afterwards.

The most consequential control is developmental staging. Because small differences in embryonic age can shift brain gene expression, the authors staged each candidate embryo using right hindlimb morphology under the Embryonic Mouse Ontogenetic Staging System (eMOSS). From 106 candidate embryos across thirteen independent litters, they selected six per group with closely matched estimated stages. This matters because an apparent sex difference could otherwise reflect one group being slightly older.

Bulk RNA-seq measures the whole tissue, not single cells

Each telencephalon was sequenced in bulk, meaning RNA from the entire dissected tissue was pooled into one measurement per embryo. The result is a per-embryo expression profile, typically reported as transcripts per million (TPM) or as raw read counts. Bulk data can reveal that a tissue differs overall, but it cannot say which cell type within the telencephalon changed.

Sequencing quality was high: more than 99% of reads mapped to the mouse reference genome in all samples, and expression distributions were similar across samples. The authors used DESeq2 with a model including sex and paternal-age condition as explanatory variables, so each factor was assessed while controlling for the other. Clustering and heatmaps used the top 100 genes ranked by statistical significance; volcano plots used edgeR with a threshold of FDR < 0.05 and |log2 fold change| ≥ 0.5.

One detail limits how much weight the clustering carries. The first three principal components explained 12.2%, 8.9%, and 5.3% of total variance, roughly 26% together. The group structure is therefore a partial view of a noisy dataset, which is consistent with the authors describing the female pattern as a tendency rather than a strong effect.

Female aged-father embryos cluster apart

In the three-dimensional principal component analysis, YFO male and AFO male samples were not clearly separated, which the authors note is consistent with their earlier work that examined male embryos only. The AFO female samples, by contrast, formed a distinct cluster separated from the other three groups.

The differential-expression heatmaps point the same way. Genes associated with sex separated male from female samples, and genes associated with paternal age separated YFO from AFO samples, indicating that both factors contribute to transcriptional variation. The volcano plots add direction: the male AFO-versus-YFO comparison showed most genes near zero fold change with low significance, while the female comparison showed a broader distribution and a tendency toward positive log2 fold change, meaning higher expression in AFO females than in YFO females.

Sex-based comparisons recovered expected sex-chromosome-linked genes, including Ddx3y, Uty, Eif2s3y, Xist, and Kdm5d, which supports that the sex grouping is biologically real. The male-female separation appeared more pronounced in the AFO condition than in the YFO condition, largely because of the changes seen in AFO females. The authors interpret this as paternal-aging-associated variation being more evident in female embryos and widening sex-dependent divergence.

High-confidence ASD genes do not shift together

Because advanced paternal age is an established epidemiological risk factor for autism spectrum disorder, the authors checked whether known high-confidence autism risk genes moved as a block. They took SFARI Category 1 genes, converted human gene identifiers to mouse orthologs, and plotted their expression across the four groups.

The result was negative in a specific and useful way. The SFARI Category 1 genes did not segregate clearly by paternal age or sex. Some genes looked relatively higher in AFO samples in both sexes, but the genes showing that trend differed between males and females, and variability between individual embryos within a group was visible. No consistent group-wide pattern emerged.

The authors conclude that the paternal-age transcriptional differences in this dataset are not readily explained by a coordinated shift in currently established high-confidence autism risk genes. That is a statement about this gene set at this stage and tissue, not a claim that paternal age is unrelated to autism biology. It redirects attention toward developmental-stage-specific signatures that predefined risk gene lists may not capture.

A tendency is not a mechanism

The paper's central claim is carefully worded: a sex-biased transcriptional response to paternal aging, described as a tendency. The evidence is clustering structure and the direction of fold changes, not a validated causal pathway. The authors do not identify which sperm-borne factor, such as de novo mutations or epigenetic changes, produced the female-specific pattern; those mechanisms are cited as proposals from prior work, not tested here.

The sample size also bounds interpretation. Six embryos per group is adequate for a data descriptor and for detecting broad structure, but it is small for ranking individual genes as robustly differentially expressed, and the authors present the volcano plots as visualization rather than as a definitive differentially expressed gene list. The partial variance explained by the top principal components reinforces that the separation is real but modest.

Scope is equally important. The findings come from one inbred strain, C57BL/6J, one paternal age contrast (3 versus 19 months), one embryonic stage (E14.5), and one brain region (telencephalon). Nothing here shows that the transcriptional tendency produces behavioral or anatomical differences in offspring, and nothing here establishes that the same pattern holds at other developmental stages or in other brain regions.

The reusable value is the sex-resolved resource

The most decision-relevant point for readers is a comparison the paper makes explicit: analyzing males alone would have missed the signal. The male AFO-versus-YFO comparison showed limited transcriptional differences, matching the authors' earlier male-only study, while the female comparison carried the visible change. For anyone planning or interpreting paternal-age neurodevelopment work, that makes sex-resolved sampling a design requirement rather than an optional add-on.

The second practical point is the staging control. Selecting embryos by hindlimb morphology rather than by mating date alone removes a confound that could masquerade as a paternal-age or sex effect. The dataset, deposited with raw reads, processed TPM values, and differential expression result files, is built to be reused with that control already in place.

The priority the paper establishes is therefore specific: treat female embryos from aged fathers as the group where paternal-age transcriptional effects are most likely to be detectable at E14.5, and treat established autism risk gene sets as unlikely to show a coordinated response in this tissue and stage. Testing whether that female-biased tendency holds at other stages, in other brain regions, or with larger samples is the next question this dataset sets up rather than answers.

Sources used in this answer