Effects of IVF on testis weight and histology. To ascertain whether IVF could influence male reproductive function, we first assessed the effect of IVF on testes’ weight and histology using our mouse model (Figure 1A). Mean body and testicular weight at both 12 and 39 weeks were significantly higher in offspring of the IVF group compared with naturally conceived controls (naturals) (Figure 1, B and C), which led to a significantly higher testicular-to-body weight ratio for the IVF group in 39-week offspring (Figure 1D). Because testes are the primary producers of the key male sex hormone, testosterone, they play a critical role in both physical and reproductive health (26). Due to testosterone’s effect on multiple systems, abnormal testosterone levels can be indicators of underlying testicular and overall adverse health conditions (27). Accordingly, we determined whether an increase in testicular weight correlated with changes in testosterone levels and expression of the androgen receptor (AR) and found that both testosterone levels and AR expression in testis were significantly reduced in the IVF group at both 12 and 39 weeks compared with naturals (Figure 1, E and F).

Mouse model and general parameters. (A) Mouse model: Top: Naturally conceived embryos (Naturals) were generated using SJL/B6 male mice and CF-1 female mice, and bottom: IVF embryos were produced using capacitated sperm from SJL/B6 GFP+ or GFP– mice and eggs from superovulated CF-1 mice. Embryos were cultured to the blastocyst stage; 10 blastocysts were transferred to pseudopregnant females. At E18.5 pregnant females from both groups were C-sectioned, and fetuses were delivered and collected for molecular analysis. For adult cohorts C-section–delivered pups at E18.5 (Naturals) or E19.0 (IVF) were fostered with wild-type dams and monitored until 12 or 39 weeks of age. Shown are (B) body weights before 6 hours of fasting, (C) testicular weights, (D) testicular/body weight ratios, (E) serum testosterone levels assayed by ELISA, and (F) representative immunoblot testicular protein levels of AR relative to GAPDH assayed by Western blots. Data are depicted as mean ± SEM; n = 10–12 per group. The black line represents the mean of each group. Statistical significance was determined by t test; *P < 0.05, **P < 0.01, and ***P < 0.001 IVF groups compared with Naturals.

Spermatogenesis is a complex process involving the proliferation and differentiation of germ cells. Several factors could influence this process, including alterations in the seminiferous epithelium (28). Previous studies showed that seminiferous epithelium changes impair elongation of spermatids, affect axoneme formation, and disturb interactions between Sertoli cells and germ cells that lead to abnormal spermatogenesis (28). Seminiferous epithelium perturbations also arrest spermatogenesis at various developmental stages in mutant mice (29). Thus, any adverse changes to the seminiferous epithelium may impair spermatogenesis and affect male fertility. We therefore histologically examined the seminiferous epithelium at 12 and 39 weeks (Figure 2, A and B). Hematoxylin-eosin staining revealed increased seminiferous tubule cross-sectional area and seminiferous epithelium area per tubule at 39 weeks but not at 12 weeks (Figure 2, C and F), whereas the height and lumen of the seminiferous epithelium were increased in both 12- and 39-week offspring (Figure 2, D and E).

Testicular morphology analysis using hematoxylin-eosin, Picrosirius red, and Masson’s trichrome staining. Testicular cross sections at (A) 12 weeks of age and (B) 39 weeks of age using hematoxylin-eosin, Picrosirius red (PR), and Masson’s trichrome (MT). (Original magnification is 4×.) Morphology of the seminiferous tubules in hematoxylin-eosin slides, including (C) whole cross-sectional area, (D) height, (E) lumen area, and (F) total seminiferous epithelium area. Percentage of positively stained area with PR (G) and MT (TC) (H) is depicted. The black line represents the mean of each group. Each data point represents an individual conceptus from a minimum of 5 litters (n = 5–10/group). Data are expressed as mean ± SEM. Statistical analysis between groups was done by t test; *P < 0.05, and **P < 0.01, when comparing IVF against Naturals.

In testes, collagen, a major component of the testicular basement membrane, is located at the base of the seminiferous epithelium (30). The basement membrane has multiple functions, including germ cell movement and communication between different cell types (31). Disruption of this matrix is associated with azoospermia (32). Additionally, excessive collagen is associated with an increase in tunica albuginea weight and thickening of the basement membrane, as seen in fibrosis and aging models (30). We therefore assessed the effect of ART on the extracellular matrix in testis using 2 staining protocols: PR, which quickly and efficiently assesses collagen fiber content, organization, and orientation (33), and MT, which not only detects the collagen fiber deposition but also can distinguish muscle fibers (red), collagen (blue), and nuclei (black) simultaneously (34). Both methods revealed increased staining intensity exclusively at 39 weeks (Figure 2, G and H). To validate this increased staining, we assayed collagen type I alpha 1 chain (Col1a1) RNA and protein by reverse transcription quantitative PCR (RT-qPCR) and immunoblotting, respectively, as well as another extracellular matrix marker, vascular cell adhesion protein 1 (Vcam1), all of which were increased in IVF offspring compared with naturals (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.188931DS1). Finally, to determine whether changes in testicular size are due to alterations in cell composition — specifically spermatogonia, spermatocytes, and round spermatids — we performed an immunohistochemistry analysis of testicular sections using anti-SYCP3, PNA-488 for acrosome staining, and DAPI. At 12 weeks, we did not observe changes in cell composition in whole testis (Supplemental Figure 2, A and B). However, at 39 weeks, we observed a decrease in all cell types (Supplemental Figure 2, C and D), which could indicate an early loss of cells essential for spermatogenesis, potentially influencing future fertility. Taken together, these results indicate an increased thickening of the tubular walls and a greater degree of tubule disorganization with higher collagen content in IVF male offspring, with no significant changes in cell composition at 12 weeks but with changes by 39 weeks. Additionally, this increase in collagen deposition contributed to the increased testicular weight in IVF offspring.

Effect of IVF on sperm count, morphology, and DNA methylation. Given the IVF-associated changes in testis morphology, we next examined the effect of IVF on several sperm parameters. Although there was no significant difference in sperm counts from 12-week-old IVF offspring, there was greater variability in sperm number (Figure 3A). When testes’ size and sperm count were correlated, IVF offspring with larger testes had lower sperm counts (Supplemental Figure 3), and the percentage of sperm with abnormal morphology was increased (Figure 3B). No data are available for the 39-week time point because sperm were immediately frozen, prohibiting characterization.

Sperm parameters and testicular DNA methylation. (A) Sperm count using a bright-field microscope; (B) sperm morphology quantification using eosin 1% staining where sperm is categorized as normal (%normal), abnormal head (%head), abnormal tail (%tail), and abnormal midpiece (%midpiece); (C) testicular global DNA methylation; and (D) sperm global DNA methylation using LUMA for 12- and 39-week natural and IVF offspring. Infinium Mouse Methylation BeadChip sperm analysis, including (E) volcano plot for DMRs at 12 weeks, 12 weeks heatmap with (F) hypermethylated DMRs and (G) hypomethylated DMRs, (H) volcano plot for DMRs at 39 weeks, and 39 weeks heatmap with (I) hypermethylated DMRs and (J) hypomethylated DMRs. Seg, segmentation analysis. Each data point represents 1 individual from a minimum of 5 litters (n = 10–15/group). Data are expressed as mean ± SEM. Statistical analysis between groups was done by t test; *, **, and *** represent significant differences of P < 0.05, P < 0.01, and P < 0.001 respectively, when comparing IVF and Naturals. Variability was calculated using an F test, and # represents a significant difference (P < 0.05) in IVF compared with Naturals.

IVF is also associated with changes in whole-genome average DNA methylation (7, 11, 24, 25). We therefore assessed global DNA methylation using a luminometric methylation assay (LUMA) on testis DNA at E18.5, 12 weeks, and 39 weeks and sperm DNA at 12 and 39 weeks. We observed statistically significant decreased DNA methylation in testes and sperm from IVF-conceived male offspring compared with naturals at all time points (Figure 3, C and D). To identify specific regions with altered DNA methylation that could be transmitted to offspring, we examined DNA methylation in sperm from 12- and 39-week-old IVF-conceived mice using the Infinium Mouse Methylation-12v1-0 BeadChip array, which assays approximately 285,000 CpG probes distributed across the genome (7, 24, 35, 36). As previously described (7, 35), we first used all probes to determine the differentially methylated loci (DMLs) that had a 5% absolute change (0.05 difference) between groups. With these DMLs, we then determined differentially methylated regions (DMRs) that are associated with a differentially methylated probe on the array. Given that changes between 10% and 15% in methylation can influence gene expression and produce a phenotype (37), we considered a DMR a genomic region that had a 10% difference in methylation between groups. Heatmap and principal component analysis (PCA) by probe passing rate, group, and age showed a differential clustering between sperm from IVF and natural offspring (Supplemental Figure 4). At 12 weeks, we observed 89 DMRs, 4 hypermethylated DMRs, and 85 hypomethylated DMRs in IVF compared with naturals (Figure 3, E–G, and Supplemental Table 3). Gene ontology analysis failed to detect a significant association with specific biological functions at 12 weeks. Finally, in 39-week sperm samples, we observed that IVF-conceived males exhibited 12,683 DMRs, 4,914 hypermethylated DMRs, and 7,769 hypomethylated DMRs compared with naturals (Figure 3, H–J, and Supplemental Table 4). Gene ontology analysis showed that the most affected pathways related to cell-cell interactions, retrotransposon silencing, regulation of membrane potential, signal transduction, cell morphogenesis, and regulation of actin filament processes (Supplemental Figure 4E). Together these results show that IVF procedures impacted both testicular and sperm DNA methylation, with possible effects on the next generation, and that age increased the number of DMRs in sperm.

Effect of IVF on testicular gene expression. To ascertain whether the changes in DNA methylation are associated with changes in gene expression of IVF-conceived males, we performed RNA-Seq on fetal and adult testes collected at E18.5, 12 weeks, and 39 weeks. PCA showed a clear separation between IVF and naturals (Supplemental Figure 5, A–C). Volcano plots revealed 57 upregulated and 78 downregulated genes at E18.5 (Figure 4A and Supplemental Table 5), 17 upregulated and 57 downregulated genes at 12 weeks (Figure 4B and Supplemental Table 6), and 64 upregulated and 29 downregulated genes at 39 weeks (Figure 4C and Supplemental Table 7). Not all DEGs at 12 weeks were observed at 39 weeks. Testes from IVF-conceived males at E18.5 had the greatest number of DEGs, and this difference did not appear to worsen with age. The heatmap for all DEGs showed tight clustering for all naturals, with IVF groups clustering together (Figure 4, D–F). Gene ontology analysis showed dysregulation in transmembrane transport and testicular function at E18.5 but failed to detect a significant association with specific biological functions at 12 and 39 weeks (Figure 4G). Finally, we determined if the DEGs in testis overlap with affected sperm DMRs. At 12 weeks, we did not observe any overlap between the DEGs and DMRs (Supplemental Figure 5D), whereas at 39 weeks, we observed that expression of 1 downregulated gene, Ttc22, and 8 upregulated genes, 1700019P21Rik, Cd24a, Gm31447, Nedd9, 8430426J06Rik, Dazl, Sycp1, and Adamts9, was linked with hypermethylated and hypomethylated DMRs, respectively (Supplemental Figure 5E). These data suggest little correlation between changes in gene expression and DNA methylation, indicating that the observed changes in gene expression may be due to epigenetic modifications not measured in this study.

Transcriptome analysis of testicular tissues by RNA-Seq. Volcano plots showing differentially expressed genes (DEGs) for (A) E18.5 gonads and testis at (B) 12 weeks and (C) 39 weeks IVF versus natural. Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs in (D) E18.5 gonads and testis at (E) 12 weeks and (F) 39 weeks. Gene ontology analysis for relevant affected pathways: (G) E18.5 testis. Color boxes at the top denote the experimental group.

Multigenerational effects of IVF on placentas from F2 concepti. Because sperm DNA methylation was perturbed in IVF-conceived males, we determined whether these changes were transmitted to their offspring. Accordingly, 12-week-old IVF-conceived males and naturals were mated with CF-1 wild-type females, and concepti (F2) were isolated at E18.5 (Figure 5A). We did not observe any differences in the time of conception when comparing pregnancies using IVF males compared with naturals; both had a conception time of 19 days. Litters sired by IVF-conceived males had fewer live pups 12 hours after birth compared with those sired by naturals (Figure 5B). Because we previously identified sex differences in fetal and placental weight in the first generation (7, 36), we performed all analyses by sex. In F2 males, although the mean placental weight was unchanged at E18.5 (Figure 5C), the mean fetal weight was decreased for IVF concepti compared with naturals (Figure 5D). In F2 females, both mean placental weight and mean fetal weight were decreased for IVF concepti compared with naturals at E18.5 (Figure 5, E and F). The reduction in fetal weight, with or without a decrease in placental weight, resulted in significantly reduced fetal/placental weight ratios for F2 IVF compared with naturals (Figure 5, G and H).

Multigenerational impacts of IVF. (A) Mouse model: Top: F2 naturally conceived offspring were generated using natural male offspring and CF-1 female mice, and bottom: F2 IVF offspring were produced using IVF male offspring and CF-1 female mice. E18.5 pregnant females from both groups were C-sectioned, and concepti were delivered and collected for molecular and histological analysis. For adult cohorts naturally delivered pups were maintained until 12 weeks of age. (B) Total number of pups is depicted for 4 consecutive litters, with data presented as summary of total pups after each litter from 4 breeding pairs/group. Placental (C) and fetal (D) weight for male offspring and placental (E) and fetal (F) weight for female offspring are shown. The fetal weight/placental weight ratio for male (G) and female (H) offspring. Data are mean ± SEM; n = 16–20 per group. Black lines represent the mean of each group. Statistical significance was determined by t test; *P < 0.05, **P < 0.01, and ***P < 0.001, when IVF groups are compared with Naturals.

To address if IVF has a multigenerational impact on the area of the junctional and labyrinth regions in the placenta, as previously observed in IVF offspring (24, 25), we first used hematoxylin-eosin staining at E18.5 (Figure 6, A–D) to assess changes in the ratio of junctional and labyrinth zone areas compared with total placental area. The percentage of labyrinth and junctional zone area was significantly lower and higher, respectively, in F2 IVF placentas compared with naturals (Figure 6, B–F) in both female and male offspring as previously observed for IVF offspring (7). To determine whether blood vessel density was impacted, we performed immunohistochemistry for CD34, a marker of fetal endothelial cells. We have previously shown that CD34 is affected during the early stages of IVF pregnancies (7, 24). We observed that the microvessel density per area was increased in placentas from both IVF F2 males (Figure 6, G and H) and IVF F2 females (Figure 6, I and J) compared with naturals, which impacted placental efficiency.

Multigenerational impacts of IVF in placenta. Percentage of placental junctional and labyrinth zone at E18.5 using hematoxylin-eosin staining. Placenta cross sections from (A) hematoxylin-eosin–stained tissues (scale bar: 850 μm) from F2 E18.5 male placentas with the percentage of (B) labyrinth zone and (C) junctional zone shown. (D) Hematoxylin-eosin histological sections from F2 E18.5 female placentas with the percentage of (E) labyrinth zone and (F) junctional zone indicated. Quantitative analysis of labyrinth fetal endothelial cells using CD34 and counterstained with hematoxylin in natural and IVF placentas at E18.5: (G) representative images of male placentas from E18.5 natural and IVF and (H) quantification of E18.5 fetal endothelial cell positive staining as a percentage of total labyrinth area using ePathology software (5, 24) for male placentas. (I) Representative images of female placentas from F2 E18.5 natural and IVF and (J) quantification of E18.5 fetal endothelial cell positive staining in females. Original magnification is 4× (G and I). RNA-Seq using F2 male and female E18.5 placentas: volcano plots showing DEGs for (K) F2 male E18.5 placentas and (L) F2 female E18.5 placentas IVF versus natural. Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs in (M) F2 male E18.5 placentas and (N) F2 female E18.5 placentas. For histology, each data point represents an individual placenta from a minimum of 4 different litters. The black line represents the mean of each group (n = 8/group/sex). Statistical significance was determined by t test; *P < 0.05 when compared groups against natural.

Finally, given our previous findings of altered gene expression in F1 placentas at E18.5 (7), we determined whether gene expression in F2 placentas from IVF offspring was similarly affected. In F2 males, PCA revealed a clear separation between the IVF and natural groups (Supplemental Figure 6A), with 635 genes upregulated and 542 genes downregulated in the IVF group compared with the natural group (Figure 6, K and M, and Supplemental Table 8). Gene ontology analysis indicated dysregulation of pathways involved in tissue remodeling, nutrient transport, and angiogenesis (Supplemental Figure 6, B and C). Similarly, in F2 females, PCA showed a distinct separation between the IVF and natural groups (Supplemental Figure 6D), with 438 upregulated and 421 downregulated genes observed in the IVF group compared with the natural group (Figure 6, L and N, and Supplemental Table 9). Gene ontology analysis highlighted dysregulation of pathways related to metabolism, transport, and inflammation (Supplemental Figure 6, E and F). Interestingly, in both female and male placentas, vascular endothelial growth factor receptor 1 (Flt1) was upregulated, which could contribute to changes in vessel formation.

When comparing these data with previous published data of E18.5 RNA-Seq, we observed that for males approximately 60 DEGs were shared whereas in F2 placentas there was a higher number of new DEGs that could reflect paternal stress contribution as well as epigenetic alterations in the sperm (Supplemental Figure 6G). For females, we observed that when compared with F1 females they had more than 100 shared DEGs but more than 700 unique ones. This difference could reflect the paternal contribution to F2 placentas whereas the F1 placental dysregulation occurs directly because of IVF procedures (Supplemental Figure 6H). Taken together, these findings demonstrate that the alterations observed in male IVF offspring have a multigenerational and persistent negative impact on placental morphology and gene expression in F2 offspring.

Effect of IVF on male F2 adult offspring metabolism. We and others have previously determined that metabolism is perturbed in IVF-conceived offspring, with previously published F1 metabolic data (7, 11, 13, 18, 19). To determine whether the F2 progeny sired by IVF-conceived males likewise exhibit adverse health outcomes, we measured the body weight of a cohort of F2 male offspring until 12 weeks of age (Figure 7A) and subsequently euthanized the mice for metabolic analyses. Male F2 offspring sired by IVF-conceived males had higher body weight compared with F2 offspring sired by naturals (Figure 7A). Serum glucose, insulin, total triglyceride, and LDL/VLDL-C were also elevated in F2 offspring relative to naturals (Figure 7, B–E). Total cholesterol was decreased, which was driven by reduced HDL (Figure 7E). We also measured organ weights and calculated their ratio to body weight. F2 offspring sired by IVF-conceived males exhibited decreased brain and testis weight relative to body weight compared with F2 offspring sired by naturals (Figure 7F), whereas pancreas and kidney weights relative to body weight were increased (Figure 7F). Similar phenotypes were also observed in the original IVF male offspring (7, 11).

Multigenerational impact of IVF on male offspring. (A) Body weights were taken from 1 to 12 weeks of age. Metabolic screening included (B) glucose, (C) insulin, (D) triglycerides, and (E) total cholesterol, HDL, and LDL/VLDL. (F) The organ weight/body weight ratio is shown. (G) Volcano plot showing DEGs for liver RNA-Seq. (H) Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs involved in cholesterol, triglyceride, insulin, and glucose metabolic pathways. Gene ontology analysis for pathways involved in metabolism using (I) upregulated DEGs and (J) downregulated DEGs. (K) Volcano plot showing DEGs for testis RNA-Seq. (L) Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs involved in spermatogenesis and testis differentiation. (M) Global DNA methylation of F2 testis at 12 weeks by LUMA. Data are shown as mean ± SEM; n = 10–15 per group. The black line represents the mean of each group. Statistical significance was determined by t test; *P < 0.05 and **P < 0.01, IVF compared with natural. For RNA-Seq, color boxes at the top denote the experimental group.

We previously showed that 12-week IVF offspring exhibited changes in liver gene expression (11). We performed RNA-Seq on a subset of liver and testis samples from male F2 progeny. For liver, PCA showed a clear separation between the IVF and natural groups (Supplemental Figure 7A), with 1,486 upregulated and 1,936 downregulated genes observed when comparing IVF and natural F2 offspring (Figure 7G and Supplemental Table 10). A subset of DEGs associated with metabolic pathways is shown in a heatmap (Figure 7H). Gene ontology analysis revealed dysregulation of pathways involved in cholesterol and phospholipid metabolism, glucose, and insulin processes (Figure 7, I and J). These results are consistent with the metabolic panel and demonstrated that the changes in metabolic measurements are the result of changes in the liver transcriptome. Additionally, we found that some liver DEGs, including Irs1, Irs2, Ripk2, Rspo1, and DNA JC genes, have been previously associated with insulin and glucose resistance (38), which would explain the higher insulin and glucose levels revealed in the metabolic panel. Additionally, we tested if there was a normal correlation between glucose and insulin. Although naturals showed a positive correlation, IVF offspring showed a more variable trend and were not correlated as expected with metabolic changes (Supplemental Figure 10A).

To determine whether DEGs observed in F1 and F2 livers were similar, we conducted an overlap analysis using our previously obtained data (11). F2 offspring had more DEGs compared with F1 (3,949 vs. 17), with no overlapping genes. Although the difference in gene expression could reflect the biology of the IVF males, we doubled the number of samples analyzed in this study, which provided more power to detect differences between the 2 groups. Nevertheless, considering the pathway analysis previously performed (11), cholesterol and triglyceride pathways are equally affected in both F1 and F2 offspring.

For testis, PCA showed a clear separation between F2 offspring from IVF and natural groups (Supplemental Figure 6B), and volcano plots comparing F2 offspring revealed 196 upregulated and 178 downregulated genes (Figure 7K and Supplemental Table 11). Some of the most affected genes involved in testicular function and development are shown in a heatmap with differential clustering between groups (Figure 7L). Gene ontology failed to detect a significant association with specific biological functions.

To determine whether the changes in testis DNA methylation persisted into the next generation, we analyzed global DNA methylation in the testis of F2 offspring. Like the male IVF offspring, total global testicular DNA methylation was decreased in F2 IVF offspring compared with the natural conception group (Figure 7M). Taken together, these findings demonstrate that the changes observed in male IVF offspring have a multigenerational and persistent negative impact on metabolism, DNA methylation, and gene expression in the liver and gonads of F2 male offspring.

Effect of IVF on female F2 adult offspring metabolism. Given the sexually dimorphic IVF mouse outcomes (7, 11, 18), we conducted a similar set of experiments to those described above for female offspring of IVF-conceived males. These mice had a higher body weight compared with offspring sired by naturals (Figure 8A). Serum glucose was decreased, whereas insulin and total triglycerides were increased and displayed a greater variability in the F2 IVF females compared with naturals (Figure 8, B–D). In contrast, cholesterol, although not statistically different, displayed greater variability, and HDL and LDL/VLDL were decreased in F2 IVF offspring compared with naturals (Figure 8E). Further, F2 offspring sired by IVF-conceived males showed a decrease in ovarian weight relative to body weight compared with naturals (Figure 8F), and as in males, pancreas and kidney weights relative to body weight were increased (Figure 8F). Most of these changes were previously observed in female IVF offspring (7, 11). As in males, we tested if there was a normal correlation between glucose and insulin (Supplemental Figure 10B). No correlation was observed, showing that both glucose and insulin are being dysregulated.

Multigenerational impact of IVF on female offspring. (A) Body weights were taken from 1 to 12 weeks of age. Metabolic screening included (B) glucose, (C) insulin, (D) triglycerides, and (E) total cholesterol, HDL, and LDL/VLDL. (F) The organ weight/body weight ratio is shown. (G) Volcano plot showing DEGs for liver RNA-Seq. (H) Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs involved in cholesterol, triglyceride, insulin, and glucose metabolic pathways. Gene ontology analysis for pathways involved in metabolism using (I) upregulated DEGs and (J) downregulated DEGs. (K) Volcano plot showing DEGs for ovary RNA-Seq. (L) Heatmap for log2-transformed expression levels obtained from RNA-Seq of DEGs involved in ovarian function. (M) Global DNA methylation of F2 ovary at 12 weeks by LUMA. Data are depicted as mean ± SEM; n = 10–15 per group. The black line represents the mean of each group. Statistical significance was determined by t test; *P < 0.05, and **P < 0.01 IVF compared with natural. For RNA-Seq color boxes at the top denote the experimental group.

We also performed RNA-Seq on a subset of liver and ovary samples to determine whether gene expression was perturbed in female F2 progeny sired by IVF-conceived males. Similar to male offspring, PCA showed a clear separation between the IVF and natural groups (Supplemental Figure 8A), and volcano plots revealed 2,579 upregulated and 2,081 downregulated genes (Figure 8G and Supplemental Table 12). A heatmap depicts DEGs involved in cholesterol, triglyceride, insulin, and glucose metabolism (Figure 8H). Gene ontology analysis for upregulated (Figure 8I) and downregulated DEGs (Figure 8J) revealed that the most affected pathways are involved in glucose, cholesterol, lipid, and energy pathways, correlating to the observed changes in the metabolic panel. As in males, these results were consistent with the metabolic panel and demonstrated that the changes in metabolic measurements are the result of changes in the liver transcriptome. Although DEGs in males were associated with insulin resistance, DEGs in females were associated with diabetic syndrome.

When RNA-Seq data from ovary were analyzed, PCA again showed a clear separation between F2 offspring sired by IVF-conceived males and those conceived by natural males (Supplemental Figure 8B), with volcano plots showing 64 upregulated and 26 downregulated genes (Figure 8K and Supplemental Table 13). As with liver, we selected a subset of DEGs involved in ovarian function and displayed them on a heatmap (Figure 8L). Gene ontology analysis revealed that ovarian metabolism and hormone production were affected (Supplemental Figure 8C). As in males, we determined if DEGs observed in F1 and F2 offspring livers were similar. Again, F2 showed a higher number of DEGs compared with F1 (4,660 vs. 12), and none of the genes overlapped. Pathway analysis performed in our previous work (11) showed that cholesterol and triglyceride pathways are affected in F1, and the same pathways were affected in our F2 offspring.

To determine whether the DEGs observed in both the liver and gonads were similar between females and males, we conducted an overlap analysis. We found that 2,843 DEGs were shared in the liver, mostly involved in glucose, cholesterol, and triglyceride pathways, consistent with the metabolism panel. Additionally, we identified 1,817 DEGs unique to females and 1,106 unique to males (Supplemental Figure 9A). In gonads, 2 DEGs were shared, with 88 unique to females and 374 to males (Supplemental Figure 9B). Finally, as in males, we assessed whether changes in male IVF offspring could impact global DNA methylation in F2 female gonads. Global DNA methylation in the ovary was decreased in F2 IVF offspring compared with naturals (Figure 8M). These ovarian changes suggest a multigenerational effect.