Age-specific DNA methylation alterations in sperm at imprint control regions may contribute to the risk of autism spectrum disorder in offspring

Results

Characteristics of study participants and semen samples

Because of variability in human populations, we studied a subgroup of the general population. Non-smoking men with a healthy reproductive profile were recruited at the Duke Fertility Center. Study design and inclusion or exclusion criteria are illustrated in Supplementary Figure 1. Socio-demographic data of our study population is shown in Table 1. In brief, a total of 63 men between 18 to 35 years old was included in our statistical analysis. Men aged less than 25 years old represented 44% (n = 28), 27% (n = 17) were between 25 and 29 years old, and nearly 29% (n = 18) were between 30 and 35 years old. Most men lacked a graduate degree (n = 46, 74.1%) and had not fathered a child (n = 55, 87.3%). The majority were healthy volunteers from outside the clinic (n = 48, 76.2%). The mean age of those that were patients was 24 years, while that of the non-patient subgroup was 30 years; this difference was significant (p-value <0.001). Abnormal clinical sperm parameters were found in 16 men (25.4%); 5 in the patient subgroup and 11 in the non-patient subgroup. About one-third (n = 20) was categorized as being overweight or obese. Having a BMI of 25 or more was strongly associated with older age (p-value = 0.001).

Table 1. Socio-demographic data of TIEGER participants and semen characteristics.

Variables	Categorization	n	%
Age (years) (median: 25.00, mean: 25.48)	18–24	28	44.4
	25–29	17	27.0
	30–35	18	28.6
BMI	Normal weight (18 ≤ BMI <25)	43	68.2
	Overweight/obese (BMI ≥25)	20	31.8
Patient at fertility clinic	Yes	15	23.8
	No	48	76.2
Highest degree of education	High school graduate/GED	6	9.7
	Some college	18	29.0
	College graduate	22	35.5
	Graduate degree	16	25.8
Biological children	Yes	8	12.7
	No	55	87.3
Total motile count (TMC)	≤39 × 106 (abnormal)	10	16.1
	>39 × 106	52	83.9
Concentration	<15 × 106 (abnormal)	3	4.8
	≥15 × 106	59	95.2
Motility	<40% (abnormal)	11	17.7
	≥40%	51	82.3
Sperm quality	normal	47	74.6
	abnormal	16	25.4
Legend: The study population includes 63 men between 18 and 35 years old. Variables included in our statistical analysis are shown. Sperm quality is categorized as “abnormal” if at least one of the sperm quality metrics (concentration, motility, TMC) is below the WHO recommended threshold. If the sum is not 63, due to missing data, the percentage was calculated on known data (missing data: TMC: 1; concentration: 1; motility: 1).

Distributions and categorization of DNA methylation outcomes in sperm

Primary analyses of DNA methylation percentages in sperm on the Illumina array, such as global means and individual means, were performed on BMIQ normalized β-values. An overview of our analytic approach is illustrated in Figure 1. In total, we evaluated 482,287 CpG sites. The mean DNA methylation at all CpG sites was 47.63% (β-value = 0.4763). We classified each CpG site by its mean β-value in the following subgroups: unmethylated (UM), hemi-methylated (HM) and fully methylated (FM), as described in the methods section. The number of CpG sites in each methylation subgroup is presented in Table 2. Most CpG sites in sperm were either unmethylated (UM) (46.95%) or fully methylated (FM) (43.02%); the remaining 10.03% was categorized as hemi-methylated (HM). Because β-values have severe heteroscedasticity at low and high methylation values, we used M-values in our validation of individual CpG outcomes. Distributions of the average M-values per site, before and after normalization, are shown as the average densities across 63 samples in Supplementary Figure 2. For the purpose of data interpretation, we also show (and discuss) β-value outcomes.

Figure 1. Workflow to assess age-related DNA methylation changes in sperm and biomarker selection for ASD. Data generated from the Illumina HumanMethylation450 BeadChip went through the following analytic procedures. (1) global DNA methylation (mean per subject) by age. (2) DNA methylation at individual CpGs by age. Note, additional tests were performed, as described in the methods section. (3) We subdivided our outcome data as follows: Unmethylated (UM, mean β-value <0.2), Hemi-methylated (HM, 0.20≤ mean β-value ≤0.80), and Fully methylated (FM, mean β-value >0.80), defined by the mean DNA methylation per CpG site (of all subjects). The top 30 of the most significant results (lowest adjusted p-value) were classified by: (4) significance (top 90), direction (positive, negative), and magnitude (absolute Delta-M >0.1). All age-related DMCs (n = 14,622) were analyzed in terms of the following approaches: (5) chromosome location, island content, functional genomic allocation, and (6) gene ontology (GO). We compared our results with: (7) similar published reports, and (8) listed data on imprinting. Finally, (9) a focus was applied on all DMCs using Simons Foundation Autism Reference Initiative database and other publicly available databases on ASD. (10) We selected a set of potential biomarkers for ASD within our set of imprinted genes.

Table 2. Frequencies and distributions of CpG sites in sperm by DNA methylation subgroup and by age.

DNA Methylation subgroup (SG)	β-value	M-value	Analyzed CpG sites		Significant DMCs by age
			N	%	N	%1	%2
All sites	All	All	482,287	100	14,622	100	3.03
					10,091−; 4,531+	69.01−; 30.99+	2.09−; 0.94+
Unmethylated (UM)	<0.2	(−inf, −2)	226,432	46.95	2,414	16.51	1.07
					2,217−; 197+	91.84−; 8.16+	0.98−; 0.09+
Hemi-methylated (HM)	(0.2–0.8)	(−2, 2)	48,391	10.03	6,390	43.70	13.20
					4,775−; 1,615+	74.73−; 25.27+	9.87−; 3.33+
Fully methylated (FM)	>0.8	(2, +inf)	207,464	43.02	5,818	39.79	2.80
					3,099−; 2,719+	53.27−; 46.73+	1.49−; 1.31+
According to the mean β-value all CpGs of the 450K are categorized by subgroup (SG): Unmethylated (UM), Hemi-methylated (HM), or Fully methylated (FM) (UM: β-value <0.2; HM: β-value (0.2–0.8); FM: β-value >0.8). Within each subgroup the number of analyzed CpG sites on the array are shown, as well as the number (and %) of significant age-related DMCs (FDR <0.05). (+) number of sites that are increased by age. (−) number of sites that are decreased by age. 1% relative to the total number of significant DMCs. 2% relative to the total number of CpGs assessed on the 450K within each subgroup.

Global DNA methylation outcomes in sperm by age

After calculating global DNA methylation at all sites, we searched for a potential correlation between age and DNA methylation. No significant relationship was found. Neither if age was used as a continuous variable in a linear regression model (p-value = 0.76) (Supplementary Figure 3), nor through stratifying by age (below and above median age) (p-value = 0.97) (Supplementary Figure 4). Next, we verified if age-associations could be found after stratification of our population by patient status or by obesity status. Although an opposite trend could be seen by these strata, these correlations were not significant (Supplementary Figures 5 and 6). We further verified if global DNA methylation differed by age within each CpG subgroup (UM, HM and FM). We found no significant correlations; p-values were 0.74 at UM sites, 0.92 at HM sites, and 0.96 at FM sites (Supplementary Figure 7).

Site specific DNA methylation outcomes in sperm by age: directions and magnitudes of change

At each CpG site of the 450K array a potential association between DNA methylation and age was evaluated using a linear regression model, adjusting for BMI, patient status and multiple testing. We found a small fraction of CpG sites (3.03%) that was significantly altered by age; 14,622 CpG sites out of 482,287 CpGs could be identified as differentially methylated sites (DMCs). Among those, most DMCs (69.01%; n = 10,091) were negatively associated with age (Table 2). A Volcano plot illustrates all DMCs that are significantly altered by age (Figure 2A). After subdividing our probes by subgroups, associations were as follows: in the UM subgroup, 91.84% of the DMCs were decreased by age (Table 2 and Figure 2B); in the HM subgroup, 74.73% DMCs were negatively associated with age (Table 2 and Figure 2C); and in the FM subgroup, significant DMCs were nearly equally distributed, with 53.3% (n = 2,719) being positively associated and 46.7% (n = 3,099) being negatively associated with age (Table 2 and Figure 2D). We present a short list of highly significant age-related DMCs (top 30 within each subgroup); with highest magnitude in change (absolute delta M-value higher than 0.1) underlined in Table 3. In the UM subgroup, 20 DMCs were highly susceptible to ageing with a high DNA methylation impact at the following genes: HOXA11; HOXA11AS, LRRC25, LHX1, LHX2, LHX5, ZIC4, GRASP, C16orf13, LOC283999, ZNF385A, LOXL3; DOK1, MAGEL2, GNA13, IRF9 and OSR2. In the HM subgroup we identified eight DMCs, annotated to the following genes: NOTCH4, RSPO1, LOXL3, HOXA5, LOC283999, HBA1 and ESAM. In the FM subgroup we found four DMCs, corresponding to LOC283999, FAM184A, as well as some unidentified (NA) genes. For instance, the gene LOC283999 showed an opposite direction in DNA methylation change by age at the FM subgroup; this may have biological consequences. Older men may have less sperm cells that are fully methylated at the corresponding site (cg09445803); a ten-year increase in age was associated with a 2.7% decrease in DNA methylation, or a drop from 82.56% to 79.86% (Table 3). As indicated in our methods section, we controlled for BMI, patient status and multiple testing. In our sensitivity analyses (excluding outliers or patients) and in our extended models (including additional co-factors) our results remained consistent with the main analysis (data not shown).

Figure 2. Volcano plots of 450K data by age in sperm. Volcano plots of our age-association study on the 450K array. X-axis: Delta β-values (Δβ) representing age-associated changes in DNA methylation (after linear regression, adjusted for BMI and patient status). Y-axis: logarithmic transformation of the adjusted p-value (BH-method). Dashed line: indicates where the adjusted p-value is 0.05; dots above the dashed line represent CpG sites where sperm DNA methylation is significantly associated with male age. (A) All sites of the 450K are included (n = 482,287) and dots above the dashed line are age-associated DMCs (n = 14,622). (B–D) Volcano plots by subgroup of DNA methylation. Colored dots: significant age-related DMCs mapped to genes selected by their potential role in inheritance of ASD from father to child; these include: OTX1, PRDM16, PTPRN2, B4GALNT4, KCNQ1, KCNQ1OT1, DLGAP2, PLAGL1, GNAS, GRB10, MAGEL2, CDH24 and FBRSL1. Each colored dot represents one DMC; multiple dots with same color are allocated to the same gene.

Table 3. Alterations at the top 90 most significant DMCs by age in sperm.

SG	Probe ID	Delta M	Mean M	p-value	Mean β	Delta β	Gene	Chr	Chr location
<0.2	cg16038003	−0.139	−2.434	2e-08	0.169	−0.023	HOXA11; HOXA11AS	7	chr7:27227520-27229043
<0.2	cg07834682	−0.099	−2.274	3.2e-07	0.179	−0.017	SIX2	2	chr2:45231211-45231482
<0.2	cg11207353	−0.104	−2.834	3.8e-07	0.131	−0.017	LRRC25	19	NA
<0.2	cg01298678	−0.132	−2.994	4e-07	0.123	−0.023	LHX2	9	chr9:126773246-126780953
<0.2	cg06490225*	−0.128	−2.693	4.2e-07	0.146	−0.021	LHX1*	17	chr17:35291899-35300875
<0.2	cg03861097	−0.138	−3.850	4.3e-07	0.073	−0.023	NA	21	chr21:34395128-34400245
<0.2	cg08178072	−0.111	−2.149	5e-07	0.193	−0.019	NA	2	chr2:45164561-45166567
<0.2	cg00896370	−0.106	−2.689	5.4e-07	0.143	−0.018	ZIC4	3	chr3:147115764-147116421
<0.2	cg01447831*	−0.158	−3.373	6.3e-07	0.102	−0.027	NA*	7	chr7:155246390-155251955
<0.2	cg22566906	−0.103	−2.292	6.3e-07	0.177	−0.018	GRASP	12	chr12:52400467-52401696
<0.2	cg25156915	−0.129	−4.268	7.2e-07	0.055	−0.018	NA	4	chr4:174427891-174428192
<0.2	cg16757281*	−0.109	−4.821	8.5e-07	0.038	−0.010	C16orf13*	16	chr16:682634-687106
<0.2	cg04856022	−0.100	−2.603	8.7e-07	0.148	−0.017	PPT2	6	chr6:32121829-32122529
<0.2	cg16872071	−0.094	−2.500	9.4e-07	0.158	−0.016	RALGDS	9	chr9:135995969-135996954
<0.2	cg24764168s	−0.143	−2.199	1.2e-06	0.196	−0.019	NA	6	chr6:1381743-1385211
<0.2	cg14699734	−0.085	−2.491	1.7e-06	0.157	−0.014	NA	19	chr19:5803734-5806023
<0.2	cg02675859	−0.102	−3.796	1.7e-06	0.072	−0.013	LOC283999	17	chr17:76228110-76228380
<0.2	cg10073842	−0.127	−2.573	1.8e-06	0.156	−0.022	MAGEL2	15	NA
<0.2	cg21619325	−0.095	−2.725	2.4e-06	0.138	−0.015	OSR1	2	chr2:19560963-19561650
<0.2	cg00260116	−0.079	−2.472	2.4e-06	0.158	−0.013	C7orf52	7	chr7:100815484-100816995
<0.2	cg08788246	−0.097	−3.373	3.1e-06	0.094	−0.015	ZNF74	22	chr22:20759743-20760923
<0.2	cg08940570	−0.125	−2.852	3.3e-06	0.133	−0.022	LOXL3; DOK1	2	chr2:74781494-74782685
<0.2	cg17465423	−0.105	−2.530	3.7e-06	0.157	−0.018	ZNF385A	12	chr12:54784900-54785238
<0.2	cg09600715	−0.086	−2.188	3.8e-06	0.186	−0.015	NA	7	chr7:27227520-27229043
<0.2	cg19556208	−0.110	−3.334	4e-06	0.098	−0.017	IRF9	14	NA
<0.2	cg18911395	−0.193	−3.374	4.2e-06	0.112	−0.028	LHX5	12	chr12:113908887-113910681
<0.2	cg17413194	−0.125	−2.593	4.2e-06	0.154	−0.022	GNA13	17	chr17:63051893-63053355
<0.2	cg15287594	−0.089	−3.758	4.3e-06	0.073	−0.010	FITM2	20	chr20:42939450-42940043
<0.2	cg03157027	−0.107	−3.804	5e-06	0.073	−0.017	OSR2	8	chr8:99960497-99961438
<0.2	cg04479580	−0.100	−2.315	5.2e-06	0.176	−0.017	NA	12	chr12:6438272-6438931
(0.2–0.8)	cg02991082	−0.086	0.009	4.7e-10	0.502	−0.010	TSC22D3	X	chrX:106959378-106959914
(0.2–0.8)	cg02494945	−0.098	−1.762	1.1e-08	0.234	−0.016	TBX4	17	chr17:59531723-59535254
(0.2–0.8)	cg26854298	−0.100	0.250	2e-08	0.542	−0.008	NA	4	chr4:174443365-174443948
(0.2–0.8)	cg11671335	−0.104	0.819	7.1e-08	0.634	−0.007	NOTCH4	6	chr6:32163292-32164383
(0.2–0.8)	cg03767475	−0.090	1.785	8.7e-08	0.770	−0.004	COMP	19	chr19:18899037-18902284
(0.2–0.8)	cg01862311s	−0.099	1.156	8.7e-08	0.685	−0.006	NA	8	chr8:26305853-26306825
(0.2–0.8)	cg15032957	−0.061	1.985	8.7e-08	0.796	−0.004	TLE2	19	chr19:3035638-3035872
(0.2–0.8)	cg24049629	−0.086	0.304	9.9e-08	0.551	−0.009	RASSF1	3	chr3:50377803-50378540
(0.2–0.8)	cg04508739	−0.088	1.142	2.3e-07	0.683	−0.006	NA	14	chr14:57264638-57265561
(0.2–0.8)	cg24114899s	−0.081	0.972	2.3e-07	0.659	−0.007	MGC12982; FOXD2	1	chr1:47902793-47905518
(0.2–0.8)	cg13054119	−0.115	−1.459	2.6e-07	0.276	−0.018	RSPO1	1	chr1:38099677-38100864
(0.2–0.8)	cg02021127	−0.100	1.404	2.6e-07	0.719	−0.005	PIP5KL1	9	chr9:130692839-130693331
(0.2–0.8)	cg04044203	−0.109	0.059	2.7e-07	0.510	−0.010	ESAM	11	chr11:124632063-124633239
(0.2–0.8)	cg10836509	−0.083	−0.834	3.1e-07	0.363	−0.012	NA	1	chr1:226297287-226298586
(0.2–0.8)	cg23674882	−0.117	−0.602	3.2e-07	0.403	−0.012	LOXL3	2	chr2:74781494-74782685
(0.2–0.8)	cg05269451s	−0.103	−1.157	3.2e-07	0.317	−0.014	NA	2	chr2:172963623-172964135
(0.2–0.8)	cg03107888	−0.084	−0.677	3.2e-07	0.388	−0.012	HOXA13	7	chr7:27238690-27240311
(0.2–0.8)	cg06097213s	−0.080	−1.002	3.2e-07	0.337	−0.013	LOC285830	6	chr6:29716468-29717158
(0.2–0.8)	cg20437892	−0.091	−0.046	3.2e-07	0.492	−0.011	NA	2	NA
(0.2–0.8)	cg00215611	−0.088	0.063	3.2e-07	0.511	−0.009	NA	2	chr2:219860927-219861242
(0.2–0.8)	cg19701577	−0.131	−0.824	3.2e-07	0.371	−0.012	HOXA5	7	chr7:27182613-27185562
(0.2–0.8)	cg10919107cr	−0.094	0.201	3.2e-07	0.534	−0.009	PMEPA1	20	chr20:56227252-56227687
(0.2–0.8)	cg04730768	−0.077	0.710	3.2e-07	0.618	−0.007	CHAT	10	chr10:50817095-50817309
(0.2–0.8)	cg23327992	−0.079	−0.472	3.2e-07	0.421	−0.012	IRF9	14	NA
(0.2–0.8)	cg20774552	−0.057	0.707	3.3e-07	0.619	−0.006	NA	11	chr11:46410921-46414687
(0.2–0.8)	cg20005923	−0.102	−0.365	3.6e-07	0.440	−0.011	LOC283999	17	chr17:76228110-76228380
(0.2–0.8)	cg02825052	−0.076	1.115	3.8e-07	0.681	−0.006	PICALM	11	NA
(0.2–0.8)	cg01704105	−0.114	−1.250	3.8e-07	0.305	−0.016	HBA1	16	chr16:226173-227254
(0.2–0.8)	cg00176496	−0.075	0.579	3.8e-07	0.597	−0.008	PITPNM3	17	NA
(0.2–0.8)	cg01344797cr	−0.067	0.977	4e-07	0.661	−0.005	PSD	10	chr10:104168558-104169153
>0.8	cg02882504s	−0.085	2.748	8.7e-08	0.866	−0.002	NA	16	chr16:3238805-3239492
>0.8	cg02680566	−0.083	2.100	2.3e-07	0.806	−0.004	OSR1	2	chr2:19560963-19561650
>0.8	cg08604523s	−0.089	2.171	4e-07	0.812	−0.003	MDK	11	chr11:46406904-46407441
>0.8	cg01943692	−0.077	2.409	7.7e-07	0.837	−0.003	NA	1	chr1:26686516-26687281
>0.8	cg00019678	−0.065	2.564	1.1e-06	0.852	−0.003	C12orf34	12	chr12:110151327-110152758
>0.8	cg09445803	−0.101	2.325	1.2e-06	0.826	−0.003	LOC283999	17	chr17:76228110-76228380
>0.8	cg12001078	0.064	2.708	1.4e-06	0.864	0.010	DAPP1	4	NA
>0.8	cg18672716	0.144	3.521	1.8e-06	0.907	0.024	NA	2	NA
>0.8	cg24321297	−0.046	2.627	2.8e-06	0.859	−0.002	ST6GALNAC2	17	chr17:74580974-74582396
>0.8	cg24888609s	0.152	3.321	3.1e-06	0.894	0.026	FAM184A	6	NA
>0.8	cg21483922	−0.067	2.220	3.1e-06	0.820	−0.003	SAMD14	17	chr17:48206663-48207601
>0.8	cg05525368	−0.062	2.950	3.8e-06	0.883	−0.002	CDC20	1	chr1:43824134-43825059
>0.8	cg11433866s	−0.070	2.122	3.8e-06	0.809	−0.003	NA	12	chr12:115124729-115125152
>0.8	cg01405985	−0.107	2.441	4.2e-06	0.835	−0.002	NA	19	chr19:36347044-36348101
>0.8	cg24073074	−0.069	2.175	4.9e-06	0.815	−0.003	WIPI1	17	NA
>0.8	cg04078221	−0.064	2.042	5.2e-06	0.801	−0.003	NA	1	chr1:47899125-47899398
>0.8	cg18231690	−0.075	2.092	5.3e-06	0.805	−0.004	ZNF592	15	chr15:85291079-85291697
>0.8	cg21690489	−0.063	2.621	5.6e-06	0.857	−0.003	CREB3L1	11	NA
>0.8	cg08692104	−0.055	3.365	6.3e-06	0.910	−0.001	ZNF823	19	chr19:11849359-11849796
>0.8	cg11277190cr	0.093	2.905	6.4e-06	0.876	0.015	FAM184A	6	NA
>0.8	cg21574186	0.079	3.519	6.4e-06	0.916	0.009	UNC5C	4	NA
>0.8	cg00471142	−0.059	3.040	6.5e-06	0.889	−0.002	LRFN1	19	chr19:39804621-39805954
>0.8	cg01109643	−0.083	2.152	6.6e-06	0.810	−0.003	NA	13	chr13:113548643-113549127
>0.8	cg24842142	−0.069	2.175	6.9e-06	0.815	−0.003	RNF39	6	chr6:30042918-30043500
>0.8	cg14101976	0.081	4.098	7.3e-06	0.942	0.010	UNC5C	4	NA
>0.8	cg12391323s	−0.046	2.435	7.5e-06	0.842	−0.003	NA	5	chr5:139027443-139030219
>0.8	cg05878073	−0.067	3.194	7.6e-06	0.899	−0.002	ABCD4	14	chr14:74769366-74769815
>0.8	cg11338156	−0.078	2.657	7.9e-06	0.858	−0.003	NA	10	NA
>0.8	cg13244241	−0.082	3.833	8.2e-06	0.931	−0.001	NA	10	chr10:102489343-102491011
>0.8	cg00730441	−0.061	2.187	8.4e-06	0.817	−0.003	TBX2	17	chr17:59485573-59485780
Probes related to DMCs by age were first sorted by subgroups (UM: β-value <0.2; HM: β-value (0.2–0.8); FM: β-value >0.8), then by adjusted p-values (Benjamini-Hochberg method; FDR <0.05). Only the top 30 most significant results by each subgroup (SG) are presented. Delta-M and Delta-β: estimated changes by age in M-value and in β-value, respectively. Annotated gene names and chromosome locations are shown according to the UCSC Genome Browser. *Probe related to CpG Islands; crcross-reactive probe; sSNP; (bold) direction is opposite (e.g., hypomethylated in FM subgroup); (underlined) highest magnitude in DNA methylation changes (Delta-M is higher than 0.1).

Genomic allocations, island content and other functional characteristics of age-associated DMCs or annotated genes

A Miami plot was created of all significant age-related DMCs across the chromosomes, separated by hypermethylated sites and hypomethylated sites (Figure 3). A remarkable observation is a general hypomethylation at sites annotated to imprinted genes. Next, we did a parallel analysis of all 14,622 DMCs in terms of their CpG content and neighborhood content (island, shore, shelf, open sea). CpG islands and open sea areas are the most frequent targets on the microarray, accounting for 30.91% and 36.28% of all analyzed CpG sites, respectively (Supplementary Table 1, Figure 4A). We detected limited age-related DMCs at islands; accounting for nearly 5% (n = 722) of the DMCs (Figure 4B), and only 0.48% of all analyzed sites (which is in large contrast to the 30.9% in the complete array) (Supplementary Tables 1 and 2). A comparison of increase (or hypermethylation) versus decrease (hypomethylation) in numbers and in ratios - by age, by island and by neighborhood content - are shown in Supplementary Table 2. Only a small difference was seen between the prevalence of negative and positive associations by age (55.7% versus 44.3 %, respectively); ratio is 1.26 if all DMCs are included, but it is 2.38 if only UM sites are considered (Supplementary Table 2). Still, an overall ratio of 1.26 at CpG islands is modest if compared to a high ratio of 24.8 at shores (hence, 24.8 times more negative associations than positive associations are measured at shores). An opposite but small ratio (0.40) is seen at open seas (or 2.5 times more positive associations than negative associations).

Figure 3. Chromosomal distribution of age-associated DMCs in sperm. Miami plot of age-associated differentially methylated CpG sites (DMCs), shown by direction of change (Delta β >0 relates to an increase in number of sperm cells that are methylated, Delta β <0 means a decrease in number of sperm cells that are methylated). Dashed line: no change (Δβ = 0). Delta β-values are calculated from M-values, as shown in the methods section $\left(\text{Δβ}=\frac{2^{({\text{M}}_0+\text{delta M})}}{1+2^{({\text{M}}_0+\text{delta M})}}-\frac{2^{{\text{M}}_0}}{1+2^{{\text{M}}_0}}\right)$. Significant DMCs correspond to FDR <0.05. Blue dots are sites allocated to imprinted genes, listed in the Geneimprint database.

Figure 4. Island content of CpGs within 450K array and age-associated DMCs. (A) Percentages of CpGs by island content are displayed with respect to the 450K (n = 482,287). (B) Percentages of CpGs by island content are displayed with respect to the number of significant age-related DMCs (n = 14,622). (C–E) By DNA methylation subgroup (Abbreviations: UM: unmethylated, mean β-value <0.2; HM: hemi-methylated, 0.20 ≤ mean β-value ≤0.80; FM: fully methylated, mean β-value>0.80).

We further examined variations in DNA methylation in relation to CpG island content by our three subgroups of DNA methylation (Supplementary Table 2 and Figure 4C–4E). Island associated DMCs were predominantly at unmethylated sites (UM), at which 13% of the age-related DMCs could be detected. Calculated ratios indicate that the direction of age-related changes is mostly negative in UM and positive in FM. Out of the 325 UM DMCs at islands, age had a positive effect on a small subset of 96 sites. And, out of the 117 FM DMCs at islands, 36 sites showed a decrease in DNA methylation. These sites -opposite to the characteristics of these subgroups (being low or high in DNA methylation, respectively)- may be interesting in the context of potential imprinting by parent of origin (discussed below). Most FM sites at shores and at shelves showed a decrease in methylation by age (approximately 95%, 1,611 sites out of 1,700 at shore sites; 81%, 776 out of 961 at shelf sites, respectively). Next, we explored functional genomic allocations (e.g., in subgroups such as promoter, body, 3′UTR, and intergenic) (Supplementary Table 3 and Figure 5). Overall, distributions of age-associated DMCs by functional allocations did not differ substantially from what was expected from the 450K array; although a higher representation of promotors within our DMCs was seen at the UM subgroup (28%) versus the FM subgroup (13%). Gene Ontology (GO) term enrichment analysis using all DMCs showed 11 significant GO terms related to male age (Supplementary Figure 8). Specifically, the following terms were captured with the highest enrichment: cell morphogenesis (biological process), and plasma membrane protein complex (cellular component).

Figure 5. Functional genomic distribution of CpGs within 450K array and age-associated DMCs. (A) Percentages of CpGs by functional genomic region are displayed; by promoter, 1stExon, 5′UTR, 3′UTR, body, and intergenic regions. CpG sites allocated to multiple regions are reported as a separate category. (A) Percentages with respect to the 450K (n = 482,287). (B) Percentages are displayed with respect to the number of significant age-related DMCs (n = 14,622). (C–E) By DNA methylation subgroup (Abbreviations: UM: unmethylated, mean β-value <0.2; HM: hemi-methylated, 0.20 ≤ mean β-value ≤0.80; FM: fully methylated, mean β-value >0.80).

Identifying age-related DMCs linked to imprinted genes and candidate ICRs

We screened our findings on age-related DMCs for their potential role in imprinting (Figure 1). First, imprinted genes were defined using Geneimprint, with a list of 227 imprinted genes in human; including 104 predicted and 123 confirmed imprinted genes [18]. Out of the 14,622 age-related DMCs, we found 271 DMCs (1.85%) that could be mapped to 95 (predicted) imprinted genes [17]; these genes are listed by subgroup in Figure 6. Because our 450K array includes 215 (potentially) imprinted genes, this means that 44% of these imprinted genes theoretically present on the array are sensitive to influences from ageing. Out of the 95 age-related imprinted genes, 49 are maternally expressed (or paternally imprinted), 42 are paternally expressed (or maternally imprinted), and imprinting is isoform dependent in four of the identified genes (Supplementary Table 4). Second, we extended our approach using a database published by Jima et al. [17], which includes a genomic map of 1,488 putative Imprint Control Regions (ICRs) or so-called Imprintome. Of our 14,622 DMCs, 747 DMCs could be linked to 380 candidate ICRs (Supplementary Table 5). More precisely, out of these 747 ICR-associated DMCs, 152 DMCs were allocated to 94 ICRs with paternal origin of methylation and 74 genes were identified; 193 DMCs were linked to 107 maternally methylated ICRs and 86 genes were found; and, the remaining 402 DMCs were associated to regions with unknown parent-of-origin methylation; corresponding to 179 ICRs or 158 genes (Supplementary Table 5). After a stringent selection of age-related genes based on a combined use of our findings above by Geneimprint list (n = 95) and by the Imprintome (n = 318), we selected 22 imprinted genes with previously reported ICRs (Figure 6). This set of genes corresponds to 94 age-associated DMCs; 79 DMCs are hypomethylated and 15 DMCs are hypermethylated (Table 4 and Figure 6). For instance, the highest magnitude at hypomethylated DMCs was measured at the PTPRN2 gene, being −28% per 10 years of ageing (cg18285788; Delta β = −0.028, p = 5.64e-04); and, the highest magnitude at hypermethylated DMCs was found at the KCNQ1 gene, being +10% per 10 years of ageing (cg17416793; Delta β = +0.010, p = 0.0127). In our search to define a set of age-related biomarkers of inheritance we could classify these 22 imprinted genes into the following categories: seven genes are paternally imprinted (or maternally expressed) (H19, KCNQ1DN, PTPRN2, KCNQ1, SVOPL, B4GALNT4, FBRSL1; the last two are yet predicted imprinted genes); twelve genes are known to be maternally imprinted (or paternally expressed) (MAGEL2, DIRAS3, FAM50B, DLGAP2, PLAGL1, ZIM2, GLI3, DLK1, CDH24, RB1, SNRPN, KCNQ1OT1); and, in 3 genes imprinting is isoform dependent (GNAS, BLCAP, GRB10). Notably, we repeated our analyses after exclusion of reported SNPs and cross-reactive probes [19, 20]. This test did not significantly change our results, with two exceptions: ZIM2 and C6orf145 (indicated in Figure 6 and Tables 3 and 4).

Figure 6. Imprinted genes affected by age in sperm and our approach for biomarker selection to predict autism spectrum disorders in offspring. A summary of (predicted) imprinted genes linked to significant age-associated DMCs is ordered by expressed allele, and significance (results with smallest p-values are on top). Upper frame: 95 (predicted) imprinted genes linked to 271 DMCs within DNA methylation subgroups (Abbreviations: UM: unmethylated, mean β-value <0.2; HM: hemi-methylated, 0.20≤ mean β-value ≤0.80; FM: fully methylated, mean β-value >0.80). Lower frame: out of the 95 (predicted) imprinted genes (upper frame) 22 genes have been mapped to ICRs. In bold, genes involved in ASD. Right frame: summary of 27 ASD-related imprinted genes (due to probe cross-reactivity, the gene C6orf145 with a single DMC was not withheld). The following superscripts are used if at least one DMC had the following characteristic: *, at CpG Islands; ^PR, at promotor region; ↓ or ↑, an opposite direction in DNA methylation change was measured (increase or decrease in DNA methylation, at UM and FM methylated DMCs, respectively); ^s, at SNPs; ^cr, at a cross-reactive probe. M, maternally expressed genes; P, paternally expressed genes; I, isoform dependent.

Table 4. Age-related DMCs and 22 allocated genes in sperm linked to ICRs and imprinting.

SG	EA	Probe ID	Delta β	Mean β	p-value	Chr	Gene
<0.2	I	cg21809160	−0.027	0.102	6.65e-05	20	GNAS
<0.2	I	cg24058407	−0.017	0.190	2.61e-04	20	GNAS
<0.2	I	cg11399589	−0.017	0.139	0.0021	20	BLCAP
<0.2	I	cg17696847	−0.008	0.079	0.0107	20	GNAS
<0.2	I	cg27006764	−0.001	0.035	0.0334	7	GRB10
<0.2	I	cg13591710	−0.016	0.168	0.0453	20	BLCAP
<0.2	I	cg11948874	−0.010	0.104	0.0482	20	BLCAP
<0.2	M	cg07439128	−0.017	0.067	2.35e-05	11	KCNQ1DN
<0.2	M	cg04937416	−0.015	0.091	3.17e-05	7	PTPRN2
<0.2	M	cg10798664	−0.012	0.179	1.40e-04	11	B4GALNT4*
<0.2	M	cg18285788	−0.028	0.054	5.64e-04	7	PTPRN2
<0.2	M	cg26461944	−0.007	0.139	0.0016	11	B4GALNT4*
<0.2	M	cg01923099	−0.020	0.122	0.0017	11	KCNQ1DN
<0.2	M	cg16083838	−0.020	0.087	0.0040	11	KCNQ1DN
<0.2	M	cg03983213	−0.009	0.074	0.0042	7	PTPRN2
<0.2	M	cg22675922	−0.005	0.060	0.0068	11	KCNQ1DN
<0.2	M	cg10726517	−0.003	0.050	0.0101	11	B4GALNT4*
<0.2	M	cg13538517	−0.017	0.150	0.0119	7	PTPRN2
<0.2	M	cg03481077	−0.014	0.193	0.0120	11	B4GALNT4*
<0.2	M	cg04876474	−0.002	0.043	0.0128	11	KCNQ1DN
<0.2	M	cg17682432	−0.003	0.073	0.0229	11	B4GALNT4*
<0.2	M	cg08242024	−0.007	0.151	0.0330	7	PTPRN2
<0.2	P	cg10073842	−0.022	0.156	1.84e-06	15	MAGEL2
<0.2	P	cg22872376	−0.014	0.100	1.20e-04	15	MAGEL2
<0.2	P	cg23076194	−0.011	0.161	0.0029	1	DIRAS3
<0.2	P	cg19447496	−0.011	0.190	0.0225	13	RB1
<0.2	P	cg23985641	−0.010	0.177	0.0473	6	FAM50B
(0.2–0.8)	I	cg07964163	−0.004	0.782	6.42e-05	20	GNAS
(0.2–0.8)	I	cg24214471	−0.006	0.681	2.87e-04	20	BLCAP
(0.2–0.8)	I	cg21733794	−0.008	0.555	6.94e-04	20	BLCAP
(0.2–0.8)	I	cg04820254	−0.005	0.723	0.0011	20	BLCAP
(0.2–0.8)	I	cg21045560	−0.011	0.351	0.0016	20	BLCAP
(0.2–0.8)	I	cg20569652	−0.009	0.282	0.0053	20	BLCAP
(0.2–0.8)	I	cg17658854	−0.011	0.207	0.0094	20	GNAS
(0.2–0.8)	I	cg01565918	−0.010	0.340	0.0128	20	GNAS
(0.2–0.8)	I	cg14235271	−0.011	0.219	0.0365	20	GNAS
(0.2–0.8)	M	cg21996245	−0.016	0.202	6.08e-05	11	B4GALNT4*
(0.2–0.8)	M	cg20846508	−0.012	0.390	1.29e-04	11	B4GALNT4*
(0.2–0.8)	M	cg24221919	−0.006	0.589	2.18e-04	7	PTPRN2
(0.2–0.8)	M	cg18628367	−0.012	0.289	4.13e-04	7	PTPRN2
(0.2–0.8)	M	cg03371125	−0.010	0.345	6.71e-04	11	KCNQ1
(0.2–0.8)	M	cg04666029	−0.004	0.641	0.0021	11	KCNQ1
(0.2–0.8)	M	cg20533553	−0.012	0.221	0.0068	11	KCNQ1
(0.2–0.8)	M	cg19698309	−0.010	0.361	0.0094	11	KCNQ1
(0.2–0.8)	M	cg20482223	−0.006	0.513	0.0177	7	SVOPL
(0.2–0.8)	M	cg19713140	−0.010	0.374	0.0378	7	PTPRN2
(0.2–0.8)	M	cg21231189	0.006	0.776	0.0403	7	PTPRN2
(0.2–0.8)	M	cg01100465cr	−0.005	0.642	0.0408	7	PTPRN2
(0.2–0.8)	M	cg15094119	0.004	0.772	0.0463	7	PTPRN2
(0.2–0.8)	M	cg15971656	−0.006	0.374	0.0481	11	B4GALNT4*
(0.2–0.8)	P	cg20808078	−0.004	0.747	2.87e-06	1	DIRAS3
(0.2–0.8)	P	cg25135755	−0.007	0.631	3.14e-06	15	MAGEL2
(0.2–0.8)	P	cg09834049	−0.012	0.291	5.29e-06	14	CDH24*
(0.2–0.8)	P	cg01152488	−0.011	0.440	9.06e-05	15	MAGEL2
(0.2–0.8)	P	cg03439898cr	0.009	0.770	0.0025	8	DLGAP2
(0.2–0.8)	P	cg20076070	−0.011	0.413	0.0056	8	DLGAP2
(0.2–0.8)	P	cg21113768	−0.008	0.464	0.0069	6	PLAGL1
(0.2–0.8)	P	cg08082351	−0.005	0.708	0.0106	8	DLGAP2
(0.2–0.8)	P	cg26939721	0.005	0.702	0.0147	15	SNRPN
(0.2–0.8)	P	cg12818159	0.006	0.785	0.0272	8	DLGAP2
(0.2–0.8)	P	cg02566775	−0.005	0.577	0.0352	6	PLAGL1
(0.2–0.8)	P	cg09212014	−0.007	0.329	0.0375	14	DLK1
(0.2–0.8)	P; M	cg03654058	−0.012	0.331	3.96e-04	11	KCNQ1OT1; KCNQ1
(0.2–0.8)	P; M	cg04762676	−0.003	0.764	0.0271	11	KCNQ1OT1; KCNQ1
>0.8	I	cg09993814	−0.003	0.830	5.19e-04	20	BLCAP
>0.8	I	cg10546626	−0.002	0.891	0.0034	20	GNAS
>0.8	I	cg17820025cr	−0.002	0.832	0.0455	20	BLCAP
>0.8	M	cg05821571	−0.003	0.825	5.53e-04	7	PTPRN2
>0.8	M	cg04799270	−0.001	0.924	0.0019	7	PTPRN2
>0.8	M	cg15012939	−0.003	0.855	0.0027	7	PTPRN2
>0.8	M	cg02855778	−0.002	0.904	0.0034	7	PTPRN2
>0.8	M	cg02773779	−0.002	0.866	0.0039	7	PTPRN2
>0.8	M	cg19100996	−0.002	0.896	0.0061	12	FBRSL1*
>0.8	M	cg22172494	0.005	0.859	0.0078	11	H19
>0.8	M	cg27629384	0.008	0.952	0.0119	7	PTPRN2
>0.8	M	cg17416793	0.010	0.942	0.0127	11	KCNQ1
>0.8	M	cg05926314	0.005	0.937	0.0129	7	PTPRN2
>0.8	M	cg19764489	0.007	0.906	0.0164	11	KCNQ1
>0.8	M	cg09350411	−0.002	0.910	0.0222	7	PTPRN2
>0.8	M	cg12001456	0.005	0.922	0.0234	7	PTPRN2
>0.8	M	cg08376924	0.005	0.913	0.0244	7	PTPRN2
>0.8	M	cg27050114cr	0.008	0.884	0.0264	11	KCNQ1
>0.8	M	cg06423822	−0.002	0.865	0.0274	7	PTPRN2
>0.8	M	cg03647659	−0.001	0.955	0.0320	11	B4GALNT4*
>0.8	M	cg24652817	−0.002	0.861	0.0363	7	PTPRN2
>0.8	P	cg07599819s	−0.002	0.899	0.0012	19	ZIM2
>0.8	P	cg03652257	−0.003	0.818	0.0033	7	GLI3
>0.8	P	cg24257495	0.005	0.930	0.0038	8	DLGAP2
>0.8	P	cg03156547	−0.002	0.839	0.0065	14	CDH24*
>0.8	P	cg06450373cr	0.008	0.889	0.0087	14	CDH24*
>0.8	P	cg01983373	−0.002	0.890	0.0178	14	DLK1
>0.8	P	cg18100008	−0.001	0.920	0.0228	7	GLI3
>0.8	P	cg03646329	−0.003	0.822	0.0243	13	RB1
>0.8	P	cg11882053	−0.003	0.827	0.0256	13	RB1
Significant age-associated DMCs and allocated genes (n = 22) are shown based on Geneimprint and the Imprintome (see methods). Probe IDs, estimated changes of β-values by age, annotated imprinted genes, and expressed allele are indicated. *Predicted imprinted gene; (EA): expressed alleles: (P) Paternally, (M) Maternally, or (I) Isoform Dependent; crcross-reactive probe; sSNP. Results are ordered by subgroup (SG): unmethylated, UM (<0.2); hemi-methylated, HM ((0.2–0.8)); fully methylated, FM (>0.8); and, sorted by adjusted p-value (FDR <0.05). Note: If DMCs were mapped to multiple genes, only the ICR-related (predicted) imprinted gene is reported.

Comparison of the current findings on age-related DMCs with results from similar studies

In a recent review by the research group of Haaf, 2,355 genes have been reported in sperm with age affected DMRs [11]. Considering the current study results and those found by four other observational studies, we retrieved 2,098 age-associated genes in sperm. Although similar techniques (Illumina platforms) were used, study populations and study conditions differed [9–12]. A Venn diagram illustrates the numbers of overlapping age-affected genes identified by each report (Figure 7). We counted 929 genes that were found by at least two studies, 83 genes that were found by at least three studies, seven genes that were found by at least four studies, and one gene that was found by all five studies. (Supplementary Table 6). In brief, six genes found by four studies are SLC22A18AS, C7orf50, UTS2R, BEGAIN, GRIN1, and PCDH15; and, a single gene identified by all is DLGAP2. Apart from C7orf50, all genes recurrently found in age-association studies over the last couple of years have been linked to the development of autism (discussed below) [21–26]. The role of UTS2R (or GPR14) in autism is unclear. It plays a role in various brain functions, but it has been considered as a candidate gene for autism [27, 28]. Overall, our findings are in line with Bernhardt et al.’s conclusion that age-induced DNA methylation alterations in sperm may contribute to the development of neurodevelopmental disorders in offspring [11].

Figure 7. Comparison of our set of age-related DMCs with earlier published datasets. Venn diagram showing the number of genes that have been identified by different authors as being differentially methylated by age in sperm; starting from a recent review by Bernhardt et al. [11], where 2,355 genes have been selected. Next to our results, the following publications have been included in this comparison: Bernhardt et al., Jenkins et al., Laurentino et al., and Oluwayiose et al. [9–12]. Seven genes (center of diagram) were identified by at least four studies (DLGAP2, SLC22A18AS, C7orf50, UTS2R, BEGAIN, GRIN1, and PCDH15); two are known imprinted genes (DLGAP2 and SLC22A18AS).

Assessment of a potential role in autism development of genes linked to age-related DMCs in sperm

Based on our results and reports by others, age-related DMCs in sperm were frequently seen at genes involved in autism. However, no earlier studies verified if these DNA methylation marks were located at sites important in epigenetic inheritance. First, we verified which of the seven recurrently reported genes (see paragraph above) are prone to imprinting. Two of the seven genes are known to be imprinted: DLGAP2 has been reported as a paternally expressed gene, and SLC22A18AS has been listed as a maternally expressed gene [18]. The estimated effects by age based on the current study at five CpG sites allocated to DLGAP2 are illustrated (Figure 8 and Supplementary Tables 4 and 5), directions varied by site; while at the three CpG sites allocated to SLC22A18AS all were hypomethylated by age (Figure 8 and Supplementary Table 4). This suggests that these genes may play an important role in autism development in children born to fathers of an advanced age. Hence, changes in DNA methylation patterns at these genes could be considered as a potential predictor for autism in the next generation.

Figure 8. DNA methylation by age at CpGs linked to DLGAP2 and SLC22A18AS. Estimates of DNA methylation in β-values by age for all significant CpG sites annotated to DLGAP2 (A–E) and SLC22A18AS (F–H) imprinted genes are shown; DNA methylation changes at these genes have been identified by at least three other age-related studies. Fitted regression lines are shown. Regression models included potential confounding factors (BMI and patient status); these were corrected for multiple testing (BH-method). The following CpG sites were annotated to DLGAP2: (A) cg03439898 (Delta β = 0.0085, p-value = 0.0025), (B) cg20076070 (Delta β = −0.011, p-value = 0.0056), (C) cg08082351 (Delta β = −0.0046, p-value = 0.7083), (D) cg12818159 (Delta β = 0.0059, p-value = 0.0272), and (E) cg24257495 (Delta β = 0.0049, p-value = 0.0038). The following CpG sites were annotated to SLC22A18AS: (F) cg08222610 (Delta β = −0.0153, p-value = 6.158515e-06), (G) cg26874323 (Delta β = −0.0097, p-value = 1.840627e-05), (H) cg23335134 (Delta β = −0.005, p-value = 6.076208e-05).

When screening the 95 age-related imprinted genes based on the Geneimprint database (Supplementary Table 4), we found 28 genes that have been reported in autism development. When considering cross-reactivity of probes, the gene C6orf145 was excluded from our final list, resulting in 27 ASD-related imprinted genes (Figure 6). Eleven of these imprinted genes have been linked to a reported ICR (Figure 6, lower right panel); five of these are included in the SFARI database (see Methods): MAGEL2 (score 1; high confidence for ASD), GNAS, and DLGAP2 (score 2; strong candidate for ASD), GRB10 (score 3; suggested evidence for ASD), and FBRSL1 (unknown score). Based on other databases (including genes found in multiplex ASD families), the following six genes are also listed: B4GALNT4, KCNQ1, KCNQ1OT1, PLAGL1, CDH24, and PTPRN2 (Figure 6) [25, 29]. Finally, these eleven genes form a first set of age-related biomarkers (Figure 6). Based on important functional characteristics of epigenetic inheritance, current and earlier findings on age-related epigenetic marks in sperm and ASD risk, we ranked our findings by several characteristics (significance, magnitude of change, overlapping findings with earlier reports, multiple age-related CpGs within the same gene, promoter or island content) (Table 5). Except for the FBRSL1 gene, all 11 genes had a “score of interest” of 3 or higher. We used this numerical value (3) to further select for potential biomarkers within the remaining set of ASD-related imprinted genes without a yet established link to an ICR. Hence, additional genes with a “score of interest” of at least 3 are: OTX1, PRDM16, and SLC26A10. The latter is listed in the NCBI Gene database as a pseudogene; hence, we do not select for this gene in our final list of potential biomarker genes. Consequently, we found thirteen candidate imprinted genes as potential biomarkers for age-related inheritance of ASD from father to child. These are illustrated in Volcano plots by DNA methylation subgroups (Figure 2), and fitted regression lines are shown for each gene (Figures 8 and 9).

Table 5. ASD-associated (predicted) imprinted genes and scoring.

Gene	Chr	Probe ID	ICR	Co	Multi CpG	Top 90	Ma	Op	Is	Pr	Score of interest
MAGEL2	15	cg10073842	1	0	1	1	1	0	0	1	5
DLGAP2	8	cg03439898	1	1	1	0	0	0	1	0	4
GNAS	20	cg21809160	1	0	1	0	1	0	1	0	4
B4GALNT4	11	cg10798664	1	0	1	0	0	0	0	1	3
GRB10	7	cg27006764	1	0	0	0	0	0	1	1	3
KCNQ1	11	cg03371125	1	0	1	0	0	0	0	1	3
KCNQ1OT1; KCNQ1	11	cg03654058	1	0	1	0	0	0	0	1	3
OTX1	2	cg10487970	0	0	0	0	0	1	1	1	3
PLAGL1	6	cg21113768	1	0	1	0	0	0	0	1	3
PTPRN2	7	cg04937416	1	0	1	0	0	0	1	0	3
PRDM16	1	cg10588310	0	0	1	0	0	1	1	0	3
CDH24	14	cg03156547	1	0	1	0	0	1	0	0	3
SLC26A10	12	cg08177625	0	0	1	0	0	1	0	1	3
ATP10A	15	cg06066676	0	0	1	0	0	0	0	1	2
FBRSL1	12	cg19100996	1	0	0	0	0	1	0	0	2
HOXB3	17	cg10585948	0	0	1	0	0	0	0	1	2
ANO1	11	cg11058904	0	0	0	0	1	0	0	0	1
C6orf145	6	cg18815879cr	0	0	0	0	0	0	0	1	1
FOXG1	14	cg18299578	0	0	0	0	0	0	0	1	1
GLIS3	9	cg13804450	0	0	1	0	0	0	0	0	1
LMX1B	9	cg13466694	0	0	1	0	0	0	0	0	1
MAGI2	7	cg00110846	0	0	1	0	0	0	0	0	1
NTM	11	cg12079699	0	0	1	0	0	0	0	0	1
OBSCN	1	cg18477163	0	0	0	0	0	0	1	0	1
UBE3A	15	cg12060334	0	0	1	0	0	0	0	0	1
ADAMTS16	5	cg26892415	0	0	0	0	0	0	0	0	0
DDC	7	cg15001032	0	0	0	0	0	0	0	0	0
RNU5D	5	cg19107296	0	0	0	0	0	0	0	0	0
ASD-associated (predicted) imprinted genes and related probes. Genes were linked to ASD disorders according to previously published databases (see methods). “Score of interest” is given by the following criteria: (ICR) listed by others as being close to an ICR; (Co) confirmed by at least four studies (including the current one); (Multi CpG) age-association was found at more than one CpG of the gene reported; (Top 90) belongs to the most significant top 90 age-related DMCs; (Ma) belongs to the highest magnitude in change by age (Delta-M > 0.1); (Op) opposite direction in methylation change; (Is) located at CpG island; (Pr) located at promoter region. Note: “1” means at least one CpG was found with this characteristic; “0” means none of the DMCs or genes met our criterium. Last column is the sum of eight criteria. crGene linked to a cross-reactive probe. Genes are listed from highest to lowest scores. For each gene, probe with highest score is shown. Note: one gene (SLC26A10) is reported by the NCBI Gene database as a pseudogene in human.

Figure 9. DNA methylation by age at CpGs linked to imprinted genes and related to ASD. Estimates of DNA methylation in β-values by age of CpG sites mapped to our selected list of ASD-related imprinted genes (with exception of CpGs mapped to DLGAP2, which are shown in Figure 8). Fitted regression lines are shown. Our regression models included potential confounding factors (BMI and patient status), and we corrected for multiple testing (BH-method). (A–E) maternally expressed genes; (F–J) paternally expressed genes; (K, L) isoform dependent transcribed gene. If more than one CpG site per gene was significant, the one with the lowest p-value is shown.

Abbreviations

ASD: Autism Spectrum Disorders; BH: Benjamini-Hochberg; BMIQ: Beta Mixture Quantile; DMC: Differentially Methylation CpG Site; FM: Fully Methylated; GO: Gene Ontology; HM: Hemi-methylated; ICR: Imprint Control Region; SFARI: Simons Foundation Autism Reference Initiative; SNP: Single Nucleotide Polymorphisms; TIEGER: The Influence of the Environment on Gametic Epigenetic Reprogramming; UM: Unmethylated.

Author Contributions

EC, JD and AS developed the research question and wrote the manuscript. JD was equally involved in the interpretation and discussion of the research results. MB performed the data cleaning and initiated data analysis. EC performed the major statistical analysis. EC designed and performed post-hoc comparisons. CD contributed to the interpretation of the data and results on outcomes related to psychiatric disorders. AS and CH developed the epidemiological study design. SM developed the experimental study design and oversaw DNA methylation data generation. AS initiated the overall study hypothesis of the TIEGER study and supervised the overall interpretation of the data analysis. RLJ, CH, SM, PA and TP edited the manuscript. All authors have critically reviewed the manuscript and have given final approval for publication.

Acknowledgments

We acknowledge the patients of the TIEGER study, Erin White, M.Sc., for her clinical contributions, Carole Grenier, B.Sc., for outstanding laboratory assistance.

Conflicts of Interest

We declare no competing interests. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Ethical Statement and Consent

The TIEGER Study was performed with approval of the Duke University Institutional Review Board (reference number: Pro00036645). Informed consent was obtained from all participants.

Funding

This work was supported by a Health and the Environment award from the Duke Cancer Institute and the Duke Nicholas School of the Environment, a research grant from FWO (G0C8523N), co-funded by the FPS Health, Food Chain Safety and Environment, Belgium. This work was conducted in the framework of the European Partnership for the Assessment of Risks from Chemicals (PARC) and has received funding from the European Union’s Horizon Europe research and innovation program under Grant Agreement No 101057014. Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the Health and Digital Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.

References

• 1. Van Opstal J, Fieuws S, Spiessens C, Soubry A. Male age interferes with embryo growth in IVF treatment. Hum Reprod. 2021; 36:107–15. https://doi.org/10.1093/humrep/deaa256 [PubMed]
• 2. Katz-Jaffe MG, Parks J, McCallie B, Schoolcraft WB. Aging sperm negatively impacts in vivo and in vitro reproduction: a longitudinal murine study. Fertil Steril. 2013; 100:262–8.e1-2. https://doi.org/10.1016/j.fertnstert.2013.03.021 [PubMed]
• 3. Milekic MH, Xin Y, O'Donnell A, Kumar KK, Bradley-Moore M, Malaspina D, Moore H, Brunner D, Ge Y, Edwards J, Paul S, Haghighi FG, Gingrich JA. Age-related sperm DNA methylation changes are transmitted to offspring and associated with abnormal behavior and dysregulated gene expression. Mol Psychiatry. 2015; 20:995–1001. https://doi.org/10.1038/mp.2014.84 [PubMed]
• 4. Pan N, Lin LZ, Wang X, Guo CH, Jing J, Li XH. Association between paternal age at childbirth and autism spectrum disorder in offspring. Zhongguo Dang Dai Er Ke Za Zhi. 2022; 24:863–8. https://doi.org/10.7499/j.issn.1008-8830.2203146 [PubMed]
• 5. Miller B, Messias E, Miettunen J, Alaräisänen A, Järvelin MR, Koponen H, Räsänen P, Isohanni M, Kirkpatrick B. Meta-analysis of paternal age and schizophrenia risk in male versus female offspring. Schizophr Bull. 2011; 37:1039–47. https://doi.org/10.1093/schbul/sbq011 [PubMed]
• 6. Grosvenor LP, Croen LA, Lynch FL, Marafino BJ, Maye M, Penfold RB, Simon GE, Ames JL. Autism Diagnosis Among US Children and Adults, 2011-2022. JAMA Netw Open. 2024; 7:e2442218. https://doi.org/10.1001/jamanetworkopen.2024.42218 [PubMed]
• 7. Ha AS, Scott M, Zhang CA, Li S, Langroudi AP, Glover F, Basran S, Del Giudice F, Shaw GM, Eisenberg ML. Sociodemographic Trends and Perinatal Outcomes in Fathers 50 Years and Older. JAMA Netw Open. 2024; 7:e2425269. https://doi.org/10.1001/jamanetworkopen.2024.25269 [PubMed]
• 8. Vervoort I, Delger C, Soubry A. A multifactorial model for the etiology of neuropsychiatric disorders: the role of advanced paternal age. Pediatr Res. 2022; 91:757–70. https://doi.org/10.1038/s41390-021-01435-4 [PubMed]
• 9. Jenkins TG, Aston KI, Pflueger C, Cairns BR, Carrell DT. Age-associated sperm DNA methylation alterations: possible implications in offspring disease susceptibility. PLoS Genet. 2014; 10:e1004458. https://doi.org/10.1371/journal.pgen.1004458 [PubMed]
• 10. Laurentino S, Cremers JF, Horsthemke B, Tüttelmann F, Czeloth K, Zitzmann M, Pohl E, Rahmann S, Schröder C, Berres S, Redmann K, Krallmann C, Schlatt S, et al. A germ cell-specific ageing pattern in otherwise healthy men. Aging Cell. 2020; 19:e13242. https://doi.org/10.1111/acel.13242 [PubMed]
• 11. Bernhardt L, Dittrich M, Prell A, Potabattula R, Drummer C, Behr R, Hahn T, Schorsch M, Müller T, Haaf T. Age-related methylation changes in the human sperm epigenome. Aging (Albany NY). 2023; 15:1257–78. https://doi.org/10.18632/aging.204546 [PubMed]
• 12. Oluwayiose OA, Wu H, Saddiki H, Whitcomb BW, Balzer LB, Brandon N, Suvorov A, Tayyab R, Sites CK, Hill L, Marcho C, Pilsner JR. Sperm DNA methylation mediates the association of male age on reproductive outcomes among couples undergoing infertility treatment. Sci Rep. 2021; 11:3216. https://doi.org/10.1038/s41598-020-80857-2 [PubMed]
• 13. Jenkins TG, Aston KI, Cairns B, Smith A, Carrell DT. Paternal germ line aging: DNA methylation age prediction from human sperm. BMC Genomics. 2018; 19:763. https://doi.org/10.1186/s12864-018-5153-4 [PubMed]
• 14. Jenkins TG, James ER, Alonso DF, Hoidal JR, Murphy PJ, Hotaling JM, Cairns BR, Carrell DT, Aston KI. Cigarette smoking significantly alters sperm DNA methylation patterns. Andrology. 2017; 5:1089–99. https://doi.org/10.1111/andr.12416 [PubMed]
• 15. Soubry A, Guo L, Huang Z, Hoyo C, Romanus S, Price T, Murphy SK. Obesity-related DNA methylation at imprinted genes in human sperm: Results from the TIEGER study. Clin Epigenetics. 2016; 8:51. https://doi.org/10.1186/s13148-016-0217-2 [PubMed]
• 16. Shnorhavorian M, Schwartz SM, Stansfeld B, Sadler-Riggleman I, Beck D, Skinner MK. Differential DNA Methylation Regions in Adult Human Sperm following Adolescent Chemotherapy: Potential for Epigenetic Inheritance. PLoS One. 2017; 12:e0170085. https://doi.org/10.1371/journal.pone.0170085 [PubMed]
• 17. Jima DD, Skaar DA, Planchart A, Motsinger-Reif A, Cevik SE, Park SS, Cowley M, Wright F, House J, Liu A, Jirtle RL, Hoyo C. Genomic map of candidate human imprint control regions: the imprintome. Epigenetics. 2022; 17:1920–43. https://doi.org/10.1080/15592294.2022.2091815 [PubMed]
• 18. Jirtle R. Imprinted Genes by Species. 2025.
• 19. Naeem H, Wong NC, Chatterton Z, Hong MK, Pedersen JS, Corcoran NM, Hovens CM, Macintyre G. Reducing the risk of false discovery enabling identification of biologically significant genome-wide methylation status using the HumanMethylation450 array. BMC Genomics. 2014; 15:51. https://doi.org/10.1186/1471-2164-15-51 [PubMed]
• 20. Peters TJ, Buckley MJ, Chen Y, Smyth GK, Goodnow CC, Clark SJ. Calling differentially methylated regions from whole genome bisulphite sequencing with DMRcate. Nucleic Acids Res. 2021; 49:e109. https://doi.org/10.1093/nar/gkab637 [PubMed]
• 21. Abrahams BS, Arking DE, Campbell DB, Mefford HC, Morrow EM, Weiss LA, Menashe I, Wadkins T, Banerjee-Basu S, Packer A. SFARI Gene 2.0: a community-driven knowledgebase for the autism spectrum disorders (ASDs). Mol Autism. 2013; 4:36. https://doi.org/10.1186/2040-2392-4-36 [PubMed]
• 22. Kuwano Y, Kamio Y, Kawai T, Katsuura S, Inada N, Takaki A, Rokutan K. Autism-associated gene expression in peripheral leucocytes commonly observed between subjects with autism and healthy women having autistic children. PLoS One. 2011; 6:e24723. https://doi.org/10.1371/journal.pone.0024723 [PubMed]
• 23. Iourov IY, Vorsanova SG, Korostelev SA, Zelenova MA, Yurov YB. Long contiguous stretches of homozygosity spanning shortly the imprinted loci are associated with intellectual disability, autism and/or epilepsy. Mol Cytogenet. 2015; 8:77. https://doi.org/10.1186/s13039-015-0182-z [PubMed]
• 24. Potabattula R, Prell A, Dittrich M, Nava C, Depienne C, Bejaoui Y, El Hajj N, Hahn T, Schorsch M, Haaf T. Effects of paternal and chronological age on BEGAIN methylation and its possible role in autism. Aging (Albany NY). 2023; 15:12763–79. https://doi.org/10.18632/aging.205275 [PubMed]
• 25. Ruzzo EK, Pérez-Cano L, Jung JY, Wang LK, Kashef-Haghighi D, Hartl C, Singh C, Xu J, Hoekstra JN, Leventhal O, Leppä VM, Gandal MJ, Paskov K, et al. Inherited and De Novo Genetic Risk for Autism Impacts Shared Networks. Cell. 2019; 178:850–66.e26. https://doi.org/10.1016/j.cell.2019.07.015 [PubMed]
• 26. Bajaj V, Markandaya M, Krishna L, Kumar A. Paternal imprinting of the SLC22A1LS gene located in the human chromosome segment 11p15.5. BMC Genet. 2004; 5:13. https://doi.org/10.1186/1471-2156-5-13 [PubMed]
• 27. Ugur C, Sertcelik M, Uneri O, Senses Dinc G, Sekmen E, Solmaz E. Evaluation of serum urotensin-II levels of children with ADHD and autism spectrum disorder. Anatolian Journal of Psychiatry. 2018; 18:7.
• 28. Valentino F, Bruno LP, Doddato G, Giliberti A, Tita R, Resciniti S, Fallerini C, Bruttini M, Lo Rizzo C, Mencarelli MA, Mari F, Pinto AM, Fava F, et al. Exome Sequencing in 200 Intellectual Disability/Autistic Patients: New Candidates and Atypical Presentations. Brain Sci. 2021; 11:936. https://doi.org/10.3390/brainsci11070936 [PubMed]
• 29. Homs A, Codina-Solà M, Rodríguez-Santiago B, Villanueva CM, Monk D, Cuscó I, Pérez-Jurado LA. Genetic and epigenetic methylation defects and implication of the ERMN gene in autism spectrum disorders. Transl Psychiatry. 2016; 6:e855. https://doi.org/10.1038/tp.2016.120 [PubMed]
• 30. Jenkins TG, Aston KI, Cairns BR, Carrell DT. Paternal aging and associated intraindividual alterations of global sperm 5-methylcytosine and 5-hydroxymethylcytosine levels. Fertil Steril. 2013; 100:945–51. https://doi.org/10.1016/j.fertnstert.2013.05.039 [PubMed]
• 31. Murphy SK, Itchon-Ramos N, Visco Z, Huang Z, Grenier C, Schrott R, Acharya K, Boudreau MH, Price TM, Raburn DJ, Corcoran DL, Lucas JE, Mitchell JT, et al. Cannabinoid exposure and altered DNA methylation in rat and human sperm. Epigenetics. 2018; 13:1208–21. https://doi.org/10.1080/15592294.2018.1554521 [PubMed]
• 32. Schrott R, Acharya K, Itchon-Ramos N, Hawkey AB, Pippen E, Mitchell JT, Kollins SH, Levin ED, Murphy SK. Cannabis use is associated with potentially heritable widespread changes in autism candidate gene DLGAP2 DNA methylation in sperm. Epigenetics. 2020; 15:161–73. https://doi.org/10.1080/15592294.2019.1656158 [PubMed]
• 33. Chertkow-Deutsher Y, Cohen H, Klein E, Ben-Shachar D. DNA methylation in vulnerability to post-traumatic stress in rats: evidence for the role of the post-synaptic density protein Dlgap2. Int J Neuropsychopharmacol. 2010; 13:347–59. https://doi.org/10.1017/S146114570999071X [PubMed]
• 34. Rasmussen AH, Rasmussen HB, Silahtaroglu A. The DLGAP family: neuronal expression, function and role in brain disorders. Mol Brain. 2017; 10:43. https://doi.org/10.1186/s13041-017-0324-9 [PubMed]
• 35. Sánchez AM, Delgado MS, Ciurana SG, Medrano J, Poo-Llanillo ME, Ishida M, Moore GE, Platas II, Simón C, Monk D. Epigenetic Symmetry of DLGAP2: Pre-Implantation Maternal Methylation Switches to a Random Monoallelic Profile in Somatic Tissues. OBM Genetics. 2018; 2:026. https://doi.org/10.21926/obm.genet.1803026
• 36. Court F, Tayama C, Romanelli V, Martin-Trujillo A, Iglesias-Platas I, Okamura K, Sugahara N, Simón C, Moore H, Harness JV, Keirstead H, Sanchez-Mut JV, Kaneki E, et al. Genome-wide parent-of-origin DNA methylation analysis reveals the intricacies of human imprinting and suggests a germline methylation-independent mechanism of establishment. Genome Res. 2014; 24:554–69. https://doi.org/10.1101/gr.164913.113 [PubMed]
• 37. Carreras-Gallo N, Dwaraka VB, Jima DD, Skaar DA, Mendez TL, Planchart A, Zhou W, Jirtle RL, Smith R, Hoyo C. Creation and Validation of the First Infinium DNA Methylation Array for the Human Imprintome. bioRxiv. 2024; 2024.01.15.575646. https://doi.org/10.1101/2024.01.15.575646 [PubMed]
• 38. Garg P, Sharp AJ. Screening for rare epigenetic variations in autism and schizophrenia. Hum Mutat. 2019; 40:952–61. https://doi.org/10.1002/humu.23740 [PubMed]
• 39. Li J, Chen W, Li D, Gu S, Liu X, Dong Y, Jin L, Zhang C, Li S. Conservation of Imprinting and Methylation of MKRN3, MAGEL2 and NDN Genes in Cattle. Animals (Basel). 2021; 11:1985. https://doi.org/10.3390/ani11071985 [PubMed]
• 40. Chen H, Victor AK, Klein J, Tacer KF, Tai DJ, de Esch C, Nuttle A, Temirov J, Burnett LC, Rosenbaum M, Zhang Y, Ding L, Moresco JJ, et al. Loss of MAGEL2 in Prader-Willi syndrome leads to decreased secretory granule and neuropeptide production. JCI Insight. 2020; 5:e138576. https://doi.org/10.1172/jci.insight.138576 [PubMed]
• 41. Gigliucci V, Busnelli M, Santini F, Paolini C, Bertoni A, Schaller F, Muscatelli F, Chini B. Oxytocin receptors in the Magel2 mouse model of autism: Specific region, age, sex and oxytocin treatment effects. Front Neurosci. 2023; 17:1026939. https://doi.org/10.3389/fnins.2023.1026939 [PubMed]
• 42. HPA. Human Protein Atlas.
• 43. Costa CIS, da Silva Montenegro EM, Zarrei M, de Sá Moreira E, Silva IMW, de Oliveira Scliar M, Wang JYT, Zachi EC, Branco EV, da Costa SS, Lourenço NCV, Vianna-Morgante AM, Rosenberg C, et al. Copy number variations in a Brazilian cohort with autism spectrum disorders highlight the contribution of cell adhesion genes. Clin Genet. 2022; 101:134–41. https://doi.org/10.1111/cge.14072 [PubMed]
• 44. Leung YK, Ouyang B, Niu L, Xie C, Ying J, Medvedovic M, Chen A, Weihe P, Valvi D, Grandjean P, Ho SM. Identification of sex-specific DNA methylation changes driven by specific chemicals in cord blood in a Faroese birth cohort. Epigenetics. 2018; 13:290–300. https://doi.org/10.1080/15592294.2018.1445901 [PubMed]
• 45. Maggio AG, Shu HT, Laufer BI, Bi C, Lai Y, LaSalle JM, Hu VW. Elevated exposures to persistent endocrine disrupting compounds impact the sperm methylome in regions associated with autism spectrum disorder. Front Genet. 2022; 13:929471. https://doi.org/10.3389/fgene.2022.929471 [PubMed]
• 46. Liu Y, Di Y, Zheng Q, Qian Z, Fan J, Ren W, Wei Z, Tian Y. Altered expression of glycan patterns and glycan-related genes in the medial prefrontal cortex of the valproic acid rat model of autism. Front Cell Neurosci. 2022; 16:1057857. https://doi.org/10.3389/fncel.2022.1057857 [PubMed]
• 47. Sands TT, Miceli F, Lesca G, Beck AE, Sadleir LG, Arrington DK, Schönewolf-Greulich B, Moutton S, Lauritano A, Nappi P, Soldovieri MV, Scheffer IE, Mefford HC, et al. Autism and developmental disability caused by KCNQ3 gain-of-function variants. Ann Neurol. 2019; 86:181–92. https://doi.org/10.1002/ana.25522 [PubMed]
• 48. Satterstrom FK, Kosmicki JA, Wang J, Breen MS, De Rubeis S, An JY, Peng M, Collins R, Grove J, Klei L, Stevens C, Reichert J, Mulhern MS, et al, and Autism Sequencing Consortium, and iPSYCH-Broad Consortium. Large-Scale Exome Sequencing Study Implicates Both Developmental and Functional Changes in the Neurobiology of Autism. Cell. 2020; 180:568–84.e23. https://doi.org/10.1016/j.cell.2019.12.036 [PubMed]
• 49. Van Wauwe J, Kemps H, Vrancaert P, Mahy A, Schellingen R, Grootaert MOJ, Beerens M, Luttun A. PRDM16, a new kid on the block in cardiovascular health and disease. Cardiovasc Res. 2025; 121:1156–72. https://doi.org/10.1093/cvr/cvaf089 [PubMed]
• 50. Baizabal JM, Mistry M, García MT, Gómez N, Olukoya O, Tran D, Johnson MB, Walsh CA, Harwell CC. The Epigenetic State of PRDM16-Regulated Enhancers in Radial Glia Controls Cortical Neuron Position. Neuron. 2018; 98:945–62.e8. https://doi.org/10.1016/j.neuron.2018.04.033 [PubMed]
• 51. Umeda T, Takashima N, Nakagawa R, Maekawa M, Ikegami S, Yoshikawa T, Kobayashi K, Okanoya K, Inokuchi K, Osumi N. Evaluation of Pax6 mutant rat as a model for autism. PLoS One. 2010; 5:e15500. https://doi.org/10.1371/journal.pone.0015500 [PubMed]
• 52. Yoshizaki K, Furuse T, Kimura R, Tucci V, Kaneda H, Wakana S, Osumi N. Paternal Aging Affects Behavior in Pax6 Mutant Mice: A Gene/Environment Interaction in Understanding Neurodevelopmental Disorders. PLoS One. 2016; 11:e0166665. https://doi.org/10.1371/journal.pone.0166665 [PubMed]
• 53. Xing M, Zhang Q, Song W. GIGYF1-disturbed IGF-1R recycling: a potential contributor to autism spectrum disorder pathogenesis? J Clin Invest. 2022; 132:e163553. https://doi.org/10.1172/JCI163553 [PubMed]
• 54. Sun W, Poschmann J, Cruz-Herrera Del Rosario R, Parikshak NN, Hajan HS, Kumar V, Ramasamy R, Belgard TG, Elanggovan B, Wong CCY, Mill J, Geschwind DH, Prabhakar S. Histone Acetylome-wide Association Study of Autism Spectrum Disorder. Cell. 2016; 167:1385–97.e11. https://doi.org/10.1016/j.cell.2016.10.031 [PubMed]
• 55. Kubota T, Miyake K, Hirasawa T. Epigenetic understanding of gene-environment interactions in psychiatric disorders: a new concept of clinical genetics. Clin Epigenetics. 2012; 4:1. https://doi.org/10.1186/1868-7083-4-1 [PubMed]
• 56. Plasschaert RN, Bartolomei MS. Tissue-specific regulation and function of Grb10 during growth and neuronal commitment. Proc Natl Acad Sci U S A. 2015; 112:6841–7. https://doi.org/10.1073/pnas.1411254111 [PubMed]
• 57. Garfield AS, Cowley M, Smith FM, Moorwood K, Stewart-Cox JE, Gilroy K, Baker S, Xia J, Dalley JW, Hurst LD, Wilkinson LS, Isles AR, Ward A. Distinct physiological and behavioural functions for parental alleles of imprinted Grb10. Nature. 2011; 469:534–8. https://doi.org/10.1038/nature09651 [PubMed]
• 58. Chiesa N, De Crescenzo A, Mishra K, Perone L, Carella M, Palumbo O, Mussa A, Sparago A, Cerrato F, Russo S, Lapi E, Cubellis MV, Kanduri C, et al. The KCNQ1OT1 imprinting control region and non-coding RNA: new properties derived from the study of Beckwith-Wiedemann syndrome and Silver-Russell syndrome cases. Hum Mol Genet. 2012; 21:10–25. https://doi.org/10.1093/hmg/ddr419 [PubMed]
• 59. Mayo S, Monfort S, Roselló M, Oltra S, Orellana C, Martínez F. In Pursuit of New Imprinting Syndromes by Epimutation Screening in Idiopathic Neurodevelopmental Disorder Patients. Biomed Res Int. 2015; 2015:341986. https://doi.org/10.1155/2015/341986 [PubMed]
• 60. Pauli S, Berger H, Ufartes R, Borchers A. Comparing a Novel Malformation Syndrome Caused by Pathogenic Variants in FBRSL1 to AUTS2 Syndrome. Front Cell Dev Biol. 2021; 9:779009. https://doi.org/10.3389/fcell.2021.779009 [PubMed]
• 61. Bukvic N, De Rinaldis M, Chetta M, Trabacca A, Bassi MT, Marsano RM, Holoubkova L, Rivieccio M, Oro M, Resta N, Kerkhof J, Sadikovic B, Viggiano L. De Novo Pathogenic Variant in FBRSL1, Non OMIM Gene Paralogue AUTS2, Causes a Novel Recognizable Syndromic Manifestation with Intellectual Disability; An Additional Patient and Review of the Literature. Genes (Basel). 2024; 15:826. https://doi.org/10.3390/genes15070826 [PubMed]
• 62. Terrinoni A, Micheloni G, Moretti V, Caporali S, Bernardini S, Minieri M, Pieri M, Giaroni C, Acquati F, Costantino L, Ferrara F, Valli R, Porta G. OTX Genes in Adult Tissues. Int J Mol Sci. 2023; 24:16962. https://doi.org/10.3390/ijms242316962 [PubMed]
• 63. Huang B, Li X, Tu X, Zhao W, Zhu D, Feng Y, Si X, Chen JG. OTX1 regulates cell cycle progression of neural progenitors in the developing cerebral cortex. J Biol Chem. 2018; 293:2137–48. https://doi.org/10.1074/jbc.RA117.001249 [PubMed]
• 64. Redies C, Hertel N, Hübner CA. Cadherins and neuropsychiatric disorders. Brain Res. 2012; 1470:130–44. https://doi.org/10.1016/j.brainres.2012.06.020 [PubMed]
• 65. Luedi PP, Dietrich FS, Weidman JR, Bosko JM, Jirtle RL, Hartemink AJ. Computational and experimental identification of novel human imprinted genes. Genome Res. 2007; 17:1723–30. https://doi.org/10.1101/gr.6584707 [PubMed]
• 66. Clair A, Baker E, Kumari M. Are housing circumstances associated with faster epigenetic ageing? J Epidemiol Community Health. 2023; 78:40–6. https://doi.org/10.1136/jech-2023-220523 [PubMed]
• 67. Wang L, Chen M, Yan G, Zhao S. DNA Methylation Differences Between Zona Pellucida-Bound and Manually Selected Spermatozoa Are Associated With Autism Susceptibility. Front Endocrinol (Lausanne). 2021; 12:774260. https://doi.org/10.3389/fendo.2021.774260 [PubMed]
• 68. Schubert C. Male biological clock possibly linked to autism, other disorders. Nat Med. 2008; 14:1170. https://doi.org/10.1038/nm1108-1170a [PubMed]
• 69. Wu S, Wu F, Ding Y, Hou J, Bi J, Zhang Z. Advanced parental age and autism risk in children: a systematic review and meta-analysis. Acta Psychiatr Scand. 2017; 135:29–41. https://doi.org/10.1111/acps.12666 [PubMed]
• 70. Levine ME, Lu AT, Quach A, Chen BH, Assimes TL, Bandinelli S, Hou L, Baccarelli AA, Stewart JD, Li Y, Whitsel EA, Wilson JG, Reiner AP, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY). 2018; 10:573–91. https://doi.org/10.18632/aging.101414 [PubMed]
• 71. Loke YJ, Hannan AJ, Craig JM. The Role of Epigenetic Change in Autism Spectrum Disorders. Front Neurol. 2015; 6:107. https://doi.org/10.3389/fneur.2015.00107 [PubMed]
• 72. Zhou J, Park CY, Theesfeld CL, Wong AK, Yuan Y, Scheckel C, Fak JJ, Funk J, Yao K, Tajima Y, Packer A, Darnell RB, Troyanskaya OG. Whole-genome deep-learning analysis identifies contribution of noncoding mutations to autism risk. Nat Genet. 2019; 51:973–80. https://doi.org/10.1038/s41588-019-0420-0 [PubMed]
• 73. Litman A, Sauerwald N, Green Snyder L, Foss-Feig J, Park CY, Hao Y, Dinstein I, Theesfeld CL, Troyanskaya OG. Decomposition of phenotypic heterogeneity in autism reveals underlying genetic programs. Nat Genet. 2025; 57:1611–9. https://doi.org/10.1038/s41588-025-02224-z [PubMed]
• 74. World Health Organization Laboratory Manual for the Examination and Processing of Human Semen. WHO Press: Geneva, Switzerland; 2010; 287.
• 75. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, Delano D, Zhang L, Schroth GP, Gunderson KL, Fan JB, Shen R. High density DNA methylation array with single CpG site resolution. Genomics. 2011; 98:288–95. https://doi.org/10.1016/j.ygeno.2011.07.007 [PubMed]
• 76. Morris TJ, Butcher LM, Feber A, Teschendorff AE, Chakravarthy AR, Wojdacz TK, Beck S. ChAMP: 450k Chip Analysis Methylation Pipeline. Bioinformatics. 2014; 30:428–30. https://doi.org/10.1093/bioinformatics/btt684 [PubMed]
• 77. Teschendorff AE, Marabita F, Lechner M, Bartlett T, Tegner J, Gomez-Cabrero D, Beck S. A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data. Bioinformatics. 2013; 29:189–96. https://doi.org/10.1093/bioinformatics/bts680 [PubMed]
• 78. Nye MD, Hoyo C, Huang Z, Vidal AC, Wang F, Overcash F, Smith JS, Vasquez B, Hernandez B, Swai B, Oneko O, Mlay P, Obure J, et al. Associations between methylation of paternally expressed gene 3 (PEG3), cervical intraepithelial neoplasia and invasive cervical cancer. PLoS One. 2013; 8:e56325. https://doi.org/10.1371/journal.pone.0056325 [PubMed]
• 79. Murphy SK, Huang Z, Hoyo C. Differentially methylated regions of imprinted genes in prenatal, perinatal and postnatal human tissues. PLoS One. 2012; 7:e40924. https://doi.org/10.1371/journal.pone.0040924 [PubMed]
• 80. Krausz C, Sandoval J, Sayols S, Chianese C, Giachini C, Heyn H, Esteller M. Novel insights into DNA methylation features in spermatozoa: stability and peculiarities. PLoS One. 2012; 7:e44479. https://doi.org/10.1371/journal.pone.0044479 [PubMed]
• 81. Kruppa J, Sieg M, Richter G, Pohrt A. Estimands in epigenome-wide association studies. Clin Epigenetics. 2021; 13:98. https://doi.org/10.1186/s13148-021-01083-9 [PubMed]
• 82. Xie C, Leung YK, Chen A, Long DX, Hoyo C, Ho SM. Differential methylation values in differential methylation analysis. Bioinformatics. 2019; 35:1094–7. https://doi.org/10.1093/bioinformatics/bty778 [PubMed]
• 83. Chen YA, Lemire M, Choufani S, Butcher DT, Grafodatskaya D, Zanke BW, Gallinger S, Hudson TJ, Weksberg R. Discovery of cross-reactive probes and polymorphic CpGs in the Illumina Infinium HumanMethylation450 microarray. Epigenetics. 2013; 8:203–9. https://doi.org/10.4161/epi.23470 [PubMed]
• 84. Hansen KD. IlluminaHumanMethylation450kanno.ilmn12.hg19: Annotation for Illumina's 450k methylation arrays. R package version 0.6.0. 2016.
• 85. Feinberg JI, Bakulski KM, Jaffe AE, Tryggvadottir R, Brown SC, Goldman LR, Croen LA, Hertz-Picciotto I, Newschaffer CJ, Fallin MD, Feinberg AP. Paternal sperm DNA methylation associated with early signs of autism risk in an autism-enriched cohort. Int J Epidemiol. 2015; 44:1199–210. https://doi.org/10.1093/ije/dyv028 [PubMed]
• 86. Bethesda (MD): National Library of Medicine (US) NCfBI. 2004.
• 87. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, Irizarry RA. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014; 30:1363–9. https://doi.org/10.1093/bioinformatics/btu049 [PubMed]
• 88. Pidsley R, Y Wong CC, Volta M, Lunnon K, Mill J, Schalkwyk LC. A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics. 2013; 14:293. https://doi.org/10.1186/1471-2164-14-293 [PubMed]
• 89. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015; 43:e47. https://doi.org/10.1093/nar/gkv007 [PubMed]
• 90. Maksimovic J, Oshlack A, Phipson B. Gene set enrichment analysis for genome-wide DNA methylation data. Genome Biol. 2021; 22:173. https://doi.org/10.1186/s13059-021-02388-x [PubMed]