Assisted reproductive technology alters deoxyribonucleic acid methylation profiles in bloodspots of newborn infants  Molly S. Estill, M.S., Jay M. Bolnick, M.D., Robert A. Waterland, Ph.D., Alan D. Bolnick, M.D., Michael P. Diamond, M.D., Stephen A. Krawetz, Ph.D.  Fertility and Sterility  Volume 106, Issue 3, Pages 629-639.e10 (September 2016) DOI: 10.1016/j.fertnstert.2016.05.006 Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Figure 1 Intracytoplasmic sperm injection and IUI compared with NAT show differential methylation of clusters and gene bodies. (A) Total counts of differentially methylated clusters between conception groups for males and females, shown in blue and pink, respectively. Red bracket indicates comparisons of FH and FZ with IUI, presenting the greater degree of differential methylation in the FH vs. IUI comparison than that of FZ vs. IUI. (B) Distribution of the change in the β-value for statistically significant clusters, as a function of conception comparison. Changes in the female and male comparisons are shown in light red and blue, respectively. (C) Pie chart labeled “Genome Annotation” indicates the proportion of the human genome that lies within exons, introns, promoters, and intergenic regions. The pie chart labeled as “Cluster Annotation” provides the proportion of all statistically significant clusters (regardless of methylation change) that overlap exons, introns, promoters, and intergenic regions. Clusters that overlay one or more features are denoted as “partial.” Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Figure 2 Metastable epialleles at DUSP22 and SPATC1L show considerable and concerted differential methylation. University of California Santa Cruz (UCSC) genome browser representation of female comparisons between conception groups for MEs located at (A) DUSP22 and (B) SPATC1L. Hypermethylated clusters and hypomethylated clusters are colored in blue and orange, respectively. Cluster heights represent the magnitude of methylation change, in scale with the y axis. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 1 Samples and conception group comparisons subject to differential methylation analysis. (A) Number of individuals analyzed in this study, according to conception type and gender. (B) Directional arrows indicate that the conception group at the source of the arrow is the methylation dataset (control) against which that at the termination of the arrow (case) is being compared. (C) Pipeline provided by ChAMP was used to filter, normalize, and apply batch correction, to obtain corrected methylation values. Aclust was then implemented to calculate methylation changes in probe clusters, followed by downstream analyses. Concurrently, the same corrected methylation values are analyzed using a linear model to calculate differential methylation of individual probes and verify the results obtained from Aclust. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 2 Estimated blood cell composition does not affect the bloodspot methylation. Singular value decomposition analysis was performed, through the ChAMP pipeline, to determine the effect of blood cell proportions, estimated using the estimatecellcount function in minifi, on sample methylation. The association of each principle component with plate, assay characteristics, and blood cell proportions is indicated using P values. A black rectangle indicates the P value slots for the various blood cell types, abbreviated as follows: CD8 T-cells (CD8T), CD4 T-cells (CD4T), natural killer cells (NK), B-cells (Bcell), monocytes (Mono), and granulocytes (Gran). Assay characteristics are abbreviated as follows: bisulfite conversion (BSC), and hybridization (Hyb). Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 3 Unsupervised clustering of bloodspot methylation using principle component analysis does not effectively segregate the different conception groups. Principle component analysis of BMIQ-normalized 450k probes, filtered to only contain the allowed 394,454 probes for all bloodspot samples. Normal contour lines, at an ellipse probability of 68%, are indicated for each group. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 4 Trends in differentially methylated clusters between males and females of identical conception groups. Hypermethylation in the female group (compared with a male control) is shown in red, whereas hypomethylation is presented in blue. (A) Counts of clusters differentially methylated between females and males of the same conception group. (B) Counts of gene bodies differentially methylated between females and males of the same conception group. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 5 Trends in non-normalized differentially methylated clusters indicates marked differences between conception groups. Differentially methylated clusters generated from non-normalized β-values for (A) males and (B) females. (C) Differentially methylated clusters generated from non-normalized β-values for comparisons between males and females. Hypermethylated and hypomethylated clusters are represented in red and blue, respectively. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 6 Trends in differentially methylated probes, as determined by limma, are similar to trends seen using Aclust. Bars indicate the total counts of differentially methylated probes as calculated using limma, for males and females (shown in blue and red, respectively), for each conception comparison. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 7 Intracytoplasmic sperm injection and IUI show considerable differential methylation in gene bodies compared with NAT. Bars indicate the counts of differentially methylated gene bodies between conception groups. Counts of differentially methylated gene bodies generated for (A) male- and (B) female-specific comparisons. Hypermethylated and hypomethylated gene bodies are represented in red and blue, respectively. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 8 Enhancers consistently altered in ICSI groups compared with NAT. Populations of hypermethylated and hypomethylated enhancers altered in the IUI, FH, or FZ vs. NAT comparison were identified and enumerated. Quantities of hyper- or hypomethylated enhancers are denoted as “Hyper” and “Hypo,” respectively. The quantity of enhancers found in common between two or more comparisons and exhibiting identical methylation trends (e.g., increased or decreased methylation in both comparisons) are indicated in the intersections of the Venn diagram. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 9 Certain imprinted genes associated with metabolism and cancer exhibit differential methylation. UCSC genome browser images of promoter and gene bodies of imprinted genes (A) H19, (B) IGF2. Hypermethylated clusters and hypomethylated clusters are colored in blue and orange, respectively. Cluster heights represent the magnitude of methylation change, in scale with the y axis. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions

Supplemental Figure 10 Metastable epialleles adjacent to SFT2D3, WDR33, and LIMS2 are hypomethylated, only in the male comparisons, when comparing infertile groups with NAT. UCSC representation of male and female comparisons between conception groups for metastable epialleles located upstream of SFT2D3, WDR33, and LIMS2. Hypermethylated clusters and hypomethylated clusters are colored in blue and orange, respectively. Cluster heights represent the magnitude of methylation change, in scale with the y axis. Fertility and Sterility 2016 106, 629-639.e10DOI: (10.1016/j.fertnstert.2016.05.006) Copyright © 2016 American Society for Reproductive Medicine Terms and Conditions