Supplementary Figure 1 Effect of HSP90 inhibition on expression of endogenous retroviruses. (a) Inducible shrna-mediated Hsp90 silencing in mouse ESCs. Immunoblots of total cell extract expressing the indicated constructs are shown. Two independent cell clones were analyzed. (b-f) A genome browser snapshot of genomic loci with indicated ERVs. RNAseq read counts under control and HSP90 inhibition conditions are plotted on the Y-axis. Mouse genomic co-ordinates from the genome build mm10 are as follows: (b) chr16: 57,805,060-57,812,042; (c) chr3: 35,413,307-35,419,237; (d) chr9:25,854,570-25,864,259. (e) chr3: 26,346,706-26,354,408; (f) chr13: 98,985,649-98,996,505. (g) The effect of HSP90 inhibition on expression of TEs in indicated cell types. The numbers of up- and down-regulated TEs (absolute fold change >2, adjusted p-value < 0.01) as annotated by RepeatMasker are indicated. (h) The effect of heat shock and two different HSP90 inhibitor concentrations (50nM and 75nM) on the expression of heat shock genes in ESCs. The same two HSP90 inhibitor concentrations were used to test the expression changes of IAPEz and MERV-L in ESCs. Fold change is shown as a mean of three independent cell-culture replicates (± SEM).
Supplementary Figure 2 Expression of genes close to ERVs is affected by HSP90 inhibition. (a) HSP90 inhibition-induced changes in expression of genes located within 200kb region around ERV insertions in NPCs. The analyzed subtypes of ERVs are shown in different colors as indicated. (b) A table showing the fold change in three selected ERVs upon HSP90 inhibition, the number of genes adjacent to these ERVs and the mean/median distance between ERVs and adjacent genes. (c,d) HSP90 inhibition-induced changes in expression of all genes in the genome (black box) in comparison with genes close to IAPEz (blue boxes in c) and close to MERVL (red boxes in d) either closer or more distant than 25kb to the closest ERV type in ESCs. Consistent with previous observations (Fig. 2a), the effect of Hsp90 inhibition on gene expression of adjacent genes trails-off with distance. (e,f) Cumulative distributions of the distance between genes and adjacent IAPEz (e) or MERVL(f). Almost 60% of the adjacent genes are within 25kB of ERV. (g) Sashimi plot of a chimeric transcript originating in an adjacent MT2B2 LTR region and continuing transcription in the Ei24 gene region. (h-k) Validation of two chimeric transcripts, Clvs1 and Rimklb, by quantitative PCR (h,i) and PCR (j,k). For qpcr, fold change is shown as a mean of three independent cell-culture replicates (± SEM).
Supplementary Figure 3 Cell-type-specific genes are misregulated after HSP90 inhibition. (a,b) Highly tissue-specific genes are ectopically activated in NPCs (a) and macrophages (b) upon HSP90 inhibition due to their neighborhood to ERVs. The heat maps were generated using RNAseq data from ENCODE detailing gene expression patterns in each of the individual mouse tissues. (c,d) Genes in (a) and (b) respectively are clustered together by common gene ontology terms as depicted.
Supplementary Figure 4 Comparison of genes upregulated in SETDB1 KO, KAP1 KO and HSP90 inhibition. (a) A venn diagram comparing genes upregulated in KAP1 KO and SETDB1 KO. Genes that are upregulated in both KOs or only one of the two KOs were then compared with genes upregulated upon HSP90 inhibition (bottom). The P-value estimated by the hypergeometric test is shown. (b,c) HSP90 inhibition affects expression of genes in the vicinity of ERVs similar to KAP1 KO and SETDB1 KO. Changes in expression of genes located within 200kb region around ERV insertions in ESCs upon KAP1 KO (b) and SETDB1 KO (c). The analyzed subtypes of ERVs are shown in different colors as indicated. (d) A Sashimi plot of the indicated locus showing chimeric reads between an ERV and an adjacent gene (co-ordinates correspond to genome build mm10). The number of chimeric reads increases upon HSP90 inhibition, and upon KAP1 deletion, buttressing the claim that HSP90 and KAP1 act in the same pathway.
Supplementary Figure 5 HSP90 inhibition upregulates genes adjacent to ERVs that are repressed by KAP1-mediated H3K9me3 and SETDB1. (a,b) A genome browser view of two loci showing an ERV and an adjacent gene, with ChIPseq and RNAseq profiles from mouse ES cells with indicated treatments and genotypes. The indicated coordinates correspond to mouse genome build mm10. At both loci, the ERV (IAPEz, green) is bound by KAP1 and shows the repressive mark H3K9me3 (brown box). The repressive mark is KAP1- dependent as the ChIPseq signal of H3K9me3 is reduced in KAP1 deletion. The adjacent gene is de-repressed in HSP90 inhibition, KAP1 deletion and for (b) also in SETDB1 deletion (RNAseq profiles in green and red).
Supplementary Figure 6 H3K9me3 and KAP1 occupancy at HSP90-target ERVs and the effect of stress on ERV transcription (a) ChIP-qPCR for HSP90 in ESCs. Enrichment is shown as the mean (± SEM). (b) ChIP-qPCR for Kap1 and H3K9me3 in control ESCs and ESCs treated with HSP90 inhibitor. Enrichment is shown as the mean of three independent cell-culture replicates (± SEM) relative to control conditions. The red dotted line indicates the signal intensity in cells not treated with HSP90 inhibitor (c,d) Effect of Kap1 knock-out on H3K9me3 modification at and around HSP90-targeted ERVs, namely IAPEz (c) and MERV-L (d). (e) Experimental design to study the effect of HSP90 inhibition on genes with different strain-specific upstream ERV elements. All genes tested have an ERV either in 129S1 or CASTEiJ strain or both, but no ERV in C57BL6. Effect of HSP90 inhibitor on gene expression was analyzed by quantitative reverse transcription-pcr. Fold Change is shown as a mean of three independent cell-culture experiments (± SEM). (f) Different conditions mimicking environmental stresses upregulate IAPEz and MERV-L in ESC as shown by quantitative PCR. Fold change is shown as means of three independent cell-culture experiments (±SEM).