Fig 2: Loss of KDM3A and KDM3B leads to differences in stable naïve state.(A) Schematic detailing derivation of naïve ESCs (serum/LIF) or ground ESCs (2i/LIF).(B) Scatterplot of splicing events called by rMATs in both WT vs DDeg+ ESCs as well as serum vs 2i/LIF ESCs (74) measured by inclusion level difference score. Pearson correlation between the inclusion level difference of specific events called in both comparisons. Blue dots represent events chosen for further confirmation.(C) Semi-quantitative PCR of splicing targets in WT ESCs, DDeg+ clones at 4 hours of degradation, and 2i/LIF ESCs. Representative IGV tracks depicting affected exon (red box) are below. Electrophoresis image was quantified using Licor software, fluorescence intensity reported as arbitrary units (a.u.) and represented in right panel. I=included isoform and S=Skipped isoform.(D) Relative expression of c-Myc or Dazl in WT or DDeg+ ESCs in serum/LIF or 2i/LIF conditions. WT cells in serum/LIF set to 1.

Fig 3: Alternative splicing defects upon acute loss of KDM3A/KDM3B occur in signaling and chromatin related genes(A) Venn overlap of DE Genes (blue) and aberrantly spliced genes (yellow) called from rMATs and Dexseq analysis.(B) Left panel: schematic of alternative splicing pathways, results above the dotted line are from rMATs and below are from Dexseq. Right panel: Number of significant events called in each class of alternative splicing pathway, FDR<0.1, inclusion level difference x>+/-0.5, and 15 read minimum.(C) Venn overlap of aberrantly spliced genes sorted by higher inclusion in WT (blue) or in DDeg+ (red). Gene ontology categories for aberrantly spliced genes shown, minimum p-value < 0.001. IGV tracks for example genes of exons with higher percent spliced in (PSI) for WT or DKO conditions. PSI value calculated in rMATs from 3 replicates in each condition. Negative value indicates event is included more in the WT condition and vice versa. The affected exon is marked in red.(D) (i) Bar graph shows the number of genes within each category, binned by gene length normalized by intron number from longest to shortest. Categories include genes with significant spliced events sorted by negative inclusion (blue) or positive inclusion (red) and plotted on the left y-axis, and the distribution of any gene detected (black) by rMATs is included as comparison on the right y-axis. (ii) Same as above except splice events are binned by decreasing expression (TPMs).(E) Clustering (k-means=5) of differentially spliced events by inclusion level difference score generated in rMATs. WT vs uninduced degron (DDeg-) and WT vs induced degron (DDeg+) lines are shown. A positive number indicates greater inclusion in WT cells. Example genes from clusters and significant gene ontology terms (minimum p-value < 0.001 ) are in right panel.

Fig 4: Acute degradation of KDM3A and KDM3B leads to differential gene expression agnostic of H3K9me2 status(A) Number of differentially expressed genes (DEG) between Double Degron (DDeg+) exposed to dTAG-13 ESCs and wildtype ESCs. Gene ontology categories for DEGs shown at minimum p-value < 0.00001.(B) Boxplot of absolute expression (TPMs) in ESCs for all annotated genes from RSEM quantification with a non-zero median TPM (All expressed genes), TPMs from up or down DEGs. Numbers indicate number of genes per condition. ****p-value < 0.0001 assessed by unpaired Welch’s t-test.(C) Boxplot of average TPMs+1 from DEGs with (H3K9me2) or without (No H3K9me2) a called consensus H3K9me2 peak (54, 55). *p-value < 0.05 assessed by unpaired Welch’s t-test(D) Left: Scatterplot of log2(DDeg+/WT) fold change of TPMs of Non-DE and DEG against H3K9me2 enrichment (normalized read counts per million and by 1kb per gene) at those genes. Right: Boxplot of total H3K9me2 enrichment at up or down DEG.(E) Four-way Venn overlap of DEG and H3K9me2 peaks in WT ESCs and KDM3A and KDM3B DKO ESCs (22).

Fig 5: KDM3A and KDM3B interact with splicing machinery in ESCs(A) Immunoblot of (i) KDM3A, (ii) FLAG in ESCs with doxycycline (dox)-inducible expression of FLAG-KDM3A. LC= loading control = RNAPII. OE=Overexpression(B) Same as above except in dox-inducible FLAG-KDM3B ESCs (i) KDM3B (ii) FLAG (iii) FLAG following immunoprecipitation with FLAG antibody. I=Input, UB=Unbound, E=Elute(C) Venn diagram of KDM3A and KDM3B interactome. Filtered for nuclear compartment using DAVID GO (https://david.ncifcrf.gov/) analysis and frequent FLAG pulldown contaminants using “Crapome” (https://reprint-apms.org/) repository at SAINT score of 0.7 (61). Gene Ontology (GO) enrichment of shared interactome are listed at enrichment score>2.5 and p-value<.00001.(D) String network analysis (https://string-db.org/) of KDM3A and KDM3B shared interactome with high confidence interaction scores (0.7 minimum) from database and experimental sources. Node size represents fold change (FC) of KDM3A over uninduced spectral counts. FC>10 = large node, 10>FC>5 = medium node, FC<5 = small node.(E) KDM3A and KDM3B interactors overlapped with functional categories of mRNP components from Singh et al. (105) or categorized as other mRNA processing proteins. Shared= black, KDM3A unique=blue, and KDM3B unique=red.(F) Single slice of 1 uM confocal images of in situ Duolink™ Proximity ligation assay (PLA) of (i) KDM3B single antibody, EFTUD2 single antibody, and KDM3B+EFTUD2 antibodies in WT ESCs or KDM3B+EFTUD2 antibodies in KDM3B KO ESCs. (ii) PLA of KDM3B and OCT4 in WT ESCs. Scale bar = 50µM with 60x objective.