SETD3 regulates endoderm differentiation of mouse embryonic stem cells through canonical Wnt signaling pathway

With self‐renewal and pluripotency features, embryonic stem cells (ESCs) provide an invaluable tool to investigate early cell fate decisions. Pluripotency exit and lineage commitment depend on precise regulation of gene expression that requires coordination between transcription (TF) and chromatin factors in response to various signaling pathways. SET domain‐containing 3 (SETD3) is a methyltransferase that can modify histones in the nucleus and actin in the cytoplasm. Through an shRNA screen, we previously identified SETD3 as an important factor in the meso/endodermal lineage commitment of mouse ESCs (mESC). In this study, we identified SETD3‐dependent transcriptomic changes during endoderm differentiation of mESCs using time‐course RNA‐seq analysis. We found that SETD3 is involved in the timely activation of the endoderm‐related gene network. The canonical Wnt signaling pathway was one of the markedly altered signaling pathways in the absence of SETD3. The assessment of Wnt transcriptional activity revealed a significant reduction in Setd3‐deleted (setd3∆) mESCs coincident with a decrease in the nuclear pool of the key TF β‐catenin level, though no change was observed in its mRNA or total protein level. Furthermore, a proximity ligation assay (PLA) found an interaction between SETD3 and β‐catenin. We were able to rescue the differentiation defect by stably re‐expressing SETD3 or activating the canonical Wnt signaling pathway by changing mESC culture conditions. Our results suggest that alterations in the canonical Wnt pathway activity and subcellular localization of β‐catenin might contribute to the endoderm differentiation defect of setd3∆ mESCs.


| INTRODUCTION
In the initial stages of embryonic development, different cell types form through a process of cleavage and differentiation, which ultimately contribute to the formation of the organism.This necessitates a precisely timed and coordinated mechanism.Embryonic stem cells (ESCs) provide an in vitro model to investigate early decisions guiding cell fate.They exhibit an indefinite self-renewal capability.Due to pluripotency, they can be guided toward specific lineages with appropriate cytokine stimulation. 1ytokines initiate a cascade of various signaling pathway activations, leading to ESC lineage commitment.Several of these pathways, such as the TGFβ, MAPK, and Wnt, function both at the ESC stage and during differentiation. 2Upon specific lineage commitment, gene networks corresponding to other lineages are inhibited. 3Activation and inhibition of gene regulatory networks during differentiation is facilitated by epigenetic factors that can finetune and influence the dynamics of lineage commitment.
SETD3 is a member of the SET domain-containing family of methyltransferases, which modify histones to modulate chromatin architecture and regulate gene expression. 4It can methylate histone 3 lysine 4 (H3K4) and histone 3 lysine 36 (H3K36) and control muscle cell differentiation by regulating myogenin expression through its interaction with MyoD. 5 It can also methylate actin in the cytoplasm, providing filament stabilization. 6etd3-deleted mice (Mus musculus) and the dSETD3 knockdown/knockout fruit fly (Drosophila melanogaster) can reach adulthood successfully. 7,8Deletion of the Setd3 gene in fruit flies had no impact on traits such as fertility and body weight; however, microarray analysis of Setd3 knockout fruit flies revealed significant changes in gene expression, particularly those associated with muscle structure or contraction. 7Setd3 knockout mice showed moderate skeletal muscle myopathy, mildly decreased lean mass, and an abnormal cardiac electrocardiogram, in addition to a smaller litter size and primary distocia. 8 another study, SETD3 upregulation was observed to contribute to the progression of post-stroke depression in rats by negatively regulating VEGF expression. 9SETD3 has been reported to have both oncogenic and tumor suppressive roles. 10,11High expression of SET domain-deleted Setd3 mRNA was observed in lymphoma and the protein was found to have more oncogenic potential. 12ur previous shRNA screen identified SETD3 as a potential critical factor for mesoderm and endoderm commitment. 13However, the role of SETD3 during mouse ESC (mESC) differentiation is not known.To explore the role of SETD3 in endoderm differentiation, we utilized time-course transcriptomic analyses.Our results indicate a role for SETD3 in the timely downregulation of the pluripotent state and response to key signaling pathways, mainly through the activity of the canonical Wnt pathway.

| MATERIALS AND METHODS
2.1 | Generation of SETD3 deletion and rescue mESCs CJ9 (wild-type) mESCs were used to prepare SETD3 knockout (setd3Δ) with previously used protocols. 13The paired primers were designed to take out the methyltransferase SET domain, which spans an approximately 10 kb region in the genome.The sgRNA primers used for this purpose were; Forward: CACCGCCATTGATACTATGTAGCCG, Forward complement: AAACCGGCTACATAGTATCAATGGC, Reverse: CACCGTGATCCCTTAGACAACGGCA, Reverse complement: AAACTGCCGTTGTCTAAGGGATCAC. mESC deletions were performed as previously described. 13mESC clones were screened using conventional PCR, validated by qPCR using primers that amplify within the deleted region (insDel primers in Table 1), and Western blotting (Figure S1).Setd3 ORF was amplified from the mouse Setd3 cDNA clone plasmid (MC203957, Origene) using the AscI-XbaI restriction enzyme cut site containing primers.A putative NLS sequence nuclear pool of the key TF βcatenin level, though no change was observed in its mRNA or total protein level.Furthermore, a proximity ligation assay (PLA) found an interaction between SETD3 and βcatenin.We were able to rescue the differentiation defect by stably re-expressing SETD3 or activating the canonical Wnt signaling pathway by changing mESC culture conditions.Our results suggest that alterations in the canonical Wnt pathway activity and subcellular localization of βcatenin might contribute to the endoderm differentiation defect of setd3∆ mESCs.

K E Y W O R D S
beta-catenin, canonical, definitive endoderm, mESCs, mouse embryonic stem cells, SET domain-containing 3 protein, Wnt pathway of SETD3 was identified by the cNLS Mapper. 14SET-domain and the putative NLS deleted from the Setd3 ORF via overlap extension PCR. 15Both constructs were cloned into the pEF1αFlagBio plasmid and verified by sequencing.
2.2 | mESC culture and endoderm differentiation mESCs were grown and maintained either in standard, or 2i4 medium, meso/endoderm differentiation was conducted as previously described. 16,17

| Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
RNA extraction, cDNA synthesis, and RT-qPCR were performed as previously described. 13Primers are listed in Table 1.Transcript levels were normalized to β-actin level.Statistical analyses (two-way ANOVA or t-test) were done on GraphPad Prism software.

| RNA-seq
Genomic DNA was eliminated using DNase (Qiagen, 79254).The RNA quality was determined on an Agilent BioAnalyzer (METU Central Lab).PolyA+ library construction and sequencing were done at Macrogen Europe (Amsterdam, Netherlands).The concentrations of library cDNA samples were analyzed using Qubit.Sequencing was performed using Illumina HiSeq2000.Accession number: GSE242033.

| Bioinformatic analyses
Bioinformatic analyses were done by Refgen (Ankara, Turkey) using tools and methods below 18 : FASTQC (Babraham Bioinformatics, USA) was used for the quality control of the acquired data.Poor-quality trimming was done using the Trimmomatic.The HISAT2 T A B L E 1 Primers used in the study.

Gene
Forward primer (5′-3′) Reverse primer (5′-3′) was used for alignment to Mus musculus reference genome M25 (GRCm38.p6).The Subread tool was used to determine the number of reads for the transcriptome elements.The R:edgeR tool and the R::limma tool were used for normalization and filtering of reads per gene and to identify genes with varying expression between groups (Differentially Expressed Genes), respectively.R scripts were used in statistical comparison studies within and between groups and data visualization applications.
2.7 | Gene Ontology (GO) analysis of differentially expressed genes (DEGs) GO analysis was conducted on the DAVID 6.8 web tool.DEGs upregulated in wild-type cells on each day of endoderm differentiation were separately analyzed.GO biological processes were ranked by statistical significance.

| Gene set enrichment analysis (GSEA)
The list of wild-type/setd3Δ genes on each day of differentiation was used and GSEAPreranked was performed using GSEA_4.1.0software.

| Short Time-Series Expression Miner (STEM) analysis
Transcript counts from each biological replicate were converted to RPKM and averaged.RPKM values were lognormalized using the STEM software.Genes with more than one missing value for four time points were filtered out of the analysis.If the absolute expression change of a gene between any time points was lower than 0.585 (=log 2 (1.5)), then it was filtered out as well.For GO analysis, only biological processes were included.The minimum GO hierarchy level was set to 6.The results were visualized using REVIGO (http:// revigo.irb.hr).

| Chromatin co-immunoprecipitation
pSetd3 (pEF1αFlagBioSetd3-puro) was stably transfected into biotinylation enzyme BirA-expressing wildtype (J1BirA) mESCs. 22pSetd3 has a BirA biotinylation sequence at the N-terminus of SETD3 and is only biotinylated in BirA-expressing cells.Parental (J1BirA) and pSetd3-expressing J1BirA mESCs were cross-linked with 1% formaldehyde for 10 min and lysed in RIPA buffer to make crosslinked whole cell extracts.Following centrifugation, chromatin is in the pellet, while all soluble material is in the supernatant.The chromatin pellets were sonicated in 0.1% SDS ChIP Buffer (5 cycles, 30s on/off, 25% amplitude).Equal amounts of chromatin were incubated with streptavidin-magnetic beads (S1420S, NEB) overnight for SETD3 immunoprecipitation.Beads were washed as described. 22For decrosslinking and elution, chromatin was incubated at 95°C for 30 min.The immunoprecipitated protein material was loaded onto an SDS-PAGE.Anti-SETD3 (NBP232136, Novus) and anti-βcatenin (sc-133240, Santa Cruz) antibodies were used for detection in Western blot as described before.

| RESULTS
To verify the shRNA screen result, we generated CRISPR/ Cas9-mediated genomic Setd3 deletion in mESCs (Figures S1A and 1A).A Western blot analysis using an antibody against the N-terminus of SETD3 showed no truncated protein expression, confirming a complete knockout (Figure S1B).SETD3 loss did not lead to any morphological aberrations (Figure 1B).Consistent with our shRNA screen results, setd3∆ mESCs failed to express mesendoderm-specific transcription factor (TF) Brachyury (Figure 1C,D) or endoderm-specific TFs Foxa2 (Figure 1E) and Sox17 (Figure 1F) during meso/endoderm differentiations.Conversely, expression of neuroectoderm-specific pioneer TFs Sox1 and Pax6 in setd3∆ cells was comparable to wild-type cells during neuroectoderm differentiation (Figure 1G,H).These results demonstrate a critical function for SETD3, specifically during meso/endoderm differentiation.
Western blot analysis showed sustained expression of SETD3 through early endodermal commitment (Figure 1I).To pinpoint the origin of the endoderm differentiation defect of setd3∆ mESCs, time-course RNA-seq was performed during endoderm differentiation.Within four days, the pioneer TFs Brachyury and its direct target Foxa2 are expressed while the pluripotency network is suppressed. 23Hierarchical clustering (data not shown) and principal component analyses (PCA) of the RNA-seq results showed tight clustering of wild-type and setd3∆ mESCs (Figure 2A).Their diversion started after the second day, coincident with the switch to an Activin-A supplemented medium.Comparison of the setd3∆ endoderm commitment path with wild-type cells revealed a delay in endoderm differentiation.The number of differentially expressed genes (DEG) gradually increased through differentiation, with the minimum difference in the ESC state (Figure 2B).Genes with the highest expression change upon SETD3 deletion showed a clear overrepresentation of early meso/endoderm marker genes such as Brachyury, Foxa2, Sox17, Cer1, Lhx1, and Gata6 (Figure 2C).
Visualization of RNA-seq reads through Integrative Genomics Viewer (IGV) revealed de-repression of these marker genes starting from day 3 (Figure 2D).Brachyury, and Foxa2 expression increased on the third day of differentiation in wild-type cells.Coincidentally, with commitment toward endodermal lineage, Foxa2 expression was stably high while Brachyury expression decreased thereafter.The subsequent activation of the endodermspecific transcription network was observed through high expression of Lhx1, Gsc, Cxcr4, c-Kit, Cer1, and Dkk1, especially on the fourth day of differentiation.In setd3Δ cells, these genes were not expressed within the same time frame, indicating improper and timely activation of the endoderm-related transcription network in the absence of SETD3.The LogFC and FDR values of the selected DEGs (|LogFC| > 1.5, FDR < 0.05) are shown in Table 2.
Consistent with mESC differentiation, the pluripotency genes Oct4, Nanog, Klf4, and Rex1 were shut off rapidly (Figure 2E).While Nanog, Klf4, and Rex1 were similarly downregulated in setd3Δ cells compared to wild-type cells, Oct4 expression remained high.Reduced endoderm commitment might also result from aberrant upregulation of neuroectodermal gene networks.Early neuroectoderm markers Neurod1, Pax6, and Sox1 showed little to no expression in the endoderm differentiation time-course of wild-type or setd3Δ cells (Figure 2F), arguing against neuroectoderm transcription network activation.
To verify that this phenotype can be directly attributed to the loss of SETD3, Flag-tagged SETD3 was re-expressed in setd3Δ cells.Rescued SETD3 expression was confirmed in both the cytoplasm and nucleus via Western blotting (Figure 3A).Wild-type and rescue (setd3Δ + pSetd3) mESCs showed similar levels of SETD3.The RT-qPCR analysis of wild-type, setd3Δ, and rescue cells during endoderm differentiation time-course verified the RNAseq results and showed that the pioneer TFs Brachyury and Foxa2 levels were rescued with a one-day delay (Figure 3B,C).While setd3∆ cells had no Lhx1, Gsc, Gata6, Cxcr4, Cer1, or Dkk1 expression, rescue cells showed the same pattern as wild-type (Figure 3D-I).Pluripotency factor Oct4 expression dropped on the fourth day in wild-type cells, yet it remained high throughout the differentiation in setd3Δ cells.Rescue cells successfully shut off Oct4 expression (Figure 3J).
To investigate whether the nuclear or cytoplasmic function of SETD3 is responsible for the observed endoderm differentiation defect, a putative nuclear localization signal (NLS) (Figure S2A) was deleted, and setd3∆ mESCs were stably transfected with this construct.SETD3 levels in nuclear and cytoplasmic fractions were validated by Western blotting (Figure S2B).SETD3 was not observed in the nuclear fraction of setd3∆ + pSetd3∆NLS mESCs, though there was overall reduced SETD3∆NLS expression.Time-course endoderm differentiation showed Brachyury expression was partially rescued by stable SETD3∆NLS re-expression (Figure S2C).Foxa2 expression was comparable to setd3∆ cells (Figure S2D).Similarly, SETD3∆NLS re-expression was not sufficient to rescue the expression of endoderm markers (Figure S2E).Overall, the re-expression of full-length SETD3 but not SETD3∆NLS fully rescued the defective endoderm differentiation phenotype.As SETD3 acts as an actin methyltransferase in the cytoplasm, actin filament (F-actin) levels of wild-type, setd3∆, and setd3∆ + pSetd3 mESCs were observed with phalloidin staining.No clear difference was observed (Figure S2F).
To investigate whether the catalytic function of SETD3 is critical for the differentiation defect, setd3∆ mESCs expressing a SET domain (105-314 amino acids)-deleted SETD3 (SETD3∆SET) were generated.The expression and length of SETD3∆SET were validated via Western blot (Figure S2G).Using the same endoderm differentiation protocol, cells were differentiated toward the endoderm.Brachyury and Foxa2 expression levels in setd3∆SET cells were similar to setd3∆ cells, suggesting that SETD3 enzymatic activity is important for successful endoderm differentiation (Figure S2H,I).
To detect time-dependent gene expression profiles associated with endoderm differentiation, Short Time-Series Expression Miner (STEM) analysis was conducted on the RNA-seq data of wild-type and setd3∆ cells.Significant profiles (p < 0.05) with upregulation or downregulation pattern during endoderm differentiation time-course in wild-type and setd3Δ groups were selected and subjected to GO analysis.Upregulated genes throughout wild-type endoderm differentiation were more strongly associated with gastrulation, such as cell migration involved in gastrulation and left/right pattern formation (Figure 4A).Pathways essential for endoderm differentiation, such as the Nodal and Activin signaling pathways, were also significantly enriched by the upregulated genes in wild-type cells (Figure 4A).None of these pathways were found in setd3Δ cells.Instead, neuroectodermal lineage-specific processes such as axonal/neuron fasciculation and regulation of action potential were upregulated.Conversely, the Activin receptor signaling pathway was enriched by the downregulated genes in setd3Δ cells, along with other development, morphogenesis, and cell differentiationrelated pathways (Figure 4A).Interestingly, the Wnt signaling pathway was among the significantly downregulated pathways in setd3Δ cells on the third day of differentiation (Figure 4B).
Wild-type co-expression networks were determined via Weighed Gene Co-expression Network Analysis (WGCNA) using the "WGCNA" package on R. The soft threshold was set to 27 (R 2 = 0.90) to construct a scalefree network.The genes with higher similarity in terms of their expression patterns were grouped in the same module.Modules that are very similar to each other were merged into a unique module (Figure 4C).After hierarchical clustering, 48 modules were obtained.1328 genes could not be assigned to any module.Obtained modules were intersected with wild-type/setd3∆ DEGs.More than 50% overlap was observed in 7 modules.These modules were subjected to functional enrichment analysis, and 4 modules were found to be significantly enriched in at least one GO biological process (adjusted p-value < 0.05).Each module represents a disjoint set of enriched biological processes (Figure 4D).The aim was to identify modules in which genes that exhibited statistically significant changes in the absence of SETD3 were coregulated.Consistent with the STEM analysis results, the most significantly affected pathways were related to gastrulation, such as axis specification and embryonic pattern specification.Additionally, endodermal lineage commitmentrelated pathways, such as BMP, Wnt, and Nodal signaling were among the most significant GO terms in the brown module.Other modules mostly contained cytoskeletonrelated terms, such as vesicle localization and actin filament depolymerization.Considering the role of SETD3 in actin modification, these results were expected.
GO analysis performed using upregulated DEGs (wildtype/setd3Δ) revealed that the Wnt signaling pathway is significantly enriched at the second, third, and fourth days of endoderm differentiation (Table 3).Gene set enrichment analysis (GSEA) also indicated that the Wnt signaling pathway is enriched by wild-type/setd3∆ upregulated genes on the second and third days of endoderm differentiation (Figure 5A).We hypothesized that the endoderm differentiation defect observed in setd3∆ cells might be due to insufficient activity of the canonical Wnt signaling pathway.Using previously published βcatenin chromatin immunoprecipitation sequencing (ChIP-seq) data in wild-type mESCs, 24 direct βcatenin target genes were identified and alterations in their expression in setd3∆ mESCs were analyzed.Figure S3A shows the list of βcatenin direct targets with an expression change in setd3∆ mESCs.Furthermore, GSEA using βcatenin direct target genes showed that these genes are enriched in wild-type/setd3∆-upregulated genes (Figure S3B).Consequently, we decided to focus on the effect of SETD3 on the canonical Wnt signaling pathway for the regulation of endoderm differentiation of mESCs.To F I G U R E 1 Differentiation phenotype of setd3∆ mESCs.(A) Western blot of SETD3 in wild-type (WT) and setd3∆ mESCs, with GAPDH as the loading control.(B) Morphology of wild-type and setd3∆ mESC colonies grown in standard medium (100X, Olympus CKX53).RT-qPCR of meso/endodermal pioneer TF Brachyury in wild-type and setd3∆ cells differentiated toward mesoderm (C), and endoderm (D).RT-qPCR of endodermal markers Foxa2 (E) and Sox17 (F) in wild-type and setd3∆ cells differentiated toward endoderm.RT-qPCR of neuroectodermal markers Sox1 (G) and Pax6 (H) in wild-type and setd3∆ cells differentiated toward neuroectoderm.Days of differentiation are shown on the x-axis (Day 0 is mESCs).Error bars indicate the ±SEM of three biological replicates.Nonsignificant p-values were not shown on the graphs.(I) Western blot of SETD3 during endoderm differentiation time-course, with GAPDH as the loading control.examine the transcriptional activity of the canonical Wnt signaling pathway, a luciferase reporter under the control of tandem TCF-binding sites was used.Consistent with the bioinformatic analyses, dual luciferase assay results showed significantly lower reporter activity in setd3∆ mESCs grown in standard medium compared to wild-type (Figure 5B).
Transcriptional activity differences can be explained by altered expression of βcatenin.There was no significant difference in either transcript or total protein levels of βcatenin between wild-type and setd3∆ mESCs grown in standard medium (Figures 5C and S3C).Since βcatenin subcellular localization can also lead to differences in its transcriptional output, 20 βcatenin levels in subcellular fractionations were assessed by Western blot analysis (Figure 5D,E).Although the cytoplasmic βcatenin levels were similar, the nuclear βcatenin pool in setd3∆ mESCs was lower than wild-type, suggesting an effect of SETD3 in βcatenin nuclear translocation and its subsequent transcriptional activity.SETD3 and βcatenin can be found in both the cytoplasm and nucleus.Interestingly, the Proximity Ligation Assay (PLA) showed an interaction between SETD3 and βcatenin in mESCs (Figure 5F), predominantly though not exclusively in the nucleus (Figures 5G and S3D).
mESCs are routinely cultured in LIF and high-serumcontaining standard medium.An alternative medium that contains a GSK3 inhibitor (CHIR99021) and a MEK1/2 inhibitor (PD0325901) along with LIF can maintain pluripotency. 25Wild-type and setd3∆ mESCs were successfully adapted to two-inhibitor and low-serum-containing medium (2i4 medium) (Figure S3E). 17Since GSK3β is a component of the βcatenin destruction complex, its inhibitor in 2i4 medium can lead to the stabilization of βcatenin, increasing its total levels in mESCs (Figure 5C).
Consistent with this increase, the interaction of SETD3 and βcatenin was increased, coincident with the DAPI signal (Figure 5F,G).In agreement with canonical Wnt signaling pathway activation by 2i4 medium, 25 luciferase reporter activity of both wild-type and setd3∆ mESCs was also dramatically increased compared to their counterparts grown in the standard medium (Figure 5B).To test whether SETD3-βcatenin interaction occurs in the nucleoplasm or on chromatin, we performed co-immunoprecipitation experiments but failed to retrieve βcatenin with SETD3.This is likely due to the transient or weak nature of the interaction.To overcome this problem, we performed immunoprecipitation of SETD3 from formaldehydecrosslinked chromatin similar to a ChIP experiment but blotted the co-immunoprecipitated material for the presence of βcatenin.Using this approach, we were able to show βcatenin pulled down with SETD3-bound chromatin fragments but not from the negative control parental chromatin fragments, suggesting at least a fraction of their interaction might occur on chromatin (Figure S3F).
To test whether the endoderm differentiation defect observed in setd3∆ cells is due to insufficient activation of the canonical Wnt signaling pathway, wild-type, and setd3∆ mESCs were cultured in 2i4 medium and differentiated toward endoderm using the same differentiation protocol.Brachyury and Foxa2 expression in setd3∆ cells were rescued to wild-type levels, suggesting increased canonical Wnt pathway activity even at the mESC stage

| DISCUSSION
SETD3 loss results in impaired endoderm differentiation, which was marked by a notable reduction in the expression of Brachyury and Foxa2, pioneer TFs for mesoderm and endoderm formation.Consequently, transcription of a multitude of endodermal marker genes also showed a significant reduction.The rescue of the differentiation defect by the stable re-expression of full-length SETD3 underlines its role in endoderm lineage commitment.On the other hand, SETD3∆NLS re-expression leads to a partial rescue, mainly of Brachyury expression, suggesting SETD3 cytoplasmic function is sufficient for initial mesodermal commitment but not further for endoderm.However, it is also possible that the partial rescue stems from lower SETD3∆NLS levels compared to full-length or endogenous SETD3.This might result from a failure to assemble with appropriate protein complex partners, leading to reduced stability in the partial deletion of proteins. 26Similar F-actin staining in wild-type, setd3∆, and setd3∆ + pSetd3 mESCs suggests that the cytoplasmic function of SETD3 is not directly affecting differentiation capability.Although we cannot completely rule out a contribution from the cytoplasmic role of SETD3 due to the overall low level of exogeneous SETD3∆NLS protein, our results suggest that nuclear SETD3 is important for endoderm differentiation.
Our rescue experiments using the SET methyltransferase domain-deleted SETD3 showed that although this construct was expressed at levels comparable to wild-type, it is not sufficient for a successful differentiation phenotype.SET domain deletion results in an enzymatically dead SETD3, affecting both the cytoplasmic and nuclear methylation functions.Though we cannot speculate on the possible histone/non-histone targets, the result suggests the catalytic function of SETD3 is important for endodermal commitment.
Since SETD3 is expressed throughout early endodermal differentiation, it is not clear what stage the differentiation defect originates from.PCA and heat-map clustering of time-course RNA-seq results showed that the samples from the second day endoderm differentiation were more similar to ESC state samples compared to the samples from the third or fourth day.This suggests that the defect arises around day two, preceding the administration of Activin A, which guides differentiation toward endodermal lineage. 27Rescue of setd3∆ endodermal differentiation defect by mESC culture in 2i4 medium shows that the SETD3 function is critical at the mESC stage rather than after pluripotency exit.Previous reports similarly showed that changes at the ESC stage can affect differentiation efficiency at later stages. 28n ESC differentiation defect could be linked to improper deactivation of the pluripotency network.Our results do not support extensive pluripotency network activation in setd3∆ cells.However, Oct4 expression remained high during differentiation.Although mainly a pluripotency factor, OCT4 is also important for proper differentiation. 3An activator of Oct4 transcription, NR5A2, is significantly upregulated in setd3∆ cells, which might explain sustained Oct4 expression. 29other reason for the differentiation defect of setd3Δ mESCs might be aberrant upregulation of neuroectodermal lineage-specific pathways.Our bioinformatic analysis showed the enrichment of many neuroectoderm-related pathways in setd3∆ cells compared to wild-type.However, RT-qPCR of key genes failed to show meaningful upregulation in setd3Δ cells (data not shown), arguing against a robust activation of the neuroectodermal network but rather suggesting its incomplete suppression.Comparable levels of neuroectodermal markers between wild-type and setd3Δ mESCs suggest that, at the time of first lineage decision, the presence of SETD3 is important for differentiation toward mesendodermal lineages rather than the neuroectodermal lineage.
A possible function of SETD3 in endoderm differentiation might be through the canonical Wnt signaling pathway.Wnt pathway activity in mESCs plays a crucial role in inducing mesendoderm differentiation, as Brachyury and Foxa2 are direct targets of the TCF/LEF-mediated canonical Wnt signaling pathway. 30,31A reduction in the canonical Wnt signaling pathway activity likely disrupts the expression of target endodermal lineage genes and might explain how SETD3 loss connects with the meso/ endodermal differentiation defect.mESCs cultured in 2i4 medium show significantly higher canonical Wnt signaling pathway activation measured through the luciferase reporter assay compared to their counterparts grown in standard medium.The observation that both wild-type and setd3∆ mESCs grown in 2i4 medium can successfully differentiate into endoderm further suggests a potential role for the canonical Wnt signaling pathway.setd3∆ mESCs grown in 2i4 medium still show lower pathway activation than wild-type cells; however, the amount of canonical Wnt activity in these cells seems sufficient for sustaining a successful mESC differentiation toward endoderm.
Our STEM analysis showed that setd3Δ cells are defective in Activin signaling.Since the differentiation protocol uses a high concentration of Activin A for the derivation of definitive endoderm cells, this means setd3∆ mESCs are somehow desensitized to it.Activin A belongs to the TGF-β superfamily of cytokines and exerts its function through SMADs.Our differential expression data suggested an upregulation of the inhibitory SMAD, Smad7, in setd3Δ cells. 32The expression levels are also confirmed via RT-qPCR (data not shown).As an inhibitor of Activin A signaling, upregulation of Smad7 levels might contribute to defective endoderm differentiation in the absence of SETD3.
Similar to many histone modifiers, SETD3 is known to catalyze the modification of non-histone proteins as well.Non-histone protein methylation can affect protein stability, localization, and their interaction with other proteins. 33,34Our PLA result revealed an interaction between SETD3 and βcatenin in the nucleus and cytoplasm of mESCs.Nuclear βcatenin pool was diminished in setd3∆ mESCs.Intriguingly, methylation of βcatenin or its interaction partners by other SET-domain-containing methyltransferases facilitates its nuclear localization. 33It would be interesting to explore whether SETD3 might target βcatenin for nuclear shuttling in a similar manner.
Our mass spectrometry analysis of nuclear SETD3 interaction partners (unpublished data from the lab) revealed SETD3 and βcatenin share common interaction partners, including BCL9, CREBBP, and TCF3, all transcriptional co-mediators of βcatenin.Additionally, H3K4me3 is associated with the activation of Wnt target genes. 35Therefore, it is plausible that SETD3 might facilitate H3K4me3 deposition on βcatenin target genes, leading to transcriptional activation by engaging in a multiprotein complex with βcatenin and their common interaction partners.

| CONCLUSION
In this study, we demonstrated that the absence of SETD3 affects various signaling pathways.We focused on the canonical Wnt signaling pathway and showed that SETD3 affects the subcellular localization of βcatenin rather than its transcript or total protein level.Reduced nuclear βcatenin levels in setd3∆ mESCs lead to a marked decrease in the canonical Wnt pathway transcriptional activity.The observed defect in endoderm differentiation resulting from the absence of SETD3 could potentially be attributed to the diminished canonical Wnt signaling pathway activity.These findings contribute to the growing body of evidence for SETD3 function and hold relevance for understanding the mESC lineage commitment mechanisms.
F I G U R E 5 SETD3 and the canonical Wnt signaling pathway.(A) Gene set enrichment analysis (GSEA) of wild-type/setd3Δ upregulated genes at the second and third days of endoderm differentiation.(B) Canonical Wnt pathway activity in wild-type and setd3∆ mESCs grown in either standard or 2i4 medium using a dual luciferase assay.Relative luciferase activity corresponds to firefly luciferase activity (Wnt reporter) normalized to transfection control renilla luciferase activity.Western blot analysis of βcatenin in (C) whole cell lysates, (D) cytoplasmic, and (E) nuclear fractions of wild-type and setd3∆ mESCs grown either in standard or 2i4 medium.GAPDH and H3 were used as loading and fractionation controls.Western blot quantification was performed using Image J software.(F) Proximity ligation assay (PLA) using SETD3 and βcatenin antibodies in wild-type mESCs grown either in standard or 2i4 medium.Red dots show SETD3βcatenin interactions, and blue (DAPI) shows the nuclei of the cells in the mESC colony.For negative control, antibody diluent without antibodies was used.(G) Quantification of total and nuclear PLA signals per cell.The colony grown in standard medium had 36 cells and 2i4 medium had 28 cells.Quantification was performed using Andy's algorithm.RT-qPCR of Brachyury (H) and Foxa2 (I) in wild-type and setd3Δ cells differentiated toward endoderm.Days of differentiation are shown on the x-axis (Day 0 is mESCs).Error bars indicate the ±SEM of at least two biological replicates.

F I G U R E 2
Endoderm differentiation time-course RNA-seq analysis of setd3∆ mESCs.(A) Principal Component Analysis (PCA) of endoderm differentiation of wild-type and setd3Δ mESCs.(B) Number of differentially expressed genes (DEGs) between wild-type and setd3Δ cells during endoderm differentiation.The overlaps among the sets of DEGs across different time points are illustrated in an upset plot (https:// upset.app/ ).Horizontal bars represent the total number of DEGs at each time point, while vertical bars represent the size of set intersections.A single dot represents DEGs unique to a particular set, and connected dots represent the intersections of corresponding sets.(C) Heat map of the top DEGs among comparison groups and their hierarchical clustering.Integrative Genomics Viewer (IGV) visualization of RNA-seq tracks of endoderm (D), pluripotency (E), and neuroectoderm-related (F) genes in wild-type and setd3∆ cells.T A B L E 2 Validated DEGs.

F I G U R E 3
Validation of RNA-seq results and rescue experiments.(A) Western blot of SETD3 in wild-type, setd3∆, and rescue (setd3Δ + pSetd3) mESC whole cell, cytoplasmic, and nuclear extracts.GAPDH and H3 were used as cytoplasmic and nuclear loading controls, respectively.The empty arrowhead shows endogenous SETD3, while the filled arrowhead indicates exogenous Flag-tagged SETD3.RT-qPCR of selected endoderm differentiation marker transcripts (B-I) and Oct4 (J) during endoderm differentiation in wild-type, setd3∆, and rescue cells.Days of differentiation are shown on the x-axis (Day 0 is mESCs).Error bars indicate ±SEM of two biological replicates.
Wnt pathway-related Gene Ontology terms.Further bioinformatic analyses.(A) Short Time-Series Expression Miner (STEM) analysis during endoderm differentiation time-course.Genes in profiles showing upregulation and downregulation were subjected to pathway enrichment analysis using the built-in GO enrichment function of STEM software.Significantly enriched GO processes were visualized.(B) Pathway enrichment analysis of DEGs upregulated in wild-type/setd3Δ during endoderm differentiation time-course.(C) Identification of gene modules with WGCNA.Dynamic colors represent the unique clusters of highly similar gene modules.Each module was identified with hierarchical clustering.(D) Functional enrichment of GO biological processes in four modules that were identified due to their significant enrichment with a high number of DEGs.Smaller adjusted p-values were represented with darker colors in the map.