UBE3D Regulates mRNA 3'-End Processing and Maintains Adipogenic Potential in 3T3-L1 Cells

Abstract

We have previously described the role of an essential Saccharomyces cerevisiae gene, important for cleavage and polyadenylation 1 (IPA1), in the regulation of gene expression through its interaction with Ysh1, the endonuclease subunit of the mRNA 3'-end processing complex. Through a similar mechanism, the mammalian homolog ubiquitin protein ligase E3D (UBE3D) promotes the migratory and invasive potential of breast cancer cells, but its role in the regulation of gene expression during normal cellular differentiation has not previously been described. In this study, we show that CRISPR/Cas9-mediated knockout of Ube3d in 3T3-L1 cells blocks their ability to differentiate into mature adipocytes. Consistent with previous studies in other cell types, Ube3d knockout leads to decreased levels of CPSF73 and global changes in cellular mRNAs indicative of a loss of 3'-end processing capacity. Ube3d knockout cells also display decreased expression of known preadipogenic markers. Overexpression of either UBE3D or CPSF73 rescues the differentiation defect and partially restores protein levels of these markers. These results support a model in which UBE3D is necessary for the maintenance of the adipocyte-committed state via its regulation of the mRNA 3'-end processing machinery.

Keywords: RNA processing; adipogenesis; cell differentiation.

FIG 1

Ube3d expression is dynamic during 3T3-L1 differentiation. (A) Oil Red O staining of 3T3-L1 cells showing accumulation of fat droplets starting on day 4 of the differentiation protocol and increasing dramatically on days 7 and 10. Representative images are shown. (B) UBE3D protein levels are high in undifferentiated 3T3-L1 cells and decrease to undetectable levels during differentiation. FABP4 is shown as a marker of adipocyte differentiation. (C) Ube3d mRNA levels decrease during differentiation. Relative expression levels normalized to 18S rRNA levels are shown (n = 3; mean ± SEM). ns, not significant (one-way analysis of variance [ANOVA] with Dunnett’s multiple-comparison test).

FIG 2

Ube3d knockout (KO) cells are unable to differentiate into mature adipocytes. (A) UBE3D levels are undetectable in Ube3d-KO cells, and FABP4 levels are greatly decreased compared to UBE3D-expressing cells (WT) after induction of differentiation for 7 days. Representative Western blotting is shown. (B) Ube3d-KO cells do not accumulate fat droplets as evidenced by a lack of Oil Red O staining on day 10 of differentiation. Representative high-power fields are shown. (C) mRNA levels of adipocyte differentiation markers are significantly decreased in Ube3d-KO versus WT cells. Relative expression levels normalized to 18S rRNA levels are shown (n = 3; mean ± SD). All values at day 7 were significantly different from WT (P < 0.0001, two-way ANOVA with Šídák's multiple-comparison test). (D) Representative images of K_i-67 and phosphohistone H3 (pHH3) staining and quantification of the percentage of positive nuclei in WT and Ube3d-KO cells 24 h after treatment with induction medium. There were no significant differences between WT and KO cells (two-way ANOVA with Šídák's multiple-comparison test; ns, P > 0.05).

FIG 3

Ube3d or CPSF73 complementation rescues differentiation in Ube3d-KO cells. (A) UBE3D and CPSF73 levels in undifferentiated WT and Ube3d-KO cells overexpressing mCherry, UBE3D, and CPSF73. Cells were grown in log phase and treated with doxycycline to induce overexpression for 4 days before collection. Representative blots are shown. (B) Withdrawal of doxycycline prior to the start of differentiation allows for a gradual decrease in overexpressed UBE3D protein levels. Ube3d-KO cells with the doxycycline-inducible UBE3D cassette were treated with doxycycline for 4 days before the start of differentiation. On day −2, when cells reach confluence, doxycycline was withdrawn, and the cells were differentiated. (C) UBE3D and CPSF73 overexpression rescues differentiation of Ube3d-KO cells. Representative wells with Oil Red O staining on day 10 are shown. (D) CPSF73 levels are decreased in Ube3d-KO cells. Representative blots of UBE3D and CPSF73, along with total protein staining as a loading control, are shown. (E) Quantification of CPSF73 levels from three independent experiments graphed as CPSF73 signal normalized to tubulin. Mean ± SEM; ***, P = 0.0004 (two-tailed t test) (F) Markers of adipocyte differentiation PPARγ2 and FABP4 are decreased in Ube3d-KO control cells overexpressing mCherry on day 7 of differentiation but restored by UBE3D overexpression. CPSF73 overexpression partially rescues FABP4 levels. The ubiquitously expressed PPARγ1 isoform is also detected by the antibody. Representative blots are shown. (G) Quantification of the PPARγ2 signal from two replicates is shown normalized to the WT.

FIG 4

RNA-seq shows that Ube3d-KO causes shifts to downstream poly(A) site use. (A) RNA sequencing of undifferentiated WT and Ube3d-KO cells was analyzed using APAlyzer to quantify alternative polyadenylation in the 3′ UTR (3′UTR APA). Relative expression difference (RED) scores are plotted for each gene, where a positive score indicates 3′ lengthening, and a negative score indicates shortening. Genes with significant changes (P < 0.05) are highlighted (positive RED scores, indicating lengthening, in pink; negative RED scores, indicating shortening, in teal). (B) RED scores (shown as log₂ fold changes) calculated by APAlyzer for genes categorized as lengthened, shortened, or undergoing no significant change. (C) Coverage plot for Hdlbp showing 3′-UTR lengthening (RED = 0.617; P_adj = 0.032). The upstream poly(A) site considered by APAlyzer is shown with a dashed vertical line. Three biological replicates are overlaid to show areas with common coverage, with read counts given on the y axis. (D) Volcano plot showing terminal exon length differences for each gene calculated by PAQR.

FIG 5

Ube3d knockout leads to readthrough transcripts. (A) Relationship between change in gene expression and change in DoG read counts. DoGs with significantly increased or decreased read counts (absolute value of log₂ fold change > 0.5; P_adj < 0.05) are shown in orange and blue, respectively. A cluster of DoG-gene pairs with significantly increased DoG levels in Ube3d-KO cells but little or no increase in gene expression are found in the region highlighted by the inset. (B) Coverage plots for Rpl9 and Src DoG regions (black bars, bottom). For each DoG region, read counts are shown on a logarithmic scale [ln(read counts + 1)], and three biological replicates are overlaid to show regions with shared coverage.

FIG 6

Analysis of ontology enrichment for genes with altered poly(A) site use and the relationship to changes in gene expression. (A) GO term enrichment of genes whose mRNAs were lengthened or shortened in the Ube3d-KO cells. Gene sets were uploaded to Metascape and analyzed for enrichment using gene lists from PaGenBase. Enriched gene lists with greater than 3 terms in the uploaded set, P values of <0.01, and an enrichment factor of >1.5 were clustered based on membership similarity and graphed according to P value. (B) Alternative polyadenylation (quantified by APAlyzer RED score) is not correlated with gene expression changes. Log₂-transformed difference in RED or gene expression between KO and WT cells are graphed on the x and y axis, respectively. Genes with RED P values of <0.05 are highlighted in blue.

FIG 7

RNA-seq shows an altered gene expression profile and dysregulated gene signatures. (A) Volcano plot showing genes with increased expression (right quadrant, magenta) and decreased expression (left quadrant, teal). (B) GO term enrichment of downregulated genes. (C) GO term enrichment for upregulated genes. In panels B and C, enrichment maps were generated using the R packages clusterProfiler and enrichplot. Nodes represent enriched GO terms (colored by adjusted P value), and edges represent the Jaccard similarity coefficient between two nodes. Representative terms are labeled from each cluster. (D) Enrichment for tissue- and cell-specific gene lists in downregulated genes (top) and upregulated genes (bottom). Gene sets were uploaded to Metascape and analyzed for enrichment using gene lists from PaGenBase. Enriched gene lists with greater than 3 terms in the uploaded set, a P value of <0.01, and an enrichment factor of >1.5 were clustered based on membership similarity and graphed according to P value.

FIG 8

Ube3d KO cells exhibit decreased markers of committed adipocyte precursors. (A) A subset of genes shown by Gupta et al. (38) to be depleted in nonadipogenic 3T3 cells (reported fold change values in open circles) are also downregulated in Ube3d-KO cells [gray bars; genes with log₂(fold change) of less than −0.5 and adjusted P value of <0.05 are marked with asterisks]. (B) Genes with known roles in the regulation of adipogenesis also exhibit decreased expression. (C) Some genes showing decreased RNA levels show decreased total protein levels in Ube3d-KO cells (C/EBPδ, HGFɑ, EFEMP1, and LPL), which are restored upon overexpression of UBE3D or CPSF73. Other genes do not show a similar change (ZFP423, SLPI, and GPRIN3). Representative Western blots are shown.