Transient deSUMOylation of IRF2BP proteins controls early transcription in EGFR signaling

Abstract

Molecular switches are essential modules in signaling networks and transcriptional reprogramming. Here, we describe a role for small ubiquitin-related modifier SUMO as a molecular switch in epidermal growth factor receptor (EGFR) signaling. Using quantitative mass spectrometry, we compare the endogenous SUMO proteomes of HeLa cells before and after EGF stimulation. Thereby, we identify a small group of transcriptional coregulators including IRF2BP1, IRF2BP2, and IRF2BPL as novel players in EGFR signaling. Comparison of cells expressing wild type or SUMOylation-deficient IRF2BP1 indicates that transient deSUMOylation of IRF2BP proteins is important for appropriate expression of immediate early genes including dual specificity phosphatase 1 (DUSP1, MKP-1) and the transcription factor ATF3. We find that IRF2BP1 is a repressor, whose transient deSUMOylation on the DUSP1 promoter allows-and whose timely reSUMOylation restricts-DUSP1 transcription. Our work thus provides a paradigm how comparative SUMO proteome analyses serve to reveal novel regulators in signal transduction and transcription.

Keywords: ATF3; DUSP1; EGFR; IRF2BP1; SUMO.

Figure 1. EGF induces rapid and transient deSUMOylation of transcriptional regulators in HeLa cells

1. Schematic representation of a quantitative proteome analysis that compares the SUMO proteome of serum‐starved HeLa cells treated for 10 min with or without 100 nM EGF (SILAC labeling and endogenous SUMO1‐ and SUMO2/3‐IPs).

2. Scatterplots represent SILAC/SUMO‐IP quantification after EGF treatment from three biological replicates. Each dot represents a protein that is either present in the same ratio between untreated and EGF‐treated cells (dark blue dots) or is significantly more present in one of the samples (red and yellow dots).

3. Left panel: Bar graph depicting mass spectrometry results of 11 proteins with altered SUMOylation (significant hits with P < 0.0001), as well as the three non‐changing proteins SUMO2, RanGAP1, and TRIM28. Right panel: Proteins highlighted in bold in the left panel were validated by SUMO‐IP/immunoblotting from serum‐starved HeLa cells without or with 10‐min EGF stimulation.

4. Time course experiment: Serum‐starved HeLa cells were treated with 100 nM EGF, and samples were harvested at indicated times and subjected to SUMO2 IP followed by immunoblotting with the indicated antibodies. IRF2BP1, IRF2BP2, and TRIM24 are rapidly but transiently deSUMOylated upon EGF treatment.

Figure 2. IRF2BP proteins are SUMOylated at their highly conserved C‐terminus

1. Schematic representation of the domain structure of IRF2BP1 and IRF2BP2. The primary sequence suggests two putative SUMO sites that are conserved among different species.

2. Identification of the endogenous SUMO sites in IRF2BP1 (left panel) and IRF2BP2 (right panel). HeLa cells were transfected with HA‐tagged wild‐type (WT) or mutant (KR) proteins, endogenous SUMO‐IPs were performed, and the HA‐signal was analyzed by immunoblotting. Mutation of the C‐terminal SUMO site abolished SUMOylation of IRF2BP1 and IRF2BP2 in HeLa cells.

3. Clustal omega analysis of IRF2BP1 and IRF2BP2 from Homo sapiens, Mus musculus, Gallus gallus, Xenopus laevis, Danio rerio, Drosophila melanogaster and Caenorhabditis elegans shows high conservation of the C‐terminal region including the SUMO site.

4. Schematic representation of the creation of stable, untagged, and siRNA‐resistant IRF2BP1 WT and K579R HeLa cells. Constructs expressing IRF2BP1 variants in an pIRES‐hrGFP II (“pIRES”) vector were transfected, selected with antibiotics, and FACS sorted for low GFP expression.

5. Stable HeLa cells expressing pIRES‐empty vector, IRF2BP1 WT, or IRF2BP1 K579R were treated with siRNA against endogenous IRF2BP1 or non‐targeting (nt) siRNA. Exogenous siRNA‐resistant IRF2BP1 was expressed at low levels similar to endogenous IRF2BP1. * refers to an unspecific band.

6. Wt and mutant IRF2BP1 localizes in the nucleus. After knockdown of endogenous IRF2BP1, stable IRF2BP1 (WT or K579R) cell lines were immunostained for IRF2BP1. Exogenous IRF2BP1 variants show a similar nuclear localization. Scale bar = 10 µm.

7. IRF2BP1 wt and mutant associate with chromatin to a similar extent. HeLa cells were lysed in 0.075% NP40 (Input). After centrifugation, the nuclei were incubated and vortexed with a nuclear extract (NE) buffer containing 170 mM NaCl. The eluates were collected, and the procedure was repeated using a NE buffer with higher salt concentrations, first 290 mM, then 420 mM. Wild‐type IRF2BP1, the SUMO‐deficient K579R mutant and the SUMOylated wild‐type form (*) all behave similarly.

Figure EV1. Transcriptome analyses of HeLa cells expressing either wt IRF2BP1 or the SUMOylation‐deficient variant IRF2BP1 K579R

1. A

   Stable IRF2BP1 cell lines were used to perform microarray analyses in asynchronously growing cells (full serum) and under serum starvation, with or without EGF treatment for 1 h.

2. B–D

   GO analysis revealed that differentially expressed genes cluster in similar biological processes, such as adhesion, proliferation, and signaling processes, irrespective of whether they were grown in full serum (B), under serum starvation (C), or under serum starvation and EGF treatment (100 ng/ml) for 1 h (D).

Figure 3. IRF2BP1 deSUMOylation contributes to transcriptional induction of immediate early genes

1. Stable IRF2BP1 cell lines (knocked down for endogenous IRF2BP1) were used to perform a microarray experiment under serum starvation and upon treatment with EGF for 1 h. A subset of 38 EGF‐dependent genes is differentially regulated in IRF2BP1 wild‐type cells compared to IRF2BP1 K579R cell lines (at least 1.5‐fold). Among them are DUSP1 (arrow), ATF3, Fos, and Egr2. Each microarray was performed in triplicates, and the bars show log₂ values of the fold changes for wt cells (black bars) and for KR cells (gray bars). For details, see Materials and Methods.

2. Chromatin IP reveals association of human IRF2BP1 with the proximal DUSP1 promoter in HeLa cells. Gene architecture of human DUSP1. The primers at ‐243/‐67 (“‐67”), ‐473/‐224 (“‐224”), and ‐1,170/‐961 (“‐916”) relative to the TSS were used for ChIP experiments. IRF2BP1 and IRF2BP2 bind to the promoter region of DUSP1 between nucleotides ‐243 and ‐67.

3. ChIP/qPCR experiments reveal preferential IRF2BP1 binding to the promoter region of DUSP1 between ‐243 and ‐67. Data show mean (bar) and individual data points from two biological replicates.

4. IRF2BP1 wild type and the K579R mutant both bind the DUSP1 promoter in the absence or presence of EGF. Stable cell lines expressing IRF2BP1 wild type or K579R were knocked down for endogenous IRF2BP1, followed by IRF2BP1 ChIP and DUSP1 qPCR of its promoter region ‐243 and ‐67. Data show means ± SEM from three biological replicates.

5. qPCR data after EGF treatment in IR2BP1 wild type and K579R cell lines after knockdown of the endogenous IRF2BP1. Data show means ± SEM from three biological replicates.

6. IRF2BP1 wild type and V578A cell lines after knockdown of the endogenous IRF2BP1 were analyzed for DUSP1 transcription after EGF treatment by qPCR (left panel), data show means ± SEM from five biological replicates (left panel), and IRF2BP1 protein levels. As expected, IRF2BP1 V578A is not SUMOylated (right panel). * refers to an unspecific band.

7. qPCR data for immediate early genes after EGF treatment in the IR2BP1 wild type and V578A cell lines after knockdown of the endogenous IRF2BP1. Data show means ± SEM from four biological replicates.

Figure EV2. Chromatin IP reveals association of human IRF2BP1 and IRF2BP2 with the proximal ATF3 promoter in HeLa cells

Gene architecture of human ATF3. The primers at ‐358/‐241 (“ATF3‐TSS‐241”) relative to the TSS were used for ChIP experiments.

Figure 4. IRF2BP proteins negatively regulate expression of the EGF receptor

1. Immunoblotting of HeLa lysates at indicated times after EGF treatment reveals enhanced DUSP1 and ATF3 expression upon knockdown of IRF2BP1. Uba2 and beta‐actin serve as independent loading controls.

2. Model. IRF2BP1 is a SUMO‐dependent transcriptional repressor of immediate early genes (IEGs). Transcription of IEGs is repressed by SUMOylated IRF2BP1, which binds to its proximal promoter. EGF receptor signaling yields at least two signals to induce IEG expression, one of which is the transient deSUMOylation of IRF2BP1.

3. Wound healing assay: HeLa cells were incubated with nt or IRF2BP1 siRNA for 2 days, grown to 90% confluency, and analyzed for wound closure with or without addition of 100 ng/ml EGF. Data show means of the relative wound density ± SEM from four biological replicates.

4. Proliferation assay: HeLa cells upon knockdown of IRF2BP1 and IRF2BP1 were analyzed for cell density over 6 days in growth medium. One representative biological experiment is shown with means ± SEM from five technical replicates.

5. IRF2BP1, IRF2BP2, and IRF2BP1 + IRF2BP2 were knocked down in HeLa cells for 72 h, and gene expression data were recorded by microarray analysis; non‐targeting siRNAs were used as a control. IRF2BP1 and IRF2BP2 have distinct and overlapping functions. Experiments were done in triplicates.

6. GSEA of the highly significant enriched EGFR signaling pathway. IRF2BP1 knockdown correlates with an increase of genes involved in EGFR signaling.

7. Validation of two candidate genes: IRF2BP1 and IRF2BP2 were knocked down for 72 h in HeLa cells, and cell lysates were analyzed by immunoblotting with the indicated antibodies.

8. FACS‐based analysis of EGFR surface expression: IRF2BP1 and IRF2BP2 were knocked down for 48 h in HeLa cells, and cells were serum‐starved for 16 h, stained with fluorescent anti‐EGFR antibodies, and analyzed by flow cytometry. Shown are means of three independent experiments ± SEM.

Figure EV3. Model

DeSUMOylation of IRF2BP1 occurs within 15 min after EGF treatment. This transient event directly correlates with the gene expression of immediate early genes.

Figure EV4. Transcriptome analyses of Hela cells with or without depletion of IRF2BP1, IRF2BP2, or both proteins

1. A

   Schematic representation of the microarray experiment that was performed in this study (three independent experiments): IRF2BP1, IRF2BP2, and IRF2BP1 + IRF2BP2 were knocked down in HeLa cells for 72 h, and gene expression data were recorded by microarray; non‐targeting siRNAs were used as a control.

2. B

   Venn diagram of genes that are differentially expressed (at least 1.5‐fold, FDR < 0.05) upon IRF2BP‐knockdown compared to control. A large number of genes are dependent on IRF2BP proteins, and they can be overlapping or distinct for IRF2BP1 and IRF2BP2.

3. C, D

   Gene set enrichment analyses (GSEA) of the microarray data using KEGG gene set. Many signaling pathways were enriched upon knockdown of IRF2BP1 (C) and IRF2BP2 (D). The first 15 hits ranked by Normalized Enrichment Score (NES) are shown.