The WWP1-JARID1B axis sustains acute myeloid leukemia chemoresistance

Abstract

To uncover substrates mediating the oncogenic activity of WWP1 in acute myeloid leukemia (AML), we performed a proteomic analysis that identified the histone demethylase KDM5B/JARID1B as a candidate target. Of note, JARID1B is indispensable for efficient recruitment of several DNA damage repair factors and for damage resolution, thus negatively influencing the sensitivity of cancer cells to chemo- and radiation therapies. Validation of JARID1B as a substrate of WWP1 revealed a positive regulation of JARID1B half-life by WWP1 through the establishment of K63-linked polyubiquitin chains. As a result, downregulation of JARID1B rising from WWP1 inactivation was associated with higher H3K4me3 enrichment at JARID1B target genes in WWP1-depleted relatively to control AML cells. Integration of RNA-seq and H3K4me3 ChIP-seq data uncovered a highly significant overlap between upregulated gene expression and enriched H3K4me3 peaks after shWWP1 inactivation. We confirmed transcriptional activation of JARID1B targets in WWP1-depleted cells, supporting a role for WWP1 in regulating JARID1B activity. Coherently, upon WWP1 inactivation, we observed a defective recruitment of repair proteins after DNA damage, with subsequent reduced DNA damage repair efficiency and enhanced sensitization of AML cells to the cytotoxic activity of chemotherapeutic drugs. All together, these data identify JARID1B as a bona fide target of WWP1 and imply that WWP1-mediated regulation of JARID1B impacts its ability to modify chromatin and to recruit DNA damage repair factors, thus ultimately affecting chemosensitivity of AML cells.

Keywords: DNA repair; acute myeloid leukemia; protein degradation; protein ubiquitination.

Fig. 1.

Identification of JARID1B as a WWP1 substrate. (A and B) Proteomic analysis (UbiScan^® Cell Signaling Technology) identifies JARID1B as a candidate substrate of WWP1. (A) Volcano plot showing intensity and enrichment of ubiquitinated peptides in shWWP1 versus control sample. Beige shading highlights putative WWP1 targets. Transcriptional regulators are shown in magenta. (B) Heatmap showing Z-scores of intensities of identified peptides belonging to transcriptional regulator set of proteins. (C and D) Representative western blots for the coimmunoprecipitation assays. HEK293T cells were transfected with Flag-WWP1 (C) or HA-JARID1B (D). Cellular extracts were pulled down with anti-Flag (C) and anti-HA (D) antibodies and then analyzed by western blot with anti-JARID1B and anti-WWP1 antibody, respectively. GADPH was used as loading control. Three independent experiments were carried out. (E) Representative western blot showing binding of endogenous JARID1B and WWP1 in NB4 cells. Cellular lysates were immunoprecipitated with anti-JARID1B or an IgG isotype control antibody and then subjected to western blot with anti-WWP1 and anti-JARID1B antibodies. Vinculin was used as loading control. Three independent experiments were carried out. (F) Representative western blots of the in vitro pull-down experiments. Purified recombinant WWP1 and JARID1B proteins were incubated overnight at 4 C. WWP1 was then immunoprecipitated by using anti-WWP1 antibody conjugated with protein A beads and the immunocomplexes were subjected to western blot with anti-JARID1B antibody. Isotypic IgG conjugated with Protein A beads were used as a negative control. Three independent experiments were carried out. (G) Representative western blots of coimmunoprecipitation assays from HEK293T cells cotransfected with Flag-WWP1 and either wild-type or the three PXY mutants of HA-JARID1B. Total cellular extracts were immunoprecipitated with anti-Flag antibody and subsequently analyzed by western blot with anti-HA antibody. Three independent experiments were carried out.

Fig. 2.

WWP1 modifies JARID1B via Lys63 polyubiquitin linkages. (A) Representative western blot analysis of the in vivo ubiquitination of JARID1B by WWP1. HEK293T cells were cotransfected with the indicated plasmids. Ubiquitinated JARID1B was assessed by immunoprecipitation with anti-JARID1B antibody, followed by western blot with a K63-linkage specific polyubiquitin antibody. Three independent experiments were carried out. (B) Ubiquitinated JARID1B was assessed in control and WWP1-depleted NB4 cells following immunoprecipitation of cellular extracts with anti-JARID1B or an IgG isotype control antibody. Immunoprecipitates were then subjected to western blot with anti-ubiquitin antibody (C) Representative in vitro ubiquitination assay. Purified recombinant JARID1B and WWP1 proteins were incubated in the presence of E1 (Ube1), E2 (UbcH5), and either wild-type ubiquitin or Ub K-R ubiquitin mutants for 2 h at 30 °C. Negative controls were included: time 0 reaction (lane 1) and reaction carried out for 2 h in the absence of ubiquitin (lane 2). In vitro ubiquitination reactions were then subjected to SDS-PAGE and immunoblotted with JARID1B antibody. (D) Schematic representation of JARID1B modification by WWP1 through the generation of K63-linked ubiquitin chains.

Fig. 3.

WWP1 silencing leads to JARID1B protein destabilization. (A) Representative western blot analysis of JARID1B protein levels in control and WWP1-depleted NB4 cells (Left panel). Two different clones of NB4/Tet-On/shWWP1 cells were analyzed. Three independent experiments were carried out. Fold changes were quantified using ImageJ software (Right panel). (B) Representative western blot analysis of JARID1B protein levels in chromatin-free soluble and in chromatin fractions extracted from control and WWP1-depleted NB4. (C) mRNA levels of KDM5B in control and shWWP1-depleted NB4 cells were analyzed by RT-qPCR. Relative abundances of transcripts were quantified and normalized to TBP. Values represent the mean ± SD from three independent experiments. (D) Detection of KDM5A, KDM5B, KDM5C, and KDM5D transcript levels by RT-qPCR. Relative abundances of mRNAs were normalized to TBP. Values represent the mean ± SD from three independent experiments. (E) Representative western blot analysis of JARID1A, JARID1B, JARID1C, JARID1D, and KDM2B protein levels following WWP1 silencing in NB4 cells (Left panels). Protein level fold changes were quantified using ImageJ software and normalized to β-actin expression (Right panel). Statistically significant differences were calculated by Student’s t test. (F) Representative western blot analysis of WWP1 and JARID1B protein levels in control and WWP1 overexpressing HEK293T cells after protein synthesis blockade with CHX (50 µg/ml). Vinculin was used as loading control. Statistically significant differences were calculated by Student’s t test. *P < 0.05, ***P < 0.001, ****P < 0.0001.

Fig. 4.

WWP1 regulates the histone demethylase activity of JARID1B. (A) Representative western blot analysis of H3K4me3 protein levels following WWP1 silencing in NB4 cells. Three independent experiments were carried out. (B) Distribution of the H3K4me3 signal around genome-wide promoters as assessed by ChIP-sequencing in WWP1-depleted relatively to control AML cells. (C) Box plots showing general H3K4me3 enrichment at JARID1B target genes in control and WWP1-knockout NB4 cells (Dataset S2). Y-axis units represent log2 H3K4me3 average coverage values in control and shWWP1 NB4 cells. (D) H3K4me3 enrichment at JARID1B selected target genes after WWP1 silencing in NB4 cells. Y-axis units represent log2 H3K4me3 coverage average values. (E) Genomic snapshots of plotted ChIP-seq tracks of H3K4me3 across the indicated JARID1B target genes. (F) Venn diagram showing overlap of upregulated DEGs in shWWP1 NB4 cells (log2FC > 1 and FDR < 0.05), increased H4K4me3 enrichment after WWP1 inactivation (shrunken log2FC > 0.3), and the genome-wide occupancy of JARID1B in K562 leukemia cells (KDM5B predicted targets from ChIP-seq data in K562 cells). The box indicates JARID1B target genes in common among the three datasets. (G) Real-time RT-PCR analysis of selected JARID1B-regulated genes following WWP1 gene silencing in NB4 cells. Relative abundances of mRNAs were normalized to TBP and expressed as fold changes of shWWP1 cells relative to controls (dotted green line). Values represent the mean ± SD from three or four independent experiments (H) Volcano plot of genes differentially expressed in WWP1 low versus WWP1 high expressing AML patients from TCGA. Significantly enriched genes are shown in magenta (enriched in WWP1 low expressing samples) or green (enriched in WWP1 high expressing samples). Padj < 0.05 and abs(log2FC) > 1 were used as thresholds. (I) Dot plot showing correlations between CDKN1A and WWP1 log2 expression levels in TCGA AML samples. Samples are colored by density of neighbors using ggpointdensity algorithm. Pearson correlation R and P value are shown. (J) Real-time RT-PCR analysis of CDKN1A expression in human peripheral blood mononuclear cells from healthy donors (n = 12), WWP1 low (n = 10), or WWP1 high (n = 20) AML samples. Relative abundances of mRNAs were normalized to TBP and expressed as relative values. Values are shown as violin plots. P value by the Kruskal–Wallis test. The Y-axis represents log2RPKM values. *P < 0.05, **P < 0.01, ****P < 0.0001.

Fig. 5.

WWP1 inactivation impairs the assembly of DDR foci in response to DNA damage. (A–C) Representative confocal images of DDR foci detected in BLM (40 µg/ml) treated control and WWP1-depleted AML cells. DNA damage foci have been detected following double immunostaining for γH2AX (purple) and one of the following repair proteins: 53BP1 (green, panel A), pDNA-PKcs (green, panel B), and BRCA1 (green, panel C). (Scale bar, 10 µm.) (D–G) Quantification of the number of γH2AX, 53BP1, pDNA-PKcs, and BRCA1 positive foci/cell. 53BP1, pDNA-PKcs, and BRCA1 foci evaluation has been performed in γH2AX positive cells (around 300 nuclei per condition); data are shown as median with quartiles and analyzed using the two-tailed unpaired t test. ****P < 0.0001. (H–K) Representation of cell population as percentage of cells with the indicated number of DNA damage foci (X-axis); data on shWWP1 condition (purple line) are shown as deviation from the Gaussian distribution set for the control sample (green line); P value has been calculated using the chi-square (Χ²) test with 2 degrees of freedom. Immunofluorescence staining has been repeated three times independently.

Fig. 6.

DNA damage resolution is delayed in WWP1-depleted cells. (A and B) DNA damage and repair activity were assessed by a comet-based assay performed at different time points following BLM washout. Representative fluorescence images. (Scale bar, 50 µm.) (A) and quantification (B) in percentage of tail DNA. The parameter % tail DNA was measured with ImageJ from 100 randomly selected cells for each condition. Three independent experiments were carried out. Values represent the mean ± SD from three or six independent experiments. Statistical significance was calculated using the one-way ANOVA test. **P < 0.01, ***P < 0.001, ****P < 0.0001. (C) Schematic representation of the EJ-RFP and DR-GFP reporter assays. The U2OS EJ-DRs cell line (38) contains integrated copies of both assays. NHEJ activity can be detected as an increase in RFP⁺ cells, while HR activation is measured as in increase in GFP⁺ cells. (D and E) The percentages of RFP⁺ and GFP⁺ cells reflect the levels of NHEJ and HR activation, respectively. Values represent the mean ± SD from three independent experiments. Statistically significant differences were calculated by the two-way ANOVA t test. *P < 0.05, ***P < 0.001.

Fig. 7.

WWP1 silencing sensitizes AML cells to DNA-damaging chemotherapeutic drugs. Control and WWP1-depleted cells were exposed to AraC (A), doxorubicin (B), and etoposide (C), and cell viability was assessed by the CellTiter-Glo^® assay. Values represent the mean ± SD from three independent experiments. Statistically significant differences were calculated by Student’s t test. **P < 0.01, ***P < 0.001. (D–F) Evaluation of apoptosis in control and WWP1-depleted NB4 cells exposed to chemotherapy. (D) Representative percentages of apoptotic cells, assessed by flow cytometry after staining with FITC annexin-V conjugates and DAPI. Three independent experiments were carried out. (E) Representative western blot analysis of cleaved PARP1 and caspase-3 proteins. (F) Densitometric quantification of cleaved (cl)/full-length (fl) ratio for PARP (Left panel) and caspase 3 (Right panel), respectively. Three independent experiments were carried out.

Fig. 8.

A model depicting the outcome of WWP1 silencing in AML cells. WWP1 inactivation reduces JARID1B ubiquitination and, in turn, decreases its cellular availability, enhancing the global levels of H3K4 trimethylation in chromatin. This change in chromatin composition would then hinder the recruitment of DNA repair factors and DNA repair efficiency, leading to accumulation of DNA damage and increased toxicity elicited by genotoxic drugs.