A truncated and catalytically inactive isoform of KDM5B histone demethylase accumulates in breast cancer cells and regulates H3K4 tri-methylation and gene expression

Abstract

KDM5B histone demethylase is overexpressed in many cancers and plays an ambivalent role in oncogenesis, depending on the specific context. This ambivalence could be explained by the expression of KDM5B protein isoforms with diverse functional roles, which could be present at different levels in various cancer cell lines. We show here that one of these isoforms, namely KDM5B-NTT, accumulates in breast cancer cell lines due to remarkable protein stability relative to the canonical PLU-1 isoform, which shows a much faster turnover. This isoform is the truncated and catalytically inactive product of an mRNA with a transcription start site downstream of the PLU-1 isoform, and the consequent usage of an alternative ATG for translation initiation. It also differs from the PLU-1 transcript in the inclusion of an additional exon (exon-6), previously attributed to other putative isoforms. Overexpression of this isoform in MCF7 cells leads to an increase in bulk H3K4 methylation and induces derepression of a gene cluster, including the tumor suppressor Cav1 and several genes involved in the interferon-alpha and -gamma response. We discuss the relevance of this finding considering the hypothesis that KDM5B may possess regulatory roles independent of its catalytic activity.

Fig. 1. KDM5B transcripts and protein isoforms.

A KDM5B transcripts as reported in Ref Seq NCBI database. RBP2-H1, PLU-1, KDM5B-240, and KDM5B-215 are the validated (NM) variants, X1 and X2 are the predicted (XM) ones. Transcripts differ at the 5′ UTRs and for the presence of alternative exons. RBP2-H1 represents the longest transcript and includes the exon-6 (in red). The arrows indicate the ATG sites, the first one in exon-1 and the second one in exon-4. B Protein domains of human KDM5B are reported above validated and predicted KDM5B protein isoforms; PLU-1 is the canonical isoform. Ab1 and Ab2 are KDM5B C-terminus and exon-6 specific antibodies, respectively. C WB analysis of KDM5B isoform expression in MCF7 lysates: on the left two bands corresponding to 175.7 and 161.5 kDa proteins are detected by Ab1 in the first blot, while on the right one single band for 161.5 kDa protein is detected by Ab2, after the stripping of the first blotting. D Preliminary investigation of KDM5B isoforms’ expression in different human cell lines and human testis, using the two antibodies Ab1 (panel up) and Ab2 (panel down), before and after the stripping, respectively. Panels A and B created with BioRender.com.

Fig. 2. 5′RLM-RACE analysis of KDM5B transcripts.

A The scheme reports the position of reverse primers used for the two RT-PCRs of 5′RACE experiment. Binding nucleotides are indicated underneath RBP2H1 sequence. See Material and Methods paragraph for experimental details. B PCR products from PCR1 and PCR2. Product lengths are as expected. C The position of first and second ATG codons are shown with green arrows. We found a new 5′-end 221 nucleotides downstream the canonical one. D Partial alignment of PCR1 and PCR2 sequences refer to KMD5B-PLU1 transcript. (For the complete alignment see Supplementary File S1). Panel A created with BioRender.com.

Fig. 3. Expression of KDM5B isoforms in different subtypes of breast cancer cell lines.

A WB analysis of KDM5B isoforms’ expression in lysates of different breast cancer cell lines shows that NTT isoform is detected in all of them; as confirmed also by the blotting using the Ab2, the exon-6-encoded residues are almost exclusively included in the truncated isoform. Cell lines are grouped according to the molecular subtypes. B No correlation between breast cancer subtypes and relative NTT expression (calculated as NTT/PLU-1 ratio) is highlighted upon the quantification.

Fig. 4. Protein stability of KDM5B isoforms.

A After CHX treatment, the KDM5B-NTT results more stable than KDM5B-PLU-1 in both MCF7 and MDA-MB-231, as shown from the blots using the anti-C-terminal KDM5B (upper panel) and the anti-exon-6 KDM5B (lower panel) antibodies. B Degradation kinetics of KDM5B isoforms in a time course in MCF7 cells (n = 3 for each time point). C Degradation kinetics of KDM5B isoforms in a time course in MDA-MB-231 cells (n = 3 for each time point). D Comparing the degradation curves of PLU-1 isoform in the two analyzed breast cancer cell lines, it was found a different rate of degradation, that is significantly faster in MDA-MB-231, where the mean half-life-time of the PLU-1 is of about 2 h, versus MCF7, where instead it is around 4 h (n = 3 for each time point). E The cycloheximide (CHX) treatment highlights the PLU-1 instability, which is rescued by MG132 treatment, demonstrating that the proteasomal activity regulates the turnover of PLU-1 isoform in MCF7 and MDA-MB-231. ns: p > 0.05, *p < = 0.05, **p < = 0.01, ***p < = 0.001; Statistics by unpaired Welch’s t-test.

Fig. 5. MS-based analysis of H3K4 methylation levels in MCF7 and MDA-MB-231 cells overexpressing the KDM5B-NTT isoform.

A The KDM5B-NTT expression in the overexpression condition in MCF7 cells at 24 h from transfection increases compared to cells transfected with the Empty vector (E), where the KDM5B-NTT expression is similar to that observed in the non-transfected cells (NT). B The KDM5B-NTT expression increases also in MDA-MB-231 cells at 24 h from transfection in the overexpression condition compared to negative controls. C Heatmap display of the log2 L/H ratios (where L: samples and H: internal standard) obtained for H3K4 methylations in MCF7 and MDA-MB-231 cells not transfected (NT), transfected with empty vector (E), or overexpressing KDM5B-NTT. The data were normalized over the average value across all the samples. D H3K4me2 and H3K4me3 levels (expressed as L/H ratios) in MCF7 and MDA-MB-231 cells overexpressing KDM5B-NTT, compared to control E. Statistical analysis was performed by repeated measures ANOVA, followed by Tukey’s multiple comparison test. **p < 0.005, ***p < 0.001.

Fig. 6. Effects of KDM5B-NTT overexpression on transcriptome.

A, B Volcano plots showing the differentially expressed genes (DEGs) in MCF7 and MDA-MB-231 cells with overexpression of KDM5B-NTT, respectively. Red dots indicate DEGs associated with p < 0.1 and log₂(FC) < −1 or log₂(FC) > 1; blue dots indicate genes with p < 0.1 and −1 < log₂(FC) < 1; green dots indicate genes with log₂(FC) < −1 or log₂(FC) > 1, and p > 0.1. C Heatmap showing in color scale the GSEA Normalized Enriched Score (NES) of gene expression profile resulting from overexpression of KDM5B-NTT in MCF7 and MDA-MB-231 cells in comparison with control cell samples. (White * = FDR < 0.01). D Cluster of KDM5B-NTT-deregulated genes, which are also significantly induced by one or both epigenetic drug (PDCA and MTA) and/or previously shown to be bound by KDM5B in ChIP experiments [19, 22, 25].

Fig. 7. Diagram illustrating the proposed mechanism for the regulatory action of KDM5B-NTT.

Transcription from Promoter 2 generates NTT mRNAs which include the exon-6, and which lead to the expression of the KDM5B-NTT protein isoform. A second ATG (ATG2) is used for translation of NTT isoform, resulting in a KDM5B N-Terminal Truncated protein isoform including 36 more residues encoded by the exon-6. Even though the NTT mRNAs are around 7% of the total KDM5B transcriptional events, the NTT protein isoform is very much more stable compared to PLU-1, targeted by the proteasome degradation system. NTT chromatin association is driven by the PHD3 domain, which preferentially recognizes the tri-methylated H3K4. This association inhibits the PLU-1 recruitment on common targets and prevents their demethylation, leading to an increase of H3K4me3 levels and resulting in gene derepression. Created with BioRender.com.