RETRACTED: Genome-wide identification of RETINOBLASTOMA RELATED 1 binding sites in Arabidopsis reveals novel DNA damage regulators

Abstract

Retinoblastoma (pRb) is a multifunctional regulator, which was likely present in the last common ancestor of all eukaryotes. The Arabidopsis pRb homolog RETINOBLASTOMA RELATED 1 (RBR1), similar to its animal counterparts, controls not only cell proliferation but is also implicated in developmental decisions, stress responses and maintenance of genome integrity. Although most functions of pRb-type proteins involve chromatin association, a genome-wide understanding of RBR1 binding sites in Arabidopsis is still missing. Here, we present a plant chromatin immunoprecipitation protocol optimized for genome-wide studies of indirectly DNA-bound proteins like RBR1. Our analysis revealed binding of Arabidopsis RBR1 to approximately 1000 genes and roughly 500 transposable elements, preferentially MITES. The RBR1-decorated genes broadly overlap with previously identified targets of two major transcription factors controlling the cell cycle, i.e. E2F and MYB3R3 and represent a robust inventory of RBR1-targets in dividing cells. Consistently, enriched motifs in the RBR1-marked domains include sequences related to the E2F consensus site and the MSA-core element bound by MYB3R transcription factors. Following up a key role of RBR1 in DNA damage response, we performed a meta-analysis combining the information about the RBR1-binding sites with genome-wide expression studies under DNA stress. As a result, we present the identification and mutant characterization of three novel genes required for growth upon genotoxic stress.

Fig 1. Characteristics of RBR1-marked domains.

(A) Position of domains (= genomic regions) marked by RBR1 with respect to genes and transposable elements (TEs). Replicate number is given in brackets. Full overlap = domain fully overlaps gene/TE; included = domain is included in gene/TE; 5’(-150) = domain overlaps 5’ of gene/TE or overlaps region 150 bp before the gene/TE; 3’(-150) = domain overlaps 3’ of gene/TE or overlaps region 150 bp after the gene; no overlap = domain does not belong to any other class. (B) The overlap of both RBR1-ChIP replicates is highly significant and defines a core set of 937 genes and 475 TEs bound by RBR1 (Genes: P(X> = 937) = 0*; TEs: P(X> = 475) = 0*). P values marked by an asterisk (*) were below the calculation limits of the software (highly significant). (C) Meta-analysis showing the distribution of RBR1-peaks with respect to genes. Each gene was divided in 50 bins (grey background), the 1 kb up- and downstream regions are shown, divided in 100bp per bin (white background). (D) Meta-analysis showing the distribution of RBR1-peaks with respect to TEs. Each TE was divided in 50 bins (grey background), the 1 kb up- and downstream regions are shown, divided in 100bp per bin (white background).

Fig 2. RBR1 binding to TEs.

(A) Pie diagram displaying TE distribution among the four main transposon classes in the entire Arabidopsis genome (left) and the RBR1-ChIP data (right). DNA transposons are highly overrepresented among TEs bound by RBR1 while retrotransposons are underrepresented (DNA transposon: P(X> = 250) = 2.17e-19, Helitron: P(X> = 205) = 0.264, LTR-retrotransposon: P(X< = 9) = 4.28e-32, non-LTR-retrotransposon: P(X< = 11) = 2.68E-05). (B) qRT-PCR analyses showing relative expression of AT1G60020, AT1G65985 and VANB (AT2G2349) in the inflorescences of rbr1-2 mutants compared to the wildtype. RAD51 is included as a positive control. Significant differences to the wildtype are marked by an asterisk (p<0.05). (C-E) Genome browser views showing RBR1-ChIP signals from sample 1 (RBR1-s1) and sample 2 (RBR-s2) associated with different TEs. (C) A Simpleguy1 transposon inserted into the second intron of AT1G65985. (D) A Simplehat1 transposon inserted directly 5’ of AT1G60020, an ORF belonging to a neighboring transposon (ATCOPIA5). (E) VANDAL21 transposon Hiun containing three ORFs (VANA, VANB, VANC). Note the RBR1 peak upstream of VANB. (F) DNA motifs detected by a MEME-ChIP analysis in the TE-associated RBR1-domains (E-value ≤ 0.01). Motifs were discovered by MEME and DREME and clustered by similarity. Only the most significant motif per cluster is shown.

Fig 3. Overlap of RBR1-ChIP with E2F-ChIP/ChAP/TChAP and RBR-RNAi data.

(A) Highly significant overlap of RBR1-bound genes with genes de-regulated in RBR1-RNAi lines as published by Horvath et al. ([35] up roots, P(X> = 38) = 1.171329E-33; down roots, P(X> = 3) = 1.97E-02) and Borghi et al. ([53] up young leaves, P(X> = 122) = 6.91E-39; down young leaves, P(X> = 26) = 8.07E-01). Transcriptional de-regulation was monitored using the Arabidopsis Ath1-Chip. For comparison, RBR1-ChIP data were therefore reduced to genes present on the Ath1-Chip. (B) Comparison of the RBR1-dataset with a time course of genes upregulated after RBR1-RNAi induction in young leaves [53] indicates that the majority of RBR1 responsive genes shows upregulation later than 6 hours after induction (hai) of silencing. (C) Overlap of RBR1-bound genes with the top 200 E2Fa-bound genes identified by ChIP, Chromatin Affinity Purification (ChAP) and Tandem Chromatin Affinity Purification (TChAP) published by Verkest at al. ([56] P(X> = 185) = 3.03E-223) and with genes showing an E2F-binding consensus site (WTTSSCSS) within 400 bp up-stream of the START-codon or beginning of the gene for non-protein coding genes (P(X> = 375) = 1.33E-106).

Fig 4. Characterization of MYB3R3- and RBR1-bound genes.

MYB3R3- and RBR1-bound genes in comparison with E2Fa-bound genes ((A) Verkest et al. [56]), core replication genes ((B) Shultz et al. [136]) and S-phase or M-phase associated genes ((C) Menges et al. [137]). Transcriptional upregulation in (C) was monitored using the Arabidopsis Ath1-Chip. For comparison, RBR1-ChIP data in (C) was therefore reduced to genes present on the Ath1-Chip. Overlap MYB3R3–RBR1: P(X> = 207) = 1.90E-217. Overlap MYB3R3–E2Fa: P(X> = 106) = 2.82E-134. Overlap MYB3R3–core replication: P(X> = 20) = 6.30E-24. Overlap RBR1–core replication: P(X> = 44) = 4.99E-55. Overlap MYB3R3–S-phase associated genes: P(X> = 30) = 1.03E-14. Overlap RBR1–S-phase associated genes: P(X> = 58) = 7.88E-24. Overlap MYB3R3–M-phase associated genes: P(X> = 66) = 3.78E-52. Overlap RBR1-M-phase associated genes: P(X> = 48) = 1.38E-15.

Fig 5. Overrepresented motifs in gene-associated RBR1-bound domains.

The analysis was done using MEME-ChIP (E-value ≤ 0.01). Motifs were discovered by MEME and DREME and clustered by similarity. Only the most significant motif per cluster is shown.

Fig 6. Meta-analysis of gene expression data from DNA stress experiments.

(A) VENN diagram displaying the number of genes upregulated at least once in 32 DNA-stress experiments (S5 Table) in comparison with RBR1-bound genes (S1 Table) and genes known to be involved in DNA repair (S4 Table). Highly significant enrichment (p < 0.0001, Fischer’s Exact test) of genes involved in DNA repair is seen when genes that are upregulated under DNA stress and bound by RBR1 (11.1% DNA repair genes) are compared with those that are upregulated but do not show RBR1 binding (0.9% DNA repair genes). (B) List of genes that are upregulated upon DNA stress in more than three out of 32 experiments and that are RBR1 targets according to the RBR1-ChIP experiment. The Y-axis displays the number of experiments in which a gene was found to be significantly upregulated under DNA-stress. For more detailed information, see S5 Table. Genes that are also present in the DNA-repair dataset (S4 Table) are labeled in red and the candidate genes for further investigation in black.

Fig 7. Root growth analysis on DNA damaging drugs.

Root growth of the wildtype and kno1 (at3g20490), flip (at1g04650), ubp21 (at5g46740) and at2G45460, represented by 2 different mutant alleles each, on 20 μM cisplatin (CiPt) and 0.3 μg/ml bleomycin (BLM). (A) Pictures after 6 days of growth. (B) Root length per day. Statistically significant differences to the wildtype are indicated by asterisks (p < 0.05).

Fig 8. DNA damage analysis in the wildtype versus kno1-1, flip-2 and ubp21-1 mutants.

(A) Root nuclei incubated with an anti-γH2AX antibody after 3 hr treatment of seedlings with cisplatin and bleomycin in comparison with mock treated samples. (B) Quantification of γH2AX foci, 50 nuclei were counted per line.

Fig 9. KNO1 is needed for RAD51 localization.

Expression of (A) CYCB1;1 and (B) RAD51 in the wildtype as well as in kno1-1, flip-2, and ubp21-1 mutant seedlings as detected by qRT-PCR after 3 hr treatment with and without 50μM cisplatin. Statistically significant differences in expression compared to the corresponding wildtype sample are marked by an asterisk (p<0.05). (C) Root cell nuclei of the wildtype and kno1-1, flip-2 and ubp21-1 mutants incubated with an anti-γH2AX antibody and an anti-RAD51 antibody after 3 hr treatment of seedlings with cisplatin and bleomycin in comparison with mock treated samples.