Trichothiodystrophy-causative pathogenic variants impair a cooperative action of TFIIH and DDX1 in R-loop processing

Abstract

The transcription factor IIH (TFIIH) is a key player in transcription and DNA repair by nucleotide excision repair. It is made of 10 subunits organized in core-TFIIH and CAK sub-complexes bridged by XPD. Pathogenic variants in the ERCC2/XPD gene give rise to xeroderma pigmentosum (XP) or trichothiodystrophy (TTD), two distinct clinical entities with opposite skin cancer proneness. Here, we show that TTD variants cause a partial dissociation of the CAK from the chromatin and from the core-TFIIH. Mass spectrometry analysis reveals that the chromatin-bound CAK, as a component of the entire TFIIH, participates in a protein assembly containing the RNA-binding proteins DDX1, SFPQ, NONO as well as RNA polymerase II (Pol II). Gene silencing experiments demonstrate that the protein assembly is required to process the DNA:RNA hybrids formed during Pol II extension and to protect the cell from transcriptional stress. TTD-specific variants in ERCC2/XPD result in TFIIH instability, altered interaction of the CAK with DDX1-SFPQ-NONO, and R-loop accumulation. Therefore, the limited amount of TFIIH that distinguishes TTD from XP gives rise to transcriptional stress and extensive gene expression deregulations, thus accounting for the wide spectrum of TTD clinical features.

Graphical Abstract

Figure 1.

The DEAD-box DDX1 is a novel interactor of CDK7. (A) Immunoblot analysis with antibodies raised against different TFIIH subunits (XPB and p62 of the core-TFIIH, the bridging factor XPD and the CDK7 and CycH subunits of the CAK), the chromatin (H3), and cytoplasm (Mek2) markers in whole extracts (WCE), 1% triton-soluble (Sol), and chromatin-enriched (Chr) fractions of primary dermal fibroblasts from healthy donors (C3PV and C8PV), PS-TTD (TTD8PV and TTD23PV), or XP (XP26VI and XP15PV) patients with pathogenic variants in ERCC2/XPD gene. Protein quantifications were obtained by normalizing each protein band on the corresponding MEK2 or H3 protein amounts for the soluble and chromatin-enriched fractions, respectively, in order to even out the loading differences. For each TFIIH subunit, the soluble (white bar) and chromatin-bound (black bar) amounts are expressed as a percentage of their sum, indicated as 100%. The graph (bottom panel) includes the data obtained by the immunoblots shown in Supplementary Fig. S1A. The percentage of chromatin-associated proteins in PS-TTD or XP primary dermal fibroblasts is compared with that observed in CTR cells. Bars indicate standard errors (**P < .01, ***P < .001; Student’s t-test). (B) Silver staining of proteins co-immunoprecipitated with CDK7 antibodies in the chromatin-enriched (Chr) fraction of MRC5 cells. (C) Table list of the novel chromatin-associated CDK7-interacting proteins in MRC5 cells identified through mass spectrometry analysis. Immunoblot analysis of DDX1 protein, the CAK (CDK7 and CycH) and XPD subunits of TFIIH in 1% triton-soluble (sol) and chromatin-enriched fractions (chr) of MRC5 cells before (Input) and after immunoprecipitation (IP) with anti-CDK7 (D), anti-DDX1 (E), or IgG (D and E) antibodies.

Figure 2.

DDX1 helicase binds to holo-TFIIH and Pol II. (A) In vitro pull-down assay of XPD or the CAK subunits CDK7 and CycH with DDX1 recombinant protein under permissive (100 mM KCl) or restrictive (300 mM KCl) salt conditions. Proteins are visualized by immunoblot analysis with antibodies raised against DDX1, XPD, CDK7, and CycH. The membranes were first hybridized with anti-XPD (*) and subsequently with anti-DDX1. Due to the similar molecular weight of the two proteins, the XPD protein band is still visible in the DDX1 immunoblotting. The input represents 10% of the total protein amount used in each pull-down reaction. In vitro pull-down assays of XPD WT, XPD fragments (B), or mutated forms of XPD (C) with DDX1 recombinant protein. XPD fragments contain specific functional domains of the protein, as depicted on the XP schematic representation (top). Mutated forms include the Arg112His and Arg722Trp substitutions causative of PS-TTD and the Arg683Trp amino acid change causative of XP. Proteins are visualized by immunoblot analysis with antibodies specific for XPD and DDX1. The input represents 10% of the total protein amount used in each pull-down reaction. (D) Immunoblot analysis with antibodies raised against the DDX1 helicase, various TFIIH subunits (XPB and p62 of core-TFIIH, the bridging factor XPD, CDK7, and CycH of the CAK sub-complex), and the RPB1 subunit of Pol II of two-step (TIP) immunoprecipitations performed first with anti-CDK7 and subsequently anti-DDX1 antibodies in 1% triton-soluble (sol) and chromatin-enriched (chr) fractions of control MRC5 cells. As a negative control, the first immunoprecipitation step was performed with IgG antibodies.

Figure 3.

Reduced DDX1 protein amount leads to impaired Pol II-mediated transcription. (A) NER efficiency by UDS analysis in MRC5 cells upon gene silencing with either scrambled (CTR) or DDX1 siRNA. The number of individual grains per nuclei was counted. The diagram reports the mean values of three independent experiments. (B) Repair efficiency by slot blot analysis of CPD or 6–4PP in untreated MRC5 cells or treated with either DDX1 or control siRNA (CTR) at 0, 1.5, 4.5, 6.5 h after 20 J/m² UV-C irradiation. The amount of CPD or 6–4PP has been normalized to the amount of loaded single-strand DNA in at least three independent experiments. No significant differences have been recorded by the Student’s t-test when comparing the mean values of DDX1 silenced cells with the corresponding controls (MRC5 or CTR siRNA). (C) Global RNA synthesis by in vivo MRC5 cells labelled with 5-ethynyluridine (EU) and EU-click reaction (red). As a negative control, cells were exposed to Actinomycin D (ACTD) before EU treatment. Nuclei were counterstained with DAPI. The diagram on the left reports the intensity of the EU nuclear fluorescent signal measured by ImageJ in cells from three independent experiments (shown by different colours). The mean value and standard error for each experiment are indicated. (D) Immunoblot analysis of DDX1, TFIIH subunits (XPB, p62, XPD, CDK7, and CycH), the α subunit of the basal transcription factor TFIIE and the RPB1 subunit of Pol II in MRC5 whole cell extract transfected with either scrambled (CTR) or DDX1 siRNA (left). The amount of DDX1 and Pol II protein levels was normalized to the amount of β-actin. The diagram reports the mean values of three independent experiments (right). (E) Immunoblot analysis with antibodies raised against DDX1, the CDK7 and CycH subunits of TFIIH, the RPB1 subunit of Pol II, and the additional novel CDK7-interactors (NONO, SFPQ, and DHX9) in TIP samples performed first with anti-CDK7 and subsequently anti-DDX1 antibodies in the chromatin-enriched fractions of control MRC5 cells. As a negative control, the first immunoprecipitation step was performed with IgG antibodies (IgG). In all the graphs of the figure, when depicted, bars indicate standard errors (**P < .01, ****P < .0001; Student’s t-test).

Figure 4.

Reduced DDX1 protein level leads to increased R-loop amount. (A) Slot blot of 0.5 and 1 μg of genomic DNA from MRC5 cells transfected with scrambled control (CTR) or DDX1 siRNA in the absence (−) or presence (+) of RNase H1 and hybridized with antibodies recognizing the RNA/DNA hybrids (S9.6 antibody) or dsDNA (left). The intensity of the bands was measured with ImageJ. The amount of S9.6 signal was normalized to the amount of the loading control dsDNA (right). The values are the mean of at least three independent experiments (*P < .05; Student’s t-test). (B) DRIP analysis with the S9.6 antibody at the β-actin locus (ACTB) of MRC5 cells treated with scrambled (CTR) or DDX1 siRNA for 120 h. The amount of RNA/DNA hybrids at a, b, c, d, and e positions, indicated as horizontal bars within the β-actin locus (schematic representation on the top), was evaluated by real-time PCR. When applied, the RNase H1 treatment is indicated. Data are expressed as fold enrichment over the input. The values are the mean of at least three independent experiments (*P < .05, **P < .01; Student’s t-test). (C) Density plot of DDX1 ChIP-seq peak distribution relative to the gene TSS in Mus musculus. The plot shows the distribution of DDX1 occupancy across 8504 genes, each showing at least one DDX1 ChIP-seq peak within −1000 to +1000 bp of TSS. Peaks were mapped separately to both the plus strand and the minus strand. The x-axis indicates the upstream and downstream distance in bp from the TSS corresponding to position 0 bp. The y-axis indicates the density of DDX1 peaks.

Figure 5.

DDX1-dependent R-loop accumulation leads to transcriptional stress. (A) Global RNA synthesis by in vivo labelling with 5-ethynyluridine (EU) and EU-click reaction in MRC5 cells 48 h after transfection with DDX1 or scrambled (CTR) siRNA as well as with the plasmid expressing the RNase H1^GFP or the empty vector. Nuclei were counterstained with DAPI. The diagram (top right) reports the intensity of the EU nuclear fluorescent signal measured by ImageJ in cells from two independent experiments (shown by different colours). The mean value and standard errors of each experiment are indicated (***P < 0.001; Student’s t-test). (B) Immunoblot analysis with antibodies raised against DDX1, XPD, γH2AX, and ORC2 in the chromatin-enriched fraction of cells transfected with either scrambled (CTR) or DDX1 siRNA (left). The amount of DDX1, XPD, and γH2AX protein levels was first normalized to the amount of the chromatin loading control ORC2 and then expressed as fold increased relative to the corresponding protein amount in CTR siRNA. The diagram reports the mean values of three independent experiments (right). Bars indicate the standard errors (**P < .01, ***P < .001; Student’s t-test). (C) Immunoblot analysis with antibodies raised against DDX1, γH2AX, and γTub in MRC5 cells transfected with DDX1 or scrambled (CTR) siRNA as well as with the plasmid expressing the RNase H1^GFP (+) or the empty vector (−) (left). The amount of DDX1 and γH2AX protein levels was normalized to the amount of the corresponding γTub loading control and reported as arbitrary units (au). The diagram reports the mean values of three independent experiments (right). Bars indicate the standard errors (*P < .05, ***P < .001; Student’s t-test).

Figure 6.

TFIIH plays a role in R-loop processing. (A) Immunoblot analysis with antibodies raised against the RPB1 subunit of Pol II and various TFIIH subunits (XPB and p62 of the core-TFIIH, XPD, CDK7, and CycH of the CAK) in MRC5 whole cell extracts transfected with scrambled control (CTR) or XPD siRNA. Protein levels were normalized to the amount of γ-tubulin and reported as arbitrary units (au). The diagram reports the mean values of three independent experiments (right). Bars indicate the standard errors (*P < .05, **P < .01, ***P < .001; Student’s t-test). (B) Slot blot of 0.5 and 1 μg of genomic DNA from MRC5 cells transfected with scrambled control (CTR) or XPD siRNAs in the absence (−) or presence (+) of RNase H1 and hybridized with antibodies recognizing the RNA/DNA hybrids (S9.6 antibody) or dsDNA (left panel). The intensity of the bands was measured with ImageJ. The amount of S9.6 signal was normalized to the amount of the loading control, the dsDNA signal (right panel). The values are the mean of at least three independent experiments. Bars indicate the standard error (*P < .05; Student’s t-test). (C) DRIP analysis with the S9.6 antibody at the β-actin locus (ACTB) of MRC5 cells treated with scrambled control (CTR) or XPD siRNA for 72 h. The amount of RNA/DNA hybrids at the a, b, c, d, and e positions of the β-actin locus was evaluated by real-time PCR. When applied, the RNase H1 treatment is indicated. Data are expressed as fold enrichment over the input. The values are the mean of at least three independent experiments. Bars indicate the standard errors (*P < .05, **P < .01; Student’s t-test). (D) Immunoblot analysis of XPD and the CDK7 subunits of TFIIH, NONO, SFPQ, and DDX1 proteins in the chromatin-enriched fractions of MRC5 cells transfected with scrambled control (CTR) or XPD siRNAs before (Input) and after immunoprecipitation (IP) with anti-CDK7 or IgG antibodies. Two independent IPs are shown (upper and lower left panels). In the cells treated with XPD-siRNA, the amount of co-immunoprecipitated proteins has been normalized to the amount of the corresponding immunoprecipitated CDK7 and expressed as fold increased relative to the sample CTR siRNA (right panel). The values are the mean of at least three independent experiments. When depicted, bars indicate standard errors (*P < .05; Student’s t-test).

Figure 7.

PS-TTD cells exhibit altered TFIIH-DDX1 interaction and impaired R-loop processing. (A) Immunoblot analysis with antibodies raised against various TFIIH subunits (XPB and p62 of the core-TFIIH, XPD, CDK7, and CycH of the CAK), DDX1, NONO, SFPQ, and the RPB1 subunit of Pol II in 1% triton-soluble (sol) and chromatin-enriched (chr) fractions of MRC5, TTD2GL, and XP102LO cells before (Input) and after immunoprecipitation with anti-CDK7 antibodies. The amount of co-immunoprecipitated DDX1, NONO, or SFPQ proteins was normalized to the amount of immunoprecipitated CDK7 (right panel). Bars indicate standard errors (*P < .05; **P < .01; ***P< .001; Student’s t-test). (B) Immunoblot analysis with antibodies raised against various TFIIH subunits (XPB and p62 of the core-TFIIH, XPD, CDK7, and CycH of the CAK), DDX1, NONO, SFPQ, and the RPB1 subunit of Pol II in 1% triton-soluble (sol) and chromatin-enriched (chr) fractions of control (C3PV), TTD23PV, and XP26VI primary dermal fibroblasts before (Input) and after immunoprecipitation (IP) with anti-CDK7 or IgG antibodies. The amount of co-immunoprecipitated DDX1, NONO, and SFPQ was normalized to the amount of immunoprecipitated CDK7. The graph (right panel) also includes data collected from the samples shown in Supplementary Fig. S8. Bars indicate standard errors (**P < .01; Student’s t-test). (C) DRIP analysis with the S9.6 antibody at the β-actin locus (ACTB) of primary dermal fibroblasts from PS-TTD with pathogenic variants in ERCC2/XPD (TTD8PV, TTD12PV, and TTD23PV) or in ERCC3/XPB (TTD6VI), from XP with pathogenic variants in ERCC2/XPD (XP15PV and XP49PV), or healthy individuals (CTR, C3PV, and C5PV). The amount of RNA/DNA hybrids at the a, b, c, d, and e positions of the ACTB locus was evaluated by real-time PCR. When applied, the RNase H1 treatment is indicated. Data are expressed as fold enrichment over the input. The values are the mean of at least two independent experiments. Bars indicate standard errors (*P < .05, **P< .01, ***P< .001; Student’s t-test).

Figure 8.

TFIIH-dependent R-loop accumulation contributes to PS-TTD transcription deregulation (A) Global RNA synthesis by in vivo labelling with 5-ethynyluridine (EU) and EU-click reaction (red staining) in C5PV and TTD7PV fibroblasts transfected for 48 h with the plasmid expressing RNase H1^GFP (green staining) or the empty vector. Nuclei were counterstained with DAPI (blue staining). The diagram (top right) reports the intensity of the EU nuclear fluorescent signal measured by ImageJ in cells from two independent experiments (shown in green and blue colours). The mean value and standard errors of each experiment are indicated (**P < .01; Student’s t-test). (B) Kernel density plot comparing the gene length distribution of the 610 TTD-specific down-regulated transcripts by RNA-seq analysis (log₂FC < −1 and False Discovery Rate-adjusted P < .05) with that of the reference list containing 37 990 transcripts that include the protein-coding genes, lncRNAs, pseudogenes, and TEC (polyA EST to be Experimentally Confirmed). The gene length is plotted in logarithmic scale (x-axis) versus the density of observations at each length (y-axis). P: Welch’s two-sample t-test.