Mechanism and Regulation of DNA-Protein Crosslink Repair by the DNA-Dependent Metalloprotease SPRTN

Abstract

Covalent DNA-protein crosslinks (DPCs) are toxic DNA lesions that interfere with essential chromatin transactions, such as replication and transcription. Little was known about DPC-specific repair mechanisms until the recent identification of a DPC-processing protease in yeast. The existence of a DPC protease in higher eukaryotes is inferred from data in Xenopus laevis egg extracts, but its identity remains elusive. Here we identify the metalloprotease SPRTN as the DPC protease acting in metazoans. Loss of SPRTN results in failure to repair DPCs and hypersensitivity to DPC-inducing agents. SPRTN accomplishes DPC processing through a unique DNA-induced protease activity, which is controlled by several sophisticated regulatory mechanisms. Cellular, biochemical, and structural studies define a DNA switch triggering its protease activity, a ubiquitin switch controlling SPRTN chromatin accessibility, and regulatory autocatalytic cleavage. Our data also provide a molecular explanation on how SPRTN deficiency causes the premature aging and cancer predisposition disorder Ruijs-Aalfs syndrome.

Keywords: DNA repair; DNA-protein crosslinks; DVC1; Ruijs-Aalfs syndrome; SPRTN; Spartan; Wss1; formaldehyde; hepatocellular carcinoma; progeria; protease; topoisomerase.

Graphical abstract

Figure 1

SPRTN/Dvc-1 Provides Resistance toward DPC-Inducing Agents in Worms and Operates Independently of FANCD2/Fcd-2 and Parallel to BLM/Him-6 (A) Domain structures and evolutionary distances of the protease domain of SPRTN/Wss1 protease family members of humans, worms, and budding yeast. SPRTN/Wss1 proteases bear interaction domains for p97/Cdc48 (SHP-box, VIM), recognition modules for ubiquitin (UBZ) or SUMO (SIM), and in metazoans a PCNA-interaction motif (PIP-box). (B) C. elegans mutant strains lacking functional SPRTN (dvc-1) are specifically sensitive to the DPC-inducing agents. Formaldehyde sensitivity was determined in synchronized L1 larvae. Cisplatin, UVC light, and IR sensitivities were assessed by measuring embryonic survival of progeny after exposure of adult animals. Error bars indicate SEM of –two to four independent experiments. (C) FANCD2 is not involved in providing formaldehyde resistance in synchronized L1 larvae. Error bars indicate SEM of two independent experiments. (D) FANCD2 provides resistance to chronic cisplatin exposure by a mechanism distinct to DPC repair by SPRTN. Viability was assessed by determination of embryonic survival of progeny of young adult animals kept on cisplatin-containing plates (200 μM) for the indicated amount of time. Error bars indicate SEM of two independent experiments. (E) Loss of XPA does not result in increased formaldehyde sensitivity in synchronized L1 worms. Error bars indicate SEM of two independent experiments. (F) Loss of SPRTN (dvc-1) results in viability defects in worms lacking the BLM helicase (him-6). Data were obtained from at least 16 animals per indicated genotype. Whiskers indicate tenth to 90th percentiles. Statistical significance was tested using an unpaired t test. (G) The BLM helicase (Him-6) is not involved in providing formaldehyde resistance in synchronized L1 larvae. Error bars indicate SEM of two independent experiments. (H) BLM (Him-6) provides resistance to cisplatin exposure by a mechanism parallel to DPC repair by SPRTN. Cisplatin sensitivity was assessed by measuring embryonic survival of progeny after exposure of adult animals. Error bars indicate SEM of two independent experiments. See also Figure S1.

Figure 2

SPRTN-Deficient Mammalian Cells Fail to Repair DPCs and Are Hypersensitive toward DPC-Inducing Agents (A) Schematic representation of the KCl/SDS precipitation assay used to measure DPC repair. Cells are lysed in denaturing conditions (1% SDS), followed by sonication and precipitation of cellular protein by the addition of KCl. Crosslinked DNA co-precipitates with the protein, whereas free DNA remains in the supernatant. The precipitate is washed several times before quantification of soluble and crosslinked DNA. (B) SPRTN-deficient MEFs fail to repair formaldehyde-induced DPCs. Sprtn^F/−, Sprtn^F/+ (untreated or treated with 4-hydroxy tamoxifen [4-OHT] for 48 hr), Fancd2^+/+, and Fancd2^−/− MEFs were treated with 200 μM formaldehyde (FA) for 1 hr to induce DPCs and lysed directly or allowed to repair. DPCs were measured as the ratio of crosslinked DNA compared to total DNA. Error bars indicate SEM of two independent experiments. (C and D) Knockdown of SPRTN results in formaldehyde sensitivity in human cells. Relative cell numbers were determined 6 days after U2OS cells transfected with SPRTN or control siRNA were treated with the indicated doses of formaldehyde, aphidicolin, or mytomicin C. Error bars represent SD of two to four replicates. See also Figure S2.

Figure 3

A DNA Switch Controls SPRTN’s Protease Activity (A) Schematic representation of recombinant GST-SPRTN-Strep variants (upper panel). Sequence of SPRTN’s active site with catalytic residues in red and tyrosine-to-cysteine replacement found in RJALS patients in blue (lower panel) are shown. (B) Autocatalytic cleavage of SPRTN is induced by DNA. SPRTN (180 nM, N-terminally GST tagged, C-terminally Strep tagged) was incubated in the absence or presence of circular ssDNA (ΦX174 virion, 10 nM). (C) Autocatalytic cleavage of SPRTN is induced by various types of DNA. GST-SPRTN-Strep (WT, E112Q, or the disease variants Y117C and ΔC, 180 nM) was incubated in the presence of different types of DNA (phage DNA [10 nM], 30-mer oligonucleotides [1.8 μM]) for 2 hr at 25°C. (D and E) SPRTN cleaves DNA-binding proteins in an ssDNA-dependent manner. GST-SPRTN-Strep (WT or the catalytically inactive E112Q variant, 480 nM) was incubated with the indicated substrates (360 nM) in the absence or presence of ss and ds phage DNA (10 nM) for 2 hr at 25°C. (F and G) SPRTN and histone H1 bind similarly to ss and ds phage DNA. Proteins (SPRTN [0.45, 0.9, and 1.8 μM] and H1 [2, 3, and 4 μM]) were incubated with DNA (50 nM) and analyzed on 0.8% agarose gels. (H) Wss1 cleaves histone H1 in an ssDNA-dependent manner. Wss1 (WT or the catalytically inactive E116Q variant, 800 nM) histone H1 (200 nM) were incubated with the indicated type of DNA (10 nM) for 2 hr at 30°C. (I) SPRTN disease variants display defects in ssDNA-dependent substrate cleavage. GST-SPRTN-Strep (WT, Y117C, ΔC or auto, 480 nM) was incubated with histone H1 (360 nM) in the absence or presence of ss phage DNA (10 nM) for 2 hr at 25°C. (J) C-terminally truncated SPRTN variants retain the ability to bind DNA. Indicated proteins (500 nM and 1 μM) were incubated with a fluorescently labeled ss oligonucleotide (250 nM) prior to gel electrophoresis in 6% PAGE gels. (K) SPRTN’s DNA-binding domain resides within aa 200–250. Indicated proteins (0.25, 0.5, and 1 μM) were incubated with a fluorescently labeled ss oligonucleotide (250 nM) prior to gel electrophoresis in 6% PAGE gels. (L) SPRTN deficient for DNA binding is deficient for ssDNA-dependent substrate cleavage. Recombinant GST-SPRTN-Strep (WT or 1–199, 480 nM) was incubated with recombinant histone H1 (360 nM) in the absence or presence of ss phage DNA (10 nM) for 2 hr at 25°C See also Figure S2.

Figure 4

Crystal Structure of the Protease Domain of SPRTN’s Fission Yeast Homolog (A) Structure of the protease domain of S. pombe Wss1b (PDB: 5JIG) in a surface (left) or cartoon (right) representation. Active site residues are displayed as sticks. Position 117 mutated in RJALS to a cysteine is highlighted in green. Numbering of residues corresponds to the human sequence. (B) Close-up view of the active site showing the octahedral coordination of Ni²⁺ by His111, His115, His130, as well as one oxygen and two water molecules. The 2F_O-F_C electron density map is contoured to 1σ, whereas the anomalous density (magenta) for Ni²⁺ is contoured to 10σ. Most likely the catalytic zinc atom has been replaced during the Ni²⁺-affinity chromatography step. See also Figure S3.

Figure 5

DNA Binding Induces a Conformational Change within SPRTN (A) SPRTN undergoes a conformational change upon DNA binding. Catalytically inactive GST-SPRTN-Strep E112Q was subjected to limited proteolytic digestion by trypsin in the presence or absence of ssDNA or dsDNA. (B) Quantification of specific proteolytic fragments observed in (A). Values represent mean ± SEM of three independent experiments. (C) SAXS analysis indicates that ssDNA binding increases the flexibility of SPRTN. Electron pair distribution shows an increase in Rg and Dmax upon ssDNA (15-mer) binding. (D) Heatmap showing H/D exchange mass spectrometry indicating differences in deuterium incorporation between SPRTN and SPRTN + ssDNA. Regions of increased protection are shown in blue and increased exposure in red. Deuterium labeling was carried out at three time points (0.3, 3, and 30 s) in triplicates. See also Figure S4.

Figure 6

Chromatin Access of SPRTN Is Controlled by a DPC-Triggered Ubiquitin Switch (A) Mono-ubiquitinated SPRTN is excluded from chromatin. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells were either lysed directly in SDS-containing loading dye (total) or subjected to fractionation in soluble and chromatin components. (B) Formaldehyde treatment induces deubiquitination of SPRTN coinciding with a complete relocalization to chromatin. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated with 1 mM formaldehyde (FA) for 2 hr. (C) SPRTN is deubiquitinated upon formaldehyde exposure in a dose-dependent manner. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated for 2 hr with the indicated dose of formaldehyde. (D) Endogenous SPRTN is deubiquitinated and relocalizes to chromatin formaldehyde exposure. U2OS cells were treated with 1 mM formaldehyde (FA) for 2 hr. Asterisk indicates an unspecific band. (E) Deubiquitination of SPRTN is specifically triggered by DNA-protein crosslinks. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated with formaldehyde (FA, 1 mM, 2 hr), UVC light (UV, 20 J/m², 2 hr before lysis), aphidicolin (Aph, 1 μM, 2 hr), or IR (3 Gy, 2 hr before lysis). (F) SPRTN is differentially recruited to chromatin depending on the type of DNA damage. Doxycycline-induced YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated with formaldehyde (FA, 0.5 mM) or UV (20 J/m²) and subjected to pre-extraction, prior to fixation and immunofluorescence. (G) SPRTN deubiquitination and chromatin recruitment upon DPC induction is independent of binding to PCNA, p97, or ubiquitin. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells expressing the indicated SPRTN variants were treated with 1 mM formaldehyde (FA) for 2 hr. (H) SPRTN-ΔC displays an aberrant subcellular localization. Doxycycline-induced YFP-SPRTN-Strep HeLa Flp-In TRex cells were analyzed using immunofluorescence. (I) SPRTN-ΔC is recruited to chromatin upon the induction of DPCs. Doxycycline-induced YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated with formaldehyde (FA, 0.5 mM) and subjected to pre-extraction, prior to fixation and immunofluorescence. (J) SPRTN-ΔC complements the formaldehyde sensitivity of SPRTN-deficient cells only partially. HeLa Flp-In TRex cells bearing the indicated doxycycline-inducible YFP-SPRTN-Strep alleles were transfected with siRNA against endogenous SPRTN and incubated in the absence or presence of doxycycline for 48 hr. Cells were then treated for 48 hr with 100 μM formaldehyde, and cell numbers were determined after an additional 4-day incubation. Values indicate cell numbers relative to untreated cells. Error bars represent SD of two to four replicates. Knockdown and doxycycline induction were confirmed by western blotting. Please note that autocleavage bands appear at similar positions as endogenous SPRTN in cells expressing WT YFP-SPRTN. See also Figure S5.

Figure 7

Autocleavage Controls SPRTN Dynamics at Sites of DNA Damage (A) SPRTN autocleavage occurs in trans. Recombinant GST-SPRTN-Strep WT and catalytically inactive YFP-E112Q-Strep were incubated in the absence or presence of ss and ds phage DNA (10 nM) for 2 hr at 25°C. (B) SPRTN autocleavage occurs in cells. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells expressing the indicated SPRTN variants were lysed and subjected to SDS-PAGE followed by western blotting against the N-terminal YFP and the C-terminal Strep tag. The asterisk indicates an unspecific band serving as loading control. (C) SPRTN autocleavage is triggered by formaldehyde. Doxycycline-inducible YFP-SPRTN-Strep HeLa Flp-In TRex cells were treated with the indicated dose of formaldehyde for 2 hr. (D) Schematic representation shows laser microirradiation and fluorescence recovery after photobleaching (FRAP) experiments. (E) Recruitment of YFP-SPRTN-Strep (WT or EQ) in HeLa Flp-In TRex cells after laser microirradiation. Data are from ≥20 cells ± SEM normalized to pre-irradiation fluorescence. (F) Representative images of HeLa Flp-In TRex cells expressing WT YFP-SPRTN-Strep from FRAP time course at indicated time following bleaching. Bleaching was achieved with 0.1-s pulse of 405-nM laser (scale bar, 10 μm). (G) FRAP from HeLa Flp-In TRex cells expressing YFP-SPRTN-Strep (WT or EQ) data are from ≥15 cells ± SEM normalized to pre-bleach fluorescence (left panel). Fitted exponential fluorescence recovery of FRAP data is shown (right panel).