USP48 specifically deubiquitinates H2ABRCA1ub. a Schematics of the fluorescence polarization screen. Recombinant nucleosome core particles are site specifically ubiquitinated using the E3 ligase named and ^TAMRAUb. Upon addition of a DUB, cleavage can be followed by a decrease in FP signal. b USP48 prefers nucleosomal substrates. Site-specific cleavage of H2A^168ub, H2A^PRC1ub, and H2A^BRCA1ub in relation to activity on minimal substrate. Observed k-values from fitting the raw data of the FP assay (k_obs(NCP)) (Supplementary Fig. ) were normalized to observed k-values on minimal substrate (k_obs(Rho)) obtained from fitting an exponential function to the traces in Supplementary Fig. ). A value above one indicates that NCPs are the preferred substrate. The mean of two technical replicates ± SEM is shown. c Time course of USP48 cleavage of H2A^168ub, H2A^PRC1ub, and H2A^BRCA1ub, anti-H2A western blotting. USP48 only cleaves efficiently when more than one ubiquitin is present on the BRCA1 site. The blot shown is representative of two experiments. d USP48 cleaves multi-monoubiquitinated H2A^BRCA1ub faster than multi-monoubiquitinated H2A^168ub and H2A^PRC1ub. Quantification of the initial reaction velocity (V₀) of USP48 measured by the liberation of free ^TAMRAUb. Gels used for quantification are shown in Supplementary Fig. . The mean of two technical replicates ±SEM is shown. e USP48 does not cleave H2A^PRC1ub when H2A^BRCA1ub is present, but all H2A^BRCA1ub is cleaved. Left panel: cleavage of NCPs, ubiquitinated on PRC1 site with unlabeled ubiquitin, and on the BRCA1 site with ^TAMRAub by USP48 was followed on gel. The TAMRA fluorescence is used as readout. Right panel: quantification of the initial reaction velocity (V₀) measured by liberation of free ubiquitin. The mean of two technical replicates ± SEM is shown

USP48 cleavage rates on H2ABRCA1ub are modulated by an auxiliary ubiquitin. a An auxiliary ubiquitin activates USP48 processivity. Gel-based assay to measure USP48 cleavage of H2A^BRCA1ub. ^TAMRAUb was used as readout. Cleavage was recorded under several different substrate and enzyme concentrations. Four (stacked) gels are shown as an example (Supplementary Fig.  for all gels). b Quantification of (a. Four out of 11 different conditions shown as an example (Supplementary Fig.  for full panel). Solid lines show fit obtained by global fitting of all 11 quantified time courses and binding data to the model defined in e using Kintek Explorer. c USP48 binds with similar affinities to ubiquitinated and unmodified NCP. USP48 is titrated to 50 nM of unmodified NCP or NCP^BRCA1ub and analyzed by native gel-shift assays. A Kd of 1 µM is estimated. d USP48 binds nucleosomes of different ubiquitination status with the same affinities. Binding of USP48 to H2A^BRCA1ub, H2A^BRCA1ub1, and unmodified nucleosomes measured by surface plasmon resonance. Nucleosomes were immobilized on the surface. Normalized RU values of 10 successive injections of different USP48 concentrations are fitted by a one-phase binding model using GraphPad Prism (raw data in Supplementary Fig. ). K_d and SE of the fit are indicated. e USP48 cleaves H2A^BRCA1ub in nucleosomes 50–150 times faster than when auxiliary ubiquitin is present. Kinetic model describing USP48’s cleavage pattern on H2A^BRCA1ub with the fitted values for k_cat(ub3), k_cat(ub2), and k_cat(ub1), and the SE of the fit. f Parameters for k_cat(ub3), k_cat(ub2), and k_cat(ub1) in e are well constrained by the data. Evaluation of the goodness of fit. The upper three panels show how well defined the lower and upper boundaries are for the individual parameters. k_cat(ub1) and k_cat(ub2) fall into a well-defined local χ² minimum. For k_cat(ub3), the lower boundaries are well defined which allows the conclusion that K_cat(ub3) should always be faster than k_cat(ub1). The lower panel shows how χ² varies when two of the fitted variables are varied against each other. Red indicates a χ² minimum

USP48 deubiquitinase activity restrains resection. a USP48 is recruited to sites of damage. BrdU-sensitized HeLa cells were subjected to laser stripe IR followed by fixation after 30 min and staining for endogenous USP48, BRCA1, and 53BP1. Scale bar = 10 µm. In the middle and lower panel, cells were treated with USP48 (exon 5 and exon 11 targeting sequences) and BRCA1 siRNA, respectively. USP48 siRNA treatment utilized exon 5 and exon 11 targeting sequences together throughout the manuscript, unless otherwise stated. b Percentage of USP48-positive laser lines in control vs BRCA1-depleted cells. Graph is mean from three independent experiments. Error bars are SEM and *p < 0.05 Student’s T-test. c, d Expression levels of USP48 in stable HeLa-FlpIn cell lines following depletion of USP48 and DOX-induced expression of siResistant EGFP-USP48^Iso1 c or Flag-USP48^Iso2-WT or Flag-USP48^Iso2-C98S d. e, f USP48 depletion increases resection lengths. Resection lengths were measured in HeLa cells depleted for USP48 and complemented as in c and d. Cells were incubated with 10 mM BrdU for 24 h with addition of 10 µM Olaparib for the final 16 h. Cells were lysed and DNA fibres spread before staining for BrdU-positive single-stranded DNA resection tracks. One hundred and twenty fibres were measured using ImageJ for each condition. Bars indicate median, also shown numerically in brackets. ***p < 0.005 Mann–Whitney test

USP48 antagonizes BRCA1-mediated resection. a Increased resection seen on USP48 knockdown requires BRCA1 and SMARCAD1. Resection lengths were measured in HeLa cells depleted as indicated. Cells were treated with 10 mM BrdU for 24 h with addition of 10 µM Olaparib for the final 16 h. Cells were lysed and DNA fibres spread before staining for BrdU-positive single-stranded DNA resection tracks. n = 190 tracks for each condition. Bars indicate median, also shown numerically in brackets. ***p < 0.005 Mann–Whitney test. b Western blottings to demonstrate protein expression levels in HeLa cells following siRNA as indicated. c Rad51 foci formation in S-phase (EdU positive) HeLa cells depleted as indicated. Cells were fixed at 2 h post 5 Gy IR. Scale bars = 10 µm (see Supplementary Fig.  for quantification). d Suppressive impact of USP48^Iso2 overexpression on resection lengths can be rescued by co-expression of an uncleavable H2A-Ub fusion. Resection lengths were measured in untransfected HeLa cells or those transfected with USP48^Iso2 and expressing either H2A or H2A-Ub fusion (KR2 denotes that lysines 13/15, 118/119, and 125/127/129 were mutated to arginines). Cells were prepared as in a and 210 tracks were measured for each condition. Bars indicate median, also shown numerically in brackets. ***p < 0.005, NS = nonsignificant, Mann–Whitney test. e USP48 restricts positioning of 53BP1 in damage foci. Images of BRCA1 and 53BP1 in cells treated with control or USP48 siRNA exposed to 2 Gy IR and fixed 8 h later. Scale bars = 2 µm. Quantification of mean relative intensity profiles for co-localizing foci. n = 25, bars = SEM. Right panel shows western blotting of USP48 protein levels

USP48 restrains homology-directed repair mechanisms. a Gene conversion (GC) was measured using U20S DR3-GFP reporter cells as illustrated. It is noteworthy that the functional GFP cannot be produced by SSA from this substrate as the template iGFP lacks the 5′- and 3′-regions of GFP. Cells depleted for USP48 were transfected with RFP, I-Sce1, and either Flag-USP48^Iso2-WT or Flag-USP48^Iso2-C98S. GFP-positive cells were normalized to RFP-transfection efficiency. %-repair is given compared with NTC. Graph shows mean, n = 5, error bars are SEM. b Single-strand annealing (SSA) was measured using U20S SA-GFP reporter cells as illustrated. The two GFP fragments of the substrate share 266 nucleotides of homology. In principle GC with crossing over could also produce functional GFP; however, these have been shown to be rare events. Cells were treated and analysed as in a. Graph shows mean, n = 3, error bars are SEM. c, d Camptothecin colony survival curves of HeLa cells depleted for USP48 and complemented with EGFP-USP48^Iso1-WT, n = 3 c or Flag-USP48^Iso2-WT or Flag-USP48^Iso2-C98S, n = 4 d. Graph shows mean % survival normalized to untreated controls, error bars are SEM. e Western blottings to demonstrate USP48, RAD51, and RAD52 protein expression levels in HeLa cells following siRNA as indicated. f, g Camptothecin colony survival curves of HeLa cells depleted for USP48, RAD51, and both USP48 and RAD51 f, and USP48 and RAD52 alone or together g. Graph shows mean % survival normalized to untreated controls, n = 3, error bars are SEM

Model of USP48 function in restricting extended DNA end resection. a USP48 cleaves the C-terminal ubiquitin modification of H2A, limiting the extent of SMARCAD1 nucleosome remodeling and 53BP1 positioning. 53BP1 in turn restrains DNA end processing and consequently direct repeats are rarely exposed either side of the DSB, favoring GC (one side of the break is illustrated). Without USP48 activity, H2A modification is unopposed and SMARCAD1 nucleosome remodeling is extended further from the break, resulting in 53BP1 positioning further from the break site. 53BP1 is unable to constrain extended resection, resulting in increased GC, and, as repeats either side of the break are more often exposed, SSA is also increased. b USP48 cannot cleave BRCA1-initiated H2A monoubiquitination efficiently, therefore SMARCAD1 can bind and initiate downstream signaling. c A second ubiquitination on H2A can activate USP48, cleaving BRCA-initiated ubiquitination; therefore, SMARCAD1 cannot engage with the nucleosome anymore and signaling is stopped