Supplementary Figure 1 Linkage preference of USP30 and USP domain architecture. a, Catalytic efficiencies determined from Coomassie-stained gel-based kinetics for all eight diub linkages (see Supplementary Data Set 1 for raw data). Error bars represent standard deviations of the mean, obtained from five (Lys6), three (Lys11, Lys48) or two (Lys27, Lys33, Lys63, Met1) time courses. b, Time course of tetraubiquitin cleavage for Lys6, Lys11, Lys48, Lys63 and Met1-linked chains, resolved by SDS-PAGE and visualised by silver staining. The gel of Lys6-tetraubiquitin from Fig. 1c is included for comparison. It shows at the earliest time points that tri- and monoubiquitin are generated, consistent with exo-dub activity (Hospenthal, M.K et al., Nat Struct Mol Biol 20, 555 565, 2013). c, Annotation of the USP domain of USP7 (PDB 1NBF, Hu, M. et al., Cell 111, 1041 1054, 2002) according to the box system with insertion points indicated (Ye, Y. et al., Mol Biosyst 5, 1797 1808, 2009). d, USP domain nomenclature indicating thumb, finger and palm subdomains (Hu, M. et al., Cell 111, 1041 1054, 2002). e, Plot of average sequence conservation of USP30 (rolling window average of 9 residues. 10: high sequence conservation, 0: low sequence conservation, obtained from ConSurf, http://consurf.tau.ac.il/2016/) and deuterium incorporation obtained from HDX experiments on USP30 c2 (compare Fig. 1d), suggesting flexibility of regions with low sequence conservation.
Supplementary Figure 2 USP30 construct optimization through insertion deletions. a, Sequence alignment of the C-terminal end of box 4 of USP30 with other USPs that have been crystallised. Of note, the corresponding residues of Phe348, Met350 and Ile353 are typically hydrophilic residues that are solvent exposed on the domain backside in other USP structures, and were mutated in USP30 to increase protein solubility (see Supplementary Table 1). b, Coomassie-stained gel of indicated USP30 proteins used for crystallography. c, Melting temperatures of apo and covalent Ub-PA complexes of indicated USP30 constructs, indicating similar melting temperatures for constructs with or without the box 4-5 insertion and with or without mutation of the three hydrophobic residues shown in a, suggesting that the protein is similarly stable once purified. In contrast, additional deletion of the box 2-3 insertion significantly increased protein stability. Crystallised proteins are indicated with an arrow. Experiments were performed in triplicate and error bars represent standard deviation of the mean. d, Representative anisotropy traces for indicated USP30 proteins in the ubiquitin-kg-tamra cleavage assay. e, Catalytic efficiencies of indicated constructs derived from ubiquitin-kg-tamra cleavage. Experiments were performed in triplicate and error bars represent standard error. f, Lys6- diubiquitin cleavage assay for indicated constructs.
Supplementary Figure 3 Electron density and SEC-MALS experiment. Electron density maps in this Figure correspond to weighted 2 F O F C electron density contoured at 1 (0.8 for the 3.6 Å USP30 c8 ~Ub-PA structure). a, Overall view of the asymmetric unit for USP30 c13 ~Ub-PA. b, Overall view of the asymmetric unit for USP30 c8 ~Ub-PA. c, Close-up view of the covalently modified catalytic triad residues of USP30 c13 ~Ub-PA (top) and USP30 c8 ~Ub-PA (bottom). d, Size-exclusion chromatography multi-angle light scattering (SEC-MALS) experiment on complex formation of USP30 c13i and Lys6-diubiquitin. While a clear shift is apparent from the elution profile, the observed complex mass is smaller than expected, indicating that the complex is not fully gel-filtration stable at the concentrations tested (80 M sample concentration, 1.2-fold molar excess of Lys6-diubiquitin). e, Overall view of the asymmetric unit for USP30 c13i + Lys6-diUb. f, Close-up view of the isopeptide bond in the USP30 active site from two different perspectives.
Supplementary Figure 4 HDX-MS analysis of ubiquitin binding to USP30. a, Deuterium incorporation measured by HDX-MS on USP30 c8 mapped onto the USP30 molecule from USP30 c13i -Lys6-diUb. Data for residues not part of the model are shown as horizontal bars. b, Deuterium incorporation measured by HDX-MS on USP30 c8 ~Ub-PA mapped as in a. c, Subtraction of data from HDX-MS runs on USP30 c8 in the presence and absence of Ub-PA mapped onto the model. The deuterium incorporation in the palm and thumb domain is virtually unchanged by ubiquitin binding. There is a significant reduction in deuterium uptake in the fingers domain, suggesting rigidification through ubiquitin binding. The box 2-3 loop insertion (residues 179 217) has regions of high deuterium uptake (>45%) in both the apo and the Ub-PA-bound state, consistent with this part not being required for ubiquitin binding and with no density being observable for this part in the 3.6 Å structure of USP30 c8 ~Ub-PA.
Supplementary Figure 5 Binding of the distal ubiquitin moiety to USP30 and characterization of USP30 active site mutations. a, Sequence alignment of crystallised USP proteins centring on Leu328 and Phe453 of USP30, showing differences in the coordination of the distal ubiquitin s C-terminal tail compared to other USPs (compare Fig. 2a). b,c, Fluorescence polarisation binding experiment of USP30 c8 C77A (one loop deleted) and USP30 c2 C77A (no loop deleted) resulting in similarly poor binding of ubiquitin-kg-tamra as compared to USP30 c13i (Fig. 2b), indicating that loop deletions do not impact on the distal ubiquitin binding site of USP30. Experiments were performed in triplicate and error bars represent standard deviation from the mean. d, Fluorescence polarisation experiment of ubiquitin-kg-tamra binding to inactive USP21 C221A, as a positive control for b, c and Fig. 2b. Experiments were performed in triplicate and error bars represent standard deviation from the mean. e, Triplicate-averaged anisotropy traces for indicated USP30 proteins with active site residues mutated in the ubiquitin-kg-tamra cleavage assay. f, Observed rate constants were obtained from fitting anisotropy-time courses plotted over enzyme concentrations. The slopes correspond to catalytic efficiencies (k cat /K M ) and are plotted in Fig. 2d.
Supplementary Figure 6 Selective recognition and cleavage of Lys6-diUb by USP30. a, Fluorescence polarisation (FP) binding experiment with FlAsH-tagged Lys6-diubiquitin and indicated USP30 proteins, showing identical affinities compared to USP30 c13i (Fig. 3a), indicating that the deleted loops do not affect binding to Lys6-diubiquitin. b, Control FP binding experiment of FlAsH-tagged diubiquitin probes used in Fig. 3a with inactive USP21 C221A, indicating that all probes can bind USP domains in principle. Mean ± standard error from three independent measurements. c, Isothermal titration calorimetry (ITC) data for asymmetric Lys6-diUb* (distal Ub: K6R, K48R; proximal Ub: K48R, Gly76) binding to USP30 c13i, confirming the affinities obtained from FP binding assays. d, Superposition of free ubiquitin (PDB 1UBQ, Vijay-Kumar, S. et al., J Mol Biol 194, 531 544, 1987), the proximal Ub of Lys6-diubiquitin (PDB 2XK5, Virdee, S. et al., Nat Chem Biol 6, 750 757, 2010) and the proximal Ub in the USP30 c13i -Lys6-diUb structure. Close-up on the 1-2 loops showing how binding to USP30 induces a distinct conformation in the proximal Ub moiety. e, Sequence alignment of structurally characterised USP domains, centred on Glu159. f, Surface representation of USP30 c13i -Lys6-diUb and close-up view of the recognition site for the isopeptide linkage. The distal Ub C-terminus and the isopeptide bond are fully engulfed by the USP domain, with Glu159 of the thumb subdomain contacting the palm subdomain and coordinating the Gly76 amide backbone. g, Specificity analysis for USP30 S477D (compare Fig. 1b). h, Diubiquitin cleavage assay with indicated substrates and USP30 Ser477 mutants at identical concentrations.
Supplementary Figure 7 Analysis of Lys6-diUb upon binding to USP30. a, Comparison of proximal versus distal ubiquitin binding sites in USP30. The USP30 c13i -Lys6-diUb complex was characterised by PISA (http://www.ebi.ac.uk/pdbe/pisa/), revealing a significantly better distal as compared to proximal binding interface: larger interface area (1652 Å 2 vs. 535 Å 2 ), larger numbers of hydrogen bonds (39 vs. 8) and salt bridges (10 vs. 2) and lower solvation free energy ( 6.4 vs. 2.0 kcal/mol), suggesting that the distal ubiquitin may bind USP30 first. b, Cartoon representation of Lys6-diubiquitin (PDB 2XK5). Lys6-diubiquitin adopts a compact conformation, with Ile36, Leu8, Leu71 of the distal ubiquitin contacting the Ile44-patch of the proximal ubiquitin (Hospenthal, M.K et al., Nat Struct Mol Biol 20, 555 565, 2013). c, Superposition of a and b on the distal Ub moiety. USP30 binds the distal ubiquitin s Ile36 patch via Trp330, which requires Lys6-diubiquitin to open for the initial binding event. d,e,g, Superposition centred on the proximal ubiquitin, analogously to a,b,c. The proximal ubiquitin binding site of USP30 recognises the ubiquitin Phe4 patch, which is exposed in Lys6-diubiquitin. This binding site, hence, could recognise compact Lys6-diubiquitin and enable it to open subsequently.
Supplementary Figure 8 Characterization USP30 mutations of the proximal ubiquitin binding site. a, Anisotropy traces for indicated USP30 proteins with proximal binding site residues mutated in the ubiquitin-kg-tamra cleavage assay. b, Observed rate constants from a plotted over enzyme concentration to derive catalytic efficiencies. c, Catalytic efficiencies of indicated USP30 proteins derived from the ubiquitin-kg-tamra cleavage assay. Replacement of His445 and Trp475 leads to a reduction of catalytic activity on ubiquitin-kg-tamra, presumably due to their close proximity to the catalytic residues. Mean ± standard error are shown, derived from three independent experiments. d, Diubiquitin cleavage assay with indicated substrates and USP30 mutants, titrated to similar activity levels. H445N and D479Q retain the preference for Lys6-diUb, whereas H445S and H445Q show a cleavage pattern similar to H445E and W475K (see Fig. 3f).
Supplementary Figure 9 CYLD recognizes Lys63-diubiquitin by a distinctly different interface. a, Open-book representation of the CYLD(C596A)-Lys63-diubiquitin binding interface, compare Fig. 3d. b, Side-on view of the proximal ubiquitin binding site in CYLD. Specific recognition of the proximal Ub is achieved by extensive contacts made by CYLD s unique 12-13 insertion, which is contacted by ubiquitin Glu64, Thr14, Thr12 and Phe4, as well as coordination of Met1, Glu16 and Glu18 by the USP domain. While Phe4 packs against the aliphatic part of the Arg824 side chain, polar contacts and salt bridges overall dominate this interface. Hence, while similar ubiquitin surfaces (Phe4 patch) are involved, the distinct orientations of the proximal ubiquitin in USP30 and CYLD dictate different linkage preferences, facilitated by unique structural elements.
Supplementary Figure 10 Parkin-mediated USP30 ubiquitination. a, Left, ubiquitination of USP30 c2 by phosphorylated Parkin, visualised by western blotting with an anti-usp30 antibody. The antibody also detects a prominent proteolytic fragment of USP30 (marked with *) formed by a cleavage in the flexible part of the box 2-3 insertion. As for USP30 c13 (Fig. 5a) ubiquitination is enhanced for inactive USP30, suggesting that USP30 can auto-deubiquitinate in trans. Mass spectrometry of these samples identified several (Lys289, Lys235, Lys310) ubiquitination sites. See Supplementary Table 2 for mass spectrometry results. Right, the same samples blotted with an anti-ubiquitin antibody, revealing extensive polyubiquitin assembly by activated Parkin. This contrasts the USP30 blot which does not show evidence for USP30 polyubiquitination. b, Gel shown in Fig. 5c. Anion-exchange chromatography allowed partial separation of F4R-ubiquinated and unmodified USP30 with coeluting E1 and Parkin species. c, Superposition of USP30 c13 ~Ub-PA and USP30 c13i -Lys6-diUb, and close-up views on (1) the coordination of Phe4 of the distal ubiquitin moiety by the fingers domain, explaining the rigidification observed in HDX-MS through ubiquitin binding in this region as well as the inability of USP30 to process Ub F4R, (2) the coordination of Lys6 of the distal ubiquitin moiety by the side chain of Glu227 and the carbonyl backbone of Phe273, consistent with the observed exo activity in Lys6-tetraubiquitin cleavage experiments (Fig. 1b).
Supplementary Figure 11 PINK1-mediated monophosphorylation of Lys6-diubiquitin on the distal Ub moiety. a, Asymmetric Lys6-diubiquitin was monophosphorylated by PhPINK1, and following PhPINK1 heat-inactivation cleaved by USP21. Deconvoluted mass spectra are shown. The major phosphorylated species corresponds to the distal Ub moiety. The small peak at 8,618 Da may contain both unphosphorylated distal Ub as well as phosphorylated proximal Ub, which could not be separated due to limited resolution of the mass spectrometer used. b, Close-up view on the distal Ub Ser65 environment of USP30 (see Fig. 6d for superposition in full). c, Deconvoluted mass spectra of the ubiquitin-kg-tamra and the Ser65-phosphorylated phosphoub-kg- TAMRA reagents. Oxidation products (+ 1 oxygen, +16 Da), present also in the untreated sample, are indicated by + [O]. d,e Triplicate averaged anisotropy time traces and observed rate constants for USP30-mediated cleavage of ubiquitin- and phosphoub-kg-tamra (see Fig. 6e for catalytic efficiencies).
Supplementary Figure 12 Characterization data on differently phosphorylated Lys6-diUb substrates. a, Sample characterisation data for assay shown in Fig. 6f. USP30 c13i was ubiquitinated by Parkin, the sample then split and one half treated with PINK1. Analysis of the ubiquitin smear indicates the presence of Ser65-phosphoubiquitin only in the PINK1-treated sample. USP30 c13i ~Ub is not recognised by either the ubiquitin or the pser65-ubiquitin antibody. b, Asymmetric Lys6-diUb chains with different phosphorylation states were obtained through assembly from phosphorylated monoub species. Deconvoluted intact protein mass spectra for reagent characterisation are shown for the chains and the respective monoub species obtained after treatment with USP21 (compare Supplementary Fig. 11a).
Supplementary Figure 13 Regulation through USP30 overexpression of Lys6-linked polyubiquitin chains on MOM proteins unaffected by USP30 knockdown. a, Immunoblotting for CISD1 in eluates of Lys6-linked polyubiquitin chain pull-downs from HeLa Flp-In T-REx cells, doxycyclineinducibly expressing wild-type Parkin. Cells treated as indicated, all samples were treated with USP21 to deplete other polyubiquitin linkages. CISD1 forms a constitutive dimer, explaining the pull-down of monoubiquitinated CISD1 protein forms. b, Experiment as in a with USP30 knock down. Immunoblotting for VDAC1, MIRO1 and CISD1. *, non-specific bands. c, Quantification from b with n = 3 independent experiments. Mean ± s.e.m. Wilcoxon-Mann-Whitney test. n.s., non-significant. d, Experiment as in a with overexpression of indicated USP30 variants via transient transfection. e, Quantification from d. Mean ± s.e.m. One-Way ANOVA (F = 48) with Dunnett s correction, n = 3 independent experiments. ***, P < 0.001.
Supplementary Figure 14 Model of localized mitophagy. USP30 displays normal levels of activity in the healthy part of the mitochondrial network. In the zone around mitochondrial damage, however, USP30 activity would be reduced due to the activity of PINK1, the presence of phosphoubiquitin and the impaired catalytic of USP30 for phosphoubiquitin substrates.
Supplementary Table 1: USP30 constructs used in this study. Construct Boundaries Mutations Box 4/5 deletion Box 2/3 deletion Solubility Used for 1 54 517 wt - - good 2 64 502 wt - - good 3 64 502 wt (358-431)SGS - poor 4 64 502 wt (358-431)SNA - poor 5 64 502 wt (376-431) - poor 6 64 502 F348Q, M350N, I353Q (358-431)SNA - moderate 7 64 502 F348D, M350S, I353E (358-431)SNA - good 8 64 502 F348D, M350D, I353E (358-431)SNA - good 3.6 Å ~Ub-PA structure 9 64 502 F348G, M350N, I353E (358-431)SNA - good 10 64 502 F348D, M350S, I353E (358-431)SNA (187-192) good 11 64 502 F348D, M350S, I353E (358-431)SNA (195-199) good 12 64 502 F348D, M350S, I353E (358-431)SNA (194-204) good 13 64 502 F348D, M350S, I353E (358-431)SNA (179-216)GSGS good 13i 64 502 C77A, F348D, M350S, I353E (358-431)SNA (179-216)GSGS moderate 2.3 Å ~Ub-PA structure 2.8 Å structure of USP30 / Lys6- diub complex
Supplementary Note 1: Sequence alignment of USP30 orthologs. The mitochondrial intermembrane part, the transmembrane helix, the polybasic signal anchor sequence and the USP boxes are indicated. Residues are coloured according to sequence identity. Select residues are marked with stars according to the legend. Of note, several USP30 orthologs have a C-terminal truncation similar to the construct used for structure determination. Strikingly, a subset of USP30 sequences in the lower part of the alignment shows hydrophilic residues (Ser, Arg) at the positions of the hydrophobic residues that were mutated along with the box 4-5 insertion deletion, suggesting that hydrophobic contacts between the insertion sequence and the USP domain at this position are not universally conserved. mitochondrial intermembrane part transmembrane helix anchor USP box 1 USP box 2 box 2/3 insertion deletion (179-216) USP box 2 USP box 3 box 4/5 insertion deletion (358-431) USP box 4 USP box 5 USP box 6 catalytic residues residues in interface for proximal Ub moiety key residues responsible for weakened distal Ub binding hydrophobic residues mutated for box 4/5 insertion deletion