Extended Data Figure 1: Protein complexes used in this study.

a–d, f, j, k, SDS–PAGE analysis of CSN variants used in the study. g, The _N8CRL4A^DDB2 substrate used for enzymatic measurements. a, The CSN5 E76A mutant catalytically inactivates CSN: activity of CSN (full-length) and CSN (CSN5(E76A)), a holocomplex carrying the active site mutant, was determined by fluorescence polarization measurements using a substrate with a PT22-labelled _N8CRL4A^DDB2. The decrease in signal for CSN (CSN5(E76A)) and in the buffer control is due to fluorophore photobleaching. Determination of steady-state kinetics using _N8CRL4A^DDB2 as substrate: initial rates observed following incubation of CSN or mutants thereof, with increasing concentrations of PT22-labelled _N8CRL4A^DDB2 substrate, as indicated on the abscissa. The fit of the observed initial velocities to the Michaelis–Menten equation is shown as a red line in b, endogenous CSN purified from HEK293T cells (CSN (HEK293T)); c, recombinant full-length CSN (CSN (full-length)); d, CSN with the boundaries used for crystallization (CSN); e, CSN (CSN6^ΔMPN), lacking the CSN6 MPN domain (comprising residues 192–327); and f, CSN (CSN6^Δloop), lacking residues 174–179 of the CSN6 Ins-2 loop. a–f, Data are the average of three technical replicates. c, e, f, Error bars show standard error of the mean (s.e.m.). l, Table of steady-state kinetic parameters (errors show ± s.e.m.). A previous study⁴⁶ described a CSN variant where the mouse CSN6 MPN domain was deleted (retaining CSN6 residues 171–324). This construct when immunoprecipitated from cells with the remainder of CSN was found to be active. For human CSN (CSN6^ΔMPN), detailed quantitative kinetic analysis revealed a ∼100-fold reduction in k_cat compared with CSN (full-length). Interestingly, mouse CSN6 171–324 retains the residues involved in the CSN4–CSN6 interface including the Ins-2 loop. Determination of steady-state kinetics using ubiquitin-rhodamine 110 as substrate: CSN variants were assayed for proteolytic release of the rhodamine 110 fluorescent group from the C-terminal glycine of ubiquitin (ubiquitin-rhodamine), using fluorescence quenching as readout. h, Wild-type CSN was not sufficiently active on ubiquitin-rhodamine to determine Michaelis–Menten parameters. To benchmark wild-type CSN against CSN (CSN5(E104A)), we assessed the relative rates at a fixed concentration of 0.5 µM ubiquitin-rhodamine substrate and 1 nM CSN (CSN5(E76A)), CSN (full-length) (23.0 ± 2.9 fmol s⁻¹) and CSN (CSN5(E104A)) (137.8 ± 2.2 fmol s⁻¹), using the CSN (CSN5(E76A)) active site mutant as a control. i, j, k, Increasing concentrations of ubiquitin-rhodamine with CSN (CSN6^Δloop) (i), CSN (CSN5(E104A)) (j) and CSN (CSN6^Δloop, CSN5(E104A)) (k) double mutant were assayed. Fit of the initial velocities to the Michaelis–Menten equation is shown as a red line. h–k, Data are the average of three technical replicates. m, Table summarizing the activity of CSN variants on ubiquitin-rhodamine (errors show ± s.e.m.). Assayed protein concentrations and V_max values are indicated.