Deconjugation of Nedd8 from Cul1 is directly regulated by Skp1-F-box and substrate, and the COP9 signalosome inhibits deneddylated SCF by a noncatalytic mechanism

Abstract

COP9 signalosome (CSN) mediates deconjugation of the ubiquitin-like protein Nedd8 from the cullin subunits of SCF and other cullin-RING ubiquitin ligases (CRLs). This process is essential to maintain the proper activity of CRLs in cells. Here, we report a detailed kinetic characterization of CSN-mediated deconjugation of Nedd8 from SCF. CSN is an efficient enzyme, with a k(cat) of ~1 s(-1) and K(m) for neddylated Cul1-Rbx1 of ~200 nm, yielding a k(cat)/K(m) near the anticipated diffusion-controlled limit. Assembly with an F-box-Skp1 complex markedly inhibited deneddylation, although the magnitude varied considerably, with Fbw7-Skp1 inhibiting by ~5-fold but Skp2-Cks1-Skp1 by only ~15%. Deneddylation of both SCF(Fbw7) and SCF(Skp2-Cks1) was further inhibited ~2.5-fold by the addition of substrate. Combined, the inhibition by Fbw7-Skp1 plus its substrate cyclin E was greater than 10-fold. Unexpectedly, our results also uncover significant product inhibition by deconjugated Cul1, which results from the ability of Cul1 to bind tightly to CSN. Reciprocally, CSN inhibits the ubiquitin ligase activity of deneddylated Cul1. We propose a model in which assembled CRL complexes engaged with substrate are normally refractory to deneddylation. Upon consumption of substrate and subsequent deneddylation, CSN can remain stably bound to the CRL and hold it in low state of reduced activity.

FIGURE 1.

Characterization of in vitro deneddylation assay components and enzymatic properties of human CSN. A, purified CSN from HEK293 cells was fractionated by SDS-PAGE and analyzed by silver staining to check for purity and stoichiometry of enzyme subunits. B, purified Cul1-Rbx1 (50 nm) conjugated with ³²P-labeled HPN8 was incubated with 1 nm CSN in a total reaction volume of 50 μl. At the indicated time points, the aliquots were quenched and evaluated by SDS-PAGE followed by autoradiography. C, PhosphorImager quantification of B. D, the initial rate of deneddylation by 0.8 nm CSN at different concentrations of substrate is plotted. K_m and k_cat were estimated by fitting the curve to the Michaelis-Menten equation.

FIGURE 2.

F-box-Skp1 and substrate inhibit deneddylation by CSN. A, purified Cul1-Rbx1 (500 nm) conjugated with ³²P-labeled HPN8 was preincubated for 10 min with 600 nm of Fbw7-Skp1 or β-TrCP-Skp1, followed by the addition of 1 nm CSN. At the indicated time points, aliquots were quenched and evaluated by SDS-PAGE (left panel) followed by PhosphorImager quantification (right panel). The rates (fmol of Nedd8 released from Cul1/s) are indicated to the right of each curve. The total reaction volume was 25 μl. B, same as A, except ³²P-labeled HPN8-Cul1-Rbx1 substrate at 50 nm was mock incubated or preincubated with 100 nm Skp1-Skp2-Cks1 in the presence or absence of 1 μm phospho-p27-cyclin E-Cdk2 (p27). Total reaction volume was 50 μl. The asterisk indicates [³²P] label incorporated into p27. C, same as A, except the ³²P-labeled HPN8-Cul1-Rbx1 substrate was at 50 nm, Fbw7-Skp1 was at 100 nm, and phospho-cyclin E-Cdk2 (CycE) was at 500 nm. D, same as C, except that Fbw7-Skp1 was omitted. E, ³²P-labeled HPN8-Cul1-Rbx1 substrate at 100 nm was preincubated 5 min with 300 nm Skp2-Skp1 plus or minus Cks1. Following assembly of SCF^Skp2 complexes, the reactions were supplemented with ubiquitylation components (1 μm ubiquitin, 400 nm E1, 100 nm Cdc34, plus or minus 500 nm phospho-p27-cyclin E-Cdk2), incubated for 10 min, supplemented with ATP and Mg²⁺ to initiate ubiquitylation, and incubated a further 20 min prior to addition of CSN (0.8 nm). The total reaction volume was 50 μl. At the indicated time points, aliquots were quenched and evaluated by SDS-PAGE followed by phosphorimaging (top panel). Quantification of the phosphorimaging scans is shown in the bottom panel.

FIGURE 3.

Proteins that bind the C-terminal domain of Cul1 inhibit deneddylation. A, ³²P-labeled HPN8-Cul1-Rbx1 (25 nm) was incubated for 10 min with 1 μm of the indicated factor prior to addition of 0.8 nm CSN. At the indicated time points, the aliquots were quenched and evaluated by SDS-PAGE followed by PhosphorImager quantification. The total reaction volume was 40 μl. B, same as A, except that substrate was 150 nm and was preincubated with the indicated final concentration of Cand1 prior to adding CSN. C, same as B, except that Cand1 (250 nm) and CSN (0.8 nm) were held constant, whereas the concentration of substrate was varied. The data were fitted to the Michaelis-Menten equation to estimate k_cat and K_m. D, ³²P-labeled HPN8-Cul1-Rbx1 (50 nm) was preincubated with 200 nm Skp1-Skp2-Cks1 for 10 min prior to addition of the indicated final concentrations of Cand1. Following a further 10 min of precincubation, CSN (0.8 nm) was added. At the indicated time points, the aliquots were quenched and evaluated by SDS-PAGE followed by PhosphorImager quantification. E, same as B, except that substrate was 50 nm and Cul1-Rbx1 was titrated.

FIGURE 4.

CSN forms a stable complex with both neddylated and unmodified Cul1. A, purified SCF^Skp2-Cks1 (600 nm) was incubated for 15 min in either the presence (top panel) or the absence (bottom panel) of 300 nm purified CSN. Complexes were passed through a Sephadex 200 size exclusion column, and every third fraction was separated by SDS-PAGE and Western blotted with antisera to the indicated proteins. B, the indicated HA-tagged Cul1 constructs were transfected into HEK293 cells. Twenty-four hours post-transfection, the lysates were generated, and ^HACul1 was immunoprecipitated with anti-HA antibody. Immunoprecipitates were fractionated by SDS-PAGE and Western blotted with antisera to the indicated proteins. Cul1 was detected with anti-HA. xRING and xSkp1 refer to point mutants of Cul1 that were deficient in binding Rbx1 and Skp1, respectively. K720R has an arginine. substitution at the Nedd8 conjugation site (lysine 720). C, the indicated FLAG-tagged Csn5 constructs were transfected into HEK293 cells. Twenty-four hours post-transfection, the lysates were generated, and ^FLAGCsn5 was immunoprecipitated (IP) with anti-FLAG antibody. Immunoprecipitates were fractionated by SDS-PAGE and Western blotted with antisera to Csn5 and Cul1 as indicated. EV refers to empty vector. ASA refers to a double point mutation that inactivates the JAMM domain of Csn5.

FIGURE 5.

CSN inhibits ubiquitylation by unmodified SCF. A, SCF^βTrCP (100 nm) and ubiquitylation components (1 μm ubiquitin, 400 nm E1, 100 nm UbcH5C, 600 nm ³²P-labeled-phospho-β catenin peptide) were incubated either in the presence or absence of 300 nm CSN for 10 min, after which ubiquitylation reactions were initiated by the addition of ATP and Mg²⁺. Time points were harvested at the indicated times, fractionated by SDS-PAGE, and subjected to PhosphorImager quantification. B, same as A, except that 100 nm Cdc34 was used in place of UbcH5C.

FIGURE 6.

Regulation of CRLs by reversible neddylation. See text for details. Transitions marked by single arrows are vectorial. The intermediates at steps 5 and 6 could re-form new CRL complexes by binding a different substrate receptor-adaptor module (dashed and curved lines, respectively).