Evidence for a physical association of the COP9 signalosome, the proteasome, and specific SCF E3 ligases in vivo

The COP9 signalosome (CSN) was initially discovered during the genetic analysis of light control of Arabidopsis seedling development and later shown to be a protein complex that is conserved among diverged organisms [1]. The accumulation of ubiquitin conjugated proteins and the failure to degrade specific proteins, such as HY5 and IAA6, in csn mutants [2–4] supports a role for CSN in the ubiquitin–proteasome pathway. In addition, CSN was found to be associated with SCF ubiquitin E3 ligases [2,5–7] and to promote the de-conjugation of NEDD8/RUB from the cullins [8]. Interestingly, both the lid subcomplex of the 26S proteasome and CSN contain eight subunits that display a remarkable subunit-to-subunit similarity [1]. Here we report evidence to support the existence of dynamic conglomerates consisting of the 26S proteasome, CSN and specific SCF E3 ligases. These substrate or E3-specific conglomerates could perform the entire substrate ubiquitination and proteasomemediated degradation in a single location within the cell, thus conferring synergy to the specificity and the efficiency of the process. Gel filtration size-fractionation was first used to examine the molecular association of CSN with known ubiquitin–proteasome pathway components. As shown in Figure 1, a small fraction of CSN (about 5–10%) co-fractionated with the 26S proteasome and SCF E3 ligase components [6] in higher molecular mass fractions, when protein phosphatase inhibitors and ATP were included in the buffer. It is possible that this interaction is labile and dynamic, therefore, our fractionation assay can, at any given moment, only detect a portion of CSN that is engaged with the proteasome and the E3 ligases. Consistent with this notion, a 30minute in vivo cross-linking treatment of the tissue with 1% formaldehyde [9] before homogenization and fractionation leads to a significant increase (up to 40%) in the portion of CSN in higher molecular mass fractions (see supplementary Figure 1). This high molecular mass CSN cofractionates with all tested subunits of the 26S proteasome and with the core subunits of SCF type ubiquitin E3 ligases. Thus, it is plausible that this cross-linking treatment probably helps to fix the transient and labile CSN conglomerates. To determine whether the cofractionating proteasome, CSN and SCF E3 ligases are physically associated, proteins were collected from the high molecular mass gel filtration fractions of extract without cross-linking treatment (fractions 2 to 4, Figure 1). These proteins were subjected to immunoprecipitation with antibodies against representative subunits of CSN, the 26S proteasome, and SCF E3 ligases. Antibodies against CSN6 co-immunoprecipitated all the tested subunits of the 26S proteasome and the SCF ubiquitin E3 ligases in addition to the CSN, whereas the pre-immune serum did not (Supplementary Figure 2). Antibodies against Rtp5, a base subunit of the 26S proteasome, also co-immunoprecipitated the subunits of CSN and SCF ubiquitin E3 ligases. To confirm that multiple SCF E3 ligases can independently form large conglomerates with CSN and the proteasome, we immunoprecipitated tagged versions of three distinct Arabidopsis F-box proteins that are known to form SCF complexes, which functionally associate with CSN [4–6]. As shown in Figure 2, all tested subunits of SCF complexes, CSN and the proteasome can be co-immunoprecipitated with the F-box proteins. This confirms that multiple SCF E3 ligases are capable of forming conglomerates with CSN and the proteasome. Together with the co-fractionation