CSN was initially identified based on the
CSN was initially identified based on the constitutitvely photormophogenic (cop) mutants from Arabidopsis thaliana . csn mutants are Cullin deneddylation-deficient, consequently accumulate neddylated Cullins, and arrest growth shortly after germination [49, 50, 51]. Weak csn mutants have defects in auxin responses, which has allowed linking CSN to SCFTIR1 regulation [22,50,52,53]. The phenotypic similarity between neddylation (e.g. axr1), deneddylation (csn) and auxin response mutants (e.g. tir1) can be explained by their inability to assemble or stabilize functional SCFTIR1 and degrade Aux/IAAs such as AXR3 .
An Arabidopsis mutant of SMAP1 (SMALL ACIDIC PROTEIN1) also displays auxin insensitive growth and interacts genetically with SCFTIR1 and its regulators . SMAP1 immunoprecipitations purify (almost) the entire COP9 signalosome and a SMAP1-related protein was identified in human CSN purifications [54, 55, 56, 57]. SMAP1 may thus constitute an additional CSN subunit but its functional role remains as yet unclear.
CAND1 is a substrate receptor exchange factor Deneddylated Cullins can be bound by CAND1 (Figure 3) [38, 39, 40]. CAND1 has a sinusoid structure and winds itself around the two ends of the Cullin. CAND1 regulates CRL assembly be preventing substrate receptor binding and Cullin neddylation . CAND1 also recruits DCN1 and can accelerate substrate receptor exchange on unneddylated Cullins by a factor of 106 [38, 39, 40,46]. In turn, CAND1 is unable to interact with neddylated Cullins, thus Cullin neddylation stabilizes CRLs, which will remain active until substrates are degraded, allowing the access of CSN to the neddylated Cullin (Figure 3). Also Arabidopsis cand1 mutants display auxin insensitivity phenotypes . In yeast, it was shown that cand1 as well as csn mutants fail to release substrate receptors . Thus, although csn mutants contain functional (neddylated) CRL complexes, they cannot respond to signals that demand the formation of new CRLs, for example when new substrate-loaded receptors become available. cand1 mutants, in turn, fail to exchange the substrate receptors and thereby become signalling-deficient.
Due to the sequence conservation between NEDD8 and ubiquitin, neddylation and ubiquitylation could be similarly complex with regard to substrate number, modifications and MMP-2/MMP-9 Inhibitor I of cellular functions. The biochemical identification of neddylated proteins, besides Cullins, is hampered by the fact that neddylated proteins, although detectable by immunoblotting, are low abundant and that neddylation and ubiquitylation have identical C-termini and leave identical mass footprints after trypsin digestion on the modified protein [2,15]. The fact that neddylated Cullins participate in ubiquitylation leads to further ramifications, as does the fact that the ubiquitylation machinery can use NEDD8 as substrate when NEDD8 is present in excess [,]. Although several laboratories have identified non-Cullin NEDD8-modified proteins, the functional relevance of these modifications has to be treated with caution, particularly in view of the recently uncovered cross-reactivity of the neddylation and ubiquitylation reactions [15,17,,]. DENEDDYLASE1 (DEN1) is a proposed NEDD8 processing enzyme . NEDD8 conjugates of a broad molecular weight range accumulate in Arabidopsis den1 mutants, suggesting the existence of non-Cullin NEDD8 conjugates when NEDD8 is present at physiological levels . Although DEN1 has NEDD8 processing activity in vitro, no NEDD8 processing defect is detected in Arabidopsis den1 mutants; rather DEN1 seems necessary for deneddylation of NEDD8 conjugates (Figure 4) . Despite their strong biochemical phenotype, den1 mutants have very subtle growth defects . AXR1 is one of two prominent NEDD8-conjugates in den1 and the auxin insensitivity phenotypes of axr1 are enhanced in axr1 den1 [16,]. Importantly, neddylation of multiple AXR1 Lys (as well as ECR1 Lys) could be observed in in vitro neddylation reactions, even in the absence of a NEDD8 conjugating enzyme, suggesting that these neddylations must be the result of non-targeted enzyme-independent auto-neddylation promoted by the highly reactive thioester-linked NEDD8 on the NAE (Figure 4) . Similarly, the other neddylation enzymes ECR1 and RCE1, that share the property of being physically close to the NAE, were also found to be neddylated. Since transgenic expression of NEDD8 could suppress the phenotypes of axr1 den1, a depletion of the pool of free NEDD8 rather than impaired AXR1 function must be at the basis of the observed phenotypes . Thus, neddylation of non-Cullin substrates as observed in eukaryotic cells may be the result of untargeted auto-neddylation rather than targeted regulatory conjugation. In this context, neddylation of enzymes of the neddylation pathway may be comparatively prominent due to the physical proximity of the protein to the NEDD8 activation process.