Novel aspects of COP9 signalosome functions revealed through analysis of hypomorphic csn mutants

Abstract

The COP9 signalosome (CSN) is a conserved eukaryotic protein complex implicated in the regulation of cullin-RING type E3 ubiquitin ligases by cleaving the small peptide RUB/Nedd8 from cullins. However, detailed analysis of CSN physiological functions in Arabidopsis has been hampered by the early seedling-lethality of csn null mutants. We and others have now identified a number of viable hypomorphic csn mutants which start to reveal novel CSN-dependent activities in adult Arabidopsis plants.¹ Here, we present a detailed comparative analysis of the csn5a-1 and csn2-5 mutants as a mean to improve understanding of CSN functions in plant cells. Our observations point to CSN-independent activities of CSN5 and suggest a role of the CSN in cytoskeleton assembly/organization.

Introduction

The control of protein turnover is crucial for virtually all physiological processes. The small peptide ubiquitin can be conjugated to target proteins through the sequential actions of ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2) and ubiquitin protein ligases (E3) usually to induce their degradation by the 26S proteasome.² E3s determine substrate specificity of the ubiquitination machinery, and the largest class of E3s is represented by the cullin-RING ligases (CRLs). According to sequence-based predictions, Arabidopsis expresses approximately 800 different CRLs potentially regulating the abundance of an even larger array of substrate proteins. CRLs all share a common architecture: The catalytically active RING protein RBX1 (Hrt1/Roc1) and a substrate-recognition subunit are brought together on a cullin protein (reviewed in ref. 3). However, cullins not only act as a passive scaffold since modification of cullins by the ubiquitin-like peptide RUB/Nedd8 is needed for CRL activities and is essential for viability in many different organisms. Recent structural work showed that major conformational changes occur upon cullin neddylation, leading to an open and catalytically active conformation.⁴ Reciprocally, mutants lacking the COP9 signalosome (CSN), which cleaves RUB/Nedd8 from cullins, exhibit pleiotropic phenotypes that culminate in Arabidopsis in early seedling lethality. CSN is an eight subunit complex conserved among eukaryotes and deletion of one of its subunit leads to complete loss of CSN activity in Arabidopsis. This has largely hampered analyses on CSN functions. The identification of viable csn reduced function (hypomorphic) mutant lines has allowed analysis of developmental and physiological responses in more mature plants and, for example, revealed the importance of CSN in responses to the plant hormone auxin.^5–7

We identified an additional hypomorphic csn mutant in a screen for enhancers of the auxin resistance phenotype of sgt1b-3 that carries a mutation in one of the two functional Arabidopsis SGT1 (suppressor of G₂ allele of skp1) genes.¹ SGT1 was initially identified in yeast as a dose-dependent suppressor of skp1-4, a mutation resulting in defects in kinetochore function, and was also shown to associate with cullin1-based SCF-type (Skp, Cullin, F-Box) CRLs in different organisms.^8–10 In Arabidopsis, the response to auxin is mediated by the well-characterized CRLs SCF^TIR1/AFB,^11,12 but the precise role of SGT1 in auxin signaling remains unclear. Our newly identified csn2-5 mutant carries a point mutation in the gene encoding the CSN2 subunit, one of the six PCI (proteasome, COP9 signalosome, eIF3) domain-containing subunits of CSN. Although the resulting mutant protein accumulates at reduced levels some intact CSN can still form in the csn2-5 mutant, seen as an intermediate accumulation of neddylated cullins compared to wild type and csn null mutants. We used this mild mutant to analyze further the regulation of SCF^TIR1 as an archetypal CRL. We show that CUL1 and the F box protein TIR1, the substrate receptor of SCF^TIR1 complex, are destabilized in csn2-5 mutant tissues, thereby providing a possible explanation for the hitherto poorly understood auxin resistance phenotype of Arabidopsis csn mutants. Our results further point to posttranslational modification of TIR1 (most likely by ubiquitination) and the proteasome-mediated degradation of TIR1, ASK1 and CUL1. Similar observations had previously been reported for other organisms.^13–15 However, while the ubiquitination and subsequent degradation of substrate receptor proteins is well established in various systems, the reduced accumulation or increased turnover of other CRL components has remained a subject of debate that could not be rationalized until now. In this addendum, we discuss some additional observations made during the characterization of csn2-5 mutant plants that provide insight to CSN functions and implicate the CSN in novel developmental processes.

Different Phenotypes ofHypomorphic csn2 and csn5 Mutants: Evidence for CSN-Independent Functionsof CSN5

The archetypal CSN present in Arabidopsis is composed of six PCI and two MPN (MPR1, PAD1 N terminal) domain subunits. Among these subunits, CSN5 has some unique features: not only does it harbor the metallopeptidase activity needed for cullin deneddylation¹⁶ but it is also the only subunit that is fully stable and detectable as a monomer or in smaller subcomplexes in wild type extracts (reviewed in ref. 17). Also, CSN5 was the only CSN subunit initially identified by sequence comparisons in budding yeast which possesses a more divergent CSN-like complex.^18,19 These points suggest that CSN5 might fulfill additional CSN-independent functions inside the cell as supported by studies performed in animal cell culture systems. For example, ectopic expression of HA-Jab1 (CSN5) led to downregulation of the cell cycle regulator p27, although it did not change cullin neddylation and was not detectably incorporated into CSN, but rather present as a monomer and part of a smaller subcomplex.²⁰ In plants, however, no indications for CSN-independent roles of CSN5 have been provided so far and no differences between csn3, csn4 and csn5 mutants could be detected by transcriptional profiling.²¹ The csn2-5 mutant and the csn5a-1 mutant (which carries a mutation in one of two Arabidopsis CSN5 loci) both accumulate reduced amounts of CSN, but while the CSN5 monomer is unaffected by csn2-5, it is barely detectable in csn5a mutant lines.^1,5,17 We compared these two mutants in different biochemical and physiological assays. Very similar results were obtained for defects in cullin deneddylation assessed by western blotting, and physiological assays for presumed CRL-mediated responses to the phytohormones auxin, jasmonate and ethylene. While csn5a-1 mutants displayed a stronger constitutive photomorphogenesis (cop) phenotype and were strongly dwarfed, csn2-5 plants were virtually indistinguishable from wild type (see ref. 1). Thus, physiological defects of csn mutants may not correlate strictly with defects in cullin deneddylation. Also, relatively strong changes in cullin neddylation patterns can be tolerated without major phenotypic consequences. As the presence of the CSN5 monomer seems to be a major difference between the two mutants, CSN-independent functions of this protein, which are retained in csn2-5 mutant but lost in csn5a-1, might account for the above phenotypic differences. Monomeric CSN5 could have important functions in general development but would act independently of CSN's deneddylation activity and CRLs. However, while this explanation for the stronger phenotype of the csn5a-1 mutant is plausible, it should be noted that different results have been reported for Schizosaccharomyces pombe: csn1 and csn2 mutants exhibit more severe phenotypes than csn5 mutant strains.²² These differences highlight the importance of studying multiple model organisms to reach a more fundamental understanding of CSN functions.

Hypomorphic csn Mutants Exhibit Phenotypes Suggesting Defectsin Cytoskeleton Assembly

Auxin is a key player controlling lateral root formation (reviewed in ref. 23). Consistent with CSN's role as an indirect regulator of auxin signaling, defects in lateral root formation were previously described for CSN5 hypomorphic mutants.^5,6 When we performed similar analyses with csn2-5 and csn5a-1, we detected differences in lateral root numbers, but these appeared to be mainly due to delayed and reduced growth of mutant primary roots, since no significant differences in lateral root density were found (data not shown). However, under these experimental conditions, we detected a skewed root growth phenotype for the csn5a-1 mutant: the average skewing angle of Col roots was approximately 3° compared to 31° of the csn5a-1 mutant (Fig. 1A; n = 14; p < 0.001). A similar phenotype was not observed for the Ler csn2-5 mutant, possibly due to more pronounced initial root skewing of wild type Ler plants (33.5° ± 7°). The basis for root skewing is currently not well understood, but auxin transport, gravity and root-gel contacts are considered to be of major importance (reviewed in ref. 24). Interestingly, exacerbated root skewing can be induced by cytoskeleton-affecting drugs, and several root skewing mutants are affected in genes regulating cytoskeleton stability. In parallel, we observed a ‘curly hypocotyl’ phenotype reminiscent of defects in cytoskeleton formation when seedlings were grown in darkness (Fig. 1B). Under these conditions, control plants grew straight, but hypocotyls of csn2-5 and csn2-5 sgt1b-3 double mutant seedlings grew in a helix-like shape performing typically 1–2 right-handed turns. In this case, a similar phenotype could not be detected for csn5a-1, but hypocotyls of csn5a-1 seedlings were extremely short because of its dwarfism and strong constitutive photo-morphogenesis (data not shown), which might mask such curly phenotype.

Figure 1.

Hypomorphic csn mutants exhibit phenotypes regularly associated with defects in microtubule assembly. (A) Plants were grown on inclined 0.1x MS plates for eight days and photographed from above. The skewing angle was measured from the root tip to a vertical vector as indicated in the pictures. (B) Plants were grown on horizontal 0.1x MS plates without sucrose for six days in darkness and transferred on a new plate for documentation. Scale bar = 0.5 cm.

Both the observed root skewing and ‘curly hypocotyls’ could result indirectly from defects in responses to hormones such as auxin and ethylene (reviewed in ref. 24). But although csn mutations confer auxin resistance, we did not detect altered gravitropism response or altered ethylene response (data not shown) in csn5a-1 and csn2-5. Besides, tir1 mutants did not present any agravitropic or modified skewing phenotypes. Therefore, it would be interesting to analyze cytoskeleton assembly and organization in hypomorphic csn mutants further and assess, for example, orientation of cellulose fibrils along the cell wall, which follows the progression of cortical microtubules (reviewed in ref. 25). Such observations would be of particular interest since CRLs and the CSN-dependent Nedd8 deconjugation pathways have been directly implicated in cytoskel eton formation in C. elegans.^26,27

Addendum to: Stuttmann J, Lechner E, Guérois R, Parker JE, Nussaume L, Genschik P, et al. COP9 signalosome- and 26S proteasome dependent regulation of SCFTIR1 accumulation in Arabidopsis. J Biol Chem. 2009;284:7920–7930. doi: 10.1074/jbc.M809069200.