Interactions of the COP9 Signalosome with the E3 Ubiquitin Ligase SCFTIR1 in Mediating Auxin Response

See allHide authors and affiliations

Science  18 May 2001:
Vol. 292, Issue 5520, pp. 1379-1382
DOI: 10.1126/science.1059776


The COP9 signalosome is an evolutionary conserved multiprotein complex of unknown function that acts as a negative regulator of photomorphogenic seedling development in Arabidopsis. Here, we show that plants with reduced COP9 signalosome levels had decreased auxin response similar to loss-of-function mutants of the E3 ubiquitin ligase SCFTIR1. Furthermore, we found that the COP9 signalosome and SCFTIR1 interacted in vivo and that the COP9 signalosome was required for efficient degradation of PSIAA6, a candidate substrate of SCFTIR1. Thus, the COP9 signalosome may play an important role in mediating E3 ubiquitin ligase–mediated responses.

The COP9 signalosome is a multiprotein complex that was discovered during the characterization of the photomorphogenic cop/det/fusmutants from Arabidopsis (1, 2). The COP9 signalosome is required for the proteasome-mediated degradation of HY5, a positive regulator of photomorphogenesis (3), and each of its eight subunits is related to one of the eight subunits that form the “lid” subcomplex of the 26S proteasome (2, 4). Thus, it may be that the COP9 signalosome is involved in protein degradation via the ubiquitin-proteasome pathway (2, 5).

We generated transgenic Arabidopsis plants with reduced levels of the COP9 signalosome subunit CSN5 (JAB1 or AJH) using the antisense and cosuppression strategy (Fig. 1A) (6–8). In thecop/det/fus mutants that affect individual subunits of the COP9 signalosome, loss of one subunit results in loss of the entire protein complex (2), and similarly, the reduction of CSN5 in the CSN5 transgenic plants leads to a reduction of COP9 signalosome levels (Fig. 1A). At the physiological level, this is reflected by the photomorphogenic phenotype of the dark-grown CSN5 transgenic seedlings (Fig. 1, A and B) (6).

Figure 1

CSN5 transgenic plants had a photomorphogenic phenotype in the dark and reduced apical dominance. (A) Immunoblot with protein extracts from a wild-type and a CSN5 transgenic (CSN5 tg) seedling using CSN5 and CSN1 antibodies (6, 27, 28). Seedlings from CSN5 antisense and cosuppression lines with similar phenotypes had similar immunoblot profiles; the result from a representative CSN5 antisense line (line J1L1 #20) is shown here. The reduction of CSN1 expression is an example for a reduction of COP9 signalosome levels (2, 28). The TATA-box binding protein (TBP) blot served as a loading control. (B) Seven-day-old dark-grown wild-type (left) and CSN5 tg (antisense J1L1 #20, right) seedlings. (C) Four-week-old wild-type plant. (D) Four-week-old CSN5 tg plant (antisense line J1L1 #20). (E) Eight-week-old CSN5 tg plant (cosuppression line X1#7). All plants are Columbia ecotype. Bar in (C) and (E) is 5 cm and in (D) is 2 cm. Arrows in (D) and (E) indicate secondary inflorescences. (F) Leaves from a 4-week-old wild-type (left) and a CSN5 tg plant (cosuppression line X1#7, right). (G and H) Toluene blue–stained (0.01%) longitudinal section of stems from wild-type (G) and CSN5 tg (cosuppression line X1#7) (H) plants. Sections were taken from 3-week-old plants at the midpoint of their primary inflorescence. Both images were taken at the same magnification.

On the basis of the pleiotropic phenotype of thecop/det/fus mutants, it has been suggested that in addition to photomorphogenesis, many other developmental processes may be affected in these mutants (9). Because the lethal phenotype of mutants that affect subunits of the COP9 signalosome has so far prevented insightful developmental studies, we used the weaker nonlethal phenotype of the CSN5 transgenic plants to study COP9 signalosome-mediated processes in plant development. The most striking phenotype of adult CSN5 transgenic plants was a strong increase in the number of secondary inflorescences (8 to 12 compared to 3 to 4 in the wild type) accompanied by a general reduction in plant size and internode length (Fig. 1, C through E). The outgrowth of secondary inflorescences in wild-type plants is inhibited by the phytohormone auxin, which is produced in the shoot apex of the primary inflorescence, in a physiological process known as apical dominance (10). Because the CSN5 transgenic plants had lost apical dominance, these plants may have become insensitive to the inhibitory auxin signal. The CSN5 transgenic plants also had strongly reduced cell size and wavy leaf morphology, phenotypes that are observed in a number of auxin-response mutants, notably the group ofaxr mutants (Fig. 1, F through H) (11–14).

We examined auxin-related phenotypes in the CSN5 transgenic plants in more detail. Whereas root growth in wild-type seedlings is inhibited by exogenous auxin application, CSN5 transgenic seedlings were more resistant to auxin (Fig. 2, A and B). Furthermore, CSN5 transgenic seedlings had fewer lateral roots, reduced root hair elongation, and reduced gravitropism response when compared to the wild type (Fig. 2, C through E). These physiological responses are known to be controlled by auxin and are affected in a qualitatively similar manner in the auxin-response mutants axr1-3 andtir1-1 (Fig. 2, A through E) (11, 15).

Figure 2

Auxin-response phenotypes of CSN5 transgenic seedlings. (A) Nine-day-old wild-type (left) and CSN5 transgenic (right) seedlings. The asterisk indicates the position of the root tip at the time of transfer to auxin-containing media. (B) Relative root growth of seedlings on 2,4-dichlorophenoxyacetic acid (2,4-D)–containing medium compared to root growth on unsupplemented medium (100%). Wild type, black bars; CSN5 tg (antisense line J1L1 #20), dark-gray bars; tir1-1, light gray bars; axr1-3, white bars (29). (C) Number of lateral roots and (D) images of roots from 12-day-old seedlings. All images were taken at the same magnification. wt, wild-type; csn5, CSN5 transgenic (antisense line J1L1 #20). (E) Root tip reorientation in degrees after a change in the gravitropic vector by 90°. ○, wild-type; •, CSN5 tg (antisense line J1L1 #20); □,tir1-1; ▪, axr1-3. Experimental procedures from (A) to (E) are as previously described (11). (F) Northern hybridization of 25 μg total RNA prepared from wild-type and CSN5 tg plants with AUX/IAA probes as indicated (30, 31). (G) The PSIAA6LUC degradation rate is reduced in CSN5 transgenic plants (31). ○, PSIAA6LUC in wild-type background; •, PSIAA6LUC in CSN5 transgenic (antisense line J1L1 #20) background; □, LUC in wild-type background; ▪, LUC in CSN5 transgenic (antisense line J1L1 #20) background.

The AUX/IAA genes form a gene family that encodes short-lived regulatory proteins (16, 17). Auxin triggers the rapid and specific transcription of most members of theAUX/IAA gene family (17). The induction of these genes is reduced in several auxin-response mutants and was also compromised in the CSN5 transgenic plants (Fig. 2F) (17,18). Thus, it is likely that similar molecular mechanisms form the basis of the auxin-reponse phenotypes observed in the CNS5 transgenic plants and other auxin mutants.

Proper auxin response in Arabidopsis is mediated by the E3 ubiquitin ligase SCFTIR1 (15). Molecular and genetic evidence suggests that AUX/IAA proteins are SCFTIR1 substrates (19). The rapid turnover of AUX/IAA proteins by SCFTIR1 appears to be an integral feature of auxin response (16), and their increased half-life in the SCFTIR1 loss-of-function mutants is the basis of their auxin phenotype (19, 20). To test whether the COP9 signalosome was involved in regulating the turnover of AUX/IAA proteins, we used a transgenic line that expresses the luciferase reporter in frame with PSIAA6, anAUX/IAA from pea (20). Using luciferase activity as a measure of PSIAA6 abundance, we found that the PSIAA6LUC degradation rate was reduced in protein extracts prepared from CSN5 transgenic lines compared to those from the wild type (Fig. 2G). At the same time, the stability of a native luciferase protein was not affected, indicating that it was the PSIAA6 moiety that specifically promoted the degradation of PSIAA6LUC. Because protein degradation appears to be a common feature to most AUX/IAA proteins (16), it is likely that the COP9 signalosome may also be required for their degradation and that increased AUX/IAA levels cause the auxin-response phenotype of the CSN5 transgenic plants.

The fact that the COP9 signalosome and SCFTIR1 were required for AUX/IAA protein degradation, and recent evidence that the cullin subunit of SCF-type E3 ubiquitin ligases from other eukaryotes can interact with the COP9 signalosome (21), prompted us to investigate a physical interaction between the COP9 signalosome and SCFTIR1. Indeed, we observed copurification of the SCFTIR1 subunit AtCUL1 in immunoaffinity-purified fractions of the COP9 signalosome (Fig. 3A). In a reciprocal experiment, an AtCUL1 antibody immunoprecipitated the COP9 signalosome and the SCFTIR1 subunit ASK1 (Fig. 3B). Finally, immunoprecipitation of TIR1, the F-box domain subunit that confers substrate-specificity to SCFTIR1 (15), yielded the entire COP9 signalosome in addition to the SCFTIR1 components AtCUL1 and ASK1 (Fig. 3C), suggesting that the COP9 signalosome and SCFTIR1 interact in vivo. In the yeast two-hybrid system, we detected direct interactions between AtCUL1 and CSN2, as well as between the SCFTIR1 subunit AtRBX1 and the COP9 signalosome subunits CSN1 and CSN6 (Fig. 3D).

Figure 3

COP9 signalosome and SCFTIR1interact in vivo and COP9 signalosome mutants accumulate preferentially RUB1-modified AtCUL1 (31). (A) Silver-stained gel of an immunoaffinity-purified COP9 signalosome from cauliflower (CSN subunits are indicated, left lane) (28). Immunoblots of the purified COP9 signalosome were probed with CSN1 and AtCUL1 antibodies (15, 27). Asterisks show the position of CSN1 and AtCUL1, respectively, • indicates the position of the RUB1-conjugated AtCUL1. (B) The AtCUL1 antibody coimmunoprecipitates SCFTIR1 components and the COP9 signalosome.Arabidopsis seedling extract (total) was used for immunoprecipitation with the AtCUL1 antibody (AtCUL1 IP) and the corresponding preimmune serum (control IP) followed by an immunoblot analysis. Four COP9 signalosome subunits that coimmunoprecipitated with AtCUL1 are shown. (C) MYC-tagged TIR1 protein immunoprecipitates SCFTIR1 subunits and the COP9 signalosome (15). (D) ArabidopsisCSN1, CSN2, and CSN6 interact with AtCUL1 and AtRBX1 in the yeast two-hybrid interaction assay (28). (E) COP9 signalosome loss-of-function mutants accumulate preferentially RUB1-modified AtCUL1. Immunoblot with seedling protein extracts: wild-type, wt; CSN5 transgenic seedlings (antisense line J1L1 #20), csn5 tg; COP9 signalosome null-mutant, cop9-1; COP1 mutant,cop1-6 (31). All techniques are as described in (28).

AXR1 is a component of an enzyme cascade that conjugates the ubiquitin-related protein RUB1 to the AtCUL1 subunit of SCFTIR1 (22). A recent observation suggests that the COP9 signalosome promotes RUB1 deconjugation (21). Indeed, we found that Arabidopsis COP9 signalosome mutants preferentially accumulated RUB1-conjugated AtCUL1, whereas wild-type extracts contained unmodified and RUB1-modified AtCUL1 (Fig. 3E). Thus, the essentially antagonistic steps of AXR1-mediated RUB1 conjugation and its subsequent COP9 signalosome–promoted deconjugation are both required for proper auxin response and act together toward the degradation of SCFTIR1 substrates (Fig. 2). The requirement of AXR1 and the COP9 signalosome for proper auxin response was further confirmed in a genetic interaction study. When we introduced a CSN5 cosuppressing transgene (csn5 tg) into a homozygous axr1-3mutant background, the auxin-response phenotypes of theaxr1-3/csn5 tg seedlings were enhanced compared to the parents (Fig. 4A). Furthermore,axr1-3/csn5 tg plants had reduced fertility (Fig. 4B), a phenotype that was observed in neither of the parent lines but in a strong mutant allele of AXR1 (11), suggesting a synergistic genetic interaction between axr1-3 and the CSN5 transgene cosuppression. Interestingly, studies centered around the ubiquitin-like modification SMT3 have revealed a parallel observation that yeast mutants deficient in SMT3 conjugation have similar phenotypes to mutants with defects in SMT3 deconjugation (23).

Figure 4

Genetic interaction between AXR1 and the COP9 signalosome (31). A weak CSN5 transgenic line (cosuppression line X1#7, no seedling phenotype but apical dominance phenotype) was crossed to the axr1-3 mutant. (A) Root growth inhibition was reduced in the axr1-3/csn5 tg plants compared to the parent lines. Wild-type, black bars; csn5 tg, dark gray bars; axr1-3, light gray bars;axr1-3/csn5 tg, white bars. (B) Theaxr1-3/csn5 tg plants have reduced fertility. Mature inflorescences of 7-week-old wild-type (Col), axr1-3, csn5 tg, and axr1-3/csn5 tg plants are shown.

The conjugation of RUB1 to cullins promotes ubiquitin-chain formation (24), and it could be that RUB1 conjugation and deconjugation cycles are important for this process. Next to a possible biochemical role, the RUB1 modification could also be essential for the physical interaction between the COP9 signalosome and SCFTIR1 and thereby regulate the nucleo-cytoplasmic distribution of SCFTIR1. This hypothesis is based on the finding that the COP9 signalosome is required for the light-dependent nucleo-cytoplasmic distribution of the putative E3 ubiquitin ligase COP1 (25).

We demonstrate a specific interaction between the COP9 signalosome and SCFTIR1, and show that the COP9 signalosome is required for protein degradation in the context of auxin response. However, in analogy to other eukaryotes, it can be assumed that AtCUL1 and AtRBX1, the SCFTIR1 subunits that directly interact with the COP9 signalosome, are core components of multiple SCF-type E3 ubiquitin ligases that differ in their F-box domain subunit (26). Thus, the Arabidopsis COP9 signalosome may also interact with other yet-to-be-identified SCF-type E3 ubiquitin ligases. Indeed, we observed several phenotypes in the CSN5 transgenic plants, such as the loss of apical dominance and the change in leaf morphology, that cannot be solely explained by a loss of SCFTIR1 function (Fig. 1, C through E). Moreover, non–SCF-type E3 ubiquitin ligases like the aforementioned putative E3 ubiquitin ligase COP1 may also interact with the COP9 signalosome (3). Thus, the function of many different E3 ubiquitin ligases could be impaired incop/det/fus mutants lacking the COP9 signalosome, and the combination of the resulting defects could account for the severe and pleiotropic phenotype of these mutants.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: xingwang.deng{at}


View Abstract

Navigate This Article