| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
American Society of Plant Biologists Ubiquitination and Auxin SignalingA Degrading StoryDepartment of Biology, University of York, Box 373, York, United Kingdom 1 To whom correspondence should be addressed. E-mail hmol1{at}york.ac.uk; fax 44-1904-434312
Signaling pathways are rarely straightforward, and auxin signal transduction is no exception. The diverse events in which auxin is involved tell of the daunting complexity behind the plant's response to auxin. This one molecule can cause changes in rates of cell division and cell elongation, alter ion fluxes across membranes, cue changes in patterning and differentiation, and affect the expression of hundreds of genes (Davies, 1995  ; Berleth and Sachs, 2001). Does auxin act through multiple pathways to these diverse ends, or is there a single pathway that is dependent on the spatial, temporal, and environmental context in which auxin is received? Or is it both? Although the answers to these and other broad questions of auxin biology remain unclear, we are now in a position to connect at least some areas of research on auxin signal transduction into a more cohesive whole. Recent advances in our understanding of the apparent hub of auxin signaling, between perception and downstream auxin-induced gene expression, has developed from the mutational and molecular analysis of three main groups of proteins: the auxin/indoleacetic acids (Aux/IAAs), the auxin response factors, and components of the ubiquitin-mediated proteolytic pathway. It seems that regulated protein degradation is central to most aspects of the auxin response.
Ubiquitin-mediated proteolysis has emerged as being as fundamentally important as phosphorylation in terms of its involvement in diverse cellular events. Target proteins are condemned to degradation in the 26S proteasome by the addition of a polyubiquitin chain in what is essentially a three-step process (Figure 1) (Voges et al., 1999 ; Jackson et al., 2000; reviewed by Pickart, 2001). First, ubiquitin is activated by the ATP-dependent formation of a thiolester bond with a conserved Cys residue of a ubiquitin-activating enzyme, E1. Activated ubiquitin then is passed to one of a family of E2 ubiquitin-conjugating enzymes, from which it is conjugated to a Lys residue of the target protein, often with the assistance of an E3 enzyme, ubiquitin protein ligase. This process is reiterated such that one or two additional ubiquitins are polymerized upon the first ubiquitin molecule. Target proteins marked with less than four ubiquitins are poor substrates of the 26S proteasome and often escape degradation. In some cases, an additional step ensures efficient proteolysis by extending the ubiquitin chain, a process that requires a multiubiquitin chain assembly factor, or E4 (Koegl et al., 1999; Azevedo et al., 2001).
Of course, ubiquitination must be directed against specific targets, and it is the E3 enzyme that mediates the critical step of substrate recognition, a fact reflected in the hundreds of E3s of various classes that are encoded by the eukaryotic genomes sequenced to date (Deshaies, 1999 ; Voges et al., 1999; Xiao and Jang, 2000; Pickart, 2001). Prominent among these, and certainly most relevant here, are the SCF ubiquitin protein ligases (for Skp1p, Cdc53p/cullin, and F-box protein in yeast and mammals) (reviewed by Deshaies, 1999; Xiao and Jang, 2000). SCF complexes are multimeric enzymes of at least four subunits built around a member of the cullin family of proteins (Mathias et al., 1996). In conjunction with a second subunit, RBX1 (also known as ROC1 or HRT1), the cullin catalyzes polyubiquitin chain formation by interacting with the E2 enzyme (Seol et al., 1999). The cullin also binds to a member of the SKP1 protein family, which, in turn, binds to an F-box protein, so called because it contains an N-terminal F-box motif required for its interaction with SKP1 (Skowyra et al., 1997; Kishi and Yamao, 1998; Patton et al., 1998). The C terminus of the F-box protein consists of any one of a number of protein–protein interaction domains that bind the relevant target protein, making it accessible to the E2-SCF complex (Kishi and Yamao, 1998; Patton et al., 1998).
Our knowledge of the SCF system is derived mainly from extensive work in yeast and mammals (Patton et al., 1998
The crucial evidence for the involvement of SCF-mediated protein degradation in auxin signaling came through the characterization of the tir1 mutants of Arabidopsis (Table 1) (Ruegger et al., 1998 ). Although the tir1 alleles were identified in a screen for resistance to inhibitors of auxin transport, the mutants were more markedly resistant to inhibitory concentrations of auxin, suggesting a defect in auxin response rather than transport (Ruegger et al., 1998). TIR1 was found to encode an F-box protein (Ruegger et al., 1998) and subsequently was shown to interact with either of the Arabidopsis Skp1-like proteins, ASK1 or ASK2, and the cullin AtCUL1 to form the functional SCFTIR1 (Gray et al., 1999, 2001). The idea that regulated protein degradation via SCFTIR1 is required for proper auxin signaling was strengthened by the analysis of mutation in ASK1 (Table 1). ask1-1 shows reduced auxin response, being slightly more auxin resistant than tir1-1 (Gray et al., 1999). This may reflect the fact that TIR1 has several close homologs in Arabidopsis, indicating the potential for redundancy in TIR1 function (Gray et al., 1999). The C-terminal domain of TIR1 and its homologs consists of Leu-rich repeats, which presumably are involved in target selection (Ruegger et al., 1998; Hsiung et al., 2001). What, then, are the targets for SCFTIR1? In the search for candidates, the Aux/IAA family of transcription factors are likely suspects.
The Aux/IAAs are a family of extremely short-lived nuclear proteins (reviewed by Abel and Theologis, 1996 ). Aux/IAA genes are induced rapidly by auxin and are found throughout the higher plants. Their importance in auxin signal transduction is manifest in the severe auxin-related phenotypes arising from semidominant mutations in several of these genes, including AXR2/IAA7, AXR3/IAA17, SHY2/IAA3, BODENLOS/IAA12, SLR1/IAA14, MSG2/IAA19, and IAA28 (Table 1) (Rouse et al., 1998; Tian and Reed, 1999; Nagpal et al., 2000; Rogg et al., 2001; reviewed by Reed, 2001). Aux/IAA proteins share a typical four-domain structure (Figure 2)
, and what is striking about these semidominant mutations is that they occur only within the highly conserved domain II, centered on a core GWPPV motif. The spectacular auxin-related phenotypic effects of these single–amino acid substitutions illustrate the critical importance of domain II to Aux/IAA function (Rouse et al., 1998; Tian and Reed, 1999; Nagpal et al., 2000; Rogg et al., 2001). Domain II is the site of the destabilization signal that confers instability on Aux/IAA members; measured half-lives range from as little as 6 min to >80 min (Abel et al., 1994; Ouellet et al., 2001).
Translational fusions of Aux/IAA N-terminal regions containing domain II with reporter proteins such as  -glucuronidase (GUS) or luciferase have shown that this lability can be conferred on otherwise stable proteins (Worley et al., 2000; Gray et al., 2001). The transferable destabilization signal within Aux/IAAs has been finely mapped to a 13–amino acid region (or so-called degron) around the core GWPPV residues in domain II. Although this 13–amino acid degron fused to luciferase excludes an essential bipartite nuclear localization signal spanning domain II, the use of a heterologous nuclear localization signal has shown that in the correct subcellular context, these 13 amino acids are sufficient to confer instability (Ramos et al., 2001). Further direct and indirect measurements of Aux/IAA stability have shown that the phenotypic consequences of mutations in domain II may be the result of stabilization of the respective IAA proteins without affecting their nuclear localization (Worley et al., 2000; Gray et al., 2001; Ouellet et al., 2001). For example, the axr3-1 mutation, a Pro-to-Leu substitution in domain II, causes an almost sevenfold increase in protein half-life (Ouellet et al., 2001). Among many auxin-related phenotypes (Table 1), this mutation causes increased apical dominance, adventitious rooting, and complete agravitropism in roots (Leyser et al., 1996). Collectively, these data stress the importance of Aux/IAAs and their instability to auxin response. So, to reverse an earlier question, what are the regulators of Aux/IAA stability?
Evidence of the involvement of SCFTIR1 in Aux/IAA degradation came from the analysis of AXR3/IAA17 reporter fusion proteins in the tir1 mutant background (Gray et al., 2001 ). This fusion protein consists of N-terminal domains I and II of AXR3, including the bipartite nuclear localization signal, fused translationally to GUS and expressed under the control of a heat-inducible promoter in transgenic Arabidopsis. After heat induction in a wild-type background, the fusion protein accumulates to relatively low levels and is depleted rapidly over time compared with controls. However, the stability of the same fusion protein was increased significantly when the construct was crossed into the tir1 background, indicating that TIR1 was required for the destabilization of AXR3 in planta (Gray et al., 2001). That TIR1 activity ultimately might result in AXR3 degradation in the 26S proteasome was supported by the finding that specific proteasome inhibitors stabilize AXR2-, AXR3-, and other Aux/IAA-reporter fusion proteins (Gray et al., 2001; Ramos et al., 2001).
The final step in defining a mechanistic link between Aux/IAA proteins and the SCFTIR1 ubiquitin ligase complex came with the direct demonstration of the interaction of AXR2 and AXR3 with SCFTIR1 in plant extracts. The use of glutathione S-transferase (GST)–tagged AXR2 or AXR3 in pulldown assays with extracts from plants expressing a c-myc epitope-tagged TIR1 yielded the entire SCFTIR1 complex, as identified by immunoblotting with c-myc, AtCUL1, and ASK2 antibodies (Gray et al., 2001
Genetic data have established the importance of regulating the stability of Aux/IAA proteins, suggesting that the plant might respond to auxin by regulating the turnover of Aux/IAAs. Although previous analysis had suggested that this was not the case (Abel et al., 1994 ), recent work examining the stability of Aux/IAA-reporter fusion proteins expressed from non-auxin-responsive promoters and using different time frames has reached different conclusions. These studies were able to uncouple any auxin-induced Aux/IAA protein destabilization from auxin-induced Aux/IAA gene expression to reveal a significant increase in the degradation of the Aux/IAA-reporter proteins in response to exogenously applied auxin (Gray et al., 2001; Zenser et al., 2001). This effect of auxin is dose dependent and extremely rapid, causing a considerable decrease in protein levels within 5 min (Gray et al., 2001; Zenser et al., 2001). In fact, it is possible that auxin might be absolutely required for Aux/IAA degradation, and if cells could be rid of basal levels of auxin, Aux/IAA proteins might be rendered extremely stable.
Perfectly consistent with the in planta measures of Aux/IAA stability, the in vitro pulldown assay with GST-AXR2 and GST-AXR3 showed an auxin-induced increase in the recovery of SCFTIR1, which also occurred in a dose-dependent manner (Gray et al., 2001
The mechanism by which auxin influences SCFTIR1-mediated Aux/IAA turnover is not understood. However, clues to what might be happening may be gleaned from the many characterized examples of SCF-mediated signaling pathways in yeast and mammals. SCF-mediated ubiquitination usually is preceded by substrate modification. These changes turn stable proteins into susceptible targets of the SCF, and by far the most common modification is phosphorylation (Deshaies, 1999 ). A good example is that of NF- B activation in mammals. NF-B normally is sequestered in the cytoplasm, because its nuclear localization signal is masked by the IB family of inhibitors. In response to inflammation and other stress stimuli, IB is phosphorylated quickly, ubiquitinated by SCF E3RSIB/-TrCP, and ultimately degraded, freeing NF-B to translocate to the nucleus to regulate gene expression (Karin and Ben-Neriah, 2000). Similarly, auxin might increase the affinity of Aux/IAAs for SCFTIR1 by stimulating their phosphorylation.
There is certainly an increasing body of biochemical and genetic evidence supporting a role for kinases in auxin signaling. A recent interesting example is the identification of an auxin-stimulated mitogen-activated protein kinase (MAPK) in Arabidopsis roots (Mockaitis and Howell, 2000
Examples of non-phosphorylation-dependent regulation of substrate–SCF interactions are beginning to emerge. One such case concerns the degradation of the human transcription factor, hypoxia-inducible factor (HIF Although without precedent elsewhere, a final possibility is that the interaction between Aux/IAAs and SCFTIR1 increases in response to auxin because of modification (perhaps phosphorylation) of the TIR1 protein.
Another point at which the regulation of Aux/IAA ubiquitination could conceivably take place is at the level of SCF abundance, assembly, or activity. This suggestion comes from studies on SCFGrr1 of yeast, which is involved in cell cycle progression and Glc and amino acid signaling (reviewed by Johnston, 1999 ). The F-box protein Grr1 is structurally similar to TIR1, both containing C-terminal Leu-rich repeats. In yeast two-hybrid studies, it has been shown that the interaction between Grr1 and Skp1, part of the process of SCFGrr1 assembly, is enhanced significantly in response to Glc (Li and Johnston, 1997). Although this would seem to set a precedent for SCF assembly in response to environmental stimuli that might be extended to SCFTIR1, recent data have suggested other possible explanations. It has since been shown that F-box proteins are inherently unstable (Galan and Peter, 1999; Mathias et al., 1999). This instability has been shown to be the result of the autoubiquitination of the F-box protein while assembled in an SCF (Galan and Peter, 1999), and it seems that SCF substrate binding may confer protection on the F-box protein (Deshaies, 1999; Galan and Peter, 1999). Thus, the apparent increased interaction of Grr1 and Skp1 in response to Glc may reflect increased Grr1 stability as a result of a Glc-induced increase in the susceptibility of SCFGrr1 targets. Similarly, the stability and abundance of TIR1 might be important for normal auxin sensitivity, because overexpression of TIR1 in transgenic plants results in increased auxin response (Gray et al., 1999). These phenotypes are dependent on additional TIR1 participating in an SCF complex, because overexpression of a TIR1 variant in which the F-box had been mutated does not result in auxin response phenotypes (Gray et al., 1999). Furthermore, loss-of-function tir1 alleles appear to be semidominant, suggesting haploinsufficiency (Ruegger et al., 1998). Therefore, auxin responses could be brought about by changes in the abundance of TIR1. This explanation could account for some of the difference in auxin responsiveness observed between tissues, because TIR1 is not expressed uniformly throughout the plant (Gray et al., 1999).
Another important event affecting the activity, rather than the assembly, of SCFTIR1 has been shown to be crucial for normal auxin signaling. It involves the modification of the cullin subunit of the SCF by its conjugation with a ubiquitin-like protein called RUB1 (NEDD8 in humans and fission yeast) that is important for optimal SCF function (del Pozo et al., 1998
AXR1 encodes an enzymatic subunit that dimerizes with a protein called ECR1 to form a RUB1-activating enzyme that is structurally and functionally homologous with the E1 ubiquitin-activating enzyme (Leyser et al., 1993
Recent results suggest that the removal of RUB1 from SCFTIR1 also is important for auxin signaling (Schwechheimer et al., 2001
The COP9 signalosome was originally identified in screens for light signaling mutants in Arabidopsis (Chamovitz et al., 1996
Consistent with this idea, the stability of an Aux/IAA-luciferase fusion protein was increased in the CSN5 antisense background (Schwechheimer et al., 2001
So what is the function of RUB1/NEDD8 conjugation to cullin and the significance of its subsequent removal? It has been shown that NEDD8 modification of cullins promotes SCF ubiquitin ligase activity (Furukawa et al., 2000
Although these are formal possibilities, the effects of CSN deficiency in fission yeast suggest that a reduction in the ubiquitin ligase activity of SCFTIR1 may not explain the increased stability of the Aux/IAA-reporter fusion protein and the auxin-related phenotypes in COP9-deficient backgrounds. Similar to the accumulation of RUB1-AtCUL1 conjugates in COP9-deficient Arabidopsis, CSN-deficient yeast accumulate NEDD8-modified cullin, but the effect is to increase rather than decrease SCF ubiquitin ligase activity (Lyapina et al., 2001
Studies in CSN5-deficient fission yeast have suggested that if COP9 does have a role in the nucleocytoplasmic partitioning of ubiquitinated substrates, this is not accompanied by any large change in the subcellular distribution of cullins (Lyapina et al., 2001
The activity of the RUB1 conjugation pathway in Arabidopsis apparently is unaffected by the addition of auxin, indicating that it is not a direct route for auxin signal transduction (del Pozo et al., 2002
The defects in auxin signaling caused by mutations in Aux/IAAs and ubiquitin pathway components demonstrate the consequences of failing to regulate the stability of Aux/IAAs. But what is the function of the strict regulation of Aux/IAA abundance?
Aux/IAAs are nuclear proteins, and their activities as transcriptional regulators appear to be based on their ability to form a variety of dimers. First, individual Aux/IAAs are able to both homodimerize and heterodimerize with other members of the Aux/IAA family (Kim et al., 1997
Aux/IAAs can also heterodimerize with members of the auxin response factor (ARF) family of transcription factors (Kim et al., 1997 ; Ulmasov et al., 1997a, 1997b). ARFs share considerable homology in their C termini with the Aux/IAAs, possessing the conserved domains III and IV (Figure 2) (Ulmasov et al., 1997a). Interaction via these ARF domains allows dimerization with Aux/IAAs and also the formation of homodimers and heterodimers with other ARFs (Kim et al., 1997). Unlike Aux/IAAs, ARFs have been shown to bind DNA directly by virtue of a large B3-type DNA binding domain toward their N termini (Ulmasov et al., 1997a, 1999b). For the ARFs characterized to date, this binding is directed specifically to auxin response elements (AuxREs) containing TGTCNC motif(s) (typically TGTCTC) in various contexts that often involve coupling elements (Ulmasov et al., 1997a; Guilfoyle et al., 1998). Synthetic palindromic or direct repeats of these six nucleotides are sufficient to bind ARFs and confer auxin regulation on the transcription of a reporter gene (Ulmasov et al., 1997b, 1999a, 1999b). Hence, ARFs appear to mediate auxin-regulated gene expression through binding to AuxREs, and Aux/IAAs have the potential to alter the transcription of auxin-inducible genes by interacting with ARFs.
The number of interactions within and between Aux/IAAs and ARFs is potentially vast. Both the Aux/IAAs and ARFs constitute large families in Arabidopsis, with >25 Aux/IAA members and >20 ARFs (Reed, 2001
Our understanding of the interactions of ARFs and Aux/IAAs that regulate transcription from auxin-inducible promoters is based largely on work with carrot cell suspension cultures, in which TGTCTC-containing AuxRE promoter–reporter fusions were used to monitor the effects of the coexpression of various ARF- and Aux/IAA-derived genes (Ulmasov et al., 1997b
When the DNA binding domain was removed from the various ARFs and these truncated versions were introduced into carrot protoplasts with an AuxRE-reporter construct, truncated ARFs with Q-rich middle regions still were able to activate transcription (Ulmasov et al., 1999a
In contrast to the activating ARFs, ARF1 lost its ability to repress transcription when it was expressed without its DNA binding domain, suggesting that part of the repression is through the occupation of AuxREs (Ulmasov et al., 1999a
Dimerization through their C-terminal domains clearly is important to the activities of ARFs on auxin-inducible promoters. Because Aux/IAAs are able to dimerize with ARFs, their potential to influence ARF-mediated transcription has also been examined in the carrot protoplast system (Ulmasov et al., 1997b
Combined with the data on the effect of auxin on Aux/IAA abundance, these observations have led to a model of auxin-regulated transcription from TGTCTC-containing AuxREs (Figure 3) . At basal levels of auxin, Aux/IAAs are relatively stable and dimerize with ARF proteins on AuxREs, reducing transcription by preventing the recruitment of activating ARFs. Increasing auxin levels cause an increasing proportion of Aux/IAAs to be degraded, allowing the formation of a greater number of ARF-ARF dimers on AuxREs, and hence higher levels of transcription of auxin-responsive genes. Although this model lacks details, such as the level of multimerization of the interacting Aux/IAAs and ARFs, it seems a reasonable place to start: a point from which the unquestioned complexity can be built in.
How far does this model go toward explaining the existing genetic and molecular data on auxin-induced gene expression? Mutations in two of the Q-rich ARFs, MP/ARF5 and NPH4/ARF7 (Table 1), confer a range of auxin-related phenotypes, including reduced expression from TGTCTC-type AuxRE-containing auxin-inducible promoters (Hardtke and Berleth, 1998 ; Harper et al., 2000). This finding is consistent with the idea that Q-rich ARFs are required to activate the expression of auxin-inducible genes in vivo. There also is good evidence that transcription from TGTCTC AuxRE-regulated genes usually is kept inactive by very unstable auxin-inactivated repressor proteins, because transcription from AuxRE-containing promoters can be activated not only by auxin but also by inhibitors of protein synthesis such as cycloheximide (Koshiba et al., 1995). The possibility that these unstable transcriptional inhibitors are Aux/IAA proteins is supported by the observation that the dominant, stabilizing mutations in domain II of the Aux/IAA proteins often result in the constitutive repression of transcription from auxin-inducible AuxRE-containing promoters (Abel et al., 1995; Oono et al., 1998; Nagpal et al., 2000; Rogg et al., 2001).
Although these observations fit the general principle of the model, at first glance, other data do not. Aux/IAA genes were defined originally by their auxin inducibility, and most contain TGTCNC-type AuxREs in their promoters (reviewed by Abel and Theologis, 1996
One explanation might be that the model is based on transcriptional activation from TGTCTC-type AuxREs and does not include the contributions of other elements that affect the auxin induction of some Aux/IAAs (Ballas et al., 1995 The Aux/IAAs are a large and diverse family; therefore, a second possibility is that in addition to the negative effect of some Aux/IAAs, others might indirectly have a positive effect on ARF-mediated gene expression. In the period after an auxin pulse, the balance of dimerization events could be biased toward allowing activating ARFs to interact on AuxREs. For example, newly synthesized individual Aux/IAAs that normally would block activating ARF dimers might themselves be sequestered and titrated out by other Aux/IAAs, which usually are at very low abundance when auxin levels are low. Similarly, nonactivating and repressing ARFs might be prevented from interfering with activating ARF dimerization at the AuxRE. This possibility would require significant temporal variation in the relative abundance of different Aux/IAAs.
Such variation is indeed reflected in the data. There are >25 different Aux/IAAs with varying temporal and dose responses (Abel et al., 1995
Coupled with the possibility of different binding affinities among the interacting components, the functional significance of these differences may be to swing the balance of dimerization events between ARFs and Aux/IAAs initially toward, and then eventually away from, promoting the interaction of activating ARFs at the AuxRE. The idea that different Aux/IAAs do different things at different times in response to auxin certainly is supported by the genetic data. The similar stabilizing mutations of Aux/IAAs confer quite different and even opposite phenotypes. For example, the axr3-1 mutant has increased adventitious rooting, whereas the axr2-1 mutant has fewer adventitious roots than the wild type (Table 1) (Leyser et al., 1996
It is possible that in the period after auxin stimulation, the activity of newly synthesized Aux/IAAs is affected by post-translational modifications that contribute to a prolonged response in the face of accumulating Aux/IAA protein. For example, phosphorylation of Aux/IAAs might alter their dimerization preferences to effect a bias toward the activation of AuxREs or even allow the direct stabilization and enhancement of activating ARF binding at the AuxRE. Indeed, if the phosphorylating activities of the auxin-induced MAPK, phytochrome A, or PINOID kinase act in any way through Aux/IAAs, it might be in modulating their activity in this manner. In these scenarios, there are parallels with the previously described activation and repression of the mammalian transcription factor NF- |
TOP
INTRODUCTION
THE UBIQUITIN PATHWAY
11 min) than that of AXR3/IAA17 (