AOBPreview originally published online on January 12, 2007
Annals of Botany 2007 99(5):787-822; doi:10.1093/aob/mcl255
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
INVITED REVIEW |
Ubiquitin, Hormones and Biotic Stress in Plants
Section of Molecular and Cellular Biology, Plant Biology Graduate Group Program, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA
* For correspondence. E-mail jcallis{at}ucdavis.edu
Received: 1 August 2006 Returned for revision: 7 September 2006 Accepted: 3 October 2006 Published electronically: 12 January 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONOVERVIEW OF UPS COMPONENTSAPPENDIX 1APPENDIX 2ACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Background: The covalent attachment of ubiquitin to a substrate protein changes its fate. Notably, proteins typically tagged with a lysine48-linked polyubiquitin chain become substrates for degradation by the 26S proteasome. In recent years many experiments have been performed to characterize the proteins involved in the ubiquitylation process and to identify their substrates, in order to understand better the mechanisms that link specific protein degradation events to regulation of plant growth and development.
Scope: This review focuses on the role that ubiquitin plays in hormone synthesis, hormonal signalling cascades and plant defence mechanisms. Several examples are given of how targeted degradation of proteins affects downstream transcriptional regulation of hormone-responsive genes in the auxin, gibberellin, abscisic acid, ethylene and jasmonate signalling pathways. Additional experiments suggest that ubiquitin-mediated proteolysis may also act upstream of the hormonal signalling cascades by regulating hormone biosynthesis, transport and perception. Moreover, several experiments demonstrate that hormonal cross-talk can occur at the level of proteolysis. The more recently established role of the ubiquitin/proteasome system (UPS) in defence against biotic threats is also reviewed.
Conclusions: The UPS has been implicated in the regulation of almost every developmental process in plants, from embryogenesis to floral organ production probably through its central role in many hormone pathways. More recent evidence provides molecular mechanisms for hormonal cross-talk and links the UPS system to biotic defence responses.
Key words: Ubiquitin, E3 ligase, RING, U-Box, SCF, CRL, ubiquitylation, regulated proteolysis, plant defence, hormonal signalling, biotic stress, pathogen response
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONOVERVIEW OF UPS COMPONENTSAPPENDIX 1APPENDIX 2ACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Plants, as sessile organisms, rely on proteomic plasticity to remodel themselves during periods of developmental change, to endure varying environmental conditions, and to respond to biotic and abiotic stresses. Regulated ubiquitin- and proteasome-mediated degradation, therefore, plays a crucial role in enabling plants to alter their proteome to maximize their chances of survival under many different circumstances.
In plants, as in all eukaryotes, the ubiquitin/proteasome system (UPS) targets proteins for degradation (note that all abbreviations mentioned in this review are listed and defined in Appendix 1). In this system, the 76-amino-acid protein ubiquitin acts as a covalent molecular tag and its attachment requires three distinct enzymatic activities. Catalysed by E1 [ubiquitin-activating enzyme (UBA)], the ubiquitin C-terminal carboxyl group is first activated by adenylation, and then forms a thioester bond with a cysteinyl sulfhydryl residue on the E1 protein itself. Ubiquitin is then transferred to a cysteinyl sulfhydryl on another protein, the E2 [ubiquitin-conjugating protein (UBC)], preserving the thioester bond. Finally, ubiquitin is typically transferred to the substrate with the help of an E3 (ubiquitin ligase, Fig. 1; reviewed in Pickart and Eddins, 2004).
|
The first ubiquitin normally forms an isopeptide linkage with a lysyl
-amino group on the substrate, although a small subset of substrates are modified at the N-terminal amino group (reviewed in Ciechanover and Ben-Saadon, 2004), and at least one substrate has been reported to bind ubiquitin on a cysteinyl sulfhydryl group (Cadwell and Coscoy, 2005). If additional ubiquitins are added, they may be attached to different substrate lysyl -amino groups or to one of seven lysyl -amino groups on the bound ubiquityl moiety. All seven lysyl residues can be used to form ubiquitin–ubiquitin linkages in budding yeast (Saccharomyces cerevisiae; Peng et al., 2003). Previous work on model substrates established that for a substrate to be recognized and then degraded by the proteasome, a chain of at least four ubiquitins linked via their lysyl residue 48 must be generated on a single substrate lysyl -amino group (Chau et al., 1989; Finley et al., 1994; Thrower et al., 2000). But experiments in mammalian cell extracts using an E2 that forms polyubiquitin chains connected through ubiquitin lysyl residue 11 (Baboshina and Haas, 1996) or assays performed with purified proteasomes using an endogenous substrate, yeast cyclin B, ubiquitylated in vitro with lysine 11 and 63 linkages (Kirkpatrick et al., 2006) suggest that other ubiquitin–ubiquitin linkages can serve as proteolytic signals, even in the complete absence of lysine 48 linkages. In addition, the attachment of a lysine-63-linked tetraubiquitin chain to lysine 48 of the ubiquitin portion of the UbDHFR fusion protein permits degradation of this model substrate by purified proteasomes (Hofmann and Pickart, 2001). Furthermore, multiple short chains of mixed lysyl linkages are formed on cyclin B in vitro by the APC (anaphase-promoting complex) E3 ubiquitin ligase. These ubiquitylated forms of cyclin B interact with substrate recognition proteins and are substrates for in vitro degradation by purified proteasomes (Kirkpatrick et al., 2006), suggesting that ubiquitin chain formation may be more complex, and that the requirements for recognition by proteasomes and associated proteins may be more relaxed than previously appreciated. After delivery to the proteasome mediated in part by ubiquitin binding proteins, the polyubiquitylated substrate can then be deubiquitylated by the proteasome's regulatory cap or associated proteases. The free ubiquitin becomes available again for attachment, i.e. it is not degraded concomitant with the substrate. The deubiquitylated substrate is fed into the proteolytic core of the proteasome where it is cleaved into small peptides. After release, these peptide fragments are further hydrolysed to free amino acids by multiple separate proteases and/or proteolytic complexes. One of the few characterized examples is tripeptidyl peptidase II, a homooligomeric complex that hydrolyses larger peptides to tripeptides and could function to process proteasome products (Book et al., 2005).
Although the best-characterized function of ubiquitylation is to target substrates for proteolysis by the proteasome, ubiquitylation has also been linked to non-proteasomal functions, currently mostly in fungi and vertebrates. Ubiquitin chains linked via ubiquitin lysine 63 (UbK63) are involved in several processes including DNA repair, protein activation (reviewed in Schnell and Hicke, 2003), endolysosomal degradation (Duncan et al., 2006) and ribosomal regulation (Spence et al., 2000), and there is some evidence that UbK6-linked chains synthesized by the heterodimeric E3 ligase BRCA1/BARD1 mediate regulation of DNA replication and repair (Morris and Solomon, 2004; Nishikawa et al., 2004). Although there is also evidence for the formation of chains using lysines 11, 27, 29 and 33 in vivo (Peng et al., 2003), knowledge of the functional importance of these alternative polyubiquitin chains is lacking. Meanwhile, monoubiquitylation is required for receptor endocytosis, transcriptional regulation and intracellular sorting (reviewed in Schnell and Hicke, 2003). The non-degradative functions of ubiquitylation are ripe areas for future research in plants.
Components of the UPS have been identified in many species of plants through mutant screens and bioinformatic searches (reviewed in Moon et al., 2004). Strikingly, over 6 % of the predicted arabidopsis (Arabidopsis thaliana ecotype ‘Columbia’) genome encodes proteins involved in the UPS (Downes and Vierstra, 2005; with selected examples in Table 1), and analyses of other plant genomes demonstrate a similar abundance of UPS-related genes. Not surprisingly, components of the UPS have been implicated in countless processes including organ initiation and patterning (Samach et al., 1999; Shen et al., 2002; Imaizumi et al., 2005), light signalling (reviewed in Hoecker, 2005) and circadian clock regulation (Han et al., 2004), but this review will focus on the role of UPS in hormone production, perception and signal transduction, and in plant defence. Although the relationship between the UPS and plant signalling became evident about 20 years ago, when the light-signalling protein phytochrome A was shown to be ubiquitylated (Shanklin et al., 1987), the broad scope of ubiquitin's involvement in plant biology was not predicted. On-going discovery, annotation and functional characterization of UPS genes continues to provide insight into the mechanisms that plants use to regulate their growth and development through targeted protein degradation.
|
There are other excellent reviews covering the UPS in plants (Moon et al., 2004; Smalle and Vierstra, 2004), hormone signalling (auxin: Schwechheimer and Schwager, 2004; Woodward and Bartel, 2005; GA: Sun and Gubler, 2004; Fleet and Sun, 2005; ABA: Himmelbach et al., 2003; BR: Vert and Chory, 2006; Z-Y Wang et al., 2006; ethylene: Chen et al., 2005; Etheridge et al., 2006; JA: Turner et al., 2002; Devoto and Turner, 2005; Lorenzo and Solano, 2005) and plant defence mechanisms (Dangl and Jones, 2001; Thordal-Christensen, 2003; Nurnberger et al., 2004; Schulze-Lefert and Bieri, 2005; Zeng et al., 2006), which cogently describe the experimental underpinnings of our current understanding of these processes. But, with the rapid pace of scientific progress in these exciting areas, new information continues to emerge. This review seeks to build upon the foundation of the previous reviews by drawing together information concerning both internally and externally regulated plant developmental processes that depend upon the UPS both to highlight some common themes and to present the most recent published findings.
|
OVERVIEW OF UPS COMPONENTS |
|---|
|
TOPABSTRACTINTRODUCTIONOVERVIEW OF UPS COMPONENTSAPPENDIX 1APPENDIX 2ACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Before discussing the specific function of the UPS in various hormone signal transduction and plant defence pathways, it is useful to highlight some of the many components that are required to regulate properly the attachment and removal of ubiquitin from potential substrates. A website dedicated to the ubiquitin pathway in arabidopsis lists the identified components (http://plantsubq.genomics.purdue.edu). In the genome of arabidopsis only two genes encode an E1, the first enzyme in the ubiquitylation pathway (Hatfield et al., 1997), whereas the E2 gene family is predicted to contain 37 members (Kraft et al., 2005). The precise function of individual members of this expanded family of ubiquitin-conjugating enzymes remains to be determined. Eight additional genes code for putative UEV (ubiquitin-conjugating E2 enzyme variant) proteins in arabidopsis (Kraft et al., 2005). UEVs resemble ubiquitin-conjugating enzymes, but lack the catalytic cysteine (Broomfield et al., 1998; Sancho et al., 1998). Not surprisingly, human UEV1 fails to promote polyubiquitylation in the presence of an E1 and cellular fractions containing E3s (Sancho et al., 1998), but UEVs do appear to function in UPS-related processes. For instance, Mms2p, a UEV from budding yeast, interacts with a canonical E2, Ubc13p, to direct the formation of UbK63-linked polyubiquitin chains (Hofmann and Pickart, 1999), whereas Tsg101 in humans binds ubiquitin and participates in endosomal sorting and HIV budding (Garrus et al., 2001; Katzmann et al., 2001; Sundquist et al., 2004). By contrast, the COP10 UEV, first identified in arabidopsis, appears to interact with a number of E2s and enhances the formation of UbK48- and UbK63-linked polyubiquitin chains in vitro, perhaps through its ability to increase E2/ubiquitin thioester formation (Yanagawa et al., 2004). Additionally, COP10 forms complexes with components of Cullin4-based E3 ligases described below (Chen et al., 2005).
The vast bulk of the arabidopsis UPS-related genes encode components of E3 ubiquitin ligases (Smalle and Vierstra, 2004). Two mechanistic classes of E3 ligases facilitate the transfer of ubiquitin from the E2 to the substrate. In the case of the HECT domain E3 ligases, ubiquitin forms a covalent thioester linkage with a cysteinyl sulfhydryl group on the HECT protein before being transferred to a lysine on the substrate. Using the conserved domain containing this cysteinyl residue, seven HECT proteins have been identified in arabidopsis (Downes et al., 2003).
The other E3 ligase class does not covalently link with ubiquitin, but rather non-covalently interacts with an E2 protein carrying ubiquitin. Currently, there are two groups within this class, the U-box domain- and RING domain-containing proteins and both domains are thought to be structurally and functionally similar. The U-box and RING domains are required for E2 interaction in their respective proteins, utilizing H-bonds or zinc chelation, respectively, to build a platform and expose the appropriate interface (Zheng et al., 2000; Ohi et al., 2003; Andersen et al., 2004). There are approximately 61 predicted U-box proteins in arabidopsis (http://www.arabidopsis.org/info/genefamily/pub.html). By contrast, over 450 proteins in arabidopsis contain one or more RING domains (Stone et al., 2005).
Cullin RING ligases in plants
RING proteins may act alone, as homo- or heterodimers, or in complex assemblies. Two RING family members, RBX1a and RBX1b (in arabidopsis; Gray et al., 2002), function redundantly to play an essential role in promoting ubiquitylation as part of the multi-subunit E3 ligases called cullin RING ligases (CRLs) (reviewed in Petroski and Deshaies, 2005) – a type universally present in eukaryotes (Willems et al., 2004; Petroski and Deshaies, 2005). Each CRL includes a cullin (CUL in arabidopsis) protein that contains a C-terminal RBX1 binding site. Distinct cullin proteins that define the different types of CRLs bind different adaptors and substrate-recruiting subunits at the cullin N-terminus (see figure in Appendix 2). Eleven CUL-like proteins have been identified in arabidopsis based on identity to vertebrate cullins (Shen et al., 2002) and six have been implicated in CRLs.
Arabidopsis CUL1 (Gray et al., 1999) forms a CRL called an SCF complex (named for the subunits Skp1, cullin1 and F-box). In arabidopsis cells, CUL1 co-immunoprecipitates with HA-tagged ASK1, one of 21 predicted arabidopsis SKP family members (called ASKs for Arabidopsis SKP1-like) (Farras et al., 2001). Both ASK1 and ASK2 interact with several F-box proteins in yeast-two-hybrid assays, including COI1, EID1, TIR1 and UFO (Gray et al., 1999; Samach et al., 1999; Dieterle et al., 2001; Xu et al., 2002) and many of these interactions have been confirmed in vitro or by co-immunoprecipitation assays using arabidopsis extracts. The phenotype of plants with single mutations in these family members suggests that ASK1 and ASK2 perform overlapping but not totally redundant functions. ask2-1 null mutants grow quite normally (Liu et al., 2004). ask1-1 mutants display some defects in vegetative and floral development, such as smaller rosette size, reduced filament length and male sterility (Zhao et al., 1999) owing to a defect in homologous chromosome separation in meiosis (Yang et al., 1999). Meanwhile, ask1 ask2 double mutants exhibit severe defects in embryogenesis and die as seedlings, plus they have higher levels of cyclin D3 (Liu et al., 2004). D-type cyclins are degraded via an SCF in humans (Yu et al., 1998) and cyclin D3 levels rise in arabidopsis seedlings with lower level of RBX1 (Lechner et al., 2002), further implicating ASK1 and ASK2 in SCF-mediated degradation. The biochemical function of the other ASKs has not been elucidated, but the varying RNA expression patterns (Zhao et al., 2003; Takahashi et al., 2004) and RNAi-induced phenotypes observed for different ASK family members suggests that they could regulate particular processes in different organs or tissues (Zhao et al., 2003), although there is also evidence that single knock-outs are often insufficient to cause phenotypic abnormalities (Takahashi et al., 2004).
The SKP/ASK proteins in turn serve as adaptor proteins by interacting with one or more of the 700-plus F-box (substrate-recruiting) proteins encoded by the arabidopsis genome (Gray et al., 1999; Gagne et al., 2002; Kuroda et al., 2002; Risseeuw et al., 2003; Takahashi et al., 2004). The F-box domain derives its name from the SKP/ASK interaction motif found in mammalian cyclin F (Bai et al., 1996). F-box proteins bind an SKP/ASK protein and a UPS substrate simultaneously, thus bringing the substrate in proximity to the CUL1-RBX1-tethered E2 enzyme with an attached activated ubiquityl group. The close proximity presumably facilitates ubiquitin transfer to the substrate, although E2 binding to the RING domain may additionally allosterically activate the E2 for ubiquitin transfer (Ozkan et al., 2005). Another protein, SGT1b, can also associate with this complex (Azevedo et al., 2002; Y. Liu et al., 2002; Gray et al., 2003), and has been shown to reduce the accumulation of at least one SCF substrate in arabidopsis (Gray et al., 2003), but its full role in plant SCFs and proteolysis in general (Holt et al., 2005; Azevedo et al., 2006) is not completely clear.
The other arabidopsis cullin most closely related to CUL1, CUL2, also shows enhanced interaction with one F-box protein when ASK1 is over-expressed in a yeast-two-hybrid assay (Risseeuw et al., 2003), and so probably forms a CRL similar to that of CUL1, but there are as yet no in planta data.
Another class of CRL utilizes the CUL3a or CUL3b protein as the primary scaffold component. Like CUL1, CUL3a and CUL3b associate with the RBX1 E2-recruiting protein, as demonstrated through yeast-two-hybrid (Dieterle et al., 2005; Gingerich et al., 2005; Weber et al., 2005), in vitro pull-down (Weber et al., 2005) and plant extract-based co-immunoprecipitation experiments (Figueroa et al., 2005). But, unlike CUL1, CUL3a and CUL3b do not interact with ASK1; instead they interact in yeast-two-hybrid and pull-down assays with BTB domain-containing proteins (Dieterle et al., 2005; Figueroa et al., 2005; Gingerich et al., 2005; Weber et al., 2005). These interactions depend on residues within the N-terminus of CUL3 and for BTB proteins, residues within the BTB domain of several family members that have been studied (Figueroa et al., 2005; Weber et al., 2005). BTB proteins associate directly with targets of the CUL3-based CRLs, combining adaptor and substrate-recruiting functions in one protein (Pintard et al., 2003b). Thus, it is not surprising that the 80 putative BTB proteins found in arabidopsis contain a number of additional domains such as ankyrin repeats (reviewed in Michaely and Bennett, 1992) and the MATH domain (Xu et al., 2003) implicated in protein–protein interactions (Dieterle et al., 2005; Figueroa et al., 2005; Gingerich et al., 2005; Weber et al., 2005) that are predicted to tether substrates to the CRL for ubiquitylation.
Forward genetic identification of CUL1 (Hellmann et al., 2003; Quint et al., 2005) and reverse genetic studies on CUL1 (Shen et al., 2002; Hellmann et al., 2003) and CUL3a and b (Dieterle et al., 2005; Figueroa et al., 2005; Gingerich et al., 2005; Thomann et al., 2005) have established their essential nature. cul1 loss-of-function mutants are early embryo lethals (Shen et al., 2002; Hellmann et al., 2003), suggesting that CUL2 cannot functionally replace CUL1. By contrast, CUL3a and CUL3b appear to be redundant as mutation of either CUL3 gene does not obviously compromise plant growth (Figueroa et al., 2005) and only mild phenotypic changes such as slightly delayed flowering and reduced hypocotyl response to far red light are reported for the cul3a-1 mutant (Dieterle et al., 2005). However, CUL3 activity is required for viability given that cul3a/cul3b double mutants are early embryonic lethals (Dieterle et al., 2005; Figueroa et al., 2005; Gingerich et al., 2005; Thomann et al., 2005). These findings suggest that both CUL1 and CUL3a/b broadly influence a wide range of essential pathways. By contrast, based on phenotypes of mutants, the individual F-box and BTB family members that potentially bring substrates to CUL1 and CUL3, respectively, appear to regulate more specific aspects of plant development and growth responses during the plant life cycle, ranging from seedling blue-light-mediated phototropism (Motchoulski and Liscum, 1999; Inada et al., 2004) to leaf senescence (Woo et al., 2001; also see below and reviewed in Moon et al., 2004; Gingerich et al., 2005), and these phenotypes are consistent with their functioning in a specific CRL. It is worth mentioning that non-CRL functions, although not documented in plants, are possible for F-box and BTB proteins. Two F-box proteins in budding yeast, Rcy1p and Ctf13p (Kaplan et al., 1997; Russell et al., 1999; Galan et al., 2001), and one BTB protein in Caenorhabditis elegans, MEL-26 (Luke-Glaser et al., 2005), appear to perform CRL-independent functions. However, the stability of the Ctf13p (Kaplan et al., 1997) and MEL-26 (Pintard et al., 2003b) proteins still depends on CRLs, suggesting that the biological function of these proteins remains influenced by the UPS.
The most recently characterized plant CRL uses CUL4 and is similar to animal CUL4 CRLs (McCall et al., 2005; Bernhardt et al., 2006; Chen et al., 2006). Two proteins, AtCUL4-L (long) and AtCUL4-S (short), encoded by the arabidopsis CUL4 gene differ in length by 50 amino acids, but both contain the N- and C-terminal cullin functional domains (Chen et al., 2006). To date, studies have focused on the long version of CUL4. AtCUL4-L interacts with RBX1 at its C-terminus, binds directly to DDB1a (from human Damaged DNA Binding-1) at its N-terminus (Bernhardt et al., 2006; Chen et al., 2006), and connects indirectly with DDB2 or DET1 via DDB1a (Bernhardt et al., 2006). One reasonable interpretation of these observations is that DDB1 serves as the adaptor subunit (analogous to ASK/SKP in CUL1 CRLs) and DDB2 and DET1 are alternative substrate-recruiting subunits (analogous to F-box proteins). However, there is also evidence for the formation of larger CUL4-based complexes. In humans, degradation of the c-Jun transcription factor requires a complex containing CUL4, ROC1 (RBX1 homologue), DDB1, DET1 and COP1, a RING E3 ligase initially identified in plants for its role in regulating photomorphogenesis (Deng et al., 1991; Wertz et al., 2004). In arabidopsis, CUL4-L appears to associate more directly with the CDD complex, composed of the UEV protein COP10, along with DDB1 and DET1, but lacking COP1 (Yanagawa et al., 2004) as indicated by yeast-two-hybrid, in vitro and plant-based co-immunoprecipitation experiments (Chen et al., 2006). However, the CDD complex shows signs of associating with COP1 in arabidopsis plants, and may regulate COP1's ubiquitylation activity. Although the DET1, COP1 and COP10 proteins all contribute to light-regulated signalling, the varying phenotypes of arabidopsis plants with reduced levels of CUL4 protein suggest that it may participate in CRLs with other substrate adaptor units to regulate the degradation of additional proteins (Chen et al., 2006).
Interestingly, there was some early evidence that DDB1 could bind either directly to a substrate (like a BTB protein) or function as an adaptor (like SKP1), but recent work favours the latter interpretation. Tandem affinity purification followed by mass spectrometry has shown interaction between human DDB1 and a large family of proteins termed DCAFs (DDB1-Cullin4-Associated Factors) (Angers et al., 2006; Jin et al., 2006). The majority of these proteins contain WD40 repeats, including DDB2, CSA and COP1 previously shown to function in CUL4-based E3 ligases (Wertz et al., 2004; Groisman et al., 2006; H. Wang et al., 2006). Mutational studies demonstrate that two conserved WDXR motifs in DCAFs are important for their binding to varying portions of a double beta propeller fold in DDB1 (Jin et al., 2006). A limited blastp search through the NCBI database using several human DCAF proteins reveals the existence of WD40-repeat-containing proteins with two or more WDXR motifs in a number of plant species, including arabidopsis, rice (Oryza sativa) and Medicago truncatula, suggesting that plant homologues of DCAFs may exist (Madden et al., 1996). In addition to binding DDB1, DCAFs presumably recruit substrate proteins and bring them into proximity with the CUL4-ROC1-tethered Ub-charged E2. For instance, the DCAF Cdt2 binds to Spd1 in fission yeast (Schizosaccharomyces pombe) and promotes its degradation (Liu et al., 2005). As previously observed for some F-box proteins (Zhou and Howley, 1998) and BTB proteins (Wee et al., 2005), DCAF proteins may also become victims of their own CRL, as ubiquitylation of the DCAF human DDB2 can be promoted by a CUL4A-DDB1-based E3 ligase (Matsuda et al., 2005). To date, it remains unknown whether DDB1 is the only adaptor protein to function in CUL4-based E3 ligases. Two human proteins, CPSF160 and SAP130, show structural similarity to DDB1 (Martinez et al., 2001; Li et al., 2006), suggesting that other CUL4-based ligases may exist. Defining the nature of CUL4 CRL complexes is currently an active area of investigation in plants and in other eukaryotes.
Humans also possess a CRL that uses the adaptor proteins elongin B/C instead of SKP to bring substrates to human CUL2 (not homologous to arabidopsis CUL2) and human CUL5 (Kamura et al., 2004; reviewed in Willems et al., 2004). One arabidopsis protein is reported to have similarity to elongin C (Risseeuw et al., 2003) but no clear homologues of the elongin B adaptor protein, nor human CUL2/CUL5 orthologues have been discovered in plants to date (Shen et al., 2002).
Finally, the APC, among its 11 or more subunits, contains a cullin-like protein, APC2, and a small RING-domain-containing protein, APC11. This CRL, found in plants (Capron et al., 2003), most likely helps to control the stability of cell-cycle regulatory proteins (Fulop et al., 2005).
Regulation of CRL activity
Cullin-based ubiquitin ligases are subject to regulation by a cycle of covalent addition and removal of the 76-amino-acid ubiquitin-like protein called RUB (Related to Ubiquitin; known as Nedd8 in fission yeast and animals). RUBs/Nedd8 attach to a conserved lysyl e-amino group near the C-terminus of all six human cullin family members (Hori et al., 1999). To date the same modification has been directly demonstrated for arabidopsis CUL1 (del Pozo and Estelle, 1999), CUL3a and CUL3b (Figueroa et al., 2005) and CUL4 is present in two electrophoretic forms, including a slower migrating form that increases in abundance in COP9 signalosome mutants (see below for relevance), strongly suggesting that it is also modified by RUB (Chen et al., 2006). RUB attachment, similar to ubiquitin attachment, occurs through the activities of E1-, E2- and E3-like enzymes. The E1-like protein is heterodimeric, composed of AXR1 and ECR1 subunits that activate RUB and pass it to the RCE1 RUB E2 conjugating enzyme. Finally, RBX1 appears to act as the RUB E3 ligase that facilitates RUB attachment to the cullin (Gray et al., 2002), although in Caenorhabditis elegans and budding yeast, an additional protein, DCN-1/Dcn1p, facilitates rubylation/neddylation in vitro and in vivo (Kurz et al., 2005).
Arabidopsis contains two closely related (RUB1 and RUB2) and one more divergent (RUB3) RUB family members (Rao-Naik et al., 1998); their precise functional roles are currently under investigation. RUB conjugation appears to stimulate CRL-based activity in vitro (Read et al., 2000; Wu et al., 2000). Curiously, de-rubylation catalysed by the COP9 signalosome (CSN) (Lyapina et al., 2001; Schwechheimer et al., 2001; Cope et al., 2002) also appears to be required for in vivo robust CRL activity in the degradation of Aux/IAA (auxin/indole-3-acetic acid) proteins (Schwechheimer et al., 2001) and HY5 (long hypocotyl5) (Osterlund et al., 2000) in plants, Spd1 in fission yeast (Liu et al., 2003), MEI-1 in C. elegans (Pintard et al., 2003a), and cyclin E in mice (Mus musculus) (Lykke-Andersen et al., 2003) and Drosophila melanogaster (Doronkin et al., 2003). By contrast, in vitro work with yeast and mammalian proteins demonstrates that the CSN effect is inhibitory rather than stimulatory on SCF activity (Lyapina et al., 2001; Zhou et al., 2001; Yang et al., 2002; Groisman et al., 2003).
These seemingly contradictory results have led to the hypothesis that a dynamic cycling of RUB/Nedd8 attachment and removal is needed for robust CRL activity (Lyapina et al., 2001). Recently, one specific mechanism was proposed to explain this observation. The CSN may be required for CRL activity in vivo by preventing proteolysis of CRL components when substrates are not present (reviewed in Wu et al., 2006).
CRL activity is further impacted by the CAND1 protein that binds to both the N- and C-termini of cullin homologues (Min et al., 2003; Goldenberg et al., 2004; Hu et al., 2004). By such interactions it interferes with the binding of substrate recruiting modules (J. Liu et al., 2002; Zheng et al., 2002; Min et al., 2003; Lo and Hannink, 2006) and attachment of RUB/Nedd8, respectively (Hwang et al., 2003; Goldenberg et al., 2004), thus functioning as a negative regulator of SCF activity in vitro. Therefore, one would expect in vivo loss of CAND1 to hyperactivate SCF activity, but the opposite is observed. Elimination of CAND1 from arabidopsis stabilizes the transcriptional regulators HY5 (Feng et al., 2004), RGA (repressor of ga1-3) (Feng et al., 2004) and IAA7/AXR2 (indole-3-acetic acid7/auxin resistant2) (Chuang et al., 2004) that are normally slated for proteasome-mediated degradation, suggesting that CAND1 acts as a positive regulator of ubiquitin-mediated degradation in vivo through its impact on CRL activity. Therefore, the precise molecular events involved in cycles of CAND1/cullin/RUB/SKP association and dissociation require further characterization. For instance, in vitro, once human CAND1 binds to unmodified human CUL1, even a suprastoichiometric amount of the Nedd8/RUB E2 enzyme in the presence of the Nedd8/RUB E1 cannot attach Nedd8 to CUL1 nor promote the displacement of CAND1 (Goldenberg et al., 2004), leaving open the question of how a new round of CAND1 release and CRL activation can be achieved. Additional recent in vitro studies demonstrate that the F-box protein Skp2 together with Skp1 can promote CAND1 dissociation from CUL1, while a small complex including the p27 substrate recognized by Skp2 protein can inhibit deneddylation of CUL1 by the CSN (Bornstein et al., 2006). These two processes can both promote SCF activity and provide an attractive explanation for increased p27 degradation as Skp2 levels rise near the end of G1. However, the authors point out that additional factors may be required for this process of CAND1 displacement and acknowledge that more work will be required to determine if these findings are broadly applicable to other CRLs (Bornstein et al., 2006).
While the biochemical characterization of CAND1 proceeds, insights into the biological function of CAND1 in plants continue to arise from the identification and characterization of cand1 mutant alleles in arabidopsis. Aberrant phenotypes observed include altered venation patterns in vegetative leaves and cotyledons (Alonso-Peral et al., 2006), modest auxin and ethylene resistance and increased ABA sensitivity compared with wild-type in roots (Cheng et al., 2004; Chuang et al., 2004; Feng et al., 2004), dark-grown hookless hypocotyls, and enhanced sensitivity to low-fluence red light in hypocotyl elongation (Chuang et al., 2004; Feng et al., 2004). Both a missense allele and T-DNA insertion alleles exhibit dwarfing as adult plants with an increased number of rosette leaves with wrinkly blades (Chuang et al., 2004; Feng et al., 2004). T-DNA insertion alleles are almost completely sterile with reduced apical dominance and delayed flowering (Chuang et al., 2004; Feng et al., 2004). These diverse phenotypes are not surprising given CAND1's ability to bind to multiple cullins and consequently to regulate the activities of many CRLs.
Proteasomal recognition and substrate degradation
Clearly, many proteins co-operate to accomplish the regulated attachment of ubiquitin to a substrate, but the story does not end there. Once the protein has been polyubiquitylated, it must be brought to the proteasome, recognized as a proteolytic substrate, deubiquitylated, threaded into the 20S core particle and hydrolysed therein. Selective and regulated deubiquitylation of proteins may play a counteracting role prior to or coincident with proteasomal docking. For instance, deubiquitylation of the human p53 protein by HAUSP inhibits p53 degradation (Li et al., 2002). Although plants contain many putative ubiquitin hydrolases (reviewed in Smalle and Vierstra, 2004), their in vivo roles remain largely uncharacterized. Mechanisms governing substrate recognition, deubiquitylation and degradation by the proteasome also provide important points of regulatory control that require further research.
Ongoing identification of the components of the UPS and characterization of their biochemical activities enhances our understanding of targeted protein degradation in plants. At the same time, a combination of forward genetic, reverse genetic and biochemical techniques enables us to examine the relationship between the UPS and various aspects of plant growth and development, including hormonal signalling and plant defence.
During functional characterization of hormone or defence pathways (discussed in detail here relative to UPS), scientists often monitor the consequences of altered levels of specific UPS components in vivo or in vitro. The molecular and phenotypic consequences depend on the nature of the UPS component (whether it promotes degradation or inhibits degradation) and on the function of the substrate in the pathway under investigation. When a particular UPS component that promotes degradation, such as an E3 ligase, is down-regulated, e.g. through mutation or transcriptional silencing, its substrate(s) should accumulate. So, in the case of a transcriptional activator, this should lead to increased expression of downstream genes (Fig. 2A). However, if the E3 ligase under consideration normally targets a transcriptional repressor for degradation, plants lacking that E3 ligase would have elevated levels of the repressor, and hence reduced transcription of downstream genes (Fig. 2A). The opposite effects would be observed upon over-expression of the E3 ligase, i.e. reduced transcription if the substrate is a transcriptional activator, owing to increased degradation and hence lower steady-state levels, but increased transcription if the substrate is a repressor (Fig. 2A).
|
Similar predictions of expected phenotypes can be developed for other potential E3 ligase substrates such as a kinase, phosphatase, hormone biosynthetic enzyme or hormone catabolic enzyme (Fig. 2B). Additionally, the effects of perturbations in the degradative components of the UPS can be examined at the level of the downstream pathway. For instance, a transcriptional activator can be either a positive or a negative regulator of a signalling pathway (e.g. in response to hormones, light, drought, pathogen attack, etc.) depending on the types of genes it turns on. Similarly, kinases, phosphatases, receptors, transporters, GTPases, acetylases, ion pumps, etc., can all function as either negative or positive regulators of pathways. If one of these proteins that acts as a positive regulator of a pathway (such as the EIN3 transcription factor in ethylene signalling – see below) is subject to UPS-mediated degradation, then a plant lacking the E3 ligase responsible for degradation of the positive signalling component should be hypersensitive to the stimulus (e.g. ethylene) or should be more disease resistant if the protein positively regulates plant defence (Fig. 2B).
In searching for and analysing novel E3 ligase/substrate interactions, these predictable phenotypes (Fig. 2) can be used to make and test hypotheses about the regulatory roles of the UPS and the substrates in the signalling pathways under investigation. Similar molecular and pathway-level predictions can be made for many components of the UPS. For example, altered levels of other proteins that promote degradation, such as 20S proteasome core subunits, AXR1 and E2 ubiquitin-conjugating enzymes, would be expected to cause similar effects to altered levels of E3 ligases. By contrast, changes in the levels of components of the UPS that act to stabilize substrates, such as deubiquitylating enzymes, would most likely promote opposite outcomes.
Auxin and the UPS
In plants, the hormone auxin appears indispensable for plant survival. Often present as indole-3-acetic acid (IAA), auxin plays a critical role in regulating cell growth, division and differentiation, and on a gross morphological scale, auxin clearly impacts apical dominance, root elongation, lateral root formation and many other processes (reviewed in Davies, 2004; Paciorek and Friml, 2006).
Many, but not all, of auxin's effects are linked to auxin-mediated alterations of the transcriptome and the UPS appears to be involved in allowing these transcriptional changes. Multiple members of the Aux/IAA family of proteins act as repressors of auxin-mediated transcription in transient transfection assays using carrot (Daucus carota) or arabidopsis protoplasts (Ulmasov et al., 1997; Tiwari et al., 2001, 2004). These proteins are subjected to rapid degradation and stabilized versions disrupt normal plant growth and auxin signalling, indicating the biological significance of rapid Aux/IAA proteolysis (reviewed in Reed, 2001). Although there is no published report of ubiquitylated forms of these extremely short-lived proteins, evidence for their degradation via the UPS is compelling. The TIR1 F-box protein (Ruegger et al., 1998), and three additional related F-box proteins called AFB1, 2 and 3 (Dharmasiri et al., 2005a, b) interact with Aux/IAA family members in vitro, suggesting that these proteins are substrates of a CUL1-based CRL, either an SCFTIR1 or an SCFAFBx. Two additional family members, AFB4 and AFB5, remain to be characterized biochemically, but the recent identification of afb5 mutants with reduced sensitivity to the structurally distinct picolinate class of synthetic auxins suggests that these proteins could act in a related pathway (Gagne et al., 2002; Dharmasiri et al., 2005b; Walsh et al., 2006).
Evidence for Aux/IAA ubiquitylation also comes from the observations that in vivo degradation of Aux/IAA fusion proteins such as IAA1:LUC (firefly luciferase) (Ramos et al., 2001), AXR2(IAA7):GUS (ß-glucuronidase) and AXR3(IAA17)NT:GUS (a fusion with the N-terminal 102 amino acids of IAA17) (Gray et al., 2001) is slowed by the proteasome inhibitor MG132 and both tagged and endogenous Aux/IAA proteins are stabilized by mutations in several components that affect CRL activity in particular, including AXR1, CAND1, CUL1 and CSN component CSN5 (Gray et al., 2001; Schwechheimer et al., 2001; Zenser et al., 2003; Chuang et al., 2004; Quint et al., 2005).
How do auxin signalling and Aux/IAA proteolysis intersect? Surprisingly, increased levels of auxin promote accelerated degradation of these Aux/IAA transcriptional repressors (Gray et al., 2001; Zenser et al., 2001) by increasing their interactions with TIR1 and related AFBs (Gray et al., 2001; Kepinski and Leyser, 2004; Dharmasiri et al., 2005b). Attempts to uncover the signalling steps between auxin perception and accelerated degradation led to the exciting discovery that auxin itself directly promotes the interaction between TIR1 and tagged forms of Aux/IAA proteins and through this increased interaction probably prompts accelerated proteolysis (Dharmasiri et al., 2005a; Kepinski and Leyser, 2005; and reviewed in Parry and Estelle, 2006). In all experiments, auxin binding occurs in the presence of both a TIR1/AFB protein and its Aux/IAA substrate, leaving open the possibility that both proteins are required to bind this crucial hormone (Dharmasiri et al., 2005a; Kepinski and Leyser, 2005). This mechanism of proteolytic regulation represents a new paradigm for regulation of SCF activity and suggests that small-molecule non-covalent modulation of CRL function may occur in other contexts.
A recent study shows that auxin directs the movement of IAA17:GFP into nuclear protein bodies (NPBs) along with components of SCFTIR1, the COP9 signalosome and the 26S proteasome core (Tao et al., 2005). In arabidopsis cells, this relocalization depends on the RAC1 GTPase, a small signalling protein. RAC1's interactions with various downstream effectors is regulated by rounds of GTP binding and hydrolysis. Expression of a dominant negative form of AtRAC1 suppresses auxin-mediated formation of these NPBs and decreases loss of IAA17:GFP protein, whereas IAA17:GFP levels drop more in cells expressing a constitutively active form of RAC1 (Tao et al., 2005). In addition, work performed in tobacco (Nicotiana tabacum) demonstrates that induction of auxin-responsive genes occurs while GFP-NtRac1 localizes predominantly to the plasma membrane, pointing to the existence of an unknown membrane-associated protein linking RAC GTPases to auxin-mediated acceleration of Aux/IAA degradation (Tao et al., 2002).
There is also evidence that mutation of the HOBBIT Cdc27-like component of the APC stabilizes the IAA17/AXR3 protein. This does not result from a global deceleration of proteolysis as the HY5 transcription factor that participates in light signalling is not similarly stabilized in hobbit mutants (Blilou et al., 2002). It is unclear if the stabilization of IAA17 in hobbit mutants results from any direct interaction between the APC and Aux/IAA proteins, but suggests a link between cell cycle regulation and Aux/IAA degradation.
Although several experiments demonstrate that auxin does not promote a global acceleration of proteolysis (Zenser et al., 2003; H. Li et al., 2004), its effects on proteolysis are not limited to the Aux/IAA proteins. For example, degradation of a phosphorylated E2FB transcription factor that helps regulate cell cycle progression is dramatically slowed in arabidopsis cultured cells incubated with exogenous auxin [(1-NAA)1-naphthalene-acetic acid] (Magyar et al., 2005). By contrast, levels of the NAC1 transcription factor, a positive regulator of auxin-responsive lateral root formation (Xie et al., 2000), drop following treatment with auxin even though its transcript levels are initially induced by auxin (Xie et al., 2000, 2002). Post-translational modification must be the cause of the lowered protein levels because plants expressing 35S:6 x Myc:NAC1 show steady levels of transcript following auxin treatment even though 6 x Myc:NAC1 protein levels drop. The authors found that the SINAT5 RING E3 ligase is capable of ubiquitylating NAC1 in vitro and affecting its accumulation in transgenic plants (Xie et al., 2002). SINAT5 is transcriptionally up-regulated by auxin more slowly than NAC1, but prior to observed decreases in NAC1 protein levels, suggesting that increased levels of SINAT5 cause increased degradation of NAC1 protein in the presence of auxin. It is not clear whether auxin has any additional role in promoting the degradation of NAC1 (e.g. by stimulating post-translational modification that increases its affinity for SINAT5) (Xie et al., 2002).
The auxin signal transduction pathway demonstrates the power of the UPS to control plant behaviour tightly in response to transient or changing signals. For instance, a pulse of auxin will rapidly trigger increased Aux/IAA degradation and turn on genes required for auxin signalling. In roots, this might lead to lateral root formation, mediated in part by the NAC1 protein. But to prevent an over-proliferation of lateral roots, auxin also promotes degradation of the NAC1 protein through increased levels of an E3 ligase that catalyses NAC1 ubiquitylation. Other hormonal and defence signalling pathways could also employ the UPS both to initiate and then to dampen a biological response.
Like SINAT5, two other components of the UPS involved in lateral root development are also transcriptionally up-regulated by auxin, namely the RING E3 ligase, XBAT32 (Nodzon et al., 2004), and the F-box protein CEGENDUO (Dong et al., 2006); however, no substrates for these proteins have been identified. It is interesting to note that the former is a positive regulator of lateral root formation while the latter inhibits lateral root development, again suggesting a potential dual stimulatory and inhibitory role for the UPS in this developmental process.
Several years ago, scientists also uncovered a link between the UPS and auxin transport. AtPIN2/EIR1, an auxin efflux carrier (Petrasek et al., 2006), does not appear to be transcriptionally regulated by auxin, but the levels of a PIN2/EIR1:GUS fusion protein drop following treatment with the auxin 1-NAA or a protein synthesis inhibitor, suggesting that the protein is short-lived and its degradation is regulated by auxin (Sieberer et al., 2000). NAA-mediated loss of EIR1:GUS is slowed by the axr1-3 mutation, intimating that degradation of this protein is regulated by a CRL (Sieberer et al., 2000). This could be through a direct mechanism, or the stability of PIN2/EIR1 could be controlled by a component of the UPS under AXR1-mediated control.
A recent report verified and extended that study by demonstrating that inhibition of proteasome activity results in accumulation of PIN2 protein and by visualizing ubiquitylated forms of PIN2 (Abas et al., 2006). Gravistimulation leads to asymmetric distribution of both endogenous PIN2 and a PIN2:GFP reporter protein that can be blocked by treatment with a proteasome inhibitor. Internalization of the PIN2 protein also appears to be affected by a proteasome-dependent process (Abas et al., 2006). Thus, the UPS affects steady-state levels of PIN2 at the plasma membrane and consequently influences the amount and direction of auxin transport within a plant, but then auxin itself appears to affect the stability of PIN2. This may constitute an important feedback loop to localize and then limit auxin in specific cells.
Gibberellins and the UPS
The gibberellins (GAs), a group of diterpenoid compounds with phytohormone activity, affect various stages of plant development, including seed germination, stem elongation, root growth, flowering and pollen tube elongation (reviewed in Davies, 2004; Swain and Singh, 2005) and like auxin signalling, gibberellin-induced changes in transcription are an important part of the response pathway (Fleet and Sun, 2005). Several features of the GA/UPS-mediated signalling cascade resemble characteristics of the Aux/IAA branch of the auxin signalling pathway. The DELLA proteins, named for highly conserved amino acids at their N-termini, act, like Aux/IAA proteins, as negative regulators of GA signalling. There are five proteins in this family in arabidopsis [GAI (GA insensitive), RGA (Repressor of ga1-3), RGL1 (RGA-like), RGL2 and RGL3 (Dill et al., 2001; Peng and Harberd, 2002)]. SLENDER RICE1 in rice (SLR1) and SLENDER1 in barley (Hordeum vulgare) (SLN1) are single DELLA proteins in these plants (Ikeda et al., 2001) and various other species also have homologous proteins (Thomas and Sun, 2004). As for Aux/IAA proteins, DELLA proteins are subject to hormone-mediated acceleration of degradation. Levels of both endogenous RGA and a GFP:RGA fusion protein drop following GA treatment of arabidopsis seedlings (Silverstone et al., 2001). Protein loss depends on an SCF complex containing the AtSLY1 (SLEEPY) F-box protein (Swain and Singh, 2005). sly1 mutants contain elevated levels of RGA and can no longer eliminate this protein following GA treatment (McGinnis et al., 2003). DELLA proteins (RGA and GAI) and SLY1 interact directly through both yeast-two-hybrid assays and in vitro pull-down experiments using GST-SLY1 (Dill et al., 2004). A single arabidopsis homologue, SNEEZY (SNE), when overexpressed, can compensate for loss of SLY1 suggesting that it also acts as a substrate-recruiting protein for DELLA proteins (Strader et al., 2004). The rice DELLA protein SLR1 also appears to be ubiquitylated through the activity of an SCF complex using the GID2 (Gibberellin-Insensitive Dwarf2) F-box protein (Itoh et al., 2003) in a GA-dependent manner, indicating a conserved function for rice GID2 and its arabidopsis orthologue SLY1.
How gibberellin promotes DELLA protein degradation is unclear. One possibility is that GA-induced post-translational modification of the DELLA proteins enhances their interactions with the F-box component of the SCF E3 ligase. The DELLA proteins do appear to become phosphorylated in the presence of elevated levels of GA in rice (Itoh et al., 2003) and only phosphorylated SLR1 can interact with the GID2 F-box protein in vitro (Gomi et al., 2004). In arabidopsis, there is also evidence for binding of phosphorylated GAI by the F-box protein SLY1 (Fu et al., 2004). But the steps connecting GA and DELLA phosphorylation have not been elucidated.
Recent identification of the first bona fide gibberellin receptors in rice (Ueguchi-Tanaka et al., 2005) and arabidopsis (Nakajima et al., 2006) indicates that the receptor plays a role in GA-induced DELLA degradation. The soluble GA receptors, called GID1 in rice and GID1a, b and c in arabidopsis, bind bioactive GAs. gid1 rice mutants exhibit classic GA insensitivity phenotypes such as dwarfed stature, dark green leaves and lack of alpha-amylase induction in seeds. Sequence homology gives little insight into the molecular mechanisms that underlie GID1-mediated signalling. These proteins resemble hormone-sensitive lipases (HSLs), but lack one of the three conserved residues in the catalytic triad, and rice GID1 cannot hydrolyse an artificial HSL substrate (Ueguchi-Tanaka et al., 2005). However, it is clear that GID1 can bind to the DELLA proteins in a GA-dependent manner recapitulated in yeast and in vitro (Ueguchi-Tanaka et al., 2005; Nakajima et al., 2006). In the rice gid1 mutant, the SLR1 DELLA protein accumulates to higher levels and SLR1 levels fail to drop following treatment with GA, demonstrating that GA-mediated SLR1 degradation depends upon GID1 (Ueguchi-Tanaka et al., 2005). Interestingly, the DELLA proteins RGA and GAI can actually enhance GID1c's affinity for 16,17-dihydro-GA4, suggesting that this DELLA/GID1 two-protein complex may actually be the biological binding target of the hormone (Nakajima et al., 2006). If this model is correct, formation of the GA-GID1-DELLA complex may be required for productive SCFSLY1/GID2-mediated ubiquitylation of the DELLA proteins, and it is unclear how phosphorylation affects this process (Nakajima et al., 2006).
In addition to contributing to GA-mediated regulation of germination, growth and flowering (Wen and Chang, 2002; Macmillan et al., 2005; reviewed in Peng and Harberd, 2002), recent studies demonstrate the importance of the DELLA signalling pathway for plant survival under salt stress. Arabidopsis seedlings grown on 100 mM salt accumulate lower levels of bioactive GA1 and GA4, which should lead to increased DELLA stability. Indeed, a 1-h treatment with 50 mM NaCl is sufficient to increase the levels of a GFP:RGA reporter protein. Salt-induced stabilization of DELLA proteins can then inhibit plant growth to enhance stress tolerance (Achard et al., 2006).
Interestingly, three other hormones have been implicated in regulation of DELLA stability. Depletion of auxin from arabidopsis seedlings through decapitation leads to reduced GA-mediated loss of GFP:RGA from root nuclei whereas application of auxin to the shoot restores rapid GA-induced reduction in GFP:RGA levels (Fu and Harberd, 2003). Ethylene exerts an opposite effect; seedlings grown in ethylene-rich air show slower loss of GFP:RGA from nuclei following GA treatment (Achard et al., 2003). This implies that a ctr1-1 mutant exhibiting a constitutive ethylene response phenotype should accumulate more DELLA proteins capable of promoting stress tolerance. In keeping with this model, ctr1-1 mutants display enhanced survival under high salt conditions. This improved salt tolerance depends at least partially on DELLA proteins as a triple mutant ctr1-1 gai-t6 rga-24 that lacks two DELLA proteins is more susceptible to salt stress than ctr1-1 (Achard et al., 2006). Abscisic acid (ABA) levels also affect DELLA accumulation. Roots treated with ABA show enhanced accumulation of a GFP:RGA fusion protein, both under basal conditions and following treatment with exogenous GA. But, plants mutant for abi1, a serine/threonine phosphatase that positively regulates ABA signalling, no longer accumulate GFP:RGA following ABA treatment (Achard et al., 2006). Intriguingly, this regulation may be cell-type-, organism- or family member-specific as SLN1 protein levels are not affected by ABA in barley aleurone cells (Gubler et al., 2002). It is unclear how these changes in auxin, ethylene and ABA levels cause differential effects on DELLA protein stability, but it does provide an intriguing example of hormonal cross-talk in which multiple pathways can converge to influence the stability of a single family of transcriptional regulators.
GA and ABA also play well-characterized antagonistic roles in regulation of hydrolase production in the aleurone layer of germinated cereal seeds through their effects on another transcription factor, GAMYB. This positive regulator of alpha-amylase expression is transcriptionally up-regulated by GA (Gubler et al., 1995) and repressed by ABA (Gomez-Cadenas et al., 2001). The SLN1 DELLA protein is implicated in the GA-mediated transcriptional regulation of GAMYB (Gomez-Cadenas et al., 2001; Gubler et al., 2002), but a more recent yeast-two-hybrid assay pulled out another potential regulator of GAMYB in barley, named GMPOZ (Woodger et al., 2004). This GAMYB interacting protein derives part of its name from its POZ domain, also known as a BTB domain. This provides a potential link between GAMYB and a CUL3-based CRL. Knock-down of GMPOZ in aleurone cells decreases GA-mediated transcription of an alpha-amylase reporter construct, suggesting that GMPOZ normally positively regulates GA signalling. At the same time, the cells show enhanced ABA-mediated stimulation of a dehydrin reporter, arguing that GMPOZ normally negatively regulates ABA signalling in the barley aleurone layer. As GAMYB protein levels were not assayed, it is unclear how GMPOZ directly affects transcription of GA- and ABA-responsive genes. But, based on the observed functions of BTB proteins in other organisms, it could potentially interact directly with elements of the basal transcription machinery, as seen for BAB1 and BAB2 in Drosophila (Pointud et al., 2001), regulate the stability of GAMYB and/or other transcription factors as seen in Keap1-mediated destabilization of the Nrf2 transcription factor in humans (Furukawa and Xiong, 2005), or even be degraded itself by a CUL3-based CRL as observed for MEL-26 in C. elegans (Pintard et al., 2003b).
Genetic evidence also suggests that a HECT-type E3 ubiquitin ligase, UPL3, influences the GA-regulated process of trichome development as upl3-1 and upl3-2 mutants possess trichomes with elevated numbers of branches, similar to some GA-signalling mutants. upl3-1 and upl3-2 mutants are moderately hypersensitive to GA in a hypocotyl elongation assay, suggesting that some positive potentiator of GA signalling is more abundant in the absence of the UPL3 protein (Downes et al., 2003).
PHOR1, a predicted E3 ligase initially identified in potato (Solanum tuberosum), also appears to participate in GA signalling (Amador et al., 2001). This protein contains a U-box domain (Hatakeyama et al., 2001; Monte et al., 2003), and a series of ARM repeats. An ARM repeat, first identified in the Drosophila protein ARMadillo, is approximately 40 amino acids in length and is typically present in a variable number of tandem copies and functions in protein–protein interactions (reviewed in Hatzfeld, 1999). Plants expressing lowered levels of PHOR1 have shorter, GA-insensitive stems, whereas plants constitutively over-expressing PHOR1 develop longer stems that are hypersensitive to treatment with low levels of GA. Both sets of experiments argue that PHOR1 is normally a positive regulator of GA-stimulated growth. Increased levels of bioactive (GA20) and inactive (GA8, GA29) gibberellins accumulate in PHOR1 antisense lines consistent with a role for PHOR1 in regulation of GA metabolism, and indicative of a lack of normal feedback inhibition in this genetic background. Although a PHOR1:GFP fusion protein can be found in the cytoplasm or nucleus of tobacco BY cells following transient transfection, the majority of PHOR1:GFP signal is localized to the nucleus following treatment with exogenous GA (Amador et al., 2001). Three arabidopsis proteins, AtPUB27, AtPUB28 and AtPUB29, are more than 50% identical to PHOR1 and may link the UPS to GA signalling in arabidopsis through degradation of unidentified substrate(s) (Monte et al., 2003).
Although there is no evidence for UPS-mediated regulation of GA transport, a recent paper presents an exciting link between light signalling, targeted protein degradation and regulation of GA biosynthesis prior to seed germination (Oh et al., 2006). For many years it has been known that light signalling through the phytochrome family of photoreceptors can promote seed germination (Shinomura et al., 1994; Hennig et al., 2002), and more recently it was discovered that this occurs through light-induced transcriptional up-regulation of GA biosynthetic enzymes (Yamaguchi et al., 1998). The PIL5 (PIF3-Like 5) transcription factor represses seed germination by regulating the abundance of transcripts for GA biosynthetic and catabolic enzymes to decrease GA levels in seeds kept in the dark. Consistent with this finding, over-expression of PIL5 blocks up-regulation of the GA biosynthetic enzymes GA3ox1 and GA3ox2 and simultaneously promotes expression of the GA2ox2 oxidase that inactivates GA4 and other biologically active GAs (Oh et al., 2006). Now there is evidence that PhyA and PhyB-mediated signalling can prompt degradation of Myc-tagged PIL5 to increase GA levels and promote germination in response to light. Higher levels of the PIL5 fusion protein accumulate in phyA mutants than in wild-type plants following a pulse of red light, and when examined over a longer period, phyB mutants also show elevated levels of Myc-tagged PIL5 relative to wild-type seedlings following a pulse of red or far-red light. The proteasome inhibitor MG132 blocks light-stimulated loss of Myc-PIL5 following a red or far-red light pulse, arguing that the UPS is required for the normal reduction of PIL5 protein levels in plants exposed to light (Oh et al., 2006). As PIL5 is degraded, GA biosynthetic genes are up-regulated while a GA catabolic gene is repressed, leading to the overall accumulation of GA required for seed germination.
Abscisic acid and the UPS
The hormone abscisic acid (ABA) is another prominent regulator of seed germination that also enables plants to respond to abiotic stresses such as drought. ABA can directly affect ion transport in guard cells to alter stomatal aperture rapidly in response to changing water availability (reviewed in Roelfsema et al., 2004), but other slower ABA responses require changes in transcription. And, as observed for other phytohormones, the UPS is implicated in regulation of ABA-responsive transcription.
The first evidence of ABA-mediated changes in protein stability was observed for the ABI5 (abscisic acid insensitive5) protein in arabidopsis. Unlike the Aux/IAA and DELLA proteins, this basic leucine zipper transcription factor acts as a positive regulator of ABA responses. Therefore, it might not seem surprising that rather than promoting degradation of ABI5, abscisic acid is required for the maintenance of high levels of the ABI5 protein. This was discovered using transgenic arabidopsis plants expressing an HA:ABI5 fusion protein under the control of the constitutive 35S promoter. Although the levels of transcript for this fusion protein remain unaltered following ABA treatment, HA:ABI5 protein levels increase substantially. ABA treatment also blocks loss of the ABI5 protein from cycloheximide-treated seedlings incapable of performing de novo protein synthesis (Lopez-Molina et al., 2001), and treatment of seedlings with proteasome inhibitors leads to the accumulation of ubiquitylated HA:ABI5 (Lopez-Molina et al., 2003), implicating the UPS in ABI5 degradation. This hypothesis is strengthened by the observation that ABI5 is stabilized in the rpn10-1 mutant that has a defect in one of the nine RPN (Regulatory Particle, Non-ATPase) subunits of the 26S proteasome lid. Interestingly, two other proteasome substrates, PhyA and HY5, are not stabilized in this mutant background, suggesting that RPN10 may promote degradation of a subset of proteasome substrates including ABI5 (Smalle et al., 2003). It is unclear how ABA stabilizes ABI5, but several experiments demonstrate that the ABI5-interacting protein (AFP) plays a role in down-regulating ABI5 protein levels in the absence of ABA (Lopez-Molina et al., 2003). Upon removal of ABA from the media, seedlings that are over-expressing AFP show a much more rapid loss of ABI5 than wild-type seedlings or afp mutants. ABI5 and AFP were initially shown to interact in a yeast-two-hybrid assay and this association was confirmed in planta through co-immunoprecipitation experiments. Fluorescently tagged AFP and ABI5 proteins also co-localize in nuclear protein bodies in transgenic arabidopsis seedlings treated with ABA. These nuclear protein bodies also include the COP1 RING E3 ligase when all three proteins are bombarded into onion epidermal cells (Lopez-Molina et al., 2003). However, to date there is no evidence that COP1 ubiquitylates ABI5, so the search is still on for the E3 ligase that acts upon ABI5, and for the abscisic acid receptor and/or other signalling components that link altered ABA levels to changes in ABI5 proteolysis.
The degradation pathway of another ABA-responsive transcription factor, ABI3, has also been studied. This B3-domain-containing transcription factor plays a prominent role in ABA-signalling. Like ABI5, an ABI3:6Myc fusion protein was recently found to be relatively short-lived under basal conditions, but to be long-lived in the presence of proteasome inhibitors (Zhang et al., 2005). It appears that ABI3 degradation is mediated, at least in part, by the AIP2 (ABI3-Interacting Protein2) RING-type E3 ubiquitin ligase. ABI3 and AIP2 interact in a yeast-two-hybrid assay and tagged versions of these proteins co-immunoprecipitate in arabidopsis extracts. Moreover, ABI3 is a substrate of AIP2-mediated ubiquitylation in an in vitro assay. Not surprisingly, over-expression of AIP2:3 x HA reduces the accumulation of ABI3 in arabidopsis seedlings, whereas levels of the endogenous ABI3 protein increase in aip2-1 mutants. ABA regulates the stability of ABI3, but unlike for ABI5, elevated levels of ABA reduce ABI3 accumulation. Similar to the NAC1/SINAT5 signalling system in the auxin pathway, it seems that ABA-mediated transcriptional up-regulation of the AIP2 RING-E3 ligase allows it to degrade more ABI3, providing a partial mechanistic explanation for the link between ABA levels and ABI3 proteolysis. Nevertheless, Zhang et al., (2005) conclude that AIP2 might target other proteins for ubiquitin-mediated degradation, and that ABI3 may also be ubiquitylated by other E3 ligases.
Yet another positive regulator of ABA transcription, the ABF2 (ABRE Binding Factor2) transcription factor, also interacts with a potential component of an E3 ligase, ARIA (ARM protein Repeat Interacting with ABF2), in both yeast-two-hybrid assays and in vitro pull-down experiments (Kim et al., 2004). Over-expression or knock-out of ARIA affects a subset of ABA-regulated processes and indicates that ARIA is a positive regulator of ABA signalling. For example, aria mutants show reduced sensitivity to ABA in root elongation assays. In addition to its ARM repeats, ARIA possesses a BTB domain. Although a GST:ARIA fusion protein can pull down ABF2 in vitro (Kim et al., 2004), it could also potentially mediate interactions with a CUL3 protein and thereby link ABF2-regulated processes to the UPS. As in the case of the GMPOZ BTB-protein involved in negative regulation of ABA signalling in barley (see above), there is no evidence of whether ARIA or its potential substrates undergo UPS-mediated degradation.
Several other sets of experiments highlight potential relationships between ABA signalling and the UPS. First, a characterization of the ATL [Arabidopsis Toxico para Levadura (toxic to fungi)] family of RING E3 ligases included a phenotypic analysis of lines with T-DNA insertions in various ATL genes. A mutation in ATL43 renders seedlings insensitive to ABA in a seed germination assay (Serrano et al., 2006). The U-box E3 ligase AtCHIP may also be involved in ABA signalling. Over-expression of the protein makes plants hypersensitive to the effects of ABA in a germination assay and in a stomatal aperture assay. AtCHIP seems capable of attaching a single ubiquitin moiety to the A subunit of protein phosphatase 2A, but it is unclear whether this is functionally related to the ABA signalling pathway (Luo et al., 2006).
The possible involvement of an SCF-complex in ABA signalling was uncovered through characterization of TLP9 (Lai et al., 2004). This member of the TUBBY-LIKE family of genes in arabidopsis contains a C-terminal tubby domain [named for a protein identified in obese mutant mice (Kleyn et al., 1996)] and an N-terminal F-box domain. Yeast-two-hybrid assays confirm that this F-box protein can interact with arabidopsis ASK1. Phenotypic analyses of two tlp9 insertional mutants indicate that this protein is required for normal ABA signalling. The mutant plants show reduced sensitivity to ABA in seedling germination assays whereas plants over-expressing TLP9 under the control of the 35S promoter exhibit ABA hypersensitivity and lower rates of germination. These phenotypes lead to the prediction that TLP9's substrate(s) are negative regulators of ABA signalling or levels (Fig. 2B), but none has been identified to date. It is interesting to note that in the abi1 ABA-insensitive mutants, TLP9 transcript levels are reduced (Lai et al., 2004).
Finally, there is new evidence that the XERICO RING E3 ligase contributes to regulation of ABA levels in arabidopsis, potentially in response to osmotic and salt stress. Transcription levels of this gene rise in response to high salt and osmotic stress – conditions that normally increase ABA levels (Ko et al., 2006; reviewed in Zhu, 2002). Over-expression of XERICO prompts an increase in ABA levels in transgenic plants relative to wild-type plants when both are grown in well-watered conditions; drought conditions further increase ABA accumulation even in the 35S:XERICO plants. Transcriptional profiling demonstrates an up-regulation of the NCED ABA biosynthetic gene in both wild-type and transgenic plants following treatment with ABA, but the level of induction is greater in the 35S:XERICO plants. XERICO interacts with the E2 ubiquitin-conjugating enzyme, UBC8, in yeast-two-hybrid assays. Unexpectedly, the TLP9 F-box protein also interacts with XERICO in this experiment. Although both appear to function in ABA signalling and show a high degree of co-expression at the transcriptional level, the functional significance of this novel RING/F-box protein interaction has not been discerned (Ko et al., 2006).
Brassinosteroids and the UPS
Brassinosteroids (BRs), a class of polyhydroxylated steroid hormones, also have pleiotropic effects on plant development and regulate processes such as cell division, cell elongation, photosynthesis, vascular differentiation and stem elongation (reviewed in Bishop and Koncz, 2002).
In the BR signal transduction pathway, the first hint that the UPS is involved in BR-mediated transcriptional responses came from analyses of the bes1-D (bri1-EMS-suppressor 1 Dominant) and bzr1-1D (brassinazole-resistant 1-1Dominant) mutants that display BR hypersensitivity (Wang et al., 2002; Yin et al., 2002). These dominant mutants harbour identical proline to leucine missense substitutions in two related proteins that are transcription factors. Analysis of bzr1-1D:CFP and BZR1:CFP fusion proteins in arabidopsis seedlings demonstrates that the mutant protein accumulates to substantially higher levels, indicating that it is stabilized relative to its wild-type counterpart. This suggests that these positive regulators of BR-mediated transcriptional changes must normally be degraded to prevent constitutive BR signalling. Subsequent work identified additional regulatory components that led to the development of a model of BR signalling with some similarity to the Wnt/GSK3/ß-catenin signalling pathway found in metazoans (reviewed in Vert et al., 2005; Cadigan and Liu, 2006). New findings, such as those regarding subcellular localization of BR signalling components, highligh