|II.||Components of the ubiquitin–proteasome system||14|
|III.||Ubiquitin-mediated degradation: a recurrent theme in the plant life cycle||18|
|IV.||Conclusion and future prospects||25|
|II.||Components of the ubiquitin–proteasome system||14|
|III.||Ubiquitin-mediated degradation: a recurrent theme in the plant life cycle||18|
|IV.||Conclusion and future prospects||25|
Ubiquitin is well established as a major modifier of signalling in eukaryotes. However, the extent to which plants rely on ubiquitin for regulating their lifecycle is only recently becoming apparent. This is underlined by the over-representation of genes encoding ubiquitin-metabolizing enzymes in Arabidopsis when compared with other model eukaryotes. The main characteristic of ubiquitination is the conjugation of ubiquitin onto lysine residues of acceptor proteins. In most cases the targeted protein is rapidly degraded by the 26S proteasome, the major proteolysis machinery in eukaryotic cells. The ubiquitin–proteasome system is responsible for removing most abnormal peptides and short-lived cellular regulators, which, in turn, control many processes. This allows cells to respond rapidly to intracellular signals and changing environmental conditions. This review maps out the roles of the components of the ubiquitin–proteasome system with emphasis on areas where future research is urgently needed. We provide a flavour of the diverse aspects of plant lifecycle where the ubiquitin–proteasome system is implicated. We aim to highlight common themes using key examples that reiterate the importance of the ubiquitin–proteasome system to plants. The future challenge in plant biology is to define the targets for ubiquitination, their interactors and their molecular function within the regulatory context.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In the 1970s, initial work aimed at understanding protein turnover indicated that different proteins had widely varying cellular half-lives, which were, in some cases, altered by exogenous stimuli (Schimke, 1973). Development in the 1980s of a cell-free lysate from rabbit reticulocyte capable of selective protein degradation in the presence of ATP allowed dissection of the process, as described in studies by Hershko & Ciechanover (1998). This work led to the identification of ubiquitin (Ub) and the discovery that covalent modification of a substrate by Ub was critical for its degradation.
The subsequent purification of a protease capable of degrading multi-ubiquitinated substrates, the 26S proteasome, established the mechanistic framework for ubiquitin dependant proteolysis via the ‘ubiquitin–proteasome system’ (UPS; Hough et al., 1987). The fundamental role of the UPS in various cellular processes was first made clear by Ciechanover et al. (1985) in mammalian cells and yeast. Genome data reveals that the UPS is most elaborate in plants. Early genetic data has shown that ablations in UPS components affects critical processes in plants and in recent years it has become apparent that this is just the tip of the iceberg.
Ubiquitin is a small globular protein (76 residues) found in all eukaryotes. Its sequence is highly conserved and only three residues differ between yeast, human and plant species (Schimke, 1973; Callis et al., 1995). Ubiquitin is the prototypical member in an expanding family of proteins (ubiquitin like proteins – Ubls) that covalently modify target proteins to alter various aspects of their regulation (Hershko & Ciechanover, 1998; Jentsch & Pyrowolakis, 2000).
Ubiquitin assumes a compact structure with a five-strand mixed β-sheet forming a cavity into which a single α-helix fits diagonally to form a characteristic ‘Ub fold’ (Vierstra, 1996). Numerous intramolecular hydrogen bonds give Ub high stability, presumably to encourage recycling rather than proteolysis during the conjugation/degradation process. The flexible C-terminus of Ub protrudes from the Ub fold and terminates with an essential glycine. The carboxyl group of this glycine functions as an initiation site for the covalent attachment of Ub to substrates.
Ubiquitin gene family members (UBQs) are synthesized as either Ub polymers in which multiples (typically four to six in Arabidopsis) of the 228 bp coding region are concatenated head-to-tail or as one of three different fusion proteins (Callis et al., 1995). The Ub-fusion genes encode either one of two different ribosomal subunits or the Ubl Related to Ubiquitin (RUB)-1 protein fused to the C-terminus of Ub (Callis et al., 1990). In all cases Ub precursors are cleaved at the terminal glycine by deubiquitinating enzymes (DUBs) to release active monomers (Amerik & Hochstrasser, 2004).
Ubiquitin contains seven lysines (K6, K11, K27, K29, K31, K48 and K63). To target substrates for degradation by the proteasome, inter-ubiquitin linkages are made to K48 (i.e. G76-K48 iso-peptide bond) to form ubiquitin chains (poly-Ub; Pickart & Fushman, 2004). Poly-Ub chains of at least four ubiquitin moieties (tetra-ubiquitin) are required to provide an efficient proteasome delivery signal (Thrower et al., 2000).
Attachment of free Ub moieties to appropriate substrates proceeds by an ATP dependent E1 – E2 – E3 enzyme conjugation cascade (Fig. 1). The cascade starts with E1 (or Ub activating enzyme). E1 catalyses the formation of an acyl phosphoanhydride bond between the adenosine monophosphate (AMP) of ATP and the C-terminal glycine carboxyl group of Ub. Activated Ub then forms a stable intermediate by binding directly to an E1 cysteine via a thiolester linkage. This activated Ub is transferred from E1 to E2 (or Ub-conjugating enzyme) by transesterification. The E2-Ub intermediate delivers Ub onto a substrate acceptor lysine using an E3 (or Ub ligase). E3 enzymes impart substrate recognition to the process and either promote direct transfer of Ub to substrates from E2 or form a final E3-Ub intermediate before transfer. The end product is a Ub-protein conjugate containing an isopeptide bond between the C-terminal glycine of Ub and lysyl ε-amino group in the substrate.
After attachment of an initial Ub moiety to a substrate, additional Ubs are ligated to specific lysine residues on the first Ub to form poly-Ub chains. Whether chains are extended by preassembled Ubs or by iterative rounds of ligation using the E3 or additional accessory factors is currently unclear.
While linkages through all seven Ub lysines have been detected in vivo, poly-Ub chains linked through lysine 48 (K48) predominate in the cell and present a proteasome targeting/recognition signal (Pickart & Fushman, 2004). Upon delivery to the proteasome, ubiquitinated substrates have poly-Ub chains removed by deubiquitinating enzymes before unfolding, import and proteolysis (Hartmann-Petersen et al., 2003).
Although ubiquitin was first identified in the context of proteolysis (Hershko & Ciechanover, 1998), it has become increasingly clear that the addition of single Ub moieties (mono-ubiquitination; Hicke, 2001) or alternative Ub chain linkage configurations can confer diverse consequences on substrates (Pickart & Fushman, 2004).
The E1 enzymes initiate the Ub conjugation cascade. In plants it is a single polypeptide of c. 1100 residues that contains a positionally conserved cysteine to bind activated Ub and a nucleotide binding motif that interacts with either ATP or AMP-Ub intermediates (Hatfield et al., 1997). There are two E1 isoforms in Arabidopsis, one of which is localized in the nucleus (Hatfield et al., 1997).
The E2 enzymes contain a diagnostic 150 residue catalytic core that surrounds the active site cysteine. Using this conserved region, 37 E2 isoforms (UBCs) have been identified in the Arabidopsis genome (Vierstra, 1996). Outwith the core E2 domain, many detected isoforms contain various N- and C-terminal extensions that are proposed to influence target recognition and localization (Hamilton et al., 2001). The elaboration of E2s is presumed to ensure equality in the distribution of activated Ub to the vast array of E3s. Individual E2 isoforms in yeast and animals have distinct functions including: cell cycle regulation, DNA repair and degradation of endoplasmic reticulum (ER) -translocated proteins (Pickart, 2001). Sequence analysis has clustered Arabidopsis E2s into 12 distinct subfamilies (Vierstra, 1996). From coexpression and interaction data it appears that different E2 enzymes showed variable preference of E3 enzymes for interaction. While AtUBC8 was shown to interact with the most E3s, AtUBC34 showed the greatest E3 specificity but the majority of subtypes currently await functional classification. Ubiquitin E2 enzyme variant (UEV) proteins although similar to E2s in both sequence and structure, lack a catalytic cysteine residue and thus are unable to form a thiol-ester linkage with ubiquitin. Phylogenetic analysis revealed that there are eight Arabidopsis UEV genes belonging to three subfamilies (Zhang et al., 2007a,b).
Subunit classes can be broadly defined by subunit composition and mechanism of action (Fig. 2).
There are currently seven types of E3 ubiquitin ligases known and these can be subdivided into two basic groups dependent on the occurrence of either a ‘Homology to E6-AP C-Terminus’ (HECT) or ‘Really Interesting New Gene’ (RING)/U-box domain. RING-containing proteins can either ubiquitinate substrates independently or function as part of a multi-subunit complex which in plants includes Skp1-Cullin-F-box (SCF), VHL-ELONGIN-CUL2/5, Cullin 3 (CUL3)-Bric a brac, Tramtrack and Broad complex/Pox virus and Zinc finger (BTB/POZ), UV-Damaged DNA-Binding protein 1 (CUL4-DDB1) and Anaphase Promoting Complex (APC). RING/U-box E3s usually act as a molecular adaptor for the E2s and the substrates (Fig. 2).
The E3s all share a common requirement for a specific E2 interaction domain and a substrate recognition domain. Approximately 1406 genes representing substrate recognition modules of the seven types of E3 ligases have been identified in the Arabidopsis genome (Vierstra, 2009). The E3s are the most diverse proteins in the ubiquitination cascade in order to confer substrate selectivity for an extensive range of substrates.
A specialized case of substrate recognition relates to degradation by the ‘N-end rule’ where the half-life of a protein is influenced by the identity of its N-terminal residue (Varshavsky, 1996). Amino-terminal residues cluster by their capacity to reduce protein half-life and are termed N-degrons (Varshavsky, 1996), the prototype in the emerging knowledge of ubiquitin linked degradation signals. Specific ubiquitin ligases have been linked to the N-end rule, the best characterized of which in Arabidopsis are PROTEOLYSIS 1 (PRT1) and PRT6 (Stary et al., 2003). PRT1 contains two RING domains and targets aromatic amino acids at N-termini of proteins. In contrast, PRT6 RING domain protein ubiquitinates proteins with arginine at the N-terminus (Garzón et al., 2007).
HECT E3s are single subunit ligases with a diagnostic 350-residue region termed the HECT domain, which was first detected in the founding member, human E6AP (Huibregtse et al., 1995). HECT E3s are unique as they form a thiol-ester intermediate E3-Ub on a conserved cysteine during Ub transfer. The N-terminal region of the HECT domain forms a stable binding pocket for the E2-Ub intermediate and the C-terminal region contains the active site cysteine (Huibregtse et al., 1995). A variety of protein–protein interaction domains upstream of the HECT domain are thought to participate in substrate recognition and localization (Smalle & Vierstra, 2004). Genome analysis has identified five HECT E3s in yeast, over 50 in humans and seven in Arabidopsis (UPL1–7; Vierstra, 2009). Arabidopsis ULP3 has been implicated as a key regulator of trichome development (Downes et al., 2003) along with UPL5, which targets WRKY53 for proteasomal degradation during leaf senescence in Arabidopsis (Miao & Zentgraf, 2010).
Multisubunit E3s are based on CULLIN and RING finger components. The RING component is involved in E2 and adaptor molecule recognition while the adaptor molecule is involved in substrate specificity. This modular design allows for greater substrate scope as different adaptors can interact with different RING fingers. The two modules associate through CULLIN, which acts as a molecular assembly platform.
SCF E3s are heterotetrameric ubiquitin ligases with subunits named after those of the founding member: SKP1, CDC53 (or CUL1/Cullin) and an F-box protein (Deshaies, 1999). A fourth subunit, RBX was subsequently discovered and found to contain a RING H2-type domain (Fig. 4c). The architecture of the SCF complex divides E2–Ub interaction, substrate recognition and complex assembly between its different subunits. The RBX subunit interacts with E2-Ub via its RING domain and as part of the Cullin-RBX-SKP1 subcomplex confers Ub transferase activity (Deshaies, 1999). Substrate specificity is provided by the F-box subunit which is anchored to SKP1 via an N-terminal F-box motif (Kipreos & Pagano, 2000) and target proteins through C-terminal protein–protein interactions motifs. F-box proteins constitute the largest single protein family in Arabidopsis (nearly 700 members; Gagne et al., 2002) with wide target specificity conferred by various C-terminal substrate recognition domains (including leucine-rich repeats, kelch, lectin binding, Armadillo, tetratricopeptide repeats, Jumonji-C and the DEAD box; Gagne et al., 2002). In many cases, substrate phosphorylation is known to be a prerequisite for recognition by F-box proteins, potentially implicating many plant kinases in the regulation of proteolysis (Deshaies, 1999). Plant SCF components show a greater degree of divergence than their mammalian counterparts (Vierstra, 2009), in Arabidopsis the diversity of F-box proteins coupled with Cullin 1, two RBX1 subunits and 21 possible SKPs (termed ASKs in Arabidopsis) could potentially assort in over 100 000 distinct SCF complexes (Smalle & Vierstra, 2004). In animals, the SCF E3 serves as a prototype for SCF-like complexes where alternative interaction partners for different Cullin proteins (CUL2, 3a, 3b and 4) have been shown to generate novel E3s (e.g. VBC and VBL; Stebbins et al., 1999).
CUL3-BTB E3 ligases are analogous to SCF E3 ligases and are based on three components: CUL3, BTB/POZ and an RBX1 domain. In the current model, BTB confers substrate specificity, CUL3 is again the assembly platform and RBX1 recruits the E2 for Ub loading. The BTB/POZ domain appears to be acting akin to the ASK1 and F-box components of SCF E3s.
The BTB/POZ domain has many functional roles including: transcriptional regulation, cytoskeleton dynamics, ion channel assembly and gating and targeting proteins for ubiquitination. In most of these roles the BTB domain mediates interactions between itself and other proteins.
In Arabidopsis 81 loci are implicated to encode BTB/POZ domains, with most genes not arising from tandem duplications. Majority of the BTB/POZ proteins contain additional upstream or downstream domains acting as protein–protein interaction motifs such as ARM, Ankyrin, Meprin And TRAF Homologous domain (MATH) and TetratricoPeptide Repeat (TPR) and DNA binding domains such as Myb repeats, Transcriptional Adaptor putative Zinc finger (TAZ), as well as the plant-specific Nonphototrophic Hypocotil domain (NPH3) and the potassium channel tetramerisation domain.
CUL4 is the molecular scaffold for a new class of E3s involved in mammalian and possibly plant DNA replication and transcription. These contain RBX1 and DDB1 proteins along with CUL4. Despite the massive conservation of CUL4 from yeast to plants and mammals CUL4 has only recently been linked to E3 ligase activity in Arabidopsis. The identification of plant CUL4-DDB1 subunits came from human orthologue data and identified two highly related genes DDB1a and DDB1b, CUL4-DDB1 is thought to be sufficient to recruit substrates but may sometimes require additional factors, including DET1, COP10, COP1 and DDB2. Many of these additional proteins actors contain a DWD motif which functions to interact with DDB1 (Lee et al., 2010).
The Anaphase Promoting Complex (APC) is the most elaborate E3 ligase known, consisting of 11 subunits (APC1–11). The APC was first identified in a yeast screen for mutants unable to degrade the mitotic cyclin Clb2 (Wäsch & Cross, 2002). Subsequently, the role of APC in degrading other crucial cell cycle regulators was discovered (Capron et al., 2003) and the name cyclosome was assigned to the complex. Arabidopsis orthologues have been detected for most APC subunits (Capron et al., 2003), which are predominantly present in single copy suggesting that a limited number of APC isoforms are assembled. The APC subunits APC2 and APC11 are related to SCF subunits CUL1 and RBX1 respectively. These subunits are presumed to have analogous structural (CUL1) and E2-Ub transferase (RBX1) roles in APC complex.
The remaining E3s interact with E2-Ub intermediates using variants of a zinc-finger structure termed the RING domain (Kosarev et al., 2002). The RING finger motif consists of four ligand pairs (either histidine or cysteine), which coordinate two zinc ions in a spatially conserved arrangement (Freemont, 2000). The RING finger assumes a cross-braced structure formed by the octet of zinc-binding histidines and cysteines. This domain is essential for ubiquitin ligase function and the two main RING domains differ at position five for the metal ligand histidine or cysteine, C3H2C3 (RING-H2) and C3H1C4 (RING-HC) respectively. Together with F-box proteins the RING-finger proteins are the most abundant E3 ligase gene family in plants.
Systematic searches have yielded 469 RING containing proteins and identified three RING-type (RING-H2, RING-Hca and RING-Hcb) and five modified RING-types (RING-C2, RING-v, RING-D, RING-S/T and RING-G; Stone et al., 2005). Categories are based on the relative spacing and nature of the metal ligands. Most sequences contain only the RING domain (c. 150 predicted proteins) or RING domain with an associated transmembrane domain (c. 120 predicted proteins). In the remaining RING proteins additional sequences are believed to be implicated in protein interaction or regulatory elements. Associated protein interaction domains include: WD40, Ankyrin, coiled-coil and zinc-finger motifs.
The U-box domain is related to the RING domain but lacks the zinc-chelating residues. From molecular models it is predicted that a structure similar to the RING domain is achieved without the ionic association using hydrogen bonds and salt bridges (Ohi et al., 2003). About 130 U-box proteins have been identified to date in the Arabidopsis genome (Kosarev et al., 2002; Mudgil et al., 2004).
The ligation of Ub to substrates is a reversible process and all known peptide linkages made from Ub moieties are efficiently cleaved by deubiquitinating enzymes (DUBs). Following the identification of various classes of DUBs it is thought that Ub removal is a dynamic process with proposed constitutive and regulated DUB activities in the cell (Amerik & Hochstrasser, 2004). The DUB enzymes perform several important functions in the Ub–proteasome pathway (Fig. 3).
To ensure normal rates of targeted proteolysis DUB enzymes maintain a sufficient pool of free ubiquitin in the cell. To achieve this, DUBs function to process precursor ubiquitin from translation products and recycle poly-Ub chains bound to proteasome regulatory particle (RP). In all cases, DUB enzymes process ubiquitin precursors via cleavage of a typical peptide bond (Amerik & Hochstrasser, 2004).
In yeast, animal and plant proteasome there are at least two associated DUB activities that mediate poly-Ub disassembly and recycling. These DUBs hydrolyse K48 linkages (ε-amino isopeptide bonds) to prevent the accumulation of ‘free’ ubiquitin chains at binding sites of the proteasome RP (Amerik & Hochstrasser, 2004).
Following activation, ubiquitin is susceptible to attack by abundant intracellular nucleophiles such as glutathione and polyamine. To prevent loss of activated Ub through such pathways DUBs function to prevent titration by these compounds (Amerik & Hochstrasser, 2004).
Beyond roles in basic Ub metabolism, DUBs also serve to negatively regulate protein degradation. The commitment of substrates to proteasomal degradation by ubiquitination can be reversed by DUBs, by altering the half-life of specific targets in response to signalling events (Amerik & Hochstrasser, 2004). More generally, DUB enzymes have been proposed to function as a final proof-reading mechanism for degradation, rescuing proteins that are inappropriately targeted to the proteasome (Lam et al., 1997). The distinct metabolic and substrate-specific roles performed by different DUBs remains to be clarified and the known DUB target repertoire should be the focus of future research.
The 26S proteasome is a 2 MDa ATP-dependent proteolysis complex that degrades ubiquitin tagged substrates. While initial characterization of the complex was derived from studies of yeast and mammalian proteasomes, subsequent studies in rice and Arabidopsis indicate a similar design (Fu et al., 1998).
The 26S proteasome comprises 31 subunits divided into two subcomplexes: the 20S core protease (CP) and 19S RP. The CP functions as a nonspecific ATP and Ub-independent protease which assumes a cylindrical structure by the assembly of four heptameric rings. The peripheral rings are composed of seven related α-subunits and the central rings are composed of seven related β-subunits in a α1–7 β1–7 β1–7 α1–7 configuration (Wolf & Hilt, 2004; Fig. 4a). Initial crystallography studies of the CP in yeast indicated a large central chamber, facing into which are protease active sites contributed by the β1, β2 and β5 subunits (Wolf & Hilt, 2004).
These three proteases generate peptidylglutamyl, trypsin-like and chymotrypsin-like activities, imparting the capacity to cleave most peptide bonds (Wolf & Hilt, 2004). Entry into the CP chamber is restricted by a pore formed at the periphery by the seven α-subunits. The narrow pore requires entering proteins to be unfolded and flexible extensions in each α-subunit provide a channel gating function to control substrate entry (Hartmann-Petersen et al., 2003). Core protease entry channel gating and the requisite unfolding of substrates provide a demarcation between protease activity and the cellular milieu. In this way, degradation of proteins is limited to those unfolded and imported into the proteasome.
The RP associates with either end of the CP and confers ATP dependence and poly-Ub recognition to the proteasome. The RP is composed of 17 subunits which form two subcomplexes termed Lid and Base. The Base sits directly over the CP α-ring channel and comprises a ring of six related AAA-ATPases (RPT1–6) and three nonATPase subunits (RPN1, 2 and 10). The Lid interacts with the Base via RPN10 (Fu et al., 1998) and contains the remaining nonATPase subunits (RPN 3, 5–9 and 11–12; Fig. 4a). The overall structure–function relationships between RP subunits remain to be clarified, but key functions have been ascribed to individual subunits (Hartmann-Petersen et al., 2003).
Cooperatively the RP Base and Lid mediate recognition of K48 linked poly-Ub chains, removal of covalently bound Ub moieties, unfolding of targeted substrates, pore gating and substrate import to the proteaseome (Fig. 4b). K48 poly-Ub recognition by RPN10 has been observed but is nonessential in yeast and Arabidopsis, suggesting that it is not the sole poly-Ub-binding determinant (Hartmann-Petersen & Gordon, 2004). RPN11 is a zinc metalloprotease with poly-Ub hydrolysis (deubiquitination) activity that disassembles/recycles Ub chains during target degradation (Verma et al., 2002). ATPase subunits in the Base (RPT1–6) contact the CP pore gating α-subunits and are presumed to facilitate substrate unfolding and pore opening.
Other RP subunits are postulated to function as receptors for poly-Ub carrier proteins and specific E3 ligase complexes. Ongoing analyses of substrate delivery and recognition by the proteasome suggests the existence of distinct receptors for different substrate types (Hartmann-Petersen & Gordon, 2004). In Arabidopsis the majority of proteasome subunits are encoded by genes in duplicate with corresponding isoforms detected in 26S proteasome preparations (Yang et al., 2004).
Two evolutionary relatives of the Lid complex in the proteaseome RP have been identified in plants and animals. These complexes (COP9 and eIF3) contain eight subunits synonymous to those in the proteasome Lid: COP9, termed the signalosome (CSN), assists in numerous eukaryotic signalling pathways (Wei et al., 1998) whereas eIF3 is involved in translational control. Experimental evidence indicates that both COP9 and eIF3 can associate with the proteasome CP to create functionally distinct particles (Dunand-Sauthier et al., 2002).
In recent years, genetic and cell biological approaches have revealed the role of the UPS in nearly all aspects of plant homeostasis, including cell division (Capron et al., 2003), plant development (Samach et al., 1999), responses to plant hormones (Sullivan et al., 2003) and signalling in response to abiotic and biotic stimuli (Smalle & Vierstra, 2004). This section discusses the variety of cellular processes regulated by the UPS.
Coordinating external and internal cues with developmental physiology requires a rapid and mobile mechanism that can be highly regulated in space and time in every cell. The mobile nature of plant hormones and the mechanistic potency of the UPS fit this purpose. It is therefore not surprising that many of the signalling pathways associated with hormones rely on the UPS-mediated degradation of downstream targets to elicit their effect. Examples described later underline the intimate relationship between the UPS and plant growth and development mediated by phytohormones.
Mutants of components of the multisubunit E3 ligases show broad pleiotropic effects on development. Although the ASK component of SCF E3 ligases contain 21 family members, ask1/ask2 double mutants show severe developmental defects in all stages of plant development (Liu et al., 2004), with a similar effect seen in mutants of the RING component RBX1 (Gray et al., 2002). In addition, the nuclear localized CUL1 component of SCFs is essential for embryogenesis, while CUL3A CUL3B act redundantly in embryo pattern formation and endosperm development (Thomann et al., 2005).
Gibberellins comprise a family of diterpene hormones and are the prototypical growth promoting hormones in plants, regulating a large number of growth related responses (de Lucas et al., 2008). The response to GA is elicited through targeted degradation of DELLA growth repressors, mediated by GID1 which binds to the DELLA repressors in a GA-dependent manner resulting in a conformational change in the DELLA protein and degradation by the 26S proteasome via SCF(SLY) (Murase et al., 2008). Jasmonates (JA) play a role in regulation of development and defence, and notably in development by responding to light-coordinated daily growth cycles and shade-avoidance behaviours (Kazan & Manners, 2011). Jasmonate, in a similar manner to GA, leads to the degradation of the jasmonate ZIM-domain (JAZ) family of proteins. The F-box component COI1 (Xie et al., 1998) binds to a JA–amino acid conjugate (Thines et al., 2007) and once bound, the SCF(COI1) complex directs the degradation of the JAZ repressors. Auxin signalling follows a similar pattern with SCF(TIR1/AFB) (Gray et al., 2001; Dharmasiri et al., 2005b) degrading the transcriptional repressor AUX/IAA family of proteins in the presence of auxin, with the F-box again being the auxin receptor (Dharmasiri et al., 2005a; Fig. 5).
Ethylene signal transduction is also regulated by the ubiquitin proteasome system, but instead of the hormone leading to degradation of transcriptional repressors, ethylene leads to the stabilization of the positive transcriptional regulators EIN3 and EIL1 by inhibiting the activity of SCF(EBF1/2) (Gagne et al., 2004) by degrading the F-box components EBF1 and EBF2, although the E3 ligase(s) responsible have yet to be confirmed (An et al., 2010). In addition to the direct involvement of E3s in hormone perception, ethylene production is regulated by the BTB protein Ethylene-Overproducer 1 (ETO1) which targets ACS5 – an enzyme acting at the rate-limiting step in ethylene production (Christians et al., 2009).
From the hormones mentioned so far there is a consistent paradigm of direct involvement of E3 ligases in hormone perception. So far, in contrast to this trend, perception of the other major plant hormone, ABA, does not rely on an E3 ligase, rather, perception is via a range of receptors (Raab et al., 2009). See the section on Abiotic stress for further discussion of ABA.
Downstream of ABA, the UPS is emerging as a complex regulator of release from dormancy and germination mostly through a diverse range of E3 ligases. The germination-related E3 ligases also appear to have a significant overlap with early development processes. The SCF component, ASK1 is important for ABA signalling as overexpression of a wheat homologue of ASK1 in Arabidopsis shows hypersensitive ABA responses in germination and growth (Li et al., 2011a); thus it is likely that ABA regulates physiological processes through altering the stability of various proteins. While the influence of ASK1 is far ranging, recently many E3 ligase components involved in ABA-mediated germination have been discovered. The RING E3 ligases AIRP1 and AIRP2 (Cho et al., 2011) and the F-box protein EDL3 (Koops et al., 2011), are responsible for reducing root growth rate in response to ABA but show divergence in other developmental processes. AIRP1 and AIRP2 are responsible for regulating stomatal closure and H2O2 production in response to ABA while ELD3 plays a vital role in early development. EDL3 transcripts are induced by ABA and stress conditions and in turn EDL3 causes chlorophyll production in etiolated plants and inhibits germination. CULBPM E3 ligases have been shown to reduce plant sensitivity under stress conditions by degrading the ABA-induced transcription factor HB6 (Lechner et al., 2011). The RING E3s RHA2a, RHA2b (Li et al., 2011b) and the BRIZ1/BRIZ2 dimer (Hsia & Callis, 2010) and the ARM E3s PUB18, PUB19 (Bergler & Hoth, 2011), PUB43 and PUB44 (Salt et al., 2011) are also responsible for inhibition of germination in response to ABA.
Post-translational regulation of protein stability is emerging as a vital mechanism of regulation in the Arabidopsis circadian clock. As the protein components of the clock are in a continual state of flux, there is a need for these components to be promptly removed once transcription has ceased. The F-box proteins ZTL (Han et al., 2004), FKF1 and LKP2 play a vital role in circadian regulation and share a functional overlap (Baudry et al., 2010). These F-box proteins all contain a distinguishing Light–Oxygen–Voltage domain that allows them to detect blue light (Kiyosue & Wada, 2000; Imaizumi et al., 2003) and translate this into a response by inhibiting their ability to ubiqutinate targets. ZTL, FKF1 and LKP2 appear to act redundantly as single mutants show only limited disruption to the circadian rhythm while triple knockout shows severe disruption (Baudry et al., 2010). As part of an SCF complex, ZTL causes the degradation of the clock component TOC1, beginning in the evening and resulting a gradual decrease over the night (Más et al., 2003). ZTL also degrades the regulator PRR5 in a stringent dark-dependent manner (Kiba et al., 2007). PRR5, in turn, has regulatory effects on the circadian clock by influencing TOC1 phosphorylation and localization (Wang et al., 2010) as well regulating flowering time (Nakamichi et al., 2007; Fig. 6).
In another branch of the circadian clock, the single subunit RING E3 ligase, SINAT5, facilitates the degradation of LHY and appears to play an integral role in flowering. In addition, SINAT5 protein levels are induced by the hormone auxin, so it serves as a branch to integrate hormonal signals into the circadian clock (Xie et al., 2002). Another protein, DET1, adds an additional layer of complexity to this relationship by protecting LHY from degradation by SINAT5 (Park et al., 2010); however, the role of DET1 is not fully understood. The interplay between these three proteins appears to be important in determining flowering time as both sinat5 (Park et al., 2010) mutants and LHY (Mizoguchi et al., 2002) over-expressers show delayed flowering.
Floral development is a complex process governed by complex regulation that has attracted much research over the last few decades. Ubiquitination appears to play a role in most if not all phases of floral development. The first phase of floral development involves determination of flowering time, which depends on a large number of factors, including input from the circadian clock via the floral regulator CONSTANS (CO). The cyclic peaks in CO appear to determine transition into flowering depending on daylength. The RING-like E3 DAY NEUTRAL FLOWERING (DNF) plays a role in regulating the sensitivity of the flowering response to daylength. Although direct evidence is lacking, it seems likely that CO is a target for DNF-mediated degradation (Morris et al., 2010). The stability of CO is also regulated by blue light via the COP1, with blue light preventing the ubiquitination by COP1. The cryptochrome CRY2 is the detector of blue light, entering an excited state in its presence. The excited CRY2 protein then binds to SPA1, an interactor of COP1 and blocks the action COP1 (Zuo et al., 2011). COP1 also targets photomorphogenic transcription factors, including HY5, HYH and LAF, for degradation (Seo et al., 2003). Thus COP1 is a hub for integrating light and circadian signals to direct initiation of photomorphogenesis at an optimal developmental stage. It appears that the light/dark localization of COP1 is regulated by the E2 COP10 through nonK48-linked polyubiquitin as well as by derubylation by the CSN that is emerging as a regulator of SCF E3 ligase activity (Smalle & Vierstra, 2004).
After ignition, development of meristem identity is initiated by the homoeotic regulator LFY, which leads to the development of floral organs. An interesting regulator of LFY has been identified, the F-box UFO. While UFO has typical E3 domains it also has DNA-binding domains and, with LFY, regulates AP3 expression, which plays a role in correct floral organ assignment. It may also regulate LFY through targeted degradation, although no direct evidence for this remains has been shown (Chae et al., 2008).
The correct development of male structures in plants depends on the action of SCFSAF, which leads to the essential thinning of the endothelial secondary wall (Kim et al., 2012). In pollen the F-box FBL17 plays a role in regulating the development of sperm by directing second mitosis of the generative microspore during pollen development. The targets of SCFFBL17 are not known, but it has been suggested that the second mitosis is achieved through degradation of cell cycle components (Gusti et al., 2009), which remains to be tested. In addition, self incompatibility in Brassica napus has been shown to be regulated by the action of the U-box E3 ARC1, which is activated by self-incompatible pollen, leading to an increase in ubiquitinated proteins in the pistil (Stone et al., 2003).
Plants face constant stresses from the abiotic conditions of their environment. These stresses come from a variety of factors, encompassing all aspects of their life cycle, with a wide range of possible effects. Changes to the physical conditions, which cause the stress, can be just for the short term, such as the differences in light and temperature intensity in normal diurnal cycles or, alternatively, they can take place or be sustained over a much longer time, as seen with seasonal effects or long-term climate changes. Abiotic stresses can also be very quick to develop, such as mechanical damage and be more troublesome for the plant to predict, such as prolonged periods of drought, or anywhere in-between. As such, it is clear that all plants must possess very diverse systems to combat all forms of abiotic stress. Some will need to be flicked on and off like a switch, while others will need to lead to long-term cellular changes. However, one constant among all of these defensive strategies is the presence of the ubiquitin system.
There are a large number of examples of the UPS affecting plant's ability to cope with abiotic stresses. These can vary across numerous signalling pathways involved in responding to all types of abiotic stresses. Drought, cold and salt are three very closely related stresses, which are common and widespread among plants in many environments. Here we will focus mainly on the action of the UPS in these areas.
Abscisic acid-mediated signalling is involved in all three stresses and allows us to map out the role of the UPS in perceiving and responding to drought, cold and salt stress via this signalling pathway. Perception of these environmental stresses can lead to the biosynthesis of ABA, which, in turn, can result in changes to the expression of hundreds of ABA-regulated genes. The promoter region of these genes contains the ABA-regulatory element (ABRE), to which certain ABA-regulated transcription factors bind (Hattori et al., 2002). Abscisic acid acts in a variety of different cell types to produce differing actions. Therefore, it is possible that there are a number of different mechanisms by which it can function. Nonetheless, the ubiquitin proteasome system is still key to controlling this pathway and bringing about abiotic stress responses.
One gene, thought to be involved in this process, is the Arabidopsis RING E3 ligase XERICO, which is upregulated in response to osmotic stress. Transgenic over-expressers were hypersensitive to osmotic stress, as well as ABA treatment and produced significantly greater amounts of ABA in the cells (Ko et al., 2006). These results suggest that XERICO is involved in the biosynthesis of ABA in response to stress. However, the major level of ubiquitin-related control in this system appears to involve the regulation of stability of the ABRE-binding transcription factors. One such transcription factor is Abscisic acid insensitive 3 (ABI3), which is regulated by the ABI3-interacting E3 ligase AIP2. Abscisic acid has been shown to induce expression of AIP2, therefore leading to a reduction in the presence of ABI3.
As mentioned previously, the ABA-induced transcription factor HB6, a negative regulator of ABA responses (Himmelbach et al., 2002), is also controlled by the UPS. The concentrations of HB6 protein increase in response to a reduction in expression of CUL3BPM E3 ligases, which have been shown to target the transcription factor for proteasomal degradation (Lechner et al., 2011). The stability of another such transcription factor, ABI5, the accumulation of which is induced by osmotic stress, is controlled by E3 ligases in an ABA-dependent manner (Lopez-Molina et al., 2001). While the mechanism for this is less well known, it has been shown that the E3 ligase KEG (Liu & Stone, 2010) as well as the E3 complex of DWA1 and DWA2 (Lee et al., 2010) are important in ABI5 regulation. The transcription factor is shown to accumulate at higher levels when either is missing, and is shown to interact directly with both, suggesting that it is targeted for proteasomal breakdown.
Drought, cold and salt stresses can all be harmful to the plant by causing cellular desiccation, and plants combat this particular aspect of all three stresses in a similar way. This is mainly achieved through ABA- and UPS-mediated signalling. It seems, therefore, that a key aspect of ABA in stress responses is in tying together of all these stresses, while also producing individual responses.
Perception of each of these stresses generates the same response of ABA, and yet not all ABA genes are upregulated in response to all stresses, suggesting that it is the action of the ABA that is producing different results. This could be related to the possible different mechanisms for ABA action, which was suggested earlier. The availability of several modes of action would allow production of the single hormone to result in different responses (Fig. 7).
While the ABA pathways mentioned play an important role in coping with cold, drought and salt stress, the plant must also rely on many ABA-independent pathways. These are also reliant on the UPS as a control system. One such example is HOS1, which was originally identified by a genetic screen (Ishitani et al., 1998) as a negative regulator of the CBF cold-response pathway, and mutation of the gene produced increased levels of a number of cold-responsive genes. More recently, HOS1 was shown to be a functional ubiquitin ligase enzyme (Dong et al., 2006). Yeast two-hybrid assays also showed a probable interaction between HOS1 and ICE1, a transcription factor, which also regulates cold responsive genes in the CBF pathway (Chinnusamy et al., 2003). Dong et al. (2006) went on to show that ICE1 could be targeted for ubiquitination by HOS1. hos1 mutants not only prevented this ubiquitination but also produced higher expression of CBF pathway components downstream of ICE1.
DREB2A is a transcription factor that binds to the Drought Response Element (DRE), a motif present in the promoter of many cold- and drought-responsive genes (Chinnusamy et al., 2003). DREB2A is thought to be proteasomally degraded by ubiquitination by DREB2A Interacting Protein (DRIP) 1 and 2, both of which are RING E3 ligases (Qin et al., 2008). DRIP double knockouts plants show increased drought tolerance, brought about by an increase in production of DREB2A-regulated stress response genes.
Another functional RING E3 ligase, Salt and Drought Induced Ring finger (SDIR1) has been shown to be involved in the response to salt, specifically in the prevention of germination under high salt conditions, although the mechanism behind this is not known (Zhang et al., 2007a,b). Interestingly, SDIR1 is also involved in ABA signalling, acting as a positive regulator of ABI5, suggesting a possibility of crossover between the two pathways.
This evidence does show clear differences in the signals used for these slightly differing stresses, which is to be expected as they do require different action, and so the plant must be able to respond to each individually. However, once again the systems are linked to some extent, both by the shared use of the DRE and, by the presence of the UPS. The intricate control possible with ubiquitination appears to be key in the crosslinking of these responses, while also keeping them, to some extent, separate and distinguishable (Fig. 7).
Without a circulatory system to carry dedicated immune surveillance cells, plants have evolved a global single-cell immune system (Spoel & Dong, 2012). Detection through ‘pattern recognition receptors’ (PRRs) of widely conserved pathogen associated molecular patterns (PAMPs) induces ‘PAMP triggered immunity’ (PTI). Whereas strain-specific recognition of effector molecules (virulence factors), or their effects on the host, leads to effector triggered immunity (ETI; Jones & Dangl, 2006). These recognition events activate defence responses such as a burst of reactive oxygen species (ROS) and hypersensitive response associated programmed cell death (HR PCD), acting to limit or inhibit pathogen spread through the host. The plant UPS is heavily implicated in plant immunity with a plethora of studies identifying ubiquitin ligases that are upregulated in response to PAMP elicitors, effector treatments or indeed pathogens themselves. These genetic upregulations are just a fraction of the vast transcriptional reprogramming that occurs in response to challenge by pathogens (reviewed by Katagiri, 2004) but indicates a clear link between the UPS and host–pathogen interactions. A simple survey of the data landscape on plant–pathogen interaction indicates that nearly every component of the ubiquitin system is implicated in plant immunity (Fig. 8). We highlight a few below with special emphasis on E3 ligases.
A triple knockout of Arabidopsis Plant U-Box genes Pub22, Pub23, and Pub24 results in enhanced resistance to bacterial and oomycete pathogens. Oxidative burst elicited by flg22 and chitin was shown to accumulate to a much greater extent in the triple knockout and enzymes involved in ROS production were also found to be upregulated. These results suggest these three U-box E3 ligases act redundantly as negative regulators of PTI in response to flg22 and chitin PAMPs (Trujillo et al., 2008).
The perception of bacterial flagellin by FLAGELLIN SENSING 2 (FLS2), a transmembrane PRR implicates two essential Plant U-Box E3 enzymes, PUB12 and PUB13. These are recruited and phosphorylated by BRI1-ASSOCIATED RECEPTOR KINASE 1 (BAK1) and this activity appears to be enhanced by another kinase BOTRYTIS-INDUCED KINASE1 (BIK1). BAK1–PUB12–PUB13 protein complex associates with FLS2 upon its binding of the conserved flagellin peptide flg22. This interaction results in PUB12- and PUB13-mediated polyubiquitination of FLS2 and its apparent internalization into the plant cytosol (Fig. 8; Robatzek, 2006; Lu et al., 2011).
ACRE74, ACRE276 and ACRE189 are E3 ligases identified as positive regulators of HR-PCD using the heterologous tobacco Cf-9/Avr9 experimental system (González-Lamothe et al., 2006; Yang et al., 2006; van den Burg et al., 2008). Silencing these E3 ligases, through virus-induced gene silencing (VIG) in tomato, reduces HR PCD induced by avirulent fungal strains of Cladosporium fulvum, leading to susceptibility.
The rice SPL11 lesion mimic mutant produces a spontaneous cell death phenotype that enhances defense against avirulent and virulent bacterial and fungal pathogens. Spl11 encodes a plant U-Box E3 with in vitro ligase activity (Zeng et al., 2004). SPL11 has been shown to interact with SPL11-interacting protein1 (SPIN1) a Signal Transduction and Activation of RNA (STAR) protein involved in regulation of flowering time. The Arabidopsis SPL11 homologue has also been shown to have a role in development underlining the apparent parallels between plant immunity and the regulation of plant development (González-Lamothe et al., 2006; Yang et al., 2006; van den Burg et al., 2008; Vega-Sánchez et al., 2008; Shikata et al., 2009).
Arabidopsis Botrytis Susceptible1 Interactor (BOI) encodes a RING E3 ligase, which interacts with the transcription factor MYB108. BOI is capable of ubiquitinating MYB108 in vitro. Knockdown boi lines displayed increased susceptibility to Botrytis cinerea. Overexpressing and RNAi lines showed no alteration in disease susceptibility to avirulent and virulent Pseudomonas strains. Using electrolyte leakage as a proxy for cell death the authors found elevated levels of cell death in virulent pathogen interactions whereas avirulent-induced HR PCD was unchanged (Lin et al., 2008; Luo et al., 2010). This could suggest that BOI is manipulated by the pathogen to degrade MYB108 in compatible interactions and perturbing BOI levels is not significant enough to disrupt this activity.
These studies clearly point to the E3 ligase group as important players in plant–pathogen interactions, This is not surprising considering the presence of a large number of Plant U-box and RING E3 encoding genes in the plant genome; however, relatively few F-boxs proteins have been implicated in plant stress responses even though they form the largest subclass within the E3s.
The messages carried by hormones are a vital part of plant immune signalling. Salicylic acid (SA), JAs and ethylene (ET) have established roles in plant defence, which include crosstalk between these pathways (Leon-Reyes et al., 2009; Salt et al., 2011). Recent findings have also implicated the classical growth and development hormones auxin and gibberellins as well as the brassinosteroids, demonstrating the integration of multiple signalling pathways (hormones in immunity reviewed by Pieterse et al. (2009). We highlight the role of SA as a typical example of how the UPS is interlinked with immunity.
Salicylic acid is a key signaller in systemic-acquired resistance (SAR; reviewed by Durrant & Dong, 2004). NONEXPRESSOR OF PR GENES1 (NPR1) is a coactivator of pathogenesis-related (PR) gene expression and is essential to SAR transduction. NPR1 contains a BTB protein–protein interaction domain that mediates formation of a CRL ubiquitin ligase. NPR1 E3 mediates the continual turnover of itself in the nucleus by the UPS (Spoel et al., 2009). NPR1 interacts with TGA2 a transcriptional activator of SA-responsive genes in SAR. Binding of NPR1 to TGA2 by its repressor domain is thought to prevent TGA2 oligomers and thus promote gene expression in the SA pathway (Boyle et al., 2009).
Systemic-acquired resistance compromised npr1 mutants have been exploited in a number of studies looking for mutants capable of recovering resistance. Li et al. (2009) discovered a suppressor of npr1-1 susceptibility, snc1, which constitutively activated defence-related responses. snc1 encodes an R protein that activates defence entirely independent of SA signalling (Zhang et al., 2003). A mutant screen for suppressors of the double mutants enhanced resistance, identified 11 MOS (modifier of snc1) components involved in this immune pathway (Monaghan et al., 2009). These include the Arabidopsis E1 enzyme UBA1 and a protein complex that is conserved in animals, although it has no known role in immunity and associates with two Plant U-box E3s (Goritschnig et al., 2007; Palma et al., 2007; Monaghan et al., 2009).
Similarly Kim & Delaney (2002) have isolated the Arabidopsis suppressor of Nim 1-1 (SON1) from a mutant screen for silencers of noninducible immunity1 NIM1 (NPR1). son1 encodes a protein containing an F-box domain. The son1 mutant similarly displays SAR-independent, constitutive resistance against both the virulent oomycete Hyaloperonospora arabidopsidis and the bacterial pathogen Pseudomonas syringae.
Screening of mutants sensitive to a putative SA precursor led to the isolation of BAH1 benzoic acid (BA) hypersensitive. Bah1 encodes a RING E3 ligase previously implicated in nitrogen limitation adaptation (Peng et al., 2007; Yaeno & Iba, 2008). The Bah1 mutant accumulated SA under BA treatment and Pseudomonas inoculation. The increase in SA was accompanied by higher resistance against P. syringae DC3000 (Yaeno & Iba, 2008).
In order to combat the immune system of host plants pathogens secrete effector molecules. The dual roles of UPS components in defence and development make them ideal targets for exploitation during infection. A number of effectors have been found to hijack the host UPS and steer it toward the degradation of host defence molecules.
Pseudomonas syringae injects effectors into host cells through Type III secretion, these include a hormone mimic (coronatine), a protease (AvrRpt2) a proteasomal inhibitor (syringolin A) and a U-box-like E3 ligase (AvrPtoB; Kim et al., 2005; Rosebrock et al., 2007; Groll et al., 2008; Katsir et al., 2008). AvrPto targets host kinases, including FLS2 and FEN, which mediates activation of basal defences through association with Prf R protein, leading to its degradation (Xiang et al., 2008). Pto, a host kinase involved in R gene AvrPto-mediated resistance has been shown to evade AvrPto ubiqitination activity via phosphorylation and hence inactivation of AvrPto E3 ligase activity (Ntoukakis et al., 2009). Coronatine mimics the active form of JA, binding to COI F-box causing repressor degradation and activation of the JA pathway antagonizing SA-mediated defence (Katsir et al., 2008). RAR1, required for RPM1-mediated ETI, is targeted by P. syringae effector AvrB in susceptible Arabidopsis leading to suppression of flg22 PAMP-triggered immunity (Shang et al., 2006).
A number of phytopathogens possess F-box effectors such as coronatine, capable of forming SCF complexes with host subunits to form a functional CRL E3 ligase. Agrobacterium tumefaciens VirF F-box interacts with the plant SKP1 proteins to degrade VIP1 and VirE2 for t-DNA integration into the host genome. The proteasomal inhibitor MG132 hinders the transformation process (Tzfira et al., 2004). The Ralstonia solanacearum effector GALA contains an F-box like domain that has been shown to interact with Arabidopsis SKP1-like ASK proteins, although activity and host targets are, as yet, undetermined (Angot et al., 2006). Similarly polerovirus PO F-box effector interacts with ASK1,2 and is essential to pathogenicity, blocking host post-transcriptional gene silencing (PTGS) defence (Pazhouhandeh et al., 2006). The virally hijacked SCF complex degrades ARGONAUTE proteins, which usually target host siRNA to viral RNA for silencing (Baumberger et al., 2007; Bortolamiol et al., 2007).
The oomycete RXLR effectors have attracted a large amount of research effort over the past decade. The Phytophthora infestans Avr3a effector has been shown to interact and stabilize U-box CMPG1 (Bos et al., 2010). This interaction was shown to suppress Avr-triggered HR PCD using an Agrobacterium transient expression system of R genes treated with Avr peptides (Gilroy et al., 2011).
A recent study dubbed ‘The effector interactome’ indicates that there may be 165 possible effector–host interactions in Arabidopsis (Mukhtar et al. (2011). A significant number of these interactions impinge on host ubiquitin machinery, indicating the high value pathogens put on the UPS as a target for defence suppression. However, despite clear evidence that the UPS is important for plant immunity our understanding of the mechanistic connections is still fragmented.
The implication of the UPS in a wide range of cellular contexts reflects the advantages conferred by selective protein degradation over other types of regulatory mechanism.
The main advantages relate to the speed and commitment of Ub-based signalling. Ub-tagged substrates can have their half-lives swiftly reduced (to the order of minutes) with rapid changes in steady-state level induced by specific stimuli. The irreversible removal of substrates also prevents the effects of inappropriate reactivation. Such features explain the frequent involvement of selective protein degradation in signalling processes requiring explicit timing control, including cell cycle progression, embryogenesis, cell lineage specification and metabolic control.
The cost to the cell of maintaining such a rapid and sensitive system to regulate protein levels is the large overall energy consumption required to continually degrade and resynthesize proteins. However, this is offset by the relatively small fraction of proteins (most of which are key regulatory proteins) that normally undergo continuous turnover in the cell. The major challenge now in plant biology is to define the targets for ubiquitination, their interactors and their molecular function within the regulatory context.
While the challenge to fully characterize the diverse signalling roles of ubiquitin is in its early stages in plants, the clear importance of its function to effect rapid and highly specific degradation of proteins in response to extracellular or developmental cues is compelling.
The Authors thank Dr Richard Ewan for helpful discussions on the manuscript figures. Work in the Sadanandom laboratory is funded by Syngenta and Biotechnology and Biological Sciences Research Council.