MAPK phosphorylation-induced stabilization of ACS6 protein is mediated by the non-catalytic C-terminal domain, which also contains the cis-determinant for rapid degradation by the 26S proteasome pathway

Authors

  • Sunjoo Joo,

    1. Department of Biochemistry, University of Missouri-Columbia, Columbia, MO 65211, USA
    Search for more papers by this author
    • Present address: Department of Biology, Washington University, St Louis, MO 63130, USA.

  • Yidong Liu,

    1. Department of Biochemistry, University of Missouri-Columbia, Columbia, MO 65211, USA
    2. Bond Life Sciences Center, University of Missouri-Columbia, MO 65211, USA
    Search for more papers by this author
  • Abraham Lueth,

    1. Department of Biochemistry, University of Missouri-Columbia, Columbia, MO 65211, USA
    2. Bond Life Sciences Center, University of Missouri-Columbia, MO 65211, USA
    Search for more papers by this author
  • Shuqun Zhang

    Corresponding author
    1. Department of Biochemistry, University of Missouri-Columbia, Columbia, MO 65211, USA
    2. Bond Life Sciences Center, University of Missouri-Columbia, MO 65211, USA
    Search for more papers by this author

(fax 573 884 9676; e-mail zhangsh@missouri.edu).

Summary

Ethylene is an important hormone in plant growth, development and responses to environmental stimuli. The ethylene-signaling pathway is initiated by the induction of ethylene biosynthesis, which is under tight regulation at both transcriptional and post-transcriptional levels by exogenous and endogenous cues. 1-Aminocyclopropane-1-carboxylic acid synthase (ACS) is the rate-limiting enzyme that catalyzes the committing step of ethylene biosynthesis. Recently, we found that ACS2 and ACS6, two isoforms of the Arabidopsis ACS family, are substrates of a stress-responsive mitogen-activated protein kinase (MAPK) cascade. Phosphorylation of ACS2/ACS6 by MPK6 leads to the accumulation of ACS proteins and the induction of ethylene. In this report, we demonstrate that unphosphorylated ACS6 protein is rapidly degraded by the 26S proteasome pathway. The degradation machinery targets the C-terminal non-catalytic domain of ACS6, which is sufficient to confer instability to green fluorescent protein and luciferase reporters. Phosphorylation of ACS6 introduces negative charges to the C-terminus of ACS6, which reduces the turnover of ACS6 by the degradation machinery. Consistent with this, other nearby conserved negatively charged amino acid residues are essential for ACS6 stability regulation. Protein degradation and phosphorylation are two important post-translational modifications of proteins. This research reveals an intricate interplay between these two important processes in controlling the levels of cellular ACS activity, and thus ethylene biosynthesis. The post-translational nature of both processes ensures a rapid response of ethylene induction, which is detectable within minutes after plants are exposed to stress.

Introduction

Ethylene, a gaseous plant hormone, plays important roles in regulating plant growth, development and responses to biotic/abiotic stresses (Abeles et al., 1992; Broekaert et al., 2006; Klee, 2004; Wang et al., 2002). Ethylene-regulated processes begin with an increase in ethylene biosynthesis (Chae and Kieber, 2005; Kende, 2001; Wang et al., 2002; Zarembinski and Theologis, 1994). There are only two reactions that are specific for the ethylene biosynthetic pathway, the conversion of S-adenosyl-l-methionine (SAM) to 1-aminocyclopropane-1-carboxylic acid (ACC), which is catalyzed by ACC synthase (ACS), and the oxidative cleavage of ACC to form ethylene, which is catalyzed by ACC oxidase (ACO; Kende, 1993; Yang and Hoffman, 1984; Zarembinski and Theologis, 1994). In general, ACS activity is very low in tissues that do not produce significant quantities of ethylene. Upon stimulation, ACS activity is rapidly induced. In contrast, ACO activity is constitutively present in most vegetative tissues. Therefore, ACS is the rate-limiting enzyme and is the major regulatory step in ethylene induction (Bleecker and Kende, 2000; Chae et al., 2003; McKeon et al., 1995; Wang et al., 2002).

Arabidopsis has nine ACS genes: encoding eight functional isoforms (ACS2, ACS4, ACS5, ACS6, ACS7, ACS8, ACS9 and ACS11) and one non-functional isoform (ACS1; Tsuchisaka and Theologis, 2004a; Yamagami et al., 2003). Although the ACS1 homodimer is inactive, it can form functional heterodimers with ACS2 and ACS6, and thus potentially contributes to the total cellular ACS activity (Tsuchisaka and Theologis, 2004b). Based on the C-terminal sequences, plant ACS isoforms are classified as Type 1 (ACS1, ACS2 and ACS6), Type 2 (ACS4, ACS5, ACS8 and ACS9) and Type 3 (ACS7 and ACS11) subgroups (Yoshida et al., 2005). ACS activity is regulated at multiple levels. Transcriptional regulation of ACS genes has been investigated in great detail (Arteca and Arteca, 1999; Tsuchisaka and Theologis, 2004a; Vahala et al., 1998). Post-transcriptional regulation has also been implicated in the upregulation of ACS activity (Chappell et al., 1984; Felix et al., 1991). Functional ACS enzymes are dimeric proteins. Formation of homodimers and heterodimers between different ACS gene products could serve as another level of fine-tuning of the cellular ACS activity (Tsuchisaka and Theologis, 2004b). Recent studies provided convincing evidence that supports yet another fundamental level of regulation, the stability of the ACS protein (Chae and Kieber, 2005; Chae et al., 2003; Liu and Zhang, 2004; Wang et al., 2004).

Several studies documented that the cellular ACS activity is rapidly turned over, with half-lives ranging from about 20 min to several hours (Kende, 1993; Kim and Yang, 1992; Spanu et al., 1990). However, the specifics, such as which ACS member(s) are involved and whether the half-life of a specific ACS protein is regulated, were not clear. Genetic screens identified ethylene-overproducing 1 (eto1), eto2 and eto3 mutants (Guzman and Ecker, 1990; Kieber et al., 1993). eto2 and eto3 have mutations in the C-termini of ACS5 and ACS9, respectively, which stabilize the ACS5 and ACS9 proteins, respectively (Chae et al., 2003; Vogel et al., 1998). Cytokinin, a plant hormone that increases ethylene biosynthesis, was found to prolong the half-life of the ACS5 protein (Chae et al., 2003; Vogel et al., 1998), providing direct evidence supporting the regulation of an ACS half-life by a specific stimulus. ETO1 encodes a BTB (Broad-Complex, Tramtrack, Bric-a-brac) domain-containing protein, which has been shown to function as a substrate adaptor in CUL3-based ubiquitin ligase (Wang et al., 2004). ETO1 interacts with both ACS5 and CUL3, targeting ACS5, and possibly other closely related ACS member(s), to degradation by the 26S proteasome pathway (Wang et al., 2004; Yoshida et al., 2005). Mutations in the C-termini of ACS5 (eto2) and ACS9 (eto3), are likely to reduce the efficiency of ETO1 binding, therefore resulting in their stabilization.

Protein phosphorylation/dephosphorylation is one of the major signaling pathways that regulates protein stability (Deshaies and Ferrell, 2001; Hardtke et al., 2000; Karin and Ben-Neriah, 2000; Tang et al., 2005). Based on studies using general kinase and phosphatase inhibitors, protein phosphorylation and dephosphorylation were implicated in the regulation of ethylene induction in plants under stress (Felix et al., 2000; Spanu et al., 1994; Tuomainen et al., 1997). Possible mechanisms underlying the phosphorylation regulation of ACS include the change in ACS activity, ACS stability, or both (Spanu et al., 1994). Recently, tomato LeACS2 was shown to be phosphorylated in vitro by an unidentified calcium-dependent protein kinase (CDPK) from pericarp tissue (Tatsuki and Mori, 2001). It was hypothesized that CDPK might be involved in the ETO1-mediated stability regulation of ACS5 (Wang et al., 2004). However, the identity of the CDPK remains to be elucidated, and in vivo functional analysis of the phosphorylation regulation is still lacking.

Mitogen-activated protein kinase (MAPK) cascades are conserved eukaryotic signaling modules downstream of sensors/receptors that transduce extracellular stimuli into cellular responses (MAPK Group., 2002; Widmann et al., 1999). Each MAPK cascade consists of three kinases with an MAPK at the bottom tier. Phosphorylation activation of an MAPK is carried out by an MAPK kinase (MAPKK, or MEK), which is activated, in turn, by an MAPKK kinase (MAPKKK, or MEKK). In plants, MAPKs have been implicated in regulating plant growth, development and stress/defense responses. Arabidopsis MPK6 and MPK3, as well as their functional orthologs in other plant species, including tobacco SIPK and WIPK, are activated in plants under a variety of stresses (reviewed in Mizoguchi et al., 1997; Tena et al., 2001; Zhang and Klessig, 2001; MAPK Group., 2002; Nakagami et al., 2005; Pedley and Martin, 2005). Based on biochemical and genetic analyses, we placed tobacco SIPK and its orthologous Arabidopsis MPK6 cascade upstream of the ethylene biosynthetic pathway (Kim et al., 2003; Liu and Zhang, 2004). ACS2 and ACS6, two members of the Type-1 ACS family, are substrates of MPK6. Phosphorylation of ACS2/ACS6 by MPK6 leads to the accumulation of ACS protein, the rate-limiting enzyme of ethylene biosynthesis, and thus elevated cellular ACS activity and ethylene production (Liu and Zhang, 2004). MAPK phosphorylation sites, which are only present in the Type-1 ACS members, are highly conserved in both dicotyledonous and monocotyledonous plants. In contrast, Type-2 and -3 ACS members lack the MAPK phosphorylation sites, suggesting differential regulation of different subgroups in the ACS family.

In this report, we demonstrate that unphosphorylated ACS6 is rapidly turned over by the 26S proteasome pathway. The phospho-mimicking form of ACS6, ACSDDD, has an apparent half-life of more than 3 h. In contrast, ACS6AAA, the ACS6 form lacking the phosphorylation sites, is very unstable, with an apparent half-life of <10 min. MAPK phosphorylation sites, which are important for ACS stabilization, are in the C-terminus of ACS6. Interestingly, the cis-determinant responsible for ACS6 instability also resides in this region. The C-terminus of ACS6 is required for proteasome-mediated degradation. In addition, the C-terminal 61-amino-acid region of ACS6 is sufficient to confer the instability and phosphorylation-dependent stabilization to reporter proteins. MAPK phosphorylation introduces additional negative charges to the ACS protein, which is important for the stabilization of ACS6. Protein degradation and phosphorylation are two important post-translational modifications of proteins. This research reveals an intricate interplay between the two important processes in controlling the levels of cellular ACS activity and ethylene biosynthesis in plants.

Results

Rapid degradation of unphosphorylated ACS6 by the 26S proteasome pathway

ACS isozymes are extremely efficient enzymes with very high specific activities (Liu and Zhang, 2004; Yamagami et al., 2003). Although ACS6 activity is readily detectable in total Arabidopsis protein extracts, the abundance of ACS6 protein is very low. Even under the induced state, we were unable to detect ACS6 protein directly by immunoblot analysis. In our previous report, we had to use an anti-Flag antibody to immunoprecipitate and enrich the Flag-tagged ACS6 protein, and then use the anti-ACS6 antibody to detect the ACS6 protein by immunoblot analysis (Liu and Zhang, 2004). This coupled immunoprecipitation–immunoblot assay is very laborious and less quantitative. To study the turnover of ACS6, we remade all constructs with ACS6 and its mutants fused to a four-copy myc-epitope tag (4myc).

Ethylene production in pooled T135S:4myc-ACS6WT, 35S:4myc-ACS6DDD and 35S:4myc-ACS6AAA transgenic seedlings was determined as previously described (Liu and Zhang, 2004). Again, we found that only the 35S:4myc-ACS6DDD seedlings overproduced ethylene (Figure 1a). In contrast, 35S:4myc-ACS6WT and 35S:4myc-ACS6AAA transgenic seedlings only produced ethylene at the low basal levels of control seedlings. Consistent with the overproduction of ethylene, 35S:4myc-ACS6DDD seedlings accumulated 4myc-tagged ACS6 protein that was detectable by direct immunoblot using anti-myc antibody (Figure 1a). In contrast, no tagged ACS6 protein was detectable in 35S: 4myc-ACS6WT or 35S:4myc-ACS6AAA transgenic seedlings, consistent with our previous publication usingFlag-epitope-tagged ACS6 (Liu and Zhang, 2004).

Figure 1.

 The proteasome inhibitor induces the accumulation of ACS6 protein and elevates ethylene production.
(a) Elevated ethylene production in 35S:4myc-ACS6DDD transgenic seedlings. GC vials with T135S:4myc-ACS6WT, 35S:4myc-ACS6DDD or 35S:4myc-ACS6AAA transgenic seedlings (2-weeks old) were flushed and capped. Ethylene levels in the headspace were determined 24 h later (upper panel). Seedlings were collected for immunoblot analysis using anti-myc antibody (lower panel). Error bars indicate the standard deviation (n = 3).
(b) Inhibition of the 26S proteasome pathway results in the accumulation of the ACS6WT protein and ethylene induction. Two-week-old T135S:4myc-ACS6WT seedlings were treated with MG132 (50 μm final concentration) or an equal volume of DMSO, the solvent for the MG132 stock, as a control. Ethylene levels in the headspace were determined 12-h later. The levels of 4myc-tagged ACS6WT protein were determined by immunoblot analysis using anti-myc antibody. Error bars indicate the standard deviation (n = 3).

To determine whether unphosphorylated ACS6 is degraded via the ubiquitin–proteasome pathway, we treated 35S:4myc-ACS6WT seedlings with MG132, a specific reversible inhibitor of the 26S proteasome. As shown in Figure 1b, MG132 treatment elevated the ethylene production by approximately 50-fold compared with the controls treated with DMSO, the solvent for the MG132 stock solution. The elevated ethylene production was associated with the accumulation of 4myc-ACS6WT protein (Figure 1b). These results demonstrate that ACS6, when unphosphorylated, is rapidly degraded by the proteasome-dependent pathway. MG132 treatment may stabilize other ACS isoforms, which could contribute to the increase in ethylene production shown in the upper panel of Figure 1b.

Phosphorylation of Ser residues in the ACS6 C-terminus stabilizes ACS6 by altering its turnover rates

The accumulation of ACS6WT protein in seedlings treated with MG132 inhibitor suggests that the instability of ACS6 is a result of rapid degradation of the unphosphorylated protein. To understand the regulation of ACS6 stability by phosphorylation, we compared the half-lives of 4myc-ACS6WT, 4myc-ACS6DDD and 4myc-ACS6AAA proteins. Because of the extreme instability, ACS6WT and ACS6AAA proteins do not accumulate to detectable levels in the transgenic Arabidopsis. Therefore, we treated the seedlings first with MG132 overnight to allow the proteins to accumulate. After 16 h, MG132 was removed by washing with fresh medium, and cycloheximide (CHX) was added to inhibit de novo protein synthesis. Because MG132 is a reversible inhibitor, we can determine the apparent half-lives of the ACS6 proteins by measuring their levels at various times (Figure 2a). As shown in Figure 2b, the wild-type (WT) 4myc-ACS6WT protein had an apparent half-life of approximately 1 h. The phospho-mimicking 4myc-ACS6DDD protein showed a lower rate of degradation, with an apparent half-life of >3 h. The levels of 4myc-ACS6AAA, the loss-of-phosphorylation ACS mutant, decreased much more rapidly, with an apparent half-life of <10 min.

Figure 2.

 Apparent half-lives of wild-type ACS6WT, phospho-mimicking ACS6DDD and loss-of phosphorylation ACS6AAAin vivo.
(a) Diagram showing the treatment scheme. Two-week-old transgenic seedlings were pre-treated with MG132 (50 μm) for 16 h. After being washed with fresh medium, to remove the proteasome inhibitor, the seedlings were transferred to medium with cycloheximide (CHX; 1 mm). Seedlings were then collected at the indicated times for measurement of the 1-aminocyclopropane-1-carboxylic acid synthase (ACS) levels by immunoblot analysis.
(b) Homozygous T335S:4myc-ACS6WT, 35S:4myc-ACS6DDD and 35S:4myc-ACS6AAA seedlings were grown in liquid medium. When the seedlings were 2-weeks old, they were treated as described in (a). The levels of 4myc-tagged ACS6 protein were determined in the total extracts by immunoblot analysis using an anti-myc antibody.
(c) Duplicate gels were stained with Coomassie-blue to confirm equal loading.

These results support the conclusion that MAPK-phosphorylation of ACS6 attenuates the degradation of ACS6, which prolongs the half-life of ACS6 protein. The estimation of apparent half-lives using this procedure is likely to be longer than the actual half-lives, because of the time needed to reverse the effect of MG132, and because of the <100% inhibition of de novo protein synthesis by CHX. The relatively long apparent half-life of ACS6WT is likely to be a result of MPK3/MPK6 activation during the washing manipulation and CHX treatment, both are known to activate MPK3/MPK6 and their orthologs in other species (Usami et al., 1995; Y.L. & S.Z., unpublished data). The very short half-life of ACS6AAA, which cannot be phosphorylated, should reflect more closely the half-life of ACS6 without MPK3/MPK6 activation.

Phosphorylation of multiple Ser residues is needed to stabilize ACS6

ACS6 has three Ser residues (Ser480, Ser483 and Ser488) in the C-terminus that can be phosphorylated by MPK6/MPK3 (Liu and Zhang, 2004; Y.L. & S.Z., unpublished data). Based on the in vitro phosphorylation assay, the three Ser residues can be independently phosphorylated at equal efficiencies, and the phosphorylation of all three residues can be achieved rapidly (Liu and Zhang, 2004). One standing question is whether the phosphorylation of all three Ser residues is required for ACS6 stabilization. To answer this question, we generated single and different combinations of double Ser-to-Asp mutants of ACS6.

As shown in Figure 3, none of the three single S-to-D mutant transgenic seedlings (35S:4myc-ACS6S480D, 35S:4myc-ACS6S483D and 35S:4myc-ACS6S488D) produced more ethylene than the 35S:4myc-ACS6WT control seedlings. The 35S:4myc-ACS6S480D/S483D seedlings failed to produce a higher level of ethylene, the 35S:4myc-ACS6S483D/S488D seedlings produced an elevated level of ethylene, and the 35S:4myc-ACS6S480D/S488D seedlings produced the highest level of ethylene among the three, although still at a much lower level than the triple S-to-D mutant 35S:4myc-ACS6DDD transgenic seedlings (Figure 3). Immunoblot analysis using an anti-myc antibody revealed a close correlation between the levels of ethylene production and the levels of ACS protein accumulation. These results indicated that the phosphorylation of multiple Ser residues is needed to stabilize the ACS6 protein. In the C-termini of Type-1 ACS isoforms from different plant species, the number of putative phosphorylation sites is evolutionarily conserved, with most having three phosphorylation sites and a few having four (Liu and Zhang, 2004; Figure S1), suggesting that the phosphorylation of multiple Ser residues is important to the functionality of Type-1 ACS regulation.

Figure 3.

 Mutation of multiple Ser residues to Asp is needed to stabilize ACS6.
GC vials with 2-week-old T135S:4myc-ACS6S480D, 35S:4myc-ACS6S483D, 35S:4myc-ACS6S488D, 35S:4myc-ACS6S480D/S483D, 35S:4myc-ACS6S480D/S488D or 35S:4myc-ACS6S483D/S488D transgenic seedlings were flushed and capped. Ethylene levels in the headspace were determined 24-h later (upper panel). Seedlings were collected for immunoblot analysis using anti-myc antibody (lower panel). Error bars indicate the standard deviation (n = 3).

The conserved acidic amino acids in front of MAPK phosphorylation sites are essential for ACS6 stabilization

ACS isozymes have a highly conserved N-terminal catalytic domain and short C-terminal extensions that are not required for their enzyme activity (Yamagami et al., 2003). Type-1 ACS isozymes, including ACS6, have longer C-terminal extensions than Type-2 and -3 isoforms. The MAPK phosphorylation sites are only present in the Type-1 ACS isoforms, with the last Ser residue situated about 4–10 amino acids from the stop codon. Besides the conserved MAPK phosphorylation sites, the C-termini of the Type-1 ACS isoforms have additional charged amino acid residues that are highly conserved, including two invariable acidic residues (Asp or Glu) located at the −4 and −5 position from the first MAPK phosphorylation site, and two clusters of basic residues further upstream (Figure 4a; Liu and Zhang, 2004; Figure S1).

Figure 4.

 D475 and D476, two highly conserved acidic amino acids in front of the mitogen-activated protein kinase (MAPK) phosphorylation sites, are essential for ACS6 stabilization.
(a) The diagram illustrates the mutants with the D475 and D476 residues substituted with either neutral or basic amino acids.
(b) Mutation of the conserved D475/D476 residues to either neutral or basic amino acids destabilizes the phospho-mimicking ACS6DDD protein, and abolishes the ethylene production. GC vials with 2-week-old T135S:4myc-ACS6WT-AA, 35S:4myc-ACS6WT-RR, 35S:4myc-ACS6DDD-AA or 35S:4myc-ACS6DDD-RR transgenic seedlings were flushed and capped. Ethylene levels in the headspace were determined 24 h later (upper panel). Seedlings were collected for immunoblot analysis using anti-myc antibody (lower panel). The controls with 35S:4myc-ACS6WT, 35S:4myc-ACS6DDD and 35S:4myc-ACS6AAA were also shown in Figure 1a. Error bars indicate the standard deviation (n = 3).
(c) Mutation of the conserved D475/D476 residues abolishes the stability regulation of ACS6WT by the MPK3/MPK6 cascade. Two-week-old T135S:4myc-ACS6WT, 35S:4myc-ACS6DDD, 35S:4myc-ACS6AAA, 35S:4myc-ACS6WT-AA, 35S:4myc-ACS6WT-RR, 35S:4myc-ACS6DDD-AA and 35S:4myc-ACS6DDD-RR transgenic seedlings in the GVG-NtMEK2DD background were treated with dexamethasone (DEX; 1 μm) or ethanol as controls. The vials were flushed and capped. After 12 h, seedlings were collected for immunoblot analysis using the anti-myc antibody.

Phosphorylation modifies the properties of a protein by introducing negative charges, which can alter either the conformation of the protein or its interaction with other proteins. For this reason, charged amino acid residues in the proximity could potentially influence the phosphorylation regulation. To determine if D475 and D476 of ACS6 are required for the phosphorylation regulation of ACS6, we mutated them to either Ala, a neutral amino acid, or Arg, a positively charged amino acid, in both ACS6WT and ACS6DDD backgrounds (Figure 4a). 35S-promoter-driven 4myc-ACS6WT-AA, 4myc-ACS6WT-RR, 4myc-ACS6DDD-AA and 4myc-ACS6DDD-RR constructs were transformed into WT and GVG-NtMEK2DD plants.

As shown in Figure 4b, 35S:4myc-ACS6WT-AA and 35S:4myc-ACS6WT-RR seedlings produced similar basal levels of ethylene as did 35S:4myc-ACS6WT seedlings, and no protein accumulation was detectable, suggesting that D475 and D476 are not required for ACS6 degradation. In contrast, mutation of these two Asp residues to either Ala or Arg abolished the accumulation of the gain-of-phosphorylation ACS6DDD, and reversed the ethylene overproduction in 35S:4myc-ACS6DDD seedlings. As shown in Figure 4b, 35S:4myc-ACS6DDD-AA and 35S:4myc-ACS6DDD-RR seedlings only produced basal levels of ethylene, and failed to accumulate ACS protein, demonstrating that these two conserved Asp residues are essential for the accumulation of phospho-mimicking ACS6DDD. Consistent with the abolishment of ACS protein accumulation and ethylene production, the typical ethylene-induced short-and-hairy root phenotype of the gain-of-function ACS6DDD transgenic seedlings was abrogated by the replacement of the two Asp residues with Ala (Figure S1).

To directly demonstrate the involvement of these two Asp residues in the phosphorylation regulation of ACS6, we examined the accumulation of the same set of ACS6 mutant transgenes in GVG-NtMEK2DD seedlings before and after dexamethasone (DEX) treatment. As shown in Figure 4c, both ACS6WT-AA and ACS6WT-RR failed to accumulate in the GVG-NtMEK2DD transgenic background after DEX treatment, which is in contrast to the accumulation of ACS6WT after MAPK activation. Also, no accumulation of ACS6DDD-AA and ACS6DDD-RR was observed. Taken together, these results demonstrate that D475 and D476 play positive roles in stabilizing the ACS6 protein, and are essential for the stability regulation of ACS6 by phosphorylation.

The conserved basic amino acid residues negatively impact the stability of ACS6

Besides the two conserved acidic residues, the ACS6 C-terminus contains two clusters of conserved basic amino acids further upstream, with one rich in Lys and the other rich in Arg (Figure 5a). Lys residues can serve as ubiquitination sites (Pickart, 2001; Smalle and Vierstra, 2004). If the residue(s) in the K454K455K456K457K458 cluster are the ubiquitination sites in ACS6, their mutation to other amino acids will prevent the attachment of ubiquitin, which in turn will stabilize the unphosphorylated ACS6 protein. To test this, we replaced the Lys cluster with neutral amino acids Ala and Ile. As shown in Figure 5b, there was no increase in ethylene production or protein accumulation when the KKKKK cluster was changed to neutral amino acids (ACS6WT-IAIAI), suggesting that they are not the ubiquitination sites in ACS6, or that there are additional sites elsewhere. In the phospho-mimicking ACS6DDD background, substitution of KKKKK with IAIAI resulted in a slight enhancement (approximately 1.5-fold) of protein accumulation and ethylene production in 35S:4myc-ACS6DDD-IAIAI transgenic plants (Figure 5c). Mutation of R472R473 to neutral Ala and Leu residues, in either ACS6WT or ACS6DDD backgrounds, gave similar results (Figure 5b,c). When both clusters were replaced with neutral residues, 35S:4myc-ACS6WT-IAIAI-AL seedlings had slightly elevated levels of ACS protein accumulation and ethylene production (Figure 5b). Based on these results, we conclude that Lys residues in the first cluster are not essential for the ubiquitin-mediated degradation of ACS6, and that these two clusters of conserved basic residues have a weakly negative impact on the stability of ACS6.

Figure 5.

 Substitution of the conserved positively charged amino acid clusters with basic Asp residues stabilizes ACS6.
(a) The diagram illustrates the mutants with the two clusters of basic amino acids substituted with either neutral or basic amino acids.
(b) Mutation of both clusters of basic amino acids to acidic Asp residues stabilizes the ACS6 protein. GC vials containing 2-week-old T1 transgenic seedlings were flushed and capped. Ethylene levels in the headspace were determined 24 h later (upper panel). Seedlings were collected for immunoblot analysis using an anti-myc antibody (lower panel). Error bars indicate the standard deviation (n = 3).
(c) Mutation of basic amino acid clusters to acidic Asp residues further stabilizes the phospho-mimicking ACS6DDD protein. GC vials with 2-week-old T1 transgenic seedlings were flushed and capped. Ethylene levels in the headspace were determined 24 h later (upper panel). Seedlings were collected for immunoblot analysis using an anti-myc antibody (lower panel). Error bars indicate the standard deviation (n = 3).

In addition to mutating Lys/Arg to neutral amino acids, we also generated mutants by changing these Lys/Arg residues to negatively charged Asp. Substitution of each cluster of basic amino acids with Asp in the ACS6WT background resulted in only a slight increase in ACS6 accumulation (Figure 5b). However, when both clusters were changed to Asp, 4myc-ACS6WT-DDDDD-DD protein accumulated without phosphorylation, which led to higher levels of ethylene production in 35S:4myc-ACS6WT-DDDDD-DD transgenic seedlings (Figure 5b). The effect of replacing positively charged residues with the negatively charged Asp is additive to the phosphorylation of Ser residues. Both 35S:4myc-ACS6DDD-DDDDD and 35S:4myc-ACS6DDD-DD plants produced much higher levels of ethylene, correlating with the much higher accumulation of ACS6 protein (Figure 5c). The seedlings with 35S:4myc-ACS6DDD-DDDDD-DD transgene produced the highest level of ethylene, at a rate of 65.2 nl g−1 h−1. In these transgenic seedlings, ACC oxidase is likely to be the limiting factor in ethylene biosynthesis, because the increase in ACS6 protein was not accompanied by a proportional increase in ethylene production (Figure 5c). Based on these findings, we conclude that the addition of negative charges to the C-terminus of ACS6 is sufficient to stabilize the ACS6 protein. It is possible that MAPK phosphorylation-induced ACS6 stabilization is a result of the addition of negative charges to the ACS6 protein.

The C-terminus of ACS6 is the targeting site of the degradation machinery

The data above demonstrated that the C-terminus of ACS6, where the MAPK phosphorylation sites reside, is important to the stability regulation mediated by the stress-response MPK6/MPK3 cascade. In the process of generating amino acid replacement mutants, we identified an early termination mutant (ACS6-ΔC) as a result of a frame shift. Surprisingly, transgenic seedlings with this ACS deletion mutant produced high levels of ethylene, and showed a hairy-root phenotype (Figure S1). We then generated five serial deletion mutants (ACS6del1–ACS6del5) by introducing a stop codon at the positions indicated in Figure 6a. We found that the deletion of the last 16 amino acids was sufficient to stabilize the ACS6 protein. The 35S:4myc-ACS6del1 seedlings produced extremely high levels of ethylene, correlating with high levels of ACS6 protein accumulation (Figure 6b). Deletion of additional sequences resulted in only a slight increase in ethylene production. Unlike 35S:4myc-ACS6WT transgenic seedlings, treatment of 35S:4myc-ACS6del4 and 35S:4myc-ACS6del5 transgenic seedlings with MG132 neither enhanced the ethylene production nor increased the accumulation of ACS6 protein. Treatment of 35S:4myc-ACS6del1, 35S:4myc-ACS6del2 and 35S:4myc-ACS6del3 transgenic seedlings with MG132 resulted in only a slight increase in protein accumulation. These results indicate that the degradation of the deletion mutants is compromised, suggesting that the degradation machinery targets the C-terminus of ACS6.

Figure 6.

 The C-terminus of ACS6 is the targeting site of the 26S proteasome degradation machinery.
(a) The diagram illustrates the deletion mutants.
(b) Deletion of the non-catalytic C-terminus of ACS6 stabilizes the protein. Two-week-old T1 transgenic seedlings with 35S:4myc-ACS6WT or its serial deletion mutants grown in GC vials were treated with MG132 (50 μm) or with DMSO as a control. The GC vials were then flushed and capped. Ethylene levels in the headspace were determined 24 h later (upper panel). Seedlings were collected for immunoblot analysis using an anti-myc antibody (lower panel). Error bars indicate the standard deviation (n = 3).

The non-catalytic C-terminal domain of ACS6 is sufficient to confer stability regulation to green fluorescence protein and luciferase reporters

Based on the above analyses, we could conclude that the cis-determinants responsible for the rapid destruction of unphosphorylated ACS6 and the MAPK phosphorylation-induced stabilization of ACS6 are both embedded in the C-terminal region. To test if this region is sufficient to provide stability regulation to a reporter protein, we made transgenic plants expressing fusion constructs between the ACS6 C-terminal domain (CTD, from amino acid residue Phe435 to Thr495) and two different reporters, GFP and luciferase (LUC).

As shown in Figure 7a (lane 1), GFP protein accumulated to a high level in 35S:GFP control seedlings. The addition of the 61-amino-acid ACS6 CTD to GFP completely abolished the accumulation of the GFP-CTDWT fusion protein (Figure 7a, lane 4), demonstrating that the 61-amino-acid CTD of ACS6 is sufficient to confer instability to the GFP reporter. When the MAPK-phosphorylation sites within the CTD are in the phospho-mimicking form (GFP-CTDDDD), we were able to detect protein accumulation (Figure 7a, lane 6). However, the level of GFP-CTDDDD accumulation was much lower than that of GFP, consistent with the observation that the phospho-mimicking form of ACS6 can still be turned over by the 26S proteasome pathway, but at a lower rate (Figure 2b). Because of the lower sensitivity of the fluorescent detection, we were unable to observe fluorescence in the 35S:GFP-CTDDDD transgenic seedlings (data not shown). To demonstrate that the ACS6 CTD-conferred instability is mediated by the proteasome pathway, we treated 35S:GFP-CTDWT, 35S:GFP-CTDDDD and 35S:GFP-CTDAAA transgenic seedlings with MG132 for 16 h, and then determined the accumulation of GFP–CTD fusions. As shown in Figure 7a, the levels of GFP-CTDWT, GFP-CTDDDD and GFP-CTDAAA proteins were all markedly increased. The vector-transformed transgenic plants (Figure 7a, lanes 2 and 3) were used as negative controls in this experiment. Only non-specific bands were present in the negative controls, which showed no accumulation after MG132 treatment.

Figure 7.

 The C-terminal domain of ACS6 is sufficient to confer stability regulation to GFP and luciferase (LUC) reporter proteins.
(a) The C-terminal domain (CTD) of ACS6 confers stability regulation to the GFP reporter. Two-week-old T135S:GFP, 35S:GFP-CTDWT, 35S:GFP-CTDDDD and 35S:GFP-CTDAAA transgenic seedlings grown in GC vials were treated with MG132 (50 μm) or with DMSO as a control. After 16 h, the seedlings were collected and pooled for immunoblot analysis using an anti-GFP antibody.
(b) The CTD of ACS6 confers stability regulation to the LUC reporter. Two-week-old T135S:LUC-CTDWT, 35S:LUC-CTDDDD and 35S:LUC-CTDAAA transgenic seedlings grown in GC vials were collected for the luciferase assay. Error bars indicate the standard deviation (n = 3).
(c) The stability regulation conferred by the ACS6 CTD is dependent on the 26S proteasome degradation pathway. Two-week-old T135S:LUC and 35S:LUC-CTDWT transgenic seedlings grown in GC vials were treated with MG132 (50 μm) or with DMSO as a control. After 16 h, the seedlings were collected and pooled for luciferase assay. Relative LUC activity was normalized to that of DMSO-treated control seedlings, which was set arbitrarily as 1. Error bars indicate the standard deviation (n = 3).

Similar results were obtained with the LUC reporter. As shown in Figure 7b, much higher luciferase activity was detected in 35S:LUC-CTDDDD transgenic seedlings than in 35S:LUC-CTDWT and 35S:LUC-CTDAAA transgenic seedlings. Upon MG132 treatment, luciferase activity accumulated in 35S:LUC-CTDWT, but not in 35S:LUC control seedlings (Figure 7c). These results suggest that the CTD of ACS6 protein is sufficient to provide the instability and phosphorylation-mediated stability regulation to both reporter proteins.

Discussion

The stress-responsive MPK3/MPK6 cascade regulates ethylene biosynthesis by directly phosphorylating ACS2 and ACS6, two Type-1 ACS isoforms (Kim et al., 2003; Liu and Zhang, 2004). ACS1, the other Type-1 Arabidopsis ACS is also likely to be an MAPK, based on the presence of conserved MAPK phosphorylation sites in its C-terminus. In this report, we demonstrate that without MAPK activation the unphosphorylated ACS6 is rapidly degraded via the 26S proteasome pathway. Phosphorylation of the Ser residues in the ACS6 C-terminus slows down the degradation of the protein, based on half-life measurements. Loss-of-phosphorylation ACS6AAA has an apparent half-life of <10 min, whereas the phospho-mimicking ACS6DDD has a half-life of more than 3 h, an increase of more than 18-fold in protein stability (Figure 3). Because (i) CHX, which is used to inhibit the de novo protein synthesis during half-life measurement, activates MPK3/MPK6 in Arabidopsis (Usami et al., 1995; Y.L. & S.Z., unpublished data), and (ii) the ratio between phosphorylated and unphosphorylated ACS6WT protein is dynamic, depending on the level of MAPK activation, it is difficult to measure the half-life of ACS6WTin vivo. The estimated apparent half-life of ACS6WT (approximately 1 h, Figure 2b) is likely to be an average of unphosphorylated ACS6 and phosphorylated ACS6. In contrast, loss-of-phosphorylation and phospho-mimicking mutants of ACS6, ACS6AAA and ACS6DDD are present as homogeneous populations in cells, which allowed us to estimate their apparent half-lives more accurately.

MAPK phosphorylation sites, which are important for ACS6 stabilization, reside in the C-terminus of ACS6. We found that the degradation machinery also targets the C-terminal non-catalytic region of ACS6. The 61-amino-acid CTD is sufficient to confer instability and phosphorylation regulation to GFP and LUC reporters. Phosphorylation of ACS6 introduces additional negative charges to the C-terminus of ACS6, which is likely to reduce the recognition efficiency of the targeting/degradation machinery. Consistent with this theory, other charged amino acids in the proximity of phosphorylation sites are also important for the ACS6 stability regulation. D475 and D476, the two invariable acidic residues upstream of the phosphorylation sites, are essential for ACS6 stabilization. In contrast, the two conserved clusters of basic residues further upstream only weakly influence the regulation of ACS6 stability. However, their mutation to acidic residues resulted in the phosphorylation-independent accumulation of the mutant ACS6 protein and ethylene overproduction, which, again, supports the notion that negative charges in the C-terminus can reduce the degradation of ACS6.

MAPK-phosphorylation of ACS6 only slows down the degradation of ACS6, which allows the ethylene production to reset back to the basal level

Ethylene induction in plants under stress is transient most of the time, suggesting the existence of mechanisms that reset the ACS activity to basal level. This can be achieved by (i) protein phosphatases that dephosphorylate the phospho-ACS6, which can then be degraded, and (ii) the phosphorylated ACS6, which can still be degraded, but at a slower rate. Based on our analysis of ACS6DDD, it is likely that the second scenario is true. ACS6DDD, the phospho-mimicking form of ACS6, accumulates to detectable levels constitutively. When 35S:4myc-ACS6DDD seedlings were treated with MG132, higher levels of ACS6DDD protein accumulated (Figure 2b), suggesting that phosphorylation of ACS6 does not prevent it from degradation completely, but only slows down its degradation by the 26S proteasome pathway. We found that the deletion of the C-terminus of ACS6 can abolish the degradation, leading to very high protein accumulation and ethylene production, which are irresponsive to proteasome inhibitor treatment (Figure 6).

The ACS protein is prone to proteolysis, in both total plant protein extracts and in bacterial extracts, during the purification of recombinant protein (Liu and Zhang, 2004; Yamagami et al., 2003). Recently, a metalloprotease was identified as the protease in the plant protein extract that removes the C-terminus of ACS (Li et al., 2005). However, it is unknown whether this protease has a similar function in vivo. According to our deletion analysis (Figure 6), once the C-terminus is removed, the truncated ACS6 protein is very stable. However, no truncated ACS6 was detected in our experimental system, suggesting that only the proteasome pathway is involved in the regulation of ACS6. Nevertheless, we cannot exclude the possibility that this protease may function in specific organ/tissue/cell types, which can provide a mechanism for cells to produce sustained high levels of ethylene.

Phosphorylation of multiple Ser residues is needed for ACS6 stability regulation

All Type-1 ACS isoforms (paralogs/homologs within a plant species and orthologs from different species) have highly conserved MAPK phosphorylation sites in their extreme C-terminal regions (Liu and Zhang, 2004). Furthermore, the number of phosphorylation sites is also conserved: with most isozymes having three phosphorylation sites, and a few with four sites. Substitution of Ser residues with phospho-mimicking Asp residues revealed that more than one Ser residue must be phosphorylated for the ACS6 protein to become stable. In addition, two negatively charged amino acids, D475 and D476, are essential for the accumulation of phospho-mimicking ACS6DDD (Figure 4b) or phosphorylated ACS6WT (Figure 4c), probably by contributing additional negative charges needed to stabilize ACS6. These data suggest that the number of negative charges in the C-terminus of ACS6 is critical to the stabilization of ACS6. As shown in Figure 5, replacing the positively charged Lys and Arg with Asp can stabilize ACS6 without phosphorylation, indicating that it probably does not matter where exactly the negative charges are, and, as long as they are in the proximity, they can slow down the degradation of ACS6.

Although seedlings with the 35S:4myc-ACS6DDD transgene, in which all three Ser residues are mutated to Asp, produce the highest level of ethylene (Figure 3), two of the double Ser-to-Asp mutant transgenic seedlings, 35S:4myc-ACS6S480D/S488D and 35S:4myc-ACS6S483D/S488D, did produce elevated levels of ethylene. Could this be another mechanism for fine-tuning the level of ethylene production in plants? Depending on the levels of MPK6/MPK3 activation, ACS6 may not be fully phosphorylated, which will lead to the partial stabilization of ACS6 and the production of intermediate levels of ethylene.

The C-terminal flexible region of ACS6 functions as a docking domain for stability regulation

Deletion of the C-terminus of ACS6 fully stabilizes the ACS protein (Figure 6b), whereas fusion of the ACS6 CTD provides instability to the GFP and LUC reporter proteins (Figure 7), suggesting that the C-terminus of ACS6 contains the targeting signal for degradation. In addition, the phospho-mimicking version of ACS6 CTD (CTDDDD) confers stability to the reporter proteins (Figure 7), demonstrating that the phosphorylation of this region is sufficient to slow down the degradation. As a result, we conclude that the C-terminus of ACS6 contains the cis-determinants for both the ACS6 stabilization by MAPK phosphorylation and the rapid degradation of ACS6 by the ubiquitin–proteasome pathway. Unphosphorylated ACS6 protein is rapidly turned over in unstimulated cells. In response to a stress stimulus, the activation of MPK6/MPK3 leads to the phosphorylation of ACS6, which slows down the degradation process, thereby resulting in elevated levels of cellular ACS activity and ethylene biosynthesis.

Phosphorylation of ACS6 does not alter the kinetic properties of the ACS6 enzyme (Liu and Zhang, 2004), suggesting that there is no change in the conformation of the core enzymatic domain of ACS protein. The C-terminus of ACS6, which is highly hydrophilic and predicted to be a flexible region, extends out of the catalytic domain. We hypothesize that this region may function as a docking domain. Depending on the state of phosphorylation, different interacting proteins are recruited. When the ACS6 is not phosphorylated, binding of protein(s) that are involved in ACS6 degradation, possibly by an F-box protein or E3 ligase, targets the ACS6 protein to degradation. When ACS6 is phosphorylated, the binding efficiency of these proteins is reduced, and therefore prolongs the stability of the phosphorylated ACS6 protein. It is also possible that phosphorylation will recruit phosphorylation-dependent binding of a protein that protects the ACS protein from being recognized by the degradation machinery. Future research in the direction of identifying protein(s) that bind ACS6 in a phosphorylation/dephosphorylation-dependent manner will be critical to our understanding of the detailed mechanism.

Phosphorylation is one of the major signals that regulate the degradation of proteins by the proteasome pathway. In previously characterized examples, phosphorylation of a substrate in response to a specific signal mostly serves as a recognition tag of the E3-ubiquitin ligase, which facilitates the protein degradation (Deshaies and Ferrell, 2001; Hardtke et al., 2000; Karin and Ben-Neriah, 2000; Tang et al., 2005). There are also cases in which the phosphorylation can prevent the protein from proteasome-mediated degradation (Dan et al., 2004; Lavin and Gueven, 2006; Rolli-Derkinderen et al., 2005). Being one of the few examples in which the phosphorylation of a protein slows down the proteasome-mediated degradation, it would be interesting to understand the detailed mechanism of how MAPK phosphorylation of ACS6 slows down the proteasome-dependent degradation.

Experimental procedures

Plant growth and treatments

Arabidopsis thaliana WT (Col-0), mutants and transgenics were maintained at 22°C in a growth chamber with a 14-h light cycle (100 μE m−2 sec−1). Seedlings were grown in 50-ml gas chromatography (GC) vials with 6 ml of half-strength MS medium, situated in a growth chamber at 22°C under continuous light (70 μE m−2 sec−1), as previously described (Liu and Zhang, 2004). Unless indicated otherwise, 2-week-old seedlings were used for the experiments. DEX was used at a final concentration of 1 μm. Ethanol (the solvent for DEX) was used as a negative control. MG132 was used at a final concentration of 50 μm and an equal volume of DMSO (the solvent for MG132) was used as a negative control. Seedlings collected at various times after treatment were frozen in liquid nitrogen and stored at −80°C until their use.

At least two independent repetitions were performed for experiments with multiple time points. For single time-point experiments, at least three independent repetitions were completed.

Generation of mutant constructs and Agrobacterium-mediated transformation

Point mutations were introduced by QuickChange site-directed mutagenesis (Stratagene, http://www.stratagene.com). Deletion mutants were generated by introducing a stop codon at the desired locations. Both were confirmed by sequencing.

The ACS6 gene and its mutants with an NdeI site added to the ATG start codon were first cloned in-frame into an intermediate vector with a 4myc epitope tag at the 5′-end and flanked by XhoI and SpeI cloning sites. The inserts were then moved into a modified pBI121 binary vector with XhoI and SpeI cloning sites. GFP and LUC fusion constructs were generated similarly. Agrobacterium tumefaciens GV3101 carrying different constructs was grown overnight in Luria Bertani (LB) medium containing 25 μg ml−1 gentamycin and 50 μg ml−1 kanamycin. Transgenic Arabidopsis plants were generated using the flower-dipping method (Clough and Bent, 1998). T1 transformants were selected in the presence of kanamycin. After selection (5–6 days), the seedlings were either transplanted to soil for morphological observation and producing seeds, or were transferred to 50-ml GC vials with 6 ml of half-strength MS medium (12 seedlings per vial) for experiments. At least three vials were used for each construct in each repetition.

Assay of ethylene biosynthesis rates

The GC vials with Arabidopsis seedlings were flushed and capped immediately after treatment or without treatment. At indicated times, ethylene levels in the headspace of the GC vials were determined by gas chromatography, as previously described (Kim et al., 2003; Liu and Zhang, 2004). Seedlings were then harvested, weighed and frozen in liquid nitrogen for future analyses.

Protein extraction and immunoblot analysis

To detect 4myc- or GFP-tagged ACS6 proteins in the pooled transgenic Arabidopsis seedlings, we used a method similar to one described earlier (Chae et al., 2003). Harvested seedlings were ground in liquid nitrogen to powder, mixed with three volumes of 1.5× SDS sample buffer (without Bromophenol Blue dye). After centrifugation (20 000 g for 20 min) to remove the debris, the total protein extract was transferred to a new tube. The protein extract was diluted 500 times in 1 mm Tris (pH 6.8), and the protein concentration was determined using Bradford Reagent with BSA as the standard. Equal quantities of total proteins (50 μg) were fractionated on a 10% SDS polyacrylamide gel and were then transferred to nitrocellulose membrane. Immunoblot detection was performed as previously described (Yang et al., 2001). Anti-myc (Sigma, http://www.sigmaaldrich.com) and anti-GFP (Invitrogen, http://www.invitrogen.com) antibodies were used at 1:10 000 and 1:5000 dilutions, respectively.

Luciferase assay

For the LUC assay, protein was extracted from pooled seedlings and was stored at −80°C (Zhang and Klessig, 1997). The concentration of protein extracts was determined using the Bio-Rad protein assay kit (Bio-Rad, http://www.bio-rad.com) with BSA as the standard. Luciferase activity was determined using the Promega lusiferase assay system and a MonoLightTM 3010 luminometer (PharMingen, http://www.bdbiosciences.com).

Acknowledgements

This work was supported by grants from NSF (IBN-0133220 and MCB-0543109). We thank Clayton Larue for his critical reading of the manuscript and Njabulo Brian Ngwenyama for being involved in generating some of the constructs.

Ancillary