Abscisic acid inhibits root growth in Arabidopsis through ethylene biosynthesis

Authors


Summary

When first discovered in 1963, abscisic acid (ABA) was called abscisin II because it promotes abscission. Later, researchers found that ABA accelerates abscission via ethylene. In Arabidopsis, previous studies have shown that high concentrations of ABA inhibit root growth through ethylene signaling but not ethylene production. In the present study in Arabidopsis, we found that ABA inhibits root growth by promoting ethylene biosynthesis. The ethylene biosynthesis inhibitor L–α–(2-aminoethoxyvinyl)-glycine reduces ABA inhibition of root growth, and multiple mutants of ACS (1–aminocyclopropane-1–carboxylate synthase) are more resistant to ABA in terms of root growth than the wild-type is. Two ABA-activated calcium-dependent protein kinases, CPK4 and CPK11, phosphorylate the C–terminus of ACS6 and increase the stability of ACS6 in ethylene biosynthesis. Plants expressing an ACS6 mutant that mimics the phosphorylated form of ACS6 produce more ethylene than the wild-type. Our results reveal an important mechanism by which ABA promotes ethylene production. This mechanism may be highly conserved among higher plants.

Introduction

Abscisic acid was originally isolated from young fruit of cotton (Gossypium hirsutum), and was called abscisin II because it accelerated abscission (Ohkuma et al., 1963). Researchers later determined that ABA promotes abscission through ethylene biosynthesis (Cracker and Abeles, 1969; Riov et al., 1990), but the molecular mechanism remains unknown. Recent studies have indicated that ABA and ethylene signals are integrated to mediate plant growth and development and plant responses to biotic and abiotic stresses (van Loon et al., 2006; Zhu and Guo, 2008; Ton et al., 2009; Lumba et al., 2010; Zhao and Guo, 2011).

Ethylene biosynthesis is catalyzed by 1–aminocyclopropane-1–carboxylate (ACC) synthase (ACS), which converts S–adenosylmethionine to ACC, and ACC oxidase, which oxidizes ACC to ethylene (Yang and Hoffman, 1984; Wang et al., 2002). ACS is the rate-limiting enzyme for ethylene biosynthesis. In Arabidopsis, ACS is encoded by a small family of nine genes (Yoshida et al., 2005). ACSs may be divided into three groups according to the presence/absence of phosphorylation sites in their C–termini (Yoshida et al., 2005). Type 1 ACSs (AtACS1, 2 and 6) have both calcium-dependent protein kinase (CDPK) and mitogen-activated protein kinase (MAPK) phosphorylation sites. Type 2 ACSs (AtACS4, 5, 8, 9 and 11) have only CDPK target sites, and the type 3 ACS (AtACS7) lacks phosphorylation sites. ACS isoforms are regulated at both the transcriptional and post-transcriptional levels in response to various stimuli, and are also expressed at various developmental stages and in specific tissues (Wang et al., 2002; Stepanova and Alonso, 2009; Li et al., 2012).

The stability of ACS isoforms is regulated by protein phosphorylation and dephosphorylation (Liu and Zhang, 2004; Han et al., 2010; Skottke et al., 2011; Li et al., 2012). Arabidopsis MPK3 and MPK6 are able to phosphorylate type 1 ACS isoforms ACS2 and ACS6, which reduces turnover of the proteins by the ubiquitin/proteasome degradation machinery, increases ACS protein levels and activity, and enhances ethylene production under stress (Liu and Zhang, 2004; Joo et al., 2008; Han et al., 2010; Li et al., 2012). Researchers recently found that the protein phosphatase 2C ABI1 (ABA-INSENSITIVE 1) and protein phosphatase 2A counteract MAPK phosphorylation of ACS2/ACS6 and promote their degradation (Skottke et al., 2011; Agnieszka et al., 2014). In addition, a 14–3–3 protein directly interacts with ACS isoforms and increases ACS stability; the 14–3–3 protein also interacts with ETO1 (ETHYLENE OVERPRODUCER 1)/EOLs (ETO1-likes) but decreases the abundance of these broad complex/tramtrack/bric-a-brac (BTB) ubiquitin ligases (Yoon and Kieber, 2013). In cotton, GhACS2 interacts with and is phosphorylated by GhCPK1 in vitro (Wang et al., 2011a). In tomato (Solanum lycopersicum), LeACS2 is phosphorylated by both LeCDPK2 and MAPK (Hernández Sebastià et al., 2004; Kamiyoshihara et al., 2010). These phosphorylated ACS isoforms are more stable than non-phosphorylated ones (Kamiyoshihara et al., 2010; Wang et al., 2011a).

ABA and ethylene antagonistically affect seed germination in Arabidopsis (Beaudoin et al., 2000; Ghassemian et al., 2000; Linkies et al., 2009; Arc et al., 2013). Seed germination of ethylene-insensitive mutants such as ein2 and etr1 is hypersensitive to ABA (Beaudoin et al., 2000; Ghassemian et al., 2000). However, ABA and ethylene cascades share some common features in terms of mediation of root growth: low concentrations promote root growth and high concentrations inhibit root growth (Joshi-Saha et al., 2011; Arc et al., 2013). Previous studies suggested that ABA inhibits root growth through the ethylene signaling pathway, but ethylene production (Beaudoin et al., 2000; Ghassemian et al., 2000).

In this study, we found that both the ethylene biosynthesis inhibitor L–α–(2–aminoethoxyvinyl)-glycine (AVG) and the ethylene perception inhibitor Ag+ reduced ABA inhibition of Arabidopsis root growth. Multiple ACS mutants were more resistant to ABA in terms of root growth than the wild-type. ABA treatment greatly increased ethylene production in seedlings. We further show that two ABA-activated CDPKs, CPK4 and 11 (Zhu et al., 2007), phosphorylated ACS6. Transgenic plants expressing a mutated ACS6 that mimics the phosphorylated form of ACS6 produced more ethylene than plants expressing wild-type ACS6. These results indicate that ABA mediates root growth in Arabidopsis by promoting ethylene production.

Results

ABA inhibition of root growth is mediated by ethylene production

High concentrations of ABA inhibit Arabidopsis root growth. We performed a genetic screen to identify mutants whose root growth was resistant to ABA. We used the Ler accession because root growth is more sensitive to ABA in this accession than in Col. We identified five allelic mutants of ETHYLENE INSENSITIVE2 (EIN2): two of the mutations created a stop codon (Trp682STOP, Trp38STOP), and three altered amino acids (Gly36Glu, Arg671Trp and Gly1254Glu). These EIN2 mutations may be useful for further dissection of EIN2 functions in future studies. EIN2 is an important component in ethylene signaling (Alonso et al., 1999; Zhu and Guo, 2008). Previous results indicated that root growth of ethylene-insensitive mutants such as ein2 and etr1 is insensitive to ABA (Beaudoin et al., 2000; Ghassemian et al., 2000). However, the ABA-insensitive mutant abi1–1 shows a similar response to ethylene to the wild-type in terms of root growth. A previous study also showed that the ethylene-biosynthesis inhibitor AVG does not suppress ABA inhibition of root growth, but the ethylene perception inhibitor Ag+ reduces the sensitivity of root growth to ABA (Ghassemian et al., 2000). Furthermore, ABA treatment does not increase ethylene production or ethylene-induced gene expression (Beaudoin et al., 2000). These studies indicate that the ethylene signaling cascade but not ethylene biosynthesis is required for ABA inhibition of root growth (Beaudoin et al., 2000; Ghassemian et al., 2000). We re-assessed the effect of AVG and Ag+ on root growth in the presence of ABA. Consistent with previous results (Ghassemian et al., 2000), the ethylene perception inhibitor Ag+ (10 μm AgNO3) suppressed the inhibitory effects of a high ABA concentration (150 μm) on root growth of the wild-type (Col) (Figure 1a,b). However, in contrast with previous results (Ghassemian et al., 2000), the ethylene biosynthetic inhibitor AVG (0.2 μm) also conferred ABA resistance to root growth of wild-type (Figure 1a,b). The AVG concentration used in the present study was 10 times lower than that used in the previous study (2 μm), and it may be that the high concentration of AVG is toxic to root growth. As inhibition of ethylene production by addition of AVG and blockage of ethylene signaling by addition of Ag+ increase the root growth relative to growth on MS medium, the results indicate that, under normal conditions, basal level of ethylene inhibits root growth. These results suggest that ABA inhibits root growth by promoting ethylene biosynthesis.

Figure 1.

Root growth inhibition by ABA is mediated by ethylene production. (a) Root growth of the wild-type (WT) and the mutants (abi1 abi2 hab1, ein2–5, etr1–1 and eto1–1 on MS medium or MS medium containing 150 μm ABA, 0.2 μm AVG, 150 μm ABA plus 0.2 μm AVG, 10 μm AgNO3 or 150 μm ABA plus 10 μm AgNO3. Five-day-old seedlings on grown MS medium were transferred to MS medium or MS medium supplemented with the various reagents and grown for 3 days. The white arrowheads indicate the location of the root tips immediately after transfered to the various media. (b) Relative root growth for the plants shown in (a). The root length is expressed relative to that of seedlings without ABA. Three independent experiments were performed with similar results. Each combination of treatment and line in each experiment is represented by three replicate plates, with 15 roots per plate. Values are means ± SD (= 3). Means with different letters are significantly different at < 0.05.

Next, we investigated the root growth of various mutants in the ABA and ethylene signaling pathways in response to AVG or Ag+ in the presence of ABA. ein2–5 and etr1–1 were used as controls because their root growth is more resistant to both ABA and ethylene than that of wild-type. ABI1, ABI2 and HAB1 are protein phosphatase 2Cs that are key negative regulators of ABA signaling (Joshi-Saha et al., 2011). Interestingly, both AVG and Ag+ treatment significantly increased the root growth of wild-type and various mutant seedlings relative to seedlings growing on MS medium, except for that of ein2–5, which was not significantly increased by Ag+. These results indicate that both ethylene production and signaling reduce root growth under normal conditions. Relative to the wild-type, the triple knockout mutant abi1 abi2 hab1 was hypersensitive to ABA (Figure 1a,b) (Rubio et al., 2009). Addition of AVG or Ag+ to the medium largely restored ABA inhibition of root growth in the abi1 abi2 hab1 mutant (Figure 1a,b). AVG but not Ag+ also alleviated the ABA inhibition of root growth of ein2–5 and etr1–1 (Figure 1a,b). The eto1 mutant produced more ethylene than the wild-type (Wang et al., 2004), and its root length was only approximately 50% of that of the wild-type without ABA treatment. ABA treatment greatly inhibited eto1 root growth, and this inhibition was largely recovered when AVG or Ag+ was added to the medium. These results indicate that ethylene signaling acts downstream of ABA signaling in mediation of root growth.

ABA promotes ethylene production

To determine whether exogenous ABA increases ethylene production, we measured the ethylene production of the wild-type and mutants in ABA signaling with or without ABA treatment at various times. As shown in Figure 2(a), wild-type seedlings released more ethylene under ABA treatment than without ABA treatment after 8 h. Consistently, the ABA hypersensitive triple mutant abi1 abi2 hab1 accumulated more ethylene than wild-type under ABA treatment or without ABA treatment at 16 and 24 h (Figure 2a). ABA treatment did not increase the ethylene production of the dominant-negative mutant abi1–1 in the Col accession (the same mutation as abi1–1 in Ler) (Leung et al., 1994; Meyer et al., 1994). These results indicate that ABA signaling is involved in mediating ethylene production, and also confirm a previous report (Jiang et al., 2000) indicating that ABA promotes ethylene biosynthesis.

Figure 2.

ABA promotes ethylene production. (a) Ethylene production in the presence of 0 or 100 μm ABA. Five-day-old seedlings were transferred to 7 ml vials with 2 ml liquid MS medium (10 seedlings per vial). After 24 h at 21°C in a growth chamber, the vials containing Arabidopsis seedlings were treated with 0 or 100 μm ABA and then immediately capped. At the indicated times, 2 ml of headspace air was removed with a syringe and injected into a gas chromatograph. The weight of seedlings in each vial was determined. Three independent experiments were performed with similar results. Values are means ± SD (= 3) from one experiment. Each combination of treatment and line in each experiment was represented by three replicate vials. Means with different letters are significantly different at < 0.05. (b) Root growth of ACS mutants CS16647 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1), CS16649 (acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1) and CS16650 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1) on MS medium with 0 or 150 μm ABA. The white arrowheads indicate the location of the root tips immediately after transfered to the various media. (c) Relative root growth for the plants shown in (b). The root length is expressed relative to that of seedlings without ABA. Three independent experiments were performed with similar results. Each combination of treatment and line in each experiment was represented by three replicate plates, with 15 roots per plate. Values are means ± SD (= 3). Asterisks indicate statistically significant differences compared with wild-type (**< 0.01).

As mentioned above, ACSs are rate-limiting enzymes in ethylene biosynthesis. We hypothesized that the root growth of null mutants of ACSs will be more resistant to ABA than that of the wild-type if ethylene production is a key factor for ABA inhibition of root growth. Because of functional redundancy, single mutants of ACS genes had a similar root growth phenotype as the wild-type when growing on MS medium containing ABA. However, the hextuple ACS mutants CS16647 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1) and CS16649 (acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1), and the septuple ACS mutant CS16650 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1) (Tsuchisaka et al., 2009) showed greatly reduced sensitivity to ABA in terms of root growth (Figure 2b). These results further indicate that ABA inhibits root growth by promoting ethylene biosynthesis.

ABA-activated calcium-dependent protein kinases CPK4 and CPK11 phosphorylate ACSs in vitro

Based on the above results, we speculated that ABA may mediate ethylene biosynthesis by modulating the activities of ACSs (Liu and Zhang, 2004; Wang et al., 2004; Yoshida et al., 2005; Joo et al., 2008). We first examined the transcriptional levels of ACS genes under ABA treatment. Of the eight ACS genes whose expression was assessed, expression of three genes (ACS2, ACS7 and ACS8) was induced and that of four genes (ACS4, ACS5, ACS9 and ACS11) was reduced after ABA treatment (Figure 3). The expression of ACS6 was not affected by ABA treatment. The results suggest that ethylene biosynthesis is mediated by ABA at the transcriptional levels; however, how much these ACSs contribute to ABA-mediated ethylene production in roots requires further investigation.

Figure 3.

Expression of ACS genes in wild-type (WT), the abi1 abi2 hab1 triple mutant and the abi1–1 (Col) dominant negative mutant under ABA treatment. Total RNA was extracted from 7-day-old seedlings treated with 0 or 100 μm ABA for 8 h. EF1α was used as a control. Three independent experiments were performed, each with three technical replicates. Values are means ± SD (= 3) from one representative experiment.

Previous studies indicated that ACS stability is tightly controlled by protein phosphorylation. Type 1 ACSs have predicted phosphorylation sites for both MAPK and CDPK protein kinases, and type 2 ACSs only have predicted CDPK phosphorylation sites at the C–terminus (Skottke et al., 2011). We focused on the CDPKs because the biological roles of CDPKs in ethylene biosynthesis are not well understood (Hernández Sebastià et al., 2004; Kamiyoshihara et al., 2010).

Arabidopsis has 34 CDPKs clustered into four sub-groups (I–IV) (Cheng et al., 2002). Among them, CPK4 and CPK11 (sub-group I) are activated by ABA (Zhu et al., 2007). CPK11 modulates the activity of the Shaker pollen inline image channel (Zhao et al., 2013). Loss-of-function mutations of CPK4 and CPK11 resulted in pleiotropic ABA-insensitive phenotypes, including root growth insensitivity to ABA (Zhu et al., 2007). We also included mutants of sub-group II members (CPK21 and CPK23) and sub-group III members (CPK8 and CPK10) as controls. cpk23 mutants are more sensitive to ABA and more drought resistant than the wild-type (Ma and Wu, 2007). Consistent with the previous results (Ma and Wu, 2007; Zhu et al., 2007), the root growth of cpk4–1, cpk11–2 and cpk4–1 cpk11–2 was more resistant to ABA, while the root growth of cpk23 was more sensitive to ABA than that of the wild-type or other mutants (Figure 4a,b). We measured the ethylene production in these mutants. Without ABA treatment, ethylene production was similar among the mutants and the wild-type. However, with ABA treatment, cpk4–1, cpk11–2 and cpk4–1 cpk11–2 produced significantly less ethylene than the wild-type, while cpk8, cpk10, cpk21 and cpk23 produced similar levels of ethylene to the wild-type (Figure 4c). These results indicate that CPK4 and CPK11 are involved in ethylene production.

Figure 4.

ABA promotes ethylene production through CPK4 and CPK11. (a) Root growth of the wild-type (WT) and cpk mutants on MS medium with 0 or 20 μm ABA. Seeds were germinated on MS medium at 21°C for 48 h. The germinated seeds were then moved to MS agar plates containing 0 or 20 μm ABA for 8 days. (b) Root length of the seedlings shown in (b). Three independent experiments were performed with similar results. Each combination of treatment and line in each experiment was represented by three replicate plates, with 10 roots per plate. Values are means ± SD (= 3) from one experiment. Means with different letters are significantly different at < 0.05. (c) Ethylene production of wild-type and cpk mutants treated with 0 or 100 μm ABA. Three independent experiments were performed with similar results. Values are means ± SD (= 3). Each combination of treatment and line in each experiment was represented by three replicate vials, with 10 seedlings per vial. Asterisks indicate statistically significant differences compared with wild-type (*< 0.05; **< 0.01).

We next examined whether CPK4 and CPK11 phosphorylate ACSs in vitro. We chose ACS6 for further study because ACS6 has been well studied in terms of MPK phosphorylation but not CDPK phosphorylation. In an in vivo assay, FLAG–CPK4 and FLAG–CPK11 proteins were first immunoprecipitated from total proteins isolated from FLAG–CPK4 and FLAG–CPK11 transgenic plants treated with ABA or not treated. Then the FLAG–CPK4 and FLAG–CPK11 proteins were analyzed for their ability to phosphorylate ACS6. The results indicate that CPK4 and CPK11 phosphorylate ACS6, and such phosphorylations are stimulated by ABA treatment (Figure 5a). When purified from Escherichia coli, the two CDPK proteins phosphorylated ACS6 in vitro (Figure 5c,d). A recent study showed that ABI1 interacts with ACS6 and reverses the phosphorylation of the ACS6 C–terminus mediated by MPK6 (Agnieszka et al., 2014). However, we found that neither ABI1 nor ABI2 affected the phosphorylation of the ACS6 C–terminus mediated by CPK4 (Figure 5b), indicating that ABI1 or ABI2 do not affect the CDPK phosphorylation of ACS6.

Figure 5.

CPK4 and CPK11 phosphorylate ACS6. (a) CPK4 and CPK11 were activated by ABA treatment. FLAG–CPK4 or FLAG–CPK11 were immunoprecipitated from transgenic plants over-expressing FLAG–CPK4 or FLAG–CPK11. Western blotting was performed using anti-FLAG antibody. ACS6 protein was used as the substrate for the phosphorylation assay. (b) ABI1 and ABI2 do not affect the CPK4 phosphorylations on ACS6. (c,d) Ability of CPK4 (c) and CPK11 (d) to phosphorylate mutant forms of ACS6. (e) Putative CDPK phosphorylation sites indicated by large black letters in the ACS6 C–terminus. The last three serines (underlined) are the phosphorylation sites of MPK3/6. (f) CPK11 purified from E. coli phosphorylates ACS5, ACS7 and ACS11. (g) Mutations of the ACS6 C–terminus or deletion of the C–terminus from ACS6 amino acid 432 does not affect ACS6 activity.

To confirm the CDPK phosphorylation sites in ACS6, we mutated the predicted CDPK phosphorylation site Ser467 at the C–terminus of ACS6 to alanine (Hernández Sebastià et al., 2004), and determined whether CPK4 and CPK11 phosphorylated ACS6 in vitro. Although the relative phosphorylation level was largely reduced, ACS6S467A was phosphorylated by GST–CPK4 and GST–CPK11 purified from E. coli. This result indicates that additional phosphorylation sites exist at the C–terminus of ACS6. To identify other potential phosphorylation sites at the C-terminus of ACS6, we generated recombinant ACS6 proteins with single, double, triple or quadruple mutations of serine (S) to alanine (A). As shown in Figure 5(c,d), mutations of each Ser residue reduced the phosphorylation level, but only the S437A/S462A/S467A/S469A quadruple mutations eliminated almost all the phosphorylation by CPK4 or CPK11. These results demonstrate that S437, S462, S467 and S469 are important for CDPK-mediated phosphorylation of ACS6. The sites phosphorylated by CDPKs are different from those phosphorylated by MPK3 or MPK6 (Liu and Zhang, 2004; Joo et al., 2008) (Figure 5e). We also found that CPK11 purified from E. coli phosphorylated the type 2 ACSs ACS5 and ACS11, as well as the type 3 ACS ACS7, which lack CDPK phosphorylation sites in their C–termini (Figure 5f).

To confirm that phosphorylation of ACS6 by CPK4 and CPK11 alters its enzymatic activity, we measured the activity of GST-tagged ACSs. The activities of ACS6AAAA and ACS6WT were similar, indicating that CDPK phosphorylation does not affect the catalytic activity of ACS6 (Figure 5g). We also found that the arginine at position 431 is the last amino acid required for full activity of ACS6, and that deletion or mutation of this amino acid greatly reduces ACS6 activity (Figure 5g). These results are consistent with previous studies on the C–terminus of ACS6, which indicated that the C–terminus is responsible for protein stability but not for catalytic activity (Liu and Zhang, 2004; Joo et al., 2008).

ACS6DDDD transgenic plants produce increased levels of ethylene

To determine the function of CDPK phosphorylation of the ACS6 C–terminus, we generated transgenic plants expressing wild-type ACS6 (ACS6WT) or the loss-of-phosphorylation form ACS6AAAA (S437A/S462A/S467A/S469A) and phosphorylation-mimicking form ACS6DDDD (S437D/S462D/S467D/S469D). The transcripts of ACS6 were several times more abundant in two wild-type ACS6 transgenic lines than in mutated ACS6 transgenic lines (Figure 6a). Ethylene production was similar in two ACS6AAAA transgenic lines and in the wild-type without ABA treatment (Figure 6b). The two ACS6WT transgenic lines produced more ethylene than wild-type plants with or without ABA treatment (Figure 6b). As was the case with the non-transgenic wild-type, ethylene production in ACS6WT and ACS6AAAA transgenic plants was increased by ABA treatment. The two ACS6DDDD transgenic lines produced more ethylene than the wild-type or ACS6WT and ACS6AAAA transgenic lines but ABA treatment did not increase ethylene production in these two ACS6DDDD transgenic lines (Figure 6b), suggesting that the ethylene production in these two lines reaches a relative high level that cannot be further increased by ABA. Because mutations of the CDPK phosphorylation sites of ACS6 do not affect its catalytic activity, we infer that CDPK phosphorylation increases ACS6 activity by increasing ACS6 stability. As the C terminus of ACS6 determines the protein stability (Liu and Zhang, 2004; Joo et al., 2008), we created transgenic plants carrying GFP or GFP fused with various forms of the C terminal domain (CTD) of ACS6 to assess the protein stability by Western blot analysis. The protein band was hardly detected in transgenic plants expressing GFP–CTDWT or GFP–CTDAAAA under normal conditions, but was detected using GFP antibody after MG132 treatment (Figure 6c). However, in the absence of MG132 treatment, the protein was weakly detected in transgenic plants expressing GFP–CTDDDDD and strongly detected in plants expressing GFP-CTD7D (in which all seven serines are mutated to aspartic acid). A band smaller than GFP protein probably representing a degraded protein was weaker in plants expressing GFP-CTD7D than in plants expressing GFP-CTDDDDD in the absence of MG132 treatment, suggesting that GFP-CTD7D is more stable than GFP-CTDDDDD. Interestingly, we also found that the stability of GFP-CTD7A (in which all sevene serines are mutated to alanine) is comparable to that of GFP-CTD7D, suggesting that, when all serines are mutated to alanine, the protein becomes more stable than wild-type, GFP-CTDAAAA or GFP-CTDDDDD. Consistent with their ethylene production level, ACS6DDDD transgenic plants had shorter roots than the wild-type or ACS6WT and ACS6AAAA transgenic plants in the absence of ABA treatment, and ACS6WT and ACS6DDDD plants were more sensitive to ABA in terms of root growth (Figure 6d,e) but more resistant to ABA in terms of seed germination than wild-type or ACS6AAAA plants (Figure 6f,g). The above results indicate that the phosphorylation-mimicking form ACS6DDDD is more stable than ACS6WT or ACS6AAAA.

Figure 6.

Ethylene production, root growth and seed germination of transgenic plants expressing ACS6 or mutated ACS6s. (a) Determination of relative expression of ACS6WT, ACS6AAAA and ACS6DDDD in various transgenic lines by quantitative RT–PCR. (b) Ethylene production by various transgenic lines with and without 100 μm ABA. Three independent experiments were performed with similar results. Values are means ± SD (= 3). Each combination of treatment and line in each experiment was represented by three replicate vials, with 10 seedlings per vial. Means with different letters are significantly different at < 0.05. (c) Western blot analyses of protein stability. The total proteins were extracted from various transgenic plants expressing GFP or GFP fused to wild-type ACS6 or ACS6 mutated at the C–terminus. GFP antibody was used for Western blotting. Actin detected using an anti-actin antibody was used as a loading control. The arrow indicates GFP–CTD. Asterisks indicate possibly degraded GFP or GFP–CTD. (d) Root growth of the wild-type (WT) and various transgenic seedlings on MS medium supplemented with 0 or 20 μm ABA. Seeds were germinated on MS medium at 21°C for 48 h before transfered to MS medium containing 0 or 20 μm ABA for 8 days. (e) Root lengths of various transgenic seedlings with and without ABA. Three independent experiments were performed with similar results. Each combination of treatment and line in an experiment was represented by three replicate plates, with 10 roots per plate. Values are means ± SD (= 3) from one experiment. Means with different letters are significantly different at < 0.05. (f) Seed germination of various transgenic lines with and without ABA. Approximately 30 seeds were sown on MS medium with 0 or 0.5 μm ABA. The germination greening rates were measured after 8 days in a growth chamber at 21°C. (g) Statistical analysis of seed germination greening of various transgenic lines with and without ABA. Three independent experiments were performed with similar results. Each combination of treatment and line in each experiment was represented by three replicate plates, with 30 seeds per plate. Values are means ± SD (= 3) from one experiment. Means with different letters are significantly different at < 0.05.

Discussion

In genetic screens for mutants with an enhanced response to ABA in terms of seed germination, or for suppressors of ABA-resistant seed germination of the abi1–1 mutant, two groups independently identified EIN2 as a key gene that negatively mediates ABA inhibition of seed germination (Beaudoin et al., 2000; Ghassemian et al., 2000). Their studies established the well-known integration between ABA and ethylene with respect to seed germination. In contrast to their antagonistic modulation of seed germination, ABA and ethylene signaling synergistically mediate root growth. However, both research groups concluded that ethylene signaling rather than ethylene biosynthesis is involved in ABA inhibition of root growth (Beaudoin et al., 2000; Ghassemian et al., 2000). In this study, we showed that root growth in the ethylene over-production mutant eto1 is more sensitive to ABA than that in the wild-type; in contrast, root growth in ethylene biosynthesis mutants is more resistant to ABA than that of the wild-type. Similarly, treatment with the ethylene biosynthesis inhibitor AVG greatly reduces the sensitivity of root growth to ABA. Furthermore, ABA treatment increases ethylene production. These results clearly indicate that high concentrations of ABA inhibit root growth by mediating ethylene production. We consider that use of different experimental conditions and the lack of multiple ethylene biosynthesis mutants explain why different research groups have drawn different inferences regarding the relationship between ABA and ethylene regulation of root growth.

In this study, we found that the ABA-activated protein kinases CPK4 and CPK11 directly phosphorylate the C–terminus of ACS6 in vitro, and that the phosphorylation- mimicking form ACS6DDDD is more stable than the wild-type or the loss-of-phosphorylatable form ACS6AAAA. Both type 1 ACSs (including AtACS1, 2 and 6) and type 2 ACSs (including AtACS4, 5, 8, 9 and 11) have putative CDPK sites. In our in vitro assay, we found that CPK11phosphorylated type 2 ACSs and even the type 3 isoform AtACS7. Although AtACS7 does not have phosphorylation sites at the C-terminus, it does have 14–3–3 omega-binding clients according to a recent proteomic study (Chang et al., 2009). Because 14–3–3 proteins typically regulate their targets in a phosphorylation-dependent manner, the proteomic results suggest that AtACS7 is also phosphorylated in plant cells. Our preliminary results suggest that CDPKs are the candidate protein kinases that target the catalytic domain of AtACS7 and also of type 1 and type 2 ACSs. Because phosphorylation of ACS6 by MPK3/6 dramatically increases the stability and accumulation of ACS6 and greatly increases ethylene production (Liu and Zhang, 2004; Joo et al., 2008), CDPK phosphorylation may only contribute to the total stability of ACSs to a small degree, especially under biotic stress conditions. However, CDPK phosphorylation may contribute to the stability of ACS6 and other ACSs to a greater extent under abiotic stress conditions. Due to the instability of ACS proteins in plant cells, and the low phosphorylation levels by CDPKs, determination of the direct ACS targets of CDPKs in vivo requires further study. Furthermore, the ABA-sensitive phenotype of cpk4, ckp11 and ckp4 cpk11 mutants is not so strong, suggesting that other CDPKs or unknown factors affect the ABA inhibition of root growth. A recent study indicated that the protein phosphatase 2C ABI1 plays crucial roles in restricting ethylene production by reversing the MPK6-mediated phosphorylation of ACS6 (Agnieszka et al., 2014). However, ABI1 does not affect the CDPK-mediated phosphorylation of ACS6. The protein phosphatase 2A is also required for dephosphorylation of the ACS6 C–terminus mediated by MPK6 (Skottke et al., 2011). Whether the PP2A acts on CDPK-mediated phosphorylation of the ACS6 C–terminus requires further study. Together, these results suggest that both CDPKs and MPKs are involved in ABA mediation of ethylene production. It is well established that ABA mediates abscission, senescence and fruit ripening by promoting ethylene production (Cracker and Abeles, 1969; Jiang et al., 2000; Finkelstein, 2013). As CDPKs are highly conserved among dicots and monocots (Hamel et al., 2014), ABA-activated CDPKs may be the primary molecules that regulate ethylene production. Our results indicate that they mediate ethylene production by phosphorylating ACSs, which are the rate-limiting enzymes in ethylene biosynthesis.

By genetic analysis using various mutants, we found that ABA acts upstream of ethylene to regulate root growth because the sensitivity of the root growth of the abi1 abi2 hab1 mutant was inhibited by blocking ethylene production or ethylene signaling. In terms of root growth, abi1–1 is insensitive to ABA, but is as sensitive as the wild-type to ethylene (Ghassemian et al., 2000). Integration between the ethylene and auxin signaling cascades has been well established (Li et al., 2004; Stepanova et al., 2005, 2007, 2008; Ortega-Martinez et al., 2007; Ruzicka et al., 2007; Swarup et al., 2007; Ivanchenko et al., 2008; Stepanova and Alonso, 2009; Strader et al., 2010). Ethylene promotes auxin biosynthesis in roots by regulating the expression of WEI2, WEI7 and WEI8, whose products are involved in auxin biosynthetic pathways (Stepanova et al., 2005, 2008). Ethylene also facilitates basipetal auxin transport to the root elongation zone and auxin distribution in roots (Ruzicka et al., 2007). Previous studies have indicated that ABA inhibits root growth by mediating auxin transport and accumulation in Arabidopsis (Wang et al., 2011b; He et al., 2012). ABA induces accumulation of reactive oxygen species through both plasma membrane-localized NADPH oxidases and the mitochondrial electron transport chain complex I (Kwak et al., 2003; He et al., 2012; Hua et al., 2012). Reactive oxygen species reduce auxin accumulation by directly oxidizing auxin and by inhibiting auxin biosynthesis (He et al., 2012; Pencik et al., 2013). These results suggest that ABA regulates root growth by acting upstream of ethylene but also by directly affecting auxin accumulation and/or auxin signaling. We therefore propose a simple model to explain the integration between ABA and ethylene (Figure 7).

Figure 7.

Proposed model of how ABA promotes ethylene biosynthesis to mediate root growth. ABA signaling activates CDPKs, which phosphorylate ACSs and thus promote ethylene biosynthesis to inhibit primary root growth.

Experimental conditions

Plant materials and growth conditions

Arabidopsis thaliana seeds were germinated and grown on MS medium containing 2% w/v sucrose and 0.8% w/v agar in a growth chamber at 22°C with 22 h light/2 h dark. Seedlings grown for 5–7 days were transplanted into pots containing a mixture of forest soil and vermiculite (2:1). The potted plants were kept under 16 h light/8 h dark in a greenhouse at 21°C.

Mutants eto1 (Wang et al., 2004), etr1–1 (Chang et al., 1993), ein2– (Alonso et al., 1999) and arf2–7 (Okushima et al., 2005) were used in this study. The following T–DNA insertion mutants were also used and were obtained from the Arabidopsis Biological Resource Center (https://abrc.osu.edu/): abi1 (SALK_072009), abi2 (SALK_015166), hab1 (SALK_002104), CS16647 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1), CS16649 (acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1) and CS16650 (acs1–1 acs2–1 acs4–1 acs5–2 acs6–1 acs7–1 acs9–1) (Tsuchisaka et al., 2009). The primers used for identification of the abi1, abi2 and hab1 mutations were as described previously (Hua et al., 2012).

Phenotypic analyses

For the relative root growth assay, 5-day-old seedlings were transferred to MS medium with or without ABA, AVG or other reagents as indicated, and the plates were oriented vertically. The root lengths were measured after a further 3 days.

Seeds were germinated after stratification on MS medium at 4°C for 3 days. After growth for 48 h, the seedlings were transferred to MS medium containing 0 or 20 μm ABA. The plates were oriented vertically, and the root lengths were measured 8 days later.

For the seed germination greening assay, seeds were placed on MS medium with or without 0.5 μm ABA. Each treatment was represented by three replicate plates, each plate with approximately 30 seeds per plate. After 8 days in a growth chamber at 22°C, the number of greening seedlings was counted.

Generation of recombinant constructs and Agrobacterium-mediated transformation

The cDNA of CPK4, CPK11, ACS6 or the mutated ACS6s was amplified by PCR and cloned into pCAMBIA1307 vectors(Cambia, http://www.cambia.org/daisy/cambia/home.html) with a 3 × FLAG epitope tag at the 5′ end. The PCR primers used are listed in Table S1. The CTD domain of ACS6 and its mutants were cloned into a modified pCAMBIA1300 vector (Cambia) with a GFP epitope tag at the 5′ end under the control of a super promoter (Pro super). Transgenic Arabidopsis plants were generated by the floral-dip method (Clough and Bent, 1998) using Agrobacterium tumefaciens strain GV3101 carrying the various constructs.

Preparation of recombinant proteins

The cDNA of CPK4, CPK11, ACS5, ACS7, ACS11, ACS6 or each mutated ACS6 was fused in-frame with GST in the pGEX 4T–1 vector (GE healthcare, http://www.gelifesciences.com), and expressed in E. coli strain BL21 (DE3). The primers used are listed in Table S1. Point mutations were introduced by QuikChange site-directed mutagenesis (Stratagene, http://www.stratagene.com/) and confirmed by sequencing. BL21 (DE3) cells transformed with pGEX 4T–1 constructs were induced using 0.5 mm isopropyl-β–d–thiogalactopyranoside for 12 h at 20°C. The fusion proteins were purified using glutathione–Sepharose 4B (GE healthcare) according to the manufacturer's instructions. GST–CPK4 or GST–CPK11 were eluted using reduced glutathione (AMRESCO, https://www.amresco-inc.com), whereas ACS5, ACS7, ACS11, ACS6 or mutated ACS6 proteins were digested from GST using thrombin (Sigma-Aldrich, http://www.sigmaaldrich.com). The protein concentration was determined using a Bio–Rad protein assay kit (http://www.bio-rad.com).

Gel electrophoresis and immunoblotting

Protein samples in SDS loading buffer were boiled for 5 min before separation by 12% SDS–PAGE. After SDS–PAGE, the proteins on the gels were electrophoretically transferred to Immobilon–P membranes (0.45 μm; Millipore, http://www.millipore.com). The membranes were blocked for 1 h at room temperature using 6% w/v BSA and 0.05% v/v Tween–20 in TBST buffer containing 10 mm Tris/HCl, pH 8.0, and 150 mm NaCl. Anti-FLAG antibody (Sigma-Aldrich) was diluted 1:10 000 in the blocking buffer, and incubated with the membranes for 1 h at room temperature. After washing three times (10 min each) in TBST buffer containing 0.05% v/v Tween–20, the membranes were incubated with goat anti-mouse Ig G-HRP (diluted 1:10 000 in TBST buffer) at room temperature for 45 min, and then washed three times (10 min each) in TBST buffer containing 0.05% v/v Tween–20. The chemiluminescence signal was detected by autoradiography. Actin, immunodetected using anti-actin antibody (Sigma-Aldrich), was used as a loading control.

Analyses of GFP and GFP fused with various CTD forms of ACS6

Seven-day-old T3 Pro super:GFP, Pro super:GFP-CTDWT, Pro super:GFP-CTDAAAA, Pro super:GFP-CTDDDDD, Pro super:GFP-CTD7A and Pro super:GFP-CTD7D transgenic seedlings were treated with 50 μm MG132 or with dimethylsulfoxide as a control. After 24 h, the seedlings were collected and ground to powder in liquid nitrogen. Total proteins were extracted in protein extraction buffer [50 mm Tris/MES pH 8.0, 0.5 M sucrose, 1 mm MgCl2, 10 mm EDTA, 5 mm dithiothreitol and CompleteMini protease inhibitor cocktail tablets (Roche, http://www.roche.com)]. Total proteins (50 μg) were used for immunoblot analysis with an anti-GFP antibody (Roche).

Immunoprecipitation

Ten-day-old seedlings of transgenic plants expressing FLAG–CPK4 and FLAG–CPK11 were treated for 0.5 h with 0 or 100 μm ABA. Total proteins were extracted by grinding 1 g whole seedlings first in liquid nitrogen and then in 4 ml protein extraction buffer (50 mm Tris/MES, pH 8.0, 0.5 M sucrose, 1 mm MgCl2, 10 mm EDTA, 5 mm dithiothreitol and CompleteMini protease inhibitor cocktail tablets). After centrifugation (12 000 g for 20 min) to remove debris, 30 μl anti-FLAG M2 affinity gel (Sigma-Aldrich) was incubated with the extraction supernatant for 3 h at 4°C. After washing the samples three times with 2 ml PBS for 10 min at 4°C, the FLAG–CPK4 or FLAG–CPK11 recombinant proteins were eluted in 50 μl PBS containing 100 μg ml−1 3 × FLAG peptide (Sigma-Aldrich), and then subjected to immunoblotting or used for assessment of kinase activity.

Kinase activity assay

The kinase activity assay was performed as described previously (Hua et al., 2012). Briefly, 0.3 μg kinase and 1.5 μg substrate were mixed with kinase assay buffer (20 mm Tris/HCl, pH 7.5, 10 mm MgCl2, 1 mm CaCl2 and 10 μm ATP) and 0.2 μl [γ-32P] ATP at 30°C for 30 min in a 20 μl volume. The reaction was terminated by adding 7 μl of 4 × SDS loading buffer. The proteins were separated by 12% SDS–PAGE, and the gels were stained with Coomassie Brilliant Blue R250. The kinase activity signals were detected by autoradiography.

Assay of ethylene biosynthesis rates

Five-day-old Arabidopsis seedlings were transferred to 7 ml vials (10 seedlings per vial) containing 2 ml of liquid MS medium, and kept for 24 h at 22°C in a growth chamber with 22 h light/2 h dark. The vials containing seedlings were subsequently treated with 0 or 100 μm ABA, and immediately capped. At the indicated time periods, 2 ml of headspace air was removed with a syringe and injected into GC-17A gas chromatograph (Shimadzu, http://www.shimadzu.com), and the ethylene content was measured as described previously (Vogel et al., 1998). At least three independent experiments were performed, and three vials were used for each line in each experiment.

ACS activity assay

The ACS activity assay was performed using 0.2 μg of fusion protein GST–ACSs in 0.5 ml reaction buffer (50 mm Tris/HCl, pH 8.0, 4 mm dithiothreitol, 20 μm pyridoxal 5′–phosphate and 100 μm S–adenosyl-l–Met) at 30°C for 0.5 h. The reaction was terminated by adding 100 μl of 20 mm HgCl2, followed by 100 μl of a 1:1 mix of saturated NaOH/NaClO (Lizada and Yang, 1979). The vials were capped immediately after addition of the NaOH/NaClO, and were incubated on ice for 2–5 min. Ethylene levels in the headspace of the vials were determined as previously described. At least three independent experiments were performed, each with three vials per line per treatment.

RNA extraction and quantitative RT–PCR analysis

Seven-day-old seedlings were treated with 100 μm ABA for 8 h. Total RNA was extracted using TRIzol reagent (Invitrogen, http://www.invitrogen.com). For quantitative RT–PCR, approximately 3 μg total RNA was digested using RNase-free DNase I (TaKaRa, http://www.takara.com) at 37°C to remove remaining genomic DNA. The digested RNA was then reverse-transcribed to cDNA using MMLV transcriptase (Promega, http://www.promega.com) in a 20 μl reaction. The cDNAs were used as templates in quantitative RT–PCR reactions using SYBR Green Master Mix (TaKaRa) with gene-specific primers and an internal control (EF1α). The primers used for quantitative RT–PCR are listed in Table S2.

Acknowledgements

This work was supported by the National Basic Research Program of China (973 Program; 2012CB114300) and National Nature Science Foundation of China (91117017 and 31121002). We thank Hongwei Guo (Peking University) for providing ein2–5, etr1–1, eto1 mutants, Dapeng Zhang (Tsinghua University) and Wei-Hua Wu for cpk mutants, and the Arabidopsis Biological Resource Center (https://abrc.osu.edu/) for ACS mutants.

Ancillary