The plant hormone cytokinin plays essential roles in many aspects of growth and development. The cytokinin signal is transmitted by a multi-step phosphorelay to the members of two functionally antagonistic classes of Arabidopsis response regulators (ARRs): type B ARRs (response activators) and type A ARRs (negative-feedback regulators). Previous studies have shown that mutations in AXR1, encoding a subunit of the E1 enzyme in the RUB (related to ubiquitin) modification pathway, lead to decreased cytokinin sensitivity. Here we show that the cytokinin resistance of axr1 seedlings is suppressed by loss of function of the type A ARR family member ARR5. Based on the established role of the RUB pathway in ubiquitin-dependent proteolysis, these data suggest that AXR1 promotes the cytokinin response by facilitating type A ARR degradation. Indeed, both genetic (axr1 mutants) and chemical (MLN4924) suppression of RUB E1 increased ARR5 stability, suggesting that the ubiquitin ligase that promotes ARR5 proteolysis requires RUB modification for optimal activity.
Cytokinins (CKs), a group of N6–substituted adenine derivatives, play a crucial role in the regulation of various aspects of plant growth and development (Mok and Mok, 2001; Kamada-Nobusada and Sakakibara, 2009; Werner and Schmülling, 2009; Perilli et al., 2010). CK signaling involves a multistep phosphorelay that is similar to two-component signaling mechanisms found in prokaryotes and lower eukaryotes (Hwang and Sheen, 2001; Heyl and Schmülling, 2003; Kakimoto, 2003; Argueso et al., 2010; Perilli et al., 2010). In Arabidopsis, CKs are perceived by ER membrane-localized histidine kinase receptors (Inoue et al., 2001; Suzuki et al., 2001; Yamada et al., 2001; Caesar et al., 2011; Wulfetange et al., 2011). Binding of CK to a receptor triggers auto-phosphorylation on a conserved His, and subsequent transfer of the phosphoryl group to a conserved Asp within the receptor's receiver domain. Histidine phosphotransfer proteins then accept the phosphoryl group from the receptor and transfer it to a conserved Asp residue in Arabidopsis response regulators (ARRs) (Hwang and Sheen, 2001; Hutchison et al., 2006; Mähönen et al., 2006; Punwani et al., 2010).
Although all components of the CK signaling system are encoded by multi-gene families, the ARR gene family is particularly large (Argueso et al., 2010). The ARR5 gene is one of the first plant response regulator genes described: Urao et al. (1998) identified ARR5 based on homology with prokaryotic response regulators, and showed that ARR5 expression is up-regulated by cold, dehydration and salt stress. Simultaneously, ARR5 was identified by differential display as a gene that is rapidly induced by CKs, but not by any other plant hormone (Brandstatter and Kieber, 1998). Subsequently, 24 ARR genes were identified and divided into two main clades, the type B and type A ARRs (D'Agostino and Kieber, 1999; Imamura et al., 1999). Type B ARRs are composed of an N-terminal receiver domain, a central Myb-like DNA binding domain, and a C-terminal transactivation domain (Sakai et al., 2000; Heyl and Schmülling, 2003). Type B ARRs activate the transcription of primary CK response genes, which include the type A ARRs that encode partially redundant negative regulators of CK signaling (Hwang and Sheen, 2001; Sakai et al., 2001; Heyl and Schmülling, 2003; To et al., 2004; Mason et al., 2005). The induction of type A ARR genes creates a negative-feedback loop that guarantees that the strength and the duration of the CK response can be controlled.
In contrast to our understanding of the general function of the components of the CK response pathway, our understanding of the proteolytic control of the CK response remains superficial. Evidence for the involvement of proteasome-dependent proteolysis in CK signaling was initially obtained through characterization of proteasome mutants (Smalle et al., 2002, 2003). The rpn12a-1 and rpn10-1 mutants have defective subunits of the 26S proteasome regulatory particle, and, as a result, have lower degradation rates of polyubiquitinated proteins (Smalle et al., 2002, 2003; Kurepa et al., 2008). Both proteasome mutants have decreased CK sensitivity, suggesting that the activity of a CK response inhibitor is controlled by proteasome-dependent proteolysis (Smalle et al., 2002, 2003). Studies involving ectopically expressed type A ARRs revealed that some family members (e.g. ARR5) are unstable proteins, and that their half-lives are extended by CK treatments (To et al., 2007; Ren et al., 2009; Ryu et al., 2009). Subsequent proteasome inhibitor studies revealed that unstable type A ARRs are degraded by the proteasome (Ren et al., 2009; Ryu et al., 2009).
Analyses of the function of the RUB (related to ubiquitin) modification pathway provided additional evidence for the involvement of ubiquitin-dependent proteolysis in the regulation of CK signaling. RUB is a peptide post-translational modifier that is attached to its targets by an isopeptide bond between its C-terminal glycine and a lysine of the target protein (Downes and Vierstra, 2005; Hotton and Callis, 2008). The first step in this reaction is activation of RUB through an ATP-dependent mechanism catalyzed by the heterodimeric E1 enzyme AXR1–ECR1 (del Pozo et al., 1998). Activated RUB is subsequently transferred from the E1 enyzyme onto the E2 conjugating enzyme RCE, and then to a target protein (Hotton and Callis, 2008). A prominent class of protein targets modified by RUB are the cullins (CULs), which that function as scaffolds for assembly of the multi-subunit ubiquitin CUL–RING E3 ligases (CRLs) (del Pozo and Estelle, 1999; Hotton and Callis, 2008). RUB modification of CULs is thought to increase CRL activity, thus accelerating the degradation of CRL target proteins (Hotton and Callis, 2008). Previous reports showed that axr1 mutants have decreased sensitivity to auxin, but also to other hormones, including CKs (Lincoln et al., 1990; Abel et al., 1995; Timpte et al., 1995; Tiryaki and Staswick, 2002). The auxin resistance of axr1 mutants is caused by stabilization of AUX/IAA proteins that repress the auxin response (Gray et al., 2001). However, the underlying reason for the CK insensitivity remains unknown. Here we show that axr1 CK resistance involves stabilization of the type A ARR member ARR5, suggesting that the decreased sensitivity results from an enhanced feedback-inhibition mechanism that over-rides response activation.
Isolation of suppressor of ipt-161 1 (soi1)
To identify additional factors that control CK signaling, we screened for mutations that suppress the developmental effects caused by increased endogenous CK production. We used the transgenic line ipt-161 (C24 background) that expresses an Agrobacterium isopentenyltransferase (IPT), an enzyme that catalyzes the first step of CK biosynthesis (van der Graaff et al., 2001). Light-grown, 7-day-old ipt-161 seedlings resemble the CK-treated wild-type: they have short roots, thick hypocotyls and small cotyledons (Figure 1a) (van der Graaff et al., 2001).
To isolate CK response mutants, we mutagenized ipt-161 seeds with ethyl methane sulfonate (EMS), plated approximately 50 000 (1 g) M2 seeds on MS/2 medium (see Experimental procedures), and rescued seedlings with cotyledons that were larger than those of ipt-161. Here we describe the extragenic ipt-161 suppressor soi1. In addition to cotyledon size, the soi1 mutation also reversed the short-root phenotype of ipt-161 (Figure 1a). However, the phenotype of mature ipt-161 soi1 plants differed from the wild-type: overall rosette size was reduced, apical dominance was decreased, and plants were semi-sterile (Figure 1b). The soi1 mutation also suppressed the accumulation of anthocyanins in the upper hypocotyl region of ipt-161 seedlings (Figure 1c). Next, we analyzed the effects of soi1 on plant development in the absence of the ipt-161 transgene. Cotyledons of 7-day-old soi1 seedlings were similar in size to those of the wild-type, but the roots were longer and resembled those of ipt-161 soi1 (Figure 1a). Mature soi1 plants were also similar to ipt-161 soi1 plants: they had smaller rosettes, a short bushy inflorescence and severely reduced seed yield (Figure 1d).
To explore the mechanism by which soi1 suppresses the ipt-161 phenotype, we first tested whether soi1 affects the primary CK response by comparing the expression level of the ARR5 gene in wild-type (C24), soi1, ipt-161 and ipt-161 soi1 seedlings (Figure 1e). ARR5 is rapidly induced by CKs, and its kinetics of induction may be used as a diagnostic tool for CK sensitivity (Brandstatter and Kieber, 1998). The ARR5 gene (and other type A ARRs) is also constitutively up-regulated in plants over-expressing CK biosynthesis genes (e.g. AtIPT4 and AtIPT8; Sun et al., 2003; Li et al., 2010) and down-regulated in Arabidopsis ipt T-DNA mutants (Nishiyama et al., 2011). Thus, type A ARR expression levels may also serve as a marker for sustained changes in CK action. Our RNA gel-blot analyses showed that the ARR5 mRNA level was approximately threefold higher in ipt-161 seedlings than in the wild-type, but similar to the wild-type in ipt-161 soi1 seedlings (90% wild-type levels), and approximately 60% of the wild-type level in soi1 (Figure 1e). These results suggested that the soi1 mutation suppresses ipt-161 either by inhibiting the primary CK response or by repressing CK accumulation.
The soi1 mutant has decreased sensitivity to cytokinins
To determine whether soi1 affects CK signaling, we first assessed the CK sensitivity of the mutant seedlings by testing the effects of kinetin on root growth (Figure 2a,b). Whereas 0.1 μM kinetin inhibited root elongation in the wild-type by approximately 70%, this dose had a significantly lower effect (approximately 30%) on the root length of soi1 seedlings. The CK insensitivity of soi1 roots was also apparent at a higher kinetin concentration, suggesting a strong defect in CK signaling (Figure 2a,b). Next, we analyzed the CK induction kinetics of the ARR5 gene in soi1 (Figure 2c). We treated wild-type and soi1 seedlings with 5 μm benzyladenine (BA), and found that the ARR5 induction levels were approximately fourfold lower in the mutant (Figure 2c). These results suggest that the soi1 mutation causes decreased CK sensitivity by suppressing the primary CK response.
The soi1 mutation is located in the AXR1 gene
Our molecular and phenotypic analyses (Figures 1 and 2) indicated that the soi1 mutation affects a gene encoding a component of the primary CK response pathway. To isolate the soi1 locus, we used a map-based cloning strategy. We crossed soi1 (C24) with the Col-0 ecotype, identified 106 F2 plants resistant to 10 μm kinetin (Figure S1), and mapped the soi1 mutation to the top of chromosome 1 (Figure S2a). One of the genes in this region is AXR1 (At1 g05180). Loss of AXR1 function causes an increase in primary root elongation and decreases in shoot apical dominance and fertility, and leads to root insensitivity to multiple hormones, including CKs (Lincoln et al., 1990). Because the soi1 mutant has similar phenotypes (Figures 1 and 2), we compared the AXR1 sequence from the mutant with that of the C24 wild-type, and found a missense mutation G3179A in the AXR1 gene of the soi1 mutant (Figure S2b). This mutation resulted in a glutamic acid (Glu) to lysine (Lys) substitution in the C-terminal domain of the AXR1 protein (Figure S2c). Based on the similarity of AXR1 to its human homolog (amyloid precursor protein binding protein 1), the C-terminal domain that carries a mutation in soi1 is required for formation of the AXR1–ECR1 heterodimer (Walden et al., 2003).
Although the best-described phenotype of axr1 mutants is their decreased sensitivity to auxin, these mutants are also insensitive to other hormones, including CKs and ethylene (Lincoln et al., 1990; Abel et al., 1995; Timpte et al., 1995; Tiryaki and Staswick, 2002). As we isolated soi1 based on its decreased CK sensitivity, we next tested soi1 responses to auxin and ethylene. Similar to previously described axr1 mutants, soi1 exhibited reduced sensitivity to indole-3-acetic acid (IAA) and 2,4-dichlorophenoxyacetic acid (2,4-D) and to the ethylene biosynthesis precursor 1-aminocyclopropane-1-carboxylic acid (ACC) (Figure S3).
Previous studies showed that axr1 mutants are less sensitive to the effects of BA in a root elongation response assay (Timpte et al., 1995; Tiryaki and Staswick, 2002). However, reduced CK sensitivity in roots may be a secondary effect of decreased ethylene sensitivity, for example (Cary et al., 1995; van der Graaff et al., 2001; Chae et al., 2003). To clarify the cause of the CK insensitivity, we tested the effects of CK on (1) rosette growth, (2) anthocyanin accumulation, and (3) type A ARR expression in the weak axr1-3 and the strong axr1-12 mutant (Figure 3). Consistent with the difference in strength of the mutations (Lincoln et al., 1990), a higher kinetin dose was required to affect the shoot growth of axr1-12 mutant plants compared to axr1-3 or the wild-type (Figure 3a). Also consistent with the strength of the mutations, axr1-3 and axr1-12 were also less responsive to BA in an anthocyanin accumulation assay (Figure 3b). Whereas wild-type seedlings grown on 0.2 μm BA accumulated approximately sixfold more anthocyanins compared to the untreated control, the anthocyanin levels were increased approximately fourfold and approximately threefold by 0.2 μm BA treatment in axr1-3 and axr1-12, respectively (Figure 3b). Finally, real-time RT-PCR analyses showed that CK induction of the ARR5 gene was attenuated in both axr1-3 and axr1-12 seedlings (Figure 3c). Thus, similar to the soi1 mutation (Figures 1 and 2), the axr1-3 and axr1-12 mutations result in decreased CK sensitivity by suppressing the primary CK response.
Next, we tested whether the axr1-3 and axr1-12 mutations also suppress the ipt-161 phenotype (Figure 4). We first introgressed the ipt-161 transgene in Col-0, and then generated the double homozygous ipt-161 axr1-3 and ipt-161 axr1-12 lines. Both axr1-3 and axr1-12 alleles restored the shoot size of ipt-161 seedlings back to the wild-type level, with no visible difference in the strength of the suppression phenotypes (Figure 4a). The anthocyanin content, which was approximately sevenfold higher in ipt-161 (Col-0) compared to the wild-type, was also restored to the wild-type level in both ipt-161 axr1-3 and ipt-161 axr1-12 (Figure 4b). We also analyzed the callus induction response of the ipt-161 axr1 lines. Simultaneous treatment of explants with a specific ratio of CK and auxin is known to promote the formation of white callus. By gradually increasing the dose of CKs, explants first develop green calli, and then shoots (Valvekens et al., 1988; Thorpe, 2006). We reasoned that a CK-over-producing line such as ipt-161 may form green calli on medium that contains only auxin. Indeed, after 40 days of incubation on 0.5 μm 1-naphthaleneacetic acid (NAA), ipt-161 hypocotyl explants developed green calli, whereas the hypocotyls from wild-type Col-0 plants did not (Figure 4c). The ipt-161 axr1-3 and ipt-161 axr1-12 hypocotyl explants also did not develop calli on auxin-supplemented medium, confirming that the effect of the increased CK content in ipt-161 was attenuated by both axr1 mutations (Figure 4c). Finally, we compared the type A ARR steady-state gene expression levels in Col-0, ipt-161, ipt-161 axr1-3 and ipt-161 axr1-12 plants (Figure 4d and Figure S4a). RNA gel-blot analyses showed that, similar to ipt-161 (C24) and the ipt-161 soi1 lines (Figure 1e), the ARR5 steady-state level in ipt-161 (Col-0) plants was higher than in the wild-type, and was suppressed by the axr1-3 and axr1-12 mutations (Figure 4d). Quantitative RT-PCR confirmed these results, and also revealed that, similar to soi1, the steady-state ARR5 level in the axr1 single mutants was lower compared to the wild-type (Figure S4b). Similar accumulation trends were detected for transcripts of the type A ARR genes ARR4 and ARR6 (Figure S4b). CK responses that are known to involve CK/ethylene and CK/auxin interactions, such as root hair formation, root elongation and the de-etiolation response (Cary et al., 1995; van der Graaff et al., 2001; Chae et al., 2003), were also suppressed by the axr1-3 and axr1-12 mutations (Figure S5). We conclude that the CK insensitivity of soi1 and axr1 mutants is caused by a defect in early CK signaling, and that this defect affects hormone sensitivity throughout plant development.
Chemical inhibition of the RUB modification pathway suppresses ipt-161
To independently test the requirement for AXR1 in CK signaling, we tested whether the ipt-161 developmental and molecular phenotypes are reversed by MLN4924, an inhibitor of the RUB E1 enzyme (Soucy et al., 2009; Bennett et al., 2010; Hakenjos et al., 2011). To that end, we have germinated and grown ipt-161 (Col-0) seedlings on medium containing MLN4924, and analyzed seedling morphology and the expression of type A ARR genes (Figure 5). The ipt-161 phenotype was partially suppressed by MLN4924: whereas a lower dose of the inhibitor (10 μm) led to near-complete restoration of cotyledon size and partial suppression of the short-root phenotype, a higher dose (25 μm) appeared to be toxic and particularly affected root growth (Figure 5a). Quantitative RT-PCR analyses revealed that MLN4924 also suppressed the constitutive up-regulation of the type A ARR genes ARR4, ARR5 and ARR6 in ipt-161 (Figure 5b). Thus, both genetic studies and pharmacological assays showed that the RUB pathway is required for early CK signaling.
Inhibition of the RUB pathway promotes ARR5 accumulation
Previous studies have shown that the AXR1–ECR1 heterodimer is inactive in strong axr1 mutants, which prevents RUB modification of the CUL subunit of CRL-type ubiquitin ligases and leads to a decrease in ligase activity (del Pozo et al., 1998; Hotton and Callis, 2008). In our case, this suggests that the decrease in CK sensitivity of axr1 mutants is caused by accumulation of a CK response repressor that is normally targeted to the proteasome by a CRL-type E3. Because several type A ARRs (e.g. ARR4, ARR5 and ARR6) have been shown to be unstable proteins that accumulate in response to treatment with a proteasome inhibitor or CKs (To et al., 2007; Ren et al., 2009; Ryu et al., 2009), we analyzed the effects of RUB pathway inhibition on type A ARR stability. We selected ARR5 as a type A ARR representative because over-expression of ARR5 typically leads to a stronger CK resistance than over-expression of other tested type A ARRs (To et al., 2007; Ren et al., 2009).
We generated anti-ARR5 antisera, analyzed the endogenous ARR5 levels in untreated and BA-treated Col-0 and arr5-1 seedlings, but were unable to detect a protein of the correct size (approximately 21 kDa) that was present in the wild-type and not in the mutant. To improve our chances of detecting ARR5, we then generated plants ectopically expressing an N-terminally FLAG-tagged ARR5 (Figure S6a). In accordance with previous studies (To et al., 2007; Ren et al., 2009), the CK sensitivity of FLAG-ARR5 over-expression (FLAG-ARR5ox) lines was decreased compared to the Col-0 wild-type (Figure S6b,c). The FLAG–ARR5 levels increased after BA and MG132 treatments, consistent with previous analyses describing the stability of Myc-tagged ARR5 (Figure S6d,e; To et al., 2007; Ren et al., 2009).
Next, we tested the stability of FLAG–ARR5 in response to treatments with the RUB pathway inhibitor. Immunoblotting analyses showed that treatment with MLN4924 promoted a gradual increase in FLAG–ARR5 (Figure 6a). Cycloheximide (CHX) chase experiments (Kurepa and Smalle, 2011) further revealed that the MLN4924-induced accumulation of FLAG–ARR5 was the result of a decrease in its degradation rate (Figure 6b). Finally, by introgressing the FLAG-ARR5 transgene into the axr1-3 background, we also found that FLAG–ARR5 accumulates in response to genetic suppression of the RUB pathway (Figure 6c). Moreover, root growth response assays confirmed that the CK resistance of FLAG-ARR5ox seedlings was indeed enhanced in the axr1-3 background (Figure 6d). Taken together, these results show that ARR5 degradation by the 26S proteasome requires a functional RUB modification pathway. In addition, the increased accumulation of a CK response repressor potentially provided an explanation for the decreased CK sensitivity of axr1 mutants.
Cytokinin insensitivity of axr1 requires ARR5
To further investigate the role of type A ARRs in the CK resistance of axr1 mutants, we generated a homozygous axr1-3 arr5-1 line, and tested the effects of CKs on type A ARR expression, root length, root hair formation and anthocyanin accumulation (Figure 7). We first analyzed the CK induction of ARR4 and ARR6 genes in the arr5-1, axr1-3 and axr1-3 arr5-1 backgrounds (Figure 7a). Surprisingly, we observed a decreased induction of ARR6 in the arr5-1 mutant. The molecular mechanism that leads to this change of ARR6 expression in arr5-1 remains to be analyzed. Consistent with the induction of the ARR5 gene (Figures 1e and 2c), ARR4 and ARR6 induction was suppressed in axr1-3 plants treated for 30 min with 5 μM BA. Although this molecular phenotype was reversed in arr5-1 axr1-3 plants, the reversion was only partial (Figure 7a). Next, we analyzed CK growth responses of the single and double mutant lines. Compared to the axr1-3 mutant, the CK sensitivity of axr1-3 arr5-1 seedlings was increased in the root elongation assay (P < 0.001), anthocyanin accumulation tests (P < 0.001), and root hair growth responses (P < 0.0001) (Figure 7–b–d). However, the addition of arr5-1 only partially suppressed the CK insensitivity of axr1-3, suggesting that accumulation of other type A ARRs also contributes to this phenotype. To test this hypothesis, we generated the pentuple mutant axr1-3 arr3,4,5,6, and compared its CK-induced growth responses to those of axr1-3 arr5-1 (Figure S7). Loss of function of ARR3, ARR4 and ARR6 led to a moderate further increase in axr1-3 suppression that was significant only at higher CK doses for all the tested responses (Figure S7a,b). Thus, comparison of the overall reversion effects of the arr5 single versus the arr3,4,5,6 quadruple mutations suggested that, whereas stabilization of multiple type A ARRs contributes to the CK insensitivity of axr1-3, the stabilization of ARR5 is its major cause.
In this study, we describe isolation of the CK-insensitive mutant soi1 that suppresses the effects of CK over-production in the ipt-161 transgenic line. The soi1 mutation (Glu487→Lys; Figure S2) causes a strong pleiotropic phenotype similar to that of the axr1-12 mutant (Gln416→stop; Leyser et al., 1993). Both mutations are localized at the four-helix-bundle domain of AXR1, which is required for formation of the AXR1/ECR1 heterodimer (del Pozo et al., 1998; Walden et al., 2003), suggesting that the RUB activating enzyme is not formed in soi1 and axr1-12. Weak mutations such as axr1-3 (Cys154→Tyr) presumably affect AXR1 function without preventing assembly of the E1 heterodimer. Currently, the best-described targets of the RUB pathway are the cullins (CULs), which serve as the backbone of the large ubiquitin CUL–RING E3 ligase (CRL) family, and require RUB conjugation for proper assembly into these multi-subunit complexes (del Pozo and Estelle, 1999; Hotton and Callis, 2008). Hence, independent of the actual molecular mechanism of RUB E1 inactivation, the overall result is expected to be stabilization of a subset of 26S proteasome targets and alteration of a spectrum of physiological processes that are governed (or fine-tuned) by the affected target proteins.
Although it has been reported that axr1 mutations affect the CK response (Timpte et al., 1995; Tiryaki and Staswick, 2002), the molecular basis of this decreased sensitivity remains unknown. Here, we show that loss of AXR1 function causes CK insensitivity by directly affecting the primary CK response pathway. More specifically, we show that loss of AXR1 leads to stabilization of the negative-feedback regulator ARR5. The degradation rate of ARR5 was attenuated by chemical suppression of the RUB conjugation pathway (Figure 6). Furthermore, ARR5 accumulated both in axr1 and wild-type seedlings treated with the RUB pathway inhibitor, thus providing an explanation for the decreased CK sensitivity caused by loss of AXR1 function. These data are complemented by the finding that the arr5-1 mutation partially suppresses the axr1 CK resistance (Figure 6). However, the higher-order mutant arr3,4,5,6 did not substantially improve this axr1 suppression phenotype (Figure S7). This suggests that ARR4 and ARR6, which have been shown to be unstable proteins targeted for proteasome-dependent proteolysis (Figure 7) (To et al., 2007; Ren et al., 2009; Ryu et al., 2009), contribute little to the CK phenotype of axr1 mutants, and that other type A ARRs are involved. Alternatively, it remains possible that part of the axr1 CK resistance is caused by increased accumulation of currently unknown CK response repressors that do not belong to the type A ARR family.
It has been shown that CKs promote both the transcription of type A ARRs and the stability of at least a subset of type A ARR proteins, thus engaging a robust feedback-inhibition mechanism that limits the strength of the response (D'Agostino et al., 2000; To et al., 2007; Ren et al., 2009; Ryu et al., 2009). We propose that, in axr1 mutants, this feedback-inhibition mechanism is enhanced due to increased ARR5 accumulation, which over-rides the response activation mechanism, thus leading to CK insensitivity. Essential to this CK resistance phenotype is that it becomes stronger in response to CK treatment.
One seemingly paradoxical aspect of the proposed mechanism is that ARR5 protein accumulation is higher in axr1 mutants, whereas ARR5 transcript abundance is lower compared to the wild-type. However, recent studies in many organisms have shown that the correlation between transcript and protein abundance is quite low for many genes, particularly genes that encode unstable regulatory proteins (Vogel and Marcotte, 2012). In fact, a low correlation may also be inferred from the cytokinin induction profile of the ARR5 gene. Cytokinin treatment of ARR5 typically leads to an induction peak that is followed by a gradual decrease in transcript level due to feedback inhibition mediated by the type A ARR proteins, and the decline in ARR5 transcription reflects the increased abundance of the ARR5 protein (D'Agostino et al., 2000). Thus, the lower ARR5 transcript levels in both untreated and CK-treated axr1 plants may be viewed as a direct consequence of the increased ARR5 protein level.
The Arabidopsis genome encodes several classes of ubiquitin ligases, some of which are represented by large gene families (Vierstra, 2009). Our results suggest that one or more members of the CRL family play a key role in type A ARR proteolysis, and that they require AXR1-dependent RUB modification of their CUL subunit for optimal activity. These hypothetical CRLs probably contain the CUL1 isoform as loss of function of the corresponding gene was reported to cause CK root insensitivity (Moon et al., 2007). In conclusion, our work confirms the importance of the CRL family in hormonal regulation of plant growth.
Plant materials and growth conditions
Depending on the line investigated, Arabidopsis thaliana Columbia-0 (Col-0) and C24 accessions were used as wild-type controls. The ipt-161 (C24) line has been described previously (van der Graaff et al., 2001). ipt-161 (Col-0) was generated by introgressing the transgene from ipt-161 (C24) into Col-0, and used for genetic and biochemical analyses after six rounds of back-crossing. Except for soi1, which was isolated in the C24 background, all other mutants used in this study [axr1-3 and axr1-12 (Lincoln et al., 1990), and arr5-1 and arr3,4,5,6 (To et al., 2004)] were in the Col-0 background, and were obtained from the Arabidopsis Biological Resource Center (https://abrc.osu.edu/). To isolate homozygous ipt-161 axr1-3, ipt-161 axr1-12, axr1-3 arr5-1 and axr1-3 arr3,4,5,6 mutants, F2 individuals were tested by PCR genotyping (Tables S2 and S3), followed by sequencing of the amplified fragments for axr1-3 and axr1-12. To generate over-expressor lines, a full-length ARR5 cDNA was amplified for Gateway cloning (Life Technologies) using gene-specific primers containing attB sites (Table S3). The cDNA was recombined into pDONR221 (BP reaction), and transferred into pEarlyGate202 (LR reaction), which carries the l-phosphinotricine (Basta) resistance gene, and allows 35S promoter-driven expression of N-terminally FLAG-tagged proteins (Earley et al., 2006). Arabidopsis Col-0 plants were transformed by the floral-dip method (Clough and Bent, 1998). For further analyses, we selected two lines. After screening F2 seedlings of a cross between axr1-3 and over-expressing FLAG–ARR5 line 2, we found no double mutants and concluded that the T-DNA insertion carrying the over-expression construct is located close to axr1-3. For further analyses, we generated a homozygous double mutant between axr1-3 and FLAG–ARR5 line 1.
For all experiments, surface-sterilized and stratified seeds were sown on half-strength Murashige and Skoog medium (pH 5.7) containing 1% sucrose and 0.8% PhytoAgar (MS/2 medium), and were grown at 22°C under continuous light at 80 μmol m−2 sec−1. For growth on soil, plants were transferred to a 1:1 mix of Miracle Grow potting soil and vermiculite, and grown at 22°C under continuous light (200 μmol m−2 sec−1). All chemicals used for the treatments [kinetin (Sigma, http://www.sigmaaldrich.com), benzyladenine (BA; Sigma), 1-naphthaleneacetic acid (NAA; Sigma), indole-3-acetic acid (IAA; Sigma), 2,4-dichlorophenoxyacetic acid (2,4-D, Sigma), 1-aminocyclopropane-1-carboxylic acid (ACC; Sigma), MG132 (Enzo Life Sciences, http://www.enzolifesciences.com), cycloheximide (CHX; Sigma) and MLN4924 (Active Biochem, http://www.activebiochem.com)] were prepared as stock solutions in dimethylsulfoxide.
EMS mutagenesis, suppressor screen and map-based cloning
EMS mutagenesis was performed as previously described (Kim et al., 2006). In brief, ipt-161 (C24) seeds were imbibed in 0.1% KCl for 12 h, mutagenized with 80 mm EMS (Sigma) for 3 h, washed twice for 15 min each in 100 mm sodium thiosulfate, rinsed in water, dried for 12 h, and sown on soil. Progeny M2 seeds were harvested, sterilized, plated on MS/2 medium at low density, kept for 4 days at 4°C, and transferred to a growth chamber. Lines with suppression phenotypes were transferred to soil and allowed to self-pollinate. The suppressor of ipt-161 (soi) mutant lines were out-crossed three times before further analysis.
For mapping, F2 progenies from a cross between soi1 (C24) and Col-0 were used. F2 seeds were sterilized, sown on MS/2 plates, and, after 3 days growth, transferred to MS/2 plates containing 10 μm kinetin. Plates were positioned vertically, and plants were grown for another 7 days. DNA was isolated from 106 kinetin-resistant seedlings, and mapping was performed using microsatellite markers (Bell and Ecker, 1994), INDEL markers (Jander et al., 2002) and markers for single nucleotide polymorphisms between Col-0 and C24 sequences (Table S1). The candidate gene was amplified from both the soi1 mutant and the C24 wild-type, and sequenced using Genome Lab™ Dye Terminator Cycle Sequencing with a Quick Start Kit (Beckman Coulter Inc.).
For hormone response assays, 4-day-old seedlings were transferred to fresh MS/2 plates containing either dimethylsulfoxide (solvent control) or CK (kinetin or BA). For root length analyses, root tips were marked at the moment of transfer to test plates, the plates were positioned vertically in a growth chamber, and the plants were photographed after 5 or 12 days of growth. The root length was measured from the transfer mark using ImageJ software. For rosette growth analysis, plants were photographed and analyzed after 18 days of growth. For anthocyanin quantification, seedlings were collected after 8 days of growth on test plates, weighed, and used for isolation of total flavonoids as described previously (Kubasek et al., 1992). The anthocyanin content was measured at 520 nm using a DTX 880 multimode detector (Beckman Coulter, http://www.beckmancoulter.com). For the callus induction assay, 4-day-old etiolated seedlings were transferred to light, grown for 2 days, and used to dissect hypocotyls. Hypocotyls were transferred to full-strength MS medium containing 0.5 μm NAA and 2% sucrose, and incubated for 40 days. For all experiments, descriptive statistics, plotting and hypothesis testing were performed using prism 5.0d or prism 6 software (GraphPad Software Inc., http://www.graphpad.com).
RNA isolation, gel-blot analyses and quantitative RT-PCR
Total RNA was prepared using Trizol reagent (Invitrogen) from Arabidopsis seedlings grown with continuous shaking (20 rpm) in liquid Gamborg's B5 medium supplemented with 1% sucrose (pH 5.7). The RNA gel-blot analyses were performed as described previously (Smalle et al., 2002). The ARR5 and ARR6 antisense probes were prepared from the Arabidopsis Biological Resource Center plasmids 103N10 and 138J22 linearized with KpnI + PstI using SP6 RNA polymerase. The hybridization signal was visualized using a Pharos FX™ Plus molecular imager system (Bio-Rad, http://www.bio-rad.com). For quantitative RT-PCR, RNA treated with TURBO DNase (Ambion, http://www.invitrogen.com) was reverse-transcribed with RNA to cDNA EcoDry™ Premix (Clontech, http://www.clontech.com). The cDNA equivalent of 20 ng total RNA was used in a 10 μl reaction with the DyNAmo Flash SYBR Green qPCR kit (Finnzymes, http://www.thermoscientificbio.com) on a StepOne real-time PCR system (Applied Biosystems, http://www.appliedbiosystems.com). The primers used are listed in Table S3. For each experiment, ACT2 (At3 g18780), GAPDH (At1 g13440), At3 g53090 (ubiquitin transferase) and At4 g33380 (expressed protein) were also amplified using previously described primers (Czechowski et al., 2005), and the best reference gene(s) was selected using geNorm (Vandesompele et al., 2002). PCR efficiency was estimated using linreg pcr software (Ramakers et al., 2003; Ruijter et al., 2009), and the mean efficiency of each amplicon group was used for calculation. Three biological replicates of each sample were tested, and reactions were performed using two technical replicates.
Protein isolation, antibody generation and immunoblotting analysis
Protein extraction and immunoblotting analyses were performed as previously described (Kurepa and Smalle, 2011). Anti-ARR5 antibodies were generated in rabbits against the C-terminal amino acids 85–184, and were affinity-purified before use (SDIX; Brown et al., 2011). For immunoblotting analyses, membranes were first incubated with the anti-ARR5 antibody, and then with horseradish peroxidase-conjugated anti-rabbit IgG goat antibodies. Immunoblotting with anti-FLAG antibodies (clone M2, Sigma) used at 1:10 000 produced a weaker signal than immunoblotting with anti-ARR5 antibodies used at the same dilution. The signal was developed using SuperSignal West Femto substrate (Thermo-Pierce, http://www.piercenet.com) using a ChemiDoc™ XRS molecular imager (Bio-Rad). The signal intensities were measured using QuantityOne software (Bio-Rad) as described previously (Kurepa and Smalle, 2011).
This work was supported by grants from the US National Institute of Food and Agriculture National Research Initiative (2005-35304-16043) and the US National Science Foundation (0919991). We thank the Arabidopsis Biological Resource Center (Columbus, OH) for providing EST clones and seeds of mutant and transgenic lines.