KEEP ON GOING, a RING E3 Ligase Essential for Arabidopsis Growth and Development, Is Involved in Abscisic Acid Signaling

The RING-ANK Family Contains a Novel Plant-Specific Member

The RING-HCa domain–containing ANK repeat family contains seven members (Stone et al., 2005), one of which (At5g13530) is predicted to also contain a kinase domain (Figure 1A ). Given the novel arrangement of this protein, we analyzed the corresponding genomic locus more closely. Attempts to amplify the open reading frame (ORF) for At5g13530 from cDNA failed using primers based on the available annotation (http://www.Arabidopsis.org). Complete sequencing of the only cDNA for At5g13530 present in the EST database (APZL47g09R) revealed that it also includes sequence from the downstream locus At5g13540, suggesting that the gene is misannotated, with the full ORF likely encompassing both the At5g13530 and At5g13540 loci (Figure 1B). This fusion was confirmed by the isolation and sequencing of products from RT-PCR generated with seedling RNA and a primer from the 5′ end of the predicted At5g13530 ORF paired with one from the 3′ end of the APZL47g09R cDNA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Reannotation of the KEG Gene.

(A) Scheme and protein similarity tree of the Arabidopsis RING-ANK and kinase-ANK families. The phylogenic tree was generated with ANK repeats. Numbers at branch points represent bootstrap values after 1000 replications. At5g40160 (ANK-only protein) was used as the outgroup. RING-ANK family members tested in in vitro ubiquitylation assays are listed in boldface; those marked with asterisks were confirmed to be active (Stone et al., 2005).

(B) Intron–exon structure of At5g13530 and At5g13540 as determined by The Arabidopsis Information Resource database (top) and reannotation to generate KEG (bottom). Exons are represented by boxes (white, At5g13530; gray, At5g13540) and introns by thick black lines. The cDNA clone APZL47g09R is depicted as a line (below). A scheme of the KEG protein is also shown.

(C) Sequence alignment of the HERC2-like repeats of KEG, an Arabidopsis uncharacterized predicted protein with a HERC region (At4g32250), and similar regions in the HERC2 protein of human (Hs, Homo sapiens) and mouse (Mm, Mus musculus).

The reannotated gene (now referred to as KEG) also has a number of changes in intron–exon organization from the previous prediction (Figure 1B; see Supplemental Figure 1 online). Most notable is the shortening of the last exon of At5g13530 (exon 10), which places the predicted stop codon for At5g13530 within an intron, thus allowing the reading frame to continue into At5g13540 (Figure 1B; see Supplemental Figure 1 online). The final annotation revealed that the KEG locus, from translation start to stop codons, spans 8751 kb with 15 introns and encodes a protein of 1625 amino acids.

Amino acid sequence alignments reveal that KEG is a novel RING protein. In contrast with the six other RING-ANK proteins in Arabidopsis, in which the RING-HCa domain is at the C terminus and is preceded by two to nine ANK repeats (Stone et al., 2005), KEG has the reverse orientation, with an N-terminal RING-HCa domain followed by a kinase domain and nine ANK repeats (Figures 1A and 1B; see Supplemental Figure 2 online). Of the Arabidopsis ANK repeat proteins with either a RING or a kinase domain (Becerra et al., 2004), the kinase-ANK family members formed a single distinct clade that does not include KEG (Figure 1A). The RING-ANK proteins did not fall into a single clade, suggesting more diversity within this subgroup (Figure 1A). The C-terminal region of KEG, which was previously encoded by the At5g13540 locus, was not reported to have any identifiable motifs. However, we detected significant similarity to the mammalian HERC2 (for HECT and RCC1-like) protein, a HECT-type E3 ligase (Garcia-Gonzalo and Rosa, 2005). The similarity between KEG and HERC2 is confined to a 61–amino acid stretch present only once in HERC2 but repeated 12 times in KEG (Figures 1B and 1C). This combination of domains within a single protein is unique to KEG and its orthologs.

Single KEG homologs were found in Oryza sativa (rice; Os KEG), Medicago tuncatula (Mt KEG), and Populus tricocarpa (Pt KEG), but not in fungi or animals, suggesting that KEG is plant-specific (last search, April 31, 2006; see Supplemental Figure 3 online). The predicted rice KEG protein has 60% amino acid identity to Arabidopsis KEG with the same domain architecture, further supporting our reannotation of KEG (see Supplemental Figures 3A and 3B online). The predicted Medicago genome (The Institute for Genomic Research International Medicago Genome Annotation Group) identified three loci that show significant similarity to KEG. The RING-HCa domain, kinase domain, and ANK repeats were encoded by the AC122163_12.1 locus, whereas the HERC2-like repeats were encoded by two downstream loci, AC122163_10.1 and AC122163_11.1. By combining all three genomic regions, a full KEG protein with 69% amino acid sequence identity to Arabidopsis KEG could be assembled, suggesting that the Medicago KEG gene, like its Arabidopsis counterpart, requires reannotation (see Supplemental Figures 3A and 3B online).

KEG Has Both Ubiquitylation and Phosphorylation Activities

The ability of KEG to direct protein ubiquitylation was tested by in vitro ubiquitylation assays (Figure 2A ) (Stone et al., 2005). Given the large size of KEG (∼180 kD), only the region encompassing the RING (R) and kinase (K) domains was assayed as a fusion to glutathione S-transferase (GST-RK). Recombinant GST-RK stimulated E2-dependent ubiquitylation in vitro, as evident by the formation of a heterogeneous collection of high molecular mass proteins detected by immunoblot analysis using anti-ubiquitin antibodies (Figure 2A). Omission of At UBC8 (−E2), GST-RK (−E3), or ubiquitin (−Ub) from the reaction resulted in a loss of protein ubiquitylation (Figure 2A). Substitution of metal ligand positions Cys-3, His-4, and Cys-5 of the RING domain (GST-RKmut) with Ala blocked the ubiquitylation activity of KEG (Figure 2A), indicating that an intact RING-HCa domain is required for ligase activity. In paired reactions with a variety of E2s, GST-RK promoted protein ubiquitylation with the most closely related members of the At UBC8 subgroup—At UBC8, At UBC10, At UBC11, and At UBC28—and very little or no activity was observed with At UBC29 or At UBC30, respectively (Figure 2B), which are more divergent At UBC8 subgroup members (Kraft et al., 2005). No activity was also observed with two other UBCs, UBC34 and UBC35 (Figure 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

KEG Has Ubiquitin Ligase and Kinase Activities.

(A) In vitro ubiquitylation assays using GST fusion protein of KEG RING and kinase regions (GST-RK) and RING mutant GST-RKmut, visualized with anti-ubiquitin antibodies. Omission of At UBC8 (−E2), GST-RK (−E3), or ubiquitin (−Ub) from the complete reactions (C) resulted in a loss of ubiquitylation. The top panel confirms the presence of GST-RK proteins in the assay using anti-GST antibody. Arrows in (A) to (C) identify the slower migrating or ubiquitylated form of GST-RK.

(B) Ubiquitylation assays as in (A) using different E2s. The bottom panel indicates the presence of GST-RK proteins in the assay using anti-GST antibody.

(C) Top, ubiquitylation assays using either ATP or 5′-adenylyl imidodiphosphate (AMP-PNP). Bottom, GST pull down of GST-RK after in vitro ubiquitylation assay followed by immuoblotting with anti-ubiquitin or anti-GST antibody.

(D) In vitro phosphorylation assays using [γ-³³P]ATP.

(E) Calf intestinal phosphatase (CIP) treatment of phosphorylated GST-RK.

(F) In vitro phosphorylation assays were performed in the presence (+) of MgCl₂ and/or MnCl₂.

(G) Phosphorylation of the artificial substrates casein (lane 1) and histone IIa (lane 2) but not myelin basic protein (MBP) (lane 3) by GST-RK. The asterisk indicates the position of MBP.

The predicted kinase domain of KEG contains all of the conserved residues that define the catalytic domain of Ser/Thr protein kinases (Hanks et al., 1988), suggesting that KEG should be active in phosphotransfer as well as ubiquitylation (see Supplemental Figure 2C online). This activity was confirmed by in vitro reactions containing recombinant GST-RK and [γ-³³P]ATP. GST-RK generated an autophosphorylated form whose abundance was sensitive to calf intestinal phosphatase (Figures 2D and 2E). Autophosphorylation was more robust in the presence of Mn²⁺ compared with Mg²⁺, suggesting that the kinase activity of GST-RK prefers Mn²⁺ (Figure 2F). When added to the nonspecific substrates casein, histone IIa, and mylein basic protein, GST-RK phosphorylated the first two substrates but not the last (Figure 2G).

In ubiquitylation assays with GST-RK, we consistently detected a slower migrating form of GST-RK, which could represent a ubiquitylated and/or a phosphorylated form of KEG (arrows in Figures 2A and 2B). This slower migrating form of GST-RK was still observed in assays containing 5′-adenylyl imidodiphosphate, a β,γ-nonhydrolyzable ATP analog that can be used for ubiquitin activation but that does not support phosphotransfer by protein kinases (Johnston and Cohen, 1991; Leventhal and Bertics, 1991), suggesting that it represents a ubiquitylated form of GST-RK (Figure 2C, top panels). To confirm the presence of ubiquitylated GST-RK, we enriched for all GST-RK forms with pull-down assays using glutathione agarose beads mixed with a completed ubiquitylation reaction and subjected the bead-bound fraction to immunoblot analysis with anti-GST and anti-ubiquitin antibodies. Both antibodies detected the slower migrating form, indicating that KEG is capable of autoubiquitylation at least in vitro (Figure 2C, bottom panels).

KEG Is Essential for Postgerminative Growth and Development

To help define the physiological functions of KEG, we analyzed three Arabidopsis (ecotype Columbia [Col-0]) T-DNA insertion lines, keg-1, keg-2, and keg-3. The keg-1 and keg-2 alleles have a T-DNA insert after nucleotide 1623 (from the ATG) within exon 2 and after nucleotide 2368 within intron 3, respectively. keg-3 has a T-DNA insert at nucleotide 5750 within exon 11, previously annotated as the intergenic region between At5g13530 and At5g13540 (Figure 3A ). To examine whether the insertions affected the accumulation of the KEG transcript, portions of the ∼5-kb ORF were amplified by RT-PCR (Figure 3A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The KEG Gene Is Essential for Postgerminative Seedling Growth.

(A) Scheme of KEG depicting the positions of T-DNA insertions. Arrows indicate the positions of primers, with letters designating primer pairs used in separate RT-PCRs. RT-PCR for KEG (top three panels; using primer pairs A, B, and C, respectively) and UBQ10 (RNA control; bottom panel) using total RNA isolated from 10-d-old Col-0 and keg-1, keg-2, and keg-3 seedlings. Mock indicates PCR conducted without reverse transcriptase.

(B) Panel a shows the phenotype of a 7-d-old light-grown keg-1 seedling compared with a wild-type sibling. Panels b to d show seedlings of 7-d-old keg-2 (b) and keg-3 (c) and 3-week-old keg-1 and wild-type Col-0 (d). The inset shows a threefold magnification of the keg-1 seedling shown in panel d. All seedlings were grown on GM with 1% sucrose.

(C) Five-day-old dark-grown keg-1, keg-2, and keg-3 and Col-0 seedlings grown on GM with (top) or without (bottom) 1% sucrose. The graph shows hypocotyl and root length after 5 d in the dark for keg-1 and keg-2 and Col-0. Error bars represent sd; n = 12 except for keg-1, with n = 8 and 7 in 0 and 1% sucrose, respectively.

Levels of the KEG transcript were absent or reduced in these insertion lines. From keg-1 seedling mRNA, no RT-PCR products were produced with primer pairs that spanned the insertion site (primer pairs A and C) or with primers that amplified a region downstream of the insertion site (primer pair B). These results suggest that very little or no keg transcript accumulates in this line. For keg-2 mRNA, a low level of product was detected with a primer pair that amplified a region upstream of the insertion site (primer pair A), but no product was detected with primer pairs B and C that span the insertion site. Because the T-DNA is in the intron, this indicates that the T-DNA is not spliced out, but low levels of RNAs with KEG sequence are produced. Similarly, for the keg-3 allele, low levels of PCR products were generated from cDNA with primers that amplify upstream of the insertion site (primer pairs A, B, and C). Given that the phenotypes of all three keg alleles are identical (see below), if a truncated protein is produced from keg-2 or keg-3 transcripts, it is insufficient to supply wild-type function.

To further support the notion that the mutant phenotypes (see below) were caused by the disruption of KEG and not by an unlinked mutation, heterozygous keg-1, keg-2, and keg-3 plants were backcrossed to wild-type Col-0. PCR-based genotyping of F2 seedlings confirmed that the phenotype conferred by keg segregated with each T-DNA insertion (data not shown). In addition, heterozygous keg plants were indistinguishable from wild-type plants, indicating that the phenotype is recessive.

Seedlings homozygous for each of the three keg alleles displayed the same mutant phenotype (Figure 3B, panels a to c). The germination rate of keg seeds, scored as radicle emergence from the seed coat, was normal (data not shown). However, soon after germination, keg mutants displayed a strong postgerminative growth arrest (Figure 3B). Their cotyledons failed to expand and remained white to very pale green compared with wild-type seedlings of the same age (Figure 3B, panels a to c). The morphology of keg roots was also affected by the mutation; however, keg-1 roots were only slightly shorter and keg-2 roots were not shorter than wild-type roots at 7 d after germination (see Supplemental 4A and 4B online). The emergence of true leaves (with trichomes) was greatly delayed in all three keg backgrounds, requiring almost 3 weeks for the first pair of true leaves to emerge. The second pair of true leaves did not emerge, and no further growth was observed (Figure 3B, panel d).

Loss of KEG Affects Dark-Grown Seedling Development

The lack of cotyledon greening and expansion in keg seedlings could be caused by defects in a number of cellular processes, including a block in photomorphogenesis and/or chloroplast biogenesis. To determine whether the keg mutations affect only light-grown seedlings, the phenotype conferred by keg was examined after 3 or 5 d of growth in the dark on growth medium (GM) containing 1% sucrose (Supplemental Figure 4C online and Figure 3C, respectively). Wild-type dark-grown seedlings undergo a skotomorphogenic growth pattern in which chloroplast development is arrested, the hypocotyl elongates rapidly, the apical hook is maintained, and cotyledon expansion is minimal (von Arnim and Deng, 1996). After 3 d of growth in the dark, hypocotyl elongation and cotyledon expansion of all three keg alleles were so minimal that the cotyledons failed to emerge from the seed coat (see Supplemental Figure 4C online). After 5 d of growth on GM, keg-1 and keg-2 hypocotyls were equivalent in length to each other (Figure 3C, top row and graph; P = 0.71 by Student's t test) but were much shorter than Col-0 hypocotyls (P < 0.0001). The lengths of dark-grown roots were also measured. Although the radicle emerged normally, keg roots were substantially shorter than wild-type roots after 3 d of growth (see Supplemental Figure 4C online). However, the growth of keg roots did recover, and by 5 d both keg-1 and keg-2 roots were similar in length to each other and to Col-0 roots (P = 0.88, 0.80, and 0.47, respectively) (Figure 3C, graph). Thus, it appears that KEG controls a general growth process active in both the light and the dark.

keg Seedlings Are Hypersensitive to Sugar

The addition of 6% glucose to GM has been shown to induce postgermination growth arrest in wild-type Arabidopsis seedlings, with a concomitant block in cotyledon expansion and chloroplast development and greening (Zhou et al., 1998; To et al., 2002). Given that keg seedlings display a similar phenotype when grown on 1% sucrose (Figure 3B), we tested whether keg seedlings are sugar-hypersensitive. As shown in Figure 4A , the addition of 4% glucose to the GM strongly exacerbated the phenotype conferred by keg in light-grown seedlings but only minimally affected wild-type seedlings. The phenotype conferred by keg was not enhanced by 1 or 4% mannitol, a nonmetabolizable sugar, demonstrating that the response was not simply the result of osmotic stress (Figure 4A). Consistent with sugar hypersensitivity, the phenotype conferred by keg was attenuated in medium lacking added sugar. The cotyledons expanded and turned green (Figure 4A), but the mutant seedlings still failed to generate new leaves at wild-type rates. Omission of sugar also affected the dark-grown phenotype of keg seedlings (Figure 3C). At 5 d, keg hypocotyls remained significantly shorter than Col-0 hypocotyls (P < 0.0001) despite the fact that keg-1, keg-2, and Col-0 hypocotyls were all longer when grown in the dark without sugar than with 1% sucrose (P < 0.003). However, the morphology and organization of the keg roots were more similar to those of wild-type roots in the absence of sugar. Thus, the sugar hypersensitivity of keg seedlings may account for the defects observed in the organization of keg roots when grown on sugar-containing medium (see Supplemental Figure 4A online).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

keg Seedlings Display Glucose Hypersensitivity, and the Severity of the Phenotype Is Modulated by Ethylene.

(A) Seven-day-old keg and Col-0 seedlings grown on GM with 0, 1, or 4% glucose or 1 or 4% mannitol in the light.

(B) keg and Col-0 seedlings grown for 7 d in the light on GM with or without 1% sucrose (top three rows and bottom row, respectively) in the presence of 100 μM silver nitrate (AgNO₃) or 50 μM 1-aminocyclopropane-1-carboxylic acid (ACC; second and third rows, respectively). All panels were photographed at the same magnification.

Ethylene Affects the Phenotype Conferred by keg

There is a considerable amount of crosstalk between glucose and ethylene signaling (Leon and Sheen, 2003). For example, ethylene-insensitive mutants such as ethylene-insensitive2 are glucose-hypersensitive (Alonso et al., 1999; Chen et al., 2002), whereas increases in ethylene levels can alleviate the inhibitory effects of high glucose on postgerminative growth (Zhou et al., 1998). Consequently, we tested whether the sugar hypersensitivity of keg seedlings was caused by changes in ethylene perception or synthesis. Blocking ethylene perception by growth in the presence of silver nitrate (Figure 4B) or blocking ethylene synthesis with aminoethoxyvinylglycine (data not shown) strikingly increased the severity of the phenotype conferred by keg in the presence of exogenous sucrose. Treatment of keg seedlings grown on 1% sucrose with the ethylene precursor 1-aminocyclopropane-1-carboxylic acid, which is readily converted to ethylene in plants, partially alleviated the inhibitory effects of sugar on keg seedling development (Figure 4B). However, treatment of keg seedlings with 1-aminocyclopropane-1-carboxylic acid in the absence of glucose did not alleviate the mutant phenotype any further (Figure 4B, bottom right). These results suggested that the KEG mutant phenotype was only partially caused by sugar hypersensitivity and that keg seedlings have a functional ethylene signaling pathway.

keg Mutants Are Hypersensitive to ABA

Glucose has been shown to influence the control of postgerminative growth by the hormone ABA via its ability to increase the expression of ABA biosynthetic and signaling genes (Cheng et al., 2002). In fact, many glucose-insensitive mutations are allelic to ABA-deficient or ABA-insensitive mutations (Finkelstein and Gibson, 2002). Furthermore, when wild-type seedlings are exposed to ABA within 24 h after germination, they undergo a developmental arrest similar to that observed for untreated keg seedlings (Figure 5A ; cf. with Figure 3B). Consequently, keg seedlings could be affected in ABA synthesis and/or perception. Treatment of keg seedlings with the ABA biosynthesis inhibitor fluridone, either during or after germination, failed to alleviate the phenotype conferred by keg (data not shown), suggesting that synthesis of the hormone was not increased. By contrast, we found that the keg roots are hypersensitive to exogenous ABA. When 3-d-old keg-1, keg-2, and keg-3 seedlings were exposed to 5 μM ABA, a stronger inhibition of root growth was apparent compared with growth in the wild type (Figure 5B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

keg Seedlings Are Extremely Sensitive to the Inhibitory Effects of ABA.

(A) Col-0 was allowed to germinate for 24 h and then transferred to GM with (right panels) or without (left panel) 5 μM ABA for 5 d. Percentages indicate seedlings with the shown phenotype (total n = 50).

(B) Col-0 and keg seedlings were grown for 3 d on GM and then transferred to GM with (+) or without (−) 5 μM ABA for 5 d. The graph illustrates the percentage difference in root length between ABA-treated (5 μM) and untreated roots.

The ABA-Responsive Transcription Factor ABI5 Accumulates in keg Seedlings

Postgerminative growth arrest induced by ABA treatment coincides with increases in ABI5 mRNA and protein, implying that ABI5 is the causal agent of this arrest (Soderman et al., 2000; Lopez-Molina et al., 2002). Consistent with previous studies (Lopez-Molina et al., 2002; Smalle et al., 2003), we found that ABI5 protein was undetectable in untreated wild-type seedlings and KEG wild-type siblings (Figure 6A ) and became barely detectable upon exposure to 50 μM ABA (Figure 6C). By contrast, seedlings homozygous for each of the three keg alleles contained extremely high levels of ABI5 even without treatment with exogenous hormone (Figures 6A and 6C). RT-PCR analysis showed that ABI5 transcript levels in keg seedlings were indistinguishable from wild-type levels (Figure 6D), indicating that the increase in ABI5 protein was most likely attributable to posttranscriptional effects, such as reduced degradation of the protein in a background of constitutive synthesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

ABI5 Protein Accumulates in keg Mutants, and KEG Interacts with ABI5.

(A) Levels of ABI5 protein in 6-d-old keg-1, keg-2, and keg-3 seedlings compared with identically grown wild-type siblings (sib), Col-0, ABI5 overexpressor (ABI5-OX), and abi5-1. Anti-PBA1 immunoblot was used as a loading control. PBA1, proteasome β subunit-α

(B) Forms of ABI5 detected in keg-1, keg-2, and keg-3 and ABA0treated rpn10-1 seedlings compared with that found in ABI5-OX seedlings.

(C) Levels of ABI5 protein present in 7-d-old Col, keg, and wild-type siblings in the presence or absence of 50 μM ABA. The left panel shows an equal protein load, and the right panel shows more Col-0 protein extract to visualize ABI5 forms. Anti-PBA1 immunoblot was used as a loading control. The asterisk indicates a form of ABI5 unique to Col-0 treated with ABA.

(D) RT-PCR analysis of ABI5 transcript levels in 6-d-old keg and Col-0 seedlings. UBQ10 transcript was used as a control.

(E) Top, scheme of KEG illustrating regions used in pull-down assays. Plant extracts from 4-d-old abi5-1 and Col-0 seedlings (1 mg of total protein; middle panel) or E. coli lysates containing HA-ABI5 (bottom panel) were used in GST pull-down assays with GST-KEG fusions. Lane − in the bottom panel represents a pull-down assay using beads only. HA-ABI5 represents lysate from E. coli expressing HA-ABI5. IB, immunoblot.

Closer examination of total keg seedling extracts subjected to SDS-PAGE revealed that multiple forms of ABI5 accumulated in the keg-1, keg-2, and keg-3 backgrounds, with apparent molecular masses of ∼50, 51, and 53 kD (Figures 6B and 6C). These forms were present even in extracts prepared under conditions that would minimize posthomogenization proteolysis (hot SDS-PAGE sample buffer), suggesting that they were not artifactually generated during sample preparation. The lower molecular mass protein comigrated with ABI5 that accumulated in an ABI5 overexpression line (ABI5-OX; Figure 6B). ABA treatment of keg seedlings did not generate any new ABI5 species but did slightly increase the accumulation of all preexisting species (Figure 6C). Smalle et al. (2003) previously showed that Arabidopsis seedlings with a mutation in the gene encoding the 26S proteasome subunit RPN10 were also hypersensitive to ABA and had increased levels of ABI5 when treated with ABA. When extracts from ABA-treated rpn10-1 seedlings were examined similarly, these seedlings contained the same three isoforms of ABI5 as seen in keg seedlings without ABA treatment (Figure 6B). Thus, it appears that the two higher mass species represent forms of ABI5 that accumulate when its breakdown is attenuated.

When Col-0 seedlings were treated with ABA, the level of ABI5 increased (Lopez-Molina et al., 2002) (Figure 6C). The migration of ABI5 in Col-0 seedlings treated with ABA was compared with that of untreated keg seedlings (Figure 6C, right). The fastest migrating ABI5 species in Col-0 seedlings comigrated with the fastest form from keg seedlings. In addition, a new slower migrating species became apparent in the ABA-treated Col-0 seedlings that did not correspond to any of the forms seen in untreated or in ABA-treated keg seedlings (Figure 6C, asterisk). Thus, keg affects the nature of the ABI5 proteins present in addition to affecting their abundance.

KEG Interacts Directly with ABI5

The dramatic increase in ABI5 levels in the three keg backgrounds even in the absence of exogenous ABA strongly suggested that KEG is responsible for catalyzing ABI5 ubiquitylation, leading to its subsequent degradation. To determine whether KEG could be directly responsible for ABI5 recognition, we performed GST pull-down assays to determine whether KEG and ABI5 interact physically. We were able to pull down ABI5 from 4-d-old wild-type seedling total protein extracts using a GST fusion protein containing the RING, kinase, and ANK (A) domains of KEG (GST-RKA) (Figure 6E, middle panel). ABI5 was not pulled down using GST-RK, GST-H (HERC repeats), or GST alone (Figure 6E, middle panel). ABI5 was also not detected in pull-down assays using GST-RKA and equivalent protein extract from 4-d-old abi5-1 seedlings (Figure 6E, middle panel), further confirming the identity of the interacting protein as ABI5.

When tested similarly using recombinant hemagglutinin (HA)-tagged ABI5, the same binding preference was observed. HA-ABI5 bound to GST-RKA but not to GST-RK, GST-H, or GST alone (Figure 6E, bottom panel). Because the only protein that interacted contained the ANK repeats, we propose that KEG recognizes ABI5 through one or more of its ANK repeats.

Loss of ABI5 Can Partially Rescue the Phenotype Conferred by keg

The phenotypic similarity of keg mutants in the absence of ABA to that of ABI5-overexpressing plants in the presence of ABA (Lopez-Molina et al., 2003) implied that the phenotype conferred by keg is caused mainly by an overaccumulation of ABI5. To confirm this possibility, we tested whether the loss of ABI5 would rescue some or all of the phenotypes conferred by keg by combining keg-1 with the strong abi5-1 mutant. Consistent with the specific role of ABI5 in regulating ABA signaling, loss-of-function abi5 alleles such as abi5-1 grow well under optimal conditions and are insensitive to concentrations of exogenous ABA that arrest wild-type seedlings (Finkelstein, 1994; Lopez-Molina and Chua, 2000). Homozygous abi5-1 plants were crossed to KEG/keg-1 plants, and F2 progeny homozygous for abi5-1 were identified by their insensitivity to ABA during germination (Finkelstein, 1994). ABA-insensitive plants (abi5-1) were then genotyped by PCR (n = 198) for the keg-1 allele.

F2 abi5-1 plants heterozygous for the KEG mutation (KEG/keg-1) were allowed to self, and F3 progeny were analyzed. Strikingly, this F3 population did not segregate for the phenotype conferred by keg when grown on medium containing 1% sucrose where the phenotype conferred by keg was readily apparent (Figures 3 and 4). By contrast, we detected segregation of a new phenotype that was much milder than that of keg-1. Among seedlings with normally expanded and green cotyledons was a population of seedlings whose hypocotyls were noticeably shorter than wild-type hypocotyls (Figure 7A ). Genotyping showed that the short seedlings (n = 16) were homozygous for keg-1, whereas the wild-type seedlings (n = 104) were either wild type or heterozygous for keg-1 (Figure 7A). Thus, loss of ABI5 was able to substantially, but not completely, rescue the growth-arrest phenotype of the keg-1 seedlings.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Loss of ABI5 Rescues the Growth-Arrest Phenotype of keg Seedlings.

(A) Top, phenotypes of seedlings recovered in the F3 generation from seeds generated by self-pollination of abi5-1 KEG/keg-1 plants. Bottom, genotypes of seedlings as determined by PCR analysis. Lines 1 and 2 refer to progeny from crosses with different KEG/keg parents.

(B) Top, growth of Col-0, abi5-1, and progeny of abi5-1 KEG/keg-1 plants in 3 μM ABA. Bottom, PCR-determined genotypes of ABA-insensitive (Ins) and ABA-sensitive (S) seedlings.

The same F3 seeds were germinated in the presence of 3 μM ABA (Figure 7B). In contrast with homozygous abi5-1 seedlings that are insensitive to this concentration of ABA, a population of ABA-sensitive seedlings was evident in progeny from an abi5-1 KEG/keg-1 parent (Figure 7B). All ABA-sensitive F3 seedlings genotyped (n = 24) were homozygous for keg-1, whereas ABA-resistant seedlings (n = 24) were KEG or KEG/keg-1 (Figure 7B). Thus, ABA sensitivity was restored in keg-1 abi5-1 seedlings. This sensitivity explains our previous inability to identify double mutant F2 seedlings, because we first used ABA insensitivity to select for seedlings homozygous at the abi5-1 locus (see above).

To confirm that the restoration of ABA sensitivity by the loss of KEG was specific for abi5 mutants and not for those affecting other components of ABA signaling, we generated a similar mutant combination with abi4-1. abi4-1 eliminates the expression of ABI4, a transcription factor that also works as a positive regulator of ABA signaling (Finkelstein, 1994; Finkelstein et al., 1998). Unlike ABI3, which forms a complex with ABI5, there is no evidence that ABI4 interacts with ABI5 (Nakamura et al., 2001; Lopez-Molina et al., 2002). F3 progeny of an abi4-1 KEG/keg-1 plant grown on medium containing 1% sucrose segregated for the phenotype conferred by keg, indicating that the loss of ABI4 cannot rescue the phenotype conferred by keg (data not shown).

Sequence Analysis and Domain Definition

The GENSCAN database and manual editing of nucleotide sequences were used to predict ORFs. Domains were identified using the SMART (Simple Modular Architecture Research Tool) and Pfam (Protein families database of alignments and HMMs) databases and by BLAST searches via the National Center for Biotechnology Information. Alignments were generated with the ClustalX program (Thompson et al., 1997); the alignments were revised using the Se-Al sequence editor (Evolutionary Biology Group, University of Oxford). Phylogenetic trees were created by PAUP* (Phylogenetic Analysis Using Parsimony) 4.0 using the neighbor-joining method with 1000 bootstrap replicates.

RT-PCR Cloning and Mutagenesis

Total RNA isolated from Arabidopsis thaliana ecotype Col-0 10-d-old seedlings using the RNeasy plant RNA extraction kit (Qiagen) according to the manufacturer's instructions was used in RT-PCR to amplify the predicted KEG ORF. For RT-PCR for ABI5 (At2g36270), the TRIzol reagent (Invitrogen) extraction method was used to isolate total RNA, according to the manufacturer's instructions. For both KEG and ABI5 cDNAs, 5 μg of total RNA was reverse-transcribed using SuperScript III RNase H reverse transcriptase (Invitrogen), and the resulting cDNAs were used in PCR (see Supplemental Table 1 online for a list of all primers used). The Gateway system (Invitrogen) was used to clone the full-length KEG cDNA (4878 bp) and partial cDNA regions of KEG encoding the RING and kinase (RK; 1383 bp), RING, kinase, and ANK (RKA; 2489 bp), or HERC2-like (H; 2390 bp) domain. KEG cDNAs were introduced into the pDEST15 vector to produce in-frame fusions with the GST tag. To obtain 3× HA-ABI5, the entire ABI5 ORF was amplified from PYAt2g36270 (ABRC) and introduced into pHB1-HA3 (Liu et al., 1998) via the NdeI/BamHI restriction sites. Arabidopsis E2s were described previously (Kraft et al., 2005). For RING mutational analysis, site-directed mutagenesis (Stratagene) was used to make a series of point mutations, Cys to Ala, within the RING region.

Wild-type and mutated KEG cDNAs and the EST cDNA clone APZL47g09R (Asamizu et al., 2000), whose partial sequence was deposited previously (GenBank accession number AV529785), were sequenced in their entirety by automated sequencing.

Protein Expression and Purification

Fusion proteins were expressed in Escherichia coli strain BL21(DE3) pLysS and GST fusion proteins purified using GST-agarose (Sigma-Aldrich) as described previously (Stone et al., 2005). Arabidopsis E2s were expressed and purified as 6× His-tagged fusion proteins as described previously (Kraft et al., 2005).

In vitro Ubiquitylation and Phosphorylation Assays

Ubiquitylation assays were performed as described previously (Hardtke et al., 2002). Briefly, reactions (30 μL) containing 50 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 0.05 mM ZnCl₂, 1 mM ATP or 5′-adenylyl imidodiphosphate (Sigma-Aldrich), 0.2 mM DTT, 10 mM phosphocreatine, 0.1 unit of creatine kinase (Sigma-Aldrich), 50 ng of yeast E1, 250 ng of purified 6× His-E2, 500 ng of eluted/bead-bound GST-E3, and 2 μg of ubiquitin (Sigma-Aldrich) were incubated at 30°C for 2 h. Reactions were stopped by adding sample buffer (125 mM Tris-HCl, pH 6.8, 20% [v/v] glycerin, 4% [w/v] SDS, and 10% [v/v] β-mercaptoethanol) and analyzed by SDS-PAGE followed by protein gel blotting using anti-ubiquitin antibodies as described previously (Stone et al., 2005).

In vitro phosphorylation assays used 250 ng of GST-RK or GST incubated at 30°C for 30 min in 30 μL of buffer (20 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM DTT, 0.1% Triton X-100, and 10 μCi of [γ-³³P]ATP) supplemented with 10 mM MnCl₂, 10 mM MgCl₂, or 5 mM MnCl₂ and 5 mM MgCl₂. Reactions were stopped by adding sample buffer and boiled for 5 min. Reactions were analyzed by SDS-PAGE, and gels were stained with Coomassie Brilliant Blue R 250 and dried onto Whatman paper. Phosphorylated proteins were detected by autoradiography or phosphoimagery. For treatment with calf intestinal phosphatase, bead-bound GST-RK was first allowed to autophosphorylate as described above. Beads were then washed in calf intestinal phosphatase buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM MgCl₂, and 1 mM DTT) and incubated with or without 0.01 unit/μL calf intestinal phosphatase for 1 h at 37°C. For substrate phosphorylation, 5 μg of casein, histone IIa, or MBP (Sigma-Aldrich) was added to the in vitro assay.

Plant Material and Growth Conditions

Seeds from Arabidopsis ecotype Col-0 or mutant plant lines surface-sterilized with 30% (v/v) bleach and 0.1% (v/v) Triton X-100 were grown on GM containing 1% (w/v) agar with 1× Murashige and Skoog medium under continuous light. GM contains 1% sucrose unless stated otherwise. Selected seedlings were transferred to soil and grown under photoperiodic cycles of 16 h of light and 8 h of dark at 16°C with 50% RH. For treatments, components were omitted or GM was supplemented as indicated in the text.

Identification of T-DNA Insertional Plants and Microscopy

T-DNA insertional lines were obtained from the ABRC (Alonso et al., 2003). keg1 to keg3 are T-DNA insertion lines (SALK_049542, SALK_018105, and SALK_133445, respectively), all in the Col-0 ecotype background. Plants and seedlings with the T-DNA were identified by PCR analysis of genomic DNA using the Extract N Amp kit (Sigma-Aldrich). Genotyping consisted of PCR using gene-specific primers for the wild-type allele and in combination with T-DNA–specific primers for the mutant allele. PCR products for each of the three lines were sequenced using the T-DNA–specific primers to determine the exact genomic location of the insert. Images of seedlings were acquired using a DC290 digital camera mounted on a Zeiss Stemi SV8 microscope.

ABA Root Assay and Screen

For root assays, wild-type Col-0 and keg seedlings were germinated and grown vertically on solid GM for 3 d. Seedlings were then transferred to vertical GM plates with or without 5 μM ABA (Sigma-Aldrich), and subsequent root growth was measured after 5 d. For ABA screens, seeds were sterilized and grown in liquid GM supplemented with or without 3 μM ABA under continuous light for 7 to 10 d.

Antibody Generation and Immunoblot Analysis

Antibodies against purified ABI5 were produced in a rabbit (Polyclonal Antibody Service, University of Wisconsin-Madison) using recombinant 6× His-ABI5 (a gift of T. Thomas) expressed in E. coli BL21(DE3) pLysS (Novagen). Log-phase cultures were induced with 1 mM isopropylthio-β-galactoside at 37°C for ∼4 h. Insoluble ABI5 protein was extracted using the Bug Buster reagent according to the manufacturer's recommendations (Novagen), solubilized in 8 M urea buffer, and then purified at room temperature via nickel-nitrilotriacetic acid agarose chromatography (Qiagen Sciences). ABI5 eluates were dialyzed, concentrated, and resolved by SDS-PAGE, and gel slices containing the antigen were injected directly into the rabbit. For immunoblot analysis, total plant protein was isolated from 6- to 7-d-old seedlings grown in liquid GM. rpn10-1 plants were treated with 1 μM ABA for 2 h before harvesting. keg-1, keg-2, and keg-3 and wild-type sibling seedlings were treated with 50 μM ABA for 1 h before harvesting. Tissue was frozen and then homogenized in 2:1 (volume:fresh weight) sample buffer, and extracts were analyzed by protein gel blotting with anti-ABI5 antibodies. Anti-PBA1 (proteasome β subunit-α) antibodies were used to confirm equal protein loads (Smalle et al., 2002).

GST Pull-Down Assays

For GST pull-down assays, bead-bound GST or GST-KEG fusions were incubated at 4°C with plant extracts from Col-0 or abi5 seedlings (1 mg of total protein) or bacterial lysate in PBS containing HA-ABI5. Plant extracts were prepared by homogenizing frozen tissue in 50 mM Tris-HCl, pH 8, 150 mM NaCl, 20 mM EDTA, pH 8, 1 mM DTT, 10% glycerol, 1% Nonidet P-40, 50 mM phenylmethylsulfonyl fluoride, and a Complete mini protease inhibitor cocktail tablet (Roche Diagnostics). Extracts were cleared by centrifugation, and protein concentration was determined by Bradford assay. After incubation, beads were washed four times in 1× PBS with 1% Triton X-100. For pull downs from plant extracts, proteins were eluted from beads in SDS sample buffer. After incubation with bacterial lysate, GST-KEG fusion and interacting proteins were eluted from beads using 100 mM glutathione in 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.01% Triton X-100. ABI5 was visualized using anti-ABI5 antibodies.

Accession Number

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number DQ315360.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure 1. Nucleotide Sequence of the Reannotated At5g13530 and At5g13540 Genomic Region.

• Supplemental Figure 2. Illustration of KEG, and Amino Acid Sequence Alignments of Each Domain.

• Supplemental Figure 3. Amino Acid Sequence Alignment of Arabidopsis KEG and Homologous Proteins.

• Supplemental Figure 4. Phenotypes of Light-Grown keg Seedling Roots and Dark-Grown keg Seedlings.

• Supplemental Table 1. Primers Used in This Study.