- © 2013 American Society of Plant Biologists. All rights reserved.
Abstract
Regulation of transcriptional processes is a critical mechanism that enables efficient coordination of the synthesis of required proteins in response to environmental and cellular changes. Transcription factors require accurate activity regulation because they play a critical role as key mediators assuring specific expression of target genes. In this work, we show that CULLIN3-based E3 ligases have the potential to interact with a broad range of ETHYLENE RESPONSE FACTOR (ERF)/APETALA2 (AP2) transcription factors, mediated by MATH-BTB/POZ (for Meprin and TRAF [tumor necrosis factor receptor associated factor] homolog)-Broad complex, Tramtrack, Bric-a-brac/Pox virus and Zinc finger) proteins. The assembly with an E3 ligase causes degradation of their substrates via the 26S proteasome, as demonstrated for the WRINKLED1 ERF/AP2 protein. Furthermore, loss of MATH-BTB/POZ proteins widely affects plant development and causes altered fatty acid contents in mutant seeds. Overall, this work demonstrates a link between fatty acid metabolism and E3 ligase activities in plants and establishes CUL3-based E3 ligases as key regulators in transcriptional processes that involve ERF/AP2 family members.
INTRODUCTION
Effective mechanisms that control the timing of developmental and physiological processes and regulate these processes in response to environmental cues are of utmost importance to plants due to their sessile lifestyle. One mechanism that allows plants to quickly and flexibly respond is the ubiquitin (UBQ) proteasome pathway (Hua and Vierstra, 2011). This pathway is highly conserved among eukaryotes and requires the concerted activities of an E1 UBQ activating enzyme, a UBQ conjugating enzyme E2, and an E3 UBQ ligase. While E1 and E2 activate the UBQ to modify target substrates, the E3 ligase binds the E2 and a substrate protein to facilitate transfer of the UBQ moiety. Upon building up a UBQ chain on the substrate, the ubiquitylated protein is marked for degradation via the 26S proteasome (Hua and Vierstra, 2011).
CULLIN3 (CUL3)-based Really Interesting New Gene (RING) E3 ligases (CRL3) have been described only recently and mainly with respect to their basic architecture (Figueroa et al., 2005; Gingerich et al., 2005, 2007; Weber et al., 2005). They are composed of a CUL3 protein, as the scaffolding subunit, that binds in its C-terminal region the RING-finger protein RING-Box protein 1 (RBX1), while its N-terminal part is recognized by proteins containing a BTB/POZ fold (Figueroa et al., 2005; Weber et al., 2005). BTB/POZ proteins comprise a diverse group of proteins within Arabidopsis thaliana and rice (Oryza sativa), containing 80 and 149 members, respectively (Gingerich et al., 2007). They have been divided into 12 subgroups based on their secondary domains (Gingerich et al., 2007). While the BTB/POZ fold is required for assembly with the cullin and to interact with other BTB/POZ proteins, the secondary domain may function as an adaptor to allow binding of a substrate and delivery to the CRL3 core for ubiquitylation.
Based on its role as the central scaffolding subunit that assembles with potentially many BTB/POZ proteins, it is not surprising that the loss of CUL3 causes an embryo-lethal phenotype. Reduced amounts of functional CUL3 protein affect red light and ethylene signaling and impair plant development (Dieterle et al., 2005; Thomann et al., 2009).
One BTB/POZ subfamily is the BTB/POZ-MATH (BPM) family, which contains a BTB/POZ fold in their C-terminal region and a MATH domain located within the first 200 amino acids of their N-terminal region. The family comprises six members in Arabidopsis, all of which have molecular masses between 40 and 50 kD (Weber et al., 2005). Recent findings indicate that substrates of BPM proteins, and, thus, of a corresponding CRL3BPM E3 ligase, are predominantly transcription factors, as assembly with RELATED TO APETALA2.4 (RAP2.4), a member of the ETHYLENE RESPONSE FACTOR/APETALA2 (ERF/AP2) family, and with HB6, a class I homeobox Leucine zipper transcription factor, have been reported (Weber and Hellmann, 2009; Lechner et al., 2011). While the role of BPM/RAP2.4 assembly remains elusive, the assembly with Homeobox protein 6 (HB6) appears to be a critical step in abscisic acid (ABA) signaling (Lechner et al., 2011).
Here, we describe that Arabidopsis BPM proteins assemble widely with ERF/AP2 transcription factors, and we demonstrate with a selected member of this family, WRINKLED1 (WRI1), that the interaction is a requirement to destabilize WRI1 in plants. BPM proteins are largely necessary for normal development, and loss-of-function mutants affected in all six members are affected in metabolism and fatty acid content. The results provide in planta proof for CRL3BPM E3 ligase activity affecting one of the major plant-specific transcription factor families, thus emphasizing their central role in plant metabolism and physiology.
RESULTS
BPM Proteins Interact Broadly with ERF/AP2 Transcription Factors
We earlier described that BPM proteins assemble with several, but not all members, of the A6 group of ERF/AP2 transcription factors (Weber and Hellmann, 2009). According to Sakuma and coworkers, the ERF/AP2 superfamily can be divided into five subgroups: AP2, RELATED TO ABI3/VP1 (RAV), DEHYDRATION-RESPONSIVE ELEMENT BINDING (DREB), ERF, and others (Sakuma et al., 2002). The A6 group belongs to the ERF subfamily. To investigate how broadly BPM proteins assemble with ERF/AP2 transcription factors, we also tested additional members outside the A6 group in yeast two-hybrid (Y2H) assays using BPM1 as prey (Figure 1A). The ERF subfamily member ERF1 (At3g23240) showed weak interaction, while the ERF subfamily members WRINKLED1 (At3g54320) and ERF4 (At3g15210) showed a strong interaction with BPM1 in the Y2H assay. WRI1 also tested positively for self-assembly in the yeast assay (Figure 1B). Finally, DREB1a (At4g25480), which belongs to the DREB subfamily, as well as RAV1 (At1g13260), a member of the RAV subfamily, strongly interacted with BPM1 (Figure 1A). Since BPMs also interact with CUL3 proteins, we tested the interaction of the different ERF/AP2 transcription factors with CUL3a and did not observe any interaction in the yeast system. We therefore concluded that a large number of ERF/AP2 transcription factors are likely recognized by BPM proteins in Arabidopsis.
Interaction Studies of WRI1 with BPM and CUL3 Proteins.
(A) BPM1 interacts with ERF4, RAV1, and DREB1a, but only poorly with ERF1, in Y2H assays. SDII, medium for transformation selection; SDIV, medium for test of interaction. Pictures of single spots were taken 7 d after transformation.
(B) WRI1 can assemble with itself in Y2H assay, as well as with representative members of the BPM family, BPM1, BPM3, BPM4, and BPM5.
(C) In vitro–translated and [35S]Met-labeled BPM1 protein was used in pull-down assay with E. coli–expressed GST and GST:WRI1.
(D) Pull-down experiments with E. coli–expressed GST:WRI1 results in the precipitation of WRI1 and CUL3, while GST alone was ineffective. Asterisks indicate GST:WRI1 band, while the band below is plant WRI1. Pulldowns were first tested with α-WRI1 and then with α-CUL3 after the membrane had been stripped. WT, the wild type.
(E) Silver-stained SDS-PAGE gel to illustrate the Isopropyl-β-D-thiogalactopyranosid (IPTG)-induced expression of purified GST and GST:WRI1 proteins from E. coli. Asterisks mark the bands of the corresponding proteins.
(F) Pull-down (PD) experiments with purified proteins show that His:WRI1 can precipitate GST:CUL3a if GST:BPM1 is present in the assay (left blot), while this is not the case when GST alone is used (right blot). Blots were first probed with an α-GST antibody, then stripped and subsequently probed with α-CUL3 and α-WRI1.
(G) IP experiments with α-WRI1 antibody shows coprecipitation of CUL3 with WRI1 from Arabidopsis wild-type protein extract. If not otherwise stated in this and subsequent figures, 30 μg of total protein extract was loaded as input, and experiments were done with 14-d-old seedlings.
(H) WRI1 is present in the plant extract used for the IP shown in (G). Ponceau in this and all subsequent figures refers to unspecific staining of blotted proteins with Ponceau S and was used as a loading control. Results shown in this and all subsequent figures are based on at least three independent biological replicates.
CUL3 and BPM Proteins Assemble in Planta with WRI1
To understand the basic principles of CRL3BPM complex assembly with substrates, we focused on a single well-described protein, WRI1, which is a key player in fatty acid and carbohydrate metabolism (Cernac and Benning, 2004; Baud et al., 2009).
In agreement with findings from the Y2H assays, pull-down experiments using a glutathione S-transferase (GST):WRI1 fusion protein demonstrated that this protein can coprecipitate in vitro–translated BPM1 from rabbit reticulolysates (Figure 1C). For further investigation of in planta complex assembly, specific peptide-based antibodies were raised against CUL3 and WRI1 (see Supplemental Figure 1 online). The antibody against CUL3 did not distinguish between CUL3a and b (see Supplemental Figures 1A and 1C online); however, transient expression experiments in Nicotiana benthamiana clearly demonstrated specificity of the α-CUL3 antibody. We only observed a single band of around 85 kD appearing on protein gel blots with wild-type or cul3 mutant plant extracts (see Supplemental Figure 1B online). Likewise, only a single band was detectable when the α-WRI1 antibody was used on total plant extracts from wild-type plants, which was missing in a wri1-3 mutant when the α-WRI1 antibody was used on total plant extracts (see Supplemental Figure 1E online).
Pull-down experiments using GST:WRI1 against wild-type plant extract showed that the fusion protein precipitates both endogenous WRI1 and CUL3; however, this was not the case with GST alone (Figure 1D). In addition, pull-down analysis with GST- and His-tag proteins expressed in and purified from Escherichia coli demonstrated that BPM proteins are necessary to bridge assembly between WRI1 and CUL3 proteins. Here, His:WRI1 is only capable of coprecipitating GST:CUL3a, if GST:BPM1 protein is present in the assay but not with GST alone (Figure 1F). Also, immunoprecipitation (IP) studies using the α-WRI1 antibody successfully precipitated CUL3 from plant extracts in the wild-type background (Figure 1G). We also observed comigration of WRI1 with CUL3 in fast protein liquid chromatography (FPLC) studies (see Supplemental Figure 2A online) and detected green fluorescent protein (GFP):WRI1 localized to the nucleus, as has been previously shown for most BPM proteins and CUL3a (see Supplemental Figure 2B online; Weber and Hellmann, 2009). Overall, these studies demonstrate that WRI1 assembles with CUL3 proteins in planta and support the working hypothesis that the assembly is mediated by BPM proteins.
WRI1 Is a CUL3-Dependent Target of the 26S Proteasome
One outcome of BPM assembly with CUL3 proteins is the proteolytic degradation of their substrates. Consequently, stability assays were performed using the translational inhibitor cycloheximide (CHX) and the proteasomal inhibitor MG132. In initial experiments, CHX treatments did not point to instability of the WRI1 protein (see Supplemental Figure 3A online), although accumulation of WRI1 was observed when plants were treated with MG132 (see Supplemental Figure 3A online).
To better understand this phenomenon, WRI1 expression was tested in plants treated with CHX or with the proteasomal inhibitor. While plants incubated with MG132 did not show any change in WRI1 expression, CHX caused a strong upregulation of the WRI1 gene (see Supplemental Figure 3B online). It was therefore decided to pretreat plants with the transcriptional inhibitor actinomycin D2 (ActD2) before CHX treatment. Under these conditions, WRI1 protein was completely gone after 6 h treatment, and its disappearance was blocked by coincubation with MG132 (Figure 2A), demonstrating that WRI1 is unstable in a 26S proteasome–dependent manner. Notably, the accumulation of WRI1 protein in samples treated with all three inhibitors is most likely due to pretreatment of plants with ActD2 and MG132 for 3 h before the addition of CHX.
Stability and Expression Level of WRI1 in the Wild Type and cul3hyp.
(A) Treatment of wild-type (WT) plants with ActD2 (6 h) and CHX (3 h) inhibitors shows instability of WRI1. This is blocked by cotreatment with MG132 (6 h).
(B) WRI1 protein accumulates in cul3hyp double mutants in comparison to the wild type.
(C) Expression of WRI1 is comparable to the wild type in cul3hyp plants.
(D) WRI1 protein is more stable in cul3hyp than in the wild type. Error bars in this and all subsequent figures represent sd.
A cul3hyp double mutant that was previously described is knocked out for CUL3b and partially functional for CUL3a (Thomann et al., 2009). We took advantage of this mutant to investigate whether WRI1 is stabilized in this genetic background and to prove that the instability is mediated by a CUL3-based complex. Protein gel blot analysis on wild-type and cul3hyp plant extracts showed that WRI1 was present in the mutant in higher amounts than in the wild type (Figure 2B). This was not based on increased transcriptional activities in the mutant since no significant difference in WRI1 expression was detectable in either plant (Figure 2C). Stability assays with ActD2 and CHX showed no considerable change in protein content in the mutant over a 6-h treatment period, while WRI1 was not detectable in wild-type extracts (Figure 2D), revealing that WRI1 is unstable in a 26S proteasome– and CUL3-dependent manner.
BPM Proteins Bridge the Assembly between WRI1 and CUL3 and Are Broadly Important for Development
Our results strongly support the idea that BPM proteins assemble with a broad range of ERF/AP2 transcription factors and that, if ERF/AP2 proteins are in complex with CUL3s, the BPMs likely function as their bridging substrate receptors. Since WRI1 is unstable, and because complex formation requires presence of a functional CUL3 protein in the plant, it was necessary to investigate whether the loss of BPM proteins is also stabilizing WRI1.
Two strategies were followed to support the hypothesis that BPM proteins function as substrate receptors and are required for mediating WRI1 instability. First, because of a lack of T-DNA insertion mutants for nearly all BPM genes, a 35S artificial microRNA construct was designed to downregulate expression of all six members, based on predictions from the WMD 2 - Web MicroRNA Designer (http://wmd2.weigelworld.org) (the lines are further denoted as 6xamiBPM). Second, the MATH domain from BPM1 was cloned under the control of a 35S promoter and behind a GFP reporter, and with (further denoted as BPM1MATH:NLS) or without (BPM1MATH) a nuclear localization signal attached to the end of the domain to affect subcellular localization. The MATH construct was generated to impose a competition in the plant where endogenous BPM proteins have reduced access to WRI1 and, thus, hypothetically cause its stabilization.
Several independent plant lines were successfully generated, and two independent lines of the T4 generation were chosen for each construct for further analysis. In both 6xamiBPM lines, a significant downregulation in gene expression of all BPMs was measurable (Figures 3A and 3B), while the BPM1MATH and BPM1MATH:NLS lines showed strong expression of the transgene (Figure 3C). In addition, based on the GFP reporter, BPM1MATH constructs were detectable throughout the cell, including the nucleus (see Supplemental Figure 4A online), while BPM1MATH:NLS was exclusively present in the nucleus (see Supplemental Figure 4B online). Analysis of T4 generation plants showed that the 6xamiBPM lines consistently had higher WRI1 protein levels comparable to cul3hyp mutants, while, surprisingly, all MATH overexpression lines had significantly less WRI1 protein (Figure 3D). Although the absolute degree of WRI1 reduction in MATH overexpression lines varied among tested plants, we never observed any levels that equaled or exceeded those in the wild type. Also of note is that this is not based on reduced WRI1 expression, since the gene is actually upregulated in BPM1MATH and BPM1MATH:NLS lines (see Supplemental Figure 4C online). Currently, the reasons for this change in WRI1 protein content in the MATH-overexpressing lines remain elusive.
Generation of 6×amiBPM and MATH-Overexpressing Lines and Their Effect on WRI1 Protein Levels.
(A) and (B) qRT-PCR analysis show significantly reduced expression levels for six BPMs is in two representative 6×amiBPM lines compared with the wild type (WT).
(C) Expression levels of BPM1MATH show significant increases in two BPM1MATH lines and two BPM1MATH:NLS lines compared with the wild type.
(D) WRI1 protein content is strongly reduced in MATH-overexpressing lines when compared with the wild type, while both 6xamiBPM lines display increased WRI1 levels. All asterisks in this and subsequent figures indicate the statistical significant difference of at least P < 0.05 (one-way analysis of variance) to the wild type.
The different plant lines were broadly affected in development. Primary root growth was significantly delayed in all six lines and most strongly in the two 6xamiBPM lines (Figure 4A). While lateral roots emerged at a lower frequency in the BPM1MATH lines, no significant changes were detectable in plants expressing the BPM1MATH:NLS construct (Figure 4B). The 6xamiBPM plants developed very low numbers of lateral roots (Figures 4B and 4C), and all transgenic lines were affected in shoot development. In addition, all were late flowering, most strongly pronounced in BPM1MATH and 6xamiBPM lines (Figures 4D and 4E), with fewer leaves present at the beginning of flowering (Figure 4F) and a reduced rosette size (Figures 4D and 4G). Besides being smaller and present in fewer numbers, the leaves of transgenic plants also had a tendency to develop wider blades than the wild type (see Supplemental Figure 5 online).
Phenotype Analysis of the Wild Type, Two 6×amiBPM Lines, and MATH-Overexpressing Lines.
(A) Primary root lengths of 2-week-old Arabidopsis seedlings are strongly reduced in all transgenic lines when compared with the wild type (WT; n = 45).
(B) Specifically, 6xamiBPM plants have reduced numbers of lateral roots development (2-week-old seedlings; n = 45).
(C) Root phenotype of wild-type and two 6×amiBPM lines. Picture was taken 14 d after germination.
(D) Overview of rosette phenotypes from wild-type and transgenic lines. Picture was taken 33 d after germination.
(E) All transgenic lines are late flowering. Data were taken for 33-d-old plants (n = 30).
(F) Rosette leaf number at time of flowering (n = 10).
(G) Rosette area of 25-d-old Arabidopsis plants for each genetic background (n = 30).
To investigate to what extent WRI1 protein stability is affected in the different lines, stability assays were performed on selected plants (Figure 5; see Supplemental Figure 6 online). The assays consistently showed that in either the 6xamiBPM or MATH-overexpressing backgrounds, WRI1 was highly stable in comparison to the wild type (Figure 5A; see Supplemental Figures 6A and 6B online).
Studies of WRI1’s Protein Level, WRI1’s Transcriptional Activity, and Complex Assembly at the DNA Level.
(A) WRI1 protein is stabilized in MATH overexpressing and 6×amiBPM lines. WT, the wild type.
(B) CUL3 could be precipitated with α-WRI1 from wild-type plant extracts but not from BPM1MATH (left half) or 6×amiBPM (right half) extracts. Input was tested for presence of WRI1 and CUL3 proteins using the respective antibodies.
(C) Protein levels of WRI1 and expression of two WRI1 target genes, GLB1 and BCCP1, in the wild type and the different transgenic backgrounds.
(D) to (G) ChIP–quantitative PCR analysis on the wild type, two 6xamiBPM lines, and wri1-3 shows enrichment of the two WRI1 target promoters proBCCP1 and proGLB1 in the wild type (D) but not in a wri1-3 mutant (G) after IP with either α-WRI1 or α-CUL3 antibodies. While no enrichment was detectable in samples derived from α-CUL3 ChIPs in two 6xamiBPM lines ([E] and [F]), significant enrichment was detectable in α-WRI1 ChIP samples when compared with the wild type ([E] and [F]). ChIP–quantitative PCR experiments were repeated at least three times independently. Error bar indicates the se. Asterisk indicates a statistical significant difference (one-way analysis of variance, P < 0.05) compared with individual control.
IP experiments were performed on two MATH-overexpressing and two 6xamiBPM lines. As shown in Figure 5B, CUL3 protein was precipitated from wild-type plant extracts, while no precipitated CUL3 was detectable in either the MATH-overexpressing or 6xamiBPM lines. These findings together with stability assays demonstrate that the BPM proteins are required (1) for assembly of WRI1 into a complex with CUL3 and (2) for mediation of the transcription factor’s degradation.
WRI1 Activity Is Affected by CRL3BPM
We showed stabilization of WRI1 and an effect on its protein content in the plant in three different genetic backgrounds. In both the cul3hyp double mutant and 6xamiBPM lines, WRI1 levels are higher than the wild type, while in the MATH lines, WRI1 amounts are lower. In 6xamiBPM lines, BPM expression is reduced, and in the cul3hyp and MATH backgrounds, BPM protein levels are likely unchanged. However, based on each genetic background, the assembly of WRI1 into a CUL3-based complex is differently affected by either reduced CUL3 availability and/or functionality (cul3hyp), reduced BPM content (6xamiBPM), or reduced accessibility of BPMs to WRI1 (MATH-overexpressing lines). To investigate whether these situations affect WRI1 transcriptional activities differently, expression of two confirmed WRI1 targets, BCCP1 and GLB1, was tested (Baud et al., 2009; Maeo et al., 2009). BCCP1 (At5g16390), which encodes for a biotin carboxyl carrier protein, and GLB1 (At4g01900), which encodes for a PII protein, are both critical players in fatty acid biosynthesis, but also participate in carbon and nitrogen metabolism (Tissot et al., 1998; Chen et al., 2006).
Quantitative RT-PCR (qRT-PCR) analysis revealed a loss of BCCP1 and GLB1 expression in the wri1-3 null mutant compared with the wild type (Figure 5C). Expression of both genes in MATH-overexpressing lines was similarly reduced. By contrast, both genes were strongly upregulated in 6xamiBPM lines correlating with changes in WRI1 protein content. Interestingly, no change in BCCP1 or GLB1 expression in comparison to the wild type was noticeable in the cul3hyp line, despite the fact that WRI1 protein levels were elevated comparable to 6xamiBPM lines. These findings indicate, based on the presumed presence (cul3hyp) or absence (6xamiBPM), that BPM proteins also negatively affect transcriptional activity of their target proteins.
CUL3 Assembles with WRI1 at the DNA Level
To investigate whether CUL3 forms a complex with WRI1 at the DNA level, we performed chromatin immunoprecipitation (ChIP) experiments (Morohashi et al. 2009). In wild-type plants, α-WRI1– and α-CUL3–based ChIP experiments resulted in a two- to threefold enrichment of WRI1 binding sites (proBCCP1 and proAtGLB1), respectively (Figure 5D), while no enrichment was detectable in the wri1-3 null mutant, which served as a negative control (Figure 5G). Interestingly, α-WRI1 ChIP in the two 6xamiBPM lines yielded higher levels of proGLB1 and proBCCP1 sites, which was in agreement with higher WRI1 protein levels in these plants (Figures 5E and 5F), as well as increased transcription of the corresponding genes (Figure 5C). Finally, ChIP using the α-CUL3 antibody in 6xamiBPM lines did not lead to any enrichment of proGLB1 and proBCCP1 sites (Figures 5E and 5F), which corroborates the finding that loss of BPMs disrupt the ability of CUL3 to assemble into a complex with WRI1 (Figure 5B). Overall, these results provide strong evidence that CUL3 proteins form a complex with WRI1 at the DNA level and that this assembly requires BPM proteins.
Reduced BPM Content Affects Fatty Acid Metabolism in Seeds
The finding that reduced BPM expression in 6xamiBPM lines increases both WRI1 protein content and expression of WRI1 target genes was intriguing as it opened up the possibility that seeds of 6xamiBPM lines may also contain elevated levels of fatty acids due to augmented levels of active WRI1 (Baud et al., 2009). In agreement with this idea, both 6xamiBPM lines showed significant increases in seed weights and size when compared with wild-type seeds (Figures 6A and 6B). However, changes were much more pronounced in 6xamiBPM #1 than in 6xamiBPM #2 plants, which may be due to different activities of the 35S promoter in seeds of the two lines. The lines also showed increased WRI1 content in seeds and elevated expression of the two target genes BCCP1 and GLB1 (Figure 6C). Similar to their increases in weight, both lines also showed altered total fatty acid contents (Figure 6D); while changes in 6xamiBPM #2 plants were only very mild and furthermore nonsignificant (∼96 μg/30 seeds in average versus ∼93 μg/30 seeds in the wild type), the total fatty acid content in 6xamiBPM #1 seeds was increased by around 50% (∼140 μg/30 seeds) when compared with the wild type. The wri1-3 line was used in these experiments as a control and showed a significant reduction in both seed weight and total fatty acid contents (∼79 μg/30 seeds) when compared with the wild type and the two 6xamiBPM lines. While changes were observable for total fatty acid content, measurements of individual fatty acids did not detect any significant changes (see Supplemental Figure 7 online). In addition, no significant changes were observed in a general metabolic profile (amino and organic acids as well as soluble sugar) for wri1-3 or the two amiBPM lines when compared with the wild type (see Supplemental Figure 7 online), indicating that the changes in seed size and weight for the mutants are primarily based on aberrant fatty acid contents. Overall, these data further underscore that BPM proteins are critical regulators of WRI1 activity and that their loss positively affects both WRI1 stability as well as its actions.
Seed Weight, Size, and Fatty Acid Content in Wild-Type, wri1-3, and 6xamiBPM Plants.
(A) and (B) In comparison to the wild type (WT), seed weight and size is significantly reduced in wri1-3, while it is increased in both 6xamiBPM lines and to a greater extend in 6xamiBPM#1 then it is in #2. Bar in (A) = 1 mm. Data in (B) represent average of n = 5 measurements of 20 seeds.
(C) Seeds in 6xamiBPM lines contain higher WRI1 levels, which correlate with expression of the WRI1 target genes BCCP1 and GLB1.
(D) Fold change in fatty acid contents of wri1-3 and the two 6xamiBPM lines in comparison to the wild type. Data represent average of n = 5 measurements of 30 seeds. The asterisk shows a statistically significant difference compared with the wild type (P < 0.05, t test).
Because seeds of the two 6xamiBPM lines varied significantly in their weight and fatty acid content, we included a third 6xamiBPM line in our analysis to ensure that loss of BPMs leads to increases in fatty acid content and seed size in a reproducible manner. As observed for the other two lines, 6xamiBPM #3 seeds are also increased in size (see Supplemental Figure 8B online) as well as in dry weight (see Supplemental Figure 8C online), and this correlated with elevated WRI1 content (see Supplemental Figure 8A online), as well as upregulated expression of BCCP1 and GLB1 (see Supplemental Figure 8D online). Likewise, we also observed a significant increase in fatty acid content in these seeds in comparison to the wild type (see Supplemental Figure 8E online), substantiating the finding for 6xamiBPM #1 plants that reduced BPM activity likely results in higher fatty acid levels in seeds.
DISCUSSION
This work shows that BPM proteins have the ability to interact with a broad range of ERF/AP2 proteins. Although in planta assembly has only been confirmed for WRI1, Y2H studies still strongly indicate that many ERF/AP2 proteins are targeted in Arabidopsis by a CRL3BPM complex. IP and pull-down studies in this work underscore that WRI1 assembles in vitro and in the plant into a complex with CUL3, and the missing CUL3-WRI1 assembly in 6xamiBPM and MATH-overexpressing backgrounds emphasizes that BPM proteins are required for this step. The studies further show that the interaction of BPMs with WRI1 results in the destabilization of their substrate. This is supported by the finding that WRI1 is stabilized in a cul3hyp background, as well as in MATH overexpression and 6xamiBPM plants. Moreover, the ChIP data strongly support our assumption that WRI1-CRL3BPM assembly occurs at the DNA level. Consequently, BPM proteins can be considered as negative regulators of WRI1 activities by mediating assembly with the CRL3 core and ultimately causing its degradation via the 26S proteasome. This is also supported by the finding that fatty acid levels are significantly increased in seeds of 6xamiBPM #1 and #3 plants. These changes resemble earlier descriptions for plants overexpressing WRI1 (Cernac and Benning, 2004). However, it is significant to note that similar changes can be accomplished in Arabidopsis without ectopically expressing a transgenic WRI1 in seeds. The fact that overall metabolic changes were mostly restricted to total fatty acid contents also indicates that the function of BPM proteins in seeds is strongly connected with WRI1 activity. In this context, it is of note that WRI1 has very recently been described as part of a small gene family with a total of four members in Arabidopsis (To et al., 2012). Although WRI1 is the primary member that regulates fatty acid biosynthesis in seeds, the other members also contribute to this pathway, but they do so in other tissues (To et al., 2012). Consequently, it will be interesting to see whether these other WRI1-related proteins also underlie the control of CRL3BPM E3 ligase activities. Our findings also support the earlier suggestion that instability of RAP2.4, another BPM interacting protein, is mediated by a CRL3BPM ligase (Weber and Hellmann, 2009). Overall, it is likely that a general consequence of BPM interaction with ERF/AP2 transcription factors is degradation of the latter.
In this context, it is also important to note that the BPM family members have very recently been established as regulators of an ABA response by targeting the homeodomain Leucine zipper transcription factor HB6 for degradation (Lechner et al., 2011). Consequently, plants with reduced levels of BPM1, 4, 5, and 6 (amiR-bpm) display aberrant responses in stomatal opening (Lechner et al., 2011); however, germinating amiR-bpm seedlings only display increased ABA resistance when combined with an HB6 overexpression background. The finding that a member of another transcription factor family is a substrate of BPM proteins further increases the number of potential substrate proteins that are targeted by CRL3BPM for degradation. Furthermore, many members of the ERF/AP2 family have been described in context with stress tolerance, including the ones tested in this study. For example, DREB1a is a classical regulator of drought and cold tolerance responses in plants (Sakuma et al., 2002; Miura et al., 2007). ERF1 is known to play a role in biotic stress (Lorenzo et al., 2003; Zhang et al., 2011), and RAV1 has been described in context with senescence and different abiotic stress conditions (Sohn et al., 2006; Woo et al., 2010; Yun et al., 2010). It is therefore likely that both MATH overexpression and 6xamiBPM plants display different sensitivities toward stress, such as cold or drought, as well as treatments with phytohormones, such as ethylene or jasmonic acid, in addition to ABA.
It is also noteworthy that degradation of WRI1 appears to occur continuously rather than being stimulated by a specific signal, and this also holds true for RAP2.4 and HB6 (Weber and Hellmann, 2009; Lechner et al., 2011). It is unlikely that the cell is always degrading these proteins to the same amount since this seems to be quite an inefficient and uneconomical approach to regulating protein amounts. Rather, one would expect that specific signals are in place that slow down turnover of CRL3BPM substrates in a similar manner to the process that occurs in ethylene signal transduction, where ethylene disrupts proteasomal degradation of EIN3 mediated by the F-box proteins EBF1 and EBF2 (Guo and Ecker, 2003; Potuschak et al., 2003). In fact, ABA treatment appears to have a stabilizing effect on HB6, but the kind of signal that may have a similar effect on WRI1 or RAP2.4 remains unclear.
The wide-ranging developmental changes in both 6xamiBPM, as well as MATH overexpression lines, emphasizes that the BPM family is widely required for plant development. amiR-BPM plants showed reduced shoot growth (Lechner et al., 2011), which we also observed for 6xamiBPM lines. Interestingly, we could not detect any problems in fertility as observed by Lechner et al. (2011) for amiR-bpm plants. This opens up the question of whether BPM3 and 4, which are not affected in amiR-bpm lines, may have specific roles in flower development. In addition, 6xamiBPM plants had strongly reduced root development and fewer leaves, and it remains open whether these changes were also seen by Lechner et al. (2011). Moreover, changes in root development were not apparent in MATH overexpression lines, indicating that reduced BPM expression and binding competition approaches affected the developmental program of the roots in the corresponding plants differently.
The different approaches employed in this work to affect the CRL3BPM-WRI1 interplay revealed two additional interesting aspects about the function of BPM proteins besides being substrate receptors to a CRL3BPM ligase. First, comparing cul3hyp and 6xamiBPM plants clearly demonstrated that in both genetic backgrounds, WRI1 protein content is increased due to greater stability of the transcription factor. However, the transcriptional activity of WRI1 was only elevated in 6xamiBPM plants but not in cul3hyp, as indicated by the changed versus unchanged transcriptional levels of GLB1 and BCCP1. Given that in the cul3hyp mutant, BPM protein levels are likely normal, these findings indicate that BPM proteins negatively interfere with WRI1 activity, most likely by binding to the transcription factor, while more active WRI1 is available in 6xamiBPM plants. Second, the reduced WRI1 amount was quite surprising and unexpected. Because WRI1 expression was upregulated in MATH overexpressors, one may suggest that the reduced WRI1 content was sensed by the cell and that changes on the transcriptional level represent a feedback-loop response. In addition, these data also clearly indicate that the MATH domain also interfered with posttranscriptional processes and thus point out that BPM proteins may have even further diverse roles in addition to targeting ERF/AP2 or HB6 transcription factors for ubiquitylation and proteasomal degradation.
Finally, it is of interest that the ChIP data strongly indicate that the CRL3BPM E3 ligase assembles with WRI1 at the DNA level, while the transcription factor is bound to its target sites. At this point, we cannot judge whether the CRL3BPM complex only assembles with WRI1, and potentially other substrates, exclusively at the DNA level or whether this also occurs independently of DNA binding. In addition, it is also undetermined whether BPMs assemble independently of CUL3 with their substrates. These open questions need to be addressed in future work.
In summary, this work reveals a link between fatty acid metabolism and CUL3-based E3 ligase activities. The work also confirms that BPM proteins function in planta as substrate receptor proteins to a CRL3BPM ligase with the purpose of destabilizing bound substrates. These findings further indicate that a large number of ERF/AP2 proteins are potential targets of BPM proteins and that this complex plays a major role in plant development and stress tolerance by broadly regulating transcriptional, and potentially posttranscriptional, processes in the plant.
METHODS
Plant Materials and Growth Conditions
Arabidopsis thaliana wild-type ecotype Columbia plants and plants of the different genetic backgrounds were grown either on Arabidopsis thaliana (AT) medium without supplement of Suc (Estelle and Somerville, 1987) in a growth chamber at 22°C with 120 μmol/m2/s light intensity or in soil in a greenhouse at 20°C under long-day conditions (16 h light/8 h dark).
Clone Constructions
The cDNAs of BPM1MATH, BPM1MATH:NLS, CUL3s, ERF/AP2s, and BPMs were cloned into pDONR221 (Invitrogen). For Y2H studies, the corresponding cDNAs were shuffled into destination vectors pACT2 (prey) and pBTM116-D9 (bait) by Gateway technology (Invitrogen) as described (Weber et al., 2005). pDEST15 (Invitrogen) and pET-58-DEST (Merck) were used to express and purify GST- and His-tagged proteins in Escherichia coli, respectively; where necessary, elution of GST proteins from glutathione-agarose beads was done following standard procedures. pMDC43 was used for expression in plants and for subcellular localization studies as described (Curtis and Grossniklaus, 2003). The NLS sequence was adopted from Howard et al. (1992), extended, and attached to the MATH domain in a four-step-based PCR process. To generate artificial microRNAs, a protocol from the WMD2 microRNA designer Web page (http://wmd2.weigelworld.org; see Supplemental Figure 9 online) was followed using pRS300 as starting vector. For expression in plants, ami constructs were first cloned into pDONR221 before being shuffled into the binary vector pGBW14. For primers used, see Supplemental Table 1 online.
Subcellular Localization Studies
The fluorescent fusion proteins GFP:WRI1, GFP:BPM1MATH, and BPM1MATH:NLS were transiently expressed in Nicotiana benthamiana epidermal cells following the method described by Sparkes et al. (2006). GFP expression was detected and documented with a Zeiss LSM 510 Meta confocal microscope.
Interaction and Complex Assembly Studies
Y2H studies were followed as described by Weber et al. (2005). SDII selection medium supplemented with uracil and His was used as transformation control, while for interaction studies, SDIV minimal medium was chosen without uracil and His supplements. Photos were taken from single spots 7 d after plating. FPLC was performed with 2 mg protein injections and a flow rate of 50 µL/min using an AKTA FPLC system (GE Healthcare Science) and as described by Leuendorf et al. (2010). Pull-down analysis and IP studies were followed as described before (Hellmann et al., 2003; Bernhardt et al., 2006). Extraction and washing buffers contained at all times 1 mM PMSF (Sigma-Aldrich) and 10 μM MG132 (Sigma-Aldrich) to prevent proteolytic and proteasomal activities, respectively. For pull-down analysis with GST- and His-tagged proteins, GST-containing proteins were first eluted from glutathione-agarose beads before incubated with His:WRI1 that remained attached to tetradentated-chelated nickel resin. In general, proteins were incubated at least 1.5 h at 4°C under shaking conditions before being centrifuged. Precipitates were washed no less than three times to remove unspecific bindings, before they were taken up in Laemmli buffer (Laemmli, 1970) and boiled (10 min, 95°C). IPs were followed as described by Bernhardt et al. (2006). In brief, 1 mg of fresh protein extracts from 2-week-old seedlings were precleaned with 30 μL protein-A-agarose beads (Santa Cruz Biotechnology; 1.5 h, 4°C). The beads were centrifuged, and the supernatant was transferred into a fresh tube and incubated first with α-WRI1 (1.5 h, 4°C) before 30 μL protein-A-agarose beads were added (1.5 h, 4°C). After brief centrifugation, four washing steps followed, after which precipitates were taken up in Laemmli buffer and boiled as described above. For pull-down and IP studies, precipitates were further analyzed by SDS-PAGE and protein gel blotting using standard procedures. Where applicable, membranes were stained with Ponceau S to detect transferred proteins. For immunodetection, custom-made (α-CUL3 [rabbit] and α-WRI1 [rabbit]; GeneScript) or commercially available antibodies (GST, secondary α-rabbit IgG-horseradish peroxidase; Santa Cruz) in combination with an ECL Plus Western Blotting Detection Kit (GE Healthcare Life Science) were used.
Stability Assays
For stability assays, Arabidopsis seedlings were cultured on solid AT medium for 2 weeks before being transferred to 5 mL AT liquid medium. Plants were incubated for 3 h with the transcriptional inhibitor ActD2 (Sigma-Aldrich; final concentration 10 µg/mL) and/or the proteasomal inhibitor MG132 (Sigma-Aldrich; 20 µM/mL, 6 h) before CHX (Sigma-Aldrich; 100 µM/mL, 3 h) was added to inhibit translation. DMSO was used as mock control and as a dissolvent for all inhibitors. Protein gel blot analysis and protein detection were conducted using standard procedures. The antibodies against WRI1 and CUL3 were designed and produced by GeneScript and used in a 1:1000 dilution. Protein detection was followed as described in the ECL Plus Western Blotting Detection Kit manual (GE Healthcare Life Science).
RNA Isolation and Expression Analysis
Total Arabidopsis RNA was extracted following the protocol of the Isolate RNA kit from Bioline; reverse transcription was done according to the manual for the high-capacity cDNA reverse transcription kit (Applied Biosystems). qRT-PCR reactions (95°C, 7 min; 95°C, 15 s; 60°C, 1 min; 40 cycles) were performed using the SYBR green method on a 7500 Fast Real-Time PCR system (Applied Biosystems). Relative gene expression analyses were calculated by the full quantification method with ACTIN2 as the internal control gene. Fourteen 2-week-old seedlings were pooled for each replicate. At least three biological replicates were performed for each individual experiment. Primers used for qRT-PCR are shown in Supplemental Table 1 online.
ChIP Assays
For ChIP assays, an established protocol was followed (Morohashi et al., 2009). In brief, 60 mg (fresh weight) of 15-d-old seedlings were harvested for each ChIP experiment and cross-linked for 10 min under vacuum in cross-link buffer containing 1% formaldehyde as described by Morohashi et al. (2009). Cross-linked samples were incubated in 100 mM Gly for 5 min under a vacuum, thoroughly washed in double-distilled water, and snap frozen in liquid nitrogen. Frozen tissues were ground into fine powder and dissolved in nuclear isolation buffer (Morohashi et al., 2009) supplemented with 1× protease inhibitor cocktail (Sigma-Aldrich). After filtering through single-layered Miracloth (Merck), the samples were centrifuged (10 min, 1200g, 4°C). Pellets were resuspended in nuclear isolation buffer supplemented with 0.3% Triton X-100 and centrifuged again (10 min, 10,000g, 4°C). After resuspension in lysis buffer, the purified nuclei were then sonicated (three times, 20 s, 9 W; Fisher Scientific Model 100 dismembrator) to yield chromatin fragments of 300 to 500 bps. Sonicated chromatin fragments (2 mg) were first precleared with protein-A-agarose beads (Sigma-Aldrich) (1.5 h, 4°C) before being incubated with specific antibodies (1 mg/mL; 1.5 h, 4°C), followed by a fresh batch of protein-A-agarose beads (30 μL; 1.5 h, 4°C) to IP protein-DNA complexes. After IP, cross-linking was reversed by incubating samples overnight at 65°C in elution buffer (1% SDS, 0.1 M NaHCO3, and 0.25 mg/mL proteinase K; Morohashi et al., 2009), after which RNaseA (1 mg/mL) was added (30 min; room temperature). As input control for data normalization, a portion of sonicated, cross-linked, and precleared DNA was treated accordingly except for undergoing an IP. Samples were further cleaned up using a DNA purification kit (NuCleoSpin Extraction II; Macherey-Nagel) and quantified to use equal amounts of template (50 ng/reaction) for qRT-PCR analysis. To amplify promoter sequences that contain an AW-box recognized by WRI1 (Maeo et al., 2009), specific primers were designed (see Supplemental Table 1 online). EF1 was selected as a reference gene for internal control. qRT-PCR reactions (95°C, 10 min; 95°C, 15 s; 60°C, 1 min; 50 cycles) were done as described above and repeated with at least three independent biological replicates.
Metabolic Analysis
Metabolic profiling of 2-week-old seedlings grown on ATS plate (100 mg fresh weight) and seed samples (50 mg dry weight) of all backgrounds used in this study were analyzed according to earlier described protocols (Roessner-Tunali et al., 2003). Seed fatty acids were extracted exactly as described before (Focks and Benning, 1998). For quantification with gas chromatography, pentadecanoic acid was used as an internal standard (Browse et al., 1985).
Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: ACTIN2, At3g18780; IAA5, At1g15580; BPM1, At5g19000; BPM2, At3g06190; BPM3, At2g39760; BPM4, At3g03740; BPM5, At5g21010; BPM6, At3g43700; DREB1a, At4g25480; ERF1, At3g23240; ERF4, At3g15210; RAV1, At1g13260; WRI1, At3g54320; BCCP1, At5g16390; and GLB1, At2g16060.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure 1. Verification of α-CUL3 and α-WRI1 Antibodies.
Supplemental Figure 2. FPLC Analysis and Subcellular Localization of WRI1.
Supplemental Figure 3. CHX Treatment Does Not Reduce the Protein Level of WRI1, but It Induces Its Transcription Level.
Supplemental Figure 4. Subcellular Localization of GFP:BPM1MATH and GFP:BPM1MATH:NLS.
Supplemental Figure 5. Rosette Leaf Phenotype on the Time of Primary Inflorescence.
Supplemental Figure 6. Stability Assays of WRI1 in Wild-Type, 6×amiBPM, and MATH Overexpressing Lines.
Supplemental Figure 7. Fatty Acid Profile and Metabolic Profile in Seeds of Wild-Type, wri1-3, and Two 6xamiBPM Lines.
Supplemental Figure 8. Seed Phenotype Analyses for 6xamiBPM #3.
Supplemental Figure 9. Predicted Target Sites for Artificial MicroRNA (in Red) on the Different BPM Genes.
Supplemental Table 1. Primers Used for Different PCR-Based Approaches.
Acknowledgments
We thank Sutton Mooney for technical support and critical reading, the National Science Foundation for support to H.H. (MCB-1020673) and S.R. (MCB-1052492), Detlef Weigel for providing the vector pRS300, and Pascal Genschik for seeds of cul3hyp. We also thank Kengo Morohashi and Erich Grotewold for technical advice on ChIP.
AUTHOR CONTRIBUTIONS
H.W. originally found WRI1 and ERF4 as interactors of BPM proteins through Y2H and pull-down studies. J.H.L. generated the 6×amiBPM and MATH-overexpressing lines and generated confocal microscopy pictures. L.C. contributed to clone constructions, Y2H studies, complex assembly analysis, ChIP, stability assays, RNA isolation, expression analysis, and Arabidopsis phenotypical analyses. S.R. helped with the FPLC experiments. T.T., S.W., and A.R.F. provided the metabolic data. H.H. is the principal investigator for this National Science Foundation–supported project; he participated in complex assembly analysis and in writing the article.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Hanjo Hellmann (hellmann{at}wsu.edu).
↵[W] Online version contains Web-only data.
Glossary
- UBQ
- ubiquitin
- Y2H
- yeast two-hybrid
- IP
- immunoprecipitation
- FPLC
- fast protein liquid chromatography
- CHX
- cycloheximide
- ActD2
- actinomycin D2
- qRT-PCR
- quantitative RT-PCR
- ChIP
- chromatin immunoprecipitation
- ABA
- abscisic acid
- AT
- Arabidopsis thaliana medium
- Received November 9, 2012.
- Revised May 22, 2013.
- Accepted June 3, 2013.
- Published June 21, 2013.