An upstream regulator of the 26S proteasome modulates organ size in Arabidopsis thaliana

Authors

  • Hung M. Nguyen,

    1. Department of Molecular Biology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    2. Max Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany
    Current affiliation:
    1. Center of Research and Development, Duy Tan University, Danang City, Vietnam
    Search for more papers by this author
    • These authors contributed equally to this paper.
  • Jos H. M. Schippers,

    Corresponding author
    1. Max Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany
    • Department of Molecular Biology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
    • These authors contributed equally to this paper.
  • Oscar Gõni-Ramos,

    1. Plant Chemetics Laboratory, Max Planck Institute for Plant Breeding Research, Cologne, Germany
    Search for more papers by this author
  • Mathias P. Christoph,

    1. Department of Molecular Biology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Hakan Dortay,

    1. Department of Molecular Biology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Renier A. L. van der Hoorn,

    1. Plant Chemetics Laboratory, Max Planck Institute for Plant Breeding Research, Cologne, Germany
    Search for more papers by this author
  • Bernd Mueller-Roeber

    Corresponding author
    1. Max Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany
    • Department of Molecular Biology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author

For correspondence (e-mail schippers@mpimp-golm.mpg.de or bmr@uni-potsdam.de).

Summary

In both animal and plant kingdoms, body size is a fundamental but still poorly understood attribute of biological systems. Here we report that the Arabidopsis NAC transcription factor ‘Regulator of Proteasomal Gene Expression’ (RPX) controls leaf size by positively modulating proteasome activity. We further show that the cis-element recognized by RPX is evolutionarily conserved between higher plant species. Upon over-expression of RPX, plants exhibit reduced growth, which may be reversed by a low concentration of the pharmacological proteasome inhibitor MG132. These data suggest that the rate of protein turnover during growth is a critical parameter for determining final organ size.

Introduction

Growth of an organism and its size determination require tight regulation of cell proliferation and cell growth (Nieuwland et al., 2009; Breuninger and Lenhard, 2010; Brioudes et al., 2010; Johnson and Lenhard, 2011; Byrne, 2012). For leaf growth, this involves progression of component cells through a succession of developmental phases: proliferation (cells divide at a rate matching their expansion to maintain cell size), expansion (cells stop dividing but continue to expand, typically to a size much larger than that of meristematic cells), and maturity (cells no longer expand) (Beemster et al., 2005). The size-control system is largely dependent on both cell size and cell number (Tsukaya, 2008). However, in plants, final organ size is not simply the sum of cell number and cell size, as evidenced by an abnormal enlargement of leaf cells that appears to be triggered by an insufficient supply of cells upon inhibition of cyclin-dependent kinase activity (De Veylder et al., 2001; Autran et al., 2002; Brioudes et al., 2010; Kawade et al., 2010). This compensation effect may be triggered through a non-cell-autonomous pathway depending on intercellular signals, but also via a cell-autonomous pathway as indicated by the over-expression of a cyclin-dependent kinase inhibitor (Kawade et al., 2010).

Our understanding of leaf growth is largely based on functional characterization of transcription factors (TFs) that interact at the genetic level (Byrne, 2012). However, the gene regulatory networks controlled by each individual TF are poorly characterized. One of the best understood genetic modules acting during the growth of leaves is the patterning and formation of stomata (MacAlister et al., 2007). Recently, the gene regulatory network of one of the three central basic helix-loop-helix TFs, FAMA, was revealed (Ohashi-Ito and Bergmann, 2006; Hachez et al., 2011), corroborating its role during the final step of stomata development. Another example of a well-established gene regulatory network is that controlled by the NAC (NAM, ATAF1,2 and CUC2) domain TF VASCULAR-RELATED NAC-DOMAIN6 (VND6), which is involved in the differentiation of vascular tissue (Ohashi-Ito et al., 2010). In eukaryotic organisms, transcription of each gene is directed by cis-regulatory sequences that are typically positioned in proximity to the transcription start site. The assembly of cis-regulatory sequences associated with each gene specifies the time and place in the organism at which the gene is to be transcribed (Tuch et al., 2008). In general, transcriptional circuits for specific genetic programs tend to be weakly conserved in distantly related organisms, as reflected by the fact that different types of TFs found in such organisms control similar genetic programs (Tuch et al., 2008).

Here we identified the plant-specific NAC TF Regulator of Proteasomal Gene Expression (RPX) as a negative regulator of leaf organ size. We found that RPX binds to the promoter of genes encoding 26S proteasomal proteins in vivo, resulting in increased gene expression and proteasome activity. Furthermore, our data indicate that RPX controls an evolutionarily conserved gene regulatory network that operates in higher plants in a similar manner to yeast. Thus, we have identified a transcriptional regulator of proteasome genes in plants that acts during leaf growth in a non-cell-autonomous way.

Results

RPX is a negative regulator of leaf size in Arabidopsis

To identify novel regulators of leaf development, we screened T–DNA knockout lines for TFs that are differentially expressed during leaf growth as determined by a previous microarray study (Beemster et al., 2005). A T–DNA insertion (rpx–1 mutant) within the last exon of a gene (At5g04410; RPX) encoding a NAC TF (Figure S1) resulted in an increased leaf area as determined by measuring the first true leaf pair of 22-day-old plants (Figure 1a,b). This phenotype was confirmed by an independent knockout line (rpx–2 mutant) with an insertion in the second exon (Figure S1a). For both knockout lines, we observed a significant increase in mesophyll cell number but cell size remained unaffected (Figures 1c,d and S1). RPX expression is induced during transition from the expansion phase to the maturation phase (Figure S1b). The expression profile and the phenotype observed here suggest that RPX is a negative regulator of leaf size.

Figure 1.

Effect of RPX on organ size. (a) Cleared first true leaf of the wild-type, rpx–1 mutant and the over-expression lines (ox and ox∆C). (b) Determination of leaf area. Values are means ± SE (n = 50). (c) Mesophyll cell size. Values are means ± SE for each data point (n = 20), with 50–100 measurements per leaf. (d) Cell number as determined from total leaf area and cell size. Values are means ± SE. Asterisks indicate a statistically significant difference compared with wild type (wt) (P ≤ 0.05; Student's t test).

To determine the effect of increased expression of RPX on leaf growth, we generated Arabidopsis plants that express the NAC TF under the control of the constitutive CaMV 35S promoter. Over-expression of RPX resulted in a decreased cell number and cell size, reducing total leaf area (Figure 1a–d). RPX is one of 13 members of the NTL (NAC with transmembrane motif 1-like) sub-group; NTL family members contain a hydrophobic C–terminal domain that may allow membrane localization (Kim et al., 2006, 2007). To determine whether the presence of the C–terminus influences the regulation of RPX, we generated transgenic lines that constitutively over-express a truncated RPX protein (oxΔC) (Figure S2). Truncation of RPX resulted in dramatic reduction of leaf area, cell size and cell number (Figure 1a–d), and affected the overall development of transgenic plants as indicated by early flowering, induction of axillary branch formation and twisted leaves (Figure S2).

Although several studies have indicated that the hydrophobic C–terminus of NTL family members directs membrane localization (Seo et al., 2008; Kim et al., 2010a), other reports suggest a different mode of action. Both NTL4, the closest homolog of RPX, and NTL9 have been shown to be located in the nucleus of Arabidopsis protoplasts, both as full-length and C–terminally truncated proteins (Kim et al., 2007, 2010b; Inzé et al., 2012). Moreover, the hydrophobic C–terminus of NTL9 interacts with calmodulin, indicating that it does not only act as a membrane domain (Kim et al., 2007). Previously, it was reported that RPX (ANAC078; AT5G04410) is a membrane-localized protein in transiently transfected onion epidermal cells (Morishita et al., 2009); however, we could not confirm this observation using Arabidopsis mesophyll cell protoplasts. Both full-length and truncated RPX fused to GFP showed prominent nuclear localization, and weak cytoplasmic localization was also observed in some cases (Figure S3).

Identification of a promoter region responsible for vascular RPX expression

We found that RPX expression is mainly restricted to the vascular tissue throughout the plant (Figure S3), in line with previous reports (Morishita et al., 2009). Transformation of tobacco with an Arabidopsis RPX promoter::GUS construct resulted in a similar expression pattern in leaves as observed for Arabidopsis (Figure S3g), indicating conserved upstream regulation of RPX. To determine the cis-elements that direct vascular expression of RPX, we performed a promoter deletion study. Shortening the RPX promoter to 385 bp upstream of the start codon (which included 163 bp of the promoter and the 222 bp 5′ UTR) did not affect vascular expression (Figure 2a). Subsequently, we used the web tool AthaMap (Bülow et al., 2009) to analyze the remaining sequence for any known cis-elements (Figure 2b). At a 90% confidence cut-off level, we identified a region from −141 to −127 bp (GGGTGTTGACGTGTC) upstream of the transcription start site that was annotated as a MYC_MYB binding site, covering multiple elements including potential bZIP and MYB recognition sites. Four novel deletion constructs were generated to test the importance of the identified region for RPX expression. Partial or complete deletion of the 5′ UTR (Figure 2a) did not affect vascular expression. However, removing the MYC_MYB region from the 163 bp promoter by shortening it to 119 bp completely abolished GUS staining. Our findings are in line with a previous genome-wide binding site analysis of the ELONGATED HYPOCOTYL 5 (HY5) TF that identified RPX as a potential target gene (Zhang et al., 2011). Interestingly, the identified HY5 binding sequence (GACGTG) (Li et al., 2010) is present within the 42 bp sequence required for RPX expression. Thus, RPX expression may at least in part depend on HY5, although other upstream TFs probably also contribute to regulate its expression.

Figure 2.

Identification of promoter sequence required for vascular expression of RPX. (a) Promoter deletion analysis identified a 44 bp region required for vascular expression. Removal of the 5′ UTR had no effect on the RPX expression pattern. (b) As indicated by AthaMap analysis, the 44 bp region contains a high-confidence region with MYC_MYB binding sites; this region also contains a perfect HY5 binding site (indicated in bold).

RPX expression levels affect root growth

So far, no direct correlation between shoot size and root size has been reported; increased shoot biomass may occur concomitantly with either increased or decreased root biomass (Mizukami and Fischer, 2000; Ribeiro et al., 2012). The length of the primary root of rpx–1 and 35S::RPX plants was significantly reduced compared to wild type (Figure 3a,b). Moreover, oxΔC plants displayed a severe reduction in root length (Figure 3b), caused mainly by halted growth of the main root and early induction of lateral roots. To assess the affect on cell division within the root, the meristem cortex cell number was determined (Ioio et al., 2007). Both knockout and full-length over-expression plants showed a significantly reduced meristem cell number compared to wild type (Figure 3c,d), with oxΔC plants showing the strongest reduction. Reduction in root length and meristem cell number is commonly observed after treatment with cytokinin (Ioio et al., 2007), suggesting a potential link between RPX function and this phytohormone.

Figure 3.

RPX is required for proper root growth. (a) Control, rpx–1 mutant and RPX over-expression plants were grown for 7 days on vertical MS agar plates. (b) Primary root lengths of 20 seedlings each of wild-type (wt) and RPX-modified plants. Values are means ± SE. (c) Cortex meristem cell number of roots from 7-day-old wild-type (wt) and RPX-modified seedlings. Values are means ± SE (n = 36 for each data point). (d) Representative images of roots used for determination of cortex meristem cell number (I, wild type; II, rpx–1; III, ox; IV and V, ox∆c). Arrows indicate the end of the meristem cortex files. In (b) and (c), asterisks indicate a statistically significant difference compared with wild type (wt) (P ≤ 0.05; Student's t test).

RPX controls the expression of 26S proteasome genes

To identify direct target genes of the RPX TF, we performed microarray experiments using a β–estradiol-inducible RPXΔC construct (Figure S4a) transformed into Arabidopsis. In this approach, treatment of the transgenic plants by β–estradiol triggers translocation of the constitutively expressed chimeric TF XVE into the nucleus (Zuo et al., 2000), inducing transcription of RPX∆C. Seedlings of RPX∆C inducible over-expression lines were either treated with β–estradiol or ethanol (mock). The optimal duration of induction was determined by quantitative real-time PCR (Figure S4b). Based on this optimization, seedlings were treated with β–estradiol for 120 min, and mRNA of three independent biological replicates was hybridized to Affymetrix ATH1 (22K) arrays (http://www.affymetrix.com/). As a result, 80 up-regulated genes were found after statistical analysis using Robin (Lohse et al., 2010) (Figure S4c and Table S1). GO classification and MapMan analysis (Usadel et al., 2005) revealed a clear over-representation of genes involved in protein degradation, including genes encoding subunits of the 26S proteasome (Figure S5). In addition, RPX induced the expression of six other TFs, including three uncharacterized NAC TFs (ANAC087, ANAC094 and ANAC103). The 26S proteasome plays a central role in regulating cellular processes and hormone responses (Fu et al., 2010). In plants and other eukaryotes, the ATP-dependent protease complex contains a central core particle that encloses the protease active site (Voges et al., 1999). The opening to the protease core is shielded by a regulatory particle that consists of a ring of six regulatory particle triple A (AAA+) ATPases and 11 non-ATPases. Arabidopsis contains 55 genes encoding subunits of the 26S proteasome and several putative UMP1-like maturation factors (Table S2). Expression of these genes was examined after 60, 90 and 120 min of induction of RPX, as well as in the over-expression and knockout lines, by quantitative real-time PCR (Figure 4 and Table S3). We found that 42 of the 55 genes tested were more than twofold induced after 120 min of β–estradiol induction. In full-length over-expression plants, 24 genes were more highly expressed than in controls, while 46 genes were up-regulated in oxΔC plants, although the level of transgene expression for full-length and truncated RPX was comparable. As RPX expression is restricted to the vascular tissue, we used petioles of the first leaf pair for expression profiling of the knockout line. This revealed a significant down-regulation of more than twofold for seven of the 55 tested genes. Interestingly, the induction of RPX by some abiotic stresses correlates with the expression of genes encoding the 26S proteasome, as shown by global transcript profiling (Kilian et al., 2007). For example, RPX is induced within 60 min of UV–B exposure and returns to basal levels after 12 h (Figure S6). The 26S proteasome-encoding genes are induced 2–4 h after RPX induction, suggesting that they are direct target genes of RPX.

Figure 4.

RPX controls the expression of genes encoding 26S proteasome subunits. (a) Heatmap profile of the expression pattern of 26S proteasome-encoding genes after 60, 90 or 120 min of induction of RPX measured by quantitative real-time PCR. The upper and lower halves represent the same subunits, but different isoforms. Values indicated by the scale-bar (log2 fold change) are presented as ΔΔCT and are based on three biological replicates. Subunits with one isoform only have a symbol in the upper half and a grey field in the lower. (b) Heatmap representation for rpx–2 and over-expression lines. AGI codes, ΔΔCT values and P values are given in Table S2. Green and red coloration indicate decreased and increased expression, respectively, compared to control, i.e. mock treatment in (a) and wild type in (b).

Identification of an evolutionarily conserved RPX binding site

To identify the DNA-binding site of RPX, we performed a motif-based sequence analysis of the upstream regions of the 55 proteasome genes. This analysis identified a novel cis-element, (T/A)(A/T/G)(A/T/C)TGGGC(C/G)(T/G/A)N, which we named proteasome-related cis-element (PRCE) (Figure 5a). We found that the PRCE motif is mainly located within the first 200 bp upstream of the transcription start site. To test whether the PRCE element is also present in other species, we analyzed 1000 bp promoter regions of proteasome subunit-encoding genes. Both in the monocot Oryza sativa (rice) and the dicot Ricinus communis, we found enrichment of the PRCE motif at a similar position as for Arabidopsis (Figures 5b and S7). In contrast, we did not observe enrichment of the PRCE motif in the green alga Chlamydomonas reinhardtii. In accordance with this, the Chlamydomonas genome does not encode NAC domain TFs (Riaño-Pachón et al., 2008). Of the 80 genes identified in the microarray, 24 genes have a perfect PRCE motif in their promoter, while the remaining genes harbor an imperfect element, with a single mismatch within the PRCE element (Table S1). To determine whether RPX binds the PRCE element, we devised an in vitro assay named infrared-mediated mapping (IMAP). In short, glutathione-S–transferase (GST)-fused RPX was immobilized on glutathione disulfide beads and incubated with infrared-labeled DNA probes based on the PRCE-containing sequence of the RPN10 promoter (RPN10 is a proteasome subunit). RPX bound to probe 1 (P1) encompassing the conserved PRCE element, but not to two control probes (P2 and P3) derived from other parts of the RPN10 promoter that lack a PRCE motif (Figures 5c and S8). In addition, RPX fusion proteins generated in vivo and in vitro were used in an electrophoretic mobility shift assay (EMSA) and shown to bind the PRCE element derived from the RPN8a and RPN10 promoters (Figures 5d and S8). Thus, the identified conserved PRCE sequence is bound by RPX in vitro.

Figure 5.

Identification of the conserved proteasome-related cis-element (PRCE). (a) The PRCE element identified by MEME motif analysis is a TGGGC core-containing cis-element that is present in 39 of 55 promoters of genes encoding 26S proteasome subunits. (b) The PRCE element is located within the first 200 bp of the transcription start site (TSS) and is present in Arabidopsis, Oryza sativa and Ricinus communis. (c) IMAP oligo binding assays with probes based on the RPN10 promoter containing the PRCE element (P1) or not (P2 and P3) C, unlabelled competitor probe. Values are means ± SE (n = 3). (d) EMSA assay with in vitro-synthesized RPX protein. p8a, probe PRCE element RPN8a; p10, probe PRCE element RPN10; RPX, RPX protein; com, 1:50 competition with unlabeled probe.

RPX interacts with the PRCE element in planta

To verify whether RPX regulates the expression of 26S proteasome-encoding genes in vivo, we assayed its transactivation capacity towards four target genes (PBE1, encoding a catalytic subunit of the 20S core; RPT6a, encoding a regulatory particle triple AATPase; RPN10 and RPN12a, encoding two regulatory particle non-ATPase subunits). Mesophyll cell protoplasts were co-transformed with a vector containing the target promoter upstream of the firefly luciferase (fLUC) coding sequence (reporter), a control vector expressing Renilla luciferase (Licausi et al., 2011), and an effector plasmid expressing RPX from the CaMV 35S promoter. A significant increase in relative fLUC activity was observed for all four promoters tested, albeit with different levels of induction (Figure 6a). Deleting the 11 bp PRCE motif from the RPN10 promoter abolished RPX-mediated transactivation of the fLUC reporter (Figure 6a), indicating that transactivation depends on this promoter motif.

Figure 6.

The PRCE motif is bound in vivo and is required for RPX to regulate expression. (a) Protoplast transactivation assays. Values are means ± SE from four independent transformations. Asterisks indicate a statistically significant difference to control assays performed in the absence of 35S::RPX effector plasmid. (P ≤ 0.05; Student's t test).(b) ChIP/quantitative PCR enrichment analysis on proteasome genes. Quantitative PCR primers were designed to span the motif, the central exon/intron or the 3′ UTR for each potential target gene. Enrichment is presented as the fold change between the no-antibody control and the anti-GFP antibody sample. Values are means ± SE of three independent biological replicates. Wild type was used as an additional control for the specificity of the antibody reaction.

Direct binding of RPX to the promoters of the 26S proteasome genes in vivo was further assessed by chromatin immunoprecipitation (ChIP)/quantitative PCR assay using a functional RPX–GFP fusion protein expressed in transgenic Arabidopsis. Six genes harboring the PRCE motif within their promoters were selected. Four genes encode 26S proteasome subunits and the other two genes encode proteasome-associated proteins (PA200 protein and UMP1-like chaperone). For the quantitative PCR analysis, we designed primers that span the motif, the central part of the coding sequence and the 3′ UTR (Figure S9). For all genes, we obtained an enrichment with primers spanning the PRCE motif that was not seen for the downstream sequences (Figure 6b). We also found that RPX binds to the 3′ UTR of RPN8a, which contains an imperfect PRCE (Table S1). These observations indicate that genes encoding 26S proteasome subunits are transcriptionally regulated by RPX through direct interaction with the evolutionarily conserved PRCE element.

The expression level of RPX correlates with proteasome activity

We next investigated whether the expression level of proteasome genes is translated into a change in proteasome activity. After 8 h of RPX induction, the protein level of PBE1, PBA1 and RPN10 was increased as determined by Western blotting (Figures 7a and S10). To assess 20S proteasome activity, we used MV151, a probe that irreversibly labels the active site of proteasome subunits in an activity-dependent manner (Gu et al., 2010). MV151-labeled proteins are detected by fluorescence scanning of protein gels, revealing the activities of the three proteolytic subunits [PBB1 (β2), PBE1 (β5) and PBA1 (β1)] within the core protease of the 26S proteasome. This assay revealed significantly increased proteasome activity after 6 h of RPX induction (Figure 7b,c). Similarly, increased proteasome activity was found after 8 h of RPX induction using 7-amino-4-methylcoumarin labeled LLVY peptides (Figure S10b). Immunoblotting of protein extracts from over-expression and knockout lines revealed an opposite effect on protein abundance for PBA1 (Figure 7d) and PBE1 (Figure S10). Constitutive over-expression of both the full-length and truncated RPX protein resulted in higher overall proteasome activity (Figures 7e,f and S10), whereas the knockout lines showed a significantly decreased activity. Altogether, our data show that the expression level of RPX correlates with proteasome activity.

Figure 7.

RPX controls proteasome activity. (a) Immunological detection of PBA1, PBE1 and RPN10 after 8 h of induction of RPX, with (+) or without (−) β–estradiol. Rubisco and histone H3 from the Western blot with RPN10 are shown as loading controls. (b) Proteasome activity profiling after estradiol treatment by labeling with fluorescent MV151 probe. (c) Intensity of the overall MV151 signals normalized to the level of the large subunit of Rubisco. Values are means ± SE (n = 4). Asterisks indicate a statistically significant difference compared with mock treated inducible RPX lines (P ≤ 0.05; Student's t test). (d) Immunological detection of PBA1 in leaves of RPX transgenic plants and control. (e) Proteasome activity levels in leaf extracts of RPX transgenic plants and control were determined by labeling with MV151. (f) The intensity of the MV151 signals was quantified and normalized to the level of the large subunit of Rubisco. Values are means ± SE (n = 4). Asterisks indicate a statistically significant difference compared with wild type (P ≤ 0.05; Student's t test).

Mild inhibition of proteasome activity reverses the RPX over-expression phenotype

To determine whether the decreased rosette biomass of RPXΔC over-expression plants is caused by increased proteasome activity, we treated plants with non-lethal concentrations of the proteasome inhibitor MG132 (Lee and Goldberg, 1998). Seedlings were grown for 4 days on Murashige & Skoog (MS) medium and subsequently transferred to medium containing either 5 or 10 μm MG132. Measuring rosette biomass at day 13 revealed reversion of the growth phenotype of the RPXΔC over-expression plants to wild type on both MG132 concentrations (Figure 8). Although a small increase in the biomass of wild-type plants was observed, this was not significant. Root length was not rescued; elongation of the main root of 35S::RPXΔC plants stopped within several days after germination. Taken together, the decreased leaf area of RPX over-expression plants may be attributed to the increased expression of 26S proteasome genes rather than other genes identified in our microarray experiment.

Figure 8.

Treatment with proteasome inhibitor MG132 restores leaf growth in 35S::RPXΔC plants. (a) Comparison of rosette growth upon MG132 treatment. (b) Determination of biomass accumulation after growth in the presence of MG132. Values are means ± SE (n = 3). Asterisks indicate a statistically significant difference compared with untreated controls (P ≤ 0.05; Student's t test). FW, fresh weight.

Discussion

26S proteasome subunits are known to be tightly co-regulated at both the protein and expression level (Kurepa et al., 2009; Webb and Westhead, 2009). Here we show that RPX controls the transcription of genes encoding proteasome subunits and proteasome-associated proteins by interacting with a conserved cis-regulatory element (PRCE). As a consequence, RPX modulates 26S proteasome activity and thereby plant size (Figure 1), thus representing a new mechanism underlying organ size determination, in which RPX acts in a non-cell-autonomous manner through the proteasome. Importantly, under controlled conditions, the final size of leaves is fixed and merely depends on genetic factors (Gonzalez et al., 2010) that control cell division and cell expansion (Potter and Xu, 2001). Loss of RPX affects cell number, while over-expression affects both cell number and size. As RPX expression appears to be restricted to the vascular tissue, a currently unknown vascular cell-derived signal may contribute to the determination of final organ size (Ranjan et al., 2011). Similarly, the promotion of adaxial cell fate by YABBY TFs occurs in a non-cell-autonomous manner (Stahle et al., 2009). YABBY genes are expressed in abaxial cells but are proposed to control adaxial identity through a short-range signaling pathway (Goldshmidt et al., 2008). RPX may act through a similar short-range mechanism.

RPX acts upstream of genes encoding 26S proteasome subunits (Table S1) and some other genes, highly similar to the situation found in yeast (Saccharomyces cerevisiae) for the TF Rpn4 (Mannhaupt et al., 1999). However, Rpn4 and RPX do not share any obvious sequence homology, indicating that plants and fungi have evolved different TFs and cis-regulatory elements for the control of similar gene regulatory networks. On the other hand, the PRCE cis-element (Figure 5) is conserved between the promoters of higher plants including eudicots (e.g. Arabidopsis) and monocots (e.g. rice), but is absent from Chlamydomonas. Of note, our PRCE element deviates from a previously reported binding site for RPX (Yabuta et al., 2011) identified by a cyclic amplification and selection of targets (CASTing) assay. Deletion of the PRCE element in RPN10 promoter abolishes transactivation by RPX (Figure 6a), thus it is unlikely that the previously reported cis-element (also present in the RPN10 promoter, at −927 bp upstream of the transcription start site) is required for RPX to control expression.

It is interesting that RPX controls the expression of UMP1-like protein-encoding genes, which are required for the assembly and activation of the 20S core of the proteasome in yeast, human and mouse (Burri et al., 2000), but so far have not been functionally characterized in plants. As expected, the differential expression of RPX correlates with proteasome activity. In previous studies, over-expression of individual subunits of the proteasome did not appear to affect its total activity (Kurepa and Smalle, 2008). Therefore, analysis of the 26S proteasome has mainly focused on loss-of-function mutants for individual subunits. Loss of individual subunits such as RPT2a, RPN10 and RPN12a causes a decrease in cell number but an increase in cell size in Arabidopsis (Kurepa et al., 2009; Sako et al., 2010). Similarly, RPT5a is a negative regulator of leaf cell size and endo-reduplication like RPT2a (Sako and Yamaguchi, 2010), while RPT5b is a positive regulator of leaf size that is involved in glucose signaling (Cho et al., 2006). In addition, RPN12a and RPN10 have been implicated in cytokinin and abscisic acid responses, respectively (Smalle et al., 2002, 2003). Over-expression of RPX drastically reduces both cell size and cell number, but loss of RPX only affects cell number. As individual subunits may both increase and decrease leaf area, it is interesting that positive transcriptional regulation of the proteasome genes by RPX mainly acts negatively on leaf area. Of note, the PRCE element is found in the promoters of RPT2a, RPN10 and RPN12a, but not in the promoter of RPN5a (which is also not affected at the transcriptional level; Figure 4), suggesting that RPX may promote assembly of a certain cell size-specific proteasome, as also indicated for RPT2a (Sako et al., 2010). Moreover, our study suggests that RPX regulates proteasome activity at the tissue-specific level, as expression of RPX is mainly restricted to vascular tissue.

Although the gene regulatory network controlled by RPX is probably evolutionarily conserved between multicellular plants, the relationship between multicellular development and proteasome function must be addressed further to unravel how proteasome activity controls organ size. One potential way by which the proteasome regulates organ growth is through its interaction with elements of the cell division machinery (Marrocco et al., 2010; Eloy et al., 2011). Another mechanism may involve endo-reduplication, leading to higher cellular ploidy levels, which often correlate with increased cell and organ sizes (Melaragno et al., 1993; Cheniclet et al., 2005). Loss of RPT2a leads to extended endo-reduplication during early leaf development (Sonoda et al., 2009) and to higher ploidy levels in Arabidopsis trichomes (Sako et al., 2010). However, ploidy levels in various organs of rpt2a, rpn10 and rpn12a mutants do not correlate with cell and organ enlargement, suggesting that the effect of altered proteasome function on cell size is not only caused by a higher nuclear DNA content (Kurepa et al., 2009). Therefore, growth regulation by RPX may involve processes other than endo-reduplication.

Although we found that additional genes are induced by RPX, we demonstrate here that the decreased accumulation of rosette biomass in RPX over-expression lines is mainly caused by increased proteasome activity (Figure 8). In most cases, target proteins first need to be ubiquitinylated before they become substrates of the 26S proteasome (Vierstra, 2009). We showed that increased proteasome activity is sufficient to drastically affect leaf development, suggesting that protein turnover not only depends on the presence of ubiquitinylated proteins, but also on the amount of active 26S proteasome complexes.

Finally, our findings indicate a role for RPX in controlling the activity of the proteasome during UV–B stress (Figure S6). For RPX, a role during high light stress through the regulation of flavonoid biosynthesis (Morishita et al., 2009) and a possible involvement of changes in proteasome gene expression have been reported previously (Yabuta et al., 2011). However, as flavonoid-related genes did not appear to be affected within 2 h of RPX induction (Table S1), we conclude that flavonoid accumulation is a secondary effect downstream of the immediate RPX gene regulatory network.

In conclusion, RPX functions as a novel regulator of organ size in a non-cell-autonomous manner through regulation of gene expression of 26S proteasome subunits. The direct effect of RPX on the expression of proteasome genes may be a useful tool to adjust proteasome activity in a cell type-specific manner.

Experimental Procedures

Plant material and growth conditions

Arabidopsis thaliana (L.) Heynh., accession Col–0, was used throughout this study. The T–DNA insertion mutants for RPX (SALK-040812C and SALK-025098) were obtained from the European Arabidopsis Stock Centre (http://arabidopsis.info/). Homozygous lines were identified via PCR screening of genomic DNA using gene-specific primers together with T–DNA-specific primers (Table S2). The exact locations of the T–DNA insertions were confirmed by sequencing the PCR products obtained using the T–DNA border primer and gene-specific primers. Plants were grown either on soil or on half-strength Murashige & Skoog (MS) medium supplemented with 1% sucrose. For soil growth, seeds were vernalized in darkness for 3 days at 4°C, and then transferred to a climate chamber set at 22°C with a 16 h light/8 h dark cycle, relative humidity of 75% and fluorescent light at 145 μE m−2 sec−1. For experiments on plates, seeds were surface-sterilized with 70% ethanol (1 min), 2% sodium hypochlorite (10 min), and rinsed with sterile water (six times). Seeds were stratified on plates for 72 h in the dark at 4°C, and then transferred to 22°C with a 16 h light (100 μE m−2 sec−1)/8 h dark period. For β–estradiol treatment, 14-day-old seedlings were selected and synchronized in half-strength MS liquid medium under long-day conditions for one night. Subsequently, fresh medium with 10 μm β–estradiol was added to the seedlings in a 12-well plate and the same amount of inducer-free medium was added to control samples as a mock treatment.

Plasmid constructs and analysis of transgenic plants

To generate the 35S::RPX and 35S::RPXΔC construct, the full-length and truncated RPX coding sequences were obtained from cDNA by PCR (Table S2) and cloned into pENTR/D (Invitrogen; www.invitrogen.com). Both products were ligated into pK7FWG2.0 (Karimi et al., 2002) using the BsrGI sites (Fermentas; www.fermentas.de). The direction of the coding sequence in the pK7FWG2.0 vector was confirmed by PCR and sequencing. To obtain an inducible over-expression vector, the RPX∆C fragment was amplified by PCR using primers with XhoI and SpeI overhangs (Table S2), and cloned into the corresponding sites of the pER8 vector (Zuo et al., 2000), yielding plasmid RPXCIOX. Successful integration of the RPX coding sequence into the vector was confirmed by sequencing.

For GUS constructs, the full-length RPX promoter covering a 1672 bp genomic fragment including the 222 bp 5′ UTR was amplified by PCR (Table S2) and cloned into pENTR/D. Similarly, the truncated promoter regions were amplified by PCR (Table S2 and Figure 2) and sub-cloned into pENTR/D (Invitrogen; www.invitrogen.com). The various RPX promoter fragments were transferred to pKGWFS7.0 by LR (Invitrogen; www.invitrogen.com) recombination (Karimi et al., 2002).

For GFP constructs, the full-length RPX coding sequence in pENTR/D was recombined into pMDC44 (Curtis and Grossniklaus, 2003) to create an N–terminally fused GFP–RPX construct. The RPXΔC coding sequence was cloned into pENTR/D, and a C–terminal fusion with GFP was established by recombination into pK7FWG2.0.

All constructs were transformed into Arabidopsis using Agrobacterium tumefaciens strain GV3101 by the floral-dip method (Clough and Bent, 1996). Non-segregating T3 transgenic lines were obtained by genetic analysis, and transgene expression was determined by quantitative real-time PCR.

RNA extraction and quantitative real-time PCR

Total RNA was extracted using an RNeasy kit (Qiagen; www.qiagen.com), treated with DNase I (Ambion; www.ambion.de) and used for cDNA synthesis (1 μg) with an oligo(dT) primer and a RevertAid first-strand cDNA synthesis kit (Fermentas) according to the manufacturer's instructions. Quantitative real-time PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems; www.appliedbiosystems.com). ACTIN2 (AT3G18780) served as the reference gene. Relative transcript abundance was calculated by the comparative CT method (Schmittgen and Livak, 2008). Sequences of oligonucleotides (Table S2) used for expression profiling were designed using quantprime (Arvidsson et al., 2008).

GUS staining and cross-sections

GUS staining was performed as previously described (Plesch et al., 2001); tissues were examined using a stereomicroscope (Leica; www.leica.com). For cross-sections, GUS-stained leaves were washed twice in 80% ethanol, once in 90% and once in 100% ethanol for 40 min each. The incubated samples were stored in 100% ethanol at −20°C for 3–4 days, and then embedded in Technovit (Heraeus Kulzer Technik; http://heraeus-kulzer-technik.de). To achieve this, the samples were incubated in solution A (Technovit A7100:ethanol, 1:1) and rotated at room temperature for 24 h. Subsequently, samples were incubated in solution B (Technovit A7100:ethanol, 15:1) and kept at 4°C for at least 2 days before sectioning using a RM2025 microtome (Leica).

Root length and meristem cell number analysis

After germination, 4-day-old seedlings were selected and placed on new plates and grown vertically. After transfer, root length was recorded daily for 8 days and analyzed using imagej software (http://rsbweb.nih.gov/nih-image). Root meristem cell number was counted at 4 days after germination using a differential interference contrast (DIC) microscope as described by Ioio et al. (2007).

Leaf cell size and cell number

The first leaf pair of 22-day-old plants was cleared by adding 1.5 ml ethanol:acetic acid (7:1). After 1 h, leaves were washed once with 100% ethanol and once with 50% ethanol for 2 h at room temperature. Subsequently, leaves were placed in 1 m KOH solution and incubated on a rotary shaker overnight at room temperature. Cell sizes were measured using ImageJ from photographs taken using a B51 microscope (Olympus; www.olympus.de) with DIC optics.

Transactivation assay

Arabidopsis mesophyll cell protoplasts were isolated and transformed according to the tape–Arabidopsis sandwich method (Wu et al., 2009). Constructs for transactivation assays were generated by cloning 1000 bp promoters of RPT6a (AT5G19990), PBE1 (AT1G13060), RPN12a (AT1G64520) and RPN10 (AT4G38630) by PCR (Table S2). After cloning into pENTR/D, the promoters were recombined into the p2GWL7 vector (Licausi et al., 2011). The full-length and truncated RPX coding sequences were cloned into the CaMV 35S over-expression vector p2GW7 (Karimi et al., 2002). Protoplasts were co-transformed with the p2GWL7 vectors, normalization vector 35S::RLUC (Licausi et al., 2011) and p2GW7 constructs, using 5 μg of each plasmid. A dual luciferase reporter assay (Promega; www.promega.com) was performed according to the manufacturer's instructions, and luminescence was measured using a GloMax 20/20 Luminometer (Promega).

Western blot analysis

Crude leaf extracts obtained using CelLytic™ P Cell Lysis buffer (Sigma; www.sigmaaldrich.com) were used for Western blotting. Protein samples were separated by 12.5% SDS–PAGE. SDS–PAGE-separated proteins were transferred onto Protran nitrocellulose membrane (Whatman; www.whatman.com). The membranes were blocked for 1 h in 1% Western blocking reagent in Tris Buffered Saline (TBS) (Roche; www.roche.de), followed by incubation for 1 h with primary polyclonal antibody directed against a 26S proteasome subunit. RPN10 was detected using the antibody DS5a (Agrisera; www.agrisera.com), and PBA1 was detected as previously described (Gu et al., 2010). Membranes were washed twice in TBS containing 0.1% Tween–20 for 10 min each, then washed twice with 0.5% blocking solution for 10 min each, and incubated for 1 h with secondary anti-horseradish peroxidase antibody (AS09 548; Agrisera). All incubations were performed at room temperature and antibodies were diluted in 0.5% blocking buffer. Detection of protein bands was performed using the ECL Plus Western blotting detection system (GE Healthcare; www.gehealthcare.com), and the signals were captured using CL–XPosure Film (Thermo Fisher Scientific; www.thermofisher.com).

Proteasome activity measurements

Detection of 26S proteasome activity by the MV151 probe (Verdoes et al., 2006) was essentially performed as previously described (Gu et al., 2010). Additionally, 20S proteasome activity was determined using the 20S proteasome activity assay kit (Chemicon; www.chemicon.com) according to the manufacturer's instructions. In short, crude protein extracts from 2-week-old seedlings of RPX∆C-IOX were used for activity measurements. Thirty microliters of the crude extracts and 10 μl of proteasome substrate were added to a 96-well microtiter plate in 1 x assay buffer for fluorescence measurements. The fluorescent signal was measured every 5 min for 2 h, with excitation and emission at 360 and 480 nm, respectively.

Infrared-mediated mapping

Binding sites of RPX were tested using infrared-mediated mapping (IMAP) as described in detail in Methods S1. Briefly, RPX fused to glutathione-S-transferase (GST) and GST alone (negative control) were expressed in vitro from the plasmids pIVEX2.3d-RPX-GST and pIVEX2.3d-GST, respectively, coupled to glutathione magnetic beads (Promega), and incubated with IRDye800CW-labeled double-stranded oligonucleotides. Probes were generated by PCR (Table S2) via annealing of the specific oligonucleotides with an IRDye800CW-labeled universal primer. After incubation, proteins and bound DNA were eluted from the beads and transferred to 96-well microtiter plates with a clear flat bottom (Carl Roth; www.carlroth.com). Signal intensities were determined by scanning at 800 nm using an Odyssey infrared imaging system (LI–COR; www.licor.com). RPX–GST signals were normalized to the GST signal and plotted.

Chromatin immunoprecipitation/quantitative PCR

Binding of RPX to target gene promoters was analyzed using the Epitect CHIP qPCR kit (Qiagen). ChIP was performed according to the manufacturer's instructions, except for the first steps in which we performed cross-linking on leaf protoplasts obtained from wild type and RPX∆C-GFP plants using anti-GFP antibody purchased from Roche. Chromatin cross-linking was performed using 1% formaldehyde. Purified DNA (in 200 μl) was used for quantitative real-time PCR (1 μl per reaction) to determine enrichment of potential RPX binding sites using the primers listed in Table S2.

Comparative phylogenetic analysis

Functional cis-acting promoter elements are subjected to evolutionary pressure and are consequently more likely to be conserved in evolution in comparison with non-functional sequences (Picot et al., 2010). For phylogenetic footprinting, we screened for the presence of the RPX binding site using the MEME analysis suite (http://meme.sdsc.edu/meme) and by sequence search in the promoters of orthologous 26S proteasome genes in Oryza sativa, Ricinus communis and Chlamydomonas reinhardtii. Orthologs of the Arabidopsis genes were identified using the phytozome database (http://www.phytozome.net/).

Microarray data

The microarray dataset has been deposited in the Gene Expression Omnibus (GEO) under accession number GSE34837.

Acknowledgements

We thank Karina Schulz for technical assistance, Dr Eugenia Maximova for help with microscopic work, and Dr Karin Koehl and her ‘Green Team’ for plant care (Max Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany). We thank the Bundesministerium für Bildung und Forschung for funding the GoFORSYS Research Unit for Systems Biology (GoFORSYS-Potsdam Golm Bundesministerium für Bildung und Forschung Forschungseinrichtung zur Systembiologie; Photosynthesis and Growth: A Systems Biology Based Approach, FKZ 0313924) and the Vietnam Ministry of Education and Training for granting a fellowship to H.M.N. O.G.R and R.V.D.H. were financially supported by the Max Planck Society.

Ancillary