Partially Redundant Activity of TCP Genes

We have previously shown that the jaw-D mutant phenotype, with epinastic cotyledons and conspicuously crinkled leaves, is caused by the overexpression of miR319a [32]. To determine the contribution of different miR319 targets to this phenotype, we identified insertional alleles for TCP2, TCP4, and TCP10. Loss-of-function alleles for all three genes had slightly epinastic cotyledons and slightly enlarged leaves (Figure 1). Loss of TCP4 function in addition caused plants to produce, on average, seven additional leaves before flowering (Figure S1), similar to the delay observed in jaw-D mutants [32]. tcp2 tcp4 double mutants showed a further increase in leaf size, with some signs of crinkling. tcp2 tcp4 tcp10 triple mutants had the most obvious defects, but were still less strongly affected than jaw-D plants were (Figure 1). Interestingly, among plants that overexpressed miR319a from a constitutive 35S promoter, weak lines had bigger, but not crinkly leaves, similar to the tcp single knockout plants (Figure S2). In summary, the similar phenotypes of tcp loss-of-function mutants and miR319 overexpressers confirmed the conclusion from microarray and other analyses, that the TCP genes are the major targets of miR319 [3,8,32]. On the other hand, that all single mutants were only weakly affected indicated partially redundant function of the different TCP genes in wild type. These general conclusions are in broad agreement with defects reported for plants expressing dominant negative alleles of TCP genes, which mimic many phenotypes of jaw-D plants [42].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Phenotypes of Plants with Altered TCP Activity

Ten-day-old seedlings (top) and fully expanded sixth rosette leaves (bottom) of wild-type, various tcp mutant combinations, and jaw-D plants, in which miR319a, a negative regulator of TCP2, TCP3, TCP4, TCP10, and TCP24, is overexpressed.

https://doi.org/10.1371/journal.pbio.0060230.g001

We also prepared plants that expressed a mutant form of TCP4 linked to GREEN FLUORESCENT PROTEIN (GFP) sequences under the control of TCP4 regulatory sequences (rTCP4:GFP). In these plants, TCP4 mRNA escapes regulation by miR319 due to synonymous changes that reduce sequence complementarity to miR319 [32]. rTCP4:GFP plants have a similar, but generally milder phenotype than rTCP4 plants [32]. Because many more survive to adulthood, we were able to analyze the effects of increased TCP levels beyond the seedling stage. Several phenotypic aspects of these plants are opposite to those seen in tcp loss-of-function or jaw-D mutants. For example, their cotyledons are hyponastic (bent upwards) and hypocotyls are longer than those of wild-type plants (Figure S3A), contrasting with the shorter hypocotyls of jaw-D (Figure S3B). The rosette leaves of rTCP4:GFP plants were smaller, more rounded, and often darker green than those of wild type (Figure S3C), which contrasts with the larger leaves of jaw-D mutants. In summary, these results indicated that a variety of leaf sizes and shapes can be obtained by manipulating the levels of miR319 and its targets, the TCP genes, similar to what has been reported for the tomato homologs [41].

Effects of Altered TCP Levels on Genome-Wide Expression Profiles

To identify potential target genes of the miR319-regulated TCPs, we analyzed the results from several microarray experiments (Table S1). We separately compared leaves and shoot apices of wild-type plants with jaw-D plants, which have increased miR319a activity and therefore reduced TCP activity. In a third comparison, we analyzed apices from tcp2 tcp4 double mutants and rTCP4:GFP plants, which have increased TCP activity. We focused on genes that are likely to be positively regulated by TCPs, as indicated by reduced expression in jaw-D or tcp2 tcp4 plants, or increased expression in rTCP4:GFP plants. Because only nine genes were significantly down-regulated in tcp2 tcp4 double mutant apices, and only two of these were not detected in one of the other three comparisons, we omitted this dataset from further analyses. The weak transcriptional effects seen in tcp2 tcp4 double mutants are consistent with the weak morphological defects when compared with those of jaw-D plants, in which three additional TCP genes are strongly suppressed.

To obtain first insights into the potential role of the TCP-responsive genes during development, we made use of a developmental microarray dataset [53]. The averaged relative expression levels of the gene sets identified as differentially expressed in each experiment were highly similar, even though there was only partial overlap between them (Figures 2A and 3A). In rTCP4:GFP plants, more genes are changed in their expression than in jaw-D. One explanation could be that in rTCP4:GFP plants, the TCP4 expression domain is expanded and hence more cells and tissues are affected than in plants with reduced TCP activity. In addition, overall TCP activity is merely attenuated in jaw-D plants, because of the incomplete clearing of TCP transcripts by miR319, and because of the partial redundancy between miR319-targeted and nontargeted TCP genes, all of which have similar expression patterns (Figure S4) and similar dominant-negative effects [42].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Genes Affected by Altered TCP Activity

(A) Averaged RNA expression levels of genes affected in different genotypes across a developmental microarray dataset [53]. Shown are genes that are apparently positively regulated by TCPs, because their expression is reduced in jaw-D, or increased in rTCP4:GFP. See Figure 3A for size and overlap of gene sets.

(B) Transcript levels of lipoxygenase genes LOX2, LOX3, and LOX4 in apices, measured by qRT-PCR (average of three independent measurements). Error bars indicate standard deviation.

(C) Averaged expression levels of hormone biosynthesis genes in shoot apices from different genetic backgrounds, normalized to wild type. Numbers of genes in each pathway given in parentheses.

(D) Expression profiles of genes that are changed in jaw-D or rTCP4:GFP, normalized across five different hormone and control treatments [73].

https://doi.org/10.1371/journal.pbio.0060230.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Identification of TCP Target Genes

(A) Identification of genes that respond to changes in TCP activity in at least two of three comparisons. Selection criteria were a combination of per-gene variance (p < 0.05, logit-T, [85]) and common variance (>2-fold).

(B) Overrepresented motifs in the promoters of the genes identified in (A). n is the number of instances across the genes indicated; p is the probability that this is a chance occurrence, corrected for multiple testing. The asterisk indicates a motif that is the reverse complement of the one above.

https://doi.org/10.1371/journal.pbio.0060230.g003

Main sites of expression of TCP-responsive genes were leaf-like organs, including cotyledons, rosette leaves, cauline leaves, and sepals, consistent with the tissues affected especially in jaw-D plants [32]. That expression of potential TCP targets persisted in leaves throughout senescence suggested that the miR319/TCP regulatory module might not only be important early in leaf development, but also during later stages.

Expression Levels of JA Biosynthetic Genes Affected by miR319/TCP Activity

Using stringent criteria (logit-T per-gene variance p < 0.025, common variance > 2-fold), only a single gene, LIPOXYGENASE2 (LOX2), was identified as being affected in the different microarray comparisons (Table S2). LOX2 was the second most suppressed gene in our original analysis of jaw-D plants, after TCP4 itself [32]. The opposite effects observed in plants with reduced and increased TCP activity, respectively, indicated that TCPs are important determinants of LOX2 expression levels in the absence of other stimuli known to affect LOX2 expression, such as wounding [54,55].

LOX2 encodes a chloroplast-localized lipoxygenase that catalyses the conversion of α-linolenic acid (18:3) into (13S)-hydroperoxyoctadecatrienoic acid, the first dedicated step in the biosynthesis of the oxylipin JA [56]. Apart from LOX2, the Arabidopsis genome encodes three other lipoxygenases that are predicted to be chloroplast-localized, LOX3, LOX4, and LOX6 [57]. The expression of LOX3 and LOX4 could not be detected by microarray analysis, but more sensitive reverse transcription followed by real-time PCR showed that expression of both genes is reduced in jaw-D plants, and increased in rTCP4 plants as well (Figure 2B).

Since JA is regulated through a positive feedback loop, with JA inducing the expression of its own biosynthetic genes [55,56,58], we examined the effect of miR319/TCP on the entire biosynthesis pathway for JA and other oxylipins, for which 19 genes have been described in Arabidopsis. The first steps in JA biosynthesis occur in the chloroplast, and only the JA precursor OPDA (or its coenzyme A [CoA] ester) are transported into the peroxisome, where several rounds of β-oxidation are carried out, leading to the final product, JA [59,60]. We plotted the average expression level of the JA biosynthesis genes against the different genotypes that were subjected to microarray analysis. The average expression of JA biosynthetic genes was approximately 2-fold reduced in jaw-D plants compared to wild type, and approximately 4-fold increased in rTCP4:GFP plants (Figure 2C). We also analyzed the pathway for the hormones cytokinin, gibberellic acid, and auxin, all of which have been implicated in leaf development or leaf physiology. None of the other three pathways showed as great a contrast between wild-type, rTCP4:GFP, and jaw-D plants as the JA pathway (Figure 2C, Table S1). When we analyzed publicly available microarray data for JA response, we found the data to be consistent with an effect of miR319-regulated TCPs on endogenous JA levels, since several genes that are either down-regulated in jaw-D and tcp2 tcp4 plants or up-regulated in rTCP4:GFP plants are induced in wild-type plants treated with methyl jasmonate (MeJA) (Figure 2D). These include genes known from the literature to be responsive to MeJA, such as PDF1.2 and COR1 [61,62] (Table S3).

Identification of a JA Biosynthetic Gene as a TCP Target

To understand how miR319a regulates the JA and oxylipin biosynthesis pathways through the TCP transcription factors, we turned again to the microarray data that we had obtained from the different tissues and genotypes with altered miR319/TCP activity, and we searched for genes that appeared to be positively regulated by TCPs. With slightly relaxed parameters (logit-T per-gene variance p < 0.05), we identified a set of 117 genes with consistently changed expression (down in jaw-D and up in rTCP4:GFP) in at least two of the three analyzed tissues (Figure 3A). In the promoters of this set, the most common motifs were GGACCA and its complement, TGGTCC, which were present at least once in 49 genes (Figure 3B and Table S1).

In parallel, we identified the preferred binding site of TCP4 by in vitro selection [63]. Of 27 clones obtained after ten rounds of selection, 25 contained a variant of the consensus motif gGGaCCAC, which includes as a core the GGACCA motif found in the promoters of TCP-response genes (Figure 4A and Figure S5). Competition experiments with unlabeled oligonucleotides confirmed the specificity of the TCP4 binding site (Figure 4B). Electrophoretic mobility shift assays (EMSAs) with oligonucleotides that contained single base pair mutations indicated some flexibility in the ability of TCP4 to bind its preferred site in vitro (Figure 4C), which may explain why the motif deduced from in silico promoter analysis is only a submotif of the one identified by stringent binding site selection. The complement of the gGGaCCAC motif is related to a sequence, G(T/C)GGNCCC, that is preferentially bound by PCF5, a protein encoded by an miR319-targeted TCP gene from rice [38].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. DNA-Binding Specificity of TCP4

(A) Consensus binding motif of TCP4, based on a common sequence found in 25 out of 27 clones selected with recombinant TCP4 protein (see Figure S5).

(B) Specificity of TCP4 in vitro binding as shown by EMSAs. Unlabelled double-stranded oligonucleotides that contain the wild-type consensus sequences (agatgggGGACCACatagatgt) or a mutated version (…GGAACAC…) were used in increasing amounts as competitors.

(C) TCP4 binding to mutant sites, based on the consensus sequence used in (B).

https://doi.org/10.1371/journal.pbio.0060230.g004

In plants, metabolic pathways are often coordinately regulated by the same transcription factors [64], and we found the TCP motif GGACCA in the promoters of eight out of 19 oxylipin biosynthesis genes (Table S2). Only two promoters were expected to have this motif by chance, using the promoters of all Arabidopsis genes to determine the background distribution of the GGACCA motif. No such overrepresentation was found in the promoters of 13 auxin, 13 cytokinin, and 15 GA biosynthetic genes, with the GGACCA motif being present in the promoters of only two GA biosynthetic and one auxin biosynthetic genes, and missing in the promoters of cytokinin biosynthesis genes (Table S2).

To investigate whether the TCP binding sites were indeed required for promoter activity of JA biosynthetic genes, we focused on the LOX2 promoter, which has four sites with at most one mismatch to the motif GGACCAC. Using double-stranded oligonucleotides covering the potential TCP binding sites in the context of the LOX2 promoter, we performed EMSAs. The in vitro studies confirmed that TCP4 can bind strongly to at least two of the consensus motifs (Figure 5A). To assess the requirement for these binding sites in planta, we constructed two LOX2:GUS (β-glucuronidase) reporters, one with the wild-type sequence and one in which the four consensus motifs were mutated. In untreated plants, the wild-type reporter had strong GUS activity throughout leaves, similar to what has been reported [55], whereas the mutant reporter had very little activity (Figure 5B). Moreover, the wild-type reporter was less active in tcp2 tcp4 tcp10 triple mutants (Figure S6), confirming that TCPs positively regulate LOX2 promoter activity. Together, our findings suggest that the miR319-targeted TCPs directly regulate expression of LOX2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Functional Analysis of TCP4 Binding Sites in the LOX2 Promoter

(A) TCP4 binding to four consensus sites in LOX2 promoter. (1) tcagatcctGGACCACtgcaataa; (2) tattaattaAGACCACtcgtaact; (3) ttttaagcaGGACCAAaacctaaa; and (4) tagatacaGTGGTCCtcctatgca. consensus motifs are underlined, flanking sequences are from LOX2 promoter.

(B) X-gluc (5-bromo-4-chloro-3-indolyl-beta-D-glucuronic acid) assay for β-glucuronidase (GUS) reporter activity. In the mutant promoter, all four TCP4 consensus sites were mutated. MeJA and wounding treatments were for 45 min. The photographs show representative results from 20 independent transgenic lines analyzed for each construct.

https://doi.org/10.1371/journal.pbio.0060230.g005

LOX2 strongly responds to wounding or treatment with MeJA [55,56]. We tested if the mutated LOX2 reporter lacking TCP binding sites was still responsive to these stimuli. Wounding of rosette leaves or external treatment of plants with MeJA led to strong activation of reporter activity within 45 min (Figure 5B), indicating that TCPs are not involved in these two responses, but rather regulate the developmental aspect of LOX2 expression.

JA Levels Affected by Reduced TCP Activity

The strongly reduced expression of JA biosynthetic genes in jaw-D plants prompted us to ask whether this is accompanied by a reduction in levels of endogenous JA. We measured the concentration of JA in planta by using gas chromatography followed by mass spectrometry [65]. JA levels are low in resting tissue of both wild-type and jaw-D plants (Figure 6). A difference between wild-type and jaw-D plants was most obvious in response to wounding, which strongly induces JA biosynthesis. Near the peak of JA induction in wild type, 90 min after wounding [66], JA levels had substantially increased in both wild-type and jaw-D plants, but were about four times lower in jaw-D plants. This result is consistent with the observation that wounding can still activate a LOX2 promoter lacking TCP4 binding sites.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. JA Content of Wild-Type and jaw-D Plants

JA concentration in rosette leaves was measured in triplicate at 0 and 90 min after wounding.

https://doi.org/10.1371/journal.pbio.0060230.g006

For the microarray analyses, we had used primary rTCP4:GFP transformants with relatively strong phenotypes, and found an increase in the expression of JA biosynthetic genes (Figure 2). Unfortunately, only plants with weak, almost wild-type–like phenotypes provided sufficient material for JA measurements, because rTCP4:GFP plants with strong phenotypes stay small, have a shorter life span than wild type, and do not produce seeds. No clear differences in JA levels were seen in rTCP4:GFP plants with mild leaf phenotypes (unpublished data).

Senescence in Plants with Altered TCP Activity

Neither allene oxide synthase (aos) mutants, which appear to be completely devoid of jasmonate [66], nor oxophytodienoate reductase3 (opr3) mutants show an obvious defect in an induced senescence assay (Figure S8). On the other hand, it is well known that exogenously applied MeJA can accelerate the final stage of leaf development, senescence (e.g., [67]), and several JA biosynthetic genes, including LOX2, are transiently induced during developmental senescence [49]. Thus, JA likely plays a role in the control of senescence, but is not essential for it, as pointed out before [47].

We had noticed that positively regulated TCP targets tend to be expressed at higher levels in older leaves of wild-type plants (Figure 2A). An opposite pattern was seen for genes that were down-regulated in rTCP4:GFP plants (Figure S7). Considering that the rTCP4:GFP samples analyzed consisted of apices with small, developing leaves, this observations suggested that the developmental age of rTCP4:GFP leaves is advanced relative to that of wild-type leaves. The up-regulated genes include several genes encoding WRKY transcription factors, so named after the first four amino acids of the conserved motif WRKYGQK, which is the hallmark of this protein family. One of these genes, WRKY53, is an important positive regulator of senescence [68,69], which is induced more than 30 times in rTCP4:GFP plants, although it lacks TCP4 consensus binding motifs in its promoter (Table S4). The precocious activation in rTCP4:GFP of genes that are normally expressed only later during leaf development is consistent with the role of the snapdragon TCP gene CIN as a regulator of the mitotic arrest front during early stages of leaf growth [40], and suggests a more general role for TCPs during leaf aging. This in turn led us to examine the hypothesis that rTCP4:GFP plants might show a premature onset of senescence, and that jaw-D plants show a delay in senescence.

Obvious effects were seen in jaw-D plants grown under long days; in these plants, leaf senescence was delayed by about a week (Figure 7A), which is similar to the effects seen in the senescence mutant oresara9 [70]. In rTCP4:GFP plants, senescence was slightly accelerated (Figure 7A), consistent with these rTCP4:GFP plants examined having only relatively mild morphological defects.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Senescence Responses of Plants with Altered TCP Activity

(A) Fraction of fifth rosette leaves that still contained chlorophyll after the leaves had stopped growing (wild type, n = 28; jaw-D, n = 31; rTCP4:GFP, n = 46).

(B) Chlorophyll contents in detached leaves of wild-type and jaw-D, floated on water in darkness (average of 12 leaves per genotype).

(C) Measurements of PSII photochemistry in detached leaves of wild-type and jaw-D (average of six leaves per genotype). Error bars indicate standard deviation.

https://doi.org/10.1371/journal.pbio.0060230.g007

Incubation of detached leaves in the dark induces senescence within days, and the onset of senescence can be accelerated by treatment with exogenous MeJA [57,67]. Although there are differences between induced and developmental senescence (e.g., [49]), we could confirm the delay observed in on-plant senescence with the in vitro assay. We monitored chlorophyll degradation and maximum efficiency of photosystem II (PSII) photochemistry (F_v/F_m) in detached jaw-D leaves incubated in the dark, and found a delayed decline in both these indicators of healthy leaves (Figure 7B and 7C).

We used the in vitro assay also to determine whether the delayed senescence in jaw-D plants is potentially caused by a lack of JA or a defect in JA signaling. When we compared jaw-D to wild type, we found that treatment with exogenous MeJA restored the senescence response (Figure 8), consistent with our previous results that TCPs regulate JA biosynthesis, rather than the JA response.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Restoration of Senescence Response in jaw-D

Detached leaves of wild-type (left leaf of each pair) and jaw-D were incubated in darkness with increasing concentrations of MeJA. Note that these assays were carried out to a different time point and in a different laboratory, compared to the experiments shown in Figure 7, which likely explains the differences in senescence progression.

https://doi.org/10.1371/journal.pbio.0060230.g008

The findings that MeJA was sufficient to restore senescence in jaw-D plants, but that JA on its own is apparently not essential for senescence, suggests that JA acts redundantly with other pathways during the control of senescence. One candidate is salicylic acid (SA) signaling, which often antagonizes the effects of JA [71]. However, jaw-D plants appeared to be largely normal in their SA response, as deduced from induction of the marker gene PR1 (Figure S9).

Regulation of Growth and JA Biosynthesis by TCPs

Using a combination of microarray meta-analysis, in vitro DNA binding experiments and reporter gene studies, we identified the LOX2 gene, which encodes an enzyme catalyzing a key step in JA biosynthesis, as being likely to be directly regulated by TCPs in vivo. The transcriptional response of other genes in the JA biosynthesis pathway and the overrepresentation of a TCP DNA binding motif in this pathway suggest that TCPs directly control additional JA biosynthetic genes. This strategy, coordinated control of metabolic pathways by the same set of transcription factors, is commonly used in plants [64].

Several previous analyses of JA biosynthetic genes, including LOX2, have focused on regulatory elements and upstream factors mediating the effects of wounding or MeJA treatment [55,56]. Mutation of the TCP binding sites in the LOX2 promoter strongly reduced its activity in the absence of stimulation by wounding or MeJA, but it did not abolish the inducibility of the promoter. Our results highlight the importance of developmental control of LOX2, and of the fact that developmental regulation can be at least partially uncoupled from transcriptional induction by wounding or JA treatment. This finding is consistent with JA playing not only a role in pathogen and stress response, but also in many developmental processes. Expression of the TCP genes themselves is not wound-, pathogen, or MeJA-responsive, as deduced from publicly available microarray data (http://www.weigelworld.org/resources/microarray/AtGenExpress/) [72,73], supporting the conclusion that the TCPs represent a pathway of JA regulation that is linked to the developmental program of the plant rather than to environmental responses.

Plants with lower JA levels due to reduced activity of the enzyme encoding genes DONGLE (DGL) and OPR3 have been reported to be larger than wild type, while plants that overexpress DGL are smaller, similar to plants treated with JA [74,75]. DGL shares overlapping activity with a homolog, DEFECTIVE IN ANTHER DEHISCENCE 1 (DAD1), in stamen maturation [76]. DAD1 in turn is a direct target of the homeotic transcription factor AGAMOUS (AG), which regulates both organ identity during early flower development and organ development during later stages [77]. In light of these related findings, the observation that TCP transcription factors and JA have parallel effects on leaf growth suggests that the oxylipin pathway potentially acts downstream of TCPs in affecting growth. Importantly, several links between JA and cell cycle progression as well as growth have previously been demonstrated (e.g., [78,79]).

Effects of TCPs and JA on Senescence

Although it is well known that exogenously applied MeJA can accelerate senescence (e.g. [67]), there have been no reports that plants with mutations in the JA biosynthetic pathway are deficient in the senescence program [59,66,76,80–84], which we have confirmed for aos and opr3 mutants using an induced senescence assay (Figure S8). Nevertheless, a bona fide effect of JA on leaf senescence can be deduced from the observation that exogenously applied MeJA fails to induce senescence in the coronatine insensitive1 (coi1) mutant, which is defective in JA signal transduction [57,75,85]. The coi1 mutant on its own, however, does not show a senescence defect either. Analyses of biosynthetic mutants as well as the coi1 signaling mutant therefore both suggest that endogenous JA is not limiting for natural senescence.

There is thus an interesting contrast between JA biosynthetic and signaling mutants on the one hand, and jaw-D, which has decreased JA levels due to reduced expression of JA biosynthetic genes, on the other hand. We initially considered the possibility that the TCP4 target LOX2 might be required for the production of additional metabolites that prevent senescence, and that LOX2 might thereby directly affect chloroplast stability. There are no reports that LOX2 catalyzes processes other than the conversion of α-linolenic acid (18:3) into (13S)-hydroperoxyoctadecatrienoic acid, but the LOX2-catalyzed step can lead to end products other than JA [66]. The functions of these other oxylipins are not well known, and it cannot be ruled out that LOX2-catalyzed products are involved in JA-independent processes that delay senescence. An observation that speaks against such a scenario is that lipoxygenase activity is almost undetectable in leaf extracts of a recently isolated loss-of-function lox2 mutant. Similar to other JA biosynthetic mutants, these lox2 mutants do not show obvious changes in their senescence program (L. Dubugnon and E. E. Farmer, unpublished data).

We therefore propose an alternative scenario, namely that miR319-regulated TCPs control leaf senescence by regulating not only JA biosynthesis, but also a second, as-yet unidentified pathway that suppresses senescence in wild-type plants. We speculate that inactivation of the endogenous JA pathway alone is not sufficient to delay precocious senescence, due to such a second, redundantly acting pathway. Because of their parallel effects on senescence, JA alone should be, however, sufficient to induce senescence, both in wild-type plants and in jaw-D plants, which presumably lack activity of both pathways. Many genes activated in rTCP4:GFP plants are progressively up-regulated during leaf development, including WRKY53, an important positive regulator of senescence [68,69], suggesting perhaps a more general role for TCPs during leaf aging.

Most conserved plant miRNAs affect transcription factor genes with important roles in development [11,12], but in vivo targets that mediate the effects of these transcription factors are largely unknown. Our identification of targets of miR319-regulated TCPs thus provides an important advance in the understanding of small RNA–controlled regulatory networks. In addition, it demonstrates that the function of miRNA-controlled transcription factors is not limited to the modulation of downstream hormonal responses [13–16,18], but that miRNAs may, in addition, regulate development through effects on hormone biosynthesis.

Plant material.

Plants were grown at 23 °C. All experiments were done under long days (16 h light), except for microarray analyses, which were with plants grown under short days (8 h light). Regular illumination was 125 μmol m⁻² s⁻¹. For low light conditions, intensity was reduced to 15 μmol m⁻² s⁻¹. Origin of tcp mutants and gene identifiers are given in Tables S5 and S6. Wild type was Columbia (Col-0), unless stated otherwise.

Microarray analyses and promoter motif discovery.

Microarray analyses using the Affymetrix ATH1 platform were performed as described [86]. For the collection of apices (including the youngest leaf primordia), plants were dissected under a stereomicroscope, and all leaves with visible petioles were removed and discarded. Tissue was harvested directly into liquid N₂. Differentially expressed genes were identified with a combination of per-gene variance (calculated using logit-T [87]) and common variance based on expression estimates using gcRMA (http://www.bioconductor.org), a modification of the robust multi-array analysis (RMA) algorithm [88]. Accession numbers for microarray experiments are GSE518 (jaw-D) [32], and E-MEXP-469 (rTCP4:GFP and tcp2 tcp4). Microarray data for hormone treatment were downloaded from http://www.arabidopsis.org; TAIR accession numbers are 1007965964 (JA), 1007965859 (auxin), and 1007966175 (GA).

Six to eight–nucleotides-long overrepresented motifs were identified using a routine implemented in Genespring GX 7.3.1 (Agilent Technologies, California). Promoters were defined as 800 nucleotides upstream of the initiation codon, and exact matches among positions −800 to −10 were considered. The frequency of each individual motif in the 117 genes that changed in at least two conditions was compared to the frequency of the same motif in promoters of other, randomly chosen genes. The Ath1_02_04 annotation was used.

Expression analyses.

Real-time RT-PCR using the Opticon Continuous Fluorescence Detection System (BioRad) was performed as described [86]. GUS staining was carried out as described [89].

Protein expression and purification.

The TCP4 expression construct pRSETC-TCP4-1, designed to express the amino-terminal, 224–amino acid fragment of TCP4 including the DNA-binding domain, was transformed into the Escherichia coli strain BL21 (DE3) pLysSpSBET. 100ml LB containing 100μg/ml ampicillin was inoculated with 1 ml of overnight culture and grown at 37 °C to mid-log phase. Recombinant protein expression was induced with 1 mM isopropyl β-L-thiogalactoside (IPTG). Cells were harvested after 3 h of induction. Cells were lysed by sonication in 2 ml of lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, 1 mg/ml lysozyme). The lysate was centrifuged and the supernatant was loaded onto a Ni-NTA spin column (Qiagen). Recombinant protein was eluted in 150 μl volume containing 500 mM imidazole. Eluted protein was dialyzed against 50 mM NaH₂PO₄, 300 mM NaCl and 10% glycerol for 6 h. Purification was monitored by protein blot using anti-His HRP conjugate antibodies (Qiagen).

Random binding site selection.

Methods described earlier [63] were used. The double-stranded oligonucleotide targets (R704), which contained random 18-mer sequences flanked by 19 bp defined sequences on both ends, were prepared by annealing oligonucleotides R704 (GGAAACAGCTATGACCATG [N]₁₈ ACTGGCCGTCGTTTTAC) and 704 (GTAAAACGACGGCCAGT) followed by primer extension with Klenow fragment. Recombinant protein was incubated with 3.6 μg of double stranded R704 in 15 μl of 1 X binding buffer containing 0.1 M KCl, 10 ng of salmon sperm DNA, and 10 μg of bovine serum albumin (BSA). The DNA-protein complex was separated by polyacrylamide gel electrophoresis, bound oligonucleotides were eluted from the gel and dissolved in 20 μl of TE. The recovered DNA was amplified by 14 cycles of PCR with primers 703 (GGAAACAGCTATGACCATG) and 704. The PCR product (30 μl) was extracted with phenol/chloroform and ether, and 10 μl was subjected to next round of selection. With each round of selection, the number of PCR cycles was reduced by one cycle to avoid generation of high–molecular weight PCR products. The DNA from the tenth round of selection was amplified by PCR, purified on a 20% polyacrylamide gel and cloned into the pGEM-T Easy vector (Promega) for sequencing.

EMSAs.

Double-stranded DNA probes were generated by annealing oligonucleotides and primer extension with [α-³²P]-dCTP using Klenow enzyme. The binding reaction was carried out in a total volume of 10 μl containing ∼10 fmol of oligonucleotide probe, 1 X binding buffer (20 mM HEPES-KOH, pH 7.8, 100 mM KCl, 1 mM EDTA, 0.1 % BSA, 10 ng herring sperm DNA, and 10% glycerol) and 5–100 ng of recombinant protein. The mixture was incubated for 30 min at room temperature and loaded on 6% native polyacrylamide gel. Electrophoresis was conducted at 4 V/cm for 45 min in 0.5 x TBE electrophoresis buffer at room temperature. The gels were autoradiographed using a phospho-imager.

Site-directed mutagenesis.

Mutagenesis was carried out using the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene), according to the manufacturers instructions. Primer sequences are available on request. The four sites and positions of mutations (with ATG as +1) are: (1) position −1173 to −179, mutated at −1174, −1176; (2) position −944 to −949, mutated at −944, −946, −949; (3) position −432 to −437, mutated at −432, −434; (4) position −300 to −305, mutated at −303, −305.

Jasmonate quantification.

Protocol 2 of Mueller and colleagues [65] with an oxygen-18 labeled internal standard was used.