- © 2018 American Society of Plant Biologists. All rights reserved.
Abstract
In Arabidopsis (Arabidopsis thaliana), reduced expression of the transcriptional regulator PEAPOD2 (PPD2) results in propeller-like rosettes with enlarged and dome-shaped leaves. However, the molecular and cellular processes underlying this peculiar phenotype remain elusive. Here, we studied the interaction between PPD2 and NOVEL INTERACTOR OF JAZ (NINJA) and demonstrated that ninja loss-of-function plants produce rosettes with dome-shaped leaves similar to those of ppd mutants but without the increase in size. We showed that ninja mutants have a convex-shaped primary cell cycle arrest front, putatively leading to excessive cell division in the central leaf blade region. Furthermore, ppd and ninja mutants have a similar increase in the expression of CYCLIN D3;2 (CYCD3;2), and ectopic overexpression of CYCD3;2 phenocopies the ppd and ninja rosette and leaf shape phenotypes without affecting the size. Our results reveal a pivotal contribution of NINJA in leaf development, in addition to its well-studied function in jasmonate signaling, and imply a new function for D3-type cyclins in, at least partially, uncoupling the size and shape phenotypes of ppd leaves.
Both between species and within the same organism, leaf morphology is diverse: leaves can be simple or compound, they can have margins that are smooth or serrated, and they can be flat or have various curvatures (Pérez-Pérez et al., 2002; Bilsborough et al., 2011; Dkhar and Pareek, 2014; Vlad et al., 2014). In Arabidopsis (Arabidopsis thaliana), leaf development starts with extensive cell proliferation throughout the leaf primordium. After a few days, cells at the tip of the leaf exit the mitotic cycle and a primary cell cycle arrest front moves in a tip-to-base direction, after which cells start to differentiate and expand (Donnelly et al., 1999; Kazama et al., 2010; Andriankaja et al., 2012). Simultaneously, meristemoid cells (i.e. small triangular cells that are dispersed throughout the leaf epidermis) can undergo up to three asymmetric divisions, allowing self-renewal and the formation of neighboring pavement cells (Bergmann and Sack, 2007; Pillitteri and Torii, 2012). Eventually, meristemoids differentiate into guard mother cells, which each divide symmetrically to produce the guard cells of a stomatal complex. Together with the primary arrest front, a secondary cell cycle arrest front that terminates the meristemoid cell divisions contributes to the total cell number and size of leaves (Geisler et al., 2000). Despite the knowledge of the cellular events determining leaf development, the molecular mechanisms governing the shape of growing tissues are still poorly understood. Genetic studies have identified a few genes that affect leaf flatness, such as the CINCINNATA-like TEOSINTE BRANCHED1/CYCLOIDEA/PCF (TCP) family involved in controlling the shape of the primary cell cycle arrest front (Nath et al., 2003; Efroni et al., 2008). Loss-of-function tcp mutants have a concave primary cell cycle arrest front, leading to excessive growth at the leaf margins compared with the central zone, resulting in enlarged and crinkly leaves. On the other hand, down-regulation of the PEAPOD genes (PPD1 and PPD2) results in plants demonstrating a typical propeller-like rosette composed of enlarged but narrow dome-shaped leaves (White, 2006; Gonzalez et al., 2015). This leaf morphology is at least partially due to excessive meristemoid asymmetric divisions hypothesized to be combined with a restricted extension capacity of the margin cells (White, 2006; Gonzalez et al., 2015). In addition, although the position of the primary cell cycle arrest front is not affected in the Arabidopsis ppd deletion (Δppd) mutant (White, 2006), the shape of the front is proposed to be convex, leading to a prolonged period of cell proliferation in the central leaf blade region (Karidas et al., 2015). Consequently, whereas the TCP proteins are important for the regulation of the shape of the primary cell division arrest front (Nath et al., 2003; Palatnik et al., 2003), the PPD proteins are hypothesized to act on both the primary (Karidas et al., 2015) and the secondary (White, 2006; Gonzalez et al., 2015) mitotic arrest fronts. However, how the PPD proteins control these cellular processes and how this ultimately results in the leaf size and shape phenotypes observed in ppd mutants remain largely elusive.
Previous studies of the PPD signaling pathway have identified PPD2 direct target genes (Gonzalez et al., 2015), PPD2 protein partners (Pauwels et al., 2010; Gonzalez et al., 2015), and proteins that mark PPD2 or its interaction partners for degradation (Wang et al., 2016; Li et al., 2018). PPD2 binds to the promoters of two of the three D3-type cyclin genes, CYCD3;2 and CYCD3;3, and represses their transcription (Gonzalez et al., 2015). CYCD3s are important for cell number determination in developing leaves (Dewitte et al., 2007), and meristemoid initiation and activity are reduced in the cycd3;1 cycd3;2 cycd3;3 triple mutant compared with the wild type (Elsner et al., 2012; Lau et al., 2014). To control CYCD3 gene expression, PPD2 acts in a complex with the transcriptional repressors KINASE-INDUCIBLE DOMAIN INTERACTING8 (KIX8) and KIX9, and reduced expression of KIX8 and KIX9 also leads to the formation of propeller-like rosettes with enlarged and dome-shaped leaves (Gonzalez et al., 2015). PPD2 is also proposed to interact with NOVEL INTERACTOR OF JAZ (NINJA; Pauwels et al., 2010; Gonzalez et al., 2015), a transcriptional repressor with a well-characterized function in the jasmonic acid (JA) signaling pathway (Pauwels et al., 2010; Acosta et al., 2013; Gasperini et al., 2015). ninja loss-of-function mutants show increased JA signaling and display shorter roots mimicking the JA-mediated growth inhibition, even in the absence of the hormone (Acosta et al., 2013; Gasperini et al., 2015). Although much is known of the role of NINJA in JA signaling in the root, it remains elusive if NINJA plays a role during leaf development or transcriptionally coregulates downstream targets of PPD2.
Here, we studied the PPD2-NINJA interaction in great detail and phenotypically and molecularly characterized ninja mutants. We found that NINJA, in addition to its known function in JA signaling and root development, has a key role in leaf development, likely due to the transcriptional regulation of CYCD3 expression. With these findings, we further elucidate the importance of the PPD signaling pathway in controlling leaf growth.
RESULTS
NINJA Interacts with PPD2 in Vitro and in Nicotiana benthamiana Leaves
In previously performed tandem affinity purification experiments in Arabidopsis cell suspension culture, NINJA was identified as a putative PPD2 protein partner (Pauwels et al., 2010; Gonzalez et al., 2015). We further investigated this interaction with an in vitro pull-down experiment (Fig. 1A). The PPD2 and NINJA proteins were fused to a GST and a His tag, respectively, and the fusion products were expressed in bacteria. Pull down of the GST-PPD2 fusion protein using glutathione-Sepharose beads and subsequent immunoblotting using an anti-His antibody confirmed that NINJA also interacts with PPD2 in vitro (Fig. 1A). To confirm the PPD2-NINJA interaction in planta, a bimolecular fluorescence complementation assay was performed. Coexpression of nYFP (N-terminal fragment of YELLOW FLUORESCENT PROTEIN)-NINJA and cYFP (C-terminal fragment of YFP)-PPD2 in N. benthamiana leaves resulted in a strong YFP signal in the nuclei of epidermal cells (Fig. 1B), confirming that PPD2 interacts with NINJA.
NINJA interacts with PPD2 in vitro and in N. benthamiana leaves. A, In a pull-down (PD) experiment, GST-PPD2 was immobilized on glutathione-Sepharose and the presence of His-NINJA was verified by immunoblotting (IB) using an anti-His antibody. GST-GUS was used as a negative control. The amount of GST-GUS or GST-PPD2 was visualized by Coomassie Brilliant Blue (CBB) staining. B, nYFP-NINJA and cYFP-PPD2 or cYFP (as a negative control) were coexpressed in N. benthamiana leaves. 4′,6-Diamino-phenylindole (DAPI) staining indicates the nuclei. The DAPI-stained and bright-field microscopic images are merged in the bottom panels. Bars = 20 µm.
Deletion of the NINJA C Domain Abolishes the Interaction with PPD2, and NINJA Loss of Function Results in Plants with Dome-Shaped Leaves
NINJA is characterized by three conserved protein domains: the A domain harboring a TOPLESS (TPL)-binding ERF-ASSOCIATED AMPHIPHILIC REPRESSION motif instrumental for transcriptional repression, the B domain with unknown function, and the C domain that mediates interaction with the JASMONATE ZIM-DOMAIN (JAZ) proteins (Pauwels et al., 2010). Similar to the JAZ proteins, the PPD proteins belong to class II of the TIFY protein family. PPD proteins are characterized by a specific N-terminal PPD domain essential for the interaction with the KIX proteins (Gonzalez et al., 2015), a ZINC-FINGER EXPRESSED IN INFLORESCENCE MERISTEM (ZIM) domain containing a highly conserved TIFY (TIF[F/Y]XG) motif and known to mediate homodimerization and heterodimerization among TIFY proteins, and a divergent C-terminal Jas domain (Bai et al., 2011; Cuéllar Pérez et al., 2014). To map the protein domains that are essential for the interaction between PPD2 and NINJA, we generated truncated versions of PPD2 that comprise different combinations of the three domains and tested these fragments for interaction with NINJA in a yeast two-hybrid (Y2H) assay. We found that the PPD2 ZIM domain is necessary and sufficient for the interaction with NINJA (Fig. 2A). A similar Y2H experiment using truncated versions of NINJA showed that PPD2 interacts with the C domain of NINJA (Fig. 2B). To uncover the potential involvement of NINJA in regulating leaf development, we phenotypically analyzed ninja-1 and ninja-2 mutants (Acosta et al., 2013; Gasperini et al., 2015). These mutants have a point mutation causing a premature stop codon in NINJA and, therefore, lack the C domain essential for interaction with PPD2 and show a significantly decreased expression of NINJA (Supplemental Fig. S1). When grown in soil for 25 d, ninja mutants produced propeller-like rosettes with narrow dome-shaped leaves (Fig. 2C) similar to those of ppd2 (Fig. 2C; Wang et al., 2016) and ami-ppd (Gonzalez et al., 2015) plants. To investigate whether leaf size was changed in ninja mutants, the leaf areas of 25-d-old plants were measured, and the results are summarized in a heat map as the average percentage difference compared with the wild type (Fig. 2D). In the ppd2 mutant, older leaves (L1–L6) were significantly larger and younger leaves (L9 onward) were significantly smaller than in the wild type. Although the younger leaves of ninja mutants were also significantly smaller (L6 onward), the older leaves had a similar size to those of the wild type (Fig. 2D).
PPD2 interacts with the NINJA C domain. A and B, PPD2 (A) and NINJA (B) truncations were tested in Y2H assays to identify the specific domains for the PPD2-NINJA interaction. Transformants containing bait and prey constructs were grown on medium lacking Leu and Trp (-L-T) or Leu, Trp, and His (-L-T-H). Protein domains are represented, and numbers indicate terminal amino acid residues. C, Columbia-0 (Col-0), ppd2, ninja-1, and ninja-2 plants grown in soil for 25 d. Side views of the seventh leaf (L7) are presented below the photographs. Bar = 1 cm. D, Area of the individual leaves of ppd2 and ninja plants grown in soil for 25 d (n = 3 biological replicates with approximately eight plants per replicate), analyzed using mixed models in the SAS Enterprise Guide. Statistically significant differences (P < 0.05) relative to the wild type are marked in boldface italic type.
In summary, we demonstrate that the C domain of NINJA interacts with the ZIM domain of PPD2 and that ninja loss-of-function mutants display a similar rosette and leaf curvature phenotype to ppd2 plants, without an increased final leaf size.
The ninja Leaf Phenotype Is Not Influenced by Alterations in JA Response or Biosynthesis
Because NINJA is a negative regulator of JA signaling (Pauwels et al., 2010; Acosta et al., 2013; Gasperini et al., 2015), ninja mutants might lack the enlarged leaf size phenotype observed in ppd2 mutants (Fig. 2D) as a result of an increased JA signaling pathway leading to growth inhibition (Acosta et al., 2013; Gasperini et al., 2015). To verify this, we used ninja plants crossed with the allene oxide synthase (aos) or coronatine insensitive1 (coi1-1) mutant, in which the JA biosynthesis or response is abolished, respectively (Feys et al., 1994; Park et al., 2002; Thines et al., 2007; Pauwels et al., 2010). The ninja-1 aos, ninja-2 aos, and ninja-1 coi1-1 double mutants were grown in soil for 25 d, and the size of the first four developing leaves (L1–L4) was measured. The leaf area of ninja aos (Fig. 3B) and ninja-1 coi1-1 (Supplemental Fig. S2B) double mutants was not changed drastically compared with that of ninja or the wild type. In addition, the propeller-like rosette and dome-shaped phenotype of the ninja aos or ninja-1 coi1-1 double mutant were indistinguishable from those of the ninja plants (Fig. 3A; Supplemental Fig. S2A; Gasperini et al., 2015).
The ninja leaf phenotype is not affected by decreased JA biosynthesis. A, Col-0, ninja-1, ninja-2, aos, ninja-1 aos, and ninja-2 aos plants grown in soil for 25 d. Side views of the seventh leaf (L7) are presented below the photographs. Bar = 1 cm. B, Area of leaves 1/2, 3, and 4 of Col-0, ninja-1, ninja-2, aos, ninja-1 aos, and ninja-2 aos plants grown in soil for 25 d (n = 3 biological replicates with approximately eight plants per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. Significant differences (P < 0.05) relative to Col-0 (a), ninja-1 (b), and aos (c) are indicated with lowercase letters.
In conclusion, we demonstrate that the propeller-like rosette and the not-enlarged and dome-shaped leaf phenotypes of ninja plants are not affected by a decreased JA response or biosynthesis.
ninja Leaves Have a Convex Primary Cell Cycle Arrest Front
In the Arabidopsis Landsberg erecta Δppd mutant, the shape of the primary cell cycle arrest front is hypothesized to be convex rather than straight, leading to excessive growth in the central region of the leaf contributing to the Δppd dome-shaped leaf phenotype (Karidas et al., 2015). Because PPD2 interacts with NINJA and ninja mutants also have a dome-shaped leaf phenotype, the underlying mechanism might be similar. To investigate alterations in the shape of the primary cell cycle arrest front in ami-ppd and ninja-2 mutants, we made use of the pCYCB1;1-DB::GUS reporter line, a marker for actively dividing cells (Andriankaja et al., 2012). pCYCB1;1-DB::GUS and homozygous ninja-2 pCYCB1;1-DB::GUS and ami-ppd pCYCB1;1-DB::GUS double transgenic plants were grown in soil for 21 d, and young developing leaves of similar size were used for GUS staining. The shape of the primary arrest front appeared to be more convex in ami-ppd pCYCB1;1-DB::GUS and ninja-2 pCYCB1;1-DB::GUS leaves compared with the pCYCB1;1-DB::GUS control (Fig. 4A). The extent of convexity was quantified by measuring the perpendicular distance between the highest point of GUS signal in the center of the leaf and a horizontal line connecting the GUS signal at the leaf margins. We found that the distance between the position of the cell cycle arrest front in the central part of the leaf compared with the margins was increased significantly by 32% in ami-ppd pCYCB1;1-DB::GUS and 125% in ninja-2 pCYCB1;1-DB::GUS compared with the pCYCB1;1-DB::GUS control (Fig. 4B).
The ninja mutants have an increased expression of CYCD3;2 and CYCD3;3, and CYCD3;2-OE plants have dome-shaped leaves. A and B, The primary cell cycle arrest front and the perpendicular distance between the highest point of the GUS signal in the center of the leaf and a horizontal line connecting the GUS signal at the leaf margins of leaf 9 or 10 of 21-d-old soil-grown Col-0, ami-ppd, and ninja-2 plants expressing the pCYCB1;1-DB::GUS construct (n = 3 biological replicates with ∼20 leaves per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. Significant differences (P < 0.05) relative to pCYCB1;1-DB::GUS are indicated with lowercase letter a. C and D, Col-0, ninja-1, and ninja-2 plants were grown in vitro, and the first leaf pair (L1/2) was harvested after 11, 13, and 15 d for RNA extraction and qRT-PCR analysis to verify the expression of CYCD3;2 and CYCD3;3 (n = 3 biological replicates with approximately five leaves per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. Significant differences (P < 0.05) between Col-0 and ninja-1 (a) or ninja-2 (b) are indicated with lowercase letters. E, Col-0, CYCD3;2-OE#1-#3, and CYCD3;3-OE#1-#2 plants grown in soil for 25 d. Fold change (FC) values of transgene expression compared with the wild type are also provided. Side views of the seventh leaf (L7) are presented below the photographs. Bar = 1 cm. F, Area of individual leaves of CYCD3;2-OE plants grown in soil for 25 d (n = 3 biological replicates with approximately eight plants per replicate), analyzed using mixed models in the SAS Enterprise Guide. Statistically significant differences (P < 0.05) relative to the wild type are marked in boldface italic type.
To summarize, the ninja-2 mutant presents a convex primary cell cycle arrest front, as also observed in ami-ppd plants.
High Ectopic Expression of CYCD3;2 Results in Plants with Dome-Shaped Leaves
Previously, we have shown that the expression of the cell cycle-related genes CYCD3;2 and CYCD3;3 is up-regulated in the ami-ppd mutant (Gonzalez et al., 2015). To verify if the expression of these cell cycle genes is also changed in the ninja mutants, ninja-1 and ninja-2 plants were grown in vitro and the first leaf pair (L1/2) was harvested after 11, 13, and 15 d for quantitative real time (qRT)-PCR analysis. For all time points, we observed increased expression of CYCD3;2 and CYCD3;3 in ninja plants compared with the wild type (Fig. 4, C and D), although for CYCD3;2, the difference in expression was only statistically significant at day 11 (Fig. 4C).
To study the potential involvement of CYCD3;2 and CYCD3;3 to control leaf flatness, we generated independent homozygous CYCD3;2 and CYCD3;3 overexpression (OE) lines in which the expression of the transgene was driven by the constitutive cauliflower mosaic virus 35S promoter. For CYCD3;2, we obtained only lines with a relatively low transgene overexpression level compared with the wild type (fold change, 5.94–7.25; Fig. 4E). Interestingly, the lines with the highest transgene overexpression (#2 and #3; Fig. 4E) produced propeller-like rosettes with narrow dome-shaped leaves, whereas the leaf shape phenotype was less pronounced in the line with a relatively lower CYCD3;2 overexpression level (#1; Fig. 4E). Upon CYCD3;3 overexpression, plants did not show an effect on leaf curvature, but growth was reduced in the transgenic line with the highest level of overexpression (#2; Fig. 4E).
To study leaf size in the CYCD3;2-OE lines, individual leaf areas of 25-d-old, soil-grown plants were measured, and the results are summarized in a heat map as the average percentage difference compared with the wild type (Fig. 4F). Although the CYCD3;2-OE#1 mutant did not show alterations in leaf size, the CYCD3;2-OE#2-#3 mutants had significantly decreased areas of the younger leaves (L7 onward), whereas the area of the older leaves was not changed significantly (Fig. 4F).
In summary, we demonstrate that the expression of CYCD3;2 and CYCD3;3 was increased in ninja plants and that transgenic lines overexpressing CYCD3;2 produced propeller-like rosettes with dome-shaped leaves. In the CYCD3;2-OE line with the lowest overexpression level, the rosette and leaf shape phenotype of ppd and ninja mutants was only poorly recapitulated, whereas the increase in CYCD3;2 expression was more pronounced than in the ninja mutants. We hypothesize that the expression of CYCD3;2 in ppd and ninja mutants may be extremely high in specific cells but cannot be identified as such because entire leaves were harvested for the expression analysis. Despite the leaf curvature phenotype, the leaf areas of CYCD3;2-OE plants were not increased compared with those of the wild type, which is in contrast to the leaf size phenotype of ppd mutants but in line with that of ninja plants.
CYCD3;2-OE Plants Have an Increased Number of Pavement Cells, But Their Meristemoid Asymmetric Division Is Not Changed Drastically
ppd loss-of-function mutants have an increased number of pavement and guard cells (White, 2006; Gonzalez et al., 2015). In addition, asymmetric divisions of meristemoids are increased in the ami-ppd mutant compared with the wild type (Gonzalez et al., 2015). If CYCD3;2 is the downstream target gene of the PPD2 signaling pathway responsible for the dome-shaped leaf phenotype, we hypothesize that similar cellular alterations might be present in CYCD3;2-OE plants compared with ppd loss-of-function mutants. To verify this, CYCD3;2-OE plants with a visibly mild (#1) and a strong (#2) leaf curvature phenotype were grown in soil for 25 d, and a cellular analysis of the lower (abaxial) epidermis of L1/2 was performed (Fig. 5). Whereas the area of L1/2 in the CYCD3;2-OE lines was not changed significantly compared with the wild type (Fig. 4F), they showed increased numbers of guard cells (#1 10%, #2 22%), pavement cells (#1 20%, #2 60%), and, consequently, total epidermal cells (#1 17%, #2 46%; Fig. 5A). Accordingly, the average pavement cell area was lower than that of the wild type (#1 −10%, #2 −25%; Fig. 5A). A cell size distribution analysis, in which cells are categorized based on their size and the relative frequency of cells within each category is determined, showed that the abaxial epidermis of the CYCD3;2-OE lines contained more small cells (less than 8,000 μm2), whereas cells with a minimal area of greater than 8,000 μm2 were less abundant than in the wild type (Fig. 5B). The stomatal index (i.e. the number of stomata per total epidermal cell number) was decreased in the CYCD3;2-OE lines (#1 −6%, #2 −16%; Fig. 5A). Because the decreased stomatal index and increased relative frequency of small cells in CYCD3;2-OE leaves might result from increased meristemoid asymmetric divisions, individual meristemoids in L1/2 of CYCD3;2-OE#2 plants were followed for three successive days (D12–D14) using leaf imprints, and the number of asymmetric divisions (Fig. 5C, top), guard mother cells, and stomata formation events were scored. In contrast to the increased meristemoid asymmetric division observed in the ami-ppd mutant compared with the wild type (Gonzalez et al., 2015), we did not detect increased meristemoid asymmetric divisions in the CYCD3;2-OE#2 plants (D12 and D13, #2 24% versus Col-0 42%; Fig. 5C, middle). Also, among meristemoids that divided asymmetrically, the rate of recurrent asymmetric division on the following day was similar in CYCD3;2-OE#2 and wild-type plants (D13 and D14, #2 17% versus Col-0 16%; Fig. 5C, bottom).
CYCD3;2-OE leaves have a higher frequency of small cells but their meristemoid asymmetric divisions are not changed drastically. A, Average guard cell number (GCN), pavement cell number (PCN), total cell number (TCN), pavement cell area (PCA), and stomatal index (SI) in the lower (abaxial) epidermis of the first leaf pair (L1/2) of CYCD3;2-OE#1 and CYCD3;2-OE#2 mutants relative to Col-0 (n = 3 biological replicates with four leaves per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. Significant differences (P < 0.05) relative to Col-0 (a) and CYCD3;2-OE#1 (b) are indicated with lowercase letters. B, For the pavement cell size distribution, cells are divided into bins of 4,000 μm2 according to their area. C, Individual meristemoids in the abaxial epidermis of L1/2 of Col-0 and CYCD3;2-OE#2 mutants were followed for three successive days (D12–D14), and asymmetric division, guard mother cell, and stoma formation events were scored. Arrows, To score the recurrent asymmetric events, only meristemoids that divided asymmetrically in the D12-D13 transition were taken into account.
In conclusion, CYCD3;2-OE leaves show increased guard and pavement cell numbers and decreased average pavement cell areas. Despite the decrease in stomatal index, meristemoid asymmetric divisions were similar, or even decreased slightly, in CYCD3;2-OE compared with the wild type.
ppd2, ninja, and CYCD3;2-OE Plants Have Dome-Shaped Leaves, Albeit To a Different Extent
To compare the leaf curvature of ppd2, ninja-1, and CYCD3;2-OE plants, they were grown in soil for 25 d, the seventh leaf (L7) was harvested, and the leaf area, length, width, and perimeter were measured before (projected) and after (real) leaves were cut to flatten them (Fig. 6A; Supplemental Fig. S3). The width and perimeter of the leaf lamina were reduced in PPD2 signaling mutants compared with the wild type (width, #1 −6%, #2 −6%, ninja-1 −19%, ppd2 −14%; perimeter, #1 −5%, #2 −4%, ninja-1 −10%, ppd2 −16%), being statistically significant for ninja-1 and ppd2 leaves, whereas the leaf length did not differ between mutants and the wild type (Supplemental Fig. S3). As a read out for the narrowness of the leaves, the length-to-width ratio was calculated, and as a proxy for the extent of leaf curvature, we calculated the percentage difference between the projected and real leaf area, length, and width. For all mutant lines, the leaf length-to-width ratio was significantly greater than for the wild type (#1 8%, #2 12%, ninja-1 27%, ppd2 16%; Fig. 6B), indicating that the leaves were narrower and more elliptic. The decreases in projected-to-real leaf area, length, and width (Fig. 6C) were significantly more pronounced in ppd2 (area, −47%; length, −20%; width, −44%) and ninja-1 (area, −41%; length, −12%; width, −41%) mutants compared with the wild type (area, −28%; length, −7%; width, −30%). A similar trend was observed in the CYCD3;2-OE#2 mutant (area, −33%; length, −6%; width, −39%), although it was significant only for the projected-to-real leaf area and width differences (Fig. 6C). Despite the parallel alterations in the leaf length, width, perimeter, length-to-width ratio, and projected-to-real leaf area, length, and width differences compared with the wild type, all findings were least pronounced in the CYCD3;2-OE lines and most pronounced in the ppd2 mutant.
PPD2 signaling mutants have narrow and dome-shaped leaves. A, The leaf area, length, and width were measured before (projected) and after (real) leaf 7 (L7) was cut to flatten. B and C, Leaf 7 length-to-width ratio (B) and projected-to-real leaf area, length, and width (C) of Col-0, CYCD3;2-OE#1, CYCD3;2-OE#2, ninja-1, and ppd2 plants grown in soil for 25 d (n = 3 biological replicates with approximately eight plants per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. Significant differences (P < 0.05) relative to Col-0 (a), CYCD3;2-OE#1 (b), CYCD3;2-OE#2 (c), and ninja-1 (d) are indicated with lowercase letters.
To summarize, ppd2, ninja-1, and CYCD3;2-OE plants have an increased leaf length-to-width ratio, reflecting the narrower leaves and decreased projected-to-real leaf area, length, and width compared with the wild type, which represents the dome-shaped phenotype.
Inactivation of CYCD3;1 and CYCD3;2 Partially Restores the ppd Leaf Phenotype
The leaf curvature and cellular phenotype in CYCD3;2-OE lines together with the increased expression of CYCD3;2 in ami-ppd (Gonzalez et al., 2015) and ninja plants suggest that elevated levels of CYCD3;2 cause the dome-shaped leaf phenotype in these mutants. To test this hypothesis, homozygous ppd2 cycd3;2 double mutants were generated and grown in soil with the corresponding single mutants to measure the areas of L1 to L4 and the curvature of L7 after 25 d (Supplemental Fig. S4). At the rosette level, the ppd2 cycd3;2 double mutant showed a clear propeller-like phenotype (Supplemental Fig. S4A). In addition, the narrow and dome-shaped leaf phenotype was not reversed in ppd2 cycd3;2 compared with ppd2, which was reflected by a similar length-to-width ratio and projected-to-real leaf area, length, and width (Supplemental Fig. S4, B and C). Finally, the ppd2 leaf size increase (Wang et al., 2016) was retained in ppd2 cycd3;2 plants (Supplemental Fig. S4D). Arabidopsis has three D3-type cyclins, which were considered previously to act redundantly (Dewitte et al., 2007). In contrast to CYCD3;2-OE plants, CYCD3;3-OE plants did not display curved leaves (Fig. 4E). On the other hand, constitutive overexpression of CYCD3;1 has been shown to lead to dramatic changes in Arabidopsis leaf morphology, including extensive curling and an increased number of small cells (Riou-Khamlichi et al., 1999; Dewitte et al., 2003). Although the expression of CYCD3;1 was not changed in L1/2 of 13-d-old ami-ppd plants (Gonzalez et al., 2015) and in L1/2 of 11-, 13-, and 15-d-old ninja mutants (Supplemental Fig. S5), CYCD3;1 was identified previously as a putative target of PPD2 (Gonzalez et al., 2015). Therefore, we transformed the cycd3;1 cycd3;2 double mutant with the ami-ppd construct (Gonzalez et al., 2015). We quantified the expression of PPD2 in L1/2 of 14-d-old in vitro-grown plants to ensure that the construct had a similar functionality in the cycd3;1 cycd3;2 double mutant to that in a wild-type background, and two independent single-locus insertion cycd3;1 cycd3;2 ami-ppd mutants (#1-#2) with a PPD2 down-regulation similar to that of the ami-ppd line were selected (Supplemental Fig. S6). Of these lines, the area of L1 to L4 and the curvature of L7 were measured and compared with those of the wild type, the ami-ppd line, and the cycd3;1 cycd3;2 double mutant. The propeller-like appearance of the rosette (Fig. 7A) and the narrow-leaf phenotype, reflected by the length-to-width ratio (Fig. 7B), were less pronounced in the cycd3;1 cycd3;2 ami-ppd mutants compared with the ami-ppd mutant (#1 11%, #2 13% versus ami-ppd 18%), although not completely restored to those of the wild type. The projected-to-real leaf area (#1 −29%, #2 −25% versus ami-ppd −47%), length (#1 −12%, #2 −14% versus ami-ppd −22%), and width (#1 −25%, #2 −26% versus ami-ppd −43%) differences were significantly higher in the cycd3;1 cycd3;2 ami-ppd mutants compared with the ami-ppd mutant (Fig. 7C), yet they were still reduced compared with the cycd3;1 cycd3;2 plants (area, −10%; length, −6%; width, −7%). Interestingly, although the areas of L1 to L4 were reduced in cycd3;1 cycd3;2 ami-ppd compared with ami-ppd (Fig. 7D), they were still increased significantly compared with those of the wild type.
Inactivation of CYCD3;1 and CYCD3;2 decreases the extent of ami-ppd leaf curvature. A, Col-0, ami-ppd, cycd3;1 cycd3;2, cycd3;1 cycd3;2 ami-ppd#1, and cycd3;1 cycd3;2 ami-ppd#2 plants grown in soil for 25 d. Side views of the seventh leaf (L7) are presented below the photographs. Bar = 1 cm. B and C, Leaf 7 length-to-width ratio (B) and projected-to-real leaf area, length, and width (C) of Col-0, ami-ppd, cycd3;1 cycd3;2, cycd3;1 cycd3;2 ami-ppd#1, and cycd3;1 cycd3;2 ami-ppd#2 plants grown in soil for 25 d (n = 3 biological replicates with approximately eight plants per replicate). Error bars represent se. Statistical significance was evaluated by ANOVA followed by Tukey’s posthoc analysis. D, Area of leaves 1/2, 3, and 4 of Col-0, ami-ppd, cycd3;1 cycd3;2, cycd3;1 cycd3;2 ami-ppd#1, and cycd3;1 cycd3;2 ami-ppd#2 plants grown in soil for 25 d relative to Col-0 (n = 3 biological replicates with approximately eight plants per replicate), analyzed using mixed models in the SAS Enterprise Guide. In C and D, statistically significant differences (P < 0.05) relative to Col-0 (a), ami-ppd (b), cycd3;1 cycd3;2 (c), and cycd3;1 cycd3;2 ami-ppd#1 (d) are indicated with lowercase letters.
In conclusion, Arabidopsis CYCD3 proteins play a crucial role in limiting the leaf curvature observed in ppd loss-of-function mutants.
DISCUSSION
The Primary Cell Cycle Arrest Front Is Convex in ninja and ppd Mutants
NINJA is a transcriptional repressor with a pivotal role in JA signaling (Pauwels et al., 2010; Acosta et al., 2013; Gasperini et al., 2015). In the absence of JA, NINJA forms a complex with JAZ, MYC2, and TPL to prevent the expression of downstream JA-responsive genes (Thines et al., 2007; Pauwels et al., 2010; Bai et al., 2011; Li et al., 2017). Upon JA perception by the SCFCOI1 ubiquitin ligase complex, JAZ proteins are degraded, the MYC2 transcription factor is released, and the expression of early JA-responsive genes is induced (Thines et al., 2007; Pauwels et al., 2010). Although the role of NINJA with regard to JA signaling has been studied intensively in both Arabidopsis and cotton (Gossypium hirsutum; Acosta et al., 2013; Gasperini et al., 2015; Wang et al., 2017), JA-independent roles of NINJA are largely unknown. NINJA has been identified as a putative interaction partner of PPD2 in a tandem affinity purification experiment performed in an Arabidopsis cell suspension culture (Gonzalez et al., 2015). In this study, we confirmed this interaction using independent in vitro, Y2H, and bimolecular fluorescence complementation experiments and showed that PPD2 interacts with the C domain of NINJA, also known to interact with other members of the TIFY protein family (Vanholme et al., 2007; Pauwels et al., 2010; Bai et al., 2011; Zhang et al., 2012). In the future, however, it will be important to verify the PPD2-NINJA interaction in planta in a native context. In this study, we also showed that ninja loss-of-function mutants display propeller-like rosettes and a convex-shaped primary cell cycle arrest front that putatively leads to the formation of dome-shaped leaves. Because the expression of PPD genes is not affected upon JA treatment (Pauwels et al., 2010) and the phenotypes of PPD mutants are not related to JA signaling (White, 2006; Gonzalez et al., 2015), NINJA appears to play a pivotal role in limiting leaf curvature in addition to its function in mediating the JA-transcriptional response (Pauwels et al., 2010; Acosta et al., 2013; Gasperini et al., 2015), probably through its interaction with PPD2. A premature stop codon in NINJA, accompanied by a down-regulation in NINJA transcripts, prevents the PPD2-NINJA interaction and, therefore, the recruitment of TPL to the PPD2 repressor complex (Pauwels et al., 2010), causing the increased expression of downstream target genes of the protein complex (Fig. 8). The interaction between PPD and NINJA proteins was found not only in Arabidopsis (Pauwels et al., 2010) but also in Medicago truncatula (Ge et al., 2016; Goossens et al., 2016), suggesting that the PPD-NINJA complex might control the shape of the primary arrest front and, consequently, leaf flatness across eudicot plant species. This control of leaf shape likely occurs through the repression of CYCD3 genes, but whether the PPD2-NINJA repressor complex binds directly or indirectly to the CYCD3 loci is a subject for future research.
Schematic overview of the involvement of the PPD complex in leaf growth and shape control. In wild-type leaves, the PPD2 protein interacts with the adaptor proteins KIX8/9 and NINJA to recruit TPL, generating a transcriptional repressor complex. The complex inhibits the expression of downstream target genes, including CYCD3s, which results in a relatively straight primary cell cycle arrest front and a flat-leaf phenotype. Upon SAP-mediated proteosomal degradation or down-regulation of members of the repressor complex, the complex is inactive. This results in the overexpression of the CYCD3 cell cycle genes, the formation of a convex primary cell cycle arrest front, and a dome-shaped leaf phenotype.
CYCD3 Family Members Have a Distinct Contribution to the Control of Leaf Flatness
The CYCD3 subfamily of D-type cyclins consists of three members: CYCD3;1, CYCD3;2, and CYCD3;3. CYCD3s are considered primarily to be redundant, and their roles have been studied mainly using the cycd3;1 cycd3;2 cycd3;3 triple mutant (Dewitte et al., 2003, 2007). For some aspects of plant organ development, this redundancy is clearly demonstrated. For instance, only triple cycd3 mutants lack a cytokinin response, being unable to produce shoots in response to high concentrations of this plant hormone (Dewitte et al., 2007). Nevertheless, the spatial and temporal expression patterns of the CYCD3s are only partially overlapping in both Arabidopsis (Dewitte et al., 2007) and tomato (Solanum lycopersicum; Kvarnheden et al., 2000). This implies that plant tissues might have different requirements for specific CYCD3 expression and that CYCD3s might have very specific functions during plant development (Riou-Khamlichi et al., 1999, 2000; Kvarnheden et al., 2000; Dewitte et al., 2003, 2007), although this remains to be fully explored. The expression of CYCD3;2 and CYCD3;3 is enhanced in the ami-ppd mutants and the double kix8 kix9 mutant, which has a decreased expression of two PPD2 interaction partners (Gonzalez et al., 2015), and, as shown here, in ninja mutants (Fig. 8). Despite the fact that CYCD3;2 and CYCD3;3 have a similar expression pattern in the leaf (Dewitte et al., 2007), ectopic expression of CYCD3;2 but not of CYCD3;3 resulted in propeller-like rosettes with dome-shaped leaves. Nevertheless, the dome-shaped leaf phenotype in ppd mutants could only be partially restored upon loss of function of CYCD3;2 and CYCD3;1, although transcript levels of CYCD3;1 were not changed in ami-ppd (Gonzalez et al., 2015) and ninja mutants at the selected time points. This indicates that there also might be a certain degree of redundancy among the CYCD3 proteins to control leaf shape. In the future, it will be interesting to uncover the individual contributions of the CYCD3 family members to the control of leaf flatness.
CYCD3;2 as a Central Node in the Control of the Cell Cycle, the Stomatal Lineage, and Leaf Shape
Arabidopsis leaves typically develop into flat structures to maximize the photosynthetic surface (Horiguchi et al., 2006; Rolland-Lagan et al., 2014). Because forming a flat structure requires uniform growth throughout the lamina, leaf flatness is strictly controlled by several regulatory pathways, including the PPD module (Nath et al., 2003; Palatnik et al., 2003; Horiguchi et al., 2006; Karidas et al., 2015). Based on the shape of the primary cell cycle arrest front in ami-ppd and ninja mutants and the increased expression of CYCD3;2 in these mutants, we hypothesized that the curvature phenotype in ppd and ninja leaves is caused by the elevated expression of central to margin cyclins. Nonetheless, CYCD3;2-OE lines expressing CYCD3;2 controlled by the cauliflower mosaic virus 35S promoter displayed dome-shaped leaves. Constitutive overexpression of CYCD3;2, however, does not imply that CYCLIN-DEPENDENT KINASE (CDK) interaction partners, which are essential for the functionality of cyclins, are present. In addition, cyclins are highly prone to posttranslational regulations, further increasing the complexity by which cyclins might be regulated in individual cells. Previously, it has been shown that CYCD3;2 expression is activated in meristemoids and declined gradually throughout the stomatal lineage, becoming undetectable in mature guard cells (Adrian et al., 2015). Correspondingly, CYCD3;2 expression has been shown to be activated by SPEECHLESS (SPCH; Lau et al., 2014), a transcription factor that positively regulates meristemoid initiation and recurrent asymmetric divisions (Bergmann and Sack, 2007; Lau and Bergmann, 2012; Pillitteri and Torii, 2012; Lau et al., 2014). The opposite regulation of CYCD3;2 by PPD2 and SPCH and the gradual increase in guard cell number upon CYCD3;2 overexpression suggest that a strict spatiotemporal control of CYCD3;2 expression is not only essential to control leaf shape but also throughout the stomatal lineage. This is also concurrent with previous data showing that ectopic overexpression of CYCLIN DEPENDENT KINASE A;1 (CDKA;1) and CYCD3;2 from a FAMA promoter induces extra guard cell divisions, resulting in abnormal stomata with three to four guard cells (Yang et al., 2014). These data indicate that a timely dampening of CDKA;1-CYCD3 activity is essential to block guard cell division (Yang et al., 2014). CYCDs promote the progression of the cell cycle during the G1 phase and the G1-to-S transition (Menges et al., 2006; Magyar et al., 2012; Zhao et al., 2012). In complex with CDKs, they bind and phosphorylate RETINOBLASTOMA-RELATED (RBR), causing its degradation (Huntley et al., 1998; Nakagami et al., 1999). Interestingly, RBR has been shown previously to regulate the asymmetric divisions of meristemoids (Weimer et al., 2012), and plants with reduced RBR transcript levels also produce propeller-like rosettes with slightly dome-shaped leaves (Dorca-Fornell et al., 2013). All these findings indicate that a precise regulation of the pattern of cell cycle arrest is a key aspect in controlling leaf shape (Nath et al., 2003; Karidas et al., 2015) and underline the importance of a proper spatiotemporal regulation of CYCD3 expression therein.
ninja and CYCD3;2-OE Plants Lack the Increased Leaf Size Observed in ppd Mutants
In this study, we demonstrated that, even though ninja and CYCD3;2-OE plants show the ppd dome-shaped leaf phenotype and increased guard and pavement cell numbers, they do not produce larger leaves. The PPD proteins have been suggested to act on both the primary (Karidas et al., 2015) and secondary (White, 2006; Gonzalez et al., 2015) mitotic arrest fronts in leaves, referring to the division of pavement and meristemoid cells, respectively. Consequently, the increased leaf area observed in ppd mutants results, at least partially, from a drastic increase in meristemoid asymmetric divisions (White, 2006; Gonzalez et al., 2015). Interestingly, however, meristemoid asymmetric divisions were not increased in the abaxial epidermis of CYCD3;2-OE plants. This suggests that the increase in pavement cell number and the decrease in stomatal index in CYCD3;2-OE leaves result solely from an altered primary cell cycle arrest phase. The dissimilarity between ppd and CYCD3;2-OE plants regarding meristemoid asymmetric divisions and leaf size implies that the role of PPD2 in leaf size and shape control could be partially uncoupled and, to a certain extent, might be regulated differently. Consequently, the leaf shape and size phenotype in ppd mutants could be the cumulative output of distinct downstream regulatory pathways. In addition to the dome-shaped leaf phenotype, PPD signaling mutants produce rosettes with a propeller-like appearance. However, we hypothesize that the rosette phenotype is not necessarily related to the dome-shaped leaf phenotype but might be caused by alterations in vasculature development, a process in which PPDs have been proposed to be involved (White, 2006). The contribution of the PPD module to vasculature development could be governed by CYCD3 genes and/or by other unidentified target genes of the complex. Mediating the expression of additional downstream target genes besides CYCD3s might be a way to control distinct pathways to regulate a variety of growth and developmental processes.
The PPD Module Controls Leaf Flatness and Is Highly Conserved in Eudicots
Several studies in different plant species have consistently shown the importance of the PPD signaling pathway in organ size and shape control, including seeds, seed pods, and leaves (Gonzalez et al., 2015; Ge et al., 2016; Wang et al., 2016; Naito et al., 2017; Yordanov et al., 2017; Kanazashi et al., 2018). The mtbs1-1 mutant harbors a point mutation in BIG SEEDS1 (BS1), encoding a putative PPD ortholog in M. truncatula (Ge et al., 2016). The mtbs1-1 mutant and VmPPD mutant in Vigna mungo (Naito et al., 2017) produce enlarged seeds and leaves with similarities to those of Arabidopsis ppd mutants (White, 2006; Gonzalez et al., 2015). Down-regulation of the putative PPD orthologs in soybean (Glycine max) using an artificial microRNA significantly increased soybean seed size, weight, and amino acid content (Ge et al., 2016; Naito et al., 2017). More recently, GmPPD1 and GmPPD2 double mutants have been generated in soybean using the CRISPR/Cas9 technology; they also displayed dome-shaped trifoliate leaves and twisted pods containing fewer but enlarged seeds compared with wild-type plants (Kanazashi et al., 2018). The PPD complex is degraded in a proteasome-dependent manner by the Arabidopsis SCF complex containing the F-box protein STERILE APETALA (SAP; Fig. 8; Wang et al., 2016; Li et al., 2018). Overexpression of SAP in Arabidopsis (Wang et al., 2016) and of the poplar SAP ortholog BIG LEAF (Yordanov et al., 2017) resulted in plants with enlarged leaves and uneven lamina growth. Despite the common importance of the PPD2/KIX8/KIX9/SAP complex for organ size and shape determination in eudicots, it is lost in Poaceae (grasses; Gonzalez et al., 2015; Nelissen et al., 2016; Wang et al., 2016). In agreement, self-renewing meristemoid-like cells and a secondary cell cycle arrest front moving through the leaf blade, in which PPD2 is involved, are fundamental for eudicot leaf development but absent in grasses (Liu et al., 2009). By contrast, NINJA is present in all land plants and is essential in JA signaling (Sheard et al., 2010; Thaler et al., 2012) as well as in controlling leaf flatness through regulation of the shape of the primary cell cycle arrest front. The observation that NINJA orthologs are present in grasses such as maize (Zea mays) and rice (Oryza sativa) further strengthens their broader role in plant development.
MATERIALS AND METHODS
In Vitro Pull-Down Assay
The coding sequence of PPD2 was inserted into the BamHI and EcoRI sites of the pGEX-4T-1 vector to obtain the GST-PPD2 construct. The coding sequence of NINJA was inserted into the EcoRI and SalI sites of the pET-28a (+) vector to construct the His-NINJA plasmid. The primer sequences are listed in Supplemental Table S1. To test the interaction of NINJA with PPD2, bacterial lysates containing ∼30 μg of GST-PPD2 fusion proteins were combined with lysates containing ∼30 μg of His-NINJA fusion proteins. A total of 20 μL of glutathione-Sepharose (GE Healthcare) was added into each combination and shaken gently at 4°C for 1 h. TGH buffer (50 mm HEPES, pH 7.5, 1.5 mm MgCl2, 150 mm NaCl, 1 mm EGTA, 10% [v/v] glycerol, 1% [v/v] Triton X-100, 1 mm PMSF, and 1× Complete protease inhibitor cocktail) was used to wash the beads. The isolated proteins were separated further by SDS-PAGE and examined by immunoblot analysis using anti-GST (Abmart; 1:5,000) and anti-His (Abmart; 1:2,000) antibodies. Signals were detected using the eECL Western Blot Kit (CWBiotech; CW0049), and images were scanned using Tanon-4500 according to the manufacturer’s instructions.
Bimolecular Fluorescence Complementation Assay
nYFP was amplified from the plasmid pSY736, fused to NINJA, and inserted into the pDONR221 vector (Invitrogen). cYFP was amplified from the plasmid pSY735, fused to PPD2, and inserted into the pDONR221 vector (Invitrogen). The primer sequences are listed in Supplemental Table S1. nYFP-NINJA and cYFP-PPD2 constructs were cloned into the Gateway binary vector pGWB414 by LR reactions (Invitrogen) and transformed into Agrobacterium tumefaciens (strain C58C1, pMP90). A. tumefaciens strains containing nYFP-NINJA, cYFP-PPD2, and cYFP plasmids were resuspended in buffer (10 mm MES, pH 5.6, 150 μm acetosyringone, and 10 mm MgCl2), mixed, and coinfiltrated into Nicotiana benthamiana leaves. After infiltration, plants were grown for 2 d and fluorescence was detected using confocal microscopy (Zeiss LSM 710).
Y2H Experiments
A deletion series of NINJA and PPD2 were cloned without stop codon into Gateway-compatible entry clones. The entry vectors were used for LR reactions (Invitrogen) with pGADT7 and pGBKT7 (or pGBT9) Y2H vectors, generating bait and prey constructs. Cotransformation of the Saccharomyces cerevisiae PJ69-4A yeast strain using different bait-prey plasmid combinations was performed according to Cuéllar Pérez et al. (2014). Transformed yeast colonies were selected on SD medium lacking Leu and Trp. To verify a possible interaction, several independent yeast colonies were grown on selective medium lacking Leu, Trp, and His for 3 to 4 d at 30°C. Cotransformation with the empty pGADT7 and pGBKT7 (or pGBT9) vectors was used as a negative control.
Plant Material and Growth Conditions
The Arabidopsis (Arabidopsis thaliana) ami-ppd line is described by Gonzalez et al. (2015). The cycd3;2 mutant (GK396C10) was obtained from the GABI-KAT T-DNA collection (Sessions et al., 2002) and genotyped by PCR to verify homozygosity. The ninja-1, ninja-2, ninja-1 aos, ninja-2 aos, and ninja-1 coi1-1 lines were kindly provided by Debora Gasperini (Leibniz Institute of Plant Biochemistry). ninja-1 and ninja-2 were described by Acosta et al. (2013) and have a mutation in Chr4:14265008 (C to T) and Chr4:14264791 (G to A), respectively, resulting in a stop codon Q225* in ninja-1 and an affected first intron/exon split in ninja-2. The cycd3;1 cycd3;2 double mutant line was kindly provided by Walter Dewitte and Jim Murray (Cardiff University). The coding sequences of CYCD3;2 and CYCD3;3 were PCR amplified from Col-0 cDNA and introduced into pDONR221. The 35S::CYCD3;2 and 35S::CYCD3;3 constructs were cloned by Multisite Gateway cloning (Invitrogen) into pK7m24GW;3_FAST and pFAST-G02, respectively (Karimi et al., 2002). The expression vectors were used for the transformation of Arabidopsis Col-0 plants with the floral dip method using A. tumefaciens (strain C58C1, pMP90; Clough and Bent, 1998). Transformants were selected on medium supplied with the corresponding antibiotic, and homozygous single-locus insertion plants were used in the experiments. For the genetic interaction studies, ami-ppd and cycd3;1 cycd3;2 ami-ppd transgenic lines with a single-locus insertion and with a similar down-regulation of PPD2 transcripts were obtained and selected for further analysis (Supplemental Fig. S6). Experiments were performed with segregating wild-type and transgenic seeds harvested from plants that were grown in parallel. For molecular experiments, plants were surface sterilized and grown in vitro for 14 d on solid (9 g L−1 agar; Sigma) one-half-strength Murashige and Skoog medium (Murashige and Skoog, 1962) supplemented with 1% (w/v) Suc. For phenotyping experiments, plants were grown in soil for 21 or 25 d. In vitro and in soil, plants were stratified for 3 d at 4°C and grown at 21°C under a 16-h-day/8-h-night regime.
Histochemical GUS Analysis
Histochemical GUS staining was performed on young developing leaves of pCYCB1;1-DB::GUS (leaf 10), ami-ppd pCYCB1;1-DB::GUS (leaf 10), and ninja-2 pCYCB1;1-DB::GUS (leaf 9) plants grown in soil for 21 d. The plant material was incubated in the dark in a staining buffer containing 1 mm 5-bromo-4-chloro-3-indolyl β-d-glucopyranoside sodium salt, 0.5% (v/v) Triton X-100, 1 mm EDTA, pH 8, 0.5 mm potassium ferricyanide [K3Fe(CN)6], 0.5 mm potassium ferricyanide [K4Fe(CN)6], and 500 mm sodium phosphate buffer, pH 7. After an overnight incubation at 37°C, leaves were decolored by replacing the staining buffer with 100% ethanol. The material was mounted in lactic acid and analyzed with a binocular microscope (Leica; MZ16). The experiment was performed in three biological replicates with ∼20 leaves each.
RNA Isolation and qRT-PCR Analysis
Total RNA was isolated from the first leaf pair (L1/2) of 14-d-old plants with Trizol (Invitrogen), purified with the RNeasy Plant Mini Kit (Qiagen), and treated with DNase I (Promega). cDNA was obtained with the iScript cDNA Synthesis Kit (Bio-Rad) according to the manufacturer’s instructions. Relative transcript abundance was determined using the Roche LightCycler 480 and the LC480 SYBR Green I Master Kit (Roche Diagnostics). The resulting cycle threshold values were converted into relative expression values using the second derivative maximum method. CDKA and CBP20 were used as reference genes for normalization. All experiments were performed in three biological replicates (approximately five leaves per replicate), each with three technical replicates. The primer sequences are listed in Supplemental Table S1.
Leaf Growth Parameters
Plants were grown in soil for 25 d. For the leaf series analysis, individual leaf areas were measured with the ImageJ software (https://imagej.nih.gov/ij/). For the leaf shape analysis, the seventh leaf (L7) was harvested and photographs were taken before and after cutting the leaves to flatten them. Projected and real leaf area, length, and width and the leaf perimeter were measured using ImageJ. For the cellular analysis, the leaves were cleared with 100% ethanol, mounted in lactic acid on microscope slides, and abaxial epidermal cells (∼300 cells) were drawn for three biological replicates (four leaves each) with a binocular microscope (Leica; DMLB) fitted with a drawing tube and a differential interference contrast objective. Average guard, pavement, and total cell numbers, as well as pavement cell area and stomatal index, were calculated as described previously (Andriankaja et al., 2012). Leaf imprints were made daily from the abaxial surface of the first leaf pair (L1/2) from day 12 to day 14 according to Kagan et al. (1992). Nail polish impressions were made and analyzed by scanning electron microscopy (Hitachi; TM-1000 Tabletop Microscope). The fate of meristemoid cells was analyzed over time by evaluating whether a meristemoid became a guard mother cell or a stoma and whether it divided asymmetrically or remained unchanged.
Statistical Tests
For the leaf series analysis, statistical tests were performed with the mixed models and plm procedure in the SAS Enterprise Guide as described previously (Blomme et al., 2014). In the presence of a significant F test (P < 0.05), pairwise comparisons among the transgenic lines or between the transgenic lines and the segregating wild type were performed using Tukey’s posthoc test. For each leaf, correction for multiple testing was done according to Dunnett. For the statistical analysis of the leaf curvature and for the cellular and expression data, ANOVA was performed in R (version 3.3.2; https://www.r-project.org) using Tukey’s posthoc analysis (P < 0.05). For all statistical tests, the transgenic line was the fixed factor in the model. Biological replicates were included as a random factor.
Accession Numbers
Accession numbers are as follows: AT4G14720 (PPD2), AT4G28910 (NINJA), AT4G34160 (CYCD3;1), AT5G67260 (CYCD3;2), AT3G50070 (CYCD3;3), AT3G24150 (KIX8), and AT4G32295 (KIX9).
Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. The expression of NINJA is decreased in ninja mutants.
Supplemental Figure S2. The ninja leaf phenotype is not affected by a decreased JA response.
Supplemental Figure S3. Leaf length, width, and perimeter in PPD2 signaling mutants.
Supplemental Figure S4. Inactivation of CYCD3;2 does not complement the ppd2 leaf phenotype.
Supplemental Figure S5. The expression of CYCD3;1 is not changed in ninja mutants.
Supplemental Figure S6. Residual PPD2 expression in the ami-ppd and cycd3;1 cycd3;2 ami-ppd mutants.
Supplemental Table S1. Primers used in this study.
Dive Curated Terms
The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:
Acknowledgments
We thank Carina Braeckman and Mieke Van Lijsebettens for assistance in generating the Arabidopsis transgenic lines, Karel Spruyt for imaging assistance, and Annick Bleys for the help in preparing the article. We also thank Debora Gasperini (ninja-1, ninja-2, aos, ninja-1 coi1-1, ninja-1 aos, and ninja-2 aos), Walter Dewitte, and Jim Murray (cycd3;1 cycd3;2) for kindly providing some of the transgenic lines used in this study.
Footnotes
↵1 This work was supported by Ghent University (Bijzonder Onderzoeksfonds Methusalem project no. BOF08/01M00408). A.B. is an FWO predoctoral fellow (no. 131019) and L.P. an FWO postdoctoral fellow (no. 1206414N).
↵2 Current address: School of Life Sciences, Beijing University of Chinese Medicine, Beijing 100029, China.
↵3 These authors contributed equally to the article.
↵6 Current address: Institut National de la Recherche Agronomique, Unité Mixte de Recherche 1332, Biologie du Fruit et Pathologie, CS20032, 33140 Villenave d’Ornon, France.
↵[OPEN] Articles can be viewed without a subscription.
- Received March 19, 2018.
- Accepted June 28, 2018.
- Published July 10, 2018.