Ethylene-Mediated Regulation of A2-Type CYCLINs Modulates Hyponastic Growth in Arabidopsis1[OPEN]

Ethylene and cell proliferation confine the amplitude of Arabidopsis hyponastic leaf movement. Upward leaf movement (hyponastic growth) is frequently observed in response to changing environmental conditions and can be induced by the phytohormone ethylene. Hyponasty results from differential growth (i.e. enhanced cell elongation at the proximal abaxial side of the petiole relative to the adaxial side). Here, we characterize Enhanced Hyponasty-d, an activation-tagged Arabidopsis (Arabidopsis thaliana) line with exaggerated hyponasty. This phenotype is associated with overexpression of the mitotic cyclin CYCLINA2;1 (CYCA2;1), which hints at a role for cell divisions in regulating hyponasty. Indeed, mathematical analysis suggested that the observed changes in abaxial cell elongation rates during ethylene treatment should result in a larger hyponastic amplitude than observed, unless a decrease in cell proliferation rate at the proximal abaxial side of the petiole relative to the adaxial side was implemented. Our model predicts that when this differential proliferation mechanism is disrupted by either ectopic overexpression or mutation of CYCA2;1, the hyponastic growth response becomes exaggerated. This is in accordance with experimental observations on CYCA2;1 overexpression lines and cyca2;1 knockouts. We therefore propose a bipartite mechanism controlling leaf movement: ethylene induces longitudinal cell expansion in the abaxial petiole epidermis to induce hyponasty and simultaneously affects its amplitude by controlling cell proliferation through CYCA2;1. Further corroborating the model, we found that ethylene treatment results in transcriptional down-regulation of A2-type CYCLINs and propose that this, and possibly other regulatory mechanisms affecting CYCA2;1, may contribute to this attenuation of hyponastic growth.

Upward leaf movement (hyponastic growth) is frequently observed in response to changing environmental conditions and can be induced by the phytohormone ethylene. Hyponasty results from differential growth (i.e. enhanced cell elongation at the proximal abaxial side of the petiole relative to the adaxial side). Here, we characterize Enhanced Hyponasty-D, an activation-tagged Arabidopsis (Arabidopsis thaliana) line with exaggerated hyponasty. This phenotype is associated with overexpression of the mitotic cyclin CYCLINA2;1 (CYCA2;1), which hints at a role for cell divisions in regulating hyponasty. Indeed, mathematical analysis suggested that the observed changes in abaxial cell elongation rates during ethylene treatment should result in a larger hyponastic amplitude than observed, unless a decrease in cell proliferation rate at the proximal abaxial side of the petiole relative to the adaxial side was implemented. Our model predicts that when this differential proliferation mechanism is disrupted by either ectopic overexpression or mutation of CYCA2;1, the hyponastic growth response becomes exaggerated. This is in accordance with experimental observations on CYCA2;1 overexpression lines and cyca2;1 knockouts. We therefore propose a bipartite mechanism controlling leaf movement: ethylene induces longitudinal cell expansion in the abaxial petiole epidermis to induce hyponasty and simultaneously affects its amplitude by controlling cell proliferation through CYCA2;1. Further corroborating the model, we found that ethylene treatment results in transcriptional down-regulation of A2-type CYCLINs and propose that this, and possibly other regulatory mechanisms affecting CYCA2;1, may contribute to this attenuation of hyponastic growth.
Plants have acquired mechanisms to adjust growth and secure reproduction under unfavorable environmental conditions. Among the strategies to avoid adverse conditions is upward leaf movement, called hyponastic growth. This leaf reorientation is driven by unequal growth rates between adaxial and abaxial sides of the petiole (Cox et al., 2004;Polko et al., 2012b). Arabidopsis (Arabidopsis thaliana) exhibits hyponasty upon several environmental signals (e.g. submergence, waterlogging, proximity of neighboring vegetation, low red:far-red light ratios, reduced blue light fluence rates, low light intensities, and high temperatures; Millenaar et al., 2005Millenaar et al., , 2009Mullen et al., 2006;Koini et al., 2009;Moreno et al., 2009;Van Zanten et al., 2009;Keuskamp et al., 2010;Keller et al., 2011;Vasseur et al., 2011;De Wit et al., 2012;Rauf et al., 2013;Dornbusch et al., 2014). Hyponasty alleviates the impact of environmental stresses (Van Zanten et al., 2010b). During submergence, it allows reestablishment of gas exchange with the atmosphere (e.g. Cox et al., 2003); at high plant densities, it positions the leaves in better lit layers of the canopy to improve light interception (e.g. De Wit et al., 2012); and at high temperatures, it improves the cooling capacity of the leaves (Crawford et al., 2012;Bridge et al., 2013). The cellular basis of hyponastic growth in Rumex palustris (Cox et al., 2004) and Arabidopsis (Polko et al., 2012b;Rauf et al., 2013) has been characterized. Ethylene causes reorientation of cortical microtubules (CMTs) in the petiole, which leads to longitudinal cell expansion in an approximately 2-mmlong epidermal cell zone at the proximal part of the abaxial side of the organ (Polko et al., 2012b).
The interactions between several hormones (e.g. ethylene, abscisic acid, GAs, and auxin) in controlling hyponasty under various conditions have been studied (Mullen et al., 2006;Benschop et al., 2007;Millenaar et al., 2009;Van Zanten et al., 2009, 2010bPeña-Castro et al., 2011). The volatile phytohormone ethylene is a key component in the complex regulatory network of hyponastic growth. Ethylene is the trigger and a positive regulator of hyponastic growth in submerged and waterlogged Arabidopsis (Millenaar et al., 2005Van Zanten et al., 2010b;Rauf et al., 2013) and a negative regulator of high temperature-induced hyponasty ), but is not involved in low lightinduced hyponastic growth in this species . Abscisic acid antagonizes ethylene-induced hyponasty (Benschop et al., 2007) and is a positive regulator of high temperature-induced hyponastic growth . The growth-promoting GAs positively regulate hyponastic response to all three environmental signals (Peña-Castro et al., 2011), whereas auxins promote low light and high temperature-induced hyponastic growth (Millenaar et al., 2005;Koini et al., 2009;Van Zanten et al., 2009), as well as low red:far-red-and low blue light-induced hyponasty (Moreno et al., 2009;Keller et al., 2011). Finally, brassinosteroids also positively regulate ethylene-induced hyponasty (Polko et al., 2013).
Despite the extensive knowledge on hormonal regulation of hyponasty, little is known about the molecular genetic mechanisms that drive this response. One notable exception is the study by Rauf et al. (2013), who showed that hyponastic growth in Arabidopsis in response to root waterlogging is controlled by the NAC (for No Apical Meristem [NAM], Arabidopsis Transcription Activation Factor) transcription factor SPEEDY HYPONASTIC GROWTH that directly affects expression of the ethylene biosynthesis gene 1-AMINOCYCLOPROPANE-1-CAR-BOXYLIC ACID (ACC) OXIDASE5.
Here, we followed a forward genetic approach to identify unique components that control hyponastic growth in Arabidopsis. From a population of activation-tagged plants (Weigel et al., 2000), we isolated Enhanced Hyponasty-D (EHY-D), which showed exaggerated hyponasty under exogenous ethylene application, low light intensities, and high temperature. We found that ectopic expression of the core cell cycle regulator CYCLINA2;1 (CYCA2;1) caused the exaggerated ethyleneinduced leaf movement of EHY-D. Mathematical analyses indicated that, besides promoting cell expansion, ethylene can also attenuate the amplitude of hyponasty by affecting differential cell proliferation in the petiole of wild-type plants. We suggest that this occurs through ethylene-dependent effects on CYCA2;1 levels, activity, or sensitivity in petioles of wild-type plants.
The ethylene-mediated transcriptional regulation of CYCA2;1 observed here could contribute to this. In EHY-D, however, ethylene-mediated effects on cell proliferation are overruled by ectopic CYCA2;1 overexpression, which consequently results in enhanced hyponasty, in accordance with the predictions of our model. Correspondingly, cyca2;1 knockout lines where ethylene cannot affect CYCA2;1-mediated cell proliferation also exhibited enhanced hyponasty. Our data therefore describe a mechanism by which hyponastic growth is kept within limits, through a bipartite role for ethylene: within the same organ, ethylene initiates hyponastic growth by promoting cell elongation, while simultaneously attenuating the response by regulation of A2-type CYCLIN-mediated cell proliferation.

Isolation and Cloning of EHY-D
To identify novel genetic components that control hyponastic growth, we conducted a forward genetic screen using a population of 35S activation-tagged Columbia (Col) plants (Weigel et al., 2000). A total of 17,500 plants were screened for their hyponastic response under 6 h of ethylene and low light treatment. The screen yielded 18 candidates with aberrant petiole angle (Polko et al., 2012a(Polko et al., , 2013. Among the isolated lines was EHY-D, which showed an initial petiole angle similar to the wild type (20.5 6 1.4 and 22.5 6 1.4, respectively; Supplemental Table S1) and an exaggerated response to ethylene (Fig. 1, A and C) and low light (Fig. 1C). No other apparent visual differences were observed (Fig. 1B). The enhanced hyponasty phenotype was confirmed by quantitative analysis of leaf movement kinetics using a time-lapse digital camera setup (Fig. 1, D-F). In addition, high temperature also resulted in an enhanced response (Fig. 1, C and F), suggesting that a general genetic determinant of hyponastic growth is affected in EHY-D.
Genes causal for observed phenotypes are often flanking or in the direct vicinity of the T-DNA insertion site (Weigel et al., 2000). Therefore, we quantified the relative transcript levels of the genes within a 15-kb region up-and downstream of the T-DNA integration site by quantitative reverse transcription (qRT)-PCR under control conditions and after application of ethylene. Some of the tested genes were mildly up-regulated after 3 h of ethylene treatment compared with control conditions in wild-type Col (Table I). This included ETHYLENE INSENSITIVE3 (EIN3)-BINDING F-BOX PROTEIN2 that was previously shown to be ethylene inducible (Potuschak et al., 2003). In EHY-D, only the two genes directly flanking the T-DNA insertion border (SHN3 and CYCA2;1) were overexpressed compared with wild-type Col, and this was true under both control and ethylene conditions (Table I). This suggests that one of these genes is causal for the exaggerated hyponastic growth phenotype of EHY-D.
Overexpression of CYCA2;1 Mimics the EHY-D Hyponastic Growth Phenotype SHN3 encodes a member of the ERF family. ERFs control many developmental and physiological processes, including several ethylene-mediated responses (Nakano et al., 2006). Using overexpression lines in the Wassilewskija (Aharoni et al., 2004) and Col backgrounds (isolated A, Leaf angle phenotype of the wild type and EHY-D after 10-h ethylene treatment. B, EHY-D and the wild-type rosette phenotype. C, Absolute petiole angles of EHY-D (gray bars) and the wild type (black bars) after 6-h control conditions, ethylene (1.5 mL L 21 ), low light (20 mmol m 22 s 21 ), and high temperature (38˚C) treatment. Significance levels (2-tailed Student's t test; ns, not significant): *,P , 0.05; **,P , 0.01; and ***, P , 0.001. D-F, Hyponastic growth kinetics of EHY-D (gray symbols) compared with the wild type (dashed lines) upon treatment with ethylene (circles; D), low light (squares; E), and high temperature (triangles; F). Angles in D to F resulted from pairwise subtraction (Benschop et al., 2007). Error bars are SEM; n . 10. G, Representation of the EHY-D transfer DNA (T-DNA) insertion site (box) on chromosome 5. Red arrowheads indicate the direction of the 35S transcriptional enhancers. Genes in the vicinity, their Arabidopsis Genome Initiative codes, and annotation are depicted as arrows, pointing in the direction of transcription. Physical distances between the genetic elements are in base pairs. from the collection described in Weiste et al., 2007), we tested if SHN3 overexpression could be responsible for the observed exaggerated hyponasty in EHY-D. However, overexpression of SHN3 did not result in enhanced hyponastic responses in either the Col or Wassilewskija background (Supplemental Fig. S1). This indicates that SHN3 overexpression in EHY-D is not causing its exaggerated hyponastic growth phenotype.
CYCA2;1, the highly up-regulated gene directly flanking the EHY-D locus (Fig. 1G), belongs to a small gene family of G2-to-M cell cycle regulators Vanneste et al., 2011). To test if CYCA2;1 overexpression could explain the EHY-D phenotype, we generated 35S::CYCA2;1 plants. As observed in EHY-D, hyponastic growth was enhanced in four independent CYCA2;1 overexpression lines (Fig. 2, A-C; Supplemental Figs. S2 and S3). The differences in hyponastic growth response between the independent lines was positively correlated with the respective CYCA2;1 expression levels (Supplemental Fig. S3). Moreover, a mutant defective in the conserved and specific A2-type CYCLIN repressor INCREASED LEVEL OF POLYPLOIDY1-2 (ilp1-2), which results in enhanced expression of all A2-type CYCLIN family members , also showed exaggerated hyponastic growth under ethylene exposure (Fig. 2D). This response was comparable with EHY-D and 35S::CYCA2;1 lines. Consistently, the ILP1-D activationtagged line with decreased expression of all four A2-type CYCLINs  showed reduced hyponastic growth (Fig. 2E).
Transcription of other A2-type CYCLIN family members (Vandepoele et al., 2002) and several other cell cycle marker genes  was not distinctly affected in whole petioles of EHY-D (Fig. 2F) in either control or ethylene conditions. Taken together, these data demonstrate that overexpression of CYCA2;1 is sufficient to explain the EHY-D hyponastic growth phenotype.
Surprisingly, when we assayed the requirement of functional CYCA2;1 for hyponastic growth, we found that two independent knockout alleles of cyca2;1 also showed an exaggerated response to ethylene, low light, and high temperature treatment (Fig. 2 Table S1). Because it has been reported that reduced CYCA2;2 expression in erecta loss-of-function mutants can be compensated for by ectopic up-regulation of CYCA2;3 (Pillitteri et al., 2007), we tested for compensatory up-regulation of other A2-type CYCLINs in cyca2;1 mutant petioles by qRT-PCR. However, our qRT-PCR experiments did not reveal ectopic up-regulation of CYCA2;2, CYCA2;3, or CYC2A2;4 in petiole tissues of the cyca2;1-2 mutant (Supplemental Fig. S4). Similar to EHY-D (Fig. 2F), these genes were also not affected in the 35S::CYCA2;1 line (Supplemental Fig. S4). Therefore, compensatory transcriptional up-regulation of other A2-type CYCLINs in the petiole probably cannot explain the enhanced hyponastic growth response of cyca2;1 mutants. However, from these data, we cannot exclude that changes in spatiotemporal expression of other A2-type CYCLINs affect the hyponastic growth response. Therefore, we also analyzed cyca2;2-1, cyca2;3-1, and cyca2;4-1 insertional mutants (Vanneste et al., 2011). In contrast to the cyca2;1 mutants ( Fig. 2, G and H), these single mutants did not show an altered hyponastic growth phenotype in response to ethylene (Supplemental Fig. S5). However, when combined with other mutations in the A2-type CYCLIN family, the exaggerated cyca2;1 hyponastic growth was lost and was in some cases even lower than the wild type (Supplemental Fig. S5), suggesting that misregulation of other CYCA2s could be related to the cyca2;1 mutant phenotype. This is consistent with the results of ILP1-D where all A2-type CYCLINs were transcriptionally down-regulated (Fig. 2E) and demonstrates involvement of other A2-type CYCLINs in the control of hyponastic growth. To address this, we analyzed the hyponastic growth of a CYCA2;2 overexpression line. Similar to the EHY-D and 35S::CYCA2;1 lines, the 35S::CYCA2;2 line also showed enhanced hyponasty in response to ethylene and low light, but not in response to high temperature (Supplemental Fig. S5). In addition, the initiation of the response to low light and high temperatures was delayed, suggesting that, besides CYCA2;1, other A2-type CYCLINs are also biochemically competent in modifying hyponastic growth responses.

Ethylene Suppresses Expression of Mitotic Genes in the Petiole
To determine if CYCA2;1 is specifically involved in hyponasty or modifies a general component in ethylene-mediated growth responses, we assayed ethylene-dependent inhibition of hypocotyl elongation in dark-grown seedlings (Guzman and Ecker, 1990). No differences in elongation were detected in EHY-D, 35S::CYCA2;1, and cyca2;1-2 compared with the wild type ( Fig. 3A) in the presence of increasing concentrations of the ethylene biosynthetic precursor ACC. This indicates that CYCA2;1 levels do not affect ethylene sensitivity of the hypocotyls. Additionally, ethylene release from vegetative rosettes of these lines was similar to that of the wild type (0.65 6 0.13 in EHY-D, 0.64 6 0.09 in 35S::CYCA2;1, and 0.52 6 0.06 in cyca2;1-2 compared with 0.65 6 0.13 nL g FW 21 h 21 in the wild type). These results suggest that CYCA2;1 regulation does not modify general ethylene-mediated growth responses, but may have a specific role in hyponastic growth.
Supplemental Fig. S7), and transverse sections (Fig. 3B) of histochemically stained petioles were variable but showed overall that ethylene results in transcriptional repression of A2-type CYCLINs after 6 h of ethylene treatment, despite the fact that the GUS protein is relatively stable and transcriptional down-regulation was not yet measurable after 3 h by qRT-PCR (Fig. 2F). These results suggest that ethylene suppresses A2-type CYCLIN expression in Arabidopsis petioles.
To corroborate these findings based on promoter::GUS analysis, we analyzed A2-type CYCLIN expression in microdissected fragments of wild-type petioles (Fig. 3C) by qRT-PCR. This revealed that CYCA2;2, CYCA2;3, and CYCA2;4 are transcriptionally down-regulated Significance levels under the bars reflect the difference between expression in air control and after ethylene treatment in the respective petiole quarter, and significance levels above the gray brackets represent the difference between the adaxial and abaxial side of the petiole fragments represented in C; ns, nonsignificant; *, P , 0.05; and **, P , 0.01. after prolonged (6 h) ethylene treatment in all four fragments (Fig. 3D). Notably, CYCA2;1 was also generally down-regulated, including in the proximal-abaxial section (fragment 1, Fig. 3, C and D), but appeared modestly up-regulated at the proximal-adaxial side (Fig. 3D).
Because A2-type CYCLINs control a cell cycle checkpoint upstream of the expression of mitotic regulators such as B-type CYCLINs (Vanneste et al., 2011), we anticipated and confirmed that ethylene also represses expression of the mitotic CYCLIN, CYCB1;1, in these tissues (Fig. 3E). These results are consistent with a suppression of proliferation by ethylene in petiole tissues.

Endoreduplication Cannot Explain Exaggerated
Hyponastic Growth in EHY-D and 35S::CYCA2;1 A2-type CYCLINs have been implicated in the control of local cell cycle progression to fine tune development (Vanneste et al., 2011). More specifically, their expression levels can affect the local balance between cell proliferation and endoreduplication, a process of consecutive rounds of DNA replication without mitosis (Yu et al., 2003;Imai et al., 2006;Yoshizumi et al., 2006) that has been associated with cells that have an increased capacity for elongation (Cheniclet et al., 2005;Roeder et al., 2010). Despite previous reports on ethylene-related changes in endoreduplication in hypocotyls (Gendreau et al., 1999;Dan et al., 2003), we did not find significant ethylenedependent differences in ploidy levels in microdissected proximal and distal fragments of wild-type petioles (Supplemental Fig. S8). However, we did observe slightly different ploidy levels between distal and proximal regions of the petiole (Supplemental Fig. S8), as well as small but significant differences in the 2n, 4n, 16n, and 32n classes between the wild type and EHY-D (P = 0.041, P = 0.036, P = 0.007, and P = 0.05, respectively; Supplemental Fig. S8). However, no significant differences were detected between the wild type and 35S::CYCA2;1 (except for the 4n class, P = 0.025; Supplemental Fig. S8), making it unlikely that the exaggerated hyponastic growth of both CYCA2;1-overexpressing lines (EHY-D and 35S::CYCA2;1) can be explained by changes in ploidy levels. Moreover, a dominant negative CYCLIN-DEPENDENT KINASE B1-1 (CDKB1;1)-overexpressing line with enhanced ploidy levels in aerial organs (Boudolf et al., 2009) showed ethylene-induced hyponastic growth that was indistinguishable from the wild type (Supplemental Fig.  S8). Together, these data argue against a major role of endoreduplication in the exaggerated hyponastic growth in EHY-D.

Ectopic Overexpression of CYCA2;1 Comprises Ethylene-Mediated Differential Cell Proliferation during Hyponastic Growth
We examined if enhanced hyponasty in 35S::CYCA2;1 is due to enhanced cell expansion in the petiole compared with the wild type. Measurements of epidermal cell lengths revealed that significant cell expansion in 35S::CYCA2;1 under ethylene treatment occurs in an approximately 2-mm epidermal zone at the proximal abaxial side of the petiole (Fig. 4, A and B). The pattern of changes in cell size strongly resembles the pattern previously observed in wild-type Arabidopsis Col plants (Fig. 4, C and D;Polko et al., 2012b;Rauf et al., 2013). Moreover, 35S::CYCA2;1 showed a similar ethyleneinduced CMT reorientation as described previously for wild-type plants (Polko et al., 2012b;Supplemental Fig. S9). Although this suggests that the exaggerated hyponastic growth in 35S::CYCA2;1 is not due to differences in cell expansion compared with the wild type, this cannot be concluded without taking cell proliferation into account (see Supplemental Text S1). However, due to experimental constraints, cell division rates cannot be derived from empirical in vivo cell length measurements (see Supplemental Text S1). This is because both time lapse imaging of cell division as well as destructive measurements directly interfere with the hyponastic response itself. Nevertheless, the dynamic petiole shape and static cell size distributions as observed from epidermal imprints (Fig. 4, A-D) together provided sufficient information to allow for a mathematical analysis that indirectly estimates the contribution of cell divisions within the petiole. Using such a mathematical approach, we calculated relative division rates between abaxial versus adaxial cells, which is sufficient to describe the role of cell division in petiole hyponasty. Theoretical details can be found in the "Materials and Methods" and Supplemental Texts S2 and S3. The mathematical analysis showed that, in wild-type plants, ethylene treatment strongly increased the bias toward adaxial cell proliferation in the proximal region of the petiole, with adaxial cell division rates during ethylene treatment being up to 80% higher than abaxial cell division rates (Fig. 4, E and F). This indicates that, in this region, ethylene triggers either increased adaxial cell proliferation or decreased abaxial cell proliferation. Our qRT-PCR and GUS analysis showed that ethylene in general suppresses cell proliferation markers (Fig. 3, B and D;Supplemental Figs. S6 and S7). The most likely scenario would therefore be a decrease in cell proliferation. The analysis predicts that a local reduction of cell proliferation rate in an approximately 2-mm-long epidermal cell zone at the proximal part of the abaxial side of the petiole is required to match the measured amplitude of hyponastic growth in the wild type. This is in accordance with Polko et al. (2012b), where a comparable analysis showed that, under the assumption that abaxial and adaxial cell proliferation rates are equal, the observed changes in abaxial cell elongation rates during ethylene treatment result in a larger hyponastic response in comparison with what was experimentally observed. We observed that the 35S::CYCA2;1 line was lacking such increased bias toward adaxial cell proliferation after ethylene treatment. Instead, in this line, the cell proliferation profile in ethylene was estimated to be highly similar to the cell proliferation profile in the untreated controls (Fig. 4, E and F). In other words, cell proliferation rates in 35S::CYCA2;1 are predicted to be comparable between control and ethylene treatment, whereas in the wild type, ethylene exposure represses abaxial cell divisions relative to adaxial cell divisions.
Together, this suggests that ethylene controls differential cell proliferation in the petiole, thereby affecting the amplitude of ethylene-induced hyponastic growth. Our calculations indicate that ethylene causes enhanced abaxial cell elongation leading to hyponastic growth, while at the same time suppressing proliferation, thus attenuating hyponasty. In 35S::CYCA2;1, this secondary control mechanism is overruled, leading to exaggerated hyponastic growth.
In Silico Modeling of Hyponastic Growth Corroborates that Absence of Differential Cell Proliferation Leads to Exaggerated Hyponasty Tissue growth can be described by the combination of cell expansion and cell division. However, cell divisions, as opposed to cell elongation, do not immediately generate volumetric tissue growth (only extra cells). Therefore, divisions can only have an indirect effect on tissue growth (Harashima and Schnittger, 2010). The effect of cell division on tissue growth depends on the specific relationship between cell size and cell expansion, as is discussed in Supplemental Text S4. Our mathematical analysis on petiole epidermal cell lengths indicated that the amplitude of ethyleneinduced hyponastic growth is mediated by a differential regulation of cell division, and that the enhanced hyponastic growth response in 35S::CYCA2;1 is correlated with an absence of reduction in cell division (Fig. 4, E and F), suggesting that CYCA2;1 has a role in the mechanism that mediates the hyponastic growth response specifically via reduced abaxial cell proliferation. To further explore the relationship between (abaxial) cell division and the hyponastic growth response, we developed an in silico model of the Arabidopsis petiole. With the model, we simulated ethylene-induced hyponastic growth for different scenarios of cell expansion: linear, exponential, logistic, Figure 4. Ethylene effect on CYCA2;1mediated cell expansion and proliferation. A and B, Average epidermal cell lengths as experimentally measured from 35S::CYCA2;1 petioles per 200-mm class, according to their distance relative to the proximal side of the petiole of adaxial (A) and abaxial (B) epidermal cells after 10-h control (black circles) and 10-h ethylene (white squares) treatment. Significance levels (2-tailed Student's t test): *P , 0.05, n = 13-15. C and D, Differential experimentally determined cell growth after 10-h ethylene treatment over the length classes in 35S::CYCA2;1 (gray lines; this study) in comparison with the wild type (black dashed line as has been published in Polko et al., 2012b;Fig. 2). Error bars are SEM. E and F, Calculated relative cell proliferation after 10-h control (E) and ethylene treatment (F) in the wild type (black circles) and 35S::CYCA2;1 (white circles), presented as the difference (ratio) between adaxial and abaxial cell proliferation rate. Values greater than 0 indicate that adaxial proliferation is predominant, and 0 means equal cell proliferation on both sides. and logarithmic growth and division rates. Since it has been shown that the epidermal cell layer is sufficient to drive and restrict plant growth (Savaldi-Goldstein et al., 2007), we modeled only the epidermal layers of the abaxial and adaxial sides of the petiole (Fig. 5A). The experimental data showed that cell size increases along the petiole (Fig. 4, A and B). We simplified the growth and cell cycle dynamics by assuming in the first instance that, apart from the adaxial-abaxial differences, the petiole is spatially homogenous (i.e. that cell expansion and division are not influenced by their proximal-distal position in the petiole; for details on the in silico modeling, see "Materials and Methods" and Supplemental Text S5).
Because the precise relationship between cell size and cell expansion dynamics is not well established (see Supplemental Text S5), we simulated hyponastic growth of the petiole for different possible scenarios of cell expansion, namely linear, exponential, logistic, and logarithmic, and combined this with different cell division scenarios.
A first round of simulations assessed the possible effect of reduced cell division in the proximal abaxial region on petiole shape for the different possible cell expansion scenarios, other than an increase in abaxial cell expansion due to ethylene treatment, and analyzed whether a reduction in cell proliferation (ranging from 0%-100%) within that region could indeed attenuate the hyponastic response that would be expected without reduction in abaxial cell proliferation (Fig. 5B). These simulations support the idea that reduction in abaxial cell proliferation leads to reduction in hyponastic petiole curvature, except when cell expansion is exponential. As explained in Supplemental Text S4, however, exponential cell expansion implies that the occurrence of cell divisions has no influence whatsoever on the tissue growth, which is an unrealistic scenario. These simulations thus indicate that a decrease in abaxial cell division is expected to reduce hyponastic growth.
Next, we explored the role of cell expansion arrest during cell division. In the simulations described earlier, we did not take into account that the cell division event itself could directly affect the cell expansion. It is very likely, however, that cell expansion is arrested for a certain amount of time when the cell goes through mitosis (Beemster and Baskin, 1998;Grieneisen et al., 2007). Figure  5C shows that prolonged periods of arrest in expansion during mitosis can counterbalance the effect of reduced abaxial cell division on petiole curvature, but only for a duration of the expansion arrest that is unrealistically long.
Next, we used this model to evaluate the impact of misregulated cell proliferation, as expected for overexpression or mutation of CYCA2;1 (Fig. 5, D-G). To capture the complete petiole shape, we used the observation that cell division and elongation are limited to the proximal 3 mm of the petiole (Fig. 4, A and B). The shape of the distal part of the petiole is therefore considered conserved over the period of the experiment (see Supplemental Text S5). For all three genetic backgrounds (wild type, 35S::CYCA2;1, and cyca2;1), we parameterized that ethylene treatment increases abaxial cell elongation (Fig. 4, A and B), and that mitosis causes a 1-h arrest in cell expansion. Following our hypothesis and the simulations described earlier, we assumed that, in the wild type, abaxial cell division decreases during ethyleneinduced hyponastic growth (Fig. 5E). Alternatively, we modeled genetic backgrounds that lack such differences between abaxial and adaxial cell division by setting overall cell division rates to be constitutively higher and constitutively lower, reflecting scenarios of overexpression and mutation of CYCA2;1, respectively (Fig. 5, F and G). As was observed in the experimental measurements, the simulations result in an increased hyponastic growth response for both 35S::CYCA2;1 and, to a slightly lesser extent, cyca2;1, except under the unrealistic scenario of exponential growth (Fig. 5D).
Taken together, our in silico model shows that reduced abaxial cell division decreases the amplitude of hyponastic growth. Furthermore, it demonstrates that when this mechanism is impaired by either constitutive CYCA2;1 overexpression or by a knockout mutation, the hyponastic growth response becomes exaggerated, as was experimentally observed in both 35S::CYCA2;1 and cyca2;1 (Fig. 2

DISCUSSION
Hyponastic growth is an adaptive response by which plants cope with adverse environmental conditions. The response is controlled by complex interactions between various phytohormones. However, since hyponastic growth induced by various independent environmental stimuli is highly similar in kinetics and amplitude, the signaling mechanisms likely converge downstream on specific functional molecular components that control the response (Van Zanten et al., 2010b). We aimed to identify unique molecular hyponastic growth regulators and isolated EHY-D, which has exaggerated amplitudes of leaf movement upon induction by ethylene, low light intensity, and high temperatures. Because EHY-D exhibited an enhanced response to each treatment investigated, the insertion likely affects a general downstream determinant of hyponasty. Our study shows that the core cell cycle regulator CYCA2;1 was overexpressed in EHY-D. Several independent A2-type CYCLIN overexpression lines and mutants showed consistently altered hyponastic growth phenotypes ( Fig. 2; Supplemental Figs. S3-S5), indicating that A2-type CYCLINs are important determinants of the hyponastic response.
Our results suggest that A2-type CYCLINs operate in a specific branch of ethylene signaling that affects differential growth, but not hypocotyl elongation. We found that prolonged (6 h) ethylene treatment results in downregulation of A2-type CYCLINs in the petiole (Fig. 3, B and  D). This down-regulation is initiated at least 3 h after the start of ethylene treatment, because up to this time point, A2-type CYCLIN transcription was unaffected (Fig. 2F). Since hyponastic growth is induced already within the first hour after ethylene application, transcriptional control of A2-type CYCLINs likely does not control the induction of hyponastic growth.
The promoter region of CYCA2;1 contains eight ERFbinding ethylene responsive elements (Richard et al., 2001), implying that ethylene could control its transcription directly through ERF transcription factors that have the ethylene responsive element as their promoter targets.

Hyponastic Growth Does Not Depend on Ploidy Levels
A-type CYCLINs are expressed at the S-to-M transition of the mitotic cell cycle, prior to activation of B-type CYCLINs (Inzé and De Veylder, 2006) and are ratelimiting factors for entry in the mitotic cell phase (Burssens et al., 2000;Dewitte and Murray, 2003;Yu et al., 2003;Vanneste et al., 2011). Down-regulation of CYCA2 levels causes a shift toward endoreduplication (Imai et al., 2006;Yoshizumi et al., 2006;Vanneste et al., 2011) and is associated with developmentally controlled cell cycle exit (Vanneste et al., 2011). This process is generally associated with differentiating cells undergoing cell expansion (Sugimoto-Shirasu and Roberts, 2003), e.g. in Values are relative to petiole curvature without reduced abaxial cell division. Results are shown for linear, exponential, logistic, and logarithmic relations between cell length and cell expansion (see Supplemental Text S4). C, Effect of cell expansion arrest due to cell division on petiole curvature. Data are shown for linear cell expansion, values are relative to petiole curvature without reduced abaxial cell division and no cell expansion arrest. D, Simulation of ethylene-induced hyponastic growth in the wild type, 35S::CYCA2;1, and cyca2;1 background. Results are shown for linear, exponential, logistic, and logarithmic growth models. Values represent differential hyponastic growth relative to control treatment. The in silico model describes the cell expansion and divisions within the proximal part of the petiole, which is initially 3 mm but grows during the simulation, whereas the cells in the distal part (7 mm) were considered to have reached their final size (see Supplemental Text S5). E to G, Graphical representation of results shown in D for logarithmic growth. The symbols represent a side-on view of the petiole (see Supplemental Text S5) at t = 0 h (triangles) and after t = 10 h of either control treatment (squares) or ethylene treatment (circles). The scheme underneath E and F represents a simplification of the proposed mechanism that controls the amplitude of ethylene-induced hyponastic growth, as implemented in the in silico model for the wild-type, 35S::CYCA2;1, and cyca2;1 genetic background.
Brassica oleracea petals (Kudo and Kimura, 2002). Consistently, ethylene has been shown to induce endoreduplication events in hypocotyls of Cucumis sativus and Arabidopsis (Gendreau et al., 1999;Dan et al., 2003). In our study, however, neither CYCA2;1 overexpression nor ethylene application strongly affected ploidy levels in Arabidopsis petioles, which suggests that ethyleneinduced hyponastic petiole growth is not regulated by CYCA2-mediated effects on the endocycle. This is in agreement with a previous study showing that petiole growth is independent of changes in ploidy levels (Kozuka et al., 2005). Possibly, the occurrence and role of endoreduplication in organ growth is less pronounced in petioles of mature plants than in hypocotyls of very young seedlings.

Differential Cell Proliferation Can Control Hyponastic Growth Amplitudes
Our finding that CYCA2;1 overexpression or mutation causes an exaggerated hyponastic growth response is difficult to explain intuitively. Therefore, we developed an in silico model based on experimentally determined parameters, incorporating the effect of cell elongation and cell proliferation on petiole shape. It was critically important that the model predicted that a lack of differential cell proliferation between abaxial and adaxial petiole sides results in stronger hyponastic growth responses.
By combining cell length data and petiole shape, we were able to assess the influence of ethylene treatment and constitutive CYCA2;1 expression on relative cell proliferation rates during hyponastic growth. This mathematical analysis showed that, in wild-type petioles, adaxial cell proliferation increases relative to abaxial cell proliferation during ethylene treatment. This can be caused by an increase in adaxial cell proliferation, a decrease in abaxial cell proliferation, or a combination of both. The scenario of decreased abaxial cell proliferation is the most likely explanation for the observed effect on hyponastic growth in the wild type, as this is in line with the observed down-regulation of A2-type CYCLIN expression following ethylene treatment. This is in accordance with a previous study indicating that ethylene can arrest the cell cycle by directly affecting core cell cycle components (Skirycz et al., 2011). Together, these data, combined with our mathematical analyses and the in silico model, suggest that the control of the amplitude of ethylene-induced hyponastic growth relies on a dual mechanism. Ethylene enhances cell elongation along the abaxial side of the petiole in wild-type plants, providing the tissue growth required for the upward movement of the petiole, while also down-regulating CYCA2;1 and CYCB1;1 expression, conceivably in a differential manner (see below), reducing abaxial cell proliferation. This reduced cell proliferation counteracts the effects of cell elongation, thereby attenuating the amplitude of the hyponastic response. When CYCA2;1 is constitutively expressed, no differential inhibition of cell proliferation occurs, leading to an exaggerated hyponastic response. Our in silico model of hyponastic growth provides a proof of concept for this mechanism. Importantly, in addition to confirming the exaggerated response when CYCA2;1 is constitutively expressed, the model also predicts that the same mechanism results in exaggerated hyponastic growth response when no CYCA2;1 is present. This was experimentally observed using cyca2;1 knockout lines, which indeed show exaggerated hyponasty under ethylene treatment (Fig. 2, G and H).
The exact molecular mechanism by which ethylene installs differential cell proliferation between the abaxial and adaxial petiole side in ethylene-treated wild-type plants remains to be elucidated. Our work suggests that CYCA2;1 is critically involved. In this context, it is essential to consider the central role of distinct CDKs in complex with CYCLINs in controlling cell cycle checkpoints during cell proliferation. Besides the association with distinct CYCLINs whose levels are controlled at the level of transcription and protein stability, CDK activity is further fine tuned via interaction with proteins such as Kip-related proteins and regulatory phosphorylation events (Inzé and De Veylder, 2006;Polyn et al., 2015). Even if such components are involved, it remains unknown how the abaxial versus adaxial differentiation comes about, and this remains an important question for future studies.
The effect of ethylene on cell proliferation is complex and largely depends on the tissue context. On the one hand, ethylene was found to stimulate proliferation in the Arabidopsis root stem cell niche (Ortega-Martínez et al., 2007), in submergence-induced adventitious root growth in rice (Lorbiecke and Sauter, 1999), and in subsidiary cells of cucumber (C. sativus) hypocotyls and vascular tissues (Love et al., 2009;Etchells et al., 2012). On the other hand, in developing leaves, ethylene was found to suppress proliferation during mild osmotic stress (Skirycz et al., 2011), similar to the suppression in petioles presented in this work. Interestingly, this osmotic stress-induced cell cycle arrest is associated with regulation of CDKA;1 activity that does not involve EIN3-mediated transcriptional changes (Skirycz et al., 2011), suggesting that the effect of ethylene on differential regulation of proliferation in the petioles could involve nontranscriptional regulation. Consistently, we found that down-regulation of A2-type CYCLINs in the petiole does not occur within 3 h of ethylene treatment (Fig. 2), whereas hyponastic growth is induced within the first hour after ethylene application.

The Bipartite Role for Ethylene in Hyponastic Response
In conclusion, we propose a dual role for ethylene in the mechanism regulating hyponastic growth. Ethylene (1) induces cell elongation in the abaxial petiole epidermal cells to power the upward leaf movement, and (2) inhibits the mitotic cell cycle, likely in part by affecting CYCA2;1 expression, in the same tissue. An in silico model confirmed that such a mechanism can explain the observed exaggerated hyponastic growth in both 35S::CYCA2;1 and cyca2;1 null mutants. The dual role for ethylene found in this work adds to an increasing number of studies that indicate both growth stimulatory and inhibitory roles for ethylene in plant development, abandoning the classic idea of ethylene simply being a growth inhibitor. The hormone rather inhibits or stimulates growth in a subtle manner that integrates information from the environment together with developmental state and cellular identity of a tissue/organ (for review, see Pierik et al., 2006). This mechanism controls the magnitude of ethylene-induced hyponastic leaf movement in an effort to optimize plant performance under stressful conditions.
Isolation of 35S::At3g25390 from the Arabidopsis TF ORF-Expression ERF ectopic expression library (Weiste et al., 2007) is described in Supplemental Materials and Methods S1. This line had an 11.7 6 0.2 times higher expression of SHN3 than the wild type, as determined by qRT-PCR.
Seeds were stratified at 4°C for 4 d, sown on a fertilized mixture of soil and perlite, and grown at 20°C, 70% (v/v) relative humidity, 200 mmol m -2 s -1 photosynthetically active radiation (9-h photoperiod) as described earlier (Millenaar et al., 2005). Thirty-day-old plants in stage 3.9 (Boyes et al., 2001) were used for all experiments, except for the hypocotyl elongation assay (below). One day before the start of the experiments, plants were transferred to the experimental setup with similar conditions to the growth chambers (Microclima 1750 growth cabinet; Snijders Scientific). To rule out effects of diurnal and circadian leaf movements, all treatments commenced 1.5 h after the start of the photoperiod.

Real-Time Reverse Transcriptase-PCR and Histochemical b-Glucuronidase Staining
Real-time reverse transcriptase-PCR was conducted as described in Millenaar et al. (2005) and is described in detail in Supplemental Materials and Methods S1. Primers are shown in Supplemental Table S2 and in Richard et al. (2001), Mariconti et al. (2002), Yoshizumi et al. (2006), andVanneste et al. (2011).
For histochemical GUS staining, tissues were harvested and placed briefly in 90% (v/v) ice-cold acetone and subsequently fixed and vacuum infiltrated with 10 mM MES, 0.3 M mannitol, and 0.3% (v/v) formaldehyde for 45 min. Tissues were rinsed in 100 mM buffer (50 mM NaHPO 4 + 50 mM Na 2 HPO 4 ; pH 7.2). The histochemical reaction was performed by incubation in 1 mM 5-bromo-4-chloro-3-indolyl-b-glucuronic acid) in dimethyl sulfoxide for 24 h at 37°C. The tissues were cleared in an ethanol series of increasing concentrations (5%-90% [v/v]) and were either hand sectioned and directly observed or first embedded in Technovit 7100 (Kulzer) and dissected using a microtome. The resulting 200-mm sections were observed and photographed with an Olympus BX50 WI mounted with an Olympus DP 70 camera.

Ethylene, Low Light, and High Temperature Treatments
Ethylene (Hoek Loos) was applied to saturating (Polko et al., 2012b) concentrations (1.5 mL L -1 , except for the hyponastic growth kinetics experiment; see below) in continuous flow through and then vented away. The concentration was regularly checked by gas chromatography analysis. The control treatment was done in the same experimental cabinet. For leaf movement kinetics analysis, 5 mL L 21 ethylene was mixed with 70% (v/v) humidified air and applied to glass cuvettes containing one plant each at a flow rate of 75 L h 21 as described in Millenaar et al. (2005).
Low light intensity was induced by decreasing the photosynthetically active radiation level from 200 to 20 mmol m 22 s 21 by switching off lamps and by covering the plants with spectrally neutral shade cloth. This did not change light quality (checked with a LI-COR 1800 spectro-radiometer). Induction of high temperature was accomplished by moderating the program of the used growth cabinet. The 30°C threshold was reached after 22 min; 38°C was reached after 49 min.

Genetic Screen and Cloning of EHY-D
For details on the genetic screen that identified EHY-D, see Polko et al. (2012a). To facilitate easy and fast screening, we first checked if wild-type plants were able to exhibit a normal low light-induced hyponastic response after 6-h ethylene treatment and an overnight recovery. Ethylene-induced hyponastic growth was, as expected, quickly reversed by removing the ethylene source (Millenaar et al., 2005), and this treatment did not interfere with low lightinduced hyponasty in the subsequent photoperiod (next day; Supplemental Fig. S10). In total, we screened 17,500 individual Cauliflower mosaic virus 35S enhancer (activation)-tagged (Weigel et al., 2000) vegetative plants in developmental stage 3.7 (Boyes et al., 2001). The plants were visually monitored for (1) petiole angle after 6 h of ethylene treatment and after overnight recovery, and (2) the petiole angle after 6-h low light treatment. To check the number of inserts, crosses were made between wild-type and the glufosinate ammonium (Basta)-resistant EHY-D. Self-pollinated F2 progeny seeds were subjected to Basta selection on agar plates containing 8 g L 21 plant agar (Duchefa), 0.22 g L 21 Murashige and Skoog (Duchefa), and 50 mg mL 21 DL-glufosinate ammonium (Basta/DL-phosphinotricin; Duchefa). After 3 weeks, survival ratios were scored. Thermal Asymmetric Interlaced-PCR was conducted to identify the T-DNA locus in EHY-D as described by Liu et al., (1995). For details, see Supplemental Materials and Methods S1.

Petiole Angle Measurements
Petiole angle kinetics was measured using an automated time-lapse camera system as described in Millenaar et al. (2005). Plants were placed in glass cuvettes with the petiole of study perpendicular to the axis of the camera. To facilitate measurement, leaves obscuring the petiole base were removed, and an orange paint droplet (Decofin Universal) was used to mark the petiole/lamina junction. This did not affect the response (data not shown). Digital images of two petioles per plant were taken every 10 min. Angles were measured between the marked point at the petiole/lamina junction and a fixed proximal point of the petiole, relative to the horizontal, using the KS400 (version 3.0) software package (Carl Zeiss Vision) and a custom-made macro. To enable continuous photography over the 24-h experimental period, no dark period was included during the experiments.
Plants used for measurements at fixed time points were manually photographed from the side. Angles were measured using ImageJ Software (Abramoff et al., 2004). Before further analysis, two petioles per plant were measured and an average was calculated. Statistical significance levels were determined using type-2 2-tailed Student's t test.
To rule out diurnal and/or circadian effects on petiole movement, a pairwise subtraction was performed on hyponastic growth data. Differential petiole angle describes a difference between the angle in control versus experimental conditions at each time point (Benschop et al., 2007). The new SE for the differential response was calculated by taking the square root from the summation of the two squared SEs. Initial petiole angles at t = 0 h of the A2-type CYCLIN-related lines are shown in Supplemental Table S1. et al. (2012b). Cell lengths were quantified using a custom-made macro in KS400 software (Zeiss). Each cell was assigned to a 200-mm class, according to its position relative to the most proximal part of the petiole.
To visualize microtubules in the 35S::CYCA2;1 background, we crossed this line with 35S::TUA6:GFP (Ueda et al., 1999). After 5 to 10 h of the ethylene/ control treatment, CMTs of petiole epidermal cells were visualized using an inverted confocal laser-scanning microscope (Leica CS SPII, 633 C-apochromat objective, excitation wavelength of 488 nm, collecting at 505-530 nm for GFP emission). Petioles were divided in quadrants depending on their distance from the base, and the abaxial and adaxial sides were observed separately. Only CMT areas at least twice as long as the cell width were taken into account and grouped in categories relative to the long cell axis: transverse (0°), oblique 30°, oblique 60°, longitudinal (90°), and randomly oriented, according to Himmelspach and Nick (2001).

Hypocotyl Elongation Assay and Ethylene Release Measurements
Hypocotyl elongation assay was conducted as described in Van Zanten et al. (2010a). Sterilized seeds were sown on petri dishes containing Murashige and Skoog-enriched plant agar (4 g L 21 plant agar [Duchefa], 0.22 g L 21 Murashige and Skoog [Duchefa], and different concentrations of ACC [Duchefa]). Plates were kept for 4 d at 4°C in dark. To induce germination, plates were transferred to 200 mmol m 22 s 21 light for 4 h and, subsequently, wrapped in aluminum-foil. Thereafter, the plates were left in darkness for 5 d at 20°C. Seedlings were photographed and hypocotyl length was measured using ImageJ software (Abramoff et al., 2004).
Ethylene release measurements were performed on 30-d-old plants, 1.5 6 1 h after the start of the photoperiod, as described in Millenaar et al. (2005Millenaar et al. ( , 2009. Whole rosettes of about 300 mg were weighed and then placed in a syringe with a volume of 1.5 mL. Ethylene was allowed to accumulate in the syringe for 15 to 20 min. Subsequently, the air was analyzed on a gas chromatograph (GC955; Synspec). This short time frame prevented wound-induced ethylene production, which started to accumulate only after 20 min.

Mathematical Analysis of Cell Proliferation Rates
The mathematical model to predict cell proliferation rates is similar to the one described in Polko et al. (2012b). The background on secondary measurements of cell proliferation is explained in Supplemental Text S1, and parameters are in Supplemental Table S3. Petiole shape was quantified by fitting a function through the measured petiole angle data describing the proximal angle (petiole emergence from the shoot) and the distal angle (intersection of petiole and leaf blade). A smooth function was fitted to the measured cell lengths along the petiole to correct for variability. Given that along the adaxial side, hardly significant differences were found in cell lengths (Fig. 4A), the adaxial cell length data were fitted collectively to a single overarching function. In contrast, the abaxial cell length data showed significant differences in the proximal part (Fig.  4B) and were fitted for each individual data set independently. Since no significant differences in cell length were found in the distal part of the petiole, an extra constraint was added that required that the maximum cell length (in the distal part of the petiole) would be the same for the different data sets (Supplemental Table S3).
The curve describing petiole shape at t = 0 h was divided into 50 sections of 200 mm. An arc was fitted to each section, and by combining the curve of the arc with the function fitted to the measured cell lengths, the number of adaxial and abaxial cells per section could be calculated (see Supplemental Text S2 and Supplemental Table S3). The cell number per section at t = 0 h, combined with the functions describing the adaxial and abaxial cell lengths for the 10-h control and ethylene treatments, allowed calculation of predicted petiole shape after 10-h treatment for the null hypothesis, which assumes no cell proliferation during the treatments. Deviation from the predicted petiole shape to the observed shape allowed prediction of adaxial or abaxial cell proliferation (Supplemental Text S3). Since overall petiole elongation was not taken into account, the obtained cell proliferation rates represent relative rather than absolute values.

In Silico Model of Hyponastic Growth
In the in silico model, the (hyponastic) growth of the petiole was simulated for 10 h using 1-h time intervals. During each time step, cells expanded and/or divided. Cells could only divide after reaching a specified minimal length, after which the probability to divide existed and was evaluated at each time step. Simulations were initiated with a petiole that consisted of cells that were randomly selected from a population of in silico growing and dividing cells. This population was generated by simulating cell elongation and cell division for 10,000 time steps, starting from a single cell. All simulations were repeated 1,000 times (for parameters of all simulations, see Supplemental Table S4). For the results shown in Figure 5, D-G, the initial (abaxial and adaxial) length of the petiole was set such that the shape of the petiole resembles observations for the wild type at t = 0 h. Supplemental Text S5 discusses the relation between cell length and elongation for linear, exponential, logistic, and exponential growth, as well as the calculation of the petiole curvature (for results shown in Fig. 5, B and C) and differential (hyponastic) growth (for results shown in Fig. 5, D-G).

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S3. Correlation analysis between CYCA2;1 expression and amplitude of hyponastic growth in response to ethylene.
Supplemental Figure S4. Hyponastic response of cyca2;1 mutants upon low light and high temperature treatment and expression of A2-type CYCLINs.
Supplemental Figure S6. Histochemical analysis of A2-type CYCLIN promoter activity in rosettes and petioles.
Supplemental Figure S7. Histochemical analysis of ethylene effects on A2-cyclin promoter activity in petioles.
Supplemental Figure S8. Effects of ethylene on endoreduplication in petioles.
Supplemental Figure S10. Ethylene treatment prior to low light treatment does not affect kinetics of low light-induced hyponastic growth.
Supplemental Table S1. Initial petiole angles at t = 0 h of A2-type CYCLINrelated mutants described in this work.
Supplemental Table S2. Primers used for real-time qRT-PCR.
Supplemental Table S3. Parameters used to profile cell proliferation rates.
Supplemental Table S4. Parameters used in the in silico model.
Supplemental Text S1. Indirect measurements of cell division and expansion rates.
Supplemental Text S2. Deriving the number of cells from petiole shape and cell lengths.
Supplemental Text S3. Profiling of cell proliferation along the petiole.
Supplemental Text S4. Influence of cell division on tissue growth.
Supplemental Text S5. Additional information for the in silico model of petiole.
Supplemental Materials and Methods S1.