Negative reciprocal interactions between gibberellin and cytokinin in tomato

Authors


Author for correspondence:
David Weiss
Tel: + 972 8 9489436
Email: weiss@agri.huji.ac.il

Summary

  • The hormones gibberellin (GA) and cytokinin (CK) exhibit antagonistic effects on various processes in many species. Previous studies in Arabidopsis have shown that GA inhibits CK signaling. Here, we have investigated the cross-talk between GA and CK in tomato (Solanum lycopersicum).
  • We altered the balance between GA and CK activities by exogenous applications and genetic manipulations, and tested an array of physiological and developmental responses.
  • GA and CK showed antagonistic effects on various developmental and molecular processes during tomato plant growth. GA inhibited all tested CK responses, including the induction of the CK primary response genes, type A Tomato Response Regulators (TRRs). CK also inhibited a subset of GA responses. In contrast with exogenous application of GA, the endogenous GA-independent GA signal generated by the loss of the DELLA gene PROCERA (PRO) did not repress CK-regulated processes, such as anthocyanin accumulation, TRR expression and leaf complexity.
  • Our results suggest a mutual antagonistic interaction between GA and CK in tomato. Although GA may inhibit early steps in the CK response pathway via a DELLA-independent pathway, CK appears to affect downstream branch(es) of the GA signaling pathway. The ratio between the two hormones, rather than their absolute levels, determines the final response.

Introduction

The plant hormones gibberellin (GA) and cytokinin (CK) play important roles in the regulation of plant development. Although both are considered to be growth-promoting hormones, they exhibit antagonistic effects on various developmental processes (Weiss & Ori, 2007). Several previous studies have shown that the two hormones interact at the biosynthesis, catabolism and signaling levels (Greenboim-Wainberg et al., 2005; Jasinski et al., 2005; Bolduc & Hake, 2009; Maekawa et al., 2009).

The signaling pathways of GA and CK have been studied intensively in recent years, and major components, as well as their order of action, have been elucidated (Muller & Sheen, 2007; Harberd et al., 2009). DELLA proteins act as important negative regulators of GA signaling (Ueguchi-Tanaka et al., 2007). They interact with the growth-promoting bHLH transcription factors, Phytochrome Interacting Factors (PIFs), and inhibit their activities (Feng et al., 2008; de Lucas et al., 2008). The Arabidopsis genome contains five DELLA genes, whereas, in the rice genome, only one family member has been identified (Itoh et al., 2002). GA binding to the soluble GA receptor GA INSENSITIVE DWARF 1 (GID1) triggers its interaction with DELLA proteins (Griffiths et al., 2007). GID1–DELLA interaction stimulates the binding of DELLA proteins to an SCF E3 ubiquitin ligase via specific F-box proteins (GID2/SLY). This leads to polyubiquitination and degradation of the DELLA protein by the 26S proteosome (Harberd et al., 2009), resulting in the stimulation of GA responses. Several studies have identified other factors that affect GA responses, including the GA response inhibitor SPINDLY (SPY) (Jacobsen & Olszewski, 1993; Filardo & Swain, 2003). Although there is no direct biochemical evidence for the interaction between SPY and DELLA proteins, genetic evidence suggests that SPY is required for full DELLA activity to repress the GA response (Silverstone et al., 2007).

The CK signaling pathway in Arabidopsis starts with the binding of the hormone to the membrane-localized histidine (His) kinase receptors and their autophosphorylation. The phosphate group is then transferred by His phosphotransfer (HPT) proteins to the nucleus to phosphorylate a set of transcriptional regulators known as type-B Arabidopsis Response Regulators (ARRs) (To & Kieber, 2008). The phosphorylated type-B ARRs induce the transcription of various CK-regulated genes, including type-A ARRs (Werner & Schmulling, 2009). The latter suppress CK responses, and thus are involved in negative feedback regulation of the pathway (To et al., 2004).

GA and CK exhibit antagonistic effects on numerous developmental processes (Greenboim-Wainberg et al., 2005; Jasinski et al., 2005; Maekawa et al., 2009). High CK and low GA signals are required for normal shoot apical meristem (SAM) function (Sakamoto et al., 2001; Jasinski et al., 2005; Yanai et al., 2005). Regulators of SAM maintenance from the KNOXI family have been shown to affect the homeostasis between the two hormones: they induce the expression of the CK biosynthesis gene IPT7 and the accumulation of the hormone in the meristem (Jasinski et al., 2005; Yanai et al., 2005; Sakamoto et al., 2006), and directly suppress the transcription of the GA biosynthesis gene GA-20 oxidase (GA20ox) in this tissue (Sakamoto et al., 2001; Hay et al., 2002; Chen et al., 2004). Moreover, KNOXI and CK induce the expression of the gene encoding the GA-deactivating enzyme GA2ox at the base of SAM, perhaps to block biologically active GAs from being transported into SAM from flanking tissues (Jasinski et al., 2005; Bolduc & Hake, 2009).

How are the relative levels of GA and CK interpreted by the plant? We have shown previously that GA inhibits CK signaling in Arabidopsis, probably through a positive regulation of CK responses by the GA signaling repressor SPY (Greenboim-Wainberg et al., 2005; Maymon et al., 2009). The effect of GA on the CK response has been shown to be independent of DELLA (Maymon et al., 2009). CK, however, has no effect on GA responses (Greenboim-Wainberg et al., 2005).

In the present study, we analyzed the interaction between GA and CK in tomato plants. We examined the effect of the hormones and their interaction on various processes previously studied in Arabidopsis (Greenboim-Wainberg et al., 2005), as well as a process unique to tomato: the control of leaf complexity.

Materials and Methods

Plant material and genetics

Tomato (Solanum lycopersicum) seedlings and plants were in the M82 (sp–) background. procera (pro) was introgressed into the M82 background (five times) and the GA biosynthesis mutant gib1 (Bensen & Zeevaart, 1990; Jasinski et al., 2008) was in the ‘Moneymaker’ background. The transactivated lines (Moore et al., 1998) pFIL>>IPT7 and pFIL> CKX3 express the CK biosynthetic gene ISOPENTENYL TRANSFERASE7 from Arabidopsis (AtIPT7) and the CK-degrading gene CK OXIDASE/DEHYDROGENASE3 (AtCKX3), respectively, under the regulation of the leaf-specific promoter FIL (Shani et al., 2010). pFIL>>AtCKX3 was introgressed by crosses into the gib1 background. pFIL>>AtIPT7 plants are not fertile, and therefore plants homozygous for the driver transgene pFIL-LhG4 or the responder transgene OP:AtIPT7 were each separately crossed to pro. F2 plants homozygous for the pro mutation with the OP:IPT7 or pFIL:LhG4 transgenes were crossed to each other to generate pFIL>>IPT7 in pro. The plants were grown in a controlled glasshouse at 25 : 18°C (day : night) under natural day length and light intensity of c. 450 μmol m−2 s−1. For the germination of gib1, seed coats were cut under a binocular and the seeds were placed in Petri dishes containing filter paper soaked with 10 μM GA3.

Hormone treatments

GA3, 6-benzylaminopurine (BA), kinetin and zeatin (Sigma-Aldrich, St. Louis, MO, USA), all with the addition of the surfactant Tween 20 (100 μl l−1), were applied to plants either by spraying or by immersing seedlings/young plants.

Anthocyanin extraction and measurements

Seedlings were weighed and anthocyanin was extracted and measured spectrophotometrically (Weiss & Halevy, 1989), and the results were normalized to fresh weight.

RNA extraction and analyses

RNA extraction and cDNA synthesis  Total RNA was isolated from tomato seedlings. Frozen tissues were ground, resuspended in guanidine HCl and phenol/chloroform was added. Samples were mixed by vortexing for 30 s and, after 30 min, were centrifuged at 4°C for 45 min. Ethanol (100%) and 1 M acetic acid were added, mixed and stored overnight at −80°C. Sodium acetate (3 M) was added and samples were washed with cold 70% ethanol. For the synthesis of cDNA, we used the Verso cDNA kit (ABgene, Epsom, Surrey, UK) and 3 μg of total RNA, according to the manufacturer’s instructions.

Semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR)  RT-PCR was performed with 300 ng of total RNA, using Superscript II RT enzyme (Invitrogen, Carlsbad, CA, USA) and polyT primer, according to the manufacturer’s instructions. For PCR with Tomato Response Regulators (TRRs), the following primers were used: TRR8/9a, forward 5′-TGCATCTCTGAGGAACATAC-3′; reverse 5′-TCTTCTATGGCCTTTCTCTTG-3′; TRR8/9b, forward 5′-GCTTGAAGTAAGAGAAACGGG-3′; reverse 5′-CTGGTTCTGTCAGGTGAAATG-3′; TRR3/4, forward 5′-AGAGTTACATGTTCAACGAGGATCG-3′; reverse 5′-TCACAAATCTTGGCTGCTCA-. For TUBULIN, the forward primer 5′-CACATTGGTCAGGCCGGTAT-3′ and reverse primer 5′-GACACATCAGTGTGCTCAGT-3′ were used.

Quantitative RT-PCR (qRT-PCR) analyses  RT-PCR analysis was performed using the SYBR Premix Ex Taq II (RR081Q) kit (Takara Bio Inc., Shiga, Japan). Reactions were performed using a Rotor-Gene 6000 cycler (Corbett Research, Sydney, Australia). A standard curve was obtained for each gene using dilutions of a cDNA sample. Each gene was quantified using the Corbett Research Rotor-Gene software. At least three independent technical repeats were performed for each cDNA sample. The relative expression of each sample was calculated by dividing the expression level of the analyzed gene by that of TUBULIN. Gene-to-TUBULIN ratios were then averaged and presented as a proportion of the control treatment, set to a value of unity. The following primers were used: for the analysis of GA-stimulated transcript 1 (GAST1), forward 5′-CGTACCGGTGTTCAAAGACA-3′; reverse 5′-AGCAAGGGCAACTTTGCTTA-3′; for TRR8/9a, forward 5′-TGCTTAGAAGAAGGGGCAGA-3′; reverse 5′-GGGGGCTTTTACATTTGGTT-3′; for TRR8/9b, forward 5′-AGTATGCCGGAAATGACTGG-3′; reverse 5′-TGGAACATTTTCCGATGACA-3′; for TRR3/4, forward 5′-CGTCCCCTAAAGCATTCTCA-3′; reverse, 5′-CGTCTTGTTGGTGATGTTGG-3′; for TUBULIN, forward 5′-CACATTGGTCAGGCCGGTAT-3′; reverse 5′-ATCTGGCCATCAGGCTGAAT-3′.

Results

Reciprocal negative interaction between GA and CK in tomato seedlings

CK promotes anthocyanin accumulation in Arabidopsis seedlings, and GA application or mutation in SPY suppresses this effect (Greenboim-Wainberg et al., 2005). To test whether similar antagonistic effects exist in tomato, tomato seedlings were treated with the CK BA, GA3 or both. CK promoted and GA suppressed anthocyanin accumulation and, when the two hormones were applied together, GA completely suppressed the effect of CK, such that the level of the pigment was comparable with that found following GA treatment alone (Fig. 1a). These results show that the two hormones exert antagonistic effects on anthocyanin accumulation in tomato seedlings.

Figure 1.

Gibberellin (GA) and cytokinin (CK) affect anthocyanin accumulation and hypocotyl length antagonistically. Tomato seedlings (M82) were sprayed three times a week with 10 μM 6-benzylaminopurine (BA) or GA3 or both. After 2 wk, the anthocyanin concentration was quantified (a) and the hypocotyl length was measured (b). Values in (a) are the means of four biological repeats, each containing 10 seedlings, ± SE. Values in (b) are the means of 40 seedlings, ± SE.

We also tested the effect of the aforementioned treatments on hypocotyl elongation. Although CK treatments had no effect on final hypocotyl length, GA treatment increased it (Fig. 1b). On co-application of the two hormones, CK antagonized the effect of GA. That CK had no effect by itself, but repressed the effect of GA, implies an interaction rather than an additive effect of the two hormones. These results also suggest that, unlike Arabidopsis, where CK has no effect on GA responses (Greenboim-Wainberg et al., 2005), in tomato there is a reciprocal negative interaction, where GA inhibits CK responses and CK inhibits GA responses.

CK inhibits GA suppression of leaf complexity and serration

Tomato plants have compound leaves with highly serrated leaflets, and both complexity and serration are suppressed by GA (Hay et al., 2002). Recently, we have shown that CK is required for leaf complexity and serration in tomato (Shani et al., 2010). To test the interaction between GA and CK with respect to leaf shape, tomato plants were treated with a range of CK and GA concentrations (1–1000 μM BA and 0.01–100 μM GA3). Treatments started at the seedling stage and lasted until the fifth true leaf was fully expanded. Exogenous BA application at all tested concentrations had no effect on leaf serration or complexity (data not shown). Similarly, treatments with other CKs (trans-zeatin and kinetin) had no effect on leaf morphology (Supporting Information Fig. S1). GA treatments (high and low concentrations) resulted in less complex leaves with smooth margins (Figs 2a,b, S2). The complexity and margin serration of 1 μM GA3-treated leaves were similar to those found in 100 μM GA3-treated leaves. When BA (100 μM) was combined with a high GA3 concentration (100 μM), leaves were simple and smooth, as in the GA treatment alone (Fig. 2a). However, when BA (100 μM) was applied together with a low GA3 concentration (1 μM), it suppressed the effect of the latter and leaves were complex with serrated margins (Figs 2b, S3). These results suggest that the ratio between the hormones determines the final response.

Figure 2.

The ratio between gibberellin (GA) and cytokinin (CK) determines their effect on tomato leaf shape. Tomato seedlings (M82) were immersed three times a week in 100 μM 6-benzylaminopurine (BA), 100 μM GA3 or both (a), or in 100 BA, 1 μM GA3 or both (b). Treatments lasted until the fifth true leaf was fully expanded. Representative leaflets taken from the same position within the fifth leaves are presented. Bars, 1 cm.

We next tested whether GA can inhibit the activity of endogenous CK on leaf morphology. To this end, we applied exogenous GA3 to transgenic transactivated tomato plants expressing the CK biosynthetic gene AtIPT7 (Miyawaki et al., 2006; Werner & Schmulling, 2009) under the regulation of the leaf-specific promoter FIL (pFIL>>AtIPT7). Expressing AtIPT7 in a wild-type (WT) background increased leaf complexity but, similar to exogenous CK, it had no or a minor effect on leaflet serration (Fig. 3; Shani et al., 2010). Treatments with low GA3 concentration (1 μM) partially suppressed leaf complexity and serration in WT and transgenic plants, and the inhibitory effect increased when a higher concentration of GA3 (100 μM) was used.

Figure 3.

 Exogenous gibberellin (GA) inhibits the effect of cytokinin (CK) biosynthetic gene ISOPENTENYL TRANSFERASE7 (AtIPT7) overexpression on leaf complexity. Transgenic pFIL>>AtIPT7 tomato plants were sprayed three times a week for 6 wk with 1 or 100 μM GA3. The ninth leaves and leaflets (taken from the same positions) from these leaves are presented. WT, wild-type. Bars, 1 cm.

Recently, we have shown that decreasing the CK contents in tomato leaves by overexpressing the CK-degrading gene CKX3 strongly reduces leaf complexity and serration (Shani et al., 2010). In the light of the current finding that CK represses the effect of GA on leaf shape, it is possible that the effect of reduced CK levels on leaf complexity is mediated by enhanced GA activity. To test this possibility, we expressed the CKX3 gene under the regulation of the leaf-specific promoter FIL (Lifschitz et al., 2006; Shani et al., 2009) in the background of the GA-deficient mutant gib1 (Bensen & Zeevaart, 1990; Jasinski et al., 2008), using the LhG4 transactivation system. Leaves of the GA-deficient gib1 mutant were much smaller than control leaves, but the complexity and serration were similar to WT (Fig. 4c). However, leaves of CKX3-overexpressing plants were much simpler and smoother (Fig. 4b). When CKX3 was expressed in the gib1 background, leaves were small but still simple and smooth (Fig. 4d). These results suggest that a lack of CK affects leaf morphology directly and not via increased GA activity.

Figure 4.

Leaf serration and complexity are promoted directly by cytokinin (CK) activity and not by CK-induced inhibition of gibberellin (GA) activity. Leaf complexity and leaflet serration in tomato wild-type (WT) M82 (a), transgenic M82 plant expressing CK-degrading enzyme CK OXIDASE/DEHYDROGENASE3 (CKX3) from Arabidopsis under the regulation of the leaf-specific promoter FIL in transactivated lines (pFIL>>AtCKX3; b), the GA-deficient gib1 mutant (c) and gib1 expressing CKX3 (pFIL>>AtCKX3 gib1; d). The fifth leaves are presented. Bars, 1 cm.

The antagonistic effect of GA and CK on gene expression

To begin addressing the mode of interaction between GA and CK, we examined the possible antagonistic effect of the two hormones on the expression of CK- and GA-induced genes. In Arabidopsis, CK induces the transcription of Type A ARR genes (Imamura et al., 1998; D’Agostino et al., 2000). Tomato database searches revealed several Type A ARR homologous genes that we named TRR. We have selected for this study three of them (numbered according to the most similar respective Arabidopsis homologs): TRR3/4 (AK326842), TRR8/9a (AC215374) and TRR8/9b (AK246560) (Shani et al., 2010). To test the effect of the hormones on the expression of these genes, tomato seedlings were immersed in BA, GA3 or both, for 50 min, followed by RNA extraction. RT-PCR analysis revealed that the expression of all three TRR genes was promoted by CK (Fig. 5). GA3 treatment slightly repressed the expression of TRR8/9a and TRR8/9b, but had no effect on TRR3/4. When GA3 was applied together with BA, the promoting effect of the latter on all three genes was suppressed. To confirm these results, we repeated the experiment four times and also performed qRT-PCR with the RNA samples of one of these experiments (Fig. S4), all with similar results.

Figure 5.

Gibberellin (GA) inhibited cytokinin (CK)-induced Tomato Response Regulator (TRR) expression. Tomato seedlings (M82) with two true leaves were immersed in 10 μM 6-benzylaminopurine (BA), GA3 or both, for 50 min, followed by RNA extraction and semi-quantitative RT-PCR. The expression of TUBULIN was used as a control to ensure that equal amounts of cDNA were used in all reactions. Shown is a representative experiment from four experiments with comparable results.

In tomato, but not in Arabidopsis, CK inhibited the effect of GA on various developmental processes (Figs 1b, 2b). We thus tested whether CK also suppressed the expression of the GA-induced gene GAST1 (Shi & Olszewski, 1998). Tomato seedlings (M82) were treated with paclobutrazol to reduce the level of endogenous GAs and then treated with BA, GA3 or both. qRT-PCR analysis showed that GA3 promoted GAST1 expression (Fig. 6), but CK did not suppress this effect of GA.

Figure 6.

Cytokinin (CK) did not inhibit gibberellin (GA)-induced GA-stimulated transcript (GAST1) expression. Tomato seedlings (M82) were treated with 10 mg l−1 paclobutrazol to reduce the level of endogenous GAs. After 2 wk, the seedlings were immersed in 10 μM of 6-benzylaminopurine (BA), GA3 or both for 50 min, and left for an additional 4 h (time required for maximum GAST1 expression after GA induction; D. Weiss, unpublished data) before RNA was extracted. Expression was analyzed by quantitative RT-PCR (qRT-PCR). The relative expression of each sample was calculated by dividing the expression level of the analyzed gene by that of TUBULIN. Gene-to-TUBULIN ratios were then averaged and presented as a proportion of the control treatment, set to a value of unity. Values are the means of three biological replicates, ± SE.

Mutation in DELLA has no effect on CK responses

Although GA suppresses CK activities in Arabidopsis, quadruple DELLA mutants (loss of four of five DELLA genes) with constitutive GA signaling had no effect on CK responses (Maymon et al., 2009). It was proposed that GA affects CK via a DELLA-independent pathway, although it is still possible that the effect is mediated via the fifth DELLA gene. Several studies have suggested that tomato has only one DELLA gene, named PROCERA (PRO) (Jasinski et al., 2005; Marti et al., 2007; Bassel et al., 2008). We have searched the recently released tomato genome sequence (http://solgenomics.net) for DELLA-like genes and found only PRO. The pro mutant exhibits a GA-independent constitutive GA signal (Bassel et al., 2008). To examine whether pro can inhibit CK responses similar to GA, we tested the effect of CK and GA on anthocyanin accumulation in pro seedlings. Seedlings were treated with 10 μM BA, GA3 or both. The initial level of anthocyanin found in pro seedlings was lower than that found in WT (Fig. 7 vs Fig. 1). However, CK promoted the accumulation of the pigment in pro seedlings, and GA strongly repressed it, similar to their effect in WT. This suggests that DELLA has a role in anthocyanin accumulation in seedlings, but the inhibitory effect of GA on CK-induced anthocyanin accumulation is probably via a DELLA-independent pathway.

Figure 7.

Mutation in PROCERA (PRO) did not inhibit cytokinin (CK)-induced anthocyanin accumulation. pro seedlings (backcrossed into the M82 background) were sprayed three times a week, for 2 wk, with 10 μM 6-benzylaminopurine (BA), GA3 or both, followed by anthocyanin extraction and measurement. Values are the means of four replicates, each containing 10 seedlings, ± SE.

We also tested whether pro, similar to GA, inhibits CK-induced TRR gene expression. M82 and pro seedlings were immersed in 10 μM BA or in water for 50 min, followed by RNA extraction. RT-PCR analysis showed similar expression of TRR3/4, TRR8/9a and TRR8/9b in M82 and pro seedlings (Fig. S5). Furthermore, the expression of all three TRR genes was promoted by CK to similar levels in M82 and pro seedlings. To confirm these results, we repeated the experiment three times and similar results were obtained. These results support our hypothesis that GA suppresses CK responses via a DELLA-independent pathway.

We also tested the effect of pro on endogenous CK activity. To this end, we compared the effect of expression of AtIPT7 under the FIL promoter (pFIL>>AtIPT7) between WT (M82) and pro plants. The increased complexity induced by AtIPT7 was not affected by pro and was similar to that found in WT expressing AtIPT7 (Fig. 8), suggesting that the GA signal generated by the loss of PRO has no effect on CK-induced leaf complexity. This is different from the results with exogenous GA, which strongly suppressed the effect of AtIPT7 on leaf complexity at high and low concentrations (Fig. 4).

Figure 8.

Mutation in PROCERA (PRO) did not inhibit the effect of endogenous cytokinin (CK) on leaf complexity. Leaf and leaflet complexity in M82 (a, e), transgenic M82 plant expressing the CK-biosynthetic gene ISOPENTENYL TRANSFERASE7 from Arabidopsis (AtIPT7) under the regulation of the leaf-specific promoter FIL in transactivation lines (pFIL>>AtIPT7; b, f), pro (c, g) and transgenic pro expressing AtIPT7 (pFIL>>AtIPT7 pro; d, h). The fifth leaves and leaflets (taken from the same positions) from these leaves are presented. Bars, 1 cm.

Our results suggest that the loss of PRO has no effect on CK activity. However, pro leaves are simpler and smoother than those of WT tomato. It is possible that the GA signal in pro is not sufficiently strong to suppress the high CK activity resulting from the exogenous application of the hormone or from IPT overexpression, but sufficient to inhibit the normal endogenous CK activity. If this is the case, exogenous application of CK to pro should suppress the effect of pro on leaf complexity and serration. We repeatedly treated pro plants with high CK concentration (100 μM), but with no effect on leaf morphology (Fig. 9), in contrast with the repressing effect of exogenous CK when co-applied with low GA3 concentration (Figs 2b, S3). This suggests that pro reduces leaf complexity and serration via a CK-independent pathway, and that GA affects leaf complexity via both a CK-dependent and CK-independent pathway.

Figure 9.

Exogenous cytokinin (CK) application does not promote leaf complexity and serration in procera (pro). pro and wild-type (WT, M82) tomato seedlings were sprayed three times a week, for 4 wk, with 100 μM 6-benzylaminopurine (BA). The fifth leaves are presented. Bars, 1 cm.

Discussion

Several recent studies have shown development-dependent interactions between GA and CK in Arabidopsis (Greenboim-Wainberg et al., 2005; Jasinski et al., 2005; Maekawa et al., 2009). The results of the present study have revealed complex mutual antagonistic interaction between these hormones in various molecular and developmental processes in tomato. This mutual interaction, however, is different from that found in Arabidopsis (Greenboim-Wainberg et al., 2005). In Arabidopsis, GA inhibits CK responses, but CK has no effect on GA responses, whereas, in tomato, some of the tested GA responses are clearly suppressed by CK. In some processes, for example leaf complexity and serration, and seedling elongation, CK had no effect by itself but suppressed the effect of GA, and, in other processes, for example TRR3/4 expression, GA had no effect but suppressed the effect of CK. Therefore, the mutual inhibition is unlikely to be an additive effect, but may represent a negative interaction between the two hormone response pathways. It is not known which steps in the signaling cascades are affected by the interaction. As GA inhibited all tested CK responses, including the induction of the CK primary response genes, type A TRRs, it is possible that GA affects the early phosphorylation cascade (Muller & Sheen, 2007) of the CK response pathway. However, CK did not affect all GA responses (i.e. it had no effect on GA-induced GAST1 expression), and thus CK may act on downstream branch(es) of the GA signaling pathway, or affect the same processes through a parallel pathway.

The results of the present study imply that the ratio between the two hormones determines the final response. Treatment of tomato leaves with high or low GA levels affected leaf morphology and resulted in smooth and simpler leaves. When high CK concentration was applied together with a high GA level, it had no effect, but when it was applied together with low GA levels, it completely suppressed the effect of the latter and promoted leaf complexity and serration.

We have recently shown that CK is required for leaf complexity and serration in tomato (Shani et al., 2010). In this study, however, we found that the exogenous application of CK had no effect on these processes. It is possible that leaf complexity and serration require CK at a specific, early developmental stage and/or cell type. Thus, application of the hormone to the entire plant, throughout development, cannot affect these processes. The promoting effect of CK on leaf complexity and serration was found, however, when co-applied with GA. Therefore, in the context of the interaction between the two hormones, endogenous and exogenous CK had the same effect.

In Arabidopsis, SPY seems to play an important role in the interaction between the GA and CK signaling pathways (Weiss & Ori, 2007). The role of this protein in the interaction between the hormones in tomato plants is not known. Although the tomato genome contains an SPY homologous sequence (Greb et al., 2002), mutations in this locus have not yet been identified. Although the current model suggests that GA and SPY regulate plant development via the GA response suppressors DELLA proteins (Silverstone et al., 2007; Harberd et al., 2009), our previous results have suggested that, in Arabidopsis, the effects of GA and SPY on CK activity are not mediated by DELLAs (Maymon et al., 2009). However, the presence of multiple DELLA proteins in Arabidopsis and the lack of complete null mutants for all DELLA genes make it difficult to reach a definitive conclusion about the relationship between GA, SPY, DELLA and CK. In tomato, only one DELLA gene, PRO, was identified. Previous studies have used various approaches, such as low-stringency hybridization to tomato genomic DNA blots and screens of genomic and cDNA libraries, to search for additional DELLA genes, and all have suggested that PRO is the only DELLA protein of tomato (Bassel et al., 2004; Marti et al., 2007). Furthermore, we searched the recently released tomato genome sequence available in the SOL Genomics Network (SGN) database (http://solgenomics.net), but did not detect any additional DELLA sequences. We therefore took advantage of the probably single DELLA gene (PRO) of tomato to determine whether GA acts via a DELLA-dependent pathway to suppress CK responses. The results showed that exogenous application of GA, but not an elevated GA signal generated by the loss of PRO, suppressed CK-induced anthocyanin accumulation and TRR gene expression in seedlings and the enhanced leaf complexity in plants expressing the CK biosynthetic gene AtIPT7, suggesting that GA suppresses CK via a DELLA-independent pathway.

pro may be a ‘leaky’ mutant with partial PRO activity (Van Tuinen et al., 1999; Bassel et al., 2008). Thus, it is possible that the GA signal in pro is mild because of partial PRO activity. This mild signal might be too ‘weak’ to suppress the high CK activity generated by exogenous application of CK or by AtIPT7 overexpression. The fact that pro leaves are simpler and smoother than those of WT may suggest that this ‘weak’ signal is sufficient to suppress endogenous CK activity. Our results, however, do not support these possibilities. If the GA signal in pro is, indeed, too ‘weak’ to suppress high CK activity, the application of high exogenous CK concentrations to pro would be expected to inhibit its leaf phenotype. However, CK application did not affect pro leaf morphology. This suggests that the effect of pro on leaf complexity and serration does not involve the inhibition of endogenous CK activity. Furthermore, the slender growth habit of pro suggests a strong endogenous GA signal. Taken together, the results suggest that exogenous GA suppresses CK responses at least partially via a DELLA-independent pathway, as suggested for the interaction between the two hormones in Arabidopsis (Maymon et al., 2009).

The antagonistic effects between GA and CK have been found in numerous plant species. For example, CK promotes and GA inhibits nodule formation in legumes (Hirsch et al., 1989; Tirichine et al., 2007; Maekawa et al., 2009), tuberization in potato (Rodriguez-Falcon et al., 2006) and flower formation in grapes (Srinivasan & Mullins, 1981). Thus, the negative interaction between GA and CK and the mechanism behind it might be conserved, with minor differences, in many, if not all, plant species. It is possible that this mechanism has evolved to flexibly tune the balance between meristem activity, where high CK and low GA levels are required (Jasinski et al., 2005; Bolduc & Hake, 2009), and cell maturation, where high GA and low CK levels are needed (Greenboim-Wainberg et al., 2005).

Acknowledgements

We thank Thomas Schmülling and Tomáš Werner for the AtCKX3 construct and Yuval Eshed for the pro mutant. This research was supported by research grants from The Israel Science Foundation (ISF grant no. 253-06 to D.W. and N.O. and 60/10 to N.O.). This work was also supported by the Pearlstein Fund for research in horticulture at the Hebrew University; we thank the donors for their help.

Ancillary