In the development of tomato compound leaves, local auxin maxima points, separated by the expression of the Aux/IAA protein SlIAA9/ENTIRE (E), direct the formation of discrete leaflets along the leaf margin. The local auxin maxima promote leaflet initiation, while E acts between leaflets to inhibit auxin response and lamina growth, enabling leaflet separation. Here, we show that a group of auxin response factors (ARFs), which are targeted by miR160, antagonizes auxin response and lamina growth in conjunction with E. In wild-type leaf primordia, the miR160-targeted ARFs SlARF10A and SlARF17 are expressed in leaflets, and SlmiR160 is expressed in provascular tissues. Leaf overexpression of the miR160-targeted ARFs SlARF10A, SlARF10B or SlARF17, led to reduced lamina and increased leaf complexity, and suppressed auxin response in young leaves. In agreement, leaf overexpression of miR160 resulted in simplified leaves due to ectopic lamina growth between leaflets, reminiscent of e leaves. Genetic interactions suggest that E and miR160-targeted ARFs act partially redundantly but are both required for local inhibition of lamina growth between initiating leaflets. These results show that different types of auxin signal antagonists act cooperatively to ensure leaflet separation in tomato leaf margins.
Leaves are the main photosynthetic organs of most plants. Diverse leaf shapes and sizes are produced by different plant species, within the same species and by individual plants. Compound leaves are composed of multiple subunits termed leaflets (Figure 1a). Tomato (Solanum lycopersicum) plants have compound leaves with one or two orders of leaflets. Tomato shows a wide diversity of leaf sizes and shapes, flexibly responding to environmental changes (Holtan and Hake, 2003; Menda et al., 2004; Brand et al., 2007; Chitwood et al., 2012, 2013). The elaboration of compound leaves depends on prolonged organogenic activity and delayed maturation of the developing leaf, mainly in its margin. Patterning at the leaf margin has been shown to involve differential spatial and temporal lamina growth and to be regulated by an array of transcription factors and hormones (Hagemann and Gleissberg, 1996; Byrne et al., 2000; Dengler and Tsukaya, 2001; Kaplan, 2001; Barkoulas et al., 2007; Blein et al., 2010; Efroni et al., 2010; Floyd and Bowman, 2010; Koenig and Sinha, 2010; Moon and Hake, 2010; Bar and Ori, 2015; Sluis and Hake, 2015).
The plant hormone auxin has been shown to be involved in the patterning of leaf margins in species with both simple and compound leaves (Barkoulas et al., 2008; DeMason and Polowicky, 2009; Koenig et al., 2009; Blein et al., 2010; Bilsborough et al., 2011; Peng and Chen, 2011; Zhou et al., 2011; Ben-Gera et al., 2012). The most studied auxin response pathway involves receptors from the TIR1/AFB family, auxin response inhibitors from the Aux/IAA family, and ARF transcription factors (Salehin et al., 2015). ARF proteins are transcriptional activators or repressors that regulate the expression of auxin-responsive genes by binding to auxin response elements (AuxREs) in their promoters (Ulmasov et al., 1997; Tiwari et al., 2003; Chandler, 2016). Aux/IAA proteins bind activator ARFs and repress their function. The TIR/AFB receptors are F-box proteins that promote the degradation of Aux/IAAs (Gray et al., 2001; Dharmasiri et al., 2005). TIR/AFBs and Aux/IAAs act as co-receptor complexes (Calderon Villalobos et al., 2012). In the absence of auxin, Aux/IAAs bind activator ARFs and repress their activity, together with co-repressor proteins such as TOPLESS (Szemenyei et al., 2008). Repression by Aux/IAAs involves recruitment of a histone deacetylase complex as well as prevention of the recruitment of SWI/SNF chromatin-remodeling ATPases by activating ARFs (Wu et al., 2015). Auxin binding to the co-receptor complex leads to the degradation of the Aux/IAA and the resulting activation of the ARF (Kepinski and Leyser, 2004; Weijers and Jurgens, 2004; Dharmasiri et al., 2005). While this basic model of auxin response is compatible with the activity of activator ARFs, the mode of activity of repressing ARFs is still not fully understood, and most repressor ARFs do not bind Aux/IAAs in yeast two-hybrid assays (Tiwari et al., 2003; Vernoux et al., 2011; Guilfoyle and Hagen, 2012). TIR1/AFBs, Aux/IAA and ARFs belong to gene families, enabling diverse auxin outputs (Okushima et al., 2005; Kieffer et al., 2010; Rademacher et al., 2011, 2012; Vernoux et al., 2011). Most ARF proteins contain a DNA-binding domain (DBD), a middle region (MR) that acts as an activation (AD) or a repression domain (RD), and a carboxy-terminal dimerization domain (CTD), which is involved in protein–protein interactions (Guilfoyle and Hagen, 2007). Recently, a dimerization domain was identified within the DBD (Boer et al., 2014). ARF proteins show diverse structures, for example, several ARFs lack the CTD. Five of the 23 Arabidopsis ARFs have been shown in protoplast assays to act as transcriptional activators (Ulmasov et al., 1999; Tiwari et al., 2003). The remaining ARFs are thought to function as repressors, but this has been shown only for a few of them. In tomato, a recent study identified eight ARFs as repressors and five as activators (Zouine et al., 2014).
In developing compound leaf primordia, the juxtaposition of an auxin maxima and a region with low auxin response is thought to promote leaflet initiation and separation. The tomato Aux/IAA gene ENTIRE (E) is expressed between initiating leaflets, where it inhibits auxin response and promotes leaflet separation. Mutant e leaves are simplified in comparison with wild-type leaves due to expanded auxin response and ectopic lamina growth between leaflets (Koenig et al., 2009; Ben-Gera et al., 2012).
NAM transcription factors were also shown to be involved in marginal patterning in several species (Nikovics et al., 2006; Blein et al., 2008; Bilsborough et al., 2011; Cheng et al., 2012). The tomato NAM transcription factor GOBLET (GOB) is expressed in leaflet boundaries and promotes leaflet specification and separation (Brand et al., 2007; Berger et al., 2009; Blein et al., 2010). NAM proteins, E and auxin interact to pattern the leaf margin by complex feedback interactions, the details of which are species specific (Hasson et al., 2010; Bilsborough et al., 2011; Ben-Gera et al., 2012).
Ectopic expression of the tomato ARF protein SlARF10A, which is targeted by miR160, was recently shown to result in very narrow leaflets (Hendelman et al., 2012). The related group of miR160-targeted ARFs includes three genes in Arabidopsis, ARF10, ARF16 and ARF17, which have been shown to be involved in an array of developmental processes (Mallory et al., 2005; Wang et al., 2005; Liu et al., 2007, 2010, 2013). In soybean, proper regulation of miR160 on ARF10, ARF16 and ARF17 was found to be necessary for nodule formation (Nizampatnam et al., 2015). In tomato, five related genes are predicted to contain a miR160-binding site, SlARF10A, SlARF10B, SlARF17, SlARF16A, SlARF16B (nomenclature according to Zouine et al., 2014). SlARF10A and SlARF10B contain all the ARF domains, while SlARF17 and SlARF16B lack the CTD. SlARF10A and SlARF17 were recently shown to act as repressors in protoplast activity assays, using the DR5 promoter fused to GFP as the reporter (Zouine et al., 2014). However, Arabidopsis ARF17 and ARF16 were shown to positively regulate the auxin response sensor DR5 in pollen and root cap, respectively (Wang et al., 2005; Yang et al., 2013), and ARF10 and ARF16 were shown to positively regulate ABI3, although not directly (Liu et al., 2013). It is therefore still unclear whether miR160-targeted ARFs act as activators, repressors or both in different developmental contexts.
Here, we investigated the role of miR160-targeted ARFs in tomato leaf development by ectopic expression of either pre-AtmiR160A, or miR160-resistant forms of miR160-targeted ARFs. We show that leaves that overexpress miR160-resistant forms of SlARF10A, SlARF10B and SlARF17 are more complex and their leaflets have reduced lamina. Furthermore, ectopic expression of miR160 in tomato leaves leads to an expanded auxin response and simpler leaves than the wild-type. Genetic and molecular analyses of the interaction between SlARF10A and E indicate that they act through parallel genetic pathways and do not interact physically, but also depend on each other for their function in lamina growth and leaflet separation. Double mutant analysis shows that when GOBLET is not active SlARF10A activity fails to promote leaflet separation.
Overexpression of miR160-targeted ARFs inhibits lamina growth and promotes leaf complexity
Previously, ubiquitous overexpression of a SlARF10A form with a mutation in the miR160-binding site was shown to lead to substantially reduced leaflet lamina in tomato (Hendelman et al., 2012). Examination of available transcriptome databases (Park et al., 2011; http://tomatolab.cshl.edu/efp/cgi-bin/efpWeb.cgi) indicated that of the five miR160-targeted ARFs, SlARF10A has the highest expression in the vegetative meristem and young leaf primordia, while SlARF10B, SlARF16A and SlARF17 have lower expression, and SlARF16B is not expressed in this tissue. To assess the specific effect of SlARF10A on leaf development, and the differential function of additional miR160-targeted ARFs, we ectopically overexpressed SlARF10A, as well as SlARF10B and SlARF17 with mutations in the miR160-binding site (SlARF10Am, SlARF10Bm and SlARF17m, respectively) (Figure S1) specifically at different stages of tomato leaf development using the transactivation system (Moore et al., 1998). The Arabidopsis pFIL promoter drives expression throughout young developing leaves, and the pBLS promoter drives expression in developing leaflets from the P4 stage. We have previously used these promoters to investigate leaf development (Lifschitz et al., 2006; Shani et al., 2009). Leaf overexpression of SlARF10Am, SlARF10Bm and SlARF17m under the control of the pFIL promoter or the pBLS promoter led to reduced leaf lamina and increased leaf complexity (Figures 1b–h, S2 and S3a). To quantify the reduction in lamina width, we calculated the ratio between the lamina area and the leaf perimeter in the fifth leaf of the different genotypes. pFIL>>SlARF10Am, pFIL>>SlARF10BmpFIL>>SlARF17m and pBLS>>SlARF10Am leaves had an approximately 50% reduction in this ratio (Figure S3b), and a reduction of approximately 28% was found in pBLS>>SlARF10Bm and pBLS>>SlARF17m (Figure S3b). Quantification of leaf complexity showed that in comparison with wild-type leaves, leaves of plants that overexpressed SlARF10Am, SlARF10Bm or SlARF17m had more secondary and intercalary leaflets and, unlike the wild-type, all had tertiary leaflets. pFIL>>SlARF10Am leaves occasionally also had a fourth order of leaflets (Figure S3c). The phenotypes caused by expression with the pFIL promoter were more severe than those caused by the expression with the pBLS promoter (Figures 1 and S3). SlARF10Am overexpression showed the strongest influence on lamina width and leaf complexity. Analysis of the endogenous expression of SlARF10A mRNA during wild-type leaf development showed dynamic expression, with elevated expression at the P4 and P9 stages, possibly coinciding with primary and secondary leaflet initiation (Figure S4). These results suggest that SlARF10A, SlARF10B and SlARF17 can all affect lamina width and leaf complexity when overexpressed ectopically in developing leaves. The differences in their overexpression phenotypes suggested that they have somewhat differential activities and expression. To analyze the spatial distribution of SlARF10A and SlARF17 in wild-type leaf primordia, we performed in situ hybridization experiments. SlARF10A and SlARF17 were found to be expressed in the margins of developing leaf primordia, with strong expression in young leaflets. Expression is observed throughout very young initiating leaflets, and later becomes restricted to the leaflet margin (Figure 2a–f). SlARF10A expression is expanded in pFIL>>SlARF10Am leaf primordia (Figure S5b). The expression dynamics of miR160-targeted ARFs, together with the phenotypes caused by their ectopic expression, suggests that miR160-targeted ARFs act in a somewhat quantitative manner to restrict lamina growth during early leaf development.
SlARF10A and SlARF17 suppress the effect of auxin on lamina growth
Auxin was found to be a positive regulator of lamina expansion (Barkoulas et al., 2008; Koenig et al., 2009; Ben-Gera et al., 2012). Belonging to the ARF family, SlARF10A and SlARF17 are putative mediators of the auxin signal. To examine the involvement of SlARF10A and SlARF17 in the regulation of auxin response, we analyzed the sensitivity of pFIL>>SlARF10Am and pFIL>>SlARF17m to auxin. Wild-type and pFIL>>SlARF10Am plants were grown on MS medium with or without 2,4-dichlorophenoxyacetic acid (2,4-D). As shown previously, external 2,4-D application to wild-type plants resulted in a simplified leaf form, with wider and smoother leaflets and increased occurrence of ectopic lamina, compared with untreated plants (Figure 3a,c). In pFIL>>SlARF10Am plants the effect of auxin was much milder than in the wild-type, as the treated leaves essentially maintained the narrow lamina and deep lobing phenotype of the pFIL>>ARF10Am leaves (Figure 3b,d). This finding suggests that SlARF10A is involved in the regulation of auxin response in the leaf, but that additional players may also regulate this process. To test the effect of endogenous auxin, we generated plants that co-express the bacterial auxin biosynthesis gene tryptophan monooxygenase (iaaM) and SlARF17m under the control of the FIL promoter. iaaM overexpression causes an increase in endogenous free IAA levels (Romano et al., 1995). pFIL>>iaaM leaves are simpler than wild-type leaves, showing mainly primary leaflets and severe reduction in secondary leaflets (Figure 3g) (Ben-Gera et al., 2012). Co-expression of SlARF17m substantially suppressed the iaaM phenotype, with the co-expressing plants having narrower lamina and more complex leaves due to secondary leaflet formation and deep lobes (Figure 3h). Thus, leaves that overexpress SlARF10Am or SlARF17m are less sensitive to auxin than wild-type leaves, indicating that these ARFs repress auxin response in leaf patterning. In addition, SlARF10A and SlARF17 act antagonistically to auxin as they have an opposite effect on lamina expansion and leaf complexity.
miR160-targeted ARFs spatially inhibit auxin response and promote leaflet separation
To further understand the role of miR160-targeted ARFs in leaf development, we overexpressed the Arabidopsis pre-miR160A in leaf primordia. As expected, qRT-PCR expression analysis showed reduced expression of SlARF10A, SlARF16A and SlARF17 mRNA in leaves of pFIL>>miR160 plants (Figure S5). pFIL>>miR160 leaves were simpler than wild-type leaves, as they had only primary leaflets that were often fused due to ectopic lamina growth between them. In addition, the margins of the leaflets were smooth (Figure 4a,b). Examination of early stages of leaf development of pFIL>>miR160 showed lamina growth in the terminal leaflet and between leaflets, leading to improper leaflet separation (Figure S6). pBLS>>miR160 leaves showed round and smooth primary leaflets and no higher-order leaflets. The primary leaflets were properly separated in accordance with the pBLS expression domain (Figure 4c). Interestingly, leaves that overexpressed miR160 showed some phenotypic similarity to mutant e leaves. In plants that co-expressed miR160 and SlARF10Am under the control of pFIL, SlARF10Am overexpression only partially suppressed the miR160 overexpression phenotype (Figure 4d), suggesting that additional miR160-targeted ARFs are necessary for proper leaf patterning, or that the mutated SlARF10A retains some miR160 sensitivity. The effects of ectopic expression of miR160 and miR160-targeted ARFs in leaves indicated that these ARFs promote leaflet separation by inhibiting lamina growth between leaflets. Examination of the spatial distribution of SlmiR160 in wild-type leaf primordia by in situ hybridization experiments revealed that SlmiR160 expression is detected mainly in provascular tissue (Figure 2g,h).
Auxin has been shown to be involved in the specification of leaflets in tomato. Furthermore, the ectopic lamina and lamina expansion seen in pFIL>>miR160 leaves was similar to the phenotype caused by auxin application or increased endogenous auxin in wild-type leaves. To assess whether miR160 targeted ARFs affect auxin response, we expressed pFIL>>SlARF10Am, pFIL>>SlARF17m or pFIL>>miR160 in the background of the auxin response sensor pDR5::VENUS (Figure 5) (Ulmasov et al., 1997; Heisler et al., 2005; Ben-Gera et al., 2012). In wild-type tomato leaves, pDR5::VENUS expression maxima marked the sites of leaflet initiation at the leaf margin, and following initiation its expression became restricted to the leaflet tip and the initiating vascular tissue (Ben-Gera et al., 2012; Figure 5a). In pFIL>>SlARF10Am and pFIL>>SlARF17m leaves, maxima of pDR5::VENUS signal preceded each leaflet or lobe initiation and accumulated at their tips following initiation, in a similar pattern to wild-type (Figure 5b,c). In contrast, in miR160-overexpressing leaves, pDR5::VENUS expression was expanded to throughout the leaf margin (Figure 5d). These results suggest that the miR160-targeted ARFs function to spatially restrict auxin response and lamina generation between leaflets, enabling the formation of distinct leaflets.
ENTIRE and SlARF10A jointly restrict lamina generation between leaflets
The results presented above suggest that miR160-targeted ARFs inhibit auxin response and lamina generation between leaflets and in leaflet margins. A similar role was shown for the Aux/IAA protein ENTIRE (E)/SlIAA9 (Koenig et al., 2009; Ben-Gera et al., 2012). This suggests that, in this case, Aux/IAAs and ARFs have similar rather than antagonistic functions, and raises the question of the relationship between these factors in mediating the effect of auxin in leaf patterning. We addressed this question genetically by combining transgenes that affect E and SlARF10A activity (Figure 6). Overexpression of a stabilized form of E (EdIIm) in leaves alone, or as a fusion with the GUS reporter, leads to smaller leaves with narrower lamina. Interestingly, overexpression of EdIIm and SlARFm both lead to lamina narrowing, but the general appearance of the leaf is different (Koenig et al., 2009; Ben-Gera et al., 2012) (Figure 6b,c). In addition, EdIIm overexpression affects lamina width strongly under both promoters, while SlARFm overexpression leads to a stronger phenotype in early leaf development under the control of the pFIL promoter, and affects lamina width more mildly when expressed at later stages of leaf development under the control of the pBLS promoter. These differences correspond to the different expression domains of these factors in developing leaf primordia, where E expression is stronger between initiating leaflets (Ben-Gera et al., 2012), while SlARF10A and SlARF17 expression is stronger in leaflets. Leaves that co-expressed EdIIm-GUS and miR160 under the control of the FIL promoter, or leaves that co-expressed EdIIm and miR160 under the control of the BLS promoter, had wider leaflets than leaves that overexpressed only EdIIm and were more compound than leaves that overexpressed only miR160 due to distinct primary and secondary leaflets formation (Figures 4 and 6d,e). entire (e) loss-of-function mutants have simplified leaves due to ectopic lamina growth, similar to miR160-overexpressing leaves (Koenig et al., 2009; Ben-Gera et al., 2012; Figure 6a). The e phenotype is almost completely epistatic to that of pFIL>>SlARF10Am (Figure 6f). Thus, the effect of EdIIm is reduced in the background of miR160 and the effect of SlARF10A is suppressed in the background of e. These results suggest that both SlARF10A and ENTIRE partially depend on each other for their function in restricting lamina generation between leaflets. Nevertheless, overexpression of EdIIm partially enabled distinct leaflet formation upon downregulation of miR160-targeted ARFs expression, while SlARF10Am activity could not compensate for the reduction of E activity in this process. Leaves that co-expressed EdIIm and SlARF10Am show a slight enhancement of both phenotypes (Figure 6g). Leaves that co-expressed e and miR160 show an enhanced phenotype of very simple leaves that are composed of a single, continuous blade (Figure 6h), suggesting that E and SlARF10A act via partially separate pathways. Inspection of early development of e pFIL>>miR160 shows continuous lamina forming throughout the leaf margin and lack of initiation of distinct leaflets from very early stages of leaf development (Figure 6i). Cumulatively, these results suggest that E and miR160-targeted ARFs are both required to restrict lamina generation, and act in this process via partly separate pathways.
SlARF10A and ENTIRE do not interact in yeaste
We used yeast two-hybrid assays to examine possible interactions between ENTIRE and miR160-targeted ARFs, as well as among these ARFs. E showed strong interaction with itself and SlARF10A showed a weaker interaction with itself (Figure 7a,b). We did not observe interactions between E and the examined miR160-targeted ARFs (Figure 7). This result is in agreement with previous research suggesting that Aux/IAAs and repressing ARFs do not interact directly in vitro (Vernoux et al., 2011; Guilfoyle and Hagen, 2012). Therefore, while both E and miR160-targeted ARFs repress auxin response and lamina growth between leaflets, and depend on each other for their activity, they likely do not directly interact, or interact via a third partner.
GOBLET mediates SlARF10A activity
Leaflet specification and separation in tomato was also found to be regulated by the NAM transcription factor GOBLET (GOB), which is expressed in leaflet boundaries (Blein et al., 2008; Berger et al., 2009). To understand the relationship between SlARF10A and GOB in leaf development, we examined genotypes with altered activity of both GOB and SlARF10A. Leaves that overexpressed miR164, the negative regulator of GOB, are simplified with only primary leaflets and smooth margins (Berger et al., 2009; Figure 6c). Strikingly, when SlARF10Am was overexpressed in the background of miR164 overexpression under the control of pFIL, the miR164-overexpression phenotype was completely epistatic to overexpression of SlARF10Am (Figure 8e), suggesting that GOB is required for SlARF10A activity. Leaves that overexpressed both a miR164-resistant GOB form (GOBm) and miR160 developed separated primary and secondary round leaflets (Figure 8f), resembling in shape leaflets of GOBm-overexpressing leaves (Figure 8d). Thus, the activity of GOBm was sufficient to rescue the process of leaflet separation even when the activity of miR160-targeted ARFs was down regulated. Leaves that co-expressed GOBm and SlARF10Am show suppression of both phenotypes, but retain characteristics of GOBm overexpression (Figure 8g), and co-expression of miR160 and miR164 led to variable phenotypes that were similar or more enhanced than the miR160 overexpression phenotype (Figure 8h). Together, these results suggest that GOB mediates the activity of SlARF10A in leaflet patterning. The slightly enhanced phenotype of co-expression of miR160 and miR164 implies that GOB can also be a mediator of additional genetic pathways that regulates leaflet separation. To examine whether GOB is a target of SlARF10A, we used qRT-PCR to analyze its relative expression in genotypes that expressed different levels of SlARF10A: pFIL>>miR160, wild-type, FIL>>SlARF10Am and pFIL>>SlARF10Am-SRDX and pFIL>>SlARF10Am-VP16 (see below) (Figure S7). GOB expression was similar among these genotypes, indicating that although GOB is developmentally required for SlARF10A activity in leaf patterning, it is not a target of SlARF10A.
SlARF10A acts mainly as a transcriptional repressor in tomato leaf patterning
Arabidopsis miR160-targeted ARFs have been classified as transcriptional repressors based on the amino acids that compose their MR (Tiwari et al., 2003; Guilfoyle and Hagen, 2007), and tomato SlARF10A and SlARF17 have been recently shown to inhibit the activating effect of auxin on the expression of pDR5::GFP in co-transfection assays in tobacco protoplasts (Zouine et al., 2014). To test genetically whether the activity of SlARF10A in inhibiting lamina generation represents transcriptional activation or repression, we expressed fusions of SlARF10Am with either the activation domain from herpes simplex virus protein VP16 (Sadowski et al., 1988; Busch et al., 1999), or the plant repression domain SRDX (Hiratsu et al., 2003) in leaves (Figure 9). The leaf phenotype of pFIL>>SlARF10Am-SRDX and pBLS>>SlARF10Am-SRDX was similar to, but more severe than, that of plants that overexpressed the unfused SlARF10Am from the same promoters (Figure 9a,b). pFIL>>SlARF10Am-VP16 and pBLS>>SlARF10Am-VP16 leaves showed a range of phenotypes but in general all transgenic lines had similar or milder phenotypes in comparison with pFIL>>SlARF10Am and pBLS>>SlARF10Am, respectively (Figures 9c,d and S8). These results suggest that in the context of tomato leaf patterning, SlARF10A acts mainly as a transcriptional repressor: fusion to the SRDX enhanced its repressing effect, while the VP16 fusion partially compromised the repression activity of SlARF10Am.
Leaf patterning involves precise spatial and temporal localization of lamina growth aided by auxin signals. Here we show that miR160-targeted ARFs affect leaf patterning by modulating lamina generation. Several recent reports have suggested that mechanisms of leaflet separation involve juxtaposition of regions with lamina growth and regions where growth is inhibited, determined in part by dispersed expression of growth regulators. In Arabidopsis leaf margins, local growth inhibition by specific expression of KRP1, led to the formation of deep lobes (Malinowski et al., 2011). The C. hirsuta homeodomain protein RCO is expressed specifically in leaflet bases where it inhibits growth and consequently promotes leaflet separation. RCO is thought to have arisen via gene duplication and acquisition of specific expression in leaflet bases. It was lost in Arabidopsis, which has simple leaves, while its paralogue LMI1 exists in both species but is not expressed in leaflet bases (Vlad et al., 2014). Variations in leaf lobing and serrations within the genus Capsella (shepherd's purse) was found involve, differences in CrRCO-A alleles (Sicard et al., 2014). Here, miR160-targeted ARFs are shown to promote leaflet separation and leaf complexity. SlARF10A and SlARF17 expression is not restricted to the leaflet bases (Figure 2) (Hendelman et al., 2012). Therefore, its growth-inhibiting effect probably depends on the level of expression and its interaction with additional factors that pattern growth. Leaflet separation requires that auxin maxima at the leaf margin will be separated by areas without auxin (Barkoulas et al., 2008; DeMason and Polowicky, 2009; Koenig et al., 2009; Bilsborough et al., 2011; Ben-Gera et al., 2012). In Arabidopsis fruits, the transcription factor INDEHISCENT (IND) is specifically expressed at the silique valve margins, where it promotes local auxin depletion, stimulating fruit opening (Sorefan et al., 2009). ENTIRE and miR160-targeted ARFs may be involved in generating a similar active local auxin response inhibition in the process of leaflet separation.
Overexpression of SlARF10Am or SlARF17m in the leaf under the control of the FIL promoter did not repress the establishment of auxin maxima and leaflet initiation. This result may suggest that the leaflet initiation sites and the intercalary spaces represent two different ‘developmental domains’ that utilize different mediators of auxin response, and that SlARF10A activity is specific to the intercalary space. Alternatively, SlARF10A may act only after leaflet initiation, or act in a somewhat quantitative manner. Expression of SlARF10Am and SlARF17m under the control of native promoters may complete this picture.
Specification of boundaries between leaflets in tomato leaves depends on the activity of the GOB gene, which marks the sites of leaflet initiation in the leaf margin, and properly spaced GOB expression was found to promote leaflet separation (Blein et al., 2008; Berger et al., 2009). Auxin was shown to mediate the activity of GOB in leaf patterning, and E expression is affected by GOB (Ben-Gera et al., 2012). Co-expression of miR164 almost completely suppressed the SlARF10Am overexpression phenotype, and miR164-resistant GOB substantially suppresses the miR160 overexpression phenotype. Taken together, these results suggest that, genetically, SlARF10A acts upstream of GOB. As GOB and E were shown to redundantly affect auxin response and leaflet separation (Ben-Gera et al., 2012), and E and SlARF10A are shown here to act together but partially redundantly in this process, these factors appear to be involved in a complex feedback mechanism (Figure 10). Comparison of the effects of altering the activity of GOB, E and miR160-targeted ARFs on leaf marginal patterning and auxin response suggests that they affect different aspects of this process. While in pFIL>>miR164 leaves the initiation of primary leaflets and auxin maxima were not affected (Ben-Gera et al., 2012), in pFIL>>miR160 and e mutants the auxin signal was expanded to throughout the leaf margin and primary leaflets were fused. In addition, overexpression of GOB, E and miR160-targeted ARFs led to distinct phenotypes. These findings may imply that these factors regulate different aspects of auxin response. This is in agreement with the expression patterns of these genes: while GOB is expressed in narrow stripes in the boundaries between leaflets and the leaf margin (Berger et al., 2009), E is expressed throughout the region between initiating leaflets (Ben-Gera et al., 2012), and the expression of SlARF10A and SlARF17 is not specific to this region. In Arabidopsis leaves, the GOB ortholog CUC2 was found to facilitate the establishment of auxin maxima in serration development by reorientation of the auxin efflux carrier PINFORMED1 (PIN1) (Bilsborough et al., 2011). This raises the possibility that GOB might be a mediator of auxin distribution also in tomato leaves. While GOB activity was found to be developmentally required for SlARF10 activity, GOB expression did not change in pFIL>>miR160, pFIL>>SlARF10Am, pFIL>>SlARF10Am-SRDX, indicating that GOB is not a target of SlARF10A. miR164 overexpression was found to suppress the phenotypes of several tomato mutants with increased leaf complexity (Bar et al., 2015) and to enhance the phenotype of several mutants with simplified leaves (Busch et al., 2011; Ben-Gera et al., 2012). Therefore, GOB emerges as a central regulator of leaflet separation, on which the effects of many other regulators of leaf complexity converge.
Comparison of the phenotypes caused by expression of a miR160-resistant SlARF10A and its SRDX and VP16 fusions suggests that in the context of tomato leaf development SlARF10A acts mainly as a transcriptional repressor. The similar but milder phenotype of SlARF10Am-VP16 suggests that the transcriptional repression activity of SlARF10Am competes with the activation activity of VP16, as also reported previously (Ohta et al., 2001; Tiwari et al., 2004; Pekker et al., 2005). Repressor domains of Aux/IAAs were previously shown to dominate activator domains of VP16 or ARF5 when combined in the same protein (Tiwari et al., 2004). The repressor activity of miR160-targeted ARFs is also supported by the expansion of the DR5 signal in pFIL>>miR160 leaf primordia. Alternatively, SlARF10A could act as both a transcriptional activator and repressor. Such a dual role has been shown for some plant transcription factors (Bonaccorso et al., 2012). It was also proposed that the same ARF can act as transcriptional activator or repressor depending on its interacting partners (Guilfoyle et al., 1998). Loss of both induction and repression of gene expression by auxin in several Arabidopsis ARF mutants also indicates that ARF proteins can act as both activators and repressors (Okushima et al., 2005; Lokerse and Weijers, 2009). Arabidopsis ARF16 was shown to activate auxin-responsive genes as well as DR5 in roots (Wang et al., 2005), and complex results were reported with respect to the activity of ARF17 (Mallory et al., 2005; Yang et al., 2013). In Arabidopsis, the miR160-targeted ARFs comprise a distinct subgroup of the ARFs proteins, and the known repressing ARFs were found to be closer to the activating ARFs than to the miR160-targeted ARFs (Remington et al., 2004). Furthermore, they do not contain known repressor domains (Lokerse and Weijers, 2009). These findings suggest that the miR160-targeted ARFs may represent a specialized ARF subclass.
The basic model for the auxin signal transduction is compatible with the activity of activating ARFs, and the activity of repressing ARFs in the context of this model is unclear. Repressor ARFs were proposed to act by competition with activating ARFs on binding to the promoters of auxin-responsive genes, direct repression of target genes, or direct repression of activator ARFs (Tiwari et al., 2003; Guilfoyle and Hagen, 2007; Lokerse and Weijers, 2009; Vernoux et al., 2011; Pierre-Jerome et al., 2013). The interaction between repressing ARFs and Aux/IAAs and consequently their auxin-responsiveness is also unclear. Aux/IAAs and repressor ARFs in general do not interact in yeast two-hybrid assays (Vernoux et al., 2011; Guilfoyle and Hagen, 2012), and Aux/IAAs were so far shown to act only as negative regulators of auxin response. In agreement, we did not detect an interaction between E and miR160-targeting ARFs in yeast two-hybrid assays. E and miR160-targeted ARFs both appear to inhibit auxin response and lamina growth between leaflets, thus promoting leaflet separation. Furthermore, E and SlARF10A depend on each other for their activity. It thus seems that SlARF10A and E act cooperatively, rather than antagonistically, in leaf patterning. How could they cooperate without physical interaction? One possibility is that they interact via a third partner. Alternatively, they could affect the same targets independently. For example, they may both inhibit the same activating ARF, or E could inhibit the transcription of a target gene through inhibition of an activating ARF, and SlARF10A could directly inhibit the transcription of the same target. However, the enhanced phenotype of the combined overexpression of miR160 and the e mutant suggests that E and miR160-targeted ARFs also act partially redundantly.
In conclusion, we show that leaflet formation depends on the inhibition of auxin response between initiating leaflets, by ENTIRE, miR160-targeted ARFs and GOBLET (Figure 10). This active mechanism is required for the formation of distinct leaflets.
Tomato seeds (Solanum lycopersicum cv. M82, sp) were sown in a commercial nursery, and transferred to a greenhouse under natural daylight. For in situ hybridization and expression analyses, plants were grown in a growth chamber under an 18 h/6 h light/dark regimen. The previously described e-3 allele (Berger et al., 2009; Ben-Gera et al., 2012) was used for the genetic analysis. Transgenic genotypes were generated by the LhG4 transactivation system (Moore et al., 1998). The Arabidopsis pFIL promoter drives expression throughout young developing tomato leaves and other lateral organs and the pBLS promoter drives expression at later stages of leaf development, from the P4 stage, in developing leaflets (Lifschitz et al., 2006; Shani et al., 2009). The following transgenic genotypes were described previously: Tomato (M82, sp) plants expressing pDR5::VENUS, pFIL>>iaaM, pFIL>>EdIIm-GUS, pBLS>>EdIIm (Ben-Gera et al., 2012), pFIL>>GOBm and pFIL>>miR164 (Berger et al., 2009).
Cloning and plant transformation
To generate op:SlARF10Am, op:SlARF10Bm and op:SlARF17m we amplified each gene from cDNA (M82 background), and used assembly PCR to introduce silent mutations (see Table S1) at the miR160-recognition site of each gene, downstream to an Operator array (Moore et al., 1998). At least five independent lines from each transgene were crossed to the driver lines, examined, and a representative line was selected for further analysis. In the case of SlARF17m, different lines showed a range of phenotypes, shown in Figure S2. To generate op:SlARF10Am-SRDX, complementary single-stranded DNA fragments of the SRDX RD (see Table S1) were mixed together for the annealing reaction (50 μm of each fragment were mixed to make 1:1 ratio at final volume of 50 μl, denaturation 10 min at 60°C, annealing 1 h at 25°C). The DNA fragments of the SRDX sequence used for the annealing reaction are indicated in Table S1. The VP16 activation domain was originally amplified from the pRG50 plasmid (Sadowski et al., 1988; Parcy et al., 1998; Pekker et al., 2005). We used ligation reactions in order to ligate the SRDX or the VP16 to the C-terminus of the op:SlARF10Amsequence through the KpnI restriction site. To generate op:miR160, we amplified the miR160A (AT2G39175) from Arabidopsis Ler genomic DNA using the primers shown in Table S1. For yeast two-hybrid experiments, we amplified the genes ENTIRE, SlARF10A, SlARF10B and SlARF17 from tomato cDNA (M82 background, sp). Each of the amplified genes was cloned into the entry vector pENTR D-TOPO of the Gateway system (Invitrogen, Carlsbad, CA, USA). The clones were LR-crossed into the gateway yeast two-hybrid destination vectors pDEST32 (BD; LEU) and pDEST22 (AD; Trp) (Invitrogen). The primers used for cloning are indicated in Table S1.
Quantification of phenotypes and statistical analysis
Quantification of lamina narrowing: the fifth leaf of at least six plants of each genotype was photographed and analyzed using the ImageJ software, and the ratio between the lamina area to the leaf perimeter of each leaf was calculated. Leaflets quantification: Leaflets of the sixth leaf of 2-month-old plants were counted. Averages and standard errors of each genotype were calculated from six different plants. Statistical analysis was performed using JMP software (SAS Institute, http://www.sas.com). The Student's t-test was used for comparison of means, which were deemed significantly different at a P <0.05.
Exogenous auxin application
Sterile wild-type and pFIL>>SlARF10Am seeds were sown on MS medium [0.44% w/v MS (M-0222.0050; Duchefa, http://www.duchefa.com), 1% sucrose w/v, 0.75% plant agar (Duchefa, Haarlem, the Netherlands), 1 mm Plant Preservative Mixture (Caisson Laboratories, Smithfield, UT, USA)]. Seven-day-old seedlings were cut 2 cm beneath the SAM, and vegetative shoot apices were transplanted to MS medium containing 1 μm 2,4-D (Sigma-Aldrich http://www.sigmaaldrich.com), or a mock treatment, for 4 weeks.
Tissue collection and RNA analysis
Second leaf primordia at the stage of P5 were collected and transferred directly into liquid nitrogen. At least three biological repeats were collected for each genotype, each containing six leaves. RNA was extracted using the Plant/Fungi Total RNA Purification Kit (Norgen Biotek, Thorold, ON, Canada) according to the manufacturer's instructions, including DNase treatment. cDNA synthesis was performed using the Verso cDNA Kit (Thermo Scientific, Waltham, MA, USA) with 1 μg of RNA. Tissue collection and RNA extraction for dynamic expression of SlARF10A at successive stages of wild-type leaf development was performed as described (Shleizer-Burko et al., 2011).
RNA in situ hybridization was performed as described (Shani et al., 2010). The SlARF10A and SlARF17 probes were synthesized with UTP-DIG and expression of miR160 was detected using a SlmiR160-LNA probe (Table S1) labeled with DIG at the 3′ and 5′ ends (Exiqon, Vedbaek, Denmark). Quantitative RT-PCR analysis was performed using a Rotor-Gene Q device from Qiagen and Absolute Blue qPCR SYBR Green ROX Master mix (ABgene, Epsom, UK). Levels of mRNA were calculated relative to the gene EXPRESSED as an internal control, which has been shown to be expressed at a similar level throughout tomato development (Exposito-Rodriguez et al., 2008).
Imaging, microscopy and GUS staining
Intact leaves were photographed using a Nikon D5200 camera. Images of early leaf development were captured using an Olympus SZX7 stereoscope microscope equipped with a Nikon DXM1200 camera and ACT-1 software or by Nikon SMZ1270 stereoscope with a Nikon DS-Ri2 camera and NIS-element software. For pDR5::VENUS expression, dissected whole-leaf primordia were placed into drops of water on glass microscope slides and covered with cover slips. The pattern of VENUS expression was detected by a confocal laser scanning microscope (CLSM model SP8; Leica, Mannheim, Germany), with the solid-state laser set at 514 nm for excitation and 530 nm for emission. Chlorophyll autofluorescence was detected at 488 nm for excitation and 700 nm for emission. LAS AF and ImageJ software were used for analysis of captured images. Scanning electron microscopy was performed using a JEOL 5410 LV microscope as described (Brand et al., 2007). Images were manipulated uniformly using Adobe Photoshop CS6 software (Adobe Systems Inc., San Jose, CA, USA).
Yeast two-hybrid assays
The PJ69-4 yeast strain was transformed following CLONTECH small-scale LiAc manual (www.clontech.com/xxclt_ibcGetAttachment.jsp?cItemId=17602). We first transformed the pDEST32, by performing separate transformations for each of the genes: E, SlARF10A, SlARF10B and SlARF17, and transformed yeast lines were selected on Sabouraud Dextrose (SD) agar plates without tryptophan. Each transformed yeast line was transformed with each of the genes in the pDEST22 plasmid in order to have all genes combinations (ExE, ExSlARF10, ExSlARF10B, ExSlARF17, SlARF10AxSlARF10A, SlARF10AxSlARF10B, SlARF10AxSlARF17, SlARF10BxSlARF10B, SlARF10BxSlARF17, SlARF17xSlARF17). As a control, we co-introduced an empty pDEST32 vector with each of the genes in pDEST22 vector. Transgenic yeast lines containing both plasmids were selected on SD agar plates without tryptophan and without leucine. Transgenic yeasts containing putatively interacting proteins were selected on SD agar plates without histidine, leucine, and tryptophan (His−/Leu−/Trp−) and also on SD agar plates without adenine, histidine, leucine, and tryptophan (Ade−/His−/Leu−/Trp−) for more stringent selection of positive interactions. The SD agar plates also contained β-galactosidase for the in vivo blue colour assay. Positive interaction was considered only for transgenic lines that grew on SD agar plates without adenine, histidine, leucine, and tryptophan (Ade−/His−/Leu−/Trp−) and showed a blue or light blue colour.
Sequence data used in this study can be found in the Sol Genomic Network (SGN) database under the following accession numbers: SlARF10A (Solyc11g069500); SlARF10B (Solyc06g075150); SlARF17 (Solyc11g013480 [Nter] + Solyc11g013470 [Cter]); AtmiR160A (AT2G39175); ENTIRE (Solyc04g076850); GOBLET (Solyc07g062840); EXPRESSED (Solyc07g025390).
We would like to thank Yuval Eshed for the miR160 driver line and, the VP16 construct, Tzahi Arazi and Yuval Eshed for critical reading of the manuscript, Dario Breitel and Louise Maor for helping with the yeast two-hybrid procedure, Shiri Goldental for technical assistance, and members of our groups for fruitful discussions. This research was supported by grants from the Israel Science Foundation (539/14), US–Israel Binational Agricultural Research and Development Fund (IS 4531-12(c)), The Israeli ministry of Agriculture (837-0140-14) and German–Israel Project Cooperation Foundation (OR309/1-1; FE552/12-1).