ENTIRE and GOBLET promote leaflet development in tomato by modulating auxin response


  • Hadas Ben-Gera,

    1. The Robert H. Smith Institute of Plant Sciences and Genetics in Agriculture and The Otto Warburg Minerva Center for Agricultural Biotechnology, Hebrew University, P.O. Box 12, Rehovot 76100, Israel
    Search for more papers by this author
  • Ido Shwartz,

    1. The Robert H. Smith Institute of Plant Sciences and Genetics in Agriculture and The Otto Warburg Minerva Center for Agricultural Biotechnology, Hebrew University, P.O. Box 12, Rehovot 76100, Israel
    Search for more papers by this author
  • Mon-Ray Shao,

    1. University of California San Diego, Section of Cell and Developmental Biology, 9500 Gilman Dr.La Jolla, CA 92093, USA
    Search for more papers by this author
  • Eilon Shani,

    1. The Robert H. Smith Institute of Plant Sciences and Genetics in Agriculture and The Otto Warburg Minerva Center for Agricultural Biotechnology, Hebrew University, P.O. Box 12, Rehovot 76100, Israel
    2. University of California San Diego, Section of Cell and Developmental Biology, 9500 Gilman Dr.La Jolla, CA 92093, USA
    Search for more papers by this author
  • Mark Estelle,

    1. University of California San Diego, Section of Cell and Developmental Biology, 9500 Gilman Dr.La Jolla, CA 92093, USA
    Search for more papers by this author
  • Naomi Ori

    1. The Robert H. Smith Institute of Plant Sciences and Genetics in Agriculture and The Otto Warburg Minerva Center for Agricultural Biotechnology, Hebrew University, P.O. Box 12, Rehovot 76100, Israel
    Search for more papers by this author

(e-mail ori@agri.huji.ac.il).


Compound leaves produce leaflets in a highly controlled yet flexible pattern. Here, we investigate the interaction between auxin, the putative auxin response inhibitor ENTIRE (E, SlIAA9) and the CUC transcription factor GOBLET (GOB) in compound-leaf development in tomato (Solanum lycopersicum). Auxin maxima, monitored by the auxin response sensor DR5, marked and preceded leaflet and lobe initiation. The DR5 signal increased, but maxima were partially retained in response to the external or internal elevation of auxin levels. E directly interacted with the auxin receptors SlTIR1 and SlAFB6. Furthermore, E was stabilized by a mutation in domain II of the protein and by the inhibition of auxin or proteasome activity, implying that E is subjected to auxin-mediated degradation. In e mutants the DR5 signal expanded to include the complete leaf margin, and leaf-specific overexpression of a stabilized form of E inhibited the DR5 signal and lamina expansion. Genetic manipulation of GOB activity altered the distribution of the DR5 signal, and the inhibition of auxin transport or activity suppressed the GOB overexpression phenotype, suggesting that auxin mediates GOB-regulated leaf patterning. Whereas leaves of single e or gob mutants developed only primary leaflets, the downregulation of both E and GOB resulted in the complete abolishment of leaflet initiation, and in a strong DR5 signal throughout the leaf margin. These results suggest that E and GOB modulate auxin response and leaflet morphogenesis via partly redundant pathways, and that proper leaflet initiation and separation requires distinct boundaries between regions of lamina growth and adjacent regions in which growth is inhibited.


Leaf shape varies enormously within and among species, and can be ascribed to one of two basic forms: simple and compound. Simple leaves comprise a single blade, whereas compound leaves are composed of leaf-like subunits, termed leaflets, which are connected by petiolules to the central leaf rachis (Figure S1). The elaboration of a compound leaf often depends on the prolonged morphogenetic activity of a specific region at the leaf margin, termed marginal blastozone, from which leaflets, lobes and serration develop (Hagemann and Gleissberg, 1996; Ori et al., 2007).

The specification of leaflet initiation sites at the leaf margins was shown to involve the plant hormone auxin and NAM-like transcription factors. Points of high auxin levels, termed auxin maxima, precede leaflet initiation in compound leaves of Cardamine hirsuta, Medicago truncatula, Pisum sativum (pea) and Solanum lycopersicum (tomato) (Barkoulas et al., 2008; DeMason and Polowicky, 2009; Koenig et al., 2009; Zhou et al., 2011). In C. hirsuta, a mutation in the PIN1 auxin transporter led to the simplification of leaf form (Barkoulas et al., 2008). The tomato SlIAA9/ENTIRE (E) protein, a putative transcriptional repressor of auxin-responsive genes from the Aux/IAA (indole-3-acetic acid) family, is required for compound leaf and fruit development (Wang et al., 2005, 2009; Zhang et al., 2007; Berger et al., 2009; Koenig et al., 2009). A major pathway of auxin signaling has been dissected in great detail in recent years. In the absence of auxin, Aux/IAA proteins interact with transcription factors from the ARF family, which regulate the transcription of auxin-response genes (Weijers and Jurgens, 2004; Guilfoyle and Hagen, 2007). Auxin promotes the binding of the Aux/IAA proteins to TIR1, an auxin receptor from the TIR/AFB family, which is a subunit of an SCF E3 ligase complex. This binding leads to the degradation of Aux/IAA and to the subsequent release of ARF (Dharmasiri et al., 2005a,b; Kepinski and Leyser, 2005; Tan et al., 2007; Santner and Estelle, 2009). Most Aux/IAA proteins contain four conserved domains. Domain I is responsible for repression activity, domains III and IV are required for dimerization, and domain II contains a 12-amino-acid motif that is necessary for the auxin-dependant degradation of the protein (Ramos et al., 2001; Reed, 2001; Guilfoyle and Hagen, 2007; Mockaitis and Estelle, 2008; Li et al., 2011). As Aux/IAA proteins belong to gene families (Nebenfuhr et al., 2000; Mockaitis and Estelle, 2008), most of the loss-of-function mutants in Arabidopsis Aux/IAAs have no or very minor effects on growth and development (Reed, 2001; Overvoorde et al., 2005). In contrast, mutations in the tomato E gene lead to a simpler leaf form and parthenocarpic fruits.

The NAM/CUC transcription factors were shown to have conserved roles in the specification of leaf marginal structures in species with either simple or compound leaves (Nikovics et al., 2006; Blein et al., 2008; Berger et al., 2009; Bilsborough et al., 2011; Hasson et al., 2011). In tomato, the miR164-regulated NAM/CUC gene GOBLET (GOB) is expressed in local, specific stripes that mark and precede the initiation of leaflets. Both reduced and expanded GOB expression leads to simpler leaves, possibly because of the lack of discrete boundaries (Blein et al., 2008; Berger et al., 2009). In Arabidopsis, interspersed distribution of auxin maxima and CUC2 activity along the leaf margin is required for the formation of marginal serrations. A computational model predicted a feedback loop, in which CUC2 promotes the polarization of PIN1 towards convergent points, leading to the formation of auxin maxima, which in turn repress CUC2 expression (Bilsborough et al., 2011).

How auxin, E and GOB interact to specify leaflet initiation and separation in the compound tomato leaf is still unclear. Furthermore, the comparison of these interactions between species with simple and compound leaves can shed light on how similar building blocks are used to pattern leaves of variable shapes. Here, we show that E physically interacts with the auxin receptors TIR1 and SlAFB6, and that a mutation in domain II stabilizes the E protein, indicating that E is likely to be degraded via the SCFTIR1/AFB-mediated pathway. Phenotypes of GOB overexpression are mediated by the altered distribution of auxin response. The downregulation of the activities of both E and GOB leads to the complete elimination of leaflet initiation and to an intensified and expanded auxin response signal, indicating that E and GOB are likely to act via partially redundant pathways to promote leaflet initiation and separation, and to restrict auxin response. These results suggest that a highly regulated interaction between auxin, E and GOB is used in the flexible patterning of leaf margins.


Auxin response maxima mark the initiation sites of leaf marginal structures

To further understand the role of auxin in tomato leaf patterning, we studied the spatial and temporal distribution of the auxin signal during tomato leaf development. Previously, the expression of DR5:GUS and Arabidopsis pPIN1:PIN1:GFP were used to sense auxin response during tomato leaf development (Koenig et al., 2009). Although these markers were very informative, the interpretation of PIN1 expression is challenging as it is both a cause and a result of auxin distribution, and its expression probably responds to other cues in addition to auxin (Krecek et al., 2009). DR5:GUS was only visible relatively late in the development of leaf primordia, suggesting low sensitivity in the shoot apical meristem (SAM) and in very young primordia. We thus generated transgenic tomato plants harboring the auxin response sensor DR5::VENUS (Ulmasov et al., 1997; Heisler et al., 2005) as a tool to sensitively monitor the distribution of auxin response in tomato. DR5::VENUS expression was detected at the site of incipient leaf primordia in the SAM flanks (Figure 1a), similar to other species (Benkova et al., 2003; Heisler et al., 2005; DeMason and Polowicky, 2009). In early P3 primordia, a maximum in the DR5::VENUS signal appeared at the future initiation site of the first leaflet, before the leaflet bulge was visible (Figure 1a). Following leaf and leaflet initiation, expression was restricted to the tips of the primordia and to the presumptive leaf vascular tissue (Figure 1a–c). Similar maxima were observed in all initiating leaflets and lobes (Figure 1d–f, arrowheads; for a general scheme of a tomato leaf, see Figure S1), and at the position of the future intercalary leaflet, between the first and second leaflets (Figure 1g, arrowhead). Thus, the distribution of auxin maxima along the leaf margin spatially and temporarily correlated with the pattern of marginal-appendage initiation. In comparison with pPIN1:PIN1:GFP expression, DR5::VENUS expression was much more localized to the initiation sites, leaflet tips and presumptive vasculature.

Figure 1.

 An auxin response sensor marks the sites of leaflet initiation.
(a–g) Confocal micrographs showing the spatial distribution of DR5::VENUS expression (green) in the fifth leaf produced by the Solanum lycopersicum (tomato) plant at successive developmental stages. (a) The shoot apical meristem (SAM) and young leaf primordia at the P2 and early P3 stages. Inset is a magnification of the DR5::VENUS peak preceding leaflet initiation. (b–g) Longitudinal views of leaf primordia. The developmental stage is indicated at the top right corner of each panel. The inset in (c) is an enlargement of the primary leaflet. Insets in (e–g) illustrate the context of the image shown within the leaf. Arrowheads point to auxin maxima in initiating lobes (d, e, f), and prior to intercalary leaflet initiation (g). (h–j) Stereoscope images showing the effect of microapplication of IAA on the distribution of the DR5::VENUS signal (green). The application site is shown by the black lanolin paste on leaf primordia at the P3 stage. The arrowhead in (j) points to the microapplication site. Treatments are indicated at the bottom left corner of each panel. P0 and P2–P6 are the respective developmental stage of the leaf primordia in plastochrons. Scale bars: 200 μm (a–g); 1 mm (h–j).

The effect of auxin on DR5::VENUS distribution was examined by microapplication of auxin to specific sites of the developing leaf. Upon auxin microapplication, the DR5::VENUS signal was intensified along the application region, but endogenous maxima were still maintained (Figure 1h–j). Auxin microapplication also caused ectopic blade or leaflet outgrowth, as reported previously (Figure S1) (Koenig et al., 2009). Interestingly, the intercalary region below the application site maintained a more or less similar length to that of the control side, resulting in an asymmetric leaf (Figure S1), implying the existence of an endogenous mechanism to adjust the gaps between leaflets.

Overexpression of the bacterial auxin biosynthesis gene iaaM leads to a simplified leaf form and to expanded DR5::VENUS expression

The Agrobacterium tumefaciens tryptophan monooxygenase (iaaM) gene encodes an enzyme that participates in the auxin biosynthetic pathway, and its overexpression in plants was shown to lead to elevated levels of free IAA (Romano et al., 1995). To test the effect of increased endogenous auxin in developing leaves, we overexpressed the iaaM gene using the FIL promoter, which drives expression throughout early leaf development (Lifschitz et al., 2006; Shani et al., 2009). The leaves were simpler than wild-type leaves, showing primary leaflets with continuous blade outgrowth and significant reduction of secondary leaflets (Figure 2a,b), in agreement with the effect of exogenous auxin application (Koenig et al., 2009). The terminal leaflet showed precocious and more uniform expansion in comparison with the wild type, with no clear initiation of the terminal-leaflet lobe. Primary leaflets initiated separately but closer to each other and to the terminal leaflet (Figure 2c–e), relative to the wild type. Auxin maxima, monitored by DR5::VENUS expression, were established normally in pFIL≫iaaM leaf primordia during primary leaflet initiation, but the signal was distributed continuously throughout the terminal leaflet margin, in accordance with continuous blade outgrowth (Figure 2f). Thus, during primary leaflet initiation, the effect of increased auxin was inhibited, possibly by auxin transport and by the inhibition of the auxin response.

Figure 2.

 Transgenic elevation of endogenous auxin levels leads to simplified leaves.
(a, b) Mature fifth leaves. (c, d) Stereoscope images of fifth leaves at the P4 stage. (e) Scanning electron micrograph of a fifth leaf at the P4 stage. Arrowheads point to initiating primary leaflets. (f) Confocal image showing the DR5::VENUS signal (green) in a young pFIL≫iaaM leaf primordium. Scale bars: 1 cm (a, b); 500 μm (c–e); 200 μm (f).

ENTIRE interacts with the auxin receptors SlTIR1 and SlAFB6, and is stabilized by a mutation in the degron motif

The tomato ENTIRE (E) gene, also called SlIAA9, encodes an Aux/IAA protein that has been shown to be involved in compound-leaf patterning (Wang et al., 2005; Zhang et al., 2007; Berger et al., 2009). To determine if E is regulated through the same mechanism of auxin-mediated degradation known for the Arabidopsis Aux/IAAs, we tested its interaction with putative tomato auxin receptors. Based on sequence homology, we identified three tomato proteins closely related to the TIR1/AFB family of auxin receptors characterized in Arabidopsis (Parry et al., 2009). The tomato protein most closely related to Arabidopsis TIR1 was designated SlTIR1; its expression pattern and overexpression phenotype has been described by others (Ren et al., 2011). Another putative tomato auxin receptor is most closely related to the AFB4/AFB5 clade of auxin receptors (66–68% amino acid identity), and was named SlAFB4. A third putative tomato receptor is 49–56% identical to Arabidopsis TIR1/AFB family members. Phylogenetic analysis indicates that this protein is distinct from other members of the Arabidopsis TIR1/AFB family, and is instead more closely related to the AFB6 clade that exists in other plant species (Figure S2). Therefore, it has been designated SlAFB6. SlTIR1 and SlAFB6 cDNAs were fused with a C-terminal myc tag, expressed under a 35S promoter in Nicotiana benthamiana (tobacco) leaves via Agrobacterium infiltration, and then the fused proteins were extracted. In in vitro pull-down assays, GST-tagged ENTIRE (SlIAA9) and the Arabidopsis homolog AtIAA7 demonstrated an auxin-dependent interaction with SlTIR1-myc and SlAFB6-myc proteins from the extract (Figure 3a), providing evidence that SlTIR1 and SlAFB6 are functional auxin receptors. Interestingly, the interaction of GST-ENTIRE with SlAFB6-myc was consistently stronger than with SlTIR1-myc, suggesting a receptor specificity component to E regulation.

Figure 3.

 ENTIRE interacts with auxin receptors and is stabilized by a mutation in domain II.
(a) In vitro pull-down assays with GST-ENTIRE, or GST-AtIAA7, against SlTIR1-myc or SlAFB6-myc, in the presence of 50 μm indole-3-acetic acid or a solvent control. Included are loading controls for SlTIR1-myc and SlAFB6-myc (middle row; detected with anti-c-myc), and GST-ENTIRE and GST-AtIAA7 (bottom row; detected with Ponceau S). (b–e) Histochemical staining of β-glucoronidase (GUS) activity in shoot apices of seedlings pretreated with MG-132, PEO-IAA or DMSO as a negative control. In each panel, genotypes are indicated at the bottom-left corner, and treatments are indicated at the top right corner. Scale bars: 500 μm (b–e).

To test whether domain II (the degron motif) of the E protein is involved in the auxin-dependent destabilization of the protein, we generated transgenic lines expressing GUS fusions with either wild-type E or with an E containing a mutation in domain II (EdIIm) (Li et al., 2011). E-GUS and EdIIm-GUS were transactivated in initiating leaves using the FIL promoter. GUS activity could only be detected in pFIL≫EdIIm-GUS, but not in pFIL≫E-GUS leaves (Figure 3b,c), indicating that similar to Arabidopsis Aux/IAAs, domain II confers instability to the E protein (Gray et al., 2001; Ramos et al., 2001; Reed, 2001). To test whether this instability is mediated by auxin and the proteasome, we treated pFIL≫E-GUS seedlings with the proteasome inhibitor MG-132 (Gray et al., 2001) or with the auxin antagonist PEO-IAA, which specifically inhibits TIR1/AFBs-mediated auxin signaling (Hayashi et al., 2008). Both treatments resulted in detectable E-GUS activity (Figure 3d,e). Interestingly, pFIL≫EdIIm-GUS leaves showed reduced GUS activity in comparison with pFIL≫GUS leaves (Figure S3), in agreement with previous reports showing that additional regions in the Aux/IAA proteins are involved in the control of their stability (Calderon-Villalobos et al., 2010). Alternatively, the fusion with E may have interfered with GUS activity. In conclusion, E stability is probably regulated through SCFTIR1/AFB-mediated degradation in an auxin-dependant manner.

ENTIRE inhibits auxin response and lamina outgrowth between initiating leaflets

Previously, the E transcript was shown to accumulate in the SAM and in the margins of P2 and P3 leaf primordia (Figure S4) (Koenig et al., 2009). As in developing entire (e) leaves, primary leaflets fuse because of ectopic lamina growth between leaflets (Figures 4g and S4) (Dengler, 1984; Koenig et al., 2009), we examined the distribution of the E transcript at later stages of leaf development. Strong E expression was detected in the intercalary area between the primary leaflets (Figures 4a and S4). This implies that E activity is regulated at both the transcriptional and the protein stability levels. The distinct accumulation of E between initiating leaflets was lost in three e alleles, which may be secondary to the developmental effect (compare Figures S4h–j with S4f). Although the pattern of the DR5::VENUS signal at the SAM flanks was similar between wild-type and e plants, the signal expanded to the whole leaf margin of e leaf primordia, compared with the distinct auxin maxima in wild-type primordia (Figure 4b–e). Interestingly, in wild-type leaves the DR5::VENUS signal and the E mRNA expression patterns were complementary (compare Figures 1d with 4a), and in e mutants the DR5::VENUS signal expanded into the E expression domain (compare Figures 4a and 4d).

Figure 4.

 E inhibits auxin response and lamina expansion.
(a) In situ hybridization with an E probe on longitudinal sections of a wild-type leaf primordium. (b–e) Confocal micrographs showing the spatial distribution of DR5::VENUS expression (green) in primordia of fifth e leaves at successive developmental stages. (b) Shoot apical meristem (SAM) and young leaf primordia. (c–e) The developmental stages are indicated at the top right corner. (f–n) Overexpression of an auxin-insensitive E form inhibits lamina expansion. (f, g, j) Mature fifth leaves. (h) A whole pFIL≫ EdIIm plant with only one leaf. (i) Scanning electron micrograph of the SAM. (k, l) Stereoscope images of fifth leaves at the P4 stage. (m, n) Scanning electron micrographs of shoot apices, where the largest leaf primordia are at the P4 stage. (o–s) DR5::VENUS expression (green) in the first primary leaflet of P6 staged leaf primordia of wild type (o, p) or pBLS≫EdIIm (q–s). (p) and (s) are magnifications of (o) and (r), respectively. In all panels, genotypes are indicated at the bottom left corner. PL, primary leaflet; TL, terminal leaflet. Scale bars: 100 μm (a); 200 μm (b–e, i, m–s); 2 cm (f–h, j); 1 mm (k, l).

Overexpression of a putative auxin-insensitive version of E, but not its native form, by a strong constitutive promoter severely affected tomato development (Koenig et al., 2009). To understand the specific role of E in leaf development we overexpressed E and EdIIm specifically in developing leaves, using the two lateral organ-specific promoters pFIL and pBLS. pFIL≫EdIIm plants initiated only one leaf, and later the SAM continued to grow with no further leaf initiation (Figure 4h,i). Flowers were aberrant, with numerous unorganized narrow organs, and were infertile (Figure S5). pFIL≫EdIIm-GUS plants showed a similar but milder phenotype in comparison with pFIL≫EdIIm (Figure S3). The BLS promoter drives expression slightly following primary leaflet initiation in marginal and distal regions (Lifschitz et al., 2006; Shani et al., 2009). pBLS≫EdIIm leaves were small with very narrow lamina and very few secondary leaflets (Figure 4j). An examination of the early development of pBLS≫EdIIm showed that leaflet expansion is inhibited relative to the wild type (Figure 4k–n). The DR5::VENUS signal was substantially reduced in pBLS≫EdIIm leaflets at the time of BLS expression, and the distinct sharp maxima were lost (Figure 4o–s), which could be secondary to the phenotypic effect. In conclusion, E inhibits lamina growth between leaflets by repressing the auxin response.

Auxin mediates GOB activity

The tomato NAM/CUC transcription factor GOB, which is negatively regulated by miR164, was shown to be involved in leaflet specification and separation (Blein et al., 2008; Berger et al., 2009). pFIL≫GOBm plants, which ectopically express a miR164-insensitive GOB form, initiate numerous leaf marginal structures that later fuse to become lobes (Berger et al., 2009). To understand the relationship between auxin and GOB in compound-leaf development, we examined the distribution of the DR5::VENUS signal in leaves of pFIL≫GOBm plants. Frequent, improperly spaced auxin maxima were observed in pFIL≫GOBm leaf primordia, in correlation with the ectopic initiation events (Figure 5a–d). The ectopic auxin maxima may be secondary to the specification of ectopic leaflet initiation sites. To test whether these auxin maxima are necessary for the ectopic initiation events in pFIL≫GOBm primordia, we tested the influence of the auxin transport inhibitor NPA or the auxin antagonist PEO-IAA on pFIL≫GOBm leaf development. Strikingly, the inhibition of auxin transport or auxin activity inhibited leaflet initiation, and the ectopic initiation events (Figure 5e–i), without affecting GOB transcript levels (Figure S6), suggesting that GOB activity in leaflet initiation is mediated by proper auxin distribution. NPA treatment also resulted in the spreading of the DR5 signal such that its sharp and intense maxima were lost, whereas PEO-IAA treatment resulted in a substantial reduction of the DR5 signal (Figure 5j–l).

Figure 5.

 Auxin mediates the effect of pFIL≫GOBm on leaf development.
(a–d) Confocal micrographs showing the spatial distribution of DR5::VENUS expression (green) during the early stages of pFIL≫GOBm leaf development. (a) Shoot apical meristem (SAM) and young leaf primordia. (b–d) Leaf primordia, the developmental stage is indicated at the top right corner. (e–i) pFIL≫GOBm leaf primordia treated with DMSO (e, f), 10 μm NPA (g, h) or 400 μm PEO-IAA (i), viewed by a stereoscope (e, g, i) or by a scanning electron microscope (f, h). (j–l) Confocal micrographs showing DR5::VENUS expression (green) in pFIL≫GOBm leaf primordia treated with DMSO (j), 10 μm NPA (k) or 400 μm PEO-IAA (l). Scale bars: 200 μm (a–d, j–l); 500 μm (e–i).

pFIL≫miR164 leaves, which overexpress miR164, a negative regulator of GOB, produce only primary leaflets (Berger et al., 2009) (see Figure 8b below). The distribution of auxin maxima along the leaf margin during primary leaflet initiation was unaffected by pFIL≫miR164 (Figure 6a–c). In contrast, the DR5::VENUS signal expanded to throughout the margins of the terminal and primary leaflets, whereas its distribution in the vasculature was unaffected (Figure 6d–g). This is reminiscent of the effect of cuc-2 on the DR5 signal in Arabiodpsis leaves (Kawamura et al., 2010; Bilsborough et al., 2011).

Figure 6.

pFIL≫miR164 modulates DR5 distribution in leaflet margins.
(a–g) Confocal micrographs showing DR5::VENUS expression (green) in pFIL≫miR164 leaf primordia. The developmental stages and genotypes are indicated. (d–g) DR5::VENUS expression in terminal leaflets (d, e) and primary leaflets (f, g). Insets indicate the context of the image shown within the leaf. Yellow arrows in (e) and (g) point to the expanded DR5::VENUS signal in the leaflet margin. (h) Influence of IAA application on GOB mRNA level, assayed by qRT-PCR. IAA was applied at a concentration of 300 μm. m-P3 is the SAM and three youngest leaf primordia. Values are means ± SEs (n = 3). Scale bars: 200 μm.

Previous reports suggested that CUC genes are negatively regulated by auxin in the Arabidopsis shoot (Vernoux et al., 2000; Heisler et al., 2005; Bilsborough et al., 2011). To test whether GOB expression is regulated by auxin in tomato shoots and leaves, we measured the effect of auxin application on the GOB mRNA level. Whereas in shoot apices the GOB mRNA level was reduced in response to IAA, in P4 or P5 leaf primordia, IAA did not affect GOB expression (Figure 6h). As a positive control, we tested the effect of the same treatment on the expression of the auxin-responsive gene SAUR, which was upregulated by IAA treatment in both shoot apices and leaf primordia, as expected (Figure S7). In conclusion, auxin mediates GOB activity in leaf margins.

ENTIRE and GOBLET inhibit auxin response and specify primary leaflet initiation via redundant pathways

To test whether GOB and E act in common or parallel pathways, we examined the molecular and genetic interactions between these genes. We first tested whether E and GOB affect the expression of each other. In P4 primordia the GOB mRNA level was similar between wild type and e (Figure 7a), in agreement with previous in situ hybridization experiments (Berger et al., 2009). Conversely, the E mRNA level was elevated in pFIL≫GOBm and Gob-4d plants, containing a dominant miR164-resistant gob allele, and was reduced in the e mutant (Figure 7b). The E mRNA level was also increased in pFIL≫miR164 leaves (Figure 7b), but this increase was not statistically significant. Examination of E expression in these genotypes by in situ hybridization showed that in pFIL≫miR164 primordia E was uniformly spread throughout the leaf and leaflet margins (Figure 7c). In pFIL≫GOBm and Gob-4d, punctuate strong expression was expressed in a less organized, altered pattern, relative to the wild type (Figure 7d,e). The upregulation of E expression in response to both increased and decreased GOB activity is puzzling, and implies complex, developmental context-dependant regulation of E expression. In both cases the effect on E expression could be secondary to the developmental alteration.

Figure 7.

GOB affects E expression.
(a, b) qRT-PCR showing the relative levels of the GOB (a) or E (b) mRNA expression in P4 staged leaf primordia of the indicated genotypes. Values are means ± SEs (n = 3). *Statistically significant differences relative to the wild-type at P < 0.05; (c–e) in situ hybridization with the E probe on longitudinal sections of the indicated genotypes. Scale bars: 100 μm.

To further understand the relationship between GOB and E, we examined the genetic interaction between e and pFIL≫miR164. Whereas leaves of both e and pFIL≫miR164 initiate a normal number of primary leaflets (Figure 8a,b and S4) (Berger et al., 2009), e pFIL≫miR164 leaves lack primary leaflets altogether, and curl backwards (Figure 8c) (Berger et al., 2009). We examined the early leaf development of e pFIL≫miR164 to understand the developmental cause of the lack of leaflets. Strikingly, leaflet initiation was completely eliminated in e pFIL≫miR164, despite the existence of a marginal blastozone, as manifested by the lack of trichomes, which are a histological marker for differentiation (Figure 8d–h). e pFIL≫miR164 leaves showed an intensified DR5::VENUS signal throughout the margins of young leaf primordia, in agreement with the lack of leaflet initiation (Figure 8j). An enhanced signal was also observed in the SAM and the youngest leaf primordia, in comparison with wild type, e, pFIL≫miR164 and pFIL≫GOBm (Figure 8i). Thus, although both GOB and E are required for secondary leaflet initiation, they act via redundant pathways in primary leaflet initiation, and both restrict the auxin response in the leaf margin.

Figure 8.

e pFIL≫miR164 leaves fail to produce leaflets.
(a–c) Mature fifth leaves. (d–g) Successive stages of leaf development, viewed by a stereoscope. (h) Scanning electron micrograph of e pFIL≫miR164 leaf primordia. The arrowhead points to the trichome-less marginal blastozone. (i,j) Confocal micrographs showing the DR5::VENUS signal (green) in the e pFIL≫miR164 shoot apical meristem (SAM) and young leaf primordia (i), and in a P4-staged leaf primordium (j). (k) A scheme showing the proposed effect of E and GOB on the auxin response and lamina growth during leaf development in tomato. Scale bars: 1 cm (a–c); 500 μm (d–h); 100 μm (i–j).


Compound-leaf patterning involves the precise regulation of local growth and differentiation. Here, we show that proper leaflet initiation and separation in tomato is achieved by the coordinated action and interaction of auxin, GOB and E.

The auxin response is regulated at multiple levels in developing tomato leaves

Previously, SlPIN1 convergence points were shown to mark the sites of leaflet initiation (Koenig et al., 2009), and together with the distribution of AtpPIN1:PIN1:GFP predicted that auxin maxima are formed at these sites prior to initiation. The current results confirm this prediction using the DR5::VENUS reporter, which provides a marker for the combined outcome of auxin biosynthesis, transport and signal transduction. AtpPIN1:PIN1:GFP shows a broader distribution than DR5::VENUS, suggesting that its expression responds to other signals in addition to auxin.

A comparison of the effect of manipulation of auxin concentrations, GOB and E on the distribution of the DR5::VENUS signal, and on leaf development, suggests that the distribution of the auxin response is regulated at multiple levels, including auxin levels, transport, and signal transduction. Leaf overexpression of iaaM affected blade growth and DR5::VENUS distribution only in leaflets, and still allowed the formation of distinct primary leaflets and, accordingly, distinct auxin maxima at the sites of their initiation. In contrast, in e mutants primary leaflets are partially fused, in accordance with an expanded DR5::VENUS signal distribution. This implies that in the presence of E, elevated auxin levels are not sufficient to cause increased auxin response and primary leaflet fusion.

E and GOB modulate auxin response and leaflet morphogenesis via redundant pathways

The suppression of the ectopic leaflet initiation phenotype of pFIL≫GOBm by NPA suggests that GOB may modulate auxin distribution. Furthermore, Arabidopsis CUC2 was proposed to promote the generation of auxin response maxima by controlling the generation of PIN1-convergence points (Kawamura et al., 2010; Bilsborough et al., 2011). The effect of GOB on E expression raises the question of whether the effect of GOB on auxin response is through E, but the further enhancement of the DR5 signal in e pFIL≫miR164 relative to e suggests that they affect auxin response through at least partially parallel pathways, and that GOB also acts to inhibit the auxin signal between leaflets.

Whereas GOB is downregulated by auxin in tomato shoot apices, its expression in leaf primordia is not affected by auxin application. This is different than the case in Arabidopsis, where auxin was proposed to negatively regulate CUC expression in both the SAM and the leaf (Vernoux et al., 2000; Furutani et al., 2004; Heisler et al., 2005; Aida and Tasaka, 2006; Bilsborough et al., 2011), and implies a differential interaction between auxin and CUC genes depending on the species and tissue.

GOB and E are expressed in overlapping domains flanking initiating leaflets, and their mutants display similar leaf phenotypes (Blein et al., 2008; Berger et al., 2009; Koenig et al., 2009). Genetic analysis indicated that GOB and E are both required for leaflet separation, and the formation of higher order leaflets and serrations, but act through partially redundant pathways with respect to primary leaflet initiation (Figure 8). In agreement, reducing the activity of either GOB or E partially affected the auxin response, but downregulation of both led to a continuous strong auxin response signal throughout the leaf margin. Recently, GOB and the boundary gene POTATO LEAF (C) were also shown to affect leaflet initiation via redundant pathways (Busch et al., 2011). The differential effect on primary and secondary leaflets suggests that the initiation of primary leaflets is more robust than that of secondary leaflets, as has also been indicated by mutant analysis (Menda et al., 2004; Brand et al., 2007; Efroni et al., 2010).

Patterning by dispersed signals

Increased and decreased GOB activity both led to enhanced DR5::VENUS signal in the margins of the leaflets (Figures 5 and 6). However, the two scenarios had distinct effects on DR5 distribution. The effect of GOB upregulation and downregulation on leaflet morphogenesis was also not simply opposite (Berger et al., 2009), and external auxin application led to either a lack of leaflets or the formation of ectopic leaflets. (Barkoulas et al., 2008; Koenig et al., 2009) (Figure S1). These observations suggest that the specific distribution of auxin, GOB and E, and distinct boundaries between their expression domains, underlie leaf margin morphogenesis rather than the mere presence or absence of these patterning agents, similar to recent observations in Arabidopsis (Bilsborough et al., 2011). Recently, the targeted restriction of growth was shown to result in increased lobing in Arabidopsis, whereas uniform growth restriction along the leaf margin led to smoother margins (Malinowski et al., 2011). This further supports the model that morphogenesis at the leaf margin requires adjacent domains of ‘growth’ and ‘non-growth’, and that either uniform growth or uniform restriction of growth inhibits such morphogenesis. It also may provide an explanation as to how a factor that is thought to restrict growth such as GOB (Aida and Tasaka, 2006) promotes leaflet initiation in tomato and tooth growth in Arabidopsis (Kawamura et al., 2010).

Context-specific patterning

Auxin and NAM/CUC genes regulate the patterning of leaf margins in both simple Arabidopsis leaves and compound tomato leaves. However, in Arabidopsis leaves marginal serrations are formed rather than leaflets or lobes (Hay et al., 2006; Nikovics et al., 2006; Hasson et al., 2011). In tomato, auxin maxima mark the initiation sites of both lobes and leaflets. Whether a leaflet or a lobe will be formed depends on events happening before and after their initiation (Kaplan, 2001). If an initiation event occurs after the leaf has commenced flattening, expansion and determination, as in the case of the terminal leaflet, a lobe or a serration rather than a leaflet will form (Floyd and Bowman, 2010). Laminar growth between initiating marginal structures is an additional process that can lead to the formation of a lobe rather than a distinct leaflet (Bharathan et al., 2002; Berger et al., 2009).

In tomato, auxin response maxima were normally generated during primary leaflet initiation in pFIL≫miR164 leaves and in the vasculature, but in the margins of the terminal and primary leaflets the signal expanded to throughout the margins, similar to the case in Arabidopsis leaves (Kawamura et al., 2010; Bilsborough et al., 2011). Thus, with respect to the regulation of marginal structure initiation by auxin and GOB/CUC2, the tomato terminal and primary leaflets behave similarly to the Arabidopsis simple leaf. This raises the possibility that these may be equivalent structures. The mutual regulation of CUC and auxin is therefore tailored to the species and the developmental context.

Regulation of E expression and activity

We provide evidence that E physically interacts with two recently identified auxin receptor homologs in tomato: SlTIR1 and SlAFB6. This indicates that the activity of E is probably regulated by an auxin-dependent mechanism analogous to the SCFTIR1/AFB-mediated degradation pathway that has been well described in Arabidopsis (Calderon-Villalobos et al., 2010). Furthermore, the E protein was stabilized by a mutation in domain II of the protein and by the inhibition of auxin or proteasome activity (Figure 3), indicating that this domain is responsible for auxin-dependant degradation, as described for Arabidopsis Aux/IAAs (Gray et al., 2001; Ramos et al., 2001; Reed, 2001; Tan et al., 2007). Interestingly, E appears to have differential binding affinities to SlTIR1 and AFB6 (Figure 3), introducing another layer of control that could be important for understanding the molecular mechanism regulating E activity and leaf marginal patterning. Further studies are required to determine the relative contribution of these two receptors to leaf development.

The expression of E between initiating leaflets, where it functions to inhibit auxin-regulated lamina growth and the effect of GOB on E transcript levels, suggests that in addition to the regulation at the protein stability level, E expression is also regulated at the transcription level. In agreement, Aux/IAAs were originally identified by the rapid induction of their transcription by auxin and their transcripts show differential tissue-specific accumulation (Abel and Theologis, 1996; Hagen and Guilfoyle, 2002; Wang et al., 2010; Rademacher et al., 2012). Thus, E activity is regulated at multiple levels, which possibly enables the precise yet flexible regulation of auxin response.


The current and previous results indicate that leaflet formation requires: (i) a factor that defines localized growth (auxin); (ii) an adjacent area that lacks the growth promoting factor (through inhibited auxin response, achieved by E and GOB activity); and (iii) a factor that defines the boundary between areas of ‘growth’ and ‘no growth’ (achieved by defined GOB expression). Both GOB and E promote leaflet separation, restrict auxin response and are required for secondary leaflet morphogenesis, but they regulate primary leaflet initiation via redundant pathways. GOB probably affects auxin distribution, and may also affect auxin response, possibly through its effect on E expression (Figure 8k).

Experimental procedures

Plant material

Tomato seeds (Solanum lycopersicum cv M82, sp) were sown in a commercial nursery and grown in the field or in a glasshouse under natural daylight. For in situ hybridization and expression analyses, plants were grown in a growth room.

All transgenic genotypes described were generated by the LhG4 transactivation system (Moore et al., 1998). e alleles and GOB genotypes have been described by Berger et al. (2009). The e-2 allele was used for genetic interactions and for analysis of the DR5::VENUS signal, and the e-3 allele was used for the quantification of E and GOB expression.

Cloning and plant transformation

The pDR5rev:3XVENUS-N7 construct, kindly provided by Elliot Meyerowitz and Marcus Heisler (California Institute of Technology; Heisler et al., 2005), was subcloned into the binary pART27 vector (Gleave, 1992), and introduced into the M82 sp tomato background. Ten independent kanamycin-resistant transgenic lines were selected and examined. Further analysis was performed with one selected line.

To generate op:iaaM and to introduce it in tomato, we cloned the iaaM gene (Romano et al., 1995) downstream to an Operator array (Moore et al., 1998). At least 12 independent kanamycin-resistant transgenic lines were examined.

To generate op:EdIIm, we amplified the E gene from cDNA (M82 background), and used assembly PCR to substitute two Proline amino acids at positions 201 and 202 of the protein sequence with two Serine amino acids. This fragment was cloned downstream to an Operator array (Moore et al., 1998). The GUS sequence was amplified from the pRITAI vector (Eshed et al., 2001) and used in assembly PCR to generate the E-GUS and EdIIm-GUS fusions. At least seven independent lines of each transgene were crossed to the pBLS and pFIL driver lines, and a representative line was selected for further analysis. In the case of E and E-GUS, 10 and nine lines, respectively, were crossed to driver lines, but no aberrant phenotype or GUS expression could be detected in any line despite the presence of the transgene, as verified by PCR analysis and kanamycin resistance. SlTIR1 and SlAFB6 cDNA (from clones LEFL1009AG09 and LEFL1073CG05, respectively; Kazusa DNA Research Institute) were fused with a C-terminal tag containing three c-myc epitope repeats. The myc-tagged sequences were inserted into the binary vector pEarlyGate100 (Earley et al., 2006) and transformed into the A. tumerfaciens strain AGL1. ENTIRE was cloned into the vector pGEX-4T-3 (adding an N-terminal GST-tag; GE Healthcare, http://www.gelifesciences.com) using the restriction site BamHI and XhoI. The GST-AtIAA7 construct was previously described (Parry et al., 2009). The primers used for cloning are given in Table S1.

Hormone treatments

Auxin (indole-3-acetic acid, IAA; I-2886; Sigma-Aldrich, http://www.sigmaaldrich.com), N-1-naphthylphthalamic acid (NPA; PS-343; Sigma-Aldrich) or PEO-IAA (kindly provided by Ken-ichiro Hayashi, Okayama University, Japan) were dissolved in dimethyl sulfoxide (DMSO). For an examination of the influence of NPA and PEO-IAA on leaf development and DR5::VENUS expression in pFIL≫GOBm leaves, sterile seeds were planted on an MS medium [0.44% w/v MS (M-0222.0050; Duchefa, http://www.duchefa.com), 1% sucrose w/v, 0.75% plant agar (Duchefa)], and after germination vegetative shoot apices were transplanted to MS medium containing 10 μm NPA, 400 μm PEO-IAA or DMSO. For microapplication, vegetative tomato shoot apices were dissected as described by Reinhardt et al. (2000), leaving either P1 and P2 or P1–P3 primordia, the SAM and a subapical region intact. IAA (1 mm) was applied as described by Reinhardt et al. (2000), except that the lanolin paste was pre-warmed to 60°C and contained 2% paraffin. For qRT-PCR analysis of IAA- or NPA-treated plants, vegetative shoot apices containing the SAM and five leaf primordia were dissected and transferred to MS medium containing 300 μm IAA, 10 μm NPA or DMSO, and after 48 h the specific shoot apices or leaf primordia were collected.

Tissue collection, RNA analysis and statistical analysis

Tissue collection and qRT-PCR analysis were performed as described by Shleizer-Burko et al. (2011). Values are means of three biological repeats, each containing 20 seedlings. 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 level of P < 0.05. RNA in situ hybridization was performed as described (Shani et al., 2010). Primer sequences used for the qRT-PCR analysis and for the E probe are detailed in Table S1.

Imaging, microscopy and GUS staining

Dissected whole-leaf primordia were placed into drops of buffer (75 mmn-propylgallate, 60% glycerol) on glass microscope slides and covered with cover slips. The pattern of VENUS expression was detected by a confocal laser scanning microscope (CLSM, model LSM510; Zeiss, http://www.zeiss.com), as described by Shani et al. (2010). Optical sections from successive focal planes of each primordial tract region were collected and projected as a reconstructed three-dimensional image using the LSM image browser (version Fluorescence stereoscope images were captured by an Olympus SZX12 zoom stereoscope using a GFP florescence filter set for VENUS. In situ hybridization slides were photographed with an Olympus 1 × 81 microscope using cellr software, as described by Berger et al. (2009). Images analyzing the early developmental stages of the whole-leaf primordia were captured using an Olympus SZX7 binocular microscope. Scanning electron microscopy was performed using a JEOL 5410 LV microscope, as described by Brand et al. (2007).

GUS staining was performed as described by Ori et al. (2000). For N-carbobenzoxyl-l-leucinyl-l-leucinyl-l-norleucinal (MG-132, ENZO BML-PI102) and PEO-IAA treatment, shoot apices of 2-week-old seedlings grown on MS medium were transferred to a solution with 200 μm MG-132, 500 μm PEO or DMSO and incubated for 7 h in a vacuum. During incubation, the vacuum was released every 15 min in the first hour, and then every hour.

Protein preparation and GST pull-down assay

GST-ENTIRE and GST-AtIAA7 constructs were transformed into Escherichia coli strain BL21 and expression was induced with 1 mm isopropyl-β-d-thiogalactoside (IPTG). Cells were lysed by sonification in pull-down buffer (20 mm Tris-HCl, pH 8.0, 200 mm NaCl, 5 mm DTT, 1 mm PMSF, 10 μm MG-132, and Roche Complete Mini tablets). The lysates were incubated with glutathione agarose (Sigma-Aldrich) for 1 h at 4°C, then washed and re-suspended with pull-down buffer. SlTIR1-myc and SlAFB6-myc constructs were infiltrated into N. benthamiana leaves as described by Sparkes et al. (2006). Whole protein extracts from infiltrated leaves were prepared by grinding in a buffer of 50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 10% glycerol, 0.1% NP-40, 1 mm PMSF, 10 μm MG-132, and Roche Complete Mini tablets. SlTIR1-myc and SlAFB6-myc pull-downs with GST-ENTIRE and GST-AtIAA7 were carried out as described (Prigge et al., 2010), except pull-down buffer as described here was used instead of binding buffer and 100 μl of SlTIR1-myc or SlAFB6-myc protein extract was combined with 30 μl of GST-ENTIRE or GST-AtIAA7 purified protein.

Phylogenetic analysis

Protein sequences were aligned using the Muscle algorithm (Edgar, 2004). Phylogenetic analysis was performed using mrbayes 3.1.2 (Ronquist and Huelsenbeck, 2003) with 200 000 generations and parameters aamodelpr = mixed, nst = 6 and rates = invgamma. Tomato TIR1/AFB amino acid sequences were derived from the accession numbers listed below; sequences of other species have been described by Parry et al. (2009). TIR1 and AFB1 in A. thaliana and Brassica rapa were constrained to account for the high divergence of AFB1, consistent with the results of Parry et al. (2009).

Accession numbers

Sequence data used in this study can be found in the GenBank/EMBL database or SGN under the following accession numbers: ENTIRE (HM210878; SolyC04g076850); GOBLET (HM210879; SolyC07g062840); EXP (SGN-U346908); AtIAA7 (AT3G23050); SlTIR1 (AK320427); and SlAFB6 (AK324204); SlAFB4 (GU079663).


We would like to thank Marcus Heisler and Elliot Meyerowitz for the pDR5rev:3XVENUS-N7 construct, Ken-ichiro Hayashi for PEO-IAA, Kazusa DNA Research Institute for cDNA clones, Yogev Burko, Eduard Belausov and Dor Russ for technical help, Naomi Nakayama for help with microapplication, and Leor Eshed Williams and members of our groups for fruitful discussions and critical reading of the article. This work was supported by grants from BARD (no. IS 04140-08C) to NO and ME, and from ISF (no. 60/10) and the Israeli Ministry of Agriculture (no. 837-0055-09) to NO. HB is funded in part by a Kaye Einstein Scholarship. ES was funded in part by a Vaadia–BARD Postdoctoral Fellowship Award, number FI-431-10, and by the Machiah Foundation/Jewish Community Federation Fellowship Program.