Evaluating auxin distribution in tomato (Solanum lycopersicum) through an analysis of the PIN and AUX/LAX gene families

Authors


(fax +1 607 254 1242; e-mail cc283@cornell.edu).

Summary

The temporal and spatial control of auxin distribution has a key role in the regulation of plant growth and development, and much has been learnt about the mechanisms that influence auxin pools and gradients in vegetative tissues, particularly in Arabidopsis. For example polar auxin transport, mediated by PIN and AUX/LAX proteins, is central to the control of auxin distribution. In contrast, very little information is known about the dynamics of auxin distribution and the molecular basis of its transport within and between fruit tissues, despite the fact that auxin regulates many aspects of fruit development, which include fruit formation, expansion, ripening and abscission. In addition, functional information regarding the key regulators of auxin fluxes during both vegetative and reproductive development in species other than Arabidopsis is scarce. To address these issues, we have investigated the spatiotemporal distribution of auxin during tomato (Solanum lycopersicum) fruit development and the function of the PIN and AUX/LAX gene families. Differential concentrations of auxin become apparent during early fruit growth, with auxin levels being higher in internal tissues than in the fruit pericarp and the pattern of auxin accumulation depended on polar transport. Ten tomato PIN (SlPIN1 to 10) and five AUX/LAX (SlLAX1 to 5) genes were identified and found to display heterogeneous expression patterns, with tissue and developmental-stage specificity. RNAi-mediated co-silencing of SlPIN4 and SlPIN3 did not affect fruit development, which suggested functional redundancy of PIN proteins, but did lead to a vegetative phenotype, and revealed a role for these genes in the regulation of tomato shoot architecture.

Introduction

Auxin mediates growth and development by its role in the coordination of cell division, expansion and differentiation. Accordingly, the spatiotemporal control of auxin levels is of fundamental importance in developmental events such as organ initiation, embryogenesis and differential root and stem growth (Vanneste and Friml, 2009). Auxin also controls many aspects of fruit development, which include the sequential stages of fruit formation, expansion, ripening and abscission (Gillaspy et al., 1993; Srivastava and Handa, 2005). Although the mechanisms by which auxin exerts this control are still poorly understood, the importance of precise regulation of auxin levels and activity during fruit ontogeny has been emphasized by several studies.

During fruit set, the growth of an otherwise static ovary is stimulated after successful pollination and fertilization. The importance of regulated auxin accumulation during this process is illustrated by experiments that show that artificially increasing in auxin levels in the ovary (Alabadi et al., 1996; Rotino et al., 1997; Vivian-Smith and Koltunow, 1999), or disruption of auxin transport (Kim et al., 1992; Dorcey et al., 2009; Serrani et al., 2010) can bypass fertilization and lead to the development of parthenocarpic (seedless) fruits. Modification of auxin sensitivity by down-regulation of auxin response regulators, such as Aux/IAAs or auxin response factors (ARFs), can lead to similar parthenocarpic phenotypes (Wang et al., 2005; Goetz et al., 2007; de Jong et al., 2009a).

After the initiation of ovary growth, fruit undergo cell division and expansion in the carpel wall and in the placental tissue that surrounds the seeds (Gillaspy et al., 1993). Early observations indicated the existence of auxin gradients during fruit growth, with the highest levels of auxin in the seeds (Gustafson, 1939), suggesting that the seeds are the predominant source of auxin which moves to other tissues to promote cell division and expansion (Nitsch, 1950; Ozga et al., 2002). However, whilst increasing levels of evidence suggests that after fruit set, precise spatial and temporal synthesis, transport and action of auxin are required for fruit development (Gillaspy et al., 1993; Sundberg and Østergaard, 2009), the dynamics of these processes are mostly unknown, in particular the molecular basis behind auxin gradients and localized auxin maxima.

A major mechanism that regulates auxin distribution is polar auxin transport mediated by PIN and AUX/LAX proteins, which control cellular auxin efflux and influx respectively (Vanneste and Friml, 2009). Their asymmetric distribution across cells and tissues results in directional auxin flow and the establishment of auxin gradients (Wisniewska et al., 2006; Bainbridge et al., 2008; Swarup et al., 2008; Petrášek and Friml, 2009). Arabidopsis contains eight PIN proteins (Paponov et al., 2005), of which several are involved directly in creation of auxin gradients that control diverse developmental processes such as embryogenesis, organ initiation, vascular tissue differentiation and tropisms (Petrášek and Friml, 2009). Similarly, there are four AUX/LAX proteins in Arabidopsis, with the best characterized, AUX1, mediating high-affinity auxin uptake (Kerr and Bennett, 2007; Bainbridge et al., 2008; Swarup et al., 2008).

In this study, we seek to elucidate the molecular bases of auxin transport in tomato (Solanum lycopersicum), a model organism for the study of sympodial growth, compound leaves and fleshy fruit development (Giovannoni, 2004; Kimura and Sinha, 2008). While PIN and AUX/LAX genes have been described in several species (Schrader et al., 2003; Schnabel and Frugoli, 2004; Paponov et al., 2005; Carraro et al., 2006; Dal Cin et al., 2009), information on the genes that control auxin fluxes in tomato during organ growth and development is extremely limited. Recently, the expression of three PIN genes has been examined in tomato fruit (Nishio et al., 2010). However, this study did not analyze comprehensively the complete PIN family throughout fruit ontogeny nor did it address function. Moreover, nothing has been reported of the tomato AUX/LAX family to allow a broader understanding of auxin transport.

Here, we present the spatial and temporal patterns of auxin accumulation and characterize the complexity and tissue-specific expression of the PIN and AUX/LAX gene families, with particular emphasis on fruit growth after initiation. We also describe a transgenic line produced via RNAi-mediated silencing of PIN gene expression in order to investigate the role of polar auxin transport during vegetative growth and fruit development.

Results

Auxin accumulation in tomato fruit is tissue and developmental-stage specific and is dependent upon polar transport

To assess the spatiotemporal patterns of auxin accumulation during tomato fruit development, indole-3-acetic acid (IAA) levels were quantified in tissues dissected from fruit at different developmental stages (Figure 1). Fruit development was characterized by rapid growth from anthesis to 21 days post anthesis (dpa), followed by slower growth until fruit reached their final size. At 37–38 dpa, the fruit reached breaker stage marking the onset of ripening. At 5 dpa, during the early exponential phase of growth, IAA levels were similar in the pericarp and the internal tissues, which comprise seeds, placental tissue and the central columella, which at this stage are difficult to separate. However, clear differences were seen between tissues at 21 dpa, after a period of sustained growth. At this stage IAA concentration was highest in the seed/locular tissue, lower in the placenta and lower still in the pericarp. The same pattern, and highest overall levels, of IAA accumulation were observed at 34 dpa, when the fruit had reached its final size. IAA levels then declined in breaker fruit and were undetectable in the placenta and pericarp (Figure 1).

Figure 1.

 Tissue-specific indole-3-acetic acid (IAA) levels at different stages of fruit development.
IAA was quantified in internal tissues (i) and pericarp (p) at 5 days post anthesis (dpa), and in seeds/locular tissue (s), placenta (pl), and pericarp (p) at 21 dpa, 34 dpa and 38 dpa (breaker stage). Harvested tissues are indicated in the inset (scale bar = 10 mm). Values represent the average ± standard deviation (SE) ( 3; *not detected).

To analyze auxin distribution in a more tissue- and cell-specific manner, a DR5::GUS reporter line (Dubrovsky et al., 2008) was used (Figure 2). DR5 is an auxin-responsive promoter, and β-glucuronidase (GUS) expression therefore indirectly reflects auxin concentration (Ulmasov et al., 1997). At 5 dpa, little or no GUS activity was seen in the fruit, although GUS staining was prominent in the vasculature of the fruit pedicel (Figure 2a) where it was maintained throughout fruit development. At 9–10 dpa, strong localized GUS staining was visible in the funiculus, the connective tissue that attaches the seeds to the rest of the fruit, and in the vascular strands that connect the fruit with the pedicel (Figure 2b,c). GUS activity in the funiculus was more prevalent at 16 dpa, and was also present in the vasculature of the placenta (Figure 2(d)). From this stage onwards, GUS activity remained in the funiculus, the locular tissue that surrounds the seeds, and the developing embryo (Figure 2e–h). No GUS activity was seen in wild-type plants (Figure S1).

Figure 2.

DR5::GUS expression during tomato fruit development.
GUS staining in fruit at 5 (a), 10 (b, c), 16 (d), 22 (e, f) and 34 (g, h) days post anthesis (dpa). GUS staining is indicated in the pedicel (p), vasculature (v), embryo (e), locular tissue (l) and funiculus (f). Effect of naphthylphthalamic acid (NPA) on DR5::GUS expression in 9 (i, j) and 15 (k, l) dpa fruit from control plants (i, k) or plants treated with 50 μm NPA (j, l). 20 dpa fruit collected after treatment with 0 μm (m), 5 μm (n), 20 μm (o) or 50 μm (p) NPA. Scale bars = 1 mm (e, f), 3 mm (a–d, h), or 10 mm (g, i–p).

The pericarp and placental tissue showed no GUS staining throughout fruit development except in vascular strands, and GUS activity was not seen in fruit younger than 9 dpa. To determine the influence of polar auxin transport on the distribution of DR5::GUS expression, plants were treated with the auxin transport inhibitor N-1-naphthylphthalamic acid (NPA). NPA caused a substantial, dose-dependent increase in DR5::GUS expression in the placenta and columella (Figure 2i–p), which indicated increased auxin activity. NPA treatment resulted in parthenocarpy and thickening of the fruit pedicel distal to the abscission zone. Fruit undergoing rapid growth or that had set after the time of NPA exposure were particularly affected (Figure 2i–l), while fruit close to the mature green stage and that were, therefore, at a slower phase of growth, showed no change in DR5::GUS expression after NPA treatment (data not shown).

The lack of GUS staining in 5 dpa fruit, was in contrast with the direct measurement of IAA at this stage (Figure 1). Therefore, a second auxin-responsive reporter, DR5rev::mRFPer (Gallavotti et al., 2008) with the DR5rev promoter (Benkováet al., 2003; Friml et al., 2003) coupled to a monomeric red fluorescent protein (mRFPer), was used (Figure 3). RFP was detected 6 days before anthesis (Figure 3b, c) in vascular strands of the ovary, as well as in ovules, confined to an area likely corresponding to the micropylar pole of the embryo sac (Gasser and Robinson-Beers, 1993). At 2 days before anthesis, RFP fluorescence remained prominent in the embryo sac but had also spread to other tissues in the ovule, particularly the integument (Figure 3e, f). A similar distribution of RFP was observed at anthesis, with pronounced expression in the embryo sac and ovule surface (Figure 3i, k, l). Fluorescence was also seen occasionally on both the inner and outer surface of the ovary wall (Figure 3h, j). After fertilization, fluorescence was widespread throughout the ovules and was no longer confined to the embryo sac (Figure 3n, o). In 6 dpa fruit, the RFP signal was strong in the funiculus (Figure 3q, r) and in the outer layers of the placenta that surrounds the seeds (Figure 3t, u).

Figure 3.

DR5rev::mRFPer expression in tomato ovaries and fruit.
Confocal images of red fluorescent protein (RFP) fluorescence at 6 (a–c) and 2 (d–f) days before anthesis, at anthesis (g–l), and 2 (m–o) and 6 (p–u) days post anthesis. Images are of ovaries (b, e, h, k, n), sections of ovules (c, f, i, o), the ovule surface (l), the ovary wall (j), a developing seed (q), the seed funiculus (r), and placental tissue (s–u). Scale bars = 5 mm (a, d, g, m, p), 500 μm (e, h, k, n, q), 200 μm (b, r–t) or 50 μm (c, f, i, l, o, u). Labels indicate seed (s), placenta (pl) and pericarp (p).

PIN and AUX/LAX genes are expressed differentially in specific tissues during fruit development

To investigate the molecular basis for the observed distribution of auxin, the expression of the PIN and AUX/LAX auxin transport facilitators was analyzed in developing fruit. Ten PIN genes (SlPIN1 to 10) were identified in the tomato unigene database (http://solgenomics.net/tools/blast/dbinfo.pl) and in the provisional release of the tomato genome at the SOL Genomics Network (SGN; Mueller et al., 2005, 2009). Phylogenetic analysis (Figure 4) revealed a similar organization for the tomato and Arabidopsis PIN proteins (Paponov et al., 2005) and orthologous relationships for some proteins between both species (PIN1, PIN2, PIN6, PIN8). Therefore, we named the tomato proteins based on their relationships to Arabidopsis PINs. We have re-named three recently described tomato PIN genes (Nishio et al., 2010) after agreement with the authors (Y. Kanayama, personal communication; Table S1). The phylogenetic analysis defined two subgroups, the first, which contained proteins with a long central hydrophilic loop between transmembrane domains at the N- and C- termini, comprised Arabidopsis AtPIN1 to 4 and AtPIN7 plus tomato SlPIN1 to 4, SlPIN7 and SlPIN9. The other group, characterized by a very short hydrophilic loop, contained tomato SlPIN5, SlPIN6, SlPIN8 and SlPIN10 and Arabidopsis AtPIN5, AtPIN6 and AtPIN8.

Figure 4.

 Phylogenetic tree of tomato (Sl) and Arabidopsis (At) PIN proteins.
Neighbour-joining tree based on full-length protein alignment that shows percentage bootstrap support at each node (= 1000). Scale bar = number of amino acid substitutions per site.

The expression of SlPIN genes during fruit development was analyzed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) in seeds/locular tissue, placenta and pericarp, and compared with their expression in vegetative tissues and flowers (Figure 5). Expression of eight SlPIN genes was detected in fruit, but none of the genes was fruit specific as they also showed expression in vegetative tissues. The SlPIN genes that showed highest relative expression in fruit could be divided into three classes based on preferential expression in seeds/locular tissue (SlPIN5), placenta (SlPIN1, SlPIN4, SlPIN7, SlPIN8), or pericarp (SlPIN6), relative to the other fruit tissues (Figure 5a–c). SlPIN genes expressed in the placenta (Figure 5b) generally showed higher expression during early stages of fruit development and lower expression during ripening. An exception was SlPIN7, which changed from preferential expression in the placenta in expanding fruit to the pericarp at breaker stage and ripe fruit. SlPIN5 expression peaked in seeds/locular tissue at 14 dpa and declined gradually to almost undetectable levels during ripening (Figure 5a). SlPIN6, the only SlPIN gene preferentially expressed in pericarp, showed highest expression at 5 dpa and much lower levels of expression later in development. The remaining genes (Figure 5d) were either not detected (SlPIN2 and SlPIN10) or displayed considerably lower expression in fruit than in vegetative tissue or flowers (SlPIN3 and SlPIN9). SlPIN2, SlPIN9 and SlPIN10 were expressed preferentially in roots whereas SlPIN3 expression was highest in flowers.

Figure 5.

 Expression of SlPIN genes.
PIN expression was quantified by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) in vegetative tissues (yl: young leaf, ml: mature leaf, r: root, st: stem), flowers (fl) and ovaries (ov) at anthesis, and fruit tissues (i: internal tissues; p: pericarp; pl: placenta; s: seeds/locular tissue) harvested at 5, 14, 21, 34 days post anthesis (dpa), breaker (Br) and red stage. Genes are divided into four categories based on preferential expression in seeds/locular tissue (a), placenta (b) or pericarp (c), compared to other fruit tissues. Category (d) contains genes with low expression in fruit relative to flowers or vegetative tissue. The average expression of each gene, calculated relative to the first biological replicate of roots ± standard error (SE) is plotted ( 3).

Five AUX/LAX family genes were identified in the provisional release of the tomato genome and named SlLAX1 to 5. A phylogenetic tree of tomato SlLAX (like-AUX1) and Arabidopsis AUX/LAX proteins (Figure 6a) revealed two distinct subfamilies, with tomato SlLAX proteins belonging to each and clustering with their Arabidopsis homologs. The Arabidopsis auxin influx facilitator, AtAUX1 (Bennett et al., 1996), groups with tomato SlLAX1 and SlLAX4, which are closely related to each other (89% amino acid identity). Arabidopsis AtLAX2 also has two homologs in tomato (SlLAX2 and SlLAX5) whereas AtLAX3 is orthologous to tomato SlLAX3.

Figure 6.

 Phylogenetic tree of tomato (Sl) and Arabidopsis (At) AUX/LAX proteins and expression of SlLAX genes.
(a) Neighbour-joining tree based on full-length protein alignment that shows percentage bootstrap support at each node (= 1000). Scale bar = number of amino acid substitutions per site.
(b) AUX/LAX expression was quantified by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) in vegetative tissues (yl: young leaf, ml: mature leaf, r: root, st: stem), flowers (fl) and ovaries (ov) at anthesis, and fruit tissues (i: internal tissues; p: pericarp; pl: placenta; s: seeds/locular tissue) harvested at 5, 14, 21, 34 days post anthesis (dpa), breaker (Br) and red stage. The average expression of each gene, calculated relative to the first replicate of roots ± standard error (SE) is plotted ( 3).

qRT-PCR analysis revealed that SlLAX1, SlLAX2 and SlLAX3 showed the highest relative expression in fruit (Figure 6b). The expression of SlLAX1 and SlLAX3, which were preferentially expressed in the placenta and pericarp respectively, increased gradually from 14 dpa to breaker stage before declining in ripe fruit. In contrast, SlLAX2 showed a gradual decrease in expression from 5 dpa onwards. SlLAX4, which was most strongly expressed in roots, stems and flowers, and SlLAX5, which was preferentially expressed in stems and leaves showed relatively little expression in fruit.

Suppressed expression of SlPIN4 and SlPIN3 alters shoot architecture with no major effects on fruit development

To investigate the potential physiological significance of PIN-mediated auxin transport in tomato, RNAi-mediated suppression of multiple SlPIN genes was performed using SlPIN4 as the primary silencing target. SlPIN4 was chosen as it was preferentially expressed during the fruit cell-expansion phase especially in the placenta, a tissue previously suggested to be the IAA export route from the seed to the plant (Hocher et al., 1992) and where DR5::GUS analysis suggested auxin hyper-accumulation after NPA treatment (Figure 2). In addition, SlPIN4 had more ESTs in the tomato EST database (http://solgenomics.net/tools/blast/dbinfo.pl) than any other SlPIN gene (19 compared with 5 for SlPIN3 which had the next highest). However, the SlPIN4 region used as the RNAi silencing trigger has significant similarity to SlPIN1, SlPIN3, SlPIN7, and SlPIN9 (Figure S2) potentially permitting silencing of multiple genes and consequently greater modification of auxin transport. Fifteen primary transformants were recovered, 10 of which displayed a phenotype that featured taller, thinner, often reclining stems, and leaf epinasty (Figure 7a, b). Four primary transformants, named SlPIN4-RNAi (1), (2), (3) and (4), were chosen for further analysis. SlPIN4 was silenced to between 7 and 16% of wild-type levels in fruit, and the closely related SlPIN3 to between 33 and 56% (Figure 7e). The expression of other SlPIN genes was not altered significantly in the SlPIN4-RNAi lines, except in SlPIN4-RNAi line (1). This line also showed evidence of moderate silencing of SlPIN1 (58% of wild-type levels), SlPIN7 (44%) and SlPIN9 (76%; Figure S3). For phenotypic characterization of SlPIN4-RNAi lines (2) to (4), individual heterozygous or homozygous plants that carry the transgene were selected in the T1 generation. Segregation analysis of SlPIN-RNAi line (1) suggested a single insertion and so a homozygote was selected in the T2 generation and the phenotype characterized in the T4 generation.

Figure 7.

 Phenotypic characteristics of SlPIN4-RNAi transgenics.
(a) Six-week-old wild-type and three independent T1SlPIN4-RNAi transgenic lines.
(b) The excised sixth-leaf of plants shown in (a).
(c) Phenotypic resemblance of 3-week-old wild-type seedlings treated with 50 μm naphthylphthalamic acid (NPA) for 2 days and SlPIN4-RNAi seedlings.
(d) Rate of fruit set in wild-type and SlPIN4-RNAi lines. Values represent the average rate per week ± standard error (SE) (= 4).
(e) Expression of SlPIN4 (grey bars) and SlPIN3 (white bars) in 7 days post anthesis fruit from wild type and SlPIN4-RNAi transgenics quantified by quantitative reverse-transcription polymerase chain reaction (qRT-PCR). The average expression of each gene, calculated relative to the first biological replicate in wild type ± SE is plotted (= 3). (*t-test P-value < 0.1, **P-value < 0.05, ***P-value < 0.01). Scale bars = 100 mm.

Despite alterations in vegetative growth, reproductive development was not apparently affected in the SlPIN4-RNAi lines and there was no evidence of altered fruit development. Fruit colour, internal structure, seed number, and the onset and progression of ripening were the same as in wild type. In addition, the growth rate and size of the fruit was not affected (Figure S4).

For evaluation of fruit set, SlPIN4-RNAi lines (1) to (4) and wild-type flowers were tagged at anthesis over a 4-week period and self-pollination promoted by mechanical vibration of the inflorescence stem. A significant reduction in fruit set was observed in only two of the transgenic lines, SlPIN4-RNAi (2) and (3), although all four lines had significantly reduced RNA levels (Figure 7d, e). Therefore we could not conclude that fruit set was affected consistently.

Disruptions in auxin transport have been associated previously with parthenocarpy (Serrani et al., 2010). However, SlPIN4-RNAi lines did not show an increased rate of parthenocarpy, regardless of whether or not flowers were emasculated prior to anthesis. In addition, exogenous application of auxin to emasculated ovaries, which can induce parthenocarpy (Serrani et al., 2008), promoted fruit set at a similar frequency as in wild type (Table S2).

To determine whether auxin distribution in fruit of the SlPIN4-RNAi lines had been altered, we examined DR5 activity in the SlPIN4-RNAi line (1) crossed with the DR5::GUS reporter (Figure S5). Fruit from SlPIN4-RNAi plants did not show any defects in DR5::GUS expression and NPA treatment induced GUS accumulation in the placenta in a similar manner as in wild type.

To investigate whether the vegetative phenotype of the SlPIN4-RNAi lines was caused by alteration of auxin transport, we compared them with wild-type seedlings treated with NPA. Wild-type seedlings watered for 2 days with 50 μm NPA closely resembled the RNAi lines, characterized by elongated shoots and epinastic leaves (Figure 7c).

SlPIN4-RNAi leaves were epinastic, and leaf margins folded inwards and gave plants a ‘wilted’ appearance (Figures 7a–c and 8). Calculation of the leaf-flattening index (de Carbonnel et al., 2010), confirmed that leaves from SlPIN4-RNAi plants were significantly more curved than those of the wild type (Figure 8c). However, leaf length and leaflet number were similar and no defects were seen in leaf indentation or vascular patterning.

Figure 8.

 Leaf epinasty in SlPIN4-RNAi transgenics.
Profile view of the third leaf of 1-month-old wild-type (a) and SlPIN4-RNAi plants (b). Insets show the abaxial surface of a detached terminal leaflet. Leaf-flattening index of SlPIN4-RNAi and wild-type terminal leaflets (c). Values represent average ± standard error (SE) (= 8). (***t-test P-value < 0.01). Scale bars = 50 mm.

In addition, SlPIN4-RNAi seedlings had a slender phenotype with taller, thinner stems that had a tendency to bend, which gave the plant a reclined appearance (Figure 9a, b). Whereas wild-type seedlings grew upwards, SlPIN4-RNAi stems showed variation in growth angle (Figure 9c) and a significant increase in the length of hypocotyls and the first two internodes (Figure 9d). Transverse sections revealed that although SlPIN4-RNAi stems had the same number and arrangement of vascular bundles as wild type, they had a contracted, less-developed layer of secondary vascular tissues (Figure 9e,f). SlPIN4-RNAi plants also exhibited increased apical dominance and reduced axillary bud outgrowth (Figure 9g,h). Images of representative plants show that whilst wild-type axillary buds had shown outgrowth, equivalent SlPIN4-RNAi buds failed to develop to the same extent (Figure 9i, j; see arrows). This phenotype persisted until floral transition, after which buds developed normally.

Figure 9.

 Shoot architecture phenotype of SlPIN4-RNAi transgenics.
Wild-type (a) and SlPIN4-RNAi plants (b) displaying the bent-stem phenotype. Angle between the base of the stem and the shoot apical meristem expressed as the deviation from 90° (c). Values represent average ± standard error (SE) (= 10). Hypocotyl and internode length of wild-type and SlPIN4-RNAi plants (d). Values represent average ± SE ( 20). Transverse sections of the first internode of wild-type (e) and SlPIN4-RNAi plants (f) that show vascular bundles (vb), secondary vascular tissues (sv), pith (p) and cortex (c). Axillary bud outgrowth in wild-type (g, i) and SlPIN4-RNAi plants (h, j). Arrows indicate axillary buds at the ninth and tenth leaf nodes. Scale bars = 1 mm (e, f), 25 mm (i, j) or 100 mm (a, b, g, h). (***t-test P-value < 0.001).

Discussion

Tissue-specific auxin distribution in fruit depends on polar transport

IAA quantification and the use of auxin-responsive reporter lines revealed spatiotemporal variations in auxin distribution during fruit development. Similar IAA levels in internal tissues and pericarp at 5 dpa (Figure 1) suggest that auxin is involved in the regulation of rapid cell division throughout the fruit during early exponential growth. The elevated auxin concentration in the pericarp at this stage is likely to increase the expression of auxin-induced cell-expansion effectors, some of which subsequently show peak expression in the pericarp during the rapid cell-expansion stage (Cataláet al., 2000). It is also consistent with a model by which an initial peak of auxin in the pericarp promotes consequent cell expansion by stimulation of gibberellin biosynthesis (Gillaspy et al., 1993; Dorcey et al., 2009; de Jong et al., 2009b).

The DR5 reporters added previously unavailable spatial resolution to our knowledge of auxin activity in fruit (Figures 2 and 3). During tomato ovary development, localized DR5rev::mRFPer expression in the embryo sac was similar to DR5::GFP expression in Arabidopsis, in which auxin controls ovule cell specification and tissue patterning (Pagnussat et al., 2009). DR5rev::mRFPer expression suggested the presence of additional domains of auxin activity in tomato, such as the integument and the inner and outer surfaces of the ovary wall (Figure 3a–o). At 6 dpa, towards the end of the cell-division phase and the start of rapid cell expansion, auxin activity in the vascular-rich funiculus may reflect the development of auxin transport streams that emanate from the seeds, whilst auxin in the outer layer of placental cells suggests auxin involvement in promotion of placental expansion to surround the seeds and fill the locular cavity (Figure 3p–u).

The internal-to-external IAA gradient, apparent during the cell-expansion stage (Figure 1), is consistent with previous studies (Gustafson, 1939; Varga and Bruinsma, 1976; Hocher et al., 1992), and with the seeds/locular tissue being an important site of auxin biosynthesis (de Jong et al., 2009b). Indeed, auxin-induced gene expression during the cell-expansion stage, monitored by DR5::GUS, suggested that there was auxin transport from the seeds, via the funiculus and vasculature of the placenta to the fruit pedicel and parent plant (Figure 2). Intense GUS staining in the placenta following NPA treatment (Figure 2), probably due to an inhibition of auxin export from the fruit, implicates polar transport in maintainance of auxin flux across the placenta and abscission zone. Highest auxin accumulation in the seeds when fruit have reached their maximum size (Figures 1 and 2f, g) is likely related to auxin regulation of the final phase of embryo development (Gillaspy et al., 1993).

Basipetal polar auxin transport from the fruit to the parent plant via the pedicel has been shown previously, and is believed to prevent premature abscission (Banuelos et al., 1987; Bangerth, 2000; Serrani et al., 2010). Auxin activity in the funiculus (Figures 2 and 3), which like the pedicel is a site of abscission at the end of fruit development (Berry and Bewley, 1991), suggests that auxin may function analogously in prevention of premature seed abscission.

Although broadly in agreement, there were some discrepancies in the patterns of auxin accumulation portrayed by direct IAA measurement and DR5 activity. Notably, IAA levels in the pericarp were not reflected by DR5::GUS activity. This may be due to low sensitivity, as exogenous IAA was able to induce GUS activity in the pericarp, although only associated with vascular strands (data not shown). The DR5 approach also has other limitations, such as its dependency on the expression of the auxin signal transduction machinery (Normanly, 2010). Novel reporters that relate more directly to auxin concentration by monitoring an early step of auxin signalling may resolve this discrepancy (Vernoux et al., 2011).

Temporal and spatial distribution of PIN and AUX/LAX expression suggests that their coordinated action regulates auxin transport during fruit development

The PIN and AUX/LAX families, with 10 and five members respectively, show similar organization in tomato as in other dicots (Hoyerováet al., 2008; Křeček et al., 2009). Orthology between Arabidopsis and tomato PIN proteins was often reflected in expression. For example, both AtPIN2 and SlPIN2 displayed preferential expression in roots (Figure 5d; Paponov et al., 2005).

Several PIN genes and SlLAX1 were preferentially expressed in the placenta (Figures 5b and 6b) suggesting complex coordinated redistribution of auxin in this tissue. With the exception of SlPIN7 and SlLAX3, PIN and AUX/LAX genes were more abundantly expressed during fruit growth than ripening (Figures 5 and 6b). Their decline in expression may be related to a reduced rate of basipetal auxin transport at the end of fruit development (Else et al., 2004) and could signal the beginning of abscission processes in the funiculus and pedicel.

Only two PIN genes showed divergent tissue specificity in fruit: SlPIN5 was preferentially expressed in the seeds/locular tissue, where auxin accumulates during fruit expansion; and SlPIN6 in the pericarp, during the early stage of fruit growth and coincided with a peak of IAA levels (Figure 5a, c). Interestingly, they correspond to a subclade of PINs that are more likely to be targeted to the endoplasmic reticulum than to the plasma membrane (Křeček et al., 2009; Mravec et al., 2009), suggesting their involvement in intracellular auxin homeostasis rather than in cell-to-cell transport. Further studies to alter SlPIN5 and SlPIN6 expression will help address whether subcellular auxin compartmentalization is a mechanism for the regulation of auxin availability during seed and pericarp development.

SlLAX3 was also expressed preferentially in the pericarp (Figure 6b), which suggests a role in the regulation of auxin influx, and in maintaining auxin sink-strength in this tissue in a similar manner to its Arabidopsis ortholog, AtLAX3, which has been shown to create cell-specific auxin sinks (Swarup et al., 2008; Vandenbussche et al., 2010).

SlPIN4-RNAi lines reveal a role for PIN genes in the control of tomato plant architecture

RNAi lines with strongly reduced expression of SlPIN4 and moderate silencing of other PIN genes showed no noticeable alterations in fruit development and no changes in auxin distribution patterns were observed based on DR5::GUS expression (Figure S5), suggesting that complex and compensatory mechanisms regulate auxin homeostasis in the fruit. Functional redundancy has been described amongst PIN genes in Arabidopsis, in which phenotypes of single mutants are often weak or not discernible (Křeček et al., 2009). The overlapping expression of several PIN genes in tomato fruit suggests a similar scenario. Although there was no evidence of compensatory up-regulation of other PIN genes in whole fruit (Figure S3), auxin homeostasis may be maintained either through cross-regulation of PIN expression in specific cell types, changes in targeting (Blilou et al., 2005; Vieten et al., 2005), or by other compensatory mechanisms that involve modification of IAA synthesis and turnover. Future experiments that analyze cell-specific changes in tomato PIN protein expression and localization in the SlPIN4-RNAi lines will clarify whether the compensatory properties of the PIN-dependent auxin distribution network (Blilou et al., 2005) are responsible for stabilization of auxin distribution in SlPIN4-RNAi fruit.

Despite lacking a fruit phenotype, SlPIN4-RNAi lines showed clear alterations in shoot architecture, which, given their resemblance to wild-type plants treated with NPA (Figure 7c), are likely to be due to disruptions in auxin transport. Auxin is involved in the stimulation of cell division in the vascular cambium and PIN-mediated auxin transport is associated with vascular development and the establishment of auxin transport streams (Petrášek and Friml, 2009; Aloni, 2010). Interestingly, SlPIN4-RNAi lines showed under-developed secondary vascular tissues and decreased secondary stem growth (Figure 9e, f) which suggests that there is altered auxin canalization in the main stem and irregularities in auxin flux. This may cause asymmetries of auxin distribution that lead to differential rates of cell elongation, which in turn cause abnormal bending and increased internode length (Figure 9a–d).

Leaf epinasty and reduced bud outgrowth (Figures 8 and 9i, j) in the SlPIN4-RNAi lines may reflect an inability of lateral organs to export auxin. Indeed, auxin export from lateral buds and the establishment of auxin transport streams is considered to be a primary determinant of bud outgrowth (Leyser, 2009). In leaves, reduced auxin export may cause hyper-accumulation and lead to epinastic curvature due to an increased growth of the leaf adaxial side, as has been described in other auxin hyper-accumulating plants (Klee et al., 1987; Romano et al., 1993; Kim et al., 2007).

Conclusions

The dynamic pattern of tissue-specific auxin accumulation through tomato fruit development is likely to be regulated by polar auxin transport. This study provides a transcriptional map for PIN and AUX/LAX genes in tomato fruit, an important first step towards unravelling the networks that control auxin transport during fruit growth. Although PIN and AUX/LAX genes showed clear tissue and developmental specificity of expression, there was often overlap in expression amongst them, suggesting a coordinated action of PIN and AUX/LAX proteins in the establishment of auxin gradients during fruit development.

Co-suppression of two of these genes, SlPIN3 and SlPIN4, did not affect fruit development, which suggested that strong co-silencing of multiple genes will be required to disrupt auxin homeostasis in fruit. Localization of PIN and AUX/LAX proteins in the fruit and targeted alteration of their expression in transgenic lines will help uncover their roles in fruit development. The vegetative phenotypes of SlPIN4-RNAi lines suggested that SlPIN4 and/or SlPIN3 are involved in the determination of plant architecture. Analysis of auxin transport and distribution in these lines should allow for a better understanding of the role of SlPIN4 and SlPIN3 in the control of auxin homeostasis. These lines provide a description of genotypes other than Arabidopsis with defective expression of PIN genes that are not orthologous to AtPIN1, and contribute to our understanding of auxin transport in the regulation of traits such as compound leaves and sympodial growth, which are not present in Arabidopsis.

Experimental Procedures

Plant material

Tomato plants (Solanum lycopersicum cv. Ailsa Craig) were grown in a greenhouse (26°C/18°C day/night conditions, 16 h photoperiod, and supplemental 400 W sodium lights). To determine fruit age, flowers were tagged at anthesis and the number of days post anthesis (dpa) was recorded. The developmental stage of ripening fruit was determined by colour, as described in Lashbrook et al. (1994). The DR5::GUS transgenic line has been described previously (Dubrovsky et al., 2008).

For RNA extraction and IAA quantification, fruit harvested at 5 dpa were separated manually into pericarp and internal tissues and older fruit were divided into pericarp, placenta and seed/locular tissue. Flowers and ovaries were harvested at anthesis. Leaves and stems were harvested from 4-week-old plants. Roots were harvested from 1-week-old seedlings grown on 0.5 × MS plates (1% sucrose w/v, 1% agar w/v, pH 5.7). To test silencing, RNA was extracted from whole 7 dpa fruits. All tissue was flash-frozen in liquid nitrogen following separation. Each biological replicate consisted of pooled tissue from four or five fruit (for fruit samples), 24 seedlings (for root samples), four or five plants (for leaf and stem samples), or a minimum of 10 flowers (for flower and ovary samples). Data points represent the average ± standard error (SE) of a minimum of three biological replicates.

DNA and protein sequences

PIN and AUX/LAX genes were identified using the BLAST program against the SGN (http://solgenomics.net; Mueller et al., 2005) tomato unigene database (ftp://ftp.sgn.cornell.edu/transcript_sequences/combined_datasets/curr/all_sgn_transcript_sequences.seq) and against the provisional release version of the tomato genome (http://solgenomics.net/genomes/Solanum_lycopersicum/genome_data.pl) using Arabidopsis PIN and AUX/LAX protein sequences as queries. Gene models were predicted using AUGUSTUS (Stanke et al., 2004) and confirmed by amplification from cDNA, cloning into pGEMT-Easy (Promega, http://www.promega.com/) and DNA sequencing using the primer sequences indicated (Table S3). Neighbour-joining phylogenetic trees were created using MEGA 4.0 (Tamura et al., 2007) after sequence alignment using ClustalW2 (Larkin et al., 2007).

RNA extraction and qRT-PCR

RNA was extracted using TRIzol (Invitrogen, http://www.invitrogen.com/) and 1 μg used for cDNA synthesis with the Verso cDNA kit (Thermo Scientific, http://www.thermoscientific.com/). qRT-PCR was performed using an ABI PRISM 7900HT instrument (Applied Biosystems, http://www.appliedbiosystems.com/) using 5 μL of 10-fold diluted cDNA, 1 ×  SYBR green master mix (Applied Biosystems), and 1 μm each of two gene-specific primers (Table S4), in a final volume of 15 μL. The thermocycling regime consisted of 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 sec at 95°C, 30 sec at 54°C (or 58°C for SlPIN10), and 30 sec at 72°C. Disassociation curves and gel electrophoresis verified amplification of a single product. CT values were calculated using SDS2.1 software (Applied Biosystems) and data was analyzed using the 2−ΔΔCT method (Livak and Schmittgen, 2001) with UBI3 (X58253) as a reference gene for normalization (Rotenberg et al., 2006). Primer efficiency was tested by standard curve analysis using serial dilutions of a known amount of template and their specificity was confirmed by amplicon sequencing. Data were expressed relative to the first biological replicate in roots (Figures 5 and 6) or in wild-type fruit (Figures 8c and S3).

IAA quantification

IAA was quantified by gas chromatography-mass spectrometry (GC-MS) essentially as described by Barkawi et al. (2008). Briefly, 100 mg of frozen tissue were homogenized in 100 μL of extraction buffer (65% isopropanol, 35% 0.2 m imidazole, pH 7.0) that contained 5 ng of a [13C6] IAA standard. After incubation on ice for 1 h, samples were centrifuged at 14 000 g for 5 min, purified using solid phase extraction, and methylated with ethereal diazomethane. GC-MS was run in select ion monitoring mode and IAA quantified by isotope dilution analysis (Barkawi et al., 2008).

GUS staining

Fruit were cut longitudinally into 5 mm thick slices, placed into GUS staining buffer (Fernandez et al., 2009) and vacuum-infiltrated twice for 15 min. After incubation in the dark at 37°C for 16 h, tissues were fixed and cleared in a 3:1 (v/v) solution of ethanol:acetic acid.

NPA treatment

NPA (Sigma, http://www.sigmaaldrich.com/) was prepared as a 100 mm stock in dimethyl sulfoxide (DMSO) and added to the watering solution of 2-month-old plants at the concentration indicated. NPA-containing and NPA-free watering solutions were alternated every 2 days for 6 days. Control plants were watered with a solution that contained an equivalent concentration of DMSO (0.05%, v/v). Flowers were tagged in anthesis at the beginning of the treatment. For comparison of SlPIN4-RNAi and NPA-treated plants, 2-week-old wild-type seedlings were watered with 50 μm NPA for 2 days.

Confocal microscopy

Tissue was sectioned (100–250 μm) using a Lancer Series 1000 vibratome and confocal images obtained using a Leica TCS SP5 microscope (Leica-Microsystems, http://www.leica-microsystems.com/). RFP fluorescence was excited using the DPSS 561 nm laser and emission was detected between wavelengths 579 and 600 nm. Leica LAS-AF software was used for image processing.

Vector construction and plant transformation

To generate SlPIN4-RNAi transgenics, a 360-bp fragment of SlPIN4 (amino acids 534–654), was amplified using primers with the following sequences: forward 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTAACCATAGAGAAGTCAATCTCC-3′ and reverse 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTATTATAATCCAAGAATAAT-3′; and cloned into pDONR201 (Invitrogen). Inverted repeats were subcloned into pHELLSGATE8 (Helliwell and Waterhouse, 2003) that contained the CaMV 35S promoter. The DR5rev::mRFPer promoter–reporter fusion was amplified from pAM1006 (Gallavotti et al., 2008) using primers with the following sequences: forward 5′-AAGTCGACTCGACGGTATCGCGCCCA-3′ and reverse 5′-ATGGATCCGCATGCCTGC-3′; and cloned into pBI101 (Clontech, http://www.clontech.com/) using SalI and BamHI restriction sites. The SlPIN4-RNAi and DR5rev::mRFPer constructs were transformed into Agrobacterium tumefaciens (LBA4404) and tomato as described by Fillatti et al. (1987). SlPIN4-RNAi line (1) was crossed to the DR5::GUS reporter line, and F3 plants that were homozygous for both transgenes were used for analysis.

Phenotypic analysis

To evaluate fruit set, flowers were tagged in anthesis and self-pollination was encouraged by vibrating the inflorescence with an electric toothbrush. The numbers of flowers that had successfully set fruit were recorded 5 days later. Fruit set was calculated over a 4-week period with the following number of flowers tagged each week for wild-type and SlPIN4-RNAi lines (1) to (4) respectively: 38, 15, 34, 29, 15 (week 1); 33, 25, 28, 32, 24 (week 2); 20, 22, 22, 10, 10 (week 3); 27, 32, 31, 18, 15 (week 4). Values plotted are the average ± SE of the four 1-week values. Fruit growth curves were generated by measurement of fruit diameter at regular intervals.

To evaluate auxin-induced parthenocarpy, flowers were emasculated 3 days before anthesis and a 10-μL aliquot of a solution that contained 2 μg IAA (Sigma), 5% ethanol and 0.1% Tween 80 was applied to the ovary 1 day before anthesis and again at anthesis. Control ovaries were treated with IAA-free solution.

For quantification of vegetative phenotypes (Figures 8 and 9), plants of SlPIN4-RNAi line (1) were compared with wild type. Leaf-flattening index, stem angle, internode length and transverse sections were examined in 1-month-old plants. The leaf-flattening index was calculated as described by de Carbonnel et al. (2010). The terminal leaflet of the third leaf was photographed from the abaxial side, then flattened manually and photographed again. The ratio of the projection area before and after flattening was calculated using ImageJ software (National Institutes of Health, http://imagej.nih.gov). Stem bending was quantified using ImageJ to calculate the angle between the base of the stem and the shoot apical meristem in profile view photographs of plants. Hypocotyl and internode lengths were measured using a ruler. Transverse sections were taken from the middle of the first internode and imaged using an Olympus SZX12 microscope (http://www.olympusamerica.com/).

Accession numbers

The Genbank accessions for SlPIN and SlAUX/LAX genes are as follows: HQ127074 (SlPIN1), HQ127077 (SlPIN2), HQ127079 (SlPIN3), HQ127078 (SlPIN4), HQ127080 (SlPIN5), HQ127082 (SlPIN6), HQ127076 (SlPIN7), HQ127083 (SlPIN8), HQ127075 (SlPIN9), HQ127081 (SlPIN10), HQ671063 (SlLAX1), HQ671064 (SlLAX2), HQ671065 (SlLAX3), HQ671066 (SlLAX4) and HQ671067 (SlLAX5). The TAIR accessions for the Arabidopsis sequences are At1g73590 (AtPIN1) At5g57090 (AtPIN2), At1g70940 (AtPIN3), At2g01420 (AtPIN4), At5g16530 (AtPIN5), At1g77110 (AtPIN6), At1g23080 (AtPIN7), At5g15100 (AtPIN8), At2g38120 (AtAUX1), At5g01240 (AtLAX1), At2g21050 (AtLAX2), At1g77690 (AtLAX3).

Acknowledgements

We thank Dr. Jennifer Normanly for help with IAA quantification, Dr. Maria Ivanchenko for the DR5::GUS tomato line, Dr. Andrea Gallavotti for the DR5rev::mRFPer construct, and Dr. Jocelyn Rose for critical reading of this manuscript. This work was supported by the National Science Foundation Plant Genome Program (Award No. DBI-0922661 to C.C.)

Ancillary