The WOX13 homeobox gene promotes replum formation in the Arabidopsis thaliana fruit




The Arabidopsis fruit forms a seedpod that develops from the fertilized gynoecium. It is mainly comprised of an ovary in which three distinct tissues can be differentiated: the valves, the valve margins and the replum. Separation of cells at the valve margin allows for the valves to detach from the replum and thus dispersal of the seeds. Valves and valve margins are located in lateral positions whereas the replum is positioned medially and retains meristematic properties resembling the shoot apical meristem (SAM). Members of the WUSCHEL-related homeobox family have been involved in stem cell maintenance in the SAM, and within this family, we found that WOX13 is expressed mainly in meristematic tissues including the replum. We also show that wox13 loss-of-function mutations reduce replum size and enhance the phenotypes of mutants affected in the replum identity gene RPL. Conversely, misexpression of WOX13 produces, independently from BP and RPL, an oversized replum and valve defects that closely resemble those of mutants in JAG/FIL activity genes. Our results suggest that WOX13 promotes replum development by likely preventing the activity of the JAG/FIL genes in medial tissues. This regulation seems to play a role in establishing the gradient of JAG/FIL activity along the medio-lateral axis of the fruit critical for proper patterning. Our data have allowed us to incorporate the role of WOX13 into the regulatory network that orchestrates fruit patterning.


In flowering plants the success of reproduction relies largely on the correct morphogenesis of the gynoecium, which after fertilization forms the fruit that protects the seeds and mediates their dispersal. The Arabidopsis fruit is a silique composed mainly of the ovary whose medio-lateral pattern is formed by two lateral valves connected to the replum through the valve margins. This latter specialized tissue plays a major role in fruit opening, or dehiscence, as it facilitates the detachment of the valves from the replum through lignification and cell–cell separation mechanisms (reviewed in Robles and Pelaz, 2005; Balanza et al., 2006; Roeder and Yanofsky, 2006; Martínez-Laborda and Vera, 2009). Antagonistic activities of transcription factors give rise to this medio-lateral pattern of the Arabidopsis fruit. The FRUITFULL (FUL, Gu et al., 1998) MADS-box gene is responsible for the development of the valves (Liljegren et al., 2004), the SHATERPROOF (SHP) MADS-box genes are required for the formation of the valve margins (Liljegren et al., 2000), and REPLUMLESS homeobox gene (RPL, also known as PENNYWISE; BELLRINGER, and VAAMANA, Byrne et al., 2003; Bao et al., 2004; Bhatt et al., 2004; Smith and Hake, 2003; Zhu et al., 2007) and the class I KNOX gene BREVIPEDICELLUS (BP, Douglas et al., 2002; Venglat et al., 2002), also known as KNAT1, Lincoln et al., 1994; Chuck et al., 1996) are key players in promoting replum identity (Roeder et al., 2003; Alonso-Cantabrana et al., 2007; Ripoll et al., 2011). The combined activities of the SHATTERPROOF (SHP1, SHP2) MADS-box genes and the bHLH genes INDEHISCENT (IND, Liljegren et al., 2004) and ALCATRAZ (ALC, Rajani and Sundaresan, 2001), and recently SPATULA (SPT, Groszmann et al., 2008; Arnaud et al., 2010) have been proven to be essential for valve margin formation and thus, proper fruit dehiscence. Genetic and molecular studies have uncovered the existence of mutually opposing relationships between all these sets of transcriptional regulators. Indeed these antagonistic interactions have been determined to be critical for proper patterning of the fruit along the medial-lateral axis (Dinneny et al., 2004; Alonso-Cantabrana et al., 2007; Girin et al., 2009; Martínez-Laborda and Vera, 2009; Ripoll et al., 2011).

It is now well established that floral organs, and therefore carpels, evolved as modified leaves (Honma and Goto, 2001; Pelaz et al., 2001; Ditta et al., 2004). Interestingly, in recent years, several genes that were characterized previously for their roles in meristem and leaf development have been incorporated into the regulatory networks that control fruit patterning in Arabidopsis (reviewed in Girin et al., 2009; Martínez-Laborda and Vera, 2009). This situation is the case for the C2H2 Zn-Finger encoding gene JAGGED (JAG; Dinneny et al., 2004; Ohno et al., 2004), and for the closely related YABBY1 (YAB1) group genes FILAMENTOUS FLOWER (FIL, Chen et al., 1999; Sawa et al., 1999) and YABBY3 (YAB3, Siegfried et al., 1999). These genes were identified initially for their roles in repressing meristematic activities in lateral organ primordia and for controlling abaxial–adaxial polarity of the leaf (Chen et al., 1999; Sawa et al., 1999; Siegfried et al., 1999; Kumaran et al., 2002; Dinneny et al., 2004; Ohno et al., 2004; Ha et al., 2010). Most recently, it has been proposed that along the medio-lateral axis of the ovary, JAG, FIL and YAB3 act redundantly in a gradient of activity (known as JAG/FIL activity) to promote valve and valve margin formation (Dinneny et al., 2005; Alonso-Cantabrana et al., 2007). In this way, high levels of JAG/FIL activity induce FUL expression in lateral positions leading to valve development, whereas lower levels promote SHP expression leading to valve margin formation. The absence of JAG/FIL activity in medial tissues allows the formation of the replum (Dinneny et al., 2005; Alonso-Cantabrana et al., 2007).

In addition to its role in the replum, RPL has been also shown to regulate meristem maintenance of the SAM by interacting genetically and molecularly with members of the class I KNOX, including BP and SHOOTMERISTEMLESS (STM, Endrizzi et al., 1996; Long et al., 1996; Byrne et al., 2003; Smith and Hake, 2003; Bhatt et al., 2004; Smith et al., 2004; Cole et al., 2006; Kanrar et al., 2006; Scofield et al., 2007; Rutjens et al., 2009). However, members of the JAG/FIL activity act antagonistically to meristematic genes including BP (Kumaran et al., 2002; Ha et al., 2010). Similarly in the fruit, JAG/FIL genes negatively regulate the expression of RPL and BP (Dinneny et al., 2004; Alonso-Cantabrana et al., 2007).

Although in Arabidopsis members of the plant-specific WUS homeobox (WOX) family of transcription factors regulate different aspects of SAM development (Chandler et al., 2008) including stem cell maintenance and cell proliferation (Mayer et al., 1998; Sarkar et al., 2007; Wu et al., 2007), they are also involved in many other plant developmental events (Matsumoto and Okada, 2001; Park et al., 2005; Wu et al., 2005; Breuninger et al., 2008; Nardmann et al., 2009; Vandenbussche et al., 2009; Zhao et al., 2009; Ji et al., 2010; Skylar et al., 2010; Suer et al., 2011; Ueda et al., 2011; Nakata et al., 2012). In recent years orthologs of many Arabidopsis WOX genes have also been shown to regulate similar developmental pathways in other plant species. In rice, the OsWOX3 gene has been shown to negatively regulate OsYAB3. In turn, OsYAB3 regulates proper leaf development (Dai et al., 2007) and is related closely to the Arabidopsis FIL and YAB3 genes, members of the JAG/FIL function (Dinneny et al., 2004).

Therefore, in this context, we undertook a reverse genetic approach to identify WOX family members that regulate pattern formation along the medial–lateral axis of the Arabidopsis fruit. The results presented in this work show that the Arabidopsis WUSCHEL-LIKE HOMEOBOX13 (WOX13) gene regulates fruit patterning by negatively controlling the expression of the JAG/FIL activity genes in the medial domain, which later allows the correct formation of the replum and, thus, the correct specification of the medio-lateral pattern.


Identification and expression pattern of WOX13

Despite the fact that many WOX transcription factors regulate many aspects of plant development, no WOX gene has been shown previously to play a role in pattern formation during carpel/fruit development. We thus decided to undertake a reverse genetic approach based on in silico gene expression studies using Genevestigator (Hruz et al., 2008) and AtGenExpress (Schmid et al., 2005) and found that WOX13 was expressed at high levels in developing carpels (Figure S1a,b).

We then confirmed these observations by studying WOX13 expression by semi-quantitative reverse transcription polymerase chain reaction (RT-PCR). High levels of expression were found in inflorescences, floral buds before and after anthesis (closed and open flowers), and low levels in siliques and leaves (Figure 1a). Additionally, we performed in situ hybridization studies that showed WOX13 expression in inflorescence and flower meristems as well as in young floral buds (Figure S2a). The WOX13 expression pattern has been shown recently, based on a WOX13::GUS transcriptional reporter line (Deveaux et al., 2008). However, we generated transgenic lines harboring a GUS reporter translational construct we called WOX13::WOX13-GUS to determine more precisely the WOX13 expression pattern in gynoecia development. We found GUS activity in the root tip, emerging lateral roots and root vasculature (Figure S2b,c), in the SAM and leaf vasculature (Figure S2d) and in the gynoecia (Figure 1b), findings similar to those described by Deveaux and collaborators (Deveaux et al., 2008), No detailed studies of WOX13 expression in carpel tissues had been performed previously, therefore we used our translational reporter for that purpose. Interestingly, we observed a dynamic expression throughout carpel development (Figure 1b–i). In young gynoecia (from stages six to seven through stage 11) WOX13::WOX13-GUS was active in lateral and medial tissues (Figure 1d–g), and stronger activity was found at the apical part (Figure 1b). Approaching anthesis (stage 12), the WOX13 reporter expression gradually became restricted towards medial positions in which the replum tissue develops (Figure 1h,i). In mature fruits the expression was largely absent in valves, valve margins and repla (Figure 1c).

Figure 1.

Expression pattern of WOX13.

(a) Reverse transcription polymerase chain reaction (RT-PCR) analysis of WOX13 in different tissues. ACTIN2 was used as a control. (b–i) WOX13::WOX13-GUS expression. (b, c) Whole-mount staining for WOX13::WOX13-GUS reporter in a cleared inflorescence (b) and in a stage 17 fruit (d). (d–i) WOX13::WOX13-GUS expression in gynoecia cross-sections. b, buds; i, inflorescences; OF, stages 12 and 13 flowers; S, siliques; L, leaves.

Misexpression or loss-of-function of WOX13 alters vegetative growth and fruit development

Because WOX family members play important roles in meristem function (Mayer et al., 1998; Wu et al., 2005; Sarkar et al., 2007), and because WOX13 is expressed in the meristem-like replum region of developing carpels, we explored the possibility that WOX13 might play a role during replum formation.

We selected two independent insertional alleles named as wox13-1 and wox13-2 to better understand WOX13 function. The wox13-1 mutant contains a T-DNA insertion at the end of the homeodomain (Deveaux et al., 2008). In wox13-2, the transposon insertion is located in the first exon of the gene (Figure S1d), which may lead to the abolition of WOX13 functions. Using semi-quantitative RT-PCR, we confirmed that wox13-2 is a loss-of-function allele as no mRNA was detected (Figure S1d). Accordingly, whereas wox13-1 plants are largely normal, with only minor defects in flowering time (Figure S2e,f, not shown and Deveaux et al., 2008), the complete loss of WOX13 activity in wox13-2 also affected fruit development. A close inspection of wox13-2 siliques revealed that repla in these plants were narrower than those of wild-type fruits (Figure 2a,b). Whereas the replum epidermal layer of wild-type fruits contains on average seven cells (cells = 7.2 ± 0.25, = 7), we observed a 20% reduction in wox13-2 repla (cells = 5.7 ± 0.25, = 12). Further, the valve margin of wox13-2 fruits seemed slightly wider and contained more lignified cells than that of wild-type fruits (Figure 2c,d). In addition to the fruit defects, we observed a reduced number of lateral roots in wox13 mutants (Figure S2h). Overall these results suggest that WOX13 participates in different plant developmental processes, including fruit patterning. Interestingly, protein alignments showed that WOX13 is the only member of the Arabidopsis WOX family that lacks the WUS-box (Haecker et al., 2004) downstream of the homeodomain, in contrast to its closest paralogs (Figure S1c). This unique feature of the WOX13 protein might explain the absence of apparent phenotypes in wox13-1 plants.

Figure 2.

Effect of wox13 and 35S::WOX13 on fruit development. (a, c) Ler, (b, d) wox13-2, (e, g, i, k) Col-0, (f, h) 35S::WOX13 and (j, l) 35S::WOX13/+ stage 17 fruits. (a, b, i, j) Toluidine blue and (c, d, e–h) Safranin (o) and Alcian blue staining. Separation layer stains light blue while lignification layer is red. (k, l) Lignification pattern of fruit cross-sections stained with phloroglucinol. r, replum; SL, separation layer; LL, lignification layer; enb, enb layer; EVL, ectopic valve lignification. Scale bars in (a–e and i–l) are 100 μm and in (f–h) are 50 μm.

To gain further insights into the functions that WOX13 has during fruit development, we generated 35S::WOX13 transgenic plants. A dose effect was observed for 35S::WOX13 transgene, as the phenotypes were more pronounced in homozygous (35S::WOX13 hereafter) than in hemizygous lines (35S::WOX13/+ hereafter), a finding that correlated tightly with WOX13 expression levels (Figure S3).

In 35S:WOX13 plants, all four floral whorls were affected (Figure S2j,k). Nevertheless, the most dramatic phenotypes were identified in carpels. In the wild type the apical closure of the gynoecium occurs around stage nine, however in 35S::WOX13 the style remained open at later stages in the majority of the fruits examined (not shown). Although in these carpels fertilization occurred at low rates, manually pollination resulted in mature fruits. Plastic thin cross-sections from mature 35S::WOX13 fruits showed a conspicuously enlarged replum (Figure 2e,f). In addition, no clear lignification was observed either in the valve margins or in the enb layer (compare Figure 2f,h with Figure 2e,g), an observation that might correlate with the indehiscence of these fruits. Traces of ectopic lignification were detected in the valves of these fruits (Figure 2h).

In 35S::WOX13/+, the floral alterations occurred mostly in the gynoecium, which also showed fertility problems but to a lesser extent than in the homozygotes. In line with our observation for 35S::WOX13 plants, carpels had a very large stigma (not shown), and cross-sections revealed that repla of these fruits were clearly enlarged (Figure 2j,l), double the size of the wild-type replum in the extreme cases (Figure 2i,k). The cells of the valve margins were disorganized, so that the separation and lignified layers were not well defined (Figure 2j,l). Moreover, while in wild-type fruits the lignified layer is connected to the enb layer (Rajani and Sundaresan, 2001), in 35S::WOX13/+ we observed a discontinuity between these tissues (Figure 2k,l).

Besides the reproductive phenotypes described, 35S::WOX13 plants were late flowering, and displayed a disorganized phyllotactic pattern (not shown). We also observed that leaf morphogenesis was disturbed in this background. In both rosette and cauline leaves the blade expansion was generally impaired and prominent notches were present in the margins (Figure S2i).

WOX13 is involved in fruit dehiscence

The fact that valve margin lignification increased in wox13 mutants and decreased in 35S::WOX13 gain-of-function fruits suggests that WOX13 might under normal circumstances act to control to some extent fruit dehiscence. To further test this idea, we quantified the total number of open siliques on the main stem of wox13-2 plants compared with its respective wild-type control (Ler). The measurements were made using 60-, 67-, 74- and 81-day-old plants after germination respectively. In 60-day-old wild-type plants, we found that approximately 13% of siliques had undergone dehiscence. In contrast, nearly twice as many of wox13 mutant siliques (24%) had dehisced at this same time point (Figure 3a). At later time points we found an increase in dehiscence in wox13 mutants, a result that suggested that WOX13 might inhibit the dehiscence process. To further confirm this hypothesis, we measured dehiscence in the overexpressor 35S::WOX13 plants (35S::WOX13/+ versus Col-0). As expected, the opposite effect was observed for 35S::WOX13/+ plants (Figure 3b). Whereas at day 60, 25% of the wild-type siliques had already open and released their seeds, none of the siliques from 35S::WOX13/+ plants had dehisced. Even at later time points, we consistently observed a decrease in dehiscence in 35S::WOX13/+ when compared with the wild type (Figure 3b). Taken together, these data suggest that WOX13 under normal circumstances functions to inhibit dehiscence, perhaps by controlling valve margin lignification. Consistent with this idea, all of the 35S::WOX13 fruits analyzed during this study were indehiscent.

Figure 3.

WOX13 affects the dehiscence process and alters valve/valve margin gene expression. (a, b) Fruit dehiscence in plants 60, 67, 74 and 81 DAG (n = 7 plants per genotype). Vertical bars correspond to standard deviation (SD). (c–g) SHP2::GUS expression in wild-type (c, f), 35S::WOX13/+ (d) and 35S::WOX13 (e, g) fruits. (h–k) KNAT6::GUS expression in wild-type (h, j) and 35S::WOX13 fruits (i, k). (l–o) FUL::GUS expression in wild-type (l), 35S::WOX13/+ (m) and 35S::WOX13 gynoecia (n, o). Red arrowheads point to ectopic GUS in valves and white arrowheads delimit replum. r, replum; v, valve.

Overexpression of WOX13 disturbs the expression pattern of valve margin genes

Our loss- and gain-of-function studies suggest that WOX13 controls fruit dehiscence and lignification, perhaps by regulating the expression of valve margin identity genes. To explore this possibility, we analyzed the expression patterns of GUS reporter lines for valve margin genes in 35S::WOX13 backgrounds.

Previous studies have shown that the combined activities of FUL in the valves and RPL in the replum restrict the expression of valve margin genes to the valve margin (Ferrandiz et al., 2000; Roeder et al., 2003; Girin et al., 2010). Thus, in ful or rpl mutants, valve margin genes become expressed ectopically in valves or replum respectively (Ferrandiz et al., 2000; Roeder et al., 2003; Liljegren et al., 2004). SHP2::GUS is usually expressed in the valve margins of wild-type siliques (Liljegren et al., 2004). In contrast, we found that the expression of SHP2::GUS was dramatically reduced in 35S::WOX13/+ (Figure 3d) and absent in the 35S::WOX13 valve margins (Figure 3e,g). Because SHP activity is required for valve margin development (Liljegren et al., 2004), the reduced expression of SHP in 35S::WOX13 plants may explain both the decrease in lignifications (Figure 2f,l) as well as the decrease in fruit dehiscence (Figure 3b). Interestingly, ectopic SHP2::GUS signal was detected in the valves of 35S::WOX13 fruits (Figure 3e,g), a situation that is the likely cause of the ectopic lignifications seen in 35S::WOX13 (Figure 2h), similar to that described in ful fruits (Gu et al., 1998; Ferrandiz et al., 2000).

The valve margin markers, KNAT6::GUS (Ragni et al., 2008 and this work) and GT140 (GUS reporter for IND gene, Ferrandiz et al., 2000; Liljegren et al., 2004) were also tested in 35S::WOX13 plants. As observed for SHP2::GUS, there was a reduction in expression of both markers in the valve margin along with ectopic expression in the valves (Figures 3h–k and S5). This finding suggested that WOX13 might regulate these valve margin genes in the same way as it does for SHP. Despite the slight increase in valve margin lignification in wox13 fruits, no obvious differences in the SHP2::GUS expression were observed (not shown). However, when the expression level of SHP1 and SHP2 was measured by RT-qPCR in wox13 fruits, we observed an increase in SHP gene expression and conversely, a reduction in 35S::WOX13 fruits (Figure S4a,b). Taken together, these data suggest that WOX13 negatively regulates valve margin development by, directly or indirectly, repressing valve margin genes, a finding that might explain the ectopic lignification seen in 35S::WOX13 and similar to that described in ful fruits (Gu et al., 1998; Ferrandiz et al., 2000).

Overexpression of WOX13 impairs FUL expression in the valves

35S::WOX13, ful mutant fruits display ectopic SHP2::GUS expression in valves (Ferrandiz et al., 2000). We found that GUS expression, driven by the enhancer trap line of FUL (ful-1/+) (Gu et al., 1998) was reduced in 35S::WOX13/+ fruits and was lowered even more dramatically in the 35S::WOX13 (Figure 3l–o). In the most severe phenotypes, FUL expression in valves was totally absent and present only in the vasculature (Figure 3n) or detected only in small patches (Figure 3o). In wox13 fruits, there were no apparent changes in reporter activity (not shown), although RT-qPCR indicated a slight increase of FUL expression in wox13 fruits and a clear reduction in 35S::WOX13/+ and 35S::WOX13 fruits (Figure S4c). These data suggest that WOX13 might also negatively regulate FUL expression in the valves.

WOX13 negatively regulates the JAG/FIL activity in fruits

We found several phenotypic similarities between 35S::WOX13 plants and those plants with reduced JAG/FIL activity, such as jagged rosette leaves, enlarged replum tissue, slight fertility defects or partial indehiscence (Figure S6; Dinneny et al., 2004, 2005). On the other hand, the altered expression patterns observed for both SHP2 and FUL in 35S::WOX13 backgrounds were also reminiscent of what has been described for mutants with impaired JAG/FIL activity (Dinneny et al., 2005). Thus, a possible scenario might be that WOX13 negatively regulates the JAG/FIL function. To test this hypothesis, we first compared the behaviour of the FUL reporter (ful-1/+) between 35S::WOX13 fruits and mutant combinations affected in JAG/FIL genes. As mentioned earlier, FUL expression decreased gradually as WOX13 activity increased (Figure 3m–o). A similar trend was observed in fil yab3 jag fruits compared with that in fil yab3 or fil single mutants (Figure 4a–c; Dinneny et al., 2005). In addition, we also detected a slight variability in the FUL reporter expression pattern in fruits from different 35S::WOX13 plants (Figure 3n,o), as it has been also found previously for fil yab3 fruits (Dinneny et al., 2005). The SHP2::GUS expression was similarly affected by 35S::WOX13 or by reduced JAG/FIL activity, with a lowered expression within the normal valve margin domain and ectopic expression in the valve (Figures 3d–g and 4d,e).

Figure 4.

WOX13 regulates the JAG/FIL gene expression. (a–c) FUL::GUS expression in fil-8 (2xCol) (a) and fil-8 yab3-2 (2xCol) (b, c) gynoecia. (d, e) SHP2::GUS expression in fil-8 (2xCol) (d) and fil-8 yab3-2 (2xCol) gynoecia (e). (f–i) JAG::GUS expression in wild type (f), 35S::WOX13/+ (g), 35S::WOX13 (h) and wox13-2 (i) on stage 12 carpels. (j–m) FIL expression by in situ hybridization on wild type (j), 35S::WOX13/+ (k), 35S::WOX13 (l) and wox13-2 (m) stage 8 gynoecia. (n–r) Genetic interaction between 35S::WOX13 and fil observed by bright field (n–p) and scanning electronic microscopy (q, r). fil-8 (2xCol) (n), fil yab3 (2xCol) (o) and 35S::WOX13 fil-8 (2xCol) flowers (p). (q) is a close-up of (p), showing ectopic valve margins (evm). (s–v) Genetic interaction between wox13-2 and jag-1 (3xLer) and fil-8. Cross-sections of stage 17 fruits of jag-1 (s), jag-1 wox13-2 (t), fil-8 (u) and fil-8 wox13 (v). Red arrowheads in (b, e) point to GUS expression in valves. Arrowheads in (o, p) point to ectopic valve margins while in (j–m) and (s–v) delimit repla. r, replum; pe, petal; st, stamen; se, sepal; v, valve; sty, style. Scale bars are 0.5 mm in (a–e and p), 50 μm in (f–m), 1 mm in (n–o), 100 μm in (q), 500 μm in (r) and 50 μm in (s–v).

To further validate our hypothesis, we examined JAG expression in 35S::WOX13 and wox13-2 carpels using the reporter line JAG::GUS, which mimics endogenous JAG expression (J. Dinneny, Carnegie Institution for Science, Stanford, CA, USA, personal communication). In wild-type gynoecia, JAG::GUS is expressed in lateral tissues (Figure 4f), whereas in the 35S::WOX13/+, the expression levels of the reporter decreased and become localized even more laterally, a factor that could account for the enlargement of the replum (Figure 4g). The effect in 35S::WOX13 plants was more dramatic as, in most fruits, JAG expression was mostly absent in the valves (Figure 4h). In contrast, JAG::GUS activity in wox13 mutants extended further towards medial positions, which could explain the slight reduction of replum size (Figure 4i).

We next tested FIL expression by in situ hybridization. As published previously, in wild-type gynoecia FIL was detected in lateral positions (Figure 4j), whereas in wox13 carpels FIL expression was extended slightly to more medial positions in which the replum forms (Figure 4m). In 35S::WOX13/+ fruits, FIL expression retracted to more lateral domains, and the overall levels were reduced significantly (Figure 4k), a finding similar to that observed for JAG. This reduction was even more pronounced in the 35S::WOX13 carpels (Figure 4l). These analyses indicate that WOX13, in fact, represses (directly or indirectly) the expression of FIL and JAG genes likely in a levels-dependent manner. We wanted to test this interaction further and introduced 35S::WOX13 into fil mutants and wox13-2 into jag and fil single mutants. The constitutive expression of WOX13 in fil mutants resulted in a stronger radialization of the sepals, petals and stamens (Figure 4p,r) as found in the fil yab3 double mutant (Figure 4o) when compared with the single fil mutant (Figure 4n) (Siegfried et al., 1999). Interestingly, in 35S::WOX13 fil gynoecia, we observed ectopic valve margins in the valves, irregular and smaller valve cells and a more twisted replum (Figure 4q). These defects were similar to the ones described for the jag fil yab3+/− combination (Dinneny et al., 2005). In addition, and supporting that WOX13 represses JAG/FIL genes, we found that when JAG or FIL were mutated in wox13-2 fruits, as in double wox13 jag or wox13 fil mutants, the wox13 reduced repla recovered a size similar to that of the wild-type fruit (Figure 4t,u). This finding indicated that at least part of the wox13 replum phenotype is due to JAG and FIL expression. Taken together, these data suggest that WOX13 downregulates JAG/FIL expression.

WOX13 controls RPL and BP expression in the replum

Previous studies have shown that medial (replum) and lateral factors interact antagonistically during fruit patterning (Dinneny et al., 2005; Alonso-Cantabrana et al., 2007; Girin et al., 2009;, Martínez-Laborda and Vera, 2009; Ripoll et al., 2011). We have shown that the replum is reduced in wox13 loss-of-function mutants and that fruits of 35S::WOX13 gain-of-function plants have oversized repla (Figure 2). We therefore used transgenic reporter constructs (BP::GUS, (Ori et al., 2000); RPL::GUS, (Roeder et al., 2003)) to examine the expression of the replum identity genes BP and RPL to determine if their misregulation correlated with these altered fruit phenotypes.

Whereas the BP::GUS reporter was almost undetectable in the replum of wild-type fruits (Figure S7a,d), this same reporter could be detected readily in 35S::WOX13 backgrounds (Figure S7b,c,e). More dramatic effects were found when the RPL::GUS reporter was tested in wox13 and 35S::WOX13 lines. While in the wild-type plants RPL::GUS was detected in the replum (Figure 5a,b; Roeder et al., 2003), in wox13 fruits GUS activity was reduced from early stages of gynoecium development (Figure 5c). A similar reduction was observed later on in wox13 fruits (Figure 3a). Most interestingly, in the 35S::WOX13 background, RPL::GUS was expressed ectopically throughout the valves (Figure 5e), a finding that suggested that WOX13 might cause RPL misexpression, directly or indirectly perhaps via JAG/FIL. Therefore, the increased replum size of 35S::WOX13 fruits is mostly likely to be because of a corresponding increase in replum factor expression, RPL and BP.

Figure 5.

WOX13 promotes replum formation. (a) RPL::RPL-GUS expression in wild type (left) and wox13-2 (right) flowers. (b–e) RPL::RPL-GUS expression in carpel cross-sections of wild type (b) and wox13-2 (c) at stage eight and wild type (d) and 35S::WOX13 at stage 10 (e). (f, g) FIL in situ hybridization in rpl (f) and 35S::WOX13 rpl (g) gynoecia at stage 7. (hj) Cross-sections of stage 17 fruits of wox13-2 (3xCol) (h) and rpl-2 wox13-2 (3xCol) (i, j). (k–n) Cross-sections of bp-9 (k), rpl-2 (l), bp-9 rpl-2 (m) and 35S::WOX13 bp-9 rpl-2 (n) fruits. Arrowheads in (f, g) point to the replum whereas in (h) delimitate the replum, asterisks mark the absent repla. Scale bars in (b–e) are 50 μm, in (f–j) are 100 μm and in (k–n) are 50 μm.

In rpl mutants, the transformation of replum into valve margin tissue (Roeder et al., 2003) is due partly to misexpression of FIL and, to a lesser extent, to JAG in the replum (Dinneny et al., 2005). Consequently, the loss of replum development in rpl single mutants can be rescued in fil rpl and jag rpl double mutants (Dinneny et al., 2005). We have shown that WOX13 seems to negatively regulate JAG/FIL expression in fruits (Figure 4). Therefore, we tested whether the overexpression of WOX13 was also able to eliminate FIL expression in the medial tissues of rpl gynoecia. Whereas in the rpl carpels FIL expression extended into medial domains from which replum would develop (Figure 5f), strikingly, in 35S::WOX13 rpl carpels, FIL was absent from this region (Figure 5g). These data indicate that in the absence of RPL, WOX13 seems to be capable of repressing JAG/FIL functions in medial tissues and, thus, allowing replum development. In agreement with these observations is the fact that wox13 mutations enhance the replum defects of rpl fruit. When we combined rpl-2, which has a mild replum phenotype (Roeder et al., 2003), with wox13-2 (3xCol) (Figure 5h) the replum of the resulting rpl-2 wox13-2 double mutants was virtually absent (Figure 5i,j), a situation similar to that described previously for bp rpl fruit (Alonso-Cantabrana et al., 2007; Figures 5m and S7h). Other rpl-related defects were further enhanced in rpl-1 wox13-2 and resembled those of bp rpl plants (Figure S7k,l; Douglas et al., 2002; Venglat et al., 2002). The similarities between bp rpl and wox13 rpl plants point to WOX13 as a positive regulator of replum formation. To further support this role for WOX13 we decided to combine 35S::WOX13 with bp and rpl mutants. Remarkably, whereas in bp rpl fruits no replum tissue is observed (Alonso-Cantabrana et al., 2007; Ripoll et al., 2011; Figures 5m and S7h), replum formation is restored in 35S::WOX13 bp rpl pistils (Figures 5n and S7i). Taken together, our results suggest that WOX13 promotes replum development independently of RPL and BP.


Here we have demonstrated that WOX13 plays an important role in controlling the medio-lateral patterning of the fruit, which is fundamental for proper maturation and seed dispersal. Thus, both loss- and gain-of-function WOX13 plants result in abnormal fruits with a variation in timing of fruit opening.

WOX13 positively regulates replum formation

In Arabidopsis, the establishment of the medio-lateral fruit pattern is controlled by factors that also regulate proper shoot development and leaf formation (Carles and Fletcher, 2003; Dinneny et al., 2005; Alonso-Cantabrana et al., 2007; Ragni et al., 2008; Girin et al., 2009; Martínez-Laborda and Vera, 2009). WOX13 is expressed in meristematic tissues, such as the SAM, and in the replum, a quasi-meristematic tissue, a distribution that suggests a role in both tissues. However, although misregulation of WOX13 in 35S::WOX13 plants delayed flowering time, the altered replum was the tissue most obviously affected by WOX13 gain- or loss-of-function. wox13 mutants generated fruits with a reduced replum size, whereas the 35S::WOX13 fruits had an enlarged replum. These phenotypes suggested that WOX13 might promote replum formation perhaps by promoting cell proliferation, similar to other members of the WOX family (van der Graaff et al., 2009). In agreement, WOX13 has been shown previously to be involved in the control of cell division (Deveaux et al., 2008). The mild phenotype of the wox13 loss-of-function mutant might indicate some redundancy at the molecular level. WOX10 and WOX14 are the closest paralogs to WOX13, however WOX10 has been described as a pseudogene (Haecker et al., 2004; Deveaux et al., 2008), and the expression patterns of WOX14 and WOX13 do not overlap in carpels (Deveaux et al., 2008). Therefore, other distant genes, within or outside the WOX family, may have a role together with WOX13 in replum formation.

WOX13 promotes replum formation by down-regulating JAG/FIL activity

The phenotype of individuals that overexpress WOX13 closely resembled plants in which JAG/FIL activity is impaired (Figures 2e–l, 3b, 4n–r and S6; Dinneny et al., 2004, 2005). Further, in both conditions, SHP2 and FUL expression patterns seem to behave similarly (Figures 3c–e, l–o and 4a–e). On the other hand, we have shown that 35S::WOX13/+ plants resembled fil or jag single mutants, while 35S::WOX13 plants looked more like either the fil yab3 and/or the fil yab3 jag mutants (Figure 4n–r), a finding that might reflect some sort of dosage-dependent regulation. In addition, 35S::WOX13 fil gynoecia were similar to that of fil jag yab3+/−, and the wox13 fil and wox13 jag repla were similar to that of the wild type. In agreement, when compared with wild-type fruits, lower levels of FIL and JAG expression were observed in hemizygous and even lower (near to zero) in homozygous 35S::WOX13 fruits. It has been shown that JAG/FIL activity is repressed in the replum by RPL; thus in rpl mutants FIL becomes expressed ectopically in this tissue (Dinneny et al., 2005; Figure 5f). However, we strikingly observed that FIL is still reduced in the absence of RPL as observed in 35S::WOX13 rpl repla. These data indicate that WOX13 is likely to promote replum development by, directly or indirectly, down-regulating the JAG/FIL genes in a RPL-independent manner.

WOX13 promotes replum formation in the absence of RPL and BP

The 35S::WOX13 plants develop an enlarged replum, likely due to the down-regulation of JAG/FIL activity and, as a consequence, the replum factors were up-regulated, a situation that leads to the overgrowth of this tissue as shown previously (Alonso-Cantabrana et al., 2007; Ripoll et al., 2011). For example, 35S::WOX13 led to ectopic misexpression of RPL (Figure 5e), which we interpreted as the result of down-regulating JAG/FIL gene expression that, in turn, caused a reduction in the activity of valve and valve margin genes. However, BP, another replum identity gene, was not expressed ectopically in 35S::WOX13 valves, likely to be because of the presence of the direct BP repressors AS1 and AS2 (Ori et al., 2000; Alonso-Cantabrana et al., 2007; Guo et al., 2008).

In bp or rpl mutants, replum forms because, most likely, closely related proteins of the KNOX/BELL group such STM or POUND-FOOLISH (PNF) respectively (Endrizzi et al., 1996; Smith et al., 2004) could interact and/or substitute for either BP or RPL (Bellaoui et al., 2001; Smith and Hake, 2003; Bhatt et al., 2004; Hackbusch et al., 2005; Cole et al., 2006; Kanrar et al., 2006). However, in bp rpl plants replum growth is totally aborted (Alonso-Cantabrana et al., 2007; Ripoll et al., 2011). Nevertheless, we observed replum development in 35S::WOX13 bp rpl fruits (Figure 5n), a finding that indicated that other proteins non-identified as replum factors, perhaps within the KNOX/BELL group, might play also a role in promoting replum formation. In the SAM, BP is partially redundant with STM (founder member of the KNOX family) in regulating stem cell function, and as both genes are expressed in the replum, STM becomes a good candidate to promote replum formation (Long et al., 1996; Byrne et al., 2002). It is known that KNOX-BELL proteins form heterodimers to regulate meristem functions (Bellaoui et al., 2001; Smith and Hake, 2003; Hackbusch et al., 2005; Kanrar et al., 2006), we can then speculate that STM could, perhaps together with PNF, promote some replum formation in the absence of RPL and BP and when JAG/FIL activity is reduced as in the 35S::WOX13 plants.

WOX13 promotes replum formation by likely establishing the JAG/FIL gradient

The current model for fruit patterning in Arabidopsis proposes that the medio-lateral tissues are generated by the antagonistic action of valve, valve margin and replum factors (Dinneny et al., 2005; Alonso-Cantabrana et al., 2007; Girin et al., 2010; Ripoll et al., 2011). High levels of JAG/FIL function allow FUL expression in the valves while intermediate levels switch on the valve margin genes, and no activity is found in the replum (Dinneny et al., 2005). Then, FUL is able to impair the activity of the replum and the valve margin identity genes in the valves (Ferrándiz et al., 2000; Ripoll et al., 2011). In the replum, in turn, the replum identity genes negatively regulate valve margin and valve factors; and the valve margin identity genes prevent the ectopic expression of the replum and valve factors in the valve margin (Roeder et al., 2003; Girin et al., 2010; Ripoll et al., 2011).

The fact that mutations in WOX13 enhanced the rpl-2 phenotype suggests that a higher de-regulation of JAG/FIL in the presumptive replum of wox13 rpl might lead to a major repression of the replum genes. Alonso-Cantabrana et al. (2007) proposed that, as occurs during vegetative growth (Kumaran et al., 2002), the YABBY genes negatively regulate the expression of the class I KNOX genes, such as BP, in the fruit. Our data lend strong support for this regulation (Figure 5i,j,m).

Our data also led us to propose that JAG/FIL gradient might be reached through the action of WOX13. WOX13 is expressed early in the whole carpel, sequentially being constrained to more medial domains. Thus, throughout gynoecium development, WOX13 is expressed at high levels in the replum and valve margin regions whereas it is transient in lateral positions. The JAG/FIL gradient would then be established between these two extremes (Figure 6). If so, the JAG/FIL levels would be reduced in 35S::WOX13 plants, in a manner similar to that in mutant combinations affected in FIL, YAB3 and JAG genes. Conversely, in the wox13 mutant, the JAG/FIL genes would reach the intermediate levels more medially and the valve margins would develop further towards the medial position, resulting in a reduced replum.

Figure 6.

WOX13 function during medio-lateral patterning of the fruit. WOX13 downregulates the expression of the JAG/FIL activity genes and therefore it might be establishing the gradient of this function along the medio-lateral axis of the fruit. In lateral domains, JAG/FIL activity is high, whereas WOX13 expression is transient in this domain. High and intermediate levels of JAG/FIL function are responsible respectively for the activation of the valve (FUL, green) and valve margin identity genes (SHP, pink). In medial tissues, levels of WOX13 expression are elevated and very low or no expression levels for JAG/FIL genes are found. This regulation allows the replum identity genes to promote replum formation. VM, valve margin; ll, lignified layer; sl, separation layer.

In conclusion, WOX13 seems to promote replum formation, independently of BP and RPL, by likely repressing the valve JAG/FIL genes in a gradient manner. Here we demonstrate that WOX13 plays an important role in control of the medio-lateral pattern of the fruit, an event that is fundamental in the dispersion of the seeds, and that both mutation and overexpression of WOX13 result in variation of timing of fruit opening.

During the past few years, orthologs for some of the regulatory genes controlling fruit development in Arabidopsis have been identified in other plant species and, interestingly, they seem to share common functions (Vrebalov et al., 2009; Chung et al., 2010; Girin et al., 2010; Karlova et al., 2011). It will therefore be interesting to determine if the WOX13 gene is conserved in other plants species and, if so, if it serves a similar function.

Experimental procedures

Growth conditions, plant materials and genetics

Plants were grown in soil at 22°C under long day conditions (16 h light/8 h dark). The wild type was Columbia (Col-0) or Landsberg erecta (Ler).

The wox13-2 allele in Ler accession was obtained from the Cold Spring Harbor Laboratory collection (CSHL, ET14042) in which GUS expression was lost. HD1 and HD4, and ET3.4 and HD4 primers (Table S1) were used for genotyping wild-type and wox13-2 respectively. For phenotypic studies, wox13-2 was backcrossed three times to Col accession.

Other mutants and transgenic lines were ful-1 (Gu et al., 1998), SHP2::GUS (Savidge et al., 1995) crossed to Col, JAG::GUS generated by J.R. Dinneny, BP::GUS (Ori et al., 2000) crossed to Col background, rpl-2, rpl-1 and RPL::GUS (Roeder et al., 2003), bp-9 (Smith and Hake, 2003), jag-1 (3XLer) (Dinneny et al., 2004), fil-8 (Ler and 2xCol) (Kumaran et al., 2002), GT140 (Liljegren et al., 2000), bp-9 rpl-2 (Smith and Hake, 2003), fil-8 yab3-2 (2xCol) (Kumaran et al., 2002) and wox13-1 (Deveaux et al., 2008).

Construction of transgenic lines

The 35S::WOX13 construct was generated after cloning full-length WOX13 cDNA into the destination vector pALLIGATOR2 (Bensmihen et al., 2004). To generate the WOX13::WOX13-GUS construct we used pHDi and HD4 primers (Table S1). The resulting fragment of 2.4 kb contains the WOX13 promoter, the first exon and the first intron of the gene, and was translationally fused to the β-glucuronidase gene in the pBI101.1 (Clontech, For the KNAT6::GUS reporter line we used oJJR322 and oJJR323 primers (Table S1). The resulting PCR product was cloned into pBI101.1 vector (Clontech). Plants were transformed using a floral dip procedure (Clough and Bent, 1998), and several independent transgenic lines were selected for follow-up experiments.

Scanning electron microscopy

Samples were fixed overnight in FAA (50% ethanol, 5% acetic acid, 3.7% formaldehyde), dehydrated through an ethanol series and critical-point dried. Samples were sputter-coated with gold and palladium and viewed using a Hitachi S-2300 electron microscope (Hitachi High Technologies Europe,


Total RNA was extracted from siliques at stages 14–17 of fruit development. For reverse transcription (RT), 2 μg of RNA was used, and quantitative (q)RT-PCR was performed as described previously (Castillejo and Pelaz, 2008). ACTIN2 was used as control in RT-PCR assays whereas UBIQ10 was used as an internal control in qRT-PCR experiments. At least two biological replicates were performed per assay. See Table S1 for primers used in the PCR reactions.

In situ hybridization

In situ hybridization was performed as described previously (Ferrándiz et al., 2000). The FIL probe was generated from pY1-Y plasmid (Dinneny et al., 2004). For the WOX13 RNA probe we used WOX13_5′ and WOX13T7 primers (Table S1).

Histological analyses and GUS assays

For phloroglucinol lignin staining and for Saffranin O and Alcian Blue staining, stage 17 fruits were fixed, embedded in paraplast and sectioned (8 μm) as described previously (Liljegren et al., 2000; Roeder et al., 2003; Ripoll et al., 2011). For thin sections, (2 μm) stage 17 fruits were fixed in 1.5% glutaraldehyde, 0.3% paraformaldehyde, 0.025 m PIPES and 0.1% Tween 20 with vacuum infiltration for 30 min and left overnight at 4°C. The tissue was then washed 3 times in 0.025 m PIPES and dehydrated through an ethanol series. Samples were included in Technovit 7100 historesin following the manufacturer's instructions. Sections were mounted on slides and stained with 0.1% toluidine blue in phosphate buffer (pH 5.5). Histological sections were analyzed using either a Nikon E-600 (Nikon Instruments, or a Zeiss Axiophot, (Carl Zeiss Microscopy, microscope.

GUS staining was carried out as described elsewhere (Blazquez et al., 1997; Ripoll et al., 2011). For the BP::GUS related assays, staining was done for 4 h. Whole-mount pictures were taken under DIC optics on a Zeiss Axiophot scope. Sections (8 μm) were photographed under dark-field illumination (Nikon E-600 microscope). All GUS analyses, except for FUL::GUS, were performed in homozygous lines.


We thank José R. Dinneny for the donation of the JAG::GUS line and F. Parcy for the pALLI2 vector. Jesús Vicente, Antonio Martínez-Laborda and NASC for seeds. Antonio Martínez-Laborda and Theodoros Zografou for critical reading of the manuscript. This work was supported by the Spanish Ministry of Science grants BFU2006-00771 and BFU2009-08325 to S.P., and by a National Science Foundation to M.F.Y(grant IOS-1121055). J.J.R. was recipient of a fellowship from the Generalitat Valenciana (BPOSTDOC06/060). S.P.'s group was recognized as consolidated by the Catalonia Government (2009 SGR 697) and the CRAG is supported by the CONSOLIDER Program (CSD2007-00036).