In the Brassicaceae, indehiscent fruits evolved from dehiscent fruits several times independently. Here we use closely related wild species of the genus Lepidium as a model system to analyse the underlying developmental genetic mechanisms in a candidate gene approach. ALCATRAZ (ALC), INDEHISCENT (IND), SHATTERPROOF1 (SHP1) and SHATTERPROOF2 (SHP2) are known fruit developmental genes of Arabidopsis thaliana that are expressed in the fruit valve margin governing dehiscence zone formation. Comparative expression analysis by quantitative RT-PCR, Northern blot and in situ hybridization show that their orthologues from Lepidium campestre (dehiscent fruits) are similarly expressed at valve margins. In sharp contrast, expression of the respective orthologues is abolished in the corresponding tissue of indehiscent Lepidium appelianum fruits, indicating that changes in the genetic pathway identified in A. thaliana caused the transition from dehiscent to indehiscent fruits in the investigated species. As parallel mutations in different genes are quite unlikely, we conclude that the changes in gene expression patterns are probably caused by changes in upstream regulators of ALC, IND and SHP1/2, possible candidates from A. thaliana being FRUITFULL (FUL), REPLUMLESS (RPL) and APETALA2 (AP2). However, neither expression analyses nor functional tests in transgenic plants provided any evidence that the FUL or RPL orthologues of Lepidium were involved in evolution of fruit indehiscence in Lepidium. In contrast, stronger expression of AP2 in indehiscent compared to dehiscent fruits identifies AP2 as a candidate gene that deserves further investigation.
In the Brassicaceae, a family of angiosperms with 338 genera and 3700 species (Warwick et al., 2010) that includes the major flowering plant model system Arabidopsis thaliana, molecular analyses have demonstrated that many morphological characteristics on which traditional systematic relationships are based are homoplasious rather than homologous. Fruit structures in particular have proven to be highly labile during evolution, and all molecular phylogenetic data consistently indicate that many species with similar fruits may be only distantly related, whereas species with dramatically different fruits may be very closely related (Mummenhoff et al., 2005, 2009; Franzke et al., 2011). This implies that developmental processes controlling fruit shape and structure are extremely plastic in evolution. The typical Brassicaceae fruit is dehiscent, and is considered to represent the ancestral fruit type in the family (Hall et al., 2002). Nevertheless, species with various indehiscent fruits are found in 20 tribes distributed over the whole Brassicaceae phylogeny (Figure 1) (Appel and Al–Shehbaz, 2003), indicating that this character evolved many times independently. Here we investigate the evolutionary shift from dehiscent to indehiscent fruits by comparing two closely related Brassicaceae species, i.e. Lepidium campestre (L.) W.T. Aiton and Lepidium appelianum Al–Shehbaz.
With about 250 species, the genus Lepidium is one of the major genera of the Brassicaceae (Al–Shehbaz and Mummenhoff, 2011). Typically, Lepidium species produce two-seeded dehiscent fruits, but the genus also comprises species with indehiscent fruits, such as L. appelianum, previously considered as a member of the genus Cardaria (Cardaria pubescens) (Mummenhoff et al., 2001, 2009; Al–Shehbaz et al., 2002). The two chosen species L. campestre (dehiscent fruits) and L. appelianum (indehiscent fruits) (Figure 1) are especially suited as models for our purpose because their close relatedness ensures that the shift of fruit type only happened quite recently during evolution. Additionally, both species are diploid (2n = 2x = 16), which simplifies genetic analyses, and their close relationship to Arabidopsis facilitates the adaptation of molecular techniques.
As has been demonstrated before (Mummenhoff et al., 2009), the anatomy of L. campestre wild-type fruits resembles A. thaliana wild-type fruits, in that they form a well-defined dehiscence zone (DZ) at the valve margin (Figure 2), resulting in a similar fruit opening mechanism. In contrast, L. appelianum fruits fail to open, only releasing the seeds during decomposition of the fruit valves. Anatomical studies show that they do not form a DZ but are surrounded by a continuous ring of lignified cells (Figure 2) comparable to indehiscent fruits of certain A. thaliana mutants, such as 35S::FUL, ful or ind ful (Ferrandiz et al., 2000b; Liljegren et al., 2004). However, because the fruits of all these mutants lack a DZ, it is not possible to unequivocally relate the fruits of L. appelianum to one of them in particular.
Arabidopsis thaliana exhibits the typical Brassicaceae fruit, termed a silique, in which the process of fruit opening is quite well understood. At a morphological level, accurate patterning of relevant tissues within the fruit is crucial to allow fruit dehiscence (Figure 2). The dominant structures of the fruit are the two fruit valves that enclose the developing seeds. They are connected to the replum by the DZ (or valve margin), which consists of a stripe of lignified cells, the lignified layer, and an adjacent region of small thin-walled cells, the separation layer (Spence et al., 1996; Rajani and Sundaresan, 2001). The cells of the lignified layer are connected to the lignified endocarp layer b located on the inside of the valves. During fruit ripening, the whole fruit dries and shrinks, and only the lignified structures stay rigid, thereby creating a spring-like tension within the fruit. At the same time, the middle lamellae of the separation layer cells degenerate, acting as a pre-determined breaking zone at which the pressure tears the valves apart from the replum, resulting in fruit opening (Meakin and Roberts, 1990, 1991; Spence et al., 1996).
At the molecular level, fruit dehiscence is controlled by several genes encoding transcription factors that are required for proper establishment of the DZ (Figure 2a). Two redundant MADS box genes, SHATTERPROOF1 (SHP1) and SHATTERPROOF2 (SHP2) are expressed in the DZ, where they activate the basic helix-loop-helix protein-encoding genes INDEHISCENT (IND) and ALCATRAZ (ALC). Additionally, SHP1 and SHP2 contribute to valve margin development autonomously (Liljegren et al., 2000, 2004). IND contributes to formation of both the lignified layer and the separation layer, especially mediating lignification of the valve margin cells, whereas ALC is essential for separation layer formation (Rajani and Sundaresan, 2001; Sorefan et al., 2009). It is crucial for correct fruit patterning that the expression of these four ‘valve margin identity genes’ is restricted to the DZ, a process that is mediated by negative regulators. One of these proteins is encoded by the MADS box gene FRUITFULL (FUL), which is expressed in the fruit valves and mainly contributes to valve cell differentiation and expansion (Gu et al., 1998; Ferrandiz et al., 2000b). Another such protein is encoded by the BEL1-like homeobox gene REPLUMLESS (RPL), which is expressed in the replum, but is not necessary for replum development per se (Gu et al., 1998; Ferrandiz et al., 2000b; Rajani and Sundaresan, 2001; Roeder et al., 2003). Note that the RPL gene is also known as PENNYWISE (Smith and Hake, 2003), BELLRINGER (Byrne et al., 2003), VAAMANA (Bhatt et al., 2004), LARSON (Bao et al., 2004) and BLH9 (Cole et al., 2006). Recently, the floral homeotic gene APETALA2 (AP2) (Bowman et al., 1989) has been identified as a negative regulator of RPL and the SHP genes (Ripoll et al., 2011).
In A. thaliana, mutants for all seven genes described above have been characterized that show impaired dehiscence. Single mutants of ind or alc and shp1/2 double mutants are all defective in DZ formation, leading to inability of the valves to detach from the replum (Liljegren et al., 2000, 2004; Rajani and Sundaresan, 2001; Wu et al., 2006). In rpl mutants, ectopic SHP1/2, IND and ALC expression induces the conversion of replum cells into valve margin-like cells, leading to partial indehiscence (Roeder et al., 2003; Sorefan et al., 2009). Moreover, both the ful knockout mutation as well as 35S::FUL over-expression induce indehiscence in A. thaliana. In ful plants, valve cells are transformed into valve margin cells due to ectopic expression of valve margin identity genes (SHP1/2, IND and ALC), but in addition the overall architecture of the fruit is highly disturbed because the fruit valves fail to elongate (Gu et al., 1998; Ferrandiz et al., 2000b). On the other hand, fruits of 35S::FUL mutants look normal except for the lack of a DZ caused by the suppression of valve margin identity genes by ectopic FUL expression (Ferrandiz et al., 2000b; Ostergaard et al., 2006). In ap2 knockout plants, increased expression of RPL, SHP2 and IND causes a delay in fruit opening due to expansion of the replum and the lignified layer (Ripoll et al., 2011). Thus, it is tempting to speculate that such simple changes in gene expression are involved in independent evolution of indehiscent from dehiscent fruits within the Brassicaceae.
In the current study, we compare the expression patterns of L. appelianum and L. campestre orthologues of the A. thaliana genes (henceforce referred to as AtALC, AtFUL, AtIND, AtRPL, AtSHP1, AtSHP2, and AtAP2 for clear discrimination from the Lepidium orthologues). Whereas the expression patterns in dehiscent L. campestre closely resemble those found in A. thaliana, we observed an absence of expression of valve margin identity genes (LaALC, LaIND and LaSHP1/2) in the valve/replum border in indehiscent L. appelianum fruits. We conclude that, at least in some wild Lepidium species, the switch from dehiscent to indehiscent fruits is indeed accompanied by changes within the regulatory pathway predicted from the A. thaliana mutant system, and that the causal genetic change has probably affected an upstream regulator of LaSHP1 and LaSHP2. Therefore, our study exemplarily clarifies the molecular genetic basis of fruit dehiscence between two closely related wild species.
Sequence analysis of fruit developmental genes
One goal of our work was to uncover differences in DNA sequence and expression patterns of putative fruit developmental genes to identify candidate genes that may have contributed to the shift from dehiscent to indehiscent fruits in Lepidium. Based on the close relationship between Arabidopsis and Lepidium, we designed primers derived from A. thaliana fruit developmental gene sequences to isolate the respective orthologues from L. campestre and L. appelianum. We isolated cDNAs of ALC, FUL, IND, RPL, SHP1 and SHP2 from both Lepidium species, and initial blast (National Center for Biotechnology Information, Bethesda, MD, USA) searches revealed high degrees of DNA sequence similarity to the respective fruit developmental genes from A. thaliana. Phylogenetic analyses of the respective gene families were performed, indicating that all genes under study are putative orthologues of the corresponding A. thaliana fruit developmental genes (Figure S1). To exclude the existence of additional closely related paralogues, we performed Southern blot hybridization on genomic DNA from L. campestre and L. appelianum using various restriction enzymes. For LcSHP1 and LaSHP2, we found two bands, most likely due to cross-hybridization (as revealed by the sizes of the obtained bands). In all other cases, only one band per lane was observed, suggesting that all genes tested are single-copy genes (Figure S2).
The amino acid sequences of all orthologues were infered from cDNA sequences to facilitate the search for structural differences between genes from indehiscent L. appelianum and dehiscent Brassicaceae species, including L. campestre. The amino acid sequence identity between the putative orthologues of A. thaliana and Lepidium ranges from 73.5% (AtALC versus LaALC) to 95.5% (AtFUL versus LaFUL). Amino acid sequence identities between orthologous Lepidium proteins are always higher than 93%. Amino acid alignments (Figure S3), also including sequences of other dehiscent Brassicaceae species, were performed in order to search for amino acid substitutions that appear only in proteins of the indehiscent L. appelianum. These are considered as candidate changes that may have contributed to the shift from dehiscent to indehiscent fruits during the evolution towards extant L. appelianum. We detected a total of 20 amino acid positions (marked by arrows in Figure S3) at which the sequence of L. appelianum differs from all other Brassicaceae species under study.
Expression of valve margin identity genes is absent from the valve/replum border of L. appelianum fruits
In order to compare the expression patterns of putative fruit valve margin identity genes (i.e. ALC, IND and SHP1/2) between L. appelianum and L. campestre, their overall expression level was analysed in six tissues (root, stem, leaf, flower, fruit stage 15/16, and fruit stage 17) by quantitative RT-PCR and Northern blot hybridization (Figure 3a–d and Figure S4). Furthermore, in situ hybridization was performed on sections of flowers and fruits (stages 10–15, according to Smyth et al., 1990) for a more detailed insight into the spatial expression patterns (Figure 4a–h). It was found that, in both species, all four valve margin identity genes are mainly expressed in flowers and fruits and only at lower levels, if at all, in roots, stems and leaves. The only exceptions are LaALC in roots and LaSHP2 in stems, which show expression levels comparable to those in flowers and fruits. For ALC, the overall expression level in fruits of L. appelianum was significantly higher than in fruits of L. campestre (Figure 3d), accompanied by a clear change in spatial distribution. In L. campestre, ALC is strongly expressed in the DZ (Figure 4e), but was exclusively detected in the ovule in L. appelianum (Figure 4a). IND is expressed at the valve/replum border in L. campestre fruits, but is not detectable in fruits of L. appelianum (Figure 4b,f), consistent with a 2.5–7-fold decrease in the overall expression level in L. appelianum flowers and fruits compared to L. campestre (Figure 3a). SHP1 and SHP2 are both expressed in the DZ and ovules in L. campestre (Figure 4g,h) but only in ovules in L. appelianum (Figure 4c,d). Regarding the overall expression level, LcSHP1 is significantly more highly expressed than LaSHP1 in flowers and fruits (stage 15/16), while LcSHP2 is significantly more highly expressed than LaSHP2 in flowers whereas expression is significantly lower in fruits (stage 17) (Figure 3b,c). In summary, we show that, in L. campestre dehiscent fruits, all four valve margin identity genes are expressed in the DZ, but that such expression patterns at the valve/replum border are absent from L. appelianum indehiscent fruits.
The spatio-temporal expression patterns of FUL and RPL in the two Lepidium species are similar
As AtFUL and AtRPL are known to be upstream regulators of valve margin identity genes in A. thaliana (Ferrandiz et al., 2000b; Roeder et al., 2003; Liljegren et al., 2004), their expression patterns were also analysed and compared between the two Lepidium species. FUL was found to be expressed in all analysed tissues except for roots, and the overall expression level was significantly higher in L. campestre stems and fruits of stage 15/16 compared to L. appelianum (Figure 3e,f). In both species, in situ hybridization shows mRNA accumulation in the fruit valves, but not in the replum (Figure 4i–n). Therefore, the expression patterns look superficially similar between L. appelianum and L. campestre, but small differences at the valve/replum border cannot be excluded, because especially in this region fruit anatomy differs significantly between the two species. RPL is expressed in all tissues analysed, with significantly higher expression levels for LcRPL in leaves and flowers and for LaRPL in fruits of stage 17 (Figure 3f) when comparing the two Lepidium orthologues. Unfortunately, spatial expression patterns for RPL were only detected via in situ hybridization at the tip of the inflorescence meristem, and in all whorls of the developing flower (Figure 4k,n). Detection by in situ hybridization was not possible in mature flowers or fruits (Figure 4j,m) although the quantitative RT-PCR results clearly show such expression.
LaAP2 is up-regulated in indehiscent L. appelianum
During preparation of this paper, the floral homeotic gene AtAP2 was identified as a negative regulator of genes controlling valve margin identity in A. thaliana (Ripoll et al., 2011). As an initial step to investigate this new candidate gene, we isolated orthologous partial cDNAs from both Lepidium species (LaAP2 and LcAP2) and analysed their expression by quantitative RT-PCR. We found a significant increase in LaAP2 expression compared to LcAP2 expression in all tissues except roots (Figure 3g).
Ectopic expression of FUL and RPL orthologues from both Lepidium species does not reveal functional changes in coding regions
As no major difference in expression pattern were detected in Lepidium between the two orthologues of FUL and, likewise, between the two orthologues of RPL, we wished to analyse whether the amino acid differences found during our sequence analysis cause a change in protein function. Therefore, cDNAs of the coding regions were cloned into a binary plasmid under the control of a CaMV 35S promoter, and transformed into A. thaliana (RPL orthologues) or L. campestre (FUL orthologues). For the RPL orthologues, the most obvious effect for all constructs was formation of multiple cauline leaves at the base of second-order inflorescence shoots (Figure 5a–d). Furthermore, in some transformants of each construct, an alteration in phyllotaxy and irregular elongation of internodes was observed (Figure S5). These phenotypic changes are consistent with those previously reported for AtRPL over-expressing mutants (Cole et al., 2006), except that no early flowering was detected in our case. Ectopic expression of the Lepidium FUL orthologues frequently caused the formation of fruits that were more difficult or impossible to open manually (Figure 5e,f). When the half-life of dehiscence of such transformants was estimated using a random impact test, a significant increase was detected for both constructs expressing FUL orthologues compared to an empty vector control (Figure 5g). Therefore, over-expression of FUL cDNAs of both Lepidium species induces fruit indehiscence, as is also the case for AtFUL (Ferrandiz et al., 2000b).
Loss-of-function mutations in the coding region of classical fruit developmental genes are probably not responsible for indehiscence in L. appelianum
We isolated orthologues of six fruit developmental genes (AtALC, AtFUL, AtIND, AtRPL, AtSHP1 and AtSHP2), which are known to be essential for pod shatter in A. thaliana, from L. campestre and L. appelianum. If mutations in these genes were involved in the evolution of indehiscence in L. appelianum, these may have led to changes in either the function of the encoded proteins or in gene expression patterns. However, the fact that, in addition to their roles in fruit dehiscence, all candidate proteins are also involved in other developmental functions in A. thaliana (Ferrandiz et al., 2000a; Byrne et al., 2003; Favaro et al., 2003; Pagnussat et al., 2005; Cai and Lashbrook, 2008) makes it unlikely, however, that dramatic changes in protein function (including total loss-of-function) contribute to evolution under conditions of natural selection due to pleiotropic effects. We found that amino acid sequences of all orthologous proteins are highly conserved between the Lepidium species and other members of the Brassicaceae (Figure S3). There is no evidence for dramatic changes in conceptual protein structure exclusively within L. appelianum that may explain the origin of fruit indehiscence. However, there are a number of small differences in amino acid sequences between analysed proteins of different Brassicaceae species, including a total of 20 amino acid changes that were exclusively detected in proteins of L. appelianum (Figure S3). We cannot completely rule out the possibility that any of these mutations is responsible for the switch from dehiscence to indehiscence in Lepidium, as it is well known that single amino acid substitutions may alter protein function dramatically. An example that is of interest in the broader context of our study is a single amino acid substitution in a predicted Myb3 transcription factor termed SH4 that caused reduced seed shattering during rice domestication (Li et al., 2006). Nevertheless, a change in the expression pattern of a gene appears a more likely scenario. Due to the modularity of many promoter and enhancer regions, alterations in gene expression in one location without affecting the function in others are quite frequent processes during evolution, e.g. during sub-functionalization events (Force et al., 1999).
Absence of valve margin identity gene expression may cause indehiscence in L. appelianum
By analysing the expression patterns of the four valve margin identity gene orthologues from L. appelianum and L. campestre, we found that dehiscent fruits of L. campestre exhibit patterns very similar to those found in dehiscent A. thaliana fruits. In both fruit types, the valve margin identity genes are expressed within a thin stripe of cells at the valve/replum border, and, in addition, ALC and SHP1/2 mRNA accumulation is found within ovules (Ma et al., 1991; Savidge et al., 1995; Flanagan et al., 1996; Rajani and Sundaresan, 2001; Liljegren et al., 2004). This conservation in spatial expression patterns suggests that gene functions may also be conserved between these two species. In A. thaliana, all four genes are important for establishing DZ tissue, which is an additional indication, together with the anatomical similarities, that fruit opening in L. campestre operates according to the same mechanisms as in A. thaliana (Liljegren et al., 2000, 2004; Rajani and Sundaresan, 2001; Mummenhoff et al., 2009). The function of AtALC in A. thaliana ovules is not yet known, but the AtSHP genes contribute redundantly to the floral organ identity ‘D function’ in A. thaliana together with SEEDSTICK, AGAMOUS and BELL1 (Western and Haughn, 1999; Pinyopich et al., 2003).
In indehiscent fruits of L. appelianum, LaALC, LaIND, LaSHP1 and LaSHP2 mRNAs were undetectable in the valve/replum border. It is already known that the lack of expression of either AtALC or AtIND or both AtSHP1 and AtSHP2 in A. thaliana knockout plants induces indehiscence (Liljegren et al., 2000, 2004; Rajani and Sundaresan, 2001). These findings strongly suggest that the combined absence of the four valve margin identity genes is sufficient to cause the indehiscent fruit phenotype in L. appelianum. On the other hand, expression of LaALC and LaSHP1/2 is retained in L. appelianum ovules, suggesting that it is not a complete gene knockout that causes the down-regulation at the valve/replum border, but that more sophisticated genetic changes are involved. This ovule-specific expression also indicates that the function in ovule development of these genes is not only conserved between A. thaliana and L. campestre but also in L. appelianum, although the significant increase in overall expression level for LaALC in L. appelianum fruits and for LaSHP2 in L. appelianum flowers may indicate functional changes requiring an increased transcript abundance.
Genetic cause of indehiscence in L. appelianum: evaluating various candidates
After having established that combined absence of expression of four valve margin identity genes from the valve/replum border is probably involved in the development of indehiscence in L. appelianum, the question arises as to which genetic changes cause this simultaneous down-regulation. From studies of A. thaliana mutants, several mutations are known to induce formation of indehiscent fruits, but only a few of these fit the expression pattern found in L. appelianum fruits.
Valve margin identity genes
The first scenario involves mutations directly within the valve margin identity genes themselves. This requires at least two independent mutations within the loci of the LaSHP genes resulting in exclusion from the valve/replum border and subsequent non-activation of the downstream targets LaIND and LaALC, assuming that the regulatory network known from A. thaliana (Liljegren et al., 2004) is conserved in Lepidium. We consider this possibility unlikely because of the requirement for two independent mutations that do not involve simple gene knockouts but specific removal of expression from a certain region (the DZ) while remaining present at another region (the ovules). However, we cannot exclude at present that changes in the regulation of SHP gene expression may have occurred during evolution of Lepidium.
FUL and RPL
A scenario that appears more likely to us involves mutations in one of the upstream regulators LaFUL or LaRPL. These genes are known to restrict valve margin identity gene activity to a thin stripe at the valve/replum border (Ferrandiz et al., 2000b; Liljegren et al., 2000; Roeder et al., 2003; Dinneny and Yanofsky, 2005). Down-regulation of either of these genes induces indehiscent fruits in A. thaliana mutants, caused by expansion of valve margin identity gene expression into the valve or replum, respectively (Gu et al., 1998; Ferrandiz et al., 2000b; Roeder et al., 2003; Liljegren et al., 2004), thereby not matching the expression patterns found in L. appelianum fruits. On the other hand, ectopic expression of FUL throughout the whole fruit in 35S::FUL transgenic A. thaliana or Brassica juncea plants causes indehiscent fruit phenotypes by eliminating expression of valve margin identity genes from the valve/replum border while leaving SHP1 and SHP2 expression in ovules unaffected (Ferrandiz et al., 2000b; Ostergaard et al., 2006), a phenotype that is strikingly similar to that found in L. appelianum. Theoretically, for plants ectopically expressing RPL, a similar indehiscent fruit phenotype is expected, because as a down-regulator of valve margin identity genes ectopic expression should equally lead to elimination of DZ formation. However, transgenic A. thaliana plants in which AtRPL expression is controlled by a CaMV 35S promoter dd not show altered fruit development (Harley Smith, University of California, Riverside, CA, USA, personal communication). Additionally, over-expression of BREVIPEDICELLUS, an activator of AtRPL, leads to ectopic expression of AtRPL in valves and valve margins, resulting in plants with enlarged repla and a reduction in valve width but no changes in valve margin and lignification patterns (Alonso-Cantabrana et al., 2007). This may indicate that ectopic expression of AtRPL alone is not sufficient to eliminate the expression of valve margin identity genes from the valve/replum border, at least in A. thaliana.
In this study, the expression patterns of FUL and RPL were analysed in L. appelianum and L. campestre but did not provide a clear indication of their involvement in the evolution of fruit indehiscence. FUL expression was located in fruit valves in both Lepidium species and did not show any major expansion to other regions of the L. appelianum fruit, so a global up-regulation as in the 35S::FUL A. thaliana plants may be excluded. Nevertheless, expansion of the LaFUL expression domain by only a few cell layers towards the replum may already be sufficient to suppress valve margin identity gene expression and thereby to prevent DZ formation and induce indehiscence. Such a small expansion may not have been recognized in our data, because the fruit anatomy of the Lepidium species differs, especially at the valve/replum border, and therefore the FUL expression domains cannot be directly compared. One might even argue that such a subtle change in FUL expression is more likely to occur under natural evolutionary conditions compared to the global expansion seen in the 35S::FUL transgenic line, because it may avoid pleiotropic effects due to interference with other FUL functions. In addition to its function in fruit development, AtFUL is also known to act as a flowering time and meristem identity gene and to contribute to cauline leaf morphology (Gu et al., 1998; Ferrandiz et al., 2000a). 35S::AtFUL A. thaliana plants show a dramatically reduced time to flowering and form terminal flowers with increased seed weight in addition to their indehiscence phenotype (Ferrandiz et al., 2000a,b). For RPL expression, dramatic differences in transcript level may be excluded on the basis of our quantitative RT-PCR results, but spatial patterns could not be compared because RPL could not be detected via in situ hybridization. This is consistent with findings in A. thaliana, where AtRPL detection via in situ hybridization was also very difficult in floral stages older than stage 8, and therefore GUS reporter systems had to be used to investigate AtRPL expression (Roeder et al., 2003). Future experiments may include RPL expression studies in Lepidium with GUS reporter constructs as performed in A. thaliana, once an efficient transformation system for both species has been established.
Recently, the RPL gene has attracted quite some attention (Gasser and Simon, 2011; Wagner and Mitchell-Olds, 2011), because a nucleotide substitution at the same position within a conserved regulatory element (Shattering element-like, Shl) that causes loss of seed shattering in certain rice cultivars (Konishi et al., 2006) was also found to have an effect on replum form in various Brassicaceae species (Arnaud et al., 2011). Even though no evidence has been published so far showing that this substitution really affects fruit dehiscence in Brassicaceae (Arnaud et al., 2011; Wagner and Mitchell-Olds, 2011), we investigated this regulatory element from L. appelianum. We found that the relevant nucleotide in L. appelianum is the same as in dehiscent L. campestre and A. thaliana, and thereby we can exclude its involvement in the evolution of indehiscence in our case (Figure S6).
During the analysis of conceptual amino acid sequences, one and seven amino acid changes were found exclusively in the L. appelianum LaFUL and LaRPL orthologues, respectively (Figure S3). As no alteration in expression pattern was detected for these genes, we analysed the possible effects of these amino acid changes on protein function by assessing their over-expression phenotypes. For both genes, these were found to be very similar for sequences of L. campestre and L. appelianum, and additionally correspond to over-expression phenotypes reported for the A. thaliana orthologues (Figure 5) (Ferrandiz et al., 2000b; Cole et al., 2006). Such similar phenotypic effects suggest a global conservation of protein function between both Lepidium species and A. thaliana, thereby indicating that a change in protein function is most probably not responsible for the evolution of indehiscence of L. appelianum fruits. Nevertheless, this possibility cannot be ruled out completely because (i) protein function was evaluated in a heterologous system and may differ in its natural environment, and (ii) RPL over-expression only causes changes in plant architecture but not in fruits. Therefore, any of the amino acid changes may compromise protein function exclusively in fruits.
AP2 as a novel candidate gene
During preparation of this paper, AtAP2 was identified as an additional member of the fruit patterning pathway in A. thaliana (Ripoll et al., 2011). As a negative regulator of AtSHP1, AtSHP2 and AtRPL, it normally prevents over-expression of these genes in the replum and valve margin. Thus, if the regulatory network as proposed for A. thaliana by (Ripoll et al., 2011) also holds in Lepidium, a simple increase in the LaAP2 expression level may cause the observed elimination of expression of L. appelianum valve margin identity genes from the valve/replum border. We found a significant increase in LaAP2 expression levels compared to LcAP2 in all tissues except for roots (Figure 3g). Nevertheless, in A. thaliana, AtAP2 is known to be regulated by miRNA172 at the translational level (Chen, 2004), implying that an increased mRNA level does not necessarily lead to an increased protein level. Due to the activity of miRNA172, 35S::AtAP2 mutant plants have normal flowers or show only mild floral defects (Chen, 2004), but whether they exhibit changes in fruit dehiscence has, to the best of our knowledge, not been reported. Therefore, a thorough investigation of the effect of AP2 over-expression on fruit opening in A. thaliana as well as in Lepidium, including a study of the L. appelianum miRNA172 pathway, is necessary in order to address whether the increased levels of LaAP2 expression are responsible for the indehiscent fruit phenotype.
Further candidate genes
Other candidates that may cause the indehiscent fruit phenotype in L. appelianum (based on the fact that mutant alleles can induce indehiscent fruits in A. thaliana) are FILAMENTOUS FLOWER (FIL), JAGGED (JAG) and YABBY3 (YAB3). These genes act redundantly in promoting valve and valve margin formation by activating AtFUL and AtSHP1/2 (Sawa et al., 1999; Siegfried et al., 1999; Dinneny et al., 2004, 2005; Ohno et al., 2004). Because of their redundant function, each of the three A. thaliana single knockout mutants shows normal fruit opening, but fil yab3 and fil jag yab3 mutants show major defects in fruit dehiscence (Dinneny et al., 2005). In these mutants, indehiscence is caused by a loss of AtSHP expression (and, as a consequence, probably also ALC and IND expression) in the valve margins, and an additional loss of AtFUL expression in valves (Dinneny et al., 2005), which only partially resembles the situation found in L. appelianum fruits. Additionally, in fil jag yab3 mutants, sepals, petals and stamens are replaced by filamentous structures with few floral characteristics (Dinneny et al., 2005). Consequently, as described for LaSHP1 and LaSHP2, complete loss-of-function mutations within these genes most likely do not cause indehiscence in L. appelianum, but slight shifts of expression domains may be involved. A retreat of LaFIL and LaYAB3 expression from the valve margin may abolish LaSHP expression while leaving LaFUL expression unchanged.
The absence of valve margin identity gene expression from the valve/replum border revealed by our data provides insights into the molecular cause of indehiscence in L. appelianum. However, it provides little direct indication as to which genetic change led to the evolution of indehiscent fruits in this species. Several candidate genes were identified based on the fruit patterning pathway of A. thaliana, but the expression data do not favour one of these candidates in particular. Future studies should focus on isolating the genomic loci of all potential candidates and analysing their influence on fruit development by means of transformation into A. thaliana and L. campestre.
Plants were grown in the greenhouse on loam/sand/pumice/compost (1:2:2:5) or seedling substrate (Kammlott, Kammlott GmbH, Erfurt, Germany)/sand/vermiculite (1–3 mm) (8:1:1), supplemented with 1 g L−1 each of Osmocote mini (www.scotts.com,The Scotts Miracle-Gro Company, Marysville, OH, USA) and Triabon (http://www.compo-expert.com, COMPO Expert GmbH, MüCnster, Germany) under a photoperiod of 16 h at 20°C and 8 h without illumination at 15°C. Flowering was induced by at least 6 weeks (L. campestre) or 13 weeks (L. appelianum) at 4°C with illumination for 8 h day−1. For morphological analysis, thin sections were prepared from paraplast-embedded fruits (http://www.leica-microsystems.com/, Leica Microsystems GmbH, Wetzlar, Germany) and stained with safranin-astra blue (http://www.sigmaaldrich.com, Sigma Aldrich Corporation, St. Louis, MO, USA).
Cloning of candidate genes
RNA was isolated using RNAiso-G+ (segenetic, (www.segenetic.de, segenetic, Borken, Germany)). cDNA was prepared using RevertAid™ Premium reverse transcriptase (Fermentas, http://www.thermoscientificbio.com/fermentas/, ThermoFisher Scientific Corporation, Waltham, MA, USA). Primer details are given in Table S2. Isolation of candidate genes was performed by 3′ and 5′ RACE (Frohman et al., 1988) or by specific PCR with primers binding to conserved regions of the A. thaliana sequences using PhusionTM high-fidelity DNA polymerase (Finnzymes, www.thermoscientificbio.com/finnzymes/, ThermoFisher Scientific Corporation, Waltham, MA, USA) and Taq DNA polymerase (Segenetic). For 5′ RACE, SP1 primers were used for gene-specific cDNA synthesis, followed by polyadenylation using terminal deoxynucleotidyl transferase (Fermentas) or terminal transferase (Roche, http://www.roche-applied-science.com Roche Applied Science, Penzberg, Germany). For nested PCR, oligo(dT) and SP2 primers (first step) and anchor primer with SP3 primer (second step) were used. Finally, full-length clones were PCR-amplified and ligated into pGEM-T vectors (Promega, www.promgega.com, Promega, Fitchburg, WI, USA).
In situ hybridization
In situ hybridization was performed essentially as described by Zachgo (2002) using flower and fruit tissues at developmental stages 10–15 (according to Smyth et al., 1990). Tissues were fixed in paraformaldehyde and embedded in paraplast. Probe templates were produced by linearizing the vector pGEM-T containing full-length clones of respective cDNAs in the correct orientation using XbaI. Antisense RNA probes were synthesized using T7 or SP6 RNA polymerases. Images were created using a stereomicroscope (Leica, DM5000B, http://www.leica-microsystems.com/, Leica Microsystems GmbH, Wetzlar, Germany) with an integrated digital camera (Leica, DCF 490); scale bars were generated using Leica Application Suite.
Total RNA was extracted using Total RNA Isolation Reagent (Biomol, http://www.biomol.de/, Biomol GmbH, Hamburg, Germany), QIAzol lysis reagent (Qiagen, www.qiagen.com, Qiagen N.V., Hilden, Germany) or the RNeasy plant mini kit (Qiagen). RNA concentration and integrity were analysed before and after DNase I digestion using a NanoVue spectrophotometer (GE Healthcare, www.gelifesciences.com, GE Healthcare, Chalfont St Giles, UK) and by gel electrophoresis. Extracts containing up to 20 μg total RNA were digested using recombinant DNase I (Roche) and subsequently extracted with phenol/chloroform. Absence of genomic DNA was tested by PCR using primers (Table S3) designed to amplify the SHP2 gene. cDNA synthesis was performed on 500 ng of DNase I-digested RNA using Transcriptor reverse transcriptase (Roche) and oligo(dT)20 primers. Quantitative RT-PCR reactions were performed in triplicate in an Mx 3005P cycler (Stratagene, www.stratagene.com, Agilent Technologies, Santa Clara, CA, USA) using Maxima™ SYBR Green/Rox qPCR Master Mix (2×) (Fermentas) with 1 μl cDNA (1:5 diluted) as template and 0.3 μm of forward and reverse primers (Table S3). The following thermal profile was used: 95°C for 10 min, 40 cycles of 95°C for 15 sec, 62–64°C for 30 sec and 72°C for 30 sec. Raw data were analysed using LinRegPCR (Ramakers et al., 2003; Ruijter et al., 2009) to obtain sample CT values and PCR efficiencies (E) for each primer pair. CT values for triplicate reactions were averaged, and relative quantities of expression for each gene were calculated as , where Cal is the sample with the lowest CT value, i.e. the highest expression level and SOI is the sample of interest [Correction added on 8 February 2013 after original online publication on 18 January 2013: (1+E) (Ctcal – Ctsoi) was changed to ]. For normalization, relative quantities of expression were divided by a normalization factor, the geometric mean of the relative quantities of expression of the normalization genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ras-related gtp-binding nuclear protein 2 (RAN2), polyubiquitin 10 (UBQ10) and arabidopsis thaliana tip41-like protein (TIP41)-like. Inter-run calibration was performed by dividing normalized expression values by the geometric mean of the normalized expression values of three inter-run calibrator samples present on all plates to be compared.
Ectopic expression of FUL and RPL orthologues
cDNA of FUL and RPL orthologues was PCR-amplified from plasmid clones using primers listed in Table S2, and placed under the control of the CaMV 35S promoter in pFGC5941 (www.chromdb.org) using restriction enzymes SmaI and NcoI. Plasmids were transformed into Agrobacterium tumefaciens strain GV3101, and used for plant transformation via floral dip (Bartholmes et al., 2008). Agrobacteria were resuspended in infiltration medium (5% sucrose; 0.02% Silwet L-77, http://www.arabidopsis.com/, LEHLE SEEDS, Round Rock, TX, USA) to an OD600 of 2.0, and dipping was repeated three times at 1 week intervals. Ripe seeds were collected, and positively transformed T0 offspring plants were selected by spraying seedlings with 0.1% (A. thaliana) or 0.01% (L. campestre) Basta solution (Aventis CropScience Deutschlang GmbH, Hattersheim, Germany). Ten surviving plants were phenotypically analysed per construct. In case of 35S:FUL, indehiscence was further quantified for three plants per construct applying a random impact test following the protocol described by Arnaud et al. (2010), with modifications. Shaking was performed in the grinding jar of an MM 400 mixer mill (Retsch, Retsch GmbH Germany, Haan, Germany) using six 5 mm steel balls (approximately 0.5 g each). Fruits were agitated at a frequency of 9 Hz for cumulative times of 5, 10, 20, 40, 80 and 160 sec. After each interval, fruits for which at least one valve had completely detached were removed and counted as dehisced.
We thank Ulrike Coja for technical assistance, the staff of the Botanical Garden Osnabrück for plant cultivation, and Lucille Schmieding for correcting style and grammar. We appreciated access to the in situ facilities of the Zachgo laboratory (Department of Biology/Chemistry, University of Osnabrück, Germany). Many thanks also to Pia Nutt and Dajana Lobbes for help with in situ hybridization analyses, to Mariana Mondragón-Palomino for introduction to quantitative RT-PCR, to Lydia Gramzow for support with bioinformatics questions, to Domenica Schnabelrauch (Departmant of Entomology, MPI for Chemical Ecology, Jena, Germany) for her sequencing efforts, and to Nicolas Arnaud and Robert Sablowski (Department of Cell & Developmental Biology, John Innes Centre, Norwich, UK) for valuable information. We are also very grateful to two anonymous reviewers for their helpful comments on a previous version of the manuscript. This work was supported by grants from the Deutsche Forschungsgemeinschaft to K.M. (MU 1137/8-1) and G.T. (TH 417/6-1), and by a grant from the Universitätsgesellschaft Osnabrück to K.M.