SIMPLE LEAF3 encodes a ribosome-associated protein required for leaflet development in Cardamine hirsuta


(e-mail or


Leaves show considerable variation in shape, and may be described as simple, when the leaf is entire, or dissected, when the leaf is divided into individual leaflets. Here, we report that the SIMPLE LEAF3 (SIL3) gene is a novel determinant of leaf shape in Cardamine hirsuta – a dissected-leaved relative of the simple-leaved model species Arabidopsis thaliana. We show that SIL3 is required for leaf growth and leaflet formation but leaf initiation is less sensitive to perturbation of SIL3 activity. SIL3 is further required for KNOX (knotted1-like homeobox) gene expression and localized auxin activity maxima, both of which are known to promote leaflet formation. We cloned SIL3 and showed that it encodes RLI2 (RNase L inhibitor 2), an ATP binding cassette-type ATPase with important roles in ribosome recycling and translation termination that are conserved in eukaryotes and archaea. RLI mutants have not been described in plants to date, and this paper highlights the potential of genetic studies in C. hirsuta to uncover novel gene functions. Our data indicate that leaflet development is sensitive to perturbation of RLI2-dependent aspects of cellular growth, and link ribosome function with dissected-leaf development.


Leaves, the main photosynthetic organs of flowering plants, show tremendous morphological variation. Leaf shape may be classified as simple, when the leaf blade is entire, or dissected (also referred to as compound), when the blade is sub-divided into distinct leaflets. Both simple and dissected leaves initiate as entire structures at the flanks of the pluripotent shoot apical meristem (SAM). However, in dissected leaves, successive generation of lateral growth axes after leaf initiation results in the formation of leaflets. Leaflets form at the flanks of a central stalk called a rachis, and their development requires re-deployment of many of the genetic pathways that drive leaf initiation at the SAM (Hay and Tsiantis, 2006; Barkoulas et al., 2008; Blein et al., 2008; Jasinski et al., 2008; Berger et al., 2009; Koenig et al., 2009; Shani et al., 2010). Genetic analyses in tomato, Cardamine hirsuta and dissected-leaved legumes have highlighted two processes with a key role in both leaf initiation and leaflet development: growth polarization and control of the timing of tissue differentiation.

Growth polarization requires the auxin efflux carrier PINFORMED 1 (PIN1) (Galweiler et al., 1998; Benkova et al., 2003). In the SAM, PIN1 activity facilitates the formation of auxin activity maxima, which underpin leaf initiation (Reinhardt et al., 2003; Heisler et al., 2005). Class I KNOX proteins (KNOXI) maintain the activity of the SAM by preventing tissue differentiation (Long et al., 1996; Vollbrecht et al., 2000). In compound leaves, PIN1 promotes a polar flow of auxin towards sites of initiating leaflets, and establishes local auxin maxima that promote leaflet outgrowth (Barkoulas et al., 2008; Koenig et al., 2009; Scarpella et al., 2010). KNOX1 proteins are also required for leaflet formation by delaying tissue differentiation, which allows cells to respond to signals that promote growth polarization, such as local auxin maxima (Hay and Tsiantis, 2006, 2010; Barkoulas et al., 2008; Shani et al., 2009). Leaflets form within a morphogenetic window defined by the antagonistic activity of the transcriptional regulators KNOX and TCP (TEOSINTE BRANCHED1/CYCLOIDEA/PROLIFERATING CELL FACTOR). In contrast to KNOX1 proteins, class II TCP proteins act in the leaf to accelerate differentiation (Efroni et al., 2008; Shani et al., 2008). Negative regulation of TCP activity is required to maintain the morphogenetic competence of compound leaves to develop leaflets and TCP over-expression prevents leaflet formation (Ori et al., 2007). KNOXI proteins are excluded from most simple-leaved seed plants, indicating that the species-specific regulation of these genes probably contributed to the evolutionary diversification of leaf form (Hay and Tsiantis, 2006; Efroni et al., 2010; Piazza et al., 2010). KNOX and TCP proteins act at least in part by influencing the activity of the hormones cytokinin and gibberellin (Hay et al., 2002; Jasinski et al., 2005; Shani et al., 2010; Fleishon et al., 2011; Yanai et al., 2011). KNOXI-mediated regulation of these two hormones is required for both SAM function and dissected-leaf development, thus defining a common genetic module underlying shoot and dissected-leaf development (Hay et al., 2002; Shani et al., 2010). Another class of shoot regulators that are required for leaflet formation are CUC (CUP-SHAPED COTYLEDONS) proteins, which de-limit developing leaflets along the leaf, thus recapitulating their earlier developmental role in organ separation at the SAM during leaf initiation (Blein et al., 2008; Berger et al., 2009). The activities of CUC and PIN1 proteins are also required for serration development at the margin of simple leaves (Hay et al., 2006; Nikovics et al., 2006; Bilsborough et al., 2011); therefore, current evidence suggests that flexible deployment of a common suite of shoot regulators (PIN1, KNOX1, TCP, CUC) underlies the development of various types of shoot outgrowths (Efroni et al., 2010).

Here, we report the isolation and characterization of the C. hirsuta mutant sil3, which we selected for study because it shows a dramatic reduction in leaflet but not leaf number, and, as such, may provide new insight into dissected-leaf development. We present evidence that SIL3 is required for leaflet initiation, leaf growth, regulation of auxin homeostasis and KNOX gene expression. We show that SIL3 encodes the ATP binding cassette (ABC-type ATPase ABCE2 or RNase L inhibitor 2 (RLI2). In eukaryotes and archaea, the highly conserved protein ABCE1 plays important roles in ribosome-driven protein biosynthesis, coupling translation termination to ribosome recycling, and eventually re-initiation (Andersen and Leevers, 2007; Barthelme et al., 2007, 2011; Verrier et al., 2008; Khoshnevis et al., 2010; Pisarev et al., 2010). Our data reveal C. hirsuta RLI2 as a novel regulator of leaflet development, and offer a route to explore the recently discovered but poorly understood inputs of ribosomal activity into leaf development.


Isolation and characterization of sil3 in C. hirsuta

To investigate whether leaf and leaflet initiation may be uncoupled, we sought to identify mutants that have severe defects in leaflet number without showing defects in leaf number. The sil3 mutant met these criteria, and converted the pinnately dissected leaf of C. hirsuta to a simple shape resembling that of A. thaliana (Figure 1a,b). Whereas leaflets develop on the rachis of successive rosette leaves of wild-type plants in a heteroblastic pattern, sil3 mutants do not initiate leaflets (Figure 1c,d). In contrast to the short petioles of A. thaliana, sil3 leaves have a long rachis, suggesting that lack of leaflet development is unlikely to reflect deletion of the proximal leaflet-producing domain of the leaf (Figure 1d). Leaf primordia of sil3 are smaller than those of wild-type during early development and show impaired outgrowth of both blade and leaflets, resulting in irregular margins with lobes on later leaves (Figure 1c–j). Fewer trichomes were also observed in sil3 leaves compared to wild-type, indicating that cell differentiation processes may be impaired in sil3 leaves (Figure 1e–j and Figure S1). The leaf surface area was reduced in mature sil3 leaves compared to wild-type throughout heteroblastic development (Figure 1k). This reduction of leaf surface area is not simply an indirect consequence of reduced leaflet number in sil3 because it was evident when comparing the first leaf, which does not produce leaflets in either wild-type or the mutant (Figure 1c,d,k). These observations indicate that the SIL3 gene product is required for both leaf growth and leaflet development.

Figure 1.

Phenotypic characterization of the sil3 mutant. (a,b) Top view of wild-type (a) and sil3 (b) plants. (c,d) Leaf silhouettes of successive rosette leaves of wild-type (c) and sil3 (d) plants. (e–j) Scanning electron micrographs of the 4th and 5th developing rosette leaves of wild-type (e,f) and sil3 (h,i), and of a dissected wild-type (g) and sil3 (j) shoot apical meristem bearing initiating leaves. (k) Leaf surface area for rosette leaves 1–8 in sil3 and wild-type plants (= 7 plants per genotype). (l) Growth rate of the 5th rosette leaf in sil3 and wild-type plants (= 19 for wild-type and 17 for sil3). (m) Total rosette leaf number in wild-type and sil3 plants (= 11 plants per genotype). (n) Time taken for sil3 and wild-type plants to initiate rosette leaves 1–8 (= 15 plants per genotype). (o) Quantification of shoot meristem size in wild-type and sil3 embryos (= 19 for wild-type and 17 for sil3 plants). < 0.01 (Student's t-test). (p,q) Scanning electron micrographs of wild-type (p) and sil3 (q) inflorescence meristems. Scale bars = 1 cm (a–d), 500 μm (e,f,h,i), 100 μm (g,j) and 50 μm (p,q).

To understand when sil3 leaf growth deviates from that of the wild-type, we tracked the length of the fifth rosette leaf of wild-type and sil3 plants after initiation from the SAM. Growth rates of sil3 and wild-type leaves were indistinguishable until primordia reached 2 cm in size, when a reduction in leaflet number was already obvious (Figure 1l). After that, the sil3 growth rate gradually declined, culminating in formation of leaves that are 34% shorter on average than wild-type. These results show that SIL3 influences the rate of leaf growth, and indicate that defects in sil3 leaflet formation are revealed before defects in the rate of growth across the proximo-distal axis.

The same numbers of rosette leaves are produced in sil3 and wild-type plants, suggesting that SIL3 may promote leaflet initiation independent of leaf initiation (Figure 1m). However, plastochron length is prolonged by a mean of 1.2 days in the sil3 mutant (Figure 1n). To determine whether this slower rate of leaf initiation in sil3 reflects a severe defect in meristem function, we measured embryonic SAM size and compared the structure of wild-type and sil3 inflorescence meristems. The size of the sil3 embryonic SAM was slightly reduced compared to wild-type, but inflorescence meristems were indistinguishable from wild-type (Figure 1o–q). These observations suggest that sil3 shoot development is characterized by a slight delay in organ initiation in the absence of gross shoot meristem defects. In conclusion, the regulation of leaflet formation and leaf growth is more sensitive than lateral organ initiation to perturbation of SIL3 activity.

SIL3 is required for cell proliferation

To understand whether a reduction in cell proliferation may account for leaf growth defects in sil3, we compared epidermal cell number in the first rosette leaf between wild-type and sil3 plants. We found a 50% reduction in cell number when comparing the surface of the first rosette leaves of sil3 plants to those of wild-type (Figure 2a). As a similar reduction (48%) was observed in the surface area of the first rosette leaves of sil3 plants, it is likely that reduced cell proliferation accounts for the reduced growth in sil3 leaves. We also observed shorter roots in sil3 mutants, associated with a reduction in cell number in the root meristem (Figure S2). In addition, leaf epidermal cell size was slightly increased (16%) in the first rosette leaves of sil3 compared to wild-type (Figure 2b), consistent with the compensatory increases in cell size that are often observed in response to reduced cell proliferation (Tsukaya, 2006; Ferjani et al., 2007; Kawade et al., 2010). Endoreplication is a process associated with increased cell size and differentiation whereby cells undergo a special cell cycle in which DNA continues to be replicated without subsequent cell divisions, leading to cellular polyploidy (De Veylder et al., 2011). Consistent with their increased cell size, we found a higher nuclear content in sil3 cells compared to wild-type, even in young leaves, indicating an earlier onset of endoreplication (Figure 2c,d and Figure S3). In summary, our data suggest that SIL3 may promote leaf growth via its effect on cell proliferation.

Figure 2.

SIL3 is required for cell proliferation. (a) Quantification of cell number in the adaxial surface of the 1st rosette leaf in wild-type and sil3 (= 5 plants per genotype). < 0.01 (Student's t test). (b) Quantification of cell size in the adaxial surface of the 1st rosette leaf in wild-type and sil3 (= 5 plants per genotype). < 0.01 (Student's t test). (c,d) DNA content calculated as a percentage of total nuclei assayed in cells of rosette leaves 1 and 2 (c) and rosette leaves 3 and 4 (d) in wild-type and sil3 (= 3–6 plants per genotype). The samples from leaves 3 and 4 represent a younger developmental stage than the samples from leaves 1 and 2, as shown in Figure S3. (e–h) In situ localization of ChHISTONE4 (e,g) and ChCYCLINB1;1 (f,h) mRNA on longitudinal sections of wild-type (e,f) and sil3 (g,h) vegetative shoot apices. Scale bars = 50 μm.

To understand whether the lack of leaflets in sil3 is associated with altered cell proliferation at sites of leaflet formation, we studied expression of the C. hirsuta CYCLIN B1;1 (CYCB1;1) and HISTONE 4 (H4) cell division markers by in situ hybridization (Dinneny et al., 2004; Hay and Tsiantis, 2006; Barkoulas et al., 2008). In contrast to wild-type, no regions of localized cell proliferation were observed in the margins of sil3 leaves (Figure 2e–h). Therefore, SIL3 contributes to the organisation of cell proliferation foci that support leaflet outgrowth.

SIL3 influences auxin homeostasis

The local auxin activity maxima that underpin leaflet outgrowth at the C. hirsuta leaf margin are specified by polar localization of the auxin efflux protein PIN1 (Barkoulas et al., 2008). To test whether PIN1 expression and auxin activity maxima were altered in the sil3 background, we analysed the expression pattern of a PIN1::PIN1:GFP reporter and an auxin activity reporter, DR5::VENUS (Heisler et al., 2005; Barkoulas et al., 2008). In wild-type leaves, PIN1 accumulated in the epidermis and vasculature of lateral leaflets (Figure 3a), and its polar localization at the margin of the leaf rachis predicted a flow of auxin towards the tips of developing leaflets (Figure 3b). In comparison to wild-type, PIN1 expression was much reduced in sil3 mutant leaves, and declined rapidly as primordia developed (Figure 3c and Figure S4). No PIN1 expression was observed in the leaf rachis where lateral leaflets emerged in wild-type, even under strongly increased laser excitation (Figure 3d). Under these imaging conditions (see 'Experimental procedures'), weak expression of PIN1 was observed in the margin of the terminal leaflet associated with developing serrations, but the low level of expression made it difficult to determine convergence points accurately (Figure 3e and Figure S4). In comparison to wild-type, no DR5 expression was observed in the rachis of sil3 leaves (Figure 3f,g) indicating that auxin activity maxima associated with leaflet initiation do not form in sil3. mRNA expression levels of ChPIN1, an auxin-responsive gene, were also reduced in sil3 seedlings, indicating a reduction in auxin-responsive gene expression (Figure 3h). To test whether sil3 showed reduced sensitivity to auxin in comparison to wild-type, we assayed root growth in response to treatment with the synthetic auxin 2,4-dichlorophenoxyacetic acid. We observed that sil3 roots were significantly less sensitive than wild-type at all auxin concentrations tested, confirming that sil3 mutants show a reduced auxin response (Figure 3i).

Figure 3.

Auxin homeostasis is perturbed in sil3 mutants. (a–e) pPIN1::PIN1-GFP expression in wild-type (a,b) and sil3 (c–e) 5th rosette leaf primordia. (b) Lateral leaflet primordium developing at the margin of the leaf rachis; arrowheads indicate a predicted flow of auxin towards a convergence point indicated by the asterisk. Close-up views of (d) the rachis and (e) the boxed section indicated in (c), imaged using increased laser excitation. (f,g) DR5:VENUS expression in wild-type (f) and sil3 (g) 5th and 6th rosette leaf primordia. The leaf rachis is bracketed and labelled ‘r’ in each panel. (h) Real-time quantitative RT-PCR analysis of relative ChPIN1 mRNA levels in sil3 seedlings compared to wild-type. < 0.01 (Student's t-test). (i) Dose–response assay comparing root growth in wild-type and sil3 in response to increasing concentrations of the synthetic auxin 2,4-dichlorophenoxyacetic acid (= 16 plants per genotype for each treatment; < 0.01, ancova). (j–m) Top view of wild-type (j), sil3 (k), pin1 (l) and sil3 pin1 (m) mutant plants. (n–q) Leaf clearings of wild-type (n,o) and sil3 (p,q), showing a fully developed 5th rosette leaf and a close-up of the vasculature architecture for each genotype. The cross in (q) indicates a vascular island. (r) Number of lateral branches produced in sil3 compared to wild-type plants (= 20 plants per genotype). Scale bars = 100 μm (a,c), 10 μm (b), 50 μm (d–g), 0.5 cm (j–m) and 500 μm (n–q).

To further investigate the relationship of SIL3 with the auxin homeostasis pathway, we generated sil3 chpin1 double mutant plants. These plants showed arrested development at the cotyledon stage or after initiation of one aberrant leaf (Figure 3j–m). Interestingly, chpin1 single mutants displayed similar defects but with lower penetrance (31.3%), suggesting that SIL3 and ChPIN1 are redundantly required for leaf initiation. We next analysed the vascular architecture of sil3 leaves, because alterations in vasculature architecture are often associated with defects in auxin response (Mattsson et al., 1999; Sieburth, 1999). sil3 mutants exhibited a number of vascular defects, including the presence of vascular islands and a reduction in secondary vein number and free-ending veinlets (Figure 3n–q and Table 1). Moreover, sil3 plants are bushier compared to wild-type, with increased axillary branching, similar to mutants with reduced auxin sensitivity (Figure 3r) (Stirnberg et al., 1999). Taken together, our data suggest that SIL3 is likely to influence auxin homeostasis.

Table 1. Vascular development differs between sil3 mutant and wild-type leaves
GenotypeNumber of secondary veinsNumber of free-ending veinlets per cm2 leaf areaPresence of vascular islands
Wild-type10.7 ± 0.2333.8 ± 1.78No
sil3 5 ± 0.2218.7 ± 1.28Yes

SIL3 is required for KNOXI gene expression

KNOXI gene expression is necessary and sufficient for leaflet initiation in C. hirsuta (Hay and Tsiantis, 2006). To investigate whether alterations in KNOXI gene expression are implicated in the simple-leaf phenotype of sil3, we analysed expression of the KNOXI genes shootmeristemless (STM ) and brevipedicellus (BP ) by promoter fusions with the uidA gene encoding β-glucuronidase (GUS) and real-time quantitative RT-PCR. Transcript accumulation for both genes was reduced in sil3, suggesting that SIL3 is required for correct expression of BP and STM in the leaf (Figure 4a–f). However, it is possible that these alterations in KNOXI gene expression are secondary effects of leaf simplification in sil3. To test whether increased KNOXI gene expression is sufficient to bypass the requirement for SIL3 in leaflet development, we analysed double mutants between sil3 and both 35S::BP and C. hirsuta asymmetric leaves1 (chas1) plants, which show elevated and broadened expression of BP and increased leaflet number (Hay and Tsiantis, 2006). These double mutants showed additive phenotypes, with simple leaves, similar to those of sil3, that were proximo-distally compressed, similar to those of chas1 and 35S::BP (Figure 4g–l). The fact that increased BP transcription in the leaf is not sufficient to bypass the leaflet formation defect in sil3 suggests that reduced KNOX expression is unlikely to be the sole cause of leaflet loss in sil3 or that SIL3 is required post-transcriptionally for KNOX activity.

Figure 4.

KNOXI expression is reduced in sil3 plants but elevated BP expression is not sufficient to bypass leaflet initiation defects in sil3. (a,b) ChSTM > > GUS expression in the 5th developing leaf of wild-type (a) and sil3 (b) plants. (c,d) ChBP::GUS expression in the 5th developing leaf of wild-type (c) and sil3 (d) plants. (e,f) Fold change of ChSTM (e) and ChBP (f) mRNA levels in sil3 compared to wild-type seedlings as quantified by real-time quantitative RT-PCR. (g–l) Silhouettes of fully developed 5th and 6th rosette leaves of wild-type (g), sil3 (h), 35S::BP (i), sil3;35S::BP (j), as1-1 (k) and sil3 as1-1 (l). Scale bars = 100 μm (a–d) and 1 cm (g–l).

Molecular cloning of SIL3

To understand the molecular basis of the sil3 mutant phenotype, we used a map-based cloning approach. We used 1022 recombinant chromosomes generated by crossing sil3, isolated in the Oxford C. hirsuta accession, with a polymorphic Greek accession, and mapped SIL3 to a 32.9 kb physical interval on C. hirsuta linkage group G (Figure 5a) This interval is syntenic to a 21.4 kb segment of A. thaliana chromosome 4 that contains six genes (AT4G19185–AT4G19220) but is interrupted between AT4G19200 and AT4G19210 by 11.5 kb of sequence homologous to A. thaliana chromosome 5 (Figure 5b). Comparative sequence analysis of the SIL3 physical interval in wild-type and the sil3 mutant revealed a C/T substitution in the ChRLI2 gene (C. hirsuta orthologue of AT4G19210) (Figure 5e). To understand whether this mutation caused the sil3 phenotype, we performed co-segregation analysis and did not observe recombination between sil3 and the ChRLI2 locus in 50 plants. Furthermore, we performed genetic complementation analysis and introduced a 6.5 kb genomic region of the ChRLI2 wild-type locus into sil3 mutant plants. We obtained ten transgenic lines, all of which had a wild-type phenotype (Figure 5f–i). We therefore conclude that SIL3 corresponds to ChRLI2. This gene belongs to the ATP-binding cassette E (ABCE) sub-family of ABC proteins that are conserved in archaea and eukaryotes (Braz et al., 2004). RLIs have an N-terminal FeS cluster and two nucleotide-binding domains that are involved in ATP binding and cleavage (Figure 5c) (Braz et al., 2004; Verrier et al., 2008). The sil3 C/T substitution results in a non-conservative P177L substitution in the first nucleotide-binding domain (Figure 5d). As such, this mutation may perturb ATP binding and subsequent conformational changes that are involved in the regulation of ribosome recycling by RLI proteins (Becker et al., 2012).

Figure 5.

Molecular cloning and characterization of SIL3. (a) Schematic representation of the linkage map around the sil3 locus generated by fine mapping of 1022 chromosomes. Scale bar = 1 cM. (b) Annotation of the candidate genes within the sil3 genetic interval predicted by sequence similarity with the A. thaliana genome. Grey shading indicates the chromosome 5 insertion. Scale bar = 1 kb. (c) Schematic representation of the ChRLI2 protein domains. (d) A C/T non-conservative substitution in sil3 converts amino acid 177 from Pro to Leu in the NBD1 protein domain of ChRLI2 (indicated by a black arrowhead). (e) Dideoxy sequence chromatographs of ChRLI2 showing the C/T mutation in sil3. (f–i) Top view of whole plant and leaf silhouettes of successive rosette leaves for sil3 (f,g) and a transgenic plant obtained by transforming 6.5 kb of the ChRLI2 wild-type sequence into the sil3 background (h,i). (j,k) In situ hybridization using the ChRLI2 antisense probe on longitudinal (j) and transverse (k) wild-type sections. (l–p) A. thaliana wild-type Col-0 (l), A. thaliana 35S::ChRLI2 (m,n), C. hirsuta wild-type (o) and C. hirsuta 35S::ChRLI2 (p) . Wild-type and transgenic plants were photographed at the same age to show shoot growth arrest in 35S::ChRLI2 plants. Scale bars = 1 cm (f–i), 50 μm (j,k) and 0.5 cm (l–p).

To investigate the spatial expression of ChRLI2, we performed in situ hybridization. We found that ChRLI2 is expressed within the meristem and young developing leaves. including the sites of leaflet initiation and the vasculature (Figure 5j,k). Thus, the expression pattern of ChRLI2 is consistent with its function in leaf development and plastochron duration. To test whether the correct spatio-temporal pattern and level of ChRLI2 expression is important for shoot development, we over-expressed ChRLI2 under the control of the CaMV 35S promoter, which confers a broad pattern and high level of expression. We found that ChRLI2 over-expression resulted in shoot growth arrest at various stages of both C. hirsuta and A. thaliana development, with the most severe phenotype being an absence of visible leaves (Figure 5l–p and Figure S5). Similar phenotypes were seen in 35S::AtRLI2 plants (Figure S5), indicating that RLI2 over-expression is detrimental to shoot development.


rli2 mutant characterization and implications for RLI function in plants

Here we describe a mutant in the C. hirsuta orthologue of ABCE2/RLI2, and demonstrate that it is a novel component of dissected-leaf development pathways. The ABCE sub-family of ABC ATPases participates in multiple aspects of ribosome action, including biogenesis, recycling, translation initiation and termination, and are conserved in archaea and eukaryotes (Zhao et al., 2004; Kispal et al., 2005; Yarunin et al., 2005; Chen et al., 2006; Barthelme et al., 2007, 2011; Khoshnevis et al., 2010; Pisarev et al., 2010;; Becker et al., 2012). The A. thaliana genome contains three ABCE/RLI genes (Braz et al., 2004) indicating an increase in RLI gene copy number with respect to some other eukaryotic taxa, which may have resulted in genetic redundancy or sub-functionalization that has so far confounded isolation of RLI mutants. An alternative reason why RLI mutants have not yet been isolated in A. thaliana may be that such mutants are non-viable. Our isolation of a probably hypomorphic allele of RLI2 in C. hirsuta, indicates that dissected leaf development is sensitive to perturbation of ribosomal activity and highlights the potential for genetic studies in C. hirsuta to uncover novel gene functions.

Genetic evidence from Drosophila suggests that RLI is important for both ribosome function and organism development, as mutants in the fly RLI orthologue pixie exhibit reduced growth rate and slender bristles, resembling the Minute class of mutants, several of which correspond to lesions in structural ribosomal genes (Coelho et al., 2005; Andersen and Leevers, 2007). Pixie interacts with eukaryotic translation initiation factor 3 (eIF3) and the 40S ribosome subunit, and is necessary for translation initiation in Drosophila, similar to its yeast orthologue Rli1p (Kispal et al., 2005; Yarunin et al., 2005), suggesting that at least some of the specific biochemical roles of RLI ATPases in ribosome function are ancient. Therefore, ChRLI2 is likely to play a similar role in ribosome function, but this has yet to be investigated.

Interpreting the sil3 phenotype via ribosomal inputs into leaf development

In A. thaliana, mutations that perturb ribosomal function lead to a broad spectrum of leaf abnormalities, including smaller, narrower and pointier leaves with fewer trichomes (Byrne, 2009; Szakonyi and Byrne, 2010, 2011). These defects have been interpreted as general growth and differentiation defects, consistent with what is reported here for sil3 and also with results reported for virus-induced silencing of RLI genes in Nicotiana (Petersen et al., 2004). However, perhaps more intriguing are indications that ribosome activity provides inputs into leaf axial patterning. For example, the distinctive piggyback phenotype in Arabidopsis is caused by mutations in several ribosomal proteins, which, when combined as double mutants with asymmetric leaves1 (as1), produce leaves with perturbed axial tissue polarity that develop outgrowths on the adaxial side (Pinon et al., 2008; Yao et al., 2008). piggyback (pgy) mutants alone have mild developmental defects, such as pointy and serrated leaves, suggesting that they act redundantly with as1 (a negative regulator of KNOX gene expression) to control leaf axial polarity.

Similar to PGY genes, ChRLI2 is likely to play a role in ribosome function and leaf development. However, in contrast to pgy mutants, sil3 has a very striking phenotype as a single mutant and fails to enhance a chas1 mutant. In fact, sil3 leaves show reduced KNOX gene expression, which suggests that sil3 and pgy mutations may perturb distinct aspects of leaf development. Alternatively, these differences may reflect species-specific gene functions in development of simple versus dissected leaves, and characterization of viable RLI2 mutants in A. thaliana will help to answer this question. However, although ribosome activity is the most likely function of RLI2, it is possible that this protein may contribute to other processes. For example, over-expression of the A. thaliana RLI2 gene in tobacco has been reported to suppress transgene-mediated silencing (Braz et al., 2004; Sarmiento et al., 2006). As gene silencing pathways may provide species-specific inputs into leaf development pathways (Yan et al., 2010), it will be useful to understand whether RLI2 links silencing with translational control and how such action influences leaf development. Answering this question will also involve understanding the precise biochemical functions of RLI2 in vivo.

Sil3 mutant phenotype reveals links between cellular growth and compound leaf shape

An interesting aspect of the sil3 leaf phenotype is that leaflet loss is associated with slow growth and reduced cell proliferation. In tomato, the LYRATE gene regulates leaflet initiation and encodes the tomato orthologue of JAGGED, which is a positive regulator of cell division and lateral organ outgrowth in A. thaliana (Dinneny et al., 2004; Ohno et al., 2004; David-Schwartz et al., 2009). The more conspicuous leaf phenotype of lyrate compared to jagged mutants supports the idea that compound-leaf development is particularly sensitive to alterations in cell proliferation. It is possible that the burst of growth and cell proliferation associated with leaflet production imposes high energy demands on cells at the sites of leaflet initiation. In this scenario, the absence of optimal ribosome function in sil3 mutants may prevent those energy demands being met, thus failing to sustain the local growth bursts that support leaflet formation. Leaflet loss in sil3 leaves is also associated with accelerated onset of endoreplication, such that cells differentiate precociously. Such premature cell differentiation in regions of the leaf where leaflets usually arise may interfere with activation of the mitotic cell cycle, which is required for leaflet production. The degree to which the sil3 leaf phenotype reflects sensitivity to reduced cellular energy levels, cessation of cell-cycle activity, or reduced expression of developmental patterning genes, such as KNOX and PIN1, remains to be determined. These explanations are not mutually exclusive, and feedback between ribosome-dependent growth and developmental gene expression is an interesting possibility to consider.

ChRLI2 is required for leaflet-associated auxin activity maxima

Leaf simplification in sil3 is accompanied by loss of auxin activity maxima in the leaf margin and loss of cell division foci that are normally associated with leaflet initiation. Although it is possible that these are secondary effects of leaf simplification, the formation of auxin activity maxima precedes leaflet initiation in the wild-type leaf margin (Barkoulas et al., 2008), and we observed that DR5 expression was absent from the rachis of sil3 mutant leaves prior to the appearance of morphological differences between sil3 and wild-type (Figure 3f,g). As DR5 expression is the earliest known marker of leaflet formation (Barkoulas et al., 2008), this result indicates that leaflet primordia in sil3 are not arrested in their growth but rather fail to initiate. Therefore, ChRLI2 provides patterning input to compound-leaf development by contributing to the formation of leaflet-associated auxin activity maxima at the leaf margin. Phenotypic analysis suggests that auxin homeostasis may be perturbed overall in sil3 plants, although given the probable requirement for ChRIL2 in the fundamental cellular process of translation, these effects may be indirect. Analysis of a double mutant between sil3 and chpin1 suggested that SIL3 acts redundantly with the auxin transporter ChPIN1 to promote leaf inception and shoot growth, consistent with the observation that ChRLI2 is required both for expression of the auxin-responsive gene ChPIN1 and a normal auxin response. Thus, in addition to having a broad role in leaf growth by supporting ribosome biogenesis, ChRLI2 may provide more specific inputs into dissected-leaf development by influencing auxin activity at the leaf margin via processes that remain unknown.

Our observations linking ribosome function with auxin homeostasis are also consistent with work in A. thaliana suggesting mechanistic links between auxin and ribosome-mediated translational control. For example, disruption of AthNUC-L1/PARALLEL1, a nucleolin implicated in multiple steps of ribosome biogenesis, results in shorter plants with reduced apical dominance and narrow pointed leaves with vascular pattern defects, which are associated with altered auxin distribution (Petricka and Nelson, 2007). Additionally, short valve1 (stv1), a mutation in the ribosomal protein L24 that is involved in translation re-initiation of polycistronic genes, exhibits defects in apical/basal gynoecium patterning and was found to act in a common genetic pathway with ETTIN/ARF3 (ETT) (Nishimura et al., 2005). Furthermore, the translational efficiency of ETT and MONOPTEROS/ARF5 (MP), which is probably influenced by 5′ leader upstream open reading frames, was disrupted in stv1. Therefore, defects in ribosome biogenesis may disrupt the translation of mRNAs necessary to determine auxin distribution in the developing leaf. Future experiments will clarify whether perturbed expression of genes influencing auxin homeostasis are responsible for the simple-leaf phenotype of sil3 in C. hirsuta.

Experimental procedures

Accession number

The C. hirsuta RLI2 cDNA sequence has been deposited in GenBank with accession number JX097073.

Plant material and growth conditions

Plants were grown in a greenhouse with supplemental lighting under 18 h days (20°C) and 6 h nights (16°C). Wild-type C. hirsuta of the reference Oxford accession [specimen voucher Hay 1 (OXF)] has been described previously (Hay and Tsiantis, 2006), and the Greek accession was collected from wild populations from Athens, Greece, and self-pollinated for seven generations before use. Construction of the ChBP::GUS reporter and the 35S::BP line in C. hirsuta, and isolation of the chas1-1 mutant, have been described previously (Hay and Tsiantis, 2006), as have the ChSTM > > GUS reporter and the chpin1-1 mutant (Barkoulas et al., 2008). The sil3 mutant was isolated from an ethane methyl sulfonate mutagenesis screen, and is monogenic recessive. sil3 mutant characterization was performed after backcrossing to wild-type four times. To introduce the reporters PIN1::PIN1::GFP, DR5::VENUS, STM > > GUS and BP::GUS into sil3, homozygous plants were crossed and expression analysis was performed in sil3 plants in the F2 and F3 generations. To generate sil3 chpin1-1 double mutants, sil3 plants from the F2 generation of a cross between heterozygous chpin1-1 and homozygous sil3 parents were self-pollinated to generate F3 families that segregated one quarter sil3 pin1-1 double mutants. To generate sil3 chas1-1 double mutants and sil3 35S::BP lines, homozygous plants were crossed and genotypes analysed in the F2 and F3 generations.


Wild-type C. hirsuta seed [specimen voucher Hay 1 (OXF)] was mutagenized by agitation with 0.2% ethane methyl sulfonate for 10 h, washed extensively with water, then sown and harvested in pools of five plants. Approximately 2500 M2 plants were screened.

Molecular biology

A ChRLI2pr::ChRLI2 genomic clone was constructed in pMDC123 (Curtis and Grossniklaus, 2003). A 6.5 kb genomic fragment, comprising 2.5 kb promoter sequence and 4 kb sequence containing the ChRLI2 open reading frame, was PCR-amplified from a BAC containing ChRLI2 (Arizona Genomics Institute Ch_0Ba6_2P20) using primers Btw48100-19210_F5 (5′-CAGATTCCCGAGAACAGCTT-3′) and 19210R9 (5′-GACAACTAGCATGAAGGTTCTTGA-3′) using high-fidelity Pfu Ultra II (Agilent Technologies,, and cloned into the intermediate vector pTOPO (Invitrogen, The genomic fragment was transferred into pMDC123 by an LR reaction, and introduced into sil3 mutants via Agrobacterium tumefaciens-mediated transformation by floral dip as described previously (Hay and Tsiantis, 2006). 35S::ChRLI2 and 35S::AtRLI2 plasmids were constructed in pART7 (Gleave, 1992). Coding sequences were PCR-amplified from cDNA template using the following primers for ChRLI2 (ChRLI2-XhoI-F, 5′-CTCGAGATGGCAGATCGATTGACG-3′; ChRLI2-HindIII-R, 5′-AAGCTTTTAATCGTCCAAGTAGTAGTATGAG-3′) and AtRLI2 (athRLI2-XhoI-F, 5′-CTCGAGATGGCAGATCGATTGACA-3′; athRLI2-HindIII-R, 5′-AAGCTTCTAATCATCCAAGTAGTAGTATGAG-3′) (restriction sites are underlined). Amplified fragments were cloned into the XhoI and HindIII sites of the intermediate vector pART7 containing a CaMV 35S promoter and the octopine synthase (OCS) terminator sequence. Expression cassettes were transferred into the binary vector pMLBART (Gleave, 1992), and transformed into wild-type C. hirsuta and A. thaliana plants by floral dip.

In situ hybridization and microscopy

Fixation and hybridization were performed as previously described on 8 μm paraffin sections using digoxigenin-labelled C. hirsuta H4 antisense and sense probes (Hay and Tsiantis, 2006). To generate a probe to C. hirsuta CYCB1;1, a 684 bp fragment was amplified from C. hirsuta shoot cDNA using primers ChCYCB1;1F (5′-CCTGTAGTCAAGAGAAACGCAG-3′) and ChCYCB1;1R (5′-CTGTTTGTGACTGTATGCATGG-3′), and cloned into pGEM-T Easy (Promega, To generate a C. hirsuta RLI2 probe, a 300 bp cDNA fragment was amplified using primers 19210Exon3F (5′-CCTCAATATGTTGACCACATCC-3′) and 19210Exon3R (5′-GCCTCAGGAGAGAACGAACA-3′), and cloned into pTOPO (Invitrogen).

Scanning electron microscopy

Tissue was fixed overnight in 3% gluteraldehyde, 25 mm phosphate buffer pH 7.0 at 4°C, washed twice in PBS and dehydrated through an ethanol series from 15 to 100% as described previously (Hay and Tsiantis, 2006). Samples were dried using a Tousimis AutoSamDri-814 critical point dryer (, sputter-coated with palladium, and viewed on a JSM-5510 microscope (Jeol,

Confocal laser scanning microscopy

Fresh samples were mounted in water with no fixation. Confocal microscopy was performed using a Zeiss 510 LSM Meta microscope ( and a 10 × objective (Plan Neofluar), and water-dipping 25 × (Plan Neofluar) and 40 × (C Apochromat) objectives, as previously described (Barkoulas et al., 2008), or a Leica TCS SP5 II microscope ( and a 10 × objective (HC PL FLUOTAR 10 × 0.30) and a water dipping 20 x objective (HCX APO LU-V-I 0 × 0.5) . Single plane sections or projections from stacks of 5-25 sections are shown in Figure 3(a–g). To image GFP, a 458 nm argon laser with 33% laser excitation was used, with a 646–700 nm filter for the chlorophyll channel and a 475–525 nm band-pass filter for GFP. To analyse pPIN1::PIN1-GFP in sil3 mutants, the laser excitation was increased to 43%. VENUS was detected using a 488 nm argon laser with a 657–743 nm filter for the chlorophyll channel and a 505–550 nm band-pass filter.

GUS staining

GUS activity was detected as described previously (Hay et al., 2004). Tissue was fixed in 90% acetone at −20°C for 1 h, washed briefly with 100 mm phosphate buffer, and stained overnight in freshly prepared 100 mm sodium phosphate buffer with 10 mm sodium EDTA, 1 mm 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid (Invitrogen), and ferrocyanide and ferricyanide salts (2 mm). Reactions were terminated using 95% ethanol, and leaves were dissected and mounted in 50% glycerol.

Light microscopy

Imaging of leaf tissue clearings and GUS-stained samples was performed using Olympus BX50 ( and Nikon SMZ 1000 ( microscopes. Differential interference contrast optics were used to enhance the contrast of the specimens. Images were captured with a digital camera using Image-Pro 6.2 or Infinity Capture1 software (Microimaging Applications Group,

Quantitative RT-PCR analysis

Total RNA (1 μg) extracted from mature leaf tissue using a Qiagen RNeasy kit ( was treated with DNase I and used for cDNA synthesis with an oligo(dT) primer and Superscript reverse transcriptase (Invitrogen). cDNA was amplified using an ABI PRISM 7300 sequence detection system (Applied Biosystems, Amplification reactions were prepared using a SYBR Green PCR Master kit (Applied Biosystems) according to the manufacturer's instructions, with 0.8 μm primers and 10 μl cDNA (diluted 1:10 with water) per reaction. Each reaction was performed in triplicate, and each experiment was repeated three times. The efficiency of each set of primers and calculation of the level of induction was determined as described by Pfaffl (2001). The error bars represent the standard error calculated on experiment repetitions. Expression levels were normalized against values obtained for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which was used as an internal reference gene as described previously (Cnops et al., 2004). The primers used were ChGAPDH qRT F (5′-TGACCACCGTCCACTCCATCAC-3′), ChGAPDH qRT R (5′-GCTCTTCCACCTCTCCAGTCCTTC-3′), ChSTM qRT F (5′-GTCAAGGCAAAGATCATGGCT-3′), ChSTM qRT R (5′-TGGGGCTCCAACCTTCTGGCA-3′), ChBP qRT F (5′-ACTCTTCCCATCAGGATTGTTGA-3′), ChBP qRT R (5′-CCATTCAAGAAGCAATGGAGTT-3′), ChPIN1 qRT F2 (5′-GTGCATCCTGACATTCTTAGCACT-3′) and ChPIN1 qRT R2 (5′-CGCAATCAGCATCCCAAATAT-3′).

Semi-quantitative RT-PCR analysis

RNA extraction and cDNA synthesis were performed as described above using the seedlings shown in Figure 5. Twenty PCR cycles were performed with 2 μl cDNA per reaction (from undiluted cDNA and a 1:10 dilution of the cDNA) using the primers GAPDH F (5′-CACTTGAAGGGTGGTGCCAAG-3′), GAPDH R (5′-CCTGTTGTCGCCAACGAAGTC-3′), 19210 F7 (5′-TATGGGAAACCAGGTGCCTA-3′) and 19210 R8 (5′-CGAGATATGCACTTGGCTCA-3′).

Leaf tissue clearings

Mature leaves or whole plants were fixed overnight in 7:1 v/v ethanol/acetic acid, and then cleared in 8:2:1 w/v/v chloral hydrate/50% glycerol/water for several days before mounting in 50% glycerol.

Aniline blue staining and confocal microscopy

Mature embryos were removed from the seed coat, stained with aniline blue as described previously (Bougourd et al., 2000), and viewed under a Zeiss 510 LSM Meta microscope ( using an argon laser (514 nm excitation). The protocol was modified such that embryos were stained for 2 h with 0.5% aniline blue. Meristem size measurements were performed on median confocal sections using Zeiss LSM Image Browser software.

Phenotypic measurements

To obtain leaf silhouettes, fully developed leaves were flattened onto clear adhesive on white paper and digitally scanned. Image Tool software (University of Texas Health Science Center, San Antonio, TX) was used to measure the mean surface area from these silhouettes for rosette leaves 1–8. The growth of the fifth rosette leaf was tracked by measuring its length each day starting from 0.4 cm in length until flowering. Plastochron length was measured by daily scoring for the emergence of new leaves starting from leaf 1 to leaf 8, and the difference between genotypes was averaged over these eight time points. To measure cell size, epidermal impressions were taken, as previously described (Hay and Hake, 2004), from the 1st rosette leaves of wild-type and sil3 plants collected after flowering. Image Tool software was then used to measure the mean cell area. Cell number was derived by dividing mean lamina area by cell size. ImageJ software (National Institutes of Health, Bethesda, MD) was used to measure root length. The number of meristematic cortex cells was measured as described by Casamitjana Martinez et al. (2003).

Auxin sensitivity assay

Seedlings were germinated and grown in continuous light on vertical plates containing Murashige and Skoog (MS) solid medium supplemented with 2,4-dichlorophenoxyacetic acid (Sigma, dissolved in dimethyl sulfoxide (Sigma). Root length was tracked daily and measured at 11 days post-germination for approximately 15 seedlings per genotype per treatment. Analysis of covariance (ancova) was used to test for differences between mutant and wild-type at each 2,4-dichlorophenoxyacetic acid concentration.

Fluorescence-activated cell sorting

Young and old rosette leaves of 5-week-old plants were finely chopped in 400 μl of nuclei extraction buffer and mixed with 1.6 ml staining buffer containing DAPI (CyStain UV precise P kit, Partec, After a short incubation, the suspension was purified by sieving through a 50 μm CellTrics filter (Partec), and the fluorescence of the remaining nuclei was analysed using a Cyflow ploidy analyser (UV-LED 365 nm) and CyView 2.2 software (Partec).


We thank M. Barkoulas for materials and assistance and J. Baker for photography. This work was supported by Biotechnology and Biological Sciences Research Council grants BB/D010977/1 and BB/F012934/1 to M.T. and BB/H01313X/1 to A.H, and by Deutsche Forschungsgemeinschaft ‘Adaptomics’ grant TS 229/1-1 to M.T. and A.H. E.K. was supported by the Greek State Scholarship Foundation, A.H. by a Royal Society University Research Fellowship, and M.T. by a Royal Society Wolfson Merit award and the Gatsby Foundation.