ABSTRACT
A clear example of interspecific variation is the number of root cortical layers in plants. The genetic mechanisms underlying this variability are poorly understood, partly because of the lack of a convenient model. Here, we demonstrate that Cardamine hirsuta, unlike Arabidopsis thaliana, has two cortical layers that are patterned during late embryogenesis. We show that a miR165/6-dependent distribution of the HOMEODOMAIN LEUCINE ZIPPER III (HD-ZIPIII) transcription factor PHABULOSA (PHB) controls this pattern. Our findings reveal that interspecies variation in miRNA distribution can determine differences in anatomy in plants.
INTRODUCTION
A key question in evolutionary genetics is how morphological diversities between species originate. Owing to the high adaptive potential conferred to plants, roots have evolved a variety of anatomical patterns. In fact, roots have largely contributed to the evolutionary success of land plants, enabling them to adapt to different soils and conditions (Fahn, 1990). The cortex is one of the tissues that contributes most to root adaptive potential. In plants living on wet soils, such as rice, secondary growth of the cortex gives rise to the aerenchyma (Fahn, 1990), a tissue controlling the air/water ratio, whereas in plants living in unfavourable conditions, such as turnip, the cortex gives rise to storage parenchyma, a tissue where carbohydrates such as starch are stored (Fahn, 1990). Cortical layer number varies in different plant species ranging from one, as in Arabidopsis, to several, as in rice in which it can vary from zero to ten, to horseradish and barley, which have four cortical layers (Heimsch and Seago, 2008; Pauluzzi et al., 2012; Kirschner et al., 2017). Hence, the cortex provides an ideal model system in which to study the genetic mechanisms giving rise to morphological differences in plants.
The cortex, together with the endodermis, forms the root ground tissue (GT). In the Arabidopsis root meristem, cortex and endodermis originate from a stem cell (the cortex endodermis initial, CEI); a first asymmetric anticlinal division of the CEI gives rise to a self-renewed stem cell (CEI) and to a daughter cell (CEID). Subsequently, a periclinal division occurs in the CEID and generates endodermis and cortex (Fig. 1A) (Dolan et al., 1993; Benfey et al., 1993; Scheres et al., 1994; Di Laurenzio et al., 1996; Pauluzzi et al., 2012). This mechanism is established during embryogenesis and continues throughout the lifespan of the plant (Dolan et al., 1993; Benfey et al., 1993; Scheres et al., 1994; Pauluzzi et al., 2012).
In Arabidopsis, GT patterning is controlled by several genetic pathways that include the genes SHORTROOT (SHR) and SCARECROW (SCR), which encode transcription factors, and CYCLIN D6;1 (CYCD6;1) encoding a cell cycle regulator (Helariutta et al., 2000; Pauluzzi et al., 2012; Sozzani et al., 2010; Cruz-Ramírez et al., 2012). The SHR protein is produced in the root vasculature and moves into the endodermis to the CEI and CEID where it is bound and sequestered to the nucleus by SCR and the BIRD zinc finger protein (Cui et al., 2007; Welch et al., 2007; Pauluzzi et al., 2012; Cruz-Ramírez et al., 2012; Long et al., 2015b; Moreno-Risueno et al., 2015; Clark et al., 2016; Long et al., 2017). The SHR/SCR complex in concert with BIRD proteins activates expression of CYCD6;1 in the CEI, triggering the periclinal division that gives rise to the cortex and the endodermis layers (Sozzani et al., 2010; Pauluzzi et al., 2012; Cruz-Ramírez et al., 2012; Long et al., 2015a,b). It was recently suggested that BIRD proteins act as GT post-embryonic organizers, determining CEI, cortical and endodermal identity (Moreno-Risueno et al., 2015).
MicroRNAs (miRNAs), small non-coding RNAs acting as morphogens in plants (Skopelitis et al., 2012; Baulcombe, 2004), are also involved in GT patterning (Miyashima et al., 2009). miRNA-machinery mutants such as ago1-102 and hyl1-2 pattern an extra GT layer during embryogenesis (Miyashima et al., 2009). In Arabidopsis, miR165/6 are involved in the spatial restriction of the HOMEODOMAIN LEUCINE ZIPPER III (HD-ZIPIII) transcription factors PHABULOSA (PHB), PHAVOLUTA (PHV), REVOLUTA (REV), CORONA (CNA) and HOMEOBOX GENE 8 (ATHB8) (Mallory et al., 2004). Failure in this restriction generates several developmental anomalies, such as lack of embryo apical/basal and leaf adaxial/abaxial polarity formation, and defects in root zonation and vasculature differentiation (McConnell et al., 2001; Grigg et al., 2009; Dello Ioio et al., 2012; Carlsbecker et al., 2010). In the Arabidopsis root, the SHR transcription factor regulates MIR165/6 expression, restricting in a dose-dependent fashion HD-ZIPIII gene expression to the stele: this generates a radial gradient of HD-ZIPIII gene expression in the root vasculature, which patterns the stele (Carlsbecker et al., 2010). It was also posited that maintenance of the HD-ZIPIII expression gradient is necessary to restrict the Arabidopsis GT to one cortex and one endodermis, as a miR165/6-resistant version of PHB promotes the formation of extra GT layers (Miyashima et al., 2011). Although these mechanisms limiting GT development to one cortex and one endodermis layer are fairly well-characterized in Arabidopsis, how differences in cortical layer number among species are generated is still uncertain. One possibility is that SHR might control layer number in multiple-cortex species, such as Oryza sativa or Brachypodium distachyon, as the expression of the SHR locus of these species generates multiple cortical layers in Arabidopsis (Wu et al., 2014). However, whether this and/or other pathways control the variability of cortical layer number between species is still not established, partly owing to the lack of a convenient multiple-cortex root model system related to Arabidopsis.
To address this question, we utilized Cardamine hirsuta, a genetically tractable close relative of Arabidopsis and a well-established model plant for comparative development studies of shoot organs (Hay and Tsiantis, 2006; Vlad et al., 2014; Rast-Somssich et al., 2015; Hofhuis et al., 2016; Vuolo et al., 2016). We demonstrate that Cardamine has two root cortical layers (outer and inner) and one endodermis layer that are pre-patterned during embryogenesis. We show that the inner root cortical layer in Cardamine originates from a tissue with mixed cortical/endodermal identity that we called CEM. Through a comparative analysis of root development, we provide evidence that a differential spatial distribution of the transcription factor PHB and of miR165/6 are responsible for the cortex anatomical differences between Arabidopsis and Cardamine regulating CEM activity.
RESULTS
Cardamine has two cortical layers patterned during embryogenesis
It was previously shown that Cardamine has an additional GT layer compared with Arabidopsis (Fig. 1A,B) (Hay et al., 2014).
Cardamine has two cortical and one endodermis layers patterned during embryogenesis. (A) Diagram of Arabidopsis root meristem. The CEI/CEID is shown in green, the endodermis in cyan, the cortex in yellow. (B) Diagram of Cardamine root meristem. The CEI is shown in green, the CEM in orange, the endodermis in cyan, the cortex in yellow. (C) Confocal image of an Arabidopsis CO2::NLS3xVENUS meristem at 4 days post-germination (dpg) (n=11). (D) Magnification of the boxed region in C. (E) Confocal image of an Arabidopsis EN7::NLS3xVENUS meristem at 4 dpg (n=8). (F) Magnification of the boxed region in E. (G) Confocal image of a Cardamine CO2::NLS3xVENUS meristem at 4 dpg (n=25). (H) Magnification of the boxed region in G. (I) Confocal image of a Cardamine EN7::NLS3xVENUS meristem at 4 dpg (n=15). (J) Magnification of the boxed region in I. Propidium iodide is in red, VENUS in green. White asterisks indicate the CEM, white arrowheads indicate the first cells without fluorescence, blue arrowheads indicate the CEI, yellow arrowhead indicates the CEID. Scale bars: 50 µm (C,E,G,I); 20 µm (D,F,H,J).
Cardamine has two cortical and one endodermis layers patterned during embryogenesis. (A) Diagram of Arabidopsis root meristem. The CEI/CEID is shown in green, the endodermis in cyan, the cortex in yellow. (B) Diagram of Cardamine root meristem. The CEI is shown in green, the CEM in orange, the endodermis in cyan, the cortex in yellow. (C) Confocal image of an Arabidopsis CO2::NLS3xVENUS meristem at 4 days post-germination (dpg) (n=11). (D) Magnification of the boxed region in C. (E) Confocal image of an Arabidopsis EN7::NLS3xVENUS meristem at 4 dpg (n=8). (F) Magnification of the boxed region in E. (G) Confocal image of a Cardamine CO2::NLS3xVENUS meristem at 4 dpg (n=25). (H) Magnification of the boxed region in G. (I) Confocal image of a Cardamine EN7::NLS3xVENUS meristem at 4 dpg (n=15). (J) Magnification of the boxed region in I. Propidium iodide is in red, VENUS in green. White asterisks indicate the CEM, white arrowheads indicate the first cells without fluorescence, blue arrowheads indicate the CEI, yellow arrowhead indicates the CEID. Scale bars: 50 µm (C,E,G,I); 20 µm (D,F,H,J).
To determine the identity of this additional GT layer, we utilized fluorescent constructs of two GT markers: CORTEX2 (CO2), which is specific for the cortical layer, and ENDODERMIS7 (EN7), which is expressed only in the endodermis (Heidstra et al., 2004). As expected, in Arabidopsis CO2::NLS3xVENUS had a maximum expression in the cortex and is excluded from the CEI, CEID and the quiescent centre (QC) (Fig. 1C,D). In Arabidopsis EN7::NLS3xVENUS plants, fluorescence is detectable in the inner GT layer including the CEI and is excluded from the QC (Fig. 1E,F). In Cardamine, CO2::NLS3xVENUS expression is detectable in both the outer and the middle layer of the GT and is excluded from CEI and QC, whereas in EN7::NLS3xVENUS plants we could detect fluorescence only in the inner layer of the GT and in the CEI but not in the QC, pointing to the presence of a single endodermis and two cortical layers (Fig. 1G-J). Consistent with these data, Cardamine roots present only one casparian strip as visualized by endodermis-specific suberin autofluorescence (Fig. S1A-F).
We also noticed that in the Cardamine post-embryonic root at a two- to three-cell distance from the QC, a GT cell divides periclinally generating the inner cortical layer (Fig. S1G,H). We analysed the identity of those cells and we observed that the GT cells preceding the division that originates the second cortical layer in Cardamine express reporters of both cortex and endodermis, suggesting that they have a mixed identity; thus, we called this domain CEM (cortex/endodermis mix) (Fig. 1B,H,J). The cells deriving from the second GT periclinal division show the fluorescence of both EN7::NLS3xVENUS and CO2::NLS3xVENUS constructs, probably because of the stability of the NLS-3xVENUS protein (Wysocka-Diller et al., 2000). At a two-cell distance from the periclinal division of the endodermis, the middle GT cells lose endodermis identity and maintain cortical identity (Fig. 1H,J).
To establish whether the Cardamine double cortical layer is patterned during embryogenesis or after its completion, we analysed GT embryonic development. Based on Arabidopsis, we considered as root meristem GT the portion of GT contained within the lateral root cap (Dolan et al., 1993; Scheres et al., 1994).
We observed that, like in Arabidopsis, in the first stages of Cardamine embryogenesis the root GT develops as a monolayer that forms one cortex and one endodermis at heart stage, and two cortical layers in the hypocotyl (Fig. 2A,B,G,H). However, unlike Arabidopsis, an additional periclinal division in the endodermis at late torpedo stage gives rise to a second cortical layer that is then maintained throughout the life of the plant (Fig. 2C-F,I,J). We also noticed that an additional periclinal division occurs in the hypocotyl of Cardamine embryos from the late torpedo stage onwards, which increases the number of cortical layers to three (Fig. 2C-F). Using high-resolution modified pseudo-Schiff-propidium iodide (mPS-PI) staining (Truernit et al., 2008), we observed that CEI division produces the outer cortical layer and inner CEM cells. The latter divides giving rise to the inner cortex and the endodermis, in a similar fashion to post-embryonic development (Fig. 2K,L).
Cardamine inner cortical layer is patterned during embryogenesis. (A-J) DIC images of Cardamine and Arabidopsis embryos at the reported developmental stages (A,C,E,G,I) and the respective magnifications of the boxed regions (B,D,F,H,J) (respectively, n=12, n=15, n=14, n=7, n=8). White arrowheads indicate the root meristem border, black asterisks indicate cortical layers, black arrowheads indicate the CEM division, red arrowheads indicate hypocotyl periclinal division. (K,L) mPS-PI images of a Cardamine embryo at the reported developmental stage (n=18) (K) and the magnification of the boxed region in K (L). Blue arrowheads indicate the CEI. White arrowheads indicate the CEM division. White asterisks indicate CEM. Scale bars: 50 µm (A,C,E,G,I,K); 20 µm (B,D,F,H,J,L).
Cardamine inner cortical layer is patterned during embryogenesis. (A-J) DIC images of Cardamine and Arabidopsis embryos at the reported developmental stages (A,C,E,G,I) and the respective magnifications of the boxed regions (B,D,F,H,J) (respectively, n=12, n=15, n=14, n=7, n=8). White arrowheads indicate the root meristem border, black asterisks indicate cortical layers, black arrowheads indicate the CEM division, red arrowheads indicate hypocotyl periclinal division. (K,L) mPS-PI images of a Cardamine embryo at the reported developmental stage (n=18) (K) and the magnification of the boxed region in K (L). Blue arrowheads indicate the CEI. White arrowheads indicate the CEM division. White asterisks indicate CEM. Scale bars: 50 µm (A,C,E,G,I,K); 20 µm (B,D,F,H,J,L).
PHB and PHV HD-ZIPIII transcription factors regulate cortex patterning in Arabidopsis
To unravel the genetic mechanisms driving the formation of the two cortical layers in Cardamine, we performed a comparative analysis with Arabidopsis. We noticed that among Arabidopsis mutants showing additional root layers, ago1 and hyl1 develop an extra GT layer during embryogenesis (Paquette and Benfey, 2005; Heo et al., 2011; Zhang et al., 2011; Miyashima et al., 2009), resembling Cardamine GT patterning. We examined the hyl1-1 mutant and an additional hypomorphic allele of ago1 (ago1-27) in Arabidopsis, which show an extra GT layer patterned during embryogenesis (Fig. 3A-D; Fig. S2A,C,E,G). Moreover, we found that mutation in HASTY, another member of the miRNA biogenesis machinery (Park et al., 2005), also generates an additional GT layer during embryogenesis (Fig. S2I,K). We characterized the identity of this additional layer in hyl1 mutants, as an example of this set of mutants, utilizing CO2::H2B:YFP and EN7::H2B:YFP (Heidstra et al., 2004), where CO2 and EN7 promoters drive the expression of the HISTONE2B (H2B) fused to yellow fluorescent protein (YFP). We could detect the fluorescent signal of the CO2::H2B:YFP in the outer GT layers and of the EN7::H2B:YFP in the inner GT layer of hyl1-1 mutants (Fig. S3), suggesting that variations in miRNA biogenesis results in formation of supernumerary cortical layers.
PHB ectopic activity outputs in additional cortical layers. (A-F) DIC images of root meristems of Wt (n=8) (A), hyl1-1 (n=18) (C), phb-13;phv-11 (n=6) (E) and hyl1-1;phb-13;phv-11 (n=14) (G) plants at 5 dpg and the respective magnifications of the boxed areas (B,D,F,H). Blue arrowheads indicate the CEI, black arrowheads the extra periclinal division, asterisks the cortical layer(s). (I,J) Confocal images of root meristems of CO2::H2B:YFP Wt (n=10) (I) and phb-1d (n=14) (J) plants at 5 dpg. (K,L) Transverse sections of the root meristem of CO2::H2B:YFP Wt (n=4) (K) and phb-1d (n=4) (L). (M,N) Confocal images of root meristems of EN7::H2B:YFP Wt (n=6) (M) and phb-1d (n=20) (N) plants at 5 dpg. (O) Magnification of the boxed region in N. (P,Q) Transverse sections of the root meristem of EN7::H2B:YFP Wt (n=4) (P) and phb-1d (n=3) (Q). White arrowheads indicate the extra periclinal division. Propidium iodide is in red, YFP in green. Scale bars: 50 µm (A,C,E,G,I,J,M,N); 20 µm (B,D,F,H,O).
PHB ectopic activity outputs in additional cortical layers. (A-F) DIC images of root meristems of Wt (n=8) (A), hyl1-1 (n=18) (C), phb-13;phv-11 (n=6) (E) and hyl1-1;phb-13;phv-11 (n=14) (G) plants at 5 dpg and the respective magnifications of the boxed areas (B,D,F,H). Blue arrowheads indicate the CEI, black arrowheads the extra periclinal division, asterisks the cortical layer(s). (I,J) Confocal images of root meristems of CO2::H2B:YFP Wt (n=10) (I) and phb-1d (n=14) (J) plants at 5 dpg. (K,L) Transverse sections of the root meristem of CO2::H2B:YFP Wt (n=4) (K) and phb-1d (n=4) (L). (M,N) Confocal images of root meristems of EN7::H2B:YFP Wt (n=6) (M) and phb-1d (n=20) (N) plants at 5 dpg. (O) Magnification of the boxed region in N. (P,Q) Transverse sections of the root meristem of EN7::H2B:YFP Wt (n=4) (P) and phb-1d (n=3) (Q). White arrowheads indicate the extra periclinal division. Propidium iodide is in red, YFP in green. Scale bars: 50 µm (A,C,E,G,I,J,M,N); 20 µm (B,D,F,H,O).
The miR165/6-insensitive mutants of the redundantly acting PHB and PHV genes, phb-1d and phv-1d, express these genes also in the GT and show an extra GT layer (Miyashima et al., 2011). Nevertheless, whether the broader expression of PHB and PHV generates the additional GT layer in miRNA biogenesis loss-of-function mutants is not known. To assess this, we analysed the activity of the XPHB sensor of miR165/6, a constitutively expressed green fluorescent protein (GFP) sensitive to miR165/6 (Dello Ioio et al., 2012) in the hyl1-1 background. Unlike in wild-type (Wt) plants, we detected GFP expression in the GT of hyl1-1, suggesting a peripheral broadening of the PHB and PHV gradient (Fig. S4). Consistent with this evidence, the triple loss-of-function mutants hyl1-1;phb-13;phv-11, hst-1;phb-13;phv-11 and ago1-27;phb-13;phv-11 display only one cortex and one endodermis layer already from embryogenesis in a similar fashion to Wt and phb-13;phv-11 mutant plants (Fig. 3A-H; Fig. S2) suggesting that ectopic expression of PHB and PHV is required to generate the extra-cortical layer in hyl1-1, hst-1 and ago1-27 Arabidopsis backgrounds.
To characterize the identity of the additional GT layer of phb-1d and phv-1d mutants, we analysed the reporter activity of CO2::H2B:YFP and EN7::H2B:YFP in those backgrounds. In both genotypes, the CO2::H2B:YFP signal was consistently detected in the additional GT layer whereas the EN7::H2B:YFP was randomly expressed in only a subset of these cells, suggesting that this additional GT layer has cortex identity (Fig. 3I-Q; data not shown). We concluded that the expansion of PHB domain in the GT generates an additional cortical layer.
HD-ZIPIII transcription factors control cortex patterning in Cardamine
Because of the capability of the PHB and PHV HD-ZIPIII transcription factors to control cortical layer number in Arabidopsis, we hypothesized that these factors might be responsible for the generation of the double cortical layer in Cardamine. We identified five orthologues of the Arabidopsis HD-ZIPIII genes in the genome of Cardamine (Gan et al., 2016), which we named ChPHB, ChPHV, ChREV, ChHB8, ChCNA (Fig. S5A). Sequence alignment showed that the ChHD-ZIPIII genes are all more than 90% similar to Arabidopsis and that all of them contain a miR165/6 recognition site domain (Fig. S5). We also found that the Arabidopsis nine miR165/6 loci are conserved in the Cardamine genome and their alignment showed that the processed miR165 and miR166 sequences are 100% homologous to the Arabidopsis counterpart (Fig. S5B). To assess whether the HD-ZIPIII proteins control double cortex formation in Cardamine, we knocked down the HD-ZIPIII genes by means of the expression of the precursor of miR165, pri-MIR165A, driven by the 35S promoter (35S::MIR165A). Out of 16 independent 35S::MIR165A Cardamine lines, 14 showed one cortex and lack of CEM in the T1 generation suggesting that HD-ZIPIII transcription factors are necessary for double cortex formation in Cardamine (Fig. 4A-C).
PHB and miR165/6 have different activity domains between Cardamine and Arabidopsis. (A,B) Confocal images of root meristem Wt (n=5) (A) and 35S::miR165A (n=16) (B) Cardamine plants at 5 dpg. White asterisks indicate the CEM, white arrowhead indicates the second periclinal division. (C) Magnification of the boxed region in B. (D) Confocal image of 2 dpg Wt Cardamine root harbouring a ChPHB:GFP construct (n=8). (E) Magnification of the boxed region in D. (F) Confocal image of 3 dpg Wt Cardamine root harbouring a ChUB10::XPHB construct (n=15). (G) Confocal image of a 2 dpg ChUB10::GFP Cardamine plant (n=10). ChUB10::GFP plants are used as control of ChUB10 expression domain. White arrows indicate the endodermis, blue arrowheads the CEI and the CEM. (H) Graph depicting the green fluorescence ratio between three cells of the CEM and three cells of endodermis in ChUB10::XPHB and in ChUB10::GFP plants. Each fluorescence value was normalized to the corresponding cell area. n=6. Error bars represent s.d. (I) Confocal image of 2 dpg Wt Arabidopsis root harbouring a AtPHB:GFP construct (n=5). (J) Confocal image of 5 dpg Wt Arabidopsis root harbouring a XPHB construct (n=10). (K) Diagram of the pattern domain of PHB (pink) and miR165/6 (green) in Arabidopsis. (L) Diagram of the distribution of PHB (pink) and miR165/6 (green) in Cardamine. White arrows indicate the endodermis. Scale bars: 50 µm (A,B,D,F,G); 20 µm (C,E). GFP signal is in green, propidium iodide signal is in red. C, cortex; E, endodermis.
PHB and miR165/6 have different activity domains between Cardamine and Arabidopsis. (A,B) Confocal images of root meristem Wt (n=5) (A) and 35S::miR165A (n=16) (B) Cardamine plants at 5 dpg. White asterisks indicate the CEM, white arrowhead indicates the second periclinal division. (C) Magnification of the boxed region in B. (D) Confocal image of 2 dpg Wt Cardamine root harbouring a ChPHB:GFP construct (n=8). (E) Magnification of the boxed region in D. (F) Confocal image of 3 dpg Wt Cardamine root harbouring a ChUB10::XPHB construct (n=15). (G) Confocal image of a 2 dpg ChUB10::GFP Cardamine plant (n=10). ChUB10::GFP plants are used as control of ChUB10 expression domain. White arrows indicate the endodermis, blue arrowheads the CEI and the CEM. (H) Graph depicting the green fluorescence ratio between three cells of the CEM and three cells of endodermis in ChUB10::XPHB and in ChUB10::GFP plants. Each fluorescence value was normalized to the corresponding cell area. n=6. Error bars represent s.d. (I) Confocal image of 2 dpg Wt Arabidopsis root harbouring a AtPHB:GFP construct (n=5). (J) Confocal image of 5 dpg Wt Arabidopsis root harbouring a XPHB construct (n=10). (K) Diagram of the pattern domain of PHB (pink) and miR165/6 (green) in Arabidopsis. (L) Diagram of the distribution of PHB (pink) and miR165/6 (green) in Cardamine. White arrows indicate the endodermis. Scale bars: 50 µm (A,B,D,F,G); 20 µm (C,E). GFP signal is in green, propidium iodide signal is in red. C, cortex; E, endodermis.
Previous reports have shown, via a GFP translational fusion (AtPHB:GFP), that in Arabidopsis PHB expression is restricted to the stele (Fig. 4I,K; Fig. S7A) (Carlsbecker et al., 2010; Miyashima et al., 2011; Dello Ioio et al., 2012). To establish whether HD-ZIPIII proteins present a different spatial distribution between Arabidopsis and Cardamine, we analysed the expression of a translational fusion ChPHB:GFP in Cardamine root, focusing on PHB because its ectopic expression is sufficient to induce the formation of an additional cortical layer in Arabidopsis (Fig. 3). We observed that ChPHB:GFP fluorescence is present in the vasculature, QC, CEI/CEID and in the CEM of Cardamine root (Fig. 4D,E), whereas in Arabidopsis root AtPHB:GFP is localized only in the vasculature (Fig. S7A) (Carlsbecker et al., 2010; Miyashima et al., 2011; Dello Ioio et al., 2012). These results suggest that the presence of the PHB transcription factor in the GT of Cardamine is responsible for the additional cortical layer (Fig. 4I,K).
miRNA165/6 distribution generates diversity in cortical layer number
We have suggested that in Cardamine the presence of PHB in the GT is responsible for the double cortical layer. Because in Arabidopsis it is the activity of miR165/6 in the GT that restricts PHB expression to the stele (Fig. 4I,K) (Carlsbecker et al., 2010; Miyashima et al., 2011), we investigated the hypothesis that the lack of miR165/6 in the Cardamine GT allows the broadening of the PHB domain. Consistent with this hypothesis, we observed that decreasing miR165/6 activity in the Arabidopsis GT, by expressing a DNA fragment mimicking the miR165/6 target sequence (35S::MIM165/6) (Todesco et al., 2010), produces a broader XPHB signal and an extra cortical layer (Fig. S6). This suggests that the expression of PHB in this background is no longer restricted to the stele and links this phenomenon to the formation of a second cortex. To establish whether miR165/6 activity is indeed absent from Cardamine GT, we generated Cardamine plants harbouring the XPHB miR165/6-sensitive GFP protein under the control of the Cardamine constitutive promoter ChUbiquitin10 (ChUB10::XPHB). We could detect GFP signal in the QC, CEI, CEID and CEM, suggesting that activity of miR165/6 is low in these cells (Fig. 4F-H). By contrast, we detected low GFP fluorescence in the endodermis, indicating that miR165/6 is active in those cells (Fig. 4F-H). Hence, the broader expression domain of PHB in Cardamine compared with Arabidopsis might be dependent on the different miR165/6 activity domain in the two species (Fig. 4K,L). To verify that the presence of the PHB protein in the Cardamine GT is actually due to the absence of miR165/6 in the endodermis and not to an intrinsic property of the Cardamine PHB protein, we analysed the expression of the ChPHB:GFP construct in Arabidopsis. In these plants, the distribution of the ChPHB:GFP protein mirrors that of AtPHB:GFP (Fig. S7A,B), suggesting that the presence of miRNA166 in the Arabidopsis GT is sufficient to restrict PHB in the stele. Conversely, expression of the miR165/6-insensitive form of PHB (AtPHB*:GFP) in Arabidopsis is expanded in the endodermis and is accompanied by the formation of extra GT layers (Fig. S7C-E).
These results suggest that the absence of miR165/6 in the Cardamine root GT allows expression of PHB in the GT, which results in the development of an extra cortical layer.
PHB controls GT patterning via CYCD6;1
It was recently reported that Arabidopsis plants expressing the rice and Brachypodium SHR locus show extra cortical layers (Wu et al., 2014). To understand whether Cardamine SHR also generates additional cortical layers in Arabidopsis, we created a construct in which ChSHR is fused to cyano fluorescent protein (CFP) (ChSHR:CFP). We observed that Arabidopsis plants harbouring ChSHR:CFP display an additional GT layer and that ChSHR:CFP is sufficient to both recover the shr-1 phenotype (55%) or induce additional GT layers in the latter background (45%) (Fig. S8A-C), suggesting a conserved function of SHR in determining cortical layer number in different species.
To assess possible links between the SHR and PHB pathways in GT patterning, we generated ChSHR:CFP;phb-13;phv-11 Arabidopsis lines. Loss of PHB and PHV activity was not sufficient to suppress the ChSHR:CFP phenotype completely (Fig. S8A-D), indicating that PHB and PHV are not required for the SHR GT-patterning action.
To determine whether PHB requires SHR to induce extra cortex formation, we generated double mutants shr-1;phb-1d. As visualized by the activity of the cortex marker CO2::H2B:YFP, these plants show a double cortical layer, similar to phb-1d single mutants (Fig. S8H-M), suggesting that the effect of PHB on GT development does not require SHR. These results suggest that the SHR and PHB pathways act in parallel in controlling cortical layer number.
The cyclin CYCD6;1 controls the periclinal division of the CEID that gives rise to cortex and endodermis (Sozzani et al., 2010). To assess whether, in analogy with SHR/SCR, PHB acts on the CYCD6;1 gene in the formation of extra cortical layers, we analysed CYCD6;1 expression in the phb-1d mutant, which expresses PHB also in the GT. qRT-PCR analysis showed an increased level of CYCD6;1 mRNA in phb-1d roots (Fig. 5A), suggesting that the expression of CYCD6;1 is regulated by PHB. To understand whether CYCD6;1 induction by PHB coincides with second cortex formation, we analysed a fluorescent transcriptional reporter of CYCD6;1 (CYCD6;1::GFP:GUS) in the phb-1d background. Whereas in Wt roots CYCD6;1::GFP:GUS has a maximum fluorescent signal in the CEI/CEID, in phb-1d we could detect high CYCD6;1 activity also at the additional division site that generates the second cortex formation in phb-1d background (Fig. 5B,C). This indicates that, although independent of one another, the PHB and SHR/SCR pathways both act on CYCD6;1 expression in patterning the GT.
PHB controls CYCD6;1 expression in Arabidopsis. (A) qRT-PCR of CYCD6;1 in Wt and phb-1d roots. n=2, ***P<0.01. Error bars represent s.d. (B,C) Confocal images of root meristem of Wt (n=8) (B) and phb-1d mutant (n=9) (C) plants at 5 dpg, carrying the CYCD6;1::GFP:GUS construct. Propidium iodide is in red, GFP is in green. White arrowhead indicates the additional periclinal division in the GT, blue arrowheads indicate the CEI. (D) Schematic of the proposed model. In Arabidopsis, the activity of SHR and SCR in the CEI (green) is necessary and sufficient to induce CYCD6;1 expression, promoting the periclinal division (dashed green line) that gives rise to cortex (yellow) and endodermis (cyan). In the CEI/CEID and endodermis, SHR induces miR165/6, which restricts PHB expression to the stele. In Cardamine, the activity of PHB in the CEI and in the CEM tissue (orange) stimulates CYCD6;1 expression in these cells, promoting an additional periclinal division (dashed green line). This additional periclinal division produces the extra cortex layer (yellow half). Subsequently, miR165/6 is active in the endodermis (cyan), restricting PHB to the stele.
PHB controls CYCD6;1 expression in Arabidopsis. (A) qRT-PCR of CYCD6;1 in Wt and phb-1d roots. n=2, ***P<0.01. Error bars represent s.d. (B,C) Confocal images of root meristem of Wt (n=8) (B) and phb-1d mutant (n=9) (C) plants at 5 dpg, carrying the CYCD6;1::GFP:GUS construct. Propidium iodide is in red, GFP is in green. White arrowhead indicates the additional periclinal division in the GT, blue arrowheads indicate the CEI. (D) Schematic of the proposed model. In Arabidopsis, the activity of SHR and SCR in the CEI (green) is necessary and sufficient to induce CYCD6;1 expression, promoting the periclinal division (dashed green line) that gives rise to cortex (yellow) and endodermis (cyan). In the CEI/CEID and endodermis, SHR induces miR165/6, which restricts PHB expression to the stele. In Cardamine, the activity of PHB in the CEI and in the CEM tissue (orange) stimulates CYCD6;1 expression in these cells, promoting an additional periclinal division (dashed green line). This additional periclinal division produces the extra cortex layer (yellow half). Subsequently, miR165/6 is active in the endodermis (cyan), restricting PHB to the stele.
In Arabidopsis, SHR and SCR control CYCD6;1 expression in the GT (Sozzani et al., 2010; Cruz-Ramírez et al., 2012). To assess whether PHB regulates CYCD6;1 expression independently of the SHR/SCR pathway, we measured CYCD6;1 mRNA levels in phb-1d; shr-1 background by qRT PCR. In the shr background CYCD6;1 expression is almost undetectable (Koizumi et al., 2012). The phb-1d mutation is sufficient to increase CYCD6;1 mRNA levels in the phb-1d;shr-1 background (Fig. S8O), suggesting that PHB controls CYCD6;1 expression independently of SHR. A 4 h induction of a dexamethasone (DEX)-inducible miRNA-resistant version of PHB (35S:GR≫PHB*) results in a mild but significant increase of CYCD6;1 mRNA level in this background (Fig. S8P), suggesting that CYCD6;1 is an early, but most likely not direct, target of PHB.
DISCUSSION
We showed that the Cardamine double cortical layer originates from a developmental domain of mixed cortical and endodermal identity, the CEM, established during embryogenesis. Our results suggest that the presence of PHB in the CEM is necessary and sufficient to generate an additional cortical layer in the Cardamine GT. In the CEM, PHB might regulate the expression of CYCD6;1 triggering an additional periclinal division in the GT (Fig. 5D). Whereas in Arabidopsis, miR165/6 activity in QC, CEI/CEID, endodermis and cortex clears out PHB mRNA from the GT, in Cardamine the PHB protein is present in the GT owing to the absence of miR165/6 activity in those cells (Fig. 5D). Thus, the difference in GT patterning between Arabidopsis and Cardamine seems to result from a difference in the localization of miR165/6, which generates a different distribution of PHB in the two species. Our data also suggest that in Cardamine post-embryonic root development the presence of miR165/6 activity in differentiated endodermis eliminates PHB from this tissue, thus suppressing the formation of additional periclinal divisions and maintaining the number of cortical layers.
Our results highlight the role of HD-ZIPIII in the control of cortical layer number. Further studies on multi-cortical layer model species will clarify whether this mechanism holds true in general for plant root cortical variability.
In Arabidopsis embryos, miR165A expression is independent of SHR (Miyashima et al., 2013). In a similar fashion, we posit that the SHR/SCR regulatory circuit might not activate miR165/6 expression in the Cardamine QC, CEI and CEM (Fig. 5D). Consistent with this hypothesis, ChSCR, a conserved SHR target (Cui et al., 2007), is expressed in tissues in which miR165/6 activity is not detectable, such as CEI and CEM, as visualized by a ChSCR regulative sequence fused to GFP targeted to the endoplasmic reticulum, ERGFP (Fig. S8N; ChSCR::ERGFP). Hence, miR165/6 expression might be SHR dependent only in the endodermis where we detect miR165/6 activity but not in the CEI/CEID, CEM and QC. Analysis of the expression of the four miR165/6 and cross-species experiments between Arabidopsis and Cardamine will help to clarify the role of SHR in regulating the expression of these miRNAs in Cardamine GT. We provided evidence in Arabidopsis that PHB regulates additional cortical layers via the control CYCD6;1 expression independently of SHR/SCR. Our results do not rule out the possibility that the PHB and SHR/SCR circuits may interconnect downstream, as a reduction in CYCD6;1 mRNA level is still detectable in shr,phb-1d background compared with phb-1d single mutants (Fig. 5A; Fig. S8O). It was recently shown that the SHR protein is detectable in rice root cortical layers, and that overexpression of Arabidopsis SHR increases the number of cortical layers in rice (Henry et al., 2017). Our results raise the possibility that PHB controls the expression of CYCD6;1 via induction of a third as-yet-unidentified component, as short induction of PHB increases CYCD6;1 expression only slightly. Further studies in Arabidopsis and Cardamine will be required for a better understanding of the connection between these two pathways in GT patterning.
It remains to be established whether, through the control of CYCD6;1, the MIR165/6/PHB circuit generates the CEM or regulates its activity, and additional work is required to elucidate fully the genetic network leading to double cortex formation in Cardamine. However, our results indicate that variations in miRNA distribution are capable of determining differences in plant anatomy.
MATERIALS AND METHODS
Growth conditions
Arabidopsis and Cardamine seeds were surface sterilized using a solution containing 70% ethanol and 0.5% Triton X-100 (Carlo Erba) for 10 min and then with a solution of 96% ethanol for 10 min. Seeds were air dried and suspended in 0.1% agarose in water. Arabidopsis seeds were plated after 3 days of cold treatment, Cardamine seeds after 5 days. Seeds were grown in a vertical position, at 22°C in 16 h light/8 h dark cycle, on ½ MS (Murashige and Skoog medium, Duchefa) supplemented with 1% sucrose, as described by Perilli and Sabatini (2010).
Plant material
All Arabidopsis material is in Columbia ecotype, all Cardamine lines are in Oxford ecotype. CO2::H2B:YFP and EN7::H2B:YFP were kindly donated by R. Heidstra (Heidstra et al., 2004). CYCD6;1::GFP:GUS line was kindly given by R. Sozzani (Sozzani et al., 2010). phb-1d/+, phv-1d/+ (used in this study), XPHB, XmPHB, PHB*:GFP, PHB:GFP, 35S:GR≫PHB* were previously described by Dello Ioio et al. (2012), phb-13;phv-11 by Prigge et al. (2005), ago1-27 by Morel et al. (2002), hyl1-1 by Lu and Fedoroff (2000), hst1-1 by Telfer and Poethig (1998) and shr-1 by Scheres et al. (1995). Primers used for genotyping are listed in Table S1. Approximately 20 plants for genotype were analysed.
Transgenic lines generation
CO2::3xNLSVENUS and EN7::3xNLSVENUS were constructed by excising the promoters from CO2::H2B:YFP and EN7::H2B:YFP plasmids utilizing XhoI and BamHI sites. The CO2 and EN7 promoter were subsequently inserted in a 3xNLSVENUS-PBJ36 plasmid. CO2::NLS3xVENUS and EN7::NLS3xVENUS cassettes were inserted in a pMLBART plasmid utilizing NotI sites.
ChPHB:GFP and ChSHR:CFP constructs were generated utilizing the Gateway system (Invitrogen). The 3 kb region upstream of the ChPHB ATG and the ChPHB genomic sequence without the stop codon were amplified from Cardamine and cloned respectively in pDONR_P4_P1R and pDONR221 Gateway vector by BP recombination (Invitrogen). Subsequently, pDONRP4_P1R-ChPHBp and pDONR221-ChPHB were recombined with pDONR P2R_P3-C-term GFP into a pB7m34GW destination vector via LR reaction (Invitrogen; Karimi et al., 2002). The 2.9 kb ChSHR region upstream of the ChSHR ATG and the ChSHR genomic sequence without stop codon were amplified from Cardamine and cloned respectively in a pDONR_P4_P1R and a pDONR221 Gateway vector by BP recombination (Invitrogen). Subsequently, pDONRP4_P1R-ChSHRp and pDONR221-ChSHR were recombined with pDONR P2R_P3-C-term CFP into a pB7m34GW destination vector via LR reaction (Invitrogen; Karimi et al., 2002).
The ChSCR::GFP was obtained as follows. The 3 kb region upstream of the ChSCR ATG was amplified from Cardamine DNA and cloned in pDONR_P4_P1R Gateway vector by BP recombination (Invitrogen). pDONRP4_P1-ChSCRp and pDONR221-ERGFP were recombined with pDONR P2R_P3-NOS into a pB7m34GW destination vector via LR reaction (Invitrogen; Karimi et al., 2002).
35S::MIM165/6 plasmid was obtained from the Nottingham Arabidopsis Stock Center (NASC, N783233). 35S::primiR165A was provided by M. Tsiantis (Max Planck Institute for Plant Breeding Research, Cologne, Germany). Plasmids were introduced into a GV3101 Agrobacterium tumefaciens strain and plants were transformed by floral dip (Clough and Bent, 1998). Primers used for plasmid construction are listed in Table S2. Number of transformants obtained are listed in Table S3.
The ChUB10::GFP was obtained as follows. The 500 bp region upstream of the ChUB10 ATG amplified from Cardamine DNA and cloned in pDONR_P4_P1R Gateway vector by BP recombination (Invitrogen). pDONRP4_P1-ChUB10p and pDONR221-GFP were recombined with pDONR P2R_P3-NOS into a pB7m34GW destination vector via LR reaction (Invitrogen; Karimi et al., 2002).
The ChUB10::XPHB was obtained as follows. The XPHB sequence was amplified from XPHB-PGII plasmid and cloned in pDONR221 Gateway vector by BP recombination (Invitrogen). Subsequently, pDONRP4_P1R-ChUB10p and pDONR221-XPHB were recombined with pDONR P2R_P3-NOS into a pB7m34GW destination vector via LR reaction (Invitrogen; Karimi et al., 2002).
Microscopy
Confocal images of the median longitudinal sections of the root meristem were taken using a Zeiss LSM 780 microscope. A 10 μg ml−1 propidium iodide (Sigma) solution was used to mount the samples and visualize the cell wall.
Differential interference contrast (DIC) with Nomarski technology microscopy (Zeiss Axio Imager A2) was used to image root meristems. Plants were mounted in a chloral hydrate solution (8:3:1 mixture of chloral hydrate:water:glycerol). CEM number is considered as the number of cells from the CEI to the second periclinal division exclusive. Fluorescent signal was measured using Fiji-Image J as described by Moubayidin et al. (2010).
Histology
Casparian strip autofluorescence was analysed as described by Long et al. (2015a).
mPS-PI staining
Ovules were isolated from developing siliques and opened to extract embryos. These were fixed in fixative solution (50% methanol and 10% acetic acid) into a multiwell plate at 4°C overnight. The modified pseudo-Schiff-propidium iodide (mPS-PI) method was performed as described by Truernit et al. (2008).
Real-time PCR
Total RNA was extracted from 5 days post-germination roots using the NucleoSpin RNA Plus (Machery-Nagel), and the first strand cDNA was synthesized using the Superscript III First Strand Synthesis System (Invitrogen). Quantitative RT-PCR analysis was conducted using the following gene-specific primers: CYCD6;1_RT_F 5′ and CYCD6;1_RT_R 5′ (Sozzani et al., 2010) and qRTPHBfw and qRTPHBrev for PHB (Carlsbecker et al., 2010). Relative expression was normalized to ORNITHINE TRANSCARBAMILASE (OTC) control (Dello Ioio et al., 2012).
PCR amplification was carried out in the presence of the double-strand DNA-specific dye SsoAdvanced Universal SYBR Green Supermix (Bio-Rad). Amplification was monitored in real time with the 7500 Fast Real Time PCR System (Applied Biosystems). Experiments were performed in triplicate from two independent root tissue RNA extractions. Data are expressed as 2−ΔΔCt value. Student's t-test was used to determine statistical significance of these data (http://graphpad.com/quickcalcs/ttest2.cfm)
Sequences analysis
CDS and protein identity percentage was analysed using ClustalW (www.ebi.ac.uk/Tools/msa/clustalw2/).
Acknowledgements
We acknowledge R. Di Mambro, C. Galinha, M. Cartolano, R. Sozzani, A. Hasson, M. Del Bianco, E. Salvi and E. Pierdonati for valuable discussions on the manuscript; R. Sozzani and Janne Lempe for kindly providing material; and L. Giustini and M. C. Giorgi for technical assistance.
Footnotes
Author contributions
Conceptualization: M.T., S.S., R.D.I.; Methodology: G.D.R., G.B., R.D.I.; Validation: G.D.R., G.B., E.P., L.P., R.D.I.; Formal analysis: G.D.R., G.B., E.P., L.P., R.D.I.; Investigation: G.D.R., G.B., E.P., L.P., R.D.I.; Resources: M.T., S.S., P.C., R.D.I.; Writing - original draft: P.C., R.D.I.; Writing - review & editing: E.P., M.T., P.C., R.D.I.; Visualization: G.D.R., G.B., R.D.I.; Supervision: R.D.I.; Project administration: R.D.I.; Funding acquisition: S.S., P.C., R.D.I.
Funding
This work was supported by a FIRB (Futuro in Ricerca 2013) project grant from the Ministero dell'Istruzione, dell'Università e della Ricerca (FIRB2013-RBFR13DCDS to G.D.R., G.B. and R.D.I.) and a European Research Council grant (260368 to E.P., L.P. and S.S.).
Data availability
Cardamine sequences are available on https://gbrowse.mpipz.mpg.de/cgi-bin/gbrowse/chi1_public/.
References
Competing interests
The authors declare no competing or financial interests.