ERECTA-family genes coordinate stem cell functions between the epidermal and internal layers of the shoot apical meristem

Aerial plant tissues are derived from a population of stem cells in the shoot apical meristem (SAM), which is located at the shoot tip (Gordon et al., 2009; Miwa et al., 2009; Yadav et al., 2010). The SAM consists of three tissue layers: the epidermal L1 layer (tunica) and the internal layers L2 and L3 (corpus). Although these different tissue layers play distinct roles during development, stem cells are spread between them in the central zone of the SAM (Meyerowitz, 1997). We currently have a poor understanding of the molecular mechanisms that regulate how stem cells in the different SAM tissue layers are coordinated to behave as one population.

WUSCHEL (WUS) is a transcription factor that promotes stem cell proliferation, and WUS is expressed in the SAM-organizing center (OC). The wus mutants fail to maintain stem cells in the SAM (Laux et al., 1996; Mayer et al., 1998). CLAVATA3 (CLV3) encodes a secreted peptide that suppresses WUS activity and is specifically expressed in the stem cells, thereby contributing to stem cell homeostasis (Brand et al., 2000; Schoof et al., 2000). Induction of the WUS expression is directly regulated by cytokinin signal transduction components, namely type-B ARABIDOPSIS RESPONSE REGULATOR proteins (ARRs) (Meng et al., 2017; Wang et al., 2017). Mathematical models have proposed that the cytokinin influences the WUS expression via cytokinin receptors expressed in the OC (Adibi et al., 2016; Chickarmane et al., 2012; Gordon et al., 2009; Gruel et al., 2016). Moreover, in turn, WUS promotes the cytokinin responsiveness of the SAM (Leibfried et al., 2005). However, the molecular mechanisms that underlie how the primary cytokinin response in the OC affects other SAM tissues remain unknown. In addition, in contrast to the well-characterized function of WUS in stem cell maintenance, we have a limited understanding of the role that WUS-independent mechanisms play in this process (Huang et al., 2015; Lee and Clark, 2015).

ERECTA (ER), ER-LIKE1 (ERL1) and ERL2 constitute the ER receptor kinase gene family (Shpak et al., 2004). All of these genes are expressed throughout the SAM and regulate its development (Bemis et al., 2013; Chen et al., 2013; Uchida et al., 2011, 2013). The er erl1 erl2 triple mutant exhibits an expanded SAM with an enlarged stem cell region (Chen et al., 2013; Uchida et al., 2013). Furthermore, loss of function of ER-family members sensitizes SAM cell proliferation to cytokinin (Uchida et al., 2013). However, it remains largely unknown how ER activity affects the CLV3-WUS and cytokinin-signaling pathways, which contribute to stem cell maintenance.

Here, we report that loss of function of ER family restores the SAM in wus mutants, as demonstrated via the epidermis-specific expression of a stem cell marker gene. This phenotype caused by loss of ER-family activity is suppressed by localized ER expression in the epidermis. Furthermore, the CLV3 and cytokinin signaling pathways are dysfunctional in a tissue layer-specific manner in ER-family mutants. This study demonstrates that the ER family is required for the coordination of stem cell behavior between different SAM tissue layers.

A previous study reported that the vegetative SAM is enlarged in er erl1 erl2 mutants compared to that in wild type (Uchida et al., 2013) (Fig. 1A,B,E,F; Fig. S1). Conversely, in wus mutants, the SAM is consumed soon after germination (Laux et al., 1996; Mayer et al., 1998; Fig. 1C,G; Fig. S1). To investigate the genetic interaction between ER family and WUS, and to address the relationship between their contrasting mutant phenotypes, we created a wus er erl1 erl2 quadruple mutant. In wus er erl1 erl2, the SAM was maintained despite WUS loss of function (Fig. 1D,H; Fig. S1), and small cells characteristic of the wild-type SAM (Fig. 1E) were observed covering the surface of the wus er erl1 erl2 SAM (Fig. 1H). Furthermore, the wus er erl1 erl2 SAM persisted during the reproductive growth stage (Fig. S2). Although the SAM was consumed in wus (Fig. 1C,G), these wus plants occasionally produced adventitious meristems in later growth stages and formed adventitious inflorescences (Laux et al., 1996; Mayer et al., 1998). Accordingly, the emergence of the inflorescence stems in wus was severely delayed compared with that in wild type (Laux et al., 1996; Fig. S2A,B). By contrast, wus er erl1 erl2 plants maintained the primary meristem (Fig. 1D,H), and bolted normally to form the primary inflorescence (Fig. S2A). Although the wus inflorescence developed few flowers (Fig. S2F,J), the wus er erl1 erl2 inflorescence continuously produced flowers, comparable with that in wild type and er erl1 erl2 mutants (Fig. S2C-E,G-I). Collectively, these observations suggest that ER-family loss of function largely alleviates the defects in both vegetative and inflorescence SAMs in wus. Furthermore, in contrast to the consistent lack of pistils in wus flowers (Laux et al., 1996), wus er erl1 erl2 flowers formed pistils (Table S1). The number of stamens was also increased in wus er erl1 erl2 compared with that in wus, whereas the numbers of sepals and petals were not recovered, suggesting that the complementation of wus flower formation imparted by ER-family lack of function was limited to the inner whorls.

To monitor cell proliferation in the SAM of wus er erl1 erl2 mutants, we performed 5-ethynyl-2′-deoxyuridine (EdU) labeling. EdU incorporates into newly synthesized DNA to label actively dividing cells. In wild type and er erl1 erl2, EdU-labeled nuclei were detected across multiple SAM tissue layers (Fig. 2A,B). However, in wus, few EdU-labeled nuclei were detected at the center of the shoot apex where the SAM was consumed (Fig. 2C,E). This result is consistent with the observation that the small cells characteristic of SAMs in the wild-type and er erl1 erl2 plants (Fig. 1E,F) were not apparent in wus (Fig. 1G). Interestingly, the wus er erl1 erl2 SAM exhibited EdU-labeled nuclei similar to the er erl1 erl2 SAM (Fig. 2B,D,E), demonstrating that cell proliferation is active in the wus er erl1 erl2 SAM despite the loss of function of WUS. When the number of EdU-labeled nuclei was normalized to the SAM area, the normalized values were very similar among wild type, er erl1 erl2 and wus er erl1 erl2 (Fig. S3), indicating that cells in the SAM proliferate at a similar rate in these plants.

CLV3pro:GUS is a known and reliable stem cell marker; however, GUS signal is not observed in the SAM of wus plants (Brand et al., 2002; Fig. 3A,B). Conversely, CLV3pro:GUS expression produced a detectable GUS signal at the periphery (Fig. 3C, arrow) and in the central region of the SAM in er erl1 erl2, which likely reflects SAM expansion in the absence of ER-family activity. Furthermore, a strong GUS signal was detected in the SAM epidermal layer in wus er erl1 erl2 plants (Fig. 3D). In situ RNA hybridization experiments also showed the epidermal expression of endogenous CLV3 in the mutants (Fig. 3E,F), indicating the consistency between the extended epidermal CLV3pro:GUS signal and the CLV3 transcript accumulation. These results demonstrate that ER-family loss of function allows expression of the stem cell marker in the SAM epidermis even in the absence of WUS activity (Fig. 3D); however, WUS is required for CLV3pro:GUS expression in the full complement of SAM tissue layers, as is observed in the wild type (Fig. 3A) and not in wus er erl1 erl2 (Fig. 3D). Thus, in the absence of ER-family activity, the dependence of stem cell marker expression on WUS is decoupled between the epidermis and the internal tissue. Recovery of stem cells in wus er erl1 erl2 was also confirmed using qRT-PCR analysis, which revealed strong expression of endogenous CLV3 in the quadruple mutant (Fig. 3G). SHOOT MERISTEMLESS (STM) expression levels (Long et al., 1996) were also examined as a marker gene indicative of undifferentiated cells. We observed high-level STM expression in wus er erl1 erl2 plants (Fig. 3H), which is consistent with the enlarged SAM observed in these plants (Fig. 1D,H).

Expression of CLV3pro:GUS produced a GUS signal in the epidermal cells at the periphery of the SAM in er erl1 erl2 mutants (Fig. 3C, arrow), and the CLV3pro:GUS expression in the epidermis was still maintained in wus er erl1 erl2 (Fig. 3D). These results suggest that, although ER is expressed throughout the SAM (Uchida et al., 2013; Fig. S4A), the ER activity in the epidermis may directly function to regulate the stem-cell-marker signals in the epidermal cells. To investigate this potential role further, we performed complementation experiments by expressing ER in er erl1 erl2 and wus er erl1 erl2 under the control of the epidermis-specific AtML1 promoter (Sessions et al., 1999; Fig. 4A). Compared with the broad CLV3pro:GUS expression pattern observed in er erl1 erl2 (Fig. 3C), we observed no GUS signal in the epidermal cells at the periphery of the SAM in er erl1 erl2 AtML1pro:ER plants (Fig. 4B), and CLV3pro:GUS expression in these complemented plants was restricted to a compact region at the center of the normally sized SAM (Fig. 4B) as seen in the wild-type plants (Fig. 3A). On the other hand, the expression of ER in er erl1 erl2 under the control of the OC-specific WUS promoter did not rescue either the CLV3pro:GUS misexpression at the SAM periphery or the enlarged SAM morphology (Fig. S4F), emphasizing the importance of the ER function in the epidermis to prevent the misexpression of the stem cell marker. Accordingly, the elevated expression of endogenous CLV3 observed in er erl1 erl2 was decreased in er erl1 erl2 AtML1pro:ER to a level comparable with that in wild type (Fig. 4D). In addition, AtML1pro:ER largely ameliorated dwarfism and small leaf phenotype of er erl1 erl2 seedlings (Fig. S4C,D), whereas WUSpro:ER did not affect the seedling phenotypes (Fig. S4E). However, the petiole length of er erl1 erl2 AtML1pro:ER leaves appeared intermediate between those of wild type and er erl1 erl2 (Fig. S4B-D), suggesting that ER functions in non-epidermal tissues may also contribute to the petiole growth.

We observed similar results for wus er erl1 erl2 AtML1pro:ER. Following AtML1pro:ER expression, the epidermal CLV3pro:GUS expression observed in wus er erl1 erl2 was no longer apparent (Figs 3D and 4C) and the high endogenous CLV3 expression observed in wus er erl1 erl2 was also considerably reduced (Fig. 4E). These results indicate that ER activity in the epidermis contributes to the maintenance of SAM homeostasis. Furthermore, AtML1pro:ER expression improved the small leaf phenotype of wus er erl1 erl2 (Fig. S4H,I), suggesting that the ER activity in the epidermis promotes leaf growth.

The CLV3 peptide, which is secreted from stem cells, plays an important role in regulating SAM homeostasis and treatment with exogenous CLV3 peptide can lead to the loss of both SAM structure and the SAM stem cell population (Kondo et al., 2006; Fig. 5A,B,E,F,M,N,Q,R, Fig. S5). Therefore, we addressed whether the ER family regulates CLV3 signaling by treating ER-family mutants with CLV3. The SAM was maintained in er erl1 erl2 in the presence of excess CLV3 peptide (Fig. 5C,D,G,H; Fig. S5), indicating that mutation of the ER family interferes with CLV3 signaling. In agreement, GUS signals arising from expression of CLV3pro:GUS as well as WUSpro:GUS, a marker of the OC (Laux et al., 1996; Mayer et al., 1998), were maintained in the er erl1 erl2 SAM following CLV3 treatment (Fig. 5K,L,O,P). By contrast, in the wild type, CLV3 application abolished the expression of both CLV3pro:GUS and WUSpro:GUS (Fig. 5I,J,M,N,Q,R).

Closer inspection of the CLV3pro:GUS expression pattern in er erl1 erl2 revealed that the effects of CLV3 treatment differed between the epidermal and the internal tissue layers (Fig. 5S-V; Fig. S6). CLV3 peptide did not affect CLV3pro:GUS expression in the epidermis (Fig. 5S-U), as the extent of epidermal GUS signal was not reduced upon CLV3 treatment in the er erl1 erl2 mutant (Fig. 5U). However, GUS signal in the internal tissue layers was significantly diminished (Fig. 5S,T,V). Thus, er erl1 erl2 stem cells situated in the internal tissue layers, but not the epidermal stem cells, retained the ability to respond to CLV3 signaling. These findings were consistent with the decreased expression level of endogenous CLV3 in er erl1 erl2 following CLV3 peptide treatment (Fig. 5W). Moreover, the SAM in clv3 er erl1 erl2 was larger than that in clv3 and er erl1 erl2 (Fig. 6; Fig. S7), suggesting that SAM development remains responsive to CLV3 signaling in er erl1 erl2, most likely via the influence of CLV3 on the internal tissue layers of the SAM. By contrast, CLV3pro:GUS expression in both the SAM epidermal and internal tissue layers in the wild type was similarly influenced by CLV3 peptide treatment (Fig. 5Q,R,U,V). Collectively, these results indicate that ER-family activity is required for the coordination of CLV3 signaling responses between the SAM epidermal and internal tissue layers.

In addition to its effect on the SAM, CLV3 treatment also affects root growth (Fiers et al., 2005; Kondo et al., 2006). In contrast to CLV3 responses in the SAM described above, the CLV3 peptide inhibited root growth in both wild type and er erl1 erl2 (Fig. S8), indicating that the ER-family mutation does not affect CLV3 signaling in roots.

Cytokinin acts as an important regulator of SAM homeostasis (Chickarmane et al., 2012; Gordon et al., 2009), and WUS promotes the responsiveness of the SAM to cytokinin (Leibfried et al., 2005). Based on the variation between the SAM epidermal and internal tissue layers in the absence of WUS activity in er erl1 erl2 mutants (Fig. 3), we hypothesized that the cytokinin signaling-mediated regulation of SAM homeostasis might also be modulated by the ER family in a tissue layer-specific manner. To investigate this, we analyzed cytokinin responses using the synthetic cytokinin response marker TCSn:GFP (Zurcher et al., 2013). Using this reporter, we detected the GFP signal in the OC of the wild-type SAM (Fig. 7A), consistent with previous reports (Chickarmane et al., 2012), and this GFP signal was maintained in the er erl1 erl2 mutant SAM (Fig. 7B). These TCSn:GFP expression patterns are consistent with the similar expression levels of cytokinin receptor and type-B ARR genes, which are positive regulators of cytokinin signaling (Meng et al., 2017; Wang et al., 2017), between er erl1 erl2 and wild type (Fig. S9). These results suggest that the primary cytokinin response is not disturbed in er erl1 erl2.

To examine the effects of ER-family activity on the stem cell behaviors under perturbed cytokinin signaling, we employed the wooden leg (wol) allele, a dominant-negative mutation in the OC-expressed cytokinin receptor gene ARABIDOPSIS HISTIDINE KINASE 4 (AHK4) that inhibits cytokinin receptor signaling (Mähönen et al., 2000). We observed a reduction in both SAM size and the extent of CLV3pro:GUS expression in the wol mutant (Fig. 7C,E), which agrees with the known phenotypes of cytokinin-deficient mutants (Werner et al., 2003). Moreover, CLV3pro:GUS expression was similarly reduced in both the SAM epidermal and internal tissue layers in wol (Fig. 7I,K), whereas in the er erl1 erl2 wol SAM, CLV3pro:GUS expression was reduced only in the internal tissue layers (Fig. 7D,F,J,L). Furthermore, a pharmacological approach based on treatment with the cytokinin receptor antagonist S-4893 (Arata et al., 2010) revealed expression patterns that corroborated those observed in wol (Fig. 7G-L). Therefore, in the ER-family mutant SAM epidermis, the expression of this stem cell marker is resistant to the loss of cytokinin signaling, whereas the maintenance of this marker expression in the SAM internal tissue layers requires cytokinin signaling.

Taken together, our data indicate that, in the maintenance of proper SAM homeostasis, the ER-family genes are required for the coordination of stem cell behaviors between the epidermal and internal tissue layers. Without ER-family activity, the influence of WUS, CLV3 and cytokinin on stem cells is decoupled between the different SAM tissue layers.

We demonstrated that the SAM stem cell loss observed in wus mutants is suppressed by loss of function of ER-family members (Fig. 1). Recent studies also reported that the mutations altered meristem program 1 (amp1) and class III homeodomain-leucine zipper (hd-zip III) partially complement the wus phenotype by promoting the production of adventitious SAMs (Huang et al., 2015; Lee and Clark, 2015). However, there are clear differences between the effects of er erl1 erl2, amp1 and hd-zip III on the wus phenotype. Although amp1 and hd-zip III contribute to the formation of adventitious SAMs in wus, the primary SAM is still consumed in these mutants. By contrast, the primary SAM is recovered in wus er erl1 erl2 (Fig. 1). Furthermore, wus er erl1 erl2 produces flowers with pistils (Table S1), whereas wus hd-zip III develops wus-like flowers that lack pistils (Lee and Clark, 2015). Thus, the ER family seems to function in a different manner from AMP1 and HD-ZIP III. This hypothesis agrees with the finding that jabba-1D, which suppresses HD-ZIP III, results in an enlarged SAM in er mutants, suggesting that HD-ZIP III and ER act in parallel (Mandel et al., 2016, 2014). As wus er erl1 erl2 maintains CLV3pro:GUS expression specifically in the SAM epidermis (Fig. 3D), which contrasts with the lack of GUS signal in all SAM tissue layers in wus (Fig. 3B), we proposed that ER suppresses WUS-independent processes that can maintain stem cells in the epidermis. We subsequently showed that localized ER expression in the epidermis is sufficient to suppress WUS-independent stem cell development (Fig. 4). Similarly, future research should focus on elucidating the SAM domains where AMP1 and HD-ZIPIII act to regulate SAM homeostasis.

In wild-type plants, CLV3pro:GUS expression was detected in the epidermal L1 layer and the internal L2 and L3 layers in the SAM center (Fig. 3A), indicating that stem cells are located within each of these layers. The stem cells in the epidermal L1 layer and the subepidermal L2 layer divide anticlinally, whereas those in the L3 layer divide both anticlinally and periclinally (Meyerowitz, 1997). All of these stem cells supply daughter cells to surrounding tissues for organ formation. In contrast to the multiple tissue-layer expression observed in the wild type, CLV3pro:GUS expression was restricted to the epidermis in the wus er erl1 erl2 SAM (Fig. 3D); however, cell proliferation was still detected in all SAM tissue layers (Fig. 2D). Moreover, our histological analyses revealed a population of small cells covering the wus er erl1 erl2 SAM (Fig. 1H).

The inconsistency between CLV3pro:GUS expression and cell proliferation within the wus er erl1 erl2 SAM can be explained in multiple ways. Periclinal division of stem cells in the wus er erl1 erl2 SAM epidermis may supply daughter cells in the subepidermal tissue layers that may then proliferate, thereby maintaining the internal tissues. Alternatively, in addition to the observed canonical epidermal stem cells, there may be stem cells that do not express the CLV3pro:GUS marker in the internal tissue of the wus er erl1 erl2 SAM. However, there is currently no evidence for the existence of CLV3-negative stem cells in the wild-type SAM, and future research to examine this possibility would require the development of alternate reliable stem cell markers. For this purpose, the reported Arabidopsis SAM transcriptome (Yadav et al., 2009, 2014) may be used alongside the materials made in this study.

Although the cytokinin response is primarily activated in the SAM OC alone (Adibi et al., 2016; Chickarmane et al., 2012; Gordon et al., 2009; Gruel et al., 2016), attenuation of cytokinin signaling results in an overall reduction in SAM size (Leibfried et al., 2005; Werner et al., 2003; Fig. 7C,E,G). This implies that a secondary signal exists that is activated following the primary cytokinin response in the OC, which would then non-cell-autonomously affect surrounding SAM cells. We showed that a normal cytokinin response was maintained in the OC of the er erl1 erl2 SAM (Fig. 7B). However, in contrast to that seen in the wild-type SAM, CLV3pro:GUS expression in the er erl1 erl2 SAM epidermis was not attenuated by cytokinin signaling (Fig. 7C-L). Thus, ER-family loss of function likely renders the epidermal stem cells independent of the non-cell-autonomous secondary signal that acts downstream of the primary cytokinin response in the OC. This hypothesis is consistent with the observed suppression of the er erl1 erl2 phenotype by ER expression in the SAM epidermis (Fig. 4). It will be interesting to characterize the molecular nature of this secondary signal in future studies. Given that WUS is known to both modulate the cytokinin responsiveness of the SAM (Leibfried et al., 2005) and regulate the expression of hundreds of genes (Busch et al., 2010), the WUS transcription factor may regulate gene expression that leads to production of a secondary signal in the SAM OC. Furthermore, given the similar phenotypes resulting from both CLV3 peptide treatment (Fig. 5Q-V) and the attenuation of cytokinin signaling (Fig. 7C-L), CLV3 signaling may also regulate production of such a secondary signal.

ER-family loss of function renders the SAM insensitive to CLV3 peptide treatment in a tissue layer-specific manner (Fig. 5). However, it is unlikely that ER-family proteins directly perceive the CLV3 signal, as direct binding of the CLV3 peptide to its corresponding receptor CLV1 has been unambiguously demonstrated (Ogawa et al., 2008). Several EPIDERMAL PATTERNING FACTOR-LIKE (EPFL) secreted peptides have been identified as ligands for ER-family proteins in stomatal patterning, inflorescence morphogenesis and leaf serration (Abrash et al., 2011; Lee et al., 2015, 2012; Tameshige et al., 2016; Uchida et al., 2012). Therefore, EPFL-family members may act as a yet-uncharacterized signal upstream of the ER family in stem cell maintenance. Accordingly, we found that EPFL1 and EPFL2 are expressed in the shoot apex (Fig. S10), although it remains unclear whether these genes act in regulating SAM functions. In this study, we show that the localized ER expression in the epidermis is sufficient to rescue the misexpression of stem cell marker CLV3 in the epidermis of ER-family mutants, whereas ER is expressed throughout the shoot apices (Uchida et al., 2013; Fig. S4A). Given that the shoot apex is a complex tissue composed of multiple domains, such as OC, the boundary region, the rib zone and the initiating primordia of lateral organs, it will be important in future research to delineate the individual functions of ER-family proteins in each domain and to elucidate whether the SAM-expressed EPFLs act through ER proteins in epidermal and/or non-epidermal domains.

Taken together, our findings indicate that ER-family receptor kinase genes coordinate stem cell behaviors between the SAM epidermal and internal tissue layers to ensure that all SAM stem cells behave as a single entity. Continuing research should focus on downstream events that regulate SAM stem cells following activation of ER-family signaling.

The mutant lines er erl1 erl2 (Shpak et al., 2004), wus (SAIL_150_G06) (Chatfield et al., 2013; Sonoda et al., 2007), wol (CS9817) (Mähönen et al., 2000), CLV3pro:GUS (Brand et al., 2002), TCSn:GFP (Zurcher et al., 2013), AtML1pro:GUS (Uchida et al., 2012), AtML1pro:ER (Uchida et al., 2012), EPFL2pro:GUS (Tameshige et al., 2016), WUSpro:GUS (Hirakawa et al., 2017) and ERpro:ER-YFP (Horst et al., 2015; Ikematsu et al., 2017) in Col have been reported previously. The mutant line clv3-2 in Ler was introgressed into Col three times and then crossed with er erl1 erl2. To engineer clv3 with functional ER, the ER genomic fragment derived from Col (Godiard et al., 2003) was introduced into clv3-2. The primers used for amplification of the WUS and EPFL1 promoter regions are listed in Table S2. WUSpro:ER was constructed according to the previously reported procedure for the AtML1pro:ER construction (Uchida et al., 2012). Plants were grown on Murashige and Skoog (MS) media at 22°C under continuous light. For chemical treatment, plants were grown on media containing either 5 μM 6-benzylaminopurine (BAP; Sigma, B3408), 5 μM CLV3 peptides (Operon) or 10 μM S-4893 (InterBioScreen, 1S-73130).

Seedling tissue was fixed with 4% FAA and then dehydrated via an ethanol series using 50-100% ethanol solutions before the ethanol was gradually exchanged with 100% acetone followed by critical point drying. Leaves were removed from the dried samples and vapor deposition was performed by ion spatter (E-1010, Hitachi). Samples were then observed using a field emission scanning electron microscope (S-4700, Hitachi).

Seedlings were treated with 90% acetone and incubated in a GUS staining solution [50 mM sodium phosphate buffer (pH 7.0), 10 mM potassium ferricyanide, 10 mM potassium ferrocyanide, 2 mM X-Gluc, 0.2% Triton-X] at 37°C. Samples were fixed with 4% FAA, embedded in Technovit7100 (Heraeus Kulzer) and sectioned using a microtome (Leica, RM2235). Sections were stained with either 0.04% Neutral Red or 0.02% Toluidine Blue. Quantitative analysis of the GUS signal in section images was performed using the ImageJ software. Cells exhibiting GUS signal in the internal tissue layers were selected using the polygonal lasso tool as shown in Fig. S6, and the selected area was quantified. GUS-stained epidermal cells were traced using the segmented line tool as shown in Fig. S6, and the length of the SAM epidermis exhibiting GUS signal was measured. In situ RNA hybridization experiments to detect CLV3 transcripts were performed according to the previous report (Uchida et al., 2013).

Total RNA was extracted from 10-day-old shoot apices using an RNeasy Plant Mini Kit (Qiagen). qRT-PCR was performed using a ReverTra Ace qPCR RT Master Mix with gDNA Remover kit (TOYOBO), a SYBR Fast qPCR kit (KAPA) and a Light Cycler 96 (Roche). The primers used for expression analyses are listed in Table S2.

Plant samples were embedded in 6% UltraPure Low Melting Point Agarose (Thermo Fisher), and then 70 μm sections were made using a vibrating microtome (Leica, VT1200S). Sections were mounted in water and stained with or without 25 μg/ml FM4-64 for counterstaining. GFP, YFP and FM4-64 fluorescence was observed by confocal microscopy (Leica, SP8 and Zeiss, LSM800) with excitation at 488 nm. Emission ranges were 495-555 nm for GFP, 500-546 nm for YFP and 580-626 nm for FM4-64.

Ten-day-old seedlings grown on solid MS media were incubated in ½ MS liquid medium containing 10 μM EdU (Click-iT EdU Alexa Fluor 488 imaging kit; Invitrogen) for 16 h. The seedlings were then treated with 90% acetone, washed three times with PBS, fixed with 4% FAA, embedded in Technovit7100 (Heraeus Kulzer), and sectioned using a microtome (LEICA, RM2235). The sections were treated with an Alexa Fluor 488 probe according to the manufacturer's protocol, and then fluorescence signals were observed by fluorescence microscopy (Zeiss, Axioimager A2 with FS 38HE filter).

We thank Dr Rüdiger Simon, Dr Bruno Müller and Dr Tatsuo Kakimoto for providing materials, Dr Ayako Miyazaki for critical reading of the manuscript, and Ms Rie Iwasaki for technical assistance. We also thank Dr Yoshikatsu Sato at WPI-ITbM Live Imaging Center for generous support in confocal microscopy.