Arabidopsis ERECTA-Family Receptor Kinases Mediate Morphological Alterations Stimulated by Activation of NB-LRR-Type UNI Proteins

Shoot apical meristems (SAMs), which maintain stem cells at the tips of stems, and axillary meristems (AMs), which arise at leaf axils for branch formation, play signiﬁcant roles in the establishment of plant architecture. Previously, we showed that, in Arabidopsis thaliana , activation of NB-LRR (nucleotide-binding site-leucine-rich repeat)-type UNI proteins affects plant morphology through modulation of the regulation of meristems. However, information about genes involved in the processes was still lacking. Here, we report that ERECTA (ER) receptor kinase family members cooperatively mediate the morphological alterations that are stimulated by activation of UNI proteins. uni-1D is a gain-of-function mutation in the UNI gene and uni-1D mutants exhibit early termination of inﬂorescence stem growth and also formation of extra AMs at leaf axils. The former defect involves modulation of the SAM activity and is suppressed by er mutation . Though the AM phenotype is not affected by a single er mutation, it is suppressed by simultaneous mutations of ER -family members. It was previously shown that trans -zeatin (tZ)-type cytokinins were involved in the morphological phenotypes of uni-1D mutants and that expression of CYP735A2 , which is essential for biosynthesis of tZ-type cytokinins, was modulated in uni-1D mutants. We show that this modulation of CYP735A2 expression requires activities of ER-family members. Moreover, the ER activity in UNI -expressing cells contributes to all morphological phenotypes of uni-1D mutants, suggesting that a cross-talk between ER-family-dependent and UNI-triggered signaling pathways plays a signiﬁcant role in the morphological alterations observed in uni-1D mutants.


Introduction
Shoot apical meristems (SAMs) and axillary meristems (AMs) play significant roles in the establishment of plant architecture. The SAMs, which are maintained at the tips of stems, are indeterminate structures and are the source of stem cells from which all post-embryonic aerial organs are derived (Brand et al. 2000, Schoof et al. 2000, Clark 2001, Lenhard and Laux 2003, Kondo et al. 2006, Muller et al. 2006, Tucker and Laux 2007, Miwa et al. 2008, Muller et al. 2008). On the other hand, AMs, which are required for branch formation, arise at leaf axils, which are boundary tissues between stems and leaves (Schmitz andTheres 2005, Aida andTasaka 2006). Changes in activities of these above-ground meristems affect plant morphology.
The uni-1D mutant of Arabidopsis thaliana harbors a gain-of-function mutation in the UNI gene, which encodes a member of the CC-NB-LRR (coiled-coil nucleotide-binding site-leucine-rich repeat) family (Igari et al. 2008). The heterozygous uni-1D/+ plants show interesting morphological phenotypes, i.e. early termination of inflorescence stem growth after formation of several fruits and formation of extra AMs at leaf axils (Igari et al. 2008), while the homozygous uni-1D plants exhibit severe growth defects soon after germination and stop growing at a vegetative stage. Although it was shown that trans-zeatin (tZ)-type cytokinins were involved in morphological alterations of uni-1D/+ plants (Igari et al. 2008), further aspects of these processes are still not clear. Because NB-LRR-type proteins function as molecular sensors to trigger intracellular signaling pathways upon stimulation by ligands (Jones and Dangl 2006, Bent and Mackey 2007, Padmanabhan et al. 2009), the uni-1D mutation probably converts UNI proteins into an active state without an activation event, which triggers downstream signaling pathways (Igari et al. 2008); however, such a ligand for activation of UNI proteins has not yet been identified. Many NB-LRR-type proteins have been reported to function in plant immunity as sensors for pathogen-derived factors to activate defense responses (Jones and Dangl 2006, Bent and Mackey 2007, Padmanabhan et al. 2009) and uni-1D plants also show some aspects of pathogenesis-related responses such as the up-regulation of PATHOGENESIS-RELATED GENE 1 (PR1) and PR5 through the salicylic acid (SA) pathway (Igari et al. 2008). However, it has been shown that inactivation of the SA pathway did not affect the morphological phenotypes of uni-1D/+ mutants (Igari et al. 2008) and so far there has been no evidence that UNI is involved in plant immunity.
The ERECTA (ER) family consists of ER, ER-LIKE 1 (ERL1) and ERL2 (Shpak et al. 2004) and encodes leucine-rich repeat receptorlike kinases in a subfamily of transmembrane-type signaling receptors in plants (Torii et al. 1996, Shpak et al. 2004). The ER-family members are broadly expressed in various tissues (Yokoyama et al. 1998, Shpak et al. 2004) and play roles in diverse aspects of plant development by mediating cell-cell signals that sense and coordinate organ development (Torii et al. 1996, Shpak et al. 2004, Tisne et al. 2008. Though a loss-of-function mutation in the ER gene affect inflorescence architecture but do not cause severe growth defects (Torii et al. 1996), simultaneous loss of activities of more than one ERfamily member confers extreme dwarfism (Shpak et al. 2004), abnormal flower development (Shpak et al. 2004, Pillitteri et al. 2007, Hord et al. 2008) and defects of stomatal patterning and differentiation (Shpak et al. 2005). Furthermore, ER has been identified as a major trait in various quantitative trait locus (QTL)/expression quantitative trait locus (eQTL) analyses of environmental stresses and developmental processes, suggesting that ER could be a modulator of signaling pathways in response to changes in external and/or internal conditions (Keurentjes et al. 2007, Tisne et al. 2008, Ghandilyan et al. 2009, El-Lithy et al. 2010, Terpstra et al. 2010, van Zanten et al. 2010a, van Zanten et al. 2010b.
In this study we report that ER-family members cooperatively mediate morphological alterations that are stimulated by activation of UNI proteins. Early termination of inflorescence stem growth observed in uni-1D/+ mutants is suppressed by er mutation. On the other hand, though the AM phenotype is not affected by single er mutations, it is suppressed by simultaneous mutations of more than one ER-family member. Cytokinin biosynthesis pathways and responses to cytokinins are modulated in uni-1D mutants, and these modulations require activities of ER-family members. Furthermore, ER activities in UNI-expressing cells contribute to all morphological phenotypes of uni-1D mutants, suggesting that a signaling cross-talk between ER-family-dependent and UNI-triggered signaling pathways plays a significant role in the morphological alterations observed in uni-1D mutants.

Results
Early termination of inflorescence stem growth caused by the uni-1D mutation is restored in the Ler background.
Thus, among the morphological phenotypes of the original uni-1D/+ mutants in the Ws background, the early cessation of inflorescence growth is restored specifically in the Ler background.
ER is required for early termination of inflorescence stem growth in uni-1D/+ plants Because the Ler accession is known to harbor a loss-of-function mutation in the ER gene (Torii et al. 1996), which functions in various developmental aspects of plants (Torii et al. 1996, Shpak et al. 2003, Shpak et al. 2004, Shpak et al. 2005, Woodward et al. 2005, van Zanten et al. 2009), we examined whether the er mutation in Ler plants is responsible for the rescue of rapid termination of inflorescence growth in uni-1D/+ (Ler) plants. To this end, a functional ER genomic fragment was introduced into uni-1D/+ (Ler) plants. The resulting uni-1D/+ ER + (Ler) plants showed early cessation of stem growth ( Fig. 2A) and the architecture of the uni-1D/+ ER + (Ler) plants was equivalent to that of the the original uni-1D/+ (Ws) plants (Figs. 1A, 2A). These observations strongly suggest that the presence of ER activities is associated with early termination of inflorescence stem growth in uni-1D/+ plants. To examine this possibility further, er loss-of-function mutants in Ws and Col backgrounds were crossed with uni-1D/+ (Ws) plants and uni-1D/+ (Col) plants, respectively. Inflorescence stems of er uni-1D/+ plants elongated well in both backgrounds (Fig. 2B, C) and, accordingly, inflorescence SAMs of er uni-1D/+ plants continuously produce flowers as in wild-type plants (Fig. 2D, F), while in uni-1D/+ plants only a few flowers were formed and then flower formation stopped (Fig. 2E) (Igari et al. 2008). When we examined expression patterns of WUS, a central player in stem cell maintenance in SAMs, WUS expression was detected in central domains of inflorescence SAMs of wild-type and er uni-1D/+ plants by in situ RNA hybridization ( Fig. 2G, I, arrowheads). In contrast, in uni-1D/+ mutants, inflorescence SAMs were apparently small and the expression level of WUS was below the detection limit (Fig. 2H, arrowhead). WUS is also expressed in floral meristems during flower development (Fig. 2G, arrow) (Laux et al. 1996, Lenhard et al. 2001, Lohmann et al. 2001) and this expression in floral meristems was not affected in uni-1D/+ plants (Fig. 2H, arrow). These observations were in accordance with the fact that uni-1D/+ plants were able to produce normal flowers once flower formation started ( Fig. 2E) (Igari et al. 2008). Thus, ER is required for early termination of inflorescence stem growth in uni-1D/+ plants.  As shown in Fig. 1G, the Ler background did not affect the formation of extra axillary buds at axils of cotyledons that was observed in the original uni-1D/+ plants in the Ws background, suggesting that the ER function is not involved in this phenotype. When we compared er uni-1D/+ plants with er mutants in the original Ws background, er mutants formed no axillary buds at the axils of cotyledons and the first pair of leaves (Fig. 2J). On the other hand, in er uni-1D/+ plants, formation of axillary buds was observed even at both the axils of cotyledons and the first pair of leaves (Fig. 2K). These results indicate that the er mutation does not affect formation of extra axillary buds at leaf axils in uni-1D/+ plants.
ER is not involved in up-regulation of PR genes in uni-1D/+ plants uni-1D/+ mutants exhibit a high level of expression of the pathogenesis-related genes PR1 and PR5 (Igari et al. 2008). We next investigated whether ER is required for up-regulation of these PR genes. As previously reported (Igari et al. 2008), PR1 and PR5 expression was increased in uni-1D/+ mutants compared with wild-type plants (Fig. 3). The same degree of PR1/5 upregulation was also detected in uni-1D/+ plants harboring the er mutation (Fig. 3). These results indicate that the ER activity does not influence PR1/5 expression in uni-1D/+ mutants. Thus, ER function is specifically involved in early cessation of inflorescence growth of uni-1D/+ plants but not in either formation of extra axillary buds at leaf axils or up-regulation of PR genes.
The ER activity in the UNI promoter active region contributes to the SAM defect of uni-1D/+ plants ER is expressed in a broad range of tissues during diverse developmental processes (Yokoyama et al. 1998, Shpak et al. 2004, Shpak et al. 2005. Because both ER and UNI have protein structures that can act as starting points to trigger intracellular signaling pathways, we hypothesized that the signaling pathways activated by them could cooperate in the same cells for the SAM defect of uni-1D/+ plants. To examine this possibility, we made constructs in which the ER coding sequence is translationally fused to a FLAG-tag at its C-terminus and is driven by the ER promoter (ProER:ER-FLAG) or by the UNI promoter (ProUNI:ER-FLAG). er mutant plants developed short inflorescence stems (Fig. 4A) and clustered flower buds at the top of the inflorescence (Fig. 4D), as previously described (Torii et al. 1996). Introduction of the ProER:ER-FLAG transgene into the er mutant rescued all er phenotypes (Fig. 4B, E) with flower buds folded on to each other to cover the SAM (Fig. 4E). When the expression pattern of ER-FLAG proteins was examined by immunohistochemistry using anti-FLAG antibodies, the signals were broadly detected in the inflorescence tip (Fig. 4H), in agreement with previous reports (Yokoyama et al. 1998, Shpak et al. 2004). The ProER:ER-FLAG transgene was then introduced into er uni-1D/+ plants. The suppression of early termination of inflorescence growth of uni-1D/+ plants by er mutation was complemented by the ProER:ER-FLAG as expected ( Fig. 4J-M).
Next, the ProUNI:ER-FLAG transgene was introduced into the er mutant. The promoter region used in this experiment was previously shown to be sufficient to induce all uni-1D/+ phenotypes when it was fused to UNI cDNA harboring the uni-1D mutation (Igari et al. 2008). The ProUNI:GUS transgenic plants showed strong b-glucuronidase (GUS) activity in regions beneath the SAM and also in the vasculature (Supplementary Fig. S1A-C). In the ProUNI:ER-FLAG transgenic plants, the ER-FLAG proteins were also detected in regions beneath the SAM by anti-FLAG immunohistochemistry (Fig. 4I), which is in accordance with the result using the ProUNI:GUS transgenic plants ( Supplementary Fig. S1B, C). The short inflorescence stems and clustered flower buds at the inflorescence tip with the er mutation were rescued by ProUNI:ER-FLAG (Fig. 4C, F) as well as ProER:ER-FLAG (Fig. 4B, E). The ProUNI:ER-FLAG transgene was then introduced into er uni-1D/+ plants. er uni-1D/+ plants harboring the ProUNI:ER-FLAG exhibited uni-1D/+ morphology (Fig. 4J, K, N), showing that the ER activity in the region where the UNI promoter is active contributes to the SAM defect observed in uni-1D/+ mutants.
ER-family members cooperatively mediate the AM phenotype of uni-1D/+ plants ER has two family genes, ERL1 and ERL2, and it has been reported that the simultaneous loss of activities of ER-family members affects various events in plant development (Shpak et al. 2004, Shpak et al. 2005, Pillitteri et al. 2007, Hord et al. 2008. Therefore, we next examined whether activities of the other ER-family members are also involved in uni-1D/+ morphological phenotypes. We first checked the effects of simultaneous loss of ERL1 and ERL2 functions. As shown in Supplementary Fig. S2, erl1 erl2 uni-1D/+ plant still showed the original uni-1D/+ morphology. Accordingly, the formation of extra axillary buds, which was not seen in wild-type plants (Fig. 5A), was observed in erl1 erl2 uni-1D/+ plants as well as in uni-1D/+ and er uni-1D/+ plants (Fig. 5B-D). We next examined the effects of other combinations of mutations among ERfamily members. Most er erl1 uni-1D/+ plants still produced extra axillary buds (Fig. 5E), while er erl2 uni-1D/+ plants did not show this phenotype (Fig. 5F), demonstrating significant roles for ER and ERL2 in the AM phenotype of uni-1D/+ plants.
Cytokinin response and CYP735A2 expression are up-regulated in uni-1D/+ mutants but are suppressed by attenuation of activities of ER-family members Among combinations of mutations of ER-family genes, we found the er erl1/+ erl2 uni-1D/+ plants showed a similar morphology to er erl1/+ erl2 plants without any sign of the morphological alterations observed in uni-1D/+ plants (Fig. 6A, Supplementary Fig. 3). It has been reported that tZ-type cytokinins accumulated in uni-1D/+ plants and, correspondingly, expression of the cytokinin-responsive ARR5 gene was upregulated in uni-1D/+ plants (Igari et al. 2008). When cytokinins were artificially reduced in uni-1D/+ seedlings, ARR5 up-regulation and also morphological alterations were suppressed, suggesting the importance of cytokinins for the phenotype (Igari et al. 2008). Therefore, we first checked the effect of attenuation of ER-family activities on the cytokinin response by determining the expression level of ARR5. As shown in Fig. 6B, ARR5 expression was up-regulated in uni-1D/+ plants compared with wild-type plants, but this ARR5 up-regulation was suppressed in er erl1/+ erl2 uni-1D/+ plants. Among active cytokinins, only tZ-type cytokinins are specifically accumulated in uni-1D/+ plants (Igari et al. 2008) and the production of the tZ-type cytokinins requires the CYP735A enzyme in their biosynthesis pathway (Takei et al. 2004). CYP735A2 expression was up-regulated in uni-1D/+ plants compared with wild-type plants (Fig. 6C) (Igari et al. 2008), but this CYP735A2 up-regulation was not observed in er erl1/+ erl2 uni-1D/+ plants (Fig. 6C), indicating that activities of ER-family members contribute to the modulation of CYP735A expression in uni-1D/+ plants.
Activities of ER-family members in the UNI promoter active region contribute to all morphological phenotypes of uni-1D/+ plants Because the ER activity in the region where the UNI promoter is active contributed to the SAM phenotype of uni-1D/+ plants (Fig. 4), we next examined whether a similar scenario could be occurring in the case of the suppression of all uni-1D/+ morphological phenotypes by attenuation of activities of ER-family members. The suppression of the uni-1D/+ phenotypes by er erl1/+ erl2 was complemented by the introduction of the ProER:ER-FLAG transgene as expected (Fig. 7). When the ProUNI:ER-FLAG transgenes were introduced into er erl1/+ erl2 uni-1D/+ plants, the ProUNI:ER-FLAG transgene had the same effect as the ProER:ER-FLAG transgene (Fig. 7). These observations show that the ER activity in the region where the UNI promoter is active is sufficient for uni-1D/+ plants to exhibit morphological alterations. Together, these data indicate that ER-family members cooperatively contribute to all morphological phenotypes observed in uni-1D/+ plants by exerting their functions in the UNI promoter active regions.

Relationships between signaling pathways triggered by UNI and those triggered by ER-family members
The structure of UNI is related to the NB-LRR family. Most well-studied NB-LRRs serve as intracellular immune receptors (Jones andDangl 2006, Caplan et al. 2008). These NB-LRRs are activated by specific pathogen effectors and they then trigger defense responses. Interestingly, uni-1D plants have morphological phenotypes that can be genetically separated from the pathogenesis responses (Igari et al. 2008). Some NB-LRR-related factors also play roles in the biological processes that are distinct from pathogenesis-related regulations (Faigon-Soverna et al. 2006, Hewezi et al. 2006, Bomblies et al. 2007, Alcazar et al. 2009) and therefore UNI could be classified as belonging to this type of NB-LRR-related factor. As UNI has a protein structure that is potentially able to serve as a molecular switch in response to a ligand that can activate UNI proteins, it would be possible for UNI proteins to sense changes in the internal or external environment, via reception of an as yet unidentified ligand, and then function as a molecular switch to trigger the signaling for morphological regulation. The unidentified ligands to activate UNI proteins could be missing in wild-type Arabidopsis plants under normal growth conditions. This speculation is in accordance with the fact that the loss of UNI function by a T-DNA insertion induced no obvious phenotype (Igari et al. 2008).
Because both ER-family proteins and UNI proteins have structures that can act as starting points to trigger intracellular signaling pathways, and also because they play roles in the UNIexpressing cells for morphological phenotypes of uni-1D/+ plants (Figs. 4, 7), it would be interesting to investigate intracellular signaling cross-talk between signaling pathways triggered by ER-family proteins and those triggered by UNI proteins. Furthermore, it would be interesting to analyze how these signaling pathways are connected to regulation of CYP735A2 expression (Fig. 6).
Different thresholds of activities of ER-family proteins for each uni-1D/+ phenotype uni-1D/+ mutants exhibit two morphological phenotypes: early termination of inflorescence stem growth and formation of extra AMs at leaf axils ( Fig. 1) (Igari et al. 2008). When only ER was mutated in uni-1D/+ plants, the former defect was suppressed (Fig. 2). On the other hand, attenuation of ER-family activities in uni-1D/+ plants resulted in suppression of not only the inflorescence defect but also the AM defect (Fig. 5). This difference in threshold of ER-family activities required for each phenotype of uni-1D/+ plants could be explained by the fact that proper development of each tissue needs different levels of ER-family activities. Loss-of-function mutations just in the ER gene can induce phenotypes in inflorescence architecture (Torii et al. 1996) though ER-family members are broadly expressed in various tissues, including inflorescences (Yokoyama et al. 1998, Shpak et al. 2004). Developmental processes of other tissues, such as flowers (Shpak et al. 2004, Pillitteri et al. 2007, Hord et al. 2008) and stomata (Shpak et al. 2005), are affected only when activities of more than one ER-family member are simultaneously lost. Thus, the development of inflorescences could be more sensitive to a reduction in total activities of ER-family proteins compared with developmental processes in other tissues. This could be a reason why the inflorescence defect of uni-1D/+ mutants was suppressed by single er mutations (Fig. 2) and the AM phenotype was more tolerant (Fig. 5).

Relationships between cytokinins and uni-1D/+ phenotypes
Though cytokinin is known to be an important hormone, playing significant roles in various developmental processes in plants (To and Kieber 2008, Werner and Schmulling 2009, Perilli et al. 2010, reports on its involvement in initiation processes of AM formation are limited. Interestingly, it was reported that infection of Arabidopsis with Rhodococcus fascians provokes ectopic formation of AMs, resulting in an overall bushy appearance (Vereecke et al. 2000, Manes et al. 2004, and that R. fascians produces cytokinin derivatives during its pathology, which are poor substrates of cytokinin degradation enzymes of host plants (Pertry et al. 2009 recognition by plant cytokinin receptors (Pertry et al. 2009), triggering fast activation of ARR5 expression (Depuydt et al. 2008). In uni-1D/+ mutants, formation of extra AMs was associated with modulation of cytokinin biosynthesis pathways and up-regulation of ARR5 expression (Fig. 6) (Igari et al. 2008), resembling the events provoked by R. fascians infection, except for the source of the cytokinins. These observations suggest the importance of cytokinin regulation to alter plant morphology through activation of AM formation. Because ER-family members were involved in the formation of extra AMs in uni-1D/+ plants (Fig. 6), it is an attractive hypothesis that ER-family members play an important role in the morphological alterations observed in R. fascians infection. It has been shown that a feedback relationship between WUS and cytokinin signaling controls meristem homeostasis (Leibfried et al. 2005, Gordon et al. 2009). WUS creates a domain of high cytokinin responsiveness by negatively regulating ARR genes that can function as inhibitors of the cytokinin response (Leibfried et al. 2005). In turn, the localized high responsiveness to cytokinin establishes a spatial domain in which stem cell fate is specified through induction of WUS (Gordon et al. 2009). In uni-1D/+ plants, expression of ARR5, an inhibitor of the cytokinin response, was up-regulated ( Fig. 6) and WUS expression was altered (Fig. 2), suggesting that the homeostasis between WUS and cytokinin signaling could be disrupted and that this disruption could be involved in morphological alterations of uni-1D/+ plants. Because ER-family members participate in this event (Fig. 6), it would be of interest to investigate whether ER-family members play a role in the feedback loop between WUS and cytokinin signaling for meristem homeostasis.
Although the levels of ARR5 and CYP735A2 transcripts in uni-1D/+ plants were reduced to the wild-type level by er erl1/ + erl2, those in wild-type plants were not significantly affected by er erl1/+ erl2 (Fig. 6). This implies that ER-family members do not play important roles in cytokinin-related regulation under normal conditions but exert their functions to affect expression of CYP735A2 and ARR5 in response to changes in growth conditions via stimuli such as activation of UNI proteins. Alternatively, the remaining activity of ERL1 in er erl1/+ erl2 plants could be enough to maintain proper regulation of the cytokinin-related genes in the wild-type background.

Possibility of coordination between environmental changes and morphological regulation via NB-LRR-related proteins and the ER-family
The functional UNI promoter (Igari et al. 2008) has activities mainly outside meristems ( Supplementary Fig. S1), although uni-1D/+ mutants exhibit meristem-related morphological phenotypes. This could suggest that activation of UNI proteins outside meristems affects the regulation of meristems. Furthermore, ER has been identified as a major trait in various QTL/eQTL analyses regarding environmental stresses and developmental processes, suggesting that ER could be a modulator of signaling pathways in response to changes in external and/or internal conditions (Keurentjes et al. 2007, Tisne et al. 2008, Ghandilyan et al. 2009, El-Lithy et al. 2010, Terpstra et al. 2010, van Zanten et al. 2010a, van Zanten et al. 2010b. Taken together, it would be attractive to hypothesize the existence of a system that connects environmental changes to morphological regulation via UNI proteins and ER-family receptor kinases. Because there are about 150 NB-LRR-related genes in the sequenced genome of A. thaliana (Meyers et al. 2003) and the biological roles of most of them have not been investigated, the possibility that some of them could function in such systems for the coordination between environmental changes and morphological regulation cannot be excluded. Future reports on this type of system involving NB-LRR-related genes should open up further aspects of the relationships between morphological regulation and environmental responses and it would be interesting to examine the involvement of ER-family receptor kinases in such systems.

Plant materials
The uni-1D allele originally identified in the Ws accession was introgressed six times into Col and Ler. pKUT196 (Ler) harboring the ER genomic fragment was described previously (Shpak et al. 2003). The er-2 mutant in the Col background (CS3401) and the er-201 mutant in the Ws background (FLAG_395D02) were obtained from ABRC and INRA, respectively. er-105, erl1-2 and erl2-1 mutants in the Col background were reported previously (Torii et al. 1996, Shpak et al. 2004).

Plasmids and plant transformation
For ProUNI:GUS construction, the 2.2 kbp promoter region upstream from the translation start codon of the UNI gene was inserted into the pBI101 vector.
To construct ProER:ER-FLAG:ERterm, PCR was performed using the primer pair ERg4359 (CAACAATGATCTGGAAGG AC) and ERECTAnostopBamHI.rc (CGCGGATCCGCGCTCAC TGTTCTGAGAAATAAC), and the amplified fragment was inserted into the SacI/BamHI site of pKUT195 to generate pNBL111, which harbors the ER genomic fragment from the 1,678 base promoter region to the unique XbaI site in the coding region, the ER cDNA fragment from the XbaI site to the end of the open reading frame without the stop codon, and the 1,964 bp ER terminator region. The 3 Â FLAG cassette with a stop codon (pNLB112) was generated by PCR using pF3PZPY122 [a gift from Dr. Xing-Wang Deng (Feng et al. 2003)] as a template with the primer pair: 3XFlag5 (GAAGTT CATTTCATTTGGAGAG) and 3XFlag3 BglII.rc (GAAGATCTTC GACTTTATCGTCATCGTC) and cloned into pCR2.1 (Invitrogen). Subsequently, the BamHI/BglII fragment of the 3 Â FLAG cassette was inserted into the BamHI site of pNLB111 to generate ProER:ER-FLAG:ERterm (pNLB115). This construct was introduced into the er-105 mutant and rescued the er phenotypes. For construction of ProER:ER-FLAG and ProUNI:ER-FLAG used in experiments described in this study, the fragment containing the ER coding region, 3 Â FLAG-tag and the ER 3 0 -untranslated region was amplified by PCR using the genome prepared from er-105 mutants harboring the original ProER:ER-FLAG transgene. The PCR products were then fused to the 2 kb ER promoter and the 2.2 kb UNI promoter, respectively, in the pBIN30 vector.
MP90 Agro strains harboring each plasmid were used for plant transformation. Ws wild-type plants and er-2 mutants in the Col accession were parent lines for transformation of ProUNI:GUS and ER-FLAG series, respectively.

Quantitative real-time PCR
RNA isolation, synthesis of first-strand cDNA and real-time PCR for ARR5 were performed as previously described. Primers used for CYP735A2 were GCTCTTCCATCCACCACAACA and CGGA TTGTGCTTCGTTAGCA.

GUS histology
The procedures for GUS staining were described in Uchida et al. (2007). The stained material was cleared with chloral hydrate and then observed.

In situ RNA hybridization
In situ hybridization was performed according to Takada et al. (2001). Templates for transcription of a WUS antisense probe were described in Hamada et al. (2000).

Immunohistochemistry
Immunohistochemistry was performed using mouse anti-FLAG M2 antibody (Sigma, F1804) and ImmPRESS reagent anti-mouse Ig (Vector Laboratories, MP-7402) according to the manufacturer's instructions with the additional treatment of sectioned tissues with proteinase K before the blocking stage. ImmPACT DAB (Vector Laboratories, SK-4105) was used as a peroxidase substrate.

Supplementary data
Supplementary data are available at PCP online.