Epidermal cell density is auto-regulated via a secretory peptide, EPIDERMAL PATTERNING FACTOR 2 in Arabidopsis leaves.

Transcription

1 Plant and Cell Physiology Advance Access published May 12, 2009 Epidermal cell density is auto-regulated via a secretory peptide, EPIDERMAL PATTERNING FACTOR 2 in Arabidopsis leaves. Category: Rapid Paper Subject area: Growth and development Running title: EPF2 peptide regulates epidermal cell density. Number of figures: 13 Number of supplementary figures : 2 Attachment: Cover page candidate Corresponding author: Tatsuo Kakimoto Department of Biological Sciences, Graduate School of Science, Osaka University. Toyonaka, Osaka , Japan. kakimoto@bio.sci.osaka-u.ac.jp TEL/FAX: Authors: Kenta Hara 1, Toshiya Yokoo 1, Ryoko Kajita 1, Takaaki Onishi, Saiko Yahata 1, Kylee M. Peterson 2, Keiko U. Torii 2, and Tatsuo Kakimoto 1 * 1. Department of Biological Science, Graduate School of Sciences, Osaka University, Toyonaka, Osaka , Japan 2. Department of Biology, University of Washington, Seattle, WA 98195, USA Abbreviations: MMC, meristemoid mother cell. RT-PCR, Reverse transcription polymerase chain reaction. GMC, guard mother cell. SLGC, stomatal lineage ground cell. MAPK, mitogen activated kinase. The Author Published by Oxford University Press on behalf of Japanese Society of Plant Physiologists. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org 1

2 Abstract Regulation of the number of cells is critical for development of multicellular organisms. During plant epidermal development, a protodermal cell first makes a fate decision of whether or not to be the meristemoid mother cell (MMC), which undergoes asymmetric cell division forming a meristemoid and its sister cell. The MMC-derived lineage produces all stomatal guard cells and a large proportion of non-guard cells. We demonstrate that a small secretory peptide, EPIDERMAL PATTERING FACTOR 2 (EPF2), is produced by MMC and its early descendants, and negatively regulates the density of guard and non-guard epidermal cells. Our results suggest that EPF2 inhibits cells from adopting the MMC fate in a non-cell-autonomous manner, thus limiting the number of MMCs. This feedback loop is critical for regulation of epidermal cell density. EPF2 resembles in its amino acid sequence to EPF1, which is known to control stomatal positioning. Overexpression of EPF1 also inhibits stomatal development, but EPF1 can act only on a later developmental process than EPF2 can do. Overexpression and promoter-swapping experiments suggested that protein functions of EPF1 and EPF2, rather than expression patterns of these genes, are responsible for the specific functions. Although targets of EPF1 and EPF2 are different, both EPF1 and EPF2 require common putative receptor components TMM, ER, ERL1, and ERL2 to function. 2

3 Introduction The epidermis of the plant shoot consists of several types of cells: pairs of guard cells that constitute stomata, pavement cells, and trichome cells (Sachs 1991). The guard cells are made in the MMC-derived lineage. The MMC has been defined as a cell that was directly formed from a protodermal cell and then begins to asymmetrically divide forming the meristemoid and its sister cell (Geisler et al. 2000). Meristemoids are visually distinguishable from their sister cells by their triangular shape and smaller size. The term "stomatal lineage cells" includes all cells derived from MMC (Geisler et al. 2000). The meristemoid may further undergo another one or two rounds of asymmetric cell divisions, known as amplifying divisions, creating another meristemoid and sister cell. The meristemoid eventually becomes the oval-shaped guard mother cell (GMC), which then divides symmetrically one time to form the paired guard cells. A sister cell may also undergo formative asymmetric cell divisions (called the amplifying cell divisions), creating a satellite meristemoid and its sister cell (Bergmann and Sack 2007; Geisler et al. 2000; Nadeau 2009). These proliferative sister cells and their descendant pavement cells are called stomatal lineage ground cells (SLGCs) (Nadeau 2009; Shpak et al. 2005). MMCs, a fraction of meristemoids, and proliferative SLGCs share the potential to form a meristemoid. Specification of MMC, GMC, and the guard cell is governed by closely related three bhlh transcription factors SPEECHLESS (SPCH), MUTE, and FAMA, respectively (MacAlister et al. 2007; Pillitteri et al. 2007). Two other bhlh proteins, ICE1 and SCREAM2, work together with SPCH, MUTE and FAMA presumably via forming heterodimers (Kanaoka et al. 2008). The frequency of asymmetric cell division is the major determinant of the numbers of both stomatal and non-stomatal epidermal cells. Stomata are always separated by at least one non-stomatal cell (called the one-cell spacing rule)(sachs 1991). The plane of asymmetric cell division in the sister cell is positioned so that a new meristemoid is formed away from pre-existing stomata or precursors, which in turn ensures the one-cell-spacing rule (Geisler et al. 2000). The plane of asymmetric cell division is controlled by a secretory peptide, EPIDERMAL PATTERNING FACTOR 1 (EPF1), which is produced in meristemoids, GMC and young guard cells (Hara et al. 2007). However we cannot eliminate the possibility that EPF1 also directly affects cell identity, so that cells close to the meristemoids, GMCs or young guard cells lose potential to be guard cells. Receptor kinases ERECTA (ER), ERECTA LIKE1 (ERL1), and ERECTA LIKE 2 (ERL2) (Shpak et al. 2005), 3

4 which have partially redundant functions, and a receptor-like protein TOO MANY MOUTHS (TMM) (Nadeau and Sack 2002) negatively regulate stomatal density and placement. These putative receptor components are required for the function of EPF1 (Hara et al. 2007), perhaps as receptors for EPF1. It is hypothesized that these putative receptor components function as a complex. Several lines of evidence suggest that a mitogen activated protein kinase (MAPK)-cascade transmits the signal from the putative receptor components to transcriptional regulators. Disruption of the MAPK kinase kinase (MAPKKK), YODA (YDA) (Bergmann et al. 2004), as well as double disruption of two MAPKs, MPK3 and MPK6 (Wang et al. 2007), or double suppression of two MAPK kinases, MKK4 and MKK5 (Wang et al. 2007), results in an increased number of stomata and violation of the one-cell-spacing rule. The yda loss-of function mutant is epistatic to tmm (Bergmann et al. 2004), and spch is epistatic to yda (MacAlister et al. 2007). Expression of constitutively active forms of YDA (Bergmann 2004), MPK4 or MPK5 (Wang et al. 2007) inhibits entry into the stomatal lineage. The MAPK cascade appears to phosphorylate and leads to disruption of the SPCH protein (Lampard et al. 2008). Genetic evidence suggested that EPF1 is upstream of this signaling pathway and regulates stomatal positioning conforming the one-cell-spacing rule, independently of SDD1 (Hara et al. 2007). SDD1 is a subtilisin-like required for stomatal spacing and limiting the density of stomata (Berger and Altmann 2000). Although disruption of genes for the putative receptor and the MAP kinase cascade affects both stomatal density (non-guard cell density has not been studied well) and placement, epf1 primarily affects stomatal placement. Therefore, there might be an unknown signaling molecule that regulates putative receptors and the MAP kinase cascade for the control of epidermal cell-density. Here we describe that EPF2, which encodes a putative secreted peptide with amino acid sequence similarity to EPF1, is expressed in MMCs and their early descendants, and negatively regulates formation of MMC and hence epidermal cell-density. Results Identification of new bioactive secretory peptides Arabidopsis has 10 genes for EPF1 homologues, including two previously unpredicted genes (Fig. 1). They all have a predicted secretory signal sequence according to PSORT ( and 6 conserved Cys residues at the C-terminal region end. To gain insight into the functions of the EPF1 homologues, we over-expressed these genes in 4

5 Arabidopsis, and found that the stomatal density was decreased in overexpressors of EPF1 (Fig.2B), At1g34245 (Fig. 2C), and of At4g14723 and At3g22820 (data not shown). Overexpression of a gene is often useful to understand the function of the gene. However, overexpression of a gene may confer phenotypes unrelated to its true function, especially in the case where transcriptional regulation plays a key role in the regulation its function. We next examined whether expression patterns of these genes are relevant to stomatal formation. In the transgenic plants harboring EPF1-GFP or EPF2-GFP, promoters of EPF1 (Hara et al. 2007) and At1g34245 (Fig. 3A, B) activated GFP in stomatal precursors. By contrast, patterns of reporter gene expression from promoters of At4g14723 and At3g22820 were not relevant to stomatal development (data not shown). We further tested whether T-DNA insertion mutants for these genes show any defects in stomatal development. The T-DNA insertion lines for At4g14723 (SALK_071065, T-DNA is present in an exon), At3g22820 (SALK_018403, T-DNA is in an intron), or their double mutant failed to show any recognizable abnormalities in epidermal development, suggesting that these genes are unlikely important to epidermal development. Thus we focused on At1g34245, being designated EPIDERMAL PATTERNING FACTOR 2 (EPF2). We name the other 8 genes EPF-LIKE 1 (EPFL1) to EPFL8. The expression pattern of EPF2 RT-PCR-analysis revealed that EPF2, as well as EPF1, is preferentially expressed in the aerial organs (Fig. 3C). An in situ RNA hybridization experiment showed that the EPF2 message is present in a scattered pattern only in small, young leaves (Fig. 3D, E), consistent with the EPF2-GFP expression. During early leaf development, the onset of EPF2-GFP signal preceded the emergence of meristemoids (Fig.3A). The GFP signal was later detected in meristemoids, their sister cells, and guard mother cells (GMCs). It was also expressed in another type of cells with a small quadrangle shape, which are perhaps cells with asymmetric cell division competency, including MMC. EPF2-GFP was not detected in guard cells and in any non-epidermal cells. The developmental window of EPF2-GFP expression is slightly earlier than that of EPF1-GFP, which expresses in a fraction of meristemoid and GMC and young guard cells, but not in the sister cells and MMC (Hara et al. 2007). To verify that EPF2 is expressed in the stomatal lineage, we examined expression of EPF2-GFP in the spch mutant background, which lacks all stomatal lineage cells. EPF2-GFP, EPF1-GFP and TMM-GFP were expressed in early stages of leaf development 5

6 in wild-type or SPCH(+/-) heterozygotes, but not in the spch homozygotes (Fig.4). On the basis of these observations, we conclude that EPF2 is expressed in MMCs and early meristemoids. However, it should be noted that we do not know whether all the EPF2-expressing cells undergo asymmetric cell division. Effects of EPF2 overexpression on numbers of guard and non-guard cells Although overexpression of either EPF1 or EPF2 decreased stomatal density, their epidermal phenotypes are different. The epidermis of EPF1 overexpressors has both small and large epidermal cells (Fig.2, B and D), with an increased number of small non-guard cells, in place of decrease in guard cells (Fig.2 D). By contrast, the epidermis of EPF2 overexpressors is devoid of small pavement cells (Fig.2, C). A pavement cell is either directly differentiated from a protodermal cell or from MMC-descendant (stomatal lineage), with the latter making a larger contribution for the number of pavement cells (Geisler et al. 2000). Those pavement cells derived from the former are larger than the latter. Thus it was possible that EPF2-overexpression blocks entry into the stomatal lineage. The epf2 loss-of-function mutant has increased number of guard and non-guard (pavement) cells. We next examined the phenotypes of a loss-of-function mutant of EPF2. In the epf2 mutant, the stomatal density was increased but most stomata were separated by at least one non-guard cell in the epidermis of cotyledons (Fig. 5A and C) and the first rosette leaves (Fig. S1B). As was previously reported, epf1 formed stomata that were in contact (Figs. 5B and S1C). Another important feature of epf2 is an increase in the density of pavement cells, which does not occur in epf1 (Fig. 5E, F). This increase did not occur uniformly but instead is increased in sectors around stomata in cotyledons (Figs. 5C) and in true leaves (Fig. S1C). The epf2 phenotype was complemented by introduction of a genomic fragment containing the EPF2 gene (Fig. 6), confirming that the observed phenotype of epf2 was due to the loss-of-function of EPF2. The phenotypes of EPF1 and EPF2 overexpression, as well as the phenotypes of EPF1 and EPF2 loss-of function mutants were both different, indicating that EPF1 and EPF2 have different functions. To further explore the functions of EPF1 and EPF2, we examined the epidermis of the epf1;epf2 double mutant. The degree of stomatal clustering of epf1;epf2 remained similar to that of epf1 (Fig. 5G), and the effects of epf1 and epf2 mutations on 6

7 stomatal density was additive (Fig.5 B-F). These results are consistent with EPF1 and EPF2 having different functions during stomatal development. EPF2 inhibits protodermal cells from being MMCs We then made a working hypothesis that EPF2 begins to be expressed when a protodermal cell acquires the potential to undergo asymmetric cell division, and functions non-cell autonomously to inhibit neighboring cells from acquiring a competency to undergo an asymmetric cell division. To test this hypothesis, we examined the effect of EPF2-overexpression on its own expression. EPF2-overexpressors had decreased number of cells that express EPF2-GFP, but the signal intensity of EPF2-GFP within a given cell was unchanged (Fig.7). This indicates that EPF2 inhibits protodermal cells from acquiring the MMC character. We further tested expression hierarchy between EPF1, EPF2 and TMM. While overexpression of EPF1 also inhibits the formation of stomata, it did not reduce the number of cells expressing EPF2-GFP. This indicates that, despite that the same CaMV35S promoter was used for the overexpression studies, EPF1 functions later than EPF2 (Fig.7). Overexpression of either EPF1 or EPF2 decreased the number of cells expressing EPF1-GFP, which normally begins expression after the meristemoids are formed (Fig. 8). Next we examined the effects of EPF1 and EPF2 overexpression on TMM, which is normally expressed in the proliferative cells of the epidermis (Nadeau and Sack 2002). Overexpression of neither EPF1 nor EPF2 affected TMM-GFP expression (Fig. 9). It is possible that TMM is expressed in young cells regardless of whether they have asymmetric cell division competency. Promoter swapping revealed functional differences between the EPF1 and EPF2 products The overexpression phenotypes suggest that the coding sequences of EPF1 and EPF2 are largely responsible for their specific functions. However, we were unable to test whether EPF2 can substitute the function of EPF1, because the overexpression of EPF2 inhibits cells from entering the stomatal-lineage, resulting in the epidermis devoid of cells that can respond to EPF1. To overcome this problem, we performed a promoter/coding-region swapping-experiment. EPF1 promoter-epf2 partially suppressed the stomatal-clustering phenotype of epf1 (Fig.10A). By contrast, EPF2 promoter-epf1 did not suppress increased 7

8 stomatal density phenotype of epf2 (Fig. 10B). These results indicate that EPF2 can, in part, substitute for EPF1, but EPF1 cannot substitute for EPF2. EPF2 requires TMM, ER/ERL1/ERL2, and YDA to inhibit stomatal formation. We previously reported that TMM receptor-like protein, at least one of ER, ERL1 and ERL2, and YDA, but not SDD1 are required for the function of EPF1 (Hara et al. 2007). Here we tested whether EPF2 also requires these genes to suppress stomatal differentiation. As shown in Fig. 11, EPF2 overexpression in tmm, er;erl1;erl2, as well as in yda failed to decrease the density of stomata, indicating that EPF2 requires TMM, at least one of ER, ERL1 and ERL2, and YDA. However, similar to EPF1, the overexpression of EPF2 reduced the stomatal density in sdd1, indicating that EPF2 does not require SDD1 (Fig.11). Genetic interactions were also examined by making multiple loss-of-function mutants. The stomatal density of epf1;epf2;tmm was similar to tmm, consistent to the idea that EPF1 and EPF2 function upstream of TMM. However, surprisingly, stomatal densities of tmm or epf1;epf2;tmm in cotyledons (Fig. 12A) and true leaves (Fig. S2) were lower than that of epf1;epf2. A possible explanation would be that TMM perceives an unidentified positive signal. The er;erl1;erl2;epf1;epf2 pentuple loss-of-function mutants exhibited stomatal density similar to er;erl1;erl2, also consistent to the idea that EPF1 and EPF2 functions upstream of ER, ERL1 and ERL2 (Fig. 12B). Discussion Formation of proper numbers of cells in diverse tissues is important for the proper development of multicellular-organisms. In plants, CLV3 is a relatively well studied-molecule that mediates a negative feedback-loop for proliferation of cells in the shoot apical meristem (Clark et al. 1996; Fletcher et al. 1999). The mature form of CLV3 is a 12-amino acid peptide, and belongs to the CLE family signaling peptides (Ito et al. 2006; Kondo et al. 2006). We here report that a putative small secreted-peptide EPF2 limits the final density of epidermal cells by acting as a key component of a negative feedback loop that limits the number of cells entering the stomatal lineage. Very recently, it was reported that EPF2 controls asymmetric cell divisions during stomatal development and controls stomatal density (Hunt and Gray 2009). We further delimited the major action-point of EPF2 in the stomatal lineage: EPF2 does not directly act on asymmetric cell division, but it inhibits cells from acquiring the MMC fate. We demonstrate this by quantitative 8

9 examination of the numbers of both guard and non-guard pavement cells (Fig.5), by overexpressing EPF2 in an EPF2-GFP expressing line (Fig.7), and by performing promoter-swapping between EPF1 and EPF2 (Fig.10). Hunt and Gray (2009) showed that epf2, tmm, and the epf2;tmm double mutant have similar stomatal densities. We here clearly demonstrate that TMM and at least one of ER, ERL1, and ERL2, and YDA are necessary for EPF2 to function, by examining phenotypes of EPF2-overexpressors in corresponding mutants (Fig. 11), and by making the epf1;epf2;tmm triple and the epf1;epf2;er;erl1;erl2 quintuple mutants (Figs. 12 and S2). Every EPF-family member has a secretory signal sequence at the amino terminus, and has 6-Cys residues at common positions at the carboxy half. EPF1, EPF2 and EPFL7 have additional two Cys residues at common positions. Functions of the family members other than EPF1 and EPF2 are not known. The ectopic overexpression of EPFL4 (At4g14723) and EPFL5 (At3g22620) also led to decreased numbers of stomata (data not shown). Our results suggest that these two EPFLs do not normally act in stomatal development, they possibly regulate some other developmental processes through similar signaling systems. It has become evident that plants possess many genes for Cys-rich small secretory peptides with potential signaling functions. For instance, recently reported LURE Cys-rich peptides of Torenia are the pollen-tube guidance molecules secreted from the synergid cells (Okuda et al. 2009). The SCR proteins of Brassica are the pollen determinants of sporophytic self-incompatibility, and they belong to another class of Cys-rich peptides (Schopfer et al. 1999). Considering the large number of genes encoding small Cys-rich secretory peptides (Okuda et al. 2009; Silverstein et al. 2007), it is possible that there are still many unknown intercellular signaling molecules. EPF1 and EPF2 are expressed in overlapping, but different cell types within the stomatal lineage. EPF2 initiates its expression in a fraction of protodermal cells at an early stage of epidermal development. These cells are likely to be MMCs. This was underpinned by the finding that EPF2 expression requires SPCH, which is required for the entry into the stomatal lineage. It is also possible that mutual lateral inhibition between protodermal cells that transiently express EPF2 results in a selection of proper numbers of cells that enter the stomatal lineage and stably express EPF2. EPF2-overexpression inhibited formation of EPF2-GFP positive cells. By contrast, epf2 loss-of-function increased the number of stomata and surrounding small pavement cells, which are most likely made in the stomatal lineage. From these results, we propose the following model (Fig.13). The putative secreted 9

10 protein EPF2 is produced in MMCs and early descendants, and diffuses to surrounding cells. With the increase in the density of MMCs and early descendants, the apoplasmic concentration of EPF2 increases. EPF2 of over a certain concentration inhibits protodermal cells from being the MMC, thus limiting the density of stomatal lineage. Because stomatal lineage makes a large contribution to the number of both guard and non-guard cells, this feedback loop is critical for regulation of epidermal cell density. Our study highlighted specific, non-redundant functions of EPF1 and EPF2: EPF1 enforces the one-cell spacing rule and EPF2 restricts population of cells from acquiring the stomatal-lineage fate. Overexpression of either EPF1 or EPF2 inhibited formation of stomata, but they did so by acting on different developmental processes. When the same CaMV 35S constitutive promoter was used for the overexpression study, EPF2 inhibited formation of EPF2-GFP positive cells, and EPF1 acted at a later stage. Also, EPF2-overexpression decreased the number of pavement cells, but EPF1-overexpression increased the number of pavement cells. Nevertheless, the effects of both EPF1 and EPF2 require the presence of TMM and ER-family receptor kinases. This indicates that yet another unknown factor is involved in the recognition-specificity. It is also possible that TMM and ER-family proteins form receptor complexes, and combination of receptor components are different in different cell types. Because three ER-family genes have overlapping yet distinct roles in stomatal development (Shpak et al., 2005), they might contribute to the specificity. EPF2 also requires YDA, suggesting that the function of EPF2 is to inhibit the entry into stomatal lineage and is mediated by the MAP kinase cascade. It has recently been reported that MAP kinase-mediated phosphorylation of SPCH destabilizes SPCH (Lampard et al. 2008). It would be interesting to know whether the effect of EPF2 is mediated by the destabilization of SPCH. It would also be interesting to know whether peptide-control of the MAP kinase cascade is widely used for the control of cell proliferation and / or cell specificity, in response to positional or environmental cues. Materials and Methods Plant materials and growth conditions Arabidopsis ecotype Columbia was used in all experiments. Plants were grown in plates with GM medium (MS salts, 1% sucrose,1/100 volume of 2.5% MES-KOH at ph 5.7, 0.3% Phytagel) under continuous light at 22 C. Mutants used in this study are as follows: yda-y295 (Bergmann et al. 2004), er, erl1-2, erl2-1 (Shpak et al. 2005), tmm 10

11 (SALK_011959) (Hara et al. 2007), sdd1(gabi, 627-D04)(Hara et al. 2007), epf1-1 (SALK_137549)(Hara et al. 2007), and epf2-3 (SALK_047918). In the epf2-3 mutant, T-DNA was inserted in an exon, with the following junction: T-DNA LB / GCGTCATCTACAGATGCA. Homozygous T-DNA insertion mutants of epf1 (SALK_137549) and epf2 were crossed with er-105 erl1-2/+ erl2-1 mutants and the offspring selfed. The ER (At2g26330) and EPF1 (At2g20875) loci are tightly linked, which matched the ratio found in this cross. PCR-based analysis was performed to confirm the correct genotypes. Sets of primers and their sequences for diagnosis of ER, ERL1, ERL2, er-105, erl1-2, and erl2-1 are available in (Shpak et al. 2004). The wild-type allele of EPF1 was detected by the primer pair SML (5'-TGACTCTCTCCTCTTCTCTA-3') and SML rc (5'-CGGAATTCACGAAGGTGAGATGATATGGTTGA-3'). The T-DNA insertion allele of epf1 was detected by the primer pair Lba1 (5'-TGGTTCACGTAGTGGGCCATCG-3') and SML rc. The wild-type allele of EPF2 was detected by the primer pair SML (5'-CACTAAGTCCGTCACATCAA-3') and SML rc (5'-CGGAATTCCGGTATGATGGAGATGGCTT-3'). The T-DNA insertion allele of epf2 was detected by the primer pair Lba1 and SML rc. Microscopy and Quantitative analysis of epidermis GFP image was acquired with a confocal microscope. The cell margin was counter-stained with FM4-64 as described before (Hara et al. 2007). For quantitative analysis, abaxial sides of cotyledons of 15-days-old plants or primary leaves of 20-days-old plants were examined according to a previous paper (Hara et al. 2007), with a modification that peeled epidermis was stained with safranin for cell counting under a microscope; except for Fig. 12B. For Fig. 12B, 20-days-old plants were fixed in 90% ethanol/ 10% acetic acid and then cleared in a chloral hydrate solution (chloral hydrate: H20: glycerol =8:1:1), and examined under a differential interference microscope. Plasmids For overexpression of EPF2, a genomic sequence was PCR-amplified with primers K150: 5'- aaaatgacgaagtttgtacgcaagtatatg-3' (lowercase letters do not match the genome sequence) and K151: 5'- CAAAACTGATATTTTAATCACAGACGTCA, and 11

12 blunt-end cloned into ptk014, which gives kanamycin resistance to plants, or into ptk016, which gives BASTA resistance to plants. ptk014 and ptk016 carry duplicated 35S enhancers and the omega leader sequence, which had been derived from pbe2113 (Mitsuhara et al. 1996). For construction of EPF2-GFP, a 2610 bps of the promoter region of EPF2 was amplified with primers 5'-TGGTCTAGAGAACAAGTGAAGTAAGCCAA-3' and 5'-ccgctcgagGTTTATAATCTTTTTTTTTAACAAGAAGAAAC-3' (lowercase letters include the XhoI site), and cloned into prk2, at a region upstream of the GFP(S65T) with an endoplasmic reticulum localization signal. For complementation of epf2, a DNA region containing 1737 bps of the 5' region, entire coding region, and 1133 bps of the 3' region of the EPF2 gene was PCR-amplified with the primers 5'- ccgctcgagctatttgacatattttcttttgtcatattt-3' and 5'- ggggtaccccaattttaggcaggttaattatctcaat-3' (the lowercase letter regions contain restriction enzyme recognition sites), and cloned in of pgwb1, which is a binary vector with Hygromycin-selectable marker in plants. For construction of TMM-GFP, a promoter region of TMM was amplified with the use of primers K136 (5'-TGGGATTATTCCATGTGCAATTTTGTTA-3') and K194 (5'-ccgctcgagTTCTTAGTTGTTGTTGTTGTGTGAATGC-3') and cloned in prk2. For the promoter-swapping experiment, the promoters and coding regions of EPF1 and EPF2 were amplified from genomic DNA using the following primer pairs: EPF1 promoter : primer 3703 and primer 4090 EPF1 coding region : primer 3716 and primer 3674 EPF2 promoter : primer 3704 and primer 4091 EPF2 coding region : primer 3714 and primer 3672 where the primer sequences are as follows: primer 3703: 5'-ccgctcgagACGACGATGTCCTCTTTTGTCTTTGAGAA-3' primer 4090: 5'-cgggatccGATATATTATCGCAAGTGGTAAAAGT-3' primer 3716: 5'-cgggatccATCATGAAGTCTCTTCTTCTCCTTG-3' primer 3674: 5'-cgggatccAAGGAAAACAAAACGGTTGAATGCATAGA-3' primer 3704: 5'-ccgctcgagTGGTCTAGAGAACAAGTGAAGTAAGCCAA-3' primer 4091: 5'-cgggatccGTTTATAATCTTTTTTTTTAACAAGAAGAAAC-3' primer 3714: 5'-cgggatccAACATGACGAAGTTTGTACGCAAGT-3' primer '-gactagtCAA AACTGATATTTTAATCACAGACGTCA-3' 12

13 Then a promoter region and a coding region were cloned in a binary vector in four combinations. RT-PCR Reverse transcription polymerase chain reaction (RT-PCR) was performed following Hara et al. (2007), using the following primers. For EPF2, primers K150 and K151. For 18S-RNA genes (duplicated genes At3g41768 and At2g01010), primers No.2095 (5'-CTGGTTGATCCTGCCAGTAGTCATATGCT-3') and No (5'-GACAGGTATCGACAATGATCCTTCCGCA-3'). In situ RNA hybridization Whole-mount in situ RNA hybridization was performed using digoxigenin-labeled crna probes and alkaline phosphatase-labeled anti-digoxigenin, following the procedures outlined in Hejatko et al. (2006). For production of crna, cdna for EPF2 was first PCR-amplified using primer pairs No / No (for production of antisense RNA) and No / No (for production of sense RNA). No. 4771: 5'- AATACTTCACACACAACACATAACACACGT-3' No. 4800: 5'- taatacgactcactatagggagttaatatcccaacaaattgtacaatag-3' No. 4799: 5'-taatacgactcactatagggagCCTCAACTATATATACACATATCTCAC-3' No. 4773: 5'-GGGCCAATAGCATTTAATTTAATATCCCAAC-3' Where lowercase letters indicate the T7 promoter sequence, which was used to produce digoxigenin-labeled crna. Acknowledgements We thank Amanda Rychel for editing. This work was supported by the grant "KAKENHI" (numbers and for T. Kakimoto), and NSF (IOB and IOB ) to K.U.T. K.M.P. was supported in part by the University of Washington Mary Gates Undergraduate Research Fellowship and NSF REU Supplement to IOB K.U.T. is a PREST investigator of JST. Sequence information for EPFL3 and EPFL7 have been deposited to DDBJ with accession numbers AB and AB499313, respectively. 13

14 References Berger, D. and Altmann, T. (2000) A subtilisin-like serine protease involved in the regulation of stomatal density and distribution in Arabidopsis thaliana. Genes Dev 14: Bergmann, D.C. (2004) Integrating signals in stomatal development. Curr Opin Plant Biol 7: Bergmann, D.C., Lukowitz, W. and Somerville, C.R. (2004) Stomatal development and pattern controlled by a MAPKK kinase. Science 304: Bergmann, D.C. and Sack, F.D. (2007) Stomatal development. Annu Rev Plant Biol 58: Clark, S.E., Jacobsen, S.E., Levin, J.Z. and Meyerowitz, E.M. (1996) The CLAVATA and SHOOT MERISTEMLESS loci competitively regulate meristem activity in Arabidopsis. Development 122: Fletcher, J.C., Brand, U., Running, M.P., Simon, R. and Meyerowitz, E.M. (1999) Signaling of cell fate decisions by CLAVATA3 in Arabidopsis shoot meristems. Science 283: Geisler, M., Nadeau, J. and Sack, F.D. (2000) Oriented asymmetric divisions that generate the stomatal spacing pattern in arabidopsis are disrupted by the too many mouths mutation. Plant Cell 12: Hara, K., Kajita, R., Torii, K.U., Bergmann, D.C. and Kakimoto, T. (2007) The secretory peptide gene EPF1 enforces the stomatal one-cell-spacing rule. Genes Dev 21: Hejatko, J., Blilou, I., Brewer, P.B., Friml, J., Scheres, B. and Benkova, E. (2006) In situ hybridization technique for mrna detection in whole mount Arabidopsis samples. Nat Protoc 1: Hunt, L. and Gray, J.E. (2009) The signaling peptide EPF2 controls asymmetric cell divisions during stomatal development. Curr Biol doi: /j.cub Ito, Y., Nakanomyo, I., Motose, H., Iwamoto, K., Sawa, S., Dohmae, N., et al. (2006) Dodeca-CLE peptides as suppressors of plant stem cell differentiation. Science 313: Kanaoka, M.M., Pillitteri, L.J., Fujii, H., Yoshida, Y., Bogenschutz, N.L., Takabayashi, J., et al. (2008) SCREAM/ICE1 and SCREAM2 specify three cell-state transitional steps leading to arabidopsis stomatal differentiation. Plant Cell 20: Kondo, T., Sawa, S., Kinoshita, A., Mizuno, S., Kakimoto, T., Fukuda, H., et al. (2006) A 14

15 plant peptide encoded by CLV3 identified by in situ MALDI-TOF MS analysis. Science 313: Lampard, G.R., Macalister, C.A. and Bergmann, D.C. (2008) Arabidopsis stomatal initiation is controlled by MAPK-mediated regulation of the bhlh SPEECHLESS. Science 322: MacAlister, C.A., Ohashi-Ito, K. and Bergmann, D.C. (2007) Transcription factor control of asymmetric cell divisions that establish the stomatal lineage. Nature 445: Mitsuhara, I., Ugaki, M., Hirochika, H., Ohshima, M., Murakami, T., Gotoh, Y., et al. (1996) Efficient promoter cassettes for enhanced expression of foreign genes in dicotyledonous and monocotyledonous plants. Plant Cell Physiol 37: Nadeau, J.A. (2009) Stomatal development: new signals and fate determinants. Curr Opin Plant Biol 12: Nadeau, J.A. and Sack, F.D. (2002) Control of stomatal distribution on the Arabidopsis leaf surface. Science 296: Okuda, S., Tsutsui, H., Shiina, K., Sprunck, S., Takeuchi, H., Yui, R., et al. (2009) Defensin-like polypeptide LUREs are pollen tube attractants secreted from synergid cells. Nature 458: Pillitteri, L.J., Sloan, D.B., Bogenschutz, N.L. and Torii, K.U. (2007) Termination of asymmetric cell division and differentiation of stomata. Nature 445: Sachs, T. (1991) Pattern formation in plant tissues. Cambridge University Press. Schopfer, C.R., Nasrallah, M.E. and Nasrallah, J.B. (1999) The male determinant of self-incompatibility in Brassica. Science 286: Shpak, E.D., Berthiaume, C.T., Hill, E.J. and Torii, K.U. (2004) Synergistic interaction of three ERECTA-family receptor-like kinases controls Arabidopsis organ growth and flower development by promoting cell proliferation. Development 131: Shpak, E.D., McAbee, J.M., Pillitteri, L.J. and Torii, K.U. (2005) Stomatal patterning and differentiation by synergistic interactions of receptor kinases. Science 309: Silverstein, K.A., Moskal, W.A., Jr., Wu, H.C., Underwood, B.A., Graham, M.A., Town, C.D., et al. (2007) Small cysteine-rich peptides resembling antimicrobial peptides have been under-predicted in plants. Plant J 51: Wang, H., Ngwenyama, N., Liu, Y., Walker, J.C. and Zhang, S. (2007) Stomatal development and patterning are regulated by environmentally responsive mitogen-activated protein kinases in Arabidopsis. Plant Cell 19:

16 Legends to Figures Fig.1. Alignment of the carboxy terminal regions of the EPF family proteins. EPFL3 and EPFL7 were previously unpredicted. The numbers of amino acid residues preceding and interrupting the aligned sequences are indicated in parentheses. Conserved Cys residues are marked with a yellow color, and conserved Gly, Ser, and Pro are marked with green, pink and blue colors. Fig. 2. Effects of overexpression of EPF1 and EPF2 on epidermal development. An abaxial side of the primary leaf of a transformant of the control vector (A), an EPF1-overexpressor (B), or an EPF2-overexpressor (C). Bar, 100 µm. (D) The density of guard cells (light-green background) and distribution of sizes of non-guard cells (mean ±SD) of transformants of the control vector (gray), EPF1-overexpressors (blue), and EPF2-overexpressors (red). EPF1- or EPF2-overexpressors with clear decrease in stomatal density were selected for this analysis. Fig. 3. Expression patterns of EPF2. (A, B) EPF2-GFP expression in abaxial surface of a leaf of (A) about 0.2 mm long and of (B) about 2mm. Note that meristemoids are not present in (A), and cells of diverse developmental stages of the stomatal lineage are present in (B). (C) RT-PCR analysis for EPF2 expression. Ro, roots. RL, rosette leaves. CL, cauline leaves. St, stem. FB, floral buds. Fl, flowers. Si, siliques. the 18S rrna gene was used as a control. (D, E) in situ RNA hybridization with EPF2 antisense probe (D) or sense probe (E). Bars, 100µm. Fig.4 Effects of spch mutation on expression of EPF1-GFP (A, D, G, J), EPF2-GFP (B, E, H, K), or TMM-GFP (C, F, I, L) in plants segregating as wild-type phenotype (+/- or +/+ for SPCH) (A-F) or in homozygous spch (-/-) (G-L). Pictures in D-F and J-L are close-up views in A-C and G-I, respectively. Bars, 100µm. Fig.5 Phenotypes of epf1, epf2, and epf1;epf2. (A-D) Abaxial sides of cotyledons. Bar, 100 µm. (A) WT. (B) epf1. (C) epf2. (D) epf1;epf2. (E and F) Effects of the epf1, epf2, and epf1;epf2 double mutations on the density of guard cells and non-guard cells (pavement 16

17 cells) in cotyledons (E) and primary leaves (F). (G) Effects of the epf1, epf2, and epf1;epf2 double mutations on stomatal clustering in primary leaves (mean±sd). Fig. 6. Complementation of epf2-3 by the EPF2 gene. Total epidermal cell densities of cotyledons are shown (mean±sd). Fig.7. Effects of overexpression of EPF1 (middle) or EPF2 (right) on the formation of EPF2-GFP positive cells. Bottom pictures are close-up views of the encircled regions. White arrows, stomata. The green arrow, EPF2-GFP positive cell in an EPF2-overexpressor. Fig.8. Effects of overexpression of EPF1 (middle) or EPF2 (right) on the formation of EPF1-GFP positive cells. Bottom pictures are close-up views of the encircled regions. White arrows, stomata. Fig.9. Effects of overexpression of EPF1 (middle) or EPF2 (right) on the formation of TMM-GFP positive cells. Bottom pictures are close-up views of the encircled regions. White arrows, stomata. Fig.10 Swapping between promoter / coding region. Each chimeric gene was introduced into epf1-1 (A) or epf2-3 (B), and primary leaves of 10 independent T1 plants were used for quantification. (A) Percentage stomata present in each cluster size in the epf1 background (mean±sd). (B) Stomatal density (mean±sd) in the epf2 background. Fig. 11. Effects of EPF2 overexpression on stomatal densities in different genetic backgrounds. Spots represent stomatal densities of independent T1 plants that had been transformed with the control vector (openspots) and an EPF2-overexpressing construct (closed spots). Abaxial sides of cotyledons of 15-d-old plants were examined. Fig.12. Genetic interactions between the epf1;epf2 double mutation and tmm (A), and the epf;epf2 double mutation and the er;erl1;erl2 triple mutation. Stomatal densities of cotyledons are shown (mean±sd). Note: Stomatal densities of the same genotypes differ in (A) and (B). This difference might have been caused by the following factors. Experiments 17

18 for (A) and all other experiments and the experiment for (B) were performed in different laboratories, and used different sample preparation method (see Materials and Methods). Fig.13. A model for the EPF2-mediated negative feedback loop for the control of the MMC-derived lineage. A protodermal cell may directly differentiate into a pavement cell (1) or enter the stomatal lineage (2). EPF2 is expressed in MMC and its early descendants. EPF2-expressing cells are shown with red colors, with darker red the higher expression. EPF2 diffuses to surrounding cells and inhibits protodermal cells from acquiring asymmetric cell division competency (3). Supplementary Figure Legends Fig. S1. Phenotypes of epf1 and epf2 in abaxial epidermis of primary leaves of 20-days-old plants. (A) Wild-type. (B) epf1-1. (C) epf2-3. Bar, 100µm. Fig.S2. Genetic interactions between epf1;epf2 and tmm. Stomatal densities of abaxial sides of the primary leaves are shown with (mean±sd). Fig. 1 18

19 Fig. 2 19

20 Fig. 3 20

21 Fig. 4 21

22 Fig. 5 22

23 Fig. 6 Fig. 7 23

24 Fig. 8 Fig. 9 24

25 Fig

26 Fig. 11 Fig

27 Fig