Shouting out loud: signaling modules in the regulation of stomatal development

Abstract Stomata are small pores on the surface of land plants that facilitate gas exchange for photosynthesis while minimizing water loss. The function of stomata is pivotal for plant growth and survival. Intensive research on the model plant Arabidopsis (Arabidopsis thaliana) has discovered key peptide signaling pathways, transcription factors, and polarity components that together drive proper stomatal development and patterning. In this review, we focus on recent findings that have revealed co-option of peptide-receptor kinase signaling modules—utilized for diverse developmental processes and immune response. We further discuss an emerging connection between extrinsic signaling and intrinsic polarity modules. These findings have further enlightened our understanding of this fascinating developmental process.


Introduction
Stomata are turgor-driven microscopic valves in the epidermis of aerial regions of land plants. The controlled opening and closing of the stomatal pores is essential to the regulation of gas exchange and water loss by the plant (Chaerle et al., 2005). In addition, maintaining proper stomatal density, distribution, and development are pivotal for plant survival. In Arabidopsis, a dicot plant, stomatal development occurs through the initiation of an entry division of a subset of undifferentiated protodermal cells called meristemoid mother cells (MMCs). The asymmetric cell division (ACD) of the MMC generates a small, triangular-shaped meristemoid (Ms), and its sister cell, a stomatal lineage ground cell (SLGC). The Ms possess a stem cell-like character and can undergo rounds of ACDs in an inward spiralling manner. Late Ms lose their stem cell-like potential and differentiate into a round guard mother cell (GMC). Finally, the GMC will further divide symmetrically to generate a pair of guard cells (GCs) that surround a pore (Nadeau and Sack, 2002;Bergmann and Sack, 2007;Pillitteri and Torii, 2012).
To optimize stomatal distribution during leaf expansion, additional spacing divisions occur away from the existing stomata. This process follows in accordance with the "onecell spacing rule," in which stomata should not directly contact each other. Consequently, young SLGCs are allowed to re-establish MMC identity, thus undergoing asymmetric spacing division to form secondary stomata (Nadeau and Sack, 2002). The rest of the protodermal cells, which are not destined to become stomata, differentiate into pavement cells, which make up the rest of the epidermis, acquiring the characteristic puzzle-like pattern (Figure 1).
The genetic control of stomatal development requires the consecutive activity of three basic helix-loop-helix (bHLH) transcription factors (TFs), SPEECHLESS (SPCH), MUTE, and FAMA, which drive stomatal initiation, proliferation, and differentiation, respectively (Figure 1; Ohashi-Ito and Bergmann, 2006;MacAlister et al., 2007;Pillitteri et al., 2007). These TFs act in coordination with two additional, redundant bHLH TFs-ICE1/SCREAM (SCRM) and SCRM2which are expressed throughout the stomatal lineage, forming heterodimer TF complexes (Kanaoka et al., 2008;Pillitteri and Torii, 2012). Receptor-like kinases (RLKs) coupled with a mitogen-activated protein kinase (MAPK) signaling cascade, mediated by YODA (YDA, MAPKKK)-MKK4/5 (MAPKKs)-MPK3/6 (MAPKs), negatively regulate these core bHLH heterodimers to enforce proper stomatal patterns (Müller et al. 2010;Meng et al., 2012). In return, bHLH heterodimers induce expression of some of the key signaling components, thereby forming a negative-feedback loop (Lau et al., 2014;Horst et al., 2015;Qi and Torii, 2018). This review aims to summarize the complex signaling pathways that orchestrate stomatal development, synthesize how peptide ligands and receptors are co-opted for different developmental programs, and explore an emerging connection between extrinsic signaling and intrinsic polarity modules in the process (Figure 1). Please see Table 1 for quick guides of key regulators of stomatal development discussed in this review article.

RLK signaling complexes ensure stomata development inhibition
Plants possess a battery of RLKs. Those RLKs with an extracellular leucine-rich repeat (LRR) domain play key roles in numerous aspects of plant development and immune response (Albert et al., 2020;Gou and Li, 2020). Stomatal development is not an exception. Three members of the ERECTA-family (ERf) LRR-RLK, namely ERECTA (ER), ERECTA-LIKE1 (ERL1), and ERL2, regulate plant organ growth, inflorescence elongation, vascular patterning, and leaf shape (Figure 1; Torii et al., 1996;Shpak et al., 2003Shpak et al., , 2004Uchida et al., 2012;Bemis et al., 2013;Chen et al., 2013;Ikematsu et al., 2017). The same ERf is in addition crucial for stomata patterning and differentiation. ER, which is highly expressed in protodermal cells, primarily acts to repress asymmetric entry division of MMCs, reflected by the increased number of ACDs in the er single mutant. Meanwhile, both ERL1 and ERL2 subsequently function to prevent M differentiation. In the erl1 erl2 double mutant, Ms prematurely differentiate into GMCs, underlining their important role in maintaining the amplifying potential of Ms (Shpak et al., 2005). Severe stomatal clustering is observed when all ERf members lose function (er erl1 erl2; Figure 2B; Shpak et al., 2005).
An LRR receptor-like protein (LRR-RLP), TOO MANY MOUTHS (TMM), lacks the characteristic intracellular kinase domain of LRR-RLKs (Yang and Sack, 1995;Bhave et al., 2009). tmm mutants show altered stomatal patterning reminiscent of er erl1 erl2 triple mutants, i.e. clustered and numerous stomata in the epidermis of cotyledons and leaves ( Figure 2B; Yang and Sack, 1995). However, in contrast to er erl erl2, hypocotyls and stems of tmm are devoid of stomata, indicative of an additional complex regulatory function of TMM in the formation of stomata (Geisler et al., 1998). Further detailed genetic analyses reveal a role for TMM activity in precursor of stomatal lineage cell fate and progression, in an organ-and region-specific manner (Bhave et al., 2009).
Taking the phenotypes of both er-family members and tmm into consideration, one might wonder about their genetic and molecular relationships. Initial genetic evidence has hinted that these proteins contribute in combination to determine stomatal-lineage cell fate, possibly through dimerization (Shpak et al., 2005). Indeed, recent structural and biochemical analyses show that ERf members are able to form constitutive heterodimer complexes with TMM, thereby pre-forming a ligand-binding pocket (Lin et al., 2017). Interestingly, depending on the specific ligands, the heterodimerization of ERf and TMM will be unfavored, and this ligand-based discrimination of receptor heterodimerization is likely the molecular basis of complex TMM function (Figures 1 and 2).
A central question, then, is identifying ligands perceived by ERf and TMM. A family of secreted cysteine-rich peptides, EPIDERMAL PATTERNING FACTOR (EPF)/EPF-LIKE (EPFL), has been identified as a regulator of stomatal development (Hara et al., 2007(Hara et al., , 2009Hunt and Gray, 2009;Kondo et al., 2009;Hunt et al., 2010;Sugano et al., 2010;Lee et al., 2015;Lin et al., 2017). The predicted mature EPF/EPFL peptides possess six or eight cysteines that form disulfide bonds, constructing a loop region and a functionally crucial scaffold. Intriguingly, structural analyses by nuclear magnetic resonance combined with loop region swapping experiments ADVANCES • Since the discovery of ERf and TMM as a co-receptor controlling stomatal development, various upstream and downstream components have been identified, aiding in the formation of specific peptide-receptor signaling modules and signal transduction. • These peptide-receptor signaling modules can act in both paracrine and autocrine manners, dependent on the developmental stage of the respective stomatal cell. • The peptide-receptor signaling modules have been co-opted to regulate arrays of developmental programs and intersect with immunity and environmental response. • The interplay of extrinsic signals and intrinsic polarity cues, that provide spatial information during cell division, are still not fully understood. However, the discovery of novel polarity proteins and regulatory mechanisms are now unravelling a molecular intersection of different protein modules during stomatal development.
reveal that the variable loop region determines the antagonistic activity of EPF2 with EPFL9/STOMAGEN (Ohki et al., 2011). EPF1 and EPF2 are both secreted from stomatal precursor cells and produce an inhibitory effect on stomatal lineage formation, while mesophyll cell-derived EPFL9/STOMAGEN peptides promote stomatal lineage proliferation through a competitive binding on selected ERf members (Hunt et al., 2010;Lee et al., 2015). Although EPF1 and EPF2 both negatively regulate stomatal development, careful genetic analyses have uncovered a more distinctive role for each signaling peptide (Hara et al., 2009;Hunt and Gray, 2009). Genetic studies have identified EPF2 as an inhibitor of protodermal cells entering the stomatal lineage. Consistently, EPF2 is expressed in early stomatal lineage precursors, MMCs, and early Ms (Hara et al., 2009). In early developmental stages, the epf2 mutant displays numerous small, undifferentiated epidermal cells, resulting in more GCs and smaller pavement cells upon their differentiation. This phenotype is similar to those observed in a dominant negative mutant form of ER or SPCH overexpression, in which uninhibited asymmetric entry divisions occur during early stages of the stomatal lineage ( Figure 2; Shpak et al., 2003;MacAlister et al., 2007;Pillitteri et al., 2007;Hara et al., 2009;Lee et al., 2012). Conversely, ectopic expression of EPF2 or application of a bioactive EPF2 peptide results in the development of an epidermis composed of only pavement cells, identical to the spch mutant (Figure 2, B and D; Figure 1 Peptide-receptor signaling throughout stomata development. Overview of the different receptor peptide signaling modules and their main downstream components regulating cell-state transitional steps within the stomatal cell lineage. Members of the EPF family, EPF1, EPF2, and EPFL9/STOMAGEN, are secreted from stomatal precursors and underlying mesophyll cells into the apoplast and competitively bind to three cell surface LRR-RLKs, ERECTA (ER), ER-LIKE 1 (ERL1), and ERL2. Together with the LRR receptor protein, TMM and BRI1-ASSOCIATED RECEPTOR KINASE/SOMATIC EMBRYOGENESIS RECEPTOR KINASE (BAK/SERK) family RLKs, they form a ternary receptor complex. This complex activates the downstream MAPK cascade (arrow, P + ) that is composed of YODA (YDA) MAPKKK, two redundant MAPKKs (MKK4 and MKK5) and two, redundant MAPKs (MPK3 and MPK6) possibly through two functional redundant members of the BSK family (question marks; cytoplasm). The activated MAPK cascade in return controls the consecutive activities of three bHLH TFs and their partner bHLH proteins SCREAM (SCRM) and SCRM2 (nucleus). While SPEECHLESS (SPCH) and MUTE are negatively regulated through TF inhibition by the MAPK cascade, FAMA is activated by the latter to inhibit further symmetric cell division and promote cell differentiation. Each TF pair is specifically expressed during the timely progression of stomata development (color indicated). In the model plant Arabidopsis, stomata development occurs by the initiation of an asymmetric division of protodermal cells called MMCs (blue). The ACD generates a small meristemoid (purple) and a large SLGC (light grey). The meristemoid can undergo one to three rounds of asymmetric divisions in an inward-spiral manner to produce additional amplicons surrounded by SLGCs. Stomata development ends when the late meristemoid differentiates into a GMC (light green), which then will further divide symmetrically to generate a pair of GCs (dark green).  Hara et al., 2007Hara et al., , 2009Dong et al., 2009;Ohki et al., 2011;Lee et al., 2012;Qi et al., 2017EPF2 At1g34245 MacAlister et al., 2007Hara et al., 2009;Ohki et al., 2011;Lee et al., 2012;Lau et al., 2014;Horst et al., 2015;Lin et al., 2017 EPFL9/STOMAGEN At4g12970 Hunt andGray, 2009;Kondo et al., 2009Kondo et al., , 2010Hunt et al., 2010;Sugano et al., 2010;Ohki et al., 2011;Lee et al., 2012 At4g37810 Tameshige et al., 2016;Kawamoto et al. 2020 EPFL4/CLL2 At4g14723 Abrash andBergmann, 2010;Abrash et al., 2011;Lin et al., 2017 EPFL6/CHAL At2g30370 Abrash andBergmann, 2010;Abrash et al., 2011;Lin et al., 2017CLE9 At1g26600 Qian et al., 2018Vatén et al., 2018CLE10 At1g69320 Qian et al., 2018Vatén et al., 2018Receptor ER At2g26330 Rédei, 1992Torii et al., 1996;Shpak et al., 2003Shpak et al., , 2004Woodward et al., 2005;Uchida et al., 2012;Bemis et al., 2013;Chen et al., 2013;Ikematsu et al., 2017;Kosentka et al., 2017;Lin et al., 2017;Hohmann et al., 2020ERL1 At5g62230 Torii et al., 1996Shpak et al., 2003Shpak et al., , 2004Uchida et al., 2012;Bemis et al., 2013;Chen et al., 2013;Ikematsu et al., 2017;Qi et al., 2020ERL2 At5g07180 Torii et al., 1996Shpak et al., 2004Shpak et al., , 2003Uchida et al., 2012;Bemis et al., 2013;Chen et al., 2013;Ikematsu et al., 2017HSL1 At1g28440 Qian et al., 2018FLS2 At5g46330 Hohmann et al., 2020Co-receptor TMM At1g80080 Yang and Sack, 1995Geisler et al., 1998;Nadeau and Sack, 2002;Bhave et al., 2009;Dong et al., 2009;Lin et al., 2017SERK1 At1g71830 Hecht et al., 2001Wang et al., 2008Wang et al., , 2015Meng et al., 2015SERK2 At1g34210 Hecht et al., 2001Wang et al., 2008Wang et al., , 2015Meng et al., 2015SERK3/BAK1 At4g33430 Hecht et al., 2001Chinchilla et al., 2007;Heese et al., 2007;Wang et al., 2008;Postel et al., 2010;Roux et al., 2011;Santiago et al., 2013;Sun et al., 2013aSun et al., , 2013b At1g45640 Wang et al., 2007;Lampard et al., 2008Lampard et al., , 2009Zhang et al., 2015Zhang et al., , 2016Putarjunan et al., 2019MPK6 At2g43790 Wang et al., 2007Lampard et al., 2008Lampard et al., , 2009Zhang et al., 2015Zhang et al., , 2016    This indicates that each individual TF is required for a specific cell differentiation step during stomata development ontogenesis. Although an increased number of ACDs in the er single mutant can be observed, only the erf triple mutant (er erl1 erl2) or loss of tmm exhibit severe stomatal clustering. Uncoupling the functional redundancy of the ERf members identified specific peptide-receptor pairs, which contribute to different steps during stomata development (C-H). SPCH-induced EPF2-dependent paracrine signaling by MMCs (light blue) is perceived by ER in neighboring protodermal cells to inhibit SPCH (dark blue) activity (C). Modulation in EPF2 activity can cause an epidermis devoid of stomata when overexpressed (D) or piled up asymmetric division in epf2 mutants (E) since SPCH activity is not downregulated by the EPF-ER signaling module. MUTE (purple) activity depends on both EPF1-dependent autocrine (light blue) and paracrine (light red) signaling. MUTE directly induces ERL1 expression, while at the same time ERL1 perceives EPF1 signal in SLGCs (paracrine) as well meristemoids (autocrine) to inhibit MUTE activity (F). This negative feedback loop ensures the right amount of MUTE activity provided by the EPF1-ERL1 signaling module for proper GC differentiation. While overexpression of EPF1 causes meristemoid arrest due to excessive activation of the ERL1-driven signaling pathway (G), loss of EPF1 results in stomata clustering due to excessive MUTE activity (H). I, EPFL9, also known as STOMAGEN, is expressed in the mesophyll (dark red) and positively regulates stomatal development through a competitive mode of action with EPF1 and EPF2 for receptor binding on ERf members. While overexpression or exogenous application of EPFL9/STOMAGEN induces paracrine signaling, leading to numerous clustered stomata (J), epfl9/stomagen confers reduced stomatal density (K). Intriguingly, the molecular control of EPFL9/STOMAGEN expression within the mesophyll is tied to auxin signaling. Repression of EPFL9/STOMAGEN is ensured by ARF5/MONOPTEROS (MP), which in return is under the control of specific AUX/IAA BODENLOS (L). Lee et al., 2012). Combined with the opposing epf2 and spch phenotypes (Figure 2, B and E) and SPCH ChIP-seq assays reporting that EPF2 is a direct SPCH target (Lau et al., 2014), these data support a model in which EPF2 is secreted from SPCH-expressing MMCs to neighboring cells and acts as a cell-to-cell (paracrine) signal downregulating SPCH activity. This negative feedback loop prevents surrounding protodermal cells from entering the stomatal lineage ( Figure 2C; Lau et al., 2014;Horst et al., 2015). Experimental evidence has elucidated how this paracrine signal is perceived by neighboring cells. Exogenous EPF2 application on plants expressing the dominant negative form of ER alleviates the effects of the latter, indicating that EPF2 and ER act in the same pathway. Consistently, biochemical and structural data have identified the preformed, heterodimeric complex of ER and TMM as a binding partner of EPF2 Lin et al., 2017). Altogether, this heterodimeric receptor-peptide signaling complex transduces the signal further into the cell (discussed below) to restrict entry into the stomata lineage through SPCH inhibition (Figure 2, C-E). While EPF2 functions during early stomatal development, EPF1 is expressed in later stages, including in late Ms, GMCs, and young GCs, and serves as a positional cue providing spatial information to maintain the "one-cell spacing rule" between developing stomata (Figure 2, F-H; Hara et al., 2007;Lee et al., 2012). Initially identified in a large genomewide screen for secreted peptides, the epf1 single mutant exhibits stomatal pairwise clustering, similar to the phenotype of plants expressing a dominant negative mutant version of ERL1 ( Figure 2H; Hara et al., 2007;Lee et al., 2012). Conversely, overexpression of EPF1, or exogenous EPF1 application, does not affect MMC formation but results in an epidermis composed of arrested Ms, resembling the loss-of-function mutation in MUTE ( Figure 2G; Pillitteri et al., 2007;Lee et al., 2012).
Initially, it was proposed that paracrine signaling from differentiating Ms to neighboring SLGCs must occur to preserve the orientation of secondary asymmetric spacing division, thus maintaining the "one spacing" rule. Indeed, studies localizing the known polarized plasma membrane marker BREAKING OF ASSYMMETRY IN THE STOMATAL LINEAGE (BASL), which predicts the future division plane at the opposing end of the future M, showed that its distribution is flawed in both tmm and epf1 mutants (Dong et al., 2009). Incorrect positioning of BASL crescent accumulation in tmm and epf1 SLGC instructs incorrect asymmetric spacing division orientation. However, the overall accumulation distal to the new M in both tmm and epf1 single mutants is still present (Dong et al., 2009). Since the double mutant combinations of tmm basl as well as epf1 basl exhibit an additive phenotype, BASL likely acts independently of both TMM and EPF1. How exactly the EPF1-ERL1 signaling pathway might influence the location of BASL polarization, and stomatal lineage cell polarity, is still not understood. Taken together, these findings challenge the notion of a paracrine signaling module from differentiating M to neighboring SLGCs.
In spite of this, a new model has been proposed which combines the traditional view of EPF/EPFL family peptides as non-cell autonomous signals with the additional contribution of EPF1 in autocrine regulation of GMC differentiation ( Figure 2F; Qi et al., 2017). MUTE directly induces ERL1 expression, while at the same time ERL1 perceives EPF1 signal to inhibit MUTE activity, creating a negative feedbackloop ensuring the presence of the right amount of MUTE for GC differentiation. This delicate balance between transcriptional regulation and signal transduction can prevent extra symmetric divisions of Ms and GMCs, and an impaired signal transduction by the tmm mutation results in stomatal pairwise clustering, independent of known polarity contributors . This is likely due to excessive MUTE activity, since MUTE overexpression could trigger numerous GMC symmetric divisions (Han et al., 2018). Moreover, this autocrine inhibition of stomata cell fate can lead to severe M arrest, possibly due to excessive activation of the ERL1driven signaling pathway, a phenotype also observed in plants with overexpression or exogenous treatment of EPF1 Qi et al., 2017). This model is also supported by the finding that absolute co-expression of ERL1 and MUTE, driven by the MUTE promoter in an er-family triple mutant background, results in the same severe phenotype. Taking the functional redundancy of the ERf members in consideration, the ERL1, ER, and ERL2 receptor populations expressed in SLGCs are likely buffering the extreme amount of EPF1 peptide ligands secreted from Ms in a paracrine manner to ensure an adequate autocrine inhibition for proper stomata development (Abrash et al., 2011;Uchida and Tasaka, 2013;Qi et al., 2017). Such receptor buffering systems are known in the context of plant development. A genetic study suggests that, during shoot apical meristem formation, BARELY ANY MERISTEM (BAM) receptors sequester peptide ligands at the flanks of the meristems. This paracrine impoundment establishes a buffer around the meristems which prevents these peptide ligands from disturbing the delicate balance necessary for stem cell maintenance (DeYoung and Clark, 2008).
The competitive binding of these peptide ligands with different activities raises the question as to which mode ERf receptor kinase responds to the individual peptides at the subcellular level. A recent study combining genetic, pharmacological, and live cell imaging analyses revealed new insights into the initial subcellular behaviors of the receptor ERL1 upon ligand perception. EPF1, which activates the inhibitory stomatal signaling cascade, triggers TMM-dependent ERL1 internalization into the intraluminal vesicles of multivesicular bodies/late endosomes for subsequent vacuolar degradation (Qi et al., 2020). Conversely, upon EPFL9/ STOMAGEN perception, ERL1 is retained at the endoplasmic reticulum, likely due to impaired ERL1 endocytosis (Qi et al., 2020). It is not clear, however, whether the retained endoplasmic reticulum accumulation of ERL1 originates from former ERL1-TMM receptor complexes recycled back to the endoplasmic reticulum and/or whether newly synthesized ERL1 receptor molecules are stalled to decrease the overall ERL1 receptor population at the cell surface. It is reasonable to predict that ERfs possess specific short-sequences within their cytoplasmic domains, which direct routing and trafficking decisions. Indeed, the dominant-negative ERL1, which lacks the entire cytoplasmic domain, is insensitive to either EPF1 or EPFL9/STOMAGEN application and predominantly remains at the plasma membrane (Qi et al., 2020). It would be interesting to examine whether particular adapter coat proteins recognize specific domains of ERfs and how post-translational modifications (PTMs), such as phosphorylation or ubiquitination, might affect peptide ligand perception or contribute to their molecular sorting. At present, functional evidence for PTMs is only inferred from sitedirected mutagenesis studies of ER (Kosentka et al., 2017;Perraki et al., 2018). Identifying these mechanisms will help illuminate the molecular relationships between antagonistic peptides at the subcellular level.

Check and balance-how CLE peptides influence stomatal-lineage division patterns
Despite the significant roles of secreted EPF/EPFL family members, it is not the only peptide family that influences stomatal patterning. In flowering plants, the CLAVATA3/ENDOSPERM SURROUNDING REGION-RELATED (CLE) peptide family of post-translationally modified dodecapeptides regulate a wide range of biological processes in stem cell homeostasis, as well as in response to phytohormone signaling (Cock and McCormick, 2001;Fletcher, 2020). The potential role of CLE peptides was first implicated from a transcriptome study: CLE9 accumulated highly and specifically in mute scrm-D seedlings, in which the epidermis was solely composed of Ms (Pillitteri et al., 2011). More recent work has shown that CLE9 and CLE10 have dual functions in roots and in leaves (Qian et al., 2018). In the root meristem, CLE9/10 peptides form signaling complexes with BAM receptors to repress the periclinal cell division of xylem precursor cells. In leaves, CLE9/10 negatively regulates the division of MMCs (Qian et al., 2018). There, CLE9/10 peptides are perceived by the LRR receptor kinase HAESA-LIKE1 (HSL1), enabling the activation of the known components of the MAPK cascade, resulting in the phosphorylation and destabilization of SPCH (Qian et al., 2018). Since the exogenous application of CLE9/10 peptide in the er erl1 erl2 triple mutant decreases the number of GCs, it seems likely that this receptor-peptide module functions independently of the known ERf-EPFs signaling pathway to modulate stomata density.
Another study suggests that CLE9/10 peptides act downstream of cytokinin signaling in a non-cell autonomous manner via SPCH (Vatén et al., 2018). In wild-type plants, both CLE9 and the cytokinin signaling effector gene ARABIDOPSIS RESPONSE REGULATOR16 (ARR16), a M-expressed gene acting as negative regulator for the latter, are directly induced by SPCH in Ms (Ren et al., 2009;Pillitteri et al., 2011;Lau et al., 2014;Vatén et al., 2018). During amplifying divisions, the induction of CLE9/CLE10 peptides in Ms suppress further SLGC division potential, possibly upstream of ARR16/17. On the other hand, the sustained ARR16/ARR17 effect in SLGCs reduces their sensitivity to cytokinin and subsequently lowers the probability of undergoing a spacing division. Intriguingly, a lack of either ARR16/17 or CLE9/10 causes high cytokinin response in SLGCs, promoting and preserving SPCH expression, indicated by an increased SLGC division potential. Whereas, these mechanisms exert minimal effects on eventual stomatal patterning, it influences the way satellite stomata are generated. The negative feedback loop between SPCH target activation, cell type-specific negative regulation of cytokinin response factors, and transcriptional repression of SPCH driving ACDs in the stomata cell lineage may ensure the physiological adaptation of leaf growth and cell fate states to a given environmental setting. How other plant hormones participate in this complex molecular network is still unclear.
Genetic evidence suggests that SERK members contribute redundantly to stomatal patterning (Meng et al., 2015). Higher-order serk1 serk2 bak1 serk4 quadruple mutants display excessive stomatal clustering and developmental growth phenotypes reminiscent of the er erl1 erl2 triple mutant, suggesting that these receptors may function together (Meng et al., 2015). Importantly, the stomatal cluster phenotypes of bak/serk higher-order mutants can be genetically uncoupled from BR signaling defects, indicating that BAK/ SERKs enforces stomatal patterning together with ERf, not with BRI1 (Meng et al., 2015). Biochemical coimmunoprecipitation assays further demonstrate that SERKs form an EPF peptide ligand-dependent multi-protein receptor complex with both ER-TMM and ERL1-TMM with EPF2 and EPF1, respectively (Meng et al., 2015).
Interestingly, SERKs not only contribute to the EPF/EPFL peptide ligand dependent pathway but also to CLE9/10 peptide ligands and their receptor, HSL1. Upon receptor complex formation and peptide ligand binding, the binding affinity between CLE9/10 and HSL1 is higher in the presence of SERK family members (Qian et al., 2018). The unique ability of SERK family members to stabilize this specific receptor peptide ligand complex suggests that one primary function of this particular interaction might be to slow down the dissociation of peptide ligands from their respective receptors, a phenomenon that has been reported for other LRR-RLKpeptide ligand complexes upon co-receptor heterodimerization (Hohmann et al., 2018).
Step by step-how downstream components integrate receptor-peptide signaling Ligand-activated heterodimerization of primary LRR-RLKs and BAK/SERKs triggers phosphorylation and association of receptor-like cytoplasmic kinase (RLCK), which bridges the further downstream signal transduction (Liang and Zhou, 2018). Originally identified as a transducer of BR signaling, BRASSINOSTEROID SIGNALING KINASE (BSK) family RLCKs regulate immunity response with FLS2 (Kim and Wang, 2010;Shi et al., 2013;Nolan et al., 2020;Wang et al., 2020). A recent study provided genetic evidence supporting the concept that the BSK family might be the missing link between EPF-mediated receptor-activation and the MAPK cascade (Neu et al., 2019). A loss-of-function double mutant, bsk1 bsk2, results in a clustered stomata phenotype resembling that of yda or er erl1 erl2 triple mutants. BSK1 interacts with the kinase domain of the MAPKKK YDA (Neu et al., 2019). Thus, upon EPF-mediated receptor activation, a tertiary complex of ERf-TMM-SERK may physically interact with BSK1/BSK2 to transduce the phosphorylation-encoded extracellular information further on to YDA (Figure 1).
Formation of the EPFs-ERfs-TMM-SERKs (-BSKs) ligandreceptor signaling complex activates the downstream MAPK cascade, and this activation leads to the degradation of each bHLH TF module (Bergmann et al., 2004;Gray and Hetherington, 2004;Wang et al., 2007;Lampard et al., 2008Lampard et al., , 2009Horst et al., 2015). Consequently, while loss-offunction of MAPK-cascade components causes stomata overproduction, expression of constitutively active versions strongly inhibits stomatal development. Although it has been known for a while that the MAPK cascade plays a significant role in inhibition of SPCH, the molecular mechanism behind the phosphorylation-dependent degradation of each bHLH TF module individually remained elusive until now. A recent structure-function study demonstrated that activated MPK3/MPK6 associate first with SCRM through its bipartite motifs, triggering the subsequent phosphorylation and degradation of the SCRM-SPCH heterodimers, thus preventing entry into the stomata lineage (Putarjunan et al., 2019). Substitutions within the SCRM-KRAAM motif abolish the association with MPK3/6, resulting in stomatal overproduction, a phenomenon observed in scrm-D mutants (which possesses a KRAAM-to-KHAAM amino-acid substitution). Intriguingly, while the putative MAPK docking motif is highly conserved among vascular and nonvascular plants, the KRAAM motif can only be found in SCRM, SCRM2, and their orthologs. This has led to the hypothesis that, while MPK3/6 regulates myriads of developmental, environmental and immunity responses (Zhang, 2018), distinct binding motifs in MAPK substrates are used to cause specific developmental responses (Figure 1; Putarjunan et al., 2019).
In biological processes, kinases and phosphatases act as a phospho-switch to modulate and fine tune the activity of their respective substrates. So far there have been two phosphates identified, which further fine tune the signaling output in stomatal development. MAP KINASE PHOSPHATASE1 (MKP1) controls the phosphorylation status of MAPKs within stomatal precursors downstream of the MAPKKK YDA (Tamnanloo et al., 2018). mpk1 mutant epidermis undergoes asymmetric entry division. However, Ms occasionally fail to differentiate, resulting in rose-petal like SLGC clusters, reminiscent of the mute mutant and EPF1 overexpression phenotypes (Tamnanloo et al., 2018). MUTE expression is diminished in mpk1 (Tamnanloo et al., 2018). These findings suggest that MPK1 counteracts with MPK3/6 in the EPF1-ERL1-mediated signaling pathway to promote M-to-GMC differentiation.
In addition to MKP1, a recent work revealed that subunits A1, A2, and A3 of the PP2A phosphatase promote stomatal development . Alteration in PP2A activity, either through higher-order pp2a mutants or through pharmacological impairment, suppresses stomatal production, indicative of a positive role of PP2A during stomatal development. Furthermore, PP2A-A subunits directly bind SPCH in vitro, suggesting that PP2A may function to regulate the phosphorylation-dependent equilibrium of SPCH protein. Future studies will hopefully illuminate how PP2A activity as well as its regulation is integrated within the ER-EPF2 receptor-peptide signaling module.

How intrinsic polarity cues contribute to ACD and stomata pattern formation
Although receptor-mediated signal-transduction is critical for the regulation of stomatal development, evidence points towards a broader, more complex network of intrinsic polarity cues integrated with extrinsic signaling components that provide spatial information during cell division. ACDs give rise to cells of different sizes and shapes, potentially with different cell fates (Abrash and Bergmann, 2010). To achieve this, the mother cell has to determine an axis of polarity prior to mitosis in a way that the nucleus possesses an asymmetric position prior to division, orienting the division plane in relation to this axis (Rasmussen et al., 2011). In addition, during asymmetric amplifying divisions, SPCH activity must be strictly regulated in both Ms and SLGCs to sustain the stem-cell-like properties of the M. The plant specific polarity protein BASL promotes asymmetry via differentially regulating SPCH activity within two daughter cells, M and SLGC, within the stomatal lineage (Figure 3; Dong et al., 2009;Zhang et al., 2015Zhang et al., , 2016. BASL displays a dynamic subcellular localization, accumulating first in the nucleus of premitotic cells, followed by a highly polarized crescent in the region distal to the future division plane and away from the migrated nucleus (Dong et al., 2009). In loss-of-function basl mutants, a high proportion of symmetric cell divisions can be observed in MMCs. Furthermore, a domain analysis of BASL identified its C-terminal region as a prerequisite to direct polarized accumulation at the cell periphery, which was sufficient for polar cell growth (Dong et al., 2009). This suggests that the initial nuclear localization serves as a reservoir of BASL protein.
BASL polar localization and function is dependent on both MPK3/6, as well as on members of the plant-specific BREVIS RADIX (BRX) family (Figure 3, A and B; Zhang et al., 2015Zhang et al., , 2016Rowe et al., 2019). One model suggests that MPK3/6-dependent phosphorylation of BASL is essential to shift from its nuclear localization toward accumulating at the plasma membrane ( Figure 3A; Zhang et al., 2015Zhang et al., , 2016. Here, BASL directly associates with YDA, acting as a scaffold protein to polarly tether the MAPK complex to the cellular cortex of SLGCs. This provides a positive feedback loop between BASL and the YDA-MAPK cascade. Although the molecular details still remain unclear, this polar localization is necessary to inhibit SPCH activity within the SLGC. Conversely, it has recently been reported that members of the BRX family are also required for ACD within the stomatal lineage, forming a co-dependent equilibrium with BASL for both polarization and localization ( Figure 3B; Rowe et al., 2019). BRXf and BASL physically interact to mutually influence localization, independent of the BASL-MAPK circuit, demonstrating the existence of a more complex core polarity system that enables the scaffolding, positioning, and segregation of additional proteins necessary for cell-fate commitment.
Polarity proteins from the POLAR LOCALIZATION DURING ASYMMETRIC DIVISION AND REDISTRIBUTION (POLAR) family function at the cortical site distal to the future division plane of premitotic MMCs and Ms in a BASLdependent manner (Figure 3, C-G; Pillitteri et al., 2011). POLAR forms a scaffolding complex that controls the polarity of the GLYCOGEN SYNTHASE KINASE3/SHAGGY-like/ BR-INSENSITIVE2 (BIN2) kinase. The scaffold complex enables the inhibition of the YDA MAPK cascade at the cell cortex before ACD via direct phosphorylation of YDA in a BIN2-dependent manner (Gudesblat et al., 2012;Khan et al., 2013;Houbaert et al., 2018). This, in turn, disrupts the balance of both nuclear and cytoplasmic MAPK signaling components, resulting in elevated accumulation of SPCH in the nucleus and promoting stomatal asymmetric division in Ms ( Figure 3C). Soon after division, BIN2 phosphorylates POLAR at the BASL defined polarity site, resulting in the disassociation of BIN2 and its re-localization to the nucleus (Houbaert et al., 2018). The combined effort of nuclear-localized BIN2 activity, which directly phosphorylates SPCH and elevates MAPK signalling, further restricts ACD in the SLGC, resembling a fine-tuning mechanism of cell specification during leaf morphogenesis ( Figure 3D; Gudesblat et al., 2012;Houbaert et al., 2018).
The intensive localization analyses of BASL-YDA-MPK3/6 as well as BASL-POLAR-BIN2 modules suggest that their ability to polarize and to form a complex in vivo emanates from BASL's ability to self-organize in a polar manner upon phosphorylation by the YDA-MPK3/6 cascade Houbaert et al., 2018). This feedback circuit implies a mechanism in which a MAPK signaling gradient is formed descending from the BASL crescent toward the nucleus in a concentration-dependent manner (Shao and Dong, 2016). The idea of constitutive signaling gradients in the context of development has been identified in various biological systems. For instance, the small GTPase RAN, a key player in the "spindle self-organization" pathway in animals, maintains a gradient within a certain threshold at discrete positions around the chromosomes, controlling the activity of spindle assembly factors (Kalab et al., 2002;Clarke and Zhang, 2008;Zhang and Dawe, 2011).
With respect to a potential signal gradient during stomatal development, a recent publication identified new MAPK substrates, which regulate stomatal production in a positive manner (Xue et al., 2020). The authors reported that the MAPK SUBSTRATE IN THE STOMATA LINEAGE (MASS) 1, 2, and 3 function at the plasma membrane, negatively impacting the MAPK signaling cascade. Similar to BASL and in contrast to MASS3, both MASS1 and MASS2 initially localize to the nucleus and are re-localized toward the cell periphery upon MPK6 phosphorylation. However, overexpression of MASS2 results in clustered stomata and disregard of the "one-spacing rule" independently of BASL, since its localization remains unaffected (Xue et al., 2020). These data might reflect a potential relationship between the regulation of division reorientation by extrinsic signals triggered by ligand-receptor interaction and/or the YDA MAPK cascade through the control of a MAPK gradient within the stomatal lineage. With recent identifications of polarity proteins both in the evolutionarily conserved or broader versus tissue-specific contexts (e.g. Muroyama and Bergmann, 2019;Van Dop et al., 2020), our understanding of the mechanism underpinning intrinsic plant cell polarity is burgeoning. Whether shared extrinsic peptide signaling modules are coopted to influence different polarity systems is an important future question.
Variations in a theme: shared modules ER was originally described as a gene regulating inflorescence architecture, found as a spontaneous mutation in the commonly used accession Landsberg erecta (Rédei 1992;Torii et al., 1996). Through the analysis of stomatal development, upstream ligands, receptor complex modules, and intracellular phosphorylation cascades have been discovered.
This raises a further question as to whether other aspects of ERf-mediated developmental processes use analogous, even shared signaling components. Indeed, two EPFL peptides, EPFL4 and EPFL6, promote inflorescence architecture as ligands for ER. Interestingly, EPFL4/6 are expressed in the stem endodermis, and phloem-expressed ER is sufficient to perceive these signaling peptides in a non-cell autonomous, paracrine manner (Figure 4; Uchida et al., 2012).
Like stomatal development, co-receptor BAK/SERKs and RLCKs BSK1 and BSK2 promote inflorescence architecture most likely with ER: higher-order bak/serk mutants as well as bsk1 bsk2 double mutants develop shorter inflorescence with characteristic flower-bud clusters, a phenotype resembling that of er (Meng et al., 2012;Neu et al., 2019). Yet another RLCK, BOTRYTIS-INDUCED KINASE1 (BIK1), has been recently reported in the contexts of ER signaling. In contrast to BSK1/2, BIK1 plays a negative role in inflorescence development ( Figure 4A; Chen et al., 2019). Loss-of-function bik1 mutant plants develop significantly longer internodes and pedicles, as well as a looser inflorescence compared to wild-type plants. This phenotype could be partially rescued upon additional loss of er, indicating an antagonistic Figure 3 Intrinsic polarity cues contribute to stomatal development. The plant-specific polarity protein BASL (green) exhibits a dual localization pattern. While BASL accumulation can be detected in the nucleus during stomatal lineage progression, it additionally forms a polarized crescent in MMCs and meristemoids in the region distal to the future division plane and away from the migrated nucleus. This crescent localization is dependent on both MPK3/6 (A), as well as on members of the plant-specific BREVIS RADIX (BRX) (light red) family (B). Right before MMC asymmetric division (C), BASL and POLAR (orange) recruit the BIN2 (light blue) to the cellular cortex. This BASL-POLAR-BIN2 polarity module enables the inhibition of the YDA MAPK cascade via direct phosphorylation of YDA. This leads to an elevated accumulation of SPCH, hence promoting stomatal asymmetric division. To prevent further ACD in SLGCs, SPCH must be down regulated. To do so, BIN2 disassociates from POLAR after phosphorylating the latter and re-localizes into the nucleus to inhibit SPCH activity while POLAR is degraded (D). Furthermore, a positive feedback loop between BASL and the YODA MAPK cascade restricts further stomatal ACD in the SLGC through direct phosphorylation of SPCH. The specific localization pattern of the BASL-POLAR-BIN2 polarity module further fine tunes stomatal development (E-G).
Altogether, these findings highlight that EPF/EPFL-ERfmediated signaling pathways in stomatal development and inflorescence growth are variations on a theme-pathways composed of the same/paralogous modules (Figure 4). Although the downstream components of such ERf-mediated developmental processes often remain unknown, these variations on a theme become even more visible when other EPF/EPFL-ERf-mediated signaling pathways are taken into consideration. During leaf margin morphogenesis, for instance, auxin responses are maintained at tips of the teeth to promote their growth (Kawamura et al., 2010;Heisler and Byrne, 2020). This auxin response represents a culmination of a feedback circuit between the EPFL2/ERf peptide-receptor module to restrict and define auxin maxima (Tameshige et al., 2016). Auxin shapes the site of EPFL2-ERf signaling at the boundary of each leaf teeth primordium by repressing EPFL2 expression from the leaf teeth tips while promoting ERL2 expression at the leaf teeth tips. The EPFL2-ERf signaling in turn restricts the auxin maxima in the tip to enable directed leaf teeth growth.
In addition to its role during leaf margin morphogenesis, a recent study found that differential expression of EPFL2 Figure 4 Signal specificity is maintained between different LRR-RLKs with shared intermediate components. ER mediates different developmental processes upon different peptide-ER-co-receptor formation. (A) During stem elongation EPFL4 and EPFL6 are secreted from endodermis cells into the apoplast to be perceived by ER in phloem cells in a noncell autonomous manner. Upon EPFL4/6-ER-SERKs formation, with the additional possibility of a higher order complex with BSKs (question mark) as well the RLCK BOTRYTIS-INDUCED KINASE1 (BIK1), a MAP kinase cascade is activated to transduce the extracellular information into the cell to further promote downstream effectors for stem elongation. Although stomata development depends also on the formation of an EPF/EPFL-ERf-SERKs complex, possibly with BSKs, to activate a MAPK cascade to overall inhibit stomata development (B), this peptide-receptor-co-receptor complex additionally requires the co-receptor TMM. Whether BIK1 also plays a role during stomata development is still unknown (question mark). Stomata development can be further fine-tuned through CLE9/10 peptides, which are perceived by the LRR receptor kinase HAESA-LIKE 1 (HSL1) and its coreceptor BAK1, enabling the activation of the same components of the MAPK cascade. It is unclear how the MAPK cascade is able to cope with the signal output of two different peptide-receptor signaling complexes for the same developmental process as well as create signal specificity between shared pathways, such as the flg22-FLS1-SERKs signaling module for plant innate immunity. and EPFL9/STOMAGEN, resulting in a reginal activation of specific ERf members, couples ovule initiation with fruit growth (Kawamoto et al., 2020). While EPFL2 acts predominantly through ERL1 and ERL2 to control the initiation and spacing of ovule primordia during gynoecium and fruit growth, EPFL9/STOMAGEN, acting from the carpel wall, mainly promotes fruit growth through ER. Intriguingly, EPFL9/STOMAGEN fail to compensate here for the loss of epfl2 when expressed under the EPFL2 promoter, indicating that EPFL9/STOMAGEN can antagonize EPFL2 functions during fruit growth, a phenomenon best described in stomatal development.
All in all, EPF/EPFL-ERf-mediated signaling pathways and their same/paralogous modules became indispensable for various aspects of plant growth and development. Many signaling components, such as BAK/SERKs, RLCKs, and MAPK cascades, are even shared with broader signaling pathways in development, environmental response, and immunity (Kim and Wang, 2010;Shi et al., 2013;Nolan et al., 2020;Wang et al., 2020). This highlights two important questions of signal discrimination and specificity. First, how can ERfs discriminate different EPF/EPFL peptides and properly ensure specific outcomes? Second, how can different signaling pathways, such as those of stomatal development and immunity, maintain specificity?
One critical factor distinguishing EPF/EPFL-ERf-mediated stomatal development from inflorescence growth is TMM (Figure 4). TMM is specifically expressed in the epidermal layer, and whereas EPF1 and EPF2 require TMM to activate ERf members, EPFL6 (also known as CHALLAH; CHAL) and EPFL4 (CHAL-LIKE2) do not (Abrash and Bergmann, 2010;Abrash et al., 2011). In the absence of TMM, ERf in epidermal cells can be activated by EPFL4/6 which bleed-through from endodermis to epidermis and inadvertently suppress stomatal development in stems and hypocotyls (Abrash and Bergmann, 2010;Abrash et al., 2011). Consistently, epfl4/5/6 triple mutation reverses the stomata-less phenotype of tmm stems and hypocotyls (Abrash et al., 2011). More recent structural biology and cell biology studies support this model: EPFL4/6 binds with 10 time higher affinity to ERL1 than to the ERL1-TMM complex (Lin et al., 2017). Likewise, ERL1 can perceive EPFL6 and rapidly undergoes endocytosis in the absence of TMM (Qi et al., 2020).
Understanding how signal specificity is maintained between different LRR-RLKs with shared intermediate components, such as stomatal development and immunity signaling, is an important outstanding question. Evidence has been provided that both the EPF-ERf and flg22-FLS2 immunity pathways exhibit antagonistic interactions at the MAPK cascade ( Figure 4B; Sun et al., 2018). On the other hand, a more recent, newly designed genetic tool that can constitutively activate BAK/SERK-dependent LRR-RLKs suggests the maintenance of basal signal specificity between ERf and FLS2 pathways when they are expressed within the context of stomatal cell lineages (Hohmann et al., 2020).
Harnessing new tools and approaches will be required to further delineate the molecular mechanisms preventing signal interference.

Conclusion and perspectives
Significant progress has been made in the past two decades to unravel the molecular mechanisms behind stomata development. The discovery of specific receptor-peptide signaling complexes, transducing their extracellular perceived peptide signals onto a MAPK cascade to further inhibit specific TFs, has dramatically enhanced our understanding of cell state transition and differentiation within the stomatal lineage. The identification of intrinsic polarity cues, contributing to ACD, has provided us a unique opportunity to understand division plane switch and the mechanisms behind protein polarization at the plasma membrane in plants. There are still very big, open questions to be addressed (see also Outstanding Questions Box). For example, what molecular mechanisms connect external signaling factors with the intrinsic polarity cues? And how, through the evolution of land plants, have different plant species co-opted and rewired these signaling and polarity modules to achieve specific stomatal patterns optimized to their ecological niche? Recent studies in monocots, including rice (Oryza sativa), grass species (Brachypodium dystachyon), and moss (Physcomitrella patens), reveal new exciting insights Chater et al., 2016;Raissig et al., 2017;Abrash et al., 2018;Hepworth et al., 2018;Lu et al., 2019;Wu et al., 2019).
The ability of plants to regulate themselves to the atmospheric balance and water cycles of our planet is dependent on the precise control of stomatal development and distribution. Since key regulators of stomata development are conserved in grass species and most land plants (see also Endo and Torii, 2019), deciphering the molecular network underlying these processes could well uncover a promising trait for agricultural application. It is therefore crucial to continue basic research to fully understand the molecular mechanisms behind these developmental processes and how plants integrate external signals to optimize stomatal formation. Applying new genetic tools, we may be able to manipulate stomatal development and patterning in our favor to increase agricultural harvests and reinforce plants to an ever-changing climate.

Accession numbers
Please see Table 1: Regulators of Stomatal Development for an overview of all accession numbers of all major genes and proteins mentioned in this review.
acknowledges the Johnson & Johnson Centennial Endowed Chair from the UT Austin Molecular Biosciences.

Funding
This work was supported by the Howard Hughes Medical Institute, Gordon and Betty Moore Foundation (GBMF-3035) to K.U.T.

OUTSTANDING QUESTIONS
• What molecular mechanisms on a subcellular level determine signal specificity upon competitive binding of antagonistic acting peptide ligands (e.g. EPF1/2 versus EPFL9/ STOMAGEN)? • How is signal specificity maintained between different LRR-RLKs with shared intermediate components? (e.g. EPFf-ERf vs. CLE-HSL1 versus flg22-FLS)? • How can different receptor peptide modules (e.g. EPF/EPFL-ERf versus CLE-HSL) integrate different external cues to further fine tune the same developmental process? • How are intrinsic polarity cues integrated with extrinsic signaling components that provide spatial information during cell division?