All land plants regulate gas exchange through the stomata, which consist of two kidney-shaped guard cells surrounding a pore. The degree of pore opening is determined by the turgor pressure of the guard cells, which depends on the movement of water and large quantities of ions and sugars between the guard cells and their neighbouring epidermal cells . High turgor pressure induces pore opening, whereas low turgor triggers its closure . The exchange of gases between the plant and the atmosphere depends on the degree of openness of stomatal pore, but also of the number of stomata on the plant surface and of their spatial distribution [2,3].
Stomatal formation in Arabidopsis and other species is preceded by a series of asymmetric cell divisions and a final symmetrical division  (Fig. 1). The meristemoid mother cell initiates this process, generating the daughters, meristemoid and stomatal lineage ground cell. The meristemoids can divide asymmetrically (amplifying divisions), and in a spiral pattern, before producing the paired guard cells. Some stomatal lineage ground cells differentiate into pavement cells, whereas others can also divide asymmetrically (spacing divisions) producing new meristemoids (satellite meristemoid) placed away from the pre-existing stoma. This process, which depends on the orientation of the planes of cell division, ensures that the stomata are not in direct contact with their stomata neighbours [4,5], guaranteeing ion flux between guard cells and neighbouring non-stomatal cells, allowing so stomatal movements.
Extensive work carried out during the last two decades supports the existence of a regulatory pathway that controls stomatal development and patterning (Fig. 1). In this pathway, peptides ligands of the EPIDERMAL PATTERNING FACTOR-LIKE family are recognized by TOO MANY MOUTHS (TMM) receptor-like protein, and the receptor-like kinases ERECTA (ER), ERECTA-LIKE1 (ERL1) and ERL2, which allow to the ERECTA-family receptor kinases (ERfs) [4,6,7]. Because TMM associates with ERL1 in vivo , the formation of heterodimeric complexes between TMM and ERfs might be required for the initiation of this signalling cascade. In any case, these receptors, activated by peptide ligands, activate a MITOGEN-ACTIVATED PROTEIN (MAP) KINASE module that contains three sequentially activated protein kinases: 1) the MAP kinase kinase kinase YODA; 2) the MAP kinase kinases MKK4, MKK5, MKK7 and MKK9; and 3) the MAP kinases MPK3 and MPK6 [9-11]. MPK3 and MPK6 phosphorylate and destabilize the basic helix-loop-helix protein SPEECHLESS (SPCH), so blocking the entry into the stomatal development pathway [12-14]. SPCH activation drives the first division of stomatal pathway . SPCH not only triggers the initiation of the development of stomata, but also prolongs the cell divisions of the meristemoids . Then, the basic helix-loop-helix MUTE, acting most probably independently of YODA and its downstream kinases , blocks the asymmetric cell divisions of the meristemoids, triggering the formation of the guard mother cell [12,13]. Finally, FAMA, which also encodes a basic helix-loop-helix protein , causes paired guard cells formation by inducing that guard mother cells divide symmetrically .
Stomatal development is regulated by environmental and endogenous factors. Three phytohormones, brassinosteroids, abscisic acid (ABA) and auxins have been implicated in the control of stomatal development [18-21]. This article delves into the role of brassinosteroids and ABA during stomatal development in Arabidopsis and other species. Interestingly, these plant regulators repress stomatal development in Arabidopsis leaves [18-20], affecting the same stages of this process of development, most probably through the regulation of the YODA-MKK4/5/7/9-MPK3/6 module. Indirect evidences also suggest that the regulation of stomatal development through these phytohormones may have evolved after the divergence of lycophytes and seed plants and before that of angiosperms and gymnosperms.
2. Brassinosteroids repress stomatal development in cotyledons and leaves
One of the most well-characterized plant pathways is regulated by brassinosteroids [22,23] (Fig. 2). These phytohormones are perceived by a plasma membrane-localized and leucine-rich-repeat receptor-like kinase named BRASSINOSTEROID INSENSITIVE 1 (BRI1) . Brassinosteroid recognition by BRI1 results in the activation of a family of kinases named BRASSINOSTEROID SIGNALLING KINASES (BSKs) . Activation of BSKs is followed by the phosphorylation and activation of the Kelch-repeat domain-containing protein phosphatase BRI1 SUPPRESSOR 1 (BSU1), leading to the dephosphorylation and inactivation of the glycogen synthase kinase3-like kinase BRASSINOSTEROID INSENSITIVE 2 (BIN2) [25-27]. The repression of BIN2 activity and action of PROTEIN PHOSPHATASE 2A (PP2A) promote the increase in the nucleus of non-phosphorylated BRI1 EMS SUPPRESSOR 1 (BES1)/BRASSINAZOLE RESISTANT 1 (BZR1), regulating many processes such as cell expansion, reproductive development, etiolation, vascular differentiation, cell division and stress responses [28-32]. Without brassinosteroids, BIN2 inhibits BES1/BZR1 function through phosphorylation.
Genetic and pharmacological studies have implicated these plant regulators in stomatal development and patterning. The surface of cotyledons of the quadruple loss-of-function mutant of BSU1-related phosphatases (bsu-q) was full of paired guard cells, lacking other epidermal cells  (Fig. 3B). In spite of this drastic phenotype, bsu-q mutant is not lethal. Brassinosteroid biosynthetic (deetiolated2-1) or insensitive (bri1-116, dominant bin2-1, and plants overexpressing BIN2) mutant/transgenic plants had cotyledons bearing more stomata than their corresponding wild type plants  (Fig. 3C). Some of these stomata were in direct contact with their stomata neighbours . In contrast, the number of stomata was reduced in deetiolated2-1 plants overexpressing BSU1-LIKE2 and in plants without BRASSINOSTEROID-signalling glycogen synthase kinase3-like kinases (bin2-3 bil1 bil2 null mutant)  (Fig. 3D). These findings support a hypothesis that brassinosteroids negatively regulate stomatal development in cotyledons . As might be expected, plants grown on medium supplemented with either brassinolide (a type of brassinosteroid) or bikinin (a repressor of BIN2 activity) developed less stomata than their controls . Interestingly, brassinosteroids seem to negatively regulate stomatal development in cotyledons independent of the BIN2 substrate BZR1, as gain-of function mutant of BZR1 had no defect, and this mutant had no effect on stomata formation in bri1-116, bsu-q and bin2-1 .
The effect of these phytohormones is not limited to cotyledons. Brassinosteroids also repress development of stomatal clusters in leaves. Plants that either lacked brassinosteroids (constitutive photomorphogenesis and dwarfism) or had reduced sensitivity to these regulators (bri1-1, dominant bin2-1 and wild type plants overexpressing a glycogen synthase kinase3/shaggy-like kinase that acts redundantly with BIN2) also developed stomatal clusters in their leaves . Indeed, wild type plants grown on medium supplemented with brassinazole, an inhibitor of brassinosteroid biosynthesis, also developed stomatal clusters .
In contrast with these results, Gudeblast et al.  found that brassinosteroids play no role, or even a very small positive role, on stomatal production in cotyledons. One possible explanation for this disparity could be related to the fact that these authors used different growth conditions to those employed by Kim et al. . Genetic and pharmacological analysis have also shown that, in the hypocotyl, these plant regulators promote stomata formation [33,34].
3. BIN2 represses YODA and MKK4 by phosphorylation
Genetic and biochemical experiments have unravelled a crosstalk between the brassinosteroid signalling pathway and the stomatal signalling cascade (Fig. 2). The expression of a YODA variant, constitutively active, in bri1-116, bsu-q or bin2-1 mutants blocked the development of stomata . This YODA variant, indirectly, seems to result in the phosphorylation and inactivation of SPCH, through the activation of the MAP kinase pathway [10,14]. These findings suggest that brassinosteroid signalling components act upstream of YODA . In agreement with this idea, the bsu-q spch-3 mutant, like spch-3, was devoid of stomata , resulting in plants that only survive under laboratory conditions.
Both in vitro and in vivo assays have demonstrated that BIN2 physically interacts and phosphorylates YODA, inhibiting its activity [18,35] (Fig. 2). As expected, the activation of BIN2 (deetiolated2 mutants) reduced MPK3 and MPK6 activity, while its repression (bikinin or brassinolide treatments in wild-type plants) triggered the opposite response, resulting in plants with increased MPK3 and MPK6 activity . Mass spectrometry analysis has shown that BIN2 also phosphorylates MKK4 (and MKK5) on at least two residues, serine 230 and threonine 234 . While the BIN2-mediated threonine 234 phosphorylation is directly related to stomatal development , the significance of serine 230 phosphorylation remains unknown. In addition, the phosphorylation of MKK4 by BIN2 represses its activity against MPK6 in vitro . Future work is required to elucidate which residues of YODA are phosphorylated by BIN2.
Previous work has shown that the output of brassinosteroid signalling is regulated by BIN2 through the control of the localization of both BES1 and BZR in specific subcellular compartments [32,36,37]. Similarly, brassinosteroid signalling may be controlled by BIN2 through the regulation of the subcellular localization of YODA . In principle, because the nucleocytoplasmic localization of MKK4 remains unchanged after exogenous application of either brassinosteroids or brassinazole , this mechanism may not apply to MKK4 .
BIN2 also represses SPCH by phosphorylation . However, it has been proposed that while BIN2-mediated phosphorylation of YODA and MKK4 takes place in the cotyledon, the regulation of SPCH by BIN2 occurs in the hypocotyl . Low levels of MAP kinases in the embryonic stem may prevent BIN2 regulation of YODA and MKK4, triggering BIN2-mediated phosphorylation of SPCH to negatively regulate stomatal development in this plant organ .
4. ABA represses stomatal formation in cotyledons and leaves
ABA has been postulated as the main regulator that controls stomatal movement . The binding of ABA to its receptors (PYRABACTIN RESISTANCE 1 (PYR)/PYRABACTIN RESISTANCE 1-LIKE (PYL)/REGULATORY COMPONENTS OF ABA RECEPTORS (RCARs)) inhibits protein phosphatases of type 2C (PP2C; notably ABA-INSENSITIVE1 (ABI1) and ABA-INSENSITIVE2 (ABI2)), allowing that SUCROSE NON-FERMENTING-1-RELATED PROTEIN KINASES (SnRK2s; remarkably OPEN STOMATA1) activate many ABA responses . These responses include the activation of the efflux ion channels (for example, SLOW ANION CHANNEL-ASSOCIATED 1 (SLAC1)), triggering stomatal closure  (Fig. 2). The role of ABA in stomatal closure has been supported by many studies involving different species, such as Vicia faba [41,42] Commelina communis , Arabidopsis thaliana , or Vitis vinifera .
Both ABA-deficient plants (aba2-2) and those that lack sensitivity to ABA (abi1-1 and abi1-2) developed more stomata on cotyledons and leaves than control plants  (Fig. 3E). Despite increasing the number of stomata, these genetic backgrounds did not develop stomatal clusters . In contrast, plants with endogenous high ABA levels (cyp707a1a3) developed fewer stomata  (Fig. 3F). These results support a hypothesis that ABA, in a similar manner as brassinosteroids, represses stomata development in cotyledons and leaves. Consistently, application of ABA rescued the phenotype of aba2-2, but not of abi1-1 . ABA not only represses stomatal development but also induces enlargement of pavement cells , contributing with both factors to decrease the number of stomata per surface. Moreover, this phytohormone, in contrast to its role on pavement cell expansion, seems to reduce guard cell expansion and so pore size in mature stomata . In Lepidium, ABA increases the amount of wax compounds with chain length longer than C26 , which is related to lower cuticular transpiration . In Arabidopsis, ABA may also regulate the composition of cuticular waxes and, consequently, the level of cuticular transpiration. Agree with this view, ABA promotes the expression CER6 gene of Arabidopsis , which encodes a condensing enzyme involved in the elongation of fatty acyl-CoAs longer than C24 . In short, ABA regulates a number of processes that reduce the loss of water from the plant. Interestingly, FOUR LIPS and/or MYB88, which encode R2R3-MYB proteins controlling the transition from guard mother cell to guard cells , might mediate ABA responses .
Interestingly, mutations in OPEN STOMATA 1 or SLAC1 do not seem to alter stomatal production [51,52], suggesting that components upstream of OPEN STOMATA 1 repress stomatal development in cotyledons and leaves.
5. MAP kinase signalling and ABA action
MAP kinase cascades are typically points integrating multiple signals that regulate one or more processes. The ABA signal is transmitted through MAP kinase signalling . For example, ABA activates MPK3, and overexpression of MPK3 increases phytohormone sensitivity in ABA-induced postgermination arrest of growth . The inhibition of MAP kinase signalling decreases sensitivity to postgermination growth arrest by this plant regulator, supporting a hypothesis that the ABA signal is transmitted through MPK3 . ABA also transiently activates MPK4 and MPK6 , and MPK6 regulates an ABA-dependent signalling cascade that results in hydrogen peroxide production and stress responses .
Hydrogen peroxide is a signalling molecule involved in ABA-induced stomatal closure . Knockout mutants of MKK1 and MPK6 blocked ABA-dependent hydrogen peroxide production in guard cells . MPK3 antisense plants were impaired in ABA inhibition of stomatal opening, but responded normally to ABA in stomatal closure . These plants were less sensitive to hydrogen peroxide, both in inhibition of stomatal opening and in promotion of stomatal closure . It is so likely that MPK3 functions downstream of reactive oxygen species (ROS) in ABA inhibition of stomatal opening but not in ABA-induced stomatal closure. Moreover, plants with mutations in both MPK9 and MPK12, but not in a single gene, showed reduced ABA promotion of stomatal closure, ABA inhibition of stomatal opening and hydrogen peroxide promotion of stomatal closure . Consistently with these results, ABA and calcium failed to activate anions channels in guard cells of mpk9-1 mpk12-1 mutants, which had enhanced transpirational water loss in leaves . Then, MPK9 and MPK12 induce redundantly stomatal closure in response to ABA and acting downstream of ROS . Interestingly, recent evidence indicates that whereas MPK9 and MPK12 mediate stomatal closure induced by ABA, biotic stress-induced stomatal closure is mediated by MPK3 and MPK6 .
MAP kinases are transmitters of an ABA-dependent cascade that controls stomatal movement , and also most probably stomatal development (Fig. 2). Certainly, MPK6 (and perhaps components upstream of MPK6) is a target of ABI1 , which controls stomatal development. It is therefore likely that MPK6-containing MAP kinase module mediates ABA-induced repression of stomatal development. Brassinosteroids, through BIN2, regulate the same MAP kinase module as ABA. ARABIDOPSIS PROTEIN PHOSPHATASE 2C 3, a member of the PP2C family, also regulates stomatal development by blocking MPK3, MPK4 and MPK6 activity . Then, brassinosteroids and ABA pathways regulating stomatal development seem to converge at the same MPK6-containing MAP kinase module, whereas ABA-induced stomatal closure is mediated by MPK9 and MPK12.
6. Brassinosteroids and ABA may control the same steps of stomatal development
Under low brassinosteroid levels, BIN2 represses both YODA and MKK4 by phosphorylation, which reduces MAP kinase activity [18,19] (Fig. 2). Low MAP kinase activity interferes with the frequency, the orientation and the polarity of asymmetric cell divisions that take place during stomatal development, resulting in clustered stomata . Then, stomatal clusters in both brassinosteroid-deficient plants and plants lacking sensitivity to brassinosteroids, most probably, derive from an increased number of meristemoid mother cells dividing asymmetrically, and also from disruptions in the orientation and the polarity of asymmetric cell divisions during stomatal development. When brassinosteroid levels get high, the alleviation of the BIN2-mediated inhibition of both YODA and MKK4 increases MAP kinase activity. High levels of MAP kinase activity prevent the asymmetric cell division of the meristemoid mother cells and the commitment of the meristemoids to becoming stomata . MPK3/MPK6 gain-of-function mutants underwent less asymmetric cell divisions than control plants, and did not develop stomata . It is then likely that brassinosteroids reduce both the number of meristemoid mother cells undergoing asymmetric cell divisions and the number of meristemoids developing into stomata. Interestingly, the early steps of the biosynthesis of sterols, which are precursors of brassinosteroids, are required for determining cell fate after the asymmetric division of meristemoid mother cells or meristemoids . Sterols function independently of brassinosteroid-mediated signalling pathway in stomatal development .
The regulation of the MUTE promoter in ABA-deficient plants (aba2-2) differs from that of wild-type plants : aba2-2 pMUTE::GFP plants increased both the number of cells expressing GFP and the temporal windows in which these cells could be visualized. The activity of MUTE promoter is confined to a few meristemoids [12,13], where MUTE probably represses asymmetric cell divisions and triggers the formation of the guard mother cell . This suggests that more cells have initiated stomatal development in aba2-2 than in the wild type. Then ABA, like brassinosteroids, seems to reduce also the number of meristemoid mother cells undergoing asymmetric cell divisions. Supporting this view, both ABA-deficient (aba2-2) and ABA-insensitive (abi1-1 and abi1-2) plants had increased transcripts levels of SPCH and MUTE compared to wild-type plants, while cyp707a1a3 decreased the level of these transcripts relative to wild-type plants . Moreover, wild-type plants had a prolonged and more extensive initiation of stomata than cyp707a1a3, supporting the role of ABA on reducing the number of cells entering into the stomatal pathway . This phytohormone may also reduce the commitment of the stomatal precursors to a stomatal fate.
The loss-of-function mute mutant lacks stomata but develops meristemoids [12,13]. The double mutant mute aba2-1 had more meristemoids than mute . This supports the view that ABA controls the entering into the stomatal pathway, while MUTE plays a later role regulating the development of the stoma from the immediate stomatal precursor, that is, the last meristemoid. The transition from the meristemoid to the guard mother cell is also regulated by auxin: studies on dynamic changes in auxin activity have shown that a decrease in the levels of this phytohormone in meristemoids triggers guard mother cell formation .
In contrast to brassinosteroid-deficient plants and/or plants lacking sensitivity to brassinosteroids, plants with disruptions in ABA production and/or signalling transduction did not develop stomatal clusters [18-20]. An unknown mechanism might prevent cluster formation in these plants. Alternatively, the levels of MAP kinase activities could be high enough to prevent cluster development in plants with disruptions in ABA production and/or signalling transduction.
7. Do brassinosteroids and ABA control stomatal development beyond Arabidopsis?
The sequence of cell divisions during stomatal development differs among plant groups, but in all them, at least two stages are constant: 1) the first asymmetric cell division that occurs in the meristemoid mother cell, generating the stomatal precursor; and 2) the transition from the guard mother cell to the paired guard cells. Orthologs of SPCH and FAMA are present in all seed plants tested, in agreement with their stomatal development .
SPCH orthologs, like SPCH, also have a MAP kinase target domain [65-67]. This domain regulates SPCH activity in response to phosphorylation by the MAP kinases . The conservation of this MAP kinase target domain in SPCH orthologs, and of the sites that are phosphorylated in SPCH, suggest that these basic helix-loop-helix proteins are also regulated by MAP kinase cascades, which, in turn, may be regulated by brassinosteroids and/or ABA. Thus, stomatal production beyond Arabidopsis may also be regulated by phytohormones. In supporting this view, ABA (or growth conditions that trigger ABA accumulation) reduces stomatal density in several plant species such as rice , basil , Populus alba , or lupin .
Interestingly, the basic helix-loop-helix proteins that regulate stomatal development in the two lower land plants, Selaginella and Physcomitrella, which seem to be capable of promoting multiple steps of stomatal development, do not contain a MAP kinase target domain [65,67]. This finding suggests that SPCH, and the hypothetical control of stomatal production through phytohormones such as brassinosteroids and/or ABA, evolved after the divergence of the lycophytes and seed plant lineages and before that of gymnosperms and angiosperms.
8. Concluding remarks
Both brassinosteroids and ABA harmonize the development and the movements of the stomata that optimize water use and photosynthesis in Arabidopsis. Although studies of the role of these regulators during stomatal development in other plant species is not as extensive as with Arabidopsis to know how much this response extends, the conservation of the MAP kinase target domain in SPCH orthologs, suggests that brassinosteroids and/or ABA control stomatal development in seed plants. ??-CARBONIC ANHYDRASE 1 and ??-CARBONIC ANHYDRASE 4 control also both processes with double mutants in these genes showing impaired CO2-regulation of stomatal movements and increased stomatal density . Similarly, strigolactone may also regulate both stomatal development and function, with strigolactone-deficient and strigolactone-signalling max mutants showing lower ABA-regulation of stomatal closure and increased stomatal density . More studies are also required to know if other factors have the ability to coordinate the development and physiology. The interaction between stomatal signalling cascade and ABA signalling pathway must also be confirmed.