Rewiring of hormones and light response pathways underlies the inhibition of stomatal development in an amphibious plant Rorippa aquatica underwater

Land plants have evolved the ability to cope with submergence. Amphibious plants are adapted to both aerial and aquatic environments through phenotypic plasticity in leaf form and function, known as heterophylly. In general, underwater leaves of amphibious plants are devoid of stomata, yet their molecular regulatory mechanisms remain elusive. Using the emerging model of the Brassicaceae amphibious species Rorippa aquatica, we lay the foundation for the molecular physiological basis of the submergence-triggered inhibition of stomatal development. A series of temperature shift experiments showed that submergence-induced inhibition of stomatal development is largely uncoupled from morphological heterophylly and likely regulated by independent pathways. Submergence-responsive transcriptome analysis revealed rapid reprogramming of gene expression, exemplified by the suppression of RaSPEECHLESS and RaMUTE within 1 h and the involvement of light and hormones in the developmental switch from terrestrial to submerged leaves. Further physiological studies place ethylene as a central regulator of the submergence-triggered inhibition of stomatal development. Surprisingly, red and blue light have opposing functions in this process: blue light promotes, whereas red light inhibits stomatal development, through influencing the ethylene pathway. Finally, jasmonic acid counteracts the inhibition of stomatal development, which can be attenuated by the red light. The actions and interactions of light and hormone pathways in regulating stomatal development in R. aquatica are different from those in the terrestrial species, Arabidopsis thaliana. Thus, our work suggests that extensive rewiring events of red light to ethylene signaling might underlie the evolutionary adaption to water environment in Brassicaceae.


In brief
The Brassicaceae amphibious plant Rorippa aquatica dramatically changes leaf morphology and function under water. Ikematsu et al. reveal that submergence triggers rapid inhibition of stomatal development through the redlight-mediated ethylene pathway. Blue light counteracts this, and jasmonic acid attenuates this adaptive response to submergence.

INTRODUCTION
For terrestrial plants, flooding is a threat to their survival. 1 Plants evolved diverse responses to cope with submergence. For example, deepwater rice plants escape from submergence by rapidly elongating their internodes. 2,3 Some plant species, on the other hand, endure submergence stress by slowing down their metabolism. 4,5 At the extreme end of the spectrum are aquatic plants-a group of land plants that have ''returned'' to the underwater environment. 6,7 Many aquatic plants perform gas exchange by diffusion through the epidermis devoid of a cuticular layer. 8 The water's edge environment, such as river and lakesides, is a harsh condition for both terrestrial and aquatic plants because the water levels tend to fluctuate easily. 9 Amphibious plants are adapted to such fluctuating environments, with many of them exhibiting phenotypic plasticity of leaf forms, known as heterophylly. 10,11 Terrestrial leaves develop larger leaf blades, whereas submerged leaves share common features of the cylindrical shape. [12][13][14][15][16][17][18][19][20] Plant growth and development are largely shaped by light and hormones, and amphibious plants are no exception. For example, a physiological study of the amphibious plant Rotala hippuris (Lythraceae) has shown that increasing red to far-red light promoted leaves with submerged characteristics, with blue light making shifts toward terrestrial leaves. 19 Ethylene is a key hormone in plant submergence responses. Because ethylene gas does not diffuse under water, it accumulates within submerged plant tissues and triggers the submergence response. 21,22 On the contrary, ABA is responsible for the drought stress response, and its biosynthesis and endogenous levels tend to be reduced under submergence. 21,23 In amphibious plants, both ethylene and gibberellins (GAs) have been shown to induce submerged leaf formation in Hygrophila difformis (Acanthaceae), Ranunculus trichophyllus (Rananculaceae), and Callitriche palustris (Plantaginaceae), 14,15,18,24 whereas ABA promotes the differentiation of leaves with terrestrial characteristics. [13][14][15]25 Stomata are cellular valves on the shoot epidermis for efficient gas exchange and water control, and the evolution of stomata is considered a key developmental innovation of land plants in the terrestrial environment. 26 Aquatic plants, on the contrary, do not develop stomata, as gas exchange through stomata does not work under water. 1,27 Zostera marina, for instance, has lost a set of key regulatory genes driving stomatal differentiation, 28 including those encoding basic helix-loop-helix (bHLH) transcription factors SPEECHLESS (SPCH), MUTE, and FAMA, originally identified in the model plant Arabidopsis thaliana. [29][30][31] Amphibious plants flexibly modulate stomatal development depending on the environment. 15,16 Stomatal development is largely influenced by light and hormonal conditions, and the integration of light/hormone signaling pathways into the core stomatal developmental program is a topic of intense study in A. thaliana as well as grass species. [32][33][34] However, the molecular circuitry that integrates light and hormone signals into stomatal development in amphibious plant species under submergence remains largely unknown.
Here, we report light and hormonal regulation of submergencetriggered repression of stomatal development in the Brassicaceae amphibious plant Rorippa aquatica. 12 R. aquatica is closely related to A. thaliana. 35 It is an allotetraploid species and shares high synteny with Cardamine hirsuta. 36,37 R. aquatica thus serves as an accessible model to study signaling pathways underpinning the developmental innovation of terrestrial water-adaptive lifestyles, although transformation and gene editing pipelines are not yet established. Furthermore, the availability of whole genome sequencing data of R. aquatica 37 enables a rapid identification of orthologs of A. thaliana developmental regulatory genes (Figure S1). We show that submergence-induced inhibition of stomatal development is independent from temperature-dependent morphological heterophylly. Through a series of physiological and pharmacological studies, we discover a central role of ethylene in the submergence-induced inhibition of stomatal development in R. aquatica and opposing actions of red and blue light on the ethylene-mediated submergence response. We unravel a novel, antagonistic role for jasmonic acid (JA) in counteracting the red-light-mediated inhibition of stomatal development under water. Our work suggests that rewiring of ethylene and light signaling that feed into the repression of SPCH and MUTE expression underlies the adaptation for underwater environment in amphibious plant R. aquatica.

RESULTS
Submergence inhibits stomatal development in R. aquatica independent of heterophylly R. aquatica exhibits typical heterophylly in response to submergence (Figures 1 and S2). To evaluate the stomatal phenotype, we first observed mature leaves from a plant grown in terrestrial or underwater conditions at ambient, optimal temperature (25 C) (Figures S2A-S2F; Table S1; STAR Methods). Unlike the terrestrial mature leaves, the submerged mature leaves developed only a few stomata ( Figures S2A-S2F). Quantitative analyses confirmed the low stomatal density and stomatal index (number of stomata/(number of stomata and nonstomatal epidermal cells) 3 100) ( Figures S2E and S2F). In addition, epidermal pavement cells from submerged mature leaves are isodiametric and devoid of characteristic lobing ( Figures S2C and S2D).
It has been shown that GA drastically influences heterophylly and that temperature via a GA pathway can mimic the submergence response. 12 To delineate the relationships between heterophylly and stomatal development, we next examined the epidermis of mature leaves grown under different temperatures ( Figure 1). As reported, 12 at low ambient temperatures (20 C), mature terrestrial leaves exhibit a highly dissected, cylindrical appearance resembling submerged leaves ( Figures 1A and  1D). In contrast, at high ambient temperatures (30 C), mature leaves exhibit simple leaf blades even under submergence ( Figures 1C and 1F). Unexpectedly, the mature leaf epidermis from plants grown in the terrestrial (air) condition formed stomata regardless of the leaf shape due to different growth temperatures ( Figures 1G-1I). Under submergence, stomatal development was strongly inhibited, regardless of the leaf shape ( Figures 1J-1L). Likewise, epidermal pavement cells remained rectangular and unlobed under submergence regardless of the leaf shape ( Figures 1J-1L). Although a subtle increase in stomatal density/index was observed for the round submerged leaves at 30 C, it was far lower than that observed in terrestrial leaves ( Figures 1M and 1N).
We further tested the effects of GA or a GA biosynthesis inhibitor, paclobutrazol (Pac), 38 on submergence-triggered inhibition of stomatal development at optimal temperature ( Figure S3).
(G-L) Representative epidermis of R. aquatica mature leaves grown under terrestrial (G-I) and submerged conditions (J-L) at 20 C (G and J), 25 C (H and K), and 30 C (I and L). Scale bars, 50 mm. (M and N) Stomatal density (M) and stomatal index (N) of mature terrestrial and submerged leaves grown under different temperature. Data were obtained from 10 leaves for each terrestrial and submerged condition and observed abaxial side epidermis. Different letters indicate significant differences between conditions (p % 0.05; two-way ANOVA; Tukey's HSD test).
(O) Representative image of young terrestrial plantlet subjected to submergence experiments. The youngest premature leaf at the beginning time (day 0, blue arrowhead) was subjected to microscopy. Scale bars, 5 mm. (P and Q) Representative images of premature leaves after 7 days of terrestrial (P) or submergent (Q) treatment at 25 C. Note that these are still young leaves and thus do not exhibit dissected morphology as mature leaves shown in (B) and (D). Scale bars, 5 mm. See also Table S1. (R and S) Time course changes of stomatal density (R) and stomatal index (S) under submerged treatment. Data were obtained from 10 leaves for each condition and observed the abaxial side epidermis. Different letters indicate significant differences between conditions (p % 0.05; two-way ANOVA; Tukey's HSD test). See also Figure S2. The heatmaps present the mean log 2 (FC) values, relative to the corresponding levels in the terrestrial controls. Asterisks indicate significant differences (FDR < 0.05). See also Data S1K.
(legend continued on next page) ll OPEN ACCESS GA or Pac treatment did not confer discernible effects on the epidermal phenotypes, stomatal density, or stomatal index of submerged leaves, while it showed some effects in the terrestrial condition ( Figures S3D-S3F and S3J-S3N). Taken together, we reveal that submergence-triggered inhibition of stomatal development can be uncoupled from temperature-dependent morphological heterophylly.
Submergence rapidly triggers the inhibition of stomatal development To investigate the dynamics of stomatal formation upon submergence, we next conducted a submergence treatment using young seedlings that were initially grown in the terrestrial condition and sequentially observed the epidermal phenotypes of their youngest immature leaves ( Figure 1O, blue arrowheads; Figures S2G-S2Q). By day 7, these leaves show serrated leaf edges ( Figure 1Q). A significant decrease in stomatal density/stomatal index is already evident on these leaves only by day 4 (Figures 1R and 1S). Under the terrestrial condition, newly developing leaves show frequent asymmetric divisions of meristemoids in a characteristic inward spiral manner, forming anisocytic stomatal complexes highly resembling those of A. thaliana ( Figures S2L-S2O). 30 In contrast, the asymmetric divisions are strongly suppressed in developing leaves under submergence ( Figures S2P and S2Q). Some stomatal linage cells are arrested, likely at the late meristemoid to the GMC stage based on their triangle to round cell shapes ( Figures S2P and S2Q, arrowheads). Together, these results suggest that a decrease in stomatal development under submergence is caused by immediate inhibition of stomatal-lineage initiation and the additional decision point of meristemoid/GMC arrests.
Submergence induces rapid reprogramming in R. aquatica To understand the molecular basis of a submergence response in R. aquatica, we performed a transcriptome analysis during the submergence response. The youngest immature leaves of seedlings grown in the terrestrial condition ( Figure 1O) were submerged for 1 h or 4 days, and treated leaves were subjected to RNA sequencing (RNA-seq) (Figure 2A; Data S1; STAR Methods). We identified 9,316 (20%) differentially expressed genes (DEGs) between submerged and control terrestrial leaves (false discovery rate [FDR] < 0.01 and expression-level log 2 (FC) > 1). We next categorized the response pattern of each DEGs by 3 3 3 self-organizing map (SOM) clustering based on their log 2 (FC) values of the three time points including the virtual time 0 (log 2 FC = 0) ( Figure S4). Further, we classify the clusters into two groups: upregulated and downregulated ( Figures 2B and 2C). Subsequently, gene ontology (GO) enrichment analysis of DEGs in each cluster was performed (Data S1; STAR Methods).
Transient expression changes in the GO categories ''shade avoidance-related genes'' (SOM6: up in 1 h; down in 4 days) and ''red or far-red light responsive genes'' (SOM4: down in 1 h; up in 4 days) suggest that the response to light environment is rapidly altered by submergence ( Figures 2B and 2C). We also noted that hormone pathways are dramatically upregulated or downregulated upon submergence. Among them, ethylene biosynthesis and signaling genes are the most highly enriched in SOM3 (up in 1 h; maintained high for 4 days), indicating the immediate induction and continuous enhancement of ethylene response upon submergence. Likewise, GA genes are also enriched in SOM3, suggesting morphological heterophylly. On the other hand, GO categories for ABA metabolism and GA homeostasis are significantly enriched in the early downregulated cluster SOM4 ( Figure 2C). We also found that JA response genes are downregulated and then maintained at low expression levels (SOM7) ( Figure 2C), suggesting a possible antagonistic role for JA signaling/response during submerged leaf development.
As expected, the GO terms associated with stomatal development were enriched among the continuously repressed gene clusters, SOM8 and SOM9 ( Figure 2C). Among them, the GO category ''asymmetric division'' was only enriched in SOM8, whereas ''guard mother cell differentiation,'' the later step of stomatal development, was enriched only in the later cluster SOM9. This likely reflects the developmental trajectories of stomatal cell lineages. Consistently, we observed overall repression of R. aquatica orthologs of Arabidopsis stomatal development genes, which are found as paralogs due to allotetraploidy ( Figure 2D): e.g., stomatal bHLH genes RaSPCH_1/2, RaMUTE_1/2, RaFAMA_1/2, RaSCRM/ICE1_1/2, and RaSCRM2_1/2 and signaling components 32 RaTMM_1/2, RaERECTA (RaER), RaER-LIKEs (ERLs), and RaEPF1 ( Figure 2D). On the other hand, the expression levels of RaEPFL9 (RaSTOMAGEN), which are not expressed in the epidermis based on the knowledge from Arabidopsis, 39,40 are unaffected by submergence ( Figure S4D). Further, the qRT-PCR analysis highlighted that, among the stomatal-lineage-specific bHLH genes ( Figure S1), RaSPCH and RaMUTE expressions are downregulated within 1 h, whereas no significant downregulation of RaFAMA was observed, even after 24 h of submergence (Figure 2E). Both paralogs of RaSPCH1/2, RaMUTE1/2, and RaFAMA1/2 respond similarly to submergence (Figures S1B and S1C). This trend is consistent with the earliest roles of SPCH to initiate stomatal cell lineages, MUTE to switch from meristemoid to GMC, and then FAMA to promote guard cell differentiation in A. thaliana. 41 Collectively, our results demonstrate that submerging R. aquatica rapidly reprograms a suite of gene expressions underlying light signaling and plant hormones, most significantly ethylene, but also GA, ABA, and JA. Subsequently, repression of stomatal development genes follows.
Ethylene plays a key role in the inhibition of stomatal development under submergence To elucidate the mechanism of submergence-induced inhibition of stomatal differentiation, we first examined the effects of ethylene because ethylene-related genes are the most immediately and significantly upregulated in response to submergence ( Figure 2B). First, we treated the ethylene precursor ACC 42 or (E) Expression levels of three stomatal-lineage bHLH genes measured by qRT-PCR in premature leaves. The seedlings were placed under submergence for 1 h (top) and 24 h (bottom). Primer pairs that detect both paralogs were used for the analysis. Expression levels with three biological replicates were normalized with respect to that of RaUBQ5. The mean ± SD is shown. *p < 0.05, ***p < 0.001; n.s., not significant, based on two-tailed Student's t test. See also Figures S1 and S6. ethylene gas for plants growing in terrestrial conditions for 7 days. Both treatments induced a strong inhibition of stomatal development ( Figure 3A) with reduced stomatal density and index ( Figures 3C and 3D). Consistent with this result, both treatments significantly downregulate RaSPCH and RaMUTE expression even after 24 h, whereas no reduction of RaFAMA is observed (Figures 3E-3G). In addition, the ACC and ethylene treatment induced the development of isodiametric pavement cells without lobing and deeply serrated leaf blades, highly resembling the submerged leaves ( Figures 3A  and S5). Thus, ethylene signaling also influences leaf morphogenesis.
Next, we investigated how stomata formation is affected when the ethylene pathway is inhibited during submergence by taking a pharmacological approach. For this purpose, we used a known inhibitor of ethylene signaling, silver nitrate, 43 and ABA, which act antagonistically to ethylene. 44 Indeed, treatment with silver nitrate or ABA reverted the inhibition of stomatal formation under submergence ( Figure 3B). RaSPCH and RaMUTE expressions are partially maintained under submerged conditions ( Figures 3E and 3F). In addition, the epidermal pavement cells became larger and more lobed, recapitulating terrestrial leaf development even under the submergence ( Figure 3B). Taken together, our results emphasize that ethylene is necessary for the inhibition of stomatal development under aquatic conditions and is sufficient for evoking the reprogramming of submerged leaf development even under aerial conditions.
Opposing roles of red and blue light on the inhibition of stomatal development under submergence Light quality plays an important role in the submergence response in some heterophyllous plant species. 19 Consistently, we discovered the immediate changes in the expression levels of R. aquatica genes belonging to the ''shade avoidance'' and ''response to red or far-red light'' categories upon submergence ( Figure 2). We thus investigated the roles of light signaling in R. aquatica stomatal development in response to submergence ( Figure 4). Under the control terrestrial (air) environment, we found that red light promotes stomatal development just as white light does. Blue light further enhanced stomatal development with significantly higher stomatal density/index under the control terrestrial condition (Figures 4A-4C). Under submergence, similar to white light, stomatal formation was strongly suppressed when red light was irradiated ( Figures 4A-4C). By contrast, blue light irradiation during submergence triggered the differentiation of stomata as well as lobed and expanded pavement cells (Figures 4A-4C). Strikingly, simultaneous irradiation of red and blue light triggered the inhibition of stomatal development ( Figures 4A-4C), suggesting that the effect of red light is epistatic to that of blue light. Consistently, the expressions of RaSPCH and RaMUTE but not RaFAMA are rapidly repressed by 1 h of red or white light treatment, and their expression is further reduced by 24 h (Figures S6A and S6B). Blue light treatment did not reduce RaSPCH/RaMUTE expression, and red light acts epistatic to the blue light, corroborating with the stomatal phenotypes (Figures S6A and S6B).
To gain insights into the effect of light during submergence on gene expression, we subsequently performed the RNA-seq analysis after 1 h of submergence under various light exposure conditions: white (W), red (R), blue (B), and simultaneous red and blue irradiation (R + B) ( Figures 4D and 4E). White light treatment during submergence resulted in an overall increase in the expression of light and shade avoidance-related genes ( Figures 2B, 2C, and 4D). Red and blue light triggered opposing effects on such genes, including RaATHB2_1, RaBBX21_1, RaPIF1_2, RaPIF4_1/2, and RaZAT10_2 ( Figure 4D). In accordance with the epistatic effects of red light over blue light on submerged leaf morphogenesis, simultaneous red and blue light irradiation overall mimicked the red light effects on the expression of these genes (Figures 4A-4D). We also found that the blue light treatment reduced the expression of some hypoxia response genes, implying that blue light under submergence not only triggers the differentiation of aerial epidermis (fully developed stomata and pavement cells) but also may diminish hypoxia response ( Figure 4E).
Quantitative measurements of visible light in terrestrial versus submerged conditions did not reveal a substantial difference in blue and red light or the red/far-red ratio ( Figures 4F and  S7). In A. thaliana, both red and blue light promote stomatal development. 32,45 In contrast, our physiological and gene expression studies of R. aquatica highlight that red and blue light act antagonistically on stomatal development under submergence.

Blue light inhibits ethylene biosynthesis, whereas red light inhibits stomatal formation in an ethylenedependent manner
We discovered that ethylene plays a central role in submerged leaf epidermal morphogenesis (Figure 3). To further understand how opposing actions of red and blue light ( Figure 4) intersect with the ethylene pathway, we surveyed the expression of ethylene-related genes from our RNA-seq experiments under different light regimes with 1 h submergence compared with white light under a terrestrial condition. Overall, ethylene biosynthesis and ethylene-response genes are upregulated under white and red light but, on the contrary, downregulated or unaffected by the blue light ( Figure 5A). Simultaneous red and blue light irradiation reverted the inhibitory effects of blue light on the expression of these ethylene-related genes ( Figure 5A). Interestingly, even under the aerial conditions, red-light-induced ethylene biosynthesis gene expression, whereas blue light suppressed it ( Figure S6C). Thus, blue light suppresses ethylene biosynthesis and response, whereas red light promotes them to adapt to a submerged environment, with the red light pathway being epistatic to the blue light pathway.
Next, we sought to delineate the relationship between ethylene and light signaling. We have shown that both ACC and ethylene treatments inhibit stomatal development of aerial leaves under ambient light conditions (Figure 3). We further tested the effects of ACC and ethylene treatments under blue light. The application of ethylene suppressed the blue-lightmediated stomatal development ( Figure 5B), reducing both stomatal density and stomatal index of aerial leaves to levels indistinguishable from those of submerged leaves grown under normal white light conditions ( Figures 5D, 5E, S6D, and S6E). Paradoxically, however, ACC treatment was only partially suppressing the blue-light-promoted stomatal development ( Figures 5B, 5D, 5E, S6D, and S6E). Thus, blue light may interfere ll OPEN ACCESS with the ethylene biosynthesis process, likely at the oxidation step of ACC to ethylene.
The submergence-triggered inhibition of stomatal development was enhanced by red light irradiations, and the red light effects are downstream of the blue light effects (Figure 4). We thus postulated that red light triggers the inhibition of stomatal development by promoting the process downstream of ethylene perception. To address this hypothesis, we performed simultaneous treatments of red light and ethylene-response inhibitors, AgNO 3 and ABA ( Figure 5). Indeed, red light inhibition of stomatal development under submergence is nullified by AgNO 3 or ABA (Figures 5C-5E). Taken together, our results suggest that blue light inhibits ethylene  Figures 1R and 1S. (E-G) Expression levels of RaSPCH (E), RaMUTE (F), and RaFAMA (G) measured by qRT-PCR in 24 h-treated premature leaves. Expression levels with three biological replicates were normalized with respect to that of RaUBQ5. The mean ± SD is shown. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant based on one-sided Student's t test.
biosynthesis, whereas red light inhibition of stomatal development is dependent on ethylene signaling.
Antagonistic role of JA on submergence-triggered inhibition of stomatal development Our time course transcriptome analysis of submergence response revealed the downregulation of JA-responsive genes ( Figure 2C). We thus analyzed the role of JA during the submergence response in R. aquatica ( Figure 6). Strikingly, MeJA treatment to submerged plants suppressed the submergence-triggered inhibition of stomatal development ( Figure 6C), significantly increasing the stomatal density and stomatal index (Figures 6E and  6F). On the other hand, when red light and MeJA were treated simultaneously, the effect of JA was attenuated, resulting in decreased but not full inhibition of stomatal development under submergence ( Figures 6D-6F). These results indicate that JA counteracts submerged leaf morphogenesis in parallel with the red light pathway.
To further delineate the antagonistic actions of JA in red-lightinduced submergence leaf epidermal morphogenesis, we examined the JA-related gene expression under submergence with white versus red light irradiations (Figures 6G-6O). Consistent with the observed stomatal development, JA treatment significantly increased the expression of both RaSPCH and RaMUTE, which was counteracted by the red light treatment (Figures 6G  and 6H). The expression of a suite of JA-responsive genes was strongly downregulated under submergent conditions compared with terrestrial conditions (Figures 6J-6O). Such  Figure 2D. See also Figure S6 and Data S2A-S2D. (J) Deflection of irradiance distribution due to submergence. Excerpt of focused wavelength bands: blue (430-500 nm) and red to far-red (640-710 nm). Measurements per 1 nm are acquired three times, and the mean value is shown. See also Figure S7. ll OPEN ACCESS genes include R. aquatica orthologs of known JA biosynthetic/ catabolic genes, 46 such as RaASO, RaJMT, and RaJOX4, as well as JA-responsive genes RaVSP1, RaVSP2, and RaPDF1.2A. Interestingly, however, we did not observe the reduction of these JA-inducible gene expressions by simultaneous red light irradiation ( Figures 6J-6O). The results indicate that the red light counteracts with the JA-induced stomatal development in submerged leaves without directly impinging on the JA-responsive gene expression. On the basis of these findings, we conclude that JA acts in an antagonistic manner to submergence-induced inhibition of stomatal development in a separate pathway from the red-light-mediated processes.

DISCUSSION
Using R. aquatica, a Brassicaceae amphibious plant, we lay the foundation for the molecular and physiological basis of the submergence-triggered inhibition of stomatal development that can be uncoupled from the morphological heterophylly (Figures 1  and 7). Recently, comparative transcriptome analyses of submerged vs. terrestrial leaves were reported using amphibious Ranunculus trichophyllus 15 and C. palustris. 18 Both studies detected reduced expression of core stomatal regulators, including homologs of SPCH, MUTE, FAMA, SCRM, EPF1, and TMM, reflecting the dramatic reduction of stomata in their submerged leaves. 15,18 Instead of molecularly characterizing the astomatous submerged leaves, we profiled rapid transcriptome changes triggered by submergence ( Figure 2) and found that RaSPCH and RaMUTE, but not RaFAMA, are significantly downregulated within 1 h of submergence ( Figure 2). Consistently, stomatal cell-cell signaling genes, many of which are direct SPCH/MUTE targets in A. thailana, 47,48 are also repressed rapidly ( Figure 2). Thus, submergence-triggered inhibition of stomatal development in R. aquatica likely occurs via signaling Expression levels with three biological replicates were normalized with respect to that of RaUBQ5. The mean ± SD is shown. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant based on one-sided Student's t test. Our study highlights that R. aquatica species have utilized the ethylene module, which plays a major role in submergent responses of terrestrial plants 21,22 to suppress stomatal differentiation. Importantly, we found that ethylene or ACC treatment of the terrestrial-grown R. aquatica plants is sufficient to fully downregulate RaSPCH and RaMUTE expression (Figures 3A, 3E, and  3F). This indicates that the activation of the ethylene pathway alone is sufficient for the inhibition of stomatal differentiation without any additional need for physical submergent stress (such as hypoxia or osmotic stress) or other endogenous hormones. This is different from submerged leaf morphogenesis in C. palustris, which requires multiple hormones, including ethylene and GA. 18 Our temperature shift experiments demonstrated that GA-mediated morphological heterophylly and ethylene-mediated inhibition of stomatal development are largely separable pathways in R. aquatica (Figure 1). It would be interesting to determine if these two processes are merged in C. palustris and other amphibious species.
In A. thaliana, treatment of ACC has been shown to increase the number of stomata, suggesting that ethylene promotes stomatal differentiation. 49 A more recent study found that a loss-offunction mutation in CONSTITUTIVE TRIPLE RESPONSE1 (CTR1), which encodes a negative regulator of ethylene signaling, reduces the asymmetric division potential of a meristemoid. 50 Consequently, the stomatal index is elevated in ctr1 mutants. 50 These data in A. thaliana collectively suggest that ethylene negatively regulates the stem-cell-like state of meristemoids but permits differentiation of stomata. Indeed, the ACC treatment had no effect on SPCH amounts in A. thaliana. 50 These observations show contrasting results for the rapid inhibition of stomatal development by ethylene in R. aquatica, suggesting that extensive rewiring events of ethylene signaling have occurred for evolutionary adaptation to water environment in Brassicaceae.
Our study further identified the complex, antagonistic actions of red and blue light on submergence-triggered inhibition of stomatal development, all feeding into the regulation of ethylene biosynthesis and/or response (Figures 4, 5, and 6). Specifically, inhibition of stomatal development by submergence was observed only in the presence of red light. In contrast, blue light suppresses the submergence response and promotes stomatal development even under water. This is likely via inhibition of ethylene biosynthesis since blue light effects can be interrupted by ethylene treatment, while ethylene precursor ACC treatment only slightly diminishes the blue light effect ( Figure 5B). Consistently, red-light-induced and blue light reduced the expression of key ethylene biosynthetic enzyme genes, RaACS7, RaACS8, and RaACO1 (Figures 5A and S6C).
We observed that red light modestly induced the expression of ethylene biosynthesis genes even in the terrestrial condition (Figure S6C). If so, why does red light not inhibit stomatal development in terrestrial leaves? This could simply be due to the physical property of water (liquid) versus air (gas). It is well known that ethylene becomes trapped inside the plant under submergence as ethylene gas does not diffuse through water. 22 Perhaps, under the terrestrial condition, a minimally elevated ethylene gas by the red light will be diffused into the air so that it will not impact stomatal development. Alternatively, other mechanisms, such as feedbacks and post-transcriptional regulations, might operate in the terrestrial leaves.
Strikingly, our findings that red light inhibits stomatal development through the ethylene-response pathway in R. aquatica is opposite from what are known for the red light regulation of stomatal development and ethylene response in a terrestrial plant. In A. thaliana, red light positively regulates stomatal development via phytochrome B (PhyB) and downstream PHYTOCHROME INTERACTING FACTOR4 (PIF4). 51 Furthermore, PhyB negatively regulates ethylene biosynthesis and signaling via red-light-triggered direct association with and subsequent degradation of PIF4/5, which upregulate ACS, and ETHYLENE INSENSITIVE3 (EIN3), a key downstream transcription factor of ethylene signaling. 52,53 Taking account of these pathways, one could envision that R. aquatica has co-opted the PhyB pathways to reinforce the activities of downstream components (e.g., PIFs and EIN3). At a high ambient temperature, which mimics shade, A. thaliana PIF4 represses SPCH expression via direct binding to the SPCH promoter region, thereby reducing the number of stomata. 54 We found that red light treatment rapidly downregulates RaSPCH but upregulates RaPIF1, RaPIF3, and RaEIN3 within 1 h (Figures 4D, 5A, and S6A). While the lack of efficient transformation pipelines in R. aquatica hampers the functional analyses of these candidate genes, future molecular genetic studies may shed light on the conservation and innovation of PhyB signaling in R. aquatica. While the effects of red light on R. aquatica are opposite to those on A. thaliana, it may not be uncommon that amphibious plants utilize red light as a cue to promote submerged leaf morphogenesis (hence to inhibit stomatal development). Indeed, in the amphibious plant Rotala hippuris, a low red/far-red ratio or blue light irradiation induces the differentiation of aerial leaves, whereas a high red/far-red ratio induces submerged leaves with the inhibition of stomatal development. 19 Because water scatters and absorbs visible light, the underwater light environment is complex and shifts to a higher red/far-red light ratio. 55 However, in our laboratory, at shallow submergence conditions, we detected only slight changes in the blue light intensity as well as the red/far-red ratio (Figures 4F and S7A). Thus, it is rather unlikely that R. aquatica solely utilizes the red/far-red ratio for submerged leaf morphogenesis. This is consistent with the natural habitat of R. aquatica, which is in clear, slow-moving, shallow yet fluctuating water. 12 Perhaps, light and ethylene signaling intersection may be critical to ''amplify'' the slight differences in light cues in R. aquatica.
We discovered that JA (MeJA) counteracts the submergencetriggered inhibition of stomatal development, conferring derepression of RaSPCH and RaMUTE expression ( Figure 6). This JA effect can be attenuated by the red light treatment, whereas JA biosynthesis and responsive gene expressions are unaffected by the red light treatment. Thus, it is likely that JA promotes stomatal development via antagonizing with red-lightmediated ethylene pathways (Figure 7). Alternatively, it may be possible that the JA response requires blue light, and the red light effect may be an indirect consequence of the absence of blue light. In the cotyledons of A. thaliana seedlings, JA signaling has been shown to inhibit stomatal development. Consistently, loss-of-function mutants in the JA receptor COI1 or higher-order loss-of-function mutants of downstream MYC transcription factors myc2, myc3, and myc4 exhibit the deregulated stomatal cluster phenotype. 56 The exact molecular mechanism of how the JA pathway leads to the downregulation of SPCH and MUTE expression in A. thaliana remains to be explored.
The evolution of aquatic and amphibious plants has occurred independently multiple times in land plant lineages. 57 As such, it is likely an example of convergent or parallel evolution. Plant submergence responses elucidated thus far share common yet unique features-critical roles of ethylene that can be reinforced by red light. Leveraging the close phylogenetic relationships between R. aquatica and A. thaliana, future comparative studies may uncover the exact molecular circuitries of light and hormone signaling underpinning the developmental innovation in the fluctuating water environment.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
The authors declare no competing interests. temperature used for the submerged treatment was adjusted to the same temperature as the experimental condition before treatment (Table S1). To minimize the circadian effects, all samples for RNA extraction were harvested at same time of the day, between 2-3 pm.

Light treatment
Thirty seedlings in the same conditions as short-term submerged treatment (described above) were used for light treatment experiment. Treatment periods were 1 hour or 7 days as continuous light condition after transfer from the control white light and terrestrial condition. The cubic glass tanks were placed under LED light panels in dark growth chamber at 25 C and 80 % RH (Table S1). The LED light treatments were as follows: red light (R), blue light (B), or combination red: blue at 1:1 (RB). Fluorescent light units of growth chamber for white light (W) was used as a control. The LED panel was 165 x 165 mm in size and contained 400 lamp beads (CCS, Japan, ISL-150X150-RHB and ISL-150X150-FR). The light intensity was 40 mmolm -2 s -1 at plant leaf surface level. For RB light condition, we used the ISL-150X150-RHB LED panel which have evenly distributed red and blue LED beads and set light intensity as red: 20 and blue: 20 mmol m -2 s -1 (total: 40 mmol m -2 s -1 ). LA-105 spectrometer (NK system, Japan) was used to measure the spectrum of the various light sources.

RNA extraction and RT-qPCR Analysis
Six premature leaves at the indicated stage were harvested and flash frozen in liquid nitrogen, and the total leaf RNA was purified using RNeasy Plant Mini Kit with DNaseI digestion (Qiagen). Quantity and integrity of RAN was checked using a spectrophotometer and Agilent 2100 bioanalyzer (Agilent). Reverse transcription was conducted using ReverTra Ace (Toyobo, Osaka, Japan). Quantitative PCR was carried out using a Syber Fast qPCR kit (Kapa Biosystems, Wilmington, MA, USA) and QuantStudio 3 system (Thermo Fisher Scientific). Data were obtained from three technical and three biological replicates. The expression level of RaUBQ5 was used as an internal control. Primer sequences are listed in Table S2.

RNA-Seq Analysis and gene expression profiling
The following samples, each treatment triplicated (3 biological replicates), were subjected to RNA-seq analysis: aerial mock 1 hr; submerged mock 1 hr; aerial mock 96 hr; submerged mock 96 hr; aerial white light 1hr; submerged white light 1 hr; submerged blue light 1 hr; submerged red light 1 hr; submerged blue and red light 1 hr (see Table S3 for further details). RNA-seq libraries were prepared using the Illumina TruSeq Stranded RNA LT kit (Illumina), according to manufacturer instructions. In total, 27 libraries were prepared and sequenced using the NextSeq500 sequencing platform (Illumina) in accordance with the manufacturer's instructions. Approximately 20 million raw reads giving more than 1-2 Gb sequence data for each sample were obtained by single-end sequencing of 76 bp length. For read count statistics, see Table S3.The obtained reads were mapped to the reference R. aquatica genome with gene annotation data (https://doi.org/10.6084/m9.figshare.19207362) by TopHat2 61 and the htseq-counts script in the HTSeq library was used to count the reads. 62 Count data were subjected to trimmed mean of M-values (TMM) normalization in EdgeR. 63,68 Transcript expression profiles and DEGs were defined using EdgeR generalized linear models (GLMs). 63 Differential expression was calculated via fitting a GLM at the gene level using either treatment conditions. The threshold for DEGs in the 1 hour and 4 days submerged treatment experiment was; false discovery rate (FDR) of < 0.01, log 2 FC (vs terrestrial control) of <1 and >1. Total 9,230 genes were obtained as DEGs under the two conditions.