The drought-responsive ZmFDL1 gene regulates cuticle biosynthesis and cuticle-dependent leaf permeability

In higher plants, the outer surface of the aerial parts is covered by the cuticle, a complex lipid layer that constitutes a barrier against damages caused by environmental factors and provides protection against non-stomatal water loss. We show in this study that cuticle deposition, during the juvenile phase of in maize (Zea mays) plant development, and cuticle-dependent leaf permeability are controlled by the MYB transcription factor ZmMYB94/FUSED LEAVES1 (ZmFDL1). Biochemical analysis showed that in fdl1-1 mutant seedlings at the coleoptile stage both cutin and wax biosynthesis and deposition were altered. Among cutin compounds, ω-hydroxy fatty acids and polyhydroxy-fatty acids were specifically affected, while the reduction of epicuticular waxes, was mainly observed in primary long chain alcohols, and to a minor extent, long-chain wax esters. Transcriptome analysis allowed the identification of novel candidate genes involved in lipid metabolism and the assembly of a proposed pathway for cuticle biosynthesis in maize. Lack of ZmFDL1 affects the expression of genes located in different modules of the pathway and correspondence between gene transcriptional variations and biochemical defects have been highlighted. A decrease in cuticle-dependent leaf permeability was observed in maize seedlings exposed to drought as well as ABA treatment, which implies coordinated changes in the transcript levels of ZmFDL1 and associated genes. Overall, our results suggest that the response to water stress implies the activation of wax biosynthesis and the involvement of by both ZmFDL1 and ABA regulatory pathways. One-Sentence Summary Cuticle biosynthesis and cuticle-mediated drought-response during the juvenile phase of maize plant growth, are regulated by the MYB transcription factor fused leaves1 (ZmFDL1) and influenced by ABA.


INTRODUCTION
In higher plants, the outer surface of aerial parts, including vegetative organs, flowers, fruits, seeds and pollen grains, is constituted by a continuous hydrophobic layer termed cuticle, which consists of two major components, the polymer cutin and cuticular waxes. Cutin is a polymer of C16 to C18 hydroxylated fatty acids, that are directly cross-esterified to each other or via glycerol backbone (Fich et al., 2016). Cutin monomers are synthetized in the endoplasmic reticulum (ER), through different reactions, including the esterification of fatty acids to coenzyme A (CoA), ω-hydroxylation and further oxidation, glycerol-3phosphate transacylation, and export through the cell wall to the surface where polymerization occurs (Fich et al., 2016). Cuticular waxes are constituted by a complex mixture of very-long-chain fatty acids (VLCFAs), with more than 20 carbon atoms, and their derivatives, which include alcohols, aldehydes, alkanes, ketones and wax esters. They also include variable amounts of cyclic compounds, such as triterpenoids and phenylpropanoids (Bernard and Joubès, 2013;Lee and Suh, 2015a). The first phase of wax biosynthesis consists in the elongation of C16 and C18 fatty acids produced in the plastids to very-long-chain fatty acids (VLCFAs), with a chain length between C22 and C38, by elongase (FAE) complexes in the endoplasmic reticulum (Haslam and Kunst, 2013). Wax components are then produced through two different pathways.
Primary alcohols and esters are produced by the alcohol forming pathway, also termed acyl-reduction pathway (Rowland et al., 2006;Li et al., 2008), while alkanes, secondary n-alcohols and ketones are produced by the alkane forming pathway, also called decarbonylation pathway (Bernard et al., 2012). All wax compounds are synthesized in the epidermal cell layer (L1) and are secreted to cover the cell wall of epidermal cells, where they are embedded in the cutin and deposited on the cuticle surface as films or wax crystals. Waxes forming the outer layer of plant tissues are called epicuticular waxes.
Cuticle formation prevents post genital fusions among organs that grow very tightly appressed to each other when enclosed in vegetative shoots or within buds (Ingram and Nawrath, 2017). In addition, cuticle constitutes a constant barrier against damages caused by environmental abiotic and biotic factors, including UV light, temperature changes, pest and pathogens and a primary waterproof barrier that controls non-stomatal water loss and gas exchange from leaves (Yeats and Rose, 2013). Several studies in various plants have shown that changes in the amount and/or distribution of cuticular waxes lead to alterations in cuticular permeability. In tomato, it was shown through mechanical and genetic manipulations of the cuticular components that aliphatic constituents of the intracuticular wax layer have a key role in limiting transpiration rate across epidermis (Vogg et al., 2004). In Sorghum bicolor, an irradiate bloomless mutant with a high reduction in epicuticular waxes appeared to have a higher rate of epidermal permeability and night-time water loss (Burow et al., 2008). In rice, impairments in the organization of crystal waxes on the leaf surfaces of the wilted dwarf and lethal 1 (wdl1) mutant were correlated with a 2.3-fold increase in the transpiration rates and higher rates of water loss (Park et al., 2010). A positive correlation between an increase of wax amount and tolerance to drought stress was reported in Arabidopsis (Arabidopsis thaliana) (Aharoni et al., 2004;Kosma et al., 2009;Cui et al., 2016). Similarly, alkanes and primary alcohols were increased by drought treatment in two Australian wheat cultivars. Their study allowed the characterization of two MYB transcription factors that were activated by drought and in turn, stimulated the expression of cuticle biosynthesis-related genes (Bi et al., 2016). In rice, the OsGL1-6 gene, highly similar to Arabidopsis AtCER1 and involved in cuticular wax accumulation, was shown to be activated by drought (Zhou et al., 2015).
Finally, in maize seedlings, both cuticle permeability and drought sensitivity were shown to be increased by a mutation in the GLOSSY6 (ZmGL6) gene, the product of which is putatively involved in intracellular trafficking of cuticular waxes (Li et al., 2019).
In Arabidopsis, deposition of cuticular waxes in both leaves and stems is regulated by the AtMYB96, AtMYB30 and AtMYB94 transcription factors, that collectively modulate the expression of wax biosynthetic enzymes (Raffaele et al., 2008;Seo et al., 2011;Lee and Suh, 2015a). ABA, drought, and high salinity activate the expression of AtMYB96 (Seo et al., 2009) which in turn mediates the activation of cuticle biosynthetic genes to increase drought tolerance (Seo et al., 2011). Similarly, the expression level of AtMYB94 is also increased by drought, and transgenic Arabidopsis lines overexpressing AtMYB94 showed an increased accumulation of cuticular waxes and, under drought conditions, a reduction in the rate of cuticular transpiration in leaves (Lee and Suh, 2015b). We showed in a previous work that the maize ZmFDL1/MYB94 transcription factor is closely related to AtMYB96, AtMYB30 and AtMYB94 and is responsible for cuticle mediated post-genital organ separation during embryo development and early phases of seedling growth.
Lack of ZmFDL1 activity, in the fused leaves1-1 (fdl1-1) recessive mutant, specifically affects seedling development at early developmental stages and results in organ fusions, due to the lack of cuticular material in the boundary between organs, and irregular distribution of wax crystals on young leaf epidermal surface (La Rocca et al., 2015).
To further gain insights into the role of the ZmFDL1/MYB94, we compared in this study the cuticle composition of mutant and wild type seedlings and analysed the impact of the mutation on the transcriptome during early phases of seedling development. We also investigated ZmFDL1/MYB94 involvement in controlling cuticular permeability and in mediating drought stress response in maize seedlings. The isolation of genes involved in biosynthesis and transport of cuticular waxes and responsive to drought are of particular interest for crop breeding. The detailed functional characterization of the maize ZmFDL1/MYB94 regulatory gene proposed in this work will contribute to unravel the genetic-molecular mechanisms at the basis of the cuticle-mediated drought stress tolerance in an important cereal crop such as maize.

ZmFDL1 regulates cuticle deposition in a phase-dependent manner
The most evident defects observed in fdl1-1 homozygous mutants were the irregular coleoptile opening and presence of fusions between coleoptile and first leaf ( Fig. 1A; La Rocca et al., 2015). To characterize the effects of the fdl1-1 mutation on cuticle composition, cutin and epicuticular waxes were extracted from mutant and wild type seedlings and analysed by gas chromatography (Domergue et al., 2010;Bourdenx et al., 2011).
At the coleoptile stage, the total cutin and wax loads of fdl1-1 mutants were considerably reduced compared with those of wild type seedlings, but not completely abolished (Fig. 1B). In fdl1-1 coleoptiles, the reduction in total waxes was more severe than that observed for cutin (94% and 47% decreases, respectively;  Fig. S1B, S1C, S2M-T) were similar in fdl1-1 and wild-type plants. Reduction in cutin content was mainly due to decreases in ω-hydroxy fatty acids and polyhydroxy-fatty acids ( Fig. 1C and Supplemental Fig. S2E-L). The ω-hydroxy fatty acids content was more impaired than the polyhydroxy-fatty acids content (65% and 44% decreases, respectively) in the coleoptiles of the homozygous fdl1-1 mutant ( Fig 1C). Such differences were detected up to the second leaf stage of the seedling development (Supplemental Fig. S2E, F, I and J). Loss of ZmFDL1 activity had no substantial effect on the very-long-chain fatty acids (VLCFA) and fatty alcohols (Fig 1C) as only a few monomers were significantly different from wild type at the early stage of seedling development (Supplemental Fig. S2A, B).
The decrease in epicuticular wax load (Fig. 1B and Supplemental Fig. S1,S2) was mainly due to reduction in primary long-chain alcohols ( Fig. 1D and Supplemental Fig. S1, S2), which represent the major components of maize seedling waxes and are dominated by the C32 primary alcohol isomer. In coleoptiles, the C32 primary alcohol content was decreased by 98% in fdl1-1 when compared to wild type (Fig. 1D). The amounts of other minor primary long-chain alcohols with 30 or more carbons (C31 to C34OH) were also lower in fdl1-1, whereas the amounts of fatty alcohols with less than 30 carbons (C28 and C26OH) showed the opposite trend and were increased in fdl1-1 with respect to wild type seedlings (Supplemental Fig. S2Q, R).
These differences disappeared with seedling development, and in the third leaf, the amounts of primary alcohol were similar in fdl1-1 mutant and wild-type seedlings (Fig S1C; S2 Q-T). Alkanes and aldehydes were also analysed (Supplemental Fig. S2M-P) and found to be reduced only in the early stages of seedling development (Supplemental Fig. S2M,N). In particular, the C32 n-aldehyde was nearly absent in fdl1-1 coleoptiles whereas alkanes were less affected (Supplemental Fig. S2M). Finally, minor long-chain wax esters (WE) were detected at the coleoptile stage and their content was much lower in the homozygous fdl1-1 mutant than in wild type seedlings (Supplemental Fig. S1D). Altogether these data suggested that ZmMYB94 is a regulatory component of both cutin and wax biosynthesis and deposition. Moreover, as previously observed for visual mutant traits, ZmMYB94 related biochemical defects undergo a progressive reversion during seedling development, further confirming that ZmMYB94 activity occurs in a specific developmental phase (La Rocca et al., 2015).

Leaf permeability is altered in the fdl1-1 mutant
Cuticle-dependent leaf permeability was assessed by means of the chlorophyll leaching assay. The second leaf of fdl1-1 homozygous seedlings displayed a higher permeability compared to non-mutant B73 siblings ( Fig. 2A). Differences observed between the mutant and wild type leaf permeability progressively decreased in the second, third and fourth leaves (Fig. 2B). The differences were maintained in different genetic backgrounds, as observed in F2 progenies obtained after introgressing the mutation in the H99 and Mo17 inbred lines (Fig. 2C). To further understand the impact of the increased leaf epidermal permeability, thermography images of second fully expanded leaves of wild type (Fig. 2E) and fdl1-1 homozygous mutant ( Fig. 2F) seedlings were analysed. The fdl1-1 presented a reduced leaf temperature of about 1.5°C compared to its control (Fig. 2D). Moreover, a water loss time course experiment was performed on 10-old-days seedlings, by estimating the loss of weight with respect to the initial seedling fresh weight. The resulting profiles showed that homozygous fdl1-1 had a higher water loss rate compared to the wild type plants (Fig.   2G). We excluded that the observed differences in epidermal leaf permeability, leaf temperature and water loss values could be due to alterations in the stomatal index (Supplemental Fig. S4A, B).

Transcriptome profile of the fdl1-1 mutant
To investigate the regulatory network associated with ZmFDL1, cDNA libraries were produced from coleoptiles of fdl1-1 and wild type seedlings and sequenced on HiSeq™ 2000 platform. Six biological replicates, three for the fdl1-1 mutant and three for the wild type background respectively, were analysed.
Statistics concerning the total number of reads produced and the proportion of reads assigned to gene models according to the Zm00001d annotation of the ZmB73_RefGen_v4 reference assembly (Jiao et al., 2017) are reported in Supplementary Table S1. Identification of differentially expressed genes (DEGs) was performed by the latest versions of DESeq2 (Love et al., 2014) and Limma (Ritchie et al., 2015). A total of 2213 and 2399 genes were considered differentially expressed by Limma and DESeq2 respectively when a significance cutoff FDR of 0.05 was applied. Importantly, 1639 genes were differentially expressed according to both methods and were considered for subsequent analyses. Of these 612 were down-regulated and 1027 were up-regulated (Fig. 3A).
Functional enrichment analyses of differentially expressed genes, according to the Gene Ontology (GO) annotation of maize gene models reported by (Wimalanathan et al., 2018), was performed in order to identify biological processes and pathways that are differentially modulated in the fdl1-1 mutant. As reported in the Supplementary Tables S4 and S5 more than 200 ontology terms were significantly enriched in both up and down-regulated genes, after FDR correction, suggesting that the depletion of ZmFDL1 is likely to elicit a complex regulatory response. Importantly, ontology terms related to phenotypic traits observed in the fdl1-1 mutant, including: "GO:0010025: wax biosynthetic process", "GO:0005618: cell-wall", "GO:0009414: response to water deprivation" and "GO:0019216: regulation of lipid metabolic process" are highly enriched in both sets of up and down-regulated genes ( Fig. S3B, C), suggesting that differentially expressed genes identified by our analyses are likely to be involved in the modulation of wax/cutin biosynthesis and deposition in maize. The most significantly enriched GO terms associated with up-regulated genes include "GO:0042542: response to hydrogen peroxide", "GO:0009408: response to heat", "GO:0010286: heat acclimation", "GO:0009651: response to salt stress" and similar (Supplementary Table S3), suggesting that the majority of the genes up-regulated in the fdl1-1 mutant, might be involved in the response to osmotic and metabolic stress caused by impaired fdl1 activity. On the other hand, down-regulated genes are highly enriched in ontology terms associated with chloroplast and pigments, including for example: "GO:0009570: chloroplast stroma"," GO:0009941: chloroplast envelope", "GO:0016117: carotenoid biosynthetic process" and "GO:0015995: chlorophyll biosynthetic process".
Functional enrichment analyses of KEGG pathways (Moriya et al., 2007;Okuda et al., 2008) identified the "zma00195: Photosynthesis" and "zma04712: Circadian rhythm" pathways as those with the most statistically significant enrichment in up-and down-regulated genes, respectively. Fatty acid biosynthesis occurs in the plastid (Fig. 4) and conceivably an impairment of the downstream usage of the early cutin and wax monomer precursors could lead to alterations in the chloroplast physiology and produce an altered photosynthetic activity.
More interestingly and in line with previous observations, the pathways related to cutin and fatty acid metabolism among were the most enriched pathways in both genes significantly down-regulated (zma01212: Fatty acid metabolism, zma00061: Fatty acid biosynthesis and zma00073: Cutin, suberine and wax biosynthesis), and up-regulated (zma00561: Glycerolipid metabolism). All in all, the high level of agreement observed between the results of our functional enrichment analyses, which were performed according to three independent ontologies, are highly consistent with the idea that genes differentially modulated in the fdl1-1 mutant are likely to mediate the phenotypic traits observed in this developmental stage, as demonstrated by the systematic enrichment of ontology terms associated with wax, cutin and lipid metabolism in general.
Accordingly, we performed a careful manual annotation of our list of differentially expressed genes to identify candidate genes directly involved in cutin metabolism. Based on a combination of Gene Ontology, KEGG pathway and orthology predictions with Arabidopsis thaliana and Oryza sativa, as available from the phytozome annotation of the ZmB73_RefGen_v4 gene models, and the ARALIP Website (Li-Beisson et al., 2013) we identified 79 candidate DEGs which were tentatively assigned to 9 distinct lipids and cuticle-related biosynthetic processes, defined by expert manual curation (Supplemental Table S1). These genes were mainly related to the biosynthesis and the transport of secondary metabolites but also to the regulation of the gene expression. According to our careful manual annotation, 15 DEGs were assigned generically to the lipid metabolism, 9 DEGs to the glycerolipid metabolism and 5 DEGs to the glycerophospholipid metabolism.
Remaining genes were assigned to the putative biosynthetic pathways described below.
To validate the reproducibility of the gene expression data as obtained by the RNA-Seq analysis, expression patterns of 12 DEGs were investigated by quantitative Real-Time PCR. Notably, RNA-Seq expression data were confirmed for all the DEGs considered and displayed high levels of correlation with Real-Time PCR data (R 2 = 0.8) (Fig. 3C). Among the 12 DEGs selected for validation, 7 DEGs, which were differentially expressed in fdl1-1 compared to wild type seedlings at the coleoptile developmental stage, are putative cuticle-related genes ( Fig. 3E).

Drought affects cuticular permeability and the expression of cuticle-related genes
Wild type B73 seedlings were grown under either drought stress (LW), which was applied by withholding irrigation and maintaining RSWC at 30%, or normal water (NW) conditions. After 14 days, the morphological parameters of seedlings grown under restricted water were similar to those of control plants (Fig. 5A), indicating that the applied stress was mild. We performed chlorophyll leaching assays and observed that treated seedlings (LW) responded to the low water content by modifying the cuticle-dependent leaf permeability, which resulted lower compared to the control plants grown under NW condition (Fig. 5B). The same response was observed in wild type seedlings treated by applying a 1µM ABA solution directly to the roots up to 14 DAS. Chlorophyll leaching assay on the second leaf revealed, also in this case, that epidermal permeability decreased in ABA-treated plants compared to mock-treated control seedlings (Fig. 5C).
To investigate transcriptomic changes induced by the drought stress stimuli, maize seedlings were grown under normal-watered (NW) condition until 5 days after germination (DAS) and the next day the drought stress was imposed (Fig. 5D). To apply moderately stronger stress, RSWC was maintained at 30% of the total (Fig. 5E) for a shorter period. As a consequence, after 7 days (13 DAS) treated plants were visibly stunted compared to control seedlings (Fig. 5D). Every day, the second leaf from LW treated and NW control seedlings were sampled and gene expression analysis was performed for every time point.
Expression levels of the drought stress marker ZmRAB18 (Mao et al., 2015), which resulted higher in seedlings grown under LW compared to those grown under NW conditions (Fig. 5F), confirmed the effectiveness of the drought treatment. The expression profile of ZmFDL1, analysed in the second leaf of seedlings grown under NW condition, revealed that ZmFDL1 transcript accumulates during leaf expansion to reach a peak around 11 DAS (Fig. 5G, black line). In LW condition (Fig. 5G, grey line), ZmFDL1 transcript profile was different since the peak of maximum expression was detected earlier, after 2 days of drought stress ( Fig.   5D, 8DAS), exactly when the RSWC reached the 30% of the total (Fig. 5E). At subsequent time points, the ZmFDL1 transcript level decreased to be lower in LW than in NW growth conditions, eventually.
The ZmGL15 transcription factor expression profile nicely mirrored that observed for ZmFDL1 (Fig.   5H). An increase in transcript accumulation was detected during leaf growth (

ABA treatment affects ZmFDL1 and wax-related genes expression
We previously showed that plants responded to ABA treatment by reducing leaf permeability. To investigate the possible involvement of ABA in the control of cuticle-dependent leaf permeability in maize seedling, we analyse a maize mutant in the VIVIPAROUS1 (ZmVP1) gene (vp1-1). The transcription factor VP1, is the orthologue of the Arabidopsis ABA INSENSITIVE3 (AtABI3) gene (Giraudat et al., 1992), and mutants in this gene have been reported to have a reduced sensitivity to ABA (McCarty et al., 1991;Suzuki et al., 2003;Cao et al., 2007). Interestingly we noticed that the vp1-1 mutant seedlings showed decreased leaf permeability compared with wild type controls (Fig. 6A). Subsequently, to investigate the possible interplay between ABA and ZmFDL1 regulated pathways, we evaluated the effect of exogenous application of the ABA hormone on the fdl1-1 mutant. After 14 days of a treatment consisting in the application of a 1µM ABA solution (directly to the root as for the wild type; Fig. 5C) no significant reduction of cuticle dependent leaf permeability was observed, according to a chlorophyll leaching assay, in fdl1-1 mutant seedling compared to mock-treated control (Fig. 6B).
Comparisons of expression profiles in ABA and control-treated wild type plants showed a consistent down-regulation of ZmFDL1 upon treatments with increasing concentrations of the hormone after both 24 and 48 hours of treatment (Fig. 6E). Importantly, seedlings of both fdl1-1 and wild type control plants responded to ABA with a decrease in growth, suggesting that the perception of ABA was not altered (Fig 6D). Consistent with our previous observations based on RNA-Seq data (Supplemental Table S1) and quantitative expression analysis (Fig. 3D), the ZmGL15 and ZmCER4 genes were down-regulated in un-treated fdl1-1 seedlings compared to un-treated control wild type plants while ZmCER1 was up-regulated ( Fig. 6F, G, H). The same trends were observed also upon treatment with ABA. It is remarkable that after 48 hours of ABA treatment, the reduction in expression of ZmGL15 and ZmCER4, compared to their untreated controls, was more pronounced. The difference between ABA and control-treated plants was about 90% and 53% for ZmGl15 (Fig. 6F), and 91% and 52% for ZmCER4 (Fig. 6G) in fdl1-1 mutant and wild type plants respectively.
Similarly, the up-regulation of ZmCER1 was higher in ABA treated fdl1-1 mutant plants (164%) than in wild type plants (102%) and was anticipated at 24 hours of treatment (Fig. 6H).

DISCUSSION
As shown in our previous study (La Rocca et al., 2015), maize plants lacking the ZmFDL1/MYB94 transcription factor are defective in seedling development since fusions, which occur specifically in embryonic primordia and seedling organs, impair coleoptile opening as well as leaf expansion. Such fusions were attributed to the irregular deposition of intervening cuticular material between juxtaposed cell walls and to the patchy distribution of epicuticular waxes on the epidermis of the young leaves. This primary phenotypic characterization suggested that ZmFDL1 may have an important role in controlling cuticle biosynthesis in maize. This hypothesis was corroborated by the fact that AtMYB96, AtMYB94 and AtMYB30, which are its closest Arabidopsis related genes, were all shown to be involved in the regulation of cuticular wax biosynthesis (Sheperd and Wynne Griffiths, 2006;Raffaele et al., 2008;Seo et al., 2011;Lee and Suh, 2015a).

ZmFDL1 is a key regulator of cuticle biosynthesis and deposition in maize
In the present work, a more thorough characterization, including biochemical as well as transcriptional analyses, has been undertaken to gain further insights into the role of ZmFDL1 in cuticle formation. The data herein obtained strongly confirm that ZmFDL1 is a key regulatory component of both cutin and wax biosynthesis and deposition in maize seedlings. Among cutin compounds, ω-hydroxy fatty acids and polyhydroxy-fatty acids were specifically affected in fdl1-1 coleoptiles. In epicuticular waxes, the reduction was mainly observed in primary long chain alcohols, although a reduction of long-chain wax esters was also detected (Supplemental Fig. S1D). In agreement with the pattern observed for the morphological traits (La Rocca et al., 2015), these biochemical defects appeared transiently in germinating mutant seedlings consistently down-regulation also by quantitative expression analysis (Fig. 3D).
The decrease in very long chain C32 primary alcohols (C32OH; Fig 1D) observed in mutant coleoptile waxes correlates well with the down regulation of ZmCER4 ( Fig. 3D and Supplemental Table S1), whose closest homolog is the AtCER4 which was shown to be involved in the alcohol forming pathway (Fig. 4) (Rowland et al., 2006). Similarly, in Arabidopsis AtMYB94 activates the transcription of AtFA3/AtCER4 and additional genes involved in wax biosynthesis (Lee and Suh, 2015b).
The ZmWSD11 gene (Fig. 4), which corresponds to the Arabidopsis FOLDED PETAL 1  Table S1) and a strong decrease in the content of the verylong-chain aldehydes (32ALD) (Supplemental Fig. SM), precursors of the very-long-chain alkanes, could possibly explain the discrepancy between the transcript level of ZmCER1, ZmWSD11 (Fig. 3D) and their relative products (Supplemental Fig. S2M, S1D).
The reduction of ω-hydroxy fatty acids and polyhydroxy-fatty acids (Supplemental Fig. S2I, E) was consistent with the low expression of genes involved in the biosynthesis of cutin precursors and cutin monomers, such as the ZmONI3, ZmHTH1, ZmCUS2 and ZmAAE16, ZmCER8, ZmLACS2 and ZmLACS4 (Fig. 4 and Supplemental Table S1).
Interestingly, in rice, OsONI3, showing high level of similarity with an Arabidopsis HOTHEAD (HTH) protein (Akiba et al., 2014), was suggested to encode an omega-hydroxy dehydrogenase catalysing the biosynthesis of the long chain α, ω-dicarboxylic fatty acids (Yephremov et al., 1999). Interestingly, the seedling morphology of the OsONI3 mutant highly resembles that of fdl1-1, showing regions of fusion involving embryo leaf primordia as well as shoot leaves. This may suggest that the proper synthesis of cutin related compounds is required to prevent post-germinative adhesion among organs.
Furthermore, the strong enrichment of down-regulated DEGs in the KEGG "zma04712: Circadian rhythm" pathway (Supplemental Table S2) could reflect the conserved function between ZmFDL1 and its Arabidopsis homolog AtMYB96, which was shown to be connected with the clock, to shape the circadian gating of abscisic acid (ABA) responses (Lee et al., 2016).

ZmFDL1 and associated cuticle-related genes mediate leaf permeability under drought and ABA treatments
We have shown in the present study that ZmFDL1, besides promoting organ separation, is required in maize seedling leaves to control cuticle-dependent leaf permeability, to reduce the trans-cuticular water loss and eventually to preserve the leaf water status (Fig. 2). Several studies have shown that cuticle formation responds to environmental conditions, suggesting that the cuticle retains an active role in the plant response to environmental stress conditions, among them water scarcity (Xue et al., 2017). Our study provides support to these findings and further indicates that in maize seedlings this response is mediated by ZmFDL1. It also provides evidence that additional regulators such as ZmGL15, as well the ABA hormone signalling pathway, are involved. Cuticle-dependent leaf permeability, as measured through chlorophyll leaching analysis on B73 maize seedlings, decreased under limited water supply as well as through the exogenous administration of the ABA hormone (Fig. 5A, B). Moreover, the application of a drought stress treatment (LW) to developing seedlings caused a change in the expression profile of ZmFDL1 and associated genes (Fig. 5, 7, and Supplemental Fig. S4).
Under limited watering conditions (LW), both ZmFDL1 and ZmGL15 showed an early peak of expression with respect to normally watered plants (NW). This observation applies also to ZmCER4, while the expression profiles of both ZmCER1, ZmWSD11 and ZmGL13 showed a neat peak that was not observed in NW condition. We may speculate that the effect of drought on the expression profile of ZmCER4 is modulated to a greater extent by the positive action exerted by ZmFDL1, which might also mediate the positive effect exerted by drought on the ZmGL13 transcript level. Differently, if we assume from our previous data that ZmFDL1 is a repressor of ZmCER1 and ZmWSD11, their up-regulation observed under drought might be independent from the action of this MYB factor. Overall our data indicate that genes involved in waxes biosynthesis (ZmCER1, ZmCER4 and ZmWSD11), and eventually the total amount of cuticular waxes, are increased in response to drought stress conditions, thus improving plant protection from water loss. This finding is in agreement with a number of studies conducted for wax-related Arabidopsis genes. One example is constituted by AtCER1, that controls the biosynthesis of very-long-chain-alkane (Bernard et al., 2012) and was correlated with drought stress responses (Bourdenx et al., 2011). Its overexpression in plants significantly increases the production of alkanes and leads to increased drought tolerance (Bourdenx et al., 2011).
Moreover, the expression of AtWAX2, which is involved in both cutin and wax biosynthesis, was shown to be induced by drought, ABA, low temperature and salinity in cucumber (Chen et al., 2003).
Our analysis suggests that ZmFDL1 and waxes-related genes, through increasing their transcriptional activity, provide a rapid response to the drought stimulus. This response seems transient, since at later stages, when the stress condition is more severe, as indicated by the high expression level of the ZmRAB18, their transcript levels are diminished. A negative feedback might be produced, which causes the repression of wax biosynthesis. The observed gene pattern might also reflect the pattern of leaf growth meaning that in leaves approaching an advanced developmental stage, cuticle biosynthesis is diminished.
The role of ABA, in mediating the response to water deficit has been known since many years (Bartels and Sunkar, 2005;Shinozaki and Yamaguchi-Shinozaki, 2007) and the effect of ABA treatment on cuticle composition and related gene expression has been reported in Arabidopsis (Kosma et al., 2009) and tomato (Martin et al., 2017). We observed in the present study that absence of ABA perception in vp1 mutant seedlings resulted in increased cuticle-dependent leaf permeability (Fig. 7), while a slight decrease in leaf permeability was obtained through the application of exogenous ABA treatment (Fig. 7), thus showing, for the first time in maize, the involvement of ABA in this process.
We also observed that ABA exogenous application to B73 wild type seedlings caused severe repression of ZmFDL1, after both 24 and 48 hours of treatment. As expected, the repression, at the transcriptional level, was also observed for both ZmGL15 and ZmCER4, although only after 48 hours of ABA treatment (Fig 6F, G). A different pattern was observed for ZmCER1 that was up-regulated in the ABA treated plants (Fig. 6E) might be due the high concentrations of the exogenous ABA applied, which are considerably higher than physiological levels. These experimental conditions, with increasing ABA concentration are plausibly mimicking severe drought stress (Dallmier and Stewart, 1992). It is remarkable that the change in gene expression of ZmGL15 and ZmCER4 caused by ABA treatment, although qualitatively similar in both wild type and fdl1-1 mutant plants, was considerably more pronounced in the latter. Taken together the data from the expression analysis strongly suggest the existence of two signalling pathways promoted by the ABA hormone for the regulation of cuticular wax biosynthesis, of which one is independent and a second requires the action of ZmFDL1. The two pathways may act additively in controlling the transcription of biosynthetic genes.
In summary (Fig. 7), we showed in this study that the ZmFDL1 transcription factor is a key regulator of cuticle deposition and mediates the active response of cuticle to water scarcity condition. ZmFDL1 sensitivity to drought stress, consisting in the up-regulation of its transcript, occurred very early during seedling leaf development, as observed in the second leaf not yet emerged (Fig. 5G). This early ZmFDL1 response, accompanied by the activation of the regulatory gene ZmGL15, may allow plants to rapidly cope with water scarcity conditions in young tissues, by stimulating the activity of genes involved in cuticle biosynthesis, and more specifically in the wax deposition. A reduction in non-stomatal evapotranspiration is thus achieved that would prevent water deficit. It appeared, however, that the FDL1 mediated response is less effective at later developmental stages as well as when the severity of the imposed stress is high. We also showed that in maize seedlings ABA perception (signalling) influences cuticle dependent leafpermeability and ABA has a negative effect on ZmFDL1 transcription. Although mechanisms of interaction between the hormone signalling pathway and the ZmFDL1 regulatory pathway remains to be elucidated, our model suggests that the two pathways interact to modulate the expression of cuticle related genes.

Plant materials and growth conditions
The maize (Zea mays) fdl1-1 mutant, first described by La Rocca et al (2015)., was introgressed into B73, H99 and Mo17 inbred lines. In all the experiments performed, homozygous mutants and their wild type control plants were from the same F2 segregating progeny. Plants were grown under long day photoperiod (16 h of light/8 h of dark) in a growth chamber with controlled temperature (25 °C night, 30°C day) and with photon fluence of 70 µmol m -2 s -1 in pots containing S.Q.10 substrate (peat, sand, compost) (Vigorplant).
For drought treatment, maize seedlings were germinated and grown in soil under normal-watered (NW) condition (RSWC=80%) until 5 days after germination (DAS) and the next day (6 DAS, day 0) the drought stress was imposed by withholding irrigation. To monitor the drought stress level, the relative soil water contents (RSWC) were measured (Janeczko et at. 2016), as previously described (Castorina et al., 2018). To keep the plants under NW or low-watered (LW) conditions the RSWC was measured every day and the pots were watered as necessary until the end of the experiments. For the cuticle-dependent epidermal permeability assay the drought (RSWC=40%) was imposed up to 18 DAS (Fig. 2D) and for each treatment 7 pots were used.
Whereas, for the expression analysis of cuticle related genes the drought (RSWC=30%) was imposed up to 13 DAS ( Fig. 6 and Supplementary Figure S4) and for each time point the second leaf from four/five independent plants per treatment were sampled.
To mimic drought stress, plants were treated with exogenous applications of the abscisic acid (ABA) hormone (Duchefa) as previously described by (Riboni et al., 2016). The fdl1-1 and B73 wild type seedlings at the coleoptile developmental stage has been treated with 1ml of ABA (1 µM) or mock control (0.001% v/v ethanol) solution. Starting from 4 DAS, the treatments have been dispensed, every other day, watering directly the roots up to 14 DAS. For each treatment (Mock and ABA) the second leaf, from a minimum of 7 independent plants per genotype, was taken (Fig. 2E, Supplementary Figure S3).
To analyse the transcript regulation effect of the ABA on ZmFDL1 and other cuticle-related genes, B73 wild type and fdl1-1 seedlings have been treated dipping only the root apparatus into an ABA (1, 10 and 100 µM) or mock control solution. After 24 and 48 hours of treatment, only the green tissues (coleoptile and leaves) from a minimum of six independent plants per genotype have been sampled for subsequent total RNA extraction.

Cuticular Cutin and Wax Analysis
Cuticle composition was analysed in fdl1-1 and wild type seedlings at succeeding developmental stages: coleoptile stages, first emerging leaf, second emerging leaf and third emerging leaf at 2, 6, 10 and 14 days after sowing, respectively. Cutin and epicuticular waxes were extracted and identified by the combination of gas chromatography on column injection (GC-OCI) and gas chromatography-mass spectrometry (GC-MS) performed according to previously described methods (Domergue et al., 2010;Bourdenx et al., 2011).

Chlorophyll leaching assay, leaf temperature and water loss
For the chlorophyll leaching analysis, second, third and fourth fully expanded leaves were taken from a minimum of five independent plants per genotype and dissected into pieces of 8 cm in length as measured from the apex. Leaf sectors were weighed, immersed in ethanol 80% and incubated in the dark at room temperature. Absorbance was measured every day at 647 and 664 nm with a spectrophotometer (Agilent Technologies Cary 60 UV-Vis) to quantify the chlorophyll released in the solution and measurements were performed until complete chlorophyll extraction. The micromolar concentration of chlorophyll was calculated using the equation: total micromoles chlorophyll = 7.93 x A664 + 19.53 x A647 (Lolle et al., 1997).
The data obtained were normalized per gram of fresh weight and area of the 8 cm long leaf pieces to be expressed as a percentage of the total chlorophyll.
Thermal images of second fully expanded leaves were taken from 14 old-days fdl1-1 and wild type seedlings with a semi-automated long-wave infrared cameras system (FLIR T650sc). The temperature of the leaves was measured using the FLIR ResearchIR Max software.
To determine the time course of the seedling water loss, 10-old-days seedlings were detached and weighed immediately. Seedling weight was then estimated at designated time intervals and water loss was calculated as the percentage of fresh weight (FW) based on the initial weight. Several biological replicates were measured for each genotype.

Cell density and stomatal index analysis
To measure stomatal density and stomatal index, a leaf surface imprint method was used as previously described (Castorina et al., 2018). We analysed second, third and fourth fully expanded leaves of fdl1-1 mutant and wild type plants. Stomatal index (SI) was determined as [number of stomata / (number of epidermal cells + number of stomata)] × 100. One-way ANOVA test was performed with the statistical package SPSS 21.0.

RNA, cDNA Preparation and quantitative gene expression analysis
Total RNA was extracted from maize tissues using the TRIzol Reagent (Life Technologies) and treated with