Drought is a threat to the quality and yield of grapevine in the world’s important wine grape growing regions (Mosedale et al., 2016; Van Leeuwen and Destrac-Irvine, 2017; Van Leeuwen et al., 2019). These regions are expected to have decreased precipitation with associated risks of developing soil water deficit in coming years (IPCC, 2014; Sherwood and Fu, 2014; Scholasch and Rienth, 2019). One adaptation strategy seen in plants to tolerate water limitation involves stomatal regulation of water loss (Hunt et al., 2010; Hughes et al., 2017; Bertolino et al., 2019; Caine et al., 2019; Dayer et al., 2020; Gambetta et al., 2020). Stomata are pores mainly located in the leaf epidermis. The opening of these pores controls leaf gas exchange (CO₂ uptake for photosynthesis and water loss via transpiration) and is regulated by changes in turgor pressure in the guard cells surrounding these pores. The two guard cells respond to a range of environmental signals, often in conflict with each other, and sometimes rapidly changing (e.g., humidity, CO₂ concentration, light). In drought-stressed grapevine, stomatal closure is triggered by hydraulic signals and maintained by abscisic acid following re-watering (Lovisolo et al., 2010; Tombesi et al., 2015). Genotypic variation for stomatal sensitivity to reduced water availability has been shown to exist in grapevine (Schultz, 2003; Soar et al., 2006; Bota et al., 2016; Villalobos-González et al., 2019; Faralli et al., 2021).

Stomatal density and distribution in the epidermal tissue also plays a critical role in determining transpiration rate per unit of leaf area (Hunt et al., 2010). Previous work focusing on natural variation for stomatal anatomical features provided evidence of a close negative relationship between plant water-use efficiency and stomatal density (Bertolino et al., 2019; Faralli et al., 2019). According to extensive studies carried out in Arabidopsis (Doheny-Adams et al., 2012; Franks et al., 2015; Hepworth et al., 2015; Lee et al., 2015), stomatal density and distribution are under the control of small cysteine-rich peptides (CRP) called epidermal patterning factors (EPFs) highly conserved in a wide range of higher plants (Lu et al., 2019). Three members of this family play a key role in the formation of stomata: EPF1, EPF2 and EPFL9. EPF2 and EPF1 are expressed in the epidermis, in the earlier and later stages of leaf development, respectively. EPF2 inhibits the formation of cells considered the precursors of stomata guard cells, while EPF1 inhibits the subsequent differentiation of these same precursors and induces asymmetric cell division (Hara et al., 2009). Epidermal Patterning Factor Like 9 (EPFL9), also known as STOMAGEN, plays an antagonist role with respect to EPF1 and EPF2 as it induces stomata formation (Kondo et al., 2010). EPF-peptides interact with two transmembrane receptors of epidermal cells, ERECTA and Too Many Mouths (TMM). While EPF1 and EPF2 activate the receptor complex which in turn induces a MAPKs (Mitogen-Activated Protein Kinases) cascade (Morales-Navarro et al., 2018; Zoulias et al., 2018) leading to the destabilization of important transcription factors involved in the formation of stomata (SPEECHLESS, MUTE, FAMA; Pillitteri et al., 2007; Chen et al., 2020), STOMAGEN inactivates it. STOMAGEN is the only known positive regulator of stomata produced in mesophyll, and was confirmed to act independently of EPF1 and EPF2 (Hunt et al., 2010; Kondo et al., 2010; Sugano et al., 2010; Ohki et al., 2011). Its activity is antagonized by that of EPF2, however, it is not well understood if the antagonistic action is due to the sharing of an identical binding site in the common receptor or to other mechanisms (Ohki et al., 2011). An evolutionary model suggests that EPFL9 may derive from the duplication of EPF1/2 with a subsequent alteration in the function (Shimada et al., 2011). This is confirmed by the fact that EPF1/2 are more widespread in higher plants compared to EPFL9 (Lu et al., 2019). Despite the different amino acid composition among the CRP different sub-classes and across species, the members of CRPs have in common a small size, a conserved N-terminal region that include an apoplast secretion signal and a functional C-terminal domain containing cysteine residues (Marshall et al., 2011).

Several functional genomics studies, based on the ectopic expression or silencing of EPF1, EPF2, or EPFL9, have recently demonstrated a highly conserved functional paradigm in Arabidopsis and cereals. In barley, Hughes et al. (2017) proved that HvEPF1 overexpression limits stomatal development. In a hexaploid bread wheat, Dunn et al. (2019) decreased stomatal density (SD) via the overexpression of TaEPF1 and TaEPF2 orthologues and demonstrated improvements in water-use efficiency without affecting yield when SD reduction was moderate. Similarly, in rice Caine et al. (2019) and Mohammed et al. (2019) elucidated the function of OsEPF1 adopting an over-expression approach. Adding to the studies on rice, Lu et al. (2019) confirmed the role of OsEPF1, OsEPF2 and OsEPF9 by a dual strategy, both over-expression and down-regulation via RNA interference. Yin et al. (2017) were the first to apply the genome editing technology in rice to disrupt OsEPFL9.

Gene editing via the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) (Jinek et al., 2012) is to date the most powerful tool for functional genomics studies in plants (Liu et al., 2016). CRISPR/Cas9 system can efficiently produce nucleotide mutations into precise positions in the genome through the combined action of a specific guide RNA and the Cas9 nuclease which cleaves the DNA eliciting the non-homologous end-joining (NHEJ) pathway for DNA repair (Podevin et al., 2013). NHEJ may produce knock-out (KO) mutants with random insertion or deletion (indels) of variable lengths at the Cas9 cleavage site causing frameshift mutations or loss of amino acids in protein-coding sequences. These KO mutants are perfect systems to prove the function of a candidate gene (Jain, 2015). This technology is steadily boosting (Hess et al., 2017; Anzalone et al., 2019) and, coupled with the advancements of in-vitro culture practices, represents a knowledge-based strategy for the genetic improvements of cultivated plants, with relevant advantages compared to traditional breeding (Chen et al., 2019).

In grapevine, CRISPR/Cas9 technology has been successfully applied to evaluate the function of genes involved in susceptibility or tolerance to diseases, mainly caused by fungal pathogens (Malnoy et al., 2016; Giacomelli et al., 2019; Li et al., 2020; Wan et al., 2020; Chen et al., 2021; Scintilla et al., 2021), or to enhance tolerance to cold stress (Wang et al., 2021).

In this study, we inactivated VvEPFL9-1 in a grapevine table grape variety, ‘Sugraone’, adopting a genome editing approach based on CRISPR/Cas9 technology. Different edited lines with a significant reduction in stomatal density were produced and three of them were analyzed to investigate how reducing stomatal density affects grapevine physiological performance under different environmental conditions.

Crops worldwide will experience warmer conditions in the next decades, followed by limited water availability and increasing atmospheric CO₂ concentration (McGranahan and Poling, 2018). Alteration of stomatal density and stomatal size through the genetic manipulation of epidermal patterning factors has been shown to be an effective approach to increase drought tolerance and reduce water loss in several species (Bertolino et al., 2019; Buckley et al., 2020). There is a lot of knowledge about EPF gene family in Arabidopsis and in domesticated grasses but in perennial crops, which present genetic and physiological differences compared to annual species due to ecological and agronomic peculiar features (Lundgren and Marais, 2020), no evidence has been collected on their role. The aim of our study was to shed light for the first time on the genetic basis of stomatal density traits in grapevine, a perennial woody fruit plant with a longer lifespan than previously studied crops (i.e., longer than 30 years).

Water conservation, higher _iWUE and enhanced tolerance to multiple stresses (e.g., drought stress combined with heat stress) were achieved in Arabidopsis and grasses overexpressing EPF1/EPF2 or down-regulating EPFL9, due to a reduction in stomatal density (Franks et al., 2015; Hughes et al., 2017; Caine et al., 2019; Dunn et al., 2019; Lu et al., 2019). Between these two reverse genetics approaches, we have chosen the second, relying on the knock-out of VvEPFL9 by the powerful CRISPR/Cas9 gene editing technology.

In the grapevine genus we found two AtEPFL9 orthologs, we named VvEPFL9-1 and VvEPFL9-2, identical at 82% in the protein region corresponding to the functional peptide and, respectively, sharing 82 and 95% identity with the same region of AtEPFL9 peptide. So far, two EPFL9 paralogs have been found in maize and rice (Yin et al., 2017; Hepworth et al., 2018; Lu et al., 2019), showing, respectively, 84 and 73% (ZmEPFL9-1 and ZmEPFL9-2) and 82 and 73% (OsEPFL9-1 and OsEPFL9-2) identities to AtEPFL9 functional peptide. It has been suggested that EPFL9 paralogs in cereals might be functionally divergent (Lu et al., 2019) but definitive evidence indicating a different function has never been produced. In the study of Lu et al. (2019), the approach used to silence OsEPF9-1 was RNA interference with a 450 bp-long hairpin RNA, which hardly discriminated between the two variants. In our study, we decided to focus on VvEPFL9-1, since at the time the experiment was designed, VvEPFL9-2 was not anchored to any chromosome in the grapevine reference genome and this uncertainty oriented our choice on VvEPFL9-1. According to our data, the knock-out of VvEPFL9-1 can reduce stomatal density by up to 60%, leading to the hypothesis that VvEPFL9-1 and VvEPFL9-2 could be both involved in stomatal induction with a redundant function. A similar approach based on CRISPR/Cas9 technology to knock-out EPFL9 in rice achieved nearly 90% of stomatal density reduction compared to control by targeting a site on the first exon encoding for the signal peptide and thus not discriminating between OsEPFL9 paralogs (Yin et al., 2017).

Our study also confirms the crucial role of cysteine residues in the C-terminal functional peptide. This is demonstrated by the lines S-epfl9KO5 and S-epfl9KO8 in which the loss of the second cysteine (due to a 3 bp-deletion or single base substitution) resulted in a stomatal density reduction similar to the one gained by a full frameshift of the coding sequence. This is consistent with the finding of Ohki et al. (2011) who observed that impairing the formation of a disulphide bond prevented the correct protein folding and function. The design of a sgRNA that directed Cas9 cleavage next to the nucleotide triplet coding for the second cysteine proved to be a good choice for effective 3- and 6- bp deletions. Moreover, our data showed that the retention of almost 50% functional VvEPFL9-1 in some transgenic lines (S-epfl9KO1 and S-epfl9KO5) due to a partial editing of the target site, with a substantial maintenance of a WT peptide, still resulted in a significant decrease of SD, suggesting that a threshold amount of peptide may be required for EPFL9-1 to be functionally effective.

Reduction in stomatal density following VvEPFL9-1 knock-out was significant, although partially compensated by an increase in stomatal size (SS, inferred by pore length measurements). The negative yet non-linear association between SD and SS has been frequently reported in many species (Franks and Beerling, 2009) and often linked to an improved economy of epidermal space allocation with the combination of low SD and high SS as a preferable strategy when low stomatal conductance is required (Doheny-Adams et al., 2012; Lawson and McElwain, 2016). In our work, however, the reduction in SD was accompanied by only a partial compensation for SS.

Stomata are the main drivers of transpiration but at the same time are pivotal for CO₂ uptake for mesophyll photosynthesis (Lawson and Blatt, 2014). For instance, in barley and wheat, a reduction in SD by 50% compared to WT led to a significant reduction in carbon assimilation (A_sat) and conductance (g_s) and to an enhanced water use efficiency ( _iWUE) under optimal growth conditions (Hughes et al., 2017; Dunn et al., 2019). Similarly, in two-months-old ‘Sugraone’ at well-watered conditions (Experiment 1), we found that a 60% reduction in SD led to a reduced A_sat for the edited lines compared to the WT. Additionally, the reduction in g_s was even greater, leading to a higher value of _iWUE (i.e., A_sat/ g_s) in edited versus WT lines. Moreover, the reduction in A_sat was not concomitant to reductions in Rubisco velocity (V_cmax) or to impairment in electron transport chain (J_max) suggesting that the knock-out of VvEPFL9-1 did not affect the photosynthetic machinery, at least at the conditions applied in this work. In an additional experiment (Experiment 3) carried out under well-watered conditions in a phenotyping platform on older plants than those used in experiment 1, iWUE confirmed to be significantly improved in the edited line (S-epfl9KO6) compared to ‘Sugraone’ WT while transpiration (WU) performances were not significantly different. The main sources of variation between the two experiments were plant age and environmental conditions. In Experiment 3 plants were older than in Experiment 1 (12- vs. 2-Months-old) and regarding light conditions, Experiment 1 was carried out under the natural fluctuating light of a greenhouse, while in Experiment 3 plants were subject to a steady-state light pattern. Our results suggest that canopy structure (over-saturation of apical leaves and basal leaves under the sub-saturating light intensities of the greenhouse) may play a role in defining the effectiveness of a reduced stomatal density phenotype. Furthermore, the conditions of dynamic light intensity such as those present in the greenhouse, may have contributed to accentuate the water saving behavior of the lines with lower stomatal density. Indeed, reducing stomatal density can limit stomatal clustering (Harrison et al., 2020) and therefore increase stomatal responsiveness to environmental cues (Faralli et al., 2021). Important differences for g_s were also observed between Experiments 1 and 3, suggesting that plant age and pot-effect significantly influences operating g_s, although the g_s values are inside the ranges shown by Lavoie-Lamoureux et al. (2017) for pot-grown grapevine. Vitis vinifera genotypes with reduced SD and, in turn, limited A_sat and greater _iWUE, may be desirable to improve plant water conservation and to delay sugar accumulation under current and future climatic scenarios (Kuhn et al., 2014; Arrizabalaga-Arriazu et al., 2021). Sugars and organic acids along with various secondary metabolites (e.g., tannins, flavonols, anthocyanins, aroma compounds) are determinants of grape berry quality and their accumulation during berry ripening is the result of the interaction between genotype and environment, a relationship made vulnerable by climate change (Bobeica et al., 2015; Rienth et al., 2021). It is known that grapevine physiology will be impacted by elevated carbon dioxide, increasing temperatures, and extreme heat events during the growing season (De Cortázar-Atauri et al., 2017; Delrot et al., 2020). In particular, high temperature and increasing CO₂ levels are already affecting viticulture (Cook and Wolkovich, 2016; Mosedale et al., 2016; Edwards et al., 2017; Droulia and Charalampopoulos, 2021) with an evident shift towards an earlier onset of phenological stages (Edwards et al., 2017; Alikadic et al., 2019) and accelerated berry ripening (Jones et al., 2005; Parker et al., 2020; Rienth et al., 2021). High temperatures and water stress slow down vine metabolism resulting in a lower accumulation of polyphenols and aromatic compounds in the berries (Tomasi et al., 2011; Jones, 2013; Pons et al., 2017; Venios et al., 2020). Thus, one of the consequences of a compressed phenology may be an earlier sugar accumulation in the berries that leads to anticipated harvest dates when the secondary metabolites content is sub-optimal (Palliotti et al., 2014; Edwards et al., 2017). Although currently several agronomic approaches of source-limitation (i.e., pre-flowering leaf removal, shading nets, anti-transpirant application, etc.) have been set up to delay sugar accumulation in ripening grapes in the field (Palliotti et al., 2014; Prats-Llinàs et al., 2020), stomatal manipulation may be a favorable genetic strategy for the future, that deserves to be further explored also under combined environmental stress and in field trials. In our study, we further applied a water stress experiment to test if and how a reduced stomatal density can affect plant behavior in drought conditions. During a progressive reduction in soil water availability, significant differences in transpiration rate were observed in edited lines compared to WT only under moderate water stress (i.e., DASA 3 and 4). Yet, under severe water stress (e.g., DASA 10–12), some trends (p < 0.1) were observed in edited lines showing higher transpiration rate followed by A_sat and g_s maintenance. Notably, the reduction in g_s and A_sat during the dry-down was evident for WT plants (p < 0.001) while this was not significant for edited lines. This conservative behavior induced by reduced SD has been previously associated with a longer period of transpiration maintenance during drought, leading to a prolonged carbon assimilation respect to WT (Caine et al., 2019). In rice, lines overexpressing the OsEPF1 gene had higher yield than WT when water-stressed at flowering stage (Caine et al., 2019) confirming that water conservation during key-stages of yield formation may be desirable for yield maintenance (Faralli et al., 2019). In addition, limiting plant transpiration could be an advantage for irrigated vineyards in terms of a reduction in water input demand (Keller et al., 2016). In view of an increase in the number of grapevine growing regions where water resources will become limited (Schultz, 2000; Santillán et al., 2020), genotypes with reduced stomatal density will require less units of irrigation water for cultivation area, thus increasing crop water productivity for farmers (Scholasch and Rienth, 2019).

To our knowledge, this is the first study describing the function of VvEPFL9-1 in a perennial fruit crop as well as the physiological advantages of epfl9-1 knocked-out phenotype under different availability of soil water. In grapevine, reducing stomatal density via VvEPFL9-1 loss of function can induce water conservation and increase _iWUE, although an impact of photosynthetic CO₂ absorbance (A_sat) was observed in some edited lines. While in several crops, reduced photosynthetic CO₂ uptake can decrease yield and biomass, we speculate that reduced A_sat and increased _iWUE may be a favorable combination of physiological attributes in grapevine, especially under future climate change scenario. However, at this stage further trials in the field under standard management conditions are required as well as additional evaluations regarding the potential effects of reduced stomatal density under natural environmental fluctuations. To conclude, this work reinforces the concept that stomatal anatomical features constitute a promising target for designing climate change-resilient crops (Franks et al., 2015; Hughes et al., 2017; Bertolino et al., 2019; Caine et al., 2019; Dunn et al., 2019; Lu et al., 2019; Buckley et al., 2020) and provides evidence of this in grapevine, the most economically important fruit crop globally.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material. The original contributions presented in the study are publicly available. This data can be found here: NCBI Sequence Read Archive, BioProject accession number: PRJNA820619.

MC performed plant transformation experiments, plant molecular analysis, phenotyping, statistical analysis, and wrote the paper. MF contributed to the dry-down experiment design, carried out Experiment 3, supervised physiological analysis and statistical elaboration of the data, and wrote the paper. JL carried out alignments and phylogenetic tree and revised the manuscript. LB performed carbon isotope composition analysis and revised the manuscript. SP performed the analysis for the T-DNA integration point determination. CV, MM, WO, and AR supervised and revised the manuscript. LDC conceived the project, designed vectors for gene editing, performed paralogs analysis, transformation experiments, plant molecular characterization, took care of the plants in greenhouse, and wrote the paper. All authors contributed to the article and approved the submitted version.

This research was funded by the Autonomous Province of Trento (Italy) in the framework of the Fondazione Edmund Mach (FEM) International PhD initiative and by the ERDF 2014–2020 Program of the Autonomous Province of Trento with EU co-financing (Fruitomics, CUP number: C49H18000000001).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.