strigolactone-miR156 module controls stomatal behaviour during drought recovery

miR156 is a conserved microRNA whose role and induction mechanisms under stress are poorly 35 known. Strigolactones are phytohormones needed in shoots for drought acclimation. They promote stomatal closure ABA-dependently and independently; however, downstream effectors for the former have not been identified. Linkage between miR156 and strigolactones under stress 38 has not been reported. We compared ABA accumulation and sensitivity as well as performances of wt and miR156- 40 overexpressing (miR156-oe) tomato plants during drought. We also quantified miR156 levels in wt, 41 strigolactone-depleted and strigolactone-treated plants, exposed to drought stress. 42 Under irrigated conditions, miR156 overexpression and strigolactone treatment led to lower 43 stomatal conductance and higher ABA sensitivity. Exogenous strigolactones were sufficient for 44 miR156 accumulation in leaves, while endogenous strigolactones were required for miR156 45 induction by drought. The “after-effect” of drought, by which stomata do not completely re-open 46 after rewatering, was enhanced by both strigolactones and miR156. The transcript profiles of 47 several miR156 targets were altered in strigolactone-depleted plants. 48 Our results show that strigolactones act as a molecular link between drought and miR156 in 49 tomato, and identify miR156 as a mediator of ABA-dependent effect of strigolactones on the after- 50 effect of drought on stomata. Thus, we provide insights into both strigolactone and miR156 action 51 on stomata. ABSTRACT miR156 is a conserved microRNA whose role and induction mechanisms under stress are poorly 86 known. Strigolactones are phytohormones needed in shoots for drought acclimation. They 87 promote stomatal closure ABA-dependently and independently; however, downstream effectors 88 for the former have not been identified. Linkage between miR156 and strigolactones under stress 89 has not been reported. 90 We compared ABA accumulation and sensitivity as well as performances of wt and miR156- 91 overexpressing (miR156-oe) tomato plants during drought. We also quantified miR156 levels in wt, 92 strigolactone-depleted and strigolactone-treated plants, exposed to drought stress. 93 Under irrigated conditions, miR156 overexpression and strigolactone treatment led to lower 94 stomatal conductance and higher ABA sensitivity. Exogenous strigolactones were sufficient for 95 miR156 accumulation in leaves, while endogenous strigolactones were required for miR156 96 induction by drought. The “after-effect” of drought, by which stomata do not completely re-open 97 after rewatering, was enhanced by both strigolactones and miR156. The transcript profiles of 98 several miR156 targets were altered in strigolactone-depleted plants. 99 Our results show that strigolactones act as a molecular link between drought and miR156 in tomato, and identify miR156 as a mediator of ABA-dependent effect of strigolactones on the after- effect of drought on stomata. Thus, we provide insights into both strigolactone and miR156 action on stomata.


INTRODUCTION 109
It has been estimated that more than 90% of water uptake in plants is lost through transpiration. 110 Stomata represent a fundamental checkpoint balancing the entry of carbon dioxide and the exit of 111 water, and their regulation is at the core of the main acclimation strategy to water scarcity 112 (Matthews, Vialet-Chabrand, & Lawson, 2017). Upon drought, tolerance mechanisms operate at 113 different spatial and temporal scales with rapid stomatal closure as the basis for preventing shoot 114 water loss (Tardieu, Simonneau, & Muller, 2018). MicroRNAs (miRNAs) and phytohormones have 115 been associated to the regulation of guard cell development and movement (Curaba, Singh, & 116 Bhalla, 2014;Ding, Tao, & Zhu, 2013). 117 miRNAs are a widespread class of endogenous, small RNA molecules (19-24 nt in length) that 118 negatively regulate gene expression at the transcriptional, post-transcriptional, and translational 119 levels (Nozawa, Miura, & Nei, 2012). In animals, under certain conditions, miRNAs are being 120 looked at as hormones, because of their cell-to-cell and also long-distance movement coupled to 121 signalling activity (Bayraktar, Van Roosbroeck, & Calin, 2017). While some plant miRNAs are 122 species-specific in terms of expression patterns and targets, others, including miR156, are very 123 conserved. The modulation of miR156 is crucial throughout development for correct leaf 124 formation, tillering/branching, plastochron, panicle/tassel architecture, and timing of age-125 dependent reproductive transition along with fruit ripening and fertility (Wang & Wang, 2015). In 126 spite of a few differences among species, the pattern of mature miR156 accumulation in response 127 to environmental stimuli is also rather conserved (Khraiwesh, Zhu, & Zhu, 2012). Indeed, miR156 is 128 consistently induced by a variety of abiotic stresses such as drought, osmotic stress, heat, cold, 129 salinity, and macro-nutrient deficiency (Cui, Shan, Shi, Gao, & Lin, 2014;Ding, Fromm, & 130 Avramova, 2012;Hsieh et al., 2009;H. H. Liu, Tian, Li, Wu, & Zheng, 2008;Stief et al., 2014). As for 131 its role under stress, it was initially postulated that miR156 increase is merely needed to stall 132 flowering until stress is over. This effect would be exerted via miR156-mediated inhibition of the 133 age-dependent pathway to flowering. In support of this view, miR156-overexpressing plants 134 (miR156-oe hereafter) are late-flowering. Under long-day, inductive conditions the flowering time 135 in Arabidopsis (Arabidopsis thaliana) tends to inversely correlate with miR156 levels in a range of 136 environmental conditions (Cui et al., 2014;May et al., 2013). However, it is also becoming clear 137 that this miRNA may have more direct functions in stress acclimation. For instance, it was shown 138 that miR156-oe plants outperform wt controls both in Medicago sativa under drought (Arshad, 139 Feyissa, Amyot, Aung, & Hannoufa, 2017) and in Arabidopsis under osmotic/high salinity stress 140 (Cui et al., 2014) or recurring heat (Stief et al., 2014). These functions may be exerted by miR156 141 through post-transcriptional repression of one or more members of the wide family of SPL 142 (SQUAMOSA PROMOTER BINDING PROTEIN-LIKE) transcription factors (Arshad et al., 2017;Cui et 143 al., 2014;Stief et al., 2014), the best characterised among miR156 targets. However, these studies 144 do not clarify which cascade of molecular events induces miR156 under stress, and whether 145 miR156 can affect stomatal regulation. 146 Abscisic acid (ABA) is the best-characterised phytohormone among several affecting stomatal 147 functioning. Its direct application, as well as endogenous synthesis in phloem companion, 148 mesophyll and guard cells, trigger stomatal closure (Bauer et al., 2013;Kuromori, Sugimoto, & 149 Shinozaki, 2014;S. A. M. McAdam & Brodribb, 2018;Merilo et al., 2018). ABA signalling is known 150 to affect miRNA production (Lian et al., 2018;Speth, Willing, Rausch, Schneeberger, & Laubinger, 151 2013;Yan et al., 2017), but to our knowledge, no miRNAs have been demonstrated to affect ABA 152 sensitivity. 153 Strigolactones, the most recently discovered class of phytohormones, also play a role in this 154 scenario. ABA and strigolactones share a carotenoid precursor, from which strigolactone synthesis 155 proceeds through a partially known series of enzymes including DWARF27 (D27), CAROTENOID 156 CLEAVAGE DIOXYGENASE7 (CCD7) and 8 (CCD8), and MORE AXILLARY GROWTH1 (MAX1) to 157 produce bioactive strigolactones (Waters, Gutjahr, Bennett, & Nelson, 2017). Strigolactones 158 modulate several aspects of plant development and interactions with rhizosphere organisms 159 (Andreo-Jimenez, Ruyter-Spira, Bouwmeester, & Lopez-Raez, 2015;Cardinale, Korwin Krukowski, 160 Schubert, & Visentin, 2018;Lanfranco, Fiorilli, Venice, & Bonfante, 2018;Lopez-Raez, 2016). 161 Furthermore, plants with defective strigolactone synthesis or signalling in several dicot plants are 162 hypersensitive to drought, salt and osmotic stress . In Arabidopsis, Lotus 163 japonicus and tomato, strigolactones positively control stomatal movements as components of a 164 systemic stress signal (Tardieu, 2016). Indeed, mutants in strigolactone biosynthesis exhibited 165 reduced stomatal closure (Ha et al., 2014;J. Liu et al., 2015;Visentin et al., 2016). Conversely, 166 enhanced stomatal closure and drought tolerance was observed in plants treated with exogenous 167 strigolactones (Lv et al., 2018;Visentin et al., 2016;Y. Zhang, Lv, & Wang, 2018) or in which shoot 168 strigolactone biosynthesis is increased by grafting onto a low-strigolactone rootstock (Visentin et 169 al., 2016). The effect of strigolactones on stomatal closure depends, at least in part, on ABA 170 synthesis, transport and/or sensitivity. Accordingly, strigolactone depletion decreases sensitivity to 171 exogenous ABA in several species (Bu et al., 2014;Ha et al., 2014;J. Liu et al., 2015;Lv et al., 2018;172 Visentin et al., 2016) and sensitivity to endogenous ABA in stressed tomato (Visentin et al., 2016). 173 On the other hand, treatment with the synthetic strigolactone analogue racemic GR24 (rac-GR24) 174 increases sensitivity to ABA in tomato (Visentin et al., 2016). It deserves attention that the above-175 described model may be confined to dicot plants, as it does not apply to rice -most strigolactone-176 biosynthetic mutants of which produce more ABA than the wild-type (wt) and thus are more 177 resistant to drought (Haider et al., 2018). 178 In this varied landscape, the possibility of a functional connection between miR156 and 179 strigolactones in the regulation of stomatal movements and drought avoidance has never been 180 investigated. miR156g was in silico predicted to directly target the transcripts of the strigolactone 181 biosynthetic gene MAX1 in Arabidopsis (Marzec & Muszynska, 2015), but experimental proof of 182 this is currently lacking. Additionally, a functional link in the context of shoot development, via 183 stabilisation of specific SPL proteins by strigolactones, is already known in gramineous plants (J. 184 Liu, Cheng, Liu, & Sun, 2017;Song et al., 2017). The fact that the transcripts of several SPL factors 185 are targeted by miR156 offers a potential integration point between hormone-and miRNA-186 mediated signalling (Kerr & Beveridge, 2017;M. Liu et al., 2017;Song et al., 2017). Nevertheless, it 187 is still unknown whether this or a similar mechanism is operational under stress. 188 In this work, we used different approaches to clarify the link between miR156 and strigolactones 189 with regards to stomatal regulation in tomato. We investigated the effect of miR156 190 overexpression on stomatal function and ABA metabolism/sensitivity as well as the transcript 191 stability of strigolactone biosynthetic genes. Moreover, the effect of strigolactones on miR156 192 levels were assessed by i) treatment with exogenous strigolactones coupled to the use of a 193 strigolactone-depleted transgenic line, and ii) the application of drought stress to increase both 194 strigolactone and miR156 levels in shoots. The results identified strigolactones as a molecular 195 component linking drought to miR156 accumulation. Furthermore, they allowed us to integrate 196 miR156 in a model that describes the connections between strigolactones and ABA in tomato 197 , thus offering insights into both strigolactone and miR156 action on 198 stomata. 199

Plant materials and growth conditions 202
The tomato CCD7-silenced line 6936 (kind gift by Dr H. J. Klee, University of Florida) (Vogel et al., 203 2010) and its wt M82 were grown in a growth room set at the following conditions: 16/8 h 204 day/night cycle, 25°C, 65% humidity, and 200 µmol s -1 m -2 of photosynthetic photon flux density 205 (PPFD). Seeds were sterilized in 4% (v/v) sodium hypochlorite containing 0.02% (v/v) Tween 20, 206 rinsed thoroughly with sterile water, and plated on MS medium with 0.8% w:v agar, pH 5.9. Ten 207 days after, seedlings were transferred to an inert substrate and pots were watered with Hoagland 208 solution twice per week. Drought-stress kinetics were performed by uprooting plants of both wt 209 and CCD7-silenced genotypes and transferring them in vermiculite hydrated with only 20 ml of 210 water (stressed group), or transferred in wet vermiculite (well-watered controls). A sub-group of 211 wt plants were leaf-sprayed with GR24 5DS (see dedicated paragraph below) 24 h before the 212 beginning of stress, while all other plants were mock-treated with a solution of 0.01% v/v acetone 213 in water. Throughout the experiment, each condition and genotype was represented by 5 plants. 214 The tomato miR156-oe line overexpressing the AtMIR156b primary transcript and its wt 215 genotype. After 10 days, once severe water stress levels were reached, the plants were watered to 222 allow for recovery. Stomatal conductance was measured throughout the experiment, while leaf 223 samples were collected for water potential measurements at day zero, ten and fifteen (see below 224 for physiological analytical methods). 225

Treatment with exogenous strigolactones 227
A 5 μM solution of GR24 5DS (StrigoLab Srl, Turin, Italy) in 0.01% v/v acetone in water was sprayed 228 on leaves of unstressed plants until runoff, while control plants were sprayed with a 229 corresponding solution of acetone only. This pure enantiomeric form of the synthetic strigolactone 230 analogue GR24 was preferred over commercial rac-GR24 due to the possibly confounding 231 bioactivity by the other enantiomer (GR24 ent-5DS ) contained in the racemic mixture (Scaffidi et al., 232 2014). Stomatal conductance (see below) was measured 2 and 24 h after treatment, while 233 samples for the quantification of mature miR156 were collected 2, 6 and 24 h after treatment (in 234 the absence of stress), deep frozen and stored at -80°C until analysis. When GR24 5DS treatment 235 was imposed on wt plants to be subsequently stressed, harvesting times were i) 24 h after 236 treatment (well-watered samples; with stress beginning immediately after harvest); ii) after 3 h of 237 stress (water-stressed samples; with re-watering and the beginning of recovery happening 238 immediately after harvest); iii) 24 h after the beginning of stress, i.e. 21 h into recovery (recovered 239 samples). 240

Gene-transcript quantification and miR156 target-site detection 242
For transcript quantification from axillary buds, at least 30 stem sections were excised at the level 243 of leaf insertion from 10-week-old MicroTom plants (eight plants each, for wt and miR156-oe 244 plants) to obtain two lots of 50 mg (fw) per genotype, and freeze-dried until analysis. For 245 quantification in shoot tissues, the same procedure was applied on 200 mg (fw) of tomato leaves. 246 Total RNA was extracted by using Spectrum™ Plant Total RNA Kit (SIGMA), and treated with DNase 247 I (ThermoScientific) at 37°C for 30 min to remove residual genomic DNA. First-strand cDNA was 248 synthesized from 1 µg of purified total RNA using the High-Capacity cDNA Reverse Transcription 249 Kit (Applied Biosystems, Monza, Italy) according to the manufacturer's instructions. 250 For targeted miR156 cDNA synthesis, a modified protocol with a stem-loop primer (Pagliarani et 251 al., 2017) was followed in samples of wt M82 and CCD7-silenced plants obtained from the 252 drought-stress experiments described above. For transcript quantification of candidate genes by 253 quantitative reverse-transcription PCR (qRT-PCR), random primers were used to reverse transcribe 254 total RNA. qRT-PCR analysis was carried out in a StepOnePlus system (Applied Biosystems) using 255 the SYBR Green (Applied Biosystems) method on 10 ng of cDNA (50 ng for SlCCD7 transcripts). For 256 loci and primers (which were used at 400 nM), see Table S1. Transcript concentrations were 257 normalised on the geometric mean of SlsnRU6 and SlEF-1α transcript concentrations used as 258 endogenous controls. Three independent biological replicates were analysed as a minimum, and 259 each qRT-PCR reaction was run in technical triplicates. Transcript amounts were quantified by the 260 2 -ΔΔCt method. Putative target genes of miR156 were predicted in silico using the psRNATarget 261 algorithm with default setting parameters (http://plantgrn.noble.org/psRNATarget/) (Dai & Zhao, 262 2011). 263 264

Stomatal conductance and aperture measurements 265
During the drought-stress time-course on miR156-oe plants and their MicroTom wt, stomatal 266 conductance was measured daily between 10:00 and 12:00 am on two randomly selected, fully 267 developed apical leaves for each plant with a portable system (SC-1 Leaf Porometer for Stomatal 268 Conductance Measurements, Decagon Device, WA, USA). Leaf water potential was measured 269 using a Scholander-type pressure chamber (Soil Moisture Equipment Corp., Santa Barbara, CA, 270 USA) (Scholander, Bradstreet, Hemmingsen, & Hammel, 1965) on one leaf per plant, immediately 271 after gas exchange quantification. During the drought stress time-course on CCD7-silenced plants 272 and their M82 wt, stomatal conductance was measured on two randomly selected, fully 273 developed apical leaves for each plant with the same portable system as above. The selected time-274 points for full measurements were: i) immediately before stress start, between 8:00 and 10:00 am; 275 ii) 3 h later, when stomatal conductance values were about 20% of the irrigated controls (after 276 which, the plants were watered to start recovering); iii) 6 and 24 h after the beginning of stress, 277 i.e. 3 and 21 h into the recovery period. In order to quantify guard-cell reactivity to exogenous ABA 278 treatments, leaves of 4-week-old wt (MicroTom) and miR156-oe plants were sprayed with varying 279 concentrations of ABA in water or with water alone until drip-off, then let dry for 1 h before 280 quantifying stomatal conductance as above. For the quantification of stomatal pore areas in the 281 two genotypes, mature excised leaves were pre-incubated in a MES-KCl buffer (10 mM MES-282 KOH/50 mM KCl, pH 6.15) under light for 2 h to promote stomatal opening before treatments. ABA] were added to each sample. After 1 h shaking in the dark at 4°C, the homogenates were 298 centrifuged (20 000 g, 10 min, 4°C), and the pellets were then re-extracted in 0.5 ml extraction 299 solvent for 30 min. The combined extracts were purified by solid-phase extraction on Oasis® HLB 300 cartridges (60 mg, 3 ml, Waters, Milford, MA, USA), then evaporated to dryness in a Speed-Vac 301 (UniEquip) and finally analysed by UPLC-ESI(-/+)-MS/MS (Turečková, Novák, & Strnad, 2009 To test the hypothesis that miR156 induction under drought is relevant to the plant water balance, 312 the physiological responses of wt (MicroTom) and miR156-oe plants were analysed during a 313 drought and recovery time-course. Well-watered miR156-oe plants displayed a significantly lower 314 stomatal conductance than wt (day zero in Fig. 1A), while water potential in the same leaves did 315 not significantly differ between the two genotypes (well-watered samples in Fig. 1B). This pattern 316 of higher stomatal conductance in the wt was maintained throughout the experiment for the 317 irrigated controls. In stressed plants, miR156 overexpression correlated with lower stomatal 318 conductance in the initial phases of the dehydration kinetics (day two, Fig 1A). At the time of the 319 most severe drought stress level the situation reversed, as stomatal conductance of miR156-oe 320 plants was slightly but significantly higher than wt (day seven and ten, Fig. 1A). At this stage, leaf 321 water potential was significantly lower in wt than miR156-oe plants (water-stressed samples in Fig  322 1B, corresponding to day ten in Fig. 1A). At day fifteen, i.e. five days after re-irrigation, leaf water 323 potential of both wt and miR156-oe plants had recovered to levels comparable to those preceding 324 the drought spell (recovered samples in Fig. 1B). However, unlike wt plants, stomatal conductance 325 of miR156-oe tomato plants had completely failed to recover at this time (day fifteen, Fig. 1A). 326 These data suggest that: i) miR156 overexpression can lead to better acclimation to drought; ii) 327 the guard cells of miR156-oe plants respond more slowly to changes in water availability than 328 those of wt; and iii) under irrigated conditions, stomatal conductance is most likely controlled by a 329 non-hydraulic, miR156-related signal in tomato. 330

331
The low stomatal conductance of miR156-oe plants is not due to increased free ABA 332 Considering the pivotal role of ABA in stomatal closure, we reasoned that the physiological effects 333 of miR156 overexpression on stomatal opening observed in the above experiments may be due to 334 increased ABA levels. Hence, we set to quantify ABA and its main catabolites in the leaves of wt 335 and miR156-oe plants under well-watered conditions (day zero in Figure 1). Unexpectedly given 336 their low-transpiration phenotype, free ABA was significantly less concentrated in the leaves of 337 miR156-oe plants than in wt ( Fig. 2A). Accordingly, transcripts of the ABA biosynthetic gene 338 SlNCED1 were somewhat lower in miR156-oe plants, albeit not significantly (Fig. 2B). While 339 glycosylated ABA levels were similar in wt and miR156-oe plants, the ABA catabolite phaseic acid 340 showed a non-significant trend towards higher concentrations in miR156-oe plants than in the wt 341 ( Fig. 2A). Consistently, transcript amounts of the ABA-hydroxylating genes SlCYP707A1 and 342 SlCYP707A2, which catalyse a key step in phaseic acid production from ABA, were significantly 343 higher in miR156-oe than wt leaves (Fig. 2B). Other related metabolites, such as dihydro-phaseic 344 acid, neo-phaseic acid and 7'-hydroxy-ABA, were below the limit of detection in both genotypes 345 (n.d. in Fig. 2A). Thus, miR156 overexpression seems to decrease endogenous ABA levels, at least 346 partly by inducing degradation. These data therefore exclude that the low stomatal conductance 347 rates observed in miR156-oe plants under well-watered conditions are due to increased ABA 348 content in total leaf tissues. 349 350 miR156 promotes stomatal closure in response to ABA 351 Since leaves of the miR156-oe genotype contained less free ABA than those of the wt and yet they 352 had more closed stomata, we hypothesized that miR156 may increase guard cell sensitivity to 353 ABA. To further address this point, we quantified stomatal conductance in wt and miR156-oe average conductance values was already significant at 1 µM ABA concentration. Also at higher ABA 358 concentrations the decrease of stomatal conductance relative to the mock-treated controls was 359 more marked in miR156-oe plants than in the wt (Fig. 3). To confirm physiological data at the 360 morphological level, we as well assessed stomatal closure by determining stomatal pore areas 361 after 1 h exposure to 10 µM exogenous ABA. In the absence of ABA in the floating medium, 362 stomatal pore areas were significantly wider in wt than miR156-oe leaves (Fig. S1A), confirming 363 the stomatal conductance differences occurring between the two genotypes under unstressed and 364 untreated conditions ( Fig. 1 and 3). Upon ABA treatment, average stomatal pore areas in miR156-365 oe plants decreased by about 60% with respect to non-treated stomata in the same genotype. By 366 contrast, mean stomatal aperture in wt plants decreased by about 30% only, compared to the 367 untreated control (Fig. S1B and C). These results indicate that guard cells have higher sensitivity to 368 exogenous ABA in miR156-oe than wt plants. This in turn supports the notion that the lower 369 stomatal conductance in well-watered miR156-oe vs wt plants might be due, at least partly, to 370 higher sensitivity to endogenous ABA, which would more than compensate for the lower ABA 371

content. 372
Strigolactones are needed and sufficient for miR156 induction in tomato, both under normal 374 conditions and under drought 375 As outlined in the introduction, both the miR156 and strigolactone biosynthetic pathways are 376 induced by drought in leaves; both are needed for full resistance to abiotic stress; and both 377 enhance guard cell sensitivity to ABA in tomato. This was shown above for miR156 and in a 378 previous work for strigolactones (Visentin et al., 2016). Thus, we set to explore if not only a 379 correlation but also a causal link exists between strigolactones and miR156 that may be relevant 380 to stomatal functioning. 381 As a first approach, we compared the levels of mature miR156 in leaf tissues 2, 6 and 24 h after 382 treatment with the strigolactone analogue GR24 5DS with mock-treated controls. Results from this 383 experiment indicate that exogenous strigolactones are sufficient to increase the concentration of 384 the mature miRNA starting a few hours after treatment and up to 24 h later (Fig. 4A). 385 We next tested whether endogenous strigolactones are required for miR156 induction under 386 stress, by quantifying mature miR156 in leaves of wt (M82) and CCD7-silenced tomato plants 387 under irrigated and stress conditions. Severe stress levels are achieved quickly (within a few hours) 388 in this experimental set-up, by transferring the plants to dry or wet vermiculite substrate for the 389 stressed and control group, respectively. The two genotypes showed a very similar, low 390 concentration of miR156 under irrigated conditions, while wt leaves treated 24 h earlier with 391 GR24 5DS had higher concentrations than mock-treated leaves (Fig. 4B), consistently with the 392 results in Fig. 4A. At the most severe stress point, the concentration of mature miR156 increased 393 in wt leaves, as expected. Conversely, in plants impaired in strigolactone biosynthesis, no 394 induction could be observed (water-stressed samples in Fig. 4B). Additionally, the results in Fig. 4B  395 clearly show a powerful synergic effect of GR24 5DS pre-treatment and drought on mature miR156 396 levels. This is obvious both in stressed samples, and especially after re-watering (for 24 h in the 397 recovered samples in Fig. 4B), when miR156 amounts decrease towards pre-stress levels in the 398 leaves of mock-treated wt plants, , while they steadily increase up to 450 folds in the GR24 5DS -pre-399 treated leaves. 400

Strigolactones promote sustained stomatal closure ("after effect") during recovery from drought 402
Based on the above data, miR156 appears to promote the after-effect of drought, and its 403 inducibility by stress seems to completely depend on strigolactones in tomato. Therefore, we 404 reasoned that if the strigolactone-miR156 module is operational in guard cells, then strigolactones 405 should positively affect the extent of the after-effect of drought as well. To elucidate this point, 406 stomatal conductance data (Fig. 4C) were collected during the same experiment reported in Fig.  407 4B. Data showed clearly that GR24 5DS -treated plants displayed an enhanced after-effect of 408 drought, thus mimicking the physiological response of miR156-oe plants. Consistently, CCD7-409 silenced plants did the opposite and recovered their stomatal conductance faster than wt plants. 410 Overall, these data attested that strigolactones promote sustained stomatal closure during 411 recovery from drought (the so-called "after-effect") in tomato, as miR156 does. 412 Finally, we tested whether exogenous strigolactones could directly induce stomatal closure. In 413 Arabidopsis, treatment with rac-GR24 induces a fast stomatal closure that has been demonstrated 414 to be ABA-independent (Lv et al., 2018). Thus, we measured stomatal conductance in unstressed, 415 wt tomato leaves treated with GR24 5DS compared with the mock-treated controls. Our results 416 clearly showed that stomatal conductance levels of wt tomato decreased significantly within 2 h of 417 GR24 5DS treatment, while they they started recovering at 24 h (Fig. 4D). 418 419

Strigolactones affect the transcript accumulation profiles of putative SPL factors 420
The above results, together with previous observations on the activation of the strigolactone 421 biosynthetic pathway in leaves under drought (Visentin et al., 2016), suggested that a 422 strigolactone-miR156 module does exist in tomato. This module may be fully operational under 423 and after drought, thanks to an initial stress-induced increase of leaf strigolactones. If so, then we 424 should be able to identify transcripts of putative miR156 targets that are dysregulated, in a stress- in wt and CCD7-silenced plants diverged from those of miR156. This finding is consistent with the 434 hypothesis that after drought -i.e. when the stomatal after-effect becomes apparent -these 435 transcripts are destabilised by miR156, which in turn is induced by strigolactones (Fig. 5). On the 436 contrary, the transcript profiles of COLORLESS NON-RIPENING (CNR, Solyc02g077920) and of the 437 orthologue of SPL9/15 in Arabidopsis (Solyc10g078700) (Silva et al., 2019) showed no significant 438 changes during drought and they both increased during recovery in the two genotypes, mirroring 439 the related miRNA expression trend (Fig. 5) The transcripts of the strigolactone-biosynthetic gene MAX1 have been predicted in silico to be 448 direct targets of miR156 in Arabidopsis (Marzec & Muszynska, 2015). This would in principle lead 449 to transcript degradation and thus lower strigolactone production in miR156-oe plants. Indeed, in 450 axillary buds of miR156-oe potatoes, the strigolactone content was lower than in wt (Bhogale et 451 al., 2014), but the molecular underpinnings were not investigated at that time. To search for all 452 possible links between miR156 and strigolactones, and given that strigolactones are known to 453 feedback regulate their own synthesis at the transcriptional level, we also checked whether 454 miR156 might affect the stability of strigolactone biosynthetic genes in tomato. We addressed this 455 issue in silico first, by searching for miR156 target sequences on transcripts of D27, CCD7, CCD8 456 and MAX1 in Arabidopsis and tomato. However, no acceptable predictions satisfying the 457 requirements for identification of miRNA targets in plants were obtained (Axtell & Meyers, 2018;458 Dai & Zhao, 2011;Jones-Rhoades & Bartel, 2004). This contrasts with available information for 459 MAX1 and miR156g in Arabidopsis (Marzec & Muszynska, 2015). 460 Additionally, we quantified transcripts of strigolactone-biosynthetic genes in roots, where they are 461 mostly expressed, and axillary buds of miR156-oe and wt plants. However, no evidence of miR156-462 driven transcript destabilisation was found, and even higher transcript concentrations in the 463 miR156-oe plants was observed for some genes in either tissue ( Fig. 6A and B). Therefore, 464 negative regulation by miR156 on strigolactone biosynthetic genes at the transcript stability level 465 was not observed in tomato. Of course, indirect effects at the protein or metabolite level cannot 466 be excluded at this stage. Nonetheless, the latter information would not change the answer to the 467 question being asked here, i.e. if miR156 directly destabilizes the transcripts of strigolactone-468 biosynthetic genes, as suggested by Marzec and Muszynska (2015). 469 470 DISCUSSION 471

miR156 induction by drought requires strigolactones in tomato 472
Almost nothing is known about the molecular cues modulating miR156 levels under any 473 conditions. Even though specific changes in the epigenetic landscape at the MIR156a and c loci are 474 reported to be important for correct miR156 expression during Arabidopsis development (Xu, 475 Zhang, & Wu, 2018), no information are available about the signalling path connecting it to stress. 476 Partly filling this gap, we found that miR156 induction by drought was completely dependent on 477 the efficient synthesis of endogenous strigolactones. We also observed that exogenous 478 strigolactones increased mature miR156 levels in tomato leaves, both in the absence and 479 especially in combination with drought stress. These findings make strigolactones the first 480 identified molecular component in the drought-triggered pathway leading to miR156 induction, 481 and to its stress-related effects (Cui et al., 2014;Stief et al., 2014). 482 483

The role of the strigolactone-miR156 module in stomata 484
Given the demonstrated effects of strigolactones on stomata in several plant species (Bu et al., 485 2014;Ha et al., 2014;J. Liu et al., 2015;Lv et al., 2018;Visentin et al., 2016), and the above 486 described functional connection between strigolactones and miR156, we sought to investigate 487 whether miR156 may act at the stomatal level specifically. When we compared the physiological 488 performances of miR156-oe with those of wt plants, stomatal conductance of the former was 489 indeed lower under irrigated conditions, suggesting a positive control of miR156 on stomatal 490 closure. This trait of miR156-oe plants is opposite to strigolactone-related mutants, and 491 reminiscent of tomato plants whose shoots experience high strigolactones. This latter situation 492 can be achieved either by treating with exogenous strigolactones (which induce miR156, as stress 493 does), or grafting a wt scion onto a strigolactone-depleted rootstock (which induces higher 494 strigolactone biosynthesis in the shoot, higher ABA sensitivity in guard cells and lower stomatal 495 conductance) (Ha et al., 2014;J. Liu et al., 2015;Visentin et al., 2016). It is noteworthy that in 496 miRNA156-oe plants subjected to severe stress, stomatal conductivity was higher than in the wt 497 (as observed in miR156-oe M. sativa by Arshad et al., 2017), reversing the pattern observed upon 498 irrigation. This could be explained by non-ABA-dependent signal(s) acting downstream of hydraulic 499 signals generated by severe drought. Consequently, miR156-oe plants lost less water under severe 500 drought. This was attested by their water potential dropping less than wt, and by the fact that 501 their better water status overweighed higher ABA sensitivity in terms of stomatal closure. 502 The effects of the strigolactone-miR156 module under recovery might be executed via the post-503 transcriptional repression of SPL genes such as Solyc10g009080, which most convincingly showed 504 a divergent transcript profile with miR156 in this phase. Their ultimate function under stress is 505 worthy of further experimental investigation. It should be noted here that, in drought stress 506 conditions, the transcripts of all four SPL genes suggested to be miR156 targets decreased in 507 CCD7-silenced leaves as much as in the wt. As in these transgenic tissues miR156 levels did not rise 508 as a consequence of drought, we must conclude that their regulation during stress is more likely to 509 occur at the transcriptional rather than post-transcriptional level, contrarily to previous 510 suggestions (M. . 511 512 miR156 is a possible mediator of ABA-dependent effects of strigolactones on stomata 513 Strigolactone effects on stomata are known to be both ABA-dependent and independent. Indeed, 514 treatment with exogenous strigolactones induces fast, ABA-independent closure of stomata but 515 also higher sensitivity to ABA (Brun et al., 2019;Li et al., 2017;J. Liu et al., 2015;Lv et al., 2018;516 Visentin et al., 2016;, while strigolactone-related mutants have impaired 517 ABA-dependent responses to osmotic stress (reviewed in Cardinale et al., 2018). 518 We found several indications rather pointing to miR156 affecting the ABA-dependent subset of 519 strigolactone effects on stomata, under irrigated conditions. Firstly, stomatal closure in response 520 to exogenous strigolactone treatment (ABA-independent in Arabidopsis) (Lv et al., 2018) was 521 achieved within 2 h in tomato, when miR156 induction was not significant yet; and started 522 decreasing by 24 h, when miR156 levels were highest (Fig. 4A vs Fig. 4D). Secondly, under irrigated 523 conditions, miR156-oe plants had lower stomatal conductance but also significantly less free ABA 524 in leaves, possibly due to weaker biosynthesis and especially accelerated ABA conversion to 525 phaseic acid. Accordingly, the activation of ABA catabolism is a reported effect of increased 526 strigolactone synthesis in the shoot, and of exogenous strigolactone treatment in different tissues 527 (Ferrero et al., 2018;Lechat et al., 2012;Toh et al., 2012;Visentin et al., 2016). Thirdly, stomatal 528 closure by exogenous ABA was more complete in miR156-oe plants than in wt, which may be 529 explained by higher sensitivity to, or more effective transport of ABA. Higher ABA sensitivity would 530 more than compensate lower ABA levels and justify reduced stomatal conductance in miR156-oe 531 plants. As in other priming phenomena, the strigolactone-dependent increase in ABA sensitivity 532 may be exerted at or downstream the perception and signalling level. Consistently, several ABA-533 responsive genes are less induced by drought in the max2 signalling mutant of Arabidopsis than in 534 the wt (Ha et al., 2014). Finally, but not less importantly, stomatal conductance of miR156-oe 535 plants recovered much less than wt following drought stress. This response was opposite to CCD7-536 silenced plants, while resembling that of plants treated with exogenous strigolactones prior to 537 stress imposition. This set of features converges on the idea that the strigolactone-miR156 538 pathway promotes the "after-effect" of drought, which is thought to be part of a wider stress-539 memory mechanism (Lämke & Bäurle, 2017). After rewatering in fact, stomatal conductance never 540 quite reaches the levels of unstressed plants even though water potential has, and this response 541 could depend on non-hydraulic signals such as ABA (Ding et al., 2012;Galmes, Medrano, & Flexas, 542 2007;Lovisolo, Perrone, Hartung, & Schubert, 2008). Thus, genetic and pharmacological evidence 543 indicates that both strigolactones and miR156 increase sensitivity to ABA in a number of species, 544 including tomato. This reinforces the possibility of a strigolactone-miR156 module setting basal 545 stomata sensitivity to ABA in accordance with the water-balance history of the plant. Of course, 546 our data do not exclude that part of miR156 effects may also be exerted ABA-independently. 547 548 In conclusion, the results of the present work provided insight into the induction of miR156 under 549 drought. They also demonstrated a cause-effect link between miR156 accumulation and the 550 regulation of water relations and stomatal functioning. Particularly, we suggest that miR156 might 551 not directly mediate the short-term, transient and likely ABA-independent stomatal closure 552 triggered upon GR24 5DS treatment. Rather, the strigolactone-miR156 module seems to set basal 553 sensitivity of guard cells to ABA. This is most obvious after a drought period has occurred, when 554 the module has been activated but hydraulic signals are no longer active. We propose that the 555 module is also relevant for the establishment of specific drought responses, since its activation 556 might promote acclimation to (recurring) water-limiting conditions by increasing sensitivity to ABA 557 and drought avoidance. Tomato plants conditionally depleted of miR156, or overexpressing it, will 558 allow discriminating between the stress-related, strigolactone-dependent effects of miR156 on 559 stomata, and the long-term developmental consequences of its constitutive dysregulation.   Fig. 1). The concentrations of (A) free ABA, abscisic-β-D-glucosyl ester (ABA-GE), phaseic acid (PA), dihydro-phaseic acid (DPA), neo-phaseic acid (neo-PA), 7'hydroxy-ABA (7'OH-ABA) and of (B) transcripts of the ABA catabolic genes SlCYP707A1 and SlCYP707A2 and of the ABA biosynthetic gene SlNCED1 were quantified. Gene transcript abundance was normalised to endogenous SlEF-1α and SlsnRU6, and presented as fold-change value over mean values of wt plants, which were set to 1. Data represent the mean ± SE of n = 5 biological replicates. * indicates significant differences between genotypes, as determined by a Student's t-test (P <0.05). n.d. = not detectable. Data are the means ± SE of n = 5 biological replicates. Different letters indicate significant differences between different treatments in the same genotype as determined by a one-way ANOVA test, while * indicates significant differences between genotypes for the same condition, as attested by Student's t-test (P <0.05). Effect of GR24 5DS (5 µM) compared to mock treatment on the concentration of mature miR156 in leaf tissues during a short-term time-course in unstressed wt plants (0, 2, 6 and 24 h after treatment, n = 3 each sample being a pool of 3 leaflets). (B) Mature miR156 levels in leaves of wt (treated with exogenous strigolactones or mock-treated), and mock-treated CCD7-silenced plants during a quick drought time-course. Stress was imposed by uprooting plants at time zero (wellwatered samples) and transferring them into dry vermiculite, with irrigated controls being transferred into wet vermiculite. Water-stressed and recovered samples were harvested 3 or 24 h (respectively) after the beginning of stress, having been re-watered right after stress peaking. For strigolactone and mock treatment, leaves were sprayed as in (A) 24 h before time zero (the beginning of stress). Data represent the mean ± SE of n = 3 biological replicates from three independent experiments. Target RNA abundance was normalised to endogenous SlEF-1α and SlsnRU6 transcripts and presented as fold-change value over mock-treated wt tissues, which were set to 1. Different letters indicate significant differences as determined by a one-way ANOVA test (P <0.05). (C) Stomatal conductance values for the experiment in (B). One-way ANOVA test (P <0.05) was applied to detect differences among genotypes within a given time point. (D) Normalised stomatal conductance of wt (M82) plants upon treatment with GR24 5DS . The aerial parts of plants were sprayed with a 5 µM GR24 5DS solution, and stomatal conductance (g s , mmol H 2 O m -2 s -1 ) was measured 2 and 24 h after the treatment. Data are presented as percentage of stomatal conductance over average g s values of mock-treated plants, which were set to 100%.
Data represent the mean ± SE of n = 4 biological replicates from three independent experiments.
Different letters indicate statistically significant differences among treatments as determined by a one-way ANOVA test (P <0.05). harvested 3 or 24 h (respectively) after the beginning of stress, having been re-watered right after stress peaking. Data are presented as fold change over the mean values for well-watered plants (time zero) of the same genotype, which were set to 1. Data represent the means ± SE of n = 5 (Solyc07g062980, Solyc10g009080 and Solyc10g078700) or n = 3 (Solyc03g114850, Solyc04g045560, Solyc01g090660 and Soly02g077920) biological replicates. Different letters indicate significant differences between conditions and genotypes within a given bar cluster as determined by a one-way ANOVA test (P <0.05). Figure 6. Effects of miR156 overexpression on transcripts of strigolactone-biosynthetic genes.
Transcript amounts of the putative (SlD27) or confirmed (SlMAX1, SlCCD7 and SlCCD8) strigolactone-biosynthetic genes were quantified in (A) roots or (B) axillary buds of wt (MicroTom) and miR156-oe plants. Gene transcripts were normalised to endogenous SlEF-1α and SlsnRU6 and presented as fold change over mean wt values, which were set to 1. In (A) data represent the means ± SE of n = 4 biological replicates from two independent experiments; in (B), each of two replicates was the pool of at least 30 axillary buds from 8 plants. SlCCD7 transcript was not detectable in samples represented in (B). * indicates significant differences between genotypes as determined by a Student's t-test (P <0.05).

SUPPLEMENTARY DATA
The following supplementary data are available: Figure S1. Effects of miR156 overexpression on stomatal reactivity to exogenous ABA expressed as pore areas. Table S1. List of primer pairs used in this work, with target gene names and/or Solyc codes.       S1. Effects of miR156 overexpress ion on stomatal rea ctivity to exogenous A BA. (A) Representative images of stomata from wt (MicroTom, left-hand panels) and miR156-oe plants (right-hand pa nels), before (upper pane ls) or after treatment (lower panels) with ABA 10 µM for 1 h. (B) Raw and (C) norma lised stomatal aperture areas measured on aba xial epiderma l peels of the two genotypes, before and after ABA treatment. Data represent the mean a nd ± SE of n > 30 biologica l replicates from two indepe ndent expe rime nts. * indicates significant differences betwee n genotypes under the same conditions, while differe nt letters mark significant diffe rences be tween treated and untreated leaves of the same genotype as determined by a Student's t-test (P <0.05).