Ectopic Expression of Distinct PLC Genes Identifies ‘Compactness’ as a Possible Architectural Shoot Strategy to Cope with Drought Stress

Abstract

Phospholipase C (PLC) has been implicated in several stress responses, including drought. Overexpression (OE) of PLC has been shown to improve drought tolerance in various plant species. Arabidopsis contains nine PLC genes, which are subdivided into four clades. Earlier, OE of PLC3, PLC5 or PLC7 was found to increase Arabidopsis’ drought tolerance. Here, we confirm this for three other PLCs: PLC2, the only constitutively expressed AtPLC; PLC4, reported to have reduced salt tolerance and PLC9, of which the encoded enzyme was presumed to be catalytically inactive. To compare each PLC and to discover any other potential phenotype, two independent OE lines of six AtPLC genes, representing all four clades, were simultaneously monitored with the GROWSCREEN-FLUORO phenotyping platform, under both control- and mild-drought conditions. To investigate which tissues were most relevant to achieving drought survival, we additionally expressed AtPLC5 using 13 different cell- or tissue-specific promoters. While no significant differences in plant size, biomass or photosynthesis were found between PLC lines and wild-type (WT) plants, all PLC-OE lines, as well as those tissue-specific lines that promoted drought survival, exhibited a stronger decrease in ‘convex hull perimeter’ (= increase in ‘compactness’) under water deprivation compared to WT. Increased compactness has not been associated with drought or decreased water loss before although a hyponastic decrease in compactness in response to increased temperatures has been associated with water loss. We propose that the increased compactness could lead to decreased water loss and potentially provide a new breeding trait to select for drought tolerance.

Introduction

Drought is one of the most impactful stresses in agriculture, affecting 75% of all harvested areas (Boretti and Rosa 2019). Climate change will only make this worse, with increasing periods of high temperature and no rain, as well as a decline in freshwater resources (Kim et al. 2019). Understanding how plants sense and respond to drought is therefore of great importance so that crops can be bred to become more tolerant.

To deal with water deprivation, plants use various strategies (Martignago et al. 2020). The escape strategy induces early flowering so that it can reproduce before dying. Some plants employ the more direct drought-avoidance strategy not only by closing their stomata and by increasing the production of compatible solutes to maintain their water potential but also by slowing down their life cycle. Lastly, with the drought tolerance strategy, plants are able to improve survival by stimulating DNA repair, reactive oxygen species (ROS) scavenging and expression of late embryogenesis-abundant (LEA) proteins and dehydrins that have a chaperone function (Singh et al. 2015). The ultimate reaction of plants to water stress is a combination of these strategies, and this may vary between species and ecotypes (Meyre et al. 2001, Bouchabke et al. 2008, Bouzid et al. 2019) and can change in combination with other abiotic or biotic stresses (Lamaoui et al. 2018, Monohon et al. 2021, Rivero et al. 2021, Sinha et al. 2021).

The initial response to drought may significantly vary, and the reason for this is largely unknown. An interesting candidate in the early sensory mechanism is represented by the lipid signaling pathway headed by phospholipase C (PLC) (Munnik and Vermeer 2010, Munnik 2014, Hong et al. 2016; Sagar & Singh, 2021, Ali et al. 2022, Fang et al. 2023). Overexpression (OE) of PLC has been shown to improve drought tolerance in various plant species, including monocots and dicots, i.e. maize, rice, tobacco, canola and Arabidopsis (Wang et al. 2008, Georges et al. 2009, Tripathy et al. 2012, Zhang et al. 2014, 2018a, 2018b, van Wijk et al. 2018, Deng et al. 2019, Chen et al. 2021).

In animal systems, PLC is a well-established signaling enzyme that becomes activated in response to receptor stimulation by various signals (Katan and Cockcroft 2020). Upon activation, PLC hydrolyzes the minor plasma membrane (PM) phospholipid, phosphatidylinositol 4,5-bisphosphate (PIP₂), into two second messengers: inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol (DAG). While the water-soluble IP₃ diffuses into the cytosol where it triggers the release of Ca²⁺ from the ER by specific binding of a IP₃ gated-Ca²⁺ channel, DAG remains at the PM where it recruits and activates members of the protein kinase C (PKC) family. The subsequent increase in cytosolic Ca²⁺ concentration and change in phosphorylation status of various target proteins trigger a host of downstream responses, affecting numerous cellular processes (Katan and Cockcroft 2020).

The PLC-signaling system has been found to be strongly conserved in flowering plants although the main downstream targets, i.e. the IP₃ channel and PKC, are typically absent (Munnik and Testerink 2009, Munnik 2014). Instead, flowering plants seem to use the phosphorylated products of the PLC hydrolysis as second messengers, i.e. inositol polyphosphates (IPPs, including IP₅, IP₆, IP₇ and IP₈) and phosphatidic acid (PA), which are generated by IPP kinases (IPKs) and DAG kinases (DGKs), respectively (Munnik et al. 2000, Munnik and Vermeer 2010, Gillaspy 2013). Plant targets for IP₅ or IP₆ include ligand-gated Ca²⁺channels, proteins involving mRNA transport (e.g. GLE1) and F-box proteins involved in auxin and jasmonate perception (e.g. TIR1 and COI1) (Munnik and Testerink 2009, Heilmann 2016a, Noack & Jaillais 2017). Similarly, IP₇ and IP₈ have been implicated in phosphate signaling through binding SPX domains (Williams et al. 2015, Laha et al. 2016, Ried et al. 2021). Targets for PA include protein kinases, phosphatases, small G-proteins, transcription factors and membrane transporters, involving processes such as endocytosis, membrane trafficking, organization of the cytoskeleton, osmoregulation and gene expression (Munnik 2001, Wang et al. 2006, Li et al. 2009, Testerink and Munnik 2011, McLoughlin and Testerink 2013, Pleskot et al. 2013, Pokotylo et al. 2014, Thomas and Staiger 2014, Hou et al. 2016, Ufer et al. 2017, Pokotylo et al. 2018, Yao and Xue 2018, Wang et al. 2019, Kolesnikov et al. 2022).

Animal PLCs predominantly hydrolyze PIP₂ but can also hydrolyze its immediate precursor, phosphatidylinositol 4-monophosphate (PIP; Levin et al. 2017). The quantities of these minor lipids are approximately equal to 1:1. In flowering plants, however, PIP₂ levels are ∼30- to 100-fold lower, while PIP concentrations are similar to those in animals (Munnik et al. 1998, Meijer and Munnik 2003, Heilmann 2016b, Noack & Jaillais 2017). So, for plants, it is more likely that PLC hydrolyzes PIP in response to stimulation (Munnik 2014). There are, however, conditions in which plants locally and temporarily increase their PIP₂ levels, i.e. upon heat or osmotic stress, during cell division, at the tip of growing root hairs and pollen tubes and in response to polyamines, which likely reflects PIP₂’s role as lipid second messenger itself (van Leeuwen et al. 2007, Ischebeck et al. 2010, Simon et al. 2014, Bloch et al. 2016, Zarza et al. 2020). During these circumstances, PLC may have an additional function, i.e. to attenuate PIP₂ signaling besides initiating IPP and PA signaling.

How OE of PLC leads to increased drought tolerance is unknown. The fact that it works in both monocots and dicots indicates that the mechanism is conserved (Wang et al. 2008, Georges et al. 2009, Tripathy et al. 2012, Zhang et al. 2014, 2018a, 2018b, van Wijk et al. 2018, Deng et al. 2019, Chen et al. 2021. van Hooren et al. 2023). While metazoan PLCs can be grouped into six subfamilies, which are characterized by conserved domains that explain their regulation by e.g. G-protein-coupled receptors or receptor tyrosine kinases, plant PLCs lack most of such domains and only contain the minimum present in all eukaryotic PLCs, i.e. a catalytic X- and Y-domain that fold over the lipid headgroup, an EF-hand and C2 domain; the latter explaining the Ca²⁺ sensitivity of the enzyme (Munnik 2014). Plant PLCs mostly resemble PLCζ, which is the only isoform that lacks the pleckstrin homology (PH) domain responsible for binding PIP₂. Metazoan PLCζ is specifically expressed in sperm cells, and it is still unknown how this isoform is regulated (Thanassoulas et al. 2022), as is the case for plant PLCs (Munnik 2014, Hong et al. 2016, Sagar & Singh, 2021, Ali et al. 2022, Fang et al. 2023).

The Arabidopsis genome encodes nine PLC genes, which can be subdivided into four clades (Hunt et al. 2004, Tasma et al. 2008). Earlier, OE of PLC3, PLC4, PLC5 and PLC7 has each been found to increase the tolerance of Arabidopsis plants to water stress (van Wijk et al. 2018, Zhang et al. 2018a, 2018b, van Hooren et al. 2023). PLC2 is the only constitutively expressed PLC and is also the highest in expression of all AtPLCs (Tasma et al. 2008). The fourth PLC clade is represented by PLC8 and PLC9, of which the encoded proteins were predicted to be catalytically inactive due to certain mutations in the catalytic X-domain (Hunt et al. 2004) although experimental evidence for this is lacking. Interestingly, PLC9-OE were reported to contain higher IP₃ levels under heat stress, while plc9 knock-out (KO) mutants exhibited reduced levels in response to heat (Zheng et al. 2012). Moreover, AtPLC9 OE increased the thermotolerance of both Arabidopsis and rice (Zheng et al. 2012, Liu et al. 2020), while plc9 mutants were more sensitive to heat stress. These experimental results imply that AtPLC9 does have activity.

Promoter–GUS fusions in Arabidopsis showed that PLC expression is most prominent in vascular tissue in the shoot (PLC1, PLC2, PLC3, PLC5 and PLC7) and either in the vasculature of developing root and root primordia (PLC3 and PLC5) or throughout the whole root (PLC1, PLC2 and PLC7). Expression has also been found in the root tip (PLC2 and PLC7), stomata (PLC2 and PLC5) and trichomes (PLC2, PLC3, PLC5 and PLC7) (Hunt et al. 2004, Kanehara et al. 2015, van Wijk et al. 2018, Zhang et al. 2018a, 2018b). Fluorescent protein (FP) fusions to Arabidopsis PLC2, PLC3, PLC5 and PLC9 have been found to localize to the PM under their native promoter (Zheng et al. 2012, Ren et al. 2017, van Hooren et al. 2023). For PLC2 and PLC5 also, constructs with the UBQ10 promoters were made, but no viable lines could be obtained with this, indicating that high amounts of these fusion proteins are lethal (van Hooren et al. 2023).

PLC-OE is typically achieved with POLYUBIQUITIN 10 (UBQ10) or 35S promoters that ectopically and constitutively express genes to high amounts (Grefen et al. 2010). Experiments performed with such lines have mainly focused on survival under severe drought stress, comparing the tolerance of OE lines of a single PLC gene with that of wild type (WT) (Wang et al. 2008, Georges et al. 2009, Tripathy et al. 2012, Deng et al. 2019, van Wijk et al. 2018, Zhang et al. 2018a, 2018b). In this study, the GROWSCREEN phenotyping platform (Jansen et al. 2009) was used to quantitatively compare the behavior of Arabidopsis PLC-OE lines from all four PLC clades under mild water stress conditions, i.e. PLC2, PLC3, PLC4, PLC5, PLC7 and PLC9, in comparison to WT plants. To investigate which tissues were most relevant for the PLC-induced drought tolerance, we constructed 13 different ‘Tissue-Specific Expression of PLC5’ (TSEP) lines and took these along in the phenotyping screen. The GROWSCREEN-FLUORO setup allowed for daily measurements of plant growth, architecture, color and chlorophyll fluorescence, measuring responses to water deprivation in a non-lethal fashion (Osmolovskaya et al. 2018). Both early and late responses to water deprivation were monitored, as well as the recovery after the post-stress irrigation of the plants. Instead of just estimating the final outcomes of the imposed water deprivation, an overview of the progress of the stress may provide new information on the plant’s response, following the dynamics of the studied parameters.

Results

PLC2-OE promotes drought stress tolerance

Of all Arabidopsis PLC genes, PLC2 is expressed the highest and is the only PLC that is constitutively expressed and not induced by stress (Hunt et al. 2004, Tasma et al. 2008). To test whether the OE of this PLC would also promote drought tolerance, distinct PLC2-OE lines were generated using pUBQ10 and p2x35S promoters. Enhanced expression was validated using quantitative polymerase chain reaction (qPCR) (∼5- to 17-fold) (Fig. 1A) and Western blot analysis of the two lines with the highest gene expression (Fig. 1B). A clear increase in a ∼66 kD protein, the expected size of PLC2, was observed as wells as two minor bands at ∼37 and ∼29 kD, which add up as 66 kD, likely representing PLC2-breakdown products. The same PLC-OE lines revealed an increase in drought tolerance after 18 days of withholding water (Fig. 1C).

Fig. 1.

Survival rates of PLC2-OE lines under water stress. (A) PLC2 expression in WT and OE lines as measured by qPCR. Values are normalized to SAND and WT and represent the means with SD of three to four biological replicates (ANOVA test, *P < 0.05; **P < 0.01; ***P < 0.001). (B) PLC2 detection using Western blot with anti-AtPLC2 1/2000 (D’Ambrosio et al. 2017) and GARPO 1/5000 on 11-d-old roots of WT, KO and OE lines of PLC2. (1) AtPLC2 ∼66 kD, (2) breakdown product AtPLC2 ∼37 kD, (3) breakdown product AtPLC2 ∼29 kD. (C) Survey of plant survival of WT and PLC2-OE lines after water stress in a single experiment. The 37-d-old plants were water deprived for 18 d. To make clear which plants survived this period, plants were rewatered and photographs were taken after 9 d.

OE of presumed inactive PLC9 increases drought tolerance too

In the literature, PLC8 and PLC9 are suggested to be catalytically inactive (Hunt et al. 2004, Tasma et al. 2008) although experimental evidence is lacking for this. To investigate this in more detail, the catalytic domains of PLC8 and PLC9 were compared in silico with those of the other Arabidopsis PLCs as well as with rat PLCδ1 (Rattus norvegicus), of which the crystal structure has been resolved (Essen et al. 1997) and several essential amino acids for activity have been identified (Ellis et al. 1998). Except for PLC8 and PLC9, all Arabidopsis PLCs contained the same key amino acids in the catalytic site as rat PLCδ1 (Supplementary Table S3). Mutations in the catalytic side for PLC8 and PLC9 were H311L, H356P, E390K, K440R and Y155R, of which the two histidine residues are located at prime positions within the catalytic center, as revealed by modeling of the catalytic sites of PLC5 and PLC9, and using the protein structure of rat PLCδ1 as a template (Fig. 2A). Mutations of these two histidines in rat PLCδ1 resulted in a dramatic reduction of PIP₂ binding, i.e. by 6,000-fold for His³¹¹ and 20,000-fold for His³⁵⁶ (Ellis et al. 1998). While these results may indicate that PLC8 and PLC9 are theoretically inactive, PLC8 and PLC9 also gained two new cationic amino acids (i.e. E390K and Y155R) that may take over. These new amino acids may explain the likely preference of PIP over PIP₂ as a substrate of plant PLCs (Munnik 2014).

Fig. 2.

Structure modeling of catalytic site of PLC9, and survival rates of PLC9-OE lines under water stress. (A) Protein structures of Arabidopsis PLC9 compared to AtPLC5 and Rat PLCδ (1DJX). Encircled are the His residues missing from PLC9. (B) PLC9 expression in WT, KO and OE lines as measured by qPCR. Values are normalized to SAND and represent the means with SD of three to four biological replicates (ANOVA test, **P < 0.01; ****P < 0.0001). (C) Survey of plant survival of WT and PLC9-OE lines after water stress in a single experiment. Three-week-old plants were water deprived for 3 weeks, after which they were watered again to show which plants survived the drought period. (D) Photographs of the plants used in Fig. 1C after 3 d of rewatering (pots with surviving plants were grouped together per line).

A number of independent PLC9-OE lines were generated, and their PLC9 expression was monitored in homozygous T3 plants using qPCR, along with two plc9-KO lines to validate the specificity of the primers (PLC8 is a very close homolog). As shown in Fig. 2B, all PLC9-OE lines revealed an increase in PLC9 expression (∼3- to 20-fold), and these lines also showed increased survival rates after 3 weeks of drought stress (Fig. 2C, D).

Shoot architecture of PLC-OE under water deprivation

For the subsequent large GROWSCREEN phenotyping experiment, PLC-OE lines with the highest OE were selected (i.e. pUBQ10 lines #4.1 and #8.1 for PLC2 and #18.4 and #23.6 for PLC9), together with two independent OE lines of PLC3, PLC4, PLC5 and PLC7.

Plant rosettes were monitored during control and water deprivation conditions between 25 and 38 d after sowing (DAS) and during recovery, until harvesting at 45 DAS. Since all PLC-OE lines exhibited increased survival during water deprivation, we were specifically interested in traits that would distinguish PLC-OEs from WT plants. Using a principal component analysis (PCA) plot, traits were visualized that differentiated PLC-OEs from WT under control (Fig. 3A) and drought conditions (Fig. 3B). In order to correct for age, and thus size of the plants, data from both PCA plots were normalized per day, for each day at control conditions, or during the period of water deprivation for the stressed plants. Arrow length and direction indicate how much and to which principal component (PC) each trait contributed. Data from the measuring devices, i.e. red/green/blue (RGB) and fluorescence, were separated into distinct PCA plots to facilitate their analysis (Supplementary Fig. S1). RGB measurements were taken at more time points, and PCA results showed greater explanatory power than the fluorescence measurements. Hence, we focused on the results obtained from RGB measurements. Data from the fluorescence measurements can be found in Supplementary Figs. S2–12.

Fig. 3.

PCA plots showing altered plant morphology responses of PLC-OE lines compared to WT in their response to water stress. PCA plots from control- (left panel, 21–45 DAS) and water-limiting conditions normalized to control (right panel; 30–38 DAS (day 5–13 of stress). Data were normalized per day and per trait to account for plants being larger at later days. Arrows size and directions indicate how much each trait contributes to each PC. Percentages of total variance represented by PC1 and PC2 are shown in parentheses. N = 40–60.

At control conditions, no specific traits were found that would distinguish PLC-OEs from WT (Fig. 3A). PLC-OE lines acted differently from each other and from WT, which is in line with earlier observations that OE of PLC did not result in striking growth phenotypes (van Wijk et al. 2018, Zhang et al. 2018a). Only PLC5-OE plants were found to be smaller and more compact, confirming earlier observations (Zhang et al. 2018b).

When looking at the effect of water deprivation, PLC-OE lines did group together and, more importantly, distinguished themselves from WT (Fig. 3B). The traits in this PCA plot can roughly be divided into three groups that clustered together: I [‘perimeter, convex hull perimeter, area by circumference and convex hull area’], II [‘compactness and stockiness’] and III [‘hue, saturation, brightness value and projected plant surface area’] (Fig. 3B).

‘Projected plant surface area’ represents the total number of pixels in RGB that was identified as plant (Fig. 4A). Under control conditions, OE of PLC7 and PLC9 exhibited a statistically significant larger ‘Projected plant surface area’ than WT. As mentioned earlier, PLC5-OE lines were slightly smaller than the WT, but at control conditions, this was not statistically significant. There were no significant changes for OE PLC2, PLC3 and PLC4. Under water deprivation, the changes between WT and PLC-OE lines became smaller and were no longer significantly different (Fig. 4B), except for PLC5-OE that was significantly smaller. By normalizing the ‘projected plant surface area’ of water deprivation data to the control conditions, we saw that the PLC-OE lines exhibited a significantly stronger growth decrease as a group than WT but not enough to have this be significant for any individual line (Fig. 4C).

Fig. 4.

PLC-OE does not change the ‘projected plant surface area’ under water-limiting conditions. Arabidopsis WT and PLC-OE lines plants were monitored using the GROWSCREEN-FLUORO system for indicated times. For each genotype, two different insertion lines with 30 replicates were used. (A) ‘Projected leaf area’ (schematically drawn in the upper left corner) at control conditions. (B) ‘Projected leaf area’ at water-withholding conditions (25–37 DAS). (C) ‘Projected leaf area’ at water-withholding conditions normalized to control. (D) Modeled growth rates at water deprivation conditions. Each individual plant was modeled, before, during and after drought stress, and then used to determine the relative growth rate per day. (E) The growth rate declines during water deprivation. Modeled growth rates for each plant during the water-withholding period were made using the formula: ax² + bx + c formula, where a was extracted as an indication of how fast growth declined in response to drought. Error bars represent a 95% confidence interval. Data were analyzed by one-way ANOVA followed by a Dunnett’s test. Statistically significant differences with WT are indicated by * (P < 0.05). (F) The ratio of growth during and after water withholding normalized to growth during and after withholding at control conditions. Error bars represent SEM.

The growth rate was determined to investigate how well the different lines continued to grow during water stress. However, as sowing dates were spread across 3 d, and measuring was performed at the same time, not all plants could be compared at the same DAS. To overcome this, growth curves were plotted for each individual plant. This modeled data could then be used to determine growth rate per day (Fig. 4D). As the growth per day is very much correlated to the initial size of the plant, we also determined the derivative of the ‘growth per day’ or ‘decline in growth per day’ for the time during water withholding (Fig. 4E). Of the different lines, only PLC9-OE plants exhibited a statistically significant bigger growth rate decline. Hence, we conclude that the drought-tolerant phenotype of PLC-OEs is not caused by a difference in growth response. Growth relative to control, during and after water withholding, was plotted to see whether a different growth strategy could be observed for the PLC-OEs (Fig. 4F). Growth during water deprivation mirrored the data from Fig. 4C, but during recovery, no differences between WT and PLC-OEs were observed.

The parameters in the PCA plots that allowed differentiation between WT and PLC-OEs were ‘convex hull perimeter’, ‘plant perimeter’, ‘convex hull area’ and ‘area by circumference’. These traits seemed to have a similar influence in separating the plant lines and also contributed greatly to the difference between WT and PLC-OEs. The ‘convex hull perimeter’ (Fig. 5) is determined by drawing a line across the extremities of each leaf to get the perimeter of the plant, which is different from the ‘plant perimeter’ (Supplementary Fig. S13) that follows the exact contours of the leaves. One might expect the ‘plant perimeter’ to be more precise, but values highly fluctuate as leaves tend to overlap with each other as plants grow older. ‘Convex hull area’ (Fig. 6) uses the lines drawn by the ‘convex hull perimeter’ to calculate the area. The ‘area by circumference’ (Supplementary Fig. S14) takes the longest leaf and draws a circle using that leaf length as the radius for the circle. In both cases, the convex hull data were more consistent, with less outliers, and revealing the same general patterns. Hence, we focused on the convex hull data. Both ‘convex hull perimeter’ and ‘convex hull area’ were slightly larger in PLC-OEs under control conditions, with the exception of PLC5-OE. For OE lines of PLC3, PLC7 and PLC9, this was statistically significant (Figs. 5A, 6A). However, this difference disappeared during water deprivation, with PLC5-OE becoming significantly smaller than WT (Figs. 5B, 6B). The fact that the PLC-OEs lose their slightly larger ‘convex hull perimeter’ and ‘convex hull area’ and thus respond more strongly to the water deprivation is clearly visible when normalizing the water deprivation data to control (Figs. 5D, 6D) and was statistically significant for PLC5-, PLC7- and PLC9-OE lines.

Fig. 5.

PLC-OE leads to a decrease in ‘convex hull perimeter’ in response to water stress. The ‘convex hull perimeter’ (schematically indicated in the upper left corner in A) was measured in (A) plants grown at control conditions; (B) plants at water-withholding conditions, with gray shading indicating a period of water-withholding (25–38 DAS); (C) water-withholding conditions normalized to control per day; (D) water-withholding conditions normalized to control conditions; (E) water-withholding conditions normalized to ‘projected plant surface area’.

Fig. 6.

PLC-OE plants exhibit decreased ‘convex hull area’ in response to water stress. (A) ‘Convex hull area’ (indicated in the upper left corner in A) at control conditions. (B) ‘Convex hull area’ at water-withholding conditions (25–38 DAS). (C)’ Projected plant surface area’ plotted against ‘convex hull area’. (D) ‘Convex hull area’ water-withholding conditions normalized to control conditions. (E) The normalized water-withholding data ‘projected plant surface area’ data plotted against the ‘convex hull perimeter’ data.

When comparing the ‘convex hull perimeter’ and ‘convex hull area’ to ‘projected plant surface area’ (Figs. 5C, 6C) under control conditions, no substantial differences were observed. However, under water-limited conditions, when comparing plants of equal sizes, the PLC-OE lines tend to have smaller ‘convex hull perimeter’. The same, although to a lesser extent, was found for the ‘convex hull area’. This was most clearly observed in the ‘projected plant surface area’ range of 50–100 mm², which corresponds to the average size of WT plants during water deprivation. When normalized to control conditions, it became clear that PLC-OEs had a greatly reduced ‘Convex hull perimeter’ and ‘convex hull area’ compared to WT, with the change being statistically significant for PLC5, PLC7 and PLC9 OEs throughout the whole stress period, while the rest of the PLCs exhibited only significant values at intermittent points (Supplementary Table S4).

The relation between ‘projected plant size’ and ‘convex hull area’ can be expressed by ‘compactness’ (Fig. 7). ‘Compactness’ is the ‘projected plant surface area’ divided by the ‘convex hull area’, so it gives the ratio of how much of the ‘Convex hull area’ is actually covered by the rosette. Similarly, plant size can be divided by ‘area by circumference’, resulting in the ‘stockiness’ of a plant. However, as ‘area by circumference’ had a higher variation than the ‘convex hull area’, ‘stockiness’ also had a higher variation than ‘compactness’ (Supplementary Fig. S15).

Fig. 7.

PLC-OE lines have increased compactness in response to water limitation. ‘Compactness’ is schematically indicated in the lower left corner in (A). (A) ‘Compactness’ at control conditions. (B) ‘Compactness’ at water-withholding conditions, with gray shading indicating the particular period (25–38 DAS). (C) ‘Compactness’ under control and water-withholding conditions. For each measurement day, ‘projected plant surface area’ is plotted against ‘compactness’. (D) ‘Compactness’ at water-withholding conditions normalized to control conditions. (E) The normalized water-withholding data ‘projected plant surface area’ data plotted against the ‘compactness’ data.

At control conditions, ‘Compactness’ reached a maximum at 28–30 DAS in all plants. Only PLC5-OE plants were significantly more compact than WT at control conditions (Fig. 7A). During water limitation, WT plants became less compact as stress prolonged. Strikingly, PLC-OEs retained their ‘compactness’ as at control conditions during the most part of the drought treatment and only decreased their compactness at the very last days of stress (Fig. 7B, D). As a group, the PLC-OEs were statistically more compact under, and in their response to, water stress than WT. PLC5-OE was always more compact than WT when comparing ‘plant size area’ to ‘compactness’ no matter the size of the plant, (Fig. 7C), while the other PLC-OEs are only more compact, relative to their size, in response to drought (Fig. 7E).

Chlorophyll content and fluorescence of PLC-OE plants under water deprivation

In this large-scale experiment, chlorophyll content was not directly measured, but indirectly using the ‘hue’ of a plant, which is correlated with its chlorophyll content (Majer et al. 2010, Liang et al. 2017, Farago et al. 2018). Since a calibration curve was lacking, only the relative chlorophyll content could be measured.

‘Hue’, ‘saturation’ and ‘brightness’ were determined by transforming RGB values of the area covered by the ‘projected plant surface area’. Under control conditions, ‘hue’ steadily increased over time, leveling off at the latest time points, indicating the greening of plants as they grow older (Supplementary Fig. S16A). Water-deprived plants initially increased their ‘hue’, but this rapidly decreased after 6–8 d of water withholding and increased again after rewatering (Supplementary Fig. S16B). Only PLC5-OE plants showed a significant difference with WT at control conditions, being slightly greener at the later timepoints. Apart from this observation, no significant differences between PLC-OEs and WT were observed in either condition, nor after normalization, or when compared to ‘projected plant surface area’ (Supplementary Figs. S16C–E). Similarly, ‘brightness’ (Supplementary Fig. S17) and ‘saturation’ (Supplementary Fig. S18) revealed no difference between WT and PLC-OE lines. Interestingly, ‘saturation’ responded very quickly to water deprivation, as 4 d after the start of the stress period already a decline was observed, while at control conditions the ‘saturation’ remained constantly increasing.

Photosynthetic activity is often measured using the F_v/F_m ratio, which indicates that how much of the photoreactive centers present are actually available for photosynthesis (Yao et al. 2018). Under stress conditions, this ratio is typically reduced to limit the production of ROS (Dalal and Tripathy 2018). Under control conditions, we observed that the F_v/F_m ratios remained more or less stable (Supplementary Fig. S6A) but indeed went down after 7 d of water stress (Supplementary Fig. S6B). In both conditions, however, no statistically significant difference between WT and PLC-OE lines was found. After normalization, some significant changes were found during the first couple of days of water deprivation, but this faded as stress prolonged (Supplementary Fig. S6C).

Biomass of PLC-OE plants after water deprivation

After 13 d of water deprivation, normal watering was resumed again. After a week of recovery, plants were harvested and ‘fresh weight’ (FW) (Supplementary Figs. S19A, B), ‘dry weight’ (DW) (Supplementary Figs. S19D, E) and ‘relative water content’ (Supplementary Figs. S19G, S19H) of the shoots determined. At control conditions, plants overexpressing PLC3, PLC7 and PLC9 were significantly heavier than WT, while PLC5-OE plants were significantly lighter than WT, in both FW and DW. The ‘relative water content’ at control conditions was only significantly different for PLC5-OE. After water stress, only PLC9- and PLC3-OE lines were significantly different in FW and DW compared to WT, and no differences in the ‘relative water content’ among the PLC-OE lines was recorded. Normalizing the water-stressed plants to their control counterparts revealed that none of the lines responded significantly different from WT to water stress (Supplementary Figs. S19C, F, I), which is in strong contrast to their survival rate in response to drought.

Phenotyping tissue-specific PLC5 lines

To investigate which cells/tissues are most important in gaining drought survival through PLC expression, a set of cell- and tissue-specific promoters was used to drive expression of PLC5 (Supplementary Table S2). This was based on earlier individual experiments in which PLC5-OE appeared to generate the strongest drought tolerance (data not shown). Normally, PLC5 is expressed in the vasculature (phloem), hydathodes and guard cells (Zhang et al. 2018b). Thirteen so-called TSEP lines were generated. An overview of the tissue specificity of the used promoters and a schematic representation of the expression of each line is given in Fig. 8A, B.

Fig. 8.

Survival under drought stress and localization of Tissue-specific Expression of PLC5. (A) Overview of promoter lines and in which tissue they are mainly expressed. (B) Schematic overview of plant tissues in which each line is expressed. The figure is adapted from Kim et al. (2017) and Vaughan-Hirsch et al. (2018). (C) Plant survival of the TSEP lines after water stress. Three-week-old plants were withheld from water for 3 weeks, after which they were watered again. Data were normalized using included WT as 0 and using the survival of the PLC5-OE line as 1. Each line was tested for two different inserts each tested two times over three different experiments.

Two independent transgenic lines for each construct were tested for their survival response to prolonged drought stress, with WT indicating the basal drought tolerance and UBQ10:PLC5-OE as a reference for the highest drought tolerance. As shown in Fig. 8C, increased survival was obtained with six promoters, i.e. DR5, E28, E49, LRC1, LRC3 and ML1-1. For the S2 promoter, only one insertion line showed the increased performance and was therefore excluded from subsequent analyses, as were TSEP lines that failed to show an increase in drought survival (i.e. A14, S1, CAB3 and CER6).

Interestingly, PCA plots of the six TSEP lines that showed enhanced drought survival were very similar to those of the PLC-OE lines (Fig. 9). At control conditions, no clear differences between WT and the TSEP lines were detected. However, TSEP lines did show distinct changes in response to water stress when compared to WT. The similarity in PCA plots between PLC-OEs and TSEPs indicated that similar phenotypes were occurring. Just like the PLC-OEs, TSEP lines revealed no significant difference in ‘projected plant surface area’ after normalization (Fig. 10A), indicating a similar response to water deprivation. At control conditions, however, these TSEP lines were significantly bigger than WT plants, with pDR5:PLC5 plants showing the most prominent increase in size (Supplementary Fig. S20A), though this was lost again during water deprivation. pLRC1:PLC5 plants were even significantly smaller in size at the final stage of water deprivation (Supplementary Fig. S20B).

Fig. 9.

Altered plant morphology of TSEP lines compared to WT in reaction to water stress. PCA plots from control- (left panel, 21–45 DAS) and water-limiting conditions (right panel; 30–38 DAS; day 5–13 of stress). Data were normalized per day and per trait to account for plants being larger at later days. Arrows indicate how much each trait contributes to each PC. Percentages of total variance represented by PC1 and PC2 are shown in parentheses. N = 40–60.

Fig. 10.

TSEP lines with increased drought survival respond similarly to PLC-OE plants under water deprivation. All data are from water-stressed plants normalized to control conditions. (A) ‘Projected plant surface area’, (B) ‘convex hull perimeter’, (C) ‘convex hull area’, (D) ‘compactness’ and (E) ‘hue’.

The stronger decrease in ‘convex hull perimeter’ and ‘convex hull area’ between WT and PLC-OE during water deprivation was again apparent for the TSEP lines (Fig. 10C, D), being statistically significant for the complete period of water deprivation, for all lines. Similarly, the corresponding ‘compactness’ was maintained during water stress for the TSEPs, only becoming less compact at the final days of water deprivation (Fig. 10D). ‘Hue’ was not affected in most of the PLC-OEs under any of the experimental conditions, except for PLC5-OE, which showed higher green values. TSEP lines #E49 and #ML1-1 also revealed this increase in ‘hue’ at control conditions, as did #LRC1 and #LRC3 at later timepoints (Supplementary Fig. S16F). During water deprivation, the #LRC1 lines exhibited significantly lower ‘hue’ (Supplementary Fig. S16G), and when the data were normalized, both #LRC1- and #ML1-1 lines responded significantly stronger than WT at later time points (Fig. 10E).

All six TSEP lines that exhibited increased drought survival (Fig. 8C) had increased biomass compared to WT under control conditions, in both FW and DW (Supplementary Figs. S21A, D). Under water deprivation, lines #DR5, #E28, #E49 and #ML1-1 were still significantly heavier than WT for both FW and DW. After normalization of the results of the water-deprived individuals to the ones of the controls, the differences in FW, DW and RWC among the tested lines were no longer apparent. This indicates that all the lines had the same relative biomass penalty after 2 weeks of water deprivation followed by a week of rewatering.

Discussion

OE of any PLC, including PLC9, improves drought tolerance

Earlier studies have shown that OE of PLC results in increased survival rates in response to water deprivation in various plant species (Wang et al. 2008, Georges et al. 2009, Tripathy et al. 2012), including Arabidopsis PLC3, PLC4, PLC5 and PLC7 (van Wijk et al. 2018, Zhang et al. 2018a, 2018b, van Hooren et al. 2023). Here, we show that OE of PLC2 and PLC9 confirmed these results, suggesting that OE of any PLC can achieve this. While this is an interesting observation, it remains unclear whether plants also use endogenous PLC signaling as a protective reaction against water stress. A number of PLCs are indeed induced in response to drought, osmotic stress and/or salinity (Hunt et al. 2004, Tasma et al. 2008, Li et al. 2015), and PLC has also been linked to ABA, an important phytohormone in drought signaling (Hirayama et al. 1995, Staxen et al. 1999, Sanchez and Chua 2001, Hunt et al. 2003, Mills et al. 2004, Munnik 2014, Zhang et al. 2018a). The mechanisms for how drought tolerance is achieved are still unknown to the PLC-OEs.

To increase our understanding of how PLC-OE leads to increased drought tolerance, a large phenotyping experiment was initiated with six distinct PLC-OEs and 13 tissue-specific PLC5-expression lines. For each of these lines, we used two distinct insertions, and for every insertion, 30 plants were used. The PCA plot (Fig. 3 and Supplementary Fig. S1) clearly revealed that in response to water deprivation, PLC-OE lines grouped differently from WT. This was especially notable for PLC9-OE lines, as the PLC9 enzyme lacks two key histidines in the catalytic site (Ellis et al. 1998). However, OE of PLC9 clearly increased the survival rate under drought stress (Fig. 2B, C). Hence, we propose that either the change in the amino acids in the catalytic site does not interfere with the enzymatic function of PLC9 or the increased survival found in PLC9-OEs is not directly linked to the activity of the enzyme. Other ways in which PLC9 could function is by forming protein complexes through protein–protein interactions or by preventing PPI signaling through shielding PPIs from binding other protein targets. Alternatively, if PLC enzymes would function as heterodimer with other PLCs, PLC9 could function as a silent partner.

That PLC9 has activity, directly or indirectly, is also indicated by earlier studies, showing that OE of PLC9 increased heat stress tolerance, while plc9 KO mutants were more sensitive (Zheng et al. 2012, Gao et al. 2014, Ren et al. 2017, Liu et al. 2020). Another explanation for the observed mutations in the catalytic domain of PLC8 and PLC9 is that these members evolved an alternative catalytic domain. While two crucial histidines were lost, two new cationic amino acids were regained, i.e. E390K and Y155R. Perhaps the latter is used to accommodate the negative charge of the phosphorylated inositol headgroup. A possible advantage of the observed mutations could be that the enzymatic activity of PLC8 and PLC9 is less pH-sensitive, as histidines lose their positive charge at pH >6, as opposed to lysine and arginine that maintain their charge at higher pH values. Heterologous PLC9 expression and in vitro enzyme activity assays will help investigating this further. Similarly, PLC8-OE lines could be generated to confirm the improved drought tolerance.

PLC-OE and drought tolerance correlate with increased compactness

PCA plots revealed that PLC5-OE grouped differently from other PLC-OEs, which was particularly evident at control conditions. This is in line with earlier observations from Zhang et al. (2018b), where a ‘semi-dwarfed’ phenotype was documented. Dwarfism is often associated with drought tolerance, as small shoots transpire less water, and hence, more water remains available for the plant, making them appear as being more drought tolerant (Nishiyama et al. 2011, Maggio et al. 2018, Kudo et al. 2019, Bhatnagar et al. 2020, Yang et al. 2020). However, none of the other PLC-OE lines exhibited this ‘semi-dwarfed’ phenotype, so the reason for the increased drought tolerance cannot be simply due to dwarfism. Moreover, at control conditions, the rest of the PLC-OE lines tend to grown bigger (Fig. 4A) and produce more biomass (Supplementary Figs. S19A, D). Under water deprivation, this difference became less significant although it was still present in some of the tested lines (Fig. 4B; Supplementary Fig. S19B, E). When specifically looking at the rate of growth decline, no change was observed with the exception of PLC9-OEs, which declined stronger (Fig. 4E). Normalization of the water-deprived plants showed that drought stress leads to a non-significant decrease in size (Fig. 4C), which was compensated after recovery (Supplementary Figs. S19C, F). Hence, we conclude that the shared increase in drought survival of the PLC-OEs is not due to a growth penalty.

Most stresses, including water stress, are associated with a decrease in photosynthetic activity and an eventual decline in the amount of chlorophyll (Chaves et al. 2009). In this experiment, we saw the same thing but found no difference between WT and OE lines in F_v/F_m ratios (Supplementary Fig. S6), nor for ‘hue’, a readout for chlorophyll content (Supplementary Fig. S16). At control conditions, PLC5-OE lines were significantly greener than WT although increased chlorophyll levels may be explained by the reduced cell size, which could also be the reason for the smaller shoot phenotype. This indicates that the increased survival in the OE lines was not caused by changes in photosynthetic capacity although PLC interference leading to increased ROS turnover and an enhanced stress response could still be a possibility. It should be noted that F_m level, which indicates the amount of active reaction centers available for photosynthesis, was higher in all the PLC-OE lines at control conditions although this was not statistically significant (Supplementary Fig. S5A).

The PCA plots did reveal changes correlating to the shoot architecture. In response to water deprivation, both ‘convex hull perimeter’ and ‘convex hull area’ decreased, an effect that was significantly greater in all PLC-OEs. Since PLC-OE lines tend to be bigger than WT, and hence are expected to have larger perimeters, the two factors appear to be correlated. However, when plotting the two parameters against each other, a smaller perimeter and area for the PLC-OEs were recorded when plants of the same size were compared, indicating that a change in the architecture has occurred (Figs. 5C, E, 6C, E). The ratio of the ‘projected plant surface area’ and ‘convex hull area’ can be given as a ‘compactness’ value, and under water-deprived conditions, the ‘compactness’ of the PLC-OEs as a group was significantly more affected than for WT (Fig. 7D).

No effect in ‘relative water content’ at any condition for any of the tested lines, as compared to WT, was found (Supplementary Figs. S19G, H, S21G, H), so it is unlikely that PLC-OE has a beneficial effect for the plants, giving them the advantage to deal better with low water potential (which is linked to the drought tolerance strategy). We do find that PLC-OE results in a stronger response related to shoot size and architecture, traits that are more associated with an increased drought-avoidance strategy.

The increased survival under water deprivation is a common characteristic feature of all tested, independent PLC-OE lines. In this study, an architectural change in response to water stress was found. This modification was not limited to the PLC-OE lines but was also found in the six TSEP lines that exhibited increased survival under water deprivation (Figs. 8, 10). The fact that the architectural changes and increased survival phenotypes can be reproduced by expressing PLC5 in various tissues indicates that the phenotypes from PLC-OEs are probably caused by ectopic expression and not from a higher expression in tissues where PLC is already expressed.

Tissue-specific ectopic expression of PLC5

Of the 13 TSEP lines constructed, six showed an increased survival phenotype in response to water deprivation in two independent insertion lines. The expression of these promoters is mainly restricted to the developing parts of the root. One of them, #E28 targets the expression of the transgene in the root pericycle and endodermis, while #E49 is specifically expressed in the cortex of the elongation zone. In contrast, #E30 line, which did not show increased survival to water deprivation, targeted the expression of PLC5 in the endodermis of the root maturation zone. #S1, similarly, gives expression in the root epidermis and pericycle but only in the late elongation zone. The different tissues in the mature root have important physiological roles, but it appears that the targeted PLC expression therein does not affect the survival rate under drought stress. Both promoters #LRC3 and #LRC1 are mostly active in the lateral root cap (LRC), but #LRC1 also triggers the expression in the root and shoot epidermis, similarly to #ML1-1. The LRC plays an important role in the early development of the root (Di Mambro et al. 2019). It is the primary anatomical element to get in contact with the surrounding environment, and thus it is the first to sense, and probably prone to react to, the changes in parameters such as temperature, water and nutrient availability or mechanical barriers.

#DR5 expression coincides with auxin maxima (Ulmasov et al. 1997) and thus, in the case of roots, with the zones of intensive growth. Consequently, it has a high overlap with the sites of expression of all the other constructs that target PLC5 expression in developing parts of the root. The absence of any effects on the survival rates of the #CAB3, #CER6 and #Q4 transgenic lines indicates that the ectopic expression of PLC5 in either mesophyll cells, guard cells or the quiescent center alone does not provide physiological advantages for better survival under water deprivation. Most of the parameters evaluated in the present study are associated with the above-ground parts of Arabidopsis; however, the observed pattern of increased survival of lines with targeted PLC5 expression in the most dynamic parts of the root (cell division and elongation zone) gives a new organ-specific direction for further research on the role of PLCs in drought perception and/or its response.

Even though there were rosette size differences in the six TSEP lines that exhibited increased survival, with #DR5 being by far the largest and #LRC1 plants having the smallest rosettes among all water-deprived individuals, they all appeared more compact under water stress. While this parameter can easily be measured, in big phenotyping experiments, it is often ignored, with a few exceptions (Skirycz et al. 2011, Prerostova et al. 2018, la Rosa N et al. 2019).

Compactness architecture and drought tolerance

Specific studies on compactness in Arabidopsis revealed that it generally decreases as plants grow older (la Rosa N et al. 2019). A circadian rhythm on compactness has also been reported, which most likely reflects the change in leaf angle during day and night, being almost horizontal, and therefore the least compact during midday and becoming more compact at night (Dhondt et al. 2014). This is unlikely to ultimately affect the data presented in this study, as plants were randomly measured.

In response to salt stress or H₂O₂, though not in response to mannitol, plants tend to become more compact (Jansen et al. 2009, Dhondt et al. 2014). A slight compactness decrease can be found in response to drought in the supplementary data of la Rosa N et al. (2019); however, there were no statistics to accompany these data. Our data confirmed that the compactness of plants decreased while growing older and that water-limiting conditions induced an even stronger decrease. Since compactness is easily measured in automated phenotyping experiments, and often already is, and is affected by (abiotic) stress, a greater focus on this trait could help future research to characterize and identify stress responses even better.

An increase in ‘compactness’ could result from different variables, including leaf shape, leaf angle and shorter petiole length. However, no significant difference in ‘leaf diameter’ measurements was found among all tested lines (Supplementary Fig. S28), and PLC-OEs plants even tend to exhibit narrower leaves, arguing against the idea that leaf shape is the main cause for the observed differences in compactness between the WT and the PLC-OEs under water deprivation.

Low turgor pressure may cause leaves to wilt, thus decreasing their standing angle and subsequently the ‘compactness’ of the monitored individual. However, if this would have been the sole explanation of the detected difference in ‘compactness’, then we would expect the alteration to occur when plants stop growing. However, the difference was already observed within the first few days of water deprivation.

Changes in petiole elongation have not been well studied in the context of drought stress. Most published data are related to its role in thermomorphogenesis, where plants grow longer petioles and increase their hyponasty when grown at temperatures above 26–28°C, which decreases their compactness substantially (Ibanez et al. 2017). One of the main mechanisms by which plants are able to cool down under heat stress is to evaporate water (transpiration). This process is limited by the water supply and air humidity. Modeling of the air flow around heat-stressed leaves predicts that a decrease in compactness would increase the air flow around the leaves, improving the removal of water-saturated air around the leaves, and enabling a steady supply of air to evaporate water into (Bridge et al. 2013). This finding is not surprising, as transpiration is exactly the same mechanism by which many mammals cool down. If increasing the petiole length can facilitate water evaporation, then the opposite, an increase in compactness, might help in conserving water. A compact canopy could reduce air flow underneath the leaves, making the air therein increasingly humid. Water evaporates badly in humid air, and therefore the stomata that remain open at the same level would lose less water (Gonzalez et al. 2012).

To study the change in compactness and its relation to drought in more detail, future studies could make use of different mutant lines. The plc-KO lines do not have decreased drought tolerance (Zhang et al. 2018a, b, van Wijk et al. 2018); if these lines do show changed compactness levels, then this might indicate whether the compactness and drought phenotype are related. Furthermore, measuring the compactness of known drought-tolerant lines, preferably lines of which the underlying reasons the tolerance is known, would indicate if the compactness phenotype is present in more drought-tolerant plants.

So far, research on ‘compactness’ has mainly centered on grasses (grains and rice), in order to grow more plants in the same space. Interestingly, also in grasses, an increase in drought tolerance has been correlated with ‘compactness’ (Honsdorf et al. 2014, Kim et al. 2020, Gano et al. 2021). Hence, this correlation might be interesting for breeding crops as well, as it simultaneously invests into drought tolerance based on drought avoidance rather than drought escape. Further research into this correlation may further substantiate the ideas brought forward here. The next challenge will be to unravel the molecular mechanism behind this phenomenon.

Materials and Methods

Plant material

Arabidopsis thaliana ecotype Col-0 was used as the genetic background for all the lines used in this research. Plants were grown for at least one generation together and synchronized seed batches were used. In the GROWSCREEN-FLUORO setup, the means of two Col-0 WTs were used: one being the line that was grown together with the PLC-OE lines for several generations and the other was freshly ordered from the Nottingham Arabidopsis Stock Centre (Scholl et al. 2000). UBQ10:PLC3 (AT4G38530) #9 and #16, UBQ10:PLC5 (AT5G58690) #2 and #3 and UBQ10:PLC7 (AT3G55940) #9 and #17 were described earlier (van Wijk et al. 2018, Zhang et al. 2018a). The 35S:PLC4 (AT5G58700) #2 and #4 line was kindly provided by Dr Rodrigo Gutiérrez (Universidad Catolica de Chile). PLC2-OE (AT3G08510) PLC9-OE (AT3G47220) and 13 TSEP lines were generated in this study and will be described later.

KO plc9-1 (SALK_025949) and plc9-2 (SALK_021982) lines were obtained from SALK (signal.salk.edu), and a plc2-1 mutant in Ws background was kindly provided by Dr. Ana Laxalt (University of Mar del Plata, Argentina; Di Fino et al. 2017).

Plant growth conditions

For genotyping, qPCR and Western blot analyses, plant material was grown on agar plates. Seeds were first surface sterilized in a 3.5 l desiccator with 20 ml of thin bleach and 600 µl of 37% w/w HCl for 3–4 h and then stratified in 0.1% (w/v) daishin agar at 4°C for 2–4 d in the dark. Stratified seeds were sown on square Petri dishes (Greiner, Kremsmünster, Austria) containing 40 ml of ½ MS medium, supplemented with 1% (w/v) sucrose, 0.1% (w/v) MES (pH 5.8) and 1% (w/v) daishin agar (Duchefa Biochemie, Haarlem, The Netherlands). Plates were placed vertically at an angle of 70° at long-day conditions (22°C, RH 70%, 16 h of light and 8 h of dark) and scanned at different time points (Epson Perfection V700, Suwa, Japan).

Construction of PLC2-OE, PLC9-OE and TSEP lines

pUBQ10:PLC2, p2x35SP:PLC2, pUBQ10:PLC9 and 13 ‘TSEP’ lines were constructed using gateway cloning. Oligonucleotide primers including Attb1 and AttB2 sites were used to PCR amplify the coding sequence (CDS) of PLC2 and PLC9 from cDNA (Supplementary Table S1). The PCR fragment was inserted into the pDONR207 vector with BPII clonase enzyme mix to generate a BOX2-vector entry clone, and entry clones for pGEM-UBQ10, pGEM-2x35S and pGEM-StopTnos vectors were kindly provided by Dr Ben Scheres (Utrecht University, The Netherlands). The three entry clones were recombined in a pGreenII-125 expression vector using the Multisite Gateway Three-Fragment Construction Kit (www.lifetechnologies.com) and the supplier’s protocol and checked by PCR and subsequent sequencing. (https://www.thermofisher.com/nl/en/home/life-science/cloning/gateway-cloning/gateway-technology.html).

TSEP lines were generated by recombining the tissue-specific promoters (Supplementary Table S2; Vaseva et al. 2018) in pENTR5 vectors, the coding sequence of PLC5 in pGreenII029JV (Zhang et al. 2018b) and pGEM-StopTnos in a pGreen0125-expression vector using Multisite Gateway Cloning. Constructs were validated for PLC5 mutations by PCR and subsequent sequencing. PLC5 was chosen since pilot studies indicated that PLC5-OE had the strongest increase in drought survival.

All constructs were cotransformed with vir-helper vector, pSOUP into Agrobacterium tumefaciens strain GV3101, which was then used to transform Arabidopsis by floral dipping (Clough and Bent 1998). At least two homozygous lines were selected by norflurazon selection in T3 generation and used in the next experiments. The norflurazon selection was used as a proxy to determine the successful integration and expression of the TSEP constructs.

RNA extraction and qPCR

PLC2 and PLC9 expression in KO and OE lines was analyzed using qPCR. Total RNA was extracted from 7-d-old seedlings that were homogenized in liquid nitrogen with mortar and pestle using a homemade TRIzol reagent. RNA (0.5 μg) was converted into cDNA using oligo-dT18 primers, dNTPs and SuperScript III Reverse Transcriptase (Invitrogen, Waltham, Massachusetts) according to the manufacturer’s instructions. qPCR was performed using HOT FIREPol EvaGreen qPCR mix Plus (ROX) (Solis Biodyne, Tartu, Estonia) with an ABI 7500 Real-Time PCR System (Applied Biosystems). Relative expression levels were determined by comparing the threshold cycle values for PLC2 (AT3G08510) or PLC9 (AT3G47220) to the reference gene, SAND (AT2G28390) (Czechowski et al. 2005). Primers for SAND, PLC2 and PLC9 (being negative for PLC8) are listed in Supplementary Table S1. Three technical replicates and three biological replicates were used, with each biological replicate consisting of 12–15 seedlings. Changes were tested for significance with a Student’s t-test.

Protein isolation and Western blotting

For protein extraction, 50 µl of ice-cold protein extraction buffer (0.5% (v/v) NP-40, 75 mM NaCl, 100 mM Tris-HCl (pH 8.0), 1% (w/v) PVPP, 100 mM DTT, 1 mM PMSF and 10 mM EDTA (pH 8.0), supplemented with protease inhibitor EDTA free (Roche; just before use) were added to 10 mg root material of 5 DAG seedlings grown on plates that was ground in liquid nitrogen. During the whole procedure, material was mixed well and kept cold. Samples were centrifuged 2 × 5 min at 2,000×g followed by 2 × 5 min at 5,000×g, each time transferring the supernatant to a new tube. Protein concentrations were estimated using Nanodrop (ThermoFisher, Ochten, The Netherlands).

SDS–PAGE gels (10%) were loaded with 4–5 µg of boiled and centrifuged protein sample (10 min at 95°C; 2 min at 16,000×g) in a sample buffer (containing 2% SDS (w/v), 10% glycerol (v/v), 5% β-mercaptoethanol (v/v), 60 mM Tris-HCl pH 6.8 and 0.02% (w/v) bromophenol blue) and ran for ∼1 h at 30 mA/gel. Gels were blotted to PVDF membrane (Immobilon-P, Merck, Rahway, New Jersey) that were then incubated for 1 h with a PLC2-antibody (1:2,000) provided by Dr Ana Laxalt (D’Ambrosio et al. 2017). Filters were washed four times for 5 min with a phosphate-buffered saline (PBS) buffer containing 0.05% Tween-20 (v/v), incubated with a second antibody, polyclonal Goat anti-rabbit IgG (H + L)/HRP (1:5,000; ThermoFisher) for 1 h, washed 6 × 10 min with PBS buffer containing 0.05% Tween-20 (v/v) and imaged using ECL.

Drought tolerance assay

Seeds were stratified in liquid for 2 nights in the dark at 4°C and thereafter directly sown in soil (pots with diameter from 3.8 to 5 cm and height of 5 cm for PLC9-OE and pots of 4.5 × 4.5 cm and height of 7.5 cm for PLC2-OE), containing 30 or 80 g of soil, respectively (Zaaigrond nr. 1, SIR 27010-15, JongKind BV, Aalsmeer, The Netherlands). Plants were grown at short-day conditions (22°C with 13 h light/11 h darkness).

In the drought experiment with PLC9-OE and TSEP lines, 15 plants per line were grown for 3 weeks and watered every other day. After these 3 weeks, the excess water was removed from the trays, and the plants were withheld from water for 3 weeks. At the end of the experiment, plants were rewatered again to allow the survived individuals to recover and turn green. During the experiment, pots were randomized twice per week, using a shuffling schedule that would distribute the plants equally on random positions at the inside and outside of a tray and in the corners. The experiment was performed three times, with each line included at least two times. In the drought experiment with PLC2-OEs, nine plants per pot were grown for 37 d in well-watered conditions, after which plants were left to dry for an additional 18 d, with pots being randomized twice a week. In both experiments, plants were rewatered again after the drought period and left to recover for 9 d before pictures were taken. The experiment was performed twice, giving similar results.

PLC9 protein structure modeling

The crystal structure PLC-δ1 of R. norvegicus (1DJX) (Essen et al. 1997) was used as a template to build models of the Arabidopsis PLC5 and PLC9 proteins using https://evcouplings.org/, which were then further visualized using Novavold® (Version 17.2; DNASTAR, Madison, WI, USA).

Phenotyping platform and drought stress setup

For the phenotyping, the GROWSCREEN-FLUORO setups in the GROW-SCREEN chamber facilities of IBG-2 at Plant Sciences of the Forschungszentrum Jülich (Jülich, Germany) were used, as have been described earlier (Jansen et al. 2009, Barboza-Barquero et al. 2015, Prerostova et al. 2018). Single seeds were sown in multi-well trays filled with a clay-peat substrate (Einheitserde Typ Mini-Tray, Einheitserdewerke Werkverband e.V., Sinntal-Altengronau, Germany) using a phenoSeeder robot (Jahnke et al. 2016). After stratification for 4 d at 4°C, trays were transferred to a climate-controlled GROWSCREEN chamber at 22°C/18°C (day/night), 50% relative air humidity and an 8/16 h light/dark period. Seed germination and seedling growth were monitored daily until transplantation using an automated imaging and image processing system. Fourteen DAS, seedlings with the largest leaf surface area per line, determined by an automated system, were transplanted to pots (7 × 7 × 8 cm) filled with peat–sand–pumice substrate (SoMi 513 Dachstauden; Hawita, Vechta, Germany) and watered to field capacity. Pots were arranged in trays that were positioned in two independent GROWSCREEN chambers, set at the same environmental conditions as for germination and seedling growth. Pots were first dried to 60–40% of field capacity and kept at that moisture level until the start of drought treatment (withholding water) at 25 DAS. Pots of the control treatment were maintained between 60% and 40% of field capacity until harvest at 45 DAS, whereas pots of the drought treatment were dried to between 8% and 5% of field capacity over a period of 13 d. Pots were then rewatered to 60% of field capacity and left to recover for 7 d until harvest at 45 DAS. Shoot biomass was measured at harvest as FW, followed by DW, after drying rosettes for 7 d at 65°C. The ‘relative water content’ was determined using the formula: (FW − DW)/FW × 100%.

Per construct, two independent insertion lines were used, and for each insertion line (and WT), 30 plants were used per treatment, making 2,460 plants in total. All plants were distributed according to a complete randomized block design across a total of 66 trays with 40 pots each. Trays for both treatments were placed in the GROWSCREEN chambers where a robot daily delivered the trays to a weighing station and randomized them by placing them back in another position. Trays were watered twice per week. The robot also delivered the trays to the GROWSCREEN-FLUORO imaging station. Imaging started at 21 DAS and was repeated every 2–3 days until harvest for RGB value imaging. From 28 to 38 DAS, during the water-withholding period, RGB imaging was performed daily. Chlorophyll fluorescence of dark-adapted plants was imaged twice per week. Image analysis was performed by in-house developed software and included the RGB images, a first rough segmentation of shoots from the substrate background using thresholds on ‘hue’, ‘saturation’ and ‘brightness’ values. This was followed by a precise segmentation based on a trained support-vector machine algorithm. The obtained phenotypic traits included ‘projected shoot surface area’, ‘shoot perimeter’, ‘convex hull area’, ‘convex hull perimeter’, ‘compactness’, ‘surface coverage’, ‘stockiness’, ‘area by circumference’ and the average ‘RGB’ values and ‘hue’, ‘saturation’ and ‘brightness’ values of all shoot pixels. Chlorophyll fluorescence data were processed as described by Jansen et al. (2009), resulting in the traits F₀, F_m and F_v/F_m.

Data processing

Data analysis was performed in R, using a script shared at https://github.com/MaxvHooren/GROWSCREEN-analysis. Because the plants from one of the climate chambers were slightly bigger, a correction was applied, which was determined by calculating the difference between the chambers per day and per treatment. These differences were averaged and then applied as a correction factor. This process was repeated for every trait measured. PCA was performed with the FactoMineR and FactoExtra packages. The data were normalized per day before doing PCA, as older/bigger plants would have had a bigger impact on the estimated differences.

As not all seeds were sown on the same day, measurements and DAS were not aligned. Therefore, it was difficult to establish the relative growth rates per day and to compare the changes in this parameter between different groups, as different numbers of plants were measured per line on each day. In order to overcome this, growth curves were fitted to ‘rosette projected-surface area’ over time for each individual plant to enable the calculation of growth rate. For data obtained during control conditions, an exponential growth curve (y = a*e^Bx) was fitted, whereas a quadratic function (y = ax² + bx + c) was used for data obtained during water withholding.

The statistical significance among the tested lines for every day was determined by an ANOVA test, followed by a Tukey’s honest significant difference test, comparing the values with the combined WT values (Supplementary Table S4). Blank timepoints correspond to times when either the line in question or WT was not measured.

After the analysis in R, bar graphs were built using GraphPad Prism version 8.4.2 for Windows (GraphPad Software, San Diego, CA, USA; www.graphpad.com).

Supplementary Data

Supplementary Data are available at PCP online.

Data Availability

The raw data used in this paperis stored at the GrowSreenChamber institution on the ‘Phenomis’ database at IBG-2 and can be supplied upon request. The script for analysis can be found at https://github.com/MaxvHooren/GROWSCREEN-analysis. The analyzed data are available at https://zenodo.org/record/7715023. All other data supporting the findings of this study are available from the corresponding author (T.M.).

Funding

Netherlands Organization for Scientific Research (867.15.020 to T.M.); European Union’s Horizon 2020 Research and Innovation Programme EPPN2020 (731013 to T.M.).

Author Contributions

M.v.H. and T.M. designed the experiments. R.v.W. created the PLC2-OE and PLC9-OE lines. M.v.H. created the tissue-specific PLC5-OE (TSEP) lines using the promoter vectors created by I.I.V. and D.V.D.S. R.v.W. performed qPCR, Western blot analysis and drought experiments on PLC2-OE. M.v.H. performed qPCR and drought experiments on PLC9-OE and TSEP lines. M.v.H. helped with sowing, transferring and harvesting of the GROWSCREEN experiment, which itself was conducted by IBG-2 at Plant Sciences of the Forschungszentrum Jülich. M.v.H. performed the data analysis, and M.v.H. and T.M. constructed the figures and wrote the article. R.v.W., I.I.V., D.V.D.S. and M.H. critically read the manuscript and made valuable corrections.

Disclosures

The authors have no conflicts of interest to declare.