The products of sexual fertilization in flowering plants (seeds, grains, fruits, etc.) provide most of the world's food and feed supply (Kern, 2002;Aizen et al., 2008;Thow and Priyadarshi, 2013;Tuncel and Okita, 2013). Successful fertilization is critically dependent on the fitness of the male gametophyte (i.e. pollen) during pollen grain maturation, dispersal, germination, tube growth and fertilization (Firon et al., 2012;Carrizo Garcia et al., 2017). Being able to determine pollen grain viability provides an important control in evaluating the reproductive potential of plants, and for basic research on fundamental questions in pollen cell biology.
Pollen development begins in the anther locules when a diploid microspore mother cell undergoes meiosis to produce a tetrad microspore. Each microspore divides mitotically once or twice (depending on the species). This is followed by a maturation stage when the pollen desiccates in preparation for dehiscing of the pollen sac and release of the mature pollen. Successful fertilization requires pollen to hydrate on the stigma of receptive carpels, grow a tube through the pistil, guide the tube toward the ovule, and discharge the two sperm cells inside the ovule to fuse with both the egg and central cells (Suzuki, 2009). Because these processes can be easily disrupted by environmental changes, pollen is extremely vulnerable to stresses such as high temperatures and low humidity, with even small changes in temperature causing pollen abortion or failure to germinate and grow a tube, resulting in reduced seed set and sometimes complete male sterility (Zinn et al., 2010;Rieu et al., 2017).
Various assays have been used to evaluate pollen fitness. The most commonly used is Alexander stain (AS), which discriminates live and dead (aborted) pollen by differential staining of red and turquoise color, respectively (Alexander, 1969;Peterson et al., 2010;Andreuzza et al., 2015). However, it has been reported that AS sometimes fails to distinguish live and dead pollen in some plant species, including Arabidopsis thaliana, perhaps due to protocol variations (Peterson et al., 2010). Other common methods use tetrazolium salts as indicators of mitochondrial dehydrogenase activity, such as thiazolyl blue tetrazolium bromide [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide; MTT] or 2,3,5-triphenyl tetrazolium chloride (TTC), that stain live pollen dark purple/black or orange, respectively (Endo et al., 2013). A significant limitation of these histochemical techniques is that they are cytotoxic, which prevents any follow-up analyses on living cells.
A non-cytotoxic staining protocol has previously been used in pollen that employs fluorescein diacetate (FDA). FDA is a cell-permeable molecule that becomes trapped in cells following hydrolysis by cytoplasmic esterases (Heslop-Harrison et al., 1984;Pacini et al., 1997;Wang et al., 2004). However, one of the limitations of this reporter is that some FDA can still be hydrolyzed to fluorescein in non-viable and thereby produce false-positive results (Clarke et al., 2001).
The limitations of prior methods led us to develop a simple fluorescent stain-based alternative that is also amenable to use in rapid analyses of large populations of pollen by flow cytometry. The method presented here utilizes dichlorodihydrofluorescein diacetate (H 2 DCFDA), which is a derivative of FDA that only becomes fluorescent after oxidation by reactive oxygen species [ROS; i.e. dichlorofluorescein (DCF)-stained]. Only live cells with metabolic activity generate enough ROS to oxidize H 2 DCFDA, which makes it a robust reporter for cell viability as well as endogenous ROS levels. Applying this approach with flow cytometry, we discovered that a population of pollen grains from Arabidopsis or tomato flowers consistently shows a bimodal distribution characterized by low-or high-ROS levels, representing low or high metabolic activity, respectively. Following purification of the two subpopulations by fluorescence-activated cell sorting (FACS), we demonstrate that under favorable conditions the ability of pollen to germinate is positively correlated with the level of ROS. We also show how this technique can be used to gain insights into how pollen responds at the population scale to heat stress (HS), oxidative stress and antioxidative treatments.
Quantifying pollen viability is important in understanding how pollen responds to a stress condition. Therefore, we compared the effectiveness of the three most widely used staining procedures for scoring pollen viability; AS and the two tetrazolium salts MTT and TTC. The staining protocol of each dye was applied to Arabidopsis pollen imbibed in vitro in pollen germination medium (PGM) under optimal temperature (22°C), heat-stressed (37°C) or neutralizing temperatures (heated at 80°C) for 90 min. The percentage of pollen deemed viable by staining was compared in parallel analyses with pollen germination frequencies in vitro. As expected, germination frequencies were significantly higher under optimal conditions compared with a severe 90% reduction or no germination following 37°C or 80°C treatments, respectively (Figure 1a). Surprisingly, about 99% of pollen was positively stained red by AS, indicating viability, regardless of the temperature treatment (Figure S1b,e). In contrast, none of the pollen grains was stained by TTC (Figure S1d,e). The failure of AS to detect dead pollen grains, and the inability of TTC to stain at all, was similar for tomato and Nicotiana benthamiana pollen, regardless of growth conditions that included: 22°C, high temperatures, heavy metals (aluminum) or an oxidative stress with tert-butyl hydroperoxide (Figures S2 and S3).
In contrast to the unreliability of AS or TTC, staining with MTT appeared to be a reliable indicator of viability. For example, the percentage of black-stained (viable) pollen was 91, 48.3 and 0% following the 22, 37 and 80°C treatments, respectively. This was consistent with results from parallel assays to determine the percentage of pollen germination and tube growth in vitro (Figures 1a and S1c, e). Additional studies with different stress conditions and pollen from other species confirmed the reliability of MTT as a viability reporter for pollen (Figures S1-S3).
H 2 DCFDA reliably stains pollen of a wide variety of plants Although MTT staining was confirmed as a reliable method for distinguishing between dead and active pollen, the staining process is ultimately cytotoxic (Riss et al., 2004). In looking for a fluorescent alternative to histochemical staining that would be more suitable for dynamic cytometric measurements in live cells, we tested the fluorescein-based dye H 2 DCFDA (an indicator for ROS). Similar to FDA, which has been used as viability stain (Heslop-Harrison et al., 1984), H 2 DCFDA is highly permeable to cells and is retained in live cells following cleavage of an acetate group by intracellular esterases. In contrast to FDA, the non-fluorescent H 2 DCFDA requires an additional oxidation step by ROS to be converted to the highly fluorescent DCF.
To verify the efficacy of H 2 DCFDA as a viability stain, we used epifluorescence microscopy to compare DCF-staining of Arabidopsis pollen under optimal conditions (22°C), or following exposure to 37°C or 80°C treatments (Figure 1a). Consistent with H 2 DCFDA-staining as an indicator of viability, most of the cells from the 22°C treatment showed significant DCF fluorescence, whereas only background autofluoresence was observed for cells subjected to the 80°C neutralizing treatment (Figure 1b). These staining results correlated with a 90% decrease in pollen tube production following 37°C treatment, and zero tubes following an 80°C treatment (Figure 1a). DCF also stained the cytoplasm of growing tubes, further demonstrating its ability to fluoresce in metabolically active cells (Figure S4). The results for DCF fluorescence were in agreement with MTT staining under the same pollen incubation conditions (Figure S1).
H 2 DCFDA staining was also tested on pollen from divergent species, including tomato, N. benthamiana, common poppy, cucumber and sweet basil producing, all of which produced strong fluorescent signals that clearly discriminated between live and dead pollen (Figure S5).
Taken together, these assays indicate that H 2 DCFDA staining is a sensitive and reliable method for scoring live pollen grains in a wide range of plant species.
To accurately evaluate viability in a large population of pollen, we tested DCF-stained pollen using flow cytometry. To this end, we used the BD FACSAria TM III for the analysis of filtered pollen grains. Using forward scatter (FSC-A) and side scatter (SSC-A) parameters (i.e. size and granularity, respectively), gating of the imbibed pollen population was performed (Figures 2a and S6a), which discriminates them from other cells and floral debris. The population of single separated grains was then discriminated from droplets containing pollen grain aggregates (i.e. application of double discrimination gating; Figure S6b). DCF fluorescence was measured using the 485-nm excitation laser and 530nm emission filter (fluorescein isothiocyanate; FITC). To determine the minimal positive DCF fluorescence that is distinguishable from pollen autofluorescence, unstained pollen was analyzed as a control in each flow cytometry experiment (see Experimental procedures for additional details).
The distribution profile for an unstained pollen population showed a narrow spread of grains along the FITC-A axis with low autofluorescence intensity levels (Figure 2b,c). In contrast, the stained pollen population spread over a wide range of fluorescence intensities concentrated around two distinct peaks. This bimodal distribution has been observed in all analyses with healthy pollen samples, and indicates that the pollen population can be largely divided into 'low-ROS' and 'high-ROS' subpopulations; the fluorescence intensities of the former partially overlap with the autofluorescence of the unstained pollen, whereas the latter possess one-two orders of magnitude higher intensity (Figure 2e,f). A bimodal distribution was also observed for tomato pollen (Figure S7), suggesting that this population-scale pattern might be a general characteristic for pollen from many plant species.
Because positive DCF-staining depends on esterase activity and cellular generation of ROS, the intensity level is also an indicator of metabolic activity. Therefore, our results suggest that the low-ROS population is comprised of pollen that are either dormant or have low metabolic activity, while the high-ROS population represents cells with higher metabolic activities. Taken together, DCF-staining coupled with flow cytometry provides a highly sensitive and robust approach to evaluating the variability of metabolic states within a population of pollen.
To examine the germination capacity of pollen from lowand high-ROS subpopulations, we performed sorting experiments to isolate and germinate pollen from each subpopulation (Figure 3). Following pollen grain hydration for 15 min and H 2 DCFDA staining for 30 min, low and high gates were established in the FACS for sorting positive DCF-stained grains according to the bimodal distribution (Figure 3a,b). Pollen from each subpopulation was collected into separate tubes and plated for overnight germination in PGM (Figure 3c-e).
Out of a total of about 800 sorted grains for each population, nearly 10% of the high-ROS pollen produced tubes, whereas only 0.3% of the low-ROS pollen formed tubes (Figure 3d). A similar 1-35 proportion of pollen germination percentage between the low-and high-ROS was also obtained with tomato pollen (Figure S8). The fact that tube formation for the high-ROS subpopulation was nearly 35fold higher than for the low-ROS population supports the conclusion that under in vitro germination conditions, the subpopulation of pollen displaying strong metabolic activity will germinate and grow first, while a low activity subpopulation remains dormant.
To evaluate whether DCF-staining coupled with flow cytometry can be used to monitor ROS dynamics under different environmental conditions, Arabidopsis pollen was exposed to oxidative stress in a dose-dependent manner (Figure 4). To this end, pollen were subjected to increasing concentrations (1 and 10 mM) of the superoxide ion (O 2 À )generating herbicide paraquat (PQ) for 15 min in vitro, stained with H 2 DCFDA and subjected to flow cytometry. Because PQ generates O 2 À upon reduction by electron donors, which in the absence of photosynthesis predominately originates from cellular respiration (Miller et al., 2009), our focus was on the perturbing and analyzing the high metabolically active subpopulation (i.e. high-ROS population). As expected, an overall ROS increase of 1.3and 2.8-fold was obtained in proportion with the degree of oxidative stress resulting from treatments of 1 and 10 mM PQ concentrations, respectively (Figure 4a,b). Importantly, the stress treatments did not change the relative bimodal distribution of cells (Figure 4c), indicating that the PQ-triggered increases in ROS did not cause a precipitous increase in cell death in the short time (15 min) of the stress treatment. Nevertheless, over time (during 4 h in germination medium) the increase in ROS level in the 1 and 10 mM PQ was inversely correlated with a five-and 37-fold decrease in pollen germination, respectively (Figure S9). In a complementary experiment, pollen was treated with increasing concentrations of caffeic acid (CA), a potent antioxidant (Gulcin, 2006). In contrast to PQ, 50 and 100 lM CA reduced the average DCF fluorescence levels by 30% to nearly twofold, respectively (Figure 4d,e). Correspondingly, pollen treated with 50 and 100 lM CA resulted in a ~50 and ~64% increase in pollen tube formation, respectively (Figure S10), indicating that the capacity of pollen to germinate in vitro can be significantly improved by a moderate reduction in the oxidative level. On the other hand, treatment with higher concentrations of CA (> 200 lM) totally inhibited pollen tube formation, suggesting that there is a physiological range of ROS concentrations below or above which germination is compromised (Figure S10).
Increases in ROS production can occur in response to unfavorable growth conditions, including HS (Choudhury et al., 2017). Large increases in ROS can lead to cellular damage that can make pollen cells unhealthy or die (Rieu et al., 2017). To evaluate whether DCF-staining coupled with flow cytometry can be used to monitor ROS dynamics in response to a HS, we conducted experiments in which pollen were exposed in vitro to elevated temperatures.
Pollen grains were imbibed for 15 min in PGM at 22°C, then supplemented with H 2 DCFDA and exposed for 30 min to 35°C before a flow cytometry analysis (Figure 5a). Compared with a control sample kept at 22°C, the HS treatment increased the mean median DCF fluorescence intensity (MFI) of the high-ROS subpopulation by 24%, indicating that HS triggered an increase in ROS production (Figure 5b). This was associated with a 17% decrease in the overall number of viable DCF-stained grains (Figure 5c), and a 60% decrease in pollen germination potential (Figure 5e). The decrease in DCF-stained grains was specifically associated with the high-ROS subpopulation (~twofold decrease), whereas cell numbers in the low-ROS subpopulation actually increased (Figure 5d). © 2019 The Authors The Plant Journal © 2019 John Wiley & Sons Ltd, The Plant Journal, (2019), doi: 10.1111/tpj.14286
The apparent vulnerability of the high-ROS grains to HS was confirmed in a similar experiment in which pollen were stained following incubation at 22, 37 or 42°C (Figure S11). Similar to the experiment in Figure 5, the HS treatment reduced the number of high-ROS stained pollen by 40 and 83% (at 37°C or 42°C, respectively; Figure S11a), and decreased germination potentials (Figure S11b). Together, these results suggest that a HS-induced death is more likely to occur in metabolically active grains (i.e. high-ROS) rather than in low metabolism/dormant (i.e. low-ROS) pollen grains.
Pollen is the only free-living cell-sized tissue in flowering plants, making it uniquely amenable to flow cytometry and cell sorting without prior treatments such as protoplasting. Yet, the number of studies that have utilized flow cytometry with pollen is relatively small, using it mainly for evaluating nuclear DNA content and ploidy (Stehlik et al., 2007;Hirano and Hoshino, 2009;De Storme et al., 2013;Kron and Husband, 2015;Loureiro and Castro, 2015), which is important in the fields of plant breeding and population biology (Becker et al., 2003). Flow cytometry and sorting by FACS methods have also been used for the detection and evaluation of fluorescent-tagged transgenic pollen (Moon et al., 2011;Borges et al., 2012), and for purification of pollen and sperm cells for characterizing the pollen transcriptome (Becker et al., 2003;Pina et al., 2005;Borges et al., 2008). The potential of flow cytometry as a tool to determine viability and monitor morphological changes in sperm cells was suggested by Zhang et al. (1992), and was also demonstrated as a tool for comparing size and autofluorescence properties between pollen varieties (Tyrer, 1981). However, the practice of flow cytometry for studying functional aspects of pollen biology, including pollen fitness, has remained untapped.
Recently, Heidmann et al. (2016) described the potential of using impedance flow cytometry (IFC) for evaluating pollen quality by distinguishing between viable and heatinactivated pollen of several species. In IFC, cells are passed through a micro-fluidic chip and their dielectric properties (impedance) are measured in a range of several radio frequencies, which can be used to analyze cell size, shape and membrane properties (Heidmann et al., 2016). Although this appears to be a good non-invasive and label-free method for evaluating quality of pollen, correlation of IFC viability score with actual pollen germination capacity was not presented. In addition, IFC is limited compared with optical flow cytometry for studying cell biology and pollen physiology, as it cannot incorporate labeling techniques and cells cannot be sorted.
Here we used the advantage of optical flow cytometry coupled with H 2 DCFDA staining as a method for evaluating pollen fitness, and as a tool for studying pollen responses to stress conditions at the population level. First, we showed that DCF fluorescence is a reliable marker for scoring pollen viability in a wide range of plant species (Figures 1,S4 and S5). This approach was then successfully applied to flow cytometry using FACS, which easily distinguished pollen grains that appeared to be dead (i.e. cells with only background auto-fluorescence) and those with both high and low DCF fluorescence (Figure 2).
Fluorescence-activated cell sorting analyses of DCF-stained pollen grains provided two lines of evidence indicating a significant amount of physiological heterogeneity in a population of mature pollen grains. First, we consistently observed a bimodal distribution of pollen into low-ROS and high-ROS populations (Figure 2), correlating with a low and high capacity to germinate and grow tubes, respectively (Figures 3 and S8). This was true for pollen from Arabidopsis, tomato and N. benthamiana. Second, the DCF fluorescence intensities of the high-ROS population spanned about one order of magnitude, indicating extensive cell-to-cell variability in metabolic activities within the subpopulation of metabolically active pollen.
Potential underlying factors promoting non-genetic variation in pollen metabolic activity include asynchronies during pollen grain maturation, which can result in grains having significant differences in water content, starch and soluble carbohydrates levels (Carrizo Garcia et al., 2017). In addition, individual pollen grains can mature with differences in epigenetic modifications (Slotkin et al., 2009;Colaco et al., 2012) that might impact gene-expression networks and contribute to metabolic variations within a pollen population. Nevertheless, it is not yet clear if the bimodal distribution of pollen observed here is due to random variation, or if there is a developmental program that evolved to ensure the production of two distinct populations.
Pollen germination assays following FACS purification revealed that under optimal conditions the high-ROS pollen germinated with a frequency that was 35-fold higher than the low-ROS pollen (Figure 3c). This is consistent with the expectation that for pollen grains to germinate they must break dormancy and become metabolically active. Thus, with analogy to the 'oxidative window' concept in seed germination (Bailly et al., 2008), some ROS in pollen is necessary. In plants, there are many examples in which some ROS has been proven to be necessary for several biological processes, including developmental programs and response to stress (Mittler, 2017). However, further increases in pollen ROS production (e.g. in response to PQ; Figure 4) decreased the percentage of pollen germination. Interestingly, a moderate reduction in ROS using an antioxidant such as CA actually improved in vitro pollen germination (Figures 4,S8 and S9), suggesting that a subset of the high-ROS population is on the verge of oxidative distress even under the control conditions used in this study.
Heat stress-induced reactive oxygen species can be bad In many flowering plants, pollen is highly sensitive to temperature stresses imposed during microsporogenesis, pollen maturation and fertilization, with even a short period of high temperatures sometimes being fatal (Zinn et al., 2010;Firon et al., 2012). H 2 DCFDA staining coupled with flow cytometry indicates that HS can increase the mean fluorescence of the high-ROS pollen population (Figure 5b), which correlates with a decrease in pollen germination frequencies (Figures 5e and S11b). In addition, HS also reduced the overall number of stained pollen grains (Figures 5c and S11a). These observations are consistent with an expectation that oxidative stress can be detrimental and lead to cell death.
The unique insight provided here through flow cytometery analyses was the observation that mature pollen sort into two populations, high-ROS and low-ROS. While it is not yet clear why a flower generates two distinct populations of pollen grains, we offer a working model in Figure 6 that suggests a potential reproductive value for a plant to maintain a subset of pollen in a delayed dormant state as an 'insurance policy' for environmental variations. In this model, the pollen cells most vulnerable to a HS are those corresponding to the most metabolically active high-ROS population. Under stress, these metabolically active cells generate even more ROS and eventually die, while the pollen from low metabolic/dormant populations can survive. Thus, the low-ROS subpopulation might represent the 'back-up reserves' that can break dormancy at a later time and provide a second opportunity to complete fertilization under different environmental circumstances.
In terms of using optical flow cytometry, we are only beginning to scratch the surface of its potential for studying pollen biology. The major advantages of employing flow cytometry/FACS are the robustness of the technique, the time it saves and the large-scale multifaceted data obtained from each analysis. In this study, we established that FACS purification can be used to fractionate and physiologically analyze different pollen subpopulations. Nevertheless, additional optimization studies are still needed for improving germination rates of FACS-purified pollen (see Experimental procedures, and Figures S12 and S13). The advantage of using H 2 DCFDA compared with a histochemical dye such as MTT is that DCF is not cytotoxic, which means that stained pollen are viable and can be used for downstream applications. For example, a mixture of mutant pollen could be subjected to a HS treatment, and the surviving pollen isolated by FACS and used for manual pollination and breeding of new plant lines. We expect there to be many pollen applications for using H 2 DCFDA staining and other reporters coupled with flow cytometry/ FACS, including a future characterization of the HS response in pollen cells that sort into either high-ROS or low-ROS populations.
Arabidopsis (A. thaliana, Columbia-0 ecotype) and N. benthamiana plants were grown in growth chambers (23°C, 16 h of light, 8 h of dark and 70% humidity). Tomato plants (Solanum lycopersicum) were either grown in a growth chamber (cv. MicroTom; 24°C day temperature, 21°C night temperature, 16 h of light, 8 h of dark and 70% humidity) or in a greenhouse (cv. M82) at the Volcani Center, Beit-Dagan, Israel. Cucurbitaceae plants and basil (Ocimum basilicum) were grown in a greenhouse in Bar-Ilan University, Israel. Poppy plants (Papaver umbonatum) grew wild in Bar-Ilan University, Ramat-Gan, Israel.
Mature open flowers of each species were harvested on the day of the experiment, soaked in PGM [10% Sucfor Arabidopsis and N. benthamiana: 100 mM KCl, 100 mM CaCl 2 , 100 mM MgSO 4 , 1% boric acid and 10% sucrose (Boavida and McCormick, 2007); for tomato: 2 mM Ca(NO 3 ) 2 , 2 mM MgSO 4 , 1 mM KNO 3 , 2 mM boric acid and 10% sucrose (Firon et al., 2006)], vortexed and centrifuged for 3 min at 10 000 rpm. This 'standard prep' of pollen was regularly used for most pollen germination assays under control conditions and in response to different treatments, and for the different staining procedures.
Colorimetric viability staining procedures. A 2-h staining period was used with the following dyes: AS variations A and B (A: 20 ml 95% ethanol, 50 ml distilled water, 40 ml glycerol, 2 ml lactic acid, 20 mg malachite green, 100 mg acid fuchsin and 2 g phenol; B: 10 ml 95% ethanol, 54.5 ml distilled water, 25 ml glycerol, 0.5 ml orange G, 1 ml malachite green, 1 ml acid fuchsin and 4 ml glacial acetic acid); 1% MTT (3-4,5-dimethylithiazolyl-2-2,5diphenyl monotetrazolium bromide; Bio Basic, Markham, Canada) in 5% sucrose; and 1% TTC (2,3,5-triphenyl tetrazolium chloride; Bio Basic Canada) in 60% sucrose. All stains were supplemented in a 1:10 volume to the pollen. Pollen was visualized by a Nikon eclipse TE200 microscope.
For Arabidopsis pollen, flowers were vortexed in a small volume of germination medium containing 5% w/v sucrose [i.e. PGM (5% Suc); approximately 20 ml flowers in a 50-ml falcon tube with 10 ml PGM], which is the minimal sucrose concentration that allows germination (Boavida and McCormick, 2007). Using minimal germination conditions is important for reducing the risk of tube formation in some of the pollen by the time of the analysis, which could clog the flow cytometer. For tomato pollen, detached anther cones from open flowers were cut in half transverse and vortexed in a small volume of tomato PGM (5% Suc; Firon et al., 2006), while for N. benthamiana pollen intact flowers were vigorously vortexed in a 50-ml falcon filled with tomato PGM (5% Suc). Suspensions of pollen were subsequently filtrated through a folded miracloth (Merck Millipore, Darmstadt, Germany) to remove tissue debris.
For the FACS pollen-purification experiments (described in Figures 3, S8 and S12), all steps from harvesting the pollen, filtering and processing them in FACS to germinating the purified pollen subpopulations were carried out in a PGM (10% Suc, instead of 5% Suc) and referred to in the figure legends as 'FACS prep'. In the supplementary data, we also discuss the impact of the H 2 DCFDA staining (Figure S12) and the FACS-prep and FACS purification (Figure S13) on tube-formation potential of the pollen.
Care was taken to ensure that the pollen suspensions were not too dilute, determined visually by the suspension having a yellow appearance. If pollen suspensions were too dilute, pollen was centrifuged at 10 000 rpm for 3 min, and resuspended in a smaller volume of germination buffer. The low-ROS subpopulation includes mostly dormant but also dead pollen that do not stain (gray), as well as weakly stained pollen that are dormant with low metabolic activity. The high-ROS subpopulation includes pollen with a wide range of metabolic activity levels (purple). Exposure to stress such as high temperatures (yellow lightning icon) rapidly causes an oxidative stress due to accumulation of ROS beyond the optimal range, seen as a shift (indicated by a yellow arrow) in DCF intensity (red). Pollen undergoing oxidative stress (red zone) experience metabolic perturbation and accumulate damage that reduce pollen fitness in the short term and increase mortality rate in a longer termthe dashed arrow indicates that pollen dying as a result from the stress will not stain and appear on the left side (gray zone) of the distribution plot.
Pollen was allowed to hydrate in germination buffer for 15 min before H 2 DCFDA was added to a final concentration of 5 lM and incubation continued for 30 min prior to flow cytometry analysis. In the moderate HS experiment (35°C; Figure 5), H 2 DCFDA was added immediately following the 15 min hydration and incubated during treatment. In contrast, in experiments presented in Figures 4 and S10, H 2 DCFDA was added following the treatments. A small sample (~1 ml) of unstained pollen was retained to calibrate the DCF fluorescence threshold. It is essential to include an unstained pollen sample in every experiment to serve as a reference point for comparisons between different experiments. Different flow cytometry sessions may result in differences in absolute fluorescent intensities. This is due to voltage changes and laser calibration in the flow cytometer, which varies between usages. As another negative staining control for optimization of the flow cytometry protocol, pollen was heated for 30 min at 80°C in sucrose-free germination buffer before the addition of the H 2 DCFDA.
All experiments included a control treatment at optimal conditions (22°C). HS treatments included exposure to 35, 37 or 42°C for either 15 or 30 min. Other treatments included incubations with PQ (methyl viologen; Sigma Aldrich, St. Louis, MO, USA; 1 or 10 mM), CA (hydroxycinnamic acid; Alfa Aeser; 50-200 lM), tertbutyl hydroperoxide (40 mM) or aluminum [20 mM Al(NO 3 ) 3 ] as indicated.
Following incubations with H 2 DCFDA, samples were processed in a BD FACSARIA TM III (BD Biosciences, San Jose, CA, USA) using a standard configuration, 85-ll nozzle, a flow rate of 9 or less and the FITC laser-filter set (488 nm laser line, BP filter 530/30) compatible with DCF. Flow cytometer droplets containing single pollen grains (i.e. singlet pollen) were identified empirically by first gating pollen from non-pollen events based on the parameters of FSC-A and SSC-A, which measure size and granularity, respectively, and then applying a double discrimination gate using a further two parameters of forward scatter (height and width, i.e. FSC-H and FSC-W, measuring detector signal maximum and duration of signal, respectively). Fluorescence in the FITC-A channel of singlet pollen was then calibrated using the unstained sample as the autofluorescence threshold. Typically, 10 000 singlet pollen grains were analyzed per sample. When sorting pollen, 5000 pollen were sorted into 1 ml of PGM (10% Suc).
All pollen grains and tubes were counted from images captured using DIC at 40 9 magnification, using the ImageJ software (version 1.51).
© 2019 The Authors The Plant Journal © 2019 John Wiley & Sons Ltd, The Plant Journal, (2019), doi: 10.1111/tpj.14286