Methodological considerations in the analysis of guard cell starch metabolism

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Introduction

Stomata are microscopic pores on the leaf surface that regulate CO₂ uptake for photosynthesis and water loss through transpiration. Their aperture is controlled by the turgor pressure of the two surrounding guard cells (GCs), which respond dynamically to environmental cues (Zhang et al., 2024). Among these, blue light (BL) is a key signal that promotes stomatal opening, primarily by activating plasma membrane H⁺‐ATPases for potassium (K⁺) uptake, water influx, and subsequent GC swelling (Inoue & Kinoshita, 2017). In Arabidopsis thaliana, BL signaling not only activates ion transport but also triggers rapid degradation of starch in GC chloroplasts, releasing glucose, part of which is subsequently converted to malate (Horrer et al., 2016; Flutsch et al., 2020). These metabolites are essential for stomatal opening, as they provide counterions for K⁺ and carbon skeletons for energy production, while also contributing to osmotic adjustment (Santelia & Lawson, 2016; Daloso et al., 2017). Arabidopsis mutants unable to degrade starch in GCs show a strongly impaired stomatal opening response to BL due to metabolic insufficiency (Horrer et al., 2016). Through this coordinated activation of proton pumping, ion transport, and starch degradation, BL efficiently couples energy availability with gas exchange and photosynthetic activity (Lawson & Matthews, 2020).

This model was challenged by a recent New Phytologist article by Bahadar et al. (2025). Using metabolic and stomatal analyses in Arabidopsis, cowpea (Vigna unguiculata), and tobacco (Nicotiana tabacum L.), the authors reported that starch in GCs was not degraded during BL‐induced stomatal opening in any of the species examined.

In Arabidopsis, GC starch degradation in response to stimuli that induce stomatal opening (e.g. BL, dark‐to‐light transition, exposure to low CO₂) has been consistently observed in different laboratories (Horrer et al., 2016; Flutsch et al., 2020; Li et al., 2020; Han et al., 2022; Shi et al., 2024; Gurrieri et al., 2025; Silva‐Alvim et al., 2026). Furthermore, starch degradation in response to light has also been observed in other species, such as lettuce (Lactuca sativa), apple (Malus domestica), or broad bean (Vicia faba) (Dang et al., 2024; Liu et al., 2025, 2026; Zhang et al., 2025). We therefore conducted a series of control experiments to investigate the basis of these discrepancies and demonstrate that they arise from differences in light growth regimes, particularly the photoperiod, as well as the experimental procedures – factors that may undermine the conclusions drawn by Bahadar et al. (2025).

Results

GC starch levels are extremely low at the end of the night in short‐day‐grown Arabidopsis

In Bahadar et al. (2025), Arabidopsis plants were grown under short‐day conditions (SD; 8 h : 16 h, light : dark), which differs markedly from the growth regimes used in most studies of GC starch metabolism, where plants were typically grown under a neutral‐day (ND) photoperiod (12 h : 12 h, light : dark). Photoperiod strongly influences leaf starch metabolism. Under SD conditions, Arabidopsis allocates a higher proportion of its photoassimilates to starch during the light and degrades it at a slower rate throughout the long night compared with plants grown under ND conditions (Lu et al., 2005; Sulpice et al., 2014). Such changes in whole‐leaf carbon metabolism may also affect GC starch dynamics and therefore influence the extent and timing of starch degradation during stomatal opening. Using a propidium iodide‐based staining method (Flütsch et al., 2018), we compared starch levels in GCs of Arabidopsis plants grown under SD or ND. As previously reported (Horrer et al., 2016), GCs of plants grown under ND had substantial amounts of starch at the end of the night (EoN), which were degraded in response to 1 h BL illumination (Fig. 1a). This BL‐induced starch degradation coincided with stomatal opening (Fig. 1b), consistent with previous observations linking GC starch turnover to BL‐driven stomatal responses. In contrast to ND, GCs of SD‐grown plants contained markedly lower starch levels at the EoN and accumulated starch after 1 h of BL illumination (Fig. 1c). BL treatment induced stomatal opening, but the response was weaker compared to plants grown under ND (Fig. 1d), as previously reported (Aoki et al., 2019; Han et al., 2022), and showed no clear association with GC starch turnover.

Fig. 1.

Guard cell (GC) starch dynamics and stomatal aperture in Arabidopsis in response to blue light (BL). (a–h) GC starch dynamics and stomatal aperture in response to 1 h BL illumination (1 h BL; blue) in Arabidopsis peels isolated at the End of the Night (EoN; grey) from plants grown under neutral day (ND; 12 h : 12 h, light : dark) or short day (SD; 8 h : 16 h, light : dark) conditions and incubated in MES‐BTP (a–d) or MES‐NaOH buffer (e–h). (i–p) GC starch dynamics and stomatal aperture in response to 1 h BL illumination (1 h BL; blue) in Arabidopsis peels isolated at the EoN (grey) from plants grown under ND (12 h : 12 h, light : dark) or short day (SD; 8 h : 16 h, light : dark) conditions and preincubated in mannitol 0.5 M for 30 min in darkness before transfer to MES‐BTP (i–l) or MES‐NaOH buffer (m–p). Data represent n independent stomata per time point and condition. Representative stomata per time point and condition are presented. Bars, 10 μm. Significant differences between EoN and 1 h BL were assessed based on a Wilcoxon test (*P < 0.05, **P < 0.01, ***P < 0.001, ns, not significant).

These results could explain why Bahadar et al. (2025) did not observe GC starch degradation during BL‐induced stomatal opening in Arabidopsis. Under long‐night conditions, GCs begin the day with extremely low starch amounts, and early‐morning BL‐driven carbon assimilation may be directed toward starch synthesis rather than degradation, potentially decoupling stomatal opening from starch turnover. This observation highlights the plasticity of GC starch metabolism and suggests that the relationship between starch turnover and stomatal opening is influenced by the growth conditions and by the initial metabolic state of GCs, particularly their starch content at dawn.

High mannitol concentrations reduce GC viability and inhibit GC starch degradation and stomatal opening

Bahadar et al. (2025) kept blended peels in a hypertonic solution (0.5 M mannitol) under dark conditions to prevent stomatal opening until sufficient material was obtained. Collected peels were then washed and transferred to an opening buffer (5 mM MES‐NaOH, pH 6.5, 5 mM KCl and 50 μM CaCl₂) for stomatal aperture and metabolite measurements. Because the prolonged exposure to a hypertonic solution could potentially affect GC functionality, we assessed cell viability in epidermal peels at the end of the BL treatment using fluorescein diacetate (FDA) and SYTOX Orange dual staining. FDA stains metabolically active cells (Lawrence et al., 2018; Rasouli et al., 2020; Brett & Ng, 2022; Fleetwood et al., 2024), whereas SYTOX Orange stains nucleic acids in cells with compromised plasma membranes (Truernit & Haseloff, 2008), thereby enabling the identification of viable and metabolically competent GCs.

The presence of MES‐NaOH in the opening buffer reduced GC viability by c. 20% compared with cells incubated in a buffer containing the less alkaline agent MES‐BTP (Fig. 2a–d, Mock). Despite this reduction in viability, GCs largely retained their ability to respond to BL (Fig. 1e–h, Mock). By contrast, the additional preincubation with 0.5 M mannitol reduced viability to only c. 15–35% (Fig. 2a–d, mannitol) and severely impaired BL‐induced starch turnover and stomatal opening. Following preincubation in the hypertonic solution, GC starch content remained unchanged after 1 h BL illumination, and stomata failed to open, irrespective of the growth photoperiod or the buffering agent used (Fig. 1i–p). The close correlation between these functional impairments and the reduction in GC viability supports the conclusion that mannitol pretreatment negatively affects GC viability and physiological competence. In line with our findings, Bahadar et al. (2025) cite a study in V. faba demonstrating that exposure to 0.5 M mannitol caused near‐complete GC plasmolysis within 10 min (Tallman & Zeiger, 1988). Nevertheless, samples subjected to mannitol pretreatment were used for metabolomic analyses in Bahadar et al. (2025). Given the potential effects of mannitol and the associated reduction in GC viability observed in our study, these experimental conditions should be considered when interpreting the resulting metabolomic data.

Fig. 2.

Guard cell (GC) viability in epidermal peels and leaf starch content. (a–d) Percentage of intact GCs in peels from plants grown under neutral day (ND; 12 h : 12 h, light : dark; a, b) or short day (SD; 8 h : 16 h, light : dark; c, d) conditions. For mock treatment, isolated peels were incubated for 30 min in darkness in MES–BTP (a, b) or MES‐NaOH buffer (c, d) before 1 h blue light (BL) treatment. For mannitol treatment, isolated peels were preincubated in 0.5 M mannitol for 30 min in darkness, washed with 1 l of water, and transferred to the indicated buffer before 1 h BL treatment (1 h BL). Samples for the live–dead staining were collected after BL treatment and stained immediately with fluorescein diacetate (FDA) or SYTOX Orange. Representative images show intact GCs in green and nonintact GCs in red. Data represent n independent stomata per condition. Significant differences between mock and mannitol treatments were assessed using a Wilcoxon test (*P < 0.05, ***P < 0.001; ns, not significant). (e) Starch content in whole leaves of plants grown under ND (12 h : 12 h, light : dark) or short day (SD; 8 h : 16 h, light : dark) was quantified at the EoN using an enzymatic assay. Significant differences in leaf starch content between ND and SD were determined by Wilcoxon test.

Our results highlight the importance of assessing both GC viability and functionality for the proper interpretation of physiological responses in isolated epidermal peel preparations.

Starch in GCs cannot be reliably measured enzymatically

In Bahadar et al. (2025), starch was quantified using an enzymatic assay that measures total starch in epidermal peels rather than starch concentrations within GCs. Pavement cells have been reported to contain small amounts of starch (Streb & Zeeman, 2012) and therefore contribute to the total starch measured in epidermal peel extracts. Furthermore, although blended peels are subjected to several washing steps before metabolite extraction, potential contamination from residual mesophyll tissue cannot be excluded. Consistent with this concern, we show that GCs of Arabidopsis contain extremely low starch levels after 16 h of darkness (Fig. 1c), while Bahadar et al. (2025) report unexpectedly high end‐of‐night starch levels (c. 8 μmol g⁻¹ FW). These values are comparable to starch concentrations typically measured at the EoN in whole leaves of plants grown under either ND or SD (Fig. 2e; see also Lu et al., 2005; Sulpice et al., 2014) and therefore are unlikely to reflect starch amounts in GCs. Bahadar et al. (2025) noted in their Discussion that their analysis ‘did not refer to the concentration of starch within GCs’, but their title nevertheless states that BL‐induced stomatal opening is not associated with starch breakdown in GCs. This conclusion is not justified by the data presented.

Quantification of starch specifically within GCs cannot be reliably achieved using bulk enzymatic assays. Such approaches lack the spatial resolution and the sensitivity required for cell‐type‐specific starch analyses and therefore fail to distinguish GC‐specific starch pools, even when epidermal peels are enriched or mechanically blended. GCs account for no more than c. 44% of cells in epidermal peel fractions (Tolleter et al., 2024) and have a cell volume that is approximately two orders of magnitude smaller than that of mesophyll cells (Chen et al., 2012). Furthermore, GCs contain at most five chloroplasts per cell, in contrast to 40–100 chloroplasts in mesophyll cells (Liu et al., 2021). During mechanical blending, mesophyll cells rupture and release insoluble starch granules into the homogenate. These granules are not efficiently removed by washing, leading to substantial mesophyll‐derived starch contamination even in the absence of intact mesophyll cells.

Alternative image‐based approaches, such as confocal laser scanning microscopy combined with propidium iodide (Flütsch et al., 2018) or safranin O staining (Liu et al., 2021), are more suitable for GC‐specific starch quantification because they enable direct visualization of starch granules within individual GCs. However, a potential limitation of these methods is that the quantification of very small starch granules from 2D images may not be fully linear at low starch levels. Although we use z‐stacks to identify the optimal focal plane, the boundaries of small granules cannot always be precisely resolved in two dimensions. While 3D reconstruction would likely improve quantification accuracy, it is currently impractical for the large numbers of granules and cells required for robust quantitative analyses. Nevertheless, imaging‐based approaches provide direct, GC‐specific measurements and have proven highly reproducible for monitoring relative changes in GC starch dynamics. We urge the stomatal research community to rely on image‐based approaches, rather than enzymatic bulk measurements, when examining starch metabolism in GCs, as methodological choices critically determine the validity of conclusions about GC starch metabolism and its role in stomatal function.

Conclusions

In conclusion, the discrepancy reported by Bahadar et al. (2025) can be explained by their choice of plant growth conditions and the experimental design. Growth conditions such as photoperiod strongly influence diel starch turnover and must be carefully considered when comparing GC starch metabolism across studies. In addition, preincubation in high mannitol concentrations can compromise GC viability and stomatal responsiveness, thereby affecting the interpretation of the metabolic measurements. Finally, bulk enzymatic assays lack the spatial precision required to quantify starch specifically within GCs and are therefore unsuitable for drawing GC‐specific conclusions about starch metabolic dynamics. Together, these factors underscore the need to routinely evaluate GC viability and functionality in isolated epidermal peel preparations and to employ image‐based approaches for GC starch quantification under physiologically appropriate conditions, thereby ensuring accurate and reproducible measurements of GC starch turnover and its role in stomatal function.

Materials and Methods

Plant material and growth conditions

Four‐week‐old Arabidopsis thaliana (L.) Heynh. Columbia (Col‐0) wild‐type plants were used in this study. Seeds were sown in soil (Profi Substrat, Einheits Erde, Classic), stratified at 4°C for 3 d in darkness, and germinated in a growth chamber for 7 d, after which seedlings of similar size were transplanted into single pots. All plants were grown in climate‐controlled chambers at 21°C : 19°C, day : night temperature, 45% : 55%, day : night relative humidity, and an irradiance of 150 μmol m⁻² s⁻¹. Plants were grown under either an ND (12 h : 12 h) or SD (8 h : 16 h) light : dark photoperiod. Plants were watered twice per week.

GC‐enriched epidermal fragments isolation

Fully expanded leaves from six Arabidopsis plants were collected for each biological replicate. Central veins were removed before tissue disruption. Leaves were blended using a Philips ProBlend 6 blender (Philips, Amsterdam, the Netherlands) in 500 ml ice‐cold water two times for 30 s each. The homogenate was filtered through a 200 μm mesh (Sefar AG, Heiden, Switzerland) to collect epidermal peels, which were then floated in 20 ml stomatal buffer and incubated in the dark for 30 min before further treatments. Two stomatal buffers were used: the MES‐BTP buffer (5 mM MES‐bistrispropane, 50 mM KCl, 0.1 mM CaCl₂, pH 6.5; Horrer et al., 2016) or the MES‐NaOH buffer (5 mM MES‐NaOH, pH 6.5, 5 mM KCl and 50 μM CaCl₂; Bahadar et al., 2025). Alternatively, peels were preincubated in 0.5 M mannitol hypertonic solution for 30 min, after which they were washed with 1 l water and transferred to the indicated buffer. Peels were then subjected to BL treatment for 1 h with 90 μmol m⁻² s−1 BL. Samples were collected at the EoN and after 1 h BL. Epidermal fragments were either fixed in 50% (v/v) methanol and 10% (v/v) acetic acid for starch quantification or mounted directly on microscopy slides for stomatal aperture analysis.

Stomatal aperture measurements under BL

Epidermal fragments were transferred from stomatal buffer to microscopic slides at the indicated time points. Images were taken using an inverted microscope (Nikon Eclipse TS100; Nikon Corporation, Tokyo, Japan) at ×40 magnification. Stomatal opening was quantified based on both area and aperture using StomataPy (https://github.com/Alias‐z/StomataPy). Per independent experiment, on average 134 stomata were measured at each time point. The experiment was repeated twice per each condition.

GC starch staining and quantification

GC starch granules in epidermal fragments were visualized using pseudo‐Schiff–propidium iodide (PS‐PI) staining following the protocol of Flütsch et al. (2018). In brief, fixed epidermal fragments were oxidized with 1% periodic acid, rinsed with water, and incubated in Schiff reagent (100 mM sodium metabisulfite, 0.15 N HCl) supplemented with propidium iodide (0.1 mg ml⁻¹). After destaining in water, samples were incubated in chloral hydrate solution for 12 h in the dark and subsequently mounted in Hoyer's solution. Imaging was performed using a confocal microscope with excitation at 488 nm and emission detection between 610 and 640 nm. Starch area relative to GC area was quantified with https://github.com/Alias‐z/StomataPy. Two independent experiments were performed, each sampling 40 GCs from four biological replicates.

Leaf starch measurements

Leaf starch content was quantified enzymatically following Smith & Zeeman (2006). Entire rosettes were harvested at the EoN and homogenized in 0.7 M perchloric acid. Insoluble material was pelleted, washed once with MQ water and twice with 80% (v/v) ethanol, and resuspended in water. Starch was gelified by boiling at 95°C for 15 min and digested using 1 U of α‐amylase (Megazyme Ltd., Bray, Ireland) and 1.4 U amyloglucosidase (Roche Holding AG, Basel, Switzerland) for 2 h at 37°C. Glucose concentrations were determined via a coupled hexokinase/glucose‐6‐phosphate dehydrogenase reaction (Roche), converting NAD to NADH, which was quantified spectrophotometrically at 340 nm (Tecan Group Ltd., Männedorf, Switzerland), based on standard curves of known glucose concentrations. Six biological replicates were analyzed in total.

Cell viability staining

At the end of the BL treatment, epidermal fragments were subjected to live–dead staining using FDA (F7378; MERCK, Burllington, MA, USA) and SYTOX™ Orange (S11368; Thermo Fisher Scientific Inc., Waltham, MA, USA). Epidermal fragments were incubated in a staining solution containing the same stomatal buffer used during the experiment, supplemented with 0.5 mM FDA and 0.25 μM SYTOX Orange. Samples were then mounted on slides and imaged within 30 min using a confocal microscope with a ×40 air‐immersion objective. Three channels were used: (1) FDA fluorescence (excitation 488 nm, emission 500–525 nm, green) to identify viable cells; (2) SYTOX Orange fluorescence (excitation 568 nm, emission 580–610 nm, orange) to mark nonviable cells; and (3) transmitted light (bright field, grey). Images from the three channels were overlaid in ImageJ to generate the composite panel shown in Fig. 2(a–d). To ensure robust quantification, only cells displaying a clear fluorescence signal (either FDA or SYTOX Orange) were included in the analysis, thereby excluding poorly stained or out‐of‐focus cells. The percentage of live cells (GC viability) was calculated as the number of FDA‐stained cells divided by the total number of stained cells per epidermal fragment. Each epidermal fragment was considered one biological replicate, with at least 10 fragments analyzed per treatment. The experiment was repeated twice per each condition.

Competing interests

None declared.

Author contributions

DS conceptualized the study. KF performed GC starch quantification experiments and stomatal aperture measurements. CP performed enzymatic starch assays. TD performed live–dead staining experiments. HZ acquired stomatal opening images, conducted data analysis, and prepared the final figures. LP performed starch quantification experiments. DS and KF wrote the manuscript with approval from all authors. Artificial intelligence (AI) tools were used solely to improve the clarity and grammar of the manuscript and to assist with R coding; all scientific content, data analysis, interpretation, and conclusions were developed and verified by the authors.

Data availability

The data generated in this study are openly available in the repository of ETH Library at doi: 10.3929/ethz-c-000802346. The R codes used for statistical analyses are provided at https://github.com/Alias‐z/GCStarchMethod.