Hydrogen peroxide sensor HyPer7 illuminates tissue-specific plastid redox dynamics

Abstract

The visualization of photosynthesis-derived reactive oxygen species has been experimentally limited to pH-sensitive probes, unspecific redox dyes, and whole-plant phenotyping. Recent emergence of probes that circumvent these limitations permits advanced experimental approaches to investigate in situ plastid redox properties. Despite growing evidence of heterogeneity in photosynthetic plastids, investigations have not addressed the potential for spatial variation in redox and/or reactive oxygen dynamics. To study the dynamics of H₂O₂ in distinct plastid types, we targeted the pH-insensitive, highly specific probe HyPer7 to the plastid stroma in Arabidopsis (Arabidopsis thaliana). Using HyPer7 and glutathione redox potential (E_GSH) probe for redox-active green fluorescent protein 2 genetically fused to the redox enzyme human glutaredoxin-1 with live cell imaging and optical dissection of cell types, we report heterogeneities in H₂O₂ accumulation and redox buffering within distinct epidermal plastids in response to excess light and hormone application. Our observations suggest that plastid types can be differentiated by their physiological redox features. These data underscore the variation in photosynthetic plastid redox dynamics and demonstrate the need for cell-type-specific observations in future plastid phenotyping.

Dynamics of peroxide can be resolved by incorporating the pH-insensitive probe HyPer7 to plastid stroma, revealing heterogeneities in peroxide accumulation and distinct epidermal plastid types.

Introduction

Plastids are modular organelles that adopt specialized functions depending on their developmental stage, tissue specificity, and environmental conditions. Early observations suggested that epidermal pavement cells were devoid of chloroplasts, a notion that has been perpetuated until relatively recently (Barton et al. 2018). Growing evidence suggests that photosynthetic plastids may be more heterogeneous than once appreciated (Beltrán et al. 2018; Mackenzie and Mullineaux 2022). In fact, observed variation in photosynthetic plastid morphologies, proteomes, and inter-organellar interactions may reflect specializations in metabolism and plastid-to-nucleus retrograde signaling. For example, the phosphoenol pyruvate translocator, CAB UNDEREXPRESSED 1 (CUE1), exhibits tissue-specific protein accumulation in sensory plastids of Arabidopsis (Arabidopsis thaliana) inflorescence stems (Beltrán et al. 2018). cue1 mutants display phenotypes in retrograde metabolite accumulation, small RNA turnover, and DNA methylation, suggesting a vital role for sensory plastids during plastid-to-nucleus retrograde signaling (Shen et al. 2009; Fang et al. 2019). Additionally, the plastid-targeted aminotransferase, ALD1, only requires expression in epidermal pavement cells to induce defense metabolite production and systemic acquired resistance to Pseudomonas syringae (Jiang et al. 2021). Beyond proteomic differences, physiological studies distinguish epidermal, vascular, and mesophyll plastids by nonphotochemical quenching (NPQ) dynamics (Omasa et al. 2009; Gorecka et al. 2014). Given increasing evidence of chloroplast heterogeneity, a more detailed study of the signaling and physiology of these photosynthetic organelles will require the use of genetically encoded biosensors that allow optical dissection of tissues and cell types.

An important component for plastid-to-nucleus retrograde signaling is the photosynthesis-derived reactive oxygen species (ROS), H₂O₂. H₂O₂ has a relatively long half-life and is thought to disseminate throughout the cell from its origin via aquaporins, making it an attractive oxidative stress messenger. Current knowledge of plastid H₂O₂ dynamics has been experimentally limited to nonspecific redox dyes, pH-sensitive genetically encoded probes, and whole-leaf or seedling phenotyping. Genetically encoded redox biosensors with high specificity and limited pH sensitivity represent a gold standard for investigations of cellular redox dynamics. Yet, despite the growing wealth of probes of this type (Müller-Schüssele et al. 2021), investigations have not addressed the potential for spatial variation in plastid redox or ROS dynamics by cell type or environmental stimulus.

Here, we utilized the pH-insensitive, genetically encoded H₂O₂ reporter, HyPer7, to monitor stromal H₂O₂ dynamics in A. thaliana (Pak et al. 2020). HyPer7 consists of a cyclically permutated yellow fluorescent protein (YFP) with N- and C-terminal OxyR-RD domain derived from Neisseria meningitidis. Following oxidation by H₂O₂, HyPer7 forms an intramolecular disulfide bridge that alters the excitation spectra, allowing for ratiometric imaging at its 2 relative maxima (400 nm maximal when reduced and 500 nm maximal when oxidized). In our hands, HyPer7 showed a remarkably dynamic range that could report photosynthesis-dependent H₂O₂ production in response to excess imaging light. Using HyPer7, we detected cell-type-specific differences in H₂O₂ production. Interestingly, pavement cell “sensory” plastids showed the highest production of H₂O₂ in both unstressed and excess-light stressed plastids that we assayed. Moreover, we characterized the in vivo reduction kinetics of HyPer7, demonstrating that probe reduction can occur on the order of minutes in plastids. Lastly, we utilized HyPer7 to compare H₂O₂ production in guard cell and sensory plastids in response to abscisic acid (ABA). Observed stromal oxidation of HyPer7 independent of light and photosynthetic electron transport suggested that guard cell chloroplasts are not a primary site of H₂O₂ production in response to ABA. Together, our data underscore the heterogeneity of photosynthetic plastids in vascular plants and the value of cell-type-specific observations, presenting an enhanced approach to the study of H₂O₂ dynamics in plants.

Results

SSU:HyPer7 monitors stromal peroxide dynamics in vivo

To target HyPer7 to the plastid stroma, we fused the A. thaliana small subunit of RuBisCO targeting peptide to the N-terminus of HyPer7 and expressed the construct under a Ubiquitin10 promoter. We observed constitutive expression throughout leaf cell types (Fig. 1, A to C) with fluorescence that could clearly be separated from chlorophyll autofluorescence (Supplemental Fig. S1). Our construct showed stable expression throughout the development with clear colocalization of HyPer7 with chlorophyll autofluorescence (Fig. 1, D to G). Additionally, we observed signal within the stroma-filled plastid stromules, which further confirmed stromal targeting (Fig. 1, H to K). With an automated imageJ macro (Outlined Supplemental Fig. S2), we attempted to estimate the dynamic range of HyPer7 via an exogenous application of H₂O₂ and dithiothreitol (DTT) to be at least 7-fold in abaxial cotyledon sensory plastids under our imaging conditions (Fig. 2D). Additionally, we monitored the HyPer7 oxidation status in sensory plastids of untreated plants and found that the stroma was largely reduced. However, 488/405 ratios of untreated seedlings exceeded DTT-treated seedlings, suggesting a basal level of peroxide production (Fig. 2D). This observation was consistent with previous reports in Nicotiana benthamiana with the probe HyPer2 (Exposito-Rodriguez et al. 2017). It has been noted that DTT is inefficient at reducing oxidized HyPer7 in Arabidopsis cytosol (Ugalde et al. 2021b), although it accelerates the reduction of HyPer7 in yeast (Saccharomyces cerevisiae) cytosol following H₂O₂ application (Kritsiligkou et al. 2021); during pilot experiments (data not included), we noted an inconsistent reduction of HyPer7 between independent samples with incubation times of <30 min and without vacuum infiltration of the mesophyll air space. With this incubation timeline, we may have achieved an indirect reduction of HyPer7 through DTT-mediated reduction of glutathione and/or enzymatic disulfide scavengers such as thioredoxins (Trxs) that are abundant in plastids (Meng et al. 2010). We attempted to measure the excitation spectra of DTT-treated seedlings to confirm complete reduction; however due to the lower quantum yield of HyPer7 in its reduced state (Pak et al. 2020), we were not confident in the signal-to-noise ratio (data not included). In light of these inherent uncertainties of reduction by DTT and the large dynamic range, we present data as either absolute 488/405 ratios or relative ratios (R-R′), calculated from the average initial 488/405 ratio during time-course experiments, rather than percent or degree of oxidation.

Figure 1.

Expression and targeting of SSU:HyPer7. A–C) Stable expression of UBQ10:SSU:HyPer7 in an A. thaliana (Col-0) true leaf excited by its 2 relative maxima, 488 and 405 nm and collected between 505 and 550 nm. Whole-leaf scanning was performed on a Zeiss LSM-510 Meta fitted with a 10× Epiplan-Neoflaur objective using the tile scan feature in Zen software. D to G) HyPer7 (collected between 505 and 550 nm), excited with a 488 nm Argon Laser and imaged with a 40× c-Plan-Apochromat objective colocalizes with chlorophyll autofluorescence (650 nm LP) in abaxial pavement cells of 8-d-old cotyledons. H to K) Max intensity projection Z-stack shows HyPer7 (Excitation, 488 nm; emission, 505 to 550 nm) localizing within stromules (white arrow). White box in H) denotes the magnified J and K). Z-stack was filtered with a 1-pixel x, y, z radius median filter prior to max intensity projection.

Figure 2.

HyPer7 reports light-dependent stromal H₂O₂ at large dynamic range. Ratiometric HyPer7 micrographs cropped using segmented plastid masks treated with either 10 mM H₂O₂ A), Untreated B), or 10 mM DTT C) for 30 min prior to imaging. D) Estimation of HyPer7 dynamic range following 30-min incubation with either 10 mM H₂O₂ or 10 mM DTT. White triangle denotes the per-plastid mean from multiple individual plants (N > 120 plastids per treatment). Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. E) Mean HyPer7 ratio, per plastid, ±1 Se from abaxial pavement cell sensory plastids imaged within 2 h of subjective dawn at indicated age and tissue. Line indicates generalized additive model ±Se fitted to each developmental stage. N > 250 plastids at each time point derived from 6 to 8 different leaves or cotyledons. Data for 2-wk-old true leaf oxidation curves were taken from the same experiment as presented in G). F) HyPer7 oxidation status relative to the first frame average (R-R′) at end of imaging time course. Letters indicate significantly different groups (Tukey-HSD test, P < 0.01). Sample means are indicated with white triangles. Data were taken from the same experiment as presented in E). Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. G) Per plastid, mean 405 or 488 channel intensities ±1 Se over the imaging time course from first and second true leaves of 2-wk-old seedlings presented in E). HyPer7 oxidation in guard cell plastids before H) and after varied 488 nm laser output using ROI + bleaching feature on LSM510. Representative micrographs of HyPer7 488/405 ratios prior to I) and post J) excitation of guard cell chloroplasts with varying laser intensities. Asterisks overlaid on micrographs in J) indicate percent laser output used for 50 iterations of excitation with 488 nm laser at each ROI; ****10%, ***5%, **0.5%, and *0%. Imaging in H), I), and J) was performed with settings used to minimize laser output and exposure time by opening the confocal pinhole to 2.2 AU on an LSM510-Meta and collecting fluorescence at 505 to 550 nm using a 20×/0.8NA Plan-Apochromat objective at 3× digital zoom, 488 nm at 0.2% output, and 405 at 2.5% output with a single average (exposure time 1.9 s). H) Quantification of HyPer7 488/405 ratio relative to the pretreatment average. ***P < 0.001, ****P <<< 0.001, Wilcoxon ranked sums test, n > 12 plastid foci derived from 4 guard cell pairs on fully expanded true leaves collected at bolting. Data are relative (R-R′) to the average preincubation ratio. White triangle denotes sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. K) HyPer7 488/405 ratio in abaxial sensory plastids from 8-d-old seedling cotyledons relative (R-R′) to pretreatment sample average before or after 10-min incubation with brightfield halogen lamp at varying output. **P < 0.01 Wilcoxon ranked sums test, n > 100 plastids derived from 7 independent leaves. White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. Abbreviations: ROI, region of interest; AU, arbitrary units; DTT, dithiothreitol.

A recent report demonstrated that cytosolic HyPer7 is sufficiently sensitive to report photosynthesis-derived H₂O₂ in Arabidopsis cytosol (Ugalde et al. 2021b). This led us to ask whether we could apply time-course imaging as an excess-light treatment to study stromal H₂O₂ production in plastids. Consistent with previous reports, we detected immediate oxidation of HyPer7 following illumination with confocal lasers (Fig. 2E, Movie 1). This outcome was a result of increased 488 channel intensity with a concomitant decrease in 405 channel intensity, aligning with the published dynamics of the probe (Fig. 2G). The oxidation was nonlinear over the time course, slowing at ∼30 s following illumination. We then compared oxidation profiles of Arabidopsis seedlings at various developmental stages (Fig. 2E), with significant differences in the relative ratios at the conclusion of imaging time courses depending on developmental stages (Fig. 2F).

To separate high light-treatment effects from observations with confocal lasers, we used 2 separate approaches. First, we noted that guard cell plastids showed the highest expression of HyPer7 (Supplemental Fig. S4); this allowed us to minimize our imaging laser output (488 nm at 99.8% AOTF attenuation and 405 nm at 97.5% acusto optic tunable filter (AOTF) attenuation) and exposure time by limiting imaging to a single average (1.9 s total scan time). We then excited whole guard cell pairs with 50 iterations of a 488 nm laser at various power outputs ranging from 0% to 10% (Fig. 2, H to J). As expected, we observed oxidation of HyPer7 at varying magnitudes that mirrored the increased power output. Secondly, we used an approach similar to Exposito-Rodriguez et al. (2017) that applies a brightfield lamp to expose seedling cotyledons to either 100 or 1,000 µE light for 10 min. Although we observed a statistically significant increase after the 1,000 µE treatment, the magnitude of change was small (Fig. 2K). This modest effect was likely due to the absorption and refraction of light through the leaf lamina to reach the focal plane—whereas confocal lasers directly excite the plastids in view. We noted the difficulty in maintaining the exact focal plane before/after treatment, which often led to the inclusion of different populations of plastids in the image analysis and contributed to noisier data. In light of these observations, we opted to use rapid, time-course confocal imaging as our experimental assay.

To assay the specificity of HyPer7 to H₂O₂, we applied the H₂O₂ scavenging antioxidant L-ascorbate (AsA) to seedlings prior to imaging. The seedlings showed significantly lower stromal oxidation compared to an NaCl osmotic control (Fig. 3, A to C). Additionally, AsA-treated plants showed a slower oxidation curve following excess-light imaging (Fig. 3D). We used the photosynthesis inhibitor, DCMU, to block the flow of electrons from PSII, creating a burst in singlet oxygen with a concomitant attenuation in H₂O₂ production at PSI, a phenomenon that has been well documented in multiple systems (Fufezan et al. 2002; Slesak et al. 2003; Ugalde et al. 2021b). As expected, pretreatment of seedlings with DCMU resulted in near-complete attenuation of oxidation by HyPer7, despite creating a more oxidizing stroma (Fig. 3, E and F). Together, these results suggest that HyPer7 remains specific to H₂O₂ in the plastid stroma.

Figure 3.

HyPer7 remains specific to H₂O₂, in vivo. A and B) Representative ratiometric images at the outset and end of imaging time course for seedlings pretreated with either 1 mM l-ascorbate A) or 1 mM NaCl B). Scale bar indicated in A) applies to all images as they were acquired with identical settings. C) Quantification of stromal HyPer7 oxidation in seedlings pretreated with either 1 mM AsA or 1 mM NaCl (osmotic control) prior to imaging stress ****P <<< 0.0001 (Wilcoxon ranked sum test). White triangle indicates per-plastid mean. White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. D) Mean HyPer7 ratio, per plastid, ±1 Se relative to the first frame from abaxial pavement cell sensory plastids over an imaging time course. Seedlings were vacuum infiltrated with a syringe and incubated with either 1 mM AsA or 1 mM NaCl (at pH = 5.7) as osmotic control for 30 min prior to imaging. N > 97 plastids coming from multiple independent plants imaged within 2 h of subjective dawn. Data are from the same experiment presented in C). E) Quantification of stromal HyPer7 oxidation in seedlings pretreated with either 20 µM DCMU or mock (DMSO) prior to imaging stress (N > 200 plastids, **P < 0.01, Wilcoxon ranked sum test). White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. F) Mean HyPer7 ratio, per plastid, ±1 Se relative to the first frame from abaxial pavement cell sensory plastids over an imaging time course. Seedlings were pretreated with either 20 µM DCMU or mock (DMSO) and imaged within 2 h of subjective dawn. Data are from the same experiment presented in E). Abbreviations: AsA, L-ascorbate; AU, arbitrary units; DCMU, N-(3,4-dichlorophenyl)-N-dimethylurea.

As previous reports of redox-sensitive fluorophores demonstrate irreversible photoconversion between reduced and oxidized states (Schwarzländer et al. 2008), we monitored the reduction dynamics of HyPer7 in the plastid stroma. Using 50 iterations of excitation via 488 nm laser at 10% output or 0% (mock) output, we saw that excited plastids could rapidly reduce stromal HyPer7 (Supplemental Fig. S3, A, D to G). Relative ratios of excited plastids were comparable to mock values within 30 min of excitation (Supplemental Fig. S3A). We monitored individual channel intensities, which showed decreased 488- and increased 405-channel intensities consistent with reduction of the probe (Supplemental Fig. S3, B and C). Additionally, imaging of root atrichoblast leucoplasts that lack photosystems showed no detectible imaging-dependent changes in HyPer7. This suggests HyPer7 does not undergo photoconversion from reduced-to-oxidized states under our imaging conditions (Fig. 4D). Lastly, we performed sequential region of interest (ROI) excitations at 10 min intervals using 20 iterations of 488 nm laser excitation at 0.5% or 0% power output to show that HyPer7 can be used to dynamically study H₂O₂ in response to excess imaging light (Supplemental Fig. S3H). Taken together, these data report the rapid reduction of HyPer7 in the plastid stroma and do not show evidence of imaging-dependent photoconversion.

Figure 4.

HyPer7 reports cell-type heterogeneity in imaging-dependent H₂O₂ production. A) Ratiometric images of stromal HyPer7 oxidation over imaging time course. Images were obtained by dividing, pixel-by-pixel a 1-pixel-radius median-filtered 488/405 channels. Plastids were cropped using a binary mask obtained by thresholding the 488 channel. B) Stromal HyPer7 oxidation prior to excess-light stress compared across cell types of 8-d-old seedlings grown on 1/2 MS media at subjective dawn. Significantly different groups were determined using a Tukey-HSD test with P-value cutoff of 0.01 (N > 55, 500, 100, and 87 for guard cell, pavement cell, petiole epidermis, and root plastids, respectively). White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. C) Stromal HyPer7 oxidation prior to excess-light stress compared across epidermal cell types of overnight-equilibrated leaf discs from Col-0 collected at bolting. Significance denoted by letters, Tukey-HSD test with P-value cutoff of 0.01 (N > 160 and 255 for guard cell plastids and pavement cell plastids, respectively). White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. D and E) Relative stromal HyPer7 oxidation over the imaging time course separated by cell type in D) 8-d-old seedlings or E) equilibrated leaf discs. Data are from the same experiments presented in B and C). Line indicates generalized additive model ±Se fitted to each cell type. F) Stromal oxidation of GRX1-roGFP2 in 8-d-old seedlings separated by cell type. Significantly different groups were determined using a Tukey-HSD test with P-value cutoff of 0.001 (N = 73, 192, and 81 for guard cell, pavement cell, and petiole epidermis, respectively). White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. G) Representative ratiometric images of stromal GRX1-roGFP2 oxidation in 8-d-old seedlings at subjective dawn. Images were obtained by dividing, pixel-by-pixel, 405/488 channels filtered with a 1-pixel radius Gaussian filter. Plastids were cropped using a binary mask obtained by thresholding a median-filtered 488 channel. Abbreviations: AU, arbitrary units.

H₂O₂ production varies across epidermal cell types

Using our excess imaging light assay, we optically dissected plastids of different epidermal cell types to ask whether whole-leaf or seedling assays ignore heterogeneities in plastid H₂O₂ production. We observed remarkably variable light-dependent H₂O₂ production depending on cell type (Fig. 4, A to E). Abaxial pavement cell sensory plastids showed the highest basal oxidation as well as the largest imaging-dependent oxidation of stroma. Leucoplasts of differentiated root epidermal cells showed almost completely reduced stroma (relative to our DTT estimate) and no detectible oxidation of the probe following imaging. Plastids of the guard cells showed the lowest basal oxidation and accumulation over imaging among the photosynthetic plastids in cotyledons (Fig. 4, B to E). In true leaves, the basal oxidation of guard cell plastids was higher than in cotyledons, but still showed attenuated H₂O₂ production compared to sensory plastids. We noted that the expression of HyPer7 was markedly brighter in guard cell cotyledons compared to pavement cells using DTT to estimate fully reduced stromal HyPer7 (Supplemental Fig. S4, A and B). Since in vitro characterization of HyPer7 shows concentration-dependent sensitivity (Pak et al. 2020), we controlled for the expression difference by monitoring dimmer, T2-generation plants hemizygous for HyPer7 (Supplemental Fig. S4, B and C). Despite the ∼10-fold difference in 405 channel intensity, we still observed similar imaging-dependent oxidation (Supplemental Fig. S4, D to F). Thus, we cannot account for differences in oxidation between sensory plastids and guard cell chloroplasts by probe expression alone. We also assayed stromal redox potential shortly after subjective dawn in these cell types with the previously published probe GRX1-roGFP2 (Ugalde et al. 2021a). Guard cell plastids showed a significantly lower 405/488 ratio (Fig. 4, F and G). This result indicates that guard cell stromal E_GSH is more reduced compared to both sensory plastids and plastids of the petiole epidermis.

ABA-induced guard cell stromal H₂O₂ can accumulate independent of photosynthesis

Because of the drastically attenuated light-dependent H₂O₂ production observed in guard cell plastids, we decided to utilize HyPer7 to monitor H₂O₂ production in guard cell chloroplasts in response to ABA application. Previous reports have provided conflicting interpretations of the role of guard cell chloroplasts during ABA-induced ROS signaling (Azoulay-Shemer et al. 2015; Iwai et al. 2019). Further complicating interpretations, ROS monitoring with epidermal peels and 2',7'-Dichlorodihydrofluorescein diacetate (H₂DCFDA) staining has shown accumulation of dye primarily in the stroma (Leshem and Levine 2013; Iwai et al. 2019). A recent application of plastid-targeted roGFP2-orp1 demonstrated that plastid H₂O₂ partially depends on mitochondrial-sourced ROS during ABA signaling (Postiglione and Muday 2022). It is well known that fluoresceine derivatives are pH-sensitive and cell- or organelle-affixed following cleavage by cellular esterases. Staining with the close analog Carboxyfluorescein diacetate (CFDA) has been used to label plastids in vivo and in vitro in a variety of species, suggesting that esterase activity could be relatively high in the plastid (Borucki et al. 2015; Brunkard et al. 2015). This led us to question the interpretation of previous H₂DCFDA staining in response to ABA.

The primary source of H₂O₂ in the plastid is the light-dependent Mehler reaction at PSI (Mullineaux et al. 2006; Waszczak et al. 2018). To test whether the stroma could accumulate H₂O₂ in response to ABA, independent of PSI, we dark-incubated detached cotyledons with and without 20 μM ABA. Consistent with previous reports, we saw significantly more oxidized stromal HyPer7 in the ABA treatments in both pavement cell sensory plastids and guard cell chloroplasts (Fig. 5, A to C). The degree of oxidation change was more pronounced in guard cell chloroplasts compared to sensory plastids. Interestingly, imaging-dependent oxidation revealed that plastids of both cell types were able to produce more H₂O₂ in the ABA-treated leaves (Fig. 5, D and E). Additionally, 1 h dark pretreatment with DCMU followed by ABA spike-in failed to inhibit stromal oxidation of the probe in both cotyledons and true leaves (Fig. 5, F, H, and I). In cotyledons, we monitored imaging-dependent oxidation of guard cell chloroplasts in DCMU or mock pretreated plants to demonstrate the effectiveness of DCMU in attenuating photosynthesis-derived H₂O₂ during our treatments (Fig. 5G). Together, these data report accumulation of H₂O₂ in guard cell chloroplast stroma independent of photosynthetic electron transport and suggest that plastids are not a primary source of ABA-induced H₂O₂ in guard cells.

Figure 5.

Stromal H₂O₂ accumulates independent of photosystems in response to ABA. A and B) Initial stromal oxidation of 8-d-old cotyledon abaxial pavement cell A) or guard cell B) plastids in response to 15 min dark pretreatment with 20 µM ABA shortly after subjective dawn. White triangle indicates mean (55 < N < 380 plastids from multiple independent plants, ****P << 0.001 and **P < 0.01, Wilcoxon ranked sums test). Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. C) Representative micrographs of guard cell plastids with pseudo-colored stromal HyPer7 488/405 ratio. 488- and 405-nm channels were filtered with 1-pixel radius Gaussian filter prior to ratio calculation. Relative (R-R′) oxidation of pavement cell D) or guard cell E) stromal HyPer7 ±Se over excess imaging light treatment following dark pretreatment with 20 µM ABA or solvent control. Data are from same experiments presented in A and B). Trend line indicates generalized additive model ±Se. F) Stromal HyPer7 oxidation status of 8-d-old cotyledon guard cells following 1 h dark pretreatment with 20 µM DCMU or DSMO and subsequent 20 µM ABA spike-in for 15 min prior to imaging. Letters indicate significantly different groups determined using a Tukey-HSD test with P-value cutoff of 0.01, n > 50 plastids from at least 5 independent plants. White triangle indicates sample mean. Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. G) Relative (R-R′) stromal HyPer7 488/405 ratio ±Se over imaging time course in 8-d-old cotyledons guard cells in the presence of DCMU and/or ABA. Trend lines indicate generalized additive model ±Se. Data are from the same experiments presented in F). H) Representative pseudo-colored micrographs of HyPer7 488/405 ratio in guard cell plastids. 488- and 405-nm channels were filtered with 1-pixel radius Gaussian filter prior to ratio calculation. I) Comparison of HyPer7 stromal oxidation in guard cells pretreated with DCMU and spiked-in with 20 µM ABA or solvent control in 2-wk-old first and second true leaves. Triangles indicate mean relative (R-R^mock) HyPer7 oxidation (N > 160 plastids from multiple independent plants, ****P << 0.001, Wilcoxon ranked sums test). Individual points reflect individual segmented plastids, box indicates interquartile range with center line indicating treatment median. Lines extending beyond box indicate 1.5× interquartile range. Abbreviations: ABA, abscisic acid; AU, arbitrary units; DCMU, N-(3,4-dichlorophenyl)-N-dimethylurea.

Discussion

HyPer7 is a specific and highly sensitive reporter of stromal H₂O₂ dynamics

Here, we report on successful targeting of the genetically encoded H₂O₂ probe, HyPer7, to the plastid stroma. With this approach, we found that excess excitation light from the confocal microscope was sufficient to rapidly oxidize the probe via photosynthesis-derived H₂O₂. Using time-course microscopy with excess imaging light, we monitored stromal peroxide dynamics of plastids to reveal a rapid increase in HyPer7 oxidation, followed by a slowed oxidation or plateau in the stroma of cotyledon sensory plastids. We speculate that the “plateau” phase reflects the outset of NPQ because the plateau observed in epidermal sensory plastids was below the experimentally estimated saturation point of the probe with exogenous H₂O₂ application.

Rapid NPQ response is largely driven by a photosynthesis-derived pH gradient across the thylakoid membrane, which complicates the interpretation of data from previous stroma-localized HyPer variants and requires side-by-side comparisons with pH sensors and peroxide-insensitive HyPer mutants (Exposito-Rodriguez et al. 2017). For example, estimates suggest a ∼32-fold change in HyPer3 488/405 ratios between pH = 6 to 8 (Pak et al. 2020). Utilizing the pH-insensitive HyPer7, therefore, provides unprecedented stromal resolution during the outset of NPQ. While the H₂O₂-specific probe roGFP2-ORP1 has similar pH insensitivity, it is estimated to detect ∼100 nM increases in H₂O₂ whereas HyPer7 is on the order of ∼30 nM (Müller-Schüssele et al. 2021). As the stroma has an extraordinary redox buffering system (E_GSH estimated at −361 mV at pH 8 [Slesak et al. 2003]), HyPer7 provides a sensitive method to detect changes in stromal H₂O₂.

Lastly, we demonstrated in vivo reduction of HyPer7 following excess imaging (Supplemental Fig. S3). We show that reduction of the probe following excess excitation light is rapid, becoming comparable to mock-imaged plastids around 30 min following excitation. Moreover, we have no evidence to support the hypothesis of irreversible, imaging-dependent photoconversion of HyPer7 such as that documented for roGFP1 (Schwarzländer et al. 2008); root leucoplasts show no reduced-to-oxidized photoconversions and reduction kinetics following excitation demonstrate that oxidation of the probe is reversible in the plastid stroma.

HyPer7 shows cell-type-specific H₂O₂ dynamics in response to excess light and hormone application

Via optical dissection, we experimentally distinguished cell-type differences in plastid H₂O₂ production at a basal (unstressed) level as well as in response to excess imaging light. Interestingly, epidermal sensory plastids showed the highest basal H₂O₂ production as well as the highest accumulation of H₂O₂ over imaging stress in both seedling cotyledons and true leaf epidermal cells (Fig. 4, A to E). On the other hand, guard cell chloroplasts showed somewhat attenuated H₂O₂ production in response to excess-light stress compared to sensory plastids. Additionally, we monitored stromal E_GSH with a previously published line targeting GRX1-roGFP2 to the plastid stroma and found that guard cell plastids showed significantly substantially more reduced roGFP2 when compared to epidermal sensory plastids and plastids of the petiole epidermis (Fig. 4, F and G). This observation suggests that heterogeneities may also exist in stromal glutathione redox buffering depending on cell type.

The highly attenuated, light-dependent H₂O₂ production by guard cell chloroplasts led us to reconsider the role of guard cell chloroplasts during ABA signaling. While reports suggest that photosynthesis is required for stomatal closure and H₂O₂ accumulates in plastids (Leshem and Levine 2013; Iwai et al. 2019), it is unclear whether plastids are a primary H₂O₂ source in response to ABA application. To test this hypothesis, we inhibited plastid-derived H₂O₂ formation via dark incubations and with DCMU incubation prior to ABA treatment. Our results suggested that while the plastid stroma accumulated more H₂O₂, this accumulation did not depend on light or photosynthesis. However, when imaging stress was applied, we saw more oxidation of the stroma, suggesting that plastids may contribute to H₂O₂ accumulation during instances of drought and excess-light stress. A recent study using roGFP2-ORP1, found that ABA application could oxidize guard cell chloroplast stroma, and this partially depended on mitochondria-sourced H₂O₂ (Postiglione and Muday 2022). We speculate that, in the instance of ABA application, RBOH-derived and potentially mitochondria-derived H₂O₂ depletes the stromal antioxidant systems, leading to the increased imaging-dependent oxidation we observed (Fig. 5). This could be an intriguing intersection for integrating multiple abiotic stress signaling pathways that will require further investigation. Additionally, we speculate that the biological relevance of H₂O₂ buffering observed in the guard cell chloroplasts could function to prevent light-directed stomatal closure to allow gas exchange during the day. Whether these trends are mirrored or differ in guard cell chloroplasts of C4 and/or CAM photosynthetic plants will be an intriguing point for future inquiry. For example, a hallmark of CAM photosynthetic plants is nocturnally opened stomata that allow gas exchange and carbon fixation at night (Males and Griffiths 2017); one could hypothesize that plastid-derived H₂O₂ could be utilized to signal stomatal closure during transitions from night to day. With advances in the transformation of the facultative-CAM species (Agarie et al. 2020), Mesembryanthemum crystallinum (L.), incorporation of redox biosensors would provide a tractable system for comparative investigations of guard cell chloroplasts during C3 and CAM photosynthesis. Our analysis focused primarily on the origin of H₂O₂ in the stroma, and we did not monitor stomatal dynamics or conductance in response to our treatments. Further investigation of the role of guard cell chloroplasts should include experimental precautions such as dark incubations with ABA to elucidate the physiological relevance of plastid localized H₂O₂ during ABA signaling. Taken together, these data support a hypothesis where a primary source of H₂O₂ during ABA signaling is RBOH proteins located at the plasma membrane, and guard cell chloroplasts may contribute to H₂O₂ accumulation under conditions of excess light (Kwak et al. 2003; Chapman et al. 2019).

HyPer7 limitations

HyPer7 is a pH-insensitive H₂O₂-specific probe, but its steady state ratio also relies on the reduction kinetics, chromophore maturation time, and turnover of the protein. The intramolecular disulfide bridge in HyPer7 is thought to be reduced by Trxs; however, the exact mechanism of cellular and stromal reduction of HyPer7 remains elusive and beyond the scope of our study. Such analysis will be challenging in planta due to the presence of both Grxs and Trxs as well as the duplication of the plastid-targeted Trxs. Moreover, some of the plastid-targeted Trxs show severe mutant phenotypes (Meng et al. 2010). The differences in basal oxidation status of HyPer7 and E_GSH in plastids from distinct cell types also raises the question of whether (and more importantly, why) heterogeneities exist in stromal enzymatic reducing systems. In guard cell chloroplasts, we demonstrated rapid reduction of stromal HyPer7 on the order of minutes following excess-light derived H₂O₂ (Supplemental Fig. S3). Such kinetic limitations have been ameliorated in redox potential probes, such as GRX1-roGFP2, where a glutaredoxin (GRX1) is fused to the probe, facilitating its reduction (Aller et al. 2013). With the limitations of reduction time in mind, caution is required while preparing samples for imaging when comparing the absolute HyPer7 ratio. To minimize the excess-light stress of samples, we focused seedlings with a <100 µE brightfield lamp.

Another important consideration when using HyPer7 is that 405 and 488 nm light scatters differently. This effect could produce artificially higher 488/405 ratios while imaging cells beneath the epidermis. In the case of roGFP2, estimates suggest a ∼20% reduction in dynamic range at a depth of 120 µm into Arabidopsis leaves (Schwarzländer et al. 2008). To compare tissues at different depths, it may be necessary to recalibrate the HyPer7 dynamic range for each imaging depth—a challenging task as HyPer7 exhibits an extraordinary dynamic range. As an alternative approach, previous reports indicate that HyPer family proteins may have different fluorescence lifetimes depending on their oxidation status (Bilan et al. 2013). If this feature is true of HyPer7, fluorescence lifetime imaging with a single excitation source or even multiphoton excitation may minimize artifacts from imaging deeper into tissues.

Materials and methods

Plant growth conditions

Arabidopsis ecotype Col-0 was sown on ½ MS media, pH 5.7 without sucrose. Cotyledons were imaged at 6 and 8 d of age, and the first true leaves were imaged at 14 d after sowing. All plants were grown at 12-h daylengths at 22°C under a constant fluence rate of 150 μE. Leaf discs were harvested from fully expanded true leaves at bolting from soil-grown plants. Leaf discs were equilibrated overnight in dH₂O and imaged the following morning.

Generation of stromal targeted HyPer7

HyPer7 was cloned using the primers listed in Supplemental Table S1 from pLifeAct:HyPer7 (Pak et al. 2020) into the entry vector pCR8/GW/TOPO (Invitrogen). Preserving the original amino acid sequence, the internal BSA1 restriction site was mutated using PCR-based site-directed mutagenesis (Supplemental Table S1). The mutated BSA1 HyPer7 was cloned into pGGD000 (Lampropoulos et al. 2013; obtained from Addgene). The targeting sequence from Arabidopsis small subunit of RuBisCo (SSU) was directly cloned into pGGC000. The final binary vector was assembled using BSA1-mediated golden gate assembly with plasmids listed in Supplemental Table S2. Arabidopsis ecotype Col-0 was transformed using the floral dip method and screened on 1/2 MS media (Without sucrose, buffered at pH 5.7 with 0.05% w/v MES) with 40 mg L⁻¹ kanamycin.

Confocal microscopy

All plant materials were mounted on coverslips of 1.5 thickness and imaged in either dH₂O or corresponding incubation buffers. To prevent crushing the tissues, 2 small drops of vacuum grease were placed on either side of the seedling or tissue. Colocalization of HyPer7 with chlorophyll used a 40×/1.2NA c-Plan-Apochromat water immersion objective fitted on a Zeiss LSM510-Meta, exciting at 488 nm. Fluorescence was collected from 505 to 550 nm (HyPer7) and 650 nm long pass (chlorophyll). For ratiometric imaging of HyPer7, we used a Zeiss LSM510-Meta (40×/1.2NA, 1.3× digital zoom) with sequential excitation, alternating between 405 and 488 nm lasers at 2.8% and 1.5% acousto-optic tunable filter power, respectively. To accommodate variation in brightness between cell types and avoid overexposure, the confocal pinhole was adjusted between 0.6 and 1 airy unit. PMT gain, offset, and laser output remained constant between replicates. PMTs were allowed to stabilize for 1 h prior to imaging. To avoid excess excitation light while focusing the samples, we focused with <100 μE illuminating samples using the brightfield lamp. Time-course experiments were conducted with near-constant laser excitation (sampled at 5 s intervals with a scan time of 3.95 s). All assays were performed within 2 h of subjective dawn in the 12-h day/night cycles that seedlings were grown. Whole-leaf scanning of HyPer7 fluorescence was achieved using the tile scan feature with a 10×/0.3NA Epiplan-Neoflaur objective and identical filter setup as described above. Light output of the brightfield halogen lamp was measured with a LI-COR quantum radiometer (Model LI-185B) secured to the sample stage. For excess brightfield light treatments, Kohler illumination was performed to ensure even illumination of the field of view. ROI excitations of guard cell pairs were performed using the regions and bleaching features on Zen software; laser powers and iterations are indicated in figure legends. Samples were imaged using a 20×/0.8NA plan-apochromat objective at 3× digital zoom. For estimating reduction kinetics, detached cotyledons were secured by placing a small nut on top of the cover slip (Supplemental Fig. S2I); to prevent crushing the cotyledon, 2 small drops of vacuum grease were placed next to the leaf.

Image analysis and processing

Confocal data were analyzed using ImageJ. An automated macro segmented individual plastids by applying a 2-pixel radius median filter to the 488 channel. The filtered image was thresholded using the “default” threshold method and segmented using the particle analyzer plugin by size (2 to 18 μm² for most plastids or 5 to 30 μm² for true leaf sensory plastids) and circularity (0.5 to 1). Segmented ROIs measured the original, unfiltered 488 and 405 images. The quotient of the mean 488 and 405 channels per plastid was used to quantify HyPer7 oxidation. Images used to illustrate stromal HyPer7 oxidation were created using a binary mask generated from thresholding the 488-channel. 488 and 405 channels were filtered with a Gaussian blur of sigma radius set to 1 pixel. A ratiometric image was produced from the Gaussian-blurred images using the Ratio Plus imageJ plugin. The binary mask was then applied to crop the stroma from background using the “AND” function on the Image Calculator plugin. The ratio min/max was set between 0 and 6 unless specified in the figures. Relative ratios presented were calculated by subtracting the initial average ratio from each subsequent image in the time course. Identical procedures were used to process GRX1-roGFP2, with the exception that 405/488 ratios were also used.

Chemical applications

ABA (Sigma, A1049) was dissolved in 100% ethanol at a 1,000× (20 mM) concentration. Shortly before incubations, the stock was diluted in dH₂O to a final concentration of 20 μM; mock treatments used an identical dilution of 100% ethanol. DCMU (diuron, Sigma 16902) was dissolved at a 500× (10 mM) concentration in DMSO; mock treatments used an identical dilution of DMSO. One millimolar L-ascorbate (Sigma A0278) was prepared fresh and adjusted to a pH of 5.7 prior to treatment using NaOH. To reduce and oxidize HyPer7 in vivo, seedlings were gently vacuum-infiltrated with dH₂O using a syringe and transferred into 10 mM H₂O₂ or 10 mM DTT and incubated for at least 30 min prior to imaging. All chemical incubations were performed in the dark.

Accession numbers

Small subunit of rubisco target peptide (SSU) was cloned from AT1G67090, ecotype Col-0.

Author contributions

I.J.D. performed cloning and transformation of HyPer7 constructs and I.J.D. and S.A.M. designed experiments. I.J.D. performed experiments with assistance from K.K. I.J.D. wrote the first draft of the manuscript and S.A.M. provided funding to the study.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Table S1. List of primers used in this study.

Supplemental Table S2. List of plasmids used in this study.

Supplemental Figure S1. HyPer7 fluorescence separation from chlorophyll autofluorescence.

Supplemental Figure S2. Automated image analysis procedure.

Supplemental Figure S3. In vivo reduction kinetics of HyPer7 in guard cell chloroplasts following high light-induced oxidation.

Supplemental Figure S4. Comparison of HyPer7 oxidation by probe expression.

Funding

This work was supported by funding from National Science Foundation (NSF) (1853519) and NIH (R01 GM134056-01) to S.A.M.

Data availability

The source of all constructions and reporters has been provided in the text.