- © 2019 American Society of Plant Biologists. All Rights Reserved.
Abstract
Autophagy is a major catabolic process in eukaryotes with a key role in homeostasis, programmed cell death, and aging. In plants, autophagy is also known to regulate agronomically important traits such as stress resistance, longevity, vegetative biomass, and seed yield. Despite its significance, there is still a shortage of reliable tools modulating plant autophagy. Here, we describe the first robust pipeline for identification of specific plant autophagy-modulating compounds. Our screening protocol comprises four phases: (1) high-throughput screening of chemical compounds in cell cultures of tobacco (Nicotiana tabacum); (2) confirmation of the identified hits in planta using Arabidopsis (Arabidopsis thaliana); (3) further characterization of the effect using conventional molecular biology methods; and (4) verification of chemical specificity on autophagy in planta. The methods detailed here streamline the identification of specific plant autophagy modulators and aid in unraveling the molecular mechanisms of plant autophagy.
Autophagy is a key regulator of cellular homeostasis and plays an essential role in abiotic and biotic stress responses, fecundity, aging, and cell death (Minina et al., 2013, 2014, 2017; Dagdas et al., 2016; Avin-Wittenberg et al., 2018; Ustun et al., 2018). The autophagy pathway comprises selective or bulk sequestration of cytosolic contents into double-membraned vesicles known as autophagosomes and their delivery to the lytic compartment, where autophagy activity (or autophagic flux) completes with the degradation of the delivered cargo (Ohsumi, 2014). Autophagy is coordinated by evolutionarily conserved autophagy-related (ATG) proteins that orchestrate sequestration of the cargo, formation and trafficking of the autophagosomes, and cargo degradation. Among these proteins, ATG8 is remarkable in the sense that it is incorporated into the membranes of forming autophagosomes, trafficked to the lytic compartment, and degraded together with the cargo, making it an ideal tool for autophagy activity readout (Klionsky et al., 2016).
The importance of autophagy for plant stress tolerance in general and crop performance in particular is of critical interest in plant biology, requiring continuous development of increasingly sophisticated tools for its investigation (Batoko et al., 2017; Avin-Wittenberg et al., 2018). The forward and reverse genetic approaches such as knockouts, knockdowns, and overexpression of ATG genes, although effective tools, have limitations (Klionsky et al., 2016; Avin-Wittenberg et al., 2018). The chemical modulation of autophagy presents an alternative option and has also been instrumental for our current understanding of the underlying mechanisms controlling autophagy (Klionsky et al., 2016). However, an increasing number of studies describe significant off-target effects of commonly used autophagy-modulating compounds. For example, 3-methyladenine, which is typically used as an inhibitor of autophagy (Klionsky et al., 2016), can also increase autophagic flux after prolonged treatments due to its transient suppression of class III phosphoinositide 3-kinases (PI3Ks; Wu et al., 2010). Another autophagy inhibitor, chloroquine, has been shown recently to induce autophagy-independent severe disorganization of the Golgi and endolysosomal systems (Mauthe et al., 2018). An enhancer of autophagy, rapamycin, extensively used for yeast and animal model systems, is known to specifically form a ternary complex with the FKBP12 (12-kD cis-trans-peptidyl-prolyl isomerase FK506-Binding Protein) and the TOR (Target of Rapamycin) kinase to induce autophagy (Lorenz and Heitman, 1995; Klionsky et al., 2016). Nevertheless, an opposite effect is caused by high concentrations of rapamycin (Zhou et al., 2012). Furthermore, although rapamycin was originally thought to be ineffective in Arabidopsis (Arabidopsis thaliana; Sormani et al., 2007), potentially due to the low capacity of plant FKBP12 proteins to form the complex, a more recent report suggests that this is not the case (Xiong and Sheen, 2012). These studies illustrate the importance and need for precise and robust screening strategies to identify autophagy modulators. Furthermore, novel modulators of autophagy may play a critical role in understanding the underlying mechanisms regulating the process.
Here, we describe the first robust pipeline for the identification of chemical compounds specifically modulating plant autophagy. The assays utilize a combination of reporter and control transgenic lines to optimize the treatment conditions for cell cultures and in planta to ensure that the compounds influence autophagic flux without causing considerable stress to other endomembrane compartments. Using this new method, we have identified several promising autophagy-modulating compounds.
RESULTS
Screen Overview
The pipeline comprises a combination of assays each having an incremental decrease in throughput capacity but an increase in the depth and reliability of the data (Fig. 1A). The assays are grouped into four phases, each providing a new more detailed level of information about investigated compounds. The combination of all phases, including the optimization of the upstream steps based on downstream results (see feedback connections in Fig. 1A), was carried out to provide reliable characterization of the compound effect in vivo.
Multiphasic screen for autophagy-modulating compounds. A, The screen comprises four interconnected phases that allow identification of the best candidate compounds and verification of their specificity. The first phase is performed in a high-throughput mode, testing a library of compounds on cells expressing a reporter construct. The assay is performed in lysates of treated cells and can provide quantitative end-point readouts. The assay of the second phase is performed in planta and has lower throughput, as it employs advanced microscopy. It provides real-time quantitative readout of autophagy activity and morphological data. While the readout in the first two phases is based on delivery of autophagosomal membranes to the lytic vacuole, the third phase quantifies the degradation of autophagic cargo (i.e. completion of the autophagic flux). Phase 4 is an essential step of the screen that verifies the specificity of the compound’s effect to autophagy in planta using transgenic endomembrane marker lines. Importantly, results obtained in each phase are used for optimization of the previous steps: optimization of the concentration range for future screens (dashed arrow i); estimation of efficacy and timing of cargo delivery versus its degradation (dashed arrow ii); and optimization of the concentration range of compounds for autophagy-specific effect (dashed arrow iii). B, In the first phase, a high-throughput screen of compounds modulating autophagy is performed using the dual-luciferase assay. Reporter represents wild-type (WT) tobacco BY-2 cells coexpressing Fluc fused to ATG8 and Rluc fused to NLS. Upon activation of autophagy, Fluc-ATG8 incorporates into autophagosome membranes and is delivered to the vacuole where Fluc activity is lost, while Rluc-NLS levels are unaffected. Control represents wild-type BY-2 cells coexpressing Fluc-NLS and Rluc-NLS, which are unaffected upon autophagy up-regulation. The readout for phase 1 is the ratio between chemiluminescence intensities of Rluc and Fluc substrates measured in cell lysates. C, In the second phase, delivery of ATG8 to the lytic vacuole is assessed using confocal microscopy. For this, the tandem tag (TT) of red (TagRFP) and green (mWasabi) fluorescent proteins that are fused to ATG8 is expressed in a wild-type background (Reporter) or in an autophagy-deficient background (Control). Upon activation of autophagy, the TT-ATG8 is delivered to the vacuole in the reporter line, where green fluorescence is quenched by the acidic pH, while red fluorescence persists. The ratio between red and green fluorescence intensities in the vacuole serves as the readout and is proportional to the rate of autophagosome delivery to the lytic compartment. D, In phase 3, the canonical GFP-cleavage western-blot assay is implemented to assess the completion of autophagic flux. For the assay, GFP-ATG8 is expressed in either a wild-type (Reporter) or autophagy-deficient (Control) background. Upon activation of autophagy in the reporter line, the GFP-ATG8 fusion is delivered to the vacuole and cleaved. Thus, the ratio of cleaved versus total GFP provides the readout for autophagy activity. A, Autophagosome; C, cargo; Cyt, cytoplasm; N, nucleus; P, phagophore; V, vacuole.
Phase 1: Hit Identification
The first phase was performed in the high-throughput mode using tobacco (Nicotiana tabacum ‘Bright Yellow 2’) cell cultures (Nagata et al., 1992) expressing chemiluminescent reporters for autophagic flux (Fig. 1B). Transgenic cv BY-2 cells were incubated and treated in 96-well format, lysed on the same plate, and used for quantitative measurement of luminescence. In this assay, the reporter line coexpressed Renilla luciferase (Rluc) fused to a nuclear localization signal (NLS) and firefly luciferase (Fluc) fused to ATG8. Upon activation of autophagy, Fluc-ATG8 was delivered to the lytic vacuole, leading to the loss of its activity and consequently resulting in an increase of the ratio of Rluc to Fluc. In the control line, both luciferases are targeted to the nuclei, where they are protected from sequestration into autophagosomes. The dual-luciferase assay permits measurement of both response (Fluc-ATG8 for the reporter line and Fluc-NLS for the control line) and reference (Rluc-NLS) luminescence in the same cells, thus removing possible bias originating from cell-to-cell fluctuations of expression or differences in cell number per sample. Additionally, absolute values of the Rluc luminescence can be used as a crude estimation of cell viability, as it generally dramatically decreases in lysates of dying cells.
The growth and treatment conditions were preliminarily optimized using known inducers and inhibitors of autophagy, such as AZD8055 (Dong et al., 2015), benzothiadiazole (Yoshimoto et al., 2009), and concanamycin A (Yoshimoto, 2004; Supplemental Fig. S1, A and B; data not shown). Importantly, data from the viability assay of phase 2 (described below) provided a critical insight into an average range of compound concentrations that are not toxic for cv BY-2 cells (Fig. 2A). Therefore, it is advisable to perform a screen of a relatively small number of compounds first to estimate the optimal average concentration range for the library of interest (feedback connection i, Fig. 1A). As a proof of concept, a set of 364 small synthetic compounds of the plasma membrane recycling compound set A (Drakakaki et al., 2011) and plasma membrane recycling inhibitors in pollen (Drakakaki et al., 2011) libraries was screened in this study, and a total of 19 enhancers and five inhibitors of autophagy were chosen for further testing (Supplemental Fig. S1B).
Phase 2: viability testing. Tobacco cv BY-2 cell cultures (A) and roots of 5-d-old Arabidopsis seedlings (B) were exposed for 4 h to a concentration gradient of C12 dissolved in 1% (v/v) dimethyl sulfoxide (DMSO; vehicle). Viability is assessed at multiple time points using differential interference contrast (DIC) light microscopy. Swollen (white arrows in A) and collapsed (black arrows in A) protoplasts are shown in cv BY-2 cells at 200 and 400 µm concentration, respectively. Arabidopsis roots with wrinkled (white arrows in B) or collapsed (black arrows in B) protoplasts are shown at 100 and 200 µm concentration, respectively. Scale bars = 20 µm.
Phase 2: Hit Optimization
To enhance the efficiency of the screen and promote an early elimination of compounds that do not reveal in vivo activity, in planta assays were performed early in the screen, already at the second phase.
Solubility Assay
The first estimation of the applicable concentration range for the compounds selected in phase 1 was performed by assessing their solubility in the medium used for growing cell cultures and seedlings under the corresponding growth conditions. For each selected compound, we tested the concentration used in the first phase and additional concentrations that were severalfold greater. Additionally, at this stage, it is advisable to examine the predicted in vitro and in vivo stability of the selected compounds under the treatment conditions (Di et al., 2006). This will allow correct handling of the compounds based on factors such as sensitivity to aqueous solutions, light, or pH during upcoming experiments and omit compounds predicted to be unstable.
Viability Assay
After estimating the range of possible concentrations of a compound, predicted by physical characteristics including but not limited to Mr, acid-base Kd, and partition coefficient (Manallack, 2007), it is essential to identify the maximal concentration that is not cytotoxic but still influences autophagy. Each compound at its maximal concentration selected based on the results of the solubility assay, along with serial dilutions, were used for treatment of cv BY-2 cell cultures and Arabidopsis seedlings. Viability of the plant material was assessed by phenotyping specimens at 0 to 24 h of the treatment (depending upon the potency of each compound) using DIC microscopy. Only the concentrations not causing any visible changes in cell morphology were considered for further assays.
An example presented in Figure 2 illustrates the dose-dependent toxicity of compound 12 (C12) after a 4-h treatment. At 50 µm, C12 had no observable effect on the cv BY-2 cells (Fig. 2A), whereas the 200 µm treatment caused swelling of the cells (white arrows, Fig. 2A) and 400 µm C12 led to protoplast collapse (black arrows, Fig. 2A). Interestingly, we observed that, on average, root cells of Arabidopsis seedlings were considerably more sensitive to the compound treatments than cv BY-2 cells (Fig. 2). For instance, a 100 µm treatment of C12 resulted in wrinkled protoplasts in the epidermal root cells of Arabidopsis seedlings (white arrows, Fig. 2B), whereas 200 µm C12 led to protoplast collapse (black arrows, Fig. 2B). The difference in the sensitivity of the model system to the same compounds is an important factor to be assessed for phase 1 optimization (feedback arrow i, Fig. 1A).
Tandem Tag Assay
The first phase provides high-throughput quantitative data about delivery rate of the ATG8 fusion to the lytic compartment upon treatment of plant cell culture. Although this readout usually correlates well with autophagy activity, it still might provide false-positive and false-negative results. The TT assay not only allows the assessment of time- and dose-dependent delivery rate of the ATG8 fusion to the lytic compartment upon treatment with the compounds in planta but also yields important morphological data on autophagosome formation and confirms that the observed changes in ATG8 localization are autophagy dependent. For this assay, ATG8 fused to the fluorescent TT consisting of TagRFP and mWasabi (Zhou et al., 2012) was expressed in wild-type and autophagy-deficient Arabidopsis plants (reporter and control lines for autophagic flux detection, respectively; Fig. 1C). In the low pH of the vacuole, mWasabi green fluorescence is quenched more efficiently than TagRFP red fluorescence, and therefore the ratio of the red to green fluorescence intensities in the vacuoles can be used as a readout for autophagy activity, with higher ratios indicating increased autophagic flux.
Importantly, the basal autophagy level and compound sensitivity vary significantly depending on the cell type and differentiation stage. We observed the most reproducible results when performing microscopy experiments on epidermal root cells located at the beginning of the differentiation zone (Fig. 3A).
Phase 2 continued: hit verification and optimization in planta. A, Roots of Arabidopsis seedlings (5-d-old) expressing the TT (TagRFP-mWasabi) fused to ATG8a are utilized for hit optimization. The epidermal layer at the beginning of the root differentiation zone (DZ), located above the elongation and meristematic zones (EZ and MZ, respectively), is scanned via confocal microscopy. Images are then processed by our script to automatically quantify red (TagRFP) and green (mWasabi) fluorescence intensities in the vacuoles and perform statistical analysis of red-to-green ratios versus time or concentration. B, The TT-ATG8a vector is expressed in wild-type (WT) and atg5atg7 Arabidopsis lines to determine the effects of each compound (C1–C16 and In4) on autophagic flux. Treatments represented by red text (In4 and C3) are performed on seedlings exposed to nitrogen starvation for 24 h. Statistical information is as follows: n ≥ 6; ***, P < 0.005; **, P < 0.01; *, P < 0.05; ns, nonsignificant (Student’s t test). Error bars represent se. Scale bars = 15 µm.
First, we performed an estimation of the optimal concentration and treatment duration. For each compound, a range of concentrations below the cytotoxicity level identified in the viability assay were used to treat the TT reporter line only. The time-resolved changes in vacuolar fluorescence were imaged using confocal microscopy. The obtained data were originally quantified manually through the selection of small regions of the vacuoles using the DIC channel and then later automatically processed by our designated ImageJ-based macro and R script (Fig. 3; Supplemental Fig. S2A, TT assay protocol). Next, the optimal treatment conditions for each compound were verified using both control (atg5atg7 TT) and reporter TT lines and the appropriate vehicle control (Fig. 3B). Figure 3B illustrates the effect of a set of compounds on red/green fluorescence, indicating increased autophagic flux. The assay revealed that under the tested conditions, several compounds cause a significant change of the TT ratio compared with the vehicle treatment in the reporter line (WT TT-ATG8a, Fig. 3B) but not in the control line (atg5atg7 TT-ATG8a, Fig. 3B). Please note that the TT assay and the later phases described below can be utilized to test the effects of potential autophagy inhibitors by subjecting the seedlings to nitrogen starvation for 18 h prior to treatment application (In4 and C3, Fig. 3B).
The final optimization of the treatment conditions is carried out using the TT assay as the last step of phase 4 (feedback loop iii, Fig. 1A).
The TT assay in cv BY-2 cultures is less straightforward compared with Arabidopsis due to the lack of corresponding autophagy-deficient control cell lines. Nevertheless, it still can be used by combining TagRFP-mWasabi-ATG8 and TagRFP-mWasabi-NLS as a reporter and a control line for autophagic flux quantification, respectively (Supplemental Fig. S2B).
Phase 3: Verification of Autophagic Flux Completion
While the first two phases provide information about autophagy-dependent delivery of ATG8 to the lytic compartment, it remains unclear whether the final step of autophagic flux (i.e. cargo degradation) also takes place. The canonical assay to assess this is the GFP-ATG8 cleavage assay (Nair et al., 2011; Klionsky et al., 2016).
Arabidopsis plants expressing GFP-ATG8a in wild-type or autophagy-deficient backgrounds are used as reporter and control lines, respectively, for autophagic flux quantification (Fig. 1D). Upon induction of autophagy, GFP-ATG8 is delivered to the vacuole at a higher rate, where it is first cleaved by lytic proteases, leading to transient accumulation of the cleavage product GFP. Western-blot analysis of protein extracts allows the estimation of a ratio between total and cleaved GFP in plant cells, thus revealing the relative level of autophagic flux (Fig. 1D).
Importantly, tissue- and organ-specific differences in the levels of 2× 35S-driven expression of GFP-ATG8 and/or basal autophagic flux can considerably skew results. For example, protein extracts from whole seedlings compared with the root differentiation zone of the same lines under the same treatment conditions reveal significant differences in the readout (Fig. 4A). We established a designated protocol for the GFP-ATG8 cleavage assay for the Arabidopsis root differentiation zone and obtained reproducible results consistent with the results of the TT assay (Figs. 3B and 4, B and C). Notably, the combination of GFP-ATG8 cleavage assay results with morphological and quantitative data obtained via the TT assay may indicate which step of the autophagy pathway is affected by the investigated compound (feedback arrow ii, Fig. 1A). For example, inhibitor 4 had an accumulation of cytoplasmic TT-positive puncta at high concentrations, suggesting that fusion of the autophagosomes to the vacuole and autophagic flux was disrupted (data not shown). Interestingly, slightly greater concentrations of putative autophagy modulators or longer incubation times were needed for efficient GFP-ATG8 cleavage assay, compared with the TT assay. This might be caused by differences in the cell types that were included in each assay (only epidermal cells for the TT assay versus all cell types for the GFP-ATG8 cleavage assay) or differences in the sensitivity of the methods. Alternatively, it might indicate that GFP-ATG8 fusion hydrolysis does not occur immediately after delivery of the autophagosomes to the lytic compartment.
Phase 3: validation of autophagic flux completion. A, GFP cleavage assay for Arabidopsis whole-seedling and root differentiation zone protein extracts from the following lines: the wild type (WT), GFP-ATG8a expressed in the wild type (WT GFP-ATG8a), and atg5 knockout (atg5 GFP-ATG8a). Treatments were applied for 6 h prior to protein extraction and included a vehicle control (1% [v/v] DMSO) and 0.5 µm AZD8055 (AZD). B, Immunoblotting with α-GFP of differentiation zone protein extracts from WT GFP-ATG8a and atg5 GFP-ATG8a lines (three independent replicates per treatment and genetic background) treated with 1% (v/v) DMSO (vehicle) or 30 µm C12 for 4 h. Bio-Rad TGX stain-free gels and α-actin are utilized as loading controls. The GFP-ATG8a fusion band is represented by the white arrow, cleaved GFP is represented by the black arrow, and the lower molecular weight band is represented by the red arrow. The dashed line in A indicates membrane splicing. C, Cleaved versus total GFP ratios comparing vehicle and 30 µm C12 treatment in WT GFP-ATG8a and atg5 GFP-ATG8a lines. Statistical information is as follows: n = 3; **, P < 0.01; ns, nonsignificant (Student’s t test). Error bars represent se.
Phase 4: Specificity Check
Autophagy is integrated tightly with endocytic and exocytic pathways and the trafficking routes to the vacuole (Deretic et al., 2012; Spitzer et al., 2015; Pečenková et al., 2017). Thus, it is probable that putative modulators of autophagy identified in the previous phases of the screen influence the endomembrane system and thus activate or inhibit autophagy indirectly and have significant off-target effects, as was recently shown for some characterized compounds (Wu et al., 2010; Mauthe et al., 2018).
To minimize this possibility, we designed the fourth phase of the screen where the effect of each candidate compound on endomembrane compartments is tested in planta using a library of fluorescent marker lines (Supplemental Table S1). The library comprises the minimal set representing the key components of the endomembrane system, including endoplasmic reticulum (ER), Golgi apparatus, trans-Golgi network, multivesicular bodies, vacuole, mitochondria, peroxisomes, plasma membrane, and endocytic and exocytic pathways.
Each compound was tested on the set of marker lines using the conditions optimized in the second and third phases. If detectable morphological changes were observed for any of the markers, a range of lower compound concentrations and treatment times were tested. This allowed us to estimate the maximum possible concentration at which a compound did not change the morphology of any endomembrane marker. The identified concentration was then tested using the TT assay to assess its impact on autophagy activity (feedback arrow iii, Fig. 1A).
Figure 5 provides an example of optimization for C12 at conditions originally optimized in phase 2. A gradient of C12 concentrations was tested using an ER marker line (Fig. 5A). Concentrations of C12 at 60 µm and above caused significant ER stress, resulting in a disrupted cisternae network and the formation of swollen GFP-positive bodies (white arrows, Fig. 5A; Supplemental Video S1). C12 at 30 µm or less did not alter ER morphology (Fig. 5A). Thus, C12 concentrations ranging from 7.5 to 30 µm were tested using the TT and GFP-ATG8a cleavage assays and showed a significant impact on autophagic flux (Fig. 5, B and C).
Phase 4: identification of autophagy-specific compounds and conditions. A, HDEL-GFP Arabidopsis seedlings exposed to a gradient of C12 concentrations to assess its phenotypic effects on the ER. Swollen ER bodies (white arrows) and disrupted ER networks occur at higher concentrations (60 and 120 µm). B, TT (TagRFP-mWasabi)-ATG8a reporter lines treated with a range of C12 concentrations. C, The GFP-cleavage assay for a gradient of C12 treatments. Western-blot detection is shown for GFP in protein extracts prepared from differentiation zone of WT GFP-ATG8a and atg5 GFP-ATG8a lines. Bio-Rad TGX stain-free gels and α-actin immunoprobing are used as loading controls. The GFP-ATG8a fusion band is represented by the white arrow, cleaved GFP is represented by the black arrow, and the lower molecular weight band is represented by the red arrow. Statistical information in B is as follows: n = 3; **, P < 0.01; *, P < 0.05; ns, nonsignificant (Student’s t test). Error bars represent se. Scale bars = 15 µm (A) and 20 µm (B).
DISCUSSION
We describe a novel robust pipeline designed for the identification of chemicals specifically modulating plant autophagy. Our approach resulted in hits with minimal off-target effects and provides data essential for further characterization of bioactive compounds. One of the most important features of the new pipeline is the combination of a high-throughput screen utilizing cell cultures with further verification steps performed in planta. This approach simultaneously enables screening of large numbers of compounds and eliminates false-positive candidates that are active in cell cultures but do not have activity in planta due to differences in cell permeability, stability in medium used for cell culture, and plant growth- or cell type-specific responses. Furthermore, combined use of tobacco cell cultures and Arabidopsis seedlings provides a first estimation of potential conservation of targets in plant species belonging to different families. Interestingly, in our experience, the average concentration range of compounds applicable for cv BY-2 cells was significantly higher than the concentration range suitable for Arabidopsis plants, indicating differences in responses between the two models. It is also possible, however, that cv BY-2 cells would respond similarly at the same concentrations with increased exposure times, but more detailed kinetic experiments would be required to test this hypothesis. Nevertheless, the difference observed is a useful starting point for future screens utilizing either of the model systems. Notably, the absolute values of the applicable concentration ranges must be individually adjusted to a specific chemical library, as the activity of compounds might vary due to purity, age, structural stability, and storage differences.
As further characterization of new autophagy modulators is costly and labor intensive (Hutson et al., 2017; Schenone et al., 2017), it is important to identify bioactive chemicals that specifically modulate autophagy under physiological conditions. To achieve this, we designed a multifaceted approach in which the first three phases involved advanced fluorescence microscopy, biochemistry, and cell and molecular biology methods. This approach provides quantitative and morphological data on autophagy pathway activity and is optimized to assess the effect of a chemical on several key aspects of the pathway. This characterization not only ensures reliability of the conclusions about effects on plant autophagy but also provides an indication of the affected stage of the autophagy pathway (e.g. phagophore and autophagosome formation, cargo delivery or cargo degradation). The ATG8 lipidation assay (Chung et al., 2010) could provide additional strong evidence for the compound effects on phagophore and autophagosome formation; however, it is much more labor intensive and time consuming than the TT and GFP-cleavage assays. Thus, it might be more useful at later stages of compound characterization, when only a few selected compounds with optimized concentrations/treatment times are used. The assays of the first three phases of our pipeline are focused on assessing the chemicals’ effect on autophagy activity, and the fourth phase puts the activity into a broader context. Due to the overwhelming complexity of the network, it is frequently overlooked that autophagy is an integral part of the endomembrane trafficking system (Ponpuak et al., 2015; Pečenková et al., 2017; Søreng et al., 2017). The assays in the fourth phase identify treatment conditions under which candidate compounds significantly influence autophagy activity without causing detectable aberrations in other components of the trafficking machinery. A recent review of chemical biology screens for autophagy modulators by Mishra et al. (2018) highlights the wide range of assays and techniques developed for yeast and mammalian cell cultures. Despite these advancements, it remains challenging to monitor tissue-specific effects in animal systems (Mishra et al., 2018). The assays described here offer a distinct advantage over animal systems, as they are time and cost effective and provide insight into tissue-specific effects of the compounds. Although informative, it should be noted that the assays do not provide a definite proof of autophagy specificity of the compound. A more detailed estimation of potential off targets of the compound can be achieved via assessment of its effect on loss-of-function mutants of the predicted target and predicted off targets (Robert et al., 2009).
The pipeline described here is highly efficient in identifying specific regulators of plant autophagy and optimizing the treatment conditions for plants grown in laboratory settings, thus providing a powerful tool for fundamental plant autophagy research. The assays are not limited to any specific class of compounds; however, researchers utilizing the assays must consider practical limitations of treatment applications to cell cultures or seedlings, including but not limited to solubility, compound stability in the culture medium, and the stress it will inflict on the specimens. Cell cultures used in the primary screen can be cultured under autophagy-stimulating conditions (e.g. starvation) or under autophagy-suppressing conditions (nutrient replete conditions). Depending on the growth conditions, the pipeline can be then used to primarily identify inducers or suppressors of autophagy. In our proof-of-concept experiments, we cultured the cells under nutrient replete conditions that generally lead to a low basal level of autophagy activity. Nevertheless, in our primary screen (Supplemental Fig. S1B), we identified several compounds that were able to suppress constitutive autophagy. For the subsequent characterization of these compounds in the assays of phases 2, 3, and 4, we artificially enhanced autophagy activity in seedlings and cell lines by subjecting them to nitrogen-starvation conditions. Further characterization of the identified compounds can include conventional methods for target identification (Dejonghe and Russinova, 2014; Schenone et al., 2017) and/or direct testing of compounds on plants grown in soil. The first approach will provide an in-depth mechanistic understanding of the modulator’s activity, information valuable for optimization of specificity and potency, as well as predicting its applicability for other species, including animal and human disease models. The second approach might potentially streamline direct application in the field, enabling the use of chemical transient modulators of plant autophagy as potent regulators of crop fitness (Avin-Wittenberg et al., 2018; Minina et al., 2018; Ustun et al., 2018), thus benefiting modern agriculture.
MATERIALS AND METHODS
Genetic Constructs
Maps and sequences of all constructs used in this study can be found in Supplemental Table S2. Information about the primers used for cloning is available in Supplemental Table S3.
To generate the entry clone AM425, the AtATG8a coding DNA sequence (NM_001084955.1) was amplified from total cDNA of Arabidopsis (Arabidopsis thaliana) seedlings using primers AM18/AM19. The obtained attB-flanked amplicon was then recombined with the Gateway pDONR/Zeo vector (ThermoFisher) using Gateway BP Clonase II enzyme mix (ThermoFisher).
Fluc and Rluc coding sequences were amplified using the primer pairs PB31/PB32 and PB33/PB34 and vectors pRD400 35S::Fluc and pRD400 35S::Rluc (Eskelin et al., 2011), respectively. The amplicons were inserted using KpnI/AscI restriction digestion sites into a pMDC32 vector (Curtis and Grossniklaus, 2003) that was previously modified to replace the hygromycin resistance gene with nptII sequence. The resulting binary Gateway-based destination vectors were named AM616 and AM618 (Supplemental Table S2).
AM439 pMDC kan Fluc AtATG8a was obtained by recombining the entry clone AM425 with AM616 pMDC kan Fluc using Gateway LR Clonase II enzyme mix (ThermoFisher).
The synthetic sequence of attB-flanked SV40 NLS (Eurofins Scientific) was recombined with the Gateway pDONRTM/Zeo vector (ThermoFisher) using Gateway BP Clonase II enzyme mix (ThermoFisher), producing entry clone CC3.
The CC3 entry clone was then recombined with the AM618 vector to produce an intermediate clone containing the sequence of 2× 35S::Renilla-NLS fusion::NosT. This sequence was amplified using the 5′-phosphorylated primers CC1 and CC2. To generate the CC8 destination clone, the amplicon was then blunt-end ligated into the AM439 destination clone linearized with the PmeI restriction digestion enzyme (NEB).
The CC10 clone was obtained by introducing two stop codons between the Fluc and AtATG8a utilizing the QuikChange site-directed mutagenesis kit (Agilent) and primers CC7/CC8.
The sequence of TagRFP-mWasabi was amplified from the plasmid kindly provided by Zhou et al. (2012) using the primers PB15/PB16. The amplicon was inserted into pMDC32 or pMDC kan to produce the destination binary vectors AM613 and AM614, respectively. These two vectors were then recombined with the entry clone AM425 using Gateway LR Clonase II enzyme mix (ThermoFisher) to produce AM434 and AM435 clones, respectively.
Plant Material
The description of stable transgenic lines used in this study is available in Supplemental Table S1.
Plant Growth and Transformation
Arabidopsis seeds were surface sterilized using a bleach solution with a final concentration of 2.7 g L−1 sodium hypochlorite (Klorin, Colgate Palmolive) with 0.05% (v/v) Tween 20 for at least 20 min, washed three times with sterile Milli-Q water, and plated on Murashige and Skoog (MS) medium: 0.5× MS medium including vitamins (M0222, Duchefa), supplemented with 10 mm MES (M1503, Duchefa), 1% (w/v) Suc, and 0.8% (w/v) plant agar (P1001, Duchefa), pH 5.8. Seeds on plates were then vernalized in darkness at 4°C for 24 to 48 h. After vernalization, plates were incubated vertically under long-day growth conditions (16 h of 150 µm light at 22°C, 8 h of dark at 20°C).
For growth in soil (2001, Hasselfors Garden), seeds were surface sterilized as described above, vernalized in Milli-Q water in Eppendorf tubes, and then transferred directly into soil. Plants were grown under long-day conditions as described above.
For transformation, Arabidopsis plants were grown in soil for approximately 5 weeks, and the first inflorescence was clipped to stimulate production of auxiliary inflorescences. All genetic constructs used for transformation were first introduced into Agrobacterium tumefaciens strain GV3101 (Van Larebeke et al., 1974). The floral dip transformation was performed as previously described (Clough and Bent, 1998). Transformants were selected on 0.5× MS medium as described above supplemented with 400 µg mL−1 timentin (T0190, Duchefa), 250 µg mL−1 cefatoxime (C0111, Duchefa), and either 35 µg mL−1 hygromycin B (H0192, Duchefa) or 50 µg mL−1 kanamycin (K0126, Duchefa). Lines showing a 3:1 segregation pattern in the T2 generation were used for establishing homozygous T3 lines.
For testing potential inhibitors of autophagy, nitrogen starvation was induced in seedlings by incubating them in liquid nitrogen-free MS medium: 0.5× MS basal salt micronutrient solution (M0529, Sigma), 1% (w/v) Suc, 3 mm CaCl2, 1.5 mm MgSO4, 1.25 mm KH2PO4, 5 mm KCl, and 0.8% (w/v) plant agar, pH 5.8. Seeds were germinated on standard 0.5× MS medium as described above. Seedlings were grown on vertically positioned plates under long-day conditions for 4 to 5 d before being transferred into wells of six-well tissue culture plates with 3 mL of nitrogen-free MS medium. The medium was gently pipetted on top of the roots to ensure their complete submersion. The nitrogen-starvation effect was clearly detectable after 18 h of treatment.
cv BY-2 Cell Culture Growth and Transformation
Tobacco (Nicotiana tabacum ‘BY-2’) suspensions were subcultured under sterile conditions every 7 to 10 d using the standard medium: 1× MS medium including vitamins (M0222, Duchefa), 0.02% (w/v) KH2PO4, 0.2 mg L−1 2,4-dichlorophenoxyacetic acid, and 3% (w/v) Suc, pH 5.6. The cultures were incubated in Erlenmeyer flasks at 22°C in darkness on a gyratory shaker at 120 rpm.
For transformation of BY-2 cells, cultures of transgenic A. tumefaciens GV3101 were grown overnight at 28°C and 180 rpm in 3 mL of Luria-Bertani medium supplemented with 50 µg mL−1 rifampicin and 50 µg mL−1 kanamycin. Bacterial cells from the overnight cultures were spun down for 15 min at 4,000g, resuspended in 2 mL of cv BY-2 medium supplemented with 150 µm acetosyringone (D134406, Sigma-Aldrich), and incubated at 28°C and 120 rpm for 1 h. The bacterial cells were then added to 4 mL of 4-d-old cv BY-2 culture to the final OD600 = 0.375. The mixture was transferred on plates with cv BY-2 medium solidified with 0.8% (w/v) agar. The plates were kept at 28°C for 48 h, after which the cells were gently transferred on plates containing 400 µg mL−1 timentin (T0190, Duchefa), 250 µg mL−1 cefatoxime (C0111, Duchefa), and either 40 µg mL−1 hygromycin B (H0192, Duchefa) or 50 µg mL−1 kanamycin (K0126, Duchefa). The first calli were typically visible after 2 weeks of growth on the selection plates.
Treatment of Arabidopsis Seedlings for Microscopy
For microscopy experiments, treatment of Arabidopsis seedlings with candidate compounds was performed on six-well tissue culture plates. Five-day-old seedlings grown on vertical plates were transferred into wells containing 3 mL of standard or nitrogen-free 0.5× MS medium supplemented with a compound or the corresponding concentration of the vehicle (DMSO). The seedlings were then submerged by gently pipetting the medium on the roots. The plates were sealed with a strip of Saran wrap and incubated in the growth cabinet under long-day conditions for the designated time. For microscopy, the seedlings were mounted in the corresponding treatment medium.
Prior to the viability assay, solubility was assessed in small volumes of MS medium to exclude concentrations at which the compound precipitated. Viability assays were performed using a Zeiss AxioImager 2 epifluorescence microscope equipped with DIC (Nomarski) optics. Images were acquired using ZEN blue software (version 2.1, Carl Zeiss).
Treatment of cv BY-2 Cells
For the primary screen, 3-d-old cv BY-2 cell culture was pipetted onto 96-well plates (140 µL per well) using wide orifice tips. The plates were kept on an orbital shaker (4 mm orbital movement) at 640 rpm, at 25°C for 24 h, to let cells recover from the mechanical stress. One microliter of each compound taken from the plasma membrane recycling inhibitors in pollen (Robert et al., 2008) and plasma membrane recycling compound set A (Drakakaki et al., 2011) libraries was pipetted onto the plates with the cv BY-2 cells. One micromolar AZD8055 and 1 mm benzothiadiazole (BTH) were used as positive controls for up-regulation of autophagy; 1% (v/v) DMSO was used as a negative control. The cells were treated while shaking at 25°C, in the dark for 24 h, after which 50 µL of 4× Passive Lysis buffer (Dual Luciferase Assay kit, E1910, Promega) was added to all wells, and the plates were sealed and frozen at −80°C to rupture the cell walls. To measure the luciferase intensity values, the plates were thawed at room temperature, vortexed for 2 min, and spun down at 4,000g for 5 min. Ten microliters of each cell lysate was used for dual-luciferase assay by applying 50 µL of freshly prepared LARII and Stop&Glo reagents of the Dual Luciferase Assay kit (E1910, Promega). The luciferase intensity values of Fluc and Rluc were measured sequentially by calculating integrated chemiluminescence intensity for a 1 s interval (using gain 240), after a delay of 2 s upon injection of each substrate (LARII and Stop&Glo, respectively), using the Synergy H4 Multi-Mode Microplate Reader (BioTek).
The obtained data were then analyzed using R (version 3.5.2; R Core Team, 2018). The compounds were ranked according to their Rluc-to-Fluc intensity ratio and to their relative luminescence unit (RLU) values for Rluc. The latter were used to infer the viability of the cells, and low values (below 150,000 RLU) were considered to correspond to cytotoxic effects from the treatment. Rluc RLU values produced by dying cell cultures will vary depending on detector sensitivity and therefore should be estimated prior to screening. Compounds inducing more than 2.5-fold increase in Rluc-to-Fluc ratio compared with the DMSO control and Rluc RLU values above 150,000 were selected as putative enhancers of autophagy. Compounds inducing a 2-fold reduction of the Rluc-to-Fluc ratio and Rluc values above 150,000 RLU were selected as putative inhibitors.
Treatment of cv BY-2 cells for further analyses was performed in six- or 24-well plate format in 1 to 3.5 mL of 4-d-old cv BY-2 culture. After transfer onto the plate and before the start of a treatment, the cells were left to recover for 4 to 18 h. During treatment, the plates were kept at 25°C and 120 rpm in the dark. For sampling, the cells from each well were harvested onto a 50-µm nylon mesh, frozen, ground in liquid nitrogen, and resuspended in 1× passive lysis buffer (Dual Luciferase Assay kit, E1910, Promega) using 2 volumes per mg of material. Ten microliters of each cell lysate was then transferred onto another plate and used for dual-luciferase assay applying one-quarter of the recommended volume of each reagent of the Dual Luciferase Assay kit (E1910, Promega). The substrates were added by an automated pump of the FLUOstar Omega Microplate Reader (BMG LABTECH) directly prior to double orbital shaking at 100 rpm for 1 s and then a 2-s reading interval for each well. The gain adjustment was estimated in a pilot experiment using lysates of cv BY-2 cells treated with 1% (v/v) DMSO, 500 nm AZD8055, and 1 µm concanamycin A.
TT Assay
The detailed protocol for the TT assay and the AuTToFlux software can be found at https://github.com/jonasoh/AuTToFlux.
Western Blot
For the western-blot analysis, Arabidopsis seedlings were grown on vertical plates as described above. Briefly, seeds were sown in two rows per plate and the seedlings were grown for 5 to 7 d until the root became 4 to 5 cm long. For treatment, the plates were placed horizontally and flooded with liquid 0.5× MS medium supplemented with the corresponding compound. Treatments were performed in the growth cabinet under long-day conditions. Upon the completion of the treatment, the excess liquid was poured off the plates and the whole seedlings were harvested and ground in liquid nitrogen. Alternatively, the differentiation zone of the root was excised from approximately 50 seedlings using a scalpel blade, gently collected with tweezers, and briefly dried on a paper tissue to be then placed into a 1.5-mL Eppendorf tube and ground in liquid nitrogen with a plastic pestle. The powdered material was mixed 1:2 (w/v) with 2× Laemmli buffer (Laemmli, 1970), and the mixture was thoroughly vortexed and boiled for 5 min at 95°C. Samples were spun down at 13,000g for 10 min, and 1 to 7.5 µL of the supernatant was loaded on precast gels (Bio-Rad). Proteins were transferred onto a polyvinylidene difluoride membrane and blotted using 1:2,000 anti-GFP (catalog no. 632381, Clontech), 1:2,000 anti-actin (AS132640, Agrisera), 1:25,000 anti-rabbit-HRP (AS09602, Agrisera), and 1:8,000 anti-mouse-HRP (NA931, Amersham) antibodies. The blots were developed using ECL Prime (Amersham) and GelDoc XR (Bio-Rad). The intensities of the bands were quantified using Image Studio Lite (version 5.2, Li-Cor) software.
Statistical Analysis and Image Preparation
Statistical analysis was performed using GraphPad Prism (version 7.03, GraphPad Software) unless otherwise stated. Figures were prepared using Adobe Photoshop and Illustrator CC (Adobe). Adjustments to brightness and contrast were applied equally to corresponding images within figures.
Time Requirements
The rate at which compounds can be screened during phase 1 largely depends on the automation system available. Phase 2 required approximately 12 h of active effort per compound (testing three concentrations and three time points per compound), which is spread over 5 d (for details, see TT assay protocol). The most time-consuming part of the TT assay is confocal laser scanning microscopy imaging, which we performed in a semi-high-throughput manner using Zeiss software options, such as position marks. Time required for this phase will mostly depend on the type of microscope available and experience of the user. We suggest to allocate at least 15 min for acquiring images for one time point/concentration of each compound. Phase 3 narrows to a single optimal concentration and time point and required 8 h of active work spread over 3 d for each molecule of interest. We utilized a Bio-Rad western-blotting workflow that included TGX stain-free precast gels and turbo-transfer to reduce preparation work and waiting times, respectively. Phase 4 featured imaging of our library of Arabidopsis endomembrane marker lines (Supplemental Table S1) using confocal laser scanning microscopy and detailed phenotyping, which in total required 20 h of work per compound spread out over 5 to 7 d.
Data Availability
The data relevant to this article will be made available upon personal request. Please note that some of the information about identified modulators of plant autophagy will be made public in upcoming publications.
Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Dual-luciferase reporter verification and results of the primary screen performed in tobacco cv BY-2 cells.
Supplemental Figure S2. TT assay.
Supplemental Table S1. Stable transgenic lines used in this study.
Supplemental Table S2. Constructs used in this study.
Supplemental Table S3. Primers used in this study.
Supplemental Video S1. Examples of phase 4 results.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Alyona Minina (alena.minina{at}slu.se).
A.N.D. performed most of the experiments for phases 2 through 4 and wrote part of the article; C.C. established and performed the primary screen in phase 1 and established transgenic lines; K.D. helped establish the transgenic lines for the study; J.A.O. contributed to designing the scripts for the automated vacuolar fluorescence intensity measurement; S.B.F. contributed to the primary screen in phase 1; S.R. contributed to conceiving the project idea and the primary screen in phase 1; G.R.H. assisted with phase 4 experiments; P.V.B. secured funding and contributed to conceiving the project idea; E.A.M. contributed to conceiving the project idea, supervised the project, performed some experiments, and wrote part of the article.
↵1 This work was supported by grants from Olle Engkvist Foundation (to P.V.B., E.A.M., and S.R.), the Carl Tryggers Foundation (to E.A.M.), MSCA IF (to E.A.M.), the Swedish Foundation for Strategic Research (to P.V.B.), the Swedish Research Councils VR and Formas (to P.V.B.), the Knut and Alice Wallenberg Foundation (to P.V.B.), and by the research program Trees and Crops for the Future at the Swedish University of Agricultural Sciences. The postdoctoral fellowship of A.N.D. was provided by the Natural Sciences and Engineering Research Council of Canada. G.R.H. was supported by the August T. Larsson Guest Researcher Programme for his visits to the Swedish University of Agricultural Sciences.
↵2 These authors contributed equally to the article.
↵4 Senior author.
- Received May 31, 2019.
- Accepted August 14, 2019.
- Published September 5, 2019.