Imaging plant cell death: GFP-Nit1 aggregation marks an early step of wound and herbicide induced cell death
© Cutler and Somerville. 2005
Received: 09 December 2004
Accepted: 29 March 2005
Published: 29 March 2005
Skip to main content
© Cutler and Somerville. 2005
Received: 09 December 2004
Accepted: 29 March 2005
Published: 29 March 2005
A great deal is known about the morphological endpoints of plant cell death, but relatively little is known about its sequence of events and / or its execution at the biochemical level. Live cell imaging using GFP-tagged markers is a powerful way to provide dynamic portraits of a cellular process that can in turn provide a descriptive foundation valuable for future biochemical and genetic investigations.
While characterizing a collection of random GFP-protein fusion markers we discovered that mechanical wounding induces rapid aggregation of a GFP-Nitrilase 1 fusion protein in Arabidopsis cells directly abutting wound sites. Time-lapse imaging of this response shows that the aggregation occurs in cells that subsequently die 30 – 60 minutes post-wounding, indicating that GFP-Nit1 aggregation is an early marker of cell death at wound sites. Time-lapse confocal imaging was used to characterize wound-induced cell death using GFP-Nit1 and markers of the nucleus and endoplasmic reticulum. These analyses provide dynamic portraits of well-known death-associated responses such as nuclear contraction and cellular collapse and reveal novel features such as nuclear envelope separation, ER vesiculation and loss of nuclear-lumen contents. As a parallel system for imaging cell death, we developed a chemical method for rapidly triggering cell death using the herbicides bromoxynil or chloroxynil which cause rapid GFP-Nit1 aggregation, loss of nuclear contents and cellular collapse, but not nuclear contraction, separating this response from others during plant cell death.
Our observations place aggregation of Nitrilase 1 as one of the earliest events associated with wound and herbicide-induced cell death and highlight several novel cellular events that occur as plant cells die. Our data create a detailed descriptive framework for future investigations of plant cell death and provide new tools for both its cellular and biochemical analysis.
Events that compromise the integrity of a single cell can threaten the well-being of an entire multi-cellular organism, a fact reflected by the diverse mechanisms that organisms use to eliminate potentially dangerous cells . For example, when injured beyond repair by high doses of irradiation or by treatment with cytotoxic agents, regulated cell death may be activated to remove damaged cells. This process is best understood in animal cells from extensive analyses of the mechanisms that underpin apoptosis, a regulated process marked by a set of stereotyped changes in cellular architecture that culminate in cell death [2, 3].
In plant cells, regulated cell death occurs in numerous contexts, such as during development of xylem elements  or as part of the hypersensitive response (HR) to pathogen attack [5–7]. Although the hypersensitive response is triggered by highly specific plant-pathogen interactions and is under genetic control [7–9], the HR exhibits many similarities to plant responses triggered by wounding. Both processes activate localized mechanisms for forming protective barriers through lignification, cross-linking of cell wall proteins and other modifications of the extracellular matrix [10, 11], which are thought to limit pathogen access. The HR and wound response display extensive overlaps in their transcriptional responses  and pathogen-derived elicitors of the HR can activate wound-induced kinase activities  suggesting that the two responses share some regulatory components. Thus, both wound and hypersensitive responses activate related defenses local to and transmitted from their primary sites of initiation, but differ in their activating signals.
There has been intensive research on the signaling mechanisms that regulate the myriad responses elicited by wounds and pathogens. Several lines of evidence suggest that the localized and transmitted components of both responses are regulated, in part, by hydrogen peroxide-mediated signaling events [14–17]. It has been proposed that H202 is a broad spectrum signaling molecule that triggers local processes like cell wall protein cross linking  and cell death  as well as long distance effects such as gene induction . Although the precise mechanisms by which H202 signals are initiated locally and then transmitted are incompletely known, they involve activation of an NADPH oxidase that generates H202 via superoxide production , analogous to the oxidative defenses mounted by macrophages of the mammalian immune system. NO signaling also appears to participate in this oxidative response [19, 20].
In contrast to the signal-mediated events that trigger these responses, relatively little is known about the downstream events that execute the orderly patterns of cellular deconstruction that accompany cell death. In animal cells many of the characteristic events are attributed to the activity of caspase proteases, which initiate and execute a cascade of proteolytic events that participate in subcellular deconstruction . Parallels between plant cell death and animal apoptosis have been suggested from observations of cellular contraction, nuclear contraction and fragmentation of DNA during HR cell death. Plants have a family of caspase-related proteins, designated as metacaspases, and numerous studies have implicated caspase-like proteases in the control of cell death activation in plants . Recently, a vacuolar protease unrelated to caspases at the amino acid sequence level has been found to posess caspase protease activity and has been demonstrated to be required for viral induced hypersentive cell death in tobacco , showing that caspase activities mediate some components of plant cell death. Nonetheless, the targets of this and other caspases that mediate the changes in subcellular architecture during plant cell death still remain elusive.
In the course of a screen for useful GFP fusion proteins , we observed that a GFP-Nitrilase 1 fusion protein exhibited a change of aggregation state in response to wounding. In order to place this phenomenon in a larger context, we initiated a series of live-cell imaging experiments to identify and characterize markers associated with the execution of physically and chemically induced cell death. Our studies reveal a series of previously undescribed events associated with wound-induced cell death in plants and raise new mechanistic questions regarding the execution of plant cell death.
In addition to nuclear contraction, a wound-induced decrease in the intensity of GFP nuclear fluorescence in the N6 marker line was evident within less than 20 min of wounding (Fig. 3A). Several explanations for the decrease in nuclear fluorescence seemed possible, including GFP destruction, alterations in pH that quenched fluorescence, or the expulsion of nuclear contents into the cytoplasm. While these possibilities are not mutually exclusive, the latter seemed particularly attractive since contraction might cause extrusion of nuclear contents. The N6 marker possesses a low background level of cytoplasmic GFP localization that might mask the release of nuclear contents and also may be partially immobilized from the bulk nucleoplasm by an association with DNA (this marker illuminates chromosomes during mitosis, ). Consequently, we examined wound responses using the nuclear marker line N7, which displays a lower level of background cytoplasmic fluorescence and does not appear to be associated with chromatin as assessed by time-lapse imaging of mitoses (data not shown). Hypocotyls of plants carrying the N7 marker were wounded and imaged. Time-lapse imaging of cells proximal to wound sites revealed that a gradual illumination of the cytoplasm occurs concomitantly with the decrease in nuclear fluorescence and nuclear contraction, suggesting that the nuclear label is lost or expelled into the cytoplasm (Fig. 3B). The simultaneous illumination of nucleoplasm and cytoplasm facilitated visualization of the formation of lobes on the nuclei, revealed by negative contrast, as the interior of these lobes excludes the GFP label (See arrows in Fig. 3).
Based on our observations that the inter-lobal space excludes cytosolic GFP, contains the lumenal ER marker mGFP5, and that the lobes are demarcated by a membrane system contiguous with the ER, we believe that the simplest explanation for the observed nuclear lobing is that the inner and outer membranes of the nuclear envelope separate during nuclear contraction. Thus, nuclear contraction does not reflect a contraction of the entire nucleus proper, but rather a contraction of the nuclear lumen and its tightly associated inner envelope.
The earliest events after wounding were fusiform body alterations, aggregation of GFP-Nit1, and nuclear hypertrophy. Following herbicide treatment, we observed that fusiform body alterations preceded GFP-Nit1 aggregation (data not shown). Shortly after these early events, nuclear contraction ensues with a concomitant separation of the nuclear envelope and nucleoplasm leakage into the cytosol. It is currently unknown if this reflects a generalized loss of nucleoplasm or if there is specificity to the contents lost. Later, after nuclear contraction has largely ceased, the remaining tubular ER undergoes a dramatic loss of integrity and degenerates, forming vesicular structures. Intense staining of nuclei by propidium iodide and extensive cell shrinkage follows this. Since propidium iodide is a charged fluorophore that is largely precluded from entering vital cells, we interpret this late event to represent a major degeneration in the integrity of the plasma membrane and consider this the likely point of cell death, although this is an operational definition. Observations of a plasma-membrane marker suggest that the plasma membrane is intact at a gross level during the early stages of the wound response, although atypical involutions and extreme photo-sensitivity have been observed (data not shown).
The events we documented occur on a relatively rapid time scale. Typically, within minutes of wounding we observed detectable effects on the shapes of nuclei, fusiform bodies and the aggregation of GFP-Nit1. Within one to two hours of wounding, cells displayed intense staining of their contracted nuclei by propidium iodide. Although we did not obtain evidence for programmed cell death (PCD), this rapid response is compatible with evidence that plant cells undergo PCD by constitutively expressed molecular machinery . It seems plausible that, by limiting possible pathogen access through a wound site, wound-induced cell death functions for the benefit of the organism as a whole.
Although we are far from a mechanistic picture of these events, our observations show that nuclear contraction can be uncoupled from cellular collapse. The known effects of benzonitrile herbicides on mitochondrial function may suggest that nuclear contraction is energy dependent. Since cellular collapse occurred after chloroxynil treatment, this may suggest that contraction is energy independent, perhaps unsurprising, since turgor pressure is generated in part through vacuolar ATPases and proton pumps.
Our focus in this study was on placing the aggregation of GFP-Nit1 in the context of cellular responses during wound-induced cell death. We could not identify any precedent in the literature for a similar transformation. We do not understand the mechanism by which GFP-Nit1 is converted from a soluble protein to a granular form. The observation that nitrilase 1 undergoes a change in sedimentation coefficient after chemically-induced injury in wild type plants suggests that the phenomenon is not an artifact of the GFP fusion protein. We do not know if nitrilase 1 is the only protein that exhibits such properties. However, we did not observe a similar response during the characterization of several hundred other GFP-protein fusions  (and unpublished results). The observation that the GFP-Nit1 aggregates have a relatively constant size that is roughly comparable to that of a secretory vesicle may suggest that the basis of aggregation is encapsulation or binding of nitrilase to a vesicular type of structure. However, the observation that treatment with Triton X100 does not disperse the aggregates indicates that they are not simple membrane-bound vesicles.
Wounding and benzonitrile herbicides both induce rapid forms of cell death that yield collapsed cells with disrupted plasma membranes. One of the earliest responses in both of these cell death systems is the redistribution of Nit1 into aggregates that are visible in vivo as GFP-Nit1 granules or biochemically as a low-speed pelletable form of Nit1. Nit1 aggregation precedes nuclear contraction and cellular collapse, two classic cytological features of plant cell death. The use of GFP-Nit1 and other markers to image plant cell death in vivo has revealed novel subcellular responses that to our knowledge have not been previously described, such as nuclear envelope separation, formation of nuclear lobes and release of nuclear contents into the cytosol. Our data have enabled a new detailed descriptive model for plant cell death and raise new mechanistic questions.
Plants were grown on agar-solidified media consisting of 0.5X MS salts (Gibco-BRL) and 0.8 % Agar (Research Organics). Prior to germination, seeds were chilled for 4 days on MS plates at 4°C then transferred to a growth cabinet and grown under continuous illumination at 300 μE m-2s-1. Unless otherwise stated, imaging experiments were performed on whole 4 – 8 day old seedlings mounted in 0.5X MS or an imaging buffer (IMB) composed of 0.5X MS salts, 25 μg/ml propidium iodide (Sigma) and 0.01% Silwet (Lehle Seeds, Tucson, AZ), which was added to facilitate propidium iodide penetration of the epidermal cuticle. To prevent specimen drift during time lapse experiments, whole seedlings in 0.5X MS were mounted by adding 2 volumes of molten 2% low melting point agarose (Research Organics), 2% high resolution 3:1 agarose (FMC) and 0.5X MS salts (Gibco-BRL) at 42°C. Cells showed active streaming after mounting, suggesting minimal stress induced by the mounting process.
Most of the transgenic lines used for imaging in this paper have been described previously [24, 37], but are described here for clarity. For GFP-cDNA lines, experiments were performed on T3 seed derived from the primary transgenic plants. Unless indicated otherwise, these lines contained a single GFP-cDNA fusion. The GFP fusions are available from the Arabidopsis Biological Resource Center (ABRC) at Ohio State University http://Arabidopsis.org.
The N7 marker line (ABRC #CS84731) contains a fusion protein between GFP and the carboxy terminus of an ankyrin-repeat containing transcription-factor-like protein (Genbank Accession CAA16704). The N6 marker line (ABRC #CS84815) contains a near full-length fusion between an HMG-delta protein (Genbank accession Y14075) and GFP. In dividing cells it illuminates chromosomes aligned along the metaphase plate and thus likely associates with chromatin (26). In interphase cells, it illuminates the nucleoplasm. This line also contains a second PCR-detectable insert that contains an out-of-frame cDNA fusion to GFP.
The mGFP5 line was generated by Wolf Scheible by transformation of the mGFP5 construct on plasmid pmGFP5-ER  into the Columbia ecotype of Arabidopsis. The Q4 ER membrane marker line (ABRC #CS84728) is a fusion between GFP and a novel protein with a predicted carboxy-terminal trans-membrane (Genbank Accession AAB71445).
The GFP-Nit1 marker comprises a full-length nitrilase 1  (Genbank Accession U47114) fused to the C-terminus of GFP. Two lines were used in this study 35S-GFP-Nit1, in which the fusion protein is driven by the 35S promoter and, N1P2E, in which expression is driven by 1.8 kb of sequence upstream from the Nit1 start codon . These two different lines possess similar expression levels in transgenic plants, however in comparison to 35S-GFP-Nit1, the N1P2E line shows reduced expression in root epidermal cells and guard cells
Agarose mounted seedlings were wounded by creating cuts through plant tissue with a razor blade or sharp forceps tips. Image data was collected after a brief (45 – 60 second) period of aligning the wound site with the field of view. During and prior to imaging experiments, agarose embedded specimens were covered with a humidifying dome to prevent desiccation. The majority of image series were obtained by collecting 20 – 25 μM deep Z-series (typically 15X 1.5 μm z-steps) for 30 – 60 time points at 120 sec intervals. 3-D time series were made from these data sets by making brightest-point reconstructions of each Z-series using the BioRad software package LaserSharp (BioRad, Hercules, CA). For some time series, reconstructions were performed manually in NIH image using brightest point reconstructions (Wayne Rasband, RSB, NIH, Bethesda Maryland). A Nikon inverted fluorescence microscope equipped with a Nikon 60X 1.2 numerical aperture water immersion objective and a Bio-Rad MRC 1024 confocal head with a krypton-argon laser. EGFP was excited at 488 nm and emitted fluorescence was collected through a 525-30 band pass filter. Chloroplast autofluorescence and propidium iodide fluorescence were obtained by excitation at 568 nm and collecting emitted fluorescence through a 596 – 615 nm band pass filter.
Chloroxynil has a higher solubility in ethanol and water than bromoxynil and was used in imaging experiments. Both herbicides show similar toxicity to Arabidopsis using a germination assay . A 1 M stock of Chloroxynil (ChemServices, West Chester, PA) was made in ethanol. For imaging experiments, a 1 mM solution of Chloroxynil in IMB was made and 100 μl of this solution was overlaid onto to seedlings mounted in 100 μl of agarose, yielding a final concentration of 500 μM.
Recombinant Nit1 was produced in E. coli as a His-tagged thioredoxin fusion protein using the vector pET32(A) (Novagen, Madison, WI) and purified by chromatography on a nickel-chelate resin, (Novagen, Madison, WI). Polyclonal serum against recombinant Nit1 protein was made by a commercial provider (Cocalico, Reamstown, PA). The Nit1 antibodies were affinity purified against 2 mg of purified recombinant Nit1 adsorbed onto a 6 × 8 cm piece of nitrocellulose membrane and eluted off the membrane by treatment with 200 mM glycine, 150 mM NaCl pH 2.0. The eluted antibodies were neutralized to pH 7.0 with Tris base, dialysed overnight against Tris-buffered saline and concentrated to 2 mg/ml using a spin column concentrator with a 15 kDa cutoff, the concentrated antibody was diluted to 1 mg/ml with glycerol.
Herbicide-treated plants were transferred into a 3X-weight volume of ice-cold granule isolation buffer (GEB, 400 mM sucrose, 75 mM KCl, 50 mM PIPES pH 6.9, 10 mM MgCl2, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml aprotinin). They were then sonicated, filtered through Miracloth and Triton-X 100 was added to a final concentration of 4%. After 10 min incubation on ice, the extract was centrifuged at 10,000 g for 10 min. The supernatant was removed and retained and the pellet was washed by with GEB + 4% Triton-X 100. Protein from the supernatant fraction was precipitated using a chloroform-methanol extraction protocol . The pellet fraction and precipitated low-speed supernatant fractions were resuspended in equal volumes of SDS-PAGE sample buffer, typically 500 μl per 100 mg input tissue (fresh weight).
For herbicide induced-pelleting, 2–3 week old Arabidopsis seedlings were transferred into a solution of ddH20, 0.01% Silwet and varying doses of Bromoxynil. After 1 h incubation at room temperature with gentle shaking, seedlings were transferred to an eppendorf tube, placed on ice and disrupted by sonication in GEB as described above.
SDS-PAGE was performed using 8 – 20 % gradient mini-gels, prepared and run as described in Ausubel et al. . For western analysis, proteins were transferred onto nitrocellulose membranes using a semi-dry electroblotter in Towbin transfer buffer . Westerns were developed using Pierce SuperSignal HRP substrate (Pierce, Rockville, IL). The cofillin and Pip2A used as control antibodies were obtained from Rose Biotechnology  and used at 1/2000 dilutions. The Nit1 antibodies were used at 1/1000 dilution.
We thank David Ehrhardt for assistance with confocal microscopy and for fruitful discussions. We also thank Farhah Assaad for valuable comments on the manuscript and figures. This work was supported by a grant from the U.S. Department of Energy (DOE-FG02-00ER20133).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.