Organized Disassembly of Photosynthesis During Programmed Cell Death Mediated By Long Chain Bases

In plants, pathogen triggered programmed cell death (PCD) is frequently mediated by polar lipid molecules referred as long chain bases (LCBs) or ceramides. PCD interceded by LCBs is a well-organized process where several cell organelles play important roles. In fact, light-dependent reactions in the chloroplast have been proposed as major players during PCD, however, the functional aspects of the chloroplast during PCD are largely unknown. For this reason, we investigated events that lead to disassembly of the chloroplast during PCD mediated by LCBs. To do so, LCB elevation was induced with Pseudomonas syringae pv. tomato (a non-host pathogen) or Fumonisin B1 in Phaseolus vulgaris. Then, we performed biochemical tests to detect PCD triggering events (phytosphingosine rises, MPK activation and H2O2 generation) followed by chloroplast structural and functional tests. Observations of the chloroplast, via optical phenotyping methods combined with microscopy, indicated that the loss of photosynthetic linear electron transport coincides with the organized ultrastructure disassembly. In addition, structural changes occurred in parallel with accumulation of H2O2 inside the chloroplast. These features revealed the collapse of chloroplast integrity and function as a mechanism leading to the irreversible execution of the PCD promoted by LCBs.

www.nature.com/scientificreports www.nature.com/scientificreports/ chloroplast 21 . Moreover, it has been recognized that photosynthetic activity is interrupted upon pathogen infection [22][23][24] . All this diverse information suggests that the chloroplast and its photosynthetic activity are involved in the PCD process.
In parallel, evidences dealing with the PCD-HR mediated by LCBs suggest an association with the chloroplast: (1) PCD elicited by fumonisin B1 (FB1, a mycotoxin that evokes LCB accumulation 25 ) is light-dependent 16 ; (2) PCD induced by FB1 and mediated by MPK6 11 promotes extensive chloroplast damage and induces H 2 O 2 formation inside the chloroplasts 11,19 . However, a study of the chloroplast and its function during HR-PCD that is mediated by LCBs has not been specifically addressed.
Therefore, these and other unknown events dealing with the chloroplast need to be identified and positioned in a single sequence in order to establish a clear functional context. For this reason, we have induced PCD in Phaseulus vulgaris using two LCB eliciting treatments: FB1 and a non-host pathogen that induces HR-PCD. Then, we explored if LCB accumulation took place in both treatments and measured some biochemical responses such as reactive oxygen species formation and MAP kinases activation. Then, we analyzed the chloroplast ultrastructure and the functioning of its light-dependent reactions. Here, we provide direct evidence that there is a direct organized and irreversible collapse of the chloroplast that leads to the PCD mediated by LCBs.

Biological material. Phaseolus vulgaris var. Canario (common bean) plants were grown in agrolite, watered
with Hoagland solution and maintained under a natural photoperiod at 28 °C in a greenhouse. Pseudomonas syringae pv. tomato DC3000 avrRPM1 (Pst), a strain that elicits a defense response in Phaseolus, was cultured in solid B King medium with rifampicin and tetracycline at 50 µg/ml. Fresh cultures were resuspended in 10 mM MgCl 2 and then leaf-infiltrated.
Samples at the infiltration sites were taken in the interval from 0 to 48 h after treatments as indicated. Treatments with MgCl 2, Silwet L-77, FB1, Pst, SN or SA were performed while the leaves were attached to the plant and then all measurements were carried out. A photographic record of the evolution of leaves upon different treatments was followed. electrolyte leakage assay. Leaf disks of 1.0 cm diameter containing the infiltration points were cut at every infiltration time, weighed and electrolyte leakage was determined with a conductimeter Conmet1, Hanna Instruments (Woonsocket, RI) 27 . Briefly, leaf disks were submerged in double distilled water under moderate stirring and medium conductance was measured at several times. Electrolyte leakage was expressed following the formula ES = EC1/EC2 × 100, where ES corresponds to final conductivity of the sample, E1 corresponds to the electrical conductivity measured at 24, 48 and 72 h and EC2 corresponds to the conductivity measured at the end of the experiment when leaf disks were boiled to release total electrolytes into the medium.

FB1 and
Pseudomonas syringae pv. tomato induce similar responses associated to a pcD-HR in Phaseolus vulgaris. In order to confirm that Phaseolus was FB1-sensitive as it occurs in many, but not all species 33,34 , leaves were infiltrated with 5 to 50 μM FB1 (Fig. 1a). It was observed that the toxin produced a delimited spot the size of which was concentration-dependent (Fig. 1a, left leaf) and started to be visible 24-48 h after 10 μM FB1 infiltration. Since non-host species can produce a hypersensitive response (HR) as part of a defense reaction, we infiltrated 10 5 -10 8 CFU/ml of a suspension from Pseudomonas syringae pv. tomato (Pst), a non-host pathogen, to bean leaves. Figure 1a (right leaf) shows that Pst suspension developed a necrotic zone the size of which was pathogen concentration-dependent and that was visible at 18-24 h after infiltration. Dead tissue remained circumscribed to the site of inoculation with FB1 or Pst, as it occurs in the classical HR and not  www.nature.com/scientificreports www.nature.com/scientificreports/ in necrosis processes. As a control, infiltration of MgCl 2 developed only a light circle caused by the slight wound (Fig. 1a, right side in both leaves). In order to corroborate that the FB1 and Pst treatments were inducing cell death, solute leakage was measured. Figure 1b shows that both, toxin and pathogen induced electrolyte release, but the kinetics and extent were faster and higher, respectively, with the Pst treatment as compared to the one elicited by the toxin. elevation of LcBs. FB1 elicits PCD and defense reactions through the elevation of LCBs as second messengers 11,16 . However, elevation of some sphingolipid species could not be detected in other non-host case 8 . Since in our system FB1 could be inducing a PCD related to an increase in endogenous LCBs, we measured LCB levels in the FB1-and Pst-infiltrated leaves. Figure 2a shows the determination of endogenous phytosphingosine, a LCB, upon FB1 or Pst infiltration. We observed phytosphingosine increases of 1.5 to 3-fold along the infiltration time course with either treatment. Although the kinetics and maximal extent of the LCB elevation were not identical for FB1 and Pst, there were coincident times of phytosphingosine rise (0.5 and 24 h), showing a 2-to 3-fold www.nature.com/scientificreports www.nature.com/scientificreports/ increase over the LCB levels measured in control leaves. These results indicated that both treatments induced LCB elevation in Phaseolus leaves. H 2 o 2 generation. Another known response in plant immunity is the early generation of H 2 O 2 17,35,36 . For this reason, we measured this reactive oxygen species with a soluble assay in both PCD inducing treatments (Fig. 2b). The results showed that FB1 and Pst treatments induced an apparent significant generation of H 2 O 2 at 8 h upon FB1 infiltration and at 4 and 8 h upon Pst infiltration. However, statistical significance was not supported by tests (see Fig. legend).
MApK activation. MAPK activation has been demonstrated to be downstream LCB surge in Arabidopsis PCD 11 . In order to get further support to the idea that the two treatments (FB1 and Pst) that induced PCD-HR were fully comparable in terms of the LCB elicitation, we investigated the possible activation of MAPKs upon FB1 and Pst infiltration of P. vulgaris leaves (Fig. 2c). In addition, we tested MAPK activation upon infiltration of SN, a LCB that elicits PCD-HR and that is precursor of phytosphingosine 11 . We detected the activation of a 48 kDa MAPK with the three treatments (Fig. 2c). The MAPK nature of the phosphorylation was corroborated by the specificity of the phosphorylated substrate i.e., myelin basic protein but not of histone or casein ( Supplementary  Fig. S1a,b) and the recognition of the 48 kDa band by antibodies directed against two typical MAPK, ERK1 and ERK2, and to their phosphorylated forms ( Supplementary Fig. S1c). MgCl 2 infiltration showed an activation of the same MAPK at 5 and 10 min that probably involved the slight wound effect due to the infiltration procedure ( Supplementary Fig. S2a). While FB1 and SN showed a MAPK activation at 45 and 240 min, Pst displayed a sustained activation as long as 480 min after infiltration (Fig. 2c, Supplementary Fig. S2a-d). The molecular mass of the detected MAPK was similar to the AtMPK6 from Arabidopsis thaliana and to the NtSIPK from Nicotiana tabaccum 37,38 . The band, strongly phosphorylated in the presence of FB1, Pst and exogenous SN was dramatically labeled upon SA treatment (  Chloroplast fluorescence is compromised upon FB1 or Pst treatments. H 2 O 2 could be formed in the chloroplast as a result of impairment of the photosynthetic function. Our first approach to explore the functional status of the chloroplast treated with FB1 or Pst was to examine chloroplast structure using its autofluorescence by confocal microscopy (Fig. 4). Figure 4a shows images recorded at 8, 18 and 24 h after MgCl 2 , FB1 or Pst treatments. As it can be observed, chloroplasts from Phaseolus leaves that were MgCl 2 -infiltrated showed similar morphology, distribution and emission of chlorophyll fluorescence throughout this time course. The FB1 treated plants (Fig. 4a) showed an apparent decrease in the chlorophyll fluorescence signal at 24 h, but the graph in Fig. 4b shows that dispersion of data gave no significant difference between the control and FB1 fluorescence intensity values at 24 h. In contrast, Pst-treated plants (Fig. 4a, right column) showed a clear decrease of the fluorescence emission at 18 h (Fig. 4b). This decrease was more evident at 24 h when the signal was very low. These results can be explained by impairment of the photosynthetic activity of the treated leaves. To analyze this hypothesis further, the fluorescence emitting area and the fluorescence intensity per pixel were measured. While the first parameter correlates with the number of functional chloroplasts (Fig. 4b), the latter is an indication of the chloroplast fitness to perform photosynthetic activity per area unit (pixel) (Fig. 4c). Both parameters were unchanged in the chloroplasts from control plants. In the case of FB1-treated plants, a significant decay in the fluorescence emitting area was observed after 24 h (30% lower than in MgCl 2 -treated plants) (Fig. 4b). In Pst-treated plants, a clear decay in the fluorescence emitting area was observed at 18 h (64% lower as compared to the MgCl 2 -treated plants), with no significant decrease at 24 h (70% lower than MgCl 2 -treated plants) (Fig. 4b). However, neither FB1-or Pst-treated plants showed a significant decay in the fluorescence intensity per pixel as compared to the control (Fig. 4c). Thus, in the case where the photosynthetic activity was compromised, it could primarily be associated to a decrease in the number of functional chloroplasts with FB1 and or Pst, rather than with an overall decay in the activity of each chloroplast.

Early and similar impairments of the photosynthetic activity are caused by FB1 and Pst treatments.
While autofluorescence measurements were a good indicator of photosynthetic impairment, we explored the effect of FB1 and Pst on the photosystem II (PSII) activity. Thus, we measured the chl a fluorescence induction curve (OJIP curve) between 0 h (pre-infiltration sampling) and 48 h (0, 8, 18, 24 and 48 h). Figure 5a-c shows the average curves of fluorescence emission. No significant changes in the control leaves (MgCl 2 -infiltrated, Fig. 5a) www.nature.com/scientificreports www.nature.com/scientificreports/ were found as indicated by the low scattering of the curves, especially in phases O-I, at all infiltration times under study. This implied that the MgCl 2 infiltration produced only slight perturbations in the photosynthetic activity at the time span of 48 h. However, Fig. 5b,c clearly showed, respectively, that the FB1 and Pst infiltration produced impairments in the photosynthetic activity as a function of the treatment time.
We compared the change in the fluorescence intensity at O, L, K, J, I and P steps as a function of the time of exposure to MgCl 2 (Fig. 5d), FB1 (Fig. 5e) or Pst (Fig. 5f). Figure 4e shows that the earliest apparent decrease on chl a fluorescence occurred at 12 h with FB1 and at 8 h with Pst. In the case of prompt chlorophyll fluorescence, the first statistically significant decrease took place at 24 h for FB1-treated plants while with Pst treatment it occurred after 12 h (Fig. 5f). JI and IP phases presented the major changes in both treatments at 24 and 48 h but were more pronounced with Pst. Statistical significance was performed for all curves by multivariate hierarchical clustering analysis (Fig. 5g-i). The decrease in the early decay of the J step with Pst can be interpreted as a decrease in the capacity to perform QA reduction (photochemical phase of PSII), meanwhile the decrease in the JI and IP phases could reflect loss of electron transfer between PSII to Photosystem I (PSI). Therefore, the most affected part of the photosynthetic process that was associated to LCB surges was the electron transport.  www.nature.com/scientificreports www.nature.com/scientificreports/ In order to know the state of the energy flow parameters, OJIP curves were analyzed by using the JIP-test model 39,40 . These parameters are presented in the form of radar plots (Fig. 6) and were calculated using the average curves in Fig. 5a-c. The comparison of data from leaves treated with FB1 or Pst for every time, i.e. 8 h, 18 h, 24 h and 48 h (Fig. 6) revealed a striking similarity in the magnitude and pattern of effects on JIP-test parameters. In the leaves exposed to FB1 and Pst for 18 h (Fig. 6), the total performance photosynthetic index (PI tot ) parameter was the most negatively affected. According to the PI tot calculation, the decrease was mainly due to a decrease in the absorption per reaction center component (ABS/RC), i.e. an alteration of the ratio of the absorbed energy and the number of functional reaction centers. Therefore, there was a decrease in the number of functional PSII units. This result was in agreement with the information given by the images of confocal microscopy (Fig. 4a,b). In addition, the values of the yield of energy dissipation per reaction center (DI/ABS and DI/RC, respectively), dramatically enlarged upon 18 h of FB1 or Pst exposure.
Another photosynthetic parameter negatively affected after 18 h of exposure to FB1 or Pst was the electron carrier pool per reaction center (EC/RC). Such result was expected as the JI and IP phases were the most impaired. Based on the JIP algorithm analysis, increases in LCB were due to either the loss in the total pool of electron carriers or to the compromised turnover capacity of PSII, reducing the carrier molecules.
Together, these results suggest that both experimental treatments decreased the number of functional PSII reaction centers and the electron carriers.

Disconnection between PSII and PSI is promoted by FB1 or Pst treatments. The OJIP-test sug-
gested that one of the major impairments of photosynthesis during PCD elicited by LCBs was the flow of electron carriers per active reaction center. This can be directly determined by measuring the linear electron flow between PSII and PSI by reflectance changes at 820 nm. Modulated light reflection (MR) is an approach to assess the oxidation state of P700 + and PC + 41-43 . MR was measured in parallel to the chl a fluorescence at pre-infiltration (0 h) and 10, 24 and 30 h after MgCl 2, FB1 or Pst treatment (Fig. 7). MR experimental curves were normalized to the values at 0 h (pre-infiltration) for each condition to simplify the comparison.
MR determinations of leaves infiltrated with MgCl 2 and with 0 h exposure to FB1 (Fig. 7a) or to Pst (Fig. 7b) leaves showed that plants behaved in similar manner. The first phase, a fast phase (MR fast ), occurring in approximately a time span of 0.001 and 0.01 s, reflects the levels of P700 + and PC + (the oxidized forms). The second phase, a slow phase (MR slow ) resolved in a span of 0.01 to 0.1 s, showed the progressive reduction of P700 + and of PC due to the electrons coming from PSII, thus illustrating the connectivity of the electron transport between the two photosystems. Hence, this signal correlates with the PSII functionality and the activity of the electron transport chain between the two photosystems. www.nature.com/scientificreports www.nature.com/scientificreports/ Fig. 7a. The fast initial phase of the MR signal was characterized by a decrease of the MR values in the span from 300 µs to 30 ms, when P700 + and PC + presented maximum accumulation. This was followed by a slow (400-500 ms) positive phase caused by the gradual reduction of P700 + . However, the curves of the leaves treated with FB1 for 24 and 30 h showed a 40% decrease of the MR fast phase compared to the 0 h curve, indicating that the total accumulation of PC + and P700 + was affected by the FB1 treatment. Additionally, the MR slow was significantly affected at 30 h by the FB1 treatment. This could be explained by a loss in the efficiency of the electron transport chain to re-reduce the pool of P700 or/and PC upon the actinic illumination. However, plants treated with FB1 after 24 h showed a maximum value of the MR recovery after the fast phase, indicating that the electron transport between PSII and PSI units was still functional but was lost later after 30 h of FB1 exposure.

Determination of MR in FB1-treated plants is shown in
In the case of the pathogen treatment, the Pst control (0 h) displayed a similar curve to the FB1 control (Fig. 7b). MR slow phase at 10 and 24 h showed a decrease in total amplitude of 40% and 50%, respectively, when compared to the 0 h curve. Contrary to the FB1 curve at 24 h, the Pst curve showed that the re-reduction of P700 after 0.03 s almost completely disappeared. This meant that the electron transport between the two photosystems was impaired, indicating loss of the electron transport capacity between PSII and PSI. Step O was measured at 50 μs, L at 150 μs, K at 300 μs, J at 2000 μs, I at 30,000 μs and P when fluorescence reached its maximum. (g-i) Statistical differences for each time point were estimated by multivariate hierarchical clustering analysis for MgCl 2 (g), FB1 (h) and Pst (i). Values of the O, L, K, J, I and P were used to estimate the Euclidian distance and were clustered by group average. The longer the distance in a cluster, the more different two time points were. To estimate the threshold of significance, the maximum distance in MgCl 2 data set was used as reference. Red dotted line describes the threshold for significance, in such case a node situated at the left of the red line is considered not significant. OJIP plots were constructed from data acquired with the HandyPEA instrument as described under Materials and methods section and processed by the Biolyzer 3.0 software, Bioenergetics Laboratory, Geneva, Switzerland.

Chloroplast integrity and ultrastructure are compromised upon FB1 and Pst treatments.
Changes in photosynthetic functionality is associated with changes in grana stacking 44 . This is manifested by alteration of the efficiency of the electron transport between PSII to PSI 45 , the collapse of the proton gradient 46 , and the impairment of the lateral heterogeneity of the thylakoid membrane components 47 . In order to explore this possibility, the details of the chloroplast structure were investigated by transmission electron microscopy (Fig. 8b,c Supplementary Fig. S3). The control samples showed typical elongated chloroplasts with intact and well stacked thylakoid membranes and very well defined outer and thylakoid membranes (Fig. 8a). In contrast, chloroplasts from the FB1-and Pst-treated leaves (Fig. 8b,c; Supplementary Fig. S3) showed a rounded shape, very highly unstructured grana and thylakoid membranes at different states of disintegration, suggesting that a progressive process of chloroplast destruction was taking place. It was clear that FB1 and Pst treatment induced alterations of the electron transport capacity of PSII that were already detectable at 8 h (Fig. 6) and progressed until a visible macroscopic phenotypical alteration appear.
We calculated the so-called L-band from the average OJIP curves in Fig. 5a-c, as an additional way to confirm that the photosynthetic impairment was caused by alterations in thylakoid ultrastructure. In case of alteration in PSII unit connectivity, a positive band between 20 µs and 300 µs (L-band) [48][49][50] should appear in the OJIP transient. Figure 8d,e shows relative changes in fluorescence between 20 µs and 300 µs (ΔW OK ), where the L-band occurs. In both figures, it was possible to observe a clear positive band at 24 h for FB1 and 18 h for Pst. This confirms the observed disconnection between PSI and PSII (Fig. 6). Altogether, MR, TEM and L-band analysis confirmed that impairment of photosynthetic activity induced by FB1 and Pst led to disruption of thylakoid organization.

Discussion
In this work, we have studied how two independent treatments that induce PCD elicited by LCBs have the same effect over the function and structure of the chloroplast. Given the role of this organelle on cell function, these results suggest that the chloroplast decay during PCD is an effective and irreversible mechanism that leads to cell death as a defense strategy against non-host pathogens. In detail, we show that the photosynthetic failure  Fig. 5 were used to calculate OJIP-test parameters. All parameters were normalized to the control value (pre-infiltrated plants) and represented in the radar plots shown. PI(abs) refers to photosynthetic performance index on absorption bases. PI(tot) refers to total photosynthetic performance index. ABS/RC refers to absorption per active PSII reaction center. TR/RC refers to trapping probability per active PSII reaction center. ET/RC refers to the probability of electron transport beyond Q A per active PSII reaction center. RE/RC refers to the probability of an electron to be used in the reduction of final acceptors per active PSII reaction center. DI/RC refers to proportional energy dissipated by heat or fluorescence per active PSII reaction center. EC/RC refers to the rate of turnover by electron carriers per active PSII reaction center (which is estimated as the complementary area over the OJIP curve). RC/ABS is reciprocal of ABS/RC (see above). TR/ABS refers to the quantum yield for primary photochemistry. ET/TR refers to the yield for electron transport. RE/ET refers to the probability of an electron traveling through the electron transport chain to be used in the reduction of end acceptors. RE/ABS refers to the total quantum yield of photosynthesis. DI/ABS refers to quantum yield for energy dissipation. Note that OJIPtest parameters were calculated from average OJIP curves in Fig. 5, hence the statistical significance of these parameters is based on Fig. 5g-i. The OJIP parameters shown in the eight radar plots were obtained from the data in Fig. 5 using the Biolyzer 3.0 software, Bioenergetics Laboratory, Geneva, Switzerland.
www.nature.com/scientificreports www.nature.com/scientificreports/ is preceded by LCB increase and MAPK activation and involves ROS formation. This indicates that the loss of functionality of the chloroplast is part of the mechanism by which PCD occurs and is one event in the non-host immunity manifested in the HR.
In order to demonstrate the relation between all these events, the experimental framework consisted of a comparison of Pst and FB1 effects on chloroplast function in Phaseolus vulgaris. Pst (Pseudomonas syringae pv. tomato) is a bacterial non-host pathogen for Phaseolus vulgaris, while FB1 is an inducer of LCB elevation that leads to PCD 11,16,24 . FB1 has proven to lead to the manifestation of an HR and defense reactions against pathogens 11,16,35 . The comparison between these treatments revealed striking similarities that included same biochemical features (LCB rise, MAPK activation and H 2 O 2 production), structural changes in the chloroplast (thylakoid unstacking and deformation of the chloroplast) and photosynthetic function impairment. The experimental comparison of both treatments allowed the identification of several pieces of the mechanism underlying plant defense under a scheme of PCD.
In Arabidopsis, FB1 induced an HR-like wound due to a LCB accumulation acting as second messengers 11,16,35 . Here, we also show that FB1 in Phaseolus vulgaris induces a LCB rise and downstream events of HR such as MPK activation and ROS production. The parallelism found here for Pst, a non-host pathogen, shows that LCBs are also involved in this type of immune response. Although, during the non-host immunity response observed in Nicotiana benthamiana with Pseudomonas cichorii, an increase in gene transcription of SPT subunits was essential to produce HR, LCB changes were undetectable 8 . The successful LCB detection in our case of non-host defense could be due to a difference in LCB extraction and/or to the frequent sampling during the intermittent and irregular LCB increase pattern that contrasts with regular LCB rises induced by pathogens in the PAMPs-triggered immunity (PTI) and in the Effector-triggered immunity (ETI) schemes 51 . Therefore, the LCB surge seems to follow a profile that is specific to the type of immunity. In the case of the PTI, the rise of LCB occurs as a unique, early and short-life peak upon exposure to the pathogen while in the ETI occurs as a sustained and high LCB surge 51 . It can be suggested that non-host immunity, that contains PTI and ETI features, shows a LCB surge profile that takes elements from both PTI and ETI, which are associated with this non-host HR. . MAP cascades are activated upon exposure to several stresses. Based on the specificity of substrate phosphorylation and antibody reactivity, the present work shows that a MAPK is activated in Phaseolus leaves upon FB1, SN and Pst exposure. This result indicates that endogenous and exogenous LCBs are able of eliciting the same response, i.e. the same MAPK activation. In the case of responses to pathogen attack, the most common MAPKs activated are MPK3 and MPK6 37,52 . In addition, the latter has been shown to be the one activated downstream LCB accumulation 11 . Therefore, the molecular mass, the activation by LCB and by SA, and the analogous time response pattern to Pseudomonas syringae pv. tomato strongly indicate that the MAPK activated is MPK6 as it occurs in other systems 11 . However, immunoprecipitation with specific antibodies directed against the Phaseolus enzyme must settle this point.
It is possible that very low levels of H 2 O 2 generated at early times and undetected by DAB stain, initiate PCD as part of the initial signaling events that include LCB surge and MAPK activation. Such ROS generation that we were unable to detect has been described as an initial H 2 O 2 burst in defense against pathogens, but takes place outside the chloroplast 11,16,36 .
After the signaling steps: LCB increases and MAPK activation, the chloroplast starts a program to stop photosynthesis. The pattern of decay of photosynthesis upon LCBs accumulation was very similar between the FB1 www.nature.com/scientificreports www.nature.com/scientificreports/ and Pst treatments. Three main characteristics were observed: (1) a decrease in the number of functional reaction centers of PSII; (2) the compromised ability of PSII to reduce the PQ and to sustain the electron transport towards PSI; (3) a loss in the PSII centers connectivity due to severe thylakoid unstacking. These shared features between the FB1 and Pst treatments make us hypothesize that the loss of function of the chloroplast is a well-organized process that has in common LCB surges and MAPK activation.
It is known that during pathogen infections the photosynthetic activity is interrupted [20][21][22] . The observed effect in this work over linear electron transport and its carriers (as shown by JIP-test and MR) indicates a size decrease of the pool of acceptors. This would lead to an increased amount of ROS accumulation in the chloroplast 53,54 . It is commonly accepted that ROS can be generated in the chloroplast via light sensitization [55][56][57] and this could be aggravated by an increase in excitation pressure. This explains why at 24 h, DAB stain detected high levels of H 2 O 2 , which occurred after the first major signs of photosynthesis inactivation started to appear.
Interestingly, it has been reported that PCD induced by FB1 requires light and it has been proposed that ROS produced in the chloroplast could be the second messengers that up-regulate SA synthesis by enhancing PAL activity leading to a PCD HR-like 58 . Also, based only on chloroplast ultrastructure, studies in Arabidopsis, one using FB1 11 and the other using Pseudomonas syringae maculicola 4326 avrRpt2 58,59 proposed that ROS formed in the organelle may cause disassembly of the photosystems. Altogether, the detected accumulation of H 2 O 2 in this work must have caused a destructive effect on the photosynthetic ensemble as revealed by the ultrastructural observations. LCB accumulation produced by FB1 and Pst caused an eventual alteration of chloroplast morphology. It was in the rounded chloroplasts where the large peroxide accumulation occurred and in which thylakoids looked unstacked and broken. In support to this structural and functional damage, the so-called L-band was observed www.nature.com/scientificreports www.nature.com/scientificreports/ in both FB1 and Pst treatments. The appearance of a positive L-band is attributed to failures in the connectivity between the photosystems [48][49][50] . For both treatments, the L-band was detected at the times when clear thylakoid unstacking was found. Therefore, the failure of connectivity is reasonably explained by the unstacking of the thylakoid membranes in the leaves exposed to the toxin or the pathogen.
Here we show that during PCD, just after the LCB accumulation and MAPK activation took place, a decrease in the chloroplast electron carriers occurred. This was followed by a large accumulation of ROS inside the chloroplast that lead to collapse the thylakoid membrane system. It is possible that ROS could catalyze the oxidation of the thylakoid membrane lipids. This is supported by the formation of H 2 O 2 inside the chloroplasts 17 and by the promotion of extensive chloroplast damage by FB1, that includes the degradation of proteins from the chloroplast stroma 57 . Degradation of core proteins from the photosynthetic complexes has been reported but in a model that involves a compatible plant-pathogen interaction 59,60 . In nature, several scavenging processes ensure a low concentration of ROS inside the chloroplast 61 . Therefore, an intermediary process is required to allow accumulation of ROS in the chloroplast; thus, a midway step between accumulation of LCBs and inactivation of the chloroplast would be required to shut down ROS scavenging processes. The decay of the photosynthetic machinery observed can be related to a silencing of the chloroplast protein expression 62,63 associated with an absence of the repair mechanism that produces the photodamage of PSII [64][65][66][67] . Therefore, the damaged photosystems would not be replaced by new ones 68 and the chloroplast would produce ROS via antenna sensitization due to excitation pressure 69 .

conclusions
We have shown that two agents inducing PCD by accumulation of LCBs led to the organized decay of the chloroplast function and structure. This is characterized by the decline of the light reactions of photosynthesis, which leads to the accumulation of ROS in the chloroplast. Thus, the observed HR-PCD involves signaling and executory steps that include the rise of LCBs, MAPK activation, collapse of both photosystems, massive ROS production and structural disintegration of the thylakoid membrane and the chloroplast. This orchestrated and programmed sequence of events leads the plant cell to its self-destruction as a defense strategy against dissemination of biotroph pathogens. The features of such molecular events may vary in other cases of PCD or in other forms of immunity but because of their impact on cell energetics, they seem to function as a stratagem to reach a point of no-return in the program of cell death in immunity.