In vascular plants, photosynthetic electron transport (PET) chain produces reductive power that is utilised by diverse acceptors involved in both chloroplast and cellular metabolism1. In the last steps of this process transfer of energy-conserving electrons beyond photosystem I (PSI) is performed by a small iron-sulphur protein ferredoxin (Fd). Fd acts simultaneously as a bottleneck and as a hub which distributes high-energy electrons to a multitude of enzymes, which include nitrite reductase, sulphite reductase, fatty acid desaturase, glutamine-2-oxoglutarate amino transferase (GOGAT), redox complexes such as cytochrome b6/f for cyclic electron transport (CET) and thioredoxins2. The dominant pathway in chloroplasts is, however, the one that produces NADPH, by the activity of ferredoxin NADP+ oxidoreductase (FNR). This is known as the linear electron transfer (LET) pathway1. Photosynthetically active FNR (sometimes classified as autotrophic) is a member of the family of soluble monomeric enzymes that contain non-covalently bound FAD (flavin adenine dinucleotide) as the prosthetic group3. Autotrophic FNR utilizes two Fds to produce one molecule of NADPH. Other members of the FNR family are involved in various metabolic pathways, such as: nitrogen fixation, isoprenoid biosynthesis, steroid metabolism, xenobiotic detoxification, iron-sulphur cluster biogenesis and oxidative stress response4. In those pathways, FNR however catalyses the opposite reaction, namely the production of reduced Fd4 and is thus sometimes classified as heterotrophic FNR. In order to function efficiently Fd/FNR redox chemistry has to be constantly poised5. Over-reduction of PET chain can damage protein complexes and result in production of the excess amounts of reactive oxygen species (ROS)6. The ROS include free radical forms (O2.−, superoxide radicals; OH., hydroxyl radical; HO2. perhydroxy radical and RO. alkoxy radical) and molecular non-radical forms (H2O2, hydrogen peroxide and 1O2, singlet oxygen)7. During photosynthesis, PSI and photosystem II (PSII) are the primary sources of 1O2 and O2.− production6,8. It is widely accepted that in conditions leading to over-reduction, pseudo-cyclic electron transfer enables electrons to flow from water to PSI, with molecular oxygen as alternative electron acceptor5. This process, known as Mehler reaction, results in formation of H2O2 which is generated by the superoxide dismutase (SOD) and converted into water and oxygen by ascorbate peroxidase9. Mehler reaction, or water-water cycle, allows functionality of the CET and subsequent synthesis of ATP needed for the Calvin cycle, which finally oxidizes NADPH to re-establish LET5. The question remains how is the partitioning of electrons between the various energy-conserving and -dissipating pathways achieved. Thylakoid rhodanase-like protein TROL has been shown to act as bona fide membrane attachment point for the FNR10,11,12, although other membrane associations of FNR have been described13,14,15,16,17,18,19. Further, the N-terminus of FNR was shown to be important for the FNR recruitment to the membrane and its association with different supramolecular complexes11. TROL consists of the centrally positioned rhodanase homology domain which is connected by proline and valine rich swivel with the C-terminal FNR membrane recruitment motif (MRM)10,18. The MRM of TROL shares high amino acid similarity with a similar domains of the inner envelope translocon protein component Tic6210,18,20. Association of pea Tic62 poly-proline type II helix MRM peptide with the FNR was shown to be pH-dependent18. However, Tic62 is a soluble protein and it’s homologues from other plant species lack FNR MRM18,21. TROL-FNR complexes are located in non-appressed regions of thylakoid membranes, in the vicinity of PSI10. TROL knock-out (trol) plants have decreased electron transport rate (ETR) and increased non-photochemical quenching (NPQ) in high-light (HL) conditions10.

Recently, a new concept for FNR binding to plant thylakoids has been proposed22. According to this model, a majority of chloroplast FNR is during periods of darkness bound to thylakoids via Tic62 and TROL proteins. In these conditions, surplus FNR is stored at the points of attachment, presumably stabilizing the FNR enzyme during the hours of photosynthetic inactivity. In the periods of changing light conditions, i.e. in the morning hours, FNR is released to stroma, where it acts as an efficient NADPH catalyst, allowing efficient LET. However, contrary to this scheme previous studies have reported that NADP+ photoreduction is very inefficient when the enzyme is not bound to the membranes23 and that thylakoids devoid of FNR cannot reproduce WT rates of NADPH production24. Such scenario cannot be avoided even when high concentrations of soluble FNR are added24. Further, in transgenic Arabidopsis plants enriched for FNR at TROL, a more rapid induction of NPQ during light transition and a higher ratio of PSI/PSII excitation were recorded, implicating enhanced CEF21.

We have previously proposed a different scheme for the dynamic FNR recruitment to TROL20,25. We proposed that in normal light conditions TROL bound FNR performs efficient NADP+ photoreduction, due to specific FNR recruitment at the vicinity of PSI, which is necessary for the directed transfer of reduced Fd. In the conditions of HL, FNR is released from TROL, probably by the transmembrane signalling involving rhodanase and proline–rich domains. Released FNR can act as efficient ROS scavenger, or Fd can partition electrons to other acceptors.

Here we test our hypothesis and investigate how the deletion of TROL and membrane detachment of FNR from this anchoring site influence partitioning of electrons beyond PSI. We posit that TROL-FNR interaction presents the branching point between electron-conserving and electron-dissipating pathways. We monitored ROS formation and subsequent induction of plant stress-relief responses. We establish that without TROL, light-dependent O2.− generation is reduced, while the generation of other ROS is enhanced. Most significantly, O2.− formation induced by MV is diminished in chloroplasts lacking TROL, when plants are pre-acclimated to dark and growth light (GL). We propose that other Fd-dependent pathways downstream of PSI and different from the LET become dominant by the dynamic detachment of FNR form TROL, thus suggesting novel mechanism of photosynthesis regulation. Alternatively, efficient scavenging of O2.− by FNR-dependent pathways can be envisaged.


Generation of ROS in isolated chloroplasts from wild-type (WT) and trol plants was monitored in light, as well as in dark, when all reaction centres are open. We have used O2.− specific spin trap DMPO (5,5-dimethyl-1-pyrroline-N-oxide) and general electron spin trap Tiron (4,5-dihydroxy-1,3-benzenedisulfonic acid) to detect the production of ROS26,27 in conditions when FNR docking partner TROL is missing from the vicinity of PSI. Moreover, we chose different light acclimation conditions to promote interactions of FNR with different complexes or membrane domains. Since the concentrations of plant antioxidants are important for the interpretation of ROS production that were investigated in EPR experiments, we have quantified the levels of ROS-scavenging enzymes and the extent of lipid peroxidation and protein carbonyls in the WT and trol plants acclimated to GL (see Supplementary Fig. S1 and Supplementary Table S1 online). Except for the peroxisomal catalase activity and the levels of H2O2, all other enzymatic and non-enzymatic components of antioxidative stress response were not significantly different between WT and trol plants (see Supplementary Fig. S1 and Supplementary Table S1 online).

Trol chloroplasts accumulate diminished levels of O2

Illuminated isolated intact Arabidopsis WT chloroplasts doped with DMPO revealed a dominant DMPO-OOH spin adduct formation that, according to the EPR spectral parameters, is typical for the superoxide radical formation (Fig. 1a). The apparent hyperfine splitting constants of DMPO-OOH (aN = 1.42 mT, a = 1.14 mT, a = 0.115 mT) are in conformity with the published literature28. In the presence of SOD the corresponding EPR signal is suppressed (29, Fig. 1a). As the lifetime of superoxide is short with respect to the time scale of the EPR experiment, the superoxide “end-product” in terms of DMPO-OH hydroxyl radical spin adduct is detected as well (aN = 1.49 mT, aH = 1.47 mT). The analysis of DMPO-OOH spin adduct formation in chloroplasts from trol plants acclimated to any of the three light conditions revealed a reduced EPR signal intensity and, thus, decreased free radical production with respect to the WT plants (Fig. 1b). Specifically, when the EPR data measured for trol chloroplasts were normalized to the WT data, which are assumed to represent a 100% radical yield, around 20% lower light dependent generation of superoxide anion could be estimated for all three acclimation conditions (Fig. 2a). These findings implicate that in the absence of TROL, photo-generated electrons less efficiently spill over to O2, as it could be expected from the present models of the PET5, or that the O2.− is to some extent scavenged. It is also important to point out that photoreduction of O2 by isolated chloroplasts is saturated already at low intensities of light (around 10 μmol photons m−2 s−1)18,30 and low concentrations of O2. Non-normalised experimental data for the DMPO EPR measurements of WT and trol chloroplasts are shown in the Supplementary Table S2.

Figure 1
figure 1

Experimental EPR spectrum.

(a) The analysis of EPR spectra of DMPO-OOH spin adduct formation in chloroplasts isolated from Arabidopsis WT plants exposed to illumination with photosynthetic light of 100 μmol photons m−2 s−1 for 30 seconds. The experimental spectrum (WT) was simulated revealing the contribution of DMPO-OOH superoxide spin adduct (75%) and DMPO-OH hydroxyl radical spin adduct (25%). In the presence of SOD the former component is suppressed. (b) EPR spectra of spin adducts of chloroplasts isolated from Arabidopsis WT and trol plants grown in different light conditions (dark, GL and HL).

Figure 2
figure 2

The EPR study of DMPO-OOH spin adduct formation in chloroplasts isolated from Arabidopsis trol plants acclimated to different light conditions (dark, GL or HL).

Chloroplasts were illuminated with photosynthetic light of 100 μmol photons m−2 s−1 for 30 seconds. (a) The superoxide anion production in mutant expressed in terms of WT plant data exposed to the same experimental conditions and taken as a reference representing 100% radical yield; (b) The superoxide anion production from trol chloroplasts exposed to MV normalized with respect to the data of WT exposed to MV which are taken as a reference for 100% superoxide radical production. Data (mean ± SE) from three independent measurements were analysed by one-way ANOVA and LSD post-hoc test. Data labelled with the same lower-case letters are not significantly different. Different letters point to the experimental data that are significantly different (p < 0.05).

Lower amounts of MV-induced O2.− can be detected in chloroplasts of dark and growth-light acclimated trol plants

To further substantiate these findings and to test previous reports that released FNR acts as an efficient scavenger of O2.− anion31, we have incubated chloroplasts with methyl-viologen (MV). In illuminated chloroplasts, MV traps almost all electrons from the PSI, specifically from the reduced Fd, at a diffusion-controlled rate32, to form its cation radicals, which generate O2.− at a rapid rate via auto oxidation. Using EPR spectroscopy, O2.− can be detected in terms of the larger DMPO-OOH spin adduct formation in the samples exposed to MV, in comparison to non-exposed ones. In HL conditions the EPR signals did not significantly differ between trol and WT chloroplasts. When the experimental data of radical production in chloroplasts non-exposed to MV were taken as referent values representing 100% of O2.− yield, it could clearly be noticed that the exposure of WT chloroplasts to MV caused significantly larger EPR signal for plants acclimated to dark (ca. 130%) or GL (ca. 170%) (see Supplementary Fig. S2 online). On the other hand, for WT plants acclimated to HL the yield of O2.− was practically not affected by the exposure to MV (see Supplementary Fig. S2 online). In chloroplasts isolated from trol plants acclimated to GL, light-induced generation of O2.− was drastically increased (ca. 170%) after addition of MV, whereas in chloroplasts isolated from trol plants acclimated to dark and HL conditions no significant impact of MV on superoxide anion generation was observed (see Supplementary Fig. S2 online). When the experimental data of radical production in WT chloroplasts treated with MV were taken as referent values representing 100% of O2.− yield, it could clearly be noticed that the exposure of trol chloroplasts to MV caused significantly smaller EPR signal for plants acclimated to dark (ca. 40% less DMPO-OOH adduct) and to GL (ca. 30% less DMPO-OOH adduct) (Fig. 2b). Observed DMPO-OOH adduct production in chloroplasts of WT and trol plants acclimated to HL remains constant despite the addition of MV (Fig. 2b, see Supplementary Fig. S2 online).

Accumulation of other ROS in trol chloroplasts

We have then asked whether observed reduction of O2.− generation is a part of a more general ROS avoidance response which is dependent on the release of FNR from TROL. To answer this question we have inspected the possibility of other ROS formation, besides the superoxide anion, by using nonspecific electron spin trap Tiron. The appearance of the EPR signal exhibiting electron hyperfine couplings of 0.36 and 0.187 mT from two non-equivalent hydrogens is indicative of Tiron semi-quinone radical (Fig. 3a) pointing to the electron addition sources in general27. Since the focus of this study relies on the comparison between the radical productions in mutant versus WT plants, the later experimental data were taken as referent ones and assumed to represent 100% radical yield. In this approach EPR signal intensity from Tiron incorporated in trol versus WT chloroplasts was at least 20% larger in either dark- or GL-acclimated plants, in both dark and light EPR experimental setups (Fig. 3b). However, Tiron-detectable ROS in HL-acclimated plants did not significantly differ from the WT (Fig. 3b). Non-normalized experimental data for the WT and trol chloroplasts in the presence of Tiron spin trap are shown in the Supplementary Table S2 online. We then performed the experiments in the presence of MV. Contrary to the experiments with DMPO, in the presence of Tiron chloroplasts of both WT and trol plants acclimated to any of the three light conditions accumulated significantly more (≥ 50%) Tiron semi-quionone radicals when exposed to MV in comparison to non-exposed, the later assumed to represent the referent 100% radical yield (see Supplementary Fig. S2 online). However, when the EPR signals from WT chloroplasts treated with MV were taken as referent values representing 100% of Tiron semi-quinone yield, it became evident that trol chloroplasts in the presence of MV do not generate more Tiron-detectable ROS then the MV treated WT (Fig. 3c). The exception are chloroplasts of GL-acclimated trol plants, which generated about 10% less Tiron-detectable ROS when the EPR spectra were recorded under illumination (Fig. 3c), however, statistically non-significantly.

Figure 3
figure 3

The EPR study of Tiron semiquinone radical formation in chloroplasts isolated from WT and trol plants acclimated to different light conditions (dark, GL or HL).

Chloroplasts were illuminated with photosynthetic light of 100 μmol photons m−2 s−1 for 30 seconds (white columns) or kept in dark for the equivalent time period (black columns). (a) The experimental and simulated EPR spectra of Tiron semiquinone radical in WT; (b) Data are normalized with respect to the EPR signal of the chloroplasts extracted from WT plants assumed to represent 100% radical yield; (c) Data from trol chloroplasts exposed to MV normalized with respect to the data from WT exposed to MV assumed to represent 100% radical yield. Data (mean ± SE) are analysed by one-way ANOVA and LSD post-hoc test. Data labelled with the same lower-case letters are not significantly different. Different letters point to the experimental data that are significantly different (p < 0.05).

Finally, we have determined the intactness rate of around 75% for both WT and trol chloroplasts by the end time point of all described EPR measurements, as assayed by oxygen evolution measurements33.

The fact that the generation of these radical species was independent of light exposure in the EPR assays, indicated that generated ROS most likely involve radicals other than the O2.− anion.


This work addresses a fundamental question in current photosynthesis research, namely the understanding of the regulation of photosynthetic electron partitioning21.

It has been shown that both overexpression of FNR34 and its release from thylakoid membranes31 increase ROS tolerance in tobacco and wheat. These findings and the distinctive HL phenotype of trol plants (Fig. 4) prompted us to investigate stress-relief mechanisms and chloroplast ROS generation in trol plants pre-acclimated to different light regimes. To study the generation of ROS and to address the fate of electrons in reactions downstream of PSI, we have employed a highly sensitive electron paramagnetic resonance (EPR) spectroscopy30. In chloroplasts, the major site of O2.− production is the thylakoid membrane-bound primary electron acceptor of PSI35. The generation of O2.− may trigger the formation of more reactive ROS like OH. and 1O29,30,35. We speculated that the binding of FNR to the TROL is important for partitioning of electrons toward energy-conserving pathway36 of NADPH production.

Figure 4
figure 4

Phenotype of A. thaliana plants.

Distinctive HL phenotype of trol plants (left) vs WT (right) which is evident after 2-day exposure to 800 μmol photons m−2 s−1. Trol plants remain green, while WT accumulates anthocyanins and starts flowering. Both conditions are considered as signs of photooxidative stress. Tray containing trol pants was randomly spiked with WT (marked by arrows).

The EPR spin-trapping data on isolated chloroplasts indicate that there are differences in free radical generation or scavenging in WT and trol samples. Trol chloroplasts generate diminished levels of O2.− in all three tested acclimation conditions. It seems that in the absence of TROL photo-generated electrons less efficiently spill over to O2. On the other hand, it is possible that in the absence of TROL, PSI photo-generated electrons are not passed over to O2 in the first place, but reduced Fd distributes them efficiently to any of the alternative pathways in chloroplasts (see Introduction). Here we have to point out that TROL deficiency does not alter CET10. This would imply that the FNR-driven production of NADPH in LET is actually a rate limiting process and that some other Fd-dependent pathway is (are) more efficient electron sink(s). That could explain the need for TROL as an FNR sequestration site in vascular plant thylakoids. Hence, FNR has to be docked in the vicinity of Fd reduction site to prevent Fd from transferring electrons to alternative pathways, although FNR has high affinity for reduced Fd37. Finally, it has been shown that Fd-reduced, bound FNR cannot catalyse the photo-reduction of O238.

It is generally accepted that during the periods of photosynthesis the role of FNR in the production of NADPH becomes dominant over its diaphorase activity (3 and references therein). In HL, stroma is over-reduced, leading to the activation of FNR diaphorase activity and efficient ROS avoidance (see scheme in 20,25) through so-far unknown mechanism. Trol chloroplasts produce approximately 20% less O2.− compared to the WT in all pre-acclimation conditions (Fig. 2a). Even more strikingly, in trol chloroplasts treated with the redox-cycling agent MV, 30-40% less O2.− can be detected, when compared to the MV-treated WT chloroplasts (Fig. 2a). Apparently, chloroplasts of dark- and GL-acclimated trol plants distribute electrons in such a way that MV-induced rapid O2.− production is efficiently outcompeted, or entirely avoided. Alternatively, in the absence of TROL-FNR interaction, soluble FNR acts as an efficient O2.− scavenger, or FNR could be bound to alternative complexes or membrane domains which allows that light-induced generation of O2.− is efficiently and rapidly diminished even in the presence of MV. It is possible that in the HL-acclimated plants the mechanisms of O2.− scavenging have reached their maximum, irrespective of TROL-FNR interaction.

In the case of the treatments with MV followed by the free radical detection with Tiron, MV consumes almost all electrons from PSI to generate superoxide anion in redox-cycling process. This radical form is subsequently quickly transformed into other radicals. Since trol chloroplasts are capable of more efficient scavenging of superoxide anion radicals produced by MV, it is conceivable to assume that the generation of other free radicals is also subsequently lower (compare Supplementary Fig. S2 online, WT-dark to trol-dark and WT-GL to trol-GL). Consequently, Tiron can detect less other radical forms. Here, for clarity, it has to be repeated that in the absence of MV trol chloroplasts produce 20% more Tiron-detectable radicals than the WT (Fig. 3a).

In conclusion, three scenarios can be envisaged: first, it is possible that scavengers utilized by the free FNR contribute to the generation of other radical forms; second, Fd efficiently passes electrons to alternative pathways when FNR is not sequestered to the TROL, which contributes to over-reduction of other pathways; third, electrons leak to various sinks upstream of the PSI, as the absence of the FNR sequestration site causes slower rates of NADP+ reduction and results in over-reduction of LET.

It has been proposed that FNR binding to the membrane recruitment motif of TROL and Tic62 is mediated by changing pH and that this binding is regulated by altering light quantity conditions11,22. Further, the assembly of different FNR-containing complexes has been implicated in regulation of electron flow39,11. Also, it has been proposed that the recruitment of FNR to TROL and to the other anchoring proteins is dynamic10,11. Such alternative binding and release of FNR or its association with different complexes, or electron donors/acceptors could be linked to the regulation and prioritization of different electron transfer pathways11,21,22,39. In our models, we suggest that trol chloroplasts are not capable of performing the dynamic light- and/or pH-dependent FNR association with the thylakoid membranes20,25. Finally, Lintala et al.12 have shown that tic62 trol double mutants displays distinct metabolic phenotype, while Goss et al.40 link FNR amounts and distribution with redox state of cellular glutathione.

To summarize, PSI is capable of reducing O2 to O2.− 2 and the probability of this occurring increases when the rate of reduction of Fd by PSI exceeds the rate of NADPH utilization by carbon fixing reactions41. The maintenance of electron transport under these conditions is particularly important, since the resulting over-acidification of the thylakoid lumen down-regulates excitation of PSII42. Present models predict that pseudo-cyclic electron flow, over H2O2 to water, is activated to restore the redox poise and sustain LET1,5. Here we demonstrate that the absence of TROL triggers much more efficient O2.− scavenging mechanisms. We propose that the dynamic binding and release of FNR from TROL represents novel and efficient mechanism that maintains LET before pseudo-cyclic flow is activated. Furthermore, we hypothesize that the FNR-TROL branch point could be the source element that modulates various downstream networks of plant ROS detoxification.


Plant material and growth conditions

Arabidopsis thaliana (L.) ecotype Columbia (Col-0) plants and At4g01050 knock-out mutant line, trol10, were grown on soil at 20 °C under 80 μmol photons m−2 s−1, a 16-h photoperiod and a relative air humidity of 60% (day) and 70% (night). Two days prior to EPR measurements plants were grown either in dark or under 80 μmol photons m−2 s−1 (GL) or 250 μmol photons m−2 s−1 (HL). Trol line has been thoroughly characterized previously10 and has been shown to have molecular and phenotypic characteristics similar to TROL antisense line.


The spin trap 5,5-dimethyl-1-pyrroline-N-oxide (DMPO), 4,5-dihydroxy-1,3-benzenedisulfonic acid (Tiron) and N,N′-dimethyl-4,4′-bipyridinium dichloride (Methyl viologen dichloride hydrate, MV) were purchased from Sigma Chemical Co. (St. Louis-Aldrich Corp., MO, USA).

Chloroplast isolation from A. thaliana

Intact Arabidopsis chloroplasts were isolated from 3-4 weeks old plants as described by Jurić et al.10. Final sedimentation was achieved by centrifugation at 1000 xg for 5 min. Sediment was resuspended in buffer containing 330 mM Sorbitol and 20 mM Tris/HCl pH 8.4. Chloroplast concentration was set to 1 mg chlorophyll per 1 ml buffer. Chloroplast intactness and functionality was assessed by oxygen evolution measurements33. The value of intactness is given in the Results section.

EPR measurements

The formation of ROS was measured by EPR spectroscopy using DMPO as a specific superoxide anion spin trap molecule and Tiron as a general spin trap22,23. Reaction mixture contained chloroplasts equivalent to 50 μg chlorophyll and 10 mM Tiron or 433 mM DMPO whereas in the experiments with MV its respective concentration was 10 mM. Immediately after admission of a specific spin trap, samples were either kept in the dark or illuminated with photosynthetic light of 100 μmol photons m−2 s−1 for 30 seconds. The experiments were performed in glass capillaries (inner diameter of 1 mm) on an X-band Varian E-109 spectrometer applying the following instrumental set-up: microwave power of 20 mW, modulation amplitude of 0.1 mT, modulation frequency of 100 kHz and scan range of 8 mT (DMPO) and 2.5 mT (Tiron). Data were collected using the supplied software43 at room temperature and analysed by EasySpin software package44. In order to compare the EPR detected radical yield in trol versus WT plants, data related to mutant were normalized with respect to the WT data, which were taken as a referent ones and assumed to represent 100% radical yield. Similarly, when the samples were exposed to MV, the results were quantified with respect to the EPR signal of chloroplasts not exposed to MV, which were taken as a referent ones and assumed to represent 100% radical yield.


All statistical analyses were carried out by using IBM SPSS Statistics software, ver. 18 (IBM Corporation, Armonk, New York, USA). Details for the particular analyses are described in the respective Figure and Table legends.

Additional Information

How to cite this article: Vojta, L. et al. TROL-FNR interaction reveals alternative pathways of electron partitioning in photosynthesis. Sci. Rep. 5, 10085; doi: 10.1038/srep10085 (2015).