Redox chains in chloroplast envelope membranes: Spectroscopic evidence for the presence of electron carriers, including iron–sulfur centers

Abstract

We have shown that envelope membranes from spinach chloroplasts contain (i) semiquinone and flavosemiquinone radicals, (ii) a series of iron-containing electron-transfer centers, and (iii) flavins (mostly FAD) loosely associated with proteins. In contrast, we were unable to detect any cytochrome in spinach chloroplast envelope membranes. In addition to a high spin [1Fe]³⁺ type protein associated with an EPR signal at g = 4.3, we observed two iron–sulfur centers, a [4Fe-4S]¹⁺ and a [2Fe-2S]¹⁺, associated with features, respectively, at g = 1.921 and g = 1.935, which were detected after reduction by NADPH and NADH, respectively. The [4Fe-4S] center, but not the [2Fe-2S] center, was also reduced by dithionite or 5-deazaflavin/oxalate. An unusual Fe-S center, named X, associated with a signal at g = 2.057, was also detected, which was reduced by dithionite but not by NADH or NADPH. Extremely fast spin–relaxation rates of flavin- and quinone-free radicals suggest their close proximity to the [4Fe-4S] cluster or the high-spin [1Fe]³⁺ center. Envelope membranes probably contain enzymatic activities involved in the formation and reduction of semiquinone radicals (quinol oxidase, NADPH-quinone, and NADPH-semiquinone reductases). The physiological significance of our results is discussed with respect to (i) the presence of desaturase activities in envelope membranes and (ii) the mechanisms involved in the export of protons to the cytosol, which partially regulate the stromal pH during photosynthesis. The characterization of such a wide variety of electron carriers in envelope membranes opens new fields of research on the functions of this membrane system within the plant cell.

The two envelope membranes that surround chloroplasts contain numerous enzymes involved in the biosynthesis of specific plastid membrane constituents: glycerolipids, prenylquinones, chlorophyll precursors, carotenoids (for review see ref. 1). The formation of polyunsaturated fatty acids and of colored carotenoids involve desaturation steps that are only poorly understood. For instance, sequential desaturation of the colorless carotenoid precursor phytoene leads to the formation of lycopene, a colored carotenoid with 11 double bonds. Quinones and factors regulating the redox state of quinones may play a major role in phytoene desaturation, whereas in vitro molecular oxygen is the terminal electron acceptor (2, 3). Most of our knowledge on fatty acid desaturation in chloroplast concerns the soluble components of the 18:0 to 18:1 desaturation system: ferredoxin, ferredoxin: NADP oxidoreductase, stearoyl-ACP desaturase (for review see ref. 4). An n-6 lipid-linked desaturase, probably involving ferredoxin:NADPH oxidoreductase, has been characterized in chloroplast envelope membranes (5). Experiments on chloroplasts (6) suggest that O₂ could be the final electron acceptor, whereas reduced ferredoxin (E′₀ = −0.4 V) could be the source of electrons for the reduction of O₂ to H₂O (E′₀ = +0.81 V). Because ferredoxin delivers only one electron at a time, the envelope desaturase has to oxidize two reduced ferredoxins, and store the first electron before the double bond is formed (4). This is possible only in the presence of a complex electron-transfer chain, which has not been detected yet in envelopes.

Desaturation is not the only envelope enzymatic process that would require an electron-transfer chain. For instance, optimization of photosynthesis in chloroplasts is dependent on the maintenance of pH gradients between the stroma and both the thylakoid lumen and the cytosol (7). In the latter case, the inner envelope probably contains an energy-transducing proton pump as a primary mechanism facilitating the formation of stroma-cytosol ΔpH (8, 9). A possible role for an envelope electron-transfer chain could be to maintain this ΔpH.

To date, the only known envelope constituents that could play a role in electron transfer are prenylquinones: α-tocopherol and plastoquinone-9 (10, 11). Because quinone radicals are often involved in redox chains and since semiquinones are paramagnetic and therefore detectable by EPR spectroscopy, we investigated envelope membranes from spinach chloroplasts by EPR spectroscopy. In this article, we report for the first time the presence in envelope membranes of: (i) several iron–sulfur proteins, (ii) semiquinones, and (iii) flavins that could be components of one or several electron-transfer chains.

MATERIALS AND METHODS

Purification of Envelope Membranes from Intact Spinach Chloroplast.

Chloroplasts were isolated from spinach (Spinacia oleracea L.) leaves and further purified by centrifugation in Percoll gradients (12). Envelope membranes, thylakoids, and stroma were purified from chloroplasts, lysed in hypotonic medium, by centrifugation through a step-sucrose gradient (12). They were stored (in liquid nitrogen) at 10 mg of protein per ml in 10 mM 4-morpholinepropanesulfonic acid (MOPS)–NaOH (pH 7.8), or lyophilized for pentane treatment. Extensive analyses of purified envelope fractions have shown (for review see ref. 12) that they were totally devoid of membranes derived either from thylakoids, mitochondria, endoplasmic reticulum, or any other extraplastidial membranes that could exhibit EPR signals.

Pentane Treatment of Chloroplast Envelope Membranes.

One milliliter of chilled (0–5°C) distilled pentane (13) was added to lyophilized envelope membranes (10 mg protein). The mixture was vortexed in a glass tube for 5 min under argon at 0–5°C and finally centrifuged at 1000 × g (Kubota, Tokyo). The supernatant was discarded. Pentane extraction of the pellet was repeated four times (14). The remaining pentane was removed from the envelope pellet by evaporation under a stream of argon during 1 hr at 0–5°C followed by 1 hr at 20°C. Envelope proteins were finally rehydrated in 1 mM MOPS–NaOH (pH 7.8).

EPR Measurements.

EPR spectra were recorded on a Varian E 109 spectrometer coupled to a Hewlett–Packard 9826 calculator and equipped with a Varian Gaussmeter and an EIP 548A microwave-frequency counter, for calibration of the magnetic field and of the frequency. The samples were cooled with a liquid helium transfer system (ESR 900, Oxford Instrument) to variable temperatures starting from 4.2 K. The temperature was measured with a gold–iron/chromel thermocouple located about 2 cm below the bottom of the EPR sample in the flowing helium gas stream. Samples of envelope membranes (150 μl, 1.5–6 mg protein) were placed in EPR quartz tubes, rapidly frozen in liquid nitrogen, and stored at 77 K.

Reduction of Chloroplast Envelope Membranes with Chemical Agents.

The samples were reduced by addition of either dithionite or 5-deazaflavin/oxalate. Aliquots of a stock solution of dithionite were progressively added to the same envelope sample to obtain a range of concentration (up to 5 mM). Envelope membranes and dithionite at given concentrations were incubated together for 3 min at 20°C and then rapidly frozen in liquid nitrogen for EPR analysis. After recording its spectrum, the sample was thawed for the next addition of dithionite. Reduction with the photoactivatable catalyst 5-deazaflavin (20 μM) in the presence of sodium oxalate (25 mM) as electron donor was performed as described by Jouanneau et al. (15). The mixture containing envelope membranes together with 5-deazaflavin/oxalate was irradiated for 30 min at 30 cm from a white light source (250 W) to start the photoreduction. Control experiments were also run in the presence of 5-deazaflavin/oxalate, but the samples were placed in the dark instead of being irradiated. After incubation, the samples were rapidly frozen in liquid nitrogen for EPR analysis.

Oxidation or Reduction of Chloroplast Envelope Membranes with Physiological Mediators.

NADH or NADPH (E′₀ = −0.32 V) and oxygen (E′₀ = +0.81 V) were used, respectively, to reduce or oxidize envelope electron carriers. NADPH or NADH were progressively added to the same envelope sample to obtain a range of concentration (up to 500 μM). After each addition, envelope membranes were incubated for 10 min at 25°C and then rapidly frozen in liquid nitrogen for EPR analysis, as described above. After recording the spectrum, the sample was thawed for the next addition of NADPH or NADH. Oxidation of envelope electron carriers was achieved under a stream of oxygen during an increasing length of time (up to 20 min). Aliquots of the suspension were taken at 0, 5, 10, and 20 min and rapidly frozen in liquid nitrogen for EPR analysis.

Carotenoid and Prenylquinone Determination.

Carotenoids and prenylquinones were analyzed as described by Block et al. (16) and Lichtenthaler et al. (10), respectively.

Flavin Determination.

Flavins were analyzed in envelope membranes as described by Faeder and Siegel (17). Protein-bound flavins were first extracted from membranes as described by Yagi (18), prior to flavin determination. Fluorescence intensities (λ_ex = 450 nm, λ_em = 530 nm) of the mixtures containing flavins were first determined at pH 7.7 with a Spex Fluoromax spectrofluorimeter, then the pH of the mixtures was adjusted to 2.6 by addition of 1 N HCl for another fluorescence analysis. The amounts of FMN and FAD present in the samples were then calculated by using the equations of Faeder and Siegel (17).

Nonheme Iron and Sulfur Determination.

Nonheme iron and sulfur were determined in envelope membranes as described by Doeg and Ziegler (19) and Beinert (20), respectively.

Protein Analyses and Determination.

Thylakoid and envelope polypeptides (corresponding to 120 μg protein) were analyzed by SDS/PAGE, as described by Chua (21). Cytochromes could be directly visualized on the gels by following their peroxidase activity with H₂O₂ and 3,3′,5,5′-tetramethylbenzidine (TMBZ), as described by Thomas et al. (22). Protein concentration was determined according to Lowry et al. (23), using BSA as a standard.

RESULTS

Chloroplast Envelope Membranes Show EPR Signals at g = 4.3 and Around g = 2.

Fig. 1A shows that native envelope membranes present a signal at g = 4.3 and another complex signal around g = 2, dominated by a major isotropic feature at g = 2.003 (Fig. 1B). Pentane-treated membranes were then used to increase the signal-to-noise ratio. Pentane is a nonpolar molecule, which does not affect the integrity of the membranes. Indeed, after pentane treatment of the membranes their EPR features remain unchanged. The g = 2 region was resolved in a series of signals with maxima at g values of 2.167, 2.077, 2.017, 1.961, 1.929, and 1.875, as well as an isotropic signal at 2.003 (Fig. 1C). Some of these could be associated with Mn²⁺ (I = 5/2), as suggested by the six-line pattern of the signal and by the hyperfine splitting value of ≈96 Gauss. Indeed, manganese was found (≈0.5 nmol per mg protein) in purified envelope membranes (24). The signals observed in the g = 2 region could also be associated with iron–sulfur proteins. In support of this suggestion, analyses of five different preparations have shown that envelope membranes contain 3.2 ± 0.3 nmol Fe and 5.3 ± 0.4 nmol S per mg protein.

Figure 1.

EPR signals of native (A and B) and pentane-treated (C) envelope membranes from spinach chloroplasts. Protein concentration: 10 mg/ml (A and B) or 45 mg/ml (C). EPR conditions: field, 2500 ± 2500 Gauss (A) or 3250 ± 400 Gauss (B and C); time constant, 0.128 sec (A) or 0.25 sec (B and C); microwave power, 1 mW; modulation amplitude, 12.5 G; frequency, 9.22 GHz; temperature, 4.2 K; gain, 10⁴. The six-line pattern under spectrum C indicates the hyperfine lines due to Mn²⁺; the distances between these lines are indicated by the different values, expressed in Gauss.

Pentane-treated envelope membranes are highly concentrated compared with native membranes (on average 40–50 mg protein per ml instead of 10–15). This is essential for EPR analyses of envelope preparations, since the signals detected have a low signal-to-noise ratio. One should keep in mind (i) that envelope membranes have a much higher lipid-to-protein ratio (1.2–1.5 mg lipid per mg protein) compared with other cell membranes (i.e., thylakoids, 0.5 mg lipid per mg protein) and (ii) that chloroplast envelope membranes are likely to have a low content of redox centers, thus making very difficult a complete characterization of the centers involved.

The Envelope EPR Signal at g = 4.3 Is Characteristic of High-Spin Fe³⁺.

This cation could be present in envelope vesicles either as “free” Fe³⁺ or associated with a protein in a [1Fe]³⁺ center. For instance, a g = 4.31 signal was observed in the oxidized form of rubredoxin, a [1Fe]³⁺ protein, from Pseudomonas oleovorans (25). In envelope membranes, the g = 4.3 signal is probably not due to contaminating free Fe³⁺, since (i) its amplitude did not decrease when envelope membranes were washed extensively by EDTA (not shown) and (ii) it was still very visible when the spectra were recorded at 15 K instead of 4.2 K (not shown). This last observation is indicative of a Fe³⁺ species with different relaxation properties from free Fe³⁺. Therefore, the Fe³⁺ center giving rise to the g = 4.3 signal in envelope membranes is likely to be covalently linked to proteins. Interestingly, a lipoxygenase, which is an iron-containing protein and therefore could contribute to such a signal, has recently been characterized in chloroplast envelope membranes (26).

Characterization of the EPR Signal at g = 2.003.

This feature is isotropic and has a band width (peak-to-peak) of 12 ± 1 Gauss (Fig. 1B). This value is indicative of a semiquinone radical, rather than of a flavin radical, also because no wings on either side of the resonance transition were visible (27). In addition, we observed that the signal at g = 2.003 was present in membranes prepared at pH ranging from 6 to 8 (not shown). The presence of such a stabilized semiquinone radical in envelope membrane is rather surprising, because semiquinone radicals are in general transient species that are formed during oxidoreduction processes. The stabilization of a semiquinone intermediate is probably dependent on its binding to a specific protein (for example see ref. 28). Quinones that are loosely bound to proteins can be extracted by pentane (14). Fig. 1C shows that after the pentane treatment the g = 2.003 signal was still visible. Its relative intensity was much smaller than in the native envelope, because the pentane-treated envelope contained 4.5 times more protein per milliliter than native envelope preparations, whereas the g = 2.003 signal was increased only by 1.2-fold. This result suggests, that either the envelope semiquinone was partially extracted by the pentane treatment or the redox state of the envelope semiquinone was affected by a change in its environment brought about by the pentane treatment. The first hypothesis is rather unlikely since we observed that pentane treatment was unable to remove envelope prenylquinones, in contrast to carotenoids which were either totally or partially extracted by pentane (not shown).

The g = 2.003 signal arises from a rapidly relaxing radical, as shown by the important decrease of its amplitude when raising the temperature from 4.2 to 46 K (Fig. 2A). Its power-saturation (from 0.01 mW to 50 mW) behavior produces a biphasic curve (Fig. 2B). One component, with the peak-to-peak width of 12 ± 1 Gauss, saturates at 2 mW and was assigned to the semiquinone radical. Beyond 2 mW, a second component is revealed, with a larger peak-to-peak width of 21 ± 1 Gauss, which does not saturate before 40–50 mW. The radical associated with this second signal was tentatively assigned to a flavosemiquinone species (for example see ref. 29).

Figure 2.

Evolution of the amplitude of the chloroplast envelope EPR signal at g = 2.003 as a function of temperature (A) and microwave power (B). Protein concentration: 50 mg/ml. EPR conditions: see Fig. 1B, except that the temperature varied from 4.2 to 46 K (A) and the microwave power varied from 0.01 mW to 50 mW; temperature, 4.2 K (B).

Chloroplast Envelope Membranes Are Devoid of Cytochromes.

The low- and high-spin hemes of cytochromes are characterized by EPR signals in the g = 3 and g = 6 regions, respectively. The lack of any signal in these regions strongly suggests that envelope membranes may be devoid of cytochromes. The presence of cytochromes in chloroplast membranes was also investigated after SDS/PAGE analysis of envelope and thylakoid polypeptides. Cytochromes b₆ and f from thylakoids were easily visualized on the gels by following their peroxidase activity with H₂O₂ and TMBZ. In contrast, no reaction was observed with purified envelope membranes (not shown). These two sets of results are in good agreement with our previous spectrophotometric observations (30); the low temperature (77 K) difference spectrum resulting from dithionite-reduced minus oxidized envelope preparations does not show any absorption peak due to cytochromes. Therefore, we have been unable to detect cytochromes in purified envelope membranes by using three different methods. This raises some doubt on the putative identification of a cytochrome as the product of the chloroplastic gene cemA, which was found to be localized in the inner envelope membrane from pea chloroplasts (31).

Reduction of Envelope Membranes by Sodium Dithionite.

Addition of dithionite to pentane-treated envelope membranes leads to major changes of the EPR spectrum. With increasing concentrations of dithionite (from 10 μM to 5 mM), we first observed the progressive disappearance of the g = 2.003 signal (not shown), probably because the semiquinone radical was reduced to the quinol state, which is EPR-silent, thus revealing more clearly a new signal, named Y, with a maximum at g = 2.017. This new signal is shown in Fig. 3 for the pentane-treated envelope recorded after addition of excess amounts (5 mM) of dithionite. Unfortunately, further identification of the center associated with the g = 2.017 signal is almost impossible, since its observation requires the complete disappearance of the semiquinone signal, which can be obtained only at high dithionite concentration.

Figure 3.

Effect of dithionite on chloroplast envelope EPR signals. Protein concentration: 35 mg/ml (A) or 45 mg/ml (B). EPR conditions: see Fig. 1B, except for B, in which the microwave power was 5 mW and the temperature 15 K. The signal centered at g = 2.003 was no longer visible in dithionite-treated samples, and a signal centered at g = 1.921 was visible in the dithionite-treated sample only when recorded at high microwave power and temperature (as indicated in B).

With increasing concentrations of dithionite (from 10 μM to 5 mM), we also noticed an increase of a broad (68 ± 2 Gauss) signal, named X, centered at g = 2.057 (Fig. 3A). Its maximum intensity occurred in the presence of 100 μM of dithionite (not shown). When the spectrum was recorded at 40 K, this signal was not observed any longer (not shown), suggesting that it could be associated with an iron–sulfur center. However, no iron–sulfur protein exhibiting such a high g_y value in the reduced state has yet been described in any other biological system. Indeed, the protein involved has EPR characteristics that are different from those associated with traditional [4Fe-4S]¹⁺, [3Fe-4S]¹⁺, or [2Fe-2S]¹⁺ centers (for review see ref. 32).

When recording the EPR spectrum of dithionite-reduced pentane-treated envelope under different conditions (temperature, 15 K; microwave power, 5 mW), signal Y, with a maximum at g = 2.017, became more intense and a new signal appeared, centered at g = 1.921 (Fig. 3B). This last signal was already visible, but very weak, in the native envelope membrane (see Fig. 1C) and was further analyzed after photoreduction with 5-deazaflavin/oxalate, which allows for reduction of iron–sulfur proteins with a very low redox potential. Similar signals have been extensively reviewed in the literature (for example see ref. 33).

Photoreduction of Envelope Membranes by 5-Deazaflavin/Oxalate.

Fig. 4 shows that after photoreduction with 5-deazaflavin/oxalate the envelope semiquinone is reduced to the quinol state since the signal at g = 2.003 is abolished, thus unmasking the g = 2.017 signal of the Y species. Also, when recording the spectrum both at a higher temperature (15 K) and a higher microwave power (5 mW), both amplitudes of the Y signal and of the signal at g = 1.921 were considerably increased. The latter could be associated with either a [4Fe-4S] or a [2Fe-2S] cluster (34). These two centers differ markedly in their relaxation properties: a [4Fe-4S] cluster has a faster relaxation and is therefore no longer visible at 30 K, whereas a [2Fe-2S] cluster relaxes more slowly and can still be detected at 70 K. This provides a diagnostic tool for distinguishing the centers, provided that magnetic interactions with other paramagnetic species are absent. Fig. 4 indicates that the signal at g = 1.921 is completely abolished at 35 K, and therefore that envelope membranes contain a [4Fe-4S] center, paramagnetic when reduced, with a global charge of +1, i.e., a [4Fe-4S]¹⁺ center.

Figure 4.

Effect of photoreduction by 5-deazaflavin/oxalate on chloroplast envelope EPR signals. Protein concentration: 10 mg/ml. EPR conditions: see Fig. 1B, except for spectrum IIb and IIc, in which the microwave power was 5 mW and the temperature 15 K and 35 K, respectively.

To provide a more physiological view of the importance of envelope components that may be involved in electron transfer, we investigated whether NADPH, NADH, and oxygen could modulate the amplitude of the chloroplast envelope EPR signals.

Oxidation and Reduction of Envelope Membranes with Physiological Mediators.

A new signal, centered at g = 1.935, was revealed upon the addition of 10 μM of NADH to pentane-treated envelope membranes (Fig. 5). This feature was still visible when the spectrum was recorded at 45 K, suggesting that it could be associated with a [2Fe-2S]¹⁺ center. Surprisingly, it was not observed after reduction with 5-deazaflavin/oxalate, which is able to reduce centers with a much lower redox potential than those reduced by NADH. This difference could be due to either to a very high affinity of the envelope [2Fe-2S]¹⁺ center for NADH (possibly via a dehydrogenase) or to a limitation in the accessibility of the center by deazaflavin. Since NADH is not, in general, a true physiological donor for most of the chloroplast enzymes, it is possible that the [2Fe-2S]¹⁺ center reduced by NADH could be localized in a compartment only accessible from the cytosol. Therefore, the protein associated with the [2Fe-2S]¹⁺ center can be located either on the outer envelope membrane or on the outer face of the inner envelope membrane.

Figure 5.

Effect of reduction with NADH on chloroplast envelope EPR signals. Pentane-treated envelope corresponding to 30 mg of protein/ml was used. EPR conditions: see Fig. 1B, except that the temperature was set at 15 K.

Table 1 summarizes some effects of NADPH and/or oxygen on the amplitude of the semiquinone signal and of the signal associated with the [4Fe-4S]¹⁺ center. First, addition of 10 μM of NADPH (in the presence of oxygen) to envelope membranes led to the appearance of the g = 1.921 signal. This signal was also observed after reduction with 5-deazaflavin/oxalate (Fig. 4), but not after reduction by NADH (Fig. 5). Another effect of NADPH addition (in the absence of oxygen) was the decrease of the semiquinone signal at g = 2.003 (Table 1). These results suggest that the semiquinone is reduced to its quinol form by NADPH, probably owing to NADPH–semiquinone reductase activity. In contrast, when NADPH was added in the presence of saturating levels of oxygen, the amplitude of the semiquinone signal increased significantly (Table 1). Finally, when envelope membranes were placed for 20 min under a stream of oxygen in the absence of NADPH, the amplitude of the semiquinone signal at g = 2.003 increased at least by a factor of 4 (Table 1). Although pure α-tocopherol is indeed able to generate an EPR-active α-tocopheroxyl radical (not shown), the possibility that the oxygen-induced increase of the g = 2.003 signal could be due to the formation of an α-tocopheroxyl radical is rather unlikely. We observed that the g = 2.003 signal formed during the oxygenation of envelope membranes was almost completely abolished after further addition of NADPH under argon (Table 1), in good agreement with our previous observations of the reduction of the semiquinone by this cofactor. In fact, α-tocopheroxyl radicals are recycled back to α-tocopherol with either ascorbate or reduced glutathione to generate either monodehydroascorbate or glutathionyl (GS^·) radicals. Furthermore, these radicals are themselves thought to be recycled by NADPH (35). Because neither ascorbate nor glutathione were present in our experiments, we can conclude that the oxygen-induced increase of the g = 2.003 signal is probably not due to the formation of an α-tocopheroxyl radical. More likely, this increase is due to an oxidation of quinol to semiquinone, probably owing to a quinol oxidase activity.

Table 1.

Effect of oxidation by oxygen and reduction by NADPH on g = 1.921 and g = 2.003 chloroplast envelope EPR signals

Signal g value	No addition	NADPH	NADPH + argon	NADPH + oxygen	Argon	Oxygen	Oxygen then NADPH + argon
1.921	≈0	++	≈0	++	≈0	≈0	≈0
2.003	++	+(+)	+	+++	≈0	+++++	+

The relative amplitude of the chloroplast envelope EPR signals is indicated by a more or less large number of plusses. 

Finally, NADH and NADPH are able to transfer two electrons, whereas iron–sulfur centers can accept only one. Therefore, a link between electron donors and acceptors is necessary: flavins that can transfer either one or two electrons (36) are good candidates for such a function. We therefore analyzed the flavin content of envelope membranes.

Analyses of the Flavin Content of Envelope Membranes.

Only between one-third and one-half of the envelope flavins could be extracted by a classical aqueous treatment, whereas they were all solubilized by 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) (33 mM). No amino acid-linked flavins were further released by proteolytic digestion of envelope proteins. This demonstrates that in envelope membranes a major proportion of the flavins is probably stabilized by proteins in a noncovalent manner. The FAD and FMN content of 17 different envelope preparations was analyzed: mean values of 300 and 45 pmol/mg protein, respectively, for FAD and FMN, were determined, making possible the presence of a flavosemiquinone radical in envelope membranes (see above).

DISCUSSION

EPR-detectable centers have been identified in chloroplast envelope membranes. First, envelope membranes contain semiquinone radicals. In vitro, the quinone seems to function mostly between the quinol and the semiquinone state. Envelope membranes probably contain several enzymatic activities involved in the formation and reduction of semiquinone radicals, namely a quinol oxidase (which catalyzes the oxygen-dependent conversion of quinol into semiquinone), an NADPH–quinone reductase (which catalyzes the reduction of quinone into semiquinone), and an NADPH–semiquinone reductase (which catalyzes the reduction of semiquinone into quinol). The semiquinone is probably stabilized because of the possible existence of an active quinol oxidase, which could transfer electrons from quinol to oxygen.

Second, envelope membranes contain a series of iron-containing electron-transfer proteins, but are certainly lacking any detectable cytochrome, which makes this membrane system quite unique among plant membranes. In addition to one (or several) [1Fe]³⁺ protein(s) which is/are associated with the EPR signal centered at g = 4.3, we observed an unusual Fe-S center, named X, associated with a signal centered at g = 2.057, which was reduced by dithionite but not by NADH or NADPH. Two additional Fe-S centers, a [4Fe-4S]¹⁺ center, associated with a feature at g = 1.921, and a [2Fe-2S]¹⁺ center, associated with a feature at g = 1.935, were detected after reduction, respectively, by NADPH and NADH. The [4Fe-4S] center, but not the [2Fe-2S] center, was also reduced by dithionite or 5-deazaflavin/oxalate. Signal Y, with a maximum at g = 2.017, remained unidentified. Third, envelope membranes contain protein-associated FAD and FMN, which could be responsible for the flavosemiquinone radical observed.

These electron carriers are likely to be associated into one or several electron-transfer chains. At this stage of the work, complete description of putative electron-transfer chains in envelope membranes is obviously speculative. However, a possible working hypothesis is shown in Fig. 6, which provides some reasonable clues for the interpretation of our results. Until thermodynamic profiles of both FAD and Q are analyzed, one cannot decide whether parallel electron-transfer chains branched at the flavin site are necessary, or if a single chain with both flavin and quinone sites functioning as n = 2 ↔ n = 1 converters (37) is present. However, some preliminary observations, such as the oxygen-dependent formation of semiquinone radicals and an oxygen-independent regeneration of the quinone from the semiquinone radical, could favor a branched electron chain.

Figure 6.

Scheme for a possible electron transfer chain in chloroplast envelope membranes. FAD/FADH^·/FADH₂ is the link between NADPH and the iron–sulfur centers X and [4Fe-4S]¹⁺ (steps 1–3). NADPH–quinone reductase (step 4) and NADPH–semiquinone reductase (step 5) allow the oxidation of FADH₂ and FADH· by the iron–sulfur centers X and [4Fe-4S]¹⁺. In this scheme the [4Fe-4S] center is shown to be involved in the reoxidation of FADH₂ into FADH^· (step 2) and the X center in the reoxidation of FADH· into FAD (step 3). The symmetric situation (i.e., steps 2 and 3 involving the centers X and [4Fe-4S], respectively) is also possible. Regeneration of the semiquinone radical by oxygen is due to a quinol oxidase (step 6). Oxidation of the semiquinone radical to its quinone form (step 7) has not yet been demonstrated.

Flavins, and especially FAD, are able to receive two electrons from NADPH. FAD is reduced by NADPH in a single step (step 1) into FADH₂ (Fig. 6). FADH₂ is then reoxidized into FAD in a two-step process via FADH^· (steps 2 and 3). Two of the iron–sulfur centers, i.e., X and [4Fe-4S], are probably involved in this reoxidation process since quinones or semiquinones are, in general, not directly involved in such reactions. For instance, the [4Fe-4S] center could be involved in the reoxidation of FADH₂ into FADH^· (step 2) and the X center in the reoxidation of FADH^· into FAD (step 3) or vice versa.

NADPH–quinone reductase (step 4) and NADPH–semiquinone reductase (step 5) are, respectively, involved in the formation of the semiquinone radical and of the quinol form. In the presence of NADPH, the (probably) high turnover of center X does not permit to follow its reduction, whereas the [4Fe-4S] center and the quinone are both observed to be at least partially reduced, suggesting that the quinone is involved in the oxidation of center X (step 4) and thus converted into semiquinone. On the other hand, the semiquinone is likely to be reduced to the quinol state by oxidation of the reduced [4Fe-4S]¹⁺ center (step 5).

The last step is the oxygen-dependent formation of semiquinone radicals (step 6). In general, cytochromes are the redox intermediates between quinol and oxygen (24, 38), since the difference between the redox potential of the quinone/semiquinone (0 V) and of the O₂/H₂O (+0.81 V) couples is too large for a direct link to exist between these molecules. Since envelope membranes are devoid of cytochrome, enzymes such as a quinol oxidase (see above) could play a role in this process. For instance, in plant mitochondria, the quinol oxidase that could be linked to the cyanide-insensitive pathway is an EPR-silent flavoprotein (39, 40). Finally, we have no information about the mechanisms that could be involved in the regeneration of the quinone from the semiquinone radical (step 7); very likely, the natural electron acceptor is missing. Since we never observed any decrease of the g = 2.003 signal in the presence of oxygen, the participation of oxygen in this process is rather unlikely.

The physiological significance of our results could be first related to the presence of desaturase activities in envelope membranes. For instance, phytoene desaturase is a FAD-containing flavoprotein (41, 42), its activity requires quinones, oxygen, and factors regulating the redox state of quinones (2, 3, 43). Concerning fatty acid desaturation, the mechanism involved in chloroplasts is probably very different from the microsomal desaturase system. In microsomes, the immediate electron donor to the desaturase, an iron-containing protein, is a b₅ cytochrome, which in turn can receive electrons from two different flavoproteins, one using NADH as a cofactor the other NADPH (4). In contrast, it seems that lipid-linked desaturation in chloroplasts requires NADPH and O₂ as the final electron acceptor and reduced ferredoxin as the source of the additional two electrons necessary to reduce O₂ to H₂O (4). No cytochrome seems to be involved in this process, which is in agreement with our results. Membrane-bound desaturases from plants, cyanobacteria, yeast, and mammals all contain homologous regions with the general sequence His-Xaa-Xaa-Xaa-His, which may provide metal-chelating ligands contributing to the binding of oxygen in the reaction center (5, 44). For phytoene desaturase and for the n-6 lipid-linked desaturase, the participation of O₂ as the final electron acceptor suggests that they have lower redox potentials than oxygen. They could therefore be involved in the oxygen-dependent oxidation of quinol to semiquinone (Fig. 6, step 6). Physiological and biochemical evidence for such a mechanism to operate in chloroplast envelope membranes remain to be obtained. In addition, desaturation is not the only envelope reaction that requires an electron flow. For instance, in the light, protons are pumped by isolated intact chloroplasts not only from the stroma into the intrathylakoid compartment, but also across the envelope into the external medium (1, 7). The processes involved in the regulation of the stromal pH during photosynthesis are still unknown (8, 9) and our results could provide some clues toward an understanding of this regulation; for example, NADPH–quinone reductase of the chloroplast envelope could be involved in the transfer of protons from the stroma to the cytosol.