A Glutathione Redox Effect on Photosynthetic Membrane Expression in Rhodospirillum rubrum^,

Abstract

The formation of intracytoplasmic photosynthetic membranes by facultative anoxygenic photosynthetic bacteria has become a prime example for exploring redox control of gene expression in response to oxygen and light. Although a number of redox-responsive sensor proteins and transcription factors have been characterized in several species during the last several years in some detail, the overall understanding of the metabolic events that determine the cellular redox environment and initiate redox signaling is still poor. In the present study we demonstrate that in Rhodospirillum rubrum, the amount of photosynthetic membranes can be drastically elevated by external supplementation of the growth medium with the low-molecular-weight thiol glutathione. Neither the widely used reductant dithiothreitol nor oxidized glutathione caused the same response, suggesting that the effect was specific for reduced glutathione. By determination of the extracellular and intracellular glutathione levels, we correlate the GSH/GSSG redox potential to the expression level of photosynthetic membranes. Possible regulatory interactions with periplasmic, membrane, and cytosolic proteins are discussed. Furthermore, we found that R. rubrum cultures excrete substantial amounts of glutathione to the environment.

INTRODUCTION

It is well established that the thiol-containing tripeptide glutathione (γ-Glu-Cys-Gly [GSH]) is the most abundant redox buffer in all eukaryotic and in many prokaryotic cells and that GSH is critically important for the defense of oxidative stress which inevitably is associated with respiratory life. Various functions and mechanisms where GSH is involved in bacteria are summarized in a recent review (22). During many of these reactions, two molecules of GSH are oxidized to glutathione disulfide (GSSG), and the reduced form has to be recycled by the enzyme glutathione reductase at the expense of NADPH. A signaling role for GSH, however, is thus far not readily established. In bacteria, the reversible formation of intramolecular or intermolecular disulfide bonds of cysteines of regulatory proteins appears to be a common motif in molecular redox switches. For example, the ArcB sensor of Escherichia coli or the PpsR- and RegB-type regulators in facultative anoxygenic photosynthetic bacteria undergo reversible thiol/disulfide switches when exposed to different redox conditions in vitro (21, 32). The latter group of bacteria (Rhodospirillaceae) has contributed significantly to the recent paradigms of redox signaling and control of gene expression due to its facultative phototrophic lifestyle, which allows these bacteria to adapt their energy metabolism to the redox conditions of the environment. Under conditions of limiting oxygen, the Rhodospirillaceae induce the biosynthesis of an extensive system of pigmented intracytoplasmic membranes (ICM) which harbor photosynthetic reaction centers and light-harvesting complexes. The major environmental factor that determines the amount of ICM is the availability of oxygen, since aerobic growth conditions repress the transcription of photosynthetic genes (puf, puc, bchl, crt, and others) and the formation of ICM (for a review, see reference 2).

In Rhodobacter capsulatus, the membrane-bound histidine kinase RegB is part of a global redox-responding two-component regulatory system that is involved in controlling gene expression for ICM biosynthesis and many other genes related to energy metabolism when oxygen becomes limiting (6). The redox signal for silencing the kinase activity of RegB has been identified to originate from the membranous ubiquinone (UQ) pool (33). In addition, a reversible intermolecular thiol/disulfide switch within the cytoplasmic region of RegB is critical for activity and implicates the perception of cytosolic redox signals in addition to membrane-localized quinones (32). In these molecular aspects, RegB is similar to the global redox sensor ArcB of E. coli, which was the first two-component histidine kinase for which a regulatory effect of UQ was demonstrated (10) and where the oxidation of critical Cys residues by UQ to an intramolecular disulfide bond was suggested based on in vitro experiments (21). In both cases, however, the cellular reductant that reconverts the thiol/disulfide switch to produce the active kinase conformation under low (anaerobic or microaerobic)-oxygen conditions has not been identified yet.

A second mechanism of control of photosynthetic gene expression is mediated by conserved cytosolic repressor proteins designated PpsR in Rhodobacter sphaeroides (7) and CrtJ in Rhodobacter capsulatus (28). Here, the importance of intramolecular disulfide bond formation as redox-sensing mechanism is somewhat controversial. Masuda et al. (23) determined the midpoint potential (E_m) for the disulfide bond formation of CrtJ to be −180 mV (at pH 7.0) by redox titration with GSH/GSSG and also demonstrated that the intramolecular disulfide bond occurs only in aerobic and not in anaerobic cells. In contrast, however, Cho et al. (4) reported that in R. sphaeroides cells, both critical cysteines of PpsR constantly exist in the reduced thiol form irrespective of the oxygen and light conditions. Strikingly, the E_m value of PpsR, determined by Kim et al. (18) is much more negative than CrtJ with a value of −320 mV.

Although the cellular reductant for reconverting the above-mentioned disulfide proteins to the thiol conformation has thus far not been addressed explicitly, an involvement of the GSH-glutaredoxin system has been shown in R. capsulatus (19). However, studies of the effects on other cellular redox systems or a quantitative estimation of the cellular redox potential of the cytoplasm was not conducted in that study, and it is still a compelling question how the occurrence of thiol/disulfide switches can function in the reducing environment of the cytoplasm.

In the present study we determined the cellular [GSH]/[GSSG] ratio as a measure of the cellular redox environment in Rhodospirillum rubrum. The starting point was a series of ICM induction experiments closely following the study of Malpica et al. (21) for the ArcB redox sensor of E. coli, which could be activated by externally supplied dithiothreitol (DTT) but not by GSH. A similar pattern would provide more evidence about the existence of analogous sensors controlling photosynthetic gene expression in R. rubrum. In R. rubrum it is particularly interesting that the amount of ICM is dramatically affected by the nature of the carbon growth substrate, a phenomenon still not completely understood at the metabolic level. A growth medium containing succinate and fructose as carbon sources (M2SF medium) results in strongly elevated expression of ICM under semiaerobic dark conditions, reaching ICM levels which are equivalent to phototrophic cultures grown at low light intensities (11).

Unlike RegB, a homologue of the cytosolic PpsR repressor has been identified in R. rubrum in a previous study (1). However, the midpoint potential of this protein and its importance for redox regulation in R. rubrum has up to now not been experimentally resolved. In addition, for the different growth media, a correlation of UQ redox states and expression levels of ICM has been demonstrated previously (13), which suggests that a quinone-sensing signaling systems might be present also in R. rubrum.

MATERIALS AND METHODS

Bacterial organism and growth conditions.

Experiments with R. rubrum were performed with strain ATCC 11170. R. sphaeroides and R. capsulatus were used for comparison (see the supplemental material). For growth experiments, M2S and M2SF media as described in reference 11 were used. M2S contains succinate as the sole carbon source; M2SF offers two carbon substrates, fructose and succinate. Bacteria were grown in shaker flasks with three baffles on a Certomat BS1 rotary shaker (Sartorius, Goettingen, Germany). The oxygen supply was set by varying the ratio between surface and culture volume. For semiaerobic cultures flasks were filled with half of their maximal volume, and for aerobic growth only 1/5 of the total volume was used. Photosynthetic cultures were grown in filled Pyrex flasks illuminated with tungsten light bulbs as described previously (12).

Quantification of cell growth and cell volume.

Cell density was monitored at 660 nm, and the production of photosynthetic membranes was determined by measuring the absorbance of photosynthetic reaction centers and light-harvesting complexes at 880 nm with a UV/Vis spectrophotometer (V-670; Jasco, Tokyo, Japan) in a 1-cm-path-length cuvette. For determination of the cellular dry weight (CDW), 10 ml of culture broth was harvested, washed with 0.98% NaCl (wt/vol), and then centrifuged for 10 min at 5,000 × g, and the cell pellet was subsequently lyophilized in a freeze-dryer (Christ, Osterode am Harz, Germany) to dryness and weighed afterwards. From the measurements a conversion factor was derived where an optical density of A₆₆₀ of 1.0 corresponds to 0.35 g CDW.

The application of the Nernst equation (see below) requires that concentrations of conjugate redox pairs are considered as molar concentrations. For this purpose, the cell volume was estimated by determining the dry weight of 5 ml of wet cell paste after freeze-drying overnight in a lyophilizator (Christ). From the measurements, a conversion factor was derived where 1 g CDW corresponded to 2.29 ml of cell volume (average of three independent determinations with a standard deviation of 0.006; no correction was made for interstitial water).

Determination of GSH and GSSG.

A modification of the glutathione reductase cycling assay (34) was applied for determination of reduced GSH and oxidized GSSG. Samples for GSH determination were obtained by centrifugation of 2 ml of culture broth at 4°C and 5,000 × g for 10 min. The cell pellets were washed two times with NaCl (0.98%, [wt/vol]) and then frozen and stored at −80°C. For the extraction of intracellular GSH, 1.25 ml of 10% (wt/vol) sulfosalicylic acid (SSA) was added and, after resuspension, three freeze and thaw cycles followed. After another centrifugation step (as described above), the supernatant was stored at −80°C until measurement, and the pellets were extracted again to ensure that all GSH was extracted. For the measurement of GSH, acid extracts were diluted with assay buffer (see below) to achieve a final concentration of 1% (wt/vol) SSA. For reduced GSH, a simple colorimetric assay was used, which can be completed using a cycling assay by NADPH and glutathione reductase to determine also the GSSG content. Assay mixtures included assay buffer (100 μl of 0.1 M HEPES, 5 mM EDTA [pH 7.5]), 10 μl of DTNB [5,5′-dithiobis(2-nitrobenzoic acid)], and 60 μl of sample with either standard or blank (assay buffer). After thoroughly mixing the absorbance was measured at 412 nm with a PowerWave microplate spectrophotometer (Biotek, Winooski, VT). This value determines the amount of reduced GSH in the samples. To determine the total amounts of GSH and GSSG, 30 μl of 10 mM NADPH, including 10 U of glutathione reductase liter⁻¹ in assay buffer was added, and the reaction kinetics were recorded. The slope of the kinetics correlates with the total amount of glutathione in the sample. To calculate the amount of GSSG, the amount of GSH has to be subtracted from the total glutathione and divided by 2.

HPLC analyses.

The concentrations of succinate in the culture supernatants were determined by HPLC at 210 nm using a quaternary Agilent 1100 high-pressure liquid chromatography (HPLC) system (Agilent, Palo Alto, CA) equipped with a photodiode array detector. A total of 10 μl of the sample was injected and separated by reversed-phase chromatography on an Inertsil ODS (5 μm, 250-by-4.6 mm column; GL Sciences, Torrance, CA) and isocratic elution with 0.1 M ammonium dihydrogen phosphate (pH 2.6) at a flow rate of 1 ml min⁻¹. Ubiquinone-10 and ubiquinol-10 were determined by HPLC as described previously (13).

Calculations.

The redox potential at pH 7.0, E′ (mV) of the cellular GSH/GSSG pool was calculated according to the Nernst equation: E′ = E⁰′ + RT/nF × lnQ, where Q is the ratio [GSH_stc]²[GSSG]/[GSSG_stc][GSH]², with [GSH_stc] and [GSSG_stc] being 1 M concentrations (standard conditions). Note that the explicit consideration of [GSH_stc] and [GSSG_stc] results in consistency of the units. E⁰′ denotes the standard redox potential at pH 7.0 (−240 mV for the GSH/GSSG couple [29]), R is the gas constant (8.31 J K⁻¹ mol⁻¹), T is temperature (303 K), F is the Faraday constant (9.65 × 10⁴ C mol⁻¹), and n is the number of electrons transferred (n = 2 for GSH/GSSG). The Gibbs free energy for the reduction of GSSG by NADPH was calculated from the results of the present study and published data (13) according to the following equation: ΔG′ = ΔG^o′ + RTln([NADP⁺][GSH]²/[NADPH+H⁺] [GSSG]), with ΔG^o′ = nFΔE^o′.

RESULTS AND DISCUSSION

Effects of redox-active agents on photosynthetic membrane production.

The starting point of the present study was to get more evidence about the molecular basis for the strongly elevated ICM expression when R. rubrum is grown in M2SF medium under semiaerobic dark conditions (11–13). We thus examined the effects of externally supplied redox compounds on R. rubrum, following the experiments performed by Malpica et al. (21) for characterizing the redox-sensing mechanism of ArcB in E. coli. In that study, stimulation with externally supplied reductants resulted in different transcription levels of reporter genes with an ArcB/ArcA regulated promoter system. Application of the membrane permeating reductant DTT strongly increased reporter gene expression even at aerobic conditions. In contrast, reduced GSH had almost no effect, a result that was explained by the inability of the lipophobic GSH molecule to permeate through the E. coli cytoplasmic membrane, thus making it unable to interact with the cytosolic Cys residues of ArcB.

It therefore was very surprising that in our experiments, the response of ICM expression in R. rubrum to a similar treatment resulted in a pattern that was different from that of the previous results for ArcB. Figure 1 shows the effects of four different redox-active agents on ICM formation in a series of cultivations with R. rubrum. Cells were grown under dark conditions with succinate (M2S medium) or succinate/fructose (M2SF medium) as carbon source(s). In all cultures, during the initial aerobic growth phase, the high ICM levels of the semiaerobically grown inoculum cultures declined to low basal levels due to oxygen repression. When diffusion of molecular oxygen to the growing culture became limiting, the semiaerobic expression of ICM was induced, indicated by now increasing A₈₈₀/A₆₆₀ ratios. Generally, with succinate as a carbon source (M2S medium), the semiaerobic dark expression of ICM yields only low basal levels (A₈₈₀/A₆₆₀ = 0.7 to 0.8), whereas the combination of succinate and fructose (M2SF medium) results in much higher ICM levels (A₈₈₀/A₆₆₀ = 0.9 to 1.1) which when using conventional media are only observed in phototrophically grown cultures. As illustrated in Fig. 1, the addition of 1 mM GSH to a culture grown with succinate at semiaerobic conditions resulted in a strong increase in the biosynthesis of ICM, while growth was nearly unaffected. The elevated ICM expression was identical for applied GSH concentrations of 1 to 5 mM (data not shown). In contrast, the addition of GSSG had no effect (data not shown).

Fig. 1.

Effects of externally supplied redox compounds on growth (A₆₆₀) and ICM expression (A₈₈₀/A₆₆₀) in R. rubrum. GSH panels: ▵, control; ■, 1 mM GSH. DTT panels: ▵, control; ■, 0.5 mM DTT; ○, 1 mM DTT; ♦, 2 mM DTT. l-Cys panels: ▵, control; ■, 1 mM l-Cys. DEM panels: ▵, control; ▴, 1 mM DEM; ♢, 4 mM DEM. The compounds were added to batch cultures at the time points indicated by black arrows. All cultivations were performed with M2S medium, except for the DEM cultivations, which were performed with M2SF medium in order to assess the inhibiting effect of DEM on ICM production.

Figure 1 also shows that the effect of DTT was different than GSH. Cells exposed to 0.5 to 2 mM DTT were severely distorted in growth. In contrast to what we expected, and also in contrast to GSH treatment, ICM levels immediately heavily dropped upon the administration of DTT. The A₈₈₀/A₆₆₀ curves in Fig. 1 suggest that DTT exerted some kind of thiol stress on R. rubrum which, however, had a transient course. After ∼20 h, ICM expression recovered, and 0.5 mM DTT yielded levels identical to those seen with the GSH-stimulated cells. Cells exposed to 2 mM DTT maintained the lowered ICM levels for the rest of the cultivation. The administration of 1 mM l-Cys did not alter growth or ICM production. In addition to l-Cys, the two other amino acids, l-Gly and l-Glu, which compose the GSH tripeptide, when applied either as single compounds (1 mM) or all three amino acids together, had no effect (data not shown), thus indicating that the observed elevation of ICM was specific for GSH.

In a complementary experiment, the glutathione-depleting agent diethylmaleate (DEM) was added. DEM is conjugated with GSH by GSH–S-transferase (3), thus depleting the cellular pool of GSH but not of GSSG. Since we wanted to see whether DEM reduces ICM expression, we tested this compound with M2SF medium, which normally produces high levels of ICM in semiaerobic dark conditions. The administration of 1 or 4 mM DEM resulted in a pronounced inhibition of growth, and ICM formation was significantly lowered to basal levels. Treatment with H₂O₂ or diamide, which are expected to oxidize the GSH pool, resulted in a similar decrease in ICM levels (data not shown).

The elevated ICM levels in GSH-stimulated cells, are strongly reminiscent to the high A₈₈₀/A₆₆₀ ratio achieved by using the M2SF cultivation medium. When this medium was used, the already high amounts of ICM could be increased only marginally by additional treatment with GSH. Furthermore, under aerobic growth conditions, GSH administration to an M2SF culture induced basal ICM biosynthesis even in the presence of normally inhibitory concentrations of molecular oxygen (data not shown).

In conclusion, the data presented in Fig. 1 demonstrate that externally added GSH enhanced ICM production in R. rubrum, whereas treatment with GSH-depleting agents resulted in attenuated ICM production.

Determination of extracellular and intracellular GSH.

The effect of GSH supplementation raises the question about the cellular targets and molecular mechanisms underlying the observed stimulation of ICM expression. Taking the impermeability of the inner cytoplasmic membrane for GSH into account, as argued by Malpica et al. (21), it seems possible that the effect may be caused by interactions with periplasmic components and subsequent transmembrane redox signal transduction. In E. coli a periplasmically located dithiol oxidase, DsbA, catalyzes the transfer of electrons from the periplasmic compartment to the membranous UQ pool via the membrane protein DsbB during oxidative protein folding (16). In the R. rubrum genome sequence, four DsbA oxidoreductase genes and one DsbB homolog are present (see Table S1 in the supplemental material for gene annotations). Consequently, it appears possible that reduction of DsbA by externally supplied GSH would allow activation of the sensor kinases of the RegB/ArcB type indirectly by reduction of the UQ pool. We therefore determined the redox state of UQ₁₀ in cell extracts by HPLC-MS as previously described (13) after GSH treatment of the culture. In the control cultures the total pool of UQ₁₀ was 42.0% reduced, whereas upon administration of 5 mM GSH, the UQ pool was in fact more than 60% reduced. We note that GSH was not able to reduce UQ₁₀ directly in isolated chromatophores due to the inaccessibility of the polar GSH molecule to the lipophilic membrane compartment (data not shown). The findings described above would be in agreement with externally applied GSH leading to increased UQ sensor kinase activity by partial reduction of the UQ pool with transfer of reducing equivalents via known periplasmic disulfide bond formation proteins. The scenario also would not require GSH to enter the cytosolic space, thus satisfying the previous argument of Malpica et al. (21) for ArcB in E. coli. However, it should be noted that a number of studies have demonstrated the occurrence of GSH uptake by different transport systems (ABC transporters [CydDC] [27], ybiK/spt gene products [25], and yli gene products [31]) during the last several years, which allow active importation, as well as excretion, of GSH from E. coli cells. A protein BLAST search revealed the presence of homologous transport systems also in the R. rubrum genome sequence (see Table S1 in the supplemental material). We therefore had to consider additional mechanisms with a cytosolic mode of action of GSH and quantitatively analyzed the cellular GSH/GSSG pool after exposure to external GSH, as well as in untreated control cultures. The data summarized in Table 1 were obtained with untreated cells as a reference data set of variations of the cellular GSH/GSSG redox status in response to three different growth conditions: aerobic cultures (M2S medium, no ICM biosynthesis), semiaerobic cultures (M2SF medium, high-level expression of ICM), and anaerobic photosynthetic cells (M2S medium). Samples were taken from batch cultures at late exponential phase, assumed to represent pseudo-steady-state conditions. It was found that the total intracellular concentration of GSH (which is [GSH] + 2[GSSG]) varied, depending on the growth conditions, between 6.7 μmol g CDW⁻¹ and 7.48 μmol g CDW⁻¹ (Table 1), thus confirming the range of a previous determination by Fahey et al. (9) for R. rubrum, who, however, did not determine reduced and oxidized forms separately. In our cultivations, we were able to detect oxidized GSSG only when M2S medium and aerobic growth conditions were applied. The ratio of [GSH]/[GSSG] was 32.2, meaning that ca. 94% of the total amount of GSH was in the reduced thiol form even under aerobic conditions where ICM biosynthesis was repressed by oxygen.

Table 1.

GSH/GSSG concentrations, redox potentials, and levels of ICM in R. rubrum in different growth conditions^a

Condition	Mean ± SD		[GSH]/[GSSG]	E′ (mV)	ICM (A880/A660)
	GSH (μmol/g CDW)	GSSG (μmol/g CDW)
Aerobic	6.30 ± 0.35	0.20 ± 0.04	32.18	−209	0.55
Semiaerobic	5.64 ± 0.23	∼0.02*	∼300*	∼−236*	0.80
Anaerobic phototrophic	7.48 ± 0.96	∼0.01*	∼300*	∼−252*	1.50

^aAerobic cells were cultivated in M2S medium (no ICM expression), and semiaerobic cells were cultivated in M2SF medium (with ICM expression induced). Phototrophic cells were cultivated with M2S medium.

^*, The concentrations of GSSG were estimated by applying an assumed [GSH]/[GSSG] ratio of 300.

In addition to the steady-state values of Table 1, the time course of [GSH] and [GSSG] was monitored during semiaerobic batch cultivations with M2S and M2SF media (Fig. 2 and 3). In the control cultivations not exposed to external GSH, the intracellular GSH pool sizes stayed between 5 and 8 μmol g CDW⁻¹ throughout the cultivations in both medium compositions (Fig. 2D and 3D), which is in accordance with the data of Table 1. GSSG constantly was below the detection limit, indicating a completely (>99%) reduced intracellular GSH pool. Interestingly, starting at the exponential growth phase, GSH was found to accumulate in the extracellular growth medium, reaching a final concentration of 6 μM in M2S medium (Fig. 2C) and even 12.6 μM in M2SF medium (Fig. 3C), respectively. Although it has been described before that some bacteria excrete GSH in substantial amounts into the environment, we are not aware that this feature has hitherto been noticed in R. rubrum or any other photosynthetic bacterium. Eser et al. (8) found that ca. 30% of the total GSH outside the cell in E. coli cultures and high concentrations (up to 10 μM) of GSH accumulated in supernatants of logarithmically growing E. coli and Salmonella enterica serovar Typhimurium, according to Owens and Hartman (24). In all cases, the biological function of GSH excretion has thus far remained enigmatic.

Fig. 2.

GSH time courses in cell extracts (GSH_ic) and in the extracellular culture fluid (GSH_ec) during batch cultivations of R. rubrum with succinate (M2S) medium. Open symbols: control cultivation without externally supplied GSH. Solid symbols: GSH was added after 21 h. Squares, A₆₆₀; triangles, A₈₈₀/A₆₆₀; diamonds, GSH extracellular (GSH_ec); circles, GSH intracellular (GSH_ic); open diamonds, GSSG intracellular (GSSG_ic) in GSH-supplemented cells. The data points for GSH_ic are mean values of three cell extracts with the standard deviations indicated as error bars.

Fig. 3.

GSH time courses in cell extracts (GSH_ic) and in the extracellular culture fluid (GSH_ec) during batch cultivations of R. rubrum with succinate/fructose (M2SF) medium. Open symbols: control cultivation without externally supplied GSH. Solid symbols: GSH was added after 21 h. Squares, A₆₆₀; triangles, A₈₈₀/A₆₆₀; diamonds, GSH extracellular (GSH_ec); circles, GSH intracellular (GSH_ic); open diamonds, GSSG intracellular (GSSG_ic) in GSH-supplemented cells; asterisks, fructose; crosses, succinate. The data points for GSH_ic are mean values of three cell extracts with the standard deviations indicated as error bars.

Figure 2 shows data obtained with M2S medium, where 1 mM GSH was added. However, due to partial oxidation of the GSH preparation, the effective concentration of GSH was only ∼0.6 mM, and the stimulation of ICM expression was therefore weaker than in Fig. 1. However, it does not affect the major results of Fig. 2. In both cultivations depicted in Fig. 2 and 3, the administered GSH was depleted from the culture medium accompanied by rapidly rising intracellular GSH levels and by the induction of ICM expression. With M2SF medium, the intracellular concentration of GSH increased from about 10 μmol g CDW⁻¹ to 50 μmol g CDW⁻¹. (Fig. 3D). In contrast to untreated cells, GSSG was now detectable in cell extracts at about constant levels with 1.6 to 1.9 μmol g CDW⁻¹ throughout the cultivation. It is interesting that with M2SF medium, the onset of GSH uptake and concomitant increase in the intracellular GSH and GSSG was initiated only after fructose was consumed almost completely from the growth medium, while succinate was still present (Fig. 3C).

This finding suggests that during the initial growth period, fructose degradation yielded sufficient reducing equivalents to lower the cellular redox potential, thus inducing high ICM production and preventing GSH from entering the cell. After the consumption of fructose, the low redox potential required for high-level production of ICM was maintained by the influx of GSH.

In conclusion, the depletion of administered extracellular GSH in combination with the >5-fold increase in intracellular GSH is a clear indication that the cells were able to take up GSH by active transport mechanisms. From the data, a specific GSH uptake rate of 0.041 mmol h⁻¹ g CDW⁻¹ was calculated for the experiment with M2SF medium (Fig. 3). Despite the >5-fold increase in the intracellular GSH concentration after external GSH addition, the direct uptake of GSH cannot account for the nearly complete depletion of GSH from the cultivation medium. Although the intracellular increase of GSH was ∼40 μmol g CDW⁻¹, a much larger amount of more than 13 mmol g CDW⁻¹ vanished from the culture supernatant during the same time period (M2SF medium, Fig. 3). We do not know what happened to the majority of the added GSH; extracellular oxidation to GSSG, however, was not detected. One possibility would be that a large portion of the supplemented GSH was degraded by extracellular and/or intracellular enzymes such as γ-glutamylpeptidases (30) to fuel salvage pathways for amino acids. The successive uptake of fructose and GSH could therefore also be interpreted as a diauxic growth phenomenon where GSH is used as a growth substrate after the preferential consumption of fructose. However, the possible role as a source for carbon or amino acids is unlikely to account for the elevated ICM production because, as mentioned above, the administration of the individual amino acids of GSH did not affect growth or ICM levels. It is also important to note that in a cultivation where GSH (20 mM) was used as the sole carbon source, R. rubrum was unable to grow (data not shown). Furthermore, supplementation with GSH did not increase the growth rate or final cell densities (Fig. 1 to 3). It is therefore unlikely that the putative utilization of GSH as a source of amino acids accounted for the elevation of photosynthetic gene expression. The results rather suggest that the accumulation of GSH in the cytosolic compartment of R. rubrum alters the cellular redox potential, thus elevating redox-dependent ICM expression. Consequently, direct interactions of GSH with soluble redox-regulatory proteins have to be considered in addition to membrane or periplasmic signal transduction.

Assessment of the redox environment in the light of the cellular GSH pool and its implication for redox control.

The ratio of [GSH] to [GSSG] has frequently been used as a quantitative measure for the cellular redox state, and its reducing capacity can be expressed in terms of redox potential using the Nernst equation as a tool. On a Nernst scale, the concentrations given in Table 1 for aerobic growth correspond to a GSH redox potential E′ of −209 mV. Applying more reducing growth conditions (semiaerobic growth and anaerobic photosynthetic growth, respectively) resulted in even higher levels for GSH and the failure to detect GSSG in cell extracts. These results strongly suggest that the [GSH]/[GSSG] ratio was shifted toward the reduced state in these cells, and GSSG levels were below the detection limit of the applied analytical method. Comparison to published data for GSH redox states in other bacteria shows that the ratio of [GSH] to [GSSG] generally covers a range from 50 to 300 (8, 14, 15). Applying the lowest ratio of 300 to our semiaerobic and anaerobic photosynthetic cultures would yield theoretical ambient redox potentials of −236 mV and −252 mV, respectively. This rough estimation would mean that the total GSH/GSSG pool size is more than 99% reduced in semiaerobic and anaerobic cells, which is consistent with published data for other bacteria, e.g., R. sphaeroides (14). In R. capsulatus, [GSH]/[GSSG] ratios were virtually unaffected by the oxygen conditions and corresponded to −222 mV aerobically and −224 mV anaerobically (23).

The intracellular GSH/GSSG redox potential calculated from the data presented in Fig. 3 covers a span of 35 mV from the highest value of −182 mV in aerobic cells (calculated at t = 18 h), where no ICM were expressed, down to −217 mV in cells with induced ICM production after 45 h of cultivation time. An important consequence of applying the Nernst equation for estimating the cellular redox environment is that, while the redox potential of most redox couples, e.g., NAD(P)H/NAD(P)⁺ or UQH₂/UQ, is determined exclusively by the ratio of the reduced and oxidized species, the redox potential of the GSH/GSSG couple depends on both the reduced/oxidized ratio and the absolute concentration of GSH. The reason for this difference is the change in molarity when one GSSG is converted to 2 GSH with the implication that GSH enters the Nernst equation as a squared term ([GSSG]/[GSH]²) (29). An important consequence for the role of GSH as a cellular redox buffer is that comparatively small alterations of the redox status of the GSH pool result in larger changes in the overall reducing capacity than would occur with other (equimolar) conjugate redox pairs. For example, the 35-mV decrease in E_h after stimulation with GSH calculated from the data of Fig. 3 was accompanied by an ∼8-fold increase in the [GSH]/[GSSG] ratio. To achieve the equivalent 35-mV change in the reducing capacity, the [NADH⁺ + H⁺]/[NAD⁺] ratio, for instance, would have to change >15-fold instead.

A correlation of ICM content and the redox status of the GSH pool is apparent from the data presented in Table 1, Fig. 2, and Fig. 3. Redox states of −209 mV (Table 1) and −182 mV (calculated from Fig. 3) were found to correspond to ICM biosynthesis being repressed by oxygen (ratio A₈₈₀/A₆₆₀ = 0.55 in virtually unpigmented cells). Since on the other hand, high levels of ICM were produced at −217 mV (Fig. 3), we conclude that the critical cytosolic redox potential for inducing photosynthetic gene expression is at about −210 mV on a GSH/GSSG scale.

The cellular redox potential in R. rubrum has been found previously to be around 310 to 320 mV for the NADH/NAD⁺ couple and 330 to 335 mV for NADPH/NADP⁺ (13). Comparison of the redox potential of the GSH/GSSG pool to these data thus demonstrates that both redox systems are not in equilibrium under the conditions of living cells. Since in R. rubrum no glucose-6-phosphate dehydrogenase as major cellular source for providing NADPH + H⁺ is present (12), glutathione reductase probably receives its NADPH substrate mainly by pyridine nucleotide transhydrogenase (17) and thus ultimately oxidizes NADH by the reduction of GSSG. For semiaerobic conditions, this pattern can be supported now by experimental data, where the extent of reduction increases in the order NADH (46.5% reduced [13]) < NADPH (73.4% reduced [13]) < GSH (>99.9% reduced (the present study). With the available data now it can be calculated that the reduction of the GSH pool at the expense of NAD(P)H is thermodynamically favorable with a ΔG′ value of −13 kJ mol⁻¹, which is crucial for keeping the GSH/GSSG pool in the reduced state.

In summary, the data presented above demonstrate that the amount of expressed ICM responds specifically to the external supplementation with GSH and is correlated with the redox status and concentration of the cytosolic GSH/GSSG pool.

At present the cellular targets and mechanisms for GSH in R. rubrum are essentially unclear. A regulatory effect rather than only a metabolic role is, however, indicated by the above-mentioned facts that GSH as a sole growth substrate was not sufficient to support growth and that amino acids (Glu, Cys, and Gly) did not affect growth or ICM levels in control experiments.

Considering the published data in combination with genome sequence database analysis (see the supplemental material), a hypothetical network of interacting redox components of different cellular compartments emerges, as proposed in Fig. 4. GSH enters the periplasmic space via outer membrane porins, and reducing equivalents are transferred from periplasmic thiol-proteins to the membranous UQ pool. The partial reduction of UQ subsequently releases the inhibition of UQ sensor kinases, thus enhancing the expression of ICM components. The further transport of GSH into the cytosol subsequently lowers the cytosolic GSH/GSSG redox potential considerably (below −210 mV) enough to reduce regulatory disulfide proteins. Although the midpoint potential of the PpsR homologue in R. rubrum has not yet been determined and its actual importance for the observed GSH effect remains speculative, the most simple and straightforward interpretation includes a direct response of the redox status of this protein. That GSH is the cellular reductant for the conversion of CrtJ into the inactive thiol conformation has already been proposed by Masuda et al. (23) for R. capsulatus, based on the demonstration that GSH can reduce CrtJ disulfide bonds in vitro. If we assume redox equilibrium between GSH/GSSG and the regulatory proteins included in Fig. 4, both RegB (E_m = −294 mV [32]) and PpsR (E_m = −320 mV [18]) would be >90% in the oxidized disulfide form over the 35-mV span of GSH/GSSG in Fig. 3. In contrast, because of its higher midpoint potential, a CrtJ-type repressor (E_m = −180 mV [23]) would be affected significantly and would switch from 76% being in the inactive reduced thiol form at aerobic GSH/GSSG values to almost complete inactivation of the repressor, with 98% being reduced under semiaerobic conditions.

Fig. 4.

Hypothetical model of metabolic and regulatory interactions of externally applied GSH, resulting in enhanced expression of ICM. DsbB was omitted for clarity. THase, transhydrogenase; GR, glutathione reductase; PP, periplasmic space; CM, cytoplasmic membrane; CS, cytosol.

It is important to note here that in both R. capsulatus and R. sphaeroides, identical treatments with GSH did not result in elevated photosynthetic complexes (see the supplemental material). In accordance with this, the intracellular levels of total GSH did not accumulate after exposure to external GSH but were maintained at ca. 4.5 μmol g CDW⁻¹ (standard deviation, 0.06) in R. sphaeroides and at ca. 4.2 μmol g CDW⁻¹ (standard deviation, 0.3) in R. capsulatus (see the supplemental material). It therefore has to be assumed that these species differ from R. rubrum with respect to the mechanisms for the efficient uptake of GSH from the environment.

We are aware that the scenario outlined in Fig. 4 is probably much oversimplified and that further experimental verification is required, e.g., by investigating mutants where components of the redox network have been deleted. We also have ignored thioredoxins and glutaredoxins, which clearly are important players in the thiol/disulfide network (19, 26) and which may be interconnected in the outlined signal transduction pathways.

In addition, it also appears possible that the GSH/GSSG redox potential acts by modulating the activity of metabolic enzymes alternatively or in parallel to transcriptional effects. It is interesting in this respect that activation of the first enzyme of the tetrapyrrole biosynthetic pathway, δ-aminolevulinic acid synthase, was demonstrated to occur by the reduction of several Cys residues using low-molecular-weight thiols in vitro (5). The R. rubrum enzyme in total contains six Cys residues (see the supplemental material for gene annotation). A decrease in the cellular GSH/GSSG redox potential could therefore be important for the posttranslational regulation of bacteriochlorophyll biosynthesis, including the stimulation of this enzyme.

The GSH effect is strongly reminiscent of the similar ICM elevation with M2SF medium, which was also not found in these species (13). An intriguing possibility would be that in R. rubrum both growth conditions may essentially induce the same redox signaling pathways and lower the GSH/GSSG redox state either by reducing equivalents originating from fructose catabolism (via transhydrogenase and glutathione reductase) or directly from the accumulation of cellular GSH. The absence of the succinate/fructose (M2SF medium) effect in the Rhodobacter species, however, makes it likely that up to now unidentified regulatory proteins and circuits, not included in Fig. 4, are involved in R. rubrum. The identification of the target proteins of GSH in R. rubrum will be a subject of further studies.

We finally mention that the present study also has two implications for biotechnology. First, the application of GSH to cultivation raises the amount of ICM and could be used to enhance ICM-associated products such as photosynthetic pigments or membrane proteins. Second, since GSH itself has become a product of industrial interest (20), the levels of the total pool sizes could be attractive compared to species that were already explored for this application. At present, the fermentative production of GSH uses predominantly Saccharomyces cerevisiae, and a recent study reports that genetic engineering could improve GSH production with S. cerevisiae to final intracellular GSH concentrations of 80 mg liter⁻¹ (35). However, much higher product yields (up to 4,300 mg liter⁻¹) are listed in a review on biotechnological GSH production (20). In combination with recently established high-cell-density cultivation of R. rubrum (36), the data in Table 1 can be extrapolated to theoretical total yields of GSH of ∼130 mg liter⁻¹ intracellularly and 221 mg liter⁻¹ excreted into the environment. Optimization of strains and medium conditions might be used to further increase the productivity.