Mobile hydrogen carbonate acts as proton acceptor in photosynthetic water oxidation

Significance

Photosynthesis by cyanobacteria, algae, and plants sustains life on Earth by oxidizing water to the O₂ we breathe and by converting CO₂ into biomass we eat, burn, or use otherwise. Although O₂ production and CO₂ reduction are functionally and structurally well separated in photosynthetic organisms, there is a long debated role of CO₂/ in water oxidation. Here we demonstrate that acts as mobile acceptor and transporter of protons produced by photosystem II, and that depletion of leads to a reversible down-regulation of O₂ production. These findings add a previously unidentified component to the regulatory networks in higher plants, algae, and cyanobacteria and conclude the long quest for the function of CO₂/ in photosynthetic water oxidation.

Abstract

Cyanobacteria, algae, and plants oxidize water to the O₂ we breathe, and consume CO₂ during the synthesis of biomass. Although these vital processes are functionally and structurally well separated in photosynthetic organisms, there is a long-debated role for CO₂/ in water oxidation. Using membrane-inlet mass spectrometry we demonstrate that acts as a mobile proton acceptor that helps to transport the protons produced inside of photosystem II by water oxidation out into the chloroplast’s lumen, resulting in a light-driven production of O₂ and CO₂. Depletion of from the media leads, in the absence of added buffers, to a reversible down-regulation of O₂ production by about 20%. These findings add a previously unidentified component to the regulatory network of oxygenic photosynthesis and conclude the more than 50-y-long quest for the function of CO₂/ in photosynthetic water oxidation.

Oxygenic photosynthesis in cyanobacteria, algae, and higher plants leads to the reduction of atmospheric CO₂ to energy-rich carbohydrates. The electrons needed for this process are extracted in a cyclic, light-driven process from water that is split into dioxygen (O₂) and protons. This reaction is catalyzed by a penta-µ-oxo bridged tetra-manganese calcium cluster (Mn₄CaO₅) within the oxygen-evolving complex (OEC) of photosystem II (PSII) (1–4). The possible roles of inorganic carbon, , in this process have been a controversial issue ever since Otto Warburg and Günter Krippahl (5) reported in 1958 that oxygen evolution by PSII strictly depends on CO₂ and therefore has to be based on the photolysis of H₂CO₃ (“Kohlensäure”) and not of water. These first experiments were indirect and, as became apparent later, were wrongly interpreted (6–8). Several research groups followed up on these initial results and identified two possible sites of C_i interaction within PSII (reviewed in refs. 9–12). Functional and spectroscopic studies showed that facilitates the reduction of the secondary plastoquinone electron acceptor (Q_B) of PSII by participating in the protonation of . Binding of (or ) to the nonheme Fe between the quinones Q_A and Q_B was recently confirmed by X-ray crystallography (3, 13, 14). Despite this functional role at the acceptor side, the very tight binding of to this site makes it impossible for the activity of PSII to be affected by changing the C_i level of the medium; instead inhibitors such as formate need to be added to induce the acceptor-side effect (15). Consequently, the water-splitting electron-donor side of PSII has also been studied intensively (for recent reviews, see refs. 11 and 12). Although a tight binding of C_i near the Mn₄CaO₅ cluster is excluded on the basis of X-ray crystallography (3, 14), FTIR spectroscopy (16), and mass spectrometry (17, 18), the possibility that a weakly bound affects the activity of PSII at the donor side remains a viable option (reviewed in refs. 10 and 19).

In the present study using higher plant PSII membranes, we specifically evaluate a recently suggested role of weakly bound , namely, that it acts as an acceptor for, and transporter of, protons produced by water splitting in the OEC (20–22).

Results

Time-resolved isotope-ratio membrane-inlet mass spectrometry (TR-MIMS) allows simultaneous on-line detection of various dissolved gas molecules with high sensitivity. This method, in which the gas molecules contained in the sample enter the vacuum of the mass spectrometer via pervaporation through a gas-permeable membrane, is therefore ideally suited to test the effect and mechanism of interaction with PSII (23, 24). The TR-MIMS experiments presented below were performed either in on-line or off-line mode. The commonly used on-line approach leads to a strong depletion of the samples owing to the consumption of CO₂ by the mass spectrometer and the interconversion of all C_i species. In contrast, the off-line approach allows the experiments to be performed at well-defined levels (ambient or depleted). The two techniques are illustrated schematically in Fig. 1.

Fig. 1.

Schematic view of the on-line and off-line TR-MIMS setup. For on-line experiments PSII membranes were loaded into the home-built MIMS cell and degassed for 20–40 min until stable baselines for all detected gases were reached (Lower Right). This leads to a nearly complete depletion of the samples of C_i. The sample was then illuminated with xenon flashes or by continuous white light from a slide projector. In off-line measurements the PSII membranes were illuminated inside of gas-tight syringes, that is, at well-defined C_i levels (ambient or depleted), and then injected into buffer of identical composition and pH that was thoroughly degassed in the MIMS cell.

Effects of Mild Depletion on PSII Activity.

To confirm that is of functional relevance for photosynthetic water splitting under our mild -depletion conditions we monitored O₂ evolution of PSII by off-line TR-MIMS at ambient and C_i-depleted conditions . In these experiments an approximately 20-fold reduction in C_i levels was achieved by bubbling sample solutions, electron acceptors, and (97.6%) with CO₂-depleted air inside a sealed septum vial (25). The sample suspensions were then transferred into gas-tight syringes and illuminated with white light for 10 s to induce O₂ formation by water splitting in PSII. The O₂ production was assayed by injecting the sample into the MIMS cell that contained a buffer of the same composition and pH, but extremely low dissolved gas concentration owing to the constant removal of gas by the mass spectrometer via the membrane inlet. Comparison of the and traces in Fig. 2 shows that the O₂ yield of the sample is 20% lower than that of the sample. Reversibility of the effect was established by exposing samples for 20 min to air before illumination in the syringe ( samples). A similar inhibition was achieved if the C_i depletion was performed by bubbling with N₂, and a rapid (≤2 min) recovery was seen after C_i was added in absence of O₂ in form of NaHCO₃ powder (Fig. S1). These experiments confirm that HCO₃⁻ enhances the activity of PSII.

Fig. 2.

Dependence of water splitting in PSII on the C_i concentration of the medium. PSII membranes were illuminated with continuous white light of a slide projector for 10 s inside of gas-tight syringes and then injected into the MIMS cell. The trace was obtained at ambient C_i-concentration, whereas the trace was recorded at a 20-times-reduced C_i concentration that was reached by purging PSII sample with CO₂ absorber (PCDA220M; Puregas). The reversibility was demonstrated by exposing a sample for 20 min to air ( trace). The final chlorophyll concentration was 50 µg (Chl) ml⁻¹, the -enrichment was 10%, and the medium contained 1 mM MES (pH 6.3), 15 mM NaCl, and 2 mM K₃[Fe(CN)₆] and 0.25 mM 2-phenyl-p-benzoquinone (dissolved in DMSO) as electron acceptors. The average of two to three repeats is presented.

Acts as Proton Acceptor.

According to a recent proposal, the activating role of on PSII demonstrated above may be caused by the ability of to act as a base for protons produced by photosynthetic water oxidation (20–22). If this proposal is correct, then the transiently formed H₂CO₃ should decay according to Reaction 1 into water and CO₂:

The CO₂ should then be released from PSII simultaneously with O₂ and must be detectable with TR-MIMS, because a maximum of four CO₂ per O₂ can be expected in the absence of any other proton acceptors. Fig. 3 displays the results obtained by using the conventional on-line approach, that is, by illuminating PSII suspensions inside the MIMS cell by a series of 50 saturating xenon flashes fired at 2 Hz. Because all mass peaks give the same results, Fig. 3 displays for clarity only the traces at mass-to-charge ratios of m/z 36 and m/z 48 , which detect CO₂ and O₂ with about the same high sensitivity (see Fig. S2 for the other traces). Under these conditions only light-induced O₂ evolution, and no CO₂ formation, is seen (Fig. 3, experiment 1). However, a 200-fold amplification of the CO₂ trace reveals a very small light-induced rise in the m/z 48 trace (Fig. 3, Upper Inset). Although small, the effect was reproducible and a light-induced heating artifact could be excluded by the inhibition of PSII turnover with the herbicide 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) (Fig. 3, experiment 2). The more than 200-times smaller amplitude of the light-induced CO₂ production vs. O₂ evolution by PSII explains in part why the CO₂ formation by PSII was not reported previously. However, this very small CO₂ production may also call into question its biological relevance.

Fig. 3.

Simultaneous on-line TR-MIMS measurements of O₂ and CO₂ production by PSII. O₂ (blue traces) and CO₂ (green traces) evolution of dark-adapted spinach PSII membranes [0.3 mg (Chl)⋅mL⁻¹] induced by 50 saturating flashes at 2 Hz (trace 1). The arrows indicate the start of illumination. The inset displays the CO₂ evolution with 200-fold amplification. Trace 2 shows the same measurements after inhibition of PSII with the herbicide DCMU (80 µM). The amplitudes of the traces for CO₂ and O₂ can be directly compared, because our set-up detects both gases with nearly equal sensitivity. The measurements were performed in a buffer containing 3 mM MES (pH 6.3) in the presence of 1 mM K₃[Fe(CN)₆] as exogenous electron acceptor. Final H₂¹⁸O enrichment was 10%. One representative result out of two to three repeats is presented.

During the above on-line TR-MIMS experiments the samples need to be degassed in the MIMS cell for 20–40 min to reach a stable baseline. This leads to an almost complete removal of dissolved CO₂ from the solution (see data in Fig. 1). Owing to the interconversion of all C_i species (Reaction 1) this also results in very low concentrations of in the sample suspension (17), which may significantly reduce light-induced CO₂ formation by PSII compared with in vivo conditions. We therefore performed off-line TR-MIMS experiments in which the dark-adapted PSII membranes were illuminated at ambient levels of CO₂/ in gas-tight syringes with 100 xenon flashes (2 Hz). For the detection of the light-induced gas formation the samples were then injected rapidly (<3 s) after the illumination into the MIMS cell containing the same buffer (but degassed). Fig. 4 displays the light-minus-dark difference signals of the and gases (see original traces in Fig. S3). Under these more natural conditions, the CO₂ evolution reaches up to 20% of the O₂ evolution and can be clearly seen at the same sensitivity scale as O₂.

Fig. 4.

Simultaneous off-line TR-MIMS measurements of O₂ and CO₂ production by PSII. Dark-adapted PSII membranes (1 mg Chl/mL) were illuminated inside of a gas tide syringe with 100 xenon flashes and then injected into the MIMS cell either as fast as possible with our current set up (<3 s, solid lines) or after 1 min (dotted line). Displayed are the light-minus-dark difference signals of O₂ (blue traces) and CO₂ (green traces), normalized to the amplitude of O₂ release (for original traces see Fig. S3). The amplitudes of the traces for CO₂ and O₂ can be directly compared, because our set-up detects both gases with nearly equal sensitivity. Conditions: (A) 1 mM MES, pH 6.3; (B) 1 mM MES, pH 5.5; (C and D) 1 mM MES (solid lines) or 10 mM MES (dashed lines), pH 6.3. In addition to MES, the medium contained 15 mM NaCl, 15% , and 2 mM K₃[Fe(CN)₆] as electron acceptor. Zero levels are off set for clarity of presentation. In all panels the average of two to three repeats is presented.

To compare the CO₂/O₂ ratios at different concentrations in the sample buffer, we performed these off-line TR-MIMS experiments at pH 6.3 (Fig. 4A) and at pH 5.5 (Fig. 4B), which approximately corresponds to the luminal pH in photosynthetic organisms under illumination (26, 27). At pH 6.3 the CO₂ and concentrations are about equal. When changing the pH of the solution to pH 5.5 under conditions that allow equilibration with air (open system), the concentration of dissolved CO₂ in the medium remains unchanged compared with pH 6.3, but the concentration is reduced about six times (28) (SI Text and Table S1). Comparison of Fig. 4 A and B shows that the CO₂/O₂ ratio is more than two times lower at pH 5.5 compared with pH 6.3. Although the relative CO₂ evolution by PSII is not directly proportional to the concentration, these data follow the trend expected if CO₂ is produced according to Reaction 1.

The idea that acts as proton acceptor was probed in air-saturated media by comparing light-induced CO₂ evolution at 10-fold different MES buffer concentrations (1 mM and 10 mM; Fig. 4C). MES at 200 mM was previously used to detect the proton release pattern of PSII by isotope-editing FTIR spectroscopy (29) and is therefore known to be a good proton acceptor for PSII that can outcompete in its suspected role. A strongly reduced CO₂ and the same O₂ production were observed at 10 mM MES compared with 1 mM MES (Fig. 4 C and D), confirming this proposal. We note that this result excludes Warburg-type mechanisms (30) in which acts directly or indirectly as a substrate of PSII, because for such mechanisms a buffer capacity independent CO₂/O₂ ratio is expected.

The competition between buffer and for protons is further illustrated by the dotted traces in Fig. 4 A and B. They show that at 1 mM MES the CO₂/O₂ ratios are significantly lower if the PSII suspension is not injected rapidly (1–3 s) after illumination into the MIMS cells, but after waiting for some time (∼1 min). This phenomenon reflects the back-conversion (Reaction 1) of CO₂ into by equilibration with water and proton acceptors such as side chains of proteins and lipids head groups of the sample or with MES according to their respective pK_a values (6.15 for MES and 6.38 for at 20 °C) and the respective free base concentrations . It is noted that the O₂ levels do not change during the 1-min waiting time (Fig. S4). The CO₂ signals in Fig. 4 are therefore lower estimates of the initial production. We further illustrate this point in Fig. S5, where by using an automated flash-injection system we were able to observed CO₂/O₂ ratios >0.5 at 1 ms between illumination and injection. Importantly, the time-dependent equilibration shows that accepts protons produced by water splitting at sites within PSII that are not equally well accessible for the larger MES buffer molecules (i.e., that the CO₂ production is primarily a PSII-specific effect, rather than caused by a simple pH change of the medium).

Discussion

Interaction with PSII in Vitro.

The data in this study were obtained with isolated PSII membranes from spinach. This made it possible to study the effects of on PSII activity without the need to consider indirect effects from related processes such as the carbon sequestration mechanisms (CCMs) operational in many algae and cyanobacteria (31). Our results clearly demonstrate that illumination of higher-plant PSII membranes with visible light induces both O₂ and CO₂ evolution, and that the yield of CO₂ depends both on the concentrations of and MES in the sample suspension. Quantitation of the signals in Fig. 4A reveals the production of about four CO₂ molecules per PSII (Methods). Therefore, we can exclude the possibility that the light-induced CO₂ production is due to the release of from its binding site at the nonheme iron. Collectively our data demonstrate conclusively that exchangeable acts as a base for protons produced by PSII during water splitting.

Special pathways or channels have been discussed for guiding protons away from the water-splitting Mn₄CaO₅ cluster (32, 33). We show that can compete for protons from water splitting with up to 300 times higher MES⁻ concentration (at 10 mM MES) and that the CO₂ yield observed is time-dependent (Fig. 4 and Fig. S5). We therefore propose that is able to penetrate more deeply into PSII to accept protons than MES molecules, which are not present in vivo. Inspection of the 1.9- PSII crystal structure (3ARC code) reveals that should indeed be able to penetrate easily into the entrance regions of all postulated channels; some channels are even wide and flexible enough to allow glycerol molecules to penetrate far inside (32, 33). In contrast, access to these channels for the even larger MES molecules would be more restricted. Once accepts a proton it decomposes according to Reaction 1, and the CO₂ is released from the channel and is either detected by MIMS or converted back to by equilibration with water and buffer groups in the lumen, or with the artificially added MES buffer. This is illustrated schematically in Fig. 5. These equilibration processes likely lead to the delayed detection of CO₂ compared with O₂ that is visible in the on-line experiment displayed in Fig. 3.

Fig. 5.

Schematic representation of the function of in PSII.

We suggest that owing to the comparatively weak binding in the channels, which is required for the proposed function as mobile proton acceptor and carrier, does not have a high enough occupancy at any particular site to be observable in the crystal structures. It is noted that in vivo the buffer capacity of the lumen of chloroplasts was estimated to be 0.8–1.0 mM (pH 6.4–8.1) and suggested to be due mostly to stationary groups such as proteins and lipid head groups (34), and not to mobile buffers such as MES added here at low concentrations to minimize the bulk pH changes. In vivo the role of for the activity of PSII may therefore be even more pronounced than observed during the experiments shown in Fig. 2.

Biological Relevance.

Higher plants take up CO₂ for carbon fixation by Rubisco via opening their stomata. Therefore, they do not possess special CCMs. In contrast, algae and cyanobacteria acquire CO₂ in the form of from water, where its concentration is often very low. This requires CCMs that consist of several transporters to channel through the various membranes, and of carbonic anhydrases that allow for efficient /CO₂ interconversion (31, 35). The discovery of the PSII-bound Cah3 carbonic anhydrase (cia3 gene product) in Chlamydomonas reinhartii (22, 36, 37) raised the question of whether this carbonic anhydrase together with are required for PSII function, or whether they are part of the CCM mechanism that provides the nearby Rubisco with CO₂. Despite numerous experiments this point remained unresolved (21, 22, 35, 38). Although PSII membranes isolated from spinach also possess some carbonic anhydrase activity (6, 8), there is presently no evidence for a carbonic anhydrase directly connected to higher-plant PSII, and there is also no need for a CCM function (discussed above). Our finding that under continuous illumination the activity of PSII is nevertheless dependent on the concentration of the buffer therefore demonstrates that directly affects the activity of PSII. It is noted that the ability of to capture protons will affect the H⁺/e⁻ ratio owing to the loss of some CO₂ into the stroma, unless this is prevented by subsequent reformation of via equilibration with stationary buffers in the lumen, which is strongly accelerated by carbonic anhydrases.

Although cannot be viewed as a cofactor of water oxidation in PSII, our experiments do clearly demonstrate that it affects its activity. As such, a feedback loop can be envisioned between CO₂ fixation in the stroma and water oxidation at the luminal side of PSII: At high CO₂ levels in leaves the gas will be taken up into the alkaline stroma of the chloroplasts and trapped by conversion into . Stromal carbonic anhydrases convert it back into CO₂ when the residual CO₂ is consumed by Rubisco for carbon fixation, or when it gets lost by diffusion into the lumen, where it will be also partially trapped as . This leaching of CO₂ through the thylakoid membrane into the lumen allows PSII to “sense” the C_i level in the stroma. At high C_i levels in the chloroplasts many electrons can be used for CO₂ fixation, and in the lumen activates PSII so that it produces electrons at maximum efficiency. In contrast, at low CO₂/ levels fewer electrons are needed, and the down-regulation of PSII by 20% (or more) helps to mitigate an overreduction of the plastoquinone pool, which reduces the risk of producing reactive oxygen species that are known to be damaging to enzymes involved in photosynthesis (39–41). Future studies will be needed to elucidate whether such a feedback loop is indeed operational.

Summary

The light-induced CO₂ formation by PSII and the dependence of CO₂ evolution on the concentrations of C_i and added buffer in the medium demonstrate conclusively that acts as mobile proton acceptor of PSII. The data in Fig. 2 show that the lack of this component leads under “stress” (continuous strong light) to reduced oxygen evolution by PSII (21, 22). Because strictly alternating electron and proton removals from the OEC seem to be an inherent and crucial part of the mechanism of water oxidation in PSII (42, 43), it seems very likely that this function of developed early during evolution and can be found in all O₂-evolving organisms (44). This dependence of PSII activity on concentration may also allow for a feedback regulation with the Calvin–Benson–Bassham cycle, which uses the electrons and protons from water splitting for CO₂ reduction to biomass.

Methods

Sample Preparation.

PSII membranes (“BBY” type) were prepared from fresh leaves of Spinacia (S.) oleracea following protocols described earlier (45, 46). Typical rates of O₂ evolution for spinach PSII membranes were 400–500 µmol (O₂)⋅mg (Chl)⁻¹⋅h⁻¹. After isolation, the PSII membranes were frozen by dropping small aliquots into liquid nitrogen. The samples were then stored at –80 °C until used. Before the measurements, the samples were thawed in the dark on ice and diluted to the desired concentrations with MES medium containing 200–400 mM sucrose, 35 mM NaCl, and 1–10 mM MES/NaOH at pH 6.3 (or 5.5).

TR-MIMS Measurements.

TR-MIMS measurements were performed with an isotope ratio mass spectrometer (Finnigan^Plus XP; Thermo) that was connected via a dry ice/C₂H₅OH cooling trap to a home-built membrane-inlet cell of 150- or 175-µL volume (depending on stir bar used) (47). The sample in the cell was separated from the vacuum (3 × 10⁻⁸ bar) of the mass spectrometer via a 1-cm-diameter inlet that was covered by a 25-µm-thick silicon membrane (MEM-213; Mempro) resting on a porous Teflon support (Ø 10 mm). In the mass spectrometer the gases were ionized by electron impact and separated by a magnetic sector field into a seven-cup Faraday detector array for simultaneous detection of (m/z 32), (m/z 34), and ³⁶Ar (m/z 36), ⁴⁰Ar (m/z 40), (m/z 44), (m/z 46), and (m/z 48). The simultaneous recording of argon (m/z 40) served as a control. The MIMS cell was thermostated to 20 °C. During the measurements the sample suspensions were stirred constantly at high speed with a magnetic stir bar. Before the assays the sample suspensions (or buffers) were degassed for 20–40 min until an only slightly sloping baseline was reached. On-line monitoring of light-induced O₂ and CO₂ evolutions (Fig. 3) was done by illumination of the dark-adapted PSII membranes inside the MIMS cell with a train of 50 short saturating flashes (2 Hz) provided by xenon flash lamp (LS-1130–4, ∼5-µs half-width; Perkin–Elmer). Off-line measurements of light-induced O₂ and CO₂ evolutions were performed by excitation of the dark-adapted PSII membranes inside a gas-tight 250-µL Hamilton syringe and subsequent injection of the illuminated sample into the cell filled with the thoroughly degassed buffer solution of the same composition and pH. For the experiments displayed in Fig. 4 50-µL PSII sample suspensions [1 mg (Chl)⋅mL⁻¹] were illuminated with 100 xenon flashes (PS 302, 2 Hz, light pack FY-604 ∼15-µs half-width; EG&G). For controls, 50-µL aliquots of dark-adapted PSII sample suspensions were injected into the MIMS cell (Fig. S3). The light-induced O₂ and CO₂ signals were calibrated by the injection of defined volumes of air-equilibrated buffer (284 µM O₂ and 15.4 µM CO₂ at 20 °C, pH 5.0) into the same medium (www.engineeringtoolbox.com/oxygen-solubility-water-d_841.html and http://co2now.org/current-co2/co2-now/). The CO₂/PSII ratio was determined by dividing the calibrated CO₂ concentration by the PSII reaction center (RC) concentration, assuming 250 Chl/RC. The analysis of the MIMS spectra was performed by using Origin software.

CO₂/-Depletion Procedure and Monitoring of C_i Levels.

C_i was removed from the PSII sample suspension containing and electron acceptors by means of 30-min flushing with CO₂-free air that was generated by a Puregas PCDA220M instrument, in which compressed air is directed through a desiccant chamber containing an adsorbent material for CO_2.This C_i-depletion procedure led to a ∼20-fold decrease of the CO₂ level (∼0.8 µM) without decreasing the oxygen level compared with samples containing ambient CO₂ and O₂ levels owing to exposure to air ( media; ∼15.4 µM). To avoid contamination of C_i-depleted media with atmospheric CO₂, the depletion procedure and all following sample handling and incubation steps were performed inside sealed septum vials. This procedure resulted in our -depleted PSII membranes. For reversibility measurements the -depleted suspension of PSII membranes (∼80 µL) was incubated for 20 min in air. For a better equilibration with atmospheric CO₂, the sample suspensions were continuously stirred in a septum vial with an opened cap.

The CO₂ and O₂ concentrations in CO₂-depleted air were determined by monitoring the oxygen and carbon dioxide content of the and sample solutions. For this, the 50-µL aliquots of the and media were injected into degassed MES buffer (pH 6.3) in the MIMS cell.