Time-resolved vibrational spectroscopy detects protein-based intermediates in the photosynthetic oxygen-evolving cycle

Abstract

Photosynthetic oxygen production by photosystem II (PSII) is responsible for the maintenance of aerobic life on earth. The production of oxygen occurs at the PSII oxygen-evolving complex (OEC), which contains a tetranuclear manganese (Mn) cluster. Photo-induced electron transfer events in the reaction center lead to the accumulation of oxidizing equivalents on the OEC. Four sequential photooxidation reactions are required for oxygen production. The oxidizing complex cycles among five oxidation states, called the S_n states, where n refers to the number of oxidizing equivalents stored. Oxygen release occurs during the S₃-to-S₀ transition from an unstable intermediate, known as the S₄ state. In this report, we present data providing evidence for the production of an intermediate during each S state transition. These protein-derived intermediates are produced on the microsecond to millisecond time scale and are detected by time-resolved vibrational spectroscopy on the microsecond time scale. Our results suggest that a protein-derived conformational change or proton transfer reaction precedes Mn redox reactions during the S₂-to-S₃ and S₃-to-S₀ transitions.

Time-resolved vibrational spectroscopy can detect chemical intermediates formed during enzymatic catalysis. Advantages include the technique’s exquisite structural sensitivity and its high temporal resolution. Recent advances in methodology and interpretation have produced insights into the catalytic mechanism in several biological systems (for examples, see refs. 1–4).

In this paper, we report the use of time-resolved IR spectroscopy to investigate the mechanism of photosynthetic water oxidation. Photosystem II (PSII) catalyzes the oxidation of water and the reduction of bound plastoquinone. Photoexcitation of PSII leads to the oxidation of the chlorophyll donor, P₆₈₀, and the sequential reduction of a pheophytin (Fig. 1A, reaction 1) and a plastoquinone, Q_A (Fig. 1A, reaction 2), in picoseconds. Q_A reduces Q_B to generate a semiquinone radical, Q_B⁻, on the microsecond time scale (Fig. 1A, reaction 3) (reviewed in ref. 5). A second photoexcitation leads to the reduction and protonation of Q_B⁻ to form the quinol Q_BH₂. The rate of reduction of Q_B is faster than the rate of reduction of Q_B⁻ (see ref. 6 and references therein), which gives rise to a characteristic period-2 oscillation in kinetics originating on the PSII acceptor side (7).

Fig. 1.

PSII electron transfer and representative IR results. (A) Diagram of electron transfer steps in PSII and the S state cycle of photosynthetic oxygen evolution. From the S₁ state, three saturating flashes (shown in red, green, and blue, respectively) are required to produce oxygen. (B) Flash dependence of transients recorded at 1,483 cm⁻¹ and 10°C from PSII, which contained 0.6 mM DCBQ and 7.2 mM potassium ferricyanide. Flash 1, generating the S₂ state, is shown in the red trace; flash 2, generating the S₃ state, is shown in the green trace; flash 3, generating the S₀ state, is shown in the blue trace; and flash 4, generating the S₁ state, is shown in the gray trace. The transients are displayed with vertical offsets, which serve to superimpose the traces at the first data point. Multiexponential fits to the data (Table 1) are shown superimposed.

The primary chlorophyll donor, P₆₈₀, oxidizes a tyrosine, Y_Z (Y161 in the D1 subunit), on the nanosecond to microsecond time scale (Fig. 1A, reaction 4). In turn, tyrosine Y_Z^· oxidizes the oxygen-evolving complex (OEC) on every flash (Fig. 1A, reaction 5) (8). Four sequential photooxidation reactions are required for oxygen production, and the oxidizing complex cycles among five oxidation states, called the S_n states, where n refers to the number of oxidizing equivalents stored (9). The rate of OEC oxidation slows as oxidizing equivalents are stored on the OEC (10, 11). This slowing gives rise to a characteristic period-4 oscillation in OEC kinetics (Fig. 1A). Recently, UV spectroscopy was used to detect an intermediate that could not proceed to the S₄ state at high oxygen pressure (12). X-ray spectroscopy was used to detect a lag in the reduction of Mn during the S₃-to-S₀ transition; this lag phase was attributed to a proton transfer event that precedes the Mn redox reaction (11). Structural models of PSII have been derived from x-ray diffraction at ≥3.0-Å resolution (13–17).

There are many proposed mechanisms for water oxidation (for recent reviews, see refs. 18 and 19). IR spectroscopy provides a method of distinguishing among proposed water oxidation mechanisms and probing new intermediate states formed during each of the S state transitions. To our knowledge, this report is the first study of the OEC with kinetic IR spectroscopy on the microsecond time scale. These studies show that a protein-derived intermediate state is formed on each S state transition.

Results

In Fig. 1B, the effects of multiple, saturating laser flashes on PSII samples containing 2,6-dichlorobenzoquinone (DCBQ) and potassium ferricyanide as electron acceptors are shown. The data exhibit an initial 1,483-cm⁻¹ bleach on the microsecond time scale on each flash. This initial bleach, which is limited by the instrument response time, is followed by a delayed increase in amplitude over baseline (Fig. 1B). Alterations in the kinetics as a function of flash number are observed in Fig. 1B and will be discussed below.

Control experiments in Fig. 2 address the origin of these IR transients. The green trace was acquired from oxygen-evolving PSII containing DCBQ and ferricyanide (repeated from Fig. 1B). As a negative control, the red trace in Fig. 2 was acquired on PSII, which contained an OEC, 3-(3,4-dichlorphenyl)-1,1-dimethylurea (DCMU), and no exogenous acceptors. DCMU is an inhibitor that will block electron transfer to Q_B (20) (Fig. 1A, reaction 3) and cause transfer of any electrons on Q_B back to Q_A (21). This reequilibration, coupled with multiple flash illuminations, will result in a majority of closed centers containing Q_A⁻ in the dark, which cannot undergo a stable P₆₈₀⁺Q_A⁻ charge separation. As observed by comparison of the red and green traces, an IR transient, which is the average of multiple flashes in a DCMU-treated sample (Fig. 2, red trace), does not show the bleach that is observed in the presence of the electron acceptors (Fig. 2, green trace). This result demonstrates that the bleach is caused by PSII electron transfer events in open PSII centers.

Fig. 2.

Control transients recorded at 1,483 cm⁻¹ and 10°C from PSII. In the green trace, OEC-containing samples contained 0.6 mM DCBQ and 7.2 mM potassium ferricyanide (repeated from Fig. 1B, flash 1). In the red trace, OEC-containing samples contained 5 mM DCMU alone. In the beige trace, OEC-inactivated samples contained 7.5 mM hydroxylamine and 100 μM DCMU. In the blue trace, OEC-inactivated samples contained 0.6 mM DCBQ and 7.2 mM potassium ferricyanide. For the red, blue, and beige traces, transients (5, 24, and 2, respectively) were averaged. Multiexponential fits to the data (Table 1) are shown superimposed.

Experiments were conducted on PSII from which the OEC had been removed with 1.8 M alkaline Tris and which contained hydroxylamine and DCMU (Fig. 2, brown trace). In this hydroxylamine- and DCMU-treated sample, the P₆₈₀⁺Q_A⁻ charge-separated state will be produced because reactions 3 and 4 in Fig. 1A are blocked (22, 23). The production of this state has been shown to be reversible after an extended incubation with hydroxylamine (22, 23). In agreement with these previous studies, we found that extended incubation reproducibly gave a photo-induced negative 1,483-cm⁻¹ signal, the production of which was reversible. A multiexponential fit to the transient gave three time constants equal to 5 ms (46%), 470 μs (19%), and 20 μs (35%). Previously, time constants of ≤20 and ≈200–300 μs have been reported under these conditions and assigned to rates of P₆₈₀⁺Q_A⁻ decay (22, 23). The observation of a 5-ms component in our experiments may be due to a difference in sample preparation. In some hydroxylamine- and DCMU-treated samples, we observed a small positive signal, which we attribute to an incomplete block of Y_Z-to-P₆₈₀⁺ electron transfer, as previously proposed (22, 23). When this signal was observed, longer incubation of the sample with hydroxylamine and DCMU gave the reproducible negative signal (Fig. 2, brown trace). The data acquired on hydroxylamine- and DCMU-treated samples (Fig. 2, brown trace) suggest that the initial bleach in the OEC sample (Fig. 2, green trace) is due to the generation of Q_A⁻.

As an additional control, DCBQ and ferricyanide were added to a Mn-depleted sample (Fig. 2, blue trace). In this sample, the OEC is inactivated with 1.8 M alkaline Tris (24), Y_Z is the terminal electron donor (25, 26), and reaction 5 in Fig. 1A is blocked. In this sample, Y_Z reduces P₆₈₀⁺ with a lifetime in the microsecond time domain (25, 26). The blue trace in Fig. 2 shows an immediate increase in amplitude, as opposed to the decrease in amplitude observed in oxygen-evolving preparations (green trace). We attribute the 1,483-cm⁻¹ amplitude increase to a contribution from Y_Z^· in these OEC-inactivated samples. This Y_Z^· assignment is in agreement with the previous assignments of MacDonald et al. (27), which were derived from analysis of the photoaccumulated Y_Z^· spectrum and have been supported by additional studies (reviewed in refs. 28 and 29).

This Y_Z^· assignment is also supported by fits to the 1,483-cm⁻¹ transient (Fig. 2, blue trace), which gave three time constants equal to 34 ms (32%), 1.8 ms (14%), and 95 μs (54%). Time constants of 50 ms (40% amplitude) and 3 s (60%) have previously been reported for the decay of the Y_Z^· UV signal (30). The 3-s time constant would not be observable with our data acquisition conditions. In Mn-depleted PSII, the relatively intense, positive 1,483-cm⁻¹ absorption from Y_Z^· masks the less intense bleach, which is expected from the formation of the P₆₈₀⁺Q_A⁻ charge-separated state. In OEC-containing PSII, the spectrum of Y_Z^· must be shifted away from the 1,483-cm⁻¹ observation frequency.

Comparison of these controls suggests that the initial bleach in the OEC-containing sample is due to the generation of the Q_A⁻ state (Fig. 1A, reactions 1 and 2). Interestingly, the initial amplitude shows period-2 dependence (Fig. 1B). If Q_A⁻ and Q_B⁻ have a similar contribution at this wavelength, then on this time scale, the acceptor side contribution will remain relatively stable. Therefore, we hypothesized that the slower processes after the initial bleach are due to an amino acid side chain that is close to or ligating the Mn cluster in the OEC. To test this hypothesis, transients were fit with a double exponential function between 12 μs and 26 ms (Fig. 1B). As observed in Fig. 3 and Table 1, the derived time constants show a four-flash dependence, with the rate slowing by a factor of three to five on the third and seventh flash. This period-4 dependence is characteristic of reactions in the OEC (for examples, see refs. 9–11).

Fig. 3.

Plot of extracted time constants versus flash number derived from biexponential fits to the IR transients recorded at 1,483 cm⁻¹. Two time constants were derived, and these time constants are plotted on different scales on the two y axes. Data acquired on the first four flashes are shown in Fig. 1B.

Table 1.

Rate constants derived from multiexponential fits to 1,483-cm⁻¹ infrared transients

Flash no.	Phase 1		Phase 2
	Percentage	Rate, s−1	Percentage	Rate, s−1
1	32	10,700	68	906
2	22	17,300	78	1,300
3	45	3,540	55	496
4	21	19,500	79	968
5	20	37,700	80	1,340
6	33	14,200	67	1,040
7	29	5,200	71	741
8	21	29,200	79	1,440

Discussion

In this paper, we present a time-resolved IR study of the photosynthetic water-oxidizing complex. The observed immediate decrease in 1,483-cm⁻¹ absorption in the OEC-containing samples (Fig. 1B) is attributed to an amino acid side chain in the vicinity of Q_A that is perturbed in frequency when Q_A is reduced (Fig. 1A, reaction 2). We propose that this effect cancels the expected positive absorption from Q_A⁻ and Q_B⁻ (31–33) on this time scale and at this frequency. From the frequency, one possible assignment is to an imidazole ring stretching vibration. Imidazole ring stretching vibrations have been reported in the range of 1,490–1,460 cm⁻¹ in ²H₂O and ¹H₂O (34–36). Histidine side chains are known to be ligands to an acceptor side, nonheme iron in PSII (15, 16). This ferrous iron is 9 Å from Q_A and is magnetically coupled to Q_A⁻, and its histidine ligands are close to Q_A and Q_B (13–17).

The overlapping 1,483-cm⁻¹ negative signal from the OEC, which shows period-4 dependence, is also assigned to an amino acid side chain. A reasonable assignment is to a histidine (see discussion above) that is bound to manganese or near the OEC (13–17, 37). To explain our results, a histidine frequency change must occur on each S state transition (Fig. 1B). The transients show an initial bleach, which is followed by an increase over the baseline and by an eventual decay (Fig. 1B). Therefore, these measurements suggest that an intermediate state in which an OEC histidine is perturbed is formed on each S state transition.

Our extracted rate constants for the OEC reactions (Table 1) can be compared with previous measurements. Through UV spectroscopy, half-times of 110, 350, 1,300, and 30 μs were derived for the S₁-to-S₂, S₂-to-S₃, S₃-to-S₀, and S₀-to-S₁ transitions, respectively (10). Through x-ray measurements, similar half-times were reported and assigned to Mn redox reactions (11), except that a lag time of 200 μs was detected before Mn reduction during the S₄-to-S₀ transition. The ²H₂O isotope effects on the S state transitions have been reported in the range of 1.3- to 1.4-fold (38).

In contrast, our measured rates exhibit both microsecond and millisecond phases on each S state transition, all of which show period-4 oscillation. This result suggests that IR spectroscopy can detect a kinetic heterogeneity that is not detectable by UV spectroscopy or x-ray fluorescence. For the S₁-to-S₂ and S₀-to-S₁ transitions, the microsecond time constants derived from our data (≈90 and 50 μs, respectively) are in agreement with the previously reported rates of Mn oxidation (11). Therefore, on these two transitions, we attribute the perturbative mechanism to the Mn oxidation reaction. However, for the S₂-to-S₃ and S₃-to-S₀ transitions, the microsecond time constants derived from our data (≈60 and 300 μs, respectively) are significantly faster than the reported rates of Mn redox reactions (11). Although we have not accounted for S state misses, these results suggest that the perturbative mechanism on these two transitions is not a Mn redox change but a conformational change or a proton transfer reaction. This conformational change or proton transfer reaction may be linked to Y_Z oxidation either through electrostatics (11) or allosterics. The idea that proton transfer reactions may precede electron transfer has important implications for the overall water oxidation reaction. Such behavior has recently been observed for the proton-coupled electron transfer reactions in cytochrome c oxidase (39) and has been suggested to occur during the S₃-to-S₀ transition, based on the lag phase observed by x-ray spectroscopy (11).

Our measurements are unique when compared with previous studies of the OEC with vibrational spectroscopy. For example, our measurements are on a much faster time scale compared with rapid-scan Fourier-transform IR (FT-IR) studies of the OEC (for examples, see refs, 40–42). Note that it has recently been suggested that a histidine perturbation may occur with S state cycling on the 10-s time scale (43). Although cryogenic, rapid-scan FT-IR spectroscopy has been used to study the S₁-to-S₂ transition, this technique cannot be applied to the other S state transitions, which do not occur at cryogenic temperature (44). Raman spectroscopy has been used to study the S₁ and S₂ states, but these experiments were designed to probe structure on a relatively long time scale under cryogenic conditions (45). Thus, to our knowledge, our measurements are the first to detect these short-lived intermediates with vibrational spectroscopy.

Materials and Methods

Spinach PSII preparations (46) with steady-state rates of oxygen evolution [>600 μmol of O₂ per mg of chlorophyll h (47)] were exchanged into a ²H₂O buffer (²H₂O, 99%; Cambridge Isotope Laboratories, Cambridge, MA) containing 0.4 M sucrose, 50 mM Mes-NaO²H (p²H 6.0), and 15 mM NaCl. In some control experiments, Mn was removed by treatment with 1.8 M alkaline Tris (24) under illumination for 30 min. Mn-depleted samples were exchanged into a buffer containing 50 mM Mes-NaO²H (p²H 6.0). ²H₂O buffers were necessary to minimize the background absorbance and to eliminate a rapid thermal response of solvent absorbance to a saturating photolysis pulse.

For the transient IR measurements (48), the PSII sample was pelleted by centrifugation and was transferred to a calcium fluoride window. The IR cell was assembled by using another calcium fluoride window and a 25-μm spacer. The 25-μm path length spacer was greased, and the sample was sealed with vacuum grease and parafilm to prevent dehydration during the experiment (42, 49). A continuous-wave diode laser (Ekips Technology, Norman, OK) emitting at 1,483 cm⁻¹ was used as the probe, a Q-switched Nd:YAG (neodymium:yttrium aluminum garnet) laser (Spectra-Physics) was used to produce a 532-nm photolysis pulse, and the transient absorbance of the probe was detected with a liquid nitrogen-cooled mercury–cadmium–telluride detector. The instrument response time was 1 μs. The IR sample was maintained at 10°C during the measurement by using a refrigerated water bath. The 532-nm photolysis pulses had a 10-ns pulse width and a 0.2-cm spot radius (38 mJ/cm²). Samples were aligned with one or more 532-nm flashes before the beginning of data acquisition and then translated to a new spot.

For the S state experiments, the sample was preflashed and dark-adapted for 15 min to achieve synchronization in the S₁ state (50). After dark adaptation, 12 (532 nm) photolysis flashes were given at an ≈1-Hz repetition rate, and an IR transient was recorded after each flash. For other PSII control experiments, kinetics transients were averaged. The transients were fit with a multiexponential function to extract amplitudes and rate constants by using igor pro software (WaveMetrics, Lake Oswego, OR). Derived time constants depended on the choice of initial and final fit points. For consistency, transients from the same experiment were fit with identical beginning and ending points.

Abbreviations

OEC

oxygen-evolving complex

PSII

photosystem II

DCBQ

2,6-dichlorobenzoquinone

DCMU

3-(3,4-dichlorphenyl)-1,1-dimethylurea.