An Intrinsically Disordered Photosystem II Subunit, PsbO, Provides a Structural Template and a Sensor of the Hydrogen-bonding Network in Photosynthetic Water Oxidation*

Background: PsbO is an intrinsically disordered subunit of photosystem II. Results: Temperature-sensitive PsbO dynamics are identified by reaction-induced Fourier transform infrared spectroscopy and the removal and reconstitution of PsbO. Conclusion: PsbO serves as an organizational template and undergoes flash-induced hydrogen-bonding changes, coupled with the catalytic cycle of water oxidation. Significance: PsbO samples a rough conformational landscape when bound to its target, the photosystem II reaction center. Photosystem II (PSII) is a membrane-bound enzyme that utilizes solar energy to catalyze the photooxidation of water. Molecular oxygen is evolved after four sequential light-driven oxidation reactions at the Mn4CaO5 oxygen-evolving complex, producing five sequentially oxidized states, Sn. PSII is composed of 17 membrane-spanning subunits and three extrinsic subunits, PsbP, PsbQ, and PsbO. PsbO is intrinsically disordered and plays a role in facilitation of the water oxidizing cycle. Native PsbO can be removed and substituted with recombinant PsbO, thereby restoring steady-state activity. In this report, we used reaction-induced Fourier transform infrared spectroscopy to obtain information concerning the role of PsbP, PsbQ, and PsbO during the S state cycle. Light-minus-dark difference spectra were acquired, monitoring structural changes associated with each accessible flash-induced S state transition in a highly purified plant PSII preparation (Triton X-100, octylthioglucoside). A comparison of S2 minus S1 spectra revealed that removal of PsbP and PsbQ had no significant effect on the data, whereas amide frequency and intensity changes were associated with PsbO removal. These data suggest that PsbO acts as an organizational template for the PSII reaction center. To identify any coupled conformational changes arising directly from PsbO, global 13C-PsbO isotope editing was employed. The reaction-induced Fourier transform infrared spectra of accessible S states provide evidence that PsbO spectral contributions are temperature (263 and 277 K) and S state dependent. These experiments show that PsbO undergoes catalytically relevant structural dynamics, which are coupled over long distance to hydrogen-bonding changes at the Mn4CaO5 cluster.

Although much attention is focused on lowest energy protein structures, as revealed by high resolution x-ray crystallography, enzymes sample an ensemble of conformational substates, called the conformational landscape. The primary sequence encodes not only information about static structure but also the dynamic range. These factors determine the function and the adaptability of the protein (see, for example, Refs. 1-3 for reviews, and references therein). Natively unfolded or intrinsically disordered proteins (IDPs) 2 are a limiting case in which a wide array of function-related structural organizations is readily detectable. This malleability has been postulated to play an important role in an IDP binding to its target protein (reviewed in Refs. 4 and 5). Binding is often observed to result in a decrease in disorder, either through a conformational selection or induced fit mechanism, although exceptions have been noted (see Ref. 5

and references therein).
Photosystem II (PSII) is a complex membrane protein consisting both of integral, membrane-spanning subunits and extrinsic subunits. A monomeric unit of PSII consists of at least 20 distinct protein subunits, which are composed of 17 integral subunits and 3 extrinsic polypeptides, PsbP, PsbQ, and PsbO ( Fig. 1A) (6,7). The hydrophobic intrinsic subunits, which bind most of the redox-active cofactors, are D1, D2, CP43, and CP47. The light-induced electron transfer pathway in the reaction center involves the dimeric chlorophyll (Chl) donor, P 680 , accessory chlorophyll molecules, quinone acceptors, and a tyrosine residue, YZ, which is Tyr-161 of the D1 polypeptide (reviewed in Refs. 8 and 9).
The oxygen-evolving complex (OEC) is a Mn 4 CaO 5 cluster (Fig. 1B, inset). Four sequential photooxidations (Fig. 1B) of the OEC are required to produce molecular oxygen (10). Oxygen release from the OEC fluctuates with period four. The OEC cycles through five sequentially oxidized states, called the S n * This work was supported by National Science Foundation Grant MCB states. A single flash given to a dark-adapted sample (S 1 state) generates the S 2 state (Fig. 1B), which corresponds to the oxidation of Mn(III) to Mn(IV) (11). Subsequent flashes advance the OEC to higher oxidation states and higher S states, and oxygen is evolved on the transition from S 3 to S 0 . S 4 is an unstable state, which is produced by flash excitation of S 3 and converts to S 0 in the dark. Calcium and chloride play important, but not completely understood, roles in the S state cycle. Flash excitation can be coupled with spectroscopic techniques to identify structural changes associated with each S state transition (for examples, see Refs. [12][13][14][15]. In addition to intrinsic subunits, PSII also contains extrinsic subunits (16,17). For example, in plants, the three extrinsic lumenal subunits (PsbO, PsbP, and PsbQ) are necessary to maintain high steady-state rates of oxygen evolution activity (17). One proposed role of the extrinsic polypeptides, PsbP and PsbQ, is in the retention of the OEC calcium and chloride ions (18). PsbO also plays a role in retention of calcium and chloride and accelerates the steady-state rate of oxygen evolution (reviewed in Ref. 16), although a low level of activity is observed in its absence (19). The PsbO subunit is natively unfolded in solution (20), and cyanobacterial PsbO is known to form a ␤ barrel (Fig. 1A) when bound to the PSII reaction center (7).
Previous work showed that PsbO stabilizes the metal cluster and acts as a catalyst for the water splitting reactions (reviewed in Ref. 16 and 17). Functional roles for PsbO carboxylic acid and amide side chains were proposed, based on studies using reaction-induced FT-IR spectroscopy to study one transition, the S 1 to S 2 transition (65,66). These experiments suggested that amino acid side chains on the PsbO subunit are involved in proton transfer reactions on the S 1 to S 2 transition at 200 K (65,66). Interestingly, the description of PsbO structural changes were altered with temperature. These data suggest that complex PsbO dynamics accompany OEC oxidation.
To describe the role of PsbO in the PSII conformational landscape during other parts of the S state cycle, here we examine reaction-induced Fourier transform infrared (FT-IR) difference spectra for each accessible S state transition (Fig. 1B). Reactioninduced FT-IR spectroscopy has been used to study catalysisinduced conformational changes in other proteins (reviewed by Ref. 67). In application to S state cycling events, this technique was described originally in Refs. 13 and 14. The PSII sample is synchronized in the S 1 state, and laser flashes are employed to advance the Mn 4 CaO 5 cluster sequentially to the S 2 , S 3 , and S 0 states. Difference FT-IR spectra are constructed for each flashinduced transition. For example, in a S 2 minus S 1 spectrum, unique vibrational bands of the S 2 state are positive and unique vibrational bands of the S 1 state are negative. Frequencies and intensities in these spectra have been reported to oscillate with period four. On the time scale employed, this technique reflects long-lived structural dynamics on the donor side of PSII (68,69).
This reaction-induced FT-IR technique has been used recently (70,71) to study the S state cycle in a plant PSII preparation, isolated with octylthioglucoside (72) and Triton X-100 (73). This preparation provides an advantage for spectroscopic studies in that it is highly resolved and stable. The PsbP, PsbQ, and PsbO extrinsic subunits have been reported to be retained by this preparation as isolated (72), but PsbP and PsbQ can be removed selectively by NaCl washes (74), and then PsbO can be removed by urea (75) or CaCl 2 (76) washing. . PsbO is oriented to illustrate its relationship to the hydrogen-bonded water network (blue spheres) that extends from the Mn 4 CaO 5 (dark gray, yellow, and red spheres) oxygen-evolving complex to the lumen. Amino acids of the CP43 (cyan sticks), D1 (green sticks), D2 (yellow sticks), PsbU (orange sticks), and PsbO (pink) subunits are color coded. The distance from the OEC calcium to the nearest PsbO residue is at least 17 Å. B, shows a model for flash-dependent photosynthetic water oxidation. A Mn 4 CaO 5 structure, including metal ligands, is shown in the inset (PDB 3ARC). For the FT-IR experiments, presented here, a single 532-nm flash followed by 20-min dark adaptation synchronizes PSII in the S 1 state. Subsequent actinic flashes promote centers through the S state cycle. Molecular oxygen is evolved on the third flash (S 3 to S 0 transition).
In this study, these PSII purification methods and reactioninduced FT-IR spectroscopy were used to identify extrinsic subunit influence on the PSII conformational landscape. These results provide evidence that the IDP, PsbO, provides a template for organizing its target, the PSII reaction center. To identify structural changes arising directly from PsbO, spinach PsbO was recombinantly expressed and 13 C-labeled, as described previously (66,77). The 12 C (natural abundance) and globally 13 C-labeled recombinant proteins bind and restore steady-state oxygen activity (66,77). This isotope-editing approach detects structural changes only in PsbO, with no spectral contribution from the rest of the PSII reaction center. These approaches provide evidence that PsbO undergoes temperature-dependent structural dynamics, which are coupled with the S state cycle.
Estimation of PsbO Content-Neville SDS-PAGE analysis of PSII samples ( Fig. 2A) was performed as described previously (81). For Triton X-100-isolated PSII membranes (BBY), 5 g of chlorophyll was loaded per lane and for n-octyl-␤-D-thioglucoside-isolated PSII, 1 g of chlorophyll was loaded per lane. Gels were stained with 0.05% Coomassie Brilliant Blue R (Coomassie stain) to detect total protein. To determine the content of PsbO, Western analysis was performed using an anti-PsbO antibody raised in rabbits (82). Unstained gels were blotted onto a polyvinylidene fluoride (PVDF) membrane. The PVDF membrane was subjected to sequential incubations with the primary anti-PsbO antibody (82) and then with a secondary anti-rabbit protein A-alkaline phosphatase conjugate (Calbiochem, San Diego, CA). The Western blot (Fig. 2B) was developed with a 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/ NBT) liquid substrate (Thermo Scientific, Waltham, MA). Densitometry was used to quantitate the amount of PsbO in samples used for FT-IR spectroscopy. A Western blot containing a standard curve of NaCl-PSII samples (data not shown) was employed (as performed previously in Ref. 79). Comparison of the integrated areas showed that urea or CaCl 2 treatments removed 82 and 92% of native PsbO, respectively. Reconstitution with 12 C-or 13 C-PsbO to urea-PSII gave quantitative (120%) rebinding. This analysis is consistent with previous results for a different PSII preparation (24,79).
Isolation and Purification of PsbO-Recombinant PsbO was expressed in and purified from Escherichia coli BL21(DE3)/ pLysS cells containing the PsbO expression plasmid (24,77,80). The recombinant protein is different in two respects from native, spinach PsbO. The recombinant protein has an N-terminal methionine and also has a Val-235 to Ala mutation, which has been reported to arise spontaneously from propagation in E. coli (24). These changes have no influence on PsbO binding at room temperature or on reconstituted steady-state activity (24,25,65,77). Global 13 C-enrichment of recombinant PsbO was accomplished with [ 13 C]glucose and minimal media (66,77). This protocol incorporates the 13 C-isotope at all PsbO carbons with at least 90% labeling, as assessed from gas chromatography/mass spectrometry analysis on acid-hydrolyzed PsbO samples (77).
Sequencing-Nucleic acid sequencing (Eurofins MWG Operon) of the recombinant PsbO-containing plasmid confirmed the presence of the N-terminal methionine and the V235A mutation. N-terminal protein sequencing (University of Minnesota, BioMedical Genomics Center) of recombinant PsbO confirmed the presence of an N-terminal methionine residue. Electrospray ionization (ESI)-mass spectrometry (Georgia Institute of Technology, BioAnalytical Mass Spectrometry Facilities) determined a molecular mass of 26,634 Da, consistent with N-terminal methionine residue and V235A mutation in recombinant PsbO, and the expected molecular mass of 26,530 Da for native, spinach PsbO.
PsbO Reconstitution-Native, 12 C-PsbO, or 13 C-PsbO was reconstituted to urea-PSII (2 M) in 400 mM sucrose, 50 mM MES-NaOH (pH 6.0), 60 mM NaCl, and 20 mM CaCl 2 supplemented with 0.33 mg ml Ϫ1 of bovine serum albumin (Fraction V). The PsbO concentration during reconstitution was 10 M, corresponding to 5 mol of PsbO/mol of PSII reaction center. The PsbO concentration was estimated using the extinction coefficient (16 mM Ϫ1 cm Ϫ1 at 276 nm; see Ref. 19). Rebinding of PsbO was carried out at room temperature, in the dark, for 1 h. Oxygen evolution assays were performed to assess the efficiency of PsbO reconstitution, as described in Table 1. Nonspecifically bound PsbO was removed by centrifugation (12,000 ϫ g, 4°C, 10 min). PsbO-reconstituted ( 12 C/ 13 C-PSII or native-PSII) samples were resuspended in the appropriate buffer and analyzed by urea-SDS-PAGE (see Fig. 2).

C-PsbO Labeling and Rebinding Assessed by FT-IR Absorp-
tion Spectroscopy-FT-IR absorption spectra were collected from 12 C-PSII and 13 C-PSII samples. An FT-IR difference ( 12 C-PSII minus 13 C-PSII) spectrum was generated (Fig. 3A). As expected, the isotope shifts from reconstituted PsbO, bound to PSII (Fig. 3A), were comparable with the isotope shifts observed in PsbO in solution (Fig. 3B). The magnitude of the amide I and II isotope shifts was consistent with the expected downshifts for the 13 C-isotopologue (65,66).
Reaction-induced FT-IR Spectroscopy-Reaction-induced FT-IR spectroscopy was performed at 263 and 277 K, as previously described (70,71). Samples (PSII, NaCl-PSII, urea-PSII, CaCl 2 -PSII, native-PSII, 12 C-PSII, or 13 C-PSII) for FT-IR experiments were suspended in 400 mM sucrose, 50 mM MES-NaOH (pH 6.0), 60 mM NaCl, and 20 mM CaCl 2 by three rounds of centrifugation (50,000 ϫ g, 4°C, 15 min) and resuspension and frozen (Ϫ70°C) until use. Samples were thawed and potassium ferricyanide (7 mM) was added, and the sample was centrifuged (50,000 ϫ g, 4°C, 15 min) to produce a pellet. The sample was spread on a calcium fluoride window and concentrated briefly under N 2 gas to give an O-H stretching (3370 cm Ϫ1 ) to amide II (1550 cm Ϫ1 ) absorbance ratio greater than 3. Samples were covered with a second CaF 2 window and the edges were sealed with high vacuum grease and parafilm to prevent dehydration. Acquisition parameters for FT-IR spectroscopy were as follows: 8 cm Ϫ1 spectral resolution; 60 kHz mirror speed; four levels of zero filling; Happ-Genzel apodization; and Mertz phase correction. Samples were preflashed with a single saturating 532 nm laser flash (40 mJ cm Ϫ2 ), followed by 20 min dark adaptation to synchronize PSII centers in the S 1 state. Dark-adapted PSII samples were then given four more (actinic) flashes. Each single actinic flash was followed by 15 s of data acquisition for each accessible S state. Therefore the flash frequency is one flash per 15 s. The reaction cycle was conducted once for each sample, so that each sample received only 5 total laser flashes. Reactioninduced difference spectra were constructed by the ratio of single-channel data, which were collected before and after each actinic flash. Rapid scan data were normalized to an amide II intensity of 0.5 absorbance units to correct for small differences in the path length (6 m) and concentration between samples.
The  . Isotope shifts caused by global 13 C labeling of PsbO, as assessed from FT-IR absorbance spectra. Difference spectra were constructed to measure the isotope-induced downshift. In A, data were derived from 12 C/ 13 C-PsbO bound to PSII ( 12 C-minus 13 C-PSII; black), and in B, data were derived from PsbO in D 2 O buffer ( 12 C-minus 13 C-PsbO; red). In A, samples were prepared in 400 mM sucrose, 50 mM MES (pH 6.0), 60 mM NaCl, and 20 mM CaCl 2 , as described under "Experimental Procedures" for reaction-induced FT-IR spectroscopy. In B, PsbO (red) samples were prepared at 100 M in 50 mM sucrose, 50 mM MES (pD 6.0), 60 mM NaCl, and 20 mM CaCl 2 . The 1450 cm Ϫ1 region for isolated PsbO (red) has overlapping H-O-D contributions and was removed for clarity. The y axis tick marks represent 5 ϫ 10 Ϫ2 absorbance units.
FT-IR samples were illuminated with a 633 nm HeNe (Melles-Griot, Carlsbad, CA) laser (2 milliwatt). The concentration of ferrocyanide, produced under 90 s of continuous illumination, was assessed from the amplitude of the 2038 cm Ϫ1 band in a 633-nm light-induced difference spectrum and a standard curve of potassium ferrocyanide. The standard curve was derived by addition of a known concentration of ferrocyanide to PSII and measurement of the 2038 cm Ϫ1 band in the dark in the FT-IR sample cell. Control experiments showed that no detectable ferrocyanide was produced by illumination of the ferricyanide-containing buffer (no PSII) and that no detectable ferrocyanide was produced in ferricyanide-containing PSII without 633 nm illumination. The FT-IR sample volume was estimated as 40 l, the samples contained 0.3 mg of Chl (antenna size 55 Chl/reaction center), and the 633 nm difference spectra were corrected for any small difference in amide II absorbance. The derived electron transfer rates were 1.1 Ϯ 0.3 mol of electrons (mg of Chl-h) Ϫ1 or 55 mol of electrons (mol of PSII reaction center-h) Ϫ1 at 263 K and 1.2 Ϯ 0.5 mol of electrons (mg of Chl-h) Ϫ1 or 61 mol of electrons (mol of PSII reaction-h) Ϫ1 at 277 K. Using the same procedure, the yield of ferrocyanide produced by the first actinic 532 nm flash at 277 K was ϳ1 mol of electron per mol of PSII reaction center, confirming the validity of the procedure. This experiment supports the conclusion that there is no difference in acceptor side electron transfer when the two temperatures are compared. Note that the electron transfer rate under continuous illumination is slower than the rate predicted by the oxygen evolution assay (see above). This difference is attributable to the rapid mixing and dilute sample conditions in the oxygen assay, as compared with the concentrated FT-IR samples.

RESULTS
The steady-state oxygen evolution rates of PSII preparations used in this study are presented in Table 1. Removal of extrinsic polypeptides was performed by washing with 2 M NaCl (PsbP, PsbQ depleted, NaCl-PSII) and then either 2.6 M urea (PsbO, PsbP, and PsbQ depleted, urea-PSII) or 1 M CaCl 2 (PsbO, PsbP, and PsbQ depleted, CaCl 2 -PSII). Removal of the PsbP and PsbQ subunits was confirmed by SDS-PAGE analysis of NaCl-PSII (Fig. 2, lane 3), compared with untreated PSII (Fig. 2, lane 2). The oxygen evolution activity of NaCl-PSII was dependent on exogenous calcium and chloride, as expected (see Table 1 and Ref. 83). Treatment with 2.6 M urea removed ϳ80% of natively bound PsbO (urea-PSII, Fig. 2, lane 4, and also see Refs. 77 and 79)) and significantly decreased steady-state rates of oxygen evolution (Table 1). CaCl 2 treatment also decreased oxygen evolution and removed 90% of PsbO (see "Experimental Procedures" for details).
These results are consistent with previous studies of other spinach PSII preparations (20,64,66,72,77,79,80). The V235A mutation employed here does not affect the solution structure, binding, or activity at room temperature (24). The use of the V235A variant has the advantage that the migration of this protein on an SDS-PAGE gel is slightly different from the wildtype, allowing a ready verification of reconstitution, as shown above (Fig. 2, lanes 5 and 6).
To monitor the effects of the removal of the extrinsic polypeptides, we performed reaction-induced FT-IR spectroscopy (S 2 minus S 1 ) on NaCl-PSII (Fig. 4B), urea-PSII (Fig. 4C), native-PSII (Fig. 4E), and 12 C-PSII (Fig. 4F). These spectra are dominated by contributions from the donor side of PSII. Double difference spectra (Fig. 5, A-E) were generated by subtraction of the S 2 minus S 1 difference spectra, associated with each treatment, from the control (untreated) PSII S 2 minus S 1 spectrum (Fig. 4A). Before construction of the double difference data, the spectra were scaled to reflect accurately the alterations induced by subunit removal. First, difference spectra were corrected for amide II absorption relative to an open beam background, reflecting any small differences in concentration and path length between samples. Second, spectra were then normalized to the intensity of the ferricyanide and ferrocyanide bands (2116 and 2038 cm Ϫ1 , data not shown), which reflect any small changes in the amount of charge separation in the preparations. These methods have been used previously for PSII spectra (70,71,84) and give control double difference spectra ( Fig. 5F) with flat baselines and no vibrational bands, as expected. Because the difference spectra are presented on the basis of total protein (amide II intensity), it was then necessary to use a correction factor to reflect the depletion of subunits. Each plant PSII reaction center is assumed to contain one copy of PsbP and PsbQ (6,85). There is considerable evidence for two copies of PsbO (reviewed in Refs. 16 and 86) per plant PSII reaction center, and our SDS-PAGE and Western analyses provide evidence that the removal and reconstitution is quantitative (see Fig. 2 and "Experimental Procedures" for details). The scaling factors employed are based on this information and the molecular masses of PsbP (16.5 kDa), PsbQ (20 kDa) (17), PsbO (26.5 kDa) (87), and PSII (350 kDa) and are 90% for NaCl-PSII (PsbP/PsbQ depleted) and 74% for urea-PSII (PsbP/PsbQ and PsbO depleted).
To establish the level of noise and baseline fluctuation in these data, we generated a control double difference spectrum. This control was produced using the relevant correction factors described above and by subtracting one-half of a data set from its other half. This control should exhibit no vibrational bands and is shown in Fig. 5F. In comparison, the CaCl 2 spectrum (Fig. 5C) exhibits vibrational bands that are significant relative to the noise at Ϫ1682, Ϫ1661, Ϫ1547, and ϩ1532 cm Ϫ1 . These are bands typical of amide I (C ϭ O stretch) and amide II (CN stretch/NH bending) bands in proteins. These bands have frequencies sensitive to secondary structure and hydrogen bonding (88). The urea-PSII spectrum also exhibited bands, significant relative to the noise (Fig. 5B). We conclude that removal of PsbO using either method results in a change in conformational sampling in the PSII reaction center during the S 1 to S 2 transition. Rebinding of native or recombinant protein reversed the majority of the effect of PsbO removal (Fig. 5, D and E). However, compared with the control (Fig. 5F), the PSII minus NaCl-PSII spectrum (Fig. 5A) exhibited no significant changes in the mid-infrared region.
This result supports the use of the recombinant protein in these spectroscopic studies and shows that PsbO removal has a significant effect on PSII structural transformations during the S 1 to S 2 transition. In agreement, previous EPR studies have concluded that although urea and CaCl 2 treatments do not sig-nificantly alter hyperfine splittings in the S 2 multiline signal, the intensity of the S 2 g ϭ 4.1 EPR signal is altered (89, 90) by PsbO removal. Note that, here, a subsequent flash to each of these preparations (S 2 to S 3 transition) gave the same spectral result in each type of preparation (data not shown).
To identify direct spectral contributions from PsbO in this PSII preparation, FT-IR difference spectra (Fig. 6) were recorded for 12 C-(black) and 13 C-PSII (red). Corresponding isotope-edited spectra were generated as 12 C-minus 13 C-PSII (from data in Fig. 6) and as a function of flash number at 263 K (Fig. 7, A-D). Oxygen evolution occurs on the first reaction cycle at this temperature, but the sample cannot carry out a second enzymatic cycle, under these flash conditions, possibly due to a limitation in water diffusion or a conformational gate (84). There is evidence that the steady-state rate of electron transfer under continuous illumination is slower in the FT-IR samples, when compared with a room temperature oxygen assay ("Experimental Procedures"). However, the flash-induced yield of ferrocyanide was ϳ1 mol/mol of reaction center and  The spectrum in F is a control double difference spectrum, which was generated by subtraction of one-half of the data in A from the other half and division by the square root of two. Difference spectra, used to generate the double differences, are shown in Fig. 4 and were normalized before subtraction as described in the text. Double difference spectra were generated from 23 (A), 20 (B), 20 (C), 25 (D), 29 (E), and 10 (F) spectral averages. The y axis tick marks represent 2 ϫ 10 Ϫ4 absorbance units. OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40

JOURNAL OF BIOLOGICAL CHEMISTRY 29061
the amplitudes of the ferrocyanide bands were similar for each S state transition (data not shown), arguing that there is no acceptor side limitation under these flash conditions. Data were corrected as described above for path length, concentration, and charge separation differences. 12 C and 13 C reconstitutions were performed using the same PSII sample, so corrections for subunit composition were not necessary. Again, construction of control double difference spectra, by subtraction of one-half of a data set from the other, provides an estimate of the background and signal to noise of the measurement (Fig. 7, E and J).
The isotope-edited ( 12 C-minus 13 C-PSII) FT-IR spectrum associated with the S 1 to S 2 transition at 263 K is displayed in Fig. 7A. To appear in this double difference spectrum, the band must be altered in frequency and/or intensity by the photooxidation reaction and also be sensitive to 13 C global labeling of PsbO. Thus, to be observed, the spectral contribution must originate from the PsbO subunit. For example, in the S 2 minus S 1 isotope-edited spectrum, the 12 C-labeled S 2 state will exhibit a positive band, and its 13 C-labeled S 2 isotopologue will exhibit a negative band. 12 C/ 13 C-labeled S 1 bands will be reversed in sign.
To enhance the signal from just PsbO, on the background of the unlabeled PSII, spectra were acquired from just one reac-tion cycle on 19 -25 FT-IR samples and then averaged. The resulting isotope-edited spectrum (Flash 1, Fig. 7A) displays bands at Ϫ1656, ϩ1637, Ϫ1597, and Ϫ1549 cm Ϫ1 , which are directly attributable to PsbO.
The 1656 and 1549 cm Ϫ1 bands have frequencies typical of the amide I and II vibrational modes in proteins (88). A frequency of 1656 cm Ϫ1 in H 2 O buffer is typical of an ␣ helical or a disordered region of secondary structure (88). Cyanobacterial PsbO is primarily ␤ sheet, with some disordered and a minority of ␣ helical regions (7), so we attribute the negative 1656 cm Ϫ1 band to a disordered region of PsbO, which increases in structure and hydrogen bonding during transition from the S 1 to S 2 state. The resulting positive S 2 band is not detected in Fig. 7A. This might occur if the PsbO ␤ sheet is formed in the S 2 state, because transition dipole coupling splits the amide band of ␤ sheet into high and low frequency components, each with lower intensity (88). This would make the positive bands more difficult to detect.
The magnitude of the expected isotope shift for amide bands can be determined from FT-IR absorption spectra (Fig. 3A). The FT-IR absorption spectrum of 12 C-PSII and 13 C-PSII were subtracted (dark, no photolysis flash, 12 C minus 13 C). As shown, global 13 C labeling is predicted to downshift both amide I (ϳ1650 cm Ϫ1 ) and amide II (1554 cm Ϫ1 ) bands by approxi- mately ϳ50 cm Ϫ1 . Note that the isotope-shifted bands are less intense in Fig. 3A, compared with the 12 C-labeled bands (77). Thus, we expect the signature of the amide isotope shift to be a band at ϳ1650 and 1554 cm Ϫ1 .
This comparison accounts for the Ϫ1656 and Ϫ1549 cm Ϫ1 bands in Fig. 7A, assigning them to the PsbO amide bond. To appear in the spectrum, the photooxidation reaction must change the frequency or amplitude of amide bands. Such a change is consistent with an alteration of PsbO amide hydrogen bonding during the S 1 to S 2 transition. This may correspond to an increase in order.
The frequency of ϩ1637 cm Ϫ1 (Fig. 7A) is consistent with an assignment to a hydrogen-bonding change to an amide side chain of Gln or Asn (91). The band from the 13 C-isotopologue may be detectable as Ϫ1597 cm Ϫ1 (Fig. 7A). Previously, a PsbO band at ϩ1632 cm Ϫ1 was observed at 277 K in a PSII preparation isolated with lauryl maltoside. This band was also assigned to Gln/Asn, based on specific isotopic labeling (66). The amide I region of that spectrum was not reported, due to the lower stability of the baseline in that sample. Thus, a Gln/Asn assignment accounts for the ϩ1637 cm Ϫ1 band described in this current study, but the band is observed at slightly lower tempera-ture in this octylthioglucoside preparation, compared with a lauryl maltoside-isolated PSII preparation.
The observation of these PsbO amide bands supports the conclusion that PsbO has a responsive structure, which is altered during the S 1 to S 2 transition. Interestingly, the S 2 to S 3 (Flash 2, Fig. 7B) and S 3 to S 0 (Flash 3, Fig. 7C) transitions also generated spectra that exhibited amide bands, but with differences in frequency and sign. The S 0 to S 1 (Flash 4, Fig. 7D) spectrum was similar to the negative control, the control double difference spectrum in Fig. 7E. Thus, at 263 K, the effects of the S state transitions on PsbO conformation are distinguishable.
The effect of temperature on PsbO dynamics was tested by performing the isotope-editing experiment at 277 K. At this temperature, the yield of potassium ferrocyanide produced on each flash was similar and also similar to the flash-induced ferrocyanide yields at 263 K. The rate of ferrocyanide production under continuous illumination was also indistinguishable at 277 and 263 K.
Isotope-edited spectra for each accessible S state transition were constructed and are shown in Fig. 7, F-I. Interestingly, the 277 K spectra are distinct when compared with the 263 K data The spectra in E and J are representative double difference spectra, which were generated by subtraction of one-half of the data in A and F from the other half and division by the square root of two. Difference spectra, used to generate the isotope-edited spectra, are shown in Fig. 6 and were normalized before subtraction as described in the text. Isotope-edited spectra were generated from 38 (A-D), 19 (E), 47 (F-I), and 22 (J) 22 spectral averages. The y axis tick marks represent 2 ϫ 10 Ϫ4 absorbance units. (Fig. 7, A-D). On the first flash (Fig. 7F), the spectra exhibited bands, some of which may be significant compared with the noise (Fig. 7J), throughout the mid-infrared region. On the second flash (Fig. 7G), the spectra were altered and exhibited significant bands at Ϫ1652, Ϫ1548, ϩ1441, Ϫ1400, and Ϫ1356 cm Ϫ1 . There was less intensity in the isotope-edited spectra on subsequent flashes (Fig. 7, H and I) at 277 K. Peptide backbone contributions can account for most of the observed frequencies in Fig. 7G (see Fig. 3A), with possible overlapping contributions from deprotonated carboxylate side chains (65,66). Like amide frequencies, carboxylate frequencies are sensitive to changes in hydrogen bonding (92).
These data provide evidence that the most significant change in the PsbO structure occurs during the S 2 to S 3 transition at 277 K. Again, we attribute this change to an alteration in PsbO hydrogen bonding. Thus, reaction-induced FT-IR spectroscopy establishes that IDP PsbO exhibits temperature-and flash-dependent dynamics, which are linked with oxidation of the Mn 4 CaO 5 cluster.

DISCUSSION
The understanding of how conformational selection contributes to enzymatic catalysis is still limited (3). Although static structures provide important information concerning the lowest energy conformer, proteins sample numerous conformational substates along the reaction coordinate (1-3). Descriptions of energy landscapes, which may facilitate enzymatic rate acceleration, are beginning to be developed using theoretical and spectroscopic (NMR (93,94) and FT-IR (95)) approaches.
The extrinsic PSII subunits are important in biological water oxidation (16,17,75,79,83,96). The PsbP and PsbQ extrinsic polypeptides have been shown to modulate the calcium and chloride requirements for efficient O 2 evolution (17,74,79,83,97). PsbO also plays a critical role. The removal of PsbO had differential effects on the lifetimes and structural changes associated with the S state transitions. PsbO-depleted PSII samples exhibited longer-lived S 2 and S 3 states (in the dark) and delayed O 2 release from the OEC (98). Alterations in S state lifetimes were observed in PsbO-deficient cyanobacteria (99) and in Arabidopsis (52), in which only one of the two isoforms of PsbO (PsbO-2) was expressed. Complete suppression of PsbP expression in Arabidopsis resulted in a highly unstable OEC in the mutants, which exhibited an increased S 2 state lifetime (100). Depletion of PsbP and PsbQ slowed the rate of electron transfer 6 -12-fold at the acceptor side of PSII; this observation was unaffected by removal of the PsbO subunit (101).
PsbO is an IDP (20). IDPs rely on their conformational promiscuity in solution to bind to a target or an array of target proteins (reviewed in Ref. 4 and 5). Previous analysis demonstrated that PsbO undergoes a secondary structural change, reducing the amount of disordered structure and increasing ␤ sheet character (77), without altering the overall size or shape (102). Mutation of the lone tryptophan residue (W241F) in spinach PsbO resulted in defective binding to PSII and reduced steadystate oxygen evolution activity, but this effect could be eliminated by deletion of the six amino-terminal residues (40). Mutation at Arg-151, Arg-161, (38), Asp-157, (41), or Leu-245 (29) also resulted in reduced binding. It was suggested that these mutations increase the structural stability of isolated PsbO, which, in turn, interferes with the inherent flexibility required for PsbO binding. These previous mutational and spectroscopic studies concluded that PsbO flexibility is essential for effective PSII binding and restoration of high O 2 evolution activity.
We show here that removal of PsbO either by urea or CaCl 2 treatment alters the conformational landscape of the PSII reaction center. This was assessed by monitoring the structural changes associated with the S 1 to S 2 transition. Both urea-and CaCl 2 -treated preparations are known to form an S 2 state. Previously, PsbO removal (with both urea and CaCl 2 ) has been shown to alter the equilibrium between two EPR signals arising from the S 2 state (for examples, see Ref. 86, 89, and 103). Our work provides evidence that this change in magnetic coupling may be caused by a conformational selection, which is mediated by PsbO binding to PSII. These experiments are consistent with the idea that PSII undergoes extensive dynamics on many time scales as previously proposed from analysis of transient infrared kinetics and FT-IR spectroscopy (68,69) and that there is a distribution of active conformations. Interestingly, in our studies, removal of PsbP and PsbQ had no significant effect on the spectra associated with the S 1 to S 2 transition.
Reaction-induced FT-IR spectroscopy, isotopic labeling, and temperature were then used to probe the conformational dynamics of PsbO in all accessible S states. We hypothesized that, when bound to PSII, PsbO can sample a broad distribution of substates in a rough conformational landscape. This hypothesis was based on previous isotope-edited FT-IR experiments, conducted for the S 1 to S 2 transition, which showed surprising temperature dependence. Also, significant alterations were observed when data obtained under steady-state illumination were compared with spectra acquired on a long time scale after a saturating flash (65,66).
To obtain more information, here, we monitored PsbO dynamics during the entire S state cycle (Fig. 1B) with a PSII preparation that exhibits enhanced signal to noise. Our results show that contributions from PsbO at 263 and 277 K are distinguishable and are most significant on the S 2 to S 3 transition at 277 K. This work provides evidence that PsbO participates in temperature-sensitive conformational dynamics, which are linked with the S state cycle. This conclusion provides a rationalization for the observation of broad PsbO isotope-edited spectra under illumination at 200 K (65). Under steady-state illumination, it is likely that many conformational substates of PsbO and PSII can be populated. The data suggest that PsbO retains its intrinsically disordered character, even when bound to its target, the PSII reaction center.
Previously, direct carboxylate deprotonation was inferred from 200 K data, but was not observed at 277 K in a lauryl maltoside-isolated PSII preparation (65,66). Protonated and deprotonated carboxylates can be easily distinguished by FT-IR spectroscopy (67,91). Protonation of the carboxylate side chain gives rise to C ϭ O stretching bands in the 1720 -1700 cm Ϫ1 region, whereas deprotonated carboxylates have delocalized asymmetric and symmetric CO stretching vibrations at ϳ1560 and 1390 cm Ϫ1 (65). The reaction-induced FT-IR spectra presented here do not exhibit significant intensity in the 1770 -conclusions: that any PsbO deprotonation event occurs in a small population of centers or that the deprotonation event is facile and therefore not observable at higher temperature under these conditions.
A previous FT-IR study concluded that no significant changes occurred in the mid-infrared region, associated with the S 1 to S 2 state transition, when PsbO was removed (104). However, spectral changes were attributed to removal of the PsbP subunit, but not the PsbQ or PsbO subunits. It was proposed that PsbP affects the peptide conformation around the OEC without alteration of the ligand structure. In our experiments, by contrast, PsbO depletion with urea or CaCl 2 resulted in significant changes in the reaction-induced FT-IR spectra, whereas PsbP and PsbQ removal did not have a significant effect, relative to a negative control. In our studies, FT-IR experiments were conducted on a highly purified PSII preparation (72), which is devoid of light harvesting complexes (105). This PSII preparation exhibits high steady-state activity and stability. In addition, samples were subjected to the minimum number of flashes, including only one reaction cycle (4 flashes plus a preflash), and the amount of time between preparing the sample and running the experiment was strictly controlled to improve stability and reproducibility. Data were obtained from 10 to 25 different FT-IR samples and averaged to achieve the final signal to noise. Also, spectra were corrected in intensity for any small changes in protein concentration, path length, charge separation, and subunit content. By contrast, in Ref. 104, there may have been lower intrinsic activity and stability in the PSII preparation employed, and there was a resulting decrease in the signal to noise ratio. Also, in Ref. 104, there was a lower number of averages (3 to 5), a small number of samples employed (1 to 2), samples were frozen/thawed to conduct multiple reaction cycles, and there was no reported correction for protein concentration, path length, charge separation, or subunit composition.
PsbO is ϳ17 Å from the Mn 4 CaO 5 cluster (Fig. 1A). How are these conformational changes in PsbO linked with oxidation of the metal cluster? PSII contains a catalytically important hydrogen-bonded water network surrounding the OEC (70). This network (Fig. 1A) extends from the OEC to the lumen, including a number of residues in PsbO, and may play a functional role in proton transport (106 -109). We propose that the hydrogenbonding changes in PsbO are propagated through this network, when manganese is oxidized. Interestingly, the S 2 to S 3 transition has been proposed to be associated with a change in manganese-manganese and manganese-calcium bond lengths (12,15). There are likely to be differences between cyanobacterial and plant PSII (see Ref. 16 for review). An example is the difference in binding stoichiometries of PsbO per PSII reaction center. However, the 1.9-Å cyanobacterial structure (7) serves as a reasonable starting point for our spectral interpretation. Previously, modeling of the PsbO structure has suggested that clusters of carboxylic acid side chains may be important in function (110). The FT-IR data, presented here, suggest that the amide bonds of PsbO are sensors for the functional, extended hydrogen-bonded water network in the OEC. Spectral differences observed at 263 K and 277 K may be due to inhibited diffusion of water or to a conformational gate at 263 K, under the conditions employed for flash excitation (84).
Studies of alcohol dehydrogenases have shown that longrange conformation selection is important in catalysis (111). The work presented here suggests that long-range conformational interactions may also play an important role in the PSII reaction center. Our studies provide additional insight into the role of a single subunit, PsbO, in navigating the conformational landscape, which is associated with photosynthetic water oxidation.