A Singular Bacteriophytochrome Acquired by Lateral Gene Transfer*

Bacteriophytochromes are phytochrome-like proteins that mediate photosensory responses in various bacteria according to their light environment. The genome of the photosynthetic and plant-symbiotic Bradyrhizobium sp. strain ORS278 revealed the presence of a genomic island acquired by lateral transfer harboring a bacteriophytochrome gene, BrBphP3.ORS278, and genes involved in the synthesis of phycocyanobilin and gas vesicles. The corresponding protein BrBphP3.ORS278 is phylogenetically distant from the other (bacterio)phytochromes described thus far and displays a series of unusual properties. It binds phycocyanobilin as a chromophore, a unique feature for a bacteriophytochrome. Moreover, its C-terminal region is short and displays no homology with any known functional domain. Its dark-adapted state absorbs maximally around 610 nm, an unusually short wavelength for (bacterio)phytochromes. This form is designated as Po for orange-absorbing form. Upon illumination, a photo-reversible switch occurs between the Po form and a red (670 nm)-absorbing form (Pr), which rapidly backreacts in the dark. Because of this instability, illumination results in a mixture of the Po and Pr states in proportions that depend on the intensity. These uncommon features suggest that BrBphP3.ORS278 could be fitted to measure light intensity rather than color.

To adapt their growth and development to light environment, plants use a large panel of photoreceptors. Among them, phytochromes utilize bilin as chromophore to control several aspects of photomorphogenesis like germination, flowering, shade avoidance, chloroplast development, etc. (1)(2). Phytochromes respond to red and far-red light via a reversible shift from a red-absorbing form (Pr) to a far-red absorbing form (Pfr). In the last decade, thanks to the accumulating genomic data, it has been recognized that phytochromes are not restricted to plants but are widely distributed among prokaryotes and eukaryotes. This finding has provided new insights into the diversity, the evolution, and the biochemical and structural description of this light sensor family (3)(4)(5)(6)(7)(8). A recent phylogenetic analysis shows that the phytochrome superfamily is organized in five different clades: the Phys 4 (plants phytochromes), Cphs (cyanobacterial phytochromes), BphPs (bacteria phytochromes), Fphs (fungi phytochromes), and a cluster of Phy-like sequences (6).
The architecture of most phytochromes is composed of a N-terminal photosensory core domain (PCD) and of a C-terminal module involved in signal transduction. The PCD, comprised of the PAS, GAF, and PHY subdomains, attaches autocatalytically a linear tetrapyrrole (or bilin). All bilins are derived from a heme molecule, converted to biliverdin IX (BV) by a heme oxygenase. Whereas BphPs use this simplest linear tetrapyrrole as a chromophore (9), the chromophores of Phys and Cphs are 3E-phytochromobilin (P⌽B) and 3Z-phycocyanobilin (PCB), respectively, modified from BV by bilin reductases (10). These bilins are covalently bound to a Cys residue located in the N-terminal region for BV and in the GAF domain for P⌽B and PCB (11,12). Although little is known about the Fph type, their similarity to BphPs and their biochemical properties are strong indications that BV is also their natural chromophore (13). The Phy-like proteins display an unusual PCD architecture characterized by the absence of PAS and PHY domains and may therefore represent a distinctive group of bilin proteins (6).
The C-terminal region of most BphPs possesses a two-component histidine kinase motif that transfers phosphate to a response regulator (RR) whose gene is often found within the BphP operon (14). However, other output modules, like PAC, RR, or GGDEF and EAL domains have been described (3). The spectral properties of BphPs are also highly disparate. Their dark-adapted ground state can be either the Pfr (bathyBphP) or the Pr form (15,16), and a short wavelength-absorbing form (Pnr) is photo-induced for RpBphP3 from Rhodopseudomonas (Rps.) palustris (17). Altogether, these results highlight the modularity and the adaptability of the BphP family.
Photosynthetic bradyrhizobia belong to the ␣-sub-branch of the Proteobacteria family. They have the capacity to form nitro-gen-fixing nodules on both roots and stems of some aquatic legumes belonging to the genus Aeschynomene (18). Besides their life in association with plant, these bacteria develop freely in aquatic environments or in soils. Recently, a bacteriophytochrome (BrBphP) whose gene is located in the photosynthesis gene cluster was shown to play a key role in the control of photosystem synthesis by antagonizing the action of the repressor PpsR (16). The recent sequencing of the genomes of two photosynthetic Bradyrhizobium strains (ORS278 and BTAi1) revealed the presence of two other BphPs in each strain. 5 One of these is common to both strains, whereas the second is specific to each strain.
In this study, we have carried out a detailed characterization of the specific BphP present in strain ORS278, denoted BrBphP3.ORS278. We give evidence that its gene was acquired by lateral gene transfer together with bphO and pcyA genes, involved in its chromophore synthesis. This is the first description of a BphP containing phycocyanobilin as chromophore. Furthermore, we show that this new type of BphP possesses a series of unusual photochemical properties, which leads us to propose that it acts as a sensor of light intensity rather than light color.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Bradyrhizobium strain ORS278 was grown in a modified YM agar medium at 35°C for 7 days (19).
Site-directed Mutagenesis-Single mutations were introduced in BrBphP3.ORS278 using appropriate primers and the QuikChange TM site-directed mutagenesis kit (Stratagene) according to the manufacturer's recommendations.
Construction of BrBphP3.ORS278 and gvpA Mutants-For the construction of the BrBphP3.ORS278 minus mutant, the gene was introduced in the pJQ200mp18 suicide vector (20). A gene fragment was deleted (476-bp SalI) and replaced by the lacZ-kmR cassette of pKOK5 (21). To construct the gvpA minus mutant, the gvpA gene was replaced by the lacZ-kmR cassette of pKOK5. The sequences flanking gvpA were ligated thanks to the following pairs of primers: upgvpA.F 5Ј-CCGGATCCTGGCGT-TAGCGCGATGTCGCAAAC-3Ј and upgvpA.R 5Ј-GGATGC-AATGTCGACCTCAATAGCCACGGACTTACCTCATGCG-TTTC-3Ј, and dwgvpA.F 5Ј-GCTATTGAGGTCGACATTGC-ATCCTAACAACGATCCTGATCCTGAG-3Ј and dwgvpA.R 5Ј-CCGGGCCCCGCGCCGGAGATCATCGATCTTG-3Ј (the primers upgvpA.R and dwgvpA.F contain an overlapping sequence of 27 bp (underlined) to facilitate overlap extension PCR). The ligated product was amplified and introduced into the pJQ200-SK plasmid thanks to restriction sites designed in each primer (in bold). The 4.7-kb SalI lacZ-kmR cassette of pKOK5 was then inserted into the SalI site designed in the overlapping sequences of upgvpA.R and dwgvpA.F primers (in italic). These two constructions were introduced and delivered by conjugation into ORS278 strain as described (19).
Mathematical Treatment-Under steady-state illumination with a saturating intensity, the fractions of Pr and Po are as shown in Equation 1. In this Equation, or and ro are the rate constants, under given illumination conditions, of the Po 3 Pr and Pr 3 Po photoconversions, respectively. Saturation requires that these photochemical rate constants are much greater than those for the dark relaxation, which can then be neglected. Expressing spectra as vectors (the nth component is the amplitude at the nth wavelength; in practice all our spectra were measured with a 1-nm interval), we denote as Pr and Po the spectra of the pure forms Pr and Po. The spectrum Sm of the sample under saturating illumination from some light source labeled m, as shown in Equation 2 is thus. In this expression, the rate constants or m , ro m are products of the number of photons absorbed per unit of time and of the probability ( o or r , respectively) that absorption results in successful conversion to the other form. Denoting as Zm the spectrum radiated by source m, the absorption rates are the dot products Zm⅐Pr and Zm⅐Po and, as shown in Equation 3, one has the following. 5 A. Vermeglio and E. Giraud, unpublished data. MARCH 9, 2007 • VOLUME 282 • NUMBER 10

A Singular Bacteriophytochrome
Thus, as shown in Equation 4, where ϭ o / r is the ratio of the quantum efficiencies.
In the experiment illustrated by Fig. 5A, one measures three spectra, S1, S2, S3, corresponding to illumination by sources of known spectra, Z1, Z2, Z3. The Pr and Po spectra are unknowns to be determined, together with the ratio of quantum yields . Three Sm spectra (at least) are necessary in this case, whereas if were known, only two spectra would suffice. We have thus a system of three vector equations such as Equation 4, with m ϭ 1, 2, 3. Because the Sm are linear combinations of Pr and Po, the latter are instances of the vector family as shown in Equation 5 V͑x͒ ϭ S1 ϩ x͑S1 Ϫ S2͒ (Eq. 5) where x is a scalar. The choice of the particular vectors S1 and S2 out of the three Sm is arbitrary; in practice, it is reasonable to pick the two most different spectra. Now, considering Equation 5, there are two particular values of x, say x o and x r , for which V(x o ) ϭ Po and V(x r ) ϭ Pr. One has thus a system of three vector equations to be solved for three scalar unknowns, x o , x r , and . The generic equation is Sm Ϫ Sm ϭ 0 (a vector with norm 0), where Sm is obtained from Equation 4, replacing Po and Pr by V(x o ) and V(x r ), respectively, as shown in Equation 6.
Because the S and Z are (noisy) experimental spectra, we are not looking for an exact solution but rather for a non-linear least square procedure (e.g. Levenberg-Marquardt) minimizing the sum of Euclidian norms N(x o ,x r ,) ϭ ͉S1 Ϫ S1͉ ϩ ͉S2 Ϫ S2͉ ϩ ͉S3 Ϫ S3͉. This is easily handled by the mathematical software Mathcad 12 from Mathsoft Inc.

BrBphP3.ORS278, a BphP Acquired by Horizontal Gene
Transfer-The BrBphP3 gene from strain ORS278 encodes a 620-amino acid protein that displays a weak identity (27%) with the Cph-like protein AphA from Anabaena sp. strain PCC 7120 and with the BphP from Pseudomonas fluorescens strain Pf-5. Prediction of the protein architecture using Pfam or Smart tools permits identification of a classical PCD with PAS-GAF-PHY subdomains, supporting its classification as phytochrome. However, unlike all the phytochromes described thus far, the C-terminal region is very short and displays no homology with any known functional domain (Fig. 1A). Furthermore, a phylogenetic analysis (Fig. 1C) using the GAF domain (or the entire PCD module, supplemental Fig. S1) indicates that this protein is distant from the five clades that constitute the phytochrome superfamily (6). All these data indicate an unorthodox character for BrBphP3.ORS278 and raise the question of its origin and function. This is strengthened by the fact that the BrBphP3.ORS278 gene is found in a region of ϳ91 kb that displays all the hallmarks of a horizontal acquisition island. Indeed, this region is characterized by an atypical G ϩ C content (62.8% compared with the average value 65.4% of the genome), a tRNA insertion site, a distinct codon usage, and the presence of integrase and tra genes known to be involved in integration and conjugal transfer of DNA mobile elements, respectively. These features, together with the absence of a homologue gene in the closely related species (Bradyrhizobium sp. BTAi1, B. japonicum, or Rps. palustris), indicate that BrBphP3.ORS278 was acquired by lateral gene transfer.
Other genes of interesting function are present in this genomic island at the vicinity of BrBphP3.ORS278: (i) a gene encoding a heme oxygenase designated bphO, (ii) a pcyA gene encoding a PCB ferredoxin oxidoreductase, and (iii) a hemA gene encoding a 5-aminolevulinate synthase, the first enzyme of the heme biosynthesis pathway (Fig. 1B). These proteins are key enzymes for the synthesis of PCB, the chromophore of Cphs. BrBphP3.ORS278 is also flanked by genes encoding two putative regulatory proteins (a transcriptional factor of the LuxR family and an anti-antifactor) (Fig. 1B). A cluster of gvp genes encoding gas vesicle proteins is also found in the same genomic island (Fig. 1B). Gas vesicles are widely distributed among prokaryotes from aquatic habitats; in particular, they enable cyanobacteria to float up toward the light (22).
BrBphP3.ORS278 Binds PCB as Chromophore-The presence of a pcyA gene in the vicinity of BrBphP3.ORS278 is a strong clue that the native chromophore of BrBphP3.ORS278 is PCB. To test this hypothesis, we overproduced, in E. coli, recombinant BrBphP3.ORS278 proteins using two different constructs. In the first construct, the BrBphP3.ORS278 gene was cloned in the pBAD-HisB expression vector and transformed in an E. coli strain containing the pPL-PCB plasmid. This plasmid, constructed by Gambetta and Lagarias (23), harbors ho1 and pcyA from Synecchocystis sp. PCC6803 for the in vivo reconstitution of the holoCph1 with PCB as chromophore. In the second construct, the bphO and pcyA genes found at the vicinity of BrBphP3.ORS278 were added to the previous pBAD-HisB::BrBphP3.ORS278 plasmid together with synthetic ribosome binding sites to ensure their expression. Both E. coli strains produced large amounts of recombinant proteins. The corresponding purified proteins were blue and presented identical spectral properties (Fig. 2B), indicating that pcyA and bphO genes found in the genomic island are functional and that BrBphP3.ORS278 binds a bilin (i.e. PCB). To gain further evidence supporting that the pcyA gene is involved in the modification of the bilin, we cloned the same synthetic operon but omitted the pcyA gene. In this case, the corresponding E. coli cell pellet was light green and the holo-BrBphP3.ORS278 protein was expressed in low amount and as inclusion bodies (data not shown). This indicates that BrBphP3.ORS278 could bind either BV or PCB as a chromophore but that a correct conformational state for the protein is obtained only with PCB.
Because in phytochromes characterized thus far BV binds to a Cys located at the N-terminal extremity whereas PCB binds to a Cys located in the GAF domain (11,12), we investigated the PCB binding site by mutating the Cys residues (Cys-7 and Cys-16) of the N-terminal region and the Cys residue (Cys-260) located in the GAF domain ( Fig.  2A). Once purified, these three mutated proteins turned out to be colored (data not shown). However, as revealed by zinc-induced autofluorescence (Fig. 2B, inset), the C260S protein failed to bind PCB covalently, at variance with the other mutants. Thus, the covalent binding of BrBphP3.ORS278 chromophore implies specifically Cys-260, which is the attachment site of PCB in cyanobacterial phytochromes (12,24). Altogether, these observations are strong indications that the native chromophore of BrBphP3.ORS278 is a PCB molecule.
BrBphP3.ORS278 Possesses Unusual Absorption Properties-The BrBphP3.ORS278 chromo-protein presents in its dark-adapted state two main absorption bands centered around 355 and 610 nm (Fig.  2B, black line). The latter lies at an atypically short wavelength for a PCB-containing chromo-protein whose Pr state usually peaks around 660 nm. We shall denote this darkadapted form as Po, with "o" standing for orange. Upon illumination with 590 nm light (Fig. 2B, red line), a partial bleaching of the 610-nm band was observed, accompanied by the appearance of a long wavelength band ϳ670 nm, indicative of a red form denoted Pr. However, despite the relatively high illumination intensity used (500 E m Ϫ2 s Ϫ1 ), only a small fraction of the Po form was photo-transformed. This is due, as made clear below, to three reasons: (i) the speed of the back-reaction to the dark-adapted state, (ii) the spectral overlap between the Po and Pr forms, and (iii) their unequal photochemical efficiencies.
The occurrence of a fast back-reaction is illustrated in Fig. 3A showing the kinetics of the absorption changes at 670 nm during and after a 3-s pulse of 590 nm light of variable intensity. The half-time for the dark recovery is ϳ460 ms at 25°C. This is several orders of magnitude faster than reported for any phytochrome or BphP studied so far, in which dark recoveries take place in time ranges spanning from minutes to days (15,24). Fig. 3B shows a plot of the steady-state changes measured in panel A, as a function of the illumination intensity. The data are well  fitted by the function expressing the competition between a photochemical process and a dark back-reaction. Even with the highest light intensity used in this experiment (666 E m Ϫ2 s Ϫ1 ), we still could not fully saturate the absorption change.
The temperature dependence of the dark recovery is examined in Fig. 3C, from 40°C (where t1 ⁄ 2 ϳ 215 ms) down to Ϫ11°C (t1 ⁄ 2 ϳ 30 s). The kinetics are roughly monoexponential. Assuming that Eyring's transition state theory is applicable to this process, one can estimate the enthalpy and entropy of the transition state P# (with respect to state Pr) from an Eyring's plot, ln(k/T) versus 1/T (Fig. 3D). The linear regression yields ⌬H# ϳ 15 kcal mol Ϫ1 and ⌬S# ϭ 7.6 cal mol Ϫ1 K Ϫ1 . We took advantage of the slower back-reaction at low temperature to demonstrate that, once formed, the Pr state can be photo-converted back to Po (Fig. 4). At the low temperature we used (Ϫ11°C), a first illumination with 590 nm light induced the Pr form, which then returned slowly to Po in the dark. When the protein was submitted to a second illumination with 680 nm light preferentially absorbed by the Pr form, a fast increase of the 600-nm absorption was observed, expressing the formation of Po. BrBphP3.ORS278 presents therefore a photo-reversible shift between the Po and Pr forms and behaves, in this respect, like a "classical" (bacterio)phytochrome.
The Two Spectral Forms of BrBphP3.ORS278 and Their Dark Equilibrium-The dark-adapted BrBphP3.ORS278 is clearly predominantly in the Po form. Although this gives a good approximation for the Po spectrum, things are different for the other form because, as mentioned earlier, we have no way of obtaining a total photoconversion. We designed a procedure that allows the determination of the spectra of Po and Pr and also of the relative photoconversion efficiencies of the two forms. The mathematical aspects are described under "Experimental Procedures." This procedure was applied to the data shown in Fig. 5A. The sample was equilibrated at 4°C to slow down the decay to the dark-adapted state, thus facilitating light saturation, which was checked by using neutral density filters. The three experimental spectra (green) were obtained with illuminations centered either in the Po region, or close to the isosbestic wavelength, or in the Pr region (gray spectra). The orange and red spectra are those of the pure forms, Po (peaking at 613 nm) and Pr (peaking at 672 nm), resulting from the mathematical treatment of the data. Interestingly, for the ratio of quantum yields of the two forms, ϭ o / r , one obtains 0.31, meaning that the photoconversion efficiency of Pr to Po is ϳ3-fold larger than the reverse process. The reconstituted spectra Sm (Equation 6) are superimposable to the corresponding experimental spectra. The solution is unique and well defined, both with respect to the Po and Pr spectra and to the relative yield ; in particular, the quality of the fit became clearly unacceptable when imposing equal efficiencies of the two forms ( ϭ 1).
Effect of Temperature on the Dark Equilibrium between the Po and Pr Forms-Unexpectedly, we noticed a clear effect of temperature on the dark-adapted spectrum of BrBphP3. ORS278. When lowering the temperature, one observes an increase of the 670-nm band, accompanied by a decrease in the 610-nm region (Fig. 5B). This is suggestive of a temperature-dependent equilibrium between the Pr and Po forms. This view was confirmed by measuring the extents of the light-induced changes caused by a saturating illumination either in the Pr or Po band. When lowering the temperature, the extent of the Po 3 Pr photoconversion induced by 590 nm light decreased, while the extent of the Pr 3 Po photo-FIGURE 2. Amino acid sequence alignment and absorption spectra of BrBphP3.ORS278. A, amino acid sequence alignment of the N-terminal extremity and subregion of the GAF domain of various phytochromes. Conserved residues by more than 50 and 90% are highlighted in gray and black, respectively. Open and closed triangles indicate the Cys residue used as BV or PCB binding sites, respectively. B, absorption spectra of the dark-adapted state (black line) and under 590 nm illumination (red line) recorded at room temperature; the blue spectrum is the light-dark difference. Inset, picture of a sample of BrBphP3.ORS278. The covalent chromophore binding was visualized by zinc-induced fluorescence for the wild type and the single mutants C7S, C16S, and C260S (lower part). For the upper part, the gel was stained by Coomassie Blue.
conversion induced by 680 nm light increased in a complementary way (not shown).
We used the calculated pure forms of Po and Pr (Fig. 5A) to estimate the fraction of both forms present in the dark, depending on temperature (data of Fig. 5B). A complication is that the Po spectrum becomes significantly shifted toward shorter wavelengths when increasing the temperature. The Pr spectrum, however, shows no significant shift with temperature (as seen from light-induced spectra; data not shown). We thus used the 680-nm region in the spectra of Fig. 5B to estimate the relative fractions of Po and Pr. The deduced equilibrium constant of the equilibrium in the dark varies from [Pr]/[Po] ϳ0.02 at 30°C to ϳ0.27 at 0°C. The theoretical dependence of K eq on T is given by K eq ϭ exp(Ϫ⌬H 0 /RT ϩ ⌬S 0 /R), where ⌬H 0 and ⌬S 0 are the enthalpy and entropy differences, respectively, between Pr and Po. This relation predicts a linear dependence of ln(K eq ) versus 1/T (van't Hoff plot) (Fig. 5C). The estimates of ⌬H 0 and ⌬S 0 obtained from the fit are but a rough approximation because of the limited T and K range. The decrease of K eq when lowering T implies that the Pr form has both the lowest enthalpy and entropy, so that the stable form is Pr at low T and Po at high T. The temperature where the inversion occurs (K eq ϭ 1) is predicted around Ϫ12°C.
Fluorescence Properties of BrBphP3.ORS278- Fig. 6A shows the excitation (continuous line) and emission (dotted line) spectra for a suspension of BrBphP3.ORS278 recorded at room temperature. The excitation spectrum is very similar to the absorption of the Po form of BrBphP3.ORS278, except that the long wavelength band peaks at slightly longer wavelength because of uncorrected instrumental distortion in the red part of the spectrum. A clear demonstration that the Po form is markedly more fluorescent than the photo-induced form Pr is exemplified by the light-induced fluorescence changes (Fig. 6B). These changes, recorded at 4°C for a suspension of BrBphP3.ORS278 subjected to two 6-s pulses of 590 nm light, imply that the formation of state Pr induces an important decrease in the fluorescence yield. The fluorescence properties of BrBphP3.ORS278 do not support a simple identification of its light-induced state with the Pr state of classical phytochromes, despite the similar spectral position. In the latter, Pr is the fluorescent state, whereas the fluorescence is quenched in the far-red form, Pfr (25). In BrBphP3.ORS278 the Po form is fluorescent, whereas the lightinduced Pr state emits no or little fluorescence. In this respect, BrBphP3.ORS278 behaves as other phytochromes, where the short wavelength form is the only fluorescent one.
Attempts to Determine the Role of BrBphP3.ORS278-Because of its localization at the vicinity of gvp genes,  BrBphP3.ORS278 might be involved in the control of the synthesis of gas vesicles. The photosynthetic activity of Bradyrhizobium ORS278 requires both light and strictly aerobic conditions, a particularity shared with the aerobic anoxygenic photosynthetic bacteria (18). One possibility is therefore that BrBphP3.ORS278 is involved in the regulation of gas vesicle synthesis by allowing the bacteria to reach the appropriate ecological niche in oxygen tension and light intensity for optimal growth. Such a situation is encountered in Halobacteria, for example, where the synthesis of gas vesicles is known to be under the control of various environmental factors, including light and oxygen tension (26,27). Several attempts were made to test such a hypothesis by cultivating Bradyrhizobium ORS278 with various growth media and under different light conditions and/or oxygen tension. However, in all these assays we failed to reveal a buoyancy phenotype. We also constructed deletion mutants in BrBphP3.ORS278 and gvpA and examined their phenotypes under various growth conditions. No obvious differences could be observed between the wild type and the mutants concerning, for example, growth rate, photosynthetic activity, etc. Similarly, the use of a lacZ reporter gene did not permit identification of specific conditions for which gvpA is expressed. In contrast, a low but significant ␤-galactosidase activity (ϳ50 Miller units), similar to those observed for different BphP genes of Rps. palustris, 6 was observed for the various conditions tested, indicating that BrBphP3.ORS278 is constitutively expressed.

DISCUSSION
We have described the properties of a new type of bacteriophytochrome present in the photosynthetic strain Bradyrhizobium sp. ORS278. Its gene is located in a horizontally acquired genomic island that displays hallmarks suggestive of recent gene transfer. To our knowledge this is the first clear evidence for the acquisition of a phytochrome gene by a prokaryote through lateral transfer. The presence of a pcyA gene, identified thus far only in cyanobacteria (28), may appear as a strong clue for a cyanobacterial origin of this genomic island. However, phylogenetic analyses ( Fig. 1C and supplemental Fig. S1) indicate that this protein is distant from the five clades of the phytochrome superfamily, including Cphs (6). In addition, phylogenetic analysis using other genes (bphO, hemA, and gvpA) 6 E. Giraud, unpublished results. FIGURE 5. Spectral deconvolution of the Po and Pr forms and temperature dependence of their dark equilibrium. A, the three green curves are (from smaller to larger in the 600-nm region) the steady-state spectra recorded at 4°C under saturating illuminations centered at 590, 633, and 680 nm, respectively. The spectral shapes of these illuminations are shown as gray bands. The orange and red spectra are those of the Po and Pr states computed as explained under "Experimental Procedures." B, spectra recorded from 0 to 35°C (the arrows indicate the evolution when increasing T). C, a van't Hoff plot ln(K eq ) versus 1/T, where K eq is the equilibrium constant [Pr]/[Po] in the dark-adapted state, determined from the contribution of the Pr form to the spectra in panel B in the 680-nm region. The line is a regression whose parameters correspond to ⌬H 0 ϭ Ϫ11 kcal mol Ϫ1 and ⌬S 0 ϭ Ϫ43 cal mol Ϫ1 K Ϫ1 (pertaining to the energy differences Pr Ϫ Po). The inset shows the free energy profile for the Pr 3 Po transition, compiling the present estimates of ⌬H 0 and ⌬S 0 with the information on the transition state P# obtained in Fig. 3D. The gray squares and blue arrows show the magnitude and direction of the entropic contribution ϪT⌬S, for T ϭ 300 K. The vertical scale is given in k B T units.