On the (un)coupling of the chromophore, tongue interactions, and overall conformation in a bacterial phytochrome

Phytochromes are photoreceptors in plants, fungi, and various microorganisms and cycle between metastable red light–absorbing (Pr) and far-red light–absorbing (Pfr) states. Their light responses are thought to follow a conserved structural mechanism that is triggered by isomerization of the chromophore. Downstream structural changes involve refolding of the so-called tongue extension of the phytochrome-specific GAF-related (PHY) domain of the photoreceptor. The tongue is connected to the chromophore by conserved DIP and PRXSF motifs and a conserved tyrosine, but the role of these residues in signal transduction is not clear. Here, we examine the tongue interactions and their interplay with the chromophore by substituting the conserved tyrosine (Tyr263) in the phytochrome from the extremophile bacterium Deinococcus radiodurans with phenylalanine. Using optical and FTIR spectroscopy, X-ray solution scattering, and crystallography of chromophore-binding domain (CBD) and CBD–PHY fragments, we show that the absence of the Tyr263 hydroxyl destabilizes the β-sheet conformation of the tongue. This allowed the phytochrome to adopt an α-helical tongue conformation regardless of the chromophore state, hence distorting the activity state of the protein. Our crystal structures further revealed that water interactions are missing in the Y263F mutant, correlating with a decrease of the photoconversion yield and underpinning the functional role of Tyr263 in phytochrome conformational changes. We propose a model in which isomerization of the chromophore, refolding of the tongue, and globular conformational changes are represented as weakly coupled equilibria. The results also suggest that the phytochromes have several redundant signaling routes.

mophore are important for the initial photoreactions. These include a conserved DIP motif in the GAF domain (3). The initial photoreactions in the immediate chromophore surroundings lead to changes in a so-called tongue extension of the PHY domain. In the Pr state, this PHY tongue adopts a ␤-hairpin conformation, whereas in the Pfr state, it adopts an ␣-helix and a coil (7)(8)(9)(10). The tongue contains a conserved PRXSF motif that binds the DIP motif of the GAF domain. In the Pr state, a conserved Asp residue of the DIP motif, Asp 207 in the Deinococcus radiodurans phytochrome (DrBphP), forms a salt bridge with an Arg residue (Arg 466 ) of the tongue (3,9,11). In the Pr3 Pfr transition, this salt bridge is broken, the entire PHY tongue refolds, and the Asp-interacting arginine is replaced by a serine (Ser 468 ) of the tongue helix (9,10,12). Light-induced refolding of the PHY tongue affects the orientation of the PHY domain, opening the entire PSM dimer (9). In full-length phytochromes, this large-scale change is then relayed by the long connecting helices to the output module, causing changes in its structure and activity (13)(14)(15). The output activity is often a kinase activity but may vary depending on the species, and its changes will lead to biological responses.
The important DIP and PRXSF motifs are closely associated with a conserved tyrosine (Tyr 263 ). The tyrosine resides close to the chromophore (Fig. 1B). A Y263F mutation obstructs the photocycle in the CBD fragment of DrBphP, leading to relatively high fluorescence quantum yield (16,17). The same mutation has also been introduced to the CBD-PHY fragment of cyanobacterial phytochrome Cph1 and plant PhyB (18,19). In this study, we show how Tyr 263 in DrBphP plays an important role in the structural and spectroscopic fine-tuning of the phytochrome activity. By a Y263F mutation, we pinpoint the multiple roles that this tyrosine has for the photocycle and for stabilizing the tongue interactions. The results indicate that the coupling between the conformational states of the chromophore and protein is redundant and weaker than previously assumed and that the two entities may under certain conditions even be uncoupled.

Absorption and fluorescence spectroscopy indicate a role of Tyr 263 in the photocycle
First, we studied UV-visible absorption properties in the CBD and CBD-PHY fragments of DrBphP and their Y263F variants. The absorption spectra of these proteins ( Fig. 2A) resemble the spectrum of the full-length DrBphP in the Pr form with absorption maxima at 698 nm (CBD), 700 nm (CBD-PHY), and 702 nm (CBD-Y263F and CBD-PHY-Y263F) (16,18,20,21).
Generally, red illumination reversibly bleaches the 700-nm absorbance, and the absorption maximum shifts to the far-red region (Fig. 2C). In the CBD-PHY fragments, the absorbance maxima of the illuminated spectra appear at around 755 nm, but in the CBD fragments, this far-red peak is blue-shifted around 10 nm. The shift due to Tyr 263 was hypsochromic in Pfr, as opposed to Pr. This indicates that the interaction of Tyr 263 with the chromophore is different in Pfr compared with Pr. Five times longer illumination times were needed for CBD-PHY-Y263F to reach photoequilibrium compared with the WT CBD-PHY (Fig. S1A). Because the dark conversion rate from Pfr to Pr was hardly affected by the mutation (Fig. S1B), we conclude that the Y263F mutation reduces the photoconversion (Pr3 Pfr) yield and possibly the absorption properties of the Pfr state.
The fluorescence decay properties of the biliverdin molecule in the binding pocket were studied by time-correlated singlephoton counting with an excitation wavelength of 660 nm and monitoring wavelength of 720 nm (Fig. 2E). The fluorescence decays were fitted with exponential functions (23). For Y263F variants, only a single decay component was needed. For the WT samples, multiexponential fitting curves were needed. The  (9). Tongue, conserved DIP and PRXSF motifs (green), Tyr 263 (red), and the chromophore (orange) are indicated. The double-headed arrows represent the changes in the PHY domain separation upon illumination. The CBD part has been shown only for the Pr-state structure for clarity. HK, histidine kinase.
An increase in fluorescence yield implies a decrease in yield of forming the first ground state intermediate, Lumi-R. Thus, we conclude that the Y263F mutation hinders the Lumi-R formation. As demonstrated in the Fig. S1B, the thermal reversion rate from the Pfr state is equally slow or slower in the case of CBD-PHY-Y263F when compared with the WT CBD-PHY. This means that the CBD-PHY-Y263F sample can be populated to a similar photoequilibrium Pfr state with prolonged illumination of the red light as the WT CBD-PHY sample, although the Lumi-R formation is hampered due to mutation of the Tyr 263 .

FTIR spectroscopy reveals reduced PHY tongue refolding in the Y263F mutant
The FTIR spectroscopic analysis gives information from the light-induced changes in biliverdin and the protein. Fig. 3 shows the FTIR difference spectra between the illuminated and the dark states of WT CBD-PHY and CBD fragments and their Y263F mutants. The biliverdin-based transitions are distinguished from those originating from the protein by using uniformly 13 C 15 N-apoprotein-labeled CBD-PHY and CBD-PHY-Y263F samples. The transitions from the biliverdin molecule remain unchanged in the 13 C 15 N-labeled samples, whereas the protein peaks are shifted. As indicated in Fig. S1C, the negative bands at 1734 and 1712 cm Ϫ1 originate from the carbonyl vibration of the biliverdin rings A and D, respectively (24). The large positive band at 1688 cm Ϫ1 in the CBD-PHY is an induced absorption in the Pfr state. Most likely, this positive band originates largely from the carbonyl vibration of the D ring. In addition, changes in the biliverdin vibrations can be observed at 1250 and 1230 cm Ϫ1 (Fig. S1C).
The FTIR difference signal of the CBD-Y263F is small compared with the WT CBD (Fig. 3A, top), which is a result of the Pr state dominating the photoequilibrium after red light irradiation, as described in the UV-visible analysis (Fig. 2B). The 3-fold multiplication of the signal however still highlights a difference signal for CBD-Y263F variant. In particular, the negative band at around 1712 cm Ϫ1 shows that the D-ring environment is affected by illumination. On the other hand, the 1734 cm Ϫ1 band is missing in the CBD-Y263F variant, thus indicating no changes in the A-ring environment under illumination. Interestingly, many FTIR features originating from the protein environment of CBD are present and similar in CBD-Y263F, but a few missing bands in the variant can be assigned as potential signals from the Tyr 263 in the CBD sample. Signals that are completely missing in the CBD-Y263F compared with the WT are found at 1514 and 1209 cm Ϫ1 . Signals with reduced intensity were at 1251, 1527, and 1598 cm Ϫ1 . These frequencies correspond well to Tyr-OH vibrations (25), except the 1527 cm Ϫ1 , which may report changes in the Asp (COO Ϫ1 ) interactions. B, absorption spectra of the variants after irradiation with the 655-nm light. C, absorption difference spectra of the variants (red-irradiated spectra have been subtracted from dark-adapted spectra). Shown are fluorescence emission spectra (D) and emission decay of the same variants (E). The samples were excited at 660 nm, and the fluorescence signal was monitored at 720 nm. The solid lines show the multiexponential fits of the data. F, fluorescence quantum yield measurements and fluorescence lifetimes. The average lifetimes are calculated from the amplitude-weighted sum of the lifetimes observed in the singlephoton counting experiments.

The (un)coupling of the chromophore and structure in DrBphP
In contrast to the CBD fragments, the FTIR difference spectrum of the CBD-PHY-Y263F variant has an amplitude similar to that of the WT CBD-PHY (Fig. 3A, middle). This reveals that after a prolonged irradiation time, these two samples have similar Pfr/Pr ratios, which is in line with the UV-visible analysis ( Fig. 2B and Fig. S1A). The similar negative features at 1712 cm Ϫ1 in the CBD-PHY and CBD-PHY-Y263F variants indicate that their D-ring carbonyls experience similar interactions in the Pr state. On the other hand, the D-ring signal in the Pfr state is downshifted to 1685 cm Ϫ1 in the CBD-PHY-Y263F variant (Fig. 3A, middle and bottom). This indicates that the D-ring carbonyl experiences a different chemical environment in the mutant than in the WT, for example due to altered hydrogen bonding. The slight upshift of the negative 1734 cm Ϫ1 carbonyl vibration of the A ring, observed both in the nonlabeled and 13 C 15 N-labeled samples, indicates that Tyr 263 influences the biliverdin-protein interaction in the Pr state. Small changes are also seen in the region of 1527 cm Ϫ1 (Ϫ), 1514 cm Ϫ1 (Ϫ), and 1506 cm Ϫ1 (ϩ), suggesting changes in the Tyr protonation state and interactions. The differences between the WT and Y263F samples at 1270 -1230 cm Ϫ1 reflect to a large extent the light-induced changes in the biliverdin-protein interactions.
Naturally, the Tyr 263 hydrogen-bonding character between the Pr and Pfr states influences these changes as well (26).
The secondary structural changes in the CBD-PHY domains during the photoconversion are more clear in the 13 C 15 Nisotope-labeled samples, as they separate better from the overlapping signal of the CBD fragment (Fig. 3A, bottom). In particular, the ␤-sheet to ␣-helix transition is now clearly pronounced as a positive-negative feature at 1612 cm Ϫ1 (ϩ)/ 1600 cm Ϫ1 (Ϫ). This can be translated to the signal of the transition in the tongue region, similar to Kennis and co-workers with a signal pair at 1655 cm Ϫ1 (ϩ)/1633 cm Ϫ1 (Ϫ) for the unlabeled samples of bacterial phytochromes from Rhodopseudomonas palustris (10,12). The difference signal amplitude of CBD-PHY-Y263F is about 60% from that of the WT. This indicates fewer secondary structural changes, either due to a larger ␣-helix population in the Pr state or larger ␤-sheet population in the Pfr state in the mutant. The smaller amplitudes of CBD-PHY-Y263F in the changes of secondary structure are also observed in the amide II region, as indicated by the broad negative peak in the double difference spectrum at 1560 cm Ϫ1 (Ϫ) and positive signal at 1506 cm Ϫ1 (ϩ) that have shifted due to isotope labeling to 1540 cm Ϫ1 (Ϫ) and 1483 cm Ϫ1 (ϩ), respec-

The (un)coupling of the chromophore and structure in DrBphP
tively. An additional change at 1580 cm Ϫ1 (ϩ) in labeled spectra may be related to the absorption of aspartate side chain, which would be likely, given the close proximity of Tyr 263 to Asp 207 . In summary, the FTIR difference spectra demonstrate that the yield of the refolding of the tongue is lower in the Y263F mutant.

X-ray solution scattering reveals large-scale changes in all PSM variants
To study the role of Y263F mutation in large-scale structural changes in solution (9), we applied difference X-ray scattering for CBD-PHY and CBD-PHY-Y263F. The results shown in Fig. 3 (B and C) indicate that both variants give a highly similar Pfr-Pr difference signal, which has been assigned to the CBD-PHY dimer opening (9,27). The signals were practically identical in shape (Fig. 3B), indicating that CBD-PHY-Y263F undergoes large-scale movements that are identical to those of the WT protein. The amplitudes of the differences signals, however, varied, which indicates a reduced yield of dimer opening when compared with the WT CBD-PHY (Fig. 3C). This reduction in yield may be due to the lower yield of forming Lumi-R or the reduced yield of tongue refolding. The effect was even more pronounced in the D207H mutant ( Fig. S1D), which has a broken Asp 207 -Arg 466 salt bridge and therefore impaired interactions between the PHY tongue and GAF domain.
Importantly, the X-ray scattering results, together with FTIR, indicate that in addition to the conserved Tyr 263 (and Asp 207 of the DIP motif), there should be other signaling routes that relay the structural signal from the chromophore to the PHY tongue. To reveal the structural details underlying the impaired photocycle and the alternative signaling routes to the tongue, we crystallized Y263F mutants of the CBD and CBD-PHY fragments, respectively.

Crystal structure of CBD-Y263F reveals reduced water interactions
CBD-Y263F was crystallized in the same conditions as the WT CBD (28). Crystal data were processed up to 1.34 Å resolution, and the electron density appears completely comparable with the corresponding WT data. The overall CBD-Y263F structure was highly similar to the WT CBD ( Fig. 4A) (28), with an overall root mean square deviation (RMSD) of 0.102 Å. The most notable difference between the structures is a missing water that resides between the pyrrole water and the Tyr 263 phenyl ring. This "transient water" (W2) is present in some of the CBD structures (17,22,28) and interacts with so-called pyrrole water (W1), Asp 207 , and biliverdin ring A (Fig. 4B). The phenyl ring of the mutated Phe 263 resides slightly closer to the pyrrole water than in the WT Tyr 263 , which may exclude the transient water from the binding pocket. The slight repositioning of the Phe 263 side chain may also be a result of reduced steric effects and the lack of hydrogen bonding of the hydroxyl moiety of this residue. The hydroxyl appears to bind to one solvent water molecule in the WT (W3). This interaction is lost in the Y263F structure, and the water density is reduced (Fig. 4,  C and D). Otherwise, no large changes between the CBD and CBD-Y263F structures were observed, and the pyrrole ring orientations seemed unchanged at this resolution.

CBD-PHY-Y263F in Pr state crystallized exclusively in Pfr-like conformation
Although the WT CBD-PHY formed Pr-state crystals, as previously reported (14), the Y263F mutant did not (Fig. S2A). To obtain Pr structures from the CBD-PHY-Y263F variant, we screened for new crystallization conditions in the dark. Although CBD-PHY-Y263F could not be crystallized in Pr-like crystallization conditions, it crystallized readily in the previously reported Pfr-like crystallization conditions (9). Pre-illumination with red light greatly facilitated the crystal growth (Fig. S2A). These conditions yielded "pre-illuminated" crystals that diffracted up to 3.3 Å resolution. Surprisingly, we found out that both WT CBD-PHY and the Y263F variant could be crystallized in this Pfr-like crystallization condition even if they were kept in the Pr state. We found an additional condition in which the CBD-PHY-Y263F mutant crystallized especially well in complete darkness and without pre-illumination. The resulting "dark" crystals diffracted to 3.6 Å resolution and had an open PSM conformation, as in the previously reported Pfr structures (Fig. 5) (9,29).
The relative tendency of the CBD-PHY variants to form the Pfr-like crystals in the dark varied, hinting at different propensities to adopt the open conformation in the dark (Fig. S2A). CBD-PHY-Y263F crystallized most readily in the "dark" crystallization condition reported here but also in a published Pfr crystallization condition (9). This propensity to crystallize implies that the CBD-PHY-Y263F adopts an open conformation more easily, probably due to the weakened interactions between the tongue and GAF domain. The WT CBD-PHY, on the other hand, crystallized faintly in Pfr-like crystallization

The (un)coupling of the chromophore and structure in DrBphP
conditions without pre-illumination. This implies that it rarely adopts a Pfr-like conformation in darkness due to stronger interactions between the GAF domain and the Pr-state tongue. On some occasions, however, even the WT CBD-PHY crystallized in these conditions. The absorption spectra of the dark crystals ( Fig. 5B) verified that the PAS-GAF-PHY variants have indeed remained in the Pr state from the biliverdin point of view.
The pre-illuminated and dark CBD-PHY-Y263F crystals had highly similar packing and dimensions with each other and with previously reported WT pre-illuminated Pfr crystals (9, 29) (see Table 2). Indeed, both crystal structures closely resembled these published structures. The CBD-PHY mutant dimers seemed to adopt an "open" conformation with separated sister PHY domains and a helical PHY tongue conformation (Fig. 5, A  and C).
Closer inspection of the dark CBD-PHY-Y263F structure revealed that the chromophore-binding pocket resembles that of the Pr-state structure, but the rest of the protein adopts a Pfr state-like conformation. Notably, three residues (Tyr 176 , His 201 , and Phe 203 ) along with Arg 222 adopted exclusively a conformation that is characteristic of the Pr state (Fig. 5D). Although the resolution was not adequate to unequivocally confirm the details of the biliverdin conformation, the chromophore was modeled in the Pr-state (15Z) isomer. The electron density distribution (Fig. S2C) and the absorption spectrum (Fig. 5B) support this biliverdin state. The overall chromophore surroundings were highly similar to our WT Pfr structure, which has mixed Pr/Pfr photochromic state and Prlike side-chain orientations (9).
In the pre-illuminated CBD-PHY-Y263F structure, the orientation of three residues (Tyr 176 , His 201 , and Phe 203 ) was different from that of the dark CBD-PHY-Y263F structure. Whereas in the dark crystals, these residues reside in a Pr-like conformation, the Pfr-like conformations predominate in the pre-illuminated structure (Fig. 5D). It appears that these residues can adopt both Pr and Pfr orientations in this crystal form and that these orientations reflect the prevalent Pr/Pfr ratio within the crystals. The Arg 222 residue close to biliverdin propionate B adopted a Pr-like conformation here, although it has been reported to have a different conformation in a saturated Pfr state (8,29). The chromophore in the pre-illuminated structure is oriented in the pocket as an intermediate of Pr-and Pfr-state biliverdins, closely resembling the Pfr structure published by us (9). Because this crystal form is a result of preillumination with red light, the chromophores were modeled as a Pr/Pfr mixture. In addition, the propionate C of the biliverdin shows some flexibility in pre-illuminated crystals, as the electron density is diffuse in this site (see Fig. S2C), which is in accordance with the results with other Pfr structure of DrBphP PSM (29).
The orientation of the Tyr 176 -Phe 203 -His 201 triplet in the Pr state led to a structural clash between His 201 and the tongue residue Ala 450 (Fig. 5D). This explains the unstructured region in the tongue near Ala 450 that could not be modeled. The dark crystals therefore have substantially

The (un)coupling of the chromophore and structure in DrBphP
more unstructured PHY tongue region than the pre-illuminated crystals (Fig. 5C).

Discussion
The amino acids surrounding the chromophore are highly conserved in phytochromes, which reflects their importance for the function of the protein. Generally, the residues participate in coupling of the chromophore state to the conformation of the rest of the protein. Specifically, a triad of residues, Asp 207 of the DIP motif, Ser 466 /Arg 468 of the PRXSF motif, and Tyr 263 , is expected to couple chromophore isomerization with conformational changes in the PHY tongue. In this paper, we confirmed by mutational, spectroscopic, and structural analysis that Tyr 263 plays a central role in the structural and spectral responses in D. radiodurans phytochrome. It influences the excited-state reactions of the biliverdin and stabilizes the tongue interactions to the GAF domain, thus affecting the structural equilibrium between Pr-and Pfr-like conformations. However, our data show that this coupling is weaker than commonly believed and that it can be modified. The Y263F mutation leads to reduced Pr3 Pfr photoconversion of the biliverdin but also favors a constitutively active conformation of the photoreceptor.

The role of Tyr 263 in early stages of the photocycle
The UV-visible absorption and fluorescence data indicate that Tyr 263 plays an important role in the phytochrome photocycle. The role of this residue is most striking in the CBD-Y263F sample, which has a barely detectable photoconversion and an elevated fluorescence quantum yield (16) (Fig. 2). The increased fluorescence yield hints at a role of Tyr 263 in the excited-state reactions of the biliverdin and therefore in the Lumi-R formation. Somewhat surprising is that the structural differences between the CBD-Y263F and the WT CBD are the missing or absent hydrogen bonds involving a few water molecules (Fig. 4). One missing water resides between Tyr 263 and pyrrole water and potentially has an important role in the photocycle. Whether the changes in the hydrogen-bonding pattern and biliverdin electrostatics are the cause of the divergent photocycle of Y263F is a matter for future research.
The mutated PSM fragment (CBD-PHY-Y263F) also has a higher fluorescence yield compared with the WT PSM, suggesting an impaired Lumi-R formation (Fig. 2). This is in accordance with the fluorescence studies with the Y263F mutant of Cph1 (18). Due to the long thermal reversion times, prolonged illumination conditions can be used to populate the sample to a Pfr state similar to that of the WT PSM. This is consistent with a previous study with cyanobacterial Cph1, where it was concluded that the aromatic character, but not the hydroxyl group, of the conserved Tyr 263 is critical for the Pfr formation (18). The hydroxyl of Tyr 263 facilitates the photocycle and Pfr formation but is not indispensable for these events. The resulting Pfr spectrum of CBD-PHY-Y263F closely resembles that of CBD-PHY, speaking for a biliverdin end state similar to that of the WT protein. This is also apparent in the crystal structures reported here (Fig. 5 and Fig. S2). The ability of CBD-PHY-Y263F to reach the Pfr state indicates that other residues in the binding pocket and/or in the PHY tongue are likely to be engaged in the photoconversion.

The coupling of the PHY tongue to the chromophore
Phytochromes switch between the Pr and Pfr states in response to light, and this change in the photoactive state reflects their output activity state. We propose that the photochromic and conformational states of phytochromes do not necessarily go hand in hand. Although the Pr/Pfr ratio of the bilin chromophore can be shifted by light, the phytochrome activity may be uncoupled from this light regulation. By mutating the Tyr 263 residue at the GAF/PHY interface, we were able to break this coupling. This resulted in a phytochrome that has its chromophore in the Pr state, but its PSM equilibrium is shifted toward a Pfr-like conformation.
This spectral and functional uncoupling has also been shown in other phytochrome species. The mutation of this conserved tyrosine in Arabidopsis thaliana PhyB (Y361F) led to greatly enhanced red light sensitivity (19). There is even evidence that the lack of the entire tongue favors the active state of the phytochrome, as an IsPadC tongue deletion mutant has high diguanylyl cyclase activity even in its dark state (30). The evidence speaks for the role of the conserved tyrosine in stabilizing the tongue interactions to the GAF domain. If this interaction is lost or weakened, the phytochrome more likely adopts a Pfr-like conformation that is in many cases physiologically active (1).
The absence of the Tyr 263 hydroxyl group in the mutant presumably weakens the tongue interactions that involve a salt bridge between the conserved DIP and PRXSF motifs (Asp 207 -Arg 466 ). This effect favors a Pfr-like tongue arrangement over the Pr-like conformation and causes a reduced light responsiveness of the phytochrome molecule. This may explain smaller signals of tongue refolding (Fig. 3A) and large-scale conformational changes of the Y263F mutant (Fig. 3, B and C). The importance of the tongue interactions in light responsiveness is further underlined by the greatly reduced difference X-ray scattering signal of CBD-PHY-D207H (Fig. S1D). This indicates that Asp 207 plays an even more prominent role in tongue stabilization than Tyr 263 .
CBD-PHY-Y263F seemed to crystallize exclusively in an open Pfr-like conformation. This indicates that the mutation makes the PSM prefer this conformation over the Pr conformation in solution. During the crystal formation, packing effects may further favor a Pfr-like conformation by forcing the remaining Pr-state proteins to adopt this form. We note that even the WT PSM molecules in the Pr state can be forced to form Pfr-like crystals by the crystallization conditions (Fig. 5B), although they would be unlikely to adopt a Pfr-like conformation spontaneously. In contrast to CBD-PHY-Y263F in D. radiodurans, the Y263F mutant of cyanobacterial Cph1 has been successfully crystallized in the Pr-like conformation (18). This Cph1 mutant, however, has a slightly different tongue arrangement compared with the WT Cph1 and other Pr-state PSM structures (7, 9, 11, 14, 30 -32). One explanation for this could be the crystal-packing effects; most phytochrome PSM structures in the Pr state, as in DrBphP CBD-PHY (9,14), have crystal contacts between the tongues of the neighboring symmetry mates. If the native tongue conformation is disrupted, The (un)coupling of the chromophore and structure in DrBphP this packing interaction is compromised. Even the slightest changes in the tongue conformation and stability would therefore make the Pr-like crystal formation unfavorable.
The published Pfr-state crystal structures indicate that the Y263F hydroxyl may in theory interact with the biliverdin D ring (Fig. S3A). However, we did not detect any increase in the rate of dark reversion in the Y263F mutant (Fig. S1B), which speaks against the role of Tyr 263 in stabilizing the Pfr state. Residue His 201 in DrBphP is shown to stabilize the Pfr state by interacting with the biliverdin D ring (29). This histidine, however, is not conserved in most prototypical phytochromes (Fig.  S3B). This additional interaction in DrBphP would explain its relatively stable Pfr state and slow dark reversion rate (9,14,20).

Redundant routes for coupling the PHY tongue to the chromophore
Upon photoconversion, three residues close to the D ring, Tyr 176 , Phe 203 , and His 201 , reorient (Fig. 5D). This change in orientation has been noted before (29), but its exact role in the phytochrome signaling remains unclear. Although the reorientation of His 201 enables an interaction with the ring D and a more stable Pfr state in DrBphP, it may have a role different from that of leucine in other phytochromes. Residue Tyr 176 and the aromatic character of Phe 203 are conserved among all phytochromes (Fig. S3B), and their mutation obstructs the photoconversion of DrBphP (21,29).
Although the residue Ala 450 in the helical tongue may affect the reorienting of the Tyr 176 -Phe 203 -His 201 triplet through steric packing interactions (arrows in Fig. 5D), this effect is too weak to affect the biliverdin spectrum in dark crystals. The three residues may allow more space for the D ring to rotate, which would enable the switching of these crystals between the Pr-and Pfr-like states without radically losing crystal integrity. Our data therefore indicate that both Pr and Pfr orientations of the triplet can co-exist in Pfr-like crystal form and that they mainly reflect the photochromic state of the biliverdin. This indicates that it is the chromophore, not the tongue conformation, that is the main determinant of the Tyr 176 -Phe 203 -His 201 orientation. It is possible that this triplet plays a role in phytochrome signaling by affecting the tongue stability; in the Pr state, His 201 points away from the chromophore-binding pocket and destabilizes the Pfr-like tongue conformation around residue Ala 450 . This is visible as missing electron density at this region in our dark Pr-state structure (Fig. 5C). In the Pfr state, on the other hand, His 201 potentially clashes with the 451 WGG 453 region of the Pr-likes tongue, making this combination unlikely. This way, the Tyr 176 -Phe 203 -His 201 route can act as a mediator by favoring the tongue conformation dictated by the chromophore. The residue triplet would facilitate the refolding process during the photoconversion and reduce the prospects of a misfolded PHY tongue. Indeed, mutating the PhyA and PhyB tyrosines that correspond to residues Tyr 176 and Phe 203 in DrBphP renders them constitutively active (33,34).
We propose that there is a ␤-sheet/␣-helix equilibrium for the tongue conformation and that it is affected by several means. The conserved Asp 207 -Arg 466 salt bridge and Tyr 263 play an important role in the stabilization of Pr conformation, whereas the Tyr 176 -Phe 203 -His 201 route may facilitate the coupling of the chromophore state to the tongue conformation. In addition, a conserved proline in the PRXSF motif (Pro 465 in DrBphP) has been proposed to play a role in the conformational trigger in the photoconversion through interaction with the biliverdin A ring (31). Together, these redundant interaction routes may couple the changes in chromophore to the tongue refolding. This coupling can be considered important for phytochromes, because the two states, dark reversion, and phototransition between them have to be supported by a single protein structure. Redundancy in the signal pathway is also consistent with the fact that these residues are highly conserved (Fig. S3B) but not entirely indispensable to the structural photoconversion.

Phytochromes function at several levels that affect each other
In light of the presented results, we propose a model in which light-induced responses in photoreceptors occur at several levels (Fig. 6). Phytochromes can be considered to switch conformation at the levels of the chromophore, the tongue, the PSM, and the entire protein. Each level resides in an equilibrium that is affected by external factors and the other levels. For example, light and tongue interactions affect the Pr/Pfr equilibrium of the biliverdin. On the other hand, the tongue is affected by the state of the chromophore and the quaternary PSM arrangement (35). The PSM arrangement finally affects the structure and hence the activity state of the output module (11,13,36).
By introducing the Y263F mutation, we were able to shift the equilibria within each level. This caused deviant spectral and structural properties but did not totally obstruct the function of the PSM. This speaks for redundant signaling routes within this signaling network, which may be an evolutionarily conserved feature of phytochromes and other signaling enzymes. These properties would offer a photoreceptor sophisticated means to In the case of CBD-PHY-Y263F, the Pr3 Pfr transition is impaired, but the ␣-helical tongue is favored over the ␤-sheet, which increases the prospects of attaining an open PSM that has its biliverdin in the Pr (ZZZ) state. Multiple arrows between the "biliverdin" and "tongue" levels illustrate multiple, partly redundant signaling routes between the levels.

The (un)coupling of the chromophore and structure in DrBphP
fine-tune the light-induced responses without compromising the function of the entire protein. A good example of this is a phytochrome from Stigmatella aurantiaca (SaBphP1), which has apparently defective photocycle (biliverdin level) but undergoes normal light-induced quaternary changes (PSM level) (27). The weak coupling and redundancy between the signaling levels has therefore enabled this phytochrome to adapt spectrally for novel purposes.

Molecular cloning and sample preparation
The constructs encoding the CBD and CBD-PHY fragments (amino acids 1-321 and 1-502) of the DrBphP bacteriophytochrome are described elsewhere (5,21). The Y263F mutation was introduced to the WT fragments with QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Technologies) and primer 5Ј-CCATGCACATGCAGTTCCTGCGG-AACATGGGCGTC-3Ј (nucleotides coding Phe 263 underlined). Protein expression and purification was done as described previously (9,12,20,23). The purified protein samples in final buffer (30 mM Tris/HCl, pH 8.0) were concentrated to 30 mg/ml and flash-frozen. Just before measurements, the frozen phytochrome samples were quickly thawed and filtered with 0.22-m centrifugal filters (Amicon Ultrafree, Millipore) and diluted to the desired concentration with the final buffer. All measurements were carried out in the dark at room temperature.

UV-visible absorption and fluorescence spectroscopy
The steady-state absorption spectra were recorded as described previously (20). The fluorescence quantum yield was determined from the steady-state absorption and emission spectra, and the fluorescence decay was characterized using a time-correlated single-photon counting system (22). The fluorescence data were analyzed using Pygspec (37).

X-ray solution scattering
The X-ray scattering data acquisition was performed at the cSAXS beamline of the Swiss Light Source as described previously (13). In brief, the sample (ϳ30 mg/ml) was pumped through a 1.0-mm quartz capillary and switched between the Pr and Pfr states with a red diode-pumped solid-state laser (671 nm, 5 mJ/mm 2 ; Altechna) and a far-red diode laser (789 nm, 7 mJ/mm 2 ; Thorlabs). While switching the sample, an overlapping X-ray beam (11.2 keV) was used to probe the difference scattering signal. More details can be found in Table 1 and in the supporting information.

FTIR spectroscopy
The samples were concentrated to approximately 2.5 mM. A sample volume of 2 l was pipetted on a CaF 2 window and sandwiched without any spacer between two window plates, resulting in ϳ10-m sample thickness. The hydration was assessed with an absolute IR absorption spectrum in each sample, which resulted in a ratio of 0.9/0.35 between amide I (at 1645 cm Ϫ1 ) and amide II region (at 1580 cm Ϫ1 ). The difference spectra were recorded with an FTIR spectrometer (Nicolet) by utilizing red ( ϭ 655 nm) and far-red ( ϭ 785 nm) lightemitting diodes in a consecutive manner, similar to a method described previously (12). The baseline was recorded during each switch before changing the illumination conditions. About 20 scans of both light-activated states (Pr and Pfr) were acquired.

Protein crystallography
The CBD-Y263F crystals were grown as described (28). The hanging-drop vapor diffusion method was applied, in which 20 mg/ml protein was mixed at a 1:1 ratio with reservoir solution (67 mM sodium acetate, pH 4.95, 3.3% PEG 400, 30% 2-methyl-2,4-pentanediol, 1 mM DTT). Large crystals formed after 24 -48-h incubation at room temperature. For crystallographic data collection, the crystals were flash-frozen with liquid nitrogen in cryoloops under green light. Diffraction data were collected at beamline ID23-2 of the European Synchrotron Radiation Facility (ESRF), under a 100 K cryostream and an X-ray wavelength of 0.873 Å. The data were processed with the XDS program package (38) and cut at 1.34 Å resolution, which corresponds to a correlation coefficient (CC1 ⁄ 2 ) value of 0.31 (39). The crystal belonged to the C121 space group with cell parameters highly similar to those of the published WT CBD structure (28). The CBD-Y263F crystal structure was solved by molecular replacement with Phaser, and structure 2O9C (40) was used as a search model. The structure was refined using REFMAC5 (41). In the refinement, the geometry was restrained with an X-ray matrix-weighting term of 0.5 and anisotropic temperature factors, resulting in R work /R free values of 0.143/ 0.182. The statistics of data collection, processing, structure determination, and refinement of all data sets are summarized in Table 2.
The pre-illuminated CBD-PHY-Y263F crystals were acquired by hanging-drop vapor diffusion by mixing (1:1) 20 The (un)coupling of the chromophore and structure in DrBphP mg/ml protein with reservoir solution (100 mM Tris/HCl, pH 8.5, 200 mM NaCl, 25% PEG 3350) in conditions identical to those described previously (9). Right after setting up the crystallization, drops were pre-illuminated with a red lightemitting diode (655 nm, 5 milliwatts/cm 2 ). The crystals were flash-frozen after a week without cryoprotection. Diffraction data were collected like the CBD-Y263F data at beamline ID23-2 of ESRF. The data were cut at 3.3 Å, which corresponded to a CC1 ⁄ 2 value of 31.2. The pre-illuminated CBD-PHY-Y263F crystal structure was solved by molecular replacement with Phaser. A modified D. radiodurans CBD-PHY structure 5C5K (29) was used as a search model. The side chains of Tyr 176 , His 201 , Phe 203 , and Arg 222 and the biliverdin were removed from the search model and modeled only at later stages of the refinement. Like the WT Pfr crystals (9), the preilluminated CBD-PHY-Y263F crystals belonged to P2 1 2 1 2 1 space group with two dimers in an asymmetric unit. In the refinement, automatically generated local noncrystallographic symmetry restraints were applied, and the geometry was restrained tightly with an X-ray matrix-weighting term of 0.005, resulting in R work /R free values of 0.234/0.267.
The dark CBD-PHY-Y263F crystals were grown by the hanging-drop vapor diffusion method against reservoir conditions (100 mM Tris/HCl, pH 8.5, 200 mM sodium acetate trihydrate, 30% PEG 4000). The Pr state of the protein was ensured by illuminating the crystallization drops briefly with a 785-nm LED (30.7 milliwatts) and keeping the drops in complete darkness at room temperature. Crystals were flash-frozen without additional cryoprotection under green-filtered light. Diffrac-tion data were collected at beamline ID23-1 of ESRF. The data were cut at 3.6 Å (CC1 ⁄ 2 ϭ 34.3). The cell parameters and packing indicated a crystal form highly similar to that of the preilluminated CBD-PHY-Y263F, and the molecular replacement was therefore conducted as above. To minimize the bias caused by the initial search model, the biliverdin and the neighboring residues that differ between Pr and Pfr states (Tyr 176 , His 201 , Phe 203 , and Arg 222 ) were removed and added only at the later stages of the refinement. The final refinement steps were conducted with tight noncrystallographic symmetry restraints between the chains and an X-ray matrix-weighting term of 0.003, which gave final R work /R free values of 0.248/0.287. All electron density maps were prepared from the final structure factor files with FFT of the CCP4 interface, and the crystallography figures were created with the PyMOL Molecular Graphics System version 2.0 (Schrödinger, LLC). The RMSD between WT CBD and CBD-Y263F was calculated with PyMOL after outlier rejection (13 of 315 atoms rejected).

Absorption spectroscopy of crystals
The crystal absorption spectra were measured using a homebuilt microfocusing device connected by optical fiber to a standard deuterium lamp and a diode array spectrometer (Ocean Optics), as in Ref. 9. The crystals were mounted in cryoloops in reservoir solution with 15% glycerol and flash-frozen with liquid nitrogen. The absorption measurements were conducted under a 100 K cryostream, which impairs photoconversion. To test the photoswitching of the crystals, the focused detection light was replaced with a red light-emitting diode (660 nm, 13 a Outer shell values are in parentheses. b Test set for R free calculation constitutes 5% of total reflections that were randomly chosen. c Number of protein molecules in an asymmetric unit: one (CBD-Y263F) and four (CBD-PHY-Y263F dark and pre-illuminated). d This includes atoms from a biliverdin, three acetates and a (4S)-2-methyl-2,4-pentanediol (CBD-Y263F), or four biliverdins (pre-illuminated and dark CBD-PHY-Y263F).
The (un)coupling of the chromophore and structure in DrBphP milliwatts; Thorlabs) for 5-15 s after the cryostream (and other light sources) was temporarily blocked. Visual inspection showed that the crystals remained intact through the brief illumination treatments.