Correlating structural and photochemical heterogeneity in cyanobacteriochrome NpR6012g4

Significance

Photosynthetic organisms measure changing light conditions with photosensory proteins. In land plants, phytochromes use the ratio of red to far-red light to detect shading by neighboring plants. Cyanobacteria use distantly related cyanobacteriochromes (CBCRs) to detect a broad range of conditions so that they can optimize light harvesting, movement, and other photobiology. Both phytochromes and CBCRs naturally occur as heterogeneous mixtures of populations, but the links between structural and photochemical heterogeneity are unclear. We have solved the structure of a model CBCR in both photostates, allowing us to explicitly connect structural heterogeneity of conserved residues to populations having distinct spectral and photochemical properties.

Abstract

Phytochrome photoreceptors control plant growth, development, and the shade avoidance response that limits crop yield in high-density agricultural plantings. Cyanobacteriochromes (CBCRs) are distantly related photosensory proteins that control cyanobacterial metabolism and behavior in response to light. Photoreceptors in both families reversibly photoconvert between two photostates via photoisomerization of linear tetrapyrrole (bilin) chromophores. Spectroscopic and biochemical studies have demonstrated heterogeneity in both photostates, but the structural basis for such heterogeneity remains unclear. We report solution NMR structures for both photostates of the red/green CBCR NpR6012g4 from Nostoc punctiforme. In addition to identifying structural changes accompanying photoconversion, these structures reveal structural heterogeneity for residues Trp655 and Asp657 in the red-absorbing NpR6012g4 dark state, yielding two distinct environments for the phycocyanobilin chromophore. We use site-directed mutagenesis and fluorescence and absorbance spectroscopy to assign an orange-absorbing population in the NpR6012g4 dark state to the minority configuration for Asp657. This population does not undergo full, productive photoconversion, as shown by time-resolved spectroscopy and absorption spectroscopy at cryogenic temperature. Our studies thus elucidate the spectral and photochemical consequences of structural heterogeneity in a member of the phytochrome superfamily, insights that should inform efforts to improve photochemical or fluorescence quantum yields in the phytochrome superfamily.

Providing reliable food sources in the face of climate change and expanding populations requires substantial improvements in food production (1). Modern agriculture relies on high planting densities of photosynthetic plants, but such densities cause competition among individual plants for light and trigger the shade avoidance response, reducing overall crop yield (2). Shade avoidance responses are controlled by plant phytochromes, large (≥1,100 aa) photoreceptors that reversibly photoconvert between red- and far-red–absorbing photostates to act as master regulators for light-dependent plant development [photomorphogenesis (3–5)]. Improvements in agricultural performance are thus linked to our ability to redirect the phytochrome shade avoidance response in crop plants.

Phytochromes are also found in diverse bacteria, eukaryotic algae, and fungi (6, 7), sharing a conserved three-domain photosensory core module (PCM) (8–11) that binds a covalently attached linear tetrapyrrole (bilin) chromophore. Photoisomerization of the bilin 15,16-double bond allows phytochromes to reversibly photoconvert between two photostates having distinct spectral and biochemical properties with a photochemical quantum yield below 30% (3, 12–15). Phytochromes also exhibit several types of heterogeneity. Known structural heterogeneity includes differences in side-chain rotamers and protonation state (16, 17). Known spectral heterogeneity includes the presence of populations with different peak wavelengths and fluorescence properties (14, 18–20). Spectrally similar phytochrome populations can also exhibit photochemical heterogeneity by having different excited-state lifetimes and quantum yields (12–14). Plant phytochromes also reveal biological heterogeneity, because different populations have different signaling activity within a single photostate (21). The connections between these types of heterogeneity are not well understood, but biological heterogeneity has important consequences for phytochrome function: maximal signaling activity in plant phytochrome A occurs in a population of the red-absorbing dark state generated by photoconversion (21), so the low quantum yield of phytochrome and the presence of other populations limit maximal signaling. Available crystal structures do not elucidate the structural basis for any facet of phytochrome heterogeneity. Solid-state NMR has allowed modeling of heterogeneity in a cyanobacterial phytochrome (17, 22) but does not provide atomic resolution, and the complete phytochrome PCM is too large for structure determination via conventional solution NMR spectroscopy (23).

Fortunately, the phytochrome superfamily also includes cyanobacteriochromes (CBCRs), a spectrally diverse family of cyanobacterial photoreceptors. CBCRs control various aspects of cyanobacterial photobiology (24–27) and exhibit a plethora of photocycles (6, 26, 28–31). Red/green CBCRs such as AnPixJg2 and slr1393g3 exhibit a red-absorbing dark state similar to that of plant phytochrome but are much smaller (6, 32, 33). Crystal structures demonstrate that the PCB chromophore adopts a C5-Z,syn C10-Z,syn C15-Z,anti geometry in the red-sensing state, as in phytochrome (8, 9, 11, 34). AnPixJg2 and slr1393g3 exhibit structural heterogeneity at residues distant from the chromophore, but detailed characterization of spectral or photochemical heterogeneity in these two proteins has not been reported. AnPixJg2 is closely related to NpR6012g4 from Nostoc punctiforme (35), which has been shown to be photochemically heterogeneous on the ultrafast timescale (36–39). NpR6012g4 has also been characterized using solution NMR, with complete chromophore chemical shift assignments and secondary structure already reported for both photostates (40–43).

We here report atomic-resolution solution structures for NpR6012g4 in both photostates. The photoproduct state resembles that of slr1393g3 [Protein Data Bank (PDB) ID code 5M82], whereas the dark state reveals unexpected structural heterogeneity at key residues close to the PCB chromophore. Thus, the protein matrix surrounding the bilin is heterogeneous in red/green CBCRs despite an apparently homogenous chromophore at the level of NMR spectroscopy (40, 41, 44, 45). Comparison of NpR6012g4 variants with the observed heterogeneity of Trp655 predicts the existence of an orange-absorbing population in the wild-type dark state. We confirm this predicted spectral heterogeneity using absorption and fluorescence excitation spectroscopy and demonstrate that the orange-absorbing population does not give rise to the photoproduct. Based on pH responses in both wild-type and variant NpR6012g4, we propose that loss of hydrogen bonds between PCB and the structurally heterogeneous Asp657 causes the spectral shift in this photochemically inactive population. This work thus explicitly links structural heterogeneity at specific residues to spectral and photochemical heterogeneity in a member of the phytochrome superfamily.

Results

Determination of Solution Structures for both Photostates of NpR6012g4.

We previously reported secondary-structure assignments (Biological Magnetic Resonance Bank accession numbers 26577 and 26582) and chromophore chemical shifts for NpR6012g4 in both photostates (40–43). We used these assignments to derive structural restraints and perform high-resolution structural analysis of both photostates. Atomic-level structures were calculated using distance restraints derived from nuclear Overhauser effect (NOE) cross-peaks (Fig. S1) and long-range orientational restraints derived from residual dipolar coupling (46) data (Fig. S2). The first 16 and last 8 residues in NpR6012g4 were dynamically disordered and are absent from the final structural ensembles, so NMR structures were resolved for 155 aa from Ser599 to Val754 using numbering for the full-length protein (GenBank accession number Npun_R6012). The 10 lowest-energy NMR structures for the red-absorbing dark state and green-absorbing photoproduct are shown in Fig. 1 A and B, with structural statistics summarized in Table S1. We obtained an rmsd of 0.61 Å (dark state) and 0.67 Å (photoproduct) for main-chain coordinates. Energy-minimized average structures for each state calculated from the ensembles in Fig. 1 have the expected GAF domain topology (Fig. S3A) with a six-stranded antiparallel β-sheet (β1, R620–F626; β2, V635–A640; β3, K645–V647; β4, N673–V676; β5, A699–A707; β6, Q710–Q719) flanked by four α-helices (α2, V601–L616; α3, H659–A669; α4, H688–Q694; α5, E728–Q747). The average NMR structures of the two NpR6012g4 photostates have an overall rmsd of 2.5 Å, and the average structure of the dark state is quite similar to the crystal structure of the red-absorbing dark state of AnPixJ (rmsd, 1.5 Å) except for the absence of the N-terminal helix and the presence of a shorter C-terminal helix. Photoconversion induces migration of a short stretch of residues around Trp655. In the dark state, the Trp655 indole ring forms a π-stacking interaction with the PCB D-ring, and the indole NH is hydrogen bonded to the A-ring carbonyl oxygen (Fig. 1C). Upon photoconversion, both of these protein–chromophore interactions are lost (Fig. 1D). The indole ring is instead extruded toward bulk solvent and is now close to the C3 side chain on the A-ring, consistent with a similar migration observed in crystal structures of slr1393g3 (PDB ID code 5M82).

Fig. 1.

Solution structures of NpR6012g4. (A) The ensemble of solution structures is shown for the red-absorbing dark state (α-helix, purple; β-strand, cyan; other, green). PCB chromophore and the covalently attached Cys687 are indicated in stick view (gray, carbon; blue, nitrogen; red, oxygen; yellow, sulfur). PDB ID code 6BHN. (B) The ensemble of solution structures is shown for the green-absorbing photoproduct as in A. PDB ID code 6BHO. (C) Trp655 is shown packed onto PCB in the dark state, with residues 653–654 and 655–657 highlighted in pink and purple, respectively. (D) The same region is shown in the photoproduct. (E) The 15Z PCB chromophore of the dark state is shown in the ensemble (Top) and as a schematic (Bottom; P, propionate). (F) The 15E PCB chromophore of the photoproduct is shown as in E.

Protein–Chromophore Interactions in NpR6012g4.

In the dark state, the PCB chromophore of NpR6012g4 adopts the expected C5-Z,syn C10-Z,syn C15-Z,anti configuration (Fig. 1E) seen in the red-absorbing states of AnPixJg2 and phytochromes (8, 9, 11, 34). The A-ring and D-ring adopt α-facial dispositions relative to the approximate plane defined by the B- and C-rings (47). The 8-propionate is well ordered, but the 12-propionate adopts several conformations (Fig. S3B). Photoconversion results in photoisomerization of the 15,16-double bond to the 15E configuration (Fig. 1 E and F) and rotation of the chromophore within the chromophore-binding pocket (Fig. S3C), reminiscent of the flip-and-rotate model proposed for phytochromes (48). The PCB D-ring remains on the α face of the bilin, but the A-ring moves from the α face to the β face while retaining the C5-Z,syn configuration (Fig. 1F). Movement of the A-ring arises due to approximate inversion of the dihedral angle for the 5,6-single bond (Table S2). The photoproduct chromophore exhibits increased tilt of the A-ring relative to the B-ring, more modest increases in B/C ring tilt and C/D ring tilt (Table S2), and increased disorder of the 8-propionate (Fig. S3D). The C17-methyl group is within 5 Å of both the C15 methine proton and the C13 methyl group, in keeping with intramolecular NOE cross-peaks specifically observed in the photoproduct state (40). Overall, the photoproduct chromophore is more twisted than that of the dark state, which should result in weaker conjugation of the bilin π system consistent with the blue-shifted photoproduct absorbance.

Some protein–chromophore interactions are retained upon photoconversion, whereas others change. His659 is proximal to the bilin C10 atom in both photostates (Table S2), consistent with the ability of H₆₅₉C NpR6012g4 to form a second linkage to the chromophore at C10 (35). In the dark state, the 8-propionate forms hydrogen bonds to His688 (10 out of 10 structures), Ser685 (7/10), and His684 (1/10). Upon photoconversion, hydrogen bonds from the 8-propionate to these three residues are still present (9/10, 9/10, and 1/10, respectively). By contrast, the 12-propionate is hydrogen bonded to Arg667 (6/10), His659 (2/10), His684 (1/10), and Tyr668 (1/10) in the dark state but to Tyr668 (7/10) and Tyr700 (9/10) in the photoproduct. There are also changes in the vicinity of the photoactive D-ring. In the dark state, the D-ring and/or C15 methine bridge are within 3.1 Å of Tyr624, Phe634, Trp655, Asp657, Leu660, Ile702, and Tyr718 in eight or more structures (Fig. 2 A and B). In the photoproduct, the D-ring and/or 15-methine bridge are within 3.1 Å of Leu646, Asp657, His688, Leu692, Phe695, and Val697 in eight or more structures (Fig. 2 C and D).

Fig. 2.

Protein–chromophore interactions around the photoactive D-ring of NpR6012g4. (A) Representative stereoview of the PCB-binding pocket of the NpR6012g4 dark state (Top). Heavy atoms of PCB chromophore and covalently attached Cys687 are shown in stick view and colored by atom type (C, cyan; N, blue; O, red; S, yellow). Residues within 3.1 Å of the PCB D-ring and/or C15 methine bridge (see text) are shown in ball-and-stick view and are color-coded by phenotype as discussed in the text (red, effects on the dark state; green, effects on the photoproduct; orange, effects on both; black, no significant effects). (B) Nuclear Overhauser effect (NOE) cross-peaks are shown for the H18² methyl carbon of PCB in the dark state. Residues within 3.1 Å of the D-ring follow the color code of A; other amino acids are indicated in gray, and PCB cross-peaks are indicated in blue. (C) A representative stereoview is shown for the PCB-binding pocket of the NpR6012g4 photoproduct as in A. (D) Nuclear Overhauser effect (NOE) cross-peaks are shown for the H18² methyl carbon of PCB in the photoproduct as in B.

A Photochemically Inert Orange-Absorbing Population in the NpR6012g4 Dark State.

Substitutions for Trp655 have given varying results (15). W₆₅₅V and W₆₅₅A variants both exhibited spectrally heterogeneous 15Z dark states containing both red- and orange-absorbing populations, but W₆₅₅H NpR6012g4 failed to bind chromophore (15). The equivalent Trp289 adopted multiple conformations in a microsecond simulation of the AnPixJg2 dark state (49), so this variation could be explained by structural heterogeneity of this Trp. Remarkably, we observed structural heterogeneity at Trp655 in the NpR6012g4 dark-state ensemble. The π-stacked orientation (hereafter, “Trp-in”) was present in 9 out of 10 structures (Figs. 1C and 3A). The other structure instead exhibited a Trp-out orientation in which the Trp side chain was rotated away from the D-ring about the χ2 angle (Fig. S3E), disrupting both π stacking and the hydrogen bond between Trp655 and the PCB A-ring (Fig. 3B). The Trp-out configuration was seen in ∼10% of the structures in multiple independent ensembles, even though the region from Trp655 to Asp657 was well ordered (Fig. 1C). The side-chain Nε1 atom of Trp655 exhibited modest but significant reductions in heteronuclear NOE (hetNOE) and T₁/T₂ ratio in the dark state compared with the backbone N atoms of residues 655–657 but not compared with those of the more mobile Thr653 and Val654 (Fig. S3F). Such differences were not observed in the photoproduct (Fig. S3G). We thus conclude that Trp655 is a site of structural heterogeneity in the NpR6012g4 dark state, with a minority population adopting the Trp-out configuration.

Fig. 3.

Heterogeneity at Trp655 and Asp657 in the NpR6012g4 dark state. (A) Stereoviews are shown for a representative Trp-in configuration of Trp655. (B) Stereoviews are shown for the Trp-out configuration. (C) Stereoviews are shown for a representative vertical configuration of Asp657, with hydrogen bonds indicated. (D) Stereoviews are shown for a representative horizontal configuration of Asp657.

W₆₅₅V and W₆₅₅A variants accumulate 15Z orange-absorbing species (15), so we reasoned that the Trp-out population in wild-type NpR6012g4 could absorb orange light. Such a population is not seen in the wild-type absorption spectrum, but we confirmed its presence using fluorescence excitation spectroscopy (Fig. 4A). Moreover, the excitation spectrum for this species was similar to that for the orange-absorbing population of W₆₅₅A NpR6012g4 (Fig. 4A), implicating a common chromophore environment for orange-absorbing species in both wild-type and variant NpR6012g4 proteins. These results confirm the presence of spectral heterogeneity in the NpR6012g4 dark state.

Fig. 4.

Spectroscopic analysis of NpR6012g4 heterogeneity. (A) Normalized fluorescence excitation spectra are shown for wild-type (λ_em, 630 nm; brick red) and W₆₅₅A (λ_em, 620 nm; dusty rose) NpR6012g4. (B) Absorption (dotted) and fluorescence excitation (λ_em, 640 nm; solid) spectra are shown for wild-type NpR6012g4 at pH 5 (red) and pH 8 (dark blue). Arrows indicate direction of change with lower pH.

We next sought to enrich the orange-absorbing population in wild-type NpR6012g4 to examine photochemical consequences of spectral heterogeneity. Unlike wild-type NpR6012g4, the W₆₅₅V variant exhibited increased orange absorption with decreasing temperature (Fig. S4 A and B). We reasoned that a similar process might occur in wild-type NpR6012g4 at cryogenic temperatures. Indeed, we observed that rapid cooling of dark-state NpR6012g4 to 150 K resulted in retention of red absorbance, whereas slow cooling resulted in formation of substantial orange absorbance (Fig. 5A). We illuminated the slow-cooled preparation with orange or green light and then compared the resulting difference spectra for primary photoconversion to that obtained upon illumination of the fast-cooled preparation with red light (Fig. 5 A and B). Illumination of slow-cooled NpR6012g4 with orange or green light should preferentially excite the orange-absorbing species, but all three conditions gave near-identical difference spectra (Fig. 5B). Therefore, preferential excitation of the orange-absorbing species did not result in formation of distinct photoproducts at 150 K.

Fig. 5.

The orange-absorbing population in the NpR6012g4 dark state does not undergo full photoconversion. (A) Absorption spectra are shown for 15Z NpR6012g4 at 150 K after fast (red) or slow (green) cooling. Sources for red (630 nm), orange (589 nm), and green (532 nm) excitation are indicated. (B) Difference spectra are shown for illumination of slow-cooled NpR6012g4 at 150 K with green (green) or orange (orange) light and of fast-cooled NpR6012g4 with red light (red). (C) Difference spectra are shown for NpR6012g4 2 ns after excitation with 532-nm light (green) or 630-nm light (red). (D) Difference spectra are shown for NpR6012g4 1 ms after illumination with the same sources as in C. The difference spectrum for photoconversion of NpR6012g4 under static illumination (blue circles) is shown for comparison.

To examine potential reactivity of the orange-absorbing species at physiologically relevant temperatures, we used time-resolved spectroscopy. Previous work had shown that robust signals could be obtained with 532-nm excitation of NpR6012g4 on a timescale of nanoseconds to milliseconds (50), so we used this approach for comparison with previously reported ultrafast spectroscopy (picoseconds to nanoseconds) using 650-nm excitation (38). In this experiment, 532-nm excitation preferentially enhances excitation of the orange-absorbing population (Fig. 5C), and early secondary spectra (<10 ns) showed enhanced negative bleach at 590 nm compared with the bleach at 650 nm (Fig. S5 A–D). Consistent with measurements at 150 K, a single positive band was observed at 680 nm regardless of excitation wavelength (Fig. S5 A and B). The bleach at 590 nm decayed much more rapidly than that at 650 nm (∼20 ns and 1 µs, respectively), but the positive band at 680 nm decayed on both timescales (Fig. S5C). This difference was also observed in the two bleach amplitudes at 2 ns – 200 ns (Fig. S5A): bleach at 590 nm decayed, but that at 650 nm did not.

We conclude that 532-nm excitation generated Lumi-O_f and Lumi-R_f primary photoproducts from the orange- and red-absorbing populations, respectively; the two photoproducts exhibited similar positive absorption but different bleaches and decay kinetics. We used the large difference in decay timescales of these intermediates to estimate the Lumi-O_f spectrum by subtracting the raw 10-ns spectrum from that at 2 ns (Fig. S5E, orange trace), revealing a positive absorption resembling the positive absorption of Lumi-R_f under 650-nm excitation (Fig. S5B). Two other transitions were also clear from the raw signals (Fig. S5 A–D): evolution of the primary photoproduct absorption into a Meta-R_y intermediate peaking at 570 nm and subsequent appearance of a Meta-R_g intermediate peaking at ∼550 nm. Difference spectra for these transitions could be approximated using 1-µs and 200-ns spectra for Meta-R_y and 1-ms and 500-µs spectra for Meta-R_g (Fig. S5E, red and green circles, respectively).

The amplitude of the positive 575-nm band for Meta-R_y formation is comparable to that of the negative 650-nm band for Lumi-R_f decay (compare 200-ns and 1-µs spectra, Fig. S5A). Assuming comparable extinction coefficients for the two primary photoproducts, which is plausible given their similar properties, we can therefore estimate the amplitude of Lumi-O_f in the transient absorption (TA) signals as equal to or greater than that of Lumi-R_f using the decay at 680 nm (Fig. S5 A and C). Were Lumi-O_f to generate a subsequent photoproduct, this species would be readily detectable before decay of Lumi-R_f. However, Meta-R_y did not appear on this timescale (e.g., 50-ns trace: Fig. S5A); instead, the additional bleach observed at 590 nm decayed without appearance of additional intermediates (Fig. 5 A and C).

We next constructed a global target model to analyze the secondary dynamics of NpR6012g4 (Fig. S6). We postulated parallel excitation of red- and orange-absorbing species, each of which generates a primary photoproduct. Consistent with the apparent absence of Meta-R_y formation during Lumi-O_f decay, we postulated that only the red-absorbing species leads to photoproduct. This model estimated a Lumi-R_f species-associated difference spectrum (SADS) nearly identical to the equivalent SADS estimated from the primary data (38) and to the cryotrapped intermediate (Fig. S5F), predicted a Lumi-O_f SADS that strongly resembles the 10-ns minus 2-ns spectrum (Fig. S5E), and accurately described biphasic decay of absorbance at 680 nm (Fig. S5C). Taken together, these results confirm the presence of an orange-absorbing population in the dark state of wild-type NpR0612g4 and demonstrate that this population does not undergo full photoconversion.

Asp657 and Photochemical Heterogeneity.

The behavior of the W₆₅₅A and W₆₅₅V variants implicates a correlation between the Trp-out population and the orange-absorbing species in the wild-type NpR6012g4 dark state but does not explain why that population would exhibit orange absorbance. We noted that the Asp657 sidechain adopts a “vertical” orientation in eight out of nine Trp-in structures but a “horizontal” orientation in the Trp-out structure and the ninth Trp-in structure (Fig. 3 C and D). The vertical Asp side chain makes one hydrogen bond each to the PCB B- and C-ring NH protons and a third to the backbone NH proton of Leu660 (Fig. 3C), whereas the horizontal Asp side chain makes one hydrogen bond each to the backbone NH protons of His659 and Leu660 and does not hydrogen bond to PCB (Fig. 3D). Two conformations are also seen for Asp657 in the photoproduct (Fig. S3 H and I), but these differences do not result in complete loss of hydrogen bonding to the B- and C-rings in the photoproduct state. The horizontal Asp orientation in the dark state results in such a loss, with no obvious counterion for the cationic bilin π system. This environment could lead to orange absorption in the dark state either via chromophore deprotonation or via altered electron density in the absence of hydrogen bonding.

Asp657 is not amenable to site-directed mutagenesis (15), and mutagenesis of Trp655 could alter the equilibrium between the horizontal and vertical configurations of Asp657 regardless of the bilin protonation state. However, the vertical Asp configuration places the anionic carboxylate close to the cationic bilin ring system, whereas the horizontal configuration does not. Therefore, protonation of the Asp side chain would be expected to destabilize the vertical configuration relative to the horizontal configuration by weakening this electrostatic interaction. Deprotonation of the chromophore would also have a similar effect, but deprotonation of PCB and protonation of Asp657 would occur under opposite pH conditions: low pH would suppress the orange-absorbing population were deprotonation of PCB to be the cause for orange absorption but would increase that population were protonation of Asp657 to be the cause.

We therefore examined pH effects both in wild-type NpR6012g4 and in Trp655 variants. Absorption spectroscopy of W₆₅₅V and W₆₅₅A NpR6012g4 clearly demonstrated decreasing red absorbance and increasing orange absorbance at pH 5 compared with pH 8 (Fig. S4 C and D). Similarly, red absorbance is decreased at low pH in wild-type NpR6012g4, whereas fluorescence from the orange-absorbing population is increased (Fig. 4B). We therefore conclude that the orange-absorbing population of NpR6012g4 is favored by protonation rather than deprotonation and cannot be assigned to chromophore deprotonation.

Discussion

We report solution structures of NpR6012g4 in both photostates. The photoproduct state resembles that recently determined for the red/green CBCR slr1393g3 from Synechocystis sp. strain PCC 6803 (PDB ID code 5M82). In both cases, the 15E chromophore adopts a twisted geometry at both the C5 and C15 methine bridges, and this twisted geometry seems the likely cause of the blue-shifted photoproduct absorption seen in this CBCR subfamily. Within the calculated ensemble of structures for the dark state, we observe multiple configurations for Trp655 and Asp657, two residues known to play key roles in chromophorylation and spectral tuning (15). Structural heterogeneity of the equivalent Trp residue in AnPixJg2 has been reported in a simulation, but the crystal structure of the AnPixJg2 dark state only contains the Trp-in conformation (34, 49). We also observe a horizontal configuration for the Asp657 side chain that ablates hydrogen bonding to the chromophore B- and C-rings. Substitution of Trp655 results in variant proteins that accumulate varying amounts of an orange-absorbing 15Z population (spectral heterogeneity), and we demonstrate the presence of a similar population in wild-type NpR6012g4.

We propose that orange absorption arises due to the horizontal Asp configuration. This minority configuration disrupts hydrogen bonding to PCB, which could cause a spectral shift either via formation of a deprotonated bilin π system (51) or via loss of hydrogen bonding to a protonated π system. Orange-absorbing species increase at low pH both in wild-type NpR6012g4 and in variants with substitutions at Trp655 (Fig. 4B and Fig. S4 C and D), demonstrating that the orange-absorbing species does not arise due to deprotonation of the bilin π system. We propose that structural heterogeneity at Trp655 or substitutions for this residue alter the equilibrium between the two configurations of Asp657. We propose that the vertical Asp orientation still occurs with lower occupancy in such variants, explaining the persistence of red-absorbing species. Similarly, the observed pH effects would be explained were the horizontal configuration to be favored by protonation of a titratable group at low pH. The orange-absorbing species exhibited primary photoisomerization but not full photoconversion, indicating that this species may be photochemically inactive in other red/green CBCRs as well. Our studies thus provide a plausible explanation for the 10-fold observed variation in forward quantum yield for these proteins (39), which impacts development of CBCRs for synthetic biology (52–54).

Our work underscores the value of multiple, complementary approaches in structural characterization of photoproteins. Spectroscopic and photochemical heterogeneity is well established in phytochromes and in CBCRs such as RcaE and NpR6012g4 (13, 14, 16, 17, 19–21, 36–39, 51, 55), but connecting this behavior to structural heterogeneity has proven difficult. Using solution NMR structures, we present evidence that even conserved residues proximal to the chromophore can exhibit structural heterogeneity and link that heterogeneity to specific effects in light perception and photochemistry. We anticipate that similar effects connect structural, spectral, and photochemical heterogeneity in other photoreceptor families (56–58). Indeed, photoreceptors and other signaling “input” domains need to be able to switch between two configurations to regulate signaling; this may make them uniquely suited to the study of structural heterogeneity and its functional consequences.

Materials and Methods

Details are described in SI Materials and Methods. This includes information on expression and purification of NpR6012g4, structure determination, and spectroscopic techniques. The PDB ID codes for this work are 6BHN (dark state) and 6BHO (photoproduct).