Protonation Heterogeneity Modulates the Ultrafast Photocycle Initiation Dynamics of Phytochrome Cph1

Abstract

Phytochromes proteins utilize ultrafast photoisomerization of a linear tetrapyrrole chromophore to detect the ratio of red to far-red light. Femtosecond photodynamics in the PAS-GAF-PHY photosensory core of the Cph1 phytochrome from Synechocystis sp. PCC6803 (Cph1Δ) were resolved with a dual-excitation-wavelength-interleaved pump-probe (DEWI) approach with two excitation wavelengths (600 nm and 660 nm) at three pH values (6.5, 8.0, and 9.0). Observed spectral and kinetic heterogeneity in the excited-state dynamics were described with a self-consistent model comprised of three spectrally distinct populations with different protonation states (P_r-I, P_r-II, and P_r-III), each composed of multiple kinetically distinct sub-populations. Apparent partitioning among these populations is dictated by pH, temperature, and excitation wavelength, Our studies provide insight into photocycle initiation dynamics at physiological temperatures, implicate the low-pH/low temperature P_r-I state as the photoactive state in vitro, and implicate an internal hydrogen-bonding network in regulating photochemical quantum yield.

Graphical Abstract

Emerging genetically encoded biological materials allow researchers to control biological activity with light at the cellular or sub-cellular level.^1–7 A common approach in developing such optogenetic materials involves fusing existing light-sensing domains to non-native output domains (e.g., histidine kinase domains or DNA binding proteins) to make novel hybrid photoreceptors. A detailed understanding of the mechanisms that couple light to biological function in existing photoreceptor proteins aids in the engineering of these optogenetic tools. Of the known photoreceptor families, bilin-binding proteins such as phytochromes have attracted interest for such efforts because of their response to long wavelengths that penetrate deeper into mammalian tissue.⁸

Phytochromes are widespread photoreceptors using covalently attached linear tetrapyrrole (bilin) chromophores to switch between red-absorbing P_r and far-red absorbing P_fr states (Fig. 1A–B).^9–12 The N-terminal PAS-GAF-PHY photosensory core module of full-length Cph1 (amino acids 1–514, hereafter termed Cph1Δ) from the cyanobacterium Synechocystis sp. PCC6803 is one of the most well studied model systems for exploring phytochrome photodynamics due to its robust recombinant expression and known crystal structure (Fig. 1C).^13–15 Full-length Cph1 includes a C-terminal histidine kinase domain, but Cph1 and Cph1Δ exhibit nearly identical fluorescence spectra, activation barriers for the Pr photoreaction, and photodynamics.¹⁶ Red illumination of the dark-adapted P_r state of Cph1Δ (Fig. 1B; red curve) initiates forward photoconversion of the phycocyanobilin (PCB) chromophore (Fig. 1A) to generate the isomerized Lumi-R primary photoproduct on a picosecond timescale.^17–22 Lumi-R evolves on the electronic ground-state surface via several intermediates to generate P_fr (Fig. 1B; dark red curve) on a >100 ms timescale.^{23, 24} The dark-adapted P_r state is then regenerated from P_fr either by far-red light (~700 nm) or via thermal reversion (>24 h).^{9, 13, 25}

Figure 1.

(A) The phycocyanobilin (PCB) chromophore of Cph1 is shown in the 15Z (right) and 15E (left) conformations. (B) Absorption spectra are shown for dark-adapted P_r with a 15Z PCB chromophore (red) and for calculated P_fr with a 15E PCB chromophore (dark red) at pH 8.⁵² (C) The crystal structure showing PAS-GAF-PHY domain architecture of Cph1.¹⁵

Under physiological conditions, most proteins continually flex, rotate and sample different metastable states on a complex multi-dimensional potential energy surface.^{26, 27} For phytochromes, as for photoreceptors in general, these dynamics result in multiple co-existing populations that often exhibit distinct spectra (“spectral heterogeneity”) and/or distinct kinetics (“temporal heterogeneity”). Both types of heterogeneity reflect structural differences in the chromophore itself, such as in protonation state or facial disposition of the bilin rings,²⁸ and/or in the surrounding protein scaffold, such as sidechain conformation²⁹ or protonation state.³⁰

Through analysis of the temperature dependence and excitation wavelength dependence of Cph1Δ fluorescence, Sineshchekov and coworkers provided evidence for the presence of multiple co-existing P_r populations.¹⁶ Using variants of Cph1Δ incorporating specific mutations, they argued that photochemical quantum yield correlates with differences in the relative activation barriers for bifurcating processes on the excited state surface or for partitioning between different sub-populations with different photoactivities.^{31, 32} This interpretation assumes that the P_r ground state is spectrally homogenous in solution.³³ However, Matysik and co-workers used magic angle NMR measurements of Cph1Δ at 243 K and pH 8 to resolve two co-existing P_r populations (P_r-I and P_r-II) with different hydrogen bonding networks around the embedded PCB chromophore.²³ Heyne and co-workers similarly argued for inhomogeneity in Cph1 based on polarized transient infrared signals.³⁴ This work demonstrated the existence of structural heterogeneity in the P_r ground state (i.e., without illumination).

Spectral heterogeneity has also been observed in wild-type Cph1 by examining the wavelength and temperature dependence of static and ultrafast transient absorption (TA) signals for Cph1Δ at pH 8, revealing two co-existing P_r populations denoted ^photoP_r and ^fluorP_r.³⁵ Excitation of each of these populations yielded multiple spectrally identical excited-state P_r* sub-populations that decayed on different timescales (temporal heterogeneity), with P_r* sub-populations arising from ^photoP_r excitation being spectrally distinct from those arising from ^fluorP_r excitation. The fluorescence excitation spectrum of Cph1Δ corresponds well with the extracted ^fluorP_r static spectrum, so the room-temperature fluorescence spectrum of Cph1Δ was assigned to this species.³⁵ This interpretation led to the conclusion that only one ^photoP_r sub-population was almost exclusively responsible for generation of the Lumi-R primary photoproduct. Remarkably, this ground state heterogeneity is dependent on a temperature-dependent dynamic equilibrium: the longer-living ^fluorP_r population is favored at elevated temperatures, resulting in inverted Årrhenius behavior for photochemistry.³⁵ Such behavior is difficult to reconcile with a single homogeneous P_r ground state.

Further support for a heterogeneous P_r ground state was obtained in a recent study of the pH dependence of the static absorption and resonance Raman spectra of Cph1Δ,³⁶ which identified three populations with differing pH-dependent amplitudes that were ascribed to different intra-protein hydrogen-bonding networks:

• P_r-I that dominates at low pH values (<pH 7)

• P_r-II that has its greatest amplitude at ~pH 8.5

• Pr_dep that dominates at higher pH values.

Despite this robust evidence for ground state heterogeneity in the P_r state of Cph1Δ, other studies continue to use homogeneous models for data analysis. However, ignoring ground state heterogeneity would lead to incorrect conclusions about Cph1Δ’s underlying ultrafast dynamics. For example, a homogeneous model was used to interpret the femtosecond stimulated Raman signals of Cph1Δ at pH 8, leading to the proposal that isomerization occurs exclusively on the excited-state surface.²² Subsequent broadband pump-dump-probe spectroscopic signals revealed this to be an erroneous conclusion.³⁷ Similarly, 2-dimensional (2D) TA spectroscopy was recently used to conclude that Cph1Δ is homogeneous at 298 K and pH 8.³⁸.

The present study was undertaken to reexamine heterogeneity of the P_r ground state of Cph1Δ in light of the reported 2D results. We consider the following questions: If Cph1Δ is heterogeneous, why do 2D signals³⁸ support a homogeneous P_r model? What is the basis of the temperature-dependent heterogeneity observed by Kim et al.³⁵ and the pH dependent heterogeneity observed by Escobar et al.?³⁶ What consequences does this heterogeneity have for the functional photoactivity of Cph1Δ? To address these questions, we characterized the excitation-wavelength dependence of ultrafast TA signals at three pHs (6.5, 8.0, and 9.0). To minimize differences in setup between datasets, we used Dual-Excitation-Wavelength-Interleaved Transient Absorption (DEWI)^39–41 to collect two alternating broadband TA signals near simultaneously (within 4 ms) on a single sample (here, a single pH value) while maintaining all other experimental conditions unchanged. This protocol removes errors caused by long-term fluctuations of excitation power, sample degradation, and different experimental conditions. As such, DEWI TA measurements represent a powerful approach to interrogate the photo-induced kinetics of underlying ground state populations with different absorption properties. Based on these measurements, we have been able to re-interpret the temperature dependent TA and static absorption signals from Kim et al.³⁵ to construct a self-consistent, heterogeneous global model describing the photoinitiation dynamics of Cph1Δ that is in strong agreement with the pH dependent vibrational signals from Escobar et al³⁶ and that can simulate the reported 2D TA spectroscopy results.³⁸

We began by examining the static absorption of our preparation of Cph1Δ as a function of pH. In agreement with previous reports,^{24, 36} the dark-adapted (P_r) absorption spectrum of Cph1Δ was pH dependent (Fig. 2). At pH 6.5, the absorption spectrum peaked at 664 nm (Fig. 2, dark red curves) with a high-energy shoulder at 610 nm. At pH 8.0, the spectrum blue-shifted slightly to 662 nm (orange curves), exhibiting a slight increase of the high-energy shoulder. The spectrum changed significantly at pH 9.0 (blue curves), with a more prominent shoulder and a further blue shift to 650 nm. These data thus can be interpreted as arising from at least two species, with one or more blue-shifted populations becoming more important contributors to the observed spectrum at high pH. This interpretation can also be applied to previous studies of pH^{24, 36} and temperature dependence.^{33, 35} Hence, it is possible that the same underlying heterogeneity of Cph1Δ in the P_r state can be resolved in both temperature-dependent and pH-dependent spectra. We therefore used DEWI measurements at different pH values to provide a comparison to our previous use of the DEWI technique at different temperatures.³⁵

Figure 2.

(A) Normalized static absorption spectra are shown for Cph1Δ in the P_r state at pH 6.5 (dark red curve), pH 8.0 (orange curve) and pH 9.0 (blue curve). (B) Detailed view of the peak region.

Selected DEWI TA spectra and kinetics of Cph1Δ at three pHs (6.5, 8.0, and 9.0) are contrasted in Figs. 3 and S1–S4. These measurements reveal a strong excitation-wavelength and pH dependence of the primary photodynamics. At pH 8, the most common pH used for ultrafast studies, the 660-nm excitation TA spectrum at 1 ps (Fig. 3B; red curve) exhibited a pronounced negative band around 675 nm, attributed to overlapping ground-state bleach (GSB) and simulated emission (SE) bands, with a broad positive excited-state absorption (ESA) band from 450 nm to 625 nm.³⁵ The 600-nm excitation TA spectrum (blue curve) exhibited similar features but had a slightly blue-shifted (10 nm) GSB/SE band and a significantly weaker ESA from 550–625 nm. The raw spectra resembled the previously described ^photoP_r* (660 nm excitation) and ^fluorP_r* (600 nm excitation) populations.³⁵ These results show that this preparation behaves similarly to that used in the previous DEWI study and confirms that at least two populations co-exist at pH 8. The ^fluorP_r* population is longer lived than ^photoP_r*, consistent with initiation of the Cph1 photocycle via one or more members of the ^photoP_r* sub-population ensemble (Fig. S5).

Figure 3.

Comparison of the DEWI TA spectra of Cph1Δ 1 ps after excitation at 600 nm (blue curves) and 660 nm (red curves) at pH 6.5 (A), pH 8.0 (B) and pH 9.0 (C). The spectra obtained at 600 nm excitation are scaled to the amplitude of the band at 660–675 nm of the spectra obtained at 660 nm excitation at pH 8.0. DEWI spectra at 0.5 ps and 5 ps are presented in Figs. S1 and S2.

Differences between excitation wavelengths were less pronounced at pH 6.5. The 660-nm excitation spectrum (Fig. 3A; red curve) strongly resembled the ^photoP_r* spectrum observed at pH 8 (Fig. 3B; red curve). The 600-nm spectrum at pH 6.5 (Fig. 3A; blue curve) also resembled ^photoP_r*, although with a weaker amplitude ^fluorP_r* population. This behavior is also reflected in the kinetics (Figs. S3 and S4). Hence, while the ^photoP_r*/ ^fluorP_r* heterogeneity observed at pH 8.0 also exists at pH 6.5, lower pH skewed the P_r population ensemble toward ^photoP_r*. In contrast, the pH 9.0 signals exhibited a more pronounced excitation-wavelength dependence. The 1-ps spectrum after 660 nm excitation at pH 9.0 (Fig. 3C; red curve) was more similar to that observed for 600 nm than that observed for 660 nm at pH 8.0 (Fig. 3B; blue and red curves, respectively). The 600-nm excitation spectrum at pH 9.0 (Fig. 3C; blue curve) exhibited a strong negative GSB at 600 nm that overlaps the GSB/SE band already attributed to ^photoP_r*. Hence, the heterogeneity observed at pH 9 clearly differs from that at pH 6.5 and 8.0 and may include a new ‘deprotonated’ population that was preferentially excited at 600 nm. Consistent with this interpretation, this new population also exhibited longer living kinetics than both ^fluorP_r* and ^photoP_r* (Fig. S4A).

The DEWI data qualitatively support the hypothesis that the pH-dependent and temperature-dependent signals reflect underlying heterogeneity arising from three spectrally distinct ground state populations, as proposed.³⁶ We can thus combine the nomenclature from this work and from the previous DEWI study,³⁵ as follows:

• P_r-I is ^photoP_r,

• P_r-II is ^fluorP_r, and

• P_r-III is the new population observed at high pH with a deprotonated PCB chromophore (previously called Pr_dep).³⁶

It is important to note that the P_r-III population is not red-absorbing; rather, it exhibits a green-yellow GSB resembling that observed for the green-absorbing 15Z state of cyanobacteriochrome (CBCR) RcaE.⁴¹ Chromophore deprotonation is known to cause a green-yellow absorption of free PCB extracted from Synechocystis sp. PCC 6803,⁴² green absorption in RcaE³⁰ and is thought to cause similar peak absorption in variants of another CBCR.²⁹ However, we have chosen to retain the designation Pr-III in the current work for simplicity.³⁶ Applying this nomenclature to our data, the P_r sample at pH 9.0 consists primarily of P_r-II and P_r-III populations, whereas the pH 8 sample consisted primarily of P_r-I, with some P_r-II, and the pH 6.5 sample was almost exclusively P_r-I.

We previously employed multi-wavelength global analysis to develop a framework for interpreting temperature-dependent DEWI.³⁵ This model successfully described data taken at pH 8, but it seemed likely that it would be insufficient at pH 9 because of the possible presence of P_r-III. We therefore extended this model to evaluate the six DEWI TA datasets from Fig. 3, with the goal of having a single, self-consistent model to describe the ultrafast photodynamics of multiple ground state populations as a function of pH, temperature, and excitation wavelength.

The proposed model includes three spectrally distinct populations (color boxes in Fig. 4), each of which is an ensemble of multiple temporally-distinct sub-populations (e.g., P_r-Ia*). The three ground state populations are in a dynamic equilibrium that is sensitive to both pH and temperature. We estimate that initial Franck-Condon cooling occurs after excitation with a 120-fs time constant, but current data do not allow us to resolve differences in Franck-Condon dynamics between the sub-populations. The three populations then give rise to three spectrally distinct relaxed excited-state populations, each of which decays as two (P_r-I and P_r-III) or three (P_r-II) spectrally identical but temporally distinct sub-populations.

Figure 4.

Integrated target model for DEWI TA signals at different pH values (6.5, 8.0, and 9.0). P_r populations P_r-I, P_r-II, and P_r-III are colored teal, green, and purple, respectively. The relative occupancies of initially excited populations under different experimental conditions are shown in Table S1. The pKa values are estimated from Escobar et al.³⁶

Global analysis was used to extract the characteristic species-associated difference spectra (SADS, Fig. 5) from this target model for P_r-I* (teal curve), fluorescent P_r-II* (green curve), and deprotonated P_r-III* (purple curve). The SADS were in good agreement with the raw DEWI TA data in Fig. 3 and the temperature dependent DEWI data of Kim et al. (Fig. S6).³⁵ Moreover, this model accurately described the observed multiphasic kinetics as a function of pH and excitation wavelength (Figs. S3 and S4) as well as describing kinetics as a function of temperature (Fig. S7).³⁵ Therefore, we conclude that a single model can describe the observed heterogeneity of Cph1Δ in the P_r state under these different conditions. However, this model would also need to describe the apparent homogeneity of Cph1Δ P_r in 2D spectroscopy.³⁸ In multi-chromophore systems with strongly congested spectral features like photosynthetic complexes,⁴³ fully resolved 2D signals allow tracking the dynamics of all spectrally distinct populations. This approach can suffer from Fourier transform artifacts such as phasing, but this does not seem to be the case for 2D study of Cph1Δ.³⁸ We therefore tested whether “apparent” homogeneous 2D data of Cph1Δ (measured at pH 8 and room temperature) can be simulated using the heterogeneous model presented in Figs. 4 and 5. The 9-ps 2D spectrum was calculated using a simple kinetic approach (see Supporting Information) from the spectral and kinetic responses for the induced dynamics under these conditions (Fig. 4). The response for each state (R_Pr-I, R_Pr-II and R_Pr-III for P_r-I, P_r-II and P_r-III states, respectively) is expanded in terms of constituent dynamics of the sub-populations (e.g. P_r-Ia and P_r-IIb for the P_r-I state):

$$R_{Pr}\left(t,{\omega }_{probe}\right)=\sum _{i=1}^n{c}_i\left(t\right){\Delta \epsilon }_i\left({\omega }_{probe}\right)$$

where i correspond to the relevant transient species in each of the three states (e.g. ^FCP_r-Ia, P_r-Ia, ^FCP_r-IIb, P_r-IIb and Lumi-R for the P_r-I state), ${\Delta \epsilon }_i$ are the estimated SADS (Fig. 5), $c_i\left(t\right)$ are the associated concentration profiles for each transient population and t is the time between excitation and probe pulses at is set to 9 ps for the simulation in Figure 6. These responses for the three states are then individually scaled by the associated absorption spectra $\sigma \left({\omega }_{excite}\right)$ for each state, estimated from temperature dependent analysis,³⁵ to simulate the 2D spectra:

$$2D\left({\omega }_{excite},t,{\omega }_{probe}\right)=B\left({\omega }_{excite},{\omega }_{probe}\right)\left[{\sigma }_{P_r-I}\left({\omega }_{excite}\right)R_{P_r-I}\left(t,{\omega }_{probe}\right)+{\sigma }_{P_r-II}\left({\omega }_{excite}\right)R_{P_r-II}\left(t,{\omega }_{probe}\right)+{\sigma }_{P_r-III}\left({\omega }_{excite}\right)R_{P_r-III}\left(t,{\omega }_{probe}\right)\right]$$

where $B({\omega }_{excite}, {\omega }_{probe})$ is a bandwidth factor to match the spectral response of the 2D experiment (not needed for TA signals) that is modeled by a 2D Gaussian with 200-nm Full Width at Half Maximum for this to matched the bandwidth of the experiment.³⁸ $R_{P_r-III}$ was not included in the simulation because the P_r-III population is negligibly populated at pH 8 and room temperature.

Figure 5.

Comparison of SADS for different excited-state P_r populations: photoactive (P_r-I*, teal curve), fluorescent (P_r-II*, green curve) and deprotonated (P_r-III*, purple curve). The model is summarized in Fig. 4 with parameters in Table S1.

Figure 6.

Simulated 2D electronic spectrum at 9 ps calculated using the model proposed in Figs. 4 and 5 with parameters in Table S1. Purple line, the diagonal; filled circles, peak of the bleach/SE band as a function of excitation frequency. The absorption spectra (left curves) and SADS (top curves) are color coded to match the P_r-I and P_r-II populations in the global model of Fig. 4.

Remarkably, the simulated 9-ps 2D spectrum strongly resembles the published experimental spectrum (Fig. 6).³⁸ In particular, the GSB/SE band (red peak) is off the diagonal and exhibits an elongation along the excitation frequency axis. Elongation of the spectra along the diagonal is considered a key hallmark of heterogeneity in 2D spectra (diagonal shown in Fig. 6 as a purple line);⁴⁴ indeed, its absence in the experimental 2D data is a key part of the rationale advanced for a homogeneous Cph1Δ P_r state.³⁸ However, the heterogeneity observed in the DEWI-TA signals is observed in the simulated 2D spectra as a slight shift in the observed peak emission frequency as a function of excitation frequency and as a shifting peak of the signal (sliced along the probe frequency) as function of excitation frequency. This peak shifts from ~425 THz at low-energy excitation frequencies (i.e., red wavelengths) where P_r-I is preferentially excited (Fig. 6, teal curves) to 447 THz at high-energy excitation frequencies (i.e., blue wavelengths) where P_r-II is preferentially excited (green curves). The same trend can be seen in the published experimental 9 ps spectrum.³⁸ The expected “elongation” is actually a shift in peak emission frequency as the excitation frequency changes, and that behavior is present in Cph1Δ. The frequency shift does not strictly follow the diagonal in Cph1Δ, because it is not continuous variation; instead, there is a discrete transition between P_r-I and P_r-II as the excitation wavelength changes. The 2D data do not demonstrate that Cph1Δ is homogeneous, but rather provide a high-quality demonstration that the underlying heterogeneity is not continuous.

The heterogeneity of the three populations, P_r-I, P_r-II, and P_r-III, arises on the ground-state surface, because these populations can be distinguished by changes in static absorbance spectroscopy and in the GSB bands in TA spectroscopy as their relative occupancies vary with changing pH and temperature. Each of these populations is composed of two to three spectrally identical sub-populations that evolve with different time constants. The structural basis for these kinetically distinct sub-populations is less clear: they may also arise from ground-state heterogeneity, or they may arise due to bifurcation on the excited-state surface. For example, it is possible that the presence of alternate side-chain rotamers for an aliphatic residue near the chromophore would not result in spectral differences but would result in different steric clashes as the chromophore undergoes photoisomerization. In this case, ground state heterogeneity would manifest in kinetically distinct but spectrally identical sub-populations. However, it is also possible that there are multiple pathways for structural evolution of the chromophore during photoisomerization that need not produce detectably large spectral shifts, such as different directions for rotation of the D-ring.⁴⁵

The protein-chromophore hydrogen bonding network in Cph1 involves the PCB chromophore, bound water molecules, and multiple protein residues (Fig. 7).¹⁵ The conserved His260 is proposed to be a key residue in this network: vibrational spectroscopy implicates differences in the protonation state of His260 in the differences between P_r-I and P_r-II,³⁶ and a combination of spectroscopic techniques and theoretical modeling has allowed proposal of structural models for the two species (Fig. 7).^{46, 47} However, the equilibrium between P_r-I and P_r-II is accompanied by a significant change in heat capacity (ΔC_p = 446 ± 53 cal mol⁻¹ K⁻¹).³⁵ It thus seems plausible that structural differences between these populations propagate beyond the immediate milieu of the chromophore.

Figure 7.

Chromophore pocket of Cph1 in the proposed P_r-I and P_r–II configurations.^{15, 46} (A) Model of the proposed hydrogen bonding network (teal dashed lines) formed when His260 is protonated, derived from solid-state NMR and resonance Raman spectroscopy.⁴⁶ (B) P_r-II crystal structure (2VEA).¹⁵ Alternative perspectives are shown in Figs. S8 and S9.

Our data support the conclusion that the low-temperature/low pH P_r-I state is the photoactive state in vitro. Based on Resonance Raman and solid-state NMR spectroscopy, the P_r-I state possesses a cationic His260 that is H-bonded directly to the C-ring propionate and held orthogonal to the positively charged chromophore (Fig. 7A). If the low-temperature/low-pH state is photochemically active, it follows that the blue-shifted higher-temperature/higher-pH P_r-II state can be assigned to the photoinactive, fluorescent population. Relative to P_r-I, P_r-II is thought to possess an altered hydrogen-bonding network with a bridging water that lies between the neutral imidazole sidechain of His260 and the C-ring propionate and hence would match the interactions seen in the crystal structure (Fig. 7B). In this regard, it is interesting to note that crystals used to determine the Cph1 structure⁴⁶ were grown at 18°C (favoring P_r-I) and slightly acidic pH (favoring P_r-II),¹⁵ so it seems plausible that either state (or both) could be present.

Finally, P_r-III possesses a deprotonated PCB chromophore with an unresolved hydrogen-bonding network. No photoproduct from this population was observed in the DEWI TA data, and it exhibits an even longer excited-state lifetime than P_r-II (Fig. 4). We therefore propose that P_r-III is even less active than P_r-II. The P_r-III state is not appreciably occupied at pH≤8. Significant pH fluctuations can occur transiently under high light illumination of cyanobacteria, leading to the idea that the CBCRs controlling complementary chromatic acclimation like RcaE are regulated by both light and pH.³⁰ The photosensory function of Cph1 may be similarly modulated when light fluence rates are high.

Our work is consistent with a critical role for His260 in dictating effective photochemical quantum yields in phytochromes. His260 is highly conserved in phytochromes and lies within 3.0 Å of the PCB chromophore (Figs. S8 and S9). Given the positive charge on the PCB chromophore, the protonated His260 sidechain in P_r-I would be expected to adopt a conformation that minimizes electrostatic repulsion of the two cationic moieties. It has been proposed that a charged Arg residue in PYP is responsible for shifting the excited-state surface to direct the excited state toward an isomerization coordinate and away from non-productive quenching coordinates (single bond rotations).⁴⁸ However, the H260Q mutant of Cph1Δ exhibits static spectral features of the P_r-I state and still photoconverts to P_fr,⁴² so the positive charge of protonated His260 is not essential for photoconversion in phytochromes.

Understanding the molecular basis for this dynamic heterogeneity also holds great potential for the rational design of efficient phytochrome-based fluorescent probes. The spectral and kinetic behavior of P_r-II strongly resemble those of the photochemically compromised and highly fluorescent Y176H variant of Cph1Δ, suggesting both cases use similar mechanisms for reduced photochemical quantum yield.^{35, 49} It was originally proposed that the Y176H substitution resulted in increased steric hindrance to photoisomerization of the PCB D-ring,⁴⁹ but this hypothesis predates the availability of phytochrome crystal structures. It now seems plausible that this substitution disrupts the hydrogen bonding network of the P_r-I population (Fig. 7A). Hydrogen bonding effects are similarly thought to be critical in modulating the excited-state dynamics of the SaBphP1 bacteriophytochrome from Stigmatella aurantiaca, in which excited-state proton transfer is thought to be responsible for lowering both the fluorescence and photochemical quantum yields for the biliverdin chromophores of this photoreceptor family.^{50, 51}

To conclude, in Cph1Δ, heterogeneity strongly affects kinetics, absorption, fluorescence, and quantum yields (both photochemical and fluorescent), with evidence implicating structural rearrangements and possibly even effects on in vivo function. Excitation-wavelength dependent and pH-dependent transient absorption measurements were used to reconcile spectral heterogeneity observed in static absorption and resonance Raman measurements with that observed in temperature-dependent and excitation-wavelength-dependent absorption and transient spectra in Cph1Δ. We ascribe this heterogeneity to heterogeneous protonation of either the bilin chromophore or residues within its immediate environment. We observe three spectrally distinct populations, each having multiple sub-populations distinguished by differing excited-state quenching kinetics. We used a single global model to describe how excitation wavelength, temperature, and pH affect the photocycle initiation dynamics of Cph1Δ. This model also successfully simulated reported 2D electronic spectra,³⁸ demonstrating that existing approaches for interpreting heterogeneity in such spectra can be misleading. Given the conservation of His260 in most phytochromes, it seems likely that similar effects are widespread in this photoreceptor family.