Photoactive Yellow Protein Chromophore Photoisomerizes around a Single Bond if the Double Bond Is Locked

Photoactivation in the Photoactive Yellow Protein, a bacterial blue-light photoreceptor, proceeds via photoisomerization of the double C=C bond in the covalently attached chromophore. Quantum chemistry calculations, however, have suggested that in addition to double-bond photoisomerization, the isolated chromophore and many of its analogues can isomerize around a single C–C bond as well. Whereas double-bond photoisomerization has been observed with X-ray crystallography, experimental evidence of single-bond photoisomerization is currently lacking. Therefore, we have synthesized a chromophore analogue, in which the formal double bond is covalently locked in a cyclopentenone ring, and carried out transient absorption spectroscopy experiments in combination with nonadiabatic molecular dynamics simulations to reveal that the locked chromophore isomerizes around the single bond upon photoactivation. Our work thus provides experimental evidence of single-bond photoisomerization in a photoactive yellow protein chromophore analogue and suggests that photoisomerization is not restricted to the double bonds in conjugated systems. This insight may be useful for designing light-driven molecular switches or motors.

* sı Supporting Information ABSTRACT: Photoactivation in the Photoactive Yellow Protein, a bacterial blue-light photoreceptor, proceeds via photoisomerization of the double CC bond in the covalently attached chromophore. Quantum chemistry calculations, however, have suggested that in addition to double-bond photoisomerization, the isolated chromophore and many of its analogues can isomerize around a single C−C bond as well. Whereas double-bond photoisomerization has been observed with X-ray crystallography, experimental evidence of single-bond photoisomerization is currently lacking. Therefore, we have synthesized a chromophore analogue, in which the formal double bond is covalently locked in a cyclopentenone ring, and carried out transient absorption spectroscopy experiments in combination with nonadiabatic molecular dynamics simulations to reveal that the locked chromophore isomerizes around the single bond upon photoactivation. Our work thus provides experimental evidence of single-bond photoisomerization in a photoactive yellow protein chromophore analogue and suggests that photoisomerization is not restricted to the double bonds in conjugated systems. This insight may be useful for designing light-driven molecular switches or motors. P hotoactive yellow protein (PYP), a 125-residue watersoluble protein, is believed to act as the primary photoreceptor in the negative phototaxis response of the Halorhodospira halophila bacterium to blue light. 1 The photocycle of PYP is initiated by an ultrafast photoisomerization of the conjugated p-coumaric acid chromophore that is covalently bound to the protein via a thioester linkage. Timeresolved crystallography 2,3 and molecular dynamics simulations 4 suggest that the photoisomerization occurs around the double bond ( Figure 1).
Chromophore analogues also undergo ultrafast deactivation in solution, but for many analogues, no photoproduct is detected. 5−10 This paradox was addressed with nonadiabatic molecular dynamics simulations of the pCK analogue in water. 11,12 These simulations suggested that photoisomerization can also occur around the single bond adjacent to the phenolate ring (SB, Figure 1) without forming a photoproduct. Despite support from high-level ab initio quantum chemistry calculations, 13−17 direct experimental evidence of single-bond isomerization is, however, still lacking. Experimentally demonstrating the existence of single-bond photoisomerization in the chromophore of PYP is important not only to validate theoretical methods but also to better understand the basic principles and versatility of photoisomerization in general.
Previously, experiments on locked analogues of important biological chromophores, in which isomerization around a selected bond is prevented by means of a covalent bridge, have provided valuable insights into the isomerization process in rhodopsins, 18 phytochromes, 19 and PYP. 20,21 It was found that when PYP is reconstituted with chromophore analogues, with either the single bond adjacent to the phenolate ring (SB, Figure 1) 21 or both single and double bonds locked, 20 the protein is still photoactive, albeit with a lower quantum yield. In contrast, gas-phase photoelectron spectroscopy experiments on a pCA analogue in which the double bond is locked suggest that photoisomerization is suppressed in favor of electron photodetachment. 22 To experimentally test whether photoisomerization around the single bond provides an alternative decay channel for the photoexcited chromophore, as suggested by computations, 11−16 we have synthesized a chromophore analog in which the double bond is locked within a rigid cyclopentenone ring ( Figure 1, DB-locked, compound 3). Details of the synthesis and structural characterization of 3 can be found in the Supporting Information (SI). Although this chromophore lacks the thioester moiety that is present in the protein, previous experiments and computations on thio-pCA and pCK have shown that the photochemistries of these chromophores are very similar. 5,13 Therefore, results obtained for the doublebond-locked (DB-locked) chromophore should be transferable to the native chromophore of PYP.
In Figure 2, we compare the steady-state absorption and emission spectra of the DB-locked chromophore (3) and of pCK in aqueous solution at pH >10. Because the pK a values of pCK and the DB-locked chromophore are around 8 and 9, respectively (SI), both chromophores are fully deprotonated in our measurements. The similarity between the absorption spectra suggests that the covalent bridge that locks the double bond has no significant effect on the electronic structure of the chromophore in the ground (S 0 ) or excited (S 1 ) state. However, comparing the fluorescence spectra reveals that the DB-locked chromophore has a smaller Stokes shift than pCK. We attribute this difference to the more rigid structure of the DB-locked chromophore (SI).
Time-resolved transient absorption experiments were performed for both the DB-locked chromophore and pCK with a ∼120 fs pump pulse centered at 395 nm. The chirpcorrected spectra are shown in Figure 2b,c and Figure 2e,f. The transient absorption spectrum of pCK has been previously reported by Changenet-Barret and coworkers. 10 Our results are in good agreement with theirs, despite a lower laser power in our experiments (SI). We chose our excitation energies to be low enough to prevent the formation of free electrons due to a two-photon photoionization process that can occur at higher laser intensities. 10 Figure 1. In PYP, the para-coumaric acid (pCA) chromophore is deprotonated and covalently attached to the protein via a thioester bond with a cysteine side chain (thio-pCA, left structure). The pCK analogue (middle) has a methyl group instead of a thioester but exhibits very similar optical properties compared with thio-pCA. 5 In this work, the double-bond (DB)-locked analogue (3, right) was studied. The cyclic structure prevents isomerization around the double bond, and only single-bond (SB) isomerization is possible. The spectra presented in Figure 2 show similar features for both chromophores. The most prominent feature is the negative signal at 420−480 nm for the DB-locked chromophore and at 440−520 nm for pCK. On the basis of the steady-state fluorescence spectra, we attribute this signal to stimulated emission. The negative bands around 400 nm are assigned to the ground-state bleach (GSB) signal, centered at ∼380 nm. However, in our measurements, the GSB was largely masked by the scattering from the pump pulse. We also observe a distinct positive feature at 410 nm for the DB-locked chromophore and at 430 nm for pCK, which grows initially but then decays, leaving no long-lived absorption signal. This transient signal is more prominent in pCK than in the DBlocked chromophore. We attribute this band to the absorption of the vibrationally hot ground state that is formed immediately after excited-state decay via a conical intersection, as in previous work. 10 A full global analysis of the transient absorption data is prevented by overlapping spectral features and their dynamic shifts, which would lead to wavelength-dependent decay kinetics. We therefore did not perform such analysis. Instead, we can immediately conclude from the raw data presented in Figure 2 that the excited-state decay is ultrafast in both chromophores: All transient signals disappear within ∼10 ps, leaving no long-lived signals associated with photoproducts. The ultrafast relaxation of the DB-locked analogue is consistent with the single-bond photoisomerization pathway predicted for pCK. 11,12 Because we also do not observe any spectral features associated with hydrated electrons, we can furthermore rule out that in water, electron photodetachment occurs, in contrast with the situation in the gas phase. 22−24 This difference was already rationalized by Gromov and coworkers, who showed that hydrogen bonding between the chromophore and water molecules brings the energy of the S 1 state significantly below that of the ionized state. 13 We therefore conclude, in line with results from computations, 11−16 that the chromophore decays via a rotation around the single bond adjacent to the phenolate ring, followed by internal conversion through a conical intersection.
To confirm that the ultrafast decay observed in our experiments is indeed due to single-bond photoisomerization, we also performed 100 nonadiabatic molecular dynamics simulations of the deprotonated DB-locked chromophore in water. (See the SI for details.) In all but one of the simulations, the chromophore rapidly twists by 90°around the single bond after photoexcitation. Figure 3 shows the time evolution of the single-bond torsion angle in one of the QM/MM trajectories. As in pCK, 12 shortly after reaching the 90°single-bond twisted conformation in the excited state, the system decays to the ground state via a conical intersection. The existence of this conical intersection in the presence of water was confirmed at a higher level of theory (SI). After the transition, which was modeled as a diabatic surface hop, 25 the chromophore either relaxes back to the initial conformation or continues the singlebond isomerization process (as in Figure 3). However, because the phenolate ring is symmetric, there is no distinction between this photoproduct and the initial conformation, and thus the isomerization quantum yield is zero. In one trajectory, a higher energy conical intersection is accessed at which the phenolate ring is puckered (SI). However, because this decay pathway was observed in only 1% of the trajectories, we consider it a minor decay channel and hence insignificant. Although the number of trajectories is too small to be statistically relevant, an exponential fit to the decay times yields an excited-state lifetime of 1.3 ps (SI), in qualitative agreement with our experiments (Figure 2c).
In summary, the results of our experiments and simulations strongly suggest that single-bond photoisomerization is possible in the chromophore of Photoactive Yellow Protein. Although the single-bond isomerization pathway was predicted more than a decade ago, 4,11,13,15,25 it is now also supported by the ultrashort excited-state lifetime of the DB-locked chromophore in our experiments. With the lock on the double bond, the only remaining candidate for the ultrafast decay process is single-bond isomerization. The lack of photoproduct for this deactivation channel is further supported by the total recovery of the original spectrum within picoseconds ( Figure  2c). Whereas it remains to be seen whether the single-bond isomerization channel is also active inside PYP, our results confirm that in addition to double-bond isomerization, singlebond isomerization can also be considered when designing light-driven molecular switches or motors. 26 ■ ASSOCIATED CONTENT