Effect of the PHY Domain on the Photoisomerization Step of the Forward Pr→Pfr Conversion of a Knotless Phytochrome

Abstract Phytochrome photoreceptors operate via photoisomerization of a bound bilin chromophore. Their typical architecture consists of GAF, PAS and PHY domains. Knotless phytochromes lack the PAS domain, while retaining photoconversion abilities, with some being able to photoconvert with just the GAF domain. Therefore, we investigated the ultrafast photoisomerization of the Pr state of a knotless phytochrome to reveal the effect of the PHY domain and its “tongue” region on the transduction of the light signal. We show that the PHY domain does not affect the initial conformational dynamics of the chromophore. However, it significantly accelerates the consecutively induced reorganizational dynamics of the protein, necessary for the progression of the photoisomerization. Consequently, the PHY domain keeps the bilin and its binding pocket in a more reactive conformation, which decreases the extent of protein reorganization required for the chromophore isomerization. Thereby, less energy is lost along nonproductive reaction pathways, resulting in increased efficiency.


Results and Discussion
Photochromism of All2699 g1 g2 The P r form of g1g2 has an absorption maximum at 637 nm, which is similart ot hat in g1, [8,33] while the P fr form is significantly red shifted (by 73 nm) to 710 nm (689 nm in g1; [8,33] Figure 2A). Thus, the observed spectral shift of P fr appears to be inducedb yi nteractions of the PCB chromophore with the "tongue" region of the g2 domain in g1g2. [32,34] Interestingly, we find that the quantum yield (QY) of the P r !P fr transition is increased from~8% to~13 %( as imilarQ Yi so bserved in the related Cph2 [35] ). Hence, it follows that while the presenceo f the g2 domain does not directly affect the spectralp roperties of the P r form it does affect its photochemistry.
The circular dichroism (CD) spectra of both proteins (g1g2 vs. g1) have similars hape and undergo as ign inversion of the Q-band CD signal upon P r $P fr switching ( Figure 2B). [8,33] The Q-band CD signals exhibit opposite signs fort he different states, negative for P r and positive for P fr .T he P fr signalo fg 1g2 is further red shifted and shows an increased extinction coefficient.
The signs of these signals indicate the orientation of the peripheral rings Aa nd Dw ith respect to the co-plane of rings B and C. [8,36] This overall orientation appearsu naffected by the presence of the "tongue", whichi si nl ine with the slight changes in the dihedrala ngle of rings Aa nd Do bserved by NMR. [32] In comparison, other phytochromes like Cph1 [37] and Cph2 [35] exhibit as imilar sign change of the Q-band, while most CBCRs and bacteriophytochromes show no sign change upon switching. [36,[38][39][40] Ultrafast dynamics of P r * and formation of Lumi-R The role of the protein environment on the ultrafast photoisomerization dynamics of the PCB chromophorew as investigated by femtosecond transient absorption (TA) measurements on g1g2 as comparedw ith the single-domain g1. [8] The TA data of g1g2 show three main features ( Figure 3A): i) ab road positive signal below 575 nm which can be assigned to excited state absorption (ESA), ii)a negative signal above 575 nm due to ground state bleach (GSB) and stimulated emission (SE), and iii)a positive photoproduct absorption (PA) appearinga tl ater times at 670 nm associated with the formation of the primary Figure 1. Schematic representation of the structural homology model of the P r state of the All2699g1g2 construct. [32] The structureo fthe GAF1 domain is based on the crystal structure of the soleGAF1 module (PDB ID 6OZA), while the GAF2 domain was modelledb ased on the crystal structure of the structurally similar Cph2 [32] (PDB ID 4BWI). The PCB chromophore (orange) embedded in the GAF1 domain(green)isi nc lose interaction with at onguelike protrusion (pink) from the GAF2 domain (yellow),w hich also shieldsthe PCB from the solvent.

Figure 2.
A) Stationary absorptionspectraoft he P r and P fr forms of g1g2 (solid lines) and g1 [8] (dotted lines). The pure P fr spectrum was obtained by conservative subtraction of 38 %o fthe pure P r spectrum from the PSS spectrum, followed by multiplication with af actor of 1.61toy ield the spectrum for the fully converted system. The extinction coefficient of P r is 79 000 m À1 cm À1 (similart og 1 [33] ), while the extinctionc oefficient of P fr is 61000 m À1 cm À1 .B)CDs pectrao fthe P r andP fr states of g1g2( solid lines) and g1 [8] (dottedl ines). The pure P fr CD spectrum was derived from the PSS CD as described in A). The lifetime distribution analysiso ft he experimental data (see Supporting Information and Ref. [41] form ethodology) gives furtheri nsight into the early ES dynamics of P r (Figure 3B). The positive-( > 690 nm) andt he negative-amplitude (< 690 nm) distributions with al ifetime of 100 fs can be assigned to ar ed shift of the SE and therefore to the departure of the ES wavepacketf rom the Franck-Condon (FC) region. The lifetimed istributions between 1psa nd 10 ps are located at the overlap of the steep edges of the GSB and SE, making this region very sensitive to slight spectralchanges.
Because there is no substantial decay of the ES on this timescale, we assign these distributions to dynamics on the ES potential energy surface. Based on the spectral position of the ESA, GSB and SE signals, the broad lifetime distributions (stretching from 30 ps to 1ns) with positive (420-575nm) and negative( 600-740nm) amplitudes can be attributed to the simultaneousd ecay of these signals, and thus to the decay of P r * . Compared with the positive-amplitude distribution representing the decay of the ESA signal, the negative-amplitude distribution, especially at 675 nm, appearss tretched in lifetime. This can be explained by an overlaid additional negative-amplitude distribution describing the rise of the primary photoproduct Lumi-R commencingw ith the ES decay.O nt he scale longer than 1ns, the negative and positive-amplitude lifetime distributions correspond to the non-decaying GSB and Lumi-R signals.
Distributed character of the P r * photoisomerization kinetics Previously,w es howed that the P r * decay kinetics in the singledomain g1 is described by broad and structureless lifetime distributions (see FigureS2B and discussion in [8] )and can be modelled well using as tretched exponential function, [42,43] thereby avoidingi ntroductiono fu nnecessary kinetic components. The lifetimed istributions describing the P r * decay of g1g2 (from 30 ps to 1nsi nF igure 3B)a re similarly broad and structureless, thus we followed our previous approachand analyzed the TA data using af our-state model in whicho ne of the states is modelled by as tretched exponent (Figure 4). This model yields an excellent fit of the data without additional kinetic components (comparet he case of the five-state model ( Figure S5)). Stretched exponentials are used to model distributed kinetics occurring in constrained environments [44][45][46] and here underline the importance of the protein in the isomerization kinetics of the PCB chromophore.
The evolutionary associated difference spectra (EADS) of the first three states ( Figure 4B)c learly show that at early times no significant change in the amplitude of the ESA occurs, reaffirming the conclusiont hat the sub-20 ps dynamics of P r * is not associated with ES decay.T herefore, similarly to the single GAF domainsA ll2699g1 [8] and Slr1393g3, [47] the EADS of S2 and S3 of g1g2 show only am inor spectral shift in the GSB andS E overlap area, indicating that the~3psc omponent ( Figure 4) is due to ES dynamics of the chromophore. The P r * relaxation and the associated PCB photoisomerization occur on the 100 ps timescale from state S3 after overcoming ab arrier on the ES potentiale nergy surface. [8] This state is modeled as a stretched exponent, showing that the P r * decay of g1g2 follows adistributed type kinetics.
Recently,i tw as reported that in g1 and in g1g2 there exists ab road distribution of ground state subconformations that rapidly interconvert in solution. [32,48] These subconformations could partially contribute to the observed distributed character of the P r * decay kinetics. However,t heir rapid interconversion denotes that they are separated by low energetic barriers, and thus cannot explain the large exciteds tate barriert hat determines the relatively slow P r * decay kinetics (100 ps timescale). [8] Such ab arrierc an be overcome only via dynamic reorganization of the system, which in turn provides the dominant contributiont ot he distributed kinetics of P r * decay.T herefore, our  results point to am ore dynamic pictureo ft he kinetics in knotless phytochromes. In contrast, distinct ES decay components, including such on the sub-50pst imescale, have been reported for other phytochromes (e.g.,C ph1, PhyA) andC BCRs,a nd were discussed in the framework of static ground state heterogeneityoft he P r form. [15,18,20,38,49,50] Comparison of the ultrafast dynamics of g1 g2 and g1:the effect of the g2 (PHY) domain The comparison of the ultrafast dynamics of the g1g2 construct with the dynamics of the sole GAF domain g1 [8] provides ad irect assessment of the effect of the g2 (PHY) domain.S trikingly,t he early dynamics of both proteins are identical and even the coherent oscillations observed in the SE region up tõ 2ps( Figure 5B)m atch in frequency and phase. The similarity in the g1g2 and g1 early kineticsi si na greement with the lack of an immediate effect of the PHY domain on the steadys tate properties of the P r form (Figure 2), and thus further supports the conclusion that the primary dynamics are indeed due to conformationalc hanges in the PCB chromophore. Only at later times,t he P r * kinetics of g1g2 deviates from that of g1. Figure 5c learly shows that in the g1g2 construct the ES decay and the formation of the primary photoproduct Lumi-R are accelerated relative to g1. This effect is also illustratedb yt he correspondingL DMs ( Figure S2). Considering the time range up to 10 ps, the lifetimed istribution structure remains the same forb oth proteins,w hilet he later lifetimed istributions (from 30 ps to 1ns) describing the ES decay are shiftedt os horter lifetimes in the case of g1g2 (for comparison, the center lifetimeo ft he correspondingg 1d istribution is indicated by aw hite dashed line in the LDM of g1g2, Figure 3B). Therefore, the direct comparison of the P r * kinetics of g1g2 and g1 reveals the impact of the g2 (PHY) domain and categorically demonstrates the critical role of the protein environment on the photoisomerization stepo ft he PCB chromophore.

Mechanistic model
Based on the analysis presented above,weproposethe following molecular picturef or the photoisomerization dynamics of g1g2 ( Figure 6). After excitation,t he PCB chromophore leaves the FC region (~100 fs)a nd undergoes ES conformationald ynamics on the earlyp st imescale (< 20 ps). This dynamics acts as at rigger for larger scale motions in the protein environment, which alleviatesr estrictions hindering furthere volution on the ES (illustrated by the barriero nt he ES potential energy surface). Interestingly,s imilar conclusions were derived in recent studies on related bacteriophytochromes [51,52] and ac yanobacteriochrome. [53] As the protein reorganizes, the barrier on the ES decreases which allowsP r * relaxation to proceed. This model explainst he distributed character of the P r * decay kinetics as it is imposed by the conformational dynamics of the protein. Eventually,P r * decays (~130 ps) yielding the primary photoproduct Lumi-R.
Previously,w ed emonstrated that the ring Dr otation of the PCB chromophorei ng 1i sh indered by an earby Tyrr esidue (Tyr142). [8] In the g1g2 (GAF-PHY) construct, the interaction of Figure 5. Comparison of the transientabsorption decays of g1g2 (orange) and g1 [8] (cyan) at selectedw avelengths within A) the ESA (523 nm), and B) the GSB/SE (655 nm) spectral regions. The transient decays were measureda fter 635 nm excitation of the P r form. Figure 6. Main reaction coordinates determining the photoconversion kinetics in the P r form of g1 and g1g2. For the larger construct, the interactionof the "tongue" regionw ith the PCB chromophore results in al ower barrier (red, solid vs. orange, dashed lines). The early (< 10 ps) dynamicso ft he PCB chromophore triggers reorganizations in the protein binding pocket, which lower the energetic barrier at latertimescales(DE 1 vs. DE 2 )and unlock the photoisomerization.
Chem. Eur.J.2020, 26,[17261][17262][17263][17264][17265][17266] www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH the "tongue" region of the PHY domain with the chromophore-binding pocket in the GAF domain limits the conformational space of rings Aa nd Do ft he PCB and drives the Tyr142 residue away from the chromophorea sc ompared with g1. [32] Therefore, it appearsa si ft he "tongue" region keeps the chromophorea nd the binding pocket in am ore reactive conformation. This decreases the extent of protein conformationr eorganization required for facilitating the isomerization of the PCB chromophore and results in an acceleratedP r * decay kinetics ( Figure 5) andamore efficient P r !P fr photoconversion (less energy being lost on nonproductive degrees of freedom).

Conclusions
Our work provides direct evidence for the essential role of the protein environment in the control of the photoisomerization kinetics of the PCBchromophore and outlines adetailed mechanistic pictureo ft he P r * photoisomerization dynamics in knotless phytochromes. From an evolutionary perspective, the PHY domain "tongue" represents ad evelopment in phytochromes that tunes the photoreception efficiency.T his is ak ey design principle for the development of optimized photoreceptorsf or biotechnological applications.