Stepwise Hydride Transfer in a Biological System: Insights into the Reaction Mechanism of the Light‐Dependent Protochlorophyllide Oxidoreductase

Abstract Hydride transfer plays a crucial role in a wide range of biological systems. However, its mode of action (concerted or stepwise) is still under debate. Light‐dependent NADPH: protochlorophyllide oxidoreductase (POR) catalyzes the stereospecific trans addition of a hydride anion and a proton across the C17−C18 double bond of protochlorophyllide. Time‐resolved absorption and emission spectroscopy were used to investigate the hydride transfer mechanism in POR. Apart from excited states of protochlorophyllide, three discrete intermediates were resolved, consistent with a stepwise mechanism that involves an initial electron transfer from NADPH. A subsequent proton‐coupled electron transfer followed by a proton transfer yield distinct different intermediates for wild type and the C226S variant, that is, initial hydride attaches to either C17 or C18, but ends in the same chlorophyllide stereoisomer. This work provides the first evidence of a stepwise hydride transfer in a biological system.

Hydride transfers (H À T) play acrucial role in awide range of biological and chemical systems.O ver 400 enzymecatalyzed H À Treactions depend on the cofactor nicotinamide adenine dinucleotide (phosphate), NAD(P)H, which acts as as ource of two electrons and one proton (equivalent to ahydride ion). [1] Therefore,reaction with asubstrate results in the oxidized form, NAD(P) + ,and the reduced substrate. [2] In theory,H À Tc onsists of three elementary steps as shown in Scheme 1. [1,3] Depending on the rate constants involving the necessary electron transfer (eT) and proton transfer (PT) reactions,the H À Tcan be entirely stepwise,partially stepwise, or purely concerted. Theformation of radical pair intermediates can only be detected if the rate of the first eT is significantly different to the rate of the second eT,w hich creates ac losed-shell system again. However,o wing to the elusive nature of short-lived radical pair intermediates,direct evidence of the stepwise reaction mechanism is sparse.Sofar, direct detection of radicals during H À Th as only been reported using NADH analogues as hydride donors and quinone derivatives or non-heme oxoiron(IV) complexes as hydride acceptors in thermally driven reactions. [4][5][6] Owing to the limitations imposed by the necessity to use rapid mixing strategies to initiate catalysis in thermally activated enzymes, no such mechanistic insights could be obtained yet.
Light-activated enzymes,w here the reaction chemistry can be triggered by ashort laser pulse,allow the study of H À T with am uch better time resolution. Thel ight-dependent NADPH:protochlorophyllide oxidoreductase (POR) catalyzes the reduction of the C 17 À C 18 double bond in ring Do f protochlorophyllide (PChlide,P )t op roduce chlorophyllide (Chlide,C). [7] Theenzyme requires the coenzyme NADPH [8] while the substrate PChlide acts as ap hotoreceptor ( Figure 1). [9] Thed ouble bond reduction was shown to proceed in as equential manner [10,11] on a mst imescale, where first H À Tf rom the pro-S face of NADPH to the C 17 position of PChlide [12,13] occurs followed by aPTtoC 18 ,most likely from ac onserved Ty rr esidue within the POR active site. [14] Recently,a na lternative mechanism was proposed for the active site mutant POR-C226S, [15] in which after an initial eT ac oncerted hydrogen atom and PT transfer complete the double bond reduction. [15] Although this suggests that stepwise H À Tm echanisms may be possible,t here is no direct evidence for this mechanism and the PChlideC À and counter NADPHC + radicals have yet to be identified.
As PChlide acts as ap hotoreceptor,t he POR-catalyzed reaction also depends on the excited state dynamics of the substrate itself.Based on photophysical studies on (un)bound PChlide,different (branched [16][17][18][19][20][21] and sequential [22,23] )models were proposed all starting from the excited singlet state,S 1 . Both models include the formation of an intramolecular charge transfer state (S ICT )t hat is suggested to induce an electron-deficient site at the C 17 À C 18 double bond of PChlide and facilitate its selective reduction. However,t he proposed dynamics of the S ICT differ and need further clarification (a detailed discussion on the individual models can be found in the Supporting Information). As link between the excited state dynamics of PChlide and POR chemistry,i tw as suggested that the S ICT branches into an on-catalytic and ac atalytic route. [24] In this model, while the former decays back into the ground state via formation of the solvated S ICT and triplet, the latter leads to the formation of the reactive S ICT followed by H À Tand PT.However,whether the reactive S ICT is directly linked to POR photochemistry or whether short-lived intermediates,f or example,r adical pairs,p recede the H À Treaction remains unclear.
We re-investigated by transient absorption (TA) and emission the photophysics and photochemistry of PChlide unbound or bound to POR. We use the following nomenclature: S P n = PChlide, S I (x) n = intermediates arising from (photo)chemistry between excited PChlide and NADPH, and S C n = Chlide,w here S is the corresponding spin multiplicity, n the electronic state number, and x the number of an individual intermediate.
As adetailed understanding of the PChlide photophysics is essential for the interpretation of the photochemistry catalyzed by POR, we begin with abrief overview of the lightinduced dynamics obtained for the unbound PChlide and when bound in the non-productive pseudoternary complex, PChlide/POR/NADP + .A ne xtended discussion and corresponding data can be found in the Supporting Information. Excitation of PChlide,e ither unbound or bound to POR in PChlide/POR/NADP + (ACN/Buffer/non-productive complex), results in dynamics that follow standard photophysics with an excited singlet, 1 P 1 ,and at riplet state, 3 P 1 ,where the 3 P 1 is partially formed from the 1 P 1 and both states decay back into the ground state, 1 P 0 (green and orange frames in Figure 3A). The 1 P 1 dynamics are accompanied by vibrational relaxation and solvent reorganization dynamics with typical lifetimes for those processes (16 ps/6 ps and 53 ps/4.2 ps).
When excited into higher excited singlet states (for example, 450 nm), 1 P 4 ,t he 1 P 4 ! 1 P 1 transition is additionally observed with af sl ifetime (112 fs/142 fs/116 fs). The 1 P 1 decays with anslifetime (4.3 ns/4 ns/3.9 ns) into 3 P 1 and 1 P 0 with atriplet quantum yield of 50 %/55 %/55 %. The 3 P 1 decay shows an expected O 2 dependency with total decay rates, k T ,o f (250 ns) À1 /(4.8 ms) À1 /(19 ms) À1 under aerobic and (170 ms) À1 /(1 ms) À1 /(73 ms) À1 under anaerobic conditions. Thus,t he intrinsic back intersystem crossing rate, k bisc ,g iven by the reciprocal lifetimes obtained from the data under anaerobic conditions,t ogether with the concentration of O 2 (c = 2.4 mmol L À1 in ACN; [25] c = 0.25 mmol L À1 in H 2 O [26] ) gives aO 2 -quenching rate of k q O2 = (k T Àk bisc )/[O 2 ] = 1.7 10 9 Lmol À1 s À1 /8.3 10 8 Lmol À1 s À1 /10 10 8 Lmol À1 s À1 .I nterestingly,i nt he non-productive complex, the O 2 -quenching rate is five times smaller compared to k q O2 found for unbound PChlide in buffer. This demonstrates aprotective role of POR against reactions with O 2 within the lifetime of the 3 P 1 ,t hus avoiding the formation of reactive oxygen species as long as the individual intermediates within the chlorophyll biosynthesis cannot be found unbound in vivo and are rather transferred from enzyme to enzyme.T he characterization of the standard photophysics (green and orange frames in Figure 3A)a llows the determination of characteristic and distinct different species associated spectra (SAS) and their corresponding dynamics for 1 P 1 and 3 P 1 ( Figure 3B,C). These are the basis for the study on the productive ternary complexes with NADPH instead of NADP + .
We then investigated the productive ternary complex, PChlide/POR/NADPH (Supporting Information, Figure S10 A). In the first 50 ps after excitation, the relaxation dynamics are comparable to those observed in the nonproductive complex (Supporting Information, Figure S11 A, B1,2), indicating no significant fast change in the electronic state populations.H owever,o nl onger timescales the picture changes dramatically.O nasub-ns timescale,t he 1 P 1 decays faster compared to the non-productive complex, and instead of the 3 P 1 spectrum, different TA features arise indicating the formation of an ew electronic species with positive TA contributions at ca. 400, 477, and between 650 to 700 nm ( Figure 2A). Compared to the non-productive complex, the 1 P 1 is quenched from 3.9 ns to 650 ps resulting in aq uantum yield of 83 %( F( 2 I 0 (1) ) = 1Àt/t 0 ,w here t 0 and t are the 1 P 1 lifetimes of the (non-)productive ternary complexes) for the first intermediate.O nasub-100 ns timescale,as econd intermediate with am ore intense absorption at ca. 477 nm and an additional small but distinct absorption band at ca. 760 nm ( Figure 2C)isobserved. To note,onthis timescale we do not observe the formation of 3 P 1 ,and thus k isc is negligible in this case.F urther,athird intermediate is observed with ac haracteristic absorption band at ca. 700 nm, which finally decays into the product Chlide, 1 C 0 ,w ith its characteristic absorption band at ca. 680 nm ( Figure 2C), as reported. [10] TheT Ad ata can be globally fitted with five exponentials yielding decay associated difference spectra (DADS) with well-separated lifetimes, t i ,of650 ps,86ns, 926 ns,129 ms, and 1 s( Figure 2E). Since the reduction of the C 17 ÀC 18 double bond requires two electrons and two protons,t he observed lifetimes are consistent with individual single steps including two eTs and two PTs.PChlide,even in its excited singlet state, lacks the characteristics of abase so that we can exclude aPT as the initial reaction step (for amore detailed discussion, see the Supporting Information). Thus,w ei nterpret the first intermediate as aradical anion ( 2 I 0 (1) )after eT from NADPH to 1 P 1 .C onsequently,P Tist he most probable step to follow, neutralizing the ionic radical pair. However,w eo bserve two similar DADS with lifetimes separated by af actor of ca. 10, which are prolonged when the deuterated cofactor NADPD is used (Supporting Information, Figure S12 A,B), indicating that in the formation of the second and third intermediate aH/D atom transfer from NADPH/D is involved. Thus,steps 2and 3are consistent with aproton coupled eT forming 2 I 0 (2) and 1 I 0 ,r espectively. 1 I 0 (3) represents as ingle addition of hydrogen to the PChlide anion, which is finally protonated from either surrounding H 2 Omolecules or acidic amino acid residues inside the protein pocket, yielding the final product 1 C 0 .T he recombination in the mechanistic sequence of the individual intermediates into their ground-state species is always allowed. Figure 3A (red frame) summarizes the proposed reaction mechanism starting from the excited PChlide singlet state.T his model was applied onto the TA data resulting in reasonable SAS for each intermediate ( Figure 3D)w ith corresponding yields of F( 2 I 0 (1) ) = 0.83, F( 2 I 0 (2) ) = 0.5, F( 1 I 0 ) = 1.0, and F( 1 C 0 ) = 1.0. Thus,t he total quantum yield of 1 C 0 formation is F total ( 1 C 0 ) = F( 2 I 0 (1) ) F( 2 I 0 (2) ) F( 1 I 0 (3) ) F( 1 C 0 ) = F( 2 I 0 (1) ) = 0.43, which is in good agreement with published data. [28] However,wewere not able to resolve the counter NADP(H) cation and neutral radicals, although they are expected to absorb in the same spectral regions (l max = 370 and 550 nm and l max = 400 and 500 nm, respectively). Compared to PChlide,t hey have ca. 10-fold smaller extinction coefficients ranging between 0.5 and 5 10 3 Lmol À1 cm À1 , [29] and, thus,are masked by the corresponding PChlide intermediates.Similar species were also reported for light-independent POR. [30] However,adirect comparison cannot be made owing to different environmental effects,that is,C oulombic interactions to an additional Fe-S cluster (for amore detailed discussion, see the Supporting Information).
Finally,w ei nvestigated the ternary complex, PChlide/ POR-C226S/NADPH, which was previously proposed to proceed via an alternative reaction mechanism compared to wild type (WT). [15] Here, 1 P 1 is quenched from 4.4 ns to 760 ps, resulting in aq uantum yield of 83 %( see argument for WT data above) for the first eT forming 2 I 0 (1) which is identical to the situation in WT.A gain, no triplet is observed following the 1 P 1 decay,i ndicating an egligible k isc .T he following kinetics are slightly altered compared to WT ( Figure 2B,D,F). Especially,t he final 1 C 0 formation is ca. 4-fold faster in the mutant. Because 1) steps 2a nd 3b ecome slower in the presence of NADPD;a nd 2) the last step becomes slower in the presence of D 2 O, the data for the mutant suggest similar chemical intermediates to those for WT.However, inspection of the involved spectral changes (Figure 2) reveals distinct differences for the second and third intermediate.C orrespondingly,a pplication of the same model as for WT (red frame in Figure 3A)r esolves distinct different SAS for the 2nd ( 2 I 0 (2') )a nd 3rd ( 1 I 0 (3') )i ntermediate,while the other SAS agree well within experimental error to the ones obtained for WT ( Figure 3E). Here,t he total quantum yield of 1 C 0 formation is F total ( 1 C 0 ) = 0.44. Heavy atom experiments showed that identical chemical reaction partners are involved. Thus,w ec onclude that the formed PChlide species in POR-C226S must differ compared to those formed in WT,f or example,byattachment of the hydride at C 18 rather than C 17 . Quantum-chemical calculations of all possible stereoisomers on aq ualitative basis indeed show that the absorption spectrum of H-C 18 (= 1 I 0 (3') )i se xpected to have ad istinct broad absorption band at ca. 550 nm, which is more intense than for H-C 17 (= 1 I 0 (3) ). Further, the expected band at ca. 700 nm for 1 I 0 (3) resolves pretty well but is less intense and redshifted for 1 I 0 (3') (Supporting Information, Figure S17). Finally, circular dichroism spectra of 1 C 0 either formed by WT or by POR-C226S are identical, indicating the same stereoisomer for 1 C 0 (Supporting Information, Figure S18 D). Therefore, we conclude that PChlide is bound turned by 1808 8 in the mutant compared to WT,allowing the formation of identical products ( Figure 3A;Supporting Information, Figure S18 A-C).
Here,w eh ave investigated the mechanism of the lightdependent POR, which reduces PChlide to Chlide.A part from the photophysical states of (un)bound PChlide,t hat is, excited singlet and triplet, we were able to resolve three intermediates consistent with as tepwise hydride transfer The data in the gray dashed rectangles contained invalid data owing to laser scatter and were replaced by the fit during the global fitting. [27] E,F) Decay associated differencespectra from global fits on data in (C) and (D). k r = radiative decay; k ic = internalconversion; k isc = intersystem crossing; k bisc = back intersystem crossing; 1 k eT (1) = 1st electron transfer; k beT = back electron transfer; k PT (1) = 1st proton transfer; k bHT = back Hatom transfer; k eT (2) = 2nd electron transfer; k bH À T = back hydride transfer; k PT (2) = 2nd proton transfer. mechanism, that is,interpreted as an initial eT from NADPH to the excited PChlide singlet state followed by ap roton coupled eT and asubsequent PT.I dentical transfer reactions were observed for the C226S variant, but with distinct different PChlide species in terms of hydride attachment at C 18 instead of C 17 ,w hich is potentially due to an altered PChlide binding.T oour knowledge,this is the first biological system that allowed the direct observation of as tepwise hydride transfer.Our study provides general understanding of how light energy can be harnessed to drive H-transfer chemistry,w ith implications for the design of light-activated chemical and biochemical catalysts.A dditionally,t he study also emphasizes how mutagenesis can fundamentally alter the reaction path of enzyme catalyzed H-transfer. Thus,our work provides important and new insight into fundamental mechanisms of H-transfer in abiological system and challenges the inference that biological hydride transfers proceed exclusively through concerted mechanisms.H owever,f or thermally activated systems the H À -transfer mechanism remains elusive owing to the lack of time resolution.