same order of magnitude and thus consistent witha measured external PL quantum yield of 3.6% forRbCsMAFA (and 2.4% for CsMAFA). Following theapproach in (21–25) [see also fig. S14 and (18)],we used the EQEEL and the emission spectrum topredict a Voc value of 1240 mV, confirming in-dependently the valuemeasured from the J-V curve.Furthermore, for higher driving currents, the

EQEEL in Fig. 3C reaches 3.8%, making the solarcell one of the most efficient perovskite LEDs aswell, emitting in the near-infrared/red spectralrange (Fig. 3C, inset) (26–28). Movie S1 shows aRbCsMAFAdevicemounted in our custom-madedevice holder. As we tuned toward maximumemission and back, we observed bright EL indaylight. For comparison, for commercially avail-able Si solar cells, EQEEL ≈ 0.5% (20). Thesevalues for our PSC devices indicate that all majorsources of nonradiative recombinationwere strong-ly suppressed and that the material has very lowbulk and surface defect density. We also inves-tigated transport behavior bymeans of intensity-modulated photocurrent spectroscopy (IMPS); thefindings suggest that the charge transport withinthe RbCsMAFA perovskite layer is substantiallyfaster than in CsMAFA, which is already muchmore defect-free thanMAFA (19) [see also fig. S15and (18)].Despite the high efficiencies and an outstand-

ing EL, this Rb-containing perovskite materialmust be able to achieve high stability. This task isnot trivial given the hygroscopic nature of perov-skite films, phase instabilities, and light sensitiv-ity (29). Interestingly, the Achilles’ heel of PSCs isnot necessarily the perovskite itself, but rather thecommonly used spiro-OMeTAD hole transportermaterial that becomes permeable (at elevatedtemperature) to metal electrode diffusion intothe perovskite, causing irreversible degradation(30, 31). This effect can be mitigated with bufferlayers or by avoiding the use of metal electrodes(30–32). Alternatively, for the combined heat-lightstress tests in this work, we found a thin layer ofpolytriarylamine polymer (PTAA) (see SEM imagein fig. S4B) to work equally well (33). We imposedthe above protocols simultaneously and ageddevices for 500 hours at 85°C under continuousillumination with full intensity and maximumpower point (MPP) tracking in a nitrogen atmos-phere. This compounded stress test exceeds in-dustrial standards (34). We show the result in Fig.3D (red curve). The device started with >17% effi-ciency at room temperature before the aging pro-tocol was applied (see fig. S16 for non-normalizedvalues of PCE, FF, Jsc, Voc, JMPP, and VMPP). Duringthe 85°C step (in which Voc is inevitably lowered),the device retained 95% of its initial performance.

REFERENCES AND NOTES

1. National Renewable Energy Laboratory, Best Research-CellEfficiencies chart; www.nrel.gov/ncpv/images/efficiency_chart.jpg.

2. N. J. Jeon et al., Nature 517, 476–480 (2015).3. M. Saliba et al., Nat. Energy 1, 15017 (2016).4. X. Li et al., Science 353, 58–62 (2016).5. H. Choi et al., Nano Energy 7, 80–85 (2014).6. J. W. Lee et al., Adv. Energy Mater. 5, 1501310 (2015).7. C. Yi et al., Energy Environ. Sci. 9, 656–662 (2016).8. Z. Li et al., Chem. Mater. 28, 284–292 (2016).

9. M. Saliba et al., Energy Environ. Sci. 9, 1989–1997 (2016).10. D. P. McMeekin et al., Science 351, 151–155 (2016).11. H. L. Wells, Z. Anorg. Chem. 3, 195–210 (1893).12. D. Weber, Z. Naturforsch. B 33, 1443 (1978).13. D. B. Mitzi, K. Liang, J. Solid State Chem. 134, 376–381 (1997).14. G. Kieslich, S. J. Sun, A. K. Cheetham, Chem. Sci. 5, 4712–4715

(2014).15. M. R. Filip, G. E. Eperon, H. J. Snaith, F. Giustino, Nat. Commun.

5, 5757 (2014).16. F. Hao, C. C. Stoumpos, D. H. Cao, R. P. H. Chang,

M. G. Kanatzidis, Nat. Photonics 8, 489–494 (2014).17. D. M. Trots, S. V. Myagkota, J. Phys. Chem. Solids 69,

2520–2526 (2008).18. See supplementary materials on Science Online.19. J. P. Correa-Baena et al., Adv. Mater. 28, 5031–5037 (2016).20. M. A. Green, Prog. Photovolt. Res. Appl. 20, 472–476 (2012).21. K. Tvingstedt et al., Sci. Rep. 4, 6071 (2014).22. D. Bi et al., Sci. Advances 2, e1501170 (2016).23. U. Rau, Phys. Rev. B 76, 085303 (2007).24. R. T. Ross, J. Chem. Phys. 46, 4590 (1967).25. W. Tress et al., Adv. Energy Mater. 5, 1400812 (2015).26. H. Cho et al., Science 350, 1222–1225 (2015).27. L. Gil-Escrig et al., Chem. Commun. 51, 569–571 (2015).28. G. Li et al., Adv. Mater. 28, 3528–3534 (2016).29. N. H. Tiep, Z. L. Ku, H. J. Fan, Adv. Energy Mater. 6, 1501420

(2016).30. K. Domanski et al., ACS Nano 10, 6306–6314 (2016).31. K. A. Bush et al., Adv. Mater. 28, 3937–3943 (2016).32. A. Mei et al., Science 345, 295–298 (2014).33. J. H. Heo et al., Nat. Photonics 7, 487 (2013).34. Y. G. Rong, L. F. Liu, A. Y. Mei, X. Li, H. W. Han, Adv. Energy

Mater. 5, 1501066 (2015).

ACKNOWLEDGMENTS

M.S. conceived, designed, and led the overall project; M.S., J.-Y.S.,A.U., and J.-P.C.-B. conducted SEM, PL, and XRD experiments onthe perovskite films; M.S. and W.R.T. performed EL and PLquantum yield experiments; A.U. conducted confocal laser

scanning fluorescence microscopy for PL mapping; M.S., K.D., andW.R.T. conducted long-term aging tests on the devices;M.S., T.M., J.-P.C.-B., and A.A. prepared and characterized PVdevices; A.H. participated in the supervision of the work; M.G.directed and supervised the research; M.S. wrote the first draft ofthe paper; and all authors contributed to the discussion and writingof the paper. Supported by the co-funded Marie Skłodowska Curiefellowship, H2020 grant agreement no. 665667 (M.S.); theEuropean Union’s Seventh Framework Programme for research,technological development, and demonstration under grantagreement no. 291771 (A.A.); the Swiss National ScienceFoundation, funding from the framework of Umbrella project (grantagreement nos. 407040-153952, 407040-153990, and 200021-157135/1); the NRP 70 “Energy Turnaround”; the 9th call proposal906: CONNECT PV; and SNF-NanoTera and the Swiss FederalOffice of Energy (SYNERGY). A.A. conducted IMPS experiments atthe Adolphe Merkle Institute, Fribourg, Switzerland. M.G. andS.M.Z. thank the King Abdulaziz City for Science and Technologyfor financial support under a joint research project. All data areavailable in the main paper and supplement. M.S., T.M., K.D.,J.-Y.S., S.M.Z., W.R.T., and M.G. are inventors on European PatentApplication 1618056.7 submitted by École Polytechnique Fédéralede Lausanne and Panasonic Corporation that covers the perovskitecompounds in this work.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/354/6309/206/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S16Tables S1 and S2Movie S1References (35–41)

18 July 2016; accepted 8 September 2016Published online 29 September 201610.1126/science.aah5557

BIOPHYSICS

Bifurcating electron-transferpathways in DNA photolyasesdetermine the repair quantum yieldMeng Zhang,1 Lijuan Wang,1 Shi Shu,1 Aziz Sancar,2 Dongping Zhong1*

Photolyase is a blue-light–activated enzyme that repairs ultraviolet-induced DNA damagethat occurs in the form of cyclobutane pyrimidine dimers (CPDs) and pyrimidine-pyrimidone(6-4) photoproducts. Previous studies on microbial photolyases have revealed an electron-tunneling pathway that is critical for the repair mechanism. In this study, we used femtosecondspectroscopy to deconvolute seven electron-transfer reactions in 10 elementary steps in allclasses of CPD photolyases.We report a unified electron-transfer pathway through a conservedstructural configuration that bifurcates to favor direct tunneling in prokaryotes and a two-stephopping mechanism in eukaryotes. Both bifurcation routes are operative, but their relativecontributions, dictated by the reduction potentials of the flavin cofactor and the substrate,determine the overall quantum yield of repair.

Photolyases, which belong to the photolyase(PL)–cryptochrome (CRY) superfamily, usea fully reduced flavin (FADH−) cofactorto repair sunlight-induced DNA lesions,including cyclobutane pyrimidine dimers

(CPDs) and pyrimidine-pyrimidone (6-4) photo-products (1–5). On the basis of sequence analyses,CPD photolyases are highly diversified and canbe subdivided into three classes (I to III) (6–8),

as well as single-stranded DNA (ssDNA)–specificPLs (9) (Fig. 1A). Thus, the molecular repair

SCIENCE sciencemag.org 14 OCTOBER 2016 • VOL 354 ISSUE 6309 209

1Department of Physics, Department of Chemistry andBiochemistry, and Programs of Biophysics, Chemical Physics,and Biochemistry, The Ohio State University, Columbus,OH 43210, USA. 2Department of Biochemistry and Biophysics,University of North Carolina School of Medicine, Chapel Hill,NC 27599, USA.*Corresponding author. Email: zhong.28@osu.edu

RESEARCH | REPORTSon A

pril 19, 2020

http://science.sciencemag.org/

Dow

nloaded from

210 14 OCTOBER 2016 • VOL 354 ISSUE 6309 sciencemag.org SCIENCE

Fig. 2. Photoinduced intra-molecular electron transfer inphotolyases and initial electron-transfer bifurcation in repaircomplexes. (A) Absorption tran-sients (DA) of AnPL, DmPL, andAtPL enzymes probed at 800 nmfor the excited-state flavin (LfH‾*).lpr, probe wavelength. (B) Absorp-tion transients of AnPL, DmPL,and AtPL enzymes probed at270 nm,mainly for the intermediate-state flavin radical (LfH•).(C) Absorption transients of thethree enzymes (dashed line)and enzyme-substrate complexes(solid lines) probed at 800 nm.In the presence of substrate, theexcited-state dynamics of AnPLbecomes drastically faster due todirect electron tunneling to theCPD substrate, whereas in class IIphotolyases the change of theexcited-state dynamics is muchsmaller and is almost negligible inAtPL. (D) Repair scheme withseven electron-transfer reactionsand two dimer-splitting processesamong 10 elementary steps. Acyclic electron transfer between thetricyclic ring (lumiflavin, Lf) and

Fig. 1. Classes of photolyases, sequence alignment, steady-state repairQYs, and local structures. (A) Unrooted phylogenetic tree of the PL–CRYprotein family and representative members.The class II PL is distant from theother subfamilies. (B) Sequence alignment of eight photolyases of differentclasses. Ec, Escherichia coli; An, Anacystis nidulans; Cc, Caulobacter crescentus;At, Arabidopsis thaliana; Mm, Methanosarcina mazei; Dm, Drosophila melano-gaster; Os, Oryza sativa. Three conserved active-site residues (R, E, and M orQ) near the substrate are highlighted in yellow. Three other critical active-siteresidues that vary between the class I PLs and the other subfamilies are high-

lighted in blue. Single-letter abbreviations for the amino acid residues are asfollows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu;M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.(C) Enzyme activities and steady-state repair QYs of EcPL (as a control), AnPL,DmPL, AtPL, CcPL, and AtCRY3 (ssDNA AtPL). Every data point was averagedover five measurements, with error bars showing 1 SD of uncertainty. (D) Localactive-site structures of class I AnPL (silver) and class II MmPL (light blue),highlighting the six critical residues and the catalytic flavin cofactor. In class IIPL, a glycine residue (G375 in MmPL) replaces asparagine (N349 in AnPL).

adenine (Ade) is intrinsic in all photolyases (gray arrows in the molecular structure). Because the free energy of charge separation is close to zero, the excitedand charge-separated states also interconvert. When the substrate is present, the electron path can bifurcate, either by direct tunneling to CPD through theintervening adenine via a superexchange mechanism or by a two-step hopping mechanism also bridged by the adenine (blue arrows). 1, LfH‾-Ade+T<>T(CPD);2, LfH‾*-Ade+T<>T; 3, LfH•-Ade‾+T<>T; 4, LfH•-Ade+T<>T•‾; 5, LfH•-Ade+T+T•‾; 6, LfH‾-Ade+T+T. h, Planck's constant; n, frequency; SP, C–C bond splitting;ER, electron return.

RESEARCH | REPORTSon A

pril 19, 2020

http://science.sciencemag.org/

Dow

nloaded from

mechanism recently obtained from the well-characterized class I photolyase fromEscherichiacoli (EcPL) (9, 10) may not be applicable to othersubfamilies, especially the distant class II PLs.We used ultrafast spectroscopy to systematicallyinvestigate class IAnPL (fromAnacystis nidulans),class IIDmPL (fromDrosophilamelanogaster) andAtPL (from Arabidopsis thaliana), and class IIICcPL (from Caulobacter crescentus) and ssDNAAtPL (AtCRY3) to search for conserved featuresof the overall photolyase repair mechanism.Class II PLs have Gly and Tyr or Phe residues

at two positions near the dimer substrate in the

active site that are occupied, respectively, by Asnand Arg residues in the other PLs, as well as oneAsn residue near the N5 position of the flavincofactor that is contributed by a different helix(Fig. 1, B and D). Otherwise, the overall active-site configuration is conserved across all PLs,especially the folded structure of the flavin co-factor with the adenine moiety in the middlebetween the tricyclic ring and the dimer sub-strate (Fig. 1D) (11–15). The different residues inthe flavin and substrate binding pockets alterthe reduction potentials of the dimer and thecofactor. In particular, the substitution of Gly in

place of Asn creates a space filled in by severalwatermolecules (Fig. 1D) (11, 12). Such reduction-potential changes could lead to distinct reactiondynamics that would account for the wide distri-bution of total repair quantum yields (QYs) ob-served across the different classes of PLs (Fig. 1C).Here we present the key results for the two distantclass I and class II PLs and then summarize ourfindings, as well as similar observations for otherPLs that are also listed in the supplementarymaterials.To analyze the CPD repair dynamics, we first

measured the absorption transients of the PLs

SCIENCE sciencemag.org 14 OCTOBER 2016 • VOL 354 ISSUE 6309 211

Fig. 3. Absorption transients of enzyme-substrate complexes probed widelyfrom the visible to the UV region todetect various intermediates andfinal products involved in repair. (A andB) Absorption transients of AnPL, DmPL,and AtPL complexes probed at 510 nm(A) and the deconvolution of LfH‾* (darkgreen dashed line) and LfH• (dark reddashed line) contributions from DmPL (B).(C to H) Absorption transients of thethree complexes probed in the UV regionmainly for detection of the thymineintermediates and final products. Thesedynamics are systematically fitted by thetotal flavin-related species (LfH‾*, LfH•,LfH‾, Ade, and Ade‾, dashed orange line),the thymine dimer intermediate T-T ‾

(dashed magenta line), the thymine anionT ‾ (dashed light blue line), and thethymine product T (dashed dark yellowline), as shown in three deconvolutionplots [(D), (F), and (H)].

RESEARCH | REPORTSon A

pril 19, 2020

http://science.sciencemag.org/

Dow

nloaded from

without substrate probed at 800 nm (Fig. 2A)and 270 nm (Fig. 2B) after excitation at 400 nm(Fig. 2D). Previous studies (16) revealed dynamicsattributed to an intramolecular electron transfer(ET) between the tricyclic ring (LfH−*) and theadenine moiety (Ade) in the conserved flavinfolded structure [forward electron transfer 1 (FET1)(Fig. 2D)] and two subsequent back electrontransfers (BETs) to the initial excited state (BET1a)and the original ground state (BET1b). The800-nmtransients in Fig. 2A reflect the pure excited LfH−*

dynamics, whereas the 270-nm transients in Fig.2B mainly capture the intermediate LfH• semi-quinone dynamics (16), which vary across thethree PLs. Together with other wavelength detec-tion (fig. S2), we systematically analyzed thesestretched dynamics that arise from the couplingwith the active-site relaxation (9, 17). We thenderived the initial charge separation (FET1) forAnPL, DmPL, and AtPL in average times of 1340,1590, and 564 ps, respectively; the reverse chargerecombination (BET1a) in 1230, 1029, and 1505 ps,respectively; and thebackelectron transfer (BET1b)to the ground state in 26, 715, and 17 ps, respec-tively, with a nearly constant deactivation life-time of 6 ns (see supplementary materials andtable S1). Both AnPL and AtPL promptly returnto the ground state via ultrafast BET1b, whereasDmPL exhibits an apparent long decay (Fig. 2A)because the forward electron transfer for DmPL(FET1 = 1590 ps) is similar to that of AnPL (FET1 =1340 ps). Thus, the slowDmPL transient in Fig. 2Ais caused by substantial reverse charge recom-bination BET1a to regenerate LfH−* because ofmuch slower BET1b (715 ps). These dynamics arecritical to determining the repair pathways and,therefore, theQYs of different classes of photolyases.

We next measured 800-nm absorption tran-sients of the three enzyme-substrate complexesafter 400-nm excitation (Fig. 2C; the dashed linesare reproduced from Fig. 2A for comparison).Notably, the dynamics in the three PL-substratecomplexes of the three photolyases are consider-ably different. AnPL, similar to the class I EcPL(18–20), displays a much faster decay in the pres-ence of the substrate, indicating that the elec-tron flows to the substrate efficiently: FET2 =209 ps (Fig. 2D), which is substantially shorterthan the bifurcated FET1 pathway (1340 ps).Thus, the repair mechanism is similar to thatobserved for the same class EcPL. The electronprimarily tunnels directly to the substrate witha favorable, negative free energy, mediated bythe intervening adenine in the fold structuralmotif (18–21). For DmPL, the overall dynamicsin the complex becomes slightly faster if weconsider only the forward FET1 process (1590 ps),resulting in an ET time of 5315 ps (FET2), whichis much longer than the 1590 ps of FET1 for theelectron transfer to the adeninemoiety. Thus, theelectron path bifurcates (Fig. 2D): Some electrons(19%) tunnel to the substrate through the ade-nine, whereas others (64%) hop to the adenine toform the anionAde− intermediate (the remaining17% goes to the ground state via deactivation; seetable S2). Surprisingly, for AtPL the excited flavindynamics are nearly the same as observed with-out the substrate with a FET1 time of 564 ps.Best-fitting of the transient indicates a tunnelingET directly to the substrate in 6500 ps (FET2),which indicates that the electron dominantly hopsto the adenine, a situation completely opposite ofthat in class I AnPL and EcPL. Thus, the structuraland chemical properties in class II PLs favor elec-

tron hopping to, rather than electron tunnelingthrough, the intervening adenine. From x-raystructures (11–15), the edge-to-edge distance be-tween the adenine and substrate CPD (Fig. 2D)is about 3 Å and, therefore, the further electronhopping (FET3 in Fig. 2D) from the anion ade-nine to CPD is ultrafast (see below), given thenegative free energy of favorable reduction po-tentials (table S3).After discovering the initial electron-transfer

pathways, we probed the repair dynamics compre-hensively from the visible to the deep ultraviolet(UV) region to completely resolve the reactionintermediates as we did in studies of EcPL (19)(Fig. 3, A to H). These dynamics show distinctprofiles. With systematic analyses, we can dissecteach transient into distinct components, frominitial reactants to various intermediates andfinal products. The repair follows the reactionscheme in Fig. 2D: After the electron from LfH−*

finally arrives in the CPD, the first C–C bondbreakage is ultrafast in less than 10 ps (19, 22),and the second C–C bond cleavage (SP) occurs intens of picoseconds (19, 23). Finally, the electronreturns to the flavin semiquinone (LfH•) to closethe entire photocycle. Specifically, in the visibleregion at 510 nm (Fig. 3A), we detected only thecofactor flavin signal with two components ofthe excited (LfH−*) and intermediate (LfH•) states(Fig. 3B). Note that the LfH• signal is a summationof contributions of three components (see equationS12 and fig. S3 in the supplementary materials). Inthe UV region (Fig. 3, C to H), the transientscontain all of the flavin signals, including those ofthe excited state of LfH−*, the immediate statesof LfH• and Ade−, and the final states of LfH−

and Ade, as well as the overall substrate signals

212 14 OCTOBER 2016 • VOL 354 ISSUE 6309 sciencemag.org SCIENCE

Fig. 4. Complete photocyclesof CPD repair by class I andclass II PLs and QYevolutionwith bifurcated ETchannels indifferent photolyases.Repair cyclesof (A) class I AnPL and (B) class IIAtPL with seven ETreactions among10 resolved elementary steps. InAnPL, FET2 is much faster than FET1,so the direct electron-tunnelingchannel is dominant. In AtPL, FET1is much faster than FET2, andhence the two-step hopping pathwaybecomes dominant. (C) Changesin bifurcated ET rates for FET2and FET1 and the resulting finalQYs of the two respective paths,QY2 and QY1, ordered frommicrobial to eukaryotic PLs.

RESEARCH | REPORTSon A

pril 19, 2020

http://science.sciencemag.org/

Dow

nloaded from

of the intermediate T-T− after the first C–C bondbreakage, T− after the second C–C cleavage, andthe final product of repaired base T. The variousdissections are shown in Fig. 3, D, F, and H, andfig. S4. Thus, we obtained the ultrafast electronhopping of FET3 in 6, 11, and 15 ps and theelectron return after repair in 437, 2890, and819 ps for AnPL, DmPL, and AtPL, respectively.Knowing the total QYs (Fig. 1C), we can alsoderive the second C–C cleavage in 87, 48, and 36ps and the futile back electron transfer BET2 in1138, 149, and 527 ps, respectively, for three PLs(table S1).To recapitulate, we have identified 10 elemen-

tary steps in the repair reaction by DNA photo-lyase, including 7 ET steps, and measured theirtime scales in real time (table S1). Consequently,we can calculate the QY of each step that con-tributes to the total QY (table S2). In Fig. 4, A andB, we show the two resolved photocycles for classI AnPL and class II AtPL, respectively, with thecorresponding reaction times of each step. Forclass I PL (Fig. 4A), the two systems we studied,AnPL and EcPL, show a dominant tunnelingpathway with the highest QYs (table S2). Forclass II PL (Fig. 4B), the two systems studiedhere, DmPL and AtPL, adopt mainly a two-stephopping route, also with good repair efficiency.For other PLs [class III CcPL and ssDNA-specificAtPL (AtCRY3)], both tunneling and hoppingchannels are operative (table S1). These detaileddynamics and time scales for seven ET reactionsinvolved in repair can be used to derive micro-scopic pictures of various reorganization ener-gies; their relevant reduction potentials; and,thus, reaction driving forces (table S3) (21, 24, 25).We did not observe clear evidence for the possibleflickering resonance for the initial electron bi-furcation, as proposed recently in a theoreticalstudy (26).Figure 4C shows the repair QYs along the

evolutionary path from the microbial class I tothe eukaryotic class II PLs, with initial electronbifurcation into the tunneling route FET2 andthe hopping path FET1 and their resulting QYs(QY2 and QY1). Clearly, the tunneling route inclass I leads to a higher repair QY. With thedecrease in the rates of tunneling, the hoppingchannel comes to dominate in class II PLs. Con-sequently, class II PLs can never reach the class Irepair QY because the electron path at Ade−

also bifurcates into the repair channel to theCPD and the futile path back to the originalground state, both of which share similar hop-ping rates. The conserved active-site configura-tion and the folded flavin structure that occuras a result of evolution in the entire photolyase-cryptochrome superfamily (11–15, 27–30) are essen-tial to ensure a unified electron-transfer mechanismthrough electron path bifurcation into two oper-ative routes for all CPD photolyases.

REFERENCES AND NOTES

1. A. Sancar, Chem. Rev. 103, 2203–2238 (2003).2. C. Lin, T. Todo, Genome Biol. 6, 220 (2005).3. S. T. Kim, A. Sancar, C. Essenmacher, G. T. Babcock, Proc.

Natl. Acad. Sci. U.S.A. 90, 8023–8027 (1993).4. D. E. Brash, Trends Genet. 13, 410–414 (1997).

5. J. S. Taylor, Acc. Chem. Res. 27, 76–82 (1994).6. S. Kanai, R. Kikuno, H. Toh, H. Ryo, T. Todo, J. Mol. Evol. 45,

535–548 (1997).7. A. Sancar et al., Cold Spring Harb. Symp. Quant. Biol. 65,

157–172 (2000).8. J. I. Lucas-Lledó, M. Lynch, Mol. Biol. Evol. 26, 1143–1153

(2009).9. D. Zhong, Annu. Rev. Phys. Chem. 66, 691–715 (2015).10. Z. Liu, L. Wang, D. Zhong, Phys. Chem. Chem. Phys. 17,

11933–11949 (2015).11. S. Kiontke et al., EMBO J. 30, 4437–4449 (2011).12. K. Hitomi et al., J. Biol. Chem. 287, 12060–12069 (2012).13. H. W. Park, S. T. Kim, A. Sancar, J. Deisenhofer, Science 268,

1866–1872 (1995).14. A. Mees et al., Science 306, 1789–1793 (2004).15. Y. Huang et al., Proc. Natl. Acad. Sci. U.S.A. 103, 17701–17706

(2006).16. Z. Liu et al., Proc. Natl. Acad. Sci. U.S.A. 110, 12972–12977

(2013).17. C.-W. Chang et al., Proc. Natl. Acad. Sci. U.S.A. 107, 2914–2919

(2010).18. Y. T. Kao, C. Saxena, L. Wang, A. Sancar, D. Zhong, Proc. Natl.

Acad. Sci. U.S.A. 102, 16128–16132 (2005).19. Z. Liu et al., Proc. Natl. Acad. Sci. U.S.A. 108, 14831–14836

(2011).20. C. Tan et al., Nat. Commun. 6, 7302 (2015).21. Z. Liu et al., J. Am. Chem. Soc. 134, 8104–8114 (2012).22. A. A. Hassanali, D. Zhong, S. J. Singer, J. Phys. Chem. B 115,

3848–3859 (2011).23. V. Thiagarajan, M. Byrdin, A. P. Eker, P. Müller, K. Brettel, Proc.

Natl. Acad. Sci. U.S.A. 108, 9402–9407 (2011).24. H. Wang et al., Science 316, 747–750 (2007).

25. D. N. LeBard, V. Kapko, D. V. Matyushov, J. Phys. Chem. B 112,10322–10342 (2008).

26. Y. Zhang, C. Liu, A. Balaeff, S. S. Skourtis, D. N. Beratan, Proc.Natl. Acad. Sci. U.S.A. 111, 10049–10054 (2014).

27. M. J. Maul et al., Angew. Chem. Int. Ed. 47, 10076–10080 (2008).28. W. Xing et al., Nature 496, 64–68 (2013).29. B. D. Zoltowski et al., Nature 480, 396–399 (2011).30. C. A. Brautigam et al., Proc. Natl. Acad. Sci. U.S.A. 101,

12142–12147 (2004).

ACKNOWLEDGMENTS

We dedicate this paper to the memory of the “father offemtochemistry,” Ahmed H. Zewail, who passed away on 2 August2016. D.Z. was trained as a student and later as a postdoctoralfellow in Dr. Zewail’s lab. We thank Y.-T. Kao, Z. Liu, X. Guo, C. Tan,Y. Qin, and N. Ozturk for help in the initial stages of this work.We also thank P. Houston for careful reading of the manuscript.This work was supported, in part, by NIH grants GM074813and GM118332 to D.Z. and GM031082 to A.S. Additional datasupporting the conclusions of this study are included in thesupplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/354/6309/209/suppl/DC1Materials and MethodsFigs. S1 to S4Tables S1 to S4References (31–36)

20 July 2016; accepted 15 September 201610.1126/science.aah6071

INFLUENZA

Role for migratory wild birdsin the global spread of avianinfluenza H5N8The Global Consortium for H5N8 and Related Influenza Viruses*†

Avian influenza viruses affect both poultry production and public health. A subtypeH5N8 (clade2.3.4.4) virus, following an outbreak in poultry in South Korea in January 2014, rapidly spreadworldwide in 2014–2015. Our analysis of H5N8 viral sequences, epidemiological investigations,waterfowl migration, and poultry trade showed that long-distance migratory birds can playa major role in the global spread of avian influenza viruses. Further, we found that thehemagglutinin of clade 2.3.4.4 virus was remarkably promiscuous, creating reassortantswith multiple neuraminidase subtypes. Improving our understanding of the circumpolarcirculation of avian influenza viruses in migratory waterfowl will help to provide early warningof threats from avian influenza to poultry, and potentially human, health.

In 2014, highly pathogenic avian influenza(HPAI) virus of the subtype H5N8 causeddisease outbreaks in poultry in Asia, Europe,and North America (1–3). Avian influenzaviruses are a threat both to global poultry

production and to public health; they have thepotential to cause severe disease in people andto adapt to transmit efficiently in human pop-ulations (4). This was the first time since 2005that a single subtype of HPAI virus had spreadover such a large geographical area and the firsttime that a Eurasian HPAI virus had spread to

North America. The rapid global spread of HPAIH5N8 virus outbreaks raised the question of theroutes by which the virus had been transmitted.The segment encoding for the hemagglutinin

(HA) surface protein of theHPAIH5N8 viruses isa descendant of the HPAI H5N1 virus (A/Goose/Guangdong/1/1996), first detected in China in1996 (5). Since then, HPAI H5N1 viruses havebecome endemic in poultry populations in sev-eral countries. The H5 viruses have developednew characteristics by mutation and by reassort-ment with other avian influenza (AI) viruses, bothin poultry and inwild birds. In 2005–2006, HPAIH5N1 spread from Asia to Europe, the MiddleEast, andAfrica during the course of a fewmonths.Although virus spread traditionally had been

SCIENCE sciencemag.org 14 OCTOBER 2016 • VOL 354 ISSUE 6309 213

*Corresponding author. Email: t.kuiken@erasmusmc.nl †Allauthors with their affiliations appear at the end of this paper.

RESEARCH | REPORTSon A

pril 19, 2020

http://science.sciencemag.org/

Dow

nloaded from

yieldBifurcating electron-transfer pathways in DNA photolyases determine the repair quantum

Meng Zhang, Lijuan Wang, Shi Shu, Aziz Sancar and Dongping Zhong

DOI: 10.1126/science.aah6071 (6309), 209-213.354Science 

, this issue p. 209Sciencerepair quantum yield seen in prokaryotes.the prokaryotic enzymes and a two-step hopping mechanism in the eukaryotic variety. This difference explains the higherto study the dynamics of this step. A bifurcation in the electron transfer pathway favors a direct tunneling mechanism in

applied ultrafast absorption spectroscopyet al.blue light absorption by a cofactor drives an electron transfer step. Zhang Photolyase enzymes repair DNA that has been damaged by ultraviolet sunlight. The repair process begins when

Two roads diverged in a yellow photolyase

ARTICLE TOOLS http://science.sciencemag.org/content/354/6309/209

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2016/10/14/354.6309.209.DC1

REFERENCES

http://science.sciencemag.org/content/354/6309/209#BIBLThis article cites 36 articles, 18 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Copyright © 2016, American Association for the Advancement of Science

on April 19, 2020

http://science.sciencem

ag.org/D

ownloaded from