TR-SFX Photolyase : Maestre-Reyna et al. vs Christou et al. (2023)

CPD Photolyase was the flagship model for Zhong’s pioneering femtobiology studies in the 2000s. This DNA repair mechanism is remarkable for its ultra-fast, light-mediated process. Previously, I briefly introduced CPD photolyase (CPDphr). 

6 Jan 2026 edits : clarified pump laser activation mechanisms. 

Here I will discuss the most recent visualization studies via TR SFX. These were two complementary papers by independent teams published in the same issue of Science in November/December 2023. Maestre Reyna et al., 2023 covered the entire catalytic process from 100ps to 200 µs, while Christou et al., 2023 focused on early events from 3ps to 100ps, though also briefly covered the entire catalytic cycle. I will compare these 2 studies briefly, starting with Maestre-Reyna et al., 2023. 

Visualization studies of CPDphr can be divided into 3 eras.  

Static Era 1990s to 2000s: X-ray crystallography

Park et al. 1995 (Science) Crystal structure of DNA photolyase from E. coli

  • First Atomic Structure
  • Resolved structure of Class I CPDphr (E. coli)
  • Established 2 chromophore architecture (FAD and MTHF) in some classes of CPDphr
  • Lacked CPD (cyclobutane pyrimidine dimer): binding mechanisms unresolved
  • PDB 1DNP 

Mees et al. 2004 (Science) Crystal structure of a photolyase-DNA complex

  • Visualised Base Flipping
  • Class I CPDphr (A. nidulans) + synthetic CPD analog
  • First visualization of CPD lesion flipped 170 degrees out of DNA backbone into active site in photolyase
  • Photolyase inserts methionine wedge into DNA helix to stabilize precarious state 
  • PDB 1TEZ

Kinetic Era 2000s to 2010s: Femtosecond Spectroscopy

Pioneered by Zhong Lab (Ohio State) – shifted from static spatial models to femtosecond dynamic models. No PDB structures produced since spectroscopy is an indirect observation. Spectroscopy measures how much light a molecule absorbs or emits at different wavelengths, expressed as bulk optical properties of real-time processes. Spatial structures at specific timepoints are then inferred from optical signals. 

Kao et al. 2005 (Science) Direct observation of thymine dimer repair...

  • Visualizing Bond Breaking
  • Zhong's team inferred stepwise bond breaking of C5-C5 (lower energy barrier), followed by C6-C6 in the CPD
  • FS-spectroscopy captured transient radical intermediates that previous X-ray crystallography could not resolve 

Liu et al. 2011 (PNAS) Dynamics and mechanism of CPD repair...

  • Mapping the Circuit
  • Spectroscopy + site-directed mutagenesis
  • Visualized how protein environment tuned FAD chromphore
  • Active site as electrostatic network optimizing redox potential to minimize back-electron transfer
  • Explained high quantum yield of CPDphr

Zhang, Wang, & Zhong 2017 (Arch. Biochem. Biophys.) Photolyase: Dynamics and electron transfer...

  • Electron path to quantum yield
  • Electrons tunnel through adenine moiety in FAD via superexchange
  • Conserving probability amplitude during quantum tunneling
  • Accelerates tunneling rate from ns to ps, increasing quantum yield 

Dynamic Era 2023 onwards – Time-Resolved Serial Femtosecond Crystallography TR SFX

Maestre-Reyna et al., 2023: PDB 7YD8–7YDG + dark/light references

Christou et al., 2023: PDB 8FQ0–8FQ6

Aspect

Maestre-Reyna et al. (2023)

Christou et al. (2023)

Focus

100ps to 200 µs (entire catalytic cycle)

3ps to 100ps (early events)

Full catalytic cycle also captured

Primary XFEL Facility

SwissFEL – short pulse suitable for capturing early events/repair phase

Pump laser 400 nm (0.98 ps pulse, 10 µJ) at SwissFEL

Secondary Facility

SACLA (RIKEN, Japan) – captures longer timescales in the recovery phase

408 nm (3 ns pulse, 150 µJ) at SACLA

—

Crystals

  • Anaerobic encapsulation to maintain protein in photoreduced state
  • Archaea MmCPD
  • Cis-syn CPD (most common CPD enantiomer)

Maestre-Reyna et al 2023 (TLDR)

  • Direct visual confirmations of Zhong’s previous spectroscopic inferences about sequential C-C bond breaking
  • 2 sets of TR SFX studies: 1, repair process ps-ns 10⁻¹² to 10⁻⁹); 2, recovery process ns- µs (10⁻⁹ to 10⁻⁶)
  • Visualised recovery process for the first time (spectrally silent in Fs spectroscopy)
  • 5-water cluster (5WC) and arginine residues in active site play dynamic roles in stabilising reaction

Since the recovery process is spectrally silent on previous fs-spectroscopy techniques, this study aimed to capture the entire catalytic process in precise temporal and spatial resolution. The catalytic phase occurs in pico to nanoseconds, while the recovery phase occurs in nano to microseconds. Hence two sets of TR-SFX experiments were performed, using a different facility specific to each timescale. The primary facility, SwissFEL, uses a 0.98ps pulse. This covered 100ps up to 10ns. The second facility, SACLA, uses a 3ns pulse. This covered 10ns up to 200 μs.

Crystal preparation and injection

Cocrystals of MmCPDII and CPD-containing DNA were grown in one large batch within 12 hours before usage at XFEL sites. The enzyme was preactivated via photoreduction, then placed in safety-light conditions (640nm, red), before mixing with CPD-containing DNA. Anaerobic tents were used to prevent oxidation (inactivation) of the activated enzymes. In-solution spectroscopy demonstrated that MmCPDII remained in the fully reduced state in anaerobic, safety-light conditions for at least 24 hours after photoreduction. No crystals older than 20 hours were used, to ensure that crystal samples were in a fully reduced state. This ensures that the cocrystal is primed for photoreactivation, but can only be activated by the pump laser pulse. 

6 Jan 2026 edits: The batch was then processed into cube-shaped microcrystals of 50×50×50 μm³, each microcrystal containing many molecules of CPD-photolyase complex. These microcrystals were then loaded into the injector. In total, 1.27 million microcystals were flowed through an injector. 

Normally, photoreactivation of the CPD photolyase repair process occurs in one step, where UV spectrum from sunlight induces CPD formation, while the blue light/near-UV spectrum simultaneously activates the FAD cofactor for the electron transfer process. However, to synchronise repair activity over many copies of the photolyase-CPD complex in one microcrystal, artificial bifurcation of this process is necessary. Hence, all FAD within the microcrystal was 'primed' beforehand into a photoreduced state before injection, and then immediately inhibited from interacting with CPD under anaerobic, safe light conditions. The pump laser of 400nm initiates time zero for each microcrystal. This additional laser excitation may create nonphysiological effects, which the authors discuss as potential confounding factors. 

Pump laser and then XFEL were aimed at one spot in the interaction region, capturing whichever microcrystal was present at that time. The pump laser was aimed at a larger area, covering multiple microcrystals at once, while the XFEL probe only hits one specific microcrystal. Each microcrystal can only be XFEL-diffracted once, as XFEL immediately destroyed the microcrystal. The continuous flow ensured that a different, pumped microcrystal was diffracted each time. Once the pumped microcrystals have flowed through the confocal area; the pump laser is activated again for fresh, unpumped microcrystals that arrive, before XFEL diffraction. To capture various timeframes, the interval is varied between the initial pump laser and the XFEL for each microcrystal. 

Pump laser-induced light contamination of unintended areas (nearby microcrystals) is a risk. This would activate nearby microcrystals unintentionally, creating inaccuracies in the timeframe of snapshots. This is controlled by precise aiming of the pump, titrating the power of the pump, and obtaining control data via XFEL from interleaved dark microcrystals. In a very simplified way, you would do the full pump-probe sequence for one set of microcrystals, then probe only for the next set as dark control, then pump-probe sequence again for the subsequent set, in an interleaved pattern.

If dark controls correlate to ground-state, this indicates that light contamination was minimal. Multiple crystals were utilised to replicate timeframes and obtain averaged data for validity. Cross-validation of the same timepoint across 2 facilities (SwissFEL and SACLA) was conducted. The reader may corroborate data with Christou et al., 2023 to evaluate external validity. 

Chronology

100ps: R256 becomes dynamic and moves to stabilize the CPD.

In the active site, the 5WC (five water cluster) serves a gatekeeping function that stabilizes fully reduced FADH- state before CPD binds. The 5WC was previously inferred by Zhong from his spectroscopy studies and directly visualized in this study. Before catalytic activity, 5WC is ordered. Upon forward electron transfer (FET) from the FADH- to the CPD lesion, R256 moves away from the 5WC to stabilize CPD. 5WC rearranges to accommodate the charge redistribution during catalytic activity.

650ps: C5-C5’ breaks first due to lower energy barriers.

Previously, Zhong’s spectroscopy studies and quantum mechanics calculations supported sequential breaking of CPD during repair. Specifically, Zhong identified a spectroscopic signal correlating with this intermediate (thymine dimer radical anion) that persisted for around 90ps. Maestre-Reyna et al. found that the magnitude of negative density feature around C5-C5’ is much higher than C6-C6’, implying that C5-C5’ is broken by 650ps, while C6-C6’ is still intact.

1ns: C6-C6’ breaks.

The negative density feature around C6-C6’ continues to grow. Both C5-C5’ and C6-C6’ difference density peak distances attain the maximum of 1.9 Å. A broken C-C bond correlates with 1.9 Å, compared to 1.5 Å for intact C-C bonds. This implied the fully repaired CPD intermediate dominates at this timepoint. The more gradual rise in negative density of C6-C6’ indicates that bond breaking is slower here.

[Difference density peak distance (ΔFo) maps are calculated by subtracting reference (dark) structure from the time-resolved snapshot. ΔFo = (Timepoint) – (Dark). They are seen as coloured, webbed polygons in diagrams. These maps show where electron density has appeared at that timepoint (positive density, green and negative density, red).]

2 and 3.35ns: Back electron transfer from thymine radical anion to FAD

Oscillatory motions of isoalloxazine ring in FAD indicate BET. BET regenerates FADH- for subsequent rounds of repair.

500ns:

Separated thymines stay in the active site for up to 500ns, while the active site undergoes relaxation. The R256 residue retracts from CPD and reconnects with 5WC to stabilize FADH-. FADH- returns to its previous, buckled geometry of the resting state. The 5WC reorders into its resting structure, connecting the FAD phosphate backbone with R256.

Since recovery is the slowest step, it is identified as the rate-limiting step of CPDphr repair.

Bubble intruding region (BIR): D428-W431-R441 triad + R429

Since photolyase flips the CPD out of the DNA backbone for repair, this creates a sharp kink in the DNA structure, leaving 2 adenines (A7, A8) unpaired. The BIR in photolyase enters the bubble and stabilizes DNA. R429 directly interacts with adenines, while the triad stabilizes the surrounding DNA structure. R441 stabilises DNA phosphate.  

When repair is complete, the thymine pair flips back into the backbone, displacing the BIR. W431 continues interacting with the restored thymines and nearby bases for some time before full release.

Multiphoton Effects

Intense laser pulses on a small crystal may cause multiphoton absorption by FADH-, exciting FADH- to higher electronic states instead of S1. Higher states may behave differently compared to physiologically activated FADH-*. However, it is believed that spontaneous decay of multiply excited FADH-* occurs much faster than CPD repair. Hence, the one-electron forward transfer to CPD should still be retained, similar to physiological processes. This is validated by repair timescale corroboration with Zhong’s spectroscopy studies.

An alternative explanation is that photoenzymes may have evolved to direct all deposited energy towards functional outcomes. Excess energy from multiphoton absorption could be dissipated safely as heat in the protein matrix, rather than reactive radicals which may affect the catalytic process. Therefore, the kinetics of the repair process remain similar to physiological processes.

Christou et al., 2023

This team used only one facility (SwissFEL) suitable for capturing the ultrafast dynamics of early events, focusing on geometric changes of FAD and the 5WC system. This is a shorter and more straightforward paper.

FAD Geometric Changes

In the dark state, FADH- exists in a bent geometry of 14°. Upon photoreactivation to FADH-*, this bend inverts to -23°. The 5WC rearrange themselves to accommodate this geometric and electrostatic change. At 3ns, when both C-C bonds have been cleaved in the CPD, and the separated thymines are still in the active site, FAD is in the process of recovering its resting geometry. 

Further reading (in chronological order)

Park, H. W., Kim, S. T., Sancar, A., & Deisenhofer, J. (1995). Crystal structure of DNA photolyase from Escherichia coli. Science, 268(5219), 1866–1872. https://doi.org/10.1126/science.268.5219.1866

Mees, A., Klar, T., Gnau, P., Freimann, U., Eichele, G., & Carell, T. (2004). Crystal structure of a photolyase bound to a CPD-like DNA lesion after in situ repair. Science, 306(5702), 1789–1793. https://doi.org/10.1126/science.1105778

Kao, Y. T., Saxena, C., Wang, L., Sancar, A., & Zhong, D. (2005). Direct observation of thymine dimer repair in DNA by photolyase. Proceedings of the National Academy of Sciences USA, 102(45), 16128–16132. https://doi.org/10.1073/pnas.0506586102

Liu, Z., Tan, C., Guo, X., Li, J., Wang, L., Sancar, A., & Zhong, D. (2011). Dynamics and mechanism of cyclobutane pyrimidine dimer repair by DNA photolyase. Proceedings of the National Academy of Sciences USA, 108(35), 14788–14793. https://doi.org/10.1073/pnas.1110927108

Zhang, M., Wang, L., & Zhong, D. (2017). Photolyase: Dynamics and electron-transfer mechanisms of DNA repair. Archives of Biochemistry and Biophysics, 632, 104–116. https://doi.org/10.1016/j.abb.2017.08.008

Maestre-Reyna, M., Wang, P. H., Nango, E., Hosokawa, Y., Saft, M., Furrer, A., Yang, C. H., Gusti Ngurah Putu, E. P., Wu, W. J., Emmerich, H. J., Caramello, N., Franz-Badur, S., Yang, C., Engilberge, S., Wranik, M., Glover, H. L., Weinert, T., Wu, H. Y., Lee, C. C., Huang, W. C., … Tsai, M. D. (2023). Visualizing the DNA repair process by a photolyase at atomic resolution. Science (New York, N.Y.), 382(6674), eadd7795. https://doi.org/10.1126/science.add7795

Christou, N. E., Apostolopoulou, V., Melo, D. V. M., Ruppert, M., Fadini, A., Henkel, A., Sprenger, J., Oberthuer, D., Günther, S., Pateras, A., Rahmani Mashhour, A., Yefanov, O. M., Galchenkova, M., Reinke, P. Y. A., Kremling, V., Scheer, T. E. S., Lange, E. R., Middendorf, P., Schubert, R., De Zitter, E., … Lane, T. J. (2023). Time-resolved crystallography captures light-driven DNA repair. Science (New York, N.Y.), 382(6674), 1015–1020. https://doi.org/10.1126/science.adj4270

TR SFX

Barends, T. R. M., Stauch, B., Cherezov, V., & Schlichting, I. (2022). Serial femtosecond crystallography. Nature reviews. Methods primers, 2, 59. https://doi.org/10.1038/s43586-022-00141-7