DNA photolyase: Is the nonproductive back electron transfer really much slower than forward transfer?

In a recent commentary, Stuchebrukhov (1) reviewed the progress in understanding DNA repair by photolyase, with special emphasis on a time-resolved study by Liu et al. (2). Among others, he raised the question of why back electron transfer (ET) to the semireduced flavin cofactor FADH° before splitting of the cyclobutane pyrimidine dimer (CPD) was slow (2.4 ns, compared with 250 ps for the preceding forward ET from the photoexcited fully reduced FADH⁻ to the CPD), and suggested two possible reasons: the Marcus inverted region slow-down effect or the electron “is locked in a state that prevents back-transfer until the reaction is completely finished.”

Before raising this question, one should ask whether back ET is really that slow. The 2.4-ns time constant was not observed directly in ref. 2. Rather it was deduced from an 87-ps dynamics attributed to the second bond cleavage in the CPD, assuming that the competition of the back ET with this cleavage is responsible for a reduction of the repair quantum yield to 82%, compared with 85% for the preceding forward ET. However, no absolute repair quantum yield was determined in ref. 2, and no reference reporting 82% was given. Values in the literature vary from at least 45% to 100% (3).

In fact, a rough estimate of the repair quantum yield can be obtained from absorption transients at 266 nm presented in ref. 2. An initial bleaching due to formation of the excited state of FADH⁻ is followed by an absorption increase due to restoration of intact pyrimidines. The ratio of the final absorption increase to the initial bleaching was ∼1 in ref. 2 but ∼2.5 in a preceding study by Thiagarajan et al. (4), indicating that the repair quantum yield under the experimental conditions of ref. 2 was even below the 52 ± 5% determined actinometrically in ref. 4. Of note, in ref. 2 a CPD without any flanking nucleotides was used, whereas in ref. 4 the CPD was flanked by four nucleotides on each side, which should improve the binding affinity. Because of the uncertainty in the repair quantum yield in ref. 2, a 2.4-ns back ET time can hardly be considered as reliable data.

Furthermore, ref. 4 provided experimental evidence (from the reduction kinetics of FADH°) that a back ET before CPD splitting occurred in fact in ∼350 ps, only moderately slower than the forward ET. A similar time constant (460 ps) was reported in an earlier study by Kao et al. (5) but was attributed to electron return after CPD splitting (the kinetics of CPD repair was not yet established at that time). In both studies (4, 5), a second, slower phase of FADH° reduction (1.55 and 1.24 ns, respectively) was observed. Ref. 4 demonstrated that the 1.55-ns phase coincided with electron release from the already split CPD.

In conclusion, the nonproductive back ET seems to be only moderately slower than the forward ET, and this effect may be simply due to different driving forces.