**3. Discussion and Conclusions**

Molecular dynamics simulations were performed on the YqhD dimer, with oxidized and reduced NADP/H cofactors bound in aqueous solution. The starting point of the simulations came from the only solved crystal structure of holo YqhD [7], which has a modified NADPH(OH)<sup>2</sup> cofactor, and Zn2+ in the active site of only one of the monomers within the functional dimer unit. Simulations were run on the functional structure of YqhD, with Zn2+ present in each monomer, and with NADP<sup>+</sup> and NADPH cofactors, using careful methodology to prepare these structures (i.e., repairing the crystal structure artifacts). Good conformational sampling of YqhD bound to NADP/H was obtained by performing five sets of simulations that were assigned different initial velocities from a Maxwell-Boltzmann velocity distribution at 300 K. The structures remain conserved throughout the simulations, regardless of the oxidation state of NADP/H cofactor. So, we compiled a combined trajectory of the 200 ns from the set of five simulations with both NADP/H cofactors and obtained sampling of various conformations representing opening and closing of the cofactor-binding interdomain cleft. Each monomer of the YqhD dimer showed coordinated cleft opening/closing, with both oxidized and reduced NADP/H, via the movement of β6/β7 and α12 regions in the interdomain cleft. The cleft remains open in the apo form of the YqhD crystal structure [7], indicating that it opens and closes for cofactor binding. Observations of cofactor release during MD simulations confirm the movement to an open-cleft conformation prior to the release of both NADP and NADPH. The sampling of open-cleft conformations depends on the cofactor oxidation state. NADPbound monomers tend to sample more open conformations, with cleft distances ranging from ~0.7–1.8 nm (cf. ~0.7–1.5 for NADPH, 1.25 nm for apoprotein crystal structure [7]).

The dynamics of cleft opening/closing were also found to depend on the oxidation/reduction of the cofactor. With both cofactors, YqhD undergoes periods of rapid cleft opening and closing, sampling open conformations similar to the apoenzyme [7]. NADPbound monomer scarcely populates conformations with a closed cleft in 10% of time points and remains in a partially closed state for 49% of time points or adopts open cleft conformation in 41% of time points. NADPH-bound monomers remain in closed conformation for 44% and spend only 17% in open state conformations. The maximum waiting time of transition from closed to open state was observed to be ~8 ns in NADP-bound monomer and 68 ns in NADPH-bound monomer. However, NADP-bound monomer underwent a total of 750 transitions between open and closed states, compared to 318 total cleft state transitions observed for NADPH-bound monomer, indicating a higher propensity for cleft opening dynamics when the enzyme is bound to the oxidized cofactor. Out of five independent simulations, cofactor release was observed in three sets of NADP-bound and only one set of NADPH-bound structures. This suggests an entropic, dynamics-based preference for cofactor release when NADP is bound.

Significant differences were observed even in the conformation of NADP/H cofactor, concomitant with conformational differences in the monomers. The differences in

cofactor-enzyme interactions and YqhD conformational dynamics may rationalize differences in the Michaelis constant, KM, which depends on cofactor binding (k<sup>f</sup> ) and release (kr) substrate-binding rates as K<sup>M</sup> = (k<sup>r</sup> + kcat)/k<sup>f</sup> . The higher number of cofactor-monomer hydrogen bonds between NADPH-bound vs. NADP-bound YqhD may lead to higher binding rates and/or slower release rates for NADPH, resulting in the higher affinity for NADPH indicated by its lower K<sup>M</sup> value (0.008 mM [11] vs. 0.150 mM [21] for NADP). Furthermore, the less frequent sampling of open conformations with NADPH-bound YqhD may hinder NADPH release, dropping k<sup>r</sup> and subsequently K<sup>M</sup> for NADPH. Meanwhile, conformational dynamics appear to promote the more facile release of NADP, following hydride transfer from NADPH to aldehydes. Thus, the conformational differences induced by cofactor oxidation state, dynamical effects of more-frequent cofactor cleft opening with NADP, and differences in hydrogen bond motifs may lead to preferential kinetics for the aldehyde reductase activity of YqhD. These findings raise questions about whether enzymes with higher alcohol dehydrogenase activity show a dynamic preference for the release of reduced cofactor. Results of this study enhance our basic understanding toward this class of enzyme, with the possible application of guiding the rational design of YqhD to enhance substrate affinity and biocatalyst efficiency.
