2.4.1. Interdomain Opening-Closing Cleft Dynamics

To characterize the cleft dynamics, we evaluated the minimum distance data of cleft opening/closing residues D159 and H363. For Yqhd monomer, we set a distance of <0.55 nm for closed states and >1.0 nm for open states. By assigning the open, partiallyclosed, and closed states a value of 0, 0.5 and 1, respectively, we were able to generate a graph of state vs time, shown in Figure 6a,b. The state-time analysis is performed on the 2 µs long aggregated trajectory involving 10 sets of NADP/H-bound monomer data. Within the aggregated data, the NADP-bound monomer is in the open state for 41% of the time, partially closed 49%, and closed for 10%. In contrast, the NADPH-bound monomer highly populates the closed state (44% of time points) and exists in the transitional conformation 39% and open state for 17% of the time points in the aggregated trajectories. The state population data indicate an efficient sampling of protein conformations during the 10 independent sets of simulations. From the simulation data, the effect of cofactor oxidation state is clearly evident on the occurrence of one state over another (see Figure 6a,b). For example, the NADP-bound monomer is prone to access open states, while NADPH-bound monomer remains in closed states more frequently. We also evaluated cleft minimum distance data to calculate waiting times for the enzyme to remain in a conformation before switching cleft state (i.e., switching conformation from closed to open, and vice versa). Figure 6c,d shows the cleft dynamics in NADP/H-bound monomers as the number of events and wait time for each instance required for switching from closed to open state and vice versa. The equilibrated starting structure of YqhD homodimer is in the closed state in both NADP/H-bound monomer. During the simulations, the monomer rapidly fluctuates between open and closed conformations, resulting in waiting times ranging from picosecond to nanoseconds for switching of states. The interdomain cleft remains more dynamic

in NADP-bound monomer with a total of 750 events of cleft opening/closing, compared to NADPH-bound domain with 318 events of state switching. The waiting time for state switching remains under 1 ns in 75% of cleft opening/closing events of NADP-bound monomers and 80% of NADPH bound monomers. For switching from closed to open state, the average wait time was 2.4 ns in NADPH-bound monomer and 409 ps in NADP-bound monomer. The longest wait time of switching from closed to open state was 8.3 ns in NADP-bound monomer and 68 ns in NADPH-bound monomer. However, switching from open to closed state occasionally required even longer average waiting times of 2.9 ns (maximum 56 ns) for NADPH-bound and of 1.4 ns (maximum 43 ns) for NADP-bound monomer. The cleft dynamics data evidences the effect of cofactor oxidation state on enzyme dynamics. Simulation results indicate that (a) the cleft dynamics are influenced by the cofactor oxidation state and (b) opening of the interdomain cleft facilitates the release of NADP cofactor (below), which is consistent with the experimentally observed properties of the biocatalysts such as the higher affinity of YqhD enzyme for NADPH cofactor indicated by its lower K<sup>M</sup> value of 0.008 mM [11] over NADP cofactor, K<sup>M</sup> = 0.150 mM [21]. **2021**, , x FOR PEER REVIEW 10 of 17

**Figure 6.** Open, partially closed/open, and closed states of the interdomain cleft observed using residue pair D159-H363 in YqhD monomer bound to (**a**) NADP and (**b**) NADPH cofactors during the simulations. Waiting time is for each instance of cleft opening and closing in YqhD monomer bound to (**c**) NADP and (**d**) NADPH.

2.4.2. Cofactor Binding and Release in YqhD Enzyme

The cofactor was found to be quite flexible during simulations, which can be quantified with variations in cofactor RMSD and Rg values. Figure 7 shows the distribution of RMSD and Rg values in the 2 µs of aggregated trajectory data, analyzing NADP/H-bound within each monomer of the YqhD dimer. The NADP cofactor shows a broader distribution of

RMSD values, with five peaks at 0.40, 0.30, 0.21, 0.13 and 0.50 nm, compared to NADPH with four peaks at 0.15, 0.30, 0.22 and 0.42 nm. The RMSD distribution for NADP cofactor is shifted to the right, sampling conformations having higher RMSD values relative to the holoprotein crystal structure. Rg values of the NADP/H cofactors indicate its compactness during the simulations. The NADPH cofactor shows a main peak at 0.73 nm that is close to the Rg of cofactor observed in the crystal structure, representing an extended cofactor confirmation bound in the intradomain region of each monomer. Additional peaks are observed at 0.60 nm and small peaks at 0.44 and 0.49 nm. However, the Rg distribution of the NADP cofactor is shifted to the left (more compact), with three main peaks at 0.58, 0.60 and 0.49 nm. During the simulations, extended confirmations of cofactor were sampled more frequently by NADPH than NADP. This detail is in agreement with the previous observation that the NADP-bound monomers sample more open and partially closed conformations, having a looser cofactor binding, with the open cleft conformation facilitating cofactor release. **2021**, , x FOR PEER REVIEW 11 of 17

**Figure 7.** Distributions of (**a**) RMSD and (**b**) Rg values are shown for NADP and NADPH cofactors in the aggregated trajectories of monomers. Cyan colored horizontal bar represents cofactor Rg in the crystal structures.

Subsequently, we observed an average minimum distance in the monomer trajectories between protein and cofactor, and cofactor and Zn2+ of 0.16 and 0.45 nm, respectively, to the crystal structure (see Figure 8a,b). The distances between cofactor and protein had significant differences during the simulations and are related to the binding of cofactor in each monomer. Zn2+ has an average distance of 0.25 nm from residues Asp194, His198, His267 and His281 during the simulations as observed in the crystal structure (see Figure 5a). A minimum of one water and Ala141 remain within 0.19 nm of the Zn2+ over the course of both NADP and NADPH simulations. Within 0.35 nm distances from Zn2+, an average of 12 ± 3 and 10 ± 3 water molecules were present in NADP and NADPH-bound YqhD monomer, respectively. A higher number of water molecules were present close to Zn2+ in open conformations than the closed one: 15 ± 2 water molecules for NADPH and 15 ± 7 for NADP-bound YqhD during the time-interval 800–1000 ns, which shows the population of open conformations indicated by higher distances between cofactor and protein in Figure 8a.

**2021**, , x FOR PEER REVIEW 12 of 17

**Figure 8.** The minimum distance is in nm between (**a**) protein and cofactor, and (**b**) cofactor and zinc ion as a function of time in the aggregated trajectory of monomer bound to NADP and NADPH. Large distances correspond to cofactor release events. (**c**) Cofactor binding site shows residues forming hydrogen bonds in the starting structure of YqhD monomer. Cofactor binding residues are colored based on their hydrogen bonding partner, being adenine moiety in orange, nicotinamide in green color, and nucleotide or phosphate group in purple colors. (**d**) Snapshots of NADP-bound YqhD monomer showing cofactor binding site starting from the equilibrated structure in Set 2 trajectory followed by every 25 ns interval up to cofactor release event in the second set of YqhD simulation.

NADP-bound YqhD showed cofactor release in three out of five sets of simulation trajectories, while NADPH-bound monomer exhibited cofactor release only once, in Set 3. The NADPH-bound monomer has a more rigid cofactor binding, which coincides with a conformational change in the cofactor binding pocket. The changes in the cofactor distances from the protein are related to the dynamic nature of cofactor in the binding site, as observed earlier with deviations in RMSD and Rg values of monomer and cofactor during the simulations. A total of 15 hydrogen bonds were observed in the starting structure of the cofactor binding site to the cofactor. Figure 8c shows the residues involved in hydrogen bonding with NADP/H cofactor. Hydrogen bonds involve the adenine moiety with residues Thr138, Tyr179, and Thr182; the dinucleotide moiety and phosphate groups with residues Gly38, Ser40, Gly95, Ser96, His281, and Lys160; and the nicotinamide moiety with residues Asp99 and Gly149. During the simulations, Lys160 has no hydrogen bond with NADP/H cofactor, while His281 shows hydrogen bond existence in 5% of NADP-bound monomer trajectory frames. Excluding these latter two hydrogen bonds present in the starting structure, NADP/H-bound structures have conserved hydrogen bonds involving

the adenine moiety, nucleotide and phosphate group in 50% of the aggregated trajectory frames. The NADP-bound monomer has only one hydrogen bond involving Leu279 and nicotinamide moiety for 6% existence in the aggregated trajectory. However, NADPH bound monomer forms three potential hydrogen bonds involving Asp99, Ser144, and Gly149 in ~30% of the aggregated trajectory. In both NADP/H-bound structures, hydrogen bonds involving the adenine moiety were observed more frequently (50% time points) than with nicotinamide moiety (6% in NADP and 30% in NADPH-bound monomers), which also supports our earlier observations of lower Rg and indicates the population of a bent conformation of NADP/H cofactor in open cleft states before release shown in Figure 8d. Overall, these results indicate that cofactor release is associated with an increase in the interdomain cleft (>0.55 nm), followed by loosened cofactor binding at cleft distances of 0.55–1.0 nm and loss of hydrogen bonds involving nicotinamide moiety, a change in cofactor conformation to a more compact structure (lower Rg), and, finally, release.
