Evolutionary analysis of Quinone Reductases 1 and 2 suggests that NQO2 evolved to function as a pseudoenzyme

Abstract

Quinone reductases 1 and 2 (NQO1 and NQO2) are paralogous FAD‐linked enzymes found in all amniotes. NQO1 and NQO2 have similar structures, and both catalyze the reduction of quinones and other electrophiles; however, the two enzymes differ in their cosubstrate preference. While NQO1 can use both redox couples NADH and NADPH, NQO2 is almost inactive with these cosubstrates and instead must use dihydronicotinamide riboside (NRH) and small synthetic cosubstrates such as N‐benzyl‐dihydronicotinamide (BNAH) for efficient catalysis. We used ancestral sequence reconstruction to investigate the catalytic properties of a predicted common ancestor and two additional ancestors from each of the evolutionary pathways to extant NQO1 and NQO2. In all cases, the small nicotinamide cosubstrates NRH and BNAH were good cosubstrates for the common ancestor and the enzymes along both the NQO1 and NQO2 lineages. In contrast, with NADH as cosubstrate, extant NQO1 evolved to a catalytic efficiency 100 times higher than the common ancestor, while NQO2 has evolved to a catalytic efficiency 3000 times lower than the common ancestor. The evolutionary analysis combined with site‐directed mutagenesis revealed a potential site of interaction for the ADP portion of NAD(P)H in NQO1 that is altered in charge and structure in NQO2. The results indicate that while NQO1 evolved to have greater efficiency with NAD(P)H, befitting an enzymatic function in cells, NQO2 was under selective pressure to acquire extremely low catalytic efficiency with NAD(P)H. These divergent trajectories have implications for the functions of both enzymes.

1. INTRODUCTION

Quinone reductases comprise an ancient family of flavoenzymes that afford the two‐electron reduction of quinones and other electrophiles as a detoxification strategy. These enzymes have a tightly bound FAD or FMN cofactor to receive a hydride from reducing cosubstrates NAD(P)H and then transfer the hydride to the quinone substrate or other electrophile (Deller et al. 2008; Vasiliou et al. ²⁰⁰⁶). Humans have two paralogous quinone reductase genes: Quinone Reductase 1, nqo1 and Quinone Reductase 2, nqo2 (Jaiswal et al. 1990; Zhao et al. ¹⁹⁹⁷). The two proteins share 47% sequence identity and a strikingly similar overall structure. NQO1 has numerous roles, and its regulation as part of the Keap1‐Nrf2 pathway provides a clear enzymatic function in the cellular response to oxidative stress (Lind et al. 1982; Ross and Siegel ²⁰²¹; Sihvola and Levonen ²⁰¹⁷).

NQO2 is unusual because, in contrast to NQO1, it cannot efficiently use the cellular redox couples NAD(H) and NADP(H) (Chen et al. 2000; Islam et al. ²⁰²²; Liao et al. ¹⁹⁶²; Liao and Williams‐Ashman ¹⁹⁶¹). Intriguingly, NQO2 can efficiently use various smaller dihydronicotinamide cosubstrates such as dihydronicotinamide riboside (NRH) and N‐propyl‐dihydronicotinamide (Liao et al. 1962; Liao and Williams‐Ashman ¹⁹⁶¹), N‐benzyl‐dihydronicotinamide (BNAH), N‐methyl‐dihydronicotinamide (NMEH), and others as sources of electrons (Knox et al. 2000). Of these smaller reducing cosubstrates, only NRH, and its oxidized form, nicotinamide riboside (NR), function in cells where they contribute to salvage pathways to regenerate NAD⁺ (Bieganowski and Brenner 2004; Giroud‐Gerbetant et al. ²⁰¹⁹; Yang et al. ²⁰¹⁹; Yang et al. ²⁰²⁰). When NQO2 was first discovered, a suggested cellular function was the consumption of cellular NRH (Liao et al. 1962). However, there is currently no evidence that the relatively low levels of NR and NRH allow them to function as a cellular redox couple to facilitate enzymatic reduction of electrophiles by NQO2. In fact, experiments in cells using a substrate (CB1954) that becomes highly toxic upon NQO2‐catalyzed reduction indicate that NQO2 is unable to catalyze reduction of CB1954 unless cells are supplied with exogenous small nicotinamide cofactors such as NRH or BNAH that can be internalized by the cell (Islam et al. 2022; Knox et al. ²⁰⁰⁰). On this basis, there does not appear to be sufficient cellular NRH, or another source of reducing equivalents, to enable efficient catalysis by NQO2. This raises the question as to how and why NQO2 would evolve to lose its ability to efficiently use the primary cellular redox couples, NADH and NADPH.

The very low catalytic efficiency of human NQO2 with NAD(P)H as cosubstrate is conserved in other mammals (Celli et al. 2006; Liao et al. ¹⁹⁶²; Zhao et al. ¹⁹⁹⁷), along with NQO2 from a reptile (Alligator mississipiensis) and bird (Anas platyrhynchos), consistent with conservation of this property throughout the amniotes (Islam et al. 2022). A phylogenetic analysis of NQO1 and NQO2 sequences pointed to a gene duplication event in a vertebrate ancestor before the split of the ray‐finned and lobe‐finned fishes, approximately 450 million years ago (MYA) (Hedges 2009; Islam et al. ²⁰²²). Gene duplication is a significant force for functional diversity in protein families (Ohno 1970; Zhang ²⁰⁰³). The most common fate of duplicated genes is the degeneration and silencing of one of the pair to a non‐functioning pseudogene (Lynch and Conery 2000; McGrath et al. ²⁰¹⁴). This seems to be what happened in the ray‐finned fish, which have only a single NQO1‐like gene and therefore lost the second quinone reductase gene. However, the lobe‐finned coelacanth, some amphibians, and the amniotes have both NQO1 and NQO2, and the properties of NQO2 have been conserved across the amniotes over millions of years of evolution. The retention of both genes that are expressed to yield functional proteins across the amniotes indicates that both NQO1 and NQO2 have been maintained through selection (Walsh 2003). On this basis, there are two possible fates for the duplicated genes: they could develop subfunctionalization, where the original function is partitioned across both copies. In the case of subfunctionalization, one might expect that NQO1 and NQO2 would differ in substrate (electrophile) specificity and/or tissue‐specific regulation, but both would maintain efficient use of the primary cellular redox couples, NAD(P)H. The alternative to subfunctionalization is neofunctionalization, where one copy acquires a new function (Zhang 2003). The difference in cosubstrate specificity observed between NQO1 and NQO2 points towards neofunctionalization, although one could argue that NQO2 has evolved to use smaller nicotinamide cofactors efficiently and in the process lost the ability to use NAD(P)H efficiently. In this regard, an important question is whether NQO2 evolved to use NRH efficiently, or whether it evolved specifically not to use NAD(P)H efficiently.

To address this question, we used ancestral sequence reconstruction to investigate the evolution of cosubstrate specificity in NQO1 and NQO2. Ancestral sequence reconstruction has been used to trace the molecular mechanisms of substrate or inhibitor specificity for many enzyme families (Harris et al. 2022; Hobbs et al. ²⁰¹²; Merkl and Sterner ²⁰¹⁶; Wilson et al. ²⁰¹⁵). We have reconstructed the predicted common ancestor of NQO1 and NQO2, along with two ancestors of each enzyme along their evolutionary paths from the common ancestor. The reconstructed NQO enzymes are catalytically active with the common quinone substrate, menadione. However, kinetic analyses show that after duplication from the common ancestor, NQO2 lost the ability to efficiently use NADH as a cosubstrate, whereas the catalytic efficiency of NQO1 improved compared to the common ancestor. In contrast to the consistent changes towards increased (NQO1) or decreased (NQO2) catalytic efficiency with NADH, all the enzymes were able to use NRH and BNAH with moderate to high efficiency. This indicates that NQO2 evolved to avoid efficient reduction by NAD(P)H rather than use NRH efficiently. The evolutionary analysis shows that NQO2 has undergone neofunctionalization rather than subfunctionalization and illuminated some of the molecular determinants for the inefficient use of NAD(P)H by NQO2.

2. RESULTS

2.1. Resurrection of ancestral NQO1 and NQO2 proteins

A phylogeny based on extant NQO1 and NQO2 sequences indicated a gene duplication event in an ancestral fish approximately 450 MYA, prior to the divergence between the ray‐finned and lobe‐finned fishes; an abbreviated phylogenetic tree is illustrated in Figure 1a with the full phylogenetic tree in Figure S1, Supporting Information. To track the evolution of cosubstrate specificity of NQO1 and NQO2, we reconstructed their common ancestor and two additional enzymes along each of the evolutionary pathways to extant NQO1 and NQO2. The ancestral sequence reconstruction method relies on multiple sequence alignment of extant related protein sequences and their phylogenetic relationships to reconstruct the ancestral protein sequences at the nodes of the phylogenetic tree. Ancestral NQO1 and NQO2 sequences were predicted using 172 extant NQO1 and NQO2 sequences and a combined optimization of the sequence alignment and phylogeny by maximum likelihood using PhyloBot (Hanson‐Smith and Johnson ²⁰¹⁶).

FIGURE 1.

Prediction of ancestral quinone reductase sequences. (a) Summary of the phylogenetic tree for NQO1 and NQO2; node points indicated with circles are sequences that are reconstructed in this study. CA or “common ancestor” was the ancestral gene that was duplicated to yield extant NQO1 and NQO2. Node points CA1NQO1 and CA1NQO2 represent the predicted sequences at the top of the NQO1 and NQO2 phylogenetic trees. Node points CA2NQO1 and CA2NQO2 are at the top of the NQO1 and NQO2 phylogenetic trees for the amniotes. The complete phylogenetic tree is provided in Figure S1. (b) Multiple sequence alignment of NQO1 and NQO2 with the maximum likelihood sequences of the resurrected ancestral proteins. The residues in orange were predicted with probability below 0.6. The secondary structure for extant human NQO1 is provided at the top of the alignment, with α, β, and η representing alpha‐helices, beta strands, and 3₁₀ helices, respectively; “TT” indicates β‐turns. NQO1 and NQO2 have similar overall structures, and therefore the secondary structure for NQO2 follows closely that of NQO1. (c) To assess the reliability of the ancestral sequence reconstructions, the maximum likelihood probability for predicted residues in the ancestral proteins are shown against the relative frequency or fraction of all residues for a given protein.

The multiple sequence alignment and the phylogenetic relationships were used to infer 171 ancestral sequences for the nodes of the phylogenetic tree. To characterize the changes in NQO1 and NQO2 from a common ancestor to the extant proteins, five predicted sequences were chosen for reconstruction (Figure 1a,b). Three of the sequences included the common ancestor (CA) and the two nodes immediately following the duplication (CA1NQO1 and CA1NQO2). An additional two nodes were chosen at the top of the phylogenetic trees for the amniotes (CA2NQO1 and CA2NQO2). For a given reconstructed ancestral sequence, each residue had a particular posterior probability that gave the likelihood of the residue at that position; these results are summarized in Figure 1c for the five reconstructed sequences, and the complete data for individual residues are provided in Tables S1–S5. The resulting five sequences were reliably predicted, with 60% of the residues in the common ancestor (CA) having a probability of 0.95–1.0, rising to 90% for the second common ancestor of NQO2 (“CA2NQO2,” Figure 1c). NQO1 (MW = 31 kDa) and NQO2 (MW = 26 kDa) differ in that NQO1 has a 43 residue C‐terminal domain and the only gap in the alignment of the NQO sequences occurs in the C‐terminal domain of the common ancestor.

2.2. Purification and characterization of the ancestral proteins

The predicted protein sequences were reverse‐translated to their corresponding nucleotide sequences for expression in Escherichia coli. Protein expression and purification followed our established method for extant NQO2 sequences (Leung et al. 2012). The ancestral sequences were efficiently expressed in E. coli and purified to homogeneity using Ni‐NTA affinity chromatography, removal of the affinity tag using TEV protease, and then ion‐exchange and gel filtration chromatography. The last step included reconstitution with FAD since our experience indicated that NQO2 produced in E. coli contained sub‐stoichiometric amounts of FAD (Leung et al. 2012). The final protein preparations were analyzed for purity by SDS‐PAGE (Figure 2a).

FIGURE 2.

Resurrection and characterization of ancestral quinone reductases. (a) SDS‐PAGE of the purified ancestral proteins. CA is the common ancestor that was duplicated to produce CA1NQO1 and CA1NQO2, the first ancestral NQO proteins in the phylogenetic tree. CA2NQO1 and CA2NQO2 are the ancestral NQO1 and NQO2 proteins at the top of the phylogenetic tree for the amniotes. (b) UV–Visible absorbance spectra of the purified quinone reductases to verify the presence of FAD (maxima at 380 and 455 nm) in the constructs. (c) The thermal stability of the CA, NQO1, and NQO2 ancestral and extant proteins was assessed using a thermal shift assay. The melting temperature is the average of at least four replicates; error bars represent the 95% confidence limits.

The common ancestor (CA) was subject to partial proteolysis during purification, indicated by a faster‐migrating band on SDS‐PAGE (Figure 2a). The predicted C‐terminal sequence in the CA contained a two‐residue gap and several residues that were predicted with a low probability compared to CA1NQO1 and CA2NQO1 (Figure 1b). Furthermore, predictions of the CA sequence using other software, such as FastML (Ashkenazy et al. 2012) or ProtASR (Arenas et al. 2017), did not include the C‐terminal domain. Since the C‐terminal domain is not present in other quinone reductases, it is likely that the CA did not have this domain, and that it was acquired after the gene duplication. On this basis, we suspect that the predicted CA C‐terminal domain was disordered and subject to proteolysis.

The preparations of ancestral quinone reductases were analyzed for FAD content and stability. In a previous study we found that NQO2 overexpressed in E. coli was catalytically active, but contained only 0.38 equivalents of FMN rather than 1 equivalent of FAD per protomer (Leung et al. 2012). Based on this observation we incorporated a reconstitution step in the purification of NQO2 to fully charge the protein with FAD. This reconstitution step was incorporated in the purification of all 5 ancestral enzymes, and UV/Vis spectroscopy was used to assess their FAD content. For NQO2 saturated with FAD, the ratio of absorbance at 273 nm to the absorbance maximum for the isoalloxazine ring at 455 nm was 7.2 (Leung et al. 2012). The absorbance spectra of the ancestral proteins were normalized at their maxima (270–275 nm) and showed maxima for the isoalloxazine ring at wavelengths of 455 nm, with A273/A455 ratios between 6.3 and 7.6, consistent with stoichiometric FAD content (Figure 2b).

To assess the stability of the ancestral enzyme constructs, we determined their melting temperature by thermal shift assay. The melting temperatures (T _M) were derived from the thermal denaturation profiles using both a fluorescence derivative plot and nonlinear fitting to Boltzmann sigmoidal curve, which showed close agreement (Vivoli et al. 2014). The mean T _M values from fluorescence derivative plots are presented in Figure 2c and Table S6. The ancestral sequences were more thermostable by at about 10°C than the corresponding extant proteins.

2.3. Evolution of catalytic efficiency of NQO1 and NQO2 with NADH

The molecular evolution of NQO1 and NQO2 with respect to their use of NADH as a reducing cosubstrate was tracked using steady‐state kinetic analyses. The NQO1 and NQO2 lineages evolved very different cosubstrate specificities, providing evidence for functional divergence following gene duplication. The common ancestor, CA, had a turnover number for NADH of 5.95 s⁻¹ and a modest catalytic efficiency of 9550 M⁻¹·s⁻¹ (Figure 3a and Table S8). Going down the NQO1 lineage, the catalytic efficiency with NADH increased first ~8‐fold with CA1NQO1 and then an additional ~11‐fold in extant NQO1 (Figure 3a and Table S7). In contrast, the catalytic efficiency of the NQO2 lineage with NADH decreased 200‐fold from the CA to CA1NQO2 and then an additional 15‐fold to extant NQO2. The inability to efficiently use NADH by the NQO2 enzymes was not only due to a decrease in k _cat, but also a large increase of the K _M, into the millimolar range such that it is difficult to measure accurately (Table S7). Extant NQO1 can use both NADH and NADPH with similar efficiency. We also tested the ability of the ancestral proteins to use NADPH as a reducing cosubstrate and found that the CA and CA1NQO1 ancestors could use NADPH with similar efficiency as with NADH. At the same time, NQO2 enzymes lost the ability to use NADPH. To summarize, after the gene duplication from a common ancestor, the sequence identities of the two enzymes decreased to 47% and NQO1 acquired a 100‐fold increase in catalytic efficiency with NADH, while NQO2 decreased by 3000‐fold in catalytic efficiency with the same cosubstrate to yield a final ratio of catalytic efficiencies of approximately 300,000 (Figure 3d).

FIGURE 3.

Evolution of quinone reductase catalytic efficiency. The resurrected ancestral and extant quinone reductases were tested for their catalytic efficiency (k _CAT/K _M) with cosubstrates NADH (a), NRH (b), or BNAH (c), using menadione as the common quinone substrate. Tables S7–S9 have the full kinetic parameters for NADH, NRH, and BNAH, respectively. (d) A graphic illustrating the divergent evolution of the two enzymes from the common ancestor (CA). The numbers indicate the percentage sequence identities and ratio of catalytic efficiencies with NADH ([efficiency(NQO1)]/[efficiency(NQO2)]) for the extant enzymes and their two ancestors, CA1 and CA2.

2.4. Efficient use of NRH and BNAH by ancestral and extant quinone reductases

Although extant NQO2 cannot use conventional dihydronicotinamide cosubstrates NAD(P)H efficiently, it can catalyze quinone reduction very efficiently given smaller nicotinamide cosubstrates such as NRH, NMEH, or BNAH, among others (Knox et al. 2000; Kwiek et al. ²⁰⁰⁴). To test the hypothesis that NQO2 has evolved to specifically use small nicotinamide cosubstrates such as NRH (Liao et al. 1962), we measured the ability of the ancestral NQO1 and NQO2 enzymes to use NRH with menadione as quinone substrate (Figure 3b and Table S8). We found that the common ancestor and the ancestral and extant NQO1 and NQO2 enzymes could use NRH as a reducing cosubstrate with relatively high catalytic efficiencies, with no evidence of systematic change over time. We also measured the oxidation of BNAH by the ancestral enzymes (Figure 3c and Table S9) to find whether both enzyme lineages exhibited similar efficiencies with a synthetic dihydronicotinamide cosubstrate. We found that all the ancestral enzymes could use the small synthetic cosubstrate BNAH with high catalytic efficiencies in the range of 10⁶–10⁷ M⁻¹·s⁻¹. The K _M for BNAH ranged from 20 to 200 μM, and the k _cat was from 158 to 1557 s⁻¹ with no clear evidence for systematic evolutionary selection in either protein lineage. There is evidence for the existence of NRH under physiological conditions (Giroud‐Gerbetant et al. ²⁰¹⁹; Yang et al. ²⁰²⁰), but the lack of systematic evolutionary changes in NRH efficiency observed in both NQO1 and NQO2 lineages, combined with their ability to use a synthetic cosubstrate, BNAH, supports the idea that NQO2 was not under selective pressure to use smaller dihydronicotinamide cosubstrates such as NRH. Instead, it appears that both enzymes can use the smaller reducing cosubstrates adventitiously, an ability shared with their common ancestor. On this basis, NQO2 evolved to use the conventional reducing cosubstrates (NAD(P)H) inefficiently.

2.5. Molecular determinants of cosubstrate specificity in NQO1 and NQO2

To gain insight into the molecular basis of NQO2 substrate specificity, we used the ancestral sequence analysis to identify residues that could contribute to high catalytic efficiency with NADH. Looking at the NQO1 evolutionary pathway, there was a ~300‐fold increase in catalytic efficiency with NADH in going from the CA to CA2NQO1 (Figure 3a and Table S7). Sixteen residues were identified that changed going from CA to CA1NQO1 or CA2NQO1, but were conserved in the CA to NQO2 lineage (Figure 4a); these residues could contribute to the increased efficiency in NQO1. Therefore, to increase the catalytic efficiency of CA1NQO2, these residues were changed to those observed in CA1NQO1 or CA2NQO1; the changes were P15R, N62D, D128S, F129Y, P130Q, L138F, E154S, P193I, Y195H, E198P, K200A, K202S, N222S, S226N, G227S, and Y228N. Conversely, looking at the NQO2 evolutionary pathway, there was a 100‐fold decrease in catalytic efficiency going from CA to CA1NQO2 (Figure 3a). Eight residues were identified that were changed going from CA to CA1NQO2 but were conserved in the CA to NQO1 lineage (Figure 4a); these could be residues that contribute to the decreased efficiency in NQO2. To increase the catalytic efficiency of CA1NQO2, these residues were changed back to the ones found in the CA; the changes were N24D, S31K, E74L, K166N, Y167V, E194G, C223F, and T224V. In deciding which residues to mutate, we also note that the active site of NQO2 has a negative electrostatic potential compared to the active site of NQO1, which could have important implications for binding the negatively charged phosphates in NAD(P)H. Five of the 26 substitutions introduced positive charges, (P15R, S31K, N162H, Q183K, and Y195H) and four neutralized negative charges in NQO2 (E74L, D128S, E154S, and E194G) while two new negative charges were introduced (N24D and N62D). To test the effect of the C‐terminal domain of NQO1 on cosubstrate specificity, a longer variant NQO2 with the C‐terminal domain of CA2NQO1 was also made. The two CA1NQO2 variant proteins, “short” (‐CTD) and “long” (+CTD), were expressed and purified following the same protocol as for the extant and ancestral proteins; the final preparations were characterized by SDS‐PAGE (Figure 4b).

FIGURE 4.

Distributed mutations enhance activity of CA1NQO2 with NADH. (a) The CA1NQO2 ancestor was modified at 26 residues to enhance its activity with NADH; the changes are indicated by boxes and bold type. Most mutations were chosen to replace residues in CA1NQO2 with residues in CA2NQO1 that appeared to be important for NADH catalytic efficiency based on sequence conservation and/or greater positive charge around the active site. Two CA1NQO2 variants were made, one without a C‐terminal domain, and the second with the C‐terminal domain of CA2NQO1. (b) SDS‐PAGE analysis of the CA1NQO2 variants. (c) The kinetic parameters of the CA1NQO2 variants are compared with the evolutionary NQO2 lineage (CA, CA1NQO2, and CA2NQO2) and extant NQO1/2. The CA1NQO2 variants exhibited a 300‐ to 500‐fold increase in catalytic efficiency with NADH (top graph) due to 150‐ to 400‐fold increases in the k _CAT (middle graph) combined with relatively small decreases in the K _M (bottom graph).

Our hope was that these mutations, chosen with guidance from our evolutionary analysis, would result in stable proteins that could use NADH more efficiently. The mutations had relatively little effect on catalytic efficiency with BNAH as cosubstrate, with both variants exhibiting high catalytic efficiencies like the parental CA1NQO2 enzyme (Tables S9 and S10). The mutations did have a pronounced effect on the efficiency with which the enzymes could use NADH, with increases of 500‐ and 300‐fold for the short and long forms, respectively, compared to parental CA1NQO2 (Figure 4c and Tables S7 and S10). The increased efficiencies with NADH were primarily due to increases in k _cat (Figure 4c). The presence of the C‐terminal domain on the variant CA1NQO2 enzyme did not enhance its ability to use NADH; this is consistent with a previous chimeric construct, where the C‐terminal domain of NQO1 was fused to NQO2 but did not improve its ability to use NADH as cosubstrate (Wu et al. 1997).

2.6. Interactions of NADH in NQO1

NRH lacks the ADP portion and negative charges of NADH. The 500‐fold increase in catalytic efficiency effected by the mutations in CA1NQO2 prompted us to look more closely at the changes in the protein, specifically at the electrostatic surfaces, given the clear difference between extant NQO1 and NQO2 (Figure 5a). The NQO1 active site has a more positive electrostatic potential around and above the FAD cofactor. This region of electrostatic potential above the isoalloxazine ring can be seen in the CA, which uses NADH with moderate efficiency, and in CA1NQO1, which has increased efficiency with NADH. However, this region is neutral or slightly negative in CA1NQO2, which has a drastically decreased catalytic efficiency with NADH as cosubstrate. On the other hand, the CA1NQO2 variant (CA1NQO2 26‐mut) has a restored basic region, like that observed with the CA, and a similar catalytic efficiency as the CA (Figure 5a).

FIGURE 5.

Electrostatics and a site of interaction for NADH in NQO1. (a) The ancestors of NQO1 and NQO2 were homology modeled using crystal structures of the extant enzymes as templates; the CA structure was modeled using NQO2 as a template. The FAD cofactors were included based on their positions in extant NQO1 and NQO2 structures and provide a common reference point for the surface representations. Electrostatic surfaces for the proteins were calculated using the APBS plugin in PyMol (Baker et al. 2001). The CA uses NADH with moderate catalytic efficiency (k _CAT/K _M = 10⁴ M⁻¹·s⁻¹), which is decreased over 200‐fold in CA1NQO2 (k _CAT/K _M = 45 M⁻¹·s⁻¹) but more than regained in the engineered CA1NQO2 with 26 mutations (CA1NQO2 26‐mut, k _CAT/K _M = 2.3·10⁴ M⁻¹·s⁻¹). These changes in catalytic efficiency correlate with the disappearance (in CA1NQO2) and the re‐appearance (in CA1NQO2 26‐mut) of the region of positive surface potential above the isoalloxazine ring of FAD. (b) NAD was modeled into the region of positive surface potential of extant NQO1; minor steric clashes were resolved by subjecting the complex to a single round of energy minimization using GROMACS (Abraham et al. 2015).

This basic region above the isoalloxazine ring of FAD is clearly present in NQO1, and the correlation between the increase in catalytic efficiency and the appearance of this basic region, in going from CA1NQO2 to the site‐directed variant, suggests that this may represent a site of interaction for NAD(P)H, perhaps by accommodating the negatively charged pyrophosphate moiety of NAD(P)H. To investigate this possibility, NAD was modeled into the positive electrostatic region in NQO1, putting the nicotinamide in position for hydride transfer between it and the isoalloxazine ring, fitting the pyrophosphate into a recessed basic pocket, and placing the adenine ring in an indentation on the NQO1 molecular surface. The model complex was subjected to energy minimization using GROMACS (Abraham et al. 2015), and the ADP portion of NAD could be accommodated into the site with no steric clashes (Figure 5b).

3. CHARACTERIZING THE INTERACTION OF NADH WITH NQO1 BY SITE‐DIRECTED MUTAGENESIS

To further explore the potential interaction of NAD(P)H with the basic region above the isoalloxazine ring, we made mutations in NQO1 to match the residues in equivalent positions of NQO2. We focused on possible interactions with the ADP portion of NAD by creating three separate variants (Figure 6a): NQO1(V73W), NQO1(G123Q), and NQO1(Y127F, T128D, Y129F). In the NQO1‐NAD model, V73 sits below the adenine ring, and in NQO1 homologues is typically a small residue such as V, S, T, or A, which is present as a bulkier residue (W, H, Y, or F) in NQO2. Residue 123 in NQO1 is either G or A and is replaced by Q in NQO2. Residues 127–129 interact with the pyrophosphate in the NQO1‐NAD model, are highly conserved as Y(T/S)Y in NQO1, but are present as FDI in human NQO2 or more generally as (F/H/Y)(D/E)(F/I/L) among NQO2 homologues. We changed the YTY sequence in NQO1 to FDF to conserve the aromatic features of the tyrosine residues in NQO1.

FIGURE 6.

Effect of mutations on NQO1 catalytic efficiency. (a) NAD was modeled into the basic region above the isoalloxazine ring of NQO1 and residues predicted to interact with the ADP portion of NAD were identified. Based on this analysis, 3 variants of NQO1 were created: NQO1(V73W), NQO1(G123Q), and NQO1(Y127F, T128D, Y129F). The NAD is shown in white sticks for the nicotinamide riboside (NR) portion and dark gray sticks for the ADP portion. The purple residues correspond to wild‐type NQO1 and the orange residues are the modeled mutations. (b) Wild‐type NQO1 and the three variants were tested for their catalytic activity using menadione as the quinone substrate, and either NADH (dark blue) or BNAH (light blue) as the cosubstrate.

We expressed and purified the three NQO1 variants and characterized their kinetic properties with reducing cosubstrates NADH and BNAH (Figure 6b and Table S11). The mutations had no significant effects on catalysis with BNAH as cosubstrate, with both wild‐type and the three variant NQO1 enzymes demonstrating high catalytic efficiencies and similar k _CAT and K _M values (Figure 6b). NQO1(V73W) and NQO1(G123Q) were able to use NADH as cosubstrate with similar kinetic parameters as wild‐type NQO1. Compared to the wild‐type, NQO1(YTY‐FDF) exhibited a 125‐fold decrease in turnover number and a small (2‐fold) increase in the Michaelis constant for an overall decrease in catalytic efficiency of approximately 200‐fold (Figure 6b), suggesting that these residues do function in accommodating the ADP portion of NADH.

4. DISCUSSION

When Williams‐Ashman and co‐workers originally purified a redox enzyme, now known as NQO2, that apparently did not catalyze the oxidation of NAD(P)H, but would catalyze the oxidation of NRH or N‐propyl‐dihydronicotinamide, they speculated that NRH is formed in cells, and the function of the NQO2 enzyme was to oxidize NRH to NR, thereby avoiding an accumulation of cellular NRH (Liao et al. 1962). The ancestral sequence reconstruction shows that the common ancestor of NQO1 and NQO2 had relatively modest catalytic efficiency with NADH and greater efficiency with NRH and the synthetic cosubstrate, BNAH. After the gene duplication, the ancestors of both NQO1 and NQO2 retained their ability to efficiently use small natural or synthetic dihydronicotinamide cosubstrates, but diverged in their use of NAD(P)H, with NQO1 gaining catalytic efficiency, and NQO2 losing catalytic efficiency. We conclude that NQO2 did not evolve to use NRH or other small nicotinamide cosubstrates efficiently, but instead has evolved to use NAD(P)H inefficiently. One mechanism for this change is that NQO2 has evolved to interact unfavorably with the ADP portion of NAD(P)H. In this regard, the evolutionary analysis has provided evidence for the role of the positive electrostatic potential in the active site of NQO1 and the potential interactions of NQO1 with the ADP portion of NADH. The evolution of NQO2 to use NAD(P)H inefficiently suggests the primary function of NQO2 is not the catalytic reduction of quinones; instead, we believe that it has evolved to function as a redox sensor or switch (Becker et al. 2011; Leung and Shilton ²⁰¹³).

The ancestral sequence reconstruction pointed towards a potential site of interaction for nicotinamide cofactors in this family of enzymes. The site‐directed mutagenesis that successfully increased the catalytic efficiency of CA1NQO2 highlighted evolutionary changes in the surface electrostatic potential of the NQO1 and NQO2 lineages. These electrostatic changes correlated with changes in catalytic efficiency with NAD(P)H and guided modeling of NAD in a potential binding site on NQO1. Although speculative, the “hand‐in‐glove” fit of NAD into this site of NQO1, combined with the projection of the adenine ribose into the solvent, which is consistent with the hallmark ability of NQO1 to use both NADH and NADPH as reducing cosubstrates, is consistent with NAD(P)H binding. This idea is further supported by the observation that, in addition to a neutral to negative electrostatic potential in the NQO2 lineage, the site is sterically occluded in extant NQO2 and its ancestors and would be unable to accommodate the ADP portion of NAD(P)H.

Crystal structure analyses of NQO1 in the presence of either NADP+ or NAD+ have been published (Grieco et al. 2024; Li et al. ¹⁹⁹⁵) and in both cases the binding site for the nicotinamide cofactors does not explain the inefficient use of NAD(P)H by NQO2. In the crystal structures, the ADP portion of NAD(P)⁺ is largely solvent exposed with few direct contacts to the protein. Of the direct contacts, the ADP portion of NAD(P)⁺ contacts a loop containing residues Q233 and F234 that is part the C‐terminal domain of NQO1. The current work and that of another group has demonstrated that the presence of the C‐terminal domain of NQO1 does not enhance the ability of NQO2 to use NAD(P)H (Wu et al. 1997). This is consistent with the fact that other NAD(P)H‐dependent quinone reductases in this family do not include the C‐terminal domain of NQO1; that is, these enzymes use NAD(P)H efficiently without the C‐terminal domain, which is a unique feature of NQO1. A second contact is made with H195, which certainly contributes to the positive electrostatic potential of the NQO1 active site. However, H165 was present as Y165 in the common ancestor (CA) and CA1NQO1, and was only changed to histidine in CA2NQO1; on this basis, the presence of the histidine at position 195 is not absolutely required for efficient catalysis using NAD(P)H. Finally, the YTY to FDF mutations in residues 127–129 in NQO1 had a drastic effect on the turnover number (k _cat) with NADH, but catalysis with BNAH was not significantly affected. In the crystal structures, none of these residues are near the ADP portion of NAD(P)H and so it is difficult to reconcile the effect of the mutations with the position of the ADP portion of NAD(P)⁺ in the crystal structures. In summary, the available crystal structures of NQO1 bound to NAD(P)⁺ do not provide a clear rationale for the inability of NQO2 to use NAD(P)H efficiently.

It is worth noting that for the original NQO1 crystal structure (Li et al. 1995), the coordinates and experimental structure factors for NQO1‐FAD with bound NADP⁺ were never deposited to the PDB, making a detailed evaluation of the structure and the modeled NADP⁺ impossible. For the more recent structure (PDB code 8RFM) (Grieco et al. 2024), only two of the four NQO1 protomers have bound NAD⁺, which was attributed to negative cooperativity in the NQO1 dimers. Both of the modeled NAD⁺ molecules have average thermal factors (B‐factors) that are almost 3 times higher those of the protein and non‐covalently bound FAD. On this basis, the two NAD⁺ molecules are either present at low occupancy and/or substantially disordered. In addition, the two NAD⁺ molecules are modeled in different conformations, consistent with dynamic and/or weak interactions with the protein. A stable Michaelis complex between oxidized NQO1 and NADH may be difficult to achieve: rapid reaction studies on the reductive half‐reaction of NQO1 indicated that the reaction is so fast that it can be modeled as a simple second‐order process with no observable formation of complex between the enzyme and pyridine nucleotide (Tedeschi et al. 1995). In summary, more work is required to reconcile the results of the biochemical and evolutionary analysis with those of the published crystal structures.

NQO1 has documented enzymatic functions in the cell, in particular the transfer of electrons to quinones and other electrophiles (Ross and Siegel 2021). The inefficient use of the common dihydronicotinamide cosubstrates NAD(P)H by NQO2 makes it unlikely that NQO2 has a similar catalytic function as NQO1. In fact, in all cell types tested NQO2 cannot function effectively as a catalyst for reduction of electrophiles unless an exogenous small dihydronicotinamide cosubstrate is supplied to the cell media (Islam et al. 2022; Knox et al. ²⁰⁰⁰). On this basis, after the gene duplication NQO1 retained the enzymatic function of the ancestral NQO enzyme, but NQO2 underwent neofunctionalization, or the acquisition of a completely different function. Additional evidence for neofunctionalization of the NQO2 paralog comes from the very different effect of knocking out NQO1 or NQO2 in mice. NQO1 knockout mice demonstrated increased sensitivity to menadione toxicity, consistent with a role for NQO1 in reduction and detoxification of menadione (Radjendirane et al. 1998). In contrast, NQO2 knockout mice showed decreased sensitivity to menadione toxicity, a difference that was even more pronounced when NRH was included in the menadione treatment (Long et al. 2002). In other words, the presence of NQO2 exacerbates the toxicity of menadione, an effect that is even stronger when a suitable exogenous dihydronicotinamide cosubstrate is supplied. Consistent with these knockout experiments, overexpression of NQO2 in Chinese hamster ovary (CHO) cells increased the toxicity of menadione compared to wild‐type cells, and the menadione toxicity was further increased when it was administered in the presence of NRH (Celli et al. 2006). These observations are consistent with a biological role for NQO2 that does not involve catalytic reduction of quinones; one possibility is that NQO2 functions as a redox sensor or switch to regulate cellular metabolism and/or responses to oxidative stress.

In conclusion, it appears that NQO2 has been under selective pressure to become extremely inefficient in its ability to use NAD(P)H. On this basis, the gene duplication and evolution of NQO1 and NQO2 provide an interesting example of gene duplication followed by neofunctionalization of one of the genes to produce a pseudoenzyme that is no longer able to effectively catalyze reduction of quinones or other electrophiles in cells. The cellular functions of NQO2 require further investigation, but there has been intriguing progress; two recent reviews are available for a detailed discussion (Islam and Shilton 2024; Janda et al. ²⁰²⁴). In the brain, NQO2 was found to be overexpressed in a model system for age‐related memory impairment and a pharmacological model of amnesia (Brouillette et al. 2007), and NQO2‐knockout mice demonstrate enhanced learning (Benoit et al. 2010). Roles for NQO2 in learning and memory, along with insight into the cellular mechanisms, have been investigated by Rosenblum and co‐workers (Gould et al. 2020; Gould et al. ²⁰²¹; Rappaport et al. ²⁰¹⁵), with new evidence that NQO2 dysfunction may contribute to metabolic stress and neurodegeneration (Gould et al. 2023). Our own interest in NQO2 began with the observation that it was an off‐target interactor with inhibitors of the kinase CK2 (Duncan et al. 2008; Leung and Shilton ²⁰¹⁵). The broad extent of NQO2 off‐target interactions with kinase inhibitors was demonstrated in a comprehensive screen of over 200 clinically used kinase inhibitors, where NQO2 was found as an off‐target interactor with over 30, by far the most frequent non‐kinase interactor discovered (Klaeger et al. 2017). This finding suggests that NQO2 may play a role in the cellular effects of these drugs, such that interaction with NQO2 is selected for during drug screening and optimization. It is notable that unlike NQO2, NQO1 was not been picked up in this off‐target screen for kinase‐targeted drugs. NQO2 has also been found to interact with antimalarials (Graves et al. 2002; Kwiek et al. ²⁰⁰⁴). Until the cellular functions of NQO2 are understood in greater detail, it is difficult to know how its interactions with these various drugs could affect cell metabolism, signaling, and viability.

5. MATERIALS AND METHODS

5.1. Reagents

Ampicillin, FAD, NADH, and NADPH were purchased from Sigma‐Aldrich. 1‐benzyl‐1,4‐dihydro‐3‐pyridinecarboxamide (BNAH) was purchased from TCI Chemicals (Portland, OR). Nicotinamide riboside (NR) chloride was purchased from Chromadex (Los Angeles, CA) as “Tru Niagen” nutritional supplements. To reduce the NR to NRH (dihydronicotinamide riboside), the contents of each capsule (263 mg of nicotinamide riboside) were dissolved in 10 mL of 1M NaHCO₃, pH 8.1, and the solution was filtered to remove insoluble microcrystalline cellulose. Following the basic protocol outlined in Zarei et al (Zarei et al. 2021). dihydronicotinamide riboside (NRH) was prepared by degassing the NR solution, adding 2.7 equivalents of Na₂S₂O₄ and incubating under nitrogen for 3 h. NRH was purified by reversed‐phase HPLC using a semi‐preparative Zorbax XD8‐C18 column (9.4 × 250 mm; Agilent).

5.2. Evolutionary analysis

The evolutionary analysis consisted of multiple sequence alignments of NQO1 and NQO2 sequences that were found using human NQO1 and NQO2 protein sequences as bait against the database of non‐redundant protein sequences using the NCBI BLASTp server. Whether a given sequence is NQO1 or NQO2 was determined by the annotation suggested by NCBI, by the percentage identity, and by the presence or absence of the C‐terminal domain of NQO1. A total of 172 sequences were collected. A preliminary multiple sequence alignment was calculated with MUSCLE (Edgar 2004) and showed that the sequences could be aligned with minimal gaps or insertions. The next stage of the evolutionary analysis was ascertaining the evolutionary relationship between the protein sequences by constructing a phylogenetic tree. Simultaneous multiple sequence alignment and phylogenetic tree construction were done using Phylobot (Hanson‐Smith and Johnson ²⁰¹⁶). The phylogenetic tree was constructed by maximum likelihood with the JTT + CAT evolutionary model (Jones et al. 1992). Finally, the multiple sequence alignment and the phylogenetic tree were used to infer the amino acids of the sequences at the nodes of the phylogenetic tree. The predicted protein sequences from Phylobot at specific nodes of interest were selected for analysis. The resurrected sequences comprised the amino acid residues with the highest probability at each position. The amino acid sequences were reverse‐translated to yield the corresponding DNA sequence.

5.3. Cloning of constructs and protein expression

Cloning of constructs was carried out using E. coli XL‐1 Blue competent cells. Ancestral DNA sequences were ordered as genes from Integrated DNA Technologies (“IDT,” Coralville, IA). The sequences were codon‐optimized for expression in bacteria using IDT‐provided tools. Two restriction sites were engineered into the sequences, KasI/Ehe1 at the 5′ end, and HindIII at the 3′ end to facilitate cloning into the pProEx‐HTA expression plasmid (Invitrogen) in a way that fuses a coding sequence for a 6xHis tag followed by a tobacco etch virus (TEV) protease cleavage site (i.e., HHHHHHDYDIPTTENLYFQ/GA) to the 5′ end of the quinone reductase gene. Proteolytic cleavage with TEV protease removes the affinity tag, leaving only an extra GA sequence at the N‐terminus of the proteins.

Cloning of the CA1NQO2 variants was accomplished by Gibson Assembly. The gene for the CA1NQO2 protein containing the 26 mutations and the C‐terminal domain of CA2NQO1 was purchased from IDT. PCR primers were ordered to make either a short NQO2 variant without the C‐terminal domain or a longer NQO2 variant with the C‐terminal domain of CA2NQO1.

The human NQO1 coding sequence was cloned into the expression vector pEBCaM (Ball and Basilone 2022) to yield a protein construct that comprises a hexa‐histidine calmodulin connected to NQO1 by a linker containing a TEV protease cleavage site. Mutagenesis to yield the NQO1(V73W), NQO1(G123Q), and NQO1(Y127F‐T128D‐Y129F) variants was accomplished by PCR and Gibson Assembly. The sequences of all protein constructs were verified with Sanger sequencing at the Robarts Research Institute, University of Western Ontario.

5.4. Protein expression and purification

5.4.1. Reconstructed ancient enzymes

Protein expression was done in E. coli BL21‐DE3 strains under ampicillin selection. For large‐scale protein production, four 1 L cultures were inoculated with a single colony starter culture and grown overnight at 37°C in Studier's ZYP‐5052 auto‐induction media (Studier 2005). The ancestral proteins were purified by previously described methods (Islam et al. 2022; Leung et al. ²⁰¹²). Bacterial cell pellets were thawed, treated with DNase and lysozyme, and dispersed using a Dounce homogenizer and then lysed using a French press. The cell lysates were supplemented with 50 mM Na‐phosphate, pH 7.4, 1M NaCl, and 25 mM imidazole. The broken cells were ultracentrifuged at 4°C at 100,000g for 1 h and 20 min to remove cell membranes and debris. After centrifugation, the pellet was discarded. The supernatant was applied to a 5 mL Ni²⁺‐NTA affinity column (IMAC Sepharose, Cytiva) equilibrated with the same buffer used to resuspend the cell pellets. The 6xHis‐tagged protein was eluted with 250 mM imidazole in 50 mM Na‐phosphate, 1M NaCl, pH 7.4. After Ni²⁺‐affinity purification, the protein was dialyzed against 100 mM Tris–HCl, 0.5 M NaCl, 5 mM DTT, and 5 mM EDTA, pH 8.2. The dialyzed and partially purified protein was then digested with TEV protease to remove the affinity tag.

Ion‐exchange chromatography was used as a second purification step. In preparation for anion exchange chromatography, NQO2 was dialyzed against 10 mM Tris, 1 mM EDTA, 0.5 mM TCEP (tris(2‐carboxyethyl)phosphine), pH 8.4. The protein was applied to a 1.6 × 15 cm column of Q‐Sepharose HP (Cytiva) equilibrated with 20 mM Tris, 1 mM EDTA, and 1 mM TCEP at pH 8.4. The column was developed with a gradient from 0 to 500 mM NaCl at a flow rate of 2 mL min⁻¹ while collecting 8 mL fractions. In preparation for cation‐exchange chromatography, NQO1 was dialyzed against 20 mM Na‐phosphate, 0.5 mM TCEP, and 1 mM EDTA, pH 5.0. The dialyzed protein was applied to a 1.6 × 15 cm column of S‐Sepharose HP (Cytiva) equilibrated with 20 mM Na‐phosphate, 1 mM EDTA, and 0.5 mM TCEP, pH 5.0. The column was developed with a gradient of from 0 to 500 mM NaCl at flow rate of 2 mL min⁻¹ flow while collecting 8 mL fractions. After ion‐exchange chromatography, the protein was pooled and concentrated by ultrafiltration (Amicon) for gel‐filtration chromatography. For FAD reconstitution, the concentrated protein samples were incubated with 1 to 3M Guanidine‐HCl, 10 mM FAD, 1 mM ZnCl₂, and 10 mM DTT at 4°C. After gentle mixing, the protein samples were centrifuged to remove any precipitate and then applied to a 2.6 × 65‐cm column of Superdex‐200 preparatory grade gel filtration resin (Cytiva), equilibrated with 50 mM Tris, pH 8.0 and 150 mM NaCl, 10 μM FAD, 10 μM Zn²⁺. After gel filtration, NQO2 was dialyzed against 50 mM HEPES, 100 mM NaCl, pH 7.4 and concentrated by ultrafiltration using an Amicon stirred cell. The purified and concentrated proteins were flash‐frozen in liquid nitrogen and stored at −80°C. Protein concentration was determined with the Bradford assay. To assess FAD content, we measured the UV/Vis absorbance spectra of the proteins at the concentration (0.3–1.5 mg mL⁻¹) in 10 mM HEPES, pH 7.4, from a wavelength of 600–250 nm with a Cary 100 spectrophotometer.

5.4.2. NQO1 and NQO1 mutated enzymes

NQO1, NQO1(V73W), NQO1(G123Q), and NQO1(Y127F‐T128D‐Y129F) were expressed as fusions of hexa‐histidine affinity tagged calmodulin and the full coding sequence of NQO1, with a linker that includes a TEV protease cleavage site. Plasmids were transformed into a BL21(DE3) background for expression. Overnight cultures were used to inoculate 2 × 1 L of LB‐Amp that were grown for 2 h at 37°C and then induced by addition of isopropyl β‐D‐1‐thiogalactopyranoside (IPTG, 0.2 mM final concentration), followed by overnight incubation at 22.5°C. The cells were harvested and resuspended in 10 mM Tris–HCl, 1 mM EDTA, pH 7.5 and lysed using an Emulsiflex homogenizer. The homogenate was supplemented with NaCl (300 mM) and imidazole (10 mM). Cell debris was removed by centrifugation at 4°C and 100,000g for 1.25 h. The cell extract was applied to a 6 mL column of Ni‐NTA resin equilibrated with 10 mM Tris–HCl, 300 mM NaCl, and 10 mM imidazole, pH 7.5; unbound proteins was washed from the column with the equilibration buffer. Bound proteins were eluted from the column by increasing the imidazole concentration to 250 mM. The calmodulin fusion partner was removed by supplementing the buffer with 5 mM dithiothreitol (DTT) and 5 μg mg⁻¹ of TEV protease; complete digestion was confirmed by SDS‐PAGE. The digested sample was dialyzed against 10 mM Tris pH 7.5, 50 mM NaCl, 1 mM EDTA, and 5 mM mercaptoethanol to remove excess salts and small molecules in preparation for anion exchange chromatography. To remove calmodulin and other impurities, the dialyzed sample was applied to a 2 × 20 cm column of DEAE Sepharose, pre‐equilibrated with 10 mM Tris–HCl, 50 mM NaCl, 1 mM EDTA, and 5 mM mercaptoethanol, pH 7.5. Under these conditions, the NQO1 proteins flowed through the column while other proteins were retained.

5.5. Thermal shift assay

To examine the stability of the constructs, the melting temperature (T _M) of the proteins was determined with the Applied Biosystems Real‐Time PCR System using the QuantStudio™ 3 Real‐Time PCR System. The thermal shift assay works by monitoring the change in fluorescence of the Sypro Orange Dye™, with an excitation maximum of 472 nm and an emission maximum of 570 nm. The PCR system thermally denatures proteins by changing the temperature in a controlled manner. The fluorescence emission from Sypro Orange™ dye increases when it binds to the exposed hydrophobic regions of unfolded proteins. Proteins, 0.2–0.3 mg mL⁻¹ in 100 mM HEPES, pH 7.4 and 200 mM NaCl, 5 mM DTT, were incubated with 1x Sypro Orange Dye™ in a volume of 20 μL in a 96‐well plate. Unfolding of each protein was measured with at least four replicates. As negative controls, the fluorescence of dye only and buffer only were measured. As FAD excitation and emission overlap with Sypro Orange™, the change in fluorescence of FAD from the protein alone was measured as a control. There was no comparable signal from the FAD at the protein concentrations used, indicating that fluorescence from the FAD would not interfere with the measurements.

The 96‐well plate was incubated on ice before starting the thermal denaturation assay. To generate the melt curves, the temperature was increased to 25°C at a rate of 1.6°C s⁻¹ and then held at 25°C for 2 min. Then, fluorescence data collection was started by increasing the temperature to 99.0°C at a rate of 0.05°C s⁻¹. The temperature was then held at 99.0°C for 2 min, after which the assay was finished. The Protein Thermal Shift™ Software v1.0 was used to analyze the melt curve data.

5.6. Steady‐state kinetics

We used a previously described protocol to determine the enzymatic activity of all five ancestral proteins (Liao et al. 1962). For kinetic assays with BNAH and menadione, the stock BNAH and menadione solutions in methanol were prepared fresh every day and kept on ice during the duration of the kinetic experiments. The concentration of each reagent was determined spectroscopically using the extinction coefficient ε _355nm of 7220 M⁻¹·cm⁻¹ for BNAH and ε _333nm of 2450 M⁻¹·cm⁻¹ for menadione. The kinetic reactions were conducted by monitoring the decrease in absorbance at 355 nm using a Cary 100 spectrophotometer to measure the rate of oxidation of the reducing cosubstrate. Enzyme assays were carried out for 2 min in 1 mL volumes in 100 mM HEPES, pH 7.4 at 25°C under constant stirring with a small magnetic stir bar. Enzymatic activity was measured for a range of BNAH (5–500 μM) and menadione (2–100 μM) concentrations, using an appropriate protein concentration from 0.5 nM to 0.5 μM. Each reaction was carried out in triplicate or greater.

For the kinetic assays with NADH, the same protocol was used. Stock NADH was dissolved in 100 mM Tris–HCl, pH 8.0, and prepared fresh before use. NADH (10 μM–2 mM) oxidation was measured by monitoring the change in absorbance at a wavelength of either 360 nm for high NADH concentrations (ε _360nm of 4147 M⁻¹·cm⁻¹) or at 340 nm (ε _340nm of 6220 M⁻¹·cm⁻¹) for lower concentrations of NADH. For the kinetic assays with NRH, the cosubstrate was dissolved in 100 mM Tris–HCl, pH 8.0 and concentrations from 10 μM to 2 mM were used in the assays. The reaction progress was monitored by measuring the change in absorbance at a wavelength of 360 nm (ε _360nm of 4360 M⁻¹·cm⁻¹) or at a wavelength of 380 nm (ε _380nm of 1070 M⁻¹·cm⁻¹). The initial rates were globally fit to a ping‐pong bi‐substrate Michaelis Menten equation (Equation (1)) to obtain K _M values for both substrates and K _I for menadione using Prism (www.graphpad.com),

$$\text{Rate}=\frac{k_{\mathrm{cat}}\text{[NADH][menadione]}}{K_{\text{NADH}}\text{[menadione]}\left(1+\frac{\text{[menadione]}}{K_{I,\text{menadione}}}\right)+K_{\text{menadione}}\text{[NADH]}+\text{[NADH][menadione]}}.$$

AUTHOR CONTRIBUTIONS

Faiza Islam: Conceptualization; investigation; methodology; validation; writing – review and editing; writing – original draft. Nicoletta Basilone: Conceptualization; methodology; validation; writing – review and editing; writing – original draft. Vania Yoo: Validation; investigation. Eric Ball: Conceptualization; investigation; validation; methodology; writing – review and editing. Brian Shilton: Conceptualization; investigation; funding acquisition; writing – original draft; writing – review and editing; validation; visualization; methodology; project administration; supervision; resources.

Supporting information

Figure S1. Phylogenetic tree for evolution of NQO1 and NQO2 after gene duplication. A multiple sequence alignment of 172 NQO1 and NQO2 sequences along with outgroup quinone reductase sequences from five archaea and three mollusks was used with Phylobot (Hanson‐Smith and Johnson ²⁰¹⁶) to construct the phylogenetic tree. Node points used for ancestral sequence reconstruction are indicated: CA is the common ancestor that underwent gene duplication to give rise to CA1NQO1 and CA1NQO2, the first common ancestors for the NQO1 and NQO2 evolutionary trajectories. CA2NQO1 and CA2NQO2 are the common ancestors at the top of the phylogenetic trees for the amniotes. Proteins identified as NQO1 or NQO2 are indicated by “Q1” or “Q2,” respectively, followed by the organism's name. Similarly, the ray‐finned fish have an NQO1‐like enzyme that is identified as “F1.”

Data S1. Supporting Information Tables S1–S5.

Data S2. Supporting Information Tables S6–S11.

Islam F, Basilone N, Yoo V, Ball E, Shilton B. Evolutionary analysis of Quinone Reductases 1 and 2 suggests that NQO2 evolved to function as a pseudoenzyme. Protein Science. 2024;33(12):e5234. 10.1002/pro.5234