Evidence that the metabolite repair enzyme NAD(P)HX epimerase has a moonlighting function

Abstract

NAD(P)H-hydrate epimerase (EC 5.1.99.6) is known to help repair NAD(P)H hydrates (NAD(P)HX), which are damage products existing as R and S epimers. The S epimer is reconverted to NAD(P)H by a dehydratase; the epimerase facilitates epimer interconversion. Epimerase deficiency in humans causes a lethal disorder attributed to NADHX accumulation. However, bioinformatic evidence suggest caution about this attribution by predicting that the epimerase has a second function connected to vitamin B₆ (pyridoxal 5'-phosphate and related compounds). Specifically, (i) the epimerase is fused to a B₆ salvage enzyme in plants, (ii) epimerase genes cluster on the chromosome with B₆-related genes in bacteria, and (iii) epimerase and B₆-related genes are coexpressed in yeast and Arabidopsis The predicted second function was explored in Escherichia coli, whose epimerase and dehydratase are fused and encoded by yjeF The putative NAD(P)HX epimerase active site has a conserved lysine residue (K192 in E. coli YjeF). Changing this residue to alanine cut in vitro epimerase activity by ≥95% but did not affect dehydratase activity. Mutant cells carrying the K192A mutation had essentially normal NAD(P)HX dehydratase activity and NAD(P)HX levels, showing that the mutation had little impact on NAD(P)HX repair in vivo However, these cells showed metabolome changes, particularly in amino acids, which exceeded those in cells lacking the entire yjeF gene. The K192A mutant cells also had reduced levels of 'free' (i.e. weakly bound or unbound) pyridoxal 5'-phosphate. These results provide circumstantial evidence that the epimerase has a metabolic function beyond NAD(P)HX repair and that this function involves vitamin B₆.

Keywords: AIBP; NAD(P)H hydrates; NAD(P)HX; NAXE; Vitamin B6; protein moonlighting.

Figure 1. NAD(P)H damage and repair reactions, and the neighborhoods of genes encoding the repair enzymes in bacteria

(A) Chemical and enzymatic damage reactions convert NAD(P)H to R and S forms of NAD(P)HX, which are reconverted to NAD(P)H by the repair reactions mediated by NAD(P)HX epimerase and NAD(P)HX dehydratase. Bacterial NAD(P)HX dehydratase uses ADP instead of ATP [5]. Both forms of NAD(P)HX can spontaneously cyclize. The enzymatic formation of NAD(P)HX is due to a side reaction of glyceraldehyde 3-phosphate dehydrogenase [2]. (B) Representative cases of clustering on bacterial chromosomes of epimerase–dehydratase fusion genes (yjeF) or standalone epimerase genes (epi) with genes predicted to encode the pyridoxal 5′-phosphate (PLP)-dependent enzymes alanine racemase (alr) and glutamate decarboxylase (gadA), or the PLP biosynthesis enzyme pyridoxine 5'-phosphate synthase (pdxJ). Clustering in seven phyla with holo-acyl carrier protein synthase genes (a) is also shown; its significance is unclear but may relate to the major consumption of NADPH in lipid synthesis. Genes with functions that are unknown or unrelated to NAD(P)H or PLP are colored gray. Phyla to which species belong are shown in parentheses.

Figure 2. Activities of recombinant E. coli native and mutant (K192A) YjeF proteins

(A) Assays (100 µl) contained 25 mM Tris-HCl, pH 8.0, 5 mM KCl, 2 mM MgCl₂, 0.1 mg mL⁻¹ BSA, 40 µM NADHX (containing approximately equal amounts of S and R forms), and 1 mM ADP. Reactions were started by adding 2 µg of native YjeF and absorbance was monitored at 340 nm at 22°C. (B) Assays were performed as above except that reactions were started by adding 2 µg of K192A YjeF (closed circles). In separate assays, 2 µg of Arabidopsis NAD(P)HX epimerase domain protein (E) was added at 4 min (open diamonds). Spontaneous epimerization of NAD(P)HX was undetectable in the conditions and time frame of the assay. Data are means of three replicates; S.E. was <0.17 nmol NADH formed for all data points.

Figure 3. NAD(P)HX accumulation in E. coli wild-type, K192A, and ΔyjeF cells

Cell extracts were analyzed by liquid chromatography–mass spectrometry with multiple reaction monitoring. Bars represent peak intensity of various forms of NADHX (A) or NADPHX (B). Data are means and S.E. of three replicates. Asterisks denote NADHX or NADPHX forms that are significantly different by ANOVA (P<0.05, Tukey HSD test) relative to wild-type. The data were log₁₀ transformed for ANOVA and are plotted on a logarithmic scale.

Figure 4. Partial least squares discriminant analysis (PLS-DA) of untargeted GC–TOF data

PLS-DA analysis was performed on three independent metabolomics datasets (A–C) consisting of E. coli native (black circles), yjeF deletant (blue triangles), and K192A point mutant (red squares) cells. Gray, blue, and red ellipses display the 95% confidence interval of the respective data points.

Figure 5. Metabolites that are present at different levels in yjeF deletant or K192A cells

Metabolites whose levels changed significantly (P<0.05; t test) and 2-fold or more in yjeF deletant and/or K192A point mutant cells relative to wild-type are shown for three independent metabolomics datasets (A–C). Amino acids and related compounds are marked + if they appear in one dataset, ++ if they appear in two, and +++ if they appear in all three. The numbered metabolites are routinely observed in metabolomics datasets but have not been positively identified; numbers are BinBase identifiers, which are randomly assigned and do not reflect any structural information [32].

Figure 6. Vitamin B₆ accumulation in E. coli wild-type, ΔyjeF, K192A, and YjeF overexpressing ΔyjeF cells

(A) Total vitamin B₆ levels. Data are means and S.E. of three replicates; PMP, pyridoxamine 5′-phosphate. Other B₆ forms (pyridoxine 5′-phosphate, pyridoxal, pyridoxamine, and pyridoxine) were present in trace amounts below the limit of quantification. Data are means and S.E. of three replicates. Asterisks denote levels that are significantly different (P<0.05; ANOVA) relative to wild-type. YjeF overexpressing ΔyjeF cells grew ∼30% more slowly than wild-type cells under these conditions. (B) Free PLP levels. Data are means and S.E. of duplicates. Asterisks denote levels that are significantly different (P<0.05; ANOVA) relative to wild-type.