Enzymes of mannitol metabolism in the human pathogenic fungus Aspergillus fumigatus – kinetic properties of mannitol-1-phosphate 5-dehydrogenase and mannitol 2-dehydrogenase, and their physiological implications
Institute of Biotechnology and Biochemical Engineering, Graz University of Technology, Austria
B. Nidetzky, Institute of Biotechnology and Biochemical Engineering, Graz University of Technology, Petersgasse 12/I, A-8010 Graz, Austria Fax: +43 316 873 8434 Tel: +43 316 873 8400 E-mail: bernd.nidetzky@TUGraz.at
The human pathogenic fungus Aspergillus fumigatus accumulates large amounts of intracellular mannitol to enhance its resistance against defense strategies of the infected host. To explore their currently unknown roles in mannitol metabolism, we studied A. fumigatus mannitol-1-phosphate 5-dehydrogenase (AfM1PDH) and mannitol 2-dehydrogenase (AfM2DH), each recombinantly produced in Escherichia coli, and performed a detailed steady-state kinetic characterization of the two enzymes at 25 °C and pH 7.1. Primary kinetic isotope effects resulting from deuteration of alcohol substrate or NADH showed that, for AfM1PDH, binding of d-mannitol 1-phosphate and NAD+ is random, whereas d-fructose 6-phosphate binds only after NADH has bound to the enzyme. Binding of substrate and NAD(H) by AfM2DH is random for both d-mannitol oxidation and d-fructose reduction. Hydride transfer is rate-determining for d-mannitol 1-phosphate oxidation by AfM1PDH (kcat = 10.6 s−1) as well as d-fructose reduction by AfM2DH (kcat = 94 s−1). Product release steps control the maximum rates in the other direction of the two enzymatic reactions. Free energy profiles for the enzymatic reaction under physiological boundary conditions suggest that AfM1PDH primarily functions as a d-fructose-6-phosphate reductase, whereas AfM2DH acts in d-mannitol oxidation, thus establishing distinct routes for production and mobilization of mannitol in A. fumigatus. ATP, ADP and AMP do not affect the activity of AfM1PDH, suggesting the absence of flux control by cellular energy charge at the level of d-fructose 6-phosphate reduction. AfM1PDH is remarkably resistant to inactivation by heat (half-life at 40 °C of 20 h), consistent with the idea that formation of mannitol is an essential component of the temperature stress response of A. fumigatus. Inhibition of AfM1PDH might be a useful target for therapy of A. fumigatus infections.
mannitol 2-dehydrogenase from Pseudomonas fluorescens
The six-carbon polyol d-mannitol (Man-ol) is ubiquitous throughout the fungal kingdom, and one of the most abundant sugar alcohols in nature. Aside from other physiological functions that have been ascribed to it, intracellular accumulation of Man-ol is a widespread mechanism by which fungi cope with different forms of external stress, including high temperature and oxidative stress . In parasitic fungi, the improved stress resistance resulting from accumulated Man-ol appears to confer a substantially enhanced ability to deal with defense strategies of the infected host [2–6]. The human pathogen Aspergillus fumigatus, which is the most common agent causing invasive aspergillosis in immunosuppressed patients, produces enough Man-ol to raise the serum Man-ol level of the infected animal [7,8]. High thermotolerance is a major component of A. fumigatus pathogenicity, and involves Man-ol indirectly . Decreased susceptibility to reactive oxygen species produced by human phagocytes in response to the microbial infection is yet another virulence factor of A. fumigatus, and includes direct participation of Man-ol as a free radical scavenger [10,11]. The inhibition of fungal Man-ol production therefore presents a potential target in advanced therapies for A. fumigatus infections. Unfortunately, little is currently known about the enzyme system of Man-ol metabolism in A. fumigatus.
A number of studies suggest that two different metabolic paths contribute to the biosynthesis of Man-ol in fungi (Scheme 1; reviewed in ). Reduction of d-fructose 6-phosphate (Fru6P) by an NADH-dependent mannitol-1-phosphate 5-dehydrogenase (M1PDH; EC 220.127.116.11) gives d-mannitol 1-phosphate (Man-ol1P), which, upon hydrolysis of phosphate ester by a phosphatase (EC 18.104.22.168), yields Man-ol (path 1). In the alternative route (path 2), Fru6P is converted to d-fructose (Fru), which, in turn, is reduced to Man-ol by NADPH-dependent or NADH-dependent mannitol 2-dehydrogenase (M2DH; EC 22.214.171.124 or 126.96.36.199). Mobilization of Man-ol occurs through quasireversal of path 2, whereby an NADP+-dependent or NAD+-dependent M2DH produces Fru, which is then phosphorylated from ATP by hexokinase (EC 188.8.131.52) to regenerate Fru6P . It is currently not clear whether stockpiled mannitol could also be utilized by going through the steps of path 1 in reverse [13–16], whereby Man-ol would have to become phosphorylated by a suitable kinase.
M1PDH and NAD+-dependent M2DH have recently been characterized from A. fumigatus [17,18]. On the basis of sequence similarity, both enzymes were classified as members of the polyol-specific long-chain dehydrogenase/reductase (PSLDR) family . The PSLDRs constitute a diverse group of NAD(P)-dependent oxidoreductases that are widespread among microorganisms but are lacking in humans. Their activity is not dependent on a metal cofactor and is exclusively targeted towards polyol/ketose substrates [12,19]. In a family-wide categorization of the PSLDRs, M1PDH and M2DH were clearly separated from each other, showing only a distant evolutionary relationship [12,19]. Structurally, PSLDRs are composed of two separate domains. The N-terminal domain adopts an expanded Rossmann fold, and provides the residues for binding the coenzyme. The C-terminal domain has a largely α-helical structure, and serves mainly in substrate binding. The active site is located in the interdomain cleft, and contains a highly conserved tetrad of residues (Lys/Asn/Asn/His), whereby Lys is the catalytic acid–base of the enzymatic reaction [20,21]. A recent study has demonstrated, at the level of both gene transcript and translated protein, that A. fumigatus M1PDH (AfM1PDH) becomes strongly upregulated during heat shock . Enhanced biosynthesis of Man-ol via AfM1PDH-catalyzed conversion of Fru6P might contribute extra robustness to A. fumigatus under high-temperature conditions.
In this study, an enzymological approach was chosen to examine the roles of AfM1PDH and A. fumigatus M2DH (AfM2DH) in the metabolism of Man-ol in A. fumigatus. A detailed steady-state kinetic characterization was performed with each enzyme, providing the basis for the construction of free energy profiles for the enzymatic reactions under physiological boundary conditions as defined from the literature. The results provide clear assignment of a biosynthetic function to AfM1PDH (path 1), which behaves kinetically as a Fru6P reductase, whereas AfM2DH is essentially a Man-ol-oxidizing enzyme (path 2, backwards). The results of inhibition studies show that the activity of AfM1PDH would not be affected by changes in the levels of ATP, ADP and AMP, suggesting that flux through Fru6P reduction is not under direct control of the cellular energy charge. Considering the results of studies on the human pathogenic fungus Cryptococcus neoformans, showing that a low Man-ol-producing mutant strain was a 5000-fold less potent agent than the wild type [2,3], we propose that inhibition of AfM1PDH might be exploitable in the development of novel therapeutic strategies against A. fumigatus infection.
Substrate specificity of AfM1PDH and AfM2DH
Purified preparations of recombinant AfM1PDH and AfM2DH were assayed for activity in the directions of alcohol oxidation by NAD+ (pH 10.0) and carbonyl group reduction by NADH (pH 7.1), with a range of possible alternative substrates. Both enzymes showed only trace activity for utilization of NADP+. Catalytic efficiencies (in terms of kcat/KNADP) were more than three orders of magnitude below those obtained with NAD+ . AfM1PDH and AfM2DH contain a conserved Asp (Asp33 and Asp77, respectively) in their coenzyme-binding pockets that is known from previous studies of a related PSLDR, M2DH from Pseudomonas fluorescens (PsM2DH), to prevent accommodation of the 2′-phosphate group of NADP+ [20,22]. Reactions of A. fumigatus enzymes that are dependent on NADP+ or NADPH were therefore not further investigated.
Above a level of 1% activity with Man-ol1P (1.0 mm), AfM1PDH did not catalyze the oxidation of Man-ol, d-sorbitol, d-ribitol, xylitol, d-xylose, l-xylose, d-glucose, d-mannose, l-arabinose, d-arabinose, d-galactose, l-fucose, and d-lyxose. The enzyme was also inactive above a level of 1% activity with Fru6P (100 mm) for reduction of Fru, l-sorbose, d-xylulose, d-fructose 1,6-bisphosphate, d-glucose 6-phosphate (150 mm), and d-glucose 1-phosphate (150 mm). Therefore, AfM1PDH appears to be fairly specific for its natural pair of substrates, Man-ol1P and Fru6P. From the highly truncated activity with Man-ol and Fru, we conclude that the phosphate moiety in Man-ol1P and Fru6P is essential for substrate recognition and/or catalysis by AfM1PDH.
Alcohol oxidation by AfM2DH was assayed across the same series of substrates utilized above with AfM1PDH. l-Arabinitol, d-arabinitol, d-ribose, 2-deoxy-d-galactose and 2-deoxy-d-glucose were additionally tested as alcohol substrates. In the reduction direction, l-sorbose, d-xylulose and dihydroxyacetone were examined. Above the 1% level of activity with Man-ol and Fru, two polyols (d-arabinitol and d-sorbitol) and two ketoses (d-xylulose and l-sorbose) gave significant conversion rates. The known regioselectivity of M2DH in the oxidation of polyols , indicated in Table 1, allows assignment of d-xylulose and l-sorbose as the products of oxidation of d-arabinitol and d-sorbitol, respectively. Table 1 summarizes the results of kinetic parameter determination for polyol–ketose substrate pairs of AfM2DH. Structural comparison of polyols that are reactive for NAD+-dependent oxidation by AfM2DH with those that are not substrates of the enzyme (Fig. S1) reveals that a d-arabo configuration is required for a polyol to become reactive. The C2 (R) configuration (Man-ol) is preferred over the C2 (S) configuration (d-sorbitol). A model of AfM2DH was generated with the crystal structure of PsM2DH as the template (Fig. S2) . Residues contributing to the substrate-binding site of PsM2DH are fully conserved in AfM2DH, explaining the observed substrate specificity of the A. fumigatus enzyme [12,20]. It can be assumed from the way in which Man-ol interacts with the substrate-binding site of PsM2DH in the crystal structure that ketose substrates must bind in their open-chain free-carbonyl form . The values of Km and kcat/Km in Table 1 are appropriately corrected for the available proportion of reactive ketose substrate present.
Table 1. Kinetic constants of AfM2DH for reactions with different polyol and ketose substrates. Km and kcat/Km data for carbonyl substrates are corrected for the available proportion of open-chain free-carbonyl forms present in aqueous solution (Fru, ∼ 1% ; d-xylulose, ∼ 20% ; and l-sorbose, 0.2% ). Numbers in parentheses show values as measured. Ketose reductions, pH 7.1; polyol oxidations, pH 10.0. ND, not detectable (enzyme could not be saturated with l-sorbose).
86 ± 5
0.60 ± 0.07 (60 ± 7)
140 ± 20 (1.4 ± 0.2)
64 ± 1
1.7 ± 0.1 (8.3 ± 0.4)
39 ± 2 (7.7 ± 0.4)
1.1 ± 0.1 (0.0022 ± 0.0001)
212 ± 3
13 ± 1
17 ± 1
162 ± 5
163 ± 13
0.99 ± 0.08
60 ± 1
680 ± 30
0.088 ± 0.004
A full steady-state kinetic characterization of AfM1PDH and AfM2DH was carried out under physiological pH conditions. Lineweaver–Burk plots of initial-rate data (Fig. 1) gave a pattern of intersecting lines, for each enzyme and in both reaction directions, consistent with a kinetic mechanism in which substrate and coenzyme must bind to the enzyme to form a ternary complex prior to the release of the first product. Kinetic parameters were obtained from nonlinear least-squares fits of Eqn (3) or Eqn (4) to the experimental data. They are summarized in Table 2, and their internal consistency was verified with Haldane relationship analysis, comparing the kinetically determined reaction equilibrium constant (Keq) from Eqn (7) with those reported in the literature. Table 2 shows the useful agreement between previously published and calculated Keq values [23,24]. It is interesting that, in terms of Km, AfM1PDH binds Man-ol1P two orders of magnitude more tightly than AfM2DH binds Man-ol.
Table 2. Kinetic parameters of AfM1PDH and AfM2DH at 25 °C and pH 7.1. NA, not applicable.
a The dimensionless app Keq at pH 7.1 was calculated with the Haldane relationship (Eqn 7), using kinetic parameters from this Table. b The given Keq is pH-independent, and was obtained from the dimensionless app Keq at pH 7.1. c Experimentally determined, pH-independent equilibrium constant.
Inhibition of AfM1PDH and AfM2DH by components of the cellular energy charge
M1PDH from Escherichia coli (EcM1PDH) is evolutionary related to AfM1PDH by common membership of the family of PSLDRs [12,19]. It is strongly inhibited by ATP, which acts as a competitive inhibitor against NADH with a Ki of ∼ 60 μm . With a Ki of ∼ 800 μm, AMP binds one order of magnitude less strongly to EcM1PDH than does ATP . To examine possible regulation of the two A. fumigatus enzymes by components of the cellular energy charge (ATP, ADP, and AMP), we measured inhibition of AfM1PDH and AfM2DH by each of these adenine nucleotides. Figure S3 shows double-reciprocal plots of initial-rate data recorded at different concentrations of inhibitor. The observed inhibition was, in each case, best described by competitive binding of coenzyme and adenine nucleotide. Inhibition constants (KiEI) were obtained from nonlinear fits of Eqn (5) to the data. They are summarized in Table 3. Both AfM1PDH and AfM2DH were inhibited weakly by adenine nucleotides as compared with the inhibition of EcM1PDH by ATP.
Table 3. Inhibition of AfM1PDH and AfM2DH by adenine nucleotides at pH 7.1 and 25 °C. KiEI is a constant describing competitive inhibition of AMP, ADP or ATP against NAD+ or NADH. ND, not detectable.
4.6 ± 0.6
1.5 ± 0.1
2.9 ± 0.7
5.6 ± 0.5
5.6 ± 1.9
2.0 ± 0.2
4.8 ± 1.3
5.7 ± 0.4
6.5 ± 1.4
1.4 ± 0.1
Kinetic isotope effects (KIEs)
Primary KIEs resulting from deuterium substitution of the hydrogen atom undergoing hydride transfer from substrate (polyol oxidation) or coenzyme (ketose reduction) were determined for AfM1PDH and AfM2DH at pH 7.1. Initial-rate data recorded with protio and deuterio substrate or coenzyme were fitted with Eqn (6), and KIEs on kinetic parameters are shown in Table 4. A nomenclature is used whereby superscript D designates the KIE (e.g. Dkcat). A KIE greater than unity means that hydrogen → deuterium replacement caused slowing down of the enzymatic reaction analyzed. NADH-dependent reduction of Fru catalyzed by AfM2DH was characterized by substrate inhibition at high Fru concentrations (KiS = 2 ± 1 m). Substrate inhibition was observed irrespective of whether the NADH concentration used was saturating or limiting in the sub-Km range. With the use of (4S)-[2H]-NADH (NADD) instead of NADH, this substrate inhibition disappeared (Fig. S4), precluding the use of a single equation for calculation of the KIEs. We therefore obtained Dkcat from a direct comparison of kcat data derived from nonlinear fits of Eqn (2) and Eqn (1) with initial rates recorded with NADH and NADD, respectively and additionally by fitting Eqn (6) to the experimental data obtained below the occurrence of substrate inhibition. Dkcat for Fru reduction in Table 4 (2.0 ± 0.3) represents a mean value and standard deviation for three independent experiments evaluated in either of the ways described above. The KIE on kcat/Km for Fru was calculated from the kcat and Km values obtained with NADH and NADD. The KIE on kcat/Km for coenzyme was calculated as the Dkcat/DKm ratio, where the value of DKm was obtained by dividing Km data for reactions with NADH and NADD. Note that Km (NADH) was invariant (± 15%) across a wide range of Fru concentrations (40–800 mm), and so the choice of the constant level of Fru was not critical for determination of DKm (1.23 ± 0.15). The KIE data in Table 4 are instrumental in delineating the kinetic mechanism of AfM1PDH and AfM2DH, as described in the Discussion.
Table 4. KIEs for reactions of AfM1PDH and AfM2DH at pH 7.1.
2.9 ± 0.2
2.4 ± 0.5
2.4 ± 0.4
1.5 ± 0.1
3.1 ± 0.4
0.8 ± 0.2
1.0 ± 0.1
1.2 ± 0.2
1.6 ± 0.2
2.0 ± 0.3
1.2 ± 0.1
1.6 ± 0.3
Free energy profile analysis for reactions catalyzed by AfM1PDH and AfM2DH
Interpretation of the kinetic properties of an enzyme in the context of its role in cellular metabolism relies on knowledge of the physiological concentrations of substrates, products, coenzymes and relevant effectors. We are not aware of a study reporting intracellular metabolite levels in A. fumigatus (except for an early study commenting on the cellular Man-ol content). However, data for the closely related fungus Aspergillus niger are available. Table 5 shows a list of metabolite concentrations calculated from values in the literature, thus defining plausible reaction conditions for AfM1PDH and AfM2DH in vivo. Because values for the intracellular concentration of Fru are not available for aspergilli, we approximated the level of Fru with the known intracellular Fru concentration of Rhizobium leguminosarum . With the assumption of the conditions in Table 5, kinetic constants from Table 2 (including the KiEI values for ATP, ADP and AMP from Table 3) were used to construct free energy profiles for the transformations catalyzed by AfM1PDH and AfM2DH. These free energy profiles (Fig. 2) show that both enzymes operate under nonequilibrium reaction conditions that involve a substantial thermodynamic driving force for reduction of Fru6P and oxidation of Man-ol. Considering the uncertainty in the in vivo levels of NADH, Fru and Man-ol1P used, we performed a sensitivity analysis in which the effects of changes in intracellular reactant concentrations on the thermodynamic boundary conditions for the action of AfM1PDH and AfM2DH were examined. The allowed concentration ranges were comprehensive: 5–150 μm NADH; 0.01–10 mm Fru; and 10–400 μm Man-ol1P. The results (see the shaded area in Fig. 2) indicate that the overall conclusion of this work, that the physiological direction of the reaction of AfM1PDH is Fru6P reduction, whereas that of AfM2DH is Man-ol oxidation, was not affected by the assumed variation in the reactant concentrations. Equations (10) and (11) are (simplified) kinetic expressions that can be used to estimate the net direction of the enzymatic reaction (knet = oxidation – reduction) with the given substrate and product concentrations. Applying the concentrations from Table 5 to Eqns (10, 11), we find, for AfM1PDH, that the direction parameter knet is negative or, in other words, Fru6P reduction is preferred. With AfM2DH, by contrast, knet is positive, indicating that the reaction proceeds in the direction of Man-ol oxidation.
Table 5. Internal metabolite concentrations from the literature.
a Calculated from μmol·g−1 dry mycelium with application of an intracellular volume of 1.2 mL·g−1 dry weight as determined for the mycelium of A. niger .
The literature suggests that the heat stress response of A. fumigatus involves upregulated production of M1PDH . Because the function of AfM1PDH at elevated temperatures might require pronounced resistance of the enzyme to inactivation by heat, we recorded time courses of irreversible loss of activity at different temperatures, and use half-life times (τH), calculated from these measurements (Fig. S5), as parameters for stability. The stability of AfM2DH was analyzed in the same way, and the results for both enzymes are summarized in Table 6. AfM1PDH is much more stable than AfM2DH, about two orders of magnitude in terms of τH at 30 °C. The stability of AfM1PDH was hardly affected by increasing the enzyme concentration in the assay from 0.006 mg·mL−1 to 0.23 mg·mL−1, whereas a comparable change in concentration for AfM2DH (0.003 mg·mL−1 to 0.5 mg·mL−1) resulted in a substantial (approximately six-fold) increase in τH. Further mechanistic analysis of AfM1PDH and AfM2DH inactivation was beyond the scope of this work. AfM1PDH displays remarkable stability at 40 °C and even at 50 °C, so it can be considered to be a thermotolerant enzyme, fitting with the thermotolerance of the organism. Interestingly, M1PDHs from aspergilli (A. niger and Aspergillus parasiticus ) that are less resistant to high temperature than A. fumigatus have stabilities (at 30 °C) about one order of magnitude below the τH of AfM1PDH.
Table 6. Thermal stability of AfM1PDH and AfM2DH; τH is the half-life of the enzyme under the indicated conditions.
τH (h) (0.23 mg·mL−1)a
τH (h) (6 μg·mL−1)a
τH (h) (0.5 mg·mL−1)a
τH (h) (3 μg·mL−1)a
a Protein concentration used.
43 ± 4
31 ± 1
93 ± 7
15 ± 1
23 ± 1
18 ± 2
3.6 ± 0.2
0.64 ± 0.03
0.26 ± 0.03
0.16 ± 0.02
0.42 ± 0.04
0.06 ± 0.01
Kinetic mechanism of AfM1PDH and AfM2DH
The theory developed by Cook and Cleland is used to deduce the kinetic mechanism of AfM1PDH and AfM2DH from KIE data in Table 4 . The pattern of KIEs observed for AfM1PDH, in which Dkcat/KmNADH was not different from unity within the limits of error, whereas Dkcat/KmFru6P had a large value of ∼ 3, indicates that NADH binds to the enzyme prior to binding of Fru6P. The absence of a KIE on kcat/Km for the substrate binding first is a clear requirement of the strictly ordered kinetic mechanism, because, at saturating concentrations of the substrate which is added second, the commitment to catalysis becomes infinite, and so the KIE is completely suppressed. It follows from the additional sets of KIE data in Table 4, where Dkcat/Km values for substrate and coenzyme are both greater than unity, that binding of Man-ol1P and NAD+ by AfM1PDH is not ordered, and that there is also randomness in the binding of substrate and coenzyme in each direction of the reaction catalyzed by AfM2DH. By way of comparison, an earlier study on M1PDH from A. niger also reported random binding of NAD+ and Man-ol1P . It is reasonable to assume that random binding of reactants by AfM1PDH and AfM2DH occurs in rapid equilibrium, and the absence of curvature in Lineweaver–Burk plots (Fig. 1) supports this notion. The parameter α in Table 2 is therefore interpreted as a substrate–coenzyme interaction coefficient, for which a value greater than unity indicates that binding of one reactant weakens the affinity of the enzyme for binding of the other reactant. Kcoenzyme and Ksubstrate are dissociation constants for binary enzyme complexes with coenzyme and substrate, respectively. KiNADH in Table 2 is the dissociation constant of AfM1PDH–NADH, whereas KmNADH represents an apparent binding constant. Scheme 2 summarizes the proposed kinetic mechanisms of AfM1PDH and AfM2DH.
A value of Dkcat/Km well above unity implies that the isotope-sensitive step of hydride transfer contributes significantly to rate limitation for the sequence of reaction steps included in the kcat/Km analyzed. For example, kcat/KmFru6P involves all steps from binding of Fru6P to AfM1PDH–NADH up to release of the first product, Man-ol1P or NAD+. In random bireactant systems, kcat/Km stands for reaction of the variable reactant with the corresponding binary enzyme–substrate complex. Inspection of the KIE data in Table 4 reveals that hydride transfer is partly rate-determining in either direction of each of the two enzymatic reactions. Comparison of KIEs on kcat and kcat/Km distinguishes datasets in which Dkcat is smaller than Dkcat/Km from others in which Dkcat roughly equals Dkcat/Km. The first case (Dkcat < Dkcat/Km) indicates that, under kcat conditions where the concentrations of coenzyme and substrate are both saturating, a reaction step not included in kcat/Km, presumably release of the second product, is partly (Dkcat > 1) or completely (Dkcat = 1) rate-determining overall. This applies to Fru6P reduction by AfM1PDH as well as Man-ol oxidation by AfM2DH. The second case (Dkcat ≈ Dkcat/Km > 1) indicates that hydride transfer is rate-determining for the overall enzymatic reaction and applies to Man-ol1P oxidation by AfM1PDH as well as Fru reduction by AfM2DH.
Now, considering a kinetic scenario for AfM1PDH in which kcat for Man-ol1P oxidation is governed by hydride transfer, whereas kcat for Fru6P reduction is partly limited by product dissociation, it is worth remarking that the reduction kcat exceeds the oxidation kcat by a factor of 12 (Table 2). These kcat conditions imply that chemical reaction of AfM1PDH in the reduction direction proceeds much faster than the corresponding chemical reaction in the oxidation direction. Under physiological pH conditions, therefore, AfM1PDH shows a clear preference for catalysis in the reduction direction, so the enzyme may be considered to be a Fru6P reductase. In AfM2DH, by contrast, the turnover for Fru reduction (kcat = 94 s−1) is limited by chemical transformation, whereas the kcat of 14.2 s−1 for Man-ol oxidation reflects slow product release. With the reasonable assumption that complete suppression of the KIE on the oxidation kcat requires product release to be minimally about 10 times slower than the hydride transfer, chemical transformation in oxidation by AfM2DH should occur with a rate constant of 142 s−1 or higher. In comparison with AfM1PDH, therefore, the kinetic properties of AfM2DH at pH 7.1 resemble much more those expected from a true dehydrogenase acting in the direction of NAD+-dependent alcohol oxidation (see later).
Inhibition by ketose substrate during NADH-dependent reduction of Fru and its absence under conditions in which NADD is used (Fig. S4) is plausibly explained by an expanded random kinetic mechanism of AfM2DH (Scheme 2), involving an abortive enzyme–NAD+–Fru complex, which releases NAD+ at a rate slow enough to partially limit the overall reaction rate. KIE data indicating that hydride transfer is rate-determining for reduction of Fru by AfM2DH suggest that the relative amount of enzyme–NAD+ available for binding of Fru at the steady state cannot be very high. However, slowing the reaction by using NADD in place of NADH will further restrict the availability of enzyme–NAD+, explaining the complete lack of substrate inhibition under these conditions.
Proposed function of AfM1PDH and AfM2DH in mannitol metabolism
The results of free energy profile analysis strongly support the suggestion that Fru6P reduction is the preferred direction of catalytic action of AfM1PDH in vivo, implying a physiological function of the enzyme in Man-ol biosynthesis via path 1 of Scheme 1. The proposed role of AfM1PDH is in good agreement with evidence from M1PDH gene disruption studies in other fungi, showing that Δm1pdh mutants accumulate no Man-ol (A. niger mycelium)  or have a five-fold to 10-fold decreased Man-ol content (Alternaria alternata and Phaeosphaeria nodorum) in the mycelium [15,31]. Furthermore, A. niger undergoing sporulation displayed 4.5-fold enhanced production of Man-ol as compared with nonsporulating mycelium, and this change was correlated with a similar, about six-fold, increase in the level of M1PDH activity . We also show in this work that AfM1PDH activity is not sensitive to submillimolar alterations in the levels of ATP, ADP or AMP, indicating that, in contrast to E. coli, where inhibition by ATP is a probable mechanism of downregulation of M1PDH activity for Fru6P reduction , in A. fumigatus the cellular energy charge exercises no control at the protein level over the rate of Fru6P conversion into Man-ol1P. The results of transcriptomic and proteomic analysis of the heat shock response in A. fumigatus resulting from a shift in growth temperature from 30 °C to 48 °C suggest that regulation of AfM1PDH activity is achieved at the level of enzyme synthesis . The marked resistance of isolated AfM1PDH to inactivation by temperatures promoting the heat shock (40 °C and 50 °C) is consistent with a possible role of the enzyme in conferring thermotolerance to the fungus via enhanced Man-ol production under temperature stress conditions.
Considering the reactant concentrations in Table 5, NAD+-dependent oxidation of Man-ol proceeds thermodynamically downhill. From its kinetic properties (Tables 1 and 2), AfM2DH appears to be well primed for catalysis for mobilization of Man-ol (not its synthesis) under these conditions. The enzyme would be almost saturated with both alcohol substrate and NAD+, whereas the assumed intracellular concentration of Fru is about two orders of magnitude below the apparent Km of 40 mm. A physiological role of AfM2DH in the utilization of intracellular Man-ol via reversal of path 2 in Scheme 1 is therefore proposed.
In light of the suggested interplay between AfM1PDH and AfM2DH, it is interesting that deletion of the m1pdh gene in some fungi resulted in poor growth (A. alternata)  or the absence thereof (P. nodorum)  on Man-ol as carbon source. The corresponding Δm2dh mutants, however, showed substantial growth under these conditions [15,31]. It is possible, therefore, that M1PDH has an additional physiological function in the catabolism of external Man-ol once the substrate has become phosphorylated during or after uptake into the cell. However, a clear requirement for AfM1PDH to change directional preference in catalytic action would be that levels of intracellular metabolites undergo a large shift from the situation portrayed in Table 5 to another one that favors Man-ol1P, NAD+ or both.
Are oxidoreductases other than AfM1PDH and AfM2DH involved in the metabolism of mannitol by A. fumigatus? We searched the genome of the organism, and identified an ORF whose translated product is 35.2% identical in amino acid sequence to the M2DH from Agaricus bisporus (AbM2DH) . This putative protein of A. fumigatus is currently annotated as Sou1-like sorbitol/xylulose reductase (UniProt/TrEMBL entry Q4WZX5). In contrast to AfM2DH, AbM2DH is an NADP+-dependent enzyme. From its sequence and three-dimensional structure, AbM2DH is classified as member of the short-chain dehydrogenase/reductase superfamily of proteins and enzymes, and shows no significant evolutionary relationship with AfM2DH. Now, the tentative existence of an NADP+-dependent M2DH in A. fumigatus made it necessary to examine the possibility that Man-ol is produced from Fru by reduction with NADPH. However, assuming a value of 0.62 for the ratio of intracellular concentrations of NADPH and NADP+ (data from A. niger ) and applying the values for the in vivo levels of Man-ol and Fru from Table 5, we reach the immediate conclusion that reduction of Fru by NADPH is not thermodynamically feasible under these conditions. In other words, AfM1PDH is the key enzyme for Man-ol biosynthesis in A. fumigatus, irrespective of a possible multiplicity of NADP+/NAD+-dependent M2DH activities in the organism. A side remark in this respect is that gene expression and gene disruption studies in P. nodorum and A. niger could possibly have overlooked the existence of the NAD+-dependent M2DH (P. nodorum, Q0UEB6; A. niger, A2QGA1) [15,33].
Features of AfM1PDH structure and function that might be exploited for inhibition
Inhibition of the fungal biosynthesis of Man-ol is a promising strategy for development of new therapies against infection by A. fumigatus. The results of this work show that antagonism of AfM1PDH presents a clear target for achieving the desired inhibition. Moreover, because the human genome does not encode proteins homologous to AfM1PDH or any other protein of the current PSLDR family, there is a good chance that compounds raised against AfM1PDH will not show substantial cross-reactivity with respect to binding of proteins of the human host. The high specificity of AfM1PDH for reactions with phosphorylated substrates could be a feature of recognition to be exploited in the design of selective substrate and/or transition state analogs. The N-terminal NAD+-binding domain of AfM1PDH could be another target for selective inhibition, given that, in PSLDRs, this domain adopts a somewhat unusual Rossmann fold that is distinct in many details from isofunctional domains in enzymes of other dehydrogenase/reductase families. Precedents from the literature, on lactate dehydrogenase  for example, are highly encouraging in showing that, by using inhibitors targeted towards the coenzyme-binding site, it may be possible to achieve inhibition that is not only selective for a particular enzyme type, as would be necessary for inhibition of AfM1PDH, but can even discriminate between the same enzyme from parasite and host.
Recombinant AfM1PDH and AfM2DH were produced in E. coli and purified as described recently [17,18]. Unless otherwise indicated, highly purified preparations of recombinant AfM1PDH and AfM2DH were used in all experiments. Enzymatic production of Man-ol1P and 5-[2H]-Man-ol1P was carried out with previously reported methods . d-Xylulose was synthesized by microbial oxidation of d-arabinitol, employing a previously described protocol . NADD was prepared with an enzymatic procedure that used glucose dehydrogenase from Bacillus megaterium and 1-[2H]-d-glucose. Deutero-NADH was purified by MonoQ anion exchange chromatography, as previously described [23,36,37]. The optical purity of NADD and its degree of deuteration (95 ± 1%) were analyzed by 1H-NMR and MS, respectively. 2-[2H]-Man-ol was prepared by enzymatic conversion of Fru, and was purified as described recently [38,39]. The isotopic purity of 2-[2H]Man-ol was determined by MS (> 99%). No residual Fru, NAD+ or NADH was detected by 13C-NMR. Man-ol, Fru, Fru6P, β-nicotinamide adenine dinucleotides (NAD+ and NADH) and adenine nucleotides (ATP, ADP and AMP) at a purity ≥ 95% were obtained from commercial sources.
Protein concentrations were determined with the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA, USA) referenced against known concentrations of BSA. Initial-rate data were collected with a DU800 spectrophotometer (Beckman Coulter, Fullerton, CA, USA) at 25 °C, and are based on the measurement of formation or depletion of NAD(P)H at 340 nm (εNADH = 6.22 cm−1·mm−1). Unless otherwise indicated, substrate screening was performed at a constant concentration of 300 mm. Tris/HCl buffer (100 mm; pH 7.1) was used for ketose reduction. Glycine/NaOH buffer (100 mm; pH 10.0) was used for polyol oxidation. Different pH conditions were chosen because alcohol oxidation by NAD+ generally proceeds best at high pH, whereas a lower pH value is normally suitable for carbonyl group reduction by NADH. The concentrations of NADH and NAD+ used in these assays were 0.2 and 2.0 mm, respectively. Apparent kinetic parameters (kcat and Km) were obtained for those compounds that had shown significant conversion rates in screening assays, using a threshold of 1% of the activity with the respective native substrate (Man-ol and Fru; Man-ol1P and Fru6P). By application of this criterion, the following substrates were selected for reaction with AfM2DH in the given concentration range: Fru (2–1000 mm), d-xylulose (0.5–450 mm), l-sorbose (20–1000 mm), Man-ol (1–400 mm), d-arabinitol (3–1300 mm) and d-sorbitol (30–1600 mm).
Tris/HCl buffer (100 mm; pH 7.1) was used in all further kinetic studies. Full kinetic characterization of AfM1PDH involved initial-rate measurements under conditions of various concentrations of Fru6P (0.17–8.7 mm) and Man-ol1P (0.02–2.3 mm) at several constant concentrations of NADH (0.0017–0.17 mm) and NAD+ (0.06–5.9 mm), respectively. Measurements performed with AfM2DH involved various concentrations of Fru (8.5–430 mm) and Man-ol (3–110 mm) at several constant concentrations of NADH (0.014–0.20 mm) and NAD+ (0.085–1.4 mm), respectively. Inhibition by adenine nucleotides (AMP, ADP and ATP) was analyzed by measuring initial rates under conditions in which the concentration of NADH (0.008–0.25 mm) or NAD+ (0.05–2.5 mm) was varied at several constant concentrations of the respective adenine nucleotide in the range 0.63–5.0 mm. The concentration of carbonyl or polyol substrates was constant and saturating.
Primary deuterium KIEs on apparent kinetic parameters of AfM1PDH and AfM2DH were obtained from a comparison of initial rates recorded with unlabeled or deuterium-labeled substrates or coenzymes. Oxidation of Man-ol1P and 5-[2H]-Man-ol1P was measured under conditions in which the concentration of NAD+ (0.08–8 mm) or Man-ol1P/5-[2H]-Man-ol1P (0.04–6.2 mm) was varied at a constant and saturating concentration of the respective other substrate (NAD+, 5.7 mm; Man-ol1P/5-[2H]-Man-ol1P, 1.0 mm). Fru6P reduction was measured under conditions in which the concentration of NADH/NADD (0.012–0.2 mm) or Fru6P (0.45–45 mm) was varied at a constant and saturating concentration of the respective other substrate (NADH/NADD, 0.2 mm; Fru6P, 45 mm). Likewise, the conditions used for determination of KIEs on kinetic parameters for AfM2DH were as follows. Oxidation: Man-ol/2-[2H]-Man-ol, 0.9–180 mm, and NAD+, 4 mm; NAD+, 0.08–4 mm, and Man-ol/2-[2H]-Man-ol, 260 mm. Reduction: Fru, 4–840 mm, and NADH/NADD, 0.25 mm; NADH/NADD, 0.002–0.2 mm, and Fru, 800 mm.
Kinetic parameters were obtained from a nonlinear fit of the appropriate equation to the data. Unweighted nonlinear least-squares regression analysis with sigma plot 9.0 (SYSTAT Software; San Jose, CA, USA) was used. In Eqns (1)-(6), v is the initial rate, kcat is the kinetic turnover number, E and S are the molar concentrations of enzyme and substrate, Km is an apparent Michaelis constant, and KiS is a substrate inhibition constant. E was obtained from the protein concentration, using molecular masses of 44.2 kDa and 57.6 kDa for AfM1PDH and AfM2DH, respectively [17,18]. Equation (3) implies ordered binding of substrates A and B in a bisubstrate reaction where KiA is the dissociation constant for A. Equation (4) is used for random bisubstrate kinetics, where KA and KB are dissociation constants for A and B, and α is a factor describing how bound A affects the binding of B. In Eqn (5), KiEI is a competitive inhibition constant, and I is the molar inhibitor concentration. Unless mentioned, KIEs were obtained by using Eqn (6) , where EV and EV/K are isotope effects minus 1 on kcat and kcat/Km, respectively. Fi is the fraction of deuterium in the labeled substrate.
Equation (7) is the Haldane relationship for an ordered bi-bi kinetic mechanism, where app Keq is the (kinetically determined) equilibrium constant of the reaction; kox and kred are kcat values for alcohol oxidation and ketose reduction; KiNADH and KiNAD are dissociation constants for NADH and NAD+; and KmRO and KmROH are Michaelis constants for ketose and polyol. Note: in the random bisubstrate mechanism, the expression α·KA·KB is used to replace KiA·KmB in Eqn (7). The combined effect of competitive inhibition by adenine nucleotides is accounted for by Eqn (8) or Eqn (9), where an asterisk indicates the apparent binding constant. Equations (10) and (11) are simplified expressions of enzyme directional preference (knet), based solely on catalytic efficiencies. Formally, they are derived from the complete rate equation for a rapid equilibrium random bireactant kinetic mechanism , by leaving out all terms in the denominator. Strictly, these equations would be applicable only for limiting reactant concentrations. However, they are qualitatively useful because, at given reactant concentrations, a positive value of knet indicates a preference for alcohol oxidation, whereas a negative value signifies a preference for ketose reduction. Note that the term α·KA·KB applies to a random mechanism, as in Eqn (10). When the mechanism is ordered, the α·KA·KB term is replaced by KiA·KmB, as in Eqn (11). Gibbs free energy (ΔG, kJ·mol−1) profiles for reactions catalyzed by AfM2DH and AfM1PDH at pH 7.1 were constructed with the use of Eqns (12)-(19). R is the gas constant (0.008314 kJ·mol−1·K−1), and T is the temperature (298.15 K). The parameter K in Eqn (12) is the mass action ratio calculated by applying intracellular concentrations of reactants. ΔGKeff is the difference between ΔG at Keq and ΔG at a given value of K. Calculation of ΔG values at ternary complexes (indicated by subscript) and the transition state (subscript TS) is shown in Eqns (16)–(19). k and h are the Boltzmann constant (1.38 × 10−26 kJ·K−1) and the Planck constant (6.63 × 10−37 kJ·s), respectively.
V. Pacher and K. Longus are thanked for expert technical assistance. We are grateful to M. Murkovic (Institute of Biochemistry, Graz University of Technology) and H. Weber (Institute of Organic Chemistry, Graz University of Technology) for MS and NMR measurements, respectively. Financial support from the Austrian Science Fund FWF (P18275-B09 to B. Nidetzky) is gratefully acknowledged.