Characterisation of the First Archaeal Mannonate Dehydratase from Thermoplasma acidophilum and Its Potential Role in the Catabolism of D-Mannose

: Mannonate dehydratases catalyse the dehydration reaction from mannonate to 2-keto-3-deoxygluconate as part of the hexuronic acid metabolism in bacteria. Bacterial mannonate dehydratases present in this gene cluster usually belong to the xylose isomerase-like superfamily, which have been the focus of structural, biochemical and physiological studies. Mannonate dehydratases from archaea have not been studied in detail. Here, we identiﬁed and characterised the ﬁrst archaeal mannonate dehydratase (TaManD) from the thermoacidophilic archaeon Thermoplasma acidophilum . The recombinant TaManD enzyme was optimally active at 65 ◦ C and showed high speciﬁcity towards D-mannonate and its lactone, D-mannono-1,4-lactone. The gene encoding for TaManD is located adjacent to a previously studied mannose-speciﬁc aldohexose dehydrogenase (AldT) in the genome of T. acidophilum . Using nuclear magnetic resonance (NMR) spectroscopy, we showed that the mannose-speciﬁc AldT produces the substrates for TaManD, demonstrating the possibility for an oxidative metabolism of mannose in T. acidophilum . Among previously studied mannonate dehydratases, TaManD showed closest homology to enzymes belonging to the xylose isomerase-like superfamily. Genetic analysis revealed that closely related mannonate dehydratases among archaea are not located in a hexuronate gene cluster like in bacteria, but next to putative aldohexose dehydrogenases, implying a different physiological role of mannonate dehydratases in those archaeal species.


Introduction
Mannonate dehydratases (EC 4.2.1.8) catalyse the conversion of mannonate to 2-keto-3-deoxygluconate (KDG) and have been studied as part of the hexuronic acid gene cluster in several organisms, such as Escherichia coli, Bacillus stearothermophilus, Bacillus subtilis and Erwinia chrysanthemi [1][2][3][4]. The hexuronate gene cluster encodes enzymes involved in the metabolism of glucuronate and galacturonate [5]. Glucuronate is a common sugar acid present in glucuronoxylan, a constituent of plant cell walls which can serve as the only carbon source for growth of some bacteria [3,6,7]. Glucuronate is also present in the mucus layer of mammals, providing a carbon source for anaerobic gut bacteria, such as E. coli [8,9]. As part of the hexuronate metabolism, mannonate dehydratase converts mannonate to KDG ( Figure 1A). In E. coli, and some species of Erwinia, KDG is phosphorylated and cleaved into pyruvate and 3-phosphoglycerate, which can be further metabolised in the tricarboxylic acid cycle, or the Entner-Doudoroff pathway [4,5].
Mannonate dehydratases are represented in different enzyme families, such as the xylose isomerase-like superfamily or the enolase superfamily. Mannonate dehydratases encoded in the bacterial hexuronate gene clusters usually belong to the xylose isomerase-like superfamily. Crystal structures have been solved for xylose isomerase-like mannonate dehydratases from Streptococcus suis, E. coli K12 and Enterococcus faecalis in native form and in complex with Mn 2+ ions [10,11]. His311 and Tyr325 in the binding pocket were identified as crucial for the activity of the mannonate dehydratases from Gram-positive bacteria, such as S. suis. However, in Gram-negative bacteria (e.g. E. coli K12), an additional inserted sequence in the binding pocket rendered the dehydratase less active [11]. Members of the enolase superfamily show a conserved structure and reaction mechanism, but differ in their physiological functions. Within their conserved barrel structure, mannonate dehydratases of this family share conserved ligand-binding sites for Mg 2+ , which are essential for the stabilisation of the enediolate intermediate [12]. Among mannonate dehydratases of the enolase superfamily, some representatives with diverse functions have been found, which are involved in the catabolism of sugar acids other than glucuronate or galacturonate [13]. Several crystal structures have been solved for mannonate dehydratases from the enolase superfamily, including enzymes from Chromohalobacter salexigens and Novosphingobium aromaticivorans [12,14]. However, to the best of our knowledge, no archaeal mannonate dehydratase has been investigated so far.
Here, we used genomic analysis to identify the first functional mannonate dehydratase from the thermoacidophilic archaeon Thermoplasma acidophilum (TaManD). The archaeal mannonate dehydratase was recombinantly-expressed in E. coli, purified and functionally characterised. TaManD showed high amino acid sequence identity to bacterial mannonate dehydratases from the xylose isomerase-like superfamily. In the genome of T. acidophilum, the gene encoding for TaManD is located adjacent to an aldohexose dehydrogenase (AldT), which has been shown previously to catalyse the oxidation of mannose to mannonate with high specificity [15,16]. However, the physiological significance of the oxidation and its products were not investigated further. We identified the products of an AldT-mediated oxidation of D-mannose using NMR spectroscopy and confirmed that TaManD is able to convert the products to KDG. This demonstrates that in principle, a mannose metabolism based on AldT and ManD is possible in T. acidophilum ( Figure 1B).

Screening for a Functional Mannonate Dehydratase in T. acidophilum
Based on the previously identified mannose-specific AldT, we searched the genomic neighbourhood of its gene locus (Ta0754) for enzymes which could display activity towards mannonate. In the proximity of Ta0754, several genes are located that encode for hypothetical dehydratases, suggesting their potential to convert mannonate to KDG (Table S1). We identified the protein product of the Ta0753 gene as a functional mannonate dehydratase from T. acidophilum by heterologous expression in E. coli, and compared its properties and genomic context to other previously studied mannonate dehydratases.
Among previously characterised mannonate dehydratases, TaManD shares highest protein sequence identity (31.5%) with enzymes in the xylose isomerase-like superfamily, such as the mannonate dehydratase from S. suis [10]. Key amino acid residues for substrate binding (His311 and Tyr325) and binding of the cofactor Mn 2+ (Asp310, Cys237, His199 and His266) in S. suis are conserved in the amino acid sequence of TaManD. A much higher amino acid sequence identity (62.9%) is shared between TaManD and putative mannonate dehydratases from closely related archaeal species, Ferroplasma acidarmanus and Ferroplasma acidiphilum, suggesting an archaeal clade of mannonate dehydratases.
Phylogenetic analysis showed that mannonate dehydratases from T. acidophilum, F. acidarmanus and F. acidiphilum are more closely related to bacterial mannonate dehydratases from the xylose isomerase-like superfamily than to bacterial or archaeal mannonate dehydratases of the enolase superfamily ( Figure 2). Mannonate dehydratases of the enolase superfamily have a substantially different structure from those in the xylose isomerase-like superfamily, and therefore, are only distantly related to TaManD. Despite fulfilling a similar function, the dihydroxy-acid dehydratase from S. solfataricus, which belongs to the IlvD/EDD superfamily, is rather unrelated to the xylose isomerase-like and enolase superfamilies [17].
For bacterial xylose isomerase-like mannonate dehydratases, a physiological function of the enzyme in the catabolism of hexuronates has been demonstrated [18,19]. Although TaManD is annotated as uxuA mannonate dehydratase, which implies a role in the metabolism of hexuronates, it is not present in a classical hexuronate gene cluster known from bacteria such as E. coli, B. subtilis or B. stearothermophilus ( Figure 3) [2,3]. In contrast, the gene encoding for TaManD is located adjacent to AldT in the genome of T. acidophilum. Similarly, the two closely related putative mannonate dehydratases in F. acidarmanus and F. acidiphilum are also located adjacent to putative aldohexose dehydrogenases. Therefore, a different physiological role can be proposed for these mannonate dehydratases. In the following, we functionally characterise TaManD and show that AldT is able to produce the substrate needed for a subsequent conversion to KDG mediated by TaManD. sequence identity (31.5%) with enzymes in the xylose isomerase-like superfamily, such as the mannonate dehydratase from S. suis [10]. Key amino acid residues for substrate binding (His311 and Tyr325) and binding of the cofactor Mn 2+ (Asp310, Cys237, His199 and His266) in S. suis are conserved in the amino acid sequence of TaManD. A much higher amino acid sequence identity (62.9%) is shared between TaManD and putative mannonate dehydratases from closely related archaeal species, Ferroplasma acidarmanus and Ferroplasma acidiphilum, suggesting an archaeal clade of mannonate dehydratases.   Phylogenetic analysis showed that mannonate dehydratases from T. acidophilum, F. acidarmanus and F. acidiphilum are more closely related to bacterial mannonate dehydratases from the xylose isomerase-like superfamily than to bacterial or archaeal mannonate dehydratases of the enolase superfamily ( Figure 2). Mannonate dehydratases of the enolase superfamily have a substantially different structure from those in the xylose isomerase-like superfamily, and therefore, are only distantly related to TaManD. Despite fulfilling a similar function, the dihydroxy-acid dehydratase from S. solfataricus, which belongs to the IlvD/EDD superfamily, is rather unrelated to the xylose isomerase-like and enolase superfamilies [17].
For bacterial xylose isomerase-like mannonate dehydratases, a physiological function of the enzyme in the catabolism of hexuronates has been demonstrated [18,19]. Although TaManD is annotated as uxuA mannonate dehydratase, which implies a role in the metabolism of hexuronates, it is not present in a classical hexuronate gene cluster known from bacteria such as E. coli, B. subtilis or B. stearothermophilus ( Figure 3) [2,3]. In contrast, the gene encoding for TaManD is located adjacent to AldT in the genome of T. acidophilum. Similarly, the two closely related putative mannonate dehydratases in F. acidarmanus and F. acidiphilum are also located adjacent to putative aldohexose dehydrogenases. Therefore, a different physiological role can be proposed for these mannonate dehydratases. In the following, we functionally characterise TaManD and show that AldT is able to produce the substrate needed for a subsequent conversion to KDG mediated by TaManD.

Substrate conformation for TaManD activity
Before acquiring enzyme kinetics and studying the characteristics of TaManD, we investigated which substrate conformation is encountered by TaManD under physiological conditions. In aqueous solutions, free sugar acids coexist with their lactone in an equilibrium, which is defined by the stability of the lactone, the temperature and the pH of the solution. Many lactones hydrolyse

Substrate Conformation for TaManD Activity
Before acquiring enzyme kinetics and studying the characteristics of TaManD, we investigated which substrate conformation is encountered by TaManD under physiological conditions. In aqueous solutions, free sugar acids coexist with their lactone in an equilibrium, which is defined by the stability of the lactone, the temperature and the pH of the solution. Many lactones hydrolyse spontaneously in water, although several lactones, including D-arabinonolactone [20], L-rhamnonolactone [21] and D-xylonolactone [22], have been reported to be hydrolysed by lactonases. Mannonate is able to form a lactone by covalent bond formation between carbon 1 and carbon 5 (δ-lactone) or carbon 1 and carbon 4 (γ-lactone) ( Figure 4A). The equilibrium of the two different envelope forms has been studied by NMR spectroscopy and it was found that the γ-lactone is strongly favoured over the δ-lactone [23]. However, these studies do not describe equilibria between lactone and free acid form in a physiological buffer. In a cellular environment mannonate can either be produced by a mannonate oxidoreductase (UxuB) as part of the catabolism of hexuronic acids or by an aldohexose dehydrogenase, such as AldT, in a hypothetical oxidative mannose catabolism ( Figure 1). In order to study the conformation of the substrate for TaManD in a physiological environment, we first acquired decoupled 13 C NMR spectra for D-mannono-1,4-lactone in physiological buffer at pH 7, in its lactone form (incubated with HCl) and after hydrolysis to the free sugar acid (incubated with NaOH) ( Figure 4B). The spectra obtained indicated that the lactone and the free sugar acid can be distinguished by their chemical shifts. In the free sugar acid form, carbon 6 yields a chemical shift at 63.51 ppm, whereas in D-mannono-1,4-lactone, carbon 6 displays a chemical shift at 63.02 ppm. D-mannono-1,4-lactone in buffer shows both chemical shifts, indicating that D-mannono-1,4-lactone and D-mannonate are present in an equilibrium at pH 7. Next, we oxidised isotope-labelled D-[1,6-13 C 2 ]mannose using recombinantly-expressed AldT and acquired 1D 13 C NMR spectra of the reaction product. The chemical shifts for carbon 6 of mannono-1,4-lactone and D-mannonate were used to analyse the reaction products of the mannose oxidation with AldT, as these were close together but well resolved and will have similar relaxation rates in both forms. The spectrum after partial AldT oxidation of D-[1,6-13 C 2 ]mannose ( Figure 4C) shows that equal quantities of D-mannono-1,4-lactone and D-mannonate are produced for the subsequent dehydration by TaManD.

Expression and Purification of TaManD
For the further biochemical characterisation, TaManD was expressed together with a tobacco etch virus (TEV) sequence, which allowed the proteolytic cleavage of the His-tag after the first metal affinity purification. In order to study the effect of bivalent metal cofactors on enzyme activity, cleavage of the His-tag from the recombinant enzyme was necessary, since cations (Ni 2+ , Mg 2+ ) might influence enzyme activity assays of a His-tagged enzyme. After His-tag cleavage, the recombinant enzyme showed a single band on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with an apparent mass of 38 kDa ( Figure S1). Peptide mass fingerprinting was performed with liquid chromatography-electrospray ionisation-tandem mass spectrometry (LC ESI MS/MS) and the National Center for Biotechnology Information (NCBI) database searches using the mascot software confirmed the identity of the purified protein as uxuA mannonate dehydratase from T. acidophilum. In addition, the apparent native molecular mass of the protein was estimated from size exclusion chromatography to be 225 kDa, indicating a hexameric structure of the protein.

Enzyme Characterisation
Initial tests showed that TaManD lost almost its complete activity after purification, suggesting that essential cofactors for activity were removed during the purification process. Accordingly, the effect of several additives was tested for their influence on the activity of TaManD with D-mannono-1,4-lactone. TaManD showed strongly enhanced activity in the presence of β-mercaptoethanol and the metal ions Mn 2+ , Ni 2+ , Mg 2+ , Ca 2+ and Co 2+ ( Figure 5). The highest activation (100%) was observed in the presence of β-mercaptoethanol, which was significantly higher than the activating bivalent cation Ni 2+ (71.5% of maximum enzyme activity, p-value: 0.0022). Mn 2+ , Mg 2+ and Co 2+ also activated TaManD and resulted in 50-60% increase in maximum activity when compared to the TaManD control (without any additive). No significant difference in activity was observed in the presence of Fe 2+ , Zn 2+ , Cu 2+ , EDTA, DTT and glutathione.
Catalysts 2019, 9, x FOR PEER REVIEW 6 of 14 oxidation with AldT, as these were close together but well resolved and will have similar relaxation rates in both forms. The spectrum after partial AldT oxidation of D-[1,6-13 C2]mannose ( Figure 4C) shows that equal quantities of D-mannono-1,4-lactone and D-mannonate are produced for the subsequent dehydration by TaManD.

Expression and purification of TaManD
For the further biochemical characterisation, TaManD was expressed together with a tobacco etch virus (TEV) sequence, which allowed the proteolytic cleavage of the His-tag after the first metal affinity purification. In order to study the effect of bivalent metal cofactors on enzyme activity, cleavage of the His-tag from the recombinant enzyme was necessary, since cations (Ni 2+ , Mg 2+ ) might influence enzyme activity assays of a His-tagged enzyme. After His-tag cleavage, the recombinant enzyme showed a single band on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with an apparent mass of 38 kDa ( Figure S1). Peptide mass fingerprinting was performed with liquid chromatography-electrospray ionisation-tandem mass spectrometry (LC ESI MS/MS) and the National Center for Biotechnology Information (NCBI) database searches using the mascot software confirmed the identity of the purified protein as uxuA mannonate dehydratase from T. acidophilum. In addition, the apparent native molecular mass of the protein was estimated from size exclusion chromatography to be 225 kDa, indicating a hexameric structure of the protein.

Enzyme characterisation
Initial tests showed that TaManD lost almost its complete activity after purification, suggesting that essential cofactors for activity were removed during the purification process. Accordingly, the effect of several additives was tested for their influence on the activity of TaManD with D-mannono-1,4-lactone. TaManD showed strongly enhanced activity in the presence of β-mercaptoethanol and the metal ions Mn 2+ , Ni 2+ , Mg 2+ , Ca 2+ and Co 2+ ( Figure 5). The highest activation (100%) was observed in the presence of β-mercaptoethanol, which was significantly higher than the activating bivalent cation Ni 2+ (71.5% of maximum enzyme activity, p-value: 0.0022). Mn 2+ , Mg 2+ and Co 2+ also activated TaManD and resulted in 50-60% increase in maximum activity when compared to the TaManD control (without any additive). No significant difference in activity was observed in the presence of Fe 2+ , Zn 2+ , Cu 2+ , EDTA, DTT and glutathione. Activity was then determined in reactions containing 10 mM D-mannono-1,4-lactone in 50 mM HEPES pH 7 incubated for 1 h at 55 ˚C before analysis with the semicarbazide assay. Activity is expressed in relation to the maximum enzyme activity. Fe 2+ + cys: Fe 2+ was prepared with 1 mM cysteine. DTT: dithioerythritol, EDTA: ethylendiaminetetraacetate, β-ME: β-mercaptoethanol.
TaManD was activated by similar bivalent ions and reducing factor, as previously studied bacterial mannonate dehydratases from the xylose isomerase-like superfamily. However, none of TaManD was activated by similar bivalent ions and reducing factor, as previously studied bacterial mannonate dehydratases from the xylose isomerase-like superfamily. However, none of the bacterial mannonate dehydratases described so far have been shown to have an equally strong activation with both metal ions, Co 2+ and Ni 2+ , as observed with TaManD. Early investigations of mannonate dehydratases focused on E. coli and showed different strengths in activation after incubation with bivalent cations [1,[24][25][26]. In summary, most studies showed that β-mercaptoethanol, Fe 2+ and Mn 2+ resulted in highest enzyme activity (90-100% of maximum enzyme activity). Incubation with other bivalent cations resulted in less activity compared to the enzyme's maximum activity (Co 2+ : 27-80%, Ni 2+ : 5-40% and Zn 2+ : 5-25%, depending on the study). More recently, crystal structures of xylose isomerase-like mannonate dehydratases from S. suis and E. coli have been solved and revealed the presence of primarily Mn 2+ and lower amounts of Mg 2+ , Ni 2+ and Zn 2+ in their binding sites [10,11].
T. acidophilum was originally isolated from a hot and acidic environment and accordingly this archaeon displays optimal temperature and pH for growth at 59 • C and pH 1-2, respectively [27]. The purified TaManD was active between 35 • C and 70 • C, with an optimal temperature for activity at 65 • C ( Figure 6A). Thermostability of the enzyme in the absence of substrate was studied at temperatures between 55 • C and 95 • C ( Figure 6B). The enzyme retains its full activity at 55 • C for at least an hour, whereas complete inactivation was observed at 75 • C, 85 • C and 95 • C within 90 min, 60 min and 15 min, respectively. The difference in the relatively high optimal temperature of the enzyme to the comparably low thermostability indicated that the enzyme is more prone to inactivation in the absence of substrate. Other enzymes obtained from this organism have been shown to display optimal temperatures in the range of 55 • C to 70 • C [15,[28][29][30]. and Zn 2+ in their binding sites [10,11].
T. acidophilum was originally isolated from a hot and acidic environment and accordingly this archaeon displays optimal temperature and pH for growth at 59 °C and pH 1-2, respectively [27]. The purified TaManD was active between 35 °C and 70 °C, with an optimal temperature for activity at 65 °C ( Figure 6A). Thermostability of the enzyme in the absence of substrate was studied at temperatures between 55 °C and 95 °C ( Figure 6B). The enzyme retains its full activity at 55 °C for at least an hour, whereas complete inactivation was observed at 75 °C, 85 °C and 95 °C within 90 min, 60 min and 15 min, respectively. The difference in the relatively high optimal temperature of the enzyme to the comparably low thermostability indicated that the enzyme is more prone to inactivation in the absence of substrate. Other enzymes obtained from this organism have been shown to display optimal temperatures in the range of 55 °C to 70 °C [15,[28][29][30]. Following the identification of the crucial metal cofactors required for TaManD activity and determination of its optimal reaction temperature, we acquired enzyme kinetic data for the purified enzyme. Considering the fact that under physiological conditions an equilibrium of both lactone and free sugar acid exists, the kinetic data was acquired in reactions with substrate provided in these two different forms. The substrate was either hydrolysed with NaOH to obtain free sugar acid, D-mannonate, or the lactone was prepared in buffer, equaling an equilibrium of lactone and free sugar acid. TaManD displayed a higher activity in reactions with D-mannonate compared to reactions with D-mannono-1,4-lactone prepared in buffer at pH 7 ( Table 1). The maximal velocity of the reaction, Vmax, with D-mannonate as substrate was slightly higher (24-27%) compared to reactions with the lactone prepared in buffer (p-value: 0.06), whereas no difference in the affinity between the two different forms of the substrate could be observed (Table 1, Figure S2). Table 1. TaManD kinetic data with D-mannonate (prepared by hydrolysis with NaOH) and with D-mannono-1,4-lactone prepared in buffer (pH 7). Non-linear fitting was performed ( Figure S2). Following the identification of the crucial metal cofactors required for TaManD activity and determination of its optimal reaction temperature, we acquired enzyme kinetic data for the purified enzyme. Considering the fact that under physiological conditions an equilibrium of both lactone and free sugar acid exists, the kinetic data was acquired in reactions with substrate provided in these two different forms. The substrate was either hydrolysed with NaOH to obtain free sugar acid, D-mannonate, or the lactone was prepared in buffer, equaling an equilibrium of lactone and free sugar acid. TaManD displayed a higher activity in reactions with D-mannonate compared to reactions with D-mannono-1,4-lactone prepared in buffer at pH 7 ( Table 1). The maximal velocity of the reaction, V max , with D-mannonate as substrate was slightly higher (24-27%) compared to reactions with the lactone prepared in buffer (p-value: 0.06), whereas no difference in the affinity between the two different forms of the substrate could be observed (Table 1, Figure S2).
The pH optimum of TaManD was dependent on the form of the substrate. TaManD displayed highest activity between pH 5 and 7 with mannonate, while with D-mannono-1,4-lactone in buffer, the enzyme displayed overall lower activities with an apparent optimum at pH 7 (Figure 7). Unlike previously studied enzymes from T. acidophilum (e.g., AldT), TaManD did not retain maximum activity above pH 7 with both substrates [15]. Table 1. TaManD kinetic data with D-mannonate (prepared by hydrolysis with NaOH) and with D-mannono-1,4-lactone prepared in buffer (pH 7). Non-linear fitting was performed ( Figure S2). highest activity between pH 5 and 7 with mannonate, while with D-mannono-1,4-lactone in buffer, the enzyme displayed overall lower activities with an apparent optimum at pH 7 ( Figure 7). Unlike previously studied enzymes from T. acidophilum (e.g. AldT), TaManD did not retain maximum activity above pH 7 with both substrates [15]. Based on the observations from experiments with different forms of substrate, we assume that D-mannonate is either the only form, or at least more accessible to TaManD than its lactone. At low pH, TaManD shows low activity with D-mannono-1,4-lactone, supposedly because the lactone does not hydrolyse to the free acid. In contrast, if D-mannono-1,4-lactone was hydrolysed to D-mannonate and used as substrate at low pH, high enzyme activity should be observed. Similarly, the difference in Vmax for D-mannonate and D-mannono-1,4-lactone in buffer can be explained by slow hydrolysis of the lactone to the free acid. Reactions with hydrolysed D-mannonate occur faster compared to the lactone in buffer, since more substrate is readily available. Figure 7. The pH optima of reactions with D-mannono-1,4-lactone prepared in buffer pH 7 (lactone in buffer) and with D-mannonate (prepared by hydrolysis with NaOH). Activity values are expressed as relative activity (%) normalised to the highest overall activity observed in the assay.

Product identification and substrate specificity
In order to identify the product of the TaManD reaction with mannonate, reactions containing the substrate and Co 2+ were analysed at several time points by high-performance liquid chromatography (HPLC) (Figure 8A). Although a complete separation of the product KDG and mannonate could not be achieved on different organic acid columns, an increase of KDG was observed. In order to test for substrate promiscuity, 11 different sugar acids ( Figure 8B) were incubated with purified TaManD and tested for the formation of 2-keto-3-deoxy analogues using the semicarbazide assay. D-mannonate and D-mannono-1,4-lactone were the only substrates that showed a positive reaction after incubation for 16 h. This is in contrast to other dehydratases, including members of the enolase superfamily and the IlvD/EDD superfamily, such as the L-fuconate dehydratase from Xanthomonas campestris [31] or the dihydroxy-acid dehydratase from Sulfolobus solfataricus [17,32], which have been shown to be active with a multitude of different sugar acids. Based on the observations from experiments with different forms of substrate, we assume that D-mannonate is either the only form, or at least more accessible to TaManD than its lactone. At low pH, TaManD shows low activity with D-mannono-1,4-lactone, supposedly because the lactone does not hydrolyse to the free acid. In contrast, if D-mannono-1,4-lactone was hydrolysed to D-mannonate and used as substrate at low pH, high enzyme activity should be observed. Similarly, the difference in V max for D-mannonate and D-mannono-1,4-lactone in buffer can be explained by slow hydrolysis of the lactone to the free acid. Reactions with hydrolysed D-mannonate occur faster compared to the lactone in buffer, since more substrate is readily available.

Product Identification and Substrate Specificity
In order to identify the product of the TaManD reaction with mannonate, reactions containing the substrate and Co 2+ were analysed at several time points by high-performance liquid chromatography (HPLC) ( Figure 8A). Although a complete separation of the product KDG and mannonate could not be achieved on different organic acid columns, an increase of KDG was observed. In order to test for substrate promiscuity, 11 different sugar acids ( Figure 8B) were incubated with purified TaManD and tested for the formation of 2-keto-3-deoxy analogues using the semicarbazide assay. D-mannonate and D-mannono-1,4-lactone were the only substrates that showed a positive reaction after incubation for 16 h. This is in contrast to other dehydratases, including members of the enolase superfamily and the IlvD/EDD superfamily, such as the L-fuconate dehydratase from Xanthomonas campestris [31] or the dihydroxy-acid dehydratase from Sulfolobus solfataricus [17,32], which have been shown to be active with a multitude of different sugar acids.

Cloning, expression and purification of enzymes
All plasmids were constructed by amplification of genes from T. acidophilum DSM 1728 genomic DNA (DSMZ, Braunschweig, Germany) using PCR, restriction and ligation into vectors according to standard protocols [33]. Primer pairs and restriction enzymes used for the construction of each expression plasmid are summarised in Table S2. E. coli strain α-select (Bioline, Sydney, Australia) was used in all initial cloning procedures. The expression plasmid pProEX HTa-Ta0753 was then used to transform BL21-CodonPlus (DE3)-RIL competent cells (Agilent Technologies, Santa Clara, CA, USA). Plasmid pETDuet-1-AldT was used to transform BL21 (DE3) cells (NEB, Ipswitch, Burlington, MA, USA). BL21-CodonPlus (DE3)-RIL strains containing the respective expression plasmid were grown in 500 ml Luria-Bertani medium (LB) supplemented with carbenicillin (100 µ g/ml) and chloramphenicol (35 µ g/ml) at 37 ˚C to an OD600 of 0.6 before induction was performed with 0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 20 °C for 16-18 h. The BL21 (DE3) cells carrying the expression plasmid pETDuet-1-AldT were grown in 500 ml LB medium containing carbenicillin (100 µ g/ml) to an OD600 of 0.5 at 37 °C before induction was performed with 0.4 mM IPTG at 37 °C for 4 h. All cultures were harvested by centrifugation at 5000 x g for 20 min, resuspended in 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 7 and lysed by three passages through a French pressure cell. Soluble fractions of the lysates were obtained by centrifugation at 15000 x g for 20 min. His-tagged enzymes were purified by subjecting the soluble extracts to a 5 ml nickel-nitrilotriacetic acid (Ni-NTA) affinity column (GE Healthcare) equilibrated with buffer containing 50 mM sodium phosphate (NaP), 300 mM NaCl and 20 mM imidazole. The proteins were eluted isocratically with a final concentration of 400 mM imidazole. The buffer of the eluates was exchanged to 50 mM HEPES pH 7 by three rounds of centrifugation in Amicon Ultra centrifugal filters (Millipore) with a 30 kDa molecular weight cut-off.
In order to obtain TaManD without a N-terminal His-tag, the enzyme obtained after pProEX HTa-Ta0753 expression and first His-tag purification was incubated with a TEV Protease (Promega,  Lactones except for D-mannono-1,4-lactone were hydrolysed by incubation in 1 M NaOH for 1 h at room temperature and then diluted in 50 mM HEPES pH 7 to obtain free sugar acids. All sugar acids were tested at a concentration of 10 mM in duplicate experiments.

Cloning, Expression and Purification of Enzymes
All plasmids were constructed by amplification of genes from T. acidophilum DSM 1728 genomic DNA (DSMZ, Braunschweig, Germany) using PCR, restriction and ligation into vectors according to standard protocols [33]. Primer pairs and restriction enzymes used for the construction of each expression plasmid are summarised in Table S2. E. coli strain α-select (Bioline, Sydney, Australia) was used in all initial cloning procedures. The expression plasmid pProEX HTa-Ta0753 was then used to transform BL21-CodonPlus (DE3)-RIL competent cells (Agilent Technologies, Santa Clara, CA, USA). Plasmid pETDuet-1-AldT was used to transform BL21 (DE3) cells (NEB, Ipswitch, Burlington, MA, USA). BL21-CodonPlus (DE3)-RIL strains containing the respective expression plasmid were grown in 500 mL Luria-Bertani medium (LB) supplemented with carbenicillin (100 µg/ml) and chloramphenicol (35 µg/ml) at 37 • C to an OD 600 of 0.6 before induction was performed with 0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 20 • C for 16-18 h. The BL21 (DE3) cells carrying the expression plasmid pETDuet-1-AldT were grown in 500 mL LB medium containing carbenicillin (100 µg/ml) to an OD 600 of 0.5 at 37 • C before induction was performed with 0.4 mM IPTG at 37 • C for 4 h. All cultures were harvested by centrifugation at 5000× g for 20 min, resuspended in 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 7 and lysed by three passages through a French pressure cell. Soluble fractions of the lysates were obtained by centrifugation at 15,000× g for 20 min. His-tagged enzymes were purified by subjecting the soluble extracts to a 5 mL nickel-nitrilotriacetic acid (Ni-NTA) affinity column (GE Healthcare) equilibrated with buffer containing 50 mM sodium phosphate (NaP), 300 mM NaCl and 20 mM imidazole. The proteins were eluted isocratically with a final concentration of 400 mM imidazole. The buffer of the eluates was exchanged to 50 mM HEPES pH 7 by three rounds of centrifugation in Amicon Ultra centrifugal filters (Millipore) with a 30 kDa molecular weight cut-off.
In order to obtain TaManD without a N-terminal His-tag, the enzyme obtained after pProEX HTa-Ta0753 expression and first His-tag purification was incubated with a TEV Protease (Promega, Madison, WI, USA). Digestion was performed according to the manufacturer's protocol with addition of 10% glycerol (v/v, final concentration) and incubation for 1.5 h at 30 • C followed by incubation at 4 • C overnight. Removal of the His-tag and the TEV Protease was achieved by a second Ni-NTA purification using a 1 mL ZetaSep Ni-NTA affinity column (emp Biotech, Berlin, Germany). The flow-through contained the pure protein and cleavage of the His-tag was verified by SDS-PAGE.

Protein Analysis
Purified proteins were separated by SDS-PAGE using 4-15% Tris-glycine polyacrylamide gels (Bio-Rad Laboratories, Hercules, USA) and visualised with Coomassie Brilliant Blue. For the identification of TaManD, the single protein band at 38 kDa was excised from the gel and analysed by LC ESI MS/MS at the Australian Proteome Analysis Facility (APAF, Macquarie University), as described elsewhere [34]. Protein identification was performed using the mascot software (2.4.1, Matrixscience, Boston, MA, USA). The native molecular mass of the recombinant protein was determined by size exclusion chromatography on an ÄKTA pure FPLC system using a Superdex Increase 200 10/300 GL column (GE Healthcare) equilibrated with 10 mM NaP buffer, 140 mM NaCl, pH 7. A calibration curve was obtained with molecular weight markers thyroglobulin (669 kDa), ferritin (440 kDa), aldolase (158 kDa), conalbumin (75 kDa) and ovalbumin (43 kDa).

Nuclear Magentic Resonance (NMR) Analysis
D-Mannono-1,4-lactone (>95%) was purchased from TCI Chemicals Co. (Tokyo, Japan). For NMR spectra of the sugar acid in different substrate forms, 1 M D-Mannono-1,4-lactone solutions were prepared in 1 M NaOH, 1 M HCl and in ultrapure water. All solutions were incubated for 1 h at room temperature, before each solution was diluted to 100 mM using ultrapure water. Then, 300 µL of each 100 mM solution (equivalent to 5 mg sugar acid) was freeze-dried and resuspended in 500 µL 0.1 M NaP buffer, pH 7. The pH of all three samples was evaluated prior to NMR analysis. The substrate incubated in HCl displayed pH 6, while the substrate in water or NaOH displayed pH 7. After transfer of each sample to 5 mm NMR tubes, 10% D 2 O and 40 µM 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid (TMSP) chemical shift standard was added. The 1D decoupled 13 C NMR spectra of non-isotopically labelled samples were acquired on a 500 MHz Bruker Avance III HD NMR equipped with a BBFO probe at 50 • C (323 K), using power-gated proton decoupling with a 90 • pulse with 512 scans and a 3 s relaxation delay between scans. Oxidation of D-[1,6-13 C 2 ]mannose was performed in reactions containing 5 mM of the substrate, 5 mM NAD + , 20 mM NaP buffer at pH 7 and 13.5 µg purified AldT.
Prior to the addition of purified enzyme and after 30 min of reaction at 50 • C, 1D power-gated proton decoupled 13 C NMR spectra were acquired using a 90 • pulse with 4 scans and a 3 s relaxation delay between scans. Visualisation of all spectra was performed with iNMR 6.0 (http://www.inmr.net). After Fourier transformation and automatic phase correction, an exponential visual weighting factor of 1.5 and a smoothing factor of 10 were applied to all spectra.

Enzyme Activity
D-mannono-1,4-lactone was either used after preparation in 50 mM HEPES at pH 7 (indicated as "lactone in buffer") or after hydrolysis to its free acid form (indicated as "mannonate" or "free sugar acid"). Hydrolysis was performed according to Lamble et al. by preparing 1 M stock solutions of D-mannono-1,4-lactone in 1 M NaOH and incubation at room temperature for 1 h before dilutions were prepared in 50 mM HEPES pH 7 [35]. For temperature optimum, pH optimum and thermostability, activity was quantified using the thiobarbituric acid (TBA) assay, according to a modification by Buchanan et al. [36,37]. Unless stated otherwise, reactions (60 µL) contained 0.5-1.5 µg pure enzyme, 10 mM mannonate, 50 mM HEPES pH 7 (adjusted at 55 • C) and 1 mM CoSO 4 . Reactions were stopped by incubation on ice or if needed by addition of 6 µL 12.5% trichloracetic acid (TCA). Next, 50 µL of the reaction mixture was oxidised using 125 µL of 25 mM periodic acid in 0.25 M H 2 SO 4 at room temperature for 20 min. The oxidation was stopped by addition of 250 µL of 2% (w/v) sodium arsenite in 0.5 M HCl. Finally, 1 mL of 0.3% TBA in water was added to the samples and the reaction mixture was boiled for 10 min. The formation of 2-keto-3-deoxy sugar acid was determined by reading A 549nm using 96-well microtiter plates in a Spectrostar Nano plate reader (BMG Labtech, Ortenberg, Germany) and quantified using the molar extinction coefficient of 67.8 × 10 3 M −1 × cm −1 .
The effect of pH on enzyme activity was analysed in duplicate reactions with 1 µg enzyme and 120 mM universal buffer [38]. Reactions were incubated for 2 h at 55 • C before they were analysed with the TBA Assay. In reactions with D-mannono-1,4-lactone in buffer, pH decreased during the course of the experiment, which was accounted for in the analysis by measurement of the pH after the reaction. The optimal temperature for enzyme activity was determined in duplicate reactions containing 1 µg enzyme. Duplicate reactions were performed for 1 h at different temperatures before being analysed with the TBA Assay. For thermostability assays, 120 mM universal buffer was used and purified enzyme (1 µg) was incubated in duplicates at temperatures between 55 • C and 95 • C in the absence of substrate. Samples were removed at different times and the residual activity was measured in reactions with 10 mM mannonate for 45 min at 55 • C.
The semicarbazide assay was used according to Wichelecki et al. [12] to determine the effect of metal ions, chelators and reducing agents, and the acquisition of Michalis-Menten kinetics (K m, V max ). Each reaction (60 µL) was incubated with 240 µL semicarbazide reagent at room temperature for 1 h. A 250nm was read in a UV transparent microtiter plate (Thermo Scientific, Waltham, MA, USA) using a Spectrostar Nano plate reader. Product formation was quantified using a standard curve of 2-keto-3-deoxygluconate (Sigma-Aldrich, St. Louis, MO, USA) prepared in the same assay buffer. K m, V max values were estimated using non-linear fitting in Prism 6 (6.0c, GraphPad software, San Diego, CA, USA). For the effect of metals, chelators and reducing agents, pure enzyme (0.6 µg) was incubated with 1 mM of each additive in 50 mM HEPES pH 7 for 1 h at room temperature. Activity was measured in duplicate reactions containing 10 mM D-mannono-1,4-lactone in 50 mM HEPES pH 7, incubated for 1 h at 55 • C and then analysed with the semicarbazide assay. Kinetic data for the determination of K m and V max were performed in duplicate reactions (60 µL), containing 1 µg purified TaManD, 50 mM HEPES pH 7, 1 mM CoSO 4 and different substrate concentrations (1-50 mM). Linear increase of reaction product was assured over 1 h reaction time. Reactions were performed for 45 min at 55 • C before product formation was determined using the semicarbazide assay and the KDG standard curve. Substrate specificity was analysed using 11 different sugar acids ( Figure 8B). Reactions were performed in duplicate and contained 0.5 µg enzyme, 50 mM HEPES pH 7, 1 mM CoSO 4 and 10 mM of each sugar acid. Reactions were incubated for 16 h at 55 • C before they were analysed using the semicarbazide assay.

High Performance Liquid Chromatography (HPLC) Analysis
Reactions catalysed by TaManD and standards of mannonate and KDG were analysed using an Agilent 1290 HPLC system connected to a refractive index detector (RID) G1362A (Agilent Technologies, Santa Clara, CA, USA). Samples were analysed on an organic acid column (Agilent HiPlex H + ) with 10 mM H 2 SO 4 as a mobile phase at a flow rate of 0.6 ml/min. The column was heated to 80 • C and the RID was set to 55 • C. Reactions were stopped by addition of 7 µL 12.5% TCA to a 60 µL reaction. After short centrifugation, 10 µL of the supernatant was used for HPLC analysis.

Statistical Analysis
Statistical analysis was performed for enzyme kinetics and effect of different metals and additives on enzyme activity, by two-tailed unpaired t-tests using Prism 6.

Conclusions
In this study, we present the first purification and characterisation of a functional archaeal mannonate dehydratase. The gene encoding for the mannonate dehydratase was found adjacent to a previously described aldohexose dehydrogenase (AldT) gene in the genome of T. acidophilum [15]. Previously, it has been shown that AldT from T. acidophilum selectively oxidises mannose, but the physiological function behind that oxidation was not investigated. Using NMR spectroscopy, we were able to show that mannonate and mannono-1,4-lactone are produced via oxidation of D-mannose by AldT. Kinetic assays confirmed that TaManD was able to convert both the sugar acid and its lactone to the central intermediate KDG at neutral pH, without further help of a lactonase. This resembles the second step of many oxidative pathways studied in archaea, including those for sugars like glucose, galactose, rhamnose, arabinose and xylose [39]. The amino acid sequence of TaManD and those of the putative mannonate dehydratases from F. acidarmanus and F. acidiphilum share high amino acid sequence identity. Although the gene annotations for those mannonate dehydratases indicate a role in the hexuronate metabolism (uxuA), they are all located adjacent to (putative) aldohexose dehydrogenases. It remains to be seen whether a non-phosphorylative pathway starting from mannose, similar to the non-phosphorylative Entner-Doudoroff pathway from glucose and galactose, exists in thermophilic archaea like T. acidophilum, F. acidarmanus and F. acidiphilum.