Diverse allosteric and catalytic functions of tetrameric d-lactate dehydrogenases from three Gram-negative bacteria

NAD-dependent d-lactate dehydrogenases (d-LDHs) reduce pyruvate into d-lactate with oxidation of NADH into NAD+. Although non-allosteric d-LDHs from Lactobacilli have been extensively studied, the catalytic properties of allosteric d-LDHs from Gram-negative bacteria except for Escherichia coli remain unknown. We characterized the catalytic properties of d-LDHs from three Gram-negative bacteria, Fusobacterium nucleatum (FNLDH), Pseudomonas aeruginosa (PALDH), and E. coli (ECLDH) to gain an insight into allosteric mechanism of d-LDHs. While PALDH and ECLDH exhibited narrow substrate specificities toward pyruvate like usual d-LDHs, FNLDH exhibited a broad substrate specificity toward hydrophobic 2-ketoacids such as 2-ketobutyrate and 2-ketovalerate, the former of which gave a 2-fold higher kcat/S0.5 value than pyruvate. Whereas the three enzymes consistently showed hyperbolic shaped pyruvate saturation curves below pH 6.5, FNLDH and ECLDH, and PALDH showed marked positive and negative cooperativity, respectively, in the pyruvate saturation curves above pH 7.5. Oxamate inhibited the catalytic reactions of FNLDH competitively with pyruvate, and the PALDH reaction in a mixed manner at pH 7.0, but markedly enhanced the reactions of the two enzymes at low concentration through canceling of the apparent homotropic cooperativity at pH 8.0, although it constantly inhibited the ECLDH reaction. Fructose 1,6-bisphosphate and certain divalent metal ions such as Mg2+ also markedly enhanced the reactions of FNLDH and PALDH, but none of them enhanced the reaction of ECLDH. Thus, our study demonstrates that bacterial d-LDHs have highly divergent allosteric and catalytic properties.

In the case of L-LDHs, it is known that allosteric types of the enzyme are widely distributed in bacteria including Lactobacilli (Garvie 1980). These enzymes commonly require fructose 1,6-bisphosphate (FBP) to exhibit high catalytic activity, and usually exhibit sigmoidal shaped substrate saturation curves unless FBP is present. L-LDHs generally have tetrameric structures comprising four identical subunits, and allosteric L-LDHs undergo cooperative structural changes in the substrate and FBP binding sites through quaternary structural changes (Iwata et al. 1994;Arai et al. 2010). On the other hand, no allosteric property has been reported for Lactobacillus D-LDHs. The Lactobacillus D-LDHs consistently have a dimeric structure comprising identical subunits, and exhibit hyperbolic shaped saturation curves for substrate pyruvate. Interestingly, it was reported that D-LDH from E. coli exhibits a sigmoidal pyruvate saturation curve, which is changed to the hyperbolic one by 2-ketobutyrate (Tarmy and Kaplan 1968). The reported results of size-exclusion chromatography suggested that this enzyme has a homotetrameric structure, although the detailed catalytic properties and 3D structure of the enzyme remain unknown. Beside the D-LDH from E. coli, the D-3-phosphoglycerate dehydrogenases (D-PgDHs) from some bacteria, such as E. coli (Tobey and Grant 1986;Grant 1989;Schuller et al. 1995) and Mycobacterium tuberculosis (Dey et al. 2005), are only known as allosteric and homotetrameric enzymes in the D-HydDH superfamily.
In this paper, we describe the detailed biochemical analysis of allosteric D-LDHs from three Gram-negative bacteria; Fusobacterium nucleatum subsp. nucleatum, an obligate anaerobe; Pseudomonas aeruginosa, an obligate aerobe; and E. coli, a facultative anaerobe, demonstrating that bacterial D-LDHs have highly divergent catalytic properties.

Enzyme assay
Enzyme activity was determined by measuring the reduction of the absorbance at 340 nm derived from NADH for 1 minute. The reactions were performed at 30°C in various 50 mM buffers containing 0.1 mM NADH and a substrate. The data that showed significant cooperative effects of the substrate were interpreted using the Hill equation (Equation 1) (Dixon and Webb 1979).
where v is the reaction velocity, [S] the ligand concentration, such as pyruvate, FBP, or Mg 2+ , k cat the turnover rate of catalysis, v min the reaction velocity with no ligand, S 0.5 the half-saturation concentration of a ligand, and n H the Hill coefficient. Kinetic parameters were obtained by curve fitting of the data with KaleidaGraph ver 3.51. The data that indicated significant substrate inhibition were interpreted using the equation for substrate inhibition (Equation 2) (Eszes et al. 1996).
where K i is the inhibition constant. The kinetic parameters for oxamate at pH 7.0 were calculated by curve fitting of the data with GraFit version 7.0.3. The data that indicated significant competitive-type and mix-type inhibition were interpreted using the equation for competitivetype (Equation 3) and mix-type inhibition (Equation 4), respectively.
where K m is the Michaelis constant, and K i ' the inhibition constant. The data of oxamate inhibition for FNLDH, PALDH, and ECLDH at pH 8.0 were interpreted using the equation for competitive-type (Equation 5), mix-type (Equation 6), and non-competitive-type (Equation 7) inhibition of allosteric enzyme, respectively. v where k cat ' is the turnover rate of catalysis in the presence of oxamate, K act the half-saturation concentration of oxamate for activation, and n H ' the hill coefficient for oxamate.

Temperature and pH stability
The pH-stability was determined by measuring the remaining activity after incubation of the enzymes (1 μM) in various 50 mM buffers at 30°C for 1 h. The thermostability was determined by measuring the remaining activity after incubation of FNLDH (1 μM) or ECLDH (1 μM) in 50 mM HEPES-NaOH buffer (pH 8.0), or of PALDH (1 μM) in 50 mM sodium acetate buffer (pH 5.0) at various temperatures for 30 min.

Basic properties
The recombinant FNLDH, PALDH and ECLDH consistently exhibited marked catalytic activity toward pyruvate, and were successively purified to homogeneous protein samples ( Figure 1a). FNLDH, PALDH, and ECLDH were stable in the pH ranges of 5.0-8.5, 4.0-8.0, and 5.0-10.0, respectively, during treatment at 30°C for 1 h (Figure 2 a, b, c). FNLDH, PALDH, and ECLDH were stable up to 39, 57, and 49°C, respectively, under favorable pH conditions, i.e., pH 8.0 for FNLDH and ECLDH, and pH 5.0 for PALDH ( Figure 2d). The purified FNLDH and PALDH samples each gave a single protein peak on size-exclusion chromatography analysis, and exhibited apparent molecular weights of 160 and 120 kDa, respectively ( Figure 1b). On the other hand, the ECLDH sample gave two peaks corresponding to molecular weights of 330 and 150 kDa. The theoretical molecular weights of FNLDH, PALDH, and ECLDH are 40.0, 37.8, and 38.5 kDa, suggesting that FNLDH and PALDH at that of homotetrameric and homotrimeric structures, respectively, and ECLDH was eluted at the position of homooctameric and homotetrameric structures. Since it has been reported that native D-LDH from E. coli cells gives only a peak that corresponds to the molecular weight of 130 kDa (Tarmy and Kaplan 1968), the apparently octameric form of ECLDH is likely an artificial product due to the overexpression, and exhibited slightly but significantly lower specific activity than the tetrameric form (data not shown). Therefore, only homotetrameric ECLDH was used for the detailed analysis below.

pH-Dependency of kinetic parameters
The kinetic parameters, k cat , S 0.5 , k cat /S 0.5 , and n H values, were determined by pyruvate reduction assaying in the pH range of 4.5-9.0 ( Figure 3). The three enzymes consistently showed virtually constant k cat values independently of pH, and PALDH and ECLDH exhibited about 5-fold larger k cat values than FNLDH. The three enzymes also consistently showed constant pyruvate S 0.5 values below pH 7.0, and PALDH and FNLDH exhibited approximately one order of magnitude smaller S 0.5 values than ECLDH. Above pH 8.0, the three enzymes exhibited increased S 0.5 values, depending on the pH, and their k cat /S 0.5 values changed mostly according to the changes in the S 0.5 values. ECLDH apparently showed lower pH-dependence of the S 0.5 value than the other two enzymes, of which PALDH showed slightly higher pHdependence than FNLDH. Under acidic conditions below pH 6.0, the three enzymes consistently showed hyperbolic shaped pyruvate saturation curves, and no marked cooperativity in pyruvate binding. Under higher pH conditions, however, FNLDH and ECLDH showed markedly sigmoidal saturation curves for pyruvate, i.e., positive homotropic cooperativity, giving maximal n H values of about 2.0. In contrast, PALDH showed negative homotropic cooperativitiy above pH 8.0, giving a minimal n H value of 0.5.

Substrate specificity
The kinetic parameters of the D-LDHs for various 2ketoacids are summarized in Table 1. For pyruvate, PALDH exhibited the highest catalytic efficiency among the three enzymes, since it exhibited a small S 0.5 like FNLDH and a large k cat like ECLDH. ECLDH showed a narrow substrate specificity toward pyruvate among hydrophobic 2-ketoacid substrates, although ECLDH consistently exhibited relatively high catalytic activity toward oxaloacetate, glyoxylate and hydroxypyruvate. PALDH showed a similar substrate specificity to that of ECLDH, exhibiting relatively high catalytic activity toward oxaloacetate, glyoxylate and hydroxypyruvate, although it generally exhibited higher catalytic efficiencies for these substrates than ECLDH. In contrast, FNLDH showed a broad substrate specificity toward relatively bulky hydrophobic 2-ketoacids, and exhibited particularly high catalytic activity toward 2-ketobutyrate and 2-ketovalerate, for which the catalytic efficiencies were comparable to that for pyruvate. FNLDH also showed relatively high activity to oxaloacetate and hydroxypyruvate, but poor activity toward glyoxylate.

Inhibition and activation by oxamate
Oxamate, an inert pyruvate analogue, inhibited the reactions of the three enzymes at pH 7.0, where the enzymes showed no marked homotropic cooperativity, in different manners, i.e. an apparently competitive manner with pyruvate for FNLDH, and a mixed manner for PALDH and ECLDH (Figure 4). This suggests that oxamate is bound not only to the catalytic site, but also to unknown allosteric sites in PALDH and ECLDH, whereas it is bound mostly to the catalytic site in FNLDH. Oxamate apparently enhanced the reactions of FNLDH and PALDH, and exhibited the highest activating effects at 20 mM and 2.5 mM, respectively (Figure 5a Figure 3 pH dependence of kinetic parameters on pyruvate reduction. White, grey, and black circles indicate FNLDH, PALDH, and ECLDH, respectively. The k cat (a), S 0.5 (b), and k cat /S 0.5 data (c) are plotted logarithmically, and the Hill coefficient (d) is plotted linearly. The buffers used for the assay were sodium acetate buffer (pH 4.5, 5.0, and 5.5; circles), MES-NaOH buffer (pH 5.5, 6.0, and 6.5; triangles), MOPS-NaOH buffer (pH 6.5, 7.0, and 7.5; squares), HEPES-NaOH buffer (pH 7.0, 7.5 and 8.0; diamonds), and Bicine-NaOH buffer (pH 8.0, 8.5, and 9.0; hexagons). We adjusted the pH of each buffer solution prior to the addition of substrate and cofactor, and confirmed that the pH value is not affected even in the presence of high concentrations of substrate.   Figure 5 Effects of oxamate on the catalytic reactions at pH 8.0. a-c) The saturation curves for oxamate. The reaction velocities for FNLDH (a), PALDH (b), and ECLDH (c) were measured in 50 mM Bicine-NaOH buffer (pH 8.0) containing 0.1 mM NADH, 2.5 mM (for FNLDH), 1.2 mM (for PALDH), or 7.5 mM (for ECLDH) pyruvate, and the indicated concentration of oxamate. The data for FNLDH, PALDH, and ECLDH were interpreted using the equation for competitive-type, mix-type, and noncompetitive-type inhibition of allosteric enzyme. The kinetic parameters were as follows; FNLDH: K i = 51 ± 10 (mM), k cat ' = 80 ± 20 (s −1 ), n H ' = 1.1 ± 0.1, and K act = 9.9 ± 3 (mM). PALDH: K i K i ' / (K i + K i ') = 2.0 ± 0.6 (mM), k cat ' = 470 ± 100 (s −1 ), n H ' = 1.7 ± 0.2, and K act = 1.7 ± 0.4 (mM). ECLDH: K i = 80 ± 3 (mM). The lines indicate the calculated saturation curves obtained with kinetic parameters. noncompetitive manner toward pyruvate (Figure 5f), suggesting that it is mostly bound to the allosteric site of ECLDH at pH 8.0.

Heterotropic activation of allosteric D-LDHs
Bacterial allosteric L-LDHs are commonly activated by FBP, which usually induces drastic improvement of the substrate S 0.5 values of the enzymes (Garvie 1980). In addition, Lactobacillus casei L-LDH requires some divalent metal ions (e.g., Mn 2+ ) (Arai et al. 2011), and the Thermus caldophilus enzyme is activated also by citrate under slightly acidic conditions (Taguchi et al. 1984). We therefore evaluated the effects of FBP, citrate and divalent metal ions at pH 8.0 in the presence of the S 0.5 pyruvate. FBP and citrate (1.0 mM) slightly enhanced the catalytic reactions of the three D-LDHs, and divalent metal ions such as Mg 2+ and Mn 2+ more markedly enhanced the reactions of FNLDH and PALDH (Figure 6a). In the case of FNLDH, Mg 2+ and FBP showed significant activation effects, and Mg 2+ gave a 2.8-fold smaller S 0.5 value (1.5 mM) than FBP (4.2 mM) for the enzyme activation ( Figure 6 and Table 2). In the presence of 10 mM FBP or 5 mM Mg 2+ , FNLDH exhibited hyperbolic pyruvate saturation curves, the pyruvate S 0.5 (K m ) value being reduced by approximately 1.4-fold or twice, respectively ( Figure 6, and Table 3 Table 2). The data of MgCl 2 saturation curve of ECLDH was not fitted with any equations used in this study. e-g) the saturation curves for pyruvate with or without FBP or MgCl 2 . The reaction velocities for FNLDH (e), PALDH (f), and ECLDH (g) were measured in 50 mM Bicine-NaOH buffer (pH 8.0) containing 0.1 mM NADH, the indicated concentrations of pyruvate with no effector (white circles), 10 mM (for FNLDH and PALDH) or 5 mM (for ECLDH) FBP (grey circles), and 5 mM (for FNLDH and PALDH) or 2.4 mM (for ECLDH) MgCl 2 (black circles). The lines indicate the calculated saturation curves obtained with kinetic parameters (Table 3).
showed activation effects, and Mg 2+ gave a 75-fold smaller S 0.5 value (0.16 mM) than FBP (12 mM) for the enzyme activation ( Figure 6 and Table 2). In this case, FBP and Mg 2+ also consistently reduced the pyruvate S 0.5 value significantly, and the negative cooperativity in pyruvate binding ( Figure 6, and Table 3). In the case of ECLDH, FBP and Mg 2+ exhibited only slight activation effects on the enzyme reaction, and the latter even markedly inhibited the reaction at high concentrations ( Figure 6, and Tables 2 and 3).

Kinetic profiles for NADH
The apparent kinetic parameters for NADH were determined at the saturation concentration of pyruvate at pH 7.0 and 8.0 (Table 3). FNLDH and PALDH exhibited virtually the same apparent NADH S 0.5 value (about 0.035 mM), which was approximately twice smaller than that of ECLDH. Since the pyruvate S 0.5 values were determined at 0.1 mM NADH (Figure 3 and Table 3), these values might be larger than the real values because of an insufficient concentration of NADH, particularly in the case of ECLDH. PALDH and ECLDH exhibited hyperbolic NADH saturation curves independently of pH, FBP or Mg 2+ . On the other hand, FNLDH exhibited a sigmoidal one at pH 8.0, and exhibited an n H value of 1.6, which was slightly reduced by FBP or Mg 2+ .

Discussion
The characteristics of FNLDH, PALDH and ECLDH demonstrate the high variety of the functions of bacterial D-LDHs. Of the three enzymes, PALDH exhibits a markedly smaller pyruvate S 0.5 value than ECLDH, and a larger k cat value than FNLDH at pH 7.0 (Table 3).
Although the crucial role of PALDH in the metabolism of P. aeruginosa remains uncertain, the enzyme is likely involved in regulation of the NAD + /NADH ratio within the cells like the heart-type L-LDH isozymes of aerobic  tissues (Holbrook et al. 1975). On the other hand, ECLDH with the large S 0.5 is likely favorable for facultative anaerobes such as E. coli, in which the pyruvate level within the cells dynamically changes between anaerobic and aerobic conditions like the muscle-type L-LDH isozymes (Holbrook et al. 1975). FNLDH exhibits a small pyruvate S 0.5 value although F. nucleatum is obligate anaerobe. In F. nucleatum cells, pyruvate is likely consumed mostly through pathways other than lactate fermentation, such as butyrate fermentation (Kapatral et al. 2002), and therefore pyruvate within the cells may be maintained at low concentrations. It is particularly notable that FNLDH prefers relatively bulky hydrophobic 2-ketoacids, such as 2-ketobutyrate and 2-ketovalerate, which give apparently equivalent or even higher k cat /K m value than pyruvate, unlike in the cases of the other two enzymes (Table 1). This substrate preference suggests that FNLDH might be used for not only pyruvate reduction but also the reduction or oxidation of other bulky 2-ketoacid or 2-hydroxy acids, respectively, although the actual physiological substrates of the enzyme remains uncertain. FNLDH apparently has a size-preference for 2-ketoacid substrates intermediate between those of conventional D-LDH and D-2-hydroxyisocaproate dehydrogenase (Hummel et al. 1985). Like usual D-LDHs, nevertheless, FNLDH has aromatic residues, Phe52 and Tyr299 (residue numbers are according to D-LDH from Lactobacillus pentosus), which are proposed to play key roles in the substrate recognition in the binding pocket for the substrate C3 side chains (Tokuda et al. 2003;Ishikura et al. 2005). It is also noteworthy that FNLDH exhibits relatively high activity toward 2-ketoisovalerate since little is known as to an enzyme that exhibits high catalytic activity toward C3-branched 2-ketoacid substrates in the D-LDH-related D-HydDH family. It is known that D-mandelate dehydrogenase from Enterococcus faecalis exhibits high catalytic activity toward hydrophobic C3branched ketoacids such as 2-ketoisovalerate and benzoyl formate (Tamura et al. 2002). Nevertheless, this enzyme belongs to another D-HydDH family, the 2-ketopantate reductase-related family, which is evolutionally separate from the D-LDH-related family (Wada et al. 2008;Miyanaga et al. 2013). Hence, FNLDH is the first case of a D-LDH-related D-HydDH that exhibits relatively high catalytic activity toward C3-branched 2-ketoacid substrates and may be useful for enzymatic production of C3-branched D-2-hydroxyacids.
For FNLDH, oxamate inhibits the reaction competitively with pyruvate at pH 7.0, and substitutes for the role of pyruvate in the homotropic enzyme activation at pH 8.0, converting the sigmoidal saturation curve for pyruvate to a hyperbolic one (Figure 5d). These effects indicate that oxamate simply competes with pyruvate for a common binding site, i.e., the catalytic site in FNLDH. It is likely that PALDH binds oxamate at both the catalytic and allosteric sites, since oxamate inhibits the reaction in a mixed manner at pH 7.0 (Figure 4), as in the case of ECLDH, but stimulates the reaction at low concentration at pH 8.0, where oxamate likely exhibits an activation effect mostly through binding to the catalytic site and inhibitory effects mostly through binding to the unknown allosteric site (Figure 5b,e). In the case of ECLDH, it was suggested previously that oxamate is likely bound mostly to a binding site other than the catalytic site (Tarmy and Kaplan 1968), this being consistent with the results of this study (Figure 4c). These results indicate that the three enzymes differ in their allosteric behaviors, and specificities to allosteric effectors, although the actual allosteric binding sites or physiological effectors of these enzymes remain uncertain.
FNLDH and PALDH are the first cases of D-LDHs that are activated by FBP, which is a common activator for bacterial allosteric L-LDHs (Garvie 1980). It is nevertheless unlikely that FBP is really the specific activator of the two D-LDHs, since the two enzymes require apparently higher concentrations of FBP than the physiological one for their activation (Figure 6b,c and Table 2). On the other hand, the Mg 2+ ion is more effective than FBP as to activation of the two D-LDHs, particularly PALDH ( Figure 6, and Tables 2 and 3). It is, however, also unlikely that the Mg 2+ ion is the specific activator of these enzymes, since some other divalent metal ions such as Sr 2+ and Mn 2+ also improve their reactions (Figure 6a). These two enzymes thus appear to be activated through the comprehensive effects of several activators with low specificities, physiologically, although it is possible that they have unknown activators with higher specificities. In the case of ECLDH, FBP and Mg 2+ only slightly enhance the catalytic reaction, and the latter even markedly reduces the reaction at higher concentrations (Figure 6d,g and Table 3). We have evaluated other possible effectors, such as L-alanine, L-glutamate and L-aspartate, for this enzyme, but have not found out any potential activator for ECLDH (data not shown). This implies that ECLDH is only negatively controlled under alkaline conditions.
The allosteric and tetrameric D-PgDHs from E. coli and M. tuberculosis were extensively studied as to their structure-function relationship (Tobey and Grant 1986;Grant 1989;Schuller et al. 1995;Dey et al. 2005;Grant 2012). The D-PgDHs have V-type regulatory mechanisms, in which L-serine inhibits the enzyme reaction in a non-competitive manner, whereas the three D-LDHs apparently exhibit K-type rather than V-type regulation. These D-PgDHs have additional functional domain(s), ACT domain and ASB domain, which undergo not only interactions with allosteric effector, L-serine, but also inter-subunit interactions with each other for the construction of tetrameric structure. In contrast, D-LDHs have no additional protein domain. The three D-LDHs from Gram-negative bacteria thus greatly differ from these D-PgDHs in both regulatory mechanism and protein structure.
It is reported that bacterial allosteric L-LDHs consistently undergo the Monod-Wyman-Changeux-type (MWC-type: pre-existing type) (Monod et al. 1965;Iwata et al. 1994;Arai et al. 2010) allosteric transition, where the active (R) state with high affinity to pyruvate and FBP, and the inactive (T) state with low affinity. However, such drastic structural changes appear to be unnecessary for the three D-LDHs, which exhibit apparently much smaller changes in catalytic activity due to the allosteric transition than the L-LDHs. In fact, the MWC model is not available for the case of PALDH, which shows negative homotropic cooperativity. In addition, FNLDH shows significant positive homotropic cooperativity also for NADH binding, which is less affected by FBP or Mg 2+ than pyruvate binding ( Table 3), suggesting that FNLDH changes in its structure through two steps, the NADH binding and following pyruvate binding steps, in the allosteric transition. The Lactobacillus D-LDHs consistently have dimeric structure, whereas the three D-LDHs from Gram-negative bacteria have tetrameric structure. Thus, tetramerization might be correlated with the allostery of D-LDHs. It is highly desirable to determine the 3Dstructures of these D-LDHs.