Hydrogen peroxide-producing pyruvate oxidase from Lactobacillus delbrueckii is catalytically activated by phosphotidylethanolamine



Pyruvate oxidase (Pox) is an important enzyme in bacterial metabolism for increasing ATP production and providing a fitness advantage via hydrogen peroxide production. However, few Pox enzymes have been characterized from bacterial species. The tetrameric non-hydrogen-peroxide producing Pox from E. coli is activated by phospholipids, which is important for its function in vivo.


We characterized the hydrogenperoxide-producing Pox from L. delbrueckii strain STYM1 and showed it is specifically activated by phosphotidylethanolamine (16:0–18:1), but not by phosphotidylcholine or phosphotidylglycerol. This activation is a mixture of K- and V-type activation as both km and enzyme turnover are altered. Furthermore, we demonstrated that the L. delbrueckii Pox forms pentamers and either decamers or dimers of pentamers in solution, which is different from other characterized Pox enzymes. Lastly, we generated a C-terminal truncation mutant that was only weakly activated by phosphotidylethanolamine, which suggests the C-terminus is important for lipid activation.


To our knowledge this is the first known hydrogenperoxide-producing Pox enzyme that is activated by phospholipids. Our results suggest that there are substantial differences between Pox enzymes from different bacterial species, which could be important for their role in biological systems as well as in the development of Pox-based biosensors.


CidC is another acetate-producing Pox enzyme from Staphylococcus aureus which transfers electrons to menaquinone and has an important role in cell death pathways [3]. Pyruvate:Quinone Oxidoreductase (PQO) from Corynebacterium glutamicum also produces acetate and transfers electrons to a quinone [5]. One characteristic of E. coli Pox, PQO, and CidC is that they are catalytically activated by phospholipids, which is thought to facilitate efficient transfer to membrane electron shuttles by activating the enzyme at a membrane peripheral position [3,5,6]. Mutants in E. coli Pox that disrupt lipid activation of the enzyme are localized to the C-terminal region of the protein. Further biochemical analysis identified a lipid activation helix in the C-terminal region (amino acids 558-568) that upon lipid interaction induce the preceding alpha helix to move out of the active site, which in turn positions phenylalanine 465 into the active site enhancing electron transfer between TPP and FAD [7][8][9]. The phospholipid activation of E. coli Pox is a hybrid of K mdriven (K-type) and velocity-driven (V-type) allosteric activation where the k m for pyruvate decreases along with an increase in the enzyme turnover rate [10]. However, in CidC the activation is V-type where the enzyme turnover rate is substantially increased [3].
Hydrogen peroxide-generating Pox enzymes are produced by multiple different bacteria. Pox enzymes from Lactobacillus plantarum and Streptococcus pneumoniae are the most well studied. These enzymes form homotetrameric structures [11,12]. The role of Pox enzymes in central metabolism is thought to be increasing ATP production in concert with acetate kinase during aerobic metabolism [13]. In S. pneumoniae, which has no tricarboxylic acid (TCA) cycle, additional ATP production from glucose is believed to come from this pathway [10]. L. plantarum Pox is considered to be an important component of the enhanced biomass that is observed in aerobic growth by generating acetyl phosphate as a substrate for acetate kinase [13]. In addition, the production of hydrogen peroxide from Pox could also be important on a community level. Since hydrogen peroxide is toxic to some bacteria, Pox activity could confer a fitness advantage to the producer organism. In S. pneumoniae, pox-deficient mutants display decreased virulence in a rat model of disease likely through decreased adhesion properties, but also potentially through decreased competition with commensal bacteria [14]. Streptococcus gordonii and Streptococcus sanguinis are known to produce enough hydrogen peroxide primarily through Pox to inhibit the oral pathogen Streptococcus mutans [15]. Recently, Lactobacillus delbrueckii strains harboring intact pox genes were shown to produce enough hydrogen peroxide to inhibit the growth of the oral pathogen Porphyromonas gingivalis [16]. The characterization of hydrogen peroxide-producing strains and their respective encoded hydrogen peroxide-producing enzymes could be beneficial in the development of novel treatments for diseases like periodontitis.
In the current study, we characterize a hydrogen peroxideproducing Pox enzyme from L. delbrueckii that is catalytically activated specifically by phosphotidylethanolamine, a common bacterial membrane component. This activation is largely dependent on the C-terminal region of the protein.
Further, we showed that the L. delbrueckii Pox adopts a pentameric structure and potentially a dimer of pentamers, which is novel for Pox enzymes. These characteristics are unique for hydrogenperoxide-producing Pox enzymes and demonstrate substantial variability in the structure and function of Pox from different bacterial species.

Characterization of L. delbrueckii pox
The E. coli Pox has been well studied, and it is known that this enzyme is catalytically activated by phospholipids [6,7,17]. In addition, CidC and PQO from S. aureus and C. gluticaticum, respectively, are other pyruvate oxidases that are lipid activated [3,5]. Another well studied Pox enzyme is from Lactobacillus plantarum [1,13,18]. This enzyme has not been shown to be catalytically activated by phospholipids indicating that there is variability in the catalytic activation of the Pox enzymes. Amino acid identity between the enzymes ranges from 25.9% between L. plantarum Pox and PQO to 46.4% between E. coli Pox and PQO (Table 1). For DNA sequence identity, the identity ranges from 50.7% between CidC and PQO to 56.3% for CidC and L. plantarum Pox (Table 1).
First, we wanted to determine the biochemical properties of the L. delbrueckii Pox enzyme. We previously purified L. delbrueckii Pox and used it for subsequent testing [16]. We determined the k m values for pyruvate and phosphate and the k cat value. The k m for pyruvate and phosphate were 342.2 μM and 8 mM respectively, and the k cat was 0.367 s − 1 ( Fig. 1 and Table 2). The optimum pH for several other pyruvate oxidase enzymes is near 5.5 and so we tested the activity of the L. delbrueckii Pox at various pHs to determine its pH activity profile. In agreement with other Pox enzymes, the L. delbrueckii Pox was most active at the pH range of 5.5 to 6 ( Fig. 1c).

L. delbrueckii pox forms pentameric structure in solution
Both the L. plantarum and E. coli Pox are known to form tetrameric structures [1,4]. To determine whether the L. delbrueckii Pox also formed a tetramer, we performed gel filtration chromatography with the purified enzyme. Interestingly, L. delbrueckii Pox eluted in two major peaks; the first with a calculated molecular weight of 694 kDa and the second with a calculated molecular weight of 345 kDa ( Fig. 2a and b). The molecular weight of an STYM1 L. delbrueckii Pox monomer is 68 kDa according to the translation of the gene sequence; therefore, the first peak corresponded to a decamer and the second peak to a pentamer. The last peak corresponded to unbound FAD (Fig. 2a). Analysis of the fractions by non-reducing SDS-PAGE revealed two distinct bands that migrated distances consistent with monomer and dimer forms of the L. delbrueckii Pox at approximately 68 kDa and 136 kDa (Fig. 2c). Fractions 13 and 14, which correspond to the last peak from gel filtration, did not contain any protein and also were yellow in color confirming the presence of unbound FAD (Fig. 2c). To confirm that the 136 kDa protein band was in fact a dimer of L. delbrueckii Pox, we analyzed the L. delbrueckii Pox on non-reducing and a reducing SDS-PAGE. Under reducing conditions, L. delbrueckii Pox migrated exclusively as a monomer indicating that the 136 kDa band is a dimer, presumably linked by a disulfide bond at position 72 since it is the only cysteine in the amino acid sequence (Fig. 2d).
Phospholipid activation of L. delbrueckii pox We hypothesized that L. delbrueckii Pox may be activated by phospholipids similar to E. coli Pox, CidC, and PQO [2,3,5]. We tested whether addition of either phosphotidylethanolamine (PE), phosphotidylcholine (PC), and phosphotidylglycerol (PG) could catalytically activate L. delbrueckii Pox. L. delbrueckii Pox was catalytically activated specifically by PE, but not PC or PG (Fig. 3a). The phospholipids used contained the same acyl chain structure (16:0-18:1) and only varied in the headgroup indicating that there is specificity in phospholipid activation of L. delbrueckii. To determine the full activation profile of L. delbrueckii Pox by PE, we measured the fold activity over a wider range of PE concentrations and observed activation upwards of 6-fold relative to the enzyme without PE (Fig. 3b). This level of activation exceeds that of the CidC Pox (3-fold activation) from S. aureus [3].
To gain mechanistic insight into how PE activates the enzyme, we tested the kinetic parameters of L. delbrueckii Pox in the presence of PE (Fig. 4). With the addition of PE, the k m for pyruvate and phosphate were 519.9 μM and 1.25 mM respectively and the k cat was 6.85 s − 1 ( Table 3). The approximately 18-fold increase in k cat and decrease in phosphate k m relative to the enzyme in the absence of PE suggest that these parameters are the primary drivers of catalytic activation, which is   consistent with the increase in k cat observed in E. coli Pox and CidC upon phospholipid activation [3,19]. The activation observed here is a hybrid of V-and K-type activation since the k cat increased and the phosphate k m decreased, which is also consistent with E. coli Pox [9,19].
We also wanted to determine whether L. delbrueckii Pox inserts into PE micelles. We performed coprecipitation experiments with L. delbrueckii Pox in the presence and absence of PE. Nearly all of the enzyme remained in the soluble fraction and did not coprecipitate with PE, which is in contrast to E. coli Pox [7] (Fig. 5).
The C-terminus of L. delbrueckii pox is important for lipid activation The C-terminus of E. coli Pox is important for lipid activation of the enzyme [7]. To test if the C-terminus was important in the L. delbrueckii Pox, we generated a Cterminal truncation that lacked the last 34 amino acids.
We purified the enzyme and tested it for its ability to be activated by PE (Fig. 6a). The enzyme had a specific activity of approximately 1.32 U/mg in the absence of PE, which is greater than the specific activity of the fulllength enzyme (Fig. 1c and Fig. 6b). This is consistent with the E. coli pox truncation having greater activity partially mimicking lipid activation [20]. Furthermore, the Δ34 enzyme was only weakly activated by PE to approximately 2-fold greater than the enzyme without PE compared to an approximately 6-fold activation with the full-length enzyme (Fig. 3b and Fig. 6c). This indicates that the C-terminal sequence is important for the lipid activation of the Pox enzyme.

Discussion
Here, we demonstrated that the hydrogen peroxideproducing pyruvate oxidase enzyme from L. delbrueckii is catalytically activated specifically by PE, but not other phospholipids, and the last 34 amino acids of the enzyme are important for this activation. To our knowledge, this is the first hydrogenperoxide-producing Pox enzyme that has been shown to be activated by phospholipids. In addition, we showed that the L. delbrueckii Pox forms a pentameric structure in solution of which at least two subunits form a disulfide bond between cysteine 72. Other Pox enzymes form tetrameric structures, and there has been no evidence of disulfide bond formation indicating that the L. delbrueckii Pox has a different structure than other Pox enzymes [1,4].
It is interesting that only PE had an impact on catalytic activity. This suggests that the size, shape, and charge of the phospholipid head group is important for catalytic activation. If the Pox enzyme is interacting with phospholipid that is already incorporated into the membrane, the head group would be exposed to the cytoplasm and be in  the closest proximity to the enzyme. The interaction with the cytoplasm-facing ethanolamine likely occurs through the C-terminal 34 amino acids. Similar to the E. coli Pox, we hypothesize that this interaction induces a conformational shift in the enzyme that enhances electron transfer between FAD and TPP and decreases the phosphate k m . This interaction is likely mediated by electrostatic forces. In fact, it is believed that CidC interaction with phospholipids is driven primarily by electrostatic forces suggesting there is interaction with the headgroup [3]. Furthermore, PE is known to be an abundant component of bacterial membranes, whereas PC is more commonly a eukaryotic and mammalian membrane component [21,22]. Thus, activation only by PE is reasonable since L. delbrueckii Pox is a bacterial enzyme. Enhanced hydrogen peroxide production by L. delbrueckii could be beneficial in a complex microbial community. In fact, hydrogen peroxide production by Pox has been shown to be important in the inhibition of dental pathogens like P. gingivalis [16]. The pentameric structure of L. delbrueckii Pox is different from what has been observed with E. coli Pox and L. plantarum Pox [1,4]. Based on our gel filtration data, the L. delbrueckii also forms a decamer or a dimer of pentamers, but it is unclear which of these may be occurring. Interestingly, we also observed formation of dimers linked by a disulfide bond which are a component of the pentameric structure since they were present in fractions with an elution volume consistent with pentameric molecular weight. Therefore, the structure of the pentamer must be comprised of either three monomers and one dimer or one monomer and two dimers.
E. coli Pox lipid activation is a mixture of K-and Vtype allosteric activation where the pyruvate K m decreases and the k cat increases [19,23]. The lipid activation observed with L. delbrueckii Pox was also a mixture of K-and V-type allosteric activation, however, the specific parameters that changed were different. The pyruvate k m increased slightly while the phosphate k m decreased. The K cat did increase substantially (18-fold) similar to the E. coli Pox. Since the changes in pyruvate and phosphate k m could offset one another, we believe that the lipid activation of L. delbrueckii Pox is primarily driven by increasing the enzyme turnover rate. This is consistent with structural data of E. coli Pox demonstrating that lipid activation's primary effect on the enzyme is to shift a phenylalanine closer to the interface between TPP and FAD to increase the efficiency of electron transfer [9]. This is thought to increase the enzyme turnover rate substantially. It is also in agreement with the activation of CidC where the activation of the enzyme was exclusively driven by a 10-fold increase in k cat [3].
E. coli Pox, CidC, and PQO are the only known lipidactivated Pox enzymes [3,5,7]. These Pox enzymes likely evolved this ability so that they are most active near or at the membrane where it can readily transfer electrons to a membrane electron shuttle [3,9]. For this reason, E. coli Pox inserts and co-precipitates with lipids in the presence of substrate [7]. Likewise, the CidC pyruvate oxidase from S. aureus interacts with phospholipids, but the interaction is believed to be mediated by electrostatic forces indicating that there does not necessarily need to be insertion into the membrane [3]. We did not observe any co-precipitation of L. delbrueckii Pox with PE suggesting that there is no insertion or that the interaction may be unable to withstand centrifugal forces. The fact that L. delbrueckii Pox does not strongly associate with or insert into the membrane may be a mechanism to regulate hydrogen peroxide production. If the association were too strong or there was insertion into the membrane, excessive hydrogen peroxide production may intoxicate the cell.
L. delbrueckii Pox does not need to interact with the membrane since it does not transfer electrons to a membrane electron shuttle, but rather to oxygen. Therefore, the utility of lipid activation of L. delbrueckii Pox is less clear than it is for other pyruvate oxidase enzymes that transfer electrons to membrane-bound carriers. One hypothesis is that the lipid activation of a hydrogen peroxide-producing Pox could be beneficial because it would position hydrogen peroxide production at a peripheral point of the cell. Passive diffusion of hydrogen peroxide from this point would subject the interior of the cell to less hydrogen peroxide and excrete more through the membrane than if the enzyme were positioned in the middle of the cell. Thus, the lipid activation of the Pox enzyme in this case may reduce the self-toxicity of hydrogen peroxide production by L. delbrueckii by activating the enzyme only at the edges of the cell.

Conclusions
In this report, we describe the first hydrogen peroxideproducing Pox enzyme that is activated by the phospholipid phosphotidylethanolamine. We also demonstrated a different oligomeric structure than other known Pox enzymes. Together, these data demonstrate that Pox enzymes from different bacterial species can vary in their structure and function. As Pox is an important enzyme in mediating bacterial interactions and microbial community structure, enhancing pathogenicity, and for certain biosensor applications, further understanding of the biochemical characteristics of Pox are warranted.

Purification of L. delbrueckii pox Δ34
The L. delbrueckii pox deletion mutant was PCR amplified from STYM1 genomic DNA using the primers listed in Table 4. The PCR product was ligated into the pFLAG-CTC vector (Sigma, St. Louis, MO) using the NdeI and XhoI (New England Biolabs, Ipswich, MA) restriction sites. The product was transformed into DH5α E. coli and the construct was confirmed by PCR and sequencing. The construct was transformed into LOBSTR E. coli (Kerafast, Boston, MA), in which the L. delbrueckii Δ34 Pox was expressed as a C-terminal His 6

Pyruvate oxidase activity assay and kinetic parameters
Pyruvate oxidase activity was measured as previously described [16]. Briefly, 50 mM sodium pyruvate, 15 μM FAD (Sigma, St. Louis, MO), 300 μM TPP (Sigma, St. Louis, MO), 0.03% N-ethyl-N-(2-hydroxy-3-sulfopropyl)-m-toluidine (Sigma, St. Louis, MO), 0.015% 4aminoantipyrine (Sigma, St. Louis, MO), and 33 μg/mL horseradish peroxidase (Sigma, St. Louis, MO) were combined with purified pox enzyme. After a 10-min incubation at room temperature, sodium phosphate pH 5.6 was added to 50 mM final concentration to initiate the reaction. The formation of a quinoneimine dye at 25°C was measured by an increase in absorbance at 550 nm and the specific activity was calculated as described. 1 unit of Pox activity is defined as the production of 1 μmol of hydrogen peroxide per minute and this was converted to μM/min for kinetic analysis. For k m determination of pyruvate and phosphate and determination of k cat , dilutions of substrate were used instead of the 50 mM concentration. Enzyme concentration was 1.76 μM. V 0 for each substrate concentration was plotted and the data were fitted using Prism software (Graphpad, San Diego, CA) kcat non-linear regression function. For k m and k cat determination in the presence of phosphotidylethanolamine (PE), PE was used at a concentration of 7.2 mM and the enzyme concentration was 170 nM.

Phospholipid activation assay
Phosphotidylethanolamine (16:0-18:1) (Avanti Lipids, Alabaster, AL), Phosphotidylcholine (16:0-18:1) (Avanti Lipids, Alabaster, AL), and Phosphotidylglycerol (16:0-18:1) (Avanti Lipids, Alabaster, AL) dissolved in chloroform were purchased commercially at a concentration of 10 mg/mL. The desired volume of 10 mg/mL phospholipid solution was evaporated under a steady stream of argon gas (Airgas, Radnor, PA) and dried in a vacuum desiccator for 1 h to evaporate residual chloroform and yield the desired weight of dried phospholipid. The phospholipids were resuspended in the appropriate volume of 20 mM Bis-Tris 150 mM NaCl pH 6 to achieve the desired concentration and incubated at 37°C with shaking at 225 rpm for 30 min. The suspension was vortexed for 30 s and vortexed immediately prior to each use.
Phospholipids were added to the pyruvate oxidase assay mixture at various concentrations and enzyme activity was measured as described above. Fold activity was calculated relative to enzyme activity without phospholipid.

Gel filtration analysis
Prior to analysis of the L. delbrueckii Pox, protein standards were mixed together and applied to the Superdex 200 column (GE healthcare, Chicago, IL) using a duoflow system (Biorad, Hercules, CA). The protein standards were thyroglobulin (670 kDa), gamma-globulin (150 kDa), ovalbumin (44 kDa), myoglobulin (17 kDa), and vitamin B12 (1.35 kDa). Protein was eluted from the column using a mobile phase of 20 mM Bis-Tris pH 6150 mM NaCl at 0.5 mL/min flow rate. A standard curve was plotted using the peak elution volume for each standard versus the Log 10 of the molecular weight in daltons. Eight hundred forty μg of purified L. delbrueckii Pox was incubated in the presence of 50 mM pyruvate, 300 μM TPP, and 15 μM FAD for 10 min at room temperature prior to application to the column. The absorbance at 280 nm was recorded over the elution. One mL fractions were collected after the void volume of 10 mL until the end of the elution. The peak volume for the Pox fractions was used to calculate the molecular weight according to the standard curve.
L. delbrueckii Pox in each fraction from gel filtration was visualized after non-reducing SDS-PAGE analysis with Coomassie staining. For reducing SDS-PAGE analysis, 5% β-mercaptoethanol (Sigma, St. Louis, MO) was included in the sample buffer.
PE co-sedimentation PE was prepared as described above for activation experiments. 3 μg of L. delbrueckii Pox was incubated with 50 mM pyruvate, 300 μM TPP, and 15 μM FAD with either 1 mM PE or no PE for 30 min at room temperature. The PE fraction was precipitated by centrifugation at 16,000 x g for 20 min at 4°C. The supernatant was aspirated as the soluble fraction and the PE pellet was resuspened in an equal amount of buffer. The fractions were visualized by Coomassie staining after SDS-PAGE.