Porphyromonas gingivalis strains used in this study are listed in Table 1, and were grown anaerobically at 37°C in a modified GAM broth (Nissui, Tokyo, Japan) or on Brucella HK agar plates (Kyokuto Pharmaceutical Industrial, Tokyo, Japan), supplemented with 5% rabbit blood. The following antibiotic concentrations were used, as appropriate: 20 μg/ml erythromycin, 0.5 μg/ml tetracycline, and/or 10 μg/ml ampicillin. E. coli DH5α and BL21 (DE3) strains were grown aerobically at 37°C in 2× YT medium (Becton Dickinson Japan, Tokyo, Japan) with 100 μg/ml ampicillin, 200 μg/ml erythromycin, or 10 μg/ml tetracycline.

Porphyromonas gingivalis deletion mutants lacking PGN_1171, PGN_0725, PGN_1341, and/or PGN_1888 gene were constructed by replacing each gene with ermF and ermB (erm cassette), as previously described (Yoshida et al., 2015). Approximately 0.7-kb-long upstream and downstream flanking fragments of each gene were PCR-amplified from genomic DNA of P. gingivalis ATCC 33277 using primers listed in Supplementary Table S1. The erm cassette was amplified from plasmid pVA2198 (Fletcher et al., 1997). The amplicons were joined using mixed PCR (Horton et al., 1989) or the In-fusion HD Cloning kit (Takara Bio, Otsu, Japan), and subcloned into pUC19 (GE Healthcare Japan, Hino, Japan). The resulting plasmids are listed in Table 1. Ligated fragments were introduced into P. gingivalis ATCC 33277 by electroporation. The constructed mutants were designated PGAGU101 (PGN_1171::erm), PGAGU104 (PGN_0725::erm), PGAGU108 (PGN_1888::erm), and PGAGU109 (PGN_1341::erm). To create double mutants, tetracycline resistance gene tetQ, amplified from plasmid pT-COW (Gardner et al., 1996), was used as an additional selectable marker. The tetQ gene and flanking fragments of the targeted genes were joined as described above, and used for electroporation of P. gingivalis single mutant strains to obtain PGAGU111 (PGN_0725::erm PGN_1888::tetQ), PGAGU114 (PGN_1341::erm PGN_1888::tetQ), or PGAGU115 (PGN_0725::tetQ PGN_1341::erm). A triple mutant strain lacking PGN_0725, PGN_1341, and PGN_1888 was also constructed, by replacing PGN_1888 in PGAGU115 with the ampicillin resistance gene cepA. The cepA gene was amplified from plasmid pCEPA, kindly provided by Dr. Naito (Nagasaki University, Nagasaki, Japan). The resultant mutant strain was designated PGAGU118 (PGN_0725::tetQ PGN_1341::erm PGN_1888::cepA). The integration of ligated PCR products at the expected chromosomal location(s) was confirmed by PCR amplification of the specific products across the upstream and downstream insertion boundaries using primers that were designed based on the flanking sequences extraneous to those used for gene targeting.

Crude enzyme extracts were obtained from P. gingivalis ATCC 33277 and its derivatives grown anaerobically in modified GAM broth to log phase. Harvested cells were washed two times with phosphate-buffered saline (PBS), suspended in PBS containing 0.1 mM N-α-tosyl-L-lysine chloromethyl ketone, 0.2 mM phenylmethylsulfonyl fluoride, and 0.1 mM leupeptin (as protease inhibitors), and then lysed by ultrasonication on ice. After ultracentrifugation at 30,000 × g to remove insoluble material, crude enzyme concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA).

Recombinant PGN_0725, PGN_1341, PGN_1888, and PGN_1171 proteins were obtained using the expression vector pGEX-6P-1 (GE Healthcare Japan), as described previously (Yoshida et al., 2015). Coding sequences were PCR-amplified from the genomic DNA of P. gingivalis ATCC 33277 using primers listed in Supplementary Table S1. The amplicons were ligated into pGEX-6P-1 vector, and the resulting plasmids were verified by sequencing and are listed in Table 1. Protein concentrations were determined, as described previously (Pace et al., 1995), and protein purity was assessed by SDS-PAGE.

CoA transferase activity in crude enzyme extracts or purified recombinant proteins was measured by determining the concentrations of acetyl-CoA, a reaction byproduct, with citrate synthase assay (Scherf and Buckel, 1991). Reaction mixtures (40 μl) consisted of 40 mM potassium phosphate buffer (pH 8.0), 200 mM sodium acetate, and 1 mM CoA derivatives (propionyl-CoA, butyryl-CoA, crotonyl-CoA, acetoacetyl-CoA, succinyl-CoA, or 3-hydroxybutyryl-CoA). After pre-warming, reactions were initiated by the addition of enzyme (25 μg/ml crude enzyme extracts, 125 ng/ml PGN_0725, 1.25 μg/ml PGN_1341, or 125 ng/ml PGN_1888), and incubated for 5 min. The reactions were terminated by the addition of 10 μl of 4.5% trichloroacetic acid. To neutralize the solution, 25 μl of 400 mM potassium phosphate buffer (pH 8.0) was added. After neutralization, assay mixtures (25 μl) containing 4 mM oxaloacetate, 4 mM 5,5′-dithiobis-2-nitrobenzoic acid (Sigma-Aldrich, St. Louis, MO, USA), and 36.8 μg/ml citrate synthase (Sigma-Aldrich) were added to the reaction mixtures for acetyl-CoA determinations. After incubation at 37°C for 30 min, the samples were examined spectrophotometrically at 412 nm (A₄₁₂). Acetyl-CoA concentrations were calculated using a standard curve. In addition to the reaction conditions described above, varying concentrations of sodium acetate (20–250 mM) and either butyryl-CoA (0.5–5.0 mM) or propionyl-CoA (0.2–3.0 mM) were used to determine the kinetic parameters of each recombinant protein. These parameters were computed from Lineweaver–Burk transformation (V^-1 vs. S^-1) of the Michaelis–Menten equation. The k_cat values were calculated from V_max and molecular weights of the proteins. Data were obtained from three independent experiments.

Recombinant protein CoA transferase activity with 4-hydroxybutyrate and acetyl-CoA was determined using PGN_0724, a succinate semialdehyde reductase producing 4-hydroxybutyrate from succinate semialdehyde (Yoshida et al., 2015), since 4-hydroxybutyrate is not commercially available. Each reaction mixture (100 μl) contained 40 mM potassium phosphate buffer (pH 8.0), 0.5 mM succinate semialdehyde, 0.5 mM NADH, 0.5 mM acetyl-CoA, 1 mM MnCl₂, and 1 μg/ml recombinant protein (PGN_0724, PGN_0725, PGN_1341, and/or PGN_1888). After incubation for 60 min at 37°C, the concentration of unconsumed NADH was determined spectrophotometrically at A₃₄₀, using a standard curve. The concentration of unconsumed acetyl-CoA was determined using a citrate synthase assay, as described above.

SCFAs in the reaction mixtures and culture supernatants were evaluated by gas chromatography/mass spectrometry (GC-MS) with a ZB-FFAP column (length, 30 m; diameter, 0.32 mm; film thickness, 0.25 μm; Phenomenex Inc., Torrance, CA, USA), as described previously (Yoshida et al., 2015). Briefly, the injection port temperature was set at 150°C. Helium was used as the carrier gas, with a linear velocity of 50 cm/s. The oven temperature was held at 80°C for 1 min, increased to 135°C at a rate of 10°C/min, further increased to a final temperature of 230°C at a rate of 30°C/min, and held for 6 min. For MS, the ionization source temperature was 200°C. The volume injected was 1 μl in splitless mode. To quantify the concentrations of individual SCFAs, experiments were performed in selected ion monitoring mode. Each enzyme reaction mixture (100 μl), containing 40 mM potassium phosphate buffer (pH 8.0), 500 ng/ml recombinant protein, 200 mM sodium acetate, and either 1 mM butyryl-CoA or 1 mM propionyl-CoA, was incubated at 37°C for 60 min. The bacterial supernatant samples were prepared by diluting the bacterial cultures with fresh modified GAM broth to optical density at 600 nm (OD₆₀₀) = 0.375 ± 0.025 (1.5 × 10⁵ CFU/μl) to normalize cell concentration. Diluted bacterial cultures were immediately centrifuged to remove bacterial cells. Aliquots (100 μl) of enzyme reaction mixtures or supernatants were added to 500 μl of acetone. After incubation at -20°C for 2 h, the samples were centrifuged to remove proteins.

Differences between groups were analyzed by one-way analysis of variance, followed by the Student–Newman–Keuls multiple-comparisons test. Differences were considered significant when P < 0.01.

Although PGN_1171 has been previously annotated as a butyryl-CoA:acetate CoA transferase of P. gingivalis ATCC 33277 (Nelson et al., 2003; Hendrickson et al., 2009), the deduced PGN_1171 amino acid sequence shared no more than 20 and 27% identity with R. hominis RHOM_13820 protein and Clostridium aminobutyricum AbfT protein, respectively, both of which were experimentally shown to function as butyryl-CoA:acetate CoA transferases (Scherf and Buckel, 1991; Charrier et al., 2006). In addition, PGN_1171 (220 aa) was much shorter than RHOM_13820 (446 aa) and AbfT (438 aa). A database search revealed greater RHOM_13820 amino acid identity with PGN_0725 (38.5%), PGN_1341 (23.7%), and PGN_1888 (36.4%) proteins, all of which have been annotated as putative CoA transferases in P. gingivalis ATCC 33277 (Naito et al., 2008), than PGN_1171. Moreover, the lengths of PGN_0725 (431 aa), PGN_1341 (498 aa), and PGN_1888 (431 aa) proteins were similar to RHOM_13820 and AbfT.

To evaluate the function of PGN_1171, PGN_0725, PGN_1341, and PGN_1888 proteins, eight deletion strains, including double and triple mutants, were constructed using erythromycin, tetracycline, and ampicillin resistance genes (Table 1). Integration of the resistance genes at the expected chromosomal locations in the resulting mutant strains was verified by PCR (Supplementary Figure S1).

Butyrate and acetyl-CoA are produced from butyryl-CoA and acetate in the last step of butyrate production (Figure 1). Butyryl-CoA:acetate CoA transferase activity of crude enzyme extracts was determined by measuring the production of acetyl-CoA from butyryl-CoA and sodium acetate. The initial velocity of acetyl-CoA production by PGAGU101 (PGN_1171::erm) crude enzyme extracts was not significantly different from that of the wild type (Figure 2A), indicating that PGN_1171 has no detectable butyryl-CoA:acetate CoA transferase activity. By contrast, disruption of PGN_0725, PGN_1341, or PGN_1888 resulted in a significant decrease in the reaction velocity, suggesting that all the encoded proteins possess the butyryl-CoA:acetate CoA transferase activity.

Likewise, propionate and acetyl-CoA were produced from propionyl-CoA and acetate (Figure 1). The abilities of crude enzyme extracts of PGAGU101, PGAGU109 (PGN_1341::erm), and the wild type strain to form acetyl-CoA from propionyl-CoA and acetate were not significantly different (Figure 2B). The initial velocities of PGAGU104 (PGN_0725::erm) and PGAGU108 (PGN_1888::erm) crude enzyme extracts were significantly lower than that of the wild type. These findings demonstrated that PGN_0725 and PGN_1888, but not PGN_1171 and PGN_1341, possess the propionyl-CoA:acetate CoA transferase activity.

To characterize PGN_0725, PGN_1341, PGN_1888, and PGN_1171 protein functions, each corresponding gene was expressed in E. coli to produce recombinant proteins. Molecular masses of the denatured PGN_0725, PGN_1341, PGN_1888, and PGN_1171 polypeptides agreed well with the respective predicted molecular masses (47.4, 54.7, 47.8, and 23.2 kDa, respectively; Figure 3).

GC-MS analysis demonstrated that incubation of butyryl-CoA and sodium acetate with any of the recombinant PGN_0725, PGN_1341, and PGN_1888 proteins led to the production of butyrate (Figure 4). Propionate was detected in reaction mixtures containing propionyl-CoA, sodium acetate, and either recombinant PGN_0725 or PGN_1888, but not PGN_1341. As expected, incubation of the recombinant PGN_1171 protein with butyryl-CoA or propionyl-CoA in the presence of sodium acetate resulted in no detectable production of butyrate or propionate, respectively (Figure 4).

We also investigated the kinetic parameters of recombinant PGN_0725, PGN_1341, and PGN_1888 using a colorimetric assay. Initial velocities were determined at either fixed sodium acetate concentrations at different butyryl-CoA concentrations, or fixed butyryl-CoA concentrations at different sodium acetate concentrations. Lineweaver–Burk (double reciprocal) plots were then constructed (Figure 5). Line series intersecting to the left of y-axes (V^-1) suggested that PGN_0725, PGN_1341, and PGN_1888 proteins act via a ternary-complex kinetic mechanism. K_m and V_max values were estimated from secondary plots (Figures 5B,C; Table 2). In addition, k_cat values were calculated from enzyme concentrations in the reaction mixtures. Kinetic properties of the recombinant PGN_0725 and PGN_1888 proteins were also determined by analyzing acetyl-CoA production from sodium acetate and propionyl-CoA (Table 3). These reactions also follow the ternary-complex mechanism (Figure 6).

To investigate substrate specificity of the recombinant proteins, four CoA derivatives (crotonyl-CoA, acetoacetyl-CoA, succinyl-CoA, and 3-hydroxybutyryl-CoA) were evaluated as substrates, in addition to butyryl-CoA and propionyl-CoA (Supplementary Figure S2). When the CoA derivatives tested were incubated with PGN_0725, PGN_1341, or PGN_1888 in the presence of sodium acetate, no significant production of acetyl-CoA was detected, suggesting that crotonyl-CoA, acetoacetyl-CoA, succinyl-CoA, and 3-hydroxybutyryl-CoA were not used as a substrate by these enzymes.

Next, sodium butyrate and sodium propionate (instead of sodium acetate) were tested as enzyme substrates in the presence of propionyl-CoA and butyryl-CoA, respectively. GC-MS analysis revealed that the incubation of a sodium butyrate/propionyl-CoA pair or sodium propionate/butyryl-CoA pair with PGN_0725, PGN_1341, or PGN_1888 resulted in no detectable production of propionate or butyrate (data not shown). These findings suggested that the three CoA transferases utilize no SCFA substrates other than acetate.

Since PGN_0725 was annotated as a 4-hydroxybutyrate CoA transferase (Nelson et al., 2003; Naito et al., 2008; Hendrickson et al., 2009), an enzyme catalyzing the formation of 4-hydroxybutyryl-CoA from 4-hydroxybutyrate and acetyl-CoA (Figure 1), we examined the 4-hydroxybutyrate CoA transferase activity of recombinant PGN_0725 (Supplementary Figure S3). In this assay, the recombinant PGN_0724 that produces 4-hydroxybutyrate from succinate semialdehyde was used (Yoshida et al., 2015), because the 4-hydroxybutyrate is not commercially available. When recombinant PGN_0724 was incubated with NADH and succinate semialdehyde, NADH was consumed, indicating 4-hydroxybutyrate production. However, incubation of a mix of recombinant PGN_0724 and PGN_0725 proteins with the reaction mixture resulted in no significant acetyl-CoA consumption, suggesting that PGN_0725 does not produce 4-hydroxybutyryl-CoA from 4-hydroxybutyrate and acetyl-CoA. Incubation of either recombinant PGN_1341 or PGN_1888 (instead of PGN_0725) in the reaction mixture did not result in acetyl-CoA consumption (data not shown). These findings suggested that PGN_0725, PGN_1341, and PGN_1888 did not function as 4-hydroxybutyrate CoA transferases.

PGAGU108 grew as fast as the wild type, as determined by OD₆₀₀ measurements of the cell cultures (Figure 7). By contrast, the growth rates of other mutant strains were obviously slower than that of the wild type, as were culture turbidities in the stationary phase. This suggested that the intermediates and end products of the butyrate, propionate, and related molecule biosynthetic pathways might be important for the metabolism of this microorganism.

To evaluate the contribution of PGN_0725, PGN_1341, and PGN_1888 to SCFA production, SCFA concentrations in stationary phase bacterial cultures were quantified using GC-MS. The OD₆₀₀ of each culture was adjusted to 0.375 ± 0.025 to normalize cell concentrations. This conversion allowed us to compare the production of SCFAs per cell number in the wild type and mutant strains (Figure 8). Butyrate concentrations in culture media of all the mutant strains were significantly lower than that of the wild type (P < 0.01). Butyrate concentrations in the culture supernatants of PGAGU104, PGAGU111 (PGN_0725::erm PGN_1888::tetQ), PGAGU115 (PGN_0725::tetQ PGN_1341::erm), and PGAGU118 (PGN_0725::tetQ PGN_1341::erm PGN_1888::cepA), all of which lacked the PGN_0725 gene, were significantly lower, not only than that of the wild type, but also than those of other mutants. Likewise, propionate concentrations in culture supernatants of PGAGU108, PGAGU111, PGAGU114 (PGN_1341::erm PGN_1888::tetQ), and PGAGU118, all of which lacked the PGN_1888 gene, were significantly lower than those of other strains. Deletion of PGN_0725, PGN_1341, or PGN_1888 resulted in a drastic decrease in isobutyrate and isovalerate concentrations in culture supernatants (Figure 8). Interestingly, the triple mutant, PGAGU118, still released butyrate, propionate, isobutyrate, and isovalerate into the culture supernatant (Figure 8).