A Porphyromonas gingivalis Periplasmic Novel Exopeptidase, Acylpeptidyl Oligopeptidase, Releases N-Acylated Di- and Tripeptides from Oligopeptides*

Exopeptidases, including dipeptidyl- and tripeptidylpeptidase, are crucial for the growth of Porphyromonas gingivalis, a periodontopathic asaccharolytic bacterium that incorporates amino acids mainly as di- and tripeptides. In this study, we identified a novel exopeptidase, designated acylpeptidyl oligopeptidase (AOP), composed of 759 amino acid residues with active Ser615 and encoded by PGN_1349 in P. gingivalis ATCC 33277. AOP is currently listed as an unassigned S9 family peptidase or prolyl oligopeptidase. Recombinant AOP did not hydrolyze a Pro-Xaa bond. In addition, although sequence similarities to human and archaea-type acylaminoacyl peptidase sequences were observed, its enzymatic properties were apparently distinct from those, because AOP scarcely released an N-acyl-amino acid as compared with di- and tripeptides, especially with N-terminal modification. The kcat/Km value against benzyloxycarbonyl-Val-Lys-Met-4-methycoumaryl-7-amide, the most potent substrate, was 123.3 ± 17.3 μm−1 s−1, optimal pH was 7–8.5, and the activity was decreased with increased NaCl concentrations. AOP existed predominantly in the periplasmic fraction as a monomer, whereas equilibrium between monomers and oligomers was observed with a recombinant molecule, suggesting a tendency of oligomerization mediated by the N-terminal region (Met16–Glu101). Three-dimensional modeling revealed the three domain structures (residues Met16–Ala126, which has no similar homologue with known structure; residues Leu127–Met495 (β-propeller domain); and residues Ala496–Phe736 (α/β-hydrolase domain)) and further indicated the hydrophobic S1 site of AOP in accord with its hydrophobic P1 preference. AOP orthologues are widely distributed in bacteria, archaea, and eukaryotes, suggesting its importance for processing of nutritional and/or bioactive oligopeptides.

Porphyromonas gingivalis, a Gram-negative obligate anaerobe, has been implicated as the causative agent of chronic periodontal disease (1,2), which is a major reason for permanent tooth loss (3)(4)(5). Recently, much attention has been paid to this bacterium due to its association with systemic diseases, such as cardiovascular disorders (6), decreased kidney function (7), and rheumatoid arthritis (8). P. gingivalis does not ferment carbohydrates but rather metabolizes amino acids to produce ATP by a putative respiratory chain of fumarate respiration without oxygen (9). In addition, it incorporates nutritional amino acids not as single amino acids but preferentially as di-and tripeptides and short oligopeptides, possibly via inner membraneassociated oligopeptide transporters (9,10).
Escherichia coli XL-1 Blue cells carrying expression plasmids were cultured in Luria-Bertani broth supplemented with 75 g/ml ampicillin at 37°C. Recombinant proteins were induced with 0.2 mM isopropyl-thiogalactopyranoside at 30°C for 4 h and then purified using Talon affinity chromatography as reported previously (27).
Measurement of Peptidase Activity-Peptidase activity was measured using peptidyl-MCA as reported previously (24,27). Briefly, the reaction was started by the addition of recombinant proteins (5-100 ng), a periplasmic cell fraction (1-5 l), or P. gingivalis cell suspensions (1-5 l) in a reaction mixture (200 l) composed of 50 mM sodium phosphate (pH 7.5), 5 mM EDTA, and 20 M peptidyl-MCA. After 30 min at 37°C, fluorescence intensity was measured with excitation at 380 nm and emission at 460 nm. In some experiments, pH values varied from 5.5 to 9.5 with 50 mM phosphate (pH 5.5-8.5) or Tris-HCl (pH 7.0 -9.5), and NaCl concentrations varied from 0 to 1.6 M. To determine enzymatic parameters, recombinant proteins were incubated with various concentrations of peptidyl-MCA. The data obtained were analyzed using a nonlinear regression curve fitted to the Michaelis-Menten equation with GraphPad Prism (GraphPad Software, Inc., La Jolla, CA). Values are shown as the average Ϯ S.D. and are calculated from four independent measurements.
Subcellular Fractionation-All procedures were carried out at 4°C according to a method reported previously (30), with a sight modification. Briefly, a 20-ml culture of P. gingivalis in the log phase was centrifuged at 6000 ϫ g for 15 min, and then the extracellular fraction was obtained from the supernatant by fil-tration with a 0.20-m membrane filter. Bacterial cells were washed with ice-cold PBS, resuspended in 4 ml of 0.25 M sucrose in 5 mM Tris-HCl (pH 7.5), and then left on ice for 10 min. Cells were precipitated at 12,500 ϫ g for 15 min, resuspended in 4 ml of 5 mM Tris-HCl, and mixed gently for 10 min to disrupt the outer membrane. The supernatant was obtained by centrifugation at 6000 ϫ g for 15 min and collected as the periplasmic fraction containing the outer membrane. The obtained spheroplasts as precipitate were resuspended in PBS and then disrupted in an ice-water bath by sonication pulse 10 times for 10 s each with 2-s intervals. The cytosol and inner membrane fractions were separately prepared by ultracentrifugation, as described previously (24).
Size Exclusion HPLC-Recombinant proteins and the periplasmic fraction were subjected to size exclusion HPLC using ÄKTA explorer 10S (GE Healthcare) with a Super-dex200 increase 10/300 column (1.5 ϫ 30 cm) equilibrated with 20 mM Tris-HCl (pH 8.0) containing 1 mM EDTA. They were separated with the identical buffer at a rate of 0.75 ml/min at room temperature, and then 0.5-ml fractions were collected. Aliquots of the fractions were subjected to a peptidase assay with Z-VKM-MCA as well as SDS-PAGE or native PAGE.
N-Terminal Sequencing of Proteins-Proteins separated on SDS-PAGE and transferred to a Sequi-Blot membrane (Bio-Rad) were stained with Coomassie Brilliant Blue and subjected to protein sequencing by use of a model Procise 49XcLC protein sequencer (Applied Biosystems).
Three-dimensional Homology Modeling-To search for an ideal template for homology modeling, the P. gingivalis AOP sequence was submitted to the Psipred server (31) using the GenThreader method (32) for domain and fold assignment. Afterward, a BLAST search (33) against the Protein Data Bank was performed to identify the closest homologue to P. gingivalis AOP with known three-dimensional structure. The model was generated via the Phyre2 server (Protein Homology/Analogy Recognition Engine) (34), using the tool "one-to-one threading," in the expert mode. The coordinates of Pyrococcus horikoshii acylaminoacyl peptidase (AAP) (35) (Protein Data Bank code 4HXE) and the P. gingivalis AOP sequence were submitted to structural alignment and further model building. The final model covered around 85% of the sequence, sharing 20% identity with the template.

PGN_1349 Responsible for Remaining Activity in NDP212-
We previously reported that Met-Leu-MCA-hydrolyzing activity was markedly decreased in a dpp4-5-7 triple knock-out P. gingivalis strain (NDP211). However, unexpectedly, that activity was reversed by introduction of a fourth dpp gene disruption to NDP211, resulting in NDP212, a dpp4-5-7-11 quadruple knock-out strain (24). These results indicated that the bacterium additionally expresses a peptidase related to Met-Leu-MCA hydrolysis in addition to DPP5 and DPP7 and suggested an up-regulation mechanism to compensate for the loss of dipeptide production. To identify this entity, the peptidase activity of NDP212 was reexamined in detail with various MCA substrates (portion of results shown in Fig. 1A). NDP212 showed the most efficient hydrolysis of Met-Leu-MCA, followed by benzyloxycarbonyl (Z)-LLL-MCA and succinyl (Suc)-LLVY-MCA, suggesting that the candidate is not a DPP but probably an oligopeptidase with a hydrophobic P1 preference. These activities were observed in a cell-associated form and not with the culture supernatant and were efficiently inhibited by diisopropyl fluorophosphates, suggesting a serine peptidase, the same as previously reported periplasmic DPPs (data not shown).
Taking into consideration these properties, along with information from the KEGG and MEROPS databases (29,36), we focused on three uncharacterized peptidases among the seven S9 family members (i.e. PGN_1349 (KEGG definition: DPP), PGN_1542 (esterase), and PGN_1878 (hypothetical protein)) and expressed them as C-terminal His 6 -tagged proteins in E. coli. Purified PGN_1349, PGN_1542, and PGN_1878 migrated as 85-, 35-, and 50-kDa species, respectively, in accordance with their calculated molecular masses (Fig. 1B). Subsequent peptidase analysis using 80 peptidyl-MCAs available in our laboratory revealed that PGN_1349 hydrolyzed several substrates. In contrast, attempts to detect peptidase activities in PGN_1542 and PGN_1878 have not been successful to date.
PGN_1349 is currently annotated as a DPP, although it was found to be hydrolyzed by both N-terminally blocked and These results indicate that PGN_1349 is not a DPP but rather an oligopeptidase that is able to liberate N-terminally blocked peptides with a P1 preference for hydrophobic residues. Interestingly, its substrate preference profile resembles that of NDP212, except for no activity of PGN_1349 against Suc-LLVY-MCA. In addition, NDP212 scarcely hydrolyzed Z-VKM-MCA, whereas PGN_1349 showed efficient hydrolysis. We speculated that the reason for the low level of Z-VKM-MCA hydrolysis by NDP212 was due to the activity of Kgp, which preceded the hydrolysis of Z-VKM-MCA into Z-Val-Lys and Met-MCA, in which the latter was scarcely degraded by PGN_1349 (Fig. 1C). In fact, the level of hydrolysis of Z-VKM-MCA was even higher than that of Z-LLL-MCA by the gingipain-null strain KDP136 (data not shown). At this time, we are unable to explain the activity against Suc-LLVY-MCA in NPD212, although it may be due to an unidentified peptidase. Taken together, we concluded that most hydrolyzing activities observed in NDP212 can be mainly attributed to PGN_1349.
Distribution of PGN_1349 Orthologues in Organisms-PGN_1349 in P. gingivalis ATCC 33277 and its homologues in other P. gingivalis strains have been annotated under various names, such as DPP (ATCC 33277) (29), prolyl oligopeptidase (W83) (10), putative DPP (TDC60) (37), and peptidase S9 A/B/C family catalytic domain protein (JCV1 SC001) (38). In the MEROPS peptidase database, PGN_1349 is classified into Clan SC, family S9, subfamily C of the peptidase S09.A77, for which the dipeptidylpeptidase family member 6 (DPF-6) from Caenorhabditis elegans (UniProt number P34422) is a represen- min -1. tative. An orthologue search of PGN_1349 produced a long list of S9 family peptidase members (supplemental Table S1). This family comprises at least 142 genera of organisms, in which most orthologues are distributed in both Gram-negative and -positive bacteria, such as the phyla Actinobacteria, Proteobacteria, Bacteroidetes, Cyanobacteria, and Firmicutes, but are also found in archaea (the phyla Euryarchaeota and Crenarhaeota), including Aeropyrum pernix, and eukaryotes (the phylum Nematoda), including C. elegans. The average amino acid length of the 477 DPF-6 members is 652 Ϯ 62 (mean Ϯ S.D.). DPF-6 from C. elegans is composed of 740 amino acids (39) and postulated to have serine-type peptidase activity, whereas no actual activity in any family members has yet been reported. Alignment of the amino acid sequences indicates that the Ser-Asp-His catalytic triad and their adjacent sequences are highly conserved (Fig. 2). The peptidase S9 family motif is predicted to be present in the C-terminal half region in both PGN_1349 and DPF-6 by the Pfam database (40). The N-terminal part corresponding to Met 16 -Lys 142 of PGN_1349 is absent in C. elegans DPF-6. Importantly, this sequence identity between PGN_1349 and C. elegans DPF-6 (27.3%, INT/OPT score ϭ 132/653) is even higher than that with C. elegans DPF-5/AAP (23.3%, 82/292), which releases acylaminoacyl amino acids, or E. coli oligopeptidase B (19.8%, 24/123), which releases acyloligopeptides from the N terminus of peptides (Fig. 2, bottom). These findings strongly suggest that the enzymatic activities of DPF-6 family members are represented by that of P. gingivalis PGN_1349 and enzymatically distinct from DPF-5/AAP and oligopeptidase B.
PGN_1349 Is a Novel Exopeptidase, Designated AOP-The peptidase activity of PGN_1349 is rearranged based on the residue numbers of 26 MCA substrates. It showed predominant release of tripeptides with N-terminal blockage. Other substrates, such as acylaminoacyl-, dipeptidyl-, and N-terminally blocked tetrapeptidyl-MCA, were slowly hydrolyzed, whereas aminoacyl-MCA was scarcely hydrolyzed (Fig. 3A). To more precisely evaluate the substrate length preference, we prepared a set of substrates with altered lengths (i.e. Z-Met-, Z-Lys-Met-, Z-VKM-, and Z-AVKM-MCA) and then determined the activities (Fig. 3B). PGN_1349 most potently hydrolyzed Z-Lys-Met-MCA, followed by Z-VKM-MCA. The hydrolysis of Z-AVKM-MCA was lower as compared with the former, and Z-Met-MCA was scarcely hydrolyzed. Taken together, we concluded that PGN_1349 is neither a DPP nor AAP, although it has acyl-oligopeptide-releasing activity. In order to more precisely evaluate the influence of N-terminal blockage on its enzymatic activity, the hydrolysis of four pairs of substrates, with the N-terminal either blocked or unblocked, was examined. As shown in Table 2, PGN_1349 consistently hydrolyzed the acylated substrates more efficiently than the substrates without N-terminal modification. Hence, we designated PGN_1349 as AOP.
The P1 position amino acid preference of AOP was further examined using synthetic dipeptidyl-MCA or N-terminally modified tripeptidyl-MCA with different P1 amino acids. When the activities for these substrates were plotted against P1 amino acid residual hydrophobicity indexes (41), a close positive relationship with P1 position hydrophobicity was demonstrated with both the di-and tripeptidyl-MCA substrates (Fig.  3C). These results clearly indicate that AOP prefers hydrophobic residues at the P1 position.
The pH profile of AOP revealed an optimum pH of 8.  gradually decreased as the concentration of NaCl increased. Therefore, to achieve direct comparison with various substrates, the activity was simply measured in 50 mM sodium phosphate buffer, pH 7.5, containing 5 mM EDTA in the absence of NaCl in the following experiments. We also tested the effects of reducing reagents (i.e. dithiothreitol, reduced glutathione, and oxidized glutathione) because it has been reported that they modulate the peptidase activity of oligopeptidase B (42). However, none (up to 1.6 mM) affected the peptidase activity of AOP (data not shown), again indicating characteristics distinct from oligopeptidase B. The enzymatic parameters of AOP for the three substrates indicate that the k cat value was maximal for Z-Lys-Met-MCA (1155 Ϯ 77 s Ϫ1 ), whereas K m was smallest for Z-VKM-MCA (4.9 Ϯ 1.0 M) and k cat /K m was maximal for Z-VKM-MCA (123.3 Ϯ 17.0 M Ϫ1 s Ϫ1 ) ( Table 3). When the values were compared with those of P. gingivalis DPPs for their best dipeptidyl-MCA substrates, the k cat /K m values of AOP (58 -123 M Ϫ1 s Ϫ1 ) were shown to be comparable with those of DPP4, DPP5, DPP7, and DPP11 (11-548 M Ϫ1 s Ϫ1 ), suggesting that AOP and the DPPs are able to function under similar substrate concentrations in the bacterium.
Peptidase Activities in aop Gene-disrupted Strain-An aopdisrupted strain, NDP600, was constructed, and the absence of AOP in the strain was confirmed by immunoblotting (Fig. 4A). Cell-associated peptidase activity of NDP600 was determined in the presence of 50 M N ␣ -p-tosyl-L-lysine chloromethyl ketone (TLCK) and 3 M E-64, under which the activity of Kgp toward Z-HEK-MCA was completely abolished in both the wild-type and NDP600 strains (Fig. 4B), whereas the addition of these gingipain inhibitors markedly enhanced Z-Lys-Met-MCA hydrolysis in the wild type. In contrast, this activity was greatly reduced in NDP600 both with and without the inhibitors (Fig. 4C). These results clearly demonstrated that AOP is the main peptidase responsible for this hydrolysis. Similarly, the hydrolysis of Z-LLL-MCA was significantly reduced in NDP600 (Fig. 4D). However, hydrolyses of Z-VKM-and Z-AVKM-MCA were modestly reduced in NDP600, presumably because of incomplete protection of the substrates from Kgp. On the other hand, that of Met-Leu-MCA was not affected by disruption of the aop gene, because this substrate was predominantly hydrolyzed by DPP5 (24) and DPP7 (25), which are active in NDP600. Taken together, these results suggest that AOP is primarily responsible for the release of di-and tripeptides with P1 hydrophobic residues from N-terminally blocked oligopeptides in P. gingivalis.
Active Serine and Endopeptidase Activity-Recombinant AOP was mainly purified as an 85-kDa species with the N-ter-minal sequence of ala Ϫ2 -ile Ϫ1 -Met 16 -Thr-Val-His-Ala-Gln-Lys-Ile 23 (ala Ϫ2 -ile Ϫ1 was derived from the vector sequence) due to release of the initial met Ϫ3 (Figs. 1 and 2). In addition, a minor 75-kDa species with the N-terminal sequences of Thr 89 -Phe-Lys-Gly-Thr-Leu 94 and Asn 102 -Val-Glu-Gln-Met-Gly 107 was occasionally obtained, indicating cleavage at the Phe 88 -Thr 89 and Gly 101 -Asn 102 bonds, respectively (Figs. 2 and 5A). These results suggested that the N-terminal region at around the 100th residue is protease-sensitive. Similarly, 75-and 70-kDa species were produced by limited proteolysis with chymotrypsin at the Phe 108 -Ser 109 and Phe 117 -Glu 118 bonds, respectively (data not shown).
We compared the peptidase activities of several forms of AOP, in which either the N-terminal 101 residues were truncated (Asn 102 form), the potentially essential Ser at position 615 was substituted by Ala (S615A form), or both. The S615A substitution completely abolished the activity (Fig. 5B). These results indicate that Ser 615 is essential, as previously proposed (36). The N-terminally truncated Asn 102 form maintained onethird of the activity, suggesting that the N-terminal 101 residues are not essential, although they have effects on peptidase activity.
Interestingly, 75-, 52-, and 42-kDa species emerged in enzymatically active forms and were increased by incubation at 37°C for 6 days (Fig. 5A), suggesting an autoproteolytic activity of AOP. In accord with the P1 hydrophobic preference of AOP activity, the cleavage sites of the 52-and 42-kDa species were Leu 290 -Tyr 291 and Ala 414 -Tyr 415 , respectively. Endopeptidase activity of AOP was further confirmed because it degraded 14.3-kDa ␣-lactalbumin into 7.5-and 8.5-kDa species at the Gln 62 -Asn 63 and Gln 58 -Ala 59 bonds, respectively (data not shown) and was estimated to be less than one-one hundredth of that of ␣-chymotrypsin when assessed with ␣-lactalbumin as a substrate.
Oligomerization and Subcellular Localization-We also examined the molecular structure of AOP in an aqueous solution using size exclusion HPLC (Fig. 6). The hydrolyzing activity of purified AOP was split into a major peak larger than 440 kDa and a minor 85-kDa peak. Fractions 14 -17 of the first HPLC consisted solely of the 85-kDa species, as shown by SDS-PAGE, whereas the second peak at fractions 22-25 consisted of both the 85-and 75-kDa entities. Native PAGE confirmed that the 85-kDa species in the first peak formed oligomers and that the 85-and 75-kDa species in the second peak were monomers (m1 and m2, respectively). Furthermore, the second HPLC of the large peak sample (fraction 15 of the first HPLC) consisting of the 85-kDa species reproduced the two-peak profile. These findingsclearlyindicatedanequilibriumofoligomericandmono- In P. gingivalis KDP136 cells, the apparent molecular mass of native AOP was found to be 78 kDa (Fig. 7), suggesting the loss of some region as compared with 85-and 75-kDa AOP, although the N terminus of the 78-kDa form could not be determined. AOP at 78 kDa was predominantly detected in the periplasm/outer membrane and cytosol fractions. When the periplasmic/outer membrane fraction was subjected to size exclusion HPLC, 78-kDa AOP was detected as a monomer at fractions 23-26. Because the cytosolic preparation contained a considerable amount of periplasmic components, P. gingivalis AOP appears to be primarily distributed in the periplasm and present mainly as a monomer in cells. In accord with these findings, a previous proteome analysis also found that PG_1004, an orthologue of PGN_1349 in P. gingivalis W83, was mainly recovered in cytosol/periplasm and periplasm fractions (43).
Modeling of Three-dimensional Structure-Psipred and Gen-Threader results indicate that AOP possesses three domains: Met 16 -Ala 126 , which has no similar homologue with known three-dimensional structure, and Leu 127 -Met 495 and Ala 496 -Phe 736 , which form ␤-propeller and ␣/␤-hydrolase domains, respectively. The three-dimensional model comprises residues Ala 138 -Lys 759 and bears the domain organization of S9 family members, as indicated by GenThread. The ␣/␤-hydrolase domain hosts the catalytic triad (Ser 615 , Asp 702 , and His 734 ) and is responsible for the enzymatic activity, whereas the ␤-propeller provides enzyme specificity (Fig. 8, A and B).
In agreement with the hydrophobic P1 preference, the S1 site is formed by the hydrophobic residues: Pro 536 , Ile 583 , Tyr 619 , Val 641 , Ile 644 , Phe 647 , and Trp 666 (Fig. 8C). Differently from DPPs, which have an N-anchor region determining the protein specificity (44, 45), our model indicates a more open active site without any steric impediment for its oligopeptidase feature. The autoproteolytic site (Gly 101 -Asn 102 ) associated with the production of the 75-kDa truncated AOP is localized in the N-terminal domain, which suggests that this region is very likely exposed to the solvent, reasonably explaining the site accessibility by another AOP molecule.
To estimate the evolutionary conservation of amino acid positions in AOP, we performed a Psi-BLAST (33) search with the AOP sequence, followed by multialignment in Cobalt (46), and mapped their conservation onto the AOP structure, using the server Consurf (47). This analysis revealed that the ␣/␤hydrolase domain is more conserved than the ␤-propeller domain (Fig. 8D), in agreement with the observation that the ␤-propeller domain is responsible for enzyme specificity. Therefore, the ␤-propeller domain is expected to present more variation among the proteins than the ␣/␤-hydrolase domain, which provides the catalytic activity shared by all members. Furthermore, because the oligomer interface region lies in the N terminus, we hypothesize that AOP might display a novel oligomer formation process. In this way, determining the threedimensional structure of AOP by an experimental method (e.g. x-ray crystallography) could shed light upon its mechanism of action.

Discussion
In the present study, we identified a novel exopeptidase encoded by the gene PGN_1349 in the P. gingivalis strain ATCC 33277. This enzyme, designated AOP, was found to possess active Ser 615 , to most preferentially degrade peptide bonds between Yaa 2 -Xaa 3 and Yaa 3 -Xaa 4 (where Yaa is a hydrophobic amino acid) of oligopeptides with N-terminally blocked oligo-  peptides, and to exist mainly in the periplasmic space as a monomer. Furthermore, AOP showed a tendency to oligomerization at the N-terminal 100-amino acid region; thus, it may bind to other molecules via interactions in this region. Our three-dimensional modeling results also indicated a conserved structure of S9 peptidases coinciding with the hydrophobic P1 preference of AOP. Peptidase activities of three of seven members categorized as P. gingivalis S9 family proteins have been identified in previous studies (i.e. DPP4, DPP5, and PTP-A) (14,15,24,27), whereas the present study identified the fourth, AOP. Whether the remaining three (i.e. PGN_1542 (annotated as esterase), PGN_1694 (Ala-DPP), and PGN_1878 (hypothetical protein)) possess hydrolyzing activity remains unknown (24) (this study); however, we confirmed here that they could not release N-acy-lated amino acids from MCA substrates available in our laboratory. Therefore, to the best of our knowledge, AOP is a unique exopeptidase that hydrolyzes nutritional N-terminally blocked peptides in P. gingivalis. The activity of AOP may provide substantial benefits for efficient utilization of amino acids by the bacterium. Notably, when considering its habitat in the subgingival space, the bacterium frequently utilizes serum proteins contained in gingival crevicular fluid, some of which are N-terminally blocked by acyltransferase or glutamyl cyclase by processing into their secretary forms (48).
Our comparison of amino acid sequences supported the kinship of AOP with C. elegans DPP-6 (S9.A77). On the other hand, sequence homologies were rather lower among AOP, C. elegans DPF-5/AAP, and E. coli oligopeptidase B, and their peptidase characteristics are obviously distant. Although both C, stick representation of catalytic triad (yellow) and the residues forming the S1 subsite (magenta). D, surface representation of AOP colored in terms of evolutionary conservation of its amino acids. Psi-BLAST, Cobalt, and Consurf were used to calculate the conservation scores. Left, side view of AOP. Middle, top view of the ␣/␤-hydrolase domain; the ␤-propeller was removed for better visualization. Right, top view of the ␤-propeller domain. The color code ranges from blue (non-conserved) to red (conserved). The figure was prepared using the program PyMOL (Schroedinger, LLC, New York).
AOP and DPF-5/AAP exhibit a hydrophobic P1 preference, a typical feature of mammalian (S09.004) and Aeropyrum-type (S09.070) AAP is removal of an N-acylated hydrophobic amino acid from oligopeptides with various N-terminal acyl groups (49,50). Because of this activity, AAP is also termed an acylamino acid-releasing enzyme. In contrast, AOP was shown to scarcely release acylamino acids but preferably released di-and tripeptides (Fig. 1). Furthermore, the P1 hydrophobic preference of AOP is clearly distinct from oligopeptidase B, which targets basic amino acid residues, such as Arg (51). On the other hand, both share common features in that these exopeptidases poorly exhibit aminopeptidase activity and possess elevated activities against N-acylated substrates (52) (this study). Therefore, DPF-6 family members, including P. gingivalis AOP, should be considered as the third exopeptidase group in the S9 family responsible for N-terminally blocked peptidyl substrates. Although AOP possessed a weak endopeptidase activity, this activity was also reported on AAP (53).
In supplemental Table S1, Bacteroides peptidases, such as Bacteroides fragilis (BF9343_3151, 23.4% amino acid identity), Bacteroides vulgatus (BVU_4083, 21.6%), and Bacteroides thetiotaomicron (BT_1838, 23.8%) are listed as homologues of P. gingivalis AOP (PGN_1349). Among them, BT_1838 and BVU_4083 are annotated as Ala-DPP in the S9 family, of which the enzymatic activity remains unknown (24). The amino acid identities of P. gingivalis Ala-DPP (PGN_1694) with BF_ 9343_3151 from B. fragilis, BVU_4083, and BT_1838 are 39.1, 39.4, and 40.2%, respectively, which are substantially higher than those with P. gingivalis AOP. In fact, we cloned BF_9343_3151 and expressed its recombinant form in E. coli as an 80-kDa protein. It did not show any peptidase activity (data not shown), as P. gingivalis Ala-DPP did not (24). Therefore, we conclude that not all proteins of supplemental Table S1 are AOP, especially those with lower similarities. The final annotation of AOP should be carefully done in future studies. In addition, a study on peptidase activity of Ala-DPP is now under way in our laboratory.
Subcellular fractionation experiments demonstrated that P. gingivalis AOP is localized in the periplasmic space as a monomer. It has been reported that S9 family members are present as various molecular forms (50), because oligopeptidase B and prolyl oligopeptidase were found to exist as monomers (42), and DPP4 (54) and PTP-A (17) were found to exist as dimers, whereas the dimeric structure of human DPP4 is indispensable for peptidase activity (54). In addition, mammalian and A. pernix AAPs are present as a tetramer (55) and dimer (56), respectively. P. gingivalis DPP5 also forms a homodimer (24). In this study, we found that recombinant AOP exists in equilibrium between a large oligomer and monomer. It is interesting to note that two forms of AOP demonstrated peptidase activity, and the specific activities of both seem to be equal. Furthermore, our findings indicated that the N-terminal region is indispensable for oligomerization, although the biological meaning of AOP oligomerization is obscure at present. This region might be related to heteromeric interactions with periplasmic components, such as other peptidases and peptidoglycan.
Genome and biochemical analyses of P. gingivalis demonstrated the presence of two oligopeptide transporters (9, 10) and a Ser/Thr transporter (57). Previous studies reported that di-and tripeptides are predominantly incorporated into cells (18,19); thus, they appear to be major cargoes for the oligopeptide transporters, and the contribution of the Ser/Thr transporter toward nutritional amino acid incorporation may be limited. Regarding this point, there is a critical difference between P. gingivalis and E. coli, a well studied Gram-negative rod that does not possess DPPs and predominantly incorporates single amino acids (58,59). It is possible that DPPs/PTP-A and oligopeptide transporters co-evolved, which may explain, at least in part, why novel exopeptidases (i.e. DPP11 (27), DPP5 (24), and AOP (this study)) have recently been discovered in P. gingivalis.
Based on our findings, we describe the metabolism of extracellular oligopeptides in P. gingivalis as follows. P. gingivalis produces oligopeptides from extracellular polypeptides by potent cysteine peptidases, Kgp and Rgp, located in the outer membrane and extracellular space. Subsequently, oligopeptides are converted in periplasmic space into di-and tripeptides by DPPs and PTP-A, respectively, with DPP5 and DPP7 functioning with oligopeptides with hydrophobic residues at the second position from the N terminus. Additionally, the P2 position hydrophobic residue enhances the activities of DPP7 (26). Those with Pro at the second and third positions are cleaved by DPP4 and PTP-A, respectively. Oligopeptides with the penultimate acidic residues Asp and Glu are metabolized by DPP11, whereas those with the basic residues Lys and Arg are metabolized by gingipains. Moreover, N-terminally blocked oligopeptides are segregated by AOP into acylated di-and tripeptides and N-terminally unblocked oligopeptides, which are then processed by DPPs and PTP-A. These di-and tripeptides are promptly incorporated via oligopeptide transporters in the inner membrane. Efficient utilization of extracellular proteins as carbon and energy sources is made possible by coordination of a series of exopeptidases and nutritional oligopeptide transporters in P. gingivalis, which may have inevitably evolved to choose and adapt its anaerobic subgingival habitat, where carbohydrates are consumed mainly by oral streptococci.