Phenylalanine 664 of dipeptidyl peptidase (DPP) 7 and Phenylalanine 671 of DPP11 mediate preference for P2-position hydrophobic residues of a substrate

Dipeptidyl peptidases (DPPs) are crucial for the energy metabolism in Porphyromonas gingivalis, a Gram-negative proteolytic and asaccharolytic anaerobic rod causing chronic periodontitis. Three DPPs, DPPIV specific for Pro, DPP7 for hydrophobic residues and DPP11 for Asp/Glu at the P1 position, are expressed in the bacterium. Like DPP7, DPP11 belongs to the S46 protease family, and they share 38.7% sequence identity. Although DPP11 is preferential for hydrophobic residues at the P2 position, it has been reported that DPP7 has no preference at the P2 position. In the present study, we defined the detailed P2 substrate preference of DPP7 and the amino acid residue responsible for the specificity. DPP7 most efficiently hydrolyzed Met-Leu-dipeptidyl-4-methylcoumaryl-7-amide (MCA) carrying hydrophobic residues at the P1 position with kcat/Km of 10.62 ± 2.51 μM−1 s−1, while it unexpectedly cleaved substrates with hydrophilic (Gln, Asn) or charged (Asp, Arg) residues. Examination with 21 dipeptidyl MCA demonstrated that DPP7-peptidase activity was dependent on hydrophobicity of the P2- as well as P1-position residue, thus it correlated best with the sum of the hydrophobicity index of P1- and P2-amino acid residues. Hydrophobicity of the P1 and P2 positions ensured efficient enzyme catalysis by increasing kcat and lowering Km values, respectively. Substitution of hydrophobic residues conserved in the S46 DPP7/DPP11 family to Ala revealed that Phe664 of DPP7 and Phe671 of DPP11 primarily afforded hydrophobic P2 preference. A modeling study suggested that Phe664 and Gly666 of DPP7 and Phe671 and Arg673 of DPP11 being associated with the P2- and P1-position residues, respectively, are located adjacent to the catalytic Ser648/Ser655. The present results expand the substrate repertoire of DPP7, which ensures efficient degradation of oligopeptides in asaccharolytic bacteria.


Introduction
Porphyromonas gingivalis, a Gram-negative proteolytic and asaccharolytic anaerobe, is a major causative organism of chronic periodontitis [1,2], which leads to loss of permanent teeth [3][4][5] and, thus, affects quality of life, especially in elderly individuals [6]. Much attention has recently been paid to this bacterium because of its close ଝ This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. relationship with systemic diseases including cardiovascular disorders [7], decreased kidney function [8] and rheumatoid arthritis [9]. P. gingivalis does not ferment glucose, cellobiose, lactose or sucrose [10], while it requires proteinaceous substrates as carbon and energy sources. Amino acids are incorporated into the bacterial cells mainly as dipeptides, not as free amino acids [11,12]. Therefore, dipeptidylpeptidases (DPPs) that release dipeptides by cleavage of the penultimate peptide bond (P1 position) from the N-terminus of an oligopeptide are crucial for bacterial metabolism [13][14][15]. If dipeptides with any sequences can be released by DPPs, it appears to be advantageous for energy production by the bacterium.
To date, three DPPs belonging to serine protease family have been identified to be expressed as cell-bound forms in P. gingivalis. The S9B-family peptidase DPPIV (Uniprot ID, B2RKU3/MEROPS ID, MER004211) is specific for Pro at the P1 position [13] and basically accepts any residues at the N-terminus position (P2 position) [16], though bulky P2-position residues are optimal for the activity [17]. DPP7 (B2RKV3/MER014366) is highly preferential for hydrophobic  residues at the P1 position and its activity appears to be not affected by the P2-position residue [14]. Our recent study indicated that DPP7 shows a rather broad specificity, in which moderate hydrolysis of Leu-Arg-and Leu-Asp-MCA was not simply explained by the hydrophobic P1 preference [18]. In addition to these DPPs, we recently found DPP11 (B2RID1/MER34628) that is specific for Asp/Glu and possesses a hydrophobic P2 preference [15]. DPP7 and DPP11 share sequence similarity, and hence, both belong to the S46 peptidase family. The optimum pH of DPPIV, DPP7 and DPP11 are pH 7.5 [13], 7-8 [14] and 7.0 [15], respectively, which seem to be adapted to its inhabiting environment.
Substitution analysis of conserved residues in respective S46 members demonstrated that Arg673 of DPP11 is primarily important for recognition of the P1 residue, Asp or Glu [15]. Similarly, Gly666 that is conserved in DPP7 and equivalent to Arg673 of DPP11 mediates P1 preference for bulky and hydrophobic residues [18].
In addition to above three DPPs, the gene (PGN 1645) encoding putative DPPIII (YP 004510152.1) has been found in P. gingivalis ATCC 33277 [19]. Although DPPIII has been considered to exclusively exist in eukaryotes [20], bacterial DPPIII has recently reported in Bacteroides thetaiotaomicron [21]. B. thetaiotaomicron DPPIII most potently hydrolyzes Arg-Arg-2-naphthylamide as does eukaryotic DP-PIII [20]. However, its N-terminal sequence indicates its localization in the cytosol [21], and hence, bacterial DPPIII may not participate in the degradation of environmental peptides, but be responsible for degradation of cytosolic substrates.
In the present study, we defined detailed P1 and P2 preferences of DPP7 using a preparation free from other P. gingivalis peptidases. For this purpose, we expressed active recombinant DPP7 and its single substitution mutants in Escherichia coli. Measurements of activity with a series of dipeptidyl 4-methylcoumaryl-7-amide (MCA) demonstrated that DPP7 does not simply prefer hydrophobic residues at the P1 position, while its hydrolyzing activity is influenced by both P1and P2-amino acids. Substitution experiments suggested that Phe664 of DPP7 and equivalent Phe671 of DPP11 were responsible for recognition of the P2-position hydrophobic residue of substrates.

Construction of expression plasmids
P. gingivalis ATCC 33277 genomic DNA was purified as previously reported [23]. A DNA fragment encoding DPP7 (PGN 1479, Met3-Ile712 was amplified by PCR with a set of primers [5 -GACTGCTAGATCTCAAATGAAATTAAAAAGTATTC and 5 -GGCGAATAGATCTGATCAACTTCAGCTC (BglII sites indicated by italics)], digested with BglII, and then cloned into the BamHI site of pQE60. Construction of the expression plasmid for P. gingivalis DPP11 was previously reported [18].

Expressions of P. gingivalis DPP7 and DPP11 in E. coli
E. coli XL1-Blue carrying a pQE60-derived expression plasmid was cultured in Luria-Bertani broth containing 75 μg/ml of ampicillin at 37 • C overnight. Protein expression was induced by dilution of the culture with two volumes (800ml in a 3-L baffled flask) of the broth containing 0.2 mM isopropyl β-d-thiogalactopyranoside and incubation at 30 • C for 4 h. Bacterial cells were harvested by centrifugation and lysed with lysis/washing buffer (20 mM Tris-HCl, pH 8.0, 0.1 M NaCl containing 10 mM imidazole) to which 0.4 mg/ml of lysozyme and 10 μg/ml of leupeptin had been added. Recombinant proteins were recovered in the cell lysate fraction and then purified by an affinity chromatography with Talon metal affinity resin (Clontech Lab. Inc., Palo Alto, CA) according to the manufacturer's protocol, except that 10 mM imidazole was included in the lysis/washing buffer. After extensive washing, bound proteins were eluted with 0.1 M imidazole (pH 8.0) containing 10% (v/v) glycerol. Expressed DPPs were stored at −80 • C until used.

Hydrolyzing activity toward dipeptidyl MCA
Recombinant DPPs (50 ng of DPP7 and 5 ng of DPP11) were incubated with 20 μM dipeptidyl MCA in 200 μl of reaction solution composed of 50 mM sodium phosphate (pH 7.0) and 5 mM EDTA at 37 • C for 1 h. Fluorescence intensity was measured with excitation at 380 nm and emission at 460 nm with a Fluorescence Photometer F-4000 (Hitachi). To determine the enzymatic parameters, different concentration of dipeptidyl MCA (0.5-100-μM) were incubated with wild type and single amino acid mutated DPP7. The results were analyzed by nonlinear regression curve fitting to the Michaelis-Menten equation using a GraphPad Prism software program (San Diego, CA, USA). Averages and standard deviations were calculated from three to nine independent measurements.

Homology modeling
Models of DPP7 and DPP11 were built using an automated homology modeling program EsyPred3D [24], with the known structure of GluV8 chain 'A' (2o8I) [25] as a template. Obtained structures were visualized using PyMOL.

Substrate specificity of P. gingivalis DPP7
Recombinant DPP7 was expressed in E. coli and purified to homogeneity with an apparent molecular mass of 70 kDa (data not shown). Substrate specificity of DPP7 was compared on six dipeptidyl MCA substrates, and catalytic rate constant (k cat ) and K m were determined from the concentration dependence of the reaction, such as shown in Fig. 1. Enzymatic parameters are listed in Table 1. DPP7 showed larger specificity constants (k cat /K m ) for the substrates carrying hydrophobic residues [Leu, Tyr and Phe with hydrophobicity indexes (H.I.) [26] [16] and DPP11 from P. endodontalis [15] were compared, and the residues conserved in more than 95% (capital letters), 80% ≤ x < 95% (capital and italic letters) and 70% ≤ x < 80% (small letters) members are described below the sequences of DPP11 and DPP7. Conserved (closed arrows) and non-conserved (open arrows) hydrophobic amino acids were substituted with Ala. Stars indicate essential His89, Asp196 and Ser648 forming an active triad. Triangles indicate maturation cleavage sites producing mature DPP7 [14] and DPP11 [15]. A plus mark indicates Gly666 of DPP7 and Arg673 of DPP11 essential for P1-position recognition [15,18]. A homologous region of GluV8 (Thr164-Pro189) to the S46 family is described, and amino acids identical to either P. gingivalis DPP7 or DPP11 are underlined. (B) Activities of DPP7 and its mutants were measured with Met-Leu-, Leu-Arg-and Gly-Phe-MCA. Inset shows expanded data for F637A and F664A. Values are shown as the mean ± SD (n = 3). (C) The ratio of the activity for Gly-Phe-MCA to that for Leu-Arg-MCA is plotted. of 97, 49 and 100, respectively], than those carrying a less hydrophobic residue, Ala (41), and hydrophilic residues, Asp (-55) and Glu (-31), at the P1 position. Furthermore, the k cat /K m values seemed to vary by the P2 residues, since H.I. of Leu in Met-Leu-MCA was comparable to Phe in Gly-Phe-MCA, whereas the figure for Met-Leu-MCA was 9fold higher than that for Gly-Phe-MCA. We thus further compared the hydrolyzing activity of DPP7 with a number of substrates at a single concentration of 20 μM.
The DPP7 activities for 21 dipeptidyl MCA substrates exhibited a rather broad spectrum (Fig. 2), in contrast to the narrow specificity of DPPIV [13] and DPP11 [15]. Among the substrates examined, Met-Leu-MCA was most efficiently cleaved, as it carries one of the most hydrophobic residues (H.I. = 97) at the P1 position. Furthermore, DPP7 unexpectedly cleaved Leu-Gln-and Leu-Arg-MCA, despite the fact that they carry non-hydrophobic P1 residues. Subsequently, Ser-Tyr, Gly-Phe-, Ala-Asn-and Lys-Ala-MCA were efficiently hydrolyzed with specific activities higher than 100 pmol min −1 μg −1 . Thereafter, Leu-Asp-, Leu-Glu-, Thr-Ser-and Arg-Arg-MCA were cleaved with that higher than 10 pmol min −1 μg −1 ; and Val-Arg-, Val-Asp-, Leu-Lys-, Thr-Asp-and Ile-Asp-MCA to lesser extents (>1 pmol min −1 μg −1 ). The remaining substrates were scarcely hydrolyzed. These results indicated that DPP7 targeted the peptides not only with hydrophobic but also with hydrophilic (Gln and Asn), as well as charged (Arg, Lys, Asp and Glu) residues at the P1 position. In particular, it was notable that DPP7 hydrolyzed Leu-Gln-MCA 4.5-fold more efficiently than Lys-Ala-MCA, and that the presence of a hydrophobic side chain (Leu) at the P2 position was common in these hydrolyzed substrates. Accordingly, we further characterized the substrate specificity of DPP7 by taking into consideration the P2-position residues.

P1-and P2-position preferences of DPP7
The preference of DPP7 for hydrophobic P1 amino acids was confirmed using a plot of activity against the H.I. of amino acid residues (Fig. 3) in consistent with the previous study [14]. Moreover, a weak but positive correlation with the hydrophobicity of P2-position residues was observed, hence the DPP activity was best correlated with the sum of H.I. of the P1-and P2-position residues.
To clearly elucidate the influence of the P2-position residues, the H.I. values of P2-position residues were compared among dipeptidyl MCA carrying common P1 residues, i.e., Asp, Arg or Lys (Fig. 4). In all cases, the activities for Xaa-Asp-(Xaa = Leu, Val, Ile, Thr or His), Xaa-Arg-(Xaa = Leu, Val, Arg or Gly) and Xaa-Lys-MCA (X = Leu or Lys) were proportional to the H.I. values of the P2-position residues, though the statistical significance of the correlations was higher than 0.05 due to the limited number of substrates available. Taken together, we concluded that the activity of DPP7 was dependent on hydrophobicity not only at the P1, but also at the P2 position. Accordingly, it was reasonably explained that Met-Leu-MCA possessing the highest H.I. value (173) was the best substrate for DPP7 (Figs. 2 and 3). Furthermore, these results revealed that the hydrophobic preference at the P2-position residue was a common characteristic of the S46 family, i.e., DPP7 and DPP11 [15].
An important role of the P2-position hydrophobicity in the DPP7 activity was also supported by the fact that small K m values were commonly observed for the substrates carrying hydrophobic residues at the P2 position, i.e., Met-Leu-, Leu-Asp-and Leu-Glu-MCA (Table 1). To the contrary, large k cat values were common for substrates carrying hydrophobic residues at the P1 position, i.e., Met-Leu-, Ser-Tyrand Gly-Phe-MCA. Taken together, these results indicated that hydrophobicity of the P1-and P2-position residues contributed mainly an increase in k cat and lowering of K m values, respectively, in DPP7.

Phe644 defines hydrophobic P2 preference of DPP7
The hydrophobic P2 preference of DPP7 led us to suspect that hydrophobic residue(s) are involved in P2-position recognition, which may be conserved in S46 family members. Alignment of the amino acid sequences of all 264 members (MEROPS database, Release 9.6) [16] demonstrated that there are nine conserved hydrophobic residues; with Leu83 and Phe664 completely conserved in all members; Trp29, Val200, Phe637, Leu693 and Leu705 highly conserved (80% ≤ x < 95%); and Met28 and Leu659 relatively conserved (70% ≤ x < 80%) (Fig. 5A). If any of these hydrophobic residues (Phe, Trp, Leu, Val and Met with H.I. values of 100, 97, 97, 76 and 74, respectively) truly participates in interactions with the P2 residue, then substitution to a less hydrophobic residue, Ala (H.I. = 41), may reduce the proteolytic activity. In particular, the reduction may be more significant for a substrate with a hydrophobic residue at the P2 position, whereas it may be limited for a substrate carrying a non-hydrophobic To test this assumption, six single substitution mutants (L83A, F664A, V200A, F637A, L693A, L659A) of nine conserved hydrophobic residues were constructed. Three residues, i.e., Met28, Trp29 and Leu705, were excluded from this analysis, because they are located near the N-or C-terminal edge of the molecule, and hence are unlikely to be involved in substrate recognition. We also prepared three additional Phe-to-Ala-substitution mutants (Phe69, Phe226 and Phe229, which are conserved less than 70%) as controls. As shown in Fig. 5B, the activities for Met-Leu-, Leu-Arg-and Gly-Phe-MCA of the mutants were consistently reduced as compared to the wild type (wt). The ratio of the activities for Gly-Phe-MCA to Leu-Arg-MCA was the highest in F664A (Gly-Phe/Leu-Arg = 4.62), followed by F637A (1.08), F659A (0.66) and L693A (0.60), while those values for the wt, F69A, L83A, V200A, F226A and F229A, were substantially lower (0.23-0.42) (Fig. 5C). These findings strongly suggest the involvement of Phe664 in hydrophobic P2 recognition. This mutation decreased specificity constant for Met-Leu-MCA through affecting both k cat and K m ( Table  1). Even though involvement of Phe637 as well as Leu659 and Leu693 was also surmised, their effects seemed to be limited as compared to Phe664.

Phe671 defines hydrophobic P2 preference of DPP11
Because DPP11 as well as DPP7 shows hydrophobic P2 preference [15], we also investigated the amino acid residue(s) of DPP11 responsible for the P2 hydrophobic preference. Recombinant DPP11 was expressed in E. coli and purified to homogeneity with an apparent molecular mass of 80 kDa, as reported previously [18]. Two residues (Leu79 and Val202) of DPP11 were omitted from comparison after considering the above-described result for DPP7. Hence, four conserved and highly hydrophobic residues (Phe644, Leu666, Phe671 and Leu700) were substituted with Ala. Since either Asp or Glu was indispensable at the P1 position, we chose Leu-Asp-(H.I. of Leu = 97) and Thr-Asp-MCA (H.I. of Thr = 13) as substrates carrying P2 residues with contrast values for this assay. As shown in Fig. 6A, the hydrolyzing activity of Leu-Asp-MCA by DPP11 wt was much higher than that of Thr-Asp-MCA, reflecting the hydrophobic P2 preference of DPP11. Since L700A showed a comparable activity to wt, Leu700 was excluded from possible residues required for substrate specificity. The activities of L666A and F671A were markedly decreased, and F644A completely lost activity toward both substrates. The ratio of the activities for Thr-Asp-MCA to Leu-Asp-MCA was the highest in F671A (Fig. 6B). Therefore, Phe671 of DPP11 probably interacts with the P2-position hydrophobic residue as observed in DPP7 (Fig.  5B). However, since disturbing the local conformation or secondary structure could cause the decrease in activity, another method, e.g., three-dimensional structure determination, would be necessary to confirm a direct interaction.

Modeling of substrate-binding sites of DPP7 and DPP11
The three-dimensional structures of the S46 family members DPP7 and DPP11 are yet to be established. Hence, we applied an automated homology modeling program, EsyPred3D [24], to examine the structural relationship between essential Ser648 and Phe664 involved in P2-residue recognition. In addition, it was considered important to examine the residue at position 666, because Arg673 in DPP11 equivalent to Gly666 of DPP7 is predicted to interact with a P1-position acidic residue [15], and Gly666 of DPP7 is required for optimal acceptance of a P1-position bulky and hydrophobic residue [18]. The modeling program automatically chose the specificity-determining C-terminal conserved region of the S46 family [18] using a scaffold template of the Staphylococcus aureus glutamyl endopeptidase GluV8 [25], which shows amino acid identity of 7.5% to DPP7 and 9.2% to DPP11 in the full sequence, and that of 15.7% to DPP7 and 16.2% to DPP11 in the modeled region. The topology of the active triad of GluV8, His51, Asp93 and Ser169, is identical to that of His89, Asp196 and Ser648 of DPP7, and residues 164-173 of GluV8 (numbering as mature form) [27] carrying active Ser169 shares sequence similarity with DPP7 and DPP11 (Fig. 5A). As a result, Asp566-Met695 of DPP7 and Asp572-Val702 of DPP11 were modeled based on the structure of GluV8. The model structure indicated that the tripeptidyl segments Phe664-Asp665-Gly666 (DPP7) and Phe671-Ap672-Arg673 (DPP11) are sterically adjacent to essential Ser648 and Ser655, respectively (Fig. 7). The importance of these segments was also indicated by complete conservation of Phe664/Phe671 and Asp665/Asp672 in the S46 family [16].

Discussion
Classification of DPPs from DPPI to DPP11 has generally been performed based on differences in substrate specificities [16], in which the P1-position residues seem to be most critical. The previous study on substrate specificity of P. gingivalis DPP7 had also focused on the P1 position, but was unable to refer to P2 position due to limited number of substrates used. Later in a review article, the authors noted that P2-position specificity of DPP7 should be carefully re-examined [28]. (Asp572-Val702) was performed with an EsyPred3D program based on the structure of GluV8 (2o8I) [25]. Essential Ser648 of DPP7 and Ser655 of DPP11 and residues involved in the hydrolytic activity are indicated.
Even though DPP11 is specific for acidic amino acids, its proteolytic activity is also influenced by the hydrophobicity of the P2position residues. For example, hydrolysis of Ile-Asp-MCA was 10fold faster than that of His-Asp-MCA [15]. Moreover, the present findings revealed the hydrophobic P2 preference of DPP7 and, accordingly, DPP7 more preferably cleaves substrates with higher H.I. values of P1 and P2 residues. In fact, the similar tendency that a hydrophobic P2 preference of P. gingivalis DPP7 was also observed in the previous data [14], as N-terminal and subsequent residues of cleavable substrates had higher H.I. values [Phe-Leu-, Ala-Phe-, Ile-Ala-, Trp-Ala-, Gly-Phe-and Lys-Ile-β-naphthylamide hydrochloride (pNA) with 197, 141, 140, 138, 100 and 76, respectively], whereas those of un-cleavable peptides were relatively lower (Trp-His-, Ala-Pro-and Gln-Lys-β-pNA with 105, -5 and -33, respectively). Thus, the principle that the hydrophobicity of the dipeptide portion determines the cleavability may be applicable to those substrates, except for Lys-Ile-(H.I. = 76), which is cleavable, and Trp-His-(H.I. = 105), which is not cleavable.
Three-dimensional structures of the S46 family members DPP7 and DPP11 are yet to be established. To date the structures of DPPs have been thoroughly determined in human DPPIII [29] that belongs to the metallopeptidase family M49 [16] and in human DPP7 [30] that belongs to the S28 DPPII family primarily specific for Pro. The present and previous studies [15,18] demonstrated that Phe664 and Gly666 of P. gingivalis DPP7/Phe671 and Arg673 of DPP11 are responsible for P2-and P1-position substrate recognition, respectively. Therefore, the three-dimensional model of DPP7 and DPP11 (Fig. 7) demonstrating adjacent locations of these residues to essential Ser is reasonably accepted.
If the presented models around essential Ser648 of DPP7 and Ser655 of DPP11 are appropriate, then Trp185 of GluV8, which is equivalent to Phe664 of DPP7, may also be involved in substrate recognition (Fig. 5A). It was previously shown that Trp185 of GluV8 is located adjacent to essential Ser169 and forms a substrate-binding pocket [25]. Moreover, W185Y substitution in GluV8, which naturally occurs in the GluV8-family member in Staphylococcus epidermidis [31], caused a reduction of the activity due to a 13.8-fold increase in the K m value [32]. These findings also support the appropriateness of the model.
In conclusion, our results clarify the detailed substrate preference of DPP7 and DPP11 taking P2 preference into consideration, and further identified Phe664/Phe671 as residues participating in the S2 subsite of both DPPs. These findings expand current knowledge regarding the repertoire of peptide substrates managed by DPPs of P. gingivalis, and accordingly, by other proteolytic and asaccharolytic bacteria expressing DPP7 and DPP11.