Broadened substrate specificity of bacterial dipeptidyl-peptidase 7 enables release of half of all dipeptide combinations from peptide N-termini

2.1 Activity enhancement by prime-side amino acids for DPP4, DPP5, and DPP7

To investigate the effects of the P1′ residue in detail, hydrolysis of HAXD-MCA, where X represents 12 different amino acids, was measured using a two-step method (Figure 1, Table S1). Under the conditions employed, when the examined DPP hydrolyzes the substrate into His-Ala and Xaa-Asp-MCA, the latter is readily converted into Xaa-Asp and 7-amino-4-methylcoumarin (AMC) by an excess amount of DPP11. DPP11 does not hydrolyze tetrapeptidyl substrates, while it exhibits activity exclusively toward Xaa-Asp-MCA. Consequently, hydrolysis of tetrapeptidyl-MCA was attributed to the first-order reaction of the target DPP (Ohara-Nemoto et al. 2022). However, one might raise a question that the activity of the first-step DPP might be affected by the second-step DPP11 activity. In fact, DPP11 prefers hydrophobic amino acids at the N-terminus (Ohara-Nemoto et al. 2011; Rouf et al. 2013b). Thus, we here show the results with HAMD- and HAED-MCA as controls among tetrapeptidyl substrates examined in Figure 1B. The hydrolyzing activity of DPP7 (10 ng) increased dependent on the amount of DPP11 and thereafter reached a plateau over 3 µg. Therefore, we concluded that the DPP7 activity is defined as the plateau stage independent of the DPP11 activity. Consequently, the amount of DPP11 was fixed at 5 µg in the following experiments.

Figure 1:

P1′ preference for DPP4, DPP5, and DPP7 (A) The two-step process on the breakdown of HAXD-MCA is shown. Fluorescent AMC group is indicated in blue (B) HAMD- and HAED-MCA were incubated with various amounts of DPP11 in the absence and presence of DPP7 (10 ng) at 37 °C for 30 min (C) His-Ala- and HAXD-MCA were digested by DPP7 (10 ng), or (D) DPP4 (25 ng) and DPP5 (25 ng) in the presence of DPP11 (5 μg) at 37 °C for 30 min. values (mean ± SD, n = 3) shown represent a typical example of three separate experiments (E) replotted hydrolysis findings for His-Ala- and HAED-MCA by DPP4, DPP5, and DPP7 shown in B and C. Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments (F) Relationships of activities for HAXD-MCA between two DPPs were compared. Regression lines for DPP4–DPP5, DPP4–DPP7, and DPP5–DPP7 are y = 1.842x + 178.6 (R² = 0.104), y = 2.678x + 59.10 (R² = 0.787), and y = 0.237x + 148.6 (R² = 0.202), respectively.

As shown in Figure 1C, DPP7 scarcely degraded His-Ala-MCA, while it hydrolyzed HAXD-MCA with P1′ Met, Leu, Ser, Ala, Val, Tyr, Gln, Asn, and Glu to varying degrees. Activities toward substrates with P1′ Lys, Ile, and Phe were minimal (Figure 1C). These results confirmed the previous finding that release of His-Ala from GLP-1 by DPP7 is increased by the presence of an appropriate P1′ residue. In addition, they are the first to show activity enhancement by the P1′ residue in DPP4 and DPP5, both of which belong to the S09 family (Figure 1D). The increase in activity was modest in DPP4 and most prominent with P1′ Met, followed by Leu, Tyr, Gln, and Ser. DPP5 activity was significantly increased with P1′ Ser, and Met, followed by Asn, Leu, and Ala. Therefore, at least towards substrates with P1 Ala, activity enhancement by a P1′ residue was shown to be common among the three DPPs. In addition, despite structural similarities between Leu and Ile, activities toward HALD-MCA were increased in the three DPPs, while those for HAID-MCA were quite low, possibly due to differences in the structural entropy of these side chains (Creamer and Rose. 1992). Similarly, significant differences in activities for HAYD- and HAFD-MCA were found. These results again indicated a substantial role for the P1′-position amino acid in regard to these activities. Furthermore, findings showing that the P1′ residue had the least effect on DPP4 activity likely reflects its strict preference for P1 Pro followed by Ala, in contrast to the more moderate specificities observed for DPP5 and DPP7.

Hydrolytic activities of the three DPPs toward His-Ala- and HAED-MCA that mimic the N-terminal sequence (underlined) of active GLP-1 (NH₂-His⁷-Ala⁸-Glu⁹---) are depicted in Figure 1E. These activities matched GLP-1 cleavage efficiency, i.e., DPP5<DPP4<<DPP7, previously demonstrated by MALDI TOF-MS analysis (Ohara-Nemoto et al. 2022). Taken together, it is concluded that the inactivation potentials of the three DPPs for GLP-1 are primarily determined by their allowance for P1′ Glu. Among the relationships between the activities of DPP4 and DPP5, DPP4 and DPP7, and DPP5 and DPP7 toward the same substrates, only those of DPP4 and DPP7 were closely related (y = 2.678x + 59.10, R² = 0.787), indicating similar P1′ preferences (Figure 1F). This latter result was unexpected, since DPP4 and DPP7 exhibit completely different P1 specificities and no amino acid sequence similarity (11.6 %). Currently, there is no appropriate explanation for the similar P1′ preferences of DPP4 and DPP7, though the findings clearly indicated no correlation between the HI of P1′ amino acids and their activities.

Involvement of the P2′-position residue in DPP activity was examined using a novel method with pentapeptidyl-MCA in combination with P. gingivalis prolyl-tripeptidyl peptidase A (PTP-A) as the second-step enzyme (Figure 2). PTP-A has been shown to specifically recognize the third Pro from the N-terminus and liberate an N-terminal tripeptide (Banbula et al. 1999; Ito et al. 2006). When HA-|-AFP-|-MCA was incubated with DPP7 (100 ng) together with increasing amounts of PTP-A, hydrolysis reached a plateau at 15 μg of PTP-A. As shown in Figure 2C, DPP4 and DPP7 activities toward HAAXP-MCA (X = Phe, Tyr, Ile, Leu, Val, and Glu) were increased as compared to His-Ala-MCA, while HAAEP-MCA was found to be hydrolyzed by DPP4 but minimally by DPP7. Consequently, both DPPs exhibited similar P2′ preferences (y = 1.779x − 117.2, R² = 0.964), with P2′ Phe showing the greatest level of activity, followed by Tyr, Ile, Leu, and Val. Moreover, a weak correlation between the activities and HI value regarding the P2′ residue was observed for both DPP4 and DPP7 (Figure 2D and E). In contrast, the activity of DPP5 was minimal with all of the substrates examined, suggesting its poor acceptance of P3′ Pro. Taken together, the present results are the first to indicate that the hydrophobic P2′ residue has a positive role in DPP4 and DPP7 activities.

Figure 2:

Involvement of P2′ residue in DPP activity, and comparisons of contributions of P1′ and P2′ residues (A) The chemical formula and cleavage sites of HAAXP-MCA used for the two-step method are shown (B) HAAFP-MCA was incubated with PTP-A in the absence (closed circle) or presence (open circle) of DPP7 (0.1 µg). DPP7 activity (blue circle) was calculated by subtraction of hydrolysis in the absence of DPP7 from that in its presence. Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments (C) His-Ala- and HAAXP-MCA were incubated with DPP4, DPP5, or DPP7 (0.1 µg) in the presence of PTP-A (15 µg). Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments (D) specific activities of DPP4 and DPP7 for HAAXP-MCA (X = Phe, Tyr, Ile, Leu, Val, or Glu) plotted against the HI values of P2′-position amino acids (E) coefficient of variation for P1′ and P2′ positions, as described in the text, was calculated using data shown in Figures 1 and 2, respectively (mean ± SD (n = 3). * p <0.001.

To examine the contributions of the P1′ and P2′ positions in DPP activity, it was speculated that if a position has an effect on the activity, then it will vary depending on the amino acids at the position. On the other hand, if the position does not affect the activity, the activity amount should remain constant regardless of the amino acids present. Thus, the importance of a substrate position should be evaluated using coefficient of variation, determined based on the standard deviation of activities of substrates with various amino acids divided by the average of specific activities toward those substrates. It was considered that obtained value would be highest for a protease at an essential position, such as the P1 position of GluV8 (Nemoto et al. 2008) and trypsin and the P1′ position of thermolysin. Analyses of the results shown in Figure 1C and D for the P1′ residue, and in Figure 2C for the P2′ residue showed that the coefficient of variation for the P1′ position of the three DPPs was consistently higher than that for the P2′ position (Figure 2E), indicating that the P1′ position has a more significant effect on DPP activity. These findings are reasonable, due to the proximity of the P1′ position to the cleavage site, and also consistent with crystallography data previously presented, which showed that residues at the P1′ position are tightly bound to DPP11, while those at the P2′ position are partially exposed (Bezerra et al. 2017).

2.2 DPP7 responsible for liberation of dipeptides with P1-neutral amino acids and hydrophilic Asn

To investigate DPP activity toward refractory P1 amino acids with an HI value lower than Ala (HI = 41), HXLD-MCA [X indicates His (HI = 8), Gln (−10), or Asn (−28)] was synthesized. In contrast to findings obtained with hydrolysis of HALD-MCA, the three HXLD-MCA compositions were highly resistant to DPP7 (Figure 3B). However, following prolonged incubation for up to 24 h with a large amount (1 μg) of DPP7, gradual hydrolysis of the substrates was observed (Figure 3C), suggesting that DPP7 has potential to hydrolyze those. Under these conditions, DPP7 predominantly hydrolyzed the three substrates, while DPP5 exhibited a lower level of hydrolysis only for HNLD-MCA and DPP4 showed no activity (Figure 3D).

Figure 3:

Hydrolysis of HXLD-MCA containing refractory P1 residue (A) Chemical formula and cleavage sites of HXLD-MCA used for the two-step method are shown (B) Dipeptidyl- and HXLD-MCA were incubated at 37 °C for 30 min with DPP7 (25 ng for Tyr-Ala- and HALD-MCA, 1 µg for others) in the presence of DPP11 (5 μg) (C) HHLD-, HQLD-, and HNLD-MCA were incubated with DPP11 in the absence (cont) or presence of DPP7 (1 μg) at 37 °C for up to 24 h (D) Tetrapeptidyl-MCA was incubated with DPP4, DPP5, or DPP7 (1 μg) in the presence of DPP11 at 37 °C for 24 h (E) Left. Peptidyl-MCA was incubated with wild-type P. gingivalis cells (10 µl of A₆₀₀ = 10) in the presence of DPP11. Right (magnified). Values for HHLD-, HQLD-, and HNLD-MCA are shown (F) Substrates were incubated with wild-type P. gingivalis and dpp-gene disrupted cells (10 µl of A₆₀₀ = 10) in the presence of DPP11. Activities are presented as percent ± SD (n = 3), with those for wild-type P. gingivalis considered to be 100 %. Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments. *¹ p <0.05, *² p <0.005 (as compared to wild-type P. gingivalis).

Hydrolysis of Phe-Met-MCA and tetrapeptidyl-MCA with P1 Ala (HALD-, HAMD-, and HASD-MCA) was observed in P. gingivalis wild-type organisms. In accordance with results of recombinant DPPs, the level of hydrolysis of substrates with P2 His and refractory P1 residue (HHLD-, HQLD-, or HNLD-MCA) were quite low in the wild-type cells (Figure 3E). When hydrolysis was evaluated using dpp-gene disrupted strains (Ohara-Nemoto et al. 2011; 2014), the dpp7-gene disrupted strain showed the greatest decrease (Figure 3F). These results suggest that peptide degradation with P1 neutral amino acid and hydrophilic Asn in the bacterium is predominantly mediated by DPP7.

The specific activities of recombinant DPPs were as follows: DPP4, 5,323 ± 1,544 pmol/min/µg (n = 5) and DPP11, 3,778 ± 473 pmol/min/µg (n = 5), and DPP5, 759 ± 342 pmol/min/µg (n = 4) and DPP7, 593 ± 173 pmol/min/µg (n = 8) (Figure S2), indicating higher specific activities for DPP4 and DPP11. These differences are likely due to the highly-restricted P1 residues of DPP4 and DPP11, in contrast to the broader specificities of DPP5 and DPP7. Accordingly, the contents of DPP4, DPP5, DPP7, and DPP11 in the wild-type P. gingivalis cells were calculated to be 0.9 ± 0.0, 2.9 ± 0.1, 9.5 ± 0.3, and 1.4 ± 0.1 ng/μl (A₆₀₀ = 10), respectively. This distribution is discussed in greater detail later.

2.3 Enhancement of DPP7 activity by subsite residues toward substrates with P1 neutral amino acids and Asn

DPP7 activity was found to be enhanced by hydrophobic P2 residue (Rouf et al. 2013a), thus the activity toward XNLD-MCA with the substitution of P2 His for hydrophobic amino acids was examined (Figure 4). DPP7 activities toward P1 Asn were significantly increased with P2 amino acids with a higher HI value, such as Leu (HI = 97), Phe (100), Met (74), and Tyr (63), as compared to P2 Glu (−31), Ala (41), Asp (−55), and His (8) (Figure 4B Left). Slight increases in these substrates due to hydrophobic P2 residues were also observed with DPP4 and DPP5 (Right). A plot showing the activities against HI of P2-position amino acids confirmed that the activities of DPP4 and DPP5, as well as DPP7 were positively corelated with the hydrophobicity of P2 amino acids, and that DPP7 exhibited higher activities than DPP4 and DPP5 for all of the examined substrates (Figure 4C). These results also reinforced findings indicating that DPP7 primarily targets neutral and hydrophilic P1 amino acids. Notably, the hydrolysis profiles in the bacterial cells for these eight P2 amino acids with XNLD-MCA were indistinguishable from that of recombinant DPP7 (compare Figure 4B–D).

Figure 4:

Preference for hydrophobic P2 residue in DPP activities (A) The chemical formula and cleavage sites of XNLD-MCA used for the two-step method are shown (B) XNLD-MCA was incubated with DPP7 (5 ng), DPP4 (40 ng), or DPP5 (40 ng) in the presence of DPP11 (5 μg) at 37 °C for 30 min (C) Linear regression analyses were performed using the log of the DPP activities and HI of P2-amino acid residues. The HI values for Leu, Phe, Met, Tyr, Glu, Ala, Asp, His were 97, 100, 74, 63, −31, 41, −55, and 8, respectively. Regression lines for DPP4, DPP5, and DPP7 are y = 0.0084x + 0.109 (R² = 0.721), y = 0.0136x + 0.415 (R² = 0.880), and y = 0.013x + 1.487 (R² = 0.379), respectively (D) XNLD-MCA was incubated with wild-type P. gingivalis in the absence or presence of DPP11 (5 μg). Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments.

To examine the synergistic effects of P2 and P1′ residues, nine tetrapeptidyl-MCA substrates with P1′ Leu with moderate enhancement of three DPPs were used, with Met found to be most preferable for DPP7 and Ser most preferable for DPP5 (see Figure 1). Those were combined with P1 Ala as a positive control or Asn as a refractory residue, P2 His, Tyr, or Leu, which vary in HI, and all with P2′ Asp (Figure 5A). DPP4 degraded HAMD- and HALD-MCA, as noted above, followed by HASD-MCA, while degradation of the other six substrates was negligible. These findings were understandable, because DPP4 is exclusively specific for P1 Pro and Ala. Furthermore, DPP5 exhibited the highest level of activity toward HASD-MCA, followed by HAMD-, HALD-, and YNSD-MCA, while its activity toward the others was negligible. DPP7 cleaved the substrates with P1 Ala, HAMD-, HASD-, and HALD-MCA, as noted above, and exclusively degraded substrates containing P1 Asn with both P2 and P1′ Leu (LNLD-MCA), and with P2 Tyr and P1′ Met (YNMD-MCA), followed by YNSD- and YNLD-MCA. As a result, the synergistic effect of P2 Tyr and P1′ Met was noted with DPP7, as compared with HNLD-, YNLD-, and YNMD-MCA while that of P2 Tyr and P1′ Ser was noted with DPP5, as compared with YNLD-, HNSD-, and YNSD-MCA though was rather limited.

Figure 5:

Synergistic effects of P2 and P1′ residues on DPP activities (A) Tetrapeptidyl-MCA substrates carrying P1 Ala or Asn with P2′ Asp were incubated with DPP4 (5–20 ng), DPP5 (5–40 ng) or DPP7 (5–20 ng) (B) Hydrolyzing activities of wild-type P. gingivalis (0.5–5 µl of A₆₀₀ = 10) in the presence of DPP11 (5 μg). Values (mean ± SD, n = 3) shown represent a typical example of three separate experiments (C) Tetrapeptidyl-MCA substrates were incubated with wild-type P. gingivalis or dpp-gene disrupted strains (0.5–3 µl of A₆₀₀ = 10). Results for HNLD- and HNSD-MCA are not shown, as their activities were quite low, while that of HNLD-MCA is presented in Figure 3F. Values represent a typical example of three separate experiments and are presented as percent ± SD (n = 3) of wild-type P. gingivalis, which was considered to be 100 %. *¹ p <0.05, *² p <0.005, *³ p <0.001 (as compared to wild-type P. gingivalis).

Hydrolyzing activities in wild-type P. gingivalis cells were comparable to those observed with recombinant DPPs (compare with findings shown in Figure 5A and B). Among the dpp-gene disrupted strains, dpp7 KO showed the most significant decreases in activity toward substrates containing P1 Asn (LNLD-, YNLD-, YNMD-MCA) (Figure 5C). In addition, a significant loss for HASD-MCA was noted for the dpp5 KO strain. Thus, peptides with P1-hydrophilic amino acid (Asn) were primarily processed by DPP7 in the bacterial cells, though cleavage of substrates with P1 Ala could be shared by DPP5 depending on the subsite residues.

Activity sharing between DPP7 and DPP5 was further compared using XNSD-MCA (X = Leu, Phe, or Tyr). For this comparison, the unique activity of DPP7 for LNLD- and FNLD-MCA was referred to as a positive control, and the quite low activity of both DPPs for HNSD-MCA as a negative control (Figure 6A). Both DPP5 and DPP7 showed significant hydrolysis for LNSD-, FNSD-, and YNSD-MCA as compared with that for HNSD-MCA. Nevertheless, the activities of DPP7 for these substrates were consistently superior to those of DPP5. In accord with these results, hydrolysis of substrates in P. gingivalis was most significantly reduced in the dpp7 gene-disrupted cells (Figure 6B Right). Taken together, it was concluded that peptides carrying penultimate P1 Asn and presumably neural amino acids were predominantly hydrolyzed by DPP7, though the efficiency was low.

Figure 6:

Degradation of tetrapeptidyl-MCA carrying P1′ Ser by DPP5 and DPP7, and in P. gingivalis cells (A) Tetrapeptidyl-MCA was hydrolyzed by DPP5 (20 ng) or DPP7 (5 ng) in the presence of DPP11 (5 µg) (B) Specific activities of wild-type P. gingivalis or dpp-gene disrupted strains (A₆₀₀ = 10) (5 µl for YNSD-MCA, 2 µl for others) in the presence of DPP11 (5 µg). Left. P. gingivalis wild-type cells. Right. Activities are presented as percent ± SD (n = 3), with those of P. gingivalis wild-type considered to be 100 %. Values shown represent a typical example of three separate experiments and are presented as percent ± SD (n = 3) of wild-type P. gingivalis (100 %). *¹ p <0.05, *² p <0.005, *³ p <0.001 (as compared to the wild-type P. gingivalis).

2.4 Examinations of remaining ambiguous amino acid residues to obtain a comprehensive 20 P1 ✕ 20 P2 amino acids matrix

Additional detailed examinations were conducted with specific amino acids to more comprehensively understand the 400 combinations of P1 and P2 dipeptide release.

5.1 Materials

All synthetic peptidase substrates used in the present study are shown in Table S1. t-Butyloxycarbonyl (Boc)-FSR-, Gly-Pro-, Phe-Met-, and Leu-Asp-MCA were purchased from Peptide Institute (Osaka, Japan), and Gly-Phe-MCA was obtained from Bachem (Bubendorf, Switzerland). His-Ala-, Tyr-Ala-, and all of the tetra- and penta-peptidyl-MCA substrates were not commercially available, thus were synthesized by Scrum (Tokyo, Japan) at a purity ≥ 90 %.

5.2 Bacterial strains and culture conditions

Construction of P. gingivalis ATCC 33277 derivatives NDP200 (Δdpp4), NDP300 (Δdpp5), NDP400 (Δdpp7), and NDP500 (Δdpp11) was performed as previously described (Ohara-Nemoto et al. 2014; Rouf et al. 2013b). P. gingivalis strains were grown anaerobically at 37 °C (80 % N₂, 10 % CO₂, 10 % H₂) in Anaerobic Bacterial Culture Medium broth (EIKEN Chemical, Tochigi, Japan) supplemented with 0.5 μg/ml of menadione. Appropriate antibiotics (ampicillin, erythromycin, tetracycline) were added for cultures of dpp-gene deleted strains. P. gingivalis cells were harvested in the early stationary phase and washed with ice-cold phosphate-buffered saline (PBS), pH 7.3, then suspended in PBS for adjustment to 10 at A₆₀₀.

5.3 Expression and purification of recombinant DPPs

Expression plasmids for P. gingivalis DPP4 (Rouf et al. 2013a), DPP5 (Ohara-Nemoto et al. 2014), DPP7 (22), DPP11 (Ohara-Nemoto et al. 2011), and PTP-A (Ito et al. 2006) tagged with a C-terminal hexa-histidine have been previously reported. Escherichia coli XL-1 Blue cells carrying an expression plasmid were cultured in Luria-Bertani broth supplemented with 75 μg/ml ampicillin at 37 °C overnight. After addition of two volumes of fresh media, recombinant proteins were induced with 0.2 mM isopropyl-β-d-thiogalactopyranoside at 30 °C for 4 h, then purified from the bacterial cell lysate using TALON affinity chromatography, as previously described (Ohara-Nemoto et al. 2011). A previous report noted that recombinant PTP-A was not adsorbed to a TALON metal affinity resin (Clontech Lab. Inc., CA, USA) for an unknown reason (29), thus PTP-A purification was performed by chromatography using DEAE-Toyopearl 650 M (TOSOH, Tokyo, Japan), with recombinant PTP-A purity shown by SDS-PAGE results to be approximately 90 %.

5.4 Peptidase activity determination

The reaction was started by addition of recombinant DPPs (0.5–100 ng) or a bacterial cell suspension (1–5 μl) in a reaction mixture (200 µl) composed of 50 mM Na-phosphate buffer (pH 7.5), 5 mM EDTA, and 20 µM peptidyl-MCA. All reactions were performed at 37 °C. For two-step measurements, target DPP (5–100 ng) or bacterial cells (0.5–1 µl of A₆₀₀ = 10) were added in the presence of an excess amount of DPP11 (5 µg) or PTP-A (15 µg) as the second enzyme, unless otherwise stated. After 30 min, fluorescence intensity was measured with excitation at 380 nm and emission at 460 nm, while details for prolonged reactions are presented in Figure 3. Decomposition of the substrate up to 10 % could be determined as a first-order reaction. Rgp, DPP4, DPP5, DPP7, and DPP11 activities in P. gingivalis cells (0.5–10 µl) were determined using Boc-FSR-, Gly-Pro-, Gly-Phe-, Phe-Met-, and Leu-Asp-MCA, respectively. DPP4, DPP5, and DPP7 activities for the various dipeptidyl-MCA substrates examined were determined using the conditions noted above, with an excess amount of the second peptidase utilized to adjust the measuring conditions in accordance with the other samples.

5.5 Protein concentration

Concentrations of the purified proteins were determined using Coomassie Brilliant Blue dye (Bio-Rad) with BSA as the standard.

5.6 Statistical analysis

For continuous variables, statistical significance was determined using Student’s t-test with Bonferonni correction. P values less than 0.05 were considered to indicate statistical significance.