The Mycobacterium tuberculosis prolyl dipeptidyl peptidase cleaves the N-terminal peptide of the immunoprotein CXCL-10

MtDPP forms a homodimer

The GenBank sequence CNF91057.1 was cloned into the pET28a + plasmid for bacterial heterologous expression. The expressed MtDPP was consistently purified to high yields of recombinant protein (approximately 30 mg per liter of cultured bacteria) that eluted as a monodispersed peak from an S200 size exclusion chromatography column (SEC), depicted as a dominant single band on SDS-PAGE gels (Figure S1).The protein’s elution volume correlates with a molecular weight (MW) of approximately 154 kDa, whereas its MW characterized upon SDS-PAGE analysis was 85 kDa (Figures S1 and S2), indicating that MtDPP forms a homodimer in solution in agreement with the quaternary arrangement found among DPPs (Aertgeerts et al. 2004; Nakajima et al. 2008; Rea et al. 2017; Roppongi et al. 2018).

MtDPP exhibits rapid post-proline DPP activity

We demonstrate that MtDPP possesses an unequivocal post-proline DPP activity (Figure 1). Neither single amino acid nor tripeptide-based substrates were proteolytically cleaved by MtDPP (Figure 1A), which indicates that the positioning of the substrates’ free N-terminus imposes stringent limits to enable catalysis. Moreover, dipeptide substrates with non-proline residues in the P₁ position yielded significantly lower (H-Lys-Ala-AMC; Figure 1A) or no catalytic activity (H-Gly-Ala-βNA; Figure 1A), which correlates with the capacity of other DPP orthologues that also catalyse the hydrolysis of substrates with alanine or serine occupying the P₁ position but at catalytic lower rates (De Meester et al. 2002; Lambeir et al. 2001; Leiting et al. 2003; Mortier et al. 2016). The stringent specificity of MtDPP for proline-based substrates is consistent with the replacement of a tyrosine residue with phenylalanine at position 566 (Tyr631 in hDPP4; Figure S3A), which was found to be crucial for substrate S₁-specificity determination in hDPP4 (Aertgeerts et al. 2004; Brandt 2002). Interestingly, although endopeptidase activity was not observed when the NCBZ-Gly-Pro-AMC substrate was used, MtDPP did exhibit marginal prolyl endopeptidase activity for Suc-Gly-Pro-AMC (3.3 ± 1.3%, normalised to h-Gly-Pro-AMC activity at 100%; Figure 1A), indicating that the active site of MtDPP can better accommodate ionized N-termini than hydrophobic ones, which aligns with its expected role as an N-terminal DPP.

Figure 1:

MtDPP exhibits N-terminal post-proline dipeptidyl peptidase activity. The enzymatic activity was measured in buffer consisting of 0.1 M Tris-HCl pH 7.8, 0.15 M NaCl, 10% glycerol (v/v) using surrogate substrates of the fluorophores βNA (excitation/emission wavelength: 330 nm/415 nm) and AMC (excitation/emission wavelength: 350 nm/450 nm), as well as the chromophore pNA (absorbance wavelength: 380 nm). Kinetic raw data was fitted using nonlinear regression to the Michaelis Menten (Y = V _max * X/(K _m  + X)) or substrate inhibition (Y = V _max * X/(K _m  + X * (1 + X/K _i ))) equations, respectively. (A) MtDPP activity toward various single amino acid, dipeptidic and tripeptidic-based substrates (white, βNA; grey, pNA; dotted, AMC). MtDPP activity toward each substrate was normalised to that of a respective h-Gly-Pro substrate that shared the same leaving group. (B) MtDPP kinetics toward h-Gly-Pro conjugated to different fluorophores or chromophore (○ βNA, ◇ AMC and ▲ pNA). The experiment was conducted at substrate concentrations of 0–2.5 mM, from which rate of catalysis was determined for respective concentrations and plotted as above. Error bars indicate the standard deviation based on three independent experiments performed in triplicates.

Characterization of P₁′ substrate position upon MtDPP hydrolysis was pursued using h-Gly-Pro substrates with different leaving groups (Figure S4), namely 7-amino-4-methylcoumarin (AMC), β-naphthylamine (βNA) and p-nitroanilide (pNA; Table 1; Figure 1B). MtDPP showed a mild difference between the observed K _M constants, whereas the turnover k _cat values depicted significant disparities (p-value < 0.001; Table 1). In contrast trend to hDPP4 kinetics (Bjelke et al. 2004; Leiting et al. 2003; Li et al. 2022; Roppongi et al. 2018), MtDPP exhibits lower turnover rate k _cat constant for the AMC surrogate in comparison to the pNA conjugated substrate (Table 1). Moreover, the three different h-Gly-Pro derivatives display substrate inhibition patterns that would suggest the relevance of the P₁′ leaving group during catalysis (Table 1). Despite that, a comparison between the magnitudes of the specificity constants k _cat/K _M for the three substrates show that they are hydrolysed at nearly diffusion rate limits (Table 1), 10⁵–10⁷ M⁻¹ s⁻¹ (Lambeir et al. 2001), indicating very rapid post-proline activity.

Thermal stability of MtDPP and influence of pH and ionic strength on its enzymatic activity

Like hDPP4 (Li et al. 2022), MtDPP displays a thermal stability characterised with a melting temperature (T _m ) of 70.4 ± 0.3 °C measured in thermal shift assays (TSA) (Figure 2A). This stability was further confirmed by MtDPP showing increased catalytic rate as temperatures rise up to 60 °C, after which activity declines (Figure 2B). Characterization of the individual kinetic parameters in a temperature range between 20 and 70 °C showed that MtDPP has its maximum k _cat /K _M at 50 °C. In this temperature range, the changes in the observed K _M values are relatively mild, at around 2.3-fold change between 30 and 50 °C; whereas the turnover rate constants increased by 2-fold in the same range of temperatures (Table 2). Unlike the kinetics assay performed using a continuous assay (Table 1), the temperature dependent kinetics assay showed no substrate inhibition (Table 2), which indicates the experimental limitations of the assay.

Figure 2:

Biochemical characterization of MtDPP. (A) First derivative curve plot of wild type MtDPP in 0.1 M Tris-HCl pH 7.8, 0.15 M NaCl and 10% glycerol (v/v). (B) Arrhenius plot depicting MtDPP temperature dependent kinetics. Red data point indicates subversion to plot linearity (70 °C). (C) MtDPP catalytic rate (left y-axis; ● line) and thermal shift (∆T _m ; right y-axis; ○ dotted) in different pH (3.0–13.0) conditions. (D) MtDPP catalytic velocity (left y-axis; grey bars) and ∆T _m (right y-axis; white bars) in varying ionic strength environments. Control was performed in 0.1 M Tris-HCl pH 7.8, 0.15 mM NaCl, 10% glycerol (v/v). All error bars depict standard deviation derived from three experiments in triplicates. TSA melt curve plots can be found in Figure S6A and B.

The enzymatic activity measured under constant ionic strength with varying pH showed that MtDPP is catalytically active within a wide range of the pH scale (Figure 2C). It displays a well-defined two peaked activity curve, with one peak centred on the near neutral pH range (pH 7.8), similar to what has been described for other MtDPP orthologues (Kumagai et al. 2000; Leiting et al. 2003; Nakajima et al. 2008); and the second peak centred towards the more alkaline range of the pH scale (pH 10.6). Interestingly, the activity of MtDPP at different pH values does not positively correlate with its thermal stability, as the most stable form of MtDPP was characterized to be in a more acidic pH range (Figure 2C), achieving maximal stability at a pH around 6.5. Previously, it has been shown that DPP activity is enhanced when the ionic strength increases up to values of 0.5 M NaCl (Polgar 1995). However, for MtDPP the increase in ionic strength resulted in an inverse correlation with the catalytic performance of the enzyme (Figure 2D, grey bars), whereas the thermal stability of MtDPP showed a direct correlation with the incremental increase in ionic strength (Figure 2D, white bars). Our characterization shows that MtDPP actually exhibits enhanced enzymatic activity in conditions that make the protein structurally more unstable (Figure 2C and D), which might indicate the relevance of the interdomain interactions during catalysis (Figure 3A). Considering its homology to human DPPs, the existence of a “Velcro” motif enabling a filter mechanism that increases selectivity for smaller peptides and the possible different conformations of the flap propeller loop next to the active site may both be structurally influencing catalysis (Fulop et al. 1998, 2000; Li et al. 2022; Polgar 1992). Therefore, conditions where these structural features on MtDPP are affected could result in more open protein conformations that enable easier substrate access to the active site and consequently an increase in catalytic activity.

Figure 3:

MtDPP structural model. (A) SWISS-MODEL predicted 3D ribbon model of MtDPP depicting the β-propeller (grey) and α/β-hydrolase (aqua) domains of the protein, superimposed on the hDPP4 structure (PDB: 2G5P; yellow) The model is available in ModelArchive at https://modelarchive.org/doi/10.5452/ma-ucdv8. Model statistics are shown in Figure S3B. (B) 13 superimposed catalytic site residues of MtDPP (blue) and hDPP4 (orange), indicating a conserved catalytic cleft. (C) 2D diagram representing interactions (light green circle, van der Waals; light green line, carbon hydrogen bond; green line/circle, conventional hydrogen bond; orange circle, attractive charge; orange line, salt bridge; pink line/circle, π-alkyl bond; red line, unfavourable donor–donor) between predicted MtDPP active site residues and a docked Gly-Pro-Ala substrate (PubChem CID: 7276371). (D) Thermal shift (∆T _m ) of MtDPP and its variants. Error bars indicate the standard deviation derived from three experiments in triplicates. TSA melt curve plots can be found in Figure S6C.

Mapping the active site of MtDPP using site-directed mutagenesis

Based on sequence homologies (Aertgeerts et al. 2004; David et al. 1993; Nakajima et al. 2008; Pei et al. 2006; Rea et al. 2017) and a predicted MtDPP structure (Figure 3A; the model is available in ModelArchive https://modelarchive.org/doi/10.5452/ma-ucdv8), thirteen MtDPP variants were generated to map the active site of the enzyme (Figure 3B). Most MtDPP variants showed protein expression yields comparable to the wild type (WT; Figure S5A). Interestingly however, the His672Ala variant, although visible on SDS PAGE analyses, could never be detected in Western Blot experiments; and variants Trp594Ala and Glu183Ala failed to yield soluble proteins that could be used for further characterization. The failure of these active site related variants to yield soluble protein is testament to the very tight relation between protein function and structure (Uversky 2021). Many of the MtDPP variants generated were inactive (Table 1). Amino acid replacements involving the catalytic triad (Ser565, Asp640 and His672) and the nucleophilic elbow (Gly563 and Gly567) resulted in a complete loss of catalytic activity. Moreover, replacement of residues Tyr597 and Asn642, that are expected to coordinate the highly conserved double-glutamic acid (Glu183 and Glu184) motif responsible for the coordination of the substrate’s N terminus (Table 1; Figures 3B and C; Aertgeerts et al. 2004) also resulted in inactive protein variants. The same inactivity was observed with the replacement of Tyr601 (Table 1; Figures 3B and C), a substrate S₁-specificity determining residue (David et al. 1993). Replacements of residues Val643, associated with the formation of the hydrophobic active site S₁ pocket, as well as Asp598 and Glu184, involved in the substrate’s coordination of the N-terminus in the hDPP4 orthologue (Abbott et al. 1999; Aertgeerts et al. 2004), were characterized with smaller k _cat and k _cat/K _M constants compared to WT, but with Asp598Ala and Glu184Ala displaying larger K _M constants (Table 1) providing indication of their relevance for the binding of substrates.

The characterization of the thermal stability of seven of the MtDPP tested variants exhibited a mild reduction as indicated by thermal shifts (∆T _m ) of at least 1 °C (Figure 3D), indicating that these residues contribute to enhancing the structural stability of the enzyme. Remarkably, variants Tyr597Ala and Val643Ser showed more pronounced reductions in thermostability, with a ∆T _m of −5.76 ± 0.21 °C (p-value < 0.001) and −6.90 ± 0.08 °C (p-value < 0.001), respectively (Figure 3D). The variant Val643Ser in particular had an unusually flat melt curve (Figure S6C), indicating that the interactions between the dye and hydrophobic regions are similarly feasible within the tested temperature range (Huynh and Partch 2015), therefore suggesting that the Val643Ser variant significantly influences the folding state of the protein. Interestingly, the catalytic triad variants Ser565Ala and His672Ala were characterized with TSA as unaffected (with no significant change if compared to control) and slightly more stable with a higher T _m value (p-value < 0.001), respectively (Figure 3D), suggesting that these residues promote a certain degree of instability in the WT enzyme.

The different protein stabilities characterized for the different protein variants support the existence of a tight relationship between the protein structure, that ensures the protein’s solubility and flexibility, and its function, that enable the catalytic process. Thus, our characterization shows that the activity of MtDPP correlates inversely with its thermal stability, and that the catalytically critical residues Ser565 and His672 counterintuitively may be identified as actual contributors to the protein’s instability.

MtDPP activity is lowered by hDPP4 inhibitors

We characterized the thermal stability of MtDPP in the presence of known hDPP4 inhibitors belonging to classes 1 (LAF237), 2 (BI-1356 and SYR-322) and 3 (MK0431) (Nabeno et al. 2013; Röhrborn et al. 2015) (Figure S4). The thermal stability of MtDPP increased considerably in the presence of the molecules MK0431 and LAF237, with a ∆T _m of approximately 4 °C (p-value < 0.001; Figure 4A). Interaction of MtDPP with BI-1356 also resulted in a milder increase in thermal stability (p-value < 0.001; Figure 4A), whereas a mild thermal destabilisation of MtDPP was observed in the presence of SYR-322 (p-value < 0.05; Figure 4A). Additional characterization of the inhibition of MtDPP activity with the same molecules showed that MK0431 and LAF237 displayed the strongest inhibition characterized with IC₅₀ constants found in the low micromolar range (Figure 4B). Whereas BI-1356 and SYR-322 were characterized with weak inhibition of MtDPP with IC₅₀ constants in the millimolar range. These results indicate that MtDPP may be more specifically inhibited by class 1 and 3 hDPP4 inhibitors, with the former interacting with the S₁ and S₂ subsites, forming a covalent bond with the catalytic serine residue, and the latter stabilising interactions with the S₂-extensive subsite for the human homologue (Nabeno et al. 2013; Röhrborn et al. 2015). Nevertheless, the weaker bindings of these inhibitors to MtDPP indicate relevant structural differences between MtDPP and hDPP4 exist, as similarly found with other bacterial DPPs (Nabeno et al. 2013; Rea et al. 2017; Roppongi et al. 2018).

Figure 4:

hDPP4 inhibitors bind to MtDPP and inhibit its activity. (A) ∆T _m of MtDPP after treatment with hDPP4 inhibitors. (B) MtDPP IC₅₀ inhibition assays with hDPP4 inhibitors depicting inhibition by compounds SYR-322 (■), BI-1356 (▽), MK0431 (◆) and LAF237 (○). In both, error bars depict standard deviation derived from three experiments in triplicates. Inhibition raw data was fitted to Y = Bottom + (Top − Bottom)/(1 + (IC50/X)ˆHillSlope) inhibition model. TSA melt curve plots can be found in Figure S6D.

MtDPP is able to truncate the chemokine CXCL-10 in vitro

Previously, hDPP4 has been shown to catalyse the truncation of a plethora of immune signalling molecules, including chemokines such as CXCL-10 and C-C motif chemokine ligand 22 (CCL-22), which are important for lymphocyte chemotaxis to sites of infection (Decalf et al. 2016; De Meester et al. 2002; Domingo-Gonzalez et al. 2016; Lambeir et al. 2001; Ou et al. 2013; Proost et al. 1999), as well as interleukin-23 (IL-23) which stimulates downstream signalling that leads to the production of IL-23 dependent cytokines such as interleukin-17 and 22 (Domingo-Gonzalez et al. 2016; Ou et al. 2013). Truncations of either of these molecules result in reduced lymphocyte chemotaxis (Mortier et al. 2016; Proost et al. 2017) and have been associated with TB infections (Blauenfeldt et al. 2018; Domingo-Gonzalez et al. 2016), therefore we hypothesized they could also be potential targets for processing by MtDPP.

To test this hypothesis, short synthetic peptides of CXCL-10 (VPLSRTVRCT-NH₂), CCL-22 (GPYGANMEDS-NH₂), IL-23 (RAVPGGSSPAKER-NH₂) and tumour necrosis factor alpha (TNFα, VRSSSRTPSD-NH₂) were incubated with MtDPP, followed by analysis using MALDI-TOF mass spectrometry (Figures 5 and S7). The results showed that MtDPP is able to cleave the N-terminus of CXCL-10 (Figure 5A), whereas the inactive MtDPP S565A variant is not (Figure 5C). As expected, the control peptide TNFα, was not processed by MtDPP (Figure 5B). Interestingly, MtDPP was unable to cleave the N-terminal dipeptide from neither the CCL-22 nor the IL-23 substrate (Figures S7A and B), strongly indicating the relevance of the P₁′ leaving group in substrate processing, and thus clearly differentiating from reported hDPP4 activity. Previous studies have correlated increased levels of truncated CXCL-10 to infection severity and effectiveness of TB treatment (Blauenfeldt et al. 2018; Juffermans et al. 1999; Wang et al. 2012) with the proposal of hDPP4 having a regulatory role in active TB on CXCL-10/C-X-C chemokine receptor 3 driven chemotaxis of CD4⁺ T cells (Blauenfeldt et al. 2018). Although yet speculative, it seems plausible that MtDPP could potentially modulate the immune response by a rapid truncation of CXCL-10 to hinder migration of CD4⁺ T cells to the site of infection.

Figure 5:

MALDI-TOF spectra analysing the processing of short chemokine and cytokine peptides by MtDPP. CXCL-10 (A and C) and TNFα (negative control, B and D) peptides were treated with wild type (A and B) or variant S565A (C and D) MtDPP. Peaks are labelled with their sequence and MW corresponding to incubated peptides or their resulting fragments. Peaks display the average relative intensities of three independent experiments.

Protein expression and purification

The coding sequence of MtDPP was cloned into plasmid pET28a(+) in frame with an N-terminal 6 × His tag (GenScript™). BL21 (DE3) Escherichia coli (TIANGEN^®) transformed with the construct was grown in 450 mL LB supplemented with 50 μg/mL kanamycin at 37 °C, 200 rpm. At an OD₆₀₀ of 0.4, incubation temperature was reduced to 16 °C for 1 h, and protein expression subsequently induced with the addition of 0.2 mM final concentration of IPTG. After 24 h incubation, cells were harvested and re-suspended in purification buffer (0.1 M Tris-HCl pH 7.8, 0.15 M NaCl, 10% glycerol [v/v]). Cells were sonicated (Qsonica Q700 Sonicator with microtip probe) at 40% amplitude for 20 min with 5 s on/off intervals, followed by centrifugation at 10,000 g, 4 °C for 10 min. The supernatant was subsequently incubated at 55 °C for 10 min. Debris and protein aggregates were finally removed via centrifugation at 40,000 g, 4 °C for 30 min. The obtained supernatant was loaded onto Profinity™ immobilized metal affinity chromatography (Bio-Rad) as an initial step to purify the non-native MtDPP. MtDPP was eluted with purification buffer supplemented with 0.3 M imidazole. SEC using HiLoad™ 16/600 Superdex™ 200 pg column (Cytiva) pre-equilibrated with the purification buffer using the ÄKTAprime Plus system (Cytiva) was performed to complete the isolation of MtDPP (Figures S1). The column was calibrated using the High MW gel filtration calibration kit (GE Healthcare) and the elution volumes and molecular weights were plotted using a linear regression via GraphPad Prism v8.0.1 to resolve MtDPP MW (Figures S2A). Purified protein was concentrated with Amicon^® Ultra-15 centrifugal filter units with a 10 kDa MW cut-off. Protein concentration was determined both via NanoDrop A280 and bicinchoninic acid assay (Beyotime^®).

Protein overexpression and purification were analysed via 10% SDS-PAGE gels and imaged via ChemiDoc™ MP Imaging System (Bio-Rad; Figures S1 and S5). Immunoblotting was performed to confirm purification results with primary anti-6xHis polyclonal mouse antibody (Beyotime^®) and secondary IRDye^® 800CW goat-anti-mouse antibody (LI-COR^®). Immunoblot images were acquired via LI-COR^® Odyssey Classic Imaging system. The relative migration protein marker (PageRuler™ Prestained Protein Ladder) was measured using ImageJ (Schneider et al. 2012). The Log of the molecular weight and the relative migration factor of the protein marker were plotted using linear regression to generate a standard curve with GraphPad Prism v8.0.1 to resolve MtDPP MW (Figure S2B).

Enzyme assays: kinetics, pH, ionic strength, and temperature dependencies

Catalytic activity was determined spectrophotometrically by measuring the emission and absorbance wavelengths of released fluorophore and chromophore conjugates of peptidic substrates, respectively. Fluorophores βNA (excitation 330 nm, emission 415 nm) and AMC (excitation 350 nm, emission 450 nm), as well as chromophore pNA (absorbance 380 nm) were used in this study (Figures S4). For all enzymatic assays, fluorescence and absorbance were measured via Varioskan^® LUX microplate reader with the SkanIt^® software (Thermo Scientific™). All assays were performed with 6 μg mL⁻¹ enzyme on 96-well plates (BIOFIL^®) in purification buffer supplemented with 0.3 mM substrate at 30 °C, shaking with low power at 600 rpm between each read (unless stated otherwise), and conducted in three experiments in triplicates. Raw data were processed with MS Excel and fitted onto plots via GraphPad Prism v8.0.1. Kinetic raw data was fitted using nonlinear regression to the Michaelis Menten (Y = Vmax * X/(K _m  + X)) or substrate inhibition (Y = V _max * X/(K _m  + X * (1 + X/K _i ))) equations, respectively.

Enzyme kinetics for both WT and variants of MtDPP were performed with substrate concentrations ranging from 0 to 2.5 mM pH dependency was performed with buffers from pH 3.0 to 13.0. The buffers were composed of 0.05 M acetic acid, 0.05 M 2-(N-morpholino)-ethanesulfonic acid and 0.1 M Tris (Ellis and Morrison 1982). Ionic strength dependency assays were performed in the purification buffer (without glycerol) at increasing concentrations of NaCl within the range 0.0–1.5 M.

Temperature dependent kinetics of MtDPP were performed at temperatures from 20 to 70 °C with 10 °C increments. The assay was performed using the Veriti™ 96-Well Thermal Cycler (Applied Biosciences™). Buffer-substrate mixtures were aliquoted into MicroAmp^® Fast Optical 96-well plates (Applied Biosciences™) and pre-incubated at respective temperatures for 5 min, followed by the addition of 75 ng of the enzyme. Catalysis was monitored for less than 5 min. Catalytic rates were determined by dividing the fluorescence detected by the duration at which the enzyme was allowed to catalyse the reaction.

Thermal shift assay

TSAs were performed with 0.4 mg mL⁻¹ MtDPP (or MtDPP variants) in respective test buffers with the inclusion of other molecules being tested, and a final concentration of 5 × SYPRO™ orange dye (Invitrogen™). Fluorescence measurements were attained by running the assay on the QuantStudio 5 real-time PCR system (Applied Biosystems™) at increasing temperatures from 25 to 95 °C at 0.1 °C s⁻¹ increments. Raw data analysis was performed via the Protein Thermal Shift software v1.3 (Applied Biosystems™). Temperature of melting values were determined using the Boltzmann fit value (Protein Thermal Shift™ Studies user guide – Applied Biosystems™); data were further processed with MS Excel and plotted onto bar charts or curves via GraphPad Prism v8.0.1. Experiments were conducted in three experiments in triplicates.

3D structure prediction and substrate docking

Sequence alignments were performed by CLUSTAL-W (https://www.genome.jp/tools-bin/clustalw) (Thompson et al. 1994) and visualised via EsPript 3.0 (https://espript.ibcp.fr) (Robert and Gouet 2014). Predicted 3D structure of MtDPP (the model is available in ModelArchive https://modelarchive.org/doi/10.5452/ma-ucdv8) was constructed based on known hsDPP4 structure (PDB: 2G5P) via SWISS-MODEL (Waterhouse et al. 2018) – analysis and visualisation were performed using Chimera 1.14 (Pettersen et al. 2004). Ligand (Gly-Pro-Ala; PubChem CID: 7276371) docking with predicted MtDPP structure was performed with Rosetta ligand docking (DeLuca et al. 2015) and mapped into a 2D diagram via BIOVIA Discovery Studio.

Site-directed mutagenesis

The NEBaseChanger v1.3.2 tool (https://nebasechanger.neb.com/) was used to design primers to generate MtDPP proteins variants. Mutant constructs were created and amplified via PCR with Q5^® High-Fidelity DNA Polymerase (NEB), and the respective primers provided by NEBaseChanger (Synbio Technologies; Table S1). Amplified constructs were treated with DpnI and polynucleotide kinase (NEB) for 10 min at 37 °C, proceeded by ligation via T4 DNA ligase (NEB) for 6 min at RT. Respective constructs were transformed into DH5α E. coli (TIANGEN^®). Site-directed mutagenesis was confirmed via plasmid sequencing (Synbio Technologies). Successfully mutated plasmids were transformed into BL21 (DE3) E. coli (TIANGEN^®), followed by purification and analyses as previously described.

hDPP4 inhibitor inhibition assay

hDPP4 inhibitors SYR-322 (MedChemExpress) (Feng et al. 2007), BI-1356 (Macklin) (Eckhardt et al. 2007), MK0431 (Aladdin) (Kim et al. 2005) and LAF237 (Aladdin) (Singh et al. 2008) (Figures S4) were added separately into enzymatic assays at concentrations ranging from 1 × 10⁻⁶ to 1 mM and incubated with purified MtDPP in aforementioned enzymatic assay conditions at RT for 15 min prior to the addition of an h-Gly-Pro-βNA substrate. Catalytic activity was determined as previously described, and normalised to a control assay without the addition of the inhibitors. Experiments were conducted in three experiments in triplicates. Raw data were processed with MS Excel and fitted onto plots via GraphPad Prism v8.0.1 using the Y = Bottom + (Top − Bottom)/(1 + (IC50/X)ˆHillSlope) inhibition model.

MALDI-TOF mass spectrometry

10–13 amino acid sequences of peptides above 1000 Da in MW were chemically synthesized (Chinese Peptides) at purities above 95%. Peptide digests were prepared as per the biochemical characterization assays, and incubated at 30 °C for 1 h (inactivated variant S565A was used as control). The reaction was terminated and diluted with the addition of TA30 (3:7, acetronitrile: 0.1% trifluoroacetic acid), and further mixed with MALDI-TOF peptide analysis matrix α-cyano-4-hydroxycinnamic acid (10 mg/mL in TA30) in a 1:1 ratio. Approximately 2 µL of the digest-matrix mixture was aliquoted onto an MTP 384 ground steel plate and left to air dry. The samples were processed via autoflex™ speed MALDI-TOF (Bruker Daltonics) with an MW filter of 700–3500 Da flexControl, flexAnaysis software (Bruker Daltonics) and mMass v5.5.0 (Strohalm et al. 2008) were used to acquire and analyse obtained spectra.