Structure, function, and inhibition of a genomic/clinical variant of Porphyromonas gingivalis peptidylarginine deiminase

Abstract

Citrullination is an essential post‐translational modification in which the guanidinium group of protein and peptide arginines is deiminated by peptidylarginine deiminases (PADs). When deregulated, excessive citrullination leads to inflammation as in severe periodontal disease (PD) and rheumatoid arthritis (RA). Porphyromonas gingivalis is the major periodontopathogenic causative agent of PD and also an etiological agent of RA. It secretes a PAD, termed Porphyromonas PAD (PPAD), which is a virulence factor that causes aberrant citrullination. Analysis of P. gingivalis genomes of laboratory strains and clinical isolates unveiled a PPAD variant (PPAD‐T2), which showed three amino‐acid substitutions directly preceding catalytic Residue H²³⁶ (G²³¹N/E²³²T/N²³⁵D) when compared with PPAD from the reference strain (PPAD‐T1). Mutation of these positions in the reference strain resulted in twofold higher cell‐associated citrullinating activity. Similar to PPAD‐T1, recombinant PPAD‐T2 citrullinated arginines at the C‐termini of general peptidic substrates but not within peptides. Catalytically, PPAD‐T2 showed weaker substrate binding but higher turnover rates than PPAD‐T1. In contrast, no differences were found in thermal stability. The 1.6 Å‐resolution X‐ray crystal structure of PPAD‐T2 in complex with the general human PAD inhibitor, Cl‐amidine, revealed that the inhibitor moiety is tightly bound and that mutations localize to a loop engaged in substrate/inhibitor binding. In particular, mutation G²³¹N caused a slight structural rearrangement, which probably originated the higher substrate turnover observed. The present data compare two natural PPAD variants and will set the pace for the design of specific inhibitors against P. gingivalis‐caused PD.

Short abstract

PDB Code(s): 6I0X

Introduction

Severe periodontal disease (PD) affects an estimated 750 million people worldwide.1 It causes tissue destruction in the gums and tooth loss, and has a chronic inflammatory background.2 It is an infectious disease caused by anaerobic Gram‐negative bacteria in the dysbiotic oral cavity, among which Porphyromonas gingivalis is a “keystone pathogen.”3, 4 Rheumatoid arthritis (RA) is a systemic autoimmune disease characterized by synovial inflammation and destruction of joint bone and cartilage.2 It affects about 1% of the general population by altering the immune response through auto‐antibodies that react with citrullinated proteins (ACPAs). Citrullination is a post‐translational modification in which the guanidinium group of protein and peptide arginines is deiminated by peptidylarginine deiminases (PADs), and four cysteine‐dependent PADs (PAD1–PAD4) carry out this physiologically indispensable function in humans.5, 6 However, deregulated PADs cause an excess of citrullinated proteins, which are massively bound by ACPAs in the joints. This leads to the release of pro‐inflammatory cytokines and the tissue breakdown characteristic of RA.2, 4

In addition to inflammation, RA and PD also show strong epidemiological, serological, and clinical associations.4 Indeed, the PD stage directly correlates with unusually high levels of ACPAs, so it may contribute to the inflammatory reaction of RA; vice versa, RA patients have twofold higher prevalence of PD.7 Finally, pathological RA and PD sites share augmented PAD2 and PAD4 levels, which cause enhanced protein citrullination.8 To make things worse, P. gingivalis secretes a unique cysteine‐dependent PAD, dubbed PPAD,9, 10, 11 as a virulence factor. The structure of PPAD is only remotely related to human PADs and, while the latter are endodeiminases, PPAD is a C‐terminal exodeiminase.12 Thus, during tissue destruction in the infected gums, PPAD originates new epitopes for ACPA recognition, which results in enhanced inflammation. This is fueled by gingipains RgpA and RgpB, which are cysteine peptidases secreted by P. gingivalis that generate peptides with C‐terminal arginines, i.e. new substrates for PPAD. Overall, this finding caused PPAD to be considered an etiological factor of RA10, 13 and a promising target against both PD and RA.10, 12, 14

Current approaches to drug design are essentially based on experimental crystal structure analysis of unbound and substrate/inhibitor complexes of disease‐related targets.15 Along this line, the recent structure of PPAD from P. gingivalis reference strain ATCC 33277, hereafter PPAD‐T1, revealed the general architecture and modus operandi of this unique bacterial PAD.12, 14 Moreover, recent studies reported that PPAD is present as point‐mutant variants differing from PPAD‐T1.16 We focused on triple mutant G²³¹N/E²³²T/N²³⁵D, hereafter PPAD‐T2, and analyzed its function in comparison with PPAD‐T1. We further tested the effect of Cl‐amidine (N ¹‐benzoyl‐N ⁵‐[2‐chloro‐1‐iminoethyl]‐l‐ornithinamide), a first‐generation inhibitor of human PADs developed by Thompson and colleagues,17 on PPAD‐T1 and PPAD‐T2 activity in comparison with human PADs. We further gained insight through the high‐resolution crystal structure of PPAD‐T2 with Cl‐amidine, which revealed the structural determinants of PPAD inhibition.

Results and Discussion

PPAD variants show disparate activity

Analysis of 31 P. gingivalis genomes and clinical isolates revealed that PPAD is present as variants encompassing between one and eight point mutations with respect to reference PPAD‐T1.16 We selected variant PPAD‐T2, which contains three mutations (G²³¹N, E²³²T, and N²³⁵D) that are close to catalytic residue H²³⁶ and are found in eight genomes and isolates. In contrast, other frequently observed substitutions (S¹⁹¹F, S²⁰³P, N²⁹¹D, and V³³⁵I) are located on the protein surface, away from the active site. Accordingly, analysis of these mutations in silico showed no potential effect on the enzymatic properties of PPAD (data not shown). To determine the outreach of the PPAD‐T2 mutations, we replaced these residues in reference strain ATCC 33277 and assayed citrullination of substrate Nα‐acetylarginine by suspensions of washed bacterial cells adjusted to the same OD₆₀₀ (Fig. 1). We found twofold higher activity for PPAD‐T2 than for the reference, which may be functionally relevant. Of note, there was only neglectable deiminating activity directly released into the growth medium. For further enzymatic and structural characterization, we produced PPAD‐T1 and PPAD‐T2 with C‐terminal hexahistidine‐tags in P. gingivalis strain ATCC 33277 as previously described for P. gingivalis proteins.12, 18

Figure 1.

Citrullinating activity in cell cultures. Activity of P. gingivalis strains containing variants PPAD‐T1 (ATCC 33277), PPAD‐T2 (ATCC T2; strain ATCC 33277 after introducing mutations G²³¹N, E²³²T, and N²³⁵D), and a control, in which the antibiotic cassette used in ATCC T2 was introduced in wild‐type ATCC 33277 (ATCC T1 Ctrl). The results are shown as relative activity determined in triplicates using three independent cultures of each strain and displayed as mean ± SD. The statistical difference between strains was analyzed by the analysis of variance; ns, not significant; *** P < 0.001.

Activity and thermal stability of PPAD‐T2 versus PPAD‐T1 in vitro

PPAD‐T1 was described to efficiently citrullinate arginines within dipeptide benzoylglycylarginine and peptides derived from bradykinin.9 Further substrates were epidermal growth factor and anaphylatoxin C5a, as well as peptides derived from α‐enolase, fibrinogen, and bacterial cell‐envelope proteins resulting from cleavage by gingipains.14, 19 To compare the citrullinating activities of recombinant PPAD‐T1 and PPAD‐T2, we assayed bradykinin‐derived peptides of sequence P‐P‐G‐F‐S‐P‐F‐R (exopeptide) and G‐F‐S‐P‐F‐R‐S‐S (endopeptide). Both PPAD variants citrullinated the exopeptide [Fig. 2(a)] but not the endopeptide [Fig. 2(b)]. This is consistent with previous results pinpointing PPAD as an exodeiminase but not an endodeiminase.12

Figure 2.

Enzymatic activity and stability. (a) Exo‐ and (b) endodeiminase activity assays of PPAD‐T1 and PPAD‐T2 against peptides of sequence P–P‐G‐F‐S‐P‐F‐R and G‐F‐S‐P‐F‐R‐S‐S, respectively. Peptides are shown before (HPLC chromatograms with solid lines) and after reaction with PPAD‐T1 (dashed lines) and PPAD‐T2 (dotted lines). Exopeptides were citrullinated, as shown by a shift in the retention time in (a), but endopeptides were not (in (b)). (c)–(e) Kinetic analysis of the citrullination reaction by PPAD‐T1 and PPAD‐T2 of peptides R‐P‐P‐G‐F‐S‐P‐F‐R (c) and I‐H‐A‐R‐E‐I‐F‐D‐S‐R (d). The reaction rates were determined by HPLC analysis, each point represents the mean value ± SD resulting from two measurements. Panel (e) depicts the values derived for V _max, K _M, k _cat, and k _cat/K _M. (f) Temperature of midtransition (T _m) of PPAD‐T1, PPAD‐T,2 and mutants G²³¹N, E²³²T, and N²³⁵D by differential scanning fluorimetry. The mean value ± SD resulting from duplicate measurements is shown. The statistical significance was determined by the analysis of variance; ns, not significant; **** P < 0.0001.

We next determined the kinetic parameters of citrullination of bradykinin peptide R‐P‐P‐G‐F‐S‐P‐F‐R and α‐enolase peptide I‐H‐A‐R‐E‐I‐F‐D‐S‐R by both variants [Fig. 2(c–e)]. We found that PPAD‐T1 bound these substrates six times stronger than PPAD‐T2. Moreover, PPAD‐T2 showed a three‐fold higher k _cat, that is turnover of substrates by this variant was faster at high substrate concentrations. Interestingly, the total catalytic efficiency (k _cat/K _M) of PPAD‐T1 was more than double that of PPAD‐T2. Accordingly, given that bacterial cell suspensions with PPAD‐T2 exhibited higher activity (see above), we conclude that efficiency of protein citrullination by both variants in vivo probably depends on physiological substrate concentrations. Finally, there was only a slight difference (~1.4‐fold) between the citrullination efficiencies of both exopeptides tested by either PPAD variant, so that no significant selectivity was observed for residues upstream of the C‐terminal arginine.

We also analyzed the thermal stability of both PPAD variants by differential scanning fluorimetry, which is directly based on changes in the ultraviolet fluorescence of tryptophan in protein samples upon heating. Overall, we found no significant difference in the midtransition temperatures [T _m; Fig. 2(f)]. However, when the individual contributions of the three mutations were analyzed, G²³¹N produced a significant reduction in T _m, which was compensated by a stabilizing effect of mutation N²³⁵D. In contrast, mutant E²³²T had no effect on stability.

General architecture of PPAD‐T2

We next obtained the three‐dimensional structure of PPAD‐T2 in complex with the inhibitor Cl‐amidine at high resolution [1.6 Å; Fig. 3(a,b,c,d)]. Overall, the structure showed great similarity with PPAD‐T1 in complex with a substrate or a substrate mimic.12 The structure of the protein resembles a tooth, including a crown with its cusp, a root and a connecting neck [see Fig. 3(a,b)]. The tooth crown contains the N‐terminal catalytic domain (CD; residues A⁴⁴‐K³⁵⁹), which includes an N‐terminal extension followed by a β‐propeller of five blades (I–V), each consisting of a three‐stranded twisted β‐sheet with an inner, a middle, and an outer strand, plus one helix. The tooth root comprises a C‐terminal immunoglobulin‐superfamily domain (IgSF; Residues G³⁶⁰‐E⁴⁶⁵) and is an irregular 4 + 5‐stranded β‐sandwich with a mixed front sheet and an antiparallel back sheet.

Figure 3.

Structure of PPAD‐T2 in complex with Cl‐amidine. (a) Ribbon‐plot of PPAD‐T2 in lateral view. The CD consists of an N‐terminal extension (in yellow) followed by a five‐bladed β‐propeller, with each blade (I–V) in one color (blue, plum, orange, red, aquamarine, respectively). The C‐terminal IgSF domain (in gray) is the tooth root. The Michaelis loop (magenta ribbon) at the tooth cusp is further depicted and labeled. (b) Close‐up top view in wall‐eye stereo of the tooth cusp with the active site. Selected residues are depicted and labeled, including the catalytic triad (C³⁵¹, H²³⁶, and N²⁹⁷). The catalytic cysteine is covalently bound to the Cl‐amidine moiety, which is presented with sandy‐brown carbons. The Michaelis loop is depicted as a magenta ribbon. (c) Detail of C³⁵¹ and the bound inhibitor superposed with the initial omit Fourier (2mF _obs–DF _calc)‐type map contoured at 1σ (green mesh). (d) Interaction network of C³⁵¹ and Cl‐amidine (carbons in salmon) in stereo, with hydrogen bonds and salt bridges as green lines and the binding distances in italics. (e) Detail of (b) in stereo (after an in‐plane 110°‐rotation and a vertical 50°‐rotation) showing only the Michalis loop (V²²⁶–H²³⁶) and the segment spanning the beginning of the helix of blade IV and the preceding loop segment (P²⁵⁵–H²⁶¹). The structures of PPAD‐T2 (carbons in plum) plus the complexes of wild‐type PPAD with a substrate (carbons in tan) and a substrate‐mimic (carbons in white) are shown. The hydrogen bond between N²³¹Nδ2 and Q²⁶⁰Oδ1 of PPAD‐T2, absent from PPAD‐T1, is pinpointed by a green arrow. A red arrow highlights the slight rearrangement caused by this hydrogen bond in segment P²⁵⁵–H²⁶¹. (f) Chemical structure of Cl‐amidine trifluoroacetate. A green arrow pinpoints the point of nucleophilic attack by C³⁵¹Sγ for covalent binding of the inhibitor.

The active‐site cleft is in the center of the β‐propeller, at the tooth cusp [Fig. 3(a,b)]. It is a narrow, funnel‐like hole, which accommodates the arginine side chain of substrates. The cleft is framed by loops connecting the propeller blades and contains the catalytic triad (C³⁵¹, H²³⁶, and N²⁹⁷), in which the cysteine sulfur is the catalytic nucleophile for catalysis.12 The main chain of substrates is tightly bound through R¹⁵², R¹⁵⁴, and Y²³³. The latter two residues explain the preferential exodeiminase activity of PPAD against C‐terminal arginines, as a C‐terminally extended endopeptidic substrate would collide with their side chains.

Of keynote importance is the loop connecting Blades III and IV, the “Michaelis loop” (residues V²²⁶–V²³⁷), which in unbound PPAD‐T1 transits between an open conformation (see Protein Data Bank [PDB] entry 4YT9) and a closed conformation in substrate and substrate‐mimic complexes (PDB 4YTG, 4YTB). In inhibitor‐bound PPAD‐T2, the Michaelis loop, which contains the three characteristic point mutations, is likewise in a closed conformation. The main difference with the PPAD‐T1 complexes is that replacement of G²³¹ with asparagine caused rearrangement of the Q²⁶⁰ side chain. In the PPAD‐T1 structures, this glutamine stabilizes the main‐chain carbonyl of P²³⁰ through its Nδ2 atom. In PPAD‐T2, Q²⁶⁰ is partially flexible and rotated about 90° around its χ ₂ bond, so that its Oδ1 atom hydrogen‐bonds N²³¹Oδ2 [Fig. 3(e)]. This led to slight displacement of the beginning of the helix of blade IV plus the preceding loop [Segment P²⁵⁵‐H²⁶¹; see Fig. 3(e)]. This rearrangement probably caused the PPAD‐T2 structure to crystallize in a new crystallographic space group with two molecules in the asymmetric unit.

Molecular determinants of PPAD‐T2 inhibition through Cl‐amidine

The inhibitor [chemical formula in Fig. 3(f)] was covalently attached to the catalytic C³⁵¹Sγ atom of PPAD‐T2. This resulted from the nucleophilic attack of the C³⁵¹ sulfur on the methylene group of the 2‐chloro‐1‐iminoethyl substituent of the N ⁵ atom of the central l‐ornithine moiety [green arrow in Fig. 3(f)] and concomitant leaving of a chloride. Thus, the inhibitor mimics a reaction intermediate (see refs. 5, 12) and is firmly anchored to the active site through 12 electrostatic and two hydrophobic interactions in addition to the covalent bond [Fig. 3(d)]. The part of the inhibitor proximal to C³⁵¹ plus the Sγ atom [the “warhead” according to ref. 20; see Fig. 3(f)] are salt‐bridged with D¹³⁰ and D²³⁸, and hydrogen‐bonded to N²⁹⁷ and H²³⁶. The central aliphatic part of the ornithine moiety (“side chain”) is hydrophobically pinched between the side chains of W¹²⁷ and I²³⁴. Finally, the inhibitor fragment mimicking the main chain of a substrate (“backbone”) is bound through seven hydrogen bonds by the side chains of R¹⁵², R¹⁵⁴, and Y²³³. This strong binding explains why PPAD‐T1 and PPAD‐T2 were significantly inhibited by Cl‐amidine in vitro, with IC₅₀ values of 8.89 ± 0.34 μM and 0.86 ± 0.09 μM, respectively.

Human PADs and PPAD share the gross β‐propeller architecture of their catalytic domains and some catalytic residues but strongly diverge in the loops shaping the active‐site cleft.5, 12, 21 PAD4 is the only human PAD for which experimental inhibitor complex structures have been published. Its complex with Cl‐amidine reveals similar binding to PPAD‐T2 of the inhibitor warhead and side chain (see PDB 2DW5 ¹⁷). However, PAD4 lacks structural counterparts of PPAD R¹⁵² and Y²³³, which explains the endodeiminase activity of human PADs versus the exodeiminase function of PPAD.

Concluding remarks and outlook

PPAD‐T2 is a clinical variant of PPAD‐T1 with enhanced activity in cell cultures that contains a triple amino‐acid substitution that affects the Michaelis loop, a key structural element for substrate binding. While no differences were found in thermal stability or substrate specificity, substrates were more weakly bound but faster citrullinated by PPAD‐T2. Overall, we will determine if the deviating activity of the two variants correlates with different levels of citrullinating activity in patients. Notably, preliminary results suggest that strains producing PPAD‐T2 may cause significantly more severe periodontal damage in patients than those secreting PPAD‐T1 (K. Gawron, M. Chomyszyn‐Gajewska, K. Łazarz‐Bartyzel and J. Potempa, personal communication).

The general inhibitor of human PADs, Cl‐amidine, blocked both PPAD‐T1 and PPAD‐T2 with IC₅₀ values in the micromolar range but PPAD‐T2 was more efficiently inhibited. The underlying mechanism for this difference will likewise be subject of future studies.

Finally, the crystal structure of the PPAD‐T2·Cl‐amidine complex revealed that extra interactions by residues that are absent from human PADs account for the characteristic exodeiminase activity of PPAD. This provides a blueprint for the design of specific PPAD inhibitors that do not target human PADs.

Materials and Methods

PPAD mutant generation

Mutations G²³¹N, E²³²T, and N²³⁵D were introduced into P. gingivalis reference strain ATCC 33277 by recombination. Fragments of DNA containing gene ppad, which encodes PPAD‐T1 (UniProt Q9RQJ2), and a tetracyclin resistance cassette were amplified with Phusion High‐Fidelity DNA Polymerase (Thermo Scientific) and cloned into the pUC19 vector using the Gibson Assembly Master Mix (New England Biolabs) to obtain plasmid pUC_T1. The ppad sequence was modified with the Gibson Assembly Method by amplifying pUC_T1 with primers encoding the desired mutations (MUT_FOR: 5′‐AACAATACTTATATCGACCATGTGGACTGTTGGGGCAAGTATTTGGC‐3′ and MUT_REV: 5′‐CACATGGTCGATATAAGTATTGTTCGGATCTTGTACCACATCATGATGTGTGATGC‐3′). The resulting plasmid was dubbed pUC_T2. Single point mutants G²³¹N, E²³²T, and N²³⁵D were obtained analogously and gave rise to plasmids pUC_T2a, pUC_T2b, and pUC_T2c. Electrocompetent P. gingivalis cells were prepared according to available protocols22. Briefly, liquid cell cultures of strain ATCC 33277 were grown to OD₆₀₀ = 0.6 in tryptic soy broth, centrifuged, washed twice in 1 mM magnesium chloride plus 10% glycerol, resuspended in the same buffer, and stored at −80°C. Plasmids were transformed by electroporation of 1 μg plasmid DNA at 2.5 kV for 4 ms. Transformed cells were incubated in tryptic soy broth overnight at 37°C in an anaerobic chamber, and plated on plates containing brain heart infusion with yeast extract, l‐cysteine (500 μg/mL), hemin (10 μg/mL), vitamin K (0,5 μg/mL), 5% defibrinated sheep blood, and tetracycline (1 μg/mL).

Protein production and purification

PPAD‐T1, PPAD‐T2, and the three single point mutants G²³¹N, E²³²T, and N²³⁵D were obtained by recombinant homologous overexpression from P. gingivalis cultures and purified as previously reported.12, 18 The resulting proteins spanned the CD and IgSF domains (residues A⁴⁴–A⁴⁷⁵), followed by a C‐terminal hexahistidine‐tag.

Activity assays

To assay PPAD activity in full cultures, suspensions of washed bacterial cells, and culture supernatants, substrate Nα‐acetylarginine was added at 10 mM final concentration in 100 mM Tris HCl pH 7.5, 5 mM 1,4‐dithiothreitol to initiate the reaction. The reaction was quenched by addition of 5 M perchloric acid after 60 min, and color was developed as described previously.23 The absorbance at 535 nm was measured using a Flexstation 3 microplate spectrophotometer (Molecular Devices).

Peptides of sequence P–P‐G‐F‐S‐P‐F‐R and G‐F‐S‐P‐F‐R‐S‐S (30 μg) were incubated at 37°C for 2 h with 1.5 mU of PPAD‐T1 or PPAD‐T2 in 100 mM Tris·HCl pH 7.5, 10 mM l‐cysteine. The peptides incubated in the same buffer served as controls. Reactions were terminated with 80 μL 0.5% trifluoroacetic acid (TFA) and subjected to HPLC chromatography in an Aeris 3.6 μm Peptide XB‐C18 100Å (150 × 4.6 mm) column (Phenomenex), which was resolved with a gradient of phase A (0.1% TFA) and phase B (80% acetonitrile, 0.08% TFA).

The kinetic parameters of deimination of peptides R‐P‐P‐G‐F‐S‐P‐F‐R and I‐H‐A‐R‐E‐I‐F‐D‐S‐R, both provided with an N‐terminal carboxyfluorescein fluorophore, by PPAD‐T1 or PPAD‐T2 (at 1.5 nM) were carried out at 37°C in 100 mM Tris·HCl pH 7.5, 5 mM 1,4‐dithiothreitol. Aliquots were taken at various time points, quenched with 0.5% TFA, and subjected to HPLC analysis as above. Peptide elution was monitored following the fluorescence of carboxyfluorescein. The amount of the modified peptides was quantified with a calibration curve for each peptide.

Inhibitory studies

The IC₅₀ values of inhibition of PPAD‐T1 and PPAD‐T2 by Cl‐amidine were determined with increasing inhibitor concentrations in a reaction buffer containing 100 mM Tris·HCl pH 7.5, 5 mM 1,4‐dithiothreitol. The reaction mixture was incubated with PPAD‐T1 or PPAD‐T2 (0.2 μM) at 37°C for 2 h prior to the addition of Nα‐acetylarginine (at 10 mM final concentration). After 60 min, reactions were quenched with 5 M perchloric acid, color was developed, and the absorbance at 535 nm was measured as above and compared with a calibrated curve to determine the concentration of citrulline. IC₅₀ values were determined using GraphPad Prism.

Thermal stability

The thermal stability of PPAD‐T1, PPAD‐T2, and PPAD mutants G²³¹N, E²³²T, and N²³⁵D was determined by differential scanning fluorimetry in a Prometheus NT.48 apparatus (NanoTemper). Proteins were diluted to 100 μg/mL in 20 mM Tris pH 8.0, loaded into nanoDSF standard grade capillaries, and heated from 30°C to 90°C. Protein unfolding was monitored by measuring the fluorescence intensity at the emission wavelengths (330 nm and 350 nm). Data were analyzed with the control software, which calculated the temperatures of midtransition (T _m).

Crystallization and diffraction data collection

Prior to crystallization, purified PPAD‐T2 was treated with 2 μM Cl‐amidine in 20 mM Tris·HCl, 150 mM sodium chloride, pH 7.5. Initial sitting‐drop crystallization assays were performed at the IBMB/IRB Automated Crystallography Platform at 4°C or 20°C. Successful conditions were scaled up to the microliter range in 24‐well Cryschem crystallization dishes (Hampton Research). The best crystals of PPAD‐T2·Cl‐amidine were obtained at 20°C with protein solution at 20–25 mg/mL and 100 mM sodium acetate, 15% polyethylene glycol 3,350, pH 4.6 as reservoir solution in 1 μL:1 μL drops. Crystals were cryo‐protected by rapid passage through drops containing increasing concentrations of glycerol (up to 15% [v/v]). Complete diffraction datasets were collected at 100 K from liquid‐N₂ flash cryo‐cooled crystals on a Pilatus 6M pixel detector (Dectris) at beam line XALOC24 of the ALBA synchrotron in Cerdanyola (Catalonia, Spain). Diffraction data were processed with the XDS and XSCALE programs.25 Crystals belonged to space group C2, contained two protein molecules per asymmetric unit, and diffracted to 1.6 Å resolution (see Table 1 for data processing statistics).

Table 1.

Crystallographic Data

Dataset	PPAD–T2 Cl‐amidine
Space group	C2
Cell constants (a, b, c, in Å; β in °)	123.84, 71.87, 104.72, 95.86
Wavelength (Å)	0.9793
No. of measurements/unique reflections	454,750/119,834
Resolution range (Å)	62.1–1.60 (1.70–1.60)a
Completeness (%)	99.4 (98.6)
R merge	0.074 (0.562)
R meas/CC1/2	0.086 (0.669)/0.998 (0.796)
Average intensity	11.1 (2.4)
B‐Factor (Wilson) (Å2)/Aver. multiplicity	24.8/3.8 (3.2)
Resolution range used for refinement (Å)	62.1–1.60
No. of reflections used (test set)	119,019 (814)
Crystallographic R factor (free R factor)	0.180 (0.193)
No. of protein residues/atoms/solvent molecules/covalent ligands/free ligands	842/6,643/1145/ 2 Cl‐amidine (40 atoms)/ 5 glycerol, 3 Na+
Rmsd from target values bonds (Å)/angles (°)	0.010/1.00
Average B‐factors (Å2) (overall/mol. A/mol. B)	24.2/19.6/25.1
All‐atom contacts and geometry analysisb
Residues in favored regions/outliers/all	820 (97.6%)/0/840
Residues with poor rotamers/bad bonds/bad angles	4 (0.6%)/0/0
All‐atom clashscore	0.49

^aData processing values in parenthesis are for the outermost resolution shell.

^bAccording to the wwPDB X‐ray Structure Validation Report.

Structure solution and refinement

The structure of PPAD‐T2 was solved by molecular replacement with the PHASER program26 and the coordinates of PPAD‐T1 (PDB 4YTB 12), for which positions 231, 232, 235, and 351 had been set to glycine or alanine to avoid model bias. Two solutions were found at the final Eulerian angles and fractional cell coordinates (α, β, γ/x, y, z) 301.1, 103.0, 328.0/−0.26, −0.50, 0.37 and 240.8, 72.9, 142.5/−0.23, −0.63, 0.14, respectively, with a final log‐likelihood gain of 19,601. These calculations were followed by automated density modification and model building with ARP/wARP.27 The resulting Fourier map was used for manual model building with the COOT program,28 which alternated with crystallographic refinement with PHENIX29 and BUSTER/TNT.30 The final model of PPAD–T2·Cl‐amidine contained residues A⁴⁴–M⁴⁶³ and A⁴⁴–E⁴⁶⁵ of molecules A and B, respectively, with Cl‐amidine moieties covalently attached to the Sγ atoms of the respective C³⁵¹ residues and one structural sodium ion for each protein molecule. Five glycerol molecules, an additional sodium ion in the bulk‐solvent region, and 1,145 solvent molecules completed the model. See Table 1 for refinement and model statistics.

Miscellaneous

Ideal atomic coordinates and parameters for the Cl‐amidine‐modified catalytic cysteine for crystallographic refinement were obtained from the PRODRG server (http://davapc1.bioch.dundee.ac.uk/cgi‐bin/prodrg/submit.html). Structure figures were prepared with the CHIMERA program.31 The experimental structure was validated with the wwPDB X‐ray Structure Validation Server (https://www.wwpdb.org/validation). The coordinates of P. gingivalis PPAD–T2·Cl‐amidine are available from the PDB at www.pdb.org (access code 6I0X).