Coagulation Factor IX is positioned at the merging point of the intrinsic and extrinsic blood coagulation cascades. Factor IXa (activated Factor IX) serves as the trigger for amplification of coagulation through formation of the so-called Xase complex, which is a ternary complex of Factor IXa, its substrate Factor X and the cofactor Factor VIIIa on the surface of activated platelets. Within the Xase complex the substrate turnover by Factor IXa is enhanced 200000-fold; however, the mechanistic and structural basis for this dramatic enhancement remains only partly understood. A multifaceted approach using enzymatic, biophysical and crystallographic methods to evaluate a key set of activity-enhanced Factor IXa variants has demonstrated a delicately balanced bidirectional network. Essential molecular interactions across multiple regions of the Factor IXa molecule co-operate in the maturation of the active site. This maturation is specifically facilitated by long-range communication through the Ile212–Ile213 motif unique to Factor IXa and a flexibility of the 170-loop that is further dependent on the conformation in the Cys168–Cys182 disulfide bond. Ultimately, the network consists of compensatory brakes (Val16 and Ile213) and accelerators (Tyr99 and Phe174) that together allow for a subtle fine-tuning of enzymatic activity.

INTRODUCTION

Blood coagulation serves to react on vessel injuries and prevent blood loss. At the same time, unspecific clogging of vessels must be avoided, which would otherwise result in thrombosis. Blood coagulation meets these multiple requirements by a hierarchical organization in a cascade and is thus able to rapidly amplify the initial triggering signal while allowing for regulation of the coagulation on different levels [1]. The enzymatic factors within the blood coagulation cascade belong to the chymotrypsin family of serine proteases, which are synthesized in an inactive zymogen form. Activation takes place within the cascade by subsequent proteolytic cleavage and is facilitated by the insertion of the newly generated neo-N-terminus into the activation pocket where it forms a salt bridge with Asp194 [2]. As a hallmark of the blood coagulation cascade, activation additionally requires complex formation with protein cofactors as well as additional exogenous modulators such as the phospholipid membrane of activated platelets [3].

Coagulation Factor IX and its cofactor Factor VIII are positioned at the intersection of the extrinsic and intrinsic arms of the coagulation cascade, underscoring their central role in blood coagulation. Both proteins co-operate in the so-called Xase complex to activate their downstream target Factor X, and the effect of the cofactor is illustrated by a 200000-fold increase in Factor X turnover by Factor IXa (activated Factor IX) upon assembly of the Xase complex [4]. The clinical relevance of these factors is also witnessed by the diseases haemophilia A and B that reflect defects in the Factor VIII and Factor IX genes respectively. Factor IX as a zymogen consists of a calcium-binding Gla domain, two EGF (epidermal growth factor)-like domains (EGF1 and EGF2), an activation peptide and the protease domain. Ten intradomain disulfide bonds contribute to the overall folding of the molecule as well as one additional disulfide bond between the EGF2 and the protease domains, keeping the light chain (Gla-EGF1-EGF2) and the heavy chain (protease domain) together after proteolytic activation and release of the activation peptide [5] (Figure 1A). The catalytic domain adopts a classical two-β-barrel arrangement, a signature fold for serine proteases that belong to the chymotrypsin family. The catalytic triad consists of His57 and Asp102 (chymotrypsin numbering) located in the N-terminal β-barrel as well as the catalytic Ser195 located in the C-terminal β-barrel. Three intradomain disulfide bonds and a conserved calcium-binding site maintain structural integrity in the protease domain. Flexible surface loops surround the substrate-binding site and facilitate substrate selectivity and distinct recognition sites for cofactors [6,7]. In Factor IXa, two such exosites have been defined: exosite I, consisting of the calcium-binding site, and exosite II, containing positively charged residues, that is known to bind heparin. Furthermore, exosite II has been suggested as an important site of interaction with Factor VIIIa in the Xase complex [5,8,9] (Figure 1B).

Domain organization and segments important in catalytic activity in Factor IX

Figure 1
Domain organization and segments important in catalytic activity in Factor IX

(A) Domain organization and amino acid sequence of human Factor IX (EC 3.4.21.22). Cysteine residues are marked in yellow and linked to their respective counterpart for disulfide bridge formation. γ, residues subjected to γ-carboxylation within the Gla domain; /\/\/\/, residues that are subject to glycosylation; P, phosphorylation; SO4, sulfurylation; OH, hydroxylation. The residue numbers refer to the full-length sequence. (B) Surface representation of the EGF2-protease domain in human Factor IXa (PDB code 1RFN) with segments in the protease domain that are shown to influence the catalytic activity marked in various colours. The corresponding chymotrypsin numbering references are indicated in bold black numbers. Overlaid in green is the P4/P4′ part of the antithrombin sequence in complex with a S195A Factor IXa variant (PDB code 3KCG) superimposed with the individual P4/P4′ residues marked in small letters. The S4/S1 sites in the protease domain and activation pocket (AP) are marked in bold black letters and the active site (AS) are marked in bold white letters.

Figure 1
Domain organization and segments important in catalytic activity in Factor IX

(A) Domain organization and amino acid sequence of human Factor IX (EC 3.4.21.22). Cysteine residues are marked in yellow and linked to their respective counterpart for disulfide bridge formation. γ, residues subjected to γ-carboxylation within the Gla domain; /\/\/\/, residues that are subject to glycosylation; P, phosphorylation; SO4, sulfurylation; OH, hydroxylation. The residue numbers refer to the full-length sequence. (B) Surface representation of the EGF2-protease domain in human Factor IXa (PDB code 1RFN) with segments in the protease domain that are shown to influence the catalytic activity marked in various colours. The corresponding chymotrypsin numbering references are indicated in bold black numbers. Overlaid in green is the P4/P4′ part of the antithrombin sequence in complex with a S195A Factor IXa variant (PDB code 3KCG) superimposed with the individual P4/P4′ residues marked in small letters. The S4/S1 sites in the protease domain and activation pocket (AP) are marked in bold black letters and the active site (AS) are marked in bold white letters.

Unlike the other homologous coagulation factors, i.e. Factor X, Factor VII, Protein C and thrombin, Factor IXa is a relatively poor protease in the absence of its cofactor, Factor VIIIa [3,4], and considerable effort has already been undertaken to identify key elements in the activation of Factor X by Factor IXa. Previous studies have identified Factor IXa variants with increased amidolytic or clotting activity by introduction of amino acid substitutions from the homologues Factor X sequence at specific sites in the protease domain such as residues 16, 98, 177 and 213 [1012] (Supplementary Figure S1). The comparison of these related enzymes together with site-directed mutagenesis for recombinant generation of Factor IXa variants and structural evaluation of different Factor IXa variants has helped to identify regions that could explain the low activity in the absence of its cofactor. These include, among others, the neo-N-terminus, the edge strand (residues 214–217), 99-loop, 162-helix (comprising residues 162–171), 134-loop, the calcium-binding site, 220-loop and the 148-loop [8,1126] (Figure 1B).

To understand the low-activity state of Factor IXa compared with related coagulation factors, we decided to build on the existing knowledge base, and investigate whether and how selected recombinant amino acid substitutions in the Factor IXa sequence lead to increased catalytic activity. In the present study, we have performed structural, kinetic and biophysical characterizations of a number of Factor IXa variants containing the V16I, K98T, R150A, Y177T, I212V and I213V substitutions or combinations thereof to evaluate the consequence of the individual amino acid substitutions and their synergetic stimulatory effect, if any. Using this approach, we demonstrate that the Ile212–Ile213 motif unique to Factor IXa is involved in the onset of increased catalytic turnover and that these changes influence a delicately balanced bidirectional network driven by essential molecular interactions in multiple regions of the Factor IXa molecule that together allow for a subtle fine-tuning of enzymatic activity.

MATERIALS AND METHODS

In the present study, we used Δ(Gla-EGF1) Factor IX (residues 87–415) produced in Escherichia coli where we engineered amino acid substitutions at Val16, Lys98, Arg150, Tyr177, Ile212 and Ile213 (chymotrypsin numbering) in different combinations. For brevity, we use the nicknames indicated in Figure 2(D). The ITTV212 variant was only used for crystallographic analysis. The R150A variant was included as a technical amino acid substitution to reduce potential autolysis in the 148-loop during production.

kcat/KM determination of Factor IXa variants

Figure 2
kcat/KM determination of Factor IXa variants

Kinetic experiments of Factor IXa variants were performed using (A) Pefa9, (B) Pefa10 or (C) Boc-IEGR chromogenic peptide substrates. Each individual value was determined from duplicate measurements and the results are means±S.E.M. (n=3). (D) Abbreviations (‘nicknames’) used for variants examined in the present study. Wt/wt, wild-type. In (A)–(C), ITTV refers to the ITTV213 variant.

Figure 2
kcat/KM determination of Factor IXa variants

Kinetic experiments of Factor IXa variants were performed using (A) Pefa9, (B) Pefa10 or (C) Boc-IEGR chromogenic peptide substrates. Each individual value was determined from duplicate measurements and the results are means±S.E.M. (n=3). (D) Abbreviations (‘nicknames’) used for variants examined in the present study. Wt/wt, wild-type. In (A)–(C), ITTV refers to the ITTV213 variant.

Engineering of expression plasmids with Factor IX variants

Plasmids for recombinant expression of Factor IX variants were constructed from a pET22b vector containing the nucleotide sequence corresponding to residues 87–415 of the EGF2 and catalytic domains from human wild-type Factor IX (EC 3.4.21.22) or a K98T Y177T variant [19]. Single site substitutions were introduced using PCR-based mutagenesis with custom-tailored primers. For double, triple and quadruple variants, an iterative process of introducing single site substitutions was followed. In addition, silent mutations were introduced in the sense primer to introduce a unique restriction site for quality control of selected clones after inverse PCR. Each primer was phosphorylated at the 5′-terminus before the inverse PCR to allow downstream ligation of the PCR product. Insertion of the expected nucleotide substitutions was verified with primer-specific restriction site analysis and nucleotide sequencing (Eurofins Genomics). The primer pairs used for variant construction are shown in Table 1. 

Table 1
Primer sequences
Mutation Direction Sequence 
R150A Sense 5′-GCATCAGCTTTAGTTCTTCAGTACCTTAGGGTACCACTTGTTGACCGAGCCACATG-3′ 
 Antisense 5′-ATTAATAGCTGCATTGTAGTTGTGGTG-3′ 
K98T Sense 5′-ACGTACAACCATGACATTGCCCTTCTGGAGCTCGACGAACCCTTAGTGCTAAACAGC-3′ 
 Antisense 5′-ATTAATAGCTGCATTGTAGTTGTGGTG-3′ 
V16I Sense 5′-ATTGTTGGTGGAGAAGATGCCAAACCCGGGCAATTCCCTTGGCAGGTTG-3′ 
 Antisense 5′-CCGAGTGAAGTCATTAAATGATTGGG-3′ 
I212V Sense 5′-GAAGGTACCAGTTTCTTAACTGGAGTTATTAGCTGGGGTGAAGAGTGTGC-3′ 
 Antisense 5′-CACTTCAGTAACATGGGGTCCCC-3′ 
I213V Sense 5′-GAAGGTACCAGTTTCTTAACTGGAATTGTTAGCTGGGGTGAAGAGTGTGC-3′ 
 Antisense 5′-CACTTCAGTAACATGGGGTCCCC-3′ 
Mutation Direction Sequence 
R150A Sense 5′-GCATCAGCTTTAGTTCTTCAGTACCTTAGGGTACCACTTGTTGACCGAGCCACATG-3′ 
 Antisense 5′-ATTAATAGCTGCATTGTAGTTGTGGTG-3′ 
K98T Sense 5′-ACGTACAACCATGACATTGCCCTTCTGGAGCTCGACGAACCCTTAGTGCTAAACAGC-3′ 
 Antisense 5′-ATTAATAGCTGCATTGTAGTTGTGGTG-3′ 
V16I Sense 5′-ATTGTTGGTGGAGAAGATGCCAAACCCGGGCAATTCCCTTGGCAGGTTG-3′ 
 Antisense 5′-CCGAGTGAAGTCATTAAATGATTGGG-3′ 
I212V Sense 5′-GAAGGTACCAGTTTCTTAACTGGAGTTATTAGCTGGGGTGAAGAGTGTGC-3′ 
 Antisense 5′-CACTTCAGTAACATGGGGTCCCC-3′ 
I213V Sense 5′-GAAGGTACCAGTTTCTTAACTGGAATTGTTAGCTGGGGTGAAGAGTGTGC-3′ 
 Antisense 5′-CACTTCAGTAACATGGGGTCCCC-3′ 

Expression, refolding and purification of zymogens

To initiate expression of the Factor IX variants, plasmid DNA was transformed into chemo-competent NiCo21(DE3) cells (New England Biolabs) using heat-shock procedures. The transformed cells were used to inoculate a pre-culture followed by overnight growth at 37°C with shaking at 250 rev./min. Large-scale expression was initiated by 1:100 dilution of the pre-culture into 800 ml of pre-warmed LB medium containing 100 μg/ml ampicillin. Cells were grown to a D600 of ∼0.8–1 (∼2 h at 37°C with shaking 220 rev./min) and expression was induced by addition of 800 μl of 1 M IPTG. In addition, 400 μl of 100 mg/ml ampicillin was added at the time of induction. Expression was continued for 4 h at 37°C with shaking at 220 rev./min and cells were harvested by centrifugation at 4000 g for 15 min and stored at −20°C until further use. IBs (inclusion bodies) of the expressed constructs were isolated from lysed cell pellets and purified by overnight incubation at room temperature in 50 mM Tris/HCl (pH 7.4), 500 mM NaCl, 20 mM EDTA and 2% Triton X-100 followed by washing three times in 50 mM Tris/HCl (pH 7.4), 50 mM NaCl and 20 mM EDTA using a Dounce homogenizer for resuspension of the IBs.

For refolding, 2 g of purified IBs was dissolved in 25 ml of 50 mM Tris/HCl (pH 8.5), 50 mM NaCl, 20 mM EDTA and 8.5 M guanidinium chloride with added 2-mercaptoethanol to 100 mM. Following unfolding, the pH of the sample was lowered to 3.5–4.0 in a single step, and the denaturing and reducing agents were removed by dialysis against 4 litres of 20 mM EDTA (pH 4.0) twice. Precipitated samples were dissolved in 25 ml of 50 mM NaCl, 8.5 M guanidinium chloride and 20 mM EDTA (pH 3.5) and refolding was initiated by dropwise dilution in two pulses of 12.5 ml of solubilized sample twice into 4 litres of pre-cooled (16°C) refolding buffer consisting of 50 mM Tris/HCl (pH 8.5), 150 mM NaCl, 20 mM CaCl2, 500 mM L-arginine, 0.3 mM cystine and 3 mM cysteine [10,27]. Each pulse was separated by 24 h and refolding continued for additional 48 h after the final pulse at 16°C with no stirring. Refolded sample was concentrated to 220 ml using a Centramate™ Tangential Flow system with 10K Omega™ membrane (Pall Corporation) and dialysed three times against 4 litres of buffer A (20 mM Tris/HCl, pH 7.4, 40 mM NaCl and 5 mM CaCl2). Precipitated non-correctly folded protein was removed by centrifugation. The lysate containing the refolded Factor IX variant zymogen was purified using a HiPrep Q FF 16/10 pre-packed column (GE Healthcare) equilibrated in buffer A. Zymogen at 90% purity could be collected in the flowthrough and the flowthrough was concentrated to 5–10 ml and purified further on a Superdex 75 16/60 column equilibrated in buffer A. Purified zymogen was concentrated to 2 mg/ml and used immediately for activation or saved in 1 ml aliquots at −20°C/−80°C until further use.

Activation and purification of Factor IXa variants

Activation of zymogen Factor IX variants was conducted using 0.003–0.007 unit of Factor XIa (HTI) per μg of Factor IX variant zymogen. The activation was conducted in 1 ml samples at 2 mg/ml at 20°C for 13–16 h. Following activation, Factor XIa and non-activated Factor IX zymogen were removed by size-exclusion chromatography using a Superdex 75 16/60 column equilibrated in either buffer X (20 mM Tris/HCl, pH 7.4, 25 mM NaCl and 2.5 mM CaCl2) if the sample was to be used for crystallization or buffer C (50 mM Hepes, pH 7.4, 150 mM NaCl and 5 mM CaCl2) if the sample was to be used for kinetic and biophysical characterization. Samples to be used for crystallization were concentrated to 6 mg/ml and stored in 100 μl aliquots at −80°C until further use. Samples to be used for kinetic and biophysical characterization were concentrated to 2 mg/ml and stored in 100 μl aliquots at −80°C until further use.

Active site titration (AST)

To ensure comparable results in kinetic assays from variants produced by refolding at different time points, AST was performed using MUGB (4-methylumbelliferyl 4-guanidinobenzoate hydrochloride monohydrate) as an active site titrant. The method was based on that of Payne et al. [28] and adapted to a 96-well plate format. Briefly, samples of a minimum volume of 15 μl of 0 (blank), 2 and 4 μM were prepared in an eight-well strip. A solution of 10 μM MUGB in buffer C and 0.1% PEG8000 was prepared and 40 μl was apportioned into selected wells of a half-area 96-well plate [NBS (non-binding surface)-treated; Corning]. The plate was briefly centrifuged to remove potential air bubbles and ensure homogeneous sample height. The assay was initiated by the simultaneous addition of 10 μl of the diluted protein samples and released 4-MU (4-methylumbelliferone) was measured for 2 h using a Tecan Infinite 200Pro plate reader at 25°C with excitation/emission wavelengths of 365/445 nm. The curves were baseline-subtracted using the blank sample and fitted using GraphPad Prism 6.0 to a single exponential equation with a linear component (to account for the low rate of deacylation) of the form:

 
formula

where parameters X and X0 are the time of completion and the time of reaction initiation, Y0 is an offset that accounts for background fluorescence in the assay system, amplitude is the amplitude of the burst phase under saturating assay conditions, and kobs is the observed first-order bulk rate constant associated with turnover of MUGB.

To calculate the active fraction of the titrated samples, the amplitudes were normalized to a standard curve of 4-MU and the fraction of total protein concentration was calculated. Exact concentrations of the 4-MU stock were determined using a molar absorption coefficient of 1900 M−1·mm−1 at 360 nm with a 1 mm path. The AST was repeated two or three times giving n=4 or 6. Active-site-titrated samples were adjusted to a concentration of 1.5 μM active site and stored in 85 μl aliquots at −80°C until further use.

Peptidolytic activity assays

Peptidolytic activity was measured against three different fluorigenic peptide substrates all containing AMC (7-amino-4-methylcoumarin) at the P1′ position: Pefa9 [MeS (methylsulfonyl)-dCHG (D-cyclohexylglycine)-Gly-Arg-AMC] (Pentapharm), Pefa10 [CH3SO2-dCHA (D-cyclohexylalanine)-Gly-Arg-AMC] (Pentapharm), and Boc-IEGR [t-butoxycarbonyl-Ile-Glu-Gly-Arg-AMC] (Bachem) (Supplementary Figure S2). All assays were carried out in a 384-well plate (NBS-treated; Corning) in a 25 μl total volume. Catalysis of the substrates was measured with 100 nM AST samples in buffer C and 0.1% PEG8000 at substrate concentrations from 0 to 1000 μM (Pefa9 and Boc-IEGR) or 0 to 500 μM (Pefa10). Released AMC was measured for 10 min using a Tecan Infinite 200Pro plate reader at 25°C with excitation/emission wavelengths of 342/440 nm. The initial velocity was obtained from linear regression analysis and transformed into nanomolar substrate concentrations using an AMC standard curve. Initial velocities were fitted to Michaelis–Menten kinetics and values for KM and kcat were obtained and used for calculation of the specificity constant kcat/KM. Each curve was obtained from duplicate measurements and the Michaelis–Menten constants obtained are the averages of three such duplicates. All data analysis was performed using GraphPad Prism 6.0.

Thermal shift assay (TSA)

Thermal stability of Factor IXa variants either in the absence or the presence of different inhibitors was assayed using a TSA with SYPRO Orange™ (Invitrogen) fluorescence as readout. The assays were conducted as previously described using a MX3005P RT (reverse transcription)–PCR instrument (Agilent Technologies) [29]. The assay was conducted in a 25 μl volume at a 0.07 mg/ml (∼2–2.2 μM total concentration) variant concentration and 5× SYPRO Orange™ in buffer C. For TSA in the presence of inhibitors, the assays were conducted in the presence of 200 μM (∼100× molar excess) of either pAB (p-aminobenzamidine), PPACK (D-Phe-Pro-Arg-chloromethylketone), EGR-CMK (H-Glu-Gly-Arg-chloromethylketone) or DEGR-CMK (1,5-dansyl-Glu-Gly-Arg-chloromethylketone) (Calbiochem) (Supplementary Figure S2). In addition, a titration series with 1×, 2×, 4×, 10×, 40× and 100× molar excess of PPACK was performed. Samples were pre-incubated with inhibitor at room temperature for 30 min before initiation of the thermal denaturation. The melting temperature (Tm) was obtained at V50 from fitting the raw data to a Boltzmann sigmoidal equation. The indicated Tm values were calculated as an average for three individual experiments.

DEGR-CMK labelling and emission analysis

Samples of Factor IXa variants were labelled with DEGR-CMK by incubation with 100-fold excess of inhibitor at 4°C overnight. Unbound excess inhibitor was removed with buffer exchange using a NAP-10 column (GE Healthcare) equilibrated in buffer C. Complete removal of unbound DEGR-CMK was verified by the absence of dansyl emission from the NAP-10 eluate from a 200 μM DEGR-CMK sample treated under equivalent conditions as the Factor IXa variants but in the absence of enzyme. DEGR-CMK-labelled Factor IXa variants were concentrated to ∼3 μM and stored until further use in 100 μl aliquots. For collection of dansyl emission spectra, 25 μl of either 1 μM free DEGR-CMK or 1 μM DEGR-CMK-labelled Factor IXa variant was added in a 384-well black microplate (NBS-treated; Corning) and emission spectra from 450 to 800 nm using an excitation wavelength of 330 nm were collected using a Tecan Infinite 200Pro plate reader. The spectra were collected at 25°C in 1 nm steps, and collected from four wells at a time to avoid evaporation. Raw data were smoothed using a second-order polynomial with 15 neighbours and the wavelengths of maximum emission were calculated using GraphPad Prism 6.0 software. The calculated emission maxima are the averages of three independent measurements.

Crystallization and data collection

Two 140/100 μl volumes of 6 mg/ml activated K98T or ITTV213 variants were diluted to 1500 μl in buffer X and, from a 0.5 M inhibitor stock in 100% DMSO of either PPACK or EGR-CMK, inhibitor was added in a 10-fold molar excess (5/4 μl) and the complex was allowed to incubate for a minimum of 2 h at 20°C. Following incubation, the samples were concentrated to their initial volume, filtered using a 0.22 μm spin filter and used for crystallization. Crystallization was set up at room temperature as sitting drops with 500 μl of reservoir as described by Zögg and Brandstetter [19]. In short, a standard grid in 0.1 M Mes (pH 6.5) and 18/20/22/24% PEG6000 was used for crystallization and 1 μl of double-distilled water was mixed with 0.7 μl of precipitant and 0.8/1/1.3 μl of protein sample at 6 mg/ml. Crystallization plates were stored at 20°C and crystals of the K98T complexes grew within 5–10 days and crystals of the ITTV213 complexes grew within 4–5 weeks. The most uniform crystals were obtained from samples in 18 or 20% PEG6000.

For crystallization of ITTV212–PPACK complex, the activated sample was added to 10-fold molar excess of PPACK directly after activation and the complex was purified on a Superdex 75 10/300 column in buffer X. The complex was eluted in two fractions of 1 ml, each of which was separately concentrated to 6 mg/ml and crystallized in the standard grid as described above.

Crystals were harvested, cryoprotected in reservoir solution and 2% PEG6000 of the crystal obtained and frozen in liquid nitrogen. The data collection of K98T–PPACK, K98T–EGR-CMK, ITTV213–PPACK and ITTV213–EGR-CMK was conducted at the ESRF (European Synchrotron Radiation Facility) (Grenoble, France) ID23-2 beamline at a wavelength of 0.873 Å (1 Å=0.1 nm) [30]. Data collection of ITTV212–PPACK was conducted at the ESRF ID29 beamline at a wavelength of 0.976 Å [31].

Data processing, refinement and structural analysis

Data collected for K98T–PPACK, K98T–EGR-CMK, ITTV213–PPACK and ITTV213–EGR-CMK complexes were processed using iMosflm [32], whereas the data used for the ITTV212–PPACK complex were obtained from the XDS autoprocessed files available from ESRF [33,34]. For scaling, molecular replacement and initial refinement, the CCP4i package was used [35]. For molecular replacement with Molrep [36], a PDB file of the Y94F K98T Y177T Factor IXa variant (PDB code 2WPH [19]) stripped of all waters and ligands was used. In addition, Rfree flagged HKL reflections from the mtz-file of this structure were used as a template to obtain unbiased reflections for the calculation of Rfree for the variant complexes. Initial refinement was conducted as a restrained refinement with REFMAC5 [37] using standard settings and subsequent refinement performed in Phenix 1.9-1692 [38]. The structure and electron density was analysed in COOT [39] and structure validation performed with Molprobity [40]. For Figure preparation and structural analysis, PyMOL 3.30 (Schrödinger) was used.

RESULTS

Kinetic evaluation of synergistic effect in V16I-, K98T- and I213V-containing Factor IXa variants

For the initial evaluation, we chose to focus on the kinetic effect of the V16I, K98T and I213V variants and combinations thereof using chromogenic peptide substrates containing arginine as the P1 residue and glycine as the P2 residue. Initially, two such peptide substrates were examined: Pefa9 and Pefa10. The two substrates differ by the size of their P3 residue and are optimized for Factor IXa (Pefa9) or Factor Xa (Pefa10) (Supplementary Figure S2), allowing us to probe whether the amino acid substitutions introduced in the various Factor IXa variants caused a change in substrate selectivity. Alignment of protease domain sequences from human Factor IX, Factor X, Factor VII, Protein C and thrombin show that Ile213 is a residue unique to Factor IX that is replaced by valine in all other homologous clotting factors [10,41] (Supplementary Figure S1). A previous study in which the native isoleucine residue was replaced by valine has demonstrated that this variant increases the catalytic efficiency (kcat/KM) of Factor IXa 1.3-fold [10]. Evaluation of R150A- and I213V-containing Factor IXa variant (VA) showed similar increases in the catalytic efficiency for amidolytic activity when the target substrate was either Pefa9 or Pefa10 with 1.5-fold and 1.2-fold increases respectively (Figures 2A and 2B).

Evaluation of the V16I variant on the R150A backbone (IA) showed no increase in the kcat/KM value for Pefa9, whereas the K98T (T) variant demonstrated a 2-fold improvement in kcat/KM (Figure 2A). The combination of the V16I variant with the I213V R150A variant (IVA) resulted in a negligible increase in activity compared with the VA variant; however, when the V16I variant was paired with the K98T R150A variant (ITA), further enhancements in activity were observed leading to a 2.5-fold increased kinetic efficiency compared with the wild-type, indicating a synergistic effect (Figure 2A). Consistent with their individual contributions, the combination of K98T, R150A and I213V variants (TVA) resulted in a 4-fold increase in kinetic efficiency. The strongest enhancement (6-fold) was observed with the ITV and ITVA variants (V16I, K98T, I213V on wild-type or R150A background). Intriguingly, the V16I, K98T, Y177T, I213V (ITTV213) variant reduced the kinetic efficiency to the 4-fold increase that was also observed with the TVA variant, suggesting that the Y177T substitution led to an activity decrease on the ITV background.

The Factor IXa variants exhibited similar kcat/KM values towards the substrate Pefa10 as observed for Pefa9 and could be ranked in the same order with respect to catalytic activity (Figure 2B). The differences in the evaluated kinetic parameters were found to be mostly due to kcat effects, whereas KM values were comparable among the different variants and substrates tested (Figures 2A and 2B, and Supplementary Table S1). This analogy suggested that neither the morpholino ring in the Pefa9 nor its shorter side chain in P3 compared with Pefa10 (Supplementary Figure S2) affected the kinetic activity. Given that the Factor IXa variants were based on amino acid substitutions from the Factor X sequence, we also evaluated whether the variants were more prone to catalysis of a peptide substrate containing the recognition sequence for Factor Xa, i.e. Ile-Glu-Gly-Arg (IEGR). The kcat/KM values evaluated for the Boc-IEGR substrate did not strictly follow the trend that was observed with the Pefa9 and Pefa10 substrates. Only the ITV and ITVA variants showed a 2-fold enhancement of the kinetic efficiency compared with wild-type (Figure 2C). However, compared with Pefa9, the kcat/KM values were 5–6-fold less for the ITV and ITVA variants with the Boc-IEGR substrate (Figures 2A and 2C, and Supplementary Table S1). Similarly, the observed turnover numbers (kcat) were less affected by the Val16, Lys98 and Ile213 variations, although they still indicated a trend similar to what was observed for the Pefa9 and Pefa10 substrates, ranking the variants in a similar order (Supplementary Table S1). The KM values of the Boc-IEGR substrate showed a significantly higher variation than those of the Pefa9 and Pefa10 substrates that partly reversed the increases observed for the turnover number (Supplementary Table S1).

Structure determination of the K98T and ITTV213 variants in complex with EGR-CMK and PPACK

To understand the molecular basis for the synergistic kinetic effect observed for the K98T variant in combination with the V16I and I213 variants, crystallization of the K98T, ITV and ITTV213 variants were initiated, which resulted in protein crystals of the K98T and ITTV213 variants in complex with either EGR-CMK or PPACK that were suitable for structure determination (Supplementary Figure S2). In addition, a crystal of a V16I K98T Y177T I212V (ITTV212) variant in complex with PPACK was obtained and the structure solved to evaluate the effect of the I213V substitution. The crystals diffracted up to 1.3–1.9 Å and were thus suitable for detailed analysis of small changes in individual residues (Table 2). The variants all adapted the same overall fold with a rigid core structure of two β-barrels surrounded by loop regions as observed in previously published structures of wild-type and Factor IXa variants [8,1719], with the exception of the 99-loop, as described previously [19] and discussed further below. Structural alignment of the protein backbone of our newly solved structures of K98T, ITTV212 and ITTV213 variants in complex with either EGR-CMK or PPACK could be aligned within 0.3 Å RMSD and within 1 Å RMSD when the alignment included the Factor IXa wild-type or S195A variant structure (PDB code 1RFN or 3KCG) (Figures 3A and 3B). The electron densities surrounding either the EGR-CMK or the PPACK inhibitor were well resolved (results not shown) and revealed a binding mode for the EGR-CMK inhibitor where the side chain of the P3 glutamate residue adopted a different orientation compared with the previously solved structure of the Y94F K98T Y177T variant in complex with EGR-CMK (PDB code 2WPM) (Figure 3A). The PPACK inhibitor adopted a similarly strained conformation as observed in the Y94F K98T Y177T variant in complex with PPACK (PDB code 2WPH; structure not shown) with the D-phenylalanine residue aligned at the entrance to the S4 site (Figure 3B).

Table 2
Data collection and refinement statistics

Statistics for the highest resolution shell are shown in parentheses.

Variant Ligand K98T EGR-CMK K98T PPACK ITTV212 PPACK ITTV213 EGR-CMK ITTV213 PPACK 
PDB ID 5JB8 5JB9 5JBA 5JBB 5JBC 
Wavelength (Å) 0.8729 0.8729 0.9763 0.8729 0.8729 
Resolution range (Å) 16.77–1.45 (1.502–1.45) 14.92–1.3 (1.346–1.3) 54.37–1.4 (1.45–1.4) 21.6–1.56 (1.616–1.56) 27.25–1.9 (1.968–1.9) 
Space group P212121 P212121 P212121 P212121 P212121 
Molecules in asymmetric unit 
Unit cell dimensions      
 α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 
a (Å) 44.23 44.27 44.04 44.31 44.34 
b (Å) 67.14 67.05 65.72 66.61 65.8 
c (Å) 96.83 97.15 96.76 97.05 97.32 
Data collection and refinement      
 Total reflections 228593 (23955) 320094 (33166) 254285 (15014) 250082 (23528) 377563 (34406) 
 Unique reflections 51455 (5079) 71402 (7087) 54056 (4991) 41579 (4122) 23122 (2258) 
 Multiplicity 4.4 (4.7) 4.5 (4.7) 4.7 (3.0) 6.0 (5.7) 16.3 (15.2) 
 Completeness (%) 99.15 (98.97) 99.43 (100.00) 96.34 (90.52) 99.73 (99.49) 100.00 (100.00) 
 Mean II 7.22 (1.80) 9.51 (1.81) 21.52 (3.49) 12.40 (2.10) 12.35 (3.14) 
 Wilson B-factor 19.46 15.58 19.04 20.48 25.07 
Rmerge 0.08072 (0.7014) 0.06481 (0.6968) 0.04709 (0.1471) 0.0698 (0.7482) 0.1505 (0.8739) 
Rmeas 0.09143 0.07304 0.05142 0.0765 0.1552 
 CC1/2 0.997 (0.545) 0.997 (0.65) 0.998 (0.98) 0.999 (0.618) 0.998 (0.706) 
 CC* 0.999 (0.84) 0.999 (0.888) 0.999 (0.995) 1 (0.874) 1 (0.91) 
 Reflections used for R-free      
  Rwork 0.1516 (0.2776) 0.1509 (0.2601) 0.1529 (0.3353) 0.1582 (0.2645) 0.1659 (0.2515) 
  Rfree 0.1852 (0.2926) 0.1762 (0.2808) 0.1790 (0.3320) 0.1901 (0.3081) 0.2130 (0.2933) 
 RMSD from ideal geometry      
  Bond length (Å) 0.014 0.016 0.015 0.016 0.005 
  Bond angles (°) 1.55 1.52 1.58 1.59 0.90 
 Ramachandran statistics (%) (favoured/allowed/outlier) 96/4/0 97/3/0 96/4/0 96/4/0 96/4/0 
 Clash score 3.13 2.62 3.72 3.34 0.90 
 Average B-factor 32.30 28.30 34.20 32.40 39.90 
Variant Ligand K98T EGR-CMK K98T PPACK ITTV212 PPACK ITTV213 EGR-CMK ITTV213 PPACK 
PDB ID 5JB8 5JB9 5JBA 5JBB 5JBC 
Wavelength (Å) 0.8729 0.8729 0.9763 0.8729 0.8729 
Resolution range (Å) 16.77–1.45 (1.502–1.45) 14.92–1.3 (1.346–1.3) 54.37–1.4 (1.45–1.4) 21.6–1.56 (1.616–1.56) 27.25–1.9 (1.968–1.9) 
Space group P212121 P212121 P212121 P212121 P212121 
Molecules in asymmetric unit 
Unit cell dimensions      
 α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 
a (Å) 44.23 44.27 44.04 44.31 44.34 
b (Å) 67.14 67.05 65.72 66.61 65.8 
c (Å) 96.83 97.15 96.76 97.05 97.32 
Data collection and refinement      
 Total reflections 228593 (23955) 320094 (33166) 254285 (15014) 250082 (23528) 377563 (34406) 
 Unique reflections 51455 (5079) 71402 (7087) 54056 (4991) 41579 (4122) 23122 (2258) 
 Multiplicity 4.4 (4.7) 4.5 (4.7) 4.7 (3.0) 6.0 (5.7) 16.3 (15.2) 
 Completeness (%) 99.15 (98.97) 99.43 (100.00) 96.34 (90.52) 99.73 (99.49) 100.00 (100.00) 
 Mean II 7.22 (1.80) 9.51 (1.81) 21.52 (3.49) 12.40 (2.10) 12.35 (3.14) 
 Wilson B-factor 19.46 15.58 19.04 20.48 25.07 
Rmerge 0.08072 (0.7014) 0.06481 (0.6968) 0.04709 (0.1471) 0.0698 (0.7482) 0.1505 (0.8739) 
Rmeas 0.09143 0.07304 0.05142 0.0765 0.1552 
 CC1/2 0.997 (0.545) 0.997 (0.65) 0.998 (0.98) 0.999 (0.618) 0.998 (0.706) 
 CC* 0.999 (0.84) 0.999 (0.888) 0.999 (0.995) 1 (0.874) 1 (0.91) 
 Reflections used for R-free      
  Rwork 0.1516 (0.2776) 0.1509 (0.2601) 0.1529 (0.3353) 0.1582 (0.2645) 0.1659 (0.2515) 
  Rfree 0.1852 (0.2926) 0.1762 (0.2808) 0.1790 (0.3320) 0.1901 (0.3081) 0.2130 (0.2933) 
 RMSD from ideal geometry      
  Bond length (Å) 0.014 0.016 0.015 0.016 0.005 
  Bond angles (°) 1.55 1.52 1.58 1.59 0.90 
 Ramachandran statistics (%) (favoured/allowed/outlier) 96/4/0 97/3/0 96/4/0 96/4/0 96/4/0 
 Clash score 3.13 2.62 3.72 3.34 0.90 
 Average B-factor 32.30 28.30 34.20 32.40 39.90 

Structure determination of K98T and ITTV213 variants and conformational changes in the Ile212–Ile213 motif

Figure 3
Structure determination of K98T and ITTV213 variants and conformational changes in the Ile212–Ile213 motif

(A) Cartoon representation of the peptide backbone from superposition of S195A Factor IXa variant in complex with antithrombin (PDB code 3KCG) and Y94F K98T Y177T Factor IXa variant (PDB code 2WPM), K98T and ITTV213 variants in complex with EGR-CMK. Only residues corresponding to P4/P4′ in antithrombin are shown in green. Colour codes are indicated. Arrows indicate the two observed conformations of the P3 glutamate side chain in PDB code 2WPM and the K98T ITTV213 variants respectively. Loop regions important for catalytic activity are marked according to chymotrypsin numbering. wt, wild-type. (B) Cartoon representation of the peptide backbone from superposition of wild-type Factor IXa variant in complex with pAB (PDB code 1RFN) and Y94F K98T Y177T Factor IXa variant (PDB code 2WPH), K98T and ITTV213 variants in complex with PPACK. Colour codes are indicated. “P3” indicates the D-phenylalanine residue in PPACK. Loop regions important for catalytic activity are marked according to chymotrypsin numbering. (C) Close-up view of the Ile212–Ile213 motif. The introduction of the I213V substitution in the ITTV213 variant resulted in a 0.5 Å displacement of its backbone co-ordinates. Additionally, this translational shift is associated with a shift of Gly197 by 0.5 Å, a 20° tilt around the backbone of Pro198, a ∼90° φ1 rotation of Ser45 and a 180° φ1 rotation of Ile212. Broken lines indicate polar interactions involving residues in the Ile212–Ile213 motif and the catalytically active Ser195. Colour codes are indicated.

Figure 3
Structure determination of K98T and ITTV213 variants and conformational changes in the Ile212–Ile213 motif

(A) Cartoon representation of the peptide backbone from superposition of S195A Factor IXa variant in complex with antithrombin (PDB code 3KCG) and Y94F K98T Y177T Factor IXa variant (PDB code 2WPM), K98T and ITTV213 variants in complex with EGR-CMK. Only residues corresponding to P4/P4′ in antithrombin are shown in green. Colour codes are indicated. Arrows indicate the two observed conformations of the P3 glutamate side chain in PDB code 2WPM and the K98T ITTV213 variants respectively. Loop regions important for catalytic activity are marked according to chymotrypsin numbering. wt, wild-type. (B) Cartoon representation of the peptide backbone from superposition of wild-type Factor IXa variant in complex with pAB (PDB code 1RFN) and Y94F K98T Y177T Factor IXa variant (PDB code 2WPH), K98T and ITTV213 variants in complex with PPACK. Colour codes are indicated. “P3” indicates the D-phenylalanine residue in PPACK. Loop regions important for catalytic activity are marked according to chymotrypsin numbering. (C) Close-up view of the Ile212–Ile213 motif. The introduction of the I213V substitution in the ITTV213 variant resulted in a 0.5 Å displacement of its backbone co-ordinates. Additionally, this translational shift is associated with a shift of Gly197 by 0.5 Å, a 20° tilt around the backbone of Pro198, a ∼90° φ1 rotation of Ser45 and a 180° φ1 rotation of Ile212. Broken lines indicate polar interactions involving residues in the Ile212–Ile213 motif and the catalytically active Ser195. Colour codes are indicated.

The Ile212–Ile213 motif is a regulatory element in catalysis

Comparison of the crystal structures did not reveal any conformational changes within the backbone of the K98T or the ITTV213 variants that could explain the differences observed in their kinetic efficiency, thus a detailed analysis on a residue to residue level was conducted. Analyses in the locale of the introduced I213V substitution showed a displacement of backbone co-ordinates in the area of the Ile213. This shift was accompanied by a conformational change in a number of surrounding residues: Gly197, Pro198, Ser45 and Ile212 leading to a packing of the contacts with Pro198 (Figure 3C). In contrast to I213V, the I212V substitution in the quadruple ITTV212 variant did not introduce any structural changes in the above-mentioned residues. The observed conformational changes surrounding the Ile212–Ile213 sequence, although subtle, were propagated further to the local environment of the catalytically active Ser195 (Figure 3C). In particular, the hydroxy group of Ser45 was involved in a hydrogen-bonding network that, via a conserved water molecule, interacts with the backbone carbonyl of Gly196 in both the K98T and ITTV213 variant structures indicating an intimate connection between the environment of Ser45 and the catalytic serine residue (Figure 3C). Ser195 was shown in all structures to adopt a static conformation due to the covalent linkage with the CMK (chloromethylketone) inhibitors. In the context of the observed conformational changes in the Ile212 and Ser45 side chains, we suggest that their hydrogen-bonding network facilitates productive orientation of the catalytically active Ser195 for optimal substrate cleavage. The conformational changes induced by the I213V variant not only affected the immediate surroundings, but also influenced changes more distant, such as the conformation of Tyr99 and Gln192. We therefore propose that a motif consisting of Ile212, Ile213, Gly197, Pro198 and Ser45, hereafter referred to as the Ile212–Ile213 motif, is a mechanistic trigger for the increased catalytic activity observed in the activity enhanced Factor IXa variants.

The 99-loop conformation controls access to the active site

Comparison of the 99-loop conformations present in wild-type, S195A or R150A Factor IXa structures (PDB codes 1RFN, 3KCG, 4Z0K, 4ZYU and 4ZAE) and previously solved structures of activity enhanced Factor IXa variants have identified two distinct conformations of the 99-loop that are independent of the co-crystalized inhibitors [8,1719]. The conformational state of the 99-loop, as observed in the previously published K98T Y177T (TT) (PDB code 2WPI) and Y94F K98T Y177T (PDB code 2WPH) structures [5,19], was also observed in the present structures of the K98T variant in complex with either the EGR-CMK or the PPACK inhibitor. Thus we have now demonstrated for the first time that the single amino acid substitution in the K98T variant is both sufficient and necessary for the observed changes in the 99-loop conformation reported in the previously published structures. An increase in catalytic activity of the K98T variant has been described previously [12] and can subsequently be correlated with the conformational state of the 99-loop; however, as shown in Figure 2, the kcat/KM values are also influenced by the identity of the substrate as revealed by a 2-fold increase in kcat/KM using the Pefa9 or Pefa10 substrates, but not the Boc-IEGR substrate. The K98T substitution was originally introduced into Factor IXa based on a sequence alignment with the more active Factor Xa and Factor VIIa coagulation factors (Supplementary Figure S1), which both contain threonine in the 98 position [12]. Structural alignment based on the K98T–PPACK and K98T–EGR-CMK structures solved in the present study with those of Factor Xa and Factor VIIa (based on PDB codes 1FAX and 2FIR respectively) showed that the Factor IXa residue at position 98 in the more active conformation of the 99-loop aligns with Thr98 in Factor Xa, but that the remaining residues, 93–97, in the 99-loop did not co-align. The alignment showed further that Thr98 in Factor VIIa aligned neither with Lys98 in the wild-type Factor IXa 99-loop conformation nor with Thr98 in the more active 99-loop conformation (Figure 4).

99-loop conformations in Factor Xa, Factor VIIa and Factor IXa and variants

Figure 4
99-loop conformations in Factor Xa, Factor VIIa and Factor IXa and variants

Left: superimposition of the 99-loop conformations in Factor Xa, Factor VIIa and Factor IXa variants. Right: sequence alignment of the 99-loop residues with residue 98 marked in green. Colour codes and PDB codes are indicated. wt, wild-type.

Figure 4
99-loop conformations in Factor Xa, Factor VIIa and Factor IXa and variants

Left: superimposition of the 99-loop conformations in Factor Xa, Factor VIIa and Factor IXa variants. Right: sequence alignment of the 99-loop residues with residue 98 marked in green. Colour codes and PDB codes are indicated. wt, wild-type.

Tyr99 and Phe174 guard access to the S4 site

Two important residues for substrate selectivity are Tyr99 and Phe174 and their side-chain conformations were influenced both by the Factor IXa backbone conformation and the P2 residue of the substrate. Comparison of Factor IXa variant structures that have been co-crystalized with a glycine residue at the P2 position showed that in activity enhanced variants such as K98T, ITTV213 and the previously reported Y94F K98T Y177T variant (PDB code 2WPM; results not shown), in complex with EGR-CMK, a virtually identical conformation of the amino acids lining the S2 and S4 sites were observed (Figure 5A). Comparison of these structures with the S195A Factor IXa antithrombin complex (PDB code 3KCG), which also contains a glycine residue in P2 did, however, show that, in this structure, Tyr99 has a small shift in the side chain. This can be attributed to its wild-type 99-loop conformation as well as a 1.8 Å shift of the Phe174 side chain that correlates with the backbone conformation of the 170-loop (Figure 5A). These observations show that the Tyr99 side chain adopts a near identical conformation, despite different 99-loop backbone conformations, when the P2 residue is glycine. The different Phe174 side-chain conformations can be linked to changes in the 170-loop backbone conformation and the P4/P3 residue that, in EGR-CMK-co-crystallized activity-enhanced Factor IXa variants, have no P4 residue and a glutamate in P3 compared with antithrombin (PDB code 3KCG) which has isoleucine and alanine in P4 and P3 respectively.

Structural comparison at the S2/S4 sites depending on ligands and variants

Figure 5
Structural comparison at the S2/S4 sites depending on ligands and variants

(A) Superimposition of key residues of the S2/S4 site in Factor IXa variants in complex with a P2 glycine residue. S195A Factor IXa–antithrombin complex, PDB code 3KCG, (dark blue), K98T–EGR-CMK (pale blue), ITTV213–EGR-CMK (lime green). Electron densities at 1σ around selected waters are shown as a mesh. wt, wild-type. (B) Superimposition of key residues of the S2/S4 site in Factor IXa variants in complex with a P2 proline residue. Porcine wild-type Factor IXa–PPACK, pFIXa (PDB code 1PFX), (light grey), K98T–PPACK (sky-blue), ITTV213–PPACK (dark green). Electron density at 1σ around the Phe174 side chain in the K98T–PPACK complex and selected waters are shown as a mesh. (C) Surface representation of Factor IXa variant complexes. The co-crystallized ligands are shown as sticks and coloured as indicated in (A) and (B). 99-Loop residues are coloured green and Tyr99 and Phe174 are coloured orange. S195A AT3, S195A Factor IXa–antithrombin complex, PDB code 3KCG. For simplicity, only the P4/P1 residues of antithrombin are shown as sticks. (D) Superimposition of key residues of the S2/S4 site in Factor IXa variants; K98T Y177T (TT, PDB code 2WPI, yellow) and ITTV212–PPACK complex (dark grey). Electron density at 1σ around the two Tyr99 rotamer side chains in the ITTV212–PPACK complex and selected waters are shown as a mesh.

Figure 5
Structural comparison at the S2/S4 sites depending on ligands and variants

(A) Superimposition of key residues of the S2/S4 site in Factor IXa variants in complex with a P2 glycine residue. S195A Factor IXa–antithrombin complex, PDB code 3KCG, (dark blue), K98T–EGR-CMK (pale blue), ITTV213–EGR-CMK (lime green). Electron densities at 1σ around selected waters are shown as a mesh. wt, wild-type. (B) Superimposition of key residues of the S2/S4 site in Factor IXa variants in complex with a P2 proline residue. Porcine wild-type Factor IXa–PPACK, pFIXa (PDB code 1PFX), (light grey), K98T–PPACK (sky-blue), ITTV213–PPACK (dark green). Electron density at 1σ around the Phe174 side chain in the K98T–PPACK complex and selected waters are shown as a mesh. (C) Surface representation of Factor IXa variant complexes. The co-crystallized ligands are shown as sticks and coloured as indicated in (A) and (B). 99-Loop residues are coloured green and Tyr99 and Phe174 are coloured orange. S195A AT3, S195A Factor IXa–antithrombin complex, PDB code 3KCG. For simplicity, only the P4/P1 residues of antithrombin are shown as sticks. (D) Superimposition of key residues of the S2/S4 site in Factor IXa variants; K98T Y177T (TT, PDB code 2WPI, yellow) and ITTV212–PPACK complex (dark grey). Electron density at 1σ around the two Tyr99 rotamer side chains in the ITTV212–PPACK complex and selected waters are shown as a mesh.

Comparison of the Tyr99 and Phe174 side-chain conformations in structures containing a P2 proline residue as is found in the PPACK-co-crystalized structures of the K98T, ITTV213 and K98T Y177T (TT) (PDB code 2WPI), and wild-type porcine Factor IXa (PDB code 1PFX), showed that the P2 proline will tend to cause a steric clash with Tyr99 that, as a consequence, induces a translocation towards the S4 site (Figure 5B). This effect is most pronounced for porcine Factor IXa where the Tyr99 side chain points directly into the S4 site. This outcome is enhanced further by small 99-loop backbone changes induced by variations in the 99-loop sequence in porcine Factor IXa (Figure 4). In the K98T–PPACK complex, the movement of the Tyr99 side chain towards the S4 site was less pronounced compared with porcine Factor IXa; nonetheless, the Tyr99 side chain is pushed toward the S4 site by the P2 proline residue (Figure 5B). In contrast, Tyr99 in the ITTV213–PPACK complex with the addition of the V16I and I213V amino acid substitutions evaded the P2 proline ‘pushing’ effect, thereby resulting in a conformation with the Tyr99 hydroxy group displaced 2.2 Å compared with Tyr99 in the K98T–PPACK structure. The phenol group of Tyr99 in the ITTV213–PPACK structure was rotated ∼32° compared with the ITTV213–EGR-CMK structure to accommodate the P2 proline residue instead of the P2 glycine residue.

Access to the S4 site was influenced not only by the Tyr99 conformation, but also by the conformation of the Phe174 side chain. In porcine Factor IXa, the Phe174 benzene ring is rotated compared with the S195A-Factor IXa–antithrombin structure (PDB code 3KCG) (Figures 5A and 5B). Together with the Tyr99 side chain, the Phe174 side chain conformation leads to a S4 site with a closed conformation as shown in orange in the surface representation in Figure 5(C). In the K98T–PPACK structure, the side chains of Phe174 were found in two rotameric states with the predominant rotamer forming a closed S4 site (Figure 5C), albeit with some flexibility as reflected in the corresponding electron density (Figure 5B). In the ITTV213–PPACK structure, Phe174 adopted a similar conformation as observed in the K98T–and ITTV213–EGR-CMK complexes (Figures 5A and 5B). Surface analysis of these structures showed that in the S195A-Factor IXa–antithrombin complex (PDB code 3KCG) as well as in both the K98T–and ITTV213–EGR-CMK complexes the S4 site adopts an open cleft-like conformation (Figure 5C). This open S4 site conformation was also observed in the ITTV213–PPACK complex showing that with a proline residue in the P2 position, Tyr99 and Phe174 shaped the S4 site into a more open conformation in the more active ITTV213 variant compared with the less active K98T variant (Figures 5B and 5C).

As described above, the ITTV212 variant in complex with PPACK did not exhibit the same conformational state attributed to the Ile212–Ile213 motif observed with the ITTV213 variant (Figure 3C). Nevertheless, this crystal structure could be appreciated as an intermediate that spanned the conformational space between the K98T or K98T Y177T (PDB code 2WPI) complexes with PPACK and that of the ITTV213–PPACK structure. In particular, the electron density supported the observation that Tyr99 occupied two distinct conformations corresponding to each of the rotameric states observed in the K98T–, K98T Y177T–and ITTV213–PPACK complexes respectively (Figures 5B and 5D). Figure 3(C) illustrates further the hydrogen-bonding network of the Ile212–Ile213 motif and shows the co-ordination with the catalytic triad and the residues lining the active-site cleft thereby influencing the conformational maturation of the substrate-recognition sites. These observations suggest that, with respect to the S4 site maturation, there is clear evidence for an induced-fit/conformational selection model that both involves the conformational state of the active-site cleft as well as the identity of the residues present in the P4/P2 positions of the substrate.

Conformational flexibility of the Cys168–Cys182 disulfide bond restricts the 170-loop and shapes the S4 site

The 162-helix (residues 162–171) precedes the so-called 170-loop (residue 169–174) and contains a disulfide bond linkage via Cys168 to Cys182 located in close proximity to the bottom of the S4 site. This disulfide bond was observed to occupy a broad conformational space, mainly defined by two rotamers of either left- or right-handed helical conformation, and, to a minor extent, a 180° flipped orientation that was only observed in complexes with EGR-CMK as well as in the K98T–PPACK structure (Figure 6A). The ITTV213 variant in complex with EGR-CMK exhibited both a right- and a left-handed helical conformation, as reflected by a smeared electron density with the right-handed conformation having the highest occupancy, whereas the disulfide was clearly defined with a right-handed configuration in the ITTV213–PPACK complex (Figure 6A). On the other hand, the K98T variant in complex with EGR-CMK only adopted the left-handed helical conformation with no signs in the electron density of a right-handed helical conformation. This was in contrast with the structure of K98T–PPACK complex where both left- and right-handed helical conformations were observed (Figure 6A). The different conformational states of the disulfide bond resulted in a dislocation of the cysteine backbone amide and carbonyl atoms, which resulted in an altered flexibility of the 162-helix depending on the nature of the Cys168–Cys182 disulfide bond as indicated by differences in the B-factors scaled against resolution (Figure 6B).

Multiple conformations of the Cys168–Cys182 disulfide and the 162-helix/170-loop

Figure 6
Multiple conformations of the Cys168–Cys182 disulfide and the 162-helix/170-loop

(A) Different conformational states of the Cys168–Cys182 disulfide and the corresponding electron density at 1σ. (B) Scaled B-factors in Factor IXa variants from residues 164–183 including the 162-helix, 170-loop and the Cys168–Cys182 disulfide. Scaled B-factors were calculated using PyMOL (Schrödinger) using the predefined preset B-factor putty script. The different Cys168–Cys182 disulfide conformations are indicated as: R, right-handed helix; L, left-handed helix; f, flipped conformation. Hydrogen-bonding distances from the backbone carbonyl of Leu169 and Phe174 to the 170-loop structural water molecule are indicated with broken lines.

Figure 6
Multiple conformations of the Cys168–Cys182 disulfide and the 162-helix/170-loop

(A) Different conformational states of the Cys168–Cys182 disulfide and the corresponding electron density at 1σ. (B) Scaled B-factors in Factor IXa variants from residues 164–183 including the 162-helix, 170-loop and the Cys168–Cys182 disulfide. Scaled B-factors were calculated using PyMOL (Schrödinger) using the predefined preset B-factor putty script. The different Cys168–Cys182 disulfide conformations are indicated as: R, right-handed helix; L, left-handed helix; f, flipped conformation. Hydrogen-bonding distances from the backbone carbonyl of Leu169 and Phe174 to the 170-loop structural water molecule are indicated with broken lines.

As discussed above, the conformation in the Phe174 side chain was, to a certain extent, defined by the residue occupying the S4 site; however, the rotameric state of the Phe174 side chain was also connected to the conformation of the 170-loop and the flexibility of the 162-helix. The 170-loop contains a water molecule that, via hydrogen-bonding to backbone atoms in Leu169 and Phe174, maintains the loop conformation. In structures where the Phe174 side chain adopted a conformation that led to an open S4 site conformation, this water molecule was very well resolved. In the K98T–PPACK complex, where the Phe174 rotamer leads to a closed S4 site conformation, the water molecule bridging the 170-loop was also very well resolved, but only partially occupied and displaced 0.6 Å which leads to a 0.5 Å increase in the hydrogen-bonding distance to the Phe174 carbonyl (Figure 6B). The bridging of the 170-loop may not only be dependent on the Phe174 carbonyl position, but also on the conformational flexibility in the entire 162-helix as reflected in differences in the hydrogen-bonding distance of the Leu169 carbonyl and the 170-loop water molecule (Figure 6B). Given the observed improvement in activity of the ITTV213 variant with D-P3-containing peptide substrates (Pefa9 and Pefa10), we suggest that the conformational transitions within the Cys168–Cys182 disulfide linkage and the bridging of the 170-loop reflects the enzyme maturation in the S4 site.

Substrate selectivity is guarded by the Gln192 side-chain conformation

Evaluation of the structures of the present study and those previously published indicate that the entrance to the S1 and S3 sites is influenced by the side chain of Gln192 as well as water networks that connected the substrate to the 148-loop. In both of the EGR-inhibited K98T and ITTV213 structures, a well-resolved electron density around the P3 glutamate side chain was observed. The Gln192 residue in these structures was found in one distinct conformation, Q192_d, with the Gln192 carbamide binding a less-resolved water molecule that links the P3 glutamate side chain with the side chain of His147. His147 was further water-bridged to Arg143; nevertheless, the distance for a water-bridged hydrogen bond was only observed in the ITTV213 complex and not the K98T complex despite a combined high degree of flexibility for Arg143 and His147 (Figure 7A).

Structural comparison at the S3 site depending on ligands and variants

Figure 7
Structural comparison at the S3 site depending on ligands and variants

Superimposition of key residues in the S1/S3 site in Factor IXa variants. Broken lines indicate selected potential polar interactions. (A) K98T–EGR-CMK (pale blue) and ITTV–EGR-CMK (lime green). Electron densities at 1σ around crystallographic waters, Arg143, His147 and Gln192 and the P3 glutamate side chain in EGR-CMK are shown as a mesh. (B) K98T–PPACK (light blue) and K98T Y177T–PPACK (TT-PPACK, PDB code 2WPI, orange) Electron densities at 1σ around crystallographic waters, Arg143, His147 and Gln192 side-chain rotamers are shown as a mesh. (C) ITTV212–PPACK (dark grey) and ITTV213–PPACK (dark green). Electron densities at 0.6σ around crystallographic waters, Arg143, His147 and the Gln192 side-chain rotamers are shown as a mesh.

Figure 7
Structural comparison at the S3 site depending on ligands and variants

Superimposition of key residues in the S1/S3 site in Factor IXa variants. Broken lines indicate selected potential polar interactions. (A) K98T–EGR-CMK (pale blue) and ITTV–EGR-CMK (lime green). Electron densities at 1σ around crystallographic waters, Arg143, His147 and Gln192 and the P3 glutamate side chain in EGR-CMK are shown as a mesh. (B) K98T–PPACK (light blue) and K98T Y177T–PPACK (TT-PPACK, PDB code 2WPI, orange) Electron densities at 1σ around crystallographic waters, Arg143, His147 and Gln192 side-chain rotamers are shown as a mesh. (C) ITTV212–PPACK (dark grey) and ITTV213–PPACK (dark green). Electron densities at 0.6σ around crystallographic waters, Arg143, His147 and the Gln192 side-chain rotamers are shown as a mesh.

In contrast with the EGR-CMK structures, for the PPACK structures evaluated, we observed different conformational states of the Gln192 side chain (Figures 7B and 7C). Electron density surrounding Gln192 indicated two rotamers exemplified by the K98T–PPACK and K98T Y177T–PPACK (PDB code 2WPI) structures that we refer to as Q192_a and Q192_b respectively (Figure 7B). Furthermore, the electron density around Gln192 in the ITTV213–PPACK structure suggested a population with an additional rotamer (Q192_c) (Figure 7C). As discussed above, the ITTV212–PPACK structure characterized an intermediate in the conformational space between the K98T–PPACK and ITTV213–PPACK structures. Indeed, the electron density indicated that both the Q192_b and Q192_c rotamers were present in the ITTV212–PPACK structure that again allowed us to probe an intermediate conformation (Figure 7C). The multiple rotamers in the ITTV212–PPACK structure indicated further that the conformational state of the Ile212–Ile213 motif could influence the conformation of the Gln192 rotamer. The Q192_c rotamer in the ITTV213–PPACK structure was stabilized via hydrogen-bonding to the Arg143 side chain, whereas a water molecule provided a salt bridge to the backbone of His147. As observed for the complexes with EGR-CMK, His147 was characterized by a high degree of flexibility as indicated by a poorly resolved electron density surrounding the side-chain imidazole. In general, the 148-loop was found to display a high degree of flexibility as underscored by the low electron density surrounding solvent-exposed side chains as well as large B-factors for the backbone atoms.

Generally, we observed diverse solvent networks that were characteristic of the respective variants and served to couple the active-site residues to the peptidyl inhibitor via multiple hydrogen bonds. In the EGR-CMK complexes, Gln192 adopted a preferred rotamer bound to the acidic P3 glutamate side chain, whereas, in the PPACK complexes, the side chain of the D-phenylalanine residue at P3 positioned at the entrance to the S4 site resulted in the absence of a potential interaction partner that allowed the side chain of Gln192 to become highly flexible.

Conformational flexibility dependent on structural motifs

To understand the conformational flexibility observed in the Ile212–Ile213 motif, the 99-loop, the 170-loop, the 162-helix and Gln192, we used our experimental electron density and refined structural models for an ensemble refinement. Ensemble refinement is a method merging molecular dynamics and structural refinements against the experimentally obtained electron density to probe the inherent flexibility within a protein crystal [4245]. Consistent with blurred electron densities, the ensemble refinement probed the 60-loop and the 148-loop as the most flexible segments (Supplementary Figure S3). Additionally, flexibility was identified in the 186-, 220- and 35-loops. The flexibility in the 170-loop correlated with motions in the 162-helix where multiple conformational states of the Cys168–Cys182 disulfide underlined its flexible nature as was also observed from the different rotameric states identified from the electron density using a single model refinement. A more detailed analysis of the Ile212–Ile213 motif showed that Ser45 adopted a number of conformations in all structures, but that, in the variant complexes, the distribution showed that the predominant rotamers corresponded to the ones observed in their static state. The conformation of Ile212 or Val212 only showed limited flexibility in the K98T and ITTV212 variant complexes (results not shown for ITTV212) respectively, whereas multiple conformations were observed for the ITTV213 variant, thus supporting the hypothesis that increased flexibility in Ile212–Ile213 motif facilitates increased catalytic turnover as observed in Figures 2(A) and 2(C). The flexibility of the Ile212–Ile213 motif also propagated to the S4 site and Gln192 with the most pronounced flexibility observed for the PPACK complexes as exemplified by multiple conformations of Tyr99 and Gln192 (Supplementary Figure S3). These observations underscored the role of Ile212 in the Factor IXa wild-type structure as a braking element that can be linked to the low activity of this enzyme.

Biophysically probing the integral molecule using melting experiments

The proteolytic activation of chymotrypsin-like serine proteases is accompanied by a major conformational disorder–order transition [2]. We hypothesized that the activation state from Factor IX zymogen to the fully active state might reflect a quasi-continuous degree of activity maturation, as present in the protein variants studied. Consequently, changes in the thermal stability of the protein variants should serve as a complementary reporter on the protease maturation status.

We therefore exploited biophysical properties of the variants that could provide an orthogonal view on their molecular properties. Using a TSA with SYPRO Orange™ as a fluorescent probe, we evaluated the thermal transition states of the different variants in the absence and presence of different inhibitors. The Tm of the apo (uncomplexed) enzymes was in the range 50–57°C (Supplementary Table S2). Ranking the variants according to increasing melting temperatures of their apo form did not result in the same order as that observed when ranking with respect to catalytic efficiency, and thus there was no correlation between the thermal stability of the apo enzyme variants and the measured kinetic constants (Figure 2, and Supplementary Tables S1 and S2).

Complex formation with the CMK-based substrate scaffold inhibitors showed variant dependent effects, albeit at different levels. Importantly, these effects reflected the trend observed in the kinetic measurements (Figures 2 and 8). The thermal stabilization was least pronounced with EGR-CMK, with a maximum increase in Tm of 3°C for the ITVA variant (Figure 8A and Supplementary Table S3). The DEGR-CMK showed more pronounced effects which fell into four categories: a ∼2°C destabilization was observed for wild-type Factor IXa and the R150A (A) variant; no changes in Tm were observed for the IA and VA variants; an intermediate increase of ∼2°C in Tm was found for the IVA, T and ITA variants; and the strongest effect was observed for the TVA, ITTV213, ITV and ITVA variants with an increase in Tm of ∼6°C (Figure 8A and Supplementary Table S3).

Structural interpretation of thermal stability of Factor IXa complexes

Figure 8
Structural interpretation of thermal stability of Factor IXa complexes

(A) ΔTm indicating the change in melting temperature relative to the apo variants (Supplementary Table S3). Factor IXa variants were analysed at 100-fold molar excess of pAB (dark blue), EGR-CMK (lime green), DEGR-CMK (orange) or PPACK (grey). (B) Emission maximum of DEGR-CMK-labelled Factor IXa variant. The horizontal broken red line corresponds to the emission maximum of the wild-type enzyme (λ=553 nm). In (A) and (B), ITTV refers to the ITTV213 variant; wt, wild-type. Results are means±S.E.M. (n=3). (C) The complex structure of DEGR-CMK with tPA (PDB code 1BDA, orange) was used as a model to visualize the proposed binding mode of DEGR-CMK in Factor IXa variants indicated with the wild-type pAB (PDB code 1RFN, dark blue), ITTV212–PPACK (dark grey) and ITTV213–EGR-CMK (lime green) complexes. Tyr99 and Phe174 with proposed interactions with the dansyl group are shown as sticks.

Figure 8
Structural interpretation of thermal stability of Factor IXa complexes

(A) ΔTm indicating the change in melting temperature relative to the apo variants (Supplementary Table S3). Factor IXa variants were analysed at 100-fold molar excess of pAB (dark blue), EGR-CMK (lime green), DEGR-CMK (orange) or PPACK (grey). (B) Emission maximum of DEGR-CMK-labelled Factor IXa variant. The horizontal broken red line corresponds to the emission maximum of the wild-type enzyme (λ=553 nm). In (A) and (B), ITTV refers to the ITTV213 variant; wt, wild-type. Results are means±S.E.M. (n=3). (C) The complex structure of DEGR-CMK with tPA (PDB code 1BDA, orange) was used as a model to visualize the proposed binding mode of DEGR-CMK in Factor IXa variants indicated with the wild-type pAB (PDB code 1RFN, dark blue), ITTV212–PPACK (dark grey) and ITTV213–EGR-CMK (lime green) complexes. Tyr99 and Phe174 with proposed interactions with the dansyl group are shown as sticks.

The effects on Tm were strongest for the PPACK complex structures and fell into two major categories. The wild-type Factor IXa as well as variants A, IA and VA showed no changes in Tm, whereas the IVA variant had only an intermediate effect of ∼3°C. The T, ITA, TVA, ITTV213, ITV and ITVA variants were strongly stabilized with Tm increases of 12–18°C. The stabilization of these variants with PPACK exhibited a biphasic concentration-dependence in thermal unfolding as shown in Supplementary Figure S4. Biphasic, i.e. two-step, thermal unfolding in multidomain proteins usually indicates an uncoupling of individual domains, e.g. the EGF2 and the catalytic domain, or the two β-barrel subdomains within the protease. To analyse potential changes in the binding mode of the DEGR-CMK inhibitor, dansyl emission spectra were collected. Compared with the emission maximum of free DEGR-CMK in buffer, all DEGR-CMK-labelled Factor IXa variants showed a blue shift, indicating that the dansyl moiety of DEGR-CMK was bound in close proximity to the enzymes. The differences in the emission maxima of the individual DEGR-CMK-labelled variants were generally small and hardly significant, indicating a similar conformation of the dansyl moiety in all variants (Figure 8B).

Stabilization of complexes with EGR-CMK could only be achieved for variants with a K98T substitution, yet the changes were much smaller compared with the stabilization introduced by DEGR-CMK and PPACK. This difference could be explained from the lack of a stabilizing interaction from a P4 residue, which in the DEGR-CMK was a dansyl moiety and in PPACK occupies the position of the D-phenylalanine residue.

Consistently, the strongest effects were observed for the latter two inhibitors. The remarkable destabilization of the DEGR-CMK complexes with wild-type Factor IXa and the R150A variant strongly points towards a steric clash of the inhibitors with these variants. Superimposition of the crystal structure of tPA (tissue plasminogen activator) in complex with a 2,5-DEGR-CMK inhibitor (PDB code 1BDA) with the structure of the wild-type Factor IXa–pAB complex (PDB code 1RFN), and the ITTV212–PPACK and ITTV213–EGR-CMK complexes indicate that this destabilization is probably mediated by the dansyl group which is found in close proximity to Phe174 in Factor IXa (Figure 8C). This conclusion is consistent with the observed flexibility of the 170-loop. The strongest stabilization for the DEGR-CMK complex was observed for the triple variants, ITV and ITVA, which underscores their relevance in substrate recognition. Of particular interest, the observed Tm effect of IA to IVA and similarly of T to TVA with DEGR-CMK supports the notion that conformational changes in the Ile212–Ile213 motif induces maturation of the S4 site.

DISCUSSION

Several different structural regions in chymotrypsin-like serine proteases are known to affect their catalytic activity [2123] and in the present study we applied structural, kinetic and thermal stability studies on a number of Factor IXa variants to evaluate whether any of these regions were correlated to activity-enhancing variants. Chymotrypsin-like serine proteases are proteolytically activated, resulting in formation of a salt bridge between the neo-N-terminus and the side chain of Asp194 [2]. Changes in the environment around this salt bridge, the so-called activation pocket, have been shown to positively and negatively affect the enzymatic activity in Factor IXa homologues [46,47]. Given the critical role of the neo-N-terminus for conformational maturation, a V16I variation has previously been introduced into Factor IXa variants and demonstrated to have increased clotting activity [11]. Val16 is unique to Factor IX, whereas all related (and more active) coagulation factors feature an isoleucine as residue 16 (Supplementary Figure S1), as do the super-active digestive enzymes chymotrypsin and trypsin. The V16I variant did not show a stimulating effect as a single variant for amidolytic substrates; however, in combination with other substitutions an increase in catalytic activity was observed (Figure 2). This behaviour was mimicked in the TSA with DEGR-CMK and PPACK, where an increase in Tm was only observed for the V16I variant in combination with additional amino acid substitutions. These observations indicate that the V16I variant did not introduce changes in the S1 pocket, but that in combination with the K98T or I213V variants, induced a stabilization in the proximity of the S4 site as indicated by the difference in stabilization between EGR-CMK (no P4 residue) with DEGR-CMK (harbouring a P4 dansyl group). This is consistent with the structural analysis where no differences in the oxyanion hole near to the P1 arginine residue were observed. In addition, analysis of the ITTV212–PPACK structure showed that the V16I variant could induce conformational changes at ∼10 Å distance as observed in Gln192 or Tyr99. The altered 99-loop conformation caused by the K98T variant could thereby provide an explanation for the increased stabilization of the ITA variant (Figure 8); however, the thermal stabilization of the IVA variant indicated that a co-operative effect from conformational changes in the Ile212–Ile213 motif and the V16I variant in the neo-N-terminus also led to a stabilization of the S4 site as indicated by similar Tm increases for the K98T, ITA and IVA variants in complex with DEGR.

The observed conformational changes in the Ile212–Ile213 motif occur in the vicinity of the neo-N-terminus insertion. Structural comparison of this motif with more active coagulation factors showed that the β-branching of Ile212 in the ITTV213 variant adopts the same conformation as in Factor VIIa and thrombin, whereas in Factor Xa Ile212 adopts neither of the conformations observed in the K98T or ITTV213 variant (Supplementary Figure S5). The residue in position 45 is a threonine in Factor Xa, Factor VIIa and thrombin (Supplementary Figure S1). In Factor Xa and Factor VIIa, the hydroxy group of this threonine residue has the same orientation as the serine hydroxy group in the K98T and ITTV212 structure, with the oxygen pointing directly towards the Pro198 side-chain ring. On the other hand, the threonine hydroxy group in thrombin assumes the same orientation as the serine hydroxy group in the ITTV213 variant where the oxygen is directed away from the Pro198 side-chain ring (Supplementary Figure S5). Thus the conformation of Ile212 and Ser45 in the Ile212–Ile213 motif in the more active ITTV213 variant resembles that of the most active coagulation factor, thrombin, supporting our hypothesis that the conformational changes observed in the Ile212–Ile213 motif of the more active variant, could be a trigger to increase catalytic activity in Factor IXa.

Increased catalytic activity is driven by a number of kinetic constants, among them the Michaelis–Menten constant, KM, and the turnover rate, kcat. The kinetic characterization of the different Factor IXa variants analysed in the present study did not show any clear trends in KM changes, indicating that the increase in catalytic activity for the most active variants was driven by kcat. Differences in kcat often reflect differences in the productiveness of substrate binding and substrate activation. The latter is to a large extent exerted by the proper orientation of the substrate, allowing optimal polarization of the scissile peptide bond and thus enables effective nucleophilic attack by the catalytic Ser195. Alignment of the substrate in a productive manner is highly influenced by S4 site maturation. In Factor IXa, this is influenced not only by the active state of the enzyme, but also by the residues that are present at the P4/P2 positions in the substrate. Superposition of Phe174 in the K98T–PPACK structure with that of S195A Factor IXa in complex with antithrombin (PDB code 3KCG [8]), which harbours an isoleucine residue in P4, would thus indicate a steric clash with Phe174 once this is in a conformation that leads to a closed S4 site (Supplementary Figure S6). This illustrates the need for a flexible side chain in S4 site maturation to facilitate binding of the P4 residue in the two physiological Factor IXa substrates: an isoleucine residue in antithrombin and a glycosylated asparagine residue in Factor X. The possibility of a steric clash is supported by the thermal destabilization of wild-type Factor IXa and the R150A variant in complex with DEGR, indicating a potential limitation of Phe174 flexibility exhibited by the low-activity variants that renders them less compatible with a dansyl group in the P4 position. The observation that key residues shown to change conformation in the more active ITTV213 variant are involved in a hydrogen-bonding network affecting the catalytically active Ser195 and Asp102 (via the Ile212–Ile213 motif) as well as aligning the peptide backbone of P4/P1 in a productive manner underscores the importance of these residues for subtle fine-tuning of Factor IXa activity.

The flexibility of Phe174 is highly linked to the conformation in the 170-loop as well as the 162-helix. Conformational changes in this helix can be correlated with the helical conformation of the disulfide between Cys168 and Cys182. Different rotamers of this disulfide have been shown in the recent high-resolution structures of wild-type, S195A and R150A Factor IXa variants in complex with various inhibitors (PDB codes 3KCG, 4Z0K, 4ZYU and 4ZAE). The side chain of Phe174 in two of these structures (PDB codes 4ZAE and 4YZU) aligns perfectly with the Phe174 in its open conformation as found in the K98T–EGR-CMK and ITTV213–EGR-CMK/PPACK structures with a water bridge to the 170-loop and the disulfide predominantly in its right-handed helical conformation. In the other two structures (PDB codes 3KCG and 4Z0K) Phe174 occupies a parallel shifted positioning of the side chain, lacks the water molecule that bridges the 170-loop and the Cys168–Cys182 disulfide adopts a left-handed helical conformation that does not correspond to the ideal geometry and thus preferred energy for disulfide bonds (results not shown). Similarly, the K98T and ITTV213 structures adapted different disulfide conformations and 170-loop bridging reflecting not only their active state, but also the bound substrate analogue. The analysis of flexibility within the structures solved in the present study using either B-factor analysis or Ensemble refinement against the calculated electron density maps identified the 162-helix as flexible along with movement of the water in the 170-loop and conformational shifts in the Cys168–Cys182 disulfide. The correlation of the 162-helix and Cys168–Cys182 disulfide integrity have been discussed for both thrombin and a Factor Xa–trypsin chimaeric molecule [48,49] and highlight the correlation of the 162-helix and Cys168–Cys182 disulfide with S4 site maturation. The conformation of the 162-helix and 170-loop has been reported to influence Factor IXa activity, affecting its S4 site and possibly also its interaction with factor VIIIa [24]. Indeed, molecular dynamics simulations of the Factor VIIIa and Factor IXa interaction suggest an intricate hydrogen-bonding network involving, among others, the 170-loop residues Thr172 and Phe174 in Factor IXa [9]. In addition, ethylene glycol stimulates Factor IXa activity and a binding site was localized in proximity to the 170-loop (Ser171) and Glu217 [19,25] that potentially affects the flexibility observed in the region of the 162-helix. Furthermore, a number of mutations in the 162-helix have been identified in patients with haemophilia B, and kinetic studies have shown that particularly the L169F substitution, found in 38 haemophilia B patients [50], abolishes amidolytic activity as well as Factor VIIIa-dependent Factor X activation [51]. Altogether, this highlights a potential role of the 170-loop and the Cys168–Cys182 disulfide, not only for substrate selectivity but also as a regulatory element in Factor VIIIa stimulation of Factor IXa activity.

Whereas the 162-helix and Phe174 comprise one edge of the S4 site, the opposite edge is characterized by the 99-loop, particularly Tyr99. In related enzymes such as tPA or Factor Xa, Tyr99 dictates a rather strict preference for glycine in P2 [5254]. However, the physiological substrate of Factor IXa exhibits threonine in P2, not glycine. Together with Phe174, Tyr99 determines the accessibility to the S2 and S4 sites, and both residues were shown to be modulated not only by the Factor IXa variants but also by its substrate. Comparison of the structures in the present study identified the Ile212–Ile213 motif as a trigger for Tyr99 to serve a regulatory role by limiting the access to either the S2 or the S4 sites. In the more active ITTV213 variant, the Tyr99 enabled access to both the S2 and the S4 sites, thus providing an explanation to how Factor IXa in its active state potentially can accept a threonine residue as found in the P2 position of the Factor X activation sequence.

One other segment that influences the catalytic activity of Factor IXa is the so-called autolysis loop, or 148-loop, that is connected to the substrate-binding site as indicated via the flexible water-bridged Arg143 and His147 and Gln192 (Figure 7). The importance of the 148-loop for activity regulation was reported previously [14,26,55]. In the present study, we suggest a mechanistic link between the 148-loop and Gln192 that is dependent on the conformation in the Ile212–Ile213 motif and the P3 residue. Gln192 frames the entrance to the S1 and S3 pockets, and its effect on substrate recognition has been highlighted in a number of studies. For instance, an E192Q variant has been shown to affect the inhibitory activity of BPTI (bovine pancreatic trypsin inhibitor) towards thrombin and variants of Gln192 in Factor IXa resulted in reduced clotting activity and changed substrate specificity for amidolytic substrates [20,5658]. The more static conformation of Gln192 in complexes with EGR-CMK, compared with its flexible nature in PPACK complexes, could suggest that the extended hydrogen-bonding network surrounding the acidic P3 side chain in the complexes of the K98T and ITTV213 variants lock the conformation of Gln192 and affect the substrate binding and turnover as reflected in the low catalytic turnover of the Boc-IEGR substrate compared with Pefa9 and Pefa10 (Figure 2).

Additional elements that have been identified previously to affect the activity of chymotrypsin-like serine proteases include, among others, the sodium-binding site that is located below the active site and is formed by the 186- and 220-loops. Whether, and under which circumstances, the sodium-binding site is occupied in Factor IXa is still under investigation [59,60]. However, in the high-resolution crystal structures of Factor IXa variants of the present study, the sodium-binding site was found to be unoccupied. The 35- and 60-loops, constituting part of the primed substrate-recognition sites, also cross-talk with the non-primed recognition sites and have been suggested as regulatory elements [61]. We found that the 60-loop possesses a high degree of flexibility as indicated by its high B-factors as well as multiple modelled conformations following an ensemble refinement procedure, and it is possible that ordering of this loop is another, although mechanistically undescribed, aspect of Factor IXa maturation.

In summary, the identified regulatory elements are not acting independently, but rather act in concert along defined communication lines on different scales. For one, there is a direct effect of the K98T variant on the S4 site which is framed by Tyr99 and Phe174 (Figure 9), in accordance with previous findings [5,13,19]. Building on these findings, we now elucidate how the V16I substitution extends this communication network further by wiring the activation pocket with Tyr99. Additionally, it affects the conformations of Gln192 and His147, thereby extending the network into the S3 site. This communication web is prolonged further by the Ile212–Ile213 motif, connecting Ser45 and the S2 and S3 sites over a 25 Å distance.

Communication lines in Factor IXa

Figure 9
Communication lines in Factor IXa

Surface and cartoon representation of the crystal structure of wild-type Factor IXa in complex with pAB (PDB code 1RFN). The P4/P4′ peptide substrate from antithrombin (PDB code 3KCG) is superimposed as a green line to mark the substrate-binding site. Residues subjected to recombinant site-directed mutagenesis in the present study are in orange (Val16, Lys98, Arg150, Tyr177, Ile212 and Ile213). The 99-loop residues (94-97and 100-101) are in light green. Residues constituting the Ile212–Ile213 motif (Ser45, Gly197, Pro198 and Ile212) are in rose. Residues on the 162-helix and the Cys168–Cys182 disulfide bond are in magenta. Arg143, His147 and Gln192 important for the S1/S3 site are in blue. Tyr99 and Phe174 important for the S2 and S4 site conformation are marked in yellow. Bold numbers indicate Factor IXa loop regions described previously to influence catalytic activity. Communication lines identified in the present study are indicated as black arrows.

Figure 9
Communication lines in Factor IXa

Surface and cartoon representation of the crystal structure of wild-type Factor IXa in complex with pAB (PDB code 1RFN). The P4/P4′ peptide substrate from antithrombin (PDB code 3KCG) is superimposed as a green line to mark the substrate-binding site. Residues subjected to recombinant site-directed mutagenesis in the present study are in orange (Val16, Lys98, Arg150, Tyr177, Ile212 and Ile213). The 99-loop residues (94-97and 100-101) are in light green. Residues constituting the Ile212–Ile213 motif (Ser45, Gly197, Pro198 and Ile212) are in rose. Residues on the 162-helix and the Cys168–Cys182 disulfide bond are in magenta. Arg143, His147 and Gln192 important for the S1/S3 site are in blue. Tyr99 and Phe174 important for the S2 and S4 site conformation are marked in yellow. Bold numbers indicate Factor IXa loop regions described previously to influence catalytic activity. Communication lines identified in the present study are indicated as black arrows.

The data presented above showed that numerous regions within the protease domain of Factor IXa are involved in the onset of enzyme maturation and that increases in catalytic activity cannot be linked with a single conformational change. The proposed induced-fit model of substrate binding with either a P2 glycine residue or a P2 proline residue adds further to the complexity of understanding the mechanism behind Factor IXa catalysis. The crystal structures of activity-enhanced variants of the present study represent snapshots towards the fully active state of Factor IXa and the observed flexibility observed in a number of key residues indicates that they are likely to obtain different conformations upon assembly of the Xase complex. Solution-based methods such as NMR to identify chemical shift changes in specific residues upon substrate binding has been reported previously for thrombin [62]. A similar analysis for Factor IXa in combination with molecular dynamics simulation of substrate binding with wild-type Factor IXa and activity-enhanced Factor IXa variants would be highly interesting to obtain a better understanding of residues involved in the proposed induced-fit mechanism upon substrate binding. Ultimately, an array of complementary methods for structural evaluation will be needed to understand the entire molecular basis of the dramatic increase in Factor IXa catalytic activity upon assembly of the Xase complex. The crystal structure analyses of the present study provide mechanistic insights into the enzymatic and biophysical characterization and reveal a delicately balanced bidirectional network that consists of compensatory brakes (Val16 and Ile213) and accelerators (Tyr99 and Phe174) which allow for subtle fine-tuning of enzymatic activity and are likely to keep Factor IXa in a low-activity state in the absence of its cofactor.

PDB codes

Crystal structures from the present study have been deposited in the PDB under the following codes: K98T–EGR-CMK, 5JB8; K98T-PPACK, 5JB9; ITTV212–PPACK, 5JBA; ITTV213–EGR-CMK, 5JBB; ITTV213–PPACK, 5JBC.

AUTHOR CONTRIBUTION

Line Hyltoft Kristensen, Grant Blouse, Ole Olsen and Hans Brandstetter designed the study. Line Hyltoft Kristensen performed the experimental work. Line Hyltoft Kristensen, Ole Olsen and Hans Brandstetter performed data analysis. Line Hyltoft Kristensen, Grant Blouse and Hans Brandstetter wrote the paper. All authors read and approved the final version of the paper.

We thank Elfriede Dall and the beamline scientists at the ESRF beamlines ID23-2 and ID29 for their help with and expert supervision during data collection. We thank Jens Breinholt for assistance with Figure 1(A).

FUNDING

This work was funded by the Novo Nordisk Science Talent Attraction and Recruitment (STAR) PostDoc programme.

Abbreviations

     
  • AMC

    7-amino-4-methylcoumarin

  •  
  • AST

    active site titration

  •  
  • Boc-IEGR

    t-butoxycarbonyl-Ile-Glu-Gly-Arg-AMC

  •  
  • CMK

    chloromethylketone

  •  
  • DEGR-CMK

    1,5-dansyl-Glu-Gly-Arg-chloromethylketone

  •  
  • EGF

    epidermal growth factor

  •  
  • EGR-CMK

    H-Glu-Gly-Arg-chloromethylketone

  •  
  • ESRF

    European Synchrotron Radiation Facility

  •  
  • Factor IXa

    activated Factor IX

  •  
  • IB

    inclusion body

  •  
  • 4-MU

    4-methylumbelliferone

  •  
  • MUGB

    4-methylumbelliferyl 4-guanidinobenzoate hydrochloride monohydrate

  •  
  • NBS

    non-binding surface

  •  
  • pAB

    p-aminobenzamidine

  •  
  • PPACK

    D-Phe-Pro-Arg-chloromethylketone

  •  
  • tPA

    tissue plasminogen activator

  •  
  • TSA

    thermal shift assay

References

References
1
Davie
E.W.
Fujikawa
K.
Kisiel
W.
The coagulation cascade: initiation, maintenance, and regulation
Biochemistry
1991
, vol. 
30
 (pg. 
10363
-
10370
)
[PubMed]
2
Bode
W.
Huber
R.
Induction of the bovine trypsinogen–trypsin transition by peptides sequentially similar to the N-terminus of trypsin
FEBS Lett
1976
, vol. 
68
 (pg. 
231
-
236
)
[PubMed]
3
van Dieijen
G.
Tans
G.
Rosing
J.
Hemker
H.C.
The role of phospholipid and factor VIIIa in the activation of bovine factor X
J. Biol. Chem.
1981
, vol. 
256
 (pg. 
3433
-
3442
)
[PubMed]
4
van Dieijen
G.
van Rijn
J.L.
Govers-Riemslag
J.W.
Hemker
H.C.
Rosing
J.
Assembly of the intrinsic factor X activating complex: interactions between factor IXa, factor VIIIa and phospholipid
Thromb. Haemost.
1985
, vol. 
53
 (pg. 
396
-
400
)
[PubMed]
5
Zögg
T.
Brandstetter
H.
Complex assemblies of factors IX and X regulate the initiation, maintenance, and shutdown of blood coagulation
Prog. Mol. Biol. Transl. Sci.
2011
, vol. 
99
 (pg. 
51
-
103
)
[PubMed]
6
Di Cera
E.
Serine proteases
IUBMB Life
2009
, vol. 
61
 (pg. 
510
-
515
)
[PubMed]
7
Page
M.J.
Macgillivray
R.T.
Di Cera
E.
Determinants of specificity in coagulation proteases
J. Thromb. Haemost.
2005
, vol. 
3
 (pg. 
2401
-
2408
)
[PubMed]
8
Johnson
D.J.
Langdown
J.
Huntington
J.A.
Molecular basis of factor IXa recognition by heparin-activated antithrombin revealed by a 1.7-Å structure of the ternary complex
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
645
-
650
)
[PubMed]
9
Venkateswarlu
D.
Structural insights into the interaction of blood coagulation co-factor VIIIa with factor IXa: a computational protein–protein docking and molecular dynamics refinement study
Biochem. Biophys. Res. Commun.
2014
, vol. 
452
 (pg. 
408
-
414
)
[PubMed]
10
Hopfner
K.P.
Brandstetter
H.
Karcher
A.
Kopetzki
E.
Huber
R.
Engh
R.A.
Bode
W.
Converting blood coagulation factor IXa into factor Xa: dramatic increase in amidolytic activity identifies important active site determinants
EMBO J.
1997
, vol. 
16
 (pg. 
6626
-
6635
)
[PubMed]
11
Milanov
P.
Ivanciu
L.
Abriss
D.
Quade-Lyssy
P.
Miesbach
W.
Alesci
S.
Tonn
T.
Grez
M.
Seifried
E.
Schüttrumpf
J.
Engineered factor IX variants bypass FVIII and correct hemophilia A phenotype in mice
Blood
2012
, vol. 
119
 (pg. 
602
-
611
)
[PubMed]
12
Sichler
K.
Kopetzki
E.
Huber
R.
Bode
W.
Hopfner
K.P.
Brandstetter
H.
Physiological fIXa activation involves a cooperative conformational rearrangement of the 99-loop
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
4121
-
4126
)
[PubMed]
13
Kolkman
J.A.
Mertens
K.
Insertion loop 256–268 in coagulation factor IX restricts enzymatic activity in the absence but not in the presence of factor VIII
Biochemistry
2000
, vol. 
39
 (pg. 
7398
-
7405
)
[PubMed]
14
Yang
L.
Manithody
C.
Olson
S.T.
Rezaie
A.R.
Contribution of basic residues of the autolysis loop to the substrate and inhibitor specificity of factor IXa
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
25032
-
25038
)
[PubMed]
15
Quade-Lyssy
P.
Abriss
D.
Milanov
P.
Ungerer
C.
Königs
C.
Seifried
E.
Schüttrumpf
J.
Next generation FIX muteins with FVIII-independent activity for alternative treatment of hemophilia A
J. Thromb. Haemost.
2014
, vol. 
12
 (pg. 
1861
-
1873
)
[PubMed]
16
Hopfner
K.P.
Lang
A.
Karcher
A.
Sichler
K.
Kopetzki
E.
Brandstetter
H.
Huber
R.
Bode
W.
Engh
R.A.
Coagulation factor IXa: the relaxed conformation of Tyr99 blocks substrate binding
Structure
1999
, vol. 
7
 (pg. 
989
-
996
)
[PubMed]
17
Parker
D.L.
Jr
Walsh
S.
Li
B.
Kim
E.
Sharipour
A.
Smith
C.
Chen
Y.H.
Berger
R.
Harper
B.
Zhang
T.
, et al. 
Rapid development of two factor IXa inhibitors from hit to lead
Bioorg. Med. Chem. Lett.
2015
, vol. 
25
 (pg. 
2321
-
2325
)
[PubMed]
18
Zhang
T.
Andre
P.
Bateman
T.J.
Chen
Y.H.
Desai
K.
Ellsworth
K.
Geissler
W.M.
Guo
L.
Hruza
A.
Jian
T.
, et al. 
Development of a novel class of potent and selective FIXa inhibitors
Bioorg. Med. Chem. Lett.
2015
, vol. 
25
 (pg. 
4945
-
4949
)
[PubMed]
19
Zögg
T.
Brandstetter
H.
Structural basis of the cofactor- and substrate-assisted activation of human coagulation factor IXa
Structure
2009
, vol. 
17
 (pg. 
1669
-
1678
)
[PubMed]
20
van de Locht
A.
Bode
W.
Huber
R.
Le Bonniec
B.F.
Stone
S.R.
Esmon
C.T.
Stubbs
M.T.
, et al. 
The thrombin E192Q-BPTI complex reveals gross structural rearrangements: implications for the interaction with antithrombin and thrombomodulin
EMBO J.
1997
, vol. 
16
 (pg. 
2977
-
2984
)
[PubMed]
21
Page
M.J.
Carrell
C.J.
Di Cera
E.
Engineering protein allostery: 1.05 Å resolution structure and enzymatic properties of a Na+-activated trypsin
J. Mol. Biol.
2008
, vol. 
378
 (pg. 
666
-
672
)
[PubMed]
22
Page
M.J.
Di Cera
E.
Role of Na+ and K+ in enzyme function
Physiol. Rev.
2006
, vol. 
86
 (pg. 
1049
-
1092
)
[PubMed]
23
Papaconstantinou
M.E.
Gandhi
P.S.
Chen
Z.
Bah
A.
Di Cera
E.
Na+ binding to meizothrombin desF1
Cell. Mol. Life Sci.
2008
, vol. 
65
 (pg. 
3688
-
3697
)
[PubMed]
24
Simioni
P.
Tormene
D.
Tognin
G.
Gavasso
S.
Bulato
C.
Iacobelli
N.P.
Finn
J.D.
Spiezia
L.
Radu
C.
Arruda
V.R.
X-linked thrombophilia with a mutant factor IX (factor IX Padua)
N. Engl. J. Med.
2009
, vol. 
361
 (pg. 
1671
-
1675
)
[PubMed]
25
Sturzebecher
J.
Kopetzki
E.
Bode
W.
Hopfner
K.P.
Dramatic enhancement of the catalytic activity of coagulation factor IXa by alcohols
FEBS Lett
1997
, vol. 
412
 (pg. 
295
-
300
)
[PubMed]
26
Pozzi
N.
Chen
R.
Chen
Z.
Bah
A.
Di Cera
E.
Rigidification of the autolysis loop enhances Na(+) binding to thrombin
Biophys. Chem.
2011
, vol. 
159
 (pg. 
6
-
13
)
[PubMed]
27
Rudolph
R.
Lilie
H.
In vitro folding of inclusion body proteins
FASEB J.
1996
, vol. 
10
 (pg. 
49
-
56
)
[PubMed]
28
Payne
M.A.
Neuenschwander
P.F.
Johnson
A.E.
Morrissey
J.H.
Effect of soluble tissue factor on the kinetic mechanism of factor VIIa: enhancement of p-guanidinobenzoate substrate hydrolysis
Biochemistry
1996
, vol. 
35
 (pg. 
7100
-
7106
)
[PubMed]
29
Niesen
F.H.
Berglund
H.
Vedadi
M.
The use of differential scanning fluorimetry to detect ligand interactions that promote protein stability
Nat. Protoc.
2007
, vol. 
2
 (pg. 
2212
-
2221
)
[PubMed]
30
Flot
D.
Mairs
T.
Giraud
T.
Guijarro
M.
Lesourd
M.
Rey
V.
van Brussel
D.
Morawe
C.
Borel
C.
Hignette
O.
, et al. 
The ID23-ID22 structural biology microfocus beamline at the ESRF
J. Synchrotron Radiat.
2010
, vol. 
17
 (pg. 
107
-
118
)
[PubMed]
31
de Sanctis
D.
Beteva
A.
Caserotto
H.
Dobias
F.
Gabadinho
J.
Giraud
T.
Gobbo
A.
Guijarro
M.
Lentini
M.
Lavault
B.
, et al. 
ID29: a high-intensity highly automated ESRF beamline for macromolecular crystallography experiments exploiting anomalous scattering
J. Synchrotron Radiat.
2012
, vol. 
19
 (pg. 
455
-
461
)
[PubMed]
32
Battye
T.G.
Kontogiannis
L.
Johnson
O.
Powell
H.R.
Leslie
A.G.
iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
271
-
281
)
[PubMed]
33
Delagenière
S.
Brenchereau
P.
Launer
L.
Ashton
A.W.
Leal
R.
Veyrier
S.
Gabadinho
J.
Gordon
E.J.
Jones
S.D.
Levik
K.E.
, et al. 
ISPyB: an information management system for synchrotron macromolecular crystallography
Bioinformatics
2011
, vol. 
27
 (pg. 
3186
-
3192
)
[PubMed]
34
Monaco
S.
Gordon
E.
Bowler
M.W.
Delagenière
S.
Guijarro
M.
Spruce
D.
Svensson
O.
McSweeney
S.M.
McCarthy
A.A.
Leonard
G.
Nanao
M.H.
Automatic processing of macromolecular crystallography X-ray diffraction data at the ESRF
J. Appl. Crystallogr.
2013
, vol. 
46
 (pg. 
804
-
810
)
[PubMed]
35
Winn
M.D.
Ballard
C.C.
Cowtan
K.D.
Dodson
E.J.
Emsley
P.
Evans
P.R.
Keegan
R.M.
Krissinel
E.B.
Leslie
A.G.
McCoy
A.
, et al. 
Overview of the CCP4 suite and current developments
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
235
-
242
)
[PubMed]
36
Vagin
A.
Teplyakov
A.
Molecular replacement with MOLREP
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
22
-
25
)
[PubMed]
37
Murshudov
G.N.
Skubak
P.
Lebedev
A.A.
Pannu
N.S.
Steiner
R.A.
Nicholls
R.A.
Winn
M.D.
Long
F.
Vagin
A.A.
REFMAC5 for the refinement of macromolecular crystal structures
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
355
-
367
)
[PubMed]
38
Adams
P.D.
Afonine
P.V.
Bunkóczi
G.
Chen
V.B.
Davis
I.W.
Echols
N.
Headd
J.J.
Hung
L.W.
Kapral
G.J.
Grosse-Kunstleve
R.W.
PHENIX: a comprehensive Python-based system for macromolecular structure solution
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
66
 (pg. 
213
-
221
)
39
Emsley
P.
Lohkamp
B.
Scott
W.G.
Cowtan
K.
Features and development of Coot
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
486
-
501
)
[PubMed]
40
Davis
I.W.
Leaver-Fay
A.
Chen
V.B.
Block
J.N.
Kapral
G.J.
Wang
X.
Murray
L.W.
Arendall
W.B.
3rd
Snoeyink
J.
Richardson
J.S.
Richardson
D.C.
MolProbity: all-atom contacts and structure validation for proteins and nucleic acids
Nucleic Acids Res
2007
, vol. 
35
 (pg. 
W375
-
W383
)
[PubMed]
41
Rawlings
N.D.
Barrett
A.J.
Finn
R.
Twenty years of the MEROPS database of proteolytic enzymes, their substrates and inhibitors
Nucleic Acids Res
2016
, vol. 
44
 (pg. 
D343
-
D50
)
[PubMed]
42
Burnley
B.T.
Afonine
P.V.
Adams
P.D.
Gros
P.
Modelling dynamics in protein crystal structures by ensemble refinement
eLife
2012
, vol. 
1
 pg. 
e00311
 
[PubMed]
43
Burnley
B.T.
Gros
P.
phenix.ensemble_refinement: a test study of apo and holo BACE1
Comput. Crystallogr. Newsl.
2013
, vol. 
4
 (pg. 
51
-
58
)
44
Forneris
F.
Burnley
B.T.
Gros
P.
Ensemble refinement shows conformational flexibility in crystal structures of human complement factor D
Acta Crystallogr. D Biol. Crystallogr.
2014
, vol. 
70
 (pg. 
733
-
743
)
[PubMed]
45
Levin
E.J.
Kondrashov
D.A.
Wesenberg
G.E.
Phillips
G.N.
Ensemble refinement of protein crystal structures: validation and application
Structure
2007
, vol. 
15
 (pg. 
1040
-
1052
)
[PubMed]
46
Kirchhofer
D.
Lipari
M.T.
Santell
L.
Billeci
K.L.
Maun
H.R.
Sandoval
W.N.
Moran
P.
Ridgway
J.
Eigenbrot
C.
Lazarus
R.A.
Utilizing the activation mechanism of serine proteases to engineer hepatocyte growth factor into a Met antagonist
Proc. Natl. Acad. Sci. U.S.A.
2007
, vol. 
104
 (pg. 
5306
-
5311
)
[PubMed]
47
Persson
E.
Kjalke
M.
Olsen
O.H.
Rational design of coagulation factor VIIa variants with substantially increased intrinsic activity
Proc. Natl. Acad. Sci. U.S.A.
2001
, vol. 
98
 (pg. 
13583
-
13588
)
[PubMed]
48
Reyda
S.
Sohn
C.
Klebe
G.
Rall
K.
Ullmann
D.
Jakubke
H.D.
Stubbs
M.T.
Reconstructing the binding site of factor Xa in trypsin reveals ligand-induced structural plasticity
J. Mol. Biol.
2003
, vol. 
325
 (pg. 
963
-
977
)
[PubMed]
49
Huntington
J.A.
Esmon
C.T.
The molecular basis of thrombin allostery revealed by a 1.8 Å structure of the “slow” form
Structure
2003
, vol. 
11
 (pg. 
469
-
479
)
[PubMed]
50
Rallapalli
P.M.
Kemball-Cook
G.
Tuddenham
E.G.
Gomez
K.
Perkins
S.J.
An interactive mutation database for human coagulation factor IX provides novel insights into the phenotypes and genetics of hemophilia B
J. Thromb. Haemost.
2013
, vol. 
11
 (pg. 
1329
-
1340
)
[PubMed]
51
Kolkman
J.A.
Lenting
P.J.
Mertens
K.
Regions 301–303 and 333–339 in the catalytic domain of blood coagulation factor IX are factor VIII-interactive sites involved in stimulation of enzyme activity
Biochem. J.
1999
, vol. 
339
 (pg. 
217
-
221
)
[PubMed]
52
Brandstetter
H.
Kühne
A.
Bode
W.
Huber
R.
von der Saal
W.
Wirthensohn
K.
Engh
R.A.
X-ray structure of active site-inhibited clotting factor Xa: implications for drug design and substrate recognition
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
29988
-
29992
)
[PubMed]
53
Lamba
D.
Bauer
M.
Huber
R.
Fischer
S.
Rudolph
R.
Kohnert
U.
Bode
W.
The 2.3 Å crystal structure of the catalytic domain of recombinant two-chain human tissue-type plasminogen activator
J. Mol. Biol.
1996
, vol. 
258
 (pg. 
117
-
135
)
[PubMed]
54
Salonen
L.M.
Holland
M.C.
Kaib
P.S.
Haap
W.
Benz
J.
Mary
J.L.
Kuster
O.
Schweizer
W.B.
Banner
D.W.
Diederich
F.
Molecular recognition at the active site of factor Xa: cation–π interactions, stacking on planar peptide surfaces, and replacement of structural water
Chemistry
2012
, vol. 
18
 (pg. 
213
-
222
)
[PubMed]
55
Kolkman
J.A.
Mertens
K.
Surface-loop residue Lys316 in blood coagulation Factor IX is a major determinant for Factor X but not antithrombin recognition
Biochem. J.
2000
, vol. 
350
 (pg. 
701
-
707
)
[PubMed]
56
Hsu
Y.C.
Hamaguchi
N.
Chang
Y.J.
Lin
S.W.
The distinct roles that Gln-192 and Glu-217 of factor IX play in selectivity for macromolecular substrates and inhibitors
Biochemistry
2001
, vol. 
40
 (pg. 
11261
-
11269
)
[PubMed]
57
Rezaie
A.R.
Esmon
C.T.
Molecular basis of residue 192 participation in determination of coagulation protease specificity
Eur. J. Biochem.
1996
, vol. 
242
 (pg. 
477
-
484
)
[PubMed]
58
Rohlena
J.
Kolkman
J.A.
Boertjes
R.C.
Mertens
K.
Lenting
P.J.
Residues Phe342–Asn346 of activated coagulation factor IX contribute to the interaction with low density lipoprotein receptor-related protein
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
9394
-
9401
)
[PubMed]
59
Gopalakrishna
K.
Rezaie
A.R.
The influence of sodium ion binding on factor IXa activity
Thromb. Haemost.
2006
, vol. 
95
 (pg. 
936
-
941
)
[PubMed]
60
Schmidt
A.E.
Stewart
J.E.
Mathur
A.
Krishnaswamy
S.
Bajaj
S.P.
Na+ site in blood coagulation factor IXa: effect on catalysis and factor VIIIa binding
J. Mol. Biol.
2005
, vol. 
350
 (pg. 
78
-
91
)
[PubMed]
61
Zögg
T.
Brandstetter
H.
Activation mechanisms of coagulation factor IX
Biol. Chem.
2009
, vol. 
390
 (pg. 
391
-
400
)
[PubMed]
62
Lechtenberg
B.C.
Johnson
D.J.D.
Freund
S.M.V.
Huntington
J.A.
NMR resonance assignments of thrombin reveal the conformational and dynamic effects of ligation
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
14087
-
14092
)
[PubMed]

Author notes

1

Line Hyltoft Kristensen is a former employee of Novo Nordisk A/S and current employee at Symphogen A/S

2

Ole H. Olsen is a former employee and shareholder at Novo Nordisk A/S

3

Grant E. Blouse is an employee and shareholder at Novo Nordisk A/S

Supplementary data