A Kunitz type dual inhibitor (TKI) of factor Xa (FXa) and trypsin was found in tamarind. It also shows prolongation of blood coagulation time. The deduced 185 amino acid sequence of TKI by cDNA cloning and sequence analysis revealed that it belongs to the Kunitz type soybean trypsin inhibitor (STI) family; however, it has a distorted Kunitz signature sequence due to insertion of Asn15 in the motif. TKI exhibited a competitive inhibitory activity against both FXa (Ki = 220 nm) and porcine pancreatic trypsin (Ki = 3.2 nm). The crystal structure of TKI shows a β-trefoil fold similar to Kunitz STI inhibitors; however, a distinct mobile reactive site, an inserted residue and loop β7β8 make it distinct from classical Kunitz inhibitors. The crystal structure of TKI-trypsin and a 3D model of TKI-FXa complex revealed that the distinct reactive site loop probably plays a role in dual inhibition. The reactive site of TKI interacts with an active site and two exosites (36 loop and autolysis loop) of FXa. Apart from Arg66 (P1), Arg64 (P3) is one of the most important residues responsible for the specificity of TKI towards FXa. Along with the reactive site loop (β4β5), loops β1 and β7β8 also interact with FXa and could further confer selectivity for FXa. We also present the role of inserted Asn15 in the stabilization of complexes. To the best of our knowledge, this is the first structure of FXa inhibitor belonging to the Kunitz type inhibitor family and its unique structural and sequence features make TKI a novel potent inhibitor.
The complete nucleotide of TKI was deposited in the NCBI gene databank with accession no. HQ385502. The atomic coordinates and structure factor files for the structure of TKI and TKI:PPT complex have been deposited in the Protein Data Bank with accession numbers 4AN6 and 4AN7, respectively
Structured digital abstract
arrowhead protease inhibitor A
activated partial thromboplastin time
Bauhinia bauhinioides cruzipain inhibitor
Bauhinia bauhinioides kallikrein inhibitor
Bauhinia ungulata factor Xa inhibitor
- BvTI and BvvTI
Bauhinia variegata trypsin inhibitor
Copaifera langsdorff1i trypsin inhibitor
Delonix regia trypsin inhibitor
Enterolobium contortisiliquum trypsin inhibitor
Erythriana caffra trypsin inhibitor
Murraya koenigii miraculin-like protein
porcine pancreatic elastase
porcine pancreatic trypsin
soybean trypsin inhibitor
tick anticoagulant peptide
tissue factor pathway inhibitor
tamarind Kunitz type inhibitor
winged bean chymotrypsin inhibitor
The formation and interactions of specific protein–protein complexes have enormous importance in biological processes [1, 2]. Protein–protein complexes of serine proteases and their inhibitors are extensively studied models . The Kunitz soybean trypsin inhibitor (STI) superfamily of serine proteases has been widely studied. The plant Kunitz type inhibitors are present in comparatively large quantities in the seeds of the Leguminosae subfamilies, Mimosoideae, Caesalpinioideae and Papilionoideae [3, 4]. These inhibitors have molecular mass of about 20 kDa with one or two polypeptide chains and few cysteines [5-7]. Most of them are a single polypeptide chain with two disulfide bridges and a single reactive site [4, 8, 9]. The Kunitz STIs are included as Kunitz-P inhibitors in the I3A family of the MEROPS database (http://merops.sanger.ac.uk) .
Several structures of Kunitz STI type inhibitors have been reported showing that they have a β-trefoil fold consisting of 12 anti-parallel β-strands connected by long loops. Their exposed reactive site loop has a characteristic canonical conformation with Arg/Lys at the P1 position [6, 11-15]. The conserved Asn (Asn13 in STI) residue on loop β1 interacts with P2 and P1′ of the reactive site loop (loop β4β5) and maintains the canonical conformation of the reactive site loop [11, 12, 16]. The structure of the STI complex with porcine pancreatic trypsin (PPT) reveals that P1 residue Arg63 enters into the S1 pocket of trypsin and makes a salt-bridge with Asp189 of the enzyme and inhibits it in a substrate-like manner . Moreover, some structural reports showed a variation in Kunitz type inhibitor structures. For example, Copaifera langsdorff1i trypsin inhibitor (CTI) has a β-trefoil fold composed of two non-covalently bound polypeptide chains with only a single disulfide bridge . Delonix regia trypsin inhibitor (DrTI) has one amino acid insertion between P1 and P2 of the reactive site distorting its conformation . Bauhinia bauhinioides cruzipain inhibitor (BbCI), a Kunitz type inhibitor, has a conservative β-trefoil fold but lacks disulfide bonds . The crystal structure of Murraya koenigii miraculin-like protein (MKMLP) from seeds of Murraya koenigii with trypsin inhibitory activity also shows a conservative β-trefoil fold but has seven cysteines forming three disulfide bridges, Asn65 at P1 instead of Arg/Lys, Asn13 replaced by Ala13 and carbohydrate moieties linked to Asn64 . Arrowhead protease inhibitor A (API-A), a double-headed arrowhead protease inhibitor similar to Kunitz STI inhibitor, has a β-trefoil fold but possesses two reactive sites .
The plant Kunitz type inhibitors are not restricted to trypsin inhibition only, but also show inhibition of several serine and other proteases. For example, beyond the trypsin inhibitory activity of STI, it acts as an anti-inflammatory by inhibiting human neutrophil elastase  and has an anti-invasive property on ovarian cancer cells . Enterolobium contortisiliquum trypsin inhibitor (EcTI) inhibits trypsin, chymotrypsin, factor XIIa, plasma kallikrein, human neutrophil elastase, plasmin and the invasion of gastric cancer cells [23-25]. Bauhinia ungulata factor Xa inhibitor (BuXI) blocks the activity of trypsin, chymotrypsin, plasma kallikrein, plasmin, factor XIIa and factor Xa (FXa) . Bauhinia bauhinioides kallikrein inhibitor (BbKI) inhibits trypsin, chymotrypsin, plasma kallikrein and plasmin [27, 28]. BbCI shows inhibitory properties against enzymes of two different classes, cysteine and serine proteases [28, 29]. Bauhinia variegata trypsin inhibitor (BvvTI) has trypsin inhibitory, anti-HIV-1-RT activity and inhibits the proliferation of nasopharyngeal cancer CNE-1 cells . These several plant Kunitz type inhibitors are important molecules to gain knowledge of the basic principle of protein–protein interactions and might have applications in pharmaceuticals as potential drugs. So, it is necessary to search for novel multivalent Kunitz inhibitors having pharmaceutical significance.
We have previously reported purification of tamarind Kunitz STI type inhibitor (TKI) and its preliminary crystallographic study as well its complex formation with trypsin and the crystallization of the complex [31, 32]. TKI was also reported as an in vitro and in vivo potential bio-insecticide against different insect pests and a weak elastase inhibitor and was shown to be a non-competitive inhibitor . However, in the present study, we show that TKI inhibits FXa and trypsin competitively. The inhibition constants (Ki) of TKI for FXa and trypsin were calculated. TKI was also assayed for its anticoagulant property. To gain structural insights into a dual inhibitory role of this novel TKI, cDNA cloning and the detailed 3D crystal structures of free TKI and its complex with trypsin were determined and described. Moreover, the complex with FXa has been modelled to study the mode of inhibition of FXa by TKI. Furthermore, we describe the role of inserted residue Asn15 in the stabilization of the complexes.
cDNA cloning of TKI and sequence analysis
The first strand of cDNA was obtained from highly pure RNA. It was amplified using forward degenerate and oligo(dT)18 primers. An approximately 700 bp fragment was obtained, which included the 3′ UTR region with poly(A+) tail. A 555 bp ORF was obtained after sequencing of the TKI gene, which coded for a polypeptide of 185 amino acids with a calculated molecular mass of 20 575 Da. The sequence of TKI showed significant similarity to reported Kunitz type inhibitors in an NCBI blast search, which demonstrates that TKI belongs to the soybean Kunitz family inhibitors. The primary amino acid sequence of TKI showed that it has a distorted Kunitz signature motif ([LIVM]-x-D-x-[EDNTY]-[DG]-[RKHDENQ]-x-[LIVM]-(x)5-Y-x-[LIVM]). The Kunitz signature motif is found in the soybean Kunitz type trypsin inhibitor family and in TKI the observed pattern is VHDTDGKPVLNNAGQYYI which is located at the N-terminal of the TKI sequence (Fig. 1). The sequence analysis revealed that the distorted Kunitz signature sequence was due to insertion of Asn at position 15. The insertion of Asn is also observed in Bauhinia variegata trypsin inhibitor (BvTI) (Fig. 1). A protein sequence blast with non-redundant database showed that TKI shares 45%, 44%, 42%, 39% and 33% sequence identity with BuXI , EcTI , BvTI , Erythriana caffra trypsin inhibitor (ETI)  and STI  respectively. The chain length of these proteins ranges from 172 to 185 residues. The four cysteines are linked by two disulfide bonds in TKI as in ETI and STI (Cys 42–86, Cys 134–144). Multiple sequence alignment of the representative Kunitz type inhibitors from different plants revealed that the TKI sequence forms a close relationship with classical Kunitz family members and the closest relationship with BuXI as shown in Fig. 1.
Inhibitory properties and Ki determination
The inhibitory activity of TKI was determined against trypsin, chymotrypsin, elastase and FXa by measuring the hydrolytic activity toward N-benzoyl-l-arginine-p-nitroanilide (BAPNA), N-benzoyl-l-tyrosyl-p-nitroanilide (BTPNA), N-Suc-(Ala)3-nitroanilide and CH3OCO-D-CHA-Gly-Arg-pNA-AcOH, respectively. TKI showed inhibition of trypsin and FXa but did not show any significant inhibitory activity against chymotrypsin and showed very weak inhibition of elastase (data not shown). The dissociation constants (Ki) for TKI against trypsin (Fig. 2B) and FXa (Fig. 2C) were determined from a Dixon plot. Analysis of the Dixon plot revealed that TKI is a competitive inhibitor with Ki value 3.2 × 10−9 m and 2.2 × 10−7 m for trypsin and FXa activity, respectively, which clearly explains that TKI is a potent inhibitor of porcine trypsin and human FXa. The inhibition of human FXa by TKI indicated that TKI could inhibit blood coagulation. Therefore the anticoagulation property of purified TKI was tested. The incubation of increasing concentrations of TKI with fresh human plasma showed significant increase in activated partial thromboplastin time (APTT) and prothrombin time (PT). TKI (10 μm) extended the APTT of normal plasma 2.6-fold and of PT 5-fold (Fig. 2A).
Quality and overall structure of TKI
The crystallographic data and refinement statistics for the free TKI structure are summarized in Table 1. The TKI structure was determined at 1.94 Å resolution and contains two molecules per asymmetric unit (TKI A and TKI B). The free TKI structure was refined to an Rcryst of 19.04% and Rfree of 24.02%. The overall model is well defined by electron density; however, the electron densities of residues 1, 153–155 in chain A, residues 1, 138–141 in chain B and C-terminal residues (177–185) from both the chains were not observed. A section (residues 122–130) of the final 2F0 − Fc electron density map is shown in Fig. 3C. The final model consists of 343 residues and 157 solvent molecules in both the molecules. The Ramachandran plot analysis was done using molprobity server . The refined model shows that 96.1% of all residues are in favoured regions and 99.7% of all residues are in the allowed regions (Table 1).
|Free TKI||TKI–PPT complex|
|Unique reflections||19 581||17 821|
|Completeness (%) (last shell)||97.5 (75.7)||92.9 (40.3)|
|Rsym (%) (last shell)||4.3 (25.7)||7.6 (40.0)|
|I/σ (last shell)||19.4 (3.6)||11.86 (2.0)|
|Multiplicity (last shell)||3.0 (2.4)||3.3 (2.0)|
|No. of reflections (working/test)||17 690 (16726/964)||16 883 (15972/911)|
|No. of residues||343||379|
|Resolution range (Å)||1.94||2.23|
|Average B factors (Å2)||A 21.95||A 26.71|
|B 23.09||B 32.78|
|Rmsd on bond lengths (Å)||0.008||0.009|
|Rmsd on bond angles (Å)||1.43||1.36|
|Ramachandran plot (%)|
The overall structure of TKI is similar to that of other Kunitz type inhibitors. It is devoid of helical structures and is composed of β-strands and turns. The protein shows a β-trefoil fold consisting of 12 anti-parallel β-strands connected with long loops similar to STI, an archeal member of the Kunitz type trypsin inhibitor family [36, 37]. These β-strands form six double stranded β-hairpins. Three β-hairpins form the barrel structure and the remaining form a cap-like structure on the barrel (Fig. 3A). TKI consists of two disulfide bonds, Cys42-Cys86 and Cys134-Cys144, at the surface that are also conserved in other Kunitz type inhibitors stabilizing the 3D structure [11-13]. The TKI structure does not have a 310-helix similar to STI due to deletion of three amino acids at the respective positions. This type of deletion is also observed in other Kunitz type inhibitors like ETI, winged bean chymotrypsin inhibitor (WCI) and DrTI [6, 12, 13].
The geometry and conformation of the reactive site loop
The TKI reactive site loop P4–P3′ (Ser63-Arg64-Ala65-Arg66-Ile67-Ser68-His69) is devoid of secondary structures and disulfide bonds. The Arg66 present at P1, which is identical to its closely related ETI and STI, is defined well in the electron density map. The reactive site loop of TKI has an overall high B factor for P1–P4 residues. The side chains of residues at P1, P3 and P3′ are flexible and may fit optimally into the active site of cognate proteases by adapting the conformation. The Asn14 present on loop β1, which is a conserved residue in the STI family (Asn13 in STI), stabilizes the reactive site loop (loop β4β5) in a similar manner as was shown in the other member of the STI family [11, 16]. The ND2 atom of Asn14 forms hydrogen bonds with the P2 and P1′ carbonyl O which are present at both sides of the scissile bond. Asn14 also interacts with P4 and P2′ and maintains the canonical conformation of the reactive site loop, which is a characteristic feature of this family. This conserved Asn is called a spacer residue and is also conserved in the other serine protease inhibitor families such as Kazal, ecotin, Streptomyces subtilisin inhibitor, grasshopper and potato II .
Comparison of free TKI with homologous structures
The structural studies demonstrated that TKI adopted an overall β-trefoil fold with two disulfide bonds similar to the Kunitz inhibitors. Although the overall fold of TKI showed structural similarity with Kunitz STIs, the two molecules in the asymmetric unit of free TKI show variations in loop regions (Fig. 3B). Two of the connecting loops (L6 residues 95–108 and L7 residues 112–119) of TKI show a high degree of deviation among themselves. Apart from connecting loops, the two β sheets B1 (129–137) and B2 (140–150) exhibit different orientations between TKI A and TKI B.
ETI , STI , CTI , WCI  and DrTI  share closest structural similarity with TKI. TKI exhibits 39%, 33% and 31% sequence identity with ETI, STI and WCI respectively. The superposition of TKI with ETI, STI and WCI shows rmsd values of 1.16 Å (for 117 Cα atoms), 0.95 Å (for 113 Cα atoms) and 1.5 Å (for 123 Cα atoms), respectively. The superposition of Cα atoms of the reactive site loop residues of TKI (P4–P3′) with ETI, STI and WCI exhibit rmsd values of 0.8 Å, 0.7 Å and 0.7 Å respectively. The superposition of the reactive site loop residues of TKI A with TKI B shows different conformations with rmsd of 0.8 Å.
CTI, an inhibitor of the Kunitz family different from classical inhibitors , shares 41% sequence identity with TKI and superposition of TKI with CTI gives an rmsd of 0.55 Å for 117 Cα atoms. DrTI is a Kunitz family member with a distorted reactive site loop  and shows 36% sequence identity with TKI and an rmsd of 1.02 Å for 107 Cα atoms.
Overall structure of TKI–PPT complex
The structure of TKI–PPT complex was determined at 2.23 Å resolution and refined to an Rcryst of 19.95% and Rfree of 24.58%. The refined model of the complex consists of one molecule per asymmetric unit. The TKI–PPT model lacks three segments (residues 26–28, 96–104 and 136–141), one N-terminal residue and 10 residues of the C-terminal of TKI in the electron density. The complex structure consists of 379 residues, 156 solvent molecules and a Ca2+ atom. The refined model shows that 95.6% of all residues are in favoured regions and 99.7% of all residues are in the allowed regions of the Ramachandran plot (Table 1).
The overall structure of TKI–PPT complex is similar to that of STI–PPT. TKI blocks the active site of PPT and P1 residue forms a hydrogen bond network with S1 pocket residues. After binding to PPT, a conformational change at the P1 (Arg66) position and also reduction in the temperature factor of the reactive site loop were observed. One Ca2+ atom is observed in the complex model which interacts with OE1 and OE2 of Glu70, carbonyl O of Asn72 and Val75 and OE2 of Glu80 residues of PPT. This Ca2+ atom is also observed at a similar position in the STI–TKI model. The overall complex structure and detailed interaction of TKI with PPT is shown in Fig. 4. The 2F0 − Fc electron density map clearly shows that the reactive site loop of TKI enters into the S1 pocket and interacts with Asp189 of PPT (Fig. 5).
Comparison of free TKI and TKI–PPT complex structure
The comparison between the free TKI and TKI–PPT complex shows slight conformational changes in the reactive site loop. The superposition of the free TKI A and TKI B with the TKI–PPT gave an rmsd of 0.4 Å (for 123 Cα atoms) and 0.5 Å (for 131 Cα atoms) respectively and its reactive site loop showed decreased B factors compared with both TKI A and TKI B. The superimposition of TKI–PPT complex with TKI A and TKI B is shown in Fig. 6. The reactive site loop of TKI A and TKI B is superimposed onto the TKI–PPT complex showing rmsd of 0.59 Å and 0.6 Å respectively. The flexible P1 residue (Arg66) of free TKI becomes more rigid with lower B factor in the TKI–PPT complex structure. This illustrates that P1 stabilizes the complex by making interactions with residues of the S1 pocket and other residues of PPT. The superposition of the reactive site loop of TKI–PPT complex with TKI A and TKI B discloses that there is immense divergence at position P1. The distance between Cα of P1 in TKI–PPT and TKI A and TKI B is 1.4 Å and 0.8 Å respectively. The difference is also examined at the side chain of P1. The side chain of Arg66 of TKI in the complex structure is ∼ 5 Å and 4 Å apart from the side chain of Arg66 of TKI A and TKI B respectively (Fig. 6A). Another residue (Arg64), P3 of the reactive site, also shows conformational variation. P3 becomes rigid after forming a complex with trypsin and participates in stabilization of the complex by interacting with residues of PPT (Fig. 6B). Compared with free TKI, His69 of the reactive site loop (P3′) of TKI–PPT shows conformational rotation. It moves away from Ser68 to prevent short contact and acquires a position to interact with Ser39 and His40 of PPT through a water molecule (W132), which indicates that it may help in stabilization of the complex (Fig. 6C).
Comparison of TKI–PPT with other Kunitz type inhibitor complexes with trypsin
The superposition of TKI–PPT on STI–PPT gives an rmsd of 0.82 Å for 311 Cα atoms and their reactive site loop gives an rmsd of 0.46 Å. This is smaller than the rmsd between the reactive site loop of free STI and TKI models. The interface between TKI and trypsin covers an area of about 1025 Å2 of TKI which is larger than the corresponding surface of STI (937 Å2) in the STI–PPT complex as calculated by the program pdbsum . There are notable differences in the interactions of the interacting residues of TKI with PPT compared with the STI–PPT structure. The interactions of Asn15, Arg64 (P3), Ser68 (P2′), His69 (P3′) and Ser116 of TKI with PPT make it different from the STI–PPT complex. The details of the interaction of TKI–PPT are discussed in the next section.
TKI also shares 31% sequence identity with Kunitz type WCI . The complex structures of WCI mutant F64Y/L65R (PDB ID 3I29) and L68R (PDB ID 2QYI) with bovine trypsin superimposed on the TKI–PPT structure shows an rmsd of 1.5 Å (for 323 Cα atoms) and 1.8 Å (for 335 Cα atoms). The interactions of residue Arg at P1 are similar. Double-headed API-A has less sequence identity (28%) with TKI and its complex with two trypsins (PDB ID 3E8L) superimposed with TKI–PPT shows an rmsd of 1.1 Å for 230 Cα atoms .
Mode of interaction between TKI and PPT
In the TKI–PPT structure, six residues of TKI (Thr3, Arg64 (P3), Arg66 (P1), Ser68 (P2′), Ser75 and Ser116) interact with PPT residues directly and four residues of TKI [Asn15, Gly17, Arg60 and His69 (P3′)] interact through water with residues of PPT (Fig. 4B). A total of 10 residues of TKI interact with 20 residues of PPT forming a dense network of hydrogen bonds to produce a stable complex. The detailed interactions between TKI and PPT are shown in Fig. 4 and Table 2. There are 21 hydrogen bonds between TKI and PPT involving four residues of the reactive site. The P1 residue Arg66 interacts with PPT and forms a large number of comprehensive hydrogen bonds (Table 2). The side chain of Arg66 enters into the S1 pocket of PPT and interacts with residues of the S1 pocket. The guanidinium group of Arg66 forms a salt-bridge with the carboxylate group of Asp189 of PPT. Arg66 of TKI interacts with Asp189, Ser190, Gly193, Ser195, Ser214 and Gly219 of PPT. Arg66 also interacts with Gln192 and Gly216 of PPT through water (W83). The scissile bond was found to be uninterrupted in electron density. The P3 residue Arg64 interacts with Gly216 and Asn97 of PPT. Arg64 also interacts with Thr98, Gln175, Gln192 and Gly218 through water. Ser68 at the P2′ position also participates in making hydrogen bonds with His40 CO and Phe41 of PPT. The P3′ residue His69 shows a slight conformational change compared with free TKI where it interacts with Ser39 and His40 of PPT.
|Thr3 OG1||Lys60 NZ||2.8|
|Asn15 NH2 and CO||W25||3.0, 3.3|
|NH2 and CO||W25||3.0, 3.3|
|NH2 and CO||W25||3.0, 3.3|
|Arg64 NH2||Asn97 CO||2.8|
|NH1 and NH2||W151||3.0, 3.4|
|NH1 and NH2||W151||3.0, 3.4|
|Arg66 CO||Ser195 N||2.9|
|Ser68 OG||His40 CO||3.5|
|Ser75 OG||Asn97 ND2||3.3|
|Ser116 OG||Tyr217 OH||3.4|
Molecular docking of TKI with FXa
Rigid body docking was performed using program cluspro. Figure 7 shows the detailed interactions of the best model of the TKI–FXa complex taken from the 30 highest ranking structures calculated by cluspro. The docking studies of TKI A and TKI B with FXa show interactions with the S1 and S4 pocket of FXa in a classical L-shaped substrate-like conformation (Fig. 7B). Our studies of TKI–FXa show that TKI B has a greater number of favourable stabilizing interactions than TKI A and the total buried surface of TKI B between the TKI–FXa complex is 1253 Å2, which is more than for TKI A as calculated by the program pdbsum . Various types of interaction of residues of TKI with FXa residues were calculated by the pic server. Mainly, residues from three loops (loop β1, loop β4β5 and loop β7β8) interact with FXa. Residues Gln18, Arg60, Arg66, His69, Glu79 and Lys136 of TKI form salt-bridges with Lys148, Glu147, Asp189, Glu39, Arg222 and Glu97 of FXa, respectively (Table 3). Of the seven residues of the reactive site loop, six residues interact with FXa residues and participate in stabilization of the complex. The detailed interactions are shown in Table 3 and Fig. 7C. The Arg66 (P1) of TKI intrudes inside the substrate binding pocket S1 of FXa making direct interactions with residues of the S1 pocket.
|TKI B||FXa||d (Å)|
|Gln18 OE1||Lys148 NZ||2.6|
|Arg60 NH2||Glu147 OE1||3.9|
|Arg66 NH2||Asp189 OD1||3.3|
|His69 ND1||Glu39 OE1||3.0|
|Glu79 OE2||Arg222 NH1||2.8|
|Lys136 NZ||Glu97 OE2||2.7|
|(B) Reactive site loop (P4–P3′) interactions|
|Ser63 OG||Tyr99 OH||3.4|
|Arg64 NH1||Glu 97 CO||2.7|
|NH2||Thr 98 CO||2.6|
|Ala65 CO||Gln192 NE2||3.0|
|Arg66 NH1||Asp 189 OD1||3.5|
|NH1||Ala 190 CO||3.1|
|CO||Ser 195 N||2.9|
|N||Ser 214 CO||3.3|
|Ser68 OG||Glu39 OE2||3.0|
|His69 ND1||Glu39 OE2||2.9|
|(C) Gln192 of FXa interaction with TKI residues|
|Gln192 NE2||Ser63 CO||3.2|
|Gln192 NE2||Ala65 CO||3.0|
|Gln192 OE1||Asn14 CO||3.5|
|Gln192 OE1||Asn15 ND2||2.9|
|(D) Additional interactions|
|Ala73 CO||Lys96 NZ||2.8|
|Ser75 OG||Glu97 OE1||2.9|
|Ser116 OG||Ser173 CO||2.8|
|Ala118 N||Glu97 OE2||3.1|
|Asn15 ND2||Glu147 OE2||3.2|
|Gly17 CO||Lys148 NZ||2.6|
The positively charged side chain of Arg64 (P3) of TKI accesses the S4 pocket of FXa parallel to the π-electron-rich aromatic ring of Phe174 of FXa and may participate in a cation–π interaction (Fig. 7B). The cation–π interaction can take place in parallel, in perpendicular or in a T-shaped orientation [39, 40]. The pic server predicts a cation–π interaction with Tyr99, Phe174 and Trp215 of the S4 pocket. Arg64 further stabilizes the complex by forming hydrogen bonds with Glu97, Thr98 and Gly216. One of the residues Gln192, reported as a vital residue in FXa for specificity and binding [41, 42], interacts with four important residues of TKI forming six hydrogen bonds (Table 3). Two of them are reactive site loop residues Ser63 (P4) and Ala65 (P2); one is a spacer residue Asn14 and one is an insertion residue Asn15 which is observed in TKI only. The residues from some of the loops also take part in stabilization of the complex, such as Ala73 and Ser75 of the β4β5 loop, Ser116 and Ala118 of loop β7β8 and Asn14, Asn15, Gly17 and Gln18 of loop β1.
A Kunitz type protease inhibitor from tamarind seeds has been cloned and characterized both biochemically and structurally. In this report, the dual protease inhibitor activities of the purified TKI against human FXa and trypsin have been investigated. The 3D structures of TKI and TKI–trypsin complex were determined. Additionally, a homology model of the TKI–FXa complex structure was constructed. The 3D structure of TKI shows a β-trefoil fold similar to other Kunitz STI type inhibitors. However, there are many distinct features of the TKI structure and interactions of the TKI reactive site with FXa and trypsin which make it a novel Kunitz type inhibitor.
Based on sequence homology, TKI has been classified as a Kunitz type protease inhibitor. The atomic structure of TKI illustrates that it possesses a β-trefoil fold similar to Kunitz (STI) type inhibitors which makes TKI a member of the Kunitz type inhibitor family. The TKI–trypsin complex was crystallized and its structure was determined to reveal the key structural features of TKI important for trypsin inhibition. Additionally, biochemical studies showed that TKI is a potent competitive inhibitor of FXa and trypsin with equilibrium dissociation constants (Ki) in the nano-molar range. The competitive inhibition of trypsin by TKI has been further confirmed by a structural study of the TKI–trypsin complex, which clearly shows the binding of the TKI reactive site loop in the active site of trypsin. This inhibitory study of TKI towards trypsin is contrary to the previously published report which showed that TKI was a non-competitive inhibitor of trypsin .
Analysis of the TKI reactive site showed that it has scant sequence identity with other Kunitz type inhibitors (Fig. 8A) as well as FXa inhibitors. Comparison of the reactive site of TKI with that of other physiological or non-physiological inhibitors of FXa such as antithrombin (AT) , protein Z-dependent protease inhibitor , tissue factor pathway inhibitor (TFPI) , BuXI , antistasin  and tick anticoagulant peptide (TAP)  shows that the reactive site loop does not have much similarity with these inhibitors except Arg at P1 as in AT, BuXI and antistasin and Arg at P3 as in TAP and antistasin that has P1 Arg and P3 Arg residues. This comparison reflects the uniqueness of the TKI reactive site loop. Furthermore, the interactions of distinct reactive site residues of TKI with trypsin in the complex of TKI–PPT are responsible for the difference in stabilization pattern compared with the STI–PPT complex (Table 2).
Comparison of the model of the TKI–FXa complex with the TKI–PPT complex structure reveals the structural features of TKI that could be accountable for the dual inhibitory property. The inhibition of FXa as well as trypsin is probably due to the unique reactive site which is flexible and can have diverse conformations. This is evident from the two molecules TKI A and TKI B present in the asymmetric unit that show structural variation in the reactive site loop. Molecular modelling studies show that six residues of the reactive site of TKI interact with S1, S4 and two secondary binding site/exosites (acidic 36 loop and autolysis loop) residues and provide specificity for FXa (Table 3). The serine residue at P4 may be crucial in FXa complex formation as it interacts with two vital residues, Tyr99 and Gln192, which define specificity for FXa [41, 42, 48]. There is an uncommon residue Arg at the P3 position in Kunitz STI inhibitors which occupies the S4 hydrophobic pocket, the prime determinant site for specificity of FXa [49, 50], and interacts with residues of the cation-binding hole of S4. The side chain of Arg64 is parallel to the hydrophobic ring of Phe174 giving rise to the possibility of forming the cation–π interaction (Fig. 7B). A similar role of Arg at P3 was observed in the TAP–FXa complex structure  and in the modelled antistasin–FXa complex . These observations support the hypothesis that the Arg64 (P3) of TKI could be one of the most significant residues, which provides the specificity towards FXa. The residue Arg at P1 which is observed in most Kunitz inhibitors is able to make contact with S1 pocket residues of FXa in a similar way as in trypsin.
Another two residues Ser and His at P2′ and P3′ that interact with exosites of FXa are non-polar in other Kunitz type inhibitors such as STI, ETI, DrTI, CTI, BuXI, BvTI, EcTI and WCI. Ser68 interacts with acidic 36 loop and autolysis loop residues of FXa. His69 makes a salt-bridge with acidic 36 loop residue (Table 3). It was reported earlier that AT forms salt-bridges and hydrogen bonds with the 36 loop and the autolysis loop residues of FXa . In the TAP–FXa complex structure, residues of TAP form hydrogen bonds and salt-bridges with the positively charged residues of the autolysis loop . Similar interactions were also shown in the modelled complex of antistasin–FXa  and in the second Kunitz domain of TFPI–FXa . Therefore Ser and His could also be the key determinants towards the specificity for FXa. These observations indicate that reactive site residues of TKI favour formation of a stable complex with FXa. Mutagenesis studies will further enlighten the effects on inhibitory activity and will provide an approach to designing an improved novel inhibitor that can act as anticoagulant in vivo. Along with Arg64 (P3), other residues such as Ser63, Ala73, Ser75, Ala118 and Lys136 of TKI also interact with residues of S4 and could further stabilize the complex. Such types of interaction are not observed in the TKI–PPT complex as ‘96-KETY-99′ in FXa for specificity is unavailable in PPT (see Fig. 8B). TKI inhibits both proteases despite less sequence identity (37%) of human FXa with porcine trypsin, as it has specificity for the S4 pocket of FXa which is not observed in most Kunitz STIs.
Besides the versatile reactive site, two other distinct features of TKI include insertion of the residue Asn15 in the Kunitz signature motif and a flexible loop β7β8. The inserted residue disrupts the Kunitz signature motif in TKI. The crystal structure of free TKI shows that Asn15 (loop β1) has different conformations in the TKI A and TKI B molecules. The Cα distance between Asn15 of TKI A and TKI B is 0.9 Å and their side chains are 2.8 Å apart from each other (Fig. 9A). The superposition of TKI A and TKI B on STI and ETI showed that Asn15 is not superimposed with any residue of ETI and STI; on the contrary it points outwards from an exposed loop at the protein surface. The possible role of this inserted residue lies in the TKI–PPT complex structure and the modelled TKI–FXa complex. In TKI–PPT complex structure, Asn15 forms a hydrogen bond with Gln192 and Asn143 through W25 and further with Lys145 and Gly148 via a second water molecule (W154). Consequently, the formation of a network of hydrogen bonds by Asn15 through two water molecules contributes to a stronger interaction of TKI with PPT and hence the stabilization of the TKI–PPT complex (Fig. 9C and Table 2). Moreover, in modelled TKI–FXa complex, Asn15 again plays a vital role as it forms three hydrogen bonds with Gln192 of FXa, a key determinant for specificity of substrate/inhibitor [41, 42] (Fig. 9D). The interaction of Asn15 with Glu147 of the autolysis loop is also helpful for providing stability with FXa (Table 3). Together, these results indicate that TKI has a key residue insertion that could be important in recognition by its proteases and stabilization of the complexes. CTI also has an insertion of Asp at the same place as Asn (Fig. 9B); however, the implication of this insertion in CTI has not been reported.
Superposition of TKI A on TKI B shows conformational variation at the loop β7β8 (Asp112-Pro119). This loop adopts a unique conformation compared with the loop of STI. The loop β7β8 of TKI is closer to the reactive site loop and its flexible nature indicates that it may interact with residues of proteases to which TKI binds and stabilizes the complex. In the TKI–PPT complex structure, Ser116 interacts with residue Tyr217 of PPT. These observations suggest that the flexible loop β7β8 probably plays an important role in the formation of a complex. However, the interaction of the β7β8 loop with PPT was not observed in STI–PPT, which indicates that it may not be very important in complex formation with PPT but it might be crucial for stabilization of complexes with other serine proteases. For example, in the modelled TKI–FXa complex, two residues of this loop interact with FXa residues (Table 3). One of them is Glu97 of the S4 pocket of FXa which demonstrates that this loop may be crucial in complex formation with FXa and could enhance the affinity of TKI towards FXa. So, conclusively, residues of loop β1, β4β5 (reactive site) and β7β8 of TKI are able to interact with S1, S4 pockets and two exosites of FXa and confer specificity for FXa along with trypsin.
In summary, comparative analyses with other Kunitz type inhibitors at sequence and 3D structural levels allow us to clarify and propose the structural and mechanistic basis for the dual targeting capability of TKI, which binds and inhibits FXa as well as trypsin. However, the distinct reactive site of TKI may possibly inhibit other trypsin like serine proteases and needs further investigation. Moreover, inhibition of FXa and prolongation of blood clotting time point towards the potential clinical application and use of TKI in antithrombotic therapy.
Materials and methods
RNA extraction and cDNA cloning
In order to obtain the complete amino acid sequence of TKI, 3-month-old Tamarindus indica seeds were collected locally. Total RNA was isolated from the seeds as described previously  with slight modifications such as the use of 4% polyvinylpyrrolidone in the extraction buffer and slightly higher pH extraction buffer . The first strand of cDNA was reverse transcribed from total RNA using oligo(dT)18 adaptor primer (5′-CCAGTGAGCAGAGTGACGACTCGAGCTCAAGCTTTTTTTTTTTTTTTTTT-3′). The TKI coding sequence was amplified using a forward degenerate primer (F1: 5′-GAYACHGTNCAYGAYACHGAYGG-3′) which was designed on the basis of the previously reported N-terminal sequence of TKI , and the oligo(dT)18 adaptor primer was used as the reverse primer (R1). Then amplified TKI gene was cloned into pGEM-T Easy Vector according to the instructions provided (Promega, Madison, WI, USA) and transformed using DH5α competent cells. The obtained recombinant plasmid containing TKI gene insert (pGEMT-TKI) was sequenced in both directions using M13 forward and reverse primers.
Analysis of TKI gene
The ORF finder tool from NCBI was used to deduce the primary amino acid sequence of TKI from the TKI gene sequence. Similar sequences available in the database were identified by the program blast . Multiple sequence alignment was performed using clustal w . The prediction of the Kunitz signature and probable motifs involved in post-translational modifications were done using the motif scan server .
Purification and crystallization of TKI
The purification and crystallization of TKI were reported earlier . However, better diffracting crystals for free TKI were obtained using 25% (v/v) polyethylene glycol 3350 (PEG 3350) and 0.2 m magnesium formate at pH 7.0.
TKI–PPT and TKI–FXa complex formation and purification
The TKI–PPT complex formation, purification and crystallization were done as described previously . Furthermore, the crystals used for this complex were obtained using 0.2 m ammonium phosphate dibasic and 20% PEG 3350. For TKI–FXa complex formation, the purified 100 μL TKI (15 mg·mL−1 in 20 mm Tris buffer pH 8.0) was mixed with 100 μL human FXa (2.5 mg·mL−1 concentrated from 1 mg·mL−1 FXa). The TKI–FXa mixture was incubated at 37 °C for 2 h. The complex was purified using a calibrated Superdex-75 GL 10/300 (GE Healthcare, Uppsala, Sweden) size-exclusion column on an AKTA Purifier FPLC system (GE Healthcare). The fractions were analysed on 15% reducing SDS/PAGE and concentrated to 4.5 mg·mL−1 using an Amicon Ultra-4 10 kDa cutoff concentrator (Millipore, Carrigtwohill, Co., Cork, Ireland).
APTT  and PT  assays were performed using normal human plasma and commercially available kits (Liquiplastin for PT and Liquicelin-E for APTT, Tulip Diagnostics Pvt. Ltd, Roorkee, India) in the presence and absence of TKI. For in vitro APTT assays, normal human plasma (100 μL), kit reagent (100 μL) and TKI (0–10 μm, previously diluted in 100 μL of 0.1 m Tris/HCl buffer pH 8.0) were mixed and incubated for 3 min at 37 °C. To initiate the coagulation process, 100 μL of 0.025 m CaCl2 was added, and the time for clot formation was recorded. To measure the in vitro PT assay, normal human plasma (100 μL) was incubated with different concentrations of TKI (0–10 μm previously diluted in 100 μL of 0.1 m Tris/HCl buffer at pH 8.0) for 2 min at 37 °C. Then, 100 μL of the kit reagent was added, and the clotting time was measured. Both the experiments were carried out in triplicate and the average value was taken.
Assay of inhibitory activity
Trypsin inhibitory activity of TKI was determined by measuring the residual hydrolytic activity of PPT towards a BAPNA substrate (Sigma-Aldrich, St Louis, MO, USA) and chymotrypsin inhibitory activity was assayed using BTPNA substrate (Sigma-Aldrich) at pH 8.0. FXa inhibitory assay was performed as described previously with modification [59-61]. 0.25 μm human FXa (Pierce, Thermo Scientific, USA) was incubated with different concentrations of TKI (1–60 μm) in 50 mm Tris buffer pH 8.0 and 300 mm NaCl at 37 °C for 30 min on a waterbath. After incubation, the chromogenic substrate CH3OCO-D-CHA-Gly-Arg-pNA-AcOH (Sigma-Aldrich Co, F3301) was added and incubated for 3 min at 37 °C on a waterbath. The reaction was stopped by adding 2% citric acid. Elastase inhibitory activity of TKI was performed as described previously  by measuring the residual activity of porcine pancreatic elastase (PPE) using the substrate N-Suc-(Ala)3-nitroanilide (Sigma-Aldrich). The colour formation by the liberated chromophoric group pNA by enzymatic hydrolysis of BAPNA/BTPNA/FXa substrate/PPE substrate was measured spectrophotometrically at 410 nm. All assays were performed in triplicate.
The inhibition constant (Ki) of TKI against both trypsin and FXa was determined from a Dixon plot [63, 64]. The enzyme inhibition was characterized at two different substrate concentrations ([S1] and [S2]). BAPNA (4 and 6 mm) and FXa chromogenic substrate (5 and 7 mm) were used as a substrate for trypsin and FXa activity, respectively. Studies were performed by pre-incubating the fixed amount of respective enzyme [trypsin (0.014 nm), FXa (0.5 nm)] with increasing concentrations of TKI (0–15 nm for trypsin inhibition and 0–500 nm for FXa inhibition) for 15 min at 37 °C followed by addition of two different substrate concentrations. The reciprocal velocity (1/v) versus inhibitor concentrations [I], for each substrate concentration [S1] and [S2], were plotted (Dixon plots). Ki was calculated from the intersection of the two regression lines.
Data collection and processing
The X-ray diffraction data were collected at 1.94 Å for free TKI and 2.23 Å resolution for the TKI–PPT complex at home source. Data sets for free TKI and TKI–PPT complex were collected at 100 K. Indexing, integration of all the diffraction images and scaling of the diffraction data were carried out using hkl2000  and the data collection statistics are summarized in Table 1.
Structure determination and refinement
The structures of free TKI and its complex with trypsin were solved by the molecular replacement method using the program molrep in the ccp4i suite . The best solution for free TKI and the complex were obtained with the crystal structure of a trypsin inhibitor from Erythriana caffra seeds (PDB ID code, 1TIE)  and the complex structure of soybean (PDB ID code, 1AVW) as the search model respectively . The rigid body refinement was followed by iterative cycles of restrained atomic parameter refinement using refmac5 [66, 67]. Further model visualization, refinement, model building and fitting of the electron density map were carried out using the molecular graphics program coot [68, 69]. The details of the refinement statistics are included in Table 1. The final structures were validated using molprobity server . Figures were generated using the pymol  and espript program .
Rigid body docking
As crystals of TKI with FXa were not obtained, rigid body docking was performed to examine the mode of interaction of TKI with FXa. The fully automated online protein–protein docking program cluspro [72, 73] was used to model the TKI–FXa complex with good surface complementarity and low desolvation energies. The PDB file for FXa (PDB code 1FAX)  was submitted as a receptor structure, and the coordinates of TKI (PDB code 4AN6) as the ligand structure. The server was executed with default parameters. The top-ranked complexes from cluspro were used for further analysis based on the prior knowledge of active site interactions. Interactions between TKI and FXa residues were calculated using the pic server .
The authors thank Macromolecular Crystallographic Facility (MCU) at IIC, IIT Roorkee for the structure determination. This work has been supported financially by the Council of Scientific and Industrial Research (CSIR), New Delhi, India (No. 38(1228)/09/EMR II dated Dec 01, 2009). DNP thank MHRD Government of India, for financial support. We are grateful to Dr S. Karthikeyan, IMTECH-Chandigarh, for providing help in the structure solution of TKI–PPT complex. Dr Madhulika, MD (pathology), helped in providing the facility for blood coagulation assays. DNP also thanks the CSIR for financial support. The authors are grateful to Shivendra Pratap for help with the anticoagulation assay and docking studies and to P. Supriya, D. Sonali, P. Nandita and P. Selva Kumar for critical reading of the manuscript.