A Positively Charged Loop on the Surface of Kallistatin Functions to Enhance Tissue Kallikrein Inhibition by Acting as a Secondary Binding Site for Kallikrein*

Kallistatin is a serine proteinase inhibitor (serpin) that specifically inhibits tissue kallikrein. The inhibitory activity of kallistatin is abolished upon heparin binding. The loop between the H helix and C2 sheet of kallistatin containing clusters of basic amino acid residues has been identified as a heparin-binding site. In this study, we investigated the role of the basic residues in this region in tissue kallikrein inhibition. Kallistatin mutants containing double Ala substitutions for these basic residues displayed a 70–80% reduction of association rate constants, indicating the importance of these basic residues in tissue kallikrein inhibition. A synthetic peptide derived from the sequence between the H helix and C2 sheet of kallistatin was shown to suppress the kallistatin-kallikrein interaction through competition for tissue kallikrein binding. To further evaluate the function of this loop, we used a 1-antitrypsin, a non-heparin-binding serpin and slow tissue kallikrein inhibitor as a scaffold to engineer kallikrein inhibitors. An a 1-antitrypsin chimera harboring the P3-P2 * residues and a sequence homologous to the positively charged region between the H helix and C2 sheet of kallistatin acquired heparin-suppressed inhibitory activity toward tissue kallikrein and exhibited an inhibitory activity 20-fold higher than that of the other chimera, which contained only kallistatin’s P3-P2 * sequence, and 2300-fold higher than that of wild-type Design of Synthetic Peptides— A synthetic peptide, HC2, was de- signed according to the sequence between the H helix and C2 sheet spanning amino acids 300–319, RWNNLLRKRNFYKKLELHLPK. The basic residues are underlined. A negative control C4 peptide derived from the surface region around the C4 sheet was synthesized, spanning amino acids 222–242, PFISSRTTPKDFYVDENTTV. All synthetic peptides were dissolved in 2% Me 2 SO for use. Effect of Kallistatin Synthetic Peptides on the Binding between Tissue Kallikrein and Kallistatin— 125 I-labeled human tissue kallikrein (1 3 10 4 cpm/20 m l) was preincubated with different concentrations of synthetic peptide at 37 °C for 10 min. The concentrations of the HC2 peptide in the assay were 12.5, 25, 50, 100, 200, and 400 m M ; and the concentrations of the control peptide, C4, were 100, 200, 300, and 400 m M . The reaction mixtures were then incubated with 0.5 m M kallistatin for 1 h. The reaction mixtures were resolved in 10% SDS-polyacryl- amide gel electrophoresis and analyzed by autoradiography.

Kallistatin is a serine proteinase inhibitor (serpin) that specifically inhibits tissue kallikrein. The inhibitory activity of kallistatin is abolished upon heparin binding. The loop between the H helix and C2 sheet of kallistatin containing clusters of basic amino acid residues has been identified as a heparin-binding site. In this study, we investigated the role of the basic residues in this region in tissue kallikrein inhibition. Kallistatin mutants containing double Ala substitutions for these basic residues displayed a 70 -80% reduction of association rate constants, indicating the importance of these basic residues in tissue kallikrein inhibition. A synthetic peptide derived from the sequence between the H helix and C2 sheet of kallistatin was shown to suppress the kallistatin-kallikrein interaction through competition for tissue kallikrein binding. To further evaluate the function of this loop, we used ␣1-antitrypsin, a nonheparin-binding serpin and slow tissue kallikrein inhibitor as a scaffold to engineer kallikrein inhibitors. An ␣1-antitrypsin chimera harboring the P3-P2 residues and a sequence homologous to the positively charged region between the H helix and C2 sheet of kallistatin acquired heparin-suppressed inhibitory activity toward tissue kallikrein and exhibited an inhibitory activity 20-fold higher than that of the other chimera, which contained only kallistatin's P3-P2 sequence, and 2300fold higher than that of wild-type ␣1-antitrypsin. The ␣1-antitrypsin chimera with inhibitory characteristics similar to those of kallistatin demonstrates that the loop between the H helix and C2 sheet of kallistatin is crucial in tissue kallikrein inhibition, and this functional loop can be used as a module to enhance the inhibitory activity of a serpin toward tissue kallikrein. In conclusion, our results indicate that a positively charged loop between the H helix and C2 sheet of a serpin can accelerate the association of a serpin with tissue kallikrein by acting as a secondary binding site.
Serine proteinase inhibitors (serpins) comprise a superfamily of single-chain proteins that share a highly conserved tertiary structure. Most of the family members act as inhibitors for serine proteinases and regulate a variety of physiological processes by inhibiting their target proteinases (1). The inhibitory activity of a serpin is determined primarily by its protruding reactive center loop that docks into the reactive cleft of its target proteinase and traps the enzyme in a covalent complex (2). For some serpins, a secondary binding site remote from the reactive center loop is required for proteinase inhibition (3)(4)(5)(6)(7)(8). The secondary binding sites interact with complementary sites and often promote the interaction between a serpin and its target proteinase. Another functional structure affecting the inhibitory specificity of a serpin is the heparin-binding site. For most of the heparin-binding serpins, heparin binding accelerates the association of serpins with their target proteinases (9 -13). The heparin-accelerated inhibition of proteinases has been explained by a ternary complex and an allosteric model (9, 10, 14 -17). For development of specific and potent inhibitors of serine proteinases, it is essential to identify the structural elements related to the inhibitory activity and understand the orchestration of these structural elements on the inhibition.
Kallistatin is a serpin that specifically inhibits human tissue kallikrein by forming a covalent enzyme-inhibitor complex (18). The high selectivity of kallistatin toward tissue kallikrein is attributed to its unique Phe-Phe pair at the P2-P1 residues in the reactive center (19). Kallistatin is also one of the heparinbinding serpins (18), which include antithrombin, protease nexin I, heparin cofactor II, plasminogen activator inhibitor, and protein C inhibitor (PCI). 1 Unlike most heparin-binding serpins whose inhibitory activities are accelerated by heparin, the inhibitory activity of kallistatin toward tissue kallikrein is abolished upon heparin binding (18). The heparin-suppressed effect is also observed in tissue kallikrein inhibition by PCI (20). Heparin-binding sites of serpins are composed of clusters of basic residues that bind to the acidic groups of heparin (21). These sites are distributed mainly in the D helix, whereas that of PCI is in the H helix (10,11,(22)(23)(24)(25). Compared with other serpins, kallistatin has a distinct heparin-binding site localized in the region between the H helix and the C2 sheet (26). The heparin-binding sites of kallistatin and PCI are close to each other in primary structure. However, the major heparin-binding residues K312:K313 of kallistatin are located at the N terminus of the C2 sheet, whereas the major heparin-binding residues R269:K270 of PCI are located in the H helix. Additionally, amino acid sequence alignment of serpins shows that kallistatin has a unique insertion of 3-4 residues in the loop between the H helix and C2 sheet. A molecular model of kallistatin indicates that these extra residues may bulge the loop toward the reactive center loop (19). The basic amino acid residues K312:K313 in this loop are thus positioned in close vicinity to the reactive site ( Fig. 1). We therefore hypothesize that the positively charged region between the H helix and C2 sheet of kallistatin could have a direct involvement in the interaction with tissue kallikrein in addition to heparin binding.
In the present study, we investigated the role of the basic residues in the region between the H helix and C2 sheet of kallistatin in tissue kallikrein inhibition by three approaches. First, we evaluated the association rates of tissue kallikrein with kallistatin variants containing double Ala substitutions for these basic residues in this loop. Second, we used a synthetic peptide derived from the sequence between the H helix and C2 sheet of kallistatin to assess its capability to compete with kallistatin for tissue kallikrein binding. Finally, we created chimeric ␣1-antitrypsin, which harbors the P3-P2Ј sequence and a sequence homologous to the region between the H helix and C2 sheet of kallistatin, to evaluate the effect of this positively charged loop on tissue kallikrein inhibition. ␣1-Antitrypsin was used as a scaffold because it has neither heparin binding activity nor significant inhibitory activity toward tissue kallikrein. This study provides evidence that the positively charged loop between the H helix and C2 sheet of kallistatin acts as a secondary binding site to accelerate the interaction of kallistatin and tissue kallikrein.

Construction, Expression, and Purification of Kallistatin Mutants Containing Mutations in the Region between the H Helix and C2 Sheet-
The kallistatin variants K307A/R308A and K312A/K313A were designed to contain Ala substitutions for the basic amino acid residues in the region between the H helix and C2 sheet. K307A/R308A and K312A/ K313A were created, expressed, and purified by the method described previously (26).   (29). Design of Synthetic Peptides-A synthetic peptide, HC2, was designed according to the sequence between the H helix and C2 sheet spanning amino acids 300 -319, RWNNLLRKRNFYKKLELHLPK. The basic residues are underlined. A negative control C4 peptide derived from the surface region around the C4 sheet was synthesized, spanning amino acids 222-242, PFISSRTTPKDFYVDENTTV. All synthetic peptides were dissolved in 2% Me 2 SO for use. Construction of ␣1-Antitrypsin Expression System-The ␣1-antitrypsin cDNA was synthesized from total RNA of human hepatocytes by reverse transcription-PCR with the primers 5Ј-ATGCCGTCTTCT-GTCTCG-3Ј and 5Ј-TTATTTTTGGGTGGGATT-3Ј. After gel purification and phosphorylation of the PCR product, the full-length ␣1-antitrypsin cDNA was cloned into pBluescript. A previous report showed that deletion of 5-10 amino acids from the N terminus of ␣1-antitrypsin increased recombinant protein expression without affecting the inhibitory activity of ␣1-antitrypsin (28). Therefore, the ␣1-antitrypsin cDNA with an 8-codon deletion from the N terminus was synthesized by PCR using synthetic oligonucleotide 5Ј-AGGTGCTAGCCAGAAGACA-GATAC-3Ј (harboring a NheI site as indicated by the underlined residues) and the T7 promoter as primers and the full-length ␣1-antitrypsin cDNA as template. After the cDNA was cut with NheI and EcoRI, it was cloned into prokaryotic expression vector pTrc His linearized by the same restriction enzymes. The pTrc His vector adds a hexa-histidine sequence at the N terminus of a recombinant protein for purification by metal-affinity chromatography. The addition of a hexa-histidine sequence for purification was applied in many studies that showed no effect on the inhibitory activity of serpins (29,30).

Effect of Kallistatin Synthetic Peptides on the Binding between Tissue Kallikrein and Kallistatin-
Construction, Expression, and Purification of ␣1-Antitrypsin Chimeras-Two ␣1-antitrypsin chimeras, denoted as ␣1-AT/KS and ␣1-AT/ KS/H, were created by site-directed mutagenesis using a sequence overlap-extension PCR method. ␣1-AT/KS was constructed with the P3-P2Ј sequence of kallistatin, and ␣1-AT/KS/H was engineered with not only the P3-P2Ј sequence but also with a sequence homologous to the region between the H helix and C2 sheet of kallistatin using ␣1antitrypsin as a scaffold. To create ␣1-AT/KS, the ␣1-antitrypsin cDNA in pBluescript was used as a template, with oligonucleotides 5Ј-GAG-GCCAAATTCTTCTCTGCACCCCCCGAG-3Ј and 5Ј-CTCGGGGGGTG-CAGAGAAGAATTTGGCCTC-3Ј as internal primers (mutant bases are indicated by the underline) and oligonucleotides 5Ј-AGGTGCTAGCCA-GAAGACAGATAC-3Ј and the T7 promoter primer as outside primers in the PCR. The mutant cDNA synthesized by PCR is digested by NheI/ EcoRI and cloned into the pTrc His expression vector. The chimera ␣1-AT/KS/H was created by the same method using the first chimera as template and oligonucleotides 5Ј-CCTGGAAAATCTCAAGAAGCGTT-TCTACAGAAGGTCTG-3Ј and 5Ј-CAGACCTTCTGTAGAAACGCTTT-TGAGATTTTCCAGG-3Ј as internal primers. The inserted and mutated nucleic acids are underlined. The mutations were confirmed by DNA sequencing. The methods for expression of the mutants were the same as those described previously (29). The recombinant ␣1-antitrypsin was first purified by nickel-affinity chromatography (29). After dialysis against 50 mM Tris-HCl (pH 8.0) and 20 mM NaCl, the sample was loaded onto an anion-exchange chromatography column (QE column) in a BioCAD TM SPRINT TM Perfusion Chromatography System. The recombinant protein was eluted with a NaCl gradient ranging from 20 -500 mM. The peak fractions were collected and analyzed by SDS-polyacrylamide gel electrophoresis.
Effect of Heparin on the Binding between Tissue Kallikrein and Kallistatin-Tissue kallikrein binding activity of the ␣1-antitrypsin  (26). Kallistatin mutants K307A/R308A and K312A/K313A containing double Ala substitutions for the basic amino acid residues in the region between the H helix and C2 sheet were created using kallistatin P1Arg as a scaffold. To investigate the importance of these basic residues in the inhibitory activity of kallistatin toward tissue kallikrein, we measured the association rate constants (k a ) of the kallistatin mutants and kallistatin P1Arg with tissue kallikrein. As shown in Table I, the k a of the mutants K307A/R308A and K312A/K313A were markedly reduced to 1.1 ϫ 10 4 and 8.0 ϫ 10 3 M Ϫ1 s Ϫ1 , which were only 35% and 21% of the k a (3.9 ϫ 10 4 M Ϫ1 s Ϫ1 ) for kallistatin P1Arg (29). Kinetic analysis of the inhibitory activity of these mutants indicates that the basic amino acid residues in the region between the H helix and C2 sheet are crucial for the inhibitory activity of kallistatin toward tissue kallikrein.
Effect of Kallistatin Synthetic Peptides on the Binding between Tissue Kallikrein and Kallistatin-To determine whether the positively charged loop between the H helix and C2 sheet of kallistatin can bind to tissue kallikrein, a synthetic peptide HC2 derived from this region was used to compete with kallistatin for tissue kallikrein binding in a kallistatin-kallikrein binding assay. The results showed that the HC2 peptide interfered with the complex formation of tissue kallikrein and kallistatin (Fig. 2). The HC2 peptide at 6.25, 12.5, 25, 50, 100, 200, and 400 M competed with 0.5 M kallistatin in a dose-dependent manner for tissue kallikrein binding and reduced the intensity of the band around 85 kDa representing the tissue kallikrein-kallistatin complex (Fig. 2A). The control peptide C4, derived from the surface region around the C4 sheet, did not significantly affect the formation of the kallistatin-kal-likrein complex even at concentration as high as 400 M (Fig.  2B). These results suggest that the region covering the H helix and C2 sheet in kallistatin can bind to tissue kallikrein.
Construction of ␣1-Antitrypsin Chimeras-To further investigate the role of the positively charged loop between the H helix and C2 sheet of kallistatin in tissue kallikrein inhibition, we engineered serpin chimeras using ␣1-antitrypsin as a scaffold. ␣1-Antitrypsin has no known heparin binding activity, and it is a very slow progressive inhibitor of tissue kallikrein with a k a of 7.7 M Ϫ1 s Ϫ1 , which is more than 3 orders of magnitude lower than that of kallistatin (31). Many studies have shown that substitution of the reactive center residues of a serpin changes its inhibitory specificity (32,33). We first constructed a ␣1-antitrypsin chimera ␣1-AT/KS containing kallistatin's reactive center sequence by replacing the P3-P2Ј sequence of ␣1-antitrypsin (IPMSI) with that of kallistatin (KFFSA) (Fig. 3). Using ␣1-AT/KS as a backbone, the other chimera, ␣1-AT/KS/H, was engineered with a positively charged sequence homologous to the region between the H helix and C2 sheet of kallistatin. The region between the H helix and C2 sheet in ␣1-antitrypsin contains fewer basic residues (only R305:R306) as compared with kallistatin, which contains R306:K307:R308 and K312:K313 in this region. In addition, ␣1-antitrypsin harbors many acidic residues, such as Asp 301 , Glu 303 , and Asp 304 , in this region that may affect the binding of this region with negatively charged molecules, such as heparin. Another feature we consider a critical factor is a 4-residue insertion in this region in kallistatin compared with ␣1-antitrypsin. According to a molecular model of kallistatin, we speculate that this insertion may protrude the loop between the H helix and C2 sheet in a more accessible conformation for interaction with other proteins (26). Based on these concerns, we introduced a LKKRFYRR sequence homologous to the heparin-binding sequence of kallistatin to replace EDRR in this region for ␣1-antitrypsin. This substitution accommodates 3 additional basic amino acids, 4 extra amino acid residues, and replacement of the acidic residues Glu 303 and Asp 304 (Fig. 3). By this sequence modification, a structural element imitating the positively charged loop between the H helix and C2 sheet of kallistatin was created in ␣1-antitrypsin.
Purification of ␣1-Antitrypsin and the Chimeras-The histidine-tagged ␣1-antitrypsin was purified by nickel-affinity chromatography to over 95% homogeneity as estimated by Coomassie Blue staining after SDS-polyacrylamide gel electrophoresis. The samples of ␣1-antitrypsin chimeras, however, still contained contaminating proteins after processed by a nickel-affinity column. The samples were further purified by anionexchange chromatography. The chimeras ␣1-AT/KS and ␣1-AT/KS/H were eluted by 170 -190 and 100 -130 mM NaCl, respectively. The recombinant proteins were able to reach 95% purity.
Effects of Heparin on Tissue Kallikrein Binding Activity of the ␣1-Antitrypsin Chimeras, ␣1-Antitrypsin, and Kallistatin-To determine whether the ␣1-antitrypsin chimeras, ␣1-   ) of kallistatin ␣1-antitrypsin, and the chimeras The k a values were determined according to the method described previously (29). AT/KS and ␣1-AT/KS/H, gain tissue kallikrein binding activity and heparin-suppressed inhibitory activity, a tissue kallikrein binding assay was carried out in the presence or absence of heparin. The results showed that both ␣1-AT/KS and ␣1-AT/ KS/H were able to form an 85-kDa complex with human tissue kallikrein in the absence of heparin (Figs. 4 and 5). The complex formation between tissue kallikrein and ␣1-AT/KS, which contained only the P3-P2Ј sequence of kallistatin, was not affected by heparin, even at a concentration as high as 50 units/ml (Fig. 4). In contrast, the complexes formed between tissue kallikrein and ␣1-AT/KS/H, which carried a homologous heparin-binding sequence of kallistatin in addition to the P3-P2Ј sequence, were inhibited by heparin in a dose-dependent manner (Fig. 4). Wild-type ␣1-antitrypsin failed to form a detectable complex with tissue kallikrein even in the absence of heparin under the experimental conditions, whereas ␣1-AT/KS and ␣1-AT/KS/H, like kallistatin, exhibited remarkable kallikrein binding activity (Fig. 5). ␣1-AT/KS/H appeared to act more like kallistatin because its tissue kallikrein binding activity was suppressed by heparin (Fig. 5). These results indicate that ␣1-antitrypsin can acquire tissue kallikrein binding activity and heparin-regulated activity by incorporation of the P3-P2Ј sequence and a homologous heparin-binding site of kallistatin between the H helix and C2 sheet. Effects of Heparin on the Inhibitory Activity of the ␣1-Antitrypsin Chimeras toward Tissue Kallikrein-The inhibition of tissue kallikrein by ␣1-antitrypsin chimeras in the presence of different concentrations of heparin was quantified by measuring residual amidolytic activities of tissue kallikrein after reaction with ␣1-antitrypsin chimeras. The results are shown in Fig. 6. After incubation for 3.5 h in the absence of heparin, the enzymatic activity of tissue kallikrein was completely inhibited by ␣1-AT/KS/H, and about 86% of the enzymatic activity was inhibited by ␣1-AT/KS. Preincubation with heparin at 10 units/ml suppressed the inhibitory activity of ␣1-AT/KS/H and restored Ͼ90% of tissue kallikrein activity. However, ␣1-AT/KS was not significantly affected by heparin, even at a high concentration of 50 units/ml, because it still inhibited approximately 80% of tissue kallikrein activity. The results indicate that the insertion of the homologous heparin-binding site of kallistatin between the H helix and C2 sheet confers heparinsuppressed tissue kallikrein inhibition to ␣1-antitrypsin chimeras.
Comparison of Association Rate Constants (k a ) of Kallistatin, Wild-Type ␣1-Antitrypsin, and the ␣1-Antitrypsin Chimeras with Human Tissue Kallikrein-The k a values were determined under pseudo-first-order conditions using at least a 50fold molar excess of inhibitors over tissue kallikrein, and the results are summarized in Table II. Wild-type ␣1-antitrypsin is a slow inhibitor for tissue kallikrein with a low k a of 7.7 M Ϫ1 s Ϫ1 (31). ␣1-AT/KS, with the P3-P2Ј sequence of kallistatin, showed a boosted k a of 8.7 ϫ 10 2 M Ϫ1 s Ϫ1 , which is 100-fold higher than that of wild-type ␣1-antitrypsin toward tissue kallikrein. By incorporation of the basic residues in the region between the H helix and C2 sheet, the inhibitory activity of ␣1-AT/KS/H was further enhanced to 1.8 ϫ 10 4 M Ϫ1 s Ϫ1 , which is approximately 20-fold higher than that of ␣1-AT/KS and 2300-fold higher than that of wild-type ␣1-antitrypsin. The k a of ␣1-AT/KS/H is promoted to a value comparable with that of kallistatin, 1.9 ϫ 10 4 M Ϫ1 s Ϫ1 (29). These results indicate that the positively charged loop between the H helix and C2 sheet in kallistatin accelerates its association with human tissue kallikrein. DISCUSSION A positively charged region between the H helix and C2 sheet of kallistatin has been identified previously as a heparin-binding site that is responsible for the heparin-suppressive kallikrein inhibition (26 tissue kallikrein. This is the first report successfully converting a non-heparin-binding serpin to a heparin-regulated serpin by insertion of a heparin-binding motif. Additionally, we are able to engineer a potent tissue kallikrein inhibitor, whose inhibitory activity increases 2300-fold to a level comparable with that of kallistatin, by transferring the P3-P2Ј sequence and a positively charged sequence homologous to the region between the H helix and C2 sheet (homologous heparin-binding sequence) of kallistatin into ␣1-antitrypsin. We have provided several lines of evidence to show that clusters of positively charged residues in the region between the H helix and C2 sheet are critical for enhancement of the inhibitory activity of kallistatin toward human tissue kallikrein. First, substitution of the basic amino acid residues with alanine in this positively charged loop resulted in a 70 -80% reduction of the inhibitory activity of kallistatin toward tissue kallikrein. Second, a synthetic peptide corresponding to the sequence between the H helix and C2 sheet of kallistatin competed with kallistatin for tissue kallikrein binding. Finally, insertion of a sequence homologous to this positively charged loop of kallistatin into the region between the H helix and C2 sheet of ␣1-antitrypsin resulted in a Ͼ20-fold increase of the association rate with tissue kallikrein. Taken together, our results indicate that the basic residues in the region between the H helix and C2 sheet of kallistatin act as a secondary binding site to accelerate the association between kallistatin with tissue kallikrein.
The reactive center loop of a serpin is a primary determinant for the inhibitory specificity toward serine proteinases. Previous studies have showed that swapping of the reactive site among serpins is insufficient to transfer the complete inhibitory property of one serpin to another, suggesting requirement of other structural elements to fully restore the inhibitory activity (33). Therefore, an efficient association between a serpin and a proteinase may rely on additional interactions to stabilize complex formation other than the reactive sites. Several lines of evidence indicated that interaction between serpins and serine proteinases, such as plasminogen activator inhibitor and plasminogen activator, heparin cofactor II and thrombin, and antithrombin III and urokinase, requires a secondary binding to promote the association (3)(4)(5)(6)(7)(8). Consistent with these previous studies, our results showed that substitution of the reactive center of ␣1-antitrypsin with that of kallistatin increased the inhibitory activity of ␣1-antitrypsin toward tissue kallikrein 100-fold but did not completely transfer the inhibitory activity of kallistatin to ␣1-antitrypsin. The inhibitory activity was further increased Ͼ20-fold to a level comparable with that of kallistatin by incorporating clusters of basic residues between the H helix and C2 sheet, indicating a critical role of this positively charged loop in tissue kallikrein inhibition. The notion that this loop may promote the inhibition by acting as a secondary binding element is further supported by the results of a peptide competition assay. The synthetic peptide derived from the sequence between the H helix and C2 sheet of kallistatin competes for tissue kallikrein binding, suggesting a direct interaction of this region with tissue kallikrein. Taken together, incorporation of the reactive center of kallistatin confers the ␣1-antitrypsin chimera an inhibitory specificity toward tissue kallikrein, whereas the positively charged loop between the H helix and C2 sheet appears to be a necessary architecture to boost the inhibitory activity.
The loop between the H helix and C2 sheet of kallistatin is bulged in close vicinity to the reactive center loop in the tertiary structure (26). When the reactive center loop of kallistatin docks into the reactive crevice of tissue kallikrein, the bulged loop between the H helix and C2 sheet of kallistatin would be in an accessible position to interact with tissue kallikrein. Tissue kallikrein is an acidic protein with a pI of 3-4, and it is highly negatively charged under neutral conditions (34). In a molecular model of human tissue kallikrein, several acidic amino acid residues were found to be distributed in loops surrounding the reactive crevice that may constitute cation-binding exosites (Fig. 7) (35). Accordingly, we speculate that the high density of positively charged residues in the loop between the H helix and C2 sheet of kallistatin may interact with these acidic amino residues in the loop surrounding the reactive crevice of tissue kallikrein. Therefore, the basic residues between the H helix and C2 sheet in a serpin could enhance tissue kallikrein inhibition by two mechanisms. First, these basic residues may direct the reactive center loop of a serpin in a suitable orientation to interact with the reactive crevice of tissue kallikrein by a secondary binding with exosites in the loop around the catalytic center. Second, these basic residues could stabilize the serpin-kallikrein complex formation by electrostatic interactions with cation-binding exosites of tissue kallikrein. The molecular model of the kallistatin-kallikrein complex predicts that Arg 308 /Lys 313 and Lys 307 /Lys 312 may have potential electrostatic interactions with Glu 98C[107] and Asp 61[69] of tissue kallikrein, respectively (Fig. 8). 2 Electrostatic interactions between these acidic residues of kallikrein and kallistatin need to be further investigated to support the hypothesis.
The association of kallistatin with tissue kallikrein can be blocked by heparin (18). The mechanisms by which heparin suppresses the association of kallikrein and kallistatin are not clear. Our study suggests that binding of heparin to the positively charged loop between the H helix and C2 sheet may interfere with the secondary binding of kallistatin with tissue kallikrein. In addition, heparin with negatively charged groups may expel the acidic charged amino acids of tissue kallikrein from association with kallistatin. Moreover, binding of heparin to this region may also generate steric hindrance that blocks the docking of kallislatin's reactive loop into tissue kallikrein's reactive center. This notion is supported by molecular modeling of a three-dimensional structure of kallistatin, which shows that the basic residues in this loop are in close vicinity to the reactive center loop (Fig. 1) (19).
Our previous study showed that kallistatin mutants K307A/ R308A and K312A/K313A displayed reduced heparin binding capacity and increased resistance to heparin inhibition upon interaction with tissue kallikrein (26). In agreement with these results, the ␣1-antitrypsin chimera ␣1-AT/KS/H, which contains an insertion of the homologous heparin-binding site of kallistatin, acquires a heparin-suppressed inhibitory activity toward tissue kallikrein. The findings further confirm the conclusion that the region between the H helix and C2 sheet in kallistatin is a major heparin-binding site responsible for heparin-suppressed proteinase inhibition. In addition, the experiments using chimera ␣1-AT/KS/H demonstrated that the heparin-binding site can be used as a functional unit to convert a non-heparin-binding serpin into a heparin-regulated serpin. The same concept and approach could also be applied on heparin-binding sites in other serpins, such as the D helix in antithrombin, plasminogen activator, and heparin cofactor, to convert a non-heparin-binding serpin to a serpin with heparinregulated activity similar to the donor.
In conclusion, we identified dual roles of the positively charged loop between the H helix and C2 sheet of kallistatin in binding to heparin and in acting as a secondary binding site for tissue kallikrein to accelerate tissue kallikrein inhibition. The success of engineering an ␣1-antitrypsin chimera with heparinregulated activity and enhanced inhibitory activity toward tissue kallikrein provides evidence supporting the importance of the basic residues in the region between the H helix and C2 sheet of kallistatin in tissue kallikrein inhibition. Moreover, the studies using chimera demonstrate that the heparin-binding site or the secondary binding site of kallistatin can be used as a functional module to design inhibitors for tissue kallikrein. These findings provide useful insights for future development of inhibitors for serine proteinases.