Characterization of C1 inhibitor-Ta. A dysfunctional C1INH with deletion of lysine 251.

Dysfunctional C1 inhibitor (C1INH)-Ta is a naturally occurring mutant from a patient with type II hereditary angioedema. This mutant has a deletion of the codon for Lys-251, which is located in the connecting strand between helix F and strand 3A, overlying beta sheet A. Deletion of this Lys modifies the amino acid sequence at this position from Asn-Lys-Ile-Ser to Asn-Ile-Ser and creates a new glycosylation site. To further characterize the mechanism of dysfunction, we have analyzed the recombinant normal and Ta proteins expressed by COS cells in addition to the proteins in serum and isolated from serum. Recombinant C1INH-Ta revealed an intermediate thermal stability in comparison with the intact and reactive center cleaved normal proteins. Analysis of the reactivity of this recombinant protein with target proteases demonstrated no complex formation with C1s, C1r, or kallikrein. Inefficient complex formation was, however, clearly detectable with beta-factor XIIa. Each protease produced partial cleavage of the recombinant mutant inhibitor. Recombinant C1INH-Ta, on 7.5% SDS-polyacrylamide gel electrophoresis and by size fractionation on Superose 12, showed a higher molecular weight fraction that was compatible in size with dimer formation. However, no multimerization of C1INH-Ta isolated from serum or of C1INH-Ta in serum, was observed. The C1INH-Ta dimer expressed the epitopes that normally are expressed only on the protease complexed or the cleaved inhibitor. These epitopes were not expressed on the monomeric inhibitor. The data suggest that the mutation in C1INH-Ta results in a folding abnormality that behaves as if it consists of two populations of molecules, one of which is susceptible to multimerization and one of which is converted to a substrate, but which retains residual inhibitory activity.

Dysfunctional C1 inhibitor (C1INH)-Ta is a naturally occurring mutant from a patient with type II hereditary angioedema. This mutant has a deletion of the codon for Lys-251, which is located in the connecting strand between helix F and strand 3A, overlying ␤ sheet A. Deletion of this Lys modifies the amino acid sequence at this position from Asn-Lys-Ile-Ser to Asn-Ile-Ser and creates a new glycosylation site. To further characterize the mechanism of dysfunction, we have analyzed the recombinant normal and Ta proteins expressed by COS cells in addition to the proteins in serum and isolated from serum. Recombinant C1INH-Ta revealed an intermediate thermal stability in comparison with the intact and reactive center cleaved normal proteins. Analysis of the reactivity of this recombinant protein with target proteases demonstrated no complex formation with C1s, C1r, or kallikrein. Inefficient complex formation was, however, clearly detectable with ␤-factor XIIa. Each protease produced partial cleavage of the recombinant mutant inhibitor. Recombinant C1INH-Ta, on 7.5% SDSpolyacrylamide gel electrophoresis and by size fractionation on Superose 12, showed a higher molecular weight fraction that was compatible in size with dimer formation. However, no multimerization of C1INH-Ta isolated from serum or of C1INH-Ta in serum, was observed. The C1INH-Ta dimer expressed the epitopes that normally are expressed only on the protease complexed or the cleaved inhibitor. These epitopes were not expressed on the monomeric inhibitor. The data suggest that the mutation in C1INH-Ta results in a folding abnormality that behaves as if it consists of two populations of molecules, one of which is susceptible to multimerization and one of which is converted to a substrate, but which retains residual inhibitory activity.
C1 inhibitor (C1INH) 1 belongs to the superfamily of serine protease inhibitors or serpins. It is the sole inhibitor of the complement proteases C1s and C1r and is the major inhibitor of plasma kallikrein and coagulation factor XIIa (1,2). Deficiency of C1INH or inheritance of a dysfunctional mutant molecule results in hereditary angioedema. Inhibition by serpins involves a region of the molecule termed the reactive center loop. Amino acids within this region mimic the substrate of the protease. However, following recognition of the reactive center residue, rather than cleavage, a stable, equimolar inhibitorprotease complex results. Amino acid substitutions within this region either cause loss of inhibitory activity or alter target protease specificity. Approximately 70% of dysfunctional C1INH proteins result from reactive center P1 (Arg-444) mutations (3,4). The role of this region in specificity determination is well known. Several non-reactive center mutations have been described. As expected from analyses of other mutant serpins, replacement of the P2 residue modifies target protease specificity (5)(6)(7). Mutations within the hinge region (P10 -P16) result in conversion of the inhibitor to a substrate (P12, P14, and one P10 mutant) (8 -11); others lead to multimerization of the inhibitor (P10) (12,13). Dysfunctional C1INH-Ta, a naturally occurring mutant from a patient with hereditary angioedema, does not inhibit C1s and C1r (14). It has deletion of Lys-251, which is widely separated, in linear sequence, from the reactive center loop. This residue is located in the connecting strand between helix F and strand 3A, and overlies ␤ sheet A. Deletion of Lys-251 modifies the amino acid sequence at this position from Asn-Asn-Lys-Ile-Ser to Asn-Asn-Ile-Ser, which creates a new glycosylation site. The additional carbohydrate together with the deletion resulted in a slightly higher molecular weight band on SDS-PAGE and an altered electrophoretic mobility on agarose gel electrophoresis in comparison with the normal inhibitor. Enzymatic deglycosylation reduced the normal and Ta mutant inhibitors to the same size but did not restore activity of the mutant Ta inhibitor.
In order to further characterize the potential role of the additional carbohydrate and to analyze the mechanism of dysfunction, we transfected COS-1 cells with expression vectors containing the cDNA for C1INH-Ta and the wild type C1INH and expressed the recombinant proteins in the presence or absence of tunicamycin. We also expressed a new mutant in which Lys-251 was replaced with a Thr; this results in glycosylation at Asn-249. Functional and structural analysis of the recombinant inhibitors indicate that the dysfunction of C1INH-Ta results from the deletion of Lys-251 and not from the additional carbohydrate. Furthermore, the data suggest that this mutant inhibitor behaves as if it consists of two conformers, one of which is converted primarily to a substrate, and one of which is susceptible to multimerization.

MATERIALS AND METHODS
Site-directed Mutagenesis-A polymerase chain reaction-based sitedirected mutagenesis as described previously was performed (6). The mutagenic oligonucleotide primer used has the following sequence, K251T (GAA CAC CAA CAA CAC GAT CAG CC GG CTG CTAGA), which corresponds to nucleotides 8443-8475, with A 3 C transversion at position 8457. This point mutation results in replacement of Lys-251 by a Thr. The presence of the mutation was confirmed by DNA sequence analysis.
Complex Formation of the Recombinant Proteins with Target Proteases-Radiolabeled recombinant C1INH wild type and Ta were incubated alone or with C1s, C1r, ␤-factor XIIa, and kallikrein for 1 h at 37°C. C1r and C1s were generous gifts from Dr. David Bing, Center for Blood Research, Boston, MA. Factor XII and kallikrein were purchased from Enzyme Research Laboratories, Elkhart, IN. The reactions were then stopped by 1 l of 0.1 M phenylmethylsulfonyl fluoride. Triton X-100 (0.5%), deoxycholic acid (0.25%), SDS (0.5%), and EDTA (5 mM) were added to each sample. Rabbit anti-human C1INH antibody (3 l) was then added, and samples were incubated at 4°C overnight. A suspension of 6 l of fixed Staphylococcus aureus (IgG sorb, Enzyme Center, Boston, MA), sonicated and washed three times (with 1% Triton X-100, 1% SDS, 0.5% deoxycholate, and 5 mg/ml bovine serum albumin in 0.02 M sodium phosphate, 0.15 M NaCl, pH 7.4), was added to each sample and incubated at 4°C for 1 h. The pellets (14,000 g for 3 min) were washed once with the above washing solution and three times with a washing solution lacking bovine serum albumin. Each precipitate was then dissolved in SDS non-reducing buffer (15 l) (4.2% SDS, 20% glycerol, 0.1 M Tris/HCl, pH 6.5, and 0.01% bromphenol blue), vortexed, boiled for 3 min, centrifuged briefly, and subjected to electrophoresis on SDS-7.5% polyacrylamide gel. Gels were fixed, dried, and exposed to x-ray film (Kodak XAR-5, Eastman Kodak Co.) at Ϫ70°C.
Thermal Denaturation-The heat stability of the recombinant C1INH was determined as described by Stein et al. (16). COS-1 cell supernatants (100 l) containing normal or C1INH-Ta were incubated at 40, 50, 60, 70, and 80°C for 120 min and then were centrifuged at 14,000 ϫ g for 30 min. The residual antigen in the supernatant was quantified by an ELISA technique as described by Aulak et al. (12).
Gel Filtration-Size fractionation of normal serum and serum from Ta, of C1INH protein purified from each serum, and of each recombinant protein was performed on Superose 12 using an FPLC system (Pharmacia Biotech Inc.). The column was equilibrated in PBS. Fractions of 300 l were collected at a flow rate of 0.5 ml/min.
Immunochemical Assays-C1INH antigen was detected using an ELISA technique. Ninety-six-well plates were coated with goat polyclonal anti-human C1INH diluted 1/1000 in PBS (from Atlantic Antibodies) and incubated overnight at 4°C. Nonspecific binding was inhibited by incubation in 2% bovine serum albumin in PBS for 1 h at 23°C. The bound antigen was then detected with a rabbit polyclonal antihuman C1INH (DAKO) diluted 1/1000 in PBS containing 0.05% Tween 20. These polyclonal antisera recognize all forms of C1INH: native, protease-complexed, and reactive center cleaved. Duplicate samples of bound antigen also were analyzed using two different mouse monoclonal antibodies (4C3 and KOK12) that react only with complexed or cleaved normal C1INH, but do not react with the intact molecule (17,18). Mouse monoclonal antibody 4C3 was a generous gift from Dr. Marc Shapira. Peroxidase-conjugated goat anti-rabbit or anti-mouse IgG antibody (Life Technologies, Inc.) was used to detect binding of the anti-C1INH antibodies, the color was developed using O-phenylenediamine dehydrochloride (Sigma) and H 2 O 2 , and the optical density was read at 492 nm. ELISA using the monoclonal antibody KII, which reacts only with reactive center cleaved C1INH, as the catching antibody was done as described previously (19,20). Microtiter plates were coated with 1 g/ml KII in PBS overnight at 4°C. Blocking was with 3% bovine serum albumin in PBS for 30 -45 min at room temperature. Culture supernatants (100 l) were incubated for 90 min at room temperature, and plates were washed four times with PBS containing 0.1% Tween 20. Adsorbed C1INH was detected with biotinylated rabbit polyclonal anti-C1INH antibody, followed by 30 min with streptavidin-coupled horseradish peroxidase, both diluted 1:1000 in PBS, 0.1% (v/v) Tween 20, 0.2% (w/v) gelatin. Development was with 3,3Ј,5,5Ј-tetramethylbenzidine, and reactions were stopped after 5-10 min with 2 M H 2 SO 4 .

Interaction of Recombinant Wild Type C1INH and C1INH-Ta with Target Proteases-Previous
experiments using enzymatic deglycosylation indicated that the additional carbohydrate was not simply blocking access to the reactive center (14). However, glycosidases do not restore the asparagine res-idue, which could be responsible for the inability to regenerate function. With peptide:N-glycosidase F, an aspartic acid is produced from deglycosylation, while with endoglycosidase F, one N-acetylglucosamine residue remains linked to the asparagine residue. Normal and dysfunctional C1INH were synthesized in the presence of tunicamycin to prevent glycosylation. This had no effect on the function of either protein; normal C1INH remained fully functional,and C1INH-Ta did not complex with C1r or C1s (data not shown).
Lys-251 was replaced with a Thr, using polymerase chain reaction-mediated site-directed mutagenesis (6). Replacement of Lys by Thr creates a new glycosylation site on Asn-249 (Asn 249 -Asn-Lys-Ile-Ser 3 Asn-Asn-Thr-Ile-Ser). The resulting recombinant protein was slightly larger than the wild type recombinant protein, which suggests that this site was glycosylated. It was identical to the wild type inhibitor in its reactivity with target proteases (Fig. 1, A and B).
Analysis of recombinant C1INH-Ta on SDS-PAGE after incubation for 1 h at 37°C with C1s, C1r, or kallikrein revealed no complex formation, but approximately 40 -70% cleavage (Fig. 1, A and B). Cleavage was observed in all experiments, but the precise proportion varied from one experiment to another. In all lanes containing C1INH-Ta, a portion of protein was present with an apparent molecular mass of approximately 200 kDa. This is consistent in size with a dimer of C1INH-Ta. This high molecular mass band was present and unchanged after incubation with C1r, C1s, ␤-factor XIIa, and kallikrein. As opposed to the reactions with the other target proteases, with Each COS-1 cell supernatant (100 l) containing [ 35 S]Met-labeled recombinant C1INH was incubated at 37°C for 1 h either alone or with protease: A, C1s (6.8 g) and C1r (8.4 g); B, kallikrein (15 g) and ␤-factor XIIa (10 g). Samples were immunoprecipitated with polyclonal antiserum to human C1INH and subjected to SDS-PAGE; autoradiography was performed as described under "Materials and Methods." ␤-factor XIIa, a minor but reproducible band was observed that corresponded in size to that of the inhibitor-protease complex (Fig. 1B). Time-course analysis of the interaction of the Ta protein with ␤-factor XIIa revealed complex formation, which was visible by 15 min and appeared to reach a maximum after approximately 1 h (Fig. 2, A and B). In comparison with the ␤-factor XIIa reaction with wild type C1INH, therefore, C1INH-Ta reacts much more slowly. In addition, only a portion of the inhibitor appears capable of participating in complex formation. Time-course analysis of C1INH-Ta with C1s showed maximal cleavage within 30 s (Fig. 2C). This amount did not appear to change over an incubation period of 1 h at 37°C, and no complex formation was observed.
Reactive Center Loop Conformation-Thermal stability of serpins is a measure of their conformation. Natural intact serpins multimerize at 50 -60°C. This multimerization is accompanied by decreased antigenicity. Serpins cleaved near the reactive center are stable (remain monomeric and retain anti-genicity) at elevated temperatures (80 -90°C) (21,22). The recombinant Ta molecule with an intact reactive center loop is more stable at elevated temperatures than the wild type protein, but it is not as stable as the reactive center cleaved molecule (Fig. 3). Cleavage of the reactive center loop with trypsin increases the thermal stability of both the wild type and Ta proteins, indicating that the cleaved mutant molecule undergoes the normal conformational rearrangement specific to serpins. The observation that reactive center loop cleavage of C1INH-Ta enhanced reactivity with the monoclonal antibody KII, which reacts with cleaved but not with native or with complexed C1INH, was consistent with this interpretation (Fig. 4). Reactive center loop cleavage was confirmed in both experiments by SDS-PAGE (data not shown).
Gel Filtration of Wild Type and the Mutant C1INH-Ta-In order to further analyze the apparent dimer formation observed on SDS-PAGE, the recombinant WT and Ta proteins were size-fractionated on Superose 12 (Fig. 5). As pointed out previously, because of its highly glycosylated nature, normal monomeric C1INH elutes earlier than expected on gel filtration, at a position corresponding to a molecular mass of nearly 400 kDa (Figs. 5 and 6) (23,24). When fractions were analyzed with the polyclonal anti-C1INH antibody, recombinant C1INH-Ta eluted in two peaks, one of which corresponded to the elution position of normal C1INH (Fig. 5). The earlier, smaller peak eluted at a position corresponding to a molecular mass of approximately 800 kDa. It is, therefore, possible that this peak represents a dimer of the Ta protein. The thermal stability of the protein within each peak was analyzed and each gave a profile the same as that shown in Fig. 3 for C1INH-Ta (data not shown). Antithrombin Rouen VI, with a mutation within helix F, and thus at a position near that of C1INH-Ta, undergoes spontaneous multimerization and also forms a stable inactive monomeric form after prolonged storage at 4°C or incubation at 41°C (25). This monomeric form probably is equivalent to the L-form of antithrombin or the latent form of plasminogen activator inhibitor-1. Therefore, we attempted to enhance these conformational changes in the Ta protein by incubation for 24 h at 37 and 41°C. No change in the interac-

FIG. 2. Kinetics of complex formation between recombinant wild type (W.T.) C1INH and ␤-factor XIIa (A) C1INH-Ta and ␤-factor XIIa (B), and both recombinant proteins and C1s (C).
Each COS-1 cell supernatant (100 l) containing wild type or Ta proteins were incubated with C1s for 0.5, 2, 5, 15, 30, 60, 120, 180, and 240 min. Samples were immunoprecipitated with polyclonal antiserum to human C1INH and subjected to electrophoresis and autoradiography as described under "Materials and Methods." tion with any of the proteases was observed (data not shown), nor was there any change in the quantity of high molecular weight C1INH-Ta (on either gel filtration or SDS-PAGE analysis). Gel filtration of the mutant Ta protein and the wild type protein in serum or purified from serum showed identical elution profiles, using the polyclonal antiserum as a probe (Fig. 6). Only one peak was observed with C1INH-Ta, and this peak corresponded to the size of the monomer, which was identical in its elution position to the native normal C1INH.
In order to clarify the conformation of recombinant C1INH-Ta and wild type C1INH, reactivity with the monoclonal antibodies 4C3 and KOK12 was analyzed. Only the experiments with KOK12 are illustrated; the reactivities with 4C3 differed in intensity, but otherwise the results were identical to those with KOK12. These monoclonal antibodies detect C1INH that has undergone structural rearrangement due either to cleavage, complex formation, or loop-sheet polymerization (12,17). Both monoclonal antibodies reacted with the higher molecular weight form of the recombinant C1INH-Ta (Fig. 7B). The reaction of each with the monomeric C1INH-Ta, like the reaction with the normal inhibitor, was minimal. This was true even though the amount of monomer present greatly exceeded the amount of the higher molecular weight form, as shown by the reactivity with the polyclonal antiserum. These data indicated that the dimer expresses the epitope that is expressed on complexed normal C1INH. The reactivity of the C1INH-Ta monomer with the monoclonal antibodies, like that of the wild type C1INH, was strikingly increased following cleavage at the reactive center with trypsin (Fig. 7, A and B). DISCUSSION The naturally occurring dysfunctional mutant serpins provide a starting point to help understand the structural requirements and the inhibitory mechanism used by these inhibitors. In this paper, we have characterized several aspects of the mechanism of dysfunction in a variant of C1INH that has deletion of Lys-251, which is situated outside the reactive center region. While the recombinant mutant protein did not form an SDS-stable complex with the target proteases C1s, C1r, and kallikrein, it demonstrated partial susceptibility to cleavage by target proteases. The intermediate thermal stability of the Ta protein together with the presence of a high molecular mass band on SDS-PAGE led to an analysis of potential multimer formation. Multimers were, in fact, detected, and both the monomer and multimer revealed intermediate thermal stability. Multimerization has been observed with several different dysfunctional serpins with mutations in several distinct regions.
The mutant Z ␣ 1 -antitrypsin, with replacement of Glu-342 by a Lys, undergoes spontaneous loop-sheet polymerization (26). This replacement disrupts a critical salt bridge between Glu-342 and Lys-290 at the junction between strand 5A and the proximal end of the reactive center loop. The result therefore FIG. 4. Analysis of the conformational rearrangement in recombinant wild type C1INH and C1INH-Ta using the monoclonal antibody KII. COS-7 cell supernatants (100 l) were incubated in the presence and absence of trypsin (5 g/ml) and tested in serial dilution using an ELISA. The monoclonal antibody KII reacts only with the reactive center cleaved C1INH as described under "Materials and Methods." E, C1INH-Ta; f, wild type C1INH; q, C1INH-Ta ϩ trypsin; Ⅺ, wild type C1INH ϩ trypsin.
FIG. 5. Gel filtration of recombinant C1INH wild type and C1INH-Ta. 200 l of COS-1 cell supernatant containing wild type or Ta recombinant protein were applied to a Superose 12 column at a flow rate of 0.5 ml/min. Fractions (300 l) were collected, and C1INH antigen was determined by ELISA using a polyclonal anti-C1INH antiserum that recognizes all forms of C1INH. Ⅺ, wild type; q, C1INH-Ta.
FIG. 6. Gel filtration of C1INH from normal serum and Ta serum. Purified normal C1INH (0.7 mg/ml) and C1INH purified from Ta serum (0.6 mg/ml) were diluted 1:200 in PBS. Each purified protein (100 l) from serum was applied to a Superose 12 column using a flow rate of 0.5 ml/min. Fractions of 300 l were collected, and C1INH was detected using an ELISA with a polyclonal anti-C1INH antiserum that recognizes all forms of C1INH. Ⅺ, wild type; q, C1INH-Ta.
probably is to open the gap in ␤ sheet A between strands 3 and 5. Initially, it was suggested that this gap provided a receptor site for the reactive center loop of a second antitrypsin molecule. Data from more recent mutants and from a crystallized antithrombin dimer suggest that overinsertion of the reactive center loop may result, and that this disrupts strand 1C (27). The reactive center loop of a second molecule then inserts into the gap in sheet C. Another example is mutant C1INH-Mo with replacement of Ala-436 by a Thr (12). This mutation is located within the hinge region amino-terminal to the reactive center and appears to interfere with the movement of the reactive center loop. The presence of Thr at this position apparently favors stable insertion of the reactive center loop into the A sheet with resulting distortion of either the A or C sheets culminating in loop-sheet polymerization. Another group of mutants that lead to multimerization are those with replacements near the carboxyl terminus, which have been described in antithrombin, ␣ 1 -antitrypsin, and C1INH (23,27). These mutations very likely disrupt the anchoring of the carboxylterminal end of the reactive center loop, resulting in overinsertion of the amino-terminal end of the loop into sheet A. Loopsheet polymerization then follows.
Prolonged incubation of antithrombin Rouen VI at 4°C or incubation at 41°C enhances formation of an inactive uncleaved molecule (the latent form) accompanied with dimer, trimer, or tetramer formation (21). The mutation in antithrombin Rouen VI results in replacement of the conserved Asn-187 with an Asp. This Asn residue is within helix F and is hydrogen-bonded to a carbonyl group within the peptide loop that connects the F helix to strand 3A. The amino acids in this peptide loop make a number of hydrophobic interactions with residues in strand 5A of ␤ sheet A. Therefore, this loop contributes to stabilizing the five-stranded sheet A. In this antithrombin mutant, the link between the F helix and the connecting peptide loop is broken, allowing movement of strands 1A, 2A, and 3A, resulting in opening of the A sheet between S3A and S5A, permitting multimerization.
C1INH-Ta has deletion of Lys-251, which creates a new glycosylation signal. This residue is located within the connecting loop between helix F and strand 3A and is hydrogen-bonded to other conserved residues within helix F and strand 5A. Our results indicate that glycosylation is not directly involved in the dysfunction of the Ta molecule. This is in agreement with previous data using enzymatically deglycosylated protein isolated from plasma and our results with the Lys-251 3 Thr substitution mutant. It is likely that the deletion of this Lys disrupts, directly or indirectly (via helix F), the structure of ␤ sheet A. This probably results in some degree of overinsertion of the reactive center loop, but not to as great an extent as the P10 Ala 3 Thr mutant or the C1INH mutants with substitutions distal to the reactive center loop (14,23). The small amount of dimer formation, the intermediate thermal stability, the susceptibility to cleavage by target proteases, and the lack of expression on the C1INH-Ta monomer of the neoepitope present on the complexed inhibitor are all consistent with this interpretation. Dimer formation was less extensive than with most of the other above described mutants and was not enhanced by incubation at either 37°C or 41°C. In addition, larger multimers were not observed. No multimers were detected in plasma from a patient with this mutation. There are several potential explanations for this apparent absence. It is possible that the multimers in plasma are simply below the detection limits of the ELISA. However, this seems unlikely because the plasma used contained similar amounts of mutant and normal protein (14). It seems most likely that the multimers are rapidly cleared from the circulation in vivo. C1INHprotease complexes have a rapid in vivo clearance rate (28,29). Since the C1INH-Ta multimer shares several characteristics with the inhibitor-protease complex, it may be rapidly cleared by a similar mechanism. It also is possible that multimers form intracellularly in vivo and, unlike with COS cells, are not secreted or that some other undefined mechanism in plasma prevents formation of the multimer. It also is theoretically possible that the multimers represent an artifact of expression of C1INH-Ta in COS cells. However, if this is true, it is specific for this mutant and therefore indicates that its conformation is different from that of the wild type protein.
Thermal denaturation analysis showed that the mutant protein was more stable at elevated temperature than the wild type. However, it was not as stable as the P10 Ala 3 Thr mutant or as reactive center cleaved wild type C1INH. Cleavage of the reactive center loop with trypsin increases the thermal stability of C1INH-Ta to a degree similar to that of the cleaved wild type protein. This further indicates that the cleavage of the mutant inhibitor results in complete insertion of the released reactive center loop. This susceptibility to cleavage by trypsin and by target proteases contrasts with the lack of FIG. 7. Gel filtration of recombinant wild type C1INH (A) and C1INH-Ta (B) after reactive center cleavage by trypsin. Each COS-1 cell supernatant (200 l) was incubated in the presence or absence of trypsin (5 g) and was applied to a Superose 12 column at a flow rate of 0.5 ml/min. Fractions (300 l) were collected and C1INH antigen was detected by ELISA using the monoclonal antibody, KOK12, which reacts only with the cleaved or complexed form of the inhibitor. E, C1INH-Ta; q, C1INH-Ta ϩ trypsin; Ⅺ, wild type C1INH; f, wild type C1INH ϩ trypsin. susceptibility of the P10 Ala 3 Thr mutant, of antithrombin Rouen and of latent plasminogen activator inhibitor (12,25,30). Therefore, the reactive center must be exposed in the Lys-251 deletion mutant as it is in the wild type inhibitor and in the P12 Ala 3 Glu and P14 Val 3 Glu mutant monomer. Finally, the lack of expression of the neoepitope on the monomer indicates the lack of complete insertion. Appearance of the neoepitope on the cleaved molecule confirms that the mutant is able to undergo the normal conformational rearrangement.
The fact that C1INH-Ta complexed with ␤-factor XIIa indicates that the molecule has a reactive center that can be recognized by protease and is capable of forming a complex, albeit inefficiently. Both the rate and extent of complex formation were considerably diminished compared with the ␤-factor XIIawild type C1INH reaction. C1INH-Ta, therefore, clearly is capable of undergoing the conformational rearrangements required for complex formation. It also can be concluded that the functional changes in C1INH-Ta are more subtle than in many of the other mutants that lead to loop-sheet polymerization and/or conversion of the inhibitor to a substrate. The only other mutant of this type that retains significant activity is antithrombin Rouen VI, which has a mutation within helix F, very near the site of the deletion in C1INH-Ta. However, the minimal activity retained by C1INH-Ta almost certainly is not physiologically relevant. Patients with hereditary angioedema express one normal C1INH gene. The small amount of activity contributed by this dysfunctional mutant would be insignificant in comparison with that contributed by the product of the normal allele.
In conclusion, the data presented here suggest that deletion of Lys-251 results in aberrant folding of the mutant molecule that results in two populations of molecules. One of these is probably characterized by loop overinsertion to the extent that it resembles a latent form. This form is susceptible to multimerization and is not recognized by target proteases. It also, at least in the dimerized form, expresses epitopes normally present on the complexed form of the protein. The other form remains in a conformation recognized by target proteases and by trypsin. This is probably the form capable of complexing with ␤-factor XIIa. It does not express the epitopes present on the protease-complexed normal molecule. Although there may be some degree of loop insertion, appropriate insertion to allow efficient complex formation appears to be prevented, or the rate of loop insertion may be slowed. This combination of characteristics (substrate-like behavior, multimerization, and retention of residual ability to complex) is unusual and perhaps unique. The analysis of this inhibitor illustrates the fine line between function and dysfunction among the serpins.