Intermolecular autocatalytic activation of serine protease zymogen factor C through an active transition state responding to lipopolysaccharide

Horseshoe crab hemolymph coagulation is believed to be triggered by the autocatalytic activation of serine protease zymogen factor C to the active form, α-factor C, belonging to the trypsin family, through an active transition state of factor C responding to bacterial lipopolysaccharide (LPS), designated factor C*. However, the existence of factor C* is only speculative, and its proteolytic activity has not been validated. In addition, it remains unclear whether the proteolytic cleavage of the Phe737–Ile738 bond (Phe737 site) of factor C required for the conversion to α-factor C occurs intramolecularly or intermolecularly between the factor C molecules. Here we show that the Phe737 site of a catalytic Ser-deficient mutant of factor C is LPS-dependently hydrolyzed by a Phe737 site–uncleavable mutant, clearly indicating the existence of the active transition state of factor C without cleavage of the Phe737 site. Moreover, we found the following facts using several mutants of factor C: the autocatalytic cleavage of factor C occurs intermolecularly between factor C* molecules on the LPS surface; factor C* does not exhibit intrinsic chymotryptic activity against the Phe737 site, but it may recognize a three-dimensional structure around the cleavage site; and LPS is required not only to complete the substrate-binding site and oxyanion hole of factor C* by interacting with the N-terminal region but also to allow the Phe737 site to be cleaved by inducing a conformational change around the Phe737 site or by acting as a scaffold to induce specific protein–protein interactions between factor C* molecules.

Horseshoe crab hemolymph coagulation is believed to be triggered by the autocatalytic activation of serine protease zymogen factor C to the active form, ␣-factor C, belonging to the trypsin family, through an active transition state of factor C responding to bacterial lipopolysaccharide (LPS), designated factor C*. However, the existence of factor C* is only speculative, and its proteolytic activity has not been validated. In addition, it remains unclear whether the proteolytic cleavage of the Phe 737 -Ile 738 bond (Phe 737 site) of factor C required for the conversion to ␣-factor C occurs intramolecularly or intermolecularly between the factor C molecules. Here we show that the Phe 737 site of a catalytic Ser-deficient mutant of factor C is LPS-dependently hydrolyzed by a Phe 737 site-uncleavable mutant, clearly indicating the existence of the active transition state of factor C without cleavage of the Phe 737 site. Moreover, we found the following facts using several mutants of factor C: the autocatalytic cleavage of factor C occurs intermolecularly between factor C* molecules on the LPS surface; factor C* does not exhibit intrinsic chymotryptic activity against the Phe 737 site, but it may recognize a three-dimensional structure around the cleavage site; and LPS is required not only to complete the substrate-binding site and oxyanion hole of factor C* by interacting with the N-terminal region but also to allow the Phe 737 site to be cleaved by inducing a conformational change around the Phe 737 site or by acting as a scaffold to induce specific protein-protein interactions between factor C* molecules.
The molecular mechanism underlying the proteolytic activation of serine protease zymogens has been established in trypsinogen, which is activated by the activator enteropeptidase through limited proteolysis of the Arg 15 -Ile 16 peptide bond in chymotrypsinogen numbering. The limited proteolysis induces the insertion of the newly appearing N-terminal Ile 16 into the activation pocket known as the Ile 16 cleft to form a salt bridge between the ␣-amino group of Ile 16 and the ␤-carboxyl group of Asp 194 . The salt bridge results in conformational changes in the substrate-binding site and the oxyanion hole to hydrolyze specific peptide bonds of substrates, whereas the conformational changes elsewhere in the molecule, including the catalytic triad of Ser 195 , His 57 , and Asp 102 , are very small (1)(2)(3).
Factor C is a serine protease zymogen involved in the hemolymph coagulation cascade of horseshoe crabs and is autocatalytically activated to ␣-factor C on bacterial LPS 4 (4 -6). The resulting ␣-factor C activates coagulation factor B to activated factor B (7,8), which activates the proclotting enzyme to the clotting enzyme (9,10) to convert coagulogen into coagulin gel (11). Alternatively, factor G autocatalytically activated in the presence of ␤-1,3-D-glucans directly activates the proclotting enzyme to the clotting enzyme (6). Factor C is also located on hemocytes as a pattern recognition receptor for LPS, and the resulting ␣-factor C triggers hemocyte exocytosis through a protease-activated G protein-coupled receptor to induce Ca 2ϩ signaling (12). On the other hand, homologs of mammalian complement factors C3 and B/C2 have been identified in hemolymph of horseshoe crabs (13), and ␣-factor C also acts as the complement C3 convertase on microbes in close cooperation with several lectins (14 -16).
Factor C is biosynthesized as a single-chain form of the zymogen containing six N-linked glycosylation sites, whereas a two-chain form of the zymogen of the H and L chains produced by cleavage of the Arg 665 -Ser 666 bond by an unknown protease is principally purified from hemocytes (6,(17)(18)(19) (Fig. 1). An LPS-binding site is located in the N-terminal Cys-rich domain, and a tripeptide sequence of Arg 36 -Trp 37 -Arg 38 in this domain is essential for LPS binding (20). Based on its amino acid sequence, ␣-factor C belongs to the trypsin family and exhibits trypsin-like amidase activity against a peptide substrate for ␣-thrombin, t-butoxycarbonyl (Boc)-Val-Pro-Arg-p-nitroanilide (pNA), or against Boc-Val-Pro-Arg-4-methylcoumaryl-7-amide but not against a peptide substrate for chymotrypsin, succinyl-Ala-Pro-Phe-4-methylcoumaryl-7amide (8). Autocatalytic activation of factor C bound to LPS occurs through proteolytic cleavage of the Phe 737 -Ile 738 bond (Phe 737 site) (Fig. 1), corresponding to the Arg 15 -Ile 16 bond in chymotrypsinogen numbering (21). It remains unknown how factor C* displays specific chymotryptic activity against the Phe 737 site despite its general trypsin-like primary substrate specificity. Previously, we prepared a recombinant factor C with an N-glycan of Man 5 GlcNAc 2 at each N-glycosylation site, expressed in an HEK293S mutant cell line lacking N-acetylglucosaminyltransferase I (GnTI Ϫ ) (22). In the absence of LPS, factor C is artificially activated by chymotrypsin through cleavage of the Phe 737 site (23). Using WT factor C, we found that chymotrypsin-activated factor C contains an additional proteolytic cleavage of the Tyr 40 -Cys 41 bond in the N-terminal Cysrich domain, designated ␤-factor C ( Fig. 1) (22). The resulting ␤-factor C exhibits amidase activity against the synthetic substrates for ␣-thrombin with ϳ70% specific activity compared with that of ␣-factor C but has neither the LPS-binding activity nor the activating activity for factor B (22). These data suggest that LPS interaction with ␣-factor C is required to maintain its proteolytic activity against factor B. Interestingly, the proteolytic conversion of factor B to activated factor B occurs through cleavage of the Ile 126 -Ile 127 bond (7); in addition, factor B is also an LPS-binding protein, and the LPS-bound form of factor B is essentially required for its proteolytic activation by ␣-factor C (8). Factor C*, however, is only speculative, and its proteolytic activity has not been validated. Moreover, it remains unclear whether the autocatalytic cleavage at the Phe 737 site occurs intramolecularly within factor C bound to LPS or intermolecularly between the molecules of factor C. Here we show that the autocatalytic activation of factor C is the unidentified intermolecular event between the factor C* molecules on the LPS surface.

Replacement of the active-site Ser 941 to Ala prevents autocatalytic activation of factor C
WT factor C prepared in the HEK293S GnTI Ϫ cell line is the two-chain form of the zymogen consisting of the H and L chains (22). The autocatalytic activation occurs through cleavage of the Phe 737 site of the L chain to be converted into the A and B chains ( Fig. 1) (19, 21). Therefore, the proteolytic activation of factor C was followed by appearance of the B chain, which was detected by using a polyclonal antibody against the B chain. The L chain on Western blotting was observed in a doublet because the fourth N-glycosylation site located in the A chain is partially modified (Fig. 1, open diamond) (8,22).
To ensure that the LPS-dependent autocatalytic cleavage reaction of factor C was accurately quantified by Western blotting under the conditions used, WT factor C was incubated at varying concentrations from 12.5 to 100 nM for 30 min at 37°C in the presence or absence of LPS and subjected to Western blotting. WT factor C was autocatalytically cleaved in the presence of LPS but not in the absence of LPS ( Fig. 2A). The relative band density of Western blotting by densitometric analysis was in direct proportion to the concentration of factor C under these conditions (Fig. 2B). After 30-min incubation with LPS, factor C at every concentration was cleaved by ϳ80% under these conditions (Fig. 2C). These results indicate that the LPSdependent autocatalytic cleavage was quantified by the densitometric analysis of Western blotting, at least under the conditions used.
To determine whether the catalytic serine residue in the protease domain of factor C is involved in the autocatalytic activation, the Ser 941 residue of factor C, corresponding to Ser 195 in chymotrypsinogen numbering, was substituted to Ala (S941A). WT factor C or the S941A mutant was incubated at 37°C in the presence of LPS and subjected to Western blotting (Fig. 3, A  and B). The autocatalytic cleavage rate was quantitated densitometrically (Fig. 3C). As expected, the L chain of WT factor C was converted to the B chain by ϳ80% within 30 min under the conditions employed (Fig. 3, A and C), but the appearance of the B chain of the S941A mutant was not detected (Fig. 3, B and C).
These results indicate that the Ser 941 residue of factor C plays a key role in the LPS-dependent autocatalytic cleavage of the Phe 737 site.

Factor C* does not exhibit intrinsic chymotryptic activity against the Phe 737 site
To examine whether factor C* recognizes the hydrophobic side chain of the Phe 737 site, the Phe 737 residue was substituted to Ala (F737A), Glu (F737E), Arg (F737R), or Pro (F737P). In addition, to examine whether the side chain of Ile 738 affects the An active transition state of zymogen factor C cleavage reaction, the Ile 738 residue was substituted to Ala (I738A). All of the mutants, except the F737R mutant, were expressed in HEK293S GnTI Ϫ cells, but the expression level of the F737R mutant was too low for purification.
LPS-dependent autocatalytic cleavage reactions for the P1 mutants (F737A, F737E, and F737P) and the P1Ј mutant (I738A) were performed using a concentration of 50 nM for each mutant within the range of the linearity of the densitometric analysis of Western blotting, as shown in Fig. 2 for WT factor C. Each of the mutants was incubated at 37°C in the presence of LPS and subjected to Western blotting with anti-B chain antibody. All of the mutants, except the F737P mutant, were autocatalytically cleaved in the presence of LPS (Fig. 4, A-D), and more than 70% of the L chain of each mutant was converted into the B chain by the 30-min incubation (Fig. 4E). Interestingly, the F737A mutant was more efficiently cleaved autocatalytically than WT factor C (Fig. 4E). These results suggest that factor C* does not exhibit an intrinsic chymotryptic activity against the Phe 737 site and that the side chain of Ile 738 has little effect on the proteolytic cleavage of the Phe 737 site.
The cleaved form of the F737A mutant exhibited high levels of specific amidase activities against Boc-Val-Pro-Arg-pNA, comparable with the levels of WT ␣-factor C, and the cleaved form of the F737E mutant also had sufficient amidase activity (Fig. 4F). In contrast, the cleaved form of the I738A mutant, corresponding to Ile 16 in chymotrypsinogen numbering, exhibited no amidase activity (Fig. 4F), indicating an essential interaction of Ile 738 with the Ile 16 cleft to induce conversion of the zymogen form to the active form. These data support the previous report of trypsin recombinants with Ile 16 mutations, demonstrating that the hydrophobic interaction of the Ile 16 side chain is the primary force to stabilize the substrate-binding site and the oxyanion hole rather than the electrostatic interaction of the Asp 194 -Ile 16 salt bridge (24). As expected, the individual incubation of the F737P mutant or S941A mutant did not result in any detectable amidase activity against the synthetic substrate in the presence of LPS (Fig. 4F).

Evidence for the existence of factor C*, the active transient state of zymogen factor C, without cleavage of the Phe 737 site
The F737P mutant prevented the cleavage reaction at the Phe 737 site in the autocatalytic activation ( Fig. 4, C, E, and F), whereas its Ser 941 residue remained intact. On the other hand, the S941A mutant lost its authentic serine protease activity (Figs. 3B and 4F), whereas its Phe 737 site remained intact. If the autocatalytic cleavage occurs intermolecularly between the factor C* molecules but not intramolecularly within individual molecules of factor C*, then the Phe 737 site of the S941A mutant  An active transition state of zymogen factor C could be cleaved by the proteolytic activity of the active transition state of the F737P mutant bound to LPS.
To examine this hypothesis, the F737P and S941A mutants were mixed and incubated for 30 min at different concentration ratios in the presence of LPS, and the aliquots were subjected to Western blotting. First, under conditions of a fixed concentration of the S941A mutant at 50 nM and increasing concentrations of the F737P mutant from 1 to 5 nM, the autocatalytic cleavage rate increased as the concentrations of the F737P mutant increased (Fig. 5, A, left, and B, left). Next, under conditions of a fixed concentration of the F737P mutant at 50 nM and increasing concentrations of the S941A mutant from 1 to 5 nM, the autocatalytic cleavage reaction was not observed (Fig. 5,  A, right, and B, right), indicating that the Phe 737 site of the S941A mutant is cleaved by the F737P mutant bound to LPS (F737P*). These results also indicate the presence of the active transition state of factor C without cleavage of the Phe 737 site.

Preparation of a factor C recombinant containing an epitope tag in an internal portion of the B chain
To obtain further evidence for the intermolecular event of autocatalytic activation, a newly developed epitope tag system, PA tag, which is a dodecapeptide derived from human podoplanin and recognized by a high-affinity mAb, NZ-1 (anti-PA antibody) (25), was introduced into the production of factor C mutants. Addition of a peptide or an epitope tag to the N terminus of WT factor C inhibits the LPS-dependent autocatalytic

An active transition state of zymogen factor C
activation of factor C (22). In contrast, with the PA tag system it is possible to insert the PA tag into internal portions of a recombinant protein or boundary portions between the domains in a mosaic protein to maintain not only the antigenicity of the PA tag but also the physiologic activities of the tag-containing proteins (26). Accordingly, the PA tag was inserted into the peptide bond between Trp 758 -Leu 759 in the B chain of WT factor C (PA factor C), corresponding to the linker position between the two ␤-barrel structures composing the catalytic domain of serine proteases (1-3).
To examine the time-dependent cleavage reaction, PA factor C or WT factor C was incubated at 37°C in the presence of LPS and then subjected to Western blotting with an anti-B chain antibody. The LPS-dependent autocatalytic cleavage of PA fac-tor C occurred in correlation with that of WT factor C (Fig. 6, A  and B). On the other hand, the autocatalytically activated PA factor C exhibited ϳ30% lower amidase activity against the synthetic peptide substrate compared with that of the WT ␣-factor C (Fig. 6C), indicating that insertion of the PA tag at this position partially inhibits the amidase activity of ␣-factor C.

Further evidence of intermolecular cleavage at the Phe 737 site in autocatalytic activation
The LPS-dependent autocatalytic conversion of PA factor C was monitored on Western blotting using the anti-PA antibody (Fig. 7A); the results corresponded to those by Western blotting with the anti-B chain antibody (Fig. 6A). To confirm that the Phe 737 site of the S941A mutant is cleaved by the F737P mutant bound to LPS (Fig. 5), the PA tag was inserted into the Trp 758 -Leu 759 bond of the F737P and S941A mutants, yielding the PA-F737P and PA-S941A mutants, respectively. As anticipated, the resulting PA-F737P and PA-S941A mutants were not autocatalytically converted by individual incubation in the presence of LPS (Fig. 7, A and B). Then, the PA-F737P and S941A mutants were mixed at a 1:1 ratio, incubated in the presence of LPS for 30 min, and subjected to Western blotting with anti-PA antibody. As expected, the L chain of the PA-F737P mutant was not converted to the B chain (Fig. 7C, center lane). Next, Western blotting experiments were carried out using a 1:1 mixture of the PA-S941A and F737P mutants in the presence of LPS, and the B chain derived from the PA-S941A mutant was detected by anti-PA antibody (Fig. 7C, right lane), clearly indicating that the Phe 737 site of the PA-S941A mutant is intermolecularly cleaved by the F737P mutant bound to LPS.

LPS is also required to allow the Phe 737 site to be cleaved by inducing a conformational change or by acting as a scaffold to induce interactions between factor C* molecules
The N-terminal tripeptide of Arg 36 -Trp 37 -Arg 38 of factor C is essential for LPS recognition, and the replacement of both Arg 36 and Arg 38 to Glu (the RE factor C mutant (factor C mutants with the replacement of both Arg 36 and Arg 38 to Glu)) causes the loss of LPS-binding activity (20). Previously, we found that incubation of factor C with LPS in the presence of the chemical cross-linker dimethyl adipimidate induces autocatalytic activation of factor C through the formation of a dimer or its multiple species of factor C, resulting in chemical crosslinking of factor C molecules (22). However, it remains unclear whether LPS-binding to the N-terminal region of factor C is also required to allow other factor C* molecules to undergo protein-protein interactions and thereby cleave the Phe 737 site. To answer this question, three types of LPS-binding activitydeficient mutants, RE-factor C, RE-F737P, and RE-S941A mutants, were prepared. These mutants were incubated for 30 min individually or in a 1:1 combined mixture in the presence of LPS and subjected to Western blotting with anti-B chain antibody. The autocatalytic conversion of WT factor C was ϳ80% under the conditions employed (Fig. 8A, lane 1, and B, column  1), whereas that of the 1:1 mixture of WT factor C and the RE-factor C mutant was reduced to ϳ40%, suggesting that the RE-factor C mutant is not involved in LPS-dependent autocatalytic activation (Fig. 8A, lane 3, and B, column 3). The RE-fac- An active transition state of zymogen factor C tor C mutant was not autocatalytically cleaved when incubated individually (Fig. 8A, lane 2, and B, column 2).
The mixture of the S941A and Phe737P mutants in the presence of LPS yielded an autocatalytic conversion of ϳ35% (Fig.  8A, lane 4, and B, column 4) because the Phe 737 site of the S941A mutant was cleaved by the F737P mutant bound to LPS. In contrast, the 1:1 mixture of the S941A and RE-F737P  An active transition state of zymogen factor C mutants in the presence of LPS did not yield the B chain (Fig.  8A, lane 5, and B, column 5). Moreover, the 1:1 mixture of the RE-S941A and F737P mutants in the presence of LPS also did not yield the B chain (Fig. 8A, lane 6, and Fig. 8B, column 5).
These results suggest that LPS is required not only to complete the substrate-binding site and oxyanion hole of factor C* by interacting with the N-terminal region but also to allow the Phe 737 site to be cleaved by inducing a conformational change around the Phe 737 site or by acting as a scaffold to induce specific protein-protein interactions between factor C* molecules.
There is a possibility that the replacement of both Arg 36 and Arg 38 to Glu impedes the expression of the enzymatic activity of the catalytic domain of the RE-factor C mutant. To confirm that the catalytic domain of the RE-factor C mutant is functional, the RE-factor C mutant was incubated with chymotrypsin. The resulting chymotrypsin-treated RE-factor C mutant (␤-RE-factor C) exhibited a level of specific amidase activity against Boc-Val-Pro-Arg-pNA equivalent to that of ␤-factor C, a chymotrypsin-treated WT factor C, indicating that the catalytic domain of ␤-RE-factor C remains intact (Fig. 8C).

Comparison of the relative contributions of ␣-factor C and factor C* to autocatalytic activation
To compare the relative contributions of ␣-factor C and factor C* to the LPS-dependent autocatalytic activation, ␣-factor C and factor C* were prepared by individually incubating WT factor C and the F737P mutant for 30 min at 37°C in the presence of LPS. An aliquot of the resulting ␣-factor C or F737P* (factor C*) was incubated with the PA-S941A mutant for the indicated duration at 37°C and subjected to Western blotting with anti-PA antibody (Fig. 9A). Interestingly, the densitometric analyses of Western blotting suggest that there was no significant difference in the degree of contribution between ␣-factor C and factor C* to the autocatalytic cleavage rate of the PA-S941A mutant (Fig. 9B).

Discussion
Proteolytic cascades in blood/hemolymph coagulation involve serine protease zymogens and are triggered through the interaction of initiator zymogens with biologic substances derived from host tissues or microbes to amplify and propagate biologic reactions (27)(28)(29)(30). In the horseshoe crab coagulation cascade, the initiator zymogen factor C is thought to be triggered by autocatalytic activation through the active transition state of factor C (factor C*) bound to LPS. However, factor C* is only speculative, and it remains unclear whether the proteolytic cleavage of the Phe 737 site of factor C required for the conversion to ␣-factor C occurs intramolecularly or intermolecularly between the factor C molecules. As for the mammalian extrinsic coagulation pathway, the catalytic activity of factor VIIa proteolytically converted from zymogen factor VII by factor Xa is accelerated by the interaction with the cofactor protein tissue factor to form a 1:1 complex on phospholipids in the presence of Ca 2ϩ (31). Higashi et al. (32) proposed a molecular model in which a tissue factor traps one of the conformational states of factor VIIa, converting the zymogen state into an active transition state.
Here we used several mutants of factor C to show that the Phe 737 site of the S941A mutant was cleaved LPS-dependently by the F737P mutant (Fig. 5), clearly indicating the presence of the active transition state of factor C with an active conformation induced by the interaction with LPS without cleavage of the Phe 737 site. Interestingly, the autocatalytic cleavage rate increased in proportion to the concentration of the F737P mutant when the concentration of the S941A mutant was kept constant (Fig. 5), indicating multiple sequential catalytic turnovers of the active transition state of the F737P mutant bound to LPS, F737P*. Moreover, the LPS-dependent proteolytic conversion of the PA-S941A mutant was clearly detected by anti-PA antibody in the mixture of the F737P and PA-S941A mutants, indicating that the Phe 737 site of the PA-S941A mutant is cleaved by F737P* (Fig. 7). Therefore, the autocatalytic activation of factor C occurs intermolecularly between the factor C* molecules but not intramolecularly.
The results of our previous studies suggested that factor C* may exhibit intrinsic chymotryptic activity to cleave the Phe 737 site despite being a member of the trypsin family (21,22). Unexpectedly, as shown in Fig. 4, the two factor C mutants with the replacement of the Phe 737 site, the F737A and F737E mutants, were autocatalytically activated with an efficient conversion rate comparable with that of WT factor C but the F737P mutant was not, indicating that factor C* does not have Figure 9. The degree of contribution of ␣-factor C and factor C* to autocatalytic activation. A, WT factor C or the F737P mutant (500 nM) was incubated for 30 min at 37°C in the presence of 6.8 M LPS to prepare ␣-factor C and factor C*. An aliquot of the resulting ␣-factor C or factor C* (50 nM) was incubated with the PA-S941A mutant (50 nM) for the indicated duration at 37°C and subjected to Western blotting with anti-PA antibody. Data are representative of three independent experiments. B, the autocatalytic cleavage rate is shown as the relative band density of the B chain to the L chain and B chain, as analyzed by ImageJ software. Error bars indicate Ϯ S.E. (n ϭ 3).

An active transition state of zymogen factor C
intrinsic chymotryptic activity against the Phe 737 site. To the best of our knowledge, there are no other examples of serine proteases targeting a specific substrate with a P1 site distinct from their characterized primary substrate specificities. Factor C* may recognize a three-dimensional structure around the Phe 737 site in the case of the LPS-dependent autocatalytic activation.
As established in trypsinogen, the proteolytic activity in serine protease zymogens can be switched on by highly localized conformational changes to complete the substrate-binding site and oxyanion hole. These changes are induced by the cleavage of the Arg 15 -Ile 16 bond, leading to salt bridge formation between the Ile 16 and Asp 194 residues (1-3). It has been proposed that salt bridge formation at the Ile 16 cleft is important for the nonproteolytic activation of prothrombin by complex formation with staphylocoagulase (33)(34)(35)(36). Staphylocoagulase, a staphylococcal nonenzymatic protein, interacts with prothrombin to form a 1:1 complex to convert the zymogen state of prothrombin into a thrombin-like active state; the N-terminal portion (Ile 1 -Val 2 -) of staphylocoagulase is inserted into the Ile 16 cleft of bound prothrombin to form a salt bridge between the ␣-amino group of Ile 1 of staphylocoagulase and the ␤-carboxyl group of Asp 194 of prothrombin, resulting in the active transition state without proteolytic cleavage of the Arg 15 -Ile 16 bond of prothrombin. We clearly identified the existence of factor C*, the active transition state of factor C without cleavage of the Phe 737 site, because the Phe 737 site of the S941A or PA-S941A mutant was LPS-dependently cleaved by a Phe 737 site-replacing mutant, the F737P mutant bound to LPS (Figs. 3 and 5).
The N-terminal Arg of factor C plays an important role in protein-protein interactions between the factor C molecules to form the dimer or its multimers on LPS for autocatalytic activation; a factor C mutant with a replacement of Arg 1 with Lys, but not replacement of Arg 1 with other amino acids residues, including Ala, Leu, or His, maintains LPS-dependent autocatalytic activation, suggesting that the positive side chain of Arg 1 or Lys 1 is essential for inducing LPS-dependent autocatalytic activation (22). Interestingly, unlike other serine protease zymogens, tissue-type plasminogen activator (tPA) is proteolytically active in a single-chain form without cleavage of the Arg 15 -Ile 16 bond in chymotrypsinogen numbering because the ⑀-amino group of Lys 156 located in the catalytic domain of tPA forms a salt bridge with the ␤-carboxyl group of Asp 194 , promoting an active conformation in the single-chain form of tPA to maintain its low zymogenicity (37). Also, in the oligomeric factor C on the LPS surface, a salt bridge may form between the basic side chain of Arg 1 or Lys 1 and the side chain of Asp 194 to promote the active transition state of factor C without proteolytic cleavage of the Phe 737 site.
Based on the above findings and our present results, we propose the following model for autocatalytic activation and initiation of hemolymph coagulation by factor C* on the LPS surface derived from Gram-negative bacteria (Fig. 10). Factor C, which is secreted by LPS-induced hemocyte exocytosis, binds to LPS through the tripeptide motif of the Arg 36 -Trp 37 -Arg 38 sequence in the Cys-rich domain to cause the conformational changes in factor C, resulting in the active transition state, factor C*. Factor C* forms the dimer or its multimers on LPS, and autocatalytic activation occurs intermolecularly between the factor C* molecules in the complex. The resulting ␣-factor C in the complex may be exchangeable with noncomplexed factor C on LPS.
Homologs of mammalian complement factors C3 and B/C2 were identified in horseshoe crabs (13)(14)(15)(16), and an arachnid homolog of complement factor B/C2 was also identified in the spider Loxosceles laeta (38). Moreover, a homolog of horseshoe crab factor C was reported from the tick Ixodes ricinus and may play a role in the tick primordial complement system (39). Therefore, the nomenclatures for the zymogens involved in the horseshoe crab coagulation cascade, including factor C, factor B, and factor G, are becoming confused because the identical or similar terms are used for the zymogens in the complement system in Chelicerata. Therefore, we propose new terms for three protease zymogens in the horseshoe crab coagulation cascade: prochelicerase C, prochelicerase B, and prochelicerase G, each of which is activated into the corresponding chelicerase in the proteolytic cascade (Fig. 11).
A Limulus reagent prepared from hemocyte lysates of horseshoe crabs is used in a sensitive assay to detect LPS. The use of this Limulus test has increased dramatically along with advancements to monitor LPS contamination in parenteral drugs, medical devices, and biologics (40,41). However, the Limulus test is totally dependent on the limited natural resource of horseshoe crab hemocytes. As an alternative approach, we recently developed a next-generation Limulus test using recombinant coagulation factors (42). We are convinced that further detailed functional studies of recombinant coagulation factors will contribute to the development of more sensitive and convenient assays for LPS.

Materials
The HEK293S GnTI Ϫ cell line was obtained from the ATCC. LPS derived from Salmonella minnesota R595 (Re) (M r ϭ ϳ2,500) was purchased from List Biological Laboratories (Campbell, CA) and used for the factor C activation assays. Boc-Val-Pro-Arg-pNA was provided by Seikagaku (Tokyo). The polyclonal antibody against the B chain of factor C was prepared previously (8).

Mutagenesis
The pCA7 plasmid encoding the full-length of factor C, which contains a six-histidine tag and a cleavage site of factor Xa at the N-terminal end, was used for expression and mutagenesis (8). Mutations were introduced using site-directed mutagenesis by inverse PCR. The PA tag (25) was inserted into the plasmid at the site between Trp 758 and Leu 759 of factor C with the In-Fusion HD Cloning Kit (Takara Bio, Shiga, Japan).

Expression of recombinant proteins
Recombinant proteins were expressed as described previously (8,22). In brief, the recombinant proteins were expressed in HEK293S GnTI Ϫ cells and secreted into culture medium. HEK293S GnTI Ϫ cells were maintained in Dulbecco's modified Eagle's medium supplemented with glutamine, penicillinstreptomycin, and 10% fetal bovine serum at 37°C under 5% CO 2 . Dulbecco's modified Eagle's medium supplemented with plasmids containing inserts of the recombinant proteins (1.8 g/ml), polyethyleneimine (2.7 g/ml), glutamine, penicillinstreptomycin, and 2% fetal bovine serum was used to transfect HEK293S GnTI Ϫ cells at 80 -90% confluence. Culture media were collected 5 days after transfection and centrifuged at 6,000 rpm for 30 min.

Purification of recombinant proteins
Recombinants were purified as described previously (8,22). In brief, collected culture media containing factor C recombi-nants were applied to a nickel-nitrilotriacetic acid-agarose column. The buffer of eluates was exchanged, treated with factor Xa, and further applied to a nickel-nitrilotriacetic acid-agarose column to remove the free histidine tag. The concentrations of recombinants were determined using the extinction coefficient of 1% solution at A 280 nm of 21.3 (4) or a Micro BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA).

Autocatalytic activation and assay for amidase activity
For autocatalytic activation, factor C recombinants were incubated with 0.68 M (1.7 g/ml) LPS at 37°C in 20 mM Tris-HCl (pH 8.0) containing 300 mM NaCl and subjected to Western blotting with anti-B chain antibody or anti-PA antibody (Wako, Osaka, Japan). In the case of amidase activity against Boc-Val-Pro-Arg-pNA, factor C recombinants were incubated with 0.68 M LPS for 30 min at 37°C in 20 mM Tris-HCl (pH 8.0) containing 300 mM NaCl. One microliter of each solution was diluted by 19 l of 20 mM Tris-HCl (pH 8.0] containing 150 mM NaCl and 1.5 M BSA, and then 5 l of 2 mM Boc-Val-Pro-Arg-pNA was added. After incubation for 5 min at 37°C, the reaction was stopped by addition of 75 l of 0.6 M acetic acid, and the liberated pNA was measured by absorbance at 405 nm. One unit of amidase activity was defined as 1 mol of digested substrate per minute, and the specific activity was expressed as units per nanomole of factor C.

Western blotting
Samples were subjected to SDS-PAGE in 10% or 12% Laemmli's gel and transferred to a polyvinylidene difluoride membrane. After blocking with 5% skim milk, the membrane was incubated with anti-B chain antibody or anti-PA antibody conjugated with peroxidase (Wako). In the case of anti-B chain antibody, horseradish peroxidase-conjugated secondary antibody (Bio-Rad) was applied, or horseradish peroxidase was conjugated with the Lightning-Link HRP Conjugation Kit (Innova Biosciences, Cambridge, UK). This was followed by development with WesternBright Quantum or Sirius (Advansta, Menlo Park, CA). Images were obtained with the Omega Figure 11. The proteolytic coagulation cascade in horseshoe crabs using newly designated terms. Newly designated terms of the zymogens and their active forms are shown in red.
An active transition state of zymogen factor C Lum G imaging system (Aplegen Life Sciences, San Francisco, CA). Precision Plus protein standards (Bio-Rad) were used to determine the apparent molecular masses.