The Structural Basis for Complement Inhibition by Gigastasin, a Inhibitor from the Giant Amazon Leech

This information is current as Siew Siew Pang, Lakshmi C. Wijeyewickrema, Lilian Hor, of September 29, 2021. Sheareen Tan, Emilie Lameignere, Edward M. Conway, Anna M. Blom, Frida C. Mohlin, Xuyu Liu, Richard J. Payne, James C. Whisstock and Robert N. Pike J Immunol 2017; 199:3883-3891; Prepublished online 23

October 2017; Downloaded from doi: 10.4049/jimmunol.1700158 http://www.jimmunol.org/content/199/11/3883

References This article cites 35 articles, 9 of which you can access for free at: http://www.jimmunol.org/ http://www.jimmunol.org/content/199/11/3883.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 29, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2017 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

The Structural Basis for Complement Inhibition by Gigastasin, a Protease Inhibitor from the Giant Amazon Leech

Siew Siew Pang,*,1 Lakshmi C. Wijeyewickrema,†,1 Lilian Hor,† Sheareen Tan,* Emilie Lameignere,‡ Edward M. Conway,‡ Anna M. Blom,x Frida C. Mohlin,x Xuyu Liu,{ Richard J. Payne,{ James C. Whisstock,*,1 and Robert N. Pike†,1

Complement is crucial to the immune response, but dysregulation of the system causes inflammatory disease. Complement is ac- tivated by three pathways: classical, lectin, and alternative. The classical and lectin pathways are initiated by the C1r/C1s (classical) and MASP-1/MASP-2 (lectin) . Given the role of complement in disease, there is a requirement for inhibitors to control the initiating proteases. In this article, we show that a novel inhibitor, gigastasin, from the giant Amazon leech, potently inhibits C1s and Downloaded from MASP-2, whereas it is also a good inhibitor of MASP-1. Gigastasin is a poor inhibitor of C1r. The inhibitor blocks the active sites of C1s and MASP-2, as well as the anion-binding exosites of the via sulfotyrosine residues. Complement deposition assays revealed that gigastasin is an effective inhibitor of complement activation in vivo, especially for activation via the lectin pathway. These data suggest that the cumulative effects of inhibiting both MASP-2 and MASP-1 have a greater effect on the lectin pathway than the more potent inhibition of only C1s of the classical pathway. The Journal of Immunology, 2017, 199: 3883–3891. http://www.jimmunol.org/ omplement is vital to host immunity (1) and an effector is constitutively active through a so-called “tickover” mechanism system that facilitates the elimination of invading path- and is amplified when Factors D and B are activated following the C ogens, but it must be tightly controlled to avoid inflam- binding of C3b to foreign surfaces (5). Once activated, all three mation and tissue damage (2). The is pathways converge at C3 and progress to the formation of the activated via three pathways (3). The classical pathway is initiated membrane attack complexes (or C5b–C9) on a target membrane (6). by binding of the C1 complex and its associated serine proteases, From a therapeutic perspective, selective inhibition of the different C1r and C1s, to ligands, such as Ag-bound Igs. The lectin pathway complement-activation pathways to attenuate disease would ideally is initiated by the MASP-1 and MASP-2 proteases associated with be required to avoid compromising the immune status of patients. In by guest on September 29, 2021 lectins, such as mannose-binding lectin (4). The alternative pathway this regard, the classical and lectin pathways present more suitable targets, given that the alternative pathway acts as an amplification loop for both of these pathways (7). Accordingly, the C1r/C1s and *Department of Biochemistry and Molecular Biology and Australian Research Coun- cil Centre of Excellence in Advanced Molecular Imaging, Biomedicine Discovery MASP-1/MASP-2 systems present as ideal targets to Institute, Monash University, Melbourne, Victoria 3800, Australia; †Department of inhibit the classical and lectin pathways, respectively. Biochemistry and Genetics and Australian Research Council Centre of Excellence in Advanced Molecular Imaging, La Trobe Institute for Molecular Science, La Trobe Previously, a novel 17-kDa C1s inhibitor, designated BD001 University, Melbourne, Victoria 3086, Australia; ‡Centre for Research, Department (which we have termed gigastasin in this article), derived from the of Medicine, Faculty of Medicine, University of British Columbia, Vancouver, British x giant Amazonian leech (Haementaria sp.) (8) was identified and Columbia V6T 1Z3, Canada; Division of Medical Chemistry, Department of { Translational Medicine, Lund University, Malmo¨ SE-221 00, Sweden; and School of found to inhibit the classical pathway. However, its role in the Chemistry, The University of Sydney, Sydney, New South Wales 2006, Australia lectin pathway was not assessed. Gigastasin includes a highly 1S.S.P., L.C.W., J.C.W., and R.N.P. contributed equally. acidic C terminus containing three sulfotyrosine (sY) residues, ORCIDs: 0000-0003-0557-5447 (S.S.P.); 0000-0001-6084-4887 (L.C.W.); 0000- which is similar to the stretch of amino acids found in complement 0002-2967-0134 (L.H.); 0000-0003-0081-0305 (E.M.C.); 0000-0002-1348-1734 (A.M.B.); 0000-0003-4096-3090 (X.L.); 0000-0002-2083-0269 (R.N.P.). component C4, the primary substrate for C1s and MASP-2. This part of C4 has more recently been shown to bind to C1s (9) and Received for publication February 6, 2017. Accepted for publication September 27, 2017. MASP-2 (10) via highly positively charged areas on the protease This work was supported by National Health and Medical Research Council of surface, termed exosites (anion-binding exosites [ABEs]). Australia Project Grant 1082090 (to R.N.P.). In this study, we have shown that gigastasin is a potent inhibitor of The coordinates and structure factors presented in this article have been sub- C1s and MASP-2, is less active against MASP-1, and is a poor mitted to the (https://www.rcsb.org/pdb/results/results.do? inhibitor of C1r. We have further determined the structure of gig- tabtoshow=Unreleased&qrid=4A0C606A) under accession number 5UBM. astasin in complex with its most preferred target, C1s, and dem- Address correspondence and reprint requests to Prof. Robert N. Pike or Prof. James C. Whisstock, La Trobe Institute of Molecular Science, Melbourne, VIC 3086, Aus- onstrated that the mechanism of inhibition involves contact with the tralia (R.N.P.) or Department of Biochemistry and Molecular Biology and ARC activation loop, the , and the ABE of the protease. Our data Centre of Excellence in Advanced Molecular Imaging, Biomedicine Discovery In- stitute, Monash University, Melbourne, VIC 3800, Australia (J.C.W.). E-mail ad- reveal that contact with the ABE regions of C1s and MASP-2 is vital dresses: [email protected] (R.N.P.) or [email protected] (J.C.W.) to its inhibitory activity. Finally, we used deposition assays to assess Abbreviations used in this article: ABE, anion-binding exosite; MBL, mannose- the effect of gigastasin. Interestingly, in contrast to our predictions binding lectin; NHS, normal human serum; PDB, Protein Data Bank; RT, room from the protease-inhibition kinetics, these data reveal that activation temperature; sY, sulfotyrosine. of the lectin pathway was more potently inhibited by gigastasin than Copyright Ó 2017 by The American Association of Immunologists, Inc. 0022-1767/17/$35.00 was classical pathway activation. www.jimmunol.org/cgi/doi/10.4049/jimmunol.1700158 3884 GIGASTASIN, A NOVEL COMPLEMENT PROTEASE INHIBITOR

Materials and Methods data were indexed, integrated, and scaled using MOSFLM (14) and Materials SCALA (15) from the CCP4 package (16). The complex structure was solved by molecular replacement [using Z–K–SBzl, Boc–LGR–AMC, and Boc–VPR–AMC were purchased from PHASER from the CCP4 suite (17)] using the C1s SP domain (1ELV, residue Bachem. 4,49-Dithiodidyridine was purchased from Sigma-Aldrich. Normal 410–668) (18) as the probe in the space group P3121. Each asymmetric unit human serum (NHS) and C1q-depleted serum were from Complement contains one single C1s/gigastasin complex. After the initial refinement [using Technology (Tyler, TX). Mannose-binding lectin (MBL)-deficient serum was REFMAC5 (19)] with the C1s SP domain, the additional densities of the from the Statens Serum Institut (Copenhagen, Denmark). Murine mono- missingCCP1andCCP2domains,aswellas gigastasin, were clearly visible. clonal anti-C4d Ab was from Quidel (San Diego, CA). Goat anti-mouse The missing structures were built in manually using COOT (20), with re- secondary Ab (Alexa Fluor 488 conjugate) was from Life Technologies. finement cycles using Phenix (21) or the CCP4 suite (16). The progress of Human dermal microvascular endothelial cells (HMEC-1) were a kind gift refinement was monitored using the Rfree value, and the final refinement from the Centers for Disease Control and Prevention (Atlanta, GA). statistics are listed in Table II. Coordinates and structure factors were sub- mitted to the Protein Data Bank (PDB) under ID 5UBM (https://www.rcsb. Purification of active plasma-derived C1s and gigastasin org/pdb/results/results.do?tabtoshow=Unreleased&qrid=4A0C606A). C1s used in the crystallization experiments was purified from expired human Expression, refolding, and purification of the recombinant plasma (Australian Red Cross Blood Service) using a modified protocol (11, 12). Briefly, the expired human plasma was clotted with 20 mM CaCl2, and Constructs for the recombinant expression of C1s, MASP-2, and MASP-1 serum was isolated by centrifugation to remove the clot. The euglobulin CCP1–CCP2–SP segment were synthesized by GenScript and cloned into fraction was precipitated by extensive dialysis in 10 mM Bis-Tris (pH 6.1), a pET-17b vector (GenScript, Piscataway, NJ). The CCP1–CCP2–SP 5 mM CaCl2, and 0.05 mM 4-nitrophenyl 4-guanidinobenzoate hydro- segment of C1s, the C1s ABE mutant, MASP-2, the MASP-2 ABE mutant, chloride. The euglobulin fraction was solubilized in the same buffer plus and MASP-1 were produced as described previously (22). Mutagenesis of 0.15 M NaCl. The C1 complement complex () was purified using the synthesized cDNA for recombinant C1s CCP12SP (residues K281 to Downloaded from IgG-Sepharose (GE Healthcare). C1 zymogen was activated on Mono S D688) was carried out as described previously (9) to introduce a cysteine (GE Healthcare) at neutral pH in the absence of 4-nitrophenyl 4-guanidi- residue at the N terminus of selected proteins. The sequences of all variants nobenzoate hydrochloride. Activation of C1 was monitored using reducing were confirmed by dsDNA sequencing. Expression, refolding, and purifi- and nonreducing SDS-PAGE. The active C1 complex was dissociated into cation of all proteins were carried out as described previously (9). Where its subcomponents C1q, C1r, and C1s by dialysis in the presence of EDTA required, C1s was activated prior to use by incubating overnight at room to remove calcium. The full-length active C1s was separated from the other temperature (RT) with C1r, as previously described (9). C1 subcomponents using Mono Q (GE Healthcare). Multiple-site mutagenesis of MASP-2 CCP12SP (K450A, K503A,

Complete cleavage of plasma-derived C1s generates several short R578A, and R583A) mutants were carried out using a QuikChange Site- http://www.jimmunol.org/ fragments from the N terminus (,15 kDa) and a disulfide-linked CCP1– Directed Mutagenesis Kit, following the manufacturer’s methods, and CCP2–SP region of C1s (∼50 kDa, residues L285–D688) (13). To produce the using splice-overlap PCR. Sequences of variants were confirmed by dideoxy plasmin-cleaved C1s for structural studies, full-length active C1s (1 mg/ml) DNA sequencing. Expression, refolding, and purification of proteins were was incubated with human plasmin (Haematologic Technologies) (/ carried out as described previously (23). substrate ratio [w/w] 1:100) in 50 mM Tris-HCl, (pH 7.2), 0.15 M NaCl, and 20 mM EDTA at 37˚C for 1 h. Cleavage of C1s was monitored using SDS- PAGE. The cleaved C1s mixture was used to make complexes with gigastasin. The activity of C1s and the C1s ABE mutant was measured using the thioester The gigastasin was synthesized using the mature primary sequence substrate Z–K–SBzl (200 mM) and the chromogenic thiol reagent 4,49 reported (8) and codon optimized for expression in insect cells. The dithiodidyridine (1 mM) (24), with an increase in absorbance observed at recombinant gene also has a honeybee melittin signal peptide for secretion,

324 nm. The activity of MASP-1 was measured using the fluorescence by guest on September 29, 2021 an N terminus 6x histidine tag for purification, and an enterokinase quenched substrate Boc–VPR–AMC (50 mM; excitation: 360 nm, emission: cleavage site for removal of the purification tag, if required. The gigastasin 460 nm). The activity of MASP-2 and MASP-2 variants was measured using gene was cloned into the pFastBac1 vector for expression in High Five Boc–LGR–AMC (200 mM; excitation: 360 nm, emission: 460 nm). insect cells. The spent insect cell media containing the secreted recombinant All assays were conducted in 20 mM Tris, 100 mM NaCl, 0.005% (v/v) gigastasin was dialyzed extensively in 20 m Tris-HCl (pH 8), 0.3 M NaCl Triton X-100 (pH 7.4) at 37˚C in a FLUOstar Omega plate reader (BMG before being loaded onto Ni Sepharose (GE Healthcare) equilibrated in the Labtech). Gigastasin was serially diluted, and the residual protease activity same buffer. The column was washed with the same buffer plus 40 mM was measured. Data were fitted to the Morrison Equation (25) using a fixed imidazole. Gigastasin was eluted with the same buffer containing 0.5 M total enzyme concentration. imidazole. The purified inhibitor was buffer exchanged into 20 mM Tris-HCl qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (pH 8), 0.3 M NaCl for storage at 4˚C. ½E þ½I þ Kapp 2 ½E þ½I þ Kapp 2 2 ½E ½I vi T T i T T i 4 T T ¼ 1 2 Complex formation between C1s and gigastasin for crystallization v0 2½ET

C1s (plasmin cleaved) was mixed with gigastasin in a molar ratio of 1:2 in app 0.1 M Tris-HCl (pH 8), 0.3 M NaCl and incubated on ice for 30 min. The Apparent equilibrium constant for inhibition values (Ki ) were converted N-terminal fragments of C1s generated by plasmin cleavage, plasmin, and to Ki using the following relationship: free C1s were removed using chromatography on Ni Sepharose. Eluted app Ki fractions were further purified using Mono Q to remove free gigastasin and Ki ¼   S C1s. The purification steps were monitored using SDS-PAGE. Fractions 1 þ K containing the C1s/gigastasin complex were pooled and concentrated by M ultrafiltration to 5 mg/ml for crystallization. Surface plasmon resonance studies Crystallization of active C1s/gigastasin complex Surface plasmon resonance studies were performed using a Biacore Initial attempts to solve the crystal structure of the full-length C1s/ T100. For these experiments, the N-terminal biotinylated and sulfated gigastasin complex were unsuccessful. The protein complex crystallized readily, but the crystals diffracted poorly and were severely anisotropic. To obtain different crystal forms, the inhibitor was complexed with plasmin- Table I. Equilibrium inhibitory constants for interaction between truncated C1s, and this was used to set up further crystallization trials using gigastasin and complement-initiating proteases the sparse-matrix approach with the hanging-drop vapor-diffusion tech- nique at both 4 and 20˚C. A single hit (0.1 M Bis-Tris, pH 5.5, 25% [w/v] Protease K (nM; mean 6 SE) PEG 3350) was obtained at 4˚C after several months. The pyramid-shaped i C1s/gigastasin complex crystals were cryoprotected with the addition of C1s 0.394 6 0.024 10% (v/v) ethylene glycol and flash-frozen in liquid nitrogen. C1sK575A, R576A, R581A, K583A 273 6 17 MASP-2 0.952 6 0.120 X-ray data collection, structure determination, and refinement MASP-2R578A, R583A 91.9 6 15.6 K450A, K503A, R578A, R583A 6 Complete x-ray diffraction data of C1s/gigastasin complex were collected MASP-2 803 74 6 on the MX2 beamline at the Australian Synchrotron (Table II). Diffraction MASP-1 12.6 1.5 The Journal of Immunology 3885 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 1. The structure of gigastasin. (A) An overview of the C1s/gigastasin complex. CCP1/2 is in purple, the C1s protease domain is in cyan, and gigastasin is in pink (N-terminal domain)/orange (antistasin domain)/green (C-terminal domain). The P1 residue (orange sticks) and sY sequence (green/ yellow/red sticks) are shown. The N and C termini are labeled. (B) Sequence alignment of gigastasin with other members of the antistasin family of inhibitors. (C) Diagram showing gigastasin alone; coloring and labeling are as in (A), with disulfide bonds shown as a yellow stick. (D) Electrostatic potential surface map of the C1s protease domain, with gigastasin shown as the gray cartoon and the amino acids that interact with C1s in a stick rep- resentation. Key interaction points are labeled. Note the extensive positively charged exosite (green dashed circle) that makes up the sY . (E) Electron density map of the loop of gigastasin that makes contact with the active site of C1s. peptide, C12S3 (Biotin-EDPNEEsYEsYDsYE), and its nonsulfated an SA sensor chip (GE Life Sciences) surface for 30 s at 5 ml/min, counterpart (Biotin-EDPNEEYEYDYE) were captured by streptavidin resulting in a final immobilization level of 419 RU. The activated on the Biacore SA chip. The peptide was synthesized essentially as forms of C1s CCP12SPS632A, C1s CCP12SPS632A, K575A, R576A, R581A, K583A, described previously (26). Unused biotin-binding sites were blocked by MASP-2S632A, and MASP-2S632A, R578A, R583A were injected at concentrations free biotin (60 s injection of 10 mM biotin at 30 ml/min). For the peptide- ranging from 0 to 1 mM of each protein for 60 s at 50 ml/min, with a 5-min immobilization strategy, peptides (380 mM) in water were injected over dissociation, followed by regeneration with 10 mM (pH 2) for 30 s at 3886 GIGASTASIN, A NOVEL COMPLEMENT PROTEASE INHIBITOR

30 ml/min and 2 M NaCl for 30 s at 30 ml/min to generate Biacore kinetic characterized to date. The analysis also reveals that the C terminus sensorgrams (Fig. 3). The experiment was repeated in triplicate for each of gigastasin contains a number of negatively charged amino protein, and sensorgrams from triplicate runs were superimposable. The ob- acids, including three sY residues. The interaction between the served maximal response unit for each concentration of enzyme was plotted against the concentration of enzyme used to yield curves (insets) shown in agent, , and is markedly strength- Fig. 3 that were fitted to a one-site, specific binding equation in GraphPad ened by the binding of electronegative residues located on the C 3 Prism 5.0, where the equation was as follows: R = Rmax [E]/(Kd + [E]). terminus of the inhibitor to the ABE of the enzyme (29, 30). Therefore, it is highly possible that the electronegative C-terminal Complement-activation assays measuring deposition of C4d on “tail” of gigastasin might also be interacting with the ABE regions endothelial cells of the complement proteases, such as C1s and MASP-2, to augment Cells were cultured between passages 17 and 20. HMEC-1 cells were cultured inhibition. in MCDB131 nutrient medium (Invitrogen) supplemented with 10% (v/v) To understand the structural basis for gigastasin inhibition, we heat-inactivated FBS, 100 U/ml penicillin/streptomycin, 1 mM sodium py- ˚ ruvate, 5 mM L-Glutamine (Life Technologies, Invitrogen), 10 ng/ml determined the 2.5-A x-ray crystal structure of the complex be- recombinant human EGF (R&D Systems), and 1 mg/ml hydrocortisone tween the CCP1–CCP2–SP fragment of plasma-derived C1s and (Tocris Bioscience) in a humidified incubator (95% O2,5%CO2)at37˚C. gigastasin (Fig. 1, Table II). The structure revealed that gigastasin HMEC-1 cells were grown to 90% confluence, washed with PBS, and consists of a three-domain protein (Fig. 1C) that binds to C1s, with starved for 16 h in serum-free media. Cells were washed with PBS and lifted a total buried surface area of 1365.6 A˚ 2. The central portion of the into suspension with Accutase (Innovative Cell Technologies), after which they were washed twice with PBS and suspended at a concentration of ∼10 inhibitor (59–91) adopted the predicted antistasin-like fold and 3 106 cells per milliliter. Then they were incubated at 37˚C for 1 h with 10% binds into the protease active site in a canonical fashion (Fig. 1A,

(v/v) NHS, C1q-depleted serum, MBL-deficient serum, or corresponding 1E; buried surface area = 788.2 A˚ 2). The primary specificity Downloaded from serum that was heated (1 h at 57˚C) to inactivate complement, in the pres- residue of gigastasin, R65 (P1), docks into the S1 specificity ence of various concentrations of gigastasin. Cells were gently centrifuged and resuspended in FACS buffer (PBS, containing 1% [w/v] BSA) supple- pocket of the enzyme (Fig. 1D, 1E) and forms extensive interac- mented with murine monoclonal anti-C4d Ab at 4 mg/ml for 1 h at 4˚C. Cells tions at the base of the pocket (including a salt bridge to D626). were gently pelleted and resuspended in 100 ml of FACS buffer with poly- However, it is also notable that gigastasin completely blocks all clonal FITC-conjugated goat anti-mouse IgG Abs (1:400; ∼5 mg/ml) for 45 other subsites of C1s and forms contacts that span from P6 to P49 min at 4˚C. The cells were finally resuspended in FACS buffer with 0.1 mg/ml (Fig. 2A). http://www.jimmunol.org/ propidium iodide and analyzed by flow cytometry using a FACS LSR system (BD Biosciences, Mountain View, CA), as previously described (27). In comparison with other members of the antistasin superfamily, gigastasin includes unique N- and C-terminal domains. The N- Complement-deposition assay terminal domain (3–58) includes five disulfide bonds. Despite Unless stated otherwise, all incubations were carried out at RT, and each step was followed by four washes with 50 mM Tris-HCl, 150 mM NaCl, and 0.1% (v/v) Tween 20 (pH 8). Microtiter plates (MaxiSorp; Nunc) were Table II. Crystallographic statistics coated with 5 mg/ml heat-aggregated human IgG, to activate the classical pathway, or with 100 mg/ml mannan, to activate the lectin pathway, in 75 mM Na2CO3 (pH 9.6) overnight at 4˚C and subsequently blocked for Data collection by guest on September 29, 2021 2 h with 1% (w/v) BSA in PBS (blocking solution). Various dilutions of Temperature (˚K) 100 gigastasin (or BSA) in GVB++ (5 mM veronal buffer [pH 7.35], 140 mM Space group P3121 NaCl, 0.1% [w/v] gelatin, 0.15 mM CaCl2, 1 mM MgCl2) were pre- Unit cell parameters (A)˚ incubated with 2% (v/v) human serum for 15 min before adding to the a, b 89.35 plates and incubated for 20 min at 37˚C. The amount of deposited C4b was c 146.87 detected with a specific rabbit polyclonal Ab against C4c (Dako), diluted Resolution (A)˚ 53.27–2.50 (2.64–2.50) 1:4000 in blocking solution, and incubated for 1 h on the plate. HRP- Completeness (%) 100 (100) conjugated anti-rabbit IgG (Dako) was diluted 1:2000 in blocking solu- Total reflections 209,827 (31,050) tion and allowed to bind for 30 min before bound enzyme was quantified Unique reflections 24,136 (3,480) using 1,2-phenylenediamine dihydrochloride (OPD) tablets (Kem-En-Tec Multiplicity 8.7 (8.9) Diagnostics), and the absorbance was measured at 490 nm. Rpim (%) 5.3 (36.0) Mean I/s (I) 10.8 (2.1) Mosaicity (˚) 0.56 Results Refinement Gigastasin inhibits proteases of the classical and lectin Nonhydrogen atoms 3,936 pathways of complement C1s atoms 2,940 Inhibitor atoms 870 Gigastasin was expressed in insect cells using baculovirus and Water 126 purified. The equilibrium-inhibitory constants for the complement Resolution (A)˚ 41.37–2.50 proteases that are able to cleave peptide substrates were determined. Rfactor (%) 17.67 The results revealed that C1s was most potently inhibited by gig- Rfree (%) 22.23 Root-mean-square deviation astasin, followed closely by MASP-2 (2.4-fold higher Ki), whereas from ideality the inhibitor was 32-fold less inhibitory toward MASP-1 (Table I). Bond lengths (A)˚ 0.004 The inhibitor only had an effect on the cleavage of C1s by the C1r Bond angles (A)˚ 0.872 protease of the classical pathway at very high concentrations, sug- Chirality (˚) 0.036 gesting that it is a poor inhibitor of C1r (data not shown). Planarity (˚) 0.004 Dihedrals (˚) 14.253 Ramachandran plot (%) Structure of the gigastasin–C1s complex reveals new insights Preferred regions 95.84 into the mechanism of inhibition Allowed regions 4.16 Bioinformatics analysis suggested that gigastasin is a distantly Disallowed regions 0.00 B factors (A˚ 2) related member of the antistasin superfamily of protease inhibitors Average all atoms 56.89 (Fig. 1) (28). These data reveal that it also contains an additional Average C1s 49.96 sequence at its N terminus that appears unique with respect to Average inhibitor 80.32 antistasin-like inhibitors or any other serine protease inhibitor Average water 45.22 The Journal of Immunology 3887

FIGURE 2. Details of the major con- tacts made between gigastasin and C1s. (A) The inhibitor makes extensive ca- nonical interactions from P6 to P49. D60 (P6) forms a hydrogen bond to Y610. F62 (P4) forms p-stacking interactions with the side chain of Y610 and also packs against P657. The main chain carbonyl of K63 (P3) forms a hydrogen bond to the backbone amide of G656. C64 (P2) forms a disulfide bond with C83 and is loosely packed into the shal- low S2 pocket. The side chain of R65 (P1) forms a buried salt bridge with D626 in the S1 pocket. The carbonyl oxygen of the P1 further forms the an- Downloaded from ticipated interactions with the (formed from backbone amides of G630, D631, and S632 [catalytic serine (S195 in numbering)]; only S632 is labeled). On the prime side, L66 (P19) is positioned such that it makes van der Waals interactions with http://www.jimmunol.org/ A460, V476, and P458 (these latter three residues are not shown). G67 (P29) and C68 (P39) form main-chain hydrogen bonds with the active site; finally, the side chain of T69 (P49) interacts with the side chain of N457. (B) Additional in- teractions made by gigastasin include W17 forming a hydrogen bond with

Y662 and the carbonyl oxygen of C54 by guest on September 29, 2021 interacting with the side chain of Q658. The C-terminal sY residues Y117 and Y119 both interact with Q493 and form ionic contacts with R576, R581, and K583.

representing the largest domain, this region only makes two sig- completely clear, we further investigated the importance of this nificant contacts with C1s (buried surface area = 265.9 A˚ 2). Most interaction in vitro. To do this, we used a mutant of C1s with the 4 notably, W17 forms extensive van der Waals contacts with the aa constituting the ABE of this enzyme mutated to alanine N-terminal (activation loop) region of C1s (Fig. 2B). In addition, residues (C1sK575A, R576A, R581A, K583A) (9, 22). Two mutants of the backbone carbonyl of C54 forms a hydrogen bond to Q658 on MASP-2 were used: one in which the two residues that were the C1s protease domain. showntobevitalforinteractionswithC4weremutatedtoal- Similarly, the C-terminal domain (92–119) of gigastasin forms anine residues (MASP-2R578A, R583A) (10) and the other in which ˚ 2 relatively limited contacts (buried surface area = 332.1 A ) with all four residues originally predicted to be interacting with C4 the protease domain. These contacts include a salt bridge between were mutated to alanine residues (MASP-2K450A, K503A, R578A, R583A) R109 and D577 on the C1s serine protease domain. Most notably, (10). The K for gigastasin with the ABE mutant of C1s was however, the far C-terminal sequence includes two sY residues i 693-fold weaker than the wild-type enzyme (Table I), indicat- that directly contact the ABE of C1s. Major polar interactions ing that the interaction between the C terminus of gigastasin made in these regards include contact with Q493, R576, R581, and the ABE is indeed of major importance to this enzyme. In and K583 (Fig. 2B). These latter three residues are known to be the case of MASP-2, the double mutant form (MASP-2R578A, R583A) the major features of the ABE found on the surface of C1s (9, 22). was inhibited 97-fold less effectively, whereas the mutant in Analysis of the interaction of the C-terminal portion of which all 4 aa of the ABE were mutated was inhibited 843- gigastasin with proteases fold less effectively, indicating that, for this enzyme, the en- Because the electron density for the amino acids from the tire ABE was required for full interaction with gigastasin C terminus of gigastasin interacting with the ABE of C1s was not (Table I). 3888 GIGASTASIN, A NOVEL COMPLEMENT PROTEASE INHIBITOR

To further delineate the details of the interaction with the ABE, a cellular surfaces. Activation within NHS will monitor activation by 12-aa peptide representing the C-terminal tail of gigastasin, C12S3, all pathways. In this context, gigastasin was able to inhibit C4d was synthesized bearing intact sulfation of the tyrosine residues deposition by ∼75% at a concentration of 1 mM, whereas 500 nM (31). The synthetic peptide was modified with an N-terminal bi- gigastasin decreased C4d deposition by ∼50% (Fig. 4). Assaying otin residue to allow immobilization to a streptavidin-coated sur- in the presence of serum depleted of MBL, a primary recognition face. Measurement of the binding kinetics of C1s and MASP-2, complex of the lectin pathway, provided an estimate of the acti- as well as the full ABE mutant of C1s and the mutant of MASP-2 vation occurring via the classical pathway. Gigastasin was able to in which two residues of the ABE were mutated, was carried out inhibit activation via this pathway by ∼50% at a concentration of using surface plasmon resonance. This showed that wild-type C1s 1.25 mM, whereas 5 mM of the inhibitor was required to decrease bound to the immobilized peptide with a Kd value of 810 nM, complement activation via this pathway by .80%. Assays in the whereas MASP-2 had a 16-fold greater affinity for the peptide presence of C1q-depleted serum gave a measure of the activation (Kd = 52 nM) (Fig. 3). The mutant forms did not display any occurring via the lectin pathway: only 0.5 mM gigastasin was binding to the molecule. The wild-type molecules also bound a required to inhibit activation by .85%, whereas 0.06 mM giga- nonsulfated form of the peptide with significantly lowered af- stasin resulted in .50% inhibition of the activation of the lectin finity (results not shown), indicating that the sulfation was im- pathway. Our findings that, in this cellular system, gigastasin fa- portant for the reaction. vors the lectin pathway is somewhat surprising, given the pre- ceding biochemical data. Therefore, we sought validation by a Complement activation assayed using C4d deposition on second ELISA approach in which we quantified the deposition of endothelial cells C4b on surfaces required for activation of the different pathways Downloaded from Activation of the early stages of complement can be monitored by in the presence and absence of gigastasin (Fig. 4D). These ex- measuring the deposition of C4 cleavage products (C4b or C4d) on periments also revealed a more prominent inhibitory effect of http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 3. Analysis of the interaction between complement proteases and gig- astasin C-terminal peptide. Sensorgrams from surface plasmon resonance mea- surements for the indicated amounts of C1s and MASP-2 flowed over peptide Biotin-EDPNEEsYEsYDsYE attached to streptavidin immobilized to the chip. The experiments were conducted in triplicate and showed excellent overlap. Insets: the response units obtained at equilibrium for each concentration of protease were plotted against the respective protease concentration and fitted using a one site, specific binding model on GraphPad Prism (regression coefficients for the fits =

0.99) to yield the Kd values shown. The Journal of Immunology 3889 gigastasin on the lectin pathway compared with the classical vation from the giant Amazonian leech, led us to propose the name pathway. gigastasin for this molecule. Intriguingly, the sequence alignments also revealed that the N terminus of the molecule bore no relation to Discussion other protease inhibitors, suggesting that it may play a hitherto The leech molecule previously designated BD001 was first de- unknown role in the mechanism of action of this inhibitor. scribed as an inhibitor of the classical pathway enzyme, C1s, some Solution of the structure of gigastasin in complex with C1s has time ago (8), but its mechanism of action and effects on the lectin given us our first insights into the detailed interactions that are pathway of complement have not been fully determined. We (9) made between an inhibitor and the active site of C1s. In addition, and others investigators (32) have described the involvement of the structural and biochemical work carried out in this study has ABEs located on the surfaces of C1s and MASP-2 in the binding validated the hypothesis that the acidic tail of gigastasin does and cleavage of their primary substrate, C4. We were intrigued to indeed bind to the ABE of C1s and plays a major role in the note that the leech molecule contained a highly negatively charged interaction between the inhibitor and this enzyme. This first C-terminal “tail” encompassing three sY residues, which is very visualization of the interaction between the sY residue and the similar to the situation found for the ABE-binding region of C4 (9, ABE of C1s is particularly interesting because it is notably very 32). This led us to hypothesize that the C-terminal tail of the leech similar to what might be expected for the interaction between the molecule might play a role in binding to the ABE of C1s. This negatively charged region of C4 and the ABE of C1s. Finally, the would be similar to the role of the negatively charged C-terminal N-terminal region of gigastasin has been shown to house an region of hirudin, which binds to the ABE of thrombin and, thus, entirely novel interaction with the activation loop of C1s. This plays a crucial role in inhibiting the enzyme (29, 30). intriguing mechanism ensures that full inhibition of the protease Downloaded from Bioinformatic analysis revealed that the central inhibitor domain only occurs with the activated form of the enzyme, because the of the leech molecule was homologous to members of the antistasin point of interaction on the enzyme is only exposed upon acti- family of protease inhibitors. This result, together with its deri- vation (22). http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 4. Gigastasin inhibits the classical and lectin complement pathways in a dose-dependent manner. (A) Complement pathways were activated by incubating HMEC-1 cells with 10% (v/v) NHS. In a similar manner, HMEC-1 cells were incubated with 10% (v/v) MBL-deficient serum or C1q-depleted serum to reflect activation of primarily the classical (B) or lectin (C) complement pathway, respectively, prior to measuring C4d deposition by flow cytometry. C4d-staining graphs (right panels) show FITC-conjugated anti-C4d fluorescence (y-axis) and cell count (x-axis) for the NHS positive control (purple), background fluorescence (orange; derived from values for heat-inactivated serum), buffer negative control (blue), and incubation conducted with gigastasin (red). Bar graphs (left panels) represent the average mean fluorescence intensity (MFI) in the presence of various gigastasin concentrations. A volume of the buffer used to dissolve gigastasin equivalent to the highest gigastasin concentration tested was used as a negative control. The average MFI for serum-activated samples was corrected for background fluorescence by subtracting the average MFI value derived from the corresponding heat- inactivated serum samples. Values are MFI 6 SEM of duplicate samples and are representative of two independent experiments. (D) Gigastasin inhibits C4b deposition. Heat-aggregated human IgG (classical pathway) or mannan (lectin pathway) was immobilized and allowed to activate human serum containing various concentrations of gigastasin or BSA as a negative control (preincubated for 15 min at RT). After 20 min of incubation at 37˚C, the plates were washed, and the deposited C4b was detected with a specific polyclonal Ab. The absorbance obtained in the absence of gigastasin was defined as 100%. The average of three independent experiments performed in duplicate is presented, with error bars indicating SD. 3890 GIGASTASIN, A NOVEL COMPLEMENT PROTEASE INHIBITOR Downloaded from http://www.jimmunol.org/

FIGURE 5. Superimposition of the structures of MASP-1 (A) and Clr (B) onto the structure of gigastasin–C1s. (A) The side chains of N451 and F593 of MASP-1 (light pink, 3GOV) form significant steric clashes with W17 in gigastasin (gray). The equivalent residues in Cls (G440 and E574, respectively) form part of the binding pocket that interacts with inhibitor residue W17. (B) Clr (orange, 1MD8) 491 loop (Pro489-His492, stick model) forms close steric clashes with gigastasin (gray, C83-C85). The equivalent loop region (G479-E482) in C1s (highlighted as a stick model in cyan) does not interfere with gigastasin binding.

Gigastasin, untested for its ability to inhibit lectin pathway inhibited activation of the classical pathway (8) and did not effect by guest on September 29, 2021 proteases to date, was shown to have a slight preference for in- activation via the alternative pathway using a cellular model of hibition of C1s, but it was also an effective inhibitor of MASP-2 complement activation. Somewhat surprisingly, gigastasin was a and, to a lesser extent, MASP-1. These data laid the basis for further far more potent inhibitor of the lectin pathway than the in vitro work investigating the ability of the inhibitor to effect activation of protease-inhibition results might have suggested. Because this was the lectin pathway of complement. We used the gigastasin–C1s a surprising result, and noting that the MBL-deficient serum structure in superposition experiments to help explain the selec- would still contain ficolin complexes that would be capable of tivity observed with regard to the different complement-initiation activating the lectin pathway, we confirmed these results using a proteases. Superposition of gigastasin–C1s with the structure of C4b-deposition assay on surfaces containing ligands that selec- MASP-2 (PDB ID: 1Q3X) (33) suggests that the inhibitor could tively activate each of the pathways of complement. Data from interact with MASP-2 in an essentially identical fashion (i.e., we these assays confirmed the results obtained using the cellular as- observed no predicted steric clashes) (data not shown). In contrast, says. C1s, and thus the classical pathway, might be suggested to be a similar analysis of the gigastasin–C1s structure with respect to the primary target of gigastasin based on the inhibitory constants MASP-1 (PDB ID: 3GOV) (34) reveals that the latter protease obtained for interaction with the complement proteases; however, would not be anticipated to be able to interact with the gigastasin when one considers that the Ki value for MASP-2 is only 2.4-fold N- and C-terminal domains in a similar fashion to C1s (Fig. 5). greater than that for C1s and that the inhibitor also targets MASP- For example, N451 and F593 of MASP-1 are predicted to block 1 to a reasonable degree, it is likely that inhibition of both acti- the interactions made by W17 in gigastasin with either C1s or vating proteases makes the lectin pathway much more susceptible MASP-2. However, we note that the central antistasin-like domain to inhibition by gigastasin. Therefore, the absence of inhibition of of gigastasin would still be anticipated to bind to the active site of C1r, the primary activating enzyme for the classical pathway, MASP-1 without significant steric clashes. Finally, in C1r (PDB would be envisaged to outweigh the slightly more potent inhibi- ID: 1MD8) (35), the 491 loop of the enzyme blocks binding of the tion of C1s in terms of the effect of the inhibitor on the activation central antistasin family domain to the active site of the enzyme, of these two pathways. The lower concentration of MASP-2 than and the other major interactions with the N and C termini of the C1s in the blood (36) might also ensure that inhibition of the lectin inhibitor are also not predicted to occur, explaining why this pathway is more efficient, although the excess of gigastasin used protease is poorly inhibited (Fig. 5). in the present experiments should have taken this concern into Following the deciphering of the mechanism of action for account. Overall, these results suggest that gigastasin is, in fact, gigastasin and understanding that it is a potent inhibitor of lectin more specific for the lectin pathway. pathway proteases in addition to C1s, we investigated the ability of In summary, this study demonstrates that the leech-derived in- the inhibitor to effect activation of the different complement hibitor, gigastasin, has a novel mechanism of action in which it pathways. As expected, and as previously described, gigastasin targets the complement proteases C1s and MASP-2 using three The Journal of Immunology 3891 regions of interaction. This novel member of the antistasin-like 19. Murshudov, G. N., P. Skuba´k, A. A. Lebedev, N. S. Pannu, R. A. Steiner, R. A. Nicholls, M. D. Winn, F. Long, and A. A. Vagin. 2011. REFMAC5 for the family of protease inhibitors shows great promise as a scaffold refinement of macromolecular crystal structures. Acta Crystallogr. D Biol. for the production of molecules that are specifically able to target Crystallogr. 67: 355–367. each of the lectin and classical pathways of the complement system 20. Emsley, P., B. Lohkamp, W. G. Scott, and K. Cowtan. 2010. Features and de- velopment of Coot. Acta Crystallogr. D Biol. Crystallogr. 66: 486–501. and, thus, ameliorate the role of these pathways in specific in- 21. Adams, P. D., P. V. Afonine, G. Bunko´czi, V. B. Chen, I. W. Davis, N. Echols, flammatory diseases associated with these pathways. J. J. Headd, L. W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve, et al. 2010. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66: 213–221. Disclosures 22. Perry, A. J., L. C. Wijeyewickrema, P. G. Wilmann, M. J. Gunzburg, The authors have no financial conflicts of interest. L. D’Andrea, J. A. Irving, S. S. Pang, R. C. Duncan, J. A. Wilce, J. C. Whisstock, and R. N. Pike. 2013. A molecular switch governs the interaction between the human complement protease C1s and its substrate, complement C4. J. Biol. Chem. 288: 15821–15829. References 23. Duncan, R. C., F. Bergstro¨m, T. H. Coetzer, A. M. Blom, L. C. Wijeyewickrema, 1. Carroll, M. C. 2004. The complement system in regulation of adaptive immunity. and R. N. Pike. 2012. Multiple domains of MASP-2, an initiating complement Nat. Immunol. 5: 981–986. protease, are required for interaction with its substrate C4. Mol. Immunol. 49: 2. Ricklin, D., G. Hajishengallis, K. Yang, and J. D. Lambris. 2010. Complement: a 593–600. key system for immune surveillance and homeostasis. Nat. Immunol. 11: 785– 24. McRae, B. J., T. Y. Lin, and J. C. Powers. 1981. Mapping the substrate binding 797. site of human C1r and C1s with peptide thioesters. Development of new sensitive 3. Duncan, R. C., L. C. Wijeyewickrema, and R. N. Pike. 2008. The initiating substrates. J. Biol. Chem. 256: 12362–12366. proteases of the complement system: controlling the cleavage. Biochimie 90: 25. Copeland, R. A. 2005. Evaluation of enzyme inhibitors in drug discovery. A 387–395. guide for medicinal chemists and pharmacologists. Methods Biochem. Anal. 46: 4. Wallis, R. 2007. Interactions between mannose-binding lectin and MASPs 1–265. during complement activation by the lectin pathway. Immunobiology 212: 289– 26. Hsieh, Y. S., D. Taleski, B. L. Wilkinson, L. C. Wijeyewickrema, T. E. Adams, Downloaded from 299. R. N. Pike, and R. J. Payne. 2012. Effect of O-glycosylation and tyrosine sul- 5. Nilsson, B., and K. Nilsson Ekdahl. 2012. The tick-over theory revisited: is C3 fation of leech-derived peptides on binding and inhibitory activity against a contact-activated protein? Immunobiology 217: 1106–1110. thrombin. Chem. Commun. 48: 1547–1549. 6. Muller-Eberhard,€ H. J. 1986. The membrane attack complex of complement. 27. Wijeyewickrema, L. C., E. Lameignere, L. Hor, R. C. Duncan, T. Shiba, Annu. Rev. Immunol. 4: 503–528. R. J. Travers, P. R. Kapopara, V. Lei, S. A. Smith, H. Kim, et al. 2016. Poly- 7. Forneris, F., D. Ricklin, J. Wu, A. Tzekou, R. S. Wallace, J. D. Lambris, and phosphate is a novel for regulation of complement by a , C1 P. Gros. 2010. Structures of C3b in complex with factors B and D give insight inhibitor. Blood 128: 1766–1776.

into complement convertase formation. Science 330: 1816–1820. 28. Corral-Rodrı´guez, M. A., S. Macedo-Ribeiro, P. J. Pereira, and P. Fuentes-Prior. http://www.jimmunol.org/ 8. West, R. R., P. O. Sheppard, and B. A. Fox. inventors; Zymogenetics, Inc., as- 2010. Leech-derived thrombin inhibitors: from structures to mechanisms to signee. Peptide and polypeptide inhibitors of complement C1s. United States clinical applications. J. Med. Chem. 53: 3847–3861. patent application PCT/US2001/019405, Publication No. WO2001098365 A2. 29. Hsieh, Y. S., L. C. Wijeyewickrema, B. L. Wilkinson, R. N. Pike, and 2001 June 18. R. J. Payne. 2014. Total synthesis of homogeneous variants of hirudin P6: a post- 9. Duncan, R. C., F. Mohlin, D. Taleski, T. H. Coetzer, J. A. Huntington, translationally modified anti-thrombotic leech-derived protein. Angew. Chem. R. J. Payne, A. M. Blom, R. N. Pike, and L. C. Wijeyewickrema. 2012. Iden- Int. Ed. Engl. 53: 3947–3951. tification of a catalytic exosite for complement component C4 on the serine 30. Winant, R. C., J. B. Lazar, and P. H. Johnson. 1991. Chemical modifications and protease domain of C1s. J. Immunol. 189: 2365–2373. substitutions in recombinant hirudin that increase hirudin-thrombin 10.Drentin,N.,P.Conroy,M.J.Gunzburg,R.N.Pike,andL.C.Wijeyewickrema. affinity. Biochemistry 30: 1271–1277. 2015. Investigation of the mechanism of interaction between mannose-binding 31. Stone, M. J., and R. J. Payne. 2015. Homogeneous sulfopeptides and sulfopro- lectin-associated serine protease-2 and complement C4. Mol. Immunol. 67: 287–293. teins: synthetic approaches and applications to characterize the effects of tyro-

11. van den Berg, C. W. 2000. Purification of complement components, regulators, sine sulfation on biochemical function. Acc. Chem. Res. 48: 2251–2261. by guest on September 29, 2021 and receptors by classical methods. Methods Mol. Biol. 150: 15–52. 32. Kidmose, R. T., N. S. Laursen, J. Dobo´, T. R. Kjaer, S. Sirotkina, L. Yatime, 12. Lane, P. D., V. N. Schumaker, Y. Tseng, and P. H. Poon. 1991. Isolation of L. Sottrup-Jensen, S. Thiel, P. Ga´l, and G. R. Andersen. 2012. Structural basis for human complement subcomponents C1r and C1s in their unactivated, proen- activation of the complement system by component C4 cleavage. Proc. Natl. zyme forms. J. Immunol. Methods 141: 219–226. Acad. Sci. USA 109: 15425–15430. 13. Villiers, C. L., G. J. Arlaud, and M. G. Colomb. 1985. Domain structure and 33. Harmat, V., P. Ga´l, J. Kardos, K. Szila´gyi, G. Ambrus, B. Ve´gh, G. Na´ray-Szabo´, associated functions of subcomponents C1r and C1s of the first component of and P. Za´vodszky. 2004. The structure of MBL-associated serine protease-2 human complement. Proc. Natl. Acad. Sci. USA 82: 4477–4481. reveals that identical substrate specificities of C1s and MASP-2 are realized 14. Battye, T. G., L. Kontogiannis, O. Johnson, H. R. Powell, and A. G. Leslie. 2011. through different sets of enzyme-substrate interactions. J. Mol. Biol. 342: 1533– iMOSFLM: a new graphical interface for diffraction-image processing with 1546. MOSFLM. Acta Crystallogr. D Biol. Crystallogr. 67: 271–281. 34. Dobo´, J., V. Harmat, L. Beinrohr, E. Sebestye´n, P. Za´vodszky, and P. Ga´l. 2009. 15. Evans, P. 2006. Scaling and assessment of data quality. Acta Crystallogr. D Biol. MASP-1, a promiscuous complement protease: structure of its catalytic region Crystallogr. 62: 72–82. reveals the basis of its broad specificity. J. Immunol. 183: 1207–1214. 16.Winn,M.D.,C.C.Ballard,K.D.Cowtan,E.J.Dodson,P.Emsley,P.R.Evans, 35. Budayova-Spano, M., W. Grabarse, N. M. Thielens, H. Hillen, M. Lacroix, R.M.Keegan,E.B.Krissinel,A.G.Leslie,A.McCoy,etal.2011.Overviewofthe M. Schmidt, J. C. Fontecilla-Camps, G. J. Arlaud, and C. Gaboriaud. 2002. CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr. 67: 235–242. Monomeric structures of the zymogen and active catalytic domain of comple- 17. McCoy, A. J., R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storoni, ment protease c1r: further insights into the c1 activation mechanism. Structure and R. J. Read. 2007. Phaser crystallographic software. J. Appl. Cryst. 40: 658–674. 10: 1509–1519. 18. Gaboriaud, C., V. Rossi, I. Bally, G. J. Arlaud, and J. C. Fontecilla-Camps. 2000. 36. Glovsky, M. M., P. A. Ward, and K. J. Johnson. 2004. Complement determi- Crystal structure of the catalytic domain of human complement c1s: a serine nations in human disease. Ann. Allergy Asthma Immunol. 93: 513–522; quiz protease with a handle. EMBO J. 19: 1755–1765. 523–525, 605.