Key Points
hC3Nb3 inhibits classical pathway–mediated hemolysis.
hC3Nb3 inhibits alternative pathway–mediated C3 fragment deposition.
hC3Nb3 provides a new tool for studies of the complement system.
Abstract
The complement system is an intricate cascade of the innate immune system and plays a key role in microbial defense, inflammation, organ development, and tissue regeneration. There is increasing interest in developing complement regulatory and inhibitory agents to treat complement dysfunction. In this study, we describe the nanobody hC3Nb3, which is specific for the C-terminal C345c domain of human and mouse complement component C3/C3b/C3c and potently inhibits C3 cleavage by the alternative pathway. A high-resolution structure of the hC3Nb3–C345c complex explains how the nanobody blocks proconvertase assembly. Surprisingly, although the nanobody does not affect classical pathway–mediated C3 cleavage, hC3Nb3 inhibits classical pathway–driven hemolysis, suggesting that the C-terminal domain of C3b has an important function in classical pathway C5 convertase activity. The hC3Nb3 nanobody binds C3 with low nanomolar affinity in an SDS-resistant complex, and the nanobody is demonstrated to be a powerful reagent for C3 detection in immunohistochemistry and flow cytometry. Overall, the hC3Nb3 nanobody represents a potent inhibitor of both the alternative pathway and the terminal pathway, with possible applications in complement research, diagnostics, and therapeutics.
Introduction
The complement system constitutes a pivotal arm of the innate immune system and serves as a surveillance system that scans our body for intruding pathogens as well as abnormal, damaged, and dying host cells. The complement system comprises a proteolytic cascade, elicited through either of the three pathways, the classical pathway (CP), the lectin pathway (LP), or the alternative pathway (AP) (1). Pattern recognition molecules (PRMs) of the CP and LP initiate these pathways upon binding of specific pathogen- or damage-associated molecular patterns. The LP possesses at least five different PRMs, which can bind to various acetyl- and carbohydrate-containing structures (2). In contrast, C1q is the sole PRM of CP and binds Ab–Ag complexes as well as a number of other patterns (3). Upon activation of the CP and LP, PRM-associated serine proteases cleave the complement component C4 into C4a and C4b that binds covalently to the activating surface through an internal thioester (TE). The C4b then recruits the serine protease C2 and, upon activation of C2, forms the C3 convertase, C4b2a. The C4b2a convertase cleaves C3 into the anaphylatoxin C3a and the opsonin C3b that, similar to C4b, binds the activating surface through a TE. The C3b molecule associates with factor B (FB) and forms the AP proconvertase C3bB. The serine protease factor D (FD) can then cleave FB, resulting in the formation of the AP C3 convertase C3bBb (4). The AP convertase cleaves C3 into C3b that can result in additional C3bBb complexes and thereby supports a positive feedback loop, which markedly amplifies the complement activation (5, 6). When exceeding a threshold density of C3b, the C3 convertases start cleaving C5 (7, 8) into C5b and the anaphylatoxin C5a. C6 captures the nascent C5b and nucleates the assembly of the C5b-9 membrane attack complex, a pore forming structure, that lyse some Gram-negative bacteria (9). In some conditions, C5b-9 can lyse, for instance, autologous RBCs as well as transfused allogenic RBCs, but when attacking nuclear cells, sublytic levels of C5b-9 may cause activation of proinflammatory signaling rather than lysis (10).
To prevent complement activation and damage on healthy host cells, a number of negative regulators prevent uncontrolled activation of the complement system. A prominent regulator of the complement system is factor H (FH) that regulates the AP activation in two ways: 1) by acting as a cofactor for the protease factor I (FI), which cleaves C3b to the inactive iC3b, and 2) by acting as a decay-accelerating factor by competing with the Bb moiety of FB for binding to C3b, which irreversibly dissociates the AP C3 convertase. Three cell bound regulators accompany FH in the regulation of the complement system: membrane cofactor protein (CD46), complement receptor (CR) 1 (CD35), and decay-accelerating factor (CD55) (11). The cleavage of C3b to iC3b by FI allows for the binding to various CRs on phagocytes, including the integrins CR3 (CD11b/CD18) and CR4 (CD11c/CD18) that enhance phagocytosis of structures with deposited C3 fragments (12). In contrast to the negative regulators, properdin (factor P [FP]) positively regulates the AP through stimulation of proconvertase formation, stabilization of the convertase, and competition with FI (13, 14).
Besides conveying protection against invading pathogens and clearing cell debris, the complement system is also important in neurodevelopment and facilitates the removal of excess synapses in a process called synaptic pruning (15). In murine models, the system aids the development of the retinogeniculate system (16, 17), neocortical circuitry (18), and refinement of spinal motor circuits (19). Whereas the complement system is critical during development, inappropriate activation later in life may be involved in neurodegenerative diseases including multiple sclerosis (20) and Alzheimer disease (21–23) as well as in the neurodevelopmental disease schizophrenia (24). Excess complement activation also plays a role in various nonneurologic diseases, many of which arise from inborn genetic errors. Loss-of-function mutations in the genes encoding FH, membrane cofactor protein, and FI as well as gain-of-function mutations in FB and C3 can lead to the life-threatening renal disease atypical hemolytic uremic syndrome (25). Furthermore, mutations in various complement components are associated with the renal disease C3 glomerulopathy (26), age-related macular degeneration (27), and paroxysmal nocturnal hemoglobinuria (28).
Inhibitors and diagnostic tools that allow studies of the complement system are highly desirable. In the pursuit of developing an arsenal of complement inhibitors, in this study, we use the nanobody technology. These single domain Abs originate from the H chain–only Abs found in members of the camelid family (29). We have previously described the selection and characterization of the two C3-specific nanobodies hC3Nb1 (30) and hC3Nb2 (31). The hC3Nb1 nanobody inhibits solely the AP, whereas the hC3Nb2 inhibits both CP, LP, and AP. In this study, we describe the selection and characterization of the hC3Nb3 nanobody that specifically inhibits the C3 convertase of the AP like hC3Nb1 but, in addition, inhibits the C5 convertases. The latter provides the first direct identification of a region in C3b that functions in the C5 convertase.
Materials and Methods
Nanobody selection
The hC3Nb3 nanobody was selected as described (30, 31). For nanobody selection, RNA was purified from PBLs from a llama (Lama glama) that was immunized with C3b. A cDNA library was prepared, and the VHH regions were cloned into phagemid vectors. Phage display was used for the selection of nanobodies with specific binding to the murine C-terminal domain of C3 (C345c) domain. One microgram of recombinant murine C3 C345c in PBS was coated in a microtiter plate for 16 h. Excess binding sites were blocked by incubating the plate in PBS supplemented with 2% (w/v) BSA. Then, 3 × 1012 nanobody-presenting M13 phages in PBS were added and incubated for 1 h at room temperature followed by 15 washes in PBS, 0.1% Tween and 15 washes in PBS. Phages were eluted by a 10-min incubation period in 0.2 M glycine (pH 2.2) followed by the addition of Tris (pH 9.1) for neutralization. The eluted phages were used in a second round of phage display; however, in this round, the wells were coated with only 0.1 μg recombinant C345c domain. After the second round of selection, C3 C345c binding nanobodies were identified by ELISA as described (30). To obtain the nanobody, individual phage-infected Escherichia coli colonies were inoculated in individual wells of a 96-well plate containing 2× Yeast Extract Tryptone medium. The cultures were grown for 6 h at 37°C, nanobody expression was induced by the addition of 0.8 mM isopropyl β-D-1-thiogalactopyranoside, and the cultures were incubated for 16 h at 30°C. Nanobodies were obtained from the supernatant, and binders were identified by ELISA. To this end, an ELISA plate was coated with 0.1 μg C345c/well. The plate was blocked by the addition of PBS-supplemented 0.1% Tween and 2% (w/v) BSA. Fifty microliters of the nanobody supernatant was transferred to the ELISA plate, and the plate was incubated for 1 h. The plate was washed six times in PBS 0.1% Tween and then incubated for 1 h in 1:10,000 anti–E-tag-HRP Ab (Bethyl Laboratories). The plate was washed three times in PBS 0.1% Tween and then supplemented with 3,3′,5,5′-tetramethylbenzidine. The reaction was subsequently quenched by the addition of 1 M HCl, and the absorbance at 450 nm was measured. Phagemids were sequenced to identify individual clones, and the nanobodies were cloned into pET-22b(+).
Nanobody, C3, C3 C345c, C4b, cobra venom factor, FB, FD, FP, mini-FH, and CR3 headpiece purification
Nanobodies were expressed in LOBSTR cells (32) as described (30, 31). Transformed cells were inoculated in 2× Yeast Extract Tryptone medium supplemented with ampicillin and grown at 37°C until an OD600 of 0.6 was reached. The nanobody expression was induced by the addition of 1 mM isopropyl β-D-1-thiogalactopyranoside, and expression was continued for 16 h at 18°C. Cells were pelleted and resuspended in 20 mM Tris (pH 8.5), 500 mM NaCl, 20 mM imidazole, 0.5 mM EDTA, and 1 mM PMSF. Resuspended cells were opened by sonication, and the cell debris was pelleted by centrifugation. The supernatant was loaded onto a HisTrap Crude Fast Flow (FF) (GE Healthcare) column for immobilized metal affinity chromatography (IMAC). The column was washed in resuspension buffer, and the nanobody was eluted by resuspension buffer supplemented with 400 mM imidazole. The eluate was dialyzed against 20 mM sodium acetate (pH 5.5) and 50 mM NaCl. The dialysate was applied to a Source 15S (GE Healthcare) column equilibrated in 20 mM sodium acetate (pH 5.5) and 20 mM NaCl, and the protein was eluted by a linear gradient from 20 to 350 mM NaCl over 35 column volumes. A subsequent polishing step was performed using a Superdex 75 (GE Healthcare) column in 20 mM HEPES (pH 7.5) and 150 mM NaCl.
The hC3Nb3 fused with GFP (hC3Nb3-GFP) in pET-22b(+) (GenScript) was expressed in LOBSTR cells (32) using a similar protocol. However, the IMAC eluate was dialyzed against 20 mM Tris (pH 8.5) and 50 mM NaCl and subsequently loaded onto a 9 ml Source 15Q (GE Healthcare) column equilibrated in 20 mM Tris (pH 8.5) and 20 mM NaCl. The hC3Nb3-GFP was eluted by a linear gradient from 50 to 500 mM NaCl over seven column volumes before being subjected to a polishing step as described above.
hC3Nb3-IgY was transiently expressed in HEK293F cells by transfecting cells with 1 μg/ml of plasmid mixed with 3 μg/ml polyethylenimine. The supernatant was harvested 5 d after transfection. The pH of the supernatant was adjusted by the addition of 50 mM Tris (pH 8.5), and 500 mM NaCl was added. The supernatant was applied to a HisTrap Excel (GE Healthcare) column equilibrated in 20 mM Tris (pH 8.5) and 500 mM NaCl. The column was washed twice in 25 ml 20 mM Tris (pH 8.5) and 500 mM NaCl and then eluted in 2 × 15 ml 20 mM Tris (pH 8.5), 500 mM NaCl, and 250 mM imidazole. The eluate was diluted 10-fold in 50 mM Tris (pH 8.5) and loaded onto a HisTrap Crude FF (GE Healthcare) column equilibrated in 20 mM Tris (pH 8.5) and 500 mM NaCl. The column was washed in 20 mM Tris (pH 8.5) and 500 mM NaCl and then eluted by 250 mM imidazole. The protein was polished on a Superdex 200 Increase (24 ml; GE Healthcare) column equilibrated in 20 mM HEPES (pH 7.5) and 150 mM NaCl.
The murine C3 C345c domain (residues 1517–1663) recombinantly fused to a hexahistidine-tagged thioredoxin (Trx) was expressed by transforming the domain-encoding pETM-20 plasmid into E.coli T7 SHuffle cells (New England BioLabs). Protein expression, harvest, sonication, and IMAC were performed as described above. The IMAC eluate was dialyzed at room temperature against 20 mM Tris (pH 8.5), 500 mM NaCl, and 0.5 mM EDTA in the presence of 1:50 (w/w) histidine-tagged tobacco etch virus (TEV) to cleave the Trx-6× His tag. The TEV-cut protein was loaded onto a HisTrap FF Crude (GE Healthcare) column equilibrated in 20 mM Tris (pH 8.5), 500 mM NaCl, and 0.5 mM EDTA to capture the TEV-his protease and uncut domain. The resulting flow through was dialyzed against 20 mM MES (pH 6.5) and 50 mM NaCl, applied to a Source 15Q (GE Healthcare) column equilibrated in 20 mM MES (pH 6.5) and 20 mM NaCl and subsequently eluted by a linear gradient from 20 to 500 mM NaCl. The domain was subjected to a polishing step as described above. The human C3 C345c domain (residues 1501–1663) was likewise expressed as a fusion protein with Trx and purified using a similar protocol.
FB and murine C3b were purchased from Complement Technology, cobra venom factor (CVF) was purified as described (33), and FD was purified as described (31). FP and FB (D279G and S699A) were both expressed and purified from HEK293 cells as described (14), whereas mini-FH was purified from HEK293 supernatants as described (13). CR3 headpiece was purified from the supernatant of stably transfected HEK293S cells as described (R.K. Jensen, G. Bajic, M. Sen, T.A. Springer, T. Vorup-Jensen, and G.R. Andersen, manuscript posted on bioRxiv, DOI: 10.1101/2020.04.15.043133). C4 was purified from outdated plasma, and C4b was generated from the purified C4 as described (34). C3 was purified from outdated plasma as described (30). C3b was generated from the purified C3 using 1% (w/w) trypsin digestion and incubation for 4 min. The resulting C3b was purified as described (31). C3 methylamine (C3MA) was likewise generated from C3 and purified as described (31). To generate C3c, 0.2% (w/w) FI (Complement Technology) and 1% (w/w) FH (Complement Technology) was added to C3b, and the mix was incubated for 24 h at 37°C followed by purification as described (31). Biotinylated iC3b was prepared by biotinylating the TE cysteine in C3b. Next, the biotinylated C3b was cleaved to iC3b through FH-mediated cleavage by FI as described (30).
Structure determination
To obtain a low-resolution structure of the hC3Nb3–C3b complex, negative stain electron microscopy (nsEM) was performed. To this end, C3b and hC3Nb3 were mixed in a 1:2 M ratio, incubated on ice for 5 min, and then applied to a Superdex 200 Increase (GE Healthcare) column equilibrated in 20 mM HEPES (pH 7.5) and 150 mM NaCl. A carbon-coated copper grid (Gilder 400-C3) was glow discharged at 25 mA for 45 s on an easiGlow Glow Discharge System (PELCO), and we added 3 μl of the complex at ∼20 μg/ml, obtained from early peak fractions, to the grid. Excess sample was blotted away, another 3 μl drop of 2% (w/v) uranyl formate was added and blotted away, and 3 μl of 2% (w/v) uranyl formate was added. The grid was incubated for 45 s, blotted, and transferred to a FEI Tecnai G2 Spirit transmission electron microscope, operated at 120 kV. Images were acquired using Leginon (35) with a defocus range from −0.7 to −1.7 μm and a magnification of ×67,000. The hC3Nb3:C3b particles were picked from micrograph images using Computational Imaging System for Transmission Electron Microscopy (36), and two-dimensional (2D) classification was performed on the picked particles using RELION (37). Stochastic gradient descent in RELION was used to obtain an initial input model for three-dimensional (3D) classification, which was likewise performed in RELION.
For crystallization of the C345c–hC3Nb3 complex, the nanobody was mixed in a 1.1-fold molar excess to murine C3 C345c domain. The complex was subjected to size exclusion chromatography (SEC) on a Superdex 75 (GE Healthcare) column equilibrated in 20 mM HEPES (pH 7.5) and 150 mM NaCl. Drops for crystallization were prepared by mixing the complex at 10 mg/ml in a 1:1 ratio with the reservoir solution containing 17.5% PEG 4K, 33 mM NaOAc (pH 4.3), 66 mM NaOAc (pH 5.3), and 0.2 M (NH4)2SO4. Crystals were grown in a sitting drop vapor diffusion tray stored at 19°C. Crystals were cryoprotected in reservoir solution supplemented with 20% glycerol before being flash frozen in liquid nitrogen. Data were collected at the ID23-2 beamline (The European Synchrotron Radiation Facility, Grenoble, France). Diffraction data were processed with XDS (38), and the structure was determined by molecular replacement using coordinates from the C345c domain of human C3b (Protein Data Bank [PDB] entry 6EHG) (30) and Nb36 (PDB entry: 5NLU) (39) without CDR3 in Phaser (40). The model was iteratively rebuilt in Coot (41) and refined using Phenix.refine (42). Structure factors and coordinates are deposited in the PDB as entry 6XZU (www.rcsb.org).
Proconvertase assembly, CVFBb cleavage, and FI/FH cleavage assays
In the proconvertase assembly assay, C3b was mixed with a 1.2-fold molar excess of the inactive, stabilized FB (D279G, S699A) in either the presence or absence of 2-fold molar excess of the hC3Nb3 nanobody. The S699A mutation inactivates the FB, whereas the D279G mutation extends the half-life of the convertase (43). The mix was incubated for 15 min in running buffer (20 mM HEPES [pH 7.5], 150 mM NaCl, and 5 mM MgCl2) and then applied to a Superdex 200 Increase column equilibrated in the running buffer.
The CVFB proconvertase was prepared by adding to stabilized FB (D279G) a 1.2-fold molar excess of CVF from Naja naja cobra venom in 20 mM HEPES (pH 7.5), 150 mM NaCl, and 2 mM MgCl2. Then, 10% (w/w) FD, relative to FB, was added to activate the proconvertase, and the mix was incubated for 15 min at room temperature and subsequently 10 min at 4°C. Meanwhile, C3 was added to a 2-fold molar excess of hC3Nb3. C3 in either the presence or absence of hC3Nb3 was added in a 10-fold molar excess to the proconvertase. The mix was incubated for 0.5, 1, 2, 4, 8, or 24 h at 4°C.
To examine FI degradation, C3b was incubated for 5 min on ice in either the presence or absence of a 1.2-fold molar excess of hC3Nb3. After incubation, we added 1% (w/w) FI (Complement Technology) and 0.2% (w/w) FH (Complement Technology) relative to C3b. The mix was incubated in 20 mM HEPES (pH 7.5) and 150 mM NaCl for 1, 2, 4, 8, or 24 h at 37°C.
C3 deposition assays
2CO3 (pH 9.6; Ampliqon) containing 15 μg/ml heat-aggregated IgG, 1 μg/ml mannan, or 20 μg/ml zymosan (Z4250; Sigma), respectively. In CP and LP assays, the plate was blocked for 1 h in TBS (10 mM Tris [pH 7.4] and 140 mM NaCl), 5 mM CaCl2, and 0.1% (w/v) human serum albumin (HSA) at room temperature, followed by three washes in TBST (TBS, 0.05% [v/v] Tween 20). Next, 100 μl 0.2% normal human serum (NHS) for CP assays or 1% NHS for LP assays in veronal buffered saline (VBS) (5 mM barbital [pH 7.4], 145 mM NaCl, 0.25 mM CaCl2, and 0.8 mM MgCl2 [Lonza: 12-624E]) supplemented with nanobody at the indicated concentration was added to each well. The plate was incubated at 37°C for 1 h or 30 min for CP or LP assays, respectively. Wells were washed three times in TBST and then incubated for 2 h at room temperature in 100 μl 0.5 μg/ml biotinylated rabbit anti-C3d Ab (A006302-2; Dako) in TBST. The wells were washed three times and then incubated for 1 h in 1 μg/ml Eu-labeled streptavidin (1244-360; PerkinElmer) in TBST-added 25 μM EDTA. The wells were subjected to three washes in TBST and then incubated in 200 μl enhancement buffer (Q99800; Ampliqon) for 2 min. Last, the fluorescence signal, read as time-resolved fluorometry, was measured on a VICTOR3 Multilabel Plate Counter (PerkinElmer). The AP assays were performed similarly; however, the zymosan-coated plates were blocked in TBS and 0.1% HSA, nanobodies were diluted in 11% NHS in VBS/EGTA/Mg2+ (VBS, 10 mM EGTA, and 4.2 mM MgCl2), and the incubation time on the plate was 1.5 h. AP assay in murine serum was performed as the AP assay in human serum; however, 5% serum from male C57Bl6 mice was used instead of NHS. The deposited murine C3 degradation fragments were detected using rat anti-mouse C3 (catalog no. CL7503NA; Cedarlane Laboratories), followed by the addition of biotin rabbit anti-rat IgG (biotinylated product E0468; Dako). The biotinylated Abs were detected using europium-labeled streptavidin as described above.
Total complement activation assay
The hC3Nb3 was diluted in PBS, Ca2+, and Mg2+ and added at concentrations from 0 to 16.6 μM to an 80% pool of lepirudin-treated (Refludan) (Celgene, Boudry, Switzerland) human plasma. Samples were prepared in duplicates, incubated for 5 min at 37°C, and immediately frozen at −80°C. The inhibitory effect of hC3Nb3 in the CP, the LP, and the APs was quantified using the WIESLAB Complement System assay kits (Svar Life Science, Malmø, Sweden), performed according to the manufacturer’s instructions. In short, samples were added to wells coated with a pathway-specific stimulant (IgM, mannan, and LPS for CP, L,P and AP, respectively). The wells were incubated and washed, and the deposition of C5b-9 was quantified by a detection Ab that recognizes a neoepitope in C9 (aE11) when incorporated in the full C5b-9 complex. The assay was performed in duplicates.
Hemolysis assay
CP-mediated hemolysis assay was performed as described (31, 44). To allow binding of nanobodies to serum C3, nanobody in 65 μl of PBS was added to 15 μl of serum and 120 μl of VBS/Ca2+ buffer (VBS, 2 mM CaCl2) and incubated for 2 h at 37°C. In a 96-well microplate setup, 60 μl of the serum–nanobody mix was added to 30 μl 6% Ab-decorated SRBCs [prepared as described (31)] and incubated for 2 h at 37°C. After incubation, further lysis was stopped by the addition of 5 mM EDTA and 0.9% (w/v) NaCl, and cells were pelleted by a 10-min centrifugation step at 2000 rpm at 4°C. The absorbance at 405 nm of the supernatant, reflecting the degree of lysis, was measured using a VICTOR3 Multilabel Plate Counter. The assay using NHS was performed in triplicates, whereas the assay using FB-depleted serum (Complement Technology) was performed in duplicates. The assay quantifying the effect of the recombinant C3 C345c domain was performed similarly and was performed in duplicates.
Bio-layer interferometry
All bio-layer interferometry (BLI) experiments were performed on an Octet Red96 (ForteBio). Running buffer was 20 mM HEPES (pH 7.5), 150 mM NaCl, and 0.05% Tween 20, and the experiments were performed at 30°C and shaking at 1000 rpm unless otherwise stated. For kinetics experiments of human C3, C3b, C3MA, C4b, and murine C3b, hC3Nb3 was immobilized on Anti-Penta-HIS sensors (HIS1K; ForteBio) through the C-terminal hexa-His tag of the nanobody. The hC3Nb3-coated sensors were transferred to human C3b, C3MA, or C3 at concentrations of 0.31, 0.625, 1.25, 2.5, 5, or 10 nM for 180 s followed by a 180-s dissociation step in running buffer. Analysis of binding of C4b was performed similarly; however, sensors were transferred to 62.5, 125, 250, 500, or 1000 nM C4b. For analysis of hC3Nb3 mutants, the L48A, the S54A, or the T102A were immobilized on sensors as above and transferred to 0.31, 0.625, 1.25, 2.5, 5, or 10 nM human C3b followed by a dissociation step as above. The hC3Nb3 Y53A mutant was analyzed similarly; however, only sensorgrams from 0.31, 0.625, 1.25, 2.5, and 5 nM C3b could be fitted. The binding of murine C3b was analyzed similarly; however, using 0.31, 0.6, 2.5, 5, 10, or 20 nM murine C3b and the running buffer in this experiment was 20 mM HEPES (pH 7.5) and 150 mM NaCl. Data were fitted using a 1:1 Langmuir binding mode using GraphPad Prism version 6 after background subtraction of the signal from a hC3Nb3-coated sensor, dipped in running buffer.
For FP competition assay, mini-FH was immobilized on amine reactive sensors (AR2G; ForteBio) as described (13). The mini-FH–coated sensors were equilibrated for 5 min in PBS supplemented with 1 mg/ml BSA and 0.05% Tween 20 prior to the experiment. Next, a 60-s association step was performed in C3b (280 nM) alone or in the presence of hC3Nb3 (714 nM), FP (94 nM), or both hC3Nb3 and FP. Upon association, the sensors were transferred to PBS supplemented with 1 mg/ml BSA and 0.05% Tween 20 for a 60-s dissociation step. The sensors were regenerated in PBS supplemented with 4 M NaCl, and the experiment was repeated.
For the hC3Nb3 mutant competition assay, C-terminally Avi-tagged hC3Nb3 was purified as described above and biotinylated by incubating the protein with a 20-fold excess (w/w) of BirA ligase in 20 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 2 mM ATP, and 0.15 mM d-biotin for 16 h at room temperature. The reaction was subsequently loaded onto a Source 15S (GE Healthcare) column equilibrated in 20 mM sodium acetate (pH 5.5) and 20 mM NaCl, and the biotinylated nanobody was eluted by a linear gradient from 20 to 500 mM NaCl.
For the initial kinetics experiments, biotinylated Avi-tagged hC3Nb3 at 1 μg/ml was immobilized on Streptavidin Biosensors (ForteBio). The nanobody-coated sensors were transferred to 9, 3, 1, or 0.333 nM C3b for 600 s and then a dissociation step was performed in 20 mM HEPES (pH 7.5) and 150 mM NaCl for 1200 s. The experiment was performed in duplicate, and data were fitted as above. For hC3Nb3 mutant competition assay, 10 nM C3b was incubated in 100 nM of the indicated hC3Nb3 mutant for 30 min at room temperature prior to association.
CR3 hC3Nb3 competition surface plasmon resonance experiment
The surface plasmon resonance (SPR) experiments were performed on a Biacore T200 instrument using CMD500M chips (XanTec Bioanalytics) equilibrated in the running buffer (20 mM HEPES [pH 7.5], 150 mM NaCl, 5 mM MgCl2, and 1 mM CaCl2). The flow cell was prepared by immobilizing streptavidin to 200 response units followed by saturation by biotinylated iC3b. The immobilized iC3b was saturated by hC3Nb3 followed by the application of the CR3 headpiece at concentrations of 1.25, 2.5, 5, 10, 12.5, 25, 50, and 100 nM. The resulting binding curves were fitted using a 1:1 binding model. The experiment was performed in triplicate.
hC3Nb3 gel shift assays
Human C3 C345c domain and hC3Nb3 were diluted in 20 mM HEPES (pH 7.5) and 150 mM NaCl and incubated separately for 30 min in either the presence or absence of 2.5% SDS. Next, human C3 C345c domain was mixed with a 1.2-fold molar excess of hC3Nb3 and incubated for 5 min at room temperature before analysis by SDS-PAGE. For the in-gel fluorescence experiment, hC3Nb3-GFP was mixed in a 2-fold molar excess with human C3 C345c domain, C3c, C3b, C3MA, C3, or C3 in plasma, assuming a C3 concentration of 1 mg/ml. The mix was incubated for 10 min at room temperature, and then the mix was subjected to SDS-PAGE. The in-gel fluorescence was scanned on a Typhoon Trio Variable Mode Imager (GE Healthcare) using excitation at 488 nm and emission at 520 nm. The gels were subsequently stained using Coomassie Brilliant Blue. To demonstrate the use of hC3Nb3-GFP in C3c purification, 600 ml plasma was incubated for 19 d in turns at 4°C and at room temperature while the C3c degeneration was monitored using hC3Nb3-GFP as described above. Upon near complete cleavage of C3 to C3c, PEG precipitation and DEAE ion exchange was performed as described (45). To identify C3c-containing fractions from DEAE ion exchange, 10 μl of the fractions were mixed with 1 μg hC3Nb3-GFP and separated on SDS-PAGE followed by in-gel fluorescence scanning as described above. For hC3Nb3-GFP–assisted C3c quantification, 3 μl of C3c was mixed with increasing amounts of hC3Nb3-GFP and then subjected to SDS-PAGE analysis and scanning as described above.
Flow cytometry
For preparation of NHS, 4 ml blood was drawn into tubes with clot activator from each of 10 random blood donors at the Department of Clinical Immunology, Aarhus University Hospital (Aarhus, Denmark). After 45 min at ambient temperature, tubes were centrifuged (2000 × g for 5 min). Serum was collected, pooled, aliquoted, and stored at −80°C until use. The use is in accordance with contemporary Danish legislation and was approved by The Danish Data Protection Agency (reference no. 1-16-02-40-12/2007-58-0010) and the Ethics Committee in Central Denmark Region (reference no. 1-10-72-127-12). Pig RBCs were obtained from venous EDTA-stabilized blood from pigs undergoing experimental surgery at Institute of Clinical Medicine, Aarhus University (Aarhus, Denmark). RBCs were fixed with glutaraldehyde as described (46). Pig RBCs present the carbohydrate moiety terminal Galα3Gal [around 6 × 104 residues per cell (47)], recognized by Abs occurring naturally in human plasma [average concentration of IgG anti-Galα3Gal ∼10 mg/l and IgM anti-Galα3Gal ∼60 mg/l (48)], and Ab binding initiates CP (49). Pig RBCs at 5000/μl were incubated in RPMI 1640 with HSA at 1 g/l and 4% NHS. Tubes containing volumes of 20 μl were incubated at 37°C for 2 h. RBCs were then washed twice in PBS by centrifugation at 200 × g for 10 min. Pig RBCs were incubated in 20 μl PBS with HSA at 1 g/l and soluble C3b or C4b prior to the addition of hC3Nb3-GFP. The mix was incubated for 30 min at room temperature in the dark before cells were resuspended in 100 μl PBS. Cells were analyzed on a NovoCyte Quanteon (ACEA Biosciences, San Diego, CA). GFP was excited at 488 nm, and emission was measured at 530/30 nm. Data are reported as median fluorescence intensity from cells.
Neuronal differentiation and complement deposition
Using the protocol of combined overexpression of the transcription factor neurogenin 2 (50) and inhibition of SMAD and WNT signaling previously described (51), induced pluripotent stem cells were differentiated into human neurons. For complement deposition assays, neurons were differentiated in a 96-well format at 30,000 cells/cm2. In the following, all reagents were warmed to 37°C prior to the addition unless otherwise stated. Wells were washed in gelatin veronal buffer supplemented with 0.5 mM MgCl2 and 0.15 mM CaCl2 (Complement Technology), and then we added 100 μl NHS (Complement Technology) or C3-depleted serum (Complement Technology) diluted to 10% in gelatin veronal buffer supplemented with 0.5 mM MgCl2 and 0.15 mM CaCl2 z-stacks spanning 2.0 μm.
Results
hC3Nb3 binds the C345c domain of C3
We immunized a llama with human C3b and performed the subsequent nanobody selection by phage display with recombinant murine C3 C345c domain comprising the amino acid residues 1517–1663 (pre–pro numbering). Thus, this strategy favors selection of nanobodies that react with both human and murine C3. In the phage display system, a single clone dominated among the reactive phages. We cloned the associated nanobody, hC3Nb3, into a bacterial expression plasmid conferring a hexa-His tag allowing recombinant expression and purification.
We used nsEM to investigate the interaction between hC3Nb3 and C3b. We formed the hC3Nb3–C3b complex, isolated the complex by SEC, and applied the complex to electron microscopy grids. In the 2D class averages obtained, the C345c domain, the macroglobulin (MG) ring, and the TE domain were readily recognized. However, the C345c domain appeared extended, in accordance with the nanobody binding to this domain (Fig. 1A). We obtained an initial model by stochastic gradient descent, which allowed 3D classification of the particles. The subsequent analysis of view distribution revealed preferred orientations among the particles (Supplemental Fig. 1A), which may distort the 3D reconstruction. Despite the limited number of orientations, we obtained a trustworthy 3D reconstruction that revealed recognizable features, including the C3b MG ring and the TE domain (Fig. 1B, Supplemental Fig. 1B). When fitting the structure of C3b (30) into the density, the CUB, the TE, and the C345c domains were outside the density, which agrees with flexibility of these domains. We therefore fitted the C345c domain and the CUB-TE moiety individually. Similar to the 2D class averages, the 3D reconstruction revealed an extra density at the C345c domain (Fig. 1B). This binding site agrees with the selection strategy, and because we immunized the llama with full-length C3b, the binding site should be accessible in C3b.
Structural characterization of the hC3Nb3–C345c complex. (A) nsEM 2D class averages of hC3Nb3–C3b complex. Asterisk (*) marks the likely position of the hC3Nb3 nanobody, bound to the C345c domain. The expanded view shows the recognizable TE domain, the MG ring, and the C-terminal C345c domain. Scale bar, 100 Å. (B) 3D reconstruction of the hC3Nb3–C3b particles. The crystal structure of C3b (green) (PDB entry 6EHG) was docked into the envelope by fitting the CUB-TE moiety, the C345c domain, and the body of the C3b individually. The asterisk (*) marks the likely position of hC3Nb3. (C) Cartoon representation of the crystal structure of hC3Nb3 in complex with the C345c domain of murine C3. Arrows mark the position of the CDR2 and CDR3 loops as well as the α1, α2, and α3 helices of the C345c domain. (D) Omit 2mFo-DFc map contoured at 1.5 σ around the hC3Nb3 CDRs. (E and F) The interface between hC3Nb3 and the C345c domain, with amino acids engaging in hydrogen bonds (dotted lines) presented as sticks. (G) Amino acid sequence alignment of the murine and human C3 C345c α1, α2, and α3 helices, which comprise the main interaction site of hC3Nb3 on C3. Green shadings mark amino acids that differ between the two species. Triangles indicate amino acids that engage in hydrogen bonds with hC3Nb3.
To gain detailed information into the nanobody–Ag complex, we crystallized a complex between the hC3Nb3 and the murine C3 C345c domain and collected diffraction data extending to 1.5 Å resolution. We solved the structure by molecular replacement and iteratively manually built and refined the structure (Table I). The structure revealed that the nanobody binds the C-terminal α3 helix of the C345c domain (Fig. 1C). Unambiguous omit 2mFo-DFc electron density were obtained for the three nanobody CDRs (Fig. 1D). This binding site agrees with the data obtained by nsEM, in which the C345c domain appeared extended in the distal direction compared with the body of the C3b molecule. The hC3Nb3 nanobody binds the C345c domain in a sideways manner in which the framework regions of the nanobody heavily participate in the binding. In this way, hC3Nb3 forms a concave binding site facilitated by CDR3 and side chains of the CDR2 that wraps around the α3 helix. This sideways binding positions the framework regions two and three of hC3Nb3 in close proximity to the α3 helix. Using the Proteins, Interfaces, Structures and Assemblies software (52), we identified six hydrogen bonds between the nanobody and the C3b C345c domain. The main interaction site lies at the Glu1654 of the C345c domain (Fig. 1E). This side chain protrudes from the α3 helix into an electropositive cavity formed by the framework region three and the CDR3 of hC3Nb3. In the cavity, Tyr38 in framework region three and Thr102 as well as the main chain of Gly100 from CDR3 form hydrogen bonds with C3 Glu1654. In the distal end of the α3 helix, C3 Tyr1659 protrudes into a hydrophobic pocket formed by the framework regions two and three of hC3Nb3. In this pocket, the Tyr1659 side chain forms a hydrogen bond to the main chain of Val49 (Fig. 1E). Between these two cavities, Tyr53 at the proximal end of CDR2 forms a hydrogen bond with the Ser1655 in the α3 helix. Last, the sole interaction outside the C3 α3 helix involves Asn1530 in the α1 helix of the C345c domain, which forms a hydrogen bond to Ser54 in CDR2 (Fig. 1F). An inspection of the amino acid sequence of the helices of human and mouse C345c domains revealed conservation of the hC3Nb3 binding site between the two species (Fig. 1G). The two species share the central residues of the binding site, Glu1654 and Ser1655. However, the Tyr1659 of murine C3 is exchanged for a Phe residue in humans, which would similarly fit into the hydrophobic cavity described above. Last, a Glu residue is found at the position of Asn1530 in human C3. In summary, our structural studies mapped the binding site of hC3Nb3 to two α-helices located in the C-terminal domain of C3, and our structure is in accordance with species cross-reactivity against human and mouse C3.
hC3Nb3 inhibits AP proconvertase assembly and delays FI cleavage
We then turned to test the functional properties of hC3Nb3. We superimposed our x-ray structure onto the current model for C3 recognition by the C3 convertases (53, 54) and the structure of C3bB (55) (Supplemental Fig. 2A, 2B). These comparisons indicated that hC3Nb3 would not interfere with substrate binding by the C3 convertase, whereas it would compete with FB for binding to C3b. We first evaluated the idea that hC3Nb3 inhibits the formation of the C3bB proconvertase by attempting to assemble C3bB in either the presence or absence of hC3Nb3. The outcomes were assessed by subjecting the reaction mixture to SEC analysis in which formation of C3bB gave rise to an increase in the absorbance and an earlier elution volume compared with C3b alone. In the presence of hC3Nb3, lower absorbance was observed, indicating that the nanobody competes with FB for binding to C3b (Fig. 2A). Subsequent SDS-PAGE analysis of the peak fractions from the SEC analysis further supported such competition as when hC3Nb3 was present, FB was absent in the early peak fractions (Fig. 2B). We next assessed whether the nanobody interferes with substrate binding by the C3 convertase. In this study, we incubated CVF from N. naja cobra venom with FB and activated the resulting CVFB proconvertase using FD. Subsequently, we added C3 with or without hC3Nb3 and monitored C3 cleavage by SDS-PAGE analysis. In agreement with the concept that hC3Nb3 does not inhibit C3 cleavage, the progression of C3b formation occurred at a similar rate with or without the nanobody (Fig. 2C).
The hC3Nb3 inhibits AP proconvertase assembly and FI degradation but not C3 cleavage. (A) AP proconvertase assembly assay. SEC elution profiles of runs containing C3b alone (dashed, black), C3b in the presence of a 1.2-fold molar excess of FB (D279G, S699A) (gray), or C3b in the presence of both FB (D279G, S699A) and a 2-fold excess of hC3Nb3 (black) are presented. (B) Fractions marked by bars in (A) were analyzed by nonreducing SDS-PAGE. This shows that hC3Nb3 prevents proconvertase assembly. (C) CVFBb cleavage assay. The C3 cleavage by CVFBb at 4°C for the indicated time was monitored by SDS-PAGE in the presence or absence of hC3Nb3. (D) FH-mediated FI cleavage assay. C3b was incubated with 1:500 (w/w) FH and 1:100 (w/w) FI for the indicated time at 37°C in either the presence or absence of a 1.2-fold molar excess hC3Nb3, and the C3b degradation was monitored by SDS-PAGE.
The binding site of hC3Nb3 on the C-terminal domain of C3 constitutes an important binding hub for regulators of the complement system, including the C3b degrading protease, FI (56). By comparison of our structure of the hC3Nb3–C345c complex with the structure of the C3b–FI–FH complex (56) (Supplemental Fig. 2C), we predicted that hC3Nb3 binding to C3b would weaken the interaction between FI and the C3b C345c domain, hence inhibiting C3b conversion to iC3b. We tested the interference with FI by incubating C3b, FI, and FH with or without hC3Nb3. In the absence of hC3Nb3, the cleavage progressed readily, and only trace amounts of intact C3b remained after 8 h of incubation (Fig. 2D). In contrast, hC3Nb3 delayed the cleavage progression and allowed only limited C3b degradation upon 8 h of incubation, whereas after 24 h of incubation, an almost complete conversion of C3b to iC3b had occurred (Fig. 2D). The data support that hC3Nb3 sterically competes with FI, but the slow cleavage in the presence of hC3Nb3 hints that degradation of C3b may occur without the interaction of FI with the C3 C345c domain. Overall, these biochemical experiments showed that hC3Nb3 prevents assembly of the AP C3 convertase and support the location of the epitope observed in the crystal structure of the hC3Nb3–C345c complex.
hC3Nb3 inhibits AP C3 degradation
The nanobody inhibited the assembly of the AP proconvertase but did not inhibit the substrate cleavage by the CVFBb C3 convertase. Hence, the data presented thus far suggest that the nanobody would inhibit the AP C3 convertase but not the CP C3 convertase. To examine this question, we performed a C3 fragment deposition assay in ELISA wells coated with either zymosan (in a buffer inhibiting CP and LP), heat-aggregated IgG, or mannan to activate AP, CP, or LP, respectively. We compared the function of the nanobody to the AP inhibitor hC3Nb1, its inactive mutant hC3Nb1 (W102A) (30), and the AP/CP/LP inhibitor hC3Nb2 (31). The data revealed that hC3Nb3 inhibited the progression of the AP-mediated C3 fragment deposition when present in a 1:1 nanobody to C3 molar ratio (Fig. 3A), whereas hC3Nb3 did not influence C3 fragment deposition through the CP (Fig. 3B) or LP (Fig. 3C). As expected, hC3Nb2 inhibited all pathways, whereas hC3Nb1 inhibited only the AP. We expected that hC3Nb3 also inhibited AP activation in mouse serum because the nanobody was selected against the murine C3 C345c domain, which is very homologous to the human counterpart (Fig. 1G). To test this, we activated the AP in murine serum and measured the C3 fragment deposition. Again, we compared the effect of the nanobody to that of two cross-reactive nanobodies, hC3Nb1 (30) and hC3Nb2 (31). Surprisingly, hC3Nb3 displayed low inhibition potential in the murine system (Fig. 3D). This may indicate that in serum, the nanobody is less efficient at binding to the murine C3 C345c domain compared with the homologous human domain.
hC3Nb3 inhibits AP-mediated C3 fragment deposition as well as CP-mediated terminal C5b-9 deposition and hemolysis. (A) hC3Nb3 inhibits AP-mediated C3 fragment deposition onto a zymosan-coated surface in 11% NHS. (B) hC3Nb3 does not inhibit the CP-mediated C3 fragment deposition onto a surface of heat-aggregated IgG in 0.2% NHS. (C) hC3Nb3 does not inhibit C3 fragment deposition mediated by LP onto a mannan-coated surface in 1% NHS. (D) As in (A), but AP was assayed in 5% murine serum. (E) hC3Nb3 inhibits C5b-9 deposition through all pathways. AP, CP, or LP were induced in an 80% lepirudin-treated plasma pool, and C5b-9 deposition was quantified. (F–H) CP was assayed on SRBCs using anti-sheep erythrocyte Abs, and the hemolysis assays was performed in 7.5% NHS (F and H) or 7.5% FB-depleted serum (G). In (A)–(D), (F), and (G), the effects of hC3Nb3 (blue curve) were compared with the AP inhibitor hC3Nb1 (gray curve) (30), inactive hC3Nb1 (W102A) mutant (green curve) (30), and the AP/CP/LP inhibitor hC3Nb2 (red curve) (31). In (H), the effect of C3 C345c (purple curve) was compared with the hC3Nb1, hC3Nb2, and hC3Nb3 using the same color scheme as in (A)–(D, (F), and (G). C3 deposition in (A)–(D) was normalized to serum without nanobodies (100% deposition). Lysis in (F)–(H) was normalized to PBS (0% lysis) and lysis by H2O (100% lysis). The black dashed lines indicate the putative C3 concentrations at the given serum dilutions, assuming a C3 concentration in undiluted human and murine serum of 5.3 μM (67) and 4.2 μM (68), respectively. Gray dashed line in (H) indicates putative C5 concentration, assuming a C5 concentration of 0.42 μM (69) in undiluted human serum. Graphs of hC3Nb1, hC3Nb1 (W102A), and hC3Nb2 in (A), (B), (F), and (G) were originally published in the Journal of Biological Chemistry. Pedersen, H., et al. (31), © the Author(s). Average and SD are shown for n = 3 experiments in (A)–(C) as well as (F) and n = 2 experiments in (D) and (E) as well as (G) and (H).
hC3Nb3 inhibits CP-mediated C5b-9 formation and erythrocyte lysis
We next investigated the effect of the nanobody on the complement terminal pathway. First, we triggered either the CP, LP, or the AP pathway and monitored the C5b-9 deposition at increasing concentrations of hC3Nb3 (Fig. 3E). To approximate physiological conditions, we used a lepirudin-stabilized pool of 80% human plasma as a complement source in these experiments. The nanobody inhibited terminal pathway progression after AP activation (Fig. 3E) as expected. However, hC3Nb3 also inhibited terminal pathway progression after CP as well as LP activation, which was unexpected because the nanobody had not influenced the CP- or LP-mediated C3 fragment deposition (Fig. 3B, 3C). To examine this finding further, we tested the effect of hC3Nb3 in a CP-mediated sheep erythrocyte hemolysis assay. SRBCs were incubated with anti-SRBC Abs before adding human serum with hC3Nb3 or one of the control nanobodies: hC3Nb1 (inhibitor of the AP C3 convertase), hC3Nb1 (W102A) (inactive version), or hC3Nb2 (inhibitor of both C3 convertases). Consistent with our C5b-9 deposition assay, we observed that hC3Nb3 inhibited CP-mediated hemolysis (Fig. 3F). In contrast, hC3Nb1 did not inhibit progression of hemolysis, which confirmed the dependence on the CP in the assay. We also performed the assay in FB-depleted serum, which does not support AP activation. Again, hC3Nb3 inhibited progression of the hemolysis when present in a 1:1 M ratio compared with C3 (Fig. 3G), confirming the inhibition of activation via CP. The inhibitory effect of hC3Nb3 on CP-mediated hemolysis may indicate an involvement of the C3 C345c domain of the auxiliary C3b in the CP C5 convertase. To test this notion, we recombinantly expressed the C345c domain of human C3 and analyzed the effect of the domain in CP-driven hemolysis. Although the hC3Nb3 and hC3Nb2 efficiently inhibited hemolysis, we were unable to detect any effect of the recombinant domain alone (Fig. 3H). Interestingly, we consistently observed a more efficient inhibition by hC3Nb2 than hC3Nb3, which may reflect their different mechanisms of inhibition. Whereas hC3Nb2 inhibits the cleavage of C3 by the CP C3 convertase early in CP-driven hemolysis, hC3Nb3 most likely inhibits the hemolysis by binding the C3b in the C5 convertase at a late step in the hemolysis reaction. Overall, these functional data demonstrated that hC3Nb3 inhibits CP-driven activation of the terminal pathway.
hC3Nb3 binds C3 and C3b with high affinity
Our functional assays demonstrated that hC3Nb3 potently inhibits the C3 fragment deposition when present in an equimolar concentration compared with C3 in human serum. To obtain insight into the binding properties of hC3Nb3, we performed BLI. To this end, we immobilized his-tagged hC3Nb3 on Anti-Penta-HIS biosensors and exposed the sensors to human C3b in concentrations ranging from 0.31 to 10 nM (Fig. 4A). We fitted the resulting sensorgrams using a 1:1 association model and calculated the binding constants. The analysis revealed that hC3Nb3 bound strongly to human C3b with a KD of 2.6 nM. We next analyzed the binding of the nanobody to native C3 and the C3 (H2O) analogue C3MA by immobilizing hC3Nb3 as above and applying 0.31–10 nM native C3 (Fig. 4B) or C3MA (Supplemental Fig. 3A). The subsequent analysis revealed that hC3Nb3 bound native C3 with a KD of 3 nM and C3MA with a KD of 6.2 nM. We analyzed the binding of the nanobody to murine C3b (Fig. 4C) and found that hC3Nb3 binds murine C3b with a KD of 3 nM. Last, we assessed the binding of hC3Nb3 to C4b by applying a concentration ranging from 62.5 to 1000 nM (Supplemental Fig. 3B). Although we were unable to unequivocally fit the resulting sensorgrams, we obtained an apparent KD of ∼1.3 μM. Fig. 4D summarizes the rate constants for binding and dissociation of hC3Nb3. Collectively, these BLI data demonstrate that the nanobody exhibits species cross-reactivity with murine C3 and that it binds human C3b as well as native C3 and C3MA with low nanomolar affinity. The data furthermore indicates that the effect of hC3Nb3 on CP-mediated hemolysis does not arise from inhibition of the CP C3 convertase because hC3Nb3 only weakly interacts with C4b.
The hC3Nb3–C3 interaction is of high affinity and inhibits association of FP. hC3Nb3 was immobilized on a BLI sensor via the C-terminal His tag and dipped in (A) 10, 5, 2.5, 1.25, 0.625, or 0.31 nM human C3b, (B) 10, 5, 2.5, 1.25, 0.625, or 0.31 nM human C3, or (C) 20, 10, 5, 2.5, 0.625, and 0.31 nM murine C3b. Gray and black sensorgrams represent curves fitted to a 1:1 binding model and experimental data, respectively. (D) Binding kinetics of hC3Nb3 derived from BLI presented in (A)–(C) and Supplemental Fig. 3A, 3B. (E) Mutant competition assay. Wild-type hC3Nb3 was immobilized on the BLI sensors via a C-terminal Avi tag, and the binding to either 10 nM C3b or 10 nM C3b mixed with a 10-fold molar excess of the indicated hC3Nb3 mutant was analyzed. (F) Properdin competition assay. Mini-FH was immobilized on BLI sensors, and the binding of C3b in the presence or absence of FP as wells as hC3Nb3 was analyzed. (G) CR3 competition assay. Biotinylated iC3b was immobilized on a SPR chip and subsequently saturated by hC3Nb3. Following saturation, CR3 headpiece at 100, 50, 25, 12.5, 10, 5, 2.5, and 1.25 nM was applied. Binding curves of FP, C3b, and C3b plus FP in (F) were published previously (13). a.u., arbitrary units; RU, response units.
We next analyzed the contribution of the individual amino acids of hC3Nb3 that engage in the C3/C3b recognition. To this end, we expressed and purified hC3Nb3 species carrying either of the mutations L48A, Y53A, S54A, or T102A predicted to interfere with epitope binding. We aimed at analyzing whether these mutants retain their ability to compete with wild-type hC3Nb3 for binding to C3b in a BLI-based setup. To avoid the competition for binding to the sensors by the C-terminally his-tagged mutants, we immobilized wild-type hC3Nb3 through a biotinylated Avi tag. First, we tested whether this immobilization strategy affected the binding kinetics of hC3Nb3. We applied C3b in concentrations from 0.3 to 9 nM and fitted the sensorgrams as above (Supplemental Fig. 3C). From this analysis, we obtained a KD of 335 pM (Supplemental Fig. 3D), confirming the high affinity binding of the nanobody. We next analyzed the binding to C3b in the presence or absence of a 10-fold molar excess of the hC3Nb3 mutants. The L48A, S54A, and T102A mutants (the latter two not shown) prevented the interaction similar to wild-type hC3Nb3 (Fig. 4E). This indicates that these mutants retained a high affinity for C3b or that the fluid phase 10-fold molar excess of these nanobodies provided sufficient competition to prevent the binding of C3b to the immobilized wild-type hC3Nb3. In contrast, the Y53A mutant was less capable of competing with the wild-type nanobody (Fig. 4E), indicating that this side chain plays an important role in Ag recognition, in accordance with its central location in the hC3Nb3:C345c interface (Fig. 1F). To gain further insight into the interaction, we immobilized the mutants and applied C3b in concentrations ranging from 0.31 to 10 nM (Supplemental Fig. 3E–H). In agreement with the competition assay, the mutants L58A, S54A, and T102A displayed similar KD as the wild-type hC3Nb3 (Supplemental Fig. 3I). In contrast, the Y53A mutation increased both the association as well as dissociation rates of the nanobody such that the KD of this mutant was 6.7 nM (Supplemental Fig. 3I). Together, the results of these biophysical experiments are in agreement with the studies of the effects of hC3Nb3 observed in human and murine serum and pinpoint Tyr53 as an important component in the hC3Nb3 paratope.
hC3Nb3 competes with properdin but not with CR3
The hC3Nb3 nanobody shares the binding site, on the C345c domain, with the positive regulator FP that similar to hC3Nb3 binds the α3 helix of the domain (13, 14, 57). We superimposed our x-ray structure onto the C3bBb–FP complex (13), which confirmed a massive overlap in the binding sites of the two proteins (Supplemental Fig. 2D). We therefore tested the competition of hC3Nb3 and FP for C3b binding. To this end, we used a BLI-based setup and immobilized mini-FH (58) through a C-terminal His tag. First, we examined whether hC3Nb3 competes with the immobilized mini-FH and found that C3b in both the presence and absence of hC3Nb3 bound the sensors (Fig. 4F). This agrees with the large spatial distance between the binding sites of FH and the C345c domain (56) (Supplemental Fig. 2C). Next, we tested for competition with FP. We were able to form the C3b–FP complex as evident by the increased response when applying both C3b and FP to the mini-FH–coated sensors compared with when adding C3b alone. However, when added together with C3b and FP, hC3Nb3 reduced the response back to that obtained with only C3b and hC3Nb3 (Fig. 4F), revealing a direct competition between hC3Nb3 and FP.
A recent structural analysis by nsEM of the iC3b–CR3 complex suggested direct contacts between regions in iC3b close to the C345c domain and the β-propeller/β I–like domain portion of the CR3 headpiece (59). We therefore investigated whether hC3Nb3 interferes with the interaction between CR3 and iC3b using SPR. We immobilized biotinylated iC3b on the SPR chip, saturated the iC3b by hC3Nb3, and subsequently applied a major part of the CR3 ectodomain called the headpiece to the immobilized iC3b–hC3Nb3 complex (Fig. 4G). We determined a KD of 41 nM for a 1:1 interaction between CR3 headpiece and the iC3b. This dissociation constant is close to the KD value of 30 nM that we obtained for the iC3b–CR3 interaction in the absence of the nanobody (R. K. Jensen et al., manuscript in preparation). Overall, these binding data demonstrated that the nanobody does not interfere with the CR3–iC3b interaction, whereas hC3Nb3 efficiently prevented C3b from interacting with FP.
The hC3Nb3–C345c complex exhibits a remarkable stability
During the characterization of hC3Nb3, we frequently observed an extra band during nonreducing SDS-PAGE analysis of the hC3Nb3–C345c complex. The extra band, at ∼40 kDa, could neither be attributed to hC3Nb3 (14 kDa) nor to C345c (17.5 kDa) alone. Intrigued by this observation, we tested whether the band arose from the formation of an SDS-resistant complex. We mixed the two components either prior to or after incubation in SDS and analyzed the complex on SDS-PAGE (Supplemental Fig. 4A). The resulting gel revealed that if we incubated either of the two components in SDS prior to mixing, they did not form an SDS-resistant complex. In contrast, when complex formation took place in the absence of SDS, the complex was stable during SDS-PAGE. These observations led us to explore the applications of this stable interaction in three different setups.
In the first setup, we evaluated the use of hC3Nb3-GFP for detection of C3 degradation products in human plasma. We mixed hC3Nb3-GFP with purified fragments of human C3, including native C3, C3MA, C3b, C3c, the human C3 C345c domain, and human plasma. We subjected these nanobody–C3 mixtures to SDS-PAGE and scanned the in-gel GFP fluorescence. The gel revealed that hC3Nb3-GFP formed stable complexes with all of the tested C3 derivatives as well as C3 in human plasma (Fig. 5A). A comparison with Coomassie Brilliant Blue staining of the gel highlighted the specificity of this method (Fig. 5A, 5B). We subsequently used hC3Nb3-GFP to guide purification of human C3c. In this study, we incubated human plasma for 19 d in turn at 4°C and room temperature to allow the degradation of C3 to C3c, whereas we monitored the degradation using the hC3Nb3-GFP (Supplemental Fig. 4B). Furthermore, we included hC3Nb3-GFP during the purification to allow the identification of fractions containing C3c (Supplemental Fig. 4C, 4D) and to titrate the C3c concentration during an early purification step (Supplemental Fig. 4E, 4F).
Applications of hC3Nb3-assisted C3 detection. (A) hC3Nb3-GFP forms an SDS-resistant complex with C3 species as detected by in-gel fluorescence. hC3Nb3 was mixed with the C3 C345c domain, C3c, C3b, C3MA, native C3, or plasma and subjected to SDS-PAGE followed by scanning of in-gel fluorescence. (B) Coomassie Brilliant Blue stain of the gel in (A). (C) Flow cytometry–based detection of C3 fragment deposition. Complement was deposited on pig RBCs by incubation in NHS, and the deposited C3 fragments were quantified using hC3Nb3-GFP in the presence of varying concentrations of purified decoy C3b (blue) or C4b (red). (D) Immunohistochemistry. Left, in vitro–differentiated neurons were incubated in C3-depleted serum. Fluorophore-conjugated hC3Nb3-IgY (green) was compared with β-III-tubulin (Tuj1, blue) and C1q (purple). Right, zoom of box indicated on left. Scale bar, 10 μm. (E) As in (D) but using NHS. Right, the zoomed image of box indicated on left shows a hC3Nb3:IgY–positive neurite with C1q-positive puncta, indicated by arrows. Scale bar, 10 μm.
In the second setup, we tested the use of hC3Nb3-GFP in flow cytometry. To this end, we provoked complement deposition on pig RBCs by incubating the cells in NHS. We subsequently detected the deposited C3b using the hC3Nb3-GFP and quantified the median fluorescence intensity from the cells. The results revealed that the hC3Nb3 efficiently bound to RBCs. To test the specificity for deposited C3 fragments, we performed the experiment in the presence of soluble human C3b or, as negative inhibitor control, soluble human C4b (Fig. 5C). Only a molar excess of C3b efficiently inhibited the hC3Nb3-GFP binding on the RBCs.
In the third setup, we tested the applicability of the nanobody for immunohistochemistry of C3 deposition on in vitro differentiated neurons. To increase compatibility with established methodologies for immunohistochemistry, we fused the hC3Nb3 to chicken IgY Fc. We incubated the differentiated neurons in either NHS or C3-depleted serum followed by staining with hC3Nb3 conjugated to 488 fluorophore and anti-C1q Ab to identify sites of complement activation and then fixed and stained the cultures for β-III-tubulin to identify neuronal processes. In C3-depleted serum, the imaging revealed recognizable neuronal processes but only sparsely distributed hC3Nb3:IgY–positive puncta (Fig. 5D), indicating low unspecific binding of the nanobody. In normal serum, the images revealed limited colocalization between β-tubulin and hC3Nb3:IgY as well as C1q (Fig. 5E), indicating limited complement activation on healthy neuronal processes. In contrast, we frequently observed hC3Nb3:IgY–positive, β-III-tubulin–negative staining in a distinct line pattern, which we ascribed to high levels of complement activation on dead or dying processes. We further observed C1q-positive puncta distributed on these processes, which substantiates the notion that this staining arises from complement activation. Collectively, these immunohistochemistry data indicate that hC3Nb3:IgY allows the detection of C3 fragments deposited in vitro on cultured neurons. Fig. 6 summarizes the properties of hC3Nb3 and compares them to hC3Nb1 and hC3Nb2.
Comparison of known anti-C3 nanobodies. (A) The table summarizes the function, KD, and interference with binding of regulators. The hC3Nb1 and hC3Nb2 are described in Jensen et al. (30) and Pedersen et al. (31), respectively. (B) The structure of hC3Nb3, guided by the hC3Nb3:C345c crystal structure, as well as the proposed position of hC3Nb2 were superimposed onto the structure of C3b in complex with hC3Nb1 (PDB entry 6EHG).
Discussion
In the current study, we describe the functional and structural characterization of the C3 specific nanobody, hC3Nb3. We show that the nanobody potently inhibits the AP by competing with FB for binding to the C3b and hence prevents AP proconvertase assembly. The nanobody similarly competes with FP, which will further contribute to inhibition of the AP. The hC3Nb3 also delays the FH-assisted FI cleavage of C3b, which would most likely also be true for other cofactors besides FH. Thus, even if the AP activity was suppressed, C3b generated by the CP C3 convertase would still be less prone to FI degradation. The net outcome of these opposing effects in an in vivo setting remains to be explored. We also demonstrated that hC3Nb3 exhibits species cross-reactivity and binds both human and murine C3b with low nanomolar affinity. This high affinity binding agrees with the selection strategy designed to favor the selection of species cross-reactive nanobodies. However, we observed that the inhibition of murine AP in serum required a higher concentration of hC3Nb3 than inhibition of the AP in human serum. Because the binding sites of FP and hC3Nb3 overlap extensively, the reduced inhibitory effect may reflect stronger competition by murine FP compared with human FP in these serum-based experiments. For practical reasons, we have not measured the affinity for native murine C3. An additional explanation is, therefore, that if the affinity for native murine C3 is much higher than the affinity for murine C3b, hC3Nb3 may be sequestered by C3 and therefore less available for C3b and inhibition of proconvertase assembly in murine serum.
We previously described the two nanobodies, hC3Nb1 that inhibits the AP (30) and hC3Nb2 that inhibits both AP, CP, and LP (31). The introduction of hC3Nb3 hence expands the toolbox of C3-specific nanobodies, which now comprises three nanobodies that exhibit a broad range of properties and bind at highly distinct sites on the C3b molecule (Fig. 6). The combined use of these nanobodies allows the investigation of the contribution of individual pathways or regulators. As an example, using complement-targeting nanobodies, we recently deciphered the pathways driving complement deposition elicited by Abs reacting with terminal galactose-α-1,3-galactose (49). Importantly, the nonoverlapping binding sites of these three C3-specific nanobodies are compatible with the combined use of these nanobodies.
Because of their small size, nanobodies are less likely to give rise to long-range steric hindrance effects compared with conventional Abs, in which even the Fab fragment is four times larger than a nanobody. For this reason, our three C3-specific nanobodies offer high precision probes for dissection of the large number of pivotal interactions formed by C3, C3b, iC3b, and C3c with other components of the complement system. An excellent illustration of how these nanobodies may help us to investigate function is the finding that hC3Nb3 inhibits the CP-driven terminal pathway. This was unexpected because the nanobody leaves C3 fragment deposition through the CP unaffected. The effect of hC3Nb3 on the terminal pathway must hence relate to the role of C3b in CP C5 convertase activity. During the progression of the complement pathways, at a threshold surface density of C3b molecules, both C3 convertases switch their specificity to C5 (7, 8), but how C3b does this remains a matter of controversy. For the AP, C5 convertase C3b has two roles; it both carries the Bb catalytic subunit and induces the shift from C3 to C5 convertase activity, whereas with respect to the CP C5 convertase, it only has the latter role. It is therefore convenient to denote C3b not carrying Bb as C3b′ to distinguish these two roles (60). One model suggests that a C5 convertase is a well-defined complex formed by a C3 convertase and C3b′, whereas as second model states that C3b′ primes C5 for cleavage by a C3 convertase while not necessarily forming a defined complex with the C3 convertases, reviewed in (60). FP similarly to hC3Nb3 inhibits the CP C5 convertase (61), which concurs with the strong overlaps of FP and hC3Nb3 binding sites but also indicates that the observed hC3Nb3 inhibition of the CP C5 convertase is not due to a missing FP contribution to the CP C5 convertase in our experiments. This further suggests that the inhibition of the CP C5 convertase by FP in part arises from its interaction with the C3b C345c domain. Whereas these results do not allow us to favor one of the two proposed roles of C3b′ for C5 convertase activity, they do for the first time, to our knowledge, indicate that the C3b′ C345c domain is in contact with either the substrate C5 or the C3 convertase subunits in the C5 convertase. However, similar to data reported by Thai et al. (62), our hemolysis data showed that recombinant C3 C345c domain did not provide significant competition for the CP C5 convertase on the surface of sheep erythrocytes. These results indicate that the role of this domain for C5 convertase activity requires its incorporation into activator-bound C3b′.
The present study also substantiates recent insight into the interaction between CR3 and iC3b (59). Our SPR data revealed that hC3Nb3 does not interfere with the interaction between CR3 and iC3b and indicate that the hC3Nb3 would allow the CR3-mediated phagocytosis of iC3b-opsonized particles. Similar to CR3, the exact binding site of CR4 remains unknown; however, nsEM data indicate that CR4 binds iC3b in the MG3–MG4 region (63). This binding site implies that the nanobody also allows the CR4-mediated phagocytosis, and this prediction in combination with the high stability of the hC3Nb3–C3b interaction and our successful usage of hC3Nb3 for flow cytometry suggests that hC3Nb3 may be used for tracking the engulfment of complement-opsonized particles.
From a research perspective, nanobodies provide greatly versatile tools that are readily adapted to tailor specific needs, and they are likely to be superior to conventional Abs in many applications. The small size of nanobodies give them high tissue penetrance compared with conventional Abs for in vivo imaging. For in vitro studies involving cell model systems, their small size allows the use of milder permeabilization strategies and hence preservation of ultrastructures, making nanobodies excellent tools for studies such as correlated light and electron microscopy (64). These properties could make hC3Nb3 a valuable reagent for the detection of C3 in intracellular compartments, thereby facilitating the analysis of multiple recently proposed intracellular roles for complement (65). In our flow cytometry and immunohistochemistry assays, we demonstrated the potential of the hC3Nb3 as a tool for specific labeling of deposited C3 fragments on biological surfaces. For our immunohistochemistry assay, we adapted the strategy of recombinantly fusing the nanobody to an IgY Fc (66). In the current study, we used direct conjugation of the fluorophore to the hC3Nb3:IgY, but the recombinant fusion of hC3Nb3 to IgY Fc also allows the detection with secondary anti-chicken Abs (66). Furthermore, the divalent appearance of the hC3Nb3:IgY adds avidity to the hC3Nb3–C3b/iC3b interactions. However, we expect that the nanobody would provide equal specificity in the absence of the IgY Fc. Our in-gel fluorescence and BLI data show that the nanobody interacts strongly with C3b and that this binding withstands harsh conditions. Our in-gel fluorescence assays further provide a rapid alternative to Western blotting for the detection of C3 and its degradation products. Overall, these properties demonstrate that hC3Nb3 can become a valuable tool for the detection of complement C3 and its degradation products in a variety of techniques.
In conclusion, hC3Nb3 potently inhibits the AP C3 convertase by inhibiting the assembly of the proconvertase. Importantly, the nanobody does not inhibit CP- and LP-mediated C3 convertase activity and C3 fragment deposition, but it inhibits the C5 convertase activity and prevents terminal C5-9 activation. In combination with our detailed structure-based mapping of its epitope, this allows us to conclude that the C-terminal domain of C3b is directly involved in CP C5 convertase activity. We also demonstrate that hC3Nb3 is an exquisite versatile tool for complement research and diagnostics, in particular because of its low nanomolar affinity for C3 as well as C3b and the ability to preserve this interaction under denaturing conditions.
Disclosures
G.R.A. declares a collaboration with Alexion Pharmaceuticals. T.E.M. has received royalties from Svar Life Science and is a member of the Scientific Advisory Board of Ra Pharmaceuticals. H.P., R.K.J., N.S.L., A.Z., D.V.P., S.T., and G.R.A. are listed as inventors on a patent describing the use of hC3Nb2 and hC3Nb3. The other authors have no financial conflicts of interest.
Acknowledgments
We acknowledge Christine Schar for assistance with SPR experiments, Thomas Boesen for help with electron microscopy, and Karen Margrethe Nielsen for technical support.
Footnotes
This work was supported by the Lundbeck Foundation (BRAINSTRUC, R155-2015-2666) and the Graduate School of Science and Technology at Aarhus University (01082017).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- AP
- alternative pathway
- BLI
- bio-layer interferometry
- C3MA
- C3 methylamine
- CP
- classical pathway
- CR
- complement receptor
- CVF
- cobra venom factor
- 2D
- two-dimensional
- 3D
- three-dimensional
- FB
- factor B
- FD
- factor D
- FF
- Fast Flow
- FH
- factor H
- FI
- factor I
- FP
- factor P
- hC3Nb3-GFP
- hC3Nb3 fused with GFP
- HSA
- human serum albumin
- IMAC
- immobilized metal affinity chromatography
- LP
- lectin pathway
- MG
- macroglobulin
- NHS
- normal human serum
- nsEM
- negative stain electron microscopy
- PDB
- Protein Data Bank
- PRM
- pattern recognition molecule
- SEC
- size exclusion chromatography
- SPR
- surface plasmon resonance
- TE
- thioester
- TEV
- tobacco etch virus
- Trx
- thioredoxin
- VBS
- veronal buffered saline.
- Received June 24, 2020.
- Accepted August 13, 2020.
- Copyright © 2020 by The American Association of Immunologists, Inc.