C3 was prepared as described in (27). Other complement components were purchased from Complement Technology Inc., Tx, US. C3(H₂O) was prepared by repeated freeze-thawing of C3 (C3(FT)) (28) or by treating native C3 with 0.2M methylamine (C3(met), treatments which both lead to disruption of the thiolester and loss of hemolytic activity (9). The anti-C3 monoclonal antibodies (mAb) 4SD17.3, 7D84.1, 7D169.1, and 7D398.1 against different neoepitopes in C3 (the C3(D) cluster) were produced according to Nilsson et al. (29).

C3(H₂O) generated by repeated freeze-thaw cycles (C3(FT) was separated by cation-exchange chromatography into native C3 and two different forms of C3(H₂O) identified in (9). Two serum preparations were also separated: the initial serum immediately after preparation and the second after 24 h incubation at 37°C. C3 (H₂O) (500 µg) or human serum (1 mL) were added to a MonoS 5/50 GL column. The rate was 0.5 mL/min with a gradient from 0 to 1 M NaCl in 20 mM Phosphate buffer pH 6.8. The chromatography was performed at room temperature (RT) (purified C3) or at +4°C (serum). Serum preparations had 1 mM benzamidine-HCl added after clotting for 45 minutes while collected fractions after separation had 1 mM phenylmethanesulfonyl fluoride (PMSF) added.

Sandwich ELISAs used the mAbs raised against C3 neoepitopes of the C3(D) epitope cluster (4SD17.3, 7D84.1, 7D169.1, and 7D398.1) (30) as capture antibodies (1-2 µg/mL) and biotinylated polyclonal (pAb) rabbit anti-C3c (10 µg/mL) followed by streptavidin-HRP (SA-HRP) from Cytiva Sweden AB (1:500) for detection.

A quantitative ELISA assessed the total amount of C3. It used rabbit anti-C3c pAb as both capture (3 µg/mL) and detection (biotinylated 3 µg/mL, followed by streptavin-HRP) as described above. The generation of C3a in supernatants was measured using a previously described C3a ELISA (31).

Blood anticoagulated with lepirudin (50 μg/mL; Schering AG, Saksa, Germany) or K2 EDTA (Greiner Bio-One, GmbH, Austria) was drawn from healthy volunteers who had been free from medication for at least 2 weeks prior to the donation. Platelet rich plasma (PRP) was prepared from whole blood by centrifugation at 150xg for 15 min at RT.

To isolate the platelets from plasma proteins, platelets were pelleted from PRP by centrifugation at 1100xg for 10 min and washed three times in Tyrode’s buffer (pH 6.5) (137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂, 0.36 mM NaH₂PO₄, 12 mM NaHCO₃, 2 mM CaCl₂, 5.5 mM glucose) containing 3.5 mg/mL bovine serum albumin (BSA), 1 μM PGE1 (both from Sigma Aldrich, St Louis, MO, USA), and 50 μg/mL lepirudin. After being washed, the platelets were pelleted and resuspended in Tyrode’s buffer without BSA, PGE1, and the platelets were then activated by the addition of thrombin activating peptide-6 (TRAP-6; 25-33 μg/mL, Sigma Aldrich) and incubated for 15 min at 37°C. After activation, the platelets were washed once and resuspended in Tyrode’s buffer. Activated and non-activated platelets were then diluted to a concentration of 200 x 10^9/L with Tyrode’s buffer. In experiments where platelets were incubated with purified C3, the protease inhibitor PMSF (1 mM) was added to the washing buffer to inhibit putative proteases that could cleave C3.

C3 (10-100µg/mL) binding to platelets was detected by FITC-conjugated rabbit pAb anti-C3c antibody (10 µg/mL, DAKO GmbH, Jena, Germany). C3 conformation after binding to the platelets, was monitored using anti-C3 mAbs specific for different neoepitopes in the C3 followed by anti-mouse Ig‐FITC (DakoCytomation). The samples were analyzed using Accuri C6 cytometer (BD) and ∼5000 platelets were analyzed for each sample. Mouse anti-human P-selectin (CD62p, 10 µg/mL, Invitrogen, Thermofisher Scientific, Sweden) was used to confirm platelet activation (22).

To confirm the exposure of phosphatidyl serine on platelets, annexin-FITC was allowed to bind to the platelets (Apoptosis detection kit, BD Pharmingen). In some experiments, platelets were preincubated with Annexin V (25 μg/ml; Sigma-Aldrich) for 30 min at RT prior to addition of native C3 (100 μg/mL) followed by another incubation for 30 mins at RT. Platelets where then washed and incubated with anti-C3c antibodies (see above) and analyzed by flow cytometry.

Isolated non-activated and TRAP6-activated platelets (100 µL of 50 × 10⁹/L platelets) were incubated with purified C3, C3b, or C3(met) (10 µg/mL) for 45 min at 37°C followed by washing with Tyrode’s buffer, pH 6.5. To evaluate the ability to generate AP convertases, isolated non-activated and TRAP6-activated platelets incubated with C3, C3b or C3(met) (10 µg/mL), as described above, were incubated together with FB (25 µg/mL) both in the absence and presence of properdin (25 µg/mL), at 37°C for 30 min. After washing once with PBS, native C3 (10 µg/mL) and FD (0.5 µg/mL) was added and incubated for 60 min. The reaction was stopped with 10 mM EDTA and the platelets were pelleted and the supernatants analyzed for C3a generation.

a. PMN: Blood was collected in 10 mL sodium heparin (17 IU/mL) tubes (BD Vacutainer) and diluted 1:2 in sterile PBS at RT followed by purification on Ficoll-Paque Plus density gradients (GE Healthcare, Uppsala, Sweden) to isolate PMN. The cells were diluted to 1 × 10⁶/mL in RPMI 1640 medium with L-glutamine (Invitrogen, Paisley, UK) and the cell purity and viability were checked with trypan blue using a Luna II automated cell counter (Logos, Biosystems, Gyeonggi-do, South Korea).

b. HDMEC: HDMEC cells (PromoCell GmbH, Heidelberg, Germany) were culture in human endothelial cell medium (ECM [³H Biomedical, Uppsala, Sweden]) containing 10% heat-inactivated FCS and the cell purity and viability were checked with trypan blue using a Luna II automated cell counter (Logos, Biosystems, Gyeonggi-do, South Korea).

c. Apoptosis: For induction of apoptosis, PMN and HDMEC were incubated in 5 µM camptothecin (Sigma Aldrich) for 4h at 37°C. After wash the HMECs were detached by trypsinization (1mg/mL), suspended at a concentration of 10⁶ cells/mL in ECM.

The different cell populations were validated by flow cytometry using Annexin-FITC to detect apoptosis (Apoptosis detection kit, BD Pharmingen) according to the instructions of the manufacturer. Non-apoptotic and apoptotic cells were incubated with either purified human C3 (100µg/mL) or human serum-EDTA (10mM) diluted 1:10 corresponding to approximately 100 µg/mL of C3 at RT for 30 min.

After washing, the cells were incubated with C3 (100µg/mL) for 30 min at RT in VBS. Anti-CD16-PerCP from BD (BD Pharmingen) were used for gating of PMN. Anti-C3c-FITC (1/100, Dako) was used for staining for C3, while FITC directly conjugated to Annexin V was used to detect apoptosis (Apoptosis detection kit, BD Pharmingen) according to the instructions of the manufacturer. The cells were fixed with 1% (v/v) PFA-PBS. Using the same detection system, competitive binding of Annexin V and native C3 was tested by sequential addition of the non-labelled Annexin V (Beckman Coulter, Netherlands) or C3 at a concentration of 10-100 µg/mL before adding the opposite labelled protein. The samples were analyzed using CytoFlex S (Beckman Coulter, Uppsala, Sweden).

Simple Western™ Wes Capillary immune electrophoresis (CIE) instrument (Bio-Techne, Minneapolis, USA) was used to analyze C3 in the different fractions using one mAb specific for the C3α-chain (C3a; 4SD17.3) and one for the C3β-chain (7D169.1) for detection. C3 bound to platelets isolated from activated and non-activated PRP was also analyzed by CIE with the mAb 7D84.1 against C3α-chain for detection and purified C3 and C3b as controls. C3 bound to non-apoptotic and apoptotic PMNs was analyzed by traditional Western blotting using pAb anti-C3c and 4SD17.3 for detection and C3, C3b and iC3b as controls. For platelets and PMNs, the protease inhibitor PMSF (1 mM) was added from the start to the washing buffers to inhibit putative serine proteases from cleaving C3.

Clots were formed in citrate plasma that had been centrifuged once at 3200 x g, leaving only trace amounts of platelets. The supernatant was transferred to new containers where clots were formed by recalcification and addition of thrombin (2 U/mL) followed by incubation at 37°C for 30 min after which the clots were snap frozen, cryosections of 4 µm thickness were prepared and stored at -80°C.

a. Immunofluorescence. Paraformaldehyde 4% fixated blood clots were blocked using 2.5% horse serum for 30 minutes. The clots were subsequently labelled for complement proteins using a primary rabbit polyclonal anti-C3c antibody diluted 1:2000 (Dako, Glostrup, Denmark) and a monoclonal anti-C3a antibody 4SD17.3, diluted 1:200. The clots were incubated overnight in a humid chamber at 4°C. After washing the clots in 10 mM phosphate buffer (Sigma-Aldrich, Saint Louis, USA), pH 7.4, secondary antibodies were applied using the VectaFluor Duet double labelling kit. In brief, cells were clots allowed to interact with a mixture of DyLight 488 anti-rabbit and 594 anti-mouse secondary antibodies (Vector Laboratories Inc., Newark, USA) for 30 minutes. Clots were washed and mounted using 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories). Negative controls included an isotype IgG1 (Dako) antibody as well as omission of the primary antibodies.

b. In situ proximity ligation assay. For the in-situ proximity ligation assay (PLA, Sigma-Aldrich), the manufacturer’s protocol was followed. In brief, the fixated clots were blocked with a Duolink blocking solution (Sigma-Aldrich) for 60 minutes at 37°C. The clots were then incubated overnight at 4°C with primary antibodies against C3c, diluted 1:2000 (Dako), and C3a, clone 4SD17:3 diluted 1:200. After washing the plus anti-rabbit and minus anti-mouse PLA probes (Sigma-Aldrich) were applied diluted 1:5 in antibody diluent (Sigma-Aldrich) and incubated for one hour at 37°C. Probes were subsequently ligated using Duolink ligation buffer (Sigma-Aldrich), diluted 1:5, and incubated on the clot for 30 minutes at 37°C. The signal from the ligation was amplified by adding in situ detection reagents specific for Texas red. Clots were washed and mounted using in situ mounting media with DAPI (Sigma-Aldrich). To ensure specificity, negative controls were established by omitting the primary antibodies individually, as well as omitting the ligase.

c. Imaging and analysis. Immunofluorescence was analysed using a Nikon Ti2-E AX confocal microscope (Nikon, Tokyo, Japan), equipped with appropriate filters. To obtain optical sections through the clots, images were collected as z-stacks with a step size of 0.5 μm, using the NIS-Elements software (Nikon).

Lipids were dissolved in ethanol, or a chloroform-methanol mixture followed by removal of the organic solvent by evaporation. Liposomes were freshly prepared by hydration of the lipid film in H₂O (MilliQ grade) followed by hand extrusion 11 times through an 80 nm diameter polycarbonate filter and 11 times through a 30 nm diameter polycarbonate filter at 50-60°C using an Avanti Mini Extruder (Birmingham, Al, USA). The used lipids and the composition of the produced liposomes (% phospholipids) are presented in Table 1. The liposomes were characterized concerning both size and zeta potential using a Malvern Zetasizer (Malvern, UK).

The quartz crystal microbalance with dissipation monitoring (QCM-D) measurements were performed using a QCM-D Pro from Biolin Scientific AB (Gothenburg, Sweden). A layer of liposomes/bilayer was first allowed to assemble on the surface of the SiO₂ sensors in the QCM-D instrument by letting the liposome suspension flow over the surface for 30 min. After priming with buffer, C3 (10 μg/mL) in veronal buffered saline (VB⁺⁺; 5 mM barbiturate, pH7.4; 145 mM NaCl; 0.15 mM Ca²⁺; 0.5 mM Mg²⁺) was added for 50 min, followed by mAb 7D84.1 (10 μg/mL) for 20 min, and finally anti-C3c antibody (10 μg/mL) for 20 min.

QCM-D studies were performed only with the liposomes with the composition molar ratio of DPPC: Cholesterol : DPPG (40:40:20). C3 (10 μg/mL) was allowed to bind to the liposome-coated sensors for 50 min, followed by the addition of a mix of FB (10 μg/mL) and properdin (5 μg/mL) for 5 min. Finally, FD (0.5 µg/mL) and additional native C3 (10 µg/mL) were added. In another experiment, SiO₂ sensors were prepared with liposome/bilayers. C3 was added (10 μg/mL) and allowed to bind for 50 min. FH (5 µg/mL) was added, followed by FB together with properdin.

Four QCM-D sensors were coated with negatively charged cholesterol liposomes, of which two sensors received liposomes with the composition DPPC: Cholesterol : DPPG (40:40:20) and the other two sensors with liposomes of composition DPPC: Cholesterol : PS (40:40:20). Next, Annexin V (0.2µg/mL) was added to two of the sensors, one of each liposome type. Finally, C3 (10 µg/mL) was added to all four sensors.

Prior to sample application, graphene oxide was applied to all Quantifoil 1.2/1.3–300 mesh grids as previously described (32). Complement C3 (0.2 mg/mL) was incubated with either DPPC: Chol : PS [40:40:20] or DPPC:X:PS [80:20] for 60 mins before grid freezing. An initial blot was performed with 3 μL of C3/liposome solution applied to the graphene oxide-containing grids for 3 minutes. Vitrification was then performed with a Vitrobot Mark IV (Thermo Fisher Scientific) at 25°C and 95% humidity. Grids were blotted in duplicate for 4 seconds and plunged into liquid ethane. Following vitrification, the grids were clipped into autogrid cartridges (Thermo Fisher Scientific) for use with autoloader systems.

All grids were screened and full datasets were collected on a Glacios 200 kV cryo-electron microscope mounted with a Falcon 4i direct electron detector with Selectris energy filter. Data were acquired in EER format at a pixel size of 1.15 Å/pixel (100 kx magnification), 56 e⁻/Ų total electron dose, −0.8 to −2.0 μm defocus range, dose rate of 0.91 electrons pixel^-1 sec^-1. Automated collection was performed using the EPU software (Thermo Scientific). A total of 4,600 micrographs were collected and had 2x binning applied and were processed using exposures fractionated into 41 frames for motion correction.

Processing was performed in cryoSPARC v4.6.0 and included initial Patch Motion Correction and Patch CTF being performed before micrograph curation (33). Micrographs were manually curated based on CFT estimations, total frame motion, defocus, ice thickness and contamination. Using the obtained template from the screening dataset, all particle picking was performed with blob picker. The picks totaled ~2 million and were 2x down sampled and extracted with a 400 x 400 box size before being Fourier-cropped to 250 x 250 pixels. Particles were classified into 150 2D classes (10 Å minimum separation distance, 170 - 220 Å circular mask inner-diameter, 60 online-EM iterations, 400 batch size per class). Following 2D class selection, this classification/selection procedure was repeated twice more which resulted in ~896,000 particles suitable for Ab-initio reconstruction. Ab-initio reconstruction was followed by heterogenous refinement using a few poor density “dummy” classes as seeds for junk particles. Two further rounds of heterogenous refinement seeding resulted in 317,410 particles suitable for masked refinement and reconstruction jobs which produced a density map at ~3.8 Å.

All experiments were repeated at least three times. Data are presented as mean ± SEM or as representative images. Data were analyzed using paired t-tests, if not more than two variables were analyzed when one-way ANOVA was applied followed by multiple comparison. Spearman correlation test was performed for the linear regression analysis. All data were analyzed using GraphPad Prism 10 for Mac OS version 10.4.0 (GraphPad Software, La Jolla, CA, USA). The calculated P-values were defined as follows: *p < 0.05, **p < 0.001, ***p < 0.001, and ****p < 0.0001.

In previous studies, we and others have identified 3–4 major peaks after separation of purified C3(H₂O) by cation exchange chromatography (9), corresponding to native C3 (peak 1), to an intermediate form of C3(H₂O) (peak 2) and finally to C3(H₂O) (peak 3) (Figure 1A). Here, this conversion was monitored by sandwich ELISAs using mAbs against conformational neoepitopes of C3 (Figure 1B). It demonstrated a continuous increase in the exposure of the C3a neoepitope detected by mAb 4SD17.3, from a weak exposure in the native C3 in Peak 1, to a strong exposure in C3(H₂O) of Peak 3, confirming a gradual conformational change in the protein (Figure 1B). The presence of the C3a epitope confirmed the presence of C3a in the C3 molecules and thus the absence of cleavage at the convertase cleavage site into C3a and C3b. The fractions were also tested with three other mAbs against conformational epitopes (clones 7D169.1, 7D84.1, 7D398.1), which all confirmed the conformational change in the molecule.

C3 in freshly prepared serum and in serum incubated for 24 h at 37°C was separated by the same method. These chromatograms did not explicitly show individual C3 peaks due to the abundance of other plasma proteins present (not shown). However, using the quantitative sandwich ELISA employing pAb anti-C3c, two distinct peaks of C3 were identified (Figure 1C; upper panel). The first major peak (peak 1) corresponded to native C3. The other peak in the position of peak 3, was displayed as a knee in the chromatogram of the 24 hr incubated serum. Using the sandwich-ELISAs, this peak exposed the neoepitops including the one in C3a (Figure 1C; lower panel). Capillary immune electrophoresis (CIE) analyses (Figure 1D), employing mAb anti-C3β-chain (7D169.1) was able to detect C3β chain present in fractions 3-6, in proportion to the total amount of C3 in these fractions (Figure 1C upper panel). Intact C3α-chain (mAb 4SD17.3) was only present in peak 1 (fraction 3) and hardly visible in peak 2 (fraction 4). The intact C3α-chain confirms the presence of native C3 in peak 1 since factor I in serum is not able to cleave native C3. By contrast, C3(H₂O) is susceptible to factor I cleavage, suggesting that generated C3(H₂O) is immediately inactivated into iC3H₂O (cf iC3b) in serum (34).

We have previously shown that intact C3 binds to activated platelets in PRP anticoagulated with EDTA or lepirudin and to isolated activated platelets (22). The tentative conclusion was that the C3 was in its C3(H₂O) form. To investigate whether it is preferably native C3 or any of the C3(H₂O) forms that bind to activated platelets, individual fractions of purified freeze-thawed C3(H₂O) isolated by Mono S cation exchange chromatography (see Figure 1A), were added, at the same concentration, to both activated and non-activated platelets (Figure 2A). Flow cytometry analysis shows that it is mainly native C3 (peak 1) that binds to activated platelets. C3(H₂O) intermediate (peak 2) also binds to some extent, while only very small amounts of C3(H₂O) (peak 3) could be detected. No C3, regardless of conformation, binds to non-activated platelets.

The conformation of C3 after binding to activated platelets, was characterized using the mAbs directed to neoepitopes described above (Figure 2B). It turned out that several of the C3 epitopes were exposed after binding to platelets, including in the C3β chain (mAb 7D169.1), C3a (mAb 4SD17.3), and C3c 20kDa α-chain polypeptide (mAb 7D398.1).

CIE was used to analyze C3 bound to platelets isolated from activated and non-activated PRP (Figure 2C). As expected, the non-activated platelets bound only trace amounts of native C3 and no C3b. The activated platelets bound C3 with an intact α-chain without any traces of α’-chain (C3b). This band was visualized with mAb 7D84.1 to C3d,g, which can detect both the α-chain and the α’-chain of C3 and C3b, respectively, ruling out the presence of C3b on the platelet surface.

Native C3, C3b, and C3(H₂O) (C3[met]) were allowed to bind to isolated non-activated and TRAP6-activated platelets in the presence and absence of properdin. The platelets were then analyzed by flow cytometry using pAb anti-C3c which is able to detect intact C3, C3(H₂O), and C3b. Again, C3 bound strongly to the activated platelets while non-activated platelets did not bind any form of C3. Interestingly, the binding was much enhanced by properdin (Figure 2D). Both C3b (Figure 2G) and C3(met) (Figure 2J) were poor binders, although their binding was slightly increased in the presence of properdin. The platelet with bound C3, C3b, and C3(met) were incubated with FB and followed by the addition FD and native substrate C3 to the formed AP convertases. The level of convertase formation was then evaluated by measuring the amount of generated C3a. The bound native C3 generated most C3a, followed by C3b and C3(met) (Figures 2E, H, K). As expected, convertase formation increased in the presence of properdin. When the amount of C3a was correlated with the amount of bound C3 on the platelet surface (non-activated and activated ± properdin), strong correlations were obtained for bound native C3 and C3b, supporting that native C3 transforms into a C3b-like form of C3 upon binding to the platelet surface (Figures 2F, I). Also, on a molar basis native C3 seemed to be equally efficient as C3b. No correlation was found between C3a generation and platelet-bound C3(met) (L), implying that C3(met) is an inefficient convertase companion.

The phospholipid-scrambled membranes of the activated platelets was shown as binding of exogenously added Annexin V (Figure 2M). Annexin V was able to partially inhibit the binding of C3 to activated platelets (Figure 2N).

To investigate whether intact C3 could be detected in in vitro formed blood clots, blood clot sections were firstly stained using proximity ligation. Figures 3A-C shows negative controls A: both antibodies no ligase, B: pAb anti-C3c alone; C: mAb anti-C3a (4SD17.3) alone. In Panel D hybridization enhanced by Texas red (purple color) shows that the C3a epitope and the rest of C3 (C3c) are located together on platelets and platelet microparticles. This was confirmed using traditional immunofluorescence staining for fibrinogen, P-selectin (CD62P), C3 (pAb anti-C3c) and C3a (mAb 4SD17.3). Figures 3E-H show that CD62 positive platelets also bind the mAb anti-C3a. In the next panels (Figures 3I–O) it is demonstrated that C3a positive platelets are also positive for C3c, corroborating the presence of intact C3 on the platelets. By using a multiplate device, the ability of intact C3 to inhibit aggregation in whole blood was tested. In Figure 3P it is shown that mAb 4SD17.3 against a neoepitope in C3a partially inhibits the aggregation. This supports that intact platelet-bound C3(H₂O) was targeted, previously shown to mediate platelet/leukocyte complex formation (26).

Apoptosis was induced in PMN and HDMEC by camptothecin. Annexin V indicated that the cells were apoptotic (Figures 4A, D). Native C3 was added to both the non-apoptotic and the apoptotic cell populations, and C3 binding was assessed by flow cytometry using pAb anti-C3c (Figures 4B, E). It clearly showed that the apoptotic cells bound much more C3 compared to the normal (non-apoptotic) cells. The same analysis was done with mAb 4SD17.3 which showed that the C3a domain of bound C3 was exposed in a similar way as on activated platelets (Figures 4C, F).

Annexin V binds negatively charged phosphatidyl serine exposed on the apoptotic cell membrane. When Annexin V was allowed to bind to the apoptotic PMN and HDMEC before C3, the C3 binding was almost completely inhibited (Figures 4G, H). Similarly, Annexin V binding to PMN was largely reduced if C3 was added before the addition of Annexin V (Figure 4I). This suggests that C3 binds to the same sites as Annexin V on the apoptotic cell membranes.

Western blotting using the pAb anti-C3c showed that C3 bound to the apoptotic cells was non-cleaved, i.e., the detected C3 had an intact α-chain (Figure 4J).

When the liposomes were tested by Quartz Crystal Microbalance with Dissipation (QCM-D), it was found that C3 were deposited on the liposome surfaces containing 20 mol% negatively charged phospholipids and cholesterol, i.e. (DPPC: Chol : DPPG; 40:40:20) and (DPPC: Chol : PS; 40:40:20). Liposomes composed of only net neutral phospholipids and cholesterol (DPPC: Chol; 60:40) and negatively charged liposomes without cholesterol (DPPC: DPPG; 80:20) did not bind any C3 at all (Figure 5A).

The conformation of C3 bound to the DPPC: Chol : DPPG and to DPPC: Chol : PS liposomes, was evaluated using neoepitope mAb anti-C3 7D84.1, indicating that the liposome binding had induced a conformational change of the C3 molecule resembling that of surface-bound C3b/iC3b (30). The amount of C3, independent of conformation, was confirmed by a pAb anti-C3c antibody (Figure 5B). Interestingly, liposomes with only 5 mol% negatively charged phospholipids and cholesterol, did not bind any C3 either. Thus, 20 mol% of negatively charged phospholipids along with cholesterol were required for C3 binding to the liposome surface. Whether the negatively charged phospholipids were of the PS or DPPG type, did not affect the outcome (Figure 5B). Both liposomes bound approximately the same amount of C3. The properties and composition of the liposomes that bound native C3 resemble those of the membranes of activated platelets.

The negatively charged and high cholesterol-containing liposome surface (DPPC: Chol : PS [40:40:20]) was also tested for its ability to support AP convertase formation. After binding of native C3, the surface was sequentially exposed to FB and properdin followed by FD and additional C3. Both liposomes were able to form AP C3 convertases, which were downregulated by FH. The AP convertase formation with DPPC: Chol : PS [40:40:20] is shown in Figure 5C and the corresponding inhibition with FH in Figure 5D.

To find out if C3 and Annexin V compete for the same type of binding sites also on the liposomes, Annexin V was first allowed to bind to negatively charged cholesterol-containing liposomes with the composition DPPC: Cholesterol : DPPG (40:40:20) and DPPC: Cholesterol : PS (40:40:20), which showed a high degree of binding (Figure 5E). When C3 was added to the Annexin V pre-coated liposome surfaces a significantly smaller mass of C3 was bound, compared to C3 allowed to bind directly to the liposomes (Figure 5F). This demonstrates that Annexin V and C3 compete for the same binding sites on the cholesterol-containing negatively charged liposome surfaces.

The surface binding of conformationally altered C3 to negatively charged and high cholesterol-containing liposomes was investigated using structural techniques. Accordingly, cryo-EM was employed to uncover the structural basis of binding and the species responsible for surface interaction. Using the lipid composition of DPPC: Chol : PS [40:40:20] and incubating the extruded liposomes with native C3 in a 1:30 ratio, we found that native, conformationally changed C3 was closely associated with the liposomal surface (Figure 6A, left panel). The electron densities for all domains were accounted for as well as a significant translocation of the anaphylatoxin C3a domain. Importantly, the thioester was the closest domain to the liposomal surface, positioned approximately 16 Å adjacent. Repeating the microscopy experiment using cholesterol absent liposomes (DPPC: PS [80:20]) maintained the native C3 conformation (35) (Figure 6A, right panel). Additionally, no protein was able to be observed near the surface of liposomes, suggesting that the presence of cholesterol is a key feature which drives this protein-surface interaction. 2D classification examples of aligned C3(H₂O)-like C3 (Figure 6, left panel) and inactive C3 (Figure 6, right panel) from cholesterol-containing and cholesterol-free environments, respectively, are presented and GSFSC curves and orientation distribution heatmaps for both are presented in Figure 6C.