Comparative protein structure modeling approach
To predict the 3D structure and C3b-binding capacity of a new multitarget complement regulator, MFHR13 (FHR11-2:FH1-4:FH13:FH19-20), we used the template-based modeling implemented in Modeller 9.1963.
High-resolution crystal and NMR structures for 17 of 20 domains of human FH are available in the Protein Data Bank (PDB). To build the model, we used the following structures as templates (PDB accessions): 3ZD2 (FHR11-2), 2WII (FH1-4 + C3b), 2KMS (FH12-13), 2G7I (FH19-20), 3SW0 (FH18-20), 2XQW (FH19-20 + C3d), and a SAXS model for FH11-14 (Accession in Small Angle Scattering Biological Data Bank (SASBDB): SASDAZ4). As the orientation between adjacent SCR domains is difficult to predict, some restrictions were added to orient the template structures appropriately. These restrictions were obtained by measuring distances Cα-Cα between adjacent SCRs in PDB structures such as 2WII, 3SW0 using the PyMOL Molecular Graphics System, Version 2.0, Schrödinger, LLC (PyMOL 2.0). Some additional restraints between FH4 and FH19 were taken into account, after superimposing the structures 2WII and 2XQW, which correspond to FH1-4 with C3b and FH19-20 with C3d, respectively.
The alignment of MFHR13 was done using also a structure for FH11-14, derived from SAXS information (SASDAZ4). FH14 was aligned with FH19 because they share a sequence identity of 30%, and it would allow orienting FH13. The module Automodel from modeler 9.19 was used to generate 100 models. The loop regions were refined using the loopmodel class, and 4 models with loop refined were built for every standard model (total: 400 models). Optimization of the objective function was done using 300 iterations with the conjugate gradient technique and molecular dynamics with simulated annealing to refine the model. The best models were chosen comparing the Discrete Optimized Protein Energy (DOPE) score. The quality of the models was evaluated using the Ramchandran Plot SAVeS Server (PROCHECK)64 (https://services.mbi.ucla.edu/SAVES/), ProsaWEB (https://prosa.services.came.sbg.ac.at/prosa.php)65. 3D structures were visualized with PyMol 2.0.
Generation of plasmid constructs
First, the vector pAct5-MFHR1I62, coding for MFHR1I62, a MFHR1 variant with isoleucine instead of valine at position 193, which corresponds to position 62 of factor H, was created. This vector was obtained via site-directed mutagenesis using the Phusion Site-Directed Mutagenesis Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. The vector pAct5-MFHR141 was used as template with the primers MFHR1_I62_fwd and MFHR1_I62_rev (Supplementary Table 1), exchanging a single nucleotide (GTA to ATA, V62I). In this vector the expression of MFHR1I62, fused to a C-terminal 8x His-tag, is driven by the 5′ region, including the 5′ intron, of the PpActin5 gene66 and the cauliflower mosaic virus (CaMV) 35S terminator. For proper posttranslational modifications, the recombinant protein was targeted to the secretory pathway by the aspartic proteinase signal peptide from P. patens, PpAP167. The hpt cassette68 is present to select the transformed plants with hygromycin.
For the production of MFHR13 the expression construct pAct5-MFHR13, based on pAct5-MFHR1I62, was generated using Gibson assembly69. For this, the sequence coding for SCR FH13, together with its natural flanking linkers, was amplified from plasmid pFH39 with the primers SCR13_overlapSCR4_fwd and SCR13_overlapSCR19_rev. This fragment was inserted in pAct5-MFHR1I62, previously amplified with primers SCR4_overlapSCR13_rev and SCR19_overlapSCR13_fwd (Supplementary Table 1), designed to exclude the linker between FH4 and FH19 and overlapping the sequence of the linkers from FH13. PCRs were performed with Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific). Assembled vectors were verified by sequencing. The plasmids are available via the International Moss Stock Center IMSC (www.moss-stock-center.org/en/) under the accession numbers P1755 and P1762 for the expression constructs pAct5-MFHR1I62 and pAct5-MFHR13, respectively.
Plant material, culture, transformation procedure, and protein production
Physcomitrella (Physcomitrium patens) was cultivated as previously described68 on Knop ME medium70. Lines producing MFHR13 or MFHR1I62 were obtained by stable transformation of the Δxt/ft moss line (IMSC no.: 40828), in which the genes coding for α1,3 fucosyltransferase and the β1,2 xylosyltransferase are knocked out45.
Moss protoplast isolation, polyethylene glycol-mediated transfection (using 50 µg of linearized plasmid DNA), regeneration and selection were performed as previously described68.
Plants surviving the selection were transferred to liquid medium and after 1 month of weekly propagation, moss lines were screened for the production of the proteins of interest. For this, 6-days-old moss suspension cultures were vacuum-filtrated and 30 mg fresh weight (FW) material were analyzed by ELISA. To analyze the time course of growth and production of the protein, lines were inoculated in triplicates at 200 mg/L DW and samples were taken every 3 days for 26 days.
The best MFHR1I62 and MFHR13 producing moss lines, respectively, were scaled up to stirred tank bioreactors (5L), operated in batch for 8–12 days with the following conditions: pH 4.5, at 22 °C, aerated with 0.3 vvm air supplemented with 2% CO2, stirred with 500 rpm, continuous light with an intensity of 160 µmol m−2 s−1 (day 0–2) and 350 µmol m−2 s−1 (day 2–8). The growth medium was supplemented with 1-naphthaleneacetic acid (10 µM NAA) at day 3 according to Top et al.41.
Purification of moss-produced recombinant proteins
MFHR13 and MFHR1 variants (MFHR1I62 and MFHR1V62), respectively, were extracted from vacuum-filtrated plant material. For this, 4 mL binding buffer (75 mM Na2HPO4, 0.5 M NaCl, 20 mM imidazole, 0.05% Tween-20, 10% glycerol, 1% protease inhibitor (P9599, Sigma-Aldrich), pH 7.0) were added per gram FW and the suspension was disrupted with an ULTRA-TURRAX® (10,000 rpm) and simultaneous sonication (ultrasonic bath) for 10 min on ice. After two consecutive centrifugation steps (4500 × g for 3 min and 20,000 × g for 10 min at 8 °C) the supernatant was filtered through 0.22 μm PES filters (Roth).
For chromatographic purification, the filtrate was loaded onto a 1 mL HisTrap FF column, using the ÄKTA system (Cytiva) at 1 mL/min. The column was washed with 30 column volumes (CV) of binding buffer and 3% elution buffer (100%: 100 mM Na2HPO4, 0.5 M NaCl, 500 mM imidazole, 10% glycerol, pH 7.4). The protein was eluted using a stepwise gradient (6% elution buffer for 10 CV, 17% 5 CV, 27% 3 CV, 100% 6 CV) and collected in 0.5 mL fractions. The first five fractions obtained with 100% elution buffer were pooled and diluted with Tris buffer pH 7.6 to reach 50 mM NaCl and loaded onto a 1 mL HiTrap Q HP column (Cytiva). MFHR13, MFHR1I62, or MFHR1V62 were eluted using a linear gradient (3–100%) and elution fraction containing the protein of interest (screened by ELISA, western blot, and Coomassie staining) were pooled and dialyzed against Dulbecco’s phosphate-buffered saline (DPBS) in Slide-A-Lyzer® MINI Dialysis Devices, 20 K MWCO (Thermo Fisher Scientific). Proteins were concentrated by ultrafiltration using Vivaspin 2, 10 kDa MWCO (PES membrane; Sartorius).
Protein quantification and immunoblotting
MFHR13 and MFHR1 variants were quantified by ELISA using a modified protocol39: In order to quantify the fusion proteins using plasma-derived FH (hFH) standard as a reference, a polyclonal antibody against the whole FH was avoided, due to the differences in the protein structure and molecular weight between FH and the fusion proteins. Instead, a detection antibody against the FH1-4, domains shared by all the proteins of interest, was used.
Microtiter plates (Nunc Maxisorp, Thermo Fisher Scientific) were coated overnight at 4 °C with GAU 018-03-02 (Thermo Fisher Scientific), a monoclonal antibody that recognizes FH20 (1:2000 in coating buffer (1.59 g/l Na2CO3, 2.93 g/l NaHCO3, pH 9.6)) and blocked with sample buffer (2% BSA in TBS (Tris Buffer Saline) supplemented with 0.05% Tween-20). Samples and hFH standard (hFH, CompTech; 12.9 pM–1.1 nM) were diluted in sample buffer and incubated for 90 min at 37 °C The proteins of interest were detected by a polyclonal anti-FH1-4 (1:15,000 in washing buffer (1% BSA in TBS supplemented with 0.05% Tween-20)71 and anti-rabbit coupled to horseradish peroxidase (HRP) (NA934; Cytiva, 1:5000 in washing buffer).
The ratio between MFHR13, MFHR1I62, and MFHR1V62 concentration was also compared with semi-quantitative western blots. SDS-PAGE on 7.5 or 10% gels, Coomassie staining and western blot analysis were performed as described before41.
Glycosylation analysis
MFHR13, purified as described above, was used for MS analysis. In brief, duplicate samples of purified MFHR13 were reduced and alkylated as previously described55, subjected to SDS-PAGE and subsequently stained with PageBlue® (Thermo Fisher Scientific). Bands corresponding to the expected size of MFHR13 were excised and destained. Digestion was performed overnight with 0.2 µg trypsin (Trypsin Gold, Promega) and 0.2 µg chymotrypsin (sequencing grade, Promega) simultaneously at 37 °C. Peptides were recovered and desalted using C18 StageTips (Thermo Fisher Scientific) and measured on a QExactive Plus Orbitrap (Thermo Fisher Scientific) as described in ref. 41. Raw data were processed with Mascot Distiller V2.7.10 and a database search was performed using Mascot server V2.7 (Matrix Science). Processed spectra from both duplicates were searched against a database containing all Physcomitrella protein models (V3.372) as well as the sequence of MFHR13 and simultaneously against a database containing sequences of known contaminants (269 entries, available on request) using a precursor mass tolerance of 5 ppm and a fragment mass tolerance 0.02 Da. As variable modifications Gln−>pyroGlu (N-term. Q) −17.026549 Da, dehydration Glu->pyroGlu (N-term. E) −18.010565 Da, oxidation +15.994915 Da (M), deamidation +0.984016 Da (N), GnGn +1298.475961 Da (N) were specified. Carbamidomethyl +57.021464 Da (C) was set as fixed modification. Search results were loaded in Scaffold5 software (Proteome Software, Inc.) and a threshold of 1% FDR at the protein level and 0.1% at the peptide level with a minimum of two identified peptides was specified.
Glycopeptides were identified from processed mgf files as described in ref. 55 using a custom Perl script. A list of searched glycopeptides is available in the Source data file. Quantitative values for identified glycopeptides were obtained from the allPeptides .txt file (available on request) from a default MaxQuant (V1.6.0.16) search on the raw data. All quantitative values were normalized against the sum of all precursor intensities from each raw file.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE73 partner repository with the dataset identifier PXD025471 and https://doi.org/10.6019/PXD025471.
Activity tests
Cofactor activity (CA)
CA was measured in fluid phase. MFHR13, MFHR1I62, MFHR1V62, or hFH (15 or 75 nM) were incubated in DPBS with 444 nM C3b and 227 nM FI at 37 °C. Samples (20 µL) were collected at different reaction times up to 20 min. The reaction was stopped by the addition of 7.7 µL 4x Laemmli buffer (Bio-Rad) with 3 µL 50 mM DTT (NuPAGE, Thermo Fisher Scientific). Proteolytic cleavage of C3b was assessed by visualizing the α-chain cleavage fragments α′68 and α′43 by SDS-PAGE in 7.5% gels under reducing conditions followed by Coomassie staining. The bands corresponding to the intact C3b α′-chain were quantified by densitometry (GelAnalyzer 19.1; www.gelanalyzer.com) and normalized with the corresponding C3b-β-chain. The ratio α′-chain/β-chain at time 0 was set to 100% intact C3b α′-chain.
Decay acceleration activity assay (DAA)
The DAA in fluid phase was performed by an ELISA-based method as previously described40. Briefly, 250 ng C3b in PBS were immobilized on Maxisorp plates overnight at 4 °C. In order to generate the C3 convertases (C3bBb), 400 ng Factor B and 25 ng Factor D (CompTech) were incubated with immobilized C3b for 2 h at 37 °C. Increasing concentrations of moss-made regulators and hFH (Comptech) were added and incubated for 40 min at 37 °C to measure their ability to displace preformed C3 convertases. The Bb fragments that remain bound to C3b were detected by an anti-factor B polyclonal antibody (Merck, Darmstadt, Germany), followed by HRP-conjugated rabbit anti-goat (Dako, Hamburg, Germany). The absorbance of the preformed C3 convertase without regulators was set to 100% intact C3 convertases and the C3 proconvertase (C3bB) formation (FB without adding FD) was included as a negative control.
Binding to complement proteins
The ability of MFHR13 and MFHR1 variants to bind to the complement proteins was tested by ELISA. For this, 5 µg/mL of C3b, C5, C6, C7, C8, C9, or the complex C5b6 (CompTech, USA) in coating buffer were immobilized on Maxisorp plates at 4 °C overnight, blocked, and subsequently incubated with increasing concentrations of MFHR13, MFHR1I62, or MFHR1V62 (0.195–50 nM) in the case of testing C3b and C5 binding or a single concentration (25 nM or 50 nM) in the case of testing C6, C7, C8, C9, and C5b6 binding, diluted in sample buffer. Bound regulators were detected with anti-His-tag antibodies (MAB050, R&D Systems, 1:1000 in washing buffer) and HRP-conjugated anti-mouse IgG sheep (NA931, Cytiva, 1:5000 in washing buffer). In order to combine all independent experiments, the absorbance was normalized with the value corresponding to the highest concentration for every binding ELISA to obtain a relative binding.
Recombinant FHR1 with a C-terminal 6x His-tag (Abcam 152006) and hFH (Comptech, USA) were included as controls. However, due to the absence of His-tag in hFH, bound-hFH was detected with a polyclonal antibody74 (1:1000 in washing buffer), and the HRP-conjugated anti-rabbit (1:5000 in washing buffer). It should also be considered, that the affinity of the anti-His-tag antibodies towards FHR1 compared to MFHR13, MFHR1I62 and MFHR1V62 might be different, due to the different lengths of the His-tags.
Microscale thermophoresis (MST)
The fusion protein MFHR13 (3 μM) was labeled with NT-647 RED-NHS (NanoTemper Technologies GmbH, Munich, Germany) as previously described75. NT-647-labeled MFHR13 (200 nM) was mixed with the ligands C5, C7, C9, and C3d, or BSA as control at dilutions ranging from 10 to 0.0005 μM. The samples were diluted in PBS supplemented with 0.05% Tween-20. Protein aggregation was minimized by centrifuging at 14,000 × g for 10 min. Thermophoresis was measured at 20% LED power and medium MST power in a Monolith NT.115 instrument (NanoTemper). All measurements were performed at RT using Premium Coated capillaries. Data were analyzed using MO.affinity Analysis Software (NanoTemper).
Heparin binding
Heparin-coated microplates (Bioworld, Dublin, Ohio, USA) were used to analyze binding of the regulators to this glycosaminoglycan (GAG) analog. Bound proteins were detected with anti-FH1-4 (1:1000 in washing buffer) and HRP-conjugated anti-rabbit IgG from donkey (1:2000 in washing buffer, Cytiva).
Overall AP regulatory activity
The ELISA-based assay used to analyze the overall ability of the regulators to inhibit TCC formation after activation of the AP with lipopolysaccharides (LPS) was performed as described previously35 with slight modifications. Briefly, increasing concentrations of MFHR13, MFHR1I62, MFHR1V62, eculizumab, or hFH (0.5–100 nM) were tested and formation of C5b-9 complex was detected using a C9 neoepitope-specific antibody (aE11, Santa Cruz Biotechnology, 1:2000 in DPBS/0.05% Tween-20), followed by HRP-conjugated anti-mouse IgG goat (NXA931, Cytiva; 1:5000 in DPBS/0.05% Tween-20). Samples with normal human serum (NHS) and without regulators were set to 100% AP activity. Heat-inactivated NHS (56 °C for 30 min) was used as a blank. A negative control to indicate spontaneous activation was included, which consisted of NHS without LPS and regulators.
The ability of the regulators to protect sheep erythrocytes from complement-mediated lysis was measured as follows: Increasing concentrations of MFHR13, MFHR1I62, MFHR1V62, eculizumab or hFH (0.3–100 nM) were incubated with 5 × 107 sheep erythrocytes followed by the addition of 30% FH-depleted serum (Comptech, USA) prepared in GVB/MgEGTA buffer (0.1% gelatin, 5 mM Veronal, 145 mM NaCl, 0.025% NaN3, 5 mM MgCl2, 5 mM EGTA, pH 7.3). The reaction was incubated for 30 min at 37 °C and stopped with GVB/EDTA (0.1% gelatin, 5 mM Veronal, 145 mM NaCl, 0.025% NaN3, 10 mM EDTA). The amount of hemoglobin released was measured at 405 nm. Samples without regulators were set to 100% hemolysis, samples lacking NHS were included as negative controls and the values were subtracted from all samples.
Regulation of MAC formation on sheep erythrocytes
The inhibition of MAC formation on sheep erythrocytes was performed as previously described13 with modifications. C5b6 (3.5 nM) was incubated with increasing amounts of the protein of interest (MFHR13, MFHR1I62, MFHR1V62, hFH, FHR1 (R&D Systems) or eculizumab, 1000 nM) for 10 min. Then a mixture of C7 (9 nM), C8 (0.667 nM), C9 (15 nM) and 5 × 107 sheep erythrocytes was added (prepared in GVB/MgEGTA buffer) in 50 µL total volume. Hemolysis was detected after 40 min at 37 °C by addition of GVB/EDTA buffer. The amount of hemoglobin released was measured at 405 nm. Samples without regulators were set to 100% MAC-induced lysis. BSA or purified extract from the parental line (Δxt/ft) were included as controls. Two negative controls without C5b6 or C9 were included, which were subtracted from all samples.
Regulation of convertase-independent activation of C5 and MAC formation on sheep erythrocytes
The ability of the regulators to inhibit convertase-independent activation of C5 and subsequent MAC formation was tested in a hemolytic assay. For this, C3b-opsonization on erythrocytes was carried out as previously described76 with some modifications. To achieve maximal C3b deposition without significant MAC formation, FB was partially inactivated in FH-depleted serum at 50 °C for 5 min and 45 µL of this pretreated serum were added to 1 mL sheep erythrocytes (109/mL in GVB/EGTA-Mg2+) and incubated at 37 °C for 35 min. Cells were washed 4 times with DPBS and resuspended in 1 mL GVB/EDTA buffer (CompTech) to prevent formation of C3 convertases in case of residual FB.
C5 (75 nM) was preincubated for 15 min with MFHR13, FHR1, hFH, or eculizumab (700 nM). BSA, cytochrome c (Sigma 2037), or purified extract from the parental line (Δxt/ft) were included as controls. C3b-opsonized erythrocytes (5 × 107 cells) together with C6 (110 nM) were added to the regulator mix and incubated for 10 min at 37 °C. Then, C7 (120 nM), C8 (70 nM), C9 (180 nM) were added to a total volume of 50 µL. After 45 min at 37 °C 100 µL GVB/EDTA were added and hemolysis was detected by measuring absorbance at 405 nm. Lysis without regulators was set to 100%. The negative control without C5 was subtracted from samples and controls.
Statistics and reproducibility
Analyses were done with the GraphPad Prism software version 8.0 for Windows (GraphPad software, San Diego, California, USA). For experiments involving a dose–response curve, logarithmic transformed data were fitted by a four-parameter logistic (4PL) nonlinear regression model to calculate the IC50 and comparison of fits was carried out using the extra sum-of-squares F test with a cutoff at P = 0.05. Number of repeated experiments was indicated in the figure legends.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
