Using the human C-terminal dimerized minimal FH molecule (hHDM-FH, FH^{1-5^18-20^R1-2}) described by Yang, et al. (53) as the design template, a new mHDM-FH construct based on mFH (UniProtKB – P06909, amino acids 1- 321 and 1050 to 1234) and FHR-E (mouse CFHR-1, Uniprot Q61406, amino acid 24-152) sequence was designed and synthesized (Invitrogen, UK; see Supplementary Figure 1 for full sequence). mHDM-FH cDNA was subsequently cloned into expression vector pDR2EF1 using XbaI and NdeI sites. This newly generated construct was sequence verified and transfected into Chinese hamster ovary cells using jetPEI (Polyplus; VWR, Leicestershire, UK) following the manufacture’s protocols. CHO cells were cultured in DMEM F-12 medium supplemented with 10% FBS (Labtech International, Heathfield, UK) and penicillin-streptomycin solution (1:100, Sigma Aldrich, UK). Stable transfections were selected in the presence of 0.6 mg/ml Hygromycin B. According to ExPASy (https://web.expasy.org/) the protein backbone is 69747 daltons, pI 6.96 and has an Ext. coefficient of 117670, Abs 0.1% (=1 g/l) = 1.687.

All proteins used in this study were produced essentially as previously outlined, i.e. hHDM-FH (53). Briefly, for mHDM-FH, a single stable expressing clone was selected for protein production and scaled up in static tissue culture before transfer to roller bottles for 10 days (in the absence of Hygromycin B). Cell debris was removed by centrifugation and supernatant was sterile-filtered prior to loading on a 2 ml HiTrap NHS-activated high-performance column (GE Healthcare), which had been previously coupled (following manufacturers guidance) with the anti-mFH monoclonal antibody 2A5 (gift from Prof C. Harris, Newcastle UK), using an AKTA START (GE Healthcare) at a flowrate of 1 ml/min. Washes of 10 column volumes were performed prior to elution. Elution was achieved using 0.1M Glycine (pH of 2.7) and eluate was collected into 1M Tris pH 9.0. All protein-containing fractions were then combined and buffer exchanged into PBS using a PD-10 column (GE Healthcare). Proteins were further polished using a gel filtration column (Superdex 600, GE Healthcare) equilibrated with PBS buffer with an ÄKTA Purifier set to a flow rate of 0.5ml/min. One ml fractions were collected and stored at -80°C until needed.

All animals were housed in specific pathogen-free conditions and procedures were performed according to the appropriate institutional guidelines, which were approved by the United Kingdom Home Office.

For the 7-, 14-, 21- and 49-day experiments, mice (n = 3 – 4) were housed in the comparative biology centre, Newcastle University. CFH^-/- mice were provided as a gift from Matthew Pickering (Imperial College London). Wild Type C57Bl/6 mice were obtained from Charles River laboratories. C3^-/- mice were bred in-house as described previously in (60). Lipopolysaccharide was removed from all reagents using a method described previously by (61). Before the start of the experiments, CFH^-/- mice were weighed and blood was extracted through tail venesection (30µl of blood was collected into 5µl 125mM EDTA– placed on ice). Mice were then given an intraperitoneal injection of 5mg/kg mHDM-FH or equal volume sterile dPBS as a control. Blood was collected every 7 days via tail venesection, or via cardiac puncture (terminal exsanguination) and tissue samples were collected into dPBS on ice. Tubes plus EDTA were weighed before and after blood collection to calculate the dilution ratio. Plasma samples were obtained by centrifuging at 500g at 4°C for 5 minutes; aliquots were stored at -80°C. Tissue samples including the liver, kidney and spleen were dried before either embedding in OCT on dry ice or fixing in formalin; OCT embedded samples were stored at -80°C.

As reported (62), hepatotropic capsids were used to enable robust hHDM-FH transgene expression in the liver following a single intra-peritoneal injection. Ten CFH^-/- and 5 wild type mice received AAV-hHDM-FH. Five CFH^-/- and 5 wild type mice received AAV-GFP, as control. Mice received 1x10¹³ viral genomes (VG)/mL in 100 µl (total injected:1x10¹² VG). Venesection was performed pre-injection and on weeks 1, 3, 6, 9. One AAV-hHDM-FH treated mouse was culled just beyond 3 weeks after becoming unwell. For the remaining mice, EDTA plasma was collected by cardiac puncture at the 12 week end-point. Haematuria and proteinuria was assessed weekly and at endpoint using Hema-Combistix (Siemens). Pre-injection and end-point plasma albumin levels were measured using a mouse albumin ELISA kit according to the manufacturer’s instructions (Cat No. E99-134, Cambridge Biosciences, UK).

Samples from individual mice were separated using SDS-PAGE under reducing conditions for mC3 and non-reducing conditions for mC5, hFH and mFH. Goat anti-mC3 (MP Biomedicals), goat anti-hC5 (Quidel), rat anti-mFH (R&D Systems) or goat anti-hFH (Quidel) were used as detection antibodies followed by goat anti-rat-HRP (Biolegend), mouse anti-goat/sheep IgG-HRP (Sigma) and rabbit anti-mouse IgG-HRP (Dako), as required. Blots were visualized using ECL Western Blotting Substrate (Pierce, Thermo).

Mouse C3 was detected using goat anti-mC3 antibody (MP Biomedicals) in combination with an HRP-conjugated goat anti-mC3 polyclonal antibody (MP Biomedicals). Results were quantified using a standard curve generated from pooled mouse sera containing a known quantity of mC3 [in house, Imperial College London and as previously described (63)]. For hFH levels, ELISA plates were coated with 2µg/ml of an anti-FH monoclonal antibody (OX24, Thermo scientific MA1-70057) in 0.2M carbonate buffer. After washing and blocking with PBS-1% BSA, mouse plasma (1:166,600) was added in PBS-0.1% Tween-20-1% BSA. After incubation and washing, plates were incubated with a polyclonal sheep anti-hFH antibody (1:20,000, Abcam) for 1 h. After washing, plates were incubated with a donkey anti-sheep HRP conjugated antibody (1:20,000, Jackson ImmunoResearch via Stratech,UK) for 1 h. After washing, enzymatic activity was developed using TMB substrate. Results were quantified using a standard curve generated using purified hFH (Complement Technologies) and interpolated from a four-parameter linear regression curve generated in GraphPad Prism9.

Kidneys were fixed in Bouin’s solution (Sigma-Aldrich, Dorset, UK) for 2-4 hours and then transferred to 70% ethanol. Sections were embedded in paraffin and stained with periodic acid-Schiff (PAS) reagent (Sigma-Aldrich, Dorset, UK) and examined by light microscopy. Slides were reviewed in a blinded fashion and a semi-quantitative analysis was undertaken with sections being scored for glomerular hypercellularity (0-4), segmental Sclerosis (0-1), and mesangial expansion (0-1). A minimum number of 10 glomeruli were scored per mouse and an average taken.

Five µm kidney cryosections were mounted on superfrost plus slides and stored at -20°C. Slides were left to air dry for 10 minutes before placing in ice cold acetone for a further 10 minutes. Slides were air dried before washing in 0.1% PBS (Triton-x) for 5 minutes on a rocker. Slides were then carefully dried before using a PAP pen around each section. 50µl of blocking solution (10% Goat/Chicken serum (depending on the secondary antibody) in 0.001% PBS; Triton-X100) was applied for 30 minutes before removal. For antibodies conjugated with biotin, an Avidin/Biotin blocking kit was used according to manufacturer’s instructions (VectorLabs - SP2001). Then 50µl of antibody diluted in blocking solution was applied for 1 h in the dark. Slides were washed thrice in PBS for 5 minutes on a rocker before applying 50µl of secondary antibody for 1 h in the dark before washing three times and mounting using VectorLab mounting media with DAPI (Vector Laboratories, 2B Scientific, UK). IF antibodies used included: C3b/iC3b/C3c - FITC-conjugated goat anti-mouse C3 - 1:200 (MP Biomedicals); C3d - goat anti-mouse C3d 1:20 (R&D system); goat anti-mouse IgG (Fc specific) – FITC (Merck – F5387); Biotinylated-OX24 1:25 (mouse monoclonal anti-human FH (SCR5 specific), generated in house); Streptavidin Alexa Fluor 546 1:400 (Thermo Fisher - 10053112) and chicken anti goat Alexa Fluor 488 1:200 (Life Technologies – A21467). Quantitative immunofluorescence analysis was performed using Leica DM4B optical microscope coupled with Leica DFC700T digital camera (Leica Microsystems) or the Zeiss Axio Imager Z2 with a Hamamatsu flash 4.0 v2 camera. The slides were blinded and randomly scored. One slide was scored per animal. Initially, categorical IF scoring was performed using a staining intensity scale (0 – normal/absent, 1 – mild, 2 – moderate, 3 – strong, 4 - intense). Where discrimination of the fluorescence between slides/studies was not easily achieved by this approach we carried out further in-depth digital analysis using Zen 3.3 and Adobe Photoshop to assign an arbitrary fluorescence unit (AFU) value to each glomerulus visualized. 100% AFU was established through analysis of >50 glomeruli from 6 (3 male and 3 female) untreated 4-month-old CFH^-/- mice and using the mean value. Inclusion of untreated animals in all experiments and their analysis allowed data to be standardized to this value.

mHDM-FH and IgG were fluorescently labelled with XenoLight CF750 fluorophore (Perkin Elmer, #125677) according to manufacturer’s instructions. The relative intensity of each asset was assessed using an In Vivo Imaging System (IVIS Spectrum, Caliper Life Sciences). Images were acquired using the following IVIS Spectrum filter sets: Excitation (Ex) 745 nm and Emission (Em) 820 nm. To minimize fluorescent signal attenuation by black fur, all mice were shaved around the abdominal region and then IVIS imaged to obtain a baseline reading. Mice were then intra-venously (I.V.) (or intra-peritoneally, I.P., see Supplementary Figure 9 ) injected with 100µl of labelled mHDM-FH or labelled IgG. Biodistribution of mHDM-FH and mIgG was achieved by performing whole body IVIS imaging at least daily and up to 96 hr (as indicated in experiments).

Fluorescent signal was quantified using Living Image 4.7.2 software. Regions of interest (ROIs) were drawn in both the left and right kidney and the liver. Signal was quantified in the physical, calibrated unit - Radiant Efficiency [p/s/sr]/[µW/cm2].

After the final in vivo scan, mice were euthanized via cardiac puncture, blood was collected in 60µl 250mM EDTA and tissue samples (liver, kidney and spleen) were dissected. Each organ was imaged ex vivo. Plasma and tissue samples were processed as previously described.

Fluid phase co-factor activity assays were carried out as follows. Briefly, 1µg mouse C3b (CompTech, Complement Technology, Inc., Tyler, Texas) was mixed with 100ng hFI (CompTech) and cofactors at 333nM: full-length hFH (hFLFH); mFH CCP 1-5 (mFH1-5); full-length mFH (mFLFH), hHDM- FH or mHDM-FH made up to a total volume of 15μl in PBS buffer before incubation at 37°C for 1 hour. The proteolytic breakdown of mC3b was assessed by a 10-20% gradient SDS-PAGE gel followed by Coomassie staining. Gels were imaged using the Licor Odyssey Fc and densitometry analysis carried out using ImageStudio Lite (Licor). Density of the mC3b alpha prime chain resulting from each test was normalized to the value obtained for a no cofactor negative control (mC3b and hFI only) included in each experiment.

Sheep red blood cells (SRBC) in Alsevier’s solution (SB068, TCS Biosciences) were washed twice in dPBS (Lonza, UK) at 1:20 by centrifugation for 5 min at 800g and 4°C. Supernatant was discarded, and the pellet was re-suspended in 300µl of dPBS, 30µl of rabbit –anti-SRBC stroma (S1389, Sigma) was added and incubated for 30 min at room temperature (RT). Cells were washed 3x with 20mL gelatin veronal buffer (GVB, Sigma G6514) as above. The final pellet was re-suspended in 3mL GVB before adding of 10µL of mouse anti-rabbit IgG (Southern Biotech) and incubation for 40 min at RT. Cells were washed once as above in GVB. The resultant pellet was diluted in GVB such that 25µl of the mix in 150µl dH20 gave a max lysis OD412nm (reference 660nm, using a Labtech plate reader (Labtech, UK)) of ~1.6. To establish mouse serum required to give ~70% haemolysis, a round bottomed 96 well ELISA plate was placed on wet ice and a serum titration (in 25µl GVB) was carried out. Twenty-five µl of GVB-SRBC was added to the plate before incubation for 30 minutes at 37°C. The reaction was stopped by addition of 150µl ice cold GVB containing 50mM EDTA. The remaining cells were pelleted by centrifugation at 800g for 5 mins at 4°C. Without disturbing the RBC pellets, 180µl of the supernatant was transferred into a flat-bottomed 96 well plate for OD readings as above. Following the process above each drug (mHDM-FH, hHDM-FH, hFLFH and Mini-mFH) was titrated from 50nM, 50nM, 500nM and 500nM, respectively, before serum giving 70% lysis in absence of inhibitor, or control (GVB or dH2O -only) buffer, was added to each well. Cells were added last and incubated for 30 mins at 37°C. Percentage lysis was calculated as is standard for these types of experiments and as previously described (61).

SPR-based experiments to determine mHDM-FH kinetics were performed using a BIAcore S200 (Cytiva, UK) according to previously defined methods (60). In brief, a 2-fold dilution series of mHDM-FH (10000 to 19 nM) was made in HBST (10 mM HEPES, 150 mM NaCl and 0.005% Tween 20, pH7.4). The concentration series was flowed across a WT murine C3b amine coupled CM5 chip (Cytiva, UK) [~1200 response units (RU)], with injections across a control blank flow cell used for data normalisation. mHDM-FH samples were injected in duplicate at a flow rate of 30 µl/min. Each cycle contains a binding phase for 200 seconds, a dissociation phase for 300 seconds and injection of regeneration buffer (10 mM Sodium acetate, 1M NaCl, pH 4.5) for 60 seconds. Data was collected at 40Hz. The dissociation constant was calculated using steady-state model provided by the BIAevaluation S200 (Cytiva, UK) package.

Analysis was performed using GraphPad Prism version 9.0 (GraphPad Software, La Jolla, CA, USA) and significance threshold was p<0.05. Normally distributed data was analyzed using a paired (for correlated data) or unpaired T Test. The Mann-Whitney U test was used to compare differences between two independent groups when the dependent variable was either ordinal or continuous, but not normally distributed. For data with unequal variance, Welch’s t-test was used. One way ANOVA and Dunn’s multiple comparison test was used when 3 or more independent groups were analyzed.

We recently published in-depth analysis of two FHR dimerized mini-FH constructs, highlighting the potential of the C terminal dimer version as a treatment option for C3G (53). To explore longer term dosing with this agent and reduce the risk of ADAs, we developed a murine version of the drug. The amino acids mediating dimerization in hFHR1 are conserved in mCFHR1 a.k.a FHR-E ( Supplementary Figure 1A ) (28, 33, 35). We used NCBI banked FHR-E and mCFH sequence to make a synthetic construct for mHDM-FH (full sequence available in Supplementary Figure 1B ). To limit immunogenicity, we used a linker region between the mFH1-5 and 18-20 domains that was consistent with FH biology (43, 64), see Figure 1A . As the C-terminal dimerized version of hHDM-FH was the most active in vivo (53), we chose this version for longer-term dosing studies in mice ( Figure 1A ). The mHDM-FH construct was readily purified by affinity chromatography and gel filtration, as shown by limited contamination via SDS-PAGE ( Figure 1B ). Using analytical size exclusion gel filtration, the new mHDM-FH construct eluted at approximately the same volume as the mFLFH, while mini-mFH eluted much later ( Figure 1C ). These data are almost identical to analysis of hHDM-FH in our previous study (53).

The new mHDM-FH construct was a highly effective cofactor for human FI mediated cleavage of mouse C3b. The only significant difference between any of the cofactors analyzed in this assay was between mFH1-5 and mFLFH (p = 0.019); mFLFH appeared most efficient and mFH1-5 the least ( Figure 2A ). Whilst hHDM-FH was highly effective on mouse C3b, the same was not observed with mHDM-FH (or other mouse FH proteins) on hC3b ( Supplementary Figure 2 ). This is consistent with the findings recently described in a humanised C3 mouse model of C3G (65), indicating unidirectional effects of interspecies/evolutionary changes between the N-terminus of FH and C3b.

Mouse haemolysis protection assays with sheep erythrocytes [human-like interaction with FH (66)] were carried out to assess function, both in the fluid phase and at cell surface. As expected, increasing concentrations of mHDM-FH were highly effective in protecting sheep erythrocytes from lysis by autologous mouse serum ( Figure 2B ). These data indicated that mHDM-FH was approximately twice as effective in protection against lysis in a mouse system compared to hHDM-FH (IC50 of 8 versus 14nM, respectively) but was 25-fold more active than mFLFH and 10-fold more active than mini-mFH (61) ( Supplementary Table I ).

Using surface plasmon resonance (SPR), we determined the binding kinetic profile of mHDM-FH on mC3b, using a previously reported experimental set up (60). On a CM5 chip, amine coupled with mC3b, a doubly diluted concentration series of mHDM-FH (19nM to 5µM) or mFLFH (7nM to 10µM) was flowed across in duplicate. The dissociation constant of mHDM-FH from mC3b was calculated to be 0.5µM, which was approximately 3x lower than that of mFLFH (K _D ~1.7µM) ( Figures 2C, D ). These data are highly comparable to the data generated for hHDM-FH [FH^{1-5^18-20^R1-2} (53)] binding to hC3b.

To evaluate whether mHDM-FH functioned in vivo in a similar manner to the human version, 5mg/kg of mHDM-FH (every 72 hrs) or PBS only as control, was administered to CFH^-/- mice for a period of 7, 14 and 21 days in consecutive experiments. A marginal increase in the presence of intact C3 was visualized in collected plasma across the experiments although this was never significantly above the background signal (see Supplementary Figures 3A, B ). However, as expected, a significant reduction in glomerular deposition of C3 was noted, with the 14- and 21-day experiments giving the largest overall reduction ( Figure 3 ). These data are in line with the dosing experiments carried out with hHDM-FH in Yang et al. (53). Next, we evaluated the effects of extending our dosing schedule to once every 7 days. We carried out two parallel studies: one dose with day 7 endpoint and 7 doses (maximum allowed by current ethics/UK Home Office) with day 49 endpoint. Here, we saw little evidence of either increased C3 levels in the blood or consistent reductions in glomerular C3 -positive staining in treated animals ( Supplementary Figures 3C, D and 4 ). After 7 doses of 5mg/kg of mHDM-FH and analysis 7 days following the last dose, there was only a small reduction in C3 deposition in CFH^-/- mice treated with mHDM-FH as compared to control ( Supplementary Figure 4 ). Analysis of ADA in wild type mice dosed in a similar manner to CFH^-/- mice suggests that some ADA can develop to mHDM-FH but these were at a low titre ( Supplementary Figure 3E ) and do not appear to have altered the outcome in the CFH^-/- animals. In combination, these data suggest that a significantly higher dose would be required to achieve control of C3 deposition over a protracted dosing schedule. Therefore, we evaluated a 20mg/kg and 40mg/kg dose in a 7-day experiment. Our analysis suggested that a 7-day dosing pattern was not compatible with mHDM-FH in the CFH^-/- mice as we could not detect significant changes in either intact C3 in the fluid phase or reduction of C3 staining in the glomerulus ( Supplementary Figure 5 ).

The previous data suggest that mHDM-FH is not entirely being deposited and retained in the kidney or is not functioning optimally long term. In order to better understand the kinetics/biodistribution of mHDM-FH in the CFH^-/- mice, we labeled mHDM-FH and control mIgG with CF750 fluorescent dye. Immunofluorescence revealed that the functionality of mHDM-FH was not affected by the conjugation with CF750 fluorophore (as shown by reduced C3 deposits in the kidney, see Supplementary Figure 6 ). Following successful labelling, mHDM-FH and control mIgG were intravenously injected into their respective mouse groups. The biodistribution of CF750 labelled mHDM-FH and control mIgG was monitored by performing whole body fluorescence imaging at 0, 2, 6 & 24 hours ( Supplementary Figure 7 ). To determine the origin and specificity of the signal, organs were harvested and imaged. Ex vivo imaging showed that mHDM-FH predominantly accumulated in the liver, kidney and spleen with no substantial fluorescence in other tissues at 30 hours ( Supplementary Figure 7 ). Quantitative analysis showed that there was a significant difference between mIgG and mHDM-FH accumulation in the liver, spleen and kidney ( Supplementary Figure 8 ) at this time point, irrespective of injection route (I.V. versus I.P., Supplementary Figure 9 ) although data was most consistent following I.V. dosing and so this injection route was favored. We next extended the analysis to 96 hours. The IVIS imaging revealed that the fluorescent signal originating from mHDM-FH was detected in both kidneys, as well as in the upper quadrant of the abdominal region where the liver is located. In contrast, the mIgG control displayed very little fluorescent signal in these anatomical regions ( Figure 4 ).

Ex vivo tissue analysis at 96hr post administration confirms mHDM-FH is strongly associated with the liver and remains visible in the kidney and the spleen ( Figure 5 ). Thus, mHDM-FH also likely binds to C3b complexes in the liver (and spleen) and that this has a significant impact on the drug availability in the fluid phase. This is underlined by the fact that even wild type mice have significant drug staining in liver ( Figure 5 ). These data suggest that C3 production and breakdown in the liver provides a significant sink for mHDM-FH and potentially many other drugs that bind to C3b or its break down products.

Given the poor efficacy of weekly dosing, we next used gene therapy to deliver the HDM-FH to the CFH^-/- animals. In this experiment, we used the human protein (hHDM-FH). We administered AAV-hHDM-FH to CFH^-/- (n=10) and WT (n=5) mice and included AAV-GFP injected CFH^-/- (n=5) and WT (n=5) mice as control groups. Following AAV-hHDM-FH injection, plasma hHDM-FH levels reached around 530µg/ml within 7 days ( Figure 6A ) except in two WT mice and one CFH^-/- mouse, where levels only reached ~100µg/ml ( Figure 6A , colored symbols). Plasma C3 levels in all the CFH^-/- mice treated with AAV-hHDM-FH reached levels comparable to those in both WT mice injected with AAV-hHDM-FH ( Figure 6B ) and WT injected with AAV-GFP ( Supplementary Figure 10B ). As expected, expression levels of hHDM-FH correlated with plasma C3 levels ( Figures 6A, B , and Supplementary Figure 10C ). Plasma C5 became detectable by western blotting in AAV-hHDM-FH treated CFH^-/- mice ( Figure 6C ). At the 12 week cull point, glomerular C3 glomerular deposition was reduced in AAV-hHDM-FH treated CFH^-/- mice with residual staining appearing diffuse and granular in appearance ( Figures 6D–G ). Glomerular C3 and C3d deposition in AAV-GFP treated CFH^-/- mice remained unchanged ( Figures 6D-G and Supplementary Figure 11A ). C3 deposition in WT mice treated with AAV-hHDM-FH was mesangial in appearance ( Supplementary Figures 11B, C ). There was also evidence of glomerular IgG deposition, but this was not different between the AAV-GFP and AAV-hHDM-FH treated CFH^-/- mice ( Supplementary Figures 10E, G ). Using mAb OX24 to detect glomerular hHDM-FH, glomerular staining was present in 4 AAV-hHDM-FH treated mice (e.g. mouse no.7, Figures 6D, E ), while the remaining 5 mice showed no detectable staining (e.g. mouse no 6, Figures 6D, E ). Using a polyclonal anti-hFH antibody there was increased glomerular staining in all AAV-hHDM-FH treated CFH^-/- mice compared to the AAV-GFP treated CFH^-/- mice ( Supplementaey Figures 10E, F ). Whether this represents increased glomerular mFHR proteins or hHDM-FH or a combination is unclear. We next measured ADA in the WT and CFH^-/- mice treated with AAV-hHDM-FH ( Supplementary Figures 10H, I ). Two WT and two CFH^-/- mice generated significant quantities of ADA ( Supplementary Figure 10H ) evident from 2 weeks after administration of ( Supplementary Figure 10I ). These animals generally had the lowest levels of hHDM-FH ( Supplementary Figure 10J ). Glomerular IgG, C3 and positive glomerular staining with OX24 was present in all four animals with high levels of ADA ( Supplementary Figure 11C ) suggesting that there could be glomerular ADA/hHDM-FH immune complexes triggering C3 activation in these animals.

As noted in the methods, one AAV-hHDM-FH treated CFH^-/- was removed from study at 3 weeks because it displayed 3+ proteinuria and 3+ haematuria ( Supplementary Figure 12 ). Renal histology showed abnormal glomerular deposition of PAS-positive material consistent with thrombosis and glomerular staining for hHDM-FH, C3 and IgG ( Supplementary Figure 12 ). The cause of this phenotype is unclear but it was an outlier as there was no significant proteinuria or haematuria in the remaining 9 CFH^-/- or 5 WT mice treated with AAV-hHDM-FH. At endpoint, mild proteinuria was detectable in the majority of mice, including WT AAV-GFP treated animals ( Supplementary Figure 13A ). Renal histology by light microscopy did not differ between AAV-GFP and AAV-hHDM-FH treated mice ( Supplementary Figure 13D ). Serum albumin was marginally lower in AAV-hHDM-FH treated compared to AAV-GFP treated CFH^-/- animals ( Supplementary Figure 13B ). When AAV-hHDM-FH levels of albumin were compared between start point of the experiment and endpoint ( Supplementary Figure 13C ), no significant change was noted in other groups.