Written informed consent was obtained from all aHUS patients or their legal guardians, and from all healthy controls of whom blood samples were used in this study. Informed consent was also obtained for isolating GMVECs from an aHUS patient (P1) who had bilateral nephrectomy in preparation to kidney transplantation. The study was performed in accordance with the appropriate version of the Declaration of Helsinki.

Six patients diagnosed with aHUS were included in this study and were genetically characterized ( Table 1 ). P1 was identified with a heterozygous deletion of CFH exon 22, CFHR3 and CFHR1 exon 1-5, indicating a genomic rearrangement resulting in a CFH/CFHR1 hybrid gene (35). This rearrangement has been described to lead to a severe aHUS phenotype (36). P2 and P4-P6 carried a pathogenic C3 variant, i.e. c.481C>T (p.Arg161Trp), resulting in decreased binding to FH and MCP, and increased binding to complement factor B, leading to a C3 convertase with increased activity (34, 37, 38). In P3, a homozygous CFHR3-1 deletion was found in combination with autoantibodies against FH (39–41). P3 (P4 in Bouwmeester et al.), P4 (P12 in Bouwmeester et al.), and P5 (P18 in Bouwmeester et al.) were described in the CUREiHUS study (18). All patients were treated according to a restrictive eculizumab protocol, as described in the CUREiHUS study (18, 43).

P1 is known with aHUS from the age of 5 months and treated until the age of 11 years with chronic plasma therapy. He then was successfully switched to treatment with eculizumab. At the age of 15 years a further deterioration of his kidney function was observed without any signs of recurrence of aHUS. His radiographic examinations revealed an acquired glomerulocystic disease, a reduced left kidney function, and an abnormal venous system of unknown origin. A bilateral nephrectomy was performed to minimize the risk for aHUS recurrence after kidney transplantation. The blood taken at the age of 15 years and used in this study was marked as day 0 (d0). The medical history of this aHUS patient (P1) was in depth studied and reported previously (36).

TMA was defined by ≥2 of the following criteria: thrombocytopenia (platelet count <150 ×10⁹/l), lactate dehydrogenase (LDH) above the upper limit of normal (>250 U/l) and low/undetectable haptoglobin (<0.3 mg/l). AKI was defined as an increase in serum creatinine of ≥26.53 µmol/l (0.3 mg/dl) in 48 hours or ≥1.5 times baseline in <7 days.

GMVECs of P1 were isolated from kidney tissue obtained after nephrectomy and subsequently conditionally immortalized as described by Satchell et al. (44). Control ciGMVECs (obtained from Satchell et al. (44)) and P1 ciGMVECs were allowed to proliferate by culturing them at 33°C with 5% (v/v) CO₂ on 96-well plates (655090, Greiner, Cellstar^®) coated with gelatin (48720, FLUKA). When confluent, cells were transferred to 37°C with 5% (v/v) CO₂ to induce growth arrest. After 7-8 days, cells were used for experiments. The growth medium, later referred to as endothelial medium, consisted of M199 medium (22340, Gibco) supplemented with 10% (v/v) heat-inactivated (HI) human serum (ISER, Innovative research), 10% (v/v) HI newborn calf serum (26010, Gibco), 2 mM L-glutamine (25030, Gibco), 100 U/ml penicillin, 100 mg/ml streptomycin (15140, Gibco), 5 U/ml heparin (Leo Pharmaceuticals), and 150 µg/ml bovine endothelial growth factor made according to the protocol described by Maciag et al. (45).

Primary GMVECs from human origin were obtained from kidney tissue after nephrectomy and cultured as previously described (46). Primary GMVECs were cultured at 37°C with 5% (v/v) CO₂ on gelatin-coated 96-well plates with endothelial medium until confluent.

HMECs-1 (ATCC^® CRL3243™) were cultured at 37°C with 5% (v/v) CO₂ on uncoated 96-well plates with MCDB 131 medium (10372, Gibco) supplemented with 50 µg/ml bovine endothelial growth factor, 10% (v/v) fetal calf serum (FCS; 10270, Gibco), 2 mM L-glutamine, 1 µg/ml hydrocortisone (H0888, Sigma-Aldrich), 100 U/ml penicillin, and 100 mg/ml streptomycin.

Blood samples of aHUS patients and healthy individuals were collected and processed as described previously (47). Pooled normal human serum (NHS) was obtained by pooling 10 or 14 healthy individual samples. To obtain HI-NHS, NHS was incubated for 30 minutes at 56°C. Samples were aliquoted and stored at -80°C until use.

Serum samples were thawed on ice and diluted to the desired serum concentration in Hanks’ balanced salt solution containing calcium and magnesium (HBSS++, 14025; Gibco) supplemented with 0.5% (w/v) bovine serum albumin (BSA; 820451, Millipore), later referred to as test medium. Where indicated, sera were pre-incubated for 5-10 minutes at room temperature with 300 µg/ml of the C5 blocker eculizumab (Soliris^®, Alexion Pharmaceuticals) and/or 0.3 mg/ml zymosan (Z4250, Sigma-Aldrich). Zymosan was pre-diluted in magnesium-ethylene glycol tetra acetic acid buffer (Mg-EGTA; 2.03 mM veronal buffer, pH 7.4, 10 mM EGTA, 7 mM MgCl₂, 0.083% gelatin, 115 D-glucose, and 60 mM NaCl). Eculizumab and zymosan concentrations are indicated per volume of undiluted serum.

Cells cultured on 96-well plates were washed twice with test medium and pre-incubated with test medium for 15 minutes at 37°C with 5% (v/v) CO₂. Next, cells were incubated with 33.3% serum in test medium with a final volume of 140 µl at 37°C with 5% (v/v) CO₂ for 2 hours, unless stated otherwise. After incubation, cells were washed twice with HBSS++, fixed with 3% paraformaldehyde/phosphate buffered saline (PBS; 10010, Gibco) for 10 minutes, and washed twice with PBS.

For immunofluorescence staining, cells were first blocked with 2% (w/v) BSA/PBS for 1 hour. Then, cells were stained with mouse anti-CD31 (1:100, 555444, BD Pharmingen) and/or rabbit anti-C5b-9 (1:200; 204903, Calbiochem), diluted in 1% (w/v) BSA/PBS with 10% (v/v) goat serum (ab7481, Abcam), for 1 hour. After washing three times, cells were incubated with the secondary antibodies (1:1000 in 1% BSA/PBS) Alexa488-conjugated goat anti-rabbit (A11008, Invitrogen), Alexa568-conjugated goat anti-mouse (ab175473, Abcam), and/or FITC-conjugated rabbit anti-C3c (1:20; F0201, Dako; recognizes the C3c part of C3, C3b and iC3b), for 1 hour. Cell nuclei were additionally stained with 5 µg/ml 4’, 6-diamidino-2-phenylindole (DAPI; D1306, Invitrogen) for 5 minutes. All steps were performed with static conditions at room temperature. Images were captured with Zeiss Axio Observer 7 with artificial intelligence sample finder in combination with ZEN imaging software (Zeiss).

For the quantification of C5b-9 depositions, an area of 4 mm² of each well was visualized, with focusing on DAPI and with equal visualization settings for all experiments. Fluorescence images were analyzed utilizing Fiji 1.51n software with a custom-made ImageJ macro to quantify C5b-9 (48).

Control ciGMVECs in 24-well plates (3524, Costar^®, Corning Incorporated) were washed twice with test medium and subsequently incubated with 10% (v/v) serum in test medium with a final volume of 280 µl at 37°C with 5% (v/v) CO₂ for 1 hour. Next, cells were washed twice with HBSS++ and fixed with 0.5% paraformaldehyde/PBS for 15 minutes. After fixation, cells were stained with rabbit anti-C5b-9 (1:100) in 0.1% (v/v) BSA/PBS at 4°C for 30 minutes. After three washes with PBS, cells were incubated with Alexa488-conjugated goat anti-rabbit (1:100) at 4°C for 30 minutes. Cells were detached with 0.05% trypsin-ethylenediamine tetra acetic acid (25300, Gibco) and dissolved in 10% (v/v) FCS/PBS. Cells were spun down and resuspended in 0.1% (v/v) BSA/PBS. C5b-9 deposition was quantified using a CytoFLEX flow cytometer (Beckman Coulter) and results were analyzed with Kaluza 2.1.3 software (Beckman Coulter).

Clinical and standard laboratory data were obtained from patients’ electronic medical records. Serum C5 levels were measured with enzyme-linked immunosorbent assay (ELISA) as described previously (49). Patients were tested for autoantibodies against FH using an in-house ELISA (50).

Data are expressed as mean ± standard deviation. All statistical analyses were performed with GraphPad Prism version 9.0.0 for Windows (GraphPad Software, San Diego, California, USA). Data were analyzed by unpaired t-test, one-way ANOVA, or two-way ANOVA, followed by post-hoc analysis as indicated in figure legends. A p-value (P) of <0.05 was considered as statistically significant.

Human endothelial cell types from different origin have been used previously to model complement deposition ex vivo. We focused on studying complement activation on glomerular endothelial cells as these are the primary target in aHUS. Therefore, we first compared the performance of primary GMVECs with ciGMVECs as cellular models for complement activation ex vivo, by investigating the endothelial C5b-9 deposition upon incubation with NHS and zymosan-stimulated NHS. A baseline level of C5b-9 deposition was observed for both primary control GMVECs and control ciGMVECs upon incubation with NHS ( Figures 1A, B ). The amount of depositions significantly increased when the cells were incubated with NHS supplemented with zymosan to simulate an activated complement system ( Figures 1A, B ). Similar results were seen for the HMECs-1 ( Supplementary Figures 1A, B ), which is the most frequently described cellular model for aHUS (27–31), thereby further validating our results on the glomerular endothelial cell model.

To determine the discriminatory power between baseline C5b-9 deposition (NHS) and the C5b-9 level after incubation with activated complement (NHS-zymosan), the fold change of NHS-zymosan to NHS was calculated for primary GMVECs and ciGMVECs ( Figure 1C ). No significant difference was measured. We continued with validating ciGMVECs as a model for complement activation on the glomerular endothelial surface.

To determine if we could adequately and reliably measure complement-mediated C5b-9 deposition on the ciGMVECs, we visualized and quantified C5b-9 on the ciGMVECs upon incubation with different test sera in which complement was activated or blocked.

First, we showed that incubation of ciGMVECs with either HI-NHS or NHS supplemented with eculizumab resulted in decreased C5b-9 depositions compared to NHS ( Figures 2A, B ). The increase in C5b-9 deposition after incubation with NHS-zymosan could also be inhibited when serum was first treated with eculizumab ( Figures 2A, B ). With flow cytometry as an alternative method for C5b-9 quantification we further confirmed our microscope imaging findings for ciGMVECs ( Supplementary Figure 2 ). In line with the C5b-9 deposition, we showed that the deposition of C3 was increased upon incubation with NHS-zymosan ( Supplementary Figure 3 ). Together, these results indicate that the generation of C5b-9 on the endothelial surface was complement-mediated. Incubation of ciGMVECs with serum of aHUS P1 resulted in increased C5b-9 deposition compared to NHS ( Figures 2A, B ). No significant differences in the C5b-9 deposition levels were observed between 1-, 2- or 3-hour serum incubations ( Supplementary Figure 4 ). Two-hour incubation with serum was used as the standard method for further experiments.

Next, we titrated eculizumab in NHS-zymosan to assess how the cellular model reflects in vivo (fluid-phase) complement inhibition by eculizumab. The target serum level of eculizumab for aHUS patients for adequate blockage of complement is above 50-100 µg/ml (51, 52). In line with this, we observed maximal blockage of C5b-9 depositions on ciGMVECs with concentrations of 100 µg/ml and higher, and partial inhibition of C5b-9 formation with 50 µg/ml eculizumab ( Figure 2C ). Concentrations of 20 µg/ml and lower did not affect the C5b-9 depositions induced by activated serum.

After validating ciGMVECs as a model to specifically measure complement-mediated C5b-9 depositions, we explored their usage for modeling aHUS by comparing aHUS patient samples obtained from different disease phases. Incubation of ciGMVECs with sera from P2, P3, P4, P5, and P6 obtained in the acute phase resulted in increased C5b-9 deposition levels, ranging from 1.37-7.69 fold the levels generated by NHS ( Table 2 ; Supplementary Figure 5 ). In all serum samples, eculizumab levels were below the detection limit of 8 µg/ml.

Surprisingly, the sample of P4 obtained during evident hematological TMA and AKI (d10; Table 2 ) showed only slightly increased C5b-9 deposition compared to NHS. On the contrary, a second acute phase sample (d211; Table 2 ), showed higher C5b-9 deposition, despite the clinically less pronounced TMA. Of note, both relapses were triggered by a viral infection.

The remission samples of P2 and P5 with an eculizumab level above the therapeutic target suggesting adequate complement blockage, showed normalized C5b-9 depositions ( Table 2 ; Supplementary Figure 5 ). The remission sera from P3 and P4, obtained while in remission after eculizumab withdrawal, also resulted in normalized endothelial C5b-9 deposition levels ( Table 2 ; Supplementary Figure 2 ).

P1 developed a severe form of aHUS from the age of 5 months, which is described in detail in the materials and methods and by Bouwmeester et al. (36). We thoroughly monitored P1 during his chronic kidney failure and eculizumab treatment with various complement (activation) and TMA markers in the blood and studied if these parameters correlated with endothelial C5b-9 depositions. Interestingly, all samples of P1 in which the eculizumab concentrations were below the therapeutic target (<50 µg/ml) showed increased C5b-9 deposition on ciGMVECs ( Figure 3A ), irrespective of the clinical disease phase the samples were taken from. Thus, also samples taken when the patient did not show active signs of TMA showed increased cellular C5b-9 deposition. During adequate complement blockage, indicated by serum eculizumab levels above 50-100 µg/ml and a CH50 <10%, C5b-9 deposition was comparable or below NHS level ( Figure 3A ).

As this patient was not a “typical” aHUS patient because of ongoing decline in kidney function (described by Bouwmeester et al. (36)) and high C5b-9 deposition levels despite “quiescent” TMA markers, we hypothesized the glomerular endothelium of this patient might be more susceptible to complement activation and/or complement-mediated damage. When this patient received a kidney transplant, we were able to isolate GMVECs from the tissue of the removed native kidney from this patient to test this hypothesis. Incubation of the same series of serum samples obtained during follow-up with the patient’s own ciGMVECs (P1 ciGMVECs) resulted in C5b-9 deposition that was similar to the deposition on control ciGMVECs ( Figure 3A ). Also, the patient cells were similarly susceptible to C5b-9 deposition induced by NHS ( Figure 3B ). The C5b-9 levels were not higher for P1 ciGMVECs compared to control ciGMVECs based on the fold change compared to NHS after incubation with different test sera ( Supplementary Figure 6 ).