The HLA typing of a male patient, aged 67, diagnosed with rapidly progressive glomerulonephritis and his donor was performed with sequence-specific oligonucleotide primer (SSOP) technology, analyzed with the IMGT/HLA allele database 3.23. The detailed description of the case is depicted in the section describing the clinical history.

The presence of HLA class I and II Abs in the patients’ serum was evaluated using the Life Screen Deluxe (LsdL), in accordance with the manufacturer’s protocol (Immucor GTI Diagnostics Inc., Waukesha, United States, lot. 3003946-3003920)) and with the Lifecodes Single Antigen Bead (LSA; Immucor Transplant Diagnostics) assay. The routinely lymphocyte-based complement-dependent cytotoxicity tests were used to test for panel-reactive and donor-specific antibodies. An overview of HLA antibody detection and endothelial cell crossmatch (ECXM) tests and timepoints is given in Figure 1.

Pre- and postimplantation biopsies were stained with endothelial-specific marker mouse anti-human CD34 (QBEnd/10,Ventana, Mannheim, Germany) and rabbit anti-human ETS-related gene (ERG) (ERP3864, Ventana, Mannheim, Germany) according to the manufacturer’s protocol, as part of the standardized pathological evaluation of biopsies. Postimplantation biopsies were stained with goat anti-human IgG (760-2680, Ventana, Benchmark automated immunostainer), goat anti-human IgM (760-2682, Ventana, Benchmark automated immunostainer), and rabbit anti-human C4d (clone SP91, Ventana, Benchmark automated immunostainer) also according to the manufacturer’s protocol, as part of the standardized pathological evaluation of biopsies. C3d staining was performed using in-house staining. Solutions were prepared using with BSA (A9647, Sigma, St Louis, United States) and PBS (17-512Q, Lonza, Wijchen, The Netherlands) In short, deparafffinization was performed using xylol and alcohol (100%, 96%, 70%) and demi-water in 6 consecutive steps. This was followed by incubation in the dark for 30 min, at 37°C, with 0,4% pepsin from porcine stomach mucosa in 37% 0.1N HCl in demi-water (pH = 2.5). A blocking step was then performed by first adding 0.01% H₂O₂ (in 1% BSA with PBS) followed by 1% BSA in PBS. The polyclonal rabbit antihuman C3d (A0063, Dako, Glostrup, Denmark) at 4°C overnight in a 1:2,500 dilution in 1% BSA in PBS was then added. This was followed by the first conjugate, HRP polyclonal goat anti-rabbit antibody (P0448, Dako), at room temperature in a 1:100 dilution in 1% BSA in PBS and the second conjugate HRP on polyclonal rabbit anti-goat antibody (P0449, Dako) at room temperature in a 1:100 dilution in 1% BSA in PBS. The substrate was then added with 0.2 mg/ml 3-amino9-ethylcarbazole (02431MH, Sigma), in 50 mM acetate buffer and 0.03% H₂O₂ (pH = 5.5) followed by a counterstaining and embedment using hematoxylin 1:2 for 5 s and a Kaiser’s glycerol gelatine embedment (1092420100, Merck, Darmstadt, Germany). An overview of all antibodies used is given in Supplementary Table S1.

To provide direct evidence for the presence of potentially pathologic (i.e., cytotoxic) alloantibodies against the machine perfusion-derived primary renal endothelial cells (MP-PRECs), a cytototic crossmatch was performed combining patient serum and primary renal endothelial cells (ECs). If non-HLA or HLA antibodies are present in patient serum, these antibodies can bind the ECs and form antibody–antigen complexes that activate the complement cascade, leading to complement-mediated cytotoxicity. The Terasaki typing trays were prepared as follows: 10 µl patient sera was added to the Terasaki wells. As positive control serum containing HLA antibodies directed against HLA-A2 (A2/A28 IgG1 and IgM, provided by Dr. F. Claas, Leiden University Medical Center, The Netherlands) expressed on the endothelial cell was used. As negative control normal human serum (NHS) from healthy volunteers with a compatible blood group and without HLA antibodies was used. Culture MP-PRECs were dissociated using nonenzymatic cell dissociation solution (C5789, Sigma^®, Zwijndrecht, The Netherlands). Cells were isolated and characterized as described before, and clinical baseline characteristics regarding the donors of the MP-PRECs can be found in our recently published paper describing the isolation and characterization of the MP-PRECs in detail (18). Experiments were performed with cells cultured up to passage 3–5. The cells were pelleted and resuspended with culture medium (Medium 200 (GIBCO, Cat#M-200-500, Grand Island, NY, United States) containing Low Serum Growth Supplement (LSGS, GIBCO, Cat#S-003-K, Grand Island, NY, United States) for a final concentration of 2000cells/µl. 1 µl of the cell suspension was added to each terasaki well. Cells were incubated for 30 minutes with sera followed by 60 minutes 5 µl rabbit complement or 60 minutes with sera followed by 120 minutes 5 µl rabbit complement at room temperature. Ethidiumbromide and acridineorange were added to stain for apoptotic and living cells. The reaction was stopped by adding sodium detaat. Results are expressed as percentage of apoptotic cells in relation to the total amount of cells.

A confocal inverted laser microscope (Leica) was used for acquisition of the staining. The fields, systematically digitized throughout the well, were acquired using a computer-based image analysis system. The staining was quantified using built-in specific functions of the software Image J (NIH, Bethesda, MD) and expressed as percentage per well.

To explore classical pathway complement activation, MP-PRECs were cultured in a 12-well culture plate until they reached confluence and were detached using cell dissociation solution (C5789, Sigma^®, Zwijndrecht, The Netherlands), 900 μl/1 ml at 37°C, collected in 4.5 ml tubes containing 2 ml cell of medium and centrifuged twice at 250×g for 6 min at 20°C. Thereafter, MP-PRECs were incubated with 25% blood group-compatible heat-inactivated NHS, or 25% blood group-compatible heat-inactivated serum containing HLA-A2 directed against the antigens on the cell membrane, or 25% of the patient’s serum, diluted in M200 medium (M200 culture medium, Ref. No. M200500, Gibco, Bleiswijk, The Netherlands) supplemented with low growth serum (supplement kit, Ref. No. S-003-K, Gibco, Bleiswijk, The Netherlands) for 45 min at 37°C. Hereafter, cells were centrifuged twice for 6 min at 250×g at 20°C in culture medium, followed by incubation with 20% NHS as a complement source for 30 min at 37°C. After incubation, MP-PRECs were washed twice with ice-cold phosphate-buffered saline (PBS)/1% bovine serum albumin (BSA) (FACS buffer) (Sigma^®, Zwijndrecht, The Netherlands) at 250×g for 6 min at 4°C. For detection of IgG(G18-145, BD Biosciences, San Jose, United States), IgM (MHM-88, BioLegend, San Diego, CA, United States) and complement activation [activated C3 (HM2168, Hycult, Uden, The Netherlands)], the MP-PRECs were incubated with antibodies depicted in Supplementary Table S1, for 30 min on ice. Cells were washed twice with ice-cold FACS buffer, centrifuged at 250×g for 6 min at 4°C, and incubated with goat anti-mouse FITC (Southern Biotech, Birmingham, United States) for 30 min on ice in the dark. Propidium iodide at 1 μg/ml (Molecular Probes, Leiden, The Netherlands) was added just before measuring to be able to exclude apoptotic and necrotic cells. Activated C3 deposition on viable nonapoptotic cells were analyzed in a FACSCalibur™ (FACSCalibur, Becton Dickinson, New Jersey, United States). The results are from at least three independent experiments. The percentage effect was calculated based on the control data (blood group and HLA-compatible serum) in mean fluorescence intensity (MFI); % binding is = (MFI test result/MFI control) × 100.

Rabbit complement (ThermoFisher Scientific CL3115, Cedarlane, Burlington, Canada) and normal human serum (NHS) from healthy volunteers having blood group AB were used separately as a complement source. NHS and rabbit complement were stored at −80°, rapidly thawed at 37°C and diluted in M200 medium to a final concentration of 25% for human complement and 20% for rabbit complement prior to use. As control, NHS was incubated at 56°C for 30 min to inactivate complement proteases.

The patient’s serum from postoperative day 3 was measured using the multiplex assay. The development, technical details, and validation of this high-throughput multiplex assay for the detection of non-HLA antibodies are described extensively in the study by Kamburova et al. (19)

Statistical tests were conducted for complement component measurements. Differences in the percentage effect and percentage apoptosis were analyzed by the unpaired t-test. Computations were performed by SPSS (IBM Statistics Chicago, United States) version 23, and GraphPad software 8.0 (Graphpad Software, San Diego, United States) was used for graphical visualization.

The patient presented in July 2015, aged 67, with rapidly progressive glomerulonephritis, a serum creatinine of 583 µmol/l and an anti-glomerular basement membrane (anti-GBM) antibody titer of >600 IU/ml without pulmonary involvement. He was diagnosed with anti-GBM glomerulonephritis based on the clinical parameters and a positive anti-GBM titer but not by kidney biopsy. He was treated with hemodialysis, plasmapheresis, cyclophosphamide, and prednisone without improvement of kidney function. During this treatment, he received one blood transfusion. The patient’s 70-year-old brother was considered eligible for kidney donation, and in November 2016, the patient received a living-related AB0 compatible kidney transplantation. Donor and recipient HLA typing was A2, A68, B62, B44, Bw6, Bw4, Cw1, Cw2, DR4, DR14, DR52, DR53, DQ8, and DQ5 and A2, A68, B7, B44, Bw6, Bw4, Cw2, Cw7, DR14, DR15, DR52, DR51, DQ5, and DQ6 for donor and recipient, respectively, resulting in a 0-1-1 HLA mismatch. The donor and recipient both tested positive for the Epstein–Barr and cytomegalo viruses. The last anti-GBM titer before transplantation was 14 IU/ml. No panel reactive and donor-specific antibodies were detected prior to transplantation in the lymphocyte based CDC test, LSdL, and LSA assay.

No relevant pre-existing damage was found in the routine preimplantation biopsy of the kidney (Figures 2A, B), and no surgical complications occurred (cold ischemia time, 142 min; warm ischemia time, 33 min) and renal perfusion, as determined by Doppler ultrasound, was excellent with immediate diuresis. Serum creatinine values declined from 940 to 514 µmol/L after transplantation. The immunosuppressive regimen consisted of mycophenolate mofetil (1,000 mg bd), tacrolimus (0.075 mg/kg bd), and prednisone (40 mg iv qd), according to the standard protocol. Within hours after transplantation, the patient developed a fever and diuresis dropped from 350 to 75 ml/h.

On postoperative day (POD) 1, nasopharyngeal, urine, and blood cultures were performed, and prophylactic ceftriaxone was started. Diuresis declined to 25 ml/h without a response to fluid challenges and cultures returned negative.

On POD 2, diuresis was 430 ml/24 h accompanied with hematuria. Doppler ultrasound showed diminished perfusion of the kidney with moderate swelling of the renal parenchyma and no signs of renal vein or artery thrombosis. Thrombocytes declined from 117 × 10 (9/L to 59 × 10⁹/L, C-reactive protein values increased to 197 mg/L, serum creatinine values increased to 650 µmol/L, and anti-GBM titer was 10 IU/ml. A Tc-99m MAG3 scintigram revealed poor perfusion of the transplanted kidney; however, during re-exploration surgery, no vascular complications were found, though the kidney had a purple discoloration and was swollen and stiff. A biopsy taken during re-exploration revealed granulocytes and lymphocytes in the glomerular capillary loops with occasional small amounts of intraluminal fibrin. Granulocytes were also present in the peritubular capillaries and associated with detachment and destruction of endothelial cells. There was extensive peritubular hemorrhage (Figures 2C, D). The tubuli appeared ischemic, with flattening of the epithelial cells. No arterial necrosis was found. Immunoglobulin G (IgG) and immunoglobulin M (IgM) staining by immunofluorescence was barely seen (data not shown), and the C4d staining was negative (Figure 3A). Additional staining for endothelial cell-specific markers CD34 and ERG showed less CD34- and ERG-positive cells compared to a posttransplantation biopsy without rejection (Figures 4A,B). This was most apparent in the peritubular capillaries, indicating a pronounced loss of endothelial cells. C3d staining was most pronounced in the peritubular capillaries and the glomerular hilus (Figure 3B). Despite the negative C4d staining, humoral rejection was strongly suspected; hence, prednisone, intravenous immunoglobulin (IVIg), and plasmapheresis was started. On POD 3, no HLA antibodies were found in the LSdL and the LSA assay, nor against A/B/DR loci, nor against C, DQ, and DP loci. In addition, serum creatinine increased to 740 µmol/L, and the patient required dialysis again. During the following days, the patient continued to be treated with plasmapheresis, IVIg, and prednisone. Despite this therapy, the kidney function did not improve, and this resulted in a transplantectomy on POD 13.

Despite the pathological signs of hyperacute ABMR with capillaritis, we could not detect HLA antibodies in patient serum obtained pretransplant or on POD3. Therefore, non-HLA-mediated rejection was considered. To determine this, we crossmatched the patient’s serum taken 1 day prior to transplantation and from POD 3 with human MP-PRECs, utilizing rabbit complement.

We crossmatched patient serum with MP-PRECs that were blood group compatible and added rabbit complement as a complement source. Crossmatching with serum taken 1 day prior to transplantation resulted in clear complement-mediated cell death; 72% of all cells appeared to be apoptotic (Figure 5). When crossmatching MP-PRECs from 3 different donors with serum taken on POD 3, extensive complement-mediated cell death occurred (p = 0.002 compared to blood group- and HLA-compatible serum) (Figure 6).

In comparison, control serum from patients with HLA class I directed against HLA antigens expressed on the MP-PRECs caused less complement-mediated cell death (p = 0.05) when compared to the patient’s serum (Figure 6). When compared to human monoclonal IgG and IgM antibodies directed against HLA-A2, a similar extent of complement-mediated cell death occurred.

Next, we performed custom endothelial flow crossmatch tests with MP-PRECs from 3 different donors, using serum from POD 3. IgM, IgG binding (recognizing human immunoglobulin subclasses IgG1, IgG2, IgG3, and IgG4), and complement activation were analyzed using flow cytometry. To better mirror the physiological situation, 20% human serum was utilized as a complement source. Crossmatching with the patient’s serum resulted in IgG and IgM binding and activated C3 to varying extents (Figure 6C). Human monoclonal IgG and IgM antibodies directed against HLA-A2 were used as positive crossmatch test controls, also showing IgG, IgM, and complement binding to varying extents. In addition, detection of non-HLA antibodies in the serum from POD 3 was measured using the multiplex assay developed by the PROCARE Consortium (20), screening for Agrin, APMAP, ARHGDIB, ARHGEF6, AT1R, ETAR, LMNB1, LPLUNC1, PECR PLA2R, PRKCZ, TUBB4B, and vimentin. Only antibodies directed against TUBB4B were found to be present in the patients’ serum (absolute mean fluorescence intensity (MFI) 1,564.5 and control ratio 2.8). However, these antibodies have not been shown to be associated with adverse outcomes after kidney transplantation.

Two years later, a retransplantation with a renal allograft from a living unrelated altruistic donor was performed. The donor and recipient both tested positive for the Epstein–Barr and cytomegalo viruses, and the donor typing was A1, A2, B7, B60, Bw6, Bw6, Cw10, Cw7, DR13, DR15,DR52, DR51, DQ6, and DQ6 resulting in a HLA A/B/DR mismatch of 1-1-0, again blood group compatible and without DSAs. Transplantation was performed after a desensitization procedure consisting of rituximab (750 mg) 9 days before transplantation and 5 sessions of plasmapheresis with exchange of 1 plasma volume per session and albumin 5% substitution. The plasmapheresis regimen was started 6 days before transplantation, followed by 0.1 g/kg IVIg (Nanogam) and 1 g/kg IVIg after the final session the day before transplantation. The last anti-GBM titer before transplantation was 1 IU/ml. Posttransplantation, plasmapheresis was continued for 3 days followed by 0.1 g/kg IVIg on the first 2 days and 1 g/kg IVIg on the 3rd day. When the patient serum obtained during desensitization treatment was analyzed for the presence of AECA utilizing MP-PRECs, a time-dependent decrease in IgG, IgM, C3, C4d, and C5b-9 was observed (Figure 7), suggesting an effective removal of non-HLA AECA. No surgical or other posttransplantation complications occurred (cold ischemia time, 152 min; warm ischemia time, 37 min), and the patient was discharged with an estimated glomerular filtration rate (eGFR) of 44 ml/min/1.73 m² (21). After 2 years, he had a stable allograft function with an eGFR of 41 ml/min/1.73 m².