CRISPR/Cas9-Engineered HLA-Deleted Glomerular Endothelial Cells as a Tool to Predict Pathogenic Non-HLA Antibodies in Kidney Transplant Recipients

Abstract

Background 

After kidney transplantation, donor-specific antibodies against human leukocyte antigen donor-specific antibodies (HLA-DSAs) drive antibody-mediated rejection (ABMR) and are associated with poor transplant outcomes. However, ABMR histology (ABMRh) is increasingly reported in kidney transplant recipients (KTRs) without HLA-DSAs, highlighting the emerging role of non-HLA antibodies (Abs).

Methods 

W e designed a non-HLA Ab detection immunoassay (NHADIA) using HLA class I and II–deficient glomerular endothelial cells (CiGEnCΔHLA) that had been previously generated through CRISPR/Cas9-induced B2M and CIITA gene disruption. Flow cytometry assessed the reactivity to non-HLA antigens of pretransplantation serum samples from 389 consecutive KTRs. The intensity of the signal observed with the NHADIA was associated with post-transplant graft histology assessed in 951 adequate biopsy specimens.

Results 

W e sequentially applied CRISPR/Cas9 to delete the B2M and CIITA genes to obtain a CiGEnCΔHLA clone. CiGEnCΔHLA cells remained indistinguishable from the parental cell line, CiGEnC, in terms of morphology and phenotype. Previous transplantation was the main determinant of the pretransplantation NHADIA result (P<0.001). Stratification of 3-month allograft biopsy specimens (n=298) according to pretransplantation NHADIA tertiles demonstrated that higher levels of non-HLA Abs positively correlated with increased glomerulitis (P=0.002), microvascular inflammation (P=0.003), and ABMRh (P=0.03). A pretransplantation NHADIA threshold of 1.87 strongly discriminated the KTRs with the highest risk of ABMRh (P=0.005, log-rank test). A multivariate Cox model confirmed that NHADIA status and HLA-DSAs were independent, yet synergistic, predictors of ABMRh.

Conclusion 

The NHADIA identifies non-HLA Abs and strongly predicts graft endothelial injury independent of HLA-DSAs.

Despite the development of potent immunosuppressive regimens, immune-mediated allograft injury remains a significant hurdle to the long-term acceptance of solid-organ transplants and, after kidney transplantation, alloimmune injuries remain major determinants of allograft failure, with subsequent reinitiation of dialysis and increased morbidity, mortality, and cost.¹ Our view of the alloimmune response is usually dichotomized into cell-mediated rejection and antibody (Ab)-mediated rejection (ABMR). More than two decades ago, the description of close associations between circulating donor-specific anti–human leukocyte antigen Abs (HLA-DSAs) and microvascular inflammation (MVI) progressively led to standardized diagnostic criteria for ABMR in the kidney allograft.² In kidney transplantation, the current ABMR classification requires the association of histologic lesions suggestive of acute tissue injury, evidence of current or recent Ab interaction with the endothelium, and serologic evidence of circulating DSAs. Many studies have described the main role of HLA-DSAs in the occurrence of ABMR. Nevertheless, the existence of ABMR-like histologic lesions in the absence of HLA-DSAs suggests alternative mechanisms, including the presence and deleterious effect of non-HLA Abs.

Further identification of the mechanisms underlying ABMR-like histologic lesions in the absence of HLA-DSAs faces several challenges. First, a dedicated clinicopathologic description of ABMR-like histologic lesions (ABMRh) in the absence of HLA-DSAs is needed to better define this particular phenotype. In fact, it has been demonstrated that 40%–60% of ABMRh cases have no circulating HLA-DSAs.^34–5 Second, a firm demonstration that ABMRh with no circulating HLA-DSAs is genuinely mediated by attacking Abs is limited given the lack of a universal assay capable of capturing the diversity of targeted non-HLA (i.e., minor) histocompatibility antigens. As a matter of fact, many targets other than HLA antigens have been previously reported.

We previously conducted a nationwide study that identified kidney transplant recipients (KTRs) with severe ABMRh in the first 3 months post-transplantation in the absence of HLA-DSAs. In that study, assessment of previously identified non-HLA Abs failed to differentiate patients with ABMRh from those who were stable. However, we demonstrated that, at the time of transplantation, patients carried preformed IgG Abs specifically targeting a conditionally immortalized human glomerular endothelial cell line (CiGEnC).⁶ These results suggested that in vitro cell-based assays are needed to assess the presence of deleterious non-HLA Abs. Given that CiGEnC express both class I and class II HLA molecules, the use of these cells for a cell-based assay was limited to patients with no circulating anti-HLA Abs. Here, we genetically modified the CiGEnC cell line to obtain CiGEnCΔHLA cells that can be used to specifically detect non-HLA Abs, even in patients with circulating anti-HLA Abs. Through validation of this cell-based assay in a large, unselected cohort of KTRs, we demonstrated its clinical utility for improving the pretransplant evaluation of immunologic risk and for designing mechanism-driven therapeutic approaches targeting non-HLA Abs.

Methods

Patients

We retrospectively reviewed all consecutive adult patients who received a kidney transplant at Necker Hospital (Paris, France) between January 2012 and June 2017, with available serum collected immediately before transplantation (day 0; D0) and available follow-up at our institution. Exclusion criteria included pretransplantation desensitization with plasma exchanges for ABO or HLA incompatibility (n=31), active infection with HIV, infection with hepatitis C or B virus (n=40), and primary nonfunction of the transplanted kidney due to surgical issues or arterial thrombosis (n=6). Ultimately, 389 KTRs were included. All participating patients provided written informed consent.

The transplantation allocation system followed the rules of the French national agency for organ procurement (Agence de la Biomédecine). Negative IgG T cell and B cell complement–dependent cytotoxicity crossmatching was required for all of the KTRs. Clinical and biologic data for the donors and recipients were retrospectively obtained from the registry Données Informatiques Validées en Transplantation (http://www.divat.fr/) and the medical records of the patients.

Histologic Assessment

Histologic assessments of all of the biopsy specimens performed during follow-up for each patient were collected. Biopsy specimens were fixed in formalin, acetic acid, and alcohol and embedded in paraffin. Tissue sections were stained with hematoxylin and eosin, Masson trichrome, Periodic acid–Schiff reagent, and Jones stain for light microscopy evaluation. C4d immunohistochemical staining was systematically performed (with a rabbit anti-human C4d monoclonal Ab [mAb]; 1:200 dilution; Clinisciences, Nanterre, France).

Protocol biopsies were performed at 3 and 12 months post-transplantation, and indication biopsies were performed for clinical indications. Biopsy specimens were classified using the Banff 2015 and 2017 updates of the Banff classification system.^7,8 The biopsy specimens were declared inadequate if the number of glomeruli was strictly below eight. The biopsy samples were graded from zero to three according to the Banff histologic parameters for ptc, g, t, i, ci, ct, cv, ah, and allograft glomerulopathy (cg).⁹

For the purpose of the study, the term “ABMRh” was used for biopsy specimens that fulfilled the first two (histologic) Banff 2015 and 2017 criteria for ABMR by combining Banff scores for g, ptc, arteritis, thrombotic microangiopathy, and C4d deposition.³

Detection of HLA-DSAs

All patients were tested for the presence of HLA-DSAs immediately before transplantation (D0), at 3 months, at 12 months, and for clinical indications. The presence of circulating HLA-DSAs against HLA-A, HLA-B, HLA-Cw, HLA-DR, HLA-DQ, and HLA-DP was determined using single-antigen flow bead assays (One Lambda Inc., Canoga Park, CA) on a Luminex platform. Beads with a normalized mean fluorescence intensity (MFI) >1000 arbitrary units were considered positive.

Cell Culture

HRECs

Normal HRECs were harvested from human nephrectomy specimens removed for renal cell carcinoma and isolated according to previously published methods, with minor modifications.^10,11 Fragments of the nonmalignant renal cortex were minced and digested with collagenase IV (250 IU/ml; Roche) for 3 h at 37°C. Cells were centrifuged, and the pellets were washed three times with PBS. Cells were then cultured in DMEM containing 10 μg/ml human apotransferrin, 500 ng/ml hydrocortisone (Sigma-Aldrich), 10 ng/ml EGF (Sigma-Aldrich), 6.5 ng/ml triiodothyronine (Sigma-Aldrich), 1% FCS, 25 IU/ml penicillin, and 25 μg/ml streptomycin (Thermo Fisher, Courtaboeuf, France), and supplemented with insulin-transferrin-selenium (Thermo Fisher). Cells were incubated at 37°C in 5% carbon dioxide and 95% air. The characterization of our cellular model has been published previously,¹⁰ confirming the proximal descent of the vast majority of the cultured tubular epithelial cells. Experiments were not performed with cells beyond the fourth passage, because it has been shown that no phenotypic changes occur up to this passage number.¹²

HK-2 Cells

HK-2 cells were obtained and cultured as previously described.¹³

CiGEnC Cells

CiGEnC cells were kindly provided by S.S.¹⁴ and were cultured in endothelial growth medium 2– microvascular (EGM2-MV; Promocell, Heideberg, Germany) in culture flasks previously coated with 0.1% gelatin (Sigma-Aldrich). CiGEnC cells proliferated at 33°C, with growth arrest and differentiation occurring after culture at 37°C for 7 days.¹⁴

CRISPR/Cas9 Genome Editing

Guide RNA (gRNA) sites in B2M and CIITA exonic loci were identified using the online optimized design software¹⁵ at http://crispor.tefor.net/. The highest scoring gRNAs, which had no off-target sequences with perfect matches in the human genome, and the nearest coding off-target exonic sites containing at least three mismatched nucleotides were selected and purchased from Thermo Fisher (TrueGuide 2-piece modified Synthetic gRNA; Thermo Fisher). The B2M crispr RNA (crRNA) targeting sequences included GAGTAGCGCGAGCACAGCTA (B2M exon 1) and AGTCACATGGTTCACACGGC (B2M exon 2). The CIITA crRNAs targeting sequences included CATCGCTGTTAAGAAGCTCC (CIITA exon 2) and GATATTGGCATAAGCCTCCC (CIITA exon 3). crRNAs and transactivating crRNA were annealed in TE buffer in a Verity thermocycler (Thermo Fisher), according to the manufacturer’s instructions, to obtain complete functional gRNAs.

Transfection

A Cas9 nuclease/gRNA/transfection reagent complex was prepared according to the manufacturer’s instructions. Briefly, a mixture containing Cas9 nuclease (TrueCut V2; Thermo Fisher), transactivating crRNAs, Opti-MEM medium, and Cas9 Plus reagent was combined with Lipofectamine CRISPRMAX Reagent (Thermo Fisher). This complex was plated on six-well plates, and cells were added and incubated for 2 days. The cells were washed and cultured for five more days.

Single-Cell Isolation of CRISPR/Cas9-Modified Cells

After transfection, CiGEnC cells were stimulated with IFN-γ (100 IU/ml; Miltenyi Biotec, Paris, France) for 2 days to upregulate HLA-I and HLA-II. Cells were harvested with trypsin and subsequently stained with Fixable Viability Dye eFluor 660 (Thermo Fisher) and directly conjugated with VioBlue anti–HLA-ABC (Miltenyi Biotec) and BV605 anti–HLA-DR (Ozyme, Montigny-le-Bretonneux, France) mAbs. B2M and CIITA loss of function was identified by live cells that did not show increased cell-surface expression of HLA-ABC or HLA-DR, respectively, with the positive threshold defined by unmodified CiGEnC cells stained with the same mAbs. These gates were then used to collect CiGEnCΔHLA cells using a 100-μm low-pressure nozzle on a BD FACSAria II (BD Biosciences, Le Pont de Claix, France) and then to deposit single cells into flat-bottomed, 96-well cell culture plates containing EGM-2/50% FBS medium. After 24 hours, the cells were refed with fresh EGM-2/5% FBS medium that was changed every other day. Colonies were scored after 14 days and serially expanded into larger vessel sizes. To analyze phenotypic stability, CiGEnCΔHLA and unmodified CiGEnC cells were separately expanded in EGM-2/5% FBS for 2 weeks. These cells were then challenged with TNF-α (100 IU/ml, Miltenyi Biotec) and IFN-γ (100 IU/ml, Miltenyi Biotec) and harvested at 24 hours for quantitative RT-PCR (RT-qPCR) analysis and at 48 hours for both FACS analysis and immunofluorescence staining.

PCR and Sanger Sequencing

Genomic DNA was isolated from clonally expanded CiGEnC cells using the QIAamp DNA Mini Kit (Qiagen, Courtaboeuf, France) according to the manufacturer’s protocol. Segments of 100- to 150-bp that contained the B2M and CIITA gRNA target sites were amplified by PCR using AmpliTaq Gold 360 DNA Polymerase (Thermo Fisher), using the primers B2M Exon 1 Forward (ATATAAGTGGAGGCGTCGCG), B2M Exon 1 Reverse (TGGAGAGACTCACGCTGGAT), B2M Exon 2 Forward (TGTCTTTCAGCAAGGACTGGT), B2M Exon 2 Reverse (ACCCCACTTAACTATCTTGGGC), CIITA Exon 2 Forward (CTGCCTCTTTCCAACACCCT), CIITA Exon 2 Reverse (CTTCTCCAGCCAGGTCCATC), CIITA Exon 3 Forward (TTTCAGCAGGCTGTTGTGTG), and CIITA Exon 3 Reverse (GCAGCAAAGAACTCTTGCCC). The PCR amplicons were then purified and submitted for Sanger sequencing using an ABI 3730XL DNA sequencer (Eurofins Genomics, Ebersberg, Germany). Unmodified CiGEnC cells were used as a control for comparison.

The highest scoring off-target site of B2M exon 2 was TGAGAGTACCAGGTGTGACG (HDHD1P2, four mismatches in 2:5:12:18). To determine if this site was mutated, the following sequencing primers were used: HDHD1P2_F (TCGTCGGCAGCGTCGTGCAGTCTGGGATTTGGGA) and HDHD1P2_R (GAGGGCCGTCTCGTGGGCTCGGTATGAGTGAGAGG).

To assess off-target cleavage of the highest scoring off-target coding site of CIITA exon 2 (JMJD4, CATCACTGCTAGGAAGCTTCAGG, four mismatches in 5:9:12:19), the following sequencing primers were used: JMJD4_F (TCGTCGGCAGCGTCATCAAAGGCTGCCTGTTCGA) and JMJD4_R (GTCTCGTGGGCTCGGTGCTCGGGCATCAACTTTGA). To assess off-target cleavage of the highest scoring off-target coding site of CIITA exon 3 (OSTF1, GATATCTGCATAACCCTTCCAGG, four mismatches in 6:7:14:17), the following sequencing primers were used: OSTF1_F (TCGTCGGCAGCGTCGGGAGATACAGCTTTGCATGC) and OSTF1_R (AAGACCAGTCTCGTGGGCTCGGTTCAGGGCAAGCA). Analyses of sequencing results were performed using Serial Cloner 2.6.1 software (http://serialbasics.free.fr/Serial_Cloner.html). Allelic analysis was performed using CRISPR-ID 1.1a¹⁶ (http://crispid.gbiomed.kuleuven.be/), and the efficacy of gene editing was assessed by Tracking of Indels by Decomposition¹⁷ web applications (https://tide.deskgen.com).

FACS Analysis

After 1 week of differentiation induced by culture at 37°C, cells were stained with VioBlue-conjugated anti–HLA-ABC, PE-conjugated anti-VEGFR2, PE-Vio770–conjugated anti-ICAM2, APC-conjugated anti–VE-cadherin, APC-Vio770–conjugated anti-Tie2 (all from Miltenyi Biotec), and BV605-conjugated anti–HLA-DR (Ozyme) Abs and analyzed on an LSR Fortessa flow cytometer (BD Biosciences, San Jose, CA), with postacquisition analysis using Kaluza software (Kaluza 2.1; Beckman Coulter, Villepinte, France). To induce HLA antigen expression, cells were exposed to IFN-γ and TNF-α (100 U/ml; Miltenyi Biotec) for 48 hours before staining. The relative fluorescence intensity was calculated by subtracting the MFI of the corresponding isotype control.

qRT-PCR

RNA was isolated using the RNeasy Mini Kit (Qiagen) and used to generate cDNA using a mixture containing an RNAse inhibitor, dNTP mix, Random Hexamer, magnesium chloride solution and MultiScribe Reverse transcription (all from Thermo Fisher). qPCR reactions were assembled with TaqMan 2x Fast Universal PCR Master Mix (Thermo Fisher) and predeveloped TaqMan gene expression probes and analyzed on a ViiA 7 Real-Time System using QuantStudio Real-Time PCR software (Thermo Fisher). The following probes used in this study were purchased from Thermo Fisher: CIITA (Hs00172106_m1), KDR (Hs00911700_m1), ICAM2 (Hs00609563_m1), CDH5 (Hs00901465_m1), TIE2 (Hs00945150_m1), B2M (Hs00187842_m1), and HLA-DR (Hs00219575_m1). For GAPDH, the following primers and probes were used: sense, 5′-CCACATCGCTCAGACACCAT-3′; antisense, 5′-TGACCAGGCGCCCAATA-3′; and probe, 5′-FAM-AGTCAACGGATTTGGTC-MGB-3′. For CXCL10, the following primers and probes were used: sense, 5′-TGTCCACGTGTTGAGATCATTG-3′; antisense, 5′-GGCCTTCGATTCTGGATTCA-3′; and probe, 5′-FAM-TACAATGAAAAAGAAGGGTGAGAA-MGB-3′. Gene expression levels were normalized to those of GAPDH. When indicated, cells were exposed to IFN-γ and TNF-α (100 U/ml) for 24 hours before RNA extraction.

Immunofluorescence

Cells grown to confluence on gelatin-coated glass coverslips were fixed in 4% formaldehyde and permeabilized in 0.1% Triton X-100 containing 3% BSA (Thermo Fisher). The cells were then incubated for 12 hours at 4°C with VioBright-515–conjugated anti-PECAM1 or APC-conjugated anti–VE-cadherin (Miltenyi Biotec) primary Abs or unconjugated anti–HLA-A/-B/-C, anti–HLA-DP/DR/DQ (Ozyme), or anti-ICAM2 (Abcam) primary Abs. Unconjugated primary Ab binding was detected using an Alexa Fluor 647–conjugated anti-mouse IgG secondary Ab (Ozyme). Negative controls were either isotypes for the fluorophore-conjugated primary Abs or the absence of the primary Ab for secondary revelation. The cells were then stained with 4′,6-diamidino-2-phenylindole, and coverslips were mounted using Fluoromount (Sigma-Aldrich); the cells were examined using a Zeiss confocal microscope (Zeiss Confocal LSM 700). Zen900 software was used to generate images, and ImageJ (Java) was used to analyze the images.

Phase-Contrast Microscopy

Unmodified CiGEnC or CiGEnCΔHLA cells at various passages were seeded in flasks at a subconfluent density and placed at either 33°C or 37°C, and morphology was examined by phase-contrast microscopy.

Cell Proliferation Assay

Unmodified CiGEnC or CiGEnCΔHLA cells were seeded at 50,000 cells per well in 12-well plates, and real-time evaluation of cell confluence was performed using the IncuCyte Live Cell Imaging System (Essen BioScience, Royston, United Kingdom). Images were acquired every 2 hours for 60 hours from nine separate regions per well using a 10× objective, and then analyzed using IncuCyte basic software.

HLA-A2 Chimeric Antigen Receptor T-Cell Generation and an xCELLigence Cytotoxicity Assay

Healthy donor PBMCs were obtained from the Etablissement Français du Sang. PBMCs were typed on the basis of the expression or absence of HLA-A2/A28 molecules, as assessed by anti–HLA-A2/A28 Ab (OneLambda) staining evaluated by FACS analysis. CD8+ T cells were sorted using a FACSAria II (BD Biosciences), transduced 2 days after activation with an HLA-A2–specific chimeric antigen receptor (CAR) construct at an MOI of 40, and incubated for 18 hour for transduction, as previously described.^18(preprint) On day 5 post-transduction, CD8+ EGFRt+ cells were sorted using a FACSAria II. CD8+ CAR-T cells were cultured in X-VIVO 20 medium containing 10% human serum AB (Biowest) supplemented with IL-2 (100 UI/ml). For the cytotoxic assay, 1 × 10⁴ unmodified CiGEnC or CiGEnCΔHLA cells were seeded in E-plate (ACEA Biosciences) wells. After 15 hours, 2 × 10⁴ CD8+ anti–HLA-A2 CAR-T cells were added to the culture. To evaluate the viability of unmodified CiGEnC or CiGEnCΔHLA cells, electrical impedance measurements were taken with an xCELLigence RTCA MP instrument (ACEA Biosciences) every 15 minutes for 10 hours. The cell indices were normalized to the reference value (measured just before adding CAR-T cells to the culture). The normalized cell index of experimental wells was normalized to the cell index of control wells containing only the corresponding endothelial cell line.

Non-HLA Ab Detection Immunoassay

Serum samples collected immediately before transplantation were tested with the non-HLA Ab detection immunoassay (NHADIA). After washing with PBS, differentiated CiGEnCΔHLA cells were trypsinized (TrypLE Express; Thermo Fisher) and washed before incubation with a Fixable Viability Dye for 20 minutes at 4°C. Then, the CiGEnCΔHLA cells were incubated with patient sera diluted 1:2 in PBS containing 0.05% BSA and 2 mM EDTA for 30 minutes. For the negative control, cells were incubated with PBS only. After two more washes, the cells were incubated with an Alexa Fluor 488–conjugated anti-human IgG Ab (AffiniPure F[ab']₂ Fragment Donkey Anti-Human IgG [H+L]; Interchim) for 20 minutes. Fluorescence was measured by flow cytometry (FACSCanto II or LSR Fortessa X-20; BD Biosciences), and geometric MFIs (Geo MFIs) were calculated using Kaluza software version 2.1. The results were calculated as the ratio of the Geo MFI_sample to the Geo MFI_{negative control}.

Statistical Analyses

Continuous variables are described as the mean±SD or median (interquartile range; IQR). Frequencies of categoric variables are presented as numbers and percentages. We compared continuous variables using the Mann–Whitney test or t test, and the proportion of categoric variables using the Fisher exact test or a chi-squared test when appropriate. P values ≤0.05 were regarded as statistically significant.

Univariate and multivariate regression analyses were performed to identify pretransplant determinants of the NHADIA. To identify the positivity threshold of the NHADIA that would best predict ABMRh independently of HLA-DSAs, time to ABMRh diagnosis was assessed in a Cox regression analysis including various thresholds of the NHADIA and HLA-DSA status at the time of transplantation. Cox proportional hazard analysis was used to associate NHADIA results with time to ABMRh. The Kaplan–Meier method was used to estimate the cumulative incidence of ABMRh, with a timescale of years post-transplantation. With a median follow-up time after transplantation of 4.6 years, time was censored at 4 years after transplantation in all of the time-to-event analyses.

Alluvial plots and dendrograms were built using the “networkD3” and “ClustOfVar” R packages, respectively. Analyses were performed with R software (R version 3.6.3 and RStudio version 1.2.5033; R Development Core Team) and GraphPad Prism software version 7.0a (GraphPad Software, San Diego, CA).

Results

Genetic Ablation of B2M and CIITA in CiGEnC

We previously observed that KTRs suspected of ABMR due to non-HLA Abs had circulating IgG specifically targeting microvascular CiGEnC compared with macrovascular endothelial cells.⁶ Nevertheless, given that CiGEnC express both class I and class II HLA molecules, the use of these cells for a cell-based assay would be limited to patients with no anti-HLA Abs in their serum. To circumvent this limitation, we aimed to suppress HLA molecule expression on the cell surface of CiGEnC to obtain a CiGEnCΔHLA cell line.

Because B2M is essential for the assembly and expression of the HLA-I complex,¹⁹ and CIITA is a crucial transactivator of HLA-II,²⁰B2M and CIITA double disruption was sequentially developed to generate CiGEnCΔHLA cells (Figure 1). A detailed description of the cell engineering is provided in Supplemental Appendix 1 and Supplemental Figure 1.

Figure 1.: CiGEnC Δ HLA cells were succesfully generated by CRISPR/Cas9 ablation of B2M and CIITA genes. (A) Schematic view of B2M disruption in CiGEnC cells. Lipofectamine reagent was used to codeliver an active Cas9 protein and two different synthetic gRNAs targeting exonic regions (exon 1 and exon 2) shared by all known splice variants of B2M into cells. (B) FACS histograms showing HLA-I expression before (left) and after (middle) transfection and after cell sorting (right). After 5 days, the cells were stimulated with IFN-γ before FACS sorting. Although >99% of unmodified CiGEnC cells upregulated HLA-I upon IFN-γ stimulation, the delivery of B2M-specific gRNA resulted in <43% HLA-I–positive cells. (C) Schematic view of CIITA disruption in CiGEnCΔB2M cells. CiGEnCΔB2M cells were transfected with two different gRNAs targeting exon 2 and exon 3 CIITA loci and an active Cas9 protein. After 5 days, the cells were stimulated with IFN-γ before FACS sorting. Delivery of CIITA-specific gRNAs resulted in <3% HLA-DR–positive cells. Nevertheless, we used single-cell FACS sorting of viable cells before clonal expansion of CiGEnCΔHLA cells. (A and C) were created with Biorender.com

Characterization of HLA Antigen Loss of Expression

After 7 days of differentiation at 37°C followed by cytokine stimulation, RT-qPCR analysis of unmodified CiGEnC revealed, as expected, basal expression of B2M that increased after cytokine stimulation, and inducible expression of CIITA and HLA-DR after cytokine stimulation. In contrast, analysis of the CiGEnCΔHLA clone revealed a >99.9% reduction in B2M, 95% reduction in CIITA, and undetectable HLA-DR mRNA expression, but equivalent mRNA levels of CXCL10, another IFN-γ–inducible gene (Figure 2A).

Figure 2.: A complete loss of HLA antigens expression was achieved in CiGEnC Δ HLA cells. (A) RT-qPCR analysis of B2M, CIITA, HLA-DR, and CXCL10 in unmodified CiGEnC and CiGEnCΔHLA cells with and without 24 hours of cytokine stimulation. The results are shown as the relative expression of the genes normalized to GAPDH expression. (B) Representative dot plots of FACS analysis of HLA-ABC and HLA-DR expression in unmodified CiGEnC (top panels) and CiGEnCΔHLA (lower panels) cells with (right panels) and without (left panels) 48 hours of cytokine stimulation. Without cytokine stimulation, >94% of unmodified CiGEnC cells expressed HLA-ABC but exhibited very limited HLA-DR expression. When both IFN-γ and TNF-α were added to the culture for 48 hours, >50% of unmodified CiGEnC cells expressed HLA-DR, as expected. In contrast, the expression of HLA-ABC and HLA-DR was completely abrogated, with <0.5% positive cells even after cytokine stimulation. (C) FACS analysis of HLA-ABC (top) and HLA-DR (lower) expression in unmodified CiGEnC and CiGEnCΔHLA cells with and without 48 hours of cytokine stimulation. Expression is presented as the relative fluorescence intensity (RFI) calculated by subtracting the MFI of the corresponding isotype control. HLA-ABC and HLA-DR antigens were completely depleted from the CiGEnCΔHLA cell surface compared with the unmodified CiGEnC cell surface. (D) Immunofluorescence analysis of HLA-ABC (yellow staining) and HLA-DR, HLA-DP, and HLA-DQ (red staining) expression in unmodified CiGEnC and CiGEnCΔHLA cells with and without 48 hours of cytokine stimulation. 4′,6-Diamidino-2-phenylindole (blue staining) was used as a nuclear counterstain. Confocal analysis showed the ablation of not only HLA-ABC and HLA-DR but also HLA-DP and HLA-DQ in CiGEnCΔHLA cells after cytokine stimulation, suggesting the complete depletion of all class I and II HLA antigens. (E) Representative dot plots of FACS analysis of HLA-ABC and HLA-A2 expression in unmodified CiGEnC and CiGEnCΔHLA cells. (F and G) Unmodified CiGEnC or CiGEnCΔHLA cells were cocultured with cytotoxic anti–HLA-A2 CAR-T cells. (F) Created with BioRender.com (G) The normalized cell index (mean±SEM) from three independent experiments is shown.

Thus, we confirmed ablation of both B2M and CIITA by monitoring the absence of HLA antigens by FACS (Figure 2, B and C) and immunofluorescence (Figure 2D) analyses of unmodified CiGEnC and CiGEnCΔHLA cells after cytokine stimulation.

Finally, we used CAR technology to redirect the antigen specificity of CD8+ T cells toward the HLA-A2 antigen expressed by CiGEnC cells (Figure 2, E and F). We cocultured HLA-A2–specific CAR-T cells with CiGEnC or CiGEnCΔHLA cells and observed high cytotoxicity of the CD8+ CAR-T cells toward unmodified CiGEnC, but not CiGEnCΔHLA, cells (Figure 2G).

CRISPR/Cas9 Editing Does Not Impair the Endothelial Phenotype in CiGEnCΔHLA Cells

CiGEnC cells have been previously characterized and are very similar to microvascular glomerular endothelial cells in terms of phenotype after 1 week of differentiation at 37°C.¹⁴

We assessed the consequences of gene editing on the endothelial phenotype of the CiGEnCΔHLA cells. RT-qPCR (Figure 3A), FACS (Figure 3, B and C), and immunofluorescence (Figure 3D) analysis of endothelial markers and optic microscopy (Figure 3, E and F) demonstrated that CiGEnCΔHLA cells remained undistinguishable from the parental cell line in terms of morphology, phenotype, and proliferative profile.

Figure 3.: CRISP/Cas9 editing does not impair the endothelial phenotype of CiGEnC Δ HLA cells. (A) RT-qPCR analysis of vWF, KDR, CDH5, and TEK after cell differentiation of unmodified CiGEnC, CiGEnCΔHLA, and HK2 cells. The results are shown as the relative expression of the genes normalized to GAPDH expression (n=3 independent experiments). Compared with human renal epithelial cells (HK2 cell line), CiGEnCΔHLA and parental CiGEnC cells showed high expression levels of glomerular endothelial genes. (B) Representative dot plots of FACS analysis of VE cadherin and ICAM2 expression (top panels) and Tie2 and VEGFR2 expression (bottom panels) in HRECs (left panels), unmodified CiGEnC cells (middle panels), and CiGEnCΔHLA cells (right panels). (C) FACS analysis of VEGFR2, Tie2, VE cadherin, and ICAM2 in HRECs and unmodified CiGEnC and CiGEnCΔHLA cells. Expression is presented as the relative fluorescence intensity (RFI) calculated by subtracting the MFI of the corresponding isotype control. Both unmodified CiGEnC cells and CiGEnCΔHLA cells exhibited similar levels of VE cadherin, ICAM2, Tie2, and VEGFR2 expression compared with HRECs. (D) Immunofluorescence analysis of PECAM1 (green staining), VE cadherin (purple staining), and ICAM2 (white staining) expression in unmodified CiGEnC (top panels) and CiGEnCΔHLA (bottom panels) cells confirmed their expression on the CiGEnCΔHLA cell surface. 4′,6-Diamidino-2-phenylindole (blue staining) was used as a nuclear counterstain. (E) CiGEnCΔHLA cells maintained morphologic features of unmodified CiGEnC cells both at the permissive temperature (33°C; top panels) and at a nonpermissive temperature (37°C; bottom panels). At both 33°C and 37°C, CiGEnC retained features of early-passage primary glomerular endothelial cells in culture, including small size, homogeneity, and the formation of “cobblestone” monolayers. Phase-contrast microscopy showed that CiGEnCΔHLA cells were viable in culture and retained the morphologic features of unmodified CiGEnC cells. (F) Cell proliferation normalized by the confluence measured at the beguining of the measurment (h0) at 33°C was analyzed for 55 hours using an IncuCyte system. The mean±SEM of three independent experiments is shown. Serially passaged CiGEnCΔHLA cells maintained the same proliferative profile as unmodified CiGEnC cells at 33°C with loss of proliferation at 37°C, as expected.

Patient Baseline Characteristics

According to our inclusion criteria, 389 serum samples collected immediately before transplantation (D0) from 389 unselected patients consecutively transplanted between January 2012 and June 2017 were available (Figure 4A). Patient and donor characteristics at the time of transplantation are summarized in Table 1. The mean±SD recipient age was 53.7±14.6 years, 64% were male, 21.6% received a kidney from a living donor, 6.7% had a previous kidney transplantation, and 79 recipients (20.3%) had preformed HLA-DSAs.

Figure 4.: The presence of non-HLA Abs before transplantation is associated with retransplantation status. (A) Design of the observational cohort study. (B) Schematic view of the NHADIA process created with BioRender.com (left) and histograms showing the NHADIA results of a negative control and a patient with serum containing non-HLA Abs. (C) Univariate (top panel) and multivariate (bottom panel) linear regression analyses of pretransplant determinants of NHADIA results measured at the time of transplantation. (D) Distribution of the normalized NHADIA results obtained with pretransplant serum samples from patients awaiting a first transplantation (top panel) or retransplantation (bottom panel). Tx, transplant.

Table 1. - Patient and transplant characteristics

Population	All (n=389)	NHADIA ≤1.87 (n=332)	NHADIA >1.87 (n=57)	P Value
Recipient characteristics
Male, n (%)	249 (64)	213 (64.2)	36 (63.2)	0.75
Age at transplantation (yr), mean±SD	53.7±14.6	54.0±14.5	52.0±15.0	0.88
Cause of ESKD				0.94
GN, n (%)	72 (18.5)	60 (18.1)	12 (21.1)
Diabetes, n (%)	42 (10.8)	35 (10.5)	7 (12.3)
Cystic/hereditary/congenital, n (%)	114 (29.3)	97 (29.2)	17 (29.8)
Hypertension, n (%)	41 (10.5)	36 (10.8)	5 (8.8)
Interstitial nephritis, n (%)	33 (8.5)	29 (8.7)	4 (7.0)
Miscellaneous conditions, n (%)	14 (3.6)	11 (3.3)	3 (5.3)
Etiology uncertain, n (%)	73 (18.8)	64 (19.3)	9 (15.8)
Previous kidney transplantation, n (%)	26 (6.7)	17 (5.1)	9 (15.8)	0.007
Pregnancy, n (%) a	87 (22.8)	76 (23.2)	11 (20.4)	0.73
Blood transfusion, n (%) a	120 (31.1)	100 (30.3)	20 (35.7)	0.44
Time since dialysis onset (mo), mean±SD	39.7±42	39.0±40.7	44.4±48.9	0.50
Transplant variables
Preemptive kidney transplantation, n (%)	66 (17.0)	56 (16.9)	10 (17.5)	0.85
Donor age (yr), mean±SD	58.1±17.1	58.3±17.0	56.9±17.9	0.60
Living donors, n (%)	84 (21.6)	72 (21.7)	12 (21.1)	>0.99
DDs, n (%)	305 (78.4)	260 (78.3)	45 (78.9)	>0.99
SCD, n (%)	123 (31.6)	104 (40.0)	19 (42.2)	0.87
ECD, n (%)	182 (46.8)	156 (60.0)	26 (57.8)	0.87
Cold ischemia time for DDs (h), mean±SD	20.1±7.4	19.9±7.4	20.7±7.7	0.56
Delayed graft function, n (%) a	71 (18.6)	63 (19.4)	8 (14.0)	0.46
Preformed DSAs, n (%)	79 (20.3)	65 (19.6)	14 (24.6)	0.38
HLA-A/B/DR mismatches, mean±SD	3.3±1.5	3.4±1.5	3.2±1.4	0.39
Induction therapy
Anti-thymocyte globulins, n (%)	149 (38.3)	118 (35.5)	31 (54.4)	0.008
Basiliximab, n (%)	227 (58.4)	204 (61.4)	23 (40.4)	0.004
None, n (%)	13 (3.3)	10 (3.0)	3 (5.3)	0.42

eGFR calculated by the Modification of Diet in Renal Disease formula. DDs, deceased donors; SCD, standard criteria donors; ECD, expanded criteria donors.

^aData unavailable for seven patients.

Pretransplant NHADIA Results Are Associated with Retransplantation Status

The cell-based NHADIA was performed using the 389 pretransplant serum samples with CiGEnCΔHLA cells as the targets, as described in the Methods section. Figure 4B shows a representative result compared with that of a negative control.

The median (IQR) value of the NHADIA was 1.26 (1.14–1.56). Linear regression analysis was performed to identify pretransplant determinants of the NHADIA value (Figure 4C), and this identified previous transplantation as the main determinant of the NHADIA result. Indeed, the median (IQR) NHADIA value was 1.24 (1.13–1.52) among the 363 patients awaiting their first transplantation (Figure 4D, top panel), and 1.62 (1.41–2.03) among the 26 patients awaiting a retransplantation (P<0.001; Figure 4D, lower panel). Of note, other pretransplant sensitizing events, such as blood transfusion and pregnancy, were not identified as independent determinants of NHADIA value in the multivariate analysis.

Pretransplant NHADIA Results Are Associated with MVI

A total of 951 adequate kidney allograft biopsies were performed during follow-up, including 510 biopsies performed at 3 months (n=298) or 12 months (n=212) post-transplantation and 441 additional indication biopsies performed at a median (IQR) time of 7.1 (1.3–14.7) months after transplantation. The biopsy specimen characteristics are summarized in Table 2.

Table 2. - Description of kidney biopsy specimens performed at 3 months, 12 months, or any other time after transplantation

Variables	All Biopsy Samples (n=951)	3 Months (n=298)	12 Months (n=212)	Others (n=441)
Indication of biopsy
Screening biopsy, n (%)	583 (64)	264 (88.9)	202 (95.3)	117 (26.8)
AKI, n (%)	278 (29.4)	30 (10.1)	6 (2.8)	242 (55.4)
BKV viremia, n (%)	23 (2.4)	3 (1.0)	0 (0)	20 (4.6)
De novo DSAs, n (%)	6 (0.6)	0 (0)	0 (0)	6 (1.4)
Control after acute rejection, n (%)	52 (5.5)	0 (0)	4 (1.9)	48 (11.0)
Other, n (%)	4 (0.4)	0 (0)	0 (0)	4 (0.9)
Time from transplant to biopsy (mo), median (IQR)	5.0 (3–12.4)	3.1 (2.9–3.2)	12.1 (11.8–12.4)	7 (1.3–14.7)
DSAs at the time of biopsy, n (%)	125 (13.1)	32 (10.7)	22 (10.4)	71 (16.1)
Pathologic lesions
Glomerulitis (g)
Score >0, n (%)	255 (27.2)	65 (22.0)	52 (24.9)	138 (31.8)
Score, mean±SD	0.4±0.7	0.3±0.6	0.3±0.6	0.5±0.8
Peritubular capilaritis
Score >0, n (%)	165 (17.5)	39 (13.1)	18 (8.6)	108 (24.7)
Score, mean±SD	0.3±0.7	0.2±0.6	0.1±0.5	0.5±0.9
MVI
Score >0, n (%)	303 (32.3)	81 (27.4)	57 (27.3)	165 (38.1)
Score, mean±SD	0.7±1.3	0.5±1.0	0.5±1.0	0.9±1.5
Tubulitis
Score >0, n (%)	173 (18.4)	36 (12.3)	26 (12.5)	111 (25.3)
Score, mean±SD	0.4±0.9	0.2±0.7	0.2±0.7	0.5±1.0
Interstitial infiltrate
Score >0, n (%)	68 (7.3)	19 (6.5)	4 (1.9)	45 (10.3)
Score, mean±SD	0.1±0.5	0.1±0.3	0.0±0.3	0.2±0.6
Vasculitis
Score >0, n (%)	33 (3.9)	8 (2.9)	0 (0)	26 (6.4)
Score, mean±SD	0.1±0.4	0.0±0.3	0.0±0.0	0.1±0.5
C4d
Score >0, n (%)	214 (23.2)	54 (18.8)	41 (20.1)	119 (27.6)
Score, mean±SD	0.4±0.8	0.3±0.6	0.3±0.7	0.5±0.9
Chronic vascular changes
Score >0, n (%)	639 (76.4)	186 (69.9)	137 (72.9)	316 (82.7)
Score, mean±SD	1.5±1.1	1.3±1.1	1.4±1.1	1.6±1.0
Arteriolar hyalinosis
Score >0, n (%)	699 (75.2)	206 (71.3)	155 (74.9)	338 (78.1)
Score, mean±SD	1.2±1.0	1.1±0.9	1.2±0.9	1.4±1.0
Allograft glomerulopathy
Score >0, n (%)	30 (3.2)	5 (1.7)	7 (3.4)	18 (4.1)
Score, mean±SD	0.0±0.3	0.0±0.1	0.0±0.2	0.1±0.4
Interstitial fibrosis
Score >0, n (%)	559 (60.3)	135 (47.0)	125 (59.8)	299 (69.4)
Score, mean±SD	1.0±1.0	0.7±0.9	1.0±1.0	1.0±1.0
Tubular atrophy
Score >0, n (%)	546 (59.1)	130 (45.6)	124 (59.6)	292 (67.7)
Score, mean±SD	1.0±1.0	0.7±0.9	1.0±1.0	1.1±1.0
Acute rejection primary diagnosis a
ABMR, n (%)	80 (8.4)	17 (5.7)	10 (4.7)	53 (12.0)
ABMRh, n (%)	183 (19.2)	41 (13.7)	30 (14.2)	112 (25.4)
TCMR, n (%)	29 (3.1)	3 (1)	0 (0)	26 (5.9)

TCMR, T cell–mediated rejection.

^aAccording to Banff 2019 criteria.

We first addressed the association of NHADIA values with allograft histology at 3 months post-transplantation. Unsupervised clustering analysis revealed the NHADIA results preferentially clustered with histologic features of ABMR (Figure 5A). Stratification of biopsy specimens according to NHADIA tertiles (Figure 5B, left) demonstrated that higher levels of non-HLA Abs were positively correlated with an increased presence of glomerulitis (P=0.002), MVI (P=0.003), and ABMRh (P=0.03). In addition, increasing severity of glomerulitis (P=0.008) and MVI (P=0.01) were associated with increased NHADIA signals (Figure 5B, right). Furthermore, biopsy specimens showing ABMRh at 3 months were associated with elevated NHADIA results (P=0.004). Finally, multivariate regression analysis demonstrated that increased NHADIA values were associated with ABMRh at 3 months, independent of HLA-DSAs (Figure 5C). Similar patterns were observed at 12 months post-transplantation (Supplemental Figure 2). Of note, acute and chronic tubulointerstitial and vascular Banff elementary lesions scores were similar among the three groups (data not shown).

Figure 5.: The pretransplant NHADIA result is associated with ABMRh lesions at 3 months post-transplantation. (A) Dendrogram representations of unsupervised hierarchic clustering analysis of NHADIA quartiles and Banff elementary lesions observed at 3 months after transplantation. The vertical axis of the dendrogram represents the distance or dissimilarity between clusters. (B) Percentages of 3-month allograft biopsy specimens with glomerulitis, peritubular capillaritis, C4d staining, MVI, or ABMRh lesions according to NHADIA tertiles (left panels; P values by chi-squared tests) and the mean±SEM values of the NHADIA results according to the corresponding histologic features (right panels; P values by the Kruskal–Wallis test). (C) Multivariate logistic regression analysis of pretransplant immunologic determinants of ABMRh.

Pretransplant NHADIA Results Predict ABMRh

We next sought to determine the best pretransplant NHADIA threshold for predicting post-transplant ABMRh. We plotted log-rank test P values of the comparison of time to ABMRh occurrence for various NHADIA thresholds, and we observed an inverted peak corresponding to a NHADIA value of 1.87 (Supplemental Figure 3). Thus, we defined this 1.87 NHADIA threshold as the indicator for non-HLA Ab presence. Figure 6A illustrates that the NHADIA threshold of 1.87 discriminated between patients who were at risk of ABMRh and those who were not (P=0.006, log-rank test).

Figure 6.: The pretransplant NHADIA result predicts ABMRh. (A) Kaplan–Meier representation of the cumulative incidence of ABMRh according to pretransplant NHADIA results. Data are based on 389 KTRs. (B) Multivariate Cox analysis of the risk of ABMRh according to the pretransplant HLA-DSA status and NHADIA status. (C) Kaplan–Meier representation of the cumulative incidence of ABMRh according to the pretransplant NHADIA status and HLA-DSA status. (B and C) Data are based on 386 KTRs due to missing HLA-DSA data. The number of events refers to the occurrence of an ABMRh, defined by biopsy specimens that fulfilled the first two (histologic) Banff criteria for ABMR.

To investigate whether non-HLA Abs assessed by the NHADIA can act in synergy with HLA-DSAs, resulting in an increased risk of ABMRh, patients were subsequently stratified into four groups according to their statuses for HLA-DSAs and non-HLA Abs (Supplemental Table 1). Multivariate Cox regression analysis demonstrated that, compared with patients with no HLA-DSAs or non-HLA Abs, patients with non-HLA Abs but no HLA-DSAs had a nonsignificant increased risk of ABMRh (hazard ratio [HR], 1.56; 95% CI, 0.92 to 2.65; P=0.10), patients with HLA-DSAs but no non-HLA Abs had a significantly increased risk of ABMRh (HR, 1.70; 95% CI, 1.00 to 2.88; P<0.05), and patients with both HLA-DSAs and non-HLA Abs at the time of transplantation had the highest risk of ABMRh (HR, 4.77; 95% CI, 2.29 to 9.91; P<0.001; Figure 6B). Figure 6C depicts the cumulative incidence of ABMRh for the four groups. Patients with neither type of Abs experienced the best outcome, with a cumulative ABMRh incidence of 26.3% at 4 years. In contrast, patients with HLA-DSAs and non-HLA Abs experienced the poorest outcome, with a cumulative ABMRh incidence of 79.5% at 4 years. Patients with HLA-DSAs but no non-HLA Abs, and patients with non-HLA Abs but no HLA-DSAs, displayed intermediate risk with cumulative ABMRh incidences of 41.5% and 38.7%, respectively, at 4 years (Figure 6C). Importantly, the median (IQR) HLA-DSA MFI values were similar in patients with or without non-HLA Abs (2797 [1322–5559] versus 1787 [1317–3610], respectively; P=0.44, Mann–Whitney test).

In addition, we performed a sensitivity study in which we considered other cutoff values of MFI (500, 2000, and 3000) for defining HLA-DSA positivity (Supplemental Figure 4). Interestingly, irrespective of the HLA-DSA MFI cutoff, we observed a very similar pattern which showed that the risk of ABMRh was dramatically increased in the combined presence of HLA-DSA and non-HLA Abs, as compared with the other groups. Again, in all comparisons, the median MFI values of the HLA-DSA remained similar between patients with and without non-HLA Abs, reinforcing the evidence for the deleterious role of non-HLA Abs in ABMRh.

As a comparison, we addressed the association between pretransplant AT1R Abs and post-transplant ABMRh. Using a threshold of 10 IU/ml, we identified 88 (22.7%) KTRs with pretransplant AT1R Abs. Comparison of pretransplant AT1R Ab levels and NHADIA values showed no correlation (r_s=0.02; Supplemental Figure 5A), and median (IQR) NHADIA values were similar in KTRs with or without AT1R Abs (1.29 [1.15–1.61] versus 1.25 [1.13–1.55], respectively; P=0.47; Supplemental Figure 5B). More importantly, whereas pretransplant NHADIA values identified KTRs at increased risk of ABMRh (Figure 6), pretransplant AT1R Abs did not (Supplemental Figure 5C).

Reclassification of Patients with Non-HLA Abs According to Banff Classification Updates

According to the Banff 2017 classification, acute rejection developed in 74 patients: 18 patients had T cell–mediated rejection, and 56 had ABMR. No evident association with NHADIA status was observed in these two entities (Figure 7A). However, the proportion of patients with a NHADIA result >1.87 was significantly greater among those with ABMRh (P=0.008) and MVI (P=0.002), two conditions that are not sufficient to diagnose ABMR according to the Banff 2017 classification, due to the absence of demonstration of pathogenic Abs (Figure 7A).

Figure 7.: The pretransplant NHADIA result could improve the Banff classification for ABMR. (A) NHADIA status in KTRs with or without T cell–mediated rejection (TCMR) or ABMR according to the Banff 2017 classification, ABMRh, and MVI (N corresponds to the numbers of patients diagnosed with the corresponding histologic features, P values by chi-squared tests). (B) All post-transplant biopsy specimens were classified according to Banff 2013 and Banff 2017 classifications. Of these, 754 biopsy specimens were not considered sABMR or ABMR in any classification. The evolution of the diagnostic categorization of the other 179 biopsy specimens across different Banff versions is depicted. The proposed reclassification according to NHADIA status is also depicted. Line colors represent the diagnostic class in the preceding Banff version.

To assess whether NHADIA status can be used to evolve definitions of ABMR, we performed a landmark analysis of all our biopsy specimens. Among the 933 included biopsy samples, 179 specimens were classified as ABMR or suspicious for ABMR (sABMR) by at least one of the 2013 or 2017 updates of the Banff classification (Figure 7B). A total of 127 biopsy specimens were classified as sABMR by the Banff 2013 classification because of morphologic and serologic evidence (v>0 and HLA-DSA positivity) or immunohistologic evidence (v>0 and C4d positivity, or at least moderate MVI). Thus, mainly due to the use of at least moderate MVI as the second criterion, the Banff 2013 classification identified the largest number of biopsy specimens diagnosed with sABMR or ABMR (n=127 and n=52, respectively). Because the suspicious category was eliminated in the Banff 2017 classification, 89 of 127 (70.1%) of the sABMR biopsy specimens classified according to Banff 2013 were reclassified as no ABMR by the Banff 2017 classification due to the absence of the second criterion (C4d positivity with positive g or peritubular capillaritis) or the third criterion (C4d negative, HLA-DSA negative with MVI). The other 38 of 127 (29.9%) Banff 2013 sABMR biopsy specimens were reclassified as ABMR by the Banff 2017 classification because of positive C4d staining and MVI in the absence of HLA-DSAs. Of the 52 Banff 2013 ABMR biopsy specimens, all remained classified as ABMR by the Banff 2017 classification. The NHADIA, by detecting the presence of non-HLA Abs directed against the glomerular endothelium, may be an alternative way to fulfill the third criterion of the Banff classification. We thus proposed reclassifying patients with an NHADIA result >1.87 but no HLA-DSAs as ABMR in an evolution of the Banff classification. By doing so, 19 biopsy specimens classified as no ABMR by Banff 2017 would be considered ABMR specimens due to the presence of non-HLA Abs (Figure 7B).

Discussion

ABMR is thought to be the main determinant of kidney allograft failure in the long term.² However, its accurate diagnosis remains challenging, as exemplified by the regular iterations of its diagnostic criteria.^8,9,21 The main diagnostic challenge lies in the lack of demonstration of pathogenic Abs in the serum of many patients with ABMRh. Indeed, recent large cohort assessments demonstrated that 40%–60% of ABMRh cases are HLA-DSA negative at the time of biopsy.^3,4,22 In this study, we developed a method that allowed for the identification of non-HLA Abs that were not only associated with ABMRh independent of HLA-DSAs but also appeared to act synergistically with HLA-DSAs to induce ABMRh.

We previously demonstrated that non-HLA Abs associated with ABMRh primarily bind in a very specific manner to glomerular endothelial cells.⁶ Our previous study used CiGEnC cells as targets, but their basal expression of HLA antigens limited their application for non-HLA Ab detection in patients without circulating anti-HLA Abs. To tackle this obstacle, we sequentially applied a CRISPR/Cas9 strategy to delete both the B2M and CIITA genes by a nonhomologous end-joining pathway to obtain a CiGEnCΔHLA clone (Supplemental Appendix 2). CiGEnCΔHLA cells remained undistinguishable from the parental cell line in terms of morphology and phenotype and allowed us to develop the NHADIA.

The evaluation of the NHADIA in an unselected cohort of KTRs revealed that non-HLA Abs were increased in patients who underwent previous kidney transplantation, supporting the role of allosensitization to minor histocompatibility antigens, a sensitization that has been shown to affect long-term graft outcomes.^23,24 A large array of non-HLA antigens, including polymorphic antigens that differ between the recipient and donor, such as the recently described genomic mismatch at the LIMS1 locus,²⁴ have already been suspected as targets for non-HLA sensitization, and other candidates have been described, including anti-AT1R Abs,^25,26 anti-ETAR Abs, polyreactive natural Abs,²⁷ and many others.^6,2829–30 More generally, the overall non-HLA mismatch burden between a donor and recipient was recently associated with poor graft survival in a genome-wide analysis.²³ Interrogating the whole genome of donors and recipients, Reindl-Schwaighofer et al.²³ revealed the large number of mismatches that might induce a humoral alloimmune response in the recipient. Indeed, the authors identified a median value of 1892 nonsynonymous single-nucleotide polymorphism mismatches in immune-accessible proteins between donors and recipients. Interestingly, the degree of nonsynonymous single-nucleotide polymorphism mismatch was independently associated with graft loss, thus suggesting that this tremendous diversity should be taken into account when addressing the extent of immunologic injury to the graft. In this respect, we believe that our cell-based assay, which is not a candidate-driven approach but integrates the diversity of plausible targets, may improve the individual risk assessment and diagnosis of non-HLA Ab–mediated cell injury.

The pretransplant NHADIA value correlated with MVI lesions on the kidney graft at 3 months and 12 months and was associated with the risk of developing ABMRh. MVI has been associated with poor kidney graft outcomes, and its deleterious effect has been observed even in the absence of HLA-DSAs.^31,32 Interestingly, we observed a biologic gradient between the pretransplant value of the NHADIA and the occurrence and severity of MVI on kidney biopsy specimens, suggesting a causal role for these non-HLA Abs as mediators of endothelial injury. Our multivariate regression analysis confirmed that pretransplant non-HLA Abs and HLA-DSAs were associated with post-transplant MVI, and that increasing the levels of either type of Ab independently increased the risk of MVI. Of note, the NHADIA value was specifically associated with MVI but not with tubulitis or interstitial inflammation, two features of T cell–mediated rejection, again supporting the instrumental role of non-HLA Abs in the development of MVI. The biologic plausibility of the causal link is further supported by the strong association between high NHADIA values and glomerulitis, which was substantially stronger than that with peritubular capillaritis, confirming that non-HLA Abs recognize antigens specifically expressed by the glomerular endothelium. By using glomerular endothelial cells, we created a test with high specificity for detecting Abs that target antigens specifically expressed on glomerular endothelial cells. These observations are in line with the recent observation obtained by single-cell RNA sequencing that, even among kidney endothelial cells, glomerular endothelial cells express distinct gene sets compared with peritubular endothelial cells.³³

The NHADIA may have consequences for firmly diagnosing a number of unresolved, suspected cases of ABMR. MVI, a key feature of ABMR, has remained a cornerstone parameter in the consecutive Banff classifications. Nevertheless, the third criterion of the Banff classification, based on serologic evidence of DSAs or indirect proof provided by C4d-positive staining or validated transcripts, remains a prerequisite to establish a definite diagnosis of ABMR.²¹ A considerable number of kidney biopsy specimens display some of the pathologic features of ABMR (ABMRh), without satisfying the third criterion, theoretically preventing a physician from diagnosing ABMR. In a recent study, Callemeyn et al.³⁴ showed that the previous entity of “suspected ABMR” in the Banff 2013 classification, now considered “no ABMR,” is mainly represented by ABMRh lacking C4d staining and HLA-DSA positivity. However, this entity was significantly associated with impaired allograft survival in their cohort, suggesting ABMRh without HLA-DSAs and/or C4d staining should not be neglected in clinical practice.³⁴ Furthermore, Shinstock et al.³⁵ showed that almost 50% of cases with ABMRh lacking HLA-DSA positivity and C4d staining were considered ABMR by clinicians and were mainly treated as such, although the Banff-based diagnoses would have concluded an absence of ABMR. This finding results from the challenge of meeting the three Banff criteria, especially the identification of a DSA. In the absence of HLA-DSAs, non-HLA Abs are suspected but often remain unrecognized, preventing the biopsy specimen from fulfilling the third criterion. To address this issue, the Banff classification allows for the indirect proof of DSAs from C4d staining, which has poor sensitivity with up to 50% negative results in HLA-DSA–associated ABMR,² and validated transcripts in a kidney biopsy specimen, which is not performed in routine practice. Hence, a portion of authentic ABMR cases are currently misdiagnosed and consequently mistreated. The NHADIA, by detecting the presence and measuring the quantity of non-HLA Abs, may be viewed as an alternative way to fulfill the third Banff criterion, confirm the involvement of Abs, and adjust treatment regimens to target Abs.

Although CiGEnCΔHLA cells express low level of AT1R (not shown; see our transcriptomic data for details, Gene Expression Omnibus (GEO) accession number GSE181398), we observed no correlation between the NHADIA results and AT1R Ab levels, suggesting the NHADIA results are not primarily explained by AT1R Abs. This result is in line with our previous study that suggested that non-HLA Abs associated with ABMRh which bind to CiGEnC cells might recognize a wide diversity of antigens.⁶ Of note, the specificity of the AT1R quantification method has been questioned.³⁶ In addition, our findings that pretransplant AT1R Ab levels do not predict subsequent rejection (Supplemental Figure 5C) are concordant with our previously published multicenter study in independent KTRs.³⁷

As a novel assay, the NHADIA needs to be further improved and validated in additional and/or external cohorts before clinical implementation. Notably, performing NHADIA at the time of biopsy and conducting a longitudinal study is required to support the deleterious role of non-HLA Abs during MVI/ABMRh. Nevertheless, NHADIA is a useful tool to address important questions regarding the role of non-HLA Abs in ABMR/endothelial injury. For instance, whether non-HLA Abs can induce complement-dependent cytotoxicity or Ab-dependent cellular cytotoxicity is still unclear. Endothelial cell phenotypic changes and intracellular signaling in response to non-HLA Ab binding are also largely unknown. These aspects are now under investigation in our laboratory. Our in vitro test also has the potential to identify new targets of non-HLA Abs. Finally, an HLA-deleted endothelial cell may be a valuable model for studying any HLA-independent cell-cell interaction/recognition.

Our study had several limitations. The relatively short follow-up period prevented us from studying the association of pretransplant NHADIA values and long-term graft outcomes. In addition, we applied the NHADIA only to pretransplant serum samples, but de novo non-HLA Abs might also develop after transplantation, a concept that will need to be addressed. In addition, developing and combining different CiGEnCΔHLAs clones derived from several donors with various genetic backgrounds could provide a wider and more mixed genetic representation, thus ensuring a more “universal” cell-based assay. It could notably help to capture more “donor-specific” non-HLA Abs targeting polymorphic proteins. Finally, our cell-based assay tested only reactivity to surface antigens. A number of intracellular proteins have been suspected to be targets of non-HLA Abs.^6,30 The role of the intracellular target candidates certainly deserves to be clarified. Whether non-HLA Abs could bind potential intracellular targets is still unclear. One may argue that, during graft lesions and endothelial damage, the endothelial intracellular compartment is accessible, and Abs recognizing intracellular proteins might bind their targets and thus participate in, and perhaps amplify, the immune injury. Additional studies are required to extend our work to cryptic intracellular antigens.

In conclusion, in addition to providing new insights into graft injury mechanisms, the NHADIA has the potential to refine risk assessment before transplantation, demonstrate the involvement of Abs at the time of alloimmune injury, improve acute rejection diagnosis, and adjust therapeutic interventions by targeting detrimental Abs.

Disclosures

D. Anglicheau reports receiving research funding from Astellas, Bristol Myers Squibb (BMS), Chiesi, and Novartis; and having consultancy agreements with, and receiving honoraria from, BMS and Novartis. C. Legendre reports receiving honoraria from Alexion, Astellas, Novartis, and Sandoz; having consultancy agreements with CSL Behring and Hansa Medical; and serving on a speakers bureau for and as scientific advisor for, or member of, Hansa Medical. S. Satchell reports receiving research funding from Evotec, Ferring, and Novo Nordisk; having consultancy agreements with Novo Nordisk; serving on the Kidney Research UK Grants Committee; and serving as a member of the UK Kidney Association. F. Terzi reports having consultancy agreements with, and receiving research funding from, ENYO; and serving as an editor and on the genetics/advisory board for Nephron Experimental Nephrology and as a member of the Swiss National Science Foundation. J. Zuber reports receiving honoraria from Alexion Pharmaceuticals and BMS Pharmaceuticals. All remaining authors have nothing to disclose.

Funding

This work was supported by the Fondation du Rein sous égide de la Fondation pour la Recherche Médicale (Prix Don de Soi-Don de Vie 2018 FdR/FRM_D.ANGLICH Subvention Transplantation et Thérapie Cellulaire 2020 FRM PME20200611626; to D. Anglicheau).