A circulating antibody panel for pretransplant prediction of FSGS recurrence after kidney transplantation

Abstract

Recurrence of focal segmental glomerulosclerosis (rFSGS) after kidney transplantation is a cause of accelerated graft loss. To evaluate pathogenic antibodies (Abs) in rFSGS, we processed 141 serum samples from 64 patients with and without primary rFSGS and 34 non-FSGS control patients transplanted at four hospitals. We screened about 9000 antigens in pretransplant sera and selected 10 Abs targeting glomerular antigens for enzyme-linked immunosorbent assay (ELISA) validation. A panel of seven Abs (CD40, PTPRO, CGB5, FAS, P2RY11, SNRPB2, and APOL2) could predict posttransplant FSGS recurrence with 92% accuracy. Pretransplant elevation of anti-CD40 Ab alone had the best correlation (78% accuracy) with rFSGS risk after transplantation. Epitope mapping of CD40 with customized peptide arrays and rFSGS sera demonstrated altered immunogenicity of the extracellular CD40 domain in rFSGS. Immunohistochemistry of CD40 demonstrated a differential expression in FSGS compared to non-FSGS controls. Anti-CD40 Abs purified from rFSGS patients were particularly pathogenic in human podocyte cultures. Injection of anti-CD40/rFSGS Ab enhanced suPAR (soluble urokinase receptor)-mediated proteinuria in wild-type mice, yet no sensitizing effect was noted in mice deficient in CD40 or in wild-type mice that received blocking Ab to CD40. In conclusion, a panel of seven Abs can help identify primary FSGS patients at high risk of recurrence before transplantation. Intrarenal CD40 (and possibly other specific glomerular antigens) is an important contributor to FSGS disease pathogenesis. Human trials of anti-CD40 therapies are warranted to evaluate their efficacy for preventing rFSGS and improving graft survival.

Copyright © 2014, American Association for the Advancement of Science.

Figures

Fig. 1. Study work flow

A schematic of the study outlining the independent patients and samples used in discovery and validation of the Ab panel for predicting rFSGS and the experimental evaluation of CD40 in the pathogenesis of rFSGS.

Fig. 2. Validation of the FAST Ab panel in rFSGS and the predictive accuracy of subsets of this panel

(A to G) ELISA assays were developed and optimized to validate rFSGS-specific Ab. Black bars correspond to serum samples collected immediately before transplantation [day 0 (D0), n = 28] and at 1 year after transplantation [year 1 (Y1), n = 26] in patients who experienced rFSGS. White bars correspond to serum samples collected immediately before transplantation (day 0, n = 31) and at 1 year after transplantation (year 1, n = 17) in patients who did not experience rFSGS during the first year. Gray bars represent non-FSGS control samples (n = 34) and consist of patients with ESRD from causes other than FSGS (obstructive uropathy, reflux nephropathy, cystinosis, and glomerulonephritis) and six normal healthy controls. Y axis represents the ratio of the optical density (OD) from MSD ELISA assay over total IgG in mg/dl for each patient. ELISA results were analyzed with a Mann-Whitney test in GraphPad Prism. Respective P values are provided on the top of each bar. A P value of <0.05 is considered significant. NS, not significant. (H) ROC analysis for three fitted logistic regression models. The outcome is recurrence versus nonrecurrence of FSGS, and the independent predictors are the log-transformed relative fluorescent signal values of seven Abs: CD40, PTPRO, FAS, CGB5, SNRPB2, APOL2, and P2RY11. The three logistic regression models fitted are shown. Model 1 is the FAST (FSGS antibody serological test) panel with all seven Abs, giving an AUC = 0.9 (CI, 0.81 to 0.99). Model 2 used three Abs (CD40, PTPRO, and CGB5), and its ROC curve has an AUC of 0.82 (CI, 0.70 to 0.95). Model 3 used only CD40 Ab data for ROC analysis, resulting in an AUC of 0.77 (CI, 0.63 to 0.92).

Fig. 3. Peptide array affinity mapping of the CD40 protein to anti-CD40/rFSGS Ab

We used the peptide microarray platform to map differential binding of anti-CD40 Ab from rFSGS to CD40 epitopes. (A) Increased binding of anti-CD40/rFSGS Ab measured in terms of relative fluorescent intensity (RFU) to specific regions of CD40 (NSQCC and ESEF), compared to binding by the anti-CD40 Ab isolated from nrFSGS controls (Ctrl). (B) Ab binding against different CD40 epitopes, each spanning 15 amino acids, and covering the full-length CD40 protein. Anti-CD40 Ab was isolated from sera of patients with rFSGS (n = 4, black circles) and nrFSGS controls (n = 4, open squares). Black bars indicate the average fold increase in Ab responses against different CD40 antigenic epitopes in rFSGS sera relative to sera from nrFSGS. Areas of increased signal with more than two adjacent peptides in the extracellular region are boxed. Epitopes with unpaired t test P < 0.05 and ≥ 2-fold increase for rFSGS/nrFSGS controls (labeled in red) were considered significant.

Fig. 4. IHC of CD40 staining in kidney biopsies from patients with rFSGS

(A) No CD40 staining is observed in the podocytes of this normal human glomerulus from the tumor-free part of a total nephrectomy for renal cell carcinoma. Vascular hilum is marked by an arrowhead. (B) Focal podocyte labeling for CD40 (arrow) in a case of rFSGS. (C) Strong CD40 signal is observed in the hyperplastic podocytes (arrows) covering an rFSGS lesion. Scale bars, 25 µm.

Fig. 5. Impact of CD40-blocking Ab on podocyte depolarization caused by human anti-CD40/rFSGS Ab

(A) Completely differentiated human podo-cytes were treated with PBS (Con, control), LPS (50 µg/ml), PAN (50 µg/ml), 2% pretransplant sera from rFSGS patients, recombinant suPAR (1 µg/ml), or CD40 autoAb from FSGS patients. After treatment, the podocytes were stained with phalloidin for F-actin analysis. Treatment of podocytes with purified CD40 autoAb or FSGS sera, which contain a high concentration of CD40 autoAb, depolarized podocytes and caused peripheral aggregation and decreased central expression of F-actin assessed in terms of mean fluorescence intensity (MFI). suPAR treatment generated a similar phenotype. (B) Completely differentiated human podocytes were treated with a monoclonal CD40 Ab to examine its effect on CD40 autoAb–induced podocyte injury. Con, PBS control; rFSGS CD40 IgG, CD40 autoAb purified from rFSGS patient sera (2 µg/ml); CD40 mAb, a CD40 monoclonal Ab used at 1:1 ratio relative to CD40 autoAb. (C) Blocking suPAR–β3 integrin pathway ameliorates podocyte depolarization caused by human anti-CD40/rFSGS Ab. Completely differentiated human podocytes were cotreated with rFSGS CD40 IgG and suPAR-blocking monoclonal Ab (suPAR mAb, 1 µg/ml) or cycloRGDfV (1 µg/ml), a small molecule inhibiting αVβ3 integrin activity. Compared to rFSGS CD40 IgG alone, cotreatment with suPAR-blocking monoclonal Ab or cycloRGDfV increased podocyte polarity (left panel) and F-actin levels (right panel). One-way analysis of variance (ANOVA) was used to calculate P values (provided in the figure), and a P value <0.05 was considered significant. Scale bars, 20 µm.

Fig. 6. Induction of proteinuria with human anti-CD40 Ab/rFSGS in mice

(A to E) Wild-type C57BL/6 mice (A to C) or CD40−/− mice (D and E) were treated with anti-CD40 Ab isolated from rFSGS (red labels) or nrFSGS (blue labels) patients. The change in ACR was greater when C57BL/6 mice were injected with two doses of anti-CD40 Ab/rFSGS compared to anti-CD40 Ab/nrFSGS; the change in ACR between day 0 and day 8 is shown in (A). With co-injection of suPAR, the increase in ACR was again greater in C57BL/6 mice injected with anti-CD40 Ab/rFSGS, and proteinuria increased more than three-fold from a baseline of 102.9 ± 7 (B). Injection of CD40-blocking Ab into C57BL/6 mice cotreated with suPAR and anti-CD40 Ab/rFSGS significantly reduced proteinuria (C). Injection of anti-CD40 Ab/rFSGS into CD40−/− mice did not cause significant proteinuria (D). CD40−/− mice cotreated with anti-CD40 Ab/rFSGS and suPAR showed a significant increase in ACR, but this was not seen when these mice were cotreated with anti-CD40 Ab/nrFSGS and suPAR (E). P values (provided in the figure) were determined with unpaired t test calculated in GraphPad Prism. A P value of <0.05 was considered significant.