Evaluation of Methodologies in Anti-nephrin Autoantibody Detection

Abstract

Introduction:

Recent studies discovered the prominent presence of anti-nephrin autoantibodies in minimal change disease (MCD) and focal segmental glomerulosclerosis (FSGS). However, widely different, and often unconventional autoantibody detection methods were used in these studies, making it challenging to standardize anti-nephrin antibody detection and quantification across different studies.

Methods:

Here, we compare methods of conventional ELISA, immunoprecipitation (IP)-based on-beads ELISA, immunoprecipitation-Western blotting (IP-WB), and cell- and tissue-based immunofluorescence staining with two cohorts totaling 169 patients and control individuals.

Results:

Different assay methods and antigen preparations led to method-specific false-positive and false-negative results. In general, high-quality antigens produced in human cells, combined with IP-based assays, yielded the most robust and reliable results. Among 63 and 24 samples from patients with FSGS or MCD, respectively, two patients with FSGS showed strong antibody signals in both ELISA-based assays and IP-WB, while approximately half of patients with MCD had weak signals detectable only by IP-WB.

Conclusion:

These findings highlight the importance of standardizing antibody detection methods.

Graphical Abstract

Lay Summary

Autoimmune antibodies against nephrin - a key protein in the kidney’s filtration system - have recently been reported in kidney diseases such as minimal change disease, steroid-sensitive nephrotic syndrome, and focal segmental glomerulosclerosis. Due to the different methods used to measure the antibody in patients’ blood, a wide range of antibody-positive rates were reported. We comprehensively evaluated different reagents and methods and identified contributing factors to the false positivity of the assay. Our analysis of 169 patients indicates that IP-WB using antigens produced in human cell lines is the most reliable assay for detecting anti-nephrin autoantibodies.

Introduction

Podocytopathy is associated with a broad spectrum of clinical conditions of podocyte effacement with disruption of the kidney filtration barrier. Besides genetic mutations, circulating factors injurious to the glomerulus have been proposed to cause some idiopathic kidney diseases,^1,2 and plasmapheresis is partially effective to reverse proteinuria and achieve remission.³ Investigation of the so-called “permeability factor” had led to the recent discovery of anti-nephrin autoantibodies in focal segmental glomerulosclerosis (FSGS), pediatric nephrotic syndrome, and minimal change disease (MCD).^4–11

Although nephrin antibodies are measured in plasma, kidney biopsies show a low degree of colocalization of antibody and nephrin. Instead, IgGs are mostly seen in intracellular puncta with nephrin, indicating rapid internalization of nephrin molecules from cell surface following binding to antibodies. However, the lack of convincing colocalization patterns between the antibody and nephrin at the slit junction constitutes a challenge for the description of the autoantibody.^{4, 8, 9, 11} Moreover, it also seems that the autoantibody titers are generally low and the reliability of conventional immune assays such as ELISA for detecting anti-nephrin remains controversial.^{5, 12} These caveats call for a comprehensive evaluation of detection thresholds, reproducibility, and feasibility in the clinic. We sought to compare methodologies including the selection of recombinant nephrin antigens, the setup of antibody assays, and in-solution versus on-cell/tissue protocols.

Methods

Anti-nephrin autoantibodies were measured in plasma from two cohorts of kidney disease and control patients using conventional ELISA, immunoprecipitation-Western blot, immunoprecipitation-based on-beads ELISA, and cell- and tissue-based immunofluorescent staining. Detailed methods are provided in the Supplementary Methods.

Results

We examined two sets of plasma samples, with the first cohort of 127 patients primarily used for comparing methods and reagents, and an additional cohort of 21 FSGS and 21 MCD patients for results validation.

We first tested conventional ELISA, and like previous studies, we began with the extracellular domain of nephrin (nephrin-ECD) as the template sequence of recombinant antigens. These include commercial His-tagged ECD produced using mouse NS0 cells (R&D nephrin-ECD-His, as in Shirai et al.⁹ and Raglianti et al.⁸) or human HEK293 cells (SB nephrin-ECD-His, similar to the lab-made antigens described in Watts et al.⁴ and Hengel et al.⁵), and lab-made FLAG-tagged ECD produced in HEK293 cells (Figure 1a, Supplementary Figure S1a – S1e).

Figure 1. Discrepancies caused by different antigen preparations for the detection of anti-nephrin autoantibodies by ELISA and IP-WB.

(a) Schematics of the testing antigens used in the study, including nephrin extracellular domain/ECD with either a His-(purchased from R&D Systems: produced in mouse NS0 cells; and Sino Biological/SB: produced in HEK293 cells) or FLAG-tag (produced in the lab using HEK293 cells). (b) Anti-nephrin antibody titers of all disease and control groups were measured in parallel by ELISA using the above antigens. Plasma samples were diluted 1:300 in the R&D antigen testing group and 1:100 in the other groups. Antibody reading of each well was obtained after subtracting that of the corresponding uncoated control well. The final autoantibody titer, as measured in unit/ml, was calculated using a standard curve generated with serially diluted commercial R&D sheep anti-human nephrin polyclonal antibody (pAb), where the titer of 200 ng/ml of the polyclonal antibody was defined as 1000 unit/ml. The highest titer in the healthy control group, plus the standard deviation (SD) of the group, was used as the cutoff for antibody positivity (above the dotted line), except for the R&D antigen results, where unexpected high titers were presented in both the healthy control and the disease groups. Note: the SB antigen and nephrin-ECD-FLAG ELISA results showed some samples with higher signals in uncoated wells than antigen-coated wells, resulting in negative values after background subtraction (refer to Supplementary Figure S6). These titers could not be calculated using the standard curve. We therefore arbitrarily assigned a value of −50 to these samples. Samples with negative values are shown as circles, while those with positive values are shown as dots. Asterisks mark samples FSGS-9 and FSGS-10 that were positive against all three antigens in ELISA. (c) IP-WB of SB nephrin-ECD-His immunoprecipitated by either antibody control (pAb: R&D sheep polyclonal anti-nephrin antibody) or individual patient plasma. To minimize variation caused by Western blot procedures, samples from different groups were blotted on the same membrane. SDS-PAGE gels were cut at 75 kDa and above 250 kDa protein marker positions. Gel slices containing samples from healthy control, Tx, IGAN, MCD, MN, primary FSGS-10 and SLE were transferred on to one PVDF membrane. Gel slices with rFSGS, nrFSGS, 2^ndFSGS, and primary FSGS 1–9 samples were transferred to a second membrane, while gel slices with the T1D samples were transferred to a third membrane. To facilitate comparison across different membranes, a positive control of pAb immunoprecipitated nephrin-ECD-His was included on each membrane. The samples marked with asterisks represent four moderate-to-strong positive plasmas selected for further analysis in panel d. (d) Comparison of IP-WB results for the selected positive plasma samples using nephrin-ECD-His or nephrin-ECD-FLAG as the antigen. His vs. FLAG tag-associated discrepancies were evident with results of SLE-9 and T1D-2, whereas true anti-nephrin reactivities of FSGS-9 and FSGS-10 were confirmed. (e) To further examine the possibility of false positivity caused by aberrant human antibodies against the His-tag, we added a synthetic 6xHis peptide to the IP with selected plasmas. The results showed that T1D-2 signals were specifically inhibited, indicating this patient naturally processed anti-His antibodies. In contrast, the SLE-9 signal remained unaffected, and the reason why this plasma recognizes His-tagged nephrin but not Flag-tagged nephrin still requires further investigation. Healthy control plasma and healthy control plasma spiked in with a commercial His-tag monoclonal antibody were used as negative and positive controls, respectively. In all the WB results shown in (c), (d) and (e), the nephrin-ECD antigens were detected using the sheep anti-nephrin pAb at 1:1000 dilution.

With our focus on FSGS, the first cohort included well-defined FSGS subclasses among a variety of additional kidney or non-kidney disease types¹³ (Supplementary Table S1). We measured plasma antibody by conventional ELISA (Figure 1b). The results showed an alarming discrepancy between mouse versus human cell-produced nephrin-ECD-His (Figure 1b), with broad positive signals detected using the mouse cell-produced antigen, including many from healthy controls. These reactivities against the mouse cell-produced antigen were further confirmed by Western blotting (Supplementary Figure S2), in contrast to only two positive samples, FSGS-9 and FSGS-10 of primary FSGS, reacting to HEK293-produced antigens. To rule out reactivity to mouse-specific glycoantigens,¹⁴ we deglycosylated the antigens using PNGase F. However, the false-positivity persisted (Supplementary Figure S3a, S3b) and therefore we only used HEK293-produced recombinant nephrin in subsequent experiments.

One of the drawbacks of ELISA is that it tends to have variation of background signals among plasma samples that could render net readings of negative values for antibody signals (as seen in Hengel et al.⁵). To investigate factors that may influence ELISA background levels, we tested blocking buffers and measured total protein and total IgG concentrations of the samples. However, none of these specifically affect either the background or antibody positivity (Supplementary Figure S4a – S4c).

Next, we performed immunoprecipitation (IP) in conjunction with Western blotting (WB) for detecting nephrin autoantibody. This IP-WB workflow confirmed the prominent presence of anti-nephrin antibodies in FSGS-9 and FSGS-10 (Figure 1c). Meanwhile, there were also samples showing notable but low IP-WB signals, including samples from control disease types, such as SLE-9 and T1D-2. To rule out tag-associated false positivity, we produced and tested a new FLAG-tagged nephrin-ECD antigen (without the His-tag). The result showed these SLE and T1D samples only interacted with the His-tagged nephrin (Figure 1d). A 6xHis peptide blocking assay and IP with other 6xHis-tagged proteins showed that the T1D-2, but not SLE-9 plasma, directly recognized 6xHis epitope, suggesting complex scenarios of His-tag-associated false positivity (Figure 1e, Supplementary Figure S5).

In agreement with prior studies,^4–7 we conclude that IP-WB with the use of optimum testing antigens shows reliable antibody results. However, IP-WB is laborious and often difficult to implement in clinical diagnosis that requires throughput and automation. Therefore, we implemented an immunoprecipitation-based enhanced on-beads ELISA assay using biotinylated nephrin-ECD (also FLAG-tagged, Figure 2a).

Figure 2. A high-throughput workflow for detection of nephrin autoantibody by in-well IP and on-beads ELISA.

(a) A schematic of the workflow for a 96-well assay. (b) Measurement of nephrin-ECD antibody titers in the 127 plasma samples by on-beads ELISA using the new biotinylated nephrin-ECD-FLAG as the antigen. FSGS-9 and FSGS-10 remained the only positive samples. (c) Using FSGS-9 and the sheep anti-nephrin pAb (1:1000 dilution) as positive controls, we tested a new patient cohort of MCD (n=21) and FSGS (n=21) by IP-WB detection of nephrin autoantibodies against nephrin-ECD-FLAG. To minimize variability introduced by Western blot procedures, all samples were transferred and blotted on the same PVDF membrane. While FSGS-9 remained strongly positive for anti-nephrin antibodies, the new cohort had samples showing only weak signals predominately in the MCD group, such as MCD-1, 2, 5, 8, 9, 11, and 16–20. (d) Quantification of nephrin autoantibody titers in the new patient cohort using the on-beads ELISA method. Notably, most weak positive samples identified in (c) were undetectable by this approach. (e) Immunofluorescence staining of HEK293 cells transfected with nephrin-FL-FLAG. Cells expressing full-length (FL) transmembrane nephrin were identified by staining of the cells with anti-FLAG antibody (in red), and co-stained with plasma samples (in 1:2 dilution) of either a healthy control, FSGS-9 or FSGS-10. In addition, the healthy control plasma was also spiked in with the sheep anti-nephrin pAb (adjusted to a titer like that of FSGS-9 and FSGS-10 in a 1:120 dilution in the healthy control plasma: Supplementary Figure S8b) as a positive control. A mixture of AlexaFluor488 labeled secondary anti-sheep and anti-human IgG antibodies was used to detect the pAb (sheep) and patient autoantibodies (human) to full length nephrin on the cell surface (in green; scale bar: 20 μm). While the pAb clearly stained nephrin in transfected cells, neither FSGS-9 nor FSGS-10 showed reactivity to the antigen, suggesting limited autoantibody binding to full-length nephrin expressed in these cells. (f) To compare ECD vs. FL nephrin, we performed IP-WB against plasma samples of a healthy control (HC), FSGS-9 and FSGS-10. The assay was performed under non-reducing conditions. Arrows and arrowheads point at monomeric and oligomeric nephrin, respectively. Both FSGS-9 and FSGS-10 showed weaker activity against the full-length antigen as compared to ECD. (g) To mitigate the inhibition of autoantibody binding to nephrin, we purified total IgG from FSGS-10 (Supplementary Figure S10), and then conjugated the IgG with AF594 fluorophore (Supplementary Methods). The labeled antibody was then used to stain normal kidney specimens. IF showed FSGS-10 IgG stained the slit diaphragm (in red, and podocin counterstain in green). scale bar: 20 μm.

IP-based on-beads ELISA yielded antibody signals consistently above those of the corresponding controls, indicating improved performance over conventional ELISA following the immunoprecipitation procedure (Supplementary Figure S6). The on-beads results were consistent with IP-WB, showing distinctively high signals in FSGS-9 and FSGS-10 (Figure 2b).

Having validated the antigen reagents and antibody methods, we then analyzed an additional cohort of 42 MCD and FSGS samples (Supplementary Table S2) using IP-WB and enhanced on-beads ELISA. Depending on cutoff thresholds, about half of all 21 MCD samples showed moderate IP-WB signals, whereas only one or two FSGS and healthy controls showed weakly positive bands (Figure 2c). However, it is important to note that the antibody signals in the second cohort were far lower than that of FSGS-9 from the original cohort. Furthermore, on-beads ELISA did not capture these changes (Figure 2d), suggesting lower sensitivity as compared to IP-WB.

To map epitope-level reactivities of nephrin autoantibodies we constructed an additional nephrin truncates of the extracellular domains (Ig1–8 series, in Supplementary Figure S7a, S7b). These truncates were tested against FSGS-9 and FSGS-10 in IP-WB. Whereas healthy control did not recognize any truncates (Supplementary Figure S7c), FSGS-9 and FSGS-10 showed distinct patterns of reactive bands of the truncates (Supplementary Figure S7d–S7e), suggesting nephrin epitopes could be patient- and/or disease-specific.

We next performed immunofluorescence (IF) staining of full-length nephrin expressed on transfected cells and human glomerular tissues.

First, we transfected HEK293 cells with FLAG-tagged full-length nephrin. Following the general concept of cell-based anti-nuclear antibody assays,¹⁵ we stained the cell specimens with FSGS-9 or FSGS-10 plasma. However, as anti-FLAG clearly marked nephrin-expressing cells, no IF signals from FSGS-9 and FSGS-10 were detected (Figure 2e). Similarly, when we applied the plasma to normal human kidney specimens, we did not observe specific staining of the slit diaphragm (Supplementary Figure S8a–S8d).

We reasoned that full-length nephrin could self-assemble into high-order complexes to sterically hinder antibody access,¹⁶ or that unspecified serum proteins inhibit antibody reactivity. To investigate, we analyzed nephrin-FL-FLAG from transfected HEK293 cells and observed oligomers (Supplementary Figure S9a, S9b) that were less reactive to FSGS-9 and FSGS-10 (Figure 2f). However, this was in contrast to native nephrin extracted from mouse kidneys that did not form intermolecular disulfide bridges (Supplementary Figure S9c). Next, we investigated serum effects on IF and observed a strong inhibition of commercial anti-nephrin antibody in the presence of 50% normal human plasma (Supplementary Figure S10a, S10b). To circumvent the suppression of antibody titer by plasma proteins, we purified total IgG from FSGS-9 and FSGS-10 and labeled them with AF594 fluorophore (Supplementary Figure S11a, S11b). We observed strong staining of the slit membrane by the IgG of FSGS-10 (Figure 2g, Supplementary Figures S11c, S12). Meanwhile, the IgG of FSGS-9 did not specifically stain the slit diaphragm. Despite the success of staining human tissue specimens with FSGS-10, and that the results also complement the earlier characterization of anti-nephrin autoantibodies in diseased glomeruli as IgG-containing microspheres,^4–7 the workflow involving IgG purification and labeling is too complex to be practically applicable as a routine antibody testing method.

Discussion

Our evaluation of anti-nephrin autoantibody detection methods highlighted the critical importance of both antigen preparation and assay selection for reliable measurements. Mouse cell-derived nephrin antigens produced high false-positive rates in both healthy controls and patient samples, rendering them unsuitable for this assay. By contrast, nephrin-ECD expressed in human cells minimized nonspecific reactivity and provided more accurate detection of autoantibodies. However, introducing a 6xHis tag to the nephrin-ECD led to tag-associated false positives in a few samples, similar to observations with PLA2R1 autoantibody assay in membranous nephropathy.¹⁷ These findings underscore the necessity of thoroughly validating reagents.

Among all assays, conventional ELISA remains the most widely used technique in published studies. With appropriate antigens, ELISA produced sensitive and reliable results for samples with low background signals. However, our study shows that ELISA failed to yield meaningful data in nearly half of the samples due to high and variable background, limiting its reliability for broader applications. In contrast, IP-WB consistently provided the most definitive evidence of true antibody binding, though its semiquantitative nature and labor-intensive workflow limit its scalability. To address the limitations of conventional ELISA, we developed a novel on-beads ELISA assay, which effectively reduces background interference. This immunoprecipitation-based antibody detection method can be fully automated or alternatively adapted to a more sensitive chemiluminescence immunoassay (CLIA) that is more compatible with clinical tests.¹⁸ Another important insight arises from our cell- and tissue-based immunofluorescence staining. We observed that full-length nephrin overexpressed in HEK293 cells was not recognized by positive plasma samples, likely due to its polymeric state rather than the monomeric form found in native kidney tissue. Moreover, the low titer of nephrin autoantibodies necessitated minimal plasma dilution for staining, but the resulting high plasma protein concentration inhibited antibody binding and generated nonspecific signals. Together, these findings suggest immunofluorescence-based staining is not suitable for nephrin autoantibody detection.

In conclusion, immunoprecipitation-based methods using human cell-produced FLAG-tagged nephrin-ECD provide robust detection of anti-nephrin autoantibodies, while on-beads ELISA achieves similar specificity but lower sensitivity compared to IP-WB. Further development of a high-throughput magnetic beads-directed assay using well-validated nephrin antigen is preferable for the screening of seropositive samples in both research and clinical settings. (A summary of all methods is in Supplementary Figure S13).

Supplementary Material

Supplementary File (PDF)

Data Sharing Statement

All data that supports the findings are included within the manuscript. Other source data related to this study are available from the corresponding author on a reasonable request.