A Nef Precursor or Benign Counterpart: High Prevalence of Healthy Subjects with Antibody Reactivity to the C3-Convertase

Key Points

• There are two anticonvertase antibody populations—Nephritic Factors which cause complement dysregulation and C3CAbs which do not.

• C3CAbs are highly prevalent in the healthy adult population (>95% prevalence).

• C3CAbs suggest a direction of study to determine the origin of Nephritic Factors.

Introduction

Nephritic Factors (Nefs) are pathogenic anticonvertase autoantibodies that dysregulate complement activity by stabilizing the convertase and antagonizing complement regulator proteins.¹ The resultant, prolonged convertase activity facilitates the accumulation of complement breakdown products, which are detectable by renal biopsy and biomarker panels. Nefs are commonly associated with complement-mediated kidney disease and are identified in 50%–80% of patients with C3 glomerulopathy.² Despite this frequency, the origin of Nefs is unknown and has not been recently investigated (the most recent significant attempts to identify an origin were published in the 1990s).^3–5 Furthermore, the prevalence of convertase-directed autoantibodies in the healthy population remains unknown. Anticonvertase autoantibodies have been found in clinically unaffected individuals with hypocomplementemia.^6–10 Although the antibodies were not associated with C3 glomerulopathy in this setting, they did promote complement hyperactivity and resulted in measurable hypocomplementemia, making them not completely benign. On the other hand, it is generally believed that anticonvertase autoantibodies are not present in healthy individuals with normal complement activity. This belief may be due to the difficulty in studying a hypofunctional anticonvertase autoantibody population.

The complement convertase complexes are relatively unstable, which forces most current routine assays to detect and quantify Nefs by their effect on convertase activity rather than through simple binding assays.^2,11 There is risk that these assays introduce a type 2 error for anticonvertase autoantibodies that do not influence underlying complement activity. Alternatively, assays that do quantify Nef binding often require incubation periods and washing steps that prevent the reporting of real-time molecular interactions.^1,2,11 Zhang et al. suggested that the ELISA-based method for Nef detection likely introduces a high rate of false-negative results because weak antibody binding in vitro is unable to survive washing steps.² Furthermore, many of the assays that statistically quantify Nefs assume that anticonvertase antibodies are absent in normal serum control samples when establishing negative reference values. Although this assumption is practical and sufficient for quantifying pathogenic Nef activity, it necessarily excludes these assays from rigorously evaluating the normal population for anticonvertase antibodies. This study aimed to determine the prevalence of convertase-directed autoantibodies in healthy, normal individuals, and in doing so, provide a possible origin for Nefs while promoting greater interest in the study of this important research topic.

Methods

Four proteins interact to form an antibody-bound alternative pathway C3-convertase complex: activated complement component C3 (C3b), Factor B (FB), Factor D (FD), and the antibody. First, FB recognizes and binds to C3b. In the presence of Mg²⁺, the proconvertase (C3bB) forms, which rearranges FB and allows FD to bind the C3bB complex. FD then cleaves FB into Ba and Bb. The Ba fragment and FD dissociate from the complex, while Bb remains bound to C3b. The result is the mature convertase, C3bBb.¹² Finally, the antibody binds to the completed complex at a newly exposed epitope on C3bBb or C3bB.

This study explored the prevalence of C3-convertase autoantibodies (C3CAbs) in the normal population using surface plasmon resonance (SPR), which measures protein–protein interactions in real time. SPR uses an immobilized binding partner (called the ligand), which is covalently bound to a chip surface, and a fluid-phase binding partner (called the analyte), which is injected over the ligand. If the analyte associates with the immobilized ligand (i.e., FB binds C3b), a proportional increase in the resonance units (RUs) is observed (Supplemental Figure 1). Persistently elevated RUs postinjection indicates the formation of a molecular complex. The interaction is continuously monitored and reported on the y-axis of a sensorgram. SPR provides two key advantages for evaluating anticonvertase autoantibodies. First, the readout from this approach measures the presence or absence of antibody recognition regardless of the functional consequence of the antibody–antigen interaction. Second, the real-time, continuous report in SPR is independent of convertase instability, which is a major limitation of typical binding assays, such as ELISA. This study adapted SPR for a specificity binding assay by measuring the association between the C3b ligand and each antibody sample in several distinct analyte conditions.

Specificity Assay

Specificity of each antibody sample (sample selection discussed below) was determined by comparing four pairs of analyte conditions against immobilized C3b. These four pairs are presented in the four panels of Figure 1 as superimposed sensorgrams. Each pair included one test analyte, which contained purified IgG, and one control analyte, which did not include purified IgG but was otherwise identical. The test analyte and control analyte were tested separately and then compared for differences in binding, which are attributed to the influence of IgG. The first analyte pair (Figure 1A) was Buffer versus IgG alone. This analyte pair identified anti-C3b autoantibodies, which have been previously reported² and could introduce false-positive results during subsequent analyte conditions. The second analyte pair (Figure 1B) was FD alone versus FD with IgG (FD+IgG). This pair measured the influence of purified IgG on FD and C3b, which are normally nonreactive reagents. If IgG introduces an interaction, this assay feature could indicate a false-positive result to the convertase forming analytes. The third analyte pair (Figure 1C) was FB alone versus FB with IgG (FB+IgG). This pair identified antibody recognition of the proconvertase. The final analyte pair (Figure 1D) was FB with FD (FB+FD) versus FB with FD and IgG (FB+FD+IgG). This pair identified antibody recognition of the convertase. For a detailed discussion on the SPR method, including injection sequence and data readout, see Supplemental Figure 1.

Figure 1.

Example Sensorgrams for a typical Nef-positive control and a sham-negative control (anti-C4b) sample for each of the four analyte pairs. Each sensorgram shows one pair of analytes (IgG-containing test analyte with its corresponding antibody-negative control analyte), which were tested separately against the C3b ligand and then superimposed in the figure for the detection of differences in binding. Significant differences between the test and control analyte are attributable to the antibody. In all sensorgrams, the x-axis represents time in seconds and the y-axis represents response in RUs. For each analyte pair (A–D), the sham-negative control anti-C4b IgG sample and a typical Nef-positive control IgG sample are shown. (A) IgG alone in buffer was tested and is shown superimposed over buffer injected with no IgG. (B) FD with IgG was tested and is shown superimposed over FD alone in buffer. (C) FB with IgG was tested and is shown superimposed over FB alone in buffer. Note that FB alone causes formation of the proconvertase. (D) FB with FD and IgG was tested and is shown superimposed over FB with FD in buffer. Note the FB with FD causes formation of the convertase. ^aUnreactive analyte pair in which neither the control analyte nor the test analyte influence binding. ^bReactive control analyte with unreactive test analyte. ^cReactive antibody-containing test analyte (FB+Nef and FB+FD+Nef); ^dcorresponding antibody-negative control analyte (FB and FB+FD). ^eΔRU calculated by Equation 1. ^fInsignificant background artifacts. FB, Factor B; FD, Factor D; Nef, nephritic factors; RU, resonance unit.

Evaluation

As seen in Figure 1, C and D, the control response for FB and FB+FD includes formation of C3bB and C3bBb. To compensate for this baseline reaction, the reactivity of the antibody was quantified using Equation 1 (represented by “e” in Figure 1):

$$∆{\text{RU}}_{\text{pair}}={\text{RU}}_{(\text{test}\hspace{0.17em}\text{analyte})}-{\text{RU}}_{(\text{control}\hspace{0.17em}\text{analyte})}$$ (1)

For example, the reactivity of the antibody to the convertase was measured as:

$$∆{\text{RU}}_{\left(\text{IgG}\right)\mathrm{C}3\text{bBb}}={\text{RU}}_{\text{FB}+\text{FD}+\text{IgG}}-{\text{RU}}_{\text{FB}+\text{FD}}$$

In addition, two approaches were used to evaluate the specificity of the antibody interaction and determine their prevalence in the sample cohort.

Approach 1: Control for error due to a false-positive reaction in IgG alone and IgG+FD (Equations 2, 3, and 4)

Convertase

$$If,\; ∆\text{RU}={\text{RU}}_{\text{FB}+\text{FD}+\text{IgG}}-{\text{RU}}_{\text{FD}+\text{IgG}}>{(\overline{\mathrm{x}}+2s)}_{\text{FB}+\text{FD}}\hspace{0.17em}\text{is}\hspace{0.17em}\text{true},$$ (2)

$$and\; ∆\text{RU}={\text{RU}}_{\text{FB}+\text{FD}+\text{IgG}}-{\text{RU}}_{\text{IgG}}>{(\overline{\mathrm{x}}+2s)}_{\text{FB}+\text{FD}}\hspace{0.17em}\text{is}\hspace{0.17em}\text{true},$$ (3)

then the sample is positive.

Proconvertase

$$If,\; ∆\text{RU}={\text{RU}}_{\text{FB}+\text{IgG}}-{\text{RU}}_{\text{IgG}}>{(\overline{\mathrm{x}}+2s)}_{\text{FB}}\hspace{0.17em}\text{is}\hspace{0.17em}\text{true}$$ (4)

then the sample is positive.

Approach 1 is at risk for reporting false-negative results in samples ran later in the study because the aging C3b-chip becomes marginally, but distinctly, less reactive over time. Therefore, a secondary independent approach was used to validate results by approach 1.

Approach 2: Secondary Evaluation (Equations 5 and 6)

Convertase

$$If,\; ∆{\text{RU}}_{\text{Sample}}={\text{RU}}_{\text{FB}+\text{FD}+\text{IgG}}-{\text{RU}}_{\text{FB}+\text{FD}}>(\overline{\mathrm{x}}+2\mathrm{s})\hspace{0.17em}\text{of}∆{\text{RU}}_{(\text{Sham}\hspace{0.17em}\text{IgG}\hspace{0.17em}\text{Population})}\hspace{0.17em}\hspace{0.17em}\text{is}\hspace{0.17em}\text{true},$$ (5)

then the sample is positive.

Proconvertase

$$If,\; ∆{\text{RU}}_{\text{Sample}}={\text{RU}}_{\text{FB}+\text{IgG}}-{\text{RU}}_{\text{FB}}>(\overline{\mathrm{x}}+2\mathrm{s})\hspace{0.17em}\text{of}∆{\text{RU}}_{(\text{Sham}\hspace{0.17em}\text{IgG}\hspace{0.17em}\text{Population})}\hspace{0.17em}\hspace{0.17em}\text{is}\hspace{0.17em}\text{true},$$ (6)

then the sample is positive.

Although independent of time, approach 2 is dependent on the assumption that (x̅+2s) for the sham IgG population ΔRU provides an accurate means of determining specificity. Using these two independent measures of specificity together provides the strongest approach for evaluating each sample for its reactivity to the convertase and proconvertase.

Samples

Two cohorts were evaluated. Cohort one included 13 monoclonal or polyclonal sham IgG samples and 13 Nef-positive purified IgG samples to serve as negative and positive populations, respectively. The sham IgG sample ligands and antibody IDs are presented in Supplemental Table 1. Cohort two included whole IgG purified from healthy donor samples to serve as the test population. Before testing by SPR, healthy donors were evaluated by traditional Nef assays (immunofixation electrophoresis and hemolytic assay) and a complement biomarker panel (C3, C3c, FB, and/or Ba and Bb, C5, and/or sC5b-9, Factor H).² Only samples that were negative for Nef activity and had normal complement activity were included in the normal human cohort, for a total of 30 individual samples. Note that normal samples 7, 15, and 18 did not undergo the C3 and C3c biomarker tests. However, these samples were still normal for FB, C5, and Factor H biomarkers and negative for Nef activity by both immunofixation electrophoresis and hemolytic assay. IgG for the Nef control population and the normal human test population was purified using Melon Gel (ThermoFisher), and all antibody samples were tested in 10 mM MgCl₂ HBS-EP running buffer.

To summarize, 56 IgG samples were independently tested for reactivity to four pairs of analyte conditions: IgG paired with buffer, FD+IgG paired with FD alone, FB+IgG paired with FB alone, and FB+FD+IgG paired with FB+FD. Of the 56 samples, 13 were negative control sham antibodies, 13 were positive control Nefs known to bind the convertase, and 30 were single-source normal human IgG with no Nef activity and normal complement biomarkers. The data were evaluated by Equations 1–6, and statistical analyses of the data were performed using GraphPad Prism 9.4.1.

Results

The results for all samples are shown in Figure 2 with statistical parameters reported in Table 1. Figure 2A presents the binding (RU) for the four control analyte conditions and two test analyte conditions: IgG alone and FD+IgG. As expected, buffer and FD did not respond to the C3b ligand, and this remained true regardless of the presence of sham antibodies, Nef antibodies, or normal human antibodies. Formation of proconvertase without antibody (FB alone) increased the mean RU from approximately 0 to 16.4 (C3b-chip 1) and 16.7 (C3b-chip 2), while formation of convertase without antibody (FB+FD) resulted in a mean binding of 84.1 (chip 1) and 65.8 (chip 2) RU. The two results correspond to two separate C3b-immobilized chips with these differences attributable to approximately 10% difference in immobilized C3b.

Figure 2.

Study results. (A) The results for each control analyte (Buffer, FD, FB, and FB+FD) show the antibody-free baseline response. Buffer and FD are unreactive, while FB forms the proconvertase and FB+FD form the convertase. All three IgG populations show low to no influence on C3b (IgG Alone) or FD+IgG, thus indicating method specificity with low levels of false-positive interactions. (B) The calculated response to the convertase (ΔRU) of each IgG population is shown. (C) The calculated response to the proconvertase (ΔRU) of each IgG population is shown. (D, E) Evaluation of reactivity for each Nef and sham control sample (D) and each normal human IgG sample (E) by approach 1 (Equations 2–4). The dotted line indicates the threshold value (x̅+2s) in each corresponding equation. IgG samples that failed to pass the threshold value are indicated by an “x.” (F) Evaluation of reactivity for each for all samples by approach 2 (Equations 5 and 6). The dotted line indicates the threshold value (x̅+2s) in each corresponding equation. IgG samples that failed to pass the threshold value are indicated by an “x.” ^aData for the control and normal IgG populations had different (x̅+2s)_FB+FD values and. For this reason, the populations are demonstrated separately (D versus E).

Table 1.

Cohort response statistics (resonance units)

Negative and positive control cohort	Buffer	Sham IgG	Nef IgG	FD alone	FD+Sham IgG	FD+Nef IgG	FB alone	FB+Sham IgG	FB+Nef IgG	FB+FD	FB+FD+Sham IgG	FB+FD+Nef IgG
Mean	−0.8	0.6	−2.4	−0.9	0.7	−2.2	16.4	20.0	62.4	84.1	85.5	120.3
SD	0.5	2.9	1.6	0.3	2.4	1.7	2.6	13.9	39.7	8.8	11.1	22.7
25% Percentile	−1.0	−0.9	−3.4	−1	−0.6	−3.3	14.6	14.6	25.7	77.1	76.4	101.2
Median	−0.8	2	−2.3	−0.9	1.7	−1.9	15.2	15.8	55.7	80.6	79.5	117.5
75% Percentile	−0.7	2.4	−1.0	−0.6	2.3	−1.1	18.6	20.8	89.9	91.7	96.2	132

Normal human cohort	Buffer	IgG		FD alone	FD+IgG		FB alone	FB+IgG		FB+FD	FB+FD+IgG
Mean	−0.5	−2.4		−0.9	0.3		16.7	42.3		65.8	91.1
SD	0.2	2.1		0.2	2.8		2.7	9.9		4.5	11.9
25% Percentile	−0.6	−3.2		−1	−0.8		15.3	34.2		63.7	84.1
Median	−0.5	−1.5		−0.9	0.4		17.0	44.7		67.1	89.0
75% Percentile	−0.3	−1.0		−0.7	1.7		18.0	49.6		68.1	98.1

Nef, nephritic factors; FD, Factor D; FB, Factor B.

Figure 2B presents the convertase ΔRU_(IgG)C3bBb as calculated by Equation 1 for the sham IgG, Nef IgG, and normal human IgG populations. To identify the differences in ΔRU between these three sample populations, the data were evaluated using a Kruskal–Wallis H test. The results show that there was a statistically significant difference in the mean ΔRU between the different antibodies, χ2(3, N=56) = 30.2, P = <0.0001, with a mean rank ΔRU of 8.00 for the sham antibody population, 41.7 for the Nef population, and 31.7 for the normal human antibody population. Further analysis using the post hoc Dunn multiple comparisons test shows that statistically significant differences exist between the sham antibody population with both the Nef population (P = <0.0001) and the normal human antibody population (P = <0.0001), whereas the difference between the Nef population and the normal human population was not statistically significant (P = 0.1924).

Figure 2C presents the proconvertase ΔRU_(IgG)C3bB data as calculated in Equation 1. The results for the Kruskal–Wallis H test showed that there was a statistically significant difference in the ΔRU between the different antibodies, χ2(3, N=56) = 20.7, P = <0.0001, with a mean rank ΔRU of 11.1 for the sham antibody population, 38.2 for the Nef population, and 32.9 for the normal human antibody population. The post hoc Dunn multiple comparisons test shows that statistically significant differences exist between the sham antibody population with both the Nef population (P = <0.0001) and the normal human antibody population (P = <0.0004), whereas the difference between the Nef population and the normal human population was not statistically significant (P = 0.7370). The statistical parameter from Equation 1 (ΔRU) for both the convertase and proconvertase is presented in Table 2.

Table 2.

Reactivity statistics for three sample populations (ΔRU)

Target Antigen	Sham IgG	Nef IgG	Normal IgG
Convertase
25% Percentile	−1.8	23.9	18.2
Median	0.0	36.6	26.3
75% Percentile	0.7	55.3	32.4
Mean	0.1	39.8	25.2
SD	2.6	19.2	12
Proconvertase (whole data)
25% Percentile	−1.2	11.2	17.2
Median	−0.2	42.3	25.9
75% Percentile	1.0	72.4	33.1
Mean	2.8	46.4	25.6
SD	13.7	38.1	9.7
Proconvertase (significant data) a
25% Percentile	−1.2	11.2	17.2
Median	−0.4	42.3	25.9
75% Percentile	0.8	72.4	33.1
Mean	−0.9	46.4	25.6
SD	3.2	38.1	9.7

RU, resonance unit; Nef, nephritic factors.

^a“Significant data” include all samples except negative control Sample 5, which was 3.24 standard deviations from the whole data mean. The Kruskal–Wallis H and Dunn comparison tests were performed with the “whole data” set.

Collectively, these results confirm that this specificity assay can identify differences in the positive control Nef and negative control sham antibody populations and demonstrate the specific interaction between Nefs and the C3-proconvertase or C3-convertase that is absent among sham antibodies. Furthermore, the results show that the normal human cohort mimics the binding profile observed in the positive control Nef cohort, thus showing that convertase-directed antibodies (C3CAbs) are present in the normal human population.

Evaluation for specificity of the control populations and the normal human population (Equations 2–6) is shown in Figure 2, D–F. Data from these ΔRU specificity models are presented in Supplemental Tables 2 and 3. Specificity for the convertase was correctly predicted (i.e., a passing result for Nefs and failing result for sham antibodies) in 21 of the 26 control samples by both approach 1 and approach 2 (Figure 2, D and F; Supplemental Table 2). In addition, every control sample was correctly predicted by at least one of the two approaches. Meanwhile, the response to the proconvertase correctly predicted specificity in 24 of 26 samples by at least one approach. One sham control sample and one Nef control sample were incorrectly predicted by both approaches, while 22 of 26 samples were correctly predicted by both approaches.

In the normal human cohort, only one sample (Sample 18) was negative for the convertase when evaluated by both approaches 1 and 2 (Figure 2, E and F; Supplemental Table 3). In addition, Sample 27 was negative for the convertase by approach 1 but was positive by approach 2. Sample 18 was also negative for the proconvertase by approach 2 but was positive for the proconvertase by approach 1. In total, this suggests that 28–29 of the 30 normal human samples (93.3%–96.7%) carry C3CAbs which share antigen specificity with Nefs. However, every sample in the normal human cohort was negative for Nef activity and had normal complement biomarkers. Therefore, this C3CAb population is functionally distinct from pathogenic Nefs.

Discussion

The identification of C3CAb IgGs in normal subjects is unexpected and challenges current thinking. The source of these antibodies and their relation to Nefs bears consideration. One possibility is that C3CAbs have affinity to the C3-convertase through autoantigen complementarity, and repeat infections cause a low proportion of C3CAbs to develop into pathogenic Nefs. Autoantigen complementarity is known to give rise to pathogenic autoantibodies,¹³ and C3G is often preceded by an infection.¹⁴ Alternatively, C3CAbs may be naturally occurring autoantibodies (NAbs). NAbs are germline autoantibodies that arise during early immune development and persist by evading clonal deletion during immune tolerance. NAbs in their germline state are not considered pathogenic. In fact, they are identified in healthy individuals and have varied housekeeping and immune functions.¹⁵ NAbs are implicated in autoantibody-driven disease when NAb-producing B cells undergo affinity maturation to produce new pathogenic autoantibodies.¹⁵

Researchers in the early 1990s found that C3-convertase autoantibodies can be produced by immortalized B cells harvested from cord–blood, and these antibodies had high homology to germline genes typical of NAbs.³ In a follow-up report, Nefs isolated from an adult patient showed somatic mutations typical of affinity maturation.⁵ The authors hypothesized that “normal Nefs” are part of the physiologic antibody repertoire produced at birth and are subject to peripheral immune control (i.e., NAbs). Failure of peripheral control could permit excess antigen-driven C3Nef production.⁴ The results in our study support the hypothesis that “normal Nefs,” which we call C3CAbs due to their benign nature, exist and are highly prevalent in the population. One hypothesis, then, is that C3CAbs are germline NAbs that become pathogenic Nefs when their B cells undergo affinity maturation and acquire the ability to dysregulate the C3-convertase. This hypothesis suggests an origin for Nefs and supports the theory that Nefs are themselves pathogenic—not just an epiphenomenon—something that has been debated.

In conclusion, taking advantage of the sensitivity and specificity of SPR, we identified a prevalent autoantibody that specifically reacts with the C3-convertase and C3-proconvertase (C3CAbs). C3CAbs bind the convertase without promoting complement over activation, distinguishing them from Nefs. Further characterizing the relationship between C3CAbs and Nefs will include determining the biochemical differences between them and identifying the source of C3CAbs in the normal population. The a priori presence of C3CAbs offers a possible origin for pathogenic Nefs and provides an opportunity to better understand Nef pathology.

Supplementary Material

kidney360-4-1615-s001.pdf ^{(243.2KB, pdf)}

Disclosures

C. Nester reports the following: Consultancy: Advisory Board -Apellis, Biocryst, Kira, Novartis, Silence Therapeutics; Research Funding: Retrophin—Pediatric Recruiting Site for the Duet Trial; Site PI—C3G Trial; Novartis—Site PI—C3G Trial; Apellis—Site PI—C3G Trial, Biocryst; Honoraria: Advisory Board—Apellis, Biocryst, Novartis; and Patents or Royalties: UpToDate—TMA Syndromes. The remaining author has nothing to disclose.

Funding

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases from 2 R01 DK110023-06A1 (C.M. Nester).

Author Contributions

Conceptualization: Christopher T. Culek, Carla M. Nester.

Data curation: Christopher T. Culek.

Formal analysis: Christopher T. Culek, Carla M. Nester.

Funding acquisition: Carla M. Nester.

Investigation: Christopher T. Culek.

Methodology: Christopher T. Culek.

Supervision: Carla M. Nester.

Writing – original draft: Christopher T. Culek.

Writing – review & editing: Carla M. Nester.

Data Sharing Statement

Partial restrictions to the data and/or materials apply. Not all complete sensorgrams are provided, although data from them are summarized and reported. We are happy to provide any data not included in the final manuscript upon request.

Supplemental Materials

This article contains the following supplemental material online at http://links.lww.com/KN9/A400.

Supplemental Figure 1. The Specificity Assay Injection Sequence and Data Collection.

Supplemental Table 1. Negative Control Antibody Samples.

Supplemental Table 2. Control Cohort Specificity Model Evaluation.

Supplemental Table 3. Normal Human Cohort Specificity Model Evaluation.