BO2C11, KM41, BOIIB2 and LE2E9 are recombinant human IgG1 specific for factor VIII originating from patients with hemophilia A (17). Trastuzumab is a humanized monoclonal IgG specific for the human epidermal growth factor receptor 2 (Herceptin^®, Roche). D02M03F05, D03M01D07 and D03M02C09 are recombinant human IgG1 originating from healthy individuals generated in our lab. Wild-type IdeS and the inactive IdeS^C94S variant were produced and purified as described (17). Intravenous immunoglobulins (IVIg) are pooled polyclonal IgG from healthy donors and were from Takeda Pharmaceuticals U.S.A., Inc. (Cambridge, MA).

Plasma (N = 80) and sera (N = 56) were prepared from blood of healthy donors (aged 18 to 85 years) collected from Etablissement Français du Sang Ile-de-France (C CPSL INSERM CENTRE EST – N°18/EFS/033). Samples numbers were empirical and based on availability at the start of the study, and are in line with a previous clinical study in healthy donors (10). The samples were kept at -80°C until use with minimal cycles of thawing and refreezing.

B-cell hybridomas were generated by Proteogenix (Schiltigheim, France) following immunization of Balb/c mice with IdeS in Freund’s adjuvant and clone 29 was selected for the highest production of anti-IdeS IgG. The VH and VL genes of clone 29 were sequenced. The cDNA was cloned in human IgG and IgA expression vectors as described (24). In the case of the recombinant IgG₁, the human hinge region was replaced by that of mouse IgG₁ to prevent cleavage by IdeS. The recombinant humanized anti-IdeS IgG^cl29 and IgA^cl29 were produced in HEK293 cells using the Expi293™ Expression System (Thermofisher). IgG^cl29 was purified from supernatant on protein A-coupled beads (GE Healthcare Life Sciences, Cytiva). IgA^cl29 was purified on jacaline-agarose beads (Thermo Scientific Pierce). Purity was assessed by SDS-PAGE and concentrations determined by nanodrop (NanoDrop™, Thermo Scientific) with extinction coefficients of 235320 M⁻¹cm⁻¹ and 152640 M⁻¹cm⁻¹, respectively.

ELISA plates (Maxisorp, Nunc) were coated with the inactive IdeS^C94S variant at 2.5 μg/ml in phosphate buffer saline (PBS, Sigma-Aldrich) overnight at 4°C, and blocked for 1 hour at 37°C using PBS-3% bovine serum albumin (BSA) or PBS-3% skim milk for detection of IgG or IgA, respectively. IVIg, human plasma, human sera, human monoclonal IgG, anti-IdeS IgG^cl29 or IgA^cl29, were then incubated in serial dilutions for an additional hour at 37°C. Bound human IgG were detected using a horseradish peroxidase (HRP)-conjugated mouse anti-human IgG Fc (9040-05, Southern Biotech), revealed using the peroxidase OPD substrate (Sigma-Aldrich) and optical densities (OD) were measured at 492 nm. Bound IgA were detected using an HRP-conjugated goat polyclonal anti-human IgA antibody (2050-05, Southern Biotech), revealed using the 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (Sigma-Aldrich) and OD were measured at 450 nm. For competition ELISA, IVIg, IgG^cl29 and donors’ plasma/serum were pre-incubated with increasing concentrations of IdeS^C94S in PBS-3% BSA for 1 hour at 37°C, prior to addition to the IdeS^C94S-coated ELISA plates.

When indicated, IgG^cl29 or IgA^cl29 were used as standards to quantify anti-IdeS IgG or IgA in human plasma/serum, respectively. The limits of blank (LOB) for the quantitative IgG and IgA ELISAs were calculated using the OD of 20 randomly selected blanks (signal from wells with blocking buffer alone) (25), as LOB = mean_blank + 1.645 x SD_blank, where SD represents the standard deviation. The limits of detection (LOD) were then estimated using the 20 samples with the lowest IgG or IgA concentrations (20Low), as LOD = LOB + 1.645 x SD_20Low. The limits of quantification (LOQ) were determined using the 20 samples with the lowest IgG or IgA concentrations ≥ LOD (20LOD), as LOQ = LOB + 1.645 x SD_20LOD. LOQ values for anti-IdeS IgG and IgA ELISAs are depicted as dotted lines on all respective graphs.

For in vitro cleavage, IVIg and BO2C11 (2000 nM each) were incubated in PBS alone or with IdeS (1–8 nM) for 6 hours at 37°C, at pH 7.4. For competitive digestion experiment, IVIg and BO2C11 (2000 nM) were incubated with 1 nM IdeS^C94S for 1 hour at 37°C in PBS pH 7.4, prior to the addition of 0. nM IdeS for an additional 6 hours. Samples were analyzed by SDS-PAGE using 12% gradient Bis-Tris protein Gel (NuPAGE, Thermosfisher) under non-reducing conditions followed by a coloration with Coomassie Blue.

The coding sequences for enhanced cyan fluorescent protein (eCFP) and Venus were obtained from Addgene (#105293 and #39813, respectively). The coding sequence for the hinge-CH2-CH3 of human IgG1 was obtained from Lecerf et al. (24). The gene encoding eCFP was fused with that encoding the Fc portion of human IgG1 from residue A231, containing two mutations at E356K and D399K. The gene encoding Venus was fused with that encoding the Fc portion of human IgG1 from residue A231, containing two mutations at K392D and K409D. The introduced mutations favor the heterodimerization of the eCFP-Fc/Venus-Fc construct by virtue of electrostatic steering (26). A GGGS linker was added between each fluorescent protein and the Fc fragment. A 6-his tag was added at the C-terminus of the eCFP-Fc monomer. The two cDNA were cloned independently in the pALL eukaryote expression vector under the control of the CMV promoter and using the murine IgG1 signal peptide (GenBank Accession Number DQ407610). The FRET substrate was expressed following co-transfection of HEK293 cells using the Expi293™ Expression System (Thermofisher). Occasionally, cells were transfected with each independent vector alone. The Venus-Fc expressed alone was purified on protein A-agarose (Cytiva). To this end, the supernatant was mixed with binding buffer (20 mM Na₂HPO₄, 150 mM NaCl pH 7). Venus-Fc was eluted with 0.1 M citric acid pH 3, neutralized with 1 M Tris pH 9. The eCFP-Fc/Venus-Fc FRET substrate and eCFP-Fc expressed alone were purified by immobilized metal ion chromatography. The supernatants were mixed with binding buffer (20 mM Na₂HPO₄, 150 mM NaCl, 20 mM imidazole pH 7) before purification and the recombinant proteins were eluted with elution buffer (20 mM Na₂HPO₄, 150 mM NaCl, 500 mM imidazole pH 7). All proteins were finally purified by size exclusion chromatography using Superdex 200 column against PBS to obtain only the heterodimer forms of the substrate or the monomeric form of the controls. The absorbance of the purified proteins was measured using a UV-Vis spectrophotometer (Agilent Technologies).

For dose-dependent digestion experiments, IdeS (0-4 µM) was incubated with the FRET substrate (2 µM) in PBS at 37°C for 45 min. For time-dependent digestion assays, IdeS (0.16 µM) and the FRET substrate (2 µM) were incubated in PBS at 37°C for 0, 5, 10, 20, 40, 80, 360 and 1440 minutes. Samples were then analyzed by SDS-PAGE 4-12% Bis-tris. When indicated, the emission spectra were recorded after excitation at 434 nm with a TECAN Infinite M200 Pro (Infinite^®, Tecan Trading AG, Switzerland).

For determination of the kinetic parameters that define the hydrolysis of the FRET substrate by IdeS, the fluorescence ratio F_{530 nm}/F_{484 nm} was calculated for the undigested substrate (R₀) and for the substrate incubated with 1 nM IdeS 24 hours at 37°C (i.e., fully digested substrate, R₂₄) following excitation at 434 nm. The FRET substrate (15 to 0.0625 µM) was then incubated with 1 nM IdeS for up to 380 sec with a reading of the fluorescence at 530 and 484 nm every 20 seconds. At each time point and for each substrate concentration, the fluorescence ratios F_{530 nm}/F_{484 nm} (R) were calculated. The concentrations of digested substrate were finally calculated using the ratiometric FRET analysis method as described in Liu et al. (27):

C=(R-R₀)/(R₂₄-R₀)xCi,

where Ci is the initial FRET substrate concentration. The initial velocity V₀ (µM/sec) was computed during the linear phase by using the fluorescence R ratio measured at 380 seconds. The K_m and V_max were calculated following plotting of the V₀ as a function of the substrate concentration, and fitting the experimental data to the Michaelis-Menten equation (Prism, version 9.4.1).

IdeS (8–24 nM) was pre-incubated alone, or with serial dilutions of plasma/serum, IVIg or purified IgG, in PBS-0.025% Tween 20, pH 7.4 for 1 hour at 37°C in MicroWell™ 96 well polypropylene black plates (Nunc). The FRET substrate (500 nM) was then added for 1 hour at 37°C in the dark. The reaction was stopped by addition of 20 mM iodoacetamide and incubation for 30 min at room temperature in the dark. The “residual IdeS activity” was calculated as the ratio of measured fluorescence ratio F_{530 nm}/F_{484 nm} for each condition over the fluorescence ratio measured in the absence of IdeS multiplied by 100. The percentage of IdeS neutralization was calculated as 100-”residual IdeS activity”.

The “clinically relevant IgG/IdeS ratio” used in the FRET IdeS neutralization assay was defined based on the available pharmacokinetic studies of therapeutic IdeS. Treatment of patients with therapeutic IdeS at 0.25 mg/kg (11, 13, 21, 28) yields a Cmax at the end of the injection of IdeS of 8.3 ± 3.7 µg/ml (10) (224 ± 100 nM, IdeS: 37 kDa). Hence, with a mean physiological concentration of circulating IgG of 13 mg/ml, the “clinically relevant IgG/IdeS ratio” corresponds to 13 mg/ml IgG for 224 nM IdeS. Accordingly, in the FRET neutralization assay, the “clinically relevant IgG/IdeS ratio” is reached when testing the serum or plasma at a 1:10 dilution with 24 nM IdeS.

For IgG purification, plasma/serum (25 µl) was diluted 1:10 in purification buffer (Pierce™ Melon™ Gel IgG Spin Purification Kit) and incubated with 100 µl of Melon Gel for 10 min at room temperature. Purified IgG was collected. For IgG depletion, plasma/serum samples (25 µl) were diluted 1:10 in PBS prior to incubation with 100 µl Protein A-coupled agarose beads (Pierce Protein A/G Agarose) for 20 min at room temperature. The IgG-depleted flow-through was collected by centrifugation. Samples were stored at -20°C until use. IgG was quantified in plasma/serum, in the purified IgG fraction and in the IgG-depleted flow through by ELISA using an unlabeled goat anti-human Ig kappa antibody (2.5 µg/mL; 2060–01 Southern Biotech) for capture, and an HRP-conjugated mouse anti-human IgG Fc (1/3000, 9040–05 Southern Biotech) for detection. A standard curve was established using serial dilutions of IVIg.

Data were analyzed using GraphPad Prism Software (San Diego, CA, USA). Outliers in both ELISAs and neutralization assay were retested at greater plasma/serum dilutions. The data were considered as not normally distributed. Statistical differences were assessed using the two-tailed non-parametric Mann-Whitney test. Correlations were assessed by two-tailed non-parametric Spearman correlation. P values below 0.05 were considered as statistically significant. Graphs depict data as means ± standard deviations, or as boxes and whiskers with the box depicting the median and 25th to 75th percentiles, and the whiskers depicting the min and max values.

We first confirmed the presence of anti-IdeS IgG in IVIg, a pool of normal human IgG. IVIg bound in a dose-dependent manner to the immobilized inactive IdeS^C94S variant by ELISA (Figure 1A). The binding of IVIg to immobilized IdeS^C94S was inhibited in a dose-dependent manner by soluble IdeS^C94S with an IC₅₀ of 108.6 ± 39.8 nM (Figure 1B). As controls, 7 human monoclonal IgG₁ that are not specific for IdeS exhibited no binding to IdeS (Figure 1C), indicating that the binding of IVIg to IdeS^C94S occurs through the Fab and not through the hinge-CH2CH3 domains in a substrate-like manner (29). We then examined the hydrolysis of IVIg by IdeS. We first used BO2C11, one of the human monoclonal IgG, as a control. BO2C11 IgG was entirely hydrolyzed into scIgG and F(ab’)₂ fragments after 6 hours at 37°C in the presence of 1 nM IdeS, the lowest concentration tested (Figure 1D, left panel, see Supplementary Figure S1 for original uncropped gel images). The scIgG was undetectable in the presence of 6 nM IdeS, indicating complete hydrolysis. In contrast, the hydrolysis of IVIg was incomplete in the presence of 1 nM IdeS, with a residual band of intact IgG and no generation of F(ab’)₂ (Figure 1D, right panel). Residual scIgG was still detectable after incubation in the presence of 8 nM IdeS. The complete hydrolysis of IVIg by 1 nM IdeS was however restored when IVIg was preincubated with IdeS^C94S prior to addition of IdeS (Figure 1E), suggesting saturation of the neutralizing anti-IdeS IgG fraction within IVIg. Conversely, addition of IdeS^C94S had no effect on IdeS-mediated BO2C11 hydrolysis. Taken together, the data indicate that IVIg contains anti-IdeS IgG and that the latter are able to neutralize the proteolytic activity of IdeS.

In order to quantify the neutralizing activity of anti-IdeS IgG in IVIg, we engineered a synthetic fluorescence resonance energy transfer (FRET) substrate for IdeS (Figure 2A; Supplementary Figure S2). The FRET substrate was cleaved by IdeS in a dose- and time-dependent manner, as shown by SDS-PAGE (Figures 2B, C, see Supplementary Figure S3 for original uncropped gel image), resulting in loss of energy transfer from eCFP to Venus as measured at 530 nm and gain in emission of eCFP at 484 nm (Figure 2D). The ratios of fluorescence intensities F₅₃₀/F₄₈₄ changed from 1.63 ± 0.26 before hydrolysis to 0.68 ± 0.03 (mean ± SD, n=4) after complete cleavage. To determine the initial velocity of the reaction, IdeS was incubated with different concentrations of FRET substrate. The F₅₃₀/F₄₈₄ fluorescence ratio was calculated after excitation at 434 nm every 20 seconds for each substrate concentration and was translated into µM of hydrolyzed substrate using the ratiometric method (Figure 2E). The initial velocity V₀ was calculated over the linear phase (i.e., 400 seconds) and plotted as a function of the initial substrate concentration (Figure 2F). Fitting the experimental data to the Michaelis-Menten equation allowed calculation of the V_max (0.012 µM/s), K_m (6.07 µM) and K_cat (12.25 s^-1), values that are similar to that previously determined for IdeS with different assays (Table 1). The FRET substrate was used to develop a functional neutralization assay (Figure 3A): IVIg (0.5–10 mg/ml) was incubated with IdeS (8–24 nM) for 1 hour prior to addition of the substrate. IVIg neutralized IdeS in a dose-dependent manner with ≥85% neutralization at the highest IgG concentration (Figure 3B). Conversely, IdeS overcame the neutralizing activity in a dose-dependent manner. Calculation of the IgG concentrations leading to 50% neutralization yielded 1.4, 5.7, 7.1 and 9.1 µg/ml for 8, 16, 20 and 24 nM IdeS, respectively (mean coefficient of variation of 6.2%, ranging from 0.3 to 23.9%, depending on the enzyme and IgG concentrations used). Taken together, these data confirm that IdeS-neutralizing IgG are present in pools of normal IgG.

In order to determine the prevalence of IdeS-binding IgG and IgA under physiological conditions, we first generated humanized monoclonal anti-IdeS IgG and IgA antibodies to be used as standards in our assays. An IdeS-specific IgG-producing B-cell hybridoma (clone 29, cl29) was obtained from Ides-immunized mice, and the VL and VH genes were cloned into human IgG or IgA expression vectors. The humanized anti-IdeS IgG^cl29 bound in a dose-dependent manner to both native IdeS and IdeS^C94S in an ELISA, and recognized IdeS in a western blot (Figure 4A, inset, see Supplementary Figure S4 for original uncropped gel image). The binding of IgG^cl29 was inhibited in a dose-dependent manner by soluble IdeS^C94S with an IC₅₀ of 60.0 ± 5.7 µg/ml for IdeS and 28.5 ± 20.3 µg/ml for IdeS^C94S (Figure 4B). The recombinant anti-IdeS IgA^cl29 bound in a dose-dependent manner to IdeS (Figure 4C). IgG^cl29 was used as a standard to quantify anti-IdeS IgG in IVIg (10 mg/ml) and yielded a value of 49.6 ± 3.8 µg/ml (0.33 ± 0.03 µM, coefficient of variation: 7.6%, Figure 1C).

We collected plasma and serum samples from 80 and 56 healthy donors, respectively. The cohort included 64% females and 36% males, with 47 (35%) individuals between 18 and 40 years of age, 45 (33%) individuals between 41 and 60 years of age, and 44 (32%) individuals between 61 and 85 years of age (Table 2). Using IgG^cl29 and IgA^cl29 as standards, we determined the concentrations of anti-IdeS IgG and IgA in the plasma or serum of the donors by ELISA. The limit of quantification (LOQ) for the IgG and IgA ELISA was calculated using 20 blank values and the 20 serum/plasma samples with the lowest values LOQ_IgG=2.34 µg/ml and LOQ_IgA=0.023 µg/ml (25).

Levels of anti-IdeS IgG and IgA were higher when measured in serum than when measured in plasma, which is consistent with the highest IgG concentration in serum (Supplementary Figure S5). In plasma, healthy donors presented with heterogenous levels of anti-IdeS IgG: 23.89 ± 22.03 µg/ml [min: 0.80 µg/ml-max: 131.30 µg/ml]; and anti-IdeS IgA: 0.993 ± 3.007 µg/ml [0.010-23.160 µg/ml] (Table 3). In serum, levels of anti-IdeS IgG were 36.91 ± 26.00 µg/ml [3.47-115.30 µg/ml] and levels of anti-IdeS IgA were 2.201 ± 4.595 µg/ml [0.0-25.740 µg/ml]. There was no difference in anti-IdeS IgG or IgA titers between males and females, or according to the age of the donors (Figures 5A–D; Supplementary Figures S6A, B). Five healthy donors (3 females and 2 males) had anti-IdeS IgG levels below the LOQ_IgG, and 20 donors (13 females and 7 males) had anti-IdeS IgA levels below the LOQ_IgA (Table 3). Hence, the prevalence of anti-IdeS IgG and IgA were 96.3% and 85.3% in our cohort, respectively. In the case of IgG, and as seen previously for IVIg and IgG^cl29, the binding of IgG to immobilized IdeS^C94S was inhibited in a dose-dependent manner in the presence of soluble IdeS^C94S, with IC₅₀ of 19.1, 10.3, 12.0 and 24.2 µg/ml, for donors 1, 2, 3 and 5, respectively (Figure 5E). Combining data from plasma and serum samples revealed a weak but statistically significant positive correlation between the levels of anti-IdeS IgG and IgA (R² = 0.068, P = 0.0022, Figure 5F).

We then investigated the presence of IdeS neutralizing antibodies in the plasma and sera from healthy donors. In this assay, we first used a concentration of 24 nM IdeS and a plasma/serum dilution of 1:2.5. Under such experimental conditions, 3 out of 80 plasma (2 females, 1 male, 3.75%) and 4 out of 56 sera (3 females, 1 male, 7.14%) had IdeS neutralizing activities >70% (Figures 6A, B: above the dotted line). IdeS neutralizing activity above 70% was detected for samples that had anti-IdeS IgG≥9.95 µg/ml or IgA≥0.05 µg/ml (Figures 6C, D), although most samples with IgG/IgA concentrations greater than the latter thresholds did not show IdeS neutralization, at least in the present experimental conditions. Of note, with a serum/plasma dilution of 1:10, only one sample showed IdeS neutralizing activities >70%. There was a positive correlation, albeit with poor goodness-of-fit, between the IdeS neutralizing activity and the levels of anti-IdeS IgG (R² = 0.042, P = 0.0164), or of anti-IdeS IgA (R² = 0.076, P = 0.0012).

To investigate whether the IdeS neutralizing activity is carried by the IgG or IgA fraction of plasma/serum, we purified IgG from 2 serum and from 2 plasma samples on protein G-agarose beads. The purification of the IgG and depletion from plasma/serum was confirmed by SDS-PAGE (Supplementary Figure S7). The starting material (plasma/serum), purified IgG and IgG-depleted fractions (flow-through) were then assessed for IdeS neutralizing activity using 8 nM IdeS. While IdeS neutralization was readily detected in plasma/serum and in the purified IgG fractions of the four donors, it was completely absent from the flow-through (Figure 6E). IgG-mediated neutralization of IdeS was dependent on IgG concentration (Figure 6E, inset).