Naturally Occurring Anti-Idiotypic Antibodies Portray a Largely Private Repertoire in Immune-Mediated Thrombotic Thrombocytopenic Purpura

This article is featured in Top Reads, p.

Life-threatening acute episodes of immune-mediated thrombotic thrombocytopenic purpura (iTTP) are the result of a severe deficiency of the von Willebrand factor (vWF)-cleaving protease ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 motifs, member 13) (1). The impaired enzymatic function of ADAMTS13 is the result of circulating autoantibodies, which either inhibit ADAMTS13 activity or increase ADAMTS13 clearance from the circulation (2, 3). Upon vessel wall damage, endothelial cells release ultra-large vWF multimers, which persist in the absence of ADAMTS13 and lead to adhesion and aggregation of platelets and ultimately to the formation of vWF platelet-rich thrombi and occlusion of the vasculature. The resulting end-organ ischemia accounts for morbidity and mortality of acute iTTP episodes (4).

Present-day treatment of acute iTTP episodes is built on three pillars: 1) daily therapeutic plasma exchange (TPE) with replacement of plasma to remove circulating anti-ADAMTS13 Abs (and acute phase reactants) and to replenish the missing protease (5–7); 2) suppression and restraining of the pathological immune responses with immune-suppressive drugs such as corticosteroids and rituximab (5, 6, 8), and as last resort splenectomy, which is effective in reducing the autoreactive B cell pool (9); and 3) targeting the vWF–platelet interaction by blocking the platelet-binding site in the vWF A1 domain with caplacizumab (10, 11), or by vWF size regulation by other means than ADAMTS13 (5–7). With modern treatment, mortality rates of acute iTTP episodes have been reduced to <5% in experienced treatment centers, although morbidity in survivors remains high (12, 13). The high risk of recurrence (at least 50–60%) (14, 15) following a first acute iTTP episode adds to the disease burden in survivors and impacts their quality of life (16–18).

How remission of an initial acute episode is achieved and factors contributing to the reemergence of anti-ADAMTS13 Abs and a relapsing course are still poorly understood. More than three decades ago, it was shown that the emergence of anti-idiotypic Abs neutralizing pathogenic inhibitory autoantibodies to coagulation factor VIII contributed significantly to the recovery and remission of patients with acquired hemophilia (19). I.v. Ig preparations of multiple healthy donors had similar properties to specific anti-idiotypic Abs, suggesting shared idiotype and anti-idiotypic networks in patients and healthy individuals (20). Such networks had been postulated as early as 1974 by Jerne (21). According to Jerne’s Ab network theory, among all naturally occurring Abs in the common immune repertoire, there are, in addition to the idiotypes, so called anti-idiotypic Abs that are able to counterbalance any immune response (22). Apart from acquired hemophilia, the existence of such idiotype–anti-idiotypic networks has been documented in autoimmune heart disease, systemic lupus erythematosus, membranoproliferative glomerulonephritis, myasthenia gravis, and type 1 diabetes (23–27).

In iTTP, the findings of highly similar anti-ADAMTS13 autoantibodies in two unrelated patients who shared H chain CDR3 motifs (28), as well as that pathogenic inhibitory anti-ADAMTS13 Abs of many unrelated iTTP patients are directed to the same epitope in the ADAMTS13 spacer domain (29, 30), point to an Ag-driven autoimmune response in iTTP. We hypothesized that a network of naturally occurring anti-idiotypic Abs to anti-ADAMTS13 autoantibodies might be relevant to achieve full biomarker remission from acute iTTP episodes and that a network imbalance might lead to the reemergence of circulating anti-ADAMTS13 autoantibodies, eventually resulting in a new acute iTTP episode. In this study, we sought evidence of an anti-idiotypic reservoir in iTTP patients to characterize potential anti-idiotypic Abs and to assess their functional properties.

Patients investigated in this study suffered from iTTP, defined as an acute thrombotic microangiopathy with thrombocytopenia (platelet count <100 × 10⁹/l), microangiopathic hemolytic anemia with schistocytes on the peripheral blood smear, and a severe ADAMTS13 deficiency (ADAMTS13 activity <10% of that in normal plasma) in the presence of a functional ADAMTS13 inhibitor and/or anti-ADAMTS13 Abs. To ensure readout in mixing studies only patients with a functional ADAMTS13 inhibitor of at least 1 Bethesda unit (BU)/ml were included.

Patients A, B, and C were splenectomized because of frequently relapsing iTTP disease courses and donated their spleens after written informed consent to our research. Patient A had her first iTTP episode at the age of 37 y and was splenectomized after the fourth iTTP episode 3 y later. Treatment of all four episodes had consisted of TPE with replacement of plasma and immunosuppression with corticosteroids. Following splenectomy, she has been in remission for >18 y. At presentation with the acute iTTP episodes, she had displayed strong functional ADAMTS13 inhibitors and high anti-ADAMTS13 Ab titers by ELISA (Supplemental Table I). Patient B had her first iTTP episode at the age of 9 y. She subsequently relapsed three times, at least twice when tapering immunosuppressants had been initiated. Treatment included TPE, corticosteroids, and, in addition, rituximab for her second, third, and fourth episodes. Six months after the last rituximab dose, when ADAMTS13 activity declined again, she was splenectomized. She has remained in remission for 15 y to date. At presentation with the acute iTTP episodes, she had displayed strong functional ADAMTS13 inhibitors and high anti-ADAMTS13 Ab titers by ELISA. Patient C experienced clinical episodes during the course of 15 y, with her first acute iTTP episode occurring at the age of 35 y. Splenectomy was undertaken after her fourth iTTP episode (age 50 y), and she has been in full remission for nearly 17 y to date. Treatment consisted of TPE and immunosuppression with glucocorticoids in all four episodes, and additionally vincristine in two episodes. ADAMTS13 parameters were evaluated only for her fourth acute episode, and they documented a severe ADAMTS13 deficiency in the presence of functional ADAMTS13 inhibitor of 1.5 BU/ml and an anti-ADAMTS13 Ab titer by ELISA of 33.0 arbitrary units/ml.

In addition, we used citrated plasma samples of 22 consecutive, unrelated iTTP patients withdrawn at presentation with an acute iTTP episode to examine the breadth of neutralizing potential of the anti-idiotypic Fabs. Patient samples were stratified into two cohorts of 11 samples each according to their functional ADAMTS13 inhibitor titers. Samples in cohort I had strong ADAMTS13 inhibitors of >2 BU/ml, whereas in cohort II functional ADAMTS13 inhibitors were 1–2 BU/ml.

Total IgG was purified from citrated plasma samples of three additional iTTP patients (patients 1–3; all having an ADAMTS13 activity <10% and a functional ADAMTS13 inhibitor of >2 BU/ml). Control IgGs were purified from standard human plasma (Siemens) and from a patient diagnosed with hereditary TTP (hTTP; patient 37.01 with baseline ADAMTS13 activity of 3.7%, a negative functional ADAMTS13 inhibitor, no anti-ADAMTS13 Abs by ELISA, and ADAMTS13 mutations c.530A>G [p.Y177C] and c.3178C>T [p.R1060W] [31]).

The study was approved by the Cantonal Ethic Committee Bern (KEK 031/06 and 123/2015).

Fourteen monoclonal anti-ADAMTS13 IgG₄ Fabs generated from splenic B cell repertoires of patients A and B and characterized previously (28) were used in this study.

Anti-ADAMTS13 IgG₄ Fabs were grouped according to their shared H chain CDR3 motifs into anti-ADAMTS13 Fabs motif 1 (Fabs 1a, 1e–g, 2c), anti-ADAMTS13 Fabs motif 2 (Fabs 2a, 2j, 4a), and anti-ADAMTS13 Fabs motif 3 (Fabs 1d, 1h, 2h, 2i, 3c, 3h). The anti-ADAMTS13 Fabs motif 1 and motif 2 use the IGHV1-3*01 gene; anti-ADAMTS13 Fabs motif 3 use the IGHV4-28*01 gene.

All anti-ADAMTS13 IgG₄ Fabs recognize an epitope in the ADAMTS13 spacer domain as evidenced by a dot-immunobinding assay on recombinant full-length ADAMTS13 and an ADAMTS13 fragment truncated after the spacer domain (MDTCS) and/or by epitope mapping using overlapping conformational peptides of the ADAMTS13 spacer domain (covering ADAMTS13 amino acid residues 556–685) and CLIPS (chemically linked peptides on scaffolds) technology (Pepscan Presto; Pepscan, Lelystad, the Netherlands). The dominant ADAMTS13 spacer domain epitope recognized by different anti-ADAMTS13 Abs of both patients A and B is ₅₉₅AVGRIGGRYVV₆₀₅ (group 1 Abs); some Abs recognize in addition residues ₅₆₄AGRAREYVTFL₅₇₄ and residues ₆₅₅IQVYRRYGEEY₆₆₅ (group 2 Abs). Finally, we included three Abs recognizing a unique epitope in the ADAMTS13 spacer domain (group 3 Abs). All monoclonal anti-ADAMTS13 IgG₄ Fabs, except Fab 1g, are strong functional inhibitors of ADAMTS13 activity (28).

After splenectomy, the whole spleens were kept in DMEM supplemented with 5% FCS, 5000 U/ml heparin, and 0.05 mg/ml DNase (Roche). Cells were scrapped from the splenic tissue and filtered through a strainer and a 100-µm filter. The cell suspensions were washed with 0.9% NaCl and splenic mononuclear cells (SMCs) were isolated by size gradient separation (Lymphoprep; Fresenius Kabi Norge). Purified SMCs were resuspended 1:1 (v:v) in 5% human albumin and freezing media, consisting of 20% dextran T70 in Hanks’ solution and 10% dimethyl sulfoxide (WAK-Chemie Medical), which were adjusted to 150–200 × 10⁶ cells per ml. Freezing was performed stepwise, that is, 2 h at −20°C followed by −80°C overnight prior to long-term storage at −150°C.

RNA was isolated from thawed SMCs using the RNeasy mini kit (Qiagen) and reverse transcribed to synthesize first-strand cDNA with the use of SuperScript III reverse transcriptase and oligo(DT)₂₀ primers (50 µM) (Invitrogen), according to the manufacturers’ protocols.

For the construction of complete IgG₁ Fab libraries of patients A and C (Supplemental Fig. 1), all gene family members encoding the IgG₁ κ/λ repertoire were amplified in three subsequent PCR rounds as previously published (32). Briefly, in a first PCR round all variable (V_H, V_κ, and V_λ) and constant (C_H1, C_κ, and C_λ) products were separately amplified with forward and reverse primers having sequence tails, creating overlap sequences later used for the assembly of the different PCR products to whole Fab sequences. The second PCR round assembled by overlap extension either V_H and C_H1 to form Fd fragments or V_κ/V_λ and C_κ/C_λ to create full-length κ/λ L chains, respectively. The H chain Fd fragments were combined with the κ/λ L chains to create full Fab PCR products in the third PCR round.

The Fab PCR products were cloned into the plasmid pComb3XSS (gift of Carlos Barbas; Addgene plasmid 63890; http://n2t.net/addgene:63890; RRID:Addgene_63890) (33) by SfiI restriction digest providing sticky overlaps on the 5′ and 3′ ends of DNA. Overnight ligation of the cut plasmid (pComb3X) with the digested and gel-purified full-length Fab DNA fragments was done by T4 DNA ligase (Promega) at room temperature. A plasmid without insert was used as background ligation control. XL1-Blue electroporation-competent cells (Agilent Technologies) were transformed with the ligated libraries using a Gene Pulser (Bio-Rad; electroporation 1.7 kV) and SOC medium (Invitrogen) for outgrowth. The propagated plasmid DNA containing one patient’s full IgG₁ κ or λ library was extracted using the QIAprep spin miniprep kit (Qiagen).

Phage display panning was done according to the protocol of Barbas et al. (34). Briefly, cloned IgG₁ κ/λ Fab libraries were amplified by XL-1 transformation, infected with VCSM13 helper phage (1 × 10¹² CFU), and selected in the presence of antibiotics (antibiotic resistances: pComb3X, ampicillin; XL-1 Blue, tetracycline; VCSM13, kanamycin). Phages were precipitated with 20% polyethylene glycol/2.5 M NaCl, and resuspended in PBS-BSA at 10 mg/ml. Panning was done on pooled anti-ADAMTS13 IgG₄ Fabs motif 1/2, or pooled anti-ADAMTS13 IgG₄ Fabs motif 3 coated at 1 µg/ml per Fab on 96-well plates (Supplemental Fig. 2). Resuspended phages were added to the preblocked wells, and phages binding to anti-ADAMTS13 IgG₄ Fabs were eluted and amplified during three to five panning rounds. For patients (A/C), both libraries (κ/λ) were panned on anti-ADAMTS13 IgG₄ Fabs motif 1/2 and motif 3 pools, resulting in eight screened and enriched Fab libraries.

Library phage DNA was digested after the last round of panning by double restriction enzyme digest (NheI and SpeI, New England Biolabs) eliminating the gene III product (pIII) whose anchors expressed Ab fragments to the phage surface. Religated phagemid DNA was transformed into XL1-Blue electroporation-competent cells and plated on ampicillin agar. Single colonies were induced with isopropyl β-d-1-thiogalactopyranoside (IPTG; Roche) for crude Fab production.

One Shot BL21 Star (DE3) chemically competent Escherichia coli (Invitrogen) was transformed with phagemid DNA of single clones and induced with 1 mM IPTG overnight at 15°C and shaking at 200 rpm. Pelleted cultures were lysed with 4 ml/g bacterial protein extraction reagent (Thermo Fisher Scientific) substituted with 5 U/ml DNAse I, 0.1 mg/ml lysozyme (both Thermo Fisher Scientific), and one tablet per 10 ml of cOmplete EDTA-free protease inhibitor cocktail (Roche). Lysates were concentrated using Amicon Ultra-15 centrifugal units (10 kDa nominal molecular mass limit; Merck Millipore).

Fabs were purified from concentrated lysates using protein G–Sepharose 4 fast flow resin (GE Healthcare Life Sciences), and batch method purification was performed according to the manufacturer’s instructions (Thermo Fisher Scientific), followed by dialysis to TBS (pH 7.4) using Slide-A-Lyzer dialysis cassettes (10 kDa molecular mass cut-off; Thermo Fisher Scientific). Yields were assessed using the Pierce BCA (bicinchoninic acid) protein assay kit (Thermo Fisher Scientific).

Phagemid DNA of the Fab fragments was Sanger sequenced using primers SeqGz (H chain, 5′-GAAGTAGTCCTTGACCAG-3′), SEQKb (L chain κ, 5′-ATAGAAGTTGTTCAGCAGGCA-3′), and SEQLb (L chain λ, 5′-CAAGTCACTTATGAACACAC-3′) (Microsynth). Obtained sequences were aligned to the most homologous IGHV and IGK/LV germline sequences using the IMGT/V-QUEST database (35).

Single Fab clones were coded in the form XY_Z. Z is an arbitrary individual clone number, and X represents the patient (A or C), and Y represents the κ or λ L chain library origin and selection CDR3 motif. Y is defined as follows: 1) κ, motif 1/2; 2) κ, motif 3; 3) λ, motif 1/2; and 4) λ, motif 3. Thus, the label A1_4 describes an individual Fab from patient A’s κ L chain library selected on anti-ADAMTS13 IgG₄ Fabs motif 1/2, with the individual clone number 4.

Equimolar pools of the selecting inhibitory anti-ADAMTS13 IgG₄ Fabs (CDR3 motif 1: 1a, 1e, 1f, 2c [25.0 nM per Fab]; motif 2: 2a, 2j, 4a [33.3 nM per Fab]; motif 3: 1d, 1h, 2h, 2i, 3c, 3h [16.6 nM per Fab] [28]), unrelated Ab fragments (anti-HIV Fab b12 [36], gift of Dennis Burton, The Scripps Research Institute, La Jolla, CA; or IgG Fc fragment; both 100 nM), and control Ags (human serum albumin and BSA [100 nM]; human serum and plasma; 1:100 [v:v] in PBS) were coated on Nunc 96-well plates. After blocking each anti-idiotypic Fab (100 nM) and equimolar anti-idiotypic Fab pools A (four Fabs, A1_4, A1_5, A1_8, A2_17 [25 nM per Fab]) or C (four Fabs, C1_3, C1_7, C3_44, C4_59 [25 nM per Fab]) were incubated for 2 h at 37°C. Bound monoclonal anti-idiotypic Fabs were detected with a mouse anti-human IgG₁ (Fab specific) (Sigma-Aldrich) followed by HRP-labeled goat anti-mouse IgG (Kierkegaard and Perry Laboratories) and developed with HRP substrate (Mabtech). The reaction was stopped by the addition of 3 M sulfuric acid, and the absorbance read at 450 nm and was corrected for baseline absorbance at 620 nm.

Anti-ADAMTS13 IgG₄ Fabs 1a, 1e, 1f, and 2c (of CDR3 motif 1; 200nM [28]) were pooled and immobilized at 2200 response units on a Biacore CM5 sensor chip (GE Healthcare Life Sciences) using 10 mM sodium acetate buffer (pH 5.0). Binding of eight single anti-idiotypic Fabs (A1_4, A1_5, A1_8, A2_17; and C1_3, C1_7, C3_44, C4_59) were assessed on a Biacore X-100 system (GE Healthcare Life Sciences).

Single anti-idiotypic Fabs were serially diluted from 500 to 31.5 nM in HBS-EP+ (GE Healthcare Life Sciences), and their binding kinetics were analyzed using 120-s contact time, 240-s dissociation time, and a flow rate of 20 µl/min.

ADAMTS13 activity (normal range 51 to >100%) was determined by the slightly modified fluorescence resonance energy transfer system (FRETS)-VWF73 assay as previously described (37, 38). Functional ADAMTS13 inhibitors (normal <0.4 BU/ml) were assessed in a Bethesda-like assay by preincubation of heat-inactivated (30 min at 56°C) patient’s plasma with standard human plasma (source of ADAMTS13) 1:1 (v:v) for 2 h at 37°C before measuring residual ADAMTS13 activity in the mixture (38, 39). One BU/ml is defined as the amount of inhibitor that results in 50% residual ADAMTS13 activity in a test mixture. Inhibitor titers were reported in BU/ml up to 2 BU/ml. Above 2 BU/ml, titers were reported as >2 BU/ml when residual ADAMTS13 activity was 11 to <25%, and as ≫2 BU/ml when residual ADAMTS13 activity was ≤10% (14).

Inhibitory anti-ADAMTS13 IgG₄ Fabs were pooled equimolar (200 nM) according to their CDR3 motifs and mixed with equimolar (200 nM) anti-idiotypic Fab pools A or C, or an equal volume of buffer control. The mixtures were preincubated for 1 h at 37°C. Then, standard human plasma was added 1:1 (v:v) and the mixtures incubated for another hour at 37°C, before residual ADAMTS13 activity was assessed.

Heat-inactivated plasma samples of 22 unrelated iTTP patients, grouped according to their functional ADAMTS13 inhibitor titers (cohort I, >2 BU/ml; cohort II, 1–2 BU/ml) were preincubated with 200 nM anti-idiotypic Fab pool A, anti-idiotypic Fab pool C, or buffer control for 1 h at 37°C. After addition of 1 vol of standard human plasma, the mixtures were incubated for another hour at 37°C, before residual ADAMTS13 activity was measured.

Standard human plasma and plasma samples of three iTTP patients (patients 1–3) and one hTTP patient (37.01) were heat-inactivated and subjected to protein G–Sepharose 4 fast flow resin and batch method purification. Purified total IgGs were dialyzed to TBS (pH 7.4) using Slide-A-Lyzer dialysis cassettes, and yields were assessed using the Pierce BCA protein assay kit.

Equimolar anti-idiotypic Fab pools A or C, rADAMTS13, or monoclonal anti-HIV Fab b12 was coated at 100 nM (5 µg/ml) on Nunc 96-well plates. After blocking, purified total IgG at a concentration of 200 or 10 µg/ml was added and incubated for 2 h at 37°C. Bound IgG was detected using a goat anti-human IgG–alkaline phosphatase-labeled Ab (Mabtech) and visualized with p-nitrophenyl phosphate (Mabtech), showing an absorbance read at 405 nm.

Four individual IgG₁-Fab phage libraries were generated with either κ or λ L chains from SMCs of patients A and C. These underwent phage display selection on anti-ADAMTS13 IgG₄ Fab motif 1/2 or motif 3 pools, resulting in eight anti-idiotypic Fab phage libraries. The size of the original libraries was 2.0 × 10⁸ CFU/ml (κ L chain) and 5.0 × 10⁸ CFU/ml (λ L chain) in patient A, compared with patient C where the sizes were 1.6 × 10⁹ CFU/ml (κ) and 1.5 × 10⁹ CFU/ml (λ). From the initially eluted anti-idiotypic phages, on average a 10- to 100-fold amplification (range 10⁴ to 10⁶ CFU/ml) over five (patient A, smaller original libraries) and three (patient C) rounds of selection was achieved. These amplifications were considerably lower than amplifications typically seen for targeted phage display Fab libraries in autoimmune diseases (amplification to 10⁶–10⁹ CFU/ml) (28, 40, 41) or viral infections (amplification to 10⁹–10¹¹ CFU/ml) (42, 43). It is noteworthy that the anti-ADAMTS13 IgG₄ Fab pools used for selection contained Abs originating from the splenic anti-ADAMTS13 repertoire of patient A, whereas those for patient C bore no relationship to the selecting Abs.

Of all 105 phage display-derived Fab clones (patient A, n = 45; patient C, n = 60), the 14 clones with the highest absorbance signal in an anti-Fab ELISA per patient were selected (an even distribution between the libraries was respected and resulted in three to four clones per library) and their phagemid DNA was sequenced. Twenty-seven of the 28 analyzed anti-idiotypic Fab clones revealed a productive H chain when aligned to the most common homologous germline gene in the IMGT/V-QUEST database (35). The anti-idiotypic Abs of patient A (n = 13) were derived from 8 genes, and those from patient C (n = 14) were derived from 11 different IGHV genes (Table I). The most abundant IGHV genes of the anti-idiotypic Fabs were V_H3-23 (n = 8; 29.6%), V_H3-15 (n = 4; 14.8%), and V_H1-69 (n = 3; 11%). Among IGHD genes, D_H1-26 (n = 5; 18.5%) and D_H3-22 (n = 3; 11%) prevailed, and among IGHJ genes, J_H4*02 (n = 12; 51.9%) and J_H5*02 (n = 9; 33.3%) prevailed.

The two patients shared four IGHV genes (V_H1-69*01, V_H3-15*01, V_H3-23*01, V_H3-49*03), two IGHD genes (D_H3-9*01, D_H3-22*01), and two IGHJ genes (J_H4*02, J_H5*02).

Of the 27 clones, 7 (4 of patient A and 3 of patient C) were paired with a productive L chain; in six instances these L chains were of type κ, whereas the seventh was of type λ. The two patients shared the IGKV1-39*01 gene (Table I).

The V region homology to the closest germline gene was on average 97.2% (range 90.1–100%) for the H chain and similar in both patients, with an average of 97.5% (range 91.3–100%) in patient A, and of 96.8% (range 90.1–98%) in patient C. The V region homology for the L chains was 91.2% (range 83.5–96.1%) Overall, we thus observed a low somatic hypermutation rate for both chains.

The anti-idiotypic Fab IGHV CDR3 regions most relevant for Ag binding were composed of 7–24 aa. CDR3 length ranged from 14 to 24 aa in patient A, and from 7 to 17 aa in patient C. We observed certain interpatient and intrapatient CDR3 similarities, with 16–35% homology when comparing clones employing the same IGHV genes.

In addition, we identified clones with identical amino acid sequences (including the CDR3) in the repertoire of patient A. Some of these clones were derived from the same library and selected on the same anti-ADAMTS13 CDR3 motif such as clones A2_18 and A2_19, or clones A4_48 and A4_51; however, clones A3_35, A4_48 and A4_51 were derived from the same library but selected on different anti-ADAMTS13 CDR3 motifs (Table II). A similar result was observed for closely related clones C1_7 and C2_16 with identical CDR3 sequences but different mutations in framework regions derived from a single library but selected on different anti-ADAMTS13 CDR3 motifs (Table II).

Six purified single anti-idiotypic Fabs (A1_4, A1_8, C1_3, C1_7, C3_44, C4_59) specifically bound to coated pools of anti-ADAMTS13 Fabs of the CDR3 motif that they had been selected on, but not to anti-ADAMTS13 Fabs of the other motif (Fig. 1A). No cross-reactivity of pooled purified anti-idiotypic Fabs (pool A: A1_4, A1_5, A1_8, A2_17; pool C: C1_3, C1_7, C3_44, C4_59) with unrelated Ab fragments (anti-HIV Fab b12 or human IgG Fc fragment) or control Ags (human serum albumin and BSA, human serum, and human plasma) was observed (Fig. 1B).

Surface plasmon resonance of single anti-idiotypic Fabs toward immobilized pooled anti-ADAMTS13 Fabs motif 1 revealed Ab affinities of 1.6 × 10⁻⁶ to 4.2 × 10⁻⁷ K_D, which is considerably lower than that determined for single anti-ADAMTS13 Fabs toward their Ag, ADAMTS13 (these had K_D ranging from 2.5 × 10⁻⁸ to 46 × 10⁻⁸) (Table III). Whereas clone C4_59, selected on anti-ADAMTS13 Fabs motif 3, showed no binding to anti-ADAMTS13 Fabs motif 1/2 in the static ELISA, this Fab demonstrated binding to the anti-ADAMTS13 Fabs motif 1 pool under surface plasmon resonance flow conditions. However, association (k_a of 2.84 × 10³) and dissociation (k_d of 78.6 × 10⁻³) were slower and faster, respectively, than showed by the other five anti-idiotypic Fabs selected on ADAMTS13 Fabs motif 1/2 (k_a of 4.79 × 10³ to 30.6 × 10³/k_d of 4.25 × 10⁻³ to 7.57 × 10⁻³) (Table III).

As previously reported, anti-ADAMTS13 Fabs of CDR3 motif 1 are strong inhibitors of ADAMTS13 activity (28). In the current modified Bethesda assay, their measured residual ADAMTS13 activity was 1%, corresponding to an inhibitor titer of ≫2 BU/ml. Preincubation with anti-idiotypic Fab pool A or C resulted in full inhibitor neutralization and restoration of ADAMTS13 activity in the mixtures (measured residual ADAMTS13 activity of 99% for pool A and of 95% for pool C) (Fig. 2).

The inhibitory capacity of anti-ADAMTS13 Fabs motif 2 was reduced compared with motif 1 Fabs, and residual ADAMTS13 activity in the modified Bethesda assay was 17% (corresponding to a functional ADAMTS13 inhibitor of >2 BU/ml). Again, preincubation with anti-idiotypic Fab pool A or C resulted in full inhibitor neutralization and restoration of ADAMTS13 activity (measured residual ADAMTS13 activity of 100% for pool A and of 113% for pool C) (Fig. 2).

Experiments with anti-ADAMTS13 Fabs motif-3 demonstrated residual ADAMTS13 activity of 1% (functional ADAMTS13 inhibitor ≫2 BU/ml), and again full restoration of ADAMTS13 activity after preincubation with anti-idiotypic Fab pools A and C (measured residual ADAMTS13 activity for both 97%).

Plasma samples withdrawn at presentation of an acute episode from 22 iTTP patients were used then to investigate a more general neutralization potential of the anti-idiotypic Fab pools A and C (Fig. 3). Samples were grouped according to their functional ADAMTS13 inhibitor titers (cohort I, >2 BU/ml; cohort II, 1–2 BU/ml). Preincubation with anti-idiotypic Fab pool A led to neutralization of the ADAMTS13 inhibitor below the clinically relevant threshold (0.4 BU/ml) in 27% (3/11) of patients in cohort I, and in 18% (2/11) of patients in cohort II. The corresponding numbers for preincubation with anti-idiotypic Fab pool C were 45% (5/11) and 36% (4/11) for patients in cohorts I and II, respectively. Somewhat surprisingly, the inhibitor titers in the nonresponding patients remained completely unchanged after preincubation with anti-idiotypic Fab pools (Fig. 3). Consequently, the anti-idiotypic Fab pools showed no general neutralization potential toward pathogenic ADAMTS13 inhibitors of unrelated iTTP patients, and only some of them (22.7% for anti-idiotypic Fab pool A, 40.9% for anti-idiotypic Fab pool C) showed a response.

To explore possible reasons of a nonresponse, the total IgG purified from plasma withdrawn during acute iTTP episodes of three additional nonresponding patients (1–3; functional ADAMTS13 inhibitor >2 BU/ml in all of them) was assessed for binding to and neutralization potential of anti-idiotypic Fab pools. Total IgG of all three iTTP patients bound in ELISAs on coated anti-idiotypic Fab pools A and C, on rADAMTS13, but not on the control anti-HIV Fab b12 (Fig. 4). Purified total IgG from standard human plasma and from an hTTP patient on regular plasma prophylaxis (patient 37.01) showed some binding to anti-idiotypic Fab pools A and C, but not to rADAMTS13 or to the anti-HIV Fab b12.

Binding to anti-idiotypic Fab pools A and C was observed at both IgG concentrations tested (200 and 10 μg/ml) for all three iTTP patients as well as standard human plasma; binding of IgG of hTTP patient 37.01 at the lower concentration, however, was minimal.

The binding of anti-ADAMTS13 Fabs motif 1, motif 2, or motif 3 pools on anti-idiotypic Fab pools A or C was higher than that of purified IgG of iTTP patients (on average 1.6-fold), of standard human plasma (2.5-fold), and of hTTP patient 37.01 (10-fold).

Despite similar binding properties of iTTP patients’ purified total IgG and of inhibitory anti-ADAMTS13 Fabs in the ELISA system, preincubation of purified total IgG of the iTTP patients with anti-idiotypic Fab pools resulted in no ADAMTS13 inhibitor neutralization (data not shown).

Idiotype–anti-idiotypic Ab networks play an important role in the regulation of pathogenic immune responses in autoimmune diseases such as systemic lupus erythematosus with cross-reactions of anti-idiotypic Abs and pathogenic autoantibodies (24), or acquired hemophilia A, where recovery from acute episodes was dependent on the presence of anti-idiotypic Abs against anti-FVIII autoantibodies (19). In this study, we demonstrated that naturally occurring anti-idiotypic Abs against anti-ADAMTS13 autoantibodies are part of the splenic Ab repertoire in iTTP patients.

Using SMCs of two patients splenectomized after several acute iTTP episodes and phage display technology, we isolated 27 distinct anti-idiotypic Fabs. Thirteen belonged to patient A, who had contributed to the anti-ADAMTS13 Ab pools used during phage display to select the anti-idiotypic Fabs, and 14 originated from the repertoire of patient C, who bore no relationship to the selecting anti-ADAMTS13 Abs, which were thus “foreign” to this patient. The monoclonal anti-ADAMTS13 Abs of patients A and B used for anti-idiotypic selection were chosen to reflect the polyclonal nature of the autoimmune response in iTTP but also to potentially produce a laboratory readout, that is, to select anti-idiotypic Abs able to neutralize functional ADAMTS13 inhibitors that typically recognize epitopes in the ADAMTS13 spacer domain (28).

Interestingly, the H chain repertoire of the anti-idiotypic Fabs of patients A and C used 15 different IGHV genes, 4 of which were used to generate Abs by both patients, and were thus shared. This is clearly different from what we observed in the splenic anti-ADAMTS13 immune repertoire of patients A and B, where the IGHV gene usage in 29 anti-ADAMTS13 Abs was restricted to only four different gene families (V_H1-3, V_H1-69, V_H3-30, and V_H4-28), which were used by both patients. Of these latter IGHV genes, only V_H1-69 was used to generate anti-idiotypic Fabs. The anti-idiotypic Fabs of patient A were less diverse than those of patient C, with 8 compared to 11 different IGHV genes used, respectively.

The variety of an individual’s Ab repertoire, created by VDJ recombination and somatic hypermutation, is massive (estimated 10¹² unique Abs), with most Ab VDJ gene recombinations being unique or private, and only a limited proportion that is shared between persons and patients (Ref. 44 and R. Arora and R. Arnaout, manuscript posted on bioRxiv, DOI: 10.1101/2020.06.18.159699). Encountering the same Ag may result in a restricted Ag-driven immune response reflected in shared VDJ genes among individuals in response to certain viral Ags but also in autoimmune diseases, as we and others have observed in iTTP (28, 45–47). We encountered highly similar anti-ADAMTS13 Abs with shared CDR3 sequences and an average somatic hypermutation rate of 12% in iTTP patients A and B (28). The somatic hypermutation rate was on average 3% and thus much lower in anti-idiotypic Fabs. It was minimal (<2%) in 15 of 27 anti-idiotypic Fab clones, including 3 clones with no IGHV mutation at all, suggesting that more than half of the anti-idiotypic Fabs were in their germline configuration (48). de Mattos Barbosa et al. (49) showed that for the immune response toward SARS-CoV-2, somatic mutations in IGHV genes of Abs directed toward the main epitope (spike protein of SARS-CoV-2) increased with neutralizing Ab titers, thus linking affinity maturation with an increased neutralization potential. The same was documented for highly mutated neutralizing HIV Abs, where canonical broadly neutralizing Abs had a substantially reduced neutralizing ability when their sequences were reverted to their germline configuration (50), or in pemphigus vulgaris, an autoimmune disease of the skin and mucosa manifesting in extensive blister formation, where pathogenic anti–desmoglein 3 Abs lost their blistering capacity when reverting to their germline configuration (51). The low IGHV mutation rate (average 3%) of the anti-idiotypic Abs in iTTP might thus be responsible for the “non-broadly” neutralization effect observed toward ADAMTS13 inhibitors in plasma of unrelated iTTP patients.

The enrichment of the anti-idiotypic Fab phage libraries was 100- to 10⁶-fold less than that of autoimmune anti-ADAMTS13 Fab phage libraries (10- to 100-fold versus 1000- to 10⁷-fold) (28). In conjunction with the low somatic hypermutation rate, the binding properties of anti-idiotypic Fabs were clearly weaker than those of the anti-ADAMTS13 Fabs for their respective Ags. Surface plasmon resonance studies of anti-idiotypic Fabs to anti-ADAMTS13 Abs revealed affinities in the range of K_D 10⁻⁶ to 10⁻⁷, compared with K_D of 10⁻⁸ to 10⁻⁹ for those observed of anti-ADAMTS13 Fabs for ADAMTS13 (28).

The Fab libraries of patients A and C contained a “public” (shared) and a “private” (individual) set of anti-idiotypic Abs. As the selection of anti-idiotypic Fabs during phage display was performed using anti-ADAMTS13 Abs of patients A and B, we had expected a larger public set of anti-idiotypic Fabs in patient C than in patient A. The higher percentage of plasma ADAMTS13 inhibitors neutralized by patient C’s anti-idiotypic Fab pool is in line with this expectation. Whereas the anti-idiotypic Fab pool of patient A was able to fully neutralize the functional ADAMTS13 inhibitors in 18% of cases with an ADAMTS13 inhibitor of 1–2 BU/ml and in 27% of cases with an ADAMTS13 inhibitor of >2 BU/ml, the corresponding numbers for patient C were 36 and 45%, respectively. Although nearly all patients have anti-ADAMTS13 Abs, recognizing a primary epitope in the ADAMTS13 spacer domain, two thirds of patients have in addition Abs recognizing epitopes in other ADAMTS13 domains. Such latter Abs were not present in our selecting anti-ADAMTS13 Abs.

The larger set of private anti-idiotypic Abs observed in our patients may explain the limited generalizability of their ADAMTS13 inhibitor neutralization potential. Additionally, genomic variations in the IGHV genes as well as posttranslational modifications, for example, glycosylation, have a profound effect on the effectivity of a person’s immune response to a specific Ag and may thus be largely individual (52).

Anti-idiotypic Abs may be directed against paratope-irrelevant regions on the idiotypes, or they may recognize the paratope and thus resemble the Ag (in our study ADAMTS13), thereby competing in binding with the Ag (22). In accordance with this, purified IgG Abs from iTTP patients 1, 2, and 3 bound effectively to the pooled anti-idiotypic Fabs in the absence of ADAMTS13 (ELISA). However, the inhibitory capacity of purified IgG remained unchanged by addition of pooled anti-idiotypic Fabs in the presence of plasma ADAMTS13 (FRETS-VWF73). Of note, in plasma of ∼5% of healthy individuals’ anti-ADAMTS13 IgG Abs, sharing epitopes on ADAMTS13 with ADAMTS13 IgG Abs of iTTP patients have been documented (53). This is consistent with our observation that purified IgG from standard human plasma obtained from several healthy donors showed weak binding on rADAMTS13 and the anti-idiotypic Fab pools.

In mice, anti-idiotypic Abs generated against anti-ADAMTS13 Abs were directed against epitopes (or idiotopes) both in the V and C region of anti-ADAMTS13 Abs (54), and in anti–vesicular stomatitis Indiana virus (VSV) Abs (55). As Fab H and L chains are composed of V and C regions, it is likely that some of the anti-idiotypic Fabs were selected against the C region (paratope-irrelevant region), rather than the paratope of the anti-ADAMTS13 Fabs. This is most likely the case for clones deriving from the same library having identical sequences but selected on different anti-ADAMTS13 CDR3 motifs (i.e., A3_35 and A4_48/A4_51, or C1_7 and C2_16). Selecting anti-ADAMTS13 Abs were of the IgG₄ subclass, which have a different C region than IgG₁, the subclass of our negative control (anti-HIV Fab b12) to which the mentioned anti-idiotypic Fabs did not bind.

Low expression levels were the reason why only a limited set of anti-idiotypic Fabs could be analyzed. For the same reason, anti-idiotypic Fab pools rather than single anti-idiotypic Fabs were used for the neutralization assays. Low affinity under flow conditions of surface plasmon resonance as well as binding to Ab regions other than to the anti-ADAMTS13 paratope are limitations of this first set of anti-idiotypic Abs. Attempts to select more potent anti-idiotypic Fabs with more stringent phage display selection protocols with double recognition panning, or negative selection panning (34) on purified IgG from plasma of a patient suffering from hTTP, failed and showed no enrichment of specific anti-idiotypic Fabs after the second panning round (data not shown).

In our attempt to identify anti-idiotypic Abs, we used SMCs of relapsing patients treated with immunosuppressant drugs in the past for and following their acute iTTP episodes. It is thus very well possible that they both were not able to mount a strong anti-idiotypic response. More suitable candidates for future studies would be nonrelapsing iTTP patients, and iTTP patients who required less intense immunosuppressive regimens to achieve remission. An assay to screen plasma samples of survivors of a first acute iTTP episode for the presence of an anti-idiotypic immune response to easily identify candidates for study would be an asset. Given the usually strong affinity of anti-ADAMTS13 Abs for ADAMTS13, abundantly present in plasma of patients in remission, such an assay is difficult to set up.

To summarize our findings, the polyclonal nature of the anti-idiotypic immune response in conjunction with the observed moderate IGHV somatic mutation rate and the overall low affinity of the anti-idiotypic Fabs suggest that phage display technology, which depends on affinity selection, is a suboptimal approach to select specific anti-idiotypes.

In other autoimmune diseases, anti-idiotypic Abs are part of regulatory networks, but the distinct role of these networks is not yet fully understood. Balanced networks where anti-idiotypic Abs keep autoantibodies in check, for example, during disease remission, can be modulated in either direction. An example is neonatal lupus, where the administration of i.v. Ig was demonstrated to modulate the Ab network in favor of the anti-idiotypic Abs, thereby improving the outcome (23). We hypothesize that during episodes of iTTP, anti-ADAMTS13 autoantibodies dominate the anti-idiotypic Abs not only in numbers, but also in binding properties. The potential neutralization effect of the anti-idiotypic Abs is undermined, and the resulting imbalance is responsible for the overt acute disease episode.

In conclusion, we report the presence of naturally occurring anti-idiotypic Abs against pathogenic anti-ADAMTS13 Abs in the immune repertoire of iTTP patients. This predominantly private repertoire is characterized by a low somatic hypermutation rate and displays only moderate neutralization potential toward anti-ADAMTS13 Abs of other iTTP patients. Understanding the definite role of anti-idiotypic Abs in remission and recurrence of acute episodes of iTTP requires further study.

For a more universal anti-idiotypic neutralization of autoantibodies, in prospect of novel treatment approaches, synthetic anti-idiotypic molecules with improved binding properties might be an essential additional choice.

We thank the patients for donating their spleens that made our research possible. We acknowledge the expert technical assistance of Isabella Aebi-Huber, Irmela Sulzer, and Schangavy Thamotharampillai, and the valuable input of Dr. Monique Vogel (Department of Immunology, Inselspital Bern). We thank Kateri A. Chambers for expert editorial assistance.

J.A.K.H. is a member of the advisory board of Takeda for the development of recombinant ADAMTS13, and of Ablynx, now part of Sanofi for the development of caplacizumab. The international hereditary TTP registry receives support through an investigator-initiated research grant (H16-36165) from Baxalta US Inc., member of Takeda group of companies, Bannockburn, IL. The other authors have no financial conflicts of interest.