Thrombotic thrombocytopenic purpura (TTP) is a rare and severe life-threatening autoimmune disorder which is mostly acquired by adult patients with an annual incidence of approximately 4 cases per million people (1). The patients suffer from systemic clumping of platelets in different organs due to the lack of activity of the metalloprotease ADAMTS13 (a disintegrin and metalloprotease with a thrombospondin type 1 motif, member 13) (1–3). The metalloprotease ADAMTS13 is responsible for the rapid processing of newly released von Willebrand factor (VWF) (4, 5). VWF is synthesized in endothelial cells and upon stimulation ultra-large VWF (UL-VWF) polymers are released in the systemic circulation. UL-VWF unfolds under high shear stress and stretches into string-like structures which enables ADAMTS13 to cleave the unfolded UL-VWF polymers thereby preventing spontaneous thrombi formation in the microvasculature (6, 7).

The diagnosis of immune TTP (iTTP) is not only based on the deficiency of ADAMTS13 activity but also on the presence of autoantibodies targeting ADAMTS13 (8–13). It was found that the majority of TTP patients develop autoantibodies that bind to the spacer domain of ADAMTS13, which is crucial for binding and subsequent processing of VWF (14–18). Autoantibodies targeting N-terminal domains neutralize the proteolytic activity of ADAMTS13 while autoantibodies directed against other domains can cause accelerated clearance of ADAMTS13 from the circulation both resulting in the accumulation of UL-VWF polymers (10, 19, 20). The retained UL-VWF strings on the surface of endothelial cells cause the formation of platelet clots within arteries and arterioles leading to microvascular thrombosis that subsequently gives rise to hemolytic anemia (disruption of red blood cells) and thrombocytopenia (systemic low level of platelets) that may cause skin hemorrhages, presenting as purple-colored spots on the skin (purpura) (1, 9, 21).

The pathology of iTTP has been extensively studied for decades; the etiology of iTTP however, remains elusive. Here we will discuss our current knowledge on processes implicated with the development of autoimmune disorders (ADs) and their relevance for the development of autoimmunity against ADAMTS13, considering both genetic and environmental factors. In this review, we will also introduce a novel model describing the role of infections and microbiota in the onset and/or relapses of iTTP.

The first suspicions concerning the involvement of genetic factors in TTP development rose from case reports describing the diagnosis of TTP in siblings (22, 23) and family members of different or same generation (mostly women) (24–27). Involvement of HLA was first suggested by the development of TTP in two HLA-identical brothers in the early 1980’s (25). The first large cohort study of HLA association in TTP was reported in the mid 1990’s by Joseph and co-workers (28). This cohort study elegantly showed the absence of the HLA class II HLA-DRB1*04 allele (encoded by the DRB4 gene), indicating it to be a protective allele for TTP (28).

After the clear separation of TTP from hemolytic uremic syndrome (HUS) (29) and the distinction between congenital TTP and iTTP (30–32) the first risk alleles for iTTP were found almost simultaneously by two independent groups (33, 34). The HLA-DRB1*11:01 and HLA-DRB1*11:04 alleles were found in different European Caucasian populations as the most prominent risk factors among the HLA-class II alleles (34, 35). The different studies also confirmed the earlier discovered protective allele HLA-DR53 (allele DRB1*04) (33–35). In later studies similar observations were done and additional HLA associations were found, which have been summarized in Table 1 . In a more recent study, it was found that the HLA risk alleles in the Japanese population were substantially different than for the European Caucasian populations. The main HLA-DRB1 allele identified as a risk factor for iTTP was found to be HLA-DRB1*08:03 (38). In contrast to HLA-DRB1*11, which is highly expressed in the European population, HLA-DRB1*08:03 is an allele unique to individuals with East Asian ancestry ( Figure 1 ). Additionally, the absence of HLA-DR3, -DR4 and -DR5 haplogroups (DR3/4/5*blank) and the higher frequency of HLA-DQA1*01:03 and HLA-DQB1*06:01 were also associated with iTTP in the Japanese population. In contrast, the haplotype HLA-DRB1*15:01/DRB5*01:01 (known to be in strong linkage disequilibrium) was identified as a protective factor in the Japanese population (38).

HLA associations have been found for several other ADs, with some of the associated HLA alleles being risk factors for one disease, but protective for other diseases (e.g. HLA-DRB1*04 is protective for iTTP in the general European Caucasian population, but a risk factor for giant cell arteritis (GCA) and Rheumatoid Arthritis) (39). Not all people that possess risk alleles develop the corresponding AD, which emphasizes the requirement of both genetic and environmental factors to lower the threshold for autoimmunity (37). To understand the role of HLA-DRB1*11 and other HLA class II alleles in the pathogenesis of iTTP, it is crucial to perform studies that provide mechanistic insight into the role of HLA class II in the onset of disease.

Other genetic factors like single-nucleotide polymorphisms (SNPs) were repeatedly implicated in several other autoimmune diseases, although the mechanism by which they influenced autoimmunity was not always clear (40). The analysis of functionally relevant SNPs in Toll-Like-Receptor-9 (TLR-9) revealed TLR-9 +2848G and TLR-9 +1174A genotypes to be more frequent in TTP patients (41). TLR-9 is an endosomal pattern recognition receptor that recognizes unmethylated-CpG-containing DNA motifs from infectious microorganisms and directs Interferon-alpha production in humans and autoantibody development in mice (42). The gene protein tyrosine phosphatase nonreceptor type 22 (PTPN22) encodes lymphoid specific phosphatase (Lyp), a potent inhibitor of T-cell activation, and the SNP C1858T is a risk factor in certain autoimmune diseases (43). However, an analysis of the PTPN22 C1858T SNP did not show any difference between TTP patients and controls (37). More recently, two SNPs were found to be associated with the development of iTTP: rs6903608 (44) and rs9884090 (45). While there is lack of functional data, in silico analyses of these SNPs revealed that rs6903608 may increase expression of the risk HLA molecules for iTTP (44), while rs9884090 is associated with reduced expression of protein O-glycosyltransferase 1 (POGLUT1), implying that post-translational modifications may shape the immune response towards ADAMTS13 (45).

Post translational modifications of antigens have long been recognized to play a role in certain autoimmune diseases (46). ADAMTS13 is an extensively glycosylated plasma protein, containing both O-glycans and N-glycans (47). It is known that alterations in glycosylation patterns may have impact on the immunogenicity of antigens (48). It is possible that in iTTP reduced O-glycosylation of serines by POGLUT1 leads to altered antigen presentation in HLA risk alleles or altered T-cell receptor (TCR) recognition, however, functional data to confirm this hypothesis is still required. It is also noteworthy that ADAMTS13 as an antigen is more extensively modified by citrullination in the context of sepsis and in the elderly suffering from underlying comorbidities (49). This raises the possibility that citrullination of ADAMTS13 is another contributing factor for the loss of tolerance towards ADAMTS13, leading to iTTP. Citrullination was earlier described as a fundamental process in driving the autoimmune processes in rheumatoid arthritis and other inflammatory conditions (50), and was found to be capable of altering the specificity of TCRs towards T-cell epitopes, even though it did not impact HLA binding (51). Citrullination through peptidyl-arginine deiminase 4 (PAD4) drives the formation of neutrophil extracellular traps (NETs), which are normally triggered by infectious agents and contribute towards thrombosis by several mechanisms, including oxidation of ADAMTS13 methionines and possibly citrullination of ADAMTS13 through PAD4 (52). Biomarkers for NETosis were found to be elevated in iTTP patients (53). The conformational change of ADAMTS13 towards an open-state was also found as hallmark of iTTP and it is promoted through the binding of autoantibodies of patients (54, 55). Interestingly, non-inhibitory anti-ADAMTS13 autoantibodies also exist in healthy (56) and obese individuals (57). Like the pathogenic autoantibodies in iTTP patients, these should also promote the opening of ADAMTS13. The removal of C-terminal-domain N-glycans in ADAMTS13 may induce this open conformation state as well (58). The modifications and the confirmational change of ADAMTS13 may result in the availability of different and cryptic epitopes which may in turn help to contribute to the onset of iTTP.

The presentation of antigens in the context of HLA-class II molecules is required for the induction of adaptive immune responses. It is important to note that HLA-class II molecules usually bind ~12-20-mer peptides, while the core peptide inside the binding cleft normally consists of only 9-aminoacids, of which a few serve as anchor residues creating specific peptide binding motifs (usually involving aminoacid positions P1, P4, P6 and P9). HLA-class II peptides are not restricted by a closed peptide binding cleft structure of the HLA molecule as is the case for HLA-class I molecules (59). HLA-class II molecules display much more flexibility, allowing for differences in positioning of peptides within the peptide binding cleft and differences in the peptide flanking residues. Therefore, HLA-II molecules are known to have a certain degree of promiscuity - that is, one HLA-II molecule binds many different peptides. On the other hand, peptides have also been found to show a degree of promiscuity towards HLA-II molecules (60, 61). Consequently, some HLA-class II restricted T-cell receptors are capable of recognizing more than one HLA class II-peptide complex with different affinities (59, 60, 62). The ADAMTS13-peptidomes for HLA-DR and HLA-DQ ( Figure 2 ) were dissected and it was revealed that the majority of eluents contained peptides derived from the ADAMTS13 CUB2 domain with the core sequence ‘FINVAPHAR’ (HLA-DRB1*11) or a different CUB2 peptide with the core sequence ‘LIRDTHSLR’. The latter was eluted from HLA-DRB1*03, an allele not formally associated with TTP (63). Peptides containing each of these core sequences were later shown to activate CD4 T-cells from TTP patients expressing the respective HLA-DRB1 alleles (65).

Elution experiments have demonstrated that promiscuity is a prominent feature of ADAMTS13 derived peptides. A notorious example is the ‘FINVAPHAR’ core peptide from the ADAMTS13 CUB2 domain which was shown to bind to different HLA-DR alleles (63). Interestingly, the core peptide ‘FINVAPHAR’ which was eluted from HLA-DRB1*11:01, was also predicted to be the highest affinity peptide for the risk factor HLA-DRB1*08:03 in the Japanese population, although the core sequence showed a shift of one amino acid, becoming ‘LFINVAPHA’ ( Figure 1 ). This strongly suggests that HLA-binding peptides containing this core structure will likely be capable of binding to both HLA-DRB1*11:01 as well as HLA-DRB1*08:03 (38). Promiscuity was also shown for other ADAMTS13 peptides (e.g. the ADAMTS13-1239-1253 peptide, or GDMLLLWGRLTWRKM - a CUB1 peptide - and the core sequence ‘LLRDPSLGA’ - a metalloprotease peptide) (64). The ADAMTS13 peptidome for HLA-DQ was also assessed, and the same type of promiscuity was observed among these alleles. Peptides with the core sequence ‘QADCAVAIG’ (CUB1 domain) were eluted from two different haplotypes (64). Peptide promiscuity is also possible between different HLA class II molecules (i.e. the same peptide binding both HLA-DR and HLA-DQ) (66). The ‘QADCAVAIG’ core peptide found in the HLA-DQ elution study was also found for HLA-DR although the predicted core was slightly different (‘CAVAIGRPL’). The same applied to the peptides with the core ‘LLRDPSLGA’ eluted for HLA-DR, which were obtained also from HLA-DQ elution although with a slightly different predicted core sequence ‘LRDPSLGAQ’ (64). Overall, in the different elution experiments more peptides were eluted from HLA-DR alleles than HLA-DQ, which is probably influenced by the expression of the HLA molecules and the lower binding affinities of ADAMTS13 peptides in HLA-DQ (64) as was also shown for coagulation Factor VIII (66).

The prominent HLA-class II association of iTTP combined with the presentation of ADAMTS13-derived peptides (including immunogenic peptides) on risk alleles for iTTP strongly supports a pivotal role for CD4 T cells in the immunopathogenesis of iTTP. Subclass and clonal analyses of anti-ADAMTS13 antibodies from iTTP patients provided evidence for isotype switching and affinity maturation (9, 10, 20, 67–69). These immunological processes further corroborate the role of CD4 T cells in the autoimmune responses against ADAMTS13. CD4 (helper) T cells are part of the adaptive immunity and are required for immune responses against pathogens or tumors. The T-cell receptors (TCR) of CD4 T cells bind to peptides in the context of HLA-class II (HLA-DR, HLA-DQ or HLA-DP) molecules on antigen-presenting cells (APC) (70, 71). The activation of CD4 T cells does not only depend on the binding of their TCR with peptide-HLA complexes but also on the binding of CD4 T cells with co-stimulatory (CD80 and/or CD86) and adhesion molecules (CD54) expressed by professional APC (72, 73). The naïve CD4 T cells differentiate upon activation by professional APC into various effector CD4 T-cell subsets, such as Th1 (74), Th2 (75), Th17 (76), T follicular (Tfh) (77) or regulatory T (Treg) cells (78). The CD4 T-cell fate is influenced by the cytokines produced by APC after which the differentiated T-cell subsets migrate to their target location to exert their function (73).

In ADs, auto-reactive CD4 T cells are used to induce the generation of autoantibodies by B cells. During T-cell development in the thymus, the T cells binding strongly to self-peptides presented in self-HLA are eliminated from the T-cell repertoire. However, a large part of the autoreactive T cells with a low or intermediate affinity escape negative selection and become part of the T-cell repertoire (71, 79–81). Autoreactive T cells involved in the development of iTTP and other ADs are required to possess a proper functional avidity to become activated by self-antigens. The avidity of T cells is determined by the TCR affinity to the peptide in the context of an HLA molecule. TCRs have been shown to bind differently to HLA-peptide complexes depending if the peptide is derived from a foreign antigen or from a self-protein. For foreign T-cell epitopes, the binding takes place with the TCR in a diagonal position relative to the HLA-peptide complex, with the CDR3 variable domains centered in the P5 position of the embedded peptide. In the case of self-peptides the docking may be altered, and examples where it may be shifted towards the N-terminal region of the peptide, or tilted in such a way that only the variable beta chain of CDR3 maintains contact with the HLA-peptide complex have been found (82). Besides the TCR binding, also the binding with stimulatory and adhesion molecules and the internal signaling transduction are important in the activation of CD4 T cells (83–85). In the context of iTTP, autoreactive CD4 T cells are hypothesized to be activated by APC that take up ADAMTS13 (using, for example, the macrophage mannose receptor (MR) or CD163) (86, 87). Upon endocytosis, ADAMTS13-derived peptides are loaded onto the HLA-class II molecules (63) and are recognized by the autoreactive CD4 T cells. Upon peptide recognition and co-stimulation, the CD4 T cells get activated and differentiate into effector CD4 T cells that secrete cytokines which subsequently stimulate antigen-specific B cells to differentiate into autoantibodies-producing plasma cells (20, 88).

The role of CD4 T cells in the development of iTTP is a clear given, as CD4 help is required in the production and affinity maturation of ADAMTS13-directed antibodies. However, the exact trigger for the activation of autoreactive CD4 T cells remains unclear. Like most ADs (86), iTTP is generally considered to be very rare (1), however the association between infection and the development of iTTP is evident. The high prevalence of infections at time of diagnosis (41%) (41) suggest that these pathogens are at least one of the triggers involved in the loss of tolerance towards ADAMTS13.

The etiology of ADs involves different factors including the genetic predisposition of the patient, but also the intricate interplay between the patient’s immune system and environmental factors (89, 90). The development of autoimmunity has often been associated with the occurrence of an infection which can be explained by the mechanism of ‘molecular mimicry’, the similarities between pathogen- and self-peptides resulting in the activation of autoreactive T or B cells in genetically susceptible individuals (89–92). Molecular mimicry is generally considered when four major criteria are met: 1) similarities between pathogen-derived antigens and self-antigens; 2) antibodies or T cells are detected that cross-react with both epitopes; 3) an epidemiological correlation/association between infection and onset of autoimmunity is present; 4) the development of the autoimmune disease is reproduced in an animal model by exposing the animal to the pathogen-derived cross-reactive antigen (89). In the past decades a variety of ADs were assumed to be caused by molecular mimicry as the antibodies or T cells demonstrated specificities directed against epitopes shared by pathogens and self-antigens. Only one autoimmune disease was found to fulfill all four listed criteria for molecular mimicry and that was Guillain-Barré syndrome resulting from C. jejuni infection (93). However, in many other ADs (type I diabetes, rheumatoid arthritis, systemic lupus erythematosus, Sjögren’s syndrome, systemic sclerosis, autoimmune thyroid disease, autoimmune liver diseases, autoimmune hepatitis, primary biliary cholangitis), and also in the context of vaccines, molecular mimicry was suggested to be the underlying mechanism but was not formally demonstrated due to lack of fulfillment of all the four criteria outlined above (89).

The aim of this review was to unravel the etiology of iTTP by addressing both genetic and environmental factors involved in the induction of autoimmunity against ADAMTS13. Although the knowledge from other ADs gives insight to the possible immunopathogenesis of iTTP, several details of the processes leading to an immune response against ADAMTS13 remain to be uncovered. The findings in earlier studies seem very promising in pleading the case of molecular mimicry as a major underlying mechanism for the development of ADs and iTTP. Associations of HLA and iTTP have been identified, as they have been known for other ADs. ADAMTS13 peptides have been identified as T-cell epitopes in risk HLA alleles, and computational studies can be performed to find potential similar peptides from microorganisms, especially those commonly associated with iTTP at diagnosis. Some outcomes of such studies already exist. However, the search has to be continued and refined, both for infectious microorganisms and for microbiota, as done for other ADs (164). The use of bio-informatic tools in the homology search and the further validation using T-cell reactivity assays could substantiate the model as proposed in this review. More insight into the TCRs of iTTP patients is also needed in this field, and an animal model that properly reproduces the immune features seen in humans would provide a substantial advancement in the case of iTTP autoimmunity mechanisms. Additionally, cross-reactive B-cell epitopes remain to be formally identified, and research here may be aided by means of 3D-structure comparison software’s like CE-BLAST (165) or PDBeFold (166), by comparing available ADAMTS13 crystal structures with those of proteins from microorganisms. Future studies will further uncover the missing pieces of the puzzle that is autoimmunity and loss of tolerance towards ADAMTS13.

AL, NG, and JV wrote the manuscript and prepared the figures. All authors contributed to the article and approved the submitted version.

Financial support was obtained from the Answering TTP foundation, Landsteiner Foundation for Blood Transfusion Research and The Netherlands Thrombosis Foundation. NG was supported by a grant from Sanquinnovate for the development of novel therapeutic compounds for treatment of iTTP. AL has been supported by a grant from the Answering TTP Foundation (grant year: 2018). NG has been supported by a grant from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement number 675746 (PROFILE) and a grant from Sanquinnovate (SQI00060).

NG and JV are inventors on a patent-application describing novel compounds for the treatment of immune TTP.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.