Preferential pairing of T and B cells for production of antibodies without covalent association of T and B epitopes

Development of a novel fully functional coagulation factor VIII with reduced immunogenicity utilizing an in silico prediction and deimmunization approach

BACKGROUND

Up to 30% of hemophilia A patients develop inhibitory antibodies against the infused factor VIII (FVIII). The development of a deimmunized FVIII is an unmet high medical need. Although improved recombinant FVIII (rFVIII) products evolved within the last years, the immunogenicity has not been solved. A deimmunized FVIII could reduce the probability of inhibitor development, providing safer therapy.

OBJECTIVE

To develop a deimmunized FVIII molecule by modifying major histocompatibility complex (MHC) class II presentation, leading to a functional but less immunogenic molecule.

METHODS

We performed (1) in silico prediction of potentially immunogenic T cell epitopes and their modification by amino acid substitutions in the FVIII sequence, (2) evaluation of functional and structural similarity of the modified rFVIII to unmodified FVIII and registered products, and (3) confirmation of the reduced immunogenicity by in vitro testing.

RESULTS

A partially deimmunized fully functional FVIII molecule incorporating 19 amino acid substitutions was generated. The substitutions led to a reduction of the immunogenicity score, indicating a reduced immunogenicity based on in silico calculations. This was confirmed in an in vitro dendritic cell (DC)--T cell assay. Using this assay, cells from healthy donors proved the significantly reduced immunogenicity of the modified FVIII variant by revealing less proliferation of T helper cells to this variant than to the unmodified FVIII.

CONCLUSION

In silico predictions resulted in a partially deimmunized FVIII. This FVIII is fully functional and was demonstrated to be less immunogenic in in vitro testing. This approach may result in a reduction of the inhibitor risk for patients with hemophilia A.

T cell recognition of the A2 domain of coagulation factor VIII in hemophilia patients and healthy subjects

Hemophilia A patients treated with coagulation factor VIII (FVIII), and also some healthy subjects, may develop anti-FVIII antibodies (Ab), whose synthesis is driven by FVIII-specific CD4+ T cells. Some Ab block the procoagulant function of FVIII (inhibitors). Many inhibitors recognize epitopes on the FVIII A2 domain. Here, we have sought to identify A2 epitopes recognized by CD4+ T cells. We tested the proliferative response of CD4+ blood lymphocytes (BL) from hemophilia patients and healthy subjects, to overlapping synthetic peptides spanning the A2 domain sequence. Many A2 peptides induced proliferative responses of CD4+ BL from one or more subjects. The peptide-induced responses were strongest in hemophilia patients with inhibitors, weakest in healthy subjects. A2 peptides comprising residues 371-400, 621-650 and 671-690 elicited frequent and strong responses in hemophilia A patients, and especially in those with inhibitors. Healthy subjects recognized frequently only the sequence 371-400. A three-dimensional model of the A2 domain suggests that these CD4+ epitope sequences have structural features typical of 'universal' CD4+ T epitopes.

Epitope repertoire of human CD4(+) T cells on the A3 domain of coagulation factor VIII

Severe hemophilia A patients treated with factor (F)VIII may develop antibodies (Ab) that block FVIII function (inhibitors). Autoimmune inhibitors may develop in subjects without congenital hemophilia, and cause acquired hemophilia. Hemophiliacs without inhibitors and healthy subjects may also have small amounts of antiFVIII Ab. FVIII-specific CD4(+) T cells induce antiFVIII Ab synthesis. Here, we have examined their epitope repertoire in hemophilia patients and healthy subjects. We used overlapping synthetic peptides, spanning the sequence of the FVIII A3 domain, to challenge blood CD4(+) T cells in proliferation assays. The epitopes recognized in hemophilia A patients with or without inhibitors, acquired hemophilia patients, or healthy subjects overlapped, yet had characteristic differences. Most members of one or more study groups recognized the sequence regions 1691-1710, 1801-1820, 1831-1850, and 1941-60. In the proposed three-dimensional structure of the A3 domain, these sequences are largely exposed to the solvent and flanked by flexible sequence loops: these are structural features characteristic of 'universal' CD4(+) T epitopes. Hemophilia A patients with inhibitors recognized prominently only the sequence 1801-1820, which overlaps a known inhibitor binding site. This is consistent with the possibility that CD4(+) T cells recognizing epitopes within residues 1801-1820 have a role in inducing inhibitor synthesis. In contrast, CD4(+) T cells sensitized to sequences 1691-1710 and 1941-60, which are recognized by healthy subjects and hemophilia A patients without inhibitors, might curb inhibitor synthesis.

Recognition of coagulation factor VIII by CD4+ T cells of healthy humans

Hemophilia A patients treated with coagulation factor (F)VIII may develop an anti-FVIII immune response. Anti-FVIII antibodies may occur also in healthy subjects. To understand the extent to which an immune response to FVIII occurs in healthy subjects, we investigated the proliferative response of blood CD4+ T cells from 90 blood donors to FVIII and to pools of overlapping synthetic peptides spanning the sequences of individual FVIII domains (A1-A3, C1-C2). Most subjects responded to FVIII and several FVIII domains. Men had stronger responses to FVIII than women, and older subjects than younger subjects. The domain-induced responses were weaker than the FVIII-induced responses, yet their intensity in individual subjects correlated with that of the response to FVIII. We examined whether Th1 and/or Th2 cells responded to FVIII in 68 subjects, by determining the CD4+ T cells that secreted interferon-gamma (IFN-gamma) or interleukin (IL)-5 after stimulation with FVIII: 25 subjects had FVIII-specific IFN-gamma-secreting cells, and seven of them had also FVIII-specific IL-5-secreting cells. None had only IL-5-secreting cells. Thus, a CD4+ T cell response to FVIII, which first involves Th1 cells, is common among subjects with a normal procoagulant function.

Human CD4+ T-cell epitope repertoire on the C2 domain of coagulation factor VIII

Approximately 25% of severe hemophilia A patients develop antibodies (Ab) that neutralize the procoagulant function of factor (F)VIII (inhibitors). Autoimmune FVIII inhibitors may develop in individuals without congenital FVIII deficiency and cause acquired hemophilia. Low titers of anti-FVIII Ab may be present in hemophilia A patients without inhibitors and in healthy blood donors. FVIII-specific CD4+ T-cells drive the synthesis of anti-FVIII Ab. We examined the epitope repertoire of CD4+ T-cells from 15 healthy subjects, 10 hemophilia A patients without inhibitors, 11 hemophilia A patients with inhibitors, and six acquired hemophilia patients. Blood CD4+ T-cells were challenged in proliferation assays with a panel 16 overlapping synthetic peptides, spanning the sequence of the FVIII C2 domain. The sequence region 2291-2330 contained the most frequently and strongly recognized peptides in each of the four subject groups. Crystallographic B factor data and the location of these peptides within the three-dimensional structure of the C2 domain confirm that this region has a high degree of solvent exposure and flexibility within the peptide backbone, which are structural features typical of immunodominant universal CD4+ epitopes. Furthermore, this sequence region overlaps inhibitor-binding sites, suggesting that CD4+ T-cells recognizing peptide sequences within this region might be involved in inhibitor synthesis. The sequence regions 2191-2210 (recognized strongly by each study group except hemophilia A patients with inhibitors) and 2241-2290 (recognized primarily by acquired hemophilia patients and healthy subjects) share the same structural features, and also overlap inhibitor-binding sites. Although similar, there appear to be important differences in the CD4+ epitope repertoires of congenital and acquired hemophilia patients.

Myasthenia gravis and its animal model: T cell receptor expression in an antibody mediated autoimmune disease

Myasthenia gravis (MG) is a prototypic antibody-mediated autoimmune disease. Since the primary target antigen of the autoimmune response is known and a well-characterized animal model is available, MG is often considered an excellent situation for the application of novel specific immunotherapies, many of which are directed at T lymphocytes. CD4+ helper T cells are required for the development of the animal model, experimental autoimmune MG (EAMG). Even though the target antigen, acetylcholine receptor (AChR) is immunologically complex, the T cell response to AChR in mice is dominated by recognition of a single peptide by about 50% of the T cells. These T cells, in turn, utilize a restricted set of TCR gene elements and conserved CDR3 regions. While specific therapy directed at the immunodominant T cells is capable of reducing the magnitude of the anti-AChR response, considerable flexibility is apparent and reveals the ability of additional T cells to provide the requisite B cell help. In human MG patients, AChR-specific T cells have been identified but in many studies the frequencies were surprisingly low. In a very few cases, AChR-specific T cells have been cloned from MG patients. Analysis reveals heterogeneity in epitope recognition and MHC restriction. Little information on TCR structure is available. Our own studies using antigen-specific as well as non-specific methods for examining clonal T cell expansions in MG have led to an alternative hypothesis concerning T-B collaboration in MG.

The serum albumin-binding region of streptococcal protein G (BB) potentiates the immunogenicity of the G130-230 RSV-A protein

BBG2Na is a protein comprising residues 130-230 of the respiratory syncytial virus subgroup A (RSV-A) G protein (G2Na) fused to the albumin-binding domain of streptococcal G protein (BB). BBG2Na was cloned, expressed in Escherichia coli and renaturated. In rodent models, this subunit RSV vaccine adjuvanted in Alhydrogel induced specific antibodies and conferred protection to RSV infection. Comparison of the antibody production in a BALB/c mouse model revealed that BBG2Na induced a stronger and earlier G2Na antibody response than G2Na alone, without altering the IgG subclass distribution. To address the role of the BB part, we explored its carrier properties and showed that it is a Th dependent antigen, generating a more potent G2Na-specific B cell memory response and able to generate Th cells that provide help for G2Na antibody production.

Anatomy of the antigenic structure of a large membrane autoantigen, the muscle-type nicotinic acetylcholine receptor

The neuromuscular junction nicotinic acetylcholine receptor (AChR), a pentameric membrane glycoprotein, is the autoantigen involved in the autoimmune disease myasthenia gravis (MG). In animals immunized with intact AChR and in human MG, the anti-AChR antibody response is polyclonal. However, a small extracellular region of the AChR alpha-subunit, the main immunogenic region (MIR), seems to be a major target for anti-AChR antibodies. A major loop containing overlapping epitopes for several anti-MIR monoclonal antibodies (mAbs) lies within residues alpha 67-76 at the extreme synaptic end of each alpha-subunit: however, anti-MIR mAbs are functionally and structurally quite heterogeneous. Anti-MIR mAbs do not affect channel gating, but are very effective in the passive transfer of MG to animals; in contrast, their Fab or Fv fragments protect the AChR from the pathogenic effects of the intact antibodies. Antibodies against the cytoplasmic region of the AChR can be elicited by immunization with denatured AChR and the precise epitopes of many such mAbs have been identified; however, it is unlikely that such antibodies are present in significant amounts in human MG. Antibodies to other extracellular epitopes on all AChR subunits are present in both experimental and human MG; these include antibodies to the acetylcholine-binding site which affect AChR function in various ways and also induce acute experimental MG. Finally, anti-AChR antibodies cross-reactive with non-AChR antigens exist, suggesting that MG may result from molecular mimicry. Despite extensive studies, many gaps remain in our understanding of the antigenic structure of the AChR; especially in relation to human MG. A thorough understanding of the antigenic structure of the AChR is required for an in-depth understanding, and for possible specific immunotherapy, of MG.

Factors that determine the severity of experimental myasthenia gravis

Based on our current information, the robust differences in responses of B6 and bm12 mice after immunization with AChR are as follows: (1) The AChR-specific T cell repertoires are strikingly different. The epitope specificities, as well as the rearranged TCR alpha and beta chains and their CDR3 domains, are virtually nonoverlapping in the two strains of mice. (2) The AChR antibody responses are quantitatively different, both to Torpedo AChR and to the autoantigen--mouse AChR. (3) The isotype distribution of AChR antibodies favors IgG2b in B6 mice, but not in bm12 mice. (4) The clinical manifestations of EAMG are qualitatively and quantitatively different in the two strains. These considerations have led to the following scheme, illustrated diagrammatically in FIGURE 2, to explain the differences in EAMG in B6 and bm12 mice: (1) The MHC Class II of B6 mice binds the alpha 146-162 peptide of Torpedo AChR with high affinity, while the genetically altered MHC Class II of bm12 mice does not, as previously suggested (see FIGURE 2). (2) The alpha 146-162/MHC Class II complex occurs only in B6 mice and interacts with T cells having appropriate TCRs, resulting in their stimulation and expansion. Although T cells of appropriate specificity are also available in the bm12 strain, the relevant peptide/MHC Class II complex is not present. Therefore, very few T cells with specificity for alpha 146-162 are stimulated, and those that are stimulated have different TCRs. T cells with specificity for other AChR peptides are also present and expanded in both strains of mice, but they have less influence on the outcome of the immune response. (3) The alpha 146-162-specific T cells of B6 mice, in turn, interact strongly with AChR-specific B cells of B6 mice. These B cells present the same epitope/MHC Class II complex as the APCs and therefore interact well with the alpha 146-162-specific T cells (FIGURE 2). Thus, T cells of this specificity appear to provide more efficient help for AChR antibody production than T cells with specificity for other Torpedo AChR epitopes. This results in production of greater amounts of AChR antibodies, including a critical subset that cross-reacts with autologous mouse AChR. The higher autoantibody levels contribute to the greater susceptibility to EAMG and to the greater severity of manifestations in the B6 strain compared with the bm12 strain. (4) There is a bias in B6 mice toward the production of AChR antibodies of IgG2b isotype. We suggest that T cells specific for alpha 146-162 may contribute to this isotype bias. The IgG2b antibodies appear to have particularly potent "myasthenogenic" effects in rats and mice. (5) Finally, it should be emphasized that these differences in immunological and clinical aspects of EAMG in B6 and bm12 mice are relative rather than absolute. T cells that respond to AChR epitopes other than alpha 146-162 can also provide help for AChR antibody production, albeit less potent. In a sense, this model represents a special case of molecular mimicry. In this case, the source of the foreign antigenic molecule is injection rather than the more usual route of infection. The antigen (Torpedo AChR) is one that these mice would never naturally encounter, and the critical amino acid (lysine 155) of the key epitope (alpha 146-162) is present only in the AChR of electric organs of electric fish and not in the AChR of mice, chickens, cows, or humans. The important point is that a detail of the structure of the foreign antigen--that is, a particular peptide of Torpedo AChR--can determine the severity of an antibody-mediated autoimmune disease, depending on how it interacts with a detail of the structure of the MHC Class II molecule and, in turn, on how the peptide/MHC Class II complex interacts with the available T cell repertoire. (ABSTRACT TRUNCATED)

How subtle differences in MHC class II affect the severity of experimental myasthenia gravis

Myasthenia gravis is an autoimmune disorder characterized by muscle weakness, due to an antibody-mediated deficit of acetylcholine receptors (AChRs) at neuromuscular junctions. We analyzed the factors that determine the severity of experimental myasthenia gravis (EAMG) induced by immunization with Torpedo AChR, in two congenic strains of mice--B6 mice, which are highly susceptible to EAMG; and bm12 mice, which are relatively resistant, and differ only in a change of three amino acids in MHC Class II. We prepared large numbers of AChR-specific T cell hybridomas from each strain and characterized their epitope specificities and T cell receptor (TCR) gene usage: Half the B6 hybridomas responded to a single AChR peptide (alpha 146-162), and their TCR genes encoded restricted V alpha and V beta chains and CDR3 motifs. bm12 hybridomas had different epitope specificities and different, less restricted TCR genes. APCs were able to present AChR or AChR-derived peptides virtually exclusively to hybridomas of their own strain. Levels of antibodies to Torpedo and autoantibodies to mouse AChR were higher in B6 mice, and were biased toward the IgG2b isotype. We conclude that the "better fit" of MHC II, peptide, and TCR in the B6 mice enhanced cognate interactions of APCs with T cells, and T cells with B cells, resulting in a more abundant and pathogenic AChR antibody response, and thus more severe EAMG.

Immunogenicity of chimeric multiple antigen peptides based on Plasmodium falciparum antigens: impact of epitope orientation

Assembly of B and T epitopes in multiple antigen peptides (MAP) can bypass genetically predisposed unresponsiveness to B epitopes. Although the underlying mechanisms are unknown, B-cell responses to such diepitope MAP are influenced by intramolecular epitope orientation. In this study, MAP constructs were synthesized, encompassing two epitopes derived from the Plasmodium falciparum antigens circumsporozoite protein (CS) and Pf332. In addition to B epitopes, the sequences comprised T epitopes restricted to mouse H-2b (CS) or to H-2d and H-2k (Pf332) haplotypes. Congenic H-2b, H-2d and H-2k Balb mice were immunized with MAP in which the two epitopes were arranged either tandemly or in parallel. Tandemly arranged (B-T)4 MAP, in which the relevant T epitope was positioned adjacent to the lysine core [(Pf332-CS)4-core for H-2b mice and (CS-Pf332)4-core for H-2d and H-2k mice], elicited the most potent antibody responses in terms of reactivity to both epitopes. Additionally, the (B-T)4 constructs were generally most efficient in recalling proliferative T-cell responses in vitro, irrespective of the MAP used for in vivo priming. As high antibody titers were generated to both epitopes, the position of B epitopes in the constructs does not appear to be critical for an efficient B-cell response. Rather, the association of strong B- and T-cell responses to the (B-T)4 MAP constructs suggests that the intramolecular position of the relevant T epitope determines the magnitude of specific antibody production.

Prevention of experimental myasthenia gravis by nasal administration of synthetic acetylcholine receptor T epitope sequences

T cell tolerization prevents and improves T cell-mediated experimental autoimmune diseases. We investigated here whether similar approaches could be used for antibody (Ab)-mediated autoimmune diseases. Myasthenia gravis, caused by IgG Ab against muscle acetylcholine receptor (AChR), is perhaps the best characterized of them. We used an animal model, experimental myasthenia gravis induced in C57Bl/6 mice by immunization with Torpedo acetylcholine receptor (TAChR), to demonstrate that nasal administration of synthetic sequences of the TAChR alpha-subunit- forming epitopes recognized by anti-TAChR CD4+ T helper cells (residues alpha150-169, alpha181-200, and alpha360-378), given before and during immunization with TAChR, causes decreased CD4+ responsiveness to those epitopes and to TAChR, reduced synthesis of anti-TAChR Ab, and prevented experimental myasthenia gravis. These effects were not induced by nasal administration of synthetic epitopes of diphtheria toxin. Secretion of IL-2, IL-4, and IL-10 by spleen T cells from TAChR immunized mice, in response to challenge with TAChR in vitro, indicated that in sham-tolerized mice only Th1 cells responded to TAChR, while peptide-treated mice had also an AChR-specific Th2 response. The TAChR peptide treatment induced also in vitro anergy to the TAChR of the spleen T cells, which was reversed by IL-2.

Generation of antibodies to human IL-12 and amphiregulin by immunization of Balb/c mice with diepitope multiple antigen peptides

Six peptide sequences derived from the human proteins/oligopeptides IL-12, amphiregulin and FALL-39 were synthesized in order to raise specific antibodies in Balb/c mice. Although peptides are valuable tools for generating specific antibodies, they are often poor immunogens due to their small size and lack of relevant T-cell epitopes. To circumvent these limitations, the human peptides were co-synthesized in diepitope multiple antigen peptides (MAP) with a known H-2d-restricted T helper-cell epitope. The importance of including a T-cell epitope in the diepitope MAPs was demonstrated by the fact that only one of the human peptides was immunogenic as a monoepitope MAP, lacking the T-cell epitope. Conversely, all diepitope MAPs generated potent antibody responses to the desired human peptides as well as to the T-cell epitope. A certain degree of variability of the antibody responses to the diepitope MAPs indicated that the alterable component, i.e. the human B-cell epitope, influenced the T-cell help elicited by the T-cell epitope. Still, the relative conformity of the B-cell responses suggests that this strategy is generally applicable for a rational production of specific antibodies. Moreover, antiserum to four diepitope MAPs recognized the corresponding full-length human protein/oligopeptide as did monoclonal antibodies made against IL-12-and amphiregulin-based MAPs.

Cooperation between transmissible gastroenteritis coronavirus (TGEV) structural proteins in the in vitro induction of virus-specific antibodies

Following infection of haplotype defined NIH-miniswine with virulent transmissible gastroenteritis coronavirus (TGEV), isolated mesenteric lymph node CD4+ T-cells mounted a specific proliferative response against infectious or inactivated purified virus in secondary in vitro stimulation. A specific, dose-dependent response to the three major recombinant viral proteins: spike (S), membrane (M), and nucleoprotein (N), purified by affinity chromatography, was characterized. Induction of in vitro antibody synthesis was analyzed. The purified recombinant viral proteins induced the in vitro synthesis of neutralizing TGEV-specific antibodies when porcine TGEV-immune cells were stimulated with each of the combinations made with two of the major structural proteins: S + N, S + M, and to a minor extent with M + N, but not by the individual proteins. S-protein was dissociated from purified virus using NP-40 detergent and then micellar S-protein oligomers (S-rosettes) were formed by removing the detergent. These occurred preferentially by the association of more than 10 S-protein trimmers. These S-rosettes in collaboration with either N or M-proteins elicited TGEV-specific antibodies with titers up to 84 and 60%, respectively, of those induced by the whole virus. N-protein could be partially substituted by a 15-mer peptide that represents a T helper epitope previously identified in N-protein (Antón et al. (1995)). These results indicate that the induction of high levels of TGEV-specific antibodies requires stimulation by at least two viral proteins, and that optimum responses are induced by a combination of S-rosettes and the nucleoprotein.

Mechanisms by which the I-ABM12 mutation influences susceptibility to experimental myasthenia gravis: a study in homozygous and heterozygous mice

The I-Abm12 mutation in C57B1/6 (B6) mice yields the B6.C-H-2bm12 (bm12) strain, which is resistant to Experimental Myasthenia Gravis (EMG) induced by immunization with Torpedo acetylcholine receptor (TAChR), while the parental B6 strain is highly susceptible to EMG. CD4+ cells from bm12 mice immunized with TAChR do not recognize three sequence regions of the TAChR alpha subunit which dominate the CD4+ cell sensitization in B6 mice. We immunized with TAChR bm12, B6 and (bm12 x B6)F1 mice. B6 and F1 mice developed EMG with comparable frequency. Their CD4+ cells recognized the same TAChR alpha subunit peptide sequences (T alpha 150-169, T alpha 181-200 and T alpha 360-378). CD4+ cells from TAChR-sensitized F1 mice were challenged with TAChR and alpha subunit epitope peptides, using F1, B6 or bm12 APC. B6 and F1 APC presented all these Ag efficiently, while bm12 APC presented TAChR and peptide T alpha 150-169 poorly and erratically. Anti-TAChR and anti-alpha subunit epitope CD4+ lines propagated from F1 and B6 mice had similar TcR V beta usage. All lines but those specific for the sequence T alpha 150-169 had unrestricted V beta usage. Anti-T alpha 150-169 lines from both B6 and F1 mice had a strong preferential usage of V beta 6. Anti-T alpha 150-169 lines from F1 mice had also a slightly higher V beta 14 usage. B6, bm12 and F1 mice developed similar anti-TAChR Ab titres, and had Ab bound to muscle AChR in comparable amounts. Therefore EMG resistance of bm12 mice must be due to a subtle shift in the anti-AChR Ab repertoire, and absence of special Ab able to cause destruction and/or dysfunction of muscle AChR. This is probably related to the absence of CD4+ cells sensitized to epitopes within the sequence T alpha 150-160, consequent to the inability of the I-Abm12 molecule to present this sequence.

Fine specificity of immune responses to epitopic sequences in synthetic peptides containing B and T epitopes from the conserved Plasmodium falciparum blood-stage antigens

Immunisation with two chemically synthesised, linear, multiple epitope peptides (MEP) containing B and T cell epitopes from two conserved blood-stage antigens of the human malaria parasite, Plasmodium falciparum, induced high levels of circulating antibodies without the use of a carrier protein. Immunisation of BALB/c mice with MEP constructs (P1 and P2) induced antibodies against the various epitope sequences included in their structures, although the immune response was focused more towards the N terminal and the middle portion of the peptides. In vitro T cell proliferation assays indicated that only one of the two Th epitopes included in P1 and P2 are functional. Both P1 and P2, based on P. falciparum sequences, cross-reacted with sera from P. yoelii-infected mice. Immunisation with P1 in CFA, but not with P2, provided partial protection to BALB/c mice against P. yoelii challenge infection. Peptide P1 was highly immunogenic in alum also, and a somewhat higher level of protection was observed as compared to CFA immunisation. We found that immunisation with P1 induced antibody responses in different strains of mice, although to different extents. These results suggest that linear, multiple epitope peptides may offer attractive alternatives as subunit vaccine candidate molecules, but at the same time highlight the fact that the design principles are far from being clear and have yet to be worked out.