Development and validation of immunoassays to quantify the half-antibody exchange of an IgG4 antibody, natalizumab (Tysabri®) with endogenous IgG4.

Natalizumab is a humanized IgG4 monoclonal antibody which binds human α4 integrin and is approved for treatment of multiple sclerosis and Crohn's disease. Assessment of the in vivo disposition of natalizumab presents a unique assay development challenge due to the ability of human IgG4 antibodies to undergo half-antibody exchange in vivo. Such exchange generates IgG4 molecules of mixed specificity comprising a natalizumab heavy-light chain pair coupled to an IgG4 heavy-light chain pair of unknown specificity. Since exchanged and non-exchanged species cannot be quantified independently using a single enzyme linked immunosorbent assay (ELISA), a novel quantitation strategy was developed employing two ELISAs: one measuring total natalizumab including both intact and exchanged molecules, and the second measuring only intact natalizumab. The presence and amount of exchanged natalizumab in serum is calculated by the difference in values obtained in the two assays. To evaluate assay performance, a control reagent was created from natalizumab and an irrelevant humanized monoclonal IgG4 antibody. Subsequent validation demonstrated that both assays are specific, accurate, and precise within the working ranges of the assays (1.5-10µg/mL for total and 0.5-12µg/mL for intact natalizumab assays). The mean accuracy, intra- and inter-assay precision for both assays were 82-113%, ≤9% and ≤20%, respectively. Additionally, the limits of detection of intact and exchanged natalizumab were established using statistical methods. The utility of the two-assay strategy was confirmed by analyzing samples from a pharmacokinetic study in rats using different variants of natalizumab administered along with another human IgG4 antibody as an exchange partner.

Resolution of disulfide heterogeneity in Nogo receptor I fusion proteins by molecular engineering.

NgRI (Nogo-66 receptor) is part of a signalling complex that inhibits axon regeneration in the central nervous system. Truncated soluble versions of NgRI have been used successfully to promote axon regeneration in animal models of spinal-cord injury, raising interest in this protein as a potential therapeutic target. The LRR (leucine-rich repeat) regions in NgRI are flanked by N- and C-terminal disulfide-containing 'cap' domains (LRRNT and LRRCT respectively). In the present work we show that, although functionally active, the NgRI(310)-Fc fusion protein contains mislinked and heterogeneous disulfide patterns in the LRRCT domain, and we report the generation of a series of variant molecules specifically designed to prevent this heterogeneity. Using these variants we explored the effects of modifying the NgRI truncation site or the spacing between the NgRI and Fc domains, or replacing cysteines within the NgRI or IgG hinge regions. One variant, which incorporates replacements of Cys²66 and Cys³°9 with alanine residues, completely eliminated disulfide scrambling while maintaining functional in vitro and in vivo efficacy. This modified NgRI-Fc molecule represents a significantly improved candidate for further pharmaceutical development, and may serve as a useful model for the optimization of other IgG fusion proteins made from LRR proteins.

Suppression of sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample preparation artifacts for analysis of IgG4 half-antibody.

Human IgG4 subtype antibodies have often been reported to have a significant portion (5-50%) of a heavy chain-light chain dimer ("half-antibody") on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), in which the heavy chain is not covalently linked through the hinge disulfides to another heavy chain. We demonstrate here that there can be artifactual sources of half-antibody. One occurred during SDS-PAGE sample preparation where rapid disulfide scrambling was initiated by preexisting free sulfhydryls in the monoclonal antibody (mAb) and by free sulfhydryl produced by destruction of disulfide bonds during heating. Inclusion of N-ethylmaleimide in the sample buffer prevented the disulfide scrambling. Presumably, cyclization of the flexible IgG4 hinge during this disulfide scrambling leads to the preferential separation of heavy chains. A second condition producing half-antibody was reoxidation after exposure to reductant, where 46% of the antibody was trapped in the intrachain disulfide form. The amount of half-antibody was reduced to 4% by reoxidation in the presence of a mixture of oxidized and reduced glutathione. When the improved sample preparation conditions were used, IgG4 mAb freshly isolated from cells contained 4.5-15% half-antibody, indicating that equilibration of the interchain and intrachain hinge disulfide pairing was not always attained in cells.

The contribution of Fc effector mechanisms in the efficacy of anti-CD154 immunotherapy depends on the nature of the immune challenge.

Blockade of the CD154-CD40 co-stimulatory pathway with anti-CD154 mAbs has shown impressive efficacy in models of autoimmunity and allotransplantation. Clinical benefit was also demonstrated in systemic lupus erythematosus (SLE) and idiopathic thrombocytopenia patients with the humanized anti-CD154 mAb, 5C8 (hu5C8). However, thromboembolic complications that occurred during the course of the hu5C8 clinical trials have proven to be a major setback to the field and safe alternative therapeutics targeting the CD154-CD40 pathway are of great interest. Recently, effector mechanisms have been shown to play a part in anti-CD154 mAb-induced transplant acceptance in murine models, while this issue remains unresolved for humoral-mediated models. Herein, aglycosyl anti-CD154 mAbs with reduced binding to FcgammaR and complement were used as a novel means to test the role of effector mechanisms in non-human primate and murine models not amenable to gene knockout technology. While aglycosyl hu5C8 mAb was relatively ineffective in rhesus renal and islet allotransplantation, it inhibited primary and secondary humoral responses to a protein immunogen in cynomolgus monkeys. Moreover, an aglycosyl, chimeric MR1 mAb (muMR1) prolonged survival and inhibited pathogenic auto-antibody production in a murine model of SLE. Thus, the mechanisms required for efficacy of anti-CD154 mAbs depend on the nature of the immune challenge.

FcR interactions do not play a major role in inhibition of experimental autoimmune encephalomyelitis by anti-CD154 monoclonal antibodies.

It has been demonstrated that anti-CD154 mAb treatment effectively inhibits the development of experimental autoimmune encephalomyelitis (EAE). However, although it appears to prevent the induction of Th1 cells and reactivation of encephalitogenic T cells within the CNS, little information is available regarding the involvement of alternative mechanisms, nor has the contribution of Fc effector mechanisms in this context been addressed. By contrast, efficacy of anti-CD154 mAbs in models of allotransplantation has been reported to involve long-term unresponsiveness, potentially via activation of T regulatory cells, and recently was reported to depend on Fc-dependent functions, such as activated T cell depletion through FcgammaR or complement. In this study we demonstrate that anti-CD154 mAb treatment inhibits EAE development in SJL mice without apparent long-term unresponsiveness or active suppression of disease. To address whether the mechanism of inhibition of EAE by anti-CD154 mAb depends on its Fc effector interactions, we compared an anti-CD154 mAb with its aglycosyl counterpart with severely impaired FcgammaR binding and reduced complement binding activity with regard to their ability to inhibit clinical signs of EAE and report that both forms of the Ab are similarly protective. This observation was largely confirmed by the extent of leukocyte infiltration of the CNS; however, mice treated with the aglycosyl form may display slightly more proteolipid protein 139-151-specific immune reactivity. It is concluded that FcR interactions do not play a major role in the protective effect of anti-CD154 mAb in the context of EAE, though they may contribute to the full abrogation of peripheral peptide-specific lymphocyte responses.

A monoclonal antibody to alpha1beta1 blocks antigen-induced airway responses in sheep.

The integrin alpha1beta1 (very late antigen-1; CD49a/CD29) is a major adhesion receptor for collagen I, IV, and VI, and its induced expression on activated monocytes and lymphocytes plays a central role in their retention and activation at inflammatory sites in autoimmune pathologies. However, the role of alpha1beta1 in allergic settings has not been explored. In this study, we show that a single 45-mg dose of aerosolized monoclonal antibody AQC2 to the alpha1 chain of human and sheep very late antigen-1, given 30 minutes before challenge, blocks both the allergen-induced late response and the associated airway hyperresponsiveness, functional indicators of allergen-induced inflammation, in sheep. AQC2 does not affect the early response. Consistent with these effects, AQC2 tended to reduce the cell response associated with local antigen instillation. An isotype-matched control antibody had no protective effects. Two humanized versions of AQC2, a wild-type IgG1 and an aglycosyl form of the same monoclonal antibody, which has reduced Fc receptor-mediated effector functions, are equally effective in blocking the antigen-induced late response and airway hyperresponsiveness in the sheep model. These data suggest that mononuclear leukocyte adhesion-dependent pathologies contribute to allergic lung disease and provide proof-of-concept that antagonists of alpha1 integrins may be useful in preventing these events.

Expression of Sonic hedgehog-Fc fusion protein in Pichia pastoris. Identification and control of post-translational, chemical, and proteolytic modifications.

We have investigated the suitability of Pichia pastoris as an expression system for the candidate therapeutic protein, Sonic hedgehog fused to an immunoglobulin Fc domain (Shh-Fc). Sonic hedgehog is a morphogen protein involved in the patterning of a wide range of tissues during animal embryogenesis. The presence of Sonic hedgehog and its receptor, Patched, in adult nervous tissue suggests possible applications for the protein in the treatment of neurodegenerative disease and injury. We have engineered the Shh-Fc fusion protein in order to improve binding affinity and increase systemic exposure in animals. N-terminal sequencing, peptide mapping, mass spectrometry, and other biochemical and biological methods were used to characterize the purified protein. These analyses revealed several unanticipated problems, including thiaproline modification of the N-terminal cysteine, cleavage by a Kex2-like protease at a site near the N-terminus, proteolysis at sites near the hinge, addition of a hexose in the CH3 domain of the Fc region, and several sites of methionine oxidation. Sequence modifications to the protein and changes in fermentation conditions resulted in increased potency and greater consistency of the product. The final product was shown to be biologically active in animal studies.

Crystal structure of the alpha1beta1 integrin I domain in complex with an antibody Fab fragment.

The alpha1beta1 (VLA-1) integrin is a cell-surface receptor for collagen and laminin and has been implicated in biological pathways involved in several pathological processes. These processes may be inhibited by the monoclonal antibody AQC2, which binds with high affinity to human alpha1beta1 integrin. To understand the structural basis of the inhibition we determined the crystal structure of the complex of a chimeric rat/human I domain of the alpha1beta1 integrin and the Fab fragment of humanized AQC2 antibody. The structure of the complex shows that the antibody blocks the collagen binding site of the I domain. An aspartate residue, from the CDR3 loop of the antibody heavy chain, coordinates the MIDAS metal ion in a manner similar to that of a glutamate residue from collagen. Substitution of the aspartate residue by alanine or arginine results in significant reduction of antibody binding affinity. Interestingly, although the mode of metal ion coordination resembles that of the open conformation, the I domain maintains an overall closed conformation previously observed only for unliganded I domains.

Long-acting forms of Sonic hedgehog with improved pharmacokinetic and pharmacodynamic properties are efficacious in a nerve injury model.

The therapeutic effects of the Sonic hedgehog (Shh) have been difficult to evaluate because of its relatively short serum half-life. To address this issue polyethylene glycol modification (PEGylation) was investigated as an approach to improve systemic exposure. Shh was PEGylated by a targeted approach using cysteines that were engineered into the protein by site-directed mutagenesis as the sites of attachment. Sixteen different versions of the protein containing one, two, three, or four sites of attachment were characterized. Two forms were selected for extensive testing in animals, Shh A192C, which provided a single site for PEGylation, and Shh A192C/N91C, which provided two sites. The PEGylated proteins were evaluated for reaction specificity by SDS-PAGE and peptide mapping, in vitro potency, pharmacokinetic and pharmacodynamic properties, and efficacy in a sciatic nerve injury model. Targeted PEGylation was highly selective for the engineered cysteines and had no deleterious effect on Shh function in vitro. Systemic clearance values in rats decreased from 117.4 mL/h/kg for unmodified Shh to 29.4 mL/h/kg for mono-PEGylated Shh A192C that was modified with 20 kDa PEG-maleimide and to 2.5 mL/h/kg for di-PEGylated Shh A192C/N91C modified with 2, 20 kDa PEG vinylsulfone adducts. Serum half-life increased from 1 h for unmodified Shh to 7.0 and 12.6 h for the mono- and di-PEGylated products. These changes in clearance and half-life resulted in higher serum levels of Shh in the PEG-Shh-treated animals. In Ptc-LacZ knock-in mice expressing lacZ under regulation of the Shh receptor Patched, about a 10-fold lower dose of PEG-Shh was needed to induce beta-galactosidase than for the unmodified protein. Therapeutic treatment of mice with PEG-Shh enhanced the regeneration of injured sciatic nerves. These studies demonstrate that targeted PEGylation greatly alters the pharmacokinetic and pharmacodynamic properties of Shh, resulting in a form with improved pharmaceutical properties.