Will sialylation change intravenous immunoglobulin therapy in the future?

Abstract

Intravenous immunoglobulin (IVIg) is a widely used treatment for immunodeficiencies and inflammatory autoimmune diseases. Multiple mechanisms of action for IVIg have been proposed, with some involving the Fab region and others related to the Fc region of the immunoglobulin (Ig)G molecule. Changing the glycan at the glycosylation site in the CH2 pocket of the Fc region (for example, via fucosylation), is known to modify the Fc effector function. In a series of publications between 2006 and 2008, it was proposed by Jeffrey Ravetch's group that the addition of a terminal sialic acid to this region was responsible for the anti-inflammatory properties of IVIg 1–3. Based on data generated in the K/BxN arthritis model, the following mechanism of action of IVIg was proposed: the sialylated Fc region first binds to splenic dendritic-cell-specific intercellular adhesion molecule 3 (ICAM-3) grabbing non-integrin (DC-SIGN in humans), which results in production of interleukin (IL)-33, leading to expansion of IL-4-producing basophils and up-regulation the expression of FcγRIIB receptors, thereby taking the inflammatory system into a non-inflammatory state 4. To test this hypothesis, we carried out studies in animal models of idiopathic thrombocytopenic purpura/immune thrombocytopenia (ITP) 5; multiple sclerosis (MS), using the experimental autoimmune encephalomyelitis (EAE) model 6; and rheumatoid arthritis (RA) 7, using the collagen antibody-induced arthritis (CAbIA) and K/BxN models. We assessed the effects of IgG sialylation in these animal models by comparing sialic acid-enriched and -depleted IgG fractions with the untreated product. To achieve the highly sialylated IgG fraction, we used affinity chromatography with sialic acid-specific Sambucus nigra agglutinin (SNA) lectin. Desialylation was achieved using neuraminidase (NAse) to produce the ‘negative’ material. There are two important factors to consider when using the SNA lectin fractionation method 8. First, IVIg fractionated over an SNA column shows increased sialylation mainly in the Fab region, as measured by glycoanalysis with high-performance liquid chromatography (HPLC), or liquid chromatography–mass spectrometry (LC-MS). The findings from both assays indicated that the SNA-binding fraction represents approximately 10% of the IgG molecules in IVIg. However, Fc-sialylation as assessed by LC-MS was increased in one subfraction, representing only 1–2% of the IgG molecule and only by a very small amount 8. Secondly, we have demonstrated that lectin chromatography can result in a skewed antibody pattern of the SNA-binding fraction 8. Antigen specificity of IgG fractions was tested by enzyme-linked immunosorbent assay (ELISA) with a range of pathogen-derived and autoantigens, and demonstrated abnormal antibody concentration patterns in the eluted material compared to the starting material. Our recent study in the antibody-induced K/BxN and CAbIA RA models in mice 7 set out to establish the role of sialylation in antibody-driven inflammation. The therapeutic effect of IVIg and its proposed mechanisms of action were studied using both prophylactic and therapeutic protocols. Prophylactic IVIg protocols have been shown in previous studies to protect against arthritis in the K/BxN model 1,3, and this study demonstrated the efficacy of prophylaxis in the CAbIA model; disease severity was reduced significantly (P < 0·01) in a dose-dependent manner. For the therapeutic protocol (which is more relevant to the treatment of human RA than prophylaxis) in the CAbIA model, disease was induced in the experimental animals with an antibody mix at day 0 and boosted with lipopolysaccharide (LPS) at day 3 to exacerbate symptoms 7. The mice were injected with either untreated, desialylated IVIg (NAse IVIg) or Fc at day 5 and clinical scores were determined based on the degree of inflammation in the digits or paws (Fig. 1). Figure 1 Lack of requirement for sialylation of intravenous immunoglobulin (IVIg) or Fc for suppression of CAbIA pathology (therapeutic protocol) (reproduced with permission from 7, copyright 2014, The American Association of Immunologists, Inc.). At day 5, mice ... With both IVIg and Fc, there was a dose-dependent improvement in clinical scores (no effect was seen with F(ab')2 in a separate test, confirming that the anti-inflammatory action of IVIg is Fc-mediated). In this therapeutic CAbIA model, no loss of treatment efficacy was observed with desialylated IVIg or Fc (Fig. 1). As these results contrasted with previous findings using a prophylactic treatment in the K/BxN animal model, we repeated the test using the prophylactic protocol with untreated versus NAse-treated IVIg or Fc in the CAbIA model, and confirmed that desialylation had no effect on the prophylactic efficacy of IVIg/Fc. Moreover, no enhanced effects were observed when using sialic acid-enriched (by SNA fractionation) Fc fractions prophylactically, compared to untreated Fc, in the CAbIA model. Of note, it was observed that highly sialylated Fc bound to Protein A in a similar manner to untreated Fc, indicating that the fraction had not been compromised by the SNA fractionation process 7. To further clarify our contrasting results in the CAbIA model with those reported previously in the K/BxN model 1,3, we repeated the prophylactic protocol in the K/BxN model. Our results confirmed that sialylation of IVIg or Fc was not required for suppression of inflammation in this model 7. Because basophils have been suggested as key cell mediators in the anti-inflammatory mechanism of IVIg 4, as described above, these were also examined in the CAbIA model 7. Mice were injected with basophil-depleting monoclonal antibody (mAb) MAR-1 twice daily for days 1–3 before beginning IVIg treatment (day 5). No difference was observed during the next 7 days between the disease response of mice treated with MAR-1 and those treated with an isotype control. In the K/BxN model, this was tested with a prophylactic approach; mice were injected with MAR-1 for 10 consecutive days, beginning 2 days prior to IVIg treatment. When disease was introduced at day 0, the IVIg prophylaxis protected the mice from arthritis as normal, as assessed by both clinical score and caliper-measurement of ankle swelling (P < 0·001). In agreement with previous reports in animal models of ITP 5 and MS 6, our study found no evidence that the mechanism of action for the anti-inflammatory effect of IVIg was dependent on Fc-sialylation in the CAbIA and K/BxN RA mouse models. In addition to the modes of action described here, we have recently identified cell surface receptor and intracellular pathways by which IVIg mediates inflammation via attenuation of interferon (IFN)-α in an in vitro model of systemic lupus erythematosus (SLE) 9. These multiple mechanisms of action indicate that further research is required in order to fully understand the anti-inflammatory effects of IVIg.