Diligent preclinical and clinical development led to the approval of the CD20 antibody rituximab (C2B8) (Grillo-Lopez et al, 1999), which has significantly improved the therapeutic outcome and has become standard of care for most B cell lymphoma patients. To build on this success for second-generation antibodies, effector mechanisms of CD20 antibodies have been intensively investigated. These studies suggested that CD20 antibodies should be grouped into two classes: while class I antibodies (e.g. rituximab, ofatumumab) are particularly effective in recruiting CD20 into lipid rafts and in mediating complement-dependent cytotoxicity (CDC), class II antibodies (e.g. tositumomab, obinutuzumab) preferentially mediate direct lymphoma cell killing (Glennie et al, 2007). Among the class I antibodies, ofatumumab proved to be more efficient than rituximab in triggering CDC against low CD20 expressing chronic lymphocytic leukaemia (CLL) cells (Teeling et al, 2004) and was subsequently approved for this indication. The classical pathway of CDC is typically initiated after complement component 1, subcomponent q (C1q) interacts with its specific binding site in the Fc region of antibodies which have bound to target antigens on cell surfaces. The efficiency of CDC is dependent on many factors - including target antigen density and antibody isotype (Bindon et al, 1988; Michaelsen et al, 1991; Dechant et al, 2002). Since CDC is considered as a relevant effector mechanism for class I antibodies (Glennie et al, 2007; Boross & Leusen, 2012), we wondered whether the CDC activity of rituximab could be improved. The current study demonstrated that at low CD20 antigen densities, especially in the case of CLL cells, a human IgG3 version of rituximab (C2B8-IgG3) is more effective in triggering CDC than the approved IgG1 antibody (C2B8-IgG1).
CD20 antibodies C2B8-IgG1 and C2B8-IgG3 were obtained from Invivogen (San Diego, CA, USA). As another independent model, we used the human leucocyte antigen (HLA) class II antibodies F3.3-IgG1 and F3.3-IgG3 (Dechant et al, 2002), which were produced in BHK-21 cells and purified using an anti-kappa affinity matrix. Biochemical analyses were performed to assess purity, assembly and aggregation of the CD20 (Figure S1A,B) and the HLA class II (Figure S2A,B) antibodies. Binding of IgG1 and IgG3 variants to their respective target antigens – as assessed by indirect immunofluorescence – was similar (Fig 1A and Figure S2C). Expression of CD20, HLA class II as well as the complement regulatory proteins CD46 and CD59 on the B lymphoma cell lines ARH-77, Ramos, Daudi and Raji was determined, demonstrating an inverse correlation between CD20 and HLA class II expression levels (Fig 1B; Figure S2D). Next, these lymphoma cells served as targets in CDC experiments using either human serum or whole blood samples from healthy donors as the effector source in 51Cr release assays. Interestingly, the IgG3 variants of C2B8 and F3.3 were significantly more effective than their IgG1 counterparts against cells expressing low target antigen levels, which were ARH-77 in case of CD20 (Fig 1C), and Ramos in case of HLA class II (Figure S2E). Given that low antigen density and superior IgG3-mediated CDC was found for both antigens against the respective target cell lines, antigen density rather than cell line specific factors appeared to be responsible for the observed differences between IgG1 and IgG3.
Subsequently, we analysed the CDC activity of C2B8-IgG1 and –IgG3 against CD5/CD19- positive CLL cells, which characteristically expressed low levels of CD20 and varying amounts of CD55 and CD59 (Fig 2A). Importantly, C2B8-IgG3 but not C2B8-IgG1 significantly triggered CDC, also when CLL cells and autologous serum from the same patient were utilized (Fig 2B,C,E). In assays with human whole blood as the effector source, antibody-mediated target cell killing was blocked by the C5 antibody eculizumab, underscoring CDC and not antibody-dependent cell-mediated cytotoxicity (ADCC) as the pivotal mechanism of action (Figs 1C and 2B+C). Interestingly, the extent of lysis correlated with CD20 expression levels (Fig 2D), but not with CD55 and CD59 expression (data not shown).
Complement-dependent cytotoxicity has been discussed as an important attack mechanism of CD20 antibodies to combat lymphoma cells (Glennie et al, 2007; Boross & Leusen, 2012). However, systematical analyses revealed rituximab not to induce CDC at low target antigen densities (van Meerten et al, 2006). Thus, several approaches to enhance CDC have been proposed, but these strategies were also not reported to induce CDC against tumour cells expressing low target antigen levels (Boross & Leusen, 2012). Previous studies demonstrated differences between the four human IgG subclasses in their ability to induce CDC (Bindon et al, 1988; Dechant et al, 2002), with human IgG3 being the most efficient isotype at low hapten densities (Michaelsen et al, 1991). Superior CDC, triggered by the IgG3 isotype against low antigen expressing target cells was considered to be evoked by the higher molecule flexibility and by the extended length of the hinge region (63 vs. 15 aa for IgG3 and IgG1, respectively). These structural differences between the two isotypes may allow the IgG3 molecule to react with more widely spaced antigens to fix C1q (Brekke et al, 1995). However, in comparison to IgG1, human IgG3 poses additional challenges for development as a pharmaceutical drug. For example, human IgG3 cannot be purified via protein A and tends to form aggregates, which we also observed to a minor extent in our C2B8-IgG3 preparation (Figure S1B). Importantly, our functional studies were confirmed with highly purified monomeric IgG3 (Figure S3). Furthermore, human IgG3 is commonly known for its reduced serum half-life compared to IgG1. This limitation can be overcome by using a recently described IgG3 allotype (H435 vs. R435), which is expected to have a similar half-life as human IgG1 (Stapleton et al, 2011).
In conclusion, the present study systematically analysed the impact of an isotype switch, from IgG1 to IgG3, on rituximab's capacity to trigger CDC against B lymphoma cells. Additional studies will be required to assess the potential of this antibody to trigger direct tumour cell killing and ADCC by different effector cell populations. Notably, the IgG3 variant of rituximab resulted in enhanced complement-mediated cytolysis at low target antigen expression and may therefore be beneficial in clinical settings where low CD20 levels are expected - such as CLL.