Differential Patterns of Gene Expression and CD20 Localisation by the Type II Anti-CD20 Antibody Obinutuzumab (GA101) Compared with the Type I Antibody Rituximab Suggests Binding to Functionally Distinct CD20 Subpopulations on B-Cell Lymphoma Cell Lines,

Abstract 3721

The anti-CD20 antibody rituximab has become central to the treatment of B-cell malignancies over the last decade. Recently, it has been shown that anti-CD20 antibodies can be divided into two types based on their mechanisms of action on B cells. Rituximab is a type I antibody that redistributes CD20 into lipid rafts and promotes complement-dependent cytotoxicity (CDC), while the type II, glycoengineered antibody GA101 has lower CDC activity but higher antibody-dependent cellular cytotoxicity and direct cell death activity. In preclinical studies GA101 was superior to rituximab in B-cell killing in vitro, depletion of B cells from whole blood, and inhibition of tumour cell growth in lymphoma xenograft models. GA101 is currently being evaluated in Phase II/III trials, including comparative studies with rituximab. To investigate the differences in direct effects of GA101 and rituximab on B-cell lymphoma signaling, we have analysed the effects of antibody binding on gene expression in different B-cell lines using a GeneChip Human Genome U133 Plus 2.0 Array (Affymetrix). Rituximab and GA101 rapidly induced gene expression changes in SUDHL4 and Z138 cells, including regulation of genes associated with B-cell-receptor activation such as EGR2, BCL2A1, RGS1 and NAB2. The effects on gene expression differed markedly between different cell lines and between the two antibodies. SUDHL4 cells showed pronounced changes in the gene expression pattern to rituximab treatment, while Z138 cells, which represent a different B-cell stage, showed less pronounced changes in gene expression. The reverse was true for GA101, suggesting not only that the signaling mediated by CD20 differs in different cell lines, but also that in a given cell line the two types of antibodies bind CD20 molecules with different signaling capacity. For each cell line, gene expression induced by other type I antibodies (LT20, 2H7, MEM97) was more like rituximab and that induced by other type II antibodies (H299/B1, BH20) was more like GA101 in terms of the number of genes regulated and the magnitude of changes in expression. Unbiased hierarchical clustering analysis of gene expression in SUDHL4 could discriminate type I from type II antibodies, confirming that the two classes of antibody recognised CD20 complexes with inherently different signalling capacities. By confocal and time-lapse microscopy using different fluorophores, rituximab and GA101 localised to different compartments on the membrane of lymphoma cells. GA101/CD20 complexes were relatively static and predominantly associated with sites of cell–cell contact, while rituximab/CD20 complexes were highly dynamic and predominantly outside areas of contact. These findings suggest that type II antibodies such as GA101 bind distinct subpopulations of CD20 compared with type I antibodies such as rituximab, accounting for the differences in mechanisms of action and anti-tumour activity between these antibodies.