Molecular Effects of Rituximab on CLL Cells In Vivo: Rapid Increase in Serum Interferon gamma, Induction of a Distinct Tumor Cell Gene Expression Signature and Loss of CD20 Expression.

Abstract

Proposed mechanisms on how the monoclonal anti-CD20 antibody rituximab (R) depletes B-cells include antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity. In vitro studies have suggested that R induced pro-apoptotic signals contribute to clinical efficacy and may sensitize cells to chemotherapy. To investigate the effect of R on tumor biology in vivo, we analyzed the molecular changes in leukemic cells of 12 previously untreated CLL patients during the first R (375mg/m2) infusion. The median reduction of circulating tumor cells within 24h was 50% (range 0–67%). We first determined whether R affects gene expression in CLL cells obtained before and at 6h and 24h after the start of R. Cells were purified by CD19+ selection and gene expression was measured on Affymetrix HU133A 2.0 arrays. A one-way ANOVA test with a stringent cutoff (false discovery rate of <20%) identified 69 genes whose expression increased >50% at 6h compared to pre treatment, and 31 genes whose expression decreased by >30%. Most of the up-regulated genes are known to be regulated by interferon (IFN) and include the pro-apoptotic genes IRF1, STAT1, FAS and OAS2. Of 12 cytokines assayed in the serum, we found that only IFNy, IL–6, IL–8, IL–10, and TNFa were induced by R with a peak at 2h. Consistent with a dominant role of IFNy on gene expression in the CLL cells, STAT1, a direct and essential mediator of IFNy signaling, was activated in circulating leukemic cells in vivo. In addition, when comparing the response between patients, IFNy serum protein levels correlated strongly with the intensity of the gene expression changes in the tumor cells (r=0.83, p=0.008). We could not detect any IFNy mRNA in CLL cells and conclude that the IFNy is most likely released by NK cells activated through FcyRIII signaling. Considering the long half-life of R, we were surprised to see that both cytokine serum levels and gene expression changes almost completely subsided by 24h. Intriguingly, among the few genes that were down-regulated by treatment, the gene encoding CD20 was the most strongly and consistently affected showing a 50% decrease in expression at 24h. We also assessed CD20 protein levels by Western blotting. Total CD20 levels were markedly decreased already at 6h and by 24h almost all CD20 had been lost. The more rapid and more pronounced decrease of CD20 protein as opposed to mRNA levels is consistent with a process previously described as shaving, during which R bound CD20 is pulled of the cell surface (Kennedy AD, J Immunol. 2004). Despite the absence of a clinical cytokine release syndrome, we observed basically identical changes in serum cytokines and gene expression with subsequent infusions in 2 patients analyzed. In summary, R induced a characteristic gene expression signature in CLL cells that is dominated by IFN response genes, many of which have well characterized pro-apoptotic functions. Thus, our data suggest that signaling for apoptosis is not so much a direct effect of R, but due to a complex immune response to the R coated CLL cells. Modified treatment schedules capable of delivering sustained pro-apoptotic signals hold promise for improved efficacy of R and should be explored.