Mechanisms of Discrepancy Between CD33 Expression and Gemtuzumab Ozogamicin-Induced Cytotoxicity in Leukemia Cells.

Abstract 4812

Gemtuzumab ozogamicin (GO) is an immunoconjugate of an anti-CD33 antibody and a toxic calicheamicin-γ₁ derivative that facilitates cellular uptake of the latter drug. CD33 is selectively expressed on myeloid cells, especially on acute myeloid leukemia (AML) cells, but usually not on non-hematopoietic cells, and therefore the clinical use is confined to AML. Despite its predominant expression on AML cells, GO induces complete remission only in 30% of recurrent AML. The major side-effects of GO include not only myelosuppression but also hepatotoxicity, although CD33 is not expressed on hepatobiliary cells. It suggests that the efficacy of GO is not strictly regulated by CD33 expression. We investigated relationship between CD33 expression and GO cytotoxicity by using leukemia cell lines, and classified them into four groups; group A defined as CD33(+) and cytotoxicity (+), group B as CD33(+) and cytotoxicity (-), group C as CD33(-) and cytotoxicity (+), and group D as CD33(-) and cytotoxicity (-). We selected four cell lines, HL-60, MEG-01, MOLT-3 and K562 which belong to group A, B, C and D, respectively, and were used for further experiment. CD33 was expressed on HL-60 but not on MOLT-3. However, GO-induced cytotoxicity was seen in both cells. We therefore compared effects of GO in HL-60 and MOLT-3. When incubated with GO for 72 hours, viability of HL-60 significantly decreased at 10 ng/mL of GO, while cytotoxic effect of GO against MOLT-3 became apparent when GO concentration was raised to 100 ng/mL. Viability of HL-60 markedly decreased when treated with GO for 30 minutes and then cultured in medium without containing GO for 72 hours, but no cytotoxic effect was observed in MOLT-3 unless culture with GO was prolonged to more than 6 hours. GO-induced apoptosis was detected in both HL-60 and MOLT-3 by Annexin V assay and TUNEL assay. The pretreatment with unconjugated anti-CD33 antibody (hP67.6) resulted in reduced cytotoxic effects of GO in HL-60, but not in MOLT-3. It indicates that GO induces apoptosis in CD33-non-expressing MOLT-3. Higher doses and longer exposure of GO treatment are necessary to induce cytotoxicity against MOLT-3 when compared with HL-60. GO might cause cytotoxicity against MOLT-3 through drug absorption pathway other than CD33-mediated endocytosis. Although HL-60 and MEG-01 expressed CD33, GO induced cytotoxicity on HL-60, but not on MEG-01. We then investigated internalization of antibody-bound CD33. After incubated with hP67.6 on ice, cells were cultured in antibody-free medium, and remaining cell surface-associated CD33 was measured. The loss of cell surface CD33 was observed in MEG-01 as well as HL-60, showing that internalization of CD33 following antibody binding occurred in MEG-01. It indicates that resistance to GO-induced cytotoxicity in MEG-01 is not associated with internalization property. Drug transport pathway other than receptor-mediated endocytosis was investigated. Cells were cultured in rhodamine 123 (Rh123)-containing medium for 30 minutes and the influx of Rh123 was assessed by the fluorescence activity of Rh123-containing cells. Cells were then cultured in medium without containing Rh123 for further 3 hours, and the efflux of Rh123 was analyzed by measuring the reduction curve of fluorescence activity. Rh123 accumulated in MEG-01 twice higher than HL-60. However, fluorescence activity of Rh123 decreased to 20% after 3 hours, while it remained unchanged in HL-60, indicating that drug efflux was activated in MEG-01. Multidrug resistance (MDR)-1 protein was strongly expressed in MEG-01. When cyclosporin-A, a modulator of MDR-1, was given with GO to MEG-01, GO-induced cytotoxicity was significantly increased, while no such effect was observed in HL-60. It suggests that p-glycoprotein-mediated efflux of GO is one of the resistance mechanisms in MEG-01.