scFv-based "Grababody" as a general strategy to improve recruitment of immune effector cells to antibody-targeted tumors

Abstract

Recruitment of immune cells to tumor cells targeted by a therapeutic antibody can heighten the antitumor efficacy of the antibody. For example, p185(her2/neu)-targeting antibodies not only downregulate the p185(her2/neu) kinase (ERBB2) but also trigger complement-dependent cytotoxicity (CDC) and antibody-dependent cellular cytotoxicity (ADCC) through the antibody Fc region. Here, we describe a generalized strategy to improve immune cell recruitment to targeted cancer cells, using a modified scFv antibody we call a "Grababody" that binds the target protein and endogenous immunoglobulins. The model system we used to illustrate the use of this platform recognizes p185(her2/neu) and includes an IgG binding domain. The recombinant scFv Grababody that was created recruited circulating human IgGs and attracted immune cells carrying Fc receptors to tumor cells that expressed p185(her2/neu). The presence of the IgG binding domain significantly enhanced CDC and ADCC activity and improved antitumor activity in vivo. Our results illustrate a novel general approach to improve antibody-like proteins for therapeutic applications.

Figure 1. Grababody 4D5scFvZZ binds to cell surface p185^her2/neu

T6-17 (A, B), SKBR3 (C, D), and BT474 (E, F) cells with different expression levels of p185^her2/neu were prepared for Fluorescence-activated cell sorting (FACS) as described under Materials and Methods. Histograms represent staining with different dose of 4D5scFv or 4D5scFvZZ, as indicated in the figure, followed by His-Probe antibody and Alexa488-conjugated goat anti-rabbit antibodies (solid lines). The gray peak represents the control staining with only the His-Probe antibody and the secondary antibody. A mouse cell line NIH3T3 (G, H), which is of the same background of T6-17, and a human cell line Jurkat (I, J) were used as control cell lines, showing no significant binding to either 4D5scFv or 4D5scFvZZ.

Figure 2. 4D5scFvZZ binds to immobilized receptor and further capture human IgG molecules

A & B: Direct binding of 4D5scFv and 4D5scFvZZ to immobilized p185^her2/neu subdomain IV, D4(508-577). To show that the 4D5scFvZZ Grababody can simultaneously bind to both the receptor and an human IgG molecule, D4(508-577)-bound 4D5scFvZZ was further exposed to human IgG (green line) or buffer. Two antibodies with human Fc were tested: C. C225 (2 mg/ml); D. 2C4 (0.05 mg/ml). Neither C225 nor 2C4 binds to D4(508-577).

Figure 3. 4D5scFvZZ induces CDC towards cells expressing p185^her2/neu

T6-17 cells were incubated with h4D5, 4D5scFv or 4D5scFvZZ in the presence of either 5% FBS or mouse serum. Cell viability was measured by standard MTT assay. Growth inhibition was determined by this formula: [(control wells−treated wells)/control wells] × 100%. t test was performed to compare statistical difference between groups. Although all treatment groups showed statistical inhibition as compared with control (p < 0.05), we consider >20% inhibition to be biologically significant. Both 4D5scFv and 4D5scFvZZ showed dose dependent inhibition activity in the presence of mouse serum.

Figure 4. ADCC activity mediated by 4D5scFvZZ

ADCC activity of human PBMC against breast cancer BT474 cells was tested in the presence of 4D5scFvZZ (1 or 10 μg/ml) and human IgG (100 μg/ml) (A). A significant increase of ADCC activity was observed when 10 μg/ml of 4D5scFvZZ was used (t test, p< 0.05) as compared with human IgG alone. In contrast, no activity was detected when 4D5scFv was used under similar conditions. ADCC activity of 4D5scFvZZ was dependent on the supplemented human IgG (B). In the absence of human IgG, only the h4D5 antibody showed very significant ADCC activity (p < 0.01). No significant ADCC activity was detected for 4D5scFvZZ on control cell lines A431 (C) and NIH3T3 (D).

Figure 5. Comparison of in vivo activity of 4D5scFvZZ and 4D5scFv

5 × 10⁴ T6–17 cells were injected s.c. into nude mice to induce tumor growth. i.p. treatments with 4D5scFv or 4D5scFvZZ were provided to mice (10 mg/kg/dose, three times per week) started one day after inoculation. Error bars represent the standard error of mean. **: The size of tumors was very significantly different from the controls (t test, p < 0.01). No statistical difference was observed for 4D5scFv treatment although more mice were used. Control: n= 9; 4D5scFv: n=11; 4D5scFvZZ: n=4.