Modifying an immunogenic epitope on a therapeutic protein: a step towards an improved system for antibody-directed enzyme prodrug therapy (ADEPT)

Abstract

Carboxypeptidase G2 (CP) is a bacterial enzyme, which is targeted to tumours by an antitumour antibody for local prodrug activation in antibody-directed enzyme prodrug therapy (ADEPT). Repeated cycles of ADEPT are desirable but are hampered by human antibody response to CP (HACA). To address this, we aimed to identify and modify clinically important immunogenic sites on MFECP, a recombinant fusion protein of CP with MFE-23, a single chain Fv (scFv) antibody. A discontinuous conformational epitope at the C-terminus of the CP previously identified by the CM79 scFv antibody (CM79-identified epitope) was chosen for study. Modification of MFECP was achieved by mutations of the CM79-identified epitope or by addition of a hexahistidine tag (His-tag) to the C-terminus of MFECP, which forms part of the epitope. Murine immunisation experiments with modified MFECP showed no significant antibody response to the CM79-identified epitope compared to A5CP, an unmodified version of CP chemically conjugated to an F(ab)(2) antibody. Success of modification was also demonstrated in humans because patients treated with His-tagged MFECP had a significantly reduced antibody response to the CM79-identified epitope, compared to patients given A5CP. Moreover, the polyclonal antibody response to CP was delayed in both mice and patients given modified MFECP. This increases the prospect of repeated treatment with ADEPT for effective cancer treatment.

Figure 1

In the first stage of ADEPT, the intravenously administered antibody–enzyme fusion protein MFECPHis is allowed to localise in the tumour. In the second stage, prodrug ZD2767P, based on a bis-iodo-phenol mustard is given after clearance of enzyme from the circulation. CP cleaves the glutamate moiety to generate the active drug ZD2767D.

Figure 2

Model of CP showing the discontinuous CM79-identified epitope (region 1 residues 157–163: KEYGVRD coloured in blue, region 2 residues 412–415 coloured in red). The graphic representation was produced with MOLMOL (Koradi et al, 1996).

Figure 3

Schematic representation of the variants of MFE-CP. Panel A illustrates the genetic constructs of the fusion proteins expressed in E. coli after transfection with pPM331 (MFECP) and pDP161 (MFEdmCP), panel B the fusion proteins expressed in P. pastoris with a C-terminal His-tag after transfection with pPIC001 (MFECPHis) and pPIC002 (MFEdmCPHis).

Figure 4

(A) SDS – PAGE and (B) Western blot of fusion proteins expressed in E. coli (1) molecular weight markers (2) MFECP (3) MFEdmCP. (C) SDS – PAGE and (D) Western blot fusion proteins expressed in P. pastoris (4) MFECPHis (5) MFEdmCPHis (6) molecular weight markers. Major band representing fusion proteins indicated by arrow.

Figure 5

(A) Inhibition of binding of biotinylated CM79 antibody by sera of patients after administration of MFECPHis and A5CP. Binding of CM79 antibody in immunised sera was compared to CM79 antibody binding in normal human serum. Statistical comparison of the reduction of CM79 antibody binding after treatment with A5CP with the reduction of CM79 antibody binding after treatment with MFECPHis was significant (P=0.00003 for the comparison of all available sera after administration of MFECPHis and A5CP; P=0.023 for the comparison of HACA-positive sera only, unpaired Student's t-test). (B) Results were confirmed on CP-coated plates (P=0.0001 for the comparison of all available sera after administration of MFECPHis and A5CP, P=0.03 for the comparison of HACA-positive sera only, unpaired Student's t-test). Bars indicate standard deviation.

Figure 6

Anti-CP antibody formation in mice after 50 μg MFECPHis (A), MFEdmCPHis (B) and A5CP (C) i.p. every 14 days for four doses. Individual bars represent mice with measurable anti-CP antibodies in each group of eight mice. Comparison of the number of injections prior to a measurable anti-CP antibody response showed no statistically significant difference between groups injected with MFECPHis and MFEdmCPHis (P=0.203; Fisher's exact test), while comparison of the groups immunised with A5CP and MFECPHis showed a statistical trend indicating that injections could be repeated more frequently after MFECPHis compared to A5CP (P=0.077; Fisher's exact test).

Figure 7

HACA formation in patients 42 days after single administration of MFECPHis, each bar represents an individual patient, the dashed line indicates the cutoff for HACA positive sera. Three standard deviations were added to the mean of a pool of pretreatment sera diluted 1/100 to give the cutoff for HACA-positive sera.