Antibody-Directed Enzyme Prodrug Therapy: A Promising Approach for a Selective Treatment of Cancer Based on Prodrugs and Monoclonal Antibodies


*Lutz F. Tietze,


The antibody-directed enzyme prodrug therapy allows a selective liberation of cytotoxic agents from non-toxic prodrugs in cancerous tissue by targeted antibody–enzyme conjugates. We have developed a series of novel glycosidic prodrugs based on the natural antibiotic CC-1065 and the duocarmycins, which are up to 4800 times less toxic than the drugs liberated from these prodrugs in the presence of the activating enzyme (e.g., β-d-galactosidase). Furthermore, the drugs show very high cytotoxicities with IC50 values of as low as 4.5 pm. In this report, we summarize our recent results on the development and biological evaluation of these novel third-generation prodrugs with higher water solubility, higher difference in cytotoxicity between the prodrugs and the corresponding drugs and improved cytotoxicity of the drugs as compared with previous compounds.

Cancer is one of the main causes of death, accounting for 13% of the mortality rate worldwide (1). Conventional cancer treatments are based on surgery, radiation therapy, and chemotherapy. Unfortunately, most chemotherapeutics in clinical use affect all fast dividing cells and thus not only cancer cells, but also normal cells with a high proliferation rate as bone marrow cells and intestinal epithelial cells. Consequently, these anticancer agents cause severe side effects like immunosuppression and gastrointenstinal disorder. These disorders are often dose-limiting or even require the termination of the treatment. Limiting the cytotoxic action of the chemotherapeutics to cancer cells is therefore highly desirable. Novel approaches aiming at higher selectivity include immunotherapy, hormonal therapy, inhibition of angiogenesis, targeting deregulated proteins of cancer cells using enzyme inhibitors, antibodies, or small peptides and the use of prodrugs by employing either drug targeting or a selective release of drugs from prodrugs in the tumor tissue (2–9).

Antibody-directed enzyme prodrug therapy (ADEPT) is a promising approach for selectively targeting cancer cells. In ADEPT, engineered conjugates of enzymes and monoclonal antibodies are administered in a first step (Figure 1) (10–20). After binding of the antibody–enzyme conjugates to tumor-associated antigens and clearance of the remaining unbound conjugates from the blood stream, a prodrug is administered in a second step. The prodrug is then transformed into a highly cytotoxic compound selectively in the tumor tissue by the targeted antibody–enzyme conjugate, thus achieving a selective killing of tumor cells.

Figure 1.

 Selective cancer therapy using antibody-directed enzyme prodrug therapy (ADEPT).

As the process of the drug’s release is catalytic, a large amount of prodrug is converted to the corresponding drug in the tumor tissue by every enzyme molecule. Hence, a high concentration of the drug is achieved locally and tumor cells are killed efficiently. Furthermore, as the drug is released extracellularly and has a low molecular weight, neighboring tumor cells, that do not express the tumor-associated antigen, are killed by drugs reaching these cells by means of diffusion. This bystander effect helps to eliminate cells that would not be susceptible to the treatment with classical antibody–drug conjugates (21). For a successful tumor therapy, the prodrug needs to display low systemic toxicity, low intake into cells, and good water solubility. Additionally, the drug should be liberated from the prodrug efficiently by an exogenous enzyme and should display high cytotoxicity, high intake into cells, and low molecular weight. Furthermore, the antibody–enzyme conjugate needs to bind specifically to the tumor cell surface and should display low immunogenicity, high enzymatic activity and should not be inactivated by the prodrug or the drug.

First- and Second-Generation Prodrugs

We started our work in the field of developing selective anticancer agents in 1981 with the synthesis of acetal glycosides like 1 (Figure 2) that are activated to give a reactive cyclophosphamide derivative as 2 at the low pH in the tumor tissue (down to pH 5.6 in cancerous tissue compared with pH 7.4 in normal tissue under hyperglycaemic conditions) (22–29) or by using a glycohydrolase–antibody conjugate binding to the tumor cells (12,28).

Figure 2.

 Activation of the acetal glycoside BE-1 (1), a first-generation prodrug, by an antibody–β-glucosidase conjugate to give the alkylating agent 2 as the active drug.

Since early investigations using prodrug 1 in the presence of an antibody–β-glucosidase conjugate did not reveal therapeutic effects, we proposed that for a successful cancer therapy the cytotoxic agent derived from the prodrug should have an IC50 value of less than 10 nm (12). Furthermore, the ratio of cytotoxicity between the prodrug and the corresponding drug, which we define as the QIC50 value [QIC50 = IC50 (prodrug)/IC50 (prodrug + activating enzyme), IC50: concentration required for 50% growth inhibition of target cells] should exceed 1000. Taking into account these premises, we developed a series of second-generation prodrugs as 3, which can be activated by the enzyme β-d-galactosidase to give the corresponding drugs such as 4 (Figure 3) (12,13). These drugs are analogs of the highly cytotoxic natural antibiotics CC-1065 (6) (30) and the duocarmycins [e.g., duocarmycin SA (5) (31)] and exert their cytotoxic action presumably through a sequence-selective alkylation of double-stranded DNA (32,33).

Figure 3.

 Second-generation prodrug 3 activated by β-d-galactosidase to provide the highly cytotoxic drug 4 as an analog of the natural antitumor agents duocarmycin SA (5) and CC-1065 (6).

Design, Synthesis, and Biological Evaluation of Novel Third-Generation Prodrugs

The spirocyclopropyl moiety as the pharmacophoric group in analogs as 9 of CC-1065 (6) and the duocarmycins can be formed in situ by an intramolecular cyclization starting from the corresponding seco-drugs 8, containing a chloromethyl- or a 1-chloroethyl-moiety and a free phenolic hydroxyl group (Figure 4). We have shown that the fast cyclization can be halted by transformation of the hydroxyl group into a glycoside as in 7, which on the contrary can be activated using a glycohydrolase to give back the seco-drug (12,13). Thus, relatively untoxic prodrugs can be generated, which can be activated to the toxic drugs selectively in the tumor tissue by a targeted enzyme.

Figure 4.

 Enzymatic activation of the prodrug 7 containing a detoxifing sugar moiety to give the seco-drug 8 followed by a fast cyclization of the seco-drug to afford the active drug 9.

In our recent investigations, we varied the pharmacophoric group and the DNA-binding unit as well as the detoxifying carbohydrate moiety to design compounds with improved characteristics for the use in ADEPT and prodrug monotherapy (34–39). Thereby, we found that the types of the DNA-binding unit, the pharmacophoric group, and the sugar moiety do not only strongly influence the cytotoxicity of the formed drugs, but also the QIC50 values, i.e., the difference in cytotoxicity between the prodrugs and the corresponding drugs.

For the synthesis of the novel prodrugs, the benzyl protected substrates 15 and 16, respectively, were prepared either as a racemic mixture, followed by chromatographic resolution using HPLC on a chiral stationary phase (35,38), or in a more efficient way by an enantioselective synthesis employing a Sharpless-Katsuki epoxidation, a metal-mediated ring-closure under epoxide-opening and an Appel reaction in the last step (Figure 5) (35,40).

Figure 5.

 Enantioselective synthesis of precursors 15 and 16, respectively.

Based on the experience with galactoside 3, the galactosidic prodrugs 19a–e containing a methyl-seco-CBI-skeleton and the galactosidic prodrugs 20a,b containing a seco-CBI-skeleton were synthesized (35,38). The prodrugs can be activated enzymatically by cleavage of the sugar moiety using commercially available β-d-galactosidase from Escherichia coli to give the corresponding seco-drugs 21a–e and 22a,b, respectively, that cyclize rapidly under loss of HCl to produce the active drugs. All of the new compounds contain a tertiary amino functionality in the side chain as part of the DNA-binding subunit to allow the formation of a salt and thus improving water solubility (41). For the synthesis, cleavage of the benzyl ether moiety in 15 or 16 was followed by glycosidation with the sugar donor 17, deprotection of the secondary amino functionality, and coupling with the respective DNA-binding subunits 18a–e in a one pot procedure (Figure 6).

Figure 6.

 Synthesis of the novel prodrugs 19a–e and 20a,b and of their corresponding seco-drug hydrochlorides 21a–e and 22a,b, respectively.

The last step in the sequence was a solvolysis of the acetic acid esters at the sugar moiety. Thus, starting from enantiopure 15 or 16, the prodrugs 19a–e and 20a,b were obtained in four steps in good 27–43% and 23–31% yield, respectively. The corresponding seco-drugs 21a–e and 22a,b were prepared as hydrochloride salts in 57–69% and 64–83% yield over two steps, respectively, by coupling of 15 to the DNA-binding subunit 18a–e, followed by debenzylation or by debenzylation of 16 and subsequent coupling to the DNA-binding subunit 18a,b.

Kinetic studies using HPLC–MS clearly revealed that the prodrugs were not activated to give the corresponding drugs in cell culture medium without serum at 37 °C in a time frame of 24 h (42). Furthermore, the prodrugs also proved to be stable in vivo as was confirmed, e.g., for 20a by pharmacokinetic studies using rats. Here, even after 6 h post i.v. administration of the prodrug, no drug could be detected in the plasma. In addition, the prodrugs were very good substrates for the exogenous enzyme β-d-galactosidase. Thus, for example, using 60-nmβ-d-galactosidase and 0.8-mm prodrug 19a in cell culture medium, the half life of 19a was less than 35 min. The drug formed from the prodrug in the course of the enzymatic reaction had a chemical half-life of about 4 h under the same conditions. Because of these promising results, we evaluated the biological activity of the new compounds in vitro as well as in vivo.

First, in vitro cytotoxicity assays were carried out using a colony-forming test based on the human tumor colony-forming ability (HTCFA) assay that reflects the proliferation capacity of single cells (Table 1). The cells used for the experiments were coherent cells of the human bronchial carcinoma cell line A549. Incubation was performed with various concentrations of seco-drug hydrochlorides 21a–e and 22a,b and of β-d-galactosidic prodrugs 19a–e and 20a,b in the absence and in the presence of β-d-galactosidase.

Table 1. In vitro cytotoxicity of β-d-galactosidic prodrugs 19a–e and 20a,b in the presence or absence of β-d-galactosidase and of the seco-drug hydrochlorides 21a–e and 22a,b against human bronchial carcinoma cells (A549)
Compound IC50 (nm)IC50 (nm) in the presence of β-d-galactosidase (4 U/mL) QIC50
  1. Cells were exposed to various concentrations of the test substance for 24 h at 37 °C; after 12 days of incubation the clone formation was compared with an untreated control assay and the relative clone forming rate was determined. β-d-galactosidase: Escherichia coli.

(+)-(1S,10R) -19a3.6 × 1030.754800
(+)-(1S,10R) -21a0.75 
(+)-(1S,10R) -19b9.4 × 1020.224300
(+)-(1S,10R) -21b0.20 
(+)-(1S,10R) -19c7.7 × 1035.91300
(+)-(1S,10R) -21c3.8 
(+)-(1S,10R) -19d1.5 × 1032.5600
(+)-(1S,10R) -21d3.7 
(+)-(1S,10R) -19e8.3 × 1020.751100
(+)-(1S,10R) -21e0.80 
(−)-(1S) -20a561.6 × 10−23500
(+)-(1S) -22a2.6 × 10−2 
(−)-(1S) -20b9.54.5 × 10−32100
(−)-(1S) -22b9.0 × 10−3 

The cytotoxicities found for the seco-drug hydrochlorides 21a–e and 22a,b were similar to those observed for the corresponding glycosidic prodrugs 19a–e and 20a,b in the presence of the enzyme β-d-galactosidase. This indicates that the enzyme is able to cleave the glycosidic bond in the prodrugs and is not deactivated during the hydrolytic process. Furthermore, nearly all of the prodrugs met our proposed requirements for a successful use in ADEPT (QIC50 > 1000; IC50 (prodrug + enzyme) <10 nm), with 19a as the best methyl-seco-CBI-prodrug and 20a as the best seco-CBI-prodrug. Both compounds contain a DMAI-DNA-binding subunit (43) and show good water solubility.

As expected, the prodrugs and seco-drugs with the seco-CBI unit have a higher cytotoxicity than the compounds containing the methyl-seco-CBI unit, as an alkylation of cellular targets is less hindered in the absence of the additional methyl group. On the one hand, the exceptionally high cytotoxicity of 22a is expected to enhance antitumor efficiency of the treatment when the corresponding prodrug 20a is applied in an ADEPT approach. Furthermore, 20a is very stable in aqueous solution, thus favoring its intravenous application. On the contrary, prodrug 19a shows a higher QIC50-value than 20a and thus the selectivity of the treatment is expected to be higher using 19a.

To compare the efficiency of our novel compounds with other cytotoxic drugs, additional HTCFA assays were performed on cells of line A549 with carmustine (23), melphalan (24), and doxorubicin (25) (Figure 7). All of these compounds are currently used as anticancer agents in the clinic.

Figure 7.

 The anticancer agents carmustin (23), melphalan (24), and doxorubicin (25).

The results clearly show that the new prodrugs 19a and 20a have a higher biological activity in the presence of the enzyme β-d-galactosidase than the anticancer drugs 23–25 (Figure 8). Furthermore, the general toxicity of the new galactosides is very low and using the prodrugs in combination with different antibody–β-d-galactosidase conjugates, very good responses could be achieved on tumor bearing mice (44).

Figure 8.

 Comparison of the in vitro cytotoxicities of various anticancer agents against human bronchial carcinoma cells of line A549 (HTCFA-assay). (□) (1S)-20a in the presence of β-d-galactosidase (4 U/mL): IC50 = 0.016 nm; (○) (1S,10R)-19a in the presence of β-d-galactosidase (4 U/mL): IC50 = 0.75 nm; (bsl00001) doxorubicin (25): IC50 = 45 nm; (•) melphalan (24): IC50 = 3.4 × 103 nm; (bsl00066) carmustine (23): IC50 = 2.6 × 104 nm.


In recent years, we have designed a series of novel and very promising galactosidic prodrugs for the use in a new approach for selectively targeting cancer cells. These prodrugs can be activated to give highly cytotoxic analogs of CC-1065 (6) and the duocarmycins (e.g., 5) by targeted antibody–enzyme conjugates. The newly synthesized third-generation compounds were prepared in diastereomerically and enantiomerically pure form with high yields and they show good water solubilities. Because of their excellent characteristics (QIC50 values of up to 4800 and a very high cytotoxicity of the corresponding drugs (e.g., IC50 = 4.5 pm), these prodrugs are superior to the compounds described so far by us and others for the use in ADEPT. The most promising candidates for clinical studies are the prodrugs 19a and 20a, which are now further investigated by two pharmaceutical companies.


This work was supported by the Deutsche Forschungsgemeinschaft, the State of Lower Saxony, the Fonds der Chemischen Industrie, the Humboldt Foundation, the DAAD, and the European Community. We are also thankful to several companies for their help as the Bayer-Schering AG and the BASF. Finally, the excellent cooperation with Prof. Dr Alves is acknowledged. B. Krewer is grateful to the Deutsche Telekom Foundation for a PhD scholarship.