In 1906, Paul Ehrlich introduced the concept of targeting cancer cells with a “magic bullet”. His magic bullet consisted of using tissue-specific carriers to deliver toxic agents to neoplastic tissues [1,2]. The development of modern cancer therapy is based on discovering and designing drugs that target cancer specific pathways with minimal side effects. Targeted toxins, also called immunotoxins or cytototoxins, are recombinant molecules that specifically bind to surface antigens or receptors overexpressed in cancer, including tumor and endothelial cells [3]. These recombinant proteins consist of a specific antibody or ligand coupled to a toxin protein. The targeted toxins bind to a surface antigen or receptor overexpressed in cancer, such as the epidermal growth factor receptor, transferrin receptor, interleukin-13 or interleukin-4 receptor. Antibodies have been used for targeting chemotherapeutic drugs, toxins, enzymes and radionuclides, with the aim of developing targeted therapies for cancer. Antibodies in the therapy of cancer have evolved from mouse antibodies, to chimeric antibodies, to humanized antibodies. Meanwhile, due to poor penetration of toxins chemically conjugated to monoclonal antibodies (mAbs) [4,5], growth factors or cytokines linked to toxins were also used for targeted cancer therapy and they have been rapidly advanced into clinic trials. The toxin part of the molecule in all clinically used cytotoxins is modified from bacterial or plant toxins, in which the cell recognition domain is replaced with a new targeting moiety from an antibody or carrier ligand. Targeted toxins have the advantage of being highly cytotoxic and easy to manipulate by genetic engineering methods [2,6,7].

Malignant brain tumors such as the glioblastoma multiforme (GBM) are highly lethal tumors and the life expectancy for patients with GBM under the current standards of care is on average 14 months from diagnosis despite maximal therapy with chemotherapy and radiation therapy [8,9]. The poor clinical prognosis associated with malignant brain tumors has led investigators to seek new, innovative methods of treatment. Targeted toxins are extremely cytotoxic to malignant GBM cell lines in vitro. Animal studies have shown prolongation of survival and complete tumor regression when targeted toxins were administered by a variety of routes [7,10,11]. The promising results seen in vivo have formed the basis for proceeding with clinical trials in humans with malignant brain tumors and leptomeningeal neoplasia, in which these agents are administered directly into the tumor or intrathecally, respectively. To date, in these clinical trials, targeted toxins have been delivered safely without significant neurological toxicity, and cytological analysis of cerebrospinal fluid and radiological findings have shown evidence of a therapeutic response. These studies have confirmed the existence of a therapeutic window between normal brain tissue and malignant cells that can be exploited with targeted therapy directed against cancer specific receptors. The successful delivery of targeted toxins directly into malignant brain tumors has established this route of administration as both practical and feasible.

This review summarizes the characteristics of target toxins and the key findings of the important clinical studies with targeted toxins in malignant brain tumor patients. Obstacles to the successful treatment of malignant brain tumors include poor penetration into tumor masses and the immune response to the toxin component. Strategies to overcome these limitations are being pursued. An outlook into future areas of development of targeted toxins will be discussed.

The toxins used in most clinical immunotoxin or cytotoxin construction are made by bacteria or plants. They are very potent in small amounts delivered by these organisms, after natural selection over millions of years. Though structurally and evolutionarily different, Diphtheria toxin (DT) and Pseudomonas aeruginosa exotoxin A (PE) share similar properties of protein synthesis inhibition either by modifying elongation factor-2 or by directly inhibiting the ribosome [12]. Once attached to the overexpressed antigens or receptors on cancer cells, the toxin is endocytosed and transferred via an endosome to either a lysosome or the Golgi apparatus. The toxin and carrier ligand are then separated, allowing the toxin to inhibit protein synthesis. Immunotoxins can inactivate over 200 ribosomes or elongation factor-2s per minute. Furthermore, other mechanisms are also involved for toxins to disrupt the host cell function; for example, AB5 subtilase cytotoxin produced by pathogenic bacteria, such as Shiga toxigenic Escherichia coli (STEC), cleaves the essential endoplasmic reticulum chaperone protein BiP/GRP78, which is key for cell survival [13,14]. A single immunotoxin can kill a cancer cell as compared to 10⁵ molecules of a chemotherapeutic drug that are needed to kill one cancer cell. So these toxins are much more potent when compared to traditional chemotherapeutic drugs.

Most toxins are polypeptides with several domains: a cell recognition chain, which binds to the receptors on the surface of the target cell; a translocation chain, which enables the toxin to cross a membrane to reach the cytosol where essential cell machinery is located; and an inactivation domain, which inactivates some vital cellular process and causes cell death [2,3]. To make an immunotoxin, the cell recognition domain is replaced with a new recognition moiety. The most commonly used toxins in the clinical trials are two bacterial toxins: Diphtheria toxin and Pseudomonas aeruginosa exotoxin A [15].

Diphtheria toxin is a 62 kDa protein secreted by Corynebacterium diphtheria [16,17]. The single polypeptide chain must be enzymatically nicked at an arginine-rich site for the A and B chain to be activated against human cells. Diphtheria toxin (DT) has a cell-binding domain at the C terminus (amino acids 482–539) and the A chain with ADP-ribosylation activity at the N terminus. The A chain catalyzes the transfer of adenosine diphosphate (ADP)-ribose to EF-2, preventing the translocation of peptidyl-t-RNA on ribosomes, thereby blocking protein synthesis and subsequently killing the cell [18,19,20]. A natural ligand for DT on the cell membrane is the heparin-binding epidermal growth factor (EGF)-like precursor [21]. DT undergoes internalization, disulfide bond reduction and proteolytic activation after cell binding, but translocation into the cytoplasm occurs directly from the acidic endocytic compartment. Recombinant DT is made by replacing the C terminal cell-binding domain with a ligand that binds to a growth factor receptor or the Fv fragment of an antibody. The native DT protein consists of 535 amino acids. Variable truncation of the binding segments resulting in 389 and 486 amino acid length toxin conjugates has resulted in the formation of toxins DAB389 and DAB486, respectively [12]. These modified DTs are unable to enter a cell without selective uptake of their carrier ligand by a receptor. Another modification of DT involves substitution of two amino acids in the B chain resulting in a new molecule cross-reacting material-107 (CRM-107) [22,23,24]. This modification reduces the non-specific binding of DT to human cells by 8000-fold, thus increasing the toxin’s tumor-specificity 10,000-fold.

Pseudomonas aeruginosa exotoxin A is a single peptide with three functional domains: domain Ia is the N terminal and cell-binding domain; domain II has translocation activity; and domain III is the C terminal and catalyses the adenosine diphosphate (ADP)-ribosylation that inactivates EF-2, which further blocks protein synthesis and causes cell death. PE binds to the low density lipoprotein receptor-related protein (LRP1), also known as alpha-2-macroglobulin receptor or CD91, which is expressed in the plasma membrane of human cells and is then internalized through clathrin-mediated endocytosis. A 37 kDa fragment (amino acids 280–612) from the C terminus is released after proteolytic cleavage between amino acids 279 and 280 and reduction of the disulfide bond at amino acids 265 and 287. The fragment is transported to the endoplasmic reticulum, translocated to the cytoplasm, and then inactivates EF-2 [25,26]. The genetic excision of domain Ia results in a molecule termed PE 40 which retains its translocation function and EF-2 inhibition properties but is unable to kill human cells [25,27]. Furthermore, removal of the Ia domain should in turn decrease the hepatoxicity of PE immunotoxins that is due to residual binding of domain Ia to the hepatocyte. A genetically engineered PE molecule PE38KDEL, has amino acids 253–364 linked to amino acids 381–608 with a change in the carboxyl end of PE (KDEL) to increase cytotoxic activity [28,29]. PE38KDEL has been fused with a targeting moiety such as the antibody Fv portion, a growth factor, or cytokine and found to have a much higher affinity for binding to cancer cell lines than the native PE immunotoxin, and was much more toxic to malignant cells [30,31].

Immunotoxins were first shown to be potent cancer cell killers in the early 1970s. Initially, unmodified toxins were injected into the topical forms of refractory metastatic cancers in early clinical studies [10,32]. Afterwards, clinical trials investigating the efficacy and toxicity of immunotoxins in treating a wide variety of hematologic malignancies have made significant progress both with peripheral blood involvement such as in leukemias and malignancy outside the vasculature such as Hodgkin’s lymphoma and multiple myeloma [3,33]. Solid tumors including brain tumors are relatively resistant to immunotoxin treatment because of decreased access to the immunotoxin and a relatively intact immune system. The first report demonstrating the efficacy of immunotoxins against primary CNS tumor cell lines was not published until 1987 [17]. The first generation of immunotoxins conjugated the toxins directly to the Fc portion of a mAb. The clinical results were unremarkable. The second generation of immunotoxins conjugated the toxins to Fab’s or synthesized fusion proteins. The clinical results were much more promising, but still did not represent a definitive treatment modality [32]. Clinical and some promising preclinical studies in brain tumors are summarized below and some major features are discussed (Table 1).

Several obstacles have influenced the therapeutic progress of immunotoxins in cancer treatment. The first concern is that immunotoxins target cell surface antigens that are highly expressed on the tumor cells but are also expressed on normal tissues, usually at a much lower level than on the tumor [2,7]. So bystander cell death can occur with an increase in either the dose or the rate of immunotoxin administration. Another example of nonspecific toxicity is vascular leak syndrome where there is fluid leakage from capillaries, a fall in the serum albumin level, fluid retention, edema and weight gain. This toxicity, owing to endothelial cell damage caused by the high concentration of immunotoxins, can usually be managed with adequate hydration, although severe vascular collapse has been observed at high doses of ricin-based immunotoxins [98,99]. The second concern is the limited access of tumor cells to targeted toxins. The diffusion rate of immunotoxins into brain tumors is affected by the blood-brain barrier [100], the high intra-tumor interstitial pressure in solid tumor [5,101], and local antigen binding [102]. The third disadvantage is that toxins are foreign proteins, and patients with solid tumors and normal immune systems may develop neutralizing antibodies which prevent retreatment [3,103]. There are multiple approaches to reduce the immunogenicity of immunotoxins, including genetic modification of key epitope recognized by T and/or B lymphocytes to generate anti-toxin antibodies without losing toxin activity. For example, seven major epitopes in the PE are recognized by B cells [104]. Mutation of these specific hydrophilic amino acids creates the new mutant PE38 proteins, which have significantly less immunogenicity but still retain full cytotoxic and anti-tumor activities [105,106]. The last and possibly the most important obstacle is the antigen or receptor heterogeneity that is due to the genetic heterogeneity of cancer, although cytotoxic effects of immunotoxins on brain tumor cell lines have been established, which means the antigens or receptors of the primary and metastatic tumors are not homogenous and are variable in density or structure [107]. For example, there is a discrepancy in receptor expression for TfR on the human medulloblastoma DAOY cell line in vitro and in vivo [108]. Furthermore, the EGFRvIII mutation is found in a large number of GBM and is the result of deletion of exons 2–7, with truncation of the extracellular portion of the protein and the subsequent inability to bind the ligand EGF, so that the immunotoxin with EGF as carrier ligand will not kill this population of GBM with EGFRvIII mutation [109].

Clinical responses to immunotoxins have mainly been observed in hematological malignancies and are not as common in solid tumors, including GBM [2,3,33]. This discrepancy may be explained by the fact that the tumor cells in the blood and bone marrow are easily accessible to the immunotoxin, while solid tumor cells such as GBM are not. A second reason for a difference in treatment response is that the immune system in hematological malignancies is impaired and damaged by previous chemotherapy administration so that anti-immunotoxin antibodies are not readily produced and therefore more than one cycle of treatment may be given. Furthermore, systemic toxicity can be caused by targeting the toxin to normal tissues that contain the same target antigen as the cancer cell. This is not a concern if immunotoxins are targeted to antigens on B- or T-cell malignancies, since normal B and T cells can be regenerated from antigen-negative stem cells. But it is a serious concern if solid tumors are targeted that have antigens present on vital organs such as the kidneys, liver or nerve cells, where the immunotoxin will kill these normal cells [110,111]. It is important to choose the target antigen or receptor, which is specifically overexpressed on the cancer cell, but not on normal cells.

The current generation of targeted toxins dates to the last several years where investigators have been engineering these drugs to bind to receptors but also to overcome two major hurdles: toxicity and immunogenicity. Further progress will depend on the identification of new antigenic targets on tumors and the production of less immunogenic immunotoxins so that patients can receive several treatment cycles. These agents may be particularly appropriate for patients with recurrent, resistant and widespread intracranial malignancy resistant to conventional therapy, including surgery, chemotherapy and radiotherapy.