Identification of CTPs

Read Also

Several monoclonal antibodies that target cell surface receptors, such as the anti-CD20 antibody Rituximab, have been approved for cancer treatment (for a review [120]). However, the large size of the antibody (160kDa), the high cost of its production and characterization as well as its relatively nonspecific binding to the reticulo-endothelial system represent major drawbacks when cytotoxic drugs or radionuclides are coupled to the antibody. On the other hand, binding to the reticulo-endothelial system can be considered as an advantage when the antibody is naked as it can induce an immunological response which can destroy tumor cells. It is expected that binding of macrophages to the Fc fragment of these antibodies will also favor the immune response. Therefore, because of their easier preparation and their good affinity or specificity, short peptides or peptido-mimetics are attractive alternative targeting agents for either cancer imaging or therapy. Various peptides that target specifically one given cell line have been identified using different techniques aimed at defining the shortest peptide with the highest specificity and affinity (Table 1).

Structure-activity relationship of ligands

The most intuitive way of designing a specific binding peptide for a given receptor is to start either from the structural data or from the structure-activity relationship study of the molecular interactions between a circulating protein and its cellular receptor. However, structural data are not always available and structure-activity relationship studies seldom advance to a linear well-defined sequence which could be directly used as a receptor-specific binding peptide, mainly because the tertiary structure scaffolding frequently hides the ligand-receptor interactions. Nevertheless, some short peptide sequences have been defined by these conventional methods [121, 122].

For instance, the erbB2 receptor has been targeted with an erbB2 receptor-binding hepta-peptide (see Table 1) attached to the pro-apoptotic alpha-tocopheryl succinate (a-TOS) [123]. The chimera between the a-TOS moiety and the receptor-binding hepta-peptide reduced breast carcinomas expressing high levels of erbB2 in transgenic mice. Another hepta-peptide binding to Neuropilin-1 (NRP), a vascular endothelial growth factor (VEGF) receptor, has been evaluated in vitro [124] and more recently in vivo [125]. The peptide used against circulating VEGF was initially identified by phage display (see next section), and then a linear 20-mer peptide was further studied by X-rays crystallography to improve its binding properties to VEGF. It was shown that this peptide was unstructured in solution but adopted a largely extended conformation upon binding to VEGF [126]. This observation confirms the dynamic process leading to high affinity binding. The initial interaction favors the next one, and so on, until the peptide secondary structure is finally changed. X-rays data also indicate that, upon binding, residues 3 to 8 of this peptide form a beta-strand structure pairing with the beta 6-strand of VEGF via six hydrogen bonds, which leads to a rather strong interaction of the peptide with VEGF. However, the parent peptide could not be shortened to less than 14 residues without dramatically reducing the binding efficiency. To our knowledge, this 14 mer-version of the VEGF binding peptide has not yet been evaluated in vivo.

 Phage display

Several peptides that target specifically cancer cells or tumor blood vessel endothelial cells have been identified using two techniques based on a combinatorial approach (i.e., phage display and One Bead One Compound). The phage-display technique relies on the combinatorial generation of short peptide sequences inserted in the extracellular protein of a filamentous phage [127]. Upon interactions of the phage with a specific extracellular receptor of a given cell type, the phage is amplified following cell infection. Limiting dilutions of the transfected suspension are then made, and one single phage type can be isolated after several rounds of selection. Subsequently, the “active” combinatorial sequence is identified by sequencing. The phage display technique allows the identification of peptide sequences ranging from 8 to 12 amino acids. Once characterized, the peptide itself is evaluated for its ability to bind with high specificity and strong affinity to the specific receptor of the targeted cell.

In terms of synthesis, the sequences highlighted by phage display are interesting and their production is not difficult. Indeed, nowadays, peptide chemistry is easy and quite cheap for peptides up to 15 amino acids, and they can be synthesized with rather good yields. Although some longer peptides have also reached the drug market (the best known among them is probably the 36 amino acid-long T20 or Fuseon, used for antiviral treatment of HIV (for a review [128])), the synthesis of such large peptides is more complicated and requires more expensive production techniques. Peptide length could also be a concern when considering the structural aspects of the peptide-ligand interactions. Except in some very specific cases, peptides of less than 15 amino acids are poorly structured. Therefore, their affinity or their specificity might be somehow reduced when extracted from the displaying protein-phage. However, by inserting cysteine residues at both ends of the identified sequence by genetic engineering, it is possible to apply a structural constraint through the formation of a disulfide bridge, leading to a cyclic peptide within the phage protein [129, 130]. Such cyclization often improves the affinity of the peptide towards the target receptor in comparison to the linear form (see below). The main inconvenience of the phage display technique as an abundant source of new cell-specific ligands is that, because of its biological origin, only natural L-amino acids can be inserted in the peptide sequence. Therefore, ligands with higher affinity can be missed out.  However, once a native sequence has been identified, it is also possible to carry out a structure-activity relationship study on analogues harboring non-natural synthons. This is probably the reason why the second technique was developed about a decade ago [131].

Chemical strategies (One Bead One Compound, ligand mimetics…)

The chemical generation of libraries, which include both D- or non-natural amino acids, has been recently developed, offering the possibility of discovering new ligands with either higher affinity or better specificity for a given cell receptor. This synthetic technique is based on the combinatorial synthesis of one compound on one single solid bead, thus generating random peptides, each with its own unique amino acid sequence on the same bead. Hence, it has been defined as the “one-bead one-compound” (OBOC) strategy [132], and has led to the discovery of a large number of specific ligands (for a recent review [133]). Once the library has been assembled, various cell lines can be grown in the presence of the beads. A cell type growing on one bead clearly indicates the interaction of a membrane receptor with the peptide assembled on that particular bead. The sequence of the peptide attached on the bead is then determined by sequencing. Competition experiments between the peptide attached on the bead and the soluble peptide allow to confirm the sequence specificity and to measure various parameters.

The main advantage of this technique is that non-natural amino acids can be inserted within the sequence. For instance, a peptido-mimetic called LLP2A showed a very high affinity for the a4b1 integrin receptor of activated T- and B-lymphomas with an IC50 of 2pM [134]. The main benefit of such non-natural ligands is their high stability against proteases. LLP2A showed an extreme resistance to proteases with no sign of degradation after 18 days of incubation in human plasma at 37°C. Indeed, this is a major advantage when developing a targeting peptide. The question of the blood clearance of such small molecules has not yet been completely resolved.