(386c) Engineered Knottin Peptides: A New Class of Agents for Non-Invasive Molecular Imaging of Tumor Biomarkers

There is a critical need for non-invasive molecular imaging probes that specifically target receptors overexpressed on tumors, for earlier cancer detection and patient-specific cancer treatment and disease management. We are developing cystine-knot (knottin) peptides as a new class of in vivo molecular imaging agents against biomarkers expressed on tumors. Knottin peptides consist of at least three interwoven disulfide bonds, which render them extremely stable in vitro and in vivo, and multiple solvent-exposed loops. These qualities are desirable for drug design and probe discovery; however, knottins as a family naturally possess diverse functions as protease inhibitors, toxins, and antimicrobials and therefore must be engineered to specifically recognize tumor-related receptors.

Previously, we used the Ecballium elaterium trypsin inhibitor II (EETI-II), from the squash family of protease inhibitors, and a fragment of the human Agouti-related protein (AgRP), as molecular scaffolds to engineer knottin peptides that bound to integrin receptors with low nanomolar affinity. The small size of these knottin peptides (~4 kDa) allowed them to clear from the body on a rapid timescale desirable for molecular imaging with short-lived isotopes, an application in which large antibodies (~150 kDa) fall short. Radiolabeled versions of these knottin peptides showed high tumor uptake and low background in non-tumor tissue (i.e. liver) as measured by positron emission tomography (PET) in mouse human tumor xenograft models. Moreover, these knottin peptides were extremely stable, and remained intact upon exposure to serum for days at 37°C.

In more recent work, we extended this technology to additional cancer targets and knottin scaffolds. Using rational and combinatorial methods, we engineered stable knottin peptides with antibody-like affinity to the tumor marker Carbonic Anhydrase IX (CAIX). CAIX is a surface protein overexpressed in poorly oxygenated (hypoxic) tumors including breast, colon, kidney, and lung. CAIX expression correlates with poor prognosis and an aggressive tumor phenotype, including increased invasiveness, metastatic potential, and resistance to chemo- and radiotherapy. Molecular imaging of CAIX would allow for tumor staging to inform clinical treatment decisions. The poor contrast provided by available small molecule-based hypoxia and CAIX imaging agents led us to pursue the development of knottin-based probes. A combination of phage display and yeast surface display was used for the directed evolution of CAIX-binding knottin peptides, which had no initial binding to this target. In this study, we used truncated versions of the knottin peptides Agatoxin, from spider venom, and the human AgRP peptide as molecular scaffolds to engineer peptides that bound to CAIX with low nanomolar affinity. Libraries of over 10 million mutants were screened using high-throughput flow cytometry to identify knottin peptides that bound to fluorescently labeled CAIX. We are currently synthesizing and testing labeled versions of these engineered knottins to assess their promise as optical and PET molecular imaging probes compared to small molecules.

Funding: NIH R21 CA143498, Center for Biomedical Imaging at Stanford, and NSF and Stanford Graduate Fellowship Programs