Combinatorial Engineering of an APPI Variant with Improved Proteolytic Resistance and KLK6 Selectivity for Cancer Therapy

Amiram Sananes amiramsa@post.bgu.ac.il ^1,2 Itay Cohen ^1,2 Anat Shahar ² Evette S. Radisky ³ Niv Papo ^1,2

¹Department of Biotechnology Engineering, Ben-Gurion University of the Negev, Beer-Sheva, Israel
²The National Institute for Biotechnology in the Negev, Ben-Gurion University of the Negev, Beer-Sheva, Israel
³Department of Cancer Biology, Mayo Clinic Comprehensive Cancer Center, Jacksonville, Florida, USA

Kallikreins, a family of 15 serine proteases, associated with a vast range of normal and pathological processes, are considered attractive targets for the development of novel therapeutics. Kallikrein 6 (KLK6) was found to be involved in several cancer malignancies: melanoma, colon, ovarian and breast cancer, where it induces metastasis and is involved in signaling pathways, leading to cancer progression. No successful attempt has been yet reported for a potent sub-nanomolar affinity inhibitor towards KLK6. Here, we used the human amyloid β-protein precursor inhibitor domain (APPI) as a scaffold to engineer a KLK6 inhibitor, with high affinity and increased specificity towards KLK6. Two libraries were generated having random mutations introduced into the APPI scaffold, both at the protease binding loop and in the entire enzyme-inhibitor interaction interface. These libraries were displayed on the surface of yeast cells and screened using high-throughput flow cytometry to isolate APPI variants with high affinity towards KLK6. We isolated a variant having 4 mutations APPI_{M17L,I18F,S19F,F34V}(APPI_4M), with a sub-nanomolar KLK6 inhibition constant (K_i) of 160 pM, making it the best in class known KLK6 inhibitor. Moreover, when tested for inhibition capability towards other closely-related serine proteases, APPI_4M showed up to 560-fold increased selectivity towards KLK6 vs other serine proteases, when compared to APPI_WT. This variant also displayed high resistance towards KLK6 hydrolysis, with a long turnover-time of 10 days. We then demonstrated that these optimized properties enabled APPI_4M to block KLK6-dependent cancer cell invasiveness. Finally, we solved both crystal structures of APPI_WT-KLK6 and APPI_4M-KLK6, allowing us to elucidate the mechanism by which APPI_4M optimally interferes with the KLK6 activity, and most importantly to understand the basis for its improved specificity and proteolytic resistance towards KLK6.