‘Chemically evolutionary screening’ of cyclic peptides for PET imaging of PD-L1 protein in tumors

Purpose/Background:Immunotherapy, such as immune checkpoint blockade therapy (ICB), has revolutionized the cancer management paradigms in recent years. Although great success has been achieved in ICB, only some patients respond to ICB. This situation calls for methods that can stratify the patients who will benefit from ICB to increase the treatment efficacy. Therefore, there is an urgent need to develop noninvasive imaging tools to quantify PD-L1 expression in tumors. Thepurpose of this study was to develop cyclic peptide-based PET radiotracers for PD-L1 and demonstrate their specificities for PD-L1 quantification in vivo in mouse models of multiple cancer types.Methods:Starting from a 22-mer peptide identified by phage display, we first applied alanine scanning and sequence truncation strategies to identify the critical residues for PD-L1 binding and the shortest sequence necessary for the high binding affinity. After that, several peptide macrocyclization approaches, such as side chain-to-side chain, tail-to-side chain, and head-to-tail cyclization, were adopted to constrain the peptides to increase the binding affinity and stability. The affinity of the peptides for PD-L1 was evaluated using SPR or an ELISA kit. The three peptides with the toppest binding affinity were conjugated with a chelator (DOTA) and then subjected to 64Cu labeling. The radiochemical yield,specific activity, and radiochemical purity of the tracers were analyzed by radio HPLC. The stability of the tracers was assayed in mouse serum. In vitro competitive inhibition assays were used to establish peptide analog affinity forPD-L1. The pharmacokinetics and imaging capacity of the tracers were assessed in multiple tumor models in immune-competent mice by PET imaging and ex vivo biodistribution and ARG. Blocking PET with an anti-PD-L1antibody as a blocking agent was utilized to verify the specificity of the tracers for PD-L1. PD-L1 expression in cell lines and tumors was validated by flow cytometry and IHC, respectively.Results:Through two rounds of screening, we successfully identified four nonadjacent residues located in the middle of the sequence as so-called “hot spots” for PD-L1 binding. Three cyclized peptides, designated HKCP-2751/2752/2753,were identified with high binding affinities at the nanomolar level (1-10 nM) for PD-L1 and outstanding stability in mouse serum. HKCP-2751 was cyclized by a side chain-to-side chain disulfide bond, and HKCP-2752/2753 was cyclizedby head-to-side chain amide bonds. [ 64Cu]Cu-HKCP2751/2752/2753 were afforded with decay-corrected RCYs of >98%, SAs of >74 GBq/μmoL (n=8), and RCPs of above 99%. All tracers showed excellent stability in vitro. The threetracers showed PD-L1-specific uptake in tumor cells. MicroPET/CT imaging in normal C57BL/6J mice showed that all tracers displayed fast blood clearance and specific uptake in immune-related organs, such as the spleen. The organwith the highest uptake is the kidney. We observed a higher liver uptake of [64Cu]Cu-HKCP2752/2752 than that of [64Cu]Cu-HKCP2751, in accordance with the lipophilicity of the tracers. PET imaging with [64Cu]Cu-HKCP2751 inB16F10, MC38, and U87MG tumor-bearing mice revealed sustained and durable tracer retention in the tumor site (4-6%ID/g from 30 min to 120 min postinjection). Blocking PET demonstrated PD-L1-specific uptake in vivo. Ex vivo biodistribution confirmed the PET observations and suggested that 90-120 min p.i. is the time window for the bestSNMMI202