Identification of a CEACAM5 targeted nanobody for positron emission tomography imaging and near-infrared fluorescence imaging of colorectal cancer.

Abstract

Here, we aim to identify a CEACAM5-targeted nanobody and demonstrate its application in positron emission tomography (PET) imaging and near-infrared (NIR) fluorescence imaging in colorectal cancer (CRC). Immunohistochemistry was applied to verify CEACAM5 expression in CRC and metastatic lymph nodes (mLNs). CEACAM5-targeted nanobodies were obtained by immunization of human CEACAM5 protein in a dromedary, followed by several rounds of phage screenings. Immunofluorescence staining and flow cytometry was carried out to determine the binding affinity of the nanobodies. The nanobodies were radiolabeled by coupling 18F-SFB for PET imaging of CRC subcutaneous xenografts and lymph node metastasis (LNM). IRDye800CW (IR800) were conjugated to form NIR probes for NIR imaging in CRC subcutaneous models. CEACAM5 was overexpressed in either human CRC tissues or mLNs. A CEACAM5 targeted nanobody, Nb41 was successfully generated, with excellent in vitro binding properties. Incorporation of albumin binding domain (ABD) did not affect the affinity of Nb41. In vivo imaging showed that both 18F-FB-Nb41 and 18F-FB-Nb41-ABD showed obvious accumulation in the tumor. Due to the longer retention in the blood, 18F-FB-Nb41-ABD enrichment in tumors was significantly delayed but higher compared to 18F-FB-Nb41. Both 18F-FB-Nb41 and 18F-FB-Nb41-ABD showed prominent LNM enrichment. Similarly, the IR800-conjugated nanobodies Nb41-IR800 and Nb41-ABD-IR800 exhibited superior imaging effects in subcutaneous models, while Nb41-ABD-IR800 exhibited higher fluorescence intensity in the tumor accompanied with a remarkedly delay compared to Nb41-IR800. Collectively, we presented the identification and in vivo validation of a CEACAM5-targeted nanobody and a fused nanobody with an ABD, which enabled to the non-invasive visualization of malignancy of CRC using PET imaging and NIR imaging in subcutaneous models as well as LNM models.