Design Of Antibody-based Fluorogenic Probes And Further Development Of Related Fluorescent Molecules

Early detection and treatment are still the most effective way to fight against cancer.Over the past decade,organic fluorescent small molecules have been widely studied in bioimaging due to their unique optical properties,low cost and chemical tractability,providing promise for early tumor diagnosis.Hitherto,several thousand small-molecule dyes have been developed,but only a small fraction can be used for tumor-related imaging.On one hand,most small-molecule fluorescent probes are designed for the detection of tumor-related enzymes,while few of them are suitable for other biomarkers of cancer including non-enzyme proteins or nucleic acids,largely due to the selectivity issue or the lack of specific ligands towards these targets.As a result,immunohistochemistry remains the most efficient method available for clinical pathological diagnosis.On the other hand,most organic fluorescent small molecules’photophysical properties may not meet the requirements for tissue or in vivo imaging,especially in view of the penetrating depth.For example,the majority of existing organic fluorescent small molecules have a maximum emission wavelength within the visible wavelength window.Moreover,there are still many limitations in the practical application of the existing NIR fluorescent molecules including the low quantum yield and poor photo stability.To address the aforementioned issues and promote the application of fluorescent probes in tumor detection,this thesis focuses on two parts,including the design of new fluorescent probes and the development of organic fluorescent dyes:（1）development of fluorogenic near-infrared antibody-based probes,which can selectively label target proteins and achieve fluorescence imaging of tumor cells;（2）artificial intelligence（AI）technology-facilitated prediction of the photophysical properties of organic fluorescent dyes,further promoting the development of small-molecule fluorophores,especially those with long emission wavelength.Antibodies,as classic molecular recognition moieties,have been playing important roles in biologic detections due to their specificity and affinity towards the antigen.To develop fluorogenic antibody-based probes which can only be turned on while binding to the target,thus allowing wash-free and real-time imaging of the antigen,we designed antibody-based probes using H-dimer fluorescent molecules which will get quenched in the dimer form.Specifically,in the absence of the antigen,two fluorophores on the N-terminal site of the antibody will be quenched upon the formation of dimer.When the probe binds to the antigen,this dimerization state will be disrupted,and the fluorescence signal is restored.In order to improve the imaging depth,we introduced a near-infrared organic fluorescent small molecule,Cy5.5,which was site-specifically labeled at the N-terminus of both the heavy and light chains of monoclonal antibodies.This position is close to the antibody-antigen binding site,which allows the fluorescence signal of Cy5.5 to be affected by the interaction between the antibody and antigen without disrupting their binding,thus indicating the expression and distribution of the target antigen.Cetuximab,a therapeutic antibody against the cell-surface protein,epidermal growth factor receptor（EGFR）,was chosen as the model antibody.Based on the interaction between cetuximab and EGFR,fluorogenic near-infrared antibody probes,Cet-Cy Me and Cet-Cy PEG,were prepared by labeling the antibody with Cy5.5 derivatives（Cy5.5-1N₃,Cy5.5-2N₃）.Live-cell imaging was subsequently carried out to validate the feasibility of this design.With the success of above probes,simultaneous detection of multiple targets in the tumor tissue and the improvement of probe performance are waiting to be achieved.The above antibody-based fluorogenic probe strategy can be used to construct antibody probes suitable for multiple targets via the combination of diverse antibodies and fluorophores at different wavelengths.However,fluorescent small molecules that can be applied to this design is currently limited,and most of these have relatively short emission wavelengths,which greatly restricts their biological applications.Therefore,developing fluorescent molecules that meet the requirements is a key in future research.Here,artificial intelligence with excellent insights was applied to predict the optical properties of organic small molecules to accelerate the development of fluorescent dyes.Specially,by constructing a database of fluorescent molecules covering the UV-Vis-NIR area,in combination with deep neural network,SMFluo1-DP platform with higher out-of-sample prediction accuracy than existing models for efficiently predicting the maximum absorption wavelength of fluorescent molecules was obtained.This platform can guide the rational design of NIR fluorescent molecules and provide technical support for the construction of antibody-based probe libraries targeting different tumor biomarkers with different wavelengths.