Isostere 18F-protein post-translational editing enables dynamic tracking of neurodegeneration biomarkers

The neurofilament light chain protein (NfL) is a suggested general marker for neuronal loss. Its release from brain parenchyma into cerebral spinal fluid, and presumed detection in blood has seen it established as a first blood-based marker of disease activity and drug efficacy in multiple sclerosis (MS) and in the presymptomatic diagnosis and assessment of disease course for other neurodegenerative disorders.1 However, the lack of characterisation of its behaviour in circulation, largely due to its antibody-dependent measurement, have hampered the biological interpretation of these measurements, especially after acute injury such as in MS relapse or head trauma.2 Here, we describe a strategy for exploiting positron emission tomography (PET) imaging using isosteric protein mimics following the installation of a fluorine-18 label that is benign enough to provide sensitive, real-time information on the dynamics and trafficking of NfL protein. This circumvents the limits of current methods that integrate 18F into proteins through the bio-conjugation of bulky, unnatural groups, which we show perturb NfLs assembly and functional properties from those in the natural state. We use a visible-light-driven reaction to access radioactive isostere proteins that are unperturbed and so closely resemble their native form. In this way, generation of [18F]fluoroalkyl radicals that can be rapidly reacted at pre-defined sites on proteins creates mimics of proteinogenic side chains bearing near-zero-size labels to probe proteins in functionally true form. These prosthetic-free, protein radiotracers can be generated in excellent radiochemical yield (up to 67%) via a semi-automated protocol in just 15 mins. High associated molar activities (precursor up to 102 GBq mol-1) allowed high sensitivity dynamic observations in blood, brain and cerebrospinal fluid, enabling even the first unambiguous observations of spinal flow kinetics using proteins. These dynamics, including the high rate of spinal flow (on the order of mm per min) and drainage of NfL from CSF into sacral lymph nodes, now provides evidence that the slow fall rate of antibody-detected markers that is observed after acute neural insults is not due to a long half-life, but rather reflects sustained neuronal loss. This discovery will now help to better correlate clinical and radiological features of disease with NfL blood levels. Our methodology now demonstrates the broad potential of a near-zero-size labelling method for the functional study of proteins in whole organisms without interfering with their biological activity and native assembly.