Progress in developing diagnostics and therapeutics for brain-based diseases has been hampered by an inability to study the underlying biochemistry in living patients. Currently, we rely upon post-mortem brain analysis, GWAS, and animal models to learn about the predisposing genes, proteins and pathways involved in these diseases. Still, we lack the ability to assess brain cells in vivo from living psychiatric and neurological patients. We would like to be able to ask questions such as what does the RNA content of a neuron from a depressed patient look like before and after treatment with an antidepressant? Or what are the proteomic changes in a bipolar patient when they are having a manic episode compared to when they feel depressed? Unfortunately, current techniques do not allow for these analyses.
One proposed method to assess the content of brain cells in living patients involves Extracellular Vesicles (EVs). EVs are small, membrane enclosed, vesicles that bud off from all cell types and contain protein and RNA. As such, they can be thought of as a molecular ‘snapshot’ of their cell of origin. By capturing EVs secreted from neurons in peripheral biofluids, researchers could noninvasively evaluate the living brain of psychiatric and neurological patients. Sampling the brain in this manner could not only lead to a better understanding of the brain and thereby potentially improve the treatment of complex brain diseases, but it could also enable the development of blood tests that diagnose the disease at a much earlier stage, i.e., before irreversible damage has already taken place.
Because of this tremendous potential, there has been a major effort in the neuroscience community to validate immunocapture procedures that would allow for analysis of these entities. Over 70 publications claim to have immunocaptured Brain-Derived EVs and to have measured the macromolecular content within. The vast majority of researchers used L1CAM, a transmembrane protein abundant on the surface of neurons, as a handle for EV immunocapture. Initially, we planned to use this technique to study internal, post-translationally modified proteins relevant to Parkinson’s Disease. However, before involving scarce patient samples we first wanted to ensure that the content of the EVs were truly neuron derived. We had two concerns. First, while L1CAM is abundant on the surface of neurons, it is also found on several non-neuronal cell types outside of the central nervous system. Second, L1CAM can also be found in cleaved and alternatively spliced isoforms making it a soluble protein. Thus, we undertook a rigorous investigation of L1CAM in human biofluids such as cerebrospinal fluid (CSF) and plasma. To do so we not only developed high sensitivity single molecule array assays to study EVs and L1CAM but also developed a framework for evaluating cell-type specific EVs.
Utilizing fractionation techniques such as size exclusion chromatography (SEC) and density gradient centrifugation (DGC), we found that L1CAM in the plasma and CSF is soluble, not EV associated. Furthermore, we demonstrate that the isoforms found in plasma are not brain-derived, as they are distinct from the isoforms found in CSF. These findings are significant as they contradict dozens of papers in this field and call into question the results of that research. This is of particular clinical importance as many trials list the contents of brain-derived EVs as markers of target engagement, including those captured using L1CAM.
Our findings highlight the importance of rigorously validating candidate cell-type specific EVs, and they outline a path for achieving this. Any protein to be used for immunocapture of EVs must be measured in fractionated biofluids using both SEC and DGC and shown to elute predominantly with the tetraspanins and not with free proteins such as albumin. Once this requirement is met, then the candidate protein should be used for immunocapture and proteinase and RNAse should be applied, and subsequently inactivated before lysis of the EVs and observation of the internal content. Finally, this content must be shown to be unique to the cell type of interest. None of the above steps were taken in the case of L1CAM leading to waste of precious patient samples, scientist time and grant funding.
Furthermore, the fact that researchers measured so many different proteins ‘internal’ to L1CAM EVs highlights an important but dispiriting fact about non-specific binding. As we demonstrate, the findings of some of these articles are due to nonspecific binding of proteins to the beads or antibodies used in the immunocapture procedure. Therefore, scientists wishing to appropriately measure the internal contents of EVs must proteinase treat their sample and inactivate the proteinase before lysis to avoid these artifacts.
As we demonstrate, L1CAM is highly abundant in plasma as it comes from many cell types in the body. This has led to an overestimation of the quantity of Brain-Derived EVs in blood. In reality, high sensitivity methods are needed to measure even total plasma EV proteins such as CD9, CD63 and CD81. Thus, we need to reevaluate our certainty that Brain-Derived EVs cross the Blood-Brain Barrier in sufficient quantities to be detected in peripheral biofluids.
Our manuscript may seem like a step back in the mission to use EVs to better understand the brain, but the techniques and framework that we establish are a foundation upon which we hope that we can collaboratively rebuild this field. We must find and rigorously validate novel markers for Brain-Derived EV immunocapture in order to realize their immense potential for helping psychiatric and neurological patients.