Most L1CAM Is not Associated with Extracellular Vesicles in Human Biofluids and iPSC-Derived Neurons

Abstract

Transmembrane L1 cell adhesion molecule (L1CAM) is widely used as a marker to enrich for neuron-derived extracellular vesicles (EVs), especially in plasma. However, this approach lacks sufficient robust validation. This study aimed to assess whether human biofluids are indeed enriched for EVs, particularly neuron-derived EVs, by L1CAM immunoaffinity, utilizing multiple sources (plasma, CSF, conditioned media from iPSC-derived neurons [iNCM]) and different methods (mass spectrometry [MS], nanoparticle tracking analysis [NTA]). Following a systematic multi-step validation approach, we confirmed isolation of generic EV populations using size-exclusion chromatography (SEC) and polymer-aided precipitation (PPT)-two most commonly applied EV isolation methods-from all sources. Neurofilament light (NfL) was detected in both CSF and blood-derived EVs, indicating their neuronal origin. However, L1CAM immunoprecipitation did not yield enrichment of L1CAM in EV fractions. Instead, it was predominantly found in its free-floating form. Additionally, MS-based proteomic analysis of CSF-derived EVs also did not show L1CAM enrichment. Our study validates EV isolation from diverse biofluid sources by several isolation approaches and confirms that some EV subpopulations in human biofluids are of neuronal origin. Thorough testing across multiple sources by different orthogonal methods, however, does not support L1CAM as a marker to reliably enrich for a specific subpopulation of EVs, particularly of neuronal origin.

Keywords: Biomarkers; Blood; Cerebrospinal fluid; Extracellular vesicles; Immunoprecipitation; Isolation methods; L1CAM; Neuron.

Fig. 1

Isolation of EVs using size-exclusion chromatography (SEC) from conditioned medium of human induced pluripotent stem cells (iPSC-) neurons (iNCM) and biofluids consistently yields vesicles showing typical EV characteristics. EVs were isolated using SEC from (a) iNCM (b) CSF and (c) plasma and subjected to phenotypic characterization to confirm the presence of EVs. (left) Size distribution profiles of EV particles isolated from iNCM (n = 3), plasma (n = 5), CSF (n = 3) using SEC measured by nanoparticle tracking analysis (NTA). Data is expressed as mean ± SEM. (middle) Representative immunoblots show the presence of classical EV markers (CD81, Flotillin-1) and absence of EV exclusion markers (Calnexin, GM130) in the EV fraction. Equal volumes of the non-EV protein fractions (to their respective EV fractions) were loaded to show absence of EV markers in them. For biofluids, positive controls include whole mouse brain lysate (m. brain) and iPSC-neuronal cell lysate (iN) and the unprocessed biofluid processed for EV isolation (input). For iNCM, whole mouse brain lysate (m. brain) and parent iPSC-neuronal lysate (input) were loaded as positive controls. Due to high sample viscosity, the conditioned medium and non-EV protein fraction could not be analyzed by immunoblotting. For each source, immunoblotting experiments were performed at least three times with similar results. (right) Neurofilament light chain (NfL) protein can be detected in iNCM, CSF and plasma, by SIMOA measurements. In iNCM-derived EVs, NfL concentration (pg/mL) was normalized to EV protein content (mg/mL), with each data point (black) representing a different iN cell line (n = 4). EVs derived from conditioned medium of iPSC-derived hematopoietic stem cells (iHCM, grey) were used as a negative control for the setup (n = 1) and show a much lower NfL signal than iNCM EVs. Due to high sample viscosity and interference of phenol-red present, the non-EV protein fraction of iNCM could not be measured. NfL levels in CSF and plasma also increase with increasing starting volume of the biofluid. The line plots show concentrations (pg/mL) of EV-associated (black) and free-floating (grey) NfL vs their starting volume, from one experiment. In all plots, the purple dashed line represents the lower limit of quantitation (LLoQ) value

Fig. 2

L1CAM is mostly found outside of EVs in its free-floating form, using size-exclusion chromatography (SEC)-based isolation and L1CAM-immunoaffinity purification. (a) Schematic of the methodological design implemented depicting that EV and non-EV protein fractions were isolated from thrombin-treated plasma by SEC and immunoprecipitated with biotin-conjugated L1CAM or Calnexin (used as a negative control). Image created using BioRender™. (b) Size distribution profiles of these plasma-derived EV fractions after IP by L1CAM (blue) or Calnexin (black) measured by NTA show no enrichment of a specific fraction of EVs by L1CAM. Data represents mean ± SEM from eight independent experiments, with x-axis indicating particle size in nanometer, while the y-axis representing particle numbers in 1 mL of plasma. (c) Representative immunoblot suggests that L1CAM is predominantly free-floating: equivalent volumes of all IP-ed fractions and pre-IP EV fraction were loaded for direct comparison. The presence of EV markers (Flotillin-1, TSG101, CD81) in the pre-IP EV fraction confirms isolation of EVs and the absence of EV exclusion marker (Calnexin) serves as a control for EV purity. The blot reveals a notably higher signal intensity for L1CAM in the free protein fraction following L1CAM IP, in contrast to its counterpart in the EV fraction. Signal for L1CAM was also intentionally overexposed (marked with *) to improve visualization. Positive controls (+ve Ctrls)—whole mouse brain lysate (m. brain) and iPSC-neuronal cell lysate (iN) and the unprocessed plasma used for EV isolation (input)—were included to validate the experimental setup. The blot was repeated at least four times with consistent findings

Fig. 3

Isolation of EVs using polymer-aided precipitation (PPT) from plasma consistently yields EVs with typical characteristics, matching findings from SEC-based isolation. (a) Size distribution profiles of EV particles isolated from plasma measured by NTA. Data is expressed as mean ± SEM; n = 6. (b) Representative immunoblot shows the presence of classical EV markers (CD81, Flotillin-1) and absence of EV exclusion markers (Calnexin, GM130) in the EV fraction. An equivalent volume of the non-EV protein fraction was also probed to check for absence of EV marker proteins. Whole mouse brain lysate (m. brain) and iPSC-neuronal cell lysate (iN) and unprocessed plasma used for EV isolation (input), were loaded as positive controls. The blot was repeated at least three times with similar results. (c) Neurofilament light chain (NfL) protein can be detected in plasma and increases with increasing starting volume. The line plots show concentrations (pg/mL) of EV-associated (black) and free-floating (grey) NfL as measured by SIMOA, with the purple dashed line denoting the lower limit of quantitation (LLoQ) value. This experiment was performed once

Fig. 4

Most L1CAM is found free-floating compared to EV-associated in plasma, using the PPT isolation and immunocapture protocol mostly used in the field. (a) Schematic overview of the workflow showing that EV and non-EV protein fractions were first treated with thrombin and isolated from plasma by polymer-aided precipitation (PPT) and subsequently immunoprecipitated with biotin-conjugated L1CAM or Calnexin (used as a negative control). Image created using BioRender™. (b) Size distribution profiles of these plasma-derived EV fractions after IP by L1CAM (blue) or Calnexin (black) measured by NTA do not indicate enrichment of a specific fraction of EVs by L1CAM. Data represents mean ± SEM from seven independent experiments, with x-axis indicating particle size in nanometer, and the y-axis showing particle concentration (number of particles/mL) of input plasma. (c) Representative immunoblot suggests that most L1CAM is free-floating: equivalent volumes of IP-ed fractions were loaded to allow direct comparison between them. Due to higher protein concentration of the pre-IP EV fraction, an approximately equivalent protein amount to IP-ed fractions was loaded to allow at least a qualitative degree of comparison. In the pre-IP EV fraction, the presence of EV markers (Flotillin-1, TSG101) confirms isolation of EVs and the absence of EV exclusion marker (Calnexin) controls for EV purity. L1CAM levels in the L1CAM-IPed non-EV protein fraction are higher than its EV fraction counterpart. Signal for L1CAM has also been shown to be overexposed (marked with *) to increase the signal-to-noise ratio and allow better visualization. Positive controls (+ve Ctrls) include whole mouse brain lysate (m. brain) and iPSC-neuronal cell lysate (iN) and unprocessed plasma used for EV isolation (input). Immunoblotting experiments were repeated at least four times and yielded similar results

Fig. 5

Higher levels of free L1CAM compared to EV-associated L1CAM are detected after L1CAM immunoaffinity purification, in conditioned medium from iPSC-neurons. iPSC-neuronal conditioned medium was subjected to SEC-based isolation to separate EVs from other irrelevant and abundantly present proteins in the medium (e.g., albumin), before specifically enriching for L1CAM (L1CAM-IP) or Calnexin (used as a control). IP-ed EV and non-EV protein fractions as well as pre-IP EV (positive control) fractions were loaded by equivalent volume to allow their direct comparison. Effective isolation of EVs is confirmed by the presence of EV markers (Flotillin-1, TSG101, CD81) and the absence of EV exclusion marker (Calnexin) in the pre-IP EV fraction. The blot shows higher signal for L1CAM in the L1CAM-IPed non-EV protein fraction compared to its EV fraction counterpart. Positive controls (+ve Ctrls)—whole mouse brain lysate (m. brain) and iPSC-neuronal cell lysate (iN) were included to validate the experimental setup, with a high protein amount (100 µg) to ensure robust signal detection. Blots have been imaged with an emphasis to enhance detection sensitivity of fainter bands in particularly relevant conditions. Thus, where required, the blot is also overexposed to enable better visualization (marked as *). This experiment was repeated at least two times with similar results

Fig. 6

CSF-derived EVs isolated by differential ultracentrifugation (UC) show enrichment for canonical EV markers but not L1CAM. Volcano plot generated from unbiased mass-spectrometry (MS)-based proteomic results from CSF-derived EVs isolated by UC, the current gold standard method for EV isolation, compared to unprocessed CSF, with x- and y-axes showing log₂ transformed fold-change of EVs over CSF and –log10 transformed p-value, respectively, and the dashed solid lines depicting fold-change of 0.5 and 2 on the x-axis and p = 0.05 on the y-axis. The plot reveals specific enrichment of canonical EV markers (blue) with encircled ones used in the present study, while no enrichment of L1CAM (orange) is observed in the EV fraction. Data represents values from a single experiment