Review Papers Claim by Claim

Scope and claim

The authors claim ICAM-5 is a neuron-specific, EV membrane surface marker enabling improved isolation of neuronally derived extracellular vesicles from human biofluids and improved detection of neuronal injury markers (UCH-L1) in mild traumatic brain injury (mTBI) serum samples ICAM-5 NDEV marker preprint

Evidence summary (data-driven)

• Discovery: neuronal EV LC-MS detected >3000 proteins, filtered to 63 brain-elevated proteins; ICAM-5 identified as a single transmembrane, neuron-restricted candidate MS discovery of ICAM-5
• SEC coelution: ICAM-5 signal coeluted with tetraspanins CD63 and CD9 in serum fractions 9-14 (serum) and fraction 12 (CSF), consistent with EV association rather than exclusively soluble form SEC coelution data
• Surface localization: dSTORM superresolution imaging reported surface detection of ICAM-5 and ATP1A3 on EVs and quantified ICAM-5+Pan-EV at ~8.61% in serum and ~20.15% in CSF; ATP1A3+Pan-EV was ~21.74% in serum and ~17.27% in CSF dSTORM quantitative surface detection
• Functional enrichment: ICAM-5 immunoprecipitated NDEVs were enriched in neuronal proteins (Tau, NSE) and lacked GFAP signal (astrocyte marker), whereas ATP1A3 and L1CAM NDEVs contained GFAP — supporting neuronal specificity of ICAM-5 capture Immunoprecipitation neuronal enrichment
• Clinical biomarker yield: In a cohort of mTBI patients (n=10), ICAM-5 NDEVs had higher UCH-L1 concentrations (mean 1520 pg/mL) than L1CAM or ATP1A3 NDEVs (ICAM-5 vs L1CAM p=0.0019; ICAM-5 vs ATP1A3 p=0.0049), and ICAM-5 NDEV UCH-L1 exceeded total EV UCH-L1 in these EV suspension measures UCH-L1 enrichment data

Strengths

• Multi-modal validation: discovery (MS) → biochemical fractionation (SEC) → imaging (EM, dSTORM) → functional IP enrichment → clinical biomarker readout (UCH-L1). Each step supports ICAM-5 EV association and neuronal enrichment Multi-modal study design
• Biological plausibility: ICAM-5 is a neuronal telencephalic immunoglobulin superfamily transmembrane protein localized to dendrites/spines, making it plausible as a neuron-specific EV membrane component (UniProt annotation supports membrane localization and telencephalic expression) ICAM-5 neuronal biology

Key limitations and potential confounders

1. Sample size and generalizability — some critical assays rely on pooled serum (N=3) or small clinical numbers (n=10 for UCH-L1), increasing risk of random effects and limiting population generalizability; results need replication in larger, demographically diverse cohorts Small sample limitations
2. Antibody and IP optimization not shown — authors used equal antibody amounts across IPs but did not optimize concentrations, affinities, or epitope accessibility; differential IP efficiency or steric hindrance could bias enrichment metrics (UCH-L1 differences may partly reflect recovery efficiency rather than true EV composition) Antibody optimization missing
3. Marker heterogeneity and proportion — dSTORM shows only ~8.6% of serum EVs are ICAM-5+, meaning capture yields will be a minority of total EVs; while CSF shows higher proportion (~20%), low absolute numbers in CSF and variable yields complicate clinical use and throughput ICAM-5 EV proportion
4. Possible soluble vs EV-associated forms — authors used SEC to separate soluble proteins from EVs and showed ICAM-5 coelution with EV markers, but soluble cleavage products were observed for other markers (ATP1A3, L1CAM) and smaller bands for ICAM-5 were also sometimes seen; proteolytic cleavage or shedding of membrane proteins could create soluble fragments that complicate IP specificity and downstream biomarker interpretation Soluble versus EV associated forms
5. Normalization and quantitative reporting — UCH-L1 values are reported per EV suspension (pg/mL of EV prep), not normalized to starting serum volume or EV particle counts; without normalization to input (serum volume) and output (EV number or total protein) comparisons across methods or cohorts are hard to interpret quantitatively Normalization caution
6. Potential cross-reactivity and truncated isoforms — smaller MW bands for ATP1A3 and L1CAM, and sometimes ICAM-5 fragments, were observed but not characterized by MS or epitope mapping; cross-reactive antibodies or truncated proteins could mislead interpretation of marker specificity Truncated isoforms issue

Interpretation and confidence

Taken together the data are consistent with ICAM-5 being present on a neuron-enriched EV subpopulation in human biofluids and that ICAM-5 immunocapture enriches neuronal proteins and improves detection of a neuronal injury biomarker (UCH-L1) in a small mTBI cohort. Confidence is moderate: orthogonal replication, larger clinical cohorts, robust antibody/IP optimization, and quantitative normalization are required before ICAM-5 can be accepted as a clinical-grade NDEV capture marker Authors conclusion and need for validation

Concrete recommendations for follow-up experiments

1. Antibody epitope mapping and IP optimization: test multiple monoclonal clones, titrate antibody:EV ratios, vary incubation times/conditions, and compare capture efficiency normalized to input serum volume and particle counts; include isotype and bead-only controls.
2. Orthogonal validation of ICAM-5 on EV surface: perform protease protection assays (proteinase K with/without detergent) and mass spectrometry on IP-captured EVs to confirm full-length ICAM-5 on intact vesicles and to map cleavage sites on smaller bands.
3. Spike recovery and specificity controls: spike cultured neuron-derived ICAM-5+ EVs into plasma from ICAM-5 negative sources (if available) and measure recovery; perform depletion experiments to confirm target EV removal.
4. Larger clinical cohorts and comparator conditions: test in larger mTBI cohorts and in non-neurological injury controls and neurodegenerative disease cohorts to measure sensitivity, specificity, and disease/temporal dynamics; include longitudinal sampling where possible.
5. Normalization standards: report biomarker loads normalized to starting serum volume, EV particle count, or total EV protein to allow comparability across methods and studies.

How this fits the field and potential applications

Improved neuronal EV capture markers are a high-value need for liquid biopsy of CNS disease because they can increase signal-to-noise for neuron-derived biomolecules in blood. ICAM-5 is biologically plausible as a neuron-specific membrane protein in the telencephalon and, if validated and standardized, could be used to enrich NDEVs for research biomarker discovery, pharmacodynamic monitoring, and potentially diagnostic assays for acute neuronal injury or neurodegeneration — but translational use will require the rigorous steps above Field relevance statement

Suggested minimal reporting for future studies

• Full antibody clone identifiers and validation images, IP protocols with optimization steps, and raw NTA/EM/dSTORM data.
• MS of IP-captured EVs (to confirm ICAM-5 peptides on EV preparations and identify coenriched neuronal proteome).
• Normalization strategy for biomarker reporting (pg per mL serum or pg per 1e9 EVs).
• Data sharing of raw MS and imaging files for reanalysis.

Interactive quantitative figure

Conclusions (evidence weighted)

The preprint provides moderate-quality, multi-modal evidence that ICAM-5 is present on a neuron-enriched EV subpopulation and that ICAM-5 immunocapture enriches neuronal proteins and improves detection of UCH-L1 in a small mTBI cohort. The claim is promising but preliminary: independent replication, antibody/IP optimization, proteomic confirmation of captured EVs, standardized normalization, and larger clinical validation are required before ICAM-5 can be considered a robust, clinically adoptable NDEV capture marker Final weighed conclusion

Author review links

Open bespoke Author Reviews on BGPT: