Paper Review — Verify Claims Fast

Executive summary

This preprint describes HALOS, an amphiphilic DNA hairpin with two cholesterol anchors that can stably embed in lipid bilayers in an axial orientation, exposing a toehold for recognition of encapsulated nucleic acid targets and transducing that recognition into extracellular reporter recruitment and membrane‑localized amplification (HCR or proximal split initiator ψ‑HCR). Claims are supported by confocal imaging on GUVs, quantitative flow cytometry on cells (n≈10,000), gel electrophoresis, and extensive 1 µs all‑atom MD simulations (NAMD3, CHARMM36). The authors report high GUV and live‑cell detection efficiencies, membrane‑localized amplification up to ~8.2× in live cells, dual‑reporter validation of transmembrane branch migration, and an MD-grounded mechanism in which cholesterol angle and axial orientation stabilize stems and enable hairpin opening upon target hybridization.

Primary experimental claims with direct textual evidence

• Transmembrane detection and outward signaling — authors: "HALOS achieves the first synthetic outward transmembrane signaling mechanism, detecting intracellular targets and amplifying signals directly on membrane without requiring cell lysis or genetic modification" (HALOS transmembrane signaling summary)
• Membrane anchoring and MD support — MD shows axial orientation confers superior stem stability (hydrogen bond means: axial ~77.2 vs lateral ~69.2 vs aqueous ~61.8), and cholesterol angle controls orientation and integrity (MD axial orientation hydrogen bonds)
• Performance metrics — GUV detection: ~85±5% show Cy3 rings within 90–120 min; flow cytometry: Cy5 mean intensity HALOS31 1289±881 a.u. vs control 374±280 a.u.; membrane HCR amplification ~4.2× (GUVs) and ~8.2× (live cells) (Performance metrics GUVs and cells)

Strengths

1. Multimodal evidence: imaging (confocal), quantitative flow cytometry (n≈10,000), gel electrophoresis, NUPACK design validation, and long MD trajectories (~1 µs) provide convergent lines of evidence for HALOS insertion, target recognition, and amplification (HCR amplification evidence)
2. Careful controls: mismatched/scrambled targets and reporters, external quencher steps for GUVs, bulky triangle reporters and streptavidin to reduce nanopore leakage, and stem‑length optimization to minimize false positives.
3. Mechanistic insight: MD and cholesterol angle analysis yield testable design rules (axial orientation, stem length ≥25–31 bp for stability) that help reproducibility and rational modification.

Critical limitations, blindspots, and risks

• Physiological relevance beyond model systems: Key demonstrations are in GUVs and two immortalized cell lines (HEK293T, A549). Lipid composition in cells is more complex (asymmetric leaflets, membrane proteins, glycocalyx, cortical cytoskeleton). Authors partially probe lipid composition effects (LPC increases fluidity and transduction) but do not test primary cells or in vivo models; extrapolation to tissues is currently unsupported (Lipid composition dependence)
• Mechanistic ambiguity about nanopores and branch migration: The model invokes toroidal nanopore formation enabling hybridization across the hydrophobic core. While dual‑reporter labeling and MD support branch migration, direct biophysical demonstration of nanopore formation (conductance measurements, cryoEM, or dye flux assays across single GUVs) is lacking; alternative explanations (membrane defects, transient peptide/lipid perturbations) are not completely excluded.
• Kinetic limitations and sensitivity: Many positive GUV responses occur on 60–120 min timescales — slow for real‑time diagnostics in some contexts. Amplification factors (4–8×) are useful but modest compared with enzymatic methods. Reported large standard deviations (e.g., flow cytometry intensities) reflect heterogeneity and raise concerns on single‑cell quantification precision (Heterogeneity and variability)
• Reproducibility concerns: While methods are described in detail and code/data are linked on GitHub, some specialized steps (in‑house DNA conjugation, GUV electroformation, image analysis pipelines, MD model building from Alphafold3 PDBs) may be nontrivial to reproduce; the MD initial models used Alphafold3 DNA PDBs with pTM ~0.4 — plausible but low confidence for structural detail.
• Potential biases and conflicts of interest: Two corresponding authors have equity interests in a related company (Exodigm Biosciences); the COI is disclosed but could bias emphasis on translational claims.

How convincing is the core claim that DNA hybridization can occur across lipid bilayers under HALOS geometry?

Evidence in favor:

• Confocal imaging: inner Cy3 decreased while membrane Cy3 increased during HALOS incubation, consistent with target relocation and hybridization (GUV imaging evidence)
• Dual‑reporter labeling showing orthogonal labeling of HALOS open site and extended target domain supports physical translocation/branch migration across bilayer.
• MD shows axial orientation reduces base-pair breaks and positions toehold for intracellular access.

Remaining doubts:

• No direct electrical or structural proof of nanopore formation is presented (e.g., single‑channel conductance, cryoEM). Therefore, while convergence of evidence is persuasive, the mechanistic claim that nanopores mediate hybridization remains plausible but not fully proven.

Suggested experiments to strengthen or falsify key claims

1. Measure single‑GUV membrane conductance before/after HALOS insertion and upon target binding to detect toroidal pore formation (patch clamp on giant vesicles); absence of conductance change would challenge the nanopore hypothesis.
2. Use cryoET or high‑resolution AFM of HALOS‑decorated membranes to directly visualize membrane deformation or pore structures.
3. Test HALOS performance in primary cells and in presence of membrane proteins and glycocalyx components to assess physiological robustness.
4. Perform time‑resolved Förster resonance energy transfer (FRET) across bilayers with fast time resolution to quantify hybridization kinetics and discriminate between direct across‑membrane hybridization and target diffusion/leakage mechanisms.
5. Mutational falsification: remove cholesterols or shorten stem length below stability thresholds and demonstrate abolished transmembrane signaling and HCR; the authors present stem length dependence, but an expanded systematic mutational matrix would be stronger evidence.

Practical considerations and opportunities

• Design rules (two cholesterols at ~180°, axial orientation, stem length ≥25–31 bp) give a clear engineering path for future sensors and multiplexed versions.
• HALOS could be adapted to sense other intracellular analytes (small RNAs, viral genomes) pending delivery methods; however, delivery of synthetic targets in vivo and immune/nuclease stability will be obstacles.
• ψ‑HCR split‑initiator approach is an elegant method to reduce background; engineering orthogonal split initiators could enable limited multiplexing on a single membrane surface.

Conclusions and confidence

The paper presents a carefully crafted body of evidence that amphiphilic DNA hairpins can be stabilized in membranes and can mediate sequence‑specific detection of encapsulated nucleic acids with extracellular readout and modest membrane‑localized amplification. The convergence of MD, imaging, gel electrophoresis, and flow cytometry provides strong support for the principal claims, but the mechanistic detail of membrane pore formation and physiological generalizability remain open. Confidence in the core experimental results (GUVs and cell lines) is high; confidence in mechanistic nanopore claims and in vivo translation is moderate pending additional targeted experiments.

Useful reproducible outputs provided

• Data and scripts available on GitHub: https://github.com/rhariadi/artificial-gpcr — valuable for reproduction of image analysis and MD work (Code and data availability)

Immediate recommendations for readers and reviewers

1. Ask for single‑GUV electrical/structural data to support nanopore claims.
2. Request expanded reproducibility data in independent labs and more primary cell types.
3. Encourage authors to publish sequences, raw flow cytometry FCS files, and MD initial PDBs so others can replicate analyses.