Characterizing the Hard and Soft Nanoparticle-Protein Corona with Multilayer Adsorption

Alberto Martinez-Serra; Giancarlo Franzese; Marco P Monopoli; Oriol Vilanova

REVIEW 4 major objections 7 minor 3 references

Characterizing the Hard and Soft Nanoparticle-Protein Corona with Multilayer Adsorption

T0 review · 4 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that the soft corona of loosely bound proteins around a nanoparticle does not settle into equilibrium, but evolves with glassy, multi-timescale dynamics, and that a coarse-grained simulation with a surface-induced…

desk verdict A useful multilayer extension of the BUBBLES corona model, but the glassy soft-corona claim rests on a single unvalidated autocorrelation curve from a calibrated potential. read the letter →

arxiv 2411.13279 v2 pith:ZSEEKKQZ submitted 2024-11-20 cond-mat.soft

Oriol Vilanova , Alberto Martinez-Serra , Marco P Monopoli , Giancarlo Franzese This is my paper · ORCID

classification cond-mat.soft

keywords proteincoronasofthardcoarse-grainedsimulationglassydynamicstransferrinnanoparticlemultilayeradsorption

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that the soft corona, the loosely attached outer layers of proteins around a nanoparticle, does not settle into a simple equilibrium but evolves in a glassy way, reorganizing over many timescales at once. Using a coarse-grained simulation of transferrin on polystyrene nanoparticles, the authors reproduce the experimentally observed multilayer adsorption and the numbers of adsorbed proteins. They find two soft-corona layers with distinct dynamics: the inner layer shows a density autocorrelation with a plateau, a signature of glassy relaxation, while the outer layer decays roughly as a power law. If this is right, the protein shell that cells actually encounter keeps changing slowly, so the biological identity of a nanoparticle depends on how long it has been in a biological fluid.

What carries the argument

The load-bearing machinery is a coarse-grained Langevin-dynamics model in which proteins are soft spheres, the nanoparticle-protein interaction is a DLVO-type potential with a deep contact minimum, and a new three-body potential U3b = -epsilon_3b exp(-di dj / $kappa^{2}$) exp(-(rij-delta)^2/(2 $omega^{2}$)) adds an attraction between two proteins only when at least one of them is near the nanoparticle surface. This three-body term stands in for the hypothesis that surface adsorption partially unfolds transferrin and exposes sticky residues; with epsilon_3b = 3.75 kBT fixed and the decay length kappa tuned for each nanoparticle concentration, it produces the second and third corona layers and the glassy relaxation. A buffer region surrounding the reaction volume maintains the protein concentration without insertion or deletion events, letting the simulations reach timescales of seconds at experimental protein-to-nanoparticle ratios.

What would settle it

Measure the population autocorrelation of the inner soft-corona layer directly, for example by fluorescence correlation spectroscopy or single-molecule tracking over the 0.01 to 0.1 second window: if it decays exponentially with a single characteristic time rather than showing a plateau, the claimed glassy dynamics are not present.

Watch

Extended reading notes

Core claim

The central discovery is that the soft corona is dominated by a glassy evolution related to crowding. In simulations of transferrin adsorbing onto carboxylated polystyrene nanoparticles, the corona organizes into three layers: an irreversibly bound hard corona, an inner soft-corona layer SC1 whose density autocorrelation function decays nonexponentially and develops a plateau, and an outer soft-corona layer SC2 whose autocorrelation decays approximately as a power law. The plateau in SC1 is interpreted as dynamical arrest caused by crowding from the outer layer, meaning the inner soft corona is trapped in local free-energy minima rather than relaxing to equilibrium. The same coarse-grained model, calibrated on the fraction of proteins bound from earlier experiments, predicts the number of adsorbed proteins per nanoparticle measured here by differential centrifugal sedimentation within about 10% at intermediate concentrations. The authors conclude that the corona's composition and structure can keep evolving over long times, which should matter for how nanoparticles interact with cells.

Load-bearing premise

The whole multilayer and glassy picture rests on the assumption that a protein adsorbed in the hard corona partially unfolds and thereby attracts other transferrin molecules through a particular three-body force, whose strength is fixed at 3.75 kT but whose range kappa has to be retuned for each nanoparticle concentration.

Editorial extensions

If this is right

If the soft corona is glassy, it does not reach equilibrium within the seconds-long observation window; its inner layer can continue to reorganize on longer timescales, so a corona composition measured at one time may not represent what a cell encounters later.
The three-layer structure means protein counts above monolayer saturation require multilayer adsorption, with the fraction-bound curve showing slope changes at layer saturation points, around [Tf]/[NP] = 320 and between roughly 700 and 1000.
Layer stabilization times grow outward: approximately 0.2 seconds for the hard corona, about twice that for the inner soft-corona layer, and two to three times longer for the outer soft-corona layer at the highest concentration studied.
At protein concentrations approaching those in blood, the glassy slowdown could become biologically relevant and should be considered when analyzing nanoparticle-cell interactions over time.
The model's adsorbed-protein counts agree with the differential centrifugal sedimentation measurements to within about 10% at intermediate concentrations, supporting the use of this coarse-grained approach in protein-rich environments.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper: if the glassy behavior is generic, many reported corona 'equilibration' times may be underestimates, and the biological identity of a nanoparticle should be reported together with its exposure history rather than as a single final composition.
Beyond the paper: the three-body interaction is a stand-in for explicit unfolding, so a direct extension would add conformational degrees of freedom to the coarse-grained protein and test whether the same plateau and power-law decay emerge without tuning kappa.
Beyond the paper: the crowding mechanism predicts a testable experimental signature, namely that the inner soft-corona layer should show slowed, collective exchange whose plateau height shifts when the outer layer is made more or less crowded.
Beyond the paper: the need to recalibrate kappa for each nanoparticle concentration suggests the model is not yet predictive across conditions without input from adsorption isotherms, so applying it to other protein-nanoparticle pairs still requires per-system experimental calibration.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things to know about this one. The paper adds a three-body protein-protein-NP attraction to the authors' existing BUBBLES coarse-grained model, and that interaction produces a three-layer corona with an inner soft layer that shows a plateau in the density autocorrelation function. That is a genuine extension beyond their 2016 hard-corona model. They also ship open-source code, a tutorial, and new DCS measurements of adsorbed protein counts, which is real work.

The structural part holds up. The radial density profile shows three distinct layers that saturate in sequence, and single-protein tracking shows the inner layer is irreversible while the outer two exchange. That is consistent with the hard/soft corona picture and a useful demonstration that a simple three-body term can generate multilayer adsorption.

The soft spots are all in the glassy claim. The three-body parameters, especially kappa, are fit to Milani's fraction-bound data and then the model is said to reproduce that data. That is calibration, not prediction. Kappa is then extrapolated to the experimental NP concentration with an ad hoc formula, so the autocorrelation in Figure 9 is computed at 1 mg/mL, not at the 0.1 mg/mL used in the experiments. That curve has no error bars and no statement of how many independent runs went into it; for the small occupancy fluctuations of one layer, the long-lag estimator is biased, and the plateau may be an averaging artifact. And the paper says 'within 10%' agreement, but its own Table 1 and Figure 11 show -33% and +18% deviations at the two ends. That is an overstatement.

None of this kills the multilayer modeling, but it does mean the headline claim — soft corona is dominated by glassy evolution — is not empirically or statistically secured. It is a model output from a fitted potential. The authors are upfront that the model is non-transferable and needs recalibration per condition, which is honest, but it also means the glassy behavior is a property of the model, not yet of the corona.

This paper is for people doing coarse-grained corona kinetics, and those who might reuse BUBBLES. It deserves a serious referee, not a desk reject, but the referee should ask for multiple independent runs with error bars on the autocorrelation, a corrected statement about validation accuracy, and a clear separation between calibrated reproduction and independent prediction. I would not cite the glassy result, but I might cite the multilayer extension and the code.

Referee Report

4 major / 7 minor

Summary. The paper introduces a coarse-grained computational model (BUBBLES) with a three-body protein–protein–nanoparticle interaction to simulate multilayer adsorption of transferrin (Tf) on polystyrene nanoparticles. The model parameters are calibrated against the fraction-bound data of Milani et al. (2012), extrapolated to the experimental nanoparticle concentration, and used to predict two soft-corona layers (SC1 and SC2) on top of the hard corona. The authors analyze the density autocorrelation of these layers and claim that the inner soft-corona layer exhibits glassy dynamics related to crowding. They attempt to validate the model with new DCS experiments measuring the size distribution and adsorbed protein counts. The central claim is that the soft corona is dominated by a glassy evolution, with consequences for the biological identity of nanoparticles.

Significance. If the glassy-corona claim were robust, it would be a valuable advance in understanding protein corona dynamics, and the open-source BUBBLES package with its tutorial is a useful community resource. However, the central claim is not empirically secured: the interaction parameters are fitted to the same fraction-bound data the model is said to reproduce, the glassy autocorrelation is computed from an apparently single trajectory with no error bars, and the DCS validation is semi-quantitative at best and internally inconsistent with the 'within 10%' summary. The manuscript is honest about the model's non-transferability, but this limits the generality of the conclusions. The structural layer predictions and the glassy dynamics are outputs of a calibrated potential rather than independent predictions, so the significance for the physical system remains conditional.

major comments (4)

[§3.1, Eq. (5), Figure 4] The parameters of the three-body interaction, ε3b and κ, are fitted to the same Milani fraction-bound data that the model is then said to reproduce; the structural layers (Figure 6) and the glassy autocorrelation (Figure 9) are outputs of this fitted potential, not independent tests. Section 2.2 explicitly states that the model is not transferable and requires preliminary adsorption isotherms for each protein–NP pair and thermodynamic condition, so the claim that the soft corona is 'dominated by a glassy evolution' cannot be read as an experimentally validated property of the physical system unless out-of-sample dynamic data are provided.
[§3.3.2, Eq. (13), Figure 9] The glassy autocorrelation is computed from what appears to be a single trajectory of one nanoparticle: no error bars, no multiple independent runs, and the long-lag estimator for a finite observation window is biased, so the distinction between a plateau (SC1) and a power law (SC2) may be an averaging artifact. With an observation window of only 0.1 s (about two decades), the nonexponential decay and plateau need statistical support before the glassy claim can be accepted.
[§3.4, Table 1, Figure 11, §4] The DCS validation is not 'within a 10% error' as claimed in the Discussion: the same section reports deviations of about −33% at [Tf]/[NP]=400 and +18% at [Tf]/[NP]=1500. In addition, the DCS-derived corona thickness (2.9–4.7 nm) is comparable to a single Tf radius (3.72 nm), not the multilayer structure extending to ~20 nm predicted by the model; thus the DCS data provide, at best, a test of adsorbed mass in an effective monolayer, not of the multilayer soft corona or its dynamics.
[§3.1, Eq. (11), §2.3] The extrapolation of κ to the experimental concentration CNP=0.1 mg/mL is based on an ad hoc function (Eq. 11) fitted to only four simulated concentrations, and the dynamic simulations used for the glassy analysis are all performed at CNP=1 mg/mL with κ=35 nm. The connection between the simulated glassy behavior and the actual experimental condition is therefore not established; either simulations at the extrapolated κ value or a sensitivity analysis over κ is needed.

minor comments (7)

[§2.2] The numerical values of the reduced surface potentials γ_Tf and γ_NP (or the zeta potentials from which they derive) are not reported; please provide them.
[Eq. (11)] The displayed formula contains garbled text ('radicaltpext/radicaltpext'); the equation should be typeset correctly.
[§3.3.2] The notation C1(t) and C2(t) is used in the text without explicitly tying it to Ci(t) from Eq. (13); please define the index i in the text.
[§3.1] It is unclear whether ε3b=3.75 kBT was also optimized or fixed a priori; the sentence 'we adjust the model's parameter and find that, by fixing ε3b=3.75 kBT and varying κ' suggests both, so clarify the fitting protocol.
[Footnote 1] The preliminary calculation for RNP=41 nm is cited to an unpublished reference ('Jareño and Delia, 2015'); please mark it as a personal communication or remove it.
[§3.4] The DCS experiments use ~110 nm diameter NPs while the simulation uses RNP=35 nm; the authors state they adjusted protein concentration to maintain the same [Tf]/[NP] ratios, but this means the surface area per NP differs, so the comparison of NAds should be justified more explicitly.
[Figure 9] The axes of the autocorrelation plot are not legible in the manuscript text; ensure the time axis with units is clearly visible.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the three-body potential is fit to the adsorption isotherm, but the structural and dynamical claims are distinct outputs and the DCS data are an out-of-sample check.

full rationale

The paper is a calibrated coarse-grained simulation study, not a first-principles derivation, and it says so explicitly: 'the strong approximations in the model make it not transferable ... preliminary experiments are necessary to measure the adsorption isotherms' (Section 2). Section 3.1 openly calibrates the three-body interaction: 'we set the NP concentration C_NP and find the corresponding κ and ε3b that best fit the experimental data (Milani et al., 2012) for the Tf fraction bound fB.' The layer-resolved density profiles (Eq. 12, Fig. 6) and the layer autocorrelation functions (Eq. 13, Fig. 9) are simulation outputs of that calibrated model, not quantities that were fitted; no equation in the paper makes C_i(t) equal to the fitted fB by construction. The DCS validation (Section 3.4, Table 1, Fig. 11) uses new experimental data, so the comparison is out-of-sample even though it measures the closely related quantity NAds; the reported deviations (-33% and +18% at the extremes) show the prediction is not trivially forced. The glassy-corona claim is admittedly not independently validated by experiment, which is an evidential weakness and a correctness risk, but it is not circularity. The self-citations (Vilanova et al., 2016) are to a published, experimentally tested prior model, and no uniqueness theorem or unverified load-bearing assertion is imported from those papers.

Assumptions & free parameters 6 free parameters · 7 assumptions · 1 invented entities

The model rests on one fitted three-body interaction (Eq. 5) whose strength epsilon_3b and range kappa are calibrated to the same Milani et al. fraction-bound data used as the experimental reference. The DLVO and Langevin frameworks are standard, but the effective parameters, including the Hamaker constant and zeta potentials, are inputs or fits. No new physical entity is proposed beyond the effective three-body force, which has no direct experimental handle. The concentration extrapolation formula Eq. 11 is an additional ad hoc assumption.

free parameters (6)

epsilon_3b (three-body interaction energy) = 3.75 kBT
Fixed ad hoc to reproduce the Milani fraction-bound data across CNP values; not derived from protein structure or experiment (Section 3.1).
kappa (three-body decay length) = 35 nm at CNP=1 mg/mL; extrapolated to 57 +/- 5 nm at CNP=0.1 mg/mL.
Fitted per CNP to match the experimental fraction bound; the CNP dependence is then fit by Eq. 11 with kappa0=(62 +/- 4) nm and C0=(0.54 +/- 0.14) mg/mL.
Hamaker constant AH = 15 kBT
Chosen to create a strongly bound hard corona and part of the DLVO potential; not independently measured in this paper (Section 2.2.1).
Born repulsion length sigma = 0.5 nm
Set as the characteristic Born repulsion in the DLVO potential; treated as an input without independent justification in this paper.
delta and omega of the Gaussian well in U3b = delta = 2 RTf = 7.44 nm; omega = delta/4 = 1.86 nm.
Chosen by hand in the three-body potential of Eq. 5; omega is small compared to the protein size.
zeta potentials or reduced surface potentials gamma_Tf and gamma_NP = Not stated numerically in the text.
The computational section says the zeta potential in PBS is tuned, and the model section says parameters are calibrated from adsorption isotherms; the numerical values are not reported.

assumptions (7)

standard math Langevin dynamics with Gaussian noise and the fluctuation-dissipation relation models the solvent effects.
Eq. 6; standard stochastic mechanics used as background.
domain assumption The protein can be represented as a rigid soft sphere with radius 3.72 nm and mass 80 kDa in implicit solvent.
Used throughout the model; ignores internal protein degrees of freedom except through the ad hoc three-body interaction (Section 2.2).
domain assumption DLVO theory with an added Born repulsion describes the protein-nanoparticle interaction potential.
Eqs. 2 to 4; the Hamaker constant and zeta potentials are effective inputs, not derived from first principles.
ad hoc to paper Proteins adsorbed in the hard corona partially unfold on the nanoparticle surface and attract other transferrin molecules.
Stated as a hypothesis in Section 2.2; this mechanism is implemented as U3b in Eq. 5 and is not directly measured.
ad hoc to paper The three-body attraction decays as exp(-di dj / kappa^2) and has a Gaussian well centered at delta = 2 RTf with width omega = delta/4.
Eq. 5; the functional form is chosen for convenience and the parameters are fit to the fraction-bound data.
domain assumption The buffer and reservoir algorithm keeps protein concentration constant in the reaction region and does not bias adsorption kinetics.
Section 2.3.2; the semipermeable wall and stochastic diffusion protocol are not verified against a full periodic system.
ad hoc to paper kappa(CNP) follows kappa0 / sqrt(1 + CNP/C0) outside the fitted concentration range, down to CNP=0.1 mg/mL.
Eq. 11 is fit to only four simulated concentrations and is used to extrapolate kappa to the experimental NP concentration; no physical derivation is given.

invented entities (1)

Surface-induced three-body protein-protein-nanoparticle attraction U3b
purpose: To make proteins near the nanoparticle attract further proteins, generating the second and third corona layers that the paper calls the soft corona.
Introduced ad hoc in Eq. 5; no direct measurement of this force is provided. It is only constrained indirectly through the fitted fraction-bound curve and the model's own layer predictions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Characterizing the Hard and Soft Nanoparticle-Protein Corona with Multilayer Adsorption." pith.science (2026). https://pith.science/paper/ZSEEKKQZ

@misc{pith2026241113279,
  author       = {Pith},
  title        = {Pith review of: Characterizing the Hard and Soft Nanoparticle-Protein Corona with Multilayer Adsorption},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZSEEKKQZ}},
  note         = {Machine review of arXiv:2411.13279}
}

read the original abstract

Nanoparticles (NPs) in contact with biological fluid adsorb biomolecules into a corona. This corona comprises proteins that strongly bind to the NP (hard corona) and loosely bound proteins (soft corona) that dynamically exchange with the surrounding solution. While the kinetics of hard corona formation is relatively well understood, thanks to experiments and robust simulation models, the experimental characterization and simulation of the soft corona present a more significant challenge. Here, we review the current state of the art in soft corona characterization and introduce a novel open-source computational model to simulate its dynamic behavior, for which we provide the documentation. We focus on the case of transferrin (Tf) interacting with polystyrene NPs as an illustrative example, demonstrating how this model captures the complexities of the soft corona and offers deeper insights into its structure and behavior. We show that the soft corona is dominated by a glassy evolution that we relate to crowding effects. This work advances our understanding of the soft corona, bridging experimental limitations with improved simulation techniques.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

[44]

doi:10.1038/nnano.2010.250 Derjaguin, B., and Landau, L. (1993). Theory of the stability of strongly charged lyophobic sols and of the adhesion of strongly charged particles in solutions of electrolytes. Prog. Surf. Sci. 43, 30 –59. doi:10.1016/0079-6816(93)90013-L Durà-Faulí, B., Bianco, V., and Franzese, G. (2023). Hydrophobic homopolymer ’s coil–globul...

work page arXiv 1993
[573]

doi:10.1038/s41467-020-20884-9 Singh, N., Marets, C., Boudon, J., Millot, N., Saviot, L., and Maurizi, L. (2021). In vivo protein corona on nanoparticles: does the control of all material parameters orient the biological behavior? Nanoscale Adv. 3, 1209 –1229. doi:10.1039/D0NA00863J Soliman, M. G., Martinez-Serra, A., Antonello, G., Dobricic, M., Wilkins,...

work page Pith review arXiv doi:10.48550/arxiv.1009.4330 2021
[6443]

S., O ’Connell, D., Mahon, E., Monopoli, M

doi:10.1038/s41598-021-84029-8 Pitek, A. S., O ’Connell, D., Mahon, E., Monopoli, M. P., Bombelli, F. B., and Dawson, K. A. (2012). Transferrin coated nanoparticles: study of the bionano interface in human plasma. PLoS One 7, e40685. doi:10.1371/journal.pone.0040685 Qiu, C., Zhang, Z., Xiao, M., Yang, Y., Zhong, D., and Peng, L.-M. (2017). Scaling carbon ...

work page arXiv 2012

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [44]

doi:10.1038/nnano.2010.250 Derjaguin, B., and Landau, L. (1993). Theory of the stability of strongly charged lyophobic sols and of the adhesion of strongly charged particles in solutions of electrolytes. Prog. Surf. Sci. 43, 30 –59. doi:10.1016/0079-6816(93)90013-L Durà-Faulí, B., Bianco, V., and Franzese, G. (2023). Hydrophobic homopolymer ’s coil–globul...

work page arXiv 1993

[2] [573]

doi:10.1038/s41467-020-20884-9 Singh, N., Marets, C., Boudon, J., Millot, N., Saviot, L., and Maurizi, L. (2021). In vivo protein corona on nanoparticles: does the control of all material parameters orient the biological behavior? Nanoscale Adv. 3, 1209 –1229. doi:10.1039/D0NA00863J Soliman, M. G., Martinez-Serra, A., Antonello, G., Dobricic, M., Wilkins,...

work page Pith review arXiv doi:10.48550/arxiv.1009.4330 2021

[3] [6443]

S., O ’Connell, D., Mahon, E., Monopoli, M

doi:10.1038/s41598-021-84029-8 Pitek, A. S., O ’Connell, D., Mahon, E., Monopoli, M. P., Bombelli, F. B., and Dawson, K. A. (2012). Transferrin coated nanoparticles: study of the bionano interface in human plasma. PLoS One 7, e40685. doi:10.1371/journal.pone.0040685 Qiu, C., Zhang, Z., Xiao, M., Yang, Y., Zhong, D., and Peng, L.-M. (2017). Scaling carbon ...

work page arXiv 2012