Arrangement and dynamics of individual cyclodextrins on the surface of core–shell micelles

I. S. Vaskan*^a, V. A. Dimitreva^ab, A. A. Piryazev^d, E. N. Subcheva^d, N. V. Bovin^a, A. B. Tuzikov^a, V. A. Oleinikov^ab and A. V. Zalygin*^ac
aShemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, Moscow, 117997, Russia. E-mail: vaskan.ivan@inbox.ru; zalygin.anton@gmail.com
^bNational Research Nuclear University Moscow Engineering Physics Institute, Moscow, 115409, Russia
^cLebedev Physical Institute, Russian Academy of Sciences, Troitsk Branch, Troitsk, Moscow 108840, Russia
^dSirius University of Science and Technology, 1 Olympic Ave, 354340, Sochi, Russian Federation

First published on 15th August 2025

---

---

Introduction

The development of nanotechnology resulted in the emergence and widespread use of various nanomaterials for drug delivery (for a review, see ref. 1), but many of them demonstrate low efficiency and serious side effects² associated with toxicity, antigenicity, and immunogenicity. Low efficiency is explained by the fact that after entering the biological environment (bloodstream), nanoparticles (NPs) bind its components. This adsorption layer, called the biomolecular corona,³ dramatically affects the functionality of NPs, causing a loss of NP targeting. Experimental studies of corona formation have shown that this process depends on the surface properties of NPs, and, therefore, by changing them, it is possible to control corona formation.⁴ Currently, the most commonly used coating material for nanoparticles is polyethylene glycol (PEG, PEGylation), which increases the circulation time of NPs in the bloodstream. However, long-term use of PEGylated therapeutics causes the formation of antibodies to PEG, followed by the effect of accelerated blood clearance (ABC).⁵ Recent work has focused on NPs with surfaces that provide them with properties such as stimulus sensitivity,⁶ low protein adsorption,^7–9 and improved targeting.¹⁰ However, overcoming the above-mentioned drawbacks one by one, along with imparting stimulus sensitivity, etc., leads to multiple, sometimes overly complex modifications, which in turn give rise to the next cascade of problems, making the practical application of NPs unpromising. It becomes obvious that the molecules that make up NPs must perform not one but several tasks at once. In this regard, cyclodextrins are a universal tool for various surface modifications due to their unique physicochemical properties.¹¹

Cyclodextrins are natural cyclic oligosaccharides composed of a(1–4)-linked glucose units produced by enzymatic degradation of starch. The most common CDs have six, seven, or eight glucose units and are named α-CD, β-CD, and γ-CD, respectively. Glucose units are closed in a cycle in such a way that all primary hydroxyls are on one rim of the cycle, while secondary hydroxyls occupy the opposite rim. Such a configuration results in a truncated cone three-dimensional structure with a hydrophilic exterior and hydrophobic interior. CDs are non-toxic and non-immunogenic, and the presence of an apolar cavity, which can host small hydrophobic molecules, provides wide opportunities for grafting various chemical compounds by means of host–guest interactions.¹² Due to these remarkable properties, CDs have gained much attention in biomaterials research, resulting in numerous CD-based nanomaterials such as micelles,^13–23 hydrogels,²⁴ nanosponges,²⁵ functional glyconanomaterials,²⁶ and CD-peptide conjugates²⁷ that have been proposed for biomedical applications. For instance, Adeli et al. utilized CDs to fabricate temperature- and pH-sensitive micelles with high drug loading and excellent anti-cancer activity.¹³ In another study, Zhang et al. developed β-CD-based micelles with long in vivo circulation time.¹⁴ Rahmani et al. fabricated novel pH-responsive β-CD grafted micelles that induce tumor cell apoptosis.¹⁵ These results indicate great potential of CDs for increasing the efficiency of delivery systems. However, despite these remarkable results, not all molecular mechanisms of the positive properties of CDs are well understood. And since they are determined not only by the structure of the CDs as such, but also by the dynamics of the NPs as a whole, the study of the surface dynamics of new nanomaterials is an essential step towards increasing the efficiency of targeting NPs.

In this paper, we study self-assembled micelle-like NPs, i.e., no stages of covalent binding of CDs or other chemical modification of the nanomatrix are required to construct the target NPs. Instead, the required NP properties are embedded in the structure of the monomer molecule belonging to the class of FSL (function–spacer–lipid) constructs,²⁸ which are conjugates of the phospholipid DOPE, a hydrophilic spacer (CMG)₄, and CDs. It should be noted that (CMG)₄ does not simply connect the F and L regions but participates in the formation of the NPs and is largely responsible for their properties. Transmission electron microscopy (TEM) and small-angle X-ray scattering (SAXS) were combined with atomistic molecular dynamics simulations in such a way that the experimental data served as a basis for creating NP models for simulations, which in turn made it possible to study the NP surface with high temporal and spatial resolution.

Our approach was based on the integration of experimental and computational methods (Scheme 1). First, using TEM and SAXS methods, the formation of nanoparticles was visually confirmed, and their structural parameters (size, radius of gyration) were determined. These data served as the basis for creating initial models of nanoparticles. The simulation results were verified by fitting the theoretical SAXS curves (derived from MD models) to experimental curves. Verified models were used to obtain topographic maps of nanoparticles, which were used to study the dynamics of the nanoparticle surface and CDs. This approach allowed us to study for the first time the structure and dynamics of CD-decorated nanoparticles, as well as the molecular properties of individual CD residues, such as localization, orientation, and mobility. This work aims to deepen our understanding of CD behavior within functionalized NPs, which is required for the rational design of drug delivery systems.

Materials and methods

Materials

All salts, solvents, lipids, and other chemicals were from Sigma (St. Louis, USA). CD-(CMG)₄-DOPE constructs were synthesized (Scheme 1) from monoamino derivatives of alpha- and beta-cyclodextrins and H-(MCMG)₄-OH amino acid, which, after hydrolysis of methyl ester groups, form a hydrophilic spacer (CMG)₄.

The introduction of the methylsulfonylethoxycarbonyl protective group (MSC) to the amino group of H-(MCMG)₄-OH, activation of the carboxyl group, conjugation with the monoamino derivative of cyclodextrin, removal of protective groups and conjugation with the lipid block DOPE-Ad-ONSu gave the desired product (Scheme 2).

H-(MCMG)₄-OH amino acid is courtesy of A. A. Formanovsky (Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry of the Russian Academy of Sciences). Aminoderivatives of cyclodextrins 6-monoamino-6-monodeoxy-α-cyclodextrin and 6-monoamino-6-monodeoxy-β-cyclodextrin were from AraChem (Netherlands). 2-(Methylsulfonyl)-ethyl-N-succinimidyl carbonate (MSC) was from Fluka, N-hydroxysuccinimide and N,N′-dicyclohexylcarbodiimide were from Merck, Sephadex LH-20 was from Sigma-Aldrich. DOPE-Ad-ONSu was synthesized as described in ref. 23.

α-CD-(CMG)₄-DOPE

TLC: R_f = 0.32 (CHCl₃/MeOH/H₂O 3:6:1).

¹H NMR (CD₃OD/D₂O 1:1, 303 K, 700 MHz) δ 5.374 (m, 4H; CHCH), 5.291 (m, 1H; 2-CH of glycerol), 5.043 (m, 6H; 6 H-1 of Gluc), 4.753 (s, HOD), 4.474 (dd, J = 12.2, 2.5 Hz, 1H; 1-CH of glycerol), 4.238 (dd, J = 12.2, 7.5 Hz, 1H; 1-CH′ of glycerol), 4.280–3.741 (m, 63H; 24H of Gluc, 32H of CMG, 7H of DOPE), 3.603 (m, 10H of Gluc), 3.440 (m, 2H; CH₂N), 3.344 (p, J = 1.7 Hz; CD₂HOD methanol), 2.368 (m, 6H; 3 CH₂CO), 2.290 (m, 2H; CH₂CO), 2.032 (m, 8H; 2 CH₂CCCH₂), 1.633 (m, 8H; 4 CH₂CH₂CO), 1.321 (m, 40H; 20 CH₂ of oleoyl), 0.912 (t, J = 7.1 Hz, 6H; 2 CH₃) ppm.

MALDI-TOF MS: found m/z 2786.218; calc. for C₁₁₅H₁₈₈N₁₄O₅₉PNa₂ [M−H + 2Na]⁺ 2786.167.

β-CD-(CMG)₄-DOPE was obtained similarly to α-CD-(CMG)₄-DOPE, the yields are similar.

TLC: R_f = 0.34 (CHCl₃/MeOH/H₂O 3:6:1).

¹H NMR (CD₃OD/D₂O 1:1, 303 K, 700 MHz) δ 5.375 (m, 4H; CHCH), 5.291 (m, 1H; 2-CH of glycerol), 5.056 (m, 7H; 7 H-1 of Gluc), 4.762 (s, HOD), 4.472 (dd, J = 12.2, 2.5 Hz, 1H; 1-CH of glycerol), 4.239 (dd, J = 12.2, 7.5 Hz, 1H; 1-CH′ of glycerol), 4.284–3.735 (m, 67H; 28H of Gluc, 32H of CMG, 7H of DOPE), 3.611 (m, 12H of Gluc), 3.439 (m, 2H; CH₂N), 3.344 (p, J = 1.7 Hz; CD₂HOD methanol), 2.374 (m, 6H; 3 CH₂CO), 2.292 (m, 2H; CH₂CO), 2.042 (m, 8H; 2 CH₂CCCH₂), 1.646 (m, 8H; 4 CH₂CH₂CO), 1.336 (m, 40H; 20 CH₂ of oleoyl), 0.915 (t, J = 7.1 Hz, 6H; 2 CH₃) ppm.

MALDI-TOF MS: found m/z 2948.268; calc. for C₁₂₁H₁₉₈N₁₄O₆₄PNa₂ [M−H + 2Na]⁺ 2948.220.

Samples were prepared by dissolving weighed portions of α-CD-(CMG)₄-DOPE and β-CD-(CMG)₄-DOPE in PBS buffer. Chemical structures of α-CD-(CMG)₄-DOPE and β-CD-(CMG)₄-DOPE monomers are shown in Fig. 1. For TEM and SAXS studies, concentration of all samples was 2 mM.

Critical micelle concentration (CMC) measurements

The critical micelle concentration (CMC) was determined by the fluorescence probe technique using pyrene as a fluorescence probe. The samples for CMC measurements were prepared by serial dilution of stock sample PBS solution with concentration varied from 250 to 0.025 μM. The pyrene concentration (0.2 μM) was kept constant during measurements. The pyrene fluorescence spectra were recorded using an RF-5301PC spectrofluorophotometer (Shimadzu, Japan): λ_ex = 334 nm, λ_em = 350–450 nm, slidwidth(ex) = slidwidth(em) = 3 nm, and step width 1 nm. The sample concentrations were plotted against the I₃₈₅/I₃₇₅ peak ratio of the fluorescence intensity and subsequently fitted with a logistic function, whereby the CMC value of each sample was determined.

Transmission electron microscopy

Sample preparation: solutions of the studied samples were vortexed for 30 seconds, then applied to the surface of a 300-mesh copper grid with a carbon/formvar coating (EMCN, China) in a volume of 2.5 μL and incubated for 1 minute. Excess liquid was drawn off with filter paper. Contrasting was performed with a 1% aqueous solution of uranyl acetate for 1 minute. The grids were dried at room temperature in air.

The image capture was carried out on a Crossbeam550 scanning electron microscope with a detector for transmission electron microscopy (Carl Zeiss, Germany) at an accelerating voltage of 30 kV.

Small angle X-ray scattering (SAXS)

SAXS experiments were performed using a XeuSS 3.0 WAXS/SAXS (Xenocs, France) machine with a GeniX3D generator (λ = 1.54 Å), producing a beam with a 300 × 300 μm size. A Pilatus300k detector was used for data collection, and the sample–detector distance was 1.3 m. The minimum projection method for a series of images was used to reduce background noise (12 frames with 1 h exposition). Norm wave vector s (s = 2sinθ/λ, where θ is the Bragg angle) was calibrated via several orders of AgBe. The data were corrected for the solvent scattering and processed using standard procedures with a program suite ATSAS.²⁹ For in-depth structural analysis of the supramers formed by constructs, we performed SAXS measurements at a concentration of 2 mM higher than the CMC value. SAXS curves of α-CD-(CMG)₄-DOPE and β-CD-(CMG)₄-DOPE micelles in a sodium phosphate buffer at pH 7.4 were found.

Molecular dynamics simulations

Atomistic molecular dynamics simulations were carried out using GROMACS package version 2024.³⁰ The starting structure consisted of spherically arranged monomers (pre-micelle) solvated in a cubic cell with SPC water and 150 mM NaCl, maintaining a distance of 1 nm between the solute and box edges (Fig. 2). The number of monomers was selected to be 190 so that the radius of gyration of simulated micelles (trajectory average) was consistent with the experimental value derived from the SAXS data. GROMOS 54A7 force field topologies for monomers were generated using Automated Topology Builder (ATB ID: AG3Y, 5BLH, KJS9).³¹ First, we performed a series of energy minimization steps (equal to the number of monomers) with turned-off non-bonded interactions using the algorithm³² (the soft-core parameter is 0.5, the soft-core power is 1, and the radius of the interaction is 0.3) to decouple each monomer in pre-micelles. Next, a short 60 ps simulated annealing stochastic dynamic run in the NVT ensemble was performed to reach the global minimum. The system was gradually cooled from 800 to 293.15 K with temperature decrease by 100 K every 10 ps. Other parameters were the same as in the production run described below. Next, after energy minimization and two 10 ps equilibration runs in NVT and NPT ensembles, 450 ns production runs were performed using the NPT ensemble with periodic boundary conditions applied. The leap-frog integrator with an integration time-step of 1.6 fs was used. Bonds to hydrogen atoms were constrained using the LINCS algorithm.³³ A cut-off of 1 nm was used for Lennard-Jones interactions. Electrostatic interactions were treated using the Particle–Mesh Ewald (PME) algorithm³⁴ in combination with a real space cut-off of 1 nm, a grid spacing of 0.12 nm, and an interpolation order of 4. Periodic boundary conditions were applied in all three dimensions. Temperature was maintained at 298 K with the V-rescale³⁵ algorithm while the C-rescale³⁶ algorithm was used to maintain pressure at 1 bar. VMD 1.9.3³⁷ was used for visualization. Fitting to the experimental SAXS curve was performed using the CRYSOL³⁸ program from the ATSAS package. First 350 ns were omitted for equilibration and remaining 100 ns (1000 frames) were used for processing, including trajectory rotational fit to remove displacement of atoms caused by nanoparticle rotation, root-mean-square deviation (RMSD) calculation for surface atoms, extraction of frames with minimal, average and maximal RMSD over trajectory, and topography map generation. Topography generation was performed similarly to previous work.³⁹ Briefly, lipid atoms were excluded from MD coordinates and the GROMACS built-in tool was used for Connolly surface computation using an algorithm⁴⁰ with a probe radius of 1.42 Å. The obtained surface was plotted in spherical coordinates.

Mixed micelles consisted of 152 DOPE lipids and 38 CD-(CMG)₄-DOPE conjugates were simulated with the same protocol, except for the production run time was 200 ns and last 75 (750 frames) ns were taken for processing.

Results and discussion

Although the studied FSLs have clearly expressed lipid and polar fragments, their association into a classical micelle-like particle did not look obvious due to the rather large size of the polar “head” (L), which could be expected to be even larger upon merging of L&S regions. To make sure that CD-containing conjugates self-assemble into micellar nanoparticles, as a first step of the study, we have determined critical micelle concentration (CMC). Results are shown in Fig. 3.

It can be seen that for both α-CD- and β-CD-containing conjugates, CMC values are around 15 μM. These values indicate high NP stability, which is crucial for in vivo applications. Low CMC values are a significant advantage of lipid-based micelles compared to their polymeric counterparts, which have CMC values around 5 mM. Next, we have applied transmission electron microscopy (TEM) to visually confirm micelle formation, and determine their sizes. Obtained images are shown in Fig. 4.

Many round-shaped particles are formed. α-CD- and β-CD-bearing micelles show a similar size of about 10–12 nm, which agrees with previous studies of similar conjugates.^31,34 To determine the nanoparticle size and internal structure under physiological conditions, the small-angle X-ray scattering (SAXS) technique was applied. Resulting SAXS curves are shown in Fig. 5. Scattering curves reveal noticeable upturns at very small angles, in the range of the scattering vectors. These results indicate that monodisperse ∼12 nm particles co-exist in solution with a moderate number of large assemblies. The chi-squared parameter for this processing has reached a value of 1.390. This observation correlates well with the TEM data (Fig. 4). It should be noted that the SAXS profiles are characteristic for scattering from micelle-like globules. Since large aggregates consisting of several globules contribute mostly to the results at the very small angles, discarding this portion of the data diminishes the influence of these particles when determining the shape of the practically monodisperse globules formed by the molecules. The particle sizes of about 12 nm obtained by this method are generally consistent with the results of other used methods.

Such curves are typical for micellar nanoparticles. The radius of gyration (R_g) derived from these curves was found to be 5.2 nm and 5.4 nm for α-CD- and β-CD-bearing micelles, respectively, which is in agreement with the TEM study. Another feature of the SAXS technique is that experimental data can be directly compared with curves from nanoparticle models obtained using molecular dynamics (MD) simulations. Thus, SAXS data can be used to validate MD models, which, in case of a good fit, allows one to study the structure of nanoparticles with the resolution of atomistic molecular dynamics. Fig. 6 also shows the fit of the curve calculated from the MD model (red line) to experimental data, with chi-square (χ²) values of 1.5 and 1.2 for α-CD- and β-CD-bearing micelles, respectively. Such χ² values indicate that simulated nanoparticles are consistent with nanoparticles in solution. Snapshots of simulated nanoparticles (trajectory average) are also shown in Fig. 6. From these models, it can be seen that micelles possess a core–shell structure; the hydrophobic core is almost completely covered by a hydrophilic spacer, with cyclodextrins located predominantly on the surface. With regard to the arrangement of cyclodextrins, two points that affect CDs’ functionality are noteworthy. First, it is evident that many cyclodextrins are located close to each other, which can lead to overlaps that prevent the binding of ligands by individual CDs. Second, cyclodextrins have different orientations relative to the external environment. To address these points and study arrangement of cyclodextrins in more detail, we have generated and analyzed topography of micelles, shown in Fig. 7.

Nanoparticles’ surface is plotted in spherical coordinates with heights over the core (shell thickness) indicated by a color. Cyclodextrin positions with three possible orientations are marked with geometrical figures. Triangles, inverted triangles, and squares correspond to the secondary rim outward, primary rim outward, and edge-on CD orientation, respectively. Monomers with these CD orientations are also shown in Fig. 7. Topography maps enable assessment of each CD and identify those that retain their functionality. By comparing the colors of the figure and its surroundings, one can judge whether the given CD is on the surface or entangled by a spacer or other CDs. It is clear that to retain the ability to bind ligands, cyclodextrins must be located on the surface, have a secondary rim outward orientation and not overlap with other molecules. From topography maps, it can be seen that only a small fraction of CD residues satisfies these conditions. We identify 41 and 45 functional CD residues (out of 190) for α-CD- and β-CD-bearing micelles, respectively. These functional CD residues are distributed throughout the surface of the nanoparticles, which indicates that the presence of a CD residue in each monomer is not mandatory from a quantitative perspective. Next, we assess the mobility of CD residues with appropriate orientation since the ability of CDs to slightly shift their position towards ligands may increase the chance of a binding event. To do this, we study the dynamics of nanoparticle topographies by comparing the extreme states between which the surface of the nanoparticles fluctuates, which corresponds to MD trajectory frames with minimal and maximal root-mean-square deviations (RMSD). The corresponding topographies are shown in Fig. 8.

The first thing to notice is that both α-CD- and β-CD-bearing micelles possess a static surface formed by a practically immobile shell. Most CD residues also stayed in the same position. However, there are several displaced CD residues, which indicate their mobility. For α-CD-bearing micelles, there are residues 37, 55, 58, 79, 104, 120, and 189 (7 in total) and in the case of β-CD-bearing micelles only residue 185 is mobile. Furthermore, some CD residues are able to change their orientation, which suggests that they also have some mobility. In the case of α-CD-bearing micelles there are residues 2, 38, 46, 78, 90, 94, 110, 121, 127, 134, 142, and 182 (12 in total). β-CD residues with a switched orientation are 41, 99, 118, 164, 167, 181, and 190 (7 in total). Thus, α-CD residues demonstrate overall greater mobility than β-CD. A possible reason for the observed static nature of the surface and low mobility of CD residues may be the formation of hydrogen bonds. Indeed, calculation of hydrogen bonds reveals that there are 2084 and 3041 H-bonds between shell moieties of α-CD- and β-CD-bearing micelles, respectively, that stabilize the surface structure.

To find out whether such dynamics of the surface structure and CD orientation are a consequence of the high density of the shell and CDs, we simulated the self-assembly of mixed micelles in which only 20% of the total number of molecules have spacer and CD residues. Dynamics of the corresponding nanoparticles’ topographies shown in Fig. 9.

Although a reduction in shell density resulted in significantly less core coverage and a thinner shell, the shell is still static with small changes in topographies and CD residues are almost equally distributed among possible orientations. As a result, only a small fraction of CD residues retains their functionality. These results indicate that CD orientation is an intrinsic property of a given combination of the spacer and CDs, and special measures should be taken to achieve a preferential secondary rim orientation of the CDs. One of the possible approaches may be the use of a shorter and less flexible spacer and/or its covalent linkage to the primary rim of CDs at more than one point. Regarding the (CMG)₄ spacer used in this work, obtained results indicate that this spacer is convenient for creating a static and stable scaffold consisting of a hydrophobic core and a thin shell decorated with functional CD residues, which will subsequently undergo further surface modifications by means of host–guest interactions to develop drug carriers with a dynamic “stealth” shell, targeting ligands and diagnostic labels.

Conclusion and perspectives

In present work, we have studied the surface structure and dynamics of α- and β-CD coated core–shell micelles self-assembled from function–spacer–lipid constructs that are conjugates of CDs, phospholipid DOPE, and a hydrophylic spacer (CMG). Of special interest were the molecular properties of individual CD residues such as localization, orientation, and mobility. It was found that CDs are distributed over the entire surface area and are mostly located on the surface. However, due to their large surface density, most CDs are found in clusters, which presumably adversely affects their ability to capture ligands. Moreover, CDs take various orientations, and only about 20% of them have their internal cavity exposed to the outside. Analysis of the dynamics of the surface structure of micelles showed that only a small part of the molecules retains mobility, which can also negatively affect their functionality. With a 5-fold decrease in the surface density of CDs and the spacer, clusters of CD residues no longer formed; however, the distribution of CDs by orientation and their mobility remained without significant changes. Obtained results suggest that CD orientation and mobility depend on a combination of the spacer, CDs, and their junctions, and that specific design efforts must be made to ensure CDs’ cavity exposure outside. These findings might be useful in further development of CD-decorated NPs towards rational design of efficient drug delivery systems. Although it was noted above that the close mutual arrangement of CD residues is a negative factor, in fact it can be useful, since many molecules that do not fit entirely into the CD cavity (modern antibiotics and other drug molecules are usually not so small) tend to form a complex with two host CD residues simultaneously; the discovery of this phenomenon stimulated the synthesis of cyclodextrin dimers to increase the affinity constant.⁴¹ This advantageous mutual arrangement is realized in our micelle-like NPs without covalent linking of two CD residues, which implies the same entropy gain as with covalent cross-linking. That is, the basic NPs, which were mostly the subject of consideration in this work, have a good prospect of use as such, without “dilution” with lipid L (when the F and S regions are absent). We also note that the “dilution” can be carried out not only with a lipid (as shown in the article) but also with an SL molecule, i.e., tuning can be carried out not only with respect to the density of CD residues but also the charge (CMG is charged negatively) of their microenvironment.

In conclusion, we draw attention to another promising feature of the NPs described here. They have two zones for loading hydrophobic drugs. The first is the above-discussed cavities of the CD residues. The second is the micelle core, which is open for loading when forming a NP through self-association. The release of CD-bound molecules will occur easily and quickly, while the release of core-bound molecules occurs only in the lysosomes of the cell. This opens up the possibility of obtaining NPs with two different ligands, or with one, releasing in a two-stage manner.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

All software programs (free to use) mentioned in the article are the official release versions without any modification. Additional data files can be shared on reasonable request.

Acknowledgements

This work was carried out with the financial support from the Russian Science Foundation, grant no. 23-24-10046. The calculations were carried out using a high-performance computing cluster of P. N. Lebedev Physical Institute of the Russian Academy of Sciences. SAXS and TEM measurements were supported by the Ministry of Science and Higher Education of the Russian Federation (agreement 075-10-2021-093).

References

1. K. Jain, Methods Mol. Biol., 2020, 2059, 1–54 CrossRef CAS PubMed .
2. S. Wilhelm, A. Tavares and Q. Dai, et al., Nat. Rev. Mater., 2016, 1, 16014 CrossRef CAS .
3. A. Salvati, Curr. Opin. Biotechnol, 2024, 87, 103101 CrossRef CAS PubMed .
4. D. Klepac, et al., Nanoscale, 2018, 10(13), 6194–6204 RSC .
5. T. T. Hoang Thi, E. H. Pilkington, D. H. Nguyen, J. S. Lee, K. D. Park and N. P. Truong, Polymers, 2020, 12(2), 298 CrossRef PubMed .
6. S. Mura, J. Nicolas and P. Couvreur, Nat. Mater., 2013, 12(11), 991–1003 CrossRef CAS PubMed .
7. M. Debayle, et al., Biomaterials, 2019, 219, 119357 CrossRef CAS PubMed .
8. M. Schroffenegger, et al., ACS Nano, 2020, 14(10), 12708–12718 CrossRef CAS PubMed .
9. I. Alberg, et al., Small, 2020, 16(18), 1907574 CrossRef CAS .
10. M. T. Manzari, et al., Nat. Rev. Mater., 2021, 6(4), 351–370 CrossRef CAS PubMed .
11. S. S. Braga, Biomolecules, 2019, 9(12), 801 CrossRef CAS PubMed .
12. S. B. Carneiro, et al., Int. J. Mol. Sci., 2019, 20(3), 642 CrossRef CAS PubMed .
13. F. Adeli, F. Abbasi, M. Babazadeh and S. Davaran, J. Nanobiotechnol., 2022, 20, 91 CrossRef CAS PubMed .
14. M. Zhang, Y. Wen, Z. Zhang, X. Song, J. Zhu and C. Thedrattanawong, et al., Nano Today, 2024, 59, 102511 CrossRef CAS .
15. A. Rahmani, F. Rahimi, M. Iranshahi, H. Kahroba, A. Zarebkohan and M. Talebi, et al., Sci. Rep., 2021, 11, 21425 CrossRef CAS PubMed .
16. I. D. Zlotnikov, D. A. Streltsov, N. G. Belogurova and E. V. Kudryashova, Life, 2023, 13, 446 CrossRef CAS PubMed .
17. M. Zhao, W. Jiang, X. Xie, Y. Jaiswal, L. Williams and M. Wei, et al., Polymers, 2022, 14, 2482 CrossRef CAS PubMed .
18. X. Li, J. Liu and N. Qiu, Polymers, 2023, 15, 1400 CrossRef CAS PubMed .
19. P. Batra and P. Shende, Colloids Surf., A, 2024, 688, 133521 CrossRef CAS .
20. K. Ramesh, C. K. Balavigneswaran, S. A. Siboro, V. Muthuvijayan and K. T. Lim, Polymer, 2021, 214, 123243 CrossRef CAS .
21. H. Ameli and N. Alizadeh, RSC Adv., 2022, 12, 4681–4691 RSC .
22. S. P. Chali, S. Azhdari, A. Galstyan, A. H. Gröschel and B. J. Ravoo, Chem. Commun., 2021, 57, 9446–9449 RSC .
23. S. S. Soni, et al., ACS Appl. Bio Mater., 2024, 7(8), 456–4866 Search PubMed .
24. Y. Zhao, et al., J. Mater. Chem. B, 2023, 12, 39–63 RSC .
25. G. Utzeri, et al., Front. Chem., 2022, 10, 859406 CrossRef CAS PubMed .
26. G. Rivero-Barbarroja, et al., Nanomaterials, 2020, 10, 2517 CrossRef CAS PubMed .
27. J. Łagiewka, T. Girek and W. Ciesielski, Polymers, 2021, 13, 1759 CrossRef PubMed .
28. D. A. Blake, N. V. Bovin, D. Bess and S. M. Henry, J. Visualized Exp., 2011, 54, 3289 Search PubMed .
29. K. Manalastas-Cantos, et al., J. Appl. Crystallogr., 2021, 54, 343–355 CrossRef CAS PubMed .
30. M. J. Abraham, et al., SoftwareX, 2015, 1, 19–25 CrossRef .
31. M. Stroet, et al., J. Chem. Theory Comput., 2018, 14, 5834–5845 CrossRef CAS PubMed .
32. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graph., 1996, 14, 33–38 CrossRef CAS PubMed .
33. B. Hess, H. Bekker, H. Berendsen and J. Fraaije, J. Comput. Chem., 1997, 18, 1463 CrossRef CAS .
34. U. Essmann, L. Perera, M. L. Berkowitz, T. Darden, H. Lee and L. G. Pedersen, J. Chem. Phys., 1995, 103(19), 8577–8593 CrossRef CAS .
35. G. Bussi, D. Donadio and M. Parrinello, J. Chem. Phys., 2007, 126, 014101 CrossRef PubMed .
36. M. Bernetti and G. Bussi, J. Chem. Phys., 2020, 153, 114107 CrossRef CAS PubMed .
37. T. C. Beutler, A. E. Mark, R. C. van Schaik, P. R. Greber and W. F. van Gunsteren, Chem. Phys. Lett., 1994, 222, 529–539 CrossRef CAS .
38. D. I. Svergun, C. Barberato and M. H. J. Koch, J. Appl. Crystallogr., 1995, 28, 768–773 CrossRef CAS .
39. I. Vaskan, et al., Biomater. Sci., 2024, 12, 798–806 RSC .
40. F. Eisenhaber, et al., J. Comput. Chem., 1995, 16, 273–284 CrossRef CAS .
41. A. M. Anderson, T. Kirtadze, M. Malanga, D. Dinh, C. Barnes, A. Campo, D. M. Clemens, R. Garcia-Fandiño, Á. Piñeiro and M. S. O’Connor, Int. J. Pharm., 2021, 606, 120522 CrossRef CAS PubMed .

This journal is © the Owner Societies 2025