Impact of crowder size, hydrophobicity, and hydration on the structure of amyloid-β oligomers

Abstract

It is being increasingly recognized that a comprehensive understanding of protein folding and aggregation requires accounting for the crowded in vivo milieu. Such a complex milieu offers a variety of soft, non-specific interactions along with the crowder volume exclusion effects that can modulate the hydration and protein aggregation processes. A clear understanding of the interplay of these effects is still lacking. Oligomerization of intrinsically disordered proteins (IDPs) forms the early stage nucleation step for fibrillation and this study investigates the structural stability of the dimer and tetramer of Aβ(16–22) IDP in the presence of molecular crowders. Molecular dynamics simulations are employed to examine the role of ethylene glycol (EG), diethylene glycol (DEG) and modified DEG (UCON) with increased hydrophobicity at crowded concentrations on the structural stability of peptide oligomers. The results show that EG destabilizes both the peptide oligomers at low and high packing fractions by enhancing the hydration of peptides at low concentration and by increasing peptide–crowder interactions at high concentration. UCON stabilizes the oligomers at low concentration by reducing peptide hydration, enhancing the peptide inter-strand interactions leading to energetic effects. Conversely, it stabilizes the oligomer structure at high packing fractions via entropic volume exclusion effects, enhancing the peptide hydration due to confinement of water around the peptide. Water molecules are confined in small volume and are observed to be disordered with anomalously slow diffusion. The results provide insights into the interplay of molecular crowders size effects on peptide hydration, regulating the oligomer structure. The findings have implications in understanding the role of crowding in shaping the free energy landscapes of IDPs.

---

1. Introduction

One of the unresolved questions in biophysics has been to understand the principles that govern protein folding, misfolding and aggregation in the living cell environment.^1–4 Soluble proteins with intrinsically disordered regions tend to undergo aggregation to form thread-like elongated fibrils and this process is termed as amyloidogenesis.^5–8 These amyloid fibrils are composed of proteins stacked in cross β-sheet conformations making them thermodynamically very stable. This process is associated with the onset of several pathological conditions such as Alzheimer's and Parkinson's neurodegenerative diseases and other disorders such as type-II diabetes.^9,10 The molecular mechanisms underlying protein aggregation are still inconclusive. Several studies have focused on investigating the effects of cosolvents, solution pH and temperature on the formation of amyloid aggregates. However, the interactions and effects of the biological fluid encountered by the self-assembling proteins remain poorly understood and have not yet been fully accounted for. The intracellular and extracellular environments are known to be heterogenous, dense spaces that consist of tightly packed large macromolecules, ions, small organic molecules or cosolutes. The decrease in the accessible volume to the biomolecules due to the crowding effects can alter the protein folding equilibria, protein dynamics, protein–protein interactions and aggregation.^11–13 These crowding effects have been widely recognized to induce compaction or folding in biomolecules as a result of volume exclusion effects exerted by the large macromolecular crowders. Such a promotion of biomolecular compaction has been shown to have an entropic origin arising due to the depletion of the crowders from the solvation shell of the test biomolecule to reduce the excluded volume of the crowders, assuming the crowder-biomolecule interactions to be entirely repulsive.^14–16 However, recent investigations have indicated that a description based entirely on steric repulsions may not be sufficient. These studies underscored the significance of accounting for the multiple soft, specific and non-specific intermolecular attractive interactions in these complex environments, leading to an emerging view of crowding where the effects have been attributed to enthalpic interactions as well.^17–26

The process of amyloid formation has three stages: a lag phase which involves formation of the critical nucleus by the protein monomers and protofibril oligomeric species, and an exponential growth phase which involves addition of monomers to the critical nucleus followed by the maturation phase where the mature fibrils are formed.²⁷ A few studies have investigated the effects of crowded environments on the amyloid formation.^11,28,29 Although crowding is traditionally believed to promote self-assembly in biomolecules, a few studies have indicated counterintuitive effects. Shtilerman et al. investigated the role of molecular crowders on fibrillation of α-synuclein using polyethylene glycol (PEG), dextran T-70 and Ficoll.³⁰ Using thioflavin T (ThioT) fluorescence spectroscopy experiments, they found that the crowding induced by 10% PEG 20 000 strikingly reduced the lag phase of aggregation, resulting in significant acceleration of protofibril formation. The lifetime of protofibrils was found to be reduced as the fibrils consumed these transient intermediates, leading to acceleration of protofibril-to-fibril conversion rate. This was majorly dependent on the molecular weight of the crowders rather than the chemical nature as all the crowders showed a similar decrease in the lag phase time. Uversky and coworkers also showed that the effects exerted by different crowders such as polymers, polysaccharides and proteins can result in acceleration of the aggregation rate of α-synuclein.³¹ Among the proteins as crowders, the presence of bovine serum albumin (BSA) protein as a crowder was found to accelerate the aggregation rate of α-synuclein compared to that in the presence of lysozyme as a crowder. They accorded this difference to the higher excluded volume effect exerted by BSA due to its larger size. Fung et al. reported enhancement of nucleation of both Aβ-40 and Aβ-42 in the presence of glucose as crowders.³² Whereas, in the presence of galactose and mannose as crowders, mature fibril formation was observed. On the contrary, in the presence of fructose at crowded concentrations, which is capable of forming favourable H-bonding interactions with the protein, the fibril formation was found to be reduced. A crowded environment is also known to affect the protein chain dynamics and translational diffusion. In a study by Heravi et al., they examined the effects of both synthetic (Ficoll70) and biological crowders (bacterial cell lysate) on the fibrillation of α-synuclein.³³ HSQC (heteronuclear single quantum coherence) NMR and fluorescence experiments indicated that the protein formed aggregates only in the cell lysate. However, the diffusion of the protein was faster in cell lysate as compared to that in Ficoll70. The rates of diffusion were found to be different for the hydrophobic residues, implying the role of soft hydrophobic interactions. Latshaw et al. examined the effects of interacting and non-interacting crowders on the fibrillation and oligomerisation of Aβ(16–22) peptides.³⁴ It was found that larger packing fractions of the smaller crowders enhanced fibrillation. The hydrophobic crowders reduced the rate of formation of fibrils, and with increasing strength of crowder interactions, the crowder–peptide interactions dominated leading to inhibition of formation of oligomers and fibrils. Therefore, these studies indicate that the chemical nature of the crowders and their interactions can play a crucial role in determining the rate of formation of IDP fibrils. Soranno et al. performed FRET analysis for examining the crowding effects of PEG of different molecular weights on aggregation of four different IDP sequences.³⁵ It was found that increasing the volume fraction as well as crowder size led to IDP compaction, highlighting the role of volume exclusion effects of PEG. Using theoretical treatments of scaled particle theory (SPT) and Flory–Huggins theory, they classified two regimes of IDP collapse. First, with shorter PEG chains that do not overlap with each other but are able to penetrate into unfolded IDP chains. Second, with longer PEG chains where the mean-field theory failed to explain the experimental observations of IDP radius of gyration reaching a constant value after a threshold PEG chain length was attained. This study underscored that the hard-sphere treatment of macromolecular crowders was insufficient to comprehensively explain the IDP compaction and the chemical interactions between crowder–crowder and crowder–protein need to be accounted for.

The complex macromolecular interactions in the crowded environment also result in altering the properties of water that may be distinct from those compared to the dilute solution.³⁶ Such altered water properties can play a crucial role in determining the biomolecular structure and function.^37–42 For instance, the decrease in water activity due to high concentrations of molecular and macromolecular crowders was found to determine the stability of biomolecules such as RNA.⁴³ A recent study correlated the modulation in water dynamics associated with human serum albumin (HSA) protein in the presence of crowded solutions of dextran-40, Ficoll-70, and PEG-35, with the stability of the protein using time-dependent fluorescence Stokes shift (TDFSS) experiments.^40,44 The water dynamics were found to be slower in the presence of dextran-40 and Ficoll-70, but were found to be faster around PEG-35. Slower water dynamics suggested rigid associated water molecules around the protein that caused entropic stabilization and enthalpic destabilization of the protein. Whereas, faster water dynamics suggested flexible water molecules causing entropic destabilization and enthalpic stabilization of the protein. The same group also explored the role of associated water dynamics in influencing the Bromelain protein activity and stability in the presence of Ficoll-70 as crowders.⁴⁵ Flexible waters at lower crowder concentration were found to increase the enzyme-substrate complex formation. The decrease in the microviscosity of water around the protein was found to control the product formation. These studies underscore the critical role of crowding-induced changes in water structure and dynamics in determining the structure of the biomolecules.

In this work, we investigate the role of molecular crowders in affecting the structure of oligomers (dimer and tetramer) of Aβ-42(16–22) peptides using molecular dynamics simulations. The crowders used are ethylene glycol (EG), diethylene glycol (DEG) and modified DEG (called UCON), which is a monomer of the 1 : 1 copolymer of ethylene glycol and propylene glycol. One of the most widely used macromolecular crowders for mimicking crowding in in vitro and in silico studies has been polyethylene glycol (PEG) since it can act as an inert and biocompatible polymeric crowder with large molecular weight inducing the volume exclusion effects in solution.^31,46,47 In order to understand, how such a polymeric crowder would exert effects on amyloid fibrillation, it is essential to understand how its monomer would affect such a process. Therefore, in this work we selected ethylene glycol, its dimer and its variant to examine molecular crowding effects on the peptide oligomer formation. In our work, the choice of ethylene glycol and its variants as molecular crowders has been to examine the role of the hydrophobic and hydrophilic interactions that the crowders can have with the protein that would modulate the oligomerization process. Breydo et al. showed that a slight change in the hydrophobicity of PEG to UCON led to differences in propensity for the protein to form amyloid oligomers or mature fibrils.⁴⁷ PEG was found to accelerate mature fibril formation, whereas UCON was found to decrease it by stabilizing the oligomers of amyloid. They also showed that the solvent accessibility of aromatic residues was enhanced in the presence of UCON but not PEG, indicating the crowding-induced changes in hydration water. In order to explore this further, we examine the interplay of the crowder chemistry, size and hydrophobicity on the structural stability of the dimer and tetramer of the Aβ-42(16–22) peptides, with a view to understand the effects of crowding-induced changes in hydration structure and dynamics on peptide oligomer conformational stability. The results indicate the differential effects of these molecular crowders on oligomer stability that are sensitive to the packing fractions of the crowders. Meanwhile, at low crowder packing fractions, the effects are linear in stabilizing the oligomer structure as a function of the crowder size and are energetic in origin. At high packing fractions, the effects are crowder-size dependent and linearly increase the structural stability of peptides with increasing crowder size. The effects of these crowders on hydration water structure and dynamics are highlighted.

2. Computational details

2.1. System setup

The system comprised aqueous solutions of dimer and tetramer of heptamer (16–22 residues) of amyloid forming Aβ(1–42) protein. Crowded solutions of ethylene glycol (EG), diethylene glycol (DEG) and modified DEG (UCON) were prepared in which the peptide oligomers were solvated. The dimer and tetramer were prepared by extracting the seven-residue heptamer of Aβ(1–42) (K–L–V–F–F–A–E) from RCSB PDB file 5OQV.⁴⁸ These residues (16–22) constitute the hydrophobic core of full protein (amyloid-β(1–42)) and have been found to play a crucial role in the formation of the mature fibrils of the protein.⁴⁹ The heptamer fragment was capped with an acetyl group and NH₂ group on the N-terminus and the C-terminus, respectively. The dimer and tetramer of this heptamer were prepared. The crowders utilized in this study are ethylene glycol (EG), diethylene glycol (DEG), and a more hydrophobic crowder denoted as UCON in which one H atom on one of the terminal CH₂ groups in DEG is replaced by a CH₃ group as shown in Fig. 1. UCON is a monomer of 1 : 1 copolymer of ethylene glycol and propylene glycol, as shown in Fig. 1. The EG, DEG, UCON molecules and the peptides were modelled using the CHARMM36m force-field.⁵⁰ Water was modelled using TIP3P potential. These crowders were packed in the solvated dimer and tetramer solutions at different packing fractions in a cubic box of 5 nm as described in Table 1 and as shown in Fig. 1.

Fig. 1 (a) Heptamer fragment of Aβ42(16–22), [lysine (K), leucine (L), valine (V), phenylalanine (F), phenylalanine (F), alanine (A), glutamic acid (E)] along with crowders employed in this study. (b) The structures of ethylene glycol (EG), diethylene glycol (DEG) and modified diethylene glycol (UCON). The box represents anadditional methyl CH₃ group in UCON. (c) and (d) Show the snapshots of the simulation boxes with the peptide dimer and tetramer in the presence of EG crowders.

Table 1 Details of the composition of different crowder solutions. The number of crowder and water molecules is denoted by N_c and N_w, respectively. The packing fractions of the crowders are denoted by ϕ_c. For each system, five independent simulations of 100 ns each were performed (Sim 2–Sim 6). Additionally, Sim 1 corresponds to a longer trajectory: 250 ns for the dimer system and 500 ns for the tetramer system

S. no.	Crowder system	N c	N w (Sim 1) (250 ns/500 ns)	N w (Sim 2) (100 ns)	N w (Sim 3) (100 ns)	N w (Sim 4) (100 ns)	N w (Sim 5) (100 ns)	N w (Sim 6) (100 ns)	Packing fraction ϕc (%)
1	DIMER only	0	3975	3970	3970	3970	3970	3970	0
2	EG + DIMER	255	2972	2984	2950	3000	2997	2965	16.6
3	DEG + DIMER	150	3021	3009	2996	3034	3020	3020	16.3
4	UCON + DIMER	150	2886	2857	2836	2848	2870	2875	19
5	EG + DIMER	710	1483	1433	1508	1485	1511	1485	45.6
6	DEG + DIMER	436	1447	1463	1447	1502	1440	1492	45.9
7	UCON + DIMER	353	1525	1566	1568	1594	1589	1568	44.5
8	TETRAMER only	0	3887	3895	3895	3895	3895	3895	0
9	EG + TETRAMER	255	2888	2935	2904	2901	2874	2900	16.6
10	DEG + TETRAMER	150	2945	2921	2913	2941	2935	2931	16.3
11	UCON + TETRAMER	150	2812	2795	2782	2800	2782	2799	19
12	EG + TETRAMER	710	1461	1432	1447	1398	1463	1420	45.6
13	DEG + TETRAMER	436	1438	1430	1438	1451	1435	1463	45.9
14	UCON + TETRAMER	353	1515	1576	1539	1557	1538	1562	44.5

---

2.2. Simulation details

All-atom molecular dynamics (MD) simulations have been performed using the GROMACS 2021.4 MD package.⁵¹ To obtain a configuration with minimum potential energy, the steepest descent energy minimization method was employed. The peptide strands were not positionally restrained, they were kept free to diffuse in solution. NVT equilibration followed by NPT equilibration for 1 ns with 2 fs time step at 300 K was performed. To ensure a better statistical average of each observable, we performed six independent simulations out of which the production runs for five of them was for 100 ns each, along with the sixth longer simulation of 250 ns for the dimeric systems and 500 ns for the tetrameric systems. In all six simulations, the system was independently packed initially, while maintaining a consistent crowder packing fraction, as listed in Table 1. This resulted in a total simulation time of 750 ns for each dimeric system and 1 μs for each tetrameric system. The production runs were performed in the NPT ensemble at 300 K and 1.0 atm, respectively. During equilibration and production runs, the temperature and pressure were maintained using the velocity-rescaling thermostat and Parrinello–Rahman barostat with time constants of 0.1 ps and 2 ps, respectively. The equations of motion were integrated using a leap-frog integrator with a time step of 2 fs. A van der Waals cutoff radius of 1.0 nm was used. The electrostatic interactions were computed using the particle mesh Ewald (PME) method with the real space cutoff of 1.0 nm and grid spacing of 0.16 nm. Bond constraints were applied on bonds involving hydrogen atoms using the LINCS algorithm.

2.3. Observables

To characterize the structural changes in the peptide dimer and tetramer and their hydration due to the effects of the molecular crowders, the following observables were computed.

Average number of hydrogen bonds between the peptide backbone of the adjacent chains was computed using GROMACS command gmx hbond. Hydrogen bonds are considered to be formed when the distance between the donor and acceptor is less than or equal to 3.5 Å, and the hydrogen-donor–acceptor angle is less than 30°. Similarly, the average number of hydrogen bonds between the peptide and water and those between peptide and crowders were also computed using the same definition.

Secondary structure: Ramachandran plots were investigated to quantify the extent of β-sheet formation in the peptide oligomers. The backbone dihedral angles ϕ and ψ were computed for each residue using VMD.⁵² These angles were then used to construct a Ramachandran plot, and the joint probability distribution P(ϕ, ψ) was obtained using kernel density estimation (KDE) in Python.

Preferential binding coefficient denoted as Γ_pc, quantifies the degree of crowder molecules binding or adsorbing on the peptide surface relative to that in the bulk.⁵³Γ_pc can be defined as⁵³

where N_w and N_c are total number of water and crowder molecules, respectively. n_w and n_c are the number of water and crowder molecules at proximal distance r from the peptide chain. This distance is determined as the shortest distance from the center of mass of the crowder or water and the protein surface. A positive value of Γ_pc indicates that the crowders are preferentially adsorbed onto the peptide surface, while a negative value suggests that the crowders are preferentially excluded from the protein surface.

Tetrahedral order parameter quantifies the local ordering of hydration water. We calculated the tetrahedral order parameter (q_tet) or orientational order parameter using,^54,55

where ψ_jk is defined as the angle formed between the central oxygen i with the neighboring oxygens j and k. A perfect tetrahedral arrangement would correspond to a q_tet value of 1. We extended the above definition by including heavy atoms of the protein chain (C, O, N) along with oxygen of water molecules in the first solvation shell (up to 0.5 nm from the protein).^56,57

Mean square displacement (MSD) was calculated to study the diffusion properties of hydration water in the crowded environment. The MSD is computed over time, which is defined as,

where r(t) is the position of a water molecule at time t, and r(0) is the initial position.

Computation of errors: the error in observables is computed by using the block-averaging method where a trajectory is divided into blocks of equal sizes and then the average and standard deviation are computed from those blocks. For computing the error in the ratio of number of water molecules in the peptide hydration shell in crowded and no crowder solutions , block averaging was used to compute the error in N_w and in N₀ and then the propagation of error was used to determine the error in their ratio. A block size of 5 ns was used, as it captures fluctuations over relevant timescales while reducing short-time correlations.

Results

Crowder effects on the structure of peptide oligomers

The stability of the β-sheet structure of the peptide dimer and tetramer is first quantified by examining the Ramachandran plots for the oligomers at low and high crowder packing fractions in Fig. 2 and 3, respectively. Fig. 2(a) and (e) show that the dimer loses the β-sheet character in pure water, whereas the tetramer retains certain probability of having β-sheet character. In the case of the dimer, addition of EG or DEG to the aqueous solution does not result in any β-sheet character as shown in parts (b) and (c). Whereas, addition of UCON results in an increase in the β-sheet character as shown in part (d). The residues show the probability of having the polyproline-II structure around (70°, 140°) in both the oligomers in all aqueous solutions, also seen in a previous study.⁵⁸ In the case of the tetramer, the residues exhibit β-sheet character in pure water (as shown in part (e)). Addition of EG to the solution decreases the probability of the β-sheet character. Whereas, addition of DEG and UCON is seen to enhance the probability of having β-sheet character as shown in Fig. 2(f)–(h). Therefore, at the low crowder packing fraction, UCON is observed to enhance the β-sheet character of the dimer as well as the tetramer as compared to that in the other two crowder solutions.

Fig. 2 Ramachandran plots for the dimer (upper row) and tetramer (lower row) of the peptide in (a) and (e) no crowder solution, (b) and (f) ethylene glycol solution, (c) and (g) diethylene glycol solution and (d) and (h) UCON solution at low packing fraction. The color bars on the top of the plots indicate low packing fraction of crowders (ranging between 16–19% as indicated in Table 1). The red square indicates the β-sheet region in the plot.

Fig. 3 Ramachandran plots for the dimer (upper row) and tetramer (lower row) of the peptide in (a) and (e) no crowder solution, (b) and (f) ethylene glycol solution, (c) and (g) diethylene glycol solution and (d) and (h) UCON solution at high packing fraction. The color bars on the top of the plots indicate high packing fraction of crowders (ranging between 44–45% as indicated in Table 1). The red square indicates the β-sheet region in the plot.

Fig. 3 shows the Ramachandran plots of the dimer and the tetramer in aqueous solutions of the crowders at high packing fractions. Addition of EG to the aqueous solution of the peptide dimer does not retain the β-sheet character of the dimer as shown in part (b). Whereas, DEG and UCON are observed to enhance the β-sheet character of the dimer (parts (c) and (d)). For the peptide tetramer, the residues do not exhibit β-sheet character in EG solution, but exhibit an increasing probability of the same in DEG and UCON solutions, with the highest being in the UCON solution (see parts (f)–(h)). Therefore, even at high crowder packing fractions, UCON is seen to enhance the β-sheet character of the oligomers.

To further explore the effect of crowder size on the stability of the β-sheet structure of the oligomers, we compute the average number of hydrogen bonds (〈NHB_pp〉) formed between the backbone of the protein chains or strands. Fig. 4(a) shows the 〈NHB_pp〉 formed between the protein strands in the dimer in different crowder solutions. In pure water, the β-sheet structure of the dimer is not maintained, as can be seen by the low value of 〈NHB_pp〉 ≈ 2 ± 0.31. At low crowder packing fractions, an increasing dependence of the 〈NHB_pp〉 is seen on the crowder size. The 〈NHB_pp〉 follows the order: EG ≈ DEG < UCON. This indicates that by adding a CH₃ group in DEG to prepare UCON, the secondary structure stability of the oligomer increases. At higher crowder packing fractions, the overall trends in 〈NHB_pp〉 follow the same order as seen in the low packing fraction of the crowders. Fig. 4(b) shows that the 〈NHB_pp〉 = 7 ± 0.43, is quite low in the no crowder solution as compared to that seen in the native tetramer structure. At low crowder packing fractions, the 〈NHB_pp〉 shows an increase with increasing crowder size, similar to that seen for the dimer with 〈NHB_pp〉 changing from 7.5 ± 0.46 to 7.8 ± 0.46 to 10 ± 0.54 for EG to DEG to UCON. However, at the higher crowder packing fractions, there is an increase in 〈NHB_pp〉 observed with increasing crowder size, implying that the larger sized and more hydrophobic UCON stabilizes the tetramer (〈NHB_pp〉 ≈ 10 ± 0.47) to a higher extent at the high crowder packing fractions. The other structural properties of the oligomers, such as radius of gyration and solvent accessible surface area, are summarized in Fig. S1–S4 and S9, S10 (ESI†), along with snapshots of the systems in Fig. S5–S8 in the ESI.†

Fig. 4 Average number of mainchain–mainchain hydrogen bonds between the peptide strands for (a) dimer and (b) tetramer systems. Light and dark shades represent low (16–19%) and high (45–46%) packing fraction of crowders, respectively, for the oligomers. The grey bar in the “no-crowder” solution denotes the number of hydrogen bonds for the native structure of the dimer and tetramer taken from PDB ID 5OQV.

The results indicate three interesting insights. First, the sensitivity of the secondary structure of the peptide oligomers on the crowder size at different crowder packing fractions. Second, the smallest sized (EG) is seen to reduce the β-sheet character of both the peptide oligomers, whereas UCON is seen to have higher stabilizing effect on both the peptide oligomers. Third, at high crowder packing fractions, the large sized crowders, DEG and UCON, are found to stabilize the peptide tetramer to a higher extent.

To gain further molecular insights into these interesting trends, we compute the preferential binding coefficients (Γ_pc) of the crowders on the peptide oligomers as shown in Fig. 5. The Γ_pc > 0 for all the crowders at low packing fractions for the dimer and tetramer is shown in Fig. 5(a) and (b) respectively. This indicates that all crowders preferentially bind on the dimer and tetramer surface at low packing fractions. UCON is seen to adsorb the most on the dimer and tetramer surface, whereas EG and DEG adsorb to similar extents. The enhanced secondary structure stability of the dimer and tetramer in UCON as seen in Fig. 4 is due to UCON having favourable interactions with the peptide. UCON has the highest hydrophobicity amongst the three crowders. The hydrophobic groups of UCON interact with the peptide surface via van der Waals interactions, while at the same time the OH groups can interact with water. However, due to its large size, it displaces more waters from the peptide solvation shell, leading to reduced peptide–water interactions and enhancing the peptide–peptide interactions in the oligomers. On the other hand, DEG being more hydrophilic than UCON, can interact favourably with the peptide that enhances peptide–crowder energetic interactions as well as has more favourable peptide–water interactions than in UCON solution, leading to destabilizing to the peptide–peptide H-bonding. These observations are explained in more detail in the next section.

Fig. 5 Preferential binding coefficients (Γ_pc) of crowders on (a) dimer and (b) tetramer systems at different crowder packing fractions. Solid lines represent low packing fraction and dashed lines represent high packing fraction.

Interestingly, Γ_pc < 0 for all the crowders for the dimer and tetramer in solutions with higher crowder packing fractions. This implies that the crowders are preferentially depleted from the dimer surface at higher packing fractions. UCON is observed to be most depleted from the dimer surface followed by DEG and EG, which is found to be least depleted. Similar trends are observed in the case of the tetramer, as shown in Fig. 5(b), where the crowders are preferentially adsorbed on the tetramer at low packing fractions but are depleted from the tetramer surface at high packing fractions. EG is least depleted from the tetramer surface and UCON and DEG are depleted similarly to a higher extent. This indicates that the volume exclusion effects of the crowders become prominent at higher crowder packing fractions that lead to depletion of crowders from the peptide surface, enhancing the peptide–peptide interactions and increasing the oligomer stability. The depletion effect of crowders is more prominent for stabilizing the tetramer to a higher extent than the dimer at high crowder packing fractions, as seen in Fig. 4.

Crowder effects on structure and dynamics of hydration water

Further insights into the structure of hydration water are explored to understand the molecular mechanisms underlying the differential effects of crowders. Fig. 6(a) shows the average number of water molecules in the first hydration shell of the oligomers in crowded solutions relative to that in no crowder solution. A decreasing trend in the ratio is observed as the crowder size increases for EG, DEG and UCON for both the dimer and tetramer at low crowder packing fraction. The ratio is seen to decrease from 0.85 in EG solution to 0.82 in DEG solution and then to 0.74 in UCON solution around the dimer. Similarly, around the tetramer, the number of hydration waters is seen to decrease in a similar way as observed for the dimer. The ratio is lower around the tetramer than around the dimer in all crowder solutions at low crowder packing fraction, indicating higher loss of water molecules around the tetramer on addition of crowders to aqueous solution. At higher crowder packing fraction, the hydration water density around both the oligomers is seen to decrease in all the crowder solutions. A slight increase in the ratio is observed with increasing crowder size, i.e. from 0.5 in EG and DEG solution to 0.55 in UCON solution around both the oligomers. At low crowder concentration, the trends are opposite and mirror those observed for variation in the number of peptide–peptide H-bonds and β-sheet structure of the peptides in Fig. 4. This indicates the role of hydration water in determining the peptide oligomer structure stability. The increasing size of crowders from EG to DEG to UCON is seen to stabilize the peptide secondary structure by decreasing the hydration of the peptides. This is supported by the slight decrease (not a significant decrease) in the number of H-bonds between peptide and water (〈NHB_pw〉) for both the oligomers when comparing EG to DEG to UCON at low crowder packing fraction, as shown in Fig. 6(b). Consequently, a decrease in the number of peptide–crowder H-bonds (〈NHB_pc〉) is also observed at low crowder packing fraction as shown in Fig. 6(c). This may be due to the increasing hydrophobicity of the crowders from EG to DEG to UCON. Interestingly, at high crowder packing fraction, the oligomers are stabilized by the increasing hydration of the peptides as reflected in the increasing 〈NHB_pw〉 as shown in Fig. 6(b). A subsequent decrease in 〈NHB_pc〉 is also seen for both the oligomers with increasing crowder size. Crowders are found to be preferentially excluded from the peptide surface at high crowder packing fractions as seen in Fig. 5, due to enhanced volume exclusion effects. This also results in decreasing 〈NHB_pc〉 with increasing crowder size. The increase in peptide–water H-bonds could potentially lead to disruption of the peptide–peptide H-bonding interactions and enthalpic destabilization of the peptide structure. But the water molecules are confined around the peptide and can be considered as rigid waters due to high crowder packing fractions, which entropically stabilizes the peptide structure.

Fig. 6 (a) Ratio of average number of water molecules around the oligomers in the first solvation shell in crowded and in no crowder solution. (b) Average number of peptide–water hydrogen bonds and (c) average number of peptide–crowder hydrogen bonds formed by the dimer and tetramer in different crowder solutions. The error bars in part (a) are within the point symbols. The light and dark shaded regions indicate the low and high crowder packing fractions, respectively.

The ordering of water molecules in the hydration shells was examined next, in terms of the tetrahedral order parameter q_tet. Fig. 7(a) and (b) show the probability distribution of the tetrahedral order parameter for a subset of water molecules, which are present within a distance of 0.5 nm from the protein chains and heavy atoms (C, O, N) of the protein chains. The q_tet of hydration waters is found to be low around both the dimer and tetramer, with a peak around 0.49 in P(q_tet) in no crowder solution, with a shoulder peak around q_tet = 0.4. These two peaks in P(q_tet) decrease on addition of the crowders at low packing fraction to the aqueous solutions of peptides and an additional peak is observed at negative q_tet values. At higher crowder packing fraction, the P(q_tet) has a peak at q_tet = 0.4 and negative q_tet values. The shift to lower q_tet values is more prominent in the case of the dimer than in the tetramer at high crowder packing fractions. This implies that the structuring of water molecules in the first hydration shell of both the peptides decreases on addition of the crowders. EG is seen to disrupt the water structure the most, followed by DEG and UCON for both the oligomers at high packing fractions. The q_tet decreases further due to higher confinement of water molecules in higher crowder concentrations.

Fig. 7 Probability distribution of the tetrahedral order parameter (q_tet) for (a) dimer and (b) tetramer systems with crowders at different packing fractions, calculated within the first solvation shell of the protein. Solid lines represent low packing fraction, while dashed lines represent high packing fraction.

The dynamics of water molecules is examined by computing the diffusivity of water molecules based on the mean squared displacement (msd) of water molecules with time. Fig. 8 shows that the variation of MSD of water molecules with introducing the crowders in solution. A linear log(MSD/time) curve with log(time) indicates normal diffusion, while a non-linear anomalous MSD is observed for t < 10 ps in each case, as expected in crowded solutions. At low packing fractions represented by solid lines, all crowders moderately affect the diffusion rates and local water structure and normal diffusion of water molecules is observed. However at high packing fractions, significant reduction in the msd of water molecules is observed with non-linear variation of msd with time. Such anomalous diffusion of water molecules is observed higher for UCON and DEG that induce slow movement of water molecules due to their larger size as compared to EG. This anomalous behaviour of water indicates that water molecules are confined due to the presence of crowders, particularly at higher concentration, which disrupts the ordered structure and hydrogen bonding network as observed in previous results also.

Fig. 8 Variation of log(MSD/time) with log(time) for (a) dimer and (b) tetramer systems with crowders at different packing fractions. Solid lines represent low packing fraction and the dashed line represents high packing fraction.

Discussion and conclusions

Using molecular dynamics simulations, we have examined the effects of molecular crowders on the structural stability of the dimer and tetramer of the heptamer of Aβ-42 protein. These crowders differ in size and hydrophobicity with EG being the smallest, DEG being the dimer of EG and UCON being the largest in size and most hydrophobic in nature amongst the three. The findings indicate that the stability of the structure of the dimer and tetramer is sensitive to the size and hydrophobicity of the crowder as well as their packing fractions, which determines the effects on the hydration of the peptides. EG is observed to be a destabilizing crowder for both the peptide oligomers at both low and high crowder concentrations. Being the smallest in size among all the crowders, it preferentially binds on the peptide surface, can form a higher number of peptide–crowder H-bonds, and can also simultaneously favourably interact with water molecules in the hydration shell, leading to denser hydration of the peptides. This results in disruption of the peptide–peptide H-bonds destabilizing the β-sheet structure of the oligomers. At the high packing fraction, although the peptide hydration density decreases as compared to that at low packing fraction, EG is capable of forming a high number of peptide–crowder H-bonds leading to destabilization of the inter-peptide interactions. Such effects of EG were also observed for disfavouring the aggregation or fibrillation of pseudo-isocyanine chloride (PIC) dyestuff via attractive, energetic interactions of EG with the dye molecules, affecting the hydration of these cationic dyes resulting in overall disfavoring of the self-assembly.^46,59 DEG is observed to either stabilize or destabilize the secondary structure of both the peptide oligomers depending on the concentration of DEG. At low packing fractions, DEG acts as a stabilizing crowder for the tetramer but acts as a destabilizing crowder for the dimer. At high packing fractions, DEG acts as a stabilizing crowder for both the oligomers. UCON, on the other hand, acts as stabilizing crowder for both the oligomers at both the crowder packing fractions. This implies that the more hydrophobic and sterically hindered crowder tends to stabilize the oligomer secondary structure to a higher extent. However, UCON affects the peptide hydration density intriguingly at low or high packing fractions. At low packing fractions, UCON is able to induce such an effect because it is more hydrophobic than all the other crowders allowing it to interact with the peptide chains via van der Waals interactions causing higher preferential adsorption on the peptide surface. However, it also induces larger steric effects that result in the displacement of water molecules from the peptide solvation shell, leading to stabilization of the inter-strand peptide interactions. Therefore, these crowder effects at low packing fractions seem to be energetic in origin. The water molecules at low crowder packing fractions could be considered as flexible waters that potentially disrupt or stabilize peptide–peptide interactions depending on the crowder, resulting in energetic stabilization of the peptide oligomers in UCON. Therefore, at low packing fraction, UCON can be considered as a stabilizing and desolvating crowder.

At high crowder packing fraction, the size effects of UCON crowders seem to dominate the stability of the oligomers. They exert volume exclusion effects leading to compaction of the peptide oligomers, resulting in depletion of the crowders and water molecules from the peptide hydration shells. Such an effect is entropic in nature as it releases or increases the configurational volume available for the solution components, since the crowders are depleted from the peptide surface relative to that in the bulk. Moreover, the water molecules in the peptide hydration shell at high crowder packing fraction can be considered to be rigid such that they can destabilize the oligomer structure enthalpically, but support the structure entropically. Therefore, UCON can be considered as a stabilizing and solvating crowder at high packing fraction. Table 2 summarizes the effects of the crowders on peptide hydration and consequent effects on peptide oligomer stability. Our results are in accordance with the experimental study by Uversky and coworkers, where they showed that UCON promoted the formation of oligomer intrinsically disordered proteins whereas PEG promoted formation of long amyloid fibrils.⁴⁷ Their other study also showed that the low and high concentrations of these crowders could induce different effects on the aggregation propensity of these proteins.⁶⁰ The classification of rigid and flexible hydration waters (although qualitative in this work), is in agreement with those described in the study by Sen and coworkers for the role of hydration waters in determining the stability of HSA protein,^40,44 and enzymatic activity of Bromelain protein.⁴⁵

Table 2 Summary of the impact of molecular crowders on the peptide dimer and tetramer stability

Crowder	Low ϕc	High ϕc
EG	Solvating and destabilizing	Desolvating and destabilizing
DEG	Desolvating and destabilizing	Solvating and stabilizing
UCON	Desolvating and stabilizing	Solvating and stabilizing

---

Moreover, the molecular crowders also affect the structure and dynamics of the hydration water around the peptides. The ordering of water molecules is found to be low in the peptide hydration shells in the presence of the crowders, irrespective of the type of crowders. At high packing fraction, the crowding molecules leave small volume for water molecules to occupy in the solvation shells of the peptide, confining them to that limited volume, leading to distortion of the tetrahedral order of the water molecules. This is also reflected in the slow diffusion of water molecules in these solutions. An anomalously slow diffusion of water molecules is observed at the short timescale of 10 ps exhibiting non-linear diffusion due to the hindrance and confinement caused by the high packing fraction of the crowders, thereby restricting the waters to a confined volume. The water diffusion is lowered in DEG and UCON solutions more than in EG solution. Such altering of the water structure and anomalously slow diffusion of water was also reported by Harada et al. for water around protein G/villin systems at crowded concentrations of these proteins.⁶¹ Our results reinforce the findings of an experimental study by Uversky and coworkers, where the crowding agents PEG and UCON were found to alter the solvent hydrogen bonding capacity. They highlighted the role of crowders in determining the dipolarity or polarizability of the solvent, and its hydrogen bond donor acidity or acceptor basicity, thereby indirectly affecting the aggregation propensity of the proteins.⁶⁰ In another recent study, the role of crowder sizes in jamming the solvent medium and its effect on polymer collapse equilibria was examined using a unified mean-field theory approach using implicit solvent.⁶² Based on the volume fraction of the model crowders, it was found that there is a limit on the size of crowders up to which they can induce polymer collapse in implicit solvent. They proposed a phase diagram of size ratios of the crowder and polymer with the crowder volume fraction. A region at high volume fraction of crowders was identified that prevented polymer collapse. It was corroborated to the jamming of the solvent by the crowders at such high volume fractions. The study supports our finding that the crowder sizes are crucial in determining the solvation of biomolecules that, in turn, can affect the biomolecular structure.

Therefore, our findings show that depending on the crowder packing fractions, the effects due to crowder size and hydrophobicity on the hydration shell of the peptides seem to be the controlling factors that can alter the stability of the peptide via peptide–water energetic interactions. Based on this, the crowders could be classified as solvating or desolvating crowders. At the high crowder packing fractions, the effect of the crowder size seems to play the dominant role in terms of the volume exclusion effects that these crowders exert entropically on the structural stability of the peptide oligomers. The crowder-concentration sensitive effects observed in our findings align with those of Latshaw et al. where they found that the crowder diameter as well as the crowder volume fraction impact the Aβ(16–22) peptide oligomerization.⁶³ They indicated that smaller crowders and larger volume fractions provide the largest enhancement effects on the rates of peptide oligomerization. Our findings provide atomistic-level insights into the impact of molecular crowders of varying chemistry on the structural organization of Aβ(16–22) oligomers, which constitute protofibrils and have been found to be possible pathological trigger agents in neurodegenerative diseases.² This Aβ(16–22) strand is proposed to be the central hydrophobic core of full Aβ 42 protein and is essential for the formation of full length Aβ amyloid aggregates.^64,65 Therefore, our findings have implications for the varying molecular crowding effects on the stability of this hydrophobic core. Our results are also in agreement with the previous studies where the role of water has been indicated to play a crucial role in determining the stability of these peptide oligomers.^41,58 In particular, our study provides atomistic insights into the influence of molecular crowders in altering the hydration shell structure and dynamics around these peptide oligomers, and effects on the stability of the hydrophobic core fragment of amyloid proteins, which is relatively less explored so far.^40,61 The results, therefore, have implications into how the interplay of the molecular crowders size, chemistry and hydrophobicity affect the hydration of peptides that can alter the free energy landscapes of the amyloid forming proteins and therefore, the fibrillation process. While the present study does not directly provide any insight into the free energy of self-association or binding of the peptides, it would be insightful to employ enhanced sampling techniques in future to quantify the crowding effects on the free energy of binding of the peptide oligomers.

Author contributions

Shivnandi contributed to the investigation, data curation, formal analysis and writing – review and editing. D. N. contributed to the conceptualization, methodology, funding acquisition, project administration, supervision and writing – original draft.

Conflicts of interest

The authors declare no competing conflicts of interest.

Data availability

The data analysis scripts and codes for the simulated systems can be found at: https://github.com/shivnandi/Abeta-oligomers-crowding.git.

Acknowledgements

DN acknowledges financial support from CSIR, India for the ASPIRE research grant (01WS(018)/2023-24/EMR-II/ASPIRE). Shivnandi thanks Prime Minister Research Fellowship (PMRF) for the funding and fellowship. The authors acknowledge the IIT Delhi HPC facility for the computational resources.

References

1. C. M. Dobson, Trends Biochem. Sci., 1999, 24, 329–332.
2. C. M. Dobson, Nature, 2003, 426, 884–890.
3. T. R. Jahn and S. E. Radford, FEBS J., 2005, 272, 5962–5970.
4. J. Greenwald, W. Kwiatkowski and R. Riek, J. Mol. Biol., 2018, 430, 3735–3750.
5. T. P. Knowles, M. Vendruscolo and C. M. Dobson, Nat. Rev. Mol. Cell Biol., 2014, 15, 384–396.
6. J. D. Sipe, M. D. Benson, J. N. Buxbaum, S.-I. Ikeda, G. Merlini, M. J. Saraiva and P. Westermark, Amyloid, 2014, 21, 221–224.
7. E. H. Koo, P. T. Lansbury Jr and J. W. Kelly, Proc. Natl. Acad. Sci. U. S. A., 1999, 96, 9989–9990.
8. F. Chiti and C. M. Dobson, Annu. Rev. Biochem., 2017, 86, 27–68.
9. V. N. Uversky, C. J. Oldfield and A. K. Dunker, Annu. Rev. Biophys., 2008, 37, 215–246.
10. M. M. Babu, R. van der Lee, N. S. de Groot and J. Gsponer, Curr. Opin. Struct. Biol., 2011, 21, 432–440.
11. S. Menon and N. Sengupta, Biochim. Biophys. Acta, Proteins Proteomics, 1867, 2019, 941–953.
12. M. C. Owen, D. Gnutt, M. Gao, S. K. Wärmländer, J. Jarvet, A. Gräslund, R. Winter, S. Ebbinghaus and B. Strodel, Chem. Soc. Rev., 2019, 48, 3946–3996.
13. S. B. Zimmerman and A. P. Minton, Annu. Rev. Biophys. Biomol. Struct., 1993, 22, 27–65.
14. S. Asakura and F. Oosawa, J. Polym. Sci., 1958, 33, 183–192.
15. R. Bhatt and S. N. Timasheff, Protein Sci., 1992, 1, 1133–1143.
16. H.-X. Zhou, G. Rivas and A. P. Minton, Annu. Rev. Biophys., 2008, 37, 375–397.
17. M. Senske, L. Törk, B. Born, M. Havenith, C. Herrmann and S. Ebbinghaus, J. Am. Chem. Soc., 2014, 136, 9036–9041.
18. M. Feig and Y. Sugita, J. Phys. Chem. B, 2012, 116, 599–605.
19. S. K. Mukherjee, S. Gautam, S. Biswas, J. Kundu and P. K. Chowdhury, J. Phys. Chem. B, 2015, 119, 14145–14156.
20. L. Sapir and D. Harries, Curr. Opin. Coll. Colloid Interface Sci., 2015, 20, 3–10.
21. D. Nayar, Phys. Chem. Chem. Phys., 2020, 22, 18091–18101.
22. D. Nayar, J. Phys. Chem. B, 2023, 127, 6265–6276.
23. C. Alfano, Y. Fichou, K. Huber, M. Weiss, E. Spruijt, S. Ebbinghaus, G. De Luca, M. A. Morando, V. Vetri, P. A. Temussi and A. Pastore, Chem. Rev., 2024, 124, 3186–3219.
24. S. L. Speer, C. J. Stewart, L. Sapir, D. Harries and G. J. Pielak, Annu. Rev. Biophys., 2022, 51, 267–300.
25. H. Garg, R. Rajesh and S. Vemparala, J. Chem. Phys., 2023, 158, 14903.
26. C. J. Stewart, G. I. Olgenblum, A. Propst, D. Harries and G. J. Pielak, Protein Sci., 2023, 32, e4573.
27. F. Chiti and C. M. Dobson, Annu. Rev. Biochem., 2006, 75, 333–366.
28. I. M. Ilie and A. Caflisch, Chem. Rev., 2019, 119, 6956–6993.
29. S. P. Patel, T. Nikam, B. Sreepathi, V. S. Karankar, A. Jaiswal, S. V. Vardhan, A. Rana, V. Toga, N. Srivastava, S. A. Saraf and S. Awasthi, ACS Chem. Biol., 2024, 19, 2118–2130.
30. M. D. Shtilerman, T. T. Ding and P. T. Lansbury, Biochemistry, 2002, 41, 3855–3860.
31. V. N. Uversky, E. M. Cooper, K. S. Bower, J. Li and A. L. Fink, FEBS Lett., 2002, 515, 99–103.
32. J. Fung, A. A. Darabie and J. McLaurin, Biochem. Biophys. Res. Commun., 2005, 328, 1067–1072.
33. S. Heravi, J. V. D. Power, A. Yethiraj and V. Booth, Protein Sci., 2024, 33, e4894.
34. D. C. Latshaw and C. K. Hall, Biophys. J., 2015, 109, 124–134.
35. A. Soranno, I. Koenig, M. B. Borgia, H. Hofmann, F. Zosel, D. Nettels and B. Schuler, Biophys. J., 2014, 106, 268a.
36. I. M. Kuznetsova, B. Y. Zaslavsky, L. Breydo, K. K. Turoverov and V. N. Uversky, Molecules, 2015, 20, 1377–1409.
37. D.-S. Yang, C. M. Yip, T. J. Huang, A. Chakrabartty and P. E. Fraser, J. Biol. Chem., 1999, 274, 32970–32974.
38. J. E. Straub and D. Thirumalai, Curr. Opin. Struct. Biol., 2010, 20, 187–195.
39. D. Thirumalai, G. Reddy and J. E. Straub, Acc. Chem. Res., 2012, 45, 83–92.
40. N. Das, E. Tarif, A. Dutta and P. Sen, J. Phys. Chem. B, 2023, 127, 3151–3163.
41. M. G. Krone, L. Hua, P. Soto, R. Zhou, B. J. Berne and J. E. Shea, J. Am. Chem. Soc., 2008, 130, 11066–11072.
42. P. Khatua, M. Gupta and S. Bandyopadhyay, J. Chem. Inf. Model., 2022, 62, 3453–3462.
43. M. Gao, D. Gnutt, A. Orban, B. Appel, F. Righetti, R. Winter, F. Narberhaus, S. Müller and S. Ebbinghaus, Angew. Chem., Int. Ed., 2016, 55, 3224–3228.
44. N. Das and P. Sen, J. Phys. Chem. B, 2020, 124, 5858–5871.
45. T. Khan, B. Halder, N. Das and P. Sen, J. Phys. Chem. B, 2024, 128, 8672–8686.
46. B. Hämisch, R. Pollak, S. Ebbinghaus and K. Huber, Chem. – Eur. J, 2020, 26, 8140–8152.
47. L. Breydo, A. E. Sales, T. Frege, M. C. Howell, B. Y. Zaslavsky and V. N. Uversky, Biochemistry, 2015, 54, 2957–2966.
48. L. Gremer, D. Schölzel, C. Schenk, E. Reinartz, J. Labahn, R. B. Ravelli, M. Tusche, C. Lopez-Iglesias, W. Hoyer, H. Heise, D. Willbold and G. F. Schröder, Science, 2017, 358, 116–119.
49. J. J. Balbach, Y. Ishii, O. N. Antzutkin, R. D. Leapman, N. W. Rizzo, F. Dyda, J. Reed and R. Tycko, Biochemistry, 2000, 39, 13748–13759.
50. J. Huang, S. Rauscher, G. Nawrocki, T. Ran, M. Feig, B. L. De Groot, H. Grubmüller and A. D. MacKerell Jr, Nat. Methods, 2017, 14, 71–73.
51. B. Hess, C. Kutzner, D. van der Spoel and E. Lindahl, J. Chem. Theory Comput., 2008, 4, 435–447.
52. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38.
53. V. Pierce, M. Kang, M. Aburi, S. Weerasinghe and P. E. Smith, Cell Biochem. Biophys., 2008, 50, 1–22.
54. J. R. Errington and P. G. Debenedetti, Nature, 2001, 409, 318–321.
55. P.-L. Chau and A. Hardwick, Mol. Phys., 1998, 93, 511–518.
56. M. Agarwal, H. R. Kushwaha and C. Chakravarty, J. Phys. Chem. B, 2010, 114, 651–659.
57. D. Nayar, M. Agarwal and C. Chakravarty, J. Chem. Theory Comput., 2011, 7, 3354–3367.
58. S. Gnanakaran, R. Nussinov and A. E. García, J. Am. Chem. Soc., 2006, 128, 2158–2159.
59. S. Rajput, R. Pollak, K. Huber, S. Ebbinghaus and D. Nayar, Soft Matter, 2023, 19, 6399–6413.
60. L. A. Ferreira, P. P. Madeira, L. Breydo, C. Reichardt, V. N. Uversky and B. Y. Zaslavsky, J. Biomol. Struct. Dyn., 2016, 34, 92–103.
61. R. Harada, Y. Sugita and M. Feig, J. Am. Chem. Soc., 2012, 134, 4842–4849.
62. Q. Chaboche, G. Campos-Villalobos, G. Giunta, M. Dijkstra, M. C. Lagomarsino and V. F. Scolari, Soft Matter, 2024, 20, 3271–3282.
63. D. C. Latshaw, M. Cheon and C. K. Hall, J. Phys. Chem. B, 2014, 118, 13513–13526.
64. C. Hilbich, B. Kisters-Woike, J. Reed, C. L. Masters and K. Beyreuther, J. Mol. Biol., 1992, 228, 460–473.
65. L. Xie, Y. Luo and G. Wei, J. Phys. Chem. B, 2013, 117, 10149–10160.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5sm00206k

---