Self-assembly coupled to liquid-liquid phase separation

Michael F. Hagan , et al.


The self-assembly of basic subunits into larger structures with well-defined architectures underlies essential functions in biological organisms, where examples of assembled structures include multi-protein filaments such as microtubules or actin [1, 2], scaffolds for vesicular budding [39], the outer shells or ‘capsids’ of viruses [1016], and bacterial microcompartments [1722] or other proteinaceous organelles [2327]. However, achieving efficient and high fidelity assembly into target architectures requires precisely tuned subunit interaction strengths and concentrations due to competing thermodynamic and kinetic effects (e.g. [2847]). The need for such precision could severely constrain the use of assembly for biological function or human engineered applications. Biological organisms employ multiple modes of biochemical and physical regulation to overcome this limitation. In this article, we investigate one such mode—how spatial heterogeneity due to formation of biomolecular condensates, or liquid-liquid phase-separation (LLPS), can dramatically enhance the speed and robustness of self-assembly.

While membranous organelles play a prominent role in compartmentalizing eukaryotic cells, it is now clear that condensates act as ‘membrane-less compartments’ to spatially organize cellular interiors within all kingdoms of life (e.g. [4869]). These compartments are implicated in diverse cellular functions, including transcriptional regulation [53, 7073], formation of neuronal synapses [7476], enrichment of specific proteins and nucleic acids [7782], cellular stress responses [8386], and cell division [79, 87]. In addition to the roles of condensates in normal cellular function, pathogenic viruses generate or exploit LLPS during various stages of their life cycles [8893]. Most relevant to this article, many viruses undergo assembly and/or genome packaging within phase-separated compartments known as virus factories, replication sites, Negri bodies, inclusion bodies, or viroplasms [88108]. In vitro studies show that viral nucleocapsid proteins and RNA molecules undergo LLPS (e.g. [99, 106, 109111], and that LLPS accelerates assembly of nucleocapsid-like particles [99]. It is hypothesized that viruses exploit LLPS to avoid host immune responses and coordinate events such as RNA replication, capsid protein translation, assembly, and genome packaging. However, the mechanisms underlying these events are poorly understood.

In addition to viruses, other examples of biological self-assembly coupled to LLPS include the formation of clathrin cages to mediate endocytosis [112]; post-synaptic densities [113] and pre-synaptic vesicles release sites (active zones) [114, 115] at neuronal synapses; observations that condensates can both accelerate and suppress aggregation of α-synuclein [116], and actin assembly in polypeptide coacervates [117].

Multiple lines of evidence suggest that condensate formation is driven by favorable interactions among their constituents combined with unfavorable interactions with the bulk exterior cytoplasm or nucleoplasm. Although condensation may be driven, destabilized, or regulated by diverse nonequilibrium effects (e.g. [54, 7073, 118121], equilibrium thermodynamics provides a starting point to model their stability, and their formation is frequently described as LLPS [48, 49, 52, 5260, 69, 74, 85, 118131]. Henceforth, we will use the term LLPS, keeping in mind that nonequilibrium effects may also be present. Consistent with equilibrium phase coexistence, the composition of the compartment interior can significantly differ from that of the cytoplasm. Thus, LLPS can provide significant spatiotemporal control over reaction processes by concentrating and colocalizing specific sets of subunit species that preferentially partition into the compartment.

These capabilities potentially enable LLPS to strongly regulate self-assembly. Yet, despite recent intense investigations into LLPS, its coupling to assembly has yet to be fully explored. Previous simulations showed that the condensed enzyme complex that forms the interior cargo of bacterial microcompartments can promote nucleation and control the size of the exterior protein shell [43, 45, 132134]. Most closely related to our work, Refs. [116, 135137] recently showed that the presence of a compartment can significantly accelerate irreversible protein aggregation into linear fibrils.

Here, we investigate the effects of LLPS on reversible self-limited assembly into target structures with finite sizes and well-defined architectures. Self-limited assembly from bulk solution is constrained by competing thermodynamic and kinetic effects—subunit interactions must be sufficiently strong and geometrically precise to stabilize the target structure, but overly high interaction strengths or subunit concentrations lead to kinetic traps (e.g. [2847]). Avoiding such kinetic traps imposes a ‘speed limit’ on assembly from bulk solution [41, 42, 138].

Using a master equation description of assembly, we show that these thermodynamic and kinetic constraints can be simultaneously satisfied by spatial heterogeneity due to phase-separated compartments. We find that LLPS can significantly accelerate assembly nucleation, consistent with previous studies of irreversible assembly [116, 135137], but also induces kinetic traps that slow assembly in certain parameter regimes. Crucially though, by enhancing nucleation only within spatially localized regions, LLPS significantly expands the range of subunit concentrations and interaction strengths over which such kinetic traps are avoided, thus promoting assembly robustness. This effect can increase by orders of magnitude the maximum rate of productive assembly into the ordered target structure. The extent of assembly acceleration and robustness enhancement are nonmonotonic functions of the key control parameters: the compartment size, and the partition coefficient of subunits between the compartment and the bulk cytoplasm. We present simple scaling estimates that capture the effect of LLPS on assembly, and reveal the underlying mechanisms that enable regulation. For example, the bulk solution acts as a “buffer” that steadily supplies free subunits to the compartment to enable rapid assembly without kinetic traps. Although we particularly focus on self-limited assembly processes that lead to finite-sized structures, our models are general and many results also apply to unlimited assembly or crystallization.



We have developed a minimal model to describe assembly in the presence of one or more liquid droplets coexisting with a background solution of different composition (Fig 1). We are motivated by processes such as virus assembly, in which viral proteins, nucleic acids and other viral components, and possibly some host proteins phase separate to form liquid compartments within the cellular cytoplasm. For this initial study, we consider only one assembling species in the limit that the assembly subunits comprise a small fraction of the compartment mass, and thus the size and composition of the compartment can be treated as independent of the subunit concentration.


Fig 1. Schematic of the model.

Subunits exchange between bulk and the phase-separated compartment (gray sphere), with equilibrium concentrations related by . Assembly can occur anywhere in the system, but occurs preferentially in the compartment when Kc > 1 due to the enhanced local subunit concentration. The volumes of the bulk Vbg and compartment Vc are related by Vr = Vc/Vbg.

We consider a system of subunits that form self-limited assemblies with optimal size N. The subunits are immersed in a multicomponent solvent which is in a state of phase coexistence, with stoichiometry such that there are one or more small compartments rich in one (or more) solvent species coexisting with a much larger background rich in the other species. While we consider protein subunits undergoing assembly, the solvent could be comprised of proteins, nucleic acids, or other macromolecules. The distinction is that ‘subunits’ form ordered (para)crystalline structures such as a capsid, while the primary compartment constituents remain amorphous and (for the regime we consider) in the liquid phase.

Our model can be viewed as a minimal starting point motivated by biological assembly coupled to LLPS. For example, in the model proposed for rotavirus assembly, two nonstructural rotavirus proteins (NSP2 and NSP5) undergo LLPS to form a viroplasm. The capsid proteins (VP2 and VP6) partition favorably into the viroplasm due to weak multivalent interactions with NSP2 and NSP5 [94, 98, 104, 107]. Here, NSP2 and NSP5 would correspond to the primary compartment constituents, and the capsid proteins (VP2 and VP6) correspond to the assembly subunits. We note that the rotavirus genomic RNA is also driven to partition into the viroplasm through interactions between RNA binding proteins and NSP2/5, but we do not explicitly consider assembly around RNA in this work to simplify the model. Our assumption of low concentration of assembly subunits in the compartment noted above is consistent with concentrations of VP2/6 in the viroplasm that are small compared to those of NSP2/5 [94, 98, 104, 107].

The driving force for subunits to enter the compartment phase is characterized by the partition coefficient Kc, which at equilibrium satisfies
with and the subunit concentrations in the compartment and background. The partition coefficient is related to the change in solvation free energy gc for a subunit that transitions from the background to the compartment as with β = 1/kBT with kBT the thermal energy. Applying standard dilute solution thermodynamics will result in a subunit solvation free energy difference with the form (see Weber et al. [
120]) gcnsχ)(Δϕ), where Δϕ is the difference in solvent composition between the background and compartment, Δχ is the difference in interaction strength (parameterized by the Flory χ parameter) for an interaction site on the subunit between the background and compartment, and ns is the number of interaction sites per subunit. The key point is that even for relatively weak interactions, a subunit with multivalency of ns ≳ 10 could have a partition coefficient as large as Kc ∼ 104 − 105, although Kc ∼ 100 may be a typical value [54].

We denote the volumes of the compartment and background as Vc and Vbg, which are related to the total system volume by Vtot = Vc + Vbg. We will present results in terms of the compartment size ratio, VrVc/Vbg. In most biological systems or in vitro experiments, the compartment volume will be small compared to the background, Vr ≪ 1. For this work we assume a fixed total subunit concentration ρT. For simplicity we will typically consider a single compartment, but we also discuss the case of multiple compartments, which might arise due to microphase separation or arrested phase separation.

Master equation models for capsid assembly kinetics

To simulate the assembly kinetics, we adapt the rate equation description originally developed by Zlotnick and coworkers [28, 29, 139] and used by others [138, 140] to describe the self-assembly of 2D polymers (capsids) in bulk solution. Denoting the concentration of an intermediate with n subunits in either phase as with α = dom, bg, the time evolution of intermediate concentrations is given by:
with the diffusive exchange between the phases given by (see Section A in
S1 Text and Refs. [135, 137])
and fn and bn as the association and dissociation rate constants for intermediates of size n. We set the initial condition as , , and .

We have made several assumptions to simplify the models, based on previous work (e.g [28, 29, 39, 41, 138140]). First, we assume that there is only one ‘average’ intermediate structure for each size n. Second, we assume that only individual subunits can associate to or dissociate from an intermediate. This assumption is based on the fact that particle-based computer simulations show that, at the dilute conditions typical of productive assembly reactions, most assembly events involve association of individual subunits [138], and that extending Eq (2) to allow for binding of higher-order oligomers does not qualitatively change the results (see the supplemental material of Ref. [138]). Third, we assume that the domain composition is independent of subunit concentration and assembly. Similarly, we assume that the diffusion coefficient is independent of subunit concentration. These simplifications are based on the assumption of low concentrations of subunits in the compartment and that the other compartment constituents are macromolecules that typically have equal or larger molecular weights as the subunits. To focus on effects of competing reactions on assembly, we also neglect the possible dependence of diffusion coefficients on intermediate size or gc, considered in Refs. [136, 137] respectively. The model can be readily extended to account for these effects.

The most important simplification is that we neglect the possibility of malformed (off-pathway) structures. While this is a good assumption under productive assembly conditions, particularly when subunit-subunit interactions have high orientational specificity, malformed structures can lead to kinetic traps at high concentrations or binding affinities [3234, 3642, 141143]. This effect will be considered by performing particle-based simulations in a future work, and we discuss its implications in the Conclusions section.

To complete the Master equation description we must specify the transition rates between intermediates. We consider two models, which consider different dependencies of rates on the partial capsid size [138].

Nucleation and growth model (NG).

We start with a simple generic model for a nucleated self-assembly process denoted as the ‘nucleation and growth (NG) model’ [138]. This can describe linear assembly with nucleation (e.g. assembly of a helical viral capsid or the equilibrium assembly of an actin filament), or polyhedral shell assembly with an initial nucleation step, followed by assembly along a single growth front until the shell closes on itself [29, 138, 140] (see Fig 2A).


Fig 2. Schematics of the two assembly pathway models.

(A) Nucleation and growth (NG) assembly pathway, Eq (4). There is one average intermediate for each size n. Subunits (capsid protein dimers in the schematic) associate or dissociate to intermediates with association rate f and dissociation rates: bnuc > 1 below the critical nucleus size (nnuc, a trimer of dimers in the schematic), belong < 1 during elongation, and bN < belong for dissociation of a subunit from a complete capsid. The rates f and bi with i ∈ {nuc, elong, N} are related to the subunit binding free energy gi by detailed balance as v0bi = f exp(βgi) with v0 the standard state volume and |gnuc| < |gelong| < |gN| accounting for the increase in number of contacts per subunit as the intermediate size grows. (B) Classical nucleation theory (CNT) assembly pathway. Capsid intermediates are represented as continuum elastic partial spherical shells, with the total binding free energy given by Eqs (5) and (6).

We consider a system of capsid protein subunits with total concentration ρT that start assembling at the time t = 0. We assume that the rate constants are the same in the compartment and background, so we simplify the presentation by omitting the specification of phase in this subsection. Our reaction is given by:
where bi is the dissociation rate constant (with i = {nuc, elong, N}), which is related to the forward rate constant by detailed balance as v0bi = f exp(βgi), with gi the change in interaction free energy upon subunit association to a partial capsid and v0 the standard state volume. The nucleation and elongation phases are distinguished by the fact that association in the nucleation phase has an unfavorable free energy change, gnuckBT ln(ρ1v0) > 0, while association in the elongation phase is favorable, gelongkBT ln(ρ1v0) < 0. For the moment, we assume that there is a single critical nucleus size nnuc.

For most results in this article, we will set gnuc = −4kBT and gelong = −17kBT and gN = 2gelong. The small value of gnuc relative to gelong accounts for the fact that the first few subunits to associate make fewer and/or less favorable contacts than subunits in larger intermediates, giving rise to a nucleation barrier, while the large value of gN accounts for the fact that in many assembly geometries the last subunit makes the largest number of contacts upon associating. We have chosen values of gnuc and gelong to be roughly consistent with binding affinity values estimated for virus capsid assembly [28, 30, 31, 144], but the results do not qualitatively change for other affinity values within a given assembly regime.

Classical nucleation theory model (CNT).

To test whether our conclusions depend qualitatively on the model geometry, we also consider transition rates based on the ‘classical nucleation theory (CNT)’ model for icosahedral capsids suggested by Zandi et al. [145]. In this model, assembly intermediates are represented as partial spheres that are missing a spherical cap. With the parameterization shown in Fig 2B, the total capsid size is N = 4πR2/a0 with R the capsid radius and a0 the subunit area, and intermediate sizes are given by n = N(1 − cos θ)/2. Subunits along the perimeter of the missing cap have fewer interactions than those in the shell interior, leading to a line tension σ, and the total binding free energy for an intermiate with n subunits is
with the perimeter of the missing spherical cap given by
with gsub the binding free energy per subunit in a complete capsid and l0 the diameter of a subunit. Following previous work [
138, 145], we set the line tension to σ = −gsub/2l0, so that a subunit adding to the perimeter of the capsid satisfies half of its contacts on average. We assume that the forward rate constant is proportional to the number of subunits on the perimeter, fn = f0ln/l0, with f0 the association rate constant for a single binding site, and we set and .

The key difference between the CNT and NG models is the dependence of the critical nucleus size on solution conditions. For the NG model the critical nucleus size nnuc is constant, provided exp(gnuc/kBT) < ρ1v0 < exp(gelong/kBT). For the CNT model, the critical nucleus size varies with subunit concentration and interaction strengths, and is given by the maximum in kBT log(ρ1v0)n + Gn, or [145]
with Γ = [gsub − ln(ρ1v0)]/σl1. Thus, the critical nucleus size continually changes over time during an assembly process for the CNT model as subunits are depleted, whereas it is constant until the very late stages of a NG assembly process.

Results and discussion

Effects of LLPS on self-assembly equilibrium

We begin by calculating how the equilibrium yield of self-assembled structures depends on subunit concentrations and interaction strengths, as well as the two key control parameters for subunit partitioning into the compartment: the partition coefficient Kc and the compartment size ratio Vr.

At equilibrium the subunit concentrations in the compartment and background are related to each other by Kc, and to the total subunit concentration ρT by mass conservation, giving

Equilibrium assembly yield.

We now calculate the effect of the compartment on assembly yields, using the well-justified approximation that intermediates have very low concentrations at equilibrium for self-limited assembly [41, 146]. Thus we consider a two-state system, with finite concentrations of only free subunits and complete assemblies with N. Mass conservation then gives
where and are the concentrations of assemblies in the compartment and background. At equilibrium these are related to the free subunit concentration by the law of mass action [
41, 147]
with α = dom, bg and gsub as the per-subunit interaction energy within a complete assembly (which we assume is the same in the compartment and background). Eqs (
10) and (11) can be easily solved numerically. However, we can make the results more transparent by following previous analysis for homogeneous assembly [41, 148] and writing the fraction of subunits in assemblies as

This simplifies to
(13) (14)

In the limit of large optimal assembly size N ≫ 1, Eq (13) satisfies the following asymptotic limits:
with ρCAC the critical assembly concentration (CAC) given by (assuming , so that all assembly occurs in the compartment)
as the CAC in a system without coupling to LLPS (i.e. Vr = 0 or Kc = 1). In all subsequent expressions, we will write the limit of no LLPS or Kc → 1 with a superscript ‘0’. The last expression in
Eq (16) assumes Vr ≪ 1 and shows that LLPS reduces the CAC by a factor KcKeff. Then using Eq (8) we arrive at the simple result that significant assembly occurs when the total subunit concentration ρT exceeds the local CAC within the compartment.

To obtain further insight, we note that KcKeff ≈ (Vr + 1/Kc)−1, giving the asymptotic limits
and that maximal enhancement of equilibrium assembly is achieved when KcVr.

Selectivity and spatial control over assembly.

We can draw two important conclusions from Eq (16). First, the presence of a compartment allows assembly under conditions where there is no bulk assembly (Fig 3). Second, there is a range of total subunit concentrations ∝ KcKeff over which assembly occurs only in the compartment, thus allowing for spatial control over assembly. As a measure of the extent to which LLPS can spatially control assembly, we define selectivity as . The equilibrium selectivity is then given by

We thus see that even a very small partition coefficient leads to strong equilibrium selectivity due to the high-valence nature of an assembled capsid. In particular, an assembled capsid has ∼N interactions with compartment components, but only has three translational degrees of freedom suppressed by partitioning into the compartment volume. However, if assembled capsids and large intermediates are not able to rapidly exchange between the compartment and background [137], the selectivity at finite times may be under kinetic control.


Fig 3. Effect of LLPS on the equilibrium and finite-time yields of self-assembly.

(A) The heat map shows the mass fraction of subunits in capsids (xN) as a function of the compartment partition coefficient (Kc) and total subunit concentration (ρT) computed from the rate equations with the nucleation-and-growth (NG) model (Eq (2)) at a finite time of 1 day. The lines show: the equilibrium critical assembly concentration (ρCAC, Eq (16), white ‘⋄’ symbols), the predicted threshold parameter values below which the median assembly timescale τ1/2 exceeds 1 day (ρnuc, Eq (29), red symbols), and the predicted locus of points corresponding to the minimum assembly timescale, beyond which monomer starvation begins to set in (ρ*, Eq (32), black ‘∘’ symbols). (B) The mass fraction of complete capsids xN as a function of total subunit concentration for no LLPS (Kc = 1). The line shows the equilibrium result (Eq (13)) and the symbols show results from numerically integrating the rate equations to 1 day (∼9 × 104 sec, red symbols) and t = 107 seconds (blue symbols). The dashed lines show ρCAC, ρnuc, and ρ*. (C) Same as (B), but in the presence of LLPS, with Kc = 36. Other parameters in (A-C) are critical nucleus size nnuc = 3, optimal size N = 120, subunit binding affinities gnuc = −4kBT, gelong = −17kBT, and compartment volume ratio Vr = 10−3.

Effect of LLPS on self-assembly kinetics

Master equation results

Assembly kinetics and yields.

Figs 3 and 4 show the effect of LLPS on assembly kinetics, as measured by the fraction of subunits in complete capsids (xN), obtained by numerically integrating the Master equation (Eq 2) with the NG model (Eq 4). Fig 4A shows xN as a function of time for several initial subunit concentrations ρT in the absence of LLPS. There is an initial lag phase during which intermediate populations build up to a quasi-steady-state, followed by rapid appearance of complete capsids, and then eventually saturation as free subunits are depleted. The duration of the lag phase decreases as 1/ρT.


Fig 4. The dependence of assembly kinetics on parameter values for the NG model with and without LLPS.

(A) The time evolution of the fraction of subunits in complete capsids xN for indicated values of the total subunit concentration ρT computed from the Master equation, with no LLPS. (B) The time evolution of xN for indicated values of Kc, for fixed ρT = 0.2μM and compartment ratio Vr = 0.001. The dashed lines show the timescale of 1 day corresponding to the red symbols in Fig 3B and 3C. Other parameter values for (A) and (B) are optimal assembly size N = 120, critical nucleus size nnuc = 3, gnuc = −4kBT, gelong = −17kBT.

Importantly, the rate of capsid production is nonmonotonic with respect to ρT—yields of complete capsids are suppressed for the highest concentration shown (ρT = 4M) by the monomer starvation kinetic trap arising from depletion of free subunits before capsids finish assembling. These results are discussed further in section Assembly timescales without LLPS and Refs. [41, 138]. This kinetic trapping effect is responsible for the low values of xN at high concentrations for finite-time results in Fig 3.

Figs 4B and 3A show how the assembly kinetics is changed by LLPS. With the lowest concentration shown in Fig 4A (ρT = 0.2 μM), xN is shown as a function of time for increasing values of the partition coefficient Kc. We see that the yields and assembly rates increase dramatically, with the duration of the lag phase decreasing and the maximum rate of capsid production (corresponding to the nucleation rate) increasing with Kc. To give a more comprehensive picture, Fig 3A shows xN as a function of both Kc and ρT. We see that assembly occurs at lower concentrations as Kc increases, and that LLPS increases the range of concentrations over which productive assembly occurs, particularly for Kc of . Similarly, Fig 3B and 3C respectively show xN as a function of concentration measured at 1 day, 107 seconds, and equilibrium. With or without LLPS, productive assembly at one day occurs over a much narrower range of concentrations than predicted by equilibrium, due to nucleation barriers at small concentrations and kinetic traps at high concentrations. Even at extremely long times the results have not reached full equilibrium due to kinetic traps at high concentrations. However, LLPS significantly broadens the range of concentrations leading to productive assembly at all timescales. The solid lines in Fig 3A and the dashed lines in Fig 3B and 3C indicate the CAC (ρCAC, Eq (16)), and scaling estimates for threshold concentrations below/above which productive assembly is impeded by large nucleation barriers or kinetic traps respectively (see section Assembly timescales without LLPS). Notice that in addition to making assembly more robust, LLPS also increases the maximum yield achievable at finite times. This increase arises because both nucleation and elongation rates can locally increase within the droplet due to the high local concentration while avoiding free subunit depletion.

The ability of LLPS to avoid kinetic traps arises because, for Vr ≪ 1, the background acts like a buffer that steadily supplies free subunit to the compartment even when the nucleation rate is large. As a measure of this behavior, Fig 5 shows the concentration of subunits in the background normalized by the total concentration, as a function of the maximum capsid formation rate (which occurs shortly after the end of the lag phase, before significant free subunit depletion has occurred). Here we have measured at the time point corresponding to the maximum rate. Results are shown for LLPS assembly for the same parameters as in Fig 4, with increasing rates corresponding to increasing values of Kc. For the case without LLPS, we achieve faster rates by increasing the subunit-subunit affinities from (gnuc, gelong) = (−4, −17)kBT to (gnuc, gelong) = (−6, −25.5)kBT. We have increased affinities rather than total concentration (as we do for other results) to simplify comparison of between the two cases. The results stop at gnuc = −6kBT because stronger affinities lead to low yields and decreasing rates due to the monomer starvation trap.


Fig 5. The background acts as a buffer of free subunits for LLPS-dominated assembly.

The plot shows the concentration of subunits in the background, , as a function of the maximum capsid formation rate (maximized over time at a given set of parameter values) for assembly with LLPS (black ‘∘’ symbols) and without LLPS (red symbols). For the LLPS case, the parameters correspond to those in Fig 4 with the increasing rate corresponding to Kc ∈ [1, 30]. For the case without LLPS, the parameters are the same except that Kc = 1 and the increasing rate is achieved by scaling the subunit-subunit affinities according to sgnuc and sgelong with s ∈ [1, 1.5]. The results stop at s = 1.5 because stronger affinities lead to kinetic traps and thus poor yields and a decreasing maximum rate.

We see that with LLPS the subunit concentration remains near ρT even for extremely high assembly rates, whereas subunits are rapidly depleted without LLPS. For higher subunit affinities |gnuc| > 6kBT depletion is so rapid that the monomer starvation trap sets in. Note that, in our Master equation description, subunits will eventually be depleted as xN → 1 even with LLPS, but in reality excluded volume effects (which are neglected in our model) would suppress assembly rates before this point unless complete capsids are expelled from the compartment.


Fig 6 shows the selectivity (xselec) measured from Master equation solutions and the equilibrium result (Eq (19)) as a function of Kc for several values of target capsid size N. We see that finite-time selectivity values closely match the equilibrium results, and that even an extremely small partition coefficient is sufficient to drive highly selective assembly in the compartment for large N.


Fig 6. Selectivity as a function of compartment partition coefficient and capsid size.

The symbols show the finite-time selectivity, , computed from the Master equation as a function of Kc for three indicated values of the optimal assembly size N at one day. The lines show the equilibrium selectivity (Eq (19)). Other parameters are the compartment volume ratio Vr = 0.001, gnuc = −4kBT, gelong = −17kBT, and ρT = 0.2μM.

Scaling estimates for the effect Of LLPS on assembly timescales

To gain insights into how LLPS can affect assembly, in this section we derive simple scaling estimates for the timescales associated with the nucleation and growth mechanism of Eq (4). We closely follow Refs. [138, 149], but we extend the analysis to include the effect of a compartment. Although we introduce a number of simplifications, in the next section we show that the resulting scaling estimates provide good approximations when these simplifications are relaxed by numerically solving the Master equation models.

Assembly timescales without LLPS.

Let us begin by summarizing the analysis of Ref [138] for assembly timescales in the absence of LLPS. As above, we consider a system of subunits with total concentration ρT that form assemblies with optimal size N subunits, and we break the assembly process into ‘nucleation’ and ‘elongation’ phases. For simplicity we assume that the association rate constant f is independent of intermediate size (except where mentioned otherwise), so that the rates of association to each intermediate are 1.

We now write the time required for an individual assemblage to form as τ = τnuc + τelong with τnuc and τelong the average times for nucleation and elongation, respectively.

The elongation timescale can generally be estimated as [138, 146]
where we have assumed Nnnuc so that NnnucN. The factor in the numerator indicates that the elongation timescale increases with optimal assembly size (i.e. α > 0) since independent subunit additions must occur. The value of the exponent α will depend on factors such as the dimensionality, the aggregate geometry, and the relative stability of intermediates, but we expect 1/2 ≤ α ≤ 2. For strongly forward-biased assembly during the elongation phase, α = 1 for the NG model and α = 1/2 for the CNT model (see [
138] and Section B in S1 Text). Except where specified otherwise, for the scaling estimates in the rest of this article we set α = 1, but the results are easily extendable to other exponent values.

The mean nucleation time at the beginning of the reaction can be estimated from the statistics of a random walk biased toward disassembly [29, 138], and can be approximately written as
where so that is the interaction free energy of the structure just below the critical nucleus size. The form of
Eq (21) can be understood by noting that the pre-critical nucleus is present with concentration , and subunits associate to the precritical nucleus with rate T. Note that the special case of nnuc = 2 corresponds to no nucleation barrier (since two subunits must associate to begin assembly), in which case and . We consider this case in Section D in S1 Text.

While Eq 21 gives the initial nucleation rate, the nucleation rate decreases over time due to subunit depletion, and asymptotically approaches zero as the concentration of complete capsids approaches its equilibrium value. Thus, we estimate the median assembly time τ1/2 (the time at which the reaction is 50% complete) by treating the system as a two-state reaction with nnuc-th order kinetics, which yields [138]
with , and xN as the equilibrium fraction of subunits in complete capsids. The factor of N−1 in
Eq 22 accounts for the fact that N subunits are depleted by each assembled capsid.

Analogous to crystallization or phase separation, there is a range of subunit concentrations and interaction strengths for which the unassembled state is metastable; i.e., the system is beyond the CAC so assembly is thermodynamically favorable, but the nucleation timescale exceeds experimentally relevant timescales. The boundary of this regime can be estimated by inverting Eq (22). Denoting the ‘relevant’ observation timescale as τobs, we can estimate the threshold subunit concentration below which nucleation will not be observed as

When elongation is fast compared to nucleation, the expressions Eq. S5 (in Section B in S1 Text) and Eq 22 respectively predict the duration of the lag phase and the median assembly time. However, these relations begin to fail above threshold values of the subunit concentration or subunit-subunit binding affinity, when nucleation and elongation timescales become comparable. Upon further increasing these parameters, nucleation becomes sufficiently fast that a significant fraction of monomers are depleted before elongation of most structures can complete. Subsequent evolution into complete assemblages then requires exchange of subunits between different intermediates (Ostwald ripening), which is an activated process and thus slow compared to assembly timescales. We describe this condition as the monomer-starvation kinetic trap. The threshold subunit concentration ρ* and interaction energies beyond which the system begins to enter the trap can be estimated by the locus of parameter values at which the median assembly time and elongation time are equal, i.e., τ1/2(ρ*) = τelong(ρ*):
and a corresponding assembly timescale

Note that corresponds to approximately the minimal timescale or maximal assembly rate (over all subunit concentrations) since both τ1/2 and τelong monotonically decrease with subunit concentration before the onset of kinetic trapping.

Effect of LLPS on maximal assembly rates and kinetic traps

We now evaluate the effect of the compartment on the propensity of the system to undergo the monomer-starvation kinetic trap, by evaluating the dependence of the elongation and median assembly timescales on the phase-separation parameters.

The median assembly timescale can be computed by the same analysis used above for the nucleation time, leading to
and similarly the threshold concentration below which nucleation does not occur on relevant timescales is
with given by
Eq (23).

The elongation time within the compartment, τelong,D or background τelong,B is given by Eq. S5 in Section B in S1 Text with the appropriate local concentration or . To estimate the onset of the kinetic trap, we must account for the numbers of capsids that are forming by both reaction channels (in the compartment or background), so we compute an average elongation time weighted by the relative number of assemblies that form the compartment or background

In the limit of strongly forward-biased elongation and so all nucleation occurs in the compartment, the elongation timescale is approximately

As discussed in section Assembly timescales without LLPS, the minimum assembly timescale occurs when the nucleation and elongation timescales are equal, τelong(Kc, Vr, ρ*) = τ1/2(Kc, Vr, ρ*); the monomer-starvation kinetic trap begins beyond this point. Using Eqs (26), (28), and (31) results in

Finally, we can approximately extend the scaling estimates of this section to the CNT model by substituting Eq (7) for nnuc.

Eq (32) shows that a key feature of preferential partitioning into the compartment is the ability of the system to buffer itself against the monomer-starvation kinetic trap while maintaining fast localized assembly in the compartment, as shown in Fig 5. We can further assess this feature in several ways as follows.

Fig 7A and 7B compare Eqs (28) and (30) to the median assembly and elongation times computed from the rate equations as a function of ρT and Kc respectively. We see that the scaling estimates and numerical results closely agree until the nucleation and elongation timescales become comparable; the threshold concentration ρ* (Eq (32)) and partition coefficient (Eq. S8 in Section C in S1 Text) are shown as vertical dashed lines in Fig 7A and 7B respectively. Fig 7C and 7D show the median assembly times computed from the rate equations as a function of ρT and Kc and Vr respectively, with the locus of parameter values leading to minimum assembly predicted by Eq (32) shown as white ∘ symbols. Fig 8 shows analogous results for the CNT model. The prediction closely tracks the minimum assembly timescale observed in the numerical results. Below this threshold the median assembly time is closely predicted by Eq (28), with the assembly timescale sped up according to snuc in Eq (26). Above this threshold the numerically computed assembly timescales rapidly increase due to overly fast nucleation and thus onset of the monomer-starvation trap. We also show the CAC and the threshold for achieving assembly within an observation time of 1 day on these plots. Notice that, at a given value of Vr, there is an optimal value of Kc (estimated below) which maximizes robustness of assembly to variations in concentration. In contrast, robustness monotonically increases with decreasing Vr.


Fig 7. Effect of LLPS on assembly timescales and the monomer-starvation kinetic trap for the NG model.

(A) The median assembly time τ1/2 and lag time calculated numerically from the Master equation (Eq (2)) and scaling estimates for the median assembly time (Eq (22)) and elongation time (Eq. S5 in Section B in S1 Text) as a function of subunit concentration, with no LLPS. The vertical dashed line indicates the scaling theory prediction for the concentration corresponding to the onset of the monomer starvation kinetic trap (ρ*, Eq (32)). (B) Same quantities, shown as a function of the partition coefficient for concentration ρT = 0.7μM. The vertical dashed line shows the estimate of the optimal value for the partition coefficient, (Eq. S8 in Section C in S1 Text) The compartment ratio is Vr = 10−3 for (A) and (B). (C, D) The median assembly time predicted by the rate equation model as a function of subunit concentration and (C) varying compartment partition coefficient with Vr = 10−3 or (D) varying Vr with Kc = 10. The white line and ‘∘’ symbols correspond to the theoretical prediction for the relationship between the subunit concentration and partition coefficient corresponding to the minimal assembly timescale (Eq (32)), beyond which the monomer-starvation kinetic trap begins to set in. The black line and ‘x’ symbols correspond to the relationship between the subunit concentration and Kc value (Eq (29)) below which nucleation will not be observed on an experimentally relevant observation timescale of τobs = 1 day. The black line and ‘◊’ symbols denote the concentration and Kc values corresponding to the CAC (Eq (16)). Other parameters are N = 120, nnuc = 3, gelong = −17kBT, and gnuc = −4kBT.


Fig 8. Effect of LLPS on assembly timescales and the monomer-starvation kinetic trap for the CNT model.

(A) The median assembly time τ1/2 and lag time calculated numerically from the Master equation (Eq (2)) and scaling estimates for the median assembly time (Eq (22)) and elongation time (Eq. S5 in Section B in S1 Text) as a function of subunit concentration, with no LLPS. The vertical dashed line indicates the concentration corresponding to the onset of the monomer starvation kinetic trap (ρ*, Eq (32)). (B) Same quantities, shown as a function of the partition coefficient for concentration ρT = 0.6μM. The vertical dashed line shows the estimate of the optimal value for the partition coefficient, (Eq. S8 in Section C in S1 Text). Parameter values are N = 120, gsub = −17kBT, and Vr = 10−3.

The maximum assembly speedup depends on volume ratio, critical nucleus size, and subunit concentration.

Given that the compartment both shifts and broadens the range of parameter values over which productive assembly can occur, it is of interest to determine parameters for which LLPS has the strongest effect on assembly times. To this end, we define the assembly ‘speedup’ as the factor by which the median assembly time decreases with LLPS relative to bulk solution: sLLPS(Kc, Vr) ≡ τ0/τ(Kc, Vr). Recalling that the minimum assembly timescale occurs at ρ* when elongation and nucleation times are equal, we can then maximize the speedup with respect to the partition coefficient to obtain (Section C in S1 Text)

Thus, for an optimal compartment partition coefficient, assembly can be sped up (i.e. τ1/2 reduced) by orders of magnitude for small Vr. The degree of speedup increases with: decreasing Vr, increasing critical nucleus size, and decreasing total subunit concentration. These trends can be understood as follows. Decreasing Vr means that subunits are not depleted as quickly within the compartment, thus allowing larger values of Kc and correspondingly higher local concentrations of subunits within the compartment without depleting subunits quickly enough to cause over-nucleation and monomer starvation. A larger critical nucleus size provides a larger separation between nucleation and growth timescales, thus enabling further concentration of subunits in the compartment without over-nucleation. The decreasing dependence on concentration arises because as the system approaches , the assembly timescale without LLPS decreases and thus so does the extent of possible speedup before over-nucleation sets in. However, note in Fig 8 that the maximum optimal concentration in the presence of LLPS exceeds the intrinsic value, , due to the extra regulation of nucleation and growth timescales allowed by a compartment. Also note that LLPS provides speedup even after monomer starvation begins to set in.

Fig 9 compares the scaling estimate for speedup (see Eq. S9 in Section C in S1 Text) to the value computed numerically from the rate equations. For the numerical value, we computed for fixed ρT and interaction parameters by numerically integrating the rate equations without LLPS, and then performed numerical minimization over Kc to obtain with respect to Kc for the same ρT and interaction parameters. Then the speedup is given by . We have presented the speedup as a function of concentration normalized by the optimal value in the absence of LLPS so that the results can be shown on the same plot. As shown in Fig 9A, the scaling estimate closely matches the numerical result until ρT exceeds the maximum value of ρ* at which point monomer starvation begins to set in. The results for the CNT model are obtained by substituting Eq (7) into Eqs. S7-S9. The agreement is reasonable but not as close as the NG estimate because Eq (7) is based on the critical nucleus size in the absence of LLPS. As noted above, we see that LLPS continues to speedup the assembly time even in the monomer-starvation regime.

Fig 9B shows the speedup as a function of Vr for fixed ρT. Here we see good agreement between the scaling estimate and numerical results, except the numerical speedup diminishes at small Vr for nnuc = 5. This occurs because the minimum assembly timescale has decreased below the diffusion limited timescale (Eq. S4 in Section A in S1 Text) in this regime, which is not accounted for in the scaling estimate.



It is well-established that efficient self-assembly in homogeneous solution is constrained to a narrow window of moderate subunit concentrations and interaction strengths, due to the competing constraints of minimizing nucleation timescales while avoiding kinetic traps [2847]. Here, we find that when subunits preferentially partition into nano- or microscale compartments, the range of parameters leading to productive assembly can be broadened by more than an order of magnitude, and the corresponding assembly timescales can be reduced by multiple orders of magnitude. Moreover, in part of this parameter range, almost all assembly occurs within the compartment interior, thus allowing spatial control over assembly. These behaviors depend sensitively on two parameters that control phase coexistence: the partition coefficient of subunits into phase separated compartments and the size ratio between the compartments and the cell. In addition, we find that the maximum degree of speedup due to LLPS increases with: decreasing compartment/cell size ratio or subunit concentration and increasing assembly critical nucleus size.

These effects arise because the compartment (or compartments) drive high local concentrations of subunits, thus minimizing the local nucleation timescales, but the small size of the compartment limits the total nucleation rate (averaged over the whole system volume). In effect, the bulk exterior acts as a subunit ‘buffer’ that, early in the reaction, steadily supplies subunits to the compartment and thereby suppresses the monomer starvation kinetic trap (see Fig 5). This mechanism has the strongest effect on robustness of assembly to variations in parameter values for small critical nucleus sizes or non-nucleated reactions, for which the homogeneous system lacks an intrinsic difference between nucleation and growth timescales and thus is most sensitive to subunit depletion. However, the decrease in assembly timescales is most dramatic for larger critical nucleus sizes, due to the high-order dependence of assembly timescales on local subunit concentration.

Relevant parameter ranges

Since these mechanisms depend on localization of subunits, the ability of LLPS to control assembly increases with decreasing compartment size (relative to the total system size). To estimate the relevance of this effect in biological systems, consider that typical compartments in eukaryotic cells range in size from ∼50nm to 10 μm [54, 123, 124, 130]. For a compartment with diameter 1 μm in a cell with diameter 20 μm, the volume ratio of the compartment relative to bulk is Vr ∼ 10−4. From Eqs (35) and (36) and Fig 7, we see that the range of subunit concentrations leading to productive assembly could increase by up to two orders of magnitude, with increases in assembly rates exceeding five orders of magnitude (Eq. S9 in Section C in S1 Text and Fig 9). These increases reflect the ability of compartmentalization to enable fast localized assembly while minimizing the rate of global depletion of subunits.


We have focused on a minimal model for this first study of the effects of LLPS on assembly robustness. There are a number of additional physical ingredients that merit further exploration. For parameter regimes that lead to high subunit concentrations within the compartment, the assumption that the subunits do not affect the equilibrium compartment size and composition will break down. Importantly, the rate equation models and scaling estimates considered here do not account for kinetic traps resulting from malformed assemblies, which can arise when subunits bound with incorrect geometries do not have time to anneal before becoming locked into place by association of additional subunits (e.g. [11, 32, 39, 41, 42, 141, 142, 150, 151]). Since association rates increase with concentration, we anticipate that malformed assemblies will occur above a threshold local concentration within the compartment, thus limiting the maximum speed up provided by LLPS. This threshold concentration increases with the geometric specificity of the subunit-subunit interactions. Thus, for sufficiently specific interactions the results described here will not qualitatively change when accounting for malformed structures—there will be a significant range of local concentrations, and thus assembly speed up, before either the malformed structure for monomer starvation kinetic trap set in. We will explore the effects of malformed assemblies on LLPS-coupled assembly in a future work.

Further, Schmit and Michaels showed that, if subunit diffusion slows with increasing subunit-compartment attraction strength (gc and Kc in our model), then there is an optimum Kc beyond which assembly slows. The results in our work arising from competing interactions are distinct from this effect. Other important effects to be incorporated include: slow diffusion into/out of the compartment [137], accounting for spatial structure and stochasticity of assembly [32, 152155], nonequilibrium effects such as synthesis of new subunits or phosphorylization-driven changes in assembly activity, selective partitioning of different species in a multicomponent assembly reaction, and the ability of the compartment to template the size and shape of assemblies, such as occurs in bacterial microcompartments [43, 45, 132, 133].

Ultimately, understanding how different combinations of these physical mechanisms enable phase separation processes to control the time, place, and rate of assembly will engender a more complete understanding of biological self-assembly, and can advance strategies for designing human-engineered nanostructured materials.


  1. 1.

    Weisenberg RC. Microtubule formation in vitro in solutions containing low calcium concentrations. Science. 1972;177(4054):1104–1105. pmid:4626639
  2. 2.

    Desai A, Mitchison TJ. Microtubule Polymerization Dynamics. Annual Review of Cell and Developmental Biology. 1997;13(1):83–117. pmid:9442869
  3. 3.

    Antonny B, Madden D, Hamamoto S, Orci L, Schekman R. Dynamics of the COPII coat with GTP and stable analogues. Nat Cell Biol. 2001;3(6):531–537. pmid:11389436
  4. 4.

    Matsuoka K, Schekman R, Orci L, Heuser JE. Surface structure of the COPII-coated vesicle. Proc Natl Acad Sci U S A. 2001;98(24):13705–13709. pmid:11717432
  5. 5.

    Doherty GJ, Mcmahon HT. Mechanisms of Endocytosis. Annu Rev Biochem. 2009;78:857–902. pmid:19317650
  6. 6.

    Kaksonen M, Roux A. Mechanisms of clathrin-mediated endocytosis. Nature Reviews Molecular Cell Biology. 2018;19(5):313–326. pmid:29410531
  7. 7.

    Kirchhausen T, Owen D, Harrison SC. Molecular structure, function, and dynamics of clathrin-mediated membrane traffic. Cold Spring Harbor perspectives in biology. 2014;6(24789820):a016725–a016725. pmid:24789820
  8. 8.

    Holland DO, Johnson ME. Stoichiometric balance of protein copy numbers is measurable and functionally significant in a protein-protein interaction network for yeast endocytosis. PLOS Computational Biology. 2018;14(3):1–34.
  9. 9.

    Guo SK, Sodt AJ, Johnson ME. Nascent clathrin lattices spontaneously disassemble without sufficient adaptor proteins. bioRxiv. 2021.
  10. 10.

    Zlotnick A, Mukhopadhyay S. Virus Assembly, Allostery and Antivirals. Trends Microbiol. 2011;19(1):14–23. pmid:21163649
  11. 11.

    Mateu MG. Assembly, stability and dynamics of virus capsids. Arch Biochem Biophys. 2013;531(1–2):65–79. pmid:23142681
  12. 12.

    Bruinsma RF, Klug WS. Physics of Viral Shells. Annu Rev Condens Matter Phys. 2015;6(1):245–268.
  13. 13.

    Perlmutter JD, Hagan MF. Mechanisms of Virus Assembly. Annu Rev Phys Chem. 2015;66(1):217–239. pmid:25532951
  14. 14.

    Hagan MF, Zandi R. Recent advances in coarse-grained modeling of virus assembly. Curr Opin Vir. 2016;18:36. pmid:27016708
  15. 15.

    Twarock R, Bingham RJ, Dykeman EC, Stockley PG. A modelling paradigm for RNA virus assembly. Curr Opin Vir. 2018;31:74–81. pmid:30078702
  16. 16.

    Zandi R, Dragnea B, Travesset A, Podgornik R. On virus growth and form. Phys Rep. 2020;.
  17. 17.

    Schmid MF, Paredes AM, Khant HA, Soyer F, Aldrich HC, Chiu W, et al. Structure of Halothiobacillus neapolitanus Carboxysomes by Cryo-electron Tomography. J Mol Biol. 2006;364(3):526–535. pmid:17028023
  18. 18.

    Iancu CV, Ding HJ, Morris DM, Dias DP, Gonzales AD, Martino A, et al. The Structure of Isolated Synechococcus Strain WH8102 Carboxysomes as Revealed by Electron Cryotomography. J Mol Biol. 2007;372(3):764–773. pmid:17669419
  19. 19.

    Kerfeld CA, Heinhorst S, Cannon GC. Bacterial microcompartments. Microbiology+. 2010;64(1):391. pmid:20825353
  20. 20.

    Rae BD, Long BM, Badger MR, Price GD. Functions, Compositions, and Evolution of the Two Types of Carboxysomes: Polyhedral Microcompartments That Facilitate CO2 Fixation in Cyanobacteria and Some Proteobacteria. Microbiol Mol Biol Rev. 2013;77(3):357–379. pmid:24006469
  21. 21.

    Chowdhury C, Sinha S, Chun S, Yeates TO, Bobik TA. Diverse bacterial microcompartment organelles. Microbiol Mol Biol Rev. 2014;78(3):438–68. pmid:25184561
  22. 22.

    Kerfeld CA, Melnicki MR. Assembly, function and evolution of cyanobacterial carboxysomes. Curr Opin Plant Biol. 2016;31:66–75. pmid:27060669
  23. 23.

    Sutter M, Boehringer D, Gutmann S, Gunther S, Prangishvili D, Loessner MJ, et al. Structural basis of enzyme encapsulation into a bacterial nanocompartment. Nat Struct Mol Biol. 2008;15(9):939–947. pmid:19172747
  24. 24.

    Nichols RJ, LaFrance B, Phillips NR, Oltrogge LM, Valentin-Alvarado LE, Bischoff AJ, et al. Discovery and characterization of a novel family of prokaryotic nanocompartments involved in sulfur metabolism. bioRxiv. 2020;.
  25. 25.

    Pfeifer F. Distribution, formation and regulation of gas vesicles. Nat Rev Micro. 2012;10(10):705–715. pmid:22941504
  26. 26.

    Kickhoefer VA, Rajavel KS, Scheffer GL, Dalton WS, Scheper RJ, Rome LH. Vaults Are Up-regulated in Multidrug-resistant Cancer Cell Lines. J Biol Chem. 1998;273(15):8971–8974. pmid:9535882
  27. 27.

    Zaslavsky BY, Ferreira LA, Darling AL, Uversky VN. The solvent side of proteinaceous membrane-less organelles in light of aqueous two-phase systems. Int J Biol Macromol. 2018;117:1224–1251. pmid:29890250
  28. 28.

    Zlotnick A, Johnson JM, Wingfield PW, Stahl SJ, Endres D. A Theoretical Model Successfully Identifies Features of Hepatitis B Virus Capsid Assembly. Biochemistry. 1999;38(44):14644–14652. pmid:10545189
  29. 29.

    Endres D, Zlotnick A. Model-Based Analysis of Assembly Kinetics for Virus Capsids or Other Spherical Polymers. Biophys J. 2002;83(2):1217–1230. pmid:12124301
  30. 30.

    Zlotnick A. Are Weak Protein-Protein Interactions the General Rule in Capsid Assembly? Virology. 2003;315:269–274. pmid:14585329
  31. 31.

    Ceres P, Zlotnick A. Weak Protein-Protein Interactions Are Sufficient to Drive Assembly of Hepatitis B Virus Capsids. Biochemistry. 2002;41(39):11525–11531. pmid:12269796
  32. 32.

    Hagan MF, Chandler D. Dynamic Pathways for Viral Capsid Assembly. Biophys J. 2006;91(1):42–54. pmid:16565055
  33. 33.

    Jack RL, Hagan MF, Chandler D. Fluctuation-dissipation ratios in the dynamics of self-assembly. Phys Rev E. 2007;76:021119. pmid:17930018
  34. 34.

    Rapaport DC. The Role of Reversibility in Viral Capsid Growth: A Paradigm for Self-Assembly. Phys Rev Lett. 2008;101:186101.
  35. 35.

    Whitelam S, Rogers C, Pasqua A, Paavola C, Trent J, Geissler PL. The Impact of Conformational Fluctuations on Self-Assembly: Cooperative Aggregation of Archaeal Chaperonin Proteins. Nano Lett. 2009;9:292–297. pmid:19072304
  36. 36.

    Nguyen HD, Reddy VS, Brooks CL. Deciphering the Kinetic Mechanism of Spontaneous Self-Assembly of Icosahedral Capsids. Nano Lett. 2007;7(2):338–344. pmid:17297998
  37. 37.

    Wilber AW, Doye JPK, Louis AA, Noya EG, Miller MA, Wong P. Reversible Self-Assembly of Patchy Particles into Monodisperse Icosahedral Clusters. J Chem Phys. 2007;127(8):085106. pmid:17764305
  38. 38.

    Wilber AW, Doye JPK, Louis AA, Lewis ACF. Monodisperse self-assembly in a model with protein-like interactions. J Chem Phys. 2009;131(17):175102. pmid:19895043
  39. 39.

    Hagan MF, Elrad OM, Jack RL. Mechanisms of Kinetic Trapping in Self-Assembly and Phase Transformation. J Chem Phys. 2011;135:104115. pmid:21932884
  40. 40.

    Cheng S, Aggarwal A, Stevens MJ. Self-assembly of artificial microtubules. Soft Matter. 2012;8(20):5666.
  41. 41.

    Hagan MF. Modeling Viral Capsid Assembly. Adv Chem Phys. 2014;155:1–68. pmid:25663722
  42. 42.

    Whitelam S, Jack RL. The Statistical Mechanics of Dynamic Pathways to Self-assembly. Ann Rev Phys Chem. 2015;66:143–63. pmid:25493714
  43. 43.

    Perlmutter JD, Mohajerani F, Hagan MF. Many-molecule encapsulation by an icosahedral shell. eLife. 2016;5:e14078. pmid:27166515
  44. 44.

    Lázaro GR, Dragnea B, Hagan MF. Self-assembly of convex particles on spherocylindrical surfaces. Soft Matter. 2018;14(28):5728–5740. pmid:29796568
  45. 45.

    Mohajerani F, Hagan MF. The role of the encapsulated cargo in microcompartment assembly. PLoS Comput Biol. 2018;14(7):e1006351. pmid:30063715
  46. 46.

    Panahandeh S, Li S, Marichal L, Leite Rubim R, Tresset G, Zandi R. How a Virus Circumvents Energy Barriers to Form Symmetric Shells. ACS Nano. 2020;14(3):3170–3180. pmid:32115940
  47. 47.

    Asor R, Schlicksup CJ, Zhao Z, Zlotnick A, Raviv U. Rapidly forming early intermediate structures dictate the pathway of capsid assembly. Journal of the American Chemical Society. 2020;142(17):7868–7882. pmid:32233479
  48. 48.

    Brangwynne CP. Soft active aggregates: mechanics, dynamics and self-assembly of liquid-like intracellular protein bodies. Soft Matter. 2011;7(7):3052.
  49. 49.

    Brangwynne C, Tompa P, Pappu R. Polymer physics of intracellular phase transitions. Nature Physics. 2015;11(11):899–904.
  50. 50.

    Hyman AA, Brangwynne CP. Beyond stereospecificity: liquids and mesoscale organization of cytoplasm. Dev Cell. 2011;21(1):14–16. pmid:21763600
  51. 51.

    Zhu F, Chen B. Monte Carlo Simulations of HIV Capsid Protein Homodimer. J Chem Inf Model. 2015;55(7):1361–1368. pmid:26107886
  52. 52.

    Bergeron-Sandoval LP, Safaee N, Michnick SW. Mechanisms and Consequences of Macromolecular Phase Separation. Cell. 2016;165(5):1067–79. pmid:27203111
  53. 53.

    Hnisz D, Shrinivas K, Young RA, Chakraborty AK, Sharp PA. A Phase Separation Model for Transcriptional Control. Cell. 2017;169(1):13–23. pmid:28340338
  54. 54.

    Banani SF, Lee HO, Hyman AA, Rosen MK. Biomolecular condensates: organizers of cellular biochemistry. Nat Rev Mol Cell Biol. 2017;18(5):285–298. pmid:28225081
  55. 55.

    Style RW, Sai T, Fanelli N, Ijavi M, Smith-Mannschott K, Xu Q, et al. Liquid-Liquid Phase Separation in an Elastic Network. Physical Review X. 2018;8(1).
  56. 56.

    Jacobs WM, Frenkel D. Phase Transitions in Biological Systems with Many Components. Biophys J. 2017;112(4):683–691. pmid:28256228
  57. 57.

    Jacobs WM. Self-Assembly of Biomolecular Condensates with Shared Components. Phys Rev Lett. 2021;126(25):258101. pmid:34241502
  58. 58.

    Joseph JA, Espinosa JR, Sanchez-Burgos I, Garaizar A, Frenkel D, Collepardo-Guevara R. Thermodynamics and kinetics of phase separation of protein-RNA mixtures by a minimal model. Biophys J. 2021;120(7):1219–1230. pmid:33571491
  59. 59.

    Perry SL. Phase separation: Bridging polymer physics and biology. Current Opinion in Colloid & Interface Science. 2019;39:86–97.
  60. 60.

    Sanchez-Burgos I, Espinosa JR, Joseph JA, Collepardo-Guevara R. Valency and Binding Affinity Variations Can Regulate the Multilayered Organization of Protein Condensates with Many Components. Biomolecules. 2021;11(2). pmid:33672806
  61. 61.

    Zhou J, Kondylis P, Haywood DG, Harms ZD, Lee LS, Zlotnick A, et al. Characterization of Virus Capsids and Their Assembly Intermediates by Multicycle Resistive-Pulse Sensing with Four Pores in Series. Anal Chem. 2018;90(12):7267–7274. pmid:29708733
  62. 62.

    Best RB. Computational and theoretical advances in studies of intrinsically disordered proteins. Curr Opin Struct Biol. 2017;42:147–154. pmid:28259050
  63. 63.

    Dignon GL, Zheng W, Best RB, Kim YC, Mittal J. Relation between single-molecule properties and phase behavior of intrinsically disordered proteins. Proc Natl Acad Sci U S A. 2018;115(40):9929–9934. pmid:30217894
  64. 64.

    Dignon GL, Zheng W, Kim YC, Best RB, Mittal J. Sequence determinants of protein phase behavior from a coarse-grained model. PLoS Comput Biol. 2018;14(1):e1005941. pmid:29364893
  65. 65.

    Dignon GL, Zheng W, Mittal J. Simulation methods for liquid-liquid phase separation of disordered proteins. Curr Opin Chem Eng. 2019;23:92–98. pmid:32802734
  66. 66.

    Ianiro A, Wu H, van Rijt MMJ, Vena MP, Keizer ADA, Esteves ACC, et al. Liquid-liquid phase separation during amphiphilic self-assembly. Nat Chem. 2019;11(4):320–328. pmid:30778139
  67. 67.

    Posey AE, Holehouse AS, Pappu RV. Phase Separation of Intrinsically Disordered Proteins. Methods Enzymol. 2018;611:1–30. pmid:30471685
  68. 68.

    Ruff KM, Pappu RV, Holehouse AS. Conformational preferences and phase behavior of intrinsically disordered low complexity sequences: insights from multiscale simulations. Curr Opin Struct Biol. 2018;56:1–10. pmid:30439585
  69. 69.

    Alberti S, Gladfelter A, Mittag T. Considerations and Challenges in Studying Liquid-Liquid Phase Separation and Biomolecular Condensates. Cell. 2019;176(3):419–434. pmid:30682370
  70. 70.

    Cho WK, Spille JH, Hecht M, Lee C, Li C, Grube V, et al. Mediator and RNA polymerase II clusters associate in transcription-dependent condensates. Science. 2018;361(6400):412–415. pmid:29930094
  71. 71.

    Chong S, Dugast-Darzacq C, Liu Z, Dong P, Dailey GM, Cattoglio C, et al. Imaging dynamic and selective low-complexity domain interactions that control gene transcription. Science. 2018;361 (6400). pmid:29930090
  72. 72.

    Henninger JE, Oksuz O, Shrinivas K, Sagi I, LeRoy G, Zheng MM, et al. RNA-Mediated Feedback Control of Transcriptional Condensates. Cell. 2021;184(1):207–225 e24. pmid:33333019
  73. 73.

    Sabari BR, Dall’Agnese A, Boija A, Klein IA, Coffey EL, Shrinivas K, et al. Coactivator condensation at super-enhancers links phase separation and gene control. Science. 2018;361 (6400). pmid:29930091
  74. 74.

    Wu X, Cai Q, Feng Z, Zhang M. Liquid-Liquid Phase Separation in Neuronal Development and Synaptic Signaling. Dev Cell. 2020;55(1):18–29. pmid:32726576
  75. 75.

    Stanishneva-Konovalova TB, Kelley CF, Eskin TL, Messelaar EM, Wasserman SA, Sokolova OS, et al. Coordinated autoinhibition of F-BAR domain membrane binding and WASp activation by Nervous Wreck. Proc Natl Acad Sci U S A. 2016;113(38):E5552–61. pmid:27601635
  76. 76.

    Del Signore SJ, Kelley CF, Messelaar EM, Lemos T, Marchan MF, Ermanoska B, et al. An autoinhibitory clamp of actin assembly constrains and directs synaptic endocytosis. eLife. 2021;10:e69597. pmid:34324418
  77. 77.

    Hernández-Vega A, Braun M, Scharrel L, Jahnel M, Wegmann S, Hyman BT, et al. Local Nucleation of Microtubule Bundles through Tubulin Concentration into a Condensed Tau Phase. Cell Rep. 2017;20(10):2304–2312. pmid:28877466
  78. 78.

    Mateju D, Franzmann TM, Patel A, Kopach A, Boczek EE, Maharana S, et al. An aberrant phase transition of stress granules triggered by misfolded protein and prevented by chaperone function. EMBO J. 2017;36(12):1669–1687. pmid:28377462
  79. 79.

    Parker R, Sheth U. P bodies and the control of mRNA translation and degradation. Mol Cell. 2007;25(5):635–46. pmid:17349952
  80. 80.

    Saha S, Weber CA, Nousch M, Adame-Arana O, Hoege C, Hein MY, et al. Polar Positioning of Phase-Separated Liquid Compartments in Cells Regulated by an mRNA Competition Mechanism. Cell. 2016;166(6):1572–1584.e16. pmid:27594427
  81. 81.

    Woodruff JB, Ferreira Gomes B, Widlund PO, Mahamid J, Honigmann A, Hyman AA. The Centrosome Is a Selective Condensate that Nucleates Microtubules by Concentrating Tubulin. Cell. 2017;169(6):1066–1077 e10. pmid:28575670
  82. 82.

    Zhang H, Zhao R, Tones J, Liu M, Dilley RL, Chenoweth DM, et al. Nuclear body phase separation drives telomere clustering in ALT cancer cells. Mol Biol Cell. 2020;31(18):2048–2056. pmid:32579423
  83. 83.

    Grousl T, Vojtova J, Hasek J, Vomastek T. Yeast stress granules at a glance. Yeast. 2022;39(4):247–261. pmid:34791685
  84. 84.

    Malinovska L, Kroschwald S, Munder MC, Richter D, Alberti S. Molecular chaperones and stress-inducible protein-sorting factors coordinate the spatiotemporal distribution of protein aggregates. Mol Biol Cell. 2012;23(16):3041–3056. pmid:22718905
  85. 85.

    Molliex A, Temirov J, Lee J, Coughlin M, Kanagaraj AP, Kim HJ, et al. Phase separation by low complexity domains promotes stress granule assembly and drives pathological fibrillization. Cell. 2015;163(1):123–33. pmid:26406374
  86. 86.

    Patel A, Lee HO, Jawerth L, Maharana S, Jahnel M, Hein MY, et al. A Liquid-to-Solid Phase Transition of the ALS Protein FUS Accelerated by Disease Mutation. Cell. 2015;162(5):1066–77. pmid:26317470
  87. 87.

    Brangwynne CP, Eckmann CR, Courson DS, Rybarska A, Hoege C, Gharakhani J, et al. Germline P Granules Are Liquid Droplets That Localize by Controlled Dissolution/Condensation. Science. 2009;324(5935):1729–1732. pmid:19460965
  88. 88.

    Brocca S, Grandori R, Longhi S, Uversky V. Liquid-Liquid Phase Separation by Intrinsically Disordered Protein Regions of Viruses: Roles in Viral Life Cycle and Control of Virus-Host Interactions. Int J Mol Sci. 2020;21(23). pmid:33260713
  89. 89.

    Etibor TA, Yamauchi Y, Amorim MJ. Liquid Biomolecular Condensates and Viral Lifecycles: Review and Perspectives. Viruses. 2021;13(3). pmid:33669141
  90. 90.
    Fernández de Castro I, Tenorio R, Risco C. In: Virus Factories; 2020.
  91. 91.

    Gaete-Argel A, Marquez CL, Barriga GP, Soto-Rifo R, Valiente-Echeverria F. Strategies for Success. Viral Infections and Membraneless Organelles. Front Cell Infect Microbiol. 2019;9:336. pmid:31681621
  92. 92.

    Lopez N, Camporeale G, Salgueiro M, Borkosky SS, Visentin A, Peralta-Martinez R, et al. Deconstructing virus condensation. PLoS Pathog. 2021;17(10):e1009926. pmid:34648608
  93. 93.

    Schoelz JE, Leisner S. Setting Up Shop: The Formation and Function of the Viral Factories of Cauliflower mosaic virus. Front Plant Sci. 2017;8:1832. pmid:29163571
  94. 94.

    Borodavka A, Desselberger U, Patton JT. Genome packaging in multi-segmented dsRNA viruses: distinct mechanisms with similar outcomes. Curr Opin Virol. 2018;33:106–112. pmid:30145433
  95. 95.

    Borodavka A, Dykeman EC, Schrimpf W, Lamb DC. Protein-mediated RNA folding governs sequence-specific interactions between rotavirus genome segments. eLife. 2017;6:e27453. pmid:28922109
  96. 96.

    Carlson CR, Asfaha JB, Ghent CM, Howard CJ, Hartooni N, Morgan DO. Phosphorylation modulates liquid-liquid phase separation of the SARS-CoV-2 N protein. bioRxiv. 2020. pmid:32637943
  97. 97.

    Garces Suarez Y, Martinez JL, Torres Hernandez D, Hernandez HO, Perez-Delgado A, Mendez M, et al. Nanoscale organization of rotavirus replication machineries. Elife. 2019;8. pmid:31343403
  98. 98.

    Geiger F, Acker J, Papa G, Wang X, Arter WE, Saar KL, et al. Liquid-liquid phase separation underpins the formation of replication factories in rotaviruses. EMBO J. 2021;40(21):e107711. pmid:34524703
  99. 99.

    Guseva S, Milles S, Jensen MR, Salvi N, Kleman JP, Maurin D, et al. Measles virus nucleo- and phosphoproteins form liquid-like phase-separated compartments that promote nucleocapsid assembly. Science Advances. 2020;6(14):eaaz7095. pmid:32270045
  100. 100.

    Kieser Q, Noyce RS, Shenouda M, Lin YJ, Evans DH. Cytoplasmic factories, virus assembly, and DNA replication kinetics collectively constrain the formation of poxvirus recombinants. PLoS One. 2020;15(1):e0228028. pmid:31945138
  101. 101.

    Luque D, Caston JR. Cryo-electron microscopy for the study of virus assembly. Nat Chem Biol. 2020;16(3):231–239. pmid:32080621
  102. 102.

    Nikolic J, Le Bars R, Lama Z, Scrima N, Lagaudriere-Gesbert C, Gaudin Y, et al. Negri bodies are viral factories with properties of liquid organelles. Nat Commun. 2017;8(1):58. pmid:28680096
  103. 103.

    Pan M, Alvarez-Cabrera AL, Kang JS, Wang L, Fan C, Zhou ZH. Asymmetric reconstruction of mammalian reovirus reveals interactions among RNA, transcriptional factor micro2 and capsid proteins. Nat Commun. 2021;12(1):4176. pmid:34234134
  104. 104.

    Papa G, Borodavka A, Desselberger U. Viroplasms: Assembly and Functions of Rotavirus Replication Factories. Viruses. 2021;13(7). pmid:34372555
  105. 105.

    Risso-Ballester J, Galloux M, Cao J, Le Goffic R, Hontonnou F, Jobart-Malfait A, et al. A condensate-hardening drug blocks RSV replication in vivo. Nature. 2021;595(7868):596–599. pmid:34234347
  106. 106.

    Savastano A, Ibanez de Opakua A, Rankovic M, Zweckstetter M. Nucleocapsid protein of SARS-CoV-2 phase separates into RNA-rich polymerase-containing condensates. Nat Commun. 2020;11(1):6041. pmid:33247108
  107. 107.

    Trask SD, McDonald SM, Patton JT. Structural insights into the coupling of virion assembly and rotavirus replication. Nat Rev Microbiol. 2012;10(3):165–77. pmid:22266782
  108. 108.

    Wang S, Dai T, Qin Z, Pan T, Chu F, Lou L, et al. Targeting liquid-liquid phase separation of SARS-CoV-2 nucleocapsid protein promotes innate antiviral immunity by elevating MAVS activity. Nat Cell Biol. 2021;23(7):718–732. pmid:34239064
  109. 109.

    Bessa Luiza M, Guseva S, Camacho-Zarco Aldo R, Salvi N, Maurin D, Perez Laura M, et al. The intrinsically disordered SARS-CoV-2 nucleoprotein in dynamic complex with its viral partner nsp3a. Science Advances. 2022;8(3):eabm4034. pmid:35044811
  110. 110.

    Chen H, Cui Y, Han X, Hu W, Sun M, Zhang Y, et al. Liquid-liquid phase separation by SARS-CoV-2 nucleocapsid protein and RNA. Cell Res. 2020;30(12):1143–1145. pmid:32901111
  111. 111.

    Iserman C, Roden CA, Boerneke MA, Sealfon RSG, McLaughlin GA, Jungreis I, et al. Genomic RNA Elements Drive Phase Separation of the SARS-CoV-2 Nucleocapsid. Mol Cell. 2020;80(6):1078–1091 e6. pmid:33290746
  112. 112.

    Day KJ, Kago G, Wang L, Richter JB, Hayden CC, Lafer EM, et al. Liquid-like protein interactions catalyse assembly of endocytic vesicles. Nat Cell Biol. 2021;23(4):366–376. pmid:33820972
  113. 113.

    Zeng M, Chen X, Guan D, Xu J, Wu H, Tong P, et al. Reconstituted Postsynaptic Density as a Molecular Platform for Understanding Synapse Formation and Plasticity. Cell. 2018;174(5):1172–1187.e16. pmid:30078712
  114. 114.

    McDonald NA, Fetter RD, Shen K. Assembly of synaptic active zones requires phase separation of scaffold molecules. Nature. 2020;588(7838):454–458. pmid:33208945
  115. 115.

    Emperador-Melero J, Wong MY, Wang SSH, de Nola G, Nyitrai H, Kirchhausen T, et al. PKC-phosphorylation of Liprin-α3 triggers phase separation and controls presynaptic active zone structure. Nature Communications. 2021;12(1):3057. pmid:34031393
  116. 116.

    Lipiński WP, Visser BS, Robu I, Fakhree MAA, Lindhoud S, Claessens MMAE, et al. Biomolecular condensates can both accelerate and suppress aggregation of α-synuclein. bioRxiv. 2022; p. 2022.04.22.489149.
  117. 117.

    McCall PM, Srivastava S, Perry SL, Kovar DR, Gardel ML, Tirrell MV. Partitioning and Enhanced Self-Assembly of Actin in Polypeptide Coacervates. Biophys J. 2018;114(7):1636–1645. pmid:29642033
  118. 118.

    Carenza LN, Biferale L, Gonnella G. Multiscale control of active emulsion dynamics. Physical Review Fluids. 2020;5(1):011302.
  119. 119.

    Plys Aaron J, Kingston Robert E. Dynamic condensates activate transcription. Science. 2018;361(6400):329–330. pmid:30049863
  120. 120.

    Weber CA, Zwicker D, Julicher F, Lee CF. Physics of active emulsions. Rep Prog Phys. 2019;82(6):064601. pmid:30731446
  121. 121.

    Zwicker D, Decker M, Jaensch S, Hyman AA, Julicher F. Centrosomes are autocatalytic droplets of pericentriolar material organized by centrioles. Proc Natl Acad Sci U S A. 2014;111(26):E2636–45. pmid:24979791
  122. 122.

    Bergeron-Sandoval LP, Heris HK, Chang C, Cornell CE, Keller SL, François P, et al. Endocytosis caused by liquid-liquid phase separation of proteins. bioRxiv. 2018; p. 145664.
  123. 123.

    Brangwynne CP. Phase transitions and size scaling of membrane-less organelles. J Cell Biol. 2013;203(6):875–881. pmid:24368804
  124. 124.

    Dignon GL, Best RB, Mittal J. Biomolecular Phase Separation: From Molecular Driving Forces to Macroscopic Properties. Annu Rev Phys Chem. 2020;71:53–75. pmid:32312191
  125. 125.

    Lin YH, Forman-Kay JD, Chan HS. Sequence-Specific Polyampholyte Phase Separation in Membraneless Organelles. Physical Review Letters. 2016;117(17):178101. pmid:27824447
  126. 126.

    Amin AN, Lin YH, Das S, Chan HS. Analytical Theory for Sequence-Specific Binary Fuzzy Complexes of Charged Intrinsically Disordered Proteins. The Journal of Physical Chemistry B. 2020;124(31):6709–6720. pmid:32639157
  127. 127.

    Lin YH, Chan HS. Phase Separation and Single-Chain Compactness of Charged Disordered Proteins Are Strongly Correlated. Biophys J. 2017;112(10):2043–2046. pmid:28483149
  128. 128.

    Falahati H, Haji-Akbari A. Thermodynamically driven assemblies and liquid-liquid phase separations in biology. Soft Matter. 2019;15(6):1135–1154. pmid:30672955
  129. 129.

    Feric M, Vaidya N, Harmon TS, Mitrea DM, Zhu L, Richardson TM, et al. Coexisting Liquid Phases Underlie Nucleolar Subcompartments. Cell. 2016;165(7):1686–97. pmid:27212236
  130. 130.

    Hyman AA, Simons K. Cell biology. Beyond oil and water–phase transitions in cells. Science. 2012;337(6098):1047–9. pmid:22936764
  131. 131.

    Zhou HX, Nguemaha V, Mazarakos K, Qin S. Why Do Disordered and Structured Proteins Behave Differently in Phase Separation? Trends Biochem Sci. 2018;43(7):499–516. pmid:29716768
  132. 132.

    Mohajerani F, Sayer E, Neil C, Inlow K, Hagan MF. Mechanisms of Scaffold-Mediated Microcompartment Assembly and Size Control. ACS nano. 2021;15(3):4197–4212. pmid:33683101
  133. 133.

    Rotskoff GM, Geissler PL. Robust nonequilibrium pathways to microcompartment assembly. Proc Natl Acad Sci USA. 2018;115:6341–6346. pmid:29866851
  134. 134.

    Li Y, Kennedy NW, Li S, Mills CE, Tullman-Ercek D, Olvera de la Cruz M. Computational and Experimental Approaches to Controlling Bacterial Microcompartment Assembly. ACS Cent Sci. 2021;7(4):658–670. pmid:34056096
  135. 135.

    Weber C, Michaels T, Mahadevan L. Spatial control of irreversible protein aggregation. Elife. 2019;8. pmid:31084715
  136. 136.

    Schmit JD, Michaels TCT. Physical limits to acceleration of chemical reactions inside phase-separated compartments. bioRxiv. 2022.
  137. 137.

    Pönisch W, Michaels TCT, Weber CA. Aggregation controlled by condensate rheology. bioRxiv. 2021; p. 2021.11.05.467474.
  138. 138.

    Hagan MF, Elrad OM. Understanding the Concentration Dependence of Viral Capsid Assembly Kinetics—the Origin of the Lag Time and Identifying the Critical Nucleus Size. Biophys J. 2010;98:1065–74. pmid:20303864
  139. 139.

    Zlotnick A. To Build a Virus Capsid—an Equilibrium-Model of the Self-Assembly of Polyhedral Protein Complexes. J Mol Biol. 1994;241(1):59–67. pmid:8051707
  140. 140.

    Morozov AY, Bruinsma RF, Rudnick J. Assembly of viruses and the pseudo-law of mass action. J Chem Phys. 2009;131(15):155101. pmid:20568884
  141. 141.

    Whitelam S, Feng EH, Hagan MF, Geissler PL. The role of collective motion in examples of coarsening and self-assembly. Soft Matter. 2009;5(6):1251–1262. pmid:23227104
  142. 142.

    Grant J, Jack RL, Whitelam S. Analyzing mechanisms and microscopic reversibility of self-assembly. J Chem Phys. 2011;135(21). pmid:22149800
  143. 143.

    Grant J, Jack RL. Quantifying reversibility in a phase-separating lattice gas: An analogy with self-assembly. Phys Rev E. 2012;85(2).
  144. 144.

    Zlotnick A, Suhanoysky MM, Teschke CM. The energetic contributions of scaffolding and coat proteins to the assembly of bacteriophage procapsids. Virology. 2012;428(1):64–69. pmid:22520942
  145. 145.

    Zandi R, van der Schoot P, Reguera D, Kegel W, Reiss H. Classical Nucleation Theory of Virus Capsids. Biophys J. 2006;90(6):1939–1948. pmid:16387781
  146. 146.

    Hagan MF, Grason GM. Equilibrium mechanisms of self-limiting assembly. Reviews of Modern Physics. 2021;93(2):025008. pmid:35221384
  147. 147.

    Hagan MF, Grason GM. Equilibrium mechanisms of self-limiting assembly. Rev Mod Phys. 2020;submitted.
  148. 148.

    van der Schoot P, Zandi R. Kinetic Theory of Virus Capsid Assembly. Phys Biol. 2007;4(4):296–304. pmid:18185007
  149. 149.

    Lazaro GR, Hagan MF. Allosteric control in icosahedral capsid assembly. J Phys Chem B. 2016;120(Bill Gerlbart Festschrift):6306–6318. pmid:27117092
  150. 150.

    Stray SJ, Bourne CR, Punna S, Lewis WG, Finn MG, Zlotnick A. A Heteroaryldihydropyrimidine Sctivates and Can Misdirect Hepatitis B Virus Capsid Assembly. Proc Natl Acad Sci U S A. 2005;102(23):8138–8143. pmid:15928089
  151. 151.

    Mohajerani F, Tyukodi B, Schlicksup CJ, Hadden-Perilla JA, Zlotnick A, Hagan MF. Multiscale Modeling of Hepatitis B Virus Capsid Assembly and Its Dimorphism. ACS Nano. 2022;16(9):13845–13859. pmid:36054910
  152. 152.

    Varga MJ, Fu Y, Loggia S, Yogurtcu ON, Johnson ME. NERDSS: A Nonequilibrium Simulator for Multibody Self-Assembly at the Cellular Scale. Biophysical Journal. 2020;118(12):3026–3040. pmid:32470324
  153. 153.

    Schwartz R, Shor PW, Prevelige PE, Berger B. Local Rules Simulation of the Kinetics of Virus Capsid Self-Assembly. Biophys J. 1998;75(6):2626–2636. pmid:9826587
  154. 154.

    Rapaport DC. Molecular dynamics simulation of reversibly self-assembling shells in solution using trapezoidal particles. Phys Rev E. 2012;86:051917. pmid:23214824
  155. 155.

    Johnson ME, Chen A, Faeder JR, Henning P, Moraru II, Meier-Schellersheim M, et al. Quantifying the roles of space and stochasticity in computer simulations for cell biology and cellular biochemistry. Molecular Biology of the Cell. 2021;32(2):186–210. pmid:33237849

Source link