Skip to main content
Biophysical Journal logoLink to Biophysical Journal
. 2025 Sep 24;124(21):3596–3609. doi: 10.1016/j.bpj.2025.09.002

Reorganization of DNA loops by competition between condensin I and a linker histone

Tetsuya Yamamoto 1,∗, Keishi Shintomi 2, Tatsuya Hirano 2
PMCID: PMC12709421  PMID: 40997801

Abstract

Condensin-mediated loop extrusion is thought to be one of the primary mechanisms underlying mitotic chromosome assembly. However, how this process is affected by other chromosomal proteins, such as histones, is not well understood. Our previous study showed that in Xenopus egg extracts codepleted of topoisomerase IIα and the histone chaperone Asf1, a highly characteristic chromatin structure called the “sparkler” is assembled. The sparkler is a compact structure assembled on nucleosome-free, entangled DNA in which multiple protrusions radiate from a core. Interestingly, condensin I is concentrated at the tips of the protrusions, whereas the linker histone H1.8 is enriched in the remaining regions of the structure. To understand the biophysical mechanisms underlying sparkler assembly, we construct a model predicting that DNA loops extruded from the entangled DNA undergo phase separation into two domains: loops enriched in condensin I remain as protrusions, whereas those enriched in H1.8 are reeled into the central region. We propose that H1.8 competes with condensin I for DNA binding, thereby reorganizing DNA loops formed by condensin I under this specialized condition.

Significance

A peculiar chromatin structure, termed a “sparkler,” assembles on nucleosome-free, entangled DNA in Xenopus egg extracts. This structure is characterized by multiple protrusions emanating from a condensed DNA mass. To elucidate the mechanism underlying sparkler assembly, we developed a theory predicting that competition between condensin I and linker histone drives phase separation within the brush of DNA loops extruded from the central region by condensin I. According to this model, condensed DNA loops enriched with linker histone are rapidly reeled into the central region, whereas swollen DNA loops enriched with condensin I remain at the periphery, thereby breaking the spherical symmetry of the system. Our results suggest that linker histone affects the stability of DNA loops extruded by condensin I.

Introduction

Upon entry into mitosis, chromosomal DNA undergoes a series of drastic structural transitions from relatively diffuse interphase chromatin to a discrete set of rod-shaped structures, each composed of duplicated sister chromatids. This process, collectively referred to as mitotic chromosome assembly, is essential for the faithful segregation of genetic information into daughter cells. Over the past two decades, it has been established that condensins, a class of structural maintenance of chromosome protein complexes, play central roles in this process (1). A recent theory, combined with computer simulations, predicted that condensins drive mitotic chromosome assembly by forming successive loops through an active mechanism called loop extrusion (2,3,4,5,6). Indeed, subsequent single-molecule assays supported the loop extrusion theory by demonstrating that budding yeast condensin translocates along DNA and forms DNA loops in an ATP hydrolysis-dependent manner (7,8). Despite these remarkable advances, the mechanism by which condensins extrude loops within crowded DNA environments where they interact with other chromosomal proteins is not well understood.

To understand the mechanisms underlying mitotic chromosome assembly, we have developed and refined a cell-free assay using Xenopus egg extracts, in which mitotic chromosome-like structures are assembled from mouse sperm nuclei (Fig. 1 a) (9,10,11). During the assembly of mitotic chromosome-like structures, protamines initially associated with sperm DNA are dissolved in the extracts, and nucleosomes are subsequently assembled along the DNA with the help of histone chaperones. When the nuclei are incubated with egg extracts depleted of the histone chaperone Asf1, individual chromosome-like structures composed of condensin-enriched central axes and fuzzy extended loops—nucleosome-depleted chromosomes—are assembled (Fig. 1 b). However, when the same nuclei are incubated with egg extracts codepleted of the histone chaperone Asf1 and topoisomerase IIα (topo IIα), the chromosomal DNA remains entangled and is converted into a highly characteristic structure, called the sparkler, in which multiple protrusions radiate from a DNA-dense core (Fig. 1, c and d). Condensin I is concentrated at the tips of the protrusions, whereas the linker histone H1.8 is concentrated at the remaining regions of the structure. The compaction of DNA is probably driven by multivalent interaction involving H1.8 aberrantly bound to nucleosome-free DNA (10). Such interactions have also been found from other variants of linker histones (12,13,14,15). A previous theory sought to explain sparkler assembly as a volume-phase transition of a DNA gel, driven by multivalent interactions between H1.8 and DNA tension generated by condensin I-mediated loop extrusion (15). However, this theory did not address the mechanism of the emergence of protrusions, one of the most prominent features of sparklers.

Figure 1.

Figure 1

Background information. (A) Chromosome/chromatin structures assembled in Xenopus egg extracts under three different conditions. In control extracts, a cluster of individual, rod-shaped chromosomes is assembled on nucleosomal DNA (a). In extracts depleted of the histone chaperone Asf1, a cluster of individual chromosome-like structures with fuzzy surfaces is formed on nucleosome-free DNA (“nucleosome-depleted” chromosomes (9)) (b). In extracts codepleted of topo IIα and Asf1, no individual chromosomes are observed, resulting in the formation of a highly characteristic chromatin structure (“sparkler”) on nucleosome-free, entangled DNA (10) (c). Although the linker histone H1.8 is undetectable in the nucleosome-depleted chromosomes (b), it is abundantly present in the sparkler (c). (B) Time course of sparkler formation (10). When mouse sperm nuclei (d) are incubated with Xenopus egg extracts codepleted of topo IIα and Asf1, entangled chromosomal DNA strands within the nuclei rapidly expand to form a cloud-like structure (e). The cloud-like structure gradually shrinks to form a bipartite architecture composed of central and peripheral regions (f). At this stage, the concentrations of DNA and condensin I are higher in the peripheral region than in the central region. As the DNA volume fraction in the peripheral region increases (g), apparent symmetry breaking occurs, resulting in the appearance of protrusions in the peripheral region (h). Simultaneously, the entire structure becomes more compact. In this final structure of sparklers (c), condensin I is concentrated at the tips of the protrusions where DNA is sparse, while H1.8 is enriched in the other regions where DNA is dense. Examples of micrographs corresponding to (c–h) are shown in Fig. S1.

The time course of sparkler assembly provides further insight (10) (Fig. 1, d–h). Mouse sperm DNA strands incubated in an extract codepeted of Asf1 and topo IIα (Fig. 1 d) swell into a uniform cloud-like structure without dissolving into the external solution over the experimental timescale (Fig. 1 e). The entangled DNA strands are subsequently reorganized into a spherical structure, in which both DNA and condensin I are more dilute in the central region than in the peripheral region (Fig. 1 f). The thickness of the peripheral region increases over time, whereas the size of the central region decreases (Fig. 1 g). Subsequently, the bulk of the DNA is lost from the peripheral region, leaving multiple DNA domains behind: the spherical symmetry of the system is broken (Fig. 1 h). The DNA lost from the peripheral region is reeled into the central region, increasing its size and DNA concentration. Subsequently, the entangled DNA strands adopt a star-like structure, with DNA domains remaining in the peripheral region as protrusions (Fig. 1 c). This series of events implies that the emergence of protrusions is driven by symmetry breaking.

Chromosomal DNA molecules present in the mouse sperm nuclei are entangled and remain entangled when the nuclei are exposed to Xenopus egg extracts depleted of topo IIα. Topoisomerase I is active in the extract, relaxing DNA supercoils if present. Here, we construct a theoretical model to study the biophysical mechanism of symmetry breaking during sparkler assembly by taking into account DNA entanglement. In polymer physics, entanglement is treated as effective crosslinks; entangled DNA strands act as a polymer gel (16,17). Unlike chemical crosslinks, DNA strands can slide through effective crosslinks (18,19,20,21). Our theory treats the central region as an entangled DNA gel, and the peripheral region as a brush of DNA loops extruded by condensin I. Our theory predicts that the radius of the central gel region decreases over time to relieve the DNA tension generated by the loop extrusion process. This in turn increases the DNA concentration in the peripheral region and promotes H1.8 binding. When the DNA concentration becomes sufficiently high, the peripheral region undergoes phase separation into domains of swollen and condensed DNA loops. Swollen DNA loops enriched in condensin I are stabilized by continuous condensin I loading and remain in the peripheral region. In contrast, condensed DNA loops enriched in H1.8 resist condensin I loading and are consequently reeled into the central gel region. Therefore, the phase separation leads to the breaking of spherical symmetry in the system. Our theory reveals a linker histone as a regulator of DNA loop stability extruded by condensin I.

Materials and methods

Model overview

We treat entangled DNA strands in a solution containing linker histone and condensin (Fig. 2). The DNA strands act as electrically neutral flexible polymers composed of Kuhn units of length b because their electric charges are neutralized at physiological salt concentrations (22). The chromosomal DNA strands are extremely long and therefore remain entangled over the experimental timescale (23). We treat DNA entanglement as effective crosslinks, where the DNA strands can slide across crosslinks (18,19,20,21). Linker histone and condensin bind to the DNA strands in a mutually exclusive manner (10,24). We treated linker histone bound to DNA as factors that give rise to the multivalent attractive interaction and condensin bound to DNA as the loop extrusion factor.

Figure 2.

Figure 2

Entangled DNA model. The center is modeled as an entangled DNA gel and the peripheral region is modeled as a brush of DNA loops extruded by condensin, which also extrudes loops within the central gel. These loops are elastically ineffective, while other parts of the entangled DNA network are elastically effective.

The central region of a sparkler intermediate is treated as a spherical, entangled DNA gel, whereas the peripheral region is treated as a brush of DNA loops extruded from the central region by condensin I. While condensin I also generates loops in the central region, these loops do not respond to deformation of the DNA gel and are therefore considered elastically ineffective (17,25). In contrast, the remaining DNA in the central region responds to gel deformation and is considered elastically effective. The numbers of DNA units in each region were determined by loop extrusion dynamics, while the total number N of DNA units in the system is conserved

Nb+Neff+Ngl=N, (1)

where Nb, Neff, and Ngl are the numbers of DNA units in the peripheral, elastically effective chains, and elastically ineffective chains, respectively.

We analyze both cases in which dynamics is limited by loop extrusion and relaxation. Relaxation of polymer gels and brushes is governed by solvent flow (26,27). The structures of the central and peripheral regions are represented by the volume fraction ϕ of DNA, the occupancy αh by linker histone, and the occupancy αc by condensin. The subscripts “c” and “h” indicate condensin and linker histone, respectively.

Free energy

The free energy of the system is given by the sum of the free energy contributions from the central and peripheral regions,

F=Fg+Fb (2)

The subscripts “g” and “b” indicate the central gel and peripheral brush regions, respectively. The free energy is a functional of the occupancy αc by condensin, the occupancy αh by linker histone, and the DNA volume fraction ϕ. The free energy of the central gel region has the form

Fg=∫0rindr4πr2(fent(αc,αh,ϕ)+fsol(αc,αh,ϕ)+fbnd(αc,αh,ϕ)), (3)

where fent(αc,αh,ϕ) is the elastic free energy due to entanglement, fsol(αc,αh,ϕ) is the solution free energy, and fbnd(αc,αh,ϕ) is the binding free energy, all per unit volume. rin is the radius of the central region. The free energy of the peripheral brush region has the form

Fb=∫rinrexdr4πr2(flop(αc,αh,ϕ)+fsol(αc,αh,ϕ)+fbnd(αc,αh,ϕ)), (4)

where flop(αc,αh,ϕ) is the elastic free energy per unit volume due to stretching of DNA loops. rex is the radius of the entire DNA structure, including both the central and peripheral regions. The free energy contributions Fg and Fb are designed to take into account the features of condensin and linker histone by extending the Flory-Rehner theory of polymer gels (28) and the Alexander-de Gennes theory of polymer brushes (29,30), respectively, with an extension based on asymptotic analysis of the Daoud-Cotton theory (31) (see also section S1 and Fig. S2 in supporting material).

The elastic free energy fent(αc,αh,ϕ) per unit volume due to entanglement has the form

fent(αc,αh,ϕ)=32G0ϕeffϕ0(Ne0Ne)2λs2, (5)

which represents the elastic free energy stored by isotropic deformation of the entangled DNA network by the swelling ratio λs from the reference state (32). The reference state is defined as the most relaxed state (15). Equation 5 is an extension of the elastic free energy of the affine tube model, which is the simplest model of entanglement (17,25). The derivation of Eq. 5 is provided previously (32). ϕeff is the volume fraction of elastically effective chains and ϕ0 is the DNA volume fraction in the reference state. Ne0 and Ne are the numbers of DNA units in a subchain between the effective crosslinks in the reference state and after deformation, respectively. The ratio Ne/Ne0 is equal to the fraction Neff/N of DNA units in the elastically effective chains. G0(=ϕ0kBT/(b3Ne0)) is the shear modulus in the reference state, where kB is the Boltzmann constant and T is the absolute temperature. The volume fraction ϕeff of the elastically effective chains has the relationship

ϕeff=ϕ0λs3NeNe0. (6)

The radius rin(=λsrR) is a function of the swelling ratio λs, where rR is the radius of the entire DNA structure in the reference state.

The solution free energy per unit volume has the form (17,27,33):

fsol(αc,αh,ϕ)=kBTb3[(1−(1+αc)ϕ)log(1−(1+αc)ϕ)−χ0ϕ2−χαh2ϕ2−μhαhϕ]. (7)

The first term of Eq. 7 is the mixing free energy, which represents the random motion of the solvent molecules; the second term is the DNA-DNA interaction and the parameter χ0 is the ratio of the magnitude of this interaction to the thermal energy kBT; and the third term is the linker histone-linker histone multivalent interaction and the parameter χ is the ratio of the magnitude of this interaction to the thermal energy kBT. The linker histone-DNA interaction is treated separately (Eq. 8). The fourth term of Eq. 7 is the contribution of the chemical potential μhkBT of linker histone. For simplicity, we assumed that the volumes of DNA units, condensin, and solvents are equal. We neglected the contribution of linker histone to the mixing free energy because their volume is much smaller than the volumes of DNA units and condensin.

The binding free energy per unit volume has the form

fbnd(αc,αh,ϕ)=kBTb3ϕ[αhlogαh+αclogαc+(1−αc−αh)log(1−αc−αh)+ϵhαh+ϵcαc]. (8)

The first, second, and third terms in Eq. 8 are the entropic contributions, and the fourth and fifth terms are the energetic contributions with respect to the binding and unbinding of the linker histone and condensin. kBTϵh and kBTϵc are the binding energy of linker histone and condensin to DNA units, respectively.

The elastic free energy per unit volume in the brush region has the form (29)

flop(αc,αh,ϕ)=32kBTb3(σb2)2ϕ, (9)

where σ(=2nl/(4πrin2)) is the number density of DNA loops in the peripheral brush region and nl is the number of loops in this region. The number density σ depends on the swelling ratio λs in the central region via rin; this represents the connectivity between this and the peripheral regions. The elastic energy in the peripheral region results from the stretching of DNA loops due to the excluded volume interaction between neighboring DNA loops rather than entanglement.

In this article, we take into account the curvature of the interface between the central and peripheral regions only minimally (supporting material, section S1 and Fig. S2 in supporting material). With this treatment, the DNA volume fraction is uniform in both of the central and peripheral regions, which greatly simplifies the model. This treatment is effective for the case in which the brush height h is smaller than the radius rin and, otherwise, it is only an approximation (see also discussion). The DNA volume fraction ϕg in the central region has the form

ϕg=b3(N−Nb)Vin, (10)

where Vin(=4πrin3/3) is the volume of the central region. The volume fraction ϕb in the peripheral region has the form

ϕb=b3Nb4πrin2h. (11)

Chemical potential of linker histone

The minimization of the free energy F with respect to the occupancy αh leads to the equality of the chemical potential of linker histone,

μh=logαh−log(1−αc−αh)+ϵh−2χαhϕ. (12)

The derivation of Eq. 12 is shown in supporting material, section S2.3. Equation 12 is composed of the contributions of entropy (first and second terms), binding energy (third term), and linker histone-linker histone multivalent interaction (fourth term). It can be applied to both of the central and peripheral regions.

Osmotic pressure

The minimization of the free energy F with respect to the DNA volume fraction ϕg leads to the osmotic pressure in the central gel region,

Πin=Πg(ϕg)+ΔΠg, (13)

where the second term results from the fact that the free energy of the peripheral region depends on the DNA volume fraction ϕg in the central region via the radius rin. The derivation of Eq. 13 is shown in supporting material, section S2.6. The first term is the excess osmotic pressure Πg(ϕg) in the central region and the second term is the osmotic pressure ΔΠg that results from the lateral osmotic pressure in the peripheral region.

The excess osmotic pressure Πg(ϕg) has the form

Πg(ϕg)=−G0(NeNe0)−4/3(ϕeffϕ0)1/3+Πsol(ϕg), (14)

where the first term is the elastic stress of the entangled DNA network and the second term is the osmotic pressure of the solution

Πsol(ϕ)b3kBT=−log(1−(1+αc)ϕ)−(1+αc)ϕ−χ0ϕ2−χαh2ϕ2. (15)

The osmotic pressure contribution ΔΠg has the form

ΔΠg=2hrinΠsol(ϕb). (16)

for h<rin/2. Equation 16 represents the projection of the lateral osmotic pressure hΠsol(ϕb) in the radial direction due to the curvature 2/rin of the interface. For h>rin/2, we use

ΔΠg=Πsol(ϕb) (17)

(supporting material, section S1 and Fig. S2 ). The osmotic pressure Πin is balanced by the hydrostatic pressures pin of the solvent

Πin+pin=0. (18)

The minimization of the free energy F with respect to the DNA volume fraction ϕb in the peripheral region leads to the form

−3kBTb3(σb2)2ϕb+Πsol(ϕb)=0. (19)

The derivation of Eq. 19 is shown in supporting material, section S2.5. This equation represents the balance between the elastic stress of the DNA loops (first term) and the osmotic pressures (second term) in the peripheral region. For simplicity, we neglect the contribution of the solvent flow to the conformation of DNA.

Loop extrusion rate and DNA tension

Single-molecule experiments have shown that the extrusion rate v decreases with increasing DNA tension f. For simplicity, we assumed the linear dependence of the extrusion rate

v(f)=v0(1−ffs), (20)

for f<fs and v(f)=0 for f>fs, motivated by the treatment of other molecular motors (34). fs is the stall force of condensin and v0 is the maximum extrusion rate.

Loop extrusion produces tension in the elastically effective chains in the central region because it decreases the number of DNA units in the subchains between effective crosslinks. The relaxation of polymer gels and brushes is governed by solvent flow (26). If DNA translocation by condensin is faster than relaxation, the volumes Vin and Vex of the central and peripheral regions are constant. The DNA tension fex, which is overcome by condensin to extrude DNA units toward the peripheral region, is derived by using the principle of virtual displacement as

fexbΔNb=∂F∂Nb|Vin,VexΔNb. (21)

Equation 21 represents that the work done by extruding ΔNb DNA units from the central to the peripheral regions (left side) is equal to the variation of the free energy due to this process with fixed volume (right side). ΔNb(=2vΔtnl) is the total number of DNA units extruded by nl condensin complexes at the interface between central and peripheral regions during the time period Δt. Thus, DNA tension has the form

fexb=∂∂NbF(ϕg,ϕb,Nb)−Πinb3ϕg. (22)

The second term of Eq. 22 results from the fact that the DNA volume fraction ϕg depends on Nb. We omitted a small term αclogαc+(ϵc−μc)αc for the central region in our numerical calculation.

Similarly, DNA tension, which is overcome by condensin to extrude DNA in the elastically effective part to the elastically ineffective part, is derived as

finb=∂Fg∂Ngl|Nb,Vin (23)

because the number of DNA units in the central region does not change with this process.

The expressions for fex and fin are shown in supporting material, section S2.7 and S2.8. Extrusion from the elastically effective to ineffective regions does not change the DNA volume fraction ϕg in the center; thus, the DNA tension fin results solely from elastic force produced by entanglement. In contrast, extrusion from the central to the peripheral regions decreases the excluded volume interaction in the central region; this slightly reduces the DNA tension fex. Note that fex and fin are instantaneous DNA tensions during extrusion steps before the DNA is relaxed.

Loop extrusion dynamics

The time evolution equation of the number Nbl(=Nb/nl) of DNA units in a loop in the peripheral region has the form:

dNbldt={2vex(t<τex)2vex−q1αcτex(t>τex), (24)

where vex(=v(fex)) is the extrusion rate of the condensin at the interface between the central and peripheral regions, and q1 is the probability that the number of condensin at the interface is unity. The probability q1 is derived as

q1=2v0τexαce2v0τexαc−1, (25)

because condensin arrives at the interface with rate 2v0αc due to the loop extrusion and are unloaded with rate τex−1 (supporting material, section S3.1). The first term of Eq. 24 is the extrusion rate of DNA by the condensin at the interface. The second term of Eq. 24 for t>τex represents that the αc−1 DNA units between condensin complexes are released toward the center by the unloading of all the condensin complexes at a loop anchor with rate q1/τex. We neglect the loop branching because the simultaneous loading of multiple condensin complexes is a rare event (4).

For simplicity, we considered cases in which each elastically ineffective loop in the central region is produced by at most one condensin. The time evolution equation of the number Ngl of DNA units in the elastically ineffective chains has the form:

dNgldt=−Nglτex+2vinngl, (26)

where vin(=v(fin)) is the extrusion rate of condensin that form elastically ineffective loops in the central region and ngl is the number of such loops. The time evolution of the number of elastically ineffective loops has the form

dngldt=−nglτex+konϕc0(1−αh)(1−ϕg)Neff. (27)

Equation 27 is an extension of the kinetic equation for binding and unbinding (35). The solution of Eqs. 26 and 27 has the form

Ngl(t)=konϕc∫0tdt′(∫t′tdt″2vin(t″))(1−αh(t′))(1−ϕg(t′))Neff(t′)e−t−t′τex
≈2konϕc0(1−αh(t))(1−ϕg(t))vin(t)τex2T(t)Neff(t) (28)

with the time factor

T(t)=1−(1+tτex)e−tτex. (29)

To derive the last form of Eq. 28, we assumed that the functions vin, αh, ϕg, and Neff, change slower than τex−1 (supporting material, section S3.2). This assumption is effective for t>τex and is only an approximate treatment for t<τex.

Condensin loading/unloading kinetics

Both topological loading and loop extrusion are driven by ATP hydrolysis. The time evolution equation of the occupancy αc by condensin in the central gel has the form

ddtαc=konϕc0(1−αh)(1−ϕg)−αcτex. (30)

The first term of Eq. 30 is the loading rate of a condensin and kon is the rate constant that accounts for this process, and the second term is the unloading rate of condensin. ϕc0 is the volume fraction of condensin in the external solution. The factors (1−αh) and (1−ϕg) in the first term of Eq. 30 account for the suppression of the loading of a condensin by linker histones and the osmotic pressure, respectively (supporting material, sections S2.2 and S2.4).

The time evolution of occupancy αc in the loop brush region has the form

ddtαc=konϕc0(1−αh)(1−ϕb)−αcτex−2v0Nbαc. (31)

The first term of Eq. 31 is the loading rate and the second term is the unloading rate of a condensing, and the third term is the rate at which a condensin is translocated to the interface between the central and peripheral regions via loop extrusion.

To derive Eqs. 30 and 31, we assume that condensin complexes loaded at the peripheral region do not penetrate the central region, motivated by the fact that condensin is most abundant at the interface during the assembly of the peripheral region is assembled. We derive the occupancy αc by condensin for the steady state, dαcdt=0.

Relaxation dynamics

Because the DNA volume fraction is larger in the peripheral region than in the central region during sparkler assembly, we assume that relaxation is governed by solvent flow across the peripheral region. For cases in which the loop extrusion process is faster than the relaxation, the time evolution of the volume of the central region is given by (26,27,36):

ddtVin=−4πrin2κ(ϕb)pinh. (32)

The solvent flow is proportional to the pressure gradient pin/h with the Darcy coefficient κ(ϕb). The mesh size of the peripheral brush region is given by the average distance σ−1/2 between the loop. We thus use the Darcy coefficient of the form (26,27,36)

κ(ϕb)=κ0(1−ϕb)2σb2 (33)

by using a constant κ0(≈b2/η) inversely proportional to the solvent viscosity η (36).

Phase separation in the peripheral region

In general, Eq. 19 has two stable solutions, where the volume fraction of one solution (swollen loops) is smaller than that of the other (condensed loops). When the peripheral region is composed of coexisting domains of such loops, the free energy of the peripheral has the form (27,37):

Fb=fb(ϕb1,σ1)h1A1+fb(ϕb2,σ2)A2h2+μlp(σ1A1+σ2A2)+Πsol(h1A1+h2A2), (34)

where fb(ϕb,σ)(=flop(αc,αh,ϕb,σ)+fsol(αc,αh,ϕb)+fbnd(αc,αh,ϕb)) is the free energy per unit volume in the peripheral region. For simplicity, we omitted the explicit description of the dependence of the free energy fb on the occupancies αc and αh. The subscripts 1 and 2 indicate the values of the swollen and condensed loops, respectively. A1 and A2 are the areas occupied by the domains of swollen and condensed loops, respectively. The first and second terms in Eq. 34 are the free energy contributions from the domains of the swollen and condensed loops, respectively. The third term is the Lagrange multiplayer to ensure that numbers of loops are constant and the fourth term is the contribution of osmotic pressure.

Minimizing the free energy with respect to σ1, σ2, A1, and A2 leads to the relationships

Πsol=ϕb12∂∂ϕb1(fb(ϕb1,σ1)ϕb1)+σ1∂∂σ1fb(ϕb1,σ1)=ϕb22∂∂ϕb2(fb(ϕb2,σ2)ϕb2)+σ2∂∂σ2fb(ϕb2,σ2) (35)
μlpNb=fb(ϕb1,σ1)+Πosmϕb1=fb(ϕb2,σ2)+Πosmϕb2. (36)

Equations 35 and 36 represent the balance of lateral osmotic pressure and chemical potential of the loops between the domains of the swollen and condensed loops (see also Eq. 14). In general, these domains are not at the thermodynamic equilibrium because ATP hydrolysis is involved in the loading and loop extrusion by condensin. Our treatment here is thus only an approximation of phase separation in the peripheral brush region.

Parameter estimates

As our model considers both loop extrusion and relaxation dynamics in a system composed of a central entangled gel region and a peripheral brush region, the number of parameters involved is rather large. Our model is described by 10 parameters: the rescaled shear modulus G0b3/(ϕ01/3kBT), the interaction parameters χ0 and χm, the binding free energy, ϵc−μc and ϵh−μh, the number of extruded DNA units 2v0τex, the rescaled stall force fsb/(kBT) of condensin, the occupancy αc0(=konτex) by condensin for ϕ=0 and αh=0, and combinations of parameters

ηm=134π(2nl)32N (37)
nh0=2v0τexnlN. (38)

Equation 37 is the ratio of the volume 4πrm3/3, of the central gel region to the total volume Nb3 of the system when all the solvent in the system is excluded, 4πrm2=2nlb2, and has been used to describe a system of similar geometry (32). The radius r0 of the dried DNA strands,

4π3r03=Nb3, (39)

represents the length of the system. Equation 38 is the ratio of DNA units extruded during a time period τex to the total N of DNA units in the system. The parameters used in numerical calculations are listed in Table 1.

Table 1.

Parameter values

Parameter Physical meaning Estimate Used
G0b3/(ϕ013kBT) shear modulus NA 0.01
χ0 DNA-DNA interaction 0.44 0.5
χm histone-histone interaction NA 2.8
ϵc−μc DNA-condensin binding energy NA 0.0
ϵh−μh DNA-histone binding energy NA 1.9
2v0τex extruded DNA length 130 100
fsb/(kBT) stall force 10 10
ηm ratio of bulk and interface 0.05 0.01
nh0 fraction of DNA units in loops 0.3 0.3
αc0 condensin occupancy 0.03–0.3 0.3

At physiological salt concentrations, the length b of each DNA unit is 100 nm (∼300 bp) (22). The excluded volume 12(1−2χ0)b3 of DNA units is ∼0.06b3 (38). With this small excluded volume, the correlation is not significant and our mean field treatment is effective (39). The values of the rescaled shear modulus G0b3/(kBT), interaction parameter χm, and binding energies, ϵc and ϵh, are not available. Thus, we set their values to be large enough for linker histone to condense the DNA strands (15).

Single-molecule experiments suggest that yeast condensin extrudes DNA 2vτex by 40 kbps (≈130 DNA units) with a stall force of fs ∼0.1 pN (8). The number of condensin loaded on DNA is estimated as 1 per 1–10 kbps (40,41), which amounts to αc0 ∼0.03–0.3. The timescale τs, with which condensin extrudes DNA units toward the steady state, is estimated as 2v0nlτs≈N. By using the experimental timescale τs ∼ 30 min (10) and the extrusion time τex ∼ 10 min of budding yeast condensin (8), the ratio nh0 is estimated as 0.3. The number of DNA loops in the peripheral region is nl ∼ 3 × 104, because the number of units in mouse sperm DNA (∼3 Gbps) is N ∼ 1 × 107 and 2v0τex ∼ 100 (Eq. 38). By using these values, the ratio ηm is estimated to be 0.05 (Eq. 37).

Results

Relaxation of the central gel region drives phase separation of the peripheral brush region

We first treat cases in which relaxation caused by solvent flow is faster than loop extrusion, pin=0. Our model predicts DNA fractions in the central and peripheral regions, DNA volume fraction, condensin occupancy, and linker histone occupancy over time (Fig. S3). The fraction Nb/N of DNA units in the peripheral region increases with time (solid line, Fig. 3 a). DNA tension is greatly reduced by relaxation and does not significantly decelerate loop extrusion of condensin because of the relatively large stall force of condensin (solid and dotted lines, Fig. 3 a). Solvent flows out from the central region to relax the DNA tension; thus, the radius rin of the central gel region decreases with time (red line, Fig. 3 b). This decreases the distance between DNA loops in the peripheral region. The DNA volume fraction ϕg in the peripheral region thus increases over time (dark green line, Fig. 3 c).

Figure 3.

Figure 3

Loop extrusion dynamics with local equilibrium. (a) Fraction of DNA units Nb/N in the peripheral brush region as a function of time t. The dotted line represents the case of ideal loop extrusion without deceleration by DNA tension and condensin unloading. (b) The radius rin of the central gel region (red) and the height h of the peripheral brush region (blue) are shown as a function of time t. These quantities are rescaled by the length scale r0 (Eq. 37). (c) DNA volume fractions in the center (light green) and peripheral region (deep green) are shown as a function of time t. ϕbbi1 and ϕbsp1 are the values of the DNA volume fraction at the binodal and spinodal lines of the peripheral region, respectively. The uniform swollen state is stable for ϕb<ϕbbi1 (solid lines) and quasistable for ϕbbi1<ϕb<ϕbsp1 (broken line) in the peripheral region. The peripheral region can show phase separation at a volume fraction in the latter window. ϕgsp1 is the DNA volume fraction at the spinodal line of the central gel. Values of parameters used for these calculations are summarized in Table 1.

DNA loops produced in the early stages are enriched in condensin and are relatively swollen (magenta line, Fig. 4 a). The probability of the binding of linker histones to the DNA loops increases with the DNA volume fraction in the peripheral region. When the DNA volume fraction in the peripheral region becomes sufficiently large, DNA loops are enriched with linker histone and condense owing to multivalent interactions. Osmotic pressure and mutual exclusion between condensin and linker histone act to suppress the loading of condensin onto condensed DNA loops rich in linker histone. Our theory predicts that the peripheral region undergoes phase separation into two domains: swollen DNA loops mainly occupied by condensin and condensed DNA loops mainly occupied by linker histone at a DNA volume fraction larger than the binodal value ϕbbi1. This phase separation is so-called multivalent bridging-induced phase separation (42). Because the uniform swollen state of the peripheral region is still metastable for ϕbbi1<ϕb<ϕbsp1, the phase separation probably does not occur immediately at ϕb=ϕbbi1, but after a lag. During this lag, loop extrusion continues to increase the DNA volume fraction ϕb. The DNA volume fraction ϕb at which phase separation occurs is determined by the nucleation kinetics of domains of condensed DNA loops.

Figure 4.

Figure 4

Peripheral brush region phase separation. (a) Lateral osmotic pressure Πsol(ϕb) in the brush region as a function of the DNA volume fraction ϕb in the peripheral region. The magenta line represents osmotic pressure when the uniformly swollen state is stable (solid) and quasistable (broken) in the peripheral brush region (see also Fig. 3). The binodal line is shown by the purple line. The black dotted curve shows osmotic pressure calculated with fixed DNA fraction, Nb/N= 0.62 and t/τex = 2.36, at ϕb=ϕbbi1 (Fig. 3c). ϕbbi1 and ϕbbi2 are the DNA volume fractions at the binodal line. (b) Fraction s of the domains of swollen loops rich in condensin complexes as a function of DNA volume fraction ϕb at which phase separation occurs. Values of parameters used for these calculations are summarized in Table 1.

The fraction s of swollen domains depends on the DNA volume fraction ϕb in the peripheral region at which phase separation occurs (Fig. 4 b). The fraction s is determined by the condition

sσsw+1−sσcon=1σ, (40)

where σswb2(=ϕbbi1Πsolbib3/(3kBT)) and σconb2(=ϕbbi2Πsolbib3/(3kBT)) are the area densities of swollen and condensed DNA loops, respectively, with Πsolbi=Πsol(ϕbbi1)=Πsol(ϕbbi2). Equation 40 is applicable to cases in which the free energy cost arising from local shear deformation of the interface due to the inequality of area per loop, σsw≠σcon, is negligible.

Condensed loops enriched in H1.8 are reeled into the central gel region, triggering symmetry breaking

We analyze the dynamics of DNA loops after phase separation. Steady loading of condensin is necessary to form a stable DNA loop. Swollen DNA loops, mainly occupied by condensin, are thus stable (magenta line, Fig. 5 a). In contrast, the number of DNA units in the condensed DNA loops mainly occupied by linker histone rapidly decrease with time because loading of condensin is suppressed by the linker histone (cyan line, Fig. 5 b). This results in breaking the spherical symmetry of the system. Therefore, our theory predicts that symmetry breaking during the assembly of the sparkler results from the phase separation in the peripheral region.

Figure 5.

Figure 5

Time evolution after peripheral brush region phase separation. (a) Numbers of units in swollen loops rich in condensin complexes (magenta) and condensed loops rich in linker histones (cyan) as a function of time Δt after phase separation. (b) DNA volume fraction ϕg (green), condensin occupancy αc (black), and linker histone occupancy αh (orange) in the central gel region as a function of time Δt. (c) Radius rin of the central gel and fraction nb of DNA units in the peripheral brush as a function of time Δt. These are cases in which the phase separation of the peripheral brush region occurs at nh = 0.74 and t/τex = 2.9, corresponding to s = 0.4. Values of parameters used for these calculations are summarized in Table 1.

We address the dynamics of the system after phase separation using the affine approximation, in which σswrin2 and σconrin2 are constant, and by neglecting the solvent exchange between the central gel region and the external solution. These approximations are effective for cases in which the decrease in DNA units in the condensed loops is faster than the solvent flow from the central region to the external solution (Fig. 5). With these approximations, the DNA volume fraction ϕg in the central gel region increases over time because the DNA in the condensed loops at the peripheral region are reeled into the central region (green line, Fig. 5 b), in turn attracting linker histone to the central gel region (orange line, Fig. 5 b). The occupancy αh by linker histone increases continuously when the binding energy ϵh−μh is smaller than a critical value, ϵhc−μhc(=2+log(1−αc0(1−2/χm))). The linker histone occupancy αh can jump at a value of DNA volume fraction ϕg for ϵh−μh>ϵhc−μhc (Fig. S4). The radius rin of the central gel region increases with time because the number of DNA units in the central gel region increase. Linker histone bound to DNA suppress the production of new loops due to mutual exclusion between linker histone and condensin. The features predicted by our theory are consistent with those of experiments (10).

The fast relaxation approximation can be effective when both the central and peripheral regions are in a swollen state

When loop extrusion is faster than relaxation, condensin extrudes DNA units to the peripheral region with fixed volume of the central region in the short timescales (Fig. S5). Because DNA tension is not relaxed in these timescales, it greatly decelerates loop extrusion. The DNA volume fraction in the peripheral region is approximately constant after initial swelling because the volume of the central region is constant (green line, Fig. S5 d). Relaxation occurs only after loop extrusion is stopped by DNA tension, dNbldt=0 (Fig. S6). Elastic stress arising from DNA tension is balanced by the hydrostatic pressure pin of the solvent in the central gel region, which creates the gradient of hydrostatic pressure across the peripheral region. This gradient drives solvent flow from the central region toward the external solution to relax the DNA tension. Thus, the radius rin of the central gel region decreases over time (red line, Fig. 6 a). The DNA volume fraction ϕb in the peripheral region increases during relaxation and the occupancy by linker histone increases over time (Fig. 6 b). Our theory predicts that the peripheral brush region undergoes phase separation into swollen DNA loops mainly occupied by condensin and condensed DNA loops mainly occupied by linker histone at DNA volume fractions in the window ϕbbi1<ϕb<ϕbsp1, similarly to the case of fast relaxation (Fig. 6 b). DNA in condensed loops is rapidly reeled into the central gel region. Therefore, our theory predicts symmetry breaking also for the case of slow relaxation.

Figure 6.

Figure 6

Dynamics limited by relaxation. (a) Radius rin of the central gel region (red) and the thickness h of the peripheral brush region (blue) as a function of time t, rescaled by the length scale r0 (Eq. 37). (b) DNA volume fractions in the center (light green) and peripheral region (deep green) as a function of time t. (c) Hydrostatic pressure pin as a function of time t, rescaled by the relaxation time τr (Eq. 39). ϕbbi1 is the DNA volume fraction at the binodal line. The uniform swollen state is stable for ϕb<ϕbbi1 (solid line) and is only quasistable for ϕbbi1<ϕb<ϕbsp1 (broken line) in the peripheral region. The peripheral region can show phase separation at a volume fraction in the latter window. Values of parameters used for these calculations are summarized in Table 1.

In contrast to many passive systems, the hydrostatic pressure pin increases with time during relaxation because loop extrusion by condensin reinforces DNA tension (Figs. 6 c and S6 a). Nevertheless, solvent flow decreases over time because the thickness h of the peripheral region increases (blue line, Fig. 6 a).

Our theory predicts that the relaxation time has the form

τr=b3r02κ0kBT. (41)

The Darcy constant κ0(≈b2/η) is estimated as 5 × 10−13 (m2/Pa s) by using the viscosity of Xenopus egg extracts, η≈ 20 mPa s (43). The relaxation time τr ∼ 1.5 min is thus ∼10-fold shorter than that of the loop extrusion τex≈ 10 min; the relaxation is rather rapid. The relaxation time τex can be longer than the timescale, with which the DNA in condensed loops is reeled to the central region after the phase separation (cyan line, Fig. 5 a), consistent with our assumption in the previous section.

Relaxation time increases with increasing thickness and DNA volume fraction in the peripheral brush region. With fast relaxation, these quantities increase with time (blue line, Fig. 3 b and dark green line, Fig. 3 c). One may think that the relaxation per unit step of loop extrusion can eventually become slower than that of the timescale of the unit extrusion step at longer timescales. By using Eq. 32, the relaxation time Δτr per unit step of loop extrusion is derived as

Δτr=hκ(ϕb)∂rin∂Nb∂Π∂Nb|Vin, (42)

where ∂Π∂Nb|Vin represents the partial derivative of Π with respect to Nb with the constraint of constant volume Vin. The derivation of Eq. 42 is shown in supporting material, section S4.

Equation 42 suggests that the relaxation time Δτr per unit step can be much longer than the step time v0−1 of loop extrusion when all DNA loops in the peripheral region are condensed and ϕb≈1. Solvent flow within the central gel region can also be prohibitively slow when the central gel region is condensed, ϕg≈1. Thus, the volume Vin of the central gel region is approximately constant if either or both the central and peripheral regions are condensed. Indeed, it is well known that the volume-phase transition of a polymer gel is a very slow process for the same reason (27,44,45,46). With the parameter values shown in Table 1, the relaxation time Δτr (<0.01 τr) per unit step is estimated as ∼10-fold less than the time between the steps of loop extrusion, v0−1≈ 0.02 τex before the phase separation (Fig. 7). However, for cases in which the shear modulus G0b3/(kBT) is small, the relaxation time Δτr per unit extrusion step can be larger than the time v0−1 per unit extrusion step at longer timescales (supporting material, section S5.10).

Figure 7.

Figure 7

Relaxation time per unit step of loop extrusion. The relaxation time Δτr per unit step is shown as a function of the fraction Nb/N of DNA units in the peripheral brush region (Eq. 39). The uniform swollen state is stable and only quasistable in the regions shown by the solid and broken lines, respectively. Values of parameters used for these calculations are summarized in Table 1.

Discussion

The sparkler is a highly characteristic chromatin structure assembled on nucleosome-free, entangled DNA in Xenopus egg extracts. In this study, we have developed a theory that predicts the mechanism of symmetry breaking observed during sparkler assembly. According to this theory, the size of the central gel region decreases to relieve the DNA tension produced by condensin I-mediated loop extrusion occurring at the central-peripheral interface (Fig. 8 a). This leads to a higher DNA volume fraction in the peripheral region, thereby increasing the likelihood of linker histone H1.8 binding to this region. When the DNA volume fraction in the peripheral region becomes sufficiently high, competition between condensin I and H1.8 drives phase separation of the peripheral region into two domains: swollen loops enriched in condensin I and condensed loops enriched in H1.8 (Fig. 8 b). Experimentally, protrusions with condensin I enriched at their tips are irregularly positioned in the final structure of a sparkler (10), suggesting a hallmark of phase separation. DNA loops enriched in condensin I are relatively stable, whereas those enriched in H1.8 are rapidly reeled into the central gel region, as H1.8 competes with condensin I and suppresses its loading onto these loops, thereby breaking the spherical symmetry of the system (Fig. 8 b). Thus, our theory predicts that H1.8 regulates the stability of DNA loops extruded by condensin I and promotes their reorganization. Other variants of linker histones likely exhibit similar functions, provided that they compete with condensin I and that their multivalent interactions are sufficiently strong. H1.8 also suppresses the formation of new DNA loops in the central gel region after reorganization. In summary, the series of events observed during sparkler formation is driven by competition between condensin I and H1.8, as well as DNA tension generated through loop extrusion. Importantly, these two key factors exert their greatest effects on the entangled, nucleosome-free DNA substrates used in the experiments (10).

Figure 8.

Figure 8

A proposed mechanism of symmetry breaking during sparkler formation. (a) DNA tension is generated in the central region by condensin I-mediated loop extrusion occurring in the peripheral region (middle). Solvent flow from the center to the surrounding solution reduces the central volume, thereby alleviating DNA tension (right). This in turn increases the DNA volume fraction in the peripheral region. (b) As the DNA volume fraction in the peripheral region increases, H1.8 is attracted to this region. The peripheral region then undergoes phase separation into two domains: swollen loops rich in condensin I and condensed loops rich in the linker histone H1.8 (middle). The DNA in the condensed loops is rapidly reeled into the central region (right).

After symmetry breaking, what mechanism drives the subsequent morphological changes into the unique sparkler structure containing protrusions? One possibility is that attractive multivalent interactions between H1.8 and DNA generate surface tension at the interface between the central gel region and the surrounding solution, whereas the swollen loop domains are not subject to such surface tension. Although surface tension acting on a uniform, closed surface stabilizes spherical shapes to minimize surface area, surface tension acting on a non-uniform surface can produce more complex shapes. One of such examples can be observed in soap films suspended in frames, which form catenoids (47). Multicomponent lipid membranes also form heterogeneous surfaces and exhibit budding, although the primary driving force in this case is heterogeneity in bending rigidity (48,49). Heterogeneous surface tension arising from phase separation likely plays a key role in the appearance of protrusions. The size of the swollen loop domains is likely determined by a combination of factors: condensin I-mediated loop extrusion at the interface, loop branching resulting from the simultaneous loading of multiple condensin I complexes, and mechanical tension at the three-phase contact line, where the swollen brush phase, condensed gel phase, and surrounding solution converge. Our current theory lays a foundation for future studies aimed at better understanding how protrusions form.

This study represents the first attempt to provide a theoretical framework for the formation of a recently described peculiar chromatin structure. As discussed below, however, we have made several assumptions to simplify our current model of symmetry breaking. First, we assumed that the DNA volume fraction ϕb is uniform in the peripheral region. For cases in which the thickness h of the peripheral region is larger than the radius rin of the central region, the fact that the DNA volume fraction ϕb decreases with the distance r from the center can be significant (31). The loading rate of condensin thus increases with the distance (Eq. 31) and it decreases the probability of the branching of DNA loops. The curvature effect of the osmotic pressure contribution ΔΠg vanishes for h>rin/2 (Eq. 17). These elements are consistent with our model. However, it is necessary to take into account the non-uniform DNA volume fraction to predict the dynamics of phase separation in the peripheral region. Second, we did not incorporate the relaxation dynamics of DNA arising from loop extrusion, although one can also take it into account using recent theoretical frameworks (50,51,52). Third, we assumed a linear relationship between the loop extrusion rate and DNA tension, although one could also use a fitted rate-tension curve derived from experimental data (8) or extend the two-state model of molecular motors (53). Fourth, we assumed that condensin I loaded onto the peripheral brush region does not penetrate the central gel region. This assumption was based on the observation that, over relatively short timescales during which the entangled DNA strands remain spherical, condensin I is most concentrated at the interface between the central and peripheral regions (10). We speculate that this is because DNA loops associated with condensin I are repelled by the osmotic pressure exerted by the central region. Fifth, we used free energy minimization, which is valid at the thermodynamic equilibrium, to predict phase separation in the peripheral region. This leads to equalities in the chemical potential and osmotic pressure between the two coexisting domains, as described in Eqs. 35 and 36, respectively. Sixth, we neglected the branching of DNA loops in the peripheral brush region because it is a rare event that becomes significant only over longer timescales (5). Finally, we assumed that DNA loops in the peripheral region consist of an equal number of DNA units. Although incorporating these factors would make the theory more quantitative, we expect that the underlying physics would remain unchanged.

Although the number of parameters involved in our model may appear large, most of them are, in principle, experimentally measurable. We derive the asymptotic analytical expressions in supporting material, section S5 and discuss the dependence of the concentration ϕc0 of condensin complexes, the shear modulus G0b3/(ϕ0kBT), and the chemical potential μh of linker histone in supporting material, section S6. Briefly, when the condensin occupancy αc0 is too small, the loops in the peripheral region are unstable or reach the steady state before phase separation occurs (Fig. S7). The ratio ϕg/ϕb of the DNA volume fractions decreases with decreasing the shear modulus G0b3/(ϕ013kBT) (see Fig. S9). Experimentally, the DNA volume fraction ϕg in the central region is smaller than that in the peripheral region ϕb, implying that the shear modulus of the entangled DNA strands is rather small. Theoretically, the uniformly swollen peripheral region becomes only quasistable at smaller DNA volume fraction ϕbbi1 as the chemical potential μh, which is the logarithm of the volume fraction of linker histone in the extract, increases because it enhances the DNA-DNA interaction via linker histone (Fig. S10). When the chemical potential μh is larger than a threshold value μhsp1, the central region becomes a condensed state from the beginning and a linker histone that occupies DNA suppresses the loading of condensed phase (magenta line, Fig. S11). In contrast, when the chemical potential μh is smaller than another threshold value μhsp2, the condensed state of the central region is unstable against the swollen state and the central region may not be condensed even after the reeling of DNA in the peripheral region (cyan line, Fig. S11). However, the technical barrier to titrate H1.8 is somewhat higher than that to titrate condensin I because the extract contains other linker histone variants.

It would be worthwhile to compare our current model of sparkler formation with our previous model of bean formation (32). The bean, a compact, round-shaped structure, is formed in Xenopus egg extracts in which topo IIα is depleted and endogenous condensins are replaced with a condensin I mutant defective in loop extrusion (11). The bean consists of a central core, where DNA and the mutant condensin I are concentrated, and a peripheral halo, where DNA is sparsely distributed. Notably, core histones are enriched in the peripheral halo, suggesting a competition between condensin I and nucleosomes. Like beans, the spherical precursors of sparklers before symmetry breaking also exhibit bipartite organization, although the spatial distributions of histones and condensin I are inverted between the two structures. A key difference between the two structures lies in their underlying assembly mechanisms: the halo of beans is formed through the free energy of nucleosome assembly (32), whereas the precursor of sparklers is assembled by condensin I-mediated active loop extrusion (this study). Despite this difference, competition between histones and condensin I likely serves as a unifying principle in both cases.

One could argue that our present theory on sparkler assembly and our previous theory on bean formation (32) are not directly relevant to our understanding of mitotic chromosome assembly because the two structures are assembled only under specialized conditions in Xenopus egg extracts. In thermodynamics, a phase diagram defines the conditions under which different states of a system emerge and serves as a powerful tool for understanding the properties of systems composed of many molecules. From a broader perspective, mitotic chromosomes, along with structures such as sparklers and beans, can be viewed as parts of a phase diagram of chromatin organization defined by the activities of essential factors. Further exploration of such a phase diagram is likely to elucidate how condensins assemble mitotic chromosomes through their interactions with other essential factors in a crowded environment.

Data and code availability

The Mathematica file “antennamodelVer155_doublelayer(brushcorrection).nb” used for the numerical calculations is available in figshare with the identifier (https://doi.org/10.6084/m9.figshare.28900841).

Acknowledgments

This work was supported by JSPS KAKENHI grant nos. JP20H05934 (to T.Y.), JP24K06969 (to T.Y.), JP23K23815 (to K.S.), and JP20H05938 (to T.H.). T.Y. thanks John Marko (Northwestern University) and Anton Goloborodko (IMBA) for fruitful discussion.

Author contributions

All authors designed the research, secured research funding, and wrote or revised the manuscript. T.Y. developed the physical model and performed analytical and numerical calculations. K.S. performed background experiments and image analysis. T.H. supervised the project.

Declaration of interests

The authors declare no conflict of interest.

Editor: Catherine Musselman.

Footnotes

Supporting material can be found online at https://doi.org/10.1016/j.bpj.2025.09.002.

Supporting material

Document S1. Figures S1–S11
mmc1.pdf (1.7MB, pdf)
Document S2. Article plus supporting material
mmc2.pdf (2.7MB, pdf)

References

  • 1.Hirano T. Condensin-based chromosome organization from bacteria to vertebrates. Cell. 2016;164:847–857. doi: 10.1016/j.cell.2016.01.033. [DOI] [PubMed] [Google Scholar]
  • 2.Alipour E., Marko J.F. Self-organization of domain structures by DNA-loop-extruding enzymes. Nucleic Acids Res. 2012;40:11202–11212. doi: 10.1093/nar/gks925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Naumova N., Imakaev M., et al. Dekker J. Organization of the mitotic chromosome. Science. 2013;342:948–953. doi: 10.1126/science.1236083. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Goloborodko A., Marko J.F., Mirny L.A. Chromosome compaction by active loop extrusion. Biophys. J. 2016;110:2162–2168. doi: 10.1016/j.bpj.2016.02.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Goloborodko A., Imakaev M.V., et al. Mirny L. Compaction and segregation of sister chromatids via active loop extrusion. eLife. 2016;5 doi: 10.7554/elife.14864. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Gibcus J.H., Samejima K., et al. Dekker J. A pathway for mitotic chromosome formation. Science. 2018;359 doi: 10.1126/science.aao6135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Terakawa T., Bisht S., et al. Greene E.C. The condensin complex is a mechanochemical motor that translocates along DNA. Science. 2017;358:672–676. doi: 10.1126/science.aan6516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Ganji M., Shaltiel I.A., et al. Dekker C. Real-time imaging of DNA loop extrusion by condensin. Science. 2018;360:102–105. doi: 10.1126/science.aar7831. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Shintomi K., Inoue F., et al. Hirano T. Mitotic chromosome assembly despite nucleosome depletion in Xenopus egg extracts. Science. 2017;356:1284–1287. doi: 10.1126/science.aam9702. [DOI] [PubMed] [Google Scholar]
  • 10.Shintomi K., Hirano T. Guiding functions of the C-terminal domain of topoisomerase IIα advance mitotic chromosome assembly. Nat. Commun. 2021;12:2917. doi: 10.1038/s41467-021-23205-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kinoshita K., Tsubota Y., et al. Hirano T. A loop extrusion-independent mechanism contributes to condensin I-mediated chromosome shaping. J. Cell Biol. 2022;221 doi: 10.1083/jcb.202109016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Gibbs E.B., Kriwacki R.W. Linker histones as liquid-like glue for chromatin. Proc. Natl. Acad. Sci. USA. 2018;115:11868–11870. doi: 10.1073/pnas.1816936115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Gibson B.A., Doolittle L.K., et al. Rosen M.K. Organization of chromatin by intrinsic and regulated phase separation. Cell. 2019;179:470–484.e21. doi: 10.1016/j.cell.2019.08.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Shimazoe M.A., Philips C., et al. Maeshima K. Linker histone H1 functions as a liquid-like glue to organize chromatin in living human cells. bioRxiv. 2025 doi: 10.1101/2025.03.05.641622. Preprint at. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Yamamoto T., Schiessel H. Loop extrusion driven volume phase transition of entangled chromosomes. Biophys. J. 2022;121:2742–2750. doi: 10.1016/j.bpj.2022.06.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Edwards S.F. The statistical mechanics of polymerized material. Proc. Phys. Soc. 1967;92:9–16. doi: 10.1088/0370-1328/92/1/303. [DOI] [Google Scholar]
  • 17.Rubinstein M., Colby R. Oxford Univ. Press; NY: 2003. Polymer Physics. [Google Scholar]
  • 18.Ball R.C., Doi M., et al. Warner M. Elasticity of entangled networks. Polymer. 1981;22:1010–1018. doi: 10.1016/0032-3861(81)90284-6. [DOI] [Google Scholar]
  • 19.Masubuchi Y. Elasticity of randomly cross-linked networks in primitive chain network simulations. J. Soc. Rheol. Jpn. 2021;49:73–78. doi: 10.1016/0032-3861(81)90284-6. [DOI] [Google Scholar]
  • 20.Likhtman A.E. Single-chain slip-link model of entangled polymers: Simultaneous description of neutron spin-echo, rheology, and diffusion. Macromolecules. 2005;38:6128–6139. doi: 10.1021/ma050399h. [DOI] [Google Scholar]
  • 21.Uneyama T., Masubuchi Y. Multi-chain slip-spring model for entangled polymer dynamics. J. Chem. Phys. 2012;137 doi: 10.1063/1.4758320. [DOI] [PubMed] [Google Scholar]
  • 22.Marko J.F., Siggia E.D. Stretching DNA. Macromolecules. 1995;28:8759–8770. doi: 10.1021/ma00130a008. [DOI] [Google Scholar]
  • 23.Rosa A., Everaers R. Structure and dynamics of interphase chromosomes. PLoS Comput. Biol. 2008;4 doi: 10.1371/journal.pcbi.1000153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Choppakatla P., Dekker B., et al. Funabiki H. Linker histone H1.8 inhibits chromatin binding of condensins and DNA topoisomerase II to tune chromosome length and individualization. eLife. 2021;10 doi: 10.7554/elife.68918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Yamamoto T., Campbell J.A., et al. Rubinstein M. Scaling theory of swelling and deswelling of polymer networks. Macromolecules. 2022;55:3588–3601. doi: 10.1021/acs.macromol.1c02553. [DOI] [Google Scholar]
  • 26.Doi M. Gel dynamics. J. Phys. Soc. Jpn. 2009;78 doi: 10.1143/JPSJ.78.052001. [DOI] [Google Scholar]
  • 27.Doi M. Oxford Univ. Press; Oxford, UK: 2013. Soft Matter Physics. [Google Scholar]
  • 28.Flory P.J., Rehner J. Statistical mechanics of cross-linked polymer networks II. Swelling. J. Chem. Phys. 1943;11:521–526. doi: 10.1063/1.1723792. [DOI] [Google Scholar]
  • 29.Alexander S. Adsorption of chain molecules with a polar head: a scaling description. J. Phys. France. 1977;38:983–987. doi: 10.1051/jphys:01977003808098300. [DOI] [Google Scholar]
  • 30.de Gennes P.G. Conformations of polymers attached to an interface. Macromolecules. 1980;13:1069–1075. doi: 10.1021/ma60077a009. [DOI] [Google Scholar]
  • 31.Daoud M., Cotton J.P. Star shaped polymers: a model for the conformation and its concentration dependence. J. Phys. France. 1982;43:531–538. doi: 10.1051/jphys:01982004303053100. [DOI] [Google Scholar]
  • 32.Yamamoto T., Kinoshita K., Hirano T. Elasticity control of entangled chromosomes: crosstalk between condensin complexes and nucleosomes. Biophys. J. 2023;122:3869–3881. doi: 10.1016/j.bpj.2023.08.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.de Gennes P.G. Cornell Univ. Press; Ithaca, NY: 1979. Scaling Concepts in Polymer Physics. [Google Scholar]
  • 34.Korn C.B., Klumpp S., et al. Schwarz U.S. Stochastic simulations of cargo transport by processive molecular motors. J. Chem. Phys. 2009;131 doi: 10.1063/1.3279305. [DOI] [PubMed] [Google Scholar]
  • 35.Langmuir I. The adsorption of gases on plane surfaces of glass, mica, and platinum. J. Am. Chem. Soc. 1918;40:1361–1403. doi: 10.1021/ja02242a004. [DOI] [Google Scholar]
  • 36.de Cagny H.C.G., Vos B.E., et al. Bonn D. Porosity governs normal stress in polymer gels. Phys. Rev. Lett. 2016;117 doi: 10.1103/PhysRevLett.117.217802. [DOI] [PubMed] [Google Scholar]
  • 37.Safran S.A. Westview Press; Boulder, CO: 2003. Statistical Thermodynamics of Surfaces, Interfaces, and Membranes. [Google Scholar]
  • 38.Bracha D., Bar-Ziv R.H., et al. Pincus P.A. Entropy-driven collective interactions in DNA brushes on a biochip. Proc. Natl. Acad. Sci. USA. 2013;110:4534–4538. doi: 10.1073/pnas.1220076110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Schaefer D.W., Joanny J.F., Pincus P. Dynamics of semiflexible polymers in solution. Macromolecules. 1980;13:1280–1289. doi: 10.1021/ma60077a048. [DOI] [Google Scholar]
  • 40.Sutani T., Yanagida M. DNA renaturation activity of the SMC complex implicated in chromosome condensation. Nature. 1997;388:798–801. doi: 10.1038/42062. [DOI] [PubMed] [Google Scholar]
  • 41.Kimura K., Hirano M., et al. Hirano T. Phosphorylation and activation of 13S condensin by Cdc2 in vitro. Science. 1998;282:487–490. doi: 10.1126/science.282.5388.487. [DOI] [PubMed] [Google Scholar]
  • 42.Forte G., Boteva L., et al. Marenduzzo D. Bridging condensins mediate compaction of mitotic chromosomes. J. Cell Biol. 2024;223 doi: 10.1083/jcb.202209113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Valentine M.T., Perlman Z.E., et al. Weitz D.A. Mechanical properties of Xenopus egg cytoplasmic extracts. Biophys. J. 2005;88:680–689. doi: 10.1529/biophysj.104.048025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Tomari T., Doi M. Hysteresis and incubation in the dynamics of volume transition of spherical gels. Macromolecules. 1995;28:8334–8343. doi: 10.1021/ma00128a050. [DOI] [Google Scholar]
  • 45.Shibayama M., Nagai K. Shrinking kinetics of poly(N-isopropylacrylamide) gels T-jumped across their volume phase transition. Macromolecules. 1999;32:7461–7468. doi: 10.1021/ma990719v. [DOI] [Google Scholar]
  • 46.Takaharu Okajima T.O., Ichiro Harada I.H., et al. Shunsuke Hirotsu S.H. Discontinuous crossover between fast and slow kinetics at the volume phase transition in poly-N-isopropylacrylamide gels. Jpn. J. Appl. Phys. 2000;39 doi: 10.1143/JJAP.39.L875. [DOI] [Google Scholar]
  • 47.Taylor G.I., Michael D.H. On making holes in a sheet of fluid. J. Fluid Mech. 1973;58:625–639. doi: 10.1017/S0022112073002375. [DOI] [Google Scholar]
  • 48.Lipowsky R. Domain-induced budding of fluid membranes. Biophys. J. 1993;64:1133–1138. doi: 10.1016/s0006-3495(93)81479-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Lipowsky R. Remodeling of membrane shape and topology by curvature elasticity and membrane tension. Adv. Biol. 2022;6 doi: 10.1002/adbi.202101020. [DOI] [PubMed] [Google Scholar]
  • 50.Yamamoto T., Sakaue T., Schiessel H. Loop extrusion drives very different dynamics for Rouse chains in bulk solutions and at interfaces. Europhys. Lett. 2019;127 doi: 10.1209/0295-5075/127/38002. [DOI] [Google Scholar]
  • 51.Yamamoto T., Schiessel H. Dilution of contact frequency between superenhancers by loop extrusion at interfaces. Soft Matter. 2019;15:7635–7643. doi: 10.1039/C9SM01454C. [DOI] [PubMed] [Google Scholar]
  • 52.Yamamoto T., Li W. Polymer brush inspired by ribosomal RNA transcription. Eur. Phys. J. E. 2023;46:61. doi: 10.1140/epje/s10189-023-00323-5. [DOI] [PubMed] [Google Scholar]
  • 53.Philips R., Kondev J., et al. Garcia H. 2nd ed. Garland Science; NY: 2012. Physical Biology of the Cell. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Figures S1–S11
mmc1.pdf (1.7MB, pdf)
Document S2. Article plus supporting material
mmc2.pdf (2.7MB, pdf)

Data Availability Statement

The Mathematica file “antennamodelVer155_doublelayer(brushcorrection).nb” used for the numerical calculations is available in figshare with the identifier (https://doi.org/10.6084/m9.figshare.28900841).


Articles from Biophysical Journal are provided here courtesy of The Biophysical Society

RESOURCES