Spatial Organization of the Cell Cytoplasm by Position-Dependent Phase Separation

During asymmetric cell division, cytoplasmic components are segregated to opposite sides of the cell. We discuss how the observed segregation can be achieved by a position-dependent phase separation mechanism controlled by a protein concentration gradient. We show that effects of even a weak gradient can be amplified by the phase transition to achieve strong segregation. We compare our theory to the segregation of germ granules observed during the divisions in the C. elegans embryo. Our study demonstrates how liquid-liquid phase separation can play a key role in the organization of the cytoplasm.

Figure 1

(a) Schematic representation of the polarized C. elegans embryo at the one-cell stage prior to asymmetric cell division. Cell polarity is established by asymmetric localization of proteins in the anterior (red) and posterior (black) cell membranes. This asymmetry is required to generate a cytoplasmic gradient of Mex-5 protein (gray). P granules (green) segregate to the posterior side where they grow, while they disappear in the anterior side. (b) False color representation of fluorescence intensity of Mex-5 tagged with green fluorescence protein (GFP) averaged over four embryos reflecting the spatial distribution of Mex-5 proteins in the cell. (c) Image of the C. elegans embryo superimposed with the fluorescence image of PGL-1 tagged with GFP labeling P granules in the cell. Note that P granules are localized at the posterior side of the cell.

Figure 2

(a) Phase diagram for a ternary fluid description of the cytoplasm, where $\psi$ denotes the volume fraction of P granule constituents and $\varphi$ denotes the volume fraction of the Mex-5 protein. In the one-phase region, all components mix. For small $\psi$, a two-phase coexistence region exists in which P granules coexist with the cytoplasm. The broken lines connect coexisting phases. The region encapsulated by the red broken line corresponds to the region of the phase diagram shown in (b). (b) The phase diagram at low $\varphi$ and low $\psi$ limits. The gray tiled line indicates schematically a range of Mex-5 concentrations corresponding to a gradient spanning from anterior ($A$) to posterior ($P$) in a polarized cell. Note that P granule segregation implies that ($A$) is in the one-phase region while ($P$) is inside the two-phase region. The case of a nonpolarized cell with P granules in the cytoplasm corresponds to a single point (red) in the two-phase region in the phase diagram. Parameter values of the phase diagram shown are ${\chi }_{\varphi \psi }=-5k_{B}T$, ${\chi }_{\varphi \theta }=3k_{B}T$, and ${\chi }_{\psi \theta }=-2k_{B}T$.

Figure 3

(Upper figure) Calculated steady-state distribution of the total P granule mass density ${\rho }_{\mathrm{tot}}(x)={\rho }_d(x)+{\rho }_c(x)$ solid line), the mass density of the droplet form ${\rho }_d(x)$ (dash line), and of the cytoplasmic form ${\rho }_c(x)$ (dotted line) as a function of position $x$. Droplets segregate to the region $x>0$ where phase separation occurs, while they dissolve in the region $x<0$. Parameter values are $D_c=10\mu {\mathrm{m}}^2/\mathrm{s}$, $D_d=0.01\mu {\mathrm{m}}^2/\mathrm{s}$, $\alpha =\beta ={10}^{-3}/\mathrm{s}$, and $L_1=L_2=20\mu \mathrm{m}$. The $y$ axis is rescaled so that the total mass $\int dx{\rho }_{\mathrm{tot}}(x)=1$. (Lower figure) Comparison of the theory with experimental data. The experimentally observed distribution ${\rho }_{\mathrm{tot}}(x)$ of P granule constituents in a polarized C. elegans embryo is shown (black circles). It is obtained from the fluorescence intensity of PGL-1 tagged with GFP. Note that data close to the cell poles are omitted due to boundary effects. As a result, the total length of the system is $40\mu \mathrm{m}$. A comparison of the steady-state solution of Eq. (4) to the experimental data is shown by the solid line. For the same solution, the mass densities for the droplet form ${\rho }_d$ (dash line) and for the cytoplasmic form ${\rho }_c$ (dotted line) are also shown. Values for the diffusion constants were chosen as are $D_c=10\mu {\mathrm{m}}^2/\mathrm{s}$, $D_d=0.01\mu {\mathrm{m}}^2/\mathrm{s}$. The fitted parameter values are $\alpha =1.5\times {10}^{-5}/\mathrm{s}$, $\beta =1.4\times {10}^{-3}/\mathrm{s}$, $L_1=24\mu \mathrm{m}$, and $L_2=16\mu \mathrm{m}$. The $y$ axis is rescaled so that the total mass $\int dx{\rho }_{\mathrm{tot}}(x)=1$.