Confinement Sets the Boundary Conditions, the Cortex Chooses the Direction

Question: How do mechanical cues from the extracellular matrix or eggshell-like confinement couple to actomyosin cortical flow and spindle orientation during oriented cell division in early embryos?

Answer

Taken together, the retrieved papers support a two-stage architecture in which confinement is a boundary condition rather than an instructive signal. Eggshell geometry and mechanical compression set the geometry and magnitude of the force landscape — by restricting admissible furrow shapes, by imposing curvature-dependent active stress, and by fixing cell size — and these changes are sufficient to bias how much of the division is asymmetric ([arxiv:2609.01393], [arxiv:2602.18909]). The actomyosin cortex then converts this landscape into oriented flow and nematic order, which both advects polarity proteins and generates anisotropic tension ([doi:10.64898/2026.03.30.715403], [doi:10.64898/2026.04.09.717607]). The direction of division is finally read out at the force-generator level, where anisotropic tension recruits NuMA to the cortex earlier and amplifies spindle oscillations, and where pulling and pushing forces compete with geometry-derived centering ([doi:10.1101/2024.10.17.617436], [doi:10.1101/2025.09.22.677921], [doi:10.64898/2026.01.23.701263]). Across these papers, confinement and matrix mechanics are therefore permissive and amplifying; what carries directional information is cortical chirality, cortical polarity, and the anisotropy of the cytoplasm — not the shell itself ([doi:10.64898/2026.02.19.706789], [doi:10.1101/2025.11.24.690087]).

Confinement changes the magnitude and asymmetry of division without specifying an axis

Curvature-dependent active stresses in the cortex are sufficient to break division symmetry in an unpolarized cell, with the contractile ring slipping off the equator to yield daughters of unequal volume; reducing cell size or tightening confinement triggers the same instability ([arxiv:2609.01393]). In that model, measured division asymmetry increases as C. elegans blastomeres shrink within the eggshell and is further amplified when the embryo is mechanically compressed, matching the model ([arxiv:2609.01393]). Confined division also has a hard geometric ceiling: confinement restricts the set of mechanically admissible furrow shapes, so pressure and axial force saturate at a confinement-imposed minimum furrow radius, with material properties only selecting the operating point inside those limits ([arxiv:2602.18909]). The eggshell's contribution is task-specific rather than global: eggshell geometry, together with CUL-3–actomyosin-dependent oriented division, is required for Phase I of AB cell rotation in the two-cell embryo, whereas Phase II proceeds independently of eggshell geometry ([doi:10.64898/2026.06.28.735104]). Confinement can also overwrite an inherited axis, as Hydra tissue spheroids in narrow channels regenerate with the body axis aligned to the channel while their supracellular actomyosin nematic fibers remodel to match ([doi:10.64898/2026.06.25.734673]).

Flow converts cortical architecture into anisotropic tension that feeds back on the axis

Compressive cortical flow aligns actin bundles in the anterior domain of the C. elegans zygote and generates anisotropic tension perpendicular to the flow direction; density-dependent contractility turns flow-driven material redistribution into mechanical negative feedback that arrests PAR domains near the midpoint ([doi:10.64898/2026.03.30.715403]). When symmetry breaking occurs laterally rather than at the pole, this anisotropic tension from aligned bundles rotates the posterior domain toward the nearest pole, so nematic architecture participates in axis convergence ([doi:10.64898/2026.03.30.715403]). The flow itself is quantitatively set by microscopic kinetics: flow speed depends biphasically on actin assembly and crosslinking rates, reflecting a competition between disrupting force transmission and reducing resistance to it ([doi:10.1101/2025.01.12.632635]). Coherent flow also requires rapid, coordinated turnover — strong reductions in either assembly or disassembly rate collapse the cortex, with distinct effects on cortex versus membrane dynamics ([doi:10.64898/2026.05.24.727551]). Cortical shape anisotropy is by itself sufficient to induce anisotropy of cortical structure and tension, with the bundled stress-fiber contribution scaling with aspect ratio while the unbundled meshwork contribution stays isotropic ([doi:10.1101/2024.10.25.620349]).

Flow mainly reinforces polarity rather than creating the division axis

In the C. elegans P1 cell, polarization depends primarily on sustained antagonistic anterior PAR activity via CDC-42 and PKC-3, with CDC-42–MRCK-1 driving the contractility that produces flow; those flows contribute only weakly to P1 polarity, mainly in the late phase ([doi:10.64898/2026.04.09.717607]). The same work shows cortical flow acts as a transport-mediated reinforcement that buffers zygotic PAR polarity against CDC-42 dosage changes, and that suppressing flow sensitizes polarity to otherwise tolerable perturbations ([doi:10.64898/2026.04.09.717607]). Perturbing tension upstream displaces the spindle rather than just slowing polarity: elevated cortical tension in mammalian oocytes produces an ectopic actomyosin-rich domain, directional cytoplasmic flows, and spindle migration that disrupts orientation ([doi:10.64898/2026.10.16.759562]). Cytoplasmic anisotropy can even substitute for shape cues — in ascidian two-cell embryos, an anisotropic ER-rich domain plus dynein-mediated cortical pulling aligns centrosomal complexes, and the ER-based mechanism can orient them in the absence of cell shape cues ([doi:10.1101/2025.11.24.690087]).

The mechanosensitive node is the cortical force generator, not the flow

Applying external stretch to epithelial tissue recruits NuMA to the cortex earlier in mitosis, coinciding with the onset and amplification of spindle oscillations, and NuMA knockdown reduces those oscillations and disrupts orientation according to stretch and cell shape ([doi:10.1101/2024.10.17.617436]). Computational integration of pulling and pushing mechanisms across Drosophila follicular epithelium, Drosophila neuroblasts, and the early C. elegans embryo indicates pulling forces drive directed orientation while pushing forces prevent stalling and promote alignment, particularly at high initial misalignment ([doi:10.1101/2025.09.22.677921]). Spindle elongation is driven by cortically anchored motors and astral microtubules and therefore scales with cell size, whereas chromosome segregation dynamics remain largely invariant — so the same size reduction imposed by confinement reshapes elongation without reshaping segregation ([doi:10.1101/2025.10.13.681585]). In sea urchin embryos, centrosomes, nuclei, and spindles start centered and progressively shift apically as cells shrink, consistent with geometry-derived centering forces decaying with cell size while apical polarity forces take over ([doi:10.64898/2026.01.23.701263]).

Handedness comes from the cortex and its substrate, not from the shell

Chiral actomyosin flows arising from active chiral torques drive the left-right asymmetric spindle skew in C. elegans, and reversing the handedness of those torques reverses chiral flow, mirrors the six-cell embryo, and produces situs inversus ([doi:10.64898/2026.02.19.706789]). In quail, the node generates an active chiral torque of about 6 µN·µm that depends on actomyosin activity and requires the underlying ventral meso-/endoderm as a mechanical substrate sustaining the counter-torque, i.e. a dorsoventrally oriented torque dipole across tissue layers ([doi:10.1101/2025.07.16.665037]). This is the clearest case in the retrieved set where a matrix-like mechanical substrate, rather than the confinement boundary, supplies the reaction needed for an oriented cortical flow ([doi:10.1101/2025.07.16.665037]). Direct evidence that matrix stiffness raises intracellular actin tension comes from FRET-based actin and α-actinin tension sensors showing higher filamentous-actin tension on stiffer substrates with increased spreading ([doi:10.64898/2026.08.02.742292]).

Conflicts and gaps

Whether confinement is instructive or merely permissive is unresolved. One model has confinement alone breaking division symmetry without any molecular polarity cue ([arxiv:2609.01393]), while the C. elegans rotation work finds eggshell geometry required for only one of two rotation phases and identifies Ras-MAPK as an eggshell-independent regulator ([doi:10.64898/2026.06.28.735104]). Separating direct force sensing from shape-mediated sensing is claimed to be achievable by stretch experiments ([doi:10.1101/2024.10.17.617436]), but the cortical tension work shows cell shape alone suffices to set tension anisotropy ([doi:10.1101/2024.10.25.620349]) — so the two accounts are not yet reconciled in embryos. Cell size is claimed to make geometry-derived centering decay so polarity dominates ([doi:10.64898/2026.01.23.701263]), yet cortical-motor-driven elongation is argued to scale with cell size across C. elegans lineages ([doi:10.1101/2025.10.13.681585]); these invoke different force populations. Most importantly, no retrieved paper measures ECM or eggshell stiffness, cortical flow, and spindle orientation simultaneously in an intact embryo: the stiffness-to-actin-tension link is established only in substrate-cultured cells ([doi:10.64898/2026.08.02.742292]), and the embryo-side evidence for matrix mechanics acting on the cortex comes from a tissue-layer substrate in the chick node ([doi:10.1101/2025.07.16.665037]) rather than from the eggshell. A cytoplasmic alternative also competes with purely cortical accounts, since ER anisotropy can orient centrosomes without shape cues ([doi:10.1101/2025.11.24.690087]) and endomembrane composition alone can render cytoplasm viscoelastic and glass-like ([doi:10.64898/2026.07.20.739612]).

What would change the answer

A decisive experiment would combine in vivo stiffening or softening of the eggshell (genetic or chemical crosslinking of eggshell/chitin proteins, or osmotic compression) with simultaneous quantification of cortical myosin flow fields and spindle/centrosome tracking, repeating the compression paradigm that already amplifies division asymmetry ([arxiv:2609.01393]). Pairing that with a genetically encoded cortical tension sensor would test whether confinement acts through cortical tension or through shape, addressing the shape-versus-force ambiguity directly ([doi:10.1101/2024.10.17.617436], [doi:10.1101/2024.10.25.620349]). Measuring cortical NuMA density as a function of locally imposed tension in an embryo — rather than in stretched epithelium — would establish whether NuMA relocalization is the conserved mechanotransduction step. Optogenetic control of cortical contractility confined to defined geometries would separate the contribution of active flow from that of the passive boundary ([doi:10.1101/2025.07.16.665037]). Finally, dechorionated or shell-free embryo preparations combined with ER-domain disruption would test whether cytoplasmic anisotropy can carry the axis when confinement is removed ([doi:10.1101/2025.11.24.690087], [doi:10.64898/2026.07.20.739612]).