Highlights d Fast-moving cells in 3D matrix establish low membrane tension at the rear d Caveolae form in response to low membrane tension and recruit the GEF Ect2 d Ect2 activates RhoA to promote F-actin organization and rear retraction d Positive feedback between membrane tension and contractility reinforces retraction

Cells going through mitosis undergo precisely timed changes in cell shape and organisation, which serve to ensure the fair partitioning of cellular components into the two daughter cells. These structural changes are driven by changes in actin filament and microtubule dynamics and organisation. While most evidence suggests that the two cytoskeletal systems are remodelled in parallel during mitosis, recent work in interphase cells has implicated the centrosome in both microtubule and actin nucleation, suggesting the potential for regulatory crosstalk between the two systems. Here, by using both in vitro and in vivo assays to study centrosomal actin nucleation as cells pass through mitosis, we show that mitotic exit is accompanied by a burst in cytoplasmic actin filament formation that depends on WASH and the Arp2/3 complex. This leads to the accumulation of actin around centrosomes as cells enter anaphase and to a corresponding reduction in the density of centrosomal microtubules. Taken together, these data suggest that the mitotic regulation of centrosomal WASH and the Arp2/3 complex controls local actin nucleation, which may function to tune the levels of centrosomal microtubules during passage through mitosis.

ESCRT-0 component HRS and actin polymerization factor WASH reside in adjacent endosomal domains. MacDonald et al. show that HRS controls WASH localization and recycling of WASH-dependent transmembrane cargo. Cargo binding to endosomal actin thus acts as sorting signal to oppose ubiquitin-mediated degradation.

The perinuclear actin cap is an important cytoskeletal structure that regulates nuclear morphology and re-orientation during front-rear polarisation. The mechanisms regulating the actin cap are currently poorly understood. Here, we demonstrate that STEF/TIAM2, a Rac1 selective guanine nucleotide exchange factor, localises at the nuclear envelope, co-localising with the key perinuclear proteins Nesprin-2G and Non-muscle myosin IIB (NMMIIB), where it regulates perinuclear Rac1 activity. We show that STEF depletion reduces apical perinuclear actin cables (a phenotype rescued by targeting active Rac1 to the nuclear envelope), increases nuclear height and impairs nuclear re-orientation. STEF down-regulation also reduces perinuclear pMLC and decreases myosin-generated tension at the nuclear envelope, suggesting that STEF-mediated Rac1 activity regulates NMMIIB activity to promote stabilisation of the perinuclear actin cap. Finally, STEF depletion decreases nuclear stiffness and reduces expression of TAZ-regulated genes, indicating an alteration in mechanosensing pathways as a consequence of disruption of the actin cap.

Cell invasion and metastasis is a multi-step process, initialised through the acquisition of a migratory phenotype and the ability to move through differing and complex 3D extracellular environments. In this study we set out to identify the parameters required for invasive cell migration in 3D environments. Cells interact with the extracellular matrix via transmembranespanning integrin adhesion complexes, which are well characterised in cells plated on 2D surfaces, yet much less is known about them in cells embedded in 3D matrices. We establish a technique to determine the composition of cell matrix adhesion complexes of invasive breast cancer cells in 3D matrices and on 2D surfaces and we identify an interaction complex enriched in 3D adhesive sites required for 3D invasive migration. Depletion of β-PIX-Myosin18A (Myo18A) abolishes cancer cell invasion, without negatively affecting matrix degradation, Rho GTPase signalling, or protrusion formation in collagen matrices. Instead, in a mechanism only seen in cells moving through 3D matrix, β-PIX and Myo18A drive the polarised recruitment of non-muscle Myosin 2A (NM2A) to the tips of protrusions. This recruitment of NM2A is required for the creation of an NM2A-NM2B isoform gradient, which ranges from the protrusion to the nucleus. We observe a requirement for active force transmission to the nucleus during invasive migration that is needed to pull the nucleus forward. We postulate that the establishment of the NM2A-NM2B actomyosin gradient facilitates the coupling of cell-matrix interactions at the protrusive cell front with nuclear movement, enabling effective invasive migration and front-rear cell polarity.