Generation of compartmentalized pressure by a nuclear piston governs cell motility in a 3D matrix

See allHide authors and affiliations

Science  29 Aug 2014:
Vol. 345, Issue 6200, pp. 1062-1065
DOI: 10.1126/science.1256965

Push me, pull you, that's the way to move

Primary cells, derived directly from human tissue, exhibit different behaviors in shape and signaling within three-dimensional (3D) or 2D spaces. When the pressure within the cell increases, cells display limb-like bumps, which they use to move through their 3D environment. Petrie et al. now show that when the complex of actin and myosin contracts, it controls the pressure within cells and therefore the shape of those protruding structures (see the Perspective by DeSimone and Horwitz). The authors measured internal pressures in migrating mammalian cells. In the 3D matrix, those cells have higher pressure that differs between the front and back of the cell, which creates a piston effect.

Science, this issue p. 1062; see also p. 1002


Cells use actomyosin contractility to move through three-dimensional (3D) extracellular matrices. Contractility affects the type of protrusions cells use to migrate in 3D, but the mechanisms are unclear. In this work, we found that contractility generated high-pressure lobopodial protrusions in human cells migrating in a 3D matrix. In these cells, the nucleus physically divided the cytoplasm into forward and rear compartments. Actomyosin contractility with the nucleoskeleton-intermediate filament linker protein nesprin-3 pulled the nucleus forward and pressurized the front of the cell. Reducing expression of nesprin-3 decreased and equalized the intracellular pressure. Thus, the nucleus can act as a piston that physically compartmentalizes the cytoplasm and increases the hydrostatic pressure between the nucleus and the leading edge of the cell to drive lamellipodia-independent 3D cell migration.

Cells moving across a flat two-dimensional (2D) surface or inside nonlinearly elastic 3D collagen use polarized signaling to direct the formation of a dendritic actin network and extend flat, lamellipodial protrusions (1, 2). When primary human fibroblasts move within a cross-linked, linearly elastic 3D structure such as dermal or cell-derived matrix, they can switch to a lamellipodia-independent migration mechanism characterized by nonpolarized signaling and blunt, cylindrical protrusions termed “lobopodia” (1). Actomyosin contractility via the RhoA–ROCK–myosin II signaling axis is required for cells to form and maintain lobopodia in response to the degree of matrix cross-linking. However, the mechanism by which increased contractility generates lobopodia is unclear.

Lobopodial cells can also be distinguished by rapid membrane blebbing along their sides, oriented perpendicular to the leading edge. Membrane blebs can be generated by elevated intracellular hydrostatic pressure, local weakening of the attachment of the plasma membrane to the underlying cortex, or both (35). We hypothesized that this lateral blebbing could result from elevated intracellular pressure during lobopodial motility. This increased pressure might result from the RhoA, ROCK, and myosin II activities required for the lamellipodia-independent migration of fibroblasts through a physiological linearly elastic 3D matrix (1).

We tested the hypothesis by directly determining intracellular pressures in primary human fibroblasts migrating on 2D surfaces compared with a 3D extracellular matrix (ECM). We used a microelectrode coupled to a servo-null micropressure system to penetrate the plasma membrane immediately in front of the nucleus (relative to the leading edge) and to measure the intracellular hydrostatic pressure exerted by the cytoplasm directly (Pic) (Fig. 1) (see supplementary materials and methods). Direct comparisons of pressure in cells migrating on top of a cell-derived matrix (CDM) and embedded within a 3D collagen matrix revealed low hydrostatic pressures in both 2D and 3D lamellipodial cells [~300 and 700 Pa on the linearly elastic 2D surface of the CDM and within nonlinearly elastic 3D collagen, respectively; see (1) for characterization of matrix elastic behavior]. In contrast, intracellular pressure was substantially elevated (~2200 Pa) in lobopodial cells migrating inside the 3D CDM. Switching these lobopodial cells to lamellipodial cells by inhibiting RhoA, ROCK, or myosin II (1) reduced hydrostatic pressure (to ~400 Pa) in each case. This inhibition distinguished lobopodia from the contractility-independent water permeation mechanism used by certain cancer cells in confined channels (6). Control cells using lamellipodia to migrate on 2D glass maintained relatively low intracellular pressure (fig. S1A), with values consistent with indirect pressure estimates for other cell types (7, 8). As expected (9), placing cells in a hypotonic medium to trigger an influx of water increased Pic, as did increasing contractility by treating cells with calyculin A. Thus, a linearly elastic, cross-linked 3D ECM activates actomyosin contractility to increase intracellular pressure and maintain the lobopodial mode of 3D cell migration.

Fig. 1 Actomyosin contractility governs intracellular pressure in a 3D ECM.

Comparison of the intracellular pressures (n ≥ 20 replicates each) of lamellipodial cells on 2D CDM and in 3D collagen, untreated lobopodial cells in 3D CDM, or cells in CDM treated overnight with inhibitors of myosin II (25 μM blebbistatin), ROCK (10 μM Y-27632), or RhoA (10 μg/ml C3 transferase). N = 3 independent experiments, *P < 0.001. Error bars indicate SEM.

To establish whether intracellular pressure is uniformly increased throughout the cytoplasm of lobopodia-bearing cells, we compared hydrostatic pressures immediately in front of and behind the nucleus (Fig. 2, A and B). Pic was significantly elevated and compartmentalized in lobopodia (to ~2400 Pa), with the nucleus separating this anterior high-pressure compartment from a low-pressure zone (~900 Pa) in the cell posterior. In contrast, low pressures were found both in front and back of the nucleus in fibroblasts using lamellipodia to migrate in 2D and 3D environments (~400 and 800 Pa, respectively).

Fig. 2 Lobopodial fibroblasts are compartmentalized into high- and low-pressure zones.

(A) Intracellular pressures were measured immediately in front of (green dot) and behind (red dot) the nucleus (N). Scale bar, 5 μm. (B) Comparison of intracellular pressures in front and behind the nucleus of fibroblasts (n ≥ 25) migrating on 2D glass (2D lamellipodia), in 3D collagen (3D lamellipodia), or in 3D CDM (3D lobopodia) (N = 3). *P < 0.01. (C) A subpopulation of PA-GFP was activated near the leading edge (dashed box), and the rate of translocation was measured at the indicated regions of interest (ROI) immediately in front of and behind the nucleus in live cells. Scale bar, 5 μm. (D) Time constants of PA-GFP accumulation show that the nucleus significantly slows the rate of PA-GFP translocation in 3D lobopodial versus 2D lamellipodial fibroblasts (n ≥ 17, N = 3). *P < 0.01. (E) Myosin II inhibition results in immediate backward movement of the nucleus in lobopodial but not lamellipodial cells (quantified in the lower panel) (n = 10, N = 3). *P < 0.01. Scale bar, 3 μm. RFP-NLS, red fluorescent protein–nuclear localization signal; NM, nuclear movement. (F) Pressure rises transiently behind the nucleus after myosin II inhibition only in lobopodial cells (n ≥ 16, N = 3). **P < 0.001 versus front Pic; *P < 0.01 versus back Pic. (G) Anterior inhibition of myosin II prevents forward movement of the nucleus (n = 10, N = 3). *P < 0.01. Error bars represent SEM.

The existence of large differences in hydrostatic pressure in front of versus behind the nucleus suggests that the nucleus physically divides the cytoplasm in lobopodial cells. We tested this prediction by measuring the diffusion and/or convection of cytoplasmic photoactivatable green fluorescent protein (PA-GFP) in live cells (Fig. 2, C and D). After photoactivation near the leading edge (Fig. 2C), fluorescent PA-GFP significantly slowed as it moved past the nucleus in lobopodial cells in 3D CDM compared with lamellipodial cells on 2D glass (Fig. 2D and fig. S2). This matrix-dependent barrier consisted of the nucleus and a dense accumulation of sheets of membranous material, including endoplasmic reticulum (fig. S3).

Given the established links between the nucleus, cytoskeleton, and the ECM (1013) and the requirement for actomyosin contractility to generate high-pressure lobopodia (Fig. 1), we hypothesized that the tight-fitting nucleus might be pulled forward like a piston to pressurize the anterior cytoplasmic compartment. Live-cell confocal microscopy showed apparent pulling forward of the nucleus, which periodically accelerated away from the cell’s trailing edge in 3D CDM (movie S1). Further, monitoring of nuclear movement and Pic after inhibition of myosin II revealed that the nucleus fell back and pressure equilibrated initially, ultimately becoming low and uniform (Fig. 2, E and F and movie S2), consistent with loss of actomyosin contraction pulling the nuclear piston forward. In direct contrast, the nucleus continued to move forward with no pressure change in 2D lamellipodial cells after myosin II inhibition. Microtubules were not required for nuclear movement in lobopodial cells (fig. S4).

Myosin IIA and vimentin intermediate filaments were polarized anteriorly in lobopodial cells (Fig. 3, A and B). Because nesprin-3, a nucleoskeleton-vimentin linker protein, is a proposed mediator of force transmission through the cytoskeleton to the nucleus and is needed for polarity and migration (14), we searched for an actomyosin–vimentin–nesprin-3 complex. Immunoprecipitation revealed a complex of actin, myosin IIA, vimentin, and nesprin-3 in primary cells (Fig. 3C). Actin and myosin II were lost from the complex after inhibiting myosin II [only 18.4 ± 12.8% (SEM) of actin and 26.7 ± 16.3% of myosin IIA remained associated with vimentin after blebbistatin treatment, versus 88.7 ± 14.9%, of the nesprin-3]. Local inhibition of myosin II in front of the nucleus (fig. S1B) blocked forward movement of the nucleus (Fig. 2G), reduced forward pressure, and retracted the leading edge (Fig. 3D), whereas myosin II inhibition behind the nucleus did not. Similarly, local permeabilization of the plasma membrane in front of the nucleus resulted in loss of forward pressure and leading edge retraction (fig. S1C). To test whether nesprin-3 provides a link between myosin contractility and forward movement of the nucleus in lobopodial fibroblasts, we compared instantaneous velocities of the nucleus and the trailing edge in control and nesprin-3 small interfering RNA (siRNA)–treated cells moving through 3D CDM (Fig. 3, E to G, and fig. S5). The nucleus and trailing edge moved independently in control cells, but not in cells with reduced nesprin-3 expression [Pearson’s correlation coefficient 0.28 ± 0.01 (SEM) in control versus 0.57 ± 0.07 in nesprin-3 siRNA-treated cells] (Fig. 3G and fig. S6). This loss of independent movement of the nucleus after depleting nesprin-3 was accompanied by a reduction in cell velocity similar to inhibiting global myosin II (Fig. 3H), suggesting that nesprin-3 and the nucleus are connected to a contractile pulling mechanism localized in front of the nucleus required for efficient 3D cell migration.

Fig. 3 Nesprin-3 associates with vimentin and actomyosin and mediates independent movement of the nucleus.

(A and B) Filamentous vimentin (A) and myosin IIA (B) are polarized to the anterior of lobopodial fibroblasts (n = 10). Scale bars, 10 μm. (C) Immunoprecipitation (IP) reveals that actin and myosin IIA form a myosin II adenosine triphosphatase–dependent complex with vimentin and nesprin-3 in primary fibroblasts (N = 6). IgG, immunoglobulin G. (D) Anterior inhibition of myosin II reduces forward pressure (n = 22, N = 3), and the leading edge retracts (n = 27, N = 5). (E to G) Instantaneous velocities of the nucleus (VN) and trailing edge (VTE) measured in lobopodial cells (n ≥ 14) expressing GFP–myosin light chain 2 and RFP-NLS [(E), scale bar, 5 μm] treated with control or nesprin-3 siRNA (F). Nesprin-3 knockdown reduces the independent movement of the nucleus relative to the trailing edge (G) (N = 3). (H) Myosin II activity and nesprin-3 are each required for efficient 3D migration (n ≥ 45, N = 3). **P < 0.0001, *P < 0.05. Error bars represent SEM.

We tested the hypothesized role of nesprin-3 in piston function of the nucleus by depleting nesprin in cells migrating in 3D CDM. Compartmentalized forward pressure was maintained in cells treated with control siRNA, whereas siRNA knockdown of nesprin-3 reduced and equalized anterior and posterior intracellular pressures (Fig. 4A; verified with independent siRNAs in fig. S5 D and E), even though the diffusion barrier remained intact (fig. S3A). Overall actomyosin contractile function remained unaffected in nesprin-3 knockdown cells, because RhoA activation by lysophosphatidic acid still increased 2D focal adhesion size in nesprin-3 siRNA-treated cells on glass (Fig. 4B and fig. S8). Removing the nucleus also reduced and equalized intracellular pressure, in addition to slowing 3D cell movement (fig. S7). Conversely, even though the RhoA effector formin mDia1 was required to increase focal adhesion size on 2D, its depletion did not affect the compartmentalized pressure or velocity of lobopodial cells (Figs. 3H and 4, A and B, and fig. S8). Thus, nesprin-3 is required for pistonlike nuclear movements and pressurizing lobopodial protrusions. This mechanism is distinct from roles of RhoA and contractility controlling focal adhesion size through mDia1.

Fig. 4 Nesprin-3 compartmentalizes intracellular pressure to mediate lobopodial 3D migration.

(A) Comparison of anterior and posterior intracellular pressure in cells treated with the indicated siRNAs (n ≥ 23, N = 3). (B) Average size of focal adhesions formed by cells treated with the indicated siRNAs and plated on 2D glass (n ≥ 13, N = 3). (C) Cortactin localization in siRNA-treated cells (n = 17). Arrowheads indicate local accumulation of the lamellipodial marker cortactin at the leading edge. Scale bars, 10 μm. (D) Quantification of lamellipodial cells in 3D CDM after siRNA treatments (n ≥ 33, N = 3). *P < 0.001. Error bars indicate SEM.

Finally, we tested whether the loss of nesprin-3 causes a loss of anterior lobopodial protrusions. Whereas control cells continued to use lobopodia to migrate through 3D CDM, the leading edges of nesprin-3 siRNA-treated cells switched to cortactin-positive lamellipodia (Fig. 4, C and D, fig. S9; confirmed with independent siRNAs in fig. S5F). This change mimics the switch from lobopodia to lamellipodia that occurs when RhoA, ROCK, or myosin II are inhibited (1), which reduced total intracellular hydrostatic pressure (Fig. 1).

We have shown that when human fibroblasts migrate through a physiological 3D ECM, their intracellular pressure becomes compartmentalized. Furthermore, actomyosin contractility and nesprin-3 act to pull the nuclear piston forward and pressurize the anterior cytoplasmic compartment. The contractility-dependent association of vimentin intermediate filaments with nesprin-3 and actomyosin—along with the known interactions of vimentin filaments, nesprin-3, and the nucleus (1416)—supports a model in which actomyosin contractility acts on vimentin filaments to pull the nucleus forward and increase compartmentalized pressure (fig. S10). This polarized intracellular pressure is associated with a nonpolarized distribution of intracellular signaling during 3D migration (1) and explains how a nucleoskeleton-cytoskeleton linker protein localized in the nuclear membrane can help to determine the structure and dynamic function of the leading edge of cells migrating in a 3D ECM. This mechanism can operate in a variety of primary human cells (fig. S11) and illustrates how the physical structure of the matrix can influence cellular physical properties to govern biological function.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

References (1723)

Movies S1 and S2

References and Notes

  1. Acknowledgments: We thank K. Carver for helping to characterize various primary human cells; D. Starr for providing the nesprin-3 antibody; and M. Hoffman, A. Doyle, W. Daley, and K. Holmberg for critical comments on the manuscript. This work was supported by the Intramural Research Program of the National Institute of Dental and Craniofacial Research, NIH. The data presented in this manuscript are found in the main paper and the supplementary materials and methods.
View Abstract

Navigate This Article