Liquids and gels can be made to flow by applying external forces at their boundaries. In this issue, Sanchez et al.1 report the observation of self-sustained flows that occur in the absence of external forces — a hallmark of living systems — in a model gel. When this 'active' gel is confined to the interior of water droplets in a water–oil emulsion, the flows resemble the streaming used by cells to circulate their fluid content. Even more remarkable is the fact that, when one of these gel-filled droplets comes into contact with a hard surface, the self-driven flows of the confined gel drive the droplet along the surfaceFootnote 1.

To build their gel, Sanchez et al. sequentially assembled ingredients extracted from cells (Fig. 1). The first — and key — components are microtubules. These stiff, cylindrical filaments are one of the constituents of the cytoskeleton, the polymer network that mediates force transmission and motility in cells. Microtubule dynamics in cells is regulated by several proteins. Among these is kinesin, a motor protein capable of 'walking' on individual microtubules by converting chemical energy from ATP fuel molecules into mechanical work. To construct the active units of their gel, the authors used a protein called streptavidin as a scaffold to assemble clusters of kinesins that could simultaneously bind to multiple microtubules.

Figure 1: Assembly of microtubule bundles.
figure 1

a, Sanchez et al.1 combined the protein streptavidin with kinesin motor proteins that had been modified to bind to streptavidin (modification not shown). The proteins self-assembled to form clusters of several kinesin molecules in complex with streptavidin. b, The authors then added microtubule filaments and polymer coils to the mix. The polymer coils generated 'depletion' forces that pushed the microtubules together, promoting the formation of microtubule bundles mediated by the kinesin clusters. The bundles formed the basis of an 'active' gel — a material that generated self-sustained, internal flows of fluid in the absence of external forces.

Full size image

Finally, Sanchez et al. added nanometre-sized polymer coils to the solution. This step was essential to promote the formation of microtubule bundles that, in the presence of ATP, are continuously remodelled by the action of the crosslinking motor proteins. The coils induce attractive forces between the microtubules through a mechanism known as depletion interaction. This interaction arises when two filaments come near to each other, because the narrow gap between them is no longer accessible to the polymer coils. This creates an osmotic pressure difference that effectively acts as an attractive force between the filaments2. Sanchez and colleagues' overall hierarchical assembly process was recently used by the same group to build artificial cilia that beat periodically and, when densely packed on a substrate, spontaneously synchronize their beating pattern to create travelling waves3.

At a moderate density, the microtubule bundles form a polymer network that is internally driven by the action of kinesins. The network flows spontaneously and exhibits mixing and enhanced transport, compared with its non-active counterpart (which is obtained when the ATP fuel runs out); this was demonstrated by the authors by tracking small particles suspended in the gel.

At scales much larger than the typical bundle length (tens of micrometres), the rich dynamics of the system resembles that of complex fluids such as liquid crystals driven by externally applied fields, but differs from them in that it occurs spontaneously as a result of the internal drive. This is the key property of active materials that are driven out of equilibrium not by forces applied at their boundaries, but rather by an input of energy on each unit, as in a suspension of swimming bacteria. Energy uptake at the microscopic scale is crucial for driving emergent phenomena and self-organization in disparate systems4 — from naturally occurring ones, such as bacterial suspensions and flocks of birds, to chemical and mechanical analogues, such as self-propelled Janus colloids (microscopic particles that have two faces with distinct properties).

When Sanchez et al. confined the microtubule network to a water–oil interface, the resulting dense, two-dimensional film again exhibited self-sustained streaming flows that seemed to be associated with bundle fracturing and healing. The complex dynamics yielded patterns resembling topological defects — structures that can be generated in liquid crystals at equilibrium by confinement or external drive.

Finally, when the researchers confined the active gel to droplets of at least 30 micrometres in diameter, the gel was spontaneously adsorbed to the inner surface of the droplets, turning into a two-dimensional active film on a curved substrate. Remarkably, the self-sustained active flows of the trapped gel drove autonomous movement of the droplet on a substrate. Although the motile droplets moved along somewhat circular trajectories, rather than travelling in a straight line, they covered about 250 micrometres in 33 minutes. These moving drops bring to mind recent theoretical work5 showing that active drops in a fluid spontaneously acquire directed motility. For drops in which the active constituents — the microtubule bundles in Sanchez and colleagues' work — form large domains and have, on average, a common orientation but no preferred direction, the theory indeed predicts rotational motion of the drops.

Reconstituted microtubule–kinesin systems have been explored before as models for active self-assembly, not least in the remarkable experiments6,7 that led the way to current studies of pattern formation in active systems. In those experiments, kinesin complexes driven by ATP organized microtubules into spirals and asters reminiscent of a cell's mitotic spindle, a star-like microtubule assembly that mediates cell division. One important difference is that the structures seen in the earlier work6,7 were essentially static, whereas Sanchez and colleagues' microtubule gel generates continuously evolving, spontaneous flows that persist as long as ATP is present — not unlike what happens in living cells. Furthermore, Sanchez et al. report that the internally generated flows in their active gel can be tuned by varying the ATP concentration, confirming the self-sustained, non-equilibrium nature of the dynamics. The fact that microtubules are assembled into bundles seems to be essential for yielding self-sustained motion (see Movie S2 in the Supplementary Information to the paper1), but the reason for this remains an open question. Also unexplained is why the behaviour of the active microtubule network is so different from that of gels composed of actin filaments and myosin motor proteins, in which activity yields spontaneous contraction8.

Sanchez and colleagues' work is a beautiful example of a growing class of experiment in biomimetic assembly, aimed at building systems that exhibit some of the features of living matter. Will it be possible to control and direct the motility of the active droplets? And can the flow-induced structures be harnessed and used as guides for the transport of particles through fluid, as those in cells are? This remains to be seen. Meanwhile, experiments of this type are beginning to shed light on the physical aspects of the complex dynamical reorganization that occurs continuously inside cells. When combined with studies of the biochemical machinery and signalling that drive such reorganization, they may ultimately lead to a quantitative understanding of the mechanics of living matter.