Henry R. Bourne and Orion Weiner

Eukaryotic cells often devote much timeand energy to business trips. Leuko-cytes crawl towards sites of infection,

soil amoebae find other amoebae with whichto form multicellular organisms, and bud-ding-yeast cells find their mating partners.How do these cells know where to go? In eachcase, an internal cellular ‘compass’ interpretsa shallow gradient of an external chemicalattractant to orientate the cell’s polarity in the correct direction; migration using this chemical compass is called chemotaxis.More than a billion years of eukaryotic evolution have conserved most componentsof the compass, yet we have just begun toscratch the surface of the network of positiveand negative feedback loops that control gra-dient interpretation and cell polarity. Onlynow are we beginning to unravel this mecha-nism to reveal how a cell’s ability to point inany direction allows it to navigate effectively.

The compass responsible for chemotaxisin eukaryotic cells differs from that of bac-teria. Bacteria swim in a jerky trajectory, usinga temporal strategy to interpret the gradient— they compare the concentration of attrac-tant at one time with that of a moment before.In contrast, eukaryotic cells take advantage oftheir larger size to implement a spatial strategy— they compare attractant concentrationsacross the cell surface, point themselvestowards the highest concentration, and movedirectly up the gradient. To do this, leukocytessuch as circulating neutrophils in mammalsamplify the shallow external gradient of theattractant by converting it to a steeper gradi-ent of internal signals, resulting in a polarizedcell with a protruding front that points up the external gradient. This polarity producesasymmetric attractant sensitivity — greater atthe cell’s protruding edge than at the trailingone. Consequently, a neutrophil confronted

with a 1807 reversal in the attractant gradientdoes not simply transform its trailing edgeinto a leading one; instead, it follows its highlysensitive ‘nose’ and executes a U-turn.

We propose that asymmetric responsive-ness to extracellular gradients involves twocomplementary mechanisms that make thecell’s ‘front’ (pseudopod) distinct from its‘back’. Each mechanism depends on actin, a cytoskeletal protein, to produce a self-organizing system for creating morphologicalpolarity. The first mechanism uses localizedpositive feedback to amplify responsiveness to the external signal at the front of the cell.The feedback loop promotes new assembly ofactin polymers, which form the pseudopod,push it forwards, and facilitate accumulationof a lipid signal in the plasma membrane.

To confine positive feedback to the leadingedge, a rapidly diffusing inhibitor acts toinhibit actin polymerization and membraneprotrusion outside the pseudopod. Thismechanism depends on assemblies of actinand the contractile protein myosin, located in the cell cortex at the rear and sides, whichinhibit protrusion. Protrusion and contrac-tion are controlled by different small enzymesthat belong to the Rho family of proteins,which hydrolyse the nucleotide GTP (and are therefore called Rho GTPases). These twomechanisms cooperate to generate and stabi-lize an internal polarity that greatly exceedsthe external gradient, allowing cells to detectand crawl persistently up shallow gradients of attractant and, in the case of neutrophils, to polarize in response to uniform attractant(usually in a randomly chosen direction).

In the positive feedback loop, the attrac-tant causes an increase in the concentrationof phosphatidylinositol-3,4,5-trisphosphate(PIP3), a membrane lipid that promotes RhoGTPase activation. In turn, the GTPases pro-mote actin polymerization in a pseudopod atthe leading edge, where the GTPases and newactin polymers together promote furtherPIP3 accumulation, amplifying the responseto attractant at the front of the cell. The feed-back loop is best documented in neutrophils,but PIP3 also accumulates at the front of theamoeba Dictyostelium and of fibroblast cells.

Inhibition at other parts of the cell con-fines the positive feedback loop to the leadingedge, by promoting cortical actin–myosinassemblies and reducing membrane accumu-lation of PIP3. Disrupting the actin–myosinassemblies makes the back of a neutrophil assensitive to attractant as the front — reversingthe gradient causes the cell simply to reverseits direction, rather than performing a U-turn.Although necessary for long-term asymmetricsensitivity to chemoattractants, this contrac-tile network is not absolutely required for

polarization itself — disruption of the net-work allows neutrophils and Dictyosteliumto form PIP3-rich membrane protrusionstowards a source of attractant, although thecells do form multiple pseudopodia at theirsides, and crawl haltingly in the correct direc-tion. In both cell types, membranes at the rearand sides fail to accumulate PIP3; in Dictyo-stelium, this resistance results from preferen-tial association of a PIP3-degrading enzymewith these membranes. The actin–myosinnetwork and mechanism of resistance to PIP3 accumulation cooperate, re-enforcingone another to interrupt the positive feedbackloop that is necessary for protrusion.

During chemotaxis, gradient interpreta-tion is intimately linked to the morphologicalpolarity required for forward movement. Thetwo essential cytoskeletal elements supportdiametrically opposite effects on PIP3 accu-mulation and attractant sensitivity, each limiting the proportion of the cell surface thatis susceptible to regulation by the other. Thisconversation between opposite ends of the cell may be a self-organizing property ofcytoskeletal systems that create cell polarity.

In one example of such a self-organizingsystem, round fragments of fish keratocytesrespond to transient pressure on one side by polarizing and crawling away. Contractileactin–myosin complexes form on the pres-sured side, and actin polymerizes on theopposite side to form a protruding pseudo-pod; polarity and motility persist long afterthe stimulus disappears. In this case, as inchemotaxis, the front and back of the cellprobably converse both mechanically (byasymmetric distribution of cortical tension)and biochemically (using Rho GTPases, PIP3and other signals). A neutrophil decides cor-rectly where to go because its ‘nose’ and itsactin–myosin ‘muscles’ reciprocally regulateone another. These clever cells would scornbrains as an unnecessary luxury. ■

Henry Bourne is in the Departments of Medicine andof Cellular and Molecular Pharmacology, Universityof California, San Francisco, California 94143, USA.Orion Weiner is in the Department of Cell Biology,Harvard Medical School, Cambridge, Massachusetts 02115, USA.

FURTHER READINGComer, F. I. & Parent, C. A. Cell 109, 541–544 (2002).Verkhovsky, A. B. et al. Curr. Biol. 9, 11–20 (1999).Weiner, O. D. Curr. Opin. Cell Biol. 14, 196–202 (2002).Zigmond, S. H. J. Cell Biol. 75, 606–616 (1977).

A chemical compassconcepts

NATURE | VOL 419 | 5 SEPTEMBER 2002 | www.nature.com/nature 21

Cell polarityOnly now are we beginning tounravel the mechanisms behind acell’s ability to point in any directionand navigate effectively.

Stretching the point: the single-celled Amoebaproteus demonstrates its locomotor pseudopodia.

EY

E O

F SC

IEN

CE

/SP

L

© 2002 Nature Publishing Group