What is the scientific explanation for creeping

People are often amazed, even startled, to hear that many of their body cells are mobile and tend to crawl about. This mobility is essential for us: otherwise wounds would not heal, the blood would not clump and close injured veins, and the immune system would not be able to attack infectious intruders.

Sometimes, however, the crawling of cells is also involved in undesirable pathological processes, such as progressive tissue-destroying inflammation and the arteriosclerotic narrowing of blood vessels, which cause the risk of a heart attack. It is also related to the fact that cancer cells spread in the body: If the degenerated cells only grew in an uncontrolled manner, i.e. no metastases would arise, the patient could be cured by surgically removing the compact tumor.

For as long as we have known about the crawling of cells, there have been all sorts of strange ideas. In 1786 the Danish biologist Otto F. Müller described a forward-striving cell as a "transparent, gelatinous body with a glass-like extension". (The terms gel, gelatine come from the Latin gelare for "solidify", "freeze".) This observation - that the mechanical state of the cell can vary and change in places (today we speak of a sol-gel transformation) - has been made Proven to be very helpful in exploring the mechanism of cell movement and better understanding the molecular components involved.

Recently, opportunities have even emerged to develop new treatment strategies for some difficult-to-fight diseases. Infections and cancer are definitely included, but possibly also cystic fibrosis or cystic fibrosis, a congenital severe metabolic disorder with the secretion of thick mucus, which is particularly noticeable in the lungs and digestive glands.

Different cell types - same principle

Cancer cells migrate comparatively slowly, at 0.1 to 1 micrometer (thousandths of a millimeter) per hour. The cells crawl in a similarly leisurely manner in healing wounds. The defense cells of the immune system, on the other hand, as well as those blood cells that have to ensure rapid wound closure, move much faster.

Against invading pathogens, a person produces more than 100 billion specific white blood cells (neutrophilic granulocytes) every day. They come from the bone marrow, from which they migrate, to flow with the blood through the body for a few hours until they crawl out of the capillaries and penetrate into other tissues. They can migrate 30 microns a minute as they scour the skin, airways, and digestive tract for microbes (Image 1). Overall, such a defense cell covers several millimeters at its destination. When added up, the total path of all neutrophils in the body per day results in twice the circumference of the earth.

The blood platelets or thrombocytes, which are responsible for the rapid closure of injured blood vessels, do not migrate properly, but they change their shape with agile movements, similar to a crawling cell, when they clog a wound with their body. As long as they are floating in the blood, they are pretty compact little discs. At their destination they quickly transform into flatter structures and develop numerous outgrowths; they now look almost like unevenly rolled pancakes. They are ideally suited to seal off injured blood vessels (Fig. 4).

In the light microscope you can see that cells stretch out the runners of their cortex when they creep and later melt them down again. Such processes, like the cortical region (the cell cortex), appear clear and homogeneous, unlike the inside of the cell, which is permeated by many different organelles.

Cells that migrate are stimulated by - often specific - external stimuli. White blood cells, for example, follow the traces of all possible signal substances from microorganisms or from infected and injured tissues. Growth factors that trigger cell division can also stimulate directed cell movement. The change in shape of the platelets is caused by the coagulation factor thrombin.

Most of these signaling substances bind to receptors on the cell surface. A chain of molecular reactions then takes place in the cell, translating the signal into instructions to respond. This process triggers changes in the cortex - and the cell begins to crawl.

Obviously, some stimuli can also induce such remodeling without switching on membrane receptors. Such a stimulus are, for example, low temperatures. Platelets change their shape irreversibly when it is cold, which is a problem for storage in blood banks: If you need them for transfusions, you must not refrigerate them beforehand.

When a cell begins to crawl, the bark appears to flow out, and in the direction of movement - on the guide border - a transparent, leaf-like foot, called a lamellipodium, is created. Fine, hair-like runners (filopodia) are also responsible for the cell membrane to grow for such a small foot, which also have the task of pulling incorporated objects towards the cell body.

The feet gain hold on the substrate by means of special attachment proteins (membrane or cell adhesion molecules that react with other molecules). This gives the cell enough pull to move a little bit forward. Then the appendage detaches itself from the ground and pushes itself forward a little further. However, all the individual steps of locomotion - formation of runners, attachment, contraction and detachment - are often so fluidly coordinated that it looks as if the cell is sliding along like a cloud in front of a mountain.

The cell body acts like a liquid droplet - a sol - that yields to a force. However, if you poke the foot with a fine needle, or if you try to suck it into a cannula, it will resist the deformation; it also behaves like a gel, i.e. like a liquid with elastic properties. The same applies to the whole cell. It deforms under mechanical tension, but has, so to speak, a memory of the previous shape, which it assumes again as soon as the deforming force ceases. The relationship between the applied stress and the measured elongation is the modulus of elasticity.

A gel also has other important properties. It holds liquid in which its components were dissolved in the sol state in the interstices of its molecular structure - like a sponge holds water in its pores - and delays the passage. The cell cortex owes both properties, elasticity and water retention capacity, to the water-soluble polymers of the cell plasma. These are the same chain-like proteins that provide the framework for the forces of contraction during movement.

Dennis C.H. Bray from the British Medical Research Council has developed a hydrostatic model for the mechanism of cell movement, in which the transitions between the sol and gel state are at the center: The cell is therefore practically a sol that is enclosed in a gel layer. The more stable edge becomes contractile in the event of irritation. Because the inside of the cell cannot be compressed as a whole, nothing changes in shape at first - until the outer layer weakens at the first or most strongly stimulated point; there the hydrostatic pressure from the inside now gently bulges the cell membrane outwards. However, because the contents of the bulge immediately gel and stabilize the new foot, it immediately becomes clear and transparent (Fig. 2). If the disturbance in the gel layer does not arise entirely on the outside of the conduction seam, but more at the base, then the bulge pulls back into the cell body.

The model can explain the creeping of cells, as far as it can be observed microscopically, well. But how is such a gel structured, and how can the cell substance involved change quickly and evenly from one state to the other?

Parallels to muscle contraction

When research into this phenomenon began in many places in the early 1970s, the proteins actin and myosin were at the top of the candidate list for contractile elements in the cell cortex. It had been known since the 1940s that these are the decisive mobile components in the skeletal muscles, and they had then also been demonstrated in other, amoeboid mobile cells. Actin makes up 10 percent of the proteins in neutrophils and even 20 percent in platelets; it is found mainly in the cortex and in the lamellipodia.

Actin is available in two forms: as a globular protein and as a thread-like polymer made up of two strands of globular subunits wound around each other (Figure 3 a). Myosin (highly schematic in Figure 3) looks much more complicated. This heaviest and longest known natural protein has globular, flexible heads on a long shaft that bind to actin filaments and can be split off from the rest of the molecule; In the muscle, such molecules aggregate into bundles that lie between the actin filaments and pull on them. As was shown in 1963, the myosin binds itself to actin with the heads at an acute angle, so that a filament covered with it looks like a harpoon under the electron microscope, which is why one end is pointed, the other is called hooked or bearded - or harpoon end.

During a muscle contraction, the two types of filaments slide along each other in opposite directions, with the actin pushing itself forward towards its tip. The energy for this is provided by adenosine triphosphate (ATP), which the myosin heads split. Calcium ions, which act on two regulatory proteins attached to actin - troponin and tropomyosin - are important for muscle contraction.

As it turned out in the mid-1970s, calcium also controls the mechanochemical activity of myosin. In non-muscular cells, it indirectly controls the phosphorylation (binding of phosphate groups) to the myosin heads, which can then exert traction on the actin. Other enzymes inactivate myosin by removing the phosphate.

All of these findings led us to suspect at the time that crawling cells, in response to a stimulus, contract the actin network in the cortex by changing the calcium level and activating myosin.

My first own work in this area was on the structure of the acting gel. I started at Harvard University Medical School in Cambridge, Massachusetts in 1974 with John H. Hartwig. The first thing we discovered was that when a white blood cell extract was stirred under certain conditions, large amounts of actin precipitated - along with an unknown high molecular weight protein that we purified and called actin binding protein (ABP).

Around the same time, Robert E. Kane of the University of Hawaii at Manoa reported that liquid extracts from sea urchin eggs gel after some time. The substance obtained in this way was filled with actin filaments. As it turned out later, extracts from very different cell types behave similarly; and together with a lot of actin, small amounts of actin-binding protein were always found, so that we made this responsible for the gelling for the time being.

In fact, we succeeded in demonstrating that the ABP can abruptly increase the elasticity of an actin solution: Even with just one molecule in 1000 acting globulins, which are present as filaments, the liquid becomes more gelatinous. No other actin-binding molecule was remotely as effective.

We now considered how the gel formation comes about. If you put a lot of stiff chopsticks in a container and shake it vigorously, they will form into parallel bundles by themselves for energetic reasons. We believed that something similar would happen with actin filaments left to their own devices in the cell cortex, whereby various cell proteins could provide the skeleton with additional support through transverse stiffening. Such tightly connected parallel struts made of actin probably give the already mentioned fine filopodia their tensile strength. In addition, cross-linked actin bundles can form complexes with cell adhesion molecules, adhesion plaques that strengthen the hold of a cell process on the substrate. This structure would probably not be suitable for building up a uniform gel in a lamellipodium.

However, this would easily be achieved by a protein that forces the filaments to cross-link three-dimensionally in uniform meshes at right angles. Since our binding protein had such a strong gelling effect, it had to be able to do just that (Fig. 5 left). When we were able to check our concept in an electron microscope in 1981, we found it confirmed (Fig. 5, photo). I'll come back to that in a moment.

Usher in the actin framework

Its structure provided further information about the properties of the binding protein. It is a very large, thread-like molecule that combines with its peers to form a strand of double length (Fig. 5, left). One end of each unit attaches itself to the actin, the other tends to attach itself to the same part of a sibling molecule so that both form a large angle. Otherwise, the subunits along the strand contain various overlapping sections that give them more strength and enable them to keep the crosslinked actin filaments at a distance.

Paul A. Janmey from Harvard University measured the mechanical properties of gels made from these building blocks. If the actin framework is held together by ABP, it is extraordinarily strong and elastic; the usual concentrations of both components in cells should easily suffice to explain the rigidity of a pre-stretched lamellipodium. Even with a low concentration of ABP, such a gel can also retain water, as Tadanao Ito from the University of Kyoto (Japan) demonstrated during a stay with us. Further reinforcement for the presumed function of the binding protein was that in white blood cells it is mainly concentrated in the cortex; this was proven by another guest researcher in our laboratory, Olle I. Stendahl from the Medical Faculty of Linköping University (Sweden), with fluorescence-labeled antibodies.

To learn more about the microscopic structure of ABP-linked acting gels, Hartwig examined them using a high-resolution technique. The sample is shock-frozen in liquid helium so that the cell structures are preserved. The frozen water is then removed by sublimation in a vacuum. The rest is vaporized with precious metals so that the textures are visible in the electron microscope.

As mentioned, it could be seen that the actin threads - at least in the test tube - in the presence of ABP arrange themselves more or less at right angles and regularly to form a network, the orientation of the filament tips being random. On average, the actin threads were a micrometer long, and side branches branched off about every hundred nanometers (millionths of a millimeter). The network in feet of white blood cells looked almost exactly the same.

The ABP molecules actually sit at the intersection of the actin filaments and protrude into both arms. In the filopodia, on the other hand, as already mentioned, all actin filaments are arranged in parallel with the bearded end pointing away from the cell body.

We were best able to test whether ABP is actually as important for the structure of the lamellipodia as we believed in cells that have no binding protein. Our colleague C. Casey Cunningham led a study of the protein composition of cells from six cultures of melanoma, the most aggressive skin cancer. Three of the cell lines contained ABP and the other three did not. The former behaved just like crawling cells otherwise: They stretched out processes and migrated towards signaling substances. The others could indeed form normal filopodia; Otherwise, however, their bark seemed to be unstable, because lamellipodia did not send them out, and they did not move forward when stimulated. It looked as if any coordination was impossible for them: bubbles appeared everywhere and immediately disappeared again (Fig. 6). Even normal cells do this occasionally, but never continuously.

We suspected that the gel in the cortex of defective melanoma cells was too weak to control the internal pressure during contraction and to force it into controlled pathways. The result is an apparently uncontrolled wobbling or bubbling of the surface. The actin filaments cannot form a uniform network in such unstable vesicles, as would be a prerequisite for a gel. But there is at least a sufficiently coherent mass that the protuberance can be retracted again. We now transplanted an active gene that codes for ABP units into these cells, and with this we actually achieved that the blistering stopped or at least significantly decreased and the cells began to crawl (Fig. 6).

The construction crew

In order for a cell to migrate, the actingel has to reorganize itself: the scaffolding has to be dismantled at the back and attached at the front in the direction of movement (Fig. 8). Typically about half of the actin in a resting cell is not polymerized; the molecules can then move freely in the sol. Filaments only arise - in response to suitable signals - in certain places. Their amount, i.e. the proportion of polymerized actin, can remain roughly constant in a crawling cell, because the build-up in one place tends to be balanced out by the breakdown in another.

For spontaneous polymerization, as Fumio Oosawa from the University of Nagoya (Japan) found out, two or three globulins have to combine to form a core - like a crystallization nucleus. This doesn't happen too often; but once a germ is present, a filament is quickly formed on it, in which further globulins accumulate on both sides. However, it grows much faster at the bearded end - this is why this is called the plus and the other minus end (Fig. 3 a).

Two general groups of control proteins take care of the breakdown and the reconstruction. One - of which there are three subclasses - binds mainly or only to the actin subunits. Vivianne T. Nachmias and Daniel Safer from the University of Pennsylvania in Philadelphia found that these proteins prevent the spontaneous nucleation of actin as well as the accumulation of further subunits at the pointed end of a filament. In addition, they slow down the growth of the hooked end, but do not completely prevent this process; the second group is responsible there.

Helen Lu Yin, now at the University of Texas at Dallas, and I found the first of these proteins in white blood cell extracts in 1979. With a calcium level, such as that found in stimulated crawling cells, it covers the bearded end like a protective cap, which makes the accumulation of further units impossible. However, it also loosens the bonds of the globulins in the filament, thus breaking it down and also embracing the newly exposed hooked end (Fig. 7 above). Because it drastically shortens the actin filaments, it can turn the gel into a sol - that's why we call it gelsolin.

As a result, a number of working groups discovered a number of other proteins that break down actin filaments, block their bearded end, or do both at the same time. They are now also assigned to three subclasses: Those of the first - a large family with a basic structure similar to that of gelsolin, which belongs to it - sit on the end like a cap, and some also cut the filament. Those of the second - they are usually called Cap-Z proteins - have a different structure; Control substances of this type were first found in amoebas and in blood platelets. A third subclass protein was first discovered in brain tissue; these now four compounds - ADF, cofilin, depactin and actophorin - are common and have a weakly decomposing effect on actin filaments.

The members of the gelsolin family can be activated by calcium. However, withdrawing it does not cause you to let go of the actin. For a long time it was not known how the bond could be released until Janmey and I learned in 1987 about an observation by the Swedish researchers Ingrid Lassing and Uno Lindberg from Stockholm University: Polyphosphoinositide - phospholipids, which belong to the cell membrane and are involved in signal conversion in the cell are - decrease the affinity of profilin for globular actin. (Profilin, which Lindberg discovered in 1977, isolates acting globulins through its binding and thus prevents the formation of nuclei for new filaments.) We were able to show that these phospholipids stop gelsolin in two ways, namely by deliberately ending its degradation activity on the fibers and by cause it to let go of the filament end (pic 7). As experiments all over the world have shown in the meantime, the phospholipids block the binding activity of almost all such proteins, which break down actin filaments and attach themselves to their bearded ends.

How the puzzle fits together

From all of this, a model can be designed that explains the creeping of cells and includes both the signal conversion in the event of a stimulus and the transformation of the acting gel in the cell cortex (Fig. 8). For example, if a cell is stimulated by a signal substance, enzymes in the cell membrane begin to synthesize polyphosphoinositides or to destroy them elsewhere. As a result of the breakdown, calcium is released into the sol of the cell (it is stored in membrane-covered vesicles), which now activates those members of the gelsolin family that tend to attach themselves to actin filament ends. This chain of events disintegrates the actin framework. Conversely, newly synthesized phospholipids cause the caps of pieces of filament near the cell membrane to loosen so that they can grow again. How effective this uncovering is depends on the chemical environment. In the case of blood platelets, it is conceivable that phase changes occur in the cell membrane when it is cold, the phospholipids orient themselves differently forever and the actin mass gels irreversibly, so to speak.

So far we have hardly said anything about the involvement of myosin. In order for a cell to crawl, it is not enough for the actin gel to reorganize. It also has to contract, and for this myosin has to pull on the actin (as in picture 3 on the right). As mentioned, calcium plays a part in this by phosphorylating myosin, i.e. providing it with energy, and at the same time dissolving the actin structure in places - by activating gelsolin and related separation proteins - because the gel has to liquefy to such an extent that the actin fibers can be moved ; on the other hand, it must not flow apart again.

D. Lansing Taylor, who now works at Carnegie Mellon University in Pittsburgh, Pennsylvania, calls this coordination "sol formation-contraction coupling". He and his colleagues checked the plausibility of the concept with mixtures of actin filaments, actin binding protein and gelsolin. When acting globulins and short filaments blocked in growth by protein caps are released in the liquefying gel, they seep through the lamellipodium to the anterior membrane. There polyphosphoinositides free the filaments from their caps so that they can grow again. The long threads are then attached to the front of the scaffolding (Fig. 8).

Various research groups have found supporting evidence for this in living cells. Fluorescence-marked actin built into crawling fibroblasts (not yet differentiated connective tissue cells) in filaments on the guiding seam; and in cells stimulated with signaling substances, the blocked bearded ends of the actin filaments lost their caps. Another consistent finding was that the breakdown of actin filaments in the cortex of blood platelets depends on the increase in intracellular calcium ion concentration. In any case, proteins from the gelsolin group seem to be involved in cell locomotion.

The mechanisms described in detail so far - involving calcium, phospholipids and actin-binding proteins - are certainly not the only ones involved. As mentioned, the acting globulins also bind ATP and also ADP (adenosine diphosphate, which is formed by splitting off a phosphate group from ATP). ADP could also still provide energy, but subunits occupied by ATP polymerize better. Acting globulins that detach from the filament exchange the ADP for the more energetic ATP. Marie-France Carlier of the laboratory of the French National Center for Scientific Research (CNRS) in Gif-sur-Yvette and others suggested that the bond between actin and the energy supplier, as well as the exchange, would be catalyzed by profilin (which shields the subunits). Profilin would thus influence the ability of actin to polymerize, but also the structure of the filaments and other regulatory proteins.

Finally, it should be considered: As widespread as the creeping of cells with emitted feet is, not all cellular surface movements are based on the remodeling of acting gels. Timothy J. Mitchison of the University of California in San Francisco suspects that after cell division, when the daughter cells diverge, a separate class of one-headed myosin molecules pulls an actin scaffold to the cell periphery. They could walk along filopodia with bundles of actin like rails. These particular myosins were discovered by Thomas D. Pollard of the Johns Hopkins University in Baltimore, Maryland and Edward D. Korn of the American National Health Institutes in Bethesda, Maryland.

Filopodia seem to arise according to a different principle than lamellipodia. Lewis G. Tilney of the University of Pennsylvania was the first to demonstrate that actin accumulates in the early 1970s, and George F. Oster of the University of California at Berkeley described a process as the mechanism for the extension of these fine extensions: which he calls "Brownian pawl" based on the Brownian molecular movement. According to this idea, thermal fluctuations in the cell membranes would be used to control the agglomeration of actin and to accelerate the expulsion of a filopodium.

Help in clinical practice

The knowledge gained from such research is of general interest. For example, it would be medically very important if it were possible to drive crawling cells or to let them move more slowly. Perhaps their behavior could be adjusted differently with altered concentrations of gelsolin and other regulating proteins.

David J. Kwiatkowski from Harvard University, Casey Cunningham and I pursued this idea on cell cultures of genetically modified mouse fibroblasts that we set up: They contained the genetic information for human gelsolin and - apart from the usual amounts of mouse gelsolin - produced it in different quantities . As our tests showed, the more of them there was, the faster they moved.

At least under laboratory conditions, such interventions in the cell machinery can be successful. It's not difficult to imagine targeted applications. For example, by forcing the movement of fibroblasts, wound healing could perhaps be accelerated. Conversely, slowing down or stopping could possibly prevent tissue destruction by white blood cells in the event of inflammation, as well as an impending arterial thrombosis with the risk of infarction from blood platelets.

A possible application in the very near future for cystic fibrosis, which was previously difficult to treat, could be traced back to research carried out in 1979. Independently of each other, Astrid Fagraeus and René Norberg at the University of Uppsala (Sweden) and Christine Chaponnier and Giulio Gabbiani from the University of Genoa (Italy) discovered that there are also substances in the blood plasma that break down actin filaments. As other researchers found, two plasma proteins work together in this case: Gc-globulin (a genetically polymorphic protein that binds to actin) and a secreted form of gelsolin.

After an injury, the blood often contains dissolved actin; the serum levels of Gc globulin and gelsolin are also noticeably lowered. According to Stuart E. Lind of Harvard University and John G. Haddad of the University of Pennsylvania, extracellular actin can be toxic to tissues - even potentially fatal to the patient - because it interferes in complex ways with blood clotting. The two plasma proteins could prevent this from happening as part of an actin capture system.

We have recently suspected that a similar interaction between actin and a substance it traps might play a role in cystic fibrosis. The processes involved in this hereditary disease, which is the most common among Europeans, are still difficult to understand. In any case, because of a genetic defect, a protein that regulates the transport of chloride to cell membranes does not work, so that excretory cells secrete abnormal amounts of concentrated secretions - also in the lungs, which thereby become the focus of infections and inflammations. As the body tries to defend it, it soon becomes full of white blood cells. When these decompose, a tough, purulent mass is created that threatens to suffocate the patient. (Part of the daily routine is therefore to cough up the phlegm in a complex procedure.)

So far, the long strands of DNA from the cell nuclei of the white blood cells have been held responsible for the firm consistency of the sputum. For this reason, attempts have recently been made to use a genetically modified DNA-degrading enzyme (deoxyribonuclease I or DNAse I). Lung mucus has been shown to become more fluid in the test tube, and according to previous observations, inhalation of the active ingredient can also improve lung function. This suggested that he was actually splitting the obstructive DNA into short pieces.

But our research suggests a completely different mechanism. As early as 1963, Lindberg had isolated a protein that was not identified at the time and that inhibits the natural variant of DNAse I. Ten years later, during a stay at the Cold Spring Harbor Laboratory in the town of the same name in the US state of New York, he and Elias Lazarides identified it as actin. DNAse I now binds tightly to actin subunits, similar to Gc globulin; if it is sufficiently concentrated, it can certainly contribute to the shortening of these polymers - namely by preventing disintegrating filaments from growing again.

As expected, we found a significant amount of actin in the sputum of cystic fibrosis patients. This was not surprising, as the white blood cells contain about as much of it as DNA. Our assumption is that this actin intercepts the natural DNAse I, so that there is no longer enough enzyme available to break down the DNA threads; so it practically inhibits DNAse function.

The toughness of the sputum could also be reduced with plasma gelsolin. Apparently it is even more effective than the DNAse (Fig. 9).

Since DNAse I and gelsolin intervene in different ways, you should try to see whether a combination of both substances has a beneficial effect. As a normal extracellular component in the body, gelsolin should actually be non-toxic in the airways and should not provoke an immune reaction.

If it works, research into the mechanisms of cell movement would have made significant progress in the treatment of cystic fibrosis. Without such dry basic research as the one presented here on the gelling of actin in creeping cells, many such medical advances would never be possible.


- The Extracellular Actin-Scavenger System and Actin Toxicity. From W.M. Lee and R.M. Galbraith in: New England Journal of Medicine, Volume 326, Issue 20, Pages 1335-1341, May 14, 1992.

- Life at the Leading Edge: The Formation of Cell Protrusions. From J. Condeelis in: Annual Review of Cell Biology, Volume 9, pages 411 to 444, 1993.

- On the Crawling of Animal Cells. By Thomas P. Stossel in: Science, Volume 260, Pages 1086-1094, May 21, 1993.

- Phosphoinositides and Calcium as Regulators of Cellular Actin Assembly and Disassembly. By Paul A. Janmey in: Annual Review of Physiology, Volume 56, pages 169 to 191, 1994.

- Reduction in Viscosity of Cystic Fibrosis Sputum in Vitro by Gelsolin. From C.A. Vasconcellos et al in: Science, Volume 263, Pages 969-971, February 18, 1994.

From: Spektrum der Wissenschaft 11/1994, page 42
© Spektrum der Wissenschaft Verlagsgesellschaft mbH

This article is contained in Spectrum of Science 11/1994