Paul Weiss was probably the first to advance the view that the guidance of cells by oriented fibrils of the extracellular matrix is a fundamental mechanism oi embryonic development. Its potential importance for directing cellular migrations during embryogenesis is now widely recognised and evidence is emerging that it may also play a much more subtle role. Albert Harris and colleagues have demonstrated that not only does the matrix influence the cells but the cells can remodel the matrix: by exerting traction on the local extracellular matrix. populations of cells can cause the surrounding matrix to become aligned by tension and thus influence the direction of locomotion of other cells over a wide region. This feedback loop may give rise to instabilities which can lead to the generation of pattern. Although we are only just beginning to understand the hitherto unsuspected properties that can emerge from such a complex dynamical system. it already seems likely that cell guidance has a much more pervasive function than merely directing some cellular migrations in the embryo; it may be a component of an extensive system of mechanical communication which plays a central role in coordinating embryonic development. A recent paper(') by Clark and colleagues from the Departments of Cell Biology and Electrical and Electronic Engineering in Glasgow describes the guidance of cultured cells on highly artificial substrata consisting of very fine and very precise parallel grooves manufactured by the latest techniques in nanotechnology. The reader who is unfamiliar with this field may be forgiven for wondering why such precision is necessary and what possible relevance such studies could have to cell guidance by the extracellular environment in vivo. Before discussing the significance of this latest work, I must therefore try to place it in the context of current problems in the field. Despite the importance of cell guidance in embryogenesis and despite the fact that it was discovered in the earliest days of tissue culture, it has received little attention and is still poorly understood in comparison with, for example, chemotaxis. We still do not knon7 whether it is the physical shape of an aligned extracellular matrix, its adhesive properties or even some mechanical property such as anisotropic elasticity that is the most important factor. It is simple enough to construct a model of an aligned extracellular matrix in
tissue culture: stretched plasma clots, drained hydrated collagen lattices and many other aligned fibrillar modcls have been shown to guide cultured cells very effectively. But it has proved surprisingly difficult to isolate the property or properties responsible for guidance even in these fibrillar models. Considerable progress has been made with other substrata, however, in discovering which properties can elicit cell guidance. Using artificial substrata that are carefully designed to have clearly defined properties, we can now be reasonably certain that the shape of the substratum and its adhesive properties can both influence the direction of cell locomotion. Guidance by anisotropic elasticity of the substratum seems probable on theoretical grounds but, although there is now some experimental evidence for this, no artificial substratum has yet been made to test it rigorously. Unfortunately, these experiments with artificial substrata have not yet succeeded in eliminating any of the properties that might be responsible for guiding cells on aligned fibrillar matrices but they have revealed guidance responses that are potentially important in their own right regardless of their relevance to fibrillar guidance. Cell guidance thus appears to be an assemblage of cellular responses to diverse properties of the substratum and more than one response may be involved in guidance by fibrillar matrices. Morphographic guidance or the response to substratum shape (sometimes inappropriately called topographic guidance) is probably the most studied and yet the least understood. Finely grooved surfaces and cylindrical surfaces are two main categories of artificially shaped substrata that guide cells very effectively. both originally introduced by Paul Weiss and colleagues, and it is still not clear whether both forms of guidance have a common mechanism. The paper by Clark et a [ . ( ' ) concerns subtrata with fine parallel grooves. These are of special interest since they have been considered models of the morphographical properties of aligned fibrillar matrices. Although they do not duplicate the shape of a fibrillar matrix very realistically, they have the advantage of rigidity which avoids the complication of unknown mechanical properties possibly contributing to the response. Clark and colleagues have found that some cells are exquisitely sensitive to grooved surfaces. The new ultrafine grooves have a repeat spacing of only 260 nm: an order of magnitude better than had been achieved previously and now truly comparable with the dimensions of extracellular fibrils. The depth of these grooves is as little as 100nm and yet they can guide some types of cell with high efficiency: MDCK cells show little sign of approaching the limit of their sensitivity. Although it has long been suspected that some cells might have such a high sensitivity, there was no firm evidence until now. The most persuasive of the revious evidence wa5 the finding of Ohara and BuckR) that very fine scratches made with metal polish would guide cells effectively but local variations in the inicroproperties of these
'random' scratches, such as scratch depth and spacing, can give rise to local variations in macroproperties such as wettability and adhesiveness. Only with substrata of the highest precision and uniformity can we be certain that any observed response is evoked directly by the morphography of the grooves and not, for example, by linear patterns of adhesiveness caused by variations in groove spacing. High precision and uniformity of grooved substrata has been achieved in the past by the use of ruling engines, similar to the machines uscd for producing diffraction gratings; and more recently by techniques of microfabrication borrowed from the microelectronics industry. Clark and colleagues have pushed the limits to thc brink of the submicroscopic domain using a new technique that is not only very simple in principle but does not require elaborate equipment such as is needed for electron beam lithography. It relies on the interference of two plane wavefronts, obtained by splitting an argon laser beam, to produce a pattern of parallel fringes on a quartz slide coated with photoresist. A fundamental limit is that the smallest repeat period obtainable cannot be less than half the wavelength of the laser but this still leaves plenty of scope for using lasers with shorter wavelengths. High precision is also necessary for investigating the mechanism of the response. Parallel grooves of rectangular cross section have three parameters: depth, width and repeat. spacing. Any hypothesis must eventually account for the variation in response in relation to thcse three variables. Only groove 'depth is varied in these latest experiments but Clark and colleagues have previously investigated much lar er repeat spacings(3) and single steps in the substratum , With single steps, a depth of 1,um appeared to be close to the limit of sensitivity for BHK cells but this is reduced to 300 nm for grooves of 6 pm repeat and to less than l00nm for the new grooves of 26Onm repeat. Referring to two current hypotheses of cell guidance by shaped substrata, the authors of this latest paper conclude that 'It is difficult to relate any of these to guidance elicited by features as small as in the present study.! To understand why, we must look at these hypotheses in some detail. The first was proposed by Heath and myself(') and arose from studies of cell guidance on cylindrical surfaces: the other category of shaped substrata known to guide cells. These are conceptually much simpler than grooved surfaces and can bc described by a single parameter: radius of curvature. It emerged that fibroblasts are highly sensitive curvature and would align on glass fibres up to 200pm in diameter. We speculated that the cell must have some sort o f accurate straightedge in order to detect such slight curvatures and actin cables, extending obliquely into the cytoplasm from sites of adhesion, were obvious candidates since they are kept perfectly straight by tension. After obtaining experimental evidence that the distribution of actin cables could indeed be influenced by the shape of
6)
the substratum, we proposed that the actin cables could not become assembled in a bent state and hence the cfficiency with which the cell could exert traction on the substratum was reduced in directions of high convex curvature; this would be expected to limit the directions in which a cell could travel on non-planar surfaces and thus account for cell guidance. Heath and I never proposed at the time that our hypothesis might also account for guidance by grooved surfaccs, bccause we thought that the fibroblasts would probably bridge over the grqoves, but later observations by Brown and myself(") showed that this was not always the case and that the Dunn and Heath hypothesis might account for guidance by grooved surfaces. The second hypothesis, proposed by Ohara and Buck@)! also assumed that actin cables and their associated adhesions or focal contacts are important. They obtained grooves with repeat spacings down to 5 p m and observed that cells could indeed bridge the grooves. They suggested that elongated focal contacts, confined to the ridge crests between grooves, would necessarily be oriented along the ridges if these were sufficiently narrow. This would be expected to influence the direction of cell movement since actin cables are generally co-aligned with their associated focal contacts. A problem with this hypothesis is that, as already mentioned, it is now known that cells do not always bridge grooves and Brown and 1(6),Brunette(7) and have found that cells can form recently Clark et focal adhesions to the side and bottom of even quite fine grooves. In their latest paper"), Clark and colleagues consider that the new 100nm dcep groovcs are too shallow to distort actin cables significantly and that the width of the ridges between them (130 nm) is too small to accomodate focal contacts (which arc commonly 1 or 2 ,um wide on plane surfaces). Furthermore, increasing the depth of the grooves up to 400nm significantly increases the alignment of BHK cells and the elongation of MDCK cells which implies that bridging of the grooves is not taking place since, in the words of the authors, 'There should be no difference between bridging a 100nm deep groove and a 400nm deep groove.' One of the team of authors, Adam Curtis, has forwardcd another hypothesis lo explain the new data. This is not published yet but is mentioned in a recent review of cell guidance by Curtis and Clark@). It apparently has some elements of the Ohara and Buck hypothesis and states that attachment sites and actin condensations form at discontinuities or sharp edges in the substratum; but insufficient details are available at present to discuss its full implications. Nevertheless, the new results do give some support to the view that focal contacts and tbcir associatcd actin filaments may yet prove to be important. The growth cones of chick cerebral neurones are apparently insensitive to the new ultrafine grooves even thou h they had pre,viously been shown by the same author$) to be highly responsive to grooves of 8,um period and
2pm depth. Though amply supplied with filopodia which have often been implicated as sensing elements; growth cones alinost totally lack well-formed focal contacts or their associated actin cables. Although Heath and 1 were never too worried by the possibility that cells lacking focal contacts or actin cables might also be guided, since such cells presumably still transmit tension to the substratum via actin filaments('), it is encouraging for our hypothesis that cells with a less organised filament system may be far less sensitive to substratum shape. Furthermore, thc case for our hypothesis is also supported by the new evidence that even these very closely spaced grooves are not bridged by the cells. Although thc 100nm dcep grooves might seem to be too shallow to distort actin cables, the fact that the response is strongly depth-dependent does imply that some part of the cell is distorted by sinking into the grooves. Indeed, the reason for thc increased efficacy of ultrafine grooves in guiding cells with focal contacts may well be that the focal contacts can no longer avoid the grooves. Each focal contact may span several grooves and become corrugated by sinking into them (see Fig. 1).If so, any actin filament attached to a part of the contact that lies within a groove would be restricted in the range of directions in which it can exert
force. Any attempt by the ccll to pull on the actin filament in a direction other than parallel with the groove would result in the filament becoming distorted and possibly disrupted by being forced into the plasma membrane supported by the wall of the groove. This would reduce t h e efficiency of a whole actin cable pulling in this direction which would lead to the dominance of other actin cables that lie parallel to the grooves and hence to the alignment of the ccll. Whatever the eventual explanation of guidance by these ultrafine grooves, this work of Clark and colleagues has revealed another important instance of the acute sensitivity of cells to their external environment and simultancously has opened up a fascinating new rcgion for exploration of the mechanism of morphographic guidance. Only time will tell which, if any, of the current hypotheses can best predict the dependence of the response on the three groove parameter?. Other groove profiles, particularly asymmetric ones such as saw tooth profiles, may be necded for more rigorous tests and may even reveal new asymmetric guidance responses that have relevance in vivo. But perhaps the central issue is still the one that most occupied Weiss: whether morphographic guidance is a unified rcsponsc with a single mcchariism, as exemplified by the Dunn and Heath hypothesis, or whether the mechanism of response to surface curvature is quite different from that of the response to ultrafine surface features.
References 1 CLLRK, P., CONNOI LY. P., CVRTIY. A. S. G . , Dow, .I. A. T. 4 m WITKIXSOK, C . 1). W.(1991). Cell guidance 11) ultrafine topography in v h ~ J.. Cell ScL. 99, 73 7Y. 2 OHARA. P. 'I. AND BUCK.K. C . (1979). Contact guitlnnce in virrn. A light.
transmission and scanning electron microscopic study. Exp. Cell Re,. 121, 235-249. 3 CLARK, P , CONPIOLLY. P.. CURTIS. A. S. G., DUw, J. A. T. AND WILKINSON, C. D. W. (1990). Topographical control of cell hehaviour: 11. Multiple grooved substrata. Dcvplo/,meni 108, 635-604. 4 CLARK;. P., CONNULLY.P., CURTIS, A. S . G . ,Dow, J. A. 7. .mri WILKINSON, C. L). 1V. (1987). 'l'opogmphicnl control of cell behauioiir: 1. Simple step cues. Development 99. 439-448. 5 DLWN,G. A. AND HEATH, .I. P. (1976). A ncw hypothcqis of contact guidancc in tirsue cells. Exp. Cd/Rrs. 101. 1-14, 6 D u n , G . A. NU BROWN. A. 1'. (1986). Alignment of fibroblaits on grooved surfaces described by a umple geometric Lraniformation. J. Cell Sci. 83, 313-36. 7 B K U N ~ I ID. E , M. (19SO). Fibroblasts on micromachined substrata orient hicrarchically to grooves of differcnt dimcnsions. E.rp. CeN Res. 164, 11-26. 8 CURTIS,A. S. G. AND C1.4RK, P. (1990). Thc cffcctr of topngraphic and mechanical properties of materials on cell behaviour. Crit. R a . . Riocr~mpor.5 , 343-362. 9 DUKN.'3. A. (1982). Contact guidance of cultured tissue cells: a survey of potentially relwanl properties o l the subatraturn. In C ' d l Hrhimiour. (ed. R. Bellairb, A. Curtis a n d G . Dunn). Cambridge Cniversity Prehs; Cambridge, pp 247-280. r----
Fig. 1. A schematic section through a cell a n d o n e o f its focal contacts on grooves of 260 nm period and 100 tim d e p t h showing how some actin filaments within t h e actin cable may be c om e distorted a n d conscquently disrupted if the cell be c om e s disoriented so t h a t t h e tension in the cable is misaligned with t h e direction of t h r grooves.
Graham Dunn 1s filth thc MRC Musclc and Cell Motility Unit, Biophysics Section, King'b Collcge London. 26-29 Drury Lane, London WC2B 5RL, UK.
tissue culture: stretched plasma clots, drained hydrated collagen lattices and many other aligned fibrillar modcls have been shown to guide cultured cells very effectively. But it has proved surprisingly difficult to isolate the property or properties responsible for guidance even in these fibrillar models. Considerable progress has been made with other substrata, however, in discovering which properties can elicit cell guidance. Using artificial substrata that are carefully designed to have clearly defined properties, we can now be reasonably certain that the shape of the substratum and its adhesive properties can both influence the direction of cell locomotion. Guidance by anisotropic elasticity of the substratum seems probable on theoretical grounds but, although there is now some experimental evidence for this, no artificial substratum has yet been made to test it rigorously. Unfortunately, these experiments with artificial substrata have not yet succeeded in eliminating any of the properties that might be responsible for guiding cells on aligned fibrillar matrices but they have revealed guidance responses that are potentially important in their own right regardless of their relevance to fibrillar guidance. Cell guidance thus appears to be an assemblage of cellular responses to diverse properties of the substratum and more than one response may be involved in guidance by fibrillar matrices. Morphographic guidance or the response to substratum shape (sometimes inappropriately called topographic guidance) is probably the most studied and yet the least understood. Finely grooved surfaces and cylindrical surfaces are two main categories of artificially shaped substrata that guide cells very effectively. both originally introduced by Paul Weiss and colleagues, and it is still not clear whether both forms of guidance have a common mechanism. The paper by Clark et a [ . ( ' ) concerns subtrata with fine parallel grooves. These are of special interest since they have been considered models of the morphographical properties of aligned fibrillar matrices. Although they do not duplicate the shape of a fibrillar matrix very realistically, they have the advantage of rigidity which avoids the complication of unknown mechanical properties possibly contributing to the response. Clark and colleagues have found that some cells are exquisitely sensitive to grooved surfaces. The new ultrafine grooves have a repeat spacing of only 260 nm: an order of magnitude better than had been achieved previously and now truly comparable with the dimensions of extracellular fibrils. The depth of these grooves is as little as 100nm and yet they can guide some types of cell with high efficiency: MDCK cells show little sign of approaching the limit of their sensitivity. Although it has long been suspected that some cells might have such a high sensitivity, there was no firm evidence until now. The most persuasive of the revious evidence wa5 the finding of Ohara and BuckR) that very fine scratches made with metal polish would guide cells effectively but local variations in the inicroproperties of these
'random' scratches, such as scratch depth and spacing, can give rise to local variations in macroproperties such as wettability and adhesiveness. Only with substrata of the highest precision and uniformity can we be certain that any observed response is evoked directly by the morphography of the grooves and not, for example, by linear patterns of adhesiveness caused by variations in groove spacing. High precision and uniformity of grooved substrata has been achieved in the past by the use of ruling engines, similar to the machines uscd for producing diffraction gratings; and more recently by techniques of microfabrication borrowed from the microelectronics industry. Clark and colleagues have pushed the limits to thc brink of the submicroscopic domain using a new technique that is not only very simple in principle but does not require elaborate equipment such as is needed for electron beam lithography. It relies on the interference of two plane wavefronts, obtained by splitting an argon laser beam, to produce a pattern of parallel fringes on a quartz slide coated with photoresist. A fundamental limit is that the smallest repeat period obtainable cannot be less than half the wavelength of the laser but this still leaves plenty of scope for using lasers with shorter wavelengths. High precision is also necessary for investigating the mechanism of the response. Parallel grooves of rectangular cross section have three parameters: depth, width and repeat. spacing. Any hypothesis must eventually account for the variation in response in relation to thcse three variables. Only groove 'depth is varied in these latest experiments but Clark and colleagues have previously investigated much lar er repeat spacings(3) and single steps in the substratum , With single steps, a depth of 1,um appeared to be close to the limit of sensitivity for BHK cells but this is reduced to 300 nm for grooves of 6 pm repeat and to less than l00nm for the new grooves of 26Onm repeat. Referring to two current hypotheses of cell guidance by shaped substrata, the authors of this latest paper conclude that 'It is difficult to relate any of these to guidance elicited by features as small as in the present study.! To understand why, we must look at these hypotheses in some detail. The first was proposed by Heath and myself(') and arose from studies of cell guidance on cylindrical surfaces: the other category of shaped substrata known to guide cells. These are conceptually much simpler than grooved surfaces and can bc described by a single parameter: radius of curvature. It emerged that fibroblasts are highly sensitive curvature and would align on glass fibres up to 200pm in diameter. We speculated that the cell must have some sort o f accurate straightedge in order to detect such slight curvatures and actin cables, extending obliquely into the cytoplasm from sites of adhesion, were obvious candidates since they are kept perfectly straight by tension. After obtaining experimental evidence that the distribution of actin cables could indeed be influenced by the shape of
6)
the substratum, we proposed that the actin cables could not become assembled in a bent state and hence the cfficiency with which the cell could exert traction on the substratum was reduced in directions of high convex curvature; this would be expected to limit the directions in which a cell could travel on non-planar surfaces and thus account for cell guidance. Heath and I never proposed at the time that our hypothesis might also account for guidance by grooved surfaccs, bccause we thought that the fibroblasts would probably bridge over the grqoves, but later observations by Brown and myself(") showed that this was not always the case and that the Dunn and Heath hypothesis might account for guidance by grooved surfaces. The second hypothesis, proposed by Ohara and Buck@)! also assumed that actin cables and their associated adhesions or focal contacts are important. They obtained grooves with repeat spacings down to 5 p m and observed that cells could indeed bridge the grooves. They suggested that elongated focal contacts, confined to the ridge crests between grooves, would necessarily be oriented along the ridges if these were sufficiently narrow. This would be expected to influence the direction of cell movement since actin cables are generally co-aligned with their associated focal contacts. A problem with this hypothesis is that, as already mentioned, it is now known that cells do not always bridge grooves and Brown and 1(6),Brunette(7) and have found that cells can form recently Clark et focal adhesions to the side and bottom of even quite fine grooves. In their latest paper"), Clark and colleagues consider that the new 100nm dcep groovcs are too shallow to distort actin cables significantly and that the width of the ridges between them (130 nm) is too small to accomodate focal contacts (which arc commonly 1 or 2 ,um wide on plane surfaces). Furthermore, increasing the depth of the grooves up to 400nm significantly increases the alignment of BHK cells and the elongation of MDCK cells which implies that bridging of the grooves is not taking place since, in the words of the authors, 'There should be no difference between bridging a 100nm deep groove and a 400nm deep groove.' One of the team of authors, Adam Curtis, has forwardcd another hypothesis lo explain the new data. This is not published yet but is mentioned in a recent review of cell guidance by Curtis and Clark@). It apparently has some elements of the Ohara and Buck hypothesis and states that attachment sites and actin condensations form at discontinuities or sharp edges in the substratum; but insufficient details are available at present to discuss its full implications. Nevertheless, the new results do give some support to the view that focal contacts and tbcir associatcd actin filaments may yet prove to be important. The growth cones of chick cerebral neurones are apparently insensitive to the new ultrafine grooves even thou h they had pre,viously been shown by the same author$) to be highly responsive to grooves of 8,um period and
2pm depth. Though amply supplied with filopodia which have often been implicated as sensing elements; growth cones alinost totally lack well-formed focal contacts or their associated actin cables. Although Heath and 1 were never too worried by the possibility that cells lacking focal contacts or actin cables might also be guided, since such cells presumably still transmit tension to the substratum via actin filaments('), it is encouraging for our hypothesis that cells with a less organised filament system may be far less sensitive to substratum shape. Furthermore, thc case for our hypothesis is also supported by the new evidence that even these very closely spaced grooves are not bridged by the cells. Although thc 100nm dcep grooves might seem to be too shallow to distort actin cables, the fact that the response is strongly depth-dependent does imply that some part of the cell is distorted by sinking into the grooves. Indeed, the reason for thc increased efficacy of ultrafine grooves in guiding cells with focal contacts may well be that the focal contacts can no longer avoid the grooves. Each focal contact may span several grooves and become corrugated by sinking into them (see Fig. 1).If so, any actin filament attached to a part of the contact that lies within a groove would be restricted in the range of directions in which it can exert
force. Any attempt by the ccll to pull on the actin filament in a direction other than parallel with the groove would result in the filament becoming distorted and possibly disrupted by being forced into the plasma membrane supported by the wall of the groove. This would reduce t h e efficiency of a whole actin cable pulling in this direction which would lead to the dominance of other actin cables that lie parallel to the grooves and hence to the alignment of the ccll. Whatever the eventual explanation of guidance by these ultrafine grooves, this work of Clark and colleagues has revealed another important instance of the acute sensitivity of cells to their external environment and simultancously has opened up a fascinating new rcgion for exploration of the mechanism of morphographic guidance. Only time will tell which, if any, of the current hypotheses can best predict the dependence of the response on the three groove parameter?. Other groove profiles, particularly asymmetric ones such as saw tooth profiles, may be necded for more rigorous tests and may even reveal new asymmetric guidance responses that have relevance in vivo. But perhaps the central issue is still the one that most occupied Weiss: whether morphographic guidance is a unified rcsponsc with a single mcchariism, as exemplified by the Dunn and Heath hypothesis, or whether the mechanism of response to surface curvature is quite different from that of the response to ultrafine surface features.
References 1 CLLRK, P., CONNOI LY. P., CVRTIY. A. S. G . , Dow, .I. A. T. 4 m WITKIXSOK, C . 1). W.(1991). Cell guidance 11) ultrafine topography in v h ~ J.. Cell ScL. 99, 73 7Y. 2 OHARA. P. 'I. AND BUCK.K. C . (1979). Contact guitlnnce in virrn. A light.
transmission and scanning electron microscopic study. Exp. Cell Re,. 121, 235-249. 3 CLARK, P , CONPIOLLY. P.. CURTIS. A. S. G., DUw, J. A. T. AND WILKINSON, C. D. W. (1990). Topographical control of cell hehaviour: 11. Multiple grooved substrata. Dcvplo/,meni 108, 635-604. 4 CLARK;. P., CONNULLY.P., CURTIS, A. S . G . ,Dow, J. A. 7. .mri WILKINSON, C. L). 1V. (1987). 'l'opogmphicnl control of cell behauioiir: 1. Simple step cues. Development 99. 439-448. 5 DLWN,G. A. AND HEATH, .I. P. (1976). A ncw hypothcqis of contact guidancc in tirsue cells. Exp. Cd/Rrs. 101. 1-14, 6 D u n , G . A. NU BROWN. A. 1'. (1986). Alignment of fibroblaits on grooved surfaces described by a umple geometric Lraniformation. J. Cell Sci. 83, 313-36. 7 B K U N ~ I ID. E , M. (19SO). Fibroblasts on micromachined substrata orient hicrarchically to grooves of differcnt dimcnsions. E.rp. CeN Res. 164, 11-26. 8 CURTIS,A. S. G. AND C1.4RK, P. (1990). Thc cffcctr of topngraphic and mechanical properties of materials on cell behaviour. Crit. R a . . Riocr~mpor.5 , 343-362. 9 DUKN.'3. A. (1982). Contact guidance of cultured tissue cells: a survey of potentially relwanl properties o l the subatraturn. In C ' d l Hrhimiour. (ed. R. Bellairb, A. Curtis a n d G . Dunn). Cambridge Cniversity Prehs; Cambridge, pp 247-280. r----
Fig. 1. A schematic section through a cell a n d o n e o f its focal contacts on grooves of 260 nm period and 100 tim d e p t h showing how some actin filaments within t h e actin cable may be c om e distorted a n d conscquently disrupted if the cell be c om e s disoriented so t h a t t h e tension in the cable is misaligned with t h e direction of t h r grooves.
Graham Dunn 1s filth thc MRC Musclc and Cell Motility Unit, Biophysics Section, King'b Collcge London. 26-29 Drury Lane, London WC2B 5RL, UK.
