- Email: [email protected]
Myosin I
Myosin ! E.D. Korn and J.A. Hammer ill Laboratory of Cell Biology, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland, USA Current Opinion in Cell Biology 1990, 2:57--61
Introduction
Heavy-chain sequences
The initial report of a myosin (hereinafter called myosin I to distinguish it from conventional myosins II) that had a single, relatively short heavy chain and did not form filaments, but did have actin-activated Mg-ATPase activity (Pollard and Kom, JBiol CIJem 1973, 248:4682-4690; Pollard and Kom, J Biol Chem 1973, 248:4691-4697), met with skepticism if not lack of interest. Retrospectively, this does not seem surprising because cell biologists had not yet focused at the molecular level, and myosin biochemists, who were fully occupied trying to understand muscle actomyosin, knew only too well how readily myosins II could be degraded by endogenous proteases to subfragment 1, the small, non-filamentous, but enzymatically active globular head that resembles myosin I. We now know, however, that Acanthamoeba castellanii contains at least three myosins I, DicO~stelium discoideum at least one, and that the 110 kD/calmodulin complex of avian intestinal brash border is also a myosin I. For a recent review of Acantbamoeba myosins I and II, see Kom et aL (J CellBiocbem 1988, 36:37-50); for a recent review of all non-muscle myosins I and II, see Kom and Hammer (Annu Rev Biophys Biophys Cbem 1988, 17:23-45); for reviews of vertebrate brash border myosin I, see [1] and Mooseker (Annu Rev Cell Bio11985, 1:209-241). Each of these five myosin I proteins contains a single heavy chain (.,,llO-140kD) composed of an .-.80kD amino-terminal domain, whose sequence is very sin'dlar to that of myosin II sub fragment 1, joined to a short carboxy-terminal domain (-.-35-50 "ld)), whose sequence shows no similarity to that of the myosin U rod-like tail; each expresses myosin-like actin-activated Mg-ATPase activity; four of the five have been shown to have myosinlike mechanochemical properties; none is capable of forming filaments; four of the five have been shown to be able to crosslink actin filaments; all are associated with the plasma membrane; one has been shown to be concentrated at the leading edge of locomoting cells. These properties account for the growing interest in the type I class of myosins and lead to interesting speculations about their possible biologic roles.
The heavy-chain sequences of Acanthamoeba myosin IB [2], myosin IC (Jung et al., Proc Natl Acad Sci USA 1987, 84:6720-6724) and DicOgstelium myosin I [3] have been deduced from their gene sequences. A bovine complementary DNA (cDNA) clone, which almost certainly encodes the bovine equivalent of chicken intestinal brash border myosin I, has also been sequenced (Hoshimam and Nakanishi, J Biol Chem 1987, 262:14625-14632). The sequence of the Acanthamoeba myosin I13 heavy chain is compared with the sequence of rat embryonic skeletal muscle myosin (Strehler et al., J Mol Biol 1986, 190:291-317), a typical myosin II, in Fig. la. The first ~80 amino-terminal residues are missing (this is a highly variable region in myosins II), but residues 1 to .-.670 are .,~60% similar to residues -,- 80-800 of muscle myosin II, which comprise the major portion of the globular head domain (subfragment 1). Regions of very high sequence conservation include the probable ATP-binding site and the reactive thiol region (although neither of the two cysteines SH-1 and SH-2 is present in the amoeba myosins I). The remainder of the myosin IB heavy chain, residues ,..671-1147, bears no resemblance to the muscle myosin II heavy chain. In particular, there is no evidence in this unconventional ,,, 50 kD carboxy-terminal domain of the seven-residue hydrophobic repeat that is characteristic of the coiled-coil, rod-like tail of myosin II. Interestingly, a segment of the myosin IB carboxy-terminal domain (residues ~910-1094) contains 48% glycine, 17% proline and 12% alanine (GPA-rich region).
Acanthamoeba myosin IC and DicO~ostelium myosin I can be aligned with Acanthamoeba myosin IB over essentially their entire lengths (see Fig. lb for percentage similarities to myosin I13). Thus, the high degree of similarity between them extends not only over their globular head domains but throughout their unconventional carboxyterminal domains. The only exception to this is the GPArich regions which, although present in the carboxy-terminal domains of all three protozoan myosins I, cannot be aligned in any unique way. What is conseived in these
Abbreviation
cDNA---complementaryDNA. (~) Current Science Ltd ISSN 0955-0674
57
58
Cytoplasm and cell motilily (a) 1
Globular head
mude I myosin II
1
ATP site
P-site
~850
Actin site I
Membrane binding ?
I
myosin IB
(b)
Acanthamoeba
I
'
~60%
myosin IC
Dictyoslelium myosin I
Brush border myosin I
I
''I
Actm site II
I
. . .0%. 902
1094 1147
--, 670
1
GPA 923 978 1034 1168 ~ 670 I ~64%
t
I
~72%
1
[
~690
922
10591111
I ~7OO/o
~80%
1 [
I
1147
1
]
1946
I
•-, 670
myosin IB
Acanthamoeba
Rod-like tail
I
~720 ~850~10301043 ~55%
Acanthamoeba
Fig. 1. Heavy-chain sequence comparisons between myosin IB and rat muscle myosin (a) and between myosin IB and the other myosin I sequences determined to date (b). The percentages refer to the degrees of similarity between the globular head sequences ([-1) and between the tail sequences [open areas; excludes the GPA-rich areas (m) and the unique areas of brush border
myosin 1 (Pa)].
regions is the unusual amino acid composition and net basic charge. The 110 kD brash border myosin I heavy chain ( l l 7 k D by deduced amino acid sequence) is composed of an ,--81kD globular head domain fused to a ,-~36kDcarboxy-terminal domain (Fig. lb). The latter region is completely distinct from myosin II but does not contain a GPA-rich region. It does contain, however, a ,-, 185residue segment which is ,--46% similar to a segment of the carboxy-terminal domain of myosin 1B [2].
Biochemical properties Some of the more salient physical and enzymatic properties of the five myosin I proteins that have been purified are summarized in Table 1. Noteworthy are the sizes of their heavy chains (consistent with the sequence data), their native molecular masses (which indicate that the molecules contain only one heavy chain) and, perhaps most important, their actin-activated Mg-ATPase activities (which we consider to be a necessatT property for classification as a myosin). Note that the actin-activated Mg-ATPase acti~Sties of tim Acantbamoeba myosins I and DicOwstelittm myosin I are regulated by phosphorylation of their heaD" chains and that, when full},phosphotTlated, their actMties are essentially equal to that of rab-
bit skeletal muscle myosin II. A specific Acantbamoeba myosin I heavy chain ldnase has been purified (Hammer et al., J Biol Ctoem 1983, 258:10168-10175) and recently shown to be activated by phosphatidylserine-enhanced phosphorylation (Brzeska et al., J Cell Bio11989, 109:276a). The locations of the major functional sites in the heavy chains of the Acantbamoeba isozymes have been determined (see Fig. la). The purine moiety of ATP can be photoaffinity-crosslinked to an amino acid about 11 kD from the amino-terminus (Lynch et aL, J Biol Ct2em 1987, 262:13842-13849; Jung el al., 1987; Atkinson et al., J Biol Cbem 1986, 261:1844-1848); sites located on both sides of an actin-protected trypsin-cleavage site ,-,64kD from the amino terminus (homologous to the 50kD/20kD junction of myosin II) contribute to the ATP-sensitive actin-binding site within the head domain (Brzeska el al., J Biol Cbem 1988, 263:427-435) [4], and the heavy chain phosphorylation site that regulates actin-activated MgATPase activity (Ser-315 in I13, Set-311 in IC, a corresponding Thr in IA [5] and probably Ser-328 in Dic~astelittm myosin I [3]) lies bem'een the actin-binding and ATP-binding sites. The sequence of the region containing this phosphorylation site has relatively little similarity to that of the corresponding region of myosin II subfragment 1, and it has been suggested that phosphorylation activates the Mg-ATPase acti~ity of actomyosin I by converting this region of its heavy chain into a con-
Myosin I Korn and Hammer
Table 1. Properties
of myosins I.
Myosin
Native
Heavy chain"
Light chain"
Acanthamoeba IAf Acan'thamoeba IBI Acanthamoeba IC§ Dictyostelium¶ Brush border"
159000 150000 162000 140 000
140000 125000 130000 117 000 110 000
17000 27000 14000 12 000 17 O00tt
K+, EDTA 22 21 15 11 0.7
Ca2+
Mg 2+
Mg 2+ + actin
2 4 4 0.6 1.1
0.3 0.3 0.2 -,- 1
18~ 17~ 20~ ,-, 10/; 0.3
0.04
"By sodium dodecyl sulfate-polyacrylamide gel electophoresis, fAIbanesi et al., I Biol Chem 1985, 260:8649-8652; Albanesi et al., l 8iol Chem 1985, 260:11174-11179 [7]. ~When heavy chain is phosphorylated. §[7]. ¶Cot6 et al., l Biol Chem 1985, 260:4543--4546. "'Conzelman and Mooseker I Cell 8iol 1987, 105:313-324. VfCalmodulin.
formation comparable to that which actosubfragment 1 has without phosphorylation [4]. As anticipated from the similarities to the sequence of muscle myosin subfragment 1, the ~ 8 0 k D amino-terminal fragments of Acanthamoeba myosin IA [4] and myosin I13 (Brzeska et al., 1988) have actin activated Mg-ATPase activities equal to those of the native molecules, and this is likely also to be true for brush border myosin I (Coluccio and Bretscher, J Cell Bio11988, 106:367-373) [6]. The Acanthamoeba myosins I contain a second actinbinding site in their carboxy-terminal domain (Lynch et al., J Biol Ctoem 1986, 261:17156-17162). This ATP-insensitive site, together with the ATP-sensitive actin-binding site in their amino-terminal domain, allows them to crosslink actin filaments (Fujisald et aL,JBiol Chem 1985, 260:11183-11189). DicO~stelium myosin I (Lynch et al., unpublished data) and brush border myosin I (Conzelman and Mooseker, J Cell Biol 1987, 105:313-324) are also capable of crosslinking actin filaments, but the location of the probable second actin-binding site has not been determined. The carboxy-terminal ,-.32kD region of brush border myosin I apparently contains the sites which bind calmodulin (Coluccio and Bretscher, 1988)
[6].
Mechanochemical properties Actomyosins II are mechanochemical enzymes, i.e. they convert the energy released by the hydrolysis of ATP into the 'walking' of myosin heads along the F-actin filaments. In vitro, this is modeled by the superprecipitat.ion of crosslinked actomyosin complexes and by the movement of myosin-coated beads, or membrane fragments with attached myosin, along actin cables. By virtue of their two actin-binding sites, Acanthamoeba myosins IA, IB and IC (Fujisald et al., 1985; Lynch et aL, 1986) [7] and DicO~astelium myosin I (Lynch et al., unpublished observations) exhibit ATP-dependent superprecipitation. Acanthamoeba myosins I also support the movement of latex beads along actin cables (Albanesi et al., J Biol Chem 1985, 260:8649-8652), albeit rather slowly (-,- 20-80 rim/s). Membrane vesicles with attached
Acanthamoeba myosin I move more rapidly ( --. 240 rim/s; Adams and Pollard, Nature 1986, 322:754-756). Similarly, beads coated with brush border myosin I move at an average rate of ,-.8nm/s [8], while intestinal microvillar membrane discs with associated myosin I move at ,-~12--(_~nm/s [9].
Intracellular localization A substantial fraction of the myosin I of Acantbamoeba is tightly associated with purified plasma membranes (Gadasi and Kom, Nature 1980, 206:452-456) [10] and possibly also with intracellular membrane vesicles (Adams and Pollard, 1986) [11]. The tight, reversible linkage (Kd ,-,30--50nmolAiter) does not seem to be via membrane-associated actin [10], although its dissociation at elevated ionic strength suggests that myosin I is a peripheral membrane protein. Acanthamoeba myosin I binds equally tightly to both membrane vesicles, from which peripheral proteins have been extracted by sodium hydroxide, and to vesicles of pure lipids [11]. It seems unlikely, however, that the latter interaction would provide the specificity that is presumably necessary biologically. The membrane-binding site on the myosin I may lie between the ATP-insensitive actin-binding site in the carboxy-terminal segment and the subfragment llike domain [11] (Doberstein and Pollard, J Cell Biol 1989, 109:86a). Furthermore, membrane-bound myosin I has at least its ATP-sensitive actin-binding sites available to bind exogenously added actin [10]. The myosin I associated with isolated brush border microvillar membranes also has an accessible ATP-sensitive actin-binding site [9], which is consistent with its function as the lateral connection between' the microvillar actin filament bundle and the plasma membrane (Matsudaira and Burgess, J Cell Biol 1979, 83:667-673). The membrane binding site in brush border myosin I probably resides within its carboxy-terminal domain [9]. Consistent with the in vitro observations, immunofluorescence studies of permeabilized Acanthamoeba demonstrate the close association of myosin I (and not myosin II) with the plasma membrane (Gadasi and Kom,
59
60
Cytoplasm and cell motility 1980; Hagen et al., J Cell Biol 1986, 103:2121-2128). Of even greater interest is the observation by Fukui et al. [12] that myosin I is concentrated at the leading edges of locomoting DicO~gstelium amoebae, i.e. in the lamellipodia and pseudopodia of chemotacting amoebae and amoebae undergoing cell division. Myosin I is also concentrated at the sites of formation of phagocytic cups. In all these regions, actin, but not myosin II, also occurs. Myosin II is concentrated at the rear of locomoting cells, where only a diffuse reaction for myosin I was observed, and in the contractile ring of dividing cells, which contains no detectable myosin I.
not, and was not intended to, exdude the essential participation of other processes at the leading edges o f motile cells. Motility almost certainly requires the integrated action o f multiple systems. Actin polymerization and filament network formation at leading edges [5,16] (Condeelis et al., Cell Motil Cytoske11988, 10:77-90; Bray and White, Science 1988, 239:883-889) will almost certainly be required, an increase in osmotic and turgor pressure [Oster and Perelson, J Cell Sci 1987, 9(suppl):156-165] may contribute, and membrane flow (Bretscher, S c i A m 1987, 257:44-50) may also be involved. Indeed, the movement of intracellular vesicles on which the latter hypothesis depends might result from the interaction of membrane-associated myosin I with actin filaments.
Perspectives
At this time, we need to learn more about the structure o f myosin I, the nature of its association with the plasma membrane and of its interaction with Factin in order to propose specific mechanisms by which membrane-myosin-actin complexes might function. We also need to determine the localization of myosin I in locomoting cells at the level of electron microscopy. We need to understand the molecular details of the association of myosin I with membranes and of its linkage to membrane receptors for the ligands that initiate motile activity. And, finally, genetic studies to correlate gene deletions with phenotypic behaviour will be important in defining the biologic role(s) of myosins I in cell motility. The latter approach will be greatly complicated, however, if all cells contain multiple myosin I isoforms, as do Acanthamoeba and probably DicO~stelium (Titus et al., J Cell Bio11989, 109:8a), in addition to myosin II and possibly other classes of myosins, especially if there is redundancy of function.
The known examples of myosin I are limited to the fwe isozymes discussed in this article. It seems likely, however, that similar enzymes are widely distributed [e.g. yeast (Goodson et al., J CellBio11989, 109:84a)] and that the subfragment 1 region of myosin, which is sufficient to produce both movement and force (Toyoshima et al., Nature 1987, 328:536-539; Kishino and Yanagida, Nature 1988, 334:74-76), is anchored to membranes or actin illaments by a variety of carboxy-terminal domains that differ from the filament-forming tail-domain that anchors myosins II. Other variations on this theme are likely to occur. Indeed, the ninaCgene of Drosophila encodes ~ ' o proteins, necessary for proper formation of photoreceptor microviUi, which comprise an amino-terminal ldnase domain followed by a subfragment 1 domain and terminating in a domain unlike that of either a myosin II or a myosin I (Montell and Ruben, Cell 1988, 52:757-772). Furthermore, a gene encoding yet another type of Acantlmmoeba myosin, distinct from both myosins I and II, has been cloned and sequenced (Horowitz and Hammer, J Cell Bio11989, 109:282a). These proteins are likely harbingers of interesting future discoveries. What are the cellular functions o f myosins I? Brush border myosin I appears to play a role in maintaining microvillar structure, but it may well play other roles in the cell, such as vesicular transport. As for the protozoan myosins I, from their differential localizations in locomoting DicO~astelium , Fukui et al. [12] suggest that actomyosin I contributes to the forces that generate extension at the leading edges to form pseudopodia and lamellipodia, while actomyosin II contracts the rear of the cell and drives the cell mass forward. Consistent with this idea is the earlier demonstration that DicO~stelium amoebae lacking the myosin II heavy chain (De tozanne and Spudich, Science 1987, 236:1086-1091; Knecht and Loomis, Science 1987, 236:1081-1086) form pseudopodia at an essentially normal rate but that expansion of pseudopodia and cell movement are inhibited [13], and the suggestion that Acanttmmoeba motility is slowed but not stopped by injection o f antibodies against myosin II [14]. Fukui etaL [12] further speculate that myosin I may function similarly in cells where its presence has yet to be demonstrated (leukocytes, macrophages, fibroblasts, neuronal growth cones). However, this proposal does
Annotated references and recommended reading • ••
Of interest Of outstanding interest
1. •
l.ot.rVm, D D: The function of the major cTtoskeletal components of the brush border. Otrr Opin Cell Biol 1989, 1:51-57.
An up-to-datere-,'iewemphasizingvillinand brush border myosin I and describing the brush border cell as a model system for examining the assemblyand dynamicsof 2 os'toskeletalnetworks: the micro~llus and the terminal web. 2. JUNO G, SCII.,.~3TCJ, tLL~.LMERJA III: Myosin I hea,,T-chain • genes of Acanthamoeba castellanii: cloning of a second and es"idence for the existence of a third isoform. Gene 1989, 82:269-280. The deduced heaw-chain amino acid sequences of 2 Acanttmmoeba myosins I are highly consen-ed throughout, reflecting the very similar ph)~ical and biochemicalproperties of the purified isoforms~Regions of sequence consen~tion ~hich are distinct from m)x3sin11sequences could prove usefulas heterologous probes to search for myosinI genes in other organisms. 3. JUNOG, SAXECL Ul, YdSt~ttLAR, It~.~t~ERJA IU: DtcO'ostelium • discoideum contains a gene encoding a myosin I hea~T chain. Proc Nail Acad Sci USA 1989, 86:6186-6190. These results prove unequivocallythat Dico~steliumcontains a myosin I. In addition to encoding a poblaeptide that is remarkablysimilar to
M y o s i n I Korn a n d H a m m e r
the myosins 1 of Acantlmmoeba, this Dico~stelium gene is upregulated during starxution-induced aggregation. The feasibility of doing gene disruptions in DicO~gstelium should now allow a direct test for the ira t,it~9 function(s) of myosin I. 4. •
BRZESKAtt, LYNCH TJ, Komq ED: The effect of actin and phosphorylation on the tO'ptic cleavage pattern o f Acanthamoeba myosin IA. J Biol O)em 1989, 17:10243-10250. ltea'.y-chain phosphoQ:lation significantly enhances F-actin protection o f a t~'ptic cleavage site ~ 3 8 k D from the amino terminus in myosin IA. This may reflect a conformational change in the myosin globular head domain that switches the conformation of actomyosin I between productive and non-productive states. 5. e
BRZESKAH, LYNCtl TJ, KOm'q ED: The localization and sequence o f the phosphoD-lation sites of Acanthamoeba myosin I: an improved method for locating the phosphorylated amino acid. J Biol Chem 1989, 264:19340-19348. This information is important to efforts directed at understanding how phosphot3~ation regulates enz)Tnatic activity. The results also provide the correct assignment of the cloned genes to the purified protein isoforms. 6. •
CAm3oxlJM, COh7~EL~'; KA, AO&~tS RA, KAISERDA, POLLME) TD, MOOSEKERMS: Structural and immunological characterization o f the myosin.like 110-kD subunit of the intestinal micro~4_llar 1 lOK-calmodulin complex: evidence for discrete myosin head and calrnodulin.binding domains. J Cell Biol 1988, 107:1749-1757. These results proxide support for the idea that brush border myosin I is composed of a myosin globular head domain, v-herein resides its actinactivated MgX+.ATPase acuity, fused on an uncom'entional carboxyterminal domain, v.tmrein resides the calmodulin binding sites. 7. •
LYr:CHTJ, BW_.eSKAtt, Koe~'~ ED: Purification and characterization o f a third isoform o f myosin I from Acanthamoeba castellaniL J Biol Chem 1989, 264:19333-19339. All 3 isoforms are very similar in their physical and biochemical properties, suggesting that their in t,it~9function(s) may overlap considerably. 8. ••
MOOSEKERMS, COLESt&'qTR: The 110-kD protein.calmodulin complex o f the intestinal microvillus (brush border myosin I) is a mechanoenzyme. J Cell Biol 1989, 108:2395-2400. An important result sho'Mng that brash border myosin I moves beads along Nitella actin cables in a unidirectional and ATP-dependent manner. This finding demonstrates that, like all other myosins, brash border myosin I is a mechanoenzyme, and opens the w-ay to exploring how calcium, calmodulin, phosphoDSation, etc. regulate movement produced by this enzyme. 9.
MOOSEKERMS, CON7.EI2,1gNKA, COLEM.&NTR, HEUSERJE, SHEE'fZ MP: Characterization of intestinal microvillar membrane disks: detergent-resistant membrane sheets enriched in associated brush border myosin I (ll0K-calmodulin). J Cell Biol 1989, 109:1153--1161. The movement of these discs, vA-fichis supported by brush border myosin I, is considerably faster than the movement of latex beads coated with purified brush border myosin I and does not require high calcium levels. •
10. ee
~,LrYATA1t, BO~XRS B, KORN ED: Plasma membrane association of Acanthamoeba myosin I. J Cell Biol 1989, 109:1519-1528.
Along with [11], this study indicates that myosin I binds to membranes directly, i.e. not through membrane-assodated actin, but directly to membrane lipid and/or integral membrane proteins. This study suggests that the membrane attachment site is near the ATP-insensi~'e actin binding site present within the unconventional carboxy-terminal domain of m)~osin I. Studies like these are cmdal to efforts directed at understanding the specific organization of myosin l-membrane-actin complexes. ADA.~t5RJ, POLLARD"I'D: Binding o f myosin I to membrane ee lipids. Nature 1989, 340:566-568. This important study demonstrates that myosin I binds directly to both Acantlmmoeba membranes from ~taich peripheral proteins have been extracted and to pure lipid vesicles. The results are in keeping with pre~ions findings, including the immunofluorescent localization of myosin I to regions at and near the plasma membrane and the obser~ution that membranous vesicles from Acantbamoeba move on actin filaments by a motor tentatively identified as myosin I. 11.
12 • •
FUKUIY, LYNCHTJ, BRZESKAH, KORN ED: Myosin I is located at the leading edges of locomoting DicO,ostelium amoebae. Nature 1989, 341:328-331. A very important result which s h o w that myosin I and myosin II have distinct Iocali7~qtions in chemotacting and dividing DicO~stelium cells. Of particular importance are the obserrations that myosin I is concentrated at the leading edge of chemotacting cells and in phagocytic cups. The authors postulate that myosin I generates a projectile force at the leading edge of chemotacting cells, while myosin It, which is concentrated in the posterior cortex of the cell, propels the cell mass forward. It is also postulated that myosin l plays a direct role in the membrane events invoked in phagoc31osis. 13.
WESSELSD, SOU. DR, KNECHT D, LoOStlS WF, DE LOZAN,'NEA, SPUDICHJ: Cell motility and chemotaxis in Dictyostelium amoebae lacking myosin heavy chain. Dev Biol 1988, 128:164-177. A detailed ana~,~is o f the motility o f DicO~stelium myosin It cells using the dynamic morphology s)~tem, revealing the wealth of information that can be gained with this s3~tem. The results reveal abnormalities in locomotion, chemotaxis and the extent of pseudopod and lamellopod formation in cells lacking myosin IL These defects are in addition to the pr~iously reported impairment of tTtokinesis. • •
14. •
SL',aStDJH, POLLARDT: Microinjection into Acanthamoeba castellanti o f monoclonal antibodies to myosin-It slows but does not stop cell locomotion. Cell Motil Cytoskeleton 1989, 12:42-52. The results of this study suggest that while myosin II plays a role in retraction o f the trailing edge of locomoting ceils, the cell either roufinely uses or can draw upon another cytoplasmic motor to maintain locomotion. A strong candidate for this other motor is myosin I. 15. SXLJa.LJV: Microfilament-based motility in non-muscle cells. • Curr Opin Cell Biol 1989, 1:75-79. An excellent review on the molecular mechanisms of cell movement, with particular emphasis on the possible role of actin pobrnerization in generating a protrusive force at the leading edge of a cell. 16. S.~tmt SJ: Neuronal cytomechanics: the actin-based motility • of growth cones. Science 1988, 242:708--715. A recent review describing the current knowledge of the neuronal grox~zh cone cytoskeleton and the role it must play in extension, navigation, branching and target location during neuronal path finding. Emphasis is placed on actin pobrnerization at the leading edge as a possible dm'ing force in growth cone movement.
61
Introduction
Heavy-chain sequences
The initial report of a myosin (hereinafter called myosin I to distinguish it from conventional myosins II) that had a single, relatively short heavy chain and did not form filaments, but did have actin-activated Mg-ATPase activity (Pollard and Kom, JBiol CIJem 1973, 248:4682-4690; Pollard and Kom, J Biol Chem 1973, 248:4691-4697), met with skepticism if not lack of interest. Retrospectively, this does not seem surprising because cell biologists had not yet focused at the molecular level, and myosin biochemists, who were fully occupied trying to understand muscle actomyosin, knew only too well how readily myosins II could be degraded by endogenous proteases to subfragment 1, the small, non-filamentous, but enzymatically active globular head that resembles myosin I. We now know, however, that Acanthamoeba castellanii contains at least three myosins I, DicO~stelium discoideum at least one, and that the 110 kD/calmodulin complex of avian intestinal brash border is also a myosin I. For a recent review of Acantbamoeba myosins I and II, see Kom et aL (J CellBiocbem 1988, 36:37-50); for a recent review of all non-muscle myosins I and II, see Kom and Hammer (Annu Rev Biophys Biophys Cbem 1988, 17:23-45); for reviews of vertebrate brash border myosin I, see [1] and Mooseker (Annu Rev Cell Bio11985, 1:209-241). Each of these five myosin I proteins contains a single heavy chain (.,,llO-140kD) composed of an .-.80kD amino-terminal domain, whose sequence is very sin'dlar to that of myosin II sub fragment 1, joined to a short carboxy-terminal domain (-.-35-50 "ld)), whose sequence shows no similarity to that of the myosin U rod-like tail; each expresses myosin-like actin-activated Mg-ATPase activity; four of the five have been shown to have myosinlike mechanochemical properties; none is capable of forming filaments; four of the five have been shown to be able to crosslink actin filaments; all are associated with the plasma membrane; one has been shown to be concentrated at the leading edge of locomoting cells. These properties account for the growing interest in the type I class of myosins and lead to interesting speculations about their possible biologic roles.
The heavy-chain sequences of Acanthamoeba myosin IB [2], myosin IC (Jung et al., Proc Natl Acad Sci USA 1987, 84:6720-6724) and DicOgstelium myosin I [3] have been deduced from their gene sequences. A bovine complementary DNA (cDNA) clone, which almost certainly encodes the bovine equivalent of chicken intestinal brash border myosin I, has also been sequenced (Hoshimam and Nakanishi, J Biol Chem 1987, 262:14625-14632). The sequence of the Acanthamoeba myosin I13 heavy chain is compared with the sequence of rat embryonic skeletal muscle myosin (Strehler et al., J Mol Biol 1986, 190:291-317), a typical myosin II, in Fig. la. The first ~80 amino-terminal residues are missing (this is a highly variable region in myosins II), but residues 1 to .-.670 are .,~60% similar to residues -,- 80-800 of muscle myosin II, which comprise the major portion of the globular head domain (subfragment 1). Regions of very high sequence conservation include the probable ATP-binding site and the reactive thiol region (although neither of the two cysteines SH-1 and SH-2 is present in the amoeba myosins I). The remainder of the myosin IB heavy chain, residues ,..671-1147, bears no resemblance to the muscle myosin II heavy chain. In particular, there is no evidence in this unconventional ,,, 50 kD carboxy-terminal domain of the seven-residue hydrophobic repeat that is characteristic of the coiled-coil, rod-like tail of myosin II. Interestingly, a segment of the myosin IB carboxy-terminal domain (residues ~910-1094) contains 48% glycine, 17% proline and 12% alanine (GPA-rich region).
Acanthamoeba myosin IC and DicO~ostelium myosin I can be aligned with Acanthamoeba myosin IB over essentially their entire lengths (see Fig. lb for percentage similarities to myosin I13). Thus, the high degree of similarity between them extends not only over their globular head domains but throughout their unconventional carboxyterminal domains. The only exception to this is the GPArich regions which, although present in the carboxy-terminal domains of all three protozoan myosins I, cannot be aligned in any unique way. What is conseived in these
Abbreviation
cDNA---complementaryDNA. (~) Current Science Ltd ISSN 0955-0674
57
58
Cytoplasm and cell motilily (a) 1
Globular head
mude I myosin II
1
ATP site
P-site
~850
Actin site I
Membrane binding ?
I
myosin IB
(b)
Acanthamoeba
I
'
~60%
myosin IC
Dictyoslelium myosin I
Brush border myosin I
I
''I
Actm site II
I
. . .0%. 902
1094 1147
--, 670
1
GPA 923 978 1034 1168 ~ 670 I ~64%
t
I
~72%
1
[
~690
922
10591111
I ~7OO/o
~80%
1 [
I
1147
1
]
1946
I
•-, 670
myosin IB
Acanthamoeba
Rod-like tail
I
~720 ~850~10301043 ~55%
Acanthamoeba
Fig. 1. Heavy-chain sequence comparisons between myosin IB and rat muscle myosin (a) and between myosin IB and the other myosin I sequences determined to date (b). The percentages refer to the degrees of similarity between the globular head sequences ([-1) and between the tail sequences [open areas; excludes the GPA-rich areas (m) and the unique areas of brush border
myosin 1 (Pa)].
regions is the unusual amino acid composition and net basic charge. The 110 kD brash border myosin I heavy chain ( l l 7 k D by deduced amino acid sequence) is composed of an ,--81kD globular head domain fused to a ,-~36kDcarboxy-terminal domain (Fig. lb). The latter region is completely distinct from myosin II but does not contain a GPA-rich region. It does contain, however, a ,-, 185residue segment which is ,--46% similar to a segment of the carboxy-terminal domain of myosin 1B [2].
Biochemical properties Some of the more salient physical and enzymatic properties of the five myosin I proteins that have been purified are summarized in Table 1. Noteworthy are the sizes of their heavy chains (consistent with the sequence data), their native molecular masses (which indicate that the molecules contain only one heavy chain) and, perhaps most important, their actin-activated Mg-ATPase activities (which we consider to be a necessatT property for classification as a myosin). Note that the actin-activated Mg-ATPase acti~Sties of tim Acantbamoeba myosins I and DicOwstelittm myosin I are regulated by phosphorylation of their heaD" chains and that, when full},phosphotTlated, their actMties are essentially equal to that of rab-
bit skeletal muscle myosin II. A specific Acantbamoeba myosin I heavy chain ldnase has been purified (Hammer et al., J Biol Ctoem 1983, 258:10168-10175) and recently shown to be activated by phosphatidylserine-enhanced phosphorylation (Brzeska et al., J Cell Bio11989, 109:276a). The locations of the major functional sites in the heavy chains of the Acantbamoeba isozymes have been determined (see Fig. la). The purine moiety of ATP can be photoaffinity-crosslinked to an amino acid about 11 kD from the amino-terminus (Lynch et aL, J Biol Ct2em 1987, 262:13842-13849; Jung el al., 1987; Atkinson et al., J Biol Cbem 1986, 261:1844-1848); sites located on both sides of an actin-protected trypsin-cleavage site ,-,64kD from the amino terminus (homologous to the 50kD/20kD junction of myosin II) contribute to the ATP-sensitive actin-binding site within the head domain (Brzeska el al., J Biol Cbem 1988, 263:427-435) [4], and the heavy chain phosphorylation site that regulates actin-activated MgATPase activity (Ser-315 in I13, Set-311 in IC, a corresponding Thr in IA [5] and probably Ser-328 in Dic~astelittm myosin I [3]) lies bem'een the actin-binding and ATP-binding sites. The sequence of the region containing this phosphorylation site has relatively little similarity to that of the corresponding region of myosin II subfragment 1, and it has been suggested that phosphorylation activates the Mg-ATPase acti~ity of actomyosin I by converting this region of its heavy chain into a con-
Myosin I Korn and Hammer
Table 1. Properties
of myosins I.
Myosin
Native
Heavy chain"
Light chain"
Acanthamoeba IAf Acan'thamoeba IBI Acanthamoeba IC§ Dictyostelium¶ Brush border"
159000 150000 162000 140 000
140000 125000 130000 117 000 110 000
17000 27000 14000 12 000 17 O00tt
K+, EDTA 22 21 15 11 0.7
Ca2+
Mg 2+
Mg 2+ + actin
2 4 4 0.6 1.1
0.3 0.3 0.2 -,- 1
18~ 17~ 20~ ,-, 10/; 0.3
0.04
"By sodium dodecyl sulfate-polyacrylamide gel electophoresis, fAIbanesi et al., I Biol Chem 1985, 260:8649-8652; Albanesi et al., l 8iol Chem 1985, 260:11174-11179 [7]. ~When heavy chain is phosphorylated. §[7]. ¶Cot6 et al., l Biol Chem 1985, 260:4543--4546. "'Conzelman and Mooseker I Cell 8iol 1987, 105:313-324. VfCalmodulin.
formation comparable to that which actosubfragment 1 has without phosphorylation [4]. As anticipated from the similarities to the sequence of muscle myosin subfragment 1, the ~ 8 0 k D amino-terminal fragments of Acanthamoeba myosin IA [4] and myosin I13 (Brzeska et al., 1988) have actin activated Mg-ATPase activities equal to those of the native molecules, and this is likely also to be true for brush border myosin I (Coluccio and Bretscher, J Cell Bio11988, 106:367-373) [6]. The Acanthamoeba myosins I contain a second actinbinding site in their carboxy-terminal domain (Lynch et al., J Biol Ctoem 1986, 261:17156-17162). This ATP-insensitive site, together with the ATP-sensitive actin-binding site in their amino-terminal domain, allows them to crosslink actin filaments (Fujisald et aL,JBiol Chem 1985, 260:11183-11189). DicO~stelium myosin I (Lynch et al., unpublished data) and brush border myosin I (Conzelman and Mooseker, J Cell Biol 1987, 105:313-324) are also capable of crosslinking actin filaments, but the location of the probable second actin-binding site has not been determined. The carboxy-terminal ,-.32kD region of brush border myosin I apparently contains the sites which bind calmodulin (Coluccio and Bretscher, 1988)
[6].
Mechanochemical properties Actomyosins II are mechanochemical enzymes, i.e. they convert the energy released by the hydrolysis of ATP into the 'walking' of myosin heads along the F-actin filaments. In vitro, this is modeled by the superprecipitat.ion of crosslinked actomyosin complexes and by the movement of myosin-coated beads, or membrane fragments with attached myosin, along actin cables. By virtue of their two actin-binding sites, Acanthamoeba myosins IA, IB and IC (Fujisald et al., 1985; Lynch et aL, 1986) [7] and DicO~astelium myosin I (Lynch et al., unpublished observations) exhibit ATP-dependent superprecipitation. Acanthamoeba myosins I also support the movement of latex beads along actin cables (Albanesi et al., J Biol Chem 1985, 260:8649-8652), albeit rather slowly (-,- 20-80 rim/s). Membrane vesicles with attached
Acanthamoeba myosin I move more rapidly ( --. 240 rim/s; Adams and Pollard, Nature 1986, 322:754-756). Similarly, beads coated with brush border myosin I move at an average rate of ,-.8nm/s [8], while intestinal microvillar membrane discs with associated myosin I move at ,-~12--(_~nm/s [9].
Intracellular localization A substantial fraction of the myosin I of Acantbamoeba is tightly associated with purified plasma membranes (Gadasi and Kom, Nature 1980, 206:452-456) [10] and possibly also with intracellular membrane vesicles (Adams and Pollard, 1986) [11]. The tight, reversible linkage (Kd ,-,30--50nmolAiter) does not seem to be via membrane-associated actin [10], although its dissociation at elevated ionic strength suggests that myosin I is a peripheral membrane protein. Acanthamoeba myosin I binds equally tightly to both membrane vesicles, from which peripheral proteins have been extracted by sodium hydroxide, and to vesicles of pure lipids [11]. It seems unlikely, however, that the latter interaction would provide the specificity that is presumably necessary biologically. The membrane-binding site on the myosin I may lie between the ATP-insensitive actin-binding site in the carboxy-terminal segment and the subfragment llike domain [11] (Doberstein and Pollard, J Cell Biol 1989, 109:86a). Furthermore, membrane-bound myosin I has at least its ATP-sensitive actin-binding sites available to bind exogenously added actin [10]. The myosin I associated with isolated brush border microvillar membranes also has an accessible ATP-sensitive actin-binding site [9], which is consistent with its function as the lateral connection between' the microvillar actin filament bundle and the plasma membrane (Matsudaira and Burgess, J Cell Biol 1979, 83:667-673). The membrane binding site in brush border myosin I probably resides within its carboxy-terminal domain [9]. Consistent with the in vitro observations, immunofluorescence studies of permeabilized Acanthamoeba demonstrate the close association of myosin I (and not myosin II) with the plasma membrane (Gadasi and Kom,
59
60
Cytoplasm and cell motility 1980; Hagen et al., J Cell Biol 1986, 103:2121-2128). Of even greater interest is the observation by Fukui et al. [12] that myosin I is concentrated at the leading edges of locomoting DicO~gstelium amoebae, i.e. in the lamellipodia and pseudopodia of chemotacting amoebae and amoebae undergoing cell division. Myosin I is also concentrated at the sites of formation of phagocytic cups. In all these regions, actin, but not myosin II, also occurs. Myosin II is concentrated at the rear of locomoting cells, where only a diffuse reaction for myosin I was observed, and in the contractile ring of dividing cells, which contains no detectable myosin I.
not, and was not intended to, exdude the essential participation of other processes at the leading edges o f motile cells. Motility almost certainly requires the integrated action o f multiple systems. Actin polymerization and filament network formation at leading edges [5,16] (Condeelis et al., Cell Motil Cytoske11988, 10:77-90; Bray and White, Science 1988, 239:883-889) will almost certainly be required, an increase in osmotic and turgor pressure [Oster and Perelson, J Cell Sci 1987, 9(suppl):156-165] may contribute, and membrane flow (Bretscher, S c i A m 1987, 257:44-50) may also be involved. Indeed, the movement of intracellular vesicles on which the latter hypothesis depends might result from the interaction of membrane-associated myosin I with actin filaments.
Perspectives
At this time, we need to learn more about the structure o f myosin I, the nature of its association with the plasma membrane and of its interaction with Factin in order to propose specific mechanisms by which membrane-myosin-actin complexes might function. We also need to determine the localization of myosin I in locomoting cells at the level of electron microscopy. We need to understand the molecular details of the association of myosin I with membranes and of its linkage to membrane receptors for the ligands that initiate motile activity. And, finally, genetic studies to correlate gene deletions with phenotypic behaviour will be important in defining the biologic role(s) of myosins I in cell motility. The latter approach will be greatly complicated, however, if all cells contain multiple myosin I isoforms, as do Acanthamoeba and probably DicO~stelium (Titus et al., J Cell Bio11989, 109:8a), in addition to myosin II and possibly other classes of myosins, especially if there is redundancy of function.
The known examples of myosin I are limited to the fwe isozymes discussed in this article. It seems likely, however, that similar enzymes are widely distributed [e.g. yeast (Goodson et al., J CellBio11989, 109:84a)] and that the subfragment 1 region of myosin, which is sufficient to produce both movement and force (Toyoshima et al., Nature 1987, 328:536-539; Kishino and Yanagida, Nature 1988, 334:74-76), is anchored to membranes or actin illaments by a variety of carboxy-terminal domains that differ from the filament-forming tail-domain that anchors myosins II. Other variations on this theme are likely to occur. Indeed, the ninaCgene of Drosophila encodes ~ ' o proteins, necessary for proper formation of photoreceptor microviUi, which comprise an amino-terminal ldnase domain followed by a subfragment 1 domain and terminating in a domain unlike that of either a myosin II or a myosin I (Montell and Ruben, Cell 1988, 52:757-772). Furthermore, a gene encoding yet another type of Acantlmmoeba myosin, distinct from both myosins I and II, has been cloned and sequenced (Horowitz and Hammer, J Cell Bio11989, 109:282a). These proteins are likely harbingers of interesting future discoveries. What are the cellular functions o f myosins I? Brush border myosin I appears to play a role in maintaining microvillar structure, but it may well play other roles in the cell, such as vesicular transport. As for the protozoan myosins I, from their differential localizations in locomoting DicO~astelium , Fukui et al. [12] suggest that actomyosin I contributes to the forces that generate extension at the leading edges to form pseudopodia and lamellipodia, while actomyosin II contracts the rear of the cell and drives the cell mass forward. Consistent with this idea is the earlier demonstration that DicO~stelium amoebae lacking the myosin II heavy chain (De tozanne and Spudich, Science 1987, 236:1086-1091; Knecht and Loomis, Science 1987, 236:1081-1086) form pseudopodia at an essentially normal rate but that expansion of pseudopodia and cell movement are inhibited [13], and the suggestion that Acanttmmoeba motility is slowed but not stopped by injection o f antibodies against myosin II [14]. Fukui etaL [12] further speculate that myosin I may function similarly in cells where its presence has yet to be demonstrated (leukocytes, macrophages, fibroblasts, neuronal growth cones). However, this proposal does
Annotated references and recommended reading • ••
Of interest Of outstanding interest
1. •
l.ot.rVm, D D: The function of the major cTtoskeletal components of the brush border. Otrr Opin Cell Biol 1989, 1:51-57.
An up-to-datere-,'iewemphasizingvillinand brush border myosin I and describing the brush border cell as a model system for examining the assemblyand dynamicsof 2 os'toskeletalnetworks: the micro~llus and the terminal web. 2. JUNO G, SCII.,.~3TCJ, tLL~.LMERJA III: Myosin I hea,,T-chain • genes of Acanthamoeba castellanii: cloning of a second and es"idence for the existence of a third isoform. Gene 1989, 82:269-280. The deduced heaw-chain amino acid sequences of 2 Acanttmmoeba myosins I are highly consen-ed throughout, reflecting the very similar ph)~ical and biochemicalproperties of the purified isoforms~Regions of sequence consen~tion ~hich are distinct from m)x3sin11sequences could prove usefulas heterologous probes to search for myosinI genes in other organisms. 3. JUNOG, SAXECL Ul, YdSt~ttLAR, It~.~t~ERJA IU: DtcO'ostelium • discoideum contains a gene encoding a myosin I hea~T chain. Proc Nail Acad Sci USA 1989, 86:6186-6190. These results prove unequivocallythat Dico~steliumcontains a myosin I. In addition to encoding a poblaeptide that is remarkablysimilar to
M y o s i n I Korn a n d H a m m e r
the myosins 1 of Acantlmmoeba, this Dico~stelium gene is upregulated during starxution-induced aggregation. The feasibility of doing gene disruptions in DicO~gstelium should now allow a direct test for the ira t,it~9 function(s) of myosin I. 4. •
BRZESKAtt, LYNCH TJ, Komq ED: The effect of actin and phosphorylation on the tO'ptic cleavage pattern o f Acanthamoeba myosin IA. J Biol O)em 1989, 17:10243-10250. ltea'.y-chain phosphoQ:lation significantly enhances F-actin protection o f a t~'ptic cleavage site ~ 3 8 k D from the amino terminus in myosin IA. This may reflect a conformational change in the myosin globular head domain that switches the conformation of actomyosin I between productive and non-productive states. 5. e
BRZESKAH, LYNCtl TJ, KOm'q ED: The localization and sequence o f the phosphoD-lation sites of Acanthamoeba myosin I: an improved method for locating the phosphorylated amino acid. J Biol Chem 1989, 264:19340-19348. This information is important to efforts directed at understanding how phosphot3~ation regulates enz)Tnatic activity. The results also provide the correct assignment of the cloned genes to the purified protein isoforms. 6. •
CAm3oxlJM, COh7~EL~'; KA, AO&~tS RA, KAISERDA, POLLME) TD, MOOSEKERMS: Structural and immunological characterization o f the myosin.like 110-kD subunit of the intestinal micro~4_llar 1 lOK-calmodulin complex: evidence for discrete myosin head and calrnodulin.binding domains. J Cell Biol 1988, 107:1749-1757. These results proxide support for the idea that brush border myosin I is composed of a myosin globular head domain, v-herein resides its actinactivated MgX+.ATPase acuity, fused on an uncom'entional carboxyterminal domain, v.tmrein resides the calmodulin binding sites. 7. •
LYr:CHTJ, BW_.eSKAtt, Koe~'~ ED: Purification and characterization o f a third isoform o f myosin I from Acanthamoeba castellaniL J Biol Chem 1989, 264:19333-19339. All 3 isoforms are very similar in their physical and biochemical properties, suggesting that their in t,it~9function(s) may overlap considerably. 8. ••
MOOSEKERMS, COLESt&'qTR: The 110-kD protein.calmodulin complex o f the intestinal microvillus (brush border myosin I) is a mechanoenzyme. J Cell Biol 1989, 108:2395-2400. An important result sho'Mng that brash border myosin I moves beads along Nitella actin cables in a unidirectional and ATP-dependent manner. This finding demonstrates that, like all other myosins, brash border myosin I is a mechanoenzyme, and opens the w-ay to exploring how calcium, calmodulin, phosphoDSation, etc. regulate movement produced by this enzyme. 9.
MOOSEKERMS, CON7.EI2,1gNKA, COLEM.&NTR, HEUSERJE, SHEE'fZ MP: Characterization of intestinal microvillar membrane disks: detergent-resistant membrane sheets enriched in associated brush border myosin I (ll0K-calmodulin). J Cell Biol 1989, 109:1153--1161. The movement of these discs, vA-fichis supported by brush border myosin I, is considerably faster than the movement of latex beads coated with purified brush border myosin I and does not require high calcium levels. •
10. ee
~,LrYATA1t, BO~XRS B, KORN ED: Plasma membrane association of Acanthamoeba myosin I. J Cell Biol 1989, 109:1519-1528.
Along with [11], this study indicates that myosin I binds to membranes directly, i.e. not through membrane-assodated actin, but directly to membrane lipid and/or integral membrane proteins. This study suggests that the membrane attachment site is near the ATP-insensi~'e actin binding site present within the unconventional carboxy-terminal domain of m)~osin I. Studies like these are cmdal to efforts directed at understanding the specific organization of myosin l-membrane-actin complexes. ADA.~t5RJ, POLLARD"I'D: Binding o f myosin I to membrane ee lipids. Nature 1989, 340:566-568. This important study demonstrates that myosin I binds directly to both Acantlmmoeba membranes from ~taich peripheral proteins have been extracted and to pure lipid vesicles. The results are in keeping with pre~ions findings, including the immunofluorescent localization of myosin I to regions at and near the plasma membrane and the obser~ution that membranous vesicles from Acantbamoeba move on actin filaments by a motor tentatively identified as myosin I. 11.
12 • •
FUKUIY, LYNCHTJ, BRZESKAH, KORN ED: Myosin I is located at the leading edges of locomoting DicO,ostelium amoebae. Nature 1989, 341:328-331. A very important result which s h o w that myosin I and myosin II have distinct Iocali7~qtions in chemotacting and dividing DicO~stelium cells. Of particular importance are the obserrations that myosin I is concentrated at the leading edge of chemotacting cells and in phagocytic cups. The authors postulate that myosin I generates a projectile force at the leading edge of chemotacting cells, while myosin It, which is concentrated in the posterior cortex of the cell, propels the cell mass forward. It is also postulated that myosin l plays a direct role in the membrane events invoked in phagoc31osis. 13.
WESSELSD, SOU. DR, KNECHT D, LoOStlS WF, DE LOZAN,'NEA, SPUDICHJ: Cell motility and chemotaxis in Dictyostelium amoebae lacking myosin heavy chain. Dev Biol 1988, 128:164-177. A detailed ana~,~is o f the motility o f DicO~stelium myosin It cells using the dynamic morphology s)~tem, revealing the wealth of information that can be gained with this s3~tem. The results reveal abnormalities in locomotion, chemotaxis and the extent of pseudopod and lamellopod formation in cells lacking myosin IL These defects are in addition to the pr~iously reported impairment of tTtokinesis. • •
14. •
SL',aStDJH, POLLARDT: Microinjection into Acanthamoeba castellanti o f monoclonal antibodies to myosin-It slows but does not stop cell locomotion. Cell Motil Cytoskeleton 1989, 12:42-52. The results of this study suggest that while myosin II plays a role in retraction o f the trailing edge of locomoting ceils, the cell either roufinely uses or can draw upon another cytoplasmic motor to maintain locomotion. A strong candidate for this other motor is myosin I. 15. SXLJa.LJV: Microfilament-based motility in non-muscle cells. • Curr Opin Cell Biol 1989, 1:75-79. An excellent review on the molecular mechanisms of cell movement, with particular emphasis on the possible role of actin pobrnerization in generating a protrusive force at the leading edge of a cell. 16. S.~tmt SJ: Neuronal cytomechanics: the actin-based motility • of growth cones. Science 1988, 242:708--715. A recent review describing the current knowledge of the neuronal grox~zh cone cytoskeleton and the role it must play in extension, navigation, branching and target location during neuronal path finding. Emphasis is placed on actin pobrnerization at the leading edge as a possible dm'ing force in growth cone movement.
61