Adv. Biophys., Vol. 27, pp. 221-226 (1991)
M Y O S I N I: A N E W I N S I G H T I N T O T H E MECHANISM AND CELLULAR SIGNIFICANCE OF ACTIN-BASED MOTILITY
K. COLLINS, .1 J. SELLERS, .2 AND P. M A T S U D A I R A .1 Whitehead Institute for Biomedical Research and Massachusetts Institute of Technology, 9 Cambridge Center, Cambridge, M A 02142 .1 and National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, M D 20892, *2 U.S.A.
The l l0K protein-calmodulin complex of intestinal brush border, structurally homologous to myosin subfragment-1 (S1), we show to have mechanochemical homology as well. Investigation of the mechanism of myosin I motility with the sliding filament in vitro assay confirms and extends results obtained with myosin II. Novel brush border myosin I structure confers the enzyme's unique regulation by calcium and tropomyosin in vitro. In vivo, the protein may have a cellular role complementary to myosin II, controlled differentially by these regulatory factors as well as by phosphorylation. In this report we discuss only vertebrate myosin I. Other myosin I proteins, those of the single-celled eukaryotes Acanthamoeba and Dictyostelium, are discussed thoroughly in several excellent reviews (1, 2). I.
MYOSIN I STRUCTURE
Myosin I was first identified in vertebrates in the microvilli of intestinal 221
222
K, COLLINS ET AL°
epithelial cells, as a complex of ll0K-protein and calmodulin. This complex was implicated as the lateral linkage of actin to the plasma membrane by ATP-induced morphological disruption and protein extraction (3) and was subsequently purified and shown to have actinactivated MgATPase activity (4-7). As purification of the complex has improved, its activity has become increasingly myosin II-like: the complex catalyzes an actin-dependent, 40-fold actin-activated rate of M g A T P hydrolysis with Vmax=0.9 sec 1 and ~m=40,,~M (8). The active complex consists of two polypeptides of l l0kD and 18kD in a molar ratio of 1:3 (unpublished observation). The 18kD polypeptide has been identified as calmodulin on the basis of enzyme activation, molecular weight shift in calcium, and partial amino acid sequence. The bovine analog of the ll0kD polypeptide gene has been cloned and sequenced (9), revealing an extensive homology with myosin S1. The partial sequence of the chicken brush border myosin I gene has also been reported (10). Another member of the vertebrate myosin I family was discovered by Montell and Rubin (11) in their analysis ofgene products important in Drosophila vision. The ninaC gene, essential for photoreceptor cell membrane structure, encodes a myosin head domain with an alternately spliced tail. Analysis of the distribution and abundance of myosin I proteins in other vertebrate cells is now gaining much experimental attention. II.
MYOSIN I MECHANISM
OF MOTILITY
To study of the mechanism of myosin I motility in vitro, we have used the sliding filament assay first developed by Kron and Spudich (12) and modified by Toyoshima et al. (13). Using this assay, we have measured the rate of myosin I-mediated motility as a function of incubation time, actin filament length, dilution of myosin enzyme, ionic strength of the motility buffer, temperature, and concentration of ATP. The effect of many of these parameters on myosin II-mediated motility has been described (14). As for myosin II, the velocity of myosin I-mediated translocation is independent of the number of actin-myosin interactions, assayed either as a function of the amount of myosin bound to the coverslip or the length of the filament translocated. Also as for myosin II, myosin I-mediated motility in vitro requires a concentration of 100 e~M ATP for maximal activation or 50-60 I,.U ATP for half-maximal activation. Rate of actin translocation by myosin I increases reversibly with temperature as does rate of translocation by
ENZYMATIC CHARACTERISTICS OF MYOSIN I MOTILITY
223
myosin II, but the log of motility rate shows a linear rather than biphasic dependence upon inverse temperature in Kelvin. Ionic strength has different effects on myosin II proteins; as for a subset of myosin II, rate of myosin I motility is largely unchanged by addition of 0-80 mM KC1 to the motility buffer. A novel and useful property of myosin I motility in vitro is its stability: the rate and regulation of motility are unchanged for over 5 weeks after purification of the enzyme and for over 2 hr on the coverslip. An unexpected similarity of myosin I and II motility arose from the calcium regulation experiments detailed in the next section. Both myosin I and II complexes lacking regulatory light chains demonstrate a drastic loss of motility and lack of regulation of MgATPase activity by increasing actin concentration (myosin II results: 15). For both myosins depleted of light chain, MgATPase activity is uncoupled from the translocation cycle. It appears that light-chain dissociation, at least in vitro, can act as an "on-off" switch common to both myosin types, despite the divergence of light chain structure and heavy chain binding sites for light chain. III.
MYOSIN
I REGULATION
We studied myosin I regulation in vitro by calcium and tropomyosin, combining measurements of ATPase activity and motility with structural assays. Because the brush border is a major site of transmembrane calcium transport and because calmodulin confers calcium regulation on most enzymes to which it binds, we expected brush border myosin I to be regulated by calcium. We find that calcium dissociates a subset of calmodulin from the complex over the micromolar range of free calcium (0-100 ~M), observed by velocity sedimentation of the complex in glycerol gradients. Over the same range of concentration, calcium also inhibits actin-activated MgATPase activity. Because myosin-mediated actin translocation is independent of myosin head concentration and because calcium-inactivated myosin I does not introduce an internal load to translocation, we see an abrupt, complete inhibition of motility in high calcium rather than a gradual decrease of rate of motility in parallel with dissociation of calmodulin. Direct inhibition of myosin I activities in high calcium is in contrast to the indirect calcium activation of most myosin II proteins, suggesting differential regulation of the two myosin types. The correlation of a structural change, partial calmodulin dissociation, with inhibition of myosin I activities suggested that change in
224
K. COLLINS ET A L .
the quaternary structure of the complex could mediate calcium regulation. To distinguish the possibilities of inhibition by calcium-mediated calmodulin dissociation from inhibition by calcium binding directly, we tested the ability of added exogenous calmodulin to rescue loss of motility and loss of MgATPase activity in high calcium buffers. Concentrations of calmodulin above 5-10 f~M completely rescue the actinactivated MgATPase activity of the enzyme in 1 mM calcium, and a calmodulin concentration between 5 and 25/~M similarly restores the uninhibited rate of actin translocation. We also studied regulation of myosin I-mediated motility by tropomyosin, using the smooth muscle isoform purified from turkey gizzard. Tropomyosin added to the actin wash of the coverslip partially inhibits actin attachment and inhibits motility completely. Tropomyosin added to the motility buffer after initiation of translocation partially dissociates actin from the coverslip and inhibits motility completely. Excess tropomyosin was necessary to mediate inhibition due to the requirement for tropomyosin-saturation of the actin filaments in the in vitro assay. The concentration of tropomyosin used (85 nM, with less than 10 nn actin) was about half to a third of the concentration required for maximal tropomyosin-mediated activation of smooth muscle myosin II ATPase activity (16). Tropomyosin inhibition occurs by formation of a tropomyosin-actin complex, not binding to myosin I directly: centrifugation of a mixture of the three proteins results in the removal of all tropomyosin-actin from solution while myosin I remains soluble in the supernatant. Thus it appears that both tropomyosin and calcium regulate myosins I and II differentially. In vivo as well as in vitro we suspect that myosin I is regulated by calcium and tropomyosin, but regulation by phosphorylation may also play a role. Both the 110K-protein and calmodulin subunits are phosphorylated by separate, partially purified kinases endogenous to brush border in an in vitro assay (unpublished observation). A putative 140kD myosin I-binding protein of the microvillus membrane may also regulate conformation of the complex in vivo. IV.
MYOSIN
I MOTILITY
IN THE
CELL
The role of myosin I in the cell is not known, but clues to the function of this enzyme are provided by its cellular localization. Several labs have determined the distribution of 110K-protein in the intestinal epithelium, revealing a concentration at the apical, brush border membrane of absorptive epithelial cells. We find the l l0K-protein in
E N Z Y M A T I C C H A R A C T E R I S T I C S OF MYOSIN I M O T I L I T Y
225
association with the basolateral membrane of the enterocyte as well, although severalfold less concentrated than at the apical membrane. A myosin molecule predominantly associated with the plasma membrane could function in intracellular vesicle transport, membrane extrusion, cytoplasmic streaming, or other cytoskeleton-regulated dynamic membrane events. Tropomyosin-containing actin structures such as the circumferential belt, microvillar rootlets, and stress fibers lack associated myosin I, perhaps because binding to tropomyosin-complexed actin is diminished in vivo as well as in vitro. The structure of the enzyme itself may also encourage membrane localization: its monomeric nature and single actin,binding site suggest that the myosin I protein, unlike myosin II, would demonstrate no preference for cytosolic actin bundles or networks. The distinct localization and regulation of myosins I and II suggest that a division of actin-based motility tasks between the two myosin types may occur. The involvement of myosin II has been demonstrated in cell shape changes, cytokinesis, and muscle contraction. Myosin II is not required, however, for motility, polarity, and formation of membrane projections in Dictyostelium amoeba (17, 18). In Acanthamoeba, myosin I is associated with cellular vesicles (19), rebinds NaOH-stripped vesicles or pure phospholipids (20), and catalyzes the motility of organelles in vitro (19). Acanthamoeba myosin I localizes to the leading edge of motile cells and to the membrane blebs that form after cytokinesis (21). Thus, the activities and localization of Acanthamoeba myosin I suggests a role for lower eukaryotic myosin I in cell polarity, motility, and formation of membrane extrusions. These events are not inhibited by lack of myosin II, supporting a myosin-type-dependent division of labor. Whether vertebrate myosin I fulfills a cellular role similar to that of Acanthamoeba myosin I remains to be tested. SUMMARY
From our work on brush border myosin I structure, activity, regulation, and function, we can begin to understand the significance of the diversification of myosin proteins. While myosin I and II proteins retain conserved elements of structure that may dictate their similar mechanisms of motility and actin-activated MgATPase activity, their unique structures may provide the basis for the distinct localization and regulation of the two myosin types. How does the tropomyosininhibited actin-binding site of myosin I differ from that of the tropomyosin-activated myosin II actin-binding site? What elements of the
226
K. COLLINS ET AL.
sites of interaction of the l l0K-protein and calmodulin contribute to the conserved, light-chain dependent coupling of MgATPase activity to translocation and which confer the novel calcium regulation of dissociation in vitro ? It seems that the evolutionary demand for diversification of cellular motility functions has been met, at least in the actinbased system, by the evolution ofisoforms tailored in structure, activity, regulation, and localization to serve complementary needs.
Acknowledgments This research was supported by funds from the American Heart Association and the Lucille P. Markey Foundation. REFERENCES 1 E. D. Korn, M.A.L. Atkinson, H. Brzeska, J. A. Hammer III, G. Jung, and T . J . Lynch, j . Cell. Biochem., 36, 37 (1988). 2 E. D. Korn a n d J . A. Hammer, III, Annu. Rev. Biophys. Biophys. Chem., 17, 23 (1988). 3 P. T. Matsudaira and D. R. Burgess, J. Cell Biol., 83, 667 (1979). 4 J. H. Collins and C. W. Borysenko, J. Biol. Chem, 259, 14128 (1984). 5 J. Krizek, L. M. Coluecio, and A. Bretscher, FEBS Lett., 225, 269 (1987). 6 H. Swanljung-Collins, J. Montibeller, and J. H. Collins, )Iethods Enzl,mol., 139, 137 (1987). 7 K.A. Conzelman and M. S. Mooseker, J . CellBiol., 105, 313 (1987). 8 K. Collins, J. R. Sellers, and P. Matsudaira, J. Cell Biol., in press (1991). 9 M. Hoshimaru and S. Nakanishi, J. Biol. Chem., 262, 14625 (1987). 10 A. Garcia, E. Coudrier, J. Carboni, J. Anderson, J. Vandekerhove, M. Mooseker, D. Louvard, and M. Arpin, J. Cell Biol., 109, 2895 (1989). 11 C. Montell and G. M. Rubin, Cell, 52, 757 (1988). 12 S.J. Kron a n d J . A. Spudich, Proc. Natl. Aead. Sci. U.S.A., 83, 6272 (1986). 13 Y. Y. Toyoshima, S.J. Kron, E. M. McNally, K. R. Niebling, C. Toyoshima, a n d J . A. Spudich, Nature, 328, 536 (1987). 14 M. P. Sheetz, R. Chasan, a n d J . A. Spudich, J. Cell Biol., 99, 1867 (1984). 15 R. D. Vale, A. G. Szent-Gy6rgyi, and M . P . Sheetz, Proe. Natl. Acad. Sci. U.S.A., 81, 6775 (1984). 16 S. Umemoto, A. R. Bengur, a n d J . R. Sellers, j . Biol. Chem., 264, 143l (1989). 17 A. De Lozanne and J. A. Spudich, Science, 236, 1086 (1987). 18 D. A. Knechat and W. F. Loomis, Dev. Biol., 128, 178 (1987). 19 R..]. Adams and T. D. Pollard, Nature, 322, 754 (1986). 20 R . J . Adams and T. D. Pollard, Nature, 340, 565 (1989). 21 Y. Fukui, T . J . Lynch, H. Brzeska, and E. D. Korn, Nature, 341, 328 (1989). Received fur publication March 28, 1990.
M Y O S I N I: A N E W I N S I G H T I N T O T H E MECHANISM AND CELLULAR SIGNIFICANCE OF ACTIN-BASED MOTILITY
K. COLLINS, .1 J. SELLERS, .2 AND P. M A T S U D A I R A .1 Whitehead Institute for Biomedical Research and Massachusetts Institute of Technology, 9 Cambridge Center, Cambridge, M A 02142 .1 and National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, M D 20892, *2 U.S.A.
The l l0K protein-calmodulin complex of intestinal brush border, structurally homologous to myosin subfragment-1 (S1), we show to have mechanochemical homology as well. Investigation of the mechanism of myosin I motility with the sliding filament in vitro assay confirms and extends results obtained with myosin II. Novel brush border myosin I structure confers the enzyme's unique regulation by calcium and tropomyosin in vitro. In vivo, the protein may have a cellular role complementary to myosin II, controlled differentially by these regulatory factors as well as by phosphorylation. In this report we discuss only vertebrate myosin I. Other myosin I proteins, those of the single-celled eukaryotes Acanthamoeba and Dictyostelium, are discussed thoroughly in several excellent reviews (1, 2). I.
MYOSIN I STRUCTURE
Myosin I was first identified in vertebrates in the microvilli of intestinal 221
222
K, COLLINS ET AL°
epithelial cells, as a complex of ll0K-protein and calmodulin. This complex was implicated as the lateral linkage of actin to the plasma membrane by ATP-induced morphological disruption and protein extraction (3) and was subsequently purified and shown to have actinactivated MgATPase activity (4-7). As purification of the complex has improved, its activity has become increasingly myosin II-like: the complex catalyzes an actin-dependent, 40-fold actin-activated rate of M g A T P hydrolysis with Vmax=0.9 sec 1 and ~m=40,,~M (8). The active complex consists of two polypeptides of l l0kD and 18kD in a molar ratio of 1:3 (unpublished observation). The 18kD polypeptide has been identified as calmodulin on the basis of enzyme activation, molecular weight shift in calcium, and partial amino acid sequence. The bovine analog of the ll0kD polypeptide gene has been cloned and sequenced (9), revealing an extensive homology with myosin S1. The partial sequence of the chicken brush border myosin I gene has also been reported (10). Another member of the vertebrate myosin I family was discovered by Montell and Rubin (11) in their analysis ofgene products important in Drosophila vision. The ninaC gene, essential for photoreceptor cell membrane structure, encodes a myosin head domain with an alternately spliced tail. Analysis of the distribution and abundance of myosin I proteins in other vertebrate cells is now gaining much experimental attention. II.
MYOSIN I MECHANISM
OF MOTILITY
To study of the mechanism of myosin I motility in vitro, we have used the sliding filament assay first developed by Kron and Spudich (12) and modified by Toyoshima et al. (13). Using this assay, we have measured the rate of myosin I-mediated motility as a function of incubation time, actin filament length, dilution of myosin enzyme, ionic strength of the motility buffer, temperature, and concentration of ATP. The effect of many of these parameters on myosin II-mediated motility has been described (14). As for myosin II, the velocity of myosin I-mediated translocation is independent of the number of actin-myosin interactions, assayed either as a function of the amount of myosin bound to the coverslip or the length of the filament translocated. Also as for myosin II, myosin I-mediated motility in vitro requires a concentration of 100 e~M ATP for maximal activation or 50-60 I,.U ATP for half-maximal activation. Rate of actin translocation by myosin I increases reversibly with temperature as does rate of translocation by
ENZYMATIC CHARACTERISTICS OF MYOSIN I MOTILITY
223
myosin II, but the log of motility rate shows a linear rather than biphasic dependence upon inverse temperature in Kelvin. Ionic strength has different effects on myosin II proteins; as for a subset of myosin II, rate of myosin I motility is largely unchanged by addition of 0-80 mM KC1 to the motility buffer. A novel and useful property of myosin I motility in vitro is its stability: the rate and regulation of motility are unchanged for over 5 weeks after purification of the enzyme and for over 2 hr on the coverslip. An unexpected similarity of myosin I and II motility arose from the calcium regulation experiments detailed in the next section. Both myosin I and II complexes lacking regulatory light chains demonstrate a drastic loss of motility and lack of regulation of MgATPase activity by increasing actin concentration (myosin II results: 15). For both myosins depleted of light chain, MgATPase activity is uncoupled from the translocation cycle. It appears that light-chain dissociation, at least in vitro, can act as an "on-off" switch common to both myosin types, despite the divergence of light chain structure and heavy chain binding sites for light chain. III.
MYOSIN
I REGULATION
We studied myosin I regulation in vitro by calcium and tropomyosin, combining measurements of ATPase activity and motility with structural assays. Because the brush border is a major site of transmembrane calcium transport and because calmodulin confers calcium regulation on most enzymes to which it binds, we expected brush border myosin I to be regulated by calcium. We find that calcium dissociates a subset of calmodulin from the complex over the micromolar range of free calcium (0-100 ~M), observed by velocity sedimentation of the complex in glycerol gradients. Over the same range of concentration, calcium also inhibits actin-activated MgATPase activity. Because myosin-mediated actin translocation is independent of myosin head concentration and because calcium-inactivated myosin I does not introduce an internal load to translocation, we see an abrupt, complete inhibition of motility in high calcium rather than a gradual decrease of rate of motility in parallel with dissociation of calmodulin. Direct inhibition of myosin I activities in high calcium is in contrast to the indirect calcium activation of most myosin II proteins, suggesting differential regulation of the two myosin types. The correlation of a structural change, partial calmodulin dissociation, with inhibition of myosin I activities suggested that change in
224
K. COLLINS ET A L .
the quaternary structure of the complex could mediate calcium regulation. To distinguish the possibilities of inhibition by calcium-mediated calmodulin dissociation from inhibition by calcium binding directly, we tested the ability of added exogenous calmodulin to rescue loss of motility and loss of MgATPase activity in high calcium buffers. Concentrations of calmodulin above 5-10 f~M completely rescue the actinactivated MgATPase activity of the enzyme in 1 mM calcium, and a calmodulin concentration between 5 and 25/~M similarly restores the uninhibited rate of actin translocation. We also studied regulation of myosin I-mediated motility by tropomyosin, using the smooth muscle isoform purified from turkey gizzard. Tropomyosin added to the actin wash of the coverslip partially inhibits actin attachment and inhibits motility completely. Tropomyosin added to the motility buffer after initiation of translocation partially dissociates actin from the coverslip and inhibits motility completely. Excess tropomyosin was necessary to mediate inhibition due to the requirement for tropomyosin-saturation of the actin filaments in the in vitro assay. The concentration of tropomyosin used (85 nM, with less than 10 nn actin) was about half to a third of the concentration required for maximal tropomyosin-mediated activation of smooth muscle myosin II ATPase activity (16). Tropomyosin inhibition occurs by formation of a tropomyosin-actin complex, not binding to myosin I directly: centrifugation of a mixture of the three proteins results in the removal of all tropomyosin-actin from solution while myosin I remains soluble in the supernatant. Thus it appears that both tropomyosin and calcium regulate myosins I and II differentially. In vivo as well as in vitro we suspect that myosin I is regulated by calcium and tropomyosin, but regulation by phosphorylation may also play a role. Both the 110K-protein and calmodulin subunits are phosphorylated by separate, partially purified kinases endogenous to brush border in an in vitro assay (unpublished observation). A putative 140kD myosin I-binding protein of the microvillus membrane may also regulate conformation of the complex in vivo. IV.
MYOSIN
I MOTILITY
IN THE
CELL
The role of myosin I in the cell is not known, but clues to the function of this enzyme are provided by its cellular localization. Several labs have determined the distribution of 110K-protein in the intestinal epithelium, revealing a concentration at the apical, brush border membrane of absorptive epithelial cells. We find the l l0K-protein in
E N Z Y M A T I C C H A R A C T E R I S T I C S OF MYOSIN I M O T I L I T Y
225
association with the basolateral membrane of the enterocyte as well, although severalfold less concentrated than at the apical membrane. A myosin molecule predominantly associated with the plasma membrane could function in intracellular vesicle transport, membrane extrusion, cytoplasmic streaming, or other cytoskeleton-regulated dynamic membrane events. Tropomyosin-containing actin structures such as the circumferential belt, microvillar rootlets, and stress fibers lack associated myosin I, perhaps because binding to tropomyosin-complexed actin is diminished in vivo as well as in vitro. The structure of the enzyme itself may also encourage membrane localization: its monomeric nature and single actin,binding site suggest that the myosin I protein, unlike myosin II, would demonstrate no preference for cytosolic actin bundles or networks. The distinct localization and regulation of myosins I and II suggest that a division of actin-based motility tasks between the two myosin types may occur. The involvement of myosin II has been demonstrated in cell shape changes, cytokinesis, and muscle contraction. Myosin II is not required, however, for motility, polarity, and formation of membrane projections in Dictyostelium amoeba (17, 18). In Acanthamoeba, myosin I is associated with cellular vesicles (19), rebinds NaOH-stripped vesicles or pure phospholipids (20), and catalyzes the motility of organelles in vitro (19). Acanthamoeba myosin I localizes to the leading edge of motile cells and to the membrane blebs that form after cytokinesis (21). Thus, the activities and localization of Acanthamoeba myosin I suggests a role for lower eukaryotic myosin I in cell polarity, motility, and formation of membrane extrusions. These events are not inhibited by lack of myosin II, supporting a myosin-type-dependent division of labor. Whether vertebrate myosin I fulfills a cellular role similar to that of Acanthamoeba myosin I remains to be tested. SUMMARY
From our work on brush border myosin I structure, activity, regulation, and function, we can begin to understand the significance of the diversification of myosin proteins. While myosin I and II proteins retain conserved elements of structure that may dictate their similar mechanisms of motility and actin-activated MgATPase activity, their unique structures may provide the basis for the distinct localization and regulation of the two myosin types. How does the tropomyosininhibited actin-binding site of myosin I differ from that of the tropomyosin-activated myosin II actin-binding site? What elements of the
226
K. COLLINS ET AL.
sites of interaction of the l l0K-protein and calmodulin contribute to the conserved, light-chain dependent coupling of MgATPase activity to translocation and which confer the novel calcium regulation of dissociation in vitro ? It seems that the evolutionary demand for diversification of cellular motility functions has been met, at least in the actinbased system, by the evolution ofisoforms tailored in structure, activity, regulation, and localization to serve complementary needs.
Acknowledgments This research was supported by funds from the American Heart Association and the Lucille P. Markey Foundation. REFERENCES 1 E. D. Korn, M.A.L. Atkinson, H. Brzeska, J. A. Hammer III, G. Jung, and T . J . Lynch, j . Cell. Biochem., 36, 37 (1988). 2 E. D. Korn a n d J . A. Hammer, III, Annu. Rev. Biophys. Biophys. Chem., 17, 23 (1988). 3 P. T. Matsudaira and D. R. Burgess, J. Cell Biol., 83, 667 (1979). 4 J. H. Collins and C. W. Borysenko, J. Biol. Chem, 259, 14128 (1984). 5 J. Krizek, L. M. Coluecio, and A. Bretscher, FEBS Lett., 225, 269 (1987). 6 H. Swanljung-Collins, J. Montibeller, and J. H. Collins, )Iethods Enzl,mol., 139, 137 (1987). 7 K.A. Conzelman and M. S. Mooseker, J . CellBiol., 105, 313 (1987). 8 K. Collins, J. R. Sellers, and P. Matsudaira, J. Cell Biol., in press (1991). 9 M. Hoshimaru and S. Nakanishi, J. Biol. Chem., 262, 14625 (1987). 10 A. Garcia, E. Coudrier, J. Carboni, J. Anderson, J. Vandekerhove, M. Mooseker, D. Louvard, and M. Arpin, J. Cell Biol., 109, 2895 (1989). 11 C. Montell and G. M. Rubin, Cell, 52, 757 (1988). 12 S.J. Kron a n d J . A. Spudich, Proc. Natl. Aead. Sci. U.S.A., 83, 6272 (1986). 13 Y. Y. Toyoshima, S.J. Kron, E. M. McNally, K. R. Niebling, C. Toyoshima, a n d J . A. Spudich, Nature, 328, 536 (1987). 14 M. P. Sheetz, R. Chasan, a n d J . A. Spudich, J. Cell Biol., 99, 1867 (1984). 15 R. D. Vale, A. G. Szent-Gy6rgyi, and M . P . Sheetz, Proe. Natl. Acad. Sci. U.S.A., 81, 6775 (1984). 16 S. Umemoto, A. R. Bengur, a n d J . R. Sellers, j . Biol. Chem., 264, 143l (1989). 17 A. De Lozanne and J. A. Spudich, Science, 236, 1086 (1987). 18 D. A. Knechat and W. F. Loomis, Dev. Biol., 128, 178 (1987). 19 R..]. Adams and T. D. Pollard, Nature, 322, 754 (1986). 20 R . J . Adams and T. D. Pollard, Nature, 340, 565 (1989). 21 Y. Fukui, T . J . Lynch, H. Brzeska, and E. D. Korn, Nature, 341, 328 (1989). Received fur publication March 28, 1990.
