Journal of Cell Science 113, 1311-1318 (2000) 1311 Printed in Great Britain © The Company of Biologists Limited 2000 JCS0832

COMMENTARY Molecular motors − a paradigm for mutant analysis

Sharyn A. Endow Department of Microbiology, Duke University Medical Center, Durham, NC 27710 USA (e-mail: [email protected])

Published on WWW 21 March 2000

SUMMARY

Molecular motors perform essential functions in the cell or by rational design. Mutants that are expected to and have the potential to provide insights into the basis provide important information about the motor of many important processes. A unique property of mechanism include ATPase mutants, which interfere with molecular motors is their ability to convert energy from the nucleotide hydrolysis cycle, and uncoupling mutants, ATP hydrolysis into work, enabling the motors to bind to which unlink basic motor activities and reveal their and move along cytoskeletal filaments. The mechanism interdependence. Natural variants can also be exploited of energy conversion by molecular motors is not yet to provide unexpected information about motor function. understood and may lead to the discovery of new This general approach to uncovering protein function by biophysical principles. Mutant analysis could provide analysis of informative mutants is applicable not only to valuable information, but it is not obvious how to obtain molecular motors, but to other proteins of interest. mutants that are informative for study. The analysis presented here points out several strategies for obtaining mutants by selection from molecular or genetic screens, Key words: Molecular motor, Mutant, Structure/function

INTRODUCTION The cellular roles of the motors and interactions between the motors are only beginning to be understood. The fact that Molecular motors carry out a myriad of functions in the cell, the cellular functions of many of the motors are as yet ranging from vesicle (Hirokawa, 1998) and RNA transport unknown means that the effects of mutants that cause loss of (Nasmyth and Jansen, 1997) to spindle assembly and function have not always been straightforward to predict, and chromosome motility in dividing cells (Endow, 1999c). The some of these effects have been unexpected. For example, finding of a large number of motors that fall into one of the mutants of the unconventional myosins, myosin VI and VII, three families of motors known to exist in eukaryotes, the actin- cause deafness in mice and humans, and are the basis of based myosins (Sellers and Goodson, 1995) or microtubule- previously described syndromes, Snell’s Walzer and shaker, based kinesins (Bloom and Endow, 1995) and dyneins that are associated with behavioral abnormalities in mice (Hirokawa, 1998), has been revolutionary in permitting (Avraham et al., 1995; Gibson et al., 1995; Weil et al., 1995). analysis of diverse biological processes as the roles of the These myosins are present in hair cells of the inner ear, where proteins in the cell are discovered. The importance of their roles in hearing are currently being investigated. molecular motors in the cell is exemplified by the process of Another example of an unexpected effect of a mutant comes cell division, in which many of the kinesin microtubule motors, from work on the kinesin-related KIF3B microtubule motor. together with cytoplasmic dynein, produce force for spindle In the KIF3B-knockout mouse, loss of KIF3B is associated assembly and chromosome attachment to the spindle (Endow, with loss of bilateral asymmetry (Nonaka et al., 1998), which 1999c). Cytoplasmic myosin then functions during cytokinesis is needed to form organs such as the heart in mammals. to complete the process of cell division by accumulating in the Hirokawa and co-workers propose that the basis of the loss cleavage furrow and providing motile force for the contractile of asymmetry is defective transport of substances required for ring (Goldberg et al., 1998). Vesicle trafficking similarly formation of cilia in the early embryo (Nonaka et al., 1998). involves motors from all three families (Hirokawa, 1998; They hypothesize that the cilia cause a leftward flow of Brown, 1999), while specific cellular roles are played by extraembryonic fluid in the area of the node, which is a myosin in muscle contraction (Cooke, 1986), dyneins in transient structure present during gastrulation, and therefore flagellar motility (Gibbons, 1996) and kinesins in regulation of play a key role in the formation of left-right morphogen microtubule dynamics (Inoué and Salmon, 1995; Waters and gradients. These unforeseen mutant effects reflect the Salmon, 1996). inability to predict the cellular roles of the newly discovered 1312 S. A. Endow myosin, kinesin and dynein motors, and underscore the different from other nucleotide-hydrolyzing proteins in necessity for cellular studies. But to fully understand the roles possessing the ability to convert chemical energy into work has of molecular motors in cells, it is necessary to determine how led to the question of how to obtain information about the motor the motors work. mechanism. Strategies that have been used successfully for other proteins, such as the G proteins, are to characterize the proteins functionally and structurally using available biochemical and THE MOTOR MECHANISM structural methods (Bourne et al., 1991), and then to mutate specific residues of the proteins and determine the effects of the Molecular motors are unique in their ability to convert the mutations on protein structure and function (Raw et al., 1997). chemical energy from ATP hydrolysis into work, which permits Findings from such structure/function studies have resulted in them to bind to and move along actin filaments or microtubules. working models of G protein function (Sprang, 1997). Further The process of energy conversion by motor proteins is not tests by state-of-the-art methods, such as time-resolved or understood, and considerable effort is now being devoted to millisecond-resolution X-ray crystallography (Scott et al., 1996; unraveling the possible mechanisms. The discovery of large Stoddard et al., 1998), could confirm the conformational changes families of related motors whose members differ from one in protein structure proposed to occur during the hydrolysis cycle, another in directionality of movement on their filament, ATPase which are based on the ‘still’ images from static crystal structures. activity and motor velocity offers an unprecedented opportunity This general approach is one that is being used to study the to perform studies to uncover the mechanism by which these mechanism of myosin function and has led to several proteins function. This information will be valuable, first, noteworthy achievements in recent years. The solution of the because it represents basic information about energy conversion first myosin motor domain structure was a breakthrough in living organisms and may lead to the discovery of new (Rayment et al., 1993) that has been rapidly followed by new principles and, second, because it will help us to understand myosin structures (Fisher et al., 1995; Dominguez et al., 1998; how the motors work in the cell. Practical uses of the anticipated Houdusse et al., 1999), as noted above. Mutant analysis coupled information will be to design mutants that disrupt specific with expression of functional myosin proteins in baculovirus is cellular functions, or to create motors for transport of drugs or underway (Ruppel and Spudich, 1995) and should provide other substances to specific cells or regions of cells. information about the myosin mechanism of function. But A milestone towards the goal of understanding the changes despite the recent surge in progress on myosin, the kinesins that take place in the motors during the conversion of chemical have continued to attract attention. The kinesins offer several to mechanical energy has been the solution of crystal structures advantages over myosin, including the smaller size of the of myosin in three different nucleotide states: a ‘rigor’ state motors compared with myosin, which is a distinct advantage for without nucleotide (Rayment et al., 1993), a putative transition structural analysis, and the ease of expressing the kinesin − state with bound ADP⋅AlF4 (Fisher et al., 1995; Dominguez et proteins in bacteria compared with the technicalities of al., 1998), and a third distinct conformational state that may baculovirus expression. On the other hand, a major barrier to correspond to a prehydrolysis state (Houdusse et al., 1999). A rapid progress on the kinesins has been the inability to comparable advance for the kinesins has been the solution of crystallize the motors in different conformations that crystal structures of monomeric and dimeric forms of the motors correspond to different nucleotide states of the motor. Although in the Mg⋅ADP state (Sack et al., 1999). The crystal structures seven crystal structures of kinesin proteins have been reported of the kinesins provide information about the dimeric forms of to date (Sack et al., 1999; Kozielski et al., 1999), all of the the motors, currently unavailable for the myosins, but are structures are in the same nucleotide state, with bound Mg⋅ADP. uninformative about the changes that take place in the motors Efforts to identify different structures of the kinesins bound to during nucleotide hydrolysis because they all correspond to the other nucleotides, including attempts to use nucleotide same state, presumably a post-hydrolysis state. Nonetheless, the analogues to grow crystals of motors in different states, have structures have been valuable in revealing similarities between not been successful − the crystals have been in the Mg⋅ADP the kinesin motors and myosin (Kull et al., 1996), and between conformation despite the bound nucleotide (Muller et al., 1999). the kinesins and G proteins (Sablin et al., 1996; Fig. 1). These This means that the crystal structures cannot be used to structural similarities provide a basis for comparing the determine the conformational changes that occur during mechanisms of protein function. The kinesin and myosin motor nucleotide hydrolysis, or to identify regions of the kinesin motor proteins share with the G proteins the ability to hydrolyze domain that move or change conformation during hydrolysis. nucleotide at a greatly enhanced rate in the presence of effector This lack of success has caused workers to consider different proteins. Specific effector proteins exist for different G proteins, approaches to uncover the conformational states of the kinesin but the effector proteins for the myosins and kinesins are actin motors. One idea is to use mutants to ‘trap’ the motor in filaments and microtubules, the cytoskeletal filaments along different conformations that reflect different nucleotide states which the motors move. This means that the very act of binding of the motor. Mutants could potentially block the motor at to their filament greatly increases the ability of the motors to different steps of the nucleotide-hydrolysis cycle − even the convert energy from nucleotide hydrolysis into the mechanical steps that normally proceed very rapidly, at millisecond rates energy that enables them to move along the filament. − and thus reveal details of the hydrolysis mechanism that would be difficult or impossible to observe using currently available biochemical methods. The task, then, is to find ANALYSIS OF MOTOR FUNCTION mutants that will be informative. This task is one that confronts both kinesin and myosin motors workers and one for which The realization that molecular motors might be fundamentally general strategies have been elusive. Molecular motors − a paradigm for mutant analysis 1313

Fig. 1. Structural similarities between a kinesin, a myosin and a G protein. The nucleotide-binding or P loop (light blue), and switch I (SwI, red) and switch II (SwII, dark green) regions of the G proteins are conserved in the kinesin and myosin motors. The crystal structures are of p21ras, the cellular Harvey-Ras p21 protein (PDB 821P; Franken et al., 1993), the kinesin Kar3p motor domain (PDB 3KAR; Gulick et al., 1998), and chicken skeletal muscle myosin S1 subfragment (MyoII, PDB 2MYS; Rayment et al., 1993). The residues mutated in the kinesin (N650K) and myosin (S474V, S465V in Dictyostelium) uncoupling mutants are depicted as ball-and-stick models in purple. The mutated residues both lie in the switch II loop/helix that follows the switch II region in the Kar3p or myosin motor domain. The microtubule- or actin-binding region of the motor is shown in dark pink. The residue numbers and sequences of the P loop, and switch I and switch II regions are shown at the bottom. PDB, Protein Data Bank. 1314 S. A. Endow

MOTOR MUTANTS correlated with the ability of the mutant motor to bind to its filament. Functionally interesting conformations of the motor The problem that has perplexed many workers is how to − for example, the force-producing steps − are likely to be only identify residues that are needed for motor function and transiently stable. By stabilizing or increasing the lifetime of mutational changes that will define essential functions without these steps, the ATPase mutants have the potential to reveal having to carry out exhaustive genetic or molecular many or all of the steps in the nucleotide-hydrolysis cycle. mutagenesis screens. Even given the available crystal Crystallization of the mutant motors could produce different structures and the identification of residues that are involved in crystal forms that reveal conformational changes that occur in nucleotide hydrolysis or binding to microtubules or actin the motors during hydrolysis. Analysis of the ATPase mutants filaments, it is not clear which residues to target or which is also expected to reveal the coupling between the nucleotide- residue to change the targeted residues to. There are also other hydrolysis cycle and interactions of the motor with its filament. problems inherent in targeted mutagenesis that complicate the At least some of the ATPase mutants should fall into the class analysis of mutants. For example, point mutants could have of ‘uncoupling’ mutants. effects that are specific to a given protein rather than affecting an activity that is required for basic motor function, or they Uncoupling mutants could inactivate the motor by destroying the motor ATPase or Uncoupling mutants were proposed to exist because of the the ability of the motor to bind to its filament, making it unusual coupling of nucleotide hydrolysis to force generation impossible to detect activity using standard ATPase or (discussed above) that is a basic property of the molecular filament-binding assays. This means that it is important to motors. The realization that uncoupling mutants might exist consider both the type of mutant and the available assays that and would be informative to study led to the search for and can be used to determine mutant effects when designing or recovery of such mutants in Dictyostelium myosin (Ruppel and selecting mutants for analysis. Fortunately for projected Spudich, 1996) and to the recent identification of a kinesin structural studies, the effects of single residue changes on uncoupling mutant (Song and Endow, 1998). The uncoupling overall structure are usually not a major concern, since mutants separate essential motor functions so that one function considerable evidence indicates that proteins exist in stably is increased or decreased, whereas another remains unchanged folded conformations that are not easy to change by single or is changed in the opposite direction. Uncoupling mutants missense mutations. Proteins maintain their folded can be either partial loss-of-function mutants or dominant conformations by extensive, relatively weak electrostatic mutants. One example is the Dictyostelium S465V (S474 of forces, including hydrogen bonds and van der Waals forces, chicken) myosin mutant (Table 1), which translocates actin together with hydrophobic interactions and stabilizing effects filaments much more slowly than wild type, at one-tenth the of disulfide bonds (Branden and Tooze, 1991). Mutation of a velocity, but has an elevated basal ATPase activity that is nearly single residue can disrupt some of these interactions, but equal to the actin-activated ATPase activity of wild type generally affects local structure rather than overall structure. (Ruppel and Spudich, 1996). The S465V mutation thus Thus, mutants resulting from single residue changes are not uncouples ATPase activity and motility. We recently reported expected to produce global alterations in protein structure and, a kinesin uncoupling mutant (Table 1) in which microtubule- in contrast, have the potential to provide valuable information stimulated ATP hydrolysis is completely blocked even though regarding possible conformations in which the protein can the basal ATPase activity of the mutant is unchanged from that exist. of wild type (Song and Endow, 1998). The basis of this defect is the uncoupling of nucleotide and microtubule binding: the ATPase mutants mutated motor binds tightly both to ADP and microtubules, Several types of molecular motor mutants are likely to be unlike wild-type kinesin motors, which bind weakly to informative about the motor mechanism of function. They microtubules in the presence of ADP. The inability of the motor include mutants that interfere with the nucleotide-hydrolysis to release nucleotide at an enhanced rate when bound to cycle, which are expected to affect both the ability of the motor microtubules is correlated with an inability to move on to bind and hydrolyze nucleotide and motor interactions with microtubules. The mutated residue is present in the microtubules or actin filaments. These mutants will be microtubule-binding region of the motor (Fig. 1), some informative not only about the mechanism of motor function but also about the cellular phenotypes that are caused by motor mutations. The mutants could show a complete inability to Table 1. Uncoupling mutants hydrolyze nucleotide, or they could affect only the filament- stimulated ATPase of the motor. Both types of mutant are ATPase Filament-binding Motility expected to alter the binding of the motor to its filament − the Myosin S465V mutant ↑ ↓ Kinesin uncoupling ↓ ↑ − effect would depend on the step of hydrolysis that is defective. mutant For example, mutants that cannot bind nucleotide and therefore Loss-of-function mutant ↓↓↓ are unable to hydrolyze ATP would be ‘rigor’ mutants that bind Null mutant − − − very tightly to their filament, whereas defective filament- Wild type +++ activated ATPase activity could result in either weaker or The myosin S465V mutant of Dictyostelium (Ruppel and Spudich, 1996) stronger binding to microtubules or actin filaments, depending and the kinesin uncoupling mutant (Song and Endow, 1998) cause an increase on the nucleotide state in which the mutants are blocked. The in one motor activity but a decrease in another, uncoupling or separating steps that are blocked can be determined by careful motor functions. The effects of loss-of-function and null mutants are shown biochemical analysis of nucleotide-hydrolyzing ability and for comparison. Wild-type motors show all three activities. Molecular motors − a paradigm for mutant analysis 1315 distance from the nucleotide-binding cleft of the motor, thus labor intensive and time consuming. A careful analysis of even the effect of the mutation on nucleotide binding was a single mutant could require several months, and ideally a unexpected. It is consistent, however, with the proposal that the number of mutants should be analyzed to test all of the structural element affected by the mutation is involved in potentially important functional regions of the motor, or communication between the nucleotide- and microtubule- residues of specific regions. Molecular mutagenesis does offer binding regions of the motor (Sablin et al., 1996). the advantage that mutational changes can be introduced into Surprisingly, despite the difference in their biochemical targeted regions of a protein, e.g. the motor domain, but this effects, the kinesin and myosin uncoupling mutations map to advantage does not outweigh the demands on labor and time. analogous structural elements in the two motors: a loop/helix This approach is therefore one that is not a method of choice that is thought to correspond to the ‘switch II’ loop/helix of the for most laboratories, given limitations on manpower and G proteins (Fig. 1), which changes in conformation on financial resources. nucleotide exchange. The myosin mutation lies in the cleft that divides the 50-kDa domain of the myosin motor into the upper Genetic screens and lower regions, which, like the structural element affected A second strategy, although one that has not yet been widely by the kinesin uncoupling mutation, has been proposed to adopted by motors workers, is to select mutants for analysis participate in communication between the ATP- and filament- from existing collections recovered from genetic screens. binding sites (Ruppel and Spudich, 1996). The difference in Several such screens have been carried out for myosins biochemical effects between the myosin and kinesin (Bejsovec and Anderson, 1988; Patterson and Spudich, 1995) uncoupling mutants indicates that the mechanisms of the and kinesins (Saxton et al., 1991; Hoyt et al., 1993). A potential myosin and kinesin motors could differ considerably despite disadvantage of using mutants recovered in genetic screens is the structural homology of the motors. The uncoupling mutants that the mutants could map to any region of the protein and thus have the potential to provide insights into motor function therefore probably include those that are specific for the that cannot be obtained from loss-of-function or null mutants. myosin or kinesin that was the target of the screen, as well as Other uncoupling mutants are likely to be discovered, since those that affect basic motor functions. This means that a there are probably several different ways of uncoupling selection step, which could itself be as time-consuming as nucleotide hydrolysis from force generation by the motors. The screening mutants obtained by molecular mutagenesis, must be effects of these mutants are difficult to predict because the performed to identify informative mutants for analysis. mechanism of coupling between nucleotide hydrolysis and Unexpectedly, two previously reported genetic screens motility is not well understood. The uncoupling mutants could yielded mutants all of which mapped to the motor domain of therefore provide further surprises as they reveal unexpected a myosin or kinesin protein. Bejsovec and Anderson (1988) aspects of motor function. screened for dominant muscle defects in C. elegans that caused paralysis, targeting the unc-54 muscle myosin heavy chain gene. They found that the most strongly dominant alleles MUTAGENESIS SCREENS AND MUTANT altered highly conserved residues in the nucleotide-binding SELECTION region of the motor domain, whereas other mutant alleles affected the actin-binding region (Bejsovec and Anderson, Molecular screens 1990). They hypothesized that the basis of the dominance and How can informative ATPase or uncoupling mutants be localization of the mutations to the motor domain was the identified? Several strategies can be used to obtain mutants effects of the mutants in disrupting assembly of thick filaments. for analysis (Table 2). These strategies include extensive The sensitivity of thick filament assembly to perturbations in molecular mutagenesis coupled with biochemical or the ability of myosin to bind to actin and hydrolyze nucleotide phenotypic analysis − for example, random mutagenesis or thus provided an effective in vivo selection for mutants alanine-scanning mutagenesis has been used to mutagenize affecting motor function. The second screen that produced specific regions of myosins (Porter and Montell, 1993; Ruppel mutants with defective motor function was for yeast and Spudich, 1996; Sasaki and Sutoh, 1998) or surface residues suppressors of the temperature-induced lethality of cin8-ts of the kinesin motor domain (Woehlke et al., 1997). Molecular kip1-∆ cells (Hoyt et al., 1993), which are null for two kinesin mutagenesis is capable of yielding valuable information, but is proteins, Cin8p and Kip1p, that overlap in function (Saunders labor intensive, involving the construction of many mutant and Hoyt, 1992). All seven extragenic suppressors recovered plasmids and the characterization of the plasmids by sequence in the screen mapped within or just adjacent to the conserved analysis. Screening the plasmids by protein expression, motor domain of Kar3p (Hoyt et al., 1993), a third kinesin purification, and biochemical or phenotypic analysis is both protein, which is thought to produce force in the mitotic spindle that opposes the force produced by Cin8p and Kip1p (Saunders and Hoyt, 1992). The recovery of kar3 mutants as suppressors ∆ Table 2. Motor mutants for analysis of cin8-ts kip1- lethality is consistent with the idea that the motors produce antagonistic forces: loss of Kar3p motor Mutant source Disadvantages Advantages function should suppress the unopposed force produced by Molecular screen Labor-intensive Target motor Kar3p in cin8 kip1 null cells. Genetic screen Target entire protein Global effects These and other available mutant collections provide ample material for analysis, but the informative mutants must still be Rational design Limited by criteria Site-specific identified from among the other mutants. Several criteria can Natural variant Limited availability Global effects be used to select mutants that are likely to be informative. 1316 S. A. Endow

These include the degree of conservation of the mutated only identify the regions of the motor that are essential for residue. The extensive sequence alignments now available for function, but also reveal the key residues that are required and the myosins (Hodge and Cope, 1996) and kinesins (Greene et tell us how the motor can be altered to produce unexpected al., 1996) can be used to determine how highly conserved the effects, such as those of the myosin and kinesin uncoupling mutated residue is. The mutants can also be mapped onto the mutants. Without the use of these mutants, it is not obvious available crystal structures to identify the structural element how to obtain information about essential motor functions − that is affected. Mutants that alter highly conserved or invariant for example, how ATP hydrolysis is stimulated by binding of residues that are present in structural elements thought to be the motor to its filament (Song and Endow, 1998). These involved in nucleotide hydrolysis or binding of the motor to mutants can reveal the coupling between nucleotide hydrolysis microtubules or actin filaments are good candidates for and motor-filament interactions by uncoupling these basic analysis. The same mutation can be expressed in different motor functions in different ways. The uncoupling mutants myosins or kinesins to determine whether the effect of the analyzed to date validate the strategy of studying mutants mutation is specific to a given motor protein or general to recovered in genetic screens by showing that the mutants can motors of its family or class. The mutations that are general to provide unexpected and important information about essential motor function should cause similar effects in different motors, motor regions and their function that could not be predicted rather than being specific to a given motor. from the crystal structures alone. Mutants can also be selected on the basis of their genetic effects. Although the effects of ATPase or uncoupling mutants Rational design in live cells cannot be predicted with certainty, mutants that A third source of mutants for analysis is rational design − have dominant effects in vivo are likely to cause interesting identifying and mutating functionally important motor regions changes in motor function. Dominant mutants are mutants that on the basis of rational considerations. These regions can often show defects in combination with a wild-type gene − the cell be identified as those containing highly conserved or invariant is still mutant even though a wild-type gene is present. residues, but are not necessarily limited to these regions. Further Although several different properties of proteins or protein- useful targets are conserved structural elements that might not protein interactions can result in dominance, one interaction consist of conserved residues (Branden and Tooze, 1991). that has been widely used to obtain information about protein Information from related proteins on which work is more function in vivo is interference of the mutant protein with the advanced, such as the G proteins, can be used to identify function of the wild-type protein. Such antagonistic protein functionally or structurally important regions in the motor interactions have been termed ‘antimorphic’ (Muller, 1932) or proteins by analogy. Based on the structural analogies between ‘dominant negative’ (Herskowitz, 1987). An example of a the G proteins, the myosins and the kinesins, the switch I and mutant recovered in a genetic screen and selected for analysis switch II regions (Fig. 1) that are proposed to undergo changes on the basis of its dominant mutant effects in vivo is the kinesin during nucleotide hydrolysis are likely to be informative. uncoupling mutant discussed above. This mutant was Mutating conserved residues in these regions of the motors originally recovered as a kar3 suppressor mutant in the genetic could trap the motors in a prehydrolysis or transition screen for suppressors of cin8 kip1 lethality (Hoyt et al., 1993). conformation, or change the kinetics of hydrolysis to reveal The kar3 mutant showed a dominant phenotype in genetic intermediate states. Determining the ability of the motor to bind tests, which can now be explained by the strong binding of the to its filament in the prehydroysis and transition states will be mutant motor to microtubules even in the presence of ADP. critically important to understanding how the motor interacts The mutant also affects a residue that is invariant among the with its filament. Although the kinesin motors are believed to known kinesin proteins, which indicates that it is likely to bind tightly to microtubules in the presence of ATP (Brady, affect basic motor function rather than an activity specific to 1985) and weakly in the presence of ADP (Ma and Taylor, Kar3p. When we tested the same mutation in another kinesin 1995; Crevel et al., 1996; Pechatnikova and Taylor, 1997), and protein, Ncd, we observed the same effects on ADP and myosin is thought to bind tightly to actin filaments in the microtubule binding (Song and Endow, 1998), thus the presence of ADP and weakly in the presence of ATP (Sellers mutational change has a general effect on the kinesin motors and Goodson, 1995), it is not certain what happens during and affects an essential motor property: the ability to hydrolyze transition states for either the kinesins or myosins. Transition ATP at an enhanced rate when the motor is bound to states in nucleotide hydrolysis are likely to correspond to the microtubules. The myosin uncoupling mutants that were force-generation steps of the motor (Howard, 1996), thus recovered previously in Dictyostelium (which is haploid) interactions between motors and microtubules or actin filaments showed ‘intermediate’ effects and were not reported to have during these states will be extremely important in understanding been tested for dominance over wild type. It is not certain that how molecular motors generate force and move along their all uncoupling mutants will be dominant, although some are cytoskeletal tracks. It is also likely that important principles will likely to have dominant negative effects because of their be learned by targeting the motor regions analogous to those unusual biochemical properties. The unc-54 myosin mutants, that are thought to change in conformation in G proteins. The all of which showed dominant effects in vivo, are likely to be current thinking is that although conserved switch regions a rich source of interesting mutants that will yield valuable exist in the kinesins, myosins and G proteins, fundamental information about myosin function. differences exist in the mechanism by which these proteins The mutants found in genetic screens will be important to function. For example, the G proteins use switch II residues to study because they should affect all of the basic motor catalyze nucleotide hydrolysis (Frech et al., 1994), whereas functions needed in live cells, rather than a subset of the myosin uses a switch I residue (Shimada et al., 1997). What functions required for motility in vitro. These mutants can not residues do the kinesins use? This question can be addressed by Molecular motors − a paradigm for mutant analysis 1317 careful design of mutants, based on structural homologies and and interpretation, together with an arsenal of available sequence considerations. methods of analysis. This will include structural methods, such as X-ray crystallography and NMR (nuclear magnetic Natural variants resonance), and sensitive fluorescence methods – e.g. A final source of informative mutants that should not be fluorescence anisotropy (Rosenfeld et al., 1996; Sweeney et al., overlooked is natural variants, proteins within a family that 1998) and FRET (fluorescence resonance energy transfer; differ from most of the other family members and arise by Suzuki et al., 1998) Ð together with high-resolution microscopy spontaneous changes followed by natural selection. Naturally methods to visualize single motors and fluorescent nucleotides occurring variants are an important source of structural in motility assays (Funatsu et al., 1995; Ishijima et al., 1998). changes coupled to changes in function, and could perform In vivo analysis and tests of mutant function should not be novel functions in the cells in which they are found. Kinesin overlooked: these studies will be extremely valuable in testing motors such as Drosophila Ncd and other members of the C- interpretations from in vitro studies. Ideally, the in vivo tests terminal motor kinesin subfamily (Endow, 1999a) can be of mutant function should include live analysis using recently considered to be natural variants of the kinesins. These motors developed methods to visualize proteins in the cell (Chalfie et differ in domain organization from other kinesin proteins in al., 1994) and permit study of the biophysics of the motors in that the conserved motor domain is C-terminal to the coiled- the cell. Finally, the approaches and strategies for mutant coil dimerization domain, instead of N-terminal, as in the design and selection described here are not limited to remainder of the kinesin family. The C-terminal motor kinesins molecular motors, but are applicable to other proteins under differ strikingly in motility from other kinesins, moving to the study whose mechanisms of function have yet to be minus end of microtubules instead of the plus end. This determined. difference in polarity of movement has allowed workers to examine the molecular basis of motor directionality (Endow, Special thanks to T. Salmon for thought-provoking and insightful 1999b). Further work is expected to provide basic insights into comments about the ideas presented here, R. Cross for reading the motor function, since directionality is a fundamental property manuscript, and H.-W. Park for notes on protein folding. Work in my of molecular motors. laboratory is supported by grants from the NIH and HFSP. Another example of a naturally occurring variant is myosin VI, which differs from other myosins in that it has two small REFERENCES insertions and a third large insertion in the conserved motor domain (Wells et al., 1999). The larger, 53 amino acid insertion Avraham, K. B., Hasson, T., Steel, K. P., Kingsley, D. M., Russell, L. B., is present in the myosin ‘converter’ region, a region at the Mooseker, M. S., Copeland, N. G. and Jenkins, N. A. (1995). The mouse junction between the α-helical rod and conserved motor Snell’s waltzer deafness gene encodes an unconventional myosin required for structural integrity of inner ear hair cells. Nature Genet. 11, 369-375. domain that was identified from crystal structures (Dominguez Bejsovec, A. and Anderson, P. (1988). Myosin heavy-chain mutations that et al., 1998; Houdusse et al., 1999). The detection of this disrupt Caenorhabditis elegans thick filament assembly. Genes Dev. 2, insertion led to the discovery that myosin VI is minus-end 1307-1317. motor that moves to the pointed end of actin filaments, in Bejsovec, A. and Anderson, P. (1990). Functions of the myosin ATP and actin contrast to all other known myosins, which are plus-end motors binding sites are required for C. elegans thick filament assembly. Cell 60, 133-140. (Wells et al., 1999). The helical rod, or putative lever arm, of Bloom, G. S. and Endow, S. A. (1995). Motor proteins 1: kinesins. Protein myosin VI appears to point in the opposite direction to that of Profile 2, 1109-1171. myosin II when the motors are bound to actin filaments. This Bourne, H. R., Sanders, D. A. and McCormick, F. (1991). The GTPase led Wells et al. (1999) to propose that the converter region acts superfamily: conserved structure and molecular mechanism. Nature 349, 117-127. as a ‘pivot point’ for the lever arm. These findings demonstrate Brady, S. T. (1985). A novel brain ATPase with properties expected for the that small structural changes can cause profound changes in fast axonal transport motor. Nature 317, 73-75. protein function and underscore the importance of careful Branden, C. and Tooze, J. (1991). Introduction to Protein Structure. Garland sequence analysis to identify functionally important regions of Publishing, Inc., New York. the motors. Brown, S. S. (1999). Cooperation between microtubule- and actin-based motor proteins. Annu. Rev. Cell Dev. Biol. 15, 63-80. Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. W. and Prasher, D. C. (1994). Green fluorescent protein as a marker for gene expression. Science 263, 802- PERSPECTIVES 805. Cooke, R. (1986). The mechanism of muscle contraction. CRC Crit. Rev. Biochem. 21, 53-118. Together with the mutants from molecular and genetic screens, Crevel, I. M.-T. C., Lockhart, A. and Cross, R. A. (1996). Weak and strong the mutants obtained by rational design or discovered as states of kinesin and ncd. J. Mol. Biol. 257, 66-76. natural variants have the potential to provide basic information Dominguez, R., Freyzon, Y., Trybus, K., M. and Cohen, C. (1998). Crystal about the motor mechanism. These mutants should allow structure of a vertebrate smooth muscle myosin motor domain and its workers to probe the mechanism of motor function in detail complex with the essential light chain: visualization of the pre-power stroke state. Cell 94, 559-571. and extend the information provided by the crystal structures Endow, S. A. (1999a). C-terminal motor kinesin proteins. In Guidebook to the and biochemistry. The analysis of these mutants should take Cytoskeletal and Motor Proteins (ed. T. Kreis and R. Vale), pp. 403-408. into account the interactions of the motors both with Oxford:Oxford University Press. microtubules or actin filaments and with nucleotides that Endow, S. A. (1999b). Determinants of molecular motor directionality. Nature Cell Biol. 1, 163-167. change during the hydrolysis cycle. To solve a problem as Endow, S. A. (1999c). Microtubule motors in spindle and chromosome complex and important as energy conversion by molecular motility. Eur. J. Biochem. 262, 12-18. motors, it will be necessary to use ingenuity in mutant design Fisher, A. J., Smith, C. A., Thoden, J. B., Smith, R., Sutoh, K., Holden, H. 1318 S. A. Endow

M. and Rayment, I. (1995). X-ray structures of the myosin motor domain identifying cold-sensitive myosin II mutants in Dictyostelium. Genetics 140, of Dictyostelium discoideum complexed with MgADP⋅BeFx and 505-515. − MgADP⋅AlF4 . Biochemistry 34, 8960-8972. Pechatnikova, E. and Taylor, E. W. (1997). Kinetic mechanism of Franken, S. M., Scheidig, A. J., Krengel, U., Rensland, H., Lautwein, A., monomeric non-claret disjunctional protein (Ncd) ATPase. J. Biol. Chem. Geyer, M., Scheffzek, K., Goody, R. S., Kalbitzer, H. R., Pai, E. F. et al. 272, 30735-30740. (1993). Three-dimensional structures and properties of a transforming and Porter, J. A. and Montell, C. (1993). Distinct roles of the Drosophila ninaC a nontransforming glycine-12 mutant of p21H-ras. Biochemistry 32, 8411- kinase and myosin domains revealed by systematic mutagenesis. J. Cell 8420. Biol. 122, 601-612. Frech, M., Darden, T. A., Pedersen, L. G., Foley, C. K., Charifson, P. S., Raw, A. S., Coleman, D. E., Gilman, A. G. and Sprang, S. R. (1997). Anderson, M. W. and Wittinghofer, A. (1994). Role of glutamine-61 in Structural and biochemical characterization of the GTPγS-, GDP¥ Pi-, and 42 the hydrolysis of GTP by p21H-ras: an experimental and theoretical study. GDP-bound forms of a GTPase-deficient Gly jVal mutant of Giα1. Biochemistry 33, 3237-3244. Biochemistry 36, 15660-15669. Funatsu, T., Harada, Y., Tokunaga, M., Saito, K. and Yanagida, T. (1995). Rayment, I., Rypniewski, W. R., Schmidt-Base, K., Smith, R., Tomchick, Imaging of single fluorescent molecules and individual ATP turnovers by D. R., Benning, M. M., Winkelmann, D. A., Wesenberg, G. and Holden, single myosin molecules in aqueous solution. Nature 374, 555-559. H. M. (1993). Three-dimensional structure of myosin subfragment-1: a Gibbons, I. R. (1996). The role of dynein in microtubule-based motility. Cell molecular motor. Science 261, 50-58. Struct. Funct. 21, 331-342. Rosenfeld, S. S., Correia, J. J., Xing, J., Rener, B. and Cheung, H. C. Gibson, F., Walsh, J., Mburu, P., Varela, A., Brown, K. A., Antonio, M., (1996). Structural studies of kinesin-nucleotide intermediates. J. Biol. Chem. Beisel, K. W., Steel, K. P. and Brown, S. D. (1995). A type VII myosin 271, 30, 212-30, 221. encoded by the mouse deafness gene shaker-1. Nature 374, 62-64. Ruppel, K. M. and Spudich, J. A. (1995). Myosin motor function: structural Goldberg, M. L., Gunsalus, K. C., Karess, R. E. and Chang, F. (1998). and mutagenic approaches. Curr. Opin. Cell Biol. 7, 89-93. Cytokinesis. In Dynamics of Cell Division (ed. S. A. Endow and D. M. Ruppel, K. M. and Spudich, J. A. (1996). Structure-function studies of the Glover), pp. 270-316. Oxford: Oxford University Press. myosin motor domain: importance of the 50-kDa cleft. Mol. Biol. Cell 7, Greene, E. A., Henikoff, S. and Endow, S. A. (1996). The kinesin home page. 1123-1136. www. blocks. fhcrc. org/~kinesin/ Sablin, E. P., Kull, F. J., Cooke, R., Vale, R. D. and Fletterick, R. J. (1996). Gulick, A. M., Song, H., Endow, S. A. and Rayment, I. (1998). X-ray crystal Crystal structure of the motor domain of the kinesin-related motor ncd. structure of the yeast Kar3 motor domain complexed with Mg¥ADP to 2. 3 Nature 380, 555-559. Å resolution. Biochemistry 37, 1769-1776. Sack, S., Kull, F. J. and Mandelkow, E. (1999). Motor proteins of the kinesin Herskowitz, I. (1987). Functional inactivation of genes by dominant negative family Structure, variations and nucleotide binding sites. Eur. J. Biochem. mutations. Nature 329, 219-222. 262, 1-11. Hirokawa, N. (1998). Kinesin and dynein superfamily proteins and the Sasaki, N. and Sutoh, K. (1998). Structure-mutation analysis of the ATPase mechanism of organelle transport. Science 279, 519-526. site of Dictyostelium discoideum myosin II. Adv. Biophys. 35, 1-24. Hodge, T. P. and Cope, J. (1996). The myosin home page. http://www. mrc- Saunders, W. S. and Hoyt, M. A. (1992). Kinesin-related proteins required lmb. cam. ac. uk/myosin/myosin. html for structural integrity of the mitotic spindle. Cell 70, 451-458. Houdusse, A., Kalabokis, V. N., Himmel, D., Szent-Gyorgyi, A. G. and Saxton, W. M., Hicks, J., Goldstein, L. S. B. and Raff, E. C. (1991). Kinesin Cohen, C. (1999). Atomic structure of scallop myosin subfragment S1 heavy chain is essential for viability and neuromuscular functions in complexed with MgADP: a novel conformation of the myosin head. Cell Drosophila, but mutants show no defects in mitosis. Cell 64, 1093-1102. 14, 459-470. Scott, W. G., Murray, J. B., Arnold, J. R. P., Stoddard, B. L. and Klug, A. Howard, J. (1996). The movement of kinesin along microtubules. Annu. Rev. (1996). Capturing the structure of a catalytic RNA intermediate: the Physiol. 58, 703-729. hammerhead ribozyme. Science 274, 2065-2069. Hoyt, M. A., He, L., Totis, L. and Saunders, W. S. (1993). Loss of function Sellers, J. R. and Goodson, H. V. (1995). Motor proteins 2: myosin. Protein of Saccharomyces cerevisiae kinesin-related CIN8 and KIP1 is suppressed Profile 2, 1323-1423. by KAR3 motor domain mutations. Genetics 135, 35-44. Shimada, T., Sasaki, N., Ohkura, R. and Sutoh, K. (1997). Alanine scanning Inoué, S. and Salmon, E. D. (1995). Force generation by microtubule mutagenesis of the switch I region in the ATPase site of Dictyostelium assembly/disassembly in mitosis and related movements. Mol. Biol. Cell 6, discoideum myosin II. Biochemistry 36, 14037-14043. 1619-1640. Song, H. and Endow, S. A. (1998). Decoupling of nucleotide- and Ishijima, A., Kojima, H., Funatsu, T., Tokunaga, M., Higuchi, H., Tanaka, microtubule-binding in a kinesin mutant. Nature 396, 587-590. H. and Yanagida, T. (1998). Simultaneous observation of individual Sprang, S. R. (1997). G protein mechanisms: insights from structural analysis. ATPase and mechanical events by a single myosin molecule during Annu. Rev. Biochem. 66, 639-678. interaction with actin. Cell 92, 161-171. Stoddard, B. L., Cohen, B., E., Brubaker, M., Mesecar, A. D. and Kozielski, F., De Bonis, S., Burmeister, W. P., Cohen-Addad, C. and Wade, Koshland, D. E. J. (1998). Millisecond Laue structures of an enzyme- R. H. (1999). The crystal structure of the minus-end-directed microtubule product complex using photocaged substrate analogs. Nature Struct. Biol. motor protein ncd reveals variable dimer conformations. Structure 7, 1407- 5, 891-897. 1416. Suzuki, Y., Yasunaga, T., Ohkura, R., Wakabayashi, T. and Sutoh, K. Kull, F. J., Sablin, E. P., Lau, R., Fletterick, R. J. and Vale, R. D. (1996). (1998). Swing of the lever arm of a myosin motor at the isomerization and Crystal structure of the kinesin motor domain reveals a structural similarity phosphate-release steps. Nature 396, 380-383. to myosin. Nature 380, 550-555. Sweeney, H. L., Rosenfeld, S. S., Brown, F., Faust, L., Smith, J., Xing, J., Ma, Y. Z. and Taylor, E. W. (1995). Mechanism of microtubule kinesin Stein, L. and Sellers, J. (1998). Kinetic tuning of myosin via a flexible ATPase. Biochemistry 34, 13242-13251. loop adjacent to the nucleotide binding pocket. J. Biol. Chem. 273, 6262- Muller, H. J. (1932). Further studies on the nature and causes of gene 6270. mutations. Proc. 6th Internat. Congress Genet. 1, 213-255. Waters, J. C. and Salmon, E. D. (1996). Cytoskeleton: a catastrophic kinesin. Muller, J., Marx, A., Sack, S., Song, Y. H. and Mandelkow, E. (1999). The Curr. Biol. 6, 361-363. structure of the nucleotide-binding site of kinesin. Biol. Chem. 380, 981- Weil, D., Blanchard, S., Kaplan, J., Guilford, P., Gibson, F., Walsh, J., 992. Mburu, P., Varela, A., Levilliers, J., Weston, M. D. et al. (1995). Defective Nasmyth, K. and Jansen, R. P. (1997). The cytoskeleton in mRNA myosin VIIA gene responsible for Usher syndrome type 1B. Nature 374, localization and cell differentiation. Curr. Opin. Cell Biol. 9, 396-400. 60-61. Nonaka, S., Tanaka, Y., Okada, Y., Takeda, S., Harada, A., Kanai, Y., Wells, A. L., Lin, A. W., Chen, L.-Q., Safer, D., Cain, S. M., Hasson, T., Kido, M. and Hirokawa, N. (1998). Randomization of left-right Carragher, B. O., Milligan, R. A. and Sweeney, H. L. (1999). Myosin VI asymmetry due to loss of nodal cilia generating leftward flow of is an actin-based motor that moves backwards. Nature 400, 505-508. extraembryonic fluid in mice lacking KIF3B motor protein. Cell 95, 829- Woehlke, G., Ruby, A. K., Hart, C. L., Ly, B., Hom-Booher, N. and Vale, 837. R. D. (1997). Microtubule interaction site of the kinesin motor. Cell 90, 207- Patterson, B. and Spudich, J. A. (1995). A novel positive selection for 216.