Dynein shifts into second gear

Although dynein, a microtubule-based molecular motor, has a respectable 40-plus-year research history, only recently has it begun to yield the innermost secrets of how it operates. Current efforts address how the bulky motor domains coordinate their activity to make processive steps along a cytoplasmic microtubule (1–3), suggest that the motor can vary its step size (4), and even demonstrate that dynein can run backward (5). The article by Shima et al. (6) in this issue of PNAS reveals another exceptional property of dynein, namely that the motor has more than one mode of force production. Such a discovery reinforces the view that dynein is more than a simple motor, perhaps indicating that further surprises are yet to come.

Dynein Is Structurally Different from Kinesin or Myosin

Dynein, myosin, and kinesin constitute the three major families of cytoskeleton-based motor proteins that operate in eukaryotic cells (7, 8). The exceptionally large mass (1–2 million Da as a stand-alone macromolecular complex) has long distinguished dynein from the other two families of motors. For both the native molecules and their motor domains, dynein is approximately four times larger than the actin-based myosin II. This size difference is more dramatic when dynein is compared with the other microtubule-based motor family, kinesin. The minimal mass required for complete dynein motor activity is ≈300–400 kDa, whereas kinesin fragments as small as 35–40 kDa support microtubule movement. Yet, despite this size difference, both microtubule-based motors do approximately the same thing. Both move along a microtubule at rates in the range of 0.5–2 μm/s, and both generate forces in the range of 1–5 pN (9). There are important differences in the details, but none that would obviously require an order of magnitude differential in size to carry out basic force production. Why then is dynein so large? The answer certainly lies in its evolutionary history, but we can at least make some sense of this size differential. Structural biology methods demonstrated a distinctive ring-like organization to the dynein motor core (10); a complementary bioinformatics comparison identified a dynein ancestry with the AAA family of proteins (11). The AAA proteins include a diverse group of macromolecular complexes that typically assemble into ring-shaped structures, a configuration that is unlike that of either myosin or kinesin. Given its other unique features (e.g., multiple nucleotide-binding motifs, an extended microtubule-binding domain), it is clear that dynein exhibits a body plan dramatically different from the other two motor families.

If dynein is built differently, then does it also work differently? The simple answer is probably not. Myosin and kinesin both generate force through a rotation of the head around the cargo-binding tail linkage. The conformational changes are directly coupled with the ATPase cycle of the motor and are best summarized as the lever-arm model of force production (ref. 12 and see also ref. 13). Structural studies have demonstrated conformational changes consistent with a similar rotation in dynein, around the head tail junction (14). These static views are supported by FRET measurements indicating ATP-dependent movement of the tail relative to the head in a cytoplasmic dynein (15). The study by Shima et al. (6) goes one step further, in showing that the ability of a dynein to rapidly move a microtubule requires a powerstroke contribution from the tail. By incorporating biotin tags into the tail and motor domain sequences, Shima et al. were able to selectively attach the motor onto a substrate and then ask what part of the motor is relevant for producing movement. Anchorage of the motor at the most distal position of the tail construct (farthest from the motor domain) provides the fastest rate of in vitro microtubule gliding (3.2 μm/s). Anchorages closer to the motor, or directly on the motor itself, produce significantly reduced rates of microtubule gliding, arguing that motion of the tail is directly coupled with force production. The study by Shima et al., together with a recent analysis of yeast dynein movement (1), provides the most direct demonstration thus far that the tail motion drives the rapid dynein-mediated microtubule movement. These findings are significant because they indicate that there may be a uniform mechanism to link nucleotide hydrolysis with linear movement in all three motor families. Even so, the tail feature in dynein is not as rigidly positioned as in kinesins or myosins (14, 16) and may be tethered in places across the face of the dynein ring (14). Thus, there remains considerable work to be done before we fully understand the mechanics of dynein's lever arm rotation.

The new level of sophistication in dynein analyses has crossed a critical threshold.

A Dynein Surprise

Perhaps the most surprising feature of the work by Shima et al. (6) is that anchorages on the motor do not completely abolish motility. Presumably without any contribution from the tail domain, motor-anchored dyneins are still able to support microtubule gliding. Although the rate is significantly slower than normal (0.03–0.06 μm/s), movement remains ATP-dependent and is generally directed toward the microtubule minus end. A similar type of movement is hinted at in the work by Reck-Peterson et al. (1). Thus, dynein appears to have a second gear to drive movement along a microtubule. Such motility could be powered by a biased Brownian motion or conformational changes in the ring or microtubule-binding stalk. Both ring and stalk likely change their positions simply as a function of the linkage between microtubule affinity and ATP catalysis. It seems unlikely that the slow mode would contribute much to the rapid dynein-driven organelle movements seen in vivo, but it may be important for maintaining tension on the cytoskeleton or dynein loading onto a microtubule.

Different Regulatory Strategies?

Although the dynein family encompasses a diverse group of motors, nearly all dyneins function within the axoneme to power ciliary and flagellar beating. A single cytoplasmic member performs most of the dynein functions in the rest of the cell. In contrast, a number of distinct kinesins carry out that motor's cytoplasmic actions. Each kinesin polypeptide is thought to contain unique sequences that provide spatial and temporal control. Although multiple kinesins could be independently regulated, how would a cell control the many functions of a single cytoplasmic dynein? The primary dynein contains an elaborate series of intermediate and light chain linkages that provide selectivity in binding cargo (17), but a good bet is that there also exists a mechanism to fine-tune motor activity for specific tasks. Indeed, dynein's rate of nucleotide hydrolysis appears sensitive to the ring geometry (18). Thus any modification or binding to the motor that alters ring geometry could in effect act to accelerate or decelerate dynein activity.

Details of the different modes of force production (forward or backward, slow or fast), the potential variations in step size, the unique motor architecture with multiple nucleotide binding domains, and the tethers between a lever arm tail and motor all are important details to consider when developing models of a dynein powerstroke. The key here is that such details may also provide glimpses of the potential means by which the motor can be regulated. A subtle tweak here or a pull there across the dynein ring could alter the motor's enzymatic rate or step distance and thus serve as a novel mechanism to turn the motor off or cause it to shift between gears. The new level of sophistication in dynein analyses, ranging from single-molecule work to image processing of motors in situ (19), has crossed a critical threshold that promises many interesting surprises in the next few years.