(A) TMR-labeled FRB-Dyn1_{331 kDa} plus rapamycin did not move processively in the single-molecule TIRF assay. Kymographs were made by horizontally stacking line scans along the axonemal axis. Some stationary dynein molecules were seen (vertical lines), but most molecules bound transiently, and no moving molecules were observed (see diagonal lines in Figure 2B).
(B) Dimerization of TMR-labeled FRB-Dyn1_{331 kDa} and FKBP12-Dyn1_{331 kDa} by rapamycin produces processive runs in the single-molecule TIRF assay. Diagonal lines in the kymograph represent dynein molecules that are moving along the axoneme over time (see also Movie S3). A histogram of rapamycin-dimerized dynein run lengths shows a run length of 1.2 ± 0.1 μm (mean ± SE; n = 223, Figure S3).

(A)GST-Dyn1_{331 kDa} labeled with a Qdot at its NH₂ terminus (center of mass of the dimer) takes discrete steps along axonemal microtubule tracks. The raw data are shown with black circles and lines connecting the data points, and steps detected by a step finding program (see Experimental Procedures) are shown in red (step size values are noted in red). Data was acquired at 4 μM ATP every 70 ms. Traces from four different dynein molecules are shown and more representative examples can be found in Figure S7.
(B) A single Qdot positioned at the COOH terminus of one motor domain in the GST- Dyn1_{331 kDa} dimer shows larger steps than NH₂-terminally labeled GST-Dyn1_{331 kDa}. Data were acquired at 10 μM ATP every 100 ms. Traces from four different dynein molecules are shown, and more representative examples can be found in Figure S8.

(A) Diagram of an axoneme and the direction of on-axis (red arrow) and off-axis (blue arrow) movement.
(B–D) Three examples of on-axis (parallel to the microtubule long axis) and off-axis (perpendicular to the microtubule long axis) stepping by head-labeled GST-Dyn1_{331 kDa}-Qdot. (B) shows an example of no off-axis stepping, (C) shows the most typical trace of a few off-axis steps, and (D) shows a rare trace of many large off-axis displacements. On-axis stepping position is represented by the black circles (raw data), and the red line shows a fit with the step-finding program, whereas off-axis stepping position is represented by blue circles (raw data) and lines. Dashed orange lines illustrate that on- and off-axis steps occur simultaneously. The panels on the right depict off-axis (x axis) versus on-axis (y axis) movements. The direction of movement is toward the top of the page. Each step is represented by a black circle, and the steps are connected with a black line. See Figure S10 for additional traces and discussion of off-axis displacement.

(A) Schematic of the native dynein heavy-chain dimer. The NH₂ terminus of the motor is labeled with GFP (green asterisk), and the COOH terminus of the motor is labeled with the HaloTag and covalently bound to tetramethylrhodamine (TMR, red asterisk).
(B) Purification of native yeast dynein expressed from its endogenous promoter. Dynein, NH₂-terminally tagged with two copies of the Protein A IgG binding domain, followed by a TEV cleavage site, was bound to IgG beads and then released from the beads with TEV protease. Subsequently, dynein was bound to microtubules in the absence of ATP, microtubule-dynein complexes were collected by centrifugation, and then dynein was released from microtubules by addition of 5 mM ATP (see Supplemental Data for details). A silver-stained gel shows the full-length dynein heavy chain after the final purification step. Tubulin is also present from the last affinity step.
(C) Tagging native yeast dynein (the construct described in [A] and [B]) does not affect its function in nuclear segregation in vivo. For monitoring nuclear segregation, the number of binucleate mitotic cells was determined after growth at 16°C for 16 hr. dyn1Δ is a yeast strain bearing a deletion in the dynein heavy-chain gene. The mean and standard error of the mean are shown (n > 300).
(D) Histogramof velocities of single native dynein molecules moving along axonemal microtubules (fit with a Gaussian function; black line). The mean velocity is 85 ± 30 nm/s (SD); n = 334.
(E) Histogram of run lengths of single dynein molecules moving along axonemal microtubules. The run length is 1.9 ± 0.2 μm (mean ± standard error [SE]; n = 334; Figure S3). See Movie S1 for images of native dynein moving along axonemes.

(A) Schematic showing dynein heavy-chain truncations and tags. A HaloTag at the COOH terminus allowed attachment of TMR for single-molecule assays. The orange oval represents GST, which was used to dimerize monomeric constructs either at the NH₂ or the COOH terminus. GFP (green rectangle) was added to the NH₂ terminus of all proteins and was used to anchor dynein molecules to glass with a GFP antibody in microtubule gliding assays. Here we functionally define the term “linker” as the ~400 amino acids, located NH₂-terminal to the start of the AAA domain, that we have found necessary for dynein motility. The scale bar represents 500 aa.
(B) SYPRO-ruby (Molecular Probes)-stained protein gel of the truncated dynein proteins. The gel shows GST-Dyn1_{331 kDa}, GST-Dyn1_{314 kDa}, and Dyn1_{331 kDa}-GST after IgG and microtubule affinity-purification steps, and GST-Dyn1_{311 kDa} is shown after the first affinity-purification step because this protein releases poorly from microtubules in the presence of ATP.
(C) The mean single-molecule run length ± SE for each of the dynein constructs is shown. GST-Dyn1_{311 kDa} and Dyn1_{331 kDa}-GST both bound to microtubules in the single-molecule assay, but no moving molecules were observed.
(D) Mean single (light gray) and multiple motor (dark gray) velocity ± SD for each of the dynein constructs. GST-Dyn1_{311 kDa} bound to microtubules and axonemes, but showed no motility in either assay. Dyn1_{331 kDa}-GST bound, but was not capable of moving on, axonemes in the single-molecule assay.
(E) The tail domain is required for the dynein power stroke. Monomeric Dyn1_{331 kDa} was anchored to glass for microtubule gliding assays via GFP attached either to its tail (NH₂ terminus; GFP-Dyn1_{331 kDa}) or end of its motor domain (COOH terminus; Dyn1_{331 kDa}-GFP). GFP-Dyn1_{331 kDa} moved microtubules along the glass (mean ± SD is shown), whereas Dyn1_{331 kDa}-GFP could bind microtubules but not move them under our standard buffer conditions (including 50 mM KAcetate). Under higher ionic strength conditions (300 mM KAcetate), very slow (3 ± 1 nm/s) movement of Dyn1_{331 kDa}-GFP was observed.
(F) GST-Dyn1_{331 kDa} purified from yeast strains lacking the light intermediate chain (Dyn3), the intermediate chain (Pac11), or the Lis 1 protein (Pac1) shows a similar run length to a yeast strain that does not harbor these deletions (GST-Dyn1_{331 kDa}). The bar graph shows the mean single-molecule run length ± SE for each strain background (Figure S3). The single-molecule velocities for dynein purified from each of these yeast strains was also similar to the control yeast strain (mean velocity ± SD = 103 ± 32, 108 ± 33, 98 ± 46, and 98 ± 32 nm/s for GST- Dyn1_{331 kDa}, dyn3Δ, pac11Δ, and pac1Δ, respectively). n values for the data in this figure are between 184 and 334.

(A) A histogram of tail-labeled Qdot-GST-Dyn1_{331 kDa} step sizes reveals a major peak at ~8 nm with a considerable tail of longer steps. Included in the histogram are 1342 steps from 27 moving dynein molecules. Backward steps make up 20% of the total. A histogram of dwell times between steps is fit with a single exponential function, and the decay constant reveals a stepping rate (k) of 0.13 ± 0.01 per second per μM ATP.
(B) A histogram of head-labeled GST-Dyn1_{331 kDa}-Qdot step sizes reveals a major peak at ~16–18 nm with a considerable tail of longer steps. Included in the histogram are 1690 steps from 29 moving dynein molecules. Backward steps make up 13% of the total steps. The dwell-time histogram is fit with a convolution of two exponential functions with equal decay constants (the fit indicates a dynein stepping rate (k) of 0.14 ± 0.01 per second per μMATP).

A stepping model for cytoplasmic dynein that takes into account the step-size and dwell-time data for head- and tail-labeled dynein. The head (COOH terminus)- labeled dynein (H) is represented by the red tag, and the center- of-mass-labeled dynein (T) is represented by the yellow tags (the AAA ring is shown schematically from this top view as a rectangle with rounded edges; the AAA rings and tubulin dimers are shown approximately to scale).
(A) In this Alternating Shuffle model, the two dynein motor domains alternate between forward and rear positions; the tail-labeled dynein takes predominantly successive 8 nm steps (the distance between tubulin dimers). In contrast, a single dynein motor domain takes a 16 nm step and then remains stationary (a 0 nm step) as its partner ring advances. Structural considerations suggest that the rings may align parallel to one another and partially overlap (see Discussion). Although dynein is shown here moving along a single protofilament, we have shown that dynein can also move on neighboring protofilaments (see Figure 6).
(B–D) Another view of the Alternating Shuffle model illustrating how dynein could take 8 nm as well as 16 nm steps from either a greater separation of the motor rings (C) or angular movements of the microtubule binding stalks (D) (a combination of these mechanisms is also possible). In (B)–(D), possible tubulin binding sites (both forward and rearward) are indicated by the blue-shaded tubulin subunits, with the darkest blue representing the most frequently observed step size of 8 nm.