View all hidden authors and organizations

Contributed by Ronald D. Vale, June 29, 2021 (submitted for review on January 22, 2021; reviewed by Zev Bryant, Samara L. Reck-Peterson, and Ahmet Yildiz)

Dynein is a dimeric motor protein that moves along microtubules through two large loop domains (AAA loops) and microtubule binding domains (MTBD) driven by 5'-adenosine triphosphate and heat. It is composed of long coils. Separate. Previous single-molecule studies tracked the position of the AAA ring during exercise, but did not track the MTBD. Here, we tracked the relative positions of two MTBDs and an AAA ring through three-color nano-resolution imaging. Observations of the two MTBDs provide a direct measurement of how dynein steps on the tubulin subunits, while observations of three fiducial markers at the same time reveal the extraordinary flexibility of movement during exercise and previously unknown movement conformations state. The techniques presented here can be used to explore the conformational dynamics of many other macromolecular complexes.

The motor protein dynein undergoes coordinated conformational changes in its domains during movement along microtubules. Previous single-molecule studies analyzed the movement of the AAA loop of the dynein homodimer, but did not analyze the distal microtubule binding domain (MTBD) that steps along the track. Here, we track two MTBDs and one AAA ring of a single dynein at the same time with nanometer precision because it has gone through hundreds of steps using three-color imaging. We show that the AAA ring and MTBD do not always step at the same time and can take steps of different sizes. This variability of movement between the AAA loop and MTBD leads to unexpectedly large numbers of dynein conformational states during exercise. Extracting the data on the conformational transition deviation, we can accurately simulate the dynein stepping. Our results indicate that the flexibility between the main dynein domains is essential for dynein movement.

The microtubule-based motor protein dynein belongs to the AAA (ATPase associated with a variety of cellular activities) family and is responsible for most of the directional movement of the negative end along the microtubules (1, 2). Dynein plays a key role in many cellular processes and maintenance of cell structure, including cargo transportation, cilia movement, and the construction of mitotic spindles (3⇓ ⇓-6). Mutations or defects of cytoplasmic dynein are associated with a variety of pathologies, including cancer and neurological diseases (7, 8).

Compared with kinesin (9, 10) and myosin (11, 12), cytoskeletal motors have a compact spherical motion domain, and dynein is larger and more complex, with a size of approximately 1.4 MDa. Dynein consists of two heavy chains and several related polypeptide chains. The relevant chain mainly binds to the N-terminal tail region, dimerizes the heavy chain, regulates the function of dynein, and connects the motor to the cargo (4, 13, 14). The remaining two-thirds of the heavy chain constitutes the motor domain, which is the driving factor of dynein movement (15). The motor domain itself is divided into several domains: linker, AAA ring, stem and microtubule binding domain (MTBD). The AAA ring is composed of six different AAA domains, which are connected together to form an asymmetric hexamer ring (AAA1 to AAA6), of which only AAA1 to 4 can bind 5'-adenosine triphosphate (ATP) (16 ⇓ ⇓ –19). At the top of the AAA ring is the N-terminal connector, which serves as a mechanical element and connects the motion domain to the N-terminal tail. The large catalytic AAA loop of dynein and the small MTBD are separated by a coiled coil about 15 nanometers long extending from AAA4, called a stem (20⇓ –22).

After ATP binds to AAA1, the motor domain is released from the microtubules, and the linker undergoes an initiation stroke (linker bending). During the start-up stroke, the AAA ring has been observed to rotate relative to the joint, thereby biasing the recombination of the MTBD towards the negative end of the microtubule (1, 23). After ATP is hydrolyzed, the free MTBD re-bonds to the microtubules, and the connector performs a force-generating power stroke by straightening back to its original configuration, and pulls the cargo together with it (1, 24⇓ ⇓ –27). Finally, with the release of adenosine 5'-diphosphate (ADP), the mechanochemical cycle can restart.

The initial dynein stepping experiment with a single fluorescent probe showed that dynein, unlike kinesin, takes lateral and backward steps (15). In addition, dynein has been shown to use a variable step size compared to a kinesin that only uses an 8 nm step size (15, 28). Two-color single-molecule experiments have shown that the two AAA rings of dynein move in an uncoordinated manner, so that one AAA ring can sometimes take multiple steps, while the other AAA ring does not have any steps (29, 30). In addition to the well-known kinesin and myosin push hands (28, 31), dynein can also move in an inch worm manner, where the leading AAA ring can move forward while the trailing AAA ring does not move, or The AAA ring at the end can step forward without passing through the leading AAA ring (29, 30). In addition, an active motor domain and an additional microtubule anchor are sufficient to achieve progressive and directional movement (32).

The previous single molecule experiment followed the AAA loop, but not the MTBD that actually stepped along the microtubule trajectory. Since the AAA ring and the MTBD are separated by a ~15-nm stem, they can take different angles relative to the MTBD (2, 33, 34), and the position and step of the AAA ring may not reflect the position and step of the MTBD. Therefore, To accurately understand the dynein stepping, it is important to directly measure the position of the MTBD relative to the microtubule trajectory and measure the position of the MTBD relative to the AAA ring.

Here, we have developed a three-color single-molecule microscope detection that can track the movement of an AAA ring and two MTBDs at the same time. In addition to extending the existing nanometer-precision distance measurement and image registration, we also use 6-nanometer small fluorescent probes (DNA FluoroCubes), which have a light stability 50 times higher than that of organic dyes (35), allowing the measurement of many dynein Step without photobleaching. Using these technological advances, we show that AAA loops and MTBD sometimes step at different times and take different size steps, which results in multiple conformations that dynein can adopt when walking along the track. The transition probability between conformations derived from our data is sufficient to generalize the directional dynein motion using Monte Carlo simulation. In summary, we conclude that dynein can adopt a variety of conformations, including some previously undescribed conformations, and the AAA loop and MTBD exhibit different stepping behaviors.

To determine how the AAA ring and MTBD move relative to each other when the dynein stepped along the microtubules, we tracked the stepping of a three-color labeled dynein, where one AAA ring and two MTBDs use three different colors of fluorescence Group mark (Figure 2). 1A). In order to accurately track all three colors with a resolution of 1 nm, we extended the previously developed two-color image registration program (36) to three colors (SI appendix, Figure S1 AD). To verify this method, we imaged three different colored dyes placed on the DNA origami nanometer ruler (37, 38) at a clearly defined distance, and found that the expected distance between the three dyes was restored with an accuracy of 1 nm (SI appendix, Fig. S1 EG).

Three-color stepping traces of dynein. (A) Schematic diagram of the design of three-color dynein. Each of the two motor domains of dynein is individually labeled and dimerized using reverse complementary single-stranded DNA [black, connected by SNAP tags (39, 41)]. The MTBD of each motion domain and one of the two AAA rings are marked with FluoroCubes (35). For a sports field, the AAA ring is marked with six colors ATTO 488 FluoroCube [green, connected via HALO tag (40)], and the MTBD (called related MTBD) is marked with six colors ATTO 647N FluoroCube [red, via YBBR-tag (41) connect]. For other motion domains, only MTBD is marked with six dye Cy3N FluoroCube [blue, connected by YBBR label (41)] (referred to as the opposite of MTBD). For more detailed information on construction design and labels, see Materials and Methods. ssDNA, single-stranded DNA. (B) The structure of yeast cytoplasmic dynein MTBD (gold) bound to tubulin (gray) [Protein Database ID code 6KIQ (56)]. The exact position where the YBBR label is inserted is shown as a red sphere. (C) The workflow for collecting step data of three-color dynein. The top photomicrograph shows a combination of all three colors, while the bottom row shows each color separately. For more detailed information on data collection, see Materials and Methods. (Scale bar, 500 nm.) (D) Raw step data, three-color dynein heterodimers (color dots) and detected steps (color lines) whose position along the axis changes with time. (Illustration) A magnified view of the area in the black and purple boxes. All other stepping traces are stored on Zenodo at https://doi.org/10.5281/zenodo.4321962 (54). (E) Step tracking in the xy space of the enlarged view in the black box of D. The original data is shown as black dots, and the fitting position is shown as a circle of the corresponding color for each field. The bottom left corner shows the superposition of the fitting step positions of all three domains.

To create a three-color labeled dynein dimer in which an AAA ring and two MTBDs are fluorescently labeled, we used the well-studied truncated yeast cytoplasmic dynein (15) and added an N-terminal SNAP Label (39), a C-end HALO label (40) and internal YBBR label (41). In this design, the HALO label is located on the top of the AAA ring, and the YBBR label is located in the flexible ring of the MTBD (Figure 1B), allowing us to mark the AAA ring and the MTBD sports field on the same ring. For a monomer motion domain, we use the six dye ATTO 488 FluoroCube (35) to label the HALO label, and the six dye ATTO 674N FluoroCube to label the YBBR labeled MTBD. For the other monomer motion domain, we only used the six dye Cy3N FluoroCube to label the YBBR-labeled MTBD (Figure 1A). In order to link the two labeled monomers into a dimer, we respectively labeled the N-terminal SNAP tag on the motor domain of the monomer with reverse-complementary single-stranded DNA. When bound, the hybridization of the single-stranded DNA produces the dimer motor as previously described (29). This three-color FluoroCube-labeled dynein has the speed and continuous synthesis ability similar to the green fluorescent protein (GFP)-labeled wild-type dynein (SI appendix, Figure S2 AC and movie S1), indicating our modification, including labeling and FluoroCube labeling does not interfere with the function of dynein.

When we compare the three-color dynein labeled with conventional organic dyes with the dynein labeled with FluoroCubes, we found that 4% of the conventional dye-labeled dynein has signals in all three channels after 50 frames, and 75% of FluoroCube-labeled dynein emits signals in all three channels (SI appendix, Figure S2 F and G). In addition, compared with traditional dyes, FluoroCube-labeled dynein produces more precise positioning under the same exposure time of 110 ms (Cy3N FluoroCube is 2.4 nm, while a single conventional dye of the same color is 7.2 nm; SI appendix, Figure S2 D and E). In short, if FluoroCubes are not used, it is not feasible to track all three domains at the same time with high resolution.

We tracked the stepping of the three-color labeled dynein along the microtubule at low ATP concentration (3 μM), and resolved a single step of all three domains with high spatiotemporal resolution (Figure 1 CE and Movie S2). For all the detected steps analyzed in this study, we found that the SD of the green-labeled AAA ring is 3.7 nm, the blue-labeled MTBD is 3.7 nm, and the red-labeled MTBD is 3.5 nm (SI Appendix, Supplementary Note 1). In order to achieve a fast acquisition of 330 ms with the smallest dead time, we optimized the acquisition sequence (see Materials and Methods and SI Appendix, Figure S3). Based on the residence time distribution (SI Appendix, Figure S4B), we estimate that the acquisition sequence allows us to detect more than 95% of all steps of MTBD and approximately 81% of all steps of the green-marked AAA ring (SI Appendix, Supplementary Note 1 ). Together, we resolved 4,500 steps from 54 dynein molecules moving along microtubules (Figure 1D and E).

Using this data set, we found that compared with previous studies (29, 30), the average step size and forward step percentage of the AAA ring are similar (Figure 2 A and B and SI appendix, Figure S4). In addition, we were able to analyze the stepping behavior of MTBD (Figure 2B). We observed that MTBDs tend not to pass each other, leading to inch worm stepping behavior (SI appendix, Figure S4), as previously described for the two-color marked AAA ring (29, 30). Also consistent with previous studies on the AAA ring (29, 30), we found that as the distance between the MTBDs along the axis increases, compared with the leading MTBD, the trailing MTBD is more likely to take the next step (for example, on the MTBD axis). The distance between the left and right MTBDs is 16 to 24 nm, and the next step is to tail the MTBD in 62% of the cases), but we did not observe this deviation between the left and right MTBDs as the off-axis distance increases (SI Appendix, Figure S4 H and I).

Two-dimensional analysis of AAA ring and MTBD stepping. (A) The microtube lattice with positive and negative ends (grey circles) and the definition of forward and backward and two-dimensional, on-axis and off-axis steps. (B) Dynein AAA ring (green), MTBD (blue) on the opposite motion domain, and MTBD (red) on the same motion domain (related MTBD) on-axis step length histogram. Other fitting parameters are shown in Table S1 in the SI Appendix. (C) Cumulative frequency plot of the on-axis step length of the dynein AAA ring (green), the opposite MTBD (blue), and the related MTBD (red). The P-value is a relative comparison between all three distributions and is calculated using Kolmogorov-Smirnov statistics. (DF) (Left) (D) Dynein AAA ring (green), (E) opposite MTBD (blue) and (F) related MTBD (red) two-dimensional stepped microtubules in xy space Lattice. (Right) Focus on the on-axis step size distribution of a single strand (purple box; no off-axis step size). The Gaussian fit step histogram shows multiple main peaks. The peak position is displayed below the corresponding histogram. The microtubule lattice is based on flat 13 filament microtubules. Use flat microtubes, because our data only reports steps in the xy (horizontal) plane, not in the z (vertical) direction. Here, each parallelogram represents a tubulin dimer, consisting of one copy of α- and β-tubulin. The yellow parallelogram represents the tubulin dimer where the domain was located before this step.

However, we also found that there is a significant quantitative difference between the steps of the MTBD and AAA loops, and the step size distributions of the two MTBDs are the same (Figure 2C). For example, compared with MTBD (the combination of two MTBDs is 18.8 nm), the AAA ring takes a slightly larger forward step (22.2 nm) on average. In addition, compared to MTBD (blue and red marked MTBD at 21% and 19%, respectively), the AAA ring takes fewer steps back (14%). In addition, focusing on the steps along a single strand without any off-axis stepping component (Figure 2 DF), we detected that the steps of MTBD have a periodicity of ~8 nm, but not all steps of the AAA ring. When we image the three domains with a higher time resolution (110 ms acquisition time instead of 330 ms), the difference in the step size distribution of the MTBD and AAA loops is also true, in order to detect nearly 100% of all Step size (SI appendix, Figure S3). In addition, we also observed a difference in the step length distribution between the AAA ring and the MTBD of dynein moving along the axon filament (SI appendix, Figure S3). Therefore, these results indicate that the two MTBDs exhibit the same motion, while the AAA ring and the MTBD do not, indicating a more complex motion rather than a simple rigid body translation of the entire motion domain.

In view of the difference between the on-axis step length and the forward and backward step length percentages of the AAA ring and the MTBD, we next checked the step time of these domains (Figure 3A). When the MTBD takes a small step (4 to 12 nm, centered on the tubulin subunit [8 nm]), the AAA ring on the same motion domain only shows an obvious synchronization step about 60% of the time. Therefore, not every step of MTBD will result in the repositioning of the AAA ring on the same motor domain. However, when the MTBD stepped a distance greater than 20 nm (corresponding to a distance of approximately three or more tubulin subunits), the probability of simultaneous stepping of the AAA loop increased to >90% (Figure 3B). In general, these results can be explained by the flexibility between these areas. When MTBD takes a small step, the stem can adjust its angle relative to the AAA ring, resulting in a small axial displacement of the AAA ring. However, the longer MTBD step can only be accommodated by the displacement of the AAA ring along the microtubule axis.

Independent stepping of AAA ring and MTBD on the same motor domain. (A) The AAA ring and MTBD on the same motor domain can be stepped at the same time (top: the two domains move along the axis) or at different steps (bottom: only the MTBD moves, while the AAA ring remains on the same axis). (B) The histogram shows the frequency of simultaneous steps of the AAA ring and the associated MTBD (red) as a function of the step length on the MTBD axis. N refers to the total number of steps for each condition. Error bars show bootstrap SEM. The P value (gray) is calculated using a two-tailed z test. (C) The correlation between the on-axis step length of the AAA ring and the related MTBD (red) when stepping simultaneously. Each point represents a step. The red line shows the linear fit. N is the sample size. m is the slope. b is the y-axis intercept. r is the Pearson correlation coefficient. (D) An example of the on-axis trajectory of each of the four quadrants (I, II, III, and IV) defined in C, with an illustration in which the opacity of the initial position display is reduced. The blue asterisk indicates the time when the AAA ring and the MTBD are moving at the same time. All traces are raw step data, the position of the three-color dynein heterodimer (color dot) along the axis with the detected step (color line) and time. The opaque line shows the SD along the axis at each step. Please note that the blue channel has been removed for clarity.

We also checked the frequency of simultaneous stepping of the MTBD on one motor domain and the AAA ring on the other motor domain, and found that, as expected, the MTBD stepping of one motor domain does not necessarily cause axial displacement on the other motor domain. AAA ring (SI appendix, Figure S5 AC).

Next, we will focus our analysis on the direction and size of the synchronization steps of the AAA ring and MTBD in the same motion domain (Figure 3B). We found that the relative step length and direction are not always the same (Figure 3 C and D). Although in most cases, both domains move forward (Figure 3 C and D, quadrant II), we observe the domain moving in the opposite direction; for example, MTBD moves a small step backward, while AAA The ring moves slightly forward (Figure 3 C and D, quadrant IV) and vice versa (Figure 3 C and D, quadrant I). The least likely event is that the MTBD moves forward and the AAA ring moves backward (Figure 3 C and D, quadrant I). In general, the correlation between the on-axis stepping of the AAA ring and the MTBD in the same motion domain shows that when the two stepping at the same time, the AAA ring takes larger forward steps and fewer back steps than MTBD on average (Figure 3C). In short, the AAA ring and the MTBD do not necessarily travel along the track at the same time and the same distance, which further supports the idea that they are not rigid connectors.

In order to study how the AAA ring and the MTBD move relative to each other, we checked the relative position between the green-marked AAA ring and the red-marked MTBD in the same motion domain. On average, the MTBD is located in front of the AAA ring (closer to the negative end of the microtubule) (SI appendix, Figure S5 DL). This finding is consistent with static electron microscopy data of dynein binding to microtubules (2, 34). However, there is no preferential position between the two MTBD or MTBD and AAA rings that are off-axis along the microtubule (SI appendix, Figure S5 DL).

Next, we calculated the angle between the stem and the microtubule based on the relative position of the AAA ring and the MTBD. To calculate the stem-microtubule angle ω, we use the on-axis distance between the AAA ring and the MTBD of the same motion domain, and assume that the length of the dynein stem is fixed (Figure 4A). We found that an average of 80.5° wide angle distribution (Figure 4B) further supports the idea of ​​high flexibility in the dynein motion domain.

The relative movement of AAA ring and MTBD in the same motion domain. (A) The schematic diagram shows the definition of the angle ω between the stem and the microtubule on the axis. Please note that the angle is only calculated for the motor domain marked with both AAA ring (green) and MTBD (red). To calculate the angle ω, we use the measured on-axis distance between the AAA ring and the MTBD (purple line) and the fixed known distance from the MTBD to the center of the AAA ring (black line). (B) Histogram of stem microtubule angle ω. (C) A diagram showing the definition of leading and trailing MTBD. (Top) Angle measurement of double-marked motor domain leading (red MTBD leading). (Bottom) Angle measurement of trailing double-marked motor field (blue MTBD leading). (D) If the dual-labeled motor domain is leading (red) or lagging (blue), the histogram of the stem-microtubule angle ω. The P value is calculated using a two-tailed t test. (E) The correlation between the stem-microtubule angle ω and the wheelbase between MTBD at the single-molecule level (gray dot). Here, a positive value refers to a state where the dual-standard motor domain is leading, and a negative value refers to a state where the dual-standard motor domain is behind. The purple line represents linear regression. N is the sample size. m is the slope. b is the y-axis intercept. r is the Pearson correlation coefficient. (F) Same data as in E, but divided into 8 nm bins (the size of a tubulin dimer). Error bars show SEM. The P value is calculated using a two-tailed t test. In B and D sample sizes (N), the average distance (μ) and its SD (σ) are given.

Next, we asked whether the stem-microtubule angle ω of the leading and trailing motor domains are different (Figure 4C). Comparing the average angles of the two cases, we find that the angle ω of the leading motor domain (red MTBD leading; 71.8°) is significantly smaller than that of the rear motor domain (red MTBD trailing; 90.0°) (Figure 4D). We also calculated the stem-microtubule angle ω as a function of the distance between the MTBD axes, and found that the angle of the rear motor domain increases as the distance between the two MTBDs increases, while the angle of the front motor domain decreases (Fig. 4 E and F ). In summary, these data reveal the flexibility of the AAA ring and the MTBD in the same motion field, and compared with the trailing motion field, the leading motion field usually adopts a sharper stem-microtubule angle.

Three-color imaging allows us to determine the relative positions of the AAA ring and the two MTBDs along the microtubule axis. Arranging all the three possible color orders results in a total of six different domain orders, each of which can be associated with the four potential conformational states of dynein (Figure 5A and SI appendix, Figure S6 A and B). In order to determine the relative frequency of the domain order, we quantified the frequency of each of the six domain orders during all step tracking. We found that dynein can adopt all six domain sequences to varying degrees (Figure 5B). The two most common domain orders are that both MTBDs are ahead of the green marked AAA ring (51% in total), followed by two domain orders, where the AAA ring is between the two MTBDs (32% in total), and the second is The order of the two domains leading the AAA ring (17% in total). Interestingly, according to the previous study (1), we will not predict the order of the two domains leading the green-marked AAA ring. In summary, observations of the sequence of multiple domains indicate that dynein can adopt multiple conformational states during exercise (SI Appendix, Figure S6 A and B).

The domain sequence frequency of the AAA ring and the two MTBDs. (A) Schematic diagram of the sequence of all six possible domains along the axis of the microtubule. (Top) The small gray circle shows tubulin, while the larger green, blue, and red circles represent the AAA ring, the opposite MTBD, and the related MTBD, respectively. For example, to be classified as the leftmost domain order, the opposite MTBD (blue) must be closest to the negative end of the microtubule, then the related MTBD (red), then the AAA ring (green). Note that the absolute distance between domains does not matter. (Bottom) One of four possible dynein conformations based on the order of the three color gamuts. Other possible conformations are shown in the SI appendix, Figure S6. (B) A histogram of the occurrence of each of the six possible domain sequences with a sample size of N. Error bars show bootstrap SEM. (C) The probability of preserving the domain order after one (orange) or two (blue) steps. Here, the step size refers to the movement of at least one of the three domains. If the transition is random, the orange and blue dashed lines indicate the probability of preserving the domain order after one (orange) or two (blue) steps. The sample size N refers to the total number of all domain step transitions that occur after the motor has taken one step from its current domain. Error bars show bootstrap SEM. A more detailed analysis of the conversion between domain sequences is given in Figure S5 of the SI Appendix.

We next asked how often dynein switches to the new domain sequence after the step has occurred. If the conformational state is random, we would expect 1/6 (∼17%) to remain in the initial domain sequence after one step, and only 1/36 (< 3%) in the first and second steps. However, we found that dynein tends to maintain the same domain order after a step (Figure 5C and SI appendix, Figure S5J), although some states are more durable than others. For example, the order of the two fields leading by the AAA ring marked in green is the most unstable, and it is more likely to transition to the order of other fields leading by MTBD. The observation of the persistence of this domain sequence is consistent with previous observations, in which it has been reported that the AAA loops of dynein rarely communicate with each other (29). We also measured the frequency of steps in the same or another domain after any step in the labeled domain. The most common result is that one MTBD moves one after another (SI Appendix, Figure S5K), which is again consistent with the observation that the two dynein AAA loops tend to step in an alternating manner (29, 30). In summary, the AAA loop of dynein and the two MTBDs often move in an alternating manner and are unlikely to pass each other, resulting in a given domain sequence persisting in multiple steps.

The observation of the sequence of the six three-color gamuts shows that when moving along the microtubules, the dynein can adopt a variety of conformations. However, we lack information about the location of the second AAA ring. Therefore, we turned to Monte Carlo simulations to gain more insights into the conformation of dynein during exercise and reveal the minimum characteristics required to describe the movement of dynein. Using our experimental data as input, we simulated the steps of two AAA loops and two MTBDs along the microtubule by assigning probabilities to the step size, step direction, and the probability of what steps will follow. (Figure 6A and Movie S3), as described in Materials and Methods. We also applied some rules based on our tricolor dynein (movie S4-S9) data. Specifically, 1) the distance-related deviation on the axis takes more forward rather than backward steps (Figure 2), and 2) the distance-related deviation to close the difference between the on-axis and off-axis motor domains when taken. There is a gap between one step (SI Appendix, Figure S4), 3) The trailing domain is more likely to take the next step instead of the leading domain (SI Appendix, Figure S4), and 4) A bias towards alternate stepping behavior (SI Appendix), Figure S4) and 5) Relative movement (change in angle ω) between AAA ring and MTBD (Figure 4). However, we did not force a specific distance between the two motor domains by setting the cut-off values ​​for the on-axis and off-axis distance; in other words, the motion domain is not constrained by the connecting tether and is stepped independently according to the above rules (Fig. 6A).

Monte Carlo simulation of dynein movement. (A) Using three-color experimental data as input, Monte Carlo simulation can be used to simulate the positions of two MTBDs and two AAA rings. (B) Comparison of Monte Carlo simulation of dynein movement with previously published experimental data in which both AAA loops are tracked (29, 30). (Left) Example step trajectory from Monte Carlo simulation of two AAA rings. (Middle) Histogram of the two-dimensional (2D) distance between two AAA rings and the previously measured 2D distance between the rings (29, 30). (Right) Compared with the previously measured data, the pass and fail steps between the two AAA rings are based on simulation (29). (C) By running Monte Carlo simulation, the importance of different rules to dynein movement can be evaluated. Here, the influence of two rules on the 2D distance between two AAA rings is tested. The influence of relative movement between AAA ring and MTBD. Run simulations with variable stem-microtubule angles (left, gray) or fixed stem-microtubule angles (left, orange). The distance-dependent offset has an effect on closing the gap between the motor domains along the off-axis. Run a simulation with deviations close to each other (right, gray) or without deviations close to each other (right, blue). Please note that the distance distribution of the flexible angle comes from the same data as in B. The effects of other rules on dynein movement are shown in the SI appendix, Figure S8. (B and C) 100 simulations were performed for each condition with more than 10,000 steps. More details about Monte Carlo simulation can be found in the Materials and Methods and the SI appendix, Figures S7 and S8, and movies S3–S10.

When we apply all these rules in the simulation process, we can reproduce the experimental data of dynein stepping very well; the step-length distribution of the simulated AAA loop is almost the same as that observed in the experimental data (Figure 2 and SI appendix) , Figure S7). Interestingly, some parameters not provided in the model are in good agreement with previous experimental observations. For example, we did not provide input about the AAA ring spacing, except for the step size distribution of the relevant MTBD determined through experiments and the relative motion between the AAA ring and the MTBD (change in angle ω) (Figure 6A). Nevertheless, our simulation The resulting distance between AAA rings (Figure 6B) is very similar to that observed in the stepping experiment reported earlier, in which both AAA rings are labeled (29, 30). In addition, when we compared the probability of passing and failing steps of the AAA ring with previous experimental data (29), we also found good agreement without directly coding this motion in the simulation (Figure 6B).

However, if we ignore any of the rules listed above during the Monte Carlo simulation, the simulated dynein motility does not match the current or previous experimental observations (SI appendix, Figure S8). For example, if we do not apply the tendency of the motor domains to approach each other along the off-axis (Figure 6C and SI appendix, Figure S8B), but allow the motor domains to move off-axis in either direction, in some simulations, the motor domain drifts >100 nm. In addition, if we fix the angle ω between the AAA ring and the MTBD, the distance between the AAA rings generated by the simulation (~27 nm) is greater than the experimental measurement (~18 nm) (29) (Figure 6C, SI appendix, Figure 5) Bigger. S6 D–L, and movies S3, S9 and S10). Therefore, encoding the rules and transition probability sets derived from the experiments listed above is sufficient and necessary to use Monte Carlo simulation to reproduce the directional dynein motion.

Since our three-color experimental data only captures the position of one AAA ring and two MTBDs, information about the position of the second AAA ring is missing. However, since our Monte Carlo simulation reproduces our data and other people's experimental data very well, we use this simulation to predict the position of the AAA ring and MTBD during exercise. As a general verification of this method, we compared the frequency of the experimental three-color gamut order (Figure 5) with the frequency of the three-color gamut order in the Monte Carlo simulation, and found a good agreement (SI Appendix , Figure S6C). Considering the four moving parts (two AAA loops and two MTBDs in the homodimer), dynein can adopt 12 potential conformations (SI appendix, Figure S6 A and B). Among the 12 possible conformations, our simulation predicts that the first three conformations account for 55% of the total, while the six least common conformations account for <20% (Figure 7 and SI appendix, Figure S6B). In general, the conformation where the stems do not cross (~76%) is more common than the conformation where the stems cross each other (~24%). Interestingly, the previous dynein motion model (1) did not predict the eight conformations where the stem-microtubule angle of at least one of the motion domains was> 90°, but according to our simulations, they accounted for about 65 of all dynein conformations. %.

Dynein conformational model. Monte Carlo simulation of dynein motion is used to determine the frequency of all 12 possible dynein conformational states (the position of the AAA loop and MTBD relative to the microtubules on the axis). Here, the appearances of all 12 states are arranged in order from the most common to the least common (from top left to bottom right). Note that the absolute distance between the four domains (the AAA ring and the two MTBDs) is irrelevant, only the relative proximity of all four domains to the negative end of the microtubule determines the classification of the conformational state. In addition, we only show the conformations that change along the axis, while ignoring the differences along the axis. We noticed that when we added experimental noise, we found that only four conformational states were sufficient to explain the experimentally observed states. Figures S6 and S9 of the SI appendix show a more detailed analysis of the influence of experimental errors on the conformational state distribution of dynein, and are discussed in more detail in the discussion and SI appendix supplementary note 1.

We next studied how experimental errors, such as positioning errors, FluoroCube rotation degrees of freedom, and steps lost due to time resolution limitations, might change the conclusions of our Monte Carlo simulations. Using average SD for step detection of about 3.5 nm and registration error of about 1 nm, we ran Monte Carlo simulation with SD of 5 nm to obtain the accuracy of each step. Using this method, we tested whether other models with fewer or other dynein conformational states can explain the experimental data. In general, we found that there are only four conformational states, plus experimental errors, can also explain the experimental data (SI Appendix, Figure S9 and Supplementary Note 1). However, without the stem-to-microtubule angle changes related to the MTBD distance observed in this study (Figure 4), we cannot summarize the experimental data (SI Appendix, Figure S9 D and E). In short, our experimental data combined with Monte Carlo simulation provides a model of the conformational distribution of dynein during exercise.

Previous dynein stepping experiments measured the position of the AAA loop (29, 30). Here, we have been able to combine an AAA ring to track the movement of two MTBDs. In order to make this measurement, several technical challenges must be overcome. First, the small 14-kDa MTBD (22) must be labeled without interfering with motor function. For this reason, the common HALO-tag (40) and SNAP-tag (39) are not ideal because they are twice the MTBD. However, we found that a 14 amino acid long YBBR tag (41) can be inserted into loop 5 of MTBD and then labeled with DNA FluoroCube (35) without interfering with the wild-type function. Secondly, because of photobleaching, it is not easy to track three colors for a long time using traditional dyes, because even the photobleaching of one dye will terminate the measurement. However, using DNA FluoroCubes, which have 50 times higher photostability than traditional organic dyes, allows long-term tracking of many steps. Third, the distance between the three colors must be measured with nanometer precision. To this end, we expanded the previously released two-color data collection and imaging analysis pipeline (36) to three colors. When the motor goes through hundreds of steps along the microtubules, all the three technological advances mentioned above are crucial to accurately measure the position of the three domains of dynein. These measurements provide insights into the stepping behavior and conformational state of dynein, as described below.

Our three-color labeled dynein experimental data combined with Monte Carlo simulations indicate that dynein can adopt a variety of conformations, many of which were not considered in the previous dynein stepping model (1, 29, 30). These conformational states presented in Figure 7 are most likely achieved by the angle change of the stem spanning between the AAA ring and the MTBD. Previously, a wide range of stem angles have been measured by electron microscopes (2, 34) and polarized light microscopes (33). Polarized light microscopy studies also observed the articulation of the stem, which allowed the AAA ring to rotate while dynein moved along the microtubules (33). This observation is very consistent with our observations of flexible movement between the AAA ring and MTBD. However, the average angle between the stem and the microtubules was determined by Imai et al. by cryo-electron microscopy. (34) and Can et al. (2) They are ~42° and ~55°, respectively, and are smaller than the values ​​we measured for the leading (72°) and trailing (90°) motion domains. One explanation for this difference may be that our C-terminal fluorescent label on the AAA ring is not in the center, but on the side of the AAA ring, which is closer to the negative end, so it may bias the angle to a larger value. Using structural information, we estimated the approximately 3 nm offset along the axis from the center of the AAA ring to the position of the HALO label (C-terminal of dynein), and applied this correction to our data analysis pipeline. Considering this correction, we measured the stem-microtubule angle of the front motor domain to be 67.1° and the rear motor domain to 83.3°. Interestingly, we estimated a stem-microtubule angle of ~65° in the recent low-temperature electron tomography structure of Dynein (42), which is very consistent with our offset correction angle value. Although this potential shift will change the angle value, it will not change the general conclusion of the considerable relative motion between the AAA ring and the MTBD.

In addition to measuring the average angle, we can also obtain information on how the stem angle depends on the separation of the two MTBDs and other parameters that vary with the dynein step (Figure 4 and SI appendix, Figures S4, S5, and S10). For example, when the MTBDs are relatively close to each other (8 nm apart, the spacing between tubulin dimers), we find that the stem-microtubule angles of the two motor domains are relatively similar (Figure 4 E and F). However, when the MTBDs are farther apart (>16 nm, two or more tubulin dimers), the motor domains tilt toward each other, resulting in a split-like conformation in which the AAA loops are closer than the MTBDs (see Figure 1). 7. The leftmost and rightmost state). In another study that measured the rotation of the AAA ring, the tilt of the two motion fields was also observed, and it was found that a larger step of the motion field has a greater rotation (33). Since we are able to track a single dynein in many steps, we can also observe that the stem-microtubule angle changes as the dynein moves along its orbit (see the dynein stepping model in the SI appendix, Figure S10). For example, when the trailing motor domain passes through the front motor domain, its angle usually changes from a steep angle to a shallower angle (similar to pivotal movement) (Figure 3 C and D). In summary, although other studies have focused on the distribution of static dynein and observed the flexibility in the dynein motion domain, this study has observed the flexibility of actively moving dynein to gain insight into movement.

By tracking an AAA loop and two MTBDs, we can obtain more information about the conformational state of dynein during exercise, rather than the information inferred from previous studies of measuring AAA loops (29, 30, 32). By combining our experimental data with Monte Carlo simulations, we estimated the frequency of 12 possible conformational relationships between two AAA loops and two MTBDs. Overall, these data indicate that the state where the MTBD is farther than the AAA ring is more common than the reverse. In addition, we found that for the leading motor domain, MTBD usually leads the AAA ring, and the AAA ring leading or MTBD leading the trailing motor domain has almost the same distribution. In addition, our data shows that the stems of dynein homodimers rarely cross; in other words, if the MTBD of motor domain 1 leads the MTBD of motor domain 2, then it is likely that the AAA loop of motor domain 1 also leads the motor domain. 2 AAA rings. Interestingly, we found that dynein changes its conformational state only after one step occurs in about 50% of the cases (Figure 4), and if dynein switches between the two states, it is likely to pass through a pivot-like Conversion (AAA ring and MTBD switch to lead in the motor domain) (Figures 5 and 7, the two states in the upper left corner). We can also prove through simulation that the ability to adopt these conformational states requires flexibility in the angle between the stem and the microtubule; if we fix the angle between the stem and the microtubule, as shown in some dynein stepping models (1) , Dynein can only adopt 2 of the 12 possible conformations (SI appendix, Figure S6 and movies S3, S9 and S10). Although the flexible movement between the AAA ring and the MTBD is essential for the interpretation of experimental data, even considering the limitations of temporal and spatial resolution (Figure 7 and SI appendix, Figure S9), some less common conformational states may be based on our The experimental data prediction can be explained by experimental noise. Our current work shows that at least four conformational states of dynein are required to interpret the experimental data (SI Appendix, Figure S9 and Supplementary Note 1). However, since we use the upper limit of experimental noise when determining the minimum number of conformational states, if we use a lower estimate of the experimental error (for example, the SD is 3 nm instead of 5 nm). Therefore, dynein may adopt 4 to 12 conformational states. Accurate confirmation of these states will require future research with higher spatial resolution and four-color markings so that two AAA rings and two MTBDs can be tracked simultaneously. Nevertheless, the general conclusion of the dynein conformational state discussed above, that MTBD is farther than the AAA loop, remains unaffected. In summary, our Monte Carlo simulations combining three-color imaging and dynein motion provide insights into the frequency and the conversion between dynein conformational states, and show that the high flexibility within the dynein motion domain is essential for achieving these conformations. important.

The stepping of MTBD is triggered by the binding of ATP and the rotation of the AAA ring, which is caused by the bending of the connector, the mechanical element of dynein (1, 23, 25⇓ –27). MTBD is then thought to perform a Brown search and then re-bind to the new tubulin subunit. By tracking the MTBD and AAA loops at the same time, we found that this search for the conversion of the entire motion domain can produce many different results. In some cases, the MTBD can move (forward or backward) without the AAA ring translation, and even if the two domains are stepped at the same time, they can also take steps of different sizes (Figures 2 and 3).

These results are most consistent with the flexibility in the dynein motion domain, rather than rigid body motion in the motion domain. For example, the relative movement between the AAA ring and the MTBD will allow the MTBD to take a small step backward or forward, while the AAA ring rotates and does not translate significantly along the long axis (Figure 3A). However, if the MTBD takes a larger step size after the Brown search, the AAA ring on the same motor domain will be forced to follow, because the step size cannot be adapted only by the angle change between the AAA ring and the MTBD. However, because our collection method may not be able to detect some AAA loop steps (SI Appendix, Supplementary Note 1), the percentage of independent MTBD steps without AAA loop translocation may be lower than what we reported in this study (Figure 1). 3B). This may also explain why we measured similar behaviors for the simultaneous steps of the AAA ring and related and opposite MTBD (Figure 3B and SI appendix, Figure S5B). Nevertheless, we have shown that the relative movement between the AAA ring and the MTBD is an important part of dynein movement (Figure 6 and SI appendix, Figure S6 and S9 and Supplementary Note 1). This relative movement can also explain why we cannot detect the steps of the 8-nanometer cycle of the AAA ring, while we detect the steps after the 8-nanometer cycle of the tubulin of MTBD (Figure 2). Therefore, our data indicates that the AAA loop is essential for initiating and driving the steps, and the Brownian search of MTBD and the flexibility of the stem determine which parts of the step and motor domain are translocated (only MTBD or MTBD and AAA loops). In future studies, it will be interesting to observe how the relative motion of the AAA ring and MTBD is related to the mechanochemical cycle model of dynein (43). One possibility is that ATP binding (joint bending) and phosphate release (joint straightening) drive the tilt of the AAA ring relative to the MTBD.

The sometimes different step size and large number of conformational states of the AAA loop and MTBD are quite different from other motor proteins (such as kinesins), which use a regular 16-nanometer motor domain and are almost completely forward steps (28, 44). The inherent flexibility of dynein and the ability to intervene in many different ways can explain why a single dynein is more effective than a single kinesin in bypassing obstacles, such as microtubule-associated proteins (45, 46).

Foerster resonance energy transfer (FRET) (47) has been extensively studied for distance measurement between protein domains. However, FRET is usually limited to short distances of 2 to 8 nm. In contrast, the multicolor measurement described here is not limited to any distance, so it can be applied to macromolecular complexes of any size to obtain a direct distance relationship. Together with DNA FluoroCubes (35), a mechanism for long-term tracking of kinetics is provided. The method described in this work can be used to study the conformational changes of other multi-domain proteins or macromolecular complexes. For example, the multicolor method can be used to study the conformational changes of the molecular chaperones during the refolding of the customer's substrate. We also envision that our method can be applied to molecular machines that run on orbits other than microtubules (such as DNA). For example, high-resolution multicolor methods can be used to study how chromatin remodeling agents interact with nucleosomes along DNA. Finally, we expect that the framework provided in this work can be extended to four colors, which will further expand the reference point for studying intra-protein and inter-protein dynamics.

Assemble the flow cell as described in (36). In short, we use a laser cutting machine to cut a custom three-cell flow chamber from a double-sided adhesive sheet (Soles2dance 9474-08x12, 9474LE 300LSE; 3M). Then, we used these three-cell flow chambers with glass slides (12-550-123; Thermo Fisher Scientific) and 170-micron thick coverslips (474030-9000-000; Zeiss) to assemble flow cells. Before assembly, the coverslips were washed in a 5% vol/vol solution of Hellmanex III (Z805939-1EA; Sigma) at 50°C overnight, and then thoroughly washed with Milli-Q water.

FluoroCubes are assembled as described in the references. 35 and SI appendix, supplementary information methods.

We use recombinant Saccharomyces cerevisiae cytoplasmic dynein (Dyn1) (amino acids 1219 to 4093) truncated at the N-terminus as the monomeric version expressed in yeast strains with the following genotypes: MATa his3-11,5 ura3-1 leu2- 3,112 ade2 1 trp-1 PEP4::HIS5 pGAL-ZZ-TEV-SNAPf-3XHA-D6-DYN1(MTBDL5:YbbR)-gsDHA is used in all our stepping experiments (VY1067 (36)). This construct has N-terminal SNAP tags (39, 41), C-terminal Halo tags (40), and YBBR tags (41). It is inserted into loop 5 of MTBD, with three glycines on each side (inserted between T3173 and L3174) Is GGG-TVLDSLEFIASKLA-GGG). In addition, we use VY208 (MATa his3-11,5 ura3-1 leu2-3,112 ade2-1 trp-1 PEP4::HIS5 pGAL-ZZ-TEV-sfGFP-3XHA-D6-DYN1-gsDHA) construct as the wild-we The speed and processing power analysis type control. The expression of dynein (VY208 or VY1067) and yeast lysis were performed as previously described (15, 36). The purification and labeling of dynein are described in detail in the SI appendix, supplementary information methods.

The tubulin used in this work was purified and polymerized as described in the references. 48. In short, we used unlabeled tubulin and biotinylated tubulin, which are in BRB80 (80 mM Pipes [pH 6.8], 1 mM EGTA, and 1 mM MgCl2) at approximately 20: Mix in the ratio of 1. To start the polymerization reaction, guanosine 5'-triphosphate was added to 1 mM, and the solution was incubated in a 37°C water bath for 15 minutes. Then, 20 µM paclitaxel (T1912; Sigma) was added, and the mixture was incubated for another 2 hours at 37 °C. At the beginning of each experiment, the microtubules were spun on a 25% sucrose cushion in BRB80 at ~160,000 × g for 10 minutes to remove unpolymerized tubulin and small filaments.

The flow cell for single molecule determination was prepared as previously described (49). In order to conduct all the experiments described in this study, we prepared four slightly different types of environments: 1) For most experiments, we use biotinylated microtubules as trajectories and low ATP concentration (3 μM), 2) for In one experiment, we used axons as a trajectory and low ATP concentration (3 μM) (SI appendix, Figure S3), 3) For another experiment, we used biotinylated microtubules as a trajectory and added 1 mM ADP to allow power Proteins bind tightly to microtubules (SI appendix, Figure S2)) and 4) For an experiment, we used biotinylated microtubules as trajectories and high ATP concentration (1 mM) (SI appendix, Figure S2). The preparation of the flow cell with dynein was modified from the reference. 35 and 36 are described in detail in the SI appendix, supplementary information methods.

To register the three channels, we used TetraSpeck beads (T7279; Thermo Fisher Scientific) based on the previously described two-color image registration protocol (36). To this end, we prepared one of the three flow cells of the flow cell with dynein (or DNA origami nanometer; SI appendix, Figure S1), and prepared another flow cell using TetraSpeck on the same flow cell. By adding 10 µL 1 mg/mL poly-d-lysine (P6407; Sigma) in Milli-Q water to the flow cell, incubating for 3 minutes and using 20 µL BRB80 (80 mM tubing [pH 6.8], 1 mM EGTA and 1 mM MgCl2). After that, we added 10 µL of 1:300 TetraSpeck beads to BRB80 and incubated for 5 minutes. Finally, clean the flow cell with 40 µL BRB80.

We designed and assembled DNA origami nanorulers according to the previously described protocol (50). The assembly, purification and flow cell preparation of DNA origami nanometers are described in detail in the SI appendix, supplementary information methods.

The microscope settings are based on the settings described in the reference. 36 It is also described in detail in the SI Appendix, Supplementary Information Methods.

In order to collect the data of dynein stepping, we prepared a chamber with TetraSpeck beads and another chamber with tricolor dynein on the same microscope slide to move along the microtubules or axons. Each data collection cycle begins by imaging a 20 × 20 grid of TetraSpeck beads. After that, we moved to the chamber with tricolor dynein and obtained six 500-frame movies with an exposure time of 110 milliseconds (see also SI appendix, Figure S3). After collecting the dynein movie, we moved back to the TetraSpeck bead chamber to collect another 20 × 20 grid, which was used as a control to test whether there were any changes in the image registration during the acquisition process (see SI appendix, Figure S1). We only accept data sets with σreg <1 nm. A more detailed description of TIRF data collection is given in the SI appendix, Supplementary Information Methods.

The data of three-color-labeled dynein and GFP-labeled wild-type dynein (SI appendix, Figure S2) were obtained using μManager (51) 2.0. Subsequently, the data was analyzed in ImageJ (52) by generating kymographs and measuring the displacement as a function of time.

Use μManager (51) 2.0 to obtain the data of tricolor dynein labeled with FluoroCubes (35) or traditional single dyes (SI appendix, Figure S2). Then, use the Spot Intensity Analysis plug-in in ImageJ (52) (https://imagej.net/plugins/spot-intensity-analysis) to locate a single molecule, set as follows: the time interval is 3.1 seconds, and the electron per ADU is 1.84 , The point radius is 3, the noise margin of FluoroCube data is 100, the noise margin of conventional dye data is 50, and the median background estimate. The number of frames to be checked is set to 20 for FluoroCube data and 10 for conventional dye data. After that, use the custom Python script described earlier to plot the data (35).

For three-color dynein step-by-step analysis, use the μManager (51) "Positioning Microscope" plug-in (SI appendix, Table S2) to install and position the dynein and TetraSpeck bead emitters. After locating all the probes, we registered the three channels (36) using the same affine-based method described previously for the two colors (SI appendix, Figure S2). Then, use the μManagers (51) "Localization Microscopy" plug-in to extract the trajectory of a single motor. Then apply a custom MATLAB (MATLAB R2019b) script to identify each step, and further analyze the data in the custom Python script. A detailed description of the data analysis can be found in the SI appendix, Supplementary Information Methods.

The image registration and distance measurement between the various dyes on the DNA Origami Nanoruler (SI Appendix, Figure S1) were performed as described previously (36). Since this is a three-color data set and not a two-color data set, we performed distance measurements on a single point pair (for example, Cy3 and ATTO 488 or Cy3 and ATTO 647N or ATTO 488 and ATTO 647N). In order to locate a single point (SI appendix, Table S2) and extract the points of the nanoruler containing all three labels, we used the μManager (51) "Positioning Microscope" plugin. To this end, we set the minimum number of frames to 18, the maximum number of lost frames to 2, the maximum distance between frames is 15 nm, the total minimum distance of the whole track is 0 nm, and the maximum distance between each dye is paired to 90 nm.

For negative staining electron microscopy, the agarose gel purified nanometer ruler was incubated on a freshly luminous carbon-coated 400-mesh copper grid for 1 minute. After that, the sample was blotted dry and immediately stained with 0.75% uranyl formate solution, and blotted dry without incubation. This staining was repeated four times, followed by the last incubation, and the staining agent was incubated for 45 seconds before blotting. The sample is air-dried before imaging. The data was collected at the University of California, San Francisco using a 4,000 × 4,000 charge-coupled device camera (UltraScan 4000; Gatan) operating under a Tecnai T12 microscope at 120 kV.

For Monte Carlo simulation, we use our experimental data as input. We defined a starting condition (the positions of the two MTBDs), followed by a continuous step-by-step simulation cycle, which ends once the dynein reaches the negative end of the microtubule lattice. A detailed description of the step loop and the entire Monte Carlo simulation conditions can be found in the custom Python script (https://doi.org/10.5281/zenodo.4321958) (53) and the SI appendix, supplementary information methods.

We discussed the inherent uncertainty due to random or systematic errors of each result and its verification in the relevant part of the article. In addition, we have included detailed information about sample size, number of independent calculations, and error bar calculations in the figure or in the corresponding figure title.

The original data set of three-color dynein stepping used in this study is hosted on Zenodo at https://doi.org/10.5281/zenodo.4321962 (54). All input data sets for the Monte Carlo simulation of dynein stepping are hosted on Zenodo at https://doi.org/10.5281/zenodo.4321958 (53). The μManager acquisition and analysis software is partly available under the Berkeley Software Distribution (BSD) license, and partly available under the GNU Lesser General Public License (LGPL). The development is hosted on GitHub at https://github.com/nicost/micro- manager. The latest version of MacOS and Windows can be downloaded here: https://micro-manager.org/Download_Micro-Manager_Latest_Release (version 2.0 gamma). Zenodo hosted a Beanshell script at https://doi.org/10.5281/zenodo.4321978 (55) to run the three-color acquisition of dynein stepping. The custom Python code for analyzing the dynein three-color stepping is hosted on Zenodo at https://doi.org/10.5281/zenodo.4321962. The Python code for the Monte Carlo simulation of dynein stepping is hosted on Zenodo at https://doi.org/10.5281/zenodo.4321958. All other codes can be obtained from the author.

We thank Christina Gladkova, Iris Grossman-Haham, and Chen Chen (University of California, San Francisco) for critical discussions of the manuscript. Andrew Carter (MRC Molecular Biology Laboratory) and Elizabeth Villa (University of California, San Diego) provided MATLAB scripts for the step-by-step detection of dynein. Part of this work forms the basis of SN's doctoral dissertation. The author is very grateful to NIH grants R01GM097312 and 1R35GM118106 (for RDV and SN) and HHMI (NS, NZ and RDV) for funding.

↵2 Current address: HHMI Janelia Research Campus, Ashburn, VA 20147.

Author contributions: SN, NS, and RDV design research; SN and NZ conducted research; SN, NS, and NZ contributed new reagents/analysis tools; SN and NS analyzed data; SN, NS, and RDV wrote this paper.

Reviewers: ZB, Stanford University; SLR-P., University of California, San Diego; and AY, University of California, Berkeley.

Competitive interest statement: RDV and SLR-P. He is the co-author of the 2018 article.

This article contains online support information at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2101391118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.