Cryo-EM of dynein microtubule-binding domains shows how an axonemal dynein distorts the microtubule

Thank you for submitting your article "Cryo-EM of dynein microtubule-binding domains shows how an axonemal dynein distorts the microtubule" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Reviewing Editor and Anna Akhmanova as the Senior Editor. The following individual involved in review of your submission have agreed to reveal his identity: Richard J McKenney (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

From the Reviewing Editor:

All three reviewers have easily reached a consensus on your work. We would like to accept it for publication after just minor changes are made. While a few experiments were mentioned in our reviews, we do not feel that any of them are required for publication. We ask only that you address all comments that can be addressed in the text. In particular toning down statements related to (1) the universality of the mechanism of microtubule binding because the axonemal dynein was linked to a cytoplasmic dynein stalk, and (2) the deformations of the microtubule because these experiments were not done on axonemal microtubules. All other suggestions or questions that can be dealt with in the text should also be addressed. I have included the full reviews from each reviewer below.

Reviewer #1:

Cytoplasmic dynein-1 traffics organelles and other cargos in the cytoplasm, while axonemal dyneins generate movement of flagella. Lacey et al. report the near-atomic resolution structures of the microtubule binding domain (MTBD) of cytoplasmic dynein-1 and an axonemal dynein, DNAH7. They solved both structures using new cryo-electron microscopy analysis methods implemented in Relion. These methods should be simpler for other labs to use compared to previous methods for solving helical structures. There are two main finding from this work. First, an updated model of cytoplasmic dynein-1's MTBD is presented. A previous structure obtained by combining lower resolution cryo-EM and molecular dynamics showed multiple rearrangements in the MTBD upon contact with the microtubule. Of these, only the movement of helix 1 is seen in the new high-resolution structure. Importantly, Lacey et al. report the helix 1 movement in both a construct containing a short region of anti-parallel coiled coil linked to the MTBD and in the context of a construct comprising the entire motor domain of cytoplasmic dynein-1. Second, they present the first high-resolution structure of an axonemal dynein MTBD bound to microtubules. This structure showed a similar overall mechanism of microtubule binding, a movement of helix 1, when compared to cytoplasmic dynein-1. An extension, or "flap", that had been observed in a solution structure of another axonemal dynein was also observed and found to interact with the neighboring microtubule protofilament. In what is arguably the most interesting finding in this paper, the authors observe distortions in the microtubule when the DNAH7 MTBD is bound to microtubules and discuss how this might relate to flagellar beating.

Essential revisions:

1) DNAH7 has low microtubule binding affinity with its own coiled-coil stalk, but high microtubule binding affinity when linked to the coiled-coil stalk of cytoplasmic dynein-1. There are at least two possible explanations for this. First, the dynein-1 stalk is better at stabilizing the high affinity state of the MTBD without a qualitative change in the MTBD's native conformations. Second, there could be more significant changes induced by the dynein-1 stalk compared to a native DNAH7 stalk, leading to artificially similar structures. Absent additional data, the authors should tone down the statement about universal conservation of the mechanism of microtubule binding.

2) For Figure 2B and D, I found it confusing that H1 wasn't represented in the stick model. It could be shown behind H3 in the darker pink color. The statement in the legend "as visible in the model" could be misinterpreted as meaning that other elements were not seen in the structures.

3) For Figure 3D, it would help the reader if each α and β tubulin density were colored in the two tones of green and labeled α and β.

4) In Figure 5A, why are the labels for DNA9, 11, and 17 in orange?

Reviewer #2:

In this manuscript, Lacey and his colleagues analyzed 3D structure of cytoplasmic and axonemal dynein microtubule-binding domains (MTBDs) binding to the microtubule, using cryo-EM technique. They carefully analyzed complex structure of MT-MTBD at unprecedented high resolution and updated the previous knowledge. They also proved that their new high resolution structure corresponds to the native dynein binding to MT.

They examined alternative binding of axonemal dynein, compared to cytoplasmic dynein, which proved that MTBD of axonemal dynein has contact to four tubulins (cytoplasmic dynein has only two contacts) and overall angle of binding is ~7degrees rotated. Moreover, they found a clue of flexible C-terminal of β-tubulin reoriented and fixed upon binding of axonemal dynein.

Another interesting finding is that binding of axonemal dynein MTBD induces distortion of MT lattice structure, making it rather elliptic instead of cylindrical.

Overall the new facts they provided by their high resolution cryo-EM study are landmark to characterize this important motor protein and will activate further research in this field. The results are presented beautifully in the graphical figures and methods are explained clearly. Considering the wealth of their findings, this manuscript deserves publication in any high journal and this reviewer enthusiastically recommends it for eLife after revising of the manuscript under the following viewpoints.

Title: Flap binding to the adjacent PF is an excellent discovery in this work. This reviewer would like to have it in the title, while the current title mentions only its influence on the MT lattice structure. Maybe "Axonemal dynein MTBD binds to 4 tubulins and distort MT structures" or "High resolution cryo-EM structure of axonemal dynein and MT revealed molecular mechanism of MTBD binding and its influence on MT lattice" will be suitable.

Subsection “Structural determination of cytoplasmic dynein-1 MTBD decorating microtubules”: This reviewer could not understand this sentence and cited figure panels. What "extension" does it mean? Does it mean that the tubulin core stretches? Or does it indicate any extended structure (like S9-S10 loop mentioned in Figure 1—figure supplement 1E) (if so, please indicate it in the Figure 1—figure supplement 2A)? Or is it phase extension to improve resolution – in this case it should read "from 4.4A to 3.6A resolution"? Please clarify this.

CTF panels such as Figure 1D should be improved. The letters in the footnote are too small and are hard to read.

Dynein c generates torque during in vitro motility assay, according to Oiwa's work, while another dynein (probably dynein f) is now to drive MT straightly. How is it interpreted to be linked with the unique interface between DNAH7 and tubulin?

Does 3D classification (of axonemal dynein and its MTBD, especially) show any sign of structural heterogeneity, for example detached flap from the adjacent PF?

Their finding may suggest that the energy landscape is more stable with elliptic MT upon binding of axonemal dyneins. Elliptic shape of B-tubule was reported in Maheshwari et al., (2015) Structure and Ichikawa et al., (2017). Is the orientation of MT flattening caused by dynein-binding consistent with the flattening of B-tubule?

Reviewer #3:

The manuscript by Lacey et al., reports high-resolution cryo-EM structures of the microtubule binding domains from cytoplasmic and axonemal dynein motors bound to microtubules. In order to accomplish this, the authors have developed novel image processing routines to aid in solving the structures of proteins bound to the microtubule lattice, providing a new tool for the structural biology community. While the structures of these dynein MTBDs have been examined previously, this is the first study to elucidate their structures bound to microtubules. The authors find substantial differences in the structural movements of the cytoplasmic dynein MTBD from that reported previously in Redwine et al., a study that relied on much lower resolution cryo structures and molecular dynamics simulations. Thus, this study provides an updated model for how the dynein MTBD binds MTs during the motor's mechanochemical cycle.

Maybe more interestingly, the authors find that an insert in the axonemal MTBD makes a distinct contact with the neighboring protofilament in the MT, and also observe a likely interaction between the B-tubulin C-terminal tail and an acidic patch on the dynein MTBD. They go on to show that the binding of this MTBD distorts the MT lattice, possibly thorough this alternative binding interface via the MTBD flap insert. They show that the axonemal MTBD prefers to bind to MTs at regions of low protofilament curvature, possibly providing an explanation for how axonemal dyneins sense curvature during flagellar beating. The final observation is very interesting; that the axonemal dyneins containing the flap insert are all closely spaced within the axonemal repeat structure, and thus all of them make contact with only a small subset of protofilaments on the B2 doublet.

Overall, I think this is an outstanding paper, of very high interest in both biological insight and technological advance. I am not qualified to review the details of their cryo-EM algorithms, so I will take the author's word for the quality of the new technique. I find the biological insights to be stimulating, and think they open up very interesting new questions about the role of distinct axonemal dynein classes in flagellar movement.

The only concern I have is that the MT distortion is observed on reconstituted MTs made from brain tubulin, an unnatural lattice for the axonemal MTBD. Since the natural lattice for this domain is the B1 doublet, how can we be sure the distortion observed is not an artifact of using single MTs? The authors should probably mention this caveat. Additionally, the authors make a fairly convincing argument that the flap binding to a neighboring protofilament is what may cause the lattice distortion. A formal test of this hypothesis would make the manuscript stronger. Mutation or deletion of the flap region could be done to examine if this restores the MT lattice to its original shape.