1. Cell Biology
  2. Microbiology and Infectious Disease
Download icon

Cryo electron tomography with volta phase plate reveals novel structural foundations of the 96-nm axonemal repeat in the pathogen Trypanosoma brucei

  1. Simon Imhof
  2. Jiayan Zhang
  3. Hui Wang
  4. Khanh Huy Bui
  5. Hoangkim Nguyen
  6. Ivo Atanasov
  7. Wong H Hui
  8. Shun Kai Yang
  9. Z Hong Zhou  Is a corresponding author
  10. Kent L Hill  Is a corresponding author
  1. University of California, Los Angeles, United States
  2. McGill University, United States
Research Article
  • Cited 0
  • Views 2,066
  • Annotations
Cite this article as: eLife 2019;8:e52058 doi: 10.7554/eLife.52058

Abstract

The 96-nm axonemal repeat includes dynein motors and accessory structures as the foundation for motility of eukaryotic flagella and cilia. However, high-resolution 3D axoneme structures are unavailable for organisms among the Excavates, which include pathogens of medical and economic importance. Here we report cryo electron tomography structures of the 96-nm repeat from Trypanosoma brucei, a protozoan parasite in the Excavate lineage that causes African trypanosomiasis. We examined bloodstream and procyclic life cycle stages, and a knockdown lacking DRC11/CMF22 of the nexin dynein regulatory complex (NDRC). Sub-tomogram averaging yields a resolution of 21.8 Å for the 96-nm repeat. We discovered several lineage-specific structures, including novel inter-doublet linkages and microtubule inner proteins (MIPs). We establish that DRC11/CMF22 is required for the NDRC proximal lobe that binds the adjacent doublet microtubule. We propose that lineage-specific elaboration of axoneme structure in T. brucei reflects adaptations to support unique motility needs in diverse host environments.

eLife digest

The parasites that cause African sleeping sickness, known as trypanosomes, propel themselves forward using a structure called a flagellum, a bit like the tail of a human sperm. But rather than connect to the body of the cell just at the base, like in a sperm, the parasite flagellum runs along the side of the cell. This means that, when it beats, the whole cell twists in a screw-like motion. The parasite flagellum beats vigorously, changes direction often, and puts the cell under lots of mechanical stress. This unusual motion likely helps the parasites to move through a thick and sticky fluid like blood.

The similarities between the parasite flagellum and the flagellum on a human sperm are down to a shared evolutionary history. Both structures contain the same basic molecular skeleton, known as the axoneme. The axoneme contains a combination of supporting proteins and molecular motors, and the molecular motors essentially pull on the supports to bend the flagellum.

The unusual movement of trypanosome parasites suggests that their axonemes may have unique structural features. But the three-dimensional structure of trypanosome axonemes had previously not been studied in great detail. Imhof, Zhang et al. now address this gap in knowledge using a technique called “cryo electron tomography” and showed that axoneme structure in trypanosomes does share many features with those of other organisms but it has extra proteins and connections for support, which could help to protect the flagellum from mechanical stress.

The similarities and differences between human and trypanosome flagella could indicate new drug targets that could be used to protect us against these parasites. A better understanding of how flagella work in general could also give insights into human genetic diseases that involve problems with these structures.

Introduction

Flagella (also called cilia) are hair-like structures that protrude from the surface of eukaryotic cells and perform motility and signaling functions (Smith and Rohatgi, 2010). These activities are essential for health, development and reproduction in humans and other multicellular organisms and to power movement of protists, including microbial pathogens that afflict nearly one billion people worldwide and present an economic burden as agricultural pests (Langousis and Hill, 2014; Gerdes et al., 2009; Ibanez-Tallon, 2003; Anvarian et al., 2019).

The structural basis for the flagellum is the axoneme, and in motile flagella the axoneme typically has a ‘9+2’ arrangement, consisting of 9 doublet microtubules (DMTs) arrayed symmetrically around a pair of singlet microtubules, with radial spokes (RS) extending inward from each DMT contacting the central pair (Khan and Scholey, 2018). Axoneme beating is driven by dynein motors and associated structures arranged in a repeating unit of 96-nm periodicity along each DMT. This 96-nm axonemal repeat is thus the foundational unit of motility for eukaryotic flagella. Canonical features of the repeat are four outer arm dyneins (OAD) (each having two or three motor domains, depending on species), seven inner arm dyneins (IAD) (one, IAD-f, having two motor domains and the others having a single motor domain), the nexin dynein regulatory complex (NDRC) inter-doublet linkage, and two or three RS (Porter and Sale, 2000). The most proximal IAD in the 96nm repeat, IAD-f, is distinguished from other IADs by having two motor domains, a large Intermediate Chain/Light Chain (IC/LC) complex that connects to the OAD and the NDRC, and extra connections to the A-tubule (Nicastro et al., 2006; Heuser et al., 2012a). Within each 96-nm repeat, dynein motors are permanently affixed to the A-tubule of one DMT and use ATP-dependent binding, translocation and release of the B-tubule on the adjacent DMT to drive microtubule sliding (Gibbons and Rowe, 1965). DMT attachment to the basal body at one end, together with ATP-independent connections, called nexin links, between adjacent DMTs, limits sliding and therefore causes DMTs to bend in response to dynein activity (Satir, 1968; Satir et al., 2014; Holwill and Satir, 1990). Precise, spatiotemporal coordination of dynein activity on different DMTs enables the bend to be propagated along the length of the axoneme, giving rise to axonemal beating (Satir, 1968; Lin and Nicastro, 2018). RS, together with the NDRC and the IAD-f-IC/LC complex, are thought to provide a means for transmitting mechanochemical signals across the axoneme as part of a complex and as yet incompletely understood system for regulating dynein activity (Porter and Sale, 2000; Satir et al., 2014; King, 2018; Viswanadha et al., 2017).

Recent advances in cryo electron tomography (cryoET) have made high-resolution, 3D structural analyses of the 96-nm repeat possible, providing insights into mechanisms of axoneme assembly and motility (Nicastro et al., 2006; Lin and Nicastro, 2018; Bui et al., 2009; Oda et al., 2014a; Jordan et al., 2018). However, such analyses have been limited to a restricted number of cell types and phylogenetic lineages. In particular, there has been no such analysis of the 96-nm repeat in any member of the Excavates (Figure 1), which includes several human and agricultural pathogens of importance to global public health. Consequently, we lack understanding of the full range of structural foundations for axoneme assembly and motility, and what structural variations underlie lineage-specific beating patterns observed in different organisms. For pathogens, such variations present potential therapeutic targets.

Phylogenetic tree of eukaryotes.

The tree is adapted from Dacks and Field (2018) and Adl et al. (2019). High-resolution structures of the 96-nm repeat of the axoneme are published for the clades indicated in blue, with the corresponding organism depicted in cartoon. T. brucei is in the clade Kinetoplastida, indicated in red, and represents the Excavates (EXC) that includes other pathogens, such as Giardia within Metamonada, also depicted in cartoon. The position of the last eukaryotic common ancestor (LECA) is indicated. AMR: Amorphea; DIA: Diaphoretickes; and EXC: Excavates are indicated.

African trypanosomes, Trypanosoma brucei (T. brucei) and related species, are parasitic protists in the Euglenozoa branch of the Excavates (Figure 1) (Koonin, 2010). They are medically and economically important pathogens of humans and other mammals (Langousis and Hill, 2014). Critical to T. brucei infection of a mammalian host (Shimogawa et al., 2018) and to their transmission via a tsetse fly vector (Rotureau et al., 2014), is motility of these parasites within and through host tissues. Motility of trypanosomes is driven by a single flagellum that is laterally connected to the cell body along most of its length (Figure 2A) (Langousis and Hill, 2014; Heddergott et al., 2012). The T. brucei flagellum consists of a 9+2 axoneme and a lineage-specific extra-axonemal structure, termed the paraflagellar rod (PFR), which runs alongside the axoneme for most of its length (Langousis and Hill, 2014; Hughes et al., 2012; Koyfman et al., 2011; Cachon et al., 1988). While the PFR exerts influence on the axoneme (Koyfman et al., 2011; Santrich et al., 1997), motility itself is driven by axoneme beating, which is transmitted directly to the cell, deforming the cell membrane and underlying cytoskeleton as the waveform propagates along the axoneme (Sun et al., 2018). Unlike most organisms, trypanosome axoneme beating propagates from the distal tip to proximal end in a helical wave, creating torsional strain and causing the cell to rotate on its long axis as it translocates with the flagellum tip leading (Heddergott et al., 2012; Walker, 1961; Walker and Walker, 1963; Rodríguez et al., 2009) (Videos 1 and 2). In essence, the entire cell rotates like an auger as it moves forward. This distinctive form of locomotion provides advantages for moving in viscous environments (Jahn and Bovee, 1968; Bargul et al., 2016) such as within human and fly tissues, and gives the genus its name, as Trypanosoma combines the Greek words for auger (trypanon) and body (soma) (Gruby, 1843).

Intact demembranated flagella from BSF T. brucei.

(A) A representative scanning electron microscope image of a procyclic form T. brucei parasite , with the cell body colored blue. The inset is a transmission electron microscope image of the flagellum from BSF T. brucei in representative transverse section, viewed from the proximal end, showing the 9+2 axoneme and PFR, enclosed within the flagellar membrane which is outlined by the yellow dotted line (adapted from Hill, 2003). (B–E) Negative stain TEM images of purified flagellum samples from BSF T. brucei, distributed on the grid with minimal clustering (B), showing that the axoneme and PFR are intact (C), with the basal body and pro-basal body on the proximal end (D), and a tapered tip at the distal end (E). The black box in (B) shows the approximate region chosen to image for cryoET. (F) Histogram of the length distribution of purified flagellum samples showing that the majority are full-length with a mean length of 25.2 microns (standard deviation = 3.5 microns). (G) A zero-degree tilted cryoEM image shows intact Axoneme, PFR and Ax-PFR connectors (arrows) from BSF T. brucei. (H–I) 6-A thick digital slice from a representative tomogram showing the sample in longitudinal (H) and the transverse (I) sections, with main structures labelled. Black line indicates one 96-nm axonemal repeat.

Video 1
Real-time video showing two T. brucei BSF parasites in culture medium.

The two parasites collide, illustrating the need for trypanosomes to accommodate interactions with other cells and tissues, which is common in the native environment of the mammalian host and insect vector.

Video 2
Real-time video showing a T. brucei BSF parasites moving in mouse blood, diluted 1:100 with culture medium.

Movement with flagellum tip leading and contact with host red blood cells is evident.

The combination of unusual locomotion mechanism, unique connections to other structures, and adaptation to diverse environmental conditions, suggests that the 96-nm repeating unit of the trypanosome axoneme might harbor lineage-specific elaborations. To investigate this possibility, we employed cryoET and sub-tomogram averaging to determine the 3D structure of the T. brucei 96-nm axonemal repeat. We report the 96-nm axonemal repeat structure for wild type parasites in bloodstream (BSF) and procyclic (PCF) stages, and for an RNAi knockdown targeting the CMF22/DRC11 subunit of the NDRC. Our results reveal lineage-specific adaptations, including novel inter-doublet linkages and microtubule inner proteins (MIPs). We also identify an NDRC subunit involved in inter-doublet connections between adjacent DMTs. We propose that lineage-specific adaptations to the 96-nm repeat may support the unique motility needs of these pathogens.

Results

3D structure of the trypanosome 96-nm axonemal repeat

A critical element of defining any structure is to ensure the sample is pristine. Our analyses demonstrated that flagellar skeletons purified from bloodstream form (BSF) trypanosomes are intact, including intact PFR, basal body and distal tip with uniform length distribution and a mean length of 25.2, + /- 3.5 µm (Figure 2B–F). Next it is critical that freezing does not distort the sample. A single zero-degree tilt image of a flagellum embedded in ice demonstrated that the axoneme, PFR and axoneme-PFR connectors remain intact following plunge freezing (Figure 2G). Having established high quality of vitrified samples, tilt series were collected from the center part of full-length flagella, spanning the middle third between the basal body and tip (Figure 2B). Major axonemal and PFR structures were resolved in slices through a single tomogram (Figure 2H,I, Video 3), indicating the 3D structure is well-preserved and relatively uncompressed (Figure 3—figure supplement 1).

Video 3
Slices through a representative tomogram reconstructed by simultaneous iterative reconstruction technique (SIRT) of BSF T. brucei.

Sub-volumes, that is particles, encompassing the 96-nm repeat of DMTs were extracted from 10 tomograms and averaged as described in Materials and methods. In total, 763 particles were averaged to determine the 3D structure of the axonemal repeat (Figure 3A–D, Video 4). The average resolution of the entire structure is 21.8 Å based on the 0.143 Fourier shell correlation criterion (Figure 3—figure supplement 2A). The resolutions at different regions vary based on visual inspection, and assessments by both local Fourier shell correlation (FSC) and ResMap (Kucukelbir et al., 2014) calculations (Figure 3—figure supplement 2A,C–F); the resolution of DMT region with MIPs reached 19.0 Å based on local FSC calculation (Figure 3—figure supplement 2A).

Figure 3 with 4 supplements see all
The 3D ultrastructure of the 96-nm repeat from intact axonemes of BSF T. brucei.

(A) A representative cross-section of a demembranated and negative-stained T. brucei flagellum, viewed from the proximal end (adapted from Hughes et al., 2012). Boxed region orients the view of the averaged 96-nm repeat along a DMT shown in B. (B) Cross-section view of the 96-nm repeat obtained by sub-tomogram averaging. Labeled are: the A- and B-tubule (At, Bt), Microtubule Inner Proteins (MIPs), Radial Spokes (RS), Inner Arm Dyneins (IAD), Nexin Dynein Regulatory Complex (NDRC), IAD-f-Intermediate Chain/Light Chain Complex (f-IC/LC), Outer Arm Dynein (OAD). The surface of the B-tubule from the adjacent DMT is visible on the right. The coloring scheme is as follows: cyan, OAD; red, IAD; blue, RS; green, NDRC; yellow, dynein f IC/LC. This scheme is consistently used throughout all main figures, figure supplements and videos unless stated otherwise. (C, D) Shaded surface rendering longitudinal views of the 96-nm repeat. Panel C shows the view from the center of the axoneme looking outward with the proximal end of the axoneme on the left and spoke heads removed for clarity (rotation relative to Panel D is shown). The surface of the B-tubule of the adjacent DMT is visible on top. Yellow and green arrows point to the inter-doublet connections formed by the f-connector and NDRC, respectively. Red arrows point to the proximal and distal holes in the inner junction between the A- and B-tubules. Panel D shows the view from the adjacent DMT, with proximal end of the axoneme on the left (rotation relative to panel B is shown). For reference, alpha (α) and beta (β) OAD are indicated, individual IADs and RS are labeled. (E) Shaded surface rendering of the averaged 96-nm repeat with the IAD, OAD and MIA complex removed, showing a massive structure at the base of the RS3 (see also Figure 3—figure supplement 1). Red arrows point to the density corresponding to the FAP59/172, 96-nm ruler (Oda et al., 2014a) between protofilaments A2 and A3. (F, G) Longitudinal (F) and transverse (G) density slices of the averaged 96-nm repeat. Red arrows in panels E and F point at the density of the FAP59/172 ruler between protofilaments A2 and A3. The red dashed line and perspective cartoon in panel F show the position and perspective of the cross-section shown in G, with the white arrow in panels F and G indicating the FAP59/172 ruler.

Video 4
3D surface rendering of the averaged 96-nm axonemal repeat from BSF T. brucei, rotated to show the structures of DMT (grey), Radial spokes (blue), NDRC (green), f IC/LC (yellow) and OAD (cyan) and IAD (red).

The 3D structure of the 96-nm repeat clearly resolved the expected major substructures, including OAD, IAD, RS, the IC/LC complex of IAD-f and the NDRC (Figure 3B–E). Individual protofilaments are well-resolved and even alpha and beta tubulin monomers within protofilaments are clearly resolved (Figure 3F). Several MIPs are also observed (Figure 3B). At this resolution, we observed a filamentous structure on the outside of the DMT that spans the entire 96-nm repeat (Figure 3E–G, red and white arrows). The location and extended conformation of this structure lead us to propose it to be the FAP59/172 molecular ruler described in Chlamydomonas that defines the 96-nm repeat (Oda et al., 2014a). Supporting this idea, the structure makes direct contact with RS, whose position depends on the FAP59/172 ruler (Oda et al., 2014a). The position of this ruler was previously determined in Chlamydomonas through mass-tagging, but the structure itself was not resolved (Oda et al., 2014a). We also observed a novel globular structure outside the B-tubule, between protofilaments B7 and B8, having a periodicity of 8 nm (Figure 3—figure supplement 3A,B blue arrow). The function of this structure is unknown, but it might influence dynein binding, because the microtubule binding domain of OADα contacts the B-tubule at this position (see Figure 4E red arrow), and its 8 nm periodicity is in the range of estimated step size for dynein and kinesin motors (Kikkawa, 2013; Reck-Peterson et al., 2006; Coy et al., 1999).

In situ structure of outer arm dyneins and novel OAD-alpha inter-doublet connector in BSF T. brucei.

(A) Shaded surface rendering, longitudinal view of the averaged 96-nm repeat. Coloring as described for Figure 3A. The box around the OAD indicates the region and perspective shown in B (red box) and D (blue box). (B, D) Shaded surface renderings of outer arm dyneins from the averaged 96-nm repeat. (B) Two adjacent OADβ dyneins. The linker and tail domains are colored yellow and the AAA+ ring is red. Cartoon overlay shows the post-powerstroke position of dynein. (D) Top view of two adjacent OADα dyneins. The linker and tail domains are colored in yellow and the AAA+ ring is colored in red. The arrow points to the OADα connector (purple), at the junction between the tail and linker domains. Cartoon overlay shows the post-powerstroke position of dynein. (C) A schematic illustrating relative DMT movement as dynein moves from pre-powerstroke one state (left) to post-powerstroke state (right). (E–F) Density slices of the averaged 96-nm repeat, viewed in cross-section, viewed from the distal tip of the axoneme. Red arrows indicate the dynein stalk domain in (E), and the OADα connector in (F), contacting the neighboring DMT.

Two holes were observed in the inner junction between the A- and B-tubules (red arrows in Figure 3C). We termed these ‘proximal’ and ‘distal’ holes, based on their position relative to the proximal end of the axoneme. The distal hole is near the site of NDRC attachment to the DMT and corresponds to the hole reported in other organisms (Nicastro et al., 2011; Pigino et al., 2012). The distal hole in Chlamydomonas is dependent on the presence of the NDRC on the external face of the DMT (Heuser et al., 2012b). The proximal hole is specific to T. brucei. Unlike the distal hole, there are no structures on the external face of the DMT at the site where the proximal hole is located. This indicates the proximal hole reflects structural properties imparted by proteins of the inner junction or inside the microtubules and is not dependent on the presence of external structures.

Interconnections were observed between substructures on the A-tubule, including between individual OADs (Figures 3D and 4), between OAD and the IAD-f complex (Figure 3B,D, Figure 3—figure supplement 3A,D). Particularly noteworthy are extensive contacts between RS3, IAD-d, and the A and B-tubules (Figure 3C, Figure 3—figure supplement 3C,D). At the base of RS3 we observed a structure that extends over four A-tubule protofilaments and attaches to the inner junction. Unlike the case for Chlamydomonas (Nicastro et al., 2006), the NDRC did not make direct contact with the OAD in T. brucei (Figure 3—figure supplement 3D), suggesting differences in mechanisms for coordinating inner and outer dynein motor activities.

Axonemal dynein arrangement in T. brucei

An earlier cryoET study of the T. brucei axoneme revealed the expected 4 OADs/repeat but did not resolve individual dynein motors (Hughes et al., 2012). With sub-tomogram averaging, the beta and alpha OAD motors are now clearly resolved (Figures 3B,D and 4A). This result provides the first direct demonstration that OADs contain two motor domains in T. brucei, making it the first protist shown to have two motors per OAD and correcting a misconception that all protists contain three motors (Lin and Nicastro, 2018). Together with three radial spokes per repeat, the entire arrangement of the T. brucei axoneme determined here therefore resembles that of humans more so than does Chlamydomonas or Tetrahymena, which are used as models for human cilium structure and function (Figure 5) (Pigino et al., 2012; Owa et al., 2019; Lin et al., 2014).

Comparison of 96-nm axonemal repeat structures across species.

(A–D) Structure of the 96-nm axonemal repeat is shown for Chlamydomonas reinhardtii (A) (Owa et al., 2019), Tetrahymena thermophilus (B) (Pigino et al., 2012), Homo sapiens (C) (Lin et al., 2014) and BSF Trypanosoma brucei (D) (this work). Longitudinal (top) and cross-sectional (bottom) views are shown for each. Canonical features of the 96-nm repeat are colored, including outer arm dyneins (cyan), inner arm dyneins (red and numbered according to convention) the IC/LC complex of inner arm dynein f (yellow), the NDRC inter-doublet linkage (green) and radial spokes (blue). The microtubule lattice is gray and the A- and B-tubules are indicated. MIP3 (red) is present in all organisms shown and is colored in the B-tubule for reference. For all structures except that from C. reinhardtii, the surface of the B-tubule from the adjacent DMT is shown. Inner dyneins and radial spokes are labeled for reference. The red dashed line indicates the position of viewing for the cross-section shown. All structures are filtered to resolution of 50 Å. Features that distinguish the T. brucei repeat include the f-connector (yellow arrow), missing dynein-c (red arrow), lineage specific MIPs within the A- and B-tubules (gray arrow), and two OAD motors in a protist (cyan arrow). Other T. brucei-specific structures, such as the OAD-alpha inter-doublet connector and b-connector are not visible in this view.

Axoneme motility is driven by rotation of the dynein AAA+ ring relative to the linker and tail domains, causing translocation of adjacent DMTs as the dynein transitions from pre-powerstroke to post-powerstroke position (Lin and Nicastro, 2018; Kikkawa, 2013; Burgess et al., 2003). The AAA+ ring, linker and tail domains were resolved in the OAD-beta dynein and are in the post-power stroke position (Figure 4B,C), consistent with the fact that samples were prepared without exogenous ATP. This result thus supports structural assignments in the averaged structure. The dynein stalk domain, which contacts the adjacent DMT is visible (Figure 4E).

Six IADs were well-resolved (Figure 3C,D) and annotated f, a, b, e, g, and d, according to standard nomenclature (Bui et al., 2012). Notably, IAD-c, which is important for movement of Chlamydomonas in high viscosity (Yagi et al., 2005), is absent from the trypanosome structure. This finding is notable, given the very viscous environments experienced by trypanosomes during movement through tissues of the mammalian host (Heddergott et al., 2012; Bargul et al., 2016; Capewell et al., 2016; Trindade et al., 2016) and tsetse fly vector (Schuster et al., 2017).

Extensive Inter-doublet connections in the T. brucei axoneme

Nexin links are connections between adjacent DMTs, that are visible in axoneme TEM thin sections. They stabilize the axoneme and are a fundamental component of the sliding filament model for axoneme motility (Satir, 1968; Satir et al., 2014; Viswanadha et al., 2017). Prior studies indicate the NDRC is the only nexin link in Chlamydomonas (Figure 5A) (Heuser et al., 2009). In T. brucei, however, we identified two prominent inter-doublet connections, the NDRC and the IC/LC complex of IAD-f (Figure 3B–D). We term this second connection the ‘f-connector’. The NDRC and f-connector each extend from the A-tubule of one DMT to contact near protofilament B9 of the adjacent DMT. NDRC contact is through the proximal and distal lobes defined by Heuser et al. (2009) and extends approximately 31 nm. The f-connector contact region extends approximately 11 nm. A structure analogous to the f-connector is observed between neighboring DMTs of three specific DMT pairs in Chlamydomonas (Bui et al., 2009). However, the prominence of the f-connector observed here in T. brucei suggests it is present between neighboring DMTs of most and perhaps all DMTs, a conclusion supported by analysis of individual DMTs (see below), indicating that nexin links in T. brucei include both the NDRC and the f-connector, as well as the OAD inter-doublet connector described below. This distinguishes the T. brucei axoneme from 3D axoneme structures from other organisms so far reported (Figure 5) (Pigino et al., 2012; Owa et al., 2019; Lin et al., 2014).

A conspicuous structure not previously reported in any organism is a large protrusion at the junction between the tail and stalk domains of OAD-alpha (Figure 4D,F). This protrusion, which we termed the ‘OAD inter-doublet connector’, extends to the space between protofilament B6 and B7 of the adjacent DMT. The OAD inter-doublet connector is thus distinguished from the OAD-alpha stalk, which extends from the AAA+ ring to the space between protofilament B7 and B8 of the adjacent DMT (Figure 4E). The OAD inter-doublet connector is present on all four OAD-alpha motors in the 96-nm repeat but is not observed in OAD-beta.

Doublet-specific features of the 96-nm repeat

The 96-nm repeat structure described above represents an average of all nine DMTs and does not reflect heterogeneity that may distinguish individual DMTs, as reported for Chlamydomonas (Bui et al., 2012). To address this, we did sub-tomogram averaging on each DMT separately. The PFR restricts axoneme orientations on the EM grid and consequently, individual DMT structures suffer from the missing wedge. This was most severe for DMT 3 and 7 and we therefore cannot comment on these DMTs (Figure 6—figure supplement 1F–G). For the remaining seven DMTs, distortion due to the missing-wedge problem obscured some details, particularly MIPs and OADs. However, main features of the 96-nm repeat were resolved (Figure 6—figure supplement 1B–E). Each DMT was distinct, but careful examination revealed some similarities, particularly in the region of IAD-b, between DMTs 1+5, 2+6 and 8+9 (Figure 6—figure supplement 1C–E). Therefore, to reduce the impact of the missing wedge, we averaged DMTs within these pairs together. We recognize that this approach may still mask some features of a single doublet, but it nonetheless reveals heterogeneity between doublets.

As shown in Figure 6 and Figure 6—figure supplement 1, we identified doublet-specific structures that were not evident in the entire averaged structure. DMT 8 and 9 are distinguished from all other DMTs in that they do not have an IAD-b. In the place of IAD-b is a previously undescribed arch-like structure that extends upward from between RS1 and RS2, which we termed ‘arch’ (Figure 6D). DMT 1 and 5 are distinguished by the presence of a novel inter-doublet connecter, which we termed ‘b-connector’, that connects IAD-b to the adjacent DMT and includes a ‘tail’ domain that connects with the ‘Modifier of Inner Arms’ MIA complex (Yamamoto et al., 2013) (Figure 6B). DMT 2 and 6 contain a b-connector that lacks the tail domain (Figure 6C). DMT 4, 8 and 9 lack the b-connector. Structural variation of the b-connector on different DMTs explains why it was not evident in the entire averaged structure. DMTs 1, 4, 5, 6, 8 and nine each have an f-connector structure. DMT two does not have a clear f-connector, but this may reflect a missing wedge artifact since the density of the NDRC connection is also reduced (Figure 6—figure supplement 1D). The analysis of individual DMTs supports the interpretation that the f-connector is present on most DMTs. Additionally, this analysis identified a new lineage specific inter-doublet connection not present in other organisms, the b-connector.

Figure 6 with 1 supplement see all
Doublet-specific structures of the BSF T. brucei 96-nm repeat.

(A) Schematic showing the numbering of individual DMTs. Colored boxes indicate the DMT pairs that were used for the averaged structures shown in panels B-D. (B–D) Panels show averaged structures for DMT pairs 1+5 (B), 2+6 (C), and 8+9 (D). Inner arm dyneins (red) and radial spokes (blue) are labeled for reference. The f-connector, b-connector and the arch that distinguish DMTs 8 and 9 are colored yellow, purple and brown, respectively.

The PFR is attached to DMT 4, 5, 6 and 7 and we therefore considered whether this attachment alters the 96-nm repeat. As detailed above, two PFR-attached DMTs, DMT 5 and 6, each show similarities to non-attached DMTs, DMT 1 and 2, that are not shared by each other (Figure 6—figure supplement 1A,C–D). Therefore, PFR attachment does not seem to correlate with specific structural changes in the 96-nm repeat, at least at the current resolution. PFR-attachment complexes themselves, have a 56 nm periodicity (Hughes et al., 2012; Koyfman et al., 2011) and therefore would not be resolved in our 96-nm repeat structure.

CMF22/DRC11 is part of the NDRC proximal lobe involved in binding the adjacent DMT

The NDRC functions in axoneme stability and motility and these functions are thought to be mediated in part through inter-doublet connections (Viswanadha et al., 2017; Olbrich et al., 2015; Wirschell et al., 2013; Ralston and Hill, 2006). The NDRC is composed of at least 11 subunits and some of these have been positioned within the complex (Heuser et al., 2009; Yamamoto et al., 2013; Ralston et al., 2006; Nguyen et al., 2013; Kabututu et al., 2010; Bower et al., 2013; Lin et al., 2011; Huang et al., 1982; Song et al., 2015; Oda et al., 2014b). However, subunits that contact the B-tubule of the adjacent DMT are unknown. We identified CMF22 as a subunit of the T. brucei NDRC (Nguyen et al., 2013), and the Chlamydomonas CMF22 orthologue is DRC11 (Bower et al., 2013). RNAi knockdown of CMF22/DRC11 abolishes forward motility in T. brucei, demonstrating the importance of DRC11 in axoneme motility (Video 5 and Video 6) (Nguyen et al., 2013). The position of CMF22/DRC11 in the NDRC is unknown, but biochemical data indicate it may be within the proximal or distal lobe structures that contact the adjacent DMT (Nguyen et al., 2013; Bower et al., 2013; Awata et al., 2015). We therefore used cryoET and sub-tomogram averaging to determine the structural basis of the CMF22/DRC11 RNAi knockdown. We used procyclic culture form (PCF) T. brucei, because loss of axonemal components is lethal in bloodstream forms (Ralston and Hill, 2006; Broadhead et al., 2006; Ralston and Hill, 2008).

Video 5
Real-time video showing wild type motility of a PCF T. brucei parasite in culture medium.

The parasite translocates using a helical movement with flagellum tip leading.

Video 6
Real-time video of a CMF22-knockdown PCF T. brucei parasite in culture medium.

The flagellum beats but is unable to drive translocation of the parasite.

The 96-nm repeat of WT PCF (Figure 7A) axonemes was very similar to that of BSF (Figures 3 and 4), including the presence of the novel OAD inter-doublet connector and the f-connector, as well as the missing IAD-c. In the CMF22 knockdown, the only structure clearly affected is the NDRC (Figure 7C–E). The entire structure of the complex is mostly preserved, but the proximal lobe of the linker region is severely reduced (Figure 7E). The affected structures encompass a large portion of the inter-doublet contact area for the T. brucei NDRC and include both regions reported to contact the adjacent DMT in the Chlamydomonas NDRC (Heuser et al., 2009). The remaining NDRC domains, including dynein contacts were not grossly affected, although connection from NDRC to the MIA complex (Yamamoto et al., 2013) might be altered. Therefore, inter-doublet connection mediated by the NDRC is critical for axoneme motility.

Comparison between averaged 96-nm repeats of wild-type and CMF22/DRC11 knockdown PCF T. brucei.

(A, C) Sub-tomogram averages of the 96-nm repeats of wild-type (A), and CMF22/DRC11 knockdown mutant (C). Yellow and green arrows point to the region of the B-tubule contacted by the f-connector and NDRC, respectively. (B, D) Zoomed-in view of the NDRC from WT (B) and CMF22/DRC11 knockdown (D). The red arrow in each panel denotes the structure most substantially affected in the knockdown. (E) Superposition of the NDRC structures shown in B and D, with WT in pink and the mutant in green. The red arrow indicates the most striking difference, corresponding to inter-doublet contacts made by the NDRC.

Extensive, lineage-specific MIPs in T. brucei

One major advance resulting from cryoET studies is the discovery that protein structures inside the microtubule, first observed in trypanosomes based on transmission EM studies more than fifty years ago (Vickerman, 1969; Anderson and Ellis, 1965), are ubiquitous in axonemal microtubules (Nicastro et al., 2011; Ichikawa et al., 2017). A striking feature of T. brucei axonemal microtubules is the presence of extensive MIP complexes not only in the A-tubule, but also in the B-tubule (Figures 3B and 8A and Supplementary file 1). Figure 8A shows a cross-section view of the averaged 96-nm repeat looking from the proximal end of the axoneme, with MIPs colored and external structures removed for clarity. The B-tubule is on top and the A-tubule is below, with 13 protofilaments of the A-tubule and 10 protofilaments of the B-tubule labeled according to convention (Figure 8A). The shape, position and periodicity of the structure inside the B-tubule, next to the inner junction between the A- and B-tubules (Figure 8A,B), indicate that this structure corresponds to MIP3 described in other organisms (Nicastro et al., 2011; Ichikawa et al., 2017). Notably however, the relationship of other MIPs in T. brucei to previously described MIPs is unclear and most TbMIPs in both the A- and B-tubules appear to be trypanosome-specific (Figure 5).

Figure 8 with 1 supplement see all
TbMIP3 and ponticulus in the B-tubule of BSF T. brucei.

(A) Guide figure showing cross-section view of the averaged 96-nm repeat, viewed from the proximal end of the axoneme with MIPs colored and densities external to the DMT removed. Red and blue lines indicate sections and viewing perspectives shown in panels (B) and (C), respectively. (B) Longitudinal view into the inside of the B-tubule showing structural variations of TbMIP3 (red, yellow and orange) described in the text, with a periodicity of 48 nm. Arrows indicate the proximal and distal holes in the inner junction. Asterisk indicates MIP3a attachment to a structure identified as MIP3c in Chlamydomonas (Owa et al., 2019). Proximal (base) and distal (tip) ends of the repeat are indicated and rotation relative to panel A is shown. (C) Longitudinal view into the inside of the B-tubule showing ponticulus complexes Pa, Pb and Pc with a periodicity of 48 nm. Arrows indicate the distal and proximal holes in the inner junction and rotation relative to panel A is shown. (D–F) Top panels show cross-sections of average density maps viewed from the axoneme's distal tip to proximal end into the DMT. A subset of protofilaments are labeled for reference and rotation relative to panel C is shown. The trypanosome-specific Ponticulus (Pa, Pb and Pc) is seen bridging the entire lumen of the B-tubule from protofilament A12 to protofilaments B3, B5, and B4, respectively. The corresponding 3D isosurface renderings, looking from the same position are shown below, with Ponticulus-Pa, Pb and Pc, colored in blue, red and yellow respectively.

When viewed in longitudinal section from within the B-tubule, TbMIP3 consists of two lobes, 3a and 3b (Figure 8B), as reported for Chlamydomonas and Tetrahymena (Nicastro et al., 2011; Ichikawa et al., 2017). There are six such TbMIP3 structures in each 96-nm repeat. Subtle structural variations in the sizes of lobe 3b and connections to lobe 3a yield a 48 nm repeating pattern of three adjacent TbMIP3 structures, colored red, gold and orange (Figure 8B). These TbMIP3 variations coincide with other structural variations within the microtubule, such as presence of inner junction holes (arrows in Figure 8B), unique contacts to Snake MIP (see Snake MIP description below), and attachment to a structure identified as MIP3c in Chlamydomonas (Owa et al., 2019) (asterisks in Figure 8B). Variation in lobe 3b between the two gold TbMIP3 structures could suggest a 96-nm repeat unit, but this variation probably results from interference from the DRC base plate on the outside of the DMT at the site of the distal hole.

Facing TbMIP3, on the opposite side of the B-tubule lumen, are several trypanosome-specific MIPs, MIP B5, B4, B2 and a MIP that extends across the entire lumen, thus corresponding to the ponticulus structure previously observed in classical thin section TEM (Figure 8C) (Vickerman, 1969; Anderson and Ellis, 1965; Vaughan et al., 2006). To our knowledge, the ponticulus was the first structure observed within the microtubule lumen in any organism and is the only structure so far described to extend across the entire microtubule. Our 3D structure shows that the ponticulus is not a single structure, but rather is comprised of 3 discrete MIPs, which we termed Pa, Pb and Pc (Figure 8C–F). Each ponticulus MIP extends across the entire B-tubule lumen, connecting the A-tubule lattice to a different B-tubule protofilament. Pa, Pb and Pc connect protofilament A12 to protofilaments B3, 5 and 4, respectively and exhibit 48 nm periodicity (Figure 8C–F). The ponticulus is assembled after construction of the axoneme (Vaughan et al., 2006). Therefore, proteins comprising these structures must be delivered into a fully formed DMT.

The A tubule also contains a diverse cohort of MIPs each with a repeating unit of 48 nm (Supplementary file 1, Figure 8A, Figure 9—figure supplement 1). Rather than constituting several isolated structures however, TbMIPs form a network of interconnected complexes, similar to, but more extensive than, that reported for Tetrahymena (Ichikawa et al., 2017). Two A-tubule MIPs are particularly notable. One, which we termed ‘ring MIP’, is unique among MIPs so far described because it forms a ring structure protruding into the microtubule lumen (Figure 9B). The ring MIP is attached to the protofilaments A8 and 9 and contacts another MIP complex on the protofilaments A8-12 termed ‘Ring Associated MIP’ (RAM) (Figure 9B,C). Another MIP, which we termed ‘snake MIP’, presents as a serpentine structure that appears to weave in and out of the A and B-tubules (Figure 10 and Video 7). The continuity of this density suggests it might be a contiguous structure, extending 48 nm and spanning multiple tubulin subunits, although we cannot rule out the possibility that protofilament subunits contribute to this structure.

Figure 9 with 1 supplement see all
The RingMIP and Ring Associated MIP (RAM) in the A-tubule of BSF T. brucei.

(A) Guide figure showing cross-section view of the averaged 96-nm repeat, viewed from the proximal end of the axoneme with MIPs colored and densities external to the DMT removed. Red and blue lines indicate sections and viewing perspectives shown in panels (B) and (C), respectively. (B–C) Longitudinal view of the A-tubule, showing the RingMIP and RAM. Left panels are sections through averaged density maps and right panels are corresponding isosurface renderings showing the same structures. The RingMIP (fuchsia), as well as its neighboring Ring Associated MIP (RAM) (red) and MIPA5-7 (cyan) are shown. The proximal (base) and distal (tip) ends of the axoneme are indicated and rotation of panel C relative to panel B is shown.

The snake MIP connects the A-tubule and the B-tubule of BSF T. brucei.

(A) Guide figure showing cross-section view of the averaged 96-nm repeat, viewed from the proximal end of the axoneme with MIPs colored and densities external to the DMT removed. Black lines 1 and 6 show the position and perspective of sections shown in B. (B) Longitudinal view of the averaged density map. A and B-tubules are labeled. Panels 1 through 6 show six 6 Å thick, consecutive digital sections (the distance between 2 sections is 6.2 Å) through the snake MIP. (C) Left panel is a guide figure showing cross-section view of the averaged 96-nm repeat, viewed from the proximal end of the axoneme with MIPs colored and densities external to the DMT removed. Black line shows position and perspective for view of snake MIP shown in the right panel. Right panel shows segmented TbMIP3 (red, yellow and orange, as described for Figure 8B) and Snake MIP (mauve). (See also Video 7.).

Video 7
3D surface rendering of the averaged 96-nm axonemal repeat from BSF T. brucei, rotated to show the structures of DMT (grey), Radial spokes (blue), NDRC (green), f IC/LC (yellow), OAD (cyan) and IAD (red).

Structures other than the Snake MIP (mauve) and TbMIP3 (red, yellow and orange) fade away to emphasize the Snake MIP structure and its connection to TbMIP3 substructures.

Discussion

The ciliary axoneme is one of the most iconic features of eukaryotic cells and is considered to have been present in the last eukaryotic common ancestor (LECA) (Khan and Scholey, 2018). To date, however, high-resolution structures of the 96-nm axoneme repeat have only been reported for two of the three eukaryotic supergroups. Here we report the 3D ultrastructure of the T. brucei 96-nm axonemal repeat. This is the first such structure reported for any pathogenic organism and first representative from the eukaryotic lineage of Excavates, a basal group that includes many pathogens of global importance to human health and agriculture (Hampl et al., 2009; Dawson and Paredez, 2013). Our studies indicate the diversity of structures comprising the 96-nm repeat is under appreciated, give insight into principles of axoneme structure and function, and identify pathogen-specific features that may support unique motility needs of trypanosomes.

The genus Trypanosoma was discovered more than 175 years ago and named for its unique cell motility (Gruby, 1843), which is driven by a single flagellum. The functional unit of the eukaryotic flagellum is the 96-nm axonemal repeat, which encompasses dynein motors and regulatory proteins that direct flagellum beating (Porter and Sale, 2000). In trypanosomes, the PFR exerts influence on the axoneme (Koyfman et al., 2011; Santrich et al., 1997; Bastin et al., 1998), but motility is powered by the axoneme, which is the focus of the current work. Despite intense study for several decades, axoneme structures that underpin the parasite’s unique mechanism of cell propulsion remained hitherto unclear. A main finding from our studies is the discovery of lineage-specific features of the T. brucei 96-nm axonemal repeat, including extensive and novel MIP structures and novel inter-doublet connections between adjacent DMTs (Figures 36 and 811). Figure 11 shows a schematic overview of the overall 96-nm structure, previously undescribed features are labeled in panel B. We hypothesize these parasite-specific structures support unique motility needs of trypanosomes and thereby contribute to the transmission and pathogenic capacity of these organisms. The T. brucei axoneme is distinguished by mechanical strain experienced due to lateral attachment to the PFR and cell body, vigorous helical beating, encounter with host tissues and frequent reversals of beat direction (Shimogawa et al., 2018; Koyfman et al., 2011; Santrich et al., 1997; Bargul et al., 2016). MIPs have been shown to stabilize the axoneme in other organisms (Owa et al., 2019; Ichikawa and Bui, 2018; Stoddard et al., 2018) and the expanded and MIP network of T. brucei may therefore help maintain stability of individual DMTs. Likewise, novel inter-doublet connections are expected to help maintain axoneme integrity under these conditions, analogous to the role of NDRC inter-doublet links in maintaining alignment of DMTs in Chlamydomonas (Bower et al., 2013).

Schematic overview of the trypanosome axoneme.

(A) Cartoon longitudinal view of the entire averaged 96-nm axonemal repeat. Major labeled structures are Outer Arm Dyneins (OAD), Inner Arm Dyneins (IAD), dynein-f IC/LC, Nexin Dynein Regulatory Complex (NDRC), Radial Spokes (RS) and Ruler. Image is oriented with proximal end (base) at the left. (B and C) Cartoon cross-section view of the axoneme (viewed from the proximal end) at roughly the position of RS1 (B) and RS2 (C). Protofilaments are numbered and structures are labelled as for panel A. Major trypanosome-specific structures described in text are labelled in pink. Note that additional T. brucei-specific structures, RingMIP, RAM MIP and b-connector are not visible in this simplified depiction. Summary of MIP structures is provided in Supplementary file 1.

The diversity and placement of T. brucei MIPs are suggestive of functions beyond stability. It is difficult to imagine for example, how a ring structure like the RingMIP, protruding into the microtubule lumen, would solely provide stability. MIPs in other organisms have been demonstrated to modulate axoneme beating (Owa et al., 2019; Stoddard et al., 2018). Given the presence of numerous trypanosome-specific MIPs, together with MIP differences reported between other species (Figure 5), we suggest that lineage-specific MIPs may provide a mechanism for fine-tuning the beating of axonemes between species that otherwise share a basic architecture. Extra connections between DMTs can also influence axoneme beating. It has been suggested that vortical beating of nodal cilia in vertebrates axoneme may involve transmission of regulatory signals from DMT to DMT, circumferentially around the axoneme (King, 2018). Extensive inter-doublet connections identified in our studies provide a means for direct interaction between DMTs and could thus contribute to helical beating that is a hallmark of T. brucei motility. Finally, given the recent demonstration that motility is critical for T. brucei virulence (Shimogawa et al., 2018), parasite-specific features of the 96-nm repeat, which is the foundational unit of motility, may present novel therapeutic targets. Future work to identify novel T. brucei MIP and connector proteins will allow these ideas to be tested directly.

By defining the structural basis of the motility defect in the CMF22/DRC11 knockdown, we demonstrate a specific requirement for inter-doublet connections in axoneme motility because the defect disrupts inter-doublet connections without affecting dyneins. This contrasts to NDRC mutants analyzed previously in Chlamydomonas, which typically exhibit structural defects in connections to dyneins or in dyneins themselves (Heuser et al., 2009; Awata et al., 2015; Bower et al., 2018). An exception is sup-pf4 (Heuser et al., 2009), but this mutant has only subtle effects on motility and beat frequency (Awata et al., 2015), which contrasts to the CMF22/DRC11 knockdown in which propulsive motility is ablated (Nguyen et al., 2013). Our CMF22/DRC11 knockdown studies therefore provide several important insights. Firstly, they demonstrate that penetrance of RNAi makes knockdown lines suitable for differential cryoET structural analysis in T. brucei. Secondly, they demonstrate CMF22/DRC11 is required for NDRC proximal lobe assembly and B-tubule attachment and, together with biochemical data (Nguyen et al., 2013; Bower et al., 2013; Awata et al., 2015), indicate that CMF22/DRC11 is part of the proximal lobe. Thirdly, because inter-doublet contacts are specifically affected, without affecting dyneins, the results demonstrate that the NDRC itself and B-tubule contacts specifically are required for control of axoneme motility. This last point is particularly significant, as dynein-independent connection between adjacent DMTs is considered to be a founding principle of the sliding filament model for axoneme motility (Satir, 1968; Holwill and Satir, 1990; Viswanadha et al., 2017), yet direct tests of this idea have been limited.

The 96-nm spacing of the axoneme is controlled by a molecular ruler (Oda et al., 2014a), which is visible in the averaged BSF 96-nm repeat structure. The T. brucei MIP repeating unit is 48 nm, suggesting existence of a separate ruler inside the DMT to guide MIP placement. Such a ruler would need to extend 48 nm, exhibit structural heterogeneity along its length, and form contacts with other MIPs. The snake MIP satisfies these criteria. Notice, for example, that structural heterogeneities along the snake MIP coincide with unique contacts to each TbMIP3a, b structure within the 48 nm repeat (Figure 8). The snake MIP appears to extend into both the A- and B-tubules, which would make it possible to establish patterns in both tubules. Extensive interconnections between MIPs (Video 7) might allow a single ruler to guide placement of all MIPs, or there might be more than one ruler, as is suggested for the outside of DMTs in Chlamydomonas (Song et al., 2018), where the 24 nm repeat of OADs is dictated by something other than the FAP59/172 ruler (Oda et al., 2014a). Besides the snake MIP, another structure inside the B-tubule (spine MIP) appears to exhibit properties required of a 48 nm molecular ruler - forming a contiguous structure, spanning 48 nm and having heterogeneities that make unique contacts to adjacent MIPs (Figure 8—figure supplement 1).

Materials and methods

Key resources

Request a detailed protocol

BSF single marker (BSSM) and PCF (Wirtz et al., 1999) T. brucei cells were used. The CMF22/DRC11 knockdown line is described (Nguyen et al., 2013).

Preparation of demembranated flagellum skeletons for cryoET

Request a detailed protocol

BSF single marker (BSSM) and PCF (Wirtz et al., 1999) T. brucei cells were cultured as described (Shimogawa et al., 2015; Saada et al., 2014) and authenticated based on selective and morphogenetic markers. Cells, 2 × 108 for BSF or 4 × 108 for PCF, were washed three times in sterile 1xPBS. Supernatant was aspirated to ensure all of the PBS is removed. To remove the cell membrane and other soluble proteins and release the DNA,160 µl Extraction buffer (20 mM HEPES pH: 7.4, 1 mM MgCl2, 150 mM NaCl, 0.5% NP40 IGEPAL CA-630 detergent, 2x Protease Inhibitors Cocktail-Sigma EDTA-free) + 1/10 vol 10x DNase buffer + 1/10 vol DNase (TURBO, Life Technologies 2 U/μl) was added and incubated at room temperature for 15 min. In order to solubilize the subpellicular microtubules, 1 mM CaCl2 (2 µl of 100 mM CaCl2) was added and incubated on ice for 30 min. Then flagellum skeletons (axoneme with PFR, basal body and FAZ filament) were centrifuged (1500 g at 4°C for 10 min) and the supernatant was removed. Then flagellum skeletons were purified away from cell body remnants and debris by one further centrifugation step over a 30% sucrose cushion at,800g at 4°C for 5 min (Extraction buffer w/o NP-40; 30% w/v sucrose). Flagellum skeletons from 200 μl of the upper fraction of the buffer-sucrose interface were collected and washed twice in 200 μl Extraction buffer, centrifugation at 1500 g at 4°C for 10 min, then resuspended in 40 µl buffer. Samples were either mixed with gold beads and plunge frozen immediately, as described below, or assessed directly for sample quality. To assess sample quality, BSF samples were negative-stained and analyzed using an FEI T12 transmission electron microscope equipped with a Gatan 2k × 2 k CCD camera. Samples were intact with uniform length distribution and a mean length of 25.2, + /- 3.5 µm (Figure 2B–F). PCF samples were examined by light microscopy to ensure uniform length distribution.

CryoET sample preparation and tilt-series acquisition

Request a detailed protocol

BSF or PCF samples in the amount of 40 µl was mixed with either 5 nm (for BSF) or 10 nm (for PCF) diameter fiducial gold beads in 12:1 ratio. An aliquot of 3 µl of the axoneme-gold beads solution was applied onto Quantifoil (3:1) holey carbon grids (for BSF) or continuous carbon-coated EM grids (for PCF) which were freshly glow-discharged for 30 s at −40 mA. Excess of the sample on the grid was blotted away with a filter paper, at a blot force of −4 and blot time of 5 s, and vitrified by immediately plunging into liquid nitrogen-cooled liquid ethane with an FEI Mark IV Vitrobot cryo-sample plunger. Axoneme architectural integrity and gold bead concentration were assessed and plunge-freezing conditions optimized by obtaining low-resolution cryoET tilt series in an FEI TF20 transmission electron microscope equipped with an Eagle 2K HS CCD camera. From these tilt series, cryoET tomograms were evaluated to ensure structural integrity of the axoneme and PFR. Vitrified cryoET grids were stored in liquid nitrogen until use.

For high-resolution cryoET tilt series acquisition, vitrified specimens were transferred with a cryo-holder into an FEI Titan Krios 300kV transmission electron microscope equipped with a Gatan imaging filter (GIF) and a Gatan K2 Summit direct electron detector. Samples were imaged under low-dose condition using an energy filter slit of 20 eV. CryoET tilt series were recorded with SerialEM (Mastronarde, 2005) by tilting the specimen stage from −60° to +60° with 2° increments. The cumulative electron dosage was limited to 100 ~ 110 e-2 per tilt series. All 4k × 4 k frames were recorded on a Gatan K2 Summit direct electron detector in counting mode with the dose rate of 8–10 e-/pixel/s. For each tilt angle, a movie consisting of 7 to 8 frames was recorded. For the PCF samples, the nominal magnification was x26,000, giving rise to a calibrated pixel size of 6.102 Å. The defocus value was targeted at −4 µm. When the BSF samples were ready to be imaged, the same instrument was upgraded with a VPP, allowing us to obtain higher contrast images at closer to focus and higher magnification conditions. To obtain tilt series for the BSF samples with VPP, we follow the procedures previously described (Fukuda et al., 2015; Si et al., 2018) and used the same GIF and K2 parameters as indicated above. Before starting each tilt series, we moved to a new VPP slot, waited for 2 min for stabilization, then pre-conditioned the VPP by illumination with a total electron dose of 12 nC for 60 s to achieve a phase shift of ~54°. Tilt series were recorded at a nominal magnification of 53,000X (corresponding to a calibrated pixel size of 2.553 Å) and a targeted defocus value of −0.6 µm. For BSF we collected a total of 50 tomograms and selected the 10 best, based on limited axoneme compression for sub-tomogram averaging. Cross sections of these 10 tomograms are shown in Figure 3—figure supplement 1, and have circularity, measured as ratio of short axis/long axis, ranging from 0.92 to 0.98. This yielded 763 particles that were averaged to determine the 3D structure of the BSF axonemal repeat. For WT PCF we collected 27 tomograms, and 17 of them were used for sub-tomogram averaging, resulting in 1177 particles averaged. For DRC11/CMF22 RNAi samples a total of 24 tomograms were collected and 19 of them were used for sub-tomogram averaging, resulting in 1726 particles averaged. For sub-tomogram averaging of individual DMT (Figure 6 and Figure 6—figure supplement 1), an additional 24 tomograms of BSF axonemes were used, for a total of 34 tomograms, yielding 297 to 339 particles averaged for each DMT (DMT1 = 339, DMT2 = 332, DMT3 = 297, DMT4 = 327, DMT5 = 311, DMT6 = 337, DMT7 = 316, DMT8 = 306, DMT9 = 309).

Data processing

Request a detailed protocol

For PCF and BSF samples, frames in each movie of the raw tilt series were drift-corrected, coarsely aligned and averaged with Motioncorr (Li et al., 2013), which produced a single image for each tilting angle. The tilt series images were reconstructed into 3D tomograms by weighted back projections using the IMOD software package (Kremer et al., 1996) in six steps. Micrographs in a tilt series were coarsely aligned by cross-correlation (step 1) and then finely aligned by tracking selected gold fiducial beads (step 2). The positions of each bead in all micrographs of the tilt series were fitted into a specimen-movements mathematical model, resulting in a series of predicted positions. The mean residual error was recorded to facilitate bead tracking and poorly-modeled-bead fixing (step 3). With the boundary box reset and the tilt axis readjusted (step 4), images were realigned (step 5). Finally, tomograms were generated by weighted back projection (step 6). Contrast transfer function (CTF) was corrected with the ctfphaseflip program (Xiong et al., 2009) of IMOD in step five above. The defocus value for each micrograph was determined by CTFTILT (Mindell and Grigorieff, 2003), and the estimated defocus value was used as input for ctfphaseflip. Note, one of the benefits of using a phase plate is that the CTF is insensitive to the sign of the defocus value being negative (underfocus) or positive (overfocus) (Fan et al., 2017).

To improve the signal-to-noise ratio and enhance the resolution, sub-tomograms containing the 96-nm axonemal repeated units along each DMT were extracted/boxed out from the raw tomograms. Sub-tomogram averaging and the missing-wedge compensation were performed using PEET program (Nicastro et al., 2006; Heumann et al., 2011) as detailed previously (Si et al., 2018), except for a new script we wrote to pick sub-volumes as outlined in the subsequent paragraphs.

In our sub-tomogram averaging scheme, each particle is defined as the 96-nm repeating unit of the DMT. We developed a MATLAB script, autoPicker, to semi-automatically pick particles and calculate their location and orientation based on axoneme geometry. Briefly, we represent the 9+2 axoneme as a cylinder. For each axoneme in a tomogram, we used IMOD to visually pinpoint 11 points and save their coordinates into a file. The first two points, pa and pb, are the center points of the two bases of the cylinder. The remaining 9 points (pi, i=1…9) identify the centers of the nine DMTs (particles) within the first 96-nm length at one end of the selected axoneme. The center is defined as the intersection point of a DMT with the middle of the three radial spokes along each particle’s 96-nm unit length. Our script reads the coordinates of the 11 points, calculates vector papb that defines the orientation of the cylinder, determine the center coordinates of all other particles within this axoneme based on the following formula:

pij=piLjpapb|papb|., where i = 1, 9; j = 1 to |papb|/L, L is the unit length (96nm)

In order to uniquely identify the orientation of each particle, autoPicker also calculates a second point, p*ij for each pij. p*ij corresponds to the middle radial spoke’s end near the central pair. This is accomplished by solving the following linear algebraic equations that both p*ij and pij must satisfy (see illustrations in Figure 3—figure supplement 4):

{papb  pijpij=0(papb ×papij ) pijpij=0|pijpij|=Length of the radial spoke (60nm)

We ran autoPicker for each axoneme in our tomograms to generate a PEET mod file that contains a list of the above described pij and p*ij pairs for all particles in that axoneme. Program stalkInit in PEET then read this mod file and generate an initial motive list file, a RotAxes file and three model files containing the coordinates for each particle. PEET then read the coordinate and orientation information from these files and automatically extracted the particles from the tomograms to perform iterative sub-tomogram averaging until no further improvement can be obtained.

Sub-tomogram averaging of the individual DMTs was performed in two steps. Step1: particles (96-nm repeat units), picked from all 9 DMTs were classified into nine classes, corresponding to the DMT from which each particle was picked, DMT 1–9. Step 2: for particles in each of the nine classes, sub-tomogram averaging was performed using PEET.

The resolutions of the sub-tomogram averages were evaluated by two different approaches, one based on Fourier shell correlation (FSC) calculated by simpleFSC in PEET (Nicastro et al., 2006; Heumann et al., 2011) and the other by ResMap (Kucukelbir et al., 2014). To calculate FSC curves, we split all particles into two of equal-sized subsets following the PEET tutorial. Specifically, particles are separated into two subsets with the PEET specific motive list file by designating each sub-volume as either ‘1’ or ‘2’ so that it would be placed into one of the two sub-sets. PEET then performed sub-tomogram averages independently for particles in each of the two equal-sized sub-sets, yielding two sub-tomogram averages of the 96-nm axonemal structure. These two independently calculated sub-tomogram averages were then used as the input maps of the simpleFSC program in the PEET package to calculate the FSC curve for the entire 96-nm axonemal repeat (Figure 3—figure supplement 2A). We also calculate FSC curves for local regions encompassing DMT with MIPs, OAD, IAD, NDRC or RS. To do so, a cuboid mask was used in ChimeraX (Goddard et al., 2018) to extract two corresponding local density regions that primarily containing either DMT with MIPs, or OAD, or IAD, or NDRC or RS from the two sub-tomogram averages. Each set of two corresponding cuboid volumes (Figure 3—figure supplement 2B) was then used as the input maps of the simpleFSC program in the PEET package to calculate an FSC curve for the local region, which is plotted as a function of spatial frequency (Figure 3—figure supplement 2A). Local resolution across the entire averaged 96-nm axonemal repeat was also evaluated with ResMap (Kucukelbir et al., 2014) using the above two independently calculated sub-tomogram averages as input maps and the result is visualized from different views in Figure 3—figure supplement 2C).

3D visualization

Request a detailed protocol

IMOD (Kremer et al., 1996) was used to visualize the reconstructed tilt-series and the 2D tomographic slices of the sub-tomogram averages. UCSF ChimeraX (Goddard et al., 2018) was used to visualize the resulting sub-tomogram averages in their three dimensions. Segmentation of densities maps and surface rendering for the different components of the 96-nm repeated unit were performed by the tools volume tracer and color zone in UCSF Chimera (Pettersen et al., 2004). GIMP 2.8.18 (GNU Image Manipulation Program) was used to color regions of interest (Figures 5, 6B–D, 8B–F and 9B–C; Figure 3—figure supplement 3C–D, Figure 8—figure supplement 1B, Figure 9—figure supplement 1B; Supplementary file 1). For rendering, no filters were applied on MIPS but we applied low pass filters on the other components to improve the clarity of individual structures described in the text. For the structures in Figure 3C–E; Figure 4A,B,D; Figure 7A–E, we filtered the DMT, NDRC, RS, IC/LC, OAD and IAD to 30 Å. For the structures in Figure 5; Figure 6; Figure 6—figure supplement 1, we filtered the entire map to 50 Å).

Trypanosome motility videos

Request a detailed protocol

Motility videos of BSF cells were obtained exactly as described in Kisalu et al. (2014). Motility videos of PCF cells were obtained exactly as described in Nguyen et al. (2013). All videos were recorded and played back at 30 frames per second. The PCF tetracycline-inducible DRC11/CMF22 RNAi knockdown line has been described previously (Nguyen et al., 2013). WT and mutant PCF videos correspond to this knockdown line cultured in the absence (WT) or presence (mutant) of 1 μg/ml tetracycline to induce RNAi.

Data availability

Request a detailed protocol

All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figure 2F and Figure 3—figure supplement 4. The cryoET sub-tomogram average maps have been deposited in the EM Data Bank under the accession codes EMD-20012, EMD-20013 and EMD-20014, for the wild-type bloodstream form, wild-type and DRC11-knock-down procyclic form, respectively.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
    Recherches et observations sur une nouvelle espèce d'hématozoaire, Trypanosoma sanguinis
    1. M Gruby
    (1843)
    Comptes Rendus Hebdomadaire Des Séances De l'Académie Des Sciences, Paris 17:1134–1136.
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
    Locomotion of Blood Protists
    1. TL Jahn
    2. EC Bovee
    (1968)
    In: D Weinman, M Ristic, editors. Infectious Blood Diseases of Man and Animals, 1. Academic Press. pp. 393–436.
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
    Studies on cilia
    1. P Satir
    (1968)
    The Journal of Cell Biology 39:77.
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
    Cilia 2010: the surprise organelle of the decade
    1. EF Smith
    2. R Rohatgi
    (2010)
    Science Signaling 4:mr1.
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
    On the surface coat and flagellar adhesion in trypanosomes
    1. K Vickerman
    (1969)
    Journal of Cell Science 5:163–193.
  87. 87
  88. 88
  89. 89
    Movement of trypanosome flagella
    1. PJ Walker
    2. JC Walker
    (1963)
    Journal of Protozoology, 10.
  90. 90
  91. 91
  92. 92
  93. 93
  94. 94

Decision letter

  1. Andrew P Carter
    Reviewing Editor; MRC Laboratory of Molecular Biology, United Kingdom
  2. John Kuriyan
    Senior Editor; University of California, Berkeley, United States
  3. Benjamin D Engel
    Reviewer; Max Planck Institute of Biochemistry, Germany
  4. Brooke Morriswood
    Reviewer; University of Wuerzburg, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study presents the first structures of the Trypanosome axoneme. Although cilia/flagella are found throughout evolution, structures are only available for a select few species. This structure from T. brucei is the first from the excavata supergroup, providing an important evolutionary comparison. Novel inter-doublet connections are observed beside the known Nexin Dynein Regulatory Complex (NDRC): connection to the IC/LC complex and a protrusion from the dynein linker. The NDRC connection is abolished in a CMF22/DRC11 RNAi knockdown, which shows this subunit is located close to the B-tubule and affects motility. Within the doublets, some T brucei MIP densities correspond to MIPs known from the T. thermophilus and C. reinhardtii structures. However, it also contains a large number of unidentified MIP densities which appear to be T brucei specific.

Decision letter after peer review:

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Cryo electron tomography reveals novel structural foundations of axoneme motility in the pathogen Trypanosoma brucei" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and four reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Benjamin D Engel (Reviewer #3); Brooke Morriswood (Reviewer #4).

Our decision has been reached after consultation between the reviewers and I regret to inform you that we felt the work could not be accepted in its current form. We felt that your manuscript's main message (the original aspects of the trypanosome axoneme) is potentially of interest to eLife readers, especially given the paucity of organisms generally studied in cilia biology. The reviewers also agreed that your paper brings interesting and novel insights. However, there was a strong consensus that your manuscript needs more work than would be reasonable to expect in a two month turn-around and hence our decision that it needs to be rejected at this stage. The reviewers comments are attached in full below. The main revisions required to make the paper suitable for eLife would be:

1) Address structural heterogeneity of the different doublet microtubules. Given the inherent asymmetric structure of the Trypanosome flagellum, with the paraflagellar rod at one side, this must be explored. You are using a small dataset and so additional tomograms may be required to resolve the different doublets.

2) Address the structure of the PFR, and thereby build on your and others prior observations of this accessory structure. Ideally you would also compare distal and proximal portions of the axoneme to account for known molecular differences (e.g. there is no PFR in the first micrometre of the axoneme).

3) Incorporate comparative images to show differences to the 96nm repeats from other organisms.

4) Redo your resolution estimations (as described by reviewer 3).

5) Overhaul your manuscript to address the reviewers comments. In particular: remove all statements about the "greatest" resolution, separate speculations from results, and include a proper summary of extant data on the 96nm repeat to place the results in proper context.

Reviewer #1:

The ciliary cytoskeleton (the axoneme) consists of nine microtubule doublets (MTDs) arranged in a nine-fold symmetric array. These doublets contain many accessory proteins such as Dyneins, Radial Spokes and microtubule inner proteins (MIPs) that are responsible for cilia motility and/or MTD stability. The proteins that are stably associated with the MTDs are recurring in regular repeats along the axoneme its length. Many efforts are and have been made to solve the structure of the axonemal repeats and inner proteins to understand cilia stability and the waveform and regulation of cilia motility.

The authors present the structure of the Trypanosoma brucei axoneme as a first representative of the excavata super group. Novel inter-doublet connections are observed beside the known Nexin Dynein Regulatory Complex (NDRC): connection to the IC/LC complex and a protrusion from the dynein linker. The NDRC connection is abolished in a CMF22/DRC11 RNAi knockdown, which shows this subunit is located close to the B-tubule and affects motility. Within the doublets, some T brucei MIP densities correspond to MIPs known from the T. thermophilus and C. reinhardtii structures. However, it also contains a large number of unidentified MIP densities which appear to be T brucei specific.

A number of axoneme structures have been solved recently. Our feeling is that there are some novel aspects to this T. brucei structure, but at the moment the findings are very descriptive. For example showing, but not identifying, the MIP densities. The authors list the differences in the 96-nm repeat, compared to other axonemal structures, but can't explain how these features relate to the function of the T.brucie axoneme.

Verifying the findings such as the other inter-doublet connections in vivo or identifying some of the MIP densities would make the manuscript more suitable for publication in eLife.

Reviewer #2:

This manuscript reports a high-quality data set on the 3D organization of the so-called 96-nm repeat structures in two developmental stages of the protist Trypanosoma brucei. Flagella were purified and analysed by cryo electron tomography and sub-tomogram averaging after plunge freezing. The resolution reached remarkable values (12-15 Å) for the bloodstream stage. In some cases, tubulin monomers could be resolved at the level of protofilaments. It revealed common features with other axonemes, especially from human cilia but also unique differences. Nice structural details of the internal composition of microtubules are reported, including the unique ponticulus and the discovery of novel microtubule internal proteins (MIP). Comparison with flagella of cells where the expression of a component of the nexin-dynein regulatory complex was knocked down revealed structural modifications. The proximal lobe of the NDRC complex appeared affected, what impaired the connection with the neighbouring doublet and could explain the motility defect.

This is a very nice study, with some new developments (MATLAB script) and will be of interest for cilia biologists and structural biologists at large.

Three points need to be clarified to fully understand the results:

1) Exponentially growing cultures of trypanosomes contain about 50% of cells with two flagella (the mature one and the growing one). Based on length distribution of bloodstream form flagella (Figure 2), it seems that mostly full-length flagella are present in the sample. How is it possible? Is it due to the purification procedure that would select long flagella? This is important because at least the distal end of growing and mature flagella are known to be different in terms of structure (Höög et al., 2014) or composition (Subota et al., 2014). Please clarify. Length measurements for procyclic form flagella are missing.

2) A recent paper (Edwards et al., 2018) showed that the docking of outer dynein arms was different along the length of the trypanosome flagellum (in PCF), with specific proximal and distal docking complexes. Subsection “3D Structure of the trypanosome 96-nm axonemal repeat”: "tilt series were collected from the center part of the flagellum, spanning the middle third between the basal body and tip". A cartoon would help to avoid ambiguity but if we understood correctly, it seems that the central portion of the flagellum was used. It is where the two docking systems are likely to overlap, hence potentially generating heterogeneity. Is the organisation the same in the short portion without paraflagellar rod (PFR)?

3) Which doublets (out of the 9) were selected for analysis? The same group has shown that doublets were not equivalent (Hughes et al., 2012), especially those connected to the PFR where dynein arms looked different.

The Title and the interpretation should be tuned down, although unique structural features (especially MIPs) are indeed reported, there is little direct evidence that they contribute to the original axoneme motility.

Writing-up. There is a lot of interpretation in the result section. The authors should either remove these and do a more exhaustive Discussion section, or write the paper with Results section and Discussion section combined.

Reviewer #3:

This study by Imhof et al., presents the first structures of the Trypanosome axoneme. Although cilia/flagella are found throughout evolution, structures are only available for a select few species. This structure from T. brucei is the first from the excavata supergroup, providing an important evolutionary comparison. The authors present several interesting findings, including the descriptions of only two dynein motors per OAD, increased connections between doublet microtubules (DMTs), and several lineage-specific microtubule inner proteins (MIPs). They also use an RNAi knockdown line to determine the position of DRC11. In principle, I support publication of this work, and I think the data is a valuable addition to the axoneme field. However, there are several major points related to the analysis that must be addressed, in particular related to resolution estimation and exploring structural variation between different DMTs. In addition, there is too much pure speculation about the potential functions of several structures, with no experimental evidence to support these functions. This speculation should be removed or heavily qualified.

Essential revisions:

1) The authors heavily promote the resolution of their structure, claiming "This resolution is the highest cryoET structure yet reported for the 96-nm repeat from any organism". However, there are some serious issues with the resolution estimation, and as a result, I believe the resolution has been overstated.

The paper's abstract claims that the DMT is resolved to 12Å. This is based on the ResMap analysis shown in Supplemental Figure S3. However, this analysis is troubling. First of all, the resolutions on this map appear to primarily range from 12-16Å, whereas the FSC curve shows a global resolution of 21Å at the 0.143 cutoff. If all the local resolutions from ResMap are averaged, the result should be close to 21Å. However, it does not appear that this will be the case, and instead ResMap is estimating resolutions that are at least 5Å better than the FSC. Even the less well-resolved appendage structures that can be seen in this image range from 15-18Å. This is a bit difficult to judge from the figure because the authors have intentionally only shown the backside of the DMT (which reports the highest resolution). They must also show the other side, with all the important accessory structures (similar to Figure 3D), as well as a cross-section slice through the DMT to show the MIPs (similar to Figure 3B). Only then can we see the range of resolutions estimated by ResMap. But even in the view that is shown, the appearance on the DMT is way too speckled, with a huge dynamic range of 12-16Å on the microtubule wall. This noisy surface is a clear sign that something is wrong with the ResMap-the surfaces should look much smoother, with less hotspots of resolution variation.

As the ResMap cannot be taken at face-value, a parallel approach should be attempted to estimate local resolution. I recommend using masks to perform two local FSCs – the doublet region and the region containing the appendage structures (OAD, IAD, RS, NRDC). How do these FSCs compare to the global FSC and to the ResMap?

For the isosurface renderings throughout the paper, the DMT looks properly filtered, but the appendage structures appear to be oversharpened or displayed at a resolution that is too high (their surfaces look noisy). I assume that these maps were uniformly filtered to the same resolution. Was it the global 21Å? I can't seem to find this information in the paper. I would expect that by around 20Å, the holes in the middle of the dynein AAA+ rings would start to become visible, at least as indentations. But in these maps, the AAA+ rings look like round egg-shaped blobs, another sign that the resolution is not as high as claimed.

The authors used a tilt-series acquisition scheme that starts at -60 degrees and thus destroys the high-resolution information before reaching low tilt, as opposed to the much preferred dose-symmetric scheme starting at zero degrees combined with dose-weighting (see the high-resolution HIV work by John Briggs and Wim Hagen). Furthermore, there currently is no way to correct the contrast transfer function (CTF) for low dose cryo-tilt series acquired with the Volta Phase Plate. The authors thus did not perform CTF-correction, meaning that resolution of the average is limited to the first zero of the power spectrum. Given the -0.6 μm target defocus and the large defocus gradient that is present in thick samples such as these axonemes, especially at higher tilts, I anticipate that this first zero would strongly limit the resolution (20-25Å sounds about right, not 12Å). Therefore, I am very cautious of the bold resolution claims made in this paper.

Finally, the authors use a "gold-standard" FSC to determine resolution, but it was not clear to me from the methods when exactly the extracted particles were split into two half-sets and averaged independently, as is required for gold-standard assessment. With only 700 total particles, getting two half-sets to 21Å might be challenging. Please explicitly describe how the averaging was performed instead of just "as described previously (75) using PEET".

2) The focus of this paper is to show the evolutionary differences of the T. brucei axoneme. Of course, by far the most distinct feature of trypanosome flagella is the paraflagellar rod (PRF). This structure seems important for axoneme stability under strong forces, a key question the authors sought to address. In Figure 2G, the authors show clear periodic connections between the axoneme and PFR (marked with arrows). It would be very valuable to compare averages of PFR-linked versus non-PRF-linked DMTs. This should be a relatively easy task, just splitting particles into those two categories based on their location with respect to the PFR. Furthermore, the PFR appears to have a fairly regular structure in Figure 2G, so is it possible to generate an average of the PFR itself? Such a structure would be something really new, and would add value to this paper.

3) The globular density with an 8-nm repeat in Figure 3—figure supplement 3 has a weak signal, and thus appears to have low occupancy in this average. Why is this? Might it have something to do with the connection to the PFR? Judging by Figure 3A, one would expect about a third or a fourth of the DMTs to have a connection to the PFR. The authors should investigate how the occupancy of this density varies between DMTs by producing averages of different DMTs using their radial position relative to the PFR for reference. Without more information, the author's proposed function of this density in regulating dynein binding (subsection “3D Structure of the trypanosome 96-nm axonemal repeat”) is far too speculative.

4) Based on the strong signal of the IAD-f IC/LC density, the authors conclude: "An IAD-f IC/LC interdoublet connection is observed between three specific doublet pairs in Chlamydomonas (17). However, the prominence of the IAD-f IC/LC connection in T. brucei suggests it is present between most and perhaps all doublet microtubules, indicating that nexin links in T. brucei include both the NDRC and IAD-f IC/LC. This distinguishes T. brucei axonemes from the known 3D axoneme structures from other organisms (36, 46)." Similar to points 2 and 3 above, if the authors want to make this claim, then they absolutely must examine the density in averages of different DMTs around the T. brucei axoneme.

5) The issues raised in the four points above (problems with resolution estimation and neglecting to analyze structural variation between different DMTs) are all related to the very limited dataset used in this paper. The primary "high resolution" structure in this paper (wild-type BSF) was generated from only 700 particles from 10 tomograms. This is only half a day of acquisition on a Titan Krios microscope (tomograms take about 1 hour each). While I understand it is not eLife policy to ask for more experiments, I think it is completely appropriate in this case for the authors to spend one more day on the microscope with their already prepared cryo-EM samples to acquire 20 more tomograms. This would produce a sufficient dataset to perform classification and look at how specific densities vary between different DMTs (see points 2-4 above). I understand that the Volta Phase Plate used in this study enables averages to be generated with less particles, but I think the 700 particles in the current average are too few to properly do this analysis, and it is not clear to me why the authors chose to proceed with such a limited dataset.

6) The proximal and distal holes in the DMTs look convincing. However, what is not convincing to me is their proposed function of allowing MIPs to be incorporated after completion of the DMT. The holes are tiny, only about as wide as a tubulin monomer (4 nm), and thus do not seem big enough to allow the free transit of MIPs, which are significantly larger than the holes. Perhaps the holes could serve as a location for the start of an "unzipping" event between the A- and B-tubules, which could allow insertion of larger MIP structures such as the ponticulus, but this is completely speculative. I don't think much can be said about the function of the DMT holes at this point.

7) The extended discussion of the RingMIP function is highly speculative and should perhaps be omitted or at least down-weighted. Its proposed mechanosensory role is not proven by the data in this paper, just speculated.

8) The authors should be careful with stating the significance of the comparison between developmental stages; there's no clear conclusion from this comparison (other than a low-resolution hint of MIPA3-4). So, putting this in the abstract without indicating the negative result could be considered false advertising. Upon reading the abstract, I assumed that there were developmental differences, and I was disappointed when I finally discovered at the end of the paper that there were not.

Reviewer #4:

Imhof, Zhang et al., present cryo-electron tomography data on preparations of isolated axonemes from the unicellular parasite Trypanosoma brucei, eukaryotic supergroup Excavata. Specifically, they provide a detailed analysis of the 96nm repeat that forms the core structural unit of the axoneme. These data were obtained from two life cycle stages of the organism – the slender bloodstream form which is found in infected mammalian hosts and whose motility is better characterised, and also the procyclic form which is found in the midgut of the tsetse fly vector. Not only are the usual structural features such as the outer dynein arms, radial spokes resolved, but also some fascinating observations of the microtubule inner proteins are provided, which appear to be considerably more abundant in the axonemes of this organism than in others imaged to date.

Given that the axoneme is an almost ubiquitous structure that was present in the last eukaryotic common ancestor, the data here are extremely significant and of relevance far beyond the trypanosome and parasitology community. As the authors note, taxonomic sampling of the ultrastructure of the 96nm axonemal repeat is limited, and by providing data from the supergroup Excavata the authors have considerably broadened the perspective onto this structure. The observations of the microtubule inner protein complexes are fascinating and could set the stage for considerable future work.

It is important to note that I have no first-hand practical experience of cryo-electron tomography and am therefore not qualified to judge the technical aspects of the manuscript relating to this technique. I am happy for the opinions of the other referee(s) to take precedence on this point.

The data appear to be of very high quality and the figures do not require more than cosmetic alteration. No extra practical work seems required, I think, and I have no major concerns. The manuscript could be considerably tightened in order to do full justice to the quality of the results, however.

In particular, the Introduction needs much more detail. The authors should define the 96nm axonemal repeat and explicitly summarise previous work on this structural unit. This is important for placing the results in context and defining the paper's original contribution. The 96nm repeat is currently first mentioned in the Results section without any prior introduction. There has been substantial previous work on the morphology and ultrastructure of the trypanosome axoneme, so it is important to emphasise that this work focuses on the structure of the axoneme's 96nm repeat unit specifically. Presently there is considerable potential for being misread, and means that the authors are actually underselling what they have.

The manuscript also frequently rushes over the results. Panels 2A-2G are not individually cited, nor are 3A-3E, and this trend continues with the other figures. Ensure that all panels are cited, and ideally in figure order (i.e. A-B-C etc) for clarity. Figure 10 is currently not cited at all in the manuscript text. Not all the supplemental figures and movies appear to be cited in the manuscript text – this needs checking. The section on the comparison of the different developmental stages (currently Figure 9) could perhaps be moved so that it comes after Figure 4.

The text would benefit from some proofreading for English, and also for style (e.g. Results should always be presented in past tense, but the present is often used here). Check also that all panel citations are correct – there is some mix-up, particular of Figure 5 and Figure 6. There is also perhaps a bit too much interpretation in the Results section that would fit better in the Discussion section.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Cryo electron tomography reveals novel structural foundations of the 96-nm axonemal repeat in Trypanosoma brucei" for consideration by eLife. Your article has been reviewed by John Kuriyan as the Senior Editor, a Reviewing Editor, and three reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Benjamin D Engel (Reviewer #1); Brooke Morriswood (Reviewer #3).

The reviewers have discussed the reviews with one another and agreed that you have done an admirable job adding significant new data and analysis to this study, including new tomograms, doublet-specific subtomogram averaging, structural comparisons between species, and much more believable resolution estimates. The manuscript is now well accessible to a general reader. All our concerns are addressed, and the manuscript requires only minor revisions. Please aim to submit the revised version within two months.

Summary:

This manuscript reports a high-quality data set on the 3D organization of the so-called 96-nm repeat structures in two developmental stages of the protist Trypanosoma brucei. This is the first representative axoneme structure of the excavata super group. Flagella were purified and analysed by cryo electron tomography and sub-tomogram averaging after plunge freezing. The resolution reached remarkable values (12-15 Å) for the bloodstream stage. In some cases, tubulin monomers could be resolved at the level of protofilaments. It revealed common features with other axonemes, especially from human cilia but also unique differences. Nice structural details of the internal composition of microtubules are reported, including the unique ponticulus and the discovery of novel microtubule internal proteins (MIP). Comparison with flagella of cells where the expression of a component of the nexin-dynein regulatory complex was knocked down revealed structural modifications.

https://doi.org/10.7554/eLife.52058.sa1

Author response

Our decision has been reached after consultation between the reviewers and I regret to inform you that we felt the work could not be accepted in its current form. We felt that your manuscript's main message (the original aspects of the trypanosome axoneme) is potentially of interest to eLife readers, especially given the paucity of organisms generally studied in cilia biology. The reviewers also agreed that your paper brings interesting and novel insights. However, there was a strong consensus that your manuscript needs more work than would be reasonable to expect in a two month turn-around and hence our decision that it needs to be rejected at this stage. The reviewers comments are attached in full below. The main revisions required to make the paper suitable for ELife would be:

1) Address structural heterogeneity of the different doublet microtubules. Given the inherent asymmetric structure of the Trypanosome flagellum, with the paraflagellar rod at one side, this must be explored. You are using a small dataset and so additional tomograms may be required to resolve the different doublets.

As requested, we have now completed additional experiments to examine heterogeneity among doublet microtubules, including analysis of 24 additional tomograms, as specified in the Materials and methods section of the revised manuscript. The presence of the PFR restricts possible orientations of the sample such that the axoneme is always positioned with either DMT3 or DMT7 facing the EM grid. This leads to the missing wedge problem for any individual doublet and is why we had restricted analysis to average of all doublets in our original submission. However, we recognize the importance of understanding heterogeneity, and in response to this request we have analyzed additional tomograms (now 34 total for BSF samples) and obtained structures for individual doublets. The analysis revealed new features on specific doublets and results are described in the revised manuscript (subsection “Doublet-specific features of the 96-nm repeat”, Figure 6 and Figure 6—figure supplement 1).

2) Address the structure of the PFR, and thereby build on your and others prior observations of this accessory structure. Ideally you would also compare distal and proximal portions of the axoneme to account for known molecular differences (e.g. there is no PFR in the first micrometre of the axoneme).

As requested by reviewers, we have compared PFR‐attached versus not‐PFR‐attached DMTs. As described in the text (subsection “Doublet-specific features of the 96-nm repeat”), this comparison demonstrated that PFR‐attachment does not correlate with alterations in the structure of the axonemal 96‐nm repeat. Thus, although there are DMT specific variations to the 96‐nm repeat (Figure 6 and Figure 6—figure supplement 1), these are not correlated with PFR attachment. Note that the PFR and PFR‐attachment complexes themselves have a 56‐nm repeat period, so would not be resolved in our analysis here, which is focused on the 96‐nm repeat.

We made a substantial effort to determine axoneme structure specifically in the proximal‐most region of the axoneme that does not contain PFR. However, this region was severely twisted (see reviewer Author response image 1 and author response video 1), thus preventing us from obtaining a high‐resolution structure. Severe twisting was not observed for the remainder of the axoneme and it is not presently known whether twisting reflects inherent features of the proximal region, or if it is imposed during sample preparation. The proximal axoneme is near the basal body/pro‐basal body complex, as well as four microtubules that extend from the basal body and wrap in a helix around the axoneme (Lacomble et al., 2009). These additional structures are expected to restrict sample orientation and if that orientation differs from the orientation allowed by the PFR, it would cause the proximal region to be twisted relative to the remainder of the axoneme. Future work will aim to address this issue. We have clarified in the text (subsection “3D Structure of the trypanosome 96-nm axonemal repeat”, Figure 2B) that the 96‐nm repeat structure presented represents the middle third of the axoneme.

Regarding the PFR itself, the PFR is a massive structure, with a repeat period (56‐nm) (Koyfman et al., 2011) that differs from that of the axonemal 96‐nm repeat period. Determining high‐resolution, 3D structure of the PFR is a focus of ongoing and future efforts, but is beyond the scope of the current work, which is focused on the axonemal 96‐nm repeat.

Author response image 1
Proximal end of the axoneme is severely twisted.

1‐Å‐thick digital slices from a representative tomogram of the BSF axoneme near the basal body (BB) region. Two different sections, separated by 12 nm, are shown (120 and 132), with example twisted regions of the DMTs indicated by the red arrows.

Author response video 1
Proximal end of the axoneme is severely twisted.

Video of slices through the tomogram presented in Reviewer Figure R1, indicating twisting of the axoneme near the basal body. Video is compressed using H.265/HEVC MP4 video format.

3) Incorporate comparative images to show differences to the 96nm repeats from other organisms.

As requested, we have added images of 96‐nm repeats from other organisms (Figure 5). We selected organisms representing the eukaryotic supergroups discussed in Figure 1, including two within SAR, Chlamydomonas and Tetrahymena, as these represent two substantially divergent lineages in this supergroup. We also include a cartoon (Figure 11) that highlights distinctive and trypanosome‐specific structures discovered in the current work. Citations are provided for readers to refer to prior publications of the other organisms for detailed description of the 96‐nm repeat in each.

4) Redo your resolution estimations (as described by reviewer 3).

We have done this as requested. (See also, our detailed responses to reviewer 3.)

5) Overhaul your manuscript to address the reviewers comments. In particular: remove all statements about the "greatest" resolution, separate speculations from results, and include a proper summary of extant data on the 96nm repeat to place the results in proper context.

We have overhauled the manuscript text as requested. The entire text has been substantially changed, so individual changes are not specifically highlighted in the text, but lines relevant to specific reviewer and editor comments are indicated in this response letter. Statements of ‘greatest’ resolution are removed, and Results are separated from speculations. To place the results in proper context, a summary of extant knowledge regarding the 96‐nm repeat is provided in the Introduction, Figure 5 is added, and a cartoon is provided (Figure 11) to highlight relevant structures discovered here that are distinctive to T. brucei.

Reviewer #1:

The ciliary cytoskeleton (the axoneme) consists of nine microtubule doublets (MTDs) arranged in a nine-fold symmetric array. These doublets contain many accessory proteins such as Dyneins, Radial Spokes and microtubule inner proteins (MIPs) that are responsible for cilia motility and/or MTD stability. The proteins that are stably associated with the MTDs are recurring in regular repeats along the axoneme its length. Many efforts are and have been made to solve the structure of the axonemal repeats and inner proteins to understand cilia stability and the waveform and regulation of cilia motility.

The authors present the structure of the Trypanosoma brucei axoneme as a first representative of the excavata super group. Novel inter-doublet connections are observed beside the known Nexin Dynein Regulatory Complex (NDRC): connection to the IC/LC complex and a protrusion from the dynein linker. The NDRC connection is abolished in a CMF22/DRC11 RNAi knockdown, which shows this subunit is located close to the B-tubule and affects motility. Within the doublets, some T brucei MIP densities correspond to MIPs known from the T. thermophilus and C. reinhardtii structures. However, it also contains a large number of unidentified MIP densities which appear to be T brucei specific.

A number of axoneme structures have been solved recently. Our feeling is that there are some novel aspects to this T. brucei structure, but at the moment the findings are very descriptive. For example showing, but not identifying, the MIP densities. The authors list the differences in the 96-nm repeat, compared to other axonemal structures, but can't explain how these features relate to the function of the T.brucie axoneme.

Verifying the findings such as the other inter-doublet connections in vivo or identifying some of the MIP densities would make the manuscript more suitable for publication in ELife.

We appreciate the reviewer comments and concise summary. Regarding the request to identify proteins that make up novel MIPs and inter‐doublet connections: Densities within the microtubule lumen were reported more than fifty years ago (Vickerman, 1969; Anderson and Ellis, 1965) and MIPs as a ubiquitous entity of cilia were described thirteen years ago (Nicastro et al., 2006), but identities of MIP proteins were determined only recently (Stoddard et al., 2018; Ichikawa and Bui, 2018; Owa et al., 2019). Likewise, nexin links were observed in the 1960s (Gibbons, 1963), but molecular identities of these inter‐doublet linkages were not determined until recently (Lin et al., 2011). Thus, while we recognize the interest in identifying proteins that comprise novel structures just discovered in the T. brucei axoneme, we consider this to be a goal of future studies and beyond the scope of the current manuscript.

Regarding the concern about being descriptive: We recognize that defining structure is by nature descriptive and emphasize that our studies have identified several novel structures not previously observed in any axoneme, despite many years of prior study. We’ve tried to better articulate the importance of our findings and how they relate to trypanosome motility in the Discussion section. Moreover, a major contribution of our work is the discovery of extensive inter‐doublet connections in T. brucei and we go further, with functional analysis combining cryoET analysis plus RNAi knockdown, to reveal that the motility defect of CMF22/DRC11 knockdowns (Nguyen et al., 2013) is due to specific loss of the inter‐doublet connection component of NDRC, without affecting dyneins. This result is important, as dynein-independent connection between adjacent DMTs is a founding principle of the sliding filament model for axoneme motility (Satir, 1968; Viswanadha, Sale and Porter 1943; Holwill and Satir, 1990), yet direct tests of this idea have been limited because most NDRC mutants also disrupt dyneins. Moreover, the CMF22/DRC11 knockdown analysis that we conducted here has identified structural defects of NDRC mutants that were not previously known. We have revised the text of the manuscript (Discussion section) to clarify these points.

Regarding the goal of relating structures identified to function of the T. brucei axoneme: A critical requirement of the trypanosome axoneme is to maintain integrity in the face of physical strain imposed on the organism due to lateral attachment to the PFR and cell body, and its particular motility mechanism – both being features that are unique to trypanosomes. Two main discoveries in the current work are extensive MIPs and inter‐doublet connections that are likewise unique to trypanosomes. The complete repertoire of functions for MIPs and inter‐doublet connections remain to be determined for any organism. However, in the few cases where data is available, MIPs and inter‐doublet connections provide axoneme stability and modulate axoneme motility (Stoddard et al., 2018; Ichikawa and Bui, 2018; Owa et al., 2019; Bower et al., 2018; Bower et al., 2013). Therefore, we consider it reasonable to suggest that lineage‐specific MIPs and inter‐doublet connections in trypanosomes provide a potential explanation for how the organism maintains axoneme integrity in the face of unique physical constraints imposed by lineage‐specific flagellum architecture and motility. We have modified the text (Discussion section) to make this link between structure and function more clear.

Reviewer #2:

This manuscript reports a high-quality data set on the 3D organization of the so-called 96-nm repeat structures in two developmental stages of the protist Trypanosoma brucei. Flagella were purified and analysed by cryo electron tomography and sub-tomogram averaging after plunge freezing. The resolution reached remarkable values (12-15 Å) for the bloodstream stage. In some cases, tubulin monomers could be resolved at the level of protofilaments. It revealed common features with other axonemes, especially from human cilia but also unique differences. Nice structural details of the internal composition of microtubules are reported, including the unique ponticulus and the discovery of novel microtubule internal proteins (MIP). Comparison with flagella of cells where the expression of a component of the nexin-dynein regulatory complex was knocked down revealed structural modifications. The proximal lobe of the NDRC complex appeared affected, what impaired the connection with the neighbouring doublet and could explain the motility defect.

This is a very nice study, with some new developments (MATLAB script) and will be of interest for cilia biologists and structural biologists at large.

We appreciate the positive comments regarding high quality of the data and recognition of the value of the contributions to a diverse research community.

Three points need to be clarified to fully understand the results:

1) Exponentially growing cultures of trypanosomes contain about 50% of cells with two flagella (the mature one and the growing one). Based on length distribution of bloodstream form flagella (Figure 2), it seems that mostly full-length flagella are present in the sample. How is it possible? Is it due to the purification procedure that would select long flagella? This is important because at least the distal end of growing and mature flagella are known to be different in terms of structure (Höög et al., 2014) or composition (Subota et al., 2014). Please clarify. Length measurements for procyclic form flagella are missing.

Thank you for pointing this out. The samples do appear to be somewhat enriched for full‐length flagella, although if 50% of the population were to have a mature flagellum and 50% have mature plus a growing flagellum, that would yield ~1/3 growing flagella, some of which may be quite close to full length. In that case, the distribution observed is not far from expected. It’s possible that the purification procedure preferentially yields long flagella, though we haven’t directly assessed that. In our experience, growing flagella tend to stay connected to the mature flagellum and if connection and overlap between filaments is too great, one cannot reliably trace each single flagellum filament, so it is possible that among the filaments that can be reliably measured, there are more mature flagella than growing flagella. In any case, the samples contain intact flagella with good preservation of structure. Regarding the variation at the distal end, this is why we avoided the distal end of the flagella for structure determination here. For PCF samples, length of flagella was not measured, but visual inspection indicated fairly uniform length distribution.

2) A recent paper (Edwards et al., 2018) showed that the docking of outer dynein arms was different along the length of the trypanosome flagellum (in PCF), with specific proximal and distal docking complexes. Subsection “3D Structure of the trypanosome 96-nm axonemal repeat”: "tilt series were collected from the center part of the flagellum, spanning the middle third between the basal body and tip". A cartoon would help to avoid ambiguity but if we understood correctly, it seems that the central portion of the flagellum was used. It is where the two docking systems are likely to overlap, hence potentially generating heterogeneity. Is the organisation the same in the short portion without paraflagellar rod (PFR)?

As requested, we’ve indicated in Figure 2 the approximate region imaged. As detailed above (response to Editor comment #2 above), we were unable to obtain a high‐resolution structure for the proximal‐most portion of the axoneme without the PFR due to extensive twisting in this region. We recognize that our analysis does not resolve differences proximal to distal in the axoneme, which is why we specify that the middle third of the axoneme is imaged.

3) Which doublets (out of the 9) were selected for analysis? The same group has shown that doublets were not equivalent (Hughes et al., 2012), especially those connected to the PFR where dynein arms looked different.

All nine doublets were averaged for the overall structure of BSF and PCF samples. Comparisons of the 96‐nm repeat in individual DMTs and PFR‐attached versus not‐attached DMTs has now been done as requested by reviewers and results described in the text (subsection “Doublet-specific features of the 96-nm repeat”). We do not see a change in the 96‐nm repeating unit that correlates specifically with presence or absence of PFR. (See also author response to Editor comment 1 and 2.) Note that in the earlier paper referred to, Hughes et al., 2012, neither the dyneins nor doublets were reported to be different, per se, but filaments from the PFR could be seen to contact dyneins. Periodicity of PFR attachment is 56 nm (Koyfman at al., 2011; Hughes et al., 2012) and this differs from the 96‐nm repeat used for averaging. Therefore, attachment complexes would not be visible in our structure, which averages a 96‐nm unit.

The Title and the interpretation should be tuned down, although unique structural features (especially MIPs) are indeed reported, there is little direct evidence that they contribute to the original axoneme motility.

As requested, we have adjusted the Title, toned down interpretations and transferred speculation to Discussion section.

Writing-up. There is a lot of interpretation in the Result section. The authors should either remove these and do a more exhaustive Discussion section, or write the paper with Results section and Discussion section combined.

As requested, we have moved interpretation into the Discussion section.

Reviewer #3:

This study by Imhof et al., presents the first structures of the Trypanosome axoneme. Although cilia/flagella are found throughout evolution, structures are only available for a select few species. This structure from T. brucei is the first from the excavata supergroup, providing an important evolutionary comparison. The authors present several interesting findings, including the descriptions of only two dynein motors per OAD, increased connections between doublet microtubules (DMTs), and several lineage-specific microtubule inner proteins (MIPs). They also use an RNAi knockdown line to determine the position of DRC11. In principle, I support publication of this work, and I think the data is a valuable addition to the axoneme field. However, there are several major points related to the analysis that must be addressed, in particular related to resolution estimation and exploring structural variation between different DMTs. In addition, there is too much pure speculation about the potential functions of several structures, with no experimental evidence to support these functions. This speculation should be removed or heavily qualified.

We appreciate the positive comments and recognition of the interest and value of the work.

Essential revisions:

1) The authors heavily promote the resolution of their structure, claiming "This resolution is the highest cryoET structure yet reported for the 96-nm repeat from any organism". However, there are some serious issues with the resolution estimation, and as a result, I believe the resolution has been overstated.

We have taken this criticism very seriously and have now used the standard Fourier shell correlation calculation server at EMDB to recalculate the resolution of two half maps. The overall resolution reported by the server is 21.8 Å while the resolution of the doublet microtubule with MIPS is 19.0 Å (Figure 3—figure supplement 2). The details are provided below and in the revised text (subsection “3D Structure of the trypanosome 96-nm axonemal repeat”; Subsection “Data processing”). The reviewer is correct that Resmap local resolution provided more optimistic estimation than the standard FSC evaluation. The original Resmap was calculated using the combined map with all particles, which should in theory have better resolution than half maps (i.e., maps from half of the full dataset). Given the recommendations of the reviewer (below), we have also performed local FSCs of specific regions of the structure as requested. The text of the manuscript (subsection “3D Structure of the trypanosome 96-nm axonemal repeat”; Subsection “Data processing”) is revised and Figure 3—figure supplement 2 is made to reflect the updated resolution calculations.

The paper's Abstract claims that the DMT is resolved to 12Å. This is based on the ResMap analysis shown in Supplemental Figure S3. However, this analysis is troubling. First of all, the resolutions on this map appear to primarily range from 12-16Å, whereas the FSC curve shows a global resolution of 21Å at the 0.143 cutoff. If all the local resolutions from ResMap are averaged, the result should be close to 21Å. However, it does not appear that this will be the case, and instead ResMap is estimating resolutions that are at least 5Å better than the FSC. Even the less well-resolved appendage structures that can be seen in this image range from 15-18Å. This is a bit difficult to judge from the figure because the authors have intentionally only shown the backside of the DMT (which reports the highest resolution). They must also show the other side, with all the important accessory structures (similar to Figure 3D), as well as a cross-section slice through the DMT to show the MIPs (similar to Figure 3B). Only then can we see the range of resolutions estimated by ResMap. But even in the view that is shown, the appearance on the DMT is way too speckled, with a huge dynamic range of 12-16Å on the microtubule wall. This noisy surface is a clear sign that something is wrong with the ResMap-the surfaces should look much smoother, with less hotspots of resolution variation.

As the ResMap cannot be taken at face-value, a parallel approach should be attempted to estimate local resolution. I recommend using masks to perform two local FSCs – the doublet region and the region containing the appendage structures (OAD, IAD, RS, NRDC). How do these FSCs compare to the global FSC and to the ResMap?

The reviewer has a good point that ResMap typically reports more optimistic resolution numbers. To address this, we have done local FSC calculations for sub‐regions as requested and the resolutions based on these new calculations are included in the revised manuscript subsection “3D Structure of the trypanosome 96-nm axonemal repeat”; Subsection “Data processing” and Figure 3—figure supplement2. We have also provided ResMap views from other angles, as well as a section through the DMT as requested (Figure 3—figure supplement 2).

For the isosurface renderings throughout the paper, the DMT looks properly filtered, but the appendage structures appear to be oversharpened or displayed at a resolution that is too high (their surfaces look noisy). I assume that these maps were uniformly filtered to the same resolution. Was it the global 21Å? I can't seem to find this information in the paper. I would expect that by around 20Å, the holes in the middle of the dynein AAA+ rings would start to become visible, at least as indentations. But in these maps, the AAA+ rings look like round egg-shaped blobs, another sign that the resolution is not as high as claimed.

The reviewer makes good points and we appreciate their evaluation and comments. Our new FSC calculation yields an average resolution of the entire BSF axoneme structure of 21.8 Å based on the 0.143 Fourier shell correlation criterion (Figure 3—figure supplement 2A). The resolutions at different regions vary based on visual inspection, and assessments by both local FSC and ResMap (34) calculations (Figure 3—figure supplement 2); the resolution of DMT region with MIPs reached 19.0 Å based on local FSC calculation(Figure 3—figure supplement 2A).The resolution of IAD is 36.1 Å which explains why a hole was not obvious.

We did not sharpen the sub‐tomogram average maps. For rendering, no filters were applied on MIPS but we applied low pass filters on the other components to improve clarity of individual structures described in the text ‐ for the structures in Figure 3C‐E; Figure 4A, B, D; Figure 7A‐E, we filtered the Microtubule, NDRC, RS, IC/LC, OAD and IAD to 30Å and for the structures in Figure 5; Figure 6; Figure 6—figure supplement 1, we filtered the map to 50Å). This information is added to the subsection “3D visualization”MATERIALS AND METHODS.

The authors used a tilt-series acquisition scheme that starts at -60 degrees and thus destroys the high-resolution information before reaching low tilt, as opposed to the much preferred dose-symmetric scheme starting at zero degrees combined with dose-weighting (see the high-resolution HIV work by John Briggs and Wim Hagen). Furthermore, there currently is no way to correct the contrast transfer function (CTF) for low dose cryo-tilt series acquired with the Volta Phase Plate. The authors thus did not perform CTF-correction, meaning that resolution of the average is limited to the first zero of the power spectrum. Given the -0.6 μm target defocus and the large defocus gradient that is present in thick samples such as these axonemes, especially at higher tilts, I anticipate that this first zero would strongly limit the resolution (20-25Å sounds about right, not 12Å). Therefore, I am very cautious of the bold resolution claims made in this paper.

The reviewer’s concern is well taken. We believe this might be because we had omitted details concerning our data processing particularly the part about CTF correction. This has been corrected by adding a detailed paragraph detailing how we performed our reconstruction with CTF correction (Subsection “Data processing” in the revised manuscript). We used in‐focus (~‐0.6um) imaging with VPP phase plate and we always targeted axoneme along the tilt axis. At high tilt angle, the defocus could reach 112nm farther from the targeted ‐0.6um defocus [i.e., from ‐0.49 to ‐0.71um. Note, one of the benefits of using a phase plate is that the CTF is insensitive to the sign of the defocus value being negative (underfocus) or positive (overfocus) (20). We recognized that during imaging, defocus determination is also challenging for VPP data, which might have limited achievable resolution of our results. Indeed, as discussed in responses above, the FSC‐based resolution of our entire 96nm repeat is 21.8 Å with that of the DMT plus MIPS region being 19.0 Å (See Figure 3—figure supplement 2A in the revised manuscript).

Finally, the authors use a "gold-standard" FSC to determine resolution, but it was not clear to me from the methods when exactly the extracted particles were split into two half-sets and averaged independently, as is required for gold-standard assessment. With only 700 total particles, getting two half-sets to 21Å might be challenging. Please explicitly describe how the averaging was performed instead of just "as described previously (75) using PEET".

We have now provided the details of averaging (Subsection “Data processing”) and FSC calculations (Subsection “Data processing”) in the revised manuscript

2) The focus of this paper is to show the evolutionary differences of the T. brucei axoneme. Of course, by far the most distinct feature of trypanosome flagella is the paraflagellar rod (PRF). This structure seems important for axoneme stability under strong forces, a key question the authors sought to address. In Figure 2G, the authors show clear periodic connections between the axoneme and PFR (marked with arrows). It would be very valuable to compare averages of PFR-linked versus non-PRF-linked DMTs. This should be a relatively easy task, just splitting particles into those two categories based on their location with respect to the PFR. Furthermore, the PFR appears to have a fairly regular structure in Figure 2G, so is it possible to generate an average of the PFR itself? Such a structure would be something really new, and would add value to this paper.

As requested, we have now examined individual DMTs and compared PFR‐attached versus not PFR‐attached DMTs – for details, please see the response to Editor comments 1 and 2 above and text of the revised manuscript (subsection “Doublet-specific features of the 96-nm repeat”). Regarding structure of the PFR itself, as discussed in response to Editor comment 2, we feel determining 3D structure of the PFR is beyond the scope of the current paper.

3) The globular density with an 8-nm repeat in Figure 3—figure supplement 3 has a weak signal, and thus appears to have low occupancy in this average. Why is this? Might it have something to do with the connection to the PFR? Judging by Figure 3A, one would expect about a third or a fourth of the DMTs to have a connection to the PFR. The authors should investigate how the occupancy of this density varies between DMTs by producing averages of different DMTs using their radial position relative to the PFR for reference. Without more information, the author's proposed function of this density in regulating dynein binding (subsection “3D Structure of the trypanosome 96-nm axonemal repeat”) is far too speculative.

The density is between protofilaments 7 and 8 of the B‐tubule, so it is not in the correct position to provide contact site for PFR. We agree that the weak signal suggests low occupancy, though we did not resolve a structure here on specific DMTs and thus cannot say whether or not it is specific to a subset of DMTs. Given that this structure is present at or near the site of OAD‐α binding, between protofilaments 7 and 8 in T. brucei (Figure 4E), and in Chlamydomonas (17), we think it is reasonable to suggest that it may influence dynein binding. We’ve removed speculation regarding impact on IFT motors.

4) Based on the strong signal of the IAD-f IC/LC density, the authors conclude: "An IAD-f IC/LC interdoublet connection is observed between three specific doublet pairs in Chlamydomonas (17). However, the prominence of the IAD-f IC/LC connection in T. brucei suggests it is present between most and perhaps all doublet microtubules, indicating that nexin links in T. brucei include both the NDRC and IAD-f IC/LC. This distinguishes T. brucei axonemes from the known 3D axoneme structures from other organisms (36, 46)." Similar to points 2 and 3 above, if the authors want to make this claim, then they absolutely must examine the density in averages of different DMTs around the T. brucei axoneme.

As requested, we have examined the structure of individual doublet microtubules (see response to Editor comment 1 and 2). The results indicate the f‐connector is present on most doublets. DMT2 is the one exception ‐ it does not have a contiguous connection, although the DMT 2‐3 interface does have a density corresponding to the site of connector attachment on the B‐tubule of DMT 3. This information is now provided in the text (Results section, Figure 6—figure supplement 1Figure).

5) The issues raised in the four points above (problems with resolution estimation and neglecting to analyze structural variation between different DMTs) are all related to the very limited dataset used in this paper. The primary "high resolution" structure in this paper (wild-type BSF) was generated from only 700 particles from 10 tomograms. This is only half a day of acquisition on a Titan Krios microscope (tomograms take about 1 hour each). While I understand it is not eLife policy to ask for more experiments, I think it is completely appropriate in this case for the authors to spend one more day on the microscope with their already prepared cryo-EM samples to acquire 20 more tomograms. This would produce a sufficient dataset to perform classification and look at how specific densities vary between different DMTs (see points 2-4 above). I understand that the Volta Phase Plate used in this study enables averages to be generated with less particles, but I think the 700 particles in the current average are too few to properly do this analysis, and it is not clear to me why the authors chose to proceed with such a limited dataset.

As requested, we have now added additional BSF tomograms to enable analysis of individual DMTs. (For details, see response to Editor comment 1, and reviewer 3 comments 2‐4 above.) For the BSF sample, we obtained a total of 50 tomograms, but selected only the 10 judged as best, based on minimal compression of the axoneme (see Figure 3—figure supplement 1), for sub‐tomogram averaging of all DMTs. As requested by the reviewer, we have now used an additional 24 tomograms (34 in total) to perform sub‐tomogram averaging of individual DMTs. (See Results section, Materials and methods section, Figure 6 and Figure 6—figure supplement 1).

Regarding resolution and number of particles averaged, we point out that the resolution achieved in our work for the averaged BSF 96‐nm repeat (21.8 Å overall resolution at 0.143 FSC criterion, with 763 particles averaged) is within range of structures reported in recent cryoET analyses of axonemes, e.g.:

Stoddard et al., 2018.

Bower et al., 2018.

Lin and Nicastro, 2018.

Dymek et al., 2019.

Owa et al., 2019.

6) The proximal and distal holes in the DMTs look convincing. However, what is not convincing to me is their proposed function of allowing MIPs to be incorporated after completion of the DMT. The holes are tiny, only about as wide as a tubulin monomer (4 nm), and thus do not seem big enough to allow the free transit of MIPs, which are significantly larger than the holes. Perhaps the holes could serve as a location for the start of an "unzipping" event between the A- and B-tubules, which could allow insertion of larger MIP structures such as the ponticulus, but this is completely speculative. I don't think much can be said about the function of the DMT holes at this point.

We have removed speculation on function of holes.

7) The extended discussion of the RingMIP function is highly speculative and should perhaps be omitted or at least down-weighted. Its proposed mechanosensory role is not proven by the data in this paper, just speculated.

We have removed speculation.

8) The authors should be careful with stating the significance of the comparison between developmental stages; there's no clear conclusion from this comparison (other than a low-resolution hint of MIPA3-4). So, putting this in the abstract without indicating the negative result could be considered false advertising. Upon reading the Abstract, I assumed that there were developmental differences, and I was disappointed when I finally discovered at the end of the paper that there were not.

We agree and have removed this comment.

Reviewer #4:

Imhof, Zhang et al., present cryo-electron tomography data on preparations of isolated axonemes from the unicellular parasite Trypanosoma brucei, eukaryotic supergroup Excavata. Specifically, they provide a detailed analysis of the 96nm repeat that forms the core structural unit of the axoneme. These data were obtained from two life cycle stages of the organism – the slender bloodstream form which is found in infected mammalian hosts and whose motility is better characterised, and also the procyclic form which is found in the midgut of the tsetse fly vector. Not only are the usual structural features such as the outer dynein arms, radial spokes resolved, but also some fascinating observations of the microtubule inner proteins are provided, which appear to be considerably more abundant in the axonemes of this organism than in others imaged to date.

Given that the axoneme is an almost ubiquitous structure that was present in the last eukaryotic common ancestor, the data here are extremely significant and of relevance far beyond the trypanosome and parasitology community. As the authors note, taxonomic sampling of the ultrastructure of the 96nm axonemal repeat is limited, and by providing data from the supergroup Excavata the authors have considerably broadened the perspective onto this structure. The observations of the microtubule inner protein complexes are fascinating and could set the stage for considerable future work.

It is important to note that I have no first-hand practical experience of cryo-electron tomography and am therefore not qualified to judge the technical aspects of the manuscript relating to this technique. I am happy for the opinions of the other referee(s) to take precedence on this point.

The data appear to be of very high quality and the figures do not require more than cosmetic alteration. No extra practical work seems required, I think, and I have no major concerns. The manuscript could be considerably tightened in order to do full justice to the quality of the results, however.

We appreciate the positive comments and recognition by the reviewer of the significance and relevance of the work to a broad readership.

In particular, the Introduction needs much more detail. The authors should define the 96nm axonemal repeat and explicitly summarise previous work on this structural unit. This is important for placing the results in context and defining the paper's original contribution. The 96nm repeat is currently first mentioned in Results section without any prior introduction. There has been substantial previous work on the morphology and ultrastructure of the trypanosome axoneme, so it is important to emphasise that this work focuses on the structure of the axoneme's 96nm repeat unit specifically. Presently there is considerable potential for being misread, and means that the authors are actually underselling what they have.

The reviewer raises a good point that our focus is the 96‐nm axonemal repeat. We have updated the text throughout to be clear on this. We also explicitly define the 96‐nm repeat in the Introduction based on previous work as well as add a new figure that directly compares the T. brucei 96nm repeat to that obtained by cyroET analysis of axonemes from other organisms (Figure 5).

The manuscript also frequently rushes over the results. Panels 2A-2G are not individually cited, nor are 3A-3E, and this trend continues with the other figures. Ensure that all panels are cited, and ideally in figure order (i.e. A-B-C etc) for clarity. Figure 10 is currently not cited at all in the manuscript text. Not all the supplemental figures and movies appear to be cited in the manuscript text – this needs checking. The section on the comparison of the different developmental stages (currently Figure 9) could perhaps be moved so that it comes after Figure 4 instead.

Thank you for pointing out these oversights on our part. We have overhauled the manuscript text to improve clarity and have cited each figure. We have referred to specific figure panels when relevant.

The text would benefit from some proofreading for English, and also for style (e.g. Results should always be presented in past tense, but the present is often used here). Check also that all panel citations are correct – there is some mix-up, particular of Figure 5 and Figure 6. There is also perhaps a bit too much interpretation in the Results section that would fit better in the Discussion section.

The manuscript text has been revised to tighten and to ensure correct in‐text figure references.

https://doi.org/10.7554/eLife.52058.sa2

Article and author information

Author details

  1. Simon Imhof

    Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Jiayan Zhang
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7514-6811
  2. Jiayan Zhang

    1. Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    2. Molecular Biology Institute, University of California, Los Angeles, Los Angeles, United States
    3. California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Contributed equally with
    Simon Imhof
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3602-1199
  3. Hui Wang

    1. Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    2. California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    3. Department of Bioengineering, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9922-7170
  4. Khanh Huy Bui

    Department of Anatomy and Cell Biology, McGill University, Montreal, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2814-9889
  5. Hoangkim Nguyen

    Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    Present address
    Teva Pharmaceuticals, California, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Ivo Atanasov

    California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Wong H Hui

    California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  8. Shun Kai Yang

    Department of Anatomy and Cell Biology, McGill University, Montreal, United States
    Contribution
    Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Z Hong Zhou

    1. Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    2. Molecular Biology Institute, University of California, Los Angeles, Los Angeles, United States
    3. California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    4. Department of Bioengineering, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    Hong.Zhou@UCLA.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8373-4717
  10. Kent L Hill

    1. Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, Los Angeles, United States
    2. Molecular Biology Institute, University of California, Los Angeles, Los Angeles, United States
    3. California NanoSystems Institute, University of California, Los Angeles, Los Angeles, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    kenthill@microbio.ucla.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6529-1273

Funding

Swiss National Science Foundation (P300PA_174358)

  • Simon Imhof

National Institutes of Health (R01GM071940)

  • Jiayan Zhang
  • Hui Wang
  • Ivo Atanasov
  • Wong H Hui
  • Z Hong Zhou

National Institutes of Health (S10RR23057)

  • Z Hong Zhou

National Science Foundation (DMR-1548924)

  • Z Hong Zhou

National Institutes of Health (AI052348)

  • Simon Imhof
  • Hoangkim Nguyen
  • Kent L Hill

Swiss National Science Foundation (P2BEP3_162094)

  • Simon Imhof

National Institutes of Health (S10OD018111)

  • Z Hong Zhou

National Institutes of Health (U24GM116792)

  • Z Hong Zhou

National Science Foundation (DBI-1338135)

  • Z Hong Zhou

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Changlu Tao for technical assistance in SerialEM operation, Robert Minahan and Masahiro Yabe for help in data processing, Neville Kisalu and Michelle Shimogawa for motility videos of BSF cells. We thank Michelle Shimogawa for critical reading of the manuscript. This research has been supported in part by grants from NIH (R01GM071940, AI052348). SNF (P300PA_174358 and P2BEP3_162094). HK was supported by NIH-NRSA fellowship GM007185.We acknowledge the use of instruments in the Electron Imaging Center for Nanomachines supported by UCLA and grants from NIH (S10RR23057, S10OD018111 and U24GM116792) and NSF (DMR-1548924 and DBI-1338135).

Senior Editor

  1. John Kuriyan, University of California, Berkeley, United States

Reviewing Editor

  1. Andrew P Carter, MRC Laboratory of Molecular Biology, United Kingdom

Reviewers

  1. Benjamin D Engel, Max Planck Institute of Biochemistry, Germany
  2. Brooke Morriswood, University of Wuerzburg, Germany

Publication history

  1. Received: September 20, 2019
  2. Accepted: November 11, 2019
  3. Accepted Manuscript published: November 11, 2019 (version 1)
  4. Accepted Manuscript updated: November 14, 2019 (version 2)
  5. Version of Record published: January 21, 2020 (version 3)

Copyright

© 2019, Imhof et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,066
    Page views
  • 429
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)