Introduction

Dyneins are microtubule (MT)-based molecular motors that perform diverse biological functions (Roberts et al. 2013). They work as large multiple bio-nanomachines (> 1 MDa) consisting of a heavy chain (HC), intermediate chain (IC), light intermediate chain (LIC), and light chain (LC). The HC possesses ATPase activity and provides the driving force for power generation to conduct a wide variety of cellular functions. The other chains are involved in regulating the HC and are therefore called accessary chains. Dyneins are classified as either cytoplasmic or axonemal on the basis of their physiological function and cellular localization, and there is a clear distinction between the two types. Cytoplasmic dyneins serve as power generators for migration and intracellular transport, whereas axonemal dyneins are located in the axoneme and are responsible for ciliary/flagellar beating. Complete genome analysis of several organisms has revealed that there are at least 15 HC genes present in most organisms; two of which encode cytoplasmic dyneins, while the others encode axonemal ones.

The structural analysis of dynein motors began with the LCs with relatively smaller molecular weights and then progressed to the larger chains, such as ICs and HCs (Table 1). The first HC structure to be published was that of dynein-c (Burgess et al. 2003), which is an isoform of an axonemal dynein purified from a green alga, Chlamydomonas reinhardtii. Although the resolution of the negatively stained electron microscopy (EM) images was not sufficient to build the atomic coordinates, the first model of the dynein power stroke was proposed from the two different EM structures with and without nucleotide. However high-resolution structural information on the HC, which is crucial to understand the molecular mechanism of the mechano-chemical coupling of dynein motors, was long awaited.

Table 1 Structures of cytoplasmic dynein available in the PDB
Full size table

Since the establishment of a method for producing functional recombinant cytoplasmic dynein motor domains (Nishiura et al. 2004), X-ray crystallographic high-resolution structures have been reported that describe detailed structural elements, such as the N-terminal linker, AAA+ (ATPases associated with various cellular activities) ring, stalk/strut coiled coils with the microtubule binding domain (MTBD), and the C-terminal non-AAA structure named “C-sequence,” as well as revealing the structural changes that occur upon ATP hydrolysis (Carter et al. 2008, 2011, Kon et al. 2011, 2012; Schmidt et al. 2012). Most recently, a single-particle cryogenic electron microscopy (cryo-EM) structure of a vast complex of cytoplasmic dynein 1 bound to dynactin and an N-terminal construct of BICD2 (total molecular mass, 1.4 MDa) has been reported (Zhang et al. 2017).

To date, many structures of cytoplasmic dynein components from different organisms have been analyzed by X-ray crystallography, cryo-EM, and NMR spectroscopy (Tables 1 and 2). As the number of solved structures of cytoplasmic dynein increases year by year, the information is becoming too complex to assess the accumulated structural data at a glance (Table 1). By contrast, only four atomic structures of axonemal dynein are available in the Protein Data Bank (Table 2) (Mullen et al. 2000; Wu et al. 2003; Kato et al. 2014).

Table 2 Structures of axonemal dynein available in the PDB
Full size table

In this short article, we review representative structural studies of the components that classify cytoplasmic and axonemal dyneins and summarize them as a structural atlas (Fig. 1) with additional updated structural data on LC1 from C. reinhardtii obtained by ourselves (Fig. 2).

Fig. 1
figure 1

Structural atlas of cytoplasmic dynein. a Schematic diagram of the cytoplasmic dynein complex. b Superposition of solved dynein structures on the schematic diagram. Atomic structures at a resolution of 4 Å or higher are shown. From left to right, pre-power stroke (dark gray) and post-power stroke (light gray) structures of the dynein motor domain, LIS1 (red), LIC (green), IC (navy) with Robl (cyan), LC8 (orange), and TcTex (yellow) are shown

Full size image
Fig. 2
figure 2

Currently available atomic structures of axonemal dynein. a Crystal structure of LC1 (PDB ID: 5YXM). LC1 crystals were grown at 4 °C via the sitting-drop vapor diffusion method by mixing 200 nL of LC1 (20 mg/mL protein) with an equal volume of reservoir solution (0.1 M ammonium phosphate monobasic, 10% (w/v) PEG3,350). LC1 crystals were soaked in cryo-protectant solution (0.1 M ammonium phosphate monobasic, 35% (w/v) PEG3,350, 10 mM Tris-HCl (pH 8.0), 100 mM NaCl) overnight, and then flash-cooled in liquid nitrogen. The X-ray diffraction experiment was performed on beamline BL44XU, SPring-8, Harima Japan. The collected images were processed by using HKL2000 software (Otwinowski and Minor 1997). Molecular replacement and refinement were performed by using Phenix (Adams et al. 2002) and COOT (Emsley and Cowtan 2004). TLS parameters were analyzed by using the TLSMD server (Painter and Merritt 2006), and 12 TLS groups were introduced in the subsequent refinement. The final structure was validated by using MolProbity (Lovell et al. 2003). The detailed crystallographic statistics information can be available in the PDB (https://pdbj.org/mine/summary/5yxm). The ribbon diagram of LC1 is shown in green. b Superposition of the X-ray (green) and NMR (magenta) structures of LC1. A representative NMR structure is shown (PDB ID: 1M9L). c Amino acid sequence alignment of LC1 with secondary structure assignments. d Superposition of the X-ray structure with anisotropic B factors and main chain conformation of the NMR structure (magenta). e NMR structure of the MTBD of dynein-c (PDB ID: 2RR7). The additional flap structure which is an insertion sequence in the MTBD of the axonemal dynein is shown in orange

Full size image

Light chain

Many structures of LCs from different organisms have been reported, including LC7, LC8, TcTex-1, and Lis1 (Table 1). It is thought that the LCs are important for dynein–cargo interactions. Several structures of LC8 (also called DYNLL or dynein light chain 1) have been determined as a complex with peptides derived from binding partners by X-ray crystallography and NMR (Table 1, Fig. 1). Details of the molecular function of LC8 remain unknown, but Chlamydomonas cells of a LC8 deletion mutant lack retrograde intraflagellar transport and display short deficient flagella (Pazour et al. 1998).

Axonemal dynein light chain-1 (LC1) in C. reinhardtii (DNAL1 in Homo sapiens), whose structure has been solved by NMR spectroscopy (Mullen et al. 2000; Wu et al. 2003) (Fig. 2), is a component of outer arm dynein (OAD) (Table 2). Knockdown of LC1 has been found to reduce beat frequency in the flatworm planarian (Rompolas et al. 2010), and the expression of an LC1 mutant shows dominant-negative effects on swimming velocity and beat frequency in C. reinhardtii (Patel-King and King 2009). These observations suggest that LC1 acts as a regulator to beat cilia/flagella. Originally, LC1 was thought to be directly bound to tubulins and to tether the OADγ HC to the microtubule (Patel-King and King 2009). Furthermore, it has been widely assumed that LC1 associates with AAA1 and AAA3 or AAA4 of the AAA+ ring in the gamma heavy chain of OAD (OADγ) in C. reinhardtii (Benashski et al. 1999). However, it was recently reported that LC1 is tightly bound to the MTBD of OADγ, which is located at the tip of stalk region in the motor domain (Ichikawa et al. 2015). This was the first report of an LC interacting with the MTBD. Moreover, it was also discovered that the binding of LC1 to the MTBD decreases the MT-binding affinity of the HC (Ichikawa et al. 2015). Because it has been reported that the ATPase activity of the HC is increased in the presence of MTs (Kon et al. 2009), both results imply that LC1 indirectly changes the ATPase activity of OADγ and regulates ciliary/flagellar beating. However, the molecular mechanism that tunes ATPase activity through the MTBD still remains poorly understood.

Although NMR structures of LC1 are available, we determined the X-ray structure of LC1 at 1.55-Å resolution to enable a more detailed discussion (Fig. 2a). As expected from a comparison of the amino acid sequences and NMR structures of LC1 (Benashski et al. 1999; Mullen et al. 2000; Wu et al. 2003), the crystal structure of LC1 shows a leucine-rich repeat conformation. However, there are large conformational differences between the X-ray and NMR structures, especially in the N- and C- terminal regions and the crystal structure differs from the NMR structures at the secondary structure level (Fig. 2b, c). In particular, the differences in the secondary structure between the X-ray and NMR structures are surprisingly large at Ala22–Glu24 and Met182–Val184 in the N-terminal and C-terminal regions, respectively (Fig. 2c). These results suggest that these two terminal regions may play the role of flexible hinges and that large conformational differences may be induced when LC binds to its partner proteins.

We also analyzed the anisotropic temperature factors of the X-ray structure with reference to the main chain conformation of the NMR structures (Fig. 2d). There were significant correlations between the anisotropic directions of the temperature factors and the structural differences between the X-ray and NMR structures, which implies that the intrinsic flexibility of LC1 is manifested in the structural discrepancy between X-ray and NMR structures.

Light intermediate chain

The LIC subunit is present in cytoplasmic dynein, but not in axonemal dynein (Inaba 2007). There are three LIC homologs in H. sapiens: LIC1, LIC2, and LIC3. On the one hand, LIC1 and LIC2 are associated with cytoplasmic dynein 1 (Hughes et al. 1995) and are thought to play important roles in cargo transport and stability of the HC (Trokter et al. 2012). On the other hand, LIC3 interacts with cytoplasmic dynein 2 (Grissom et al. 2002). Sequence analysis indicates that the LICs are divided into two domains: a conserved N-terminal domain and the other domain. The only known structure of LIC is the structure of the conserved N-terminal domain of LIC from a thermophilic hyphal fungus, Chaetomium thermophilum (Schroeder et al. 2014). Although the structure shows a Ras-like G-protein fold, the nucleotide pocket is empty. Biochemical experiments confirmed that this fungus LIC does not bind nucleotide, whereas human LIC1 does bind nucleotides (Schroeder et al. 2014). To clarify the differences in LICs by species and isoform, further structural studies and biochemical experiments will be needed.

Intermediate chain

There are four IC homologs in H. sapiens. DYNC1I1 (cytoplasmic IC1) and DYNC1I2 (cytoplasmic IC2) associate with cytoplasmic dynein HC, while DNAI1 (axonemal IC1) and DNAI2 (axonemal IC2) interact with axonemal HCs. All IC homologs possess a conserved WD40 domain in the C-terminus that interacts with HCs (Tynan et al. 2000). A secondary structure analysis using Jpred and RONN predicted that the N-terminal region of cytoplasmic IC1 and IC2 comprise coiled coil and highly disordered regions (Williams et al. 2012). However, the secondary structure prediction analysis also indicated that the N-terminal region of IC1 is highly disordered but that of IC2 possesses a folded structure in axonemal ICs (Williams et al. 2012).

In terms of atomic structure, so far, there is no high-resolution structure of either the whole IC or its WD40 domain (Tables 1 and 2). However, crystal structures of TcTex1 and LC8 with IC peptides containing the interaction sites for LCs have been determined and explain how IC interacts with LCs (Williams et al. 2007; Hall et al. 2009). Moreover, the whole structure of a cytoplasmic dynein complex determined by cryo-EM has revealed the structural arrangement of IC within the cytoplasmic dynein complex (Zhang et al. 2017). However, the precise site of the IC–HC interaction remains unclear due to the low resolution. Thus, more work is needed to gain structural insights into the dynein ICs.

Heavy chain

The HC is the largest polypeptide of the dynein complex and contains the motor domain, which is the minimum component needed for ATP-dependent motor activity. In 2008, Carter and colleagues determined the first crystal structure of the MTBD of an HC fused to the seryl–tRNA synthetase (SRS) from Thermus thermophilus (Carter et al. 2008). Several years later, a more complete structure of the stalk coiled coil with the MTBD was reported by our group (Nishikawa et al. 2014, 2016). Since then, many crystal structures of the cytoplasmic dynein motor domain in different nucleotide states have been determined (Carter et al. 2011; Kon et al. 2011, 2012, Schmidt et al. 2012, 2014; Bhabha et al. 2014). According to these structures, the HC is composed of multiple functional units, including the tail, linker, AAA+ ring, stalk/strut, and C-sequence. Each unit possesses distinct functions to drive force generation in the motor. In addition to X-ray crystallography, cryo-EM has more recently revealed the structure of a cytoplasmic dynein complex including the HC, IC, LIC, and LC, both alone and together with dynactin–BICD2 (DDB) (Zhang et al. 2017). The cryo-EM structures have revealed the relative arrangement of the cytoplasmic dynein components in an inhibitory state and provide insights into how cytoplasmic dynein is inhibited and activated.

In contrast to the genes encoding cytoplasmic dynein HCs, those encoding axonemal dyneins are many and diverse. The arrangement of the HC and the characteristics of the motor activity along the MT differ completely between axonemal dynein and cytoplasmic dynein. Cytoplasmic dyneins work as a dimer, whereas the functional oligomeric states of axonemal dynein HCs include monomers, dimers, and trimers. Moreover, an MT gliding assay has revealed that some axonemal dyneins display clockwise translocation of MTs (Kikushima and Kamiya 2008; Yamaguchi et al. 2015). These findings indicate that axonemal dyneins are highly diverse proteins in terms of the functional properties of their HCs. Among the axonemal dynein HCs, only the NMR structure of the MTBD of dynein-c from C. reinhardtii has been determined so far (Kato et al. 2014) (Fig. 2e). As compared with cytoplasmic dynein, the molecular mechanism underlying the motor activities of axonemal dyneins remains relatively unclear. Clearly, structural and functional studies of axonemal dynein HCs need to be addressed as soon as possible.

Conclusions and future prospects

Dynein motors are biologically important bio-nanomachines. In parallel with recent developments in structural biology, such as single-particle cryo-EM and synchrotron-based X-ray nano-crystallography, more and more fascinating three-dimensional structures of dyneins have become available. Based on the survey of the structures available at atomic resolutions shown above, we would like to point out two important directions for future research. One is the imbalance in structural information between cytoplasmic and axonemal dynein. Atomic data are very much focused on cytoplasmic dyneins and remarkably less structural work on axonemal dyneins has been reported. The oligomeric states of axonemal dyneins are so diverse that each dynein is likely to have a specific structural role. The expansion of structural information on axonemal dyneins is greatly anticipated.

The second point is that structures of dynein on MTs are lacking. This point is important because the dynein that walks along the MT is really the functional molecule. For the other cytoskeletal motors, kinesin and myosin, not only structures in different nucleotide states but also structures in complex with the α, β-tubulin dimer or actin filament have been reported. Two structures of the dynein MTBD and MT complex have been solved by cryo-EM using a helical averaging technique (Table 1), one of which was done by a collaborative team including one of the authors. However, the resolutions of the two structures are 9.7 and 8.2 Å, and only flexible docking based on the available crystal structures is applicable at those resolutions. In 2014, Imai and his colleagues reported the structure of dynein walking on MTs using engineered chimeric dynein construct (Imai et al. 2015), but the resolution is not high enough to discuss the structure at the residue level. We await with impatience a high-resolution structure of dynein walking on MTs.