Assembly and activation of dynein-dynactin by the cargo adaptor protein Hook3

Abstract

Metazoan cytoplasmic dynein moves processively along microtubules with the aid of dynactin and an adaptor protein that joins dynein and dynactin into a stable ternary complex. Here, we examined how Hook3, a cargo adaptor involved in Golgi and endosome transport, forms a motile dynein-dynactin complex. We show that the conserved Hook domain interacts directly with the dynein light intermediate chain 1 (LIC1). By solving the crystal structure of the Hook domain and using structure-based mutagenesis, we identify two conserved surface residues that are each critical for LIC1 binding. Hook proteins with mutations in these residues fail to form a stable dynein-dynactin complex, revealing a crucial role for LIC1 in this interaction. We also identify a region of Hook3 specifically required for an allosteric activation of processive motility. Our work reveals the structural details of Hook3's interaction with dynein and offers insight into how cargo adaptors form processive dynein-dynactin motor complexes.

Figure 1.

Dynein LIC binds the Hook domain. (A) The domain architectures of human LIC1 and Hook3. See text for details. (B) GST-tagged human full-length or truncated LIC1 bound to glutathione resin were incubated with sfGFP-tagged Hook3_1–552, centrifuged and probed for Hook3 with an anti-GFP antibody. LIC1 in the pelleted beads was detected using an anti-GST antibody. Negative (Neg) control lacks LIC1 on the beads. (C) Ratio of band intensity to the full-length LIC1 signal in B; mean and SD from n = 3 independent experiments. (D) Two sfGFP-tagged Hook3 constructs were tested for LIC1 binding using the assay described in B. Also tested was a Hook domain artificially dimerized using a GCN4 sequence (Hook3_1–160-GCN4). (E) The ratio of band intensity to the Hook3_1–160 signal in B; mean and SD from n = 3 independent experiments.

Figure 2.

The structure of the Hook domain exhibits an extended α-helix and restricted conservation. (A) The 1.7-Å structure of the Hook domain (aa 9–158) from human Hook3 with the helices labeled A–H. Colors (helices A–G) denote the canonical CH domain. (B) The conservation of residues on the surface of the structure in A is shown with red representing the most conserved and white depicting the least conserved. Highly conserved residues are labeled.

Figure 3.

Helix H in the Hook domain contains a LIC-binding interface. (A) Patches of conserved residues in the Hook domain were mutated in separate constructs. Each patch of residues is denoted by a different color. (B) GST-LIC1_389–523, bound to glutathione resin, was incubated with sfGFP-Hook3_1–239 mutants. The beads were centrifuged, and then Hook3 binding was assessed by immunoblot analysis using an anti-GFP antibody. The presence of the bait GST-LIC1_389–523 was verified using an anti-GST antibody. Negative control lacks LIC1 on the beads. (C) Ratio of band intensity to the WT Hook3_1–239 signal in B; mean and SD from n = 3 independent experiments.

Figure 4.

Two conserved Hook3 residues are critical for the assembly and motility of dynein–dynactin. (A) Single-point mutations Q147A, M151A, and I154A in sfGFP-tagged Hook3_1–552 were compared with WT and tested for binding to human GST-LIC1_389–523 as in Fig. 3 B (representative of triplicate experiments). Negative control lacks LIC1 on the beads. (B) Ratio of band intensity to the WT Hook3_1–552 signal in A; mean and SD from n = 3 independent experiments. (C) StrepII-Hook3 constructs, bound to Strep-Tactin resin, were incubated with porcine brain lysate; the beads were centrifuged; and the resin analyzed by immunoblotting for the dynein intermediate chain (IC) and the p150 subunit of dynactin. Negative control lacks Hook3 on the beads. The amount of each Hook3 construct was assessed by Coomassie stain. (D) Ratio of band intensity to the WT Hook3_1–552 signal in C; mean and SD from n = 3 independent experiments. (E) WT and single-point mutants were incubated with affinity-purified human dynein–dynactin and 1 mM ATP. SfGFP-tagged Hook3_1–552 was visualized by TIRF microscopy and classified as processive if it moved unidirectionally for >1 µm along microtubules. All constructs were normalized by dividing the total number of processive motors by the total length of microtubules in the field of view and the time of the movie (movements/µm per min). The ratios of the mutants to WT were calculated from side-by-side experiments performed on the same day. Shown are the mean ± SD of the ratios from three independent experiments performed on different days. The mean number of motile WT Hook3 molecules/µm per min was 0.039 ± 0.016. (F) Representative kymographs are shown for each construct that displayed motility. The kymographs are displayed using the same brightness and contrast.

Figure 5.

Hook3 truncations that bind dynein–dynactin are not sufficient for motility. (A) Truncations of strepII-Hook3 were tested for binding to endogenous dynein–dynactin in porcine brain lysate as in Fig. 4 C. (B) The ratio of band intensity to the WT Hook3_1–552 signal in A; mean and SD from n = 3 independent experiments. The truncations not shown were measured to be the same as or less than the signal of the negative control. (C) C-terminal strepII-Hook3 truncations were tested for binding porcine brain dynein–dynactin as in A. The intermediate chain (IC) band in the lane for Hook3_1–402 is skewed because the IC and this Hook truncation run at the same molecular weight. (D) The ratio of band intensity to the WT Hook3_1–552 signal in C; mean and SD from n = 3 independent experiments. The following p-values are given for the truncations that differ statistically from Hook3_1–552: dynactin signal–Hook3_1–239, P < 0.0001; Hook3_1–310, P < 0.001; Hook3_1–348, P < 0.0001; Hook3_1–402, P < 0.0001; Hook3_1–440, P < 0.001; IC signal–Hook3_1–239, P = 0.03; and *Hook3_1–402, P < 0.05; *, The IC signal is disrupted by the similar size of Hook3_1–402. (E) Hook3 constructs were tested for their ability to activate motility of the dynein–dynactin complex in the presence of ATP (see Fig. 4 E). Representative kymographs are shown for each construct that moved. The kymographs are displayed using the same brightness and contrast. (F) Ratios of the motile shorter constructs to Hook3_1–552 were calculated from side-by-side experiments performed on the same day. Shown are the mean and SD of the ratios from three independent experiments performed on different days. The mean number of motile Hook3_1–552 molecules/µm per min was 0.070 ± 0.060. (G) The indicated truncations of sfGFP-Hook3 were incubated with affinity-purified human dynein–dynactin, and fluorescence binding to surface-immobilized microtubules was assessed in the absence of ATP; overlay shows Hook3 in green and microtubules in blue (images are displayed using the same brightness and contrast). (H) The fluorescence quantification for each condition is shown (mean fluorescence intensity [arbitrary units] of Hook3 per micrometer of microtubule). For each condition, >30 microtubules were quantified, and three replicate experiments were performed on different days (mean and SD, with the SD representing the variation in the ratio of intensity per micrometer).

Figure 6.

Model of assembly and activation of the dynein–dynactin–Hook3 complex. Short Hook constructs (e.g., Hook3_1–402) are able to assemble the tripartite motor complex by binding the LIC1 C-terminal domain and part of the dynactin Arp1 filament and dynein heavy chain. However, the complex is inactive for motility. We speculate that the longer coiled coil of Hook3_1–552 releases the CAP-Gly domain of p150 from an autoinhibited state to enable its binding to microtubules, thus enhancing the initiation of processive motility. A change in the orientation or other allosteric change in the motor domains, based on the work of Urnavicius et al. (2015) and Torisawa et al. (2014), also might be promoted by the longer Hook3 constructs. The illustration of the dynein–dynactin complex is based on work by Urnavicius et al. (2015) and Chowdhury et al. (2015). The length of Hook3_1–402, which contains ∼240 residues of coiled coil, is estimated based on the dynein–dynactin–BicD2 cryo-EM structure (Urnavicius et al., 2015), which contained a 270-residue coiled coil. The illustration of the Hook3_1–552 coiled coil was then made proportional to the length of Hook3_1–402 based on the ratio of residues.