Hook is an adapter that coordinates kinesin-3 and dynein cargo attachment on early endosomes

Abstract

Bidirectional membrane trafficking along microtubules is mediated by kinesin-1, kinesin-3, and dynein. Several organelle-bound adapters for kinesin-1 and dynein have been reported that orchestrate their opposing activity. However, the coordination of kinesin-3/dynein-mediated transport is not understood. In this paper, we report that a Hook protein, Hok1, is essential for kinesin-3- and dynein-dependent early endosome (EE) motility in the fungus Ustilago maydis. Hok1 binds to EEs via its C-terminal region, where it forms a complex with homologues of human fused toes (FTS) and its interactor FTS- and Hook-interacting protein. A highly conserved N-terminal region is required to bind dynein and kinesin-3 to EEs. To change the direction of EE transport, kinesin-3 is released from organelles, and dynein binds subsequently. A chimaera of human Hook3 and Hok1 rescues the hok1 mutant phenotype, suggesting functional conservation between humans and fungi. We conclude that Hok1 is part of an evolutionarily conserved protein complex that regulates bidirectional EE trafficking by controlling attachment of both kinesin-3 and dynein.

Figure 1.

Hok1 is required for EE motility. (A) Morphology of wild type and Δhok1 mutants. (B) EE distribution and motility in wild type, EMD5 mutants expressing hok1 (EMD5 + Hok1), and Δhok1 mutants (Δhok1). Deletion or mutation of hok1 abolishes motility and induces EE clusters (arrowheads; asterisk in kymograph). Contrast-inverted kymographs show GFP-Rab5a fluorescence (EEs). See also Video 1. (C) Domain organization of selected Hook proteins. For accession numbers, see Materials and methods. (D) Contrast-inverted kymographs showing bidirectional motility of mCherry-Rab5a–labeled EEs and Hok1-GFP. Hok1 locates on all moving organelles (yellow lines in merged image). See also Video 2. Images in A, B, and D were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and vertical bars are in seconds.

Figure 2.

The C-terminal region of Hok1 targets to endosomes. (A) The organization of human Hook3, Hok1, and the truncated proteins Hok1^1–224, Hok1^225–624, and Hok1^625–930. (B) Morphology of Δhok1 expressing Hok1^1–224, Hok1^225–624, and Hok1^625–930. (C) Colocalization of Hok1^1–224, Hok1^225–624, Hok1^625–930, and Hok1^1–624 and mCherry-Rab5a–labeled EEs in Δhok1. The cell edge is indicated in blue. Only the C-terminal fragment of Hok1 localizes to apical EE clusters. (D) Organization of Hok1^1–624 and Hok1^1–624PX. The latter carries an EE-targeting PX domain (Wedlich-Söldner et al., 2000). (E) Morphology of control cells and Δhok1 that express Hok1, Hok1^1–624, and Hok1^1–624PX. Targeting of truncated Hok1 to EEs partially rescues the growth defect. (F) Bidirectional motility of mCherry-Rab5a–labeled EEs and Hok1^1–624PX. The upper two kymographs are contrast inverted. Hok1^1–624PX locates on the moving organelles (yellow lines in merged image). See also Video 3. Images in B, C, E, and F were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and the vertical bar is in seconds.

Figure 3.

The N-terminal region of Hok1 mediates dynein binding to EEs. (A) GFP₃-labeled dynein heavy chain (dynein) and mCherry-Rab5a–labeled endosomes (EEs) in Δhok1 mutants. Dynein (asterisk) accumulates at the cell tip (tip), from where it leaves without EEs. Cells were photobleached (bleach) to reduce signal interferences, which did not affect EE or motor motility (Schuster et al., 2011c). See also Video 4. (B) Estimated dynein numbers in moving GFP₃-Dyn2 signals in control and Δhok1 cells. Estimation used an internal calibration standard, assuming that the GFP₃-labeled dynein heavy chain forms dimers (Schuster et al., 2011c). Data represent two experiments. The red line shows a normal distribution curve. (C) Retrograde dynein flux in control and Δhok1 cells at ∼10 µm behind the tip. Bars represent data from two experiments and are means ± SE; sample size is indicated. No significant difference was found, P = 0.09. (D) Images and linescan plot of dynein and EE colocalization in Δhok1 hyphal tips after disruption of the MTs (+benomyl). The dotted line in merged image indicates the region of intensity scan. a.u., arbitrary unit. (E) Motility of Hok1^1–624 and mCherry-Rab5a–labeled EEs in Δhok1. Hok1^1–624 moves in retrograde direction, whereas EEs remain stationary. See also Video 5. (F) Retrograde flux of Hok1 and Hok1^1–624 at ∼10 µm behind the cell tip. Bars represent data from two experiments and are means ± SE; sample size is indicated. Significant difference is indicated: ***, P < 0.0001. (G) Co-migration of Hok1^1–624 and mCherry₃-labeled dynein heavy chain. Kymographs were slightly misaligned to better show colocalization. See also Video 6. (H) Domain organization of proteins used in immunoprecipitation and mass spectrometry experiments. Images in A, D, E, and G were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and vertical bars are in seconds.

Figure 4.

A conserved region in the first CC is essential for Hok1 function. (A) Organization of truncated Hok1^Δ225–624, Hok1^1–224PX, which carries an EE-targeting PX domain, and Hok1^225–930. (B) Colocalization of truncated Hok1 proteins and mCherrry-Rab5a–labeled EEs in Δhok1 and colocalization of Hok1-GFP and EEs in cells expressing the peptide pHk^361–385. Cell edges are indicated in blue. All constructs localize to EEs without rescuing Δhok1. (C) Morphology of Δhok1 mutants expressing Hok1^Δ225–624, Hok1^1–224PX, Hok1^225–930, and Hok1^Δ361–385. Neither rescues the morphology defect. (D) Localization of a highly conserved region within the first CC of Hok1 and human Hook3. Identical amino acids are shown in light blue, and similar amino acids are shown in pink. The alignment was generated in ClustalW. H.s., Homo sapiens; U.m., U. maydis. (E) Organization of the truncated proteins Hok1^Δ361–385 and Hok1^Δ333–355. (F) Anterograde and retrograde flux of mCherry-Rab5a–labeled EEs in control cells and Hok1^Δ361–385 and Hok1^Δ333–355 (Δ333–355 and Δ361–385) at ∼10 µm behind the cell tip. Bars represent data from two experiments and are means ± SE; sample size is indicated. (G) Anterograde and retrograde flux of mCherry-Rab5a–labeled EEs in control cells and cells expressing the peptide pHk^361–385 (see D for sequence). Bars represent data from two experiments and are means ± SE; sample size is indicated. Images in B and C were adjusted in brightness, contrast, and γ settings. Significant difference is indicated: ***, P < 0.0001. Bars are in micrometers.

Figure 5.

Hok1 is required for cargo binding of kinesin-3. (A) Localization of an EE cluster, labeled with GFP-Rab5a (arrowhead) in a cell expressing the peptide pHk^361–385 for 2 h. Note that clusters initially appear subapically but later shift to the cell end (tip). (B) Motility of GFP-Rab5a–labeled EEs in a cell expressing the peptide pHk^361–385 for 2 h. A subapical cluster is indicated by an asterisk. Cell end is indicated by tip and the dotted line. See also Video 7. (C) Mean run length of EEs in control cells and cells expressing pHk^361–385 for 2 h. Bars represent data from two experiments and are means ± SE; sample size is indicated. Significant difference is indicated: ***, P < 0.0001. (D) Contrast-inverted kymographs showing anterograde motility of kinesin-3–GFP in control cells, after expressing the peptide pHk^361–385, and in Δhok1 and Δhok1 cells expressing Hok1^Δ361–385. Kinesin-3 signals are strong in the control but weak in all mutants. (E) Kinesin-3-GFP numbers in control cells, after expressing the peptide pHk^361–385, and in Δhok1– and hok1–null cells expressing Hok^Δ361–385. Note that native kinesin-3 levels are shown. All numbers represent two experiments and are given as means ± SD; sample size is indicated. Motor number estimation is based on comparison with an internal calibration standard (Schuster et al., 2011c) and assuming that kinesin-3 is a dimer (Hammond et al., 2009). Red dotted lines indicate medians. Images in A, B, and D were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and vertical bars are in seconds.

Figure 6.

Kinesin-3 numbers drop before anterograde to retrograde turning of EEs. (A) Numbers of Hok1-GFP and kinesin-3–GFP in signals moving bidirectionally. Bars are means ± SE of two experiments; sample size is indicated. No significant difference is found, P = 0.391. Numbers were estimated using an internal calibration standard (Schuster et al., 2011c) and assume proteins are dimers (Xu et al., 2008; Hammond et al., 2009). (B) Numbers of kinesin-3–GFP in anterograde and retrograde signals. Bars are means ± SE of two experiments; sample size is indicated. No significant difference is found, P = 0.811. Numbers were estimated using an internal calibration standard (Schuster et al., 2011c) and assume proteins are dimers (Hammond et al., 2009). (C) Pausing time of EEs and dynein before retrograde motility. Bars are means ± SE of two experiments; sample size is indicated. (D) Kymographs showing the binding of GFP₃-dynein (green) to a mCherry-Rab5a–labeled EE (red). An anterograde (antero)-moving organelle pauses (pause) before it binds dynein and turns to retrograde motility (retro). (E) False-colored kymographs showing Hok1-GFP and kinesin-3-GFP during anterograde to retrograde turning. Kinesin-3 signals drop just before the organelle pauses (arrowheads). Recovery of the fluorescent signal begins during pause (yellow arrowhead). Note signal variations caused by changes in focal plane and that stationary signals are brighter as they are more focused. (F) Mean intensities of Hok1-GFP and kinesin-3-GFP during anterograde to retrograde turning of EEs. Hok1 numbers remain stable, whereas kinesin-3 signals drop before the pausing phase. Data points are means ± SE from a representative experiment, and sample sizes are indicated. Significant difference is indicated: *, P < 0.05; ***, P < 0.0001. Images in D and E were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and vertical bars are in seconds.

Figure 7.

The function of Hok1 is conserved. (A) Phylogenetic tree of Hook and closest non-Hook proteins and the correlation with the presence of kinesin-3 motors. No Kif1A-like sequence was found at NCBI for A. fumingatus, and no Hook was listed for A. nidulans. The tree is based on Hook domains, identified in Pfam, and calculated in MEGA 5.10 (Tamura et al., 2011). (B) Domain organization of human Hook3 and the chimeric proteins Hok1^HsH3_1–600, Hok1^{HsH3_166–436}, and Hok1^{HsH3_293–345}. (C) Colocalization of Hok1^HsH3_1–600 and Hok1^{HsH3_166–436} with mCherry-Rab5a on apical EEs. (D) Morphology of Δhok1 expressing Hok1, Hok1^HsH3_1–600, Hok1^{HsH3_166–436}, and Hok1^{HsH3_293–345}. The conserved region of human Hook3 restores the mutant phenotype in U. maydis. (E) Co-motility of Hok1^{HsH3_293–345} and mCherry-Rab5a–labeled EEs. See also Video 8. (F) Bidirectional flux of EEs in wild type and Δhok1 expressing Hok1^{HsH3_293–345}. Bars represent two experiments and are means ± SE; sample size is indicated. No significant difference was found (P = 0.285, Kruskal–Wallis test). (G) Endocytic sorting defect in Δhok1 and dynein (Dyn2^ts) mutants treated with the dye FM4-64. In wild type, the dye appears in the vacuoles (stained with Cell Tracker Blue CMAC), whereas in the mutants, it accumulates next to GFP-Rab5a–carrying EEs and does not reach the vacuoles (asterisks). The arrowhead indicates the FM4-64–stained “cloud.” (H) The late endocytic compartment in Δhok1. In wild-type cells, the late endosomes marker Rab7 localizes to vacuoles and motile structures (arrowhead). In Δhok1, Rab7 colocalizes with the apical FM4-64–positive cloud, whereas the localization on vacuoles is not affected. Images in C, D, E, G, and H were adjusted in brightness, contrast, and γ settings. Horizontal bars are in micrometers, and the vertical bar is in seconds.

Figure 8.

Model of Hok1 in controlling EE motility. (A) Fts1, Fhp1, and Hok1 form a complex that may contain additional, yet unknown, adapter proteins that link dynein and kinesin-3 to this Hok1 complex (b and c). Kinesin-3 binds to Hok1 and, to a lesser extent, directly to EE membranes. (B) Before dynein binding, the Hok1 complex releases kinesin-3. The cargo stops moving but remains attached to the MT. This may reflect the ability of Hok1 to bind MTs, as described for human Hook proteins (Walenta et al., 2001). (C) Dynein leaves the MT plus ends and travels toward the pausing EE, where it interacts with the Hok1 complex. This could involve the dynactin complex, as shown in the fungus A. nidulans (c; Zhang et al., 2011). (D and E) During pausing and while dynein moves the cargo to minus ends of MTs, kinesin-3 rebinds to the Hok1 complex. Speculative parts of the model are indicated by question marks.