HookA is a novel dynein-early endosome linker critical for cargo movement in vivo

Abstract

Cytoplasmic dynein transports membranous cargoes along microtubules, but the mechanism of dynein-cargo interaction is unclear. From a genetic screen, we identified a homologue of human Hook proteins, HookA, as a factor required for dynein-mediated early endosome movement in the filamentous fungus Aspergillus nidulans. HookA contains a putative N-terminal microtubule-binding domain followed by coiled-coil domains and a C-terminal cargo-binding domain, an organization reminiscent of cytoplasmic linker proteins. HookA-early endosome interaction occurs independently of dynein-early endosome interaction and requires the C-terminal domain. Importantly, HookA interacts with dynein and dynactin independently of HookA-early endosome interaction but dependent on the N-terminal part of HookA. Both dynein and the p25 subunit of dynactin are required for the interaction between HookA and dynein-dynactin, and loss of HookA significantly weakens dynein-early endosome interaction, causing a virtually complete absence of early endosome movement. Thus, HookA is a novel linker important for dynein-early endosome interaction in vivo.

Figure 1.

Phenotype of the eedA1 mutant and rescue of the mutant phenotype by the gene encoding HookA. (A) A schematic diagram depicting the phenotype of the Δp25 mutant in comparison to the ΔnudA (dynein HC) mutant. Note that early endosomes abnormally accumulate at the hyphal tip in both the ΔnudA and Δp25 mutants but a nuclear distribution phenotype is shown only in the ΔnudA mutant. Red, early endosomes. Dark blue, nuclei. Black lines, microtubules. (B) A brief outline of the mutant-screening procedure. (C) Colony phenotypes of the eedA1 mutant and a wild-type strain. (D) Microscopic images showing the distributions of mCherry-RabA–labeled early endosomes (mCherry-RabA) and GFP-labeled dynein HC (GFP-HC). The same cells are shown for both the mCherry-RabA and GFP-HC images. Although bidirectional movements of mCherry-RabA–labeled early endosomes are not completely abolished, ∼83% of eedA1 hyphal tips show obvious accumulation of mCherry-RabA signals (n = 140), whereas none of the wild-type hyphal tips show this accumulation (n = 100). Dynein comets are present in all wild-type and mutant cells. (E) Images of nuclei stained by a DNA dye, DAPI, in wild type and the eedA1 mutant. The pattern of nuclear distribution in the eedA1 mutant is normal, as none of the mutant cells show any cluster of four or more nuclei when grown under the same conditions that allow us to see the hyphal tip mCherry-RabA accumulation (n > 100 for wild type, and n > 100 for the mutant). (F) Rescue of the eedA1 mutant phenotype with the HookA-encoding DNA. (left) Colony phenotypes of the eedA1 mutant, eedA1 mutant transformed with the HookA-encoding gene, and a wild-type strain. (right) Distribution of mCherry-RabA–labeled early endosomes in an eedA1 mutant transformed with the HookA-encoding gene. None of the hyphal tips show accumulation of mCherry-RabA signals (n = 30). Bars, 5 µm.

Figure 2.

Sequence and functional analyses of HookA. (A) A sequence alignment of the N-terminal putative microtubule-binding domain of HookA (A. nidulans), Hook (Drosophila), and Hook3 (human_HK3). The alignment was performed using CLUSTALW (Pôle BioInformatique Lyonnais Network Protein Sequence Analysis). Residues that are identical (asterisks), strongly similar (double dots), or weakly similar (single dots) are shown as red, green, and blue characters, respectively. Also see Fig. S1. (B) Colony phenotype of the ΔhookA mutant. (C) Distributions of mCherry-RabA–labeled early endosomes in the ΔhookA mutant (also see Videos 1 and 2) and the ΔclipA mutant. An obvious accumulation of mCherry-RabA signals is found at ∼80% of the hyphal tips in the ΔhookA mutant (n = 326), whereas none of the hyphal tips in wild type (n = 240) or the ΔclipA mutant (n = 73) show the same accumulation. (D, top) GFP–dynein HC (GFP-HC) signals in wild type and the ΔhookA mutant. Maximal signal intensities (arbitrary units) of the plus-end GFP-HC comets in wild-type cells and in ΔhookA cells are 276 ± 262 (n = 25) and 268 ± 225 (n = 30), respectively, and there is no statistical difference between the values at P = 0.05. (Bottom) Images of nuclei stained by DAPI in wild type and the ΔhookA mutant. The pattern of nuclear distribution in the ΔhookA mutant is normal, as none of the mutant cells shows any cluster of four or more nuclei when grown under the same conditions that allow us to see the hyphal tip mCherry-RabA accumulation (n > 100 for wild type, and n > 100 for the mutant). Bars, 5 µm.

Figure 3.

Colocalization of HookA-GFP signals with mCherry-RabA–labeled early endosomes. (A) Images of HookA-GFP and mCherry-RabA in the same cell. (B) Kymographs of the GFP and mCherry signals obtained via duel-view imaging. Arrows are shown to indicate that some HookA-GFP signals are associated with motile early endosomes. (C) HookA-GFP and mCherry-RabA in the ΔkinA mutant. HookA-GFP signals were concentrated at every hyphal tip where early endosomes accumulate (n = 50). (D) HookA-GFP and mCherry-RabA in the Δp25 mutant. HookA-GFP signals were concentrated at every hyphal tip where early endosomes accumulate (n = 50). The same minimal medium containing 1% glycerol as a carbon source was used for cells shown in A, C, and D. For the medium used for growing the cells viewed by the dual-view imaging (B), 0.1% fructose instead of 1% glycerol was used as a carbon source to reduce the intensity of the mCherry signals. Bars, 5 µm.

Figure 4.

The C-terminal deletion mutants of HookA exhibit a defect in the HookA–early endosome interaction. (A) A diagram showing the wild-type HookA protein and the two C-terminal deletion mutants, ΔC-HookA and ΔC1-HookA, in which different amino acids are deleted. The red box indicates the putative microtubule-binding domain, the blue boxes indicate the three predicted coiled-coil domains, and the brown box indicates the C-terminal cargo-binding domain. (B) The ΔC-HookA and ΔC1-HookA mutants exhibit the same colony phenotype as that exhibited by the ΔhookA mutant. (C) The ΔC-HookA and ΔC1-HookA mutants show an obvious accumulation of mCherry-RabA–labeled early endosomes at the hyphal tip. The accumulation can be seen in ∼75% of the hyphal tips of both the ΔC-HookA and the ΔC1-HookA mutants (n = 98 for the ΔC-HookA mutant, and n = 102 for ΔC1-HookA mutant). See Video 3 for the phenotype of the ΔC-HookA mutant. (D) Kymographs showing an obvious accumulation of mCherry-RabA–labeled early endosomes at the hyphal tip in the ΔC-HookA and ΔhookA mutants and nonmotile early endosomes along the hyphae. (E) ΔC-HookA–GFP or ΔC1-HookA–GFP do not colocalize with the hyphal tip–accumulated early endosomes (100%, n = 50 for each mutant). (F) Western blots are shown to demonstrate that the ΔC-HookA–GFP or ΔC1-HookA–GFP proteins are expressed and stable. By measuring protein signal intensity on the Western blots in relation to protein loading as indicated by Ponceau S staining, we found that the level of ΔC-HookA–GFP relative to HookA-GFP is 1.17 ± 0.28 (mean ± SD; n = 3) and that of ΔC1-HookA–GFP relative to HookA-GFP is 0.99 ± 0.18 (mean ± SD; n = 3). There is no significant difference between the value of either mutant and that of the wild type at P = 0.05. Bars, 5 µm.

Figure 5.

The C-terminal domain of HookA is capable of interacting with early endosomes. (A) Two examples showing that the C-HookA–GFP signals are concentrated at the hyphal tip where mCherry-RabA–marked early endosomes accumulate and the GFP and mCherry signals largely overlap (100% hyphal tips that show the concentrated GFP signals show the mCherry-RabA accumulation; n = 50). (B) A Western blot showing that the protein level of C-HookA–GFP expressed under the gpdA promoter (gpdA-C-HookA–GFP) is much higher than that expressed under the endogenous hookA promoter (C-HookA–GFP). The proteins were pulled down by the anti-GFP antibody, and the Western blot was probed by the anti-GFP antibody. Quantitation of the Western blots suggests that the level of gpdA-C-HookA–GFP is significantly higher than that of C-HookA–GFP (P < 0.005, n = 3). If we set the values of C-HookA–GFP as 1, the mean ± SD value of gpdA-C-HookA–GFP is 3.1 ± 0.4. This is likely to be an underestimate of the gpdA-C-HookA–GFP protein level as there seems to be a lot of degradation products, which are hard to include in the measurements. (C) Phenotypic analysis of the diploids showing that overexpression of C-HookA (gpdA-C-HookA–GFP) in the wild-type background produced a dominant-negative phenotype in early endosome distribution. Bars, 5 µm.

Figure 6.

The N-terminal domain of HookA is critical for the HookA–dynein–dynactin interaction. (A) A diagram showing the wild-type HookA protein and the N-terminal deletion mutant protein, ΔN-HookA, in which the putative microtubule-binding domain is deleted. (B) The ΔN-HookA mutant exhibits the same colony phenotype as that exhibited by ΔhookA. (C) The ΔN-HookA mutant showed an obvious accumulation of mCherry-RabA–labeled early endosomes at the hyphal tip (∼71% of the hyphal tips show this accumulation, n = 112. Also see Video 6). (D) Kymographs showing an obvious accumulation of mCherry-RabA–labeled early endosomes at the hyphal tip in the ΔN-HookA and ΔhookA mutants. An arrowhead indicates one early endosome that moved away from the hyphal tip in the ΔN-HookA mutant. (E) The ΔN-HookA–GFP signals were concentrated at the hyphal tip where mCherry-RabA–marked early endosomes accumulate, and the GFP and mCherry signals largely overlap (100% hyphal tips that show the concentrated GFP signals show the mCherry-RabA accumulation; n = 50). (F) The dynein HC, the p150 subunit of dynactin, and NudF/LIS1 can be pulled down with HookA-GFP, ΔC-HookA–GFP, and ΔC1-HookA–GFP, but the amounts of these proteins pulled down with ΔN-HookA–GFP were obviously decreased. (G) A quantitative analysis of the Western results shown in F. The ratio of pulled down dynein HC, dynactin p150, or NudF/LIS1 to HookA-GFP was calculated. Values of all the mutants are relative to the wild-type values, which are set at 1. Mean and SD values were calculated from multiple independent pull-down experiments, and the number of experiments is indicated as n. For the ratio of dynein to HookA (dynein/HookA), the mean ± SD value for ΔN is 0.08 ± 0.11 (n = 4, P < 0.001), and the values for ΔC and ΔC1 are 1.77 ± 0.96 (n = 4) and 1.26 ± 0.71 (n = 3), respectively. Note that a p-value is provided only when the values are statistically different from the wild-type value, and the values of ΔC and ΔC1 are not different from the wild-type value at P = 0.05. For the ratio of dynactin to HookA (dynactin/HookA), the mean ± SD value for ΔN is 0.22 ± 0.18 (n = 4, P < 0.001), and the values for ΔC and ΔC1 are 1.5 ± 0.28 (n = 4, P < 0.05) and 1.6 ± 0.87 (n = 3), respectively. For the ratio of NudF/LIS1 to HookA (LIS1/HookA), the mean ± SD value for ΔN is 0.15 ± 0.1 (n = 4, P < 0.001), and the values for ΔC and ΔC1 are 1.37 ± 0.21 (n = 4, P < 0.05) and 2.0 ± 1.1 (n = 3), respectively. (H) A diagram showing the wild-type, ΔC-HookA, ΔN-ΔC-HookA, and ΔN1-ΔC-HookA mutant proteins. (I) Dynein HC and dynactin p150 could be pulled down with ΔC-HookA–GFP, but the amounts of these proteins were obviously diminished when the pull-down was performed with ΔN-ΔC-HookA–GFP and were nearly undetectable when the pull-down was performed with ΔN1-ΔC-HookA–GFP. (J) A quantitative analysis of the Western results shown in I. Values of ΔN-ΔC-HookA–GFP and ΔN1-ΔC-HookA–GFP are relative to the ΔC-HookA–GFP values, which are set at 1. Values of ΔN-ΔC-HookA–GFP and ΔN1-ΔC-HookA–GFP are significantly lower than that of ΔC-HookA–GFP. For the ratio of dynein to HookA (dynein/HookA), the mean ± SD values for ΔN-ΔC and ΔN1-ΔC are 0.2 ± 0.3 (n = 3, P < 0.05) and 0 ± 0 (n = 3, P < 0.001), respectively. For the ratio of dynactin to HookA (dynactin/HookA), the mean ± SD values for ΔN-ΔC and ΔN1-ΔC are 0.16 ± 0.2 (n = 3, P < 0.005) and 0.02 ± 0.03 (n = 3, P < 0.001), respectively. However, the values for ΔN-ΔC and ΔN1-ΔC are not significantly different from each other at P = 0.05. Bars, 5 µm.

Figure 7.

Dynein and p25 of dynactin are codependent for the HookA–dynein–dynactin interaction. (A) A diagram showing the dynactin complex (Schroer, 2004) and the heavy chain (HC) and intermediate chain (IC) of dynein. (B) Western blots showing that HookA–dynein–dynactin interaction is defective in a dynein HC conditional null mutant, alcA-nudA^HC, and an Arp1 conditional null mutant, alcA-nudK^Arp1. Cells were grown in rich medium YG that contains glucose to prevent expression of the nudA dynein HC gene and the nudK Arp1 gene. (C) A quantitative analysis of the Western results shown in B. The ratio of pulled down dynein HC or dynactin to HookA-GFP was calculated. Values of all the mutants are relative to the wild-type values, which are set at 1. Mean and SD values were calculated from multiple independent pull-down experiments, and the number of experiments is indicated as n. For the ratio of dynein to HookA (dynein/HookA), the mean ± SD values for alcA-nudA^HC and alcA-nudK^Arp1 are 0.1 ± 0.17 (n = 3, P < 0.001) and 0.08 ± 0.13 (n = 3, P < 0.001), respectively. For the ratio of dynactin to HookA (dynactin/HookA), values for alcA-nudA^HC and alcA-nudK^Arp1 are 0.04 ± 0.07 (n = 3, P < 0.001) and 0.01 ± 0.01 (n = 3, P < 0.001), respectively. (D) Western blots showing that the HookA–dynein–dynactin interaction is defective in the alcA-nudA^HC and the Δp25 mutant. (E) A quantitative analysis of the Western results shown in D. Values of all the mutants are relative to the values of ΔC1-HookA in the wild-type background, which are set at 1. For the ratio of dynein to ΔC1-HookA (dynein/ΔC1-HookA), the mean ± SD values for alcA-nudA^HC and Δp25 are 0.04 ± 0.08 (n = 3, P < 0.001) and 0.01 ± 0.03 (n = 4, P < 0.001), respectively. For the ratio of dynactin to ΔC1-HookA (dynactin/ΔC1-HookA), values for alcA-nudA^HC and Δp25 are 0.1 ± 0.09 (n = 3, P < 0.001) and 0.06 ± 0.12 (n = 4, P < 0.001), respectively. (F) Western blots showing that the HookA–dynein–dynactin interaction is defective in the Δp25 mutant. For the pull-down experiments presented in F, a supernatant of the 100,000 g high-speed centrifugation was used. (G) A quantitative analysis of the Western results shown in F. Values of the mutants are relative to the values of ΔC1-HookA in the wild-type background, which are set at 1. For the ratio of dynein to ΔC1-HookA (dynein/ΔC1-HookA), the mean ± SD value for Δp25 is 0.12 ± 0.2 (n = 3, P < 0.001). For the ratio of dynactin to ΔC1-HookA (dynactin/ΔC1-HookA), the mean ± SD value for Δp25 is 0.02 ± 0.03 (n = 3, P < 0.001).

Figure 8.

Loss of HookA or the HookA–dynein–dynactin interaction significantly weakens the interaction between dynein and early endosomes. (A) Western blots showing that the amount of mCherry-RabA–labeled early endosomes pulled down with GFP–dynein HC from the ΔhookA mutant extract is significantly lower than that from the wild type. The mCherry-RabA signals in the extracts used for the pull-down experiments are also shown. (B) A quantitative analysis of the Western results (shown in A). The ratio of mCherry-RabA to GFP–dynein HC (RabA/dynein) was calculated. Values are relative to the wild-type value, which is set at 1. The mean ± SD value for the ΔhookA mutant is 0.19 ± 0.22 (n = 4, P < 0.001). (C) The HookA^L150P,E151K-GFP signals were concentrated at the hyphal tip where mCherry-RabA–marked early endosomes accumulate, and the GFP and mCherry signals largely overlap. Bars, 5 µm. (D) The dynein HC and the p150 subunit of dynactin could be pulled down with HookA-GFP, but the amounts of these proteins pulled down with HookA^L150P,E151K-GFP were obviously decreased. For the ratio of dynein to HookA, if we set the wild-type mean values to 1, the mean ± SD value for HookA^L150P,E151K was 0.05 ± 0.09 (n = 3, P < 0.001). For the ratio of dynactin to HookA, if we set the wild-type mean values to 1, values for HookA^L150P,E151K were 0.09 ± 0.16 (n = 3, P < 0.001). In contrast, the amount of mCherry-RabA–labeled early endosomes pulled down was apparently not decreased. The mCherry-RabA signals in the extracts used for the pull-down experiments are also shown. (E) Western blots showing that the amount of mCherry-RabA–labeled early endosomes pulled down with dynein HC-GFP from the HookA^L150P,E151K mutant extract is significantly lower than that from the wild type. The mCherry-RabA signals in the extracts used for the pull-down experiments are also shown. (F) A quantitative analysis of the Western results (shown in E). The ratio of mCherry-RabA to GFP–dynein HC (RabA/dynein) was calculated. Values are relative to the wild-type value, which is set at 1. The mean ± SD value for the ΔhookA mutant is 0.25 ± 0.15 (n = 4, P < 0.001). (G) A working model showing that HookA on an early endosome links dynein–dynactin to the cargo for its movement along the microtubule track. Several dimers of HookA are depicted. A possibility not excluded is that HookA also facilitates cargo–track interaction, which is likely to be dynamic rather than static. For simplicity, HookA is depicted as the only protein linking dynein–dynactin to the early endosome, but it is likely that additional proteins are required for bridging the HookA–dynein–dynactin and HookA–early endosome interactions.

Figure 9.

Distribution of PexK-GFP, a peroxisomal marker, is abnormal in the ΔhookA mutant. A wild-type control and several ΔhookA cells are shown. For the ΔhookA cell presented on the top right, PexK-GFP, mCherry-RabA, and their merger are all shown. For the ΔhookA cells presented on the bottom, PexK-GFP and the merger of PexK-GFP and mCherry-RabA are shown. Arrows indicate the abnormal accumulation of PexK-GFP signals. Bars, 5 µm.