LIS1 regulates cargo-adapter–mediated activation of dynein by overcoming its autoinhibition in vivo

Cytoplasmic dynein-1 (called dynein hereafter) is a microtubule (MT) motor that transports a variety of cargos in eukaryotic cells, and defects in dynein-mediated transport are linked to devastating neurodegenerative diseases and brain developmental disorders (Maday et al., 2014; Jaarsma and Hoogenraad, 2015; Bertipaglia et al., 2018). The dynein–cargo interaction requires the multicomponent dynactin complex as well as specific cargo adapters (Schroer, 2004; Akhmanova and Hammer, 2010; Fu and Holzbaur, 2014; Reck-Peterson et al., 2018; Olenick and Holzbaur, 2019). Importantly, dynactin and cargo adapters also activate the motility of cytoplasmic dynein in vitro (McKenney et al., 2014; Schlager et al., 2014; Reck-Peterson et al., 2018; Olenick and Holzbaur, 2019). The mechanism underlying this activation was suggested by a recent cryo-EM analysis (Zhang et al., 2017a). Specifically, the two motor domains of the dynein heavy chain (HC) dimer are held together in an inactive “phi” conformation (Torisawa et al., 2014; Zhang et al., 2017a), which is in equilibrium with an “open” conformation in which the two domains are separated. While dynein in the “open” conformation is still not configured properly to move directionally along MTs by itself, its binding to dynactin and cargo adapter causes the HC dimer to become parallel for directional movement along MTs (Zhang et al., 2017a). Moreover, some cargo adapters facilitate the recruitment of a second dynein dimer to dynactin (Grotjahn et al., 2018; Urnavicius et al., 2018), further enhancing dynein force and speed (Urnavicius et al., 2018). While these are important steps toward understanding dynein regulatory mechanisms, the cargo-adapter-mediated dynein activation has never been analyzed in vivo, and it is especially unclear whether this process is regulated by other proteins in vivo.

Two of the most well-known yet enigmatic dynein regulators are LIS1 and its binding partner, NudE (Kardon and Vale, 2009; Vallee et al., 2012; Reck-Peterson et al., 2018; Olenick and Holzbaur, 2019). LIS1 is encoded by the lis1 gene, whose deficiency causes type I lissencephaly, a human brain developmental disorder (Reiner et al., 1993). Fungal genetic studies first linked LIS1 to dynein function (Xiang et al., 1995a; Geiser et al., 1997; Willins et al., 1997). In the filamentous fungus Aspergillus nidulans, the LIS1 homologue NudF is critical for dynein-mediated nuclear distribution (Xiang et al., 1995a). A. nidulans genetics also led to the identification of NudE, a NudF/LIS1-binding protein (Efimov and Morris, 2000; Feng et al., 2000; Niethammer et al., 2000; Sasaki et al., 2000), whose homologues participate in dynein function in various systems (Minke et al., 1999; Liang et al., 2004, 2007; Li et al., 2005a; Stehman et al., 2007; Yamada et al., 2008; Kardon and Vale, 2009; Ma et al., 2009; Zhang et al., 2009; Lam et al., 2010; Pandey and Smith, 2011; Wang and Zheng, 2011; Zyłkiewicz et al., 2011; Vallee et al., 2012; Raaijmakers et al., 2013; Wang et al., 2013; Klinman and Holzbaur, 2015; Kuijpers et al., 2016; Simões et al., 2018; Olenick and Holzbaur, 2019). The mechanism by which these two proteins regulate dynein remains unclear. The dynein HC contains six AAA domains in its motor ring (King, 2000; Asai and Koonce, 2001), and LIS1 binds AAA3/AAA4 and the stalk leading to the MT-binding domain (Huang et al., 2012; Toropova et al., 2014; DeSantis et al., 2017). Intriguingly, purified LIS1 inhibits dynein motility in vitro (Yamada et al., 2008; McKenney et al., 2010; Huang et al., 2012), unless ATP hydrolysis at AAA3 is blocked (DeSantis et al., 2017). NudE/Nudel (also called NDEL1) relieves the inhibitory effect of LIS1 (Yamada et al., 2008; Torisawa et al., 2011), and the NudE–LIS1 complex enhances dynein force production (McKenney et al., 2010; Reddy et al., 2016). Moreover, dynactin partially relieves the inhibition of LIS1 on dynein motility (Wang et al., 2013), and when both dynactin and cargo adapter are present, LIS1 no longer inhibits but mildly enhances the dynein movement (Baumbach et al., 2017; Gutierrez et al., 2017; Jha et al., 2017). However, the exact molecular mechanism of LIS1 action on dynein regulation is not known (Reck-Peterson et al., 2018; Olenick and Holzbaur, 2019).

We have been using the fungal model organism A. nidulans to investigate dynein regulation in vivo. Unlike budding yeast, where dynein is required almost exclusively for nuclear migration/spindle orientation (Eshel et al., 1993; Li et al., 1993), dynein, dynactin, and LIS1 in filamentous fungi are required not only for nuclear distribution (Plamann et al., 1994; Xiang et al., 1994, 1995a) but also for transporting a variety of other cargos, including early endosomes and their hitchhiking partners (Wedlich-Söldner et al., 2002; Lenz et al., 2006; Abenza et al., 2009; Zekert and Fischer, 2009; Baumann et al., 2012; Bielska et al., 2014a; Higuchi et al., 2014; Egan et al., 2015; Guimaraes et al., 2015; Pohlmann et al., 2015; Salogiannis et al., 2016; Peñalva et al., 2017; Otamendi et al., 2019). In filamentous fungi and budding yeast, dynein, dynactin, and LIS1–NudE all accumulate at the MT plus ends (Han et al., 2001; Efimov, 2003; Lee et al., 2003; Sheeman et al., 2003; Zhang et al., 2003; Li et al., 2005a; Lenz et al., 2006; Moore et al., 2008; Callejas-Negrete et al., 2015). The MT plus-end accumulation of dynein is important for spindle-orientation/nuclear migration and for early endosome transport (Lee et al., 2003; Sheeman et al., 2003; Lenz et al., 2006; Omer et al., 2018; Xiang, 2018). In A. nidulans and Ustilago maydis, plus end dynein accumulation depends on dynactin and kinesin-1, but not NudF/LIS1 (Zhang et al., 2003, 2010; Lenz et al., 2006; Egan et al., 2012; Yao et al., 2012). This differs from the situation in budding yeast, where LIS1 is critical for dynein’s plus-end accumulation, and in mammalian cells, where both LIS1 and dynactin are critical (Lee et al., 2003; Sheeman et al., 2003; Splinter et al., 2012). Recently, the interaction between fungal dynein and early endosome has been found to be mediated by dynactin as well as the Fhip–Hook–Fts complex (Walenta et al., 2001; Xu et al., 2008; Zhang et al., 2011, 2014; Bielska et al., 2014b). Within the Fhip–Hook–Fts complex, Hook (HookA in A. nidulans and Hok1 in U. maydis) interacts with dynein–dynactin and Fhip interacts with early endosome (Bielska et al., 2014b; Yao et al., 2014; Zhang et al., 2014; Guo et al., 2016; Schroeder and Vale, 2016). The function of dynactin and the Hook complex in early endosome transport is evolutionarily conserved (although multiple Hook proteins in mammalian cells participate in even more functions of dynein; Yeh et al., 2012; Guo et al., 2016; Dwivedi et al., 2019; Olenick et al., 2019), and importantly, mammalian Hook proteins activate dynein in vitro (McKenney et al., 2014; Olenick et al., 2016; Schroeder and Vale, 2016).

Here, we developed a new assay to examine HookA-mediated dynein activation in A. nidulans and revealed the role of LIS1 in this context. Specifically, we overexpressed HookA lacking the C-terminal early endosome-binding site (ΔC-HookA), which binds dynein–dynactin, but not early endosome (Zhang et al., 2014). In contrast to the MT plus-end accumulation of dynein–dynactin in wild-type cells, ΔC-HookA overexpression shifts the accumulation to the MT minus ends. LIS1 and its binding protein, NudE, are both required for this cargo-adapter–mediated dynein relocation in vivo. Interestingly, dynein mutations that open up the autoinhibited phi conformation of dynein allow the requirement of LIS1 or NudE to be bypassed to a significant extent in vivo. We suggest that the function of LIS1 is linked to a key step of dynein activation: shifting from the autoinhibited phi conformation to an open conformation that allows dynein to be fully activated.

To examine cargo-adapter–mediated dynein activation in vivo, we sought to create A. nidulans cells where dynein and dynactin are occupied by cytosolic cargo adapters. To do that, we replaced the wild-type hookA allele with the gpdA-ΔC-hookA-S allele so that ΔC-HookA (missing its cargo-binding site) is overexpressed under the constitutive gpdA promoter (Fig. 1, A and B; and Fig. S1, A and B; Pantazopoulou and Peñalva, 2009; Zhang et al., 2011). Overexpression of ΔC-HookA did not inhibit colony growth significantly (Fig. 1 C) but caused a partial nuclear-distribution defect (Fig. S1, C and D), possibly due to a loss of dynein’s MT plus-end accumulation (see below; Xiang, 2018). In wild-type strains, GFP-labeled dynein and dynactin (p150 subunit) formed comet-like structures near the hyphal tip, representing their MT plus-end accumulation (Han et al., 2001; Zhang et al., 2003; Fig. 1, D–F). Upon overexpression of ΔC-HookA, the plus-end comets of dynein or dynactin disappeared from almost all hyphal tips (Fig. 1, D–F), although signals along MT-like tracks were seen in some hyphae (Figs. 1 D and S2 A). This was not caused by a defect in MT organization, because mCherry-labeled ClipA/Clip170 (Zeng et al., 2014) formed plus-end comets in the same cells (Fig. 1 D). A dominant feature in these cells is the accumulation of dynein and dynactin at septa, which are structures known to contain active MT-organizing centers (MTOCs; Konzack et al., 2005; Xiong and Oakley, 2009; Zekert et al., 2010; Zhang et al., 2017b; Fig. 1, E and F). In some of these cells, movement of GFP-dynein toward a septum can be observed (Fig. S2 A and Video 1). Dynein at septa was found in wild-type cells (Liu et al., 2003), but the signals were much less obvious compared with the strong accumulation upon ΔC-HookA overexpression (Fig. 1 E). Besides septal MTOCs, the nuclear envelope–associated spindle-pole body (SPB) represents the earliest-discovered MTOC (Oakley et al., 1990). Previously, we were only able to detect dynein signals at the mitotic spindle poles during anaphase (Li et al., 2005b). To determine if dynein or dynactin is at the interphase SPBs in ΔC-HookA–overexpressed cells, we used the NLS-DsRed fusion that labels nuclei during interphase (Shen and Osmani, 2013). We found clear SPB-like signals of dynein and dynactin on interphase nuclei in some ΔC-HookA–overexpressed cells (Figs. 1 G and S2 B), which were never observed in wild-type cells. However, the septal signals of dynein–dynactin in the gpdA-ΔC-hookA-S cells were brighter and more consistently observed than the SPB signals (Fig. S2 C), and thus, we used the septal signals to indicate MT minus-end accumulation in the rest of the work.

The plus-end to minus-end (MTOC) relocation of dynein–dynactin is fully consistent with cargo-adapter–mediated dynein activation observed in vitro (McKenney et al., 2014; Schlager et al., 2014; Olenick et al., 2016; Schroeder and Vale, 2016; Baumbach et al., 2017; Jha et al., 2017). In the budding yeast, dynein also uses its motor activity to move from the plus end toward the minus end upon overexpression of the coiled-coil domain of the cortical dynein anchor Num1 (Lammers and Markus, 2015). Despite these consistent data, we sought to further confirm that the relocation from the plus end to the minus end needs functional dynein in A. nidulans. To do that, we examined the effect of a previously identified dynein loss-of-function mutation, nudA^F208V. The nudA^F208V mutation in the dynein tail impairs dynein-mediated nuclear distribution and early endosome transport but does not affect dynein–dynactin interaction or dynein–early-endosome interaction (Qiu et al., 2013), consistent with the importance of the tail in dynein motor activity (Ori-McKenney et al., 2010; Rao et al., 2013; Hoang et al., 2017). Upon overexpression of ΔC-HookA, plus-end comets formed by GFP-dynein with the nudA^F208V mutation were still detected (Fig. 1 H), confirming that functional dynein is needed for this relocation.

To examine the role of NudF/LIS1 in ΔC-HookA–mediated dynein activation in vivo, we used the temperature-sensitive nudF6 mutant in which the NudF/LIS1 protein is unstable at its restrictive temperature (Xiang et al., 1995a). The nudF6 mutation (identified as nudF^L304S in this work) caused a significant defect in ΔC-HookA–activated relocation of dynein and dynactin from the MT plus ends to the septa. Specifically, while the MT plus-end comets formed by dynein or dynactin disappeared and were replaced by the septal accumulation of these proteins in gpdA-ΔC-hookA-S cells, introducing the nudF6 allele into this background caused dynein and dynactin to form plus-end comets instead of the septal accumulation (Fig. 2, A–C). Similar to nudF6, another temperature-sensitive mutation, nudF7 (Xiang et al., 1995a), and a conditional-null mutation, alcA-nudF, which allows NudF/LIS1 expression to be shut off by glucose-mediated repression at the alcA promoter (Xiang et al., 1995a), also retained dynein at the MT plus ends upon overexpression of ΔC-HookA (Fig. 2 D). Together, these results indicate that NudF/LIS1 is critically required for ΔC-HookA–mediated dynein activation in A. nidulans.

Unlike dynein or dynactin, NudF/LIS1-GFP did not accumulate at septa upon overexpression of ΔC-HookA (Fig. 3 A). Instead, the plus-end comets formed by NudF/LIS1-GFP were still observed upon overexpression of ΔC-HookA. However, the comet intensity was significantly lower than that in wild-type cells, although the NudF/LIS1-GFP protein level was not decreased apparently (Fig. 3, A–C). Thus, ΔC-HookA may drive some NudF/LIS1 proteins to leave the MT plus end with dynein–dynactin, but the association is not maintained. This is consistent with previous studies showing that dynein and dynactin associate with the early endosome undergoing dynein-mediated transport but that LIS1 dissociates from it after the initiation of transport (Lenz et al., 2006; Egan et al., 2012).

We next sought to address whether dynein can undergo cargo-adapter–mediated activation only at the MT plus ends. In hyphae of U. maydis, although many early endosomes start their dynein–dynactin–LIS1–dependent minus-end–directed transport from the MT plus end after being delivered there by kinesin-3 (Lenz et al., 2006), dynein-dependent early endosome transport often initiates in the middle of a MT before the early endosome reaches the plus end (Schuster et al., 2011). In fungi and higher eukaryotic cells, including neurons, plus-end–directed kinesins are required for the accumulation of dynein at the MT plus ends (Zhang et al., 2003; Carvalho et al., 2004; Lenz et al., 2006; Arimoto et al., 2011; Roberts et al., 2014; Twelvetrees et al., 2016). In the ΔkinA (Kinesin-1) mutant of A. nidulans (Requena et al., 2001), dynein fails to arrive at the MT plus ends but locates along MTs (Zhang et al., 2003, 2010; Egan et al., 2012). We found that GFP-dynein in cells with the ΔkinA and gpdA-ΔC-hookA-S alleles accumulated at septa (Fig. 4 A), suggesting that dynein along MTs can be activated before arriving at the plus ends as long as cargo adapters are available globally. However, adding the nudF6 mutation to the genetic background with the ΔkinA and gpdA-ΔC-hookA-S alleles caused the septal accumulation of dynein to be significantly decreased and dynein along MTs to be more obvious (Fig. 4, A and B). Thus, NudF/LIS1 is important for cargo-adapter–mediated dynein activation even when dynein is not at the MT plus ends. This is consistent with the finding that neuronal LIS1 is important for transport initiation not only in the distal axon containing dynamic MT plus ends but also in the midaxon with much more stable MTs (Moughamian et al., 2013).

We next used a biochemical pull-down assay to determine if the defect in dynein activation in cells lacking NudF/LIS1 is caused by a defect in the formation of the dynein–dynactin–ΔC-HookA complex. For this assay, we combined the ΔC-HookA-GFP fusion under the control of the endogenous hookA promoter (Zhang et al., 2014) with the alcA-nudF mutant in which NudF/LIS1 expression is shut off by glucose-mediated repression at the alcA promoter (Xiang et al., 1995a). The Δp25 mutant was used as a negative control, because p25 of dynactin is important for the formation of the dynein–dynactin–ΔC-HookA complex (Zhang et al., 2014; Qiu et al., 2018). In the alcA-nudF mutant, where NudF/LIS1 is undetectable, both dynactin and dynein were still pulled down with ΔC-HookA-GFP, and only a very mild reduction in the ratio of pulled-down dynein to ΔC-HookA compared with that of the wild type was detected (Fig. 4, C and D). Thus, the role of NudF/LIS1 in dynein activation cannot be simply explained by its involvement in the formation of the dynein-dynactin-cargo adapter complex in vivo, suggesting that NudF/LIS1 must be critical for another step of dynein activation.

A key step of dynein activation is the switch of dynein from the autoinhibited phi–dynein conformation (Torisawa et al., 2014; Zhang et al., 2017a) to an open conformation that can undergo further activation by dynactin and cargo adapter in vitro to become a processive motor (Zhang et al., 2017a). Formation of phi–dynein depends on interactions between the two dynein HCs, including the ionic interactions between the linker domain and AAA4, and mutating two linker residues opens up phi–dynein (Zhang et al., 2017a). In mammalian cells, dynein with these phi-opening mutations is enriched at centrosomes/spindle poles together with dynactin (Zhang et al., 2017a). To investigate the relationship between phi–dynein and LIS1 function, we constructed an A. nidulans strain containing the two analogous phi-opening mutations nudA^{R1602E, K1645E} (Fig. 5 A). In addition, we also obtained the nudA^R1602E mutant containing only one of the two phi-opening mutations due to a homologous recombination between the two sites. In either case, the wild-type nudA allele is replaced by the mutant allele (confirmed by sequencing of the genomic DNA). The nudA^{R1602E, K1645E} mutant showed a partial defect in nuclear distribution (Fig. S3, A and B), and it formed a colony smaller than wild type, especially at a higher temperature, such as 37°C or 42°C, but the colony phenotype of the nudA^R1602E mutant was not as obvious (Figs. 5 B and S3 C). We also found that combining gpdA-ΔC-hookA-S with nudA^{R1602E, K1645E} made the colony almost inviable (Fig. S3 D).

The nudA^{R1602E, K1645E} mutant exhibited a striking accumulation of GFP-dynein at septa together with mCherry-RabA–labeled early endosomes (Figs. 5 C and S4 A). Interestingly, GFP-dynein with the nudA^R1602E single mutation also accumulated at septa together with early endosomes (Fig. 5 C), suggesting that this mutation must have opened the phi–dynein at least partially.

As the full-length HookA is associated with early endosome, it is most likely that HookA and dynactin activate the open dynein, a notion consistent with the in vitro result that open dynein by itself is not able to walk along the MT processively without dynactin and cargo adapter (Zhang et al., 2017a). By using the ΔhookA mutant, we showed that the open dynein is able to localize to the MT plus end. Specifically, we introduced the ΔhookA allele into the strain containing nudA^{R1602E, K1645E} and found bright plus-end dynein comets in this strain (Fig. S4 B). It is also interesting to note that loss of hookA did not fully eliminate the septal dynein accumulation, although it abolished the dynein–early endosome colocalization (Fig. S4 B), consistent with the previously observed movement of early-endosome–free dynein (Schuster et al., 2011). It needs to be addressed in the future whether these septal dynein molecules have been activated by other dynein cargoes whose transport is Hook independent and mediated by adapters waiting to be identified (Peñalva et al., 2017).

To determine if the phi-opening mutations can bypass NudF/LIS1 function in vivo, we introduced these mutations into the ΔnudF (nudF-deletion) mutant background. Amazingly, both the nudA^R1602E and nudA^{R1602E, K1645E} mutations enhanced growth of the ΔnudF mutant colony (Fig. 5 B). Moreover, the nudA^{R1602E, K1645E} mutations allowed dynein and early endosomes to be seen at septa of the ΔnudF mutant (Fig. 5 D). Thus, artificially opening phi–dynein allows the requirement of NudF/LIS1 to be partially bypassed. We then performed a more detailed imaging analysis using the temperature-sensitive nudF6 mutant containing either nudA^R1602E or nudA^{R1602E, K1645E} (note that the nudF6 mutant is much healthier than ΔnudF at a lower temperature, allowing us to obtain enough spores for quantitative imaging). In the nudF6 mutant grown at its restrictive temperature, GFP-dynein with nudA^R1602E mainly formed plus-end comets, but GFP-dynein with nudA^{R1602E, K1645E} accumulated at septa together with early endosomes, even though plus-end comets did not completely disappear (Fig. 6, A–C). This result suggests that the nudA^R1602E single mutation must have only opened phi–dynein partially and is not able to compensate for the loss of NudF/LIS1 as effectively as the nudA^{R1602E, K1645E} allele.

It was found previously that phi opening enhances the formation of the mammalian dynein–dynactin–cargo-adapter complex in vitro (Zhang et al., 2017a). To determine if this happens in A. nidulans, we performed a biochemical pull-down assay using strains carrying GFP-dynein with the phi-opening mutations and the ΔC-HookA-S fusion expressed under the control of the endogenous hookA promoter. Interestingly, both nudA^R1602E and nudA^{R1602E, K1645E} caused an increase in the amount of ΔC-HookA pulled down with GFP-dynein, although the increase caused by nudA^R1602E is slightly less significant than that caused by nudA^{R1602E, K1645E} (Fig. 6, D and E). In addition, nudA^{R1602E, K1645E} caused a mild increase in the amount of dynactin pulled down with GFP-dynein, but no significant change in the amount of NudF/LIS1 pulled down was detected (Fig. 6, D and E).

Importantly, upon overexpression of ΔC-HookA, GFP-dynein with nudA^R1602 accumulated at septa in the nudF6 background, which is in sharp contrast to the plus-end accumulation of wild-type dynein in the same genetic background (Fig. 7, A and B). This result suggests that the cargo adapter works synergistically with the partial phi-opening mutation to overcome the inhibition in dynein activation caused by loss of NudF/LIS1.

Previous studies have suggested a role of NudE in enhancing LIS1 function by recruiting LIS1 to dynein (Efimov, 2003; Shu et al., 2004; Li et al., 2005a; McKenney et al., 2010; Zyłkiewicz et al., 2011; Wang et al., 2013). However, NudE was also thought to relieve the inhibition of LIS1 on dynein motility (Yamada et al., 2008). More interestingly, NudE and the p150 subunit of dynactin both bind to the N-terminus of dynein intermediate chain and may compete for the binding site (Karki and Holzbaur, 1995; Vaughan and Vallee, 1995; King et al., 2003; McKenney et al., 2011; Wang et al., 2013; Jie et al., 2017). The role of NudE in cargo-adapter–mediated dynein activation has never been addressed in vivo or in vitro. To address this, we first introduced GFP-dynein and the gpdA-ΔC-hookA-S allele into the ΔnudE background. We found that although overexpression of ΔC-HookA drives dynein relocation from the MT plus ends to the minus ends, this does not happen in the ΔnudE mutant (Fig. 8 A). Thus, just like NudF/LIS1, NudE is also required for cargo-adapter–mediated dynein activation.

Furthermore, we found that the phi-opening mutations of dynein allow the function of NudE to be bypassed to a significant extent. In the ΔnudE mutant, dynein accumulates at the MT plus end as shown previously (Efimov, 2003), and mCherry-RabA–labeled early endosomes accumulate abnormally at the hyphal tip (Fig. 8 B), as similarly observed upon loss of NudF/LIS1 (Lenz et al., 2006; Zhang et al., 2010; Egan et al., 2012). Importantly, the presence of the phi-opening mutations caused a significant (although not complete) relocation of dynein and early endosome to the septa in the ΔnudE mutant (Fig. 8 B and Fig. S5, A–C). The result that the nudA^R1602E single mutation allowed a significant septal accumulation of dynein in the ΔnudE mutant (Figs. 8 B and S5 C), but not in the nudF6 mutant (Fig. 6, A and C), is consistent with the notion that the function of NudF/LIS1 is only partially lost upon loss of NudE in fungi (Efimov and Morris, 2000; Li et al., 2005a; Efimov et al., 2006).

In this study, we developed a robust in vivo assay for cargo-adapter–mediated dynein activation, which allowed us to dissect the function and mechanism of the dynein regulator NudF/LIS1 (called LIS1 hereafter). We found that both LIS1 and its binding protein, NudE, are critical for cargo-adapter–mediated dynein activation in vivo. Remarkably, the requirement for LIS1 or NudE in vivo is bypassed to a significant extent if the autoinhibited phi–dynein is opened up artificially. Our results provide the in vivo evidence to suggest that LIS1–NudE may promote the opening of phi–dynein, a key step of dynein activation.

LIS1 is required for cytoplasmic dynein function in many different cell types (Kardon and Vale, 2009; Vallee et al., 2012; Olenick and Holzbaur, 2019). Possibly, it also regulates dynein inside cilia/flagella, where dynactin is absent, since LIS1 is associated with outer-arm dynein required for flagellar beating, and the intraflagella transport dynein-2 is regulated by a phi-like autoinhibited state (Pedersen et al., 2007; Rompolas et al., 2012; Toropova et al., 2017; Roberts, 2018). Our results on the positive role of LIS1 in dynein activation are consistent with the results that LIS1 enhances the speed and/or frequency of dynein motility to varying extents in different in vitro motility assays in the presence of dynactin and the N-terminal portion of the cargo adapter BicD2 (Baumbach et al., 2017; Gutierrez et al., 2017; Jha et al., 2017). Nevertheless, the results from A. nidulans are more striking. While the dynein–dynactin–cargo-adapter complex moves robustly toward the MT minus end in the absence of LIS1 in vitro (McKenney et al., 2014; Schlager et al., 2014), LIS1 is critical for the ΔC-HookA–activated dynein relocation in vivo. We postulate that the intracellular environment may require dynein to operate under higher tension and with more complicated regulations. For example, tension applied to the dynein linker domain in vitro alters the regulatory requirement for dynein motility (Nicholas et al., 2015), as revealed during analyzing the AAA3 domain of dynein (Bhabha et al., 2014; DeWitt et al., 2015; Nicholas et al., 2015).

In A. nidulans, dynein carrying the phi-opening mutations dramatically accumulates at septal MTOCs with early endosomes, suggesting that opening the phi–dynein must have allowed dynein to bind more robustly to dynactin and cargo adapter (Fig. 6, D and E; Zhang et al., 2017a) and/or switch to the active conformation more effectively after cargo binding. Interestingly, only the two mutations that open phi–dynein more completely can partially bypass the requirement of LIS1 function to allow dynein to be accumulated at the septa. However, the single nudA^R1602E mutation, which presumably opens phi–dynein incompletely, also allows the requirement of LIS1 for dynein activation to be partially bypassed when ΔC-HookA is overexpressed. Thus, binding of cargo adapter can further promote the open state of dynein to compensate for LIS1 loss. This is consistent with the model that dynactin and cargo adapter further switch dynein to the active state, thereby preventing the equilibrium from being shifted toward the phi state (Zhang et al., 2017a). Without dynactin and cargo adapter, LIS1 may still promote the open state, but dynein would not be fully functional to move along MTs (Zhang et al., 2017a). Both phi opening and LIS1 promote the dynein–dynactin–cargo-adapter complex formation in vitro (Baumbach et al., 2017; Zhang et al., 2017a), and LIS1 also enhances the dynein–dynactin interaction in Drosophila melanogaster and Xenopus laevis egg extracts (Dix et al., 2013; Wang et al., 2013). However, this role of LIS1 is not obvious in A. nidulans in the presence of the cytosolic ΔC-HookA expressed under the control of hookA’s endogenous promoter (Fig. 4, C and D). It is possible that the concentrations of ΔC-HookA, the dynein complex, and the dynactin complex in A. nidulans are high enough to allow the formation of the dynein–dynactin–ΔC-HookA complex without LIS1. Together, our results suggest that promoting the switch from the autoinhibited phi–dynein to open dynein rather than enhancing the dynein–dynactin–cargo-adapter complex formation per se is one key function of LIS1 in A. nidulans (Fig. 9).

Recently, LIS1’s role in promoting the switch from phi–dynein to open dynein has also been supported by results from in vitro and yeast genetic studies, and structural studies further suggest that the binding of LIS1 to the dynein motor domain at AAA3/AAA4 and/or stalk stabilizes the open dynein (Huang et al., 2012; Toropova et al., 2014; DeSantis et al., 2017; Zhang et al., 2017a; Elshenawy et al., 2019 ,Preprint; Olenick and Holzbaur, 2019; Htet et al., 2019 ,Preprint; Marzo et al., 2019 ,Preprint). This function of LIS1 would in turn facilitate the cargo-adapter–dynactin–mediated switch of dynein to a fully functional state with the two dynein HCs being in a parallel configuration (Zhang et al., 2017a). Interestingly, the LIS1-binding protein NudE is also required for dynein activation in vivo, and the requirement of NudE is also bypassed to a significant extent by the phi-opening mutations. This is consistent with the notion that NudE enhances LIS1 function. NudE binds to the dynein intermediate chain in the dynein tail, a site also required for dynactin binding (McKenney et al., 2011; Wang et al., 2013; Jie et al., 2017). As the two dynein tails are held together at several positions in the phi–dynein conformation (Fig. 9; Zhang et al., 2017a), it cannot be excluded that NudE may participate in phi opening from the tail side to promote LIS1 function. However, because overexpression of LIS1 totally compensates for the loss of NudE (Efimov, 2003), we favor the possibility that NudE is not directly involved in phi opening but simply helps bring LIS1 close to its site of action.

Since the two phi-opening mutations allow LIS1 function to be bypassed to a significant extent, but not completely, we would not rule out the possibility that LIS1 has additional roles in dynein regulation besides shifting phi–dynein toward an open conformation. This is agreeable with recent data from other laboratories, which are consistent with LIS1’s role in promoting the open conformation of dynein but show that the function of LIS1 is not fully mimicked by the phi-opening mutations (Elshenawy et al., 2019 ,Preprint; Htet et al., 2019 ,Preprint; Marzo et al., 2019 ,Preprint). We should also point out that constitutively opening up phi–dynein as achieved by the phi-opening mutations has a clear negative effect inside cells, as it causes mitotic defects in mammalian cells (Zhang et al., 2017a) and a defect in nuclear distribution in A. nidulans (Fig. S3, A and B). Thus, phi opening must be regulated for normal dynein function in vivo, and identification of the regulatory factors will be an important task in the future.

One important issue is the spatial regulation of dynein activation mediated by dynactin, cargo adapter, and LIS1 in cells. In fungal hyphae, dynein and dynactin accumulate at the MT plus end before interacting with cargo adapters, and the plus-end accumulation of LIS1 should facilitate dynein activation at the plus end. However, although the MT plus end has been considered as a cargo-loading site in various cell types (Vaughan et al., 2002; Lenz et al., 2006; Lomakin et al., 2009; Moughamian et al., 2013), dynein molecules along a MT may also be involved in cargo-adapter–mediated dynein activation that needs dynactin and LIS1 (Schuster et al., 2011; Moughamian et al., 2013). In this study, we show that the LIS1-involved dynein activation may also occur before dynein and dynactin are transported to the MT plus end by kinesin-1 when cytosolic cargo adapters are globally available (Fig. 4, A and B). In this scenario, some dynein or dynein-dynactin complexes may be bound to LIS1 before reaching the plus end. The interaction between LIS1 and dynein appears quite dynamic. LIS1 could fall off the motile dynein–dynactin–cargo-adapter complex as suggested by previous studies (Lenz et al., 2006; Egan et al., 2012; Jha et al., 2017) and recent data (Fig. 3, A-C; Elshenawy et al., 2019 ,Preprint; Htet et al., 2019 ,Preprint), although a more stable association with the motile complex has also been observed (Baumbach et al., 2017; Gutierrez et al., 2017). What regulates LIS1’s interaction with dynein during its movement needs to be further addressed.

During this work, we found an intriguing phenomenon that the septal MTOCs in A. nidulans are more consistently occupied with activated dynein than the SPBs. A previous study has suggested that the SPB-generated MTs in A. nidulans are tyrosinated and unstable during mitosis whereas the detyrosinated MTs are stable and possibly generated from the septal MTOCs (Zekert and Fischer, 2009). It would be worthwhile to further determine whether MTs from these MTOCs are modified differently, which may potentially affect MT length/stability and/or the interaction with dynein at the minus ends. Indeed, tyrosination/detyrosination affects the interaction of dynein–dynactin or kinesin-3 with MTs, although the mechanisms may not be conserved and the details differ in different organisms (Zekert and Fischer, 2009; Seidel et al., 2013; Steinberg, 2015; McKenney et al., 2016; Nirschl et al., 2016; Tas et al., 2017).

A. nidulans strains used in this study are listed in Table 1. Genetic crosses were done by standard methods, and progeny with desired genotypes were selected based on colony phenotype, imaging analysis, Western analysis, diagnostic PCR, and/or sequencing of specific regions of the genomic DNA. All images were captured using an Olympus IX73 inverted fluorescence microscope linked to a PCO/Cooke Corporation Sensicam QE cooled charge coupled device camera. A UPlanSApo 100× objective lens (oil) with a 1.40 numerical aperture was used. A filter-wheel system with GFP/mCherry-ET Sputtered series with high transmission (Biovision Technologies) was used. IPLab software was used for image acquisition and analysis. Image labeling was done using Microsoft PowerPoint and/or Adobe Photoshop. Quantitation of signal intensity was done as described previously (Zhang et al., 2014). Specifically, a region of interest (ROI) was selected and the Max/Min tool of the IPLab program was used to measure the maximal intensity within the ROI. The ROI box was then dragged outside of the cell to take the background value, which was then subtracted from the intensity value. Hyphae were chosen randomly from images acquired under the same experimental conditions. For measuring the signal intensity of a MT plus-end comet formed by GFP-dynein or NudF/LIS1-GFP proteins, only the comet closest to hyphal tip was measured. For measuring GFP-dynein signal intensity at septa, usually only the septum most proximal to the hyphal tip was measured, although sometimes, two septa close to each other in the same hypha were present, in which case both were measured. Images were taken at room temperature immediately after the cells were taken out of the incubators. Cells were cultured overnight in minimal medium with 1% glycerol and supplements at 32°C or 37°C (all experiments using nudF6 strains and controls were done at 37°C). Note that the nudF6 mutant is temperature sensitive; it forms a tiny colony lacking asexual spores at a higher temperature (typical of a nud mutant), but some spores are produced at its semipermissive temperature of 32°C. Thus, for experiments involving nudF6, we harvested spores at 32°C and cultured them at 37°C for imaging analysis. The nudF6 mutant is much better than ΔnudF for imaging analysis, because we can harvest enough spores from the nudF6 mutant at 32°C, whereas the ΔnudF mutant is sick and does not produce spores at any temperature. For a few experiments using strains containing the alcA-nudF (conditional null) allele, we harvest spores from the solid minimal medium containing 1% glycerol and cultured them in liquid minimal medium containing 0.1% glucose for imaging analysis. Yeast extract and glucose–rich medium was used for growing cells for protein pull-down experiments.

For constructing the gpdA-ΔC-hookA-S strain, we cotransformed into the A. nidulans strain XX357 the fragment containing ΔC-hookA-S-AfpyrG (Zhang et al., 2018) with another fragment containing the ∼1.2-kb gpdA promoter inserted in between the N-terminal HookA coding sequence and its upstream sequence. The gpdA promoter was inserted in this region using fusion PCR with primers 41U (5′-CAT​GCT​TGC​TTC​CTC​TTG​C-3′), HKpr2 (5′-GGA​TAT​GTC​CAA​GTA​ATC​GCT​G-3′), gpdAF2 (5′-CAG​CGA​TTA​CTT​GGA​CAT​ATC​CGA​CTC​GAG​TAC​CAT​TTA​ATT​CTA​T-3′), gpdAR2 (5′-ACG​GTA​CGC​TCC​GAC​TCC​ATT​GTG​ATG​TCT​GCT​CAA​GC-3′; these four primers were described previously; Zhang et al., 2014), HKN2 (5′-AGT​CGG​AGC​GTA​CCG​T-3′), and HKgR (5′-TCA​GCC​TCA​AGG​TTT​TGG​TTC-3′). Primers 41U and HKgR were also used for PCR to confirm the correct integration of the gpdA promoter in the selected transformants, and overexpression of the ΔC-HookA-S protein was confirmed by western analyses (Figs. 1 B and S1 B).

We used fusion PCR to make a DNA fragment of nudA containing both the R1602E and the K1645E mutations using the following primers: 1602F (5′-CGA​GCG​AGT​TCC​AGA​ATA​TCA​ACT​CAG​AAT​TCT​TCG-3′), 1602R (5′-AGT​TGA​TAT​TCT​GGA​ACT​CGC​TCG​ATT​CCA​GGG​GAA​GAA-3′), 1645F (5′-GCT​GCT​TAA​CGA​AAT​CCA​GAA​AGC​TCT​CGG​TGA​ATA​C-3′), 1645R (5′-GCT​TTC​TGG​ATT​TCG​TTA​AGC​AGC​TCG​GCC​AG-3′), NudA54 (5′-GTG​GAT​GAA​CTC​ATT​CCA​AGA-3′), and NudA36 (5′-TTG​GAT​CTA​CCA​GCA​TAG​CCA-3′). The fragment was cotransformed with a selective marker pyrG fragment into the RQ2 strain containing GFP-dynein HC (NudA) and mCherry-RabA. More than 200 pyrG+ transformants were examined under the microscope, and several of them were selected because they exhibited clear septal enrichment of both the GFP and mCherry signals. Our sequencing analysis indicated that several strains contain both mutations (nudA^{R1602E, K1645E}), but one strain contains only the nudA^R1602E mutation due to homologous recombination between the two sites.

The following oligos were used for fusion PCR to create the gpdA-nudF-GFP-AfpyrG fragment: F5F (5′-ATC​AGA​CTG​GAC​GAA​GCC-3′), F5R (5′-CAA​ATA​GAA​TTA​AAT​GGT​ACT​CGA​GTC​GGT​TGT​TTG​TGT​TCG​CAA​AT-3′), gpdF (5′-GAC​TCG​AGT​ACC​ATT​TAA​TTC​TAT​TTG-3′), gpdR (5′-TGT​GAT​GTC​TGC​TCA​AGC​G-3′), FF (5′-CGC​TTG​AGC​AGA​CAT​CAC​AAT​GAG​CCA​AAT​ATT​GAC​AGC​TCC-3′), FR (5′-GCC​TGC​ACC​AGC​TCC​GCT​GAA​CAC​CCG​TAC​AGA​GTT-3′), GFPF (5′-GGA​GCT​GGT​GCA​GGC-3′), GFPR (5′-CTG​TCT​GAG​AGG​AGG​CAC​TG-3′), F3F (5′-CAG​TGC​CTC​CTC​TCA​GAC​AGG​TCG​CGA​TCT​TCA​TCA​CAG​TT-3′), and F3R (5′-CGA​CAG​AAT​GGA​ACG​GGA​AA-3′). This fragment was transformed into the RQ54 strain, and progeny with MT plus-end comets formed by NudF/LIS1-GFP were selected. For this study, we only used a transformant (RQ165) containing NudF-GFP, but not the gpdA-NudF-GFP fusion protein.

The μMACS GFP-tagged protein isolation kit (Miltenyi Biotec) was used to pull down proteins associated with the GFP-tagged protein. This was done as described previously (Zhang et al., 2014). Specifically, ∼0.4 g of hyphal mass was harvested from overnight culture for each sample, and cell extracts were prepared using a lysis buffer containing 50 mM Tris-HCl, pH 8.0, and 10 µg/ml of a protease inhibitor cocktail (Sigma-Aldrich). Cell extracts were centrifuged at 8,000 g for 15 min and then 16,000 g for 15 min at 4°C, and supernatant was used for the pull-down experiment. To pull down GFP-tagged proteins, 25 µl anti-GFP MicroBeads was added into the cell extracts for each sample and incubated at 4°C for 30–40 min. The MicroBeads/cell extracts mixture was then applied to the μColumn followed by gentle wash with the lysis buffer used above for protein extraction (Miltenyi Biotec). Preheated (95°C) SDS-PAGE sample buffer was used as elution buffer. Western analyses were performed using the alkaline phosphatase system, and blots were developed using AP color development reagents (Bio-Rad). Quantitation of the protein band intensity was done using IPLab software as described previously (Qiu et al., 2013). Specifically, an area containing the whole band was selected as a ROI, and the intensity sum within the ROI was measured. Then, the ROI box was dragged to the equivalent region of the negative control lane or a blank region without any band on the same blot to take the background value, which was then subtracted from the intensity sum. The rabbit polyclonal antibody against GFP (used for Western blots presented in Figs. 3 C, 4 C, and 6 D) was purchased from Takara Bio (catalog number 632592). The rabbit monoclonal antibody against the S-tag (used for Western blots presented in Figs. 1 B, S1 B, and 6 D) was from Cell Signaling Technology (catalog number 12774S). Polyclonal antibodies against dynein HC (Fig. 4 C), dynactin p150 (Figs. 4 C and 6 D), and NudF/LIS1 (Figs. 4 C and 6 D) were generated in previous studies by injecting proteins produced in bacteria into rabbits followed by affinity purification of the antibodies (Xiang et al., 1995a,b; Zhang et al., 2008).

All statistical analyses were done using GraphPad Prism 8 for Mac (version 8.0.0, 2018). The D’Agostino and Pearson normality test was performed on all datasets except Western blot datasets with small n (n = 3 or n = 4). For Western blot data quantitation presented in Fig. 4 D and Fig. 6 E, data distribution was assumed to be normal, but this was not formally tested. A Student’s t test (unpaired, two tailed) was used to analyze data in Fig. 4 D, and an ordinary one-way ANOVA (unpaired) was used to analyze data in Fig. 6 E. The datasets presented in Fig. 4 B passed the D’Agostino and Pearson normality test (α = 0.05), and thus, they were analyzed using a Student’s t test (unpaired, two tailed). For all other datasets, nonparametric tests were used without assuming Gaussian distribution. Specifically, the Kruskal–Wallis ANOVA test (unpaired) with Dunn’s multiple comparisons test was used for analyzing multiple datasets presented in Fig. 2 B; Fig. 3 B; Fig. 6, B and C; Fig. 7 B; Fig. S1 D; and Fig. S5, B and C. The Mann–Whitney test (unpaired, two tailed) was used to analyze the two datasets presented in Fig. S3 B. Note that adjusted P values were generated from either the ordinary one-way ANOVA test or the Kruskal–Wallis ANOVA test with Dunn’s multiple comparisons test. In all figures, **** indicates P < 0.0001, *** indicates P < 0.001, ** indicates P < 0.01, and * indicates P < 0.05. If the P value is >0.05, the difference is considered not significant, which is not indicated in any figure.

Fig. S1 shows a western analysis of the HookA-S and ΔC-HookA-S proteins as well as the partial defect in nuclear distribution caused by ΔC-HookA-S overexpression. Fig. S2 shows the movement of GFP-dynein toward a septum and dynactin localization at the SPBs, with the accumulation of dynein being more obvious at septa than at the SPBs upon ΔC-HookA-S overexpression. Fig. S3 shows the nuclear-distribution and colony-growth phenotypes of the phi-opening mutants of dynein. Fig. S4 shows the accumulation of early endosomes at septa, but not on nuclei, in the phi-opening mutant as well as the localization of open dynein in the ΔhookA mutant. Fig. S5 shows a quantitative image analysis on the localization of open dynein in the ΔnudE mutant. Video 1 shows the movement of GFP-dynein toward a septum upon ΔC-HookA overexpression.

We thank Reinhard Fischer (Karlsruhe Institute of Technology, Karlsruhe, Germany), Bo Liu (University of California, Davis, Davis, CA), Berl Oakley (University of Kansas, Lawrence, KS), Aysha Osmani (The Ohio State University, Columbus, OH), Stephen Osmani (The Ohio State University, Columbus, OH), Martin Egan (University of Arkansas, Fayetteville, AR), Samara Reck-Peterson (University of California, San Diego, San Diego, CA), and Miguel Peñalva (Centro de Investigaciones Biológicas CSIC, Madrid, Spain) for sharing Aspergillus strains. We thank Stephen Osmani and Tian Jin for critical comments on the manuscript and Ahmet Yildiz and Samara Reck-Peterson for sharing unpublished results on LIS1.

This work was funded by the National Institutes of Health (grant R01GM121850-01A1 to X. Xiang).

The authors declare no competing financial interests.

Author contributions: R. Qiu, J. Zhang, and X. Xiang designed the experiments, performed the experiments, and analyzed the data. X. Xiang wrote the paper with edits from R. Qiu and J. Zhang.