Article Figures and data Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Kinesin superfamily proteins are microtubule-based molecular motors driven by the energy of ATP hydrolysis. Among them, the kinesin-4 family is a unique motor that inhibits microtubule dynamics. Although mutations of kinesin-4 cause several diseases, its molecular mechanism is unclear because of the difficulty of visualizing the high-resolution structure of kinesin-4 working at the microtubule plus-end. Here, we report that KLP-12, a C. elegans kinesin-4 ortholog of KIF21A and KIF21B, is essential for proper length control of C. elegans axons, and its motor domain represses microtubule polymerization in vitro. The crystal structure of the KLP-12 motor domain complexed with tubulin, which represents the high-resolution structural snapshot of the inhibition state of microtubule-end dynamics, revealed the bending effect of KLP-12 for tubulin. Comparison with the KIF5B-tubulin and KIF2C-tubulin complexes, which represent the elongation and shrinking forms of microtubule ends, respectively, showed the curvature of tubulin introduced by KLP-12 is in between them. Taken together, KLP-12 controls the proper length of axons by modulating the curvature of the microtubule ends to inhibit the microtubule dynamics. Editor's evaluation In their study, Taguchi et al. aim to determine how a member of the kinesin-4 family is able to stabilize the tips of microtubules to suppress both their growth and shrinkage, a process important for normal development. This paper provides convincing data on KLP-12 by combining in vivo C. elegans work with in vitro single-molecule analysis and structural studies of the motor domain. The structure shows that KLP-12 bends tubulin heterodimers to a level that lies in between the extremes of bending by KIF5B (lattice stabilizer) and KIF2C (lattice destabilizer). This important study will be of interest to those in the fields of neuronal development and cytoskeletal dynamics. https://doi.org/10.7554/eLife.77877.sa0 Decision letter Reviews on Sciety eLife's review process eLife digest From meter-long structures that allow nerve cells to stretch across a body to miniscule 'hairs' required for lung cells to clear mucus, many life processes rely on cells sporting projections which have the right size for their role. Networks of hollow filaments known as microtubules shape these structures and ensure that they have the appropriate dimensions. Controlling the length of microtubules is therefore essential for organisms, yet how this process takes place is still not fully elucidated. Previous research has shown that microtubules continue to grow when their end is straight but stop when it is curved. A family of molecular motors known as kinesin-4 participate in this process, but the exact mechanisms at play remain unclear. To investigate, Tuguchi, Nakano, Imasaki et al. focused on the KLP-12 protein, a kinesin-4 equivalent which helps to controls the length of microtubules in the tiny worm Caenorhabditis elegans. They performed genetic manipulations and imaged the interactions between KLP-12 and the growing end of a microtubule using X-ray crystallography. This revealed that KLP-12 controls the length of neurons by inhibiting microtubule growth. It does so by modulating the curvature of the growing end of the filament to suppress its extension. A 'snapshot' of KLP-12 binding to a microtubule at the resolution of the atom revealed exactly how the protein helps to bend the end of the filament to prevent it from growing further. These results will help to understand how nerve cells are shaped. This may also provide insights into the molecular mechanisms for various neurodegenerative disorders caused by problems with the human equivalents of KLP-12, potentially leading to new therapies. Introduction Kinesin superfamily proteins (KIFs) are microtubule-based molecular motors driven by the energy of ATP hydrolysis (Hirokawa et al., 2009a). Most KIFs move along microtubules to transport various cargos, including membranous organelles, protein complexes, and mRNAs (Guedes-Dias and Holzbaur, 2019; Hirokawa et al., 2009b). In addition to transporting cargos, some kinesins possess the ability to regulate microtubule dynamics, such as elongation (polymerization), catastrophe or shrinkage (depolymerization), in diverse ways. Kinesin-1, the founding member of the KIFs, changes the conformation of unstable GDP microtubules into a conformation resembling the GTP microtubules (Muto et al., 2005; Shima et al., 2018). Conversely, kinesin-13 destabilizes both the plus and minus ends of microtubules to induce catastrophe or depolymerization (Desai et al., 1999; Ogawa et al., 2004). Kinesin-8 moves processively toward the microtubule plus-end, where it depolymerizes microtubule (Gupta et al., 2006; Niwa et al., 2012; Varga et al., 2006; Wang et al., 2016). Kinesin-4, another family of kinesins, is known to inhibit microtubule dynamics and is classified into three subfamilies: the KIF4 subfamily, the KIF7 subfamily, and the KIF21 subfamily (Yue et al., 2018). The KIF21 subfamily has attracted considerable attention because its genetic alterations are linked with several diseases. Point mutations of KIF21A cause congenital fibrosis of the extraocular muscle type 1 (CFEOM1) (Yamada et al., 2003). Polymorphisms in the KIF21B gene are associated with several inflammatory diseases, such as multiple sclerosis or Crohn's disease (Barrett et al., 2008; Goris et al., 2010). Increased expression of KIF21B is also linked to the progression of neurodegenerative disorder (Kreft et al., 2014). Kif21b knockout mice were reported to exhibit behavioral changes involving impaired learning and memory (Muhia et al., 2016). The molecular mechanisms of how kinesin-4 affects microtubule dynamics have been studied for more than a dozen years, demonstrating the main contribution of their motor domains to microtubule dynamics inhibition. Xklp1/KIF4, a fast processive motor, was first reported to reduce both microtubule growth and the catastrophe rate (Bieling et al., 2010; Bringmann et al., 2004). Its motor domain is able to bind not only to the microtubule lattices for microtubule-based motility but also to the curved tubulin dimers for inhibition of microtubule dynamics. Nonmotile KIF7 was reported to reduce the microtubule growth rate but enhance catastrophe to organize the tips of ciliary microtubules (He et al., 2014; Yue et al., 2018). The processive motor KIF21A/B reduces microtubule growth and catastrophes similar to Xklp1/KIF4 (van der Vaart et al., 2013; van Riel et al., 2017). These studies suggest that the motor domains of kinesin-4 family proteins play a crucial role in reducing the growth rate of microtubules. In other words, minor alterations in kinesin-4 motor domains affect their conserved functions to suppress microtubule dynamics by displaying strikingly distinct motility characteristics (van der Vaart et al., 2013; van Riel et al., 2017). The other domains of kinesin-4 are also known to be involved in microtubule dynamics inhibition by regulating or supporting motor domain functions. The coiled-coil region of KIF21A in which CFEOM1-associated mutations are localized operates as an autoinhibitory domain by interaction with the motor domain (Bianchi et al., 2016; Cheng et al., 2014; van der Vaart et al., 2013). The dominant character of CFEOM1 syndrome is thus connected to the increased activity of the mutant KIF21A kinesin caused by the loss of autoinhibition. The WD40 domain of KIF21B holds on to the growing microtubule tip and induces its pausing (van Riel et al., 2017), which is required for the sustained action of the motor domain on the microtubule plus-end to inhibit microtubule dynamics. We previously reported the first crystal structure of the KIF4 motor domain (Chang et al., 2013) and showed molecular mechanisms of ATP-induced motion; however, because of the lack of functional and structural information on kinesin-4 on the microtubule plus-end, the mechanism by which kinesin-4 motors inhibit microtubule dynamics is still obscure. Here, we investigated the functional and structural analyses of microtubule dynamics inhibition by kinesin-4 KLP-12, a Caenorhabditis elegans (C. elegans) ortholog of KIF21A and KIF21B (Figure 1A; Figure 1—figure supplement 1). Genetic analyses and in vitro TIRF (Total Internal Reflection Fluorescence microscopy) assays showed that KLP-12 regulates axonal length through inhibiting microtubule dynamics at its plus-end, similar to KIF21A and KIF21B. The crystal structure of KLP-12 complexed with curved α-, β- tubulin dimers suggested the structural model of microtubule dynamics inhibition by the kinesin-4 motor domain; kinesin-4 precisely controls the curvature of tubulin dimers at the plus-end, which is larger than that decorated by plus-end stabilizing kinesin-1 and smaller than that decorated by destabilizing kinesin-13. This precise control was achieved by the specific interactions on the microtubule interfaces conserved among kinesin-4 motors. Figure 1 with 2 supplements see all Download asset Open asset KLP-12 is an ortholog of KIF21A and KIF21B that regulates axonal length. (A) Schematic presentation of the domain organization of the KIF21 subfamily: C. elegans KLP-12, human KIF21A, and KIF21B consist of a motor domain (MD; magenta), neck linker (blue), coiled-coil domains (CC; black), and WD40 domain (WD40; gray). Phylogenetic tree and sequence alignment of kinesin-4 family are available in Figure 1—figure supplement 1 and Figure 1—figure supplement 2, respectively. (B) Schematic presentation of the genomic structure of C. elegans klp-12 and mutant alleles used in this study. tm10890 is a deletion mutant that induces frameshift, whereas tm5176 deletes WD repeats domain. (C) Schematic presentation of the mechanosensory neurons in C. elegans. Areas observed in panels (D) and (F) are shown by magenta boxes. ALM: anterior lateral mechanosensory, AVM: anterior ventral mechanosensory, PLM: posterior lateral mechanosensory, and PVM: posterior ventral mechanosensory neurons. (D and E) The tiling of ALM and PLM neurons. (D) Representative images of ALM and PLM neurons in wild type and klp-12 mutant worms. The axonal tip of PLM neurons does not overlap with the cell body of ALM neurons in wild type, while the axonal tip of PLM neuron overlaps with the ALM cell body in klp-12(tm5176) and klp-12(tm10890) mutant alleles. Bars, 50 µm. (E) The percentage of ALM and PLM overlap in wild-type, klp-12(tm5176), and klp-12(tm10890) mutant alleles in day 3 adult worms. n=55 in wild type, 60 in klp-12(tm5176), and 58 in klp-12(tm10890) worms. (F and G) Overexpression of KLP-12 suppresses the elongation of mechanosensory axon. (F) The morphology of wild type and klp-12-overexpressed PLM neurites. Bar, 50 µm. (G) The lengths of PLM neurites are plotted. Each dots show the length of axons in each worm. Bars represent mean ± standard deviation. n=10 in wild-type and 12 in KLP-12-expressing axons. ****, p<0.0001, Welch's t test. Figure 1—source data 1 The length of neurons in wild-type (WT) and KLP-12-overexpressing (OE) C. elegans. https://cdn.elifesciences.org/articles/77877/elife-77877-fig1-data1-v1.xlsx Download elife-77877-fig1-data1-v1.xlsx Results KLP-12 regulates the length of axons in C. elegans neurons KLP-12 is predicted a worm orthologue of KIF21A and KIF21B, which regulates axonal length. However, the function of KLP-12 remains to be elusive. Thus, we firstly analyzed the phenotype of klp-12 mutants. We used two independent deletion mutant alleles of klp-12, klp-12(tm10890) and klp-12(tm5176) (Figure 1B). klp-12(tm10890) was considered to be a null allele because the mutation induces deletion of exon 4–6, resulting in a frameshift. klp-12(tm5176) had a deletion mutation in exons encoding the tail domain. We observed the development of two mechanosensory neurons, anterior lateral mechanosensory (ALM) and posterior lateral mechanosensory (PLM) neurons (Figure 1C) because the tiling between PLM and ALM neurons are strictly regulated by microtubule regulating factors, such as kinesin-13. Previous studies have shown worm mutants with more stable microtubules have defects in tiling between PLM and ALM (Puri et al., 2021). In wild-type animals, the axonal tip of PLM neurons does not reach the cell body of ALM without overlapping with each other (wild-type in Figure 1D and E; Gallegos and Bargmann, 2004). On the other hand, more than 30% of the PLM axons in C. elegans with klp-12(tm5176) overtook the cell body of ALM (klp-12(tm5176) in Figure 1D and E). klp-12(tm10890) showed a more severe phenotype; more than 50% of neurons overlapped, and thin warped axons were observed (klp-12(tm10890) in Figure 1D and E). We also investigated the effect of overexpression of wild-type KLP-12 in C. elegans neurons. Compared to the wild type, the PLM axon overexpressing KLP-12 became strikingly shorter and thinner (Figure 1F and G). Together with these results, the appropriate activity of KLP-12 is necessary to achieve proper length control of axons, suggesting that the function of KIF21/KLP-12 family proteins (Figure 1A) is evolutionarily conserved. KLP-12 is a plus-end directed motor that represses microtubule polymerization Mammalian orthologs of KLP-12, KIF21A, and KIF21B, regulate the axon length by inhibiting the microtubule polymerization (van der Vaart et al., 2013; van Riel et al., 2017). Thus, KLP-12 may also inhibit the microtubule polymerization to restrict the length of axons. To directly visualize the KLP-12 function on the microtubules, we performed the in vitro TIRF assays. Since the neck-coiled-coil sequence of KLP-12 is not sufficient to form a stable dimer to display the processive motility in vitro, we intended to introduce the leucine zipper (LZ) of GCN4 and GFP right after KLP-12 (1-393) for dimerization (Tomishige et al., 2002), which is a similar technique used in the previous kinesin-4 family motor study (Yue et al., 2018; Figure 2A: KLP-12–LZ–GFP). Single KLP-12–LZ–GFP motors successfully showed plus-end directed processive movements on GDP microtubules stabilized by paclitaxel (taxol-stabilized microtubules) with an average velocity of 0.81±0.32 µm/s and an average run length of 1.30±0.89 µm, which is a similar speed as the processive fast motor kinesin-1 or KIF4 (Figure 2B–D; Shima et al., 2018; Yue et al., 2018). The average velocity and run length of KLP-12–LZ–GFP motors were also similar on dynamic GTP microtubules (Figure 2C–D). Figure 2 with 1 supplement see all Download asset Open asset KLP-12 is a plus-end directed motor that represses microtubule polymerization. (A) Schematic presentation of the KLP-12 constructs. KLP-12(FL): full-length KLP-12, KLP-12–LZ–GFP: KLP-12 (1-393) with GFP connected with a leucine zipper, KLP-12(M): KLP-12 motor domain (1-365), KLP-12–DARPin: KLP-12(M) with DARPin connected with a flexible linker. (B) A representative kymograph showing the motility of KLP-12–LZ–GFP on taxol-stabilized microtubules. Horizontal and vertical bars show 10 μm and 10 s, respectively. (C) Histogram showing the velocity of KLP-12–LZ–GFP on taxol-stabilized (green) or dynamic (magenta) microtubules. 0.81±0.32 µm/s (n=407) and 0.82±0.31 µm/s (n=215) on taxol-stabilized and dynamic microtubules, respectively. Mean ± standard deviation. No statistically significant differences were detected by Student's t-test. (D) Histogram showing the run length of KLP-12–LZ–GFP on taxol-stabilized (green) or dynamic (magenta) microtubules. n=407 molecules. 1.30±0.89 µm (n=407) and 1.11±0.57 µm (n=215) on taxol-stabilized and dynamic microtubules, respectively. Mean ± standard deviation. No statistically significant differences were detected by Student's t-test. (E) Representative kymographs showing the microtubule dynamics and the motility of KLP-12–LZ–GFP. 10 μM of fluorescently labeled microtubules were polymerized from GMPCPP stabilized microtubule seeds fixed on the cover glass in the presence of 0, 60, or 600 nM KLP-12–LZ–GFP at 30 °C. Horizontal and vertical bars show 5 μm and 60 s, respectively. (F–I) The effect of KLP-12–LZ–GFP on microtubule dynamics. 10 μM of fluorescently labeled microtubules were observed in the presence of indicated concentrations of KLP-12–LZ–GFP at 30 °C. (F) Microtubule growth rate in vitro in the presence of KLP-12–LZ–GFP. Green bars show mean ± standard deviation. **, Adjusted p=0.0022, ***, Adjusted p=0.0001, ****, Adjusted p<0.0001, compared with control (0 nM). One-way ANOVA followed by Dunnett's multiple comparisons test. n=52 microtubules. (G) Frequency of microtubule catastrophe events. The number of microtubule catastrophe in vitro was normalized by minute. mean ± standard deviation. ****, Adjusted p<0.0001, compared with control (0 nM). One-way ANOVA followed by Dunnett's multiple comparisons test. n=101 microtubules. (H) Microtubule depolymerization rate in vitro in the presence of KLP-12–LZ–GFP. Green bars show mean ± standard deviation. No statistically significant differences were detected by One-Way ANOVA. n=30 microtubules. (I) Frequency of microtubule rescue events. The number of microtubule rescue events in vitro was normalized by minute. ****, Adjusted p<0.0001, compared with control (0 nM). One-way ANOVA followed by Dunnett's multiple comparisons test. n=99 microtubules. The effect of microtubule growth rate by kinesin-4 family motors is available in Figure 2—figure supplement 1. Figure 2—source data 1 Source data of microtubule growth rate in vitro in the presence of KLP-12–LZ–GFP (Figure 2C, D, F, G, H and I). https://cdn.elifesciences.org/articles/77877/elife-77877-fig2-data1-v1.xlsx Download elife-77877-fig2-data1-v1.xlsx Next, we observed the microtubule dynamics and the localization of KLP-12–LZ–GFP at a series of different concentrations (Figure 2E). KLP-12–LZ–GFP did not accumulate the plus-end tips of microtubules but detached from them upon arrival of the tips, consistent with the previous results of KIF21A and KIF21B (van der Vaart et al., 2013; van Riel et al., 2017). KLP-12–LZ–GFP represented a concentration dependent decrease of the microtubule growth rate at its plus-end. In the absence of the KLP-12–LZ–GFP, the microtubule grew at the rate of 1.19±0.35 µm/min (Figure 2F). Increasing amounts of KLP-12–LZ–GFP inhibited the microtubule growth rate, with mean rates of 1.00 µm/min at concentrations of 30 nM and 0.77 µm/min at 300 nM, respectively (Figure 2F). On the other hand, microtubule depolymerization rate was not affected by KLP-12–LZ–GFP (Figure 2H). The frequencies of microtubule catastrophe events and rescue events were slightly reduced in the presence of KLP-12–LZ–GFP (Figure 2G1). It rises two possibilities; one is that KLP-12 reduces microtubule growth with a longer period of stabilization, and another is the indirect effect induced by reduced MT catastrophe events. We further compared the inhibitory effect of KLP-12 with the previously reported dynamics with other kinesin-4 family motors, illustrating that the growth inhibition rate of KLP-12–LZ–GFP was very similar to that of KIF21A (Figure 2—figure supplement 1), a closely related family with 59.9% sequence identity to KLP-12 (Figure 1—figure supplement 1). These results showed that KLP-12–LZ–GFP possesses a suppression effect on the microtubule growth rates similar to the other members of the kinesin-4 family proteins, especially to KIF21A/B. The motor domain of KLP-12 binds to both the microtubule lattice and ends to catalyze ATP Kinesin-4 moves along the microtubule until it reaches the plus-end, at which it inhibits the attachment and release of tubulin-dimers to/from the microtubule end. To achieve these dual functions, kinesin-4 must bind to the microtubule lattice and the microtubule end to catalyze ATP. We thus next focused on the monomeric motor domain of the KLP-12 (KLP-12(M)) (Figure 2A), investigating its ATPase activity in the presence of microtubules or tubulin heterodimers. We used GTP- and GDP- tubulin heterodimers to investigate the biochemical properties of KLP-12 at the plus-end of microtubules since the plus-end of microtubules is curved due to the lack of lateral interactions between protofilaments, as reported previously (Hunter et al., 2003). Steady-state ATPase kinetics of KLP-12(M) were examined in the presence of taxol-stabilized microtubules (=microtubule lattice), GTP-tubulins (mimic of the growing microtubule end), and GDP-tubulins (mimic of the shrinking microtubule end) (Figure 3A). The basal ATPase activity of KLP-12(M) in the absence of tubulins or microtubules was 0.0039±0.00095 s–1, comparable with the other kinesin motors (Hunter et al., 2003; Wang et al., 2016). The ATPase activity of KLP-12(M) was stimulated ~33 times by microtubules to reach a maximum rate of 0.13 s–1. The stimulation by free GTP- and GDP- tubulin dimers reached more than ~69 and~100 times to achieve a maximum rate of 0.27 and 0.39 s–1, respectively. KM,microtubules, KM,GTP-tubulin, and KM,GDP-tubulin were 2.6 μM, 5.6 μM, and 22.2 μM, respectively (Figure 3A). These results indicate that KLP-12(M) binds similarly to the microtubules and the GTP-tubulin-dimers but shows considerably weaker binding to the GDP-tubulin heterodimers than to the microtubules or the GTP-tubulin heterodimers. We should note that the microtubule-activated ATPase rate was significantly lower than expected from the KLP-12–LZ–GFP velocity (Figure 2B). We thus examined the microtubule- and tubulin- activated ATPase rates of KLP-12–LZ–GFP, resulting in similar rates with those of KLP12(M), albeit ~ten times higher affinity to the tubulin or microtubule (Figure 3—figure supplement 1). We concluded that KLP-12 binds to both the microtubule lattice and the growing microtubule plus-end and activates its ATPase, thus achieving microtubule motility and growing microtubule stabilization. The discrepancy between the ATPase rate and the velocity is discussed in the Discussion section. Figure 3 with 2 supplements see all Download asset Open asset Crystal structure of KLP-12-tubulin complex. (A) The steady-state ATPase activity of KLP-12(M) measured with GDP tubulin heterodimers, GTP tubulin heterodimers, and microtubules at 30 °C. Error bars represent standard deviation. Tubulin or microtubule GTPase effect was canceled by subtracting control without KLP-12(M). (B) Size exclusion chromatography (SEC) of tubulin mixed with KLP-12(M) and DARPin. (C) SDS–PAGE analysis of the SEC peaks of tubulin mixed with KLP-12(M) and DARPin. SDS-PAGE analysis of the SEC peaks of tubulin combined with KLP-12–DARPin is available in Figure 3—figure supplement 2. (D) Crystal structure of the tubulin–KLP-12–DARPin complex. Disordered linkers were drawn as a dotted line. α-tubulin is colored in light green, β-tubulin is in light blue, DARPin is in yellow, and KLP-12 is in magenta. See also Video 1 for details. (E) KLP-12 structure viewed from the interface with tubulin. The residues of the important structure are shown in color. β5a-L8-β5b is green, switch I is yellow, switch II is blue, α6 is red, and the neck linker is cyan. (F) Nucleotide binding pocket of KLP-12. The 2Fo-Fc map around AMP-PNP was calculated with coefficient 2Fo− Fc and contoured at 2.0 σ. Figure 3—source data 1 The ATPase activity of KLP-12(M) with microtubule. https://cdn.elifesciences.org/articles/77877/elife-77877-fig3-data1-v1.xlsx Download elife-77877-fig3-data1-v1.xlsx Figure 3—source data 2 The ATPase activity of KLP-12(M) with GTP tubulin. https://cdn.elifesciences.org/articles/77877/elife-77877-fig3-data2-v1.xlsx Download elife-77877-fig3-data2-v1.xlsx Figure 3—source data 3 The ATPase activity of KLP-12(M) with GDP tubulin. https://cdn.elifesciences.org/articles/77877/elife-77877-fig3-data3-v1.xlsx Download elife-77877-fig3-data3-v1.xlsx Figure 3—source data 4 The ATPase activity of KLP-12(M). https://cdn.elifesciences.org/articles/77877/elife-77877-fig3-data4-v1.xlsx Download elife-77877-fig3-data4-v1.xlsx Figure 3—source data 5 SDS–PAGE gel of the SEC peaks of tubulin mixed with KLP-12(M) and DARPin. https://cdn.elifesciences.org/articles/77877/elife-77877-fig3-data5-v1.pdf Download elife-77877-fig3-data5-v1.pdf Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Crystal structure of KLP-12-tubulin complex. Structure determination of the tubulin–KLP-12–DARPin complex The binding ability of the KLP-12(M) to the soluble GTP-tubulin heterodimer was further analyzed by size exclusion chromatography (SEC). To prevent the self-assembly of tubulins, a designed ankyrin repeat protein (DARPin) that binds to the longitudinal interface of β-tubulin was used (Ahmad et al., 2016). Equimolar tubulin dimers, KLP-12(M) with the ATP analog AMP-PNP (adenylyl-imidodiphosphate), and DARPin were injected and analyzed by SEC representing three peaks (Figure 3B). The first peak consisted of all components, tubulin, KLP-12(M), and DARPin (Figure 3C). The second and third peaks correspond to KLP-12(M) and DARPin, respectively. KLP-12(M) and DARPin shifted to the left side to form a triple complex with tubulin. Thus, KLP-12(M) has a binding ability to a soluble GTP-tubulin heterodimer, consistent with steady-state ATPase assays. Next, we proceeded to crystallize the KLP-12–GTP-tubulin complex to elucidate the molecular mechanism of microtubule stabilization by kinesin-4. For crystallization, to stabilize complex formation between KLP-12(M) and tubulin, KLP-12(M) was fused to DARPin by a long linker. According to a previous study (Wang et al., 2017), the linker length between the C-terminus of KLP-12(M) and the N-terminus of DARPin was optimized. Four G4S repeats with one G2S (KLP-12–DARPin fusion construct) were used to form the tubulin–KLP-12–DARPin complex (Figure 2A). Finally, we formed a stable 1:1 complex of KLP-12–DARPin and crystallized GTP-tubulin dimers for structure determination (Figure 3—figure supplement 2). We solved the crystal structure of the GTP-tubulin–KLP-12–DARPin complex at 2.9 Å resolution (Figure 3D and E; Table 1). KLP-12 forms an ATP conformation in which switches I and II take the closed conformation to hydrolyze ATP, and the neck linker docks to the motor core. The linker between KLP-12 and DARPin, which follows the neck linker, was not observed because of its intrinsic flexibility. In the nucleotide-binding pocket of KLP-12, the density corresponding to AMP-PNP was found with the Mg2+ ion (Figure 3F). The highly conserved Ser217 of switch I and Gly264 of switch II are coordinated to the γ-phosphate of AMP-PNP. The back door between switch I Arg218 and switch II Glu266 was also closed, representing the pre-hydrolysis state during ATP hydrolysis. Table 1 Data collection and refinement statistics. Tubulin–KLP-12–DARPinBeam lineSPring-8 BL32XUData collection Space groupP 21 Cell dimensionsa, b, c (Å)82.33, 80.98, 117.75α, β, γ (°)90.00, 93.51, 90.00 Resolution (Å)50–2.88 (3.05–2.88) * Rmeas (%)35.5 (197.0) I / σI19.28 (3.39) CC1/297.7 (51.3) Completeness (%)98.2 (98.8) Redundancy87.1 (88.7)Refinement Resolution (Å)49.25–2.88 No. reflections33796 Rwork/Rfree0.211/0.296 No. atomsProtein10301Ligand/ion93Water0 B-factorsProtein54.67Ligand/ion42.97Water0 R.m.s. deviationsBond lengths (Å)0.010Bond angles (°)1.40 * Values in parentheses are for the highest-resolution shell. Structural comparisons among three types of kinesin motors, kinesin-4, kinesin-1, and kinesin-13 To investigate the structural differences of KLP-12 among kinesin-4 subfamily and kinesin superfamily proteins (KIFs), the crystal structure of KLP-12(M) was compared to the structures of the previously reported kinesin-4 members Mus musculus KIF4 (PDB ID: 3ZFC) (Chang et al., 2013) and KIF7 (PDB ID: 6MLR) (Jiang et al., 2019), as well as the well-studied plus-end motor Homo sapiens kinesin-1 (KIF5B; PDB ID: 1MKJ, 4HNA, 3J8Y) (Sindelar et al., 2002; Gigant et al., 2013; Shang et al., 2014), and the microtubule destabilizer Homo sapiens kinesin-13 (KIF2C; PDB ID: 5MIO) (Wang et al., 2017; Figure 4A). The degree of similarity was estimated by comparing RMSDs of main chain residues of KLP-12 with other motors (Figure 4 and Figure 4—figure supplement 1A). Note that all structures are in the ATP or ATP-like conformation but structures were determined in different conditions; KLP-12, KIF5B (4HNA), and KIF2C are bound to tubulin dimers, KIF4 and KIF5B (1MKJ) are the motor domain only, and KIF7 and KIF5B (3J8Y) are bound to the microtubule. Figure 4 with 1 supplement see all Download asset Open asset Structural comparison of KLP-12 motor domain with the other kinesins. (A) Structural comparison of KLP-12 with the other reported kinesin motor domain structures in various states. (B) Superimposition of KLP-12 and KIF4 (Light green). (C) Superimposition of KLP-12 and KIF7 (Pink). (D) Superimposition of KLP-12 and KIF5B (Green). (E) Superimposition of KLP-12 and KIF2C (Cyan). The structure of the KLP-12 (KIF21 subfamily) is very similar to KIF4 among the kinesin-4 family members (Figure 4B). Almost no structural differences exist in the main chains, including nucleotide-dependent switch regions, except for the passive conformational changes in α1a and β5a-L8-β5b upon tubulin binding, reflecting the conserved mechanisms of microtubule dynamics inhi
Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The microtubule-associated protein, doublecortin-like kinase 1 (DCLK1), is highly expressed in a range of cancers and is a prominent therapeutic target for kinase inhibitors. The physiological roles of DCLK1 kinase activity and how it is regulated remain elusive. Here, we analyze the role of mammalian DCLK1 kinase activity in regulating microtubule binding. We found that DCLK1 autophosphorylates a residue within its C-terminal tail to restrict its kinase activity and prevent aberrant hyperphosphorylation within its microtubule-binding domain. Removal of the C-terminal tail or mutation of this residue causes an increase in phosphorylation within the doublecortin domains, which abolishes microtubule binding. Therefore, autophosphorylation at specific sites within DCLK1 has diametric effects on the molecule’s association with microtubules. Our results suggest a mechanism by which DCLK1 modulates its kinase activity to tune its microtubule-binding affinity. These results provide molecular insights for future therapeutic efforts related to DCLK1’s role in cancer development and progression. Introduction Growth is an essential process of life. Unchecked cellular growth, however, is a hallmark of cancer. Therefore, the process of cell division is often a target of cancer therapeutics (Steinmetz and Prota, 2018; Wieczorek et al., 2016). The macromolecular machine responsible for accurately segregating chromosomes during eukaryotic cell division is the bipolar mitotic spindle, a structure composed of dynamic microtubules organized by a multitude of microtubule-associated proteins (MAPs) (Hornick et al., 2010; Barisic and Maiato, 2016). Doublecortin-like kinase 1 (DCLK1), formerly known as DCAMKL1 and KIAA0369, is one such MAP that is also upregulated in a range of cancers, such as pancreatic, breast, bladder, colorectal, gastric, and hepatocellular carcinoma (Burgess et al., 1999; Lin et al., 2000; Li and Bellows, 2013; Meng et al., 2013; Qu et al., 2015; Liu et al., 2016; Fan et al., 2017; Kadletz et al., 2017; Jiang et al., 2018; Zhang et al., 2017). As a consequence, many studies have focused on developing small-molecule inhibitors against DCLK1 kinase activity in an effort to control cancer growth (Westphalen et al., 2017; Weygant et al., 2014; Ferguson et al., 2020). However, it is currently unclear if DCLK1 kinase activity, microtubule-binding activity, or both are involved in the molecule’s functions during cell division. Therefore, a mechanistic understanding of DCLK1, both at the molecular and biological level, is currently lacking. DCLK1 is a member of the doublecortin (DCX) superfamily, which also includes DCX, DCDC2, and retinitis pigmentosa 1 (RP1), all of which are implicated in human disease (Westphalen et al., 2017; Reiner et al., 2006; Sullivan et al., 1999; Meng et al., 2005; Gleeson et al., 1998; Francis et al., 1999). At its N-terminus, DCLK1 contains two tandem DCX domains (DC1 or N-DC: aa 54–152 and DC2 or C-DC: aa 180–263) (Figure 1A), which are highly conserved among other family members (Lin et al., 2000; Reiner et al., 2006; Sapir et al., 2000; Kim et al., 2003a; Taylor et al., 2000). DCLK1 and its paralog, DCX, were originally identified and characterized for their functions during neuronal development, including neurogenesis and neuronal migration (Burgess et al., 1999; Gleeson et al., 1998; Francis et al., 1999; Bai et al., 2003; Burgess and Reiner, 2000; Jean et al., 2012). Although the roles of DCLK1 and DCX in neurodevelopment have been phenotypically described in vivo, the molecular basis for these observations remains ill-defined. Prior studies have shown that DCLK1 and DCX may act as microtubule stabilizers, nucleators, and regulators of microtubule-based transport (Liu et al., 2012; Moores et al., 2004; Moores et al., 2006; Bechstedt and Brouhard, 2012; Bechstedt et al., 2014; Lipka et al., 2016; Monroy et al., 2020; Ettinger et al., 2016; Patel et al., 2016). Dissecting the mechanisms by which DCLK1 binds to the microtubule can therefore provide insights into the microtubule-binding behaviors of other DCX family members and how they may be subverted in disease. Figure 1 with 2 supplements see all Download asset Open asset The C-terminal domain of DCLK1 regulates autophosphorylation and microtubule binding. (A) Diagram of domains and motifs of human doublecortin-like kinase 1 (DCLK1) (UniProt O15075) that are conserved in the mouse DCLK1 used in this study. DC1, N-terminal doublecortin-like (DCX) domain; DC2, C-terminal DCX domain; kinase domain. Motifs enriched in PEST (proline/P, glutamic acid/E, serine/S, threonine/T) and DEND (aspartic acid/D, glutamic acid/E, asparagine/N, aspartic acid/D) based on Burgess and Reiner, 2001 and Nagamine et al., 2011. Below: model of human DCLK1. DC1 domain (1mg4; Kim et al., 2003a), DC2 domain modeled by homology to DCX-DC2 (5ip4; Burger et al., 2016), kinase domain (5jzj; Patel et al., 2016). DCLK1 is shown as a full-length pseudo-model, with projection domains/tails (and domain linkers) modeled as unfolded to visualize the length and convey the intrinsic disorder predicted for those regions. The mouse DCLK1 (1–740) used in this paper has the same amino acid boundaries as that of humans. (B) Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) Phos-tag gel of purified wild-type (WT), ΔC, and kinase-dead (D511N) DCLK1 proteins separated by phosphorylation level. Representative gel from n = 3 independent experiments. (C) Total internal reflection fluorescence microscopy (TIRF-M) images of 3 nM and 25 nM sfGFP-DCLK1 WT, ΔC, and D511N (green), expressed in bacteria under standard conditions, binding to taxol-stabilized microtubules (blue). Scale bars: 2.5 μm. (D) Quantification of microtubule-bound sfGFP-DCLK1 fluorescence intensity. Means ± sd: 2748.9 ± 2073.6 for 3 nM WT, 16119.6 ± 4324.3 for 25 nM WT, 1.2 ± 31.6 for 3 nM ΔC, 3.8 ± 101.9 for 25 nM ΔC, 9072.4 ± 3380.1 for 3 nM D511N, and 19666.6 ± 3345.3 for 25 nM D511N (n>100 microtubules from n = 3 independent trials for each concentration of each protein). Gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial. ***p<0.0001 and p = 0.3240 for 25 nM WT vs 25 nM D511N, calculated using Student’s t-test. p-values were calculated using n = 3. (E) Coomassie blue-stained SDS-PAGE Phos-tag gel of purified DCLK1-WT and -ΔC incubated with lambda phosphatase (λPP) or incubated in buffer alone for 1 hr at 30°C. Representative gel from n = 3 independent experiments. (F) TIRF-M images of 3 nM sfGFP-DCLK1 WT and ΔC (green) after treatment with λPP, binding to taxol-stabilized microtubules (blue). Scale bars: 2.5 μm. (G) Quantification of microtubule-bound sfGFP-DCLK1 fluorescence intensity. Means ± sd: 15090.7 ± 5285.6 for 3 nM WT + λPP, 18155.3 ± 3833.5 for 25 nM WT + λPP, 12004.2 ± 3490.3 for 3 nM ΔC + λPP, and 21240.6 ± 3413.5 for 25 nM ΔC + λPP (n>100 microtubules from n = 3 independent trials for each protein concentration; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial.). D511N data are reproduced from (D) for comparison. p = 0.3566 for 25 nM WT + λPP vs 25 nM ΔC + λPP, p = 0.6341 for 25 nM WT + λPP vs 25 nM D511N, p = 0.5989 for 25 nM ΔC + λPP vs 25 nM D511N, and p = 0.4462 for 3 nM WT + λPP vs 3 nM ΔC + λPP, calculated using Student’s t-test. p-values were calculated using n = 3. For all experiments, at least two separate protein purifications were used. Figure 1—source data 1 Uncropped gels for the associated panels in Figure 1. The red box indicates how the gel was cropped. https://cdn.elifesciences.org/articles/60126/elife-60126-fig1-data1-v2.pdf Download elife-60126-fig1-data1-v2.pdf Figure 1—source data 2 Uncropped gels. https://cdn.elifesciences.org/articles/60126/elife-60126-fig1-data2-v2.zip Download elife-60126-fig1-data2-v2.zip The C-terminal portion of DCLK1 contains a serine/threonine kinase domain and an unstructured C-terminal tail that shares sequence similarities with calcium/calmodulin-dependent protein kinase I (CaMKI) (Shang et al., 2003; Edelman et al., 2005). For both DCLK1 and CaMKI, removal of a distal C-terminal ‘tail’ region results in an increase in kinase activity (Patel et al., 2016; Shang et al., 2003; Edelman et al., 2005; Goldberg et al., 1996). This mode of regulation has been well-studied for CaMKI, whose C-terminal tail makes direct contact with the kinase domain, directly inhibiting its enzymatic activity (Goldberg et al., 1996). However, it is unclear if, or how, the C-terminal tail of DCLK1 regulates its kinase domain. In addition, the physiological significance of DCLK1 kinase activity is unknown, even though it is a target for the development of kinase inhibitors due to its prominent role in cancer (Westphalen et al., 2017; Weygant et al., 2014; Ferguson et al., 2020). Additional information on the functional role of the DCLK1 kinase domain and how it is controlled would therefore be valuable for understanding how drugs can effectively target DCLK1 for therapeutic purposes. Here we present a detailed examination of the microtubule-binding properties of DCLK1 and how they are regulated by its kinase activity. We find that DCLK1 autophosphorylates one key residue (T688) within its C-terminal tail via an intramolecular mechanism to strongly modulate its microtubule-binding affinity. Removal of the C-terminal tail or mutation of T688 results in an increase in phosphorylation of residues within both the DC1 and the DC2 domains, which in turn decreases microtubule binding. Furthermore, we observed that mutating four key phosphosites within DC1 of DCLK1 rescues microtubule binding in the construct lacking the C-terminal tail. Overall, our data led to a model in which DCLK1 autophosphorylates its C-terminal tail to modulate the activity of its own kinase domain and, subsequently, the level of phosphorylation within its microtubule-binding domains. To our knowledge, this is the first example of a self-regulatory MAP that can tune its microtubule-binding properties based on autophosphorylation state. Our results uncover a novel intramolecular regulation of microtubule binding within a prominent family of MAPs and may have implications for DCLK1’s known roles in tumor development and cancer progression. Results Previous results have suggested that phosphorylation of DCLK1 occurs in part via autophosphorylation (Patel et al., 2016; Shang et al., 2003). To determine if DCLK1 phosphorylation is mediated by an inter- or intramolecular mechanism, we utilized an established kinase-dead mutant of DCLK1 (D511N) (Patel et al., 2016; Patel et al., 2021) and an active wild-type (WT) DCLK1 enzyme, both purified from bacteria (Figure 1—figure supplement 1A and Figure 1—figure supplement 2A). We did not observe trans-phosphorylation of DCLK1-D511N upon incubation with DCLK1-WT, although DCLK1-WT efficiently autophosphorylated itself in this assay (Figure 1—figure supplement 2B). Thus, under the conditions in our experiments, DCLK1 phosphorylation occurs via an intramolecular mechanism. Removal of the C-terminal region of DCLK1 that follows the kinase domain results in an increase in kinase activity (Shang et al., 2003). How this region regulates enzymatic activity and autophosphorylation of DCLK1 and how phosphorylation of the molecule affects its microtubule-binding properties are open questions. We first compared the mobility of full-length mouse DCLK1-WT (aa 1–740) and a truncated DCLK1 lacking the C-terminal tail (ΔC: aa 1–648) to full-length kinase-dead DCLK1-D511N on a Phos-tag gel, which enhances the separation of differentially phosphorylated proteins (Figure 1B and Figure 1—figure supplement 1A-B; Kinoshita et al., 2009). We found that bacterially expressed DCLK1-WT and DCLK1-ΔC proteins migrated more slowly into the Phos-tag gel, indicative of higher levels of phosphorylation, compared to the non-phosphorylated DCLK1-D511N (Figure 1B). Using total internal reflection fluorescence microscopy (TIRF-M), we imaged sfGFP-tagged WT, ΔC, and D511N proteins binding to taxol-stabilized microtubules (Figure 1C) at concentrations differing by eightfold. Strikingly, DCLK1-ΔC did not bind to microtubules at either concentration tested, in stark contrast to DCLK1-WT and DCLK1-D511N, which both robustly bound to microtubules (Figure 1C–D). Notably, DCLK1-D511N bound microtubules more robustly at lower concentrations than DCLK1-WT. This is consistent with the prior work showing that D511N robustly stimulates tubulin polymerization (Patel et al., 2016). These experiments suggest that the C-terminal region regulates DCLK1 autophosphorylation, which in turn directly modulates its microtubule-binding affinity. To test this possibility, we sought to evaluate the microtubule-binding behaviors of dephosphorylated DCLK1-WT and DCLK1-ΔC. We incubated the proteins with the Mn2+-dependent protein phosphatase, lambda phosphatase (λPP), which strongly dephosphorylated DCLK1-WT and DCLK1-ΔC as evidenced by the marked shifts on a Phos-tag gel without phospho-intermediate bands, but had little effect on the migration of D511N (Figure 1E and Figure 1—figure supplement 2C). The similar migration of DCLK1-WT and DCLK1-ΔC in the absence of phosphatase, coupled with the relatively larger migration shift of DCLK1-ΔC after treatment, suggests that DCLK1-ΔC is hyperphosphorylated compared to the WT protein, in agreement with previous results (Shang et al., 2003). In addition, anion exchange chromatograms also revealed a greater shift in the elution volume between phosphorylated and non-phosphorylated DCLK1-ΔC compared to the shift observed for phosphorylated vs non-phosphorylated WT protein, consistent with DCLK1-ΔC having a higher negative charge due to being hyperphosphorylated (Figure 1—figure supplement 1C). Using TIRF-M, we found that λPP-treated DCLK1-WT and DCLK1-ΔC bound to microtubules similarly to DCLK1-D511N (Figure 1F–G), similar to prior results showing that phosphatase-treated DCLK1 robustly stimulates tubulin polymerization (Patel et al., 2016). This further suggests that autophosphorylation modulates the microtubule-binding affinity of DCLK1 and that hyperphosphorylation of DCLK1-ΔC largely abolishes microtubule binding. The high phosphorylation levels observed for both DCLK1-WT and DCLK1-ΔC suggested that these proteins phosphorylate themselves during bacterial expression. To determine the contributions of the C-terminal tail to DCLK1 function, we used a previously defined strategy to control the levels of autophosphorylation during expression (Patel et al., 2016; Patel et al., 2021). We co-expressed DCLK1-WT and DCLK1-ΔC with λPP in bacteria, followed by the subsequent removal of λPP from the DCLK1 preps via affinity and ion exchange chromatography. We compared DCLK1 proteins prepared in the absence or presence of λPP on a Phos-tag gel and observed that λPP co-expression substantially reduced phosphorylation levels of both DCLK1-WT and DCLK1-ΔC (Figure 2A). For all subsequent experiments, all DCLK1 protein variants were co-expressed with λPP. Upon incubation of dephosphorylated DCLK1 proteins with adenosine triphosphate (ATP), both DCLK1-WT and DCLK1-ΔC exhibited an increase in phosphorylation, but DCLK1-ΔC appeared to be entirely phosphorylated by 30 min, whereas DCLK1-WT displayed a number of phosphorylated intermediates even at 60 min (Figure 2A–B; 93.7% of DCLK1-ΔC protein shifts to the uppermost band after a 30-min incubation with ATP compared with 41.8% of DCLK1-WT protein after a 60-min incubation with ATP). Using TIRF-M, we determined the microtubule-binding affinities for DCLK1-WT and DCLK1-ΔC in the absence and presence of ATP (Figure 2C–E). We found that, in the absence of ATP, both proteins exhibited relatively similar microtubule-binding affinities (Figure 2C–E). After a 30-min incubation with ATP, the microtubule-binding affinity of DCLK1-WT moderately weakened, as evidenced by an ~2.5-fold increase in KD (Figure 2D). However, incubation with ATP resulted in a dramatic approximately forty-onefold decrease in microtubule affinity of DCLK1-ΔC (Figure 2E). Interestingly, in the presence of ATP, DCLK1-ΔC was still present at regions of microtubule curvature, consistent with prior results that doublecortin proteins have a higher affinity for these regions in vitro (Figure 1—figure supplement 2D; Bechstedt et al., 2014). We also analyzed the binding behaviors of WT and ΔC on non-stabilized guanosine diphosphate (GDP) microtubule lattices grown from GMPCPP seeds and observed a similar decrease in bound DCLK1-ΔC in the presence of ATP (Figure 2F–G). Finally, we performed a microtubule co-sedimentation assay with WT and ΔC and found that while similar amounts of protein pelleted with microtubules in the absence of ATP, significantly less ΔC co-pelleted with microtubules in the presence of ATP (Figure 1—figure supplement 2E). These results indicate that the loss of its regulatory C-terminal tail results in aberrant DCLK1 hyperphosphorylation, leading to a dramatic loss of microtubule-binding affinity. Thus, the kinase activity of DCLK1 directly controls its association with microtubules via intramolecular phosphorylation, which in turn is regulated by the C-terminus of the protein. Figure 2 Download asset Open asset Hyperphosphorylation of DCLK1-ΔC prohibits microtubule binding. (A) Coomassie blue-stained sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) Phos-tag gel of purified doublecortin-like kinase 1 wild-type (DCLK1-WT) and -ΔC proteins separated by phosphorylation level. The first and fifth lanes contain DCLK1-WT and -ΔC expressed in bacteria under standard conditions. All other lanes contain DCLK1-WT and -ΔC that were co-expressed with lambda phosphatase (λPP), which was subsequently separated from DCLK1. Incubation of λPP-treated DCLK1-WT and -ΔC with 2 mM adenosine triphosphate (ATP) at the indicated times reveals a band shift, indicative of an increase in phosphorylation. (B) Quantification of the average percent of total DCLK1 protein that is phosphorylated in each condition. Averages are derived from n = 3 independent experiments. (C) Total internal reflection fluorescence microscopy (TIRF-M) images of sfGFP-DCLK1-WT and -ΔC (co-expressed in bacteria with λPP) at indicated concentrations (green) binding to taxol-stabilized microtubules (blue) after a 30-min incubation in the absence or presence of 2 mM ATP. Scale bars: 2.5 μm. (D) Quantification of microtubule-bound sfGFP-DCLK1-WT fluorescence intensity plotted against concentration after a 30-min incubation in the absence or presence of ATP (WT without ATP, KD = 2.1 nM, and WT with ATP, KD = 5.4 nM, derived from at least n = 3 independent trials per condition). (E) Quantification of microtubule-bound sfGFP-DCLK1-ΔC fluorescence intensity plotted against concentration after a 30-min incubation in the absence or presence of ATP (ΔC without ATP, KD = 3.9 nM, and ΔC with ATP, KD = 161.0 nM, derived from n = 3 independent trials). (F) TIRF-M images of 10 nM sfGFP-DCLK1-WT or -ΔC (green, co-expressed in bacteria with λPP) binding to non-stabilized GDP microtubules grown from GMPCPP seeds (blue) after a 30-min incubation in the absence or presence of 2 mM ATP. Scale bars: 2.5 μm. (G) Quantification of microtubule-bound sfGFP-DCLK1 fluorescence intensity. Means ± sd: 15004.4 ± 6503.8 for WT, 12535.9 ± 3247.5 for WT + ATP, 12111.3 ± 3534.0 for ΔC, and 1579.3 ± 866.5 for ΔC + ATP (n>60 microtubules from n = 2 independent trials for each condition; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial; p = 0.6360 for WT vs ΔC and p = 0.0440 for WT + ATP vs ΔC + ATP, calculated using Student’s t-test; p-values were calculated using n = 2). For all experiments, at least two separate protein purifications were used. Figure 2—source data 1 Uncropped gel for the associated panel in Figure 2. The red box indicates how the gel was cropped. https://cdn.elifesciences.org/articles/60126/elife-60126-fig2-data1-v2.pdf Download elife-60126-fig2-data1-v2.pdf Figure 2—source data 2 Uncropped gels. https://cdn.elifesciences.org/articles/60126/elife-60126-fig2-data2-v2.zip Download elife-60126-fig2-data2-v2.zip In order to determine how phosphorylation regulates the microtubule-binding affinity of DCLK1-ΔC, we performed liquid chromatography with tandem mass spectrometry (LC-MS/MS) of phosphorylated DCLK1-WT and DCLK1-ΔC proteins. For each phosphorylated residue identified, we counted the total number of DCLK1-WT and DCLK1-ΔC peptides containing the phosphorylated residue, and then calculated the percent of those peptides whose spectra revealed phosphorylation at that residue. Within the microtubule-binding region of DCLK1 (aa 44–263), we found that nine sites were more frequently phosphorylated in DCLK1-WT samples and 17 sites were more frequently phosphorylated in DCLK1-ΔC samples (Figure 3A–B). Of the 17 phosphorylation sites in DCLK1-ΔC, five either directly contact tubulin or are adjacent to residues that directly contact tubulin within the lattice (Manka and Moores, 2019). The architecture of DCLK1 suggests that it likely has the flexibility to autophosphorylate its N-terminal half due to an intrinsically disordered region between the DC2 domain and the kinase domain (Figure 3C; aa 263–374). These results indicate that the loss of microtubule binding of DCLK1-ΔC is due to an increase in phosphorylation at multiple sites, as opposed to a single site whose phosphorylation status dictates microtubule binding. Figure 3 with 1 supplement see all Download asset Open asset DCLK1-ΔC aberrantly autophosphorylates at multiple sites within the microtubule-binding region. (A, B) Visualization of changes in phosphorylation due to deletion of C-terminal domain. Experiment compared peptides from wild-type (WT) and ∆C constructs: data are expressed as fold-change increases in phosphorylation in one construct over the other based on the percent of total peptides that exhibited phosphorylation at a particular site (n = 751 and 637 total peptides analyzed for WT and ∆C, respectively, from n = 3 independent experiments). Darker colors indicate a higher fold change in phosphorylation; that is darker blue indicates that a site is more commonly phosphorylated in the WT construct, while darker orange indicates a site is more commonly phosphorylated in the ∆C construct. (A) Lollipop plot summarizes changes in phosphorylation (≥1.5 fold change) mapped onto a diagram of the mouse doublecortin-like kinase 1 (DCLK1) used in this study, but all listed residues are conserved in human DCLK1. (B) DC1 domain (1mg4; Kim et al., 2003a) and DC2 domain modeled by homology to DCX-DC2 (5ip4; Burger et al., 2016). Domain structures are aligned and shown as ribbon representations with labeled S/T residues visualized as CPK/balls. Level of saturation in color indicates fold change in phosphorylation of those residues: increase in WT (blue colors) and increase in ∆C (orange/red colors). (C) Architecture of DCLK1 protein with the per-residue IUPRED2A (Mészáros et al., 2018) disorder prediction score shown in the corresponding plot with a cutoff value of 0.5 indicated by the dashed line. Residues scored above this value are predicted to be disordered. The decrease in microtubule-binding affinity we observed for the DCLK1-ΔC construct in the presence of ATP could be due to a general increase in phosphorylation throughout the entire microtubule-binding region (aa 46–263) or due to phosphorylation at specific sites within the microtubule-binding DC domains. In order to determine the relative contributions of phosphorylation within the DC1 and DC2 domains to the decrease in the microtubule-binding affinity of DCLK1-ΔC, we mutated four conserved residues that showed the highest increase in phosphorylation within the DC1 (C4A-DC1: S77, S83, S96, T143) or the DC2 domain (C4A-DC2: T189, S193, T218, S228) (Figure 4 and Figure 1—figure supplement 1A-B, Figure 3—figure supplement 1A-B; Manka and Moores, 2019). We reasoned that if phosphorylation of these residues is responsible for the decreased microtubule-binding affinity of DCLK1-ΔC, then mutating these residues to alanines, which cannot be phosphorylated, should rescue the microtubule-binding defect of this construct in the presence of ATP. If, however, phosphorylation of these residues is not responsible for the decreased microtubule-binding affinity, then there should be no difference in binding between the DCLK1-ΔC and ΔC4A-DC1 or C4A-DC2 regardless of the presence of ATP. Figure 4 Download asset Open asset Phosphonull mutations within DC1 restore microtubule binding of ΔC. (A) Diagrams depicting the domains, amino acid boundaries, and mutations relevant to the doublecortin-like kinase 1 (DCLK1) constructs used. ∆CDC1-4A indicates the four residues in DC1 that were mutated to alanines S77, S83, S96, and T143. ∆CDC2-4A indicates the four residues in DC2 that were mutated to alanines T189, S193, T218, and S228. (B) Total internal reflection fluorescence microscopy (TIRF-M) images of sfGFP-DCLK1 ΔC, ∆CDC1-4A, and ∆CDC2-4A, co-expressed in bacteria with lambda phosphatase (λPP), at indicated concentrations binding to taxol-stabilized microtubules (blue) in the absence or presence of adenosine triphosphate (ATP). Scale bars: 2.5 μm. (C) Quantification of microtubule-bound 5 nM sfGFP-DCLK1 fluorescence intensity. For 5 nM concentrations in the absence of ATP, means ± sd: 12883.5 ± 2881.6 for ΔC, 12245.5 ± 3283.0 for ∆CDC1-4A, 8552.7 ± 2097.3 for ∆CDC2-4A (n>100 microtubules per condition from n = 3 independent trials; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial; p = 0.8128 for ΔC vs ∆CDC1-4A and p = 0.1031 for ΔC vs ∆CDC2-4A calculated using Student’s t-test; p-values were calculated using n = 3). For 5 nM concentrations in the presence of ATP, means ± sd: 987.6 ± 202.5 for ΔC + ATP, 4042.6 ± 1624.6 for ∆CDC1-4A + ATP, 2482.0 ± 1058.3 for ∆CDC2-4A + ATP (n>100 microtubules from n = 3 independent trials; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial; p = 0.0319 for ΔC vs ∆CDC1-4A and p = 0.0742 for ΔC vs ∆CDC2-4A calculated using Student’s t-test; p-values were calculated using n = 3). (D) Quantification of microtubule-bound 20 nM sfGFP-DCLK1 fluorescence intensity. For 20 nM concentrations in the absence of ATP, means ± sd: 23634.3 ± 1725.1 for ΔC, 22277.3 ± 1334.9 for ∆CDC1-4A, 19912.2 ± 5408.6 for ∆CDC2-4A (n>100 microtubules per condition from n = 3 independent trials; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial; p = 0.3419 for ΔC vs ∆CDC1-4A and p = 0.3196 for ΔC vs ∆CDC2-4A calculated using Student’s t-test; p-values were calculated using n = 3). For 20 nM concentrations in the presence of ATP, means ± sd: 1579.8 ± 585.1 for ΔC + ATP, 12556.9 ± 3419.9 for ∆CDC1-4A + ATP, 3503.4 ± 826.7 for ∆CDC2-4A + ATP (n>100 microtubules from n = 4, 6, and 5 independent trials for ΔC, ∆CDC1-4A, and ∆CDC2-4A, respectively; gray dots indicate individual microtubule intensities, while colored dots represent the averages from each trial; p = 0.0002 for ΔC vs ∆CDC1-4A and p = 0.0058 for ΔC vs ∆CDC2-4A calculated using Student’s t-test; p-values were calculated using n = number of independent trials as stated above). For all experiments, at least two separate protein purifications were used. Using TIRF-M, we imaged DCLK1-ΔC, ΔC4A-DC1, and C4A-DC2 binding to taxol-stabilized microtubules in the presence or absence of ATP (Figure 4B–D). In the absence of ATP, all three DCLK1 proteins bound robustly to microtubules at both 5 nM and 20 nM. For all proteins at 5 nM, in the presence of ATP, there was substantially less DCLK1 on the microtubule (Figure 4B–C), similar to our previous results (Figure 2); however, there was a small, but significant increase in the amount of ΔC4A-DC1 bound to the microtubule (Figure 4B–C). At 20 nM, in the presence of ATP, we observed significantly more ΔC4A-DC1 and ΔC4A-DC2 on the microtubule compared with ΔC, but this result was far more striking for ΔC4A-DC1, which exhibited an approximately eightfold higher fluorescence intensity on the microtubule than ΔC (Figure 4B,D). These results indicate that mutating the residues in DC1 that are abnormally phosphorylated in the ΔC construct rescued microtubule binding by preventing phosphorylation within this domain. Mutating the phosphosites within DC2 also rescued microtubule binding, but to a lesser extent than the DC1 mutations. Therefore, aberrant phosphorylation of these residues in DCLK1-ΔC could indeed be responsible for the dramatic attenuation of microtubule binding. We next wanted to determine the mechanism by which the C-terminal region of DCLK1 prevents hyperphosphorylation of the DC domains. We examined autophosphorylated DCLK1-WT by LC-MS/MS and identified two threonine residues in the C-terminal region (T687 and T688) that were consistently phosphorylated (Figure 3—figure supplement 1C). In order to understand how the C-terminal tail contributes to autophosphorylation, we mutated T687 and T688 to alanines individually (Figure 5A). For all of the experiments with these mutants, we co-expressed DCLK1 proteins with λPP to obtain a dephosphorylated protein preparation. We first evaluated the ability of these mutan
In pancreatic islet beta cells, molecular motors use cytoskeletal polymers microtubules as tracks for intracellular transport of insulin secretory granules. Beta-cell microtubule network has a complex architecture and is non-directional, which provide insulin granules at the cell periphery for rapid secretion response, yet to avoid over-secretion and subsequent hypoglycemia. We have previously characterized a peripheral sub-membrane microtubule array, which is critical for withdrawal of excessive insulin granules from the secretion sites. Microtubules in beta cells originate at the Golgi in the cell interior, and how the peripheral array is formed is unknown. Using real-time imaging and photo-kinetics approaches in clonal mouse pancreatic beta cells MIN6, we now demonstrate that kinesin KIF5B, a motor protein with a capacity to transport microtubules as cargos, slides existing microtubules to the cell periphery and aligns them to each other along the plasma membrane. Moreover, like many physiological beta-cell features, microtubule sliding is facilitated by a high glucose stimulus. These new data, together with our previous report that in high glucose sub-membrane MT array is destabilized to allow for robust secretion, indicate that MT sliding is another integral part of glucose-triggered microtubule remodeling, likely replacing destabilized peripheral microtubules to prevent their loss over time and beta-cell malfunction.
Abstract Centrosome separation in Drosophila larval neuroblasts and asymmetric transport of embryonic determinants in oocytes are both microtubule-dependent processes that require Kinesin-1 activation by Ensconsin/microtubule-associated protein 7 (MAP7). However, the molecular mechanism used by Ensconsin to activate Kinesin-1 remains elusive. Ensconsin/ MAP7 contains an N-terminal microtubule-binding domain (MBD) and a C-terminal Kinesin-binding domain (KBD). Using rescue experiments in live flies, we show that KBD expression alone is sufficient to fully rescue Ensconsin-dependent centrosome separation defects, but not the fast oocyte streaming and the localization patterns of Staufen and Gurken proteins. Interestingly, we show here for the first time that KBD binds and stimulates Kinesin-1 binding to Mts in vivo and in vitro . We propose that the KBD/Kinesin-1 motor represents a minimal activation module that stimulates Kinesin-1 binding to Mts. Addition of the MBD, present in the full length Ensconsin allows this activation to occur directly on the Mt. Our data also suggest that in a very large cell with a complex microtubule network, but not in smaller cells, this dual activation by Ensconsin is essential for optimal Kinesin-1 targeting to the microtubule cytoskeleton.
A functional nervous system is built upon the proper morphogenesis of neurons to establish the intricate connection between them. The microtubule cytoskeleton is known to play various essential roles in this morphogenetic process. While many microtubule-associated proteins (MAPs) have been demonstrated to participate in neuronal morphogenesis, the function of many more remains to be determined. This study focuses on a MAP called HMMR, which was originally identified as a hyaluronan binding protein and later found to possess microtubule and centrosome binding capacity. HMMR exhibits high abundance on neuronal microtubules and altering the level of HMMR significantly affects the morphology of neurons. Instead of confining to the centrosome(s) like cells in mitosis, HMMR localizes to microtubules along axons and dendrites. Furthermore, transiently expressing HMMR enhances the stability of neuronal microtubules and increases the formation frequency of growing microtubules along the neurites. HMMR regulates the microtubule localization of a non-centrosomal microtubule nucleator TPX2 along the neurite, offering an explanation for how HMMR contributes to the promotion of growing microtubules. This study sheds light on how progenitor cells utilize proteins involved in mitosis for non-mitotic functions.
Article Figures and data Abstract Editor's evaluation Introduction Results and discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract A challenge in analyzing dynamic intracellular cell biological processes is the dearth of methodologies that are sufficiently fast and specific to perturb intracellular protein activities. We previously developed a light-sensitive variant of the microtubule plus end-tracking protein EB1 by inserting a blue light-controlled protein dimerization module between functional domains. Here, we describe an advanced method to replace endogenous EB1 with this light-sensitive variant in a single genome editing step, thereby enabling this approach in human induced pluripotent stem cells (hiPSCs) and hiPSC-derived neurons. We demonstrate that acute and local optogenetic EB1 inactivation in developing cortical neurons induces microtubule depolymerization in the growth cone periphery and subsequent neurite retraction. In addition, advancing growth cones are repelled from areas of blue light exposure. These phenotypes were independent of the neuronal EB1 homolog EB3, revealing a direct dynamic role of EB1-mediated microtubule plus end interactions in neuron morphogenesis and neurite guidance. Editor's evaluation In their manuscript, Dema et al. showcase a valuable tool to study the role of the microtubule end-binding protein, EB1. This important study replaces endogenous EB1 with a light-sensitive variant, which they use to locally inactivate EB1 in human iPSC-derived neurons. They find that EB1 inactivation induces microtubule depolymerization in the growth cone and neurite retraction. The data is of high quality, the evidence supporting the conclusions is solid, and the findings of this work will be of interest to cell biologists and neurobiologists, while the methods utilized will have even broader general interest. https://doi.org/10.7554/eLife.84143.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Microtubules (MTs) are essential components of the cytoskeleton in both mature neurons, in which polarized MT tracks support axonal transport, as well as during nervous system development (Atkins et al., 2023; Kapitein and Hoogenraad, 2015; van de Willige et al., 2016). During neuronal differentiation, the expression of many microtubule-associated proteins (MAPs) is upregulated, and numerous neurodevelopmental diseases are being linked to genetic alterations in MAPs, MT motors, and tubulin itself (Franker and Hoogenraad, 2013; Maillard et al., 2023). Developing neurites must elongate over long distances integrating extrinsic cues to achieve correct nervous system topology. Neurite growth and pathfinding are driven by adhesion and F-actin dynamics in the growth cone, the advancing terminal structure of growing neurites. Analogous to the role of MTs in migrating cells, dynamic MTs enter the F-actin rich growth cone periphery and these so-called 'pioneer' MTs participate in growth cone advance and guidance downstream of extracellular cues (Liu and Dwyer, 2014; Vitriol and Zheng, 2012). Proteins that associate with growing MT plus ends (+TIPs), such as spectraplakins (MACF1), the adenomatous polyposis coli protein APC, CLASPs, and neuron navigator proteins (Nav1), play critical roles in neuron morphogenesis (Cammarata et al., 2016; Coles and Bradke, 2015). For example, Nav1 mediates interactions between dynamic MT plus ends and F-actin filaments (Sánchez-Huertas et al., 2020) and CLASPs are involved in cell-matrix adhesion site remodeling (Stehbens et al., 2014). These +TIPs bind to growing MT plus ends through small adaptor proteins of the end-binding (EB) family. In humans, EBs are encoded by three MAPRE genes, and although EB1 is the best characterized +TIP adaptor, it remains unclear to what extent EB functions overlap. For example, EB3, which is upregulated in neurons and other differentiated cell types, coordinates MT and F-actin dynamics in developing neurons independent of EB1 (Gordon-Weeks, 2017). The role of EB1-dependent+TIPs in neuron morphogenesis is inferred mostly from genetic knockout phenotypes in model systems from different species. It has not been tested how +TIP association with growing MT ends participates in controlling growth cone dynamics in real time in developing human neurons in part due to an absence of technology to acutely inhibit these interactions. Here, we utilize our recent optogenetic tool to disrupt EB1 function and thus +TIP MT-binding in growth cones of hiPSC-derived developing cortical neurons. In addition to demonstrating a novel strategy to generate photosensitive variants of multidomain proteins in a single genome editing step, we show that EB1 is necessary to stabilize MTs in the growth cone periphery and to maintain growth cone advance, and that this function of EB1 cannot be compensated by EB3. Results and discussion One-step genome editing to generate photosensitive protein variants We recently reported an optogenetic system to inactivate end-binding protein 1 (EB1) with high spatial and temporal accuracy in living cells by inserting a light-sensitive LOV2/Zdk1 dimerization module between the N-terminal +TIP and C-terminal MT-binding domains of EB1 (Figure 1A). We named this photo-inactivated construct π-EB1 and demonstrated effects on MT dynamics and function in interphase and mitotic human cancer cells (Dema et al., 2022b; van Haren et al., 2018). However, our initial method of generating π-EB1 cell lines required sequential genetic knockouts and re-expression of photosensitive EB1 variants. To enable and demonstrate the utility of this LOV2/Zdk1-mediated multidomain splitting strategy to photo-inactivate proteins expressed at endogenous levels in more complex cell systems such as an hiPSC model of neuronal morphogenesis, we devised a CRISPR/Cas9-mediated genome editing strategy to directly insert the LOV2/Zdk1 dimerization module into the EB1 gene and thus convert endogenous EB1 into the photosensitive π-EB1 variant in one genome editing step. Figure 1 with 5 supplements see all Download asset Open asset One-step genome editing to replace EB1 with a photo-sensitive variant. (A) AlphaFold2 model of the π-EB1 tetramer (Mirdita et al., 2022). Note that AlphaFold2 does not correctly predict relative domain positions and did not capture the LOV2/Zdk1 interaction correctly although a structure of the LOV2/Zdk1 dimer has previously been determined (Wang et al., 2016). (B) Overview of the one-step CRISPR/Cas9-mediated insertion of a π-element construct containing the photosensitive LOV2/Zdk1 module, a fluorescent protein marker, and an internal EF1α promoter. Arrows indicate the location of PCR primers. Lowercase letters indicate mutations introduced to make the homology-directed repair (HDR) template resistant to Cas9 cleavage. (C) Genomic PCR to validate π-element integration into the endogenous EB1 locus with primers as indicated in (B). The two clones shown are homozygous as there is no short product in PCR2, which corresponds to the non-edited EB1 locus. (D) Immunoblots with antibodies as indicated of control and π-EB1 i3N clones before and after 2 days of neuron differentiation showed replacement of EB1 by the photosensitive π-EB1 variant and expected +TIP expression level changes associated with neuron differentiation. (E) RT-qPCR analysis of the expression levels of the π-element N- and C-terminal halves relative to EB1 expression in wild-type (Ctrl) i3N hiPSCs. Shown are the mean and data points from individual qPCR reactions. (F) Comparison of nuclear Oct4 staining (white) as a pluripotency marker in control and π-EB1 i3N hiPSC colonies. Nuclei are identified with DAPI (blue). (G) Image of a π-EB1 i3N hiPSCs colony with magnified images on the right showing dissociation of EGFP-Zdk1-EB1C from growing MT ends in blue light. (H) Immunoblot of control and π-EB1 and EB3-/- i3Neurons showing expression of π-EB1 and deletion of both EB1 and EB3. (I) π-EB1 i3N hiPSCs transiently expressing a mScarlet-tagged EB1N MT-binding domain before and during blue light exposure. Maximum intensity projections in alternating green and magenta over 20 s at 3 s intervals illustrate attenuation of MT growth during blue light exposure. (J) Quantification of the median MT growth rate per cell before and during blue light exposure in control and π-EB1 i3N hiPSCs. Gray lines connect data points from the same cell. Statistical analysis by paired t-test for each i3N hiPSC line. Figure 1—source data 1 Original DNA gel and immunoblot images. https://cdn.elifesciences.org/articles/84143/elife-84143-fig1-data1-v2.zip Download elife-84143-fig1-data1-v2.zip Evaluating different designs of what we named π-elements in H1299 human cancer cells, we found that homozygous integration of a π-element in which an internal EF1α promoter drives expression of the C-terminal π-EB1 half while the N-terminal half remains under the control of the endogenous promoter (Figure 1B) resulted in balanced expression of both π-EB1 parts and in homozygous edited clones yielded π-EB1 protein levels that were similar to EB1 in control cells (Figure 1—figure supplement 1). In contrast, an internal ribosome entry site (IRES) resulted in very poor expression of the C-terminal half. Of note, self-cleaving 2 A peptides also did not work because the C-terminus of the LOV2 domain cannot be modified without greatly inhibiting Zdk1 binding (Wang et al., 2016). We next tested π-element integration in i3N cells, an hiPSC line that expresses Ngn2 under the control of a doxycycline-induced promoter to allow inducible differentiation into cortical glutamatergic i3Neurons (Wang et al., 2017). Homozygous π-element integration into both copies of exon 5 of the endogenous MAPRE1 (EB1) gene was validated by genomic PCR, sequencing, and immunoblot (Figure 1C and D). Because the N-terminal π-EB1 half is expressed from the endogenous EB1 promoter, compared with EB1 in control cells, EB1N-LOV2 expression levels were similar and only mildly reduced in both i3N hiPSCs and i3Neurons (Figure 1D and E). In contrast, expression of Zdk1-EB1C that is driven by the EF1α promoter was markedly increased in i3N hiPSCs but was lower in i3Neurons (Figure 1D). This indicates a change in the activity balance of these two promoters during i3N neuronal differentiation and suggests that the π-element design could be improved by also using the native promotor to drive expression of the C-terminal half. However, we did not try this, because proliferation, stem cell properties (Figure 1F), and most importantly differentiation into i3Neurons appeared normal as indicated by expected changes in expression levels of marker proteins such as EB3, DCX, and KIF2C that increase or decrease, respectively, during neuronal differentiation (Figure 1D, Figure 1—figure supplement 2A and B) (Blair et al., 2017). Because EB1 and EB3 are very similar and may be at least partially functionally redundant, we also removed EB3 expression by introducing a frameshift and premature stop codon near the start of the EB3 open reading frame in π-EB1 i3N hiPSCs (Figure 1H; Figure 1—figure supplement 3) to be able to ask how EB1 contributes to neuron morphogenesis and compare the relative contributions of EB1 and EB3. To visualize π-EB1 photodissociation in genome-edited i3N cells, we chose to tag the π-EB1 C-terminal part with EGFP so that longer wavelength imaging channels remain available. Clonal π-EB1 i3N hiPSC colonies homogenously expressed EGFP-Zdk1-EB1C, and as expected, EGFP-Zdk1-EB1C rapidly dissociated from growing MT ends during blue light exposure in both i3N stem cells (Figure 1G) and differentiating i3Neurons (Figure 1—figure supplement 2D). In addition, and similar to what we previously observed in interphase human cancer cells (van Haren et al., 2018), π-EB1 photoinactivation significantly attenuated MT growth in i3N hiPSCs transiently expressing EB1N-mScarlet-LOV2, which remains on MT ends and enables MT growth rate measurements before and during blue light exposure (Figure 1I and J, Figure 1—video 1). EB1 is required to stabilize growth cone microtubule growth Upon Ngn2-induced differentiation, π-EB1 EB3-/- i3Neurons sprouted neurites normally (Figure 1—figure supplement 3D, Figure 1—video 2), and as expected, the EGFP-tagged C-terminal half of π-EB1 associated with growing MT ends in neurites with the majority of growing MT ends localized to growth cones. Of note, because MT growth is slower in neurons than in proliferating cells (Stepanova et al., 2003), EGFP-Zdk1-EB1C comets were less elongated in i3Neurons and appeared more punctate compared with i3N hiPSCs. Nevertheless, EGFP-Zdk1-EB1C quickly and reversibly dissociated from MT ends upon blue light exposure in π-EB1 EB3-/- i3Neurons (Figure 2A, Figure 2—video 1). Figure 2 with 2 supplements see all Download asset Open asset π-EB1 photoinactivation destabilizes MTs in i3Neuron growth cones. (A) Growth cone of a π-EB1 EB3-/- i3Neuron in which MTs were labeled with the far-red cabazitaxel derivative 4–610 CP-CTX. EGFP-Zdk1-EB1C dissociates from growing MT ends within seconds of blue light exposure. (B) MTs labeled with SPY555-tubulin in growth cones of i3Neurons with the π-EB1 genotype indicated on the left. Note that microtubules (MTs) continue to dynamically extend into the growth cone periphery in control i3Neurons but frequently depolymerize upon blue light exposure in both π-EB1 and π-EB1 EB3-/- i3Neuron growth cones. Images are shown in inverted contrast for better visibility. (C) Quantification of the length change of 3–4 MTs per growth cone of MT ends that were clearly visible before and during blue light exposure. Gray lines are individual MTs. Blue line is the average of all MTs and the shaded area indicates the 95% confidence interval. The orange dashed line indicates no change. Statistical analysis by paired t-test at 20 s before and during blue light exposure. (D) Quantification of the growth cone MT growth rate after 60 s of blue light exposure by tracking SPY555-tubulin-labeled MT ends. Data points represent the average of >100 frame-to-frame growth rate measurements from multiple MTs per growth cone. Statistical analysis by ANOVA and Tukey-Kramer HSD. To better show individual growth cone MTs, the gamma of the tubulin channels was adjusted non-linearly. Using a fluorescently labeled EB1 MT-binding domain, we previously reported in interphase cells and here in proliferating i3N hiPSCs that π-EB1 photoinactivation attenuated MT growth. Because we were unable to reliably transfect differentiating i3Neurons, we instead used new fluorogenic taxane derivatives as live MT labels (Bucevičius et al., 2020; Lukinavičius et al., 2014) at very low concentrations to analyze length changes of growth cone MTs that were visible both before and during blue light exposure of the entire growth cone. In control i3Neuron growth cones, MTs dynamically extended from the central domain into the growth cone periphery and the average length of these MTs was not affected by blue light. In contrast, in π-EB1 i3Neurons MTs in the growth cone periphery frequently shortened in response to blue light exposure (Figure 2B and C; Figure 2—video 2). While π-EB1 photoinactivation-induced MT shortening was most pronounced in π-EB1 EB3-/- i3Neurons, MTs also shortened in π-EB1 i3Neurons that still expressed EB3 and in both cases the relative MT length after 20 s of blue light was significantly reduced compared with control i3Neurons (p<0.001 by ANOVA and Tukey-Kramer HSD). Although taxane-based probes label growing MT ends more dimly than the rest of the MT possibly due to slow redistribution kinetics on MTs in cells (Ettinger et al., 2016), we were able to measure growth cone MT polymerization rates during blue light exposure by manual tracking MT length changes. Even though MT growth rates in neurons are lower compared with other cells (Stepanova et al., 2003), it is important to note that absolute MT growth rates in this experiment are further underestimated because of the reduced temporal resolution compared with EB1 tracking (Gierke et al., 2010). Nevertheless, compared with control i3Neurons, after 1 min of blue light exposure, growth cone MT growth rates in both π-EB1 lines were reduced at least fourfold (Figure 2D). Thus, taken together these data show that EB1-mediated MT end interactions are required to stabilize and sustain net growth cone MT growth. In addition, our results indicate that endogenous EB3 cannot compensate for the acute loss of EB1 function in i3Neuron growth cones. Because we previously observed that π-EB1 photoinactivation can partially displace EB3 from growing MT ends (van Haren et al., 2018), we cannot completely exclude a dominant negative effect of π-EB1 photoinactivation on EB3. However, because the EB1N-LOV2 expression level in i3Neurons is below that of EB1 in control cells, EB1N-LOV2 likely does not interfere with EB3 binding to MTs by direct competition for binding sites. Our findings are however consistent with not fully understood functional differences between EB1 and EB3 in developing neurons that relate both to their binding sites on MT ends as well as differences in interaction partners. Specifically, EB3 trails behind EB1 on growing MT ends and only EB3 binds to the F-actin regulator drebrin possibly coordinating F-actin and MT dynamics independently of directly controlling MT growth (Poobalasingam et al., 2022; Roth et al., 2018). Alternatively, EB3 may also be more important during later stages of neuron development when EB3 is more highly expressed (Jaworski et al., 2009; Leterrier et al., 2011). Acute EB1 inactivation does not immediately change F-actin dynamics Because 'pioneer' MTs that enter the growth cone periphery participate in growth cone guidance (Buck and Zheng, 2002), and MT and F-actin dynamics are coupled both mechanically and biochemically through Rho GTPase signaling, we next asked how π-EB1 photoinactivation affected F-actin dynamics. MT growth is thought to activate Rac1 which drives leading-edge actin polymerization, while MT shortening activates RhoA that increases actomyosin contractility through the release of regulatory MT-bound factors (Garcin and Straube, 2019; Wittmann and Waterman-Storer, 2001). To investigate growth cone F-actin dynamics in response to π-EB1 photodissociation-mediated growth cone MT shortening, we used SPY650-FastAct, a fluorogenic jasplakinolide derivative that binds strongly to F-actin filaments and at very low concentrations forms fluorescent speckles that serve as reporters of F-actin flow dynamics (Figure 3A; Danuser and Waterman-Storer, 2006). On kymographs perpendicular to the growth cone edge (Figure 3B), we measured the same F-actin retrograde flow rate in control (2.8+/-0.4 µm/min; mean +/-standard deviation) and π-EB1 EB3-/- i3Neurons (2.8+/-0.3 µm/min) in the dark, similar to previous measurements in growth cones from other vertebrate species (Flynn et al., 2012; Geraldo and Gordon-Weeks, 2009; Gomez and Letourneau, 2014). Although growth cone morphology was highly dynamic (Figure 3—video 1), the growth cone F-actin retrograde flow rate remained remarkably constant during blue light exposure (Figure 3C), and we also did not observe consistent changes in overall F-actin dynamics. Because retrograde F-actin flow is directly related to the rate of leading-edge actin polymerization, this indicates that EB1-mediated MT growth or +TIP interactions do not immediately influence growth cone actin polymerization dynamics. Figure 3 with 1 supplement see all Download asset Open asset F-actin dynamics in π-EB1 neuron growth cones. (A) Growth cones of control and π-EB1 EB3-/- i3Neurons labeled with SPY650-FastAct before and during blue light exposure. Apparent relocalization of F-actin to the middle of the growth cone is observed in both conditions and is likely related to the photobleaching of the probe. (B) Kymographs along the filopodia indicated by orange arrowheads in A illustrating F-actin retrograde flow. (C) Quantification of the F-actin retrograde flow rate before and during blue light exposure. Each data point represents the average of at least three flow measurements per growth cone. Gray lines connect data points from the same growth cone. Statistical analysis by paired t-test. In addition, ANOVA of all four groups showed no significant difference between control and π-EB1 EB3-/- i3Neurons (p>0.97 for all pairwise comparisons with Tukey-Kramer HSD). EB1 is required for growth cone advance To test how π-EB1 photoinactivation affected neurite dynamics, we next tracked growth cone position over longer periods of time. Within a 15 min observation window, >90% of π-EB1 neurites visibly retracted when the entire growth cone was exposed to blue light (Figure 4A and B; Figure 4—video 1), regardless of whether these cells expressed EB3 or not, again indicating that EB3 is not able to compensate for the acute loss of EB1 function in early neurite development. In contrast and as expected, control i3Neurons were insensitive to blue light exposure with only ~40% of neurites shortening. Similarly, neurites from π-EB1 i3Neurons were repelled by blue light barriers placed in front of the growth cone while control i3Neurons were unaffected (Figure 4C), and in long term, time-lapse experiments π-EB1 i3Neurons were unable to cross a blue light barrier sometimes after repeated growth attempts over several hours (Figure 4D; Figure 4—video 2). Figure 4 with 2 supplements see all Download asset Open asset π-EB1 photoinactivation blocks growth cone advance. (A) Control and π-EB1 EB3-/- i3Neuron neurites in which MTs were labeled with 4–610 CP-CTX before and during blue light exposure illustrating the retraction of the π-EB1 neurite in blue light while the control neurite continues to advance. In this experiment, the entire growth cone and adjacent neurite were exposed to blue light. (B) Quantification of the neurite length change before and during blue light exposure. Gray lines indicate individual neurites. Blue line is the average of all neurites, and the shaded area indicates the 95% confidence interval. The orange dashed line indicates no change. Statistical analysis by ANOVA and Tukey-Kramer HSD at 15 min of blue light exposure. (C) Quantification of the retraction response of control and π-EB1 EB3-/- i3Neuron growth cones that encounter a blue light barrier. Statistical analysis by Fisher's exact test. (D) Long-term phase contrast time-lapse sequence of a π-EB1 neurite advancing upward on a 10 µm wide stripe of laminin illustrating growth cone retraction every time the growth cone attempts to cross the virtual blue light barrier. Elapsed time is indicated in hours:minutes. Custom-engineered neuron networks with defined connectivity between individual neurons are a potential key to better understand nervous system information processing (Aebersold et al., 2016), and optogenetic guidance for developing neurites could be a useful tool to build such neuron networks from the bottom up. We, therefore, tested if more precise blue light exposure to small regions inside growth cones targeting only a few MTs could be used to control the direction of growth cone advance (Figure 5A). However, this experiment was technically very challenging. Although π-EB1 photo-dissociation remained sharply localized to small blue light-exposed regions (Figure 5—figure supplement 1), it was difficult to correctly place these regions in dynamic, moving growth cones. Hence, most π-EB1 EB3-/- neurites (29 out of 58) still retracted even with localized blue light exposure, but ~38% (22 growth cones, six of which also eventually retracted) turned away from the blue light-exposed region (Figure 5). In contrast, control i3Neurons did not respond and showed no turning bias relative to localized blue light exposure. Figure 5 with 1 supplement see all Download asset Open asset Growth cone turning in response to local π-EB1 photoinactivation. (A) Time-lapse of control and π-EB1 EB3-/- i3Neuron growth cones labeled with SPY555-tubulin (white) and SPY650-FastAct (magenta). The gamma of the tubulin channel was adjusted to 0.6 to better visualize growth cone microtubules (MTs). The blue circle indicates the light-exposed area. (B) Quantification of the relative turning angle in response to local blue light exposure. Gray lines are individual growth cones. Blue line is the average of all growth cones measurements, and the shaded area indicates the 95% confidence interval. The orange dashed line indicates the 0° angle. Statistical analysis by unpaired t-test at 5 min of local blue light exposure. In summary, we present a novel approach to replace the central MT regulator EB1 with a photosensitive variant by inserting a light-sensitive protein interaction module encoded by a short genetic cassette – that we named π-element – into an inter-domain linker by CRISPR/Cas9-mediated genome editing. It is important to note that the π-element only needs to contain ORFs encoding LOV2 and Zdk1 and the required regulatory elements, while the flanking homology arms direct it to the correct genomic locus. The rest of the protein of interest remains encoded by the endogenous gene. Here, we also include a fluorescent protein tag to monitor π-EB1 photodissociation and a leucine zipper coiled-coil to retain dimerization of the π-EB1 N-terminal half during blue light exposure. While we demonstrate the utility of this approach for the MAPRE1 gene in hiPSCs, we believe that a similar strategy could be adapted to many other multidomain proteins supplementing the optogenetic toolbox to investigate localized protein functions in real-time (Wittmann et al., 2020), but it should be noted that photodissociation kinetics may be different with monomeric proteins. The LOV2/Zdk1 module interacts in the dark, which is opposite to all other optogenetic dimerization modules and therefore, enables an unperturbed dark state and acute functional knockout of a given protein activity through blue light exposure. Thus, replacing an endogenous protein with a photosensitive variant in a single genome editing step as presented here has important advantages by allowing modification of proteins for which genetic knockouts might be lethal and further enabling optogenetics in complex cell systems that are not amenable to transient genetic manipulations. This approach allowed us to directly demonstrate that EB1 in developing human cortical neurons is required for sustained growth cone MT growth and growth cone advance. Although we do not completely dissect the molecular mechanism underlying the π-EB1 photodissociation-mediated neurite retraction response, unchanged F-actin polymerization dynamics during blue light indicate that retraction likely results from a loss of growth cone adhesion or an increase in neurite actomyosin contractility consistent with MT-shortening induced RhoA activation (Joo and Olson, 2021). An increase in neurite contractility that does not remain localized to the region of blue light exposure may also explain our difficulty in using π-EB1 photodissociation to control growth cone guidance more precisely. Thus, although other optogenetic tools that provide attractive stimuli could be a promising avenue to control growth cone guidance (Harris et al., 2020), inhibitory stimuli that induce MT depolymerization may be too difficult to control to be practically useful in synthetic biology approaches building defined neuron networks. Nevertheless, it will be interesting to see how π-EB1 in i3Neurons or other types of neurons can be used to analyze how MTs and +TIPs contribute dynamically to other aspects of neuron morphogenesis such as branching, dendritic spine dynamics or synaptic plasticity (Dent et al., 2004; Jaworski et al., 2009). Materials and methods Key resources table Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional informationCell line (Homo sapiens)i3N induced pluripotent stem cellsWang et al., 2017Doxycycline-induced differentiation into cortical i3NeuronsCell line (Homo sapiens)NCI-H1299ATCCCRL-5803Antibodyanti-EB1 (N-terminal epitope, mouse monoclonal)Thermo Fisher ScientificCat# 41–2100, RRID:AB_25335001:1000 (WB)Antibodyanti-EB1 (C-terminal epitope, mouse monoclonal)BD BiosciencesCat# 610534, RRID:AB_3978911:1000 (WB)Antibodyanti-EB3 (rat monoclonal)AbseaCat# KT361:1000 (WB)Antibodyanti-KIF2C (mouse monoclonal)Santa Cruz BiotechnologyCat# sc-81305, RRID:AB_21320511:1000 (WB)Antibodyanti-Oct-3/4 (mouse monoclonal)Santa Cruz BiotechnologyCat# sc-5279, RRID:AB_6280511:500 (IF)Recombinant DNA reagentEB1N-LZ-LOV2 (plasmid)van Haren et al., 2018Addgene plasmid 107614Recombinant DNA reagentMSCV-PIGScott Lowe, unpublishedAddgene plasmid 18751IRES plasmidRecombinant DNA reagentpmCherry-Zdk1-EB1Cvan Haren et al., 2018Addgene plasmid 107695Recombinant DNA reagentpSpCas9(BB)–2A-GFPRan et al., 2013Addgene plasmid 48138Cas9 plasmidSequence-based reagentEB1 Exon 1–2Integrated DNA TechnologiesRT-qPCR primers, Hs.PT.58.1854993Sequence-based reagentEB1 Exon 6–7Integrated DNA TechnologiesRT-qPCR primers, Hs.PT.58.24290642Sequence-based reagentEB3Integrated DNA TechnologiesRT-qPCR primers, Hs.PT.58.20604386Sequence-based reagentDCXIntegrated DNA TechnologiesRT-qPCR primers, Hs.PT.58.118505Chemical compound, drugSPY555-tubulinSpyrochrome / Cytoskeleton Inc.CY-SC2031:2000Chemical compound, drugSPY650-FastActSpyrochrome / Cytoskeleton Inc.CY-SC5051:3000Chemical compound, drug4–610 CP-CTXBucevičius et al., 20205 nM Molecular cloning and genome editing Construct
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)