H. Hidekazu, Structural Biology Research Center and Division of Biological Science, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8601, Japan Fax: +81 52 789 4535 Tel: +81 52 747 6471 E-mail: email@example.com
Katanin p60 (p60-katanin) is a microtubule (MT)-severing enzyme and its activity is regulated by the p80 subunit (adaptor-p80). p60-katanin consists of an N-terminal domain, followed by a single ATPase associated with various cellular activities (AAA) domain. We have previously shown that the N-terminal domain serves as the binding site for MT, the substrate of p60-katanin. In this study, we show that the same domain shares another interface with the C-terminal domain of adaptor-p80. We further show that Ca2+ ions inhibit the MT-severing activity of p60-katanin, whereas the MT-binding activity is preserved in the presence of Ca2+. In detail, the basal ATPase activity of p60-katanin is stimulated twofold by both MTs and the C-terminal domain of adaptor-p80, whereas Ca2+ reduces elevated ATPase activity to the basal level. We identify the Ca2+-binding site at the end of helix 2 of the N-terminal domain, which is different from the MT-binding interface. On the basis of these observations, we propose a speculative model in which spatial rearrangement of the N-terminal domain relative to the C-terminal AAA domain may be important for productive ATP hydrolysis towards MT-severing. Our model can explain how Ca2+ regulates both severing and ATP hydrolysis activity, because the Ca2+-binding site on the N-terminal domain moves close to the AAA domain during MT severing.
ATPase associated with various cellular activities
maltose binding protein
microtubule interacting and trafficking
variant MIT domain of p60-katanin
C-terminal domain of adaptor-p80
Katanin, spastin and fidgetin are three major microtubule (MT)-severing enzymes [1–5]. Katanin is a heterodimeric ATPase composed of two subunits, p60 (p60-katanin) and p80 (adaptor-p80). p60-katanin is the catalytic factor responsible for breaking MTs in an ATP-dependent manner and contains a single ATPase associated with various cellular activities (AAA) domain. Adaptor-p80 is a regulatory factor containing six WD40 repeats, and is not essential for severing activity (Fig. 1) [6,7]. The MT-severing function of katanin is specific for certan cell types and cell cycles. For example, katanin promotes mitosis in proliferating cells, where it severs MTs at the mitotic spindle poles and increases the number of minus ends, resulting in the accumulation of γ-tubulin at the mitotic centrosomes . However, katanin exhibits severing of the metaphase spindles at the spindle poles .
Recently, MT severing by katanin in neuronal cells has received increased attention, because of its contribution to neurite outgrowth and its implication in neurodegenerative diseases. In nascent neurons, it seems critical that the MTs are chopped by katanin at the cell body prior to being translocated into axons; subsequently, the chopped MT fragments disassemble into tubulin dimers that then reassemble into long MTs . By contrast, excessive MT severing by exogenous overexpression of p60-katanin results in a shortening of the total process length of MTs in the hippocampal neuron . In a model of neurodegenerative disease, MT breakdown by katanin triggers the loss of neurite spikes, but not the apoptosis of neurons, while the microtubule-associated protein, tau may protect MTs against attack by katanin . Thus, the mechanism of p60-katanin regulation by adaptor-p80 or other factors such as tau in various cells and tissues, and at different developmental stages, is an important and unresolved issue.
We have previously shown that the substrate recognition mechanism is evolutionarily conserved between p60-katanin and its related AAA ATPase Vps4 [13,14]. Vps4 is responsible for the dissociation of membrane skeleton ESCRT-III fibrils. Note that p60-katanin and Vps4 share a common domain organization typical of type I AAA-ATPases, which consists of an N-terminal substrate-binding region, followed by a single AAA domain at the C-terminus. Despite the low (∼ 20%) amino acid identity, we have shown that the 3D structure of the N-terminal domain of p60-katanin is strikingly similar to that of the microtubule interacting and trafficking (MIT) domain of Vps4 and other MIT domains. Therefore, we named this domain ‘variant MIT’ (vMIT). Helices 2 and 3 of p60-vMIT make up the interface with MTs, which is similar that Vps4-MIT makes with its substrate ESCRT-III . We further predicted that the binding site for p60-katanin in MTs is located at helix 12 of the α-tubulin subunit. Interestingly, this helix was recently proposed as the binding site for tau . On the basis of our structural model of the p60–MT complex, competition between tau and katanin may occur at this 12th helix of α-tubulin. This unexpected coincidence is, however, consistent with the hypothesis proposed by Qiang et al. , in which the physiological role of tau is to protect the MT bundle in axons against the MT-severing activity of p60-katanin. Sudo & Baas  recently proposed this competition between tau (MT-stabilizing) and katanin (MT-severing) as a new therapeutic target for dementia and other neurodegenerative diseases.
In this study, we focused in detail on the role of the vMIT domain in regulating the ATPase activity of p60-katanin. It is known that either MT (substrate) alone or MT + p80 (adaptor) enhances the ATPase activity of p60-katanin [6,17,18], although the underlying molecular mechanism remains unclear. We therefore hypothesized that substrate/adaptor-dependent ATPase activation originates from p60-vMIT via a change in the location of this domain relative to the AAA domain. During the course of biochemical studies, we found that Ca2+ ions inhibit the MT-severing activity of p60-katanin. Furthermore, we found that Ca2+ directly affects the vMIT domain and then cancels the elevated ATPase activity of p60-katanin that occurs in the presence of either MT or the C-terminal domain of adaptor-p80 (p80-CTD). Given that Ca2+ is a key intracellular signal that indirectly regulates MT stability, as well as rearranging the cytoskeleton in vivo, we discuss the physiological implications of this finding.
p80-CTD interacts with helix 1 of p60-vMIT
Previous biochemical studies by McNally and co-workers [7,19] showed that the N-terminal half of p60-katanin interacts with the C-terminal half of adaptor-p80, and in a previous study we showed from the 3D structure of p60-vMIT that the MT interface of p60-vMIT is located on the surface of helix 2/3. Therefore, we first examined the amino acid sequence of adaptor-p80 to identify the boundary of its C-terminal p60-binding domain. Figure 1 summarizes the domain organization of p60-katanin and adaptor-p80, and presents multiple sequence alignments of the two domains (p60-vMIT and p80-CTD) that interact with each other. Both domains are genetically conserved among many higher eukaryotes. Finally, we chose residues 480–655 of human p80 and 483–658 of mouse p80 for expression and purification because this region is markedly conserved among vertebrates (Fig. 1B).
Next, we determined the interface between p60-vMIT and p80-CTD using 1H-15N HSQC experiments. Binding of p80-CTD induced drastic line broadening of NH signals from 15N-labeled p60-vMIT (Figs 2A and S1A). Mapping of the broadened signals upon binding to p80-CTD suggested that a wide area covering the surface of helices 1/2 and 1/3 of p60-vMIT might form the binding site for p80-CTD (Figs 2B and S1B; Table S1). We further examined the key residues involved in p80 binding by using Ala-substituted mutants of p60-vMIT in an in vitro pull-down assay (Fig. 2C). Wild-type p60-vMIT bound to p80-CTD in a ratio of almost 1 to 1, as judged from the density of the bands in SDS/PAGE (Fig. 2C, lanes 3 and 14). All Ala mutants also showed similar binding to p80-CTD. Therefore, we assumed that a polar single mutation does not completely inhibit the interaction between p60-vMIT and p80-CTD, probably because the interface is multivalent and covers a wide area, including helix 1.
Binding of either MT or p80-CTD to p60-vMIT stimulates the ATPase activity of p60-katanin
The ATPase activity of sea urchin and plant p60-katanin is stimulated by MTs alone, or MTs and adaptor-p80 [6,17,18]. In addition, the MT-severing activity of p60-katanin is stimulated by p80/con80 (412–655), according to a 4′-6-diamidino-2-phenylindole assay [6,7,18,20]. We performed experiments using mammalian full-length p60-katanin and p80-CTD, the domain that we isolated above. We investigated the ability of p60-katanin to hydrolyze ATP in the presence of either MTs or p80-CTD in vitro (Fig. 3). Under both conditions, the basal ATPase activity of p60-katanin was stimulated by MTs or p80-CTD in a concentration-dependent manner. In the presence of MTs, stimulation of the ATPase activity of p60-katanin reached almost a maximum at 0.5 μm MTs, corresponding to a p60/MT molar ratio of 0.8. We also found that the ATPase activity of p60-katanin is stimulated by the addition of p80-CTD alone. Stimulation of the ATPase activity of p60-katanin reached almost a maximum at 0.4 μm p80-CTD, corresponding to a p60/p80-CTD molar ratio of 1.0. Similar stimulation of p60-katanin was observed in the presence of both MTs and p80-CTD (Fig. S2A). These results show that binding of either MTs or p80-CTD to the N-terminal domain of p60-katanin stimulates the basal ATPase activity of p60-katanin. In other words, p60-vMIT serves as a sensory regulator for the ATPase activity of p60-AAA.
Ca2+ negatively regulates the ATPase activity enhanced by MT or p80-CTD
Next, we examined the effect of Ca2+ on the ATPase activity of p60-katanin. Interestingly, we found that Ca2+ reduced the increase in ATPase activity induced by MTs or p80-CTD to the basal level (Fig. 4). Basal ATPase activity of p60-katanin in the absence of MTs or p80-CTD was not affected by Ca2+ (Fig. 4A), even at a higher concentration such as 10 mm (data not shown). In the presence of MTs, however, the ATPase activity of p60-katanin decreased as the Ca2+ concentration increased, finally decreasing to the basal level at 5 mm Ca2+ (Fig. 4B). A similar tendency was observed in the presence of p80-CTD, where the ATPase activity decreased to the basal level at 2 mm Ca2+ (Fig. 4C). These results suggest that Ca2+ regulates a putative On/Off mechanism for the ATPase activity of p60-katanin, which may be associated with either the vMIT or AAA domain.
Ca2+ inhibits the MT-severing activity of p60-katanin, but does not inhibit its binding to MTs
The dynamics of the polymerization–depolymerization of MTs is called ‘dynamic instability’ . Growing MTs suddenly start depolymerizing rapidly, known as a ‘catastrophe’. However, shortening MTs often are ‘rescued’ and start polymerizing again . In vitro, higher concentrations of Ca2+ in the mm range induce destabilization of MTs [23,24]; therefore, Ca2+ has been used for biochemical experiments to inhibit polymerization or induce the depolymerization of MTs. Figure 5A shows the effects of Ca2+ on the interaction between full-length p60-katanin and MTs in a cosedimentation assay. We quantified a normalized amount of the degraded tubulin fragments from taxol-stabilized MTs (Fig. 5B, left) and p60-katanin cosedimented with the remaining MT filaments (Fig. 5B, right). In the absence of p60-katanin, marked depolymerization of MT was observed in a Ca2+-dependent manner, probably caused by a catastrophe effect  (Figs 5A, lanes 2–7 and 5B, left). In the presence of wild-type p60-katanin, most of the MTs were severed by p60-katanin (Figs 5A, lanes 8–11 and 5B, left). However, MT severing by p60-katanin was mostly inhibited by the presence of 10 mm Ca2+ (Figs 5A, lanes 12 and 13 and 5B, left). We observed a partial inhibition in the presence of 5 mm Ca2+ in multiple independent experiments (data not shown), thus we concluded that 5 mm of Ca2+ is an approximate threshold concentration for the inhibition of MT severing.
We further examined the MT-binding activity of p60-katanin using a K255A mutant that lacks ATPase activity (data not shown). K255A bound to MTs regardless of the Ca2+ concentration (Figs 5A, lanes 14–19 and 5B, right). Furthermore, K255A might stabilize MTs, because the Ca2+-dependent catastrophe of MTs was suppressed (Figs 5A, left, lanes 14–19 and 5B, left). Recently, McNally & McNally  showed that ATPase-deficient katanin promoted the assembly of meiotic spindles. Our observation of K255A is partially consistent with their result. In addition, we examined the MT-binding activity of p60-katanin with ADP in the absence of Mg2+. p60-katanin with ADP bound to MTs without MT severing regardless of the Ca2+ concentration (Fig. 5A, right). Taken together, we concluded that Ca2+ inhibits the MT-severing activity of p60-katanin without affecting its MT-binding activity.
Ca2+ directly binds to p60-vMIT different from the interface with MT
With the aim of obtaining information on the possible involvement of metal ions in the action mechanism of p60-vMIT, we examined its interaction with Ca2+, and with Ce3+ as an ideal paramagnetic probe for Ca2+ (Figs 6A, S3 and S4A) [25,26]. NMR titration analyses of p60-vMIT at various Ca2+ concentrations are shown in Fig. S3. In the case of Ca2+, residues showing a major chemical shift perturbation were V66, I69 and K71. Because these residues are adjacent to the C-terminus of p60-katanin vMIT domain (residues 1–72), this binding site is assumed to be an artifact of the use of the engineered domain-only construct. When using paramagnetic Ce3+ to enhance the chemical shift changes upon metal binding, residues showing a large change, D23 and S24, were additionally found (Figs 6A and S4A). We succeeded in determining this Ce3+ binding site by analyzing pseudocontact shifts due to Ce3+ (Fig. S4B). The metal was chelated close to D23 and S24. From these results, we assumed that D23 and S24, which are evolutionarily conserved among many higher eukaryotes, form the binding site for Ca2+. This Ca2+-binding site differs from the putative interface between p60-vMIT and tubulin/MT, and so is consistent with our observation that Ca2+ did not inhibit MT binding.
From the structural model of the p60–MT complex, we hypothesized that the Ca2+-binding site is close to the putative interface between the vMIT domain and hexameric AAA ATPase domains (Figs 6B,C and S5), enabling us to propose a model for the regulatory role of the vMIT domain in MT severing (Fig. 7). p60-katanin may form a putative ring-shaped hexamer, as indicated by Hartman & Vale . We assumed that hexameric p60-katanin possesses flexible linkers between the vMIT and AAA domains, and may move over MTs using them freely. In the absence of Ca2+, the ATPase activity of p60-katanin is elevated upon binding to an MT through contact between the vMIT and AAA domains. The enhanced activity of p60-katanin may lead to MT severing. After the reaction, p60-katanin is released from the MT and goes back to the basal state. In the presence of Ca2+, however, p60-katanin does not adopt the activated state regardless of MT binding.
The function of p80-CTD to capture and release p60-vMIT
Our NMR and mutation studies suggest that p80-CTD binds to a wide surface of p60-MIT, especially helix 1/2 and helix 1/3, which differs from the MT-binding surface, helix 2/3 (Figs 2B and S1B) . If so, p80-CTD and MT could simultaneously bind to p60-vMIT, forming a ternary complex. However, our ATPase assay showed only a limited additive effect of p80-CTD and MT on ATPase activity (Fig. S2A). In addition, p80-CTD was excluded from the p60/MT complex (Fig. S6, lanes 9–12). Thus, in our experiments, we did not observe a ternary complex of p80-CTD, p60-vMIT and MT. Steric hindrance caused by p80-CTD binding to the surfaces of helices 1/2 and 1/3 may occur. This unique activity of p80-CTD may explain how adaptor-p80 recruits p60-katanin to MTs. That is, using the N-terminal WD40 domain, adaptor-p80 might localize on the γ-tubulin ring complex at the minus end of an MT , whereas p80-CTD captures p60-vMIT. p60-katanin would then relocate to the MT, whereas p80-CTD releases p60-vMIT. Note that, the in vivo activity of adaptor-p80 against p60-katanin also remains controversial, because adaptor-p80 localizes p60-katanin to the γ-tubulin ring complex and stimulates MT severing , whereas overexpression of adaptor-p80 suppress the MT-severing activity of p60-katanin in certain types of cell .
Existence of an MT-/p80-activated state of p60-katanin
The phenomenon that MT (substrate) as well as p80 (adaptor) binding to p60-katanin increases the ATPase activity of p60-katanin has been reported previously [6,17,18], although the molecular mechanism remains unclear. One explanation might be an equilibrium shift between an inactive dimeric state and an active hexameric state that forms upon binding to the substrate and/or adaptor. For example, the katanin-related type I AAA enzyme Vps4 has been shown to act in a dimer–hexamer equilibrium in a nucleotide-dependent manner, although substrate-dependent ATPase activation of Vps4 has not been reported to date [27,28]. Because multivalent binding between active katanin and MT is expected, MT-binding may promote hexamer formation in p60-katanin. We can rule out this possibility, however, because activation of ATP hydrolysis was observed even upon p60-katanin binding to p80-CTD, which was monomeric in solution (data not shown). This suggests the existence of a unique MT-/p80-activated state of p60-katanin. This activated state of p60-katanin is more likely attributed to an allosteric effect depending on the intramolecular spatial rearrangement between the N-terminal vMIT and the central AAA domains, rather than an equilibrium shift.
The regulatory role of p60-vMIT on its own AAA domain has not been predicted. Nevertheless, it is not surprising if the analogy to other AAA-ATPases, including type II rather than type I enzymes, is considered. In general, the N-terminal domains preceding the AAA domain of type II AAA enzymes may perform three molecular functions: (a) provision of an interface for substrates for ATP hydrolyzing energy; (b) provision of an interface for adaptor proteins that mainly contribute to subcellular localization; and (c) modulation of ATPase activity. We previously reported that p60-vMIT serves as an interface for MTs . We also reported that the N-terminal domain of both valosin-containing protein (VCP)/p97 and N-ethylmaleimide-sensitive fusion protein (NSF) serves as the substrate-binding site, as well as the surface for organelle membrane binding . However, it has been reported that binding to substrates or membranes regulates the ATPase activity of these proteins [30,31]. In VCP/p97, the residues at the contact site between the N-domain and the first AAA domain (D1-domain) were identified in conjunction with Paget’s disease of the bone and front temporal dementia (e.g. R155, A232, T262, and N387) [30,31]. A nucleotide-dependent conformational change in the N-domain relative to the D1-domain was observed by X-ray crystallography [30,31]. A mutation on A232 (A232E), one of the key residues, is associated with increased disease severity . Biochemically, an A232E mutant exhibits elevated ATPase activity . In NSF, a similar mechanism in which the contact site between the N-domain and D1 is associated with enhancement of ATPase activity has also been identified . Note that, although both the sequences and the structures of the N-terminal domains of katanin and the VCP/NSF family are rather different, the AAA domains are highly conserved. Finally, we illustrate the MT-activated state of p60-katanin in Fig. 6C, assuming that the edge of p60-vMIT comes close enough to these interfacial residues on the AAA domain with a minimum structural change upon MT binding. In this model, residues E397 and G237 of p60-katanin are highlighted, and correspond to the interfacial residue N387 and the key residue A232 in VCP, respectively.
Ca2+ alters the spatial domain rearrangement of p60-katanin in its MT-/p80-activated state
In this study, we showed that Ca2+ inhibits the MT-severing activity of katanin, as well as reducing its MT-elevated ATPase activity to basal levels. Although the mechanism of this modulation seems complicated, studying the effect of Ca2+ on p60-katanin may provide clues to the working model of MT severing, which is driven by ATP hydrolyzing energy and catalyzed by katanin, as shown in Fig. 6C. First, we showed that Ca2+ affects p60-katanin activity by directly binding to the vMIT domain (Fig. 6). By contrast, Ca2+ does not affect either MT binding or p80 binding by the domain (Fig. 5, data not shown). Because the Ca2+-binding site on p60-vMIT is different from its MT-binding site, these observations are consistent. Next, we found that Ca2+ does not affect the basal ATPase activity level of p60-katanin in the absence of MT and/or adaptor-p80 (Fig. 4A). By contrast, Ca2+ has an effect only on the MT- and/or p80-elevated ATPase activity, but not the basal activity (Figs 4B,C and S2B). Because a flexible linker (∼ 114 amino acids) connects the N-terminal vMIT domain and the AAA domain, this Ca2+ effect is difficult to explain unless these two domains move spatially close to each other. As discussed above, we assume the existence of the MT-/p80-activated state of p60-katanin in such a domain rearrangement. It is likely that Ca2+ affects this MT-/p80-activated state, while the relative orientation of the two domains, which may be a trigger to elevate ATP hydrolysis, is altered by Ca2+. This idea is further assessed by visualizing the spatial locations of the vMIT domain, Ca2+ ion, and the putative interfacial residues on the AAA domain (Figs 6C and S5).
Biological implications of the regulatory role of Ca2+ in MT-severing by katanin
In this study, Ca2+ negatively regulated the MT-severing function of katanin. In other words, Ca2+ contributed to the stabilization of MTs by protecting MTs from attack by katanin. The interaction of Ca2+ with MTs has been extensively studied because the inhibitory effect of Ca2+ on the in vitro assembly of MTs was reported by Weisenberg . Ca2+ has been shown to destabilize polymerization of MTs directly or indirectly. Serrano et al.  determined the Ca2+-binding sites on both α- and β-tubulin molecules, which are major components of MTs, and demonstrated that tubulin deprived of these sites could still polymerize in the presence of high concentrations of Ca2+. Ca2+ is also known to affect dynamic instability in the assembly of MTs: it enhances the rate of catastrophic degradation of MT . Lefèvre et al.  showed that the Ca2+-binding sites of tubulin overlap with the binding sites for microtubule-associated proteins, and thus Ca2+ competed with microtubule-associated proteins, resulting in the disassembly of MTs.
The indirect effect of Ca2+ on MT destabilization via one of the microtubule-associated proteins, tau, has received increased attention . Calpain, a calcium-dependent nonlysosomal cysteine protease, plays critical roles in the pathway. Tau is highly relevant to regulating MT polymerization, or something of that nature. Hyperphosphorylation of tau, however, abolishes its ability to bind tubulin and promote MT assembly. When MTs release hyperphosphorylated tau, the protein aggregates into paired helical filaments, which are the neuropathological hallmarks of Alzheimer’s disease. There are at least three protein kinases responsible for the hyperphosphorylation; cyclin-dependent-kinase 5, glycogen synthase kinase 3β and dual-specificity tyrosine-phosphorylation-regulated kinase 1A [37–39]. The activity of all the three kinases is regulated by calpain, as follows. In the presence of high concentrations of Ca2+, glycogen synthase kinase 3β is truncated by calpain, generating two fragments of ∼ 40 and 30 kDa. Glycogen synthase kinase 3β truncation augments its kinase activity . Cyclin-dependent-kinase 5 is also activated by proteolytic cleavage of its specific activator p35 by calpain . Recently, dual-specificity tyrosine-phosphorylation-regulated kinase 1A was shown to be proteolyzed and activated by calpain in Alzheimer disease brain . In addition, calpain directly degrades tau, which also causes its loss of MT-stabilizing activity [43,44].
To date, Ca2+ has not been thought to stabilize MTs, but rather to promote their disassembly, as mentioned above. Our new findings on katanin, however, show that Ca2+ can slow the disassembly of MTs, suggesting that the assembly and disassembly of MT is managed by a more complicated system than expected. Excess Ca2+ is generally harmful to MTs, both directly and indirectly. To survive an accidental increase in Ca2+, cells may have developed various fail-safe mechanisms, one of which might be the Ca2+ inhibition of katanin, which would function to avoid the rapid breakdown of MTs for the survival of neuronal cells upon a transient Ca2+ flux.
In conclusion, we have shown that p60-katanin, a type I AAA ATPase, possesses a novel mechanism for regulating its own ATPase activity, probably via a 3D domain rearrangement between the N-terminal vMIT domain and the AAA domain. Ca2+ binds p60-vMIT at the edge of this domain, close to the contact site between vMIT and AAA. This putative arrangement may well explain why Ca2+ inhibits MT severing without inhibiting either ATPase activity or MT-binding.
Production of p80-CTD
We generated a multiple sequence alignment of the C-terminal sequences of adaptor-p80 excluding the N-terminal WD40 repeats (residues 1–269) using clustal x , from which we decided to clone sequences 480–614, 480–630 and 480–655 of human and mouse p80. Expression vectors for the recombinant glutathione S-transferase (GST)-tagged form of the human and mouse p80 fragments were constructed using PRESAT vector methodology , as derived from the pGEX-4T3 vector (GE Healthcare, Buckinghamshire, UK). Finally, p80 (480–655), termed p80-CTD, was used for further analysis because this construct produced a more soluble protein than the others. Human p80-CTD, prepared by expression in Escherichia coli BL21 (DE3), followed by affinity purification on glutathione–Sepharose (GE Healthcare, Buckinghamshire, UK) and thrombin digestion, was used for ATPase assays. The expression vector for the recombinant MBP-tagged p80-CTD of mouse was also constructed by a standard protocol using PCR, and ligated into the EcoRI–HindII sites of pMAL-c2X (NEB). The fusion protein was prepared by expression in E. coli BL21 (DE3), followed by affinity purification on amylose-sepharose (New England Biolabs, Ipswich, MA, USA), and used for NMR experiments and binding experiments with p60-vMIT.
Production of p60-katanin
The expression vector for the recombinant GST-tagged full-length p60-katanin of mouse was constructed by a standard protocol using PCR, and ligated into the BamHI–SalI sites of pGEX-6P3 (GE Healthcare, Buckinghamshire, UK) . An Ala-substituted mutant (K255A) was engineered with the QuikChange site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA). Two complementary oligonucleotides with mutated sequences for each mutant were used as primers (Table S2). The fusion proteins were produced in E. coli JM109. Expression was induced with 0.1 mm isopropyl thio-β-d-galactoside, and Luria–Bertani cultures were grown overnight at 20 °C. For cosedimentation assays, GST-tagged proteins were purified with glutathione–Sepharose and eluted in elution buffer (50 mm Tris/HCl, pH 7.4, 100 mm NaCl, 1 mm dithiothreitol, 50 mm reduced glutathione, and 5% glycerol). For ATPase assays, proteins were eluted in elution buffer (20 mm Tris/HCl, pH 7.5, 100 mm NaCl, 1 mm dithiothreitol, 1 mm EGTA, 2 mm MgCl2, 0.25 mm ATP, 0.02% Triton X-100 and 5% glycerol) after PreScission Protease digestion on the column.
Production of the recombinant GST-tagged p60-vMIT domains of mouse, including wild-type and Ala-substituted mutants, was carried out as described previously (Table S2) . The fusion proteins were expressed in E. coli BL21 (DE3), affinity-purified on glutathione–Sepharose and dialyzed. These fusion proteins were used for MBP-tagged p80-CTD binding assays. For NMR spectroscopy, a 1-L culture was incubated with [15N]-ammonium chloride as the sole nitrogen source by following a standard fermentation protocol at 25 °C. Purification of 15N-labeled p60-vMIT was achieved by glutathione–Sepharose affinity chromatography followed by thrombin digestion, benzamidine–Sepharose chromatography and gel filtration using a Superdex 75 column (GE Healthcare, Buckinghamshire, UK). The complex of 15N-labeled p60-vMIT with p80-CTD was purified from a mixture of 15N-labeled GST-tagged p60-vMIT and non-labeled MBP-tagged p80-CTD prepared following FactorXa and thrombin digestion, respectively.
Continuous ATPase assays of p60-katanin
ATPase activity was measured using an ATP regenerating system. The reaction mixture contained 50 mm Tris/HCl, pH 7.5, 50 mm KCl, 2 mm MgCl2, 2 mm phosphoenolpyruvate, 1 mm ATP, 50 μg·mL−1 pyruvate kinase, 50 μg·mL−1 lactate dehydrogenase and 0.2 mm NADH. The reactions were performed in the presence of 0.4 μm p60-katanin, and ATPase activity was measured by monitoring the decrease in NADH absorption at 340 nm at room temperature using a UV–Vis spectrophotometer (UV mini-1240; Shimadzu, Tokyo, Japan). The shifts in absorption at 340 nm caused by the addition of taxol-stabilizing MTs and p80-CTD were measured. The absorbance data were representative of two or three independent experiments and were normalized relative to the level of the time-point zero. ATP hydrolysis activity was then calculated and represented in the unit of micromoles of ATP hydrolyzed per min per mg of p60-katanin (units·mg−1).
In vitro binding experiments
MBP-p80 bound to amylose resin was incubated with lysate of GST-tagged p60-vMIT domains in binding buffer (20 mm Tris/HCl, pH 7.4, 200 mm NaCl and 1 mm EGTA) for 2 h at 4 °C. The beads were washed four times in wash buffer (20 mm Tris/HCl, pH 7.4, 200 mm NaCl and 1 mm EGTA). The associated proteins were analyzed by SDS/PAGE.
Cosedimentation assays for MT-binding and MT-severing activity of p60-katanin
Taxol-stabilized MTs (1 μm) and GST-tagged p60-katanin (3.5 μm) were incubated for 30 min at 25 °C in binding buffer (80 mm Pipes, pH 7.0, 2 mm MgCl2, 1 mm EGTA, 1 mm ATP and 20 μm taxol). Reaction mixtures (50 μL) were spun through a glycerol cushion buffer (50% v/v glycerol in binding buffer; 100 μL) for 30 min at 100 000 g. Supernatants and pellets ware analyzed by SDS/PAGE. The band intensity of proteins was quantified with IMAGEJ (National Institutes of Health, Bethesda, MD, USA) (http://rsbweb.nih.gov/ij/).
1H–15N HSQC spectra of 15N-labeled p60-vMIT in the presence or the absence of p80-CTD were acquired at 25 °C on a Bruker AVANCE III 600 MHz spectrometer equipped with a cryogenic probe. These samples were dissolved in 20 mm sodium phosphate, pH 7.5, and 150 mm NaCl, and the protein concentrations were below 50 μm because of the low solubility of p80-CTD. 1H-15N HSQC spectra for Ce3+ titration experiments were recorded using 0.1 mm15N-labeled p60-vMIT at 25 °C on a Bruker AVANCE 500 MHz spectrometer equipped with a cryogenic probe. 1H-15N HSQC spectra for Ca2+ titration experiments were similarly performed (Fig. S3). The samples were dissolved in 20 mm Hepes, pH 7.5, and 25 mm NaCl. All data were processed using nmrpipe  and sparky  software. The position of the coordinating Ce3+ ion was calculated by using the program fantasian  coupled with an inhouse grid-search program with 144 pseudocontact shift value of NH signals. For calculation the coordinates of hVps4b-MIT (PDB: 1wr0) and p60-vMIT (PDB: 2rpa) were used. All of the figures were prepared by molmol .
A hexameric ring model of the AAA domains of p60-katanin was generated on the basis of the hexameric ring structure of p97 D1 (PDB: 1s3s) using modeller (version 9v6) (http://salilab.org/modeller/), as described previously . A molecular model of the complex of p60-vMIT with a tubulin oligomer (PDB: 3du7) was constructed on the basis of the complex between Vps4a-MIT and CHMP1a (PDB: 2jq9)  by replacing each component corresponding to p60-vMIT and helix 12 of α-tubulin, respectively . Finally, the complex model of full-length p60-katanin and part of an MT was built by connecting these two models by a flexible linker corresponding to residues 91–204, and by placing the hexameric AAA domain close to the vMIT domain. The model was adjusted and visualized using molmol .