Pathogenic mutations in the ALS gene CCNF cause cytoplasmic mislocalization of Cyclin F and elevated VCP ATPase activity

Abstract

Amyotrophic lateral sclerosis (ALS) is an adult-onset motor neuron disease characterized by a progressive decline in motor function. Genetic analyses have identified several genes mutated in ALS patients, and one of them is Cyclin F gene (CCNF), the product of which (Cyclin F) serves as the substrate-binding module of a SKP1–CUL1–F-box protein (SCF) ubiquitin ligase complex. However, the role of Cyclin F in ALS pathogenesis has remained unclear. Here, we show that Cyclin F binds to valosin-containing protein (VCP), which is also reported to be mutated in ALS, and that the two proteins colocalize in the nucleus. VCP was found to bind to the NH2-terminal region of Cyclin F and was not ubiquitylated by SCF^{Cyclin F} in transfected cells. Instead, the ATPase activity of VCP was enhanced by Cyclin F in vitro. Furthermore, whereas ALS-associated mutations of CCNF did not affect the stability of Cyclin F or disrupt formation of the SCF^{Cyclin F} complex, amino acid substitutions in the VCP binding region increased the binding ability of Cyclin F to VCP and activity of VCP as well as mislocalization of the protein in the cytoplasm. We also provided evidence that the ATPase activity of VCP promotes cytoplasmic aggregation of transactivation responsive region (TAR) DNA-binding protein 43, which is commonly observed in degenerating neurons in ALS patients. Given that mutations of VCP identified in ALS patients also increase its ATPase activity, our results suggest that Cyclin F mutations may contribute to ALS pathogenesis by increasing the ATPase activity of VCP in the cytoplasm, which in turn increases TDP-43 aggregates.

Introduction

Amyotrophic lateral sclerosis (ALS) is an adult-onset progressive neurodegenerative disease that is characterized by loss of both upper and lower motor neurons, resulting in muscle weakness and eventual death from respiratory insufficiency within 3–5 years after diagnosis (1,2). Cytoplasmic aggregation of TAR DNA-binding protein 43, a nucleic acid-binding protein, is a pathological hallmark of ALS observed in degenerating neurons of more than 90% patients, and these aggregates are supposed to contribute to ALS pathogenesis (3). However, molecular mechanisms on how these aggregates form remain largely unclarified.

Genetic analyses demonstrated that only about 10% of ALS are hereditary with dominant inheritance (familiar ALS), whereas the remaining large proportions of ALS patients are sporadic (4). To date, mutations in several genes were identified in sporadic and familial ALS patients (5), and the functions of proteins encoded by these genes have been associated with varieties of biological processes including RNA metabolism and protein homeostasis, suggesting the molecular mechanisms of disease pathogenesis but not leading to conceptual understanding of ALS (3–6).

Mutations in the Cyclin F gene (CCNF) encoding Cyclin F are recently identified in European, American, Australian and Asian ALS patients (7). Cyclin F is a ubiquitously expressed ~ 100kDa protein that contains an F-box domain, a Cyclin-box domain, two nuclear localization signals (NLSs) and a PEST (proline-, glutamic acid-, serine- and threonine-rich) domain (8,9). The F-box domain mediates binding to the SKP1–CUL1–ROC1 catalytic module of the SCF^{Cyclin F} ubiquitin ligase complex, whereas the Cyclin box recognizes an RXL motif (where X is any amino acid) in substrate proteins and thereby triggers their ubiquitylation. On the one hand, the degradation of Cyclin F is thought to be mediated through PEST domain by a metalloprotease or β-TrCP ubiquitin ligase in the M phase of the cell cycle (10,11). Furthermore, destruction boxes (D boxes) identified within the Cyclin-box domain render Cyclin F susceptible to degradation by the anaphase-promoting complex/cyclosome ubiquitin ligase in the G₁ phase (12). By these mechanisms as well as regulation at the transcriptional level (13), the amount of Cyclin F protein is strictly controlled at a low level in the M–G₁ phases and upregulated in the S–G₂ phases, suggesting that Cyclin F functions mainly in the S–G₂ phases. Therefore, most functional studies of Cyclin F focused on the ubiquitylation of proteins regulating the S phase progression (12, 14–16), the DNA damage response (17,18) and microtubule organization (19,20). Whereas most neurons reside in the G₁ phase, the function of Cyclin F in neurons has remained largely unexplored.

The sites of amino acid changes in Cyclin F that are associated with ALS have been found to be distributed throughout the protein molecule (from S3G to I772T of the 786-amino acid protein). Although one CCNF frameshift variant (p.L372 fs) has been identified, it did not segregate with disease (7). Given that all of other mutations of CCNF that have been associated with ALS are not protein disrupting (such as nonsense, frameshift and splice site), the identified missense mutations likely confer a gain of function that is cytotoxic. Furthermore, mice heterozygous for Cyclin F ablation do not manifest signs of motor dysfunction, whereas homozygous loss of Cyclin F is embryonic lethal (21).

To elucidate the molecular mechanism by which dysfunction or dysregulation of Cyclin F due to amino acid substitutions identified in ALS patients contributes to disease pathogenesis, we screened ALS-related proteins for their ability to bind to Cyclin F. We identified valosin-containing protein (VCP), an AAA+ ATPase that unfolds and segregates client proteins from macromolecular complexes or membranes (22) as one such binding partner and found that Cyclin F enhances ATPase activity of VCP, which was also observed by point mutations in VCP gene identified in ALS patients (23,24). Four ALS-associated mutations in CCNF located in the VCP binding region (S3G, K97R, T181I and S195R) caused the mislocalization of Cyclin F to the cytoplasm and increased the association of Cyclin F with VCP, resulting in further activation of VCP ATPase. We then provided evidence that VCP ATPase activity plays a positive role in TDP-43 aggregate formation. Taken together, our data indicate that Cyclin F and VCP contribute to the same molecular pathway that facilitates TDP-43 aggregation.

Results

ALS-associated mutations of Cyclin F do not disrupt formation of the SCF^{Cyclin F} ubiquitin ligase complex

Recent genetic analyses of familial as well as sporadic ALS patients have identified 10 missense mutations in CCNF (Fig. 1A) (7). Given that the most studied function of Cyclin F is that of a substrate adaptor in an SKP1–CUL1–F-box protein (SCF) ubiquitin ligase complex, we first examined whether Cyclin F mutants are able to form the SCF^{Cyclin F} complex. HEK293T cells were transfected with expression vectors for FLAG epitope-tagged WT or mutant forms of Cyclin F, and extracts of the transfected cells were then subjected to immunoprecipitation with the antibody to FLAG. Consistent with previous reports, immunoprecipitates prepared from cells expressing FLAG-tagged WT Cyclin F contained SKP1, CUL1 and ROC1 in addition to FLAG-Cyclin F (WT) (Fig. 1B). Immunoprecipitates prepared from cells expressing each of the 10 Cyclin F mutants also contained SKP1, CUL1 and ROC1 (Fig. 1B), indicating that ALS-associated mutations of Cyclin F do not disrupt formation of the SCF^{Cyclin F} complex.

Figure 1

Cyclin F mutants form an SCF complex and are not abnormally stabilized. (A) Schematic representation of the human Cyclin F protein showing its domain organization and the location of amino acid substitutions identified in ALS patients. NLS, nuclear localization signal (amino acids 20–28 and 568–574); F, F-box domain (amino acids 35–76); Cyclin, Cyclin-box domain (amino acids 271–549); PEST, PEST domain (amino acids 582–766). (B) Extracts prepared from HEK293T cells transiently expressing FLAG-tagged WT or the indicated mutant forms of Cyclin F were subjected to immunoprecipitation (IP) with anti-FLAG, and the resulting precipitates were subjected to immunoblot analysis (IB) with the indicated antibodies. The asterisk indicates a nonspecific band. (C) HeLa cells transiently expressing FLAG-tagged WT or mutant forms of Cyclin F were incubated in the absence or presence of cycloheximide (CHX) for 4 h and then subjected to immunoblot analysis with antibodies to FLAG and to tubulin (loading control). (D) SH-SY5Y cells expressing EGFP-tagged WT or mutant forms of Cyclin F were incubated in the absence or presence of cycloheximide (CHX) for indicated times and then subjected to immunoblot analysis with antibodies to EGFP and to tubulin (loading control).

ALS-associated mutations do not induce abnormal stabilization of Cyclin F

CCNF mutations have been thought to result in a gain of toxic function for the encoded protein (7). Given that Cyclin F is a short-lived protein (10) and that some ALS-associated mutations relate to protein stability (25–27), we examined whether Cyclin F mutants are more stable than Cyclin F (WT). We treated HeLa cells transiently expressing the FLAG-tagged WT or mutant proteins with cycloheximide (CHX) to inhibit protein synthesis and then examined the protein amount of exogenously expressed Cyclin F by immunoblot analysis. Both the WT and mutant forms of Cyclin F were almost completely degraded after 4 h (Fig. 1C). We also examined the stability of Cyclin F, which was expressed in neuronal cells. We infected SH-SY5Y cells with lentiviral vector, which permits doxycycline-regulated expression of enhanced green fluorescent protein (EGFP)-tagged Cyclin F (either WT or mutants) and treated the cells with doxycycline followed by CHX. Immunoblot analysis revealed no significant difference in the degradation rate between WT and mutant forms of Cyclin F (Fig. 1D), suggesting that ALS-associated mutations do not induce abnormal stabilization of Cyclin F.

Cyclin F binds to VCP

ALS is a neurodegenerative disease that usually manifests in adulthood when neurons are in non-dividing states. Given that the identified substrates of SCF^{Cyclin F} participate in cell cycle progression, DNA repair or microtubule organization in the S, G₂ or M phase, the function of Cyclin F in G₁-phase cells such as neurons has remained largely unexplored. To provide insight into the function of Cyclin F relevant to ALS pathogenesis, we first screened ALS-related proteins for their ability to bind Cyclin F. HEK293T cells were transfected with expression vectors for FLAG-tagged Cyclin F and a series of hemagglutinin (HA) epitope-tagged ALS-related proteins, and extracts prepared from the transfected cells were then subjected to immunoprecipitation with anti-FLAG. Among the nine proteins tested, VCP (also known as p97 and CDC48)—followed by SQSTM1 (also known as p62) and TDP-43—showed the greatest ability to bind to Cyclin F (Fig. 2A), prompting us to focus on VCP in the present study. We confirmed the interaction between endogenous Cyclin F and VCP proteins in HEK293T cells and in mouse brain by immunoprecipitation and immunoblot analysis (Fig. 2B). Immunofluorescence staining also revealed that the two endogenous proteins largely colocalized in the nucleus of U2OS cells (Fig. 2C), suggesting that Cyclin F binds to VCP in the nucleus.

Figure 2

Cyclin F binds to VCP. (A) Extracts of HEK293T cells transiently transfected with plasmids encoding FLAG-Cyclin F and the indicated HA-tagged ALS-related proteins were subjected to immunoprecipitation with anti-FLAG, and the resulting precipitates as well as the original cell extracts (Input) were subjected to immunoblot analysis with the indicated antibodies. (B) HEK293T cells were treated with 10 μm of MG132 for 6 h (to induce accumulation of endogenous Cyclin F), lysed and subjected to immunoprecipitation with anti-Cyclin F or control IgG (upper panel) or whole brain lysate were subjected to immunoprecipitation with anti-Cyclin F or control IgG (lower panel). The resulting precipitates as well as the original cell or tissue extract (1% Input) were subjected to immunoblot analysis with antibodies to VCP and to Cyclin F. (C) U2OS cells treated with 10 μm of MG132 for 6 h (to induce accumulation of endogenous Cyclin F) were fixed, permeabilized and subjected to immunofluorescence staining with antibodies to Cyclin F and to VCP. Nuclei were also stained with DAPI. Scale bar, 10 μm.

VCP exists in cells as a homohexamer that is formed through interactions among the two ATPase domains (D1 and D2) of each monomer, whereas the NH₂-terminal domain (NTD) of each subunit is extruded from the ATPase core and associates with various cofactors required for specific functions (22). To delineate the region of VCP that mediates binding to Cyclin F, we constructed expression vectors for ATPase domain-deleted (N fragment) or NTD-deleted (C fragment) forms of VCP (Fig. 3A). We detected interaction of Cyclin F with full-length VCP as well as with the C fragment, but not with the N fragment, in HEK293T cells, suggesting that Cyclin F binds to the ATPase domains of VCP (Fig. 3B). Cyclin F was previously shown to recognize an RXL motif in substrate proteins. VCP contains a single RXL motif in the D2 ATPase domain, suggesting that Cyclin F might bind to VCP via this motif. However, we found that mutation of the RXL sequence of VCP to AXA (Mut.) did not affect binding to Cyclin F (Fig. 3C), indicating that the RXL motif is not required for the association of VCP with Cyclin F.

Figure 3

The NH₂-terminal region of Cyclin F binds to the ATPase domains of VCP. (A) Schematic representation of human full-length (FL) VCP as well as of mutants thereof examined for their ability to bind to Cyclin F. NTD, NH₂-terminal domain; Mut, RXL motif mutant. (B, C) Extracts prepared from HEK293T cells transiently expressing FLAG-tagged Cyclin F and HA-tagged WT or mutant forms of VCP were subjected to immunoprecipitation with anti-FLAG, and the resulting precipitates as well as the original cell extracts (Input) were subjected to immunoblot analysis with the indicated antibodies. (D) Schematic representation of human WT Cyclin F as well as of deletion mutants thereof examined for their ability to bind to VCP. (E) Extracts prepared from HEK293T cells transiently expressing HA-tagged VCP and FLAG-tagged WT or mutant forms of Cyclin F were subjected to immunoprecipitation with anti-FLAG, and the resulting precipitates as well as the original cell extracts (Input) were subjected to immunoblot analysis with the indicated antibodies.

We next generated various deletion mutants of Cyclin F (Fig. 3D) and found that deletion of the NH₂-terminal region (ΔN) disrupted its binding to VCP (Fig. 3E), indicating that this region of Cyclin F is responsible for the association with VCP. The binding of Cyclin F to VCP was also attenuated by deletion of the PEST domain (ΔPEST), and this attenuation was partially reversed by additional deletion of the Cyclin-box domain (ΔCΔP) (Fig. 3E). These results thus suggested that the NH₂-terminal region of Cyclin F binds to the ATPase domains of VCP and that this interaction is inhibited by the Cyclin-box domain of Cyclin F in the absence of the PEST domain.

VCP is not likely a substrate of the SCF^{Cyclin F} ubiquitin ligase

Cyclin F functions as a substrate adaptor in the SCF^{Cyclin F} ubiquitin ligase complex. Given that the F-box domain of Cyclin F is responsible for its association with the SKP1–CUL1–ROC1 core module of this complex, a Cyclin F mutant lacking this domain (ΔF) would be expected to act in a dominant negative manner. Consistent with this notion, the ubiquitylation level of RRM2, a known Cyclin F substrate (14), was markedly enhanced by ectopic expression of Cyclin F (WT), whereas it was slightly reduced by expression of Cyclin F (ΔF), in HEK293T cells. (Fig. 4A). We next examined whether VCP is a substrate of SCF^{Cyclin F} under the same experimental conditions and found that expression of neither WT nor ΔF mutant forms of Cyclin F had a substantial effect on the extent of VCP ubiquitylation (Fig. 4B). These data thus indicated that VCP is not a substrate for ubiquitylation by SCF^{Cyclin F}.

Figure 4

VCP is not likely a substrate of the SCF^{Cyclin F} ubiquitin ligase. HEK293T cells transfected with plasmids encoding Myc epitope-tagged Cyclin F (WT or ΔF), FLAG-tagged RRM2 (A) or VCP (B) and HA-tagged ubiquitin, as indicated, were treated with 10 μm of MG132 for 5 h, lysed and subjected to immunoprecipitation with anti-FLAG under denaturing conditions (0.1% SDS). The resulting precipitates as well as the original cell extracts (Input) were then subjected to immunoblot analysis with the indicated antibodies.

Cyclin F increases the ATPase activity of VCP

Genetic analysis of ALS patients has revealed several missense mutations in VCP (28–30). Although most of these mutations are located in the region encoding the NTD, some were found to increase the ATPase activity of VCP, an effect that might be relevant to ALS pathogenesis (23,24). On the basis of these previous observations, we examined the possible effect of Cyclin F on the ATPase activity of VCP. FLAG-tagged forms of VCP and Cyclin F were purified separately from transfected HEK293T cells, and the ATPase activity of FLAG-VCP was measured with an enzyme-coupled assay in vitro (31). The ATPase activity of VCP was significantly increased by the addition of Cyclin F (WT) (Fig. 5). As might be expected, loss of its VCP binding region (ΔN) significantly impaired the ability of Cyclin F to enhance the ATPase activity of VCP (Fig. 5).

Figure 5

Cyclin F enhances the ATPase activity of VCP. The ATPase activity of FLAG-tagged VCP was measured with the use of an enzyme-coupled assay in the absence or presence of FLAG-tagged WT or ΔN mutant forms of Cyclin F isolated from transfected HEK293T cells. Activity is expressed relative to that measured in the absence of Cyclin F. Individual value points are represented by dots, and data are means ± SD from three independent experiments. ^*P < 0.05 (one-way ANOVA followed by Tukey’s post-hoc test).

Enhanced ability of ALS-associated Cyclin F mutants to bind to and activate VCP in the cytoplasm

Then, we examined the effect of ALS-associated mutations on the ability of Cyclin F to upregulate the ATPase activity of VCP. All five Cyclin F mutants tested (whose mutations are located in the VCP binding N-terminal region of Cyclin F) either showed a significant increase in binding activity to endogenous VCP or tended to bind to it to a greater extent than did the WT protein in HEK293T cells (Fig. 6A). In addition, two Cyclin F mutants were tested and showed increased ATPase activity of VCP significantly compared with Cyclin F (WT) (Fig. 6B). These data thus indicated that ALS-associated amino acid substitutions located in the NTD of Cyclin F enhance its ability to bind to and to activate VCP.

Figure 6

Increased ability of ALS-associated Cyclin F mutants to bind to and activate VCP in the cytoplasm. (A) Extracts prepared from HEK293T cells transfected with plasmids encoding HA-tagged WT or the indicated mutant forms of Cyclin F were subjected to immunoprecipitation with anti-HA, and the resulting precipitates as well as the original cell extracts (Input) were subjected to immunoblot analysis with the indicated antibodies (upper panel). The intensity of each VCP band in the immunoprecipitates was determined by densitometry and normalized by that of the corresponding HA-Cyclin F band (lower panel). Data are expressed relative to the normalized value for Cyclin F (WT) and are means ± SEM from three independent experiments. ^*P < 0.05, ^**P < 0.01 versus WT (one-way ANOVA followed by Tukey’s post-hoc test). (B) The relative ATPase activity of VCP in the absence or presence of WT or mutant forms of Cyclin F was measured as in Figure 5. Individual value points are represented by dots, and data are means ± SD from three independent experiments. ^**P < 0.01 versus WT (one-way ANOVA followed by Tukey’s post-hoc test). (C) U2OS cells transiently expressing FLAG-tagged Cyclin F (WT or mutants) were fixed, permeabilized and subjected to immunofluorescence staining with antibodies to FLAG. Nuclei were also stained with DAPI. Scale bar, 20 μm.

To determine where in the cell Cyclin F mutants activate VCP, we examined the intracellular localization of these mutants. Immunofluorescence analyses of U2OS cells transfected with expression vectors encoding FLAG-tagged Cyclin F with a series of mutations in the VCP binding region revealed the mislocalization of Cyclin F mutants in the cytoplasm with apparent aggregate structure (Fig. 6C). Consistent with the endogenous Cyclin F, WT Cyclin F was found to localize primarily in the nucleus (Fig. 6C). These results indicated that Cyclin F mutants activate VCP in the cytoplasm.

The ATPase activity of VCP promotes cytoplasmic aggregation of TDP-43

We then sought to investigate the potential role of increased VCP ATPase activity in ALS pathogenesis. It is reported that the degenerating neurons in patients with ATPase hyper-active VCP mutations exhibit cytoplasmic TDP-43 aggregates (28), suggesting that the enhanced VCP ATPase activity might cause cytoplasmic TDP-43 aggregation and neuronal cell death. To test this possibility, we first generated U2OS cells, which express NLS-mutated TDP-43 tagged with FLAG in a manner dependent on doxycycline treatment (to circumvent the toxicity of overexpression of NLS-mutated TDP-43), and then we treated the cells with hydrogen peroxide (H₂O₂) to induce cytoplasmic TDP-43 aggregation (32). Overexpression of wild-type VCP reduced cells with cytoplasmic TDP-43 aggregates, and this suppressive function was independent of its ATPase activity because ATPase inactive E578Q mutant also had the same effect (Fig. 7A). In contrast, ATPase hyper-active mutant R191Q did not show any significant effect on TDP-43 aggregation (Fig. 7A), suggesting that the ATPase-independent function of VCP in suppressing TDP-43 aggregation was canceled by enhanced ATPase activity. This hypothesis was confirmed by the treatment of NMS-873, a specific inhibitor of VCP ATPase activity (33), which magnificently reduced cytosolic TDP-43 aggregates (Fig. 7B). These results indicated dual functions of VCP in formation of TDP-43 aggregates: ATPase-dependent acceleration and ATPase-independent prevention. Together, our data indicated that mutations in the VCP binding region in CCNF cause abnormal localization of Cyclin F in the cytoplasm where Cyclin F mutants activate VCP ATPase activity, resulting in the aggregate formation of TDP-43 (Fig. 8).

Figure 7

The ATPase activity of VCP enhances cytoplasmic TDP-43 aggregation. (A) Inducible TDP-43 expressing U2OS cells were transiently expressed as HA-tagged wild-type, E578Q (ATPase inactive) or R191Q (ATPase hyper-active) mutant forms of VCP. Then, FLAG-tagged NLS-mutated TDP-43 was induced to express by treatment of H₂O₂ (1 μm for 2 h) leading to cytoplasmic TDP-43 aggregation. Cells were fixed, permeabilized and subjected to immunofluorescence staining with antibodies to FLAG (red) and to HA (green). Nuclei were also stained with DAPI. Scale bar, 20 μm. Cells containing cytoplasmic TDP-43 aggregates with HA-VCP expression (indicated by orange arrowheads) or without HA-VCP expression (indicated by white arrowheads) were manually counted in the same field to adjust the experimental conditions. More than 100 cells were counted for each cell, and the ratio of cells with cytoplasmic TDP-43 aggregates was calculated. Data are expressed relative to the HA-VCP non-expressed cells and are means ± SEM from three independent experiments. ^*P < 0.05 (Student’s t test). (B) Inducible TDP-43 expressing U2OS cells were induced to express FLAG-tagged NLS-mutated TDP-43 and treated with NMS-873 (indicated concentrations for 30 min) followed by H₂O₂ (1 μm for 2 h). Cells were fixed, permeabilized and subjected to immunofluorescence staining with antibodies to FLAG. Nuclei were also stained with DAPI. Scale bar, 20 μm. Cytoplasmic TDP-43 aggregates were indicated by the yellow arrowheads.

Figure 8

Summary. ALS-associated point mutations in CCNF gene cause nuclear-to-cytoplasmic translocation of Cyclin F, leading to abnormal activation of cytoplasmic VCP ATPase activity and consequent cytoplasmic aggregation of TDP-43.

Discussion

As the number of genetic mutations identified in ALS patients increases, the molecular characterization of these mutations in relation to disease pathogenesis is failing to keep pace. One such example is provided by mutations in CCNF, with studies of the encoded protein having focused almost entirely on its role in cell cycle progression as a component of the SCF^{Cyclin F} ubiquitin ligase complex. This complex catalyzes ubiquitylation of substrates in the S, G₂ and M phases of the cell cycle, but the function of Cyclin F in the G₁ phase has remained unclear. There are 10 amino acid-altering mutations identified in CCNF. Recently, it was demonstrated that substitution of serine 621 to glycine lost phosphorylation site resulting in increase in the ubiquitylation activity of SCF^{Cyclin F} by still unknown mechanisms (34,35). However, whether other CCNF mutations also have the same effect and how this is related to ALS pathogenesis remain unclear. In this study, we have identified VCP as a binding partner of Cyclin F and the ATPase activity of VCP increased by binding of Cylin F. This effect on VCP seems to be independent of ubiquitylation by Cylin F. As far as we are aware, Cyclin F is the first protein identified to affect the ATPase activity of VCP.

VCP is a highly abundant protein, with an estimated > 1 × 10⁶ molecules being present in a single HeLa cell (36), and it has been shown to regulate protein homeostasis by extracting specific client proteins from the endoplasmic reticulum, mitochondria and chromatin either for their degradation by the proteasome or autophagy or in order to facilitate their recycling (22,37). This segregase activity relies on conformational changes in VCP driven by its ATPase activity. We have now shown that WT Cyclin F and VCP colocalize in the nucleus, suggesting that Cyclin F might be required for the efficient removal of client proteins from chromatin on the basis of its ability to facilitate cycles of adenosine triphosphate (ATP) binding to and hydrolysis by VCP. In contrast to WT Cyclin F, our immunofluorescence images also revealed cytoplasmic localization of Cyclin F mutants, which bind to VCP, indicating that these Cyclin F mutants abnormally enhance VCP ATPase activity in the cytoplasm. From this perspective, VCP ATPase activity has been demonstrated to engage in elimination of aggregated cytoplasmic TDP-43 by autophagy (38,39). In addition, VCP mutants identified in IBMPFD (inclusion body myopathy with early-onset Paget disease and frontotemporal dementia) and ALS were shown to induce TDP-43 distribution from the nucleus to the cytosol with aggregate-like structure (40), and some of these VCP mutants were demonstrated to show higher ATPase activity (23,24). Combined with these previous findings, our data presented here supported the notion that enhanced VCP ATPase activity in the cytoplasm, resulting from CCNF mutations, causes TDP-43 cytoplasmic aggregation. As noted above, pathogenic VCP mutations are not restricted to ALS but have also been identified in individuals with IBMPFD (41,42), the affected cells of which also contain TDP-43 aggregates (43). The possible involvement of Cyclin F in IBMPFD remains to be investigated.

Amino acid substitutions identified in Cyclin F mutants associated with ALS are distributed throughout the protein. Given that all mutants we tested showed increased binding to VCP, these mutations might affect protein structure in such a way as to promote such binding. The extent of the increase in binding of Cyclin F mutants to VCP did not appear to be correlated with that of the increase in VCP ATPase activity, suggesting that the mechanism of VCP activation cannot be explained simply on the basis of Cyclin F binding. With analogy to an allosteric inhibitor of VCP that binds to the cavity formed by the D1 and D2 domains of adjacent VCP monomers and thereby increases the affinity of the VCP hexamer for ADP (33), we speculate that Cyclin F might facilitate the dissociation of ADP from the ATPase domains of VCP.

Our data indicate that Cyclin F, VCP and TDP-43 converge on the same molecular pathway in relation to ALS pathogenesis, with this pathway not being dependent on the ubiquitylation function of Cyclin F. Of note, several gene products mutated in ALS patients play a key role in protein degradation mediated by the ubiquitin-proteasome or autophagy-lysosome systems (44), and we therefore do not exclude the possibility that Cyclin F mutants also disrupt or promote protein degradation mediated by the SCF ubiquitin ligase, as demonstrated recently for S621G mutants (34,35).

In addition to VCP and TDP-43, we detected the interaction of Cyclin F with SQSTM1, which was also recently reported by another research group (35). SQSTM1, a selective autophagy receptor (45), is a component of neuronal inclusions that form in patients with a large expansion of a hexanucleotide repeat in C9ORF72 (46), which is the most prevalent mutation identified in ALS patients to date (47). Since the dysregulation of autophagy has been implicated in ALS pathogenesis (48), whether and how Cyclin F might regulate SQSTM1 in relation to ALS warrant further investigation.

Materials and Methods

Cell culture, transfection and infection

HEK293T, HeLa, U2OS and SH-SY5Y cells were purchased from ATCC and cultured under 5% CO₂ at 37°C in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, penicillin (50 U/ml), streptomycin (50 μg/ml), 2 mm of l-glutamine, 1% minimum Eagle’s essential medium nonessential amino acids and 1% sodium pyruvate (all from Thermo Fisher Scientific). HEK293T, HeLa or U2OS cells were transiently transfected with plasmid DNA with the use of the PEI MAX reagent (Polyscience). SH-SY5Y cells or U2OS cells, which express EGFP-tagged Cyclin F (WT or mutants) or FLAG-tagged TDP-43 (NLS mutant, which contains K82A, R83A and K84A substitutions), respectively, in response to doxycycline (1 μg/ml for 48 h), were generated by infection of lentivirus encoding rtTA (driven by UbC promoter) and EGFP-Cyclin F or FLAG-TDP-43 (driven by TRE promoter), respectively, followed by G418 selection (500 μg/ml for at least 7 days). Lentiviruses were produced by HEK293T cells transfected with psPAX2 (addgene plasmid 12260), pMD2.G (addgene plasmid 12259) and pSLIK-neo (addgene plasmid 25735), which contain cDNA encoding EGFP-Cyclin F or FLAG-TDP-43.

Construction of plasmids

Complementary DNAs encoding Cyclin F, FUS, OPTN, RRM2, superoxide dismutase 1, SQSTM1, TAF15, TBK1, TDP-43, UBQLN2 or VCP were amplified from HEK293T or U2OS cells, cloned into the pENTR vector (Thermo Fisher Scientific) and verified by sequencing. The resulting pENTR vectors were subjected to recombination with the p3×FLAG-CMV, pcDNA3-3myc or pcDNA3-HA destination plasmid (49) with the use of LR clonase II (Thermo Fisher Scientific) in order to obtain expression vectors for the corresponding proteins. Point and deletion mutants were generated by polymerase chain reaction-based mutagenesis. cDNAs encoding EGFP-Cyclin F (WT or mutants) or FLAG-TDP-43 (containing K82A, R83A and K84A substitutions) were cloned into the pEN_TTGmiRc2 vector (addgene plasmid 25 753) in which EGFP and miR-30a coding region was removed by NcoI and MunI restriction enzymes and verified by sequencing. The resulting pEN_TTGmiRc2 vectors were subjected to recombination with the pSLIK-neo destination plasmid.

Antibodies

Anti-VCP (ab11433) and horseradish peroxidase (HRP)-conjugated anti-FLAG (ab49763) were obtained from Abcam. Anti-SKP1 (#12248) and HRP-conjugated anti-HA (11667475001) were from Cell Signaling Technology and Roche, respectively. Anti-FLAG (F1804) and anti-tubulin (T6074) were from Sigma. Anti-CUL1 (32-2400) and anti-EGFP (A-11122) were from Thermo Fisher Scientific, and anti-Myc (sc-40), anti-Cyclin F (sc-952) and normal rabbit IgG (sc-3888) were from Santa Cruz Biotechnology. Anti-ROC1 was kindly provided by Y. Xiong (University of North Carolina at Chapel Hill) (38). HRP-conjugated anti-mouse IgG (W4021) and HRP-conjugated anti-rabbit IgG (W4011) were from Promega. Alexa Fluor 488-conjugated anti-rabbit IgG (A-11008) and Alexa Fluor 546-conjugated anti-mouse IgG (A-11030) were from Thermo Fisher Scientific. The specificity of anti-CUL1 was validated by western blot analyses with extracts prepared from cells in which CUL1 was knocked down by transient transfection of Cas9 targeted to Cul1 gene. That of anti-VCP, anti-SKP1 and anti-Cyclin F antibodies was previously validated by other research groups (19,50,51).

Immunoprecipitation and immunoblot analysis

Cells or whole brain taken from 2-month-old C57BL/6N male mice were washed with phosphate-buffered saline (PBS) and lysed for 10 min at 4°C in NP-40 lysis buffer [0.5% Nonidet P-40, 50 mm of Tris–HCl (pH 7.5), 150 mm of NaCl, 10% glycerol, aprotinin (10 μg/ml), leupeptin (10 μg/ml), 1 mm of Phenylmethanesulfonyl fluoride (PMSF), 0.4 mm of sodium orthovanadate, 0.4 mm of EDTA, 10 mm of NaF and 10 mm of sodium pyrophosphate]. The crude lysates were centrifuged at 20 000g for 15 min at 4°C, and the resulting supernatants were incubated with Dynabeads protein G (Thermo Fisher Scientific) that had been conjugated with the indicated antibodies. The immune complexes were washed three times with wash buffer (0.1% Triton X-100 and 10% glycerol in PBS) and then subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) for immunoblot analysis with the indicated primary antibodies. For direct immunoblot analysis, total cell extracts were prepared with Radio-Immunoprecipitation Assay (RIPA) buffer [50 mm of Tris–HCl (pH 8.0), 0.1% SDS, 150 mm of NaCl, 1% Nonidet P-40 and 0.5% sodium deoxycholate] and then centrifuged at 20 000g for 15 min at 4°C. The resulting supernatants were subjected to SDS-PAGE for immunoblot analysis. Mice were maintained in a specific pathogen-free facility at the Institute of Animal Experimentation, Graduate School of Medicine, Tohoku University, and they were provided with water and rodent chow ad libitum and treated according to The Standards for the Humane Care and Use of Laboratory Animals of Tohoku University and to Guidelines for Proper Conduct of Animal Experiments of the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Ubiquitylation assays

Ubiquitylation assays were performed as described previously (52). HEK293T cells transfected with vectors for FLAG-VCP or FLAG-RRM2, Myc-Cyclin F (WT or ΔF mutant) and HA-ubiquitin were treated with 10 μm of MG132 for 5 h, lysed with NP-40 lysis buffer containing 0.1% SDS to disrupt non-covalent protein–protein interactions and then subjected to immunoprecipitation with anti-FLAG followed by immunoblot analysis with the indicated antibodies.

CHX chase analysis

CHX chase analysis was performed as described previously (53), with some modifications. HeLa cells were transfected with expression vectors for FLAG-tagged WT or mutant forms of Cyclin F 24 h before exposure to CHX (25 μg/ml) for 4 h and subjected to immunoblot analysis. SH-SY5Y cells containing pSLIK-neo-EGFP-Cyclin F vector were treated with doxycycline (1 μg/ml) for 48 h before exposure to CHX (25 μg/ml) for up to 1 h and subjected to immunoblot analysis.

Immunofluorescence staining

Immunofluorescence staining was performed as described previously 53, with some modifications. U2OS cells grown on glass coverslips were fixed in 4% paraformaldehyde for 10 min, washed with PBS and permeabilized for 10 min with PBS containing 0.5% Triton X-100. After exposure to 1% bovine serum albumin (BSA) in PBS, the cells were incubated with primary antibodies for 16 h at 4°C, washed three times with PBS containing 0.1% Tween 20, incubated with Alexa Fluor-conjugated secondary antibodies for 45 min at room temperature and washed again before exposure to 4',6-diamidino-2-phenylindole (DAPI) (5 μg/ml) for 1 min. The cells were then examined with an Axio Imager 2 microscope (Zeiss). The number of cells with TDP-43 aggregates was manually counted.

Measurement of ATPase activity

The ATPase activity of VCP was assayed by an enzyme-coupled method as described previously (31), with some modifications. FLAG-VCP and FLAG-Cyclin F (WT or mutants) were separately expressed in HEK293T cells, immunoprecipitated with anti-FLAG and eluted with the 3×FLAG peptide (Sigma). The purity and concentration of the proteins were estimated by SDS-PAGE with BSA as a quantitation standard. The rate of ATP hydrolysis was calculated from the linear phase of the decrease in A₃₄₀ of NADH at 37°C as measured in 60 μl of assay buffer [50 mm of Tris–HCl (pH 9.0), 150 mm of KCl, 2 mm of MgSO₄, 3 mm of phosphoenolpyruvate, 0.25 mm of NADH and 2 mm of ATP] supplemented with enzymes (1.5 U of pyruvate kinase, 1.0 U of lactate dehydrogenase and 1 μg of FLAG-VCP) and FLAG-Cyclin F (0.5 μg).

Statistical analysis

Quantitative data are presented as means ± standard deviation (SD) or standard error of the mean (SEM) and were analyzed with Student’s t test or with one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test. A P value of < 0.05 was considered statistically significant.

Acknowledgements

We thank Y. Xiong for providing antibodies to ROC1 as well as K. Murayama and laboratory members for discussion. This work was supported by JSPS KAKENHI [Grant Numbers 15K18365 and 17K14955 to T.N. and 17H04035 to K.N.] and the China Scholarship Council to Y. J.

Conflict of Interest statement. None declared.

Author notes

Present address: Department of Biobank, Tohoku Medical Megabank Organization, Tohoku University, Sendai, Miyagi 980-8573, Japan