Loss of function of VCP/TER94 causes neurodegeneration

Variants in several genes have been found to cause frontotemporal lobar degeneration (FTLD). Although genetic variants in the tau (also known as MAPT) gene cause pathological changes that are associated with the formation of tau-positive inclusions, variants in other genes, such as progranulin, VCP, CHMP2B and TDP43 (also known as TARDBP), cause diseases associated with ubiquitin-positive and tau-negative inclusion bodies that also include TDP43 (Josephs, 2008; Olney et al., 2017). In particular, variants in VCP cause a specific type of FTLD, inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia (IBMPFD) and amyotrophic lateral sclerosis (ALS), also referred to as multisystem proteinopathy type 1 (MSP1) (Watts et al., 2004; Pfeffer et al., 2022; Scarian et al., 2022).

VCP (also called TERA/p97/CDC48) is a member of the AAA ATPase family and is involved in diverse cellular functions. VCP contains two ATPase catalytic domains called the D1 and D2 domains and regulatory N-terminal and C-terminal domains. It forms a symmetric hexametric structure and couples ATP hydrolysis to multiple unrelated cellular activities through interactions with binding partners such as p47 (also known as PLEK) and the UFD1–NPL4 (also known as NPLOC4) complex (Woodman, 2003; Xia et al., 2016). For instance, VCP facilitates the translocation of misfolded proteins across the endoplasmic reticulum (ER) membrane in ER-associated degradation (ERAD) and mitochondria-associated degradation (Ye et al., 2001; Heo et al., 2010). VCP is implicated in membrane fusion events, including the assembly of the nuclear membrane, Golgi apparatus and transitional ER (Kondo et al., 1997; Roy et al., 2000; Hetzer et al., 2001). VCP is also involved in autophagy (Ju et al., 2009; Tresse et al., 2010; Ferrari et al., 2022).

Variants in the N-terminal domain, the D1 ATPase domain, the D2 ATPase domain and the N-D1 linker of VCP have been found to be associated with IBMPFD and ALS (Pfeffer et al., 2022; Scarian et al., 2022). However, VCP variants in the N-terminal domain or D1 ATPase domain have been suggested to elevate ATPase activity in vitro (Weihl et al., 2006; Halawani et al., 2009; Manno et al., 2010). Consistently, the expression of mutant VCP in cultured cells causes the accumulation of ubiquitin, decreases proteasome activity, impairs ERAD, elevates ER stress and eventually leads to cell death (Weihl et al., 2006; Ju et al., 2008; Gitcho et al., 2009). However, RNA interference (RNAi)-mediated knockdown (KD) of VCP induces the accumulation of ubiquitinated proteins and cytoplasmic vacuolization in Drosophila S2 cells and human HeLa cells (Wojcik et al., 2004), and ER vacuolization and impaired protein trafficking in COS-7 cells (Mimnaugh et al., 2006). Expression of dominant-negative VCP causes the accumulation of ubiquitinated proteins, ERAD impairment and swelling of the ER (Dalal et al., 2004). Therefore, it is unclear whether these variants cause disease phenotypes through gain of toxic function or loss of normal function.

At the animal level, transgenic and knock-in mouse models have been generated. Both types of mouse model show degeneration of the brain associated with ubiquitin- and TDP43-positive protein inclusions (Weihl et al., 2007; Badadani et al., 2010; Custer et al., 2010). These changes might be explained by gain of abnormal function. However, the critical issue with these mouse experiments was that they lacked rescue experiments that could exclude the artificial effects of mutant protein overexpression. Consistent with this notion, mice with conditional knockout of VCP in neurons show phenotypes similar to disease phenotypes, such as brain atrophy and TDP43 pathology (Wani et al., 2021). Although the expression of the VCP mutant R115C failed to rescue these phenotypes in the conditional knockout mice, it is also necessary to compare these effects with the rescuing effect of wild-type VCP to assess whether mutant VCP retains its rescuing ability.

Recently, we analyzed VCP T262A knock-in mice (Fujita et al., 2013; Homma et al., 2021). Our data revealed that developmental microcephaly was caused by G1/S arrest in neural stem cells. Insufficient DNA damage repair in neural stem cells was caused by elevated phosphorylation of mini-chromosome maintenance helicase (MCM)3, which is essential for the G1/S cell cycle transition. In utero gene therapy inducing the expression of wild-type VCP or nonphosphorylated mutant MCM3 successfully rescued DNA damage, neuronal necrosis and TDP43 aggregation in adult knock-in mice (Fujita et al., 2013; Homma et al., 2021). These data suggest that early-stage developmental abnormalities in G1/S cell cycle regulation underlie the pathology of FTLD.

In this study, we aimed to determine whether the disease-associated mutation in VCP is a loss-of-function or a gain-of-function mutation by using a Drosophila model. Drosophila has an evolutionally conserved ortholog of VCP called TER94 (Pintér et al., 1998). This Drosophila ortholog was knocked down with RNAi, and we found that partial deficiency of TER94 caused neurodegeneration in the brain and eye. In the rescue experiment, normal TER94 rescued these phenotypes. In contrast, the disease-linked A229E mutant barely rescued the phenotypes. Our data strongly suggest that loss of the cellular functions of VCP is the pathological basis of the disease linked to the VCP gene. Our data from the Drosophila model also suggest that defects in neuronal proliferation during development underlie the pathology of FTLD.

We examined the phenotypes caused by RNAi-mediated KD of TER94 driven by the combination of UAS-dsRNA and Gal4 drivers (Brand and Perrimon, 1993). First, we quantified the effect of KD by real-time quantitative PCR. Ubiquitous Actin5C-Gal4-driven KD of TER94 resulted in a significant decrease in the transcription of TER94 in whole first-instar larvae (Fig. 1A), whereas Actin5C-Gal4-driven KD of green fluorescent protein (GFP), which was used as a control, did not cause a decrease in TER94 transcripts. Western blot analysis using an antibody against TER94 confirmed that Actin5C-Gal4-driven KD of TER94 caused a decrease in the expression of the TER94 protein (Fig. S1A,B). The Actin5C>TER94-KD flies showed complete lethality during development (Fig. 1B). We also examined the effects of elav-Gal4- and Cha-Gal4-driven TER94 KD. elav-Gal4 induces weak expression in all neurons (Luo et al., 1994). Cha-Gal4 contains a promoter of the Choline acetyl transferase (Cha; also known as ChAT) gene and induces expression in all cholinergic neurons, which constitute the majority of the excitatory neurons in the Drosophila nervous system (Kitamoto, 2001). elav>TER94-KD flies showed reduced viability during development (Fig. 1B), and almost all the flies died within 4 days after eclosion (Fig. 1C). KD of GFP driven by elav-Gal4, which was performed as a control experiment, did not affect viability or lifespan (Fig. S2A,B), suggesting that the expression of double-stranded RNA (dsRNA) itself does not have significant effects. Cha-Gal4-mediated TER94 KD did not affect viability during development. However, most flies died 20-30 days after eclosion (Fig. 1D). Thus, TER94 KD clearly affected the survival of the animals.

Because elav>TER94-KD flies, in which TER94 is knocked down in all neurons, died several days after eclosion, we examined the morphology of the brains of these flies immediately after eclosion. They were found to have a decreased brain volume compared with that of wild-type flies (Fig. 2A,C). Although the overall volume of the brain was decreased, its structure seemed not to be seriously disrupted. The volume of the central brain was decreased, but the optic lobes were not affected (Fig. 2D,E). The morphology of mushroom bodies was further examined by labeling mushroom body neurons by immunostaining using an anti-Fasciclin II (FasII; also known as Fas2) antibody (Crittenden et al., 1998; Awasaki et al., 2000). The morphology of mushroom bodies in the TER94 KD flies was severely affected, resulting in loss of almost all their structures, including all the lobes (Fig. 2B,F). These data indicated that TER94 KD caused neurodegeneration.

We also performed similar experiments on the compound eye. Compound eye-specific GMR-Gal4-driven KD of TER94 resulted in progressive loss of pigments in the compound eye (Fig. S3A). Examination of the internal morphology of the compound eye revealed progressive deterioration of tissues accompanied by vacuolization and loss of rhabdomeres (Fig. S3B,D). The thickness of the eye was decreased at eclosion and in aged flies (Fig. S3C). These data indicated that TER94 KD caused progressive tissue degeneration in the compound eye.

Next, we examined whether TER94 KD affects the nuclear localization of TBPH, the Drosophila ortholog of mammalian TDP43 (Lukacsovich et al., 1999), because changes in the localization of TDP43 are a well-defined pathological feature of IBMPFD. We knocked down TER94 in third-instar larval motor neurons in the ventral nerve cord using the OK6-Gal4 driver (Shiraishi et al., 2014). Immunostaining with an antibody against TBPH revealed that the localization of TBPH in the nucleus was significantly diminished by TER94 KD (Fig. 2G,H), but the amount of TBPH mRNA was not significantly changed (Fig. S4A). Western blotting also revealed that the level of TBPH protein was decreased by TER94 KD (Fig. S4B).

Our data showed that decreased expression of TER94 resulted in deterioration of neural tissues. Our findings raised the possibility that IBMPFD-associated mutations in VCP cause loss of its function and result in a dominant disease due to haplo-insufficiency or through a dominant-negative effect. To test this hypothesis, we examined whether the phenotypes of TER94 KD flies could be rescued by additional expression of wild-type and disease-associated mutant TER94.

We first examined the effect of ectopic expression of wild-type and A229E mutant TER94 (Ritson et al., 2010) (Fig. 3A), because we could test their rescuing ability under the prerequisite that ectopic expression itself does not cause a phenotype. The site of the A229E mutation of Drosophila TER94 is close to the D1 ATPase domain, and this mutation is a homolog of the A232E variant of human VCP. Ectopic expression of wild-type or mutant TER94 by elav-Gal4 did not affect viability (Fig. 3B), lifespan (Fig. 3C), central brain volume (Fig. S5A,B) or mushroom body volume (Fig. 4A,B). Similarities in lifespan among elav-Gal4>+, elav-Gal4>wild-type TER94 and elav-Gal4>mutant TER94 flies also confirmed the homogeneity of the genetic background in these flies (Fig. 3C). However, GMR-Gal4-driven expression of wild-type and mutant TER94 caused eye degeneration and lethality, respectively (Fig. S6A). Quantitative PCR revealed that the expression of GFP driven by GMR-Gal4 was far higher than that driven by elav-Gal4 (Fig. S6B). This suggested that when TER94 was expressed by GMR-Gal4, overexpression itself caused phenotypes. For this reason, we used only elav-Gal4 in the rescue experiments.

We tested whether the phenotypes resulting from TER94 KD driven by elav-Gal4 could be rescued by additional expression of wild-type or mutant TER94. Both showed an ability to rescue the developmental semilethality of TER94 KD (Fig. 3D). Whereas wild-type TER94 showed a remarkable rescue effect on the decrease in lifespan, mutant A229E TER94 showed a very weak rescue effect (Fig. 3E). Co-expression of GFP did not rescue the developmental semilethality or lifespan shortening, as was observed in the control experiment (Fig. S2C,D). Wild-type TER94 had a remarkable rescue effect on the decreases in central brain volume and mushroom body volume, whereas mutant TER94 showed a relatively weak rescue effect (Fig. 4A,C; Fig. S5A,C). Western blot analysis using an anti-TER94 antibody revealed that mutant TER94 expression was induced at a level comparable to that of wild-type TER94 expression (Fig. 3F; Fig. S7). These data indicated that the A229E mutation caused loss of the physiological functions of the TER94 gene.

Our previous study investigating VCP knock-in mice revealed that a VCP mutation causes developmental microcephaly because of G1/S arrest in neural stem cells. We found that the phosphorylation of MCM3 and MCM2 was elevated in VCP T262A knock-in mice (Homma et al., 2021). MCM2 and MCM3 are components of a protein complex that forms the core of the replicative DNA helicase and is involved in the licensing of DNA replication origins (Remus et al., 2009; Li et al., 2015). Our previous study also suggested that early developmental abnormalities in G1/S cell cycle regulation underlie the pathology of FTLD because induced expression of wild-type VCP or nonphosphorylated mutant of MCM3 successfully rescued DNA damage, neuronal necrosis and TDP43 aggregation in adult knock-in mice (Homma et al., 2021).

Therefore, we examined whether overexpression of Mcm2 or Mcm3 rescues the TER94 KD phenotypes. To test a prerequisite of the rescue experiment, we examined the effect of ectopic expression of Mcm2 and Mcm3 driven by neuronal elav-Gal4. Both did not have any effect on viability (Fig. S8A). However, when Mcm2 was overexpressed, ∼20% of the flies died within 4 days of eclosion. When Mcm3 was overexpressed, lifespan was shortened to ∼40 days (Fig. S8B). Therefore, we decided to test the rescuing ability of Mcm2 and Mcm3 before eclosion and in day 1 adult flies.

We examined whether developmental semilethality caused by TER94 KD could be rescued. Expression of Mcm2 fully rescued fly viability, whereas expression of Mcm3 did not (Fig. S9B). We also examined whether the reduction in mushroom body volume could be rescued. Expression of Mcm2 significantly rescued the mushroom body phenotype, whereas expression of Mcm3 did not (Fig. S9A,C,D, Fig. S10A). Our real-time PCR data suggested that overexpression of Mcm2 does not affect TER94 KD itself (Fig. S10B). These data indicated that overexpression of Mcm2 partially rescued the phenotypes of TER94 KD.

Next, we examined whether the decrease in mushroom body volume resulted from abnormalities in the pupal developmental stage. We examined the mushroom body volume of pupae 72 h after puparium formation (APF). TER94 KD caused marked reduction in mushroom body volume, which was significantly rescued by overexpression of Mcm2 (Fig. 5A-C). Thus, our data suggested that the phenotypes of TER94 KD resulted from abnormalities in G1/S cell cycle transition in the developmental stage.

Our data show that RNAi-mediated depletion of TER94, an ortholog of human VCP, caused brain degeneration, eye degeneration and premature lethality. KD driven by the ubiquitous and strong driver Actin5C-Gal4 resulted in developmental lethality. elav-Gal4-mediated KD in all neurons resulted in developmental semilethality, decreased brain volume, decreased mushroom body volume and early death within 4 days of eclosion. Compound eye-specific KD mediated by GMR-Gal4 caused severe degeneration of tissue. Remarkable cell loss and vacuolization were observed in adult flies just after eclosion. These pathological changes progressed rapidly toward complete deterioration of tissue structure. In addition, TER94 KD caused a decrease in endogenous TBPH, an ortholog of TDP43, in the nuclei of larval motor neurons, and this could have been caused by loss of the physiological function of TER94 in the nuclear entry of TBPH as previously suggested (Kushimura et al., 2018). These observations indicated that the KD of TER94 recapitulates the pathological changes caused by VCP mutations in humans.

Importantly, the phenotypes of TER94 KD flies were rescued by wild-type TER94 but were rescued to a lesser extent by mutant TER94. All of the phenotypes caused by elav-Gal4-driven TER94 KD were successfully rescued by co-expression of wild-type TER94, suggesting that these phenotypes were caused by KD of the TER94 gene. Because our data showed that RNAi-mediated depletion of TER94 caused neurodegeneration, we assessed the relevance of these phenotypes to the pathological changes caused by genetic mutations associated with IBMPFD. We performed genetic rescue experiments by inducing additional expression of wild-type and mutant TER94. Additional expression of wild-type TER94 had remarkable rescue effects on semilethality, early death after eclosion, decreased central brain volume and decreased mushroom body volume. Among these measures, viability and central brain volume were restored to wild-type level, whereas lifespan and mushroom body volume were not completely restored to wild-type level. However, this could be attributed to the weak expression induced by elav-Gal4- and RNAi-mediated KD of the induced transcripts. Although our western blot analysis revealed that mutant TER94 expression was induced at a level comparable to that of wild-type TER94, it may be that the expression of both exogenous and endogenous TER94 varies among animals. This made it difficult to assess how much of the mRNA or protein was overexpressed. The fact that our phenotypic data clearly revealed that exogenous co-expression of wild-type TER94 with the dsRNA resulted in significant recovery of the KD-induced phenotypes suggests that these rescuing effects have been caused by additive expression of exogenous TER94, which may have caused the TER94 expression level to surpass the threshold dosage for generating the mutant phenotypes. In addition, our experiments compared the effects of co-expression of wild-type TER94, which resulted in significant rescue effects, with the those of co-expression of mutant TER94 or GFP, which resulted in weak or no rescue effects. This comparison ruled out the possibility that titration of UAS caused the KD to be less effective, resulting in rescuing effects.

In our rescue experiments, the A229E mutant TER94 rescued only weakly the phenotypes of TER94 KD. If A229E mutant TER94 newly gains a toxic function, the phenotypes would have been worsened in the rescue experiments. However, we did not observe such results in our experiments. Thus, our rescue experiments indicated that FTLD caused by the A229E mutation is due to loss of function of the TER94 protein. These findings further suggested that human IBMPFD is caused by loss of function of VCP due to the A229E mutation.

Our results are consistent with the phenotypes of conditional VCP knockout mice. Neuronal VCP loss of function results in phenotypes including cortical brain atrophy and the formation of TDP43 inclusions, resembling the pathology of VCP-associated disease (Wani et al., 2021). The finding that induced expression of R115C mutant VCP in conditional knockout mice fails to rescue the knockout phenotypes is consistent with the results of our rescue experiments, although it is necessary to compare the effect of this mutant with that of wild-type VCP to assess its rescue ability.

All VCP mutations associated with human IBMPFD are missense mutations, and they are located in the N-terminal domain or the D1 ATPase domain in many cases (Pfeffer et al., 2022; Scarian et al., 2022). Previous studies suggested that disease mutations alter the conformations of the N-terminal domain and the D1 ATPase domain (Tang et al., 2010; Tang and Xia, 2013). The N-terminal domain is suggested to bind with co-factors and to link ATP hydrolysis with the unfolding or disassembly of substrate proteins (Uchiyama et al., 2002; Dreveny et al., 2004). Thus, disease mutations may alter co-factor binding and co-factor-regulated ATP hydrolysis (Zhang et al., 2015). These findings could provide links between missense mutations and loss of physiological functions.

Our data suggest that the phenotypes of TER94 KD began in the pupal developmental stage because elav>TER94-KD pupae showed a decreased mushroom body volume. This developmental defect was suggested to be at least partly caused by G1/S arrest in neuroblasts because the pupal phenotype was rescued by overexpression of Mcm2. Although elav>TER94-KD flies showed markedly decreased mushroom body volume, the mushroom bodies seemed to be almost specifically affected, because the overall morphology of the brain was not severely disrupted and the optic lobe volume was not affected. This almost specific loss of mushroom bodies is similar to that reported in a previous study in which the proliferation of neuroblasts was blocked by hydroxyurea, which interferes with DNA synthesis (Sweeney et al., 2012). In our previous study of VCP knock-in mice, the phosphorylation of MCM2 and MCM3 was found to be elevated, and these molecules were found to be involved in G1/S arrest in neural stem cells (Homma et al., 2021). Our previous work suggested that phosphorylation of MCM2 was elevated in VCP knock-in mice. If mammalian VCP and Drosophila TER94 share similar molecular functions, TER94 KD may cause hyperphosphorylation of Mcm2, and this could be rescued by overexpression of Mcm2. Despite our knock-in mouse study identifying MCM3, but not MCM2, as a critical factor in G1/S arrest, our rescue experiments identified Mcm2, but not Mcm3, as a causative factor of the phenotypes of TER94 KD. Although these data suggest the evolutionally conserved importance of the hexametric MCM complex, which consists of MCM2-7 and plays critical roles in DNA replication origin licensing, in the causative mechanisms of VCP-associated disease that affect the proliferation of neural stem cells, the role of each subunit may be different between Drosophila and mammals. In our previous study of knock-in mice, gene therapy in the developmental stage rescued adult phenotypes (Homma et al., 2021). Thus, our Drosophila data also support the conclusion that pathological changes in the cell cycle in neural stem cells in early development underlie late-onset FTLD pathology.

In conclusion, our study suggests that loss of function rather than functional gain of function of TER94/VCP is linked to neurodegeneration. Our genetic rescue experiments further revealed that loss of VCP function underlies the pathogenetic mechanisms of the dominantly inherited disease IBMPFD.

Flies were raised on yeast–cornmeal medium (7.5% cornmeal, 3.8% yeast, 9.4% glucose, 3.0% wheat germ, 0.24% n-butyl p-hydroxybenzoate, 0.09% calcium chloride, 1.1% potassium tartrate hemihydrate and 0.9% agar). All fly stocks were maintained at 25°C under a 12:12 h light–dark cycle. All recombinant DNA experiments were approved by the Toho University Safety Committee for Recombinant DNA Experiments (20-454). Canton-S was used as the wild type. UAS-TER94-WT and UAS-TER94-A229E fly strains were a gift from Dr J. P. Taylor (St Jude Children's Research Hospital, Memphis, TN, USA). The UAS-TER94-KD strain (P{GD9777}v24354) was obtained from the Vienna Drosophila Resource Center. The UAS-GFP-KD strain (GFP-IR-1) was obtained from the National Institute of Genetics (NIG)-FLY stock center. The UAS-Mcm2 strain (F001810) was obtained from the FlyORF Zurich ORFeome Project (Bischof et al., 2013). Expression of Mcm3 was induced by the UAS element in the P(EPg)HP10020 transgene in a fly strain that was obtained from Bloomington Drosophila Stock Center (stock number 21855).

For measurement of TER94 expression by quantitative PCR, ten whole first-instar larvae, ten compound eyes from day 1 female adults, ten brains and ventral nerve cords from wandering third-instar larvae, and five to ten brains from day 1 female adults were collected and pooled in 50 μl RNA later solution (Merck) to obtain one sample. The samples were incubated at 4°C overnight and then stored at −30°C. The larvae were homogenized using Biomasher II (Nippi). Total RNA was isolated with an RNeasy mini kit (Qiagen), and cDNA was synthesized with a ReverTra Ace qPCR RT kit (Toyobo). Quantitative PCR analysis was performed with a 7500 Real-time PCR instrument (Applied Biosystems) using a TER94 Taqman probe and primers for exon boundary 2 to 3 (Dm01844813_g1), an Mcm3 Taqman probe and primers for exon boundary 1 to 2 (Dm01813006_g1), a TBPH Taqman probe and primers for exon boundary 4 to 5 (Dm01820179_g1), an Actin5C Taqman probe and primers (Sone et al., 2009), and Taqman Universal Master Mix II, no UNG (Thermo Fisher Scientific). To measure the expression of GFP, RNA was isolated from whole Actin5C-Gal4>GFP first-instar larvae, the brains of elavGal4>GFP, Cha-Gal4>GFP and +>GFP adults, the ventral nerve cords of OK6-GFP>GFP wandering third-instar larvae, and the compound eyes of GMR-Gal4>GFP adults. Quantitative PCR was performed on a 7300 Real-time PCR instrument (Applied Biosystems) using GFP primers (forward primer, 5′-TCTGTCTCCGGTCAAGGTGAAG-3′; reverse primer, 5′-GGCATGGCAGACTTGAAAAAG-3′) and a Ras64B Taqman Probe and primers for exon boundary 2 to 3 (Dm01822072_g1). Ras64B was used instead of Actin5C because expression of Actin5C was different between heads and compound eyes.

Viability and lifespan were measured as previously described (Furotani et al., 2018). All chromosomes were introduced into the same genetic background by consecutive backcrossing for more than three generations. The viability of female flies was measured by counting the number of eclosed flies on the day of eclosion and comparing it with the number of flies that had balancer chromosomes. For the measurement of lifespan, one to 20 female flies were reared at 25°C in a standard plastic vial.

To measure the volume of the brains and mushroom bodies, dissection and immunostaining of pupal and adult brains were performed essentially as previously described (Saito et al., 2021). Female pupae were collected at 72 h APF, and female adult flies were collected at 1 day after eclosion. The flies were rinsed in 70% ethanol for 30 s. Then, their brains were dissected in phosphate-buffered saline (PBS). The brains were fixed in 4% paraformaldehyde in PBS for 1 h on ice. The samples were incubated with mouse Mab 1D4 anti-Fasciclin II monoclonal antibody (Developmental Studies Hybridoma Bank, Iowa University, Iowa City, IA, USA; diluted 1:50) in PBS containing 0.4% Triton X-100 overnight at 4°C. The samples were then incubated with a Cy3-conjugated anti-mouse IgG antibody (Jackson ImmunoResearch; diluted 1:200), mounted with Vectashield with DAPI (Vector Laboratories, H-1200). A coverslip was mounted on the other coverslips as pillows. Images were collected with a confocal microscope (Olympus, FV-1000) equipped with a UPLSAPO 20× objective (0.75 NA) and a UPLSAPO 60× oil-immersion objective (1.35 NA). Cy3 signals and green autofluorescence signals were scanned separately. Signals from 2 μm consecutive optical sections were collected from the entire brain. Photographs were reconstructed, and the volumes of the brains and mushroom bodies were measured using Imaris software (Zeiss).

Images of the compound eye were collected with an MZ16 microscope (Leica) equipped with an IC80HD camera (Leica). The internal structures of the compound eyes were examined by preparing semithin sections as previously described (Sone et al., 2009; Arimoto et al., 2020). The heads of female flies were collected in the daytime, cut longitudinally into halves and fixed in fixative (2.5% glutaraldehyde and 2% formaldehyde buffered with 0.1 M sodium cacodylate to pH 7.3) for more than 2 h at room temperature. They were then rinsed with 4% sucrose solution in the same buffer for 5 min three times on ice, postfixed with 1% osmium tetroxide in the same buffer for 1.5 h on ice, and rinsed with distilled water for 5 min three times on ice. Then, the samples were dehydrated in 50% ethanol, 70% ethanol and 90% ethanol each for 5 min each on ice, in dry ethanol for 5 min three times, and in propylene oxide for 5 min twice. The samples were then incubated with a 1:1 mixture of propylene oxide and Epon (TAAB) overnight and embedded in Epon. Semithin sections (500 nm thick) were cut with a diamond knife using an ultramicrotome (Leica). The sections were stained with Toluidine Blue, and photographs were taken using an Axioskop2 microscope (Zeiss) equipped with a DP73 camera (Olympus). The number of rhabdomeres in each ommatidium was counted by examining these photos. For the measurement of the thickness of the retina, paraffin sections of the compound eye were prepared as previously described (Sone et al., 2009). The heads of female flies were collected, the probosces were removed, and the samples were fixed in 4% paraformaldehyde in PBS overnight. The samples were then dehydrated with 70% and 90% ethanol for 15 min; 35% n-butanol, 50% ethanol and 15% water for 15 min; 55% n-butanol, 40% ethanol and 5% water for 15 min; 75% n-butanol and 25% ethanol for 15 min; and then 100% n-butanol for 15 min. Subsequently, the samples were embedded in paraffin, and 6 μm frontal sections of the entire brain were prepared and stained with Hematoxylin and Eosin. Images were captured using an Axioskop2 microscope (Zeiss) equipped with a DP73 camera (Olympus), and the thickness of the retina was measured using CellSens Dimension software (Olympus).

Dissection and immunostaining of larval ventral nerve cords were performed essentially as previously described (Furotani et al., 2018). Female wandering third-instar larvae were collected, dissected and fixed for 20 min in 4% paraformaldehyde in PBS. The fixed larvae were subsequently incubated at 4°C overnight with an antibody against TBPH (Lin et al., 2011; diluted 1:200) in PBS containing 0.2% Tween 20 and 5% skim milk. The samples were then labeled with a secondary antibody conjugated to Cy3 (Jackson ImmunoResearch). The ventral nerve cords were subsequently dissected and mounted with Vectashield with DAPI (Vector Labs, H-1200). The samples were observed using a confocal microscope (Nikon, AX) with a 60× oil-immersion objective lens (1.42 NA). Single-plane images were obtained. The signal intensities in the nuclei were measured using CellSens Dimension software (Olympus).

Female adult flies at day 1 after eclosion were collected, and their heads were isolated using forceps. The anterior third of wandering third-instar larvae containing the brain and the ventral nerve cord were collected. Otherwise, first-instar larvae were collected. The samples were then frozen in liquid nitrogen. Thirty heads or larvae were collected in SDS-PAGE sample buffer (125 mM Tris-HCl, pH 6.8, 4% SDS, 2.5% 2-mercaptoethanol, 10% glycerol and 0.005% Bromophenol Blue) containing Complete Protease Inhibitor (Sigma-Aldrich) and pooled to obtain one sample. The heads were homogenized using Biomasher II (Nippi), boiled, and electrophoresed on 5-20% gradient SDS polyacrylamide gels. Protein lysates from approximately four heads were loaded in each lane. The gel was electroblotted onto an Immobilon P membrane (Millipore). The membrane was blocked in TBST (10 mM Tris-HCl, pH 8.0, 62.5 mM NaCl, 0.05% Tween 20) containing 5% skimmed milk for 1 h, and incubated with rabbit anti-VCP antibody (Cell Signaling Technology, 2648S; diluted 1:300), rabbit anti-TBPH antibody (a gift from Dr C.-K. James Shen, Academia Sinica, Taipei, Taiwan; diluted 1:1000) or mouse anti-α tubulin antibody (Sigma-Aldrich, T6199; diluted 1:4000). Then, the membrane was incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit IgG antibody (GE Healthcare, NA9345V; diluted 1:2500) or anti-mouse IgG antibody (GE Healthcare, NA931VD; diluted 1:5000) for 1 h. The signal was detected with an ECL prime system (Amersham) and LAS-4000 (GE Healthcare).

All data are shown as the mean±s.e.m. We performed statistical analysis using SPSS 27 software. When the ratios of eclosed fliles were compared, we used a chi-square test. When the averages of two groups were compared, we used a unpaired two-tailed Student's t-test. When the averages of more than two groups were compared, we used one-way ANOVA with post hoc tests. As a post hoc test, we used the Games-Howell test if the variance was significantly different among the genotypes. Otherwise, we used the Dunnett test.

We thank Dr Paul Taylor for the VCP transgenic fly strains; Dr C.-K. James Shen for the anti-TBPH antibody; the Bloomington Drosophila Stock Center, the Vienna Drosophila Resource Center, the NIG-FLY stock center and the FlyORF Zurich ORFeome Project for providing fly stocks; Tayoko Tajima for technical assistance; and Ayumi Echigo, Natsumi Tashiro, Nozomi Kubota, Satomi Ito, Takuya Tamura and Mikio Hoshino for helpful comments and discussion.