To develop an in vivo model of VCP-mediated degeneration, we used Drosophila melanogaster, in which the gene TER94 encodes the highly conserved orthologue of VCP. The Drosophila VCP orthologue, hereafter referred to as dVCP, is 92% similar and 83% identical to human VCP at the amino acid level. Conservation is even higher (95% similar, 84% identical) across the 250–amino acid N-terminus, which hosts most known disease-causing mutations, and all amino acid residues altered in disease are perfectly conserved. We introduced R152H and A229E mutations into dVCP to create homologs of the R155H and A232E mutations, which cause the most common and most severe forms of IBMPFD, respectively, and generated transgenic flies. When these mutants were expressed in the fly eye or brain using the UAS/GAL4 system (Brand and Perrimon, 1993) we observed mutation-dependent degenerative phenotypes despite equivalent levels of dVCP expression (Fig 1, Supplemental Fig 1). In the eye, expression of WT dVCP caused a very modest phenotypic change, whereas matched expression of the R152H and A229E mutants caused severe external rough eye phenotypes with necrotic patches and histologically evident marked vacuolar degeneration. The more severe phenotype seen in flies expressing the A229E mutant is consistent with the more severe disease in patients with the A232E mutation.

VCP participates in a range of cellular activities and thus in diverse biological pathways. The specific pathways underlying pathogenesis of IBMPFD, however, are unknown. As an unbiased approach to identify molecules and pathways involved in mutant VCP–mediated degeneration, we performed a dominant-modifier screen, using the moderate dVCP mutant phenotype of dVCP R152H to maximize the likelihood of visualizing enhancement or suppression of the mutant dVCP phenotype. To carry out the screen, we obtained a deficiency (Df) collection representing all four chromosomes from the Bloomington Drosophila Stock Center in order to scan the maximum proportion of genome (~80%) with the smallest number of lines. From the primary screen, seventy-four deficiencies were identified as dominant modifiers (enhancers or suppressors) of the dVCP R152H eye phenotype. After the secondary screen, validation studies and individual gene interrogation by double-strand RNAi lines, we identified three related genes: TBPH (CG10327), xl6 (CG10203) and Hrb27C (CG10377) that dominantly suppress the degenerative phenotype (Figure 2A-B). RNAi-mediated knockdown of these genes in the eye did not result in a phenotypic change independent of mutant dVCP expression (Supplementary Figure 2A-D). However, in flies expressing dVCP R152H, knockdown of these genes suppressed mutation-dependent degeneration (Figure 2B). Suppression of degeneration in VCP-mutant flies was corroborated by seeing a significant reduction in the blinded phenotypic severity score (Figure 2C) and by using additional RNAi lines and classical alleles (Supplementary Figure 2E-F). We also generated transgenic lines over-expressing TBPH, which resulted in a degenerative phenotype evident externally and histologically when targeted to the eye. When exogenous TBPH was co-expressed with dVCP R152H, degeneration associated with mutant VCP was enhanced, confirming the genetic interaction (Fig 2D). Hrb27C, xl6 and TBPH correspond to the human genes DAZAP1, 9G8 and TDP-43, respectively. These are all RNA recognition motif (RRM)-containing RNA-binding proteins that shuttle between the nucleus and cytoplasm (Huang and Steitz, 2001; Lin and Yen, 2006; Ayala et al., 2008). Furthermore, all three have been shown to regulate multiple aspects of RNA metabolism, including transcription, export, splicing and translation (Elvira et al., 2006b; Swartz et al., 2007; Yang et al., 2009). We focused on TDP-43 for further assessment, because cytoplasmic deposition of this protein is a prominent feature of IBMPFD and other degenerative diseases (Geser et al., 2009).

In the brain, TDP-43 is found predominantly in the nuclei of neurons and some glial cells (Fig. 3A), although dynamic studies in vitro have shown TDP-43 to shuttle between the nucleus and cytoplasm (Ayala et al., 2008). In IBMPFD, affected brain regions show gross abnormalities in TDP-43 localization, including clearance of TDP-43 from many nuclei and accumulation in the cytoplasm; some neurons show dense nuclear inclusions of TDP-43 (Fig. 3B). To examine the subcellular localization of TDP-43 in vitro, we first investigated whether VCP mutation-dependent toxicity could be recapitulated in primary mouse cortical neurons transfected with FLAG-conjugated WT VCP or mutant VCP (R155H and A232E). At 24 or 48 hours post-transfection, a blinded examiner scored VCP-positive neurons for changes in MAP2 staining and nuclear morphology (changes in DAPI staining) to assess cytotoxicity (Figure 3C). Quantitative analysis of the data revealed significant mutation-dependent toxicity in primary neurons 48 hours post-transfection (Figure 3D). This time-dependent cytotoxicity caused by mutant VCP was also demonstrated by fluorescence-activated cell sorting for living vs. dead HEK293T cells. Exogenous expression of mutant but not WT VCP caused cell death despite equivalent levels of VCP expression (Supplemental Fig. 3A-B).

To further investigate the potential role of TDP-43 mislocalization in mutant–VCP-associated degeneration in vivo, we generated multiple transgenic flies expressing human WT or mutant TDP-43, using the UAS/GAL4 system. To target TDP-43 specifically to the nucleus or the cytoplasm, we used TDP-43 with previously characterized mutations that disrupt the nuclear export sequence (NES) or the nuclear localization sequence (NLS)(Winton et al., 2008). Five independent transgenic lines expressing WT TDP-43 in the fly eye showed a modest degenerative phenotype (Figure 4 and Supplemental Fig 4A-B and 6A). A normal eye phenotype was observed in 5 independent transgenic lines expressing NES-mutant TDP-43 (Fig 4 and Supplemental Fig 4C-D and 6A). By contrast, expression of NLS-mutant TDP-43 in the fly eye in 6 independent lines resulted in a strong degenerative phenotype and high-molecular-weight smears of TDP-43 in immunoblots (Supplemental Fig. 5A-B and 6). Immunostaining of NLS-mutant TDP-43 confirmed predominantly cytoplasmic localization, while WT and NES-mutant TDP-43 remained predominantly nuclear (Fig. 4). To confirm that the degenerative phenotype was caused by cytoplasmic localization of TDP-43 rather than another effect of the NLS mutation, we also generated lines with truncated TDP-43 (80-414). Deletion of the first seventy-nine amino acids, which disrupts the NLS, led to cytoplasmic localization of TDP-43 and a degenerative phenotype indistinguishable from that observed with NLS-mutant TDP-43 (Supplemental Fig. 5C and data not shown). These findings indicate that accumulation of TDP-43 in the cytoplasm is sufficient to cause degeneration. Together with the observation that knockdown of TBPH in flies causes no phenotypic change alone but suppresses the mutant dVCP phenotype, these results indicate that a toxic gain of cytoplasmic TDP-43 function underlies pathogenesis. However, these observations do not preclude the possibility that a loss of TDP-43 nuclear function contributes to toxicity.

Missense mutations in TDP-43 have recently been identified in association with dominantly inherited and sporadic ALS. To assess the effect of mutant TDP-43 in vivo, we generated transgenic lines expressing TDP-43 with the ALS-causing mutation M337V (Sreedharan et al., 2008). In 6 independent lines evaluated, we observed degenerative phenotypes ranging from modest to severe (Fig 5A and Supplemental Fig 6A). The severity of the phenotype correlated with the amount of abnormal TDP-43 species, including a C-terminal 25-kDa fragment and high molecular weight bands (Fig 5 and Supplemental Fig 6B). This is similar to the pathogenic biochemical signature seen in patient tissue with TDP-43 proteinopathy (Neumann et al., 2007). Immunohistochemistry for TDP-43 in these eyes showed a marked cytoplasmic distribution of TDP-43, consistent with a recent report that this mutation increases cytoplasmic TDP-43 inclusions in vitro (Nonaka et al., 2009).

To assess the relationship between dVCP and TDP-43 in vivo, we performed an epistasis study by co-expressing dVCP (WT or mutant) with TDP-43 (WT or mutant) and analyzing changes to external eye phenotype by light microscopy (Fig 6A) and in subcellular localization of TDP-43 by immunohistochemistry (Fig 6B). Crosses with most of our NLS-mutant TDP-43 lines resulted in lethality; we therefore used line #5 that exhibits a very mild phenotype (all lines are shown in Supplemental Fig. 5A). Due to the severe degradative loss of internal eye tissue in some genotypes, immunohistochemistry of TDP-43 in the eye (Fig 6B) was corroborated with quantitative immunofluorescence analysis of TDP-43’s subcellular distribution in larval salivary glands (Fig 6C-D).

The mechanism by which mutations in VCP influence the subcellular distribution of TDP-43 is unknown, but we outline several possibilities here (Figure 7). First, the nuclear depletion and cytoplasmic accumulation of TDP-43 might reflect a defect in the well-established role of VCP in ubiquitin-dependent segregation of substrates from multiprotein complexes. It is known that TDP-43 is present in the cytoplasm at low levels normally, where it is found in ribonucleoprotein complexes implicated in translation regulation (Freibaum et al.; Elvira et al., 2006a; Wang et al., 2008; Moisse et al., 2009b). Perhaps VCP activity is necessary for removal of TDP-43 from ribonucleoprotein complexes to permit recycling and impairment of this activity by disease-causing mutations leads to progressive accumulation of TDP-43 in the cytoplasm. A second possibility is that VCP directly participates in nuclear import of TDP-43. VCP was previously shown to regulate the nuclear import of the TSAd protein (T cell-specific adaptor protein) in T cell signal transduction (Marti and King, 2005). This aspect of VCP function may involve the adaptor Npl4 (nuclear protein localization-4), originally discovered in a yeast screen for mutants deficient in nuclear protein import (DeHoratius and Silver, 1996) (Fabre and Hurt, 1997). Perhaps VCP regulates nucleocytoplasmic shuttling of additional proteins, including TDP-43, and that disease mutations impair this activity. A third possibility relates to the recently discovered role of VCP in autophagy, and the finding that disease-causing mutations in VCP impair autophagy (Ju and Weihl; Tresse et al.). Since autophagy may be important for turnover of cytoplasmic TDP-43 (Wang et al.), accumulation of TDP-43 in the cytoplasm may simply reflect a defect in this degradation pathway. These three possibilities are not mutually exclusive and it is also possible that redistribution of TDP-43 in IBMPFD reflects defects in VCP functions that are presently unknown.