Inactivation of Drosophila Huntingtin affects long-term adult functioning and the pathogenesis of a Huntington’s disease model

Huntington’s disease (HD) is an autosomal dominant, progressive neurodegenerative disorder characterized clinically by deteriorating choreic movements, psychiatric disturbances and cognitive deficits (Gusella and MacDonald, 1995; Martin and Gusella, 1986; Vonsattel et al., 1985). HD is caused by an abnormal expansion of a polyglutamine (polyQ) tract at the N-terminus of a large cytoplasmic protein, huntingtin (Htt) (The Huntington’s Disease Collaborative Research Group, 1993). The polyQ tract contains between 6 and 35 repeats in the wild-type Htt protein, whereas it is expanded to beyond 36 repeats in HD (The Huntington’s Disease Collaborative Research Group, 1993). Numerous studies have demonstrated that mutant Htt containing an expanded polyQ tract is toxic to neurons (Cattaneo et al., 2001; Gusella and MacDonald, 2000). PolyQ expansion is also linked to at least eight other neurodegenerative disorders, collectively referred to as polyQ diseases (Riley and Orr, 2006; Zoghbi and Orr, 2000). Although Htt is ubiquitously expressed in the brain, HD mainly affects medium-sized spiny neurons in the striatum and to a lesser extent cortical pyramidal neurons that project to the striatum, suggesting that other cellular factors also contribute to pathogenesis (Cattaneo et al., 2001; Vonsattel and DiFiglia, 1998). Recent studies indicate that an alteration of wild-type Htt function might contribute to this specificity and to subsequent disease progression (Cattaneo et al., 2001). For example, mutant Htt can sequester wild-type Htt into insoluble aggregates, thereby exerting a dominant negative effect (Huang et al., 1998; Kazantsev et al., 1999; Narain et al., 1999; Preisinger et al., 1999; Wheeler et al., 2000). In addition, wild-type Htt can suppress the cell death induced by mutant polyQ-expanded Htt in vitro (Leavitt et al., 2001; Van Raamsdonk et al., 2005). Furthermore, wild-type Htt is proposed to have a neuroprotective role as expression of Htt can protect cultured striatal neurons from stress- and toxin-mediated cell death (Rigamonti et al., 2000).

Since its identification, the normal function of Htt has been subject to extensive investigation (Cattaneo et al., 2001; Harjes and Wanker, 2003). The murine Htt homolog (also known as Hdh) is essential during early mouse development, as Htt-null mice die during gastrulation at embryonic day 7.5 (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995). Chimeric analysis demonstrated that the early embryonic lethality is the result of a crucial role of Hdh in extraembryonic membranes, as this lethality can be rescued by providing wild-type Hdh function in extraembryonic tissue (Dragatsis et al., 1998). Conditional knockout of Hdh in the mouse forebrain at postnatal or late embryonic stages causes a progressive neurodegenerative phenotype, lending support to the hypothesis that depletion of normal Htt activity during disease progression contributes to HD pathogenesis (Dragatsis et al., 2000). A more recent study in zebrafish, which used morpholino oligos to transiently knockdown endogenous Htt, suggests that Htt has a role in normal blood function and iron utilization (Lumsden et al., 2007). Currently, little is known about the normal biological function of wild-type Htt (Cattaneo et al., 2005).

Htt encodes a large cytoplasmic protein of 350 kDa. Structural analysis of Htt proteins identified the presence of many HEAT (huntingtin, elongation factor 3, the A subunit of protein phosphatase 2A and TOR1) repeats, which are approximately 40-amino acid (a.a.) long structural motifs, composed of two anti-parallel helices, of unknown function (Andrade and Bork, 1995). No other domains have been identified in Htt to suggest a biological function for the protein. Functional studies in mammalian systems, mainly from protein interaction assays, have associated Htt with diverse cellular processes including: endocytosis; modulation of synapse structure and synaptic transmission; transcriptional regulation, especially of the brain-derived neurotrophic factor (BDNF) which is essential for the survival of the striatal neurons affected in HD; axonal transport of BDNF and vesicles; and apoptosis (Cattaneo et al., 2001; Cattaneo et al., 2005; Harjes and Wanker, 2003; Zuccato et al., 2001; Zuccato and Cattaneo, 2007). Importantly, only a few of these proposed functions of Htt have been directly tested in vivo owing to the early embryonic lethality associated with Htt-null mutant mice.

In an extensive search for Htt homologs in other species, Li et al. (Li et al., 1999) identified a single HTT homolog in Drosophila (htt, hereafter referred to as dhtt). By sequence comparison, the homologous regions between Drosophila and human Htt proteins are mainly located within five discrete areas, including three relatively large continuous regions and two small segments, that cover about one-third of the total protein length (supplementary material Fig. S1A). At the amino acid level, these homologous regions, which are comprised of about 1200 a.a. residues in Drosophila Htt (dHtt), share around 24% identity and 49% similarity (Li et al., 1999). In addition to the sequence similarity, other shared features support the proposal that the Drosophila gene identified in the study by Li et al. is indeed the fly homolog of human Htt (Li et al., 1999). For example, in terms of the protein size, both the Drosophila and human Htt proteins are unusually large and contain 3583 and 3144 a.a. residues, respectively (Li et al., 1999). In addition, the regions with a relatively high level of conservation are not only clustered in large continuous stretches, but are also located in the same order and distributed over the entire length of the proteins. Moreover, dhtt and mammalian HD genes share similar patterns of gene expression (Li et al., 1999) (Fig. 1). Interestingly, although an HTT homolog exists in Drosophila, no HTT-like gene has been found in other less complex eukaryote species such as C. elegans or the yeast S. cerevisiae (Li et al., 1999).

The identification of a Drosophila Htt homolog provides a unique opportunity to evaluate the role of Htt in this well-established genetic model system. Several cellular processes implicated in Htt function, including axonal transport and synapse formation, have been well-characterized in Drosophila, allowing an in vivo evaluation of their relationship with Htt. Further, as fly models of HD have been well-established, this model allows an in vivo examination of the function of endogenous Htt in HD pathogenesis (Marsh and Thompson, 2006; Steffan et al., 2001). In this study, we report the isolation of a dhtt mutant and describe its phenotype. Further, we examine how the removal of endogenous dhtt affects several cellular processes that have previously been implicated with Htt, and test how the loss of endogenous dhtt affects the pathogenesis associated with an established Drosophila model of polyQ toxicity (HD-Q93).

Considering the limited sequence homology between mammalian and fly Htt, it is important to examine the extent of the structural similarity between these proteins. In the Htt family proteins, the HEAT repeat is the only identifiable structural motif (Andrade and Bork, 1995; Cattaneo et al., 2005). A previous phylogenetic study identified 16 HEAT repeats in human Htt and, notably, 14 of these 16 repeats were also found in insect Htt proteins including dHtt (Tartari et al., 2008). A less stringent structural analysis predicted up to 40 HEAT repeats (including the AAA, ADB and IMB subgroups) in human Htt (see Methods). Interestingly, using the same parameter, 38 HEAT repeats could be identified in dHtt (see supplementary material Fig. S1 for details of the predicted HEAT repeats). Further, these HEAT repeats span the entire length of each protein and have a similar distribution, clustering in four groups at the N-, middle- and C-terminal regions, which have a large overlap with their segments of homologous sequences (supplementary material Fig. S1B). Although further studies are needed to elucidate the structure of the Htt proteins, this rather remarkable similarity raises the possibility of a conserved secondary structure among Htt family proteins and that both human and Drosophila Htt proteins are composed largely of repeated HEAT motifs.

Previous analysis has shown that dhtt is widely expressed during all developmental stages from embryos to adults (Li et al., 1999). We confirmed the expression of the dhtt transcript in adults using reverse transcription (RT)-PCR (data not shown) (Fig. 5C). To examine the tissue-specific dhtt expression, we performed whole-mount RNA in situ hybridization. Staining with two DIG-labeled antisense probes, targeting different regions of dhtt, revealed similar ubiquitous dhtt expression at different stages of fly embryogenesis and in larval tissues (Fig. 1A–F and data not shown), whereas a positive control performed in parallel gave rise to robust in situ signals (supplementary material Fig. S2). Importantly, of the two negative controls included in the assay, one using sense dhtt RNA probes on wild-type samples (Fig. 1B) and the other using the same set of antisense dhtt RNA probes against tissues from a dhtt deletion mutant that had been generated subsequently (Fig. 1F), both produced much weaker background signals. Together, these data indicate that dhtt is widely expressed at low levels during all stages of Drosophila development.

Mammalian Htt proteins are largely cytoplasmic with a widespread expression pattern (DiFiglia et al., 1995; Gutekunst et al., 1995; Sharp et al., 1995). To determine the expression and subcellular localization of the endogenous dHtt protein, we developed an affinity-purified polyclonal antibody against dHtt. The specificity of this antibody was confirmed by its ability to recognize ectopically expressed dHtt protein in transfected Drosophila S2 cells (Fig. 1G,H) and larval tissues (Fig. 1I–K). When the dHtt protein was ectopically expressed from a UAS-dhtt transgene by using a patched-Gal4 driver, our anti-dHtt antibody could easily detect the striped pattern of dHtt expression in the middle of imaginal discs, which is the characteristic domain of Patched expression (Fig. 1I). In wild-type animals, use of the anti-dHtt antibody resulted in low-level ubiquitous staining in embryos, larval and adult tissues, with no specific pattern of protein expression (data not shown). At the subcellular level, ectopically expressed dHtt was found predominantly in the cytoplasm in transfected S2 cells and in larval tissues (Fig. 1G–K and data not shown), suggesting that dHtt, similar to its human counterpart, is mainly a cytoplasmic protein.

Similar to human Htt, dhtt encodes an unusually large protein of 3583 a.a. residues. The cDNA for the dhtt gene is 11,579 base pairs (bp) long and is derived from 29 exons in a 38 kb transcribed genomic region at cytological interval 98E2 (Li et al., 1999) (Fig. 2A,C). No null mutations in dhtt have been isolated previously. To generate a null mutant for dHtt, we selected two FRT-bearing insertion lines surrounding dhtt: p-element d08071, which is inserted at the 5′ end of the neighboring CG9990 gene, and piggyBac insertion f05417, which is located inside the intron between dhtt exon 27 and exon 28 near the 3′ end of the gene (Fig. 2A). Using flipase (FLP)–FRT-mediated recombination (Parks et al., 2004), we generated a precise deletion of 55 kb between the two insertions (see Methods). This deletion allele, termed Df(98E2), removed most of the CG9990 gene and 34 kb of the 38 kb genomic-coding region for dhtt, with only the last two exons of the dhtt gene remaining (Fig. 2A–C). The deletion was confirmed by inverse PCR from extracted genomic DNA and by DNA sequencing (data not shown).

Drosophila containing the Df(98E2) deletion, which removes both CG9990 and dhtt, are homozygous lethal at the embryonic stage. CG9990 encodes a previously uncharacterized protein belonging to the ABC transporter superfamily (Dean et al., 2001). To separate the mutant phenotype of dhtt from CG9990, we generated transgenic flies carrying a CG9990 genomic rescue transgene in the Df(98E2) background (see Methods). These lines, referred to as dhtt-ko (dhtt-knockout only), carry both the CG9990 rescue transgene and the Df(98E2) deletion, and thus are mutant only for dhtt (Fig. 2B–D).

The dhtt-ko allele removes 27 of the 29 exons of dhtt (Fig. 2). Since Htt homozygous knockout mice die during early embryogenesis (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995), we expected that loss of dhtt in the fly would be associated with prominent developmental defects. However, dhtt-ko flies are homozygous viable, demonstrating that the lethality observed following the Df(98E2) deletion is caused by loss of CG9990. To verify that the dhtt gene is indeed deleted in dhtt-ko flies, we extracted genomic DNA from homozygous dhtt-ko adults and performed a Southern blot analysis. As shown in Fig. 2D, all genomic DNA containing the dhtt gene was removed in dhtt-ko flies except for the final two 3′ exons.

Further analyses demonstrated that homozygous dhtt-ko flies develop at a similar rate to wild-type flies and give rise to fertile adults with no discernible morphological abnormalities. The progeny derived from homozygous dhtt-ko flies did not show a reduction in viability or display other obvious developmental defects, suggesting that maternally contributed dhtt has no significant effect on animal development or function.

To determine whether developmental defects are present in dhtt-ko animals, we characterized dhtt-ko mutants using a variety of cellular and neuronal markers. These studies failed to reveal any obvious developmental abnormalities during embryogenesis, or during larval and adult stages (Fig. 3 and data not shown). In particular, the embryonic central nervous system (CNS) (Fig. 3A) and muscles, and the larval muscles, CNS, eye and other imaginal discs all appeared normal (Fig. 3B and data not shown). Further, in aged adults (40-day-old flies), the external eye morphology was normal and the eight neuronal photoreceptor cells in each ommatidium were clearly present, together with their accessory cells (Fig. 3C,D). We note that our finding is different from a recent study that examined dhtt function using RNA interference (RNAi), which implicated a role for dhtt in axonal transport and eye integrity (Gunawardena et al., 2003). The exact nature of this phenotype discrepancy is not clear. Given the experiment was carried out at 29°C, it is possible that the observed RNAi phenotypes were the result of cellular toxicity caused by the high-level expression of Gal4 (Gunawardena et al., 2003). The more severe phenotypes might also be the result of the non-specific RNAi off-target effects caused by the knockdown of unrelated genes (Kulkarni et al., 2006). Nevertheless, our results suggest that dhtt is dispensable for normal Drosophila development. This result is in contrast to the essential role of Htt in mouse (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995). These phenotypic differences are probably the result of differences in mouse and fly embryogenesis, as the early lethality of Htt-null mice is due to the function of Hdh in extraembryonic membranes, for which there are no equivalents in Drosophila (Dragatsis et al., 1998) (see Discussion).

Htt has been reported to interact with a diverse group of proteins whose functions have been directly or indirectly linked to synapse organization and synaptic activity (Harjes and Wanker, 2003), including proteins that regulate cytoskeleton dynamics and clathrin-mediated endocytosis [e.g. HIP1 (Sla2p), HIP12, PACSIN/syndapin 1 and endophilin 3] (Chopra et al., 2000; Higgins and McMahon, 2002; Kalchman et al., 1997; Modregger et al., 2002; Seki et al., 1998; Singaraja et al., 2002; Sittler et al., 1998; Wanker et al., 1997); axonal vesicle transport (e.g. HAP1) (Engelender et al., 1997; Gunawardena et al., 2003); and dendritic morphogenesis and synaptic plasticity (e.g. the postsynaptic density protein DLG4/PSD95 and the adaptor proteins GRB2 and TRIP10/CIP4) (Holbert et al., 2003; Liu et al., 1997; Sun et al., 2001). To test whether dhtt plays a role in synapse organization, we examined the formation of glutamatergic neuromuscular junctions (NMJs) in third instar larvae, a well-characterized system for studying synapse formation and function in Drosophila (Budnik and Gramates, 1999). Examination with a panel of synapse markers showed that axonal pathfinding, muscle innervation and overall synapse structure are normal in dhtt mutants (Fig. 3 and data not shown). Double-labeling with the axonal membrane marker anti-HRP and the postsynaptic density marker anti-Dlg (the Drosophila PSD95 homolog) revealed well-organized presynaptic and postsynaptic structures (Fig. 3E1–E8), as well as an enrichment of synaptic vesicles at the synapses, with no obvious synapse retraction phenotype (data not shown). Moreover, dhtt mutants showed normal organization of the presynaptic microtubule (MT) cytoskeleton, with the presence of stable MT bundles traversing the center of NMJ branches and a dynamically reorganized MT network at distal boutons (data not shown). Quantification of NMJ bouton number and axonal branching did not reveal a significant difference between wild-type controls and dhtt mutants (Fig. 3G–I). Finally, dhtt mutants displayed the stereotypical complementary pattern of active zones surrounded by the honeycomb-like organization of periactive zones, and examination of multiple synaptic proteins such as nc82 (an active zone marker) and fasciclin II (FasII, a periactive zones marker) showed normal synaptic localization (Fig. 3F1–F8). As a control, we also examined synapse organization in CG9990 transgenic animals, which displayed similar well-organized NMJ structures (supplementary material Fig. S3 and data not shown). Together, these results suggest that dhtt is not essential for synapse formation and organization at NMJs.

Htt has also been proposed to regulate axonal vesicle transport because one of its binding partners, HAP1, interacts directly with p150Glued, which is an essential subunit of the dynactin complex involved in regulating dynein-mediated retrograde axonal transport (Engelender et al., 1997; Gunawardena et al., 2003; Harjes and Wanker, 2003). In Drosophila, individuals that are defective for essential components of the axonal transport machinery often display characteristic mobility phenotypes, such as tail flipping during larva crawling, and progressive lethargy and paralysis (Gindhart et al., 1998; Gunawardena and Goldstein, 2001; Martin et al., 1999). Such mutant animals also develop an axonal swelling phenotype, owing to the abnormal accumulation of synaptic vesicles along axons, and fail to properly deliver and localize synaptic components to the termini (Gindhart et al., 1998; Gunawardena and Goldstein, 2001; Martin et al., 1999). dhtt mutants showed normal crawling behavior during larval stages. Furthermore, the distributions of the synaptic vesicle markers anti-synaptotagmin (Syt) and anti-cysteine string protein (CSP) were normal, revealing no obvious accumulation or axonal swellings (Fig. 4 and data not shown). Further, immunostaining with other synaptic components, including synapsin (a reserved vesicle pool marker), FasII (a periactive zone marker) and Dlg, demonstrated that synaptic components were properly delivered to synapses (Fig. 3E–F and data not shown). Similarly, examination of CG9990 transgenic animals failed to reveal discernible larval mobility or axonal transport defects (supplementary material Fig. S3). Together, these results suggest that dhtt does not play an essential role in axonal transport.

We next investigated whether dhtt might function in adult animals. We examined whether newly emerged dhtt adults were hypersensitive to stress tests including prolonged heat and cold exposure, vortexing and feeding with the oxidative stress compound paraquat. In these tests, dhtt mutants showed similar responses to wild-type controls (data not shown). Thus, unlike flies that are mutated for Parkinson’s disease genes such as parkin and pink1, which are sensitive to multiple stress challenges (Clark et al., 2006; Greene et al., 2003; Park et al., 2006), loss of dhtt does not render young adult animals more vulnerable to environmental stresses.

Next, we followed the activity and viability of dhtt animals throughout the adult life cycle. Although no discernible difference in the activity and survival rate was observed between dhtt-ko and wild-type young adults, we observed striking defects in older adult dhtt-ko flies. dhtt-ko animals showed similar spontaneous locomotion to that of wild-type controls at day 15 and earlier (Fig. 5A; supplementary material Movie 1). However, as flies aged, dhtt-ko mutants displayed a rapidly declining mobility, which was evident by day 25. By day 40, almost all dhtt mutants showed severely compromised mobility (Fig. 5A; supplementary material Movie 2) and their viability declined quickly. Whereas, on average, half of the wild-type controls died around day 59 but could live for up to 90 days, half of the dhtt-ko flies died around day 43 and almost all by day 50 (Fig. 5B).

To verify that the late-onset mobility and viability defects were because of loss of dhtt, we constructed a dhtt minigene that expressed full-length dhtt under the control of its endogenous regulatory region (see Methods). In the presence of this mini-dhtt transgene, the expression of the dhtt gene was restored in dhtt-ko mutants, as confirmed by RT-PCR (Fig. 5C) Both the late-onset mobility and viability phenotypes observed in dhtt-ko mutants were rescued by the dhtt transgene (Fig. 5A,B; supplementary material Movie 3). Importantly, introduction of an unrelated transgene construct, such as elav-Gal4 (Fig. 5B) or UAS-eGFP (data not shown), into the dhtt-ko background could not rescue the mobility and viability phenotypes of dhtt-ko mutants, confirming that the rescue was because of the restored expression of dhtt. Thus, although dhtt is not essential for normal Drosophila development, its function is important in maintaining the long-term mobility and survival of adult animals.

The observed mobility and viability phenotype could be because of an underlying neurotransmission defect in dhtt mutants as several proposed Htt functions, such as axonal vesicle transport and clathrin-mediated endocytosis in neurons, are essential for the delivery and recycling of synaptic vesicles at nerve terminals to ensure effective neurotransmission (Cattaneo et al., 2001; Eaton et al., 2002; Harjes and Wanker, 2003; Hinshaw, 2000; Slepnev and De Camilli, 2000). To test the neuronal communication in dhtt mutants, we measured synaptic physiology at the well-characterized third instar larval NMJ. We quantified the amplitude of evoked excitatory junctional potentials (EJPs), resting membrane potential and paired-pulse facilitation (PPF) (Fig. 6A–D and data not shown). In adult animals, we recorded electroretinogram (ERG) responses in the eye and quantified DLM (dorsal longitudinal flight muscle) bursting activity in the giant fiber flight circuit (Fig. 6E and data not shown). Following exhaustive analysis, we found no significant defect in either synaptic transmission or short-term plasticity in dhtt mutants. Together, these data suggest that dhtt is not essential for neurotransmission.

To investigate whether the observed mobility and viability phenotypes in aged dhtt mutants are correlated with neuronal communication defects, we screened for electrophysiological phenotypes in animals aged 40–45 days, when behavioral motor abnormalities are prominent. Aged dhtt mutant adults did not display abnormal seizure activity in extracellular recordings from DLM flight muscles in the giant fiber escape pathway, as has been observed in temperature-sensitive Drosophila mutants that have altered synaptic transmission (Guan et al., 2005). Aged mutants also showed normal visual transduction and synaptic transmission in the visual system at room temperature (Fig. 6E and data not shown), consistent with the observation that dhtt does not play a crucial role in neurotransmission.

Previous studies have shown that mutations affecting synaptic function often manifest a more prominent phenotype under temperature-induced stress (Atkinson et al., 1991; Coyle et al., 2004). To assess whether phototransduction can be maintained under an elevated temperature in dhtt mutants, we performed ERG recordings at 37°C. Although dhtt mutants aged 1–3 days displayed normal ERGs at both 20°C and 37°C, aged dhtt mutants displayed abnormal sensitivity to 37°C, with over 60% of aged mutants losing light-induced phototransduction and the on/off transients at 37°C compared with only 20% of control aged adults (Fig. 6E,F), indicating an important role of dhtt in maintaining phototransduction under temperature-induced stress. Given that aged dhtt mutant adults did not show additional motor defects when placed at 37°C, we hypothesize that the loss of photoreceptor depolarization found in ERG recordings reflects temperature-sensitive defects in the phototransduction cycle, rather than in synaptic transmission. These results suggest that aged dhtt mutants are more sensitive to stress than young dhtt animals or aged controls.

To further test for dhtt function in the adult brain, we examined dhtt brain morphology using a series of cellular markers. The overall patterning and gross morphology of dhtt-ko brains appeared normal, as shown by staining with antibodies such as the glial cell marker anti-Repo, the pan-neuronal marker anti-Elav and the synaptic vesicle marker anti-CSP (data not shown). In addition, within the ventral nerve cord, the neuropile was similarly enriched with synaptic vesicles. Further, the overall axonal morphology of dhtt-ko mutants appeared normal with no clear axonal blebbing or defasciculation phenotypes (data not shown). Interestingly, examination of the mushroom bodies (MBs), which are involved in learning and memory in flies, by anti-FasII staining revealed that the signal intensity was weaker in dhtt mutants than in wild-type controls, despite appearing morphologically normal (Fig. 7J,M). Quantification of MB size and FasII staining signals revealed that, although there was a slight reduction in the average size of MBs in dhtt-ko mutants compared with wild-type controls (Fig. 7P) (average area covered by each MB: wild type=13,352±156 μm², dhtt-ko mutant=11,927±310 μm²; P=0.0003), the average signal intensity of FasII signals in the MBs of dhtt-ko mutants was decreased by ~50% (Fig. 7Q) (relative signal intensity of MBs: wild type=100±5.8/μm², dhtt-ko mutant=47.7±6.1/μm²; P<0.0001; total number of MBs quantified: wild-type control, n=13; dhtt-ko mutants, n=10).

To examine the effect of dhtt loss-of-function on the detailed structure of individual neurons in the brain, we used the A307-Gal4 line, which labels the pair of giant fiber (GF) neurons and a small number of other neurons of unknown identity in the adult brain, to examine axonal projection patterns and the fine axonal terminal structure of individual neurons (Phelan et al., 1996) (Fig. 7). Among the A307-Gal4 labeled neurons, one pair, located at the dorsal-lateral edge of the brain, projects a prominent axon tract along the dorsal-posterior surface to the dorsal-central region of the brain, forming extensive dendritic connections (Fig. 7B–D; supplementary material Fig. S4). These neurons further extend their projections anteriorly, establishing a complex axon terminal structure with extensive varicosities and fine branches above the antennal lobe region of the brain (Fig. 7E,F,K; supplementary material Fig. S4). In dhtt-ko mutants, the axonal projections of A307-positive neurons follow the same path and their axons terminate at similar locations to those in wild-type flies; this is consistent with the observation that dhtt does not affect axonal integrity or pathfinding (Fig. 7G,H). Interestingly, axonal termini from both wild-type and dhtt-ko flies show a similar age-dependent maturation process, whereas axonal termini in young adult brains are mainly composed of a network of variable thin branches with no clearly recognizable synaptic boutons (supplementary material Fig. S4C,F) (3-day-old flies). Mature boutons develop as the animals age, with many prominent boutons being easily identifiable in the brains of 40-day-old flies (Fig. 7F,I; compare with supplementary material Fig. S4C,F). Owing to the significant variation in their structure and the lack of recognizable boutons, it is difficult to directly quantify and compare the size of these axonal termini in young adults. However, in aged dhtt-ko mutants, it is apparent that the axonal termini contain a significantly reduced number of varicosities and branches (Fig. 7H,I,N). Quantification of the total area covered by each axonal terminus revealed that the A307-positive axonal termini in 40-day-old dhtt-ko mutants cover about half of the area compared with controls (Fig. 7R) (average area covered by each axonal terminus: wild-type control=168.7±8.0 μm², dhtt-ko mutant=86.1±7.2 μm²; P<0.0001; total number of A307-positive axonal termini quantified: wild-type control, n=18; dhtt-ko mutants, n=17). To rule out the possibility that this reduced complexity was because of the accelerated aging process or a secondary effect associated with the reduced mobility of dhtt-ko mutants, we examined the axonal termini of the A307-positive neurons in 83-day-old wild-type flies, as animals at this age are near the end of their life span and have severely reduced mobility. The structure of the axonal termini in these flies is similar to that of 40-day-old flies and shows no obvious reduction in terminal complexity (supplementary material Fig. S5).

Using the membrane-bound mCD8-eGFP reporter driven by the GF-specific A307-Gal4 line, we further analyzed the axonal projection and terminal morphology of the GF neurons in dhtt-ko mutants. The GF neurons are a pair of large interneurons located in the central brain and project their prominent axons, which are unbranched, to the mesothoracic neuromere (T2) in the ventral nerve cord, where they bend laterally and synapse with other interneurons and motor neurons (Phelan et al., 1996). In both young (3-day-old) and aged (40-day-old) dhtt-ko mutants, GF neurons project normally to the T2 neuromere and form the characteristic terminal bends, resembling those observed in wild-type controls (supplementary material Fig. S6). In 3-day-old animals, the signal intensity of the mCD8-eGFP reporter at the GF axonal termini was similar between dhtt-ko mutants and wild-type controls (supplementary material Fig. S6B,C) (total number of 3-day-old GF axonal termini examined: wild-type control, n=10; dhtt-ko mutants, n=12). However, in 40-day-old animals, the GF axonal termini in wild-type controls showed a much stronger enrichment for the mCD8-eGFP reporter than in dhtt mutants (supplementary material Fig. S6D–G) (total number of 40-day-old GF axonal termini examined: wild-type control, n=16; dhtt-ko mutants, n=20). The exact nature behind such a difference remains to be clarified, but might represent subtle alterations in axonal transport of membrane proteins in aged neurons. Nevertheless, these results suggest that dhtt does not affect axonal pathfinding and overall brain organization, but has a functional role in regulating the complexity of axonal termini in the adult brain.

Earlier studies in cell culture and Htt mutant mice have shown that wild-type Htt has a protective role for CNS neurons (Dragatsis et al., 2000; O’Kusky et al., 1999; Rigamonti et al., 2000; Van Raamsdonk et al., 2005). Further, existing evidence suggests that normal Htt activity can be inactivated by mechanisms such as abnormal sequestration into insoluble aggregates (Huang et al., 1998; Kazantsev et al., 1999; Narain et al., 1999; Preisinger et al., 1999; Wheeler et al., 2000). These and other observations have lead to the hypothesis that the perturbation of endogenous Htt function, such as by late-onset inactivation of endogenous Htt, contributes to HD pathogenesis (Cattaneo et al., 2001). Attempts to test this directly in mouse Hdh mutants have been complicated by the crucial role of Hdh in early mouse embryogenesis and in the later stages of brain development (Auerbach et al., 2001; Dragatsis et al., 2000; Duyao et al., 1995; Leavitt et al., 2001; Nasir et al., 1995; Van Raamsdonk et al., 2005; White et al., 1997; Zeitlin et al., 1995). However, we have a unique opportunity to examine this in Drosophila because dhtt is not essential for fly embryogenesis and dhtt mutants appear to behave normally at young ages, with only a mild axon terminal defect in the brain. We used a well-established fly HD model for polyQ toxicity (HD-Q93), in which the human HTT exon 1, with 93 glutamine repeats, is expressed in all neuronal tissues (genotype: elav-Gal4/+; UAS-Httexon1-Q93/+) (Steffan et al., 2001). The HD-Q93 flies develop age-dependent neurodegenerative phenotypes in adults, which manifest as initial hyperactivity followed by a gradual loss of coordination and a decline in locomotor ability, with eventual death at around 20 days of age (Fig. 8F–H). The HD-Q93 flies also develop a progressive degeneration of both the brain and other neuronal tissues, most prominently in the photoreceptor cells in the eye (Steffan et al., 2001).

After introducing HD-Q93 into the dhtt mutant background (‘HD-Q93; dhtt-ko’ flies; genotype: elav-Gal4/+; UAS-Httexon1-Q93/+; dhtt-ko/dhtt-ko), we examined the eye degeneration phenotype by quantifying the number of rhabdomeres in each ommatidia as the animals age. The loss of photoreceptor cells in HD-Q93 flies was not significantly enhanced in the absence of endogenous dhtt (Fig. 8A–E). For example, at 11 days of age, approximately 40% of ommatidia lost three photoreceptors in both HD-Q93 and HD-Q93; dhtt-ko flies (Fig. 8E) (number of ommatidia with four photoreceptors: HD-Q93=42.6±2.9%, n=562, eight adult eyes analyzed; HD-Q93; dhtt-ko=37.6±4.1%, n=266, seven adult eyes analyzed; the difference was statistically insignificant, P>0.5). Further, the overall profile of the remaining photoreceptors per a declining mobility that could be detected as early as 5 days of age ommatidia was also similar between these flies (Fig. 8E).

Interestingly, although these HD-Q93; dhtt-ko flies showed normal mobility at the beginning of their adult life, they displayed a declining mobility that could be detected as early as 5 days of age and that rapidly deteriorated over the next few days (supplementary material Movies 4–8). During this time, the flies become progressively uncoordinated, showing an increasing frequency of both faltering while walking and falling during climbing (Fig. 8G; supplementary material Movies 4–8). In a standard climbing assay, almost all 3-day-old HD-Q93; dhtt-ko flies could successfully climb to the top of a vial, which is similar to that observed for wild-type controls. However, by day 11, only around 10% of viable HD-Q93; dhtt-ko flies could make it to the top, whereas the success rate was 61% for HD-Q93 flies, and more than 94% for wild-type and dhtt-ko flies (Fig. 8F) [based on an average of at least five independent assays of 20 flies for each genotype at the given age; compared with wild-type controls on day 11, the results from dhtt-ko flies were not significant (P=0.30), whereas the results from HD-Q93 and HD-Q93; dhtt-ko flies were significant (P<0.0001)]. Similarly, when spontaneous locomotion was tested, the HD-Q93; dhtt-ko flies showed a more rapid decline in motility over time, and by day 9, these flies were less than half as active as those in the other groups (Fig. 8G) (activity was measured by the number of spontaneous turns performed every 4 minutes; on day 9, HD-Q93; dhtt-ko=26.4±5.9, n=9; HD-Q93=66.5±5.0, n=8; elav-Gal4; dhtt-ko control=60.6±4.3, n=9; UAS-Httex1Q93; dhtt-ko control=66.2±3.0, n=9; the HD-Q93; dhtt-ko results were statistically significant when compared with UAS-Httex1Q93; dhtt-ko control flies, P=0.00002). Furthermore, the life span of HD-Q93; dhtt-ko flies was shortened significantly, with half of them dying by day 8 and almost all of them by day 14. In contrast, only about 7% of the HD-Q93 flies died at day 8 and half of them by day 14 (Fig. 8H) [viability at day 8: HD-Q93; dhtt-ko=45.0±3.2%, n=409 (from four different crosses); HD-Q93=93.1±1.7%, n=1082 (from eight different crosses); the difference between HD-Q93 and HD-Q93; dhtt-ko flies was statistically significant, P=0.0003]. Notably, at day 15, dhtt-ko mutants were healthy and displayed similar mobility to wild-type controls (Fig. 5A,B).

To understand the underlying pathology of these phenotypes, we further examined the brain structure of these flies. Compared with HD-Q93 flies of the same age, the brains of HD-Q93; dhtt-ko flies had already developed a more severe pathology at 5 days. Notably, the MBs appeared less organized – the characteristic bulged tip of the vertical α-axonal lobes, which were prominent in HD-Q93 flies (Fig. 9C, white arrows) (n=14) and other controls, was largely unrecognizable in 95% of HD-Q93; dhtt-ko flies (n=17/18) (Fig. 9H, white arrows) (supplementary material Figs S7 and S8). The clear separation, along the midline, between the pair of medially projected β-lobes, which was obvious in HD-Q93 flies (Fig. 9C, white arrow) (supplementary material Figs S7 and S8) (n=14) and wild-type flies (Fig. 7J), also became less distinct and often appeared to be merged in HD-Q93; dhtt-ko flies (n=13/18) (Fig. 9H, white arrow) (supplementary material Figs S7 and S8). In addition, the anti-FasII staining of the MBs was not as strong in HD-Q93; dhtt-ko flies as in HD-Q93 flies (Fig. 9C,H; supplementary material Figs S7 and S8). Quantification of MB size and FasII staining signals revealed that there was approximately a 7% reduction in the average size of MBs in HD-Q93; dhtt-ko mutants compared with HD-Q93 controls (Fig. 9K). The γ-lobe signals were too weak to be reliably tracked in both HD-Q93 and HD-Q93; dhtt-ko flies, so only the α- and β-lobes in each MB were measured (average MB size: HD-Q93=11,137±314 μm², n=6; HD-Q93; dhtt-ko=10,412±212 μm², n=7; P=0.04). The average signal intensity of FasII signals in the MBs of dhtt-ko mutants was also decreased, by around 31% (Fig. 9L) (relative signal intensity of α- and β-lobes in the MBs: HD-Q93=100±6.7 μm²; dhtt-ko mutant=69.3±4.0 μm²; P=0.0004). Furthermore, the HD-Q93; dhtt-ko brains showed larger areas that were devoid of neuronal cells (compare Fig. 9A and 9D with Fig. 9F and 9I, respectively; also see supplementary material Figs S7 and S8). The results obtained from quantifying the brain size of these flies suggested that, although the overall size of the brains was similar between HD-Q93 and HD-Q93; dhtt-ko flies (Fig. 9M) (average brain size: HD-Q93=119,279±2015 μm², n=7; HD-Q93; dhtt-ko=115,389±2412 μm², n=7; P=0.003), the total area that was devoid of neuronal cells was increased by about 25% (Fig. 9N) (total area devoid of neurons cells: HD-Q93=26,324±1869 μm², n=7; HD-Q93; dhtt-ko=32,945±2494 μm², n=7; P=0.03). Together, these results suggest that there is increased disorganization and an increase in neuronal loss in the brains of HD-Q93; dhtt-ko flies. Thus, loss of endogenous dhtt renders animals more vulnerable to the toxicity associated with polyQ-expanded Htt.

Htt has been characterized extensively in mammalian cell culture and in mouse systems (Cattaneo et al., 2001; Harjes and Wanker, 2003). However, only a few functional studies have been performed on its homologs in other model organisms. By deleting the dhtt gene, we demonstrate that Htt is not required for normal development of Drosophila, but instead has an essential role for the long-term mobility and survival of adult animals. Subsequent analyses revealed that loss of dhtt mildly affects the integrity of adult brains. Further, in the absence of endogenous dhtt, the neurodegenerative phenotypes associated with a Drosophila model of polyQ toxicity were enhanced significantly.

Earlier studies showed that mice lacking Htt die during early embryogenesis (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995). It is surprising to find that dhtt is dispensable during Drosophila development. As no other Htt homolog exists in the fly genome (Li et al., 1999), such a mild phenotype is unlikely to be caused by a functional redundancy resulting from another Htt-like gene in Drosophila. Considering the evolutionary distance between Drosophila and mammals, such an observation might indicate that the Drosophila and mammalian Htt proteins, with their relatively restricted sequence homology, are not functionally conserved. However, this phenotypic discrepancy might also reflect intrinsic differences during mouse and fly embryogenesis. In a chimeric analysis of Htt mutant mice, Dragatsis et al. showed that the early embryonic lethality of Htt-null mice was primarily the result of a crucial role of Hdh in extraembryonic membranes (Dragatsis et al., 1998). It is probable that Drosophila does not require equivalent tissues, such as extraembryonic membranes, to support its early development and that its embryogenesis can proceed normally in the absence of dhtt. Although an Htt homolog is found in Drosophila and vertebrates, no Htt homolog has been found in yeast or C. elegans (Li et al., 1999). The absence of an Htt homolog in C. elegans suggests that Htt does not have a function that is essential for the development of invertebrates in general, which is in agreement with our observation that dhtt is dispensable for normal development of Drosophila. Interestingly, a phylogenetic comparison of Htt proteins from different species postulates that Htt in the protostome (which includes Drosophilids) might be dispensable because, when compared with the deuterostome branch, evolution of the Htt genes along the protostome branch is more heterogeneous (Tartari et al., 2008). Further studies will be required to determine whether the mammalian Htt gene can rescue Drosophila dhtt-null phenotypes.

Analyses of dhtt-ko mutants suggest that loss of dhtt does not affect synapse formation, neurotransmission or axonal transport (Figs 3–6), which is in agreement with the absence of an Htt homolog in the worm, in which the essential components of these cellular processes are conserved. It is possible that Htt still regulates these cellular processes, but with a minor role. Extrapolating, there is a possibility that the function of Htt is associated with a novel cellular process and/or animal function that has been acquired during evolution. For example, although dhtt is dispensable for Drosophila development, it is important for maintaining the long-term mobility and viability of adult animals (Figs 5 and 6). Compared to the worm, Drosophila has a relatively long life span, a more complex nervous system and a more active life cycle, raising an intriguing possibility that the function of Htt might be directly related to these higher functions in adult animals.

Although dhtt is dispensable for normal Drosophila development, dhtt mutants show significantly reduced mobility and viability as they age, indicating an important role of dhtt in maintaining the long-term functioning and survival of adult animals (Fig. 5). Analysis of dhtt mutants revealed a mild abnormality in MB structure and a reduction in the complexity of axonal termini in the brain (Fig. 7). Similarly, Htt mutant mice with a reduced level of Htt expression display severe brain abnormalities, even though non-neuronal tissue forms normally (White et al., 1997). Targeted inactivation of Hdh in the mouse forebrain also causes a progressive neurodegeneration phenotype, suggesting that Hdh is required in the development and survival of neuronal cells (Dragatsis et al., 2000). It remains to be determined whether common underlying molecular mechanisms are responsible for the observed brain phenotypes in the fly and mouse. Nonetheless, considering the evolutionary distance between Drosophila and the mouse, the existence of structural defects in the adult brain in both species could indicate a conserved role of Htt in maintaining neuronal integrity.

The axonal terminal phenotype in dhtt mutants is also reminiscent of that observed in a mouse knockout model for Htt (Hdh^ex5), in which exon 5 has been deleted (Nasir et al., 1995). Although mice homozygous for this Hdh deletion are early embryonic lethal, Hdh^ex5 heterozygotes survive to adulthood and display increased neuronal loss, motor and cognitive deficits, and a significant loss of synapses in specific regions of the brain (Nasir et al., 1995; O’Kusky et al., 1999). Synapse complexity can be modulated by many factors, including neuronal activity and membrane and cytoskeleton dynamics. In the future, many questions remain to be answered, such as the significance of this axonal terminal phenotype, the exact function of Htt in axonal termini complexity and the possibility of a causal link between this brain defect and the observed adult mobility and viability phenotypes.

Extensive studies on HD and other polyQ diseases have demonstrated that an expanded polyQ tract can itself be neurotoxic (Zoghbi and Orr, 2000). Given that the distinctive neuronal loss observed in each polyQ disease is caused by otherwise unrelated disease genes that are widely expressed, it has been hypothesized that other cellular factors affect disease pathogenesis (Cattaneo et al., 2001; Zoghbi and Orr, 2000). In HD, accumulating evidence from cell culture and mutant mouse studies suggest that wild-type Htt has a neuroprotective function (Cattaneo et al., 2001). When endogenous wild-type Hdh is replaced by yeast artificial chromosomes (YACs) containing full-length human Htt with an expanded polyQ tract (YAC46, YAC72 and YAC128) (Leavitt et al., 2001; Van Raamsdonk et al., 2005), the mice develop massive cell death in the testes that can be suppressed by the wild-type Hdh gene, suggesting that the normal function of Htt might mitigate the cellular toxicity associated with the polyQ-expanded mutant Htt protein (Leavitt et al., 2001; Van Raamsdonk et al., 2005). In addition, a genome-wide study of HD animal models and postmortem tissues has shown that neuronal genes regulated by the transcriptional repressor REST/NRSF, including the gene encoding BDNF, could be similarly repressed by either the presence of the polyQ-expanded Htt protein or by depleting endogenous wild-type Htt (Zuccato et al., 2007). Together, these and other studies support the hypothesis that loss of normal Htt function affects HD pathogenesis.

We were able to examine the effect of removing endogenous dhtt on the pathogenesis of an established Drosophila HD model for polyQ toxicity (HD-Q93). Our results show that the phenotypes of HD-Q93 flies, including mobility, viability and brain pathology, are significantly exacerbated in the absence of endogenous dhtt, providing in vivo evidence that loss of normal Htt function can accelerate HD pathogenesis. It should be noted that our results do not directly demonstrate that loss of the endogenous Htt protein specifically affects HD pathogenesis, because the enhanced pathology might be the result of an additive effect of two detrimental factors in the animals, namely the presence of a toxic polyQ tract and the disturbance of the normal function of Htt. Although dhtt-ko mutants exhibit only a mild age-dependent adult phenotype, it is possible that, in the absence of endogenous dhtt, some undetected cellular defects might develop that render the animals more susceptible to other cellular attacks. In the presence of a toxic polyQ tract, this vulnerability might be exposed further, leading to additive phenotypes. Our result is in agreement with the early hypothesis that HD might be caused by the combination of an acquired toxicity, conferred by the expanded polyQ tract in the mutated Htt, and an incurred neuronal vulnerability, arising from the loss of endogenous Htt function (Cattaneo et al., 2001). It is important to note that previous studies have demonstrated that expansion of polyQ tracts within the Htt protein does not abolish its endogenous function, as full-length Htt with an expanded polyQ tract can fully support the development of Htt-null mice (Leavitt et al., 2001; Van Raamsdonk et al., 2005; White et al., 1997).

Recently, RNAi-mediated depletion of the polyQ-expanded Htt protein has been proposed as a therapeutic approach against HD (Farah, 2007). Our data, together with previous mammalian studies, suggest that the normal function of Htt has a conserved neuroprotective role in the brain and that its depletion could render neuronal cells more vulnerable to the toxicity associated with the polyQ tract (Auerbach et al., 2001; Cattaneo et al., 2001; Dragatsis et al., 2000; Leavitt et al., 2001). Application of such an RNAi-based strategy requires a consideration that the normal Htt function be preserved. Given the devastating consequence of fully progressed HD, together with the observations that Htt-null neurons can develop and survive in the adult mouse brain and that dhtt-ko animals are largely normal with only mild adult brain defects, our data indicate that the benefit gained from balanced administration of RNAi knockdown therapy in the adult brain might justify the relatively mild loss caused by depletion of the Htt-associated neuroprotective function.

Flies were maintained at 25°C and raised on standard Drosophila medium unless specified otherwise. To establish transgenic animals, DNA constructs were injected into w¹¹¹⁸ embryos together with the pπ25.7wc helper plasmid to generate germ line transformation; transformants were then selected in the next generation, according to standard procedures. Unless specified otherwise, flies of the genotype w¹¹¹⁸/w¹¹¹⁸ were used as controls in all assays because the dhtt-ko mutant allele was generated from this w¹¹¹⁸ genetic background during the genetic crosses.

The p-element line d08071, piggyBac insertional line f05417 and the FLP recombinase transgene line [genotype: y, w, pCaspeR-(hs-flipase)] were from Exelixis (Parks et al., 2004).

To analyze the A307-Gal4 labeled neurons in adult brains, flies with genotypes of A307-Gal4/+; dhtt-ko/TM6C, Tb and UAS-mCD8-eGFP/+; dhtt-ko/TM6C, Tb were generated and crossed together; their adult progenies carrying A307-Gal4/UAS-mCD8-eGFP; dhtt-ko/dhtt-ko were selected and analyzed. The controls were the progenies from crossing the A307-Gal4 line with the UAS-mCD8-eGFP line.

HD-Q93 flies were obtained by crossing the pan-neuronal line elav-Gal4 (C155) with the UAS-Httexon1-Q93 line (P468) provided generously by L. Thompson and J. L. Marsh, respectively (Steffan et al., 2001). To generate ‘HD-Q93; dhtt-ko’ flies, flies with genotypes of elav-Gal4(c155)/+; dhtt-ko/TM6C, Tb and UAS-Httexon1-Q93/Cyo; dhtt-ko/TM6C, Tb were established and then crossed together; adult progenies with the genotype of elav-Gal4/+; UAS-Httexon1-Q93/+; dhtt-ko/dhtt-ko were then selected and analyzed. Controls flies had the following genotypes: elav-Gal4 (C155)/+; UAS-Httexon1-Q93/+; elav-Gal4 (C155)/+; dhtt-ko/dhtt-ko; and UAS-Httexon1-Q93/+; dhtt-ko/dhtt-ko.

The Df(98E2) deficiency, in which both CG9990 and dhtt are deleted, was generated by following the Flp-FRT-based procedure, as described previously (Parks et al., 2004). Briefly, the p-element line d08071 (inserted at the 5′ end of the neighboring CG9990 gene) was crossed to virgin flies carrying an FLP recombinase transgene [genotype: y, w, pCaspeR-(hs-flipase)]. For the next generation (F1), the male progenies carrying both the d08071 and the hs-flipase lines were selected and mated with virgins carrying the f05417 line, inserted near the 3′ end of the dhtt gene. The progenies (i.e. the F1 flies) from the above crosses were heat-shocked at 37°C for 1 hour, 48 hours after egg-laying, and were then heat-shocked for a further 1 hour during each of the following 4 days. In the third generation (F2), the virgin females were collected and crossed to males carrying balancer chromosomes (w; TM3, Sb/TM6B, Tb, Hu). In the fourth generation (F3), progeny flies carrying the Df(98E2) deletion from the above crosses (between F2 flies) were selected based on their darker eye color, and crossed with flies carrying balancer chromosomes to establish individual fly lines. The presence of the deletion in these lines was confirmed by PCR analysis of extracted genomic DNA and by DNA sequencing.

To generate the genomic rescue construct for the CG9990 gene, a 24.7 kb genomic DNA fragment, covering the CG9990 genomic region, was isolated from the bacterial artificial chromosome (BAC) clone BACR10P23 (CHORI) following double digestion with the restriction enzymes XbaI and XmaI. This 24.7 kb genomic DNA fragment starts at an XbaI site near the end of the neighboring CG9989 gene, 2.66 kb from the inserted site of the p-element line d08071, and ends at the XmaI site within the second exon of the dhtt gene, thus covering the whole genomic region of the CG9990 gene (Fig. 1B). To generate the CG9990 rescue transgene, this 24.7 kb genomic DNA fragment was cloned into the NotI and SmaI sites in the pCaspeR-4 transgenic vector. The DNA for the pCaspeR-4–CG9990 genomic rescue construct was injected into w¹¹¹⁸ embryos and transformants were then selected following standard procedures. Three independent CG9990 transgenic lines were established and then tested by crossing them into the Df(98E2) deletion.

Flies with the Df(98E2) deletion, which removes both CG9990 and dhtt, are homozygous lethal at the embryonic stage. Through genetic crossings, we reintroduced the CG9990 genomic rescue transgene back into the Df(98E2) deletion background, generating fly lines that are defective at the molecular level for only the dhtt gene (referred to as dhtt-ko flies). Three independent genomic CG9990 transgenic lines were tested by crossing them into the Df(98E2) deletion; all of them rescued the embryonic lethality of the Df(98E2) deletion, producing viable adults. Thus, dhtt-ko flies, which carry both the Df(98E2) and the CG9990 transgene, are homozygous viable and can fully develop into adulthood, demonstrating that the lethality observed with the Df(98E2) deletion is caused by the loss of CG9990. To verify further that the dhtt gene is indeed deleted, as expected, in this dhtt-ko allele, we extracted genomic DNA from the homozygous dhtt-ko and wild-type control adults and performed Southern blots.

Genomic DNA was extracted from the dhtt-ko adults or from control w¹¹¹⁸ adults and digested with the restriction enzyme BamHI. DNA fragments were separated on a 1% agarose gel and transferred onto a nitrocellular membrane, according to the standard protocol for Southern blotting. DNA fragments specifically targeting all exons of the dhtt gene were labeled with ³²P-dCTP using the Klenow polymerase and random-hexamer primer method (Amersham) and used as probe. Hybridization was performed overnight at 65°C and the fragments were subsequently washed with SDS/SSC buffers, following standard procedures for Southern blotting.

The size of the genomic region covering dhtt is about 43 kb (Li et al., 1999), which is too large to be cloned into a transgenic vector by conventional approaches. Therefore, we engineered a dhtt minigene rescue construct that expresses full-length dhtt under the control of its own endogenous regulatory region. In Drosophila, the regulatory elements controlling the endogenous expression pattern of a gene are normally located at the 5′ region of the gene, within the upstream 5′ untranscribed region and in the first few introns. In addition, the expression level of a gene is also affected by its 3′ untranslated region. As a result, we isolated a 14.9 kb genomic DNA fragment from the BAC clone BACR10P23 (CHORI) following double digestion with the restriction enzymes XbaI and AscI. The XbaI site was located at the end of the CG9990 gene and the AscI site was within the tenth exon of the dhtt-coding region, thus the DNA fragment covered all of the 5′ untranscribed region of the dhtt gene together with the first ten introns and exons of the gene. We also isolated a 1.2 kb genomic DNA fragment, by double digestion with StuI-EcoRI, that covers the 3′ end of the dhtt gene, including the last exon of dhtt and the remaining polyA sites and untranscribed region at the 3′ end of the gene. Next, we assembled and isolated a 6.6 kb dhtt cDNA fragment, by digestion with AscI and StuI, that covered most of the 3′ cDNA region of dhtt, from the single AscI site within the tenth exon to the StuI site within the last exon. The dhtt minigene rescue construct was generated by ligating and fusing, in frame, the three fragments that cover all of the dhtt regulatory and coding regions: (1) the 5′ 14.9 kb genomic DNA fragment that covered all of the 5′ untranscribed region of the dhtt gene, as well as the first ten introns and exons of dhtt; (2) the middle 6.6 kb of the dhtt cDNA fragment that covered from exon 10 to exon 29; and (3) the 3′ 1.2 kb genomic DNA fragment that covered exon 29 and the polyA sites at the 3′ end of the gene, as well as the remaining 3′ untranscribed region. The 22.7 kb dhtt minigene was cloned into the NotI and XbaI sites in the pCaspeR-4 transgenic vector. After generating transgenic flies with this dhtt minigene rescue construct, we crossed these animals into the dhtt-ko mutant background to test whether the mobility and viability phenotypes of dhtt-ko mutants could be rescued.

Based on the published dhtt cDNA sequence (Li et al., 1999), we isolated overlapping fragments covering the full-length of the dhtt cDNA by PCR amplification using Pfu polymerase. We then assembled a full-length dhtt cDNA construct using these sequencing-verified dhtt fragments, cloned it into a pUASP vector, and generated transgenic fly lines by following standard protocols.

Total RNA samples were isolated, using Trizol reagent (Invitrogen), from adult animals of the following genotypes: w¹¹¹⁸ wild-type control, homozygous dhtt-ko mutants and dhtt-ko Rescue. RT-PCR reactions were performed using Superscript one-step RT-PCR with Platinum Taq (Invitrogen) following the manufacturer’s instructions. PCR primers were designed as follows: rp49 control: forward (in exon 1 of rp49) 5′-ACCATCCGCCCAGCATACAGG-3′, reverse (in exon 2 of rp49) 5′ -TTGGCGCGCTCGACAATCTCC-3′; dhtt-N: forward (in exon 5 of dhtt) 5′ -GCCAATGTAGCCAGAGTCTG-3′, reverse (in exon 6 of dhtt) 5′ -CGCATTCGCTGATGCTGCGTG-3′; dhtt-M: forward (in exon 13 of dhtt) 5′-AAGCTATTCGAGCCGATGGTC-3′, reverse (in exon 15 of dhtt) 5′-GCACCAGGAATCTCAGCATGG-3′ ; dhtt-C: forward (in exon 23 of dhtt) 5′-TCGGGAATTGACTTTCGCAGC-3′, reverse (in exon 24 of dhtt) 5′-TGCAGTTTGAGGCAGCGTTCC-3′.

Using the motif-predicting program developed by M. A. Andrade at the EMBL (http://www.embl-heidelberg.de/~andrade/papers/rep/search.html), both dHtt and human Htt were analyzed for the presence of HEAT repeats. Using the default parameters at the site and including the AAA, ADB and IMB groups of HEAT repeats (Andrade et al., 2001), a total of 40 and 38 HEAT repeats were identified for Htt and dHtt, respectively.

Sample preparation and staining for embryos, larval tissues and adult fly eyes have been described previously (Sullivan et al., 2000). To analyze NMJs, wandering third instar larvae were dissected in Ca²⁺-free saline (128 mM NaCl, 2 mM KCl, 0.1 mM CaCl₂, 4 mM MgCl₂, 35.5 mM sucrose, 5 mM Hepes at pH 7.2, 1 mM EGTA) and stained, as described (Sullivan et al., 2000). 1–2 μM thin sections of embedded adult eyes were cut using a microtome and imaged directly without further dye staining.

To stain and image the adult brains, flies were dissected in 1× PBS or in Drosophila M3 medium, to remove cuticles and external eye tissues, then fixed in 4% formaldehyde in 1× PBS for 1 hour at room temperature (RT), washed six times with 1× PBT for 1 hour, and stained with primary and secondary antibodies.

Collection and fixation of Drosophila embryos, larvae tissue dissection and fixation, as well as RNA in situ hybridization were carried out according to standard procedures (Hauptmann and Gerster, 2000). The DIG-labeled RNA probes were generated according to the manufacturer’s instructions (Roche, Indianapolis, IN). The dhtt exon 8 and exon 20 sequences were used to generate dhtt-specific in situ probes, which gave similar in situ results. The samples for in situ hybridization were analyzed with a Zeiss Axiophot 2 compound microscope.

To generate the antibody against dHtt, a cDNA fragment corresponding to the N-terminal 459 amino acids of dHtt was cloned into the pGEX-4T1 vector; the corresponding GST-fusion protein was purified according to manufacturer’s instructions (Promega) and used to produce polyclonal antibody in a rabbit (Convance). The antiserum was affinity-purified and used at a final dilution of 1:200.

Primary antibodies were applied at 4°C overnight at the following dilutions: rabbit anti-GFP (1:1000, Molecular Probes); rabbit anti-fasciclin II (1:400) and anti-DIG (1:40,000), both generously provided by Mary Packard and Vivian Budnik (Budnik et al., 1996; Koh et al., 1999); rabbit anti-synaptotagmin (1:500) (Littleton et al., 1993); mouse anti-α-tubulin (1:10,000, Sigma); monoclonal mouse anti-fasciclin II (1D4, 1:20), anti-DIG (1:20), anti-CSP (1:20), anti-synapsin (1:50), anti-α-spectrin (1:20), anti-Armadillo (1:100), nc-82 (1:50) and anti-Futsch (22C10, 1:100) (all from Developmental Studies Hybridoma Bank) (Hummel et al., 2000; Roos et al., 2000); goat anti-HRP (1:500, Jackson Labs); Alexa488-, Alexa594- and Alexa647-conjugated secondary antibodies (all used at 1:500, Molecular Probes). Rhodamine Red X-conjugated goat anti-HRP and CY5-conjugated secondary antibodies were used at 1:200 (both from Jackson Labs). DAPI (0.2 μg/ml, Molecular Probes) and TRITC-conjugated phalloidin (10 ng/ml, Sigma) were applied in PBS-Tween (PBST) for 30 minutes to label nuclei and F-actin, respectively.

To quantify MBs, adult brains were prepared for imaging as described above. Dissected adult brains from both wild-type controls and dhtt-ko mutants were stained side by side in a 12-well plate with mouse anti-Fas II antibody (1D4, 1:20) overnight at 4°C. After washing six times with 1× PBT for 2 hours, samples were stained with Alexa-594-conjugated secondary antibodies (1:500) at RT for 2 hours, followed by extensive washing with 1× PBST for 2 hours at RT. To ensure that the stained samples had a similar background signal during imaging analysis, brains from both wild-type controls and dhtt-ko mutants were mounted onto opposite ends of the same slide. The samples were photographed at 20× magnification, under the same optical parameters, using a Zeiss fluorescent microscope (Axioskop 2 Mot Plus). The boundaries of MBs were traced manually using AxioVision Rel 4.5 software (Zeiss), and both the overall size of each MB and the total intensity of FasII staining signals within the MB boundary were computed using the same software. FasII staining signals from outside the MB boundary were also measured and used as a reference to subtract out the background signals. Since we could not confidently measure the thickness of the MBs in these samples, only the overall area covered by each MB was measured. The average FasII signal intensity in each MB was calculated by dividing the ‘total intensity of FasII staining signals within the MB boundary’ by ‘the overall size of the MB’. To calculate the relative signal intensity, the average signal intensity of FasII signals from all wild-type brains was calculated and its mean was set as a reference point of 100.

To measure the overall brain size and the regions that were devoid of neurons in the brain, adult flies were dissected and stained with rat anti-Elav antibody (1:40), which labels all neuronal cells. The boundary of the whole protocerebrum in the brain and the regions devoid of neurons were tracked manually and quantified with AxioVision Rel 4.5 software, as described above.

To quantify the size of the A307-positive axonal terminals, adult brains from both wild-type controls and dhtt-ko mutants were dissected, fixed and stained side by side with an anti-GFP antibody, as described above, then mounted on the two sides of the same slide to ensure that they had a similar background for imaging analysis. Both the wild-type control and dhtt-ko mutant samples were imaged by confocal microscopy at 40× magnification under the same optical parameters. A Z-serial section of confocal images covering the entire depth of each axonal terminus was collected using a Leica TCS confocal microscope and then projected into one merged image (using the Leica software) to generate the whole picture of each axonal terminus. Since the branches and boutons in the brain were too small to distinguish clearly, only the overall area covered by each axonal terminus was measured.

Fluorescent images were analyzed and captured by fluorescent microscopy (Axioskop 2 Mot Plus, Zeiss) or by confocal microscopy (Leica TCS SP2 AOBS system). Confocal images were analyzed and projected using Leica confocal software (LCS).

Drosophila viability was measured by placing 30 newly hatched female flies of each genotype into individual vials, containing fly food, at 25°C. The number of dead flies was recorded daily. Flies were transferred into a new food vial every 3–4 days to prevent them from sticking to old food. For each genotype, at least two independent cohorts of flies, raised at different times from independent crosses, were tested and the results were averaged.

Female flies were chosen in all the mobility tests. Fly videos were captured using a Sony digital camcorder (DCR-TRV140) and edited in Apple’s iMovie program.

Climbing assays were performed as described previously (Ganetzky and Flanagan, 1978; Le Bourg and Lints, 1992). Briefly, 20 flies of a specified age were knocked to the bottom of a plastic vial. The number of flies that could climb to the top of the vial after 18 seconds was counted. The test was repeated 6–7 times for each genotype at the specified age.

The assay was performed as described previously (Feany and Bender, 2000; Joiner and Griffith, 1999; Wang et al., 2004). For each genotype, 6–14 flies of a specific age were tested.

Flies were fed 30 mM of methyl viologen (Sigma) in instant Drosophila medium (Carolina), providing a dosage of paraquat that kills about 50% of wild-type flies after 48 hours of exposure. Control flies were fed instant medium only. Adult flies that had eclosed within the last 24 hours were kept in 30 mM paraquat or in drug-free medium for 48 hours after eclosion. The number of survivors was counted at 48 hours after the start of paraquat treatment.

An electrophysiological analysis of wandering-stage third instar larvae was performed in Drosophila HL3.1 saline (NaCl, 70 mM; KCl, 5 mM; MgCl₂, 4 mM; CaCl₂, 0.2 mM; NaHCO₃, 10 mM; trehalose, 5 mM; sucrose, 115 mM; HEPES-NaOH, 5 mM; pH 7.2) using an Axoclamp 2B amplifer (Axon Instrument) at 22°C. Recordings were performed at muscle fiber 6/7 of segments A3 to A5 under current clamp. PPF was measured by determining the peak amplitude responses (P2/P1) of two stimuli separated by the indicated latency. All error bars are s.e.m. In adults, extracellular field potentials were recorded by placing a sharp glass electrode near the longitudinal flight muscles after piercing the cuticle, with a reference electrode placed in the fly head. ERGs were performed as described previously (Rieckhof et al., 2003). Temperature shifts were performed by heating mounting clay, which encompassed the fly, to the desired temperature with a peltier heating device.

For measurements of evoked EJP amplitude (Fig. 6A,B), the number of NMJs examined were: control, n=8; dhtt-ko, n=23. Voltage traces of evoked EJPs were recorded from muscle fiber 6 in third instar larvae in 0.2 mM extracellular calcium. Average resting potential: 60.2±1.2 mV in control animals and 62.4±0.8 mV in dhtt mutants. Average EJP amplitude: 19.5±1.2 mV in control animals and 17.8±1.0 mV in dhtt mutants.

Measurements of voltage traces of PPF (Fig. 6C) were performed at 25 millisecond intervals in control (rescued) and dhtt mutant third instar larvae in 0.2 mM extracellular calcium. Quantification of PPF (the amplitude of EJP 2 divided by the amplitude of EJP 1) was performed in control and dhtt mutants for 25-, 50- and 75-millisecond intervals (Fig. 6D). The number of preparations analyzed were: control, n=9; dhtt-ko, n=8.

The GF flight circuit can be activated by stimulation of the brain, and extracellular recordings can be made from the DLMs. Neither the control nor the dhtt-ko mutants displayed abnormal activity in the DLM flight muscles (data not shown).

To measure ERGs at the elevated temperature (37°C), flies were rapidly heated from 20°C to 37°C, with test light pulses (black bar below trace in Fig. 6E) given at regular intervals. A total of ten dhtt-ko mutants, aged 1–3 days, were tested and all showed normal ERGs at 20°C and 37°C (Fig. 6E). The number of preparations for the 40–45-day-old animals analyzed was: control, n=10; dhtt-ko, n=16.

We thank Leslie M. Thompson and J. Lawrence Marsh for providing advice and Drosophila stocks; Mary Packard for critical discussions and for generously providing experimental materials and technical help; Pengyu Hong for helping to quantify the size of the A307-positive axonal termini; Bun Chan for assistance in capturing and editing videos; Rodney Stewart, Richard Binari, Bernard Mathey-Prevot and Carl Johnson for advice and critical reading of the manuscript; and members of the Perrimon lab for their assistance, in particular Jianwu Bai for assistance in muscle analysis and Christian Villalta for fly injections. We thank the anonymous reviewers for their insightful comments. S.Z. gratefully acknowledges support in the form of a fellowship from the Harvard Center for Neurodegeneration and Repair (HCNR), as well as the Lieberman Award from the Hereditary Disease Foundation (HDF). N.P. is an Investigator of the Howard Hughes Medical Institute.