in Drosophila - Semantic Scholar

13 downloads 164 Views 1MB Size Report
Chemistry, D240 Medical Sciences I, University of California—Irvine, Irvine, CA 92697-1700, USA. Received 23 ...... Microscopy Sciences, Fort Washington, NY). ... Ross, C.A. (1997) Intranuclear neuronal inclusions: a common pathogenic.
© 2000 Oxford University Press

Human Molecular Genetics, 2000, Vol. 9, No. 1

13–25

ARTICLE

Expanded polyglutamine peptides alone are intrinsically cytotoxic and cause neurodegeneration in Drosophila J. Lawrence Marsh, Heli Walker, Heidi Theisen+, Ya-Zhen Zhu1, Tom Fielder1, Judy Purcell and Leslie M. Thompson1,§ Developmental Biology Center and Department of Developmental and Cell Biology, and 1Department of Biological Chemistry, D240 Medical Sciences I, University of California—Irvine, Irvine, CA 92697-1700, USA Received 23 August 1999; Revised and Accepted 15 October 1999

Several dominant, late-onset neurodegenerative diseases (e.g. Huntington’s disease) are caused by expansion of polyglutamine (polyQ) repeats within specific proteins. The diverse, yet overlapping, pathology of these diseases could be due to novel deleterious functions unique to each protein or to a common pathophysiology mediated by the long polyQ chains themselves. By engineering Drosophila to express different polyQ peptides, we find that expanded polyQ chains alone are intrinsically cytotoxic and cause neuronal degeneration and early adult death. We further find that this intrinsic toxicity is dependent on cell type and polyQ length and that the inclusion of other amino acids modifies and reduces toxicity. This is the first in vivo evidence that polyQs, when removed from their disease gene context, cause neurotoxicity. These studies provide a basis for understanding the diverse clinical presentations in terms of the intrinsic cytotoxic effect of polyQ peptides being modulated by protein context. Parallel experiments in which cytotoxic polyQ expansions were engineered into Dishevelled, a Drosophila protein containing a naturally occurring polyQ tract, strongly suggest that the effect of a toxic polyQ peptide can be neutralized by protein context. This animal model provides a simple and effective means of screening for therapeutics that relieves the polyQinduced lethality, independent of any particular disease gene. By quantifying the degree of lethality in several transgenic lines, we have identified a number of genetically modified strains that are suitable for eventual testing of compounds or genes that ameliorate the pathology of polyQ peptides.

INTRODUCTION Eight progressive, inherited neurodegenerative disorders are caused by an expansion of the naturally occurring CAG tract that codes for a polyglutamine (polyQ) repeat within the coding region of the corresponding protein. These diseases include Huntington’s disease (HD), spinal and bulbar muscular atrophy (SBMA), dentatorubral-pallidoluysian atrophy, spinocerebellar ataxia type 1 (SCA1), SCA2, SCA6, SCA7 and Machado–Joseph disease (MJD/SCA3) (1). With the exception of SCA6 (CACNL1A4) (2), which is characterized by a minimal repeat expansion, affected individuals show a similar range of repeat expansion above 35 repeats (3). Each disorder is inherited as an autosomal dominant (or X-linked in the case of SBMA), neurological syndrome with selective, neuronal cell death resulting in distinct, but overlapping, clinical and pathological manifestations (4). Age of onset is normally in mid-life; however, longer repeat ranges can cause more severe presentation of the disease with an earlier age of onset. Genetic +Present §To

studies provide evidence that inactivation of a single allele does not result in disease (5,6). In addition, mouse models for HD, SCA-1 and MJD (1,7–10), carrying expanded repeat transgenes in a background with two normal alleles, show phenotypes resembling the corresponding disease suggesting a true dominant effect. The appearance of neuronal intranuclear inclusions that contain huntingtin and ubiquitin, in mice transgenic for exon 1 of huntingtin, implicates protein misfolding and aggregation as potential mediators of neuronal pathogenesis (11). These insoluble neuronal aggregates and nuclear inclusions have been described for many of the polyQ repeat diseases, having been seen in the affected regions of brains from patients (3,12– 14) and in most of the transgenic mouse models (13,14). A role for nuclear localization of expanded polyQ repeat-containing disease proteins, independent of aggregation, has also been implicated in the initiation of disease and neurodegeneration (15,16). In contrast, the presence of cytosolic aggregates in

address: Department of Biological Science, California State University—Fullerton, PO Box 6080, Fullerton, CA 92834-6080, USA whom correspondence should be addressed. Tel: +1 949 824 6756; Fax: +1 949 824 2688; Email: [email protected]

14

Human Molecular Genetics, 2000, Vol. 9, No. 1

dystrophic neurites and neuropils in HD brain sections and in HD transgenic mice may reflect a pathogenic role for non-nuclear localization and aggregation (12,17,18). The aggregation phenomenon has been reproduced in vitro in a protein concentration and repeat length-dependent manner (19), demonstrating that aggregation is a property mediated by the expanded polyQ. The structure and behavior of polyQ repeats, both isolated and in protein contexts, have been examined in vitro; these studies argue that a structural transition associated with increased length occurs to mediate aggregation (20). A central issue is to determine the role of the expanded polyQ repeats in pathogenesis. Is toxicity of an expanded polyQ tract independent of the protein in which it is embedded or does incorporation of additional glutamines cause a gain of function or altered function of the resident protein itself that is toxic? Observations consistent with both models exist (21,22), but since functions of only two of the CAG repeat disease genes are known [the androgen receptor (SBMA) (23) and a voltage-dependent calcium channel (SCA6) (2)], a distinguishing, systematic test based on protein function has not been possible. Recently, fragments of human disease genes with expanded polyQ repeats have been expressed in Drosophila, demonstrating late-onset neuronal degeneration (24) and degeneration of photoreceptor neurons (25). In the absence of more information about the normal interaction and function of, for instance, the HD protein, a role for aberrant function of the mutant protein cannot be ruled out. Even though only fragments of the relevant human disease proteins have been used in the Drosophila models and in many transgenic mouse models, these fragments may influence the disease progression in these systems. To distinguish between protein context and direct polyQ toxicity models, we have developed a general model system, using D.melanogaster, to assess the effect of polyQ repeats on cell function and protein localization. Specifically, we have engineered transgenic flies to express polyQ tracts alone as free or epitope-tagged peptides and in parallel have modified the naturally occurring polyQ tract of a characterized Drosophila gene, dishevelled (dsh), to contain expanded repeats. By quantifying the phenotypes of the transgenic lines, we document a correlation between the severity of neuronal degeneration and age of onset of lethality. RESULTS Expression of polyQ-containing transgenes and peptides To determine the effects of an expanded polyQ peptide independent of the effects caused by an associated protein, we expressed polyQ peptides under the control of an upstream activator sequence (UAS) promoter in Drosophila. In this binary expression system, a transgene is placed under the control of the yeast UAS and is not expressed until crossed to strains of flies that express the yeast GAL4 transcription activator in different tissues and different cell types (26). Two measures were used to assess the effects of polyQ expression: (i) overall viability under different promoter drivers; and (ii) dominant phenotypic effects of expression on particular cells and tissues. The level of activity of p-element transgenes in Drosophila is influenced by the chromosomal location of

the insert (27); thus, for each construct, several transgenic lines with different expression levels were tested. As described below, quantitation of protein levels for all transgenes was not possible with the available antibodies; therefore, we tested a series of inserts for each construct under the assumption that the range of levels of expression would be similar for each construct. In addition, a given chromosomal insert location was tested with several different GAL4 promoters/drivers to allow comparisons of the effects of expressing polyQs in different tissues. Many of the triplet repeat disease genes are widely, if not ubiquitously, expressed and yet each disease exhibits a different range of tissue sensitivity (21). With the Drosophila system, we have used a series of promoters to target all neurons, subsets of neurons or non-neuronal tissues in developing epithelia. On examination of the viable offspring produced by crosses between UAS-transgene and GAL4 driver lines, we observed phenotypes, including lethality, not readily explicable by the reported tissue expression. Therefore, we crossed each of the drivers to a UAS:green fluorescent protein (GFP) strain of flies and monitored the expression in larval tissues. In some cases, this revealed expression in larval tissues other than those previously reported. Figure 1 shows both published and experimentally determined additional expression patterns for each driver used here. To determine the effects on viability, we expressed different polyQ-containing transgenes under the control of the elav, sev, dppblk and gmr promoters driving GAL4 (Fig. 2A). In some protein localization experiments, we used a ptc-GAL4 driver that directs expression in a pattern very similar to that of dppblkGAL4, except that it is expressed in all cells posterior to the furrow of the eye (28). PolyQ alone is cytotoxic To determine the effect of polyQ peptides themselves, we engineered transgenes containing 22 or 108 glutamines flanked by only 6 and 4 amino acids on the N- and C-terminal sides, respectively (Fig. 2A). Expression of a short polyQ peptide, Q22, has no effect on viability with any of the promoters tested (Fig. 3A). In contrast, expression of an expanded polyQ peptide, Q108, has a dramatic dominant effect on viability. When expressed in all neurons from embryogenesis onwards (elav-GAL4), lethality is between 77 and 100%, depending on the insert line examined (measured as a percentage of expected viable offspring). When expressed by the sev promoter, which expresses ubiquitously at low levels and in many neurons and some non-neuronal tissues, a dominant lethality of 100% is observed with all lines examined. When expressed by the dppblk promoter, which targets developing discs in late stages, lethality again approached 100%. Even under the control of the gmr promoter, which is almost exclusively in the late-stage eye disc, some lines exhibit reduced viability when the Q108 peptide is expressed. The dominant lethality caused by this promoter, which is thought to be expressed only in the eye, could be due to expression that has not previously been noted in tissues other than the eye. One line containing the UAS-Q108 peptide was fully viable with all the GAL4 drivers used, in contrast to the lethality observed with the other UAS-Q108 lines. PCR analysis revealed that the CAG repeat had undergone an internal dele-

Human Molecular Genetics, 2000, Vol. 9, No. 1

15

Figure 1. Expression of GAL4 drivers used. For each GAL4 driver used, reported expression patterns are listed as well as a cartoon showing the expression pattern in different cells of the emerging eye. The gmr promoter expresses GAL4 in all cells of the developing eye behind the morphogenetic furrow (46). The elav promoter is expressed in all neurons of the peripheral and central nervous systems beginning in embryonic stages and continuing into adult life (47). This driver provides chronic expression of expanded polyQs in neurons. The sev-GAL4 construct expresses at a low level in a ubiquitous pattern due to residual heat shock promoter elements and also expresses transiently at high levels in selected neurons such as the photoreceptors of the eye (50). The dppblk-GAL4 driver expresses GAL4 in the growing epithelia of imaginal discs in a pattern reminiscent of normal dpp, namely along the apical/posterior (A/P) border of wing discs, the dorsal A/P border of leg discs and the ventral A/P border of antennal discs (48). Notably, dpp-GAL4 is not expressed in the morphogenetic furrow of the eye disc, but is expressed in a ring around the eye. This promoter and the sev promoter also direct expression in the salivary glands. To check for non-reported patterns of expression, transgenic animals containing UAS–GFP were crossed to the indicated fly lines. Additional tissues with expression are listed. Eye expression column: expression is shown in one ommatidium (the repeating unit of the compound eye).

tion to a size nearing the Q22 (data not shown). The observation that the single apparent exception was accompanied by a deletion that reduced the number of glutamines produced underscores the strongly deleterious effect of expanded polyQ alone on cellular function. Toxicity is reduced by additional amino acids appended to expanded polyQs To monitor the fate of expressed peptides, we engineered the Q22 and Q108 expression vectors to produce polyQ peptides that contain 26 additional amino acids in the form of associated epitope tags (myc and flag epitopes) (Fig. 2A). Expression of short Q22 peptides alone causes no deleterious effects and that observation is not altered by the addition of the myc/flag epitope. However, in every case examined, the addition of the 26 amino acid myc/flag epitope to the expanded polyQ peptide, Q108, dramatically reduced its lethality (Fig. 3A). For example, when a Q108 peptide is expressed in all neurons (elav), lethality is reduced from an average of 87% for the Q108 lines to 0% for Q108 with a myc/flag tag. When expressed by the sev promoter, the presence of the epitope tag reduces lethality from 100 to 0–66%, depending on the insert

line. When expressed by the dppblk promoter, lethality is again reduced from an average of 97% to near zero for all lines. PCR and sequence data from genomic DNA from two of the Q108 myc/flag transgenic lines indicated that no deletions occurred in the epitope tagged constructs (data not shown). Thus, the addition of even a short peptide to an expanded polyQ peptide can dramatically alter its cytotoxicity. The phenotypes seen in animals with the tagged polyQ peptides are very similar to those seen in the few surviving animals without the epitope tags suggesting that it is the severity of the toxicity which is altered rather than a change in the cells that are sensitive. For example, when expressed by the dppblk promoter, addition of the myc/flag epitope to the expanded Q108 virtually eliminates the dominant lethality, but many of the surviving animals now exhibit the split thorax phenotype previously seen only in dying adults in their pupal cases with the non-tagged polyQ. Also, expression of the Q108 transgenes by the gmr promoter produces disrupted eyes both with and without the myc/flag tag. Although phenotypic comparisons have not been made due to the complete lethality of the Q108 peptide when expressed by the sev promoter, the addition of the myc/flag epitope to the peptide dramatically

16

Human Molecular Genetics, 2000, Vol. 9, No. 1

Figure 2. PolyQ-containing constructs. PolyQ peptides and dsh constructs were engineered to be expressed under the control of the UAS-GAL4 system. (A) UAS driving expression of 22 and 108 polyQ repeat peptides. The sequence of flanking amino acids both with and without the myc/flag epitope tag are shown. (B) UAS driving expression of constructs encoding 0, wild-type range (28 polyQ repeat tract) and 108 repeats within the dsh coding sequence.

increases survival, but all surviving animals exhibit a neuronalloss phenotype manifested by loss of large sensory bristles. Thus, in all UAS-GAL4 lines tested, addition of the myc/flag epitope dramatically reduces but does not eliminate the pathology of the expanded polyQ peptides. Expanded polyQ-induced pathology depends on cell type Examination of the developmental defects caused by expression in different tissues and cells reveals striking differences in the sensitivity of different cell types to the expanded polyQ-induced pathology. Effects on eye development. Development of the eye provides a sensitive assay for effects on cell differentiation and viability of both neuronal and non-neuronal cells. The cells that will give rise to the eye grow during the larval period as uncommitted cell types. During the third larval instar, a morphogenetic wave passes across the eye and as it passes, cells become committed to particular fates in the eye (29). These fates include neuronal cells (both photoreceptors and mechanoreceptors), cone cells and several pigment cells (Fig. 1). With the GAL4 drivers used here, the transgenes are expressed in different cells and for different lengths of time during eye development (Fig. 1). Expression of the Q108 peptide by elav-GAL4, which is expressed in all neuronal cells behind the furrow of the eye, with or without the epitope tag, resulted in an eye with normal external morphology and normally organized cone cells (lens)

and mechanosensory hair cells. However, the internal organization of the eye showed extensive disruption and this disruption was less severe with the tagged peptides than with the naked polyQ peptide. Analysis of the internal organization by sectioning of eyes embedded in plastic revealed extensive effects on neuronal development. In the eyes of Q108 flies, only two or sometimes three photoreceptor cells remained out of the eight found in normal eyes (see Fig. 1 for normal organization of eye) whereas in flies with the epitope-tagged Q108, six or seven photoreceptor cells typically remained (Fig. 3B). In both types of fly, large gaps in the tissue were evident with the overall number of gaps being greater in the Q108 than the epitope-tagged Q108. Thus, the disruptions to neuronal cells are less severe with the epitope-tagged Q108 than with the untagged peptide. Notably, neither peptide affects the survival of the mechanosensory hair cells whereas the adjacent photoreceptor cells are dramatically affected. In contrast to the normal external appearance of the eyes above, the eyes of flies with Q108 expressed by the gmr driver exhibit massive degeneration of the eye. In these animals, the transgene peptides are being expressed in all cells behind the morphogenetic furrow, not just the neuronal cells. External examination reveals that non-neuronal pigment cells have degenerated (Fig. 3C) and the organization of the ommatidia is highly abnormal, although the mechanosensory cells remain intact. The internal integrity of the eye is also compromised. On dissection or attempting to embed the eyes in plastic, the eyes collapse and separate easily from the surrounding cuticle (Fig. 3C). Crosses were made to determine whether the white gene (w+) could rescue the loss of pigment phenotype of Q108 driven by gmr-GAL4. However, the phenotype appears to be caused by a complete degeneration of the pigment cells in the eyes (data not shown), indicating sensitivity of a non-neuronal cell type to the presence of expanded polyQ peptides. Other developmental effects. We asked whether other nonneuronal tissues also exhibit sensitivity to expanded polyQ peptides. The dppblk-GAL4 driver expresses transgenes along the anterior/posterior compartment boundary in imaginal discs [discs are progenitor tissues that will give rise to the exoskeleton (cuticle) of the adult]. When dppblk drives Q108, most of the animals die (Fig. 3A). However, examination of dying adults in the pupal cases reveals morphogenetic abnormalities. Animals expressing Q108 under the control of dppblk are missing entire third legs, have defects in head capsule evagination and development of the mesothorax (notum) is highly abnormal. This aberrant morphology was seen in all dying pupae dissected (n > 10). In contrast, expression of the same Q108 with the myc/flag epitope attached led to a dramatic reduction in severity of the phenotype. One line showed 100% survival and exhibited none of these phenotypes. In another line, viability was increased from 0 to 84% and, of those surviving animals, only 12% exhibited a moderate thoracic phenotype that was less severe. Thus, imaginal disc epithelia can also be sensitive to polyQ toxicity and toxicity is reduced by additional amino acids. Interestingly, dppblk drives high levels of expression in the polytene cells of the salivary glands and we see clear evidence of expression and accumulation of the polyQ peptides in these cells; however, no cellular degeneration is apparent in any of the expressing cells (Figs 3A and 4). The polytene cells of the

Human Molecular Genetics, 2000, Vol. 9, No. 1

17

Figure 3. Percent lethality and phenotypic characterization of GAL4 lines expressing polyQ repeat peptides. (A) Epitope-tagged and non-epitope-tagged polyQ repeat peptide-containing lines were crossed to elav, sev, gmr and dppblk-GAL4 lines and % lethality and external phenotype of progeny were scored. (B) Toluidine blue-stained sections through the eyes of adult wild-type flies and flies containing either the 108Q peptide or the 108Q + epitope tag peptide driven by elav-GAL4, {w; P[w+, elav-GAL4]/+; P[w+, UASQ 108]/+}. Wild-type ommatidia show regular trapezoidal arrangement of seven visible rhabdomeres, whereas many ommatidia expressing Q108 have only one or two rhabdomeres remaining (see Fig. 1 for normal ommatidia structure). These eyes are also characterized by many large holes and loss of cell integrity in the tissue. Ommatidia expressing Q108myc/flag have 4–6 rhabdomeres remaining and also exhibit loss of tissue integrity, but to a lesser extent than Q108. (C) External appearance of eyes from adult flies expressing peptides under control of the gmr-GAL4 chromosome. Dissecting light micrographs of flies expressing gmr-GAL4 alone and UAS108Q under the gmr promoter. The latter shows complete lack of pigment, suggesting degeneration of pigment cells, and a disordered eye phenotype that decreases in severity when the epitope tag is added (data not shown).

salivary gland replicate their DNA and grow to a large size by cell enlargement rather than by cell division. Thus, the polyQ peptides have several days to accumulate as they are being chronically expressed in these cells by the dppblk driver. Despite this extended exposure, no degeneration is observed.

Other neuronal cell types of the peripheral nervous system (PNS) also exhibit differential sensitivity to expanded polyQinduced toxicity (Fig. 3A). When Q108 peptides are expressed by the sev driver, lethality is almost complete (only two survivors with no defects from >300 survivors expected). However,

18

Human Molecular Genetics, 2000, Vol. 9, No. 1

Figure 4. Overexpression of modified dsh transgenes by UAS/GAL4. Dsh wt, Q and 108Q under UAS control were crossed to indicated GAL4 drivers and scored for % lethality and phenotype of survivors. The lethality is a combination of polyQ-induced cytotoxicity and lethal developmental defects caused by ectopic activation of Wg signaling. Crosses were of the form UAS–Transgene/Balancer  GAL4 line/Balancer. A quarter of the offspring are expected to contain both the UAS and GAL4 chromosomes. The percent lethality was calculated by: 1 – non-balancer survivors/(both balancer types/2).

when the myc/flag epitope is appended to the expanded polyQ, lethality caused by the sev-GAL4 driver is reduced such that between 34 and 100% of animals survive and these animals lack sensory bristles. These large bristles, termed macrochaetae, are sensory hair cells produced following two divisions of a sensory mother cell that will yield a neuron, a glia, a sensory hair cell and a socket cell. Only the bristle phenotype is detected, whereas the socket cell is unaffected. Thus, some but not all neuronally derived cells are sensitive to expanded polyQ-induced toxicity. We attempted to examine whether the Q108 peptides without the tag shows the same loss of sensory bristles; however, these flies do not survive long enough to examine the bristle phenotype. Progressive degeneration Finally, progressive neuronal toxicity is suggested both by the reduced viability observed with chronic expression of the Q108 peptide in the nervous system by elav-GAL4 and by an early death seen in surviving animals. Immediately after flies expressing Q108 emerge from the pupal case they begin to die so that within 2 weeks, no surviving adults remain (Fig. 5). In contrast, normal flies show a life span of 2–3 months in our hands. The epitope-tagged Q108 peptides also show an early death phenotype, although length of survival is increased compared with the polyQ alone (50% death at 26 versus 7 days, respectively). This early-death phenotype is consistent with a progressive neuronal dysfunction mediated by the presence of the expanded repeat peptide. We have looked for evidence of progressive degeneration in the eye. To this end, we examined eyes from gmr-GAL4:UAS Q22 and Q108 from 1 to 30 days of age. Although we did not see evidence of progressive degeneration, the animals already exhibited extensive degeneration by the time they emerged as adults from the pupae. In these strains, the peptide-induced toxicity seems to be sufficiently severe at the outset that there is no apparent room for further degeneration.

Figure 5. Early and progressive lethality of flies expressing 108Q chronically in the CNS and PNS. Flies are of the genotype w; P[w+, elav-GAL4]/+; P[w+, UASQX]/+. Open circles, flies expressing the Q22 peptide (n = 103); gray circles, 108Q peptide with the epitope tag (n = 100); black circles, 108Q peptide (n = 138) without tag. Flies were passed to fresh food vials daily and the number of dead flies recorded until none remained.

Protein context is important for polyQ repeat toxicity A second objective of this study is to determine the effect of expanded polyQ repeats on the activity of associated proteins. Several proteins in Drosophila and other organisms contain repeated sequences that encode polyQ tracts. In particular, the Drosophila Dishevelled (Dsh) protein, which is required for transduction of the Wnt oncogene signal in embryonic and adult development (30,31), contains 28 glutamines near the Nterminus of the protein, analogous to the polyQ tract found in the N-terminal portion of the huntingtin protein (32). Like the HD gene (33), the dsh gene is ubiquitously expressed, including widespread expression in the brain and PNS of both flies (31) and mammals (34). Therefore, dsh represents an excellent paradigm to study the effects of expanded polyQ tracts on protein structure and function. We engineered dsh

Human Molecular Genetics, 2000, Vol. 9, No. 1

19

Figure 6. Rescue of lethality by dsh constructs. (A) Lines carrying modified dsh transgenes were crossed to a null allele of dsh and scored for percent rescue of lethality and phenotype. Rescue was tested by crossing y w dshV26/FM7 females to w/Y; P[dshxQ, w+]/P[dshxQ, w+] males. Percent rescue was calculated by dividing the number of non-FM7 sons surviving by the number of non-FM7 daughters. (B–E) Phenotypic characteristics of survivors from several crosses. When Dsh with 27Q is present, animals are fully rescued and wild-type [e.g. see leg in (E)]. When Dsh108 is the only source of active Dsh, phenotypes include notched wings (B) and disordered bristles (C), polarity defects characteristic of loss of Dsh and or Wingless function. Additional phenotypes include bifurcated legs (E) typical of loss of Wg activity (31). Thus, DshQ108 reduces but does not eliminate Dsh function.

genes to contain 0, 22 and 108 polyQ repeats (Fig. 6) and these were introduced into the germline by transformation. To test the effect of the polyQs on Dsh activity, we measured the ability of the engineered transgenes to rescue lethal mutants of dsh. In these experiments, the dsh transgene is under the control of a native, but truncated, dsh promoter and therefore is expressed ubiquitously at a moderate level. In this test, Dsh with a perfect run of 27 glutamines is able to fully rescue dsh mutants and is indistinguishable from the native dsh gene (data not shown). Complete removal of the polyQ tract does not reduce the rescuing ability of the Dsh protein much, if at all, although mild wing and bristle phenotypes suggest that this modified Dsh has somewhat reduced function (Fig. 6A). In contrast, expansion of the polyQ tract to Q108 leads to a marked reduction in rescuing ability, showing an average of

only 24% rescue of lethality (Fig. 6A) and the surviving animals show phenotypes similar to those produced by reductions in Dsh activity. For example, one sees bifurcated third legs, notched wings and aberrant bristle polarity (Fig. 6B–E). Thus, although deleterious, to a much greater degree than complete removal of the polyQs, expanded polyQ reduces but does not completely compromise protein function. We have shown above that polyQ alone is toxic and that expanded polyQ compromises but does not eliminate Dsh function. We next asked whether high level misexpression of Dsh with modified polyQs exhibited changes in Dsh-mediated abnormalities and/or exhibited polyQ-induced toxicity similar to that observed with the polyQ peptides alone. We produced dsh transgenes driven by the UAS promoter in which the polyQ region was left intact, removed or expanded to Q108

20

Human Molecular Genetics, 2000, Vol. 9, No. 1

(Fig. 2B) and then drove expression of these transgenes with GAL4 drivers. Overexpression of dsh itself can lead to developmental defects and lethality due to ectopic activation of the Wingless pathway; thus we compared the effects with those of wild-type dsh overexpression. In all cases examined, the pathology of the expanded polyQ was altered when associated with the Dsh protein. Specifically, in surviving flies expressing dsh constructs, all of the associated phenotypes resemble those caused by overexpression of dsh and do not resemble those caused by polyQ-mediated toxicity. For example, when expressed chronically in the nervous system (elav), an expanded polyQ peptide produces a dominant lethality of between 80 and 100% whereas expression of the same expanded polyQ in the context of the Dsh protein shows an average of only 20% lethality (Fig. 4). This compares very well with the range of lethality observed when native dsh is expressed by the same promoter. This effect is most clearly demonstrated when dsh is overexpressed by dppblk-GAL4. When dpp is used to drive the expression of a native dsh transgene, toxicity is almost 100% with the dying pupae showing short, thick legs and margin bristles in the wing blades (Fig. 4). In contrast, when dsh containing an expanded 108Q repeat is expressed, survival is 100% and there is no phenotype other than a mild wing vein phenotype. Thus, addition of expanded polyQs reduces the effect of ectopic Dsh activity, presumably by its deleterious effect on protein function; the combination of the Dsh protein with Q108 reduces the toxicity of the polyQ from nearly 100 to almost 0%, with no overlap in phenotype. Similar results were obtained for other drivers (data not shown). Subcellular localization of polyQ transgenes We attempted to determine the subcellular localization of expanded polyQ peptides using a variety of antibodies. When Dsh with 27 or 108 polyQ repeats was expressed in the dppblk pattern in discs (Fig. 7, top panels), staining (green) was found at the apical membrane of discs and in punctate patterns in the cytosol (Fig. 7, middle panels). In the salivary gland cells, staining was cytosolic and punctate (Fig. 7, bottom panels). This punctate, membrane-associated staining is characteristic of Dsh and has been described previously (35). The DshQ108 staining pattern is indistinguishable from the DshQ27 pattern. The similarity in cytosolic staining patterns for the Dsh transgene with 27 or 108 glutamines indicates that the presence of the expanded repeat is not sufficient to drive a protein into the nucleus; neither diffuse nor punctate staining is found in the nucleus for any UAS-DshQ108-expressing tissue examined. The subcellular distribution of polyQ peptides with the myc/ flag epitope was different than the Dsh polyQ proteins and much more difficult to detect. When the epitope-tagged Q108 peptide is expressed in the dppblk pattern in discs (Fig. 7, top panels), its cellular localization is punctate, presumably through expanded polyQ-mediated aggregation, with some apparent nuclear and perinuclear localization (Fig. 7, middle panels). In salivary glands, staining of the tagged Q108 appears to be exclusively nuclear and non-diffuse as it is excluded from the regions of the nucleus occupied by the polytene chromosomes (Fig. 7, bottom panels). Thus, expanded polyQs with an epitope tag can enter the nucleus

Figure 7. Subcellular localization of polyQ peptide transgenes and Dsh transgenes. UAS-transgenes were expressed by the ptc-GAL4 driver line {w; P[w+, ptc-GAL4]/+; P[w+, UAS(polyQ)]/+}. ptc-GAL4 expresses in a pattern similar to dppblk-GAL4 (27). Tissues were fixed and stained with various antibodies as described in Materials and Methods. The first two columns were stained with anti-Dsh antibodies. Confocal sections of wing discs (top row), close-ups of wing discs (middle rows) and views of the salivary glands (bottom row) from late third instar larvae are shown. Dsh protein with either short (Q27) or long (Q108) polyQ stretches localizes exclusively to the cytosol in both salivary gland cells and imaginal disc cells (green). Note the dark, nonstaining nuclei in the optical cross section of the wing disc (middle row) and the salivary glands (bottom row). Green fluorescence is either anti-Dsh (left two columns) or anti-flag M2 in the myc/flag tagged Q108 peptide column (right). In the wing disc, the Q108 myc/flag peptide appears to form larger aggregates than does DshQ108. Some of these can be found in the cytosol and others appear nuclear and/or perinuclear within the limits of microscopy. However, in the salivary gland, the Q108 myc/flag peptide localizes exclusively to the nucleus. Red fluorescence represents stain for filamentous actin, and yellow fluorescence represents overlap of green and red. DAPI staining was used to identify nuclei (data not shown).

with the ratio of nuclear to cytoplasmic material being cell type-specific. We were unable to detect any staining with the Q22 repeat peptides, either with or without epitope tags. Presumably, the extremely small size of the peptide renders it difficult to detect in fixed tissue or by western blotting. The Q22 coding sequence was detected in genomic DNA samples, indicating that the appropriate transgene had integrated. Similarly, it was extremely difficult to localize the Q108 peptide without the epitope tag by immunofluorescence. We tried several polyclonal antibodies which preferentially recognize the expanded polyQ repeat-containing huntingtin exon 1 as well as the IC2 monoclonal antibody raised against Tata binding protein (36) which preferentially recognizes expanded polyQ repeats. Only the IC2 antibody produced faint signals and showed very weak nuclear staining in salivary glands and punctate staining in wing discs (data not shown). DISCUSSION Understanding the pathogenesis of late-onset triplet repeat diseases such as HD has been complicated by the presentation

Human Molecular Genetics, 2000, Vol. 9, No. 1

of an array of shared features together with a number of unique features (1,4,21). For most of the disorders, the causative genes have unknown function, show no similarity to each other apart from the CAG repeat and are often expressed in a wide range of tissues. One model to explain the diverse features of these diseases would be that the expanded polyQs create a series of unique gain of function activities by modifying the action of each protein in which they occur. An alternative hypothesis is that pathology is due to intrinsic cytotoxicity of long polyQ chains themselves. We have developed a general Drosophila model of expanded polyQ pathogenesis to distinguish between these hypotheses. Our results suggest that polyQ chains are intrinsically toxic in a strongly cell type-dependent manner. Further, we find that this intrinsic toxicity is modified by the inclusion of other amino acids. Therefore, the differential clinical presentation of diseases may be partially due to the differing ability of the associated proteins to ameliorate the otherwise strong toxicity of the polyQs themselves. In addition, the results are consistent with a role for nuclear localization and/or protein processing in pathogenesis. Long polyQ peptides induce neuronal degeneration Expression of long polyQ peptides results in neuronal degeneration of the photoreceptor neurons of the eye of Drosophila whereas short polyQ peptides have no effect. Chronic expression in the central nervous system (CNS) and PNS (elav) leads to extensive early lethality in the larval stages. In addition, the few animals that survive to adult stages while expressing polyQ in the CNS and PNS show a significantly shortened life span (1 versus ~12 weeks). Expression under another promoter (sev) leads to degeneration of the sensory bristles. Thus, in several tests, polyQ peptides cause a neuronal degeneration similar to that seen in HD and related diseases. PolyQ-induced toxicity is cell type dependent Despite the extensive impact of polyQ on many neuronal cell types, not all neuronal cells are sensitive. Whereas the photoreceptor neurons of the eye are highly sensitive, the mechanosensory neurons that produce the hairs of the eye are not affected, despite the fact that both the elav and gmr promoters drive expression in these cells. One cannot conclude that simply all mechanosensory hair cells are resistant, since the sev promoter leads to loss of the large sensory bristles in the thorax. Cytotoxicity is not limited to neurons as both neuronal and non-neuronal tissues show cell death. For example, the nonneuronal pigment cells of the eye are very sensitive to degeneration, as they are completely lost when the gmr promoter is driving expression of expanded polyQ, as are most recognizable cell types in the eyes of the flies (data not shown). Expression driven by the dppblk promoter leads to almost complete lethality and yet there is no expression in neuronal tissues. Even with this extensive lethality, degeneration is not immediately evident in the imaginal discs in the larvae. Nevertheless, adult structures, that will arise 5 days later after metamorphosis of these discs, are clearly affected (e.g. loss of third legs, dysmorphology of the thorax, etc.). Notably, expression of a 78Q-containing fragment of ataxin-3 under the same dppblk promoter did not cause any obvious phenotype

21

(24). In those experiments, the 78Qs were flanked by 55 amino acids and an N-terminal hemagglutinin tag. Whether the altered toxicity is due to the shorter repeat length (78 versus 108) or to the ameliorating effect of the adjacent amino acids is not clear. The dppblk promoter also drives high levels of expression in the salivary gland, although there is no evidence of degeneration of these cells. It appears that salivary gland cells are resistant to the cytotoxic effects of polyQ despite the fact that these cells do not divide but grow by cell enlargement and in the process have ample time to accumulate and be exposed to polyQ and any cytotoxic effects that might generate. PolyQ cytotoxicity is modified by protein context The cytotoxic impact of expressing Q108 peptides is dramatic. However, when 26 amino acids in the form of a myc/flag epitope are added to the peptide, there is a profound reduction in toxicity. In every case examined, the addition of these 26 amino acids reduced lethality dramatically, e.g. from 100% lethality to near zero lethality with the elav or dppblk drivers. Furthermore, the severity of the phenotypes of surviving animals was reduced with the tagged peptide. When the same sized polyQ peptide is incorporated into a 69 kDa protein (Dsh), the effect is even further reduced. Thus, the toxic effect of Q108 can be profoundly altered by the inclusion of additional amino acids. The mechanism for the reduction in polyQ toxicity by additional amino acids is unclear. The epitope tag may increase the solubility or alter the pI of the protein or decrease the rate and extent of aggregation. In contrast, the Dsh context may confer a different protein processing pathway than occurs for the disease proteins. PolyQs reduce but do not eliminate protein function One hypothesis to account for expanded repeat effects is that expanded repeats either eliminate protein function or cause a protein to gain an aberrant function. Expanded repeats have been shown to compromise androgen receptor function (SBMA) (37) in a manner characteristic of that particular loss of function, whereas the repeat does not appear to compromise huntingtin function (38). To test this, we engineered the Dsh protein to have expanded polyQ, a normal-range polyQ or no polyQ, and used percentage rescue of a dsh null allele as a measure of the effect on Dsh protein activity. Even with a minimal promoter, native Dsh and Dsh with the polyQs deleted were both capable of effecting nearly complete rescue, although the Dsh without polyQs exhibited a small percentage of animals with polarity and patterning defects suggestive of reduced Dsh function. In contrast, Dsh with Q108 was very poor at rescuing dsh null alleles and those animals that were rescued uniformly exhibited polarity and patterning defects similar to loss of Dsh function. Although low level expression of dsh is capable of rescuing null alleles, overexpression of dsh can activate the Wingless signaling pathway in the absence of ligand. Thus, when expressed under various promoters, dsh is often lethal. However, Dsh containing Q108 is always less deleterious than native Dsh implying that addition of the polyQs reduces, but does not eliminate, Dsh function and that addition of Q108 does not cause a dominant gain of function that is cytotoxic.

22

Human Molecular Genetics, 2000, Vol. 9, No. 1

These observations can help to explain the genetics of HD (39) and related diseases. A single copy of expanded repeat protein would lead to progressive degeneration due to effects of the polyQ itself. Homozygotes for expanded polyQ repeats would not be worse than heterozygotes, since at least some protein function remains which is sufficient for viability. Different clinical presentation would be caused by the different effects that the associated proteins confer. Some patterns of polyQ expression exhibit progressive deterioration In most cases, we do not see progressive deterioration of affected tissues. However, this is likely due to the fact that the effect observed with most drivers is so severe at the outset that it cannot get much worse. However, in some settings, such as chronic expression in the neurons of the CNS and PNS, we do see early death that mimics the progressive deterioration and early death seen in human HD. Subcellular localization The partitioning between the nucleus and cytoplasm is complex as polyQ peptides appear to be present in both compartments. Nuclear localization has been described as an important component of polyQ repeat pathogenesis (1,14,15). The results presented here support a role for protein processing, leading to aggregation and/or nuclear localization. When an expanded repeat is engineered into an endogenous Drosophila gene, dsh, no expanded polyQ repeat toxicity is observed. This is in marked contrast to the hprt mouse model described by Ordway et al. (22). In this model, an endogenous mouse gene, hypoxanthine-guanine phosphoribosyl transferase (hprt), which normally does not contain a CAG repeat, was engineered to contain an expanded repeat within the fulllength coding region. Mice expressing this modified hprt develop a neurologic disorder with corresponding nuclear localization of a normally cytosolic protein and the presence of nuclear inclusions. The most obvious difference between the two models is the exclusively cytosolic localization of DshQ108. In this context, an increased number of polyQ repeats is not sufficient for nuclear localization or any apparent aggregation. This failure of DshQ108 to cause cytotoxicity when overexpressed is consistent with a lack of altered subcellular distribution relative to wild-type Dsh. Hprt is a 48 kDa protein, which is below the 50 kDa nuclear pore exclusion limit (4); therefore, it may be sufficiently small to enter the nucleus and produce a phenotype, whereas there is evidence that the larger disease proteins are cleaved to smaller fragments that may be a prerequisite for nuclear entry. It is unclear whether expanded polyQs themselves promote nuclear entry or whether the expanded polyQs are responsible for retention of the mutant protein in the nucleus. When an expanded polyQ repeat is expressed in Drosophila as an epitope-tagged peptide, nuclear localization is observed in several tissues. Nuclear localization alone, however, is not sufficient for pathogenesis as no phenotype is observed in the salivary gland, where the peptide is exclusively nuclear. Whether the absence of nuclear inclusions in this tissue plays a role in the lack of a phenotype or whether salivary gland is not susceptible to the toxic effects of increased numbers of polyQs is unclear. In the developing wing disc, where cytosolic and

nuclear aggregates are visible, a developmental phenotype was observed. It will be of interest to determine whether driving Dsh108Q into the nucleus, by engineering in a nuclear localization signal, is sufficient to produce a polyQ-derived cytotoxicity. Comparison with other models Several mouse models of triplet repeat diseases have been described. In each case, these models have expressed expanded polyQ peptides in the context of all or part of an affected human disease gene (1,7–10) and have produced different levels of neurodegeneration. A Caenorhabditis elegans model has also been described (40). In Drosophila, two different models have been described, also expressing expanded repeats in a disease gene context. Using targeted overexpression of truncated ataxin-3, Warrick et al. (24), demonstrated nuclear inclusion formation and late-onset neuronal degeneration, whereas Jackson et al. (25) expressed a portion of human huntingtin in the eye using direct regulation by the gmr enhancer and showed degeneration of photoreceptor neurons and nuclear localization and aggregation. The range of severity in these models directly correlates with the length of associated protein sequences. If compared for gmr expression alone, the Q108 peptide (repeat + 10 amino acids) alone produces the most severe effects, followed by the epitope-tagged Q108 peptide (repeat + 36 amino acids) and truncated ataxin-3 (Q78 + 64 amino acids), and the least severe effects were produced by an N-terminal fragment of huntingtin, which encodes the longest protein sequence (Q120 + 142 amino acids). The polyQ peptide, which has a repeat size within the range of juvenile onset disease, may be producing a widespread ‘glutaminopathy’ (14). The range of cells affected and cytotoxicity of this glutamine-mediated pathology is reduced when surrounded by protein sequence, such as an epitope tag or disease protein fragment, and increased with longer polyQ repeat lengths. Juvenile-onset HD, for instance, is caused by very large expansions of polyQ repeats, from 65 to >100 repeats, and more widespread neuronal degeneration is observed in brain tissue from these patients as compared with that observed in brain tissue from patients with medium range repeats (41,42). In addition, there is greater overlap with other diseases; juvenile HD patients have Purkinje cell involvement, characteristic of SCAs, not normally observed in adult-onset HD (41). Concluding remarks The experiments reported here suggest that a combination of expanded polyQ-mediated cytotoxicity and protein context directs the neuronal-specific pathogenesis observed in the polyQ repeat neurodegenerative diseases. The dominant genetics may be explained by the pathologic effects of an expanded polyQ repeat alone. It is likely that the mutant protein, containing expanded polyQs, can still execute sufficient levels of their normal function. The cell type specificity suggests that there may be intrinsic mechanisms in each cell which make either some cells resistant or some cells sensitive. In summary, the polyQ peptide Drosophila model presented here provides a powerful system to screen for agents that interrupt the dominant cytotoxic effects of expanded

Human Molecular Genetics, 2000, Vol. 9, No. 1

repeat polyQs, independent of unique protein interactions mediated by the specific disease genes. MATERIALS AND METHODS Synthesis of dsh and polyQ peptide constructs dsh constructs. Full-length genomic dsh with 184 bp of a minimal promoter element (31) was cloned as a BglII–EcoRI fragment into the BamHI–EcoRI sites of a modified Bluescript KS+ vector lacking the unique HindIII site. In vitro mutagenesis (Transformer Mutagenesis kit; Clontech, Palo Alto, CA) was used to create HindIII sites immediately adjacent to an endogenous sequence coding for a 28 amino acid polyQ tract interrupted by two histidines, thus producing a HindIII cassette containing the repeat (primers: Dsh MutA, 5-CCAACAGGAAGCTTCAGCAGCAACAGC-3; Dsh MutB, 5-CAACAGCAGCAGAAGCTTCAGCCTGTCCAGCTGG-3). The creation of these HindIII sites resulted in no change in predicted amino acid sequence of the N-terminus and a QK and VL change at the C-terminus adjacent to the repeat. This construct was tested in rescue experiments and found to have activity equivalent to wild-type dsh, therefore was subsequently used as the wild-type control in crosses. The following constructs were derived from this parental construct by deletion/insertion of HindIII cassettes: DshQ, deleted for the HindIII cassette; Dsh27Q, containing a perfect stretch of 27 CAGs substituted for the endogenous repeat; Dsh108Q, containing a stretch of 108Qs substituted for the endogenous repeat. The 27Q cassette was created by using a primer with 27 CAGs and HindIII linkers at the ends [primer: 5-CCACCAAGCTT(CAG)27AAGCTTCCACC-3]. A fill-in primer (5-GGTGGAAGCTTCTGC-3) was subsequently annealed to the CAG repeat oligonucleotide and used for second strand synthesis using Klenow to create a double-stranded DNA fragment. This fragment was then digested with HindIII and ligated into DshQ. The 108Q cassette was created by in vitro mutagenesis of a mouse–human hybrid huntingtin exon 1 construct, which contains an expanded CAG repeat derived from a juvenile-onset HD patient (kindly provided by Scott Zeitlin, Columbia) to create a HindIII cassette [primers: HD94HL-2, 5-GATGAAGGCCTTCGAGTCCCTCAAGCTT(CAG)6-3; HD94HR-2, 5-CAGCAGCAGAAGCTTCAGCCGCCACCG(CCG)6CCTCCTCAG-3]. This CAG repeat was subsequently found to contain a PCR error that creates a substitution of an arginine for a glutamine. This results in a cassette containing 43QsR-65Qs; however, this repeat range has been determined to cause polyQ-mediated toxicity in several cell model systems and in a mouse model system (S. Zeitlin, personal communication). The dsh constructs were then cloned via EcoRI/XbaI into a Drosophila expression vector (pCasper) for use in subsequent transformations. UAS-dsh constructs were produced using in vitro mutagenesis to create a new EcoRI site 25 bp upstream of the AUG in dsh in DshQ, Dsh27Q and Dsh108Q (primer: 91-2RI, 5-GTTTTCCCGTGGAATTCCGCAGTC-3]. These constructs were then subcloned into the EcoRI/XbaI sites of pUAST (26). Peptide constructs. Vectors expressing polyQ peptides were constructed by inserting the HindIII cassettes described above into a UAS expression vector (pUAST AUG myc/flag; kindly provided by Drs Mike Boedigheimer and Peter Bryant, University of California, Irvine, CA) containing an AUG immediately

23

upstream of a single BglII site and a myc/flag tag downstream. In vitro mutagenesis was used to create BglII sites adjacent to the HindIII sites. One set of constructs contained a stop codon to eliminate the C-terminal myc/flag epitope tags (primers: AUGH3Bgl25B, 5- GGGTACCGGGCCCCCCCTCGAGGTCGAAGATCTCGAAAGCTTCAGCAGCAG-3; AUGH3Bgl23B, 5-CAGCAGCAGAAGCTTAGATCTAATTCCTG- CAGCCCGGGGG-3; AUGH3Bgl23Bstop, 5-CAGCAGCAGAAGCTTTAAAGATCTAATTCCTGCAGCCCGGGGG-3). Fly stocks and crosses Germ-line transformants were generated as described by Rubin and Spradling (43) and Spradling and Rubin (44) with modifications from Park and Lim (45). The following GAL4 stocks were obtained from the National Drosophila Stock Center (Bloomington, IN): w*; P{w + mC = GAL4ninaE.GMR}12 (no. 1104) (46) and P{ry + t7.2 = GAL4Hsp70.sev}2/CyO; ry* (no. 2023). Other GAL4 stocks were received from Dr Michael O’Connor (University of Minnesota): w; P{w+; elav-GAL4}/CyO} (46), w; dppblk; P{GAL4dpp.blk1 w + mW.hs}39B2/TM6B (47) and w; P{w + mW.hs = GawB}ptc-559.1 (28). GAL4 responsive y w; P{w+; UAS–GFP-S65T}T2-1/CyO,y+ flies were a gift from Dr Daisuke Yamamoto (Mitsubishi Kasei, Tokyo, Japan). Dsh mutant lines were previously described (31). Flies were grown on cornmeal molasses medium at 25C. Histology Adult heads (mutant, wild-type Oregon R) for light microscopy were fixed in 1% osmium tetroxide, 1% glutaraldehyde and phosphate-buffered saline (PBS) for 30 min on ice, followed by a fix in 1% osmium tetroxide and PBS for 2 h on ice. Tissues were dehydrated through a graded series of ethanol and embedded in Durcupan (Fluka, Buchs, Switzerland). Serial sections were cut at 2 m, stained with 1% toluidine blue and 1% borax and mounted in DPX (Electron Microscopy Sciences, Fort Washington, NY). Antibody staining Wandering third instar larvae were dissected in ice-cold PBS, fixed in 4% formaldehyde and PBS for 30 min at room temperature, washed briefly in PBS and blocked in 3% BSA, 0.5% NP-40 and PBS for 2 h, all at room temperature. Primary antibodies were rat anti-Dsh (49) at 1:500 dilution, mouse antiFLAG M2 (Sigma, St Louis, MO) at 1:1000 and mouse IC2 (provided by Dr Jean-Louis Mandel, IGBMC, Strasbourg, France) at 4:1000. We were not successful in detecting the expanded polyQ peptide with mouse anti-HD (provided by Dr Danilo Tagle, National Institutes of Health, Bethesda, MD) or rabbit anti-CAG 53B (provided by Dr Erich Wanker, Max Planck Institute, Germany) antibodies. Secondary antibodies were FITC-conjugated goat anti-rat IgG (Zymed, San Francisco, CA) or goat anti-rabbit IgG (Molecular Probes, Eugene, OR) at 1:100. Tissues were incubated with primary antibodies overnight at 4C and with secondary antibodies for 1–2 h at room temperature. In some cases, an intermediate unlabeled rabbit anti-mouse IgG antibody (Jackson

24

Human Molecular Genetics, 2000, Vol. 9, No. 1

ImmunoResearch, West Grove, PA), at 1:200, was used against the primary antibody to further amplify the signal. Rhodamine–phalloidin (Molecular Probes) was added at 1:100 for detection of filamentous actin in order to outline cells, and DAPI was added at 1:1000 to label chromatin. These stains were added to the tissues along with the secondary antibody. The samples were analyzed with a BioRad (Hercules, CA) MRC 1024 scanning confocal microscope. ACKNOWLEDGEMENTS This work was supported by grants from the Hereditary Disease Foundation to L.M.T. and J.L.M., NIH Research grants RO1 HD36081 and RO1 HD36049 to J.L.M. and PHSNIH Research grant NS25631 to John J. Wasmuth/L.M.T. Co-PI. H.T. was supported in part by a PHS training grant NO 5T32 GM07311-17. The authors gratefully acknowledge the resources of the National Drosophila Stock Center (Bloomington, IN) and appreciate the assistance received from R. Taireh, G. Jackson, L. Zipursky, E. Wanker, D. Tagle, J.-L. Mandel and S. Zeitlin. This manuscript is dedicated to the memory of John J. Wasmuth whose enthusiasm and generosity were inspiring. REFERENCES 1. Reddy, H.P., Williams, M. and Tagle, D.A. (1999) Recent advances in understanding the pathogenesis of Huntington’s disease. Trends Neurosci., 22, 248–255. 2. Zhuchenko, O., Bailey, J., Bonnen, P., Ashizawa, T., Stockton, D., Dobyns, W.B., Subramony, S.H., Zoghbi, H.Y. and Lee, C.C. (1997) Autosomal dominant cerebellar ataxia (SCA6) associated with small polyglutamine expansions in the a1A-voltage-dependent calcium channel. Nature Genet., 15, 62–69. 3. Kakizuka, A. (1998) Protein precipitation: a common etiology in neurodegenerative disorders? Trends Genet., 14, 396–402. 4. Ross, C.A. (1997) Intranuclear neuronal inclusions: a common pathogenic mechanism for glutamine-repeat neurodegenerative diseases. Neuron, 19, 1147–1150. 5. Duyao, M.P., Auerbach, A.B., Ryan, A., Persichetti, F., Barnes, G.T., McNeil, S.M., Ge, P., Vonsattel, J.-P., Gusella, J.F., Joyner, A.L. et al. (1995) Inactivation of the mouse Huntington’s disease gene homolog Hdh. Science, 269, 407–410. 6. Zeitlin, S., Liu, J.-P., Chapman, D.L., Papaioannou, V.E. and Efstratiadis, A. (1995) Increased apoptosis and early embryonic lethality in mice nullizygous for the Huntington’s disease gene homologue. Nature Genet., 11, 155–163. 7. Burright, E.N., Clark, H.B., Servadio, A., Matilla, T., Feddersen, R.M., Yunis, W.S., Duvick, L.A., Zoghbi, H.Y. and Orr, H.T. (1995) SCA1 transgenic mice: a model for neurodegeneration caused by an expanded CAG trinucleotide repeat. Cell, 82, 937–948. 8. Hodgson, J.G. et al. (1999) YAC mouse model for Huntington’s disease with full-length mutant huntingtin, cytoplasmic toxicity, and selective striatal neurodegeneration. Neuron, 23, 181–192. 9. Ikeda, H., Yamaguchi, M., Sugai, S., Aze, Y., Narumiya, S. and Kakizuka, A. (1996) Expanded polyglutamine in the Machado–Joseph disease protein induces cell death in vitro and in vivo. Nature Genet., 13, 196–202. 10. Mangiarini, L., Sathasivam, K., Seller, M., Cozens, B., Harper, A., Hetherington, C., Lawton, M., Trottier, Y., Lehrach, H., Davies, S.W. et al. (1996) Exon 1 of the HD gene with an expanded CAG repeat is sufficient to cause a progressive neurological phenotype in transgenic mice. Cell, 87, 493–506. 11. Davies, S.W., Turmaine, M., Cozens, B.A., Difiglia, M., Sharp, A.H., Ross, C.A., Scherzinger, E., Wanker, E.E., Mangiarini, L. and Bates, G.P. (1997) Formation of neuronal intranuclear inclusions underlies the neurological dysfunction in mice transgenic for the HD mutation. Cell, 90, 537–548.

12. DiFiglia, M., Sapp, E., Chase, K.O., Davies, S.W., Bates, G.P., Vonsattel, J.P. and Aronin, N. (1997) Aggregation of huntingtin in neuronal intranuclear inclusions and dystrophic neurites in brain. Science, 277, 1990–1993. 13. Bates, G.P., Mangiarini, L. and Davies, S.W. (1998) Transgenic mice in the study of polyglutamine repeat expansion diseases. Brain Pathol., 8, 699–714. 14. Paulson, H.L. (1999) Human Genetics ’99: trinucleotide repeats. Protein fate in neurodegenerative proteinopathies: polyglutamine diseases join the (mis)fold. Am. J. Hum. Genet., 64, 339–345. 15. Klement, I.A., Skinner, P.J., Kaytor, M.D., Yi, H., Hersch, S.M., Clark, H.B., Zoghbi, H.Y. and Orr, H.T. (1998) Ataxin-1 nuclear localization and aggregation: role in polyglutamine-induced disease in SCA1 transgenic mice. Cell, 95, 41–53. 16. Saudou, F., Finkbeiner, S., Devys, D. and Greenberg, M.E. (1998) Huntingtin acts in the nucleus to induce apoptosis but death does not correlate with the formation of intranuclear inclusions. Cell, 95, 55–66. 17. Gutekunst, C.A., Li, S.-H., Yi, H., Mulroy, J.S., Kuemmerle, S., Jones, R., Rye, D., Ferrante, R.J., Hersch, S.M. and Li, X.-J. (1999) Nuclear and neuropil aggregates in Huntington’s disease: relationship to neuropathology. J. Neurosci., 19, 2522–2534. 18. Li, H., Li, S.H., Cheng, A.L., Mangiarini, L., Bates, G.P. and Li, X.J. (1999) Ultrastructural localization and progressive formation of neuropil aggregates in Huntington’s disease transgenic mice. Hum. Mol. Genet., 8, 1227–1236. 19. Scherzinger, E., Sittler, A., Schweiger, K., Heiser, V., Lurz, R., Hasenbank, R., Bates, G.P., Lehrach, H. and Wanker, E.E. (1999) Selfassembly of polyglutamine-containing huntingtin fragments into amyloidlike fibrils: implications for Huntington’s disease pathology. Proc. Natl Acad. Sci. USA, 96, 4604–4609. 20. Perutz, M.F. (1999) Glutamine repeats and neurodegenerative diseases: molecular aspects. Trends Biochem. Sci., 24, 58–63. 21. Koshy, B.T. and Zoghbi, H.Y. (1997) The CAG/polyglutamine trace diseases: gene products and moleuclar pathogenesis. Brain Pathol., 7, 927–942. 22. Ordway, J.M., Tallaksen-Greene, S., Gutekunst, C.-A., Bernstein, E.M., Cearley, J.A., Wiener, H.W., Dure IV, L.S., Lindsey, R., Hersch, S.M., Jope, R.S. et al. (1997) Ectopically espressed CAG repeats cause intranuclear inclusions and a progressive late onset neurological phenotype in the mouse. Cell, 91, 753–763. 23. La Spada, A.R., Wilson, E.M., Lubahn, D.B., Harding, A.E. and Fischbeck, K. (1991) Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy. Nature, 352, 77–79. 24. Warrick, J.M., Paulson, H.L., Gray-Board, G.L., Bui, Q.T., Fischbeck, K.H., Pittman, R.N. and Bonini, N.M. (1998) Expanded polyglutamine protein forms nuclear inclusions and causes neural degeneration in Drosophila. Cell, 93, 939–949. 25. Jackson, G.R., Salecker, I., Dong, X., Yao, X., Arnheim, N., Faber, P.W., MacDonald, M.E. and Zipursky, S.L. (1998) Polyglutamine-expanded human huntingtin transgenes induce degeneration of Drosophila photoreceptor neurons. Neuron, 21, 633–642. 26. Brand, A. and Perrimon, N. (1993) Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development, 118, 401–415. 27. Spradling, A.C. and Rubin, G.M. (1983) The effect of chromosomal position on the expression of the Drosophila xanthine dehydrogenase gene. Cell, 34, 47–57. 28. Speicher, S.A., Thomas, U., Hinz, U. and Knust, E. (1994) The Serrate locus of Drosophila and its role in morphogenesis of the wing imaginal discs: control of cell proliferation. Development, 120, 535–544. 29. Wolff, T., Martin, K.A., Rubin, G.M. and Zipursky, S.L. (1997) The development of the Drosophila visual system. In Jessell, T.M. and Zipursky, S.L. (eds), Molecular and Cellular Approaches to Neuronal Development. Oxford University Press, New York, NY, pp. 474–508. 30. Klingensmith, J., Nusse, R. and Perrimon, N. (1994) The Drosophila segment polarity gene dishevelled encodes a novel protein required for response to the wg signal. Genes Dev., 8, 118–130. 31. Theisen, H., Purcell, J., Bennett, M., Kansagara, D., Syed, A. and Marsh, J.L. (1994) Dishevelled is required during wingless signaling to establish both cell polarity and cell identity. Development, 120, 347–360. 32. The Huntington’s Disease Collaborative Research Group (1993) A novel gene containing a trinucleotide repeat that is expanded and unstable on Huntington’s disease chromosomes. Cell, 72, 971–983.

Human Molecular Genetics, 2000, Vol. 9, No. 1

33. Strong, T.V., Tagle, D.A., Valdes, J.M., Elmer, L.W., Boehm, K., Swaroop, M., Kaatz, K.W., Collins, F.S. and Albin, R.L. (1993) Widespread expression of the human and rat Huntington’s disease gene in brain and nonneural tissues. Nature Genet., 5, 259–265. 34. Sussman, D.J., Klingensmith, J., Salinas, P., Adams, P.S., Nusse, R. and Perrimon, N. (1994) Isolation and characterization of a mouse homolog of the Drosophila segment polarity gene dishevelled. Develop. Biol., 166, 73–86. 35. Axelrod, J.D., Miller, J.R., Shulman, J.M., Moon, R.T. and Perrimon, N. (1998) Differential recruitment of Dishevelled provides signaling specificity in the planar cell polarity and Wingless signaling pathways. Genes Dev., 12, 2610–2622. 36. Stevanin, G., Trottier, Y., Cancel, G., Dürr, A., David, G., Didierjean, O., Bürk, K., Imbert, G., Saudou, F., Abada-Bendib, M. et al. (1996) Screening for proteins with polyglutamine expansions in autosomal dominant cerebellar ataxias. Hum. Mol. Genet., 5, 1887–1892. 37. Warner, C.L., Griffin, J.E., Wilson, J.D., Jacobs, L.D., Murray, K.R., Fischbeck, K.H., Dickoff, D. and Griggs, R.C. (1992) X-linked spinomuscular atrophy: a kindred with associated abnormal androgen receptor binding. Neurology, 42, 2181–2184. 38. White, J.K., Auerbach, W., Duyao, M.P., Vonsattel, J., Gusella, J.F., Joyner, A.L. and MacDonald, M.E. (1997) Huntingtin is required for neurogenesis and is not impaired by the Huntington’s disease CAG expansion. Nature Genet., 17, 404–410. 39. Wexler, N.S. et al. (1987) Homozygotes for Huntington’s disease. Nature, 326, 194–197. 40. Faber, P.W., Alter, J.R., MacDonald, M.E. and Hart, A.C. (1999) Polyglutamine-mediated dysfunction and apoptotic death of a Caenorhabditis elegans sensory neuron. Proc. Natl Acad. Sci. USA, 96, 179–184.

25

41. Harper, P.S. (ed.) (1991) Huntington’s Disease. Major Problems in Neurology, Vol. 22. W.B. Saunders, London, pp. 141–178. 42. Penny Jr, J.B., Vonsattel, J.-P., MacDonald, M.E., Gusella, J.F. and Myers, R.H. (1997) CAG repeat number governs the development rate of pathology in Huntington’s disease. Ann. Neurol., 41, 689–692. 43. Rubin, G.M. and Spradling, A.C. (1982) Genetic transformation of Drosophila with transposable elements vectors. Science, 218, 348–353. 44. Spradling, A.C. and Rubin, G.M. (1982) Transposition of cloned P elements into Drosophila germ line chromosomes. Science, 218, 341– 347. 45. Park, S. and Lim, J.K. (1995) A microinjection technique for ethanoltreated eggs and a mating scheme for detection of germ line transformants. Dros. Info. Serv., 76, 197–199. 46. Ellis, M.C., O’Neill, E.M. and Rubin, G.M. (1993) Expression of Drosophila glass protein and evidence for negative regulation of its activity in non-neuronal cells by another DNA-binding protein. Development, 119, 855–865. 47. Robinow, S. and White, K. (1988) The locus elav of Drosophila melanogaster is expressed in neurons at all developmental stages. Dev. Biol., 126, 294–303. 48. Staehling-Hampton, K., Jackson, P.D., Clark, M.J., Brand, A.H. and Hoffmann, F.M. (1994) Specificity of bone morphogenetic protein-related factors: cell fate and gene expression changes in Drosophila embryos induced by decapentaplegic but not 60A. Cell Growth Differ., 5, 585–593. 49. Yanagawa, S., van Leeuwen, F., Wodarz, A., Klingensmith, J. and Nusse, R. (1995) The dishevelled protein is modified by wingless signaling in Drosophila. Genes Dev., 9, 1087–1097. 50. Grieder, N.C., Nellen, D., Burke, R., Basler, K. and Affolter, M. (1995) Schnurri is required for Drosophila Dpp signaling and encodes a zinc finger protein similar to the mammalian transcription factor PRDII-BF1. Cell, 81, 791–800.