Glial Cell Lineage Expression of Mutant Ataxin-1 and Huntingtin Induces Developmental and Late-Onset Neuronal Pathologies in Drosophila Models

Background In several neurodegenerative disorders, toxic effects of glial cells on neurons are implicated. However the generality of the non-cell autonomous pathologies derived from glial cells has not been established, and the specificity among different neurodegenerative disorders remains unknown. Methodology/Principal Findings We newly generated Drosophila models expressing human mutant huntingtin (hHtt103Q) or ataxin-1 (hAtx1-82Q) in the glial cell lineage at different stages of differentiation, and analyzed their morphological and behavioral phenotypes. To express hHtt103Q and hAtx1-82Q, we used 2 different Gal4 drivers, gcm-Gal4 and repo-Gal4. Gcm-Gal4 is known to be a neuroglioblast/glioblast-specific driver whose effect is limited to development. Repo-Gal4 is known to be a pan-glial driver and the expression starts at glioblasts and continues after terminal differentiation. Gcm-Gal4-induced hHtt103Q was more toxic than repo-Gal4-induced hHtt103Q from the aspects of development, locomotive activity and survival of flies. When hAtx1-82Q was expressed by gcm- or repo-Gal4 driver, no fly became adult. Interestingly, the head and brain sizes were markedly reduced in a part of pupae expressing hAtx1-82Q under the control of gcm-Gal4, and these pupae showed extreme destruction of the brain structure. The other pupae expressing hAtx1-82Q also showed brain shrinkage and abnormal connections of neurons. These results suggested that expression of polyQ proteins in neuroglioblasts provided a remarkable effect on the developmental and adult brains, and that glial cell lineage expression of hAtx1-82Q was more toxic than that of hHtt103Q in our assays. Conclusion/Significance All these studies suggested that the non-cell autonomous effect of glial cells might be a common pathology shared by multiple neurodegenerative disorders. In addition, the fly models would be available for analyzing molecular pathologies and developing novel therapeutics against the non-cell autonomous polyQ pathology. In conclusion, our novel fly models have extended the non-cell autonomous pathology hypothesis as well as the developmental effect hypothesis to multiple polyQ diseases. The two pathologies might be generally shared in neurodegeneration.


Introduction
Effects of glial expression of mutant proteins (or non-cell autonomous effects) on the pathology have been suggested in amyotrophic lateral sclerosis (ALS) and in spinocerebellar ataxia type 7 (SCA7). Primary mouse spinal motor neurons expressing mutant superoxide dismutase 1 (SOD1) do not provoke motor neuron degeneration, whereas when motor neurons were generated from embryonic stem cells (ESCs), co-culture with primary glial cells from transgenic mice expressing mutant SOD1 induces neurodegenerative changes of co-cultured neurons [1,2]. Furthermore, conditioned media by astrocytes but not fibroblasts, microglias, or cortical neurons induces cell death of motor neurons, suggesting soluble factor(s) mediate the non-cell autonomous effect. These effects in vitro were confirmed also in vivo. Selective Cre-mediated gene excision of mutant SOD1 in astrocytes improved the survival of SOD1 expressing mice [3], while mutant SOD1 expression in cell types other than motor neurons and oligodendrocytes seem to accelerate the onset of motor neuron disease phenotype [4]. Specifically in the case of mutant SOD1 transgenic mice, secreted mutant SOD1 protein from reactive astrocytes might be a mediator of the glial toxicity [5].
In polyglutamine (polyQ) diseases, the La Spada group reported that mutant ataxin-7 in Bergmann glias induces ataxia and neurodegeneration of Purkinje cells in mice [6]. The Orr group also reported that embryonic expression of human ataxin-1 (hAtx1), the causative gene product of SCA1, which interacts with retinoic acid orphan receptor alpha (ROR-a) and disturbs Purkinje cell development, affects the pathology in adulthood [7]. It indicates that the mutant gene expression and its interaction with mediator molecules during development are critical factors for the SCA1 pathology. All these studies suggested that the noncell autonomous effect of glial cells might be a common pathology shared by multiple neurodegenerative disorders.
In this study, to test this hypothesis of non-cell autonomous effects by glial cells as well as by stem/progenitor cells, we generated and analyzed phenotypes of Drosophila models expressing human mutant Htt or Atx1 in the glial cell lineage at different stages of differentiation. In both cases, severe phenotypes are observed in development, locomotive activity and survival of flies. Morphological analyses of these flies revealed severe degeneration of neurons in the brain. These results further support the concept that glial cells expressing mutant proteins provides a non-cell autonomous effect on neurons. In addition, the fly models would be available for analyzing molecular pathologies and developing novel therapeutics against the non-cell autonomous polyQ pathology.

Generation of fly models expressing mutant polyQ proteins in glial cell lineage
We used two Gal4 drivers, repo-Gal4 and gcm-Gal4 to express polyglutamine proteins in glial lineage cells. Glial Cell Missing (gcm) is a transcription factor that regulates glial cell lineage commitment to longitudinal glioblast cells [8]. After the commitment, gcm expression decreases rapidly but sustains until immature glial cells [8] (Fig. 1A). Exceptionally, a small percentage of mature neurons in ventral nerve cord seem to express gcm [9] beside glial cell lineage. Reversed Polarity (repo), a homeobox gene required for the differentiation and maintenance of glia function, is another marker of glial cell lineage [10]. Repo is expressed from glioblasts, immature to mature glial cells but not in neuroglioblasts (Fig. 1B). All glial cells except midline glias express repo [8,10]. GCM protein is known to regulate transcription of repo in glial cells [11,12].
In Figure 1, such differences between gcm and repo expression patterns are summarized. Gcm is mainly expressed around commitment to glial cell lineage. Expression level of repo is from immature to mature glial cells is far higher than that of gcm. Therefore, in brief, we can induce glial cell-specific gene expression by repo driver and stem/progenitor cell-specific gene expression by gcm driver in the glial cell lineage (Fig 1A, B). Existence of different types of cells in glial cell lineage or neuronal cell lineage is also summarized in Table (Fig. 1C).

Behavioral and survival effects of glial cell-lineage expression of mutant huntingtin
We crossed male repo-Gal4 flies balanced by a balancer chromosome (TM3, sb) with UAS-hHtt103Q virgin females because homo repo-Gal4 fly is lethal. The number of F1 flies was not different between repo-Gal4 positive and negative flies ( Table 1), suggesting that embryogenesis of the flies expressing mutant human Htt in glial cells was largely normal. Therefore, we tested lifespan and motor activity of adult repo-Gal4/ UAS-hHtt103Q (repo; hHtt103Q) flies, and found lifespan shortening and abnormal motor activities ( Fig. 2A, B). We also observed spontaneous activity of the flies, which was basically in accordance with the Light-Dark cycle but partially arrhythmic ( Fig. 2C vs 2D). In addition, their spontaneous activities decreased gradually during the test period (Fig. 2D).
We next employed gcm-Gal4 driver for expression of mutant Htt. Because the homo gcm-Gal4 flies were lethal, gcm-Gal4 flies were balanced by a balancer chromosome (Gla). Then we crossed them with virgin female UAS-hHtt103 flies like generation from repo-Gal4 flies. In this case, the birth rate of gcm-Gal4/+; UAS-hHtt103Q/+ (gcm; hHtt103Q) fly was remarkably reduced ( Table 1), suggesting that developmental abnormalities were induced by mutant Htt expression in neuroglioblasts/glioblasts. Although the number of the gcm; hHtt103Q adult flies were small, all the flies were available for behavioral analyses. We found their lifespan to be shortened remarkably ( Fig. 2A). Their anti-gravity climbing activity was not impaired within 24 hrs after eclosion, but they became unable to move within few days (Fig. 2B). Because of the small number of adult flies and their very short lifespan we could not calculate their daily activities.

Pathological effects of glial cell-lineage expression of mutant huntingtin
We next performed morphological analysis of these fly models. Expression of human mutant Htt protein with 103 polyQ repeats (hHtt103Q) was confirmed by immunohistochemistry with N-18 anti-htt antibody (Fig. 3A, N-18). The distribution pattern of the mutant hHtt inclusion bodies in the brain at Day 2 was consistent with that of glial cells (Fig. 3A, repo). We also performed double staining of repo and mutant hHtt and the validity of the repo-Gal4 driver was confirmed ( Developmental effects of glial cell-lineage expression of mutant ataxin-1 No adult fly was obtained in expression of mutant human ataxin-1 (hAtx1-82Q) under the control of gcm-Gal4 or repo-Gal4 driver. Therefore, we tried to determine the stage when the development of the expressers was inhibited. To discriminate the flies expressing hAtx1-82Q, we employed GFP-balancers. As expected, we could distinguish hAtx1-82Q/+ (or Y); gcm-Gal4/+ flies from hAtx1-82Q/+ (or Y); CyO,GFP/+ flies by fluorescence even at larval stage. hAtx1-82Q/+ (or Y);;repo-Gal4/+ and hAtx1-82Q /+ (or Y);;TM3, GFP sb/+ could be also distinguished by GFP.
Interestingly we found a part of pupae of hAtx1-82Q /+ (or Y); gcm-Gal4/+ (gcm; hAtx1-82Q) to have remarkably small heads (Fig. 4B). To quantitatively evaluate the small head phenotype, the ratio between head width and length of pupa was calculated ( Fig. 4B and C). The histogram of the ratio in gcm; hAtx1-82Q Q flies was clearly biphasic (Fig. 4C), and nearly 50% of pupa possess extremely small heads.

Pathological effects of glial cell-lineage expression of mutant ataxin-1
The morphologies of pupal brains were remarkably changed both in gcm; hAtx1-82Q and repo; hAtx1-82Q pupae ( Fig. 5A and B). As shown in Fig. 4C, shrinkage of the head size by gcm-Gal4 driven expression of Atx1-82Q exhibits biphasic distribution. In mild cases, the gross structure of the central nervous system is relatively preserved, whereas the size of each brain structure became 50-60% of normal in length (Fig. 5B). In some parts of the brain, vacuolations were observed in the tissue (Fig. 5B). In severe cases, however, the brain structure was extremely destroyed and the relationship to surrounding head tissues was also distorted (Fig. 5B).
In the repo; hAtx1-82Q pupae, the gross structure of the brain was relatively preserved. Although the number of glial cells stained with anti-repo antibody did not change remarkably, the neuropils became coarse and the number of neurons stained with anti-elav antibody was reduced (Fig. 6B). The reduction of neurons (photoreceptor neurons) was most prominent in retina (Fig. 6B). In addition, anti-elav staining revealed that the number and density of neurons surrounding optic lobes decreased remarkable in the gcm; hAtx1-82Q and repo; hAtx1-82Q pupae.
When axons from photoreceptor neurons were stained with a specific antibody (MAb24B10), the number of axons was reduced and the axonal pathway from retina to lamina and that from lamina to medulla were remarkably distorted (Fig. 7).

Discussion
In this study, we asked generality of the glia-mediated non-cell autonomous effect and the stem/progenitor cell-mediated developmental effect in the pathology of neurodegeneration by using  two different polyQ genes in Drosophila system with two different drivers for gene expression. Glial cell lineage expression of mutant polyQ proteins by the two drivers evoked profound effects on development, survival, behavior and pathology. Although the severities and the patterns of pathological/behavioral changes were different, both mutant hHtt and hAtx1 induced remarkable outcomes either through stem/progenitor cell-specific or glial cellspecific expression. These results basically support the hypothesis that developmental and non-cell autonomous (glial) effects are shared in multiple neurodegenerative disorders. Previously, two groups performed homologous studies. Kretzschmar and colleagues expressed ataxin-3 in Drosophila neurons and glias using APPL-Gal4 and M1B-Gal4 (a Gal4insertion in the repo gene), respectively [13]. They beautifully showed ataxin-3 aggregation in neurons and glial cells, as well as vacuolar changes of glial cells. The Birman's group analyzed effects of repo-Gal4-driven expression of Htt93Q on behaviors [14]. They nicely showed life span decrease and locomotor defects. However, these analyses were limited to a single gene. Both studies did not employ two drivers conducting glial cell lineage expression at different developmental stages. Therefore, our study added novel insights to the polyQ pathologies, including developmental effects of mutant polyQ proteins expressed in glial cell lineage stem cells and distinct effects of huntingtin and ataxin-1.
Gilal cell-specific expression of mutant hHtt by the repo driver leads to the lifespan shortening and the abnormal motor activities ( Fig. 2A, B). Neurons, in addition to glial cells, were affected morphologically in those flies (Fig. 3B). Similarly, remarkable effects on neuronal functions and developmental defects, including neuronal loss in the retina and optic lobes, were induced by repodriven exression of mutant hAtx1 (Fig. 6B). Thus, expression of the two mutant polyQ proteins in glial cells definitely induces a kind of non-cell autonomous effect on neurons.
Meanwhile, gcm-driven expression of mutant proteins in stem/ progenitor cells of glial cell lineage induced more profound effects. In the case of hHtt, gcm-driven expression, either in neuroglioblasts or gliobalsts, before L1 stage during development (Fig. 1) permitted development to pupae and adult flies although the number of adult flies was reduced ( Table 1). The adult flies survived developmental stress have short lifespan ( Fig. 2A), suggesting that impairment of stem/progenitor cells induces certain delayed effect(s) on the nervous tissue. Our result is consistent with the findings by the other group  that hAtx1 interaction with ROR-a induced delayed effect in the model mice [7].
The two disease genes driven by the same driver induced different effects. Mutant hAtx1 expression did not permit development to adult flies in either driver (Fig. 4A). The severities of phenotypes by repo and gcm drivers were almost similar. On the other hand, mutant hHtt expression driven by repo and gcm drivers permitted development to adult flies. The survival of adult flies was shorter in gcm-driven expression than in repo-driven expression ( Fig. 2A).
The discrepancy in phenotypic severities between hHtt and hAtx1 might come from a higher conservation of Atx1 and its interacting molecules in Drosophila. Drosophila Htt is remarkably different from human Htt, while Atx1 possesses 44% of homology in the AXH domain between Drosophila and human [15]. The AXH domain mediates neurotoxicity of hAtx1 through interaction with Gfi-1/Senseless, which is conserved in human and Drosophila [15]. Capicua, a critical interacting molecule to hAtx1 that mediates physiological functions of hAtx1 through binding to phosphorylated Ser776 proximal to the AXH domain [16,17], is also highly conserved between human and Drosophila. Therefore, not only gain of abnormal function but also loss of physiological function might be conserved in Drosophila models similarly to human SCA1 pathology. Another possibility is that exon-1 of mutant hHtt was expressed in our flies. The partial molecule might be less toxic than the full-length hAtx1. On the other hand, as our fly models are expressing the partial molecule, loss of physiological function of hHtt [18] is not highly plausible in our case.
Diversity in the phenotypes of the gcm-driven hAtx1 expressing flies is an open question. As the neuronal pathways are relatively preserved in the mild cases of gcm; hAtx1-82Q flies and survivors show relatively normal glial cells without mutant hAtx1, we might be able to assume that the stem/progenitor cell survival against mutant hAtx1 is stochastically regulated by a certain gene. As the ratio between severe and mild cases was nearly 1:1, such a single gene might regulate the phenotype diversity in gcm; hAtx1-82Q flies. It would be interesting to identify the gene through genetic screening. Although we mainly analyzed adult flies in the case of mutant hHtt expression, the birth rate of the hHtt expressers with the gcm-Gal4 driver was remarkably lower than that with the repo- Gal4 driver ( Table 1), suggesting that embryonic death also occurs in mutant hHtt expression. Identification of such a modifier gene might provide us some hints for therapeutics development.
In conclusion, our fly models expressing mutant hHtt and hAtx1 in glial cells or glial lineage stem/progenitor cells have extended the non-cell autonomous pathology hypothesis as well as the developmental effect hypothesis to multiple polyQ diseases. The two pathologies might be generally shared in neurodegeneration.

Fly stocks and rearing conditions
All flies were raised on a corn-meal medium without propionic acid and were maintained at 25uC and 60% humidity under a 12:12 hr light-dark cycle. pUAST-hHtt 103Q plasmid was generated by subcloning human HD exon 1 cDNA digested from pTL1HA3-HD90Q [19] with EcoRI and NotI, into pUAST vector. During the subcloning, CAG repeats were expanded. Transgenic flies of mutant Atx1 containing human full-length Atx1 with 82Q (y 1 w 1118 UAS:ATX1 82Q) were described previously [20,21]. w 1118 ; P{GAL4}repo/TM3, Sb 1 and gcm-Gal4/Gla flies were obtained from the Bloomington Drosophila stock center. Canton-S strain was used as the wild-type control in this study.

Calculation of toxicity of hAtx1 during larval and pupa stage
Males of gcm-Gal4 and repo-Gal4 driver flies which were balanced by GFP-balancers, gcm-Gal4/CyO, GFP and repo-Gal4/TM3, GFP Sb were crossed with UAS-hATX1 82Q homozygous virgin females respectively. We randomly picked up F1 pupa or larvae and checked their GFP under a stereo fluorescent microscope (LEICA, MZFLIII). Ratio of GFP (2)/GFP (+) larva or pupa were calculated, if expression of hAtx1-82Q at glial cells is not toxic the ratio should be 1.

Survival assay
For measurement of lifespan, about 25 virgin females were reared in a food vial and transferred to fresh food vials every 2 or 3 days. Numbers of dead flies were counted every 1-2 days.

Spontaneous activity assay
For spontaneous activity analysis virgin males were placed individually in glass tubes with one end filled with medium and another end cotton. Their motion is detected and counted by infrared light beam breaks every 30 min using a Drosophila Activity Monitoring System (Trikinetics, Waltham, MA). To calculate synchronized zeitgeber rhythm the flies were kept under 12 hour light-12 hour dark cycles (LD) for 10 days. Negative geotaxis assay Individual female fly was transferred to a test column (150 mm in length and 25 mm in diameter) lined with nylon mesh. To evaluate climbing ability of flies, we used the startle-induced climbing assay, which had been developed by Bainton et al [22]. The bottom of test column was tapped against a soft surface of the bench top to drop flies to the bottom after they had been placed there for 20 sec. Although almost all flies in column were dropped to the bottom by one tap, we used three taps to drop all flies to the bottom. The time that flies reach 50 mm from the bottom were counted. All flies tested reached at 50 mm within 10 sec.

Frozen sections and Immunohistochemistry
Mature pupae were fixed in 4% paraformaldehyde in PBS for 1.5 hours followed by successive incubations in 5% and 10% sucrose for 30 minutes, 15% and 20% sucrose for 1 hour, and 30% sucrose overnight at 4 degree. All sucrose solutions were in PBS. After the heads or mature pupae were frozen in dry ice/ nhexane, 10 mm frontal sections were cut with a cryostat microtome. They were then stained with anti-chaoptin antibody (24B10) (developmental studies hybridoma bank, Iowa University, diluted 1:200), Alexa488-conjugated secondary antibody (Jackson).

Toluidine Blue Staining
Toluidine blue staining of adult heads or mature pupae was performed as we described previously [21,23]. Briefly, paraffin sections (6 mm) were stained with 0.5% toluidine blue (Merck) plus 0.5% Borax after re-hydration.