Recent evidence suggests that DRPLA may be caused by the accumulation of truncated fragments of atrophin-1 in specific neuronal nuclei (Schilling et al. 1999
). These truncated fragments contain the intact NH2
terminus of atrophin-1 (Wood, J.D., unpublished observations). To date, the function(s) of atrophin-1 in the nucleus are unknown. We used the yeast two-hybrid system to find proteins that interact with the NH2
terminus of atrophin-1 and identified ETO/MTG8, a protein known to be associated with transcriptional corepressor complexes. ETO/MTG8 coimmunopurified with atrophin-1 from brain extracts, suggesting that the interaction is physiologically relevant. When cotransfected into Neuro-2a cells, atrophin-1 and ETO/MTG8 colocalized in discrete nuclear matrix–associated structures, which immunostained for the endogenous corepressor complex proteins mSin3A, HDAC1, and HDAC2. Furthermore, atrophin-1 and ETO/MTG8 cofractionated in the nuclear matrix fraction of mouse brain nuclei. We also demonstrated that atrophin-1 can repress transcription in a reporter gene model system, and that repression was enhanced by cotransfection with ETO/MTG8. Together, these findings show that atrophin-1 can be a component of transcriptional corepressor complexes and may be involved in the regulation of gene transcription in the nucleus. Hence the accumulation of truncated atrophin-1 fragments in specific neuronal nuclei in DRPLA could interfere with transcriptional control and contribute to the disease phenotype.
An emerging link between the disease-associated polyglutamine proteins is nuclear matrix association. Ataxin-1 and ataxin-3 both have been shown to be associated with the nuclear matrix (Skinner et al. 1997
; Tait et al. 1998
). Ataxin-1 interacts in a polyglutamine tract-length–dependent manner with the cerebellar leucine-rich acidic nuclear protein in discrete nuclear matrix-associated structures (Matilla et al. 1997
). Here we demonstrate that both normal (unexpanded) and various forms of mutant atrophin-1 (truncated, full-length, and high molecular weight modified) are associated with the nuclear matrix. These forms of mutant atrophin-1 accumulate in DRPLA transgenic mice as the phenotype progresses (Schilling et al. 1999
), so it is possible that these nuclear matrix-associated forms underlie the progression of neuropathology.
In our DRPLA transgenic mice, we see both time-dependent formation of neuronal intranuclear inclusions and accumulation of nuclear matrix–associated forms of atrophin-1. There is a close correlation between the threshold for polyglutamine aggregation in vitro and the threshold for polyglutamine diseases in vivo, and macromolecular aggregates are a good marker for disease. A number of publications suggest that cellular dysfunction, in both cell and animal models, can occur in the absence of polyglutamine aggregation (Hodgson et al. 1999
; Klement et al. 1998
; Saudou et al. 1998
). Hence the correlation between thresholds may merely represent an ability of expanded polyglutamine proteins to adopt a different conformation, which leads to altered and novel protein–protein interactions. In the three studies mentioned above, nuclear localization of the expanded polyglutamine protein was required for neurodegeneration. Demonstration that the neurotoxic expanded polyglutamine proteins are associated with the nuclear matrix in these other models, as in our DRPLA model, would strengthen our argument that dysfunction in the nuclear matrix underlies polyglutamine pathogenesis. The ETO chromosomal translocation in AML misdirects AML1 away from its normal subnuclear domain to alternative subnuclear domains (McNeil et al. 1999
). One could hypothesize that expanded polyglutamine proteins also cause misrouting of gene regulatory factors in an analogous fashion to translocation-associated leukemias. There may already be a precedent for this in the case of SCA1, where mutant ataxin-1 has been shown to influence PML localization (Skinner et al. 1997
The gene products for five, possibly six, genes mutated in polyglutamine neurodegenerative disorders can now be linked to transcriptional control via nuclear receptor signaling. The androgen receptor is a DNA-binding transcription factor that contains an unstable polyglutamine tract that is expanded in patients with X-linked spinal and bulbar muscular atrophy (SBMA) or Kennedy's disease (La Spada et al. 1991
). Here we show that atrophin-1, the DRPLA gene product, represses transcription and interacts with ETO/MTG8 in nuclear matrix-associated structures that contain mSin3 and histone deacetylases. ETO/MTG8 interacts with N-CoR, while N-CoR interacts with huntingtin, the Huntington's disease gene product, in a polyglutamine repeat length–dependent manner (Boutell et al. 1999
). Nuclear receptors such as the retinoic acid receptor (RARα) and retinoid X receptor (RXRα) can associate with PODs. The function of PODs remains enigmatic, although they may be related to gene transcription and cell death (Quignon et al. 1998
; Wang et al. 1998b
). The spinocerebellar ataxia type 1 and 3 gene products have both been linked to PODs: mutant ataxin-1 redistributes PML and colocalizes with PODs to some extent, whereas mutant ataxin-3 aggregates colocalize with PODs (Chai et al. 1999
; Skinner et al. 1997
). We have found that atrophin-1– and ETO/MTG8-containing nuclear complexes appear to be distinct from PODs, but overexpression of atrophin-1 and ETO/MTG8 with PML led to apparent redistribution of PML. Unlike with ataxin-1, PML redistribution was independent of the length of the glutamine repeat in atrophin-1. It should also be noted that expansion of the polyglutamine tract in the TATA-binding protein, a basal transcription factor, may cause neurological disease (Koide et al. 1999
It is apparent that the normal function of a number of the polyglutamine disease proteins may include transcriptional regulation, and there is evidence that altered gene transcription may be a major cause of cellular dysfunction in these diseases. Huntington's disease transgenic mice show alterations in the levels of mRNA for components of neurotransmitter, calcium, and retinoid signaling pathways at early and late symptomatic time points (Cha et al. 1998
; Luthi-Carter et al. 2000
), suggesting that transcriptional regulatory mechanisms are affected in the brains of these mice. Such effects are not restricted to Huntington's disease as polyglutamine expansion in ataxin-1 downregulates specific neuronal genes before pathologic changes occur in SCA1 mice (Lin et al. 2000
). Part of each disease phenotype could reflect interference with the normal function of each polyglutamine protein. However, all of the polyglutamine diseases show autosomal dominant inheritance, suggesting that these diseases are caused by novel gains of function of the expanded polyglutamine proteins, and not simply losses of function. Consistent with such a mechanism, we have found that mutant expanded huntingtin and atrophin-1, but not the normal unexpanded proteins, sequester CBP and interfere with CBP-mediated transcription, which adversely effects neuronal survival (Nucifora, F.C., Jr., M. Sasaki, M.F. Peters, H. Huang, J. Troncoso, V.L. Dawson, T.M. Dawson, and C.A. Ross, manuscript submitted for publication). This type of mechanism could be common to the polyglutamine disorders and distinct from the normal transcriptional role of each protein.
In summary, we have shown that atrophin-1 interacts with ETO/MTG8, that atrophin-1– and ETO/MTG8-containing complexes are associated with the nuclear matrix, and that atrophin-1 can repress transcription. These findings shed considerable new light on the function of atrophin-1 in the nucleus and may also be relevant to pathological neuronal dysfunction in DRPLA. The emerging links between the neurodegenerative disease–associated polyglutamine proteins, translocation-leukemia proteins, nuclear receptor corepressor complexes, and the nuclear matrix may have wider repercussions for the whole family of polyglutamine diseases.