Recently, we proposed the possibility that in addition to toxic gain-of-function due to polyglutamine-expanded ATXN1, a concomitant partial loss of ATXN1 function might contribute to SCA1 pathogenesis 
. It is challenging to establish the extent of the contribution of a potential loss-of-function mechanism to SCA1 pathogenesis in models carrying the mutant protein, since the severe gain-of-function effects might mask any subtle loss-of-function component, thus confounding the interpretation of the results. In the present study, we sought to distinguish between gain- and loss-of-function mechanisms by focusing on transcriptional defects in Atxn1−/−
mice, and comparing them to the knock-in model of SCA1, Atxn1154Q/+
mice. Using this approach, we identified several molecular changes that could be attributable to loss of ATXN1 function in SCA1.
We found that loss-of-function of Atxn1
in mice is sufficient to cause many transcriptional changes common to the Atxn1154Q/+
knock-in mice, a model of SCA1 that faithfully replicates many features of the disease, and with SCA1[82Q]
transgenic mice. It has been reported that ATXN1 interacts with several factors involved in transcriptional regulation, including CIC, SMRTER, HDAC3, Gfi1 and RORα 
. Therefore, these shared expression changes might be indicative of altered transcriptional functions of ATXN1 in SCA1 pathogenesis. Furthermore, we showed that a majority of the shared transcriptional changes go in the same direction in both Atxn1−/−
mice, strongly arguing that part of the transcriptional dysregulation in SCA1 might be explained by a partial loss-of-function of Atxn1. The up-regulation of direct targets of Atxn1-Cic provides evidence for this concept. Another important finding of this study is that there are many transcriptional changes that are unique to the Atxn1154Q/+
model, which could potentially be related to toxicity of polyglutamine-expanded Atxn1.
We propose that a combination of toxic gain-of-function and mild loss-of-function mechanisms contribute to SCA1 pathogenesis, with the partial loss-of-function of ATXN1 being sufficient to cause some transcriptional changes that are pathogenic in the cerebellum. Previous studies using microarray analysis reported on the down-regulation of the dopamine receptor D2 (Drd2
) in Atxn1−/−
mouse cerebella 
. However, with the exception of a couple of genes (e.g. Pafahb3
), we were unable to find extensive overlap between the changes reported by Goold et al.
and the microarray analysis presented in this study. The differences in genetic background, microarray platform, and the age of the Atxn1−/−
animals (5-week-old) in the Goold et al.
, compared to 16 weeks in our studies, are likely to contribute to the minimal overlap in gene expression changes in the two studies.
Bioinformatics analyses of the genes commonly altered in Atxn1−/−
cerebella show enrichment for categories associated with pathological pathways involved in neurodegeneration (Alzheimer's disease), and also pathways previously implicated in pathogenesis both in knock-in and transgenic SCA1 mouse models, such as the phosphatidylinositol and calcium signaling pathways 
. These results strongly suggest that Atxn1−/−
mice have dysfunctional cerebella due to a loss of endogenous Atxn1 function. We also found that Atxn1−/−
mice share significant overlap in cerebellar transcriptional profiles with staggerer
mice, which have a spontaneous loss-of-function mutation in the gene encoding the transcription factor Rorα. Rorα-regulated genes involved in calcium signaling (Itpr1
) and glutamatergic signaling (Grm1
) are significantly down-regulated in Atxn1−/−
cerebellum, as determined by microarray analysis and real-time qRT-PCR. It is noteworthy that loss-of-function mutations in several of these genes result in ataxic phenotypes (e.g. Itpr1
, and Grm1
, raising the possibility that simultaneous down-regulation of several of these genes could contribute to the motor coordination impairments observed in Atxn1−/−
mRNA transcript and protein levels appear normal in Atxn1−/−
, ruling out that changes in Rorα targets are due to reduced Rorα protein levels in Atxn1−/−
mice. Since ATXN1 and Rorα physically interact via Tip60, it is conceivable that loss of Atxn1 affects Rorα-dependent transcription directly 
. Altogether, these data support two important conclusions: first, that Atxn1−/−
cerebellum exhibits pathological molecular changes, even in the absence of progressive neurodegeneration, and second, that transcriptional changes in the loss-of-function model of Atxn1
could identify endogenous pathways that might also be altered by the expression of mutant Atxn1.
We previously described a reduction of Atxn1-Cic complexes in Atxn1154Q/+
cerebella, with Atxn1[154Q] favoring the formation of enhanced toxic gain-of-function complexes with RBM17 
. It is interesting that among the genes up-regulated both in Atxn1−/−
cerebella, we identified two potential direct targets of Cic-dependent repression, Ccnd1
, the genes encoding for Cyclin D1 and Ets variant 5, respectively (
and J. Fryer, unpublished data). We demonstrated that wild-type Atxn1 and Cic are bound to the promoter regions of Ccnd1
. Interestingly, we failed to detect mutant Atxn1[154Q] on these promoters in mice only expressing expanded Atxn1 (Atxn1154Q/−
). One interpretation of this result is that Atxn1[154Q] has diminished association to the promoters, resulting in reduced Atxn1-Cic dependent repression. Alternatively, it is possible that polyglutamine-induced conformational changes make the Atxn1[154Q]-Cic complexes less accessible for antibody recognition, resulting in reduced chromatin immunoprecipitation. Irrespective of the basis of the inability to detect Atxn1[154Q] binding on these promoters, the data strongly suggest that polyglutamine-expanded Atxn1 and Cic have reduced transcriptional repression function on these specific promoters in vivo
. These results provide a mechanistic explanation on how diminished Atxn1-Cic function can contribute to transcriptional defects in SCA1.
Mild overexpression of the evolutionarily conserved gene Atxn1L
partially suppresses the neuropathology caused by polyglutamine-expanded ATXN1 in flies and mice 
. Increased Atxn1L levels induce the sequestration of polyglutamine-expanded Atxn1 into nuclear inclusions, leading to a proposed model in which Atxn1L suppresses toxicity by displacement of mutant Atxn1 from its major endogenous complexes that contain Cic 
. In the present study, we show an additional mechanism contributing to this rescue, by demonstrating that mild overexpression of Atxn1L
suppresses several molecular and behavioral phenotypes in Atxn1−/−
mice, potentially by replacing Atxn1 in Cic-containing complexes (). The motor coordination and learning deficits suppressed by Atxn1L
are common to both Atxn1−/−
and polyglutamine-expanded Atxn1154Q/+
mouse models. Therefore, these data provide evidence for an additional mechanism in which Atxn1L
can functionally compensate for a partial loss of Atxn1
function to suppress SCA1 pathogenesis. Although our studies demonstrate that Atxn1L
is a functional homolog of Atxn1
in Cic-mediated transcriptional repression, we cannot rule out that Atxn1L
overexpression can restore other yet to be determined Atxn1
-related functions not addressed in these studies. In sum, based on our previous data and this study, we propose that partial loss of ATXN1 function actively contributes to SCA1 pathogenesis as part of a two-pronged mechanism, in which enhanced toxic gain-of-function of polyglutamine-expanded ATXN1 leads to neurodegeneration, while a simultaneous loss-of-function of other stable endogenous protein complexes, Atxn1-Cic, contributes to the SCA1 phenotypes ().
Model of Atxn1L rescue through Cic stabilization in Atxn1−/− mice.
It was previously reported that reduction of normal functions of genes involved in other polyglutamine diseases results in enhanced pathology, providing evidence for concomitant gain- and loss-of-function mechanisms in polyglutamine disorders 
. In SBMA transgenic mouse models expressing polyglutamine-expanded androgen receptor (AR), loss of endogenous AR protein resulted in accelerated motor neuron degeneration 
. These studies suggested two independent pathways contributing to SBMA pathogenesis: gain-of-function due to mutant AR nuclear toxicity and loss of AR trophic effects on motor neurons 
. Conditional deletion of Htt
in mouse forebrain leads to several features reminiscent of Huntington disease, including motor deficits, tremors and progressive degeneration of the striatum and cortex, hinting that loss of htt function could contribute to these phenotypes in HD 
. Moreover, loss of wild-type Htt function leads to enhanced neurodegeneration in transgenic models expressing polyglutamine-expanded Htt, while overexpression of wild-type Htt reduces toxicity caused by mutant htt 
. The mechanism by which loss of Htt contributes to HD is unclear at this time; it might involve either its anti-apoptotic properties, its role in BDNF-mediated neuroprotection, both, or some other yet to be determined function 
. Our studies comparing Atxn1−/−
and SCA1 knock-in mice pinpoint Cic-dependent transcriptional repression as one of the molecular pathways mediating the partial loss-of-function component in SCA1 pathogenesis.
Loss of normal endogenous function of mutant proteins may also play a role in other dominant neurodegenerative diseases caused by gain-of-function mutations. The Parkinson's disease model mice overexpressing the mutant A53T SNCA
gene lacking endogenous alpha-synuclein, exhibit worsened synucleinopathy when compared to littermates carrying wild-type Snca
. In Alzheimer disease (AD), increased aggregation of the amyloid beta peptides induced by AD-related presenilin mutations is thought to be a consequence of a dominant gain-of-function mechanism 
. However, loss of function of presenilin in the mouse brain results in phenotypes strikingly reminiscent of AD (progressive memory loss and neurodegeneration) in the absence of beta-amyloid deposition 
. These results suggest that altered pathways leading to Alzheimer disease can be caused from a combination of dominant gain of function and/or loss of function mechanisms. The potential prevalence of mutations that lead to both loss- and gain-of-function in human neurological diseases underscores the importance of understanding the endogenous functions of causative genes through the careful analysis of loss-of-function models, which may uncover critical pathways leading to pathogenesis.