Multiple inherited human neurological disorders are now attributed to expansion of short tandem repeats either in coding or non-coding regions of genes 
. Genetic and molecular analysis of these disorders have revealed that the repeat expansion can result in either a loss of function of the gene (Fragile-X syndrome and Friedreich's ataxia) or a gain of function of the encoded protein (SCA1, SCA2, SCA3, SCA6, SCA7, SCA17, Huntington's disease, DRPLA, and oculopharyngeal muscular dystrophy) 
. RNA-mediated pathogenesis is believed to play a critical role in several other repeat expansion disorders, including Myotonic Dystrophy Type 1 (DM1) and Type 2 (DM2), SCA8, SCA12, Huntington's disease like 2 (HDL2), and fragile X tremor ataxias syndrome (FXTAS) 
. However, the pathogenic mechanism of DM1, SCA8, SCA12, HDL2 and FXTAS, which are caused by trinucleotide repeat expansions, may also involve qualitative or quantitative alterations of the protein products of the respective genes or genes on the opposite strand 
. In contrast, SCA10 is the only human disorder proven to be caused by an expansion of a pentanucleotide repeat. Like the DM2 CCTG tetranucleotide repeat, the SCA10 ATTCT repeat shows repeat-number polymorphism, which makes these non-trinucleotide repeat highly unlikely to encode protein sequences from either strand. Furthermore, we have shown that the intronic repeat expansion does not alter ATXN10 transcripts 
. Thus, SCA10 is likely to be a disorder solely caused by RNA-based mechanism, unlike most disorders that are caused by trinucleotide repeat expansions.
In the present study we provide evidence that SCA10 pathogenesis results from a trans-dominant gain-of-function of AUUCU repeats. First, transcription of the mutant allele produces transcripts that form aggregates in the nucleus and cytoplasm of the SCA10 cells and in transgenic mouse brain. Second: the expanded AUUCU repeat complexes with hnRNP K, leading to the loss of function of hnRNP K. Third, expression of expanded AUUCU repeat results in the accumulation of PKCδ in the mitochondria and caspase-3 mediated activation of apoptosis. Fourth, diminished hnRNP K activity recapitulates these events caused by expanded AUUCU repeats. And finally, over-expression of hnRNP K, as well as down-regulation of transcripts of expanded ATTCT repeat, rescues cells from apoptosis caused by expanded AUUCU repeats. Based on these findings, we conclude that the AUUCU RNA binds to and inactivates hnRNP K, triggering caspase-3-mediated apoptosis via translocation of PKCδ to mitochondria. Previous reports suggest that the presence of PKCδ in the mitochondria results in decreased membrane potential, release of cytochrome C, and activation of caspase-3 
, further supporting our conclusion. Moreover, caspase-3 activates PKCδ and activated PKCδ further activates caspase-3 
, and proteolytically activated PKCδ down-regulates hnRNP K protein in a proteasome-dependent manner 
. Hence, positive feedback loops involving hnRNP K, PKCδ and caspase-3 may enhance this pathogenic pathway in SCA10. Since apoptosis is considered to be a major mechanism of cell death in a variety of human neurodegenerative disorders 
, the novel pathway of apoptosis induced by the mutant ATXN10
RNA is relevant to the neurodegenerative phenotype of SCA10. Our results provide strong evidence that this novel mechanism of trans-dominant RNA gain of function contributes to the pathogenic mechanism in SCA10.
The formation of aggregates may not necessarily be a required event for the mutant RNA to exert its toxicity. Binding of the soluble form of the mutant RNA to hnRNP K may be sufficient to cause the loss of function of hnRNP K with a release of PKCδ, and the aggregate formation could be a secondary phenomenon. We hypothesize that expanded AUUCU RNA pathologically binds to hnRNP K and prevents PKCδ from binding to the hnRNP K, mimicking over-expression of PKCδ within the cell. Previous studies have shown that hnRNP K is constitutively bound to PKCδ, but upon binding to nucleic acids, hnRNP K can no longer interact with PKCδ 
. Translocation of PKCδ to mitochondria in SCA10 cells, fibroblasts expressing expanded AUUCU repeat, and fibroblasts treated with hnRNP K siRNA argues for this mechanism. Studies have shown multiple apoptotic activators, including oxidative stress and over-expression of PKCδ, induce PKCδ translocation to the mitochondria, 
. The mitochondrial translocation of PKCδ has been shown to cause an alteration in calcium signaling events and mediates the H2
-mediated loss of membrane potential, release of cytochrome c, and activation of caspase-3 
. While it is possible that the expanded AUUCU repeat causes PKCδ translocation via other mechanisms, our data showing that over-expression of hnRNP K rescues AUUCU-mediated apoptosis argue for the mechanism mediated by a loss of function of hnRNP K.
Our present data do not rule out the possibility that additional proteins interact with the mutant ATXN10 transcripts. Also, expression of hnRNP K is ubiquitous within the cell, and diminished hnRNP K could lead to altered regulation of transcription, splicing and cell signaling, which may account for the phenotypic variability in SCA10 as illustrated in . We are investigating these mechanisms. However, our current data convincingly show that the hnRNP K inactivation and PKCδ mitochondrial translocation are a key pathogenic pathway mediating the RNA gain of toxic function in SCA10.