Stress granules represent a conserved component of the cellular response to stress and are crucial for cell survival . In the current study, we have shown that expanded ATXN7 affects the expression and/or behaviour of TDP-43, TIA1 and G3BP1, three RNA-binding proteins involved in the stress granule response. We could show that phosphorylation of S409/S410 in TDP-43 is increased and that TDP-43 co-localises with ATXN7 aggregates in SCA7 cells. This is consistent with previous studies identifying increased TDP-43 phosphorylation and sequestration in SCA7 mice and patients, as well as in several other neurodegenerative diseases [37–41, 34]. Findings in other diseases have indicated that TDP-43 exits the nucleus during stress, [42, 27] and that TDP-43 phosphorylation exacerbates TDP-43 related neuropathology [35, 43]. However, while Coudert et al. found that TDP-43 phosphorylation was secondary to aggregation, Hasegawa et al. observed that TDP-43 is more prone to multimerise into filaments after phosphorylation by Casein kinase 1 [39, 41]. Interestingly, we found a higher co-localisation of TDP-43 with cytoplasmic, rather than nuclear, ATXN7 aggregates. Therefore, we hypothesise that the stress caused by expanded ATXN7 in our SCA7 cell model could cause TDP-43 to exit the nucleus, interact with cytoplasmic polyQ fragments, and become phosphorylated, which strengthens the interaction and causes accumulation of TDP-43 into aggregates.
Besides TDP-43, we could also observe a clear co-localisation of TIA1 with both nuclear and cytoplasmic ATXN7 aggregates. This together with the fact that TDP-43 has been shown to positively regulate the expression of G3BP1 [44, 27], initially led us to expect decreased soluble levels of TDP-43, TIA1 and G3BP1 in our SCA7 cell model. However, western blot and IF analyses instead revealed similar or statistically significantly increased expression level of these RBPs, suggesting that the stress caused by mutant ATXN7 triggers the stress granule response pathway. As a matter of fact, increased expression of stress granule nucleating proteins has previously been observed in Huntington’s disease patients and models [45, 46]. Not only were we able to observe an increased level of G3BP1 in our expanded ATXN7 expressing cells, but we also found that the behaviour of G3BP1 changed. Although fully formed G3BP1 positive SGs were rare in the SCA7 cells, G3BP1 displayed a statistically significant increase in signal texture and speckling, indicating condensation of G3BP1 into pre-SG structures in ATXN7 cells. Furthermore, following treatment with the SG inducing compound arsenite, a trend towards more SGs in mutant ATXN7, versus control cells, could also be observed. The fact that the arsenite induced effect on SG formation in mutant versus control cells was smaller than the observed difference in G3BP1 speckling levels before arsenite treatment, is most likely due to the close to maximal induction of the SG response by arsenite, as mentioned previously. Hence, despite the sequestration of TDP-43 and TIA1, the ability of SCA7 cells to induce SGs is not reduced and in fact mutant ATXN7 appears to induce the SG response.
Although we could not see any sequestration of G3BP1 into mutant ATXN7 aggregates, a clear co-localisation of mutant ATXN7 to SGs could be observed upon arsenite mediated induction of stress granules in both SCA7 PC12 cells and patient fibroblasts. Surprisingly, endogenous and transgenic wild-type ATXN7 also localised to SGs to some extent. To our knowledge this is the first time ATXN7 has been linked to SGs, though another STAGA DUB subunit, sus1, has been shown to localise to SGs in yeast . Moreover, the yeast ATXN7 ortholog sgf73, along with the STAGA DUB module, were recently linked to export of essential transcripts in response to stress . Further studies to investigate the connection between wild-type ATXN7, the STAGA DUB module, and the stress granule response would thus be interesting.
Despite both endogenous and mutant ATXN7 localising to SGs, we could observe a difference in the shape of the SGs when comparing expanded ATXN7 expressing cells to non-induced control cells, indicating that expanded ATXN7 does not interact with SGs identically to wild-type. An interesting recent paper has compared SG structure after overexpression of several SG proteins, such as TIA1, G3BP1, TDP-43, FMR1 etc. . They noted that when different SG proteins were overexpressed to induce SG formation, granules of two classes were formed, which they designated as rough and smooth granules. G3BP1 and TDP-43 overexpression resulted in smooth granules, while increased TIA1 levels produced rough granules. While we can speculate that the partial sequestration of TDP-43 and/or TIA1 into polyQ aggregates, as well as the co-localisation of mutant ATXN7 with stress granules could affect granule shape and function, it is important to note that SGs are incredibly complex structures. The nucleation is performed by numerous proteins and mRNAs , and the resulting granule is composed of both a stable core and a more dynamic shell . It is therefore difficult to predict how small subtractions or additions of specific proteins will affect shape or function of SGs. In fact, from our results it is clear that stress granule induction, as well as disassembly do not appear to be negatively affected by mutant ATXN7 expression. Furthermore, SGs have been suggested to act as platforms able to nucleate polyQ aggregation [52, 23]. However, we could not observe any statistical increase in ATXN7 aggregation following SG induction, although there was a trend towards increased insoluble ATXN7 levels captured by filter trap. Nevertheless, considering that SCA7 is a chronic disease and the SG response is presumably continuously activated in SCA7 patients as a result of the DNA damage, oxidative stress and overloaded proteasome/autophagy systems, that can be observed in disease models [11, 19–21], it is plausible that SGs in the long run could potentiate polyQ aggregation. In fact, SGs formed in response to chronic stress have been suggested to promote cell death, rather than survival, for review see . On the other hand, the long-term accumulation of toxic expanded ATXN7 species [54, 1, 55], could with time result in sufficient disruption of PrLD containing RBPs, both potentially within SGs and elsewhere, resulting in deleterious dysfunctions of the stress response.
In conclusion, we find that although the stress granule proteins TIA1 and TDP-43 are sequestered into mutant ATXN7 aggregates, this does not inhibit stress granule formation or disassembly. In contrast, a heightened condensation behaviour of G3BP1 and trend towards more stress granules could be observed in SCA7 cells, suggesting that expansion of ATXN7 induces stress that triggers the SG response. Most interestingly, it seems that ATXN7 localises to SGs and that association of the mutant protein into SGs, alters the shape of the SGs. The role of this association, as well as the molecular consequences of an altered SG shape in SCA7 disease, needs to be clarified in further studies.