TBP fragments are present in SCA17 cells and rat cerebellum
The occurrence of TBP fragments has been reported not only for brain tissue of SCA17 mice, but also of AD patients [14, 15]. To examine whether these fragments are also detectable in our SCA17 models, we performed western blot analysis of protein homogenates extracted from HEK 293T cells transfected with myc-tagged full-length human TBP (myc-TBP) with 38Q or 64Q and from the cerebellum of our TBPQ64 rats. Along with these samples, extracts of untransfected cells and wild-type (WT) rats were assayed as controls. To achieve good coverage of the entire TBP, we applied a selection of antibodies with epitopes distributed from the myc-tagged N-terminus to the C-terminal DNA-binding core domain of the overexpressed protein (Fig. 1a). Immunodetection using the c-myc-specific antibody stained overexpressed full-length TBP in cells and rat cerebellum (Fig. 1b), whereas antibodies N-12, #8515 and 58C9 additionally detected the endogenous full-length protein (Fig. 1c, d, g). Antibody D5G7Y stained all human TBP full-length species in our samples but failed binding to endogenous TBP in rat cerebellum (Fig. 1f). The same applied to polyQ-specific antibody 1C2 (Fig. 1e), which detects polyQ stretches greater than or equal to 38Q.
Two potential C-terminal TBP fragments, c1 (migrating above 28 kDa) and c2 (migrating below 28 kDa), were detected by the antibody 58C9 in both transfected HEK 293T cells and TBPQ64 rat cerebellum (Fig. 1g). These fragments lack the polyQ stretch as they neither presented corresponding size shifts nor were detected by the polyQ-specific antibody 1C2 (Fig. 1e). Untransfected HEK 293T cells and WT rats did not feature both fragment bands, which might be due to the comparably lower expression levels of endogenous TBP. Fragment c1 was also detected by antibody D5G7Y (Fig. 1f), suggesting that it spans from the D5G7Y epitope, located around Ala110 (human reference isoform 1; UniProt identifier: P20226-1), to the C-terminus of TBP, which is detected by antibody 58C9. Surprisingly, none of the other antibodies with specificities for the N-terminal portion of TBP were able to stain further TBP fragments, neither on nitrocellulose nor PVDF membranes (Fig. 1b–e, Suppl. Fig. S1a and b, Supplementary File 1), prompting the question of the fate of the N-terminal, polyQ stretch-containing counterparts of fragments c1 and c2.
In silico tool predicts multiple calpain cleavage sites within TBP
Previous attempts have failed to associate proteases with the fragmentation of TBP, while an N-terminal fragment was shown to originate from alternative splicing in AD patient brains [16, 19]. As calpains have been associated with the fragmentation of multiple disease proteins in neurodegenerative disorders [21, 23,24,25, 45], we hypothesized that these enzymes are also accountable for the observed fragments in our models and, consequently, for the cleavage of polyQ-expanded TBP in SCA17.
Before addressing this question on a biological level, we tested the potential cleavage likelihood of TBP using the in silico calpain cleavage site prediction tool GPS-CCD . Based on the canonical sequence of human TBP (UniProt identifier: P20226-1), the prediction tool localized 20 putative calpain cleavage sites along the entire sequence and a cleavage site cluster in the alanine/glutamine-rich stretch between Ala96 and Glu117, which surpasses the maximum default cut-off value of 0.654 (Fig. 2a). A comprehensive listing of predicted calpain cleavage sites can be found in Supplementary Table S3, Supplementary File 1. Analogously to our previous study on the cleavage of the MJD protein ataxin-3 , the polyQ stretch of TBP yielded high cleavage likelihood scores, which may be considered a prediction artefact.
In vitro cleavage assays indicate TBP as a calpain substrate
Based on our computational approach, TBP cleavage by calpains in a biological system seemed probable. To test this, we performed in vitro calpain cleavage assays by incubating protein extracts of HEK 293T cells expressing myc-TBP 38Q or 64Q and of WT and TBPQ64 rat cerebellum with purified calpain-1 or calpain-2 plus CaCl2, for up to 30 min. For reasons of comparability, the reaction speed was adjusted by selecting corresponding amounts of both proteases, based on the cleavage of the calpain substrate α-tubulin  (Suppl. Fig. S2a and b, Supplementary File 1). To ascertain the specificity of the reaction, we additionally supplemented 30 min-control samples with calpain inhibitor III (CI-III) to block calpain-dependent cleavage. Western blotting was performed to analyze TBP fragmentation in obtained samples using the previously established antibodies N-12, directed against the N-terminus, and 58C9, binding the C-terminal core domain of TBP. Immunostaining with antibody 58C9 detected the previously observed fragments c1 and c2 in protein extracts of HEK 293T cells expressing myc-TBP 38Q or 64Q. Strikingly, incubation with calpain-1 (Fig. 2b) and calpain-2 (Fig. 2c) showed a time-dependent accumulation of both fragment bands, which was abolished by the co-administration of CI-III. Concurrently, full-length endogenous and overexpressed TBP reduced over time, except for the inhibitor-treated samples. Replication of the experiment using protein extracts of WT and TBPQ64 rat cerebellum incubated with calpain-1 (Fig. 2d) and calpain-2 (Fig. 2e) yielded corresponding results. No differences were found between fragmentation patterns of calpain-1 or calpain-2, pointing to conserved cleavage sites of both protease isoforms within TBP. Moreover, fragments c1 and c2 in both human and rat samples showed comparable sizes, which can be attributed to the nearly perfect interspecies sequence identity of TBP’s C-terminal portion adjacent to the polyQ stretch. Interestingly, after an initial increase until 15 min, fragment c1 appeared to reduce overtime while fragment c2 levels continued to rise. This might indicate a further degradation of c1 into smaller breakdown products like c2, which was substantiated by additional double labelling of c1 by antibodies D5G7Y and 58C9, showing that c2 lacks the D5G7Y’s epitope around Ala110 (Suppl. Fig. S2c, Supplementary File 1). As demonstrated in the previous detections, antibody N-12 stained full-length TBP and verified its time-dependent reduction but failed to prove the occurrence of corresponding N-terminal, polyQ stretch-containing TBP fragments (Fig. 2b, c).
Cell-based calpain activation leads to fragmentation of wild-type and polyQ-expanded TBP
To confirm our findings on calpain-mediated TBP cleavage in the more physiological context of an intact cell, we performed cell-based calpain assays by incubating HEK 293T cells expressing myc-TBP 38Q or 64Q with the calcium ionophore ionomycin. Treatment with this compound leads to an elevation of cytoplasmic calcium concentrations, thereby activating endogenously expressed calpains .
Transfected HEK 293T cells were incubated with ionomycin plus CaCl2 for up to 2 h. For specificity controls, cells were pre-treated with CI-III 1 h before administration of ionomycin. Western blot analysis was performed to investigate the levels of calpain activation and TBP cleavage. First, the effectiveness of the ionomycin-dependent calpain activation was assessed by detecting autoproteolysis of calpain-1 and cleavage of the calpain substrate α-spectrin. Both markers showed increased and time-dependent processing, which was abolished by the pre-incubation with CI-III (Fig. 3a). Immunostaining of TBP using the antibody 58C9 showed a corresponding accumulation of the previously observed fragments c1 and c2 (Fig. 3b). Compared to in vitro calpain cleavage assays, a reduction of full-length TBP was not apparent, as indicated by staining with antibody N-12 and 58C9, which—together with the relatively lower fragment levels—is explained by the lesser degree of calpain-mediated cleavage in living cells. To measure both the levels of induced TBP fragmentation and the efficacy of calpain inhibition by CI-III, we repeated the experiments using a single time point of 2 h (Fig. 3c). Densitometric analysis of the TBP fragment bands c1 and c2 showed a significant 50–60% level increase upon ionomycin treatment, whereas combined pre-incubation and co-treatment with CI-III effectively reduced fragmentation by approximately 50% compared to baseline levels (Fig. 3d). As we sought to reproduce this experiment in a different cell model, we repeated ionomycin and CI-III treatments using a PC12 cell model of SCA17, expressing TBP with 13 or 105Q . Western blot analysis showed similar results for both α-spectrin and TBP fragmentation, including the occurrence of fragment bands c1 and c2, but with observable lowering effects on the respective full-length proteins (Fig. 3e).
Together with the in silico and in vitro analyses, our cell-based assays confirmed TBP as a calpain substrate and calpain-mediated cleavage as a source for the TBP fragments observed at baseline in SCA17 cells and animals.
C-terminal fragments of TBP accumulate in the cytoplasm
Previous studies have identified N-terminal, polyQ stretch-containing fragments of TBP in tissues of mice and humans [14, 15]. Interestingly, we could not identify these specific fragments using our approaches but instead robustly detected their potential C-terminal, calpain cleavage-derived counterparts c1 and c2. To scrutinize these fragments’ origin and fate, we sought to determine their intracellular localization, compared to full-length TBP and members of the calpain system. For this, we performed cytoplasmic-nuclear fractionation of untransfected and myc-TBP 38Q or 64Q-transfected HEK 293T cells and screened the obtained fractions for the occurrence of full-length and cleaved TBP via western blotting using antibody 58C9. As expected, we found endogenous and overexpressed TBP having a predominantly nuclear localization (Suppl. Fig. S3a, Supplementary File 1; Fig. 4a–d). A smaller but distinct proportion of the full-length protein was present in the cytoplasmic fraction. In addition, we confirmed the cellular distribution of TBP using microscopy and immunostaining with antibody N-12 (Fig. 4e). Most surprisingly, fragments c1 and c2 showed for both endogenous (Suppl. Fig. S3a, Supplementary File 1) and overexpressed TBP variants with 38Q (Fig. 4a, b) or 64Q (Fig. 4c, d) a localization pattern opposite to the ones of the full-length proteins. To compare localization of TBP with the distribution of members of the calpain system, we likewise performed microscopical (Fig. 4f) and cytoplasmic-nuclear fractionation (Suppl. Fig. S3b, Supplementary File 1) analysis for calpain-1, calpain-2 and CAST. The proteases and their inhibitor exhibited a mainly cytoplasmic localization, although calpain-1 showed an additional characteristic granular occurrence within the nuclei of HEK 293T cells. This suggests that the calpain-mediated fragmentation may rather take place in the cytoplasm. A further hint for active cytoplasmic retention of fragments c1 and c2 was given by performing an in silico prediction for leucine-rich nuclear export signal (NES) using the NetNES tool , which indicates a potential NES around amino acid Leu275 within the C-terminus of human TBP (UniProt identifier: P20226-1) (Suppl. Fig. S3c and Suppl. Table S4, Supplementary File 1).
Our fractionation analyses intriguingly demonstrated that the calpain cleavage-derived fragments c1 and c2 are enriched in the cytoplasmic fraction of the cell, which stands in contrast to its source, full-length TBP, but matches the distribution of members of the calpain system, such as calpain-1 and CAST.
The calpain system is overactivated in SCA17 cells and rat cerebellum
In parallel to calpain-mediated fragmentation of disease proteins, many neurodegenerative disorders exhibit a pathological overactivation of calpains, which trigger detrimental cascades in affected tissues . To evaluate if this kind of overactivation is present in our SCA17 cell and animal models, we performed western blot analysis of cells expressing TBP with 13Q or 105Q and WT or TBPQ64 rat cerebellum. To assess the state of calpain activation, we detected the autoproteolysis of calpain-1, levels of the endogenous inhibitor CAST and cleavage of the calpain substrate α-spectrin. In PC12 and HEK 293T cells expressing TBP with 105Q, we found significantly reduced CAST levels as well as elevated calpain-1 and α-spectrin fragmentation, pointing towards calpain overactivation (Fig. 5a, b; Suppl. Fig. S4, Supplementary File 1). Next, we analyzed brain tissue of TBPQ64 rats at 10 months of age, which represents a terminal stage of phenotypic progression in our SCA17 animal model . Here, we observed the same significant effects on all three calpain activation markers in the TBPQ64 rat cerebellum compared to WT samples (Fig. 5c, d). Importantly, striatum and cortex of TBPQ64 rats, which feature only very low or low myc-TBP 64Q transgene expression (Suppl. Fig. S5a, Supplementary File 1)  and, consequently, lack the SDS-insoluble TBP aggregates as a disease hallmark (Suppl. Fig. S6, Supplementary File 1), showed no evidence of calpain overactivation (Suppl. Fig. S7, Supplementary File 1). These results prove that the expression of a polyQ-expanded TBP drives the elevated activity of calpains. Interestingly, endogenous TBP was reduced upon transgene expression in the TBPQ64 rat cerebellum, pointing to a potential compensatory effect on the regulation of total TBP levels (Suppl. Fig. S5b and c, Supplementary File 1). Moreover, we observed an increased occurrence of fragments c1 and c2 in the TBPQ64 rat cerebellum (Fig. 5e). Fragment levels were approximately threefold higher than in WT tissue, which is mainly due to the transgene-dependent rise in total full-length TBP levels (Suppl. Fig. S5c, Supplementary File 1). However, when normalizing fragment to full-length TBP levels, a significant increase in c1 and c2 levels by approximately 20% was still detectable (Fig. 5f), which might be a consequence of the upregulated calpain activity in TBPQ64 rat cerebellum.
Our findings in PC12 and TBPQ64 rats substantiate calpain overactivation as a new hallmark in models of SCA17, potentially causing further consequences on its molecular pathology.
Calpain overactivation leads to cleavage and dysregulation of synaptic proteins
In previous studies of our laboratory and other research groups, pathological overactivation of calpains in animal models of various neurodegenerative diseases caused excessive and potentially deleterious fragmentation of neuronal substrates, including synaptic proteins [27, 31, 36, 40, 48]. As we observed elevated calpain activity in the cerebellum of our TBPQ64 rats, we sought to analyze its impact on a selection of known neuronal calpain substrates. For this, we have performed western blot analysis of protein extracts from WT and TBPQ64 rat cerebellum and detected synapsin-1a/b, synapsin-2a and PSD-95, as well as p35. Both synapsins showed a distinct accumulation of fragments accompanied by a reduction of full-length synapsin-1a/b (Fig. 6a, c). In the TBPQ64 cerebellum, fragment levels were approximately 70% higher for synapsin-1a/b and nearly 60% higher for synapsin-2a, whereas full-length levels of synapsin-1a/b were slightly but significantly decreased, and full-length synapsin-2 presented a trend towards lowering when compared to WT samples (Fig. 6b, d). Correspondingly, full-length PSD-95 levels were reduced by half, but we could not detect breakdown products despite showing that calpain-1 readily cleaved this protein (Suppl. Fig. S8a, Supplementary File 1). On the other hand, conversion of p35 to p25 was markedly elevated by almost 90% (Fig. 6e, f), as well as cleavage of the neuron-specific and confirmed calpain substrate βIII-tubulin (Fig. 6g, h; Suppl. Fig. S8b, Supplementary File 1). These observations reveal that the detected calpain overactivation causes strong dysregulations of synaptic proteins and other neuronal proteins in the cerebellum of TBPQ64 rats.
SCA17 PC12 cells show disturbances of the synaptogenesis and calcium signaling pathways
To scrutinize the connection between polyQ-expanded TBP expression and the overactivation of calpains, we performed a 3’ RNA sequencing of PC12 cells expressing TBP 13Q and 105Q. We detected a total of 1915 differentially expressed genes (DEGs) between both cell lines, of which 795 were up- and 1120 downregulated (Supplementary File 2). Interestingly, genes coding for members of the calpain system including CAST did not show major alterations in their expression levels (Suppl. Fig. S9d, Supplementary File 1), ruling out a direct impact of polyQ-expanded TBP on their transcription. Using the list of identified DEGs, Ingenuity Pathway Analysis revealed synaptogenesis and calcium signaling amongst the top five regulated canonical pathways (Fig. 7a), both showing a significant inhibition. These effects were substantiated by perturbations of mainly downregulated DEGs enriched in both pathways (Supplementary File 2; Fig. 7b, c). Overall, PC12 cells expressing TBP 105Q demonstrated both higher magnitude and significance of DEGs featuring a reduced expression (Fig. 7d). Synaptotagmin 11 (Syt11; log2fc = 1.60) and cadherin 2 (Cdh2; log2fc = −7.58) represented the most dysregulated genes in the synaptogenesis pathway (Suppl. Fig. S9a, Supplementary File 1), and ryanodine receptor 2 (Ryr2; log2fc = 1.61) and sarcoplasmic/endoplasmic reticulum calcium ATPase 3 (Atp2a3; log2fc = −5.96) were most changed in the calcium signaling pathway (Suppl. Fig. S9c, Supplementary File 1). Both canonical pathways shared several DEGs, including calmodulin 2 (Calm2; log2fc = 0.61) and glutamate ionotropic receptor AMPA type subunit 2 (Gria2; log2fc = −5.01) (Suppl. Fig. S9b, Supplementary File 1). We validated the detected expression alterations in these genes by reverse transcription quantitative real-time PCR, which yielded nearly identical fold changes in comparison to our RNA sequencing data (Fig. 7e–g). These results suggest that the observed overactivation of calpains in SCA17 models may be triggered by polyQ-expanded TBP-dependent disturbances of calcium signaling and homeostasis. The transcriptional changes of the synaptogenesis pathway observed in SCA17 PC12 cells, however, may amplify the dysregulation of synaptic proteins as detected in the TBPQ64 rat cerebellum.
Inhibition of calpains lowers TBP cleavage and aggregation improving cell viability
Calpain inhibition by overexpression of CAST or pharmacological means has been proven effective in reducing levels of toxic fragments and aggregates of mutant proteins, thereby attenuating the overall pathology in models of neurodegenerative diseases including HD, MJD and PD [28, 34,35,36, 49, 50]. Therefore, we tested both genetic and pharmacological approaches on our SCA17 cell models for evaluating their impact on TBP cleavage and aggregation as well as cell viability. In our first attempt, we overexpressed human CAST in HEK 293T cells together with TBP 13Q or 105Q for 72 h and analyzed the obtained protein extracts via western blotting. We achieved a threefold increase of CAST resulting in an approximately 30% reduction of α-spectrin cleavage, which was detected as a marker of overall calpain activity (Fig. 8a, b). By immunodetection of TBP with antibodies N-12 and 58C9, we observed a significant 40–50% lowering of the C-terminal fragments c1 and c2 upon CAST overexpression (Fig. 8c, d), again confirming the calpain-dependent origin of these TBP breakdown products. To test how this reduction manifests on the subcellular level, we performed a cytoplasmic-nuclear fractionation of HEK 293T co-expressing CAST and myc-TBP 38Q or 64Q (Suppl. Fig. S10a, Supplementary File 1). Quantitative analysis reproduced the previous findings of the mainly cytoplasmic localization of the fragments c1 and c2, showing that levels of both breakdown products were significantly lowered in the cytoplasm but not in nuclear fraction upon CAST overexpression (Suppl. Fig. S10b, Supplementary File 1). To reproduce the findings using a pharmacological strategy in an additional cell model, we treated PC12 cells expressing TBP 13Q and 105Q with CI-III for 24 h or 48 h. Western blot analysis using the 58C9 antibody demonstrated a markedly reduced fragmentation of both TBP 13Q (Fig. 8e, f) and 105Q (Fig. 8g, h), which appeared to be slightly more effective for the polyQ-expanded TBP (Fig. 8f, h).
As fragment levels for expanded- and unexpanded TBP were modulated likewise by the calpain inhibition, we then assessed its repercussion on TBP aggregate formation and cell viability. For aggregate analysis, filter retardation assays and denaturing detergent agarose gel electrophoresis (DD-AGE) of protein extracts from HEK 293T expressing TBP 105Q and CAST (Fig. 9a–c; Suppl. Fig. S11, Supplementary File 1) as well as from PC12 cells expressing TBP 105Q treated with CI-III for 24 h or 48 h (Fig. 9d, e) were performed, using the aggregate-detecting TBP antibody N-12. Results from both cell lines showed a significant decrease in SDS-insoluble species of polyQ-expanded TBP. In HEK 293T cells, this effect was accompanied by a significant rescue of the impaired viability in cells co-expressing TBP 105Q and CAST, as detected by a resazurin-based assay (Fig. 9f).
Taken together, we demonstrated that calpain inhibition ameliorates TBP cleavage and aggregation, and rescues impaired cell viability, which all represent typical hallmarks of SCA17’s molecular pathology.