Isolation and characterization of EVs derived from R6/1 mouse striatum.
To investigate the potential effect of the cortico-striatal pathway activation, via motor skill learning, on striatal EVs profile, we subjected WT and R6/1 mice to the accelerating rotarod test, for 3 consecutive days. Half of the animals, grouped as naïve, were presented to the rod the first day but no training was performed (Fig. 1A).
Only four animals per group were sufficient to significantly reproduce the disease-associated deficits in the rotarod task, as expected, in line with our own previous work. We observed that both WT and R6/1 mice improved their performance per day, confirming they were properly trained, but the R6/1 mice had motor learning deficits, compared to WT, since the latency to fall was shorter, as previously described29 (Fig. 1B & C).
Ninety minutes after the last rotarod trial, mice were sacrificed, and striatal tissue was dissected out. Then, EVs were isolated from the striatum of both WT and R6/1 mice, either naïve or trained, by a first step of sequential UC followed by a purification by SEC, obtaining a final pool of the fractions that correspond to the peak of protein (F10-20) (Fig. 2A). We showed that the protein peak overlapped the EVs peak, as judged by NTA analysis of the particle’s concentration combined with the protein measurements (Fig. 2B). Moreover, we confirmed the size and shape of small EVs using TEM, in our four conditions (WT / R6/1 ± training) (Fig. 2C). Furthermore, we characterized the different fractions obtained in the purification steps biochemically, by WB (homogenates (Hom), apoptotic bodies (P2000), large EVs (P10K) and small EVs). We confirmed that the EVs fraction was enriched in Alix, Flotillin-1 and TSG101, specific EVs markers, in comparison to the other fractions. Note that Alix and TSG101 are specific markers for exosomes, while Flotillin-1 can be found both in exosomes and in microvesicles46. The EVs fraction was also negative for the mitochondrial protein TOMM20 (Fig. 2D). Importantly, EVs fraction would contain EVs derived from all the neural cells naturally present in the striatum, cortical afferents, and striatal neurons but also astrocytes, oligodendrocytes, and microglia47.
Motor learning differently modulates the size and the concentration of striatal R6/1 EVs in comparison to WTs.
To further characterize EVs populations in WT and R6/1 mice, with or without physical training, we assessed the distribution in size and particle concentration of the four groups by NTA (Fig. 3A). Although total particle concentration did not show differences between groups (Fig. 3B), we observed that R6/1 mice presented a lower mean size of the EVs particles than WT, and motor training mildly corrected this size alteration in R6/1 (Fig. 3C). In the literature, many different types of EVs have been described, mostly classified by biogenesis and size as oncosomes, apoptotic bodies, microvesicles, large exosomes, small microvesicles and exomeres (Fig. 3D)26. Exclusively considering the size classification, our EVs samples mostly contain microvesicles (0.1-1 µm), large exosomes (90–120 nm), and small exosomes (60–80 nm), as reported by the size distribution of the four groups of EVs (Fig. 3A). Considering the particles in the range of 65 to 85 nm as small exosomes, we observed that R6/1 mice showed an increase in the concentration of this population in the striatum, in comparison to WT mice. This alteration was completely corrected when R6/1 mice learned the motor task (Fig. 3E). On the other hand, the concentration of the large exosome’s population (vesicles in the range of 85 nm to 125 nm) was higher in the R6/1 mice versus WT but was insensitive to motor skill learning in both genotypes (Fig. 3F).
Striatal EVs proteomic signature reflects the signaling and metabolic alterations in R6/1 mice.
To investigate whether WT and R6/1 mice striatal EVs differ in their protein cargo, we assessed the proteome of naïve WT and R6/1 striatal EVs. When we compared the whole proteomic signature, we found a significant separation of the two groups in the PCA, constructed with top variables based on a PLS-DA analysis (Figure S1A). Indeed, the heatmap summarizes all the differentially expressed proteins in striatal EVs from the two naïve groups (Fig. 4A1). Remarkably, the most overexpressed proteins in R6/1 striatal EVs were ferritin, dihydropyrimidinase-like 3 protein (DPYSL3) and albumin.
Using the KEGG database48–50 with all the protein data, we extracted the biological pathways that were significant: long-term potentiation, long-term depression, ErbB/ERK signaling pathway, cAMP signaling pathway and pathways of neurodegeneration (Fig. 4A2, Supplementary Table 1). Interestingly, the alteration of these pathways has a crucial role in the pathogenesis of HD51,52.
To study the effect of motor learning on R6/1 mice, we compared the protein cargo of striatal EVs from naïve or trained R6/1 mice. Again, PCA plots revealed that motor training was sufficient to modulate the protein content of EVs in R6/1 mice (Figure S1B). The heatmap showed a general upregulation of differentially expressed proteins after the rotarod training in the R6/1 animals (Fig. 4B1). In this case, we found significant alterations in metabolic pathways (Fig. 4B2, Supplementary Table 2). Indeed, the proteins that presented higher levels in striatal R6/1 EVs were the muscle isoenzyme phosphofructokinase (PFKM) and phosphoglycerate mutase 1 (PGAM1), both involved in the glycolytic pathway. Interestingly, decreased levels of PGAM1 have been found in the brain of HD patients (Huntington’s Disease_CNS-Brain (MMHCC)_GSE857, Harmonizome 3.0), revealing a potential beneficial function of motor learning in the modulating the molecular composition of striatal EVs.
Hence, we showed that motor skill learning did not mask HD alterations in metabolism53 in the EVs from the trained R6/1 mice.
To investigate whether motor learning could also influence striatal EVs protein cargo in WT mice, we assessed EVs protein content of naïve and trained WT striatal EVs. PCA plot revealed that motor learning could not separate striatal EVs from naïve or trained WT mice, as judged by the lack of sample group clustering (Figure S1C). However, pairwise comparisons of the proteomic data of naïve and trained WT striatal EVs identified several differentially expressed proteins in EVs after the training (Figure S2). Although we did not find significant alterations in general biological pathways (Supplementary Table 3), we observed that after learning the motor task, there was a lower expression of proteins involved in protein translation, such as seryl-aminoacyl-tRNA synthetase (SerRS)54, or in plasticity and metabolism such as synaptosomal-associated protein 25 (SNAP25), phosphoglycerate kinase 1 (PGK1), protein kinase cAMP dependent regulatory (PRKAR2B) and nipsnap2 homolog 2 (NIPSNAP2)55–58 (Figure S2).
Interestingly, when we compared trained WT and R6/1 groups, PCA plot confirmed that the two groups did not differ in the protein content (Figure S1D). The heatmap revealed mostly upregulated proteins (Fig. 4C1), that resulted in an alteration in pathways related with neurodegeneration and Parkinson’s disease (Fig. 4C2, Supplementary Table 4).
When we plotted the four groups together (WT / R6/1 ± training), the PCA in three dimensions (3D) completely clustered EVs content per genotype (naïve WT and naïve R6/1) but not by motor learning, meaning that acquiring the task brings closer the protein content of R6/1 EVs to either the naïve or the trained WT EVs (Fig. 5).
Indeed, the pairwise comparison of naïve WT and trained R6/1 derived striatal EVs showed no clustering regarding EVs protein content, suggesting, again, an evident effect of motor training in R6/1 mice EVs proteomic composition (Figure S3).
Motor learning training restores normal levels of ERK2 and β-globin proteins in striatal EVs and has a mild effect on cell survival and synaptic plasticity pathways.
To further investigate the potential beneficial role of motor learning via EVs, we assessed the levels of the proteins that were shared between the four groups of study. Using an UpSet plot, we reported two proteins that were shared in both comparisons of interest, that resulted to be ERK2 (Mapk1) and β-globin (Hbb-bs) (Fig. 6A). We observed that both proteins were reduced in striatal EVs from naïve R6/1 mice, but motor learning reverted their levels (Fig. 6B & C). These results highly indicate that learning a motor task affects directly the striatal EVs content and corrects specific signaling deficits in an HD mouse model.
Since R6/1 mice striatal EVs showed a disruption in biological pathways involved in synaptic plasticity and cell survival51,52 (Fig. 4A2), we investigated whether we could observe these effects in the recipient structure, the striatum, from the same animals, by WB. We could not observe significant differences in survival/plasticity readouts59–61, such as the phosphorylated levels of ERK (Fig. 7A) in the striatal homogenates of the four groups (WT / R6/1 ± training). Although the levels of phospho-ERK1 remained unaltered between conditions (Fig. 7A1), we observed non-significant mild tendencies in the recovery of phospho-ERK2 after training in the R6/1 mouse group (Fig. 7A2), in line with our observations of the ERK2 levels in striatal EVs (Fig. 6C). Interestingly, we confirmed the expected elevated levels of phospho(S473)-Akt in R6/1 mice striatal lysates29,62, and this was partially corrected in the R6/1 mice after learning a motor skill (Fig. 7B1). Finally, we observed that phosphorylation of RPS6 (Ser235/236) was sensitive to motor learning in both WT and R6/1 mice, independently of their genotype (Fig. 7B2).
These results indicate that motor learning tasks in R6/1 mice directly influences the striatal EVs composition, which could affect their function, and therefore might have a resilient impact on cell survival and synaptic plasticity pathways.