Formation of Toxic Oligomeric Assemblies of RNA-binding Protein: Musashi in Alzheimer’s disease

Alzheimer’s disease (AD) is the most common neurodegenerative disorder associated with structural and functional alterations of brain cells causing progressive deterioration of memory and other cognitive functions. Recent studies demonstrate that several neurodegenerative diseases, including AD, exhibit RNA-binding proteins (RBPs) pathologies, such as TAR DNA- binding protein (TDP-43), fused in sarcoma (FUS), superoxide dismutase 1 (SOD1) and T-interacting antigen-1 (TIA-1), highlighting the role of RBPs in neurodegeneration [12, 34, 40]. Musashi proteins belong to a group of RBPs that regulate the translation of target mRNAs during neuronal development, originally observed to regulate asymmetric stem cell division in Drosophila melanogaster [42]. The role of this group of proteins is crucial to maintain the pool of adult neuronal stem cells in mammals [45]. They appear to function as translational repressors of target mRNAs encoding cell cycle inhibitory proteins, thus permitting stem cells to maintain an undifferentiated state. Pathological up-regulation of Musashi proteins has been observed in cellular transformation by repressing target mRNAs involved in the inhibition of cell proliferation, as reported in a variety of tumor cells [46], including cancer of neuronal origin [21, 50, 54]. Although the role of Musashi proteins in mRNAs regulation is clearly established, their precise subcellular location is still unclear [43]. In mammals, the two Musashi proteins: MSI1 and MSI2, are composed of 362 and 328 amino acid residues, respectively. Both MSI1 and MSI2 have two RNA-recognition motifs, RRM1 and RRM2. The RRM1 of MSI1 protein contains 20–110 amino acid residues and RRM2 contains 109–186 amino acid residues with a poly-alanine stretch of 274–281 amino acid residues. The RRM1 and RRM2 of MSI2 contain 21–111 amino acid residues and 110–187 amino acid residues with a poly-alanine stretch of 253–260 residues (Fig. 1) [30]. MSI1 is found in both cytoplasm and nucleus, whereas, MSI2 is reported to be associated with the polysomes in the cytoplasm [25, 44]. These proteins are mostly diffused throughout the cytoplasm, but can be nuclear or localized in perinuclear region as well depending on cell types [37]. The mechanisms regulating nuclear localization of Musashi proteins during differentiation are not determined yet [37]. It is still unclear if the nuclear sequestration of Musashi facilitates cytoplasmic target mRNA translation or if MSI1 and − 2 have distinct nuclear functions. Both Musashi proteins are involved in the process of maturation of exon 10+ tau transcripts in neuronal cell lines, indicating potential roles in alternative splicing of certain pre-mRNAs [14]. Among the two paralogs, the functional aspect of MSI1 protein is more extensively studied than MSI2. MSI1 is shown to bind to 3′-untranslated region of its target mRNAs and repress their translational processes [4, 22]. In addition, MSI1 has been found to control the splicing of photoreceptor-specific exons in the retina of vertebrates [41] as well as to regulate the splicing of factors involved in epithelial-luminal state [27]. MSI2 also acts as a translational inhibitor, regulating the function of hematopoietic stem cells [15]. It has also been demonstrated that MSI1 protein regulates memory loss as a part of behavioral plasticity in C. elegans [20]. In the past few years, several different RBPs have been identified demonstrating their altered functions and aggregation properties in neurodegenerative diseases [12], among which TDP-43, FUS and TIA-1 are extensively studied (Fig. 1) [11, 40, 47‐49]. Abnormal accumulation of tau, a micro-tubule binding protein pathologically characterizes a group of neurodegenerative diseases, known as tauopathies [10]. Tau is believed to bind to RNA and play a role in the quality control of RNA [24, 52, 55]. Moreover, tau interacts with RBPs, such as TIA-1 [3, 51]. More than a decade ago, MSI1 protein was found to be present in tau inclusion-bearing neurons in AD and Pick’s disease (PiD) [36]. Nevertheless, there is no study reporting the involvement of MSI2 protein in neurodegeneration. Co-accumulation of tau oligomers with TIA-1 and other RBPs has been demonstrated in a tauopathy animal model and it has been shown that reduction of TIA-1 levels led to reduced tau oligomers formation, rescuing behavioral deficits in these animals [3, 51]. Another RBP, U1 small nuclear ribonucleoprotein 70 kDa (U1-70 K) protein has also been demonstrated to co-aggregate with tau in both sporadic and familial forms of AD [5].

Herein, we sought to investigate whether the Musashi family members, MSI1 and MSI2 proteins can form aggregates in disease pathology. We observed for the first time in vitro and in ex vivo human AD brains, that these two RBPs are present in their soluble aggregated, i.e., oligomeric forms. These oligomers have been detected in mature neurons that furthermore, co-localize with oligomeric tau. Interestingly, we also observed a change in the patterns of Musashi protein signal, representing cellular distribution of these proteins depending on their association with tau protein. Our observations here suggest that Musashi proteins may have altered cellular localization and possibly dysregulated functions in AD brains, thus contributing to the deleterious effects of aggregated tau.

Over the past decades, RBPs, their dysregulation and toxic roles in neurodegenerative diseases are being actively investigated. Aggregation of many RBPs, such as TIA-1 [18], FUS, TDP43, hnRNPA1 and hnRNPA2 in Amyotrophic lateral sclerosis/frontotemporal dementia (ALS/FTD) proteinopathies are mediated by prion-related domain (PRD) [29, 39]. RBPs also possess RNA-Recognition Motifs (RRMs) by which they interact with RNA molecules. These motifs are found to be conserved for each protein. Apart from such conserved RRMs, the RBPs also possess a glycine-rich domain that is also conserved. This glycine-rich domain is hydrophobic in nature, allowing the reversible aggregation of these proteins as shown with FUS and TDP-43. These two RBPs are strongly implicated in neurodegenerative diseases, such as ALS and AD [16, 23]. Although they are nuclear proteins, cytoplasmic localization of these proteins is noted in stress granules containing aggregates. Apart from AD, the pathological inclusions of tau protein also characterize a group of neurodegenerative diseases, collectively known as tauopathies [10]. Interactions between tau and other RBPs have been demonstrated in neurodegenerative diseases. The Musashi proteins, another group of RBPs are mostly studied to play roles during neurogenesis [26]. There is only one study that had demonstrated the existence of MSI1 protein in neurons bearing tau inclusions in AD and PiD pathologies [36]. However, the occurrence of MSI2 protein and their toxic form of aggregation, i.e., oligomers have not been investigated yet in neurodegeneration. To the best of our knowledge, this is the first study demonstrating Musashi proteins’ aggregation state, specifically the oligomers in AD pathophysiology and their co-occurrence with tau oligomers. We have demonstrated that recombinant MSI1 and MSI2 proteins can be aggregated in vitro as shown for other amyloidogenic proteins, following our published protocol [31].

It is suggested that proteins present in supersaturated concentration in the cellular environment are driven to form aggregates [9]. In our study, we have observed an elevated level of MSI1 and MSI2 protein in AD brain tissues compared to the age-matched controls. Musashi proteins are predominantly present in the cytoplasm with spot distribution but are also expressed in the nucleus [37]. The sub-cellular localization of these proteins is also supported by the presence of nuclear localization and nuclear export signals. We propose two possible hypotheses by which the Musashi oligomers could exist in AD pathology and interact with tau oligomers (Fig. 8). It is known that Musashi proteins shuttle between nucleus and cytoplasm. Therefore, MSI2 might interact with tau in the cytoplasm as observed here and this may drive tau into the nucleus. We consider this pathway as “cytoplasmic hypothesis”. In AD pathology, such interaction eventually might lead to the aggregation of tau forming tau oligomers co-localizing with MSI2 protein in the cytoplasm. Proteins that are intrinsically disordered and have low complexity regions are shown to drive Liquid-Liquid Phase Separation (LLPS) and form stress granules. This phenomenon has been suggested as a compelling mechanism for the aggregation of such disordered proteins in the granules and thus, causing toxicity [38]. Tau is also reported to undergo such phase separation forming tau droplets [2]. However, whether such interaction is occurring in the stress granules or other RNA granules while carrying RNA transcripts has not been investigated in this study. Studies have highlighted the role of RBPs in toxic tau aggregation in the disease pathology. Tau is shown to bind with TIA-1, another core stress granule RBP leading to its cytoplasmic localization and aggregation. Such interaction modulates the interaction of tau with other RBPs [3, 51]. As a second hypothesis, sub-cellular localization of tau protein might also play role in Musashi protein aggregation. Though, tau is a microtubule binding protein and primarily present in the cytoplasm, the nuclear localization of tau has also been suggested to play role in AD pathology [6, 8]. Alternately, tau protein could mis-localize into the nucleus and interact with nuclear MSI1 and MSI2. Such mis-localization of tau can cause the aggregation of Musashi proteins and hinder their normal functionality, as MSI2 aggregates were observed in the cytoplasm away from the nucleus. We consider this pathway as “nuclear hypothesis”.

Tau is observed to be present in the RNA granules formed under stress, known as stress granules [7]. In a healthy state, Musashi proteins can be found in polysomes [28], and being shuttled in and out of the nucleus. We observed the accumulation of Musashi proteins in the cytoplasm and nucleus, thus indicating their altered functions, which could be attributed to the mis-localization of tau into nucleus. Alternatively, under stress condition, Musashi proteins might interact with tau in the cytoplasm. Interaction of Musashi proteins with tau might initiate the aggregation of the Musashi proteins (Fig. 8), as observed for TIA-1 protein. The role of Musashi proteins’ structures influencing their aggregation process` in neurodegeneration is not yet elucidated. Thus far, there is no study reporting the occurrence of any PRD or low complexity domain (LCD) in Musashi proteins that would indicate their aggregation propensity. Here in, we have attempted to address our observation of Musashi oligomer formation in vitro and as well as in AD brain tissues by analyzing their sequences. To identify if there is any Prion-like domain present in Musashi proteins or not, we analyzed the amino acid sequences of MSI1 and MSI2 proteins in Prion-Like Amino Acid Composition (PLAAC) software [1, 13]. The output results reveal that a stretch of 169–247 amino acid residues in MSI2 protein might have prion-like properties which can cause the aggregation of this protein (Fig. 9). However, MSI1 protein sequence did not show the presence of any such prion-like domain from the PLAAC analysis. Such observation requires further studies to confirm the presence of a prion-like region and its role in the aggregation process. Taken together, our findings suggest that Musashi proteins can form soluble amyloid aggregates and may have dysregulated functions in AD brains, thus contributing to the deleterious effects of tau protein aggregation. This study provides a novel insight in the complexity of protein aggregation in neurodegeneration and demonstrates the role of a significant but less studied group of RBPs, the Musashi proteins in brain diseases.