Stress Granules and Neurodegenerative Disorders: A Scoping Review

Cytoplasmic ribonucleoproteins called stress granules (SGs) are considered as one of the main cellular solutions against stress. Their temporary presence ends with stress relief. Any factor such as chronic stress or mutations in the structure of the components of SGs that lead to their permanent presence can affect their interactions with pathological aggregations and increase the degenerative effects. SGs involved in RNA mechanisms are important factors in the pathophysiology of neurodegenerative disorders such as amyotrophic lateral sclerosis (ALS), frontotemporal degeneration (FTD), and Alzheimer's diseases (AD). Although many studies have been performed in the field of SGs and neurodegenerative disorders, so far, no systematic studies have been executed in this field. The purpose of this study is to provide a comprehensive perspective of all studies about the role of SGs in the pathogenesis of neurodegenerative disorders with a focus on the protein ingredients of these granules. This scoping review is based on a six-stage methodology structure and the PRISMA guideline. A systematic search of seven databases for qualified articles was conducted until December 2020. Publications were screened independently by two reviewers and quantitative and qualitative analysis was performed on the extracted data. Bioinformatics analysis was used to plot the network and predict interprotein interactions. In addition, GO analysis was performed. A total of 48 articles were identified that comply the inclusion criteria. Most studies on neurodegenerative diseases have been conducted on ALS, AD, and FTD using human post mortem tissues. Human derived cell line studies have been used only in ALS. A total 29 genes of protein components of SGs have been studied, the most important of which are TDP-43, TIA-1, PABP-1. Bioinformatics studies have predicted 15 proteins to interact with the protein components of SGs, which may be the constituents of SGs. Understanding the interactions between SGs and pathological aggregations in neurodegenerative diseases can provide new targets for treatment of these disorders.


INTRODUCTION
Cell function is divided between the organelles sited inside the cell. Based on the presence of lipid membrane, organelles can be divided into main two groups. Nucleus, mitochondria, endoplasmic reticulum, and Golgi apparatus are the major membranous organelles. Membraneless organelles are ribosomes (Turi et al., 2019), stress granules (Arrigo et al., 1988), p-body (Sheth and Parker, 2003), and nucleolus (Shaw and Jordan, 1995) that are formed during a process called the liquidliquid phase separation phase (Marnik and Updike, 2019). The existence of stress at the cellular level can lead to a variety of responses, including global translational inhibition, leading to the formation of stress granules (SG) (Aulas et al., 2018). SGs, membraneless ribonucleoproteins containing mRNA, are cytoplasmic accumulations being stopped at the initiation of translation and disappear after the end of stress induction (Boncella et al., 2020). These stresses in mammalian cells include viral infections (biotic stress), induction of redox stress with sodium arsenite, heat and UV radiation which are environmental stress conditions (Kedersha et al., 2013). Three groups can form the protein component of stress granules: RNA-binding proteins, translation initiation factors, and non-RNA-binding proteins (Cao et al., 2020;Samadian et al., 2021). Stoppage at the critical stage of translation initiation due to biotic or environmental stress leads to the isolation of the translating polysomes resulting in the creation of a huge reservoir of RNA and related proteins that build and increase the number of SGs. On the other hand, relieving stress and increasing translated mRNAs is associated with disassembly and reduction in the number of these granules (Panas et al., 2016;Marnik and Updike, 2019).
There are two mechanisms for stopping translation initiation at the cellular level. Phosphorylation of the α subunit of eIF2 transcription initiation factor and prevention of the eIF-4F complex assembly. eIF2 is present in the ternary complex and is responsible for transferring the initiator tRNA to the preinitiation complex at the 5'-ends of mRNAs. The result of eIF2 phosphorylation is reduction in its binding to GTP and loss of its ability to transfer the initiator tRNA to ribosomes for start codon recognition. Four stress associated kinases (HRI, PERK, PKR, GCN2) have the ability to phosphorylate the α subunit in the eIF2 factor (Aulas et al., 2017;Wolozin and Ivanov, 2019). eIF-4F complex is responsible for detecting the structure of the cap at the 5' mRNA end, and assembly of this complex is controlled by the PI3K-mTOR kinase cascade. eIF4E is inactivated in phosphorylated form and this phosphorylation is performed by mTOR, a member of the phosphatidylinositol 3-kinase-related Abbreviations: SGs, Stress granules; ALS, amyotrophic lateral sclerosis; FTD, frontotemporal degeneration; AD, Alzheimer's diseases; MS, multiple sclerosis; PRISMA-ScR, Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews; MeSH, medical subject heading; GO, Gene ontology; FTLD, frontotemporal lobar degeneration; NIFID, neuronal intermediate filament inclusion disease; BIBD, basophilic inclusion body disease; MND, motor neuron disease; IHC, immunohistochemistry; IP, immunoprecipitation; MNs, motor neurons; LMN, lower MNs; TDP-43, TAR DNA-binding protein 43; HRE, Hexanucleotide repeat expansion; GEF, guanine nucleotide exchange factor; hnRNPA1, Heterogeneous nuclear ribonucleoprotein A1. kinase family of protein kinases (Mitra et al., 2015). eIF4E in active mode prevents the eIF-4F complex assembly and the translation process halts at the initiation point (Gingras et al., 1999;Aulas et al., 2017;Wolozin and Ivanov, 2019).
No specific function can be considered for SGs, but they can be considered as a "decision point" (Wolozin and Ivanov, 2019) for the fate of mRNA trapped in its structure. SG can define the fate of an mRNA that could be subjected to storage, degeneration, or re-initiation of translation. The probability of mRNA being in the SGs structure is sequence-independent but is directly related to the low translatability and the increase in the length of encoded region as well as the UTR region in the mRNA (Khong et al., 2017). Since proteins and RNAs with important roles can be included in SGs, this would affect biological interactions (Arimoto-Matsuzaki et al., 2008).
Traces of SGs have been found in many diseases, such as cancer (Gao et al., 2019), neurodegenerative disorders (Wolozin, 2012) and autoimmune conditions (McCormick and Khaperskyy, 2017). Neurodegenerative diseases include as amyotrophic lateral sclerosis (ALS) (Baron et al., 2013), Alzheimer's disease (AD) (Ash et al., 2014), and multiple sclerosis (MS) (Salapa et al., 2018). The neurodegeneration process involves atrophy and loss of neuronal activity (Wolozin and Ivanov, 2019). In neurodegenerative diseases, many mutations and misfolding events have been identified in this protein components of SGs, which can lead to the accumulation of abnormal proteins and the formation of SGs. Pathological symptoms appear when the presence of SGs becomes permanent due to an increase in their formation resulting from chronic stress (Brown et al., 2020) and a decrease in their deletion due to mutations in genes involved in the process of autophagy (Chitiprolu et al., 2018;Brown et al., 2020).
So far, many studies have been done on the nature of SGs, their components, structures and their pathological characteristics in various neurodegenerative diseases, and useful results have been obtained. In this study, we tried to establish a strong correlation between clinical evidence and genetic characteristics in neurodegenerative diseases in the form of a scoping review study by summarizing all human clinical studies and humanderived cell lines in the field of SGs.

General Framework for Review
The writing strategy of this article is based on the methodology proposed by Arksey and O'Malley (2005). This strategy was later improved by Levac et al. (2010) and Colquhoun et al. (2014). In this review, 5 steps of the 6-step framework have been followed which includes (1) Development of research questions, (2) Search strategy, (3) Study eligibility criteria, (4) Data extraction, (5) Collating, summarizing and reporting the results. The sixth step, consultation, is optional and is not included in this article. Also, in order to observe the principles of clarity and transparency in writing the article, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) Checklist has been used well (Tricco et al., 2018).

Development of Research Questions
To survey, summarize and discuss the studies on SGs in human and human-derived cell lines in neurodegenerative diseases, our review was guided by the following questions:

Study Eligibility Criteria
Studies in neurodegenerative diseases in relation to SGs in humans or in human-derived cell lines were screened from publications obtained during the search process. All types of publications were reviewed, including journal articles, conference presentations, conference abstracts and reports. Non-English articles with English abstracts were also included. Screening was done in two stages. At first, both researchers (MRA, MSM) screened articles separately based on title and abstract, according to the inclusion criteria mentioned above. In the next step, full texts of the selected articles were investigated to measure its relevance to the research question. Finally, appropriate articles were selected based on the eligibility criteria. Any contradiction was resolved in agreement with the opinion of the third person.

Data Extraction
Two separate charts were designed for human samples and human-derived cell lines in Microsoft Excel to help extracting the data. Chart related to human sample articles included author's name, year of publication, diagnosed neurodegenerative disease (number of patients), age and country, sample and method of analysis, protein component of SGs and major findings. The chart related to human-derived cell lines articles included the author's name, year of publication, origin of cell lines, age at biopsy, country, sample, method of analysis, mutation, and major findings. Two researchers (MRA, MSM) separately extracted data from articles based on charts.

Collating, Summarizing, and Reporting the Results
Quantitative and qualitative analysis was performed on the data obtained from the publications. In the quantitative analysis section, a descriptive numerical summary of the extent, nature and distribution of the studies were reviewed. In the qualitative analysis section, based on the research question mentioned earlier, a narrative review was performed on the available information with affirmation on the broader context suggested by Levac et al. (2010).

Bioinformatics Analysis
Two disease-protein interaction networks were designed using Cytoscape v3.8.0 software (Shannon et al., 2003) based on data extracted from articles. One was the network of neurodegenerative diseases with protein components of SGs and their interactions according to the results of human studies and the other was the protein components of SGs studied in humanderived cell lines from ALS. Therefore, the gene/proteins found from the literature search were used as input. The output was the interaction network between these proteins. Gene ontology (GO) analysis was performed using Enrichr's web-based tools and services (Kuleshov et al., 2016) on the genes of SGs protein components in neurodegenerative diseases and ALS. Proteinprotein interaction was predicted using the string-db cytoscape plugin (Doncheva et al., 2019) on the data extracted from the articles. The analyzed graph of GO was ranked based on p < 0.05.

RESULT
Keyword-based searches in various databases yielded 1,087 related records. Also, 7 records from other sources were added to the results. Of these, 758 duplicates were identified. After excluding the duplicates, the number of articles with potential to be related to the research question reached 329. A total of 262 articles were excluded from the study by screening titles and abstracts and 61 articles remained. By reviewing the full text of the remaining articles, 49 articles were included in the study, of which 5 articles were conference abstracts. The process of selecting eligible studies is detailed in the flowchart in Figure 1. Table 1 provides an overview of the included articles.
• TIA-1-positive SGs were observed in both microglia, and astrocytes cell types. • Ataxin-2 is present in the structure of SGs, which rapidly respond to stress and insults that affect the cell, and preventing translation of incorporate mRNA. • In controls, ataxin-2 was rarely observed in FUS inclusions Whereas in the ALS's, ataxin-2 was often seen with FUS cytoplasmic inclusions. • IP revealed that FUS was co-precipitated with ataxin-2 and not in controls. • Interaction between ataxin-2 and FUS is also RNA-independent, which was completed with RNase treatment.    • Examination of SFPQ in AD's brain tissue showed down-regulation. • SFPQ in AD's brain showed nuclear depletion and cytoplasmic colocalization with TIA-1. • SFPQ showed extra-nuclear colocalization with p-Tau in AD's brain legions. • There is probably a link between SFPQ and Tau oligomers in oligomerization and misfolding.

SGs: From Assembly to Disequilibrium and Pathogenesis
Inhibition of translation due to stress is the main factor for SGs assembly. The size of the SGs can vary from 100 to 1,000 nm (Kedersha et al., 1999). SG assembly needs inhibition of the assembly of polysomes so that only the 40S ribosome subunit remains attached. Non-translated mRNAs are attached to other RNA binding proteins at the nucleation phase in the SGs formation. These proteins at the nucleation phase, also known as SG nucleators, are generally multi-domain proteins in which the IDR "intrinsically disordered region" and RBD domains are prominent (Kedersha et al., 2005) and facilitate the assembly of SGs (Gilks et al., 2004). The nucleation phase continues with the formation of a repository of mRNPs in situ, where it is accompanied by interactions between RNA-RNA and RNA-protein to create a stable form of SG cores. Increasing the interactions with the growth of these cores brings the concentration to the critical level, initiates the phase changes (Kedersha et al., 2005;Molliex et al., 2015;Patel et al., 2015) and creates the primary biphasic core/shell structures. The composition of biphasic SGs creates a larger mature assembly during the microtubule-dependent process (Wheeler et al., 2016). The opposite point of assembling SGs is disassembly. In general, disassembly involves the return of non-translated mRNAs trapped in the SG structures to the translation process and is caused by a variety of factors including chaperones in the stress relief phase (Mazroui et al., 2007), microtubules (Loschi et al., 2009), autophagy mechanisms (Dormann et al., 2010), and posttranslational modifications . Assembling and disassembling of SGs are in equilibrium with polysomes (Kedersha et al., 2000). Since biotic or environmental stress stops translation and increases the number of SGs, relieving stress and increasing the number of translated mRNAs is associated with return to equilibrium and disassembly of SGs (Panas et al., 2016;Marnik and Updike, 2019). When the condition progresses toward disequilibrium, by increasing the assembly of SGs and decreasing the clearance, pathogenic processes would be evolved (Wolozin, 2012;Chen and Liu, 2017) (Figure 3). This pathogenesis links SGs to a wide range of neurodegenerative diseases including ALS (Colombrita et al., 2009;Volkening et al., 2009;Farg et al., 2013;Manghera et al., 2016;Dreser et al., 2017;Chen and Cohen, 2019), AD (Castellani et al., 2011;Vanderweyde et al., 2012;Ivanov et al., 2016;Maziuk et al., 2018;Silva et al., 2019;Younas et al., 2020), FTD (Hirsch-Reinshagen et al., 2017;Mackenzie et al., 2017;Hock et al., 2018;Montalbano et al., 2020), and FTDP (Vanderweyde et al., 2012). Many of the protein components of SGs such as TIA-1 (Fujita et al., 2008;Volkening et al., 2009;Liu-Yesucevitz et al., 2010;Vanderweyde et al., 2012;Dafinca et al., 2016;Gal et al., 2016;Hirsch-Reinshagen et al., 2017;Mackenzie et al., 2017;Salapa et al., 2018;Chen and Cohen, 2019;Younas et al., 2020), PABP-1 (Fujita et al., 2008;Dormann et al., 2010;Bentmann et al., 2012;McGurk et al., 2014;Dafinca et al., 2016; (Farg et al., 2013;McGurk et al., 2014;Japtok et al., 2015;Lenzi et al., 2015;Daigle et al., 2016;Lim et al., 2016;Lo Bello et al., 2017;Hock et al., 2018;Kamelgarn et al., 2018;Arenas et al., 2020) with SGs in neurodegeneration, indicating a strong association of SGs with the pathogenic mechanisms. In this study, by reviewing all studies on SGs and neurodegenerative diseases in humans, we tried to answer the question that SGs can act as nests or sources for these pathological aggregations, or disruption and mutations in the main components of these accumulations can disrupt the balance of SGs.

ALS and FTD Disorders and SGs
ALS is a neurodegenerative disease specific to motor neurons (MNs), and with progressive loss of upper MNs (UMN) in the motor cortex of the brain and lower MNs (LMN) in the brain stem and the spinal cord (Robberecht and Philips, 2013), muscle weakness and atrophy appear (Hardiman et al., 2017). The main pathological signature of ALS is the presence of inclusion bodies in the cytoplasmic region of MNs. The key component of these inclusions in 95% of cases is TDP-43 in the form of hyperphosphorylated, Ubiquitinated, and truncated (Neumann et al., 2006;Nonaka et al., 2016). FUS and SOD1 are other proteins that can be involved in the formation of these inclusions (Volkening et al., 2009;Farg et al., 2013). TDP-43 and FUS are RNA-binding proteins that are often nuclear localized but can commute between the nucleus and the cytoplasm (Kapeli et al., 2017). The nuclear activity of these proteins is summarized in transcription, pre-mRNA splicing and processing non-coding RNAs (Ederle and Dormann, 2017). In the cytoplasm, these proteins can contribute in the regulation of mRNA stability, mRNA transport, translation, autophagy, and stress response and LLPS (Birsa et al., 2020). SOD1 is a superoxide dismutase enzyme. Mutations in this gene can result in these accumulation (Volkening et al., 2009). Based on bioinformatics studies and overlap in the function of these proteins in pathological aggregations and impairment of mRNA mechanisms, these proteins can be considered as pathological factors in these disorders (Liu-Yesucevitz et al., 2010;Wolozin, 2012;Ramaswami et al., 2013). One of the symptoms of impaired mRNA mechanisms is the disequilibrium of SGs. Numerous studies have been performed on the presence of SGs and their association with pathological aggregations in humans using post mortem tissue (brain, spinal cord) and blood, which are summarized in Table 2. To investigate the presence of SGs in the target tissue, SG markers were used, which include TIA-1, G3BP1, PABP-1, TIAR, HuR, and TTP. Studies of human-derived cell lines have been mostly performed using autopsy skin fibroblasts and differentiation into motor (Dimos et al., 2008). These studies have allowed the study of mutations in the TDP-43, FUS, SOD1, and C9orf72 genes and their effects on cell and other components of pathological aggregations and SGs, which are summarized in Table 3.
TAR DNA-binding protein 43 (TDP-43), product of the TARDBP gene, plays the largest role in the formation of pathological aggregations. Most of the changes observed in this protein in aggregations include phosphorylation (Liu-Yesucevitz et al., 2010;Bentmann et al., 2012;Cohen et al., 2015;Hirsch-Reinshagen et al., 2017;Ratti et al., 2020), acetylation (Cohen et al., 2015), and cleavage at N/C terminals (Bentmann et al., 2012;Cohen et al., 2015). General observations suggest that hyperphosphorylation and acetylation may predispose TDP-43 to accumulation. TDP-43 has been studied in both full-length and cleaved (Cohen et al., 2015)   • Abnormal accumulation of proteins and stress granules was observed in c9orf72 IPSC-derived MNs. • Decreased survival of these cells may be due to disruption of mitochondrial membrane potential and calcium homeostasis, increased ER stress and decreased BCL2 protein levels. Dafinca et al., 2016 (Continued) Frontiers in Aging Neuroscience | www.frontiersin.org  TDP-43 • In the early stages of neuronal differentiation, no difference was seen between TDP-43 mutant cell lines and WTs. • Neuronal loss and pathological neurofilament abnormalities were seen in the aging stage in mutant TDP-43 cell lines. • Abnormal phenotypes in terms of shape, size, and motility were observed in mitochondria and lysosomes that were not due to mis-localization or accumulation of TDP-43 in the motor neurons carrying the TDP-43 mutations in the aging phase. • Axon trafficking in motor neurons was improved by D-sorbitol, but no TDP-43 accumulations or mis-localizations were observed. • S393L and G294V Mutations can cause motor neuron degeneration but are independent of the cytoplasmic accumulation of TDP-43. Kreiter et al., 2018 (Continued) Frontiers in Aging Neuroscience | www.frontiersin.org • Genetics play a key role in responding to cellular stress and the size and number of SGs.
• Disruption of autophagy mechanisms may be involved in TDP-43 aggregations.
• Arsenite stress relief was associated with a decrease in number of SGs and TDP-43 aggregations in 72 h, but P62 remained, so Disruption of autophagy mechanisms may be involved in TDP-43 with SGs, it is most often colocalized with TIA-1, PABP-1, G3BP1, TIAR, and HuR, which are markers for the presence of SGs in cell and are the core proteins in nucleation phase in the assembly of SGs (Kedersha et al., 2005). In addition, in MNs differentiated from fibroblasts, the effect of TDP43 on nucleation phase and interaction with G3BP1 in CORE formation has been determined (Orrù et al., 2016). TDP-43 is colocalized with STAUFEN and FMRP, which play important roles in mRNA mechanisms. STAUFEN (Volkening et al., 2009;Vassileff et al., 2020), encoded by the STAU1 gene, is involved in the transport of mRNA to different subcellular compartments and organelles (Thomas et al., 2005). FMRP is involved in the nucleocytoplasmic shuttling and dendritic localization of mRNA (Antar et al., 2005). So far, the effects of TDP-43 mutants including S393L (Kreiter et al., 2018), G294V (Kreiter et al., 2018), M337V (Loginov et al., 2018;Feneberg et al., 2020), I383T (Loginov et al., 2018) and TARDBP-A382T (Orrù et al., 2016;Loginov et al., 2018;Ratti et al., 2020) mutations in human-derived cell line studies have been considered. S393L, G294V missense mutations can cause neurodegeneration in MNs in a TDP-43 accumulation independent manner (Kreiter et al., 2018). The TDP-43 M337V mutant can increase the assembly of SGs by interfering with the function of eIF4A1 and endoplasmic reticulum chaperone Grp78 (Feneberg et al., 2020). In contrast, TARDBP-A382T mutation due to loss of TDP-43 function significantly has reduced the number of SGs in cells (Orrù et al., 2016). FUS (fused in sarcoma) is present in pathological aggregations with a lower percentage than TDP-43 and its association and effects on SGs have been studied. The mechanism of FUS toxicity is not fully understood, but due to its cytoplasmic localization, loss of nuclear activity and acquisition of cytoplasmic function might be involved (Kino et al., 2015). Studies in post-mortem tissue have shown the association of FUS aggregations with SGs through colocalization with ATAXIN2 (Farg et al., 2013) and PABP-1 (McGurk et al., 2014). FUS aggregations are directly affected by the type of mutation, benign and malignant (Japtok et al., 2015). The amount of mislocalization and recruitment in the structure of SGs is related to the type of mutation (Lenzi et al., 2015). FUS in the cell carrying the P525L mutation has more cytoplasmic localization than in control, and when exposed to stress, this localization increases leading to nuclear depletion (Lim et al., 2016). After stress relief, the mutant carrier cell needs more time to return to normal than control. SGs are more numerous in this cell and have a longer persistence, which indicates the direct effect of FUS mutations on SGs (Lo Bello et al., 2017). Post-transcription modifications, such as acetylation in lysine 510, which is located in the NLS sequence, disrupts the interaction between FUS and transportin1, causing its cytoplasmic mislocalization, which is more common in the pathogenesis of ALS than in controls (Arenas et al., 2020). The localization of SOD1 has been confirmed in both mutant and WT form with TDP-43 accumulations in spinal cord motor neurons (Volkening et al., 2009). Mutations that occur in SOD1 can also affect the dynamics of SGs. The mutant types SOD1-L144F (Gal et al., 2016) and SOD1-L39R (Rajpurohit et al., 2020) colocalize with G3BP1 in fibroblast-derived motor neurons, whereas SOD1-WT does not colocalize with SGs (Gal et al., 2016;Rajpurohit et al., 2020).
Hexanucleotide repeat expansion (HRE) is another common mutation in ALS that is associated with an increase in the number of G4C2 repeats in the C9orf72 gene. The number of repeats in normal people are between 20 and 30, but in people with mutations, the number of repeats increases to hundreds (Khan et al., 2012). Three mechanisms explain the effect of C9orf72 mutations on SGs.
1. C9orf72-related RNA transcripts accumulate in the nucleus and cytoplasm causing sequestration of RNA-binding proteins, including proteins involved in SG dynamics (Rossi et al., 2015;Dafinca et al., 2016). Colocalization of PABP-1 with TDP-43 inclusions was higher in ALS patients with C9orf72 mutation (67%) than in patients with ATXN2 mutation (47%) and patients without any mutation (36%) (McGurk et al., 2014). 2. The effect of mutations on C9orf72 protein function and the destruction of interactions with other proteins is another proposed mechanism. DENNL72 is another name for C9orf72, which stands for differentially expressed in normal and neoplastic cells and One of the molecular roles envisaged for it is acting as a guanine nucleotide exchange factor (GEF) (Levine et al., 2013). The interaction of C9orf72 as GEF with Rab proteins (Tang, 2016) is involved in autophagy and cellular trafficking mechanisms such as Rab7 and Rab11 (Farg et al., 2014) and disruption of these pathways can lead to cytoplasmic accumulations of TDP-43 (Ratti et al., 2020) and decreased clearance of SGs (Monahan et al., 2016). 3. The transcript of the mutant C9orf72 gene containing GGGGCC repeats can be translated during the non-ATG translation mechanism and produce 5 different types of dipeptide repeats (DPRs) (Mori et al., 2013). Among these, arginine-containing dipeptide repeats can interact with a number of SG protein components that have the IDR domain. DPRs containing glycine and proline also play a role in the assembly of SGs by inhibiting the translation and phosphorylation of eIF2a and G3BP1 .
FTD is one of the most common types of dementia that affects people under the age of 65 (Bird et al., 1999) and represents a diverse range of subtypes with neurodegenerative disorders such as FTLD (Faber, 1999). Due to common characteristics in clinical observations, ALS and FTD now form a broad continuum of neurodegeneration that can occur in an individual or a family. Like ALS, studies on post-mortem tissue, specifically the brain of FTD patients, have revealed the association and development of the disease with SGs (Dormann et al., 2010;Liu-Yesucevitz et al., 2010;Hirsch-Reinshagen et al., 2017;Mackenzie et al., 2017). Most pathological mechanisms and aggregations such as TDP43, FUS, and C9orf72 mutations overlap with ALS. But FTD has some distinctive features. No colocalization was observed between PABP-1 and eIF4G with cytoplasmic neuronal inclusions in FTLD-TDP brain (Dormann et al., 2010), and acetylated and full-length pTDP-43 have no effects in the pathogenesis of FTLD-TDP (Cohen et al., 2015).

AD and SGs
AD is a chronic neurodegenerative disease that is accompanied by death of neurons and loss of synapses in the cerebral cortex and certain subcortical regions of the brain (Tiraboschi et al., 2004;Burns and Iliffe, 2009). The most common form of dementia is AD, which occurs with abnormal structures (Wang et al., 2018), extracellular senile plaques, composed mainly of small proteins called Aβ42 (Bate et al., 2006), and intraneuronal neurofibrillary tangles, which are the result of accumulations of hyperphosphorylated Tau proteins (Bancher et al., 1989). Tau proteins are a group of six protein isoforms produced by alternative splicing of the MAPT (microtubule-associated protein Tau) gene (Goedert et al., 1989). The main function of Tau proteins is to maintain the stability of microtubules in axons due to their high expression in CNS neurons (Haritani et al., 1994), which are highly soluble in the cytoplasm and contribute to the dynamic and functions of microtubules (Arendt et al., 2016). Tau undergoes many post-translational modifications, including hyperphosphorylation (Grundke-Iqbal et al., 1986), acetylation (Min et al., 2010), C-terminal truncation (Zhao et al., 2016), and n-glycosylation (Wang et al., 1996), which can play a role in regulating its localization and functions. Usually, Tau is nonphosphorylated in interaction with microtubules in the axons. When stress is applied, it is phosphorylated near the microtubule binding domain and loses its ability to bind to microtubules (Trinczek et al., 1995). The hyperphosphorylated form of Tau is seen in all six isoforms in neurofibrillary tangles (Hernández and Avila, 2007). Hyperphosphorylation increases the affinity of Tau proteins for each other and binds together to form oligomers and misfolded Taus. Oligomers bind to other units to form Tau deposits, which are the Elements of NFTs that bind together to form NFTs (Shafiei et al., 2017). Tau can act as a negative regulator of protein translation by binding to ribosomes and reducing protein synthesis (Meier et al., 2016). Stopping translation and providing RNA-binding proteins to mRNA are key elements in the formation of SGs. TIA-1 and TTP are among the core nucleation proteins. The size of Tau aggregations is directly related to TIA-1 colocalization. The larger the Tau aggregations, the greater the rate of TIA-1 colocalization (Vanderweyde et al., 2012). PABP-1 and DDX6 are also proteins found near Tau aggregations in the temporal cortex neurons of AD (Silva et al., 2019). The association of aggregations between RNA-binding proteins and the formation of SGs with Tau aggregates has not been well-studied. Tau aggregations might promote the dynamic equilibrium of SGs toward further assembly or to SGs disequilibrium pave the way for the accumulation of Tau proteins. The formation of RBPs pathological aggregations close to the Tau pathological aggregates supports this hypothesis (Maziuk et al., 2018). Tau proteins have interaction with rps6, which is a component of the 40s ribosome complex, and affects translation inhibition (Koren et al., 2019). Notably, hippocampal neurons in AD patients have more positive rps6 granules than controls (Castellani et al., 2011). Moreover, the colocalization of rps6 with TIA-1 and PABP-1 in basophilic inclusions may indicate one of the main shared mechanisms between Tau aggregations and stress granules (Fujita et al., 2008).
The association of TDP-43 oligomers with Tau aggregations has also been suggested as one of the interactions involved in the pathogenesis of Tau (Montalbano et al., 2020).

MS and SGs
MS is a demyelinating disease in which the cover of nerve cells in the brain and spinal cord is damaged (Noyes and Weinstock-Guttman, 2013). The causative mechanism of the disease is summarized in the destruction of myelin sheath by the immune system and defects in myelin-producing cells (Nakahara et al., 2011). The disease has three main features, including the formation of lesions in the CNS, inflammation and destruction of the myelin sheath (Compston and Coles, 2008). Neurodegeneration is also an important feature in the pathology of MS (Frohman et al., 2006;Lassmann and van Horssen, 2011), but so far, no specific mechanism has been proposed for it. The impact of SGs on neurodegeneration and neurodegenerative diseases has been fully discussed so far. Heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1) is a major component in the formation of SGs (Guil et al., 2006) and is discussed as a factor in the pathogenesis of autoimmune mediated CNS neurological diseases and as a link between SGs and autoimmune neurological diseases (Douglas et al., 2016). The interaction between TIA-1 and hnRNPA1 in cytoplasmic granules and its nuclear depletion in MS patients is significant. The aggregations of TIA-1 in the structure of large SGs in the cytoplasm can be a link between degeneration in neurons and MS (Salapa et al., 2018). Nuclear depletion of TDP-43 and interaction with hnRNPA1 and colocalization of both in the structure of SGs also emphasizes the importance of SGs in MS Salapa et al., 2020).

Bioinformatics Perspective
GO classifies relationships between genes by annotating and categorizing them into three levels: biological process, molecular function, and cellular component. Biological process describes the cellular or physiological role performed on a larger scale by a gene in relation to other genes. The molecular function describes the molecular activity of the desired gene, and the cellular component determines where the gene product executes its function (The Gene Ontology Consortium., 2019). GO analysis was performed on the list of genes associated with SGs in neurodegenerative diseases ( Figure 3A) and the list of genes associated with SGs in ALS disease was extracted from human derived cell lines studies ( Figure 3C) using Enrichr's web-based tools and services (Kuleshov et al., 2016). Most of the proteins in the structure of SGs are RBPs (Kedersha et al., 2005), and GO analysis confirms this. According to the GO biological process, the connections that can be made in one biological pathway with other proteins are more involved in the RNA metabolic pathways and, as expected, all of these proteins act in conjunction with RNAs, including cytoplasmic mRNA, body assembly, and RNA splicing. Further use of these proteins in FIGURE 4 | The interactions between SGs protein components and predicted proteins and GO analysis on predicted ones in neurodegenerative disease. (A) Fifteen new proteins were predicted by interaction with SGs protein components using the Cytoscape string-db plugin on data extracted from articles. (B) Fifteen new proteins were predicted by interaction with SGs protein components using the Cytoscape string-db plugin on data extracted from articles. GO analysis was performed on the predicted proteins in three biological processes: molecular function and cellular component. The length of each bar indicates the importance in that particular category sorted by p-value. Note that the lower the color intensity of the bars, the greater the relationship with that category. the structure of ribonucleoprotein granules, cytoplasmic SGs and cytoplasmic ribonucleoprotein granules was confirmed by GO cellular component analysis. In ALS, the biological association of these proteins with other proteins, based on the GO biological process, significantly confirmed their role in RNA export from nucleus. This was expected, given that these proteins are mostly shuttling between nucleus and cytoplasm. Gene prediction was performed on the gene list extracted in neurodegenerative disease in Table 3 by string-db cytoscape plugin (Doncheva et al., 2019). GO biological process showed these proteins interact with other proteins in p-bodies, cytoplasmic ribonucleoprotein granules and cytoplasmic SGs. Overall, bioinformatics analysis determined the association of these proteins and genes with RNA-related mechanisms that are specifically involved in the formation and assembly of ribonucleoprotein granules. Their functional position is also shared between nucleus and cytoplasm and is an evidence to trafficking and shuttling of SGs protein components between them (Figure 4).

Conclusions
Equilibrium is the most important point in SGs. SGs are in the nature of the cell and are considered as cell solutions to stress. They are temporary constructions and when the stress is relieved, they are disassembled and reduced in number, and the cell condition returns to normal. If under any circumstances the presence of SGs becomes permanent and leads to disequilibrium, in interaction with pathological aggregations such as TDP and FUS aggregations, they can lead to pathological conditions such as neuron degeneration. SGs have been studied in many neurodegenerative diseases in humans, including ALS, FTD, AD, and MS. These studies have indicated common features in SG biology among neurodegenerative diseases. At least 15 proteins have been predicted to interact with the protein components of SGs in mentioned neurodegenerative disorders. Therefore, it seems that SG biology is common between these disorders. Yet, some components of SGs might be specific to these disorders. Based on the rarity of comparative analysis between these disorders, it is not possible to male conclusive interpretations in this regard. We have tried to provide a comprehensive summary of these studies and an overview of SGs in neurodegenerative diseases. To conclude, more studies can be done in diseases such as AD and MS and the association of the Tau protein with other protein components of SGs or the association of inflammatory pathways with the formation of SGs could also be assessed.

AUTHOR CONTRIBUTIONS
MT, SG-F, and MR wrote the manuscript and contributed in study design. MA, MS, HS, and AJ contributed in the data collection, designed the tables and figures. All authors approved the manuscript.