There are several known human NDs caused by the expansion of CAG trinucleotide repeats in the coding region of the corresponding genes. These diseases are often called “polyglutamine (polyQ) diseases” because the CAG trinucleotide encodes the amino acid glutamine (Q). To date, glutamine expansions have been described in the following human proteins: huntingtin (Htt); Sca1, Sca2, Sca3, Sca6, Sca7, and Sca17 (different types of spinocerebellar ataxia); androgen receptor protein (spinobulbar muscular atrophy); and atrophin-1 (dentatorubro-pallidoluysian atrophy). The features of the clinical manifestations characteristic of each of these diseases, determined by the type of neurons in which inclusions formed as a result of amyloidization of the described proteins, are observed ^[1][2][3][47,48,49]. Using yeast as a model system, the aggregation of the huntingtin protein has been studied in the most detail ^[4][50].

The human huntingtin locus is large, spanning 180 kb and consisting of 67 exons ^[5][51]. This locus encodes a fairly large protein (350 kDa) that performs important, but not fully understood, functions, in particular, those necessary for the normal functioning of synapses ^[6][52]. The first exon may contain a variable number of CAG repeats, normally ranging from 6 to 35. Proteins with 36 or more repeats (up to 180 repeats have been described) tend to form amyloid aggregates, which correlates with the development of HD. Importantly, the hallmark of HD is the proteolytic production of an N-terminal fragment of the huntingtin protein (containing a polyQ repeat), which forms aggregates in the nucleus and cytoplasm of affected neurons ^[7][8][53,54].

To study the influence of various factors on the aggregation of the human huntingtin protein in yeast cells, single-copy plasmids containing DNA fragments corresponding to the N-terminal region (amino acids 1-68 of the wild-type protein) of the investigated protein fused to the GFP gene were generated. These vector constructs differed from each other in the number of CAG trinucleotide repeats encoding glutamine (25, 47, 72, or 103 residues). Expression was performed under the control of a strong constitutive yeast promoter (GPD). The results showed that the degree of aggregation varied with polyQ repeat length; in particular, a protein containing 25 glutamine repeats showed no signs of aggregation, whereas when a protein containing 103 glutamine repeats was expressed, one large aggregate was formed ^[9][55]. Mutations in yeast cells inhibiting the ubiquitin/proteasome pathway did not affect the aggregation of the studied protein fragments in the series of experiments. At the same time, changes in the expression level of chaperone proteins, in particular, overexpression of Hsp40, Hsp70, and Hsp104, modulated the aggregation of fragments containing 72 and 103 glutamine repeats ^[10][11][56,57].

In the amino acid sequence of the huntingtin protein, a proline-rich region is located next to the polyQ site. Deletion of this proline-rich region effects on the number and shape of formed aggregates ^[12][13][58,59].

Interestingly, when the dependence between the intensity of huntingtin protein aggregate formation in yeast cells and the age of the cell culture was studied, the results showed that proteins containing 25Q did not aggregate during the logarithmic stage of cell growth or in aged cells. Proteins containing 103Q aggregated in both young and aged cells, while proteins containing a moderate amount of glutamine (47Q) aggregated only during cell culture senescence. Mutations in two genes (SIR2 and HSF1), the expression levels of which correlate with aging, affected the dynamics of huntingtin protein aggregation in yeast cells ^[14][60].

Currently, when yeast is used as a model organism, transcriptomic, proteomic, and metabolomic analyses are the most promising modern methodological approaches for understanding the molecular mechanisms of the formation of aggregates of proteins containing polyQ tracks, particularly the huntingtin protein ^{[15][16][17][18][19]}[61,62,63,64,65].

Parkinson’s disease, certain forms of dementia, and multisystem atrophy are characterized by the accumulation of aggregated α-synuclein in protein inclusions called Lewy bodies (LBs) ^[20][66]. This class of ND is generally referred to as synucleinopathies ^[21][22][67,68]. It is currently unclear whether the formation of LBs causes cell pathologies or acts as a protective cellular mechanism, leading to the inactivation of soluble prefibrillar oligomeric forms of α-synuclein ^{[23][24][25][26][27]}[69,70,71,72,73].

Under physiological conditions in human cells, α-synuclein is considered a presynaptic protein ^[28][74] associated with vesicles and membranes ^[29][75]. However, its exact function is not yet well understood ^[30][76]. Several studies have suggested that α-synuclein plays a critical role in enabling vesicle fusion; others have reported a chaperone-like activity ^{[31][32][33][34]}[77,78,79,80].

Human α-synuclein is encoded by the SNCA gene. The protein consists of 140 amino acids and has a molecular mass of 17 kDa. The protein has N-terminal imperfect repeats based on the amino acid motif KTKEGV. The central region of α-synuclein is strongly hydrophobic, leading to protein dimerization ^[35][36][81,82]. The N-terminus of α-synuclein is critical for oligomerization ^[37][83], whereas the C-terminus is highly acidic, facilitating protein–protein interactions ^[38][84] and possibly inhibiting aggregation ^[39][85]. The α-synuclein protein is intrinsically unfolded; in solution, it lacks a stable tertiary structure and instead forms α-helices, β-sheets, and various complex multimeric structures, such as fibrils, fibers, and aggregates ^[40][86].

Multiple wild-type alleles at the SNCA locus, such as duplications or triplications, have been shown to be associated with the familial form of the disease ^[41][42][87,88]. In addition, three missense mutations, A30P, A53T, and E46K, correlate with the autosomal dominant early form of Parkinson’s disease ^[43][44][89,90]. These mutations render α-synuclein more prone to forming amyloid fibrils ^[45][91].

It has been shown that α-synuclein in human nerve cells can undergo various posttranslational modifications, namely, phosphorylation ^[46][92], ubiquitination ^[47][48][93,94], SUMOylation ^[49][95], acetylation ^[50][96], nitration ^[51][97], and/or glycosylation ^[52][98]. Posttranslational modifications of α-synuclein are known to affect its toxicity and ability to form aggregates, but the exact contribution of different posttranslational modifications to the disease mechanism is still unclear.

There is no homolog of the human SNCA gene encoding α-synuclein in the yeast genome. It has been shown that expression of the human α-synuclein gene in yeast affects vesicle trafficking ^[53][54][55][99,100,101], induces oxidative stress ^[56][102], and causes mitochondrial dysfunction ^[57][58][103,104].

It has been demonstrated that when the human α-synuclein gene is heterologously expressed in yeast cells, the resulting protein is transported via the secretory pathway to the plasma membrane, where it forms small aggregates ^[59][60][105,106]. As the level of α-synuclein expression increases, its localization changes significantly, leading to the formation of cytoplasmic inclusions similar to the LBs found in Parkinson’s disease neurons. This is accompanied by an increase in toxicity, defined as reduced cell growth followed by cell death. The increase in toxicity caused by α-synuclein is dependent on dosage. Studies have shown that the expression of three copies of the wild-type α-synuclein gene or two copies of the α-synuclein gene with the A53T mutation under the control of the yeast GAL1 promoter leads to growth inhibition and protein aggregation ^[61][107].

Like most NDs, ALS and frontotemporal lobar degeneration (FTLD) are characterized by the misfolding and aggregation of proteins in affected neurons ^[62][108]. The aggregates formed in these diseases contain proteins such as TDP-43 (trans-activation response DNA-binding protein 43), FUS (fused in sarcoma), SOD1 (superoxide dismutase), and C9orf72 that are well-known pathologic hallmarks of ALS/FTLD ^{[63][64][65][66][67][68][69][70]}[109,110,111,112,113,114,115,116]. Using yeast as a model system, the aggregation of TDP-43, FUS, and SOD1 has been studied in the most detail. It was shown that the expression of these proteins in yeast reproduces the main features of ALS/FTLD pathology, including aggregation, protein mislocalization, and cellular toxicity ^{[71][72][73][74]}[117,118,119,120].

FUS and TDP-43 have similar domain structures and serve many similar functions. TDP-43, which is 414 amino acid residues in length, has been shown to bind to both DNA and RNA and has multiple functions in transcription repression, pre-mRNA splicing, and translation regulation. FUS, which is 526 amino acid residues in length, plays an important role in various cellular processes, such as the regulation of transcription, RNA splicing, RNA transport, and DNA repair. In neuronal cells, it plays a critical role in dendritic outgrowth formation and stability, RNA transport, mRNA stability, and synaptic homeostasis ^[75][121].

Both of these proteins have highly conserved RNA recognition motifs (RRMs) and glycine-rich domains ^[75][121]. Moreover, using a bioinformatics algorithm originally developed to identify yeast prion domains ^[76][122], “prion-like” domains were identified in the N-terminal part of FUS (amino acids 1–239) and in the C-terminal part of TDP-43 (amino acids 277–414) ^[75][76][121,122]. Similar to the prion domains of yeast prion proteins such as Sup35, Ure2, and Rnq1, these domains are rich in uncharged polar amino acids ^[77][78][123,124].

Expression of human FUS and TDP-43 proteins in yeast leads to cytoplasmic aggregation of these proteins and toxicity, thus modeling key aspects of ALS- and FTLD-related proteopathies. The RRM and prion-like domains have been shown to be key to the aggregation of the proteins studied. The patterns of localization of FUS and TDP-43 protein aggregates in yeast cells in terms of size, shape, and number of foci in the cytoplasm are strikingly similar. Indeed, when FUS-YFP and TDP-43-CFP were coexpressed in the same cell, the fluorescence signals colocalized in the same cytoplasmic foci ^[75][121].

The aggregates formed by FUS and TDP-43 in yeast cells have been shown not to react with amyloid diagnostic dyes and are SDS-soluble. Thus, FUS and TDP-43 in yeast cells form aggregates that are probably not amyloid in nature, as are aggregates of these proteins observed in patients with ALS and FTLD ^[75][79][80][121,125,126].

To date, many mutations in the human FUS and TDP-43 genes have been described and have been associated with some familial and sporadic ALS cases ^[81][127]. Using a yeast model, the influence of these mutations on the aggregation and toxicity of the proteins under study was tested. The mutations in TDP-43 significantly increased the degree of aggregation and toxicity of this protein in yeast cells. Moreover, almost all ALS-linked mutations in the TDP-43 gene are located in the DNA sequence encoding the prion-like domain of this protein. However, ALS-linked FUS mutations do not promote FUS aggregation in yeast ^[75][121].

Superoxide dismutase 1 (Sod1) is an enzyme that, in humans, is encoded by the SOD1 gene located on chromosome 21. Sod1 is 154 amino acid residues long, and alternative splicing results in five forms of the enzyme that differ in their localization in the human body ^[82][128]. This enzyme is a 32 kDa homodimer containing an intramolecular disulfide bond and a binuclear Cu/Zn site in each subunit. SOD1 binds copper and zinc ions and is one of three superoxide dismutases responsible for the destruction of free superoxide radicals in human cells. SOD1 is a soluble cytoplasmic and mitochondrial intermembrane protein that converts naturally occurring but harmful superoxide radicals into molecular oxygen and hydrogen peroxide. Recently, it has been shown that under oxidative stress, SOD1 can localize not only in the cytoplasm but also in the nucleus ^[83][84][129,130]. Mutations in this gene (more than 150 identified to date) are associated with familial ALS1 ^[85][131]. The protein (both wild-type and ALS1 variants) has a propensity to form fibrillar aggregates in the absence of intramolecular disulfide bonding or bound zinc ions. These aggregates can exert cytotoxic effects. Zinc binding promotes dimerization and stabilizes the wild-type form.

Yeast has been shown to have a homolog of the human SOD1 gene. Expression of the human SOD1 gene in yeast cells lacking the homologous gene showed that human SOD1 can fully complement the biological function of yeast SOD1 ^[86][132]. A series of studies have been conducted in which the wild-type human SOD1 gene and SOD1 genes containing various ALS-related mutations were expressed in yeast cells ^[87][88][89][133,134,135]. Interestingly, neither the wild-type human protein nor any of the mutant variants of this protein corresponding to ALS-related mutations in yeast cells caused impairment of the cell culture growth rate ^{[87][88][89][90][91]}[133,134,135,136,137]. However, it has been shown that the wild-type SOD1 protein, as well as ALS-associated SOD1 mutants, tends to selectively aggregate near mitochondria, where the protein may exert toxic effects that are not yet fully understood ^[74][120]. It has been shown that the aggregates formed are SDS-soluble and of a nonamyloid nature ^[86][92][132,138].

Integrated transcriptomic and metabolomic analysis is the most recently developed and promising methodological approach that uses yeast as a model system to understand the biological role of the described proteins in the deregulation of pathways associated with the disease. To date, only preliminary, but encouraging, results have been obtained, and in the ouresearchers' view, further research is warranted ^[93][94][139,140].

Tau, also known as microtubule-associated protein tau (MAPT), is expressed predominantly in cells of the nervous system. The main function of this protein is the stabilization of microtubules, which occurs through a kiss-and-hop mechanism ^[95][96][141,142]. The human MAPT gene contains 16 exons. In cells of the nervous system, alternative splicing yields six tau isoforms, ranging in size from 352 to 441 amino acids. The molecular mass of the tau isoforms ranges from 48 kDa to 68 kDa ^[97][143]. Numerous human NDs have been shown to be associated with tau protein aggregation (see Table 1). Although tau is a “naturally unfolded protein”, conformational changes have been detected and could be essential in the formation of aggregates ^[98][144]. Various posttranslational modifications of the tau protein (phosphorylation, acetylation, ubiquitination, SUMOylation, methylation, glycosylation, glycation, proteolysis, and numerous others) occur in the brain cells of both healthy people and patients with NDs ^[99][145]. Posttranslational modifications play an important role in tau protein aggregation, and phosphorylation is of particular importance.

Misfolding Protein	Origin	Experimental Conditions
Luciferase	Photinus pyralis	Heat shock
Guk 1-7 (guanylate kinase temperature-sensitive) temperature-sensitive	Saccharomyces cerevisiae	Heat shock
Gus 1-3 (glutamyl-tRNA synthetase) temperature-sensitive	Saccharomyces cerevisiae	Heat shock
Pro 3-1 (delta 1-pyrroline-5-carboxylate reductase) temperature-sensitive	Saccharomyces cerevisiae	Heat shock
Ubc9ts (SUMO-conjugating E2 enzyme) temperature-sensitive	Saccharomyces cerevisiae	Heat shock
VHL (von Hippel–Lindau tumor suppressor) temperature-sensitive	Homo sapiens	Constitutively unfolded in yeast cells (absent binding partner)
ΔssCPY* (mutated form of carboxypeptidase Y)	Saccharomyces cerevisiae	Constitutively unfolded
tGnd1 (truncated phosphogluconate dehydrogenase)	Saccharomyces cerevisiae	Constitutively unfolded
DegAB (the entire Ndc10 degron)	Saccharomyces cerevisiae	Constitutively unfolded

Expression of tau in yeast does not result in a negative phenotype ^[100][146]. Tau protein expressed in yeast under normal conditions, either under mild proteotoxic stress or in mutants with impaired proteotoxic stress response pathways, did not result in toxicity or the formation of obvious aggregates. In chronologically aged cells, no appreciable Tau aggregates were formed ^[101][147].

In S. cerevisiae, tau phosphorylation and hyperphosphorylation are under the control of two protein kinases, Mds1 and Pho85, which are orthologs of two human protein kinases, GSK3β and cdk5, respectively. It has been shown that when tau is expressed in yeasts lacking Pho85, the tau phosphorylation and aggregation levels increase, indicating that Pho85 (and possibly cdk5 in human neuronal cells) is a negative regulator of tau phosphorylation. On the other hand, deletion of the Mds1 gene leads to a decrease in the amount of hyperphosphorylated tau protein and, accordingly, to a decrease in the amount of the aggregated form of this protein. Thus, the tau protein expressed in yeast lacking Pho85 undergoes posttranslational modification by phosphorylation and, as a consequence, changes its conformation and forms aggregates. The hyperphosphorylation and aggregation of the tau protein in yeast cells is similar to processes that are characteristic of neurodegenerative tauopathies, including Alzheimer’s disease ^[102][148].

The yeast model systems described above, which allow the study of the aggregation of proteins whose aggregation was originally identified in studies of human NDs, can be used to screen potential drugs that prevent the aggregation and toxicity of these proteins. Indeed, yeast has been shown to be a very useful platform for drug screening to identify potential therapeutic molecules. Importantly, many of these results initially obtained in yeast have subsequently been confirmed in mammalian systems, confirming yeast as reliable. For review, see, e.g., ^{[103][104][105][106]}[28,33,149,150].

Although yeast cells are well-suited for high-throughput screening in search of chemical compounds that inhibit aggregation and/or toxicity, one limitation is the inability of many drugs to cross cell membrane and/or cell wall barriers or to accumulate within the cell. This problem can be overcome by using yeast strains containing mutations in the ERG6 or PDR5 genes. Erg6 mutants have increased membrane permeability; pdr5 mutants more easily accumulate drugs intracellularly. For review, see, e.g., ^[104][33].