2. The Molecular Basis of FXPAC
The PM alleles are characterized by increased levels of
FMR1 mRNA, which correlate with the length of the repeat tract, in both male and female carriers of a PM allele
[6][76][77]. Although the elevated mRNA levels result from an increase in transcriptional gene activity
[78], a CGG-repeat-length-dependent decrease in the expression of the
FMR1 protein, FMRP, likely results from the impaired scanning of ribosomal preinitiation complexes through CGG-repeat tracts
[6][76][79][80]. The increased expression of the
FMR1 mRNA (up to six-to-eight-fold of that seen in normal alleles) leads to transcriptionally activated cellular stress pathways, RNA-mediated toxicity triggering CGG-binding-protein sequestration, and repeat-associated non-AUG-initiated (RAN) translation, which are the current basic and central molecular mechanisms proposed to explain the pathogenesis of FXTAS.
2.1. Molecular Basis of the FMR1 Locus
The PM alleles in females are unstable and prone to expansion on intergenerational transmission, with expansion into alleles harboring greater than 200 CGG repeats, leading to FXS. Generally, one or two AGG interruptions are observed within the repeat tract of normal and intermediate
FMR1 alleles (6–44 CGG and 45–54 CGG repeats, respectively), while one or none are observed in PM alleles, and they are known to influence the stability of the repeats during parental transmission. Specifically, the presence of AGG interruptions decrease the intergenerational instability of the CGG repeats, thus decreasing the risk of expansion to a full-mutation allele
[81][82]. In addition to AGG interruptions, other factors that increase the risk of expansion to full-mutation alleles during maternal transmission include the maternal CGG repeat number and age
[81][82]. Interestingly, no association was found to correlate with either the transcriptional or translational activity of the gene
[78][80][83][84].
As observed in other trinucleotide disorders, a bidirectional transcription at the
FMR1 locus has been demonstrated and specific alternative splicings of the antisense
FMR1 (
ASFMR1) gene have been identified
[85]. The
ASFMR1 gene is expressed in all tissues, with high expression observed in the brain, spans approximately 59 kb of genomic DNA, and contains 13 exons and 45
ASFMR1 isoforms that are identified, 19 of which are expressed only in the PM
[86]. Some of these isoforms are, as for the
FMR1 gene, highly expressed in the PM as compared to the controls
[87]. Although the
ASFMR1 has been suggested to play a critical role in the pathogenesis of FXTAS
[88][89], further studies are warranted to shed light on the contribution of the
ASFMR1 in the clinical phenotypes of FXTAS.
Recently, it has been demonstrated that alleles in the PM range can be somatically unstable in both male and female carriers of a PM allele
[86][90]. As observed with intergenerational instability, it was demonstrated that the extent of somatic instability directly correlates with the number of CGG repeats and, inversely, with the number of AGG interruptions. Increased levels of somatic expansion are observed over time in blood (PBMCs) derived from female carriers of a PM allele
[86], and are mainly due to unmethylated
FMR1 alleles and, therefore, limited to the active X chromosome. Recent evidence suggests that DNA repair factors
FAN1 and
MSH3 are both also modifiers of the expansion risk in females with specific genotypes associated with increased somatic instability
[86] (these genes have also been implicated in other repeat expansion disorders (Genetic Modifiers of Huntington’s Disease Consortium)), suggesting a common expansion mechanism. Genetic factors that affect the somatic expansion risk may contribute to the variable penetrance for FXPAC that is seen. The extent of somatic instability in female PM carriers has shown a significant correlation with a diagnosis of ADHD
[91], and may also affect the risk of various PM conditions in both males and females.
Allelic instability, observed in individuals with
FMR1 mutations, leads to both intra- and intertissue mosaicism (PBMCs, fibroblasts, and brain tissues), and may account for some of the variability observed in the clinical phenotype of individual carriers of the PM
[90]. During the International Premutation Conference, new data were presented about allelic instability within the
FMR1 gene, confirming its occurrence between and within different tissues derived from the same individuals. Unstable alleles were exhibited among the majority of both female and male PM carriers. In addition, diverse allele profiles were displayed between PBMCs and fibroblasts from the same individuals among PM males, in accordance with previous studies
[90][92][93][94][95]. Allelic instability affirms the complexity of
FMR1 mutations and may relate to diverse phenotypes, including cognitive abilities and behavioral features observed in both FXS and PM disorders
[96], specifically in female carriers of a PM allele with ADHD and depression
[91]. The activation ratio (AR) is a clinically relevant parameter for females with both full-mutation
[97] and PM conditions
[86], as it reflects the fraction of normal alleles present on the active X chromosome
[98]. The X-inactivation process is widely recognized as a factor that can influence the symptoms and severity of many diseases
[99]. In FXS, although the size of the CGG repeat in the promoter region of the
FMR1 gene is a significant factor, it is not sufficient to entirely determine the functionality of the gene. Hence, factors such as the AR and methylation status of the gene in females carrying an
FMR1 mutation may also contribute to the regulation of FMRP levels. Therefore, to accurately interpret phenotypic characteristics in individuals with both FXS and FXPAC, it is necessary to assess methylation-status analyses
[100][101][102][103].
The extent of phenotypic variation based on the AR is demonstrated by the observation that approximately 30% to 50% of females carrying a full mutation and exhibiting normal intelligence have the mutation primarily on their inactive X chromosome
[103]. Moreover, studies have indicated that female PM carriers with a higher AR exhibit a significantly lower FXTAS incidence
[35][104][105]. On the other hand, individuals with a normal allele that is predominantly methylated, and therefore inactive, may be at a higher risk of developing FXTAS. Additionally, several studies have suggested that lower AR values could be linked to cognitive and behavioral challenges in female PM carriers
[97][106][107][108][109][110], potentially affecting the risk, severity, and age of onset of FXPAC. Despite numerous studies investigating the role and impact of the AR in PM carriers, there are discrepancies among their findings that may be partially attributable to technical variability, as previously reported
[86][111][112][113][114], or to differences in the methods employed to calculate the AR (as discussed in Protic et al., this special issue
[115]). At the International Premutation Conference, novel data were presented demonstrating a noteworthy correlation between clinical measures and the AR. As anticipated, the study revealed that higher ARs were linked to reduced
FMR1 transcript levels for any given repeat length, and associated with enhanced performance, verbal, and full-scale IQ scores, as well as lower levels of depression, and a smaller number of medical conditions. Based on this evidence, it is advisable to evaluate the methylation status, including the AR in females with both PM and full-mutation alleles of the
FMR1 gene, to better understand their clinical phenotypes.
2.2. Molecular Mechanisms Leading to FXTAS Pathology—RNA Toxicity and RAN Translation at CGG Repeats: Mechanistic Insights and Their Contribution to Disease Pathology
There are currently three nonexclusive models for how CGG repeats elicit pathogenesis in FXTAS (Figure 1).
Figure 1. Molecular mechanisms leading to
FMR1-PM-associated conditions. Three nonexclusive models are proposed for how CGG repeats contribute to the pathogenesis of PM conditions, including FXTAS. First, the cotranscriptional R-loop formation, which compromises genomic stability and triggers a DNA-damage response that can activate inflammatory cascades
[116][117]. Second, CGG-repeat RNAs can elicit a gain-of-function toxicity through RNA gelation into nuclear foci and sequestration of various rCGG-repeat-binding proteins, leading to their functional depletion
[25][26]. Third, repeat-associated non-AUG (RAN)-initiated translation generates potentially toxic proteins that accumulate within intranuclear neuronal inclusions in FXTAS patients. The relative contribution from each mechanism to downstream sequelae, such as mitochondrial dysfunction and neuronal death, and their potential synergies in disease pathogenesis, are areas of ongoing research in the field. Adapted from
[118].
In one, CGG-repeat RNAs elicit a gain-of-function toxicity through both RNA gelation into nuclear foci and sequestration of various rCGG-repeat-binding proteins
[25][26]. Mass-spectrometric and immunohistochemical analyses have identified over 20 proteins in the frontal cortex inclusions of FXTAS patients, including RNA-binding proteins (RBPs), HNRNP A2/B1 (heterogeneous nuclear ribonucleoprotein A1), and MBNL1 (muscleblind-like protein 1), as well as some neurofilament proteins, such as lamin A/C and α-internexin. These proteins are involved in various neurological disorders
[119]. Pur α and HNRNP A2/B1 bind directly to rCGG repeats in inclusions, and their overexpression in a
Drosophila model expressing PM CGG repeat expansions suppressed neurodegeneration phenotypes
[25][26]. Sequestration of other proteins, such as CUGBP1 (CUGBP Elav-like family member 1), SAM68 (Src-associated substrate during mitosis of 68-kDa), Rm62 (ATP-dependent RNA helicase p62), and DGCR8 (DiGeorge syndrome critical region 8), leads to altered mRNA splicing and transport, as well as dysregulated microRNAs, supporting a toxic RNA gain-of-function mechanism mediated by the expanded CGG repeats in
FMR1 [24][26][120][121][122].
HNRNPA2/B1 is present in intranuclear inclusions of FXTAS patients and it binds directly to rCGG repeats. Its overexpression, along with its two homologs in
Drosophila, suppresses the neurodegenerative eye phenotype caused by the rCGG repeat
[26]. HNRNP A2/B1 also mediates the indirect interaction between CGG repeats and CUGBP1 involved in myotonic dystrophy type 1 (DM1). Overexpression of CUGBP1 suppresses the FXTAS phenotype in
Drosophila. Pur α, another protein found in intranuclear inclusions of FXTAS patients, plays a crucial role in DNA replication, neuronal mRNA transport, and translation. Pur α knockout mice show developmental delays and altered expression, and the distribution of axonal and dendritic proteins
[123][124]. Overexpression of Pur α in a
Drosophila model suppresses rCGG-mediated neurodegeneration in a dose-dependent manner. Sequestration of SAM68 in particular causes pre-mRNA alternative splicing misregulation in CGG-transfected cells and FXTAS patients, thus contributing to FXTAS pathogenesis via a splicing alteration mechanism
[120]. TDP-43 (TAR DNA-binding protein 43), an ALS-associated RBP, has reduced association with ribosomes in the cerebellar Purkinje neurons of mice expressing 90 CGG repeats
[125]. In the same study, the authors went on to find that, in the
Drosophila model of FXTAS, wild-type TDP-43 expression leads to suppression of neurodegeneration, while the knockdown of the endogenous TDP-43 fly ortholog, TBPH, enhanced the eye phenotype.
Another study also independently reported that TDP-43 suppressed CGG-repeat-induced toxicity in a
Drosophila model of FXTAS
[126]. Interestingly, this suppression was shown to depend on HNRNP A2/B1, such that the deletion of the C-terminal domain of TDP-43, and thereby the prevention of interactions with HNRNP A2/B1, led to the abrogation of the TDP-43-dependent rescue of CGG repeat toxicity
[126]. Finally, DGCR8, a protein binding to PM rCGG repeats, causes partial sequestration of DGCR8 and its partner, DROSHA (drosha, ribonuclease 3), within PM RNA aggregates. DGCR8 and DROSHA play a critical role in microRNA biogenesis. Sellier and colleagues found that the sequestration of DGCR8 and DROSHA precludes them from their normal functions, leading to reduced processing of pri-miRNAs in cells expressing expanded CGG repeats. Consequently, levels of mature miRNAs are also reduced in the brains of FXTAS patients
[24].
Alternatively, the CGG repeats in 5′ UTR of
FMR1 mRNA may be translated into toxic proteins through a process known as RAN translation. Initially described at CAG repeats in spinocerebellar ataxia type 8 (SCA8) and DM1
[127], noncanonical translation of short tandem repeats into proteins may occur in the absence of an AUG initiation codon when repeat-containing RNAs form stable secondary structures. RAN translation has been observed on repeats associated with ten disorders: SCA8, DM1, DM2, HD, FXTAS, C9orf72 amyotrophic lateral sclerosis and frontotemporal dementia (C9 ALS/FTD), FXPOI, SCA31, and Fuchs endothelial corneal dystrophy (FECD) (reviewed in
[118][128]). In many of these diseases, RAN translation occurs in different reading frames on both sense and antisense transcripts, and the RAN products are detected in patient tissues.
In FXTAS, it is thought that CGG repeats form secondary structures that lead to the impairment of ribosomal scanning, reduced start codon fidelity, and, in consequence, aberrant translation initiation at near-cognate or noncognate codons located upstream or within the repeats
[129]. Depending on the reading frame, different toxic proteins containing long mono-amino-acid tracts are produced: polyglycine (FMRpolyG), polyalanine (FMRpolyA), and polyarginine (FMRpolyR)
[28][129]. Additionally, there is evidence that RAN translation also can occur on the CCG antisense transcript
[130] to produce additional homopolymeric proteins. Translation through the repeat may also trigger frameshifting to produce chimeric RAN proteins
[131]. The translation of FMRpolyG is the most efficient, and this protein is detected in FXTAS patient brains by both immunohistochemistry and mass spectrometry, colocalizing with p62 and ubiquitin-positive inclusions
[11][18][28][119][130][132][133][134][135]. However, quantitation of this and other RAN-translation-generated proteins remains challenging due to their low abundance, solubility, multiple initiation sites, and early translation termination—all of which hamper its detection by antibodies targeting either the N- or C-terminus
[21][134][135]. FMRpolyG was found to interact with the nuclear lamina protein LAP2β, leading to the impairment of the nuclear lamina architecture
[132]. Additionally, FMRpolyG is capable of cell-to-cell propagation via exosomes in cell-culture studies and glia-to-neuronal propagation in mouse-model systems. Similar prion-like propagation is thought to play a central role in the pathogenic spread of alpha synuclein in Parkinson’s disease and of Tau in Alzheimer’s disease (PMID: 30917002). However, the role of this phenomenon in FXTAS pathogenesis remains unclear
[134][136].
Whether RAN products generated from CGG repeats are drivers of toxicity or if there is instead a synergy between CGG-repeat RNA and RAN proteins remains unknown. Studies in overexpression systems in cells, flies, and mice suggest that near-cognate codons 5′ to the repeat that support the RAN translation of FMRpolyG are requisite to elicit maximal toxicity
[28][132][137][138]. However, FMRpolyG inclusions can persist even as phenotypes resolve when the repeat is transcriptionally silenced
[139]. Moreover, FMRpolyG production absent the repeat RNA is less toxic in neurons than is a RAN-competent CGG repeat
[131].
The exact mechanism by which RAN translation occurs remains enigmatic, and may vary in different repeats (and even different reading frames of the same repeat). However, several recent studies reported modifiers of RAN translation that provide some clues. Unwinding the structured RNA is crucial for RAN translation, as it is shown that several RNA helicases, such as DDX3X (ATP-dependent RNA helicase DDX3X), DHX36 (ATP-dependent DNA/RNA helicase DHX36), eIF4A/B (eukaryotic initiation factor 4A-I/B), and H, are directly involved in the regulation of this process enabling proper ribosomal scanning
[129][137][140]. In addition, the presence of RAN proteins, together with structured RNAs with CGG repeats, leads to the activation of the integrated stress response (ISR) and the phosphorylation of eIF2α, which, in a feed-forward-loop mechanism, shut down the global translation but selectively enhance RAN translation
[141]. Proteins which interact with CGG-repeat RNAs may also influence the RAN translation, as SRSF1 (serine/arginine-rich splicing factor 1) mediates the nuclear retention of CGG-repeat RNAs to prevent these transcripts from becoming a template for RAN translation
[118].
2.3. Therapeutic Perspectives to FXTAS from a RAN-Translation Perspective
There are currently no FDA (Food and Drug Administration Agency)-approved drugs to slow FXTAS progression or delay its onset. An emphasis point that was raised during the International Premutation Conference was that there is a critical need for the discovery of reliable, robust biomarkers to accurately understand predisease onset states and readouts for clinical progression. Some promising work suggests that metabolomic and/or proteomics biomarkers may serve this purpose
[142][143][144]. Indeed, a small open-label pilot study in patients with validation studies in patient fibroblasts indicated that the mitochondrial activator sulforaphane showed some correction of these biomarkers that could serve as a precursor for a larger study
[143].
Antisense oligonucleotides (ASOs) hold promise for FXTAS treatment by blocking RAN translation in neurons without degrading the
FMR1 mRNA, enhancing FMRP expression, and improving neuron survival
[138]. In rodent models, ASOs reduce FMRpolyG biosynthesis, correct disease-related traits, and normalize transcriptomic effects
[145]. A recent study suggests that the ubiquitin–proteasome system may be an interesting therapeutic target based on the presence of PSMB5 (proteasome subunit beta type-5) polymorphisms as disease-onset modifiers in patients and the suppression of disease-relevant phenotypes in
Drosophila with a genetic knockdown of this proteasomal subunit
[146]. This factor also changes how CGG RAN translation happens in cell tests. For example, it affects how a molecule called CMBL4c can attach to CGG-repeat RNA structures and lower FMRpolyG levels
[147]. The ISR targeted by protein kinase R (PKR) rescue in a mouse model of FXTAS
[143], along with multiple CGG-repeat-associated RBPs, offer potential treatments targeting RAN-translation modifiers, although clinical application awaits further research due to the early stage of
Drosophila, cell-based, and mouse experiments.
2.4. Genetic Modifiers in Fragile X-Associated Tremor Ataxia Syndrome (FXTAS)
The underlying neurobiological mechanisms of FXTAS are complex and not fully understood. As mentioned above, several mechanisms that have been proposed to explain the pathogenesis of FXTAS, including RNA toxicity, RAN translation producing the accumulation of the FMR PolyG polypeptide, and damage response, are linked to white-matter-tract connectivity in the brain, called white-matter hyperintensities, and strongly associated to the clinical impairment observed in FXTAS
[6][28][148]. However, not all individuals who carry a PM allele will develop PM conditions, including FXTAS, in their older adulthood, which indicates the incomplete penetrance pattern of the disease. Therefore, nowadays, some studies have been dedicated to a plausible mechanism and exploring predisposing factors, including genetic modifiers, that may contribute to the occurrence of FXPAC. Investigations of genetic modifiers of clinical manifestation of diseases have also become a new research interest in FXTAS. They sought to provide an answer to the wide diversity and severity of clinical major criteria (intention tremor and gait ataxia) and minor criteria (cognitive impairment)
[33][146][149]. Various genetic variants may contribute to cognitive impairment, including the APOe4 allelic variant, which represents the strongest risk factor of late-onset Alzheimer’s disease (AD), the most common type of dementia, in all ethnic groups
[150]. The prevalence of the APOe4 allele is 13.7% in the general population; having one copy of the APOe4 allele increases the risk by around 3 times compared to individuals without the APOe4 allele, while having two copies boosts the risk of AD by 8–12 times
[151].
APOE is an important cholesterol and lipid transporter that plays a critical role in a variety of signaling pathways in the development, maintenance, and repair mechanisms of the central nervous system (CNS)
[152]. The APOe4 allele triggers β-amyloid (Aβ) accumulation/amyloidosis in oligodendrocytes and their myelin that leads to the slowing of brain electrical signaling, which is associated with cognitive impairment
[153]. Postmortem examination of FXTAS brain tissue showed the presence of cortical amyloid plaques and neurofibrillary tangles, combined with presence of intranuclear inclusions in those with FXTAS and AD, which is additional evidence of the involvement of other genes that may modify the FXTAS phenotype
[154]. Among the
FMR1 PM carriers, the APOe4 allele frequency is higher (31.8%) in patients with FXTAS compared to the general population and increases the risk by more than 12 times to develop the disease
[155]. During the International Premutation Conference, data on 180 PM males, aged over 50 years, were presented which showed that the APOe4/APOe2 and APOe4/APOe3 genotypes were more frequent in PM males with FXTAS compared to those without FXTAS (2% vs. 0% and 10.6% vs. 2.4%, respectively).
Recently, to identify the genetic modifiers of FXTAS, a large number of PM carriers were recruited for whole-genome sequencing (WGS), which was further combined with
Drosophila genetic screening. It was demonstrated that using FXTAS
Drosophila as a genetic screening tool can be powerful in the validation of candidate genes from WGS. Eighteen genes were identified as potential genetic modifiers of FXTAS. One of such candidate genes is the proteasome subunit beta-5 (PSMB5), which genetically modulates CGG-associated neurotoxicity in
Drosophila as a strong suppressor of CGG-associated neurodegeneration. PM individuals who carry the variant PSMB5rs11543947-A, which is associated with decreased expression of PSMB5 mRNA, may be protected against FXTAS. In addition, there is a strong suppression of CGG-associated neurodegeneration through diminishing RAN translation in the
Drosophila knockdown of PSMB5
[146]. The metabolomic approach to determine a genetic modifier in an FXTAS mouse model found metabolic changes and demonstrated that Schlank (ceramide synthase), Sk2 (sphingosine kinase), and Ras (IMP dehydrogenase), which encode enzymes in the sphingolipid and purine metabolism, respectively, were significantly related with FXTAS-CGG-associated neurodegeneration pathogenesis
[149].
Finally, more studies are needed to identify possible genetic modifiers associated with FXTAS development and progression for better management of the disease and for the development of therapeutic strategies.
2.5. The Use of Human Pluripotent-Stem-Cell-Based Neurodevelopmental Models for FXTAS
Human models of FXPAC are essential tools for studying disease-specific mechanisms, such as RNA toxicity, RAN translation, and CGG somatic instability. However, generating improved model systems for all these pathologies requires patients’ disease-relevant cell cultures. In the case of FXTAS, this is especially challenging because postmortem brain samples are rarely available, limited to a small amount of biological material, and represent only the final stage of the disease.
Overcoming these limitations can be achieved by utilizing mutant human pluripotent stem cells (hPSCs) in conjunction with in vitro differentiation towards affected tissues (neurons). This approach provides a powerful tool for both fundamental and applied research, offering an excellent opportunity to investigate the disease’s pathogenic mechanisms and to identify potential targets for therapeutic intervention.
There are two types of pluripotent stem cell lines that can be utilized for FXTAS disease modeling: human embryonic stem cell (hESC) lines derived from genetically affected embryos that can be obtained by preimplantation genetic diagnosis (PGD) procedures
[156], and patient-derived induced pluripotent stem cells (iPSCs), established by reprogramming somatic cells obtained from patients (e.g., blood, skin fibroblasts)
[157]. Both PGD-derived hESCs and patient-derived iPSCs carry the disease-causing PM and can reproduce disease cellular phenotypes in vitro, and allow following dynamic processes that are misregulated during development and aging in patients.
The first in vitro model of FXTAS using pluripotent stem cells (PSCs) showed that differentiated neurons from iPSCs recapitulate the cellular phenotypes of FXTAS, including reduced synaptic puncta density, neurite length, and increased calcium transients
[157]. FXTAS iPSCs were also used to discover a toxic mechanism linked to FMRPolyG proteins via RAN translation
[132]. Additionally, human neurons derived from patient iPSCs were used to validate a therapeutic approach that selectively blocked CGG RAN initiation sites using noncleaving ASOs. ASO blockade improved endogenous FMRP expression, suppressed repeat toxicity, and prolonged survival in human neurons, showing the therapeutic potential of modulating RAN translation in FXTAS
[138].
Nevertheless, despite recent progress, the currently available human iPSC-based models for FXTAS are insufficient in reproducing the full complexity of the disease. This is because these models are based on monolayer cell cultures, which restrict the analysis to less-mature and single-cell types. To gain a comprehensive understanding of the interactions between various cell populations in the brain, and to examine the contribution of each pathogenic mechanism associated with FXTAS during early brain development, a higher level of complexity than monolayer cell cultures, such as brain organoids, would be necessary.
Brain organoids are three-dimensional mini-organs derived from PSCs that mimic the cellular composition and architecture of specific brain regions
[158]. As such, they are expected to provide a powerful tool for identifying critical molecular events in the development of FXTAS, much before the clinical signs appear in patients. Moreover, brain organoids could extend our knowledge on other aspects of the disease, such as CGG somatic instability and the generation of mosaicisms for expansion size and/or methylation, in a multicellular setting that more closely resembles the developing human brain.
2.6. Shared Molecular Mechanism with Other Repeat Expansion Disorders
FXTAS is a repeat expansion disorder that displays clinical symptoms similar to those observed in other diseases caused by repeat expansions. Parkinsonism, a varied array of cognitive impairments that can progress to dementia, and amyotrophic lateral sclerosis (ALS)-like phenotypes, including frontotemporal dementia and progressive supranuclear palsy, have all been reported in FXTAS
[159]. Tremor and ataxia, which are also hallmark symptoms of other repeat expansion disorders, such as spinocerebellar ataxias, are commonly observed in FXTAS.
The genetic basis of the FXTAS repeat expansion is similar to other repeat expansions observed in several diseases, including C9orf72 ALS/frontotemporal dementia (GGGGCC-repeat), myotonic dystrophy type 1 (CTG-repeat), NOTCH2NLC (CGG-repeat), Huntington’s disease (CAG-repeat), and spinocerebellar ataxias (SCA-CAG-repeat). Regional aggregation of cytosolic, nuclear, or extracellular proteins is a common feature observed in these diseases and disrupts neuronal function
[160]. Intranuclear eosinophilic ubiquitin-positive inclusions in neurons and astrocytes are characteristic of FXTAS pathology and have been observed in other trinucleotide disorders
[161]. TDP-43 in ALS/frontotemporal dementia and poly (amino acid)/polypeptides in FXTAS, Huntington’s disease, and spinocerebellar ataxias are examples of the types of aggregates that result from the expansion of trinucleotide repeats
[162].
The most common genetic cause of ALS/frontotemporal dementia is an expanded GGGGCC-repeat in the C9orf72 gene. Similar to FXTAS, RAN translation and the accumulation of toxic peptides in neurons and astrocytes (TDP-43) are the main pathological mechanisms in C9orf72 ALS/frontotemporal dementia
[163]. The accumulation of toxic polypeptides resulting from expanded trinucleotide repeats is also observed in Huntington’s disease (CAG-repeat) and spinocerebellar ataxias (CAG-repeat)
[164].
The NOTCH2NLC pathogenic CGG-expansions, located in the 5′ UTR (66-517) and having GGA or AGC interruptions, are particularly similar to those observed in FXTAS. They cause a late-onset disorder with a clinical variability that includes muscle weakness, dementia, parkinsonism, tremor, and ataxia. The molecular mechanisms of OTCH2NLC lead to neuronal intranuclear eosinophilic inclusions, and the antisense isoform has been hypothesized to be a pathological mechanism
[165].
Anticipation, somatic instability, and clinical severity associated with the number of repeats has been described in many repeat expansion disorders, including, HD, DM1, FXTAS, ALS, and others
[166].
Aside from these, FXTAS resembles DM1 in many respects. Firstly, because the primary mechanism for both pathologies is RNA toxicity
[25][26][132][167][168][169][170]. Secondly, and as mentioned above, both affected loci exhibit RAN translation potential, leading to the production of toxic polyglycine, polyalanine and polyarginine containing proteins by CGG expansion in the PM range in
FMR1 [28][132][171], and polyalanine- and polyserine-containing proteins by CTG expansions in DM1-affected cells
[127][172]. To add further complexity, both disorders exhibit a decrease in protein levels, albeit through distinct mechanisms
[6][76][173]. Lastly, both expansions in
FMR1 and DM1 display maternal anticipation/expansion, giving rise to distinct phenotypes (namely, FXS in
FMR1 and congenital myotonic dystrophy type 1 in DMPK) and to DNA hypermethylation. Altogether, the clinical presentation of individuals carrying the
FMR1 PM is highly heterogeneous and shares similarities with the phenotypic heterogeneity observed in DM1 and other nucleotide repeat disorders. This variability likely results from the involvement of the multiple mechanisms that, together with modifier genes and environmental factors, contribute to disease pathology to varying degrees.
2.7. Mitochondrial Dysfunction in PM Carriers
Recently, studies on cultured cell lines, animal models, and human subjects have implicated mitochondrial dysregulation in the pathogenesis and progression of FXTAS. Using magnetic resonance imaging (MRI), Rizzo et al. (2006)
[174] first described lactate accumulation in the lateral ventricles, as well as decreased ATP levels in the calf muscles of a patient with FXTAS. Subsequent studies on cultured fibroblasts from PM carriers and mouse models have confirmed impaired ATP production and the pathogenic role of expanded CGG repeats on mitochondrial functions
[175][176]. Finally, clinical studies on living patients with FXTAS and postmortem brain tissues with the disease have showed altered Krebs-cycle intermediates, neurotransmitters, and neurodegeneration markers, as well as reduced mitochondrial DNA copy numbers in specific brain regions, such as the cerebellar vermis, parietal cortex, and hippocampus
[32][177]. Finally, unlike the earlier results from human brain tissue, studies in Epstein–Barr-virus (EBV)-transformed blood lymphoblasts showed that mitochondrial respiratory activity was significantly elevated in FXTAS compared with controls. Specifically, altered complex I activity and ATP synthesis, accompanied by an altered mitochondrial mass and membrane potential, were observed, and were significantly associated with the white-matter-hyperintensity (WMH) scores in the supratentorial regions
[178]. In addition, an elevation of AMP combined with the reduction of TORC in both FXTAS and non-FXTAS categories of PM carriers was reported
[179]. In the later study, correlations between measures of mitochondrial and nonmitochondrial respiratory activity, AMPK, and TORC1 cellular protein kinases, and the scores representing motor, cognitive, and neuropsychiatric impairments, were found with the CGG repeat size, and a hyperactivity of cellular bioenergetic components was significantly associated with motor-impairment measures, including tremor–ataxia, parkinsonism, and neuropsychiatric changes, predominantly in the FXTAS subgroup
[180]. Moreover, an elevation of AMPK activity and a decrease in TORC levels were significantly related to the size of the CGG expansion. All the above studies have suggested that the bioenergetic changes in blood lymphoblasts are biomarkers of the clinical status of
FMR1 carriers. Furthermore, a decreased level of TORC1—the mechanistic target of the rapamycin complex—suggested a possible future approach to therapy in FXTAS.
Several molecular mechanisms have been proposed as mediators of abnormal mitochondrial function in FXTAS. RNA toxicity was the first model described, according to which the expanded CGG repeats in
FMR1 mRNA binds and titrates specific RNA-binding proteins, resulting in loss of their normal functions
[26]. Among these proteins, the pre-mRNA splicing factor TRA2A has gained significant attention, since it is also present in the pathognomonic ubiquitin inclusions of FXTAS
[181]. Additionally, miRNAs are increasingly recognized as major determinants of normal mitochondrial function. One of their biogenesis regulators, the DROSHA/DGCR8 enzymatic complex, was found sequestered within the expanded CGG RNA foci, leading ultimately to the loss of its normal function
[24][182][183]. Moreover, altered zinc and iron metabolism, a pivotal neuromodulator, and an essential element in maintaining mitochondrial physiology, respectively, may be additional contributing factors in FXTAS pathogenesis. Fibroblasts from PM carriers have been shown to express abnormal zinc transporter levels, thereby leading to altered zinc homeostasis
[30], whereas increased iron levels were also observed in neurons and oligodendrocytes of the putamen of carriers of a PM
[19]. Finally, among the functions of FMRP, the product of the
FMR1 gene, is the binding to superoxide dismutase 1 (SOD) mRNA and the regulation of its levels. Consequently, lower expression of FMRP may result in decreased levels of SOD1, thereby leading to increased reactive oxygen species (ROS) levels and impaired oxidative phosphorylation
[184].
More recently, emerging evidence has implicated the role of abnormal electron-transport-chain enzyme complexes in FXTAS pathogenesis. Gohel et al. had first observed defective complex activity in human cell lines and a transgenic mouse model
[185]. Additionally, a recent study, presented at the International Premutation Conference, utilizing brain-derived extracellular vesicles, a novel and powerful platform for biomarker development for brain diseases, from plasma and from postmortem brain tissues from patients with FXTAS, found a decreased quantity and activity of complex IV and V, thus further validating this pathogenic process
[143].
2.8. Omics Studies (Metabolomics and Proteomics) in PM Carriers
The development of targeted therapeutics for rare age-dependent neurodegenerative disorders encounters numerous challenges, encompassing the absence of biomarkers for early diagnosis and disease progression, intricate underlying molecular mechanisms, heterogeneous phenotypes, limited historical data, and the difficulties posed by conducting clinical trials with small patient populations, which restrict enrollment. In this context, contemporary omics studies, including metabolomics and proteomics, have emerged as promising tools for investigating global changes within a given sample, employing extensive data mining and bioinformatic analysis
[186]. Recent advancements in metabolomic- and proteomic-profiling technologies and processing have enabled the efficient and precise analysis of several hundred metabolites/proteins, facilitating the identification of biomarkers associated with disease development and progression
[187].
Giulivi et al. (2016) conducted a comprehensive analysis of the plasma metabolic profile in human PM carriers with FXTAS, comparing them to healthy noncarrier controls. Their findings identified a panel of four core serum metabolites (phenethylamine, oleamide, aconitate, and isocitrate) that exhibited high sensitivity and specificity in diagnosing PM carriers with and without FXTAS. Notably, the presence of oleamide/isocitrate was identified as a specific biomarker for FXTAS. Moreover, based on these plasma metabolic profiles, the researchers reported evidence of mitochondrial dysfunction, neurodegeneration markers, and proinflammatory damage in FXTAS PM carriers
[32]. In a separate investigation, Song et al. (2016) reported increased mitochondrial oxidative stress in primary fibroblasts obtained from PM carriers compared to age- and sex-matched controls
[46]. Napoli et al. (2016) examined peripheral blood mononuclear cells (PBMCs) derived from controls and carriers of a PM allele, with and without FXTAS, to investigate the presence of the Warburg effect. Their study revealed alterations in glycolysis and oxidative phosphorylation, indicating the involvement of the Warburg effect in FXTAS
[188]. Using a PM murine model, Kong et al. (2019) investigated metabolic changes associated with FXTAS in the cerebellum. Their findings demonstrated significant alterations in sphingolipid and purine metabolism in the cerebellum of the mice. Furthermore, they identified genetic modifiers (Cers5, Sphk1, and Impdh1) of CGG toxicity in
Drosophila [149]. In a 12-week open-label intervention study involving six males with FXTAS, Napoli et al. (2019) evaluated the effect of allopregnanolone on lymphocytic bioenergetics and plasma pharmacometabolomics. They observed the significant impact of allopregnanolone treatment on oxidative stress, the GABA metabolism, and certain mitochondria-related outcomes. These findings suggested the potential therapeutic use of allopregnanolone for improving cognitive function and the GABA metabolism in patients with FXTAS
[189]. A more recent study by Zafarullah et al. (2020) aimed to identify metabolic biomarkers for early diagnosis and disease progression in FXTAS. Through characterization of individuals who developed FXTAS symptoms over time, alterations in the lipid metabolism, particularly in mitochondrial-bioenergetics-related pathways, were identified as significant contributors to FXTAS
[88]. Subsequently, Zafarullah et al. (2021) established a significant correlation between the identified metabolic biomarkers and the area of the pons in individuals who developed FXTAS over time. They also demonstrated a notable association between these biomarkers and disease progression, highlighting their role within the context of the dysregulated lipid and sphingolipid metabolism
[142].
In addition, the effort to identify the metabolic changes associated with FXPOI is ongoing, and preliminary data of a nontargeted metabolomic profiling of FXPOI patient plasma by LC/MS were presented during the International Premutation Conference. Initial differential abundance analyses revealed the altered abundance of compounds in the omega-6 fatty-acid (n-6 FA) metabolism and arachidonic acid formation between females with a FXPOI diagnosis and female carriers of a PM without POI across both cohorts. Pathways downstream of FA and the arachidonate metabolism were also identified, including prostaglandin synthesis and the formation of proinflammatory metabolites from the AA. Further investigation of metabolic changes associated with FXPOI is likely to provide critical information about the mechanism of dysfunction in PM ovaries.
In recent years, Ma et al. (2019) conducted an LC-MS/MS-based proteomics analysis of intranuclear inclusions isolated from the postmortem brain tissue of individuals with FXTAS. Their findings revealed the presence of over 200 proteins within the inclusions, with significant abundance of SUMO2 and p62/sequestosome-1 (p62/SQSTM1). These results support a model where inclusion formation is a consequence of increased protein loads and heightened oxidative stress
[134]. Subsequently, Holm et al. (2020) characterized the proteomic profile of the FXTAS cortex compared to that of healthy controls (HCs). They observed a notable decrease in the abundance of proteins, such as tenascin-C (TNC), cluster of differentiation 38 (CD38), and phosphoserine aminotransferase 1 (PSAT1), in the FXTAS samples. Additionally, the authors confirmed a significantly elevated abundance of novel neurodegeneration-related proteins and small ubiquitin-like modifier 1/2 (SUMO1/2) in the FXTAS cortex compared to HCs
[27]. Furthermore, Abbasi et al. (2022) reported changes in the level of multiple proteins, including amyloid-like protein 2, contactin-1, afamin, cell-adhesion molecule 4, NPC intracellular cholesterol transporter 2, and cathepsin, by comparing the cerebrospinal fluid (CSF) proteome of FXTAS patients with HCs. Alterations in acute-phase-response signaling, liver X receptor/retinoid X receptor (LXR/RXR) activation, and farnesoid X receptor (FXR)/RXR activation pathways were also observed
[190]. In an ongoing study, the Tassone lab performed blood proteome profiling of PM-allele carriers who developed FXTAS over time and compared it to HC samples. Through this analysis, they identified potential proteomic biomarkers for early diagnosis and reported altered protein pathways between the groups, suggesting their involvement in the pathogenesis of the disorder
[144]. However, due to the limitations of a small sample size, further studies with larger cohorts are necessary to validate the initial findings and elucidate the role of the identified markers and pathways.
2.9. CGG Short Tandem Repeat (STR) Expansions
It has been outlined that the molecular cause of FXTAS is the presence of a PM ranged (55–200 units) expansion of the CGG short-tandem-repeat (STR) locus located within the 5′-UTR of the
FMR1 gene
[7]. In recent years, several other neurodegenerative disorders have been associated with a PM ranged CGG STR expansion as their genetic cause
[191][192][193][194][195]. These diseases include neuronal intranuclear inclusion disease (NIID), oculopharyngodistal myopathy (OPDM), and oculopharyngeal myopathy with leukoencephalopathy (OPML). These PM expansion loci are localized within the following genes and ncRNA: LRP12 (OPDM type 1), GIPC1 (OPDM type 2), NOTCH2NLC (OPDM type 3/NIID), RILPL1 (OPDM type 4), and LOC642361 (OPML). All of these disorders share a striking level of clinical similarity with FXTAS, suggesting a shared or similar molecular mechanism of pathology leading to a neurodegenerative phenotype. In search of potential additional disease loci, Annear and colleagues (2021) performed a bioinformatic in silico analysis of the reference genome and identified approximately 6000 additional CGG STR loci. When large population datasets were analyzed (
n > 12,000), 99% of these novel loci were demonstrated as displaying at least some degree of polymorphism across the human population, and approximately 15% of all CGG loci were observed to expand up to or beyond the 55-unit PM breakpoint
[196]. How many of these loci may be involved in neurodegenerative disease remains an enigma. While the repeat length is unlikely the only factor affecting the pathogenic potential of a given repeat, it is no doubt a core component. Moreover, half of these CGG STRs displayed characteristics similar to the known disease-linked repeats
[196]. This included high rates of polymorphism and a genetic localization within the 5′ UTR and gene promoter regions, a typical characteristic of disease-linked CGG STRs. However, there may be further factors at play, such as cis elements flanking the repeat and the reading frame of the repeat in reference to the localized gene
[132][197]. In each case, it cannot be excluded that additional expansions of CGG STRs may play a role in progressive neurodegeneration disorders with FXTAS and FXPOI-like phenotypes. While additional expansions are not detected in routine diagnostics using current short-read-based detection methods, the future introduction of long-read sequencing may expose potential additional loci in the clinic.
Fragile X-premutation-associated-condition involvement across the lifespan is presented in Figure 2.
Figure 2. FXPAC involvement across the lifespan.