The RNA-binding fragile-X mental retardation protein and its role beyond the brain

Abstract

It is well-established that variations of a CGG repeat expansion in the gene FMR1, which encodes the fragile-X mental retardation protein (FMRP), cause the neurocognitive disorder, fragile-X syndrome (FXS). However, multiple observations suggest a general and complex regulatory role of FMRP in processes outside the brain: (1) FMRP is ubiquitously expressed in the body, suggesting it functions in multiple organ systems; (2) patients with FXS can exhibit a physical phenotype that is consistent with an underlying abnormality in connective tissue; (3) different CGG repeat expansion lengths in FMR1 result in different clinical outcomes due to different pathogenic mechanisms; (4) the function of FMRP as an RNA-binding protein suggests it has a general regulatory role. This review details the complex nature of FMRP and the different CGG repeat expansion lengths and the evidence supporting the essential role of the protein in a variety of biological and pathological processes.

Introduction

RNA-binding proteins (RBPs) have emerged as being important proteins in posttranscriptional gene regulation, playing an essential role in cellular physiology. Pathogenic variants in several genes encoding RBPs can cause a variety of diseases in multiple systems, ranging from neurological disorders to cancer (Gerstberger et al. 2014). As the role of RBPs involves coordinating multiple steps in RNA processing such as splicing, mRNA export, and translation, any changes in the RNA-binding ability of these proteins can impact the expression of many different genes and pathways, leading to complex disease phenotypes (Lukong et al. 2008).

A well-studied example of pathogenic variants in a gene coding for an RBP causing a complex disease phenotype is the CGG repeat expansion in the gene fragile-X mental retardation 1 (FMR1). FMR1 encodes the fragile-X mental retardation protein (FMRP) which is a highly conserved multidomain, multifunctional RNA-binding protein that is highly expressed in the brain and plays an important role in synaptic plasticity (Darnell et al. 2011; Huber et al. 2002; Sidorov et al. 2013). Variable expression of FMR1 results in fragile-X syndrome (FXS), an X-linked inherited condition which causes mild-to-severe intellectual disability and is the most commonly known cause of autism (Timothy and Berry-Kravis 2014). While the changes to the expression of FMR1 have been well-established to affect the brain and cognitive function, FMRP is ubiquitously expressed in the body, which suggests it has specific roles in other organs (Verheij et al. 1995). This idea is further supported by the connective tissue dysplasia that can accompany the neurological symptoms of FXS and the development of two clinically distinct conditions which also arise from different variations in the FMR1 CGG repeat length: fragile-X-associated tremor/ataxia syndrome (FXTAS) and fragile-X-associated primary ovarian insufficiency (FXPOI). Therefore, elucidating the functional consequences of FMRP-RNA binding in tissues other than the brain may lead to novel insights into the pathogenesis of several diseases. This review will explore the functional complexity of FMR1 and its FMRP spice-variant products and will summarize the current literature supporting the role of FMR1 variants in disease processes outside the central nervous system.

FMRP tissue distribution

FMRP is found throughout the human body with variable tissue expression, with the highest levels observed in the brain and testes (Khandjian et al. 1995). FMRP expression has also been demonstrated in epithelial tissues, with the highest levels observed in basal dividing cells specifically seen in the skin, oesophagus, stomach, small intestine, large bowel, and respiratory epithelium (Devys et al. 1993).

Within the brain, the most intense staining of FMRP is within neuron-rich regions, including the cerebellum and hypothalamus, with a clear signal seen in the cell body of the neurons but with no significant staining in the nucleus (Devys et al. 1993; Hinds et al. 1993). The white matter shows extremely minimal staining, indicating that cells of the white matter including astrocytes and oligodendrocytes, as well as axons, likely contain very little FMRP (Devys et al. 1993).

FMRP expression in muscle is not well-characterized. Two early studies have yielded contradictory results, with one study demonstrating the presence of FMRP transcripts in human heart tissue, although the transcript was shorter than those observed in other tissues (Hinds et.al 1993). However, a second study found that human adult tissues of mesodermal origin including the heart, skeletal and smooth muscle, and dermis showed FMRP expression only under pathological conditions (Devys et al. 1993).

In adult mice, FMRP expression has been noted in the brain, lung, kidney, intestine, colon, spleen, eye, placenta, and ovary (Khandjian et al. 1995). While similarities were observed in expression between the murine and human brain, with the most intense staining in the neuron-rich regions, in contrast to the human brain, FMRP expression was also observed in glial cells in the developing murine brain at early and mid-postnatal developmental stages (Gholizadeh et al. 2015). In other murine tissues, low levels of FMRP expression have been observed in the heart, skeletal and abdominal muscle, with a predominance of shorter transcripts (Khandjian et al. 1995), although an earlier study failed to demonstrate FMRP expression in murine cardiac or aortic tissue (Hinds et al. 1993), possibly due to methodological differences.

FMRP structure and function

The FMR1 gene and its protein product FMRP

FMR1 is located on the X-chromosome (Xq27.3) and is made up of 17 exons that span approximately 38-kb of genomic DNA, including a 1.8-kb protein coding region (Fig. 1). Located within the 5′UTR is a trinucleotide CGG repeat of variable length that is transcribed into mRNA, but is not translated in the final protein product. Expansion of this CGG repeat results in methylation of a CpG island consisting of 52 CpG dinucleotides, located in the promoter region of the gene. The CGG expansion and methylation of the FMR1 promoter will be discussed later in more detail.

Fig. 1.

FMR1 gene and FMRP protein structure. FMR1 contains a coding region composed of 17 exons. The 5′UTR region contains a CGG repeat of variable length (CGG(n)), with differences in repeat length the main cause of FMR1-related pathologies. Greater than 200 repeats lead to DNA methylation of an upstream CpG island located in the promoter region of the gene. The coded protein, FMRP, is composed of a nuclear localization signal (NLS) and nuclear export signal (NES), two K-homology (KH) motifs (KH1, KH2), and a glycine-arginine-rich domain (RGG)

The FMR1 4.4-kb transcript encodes FMRP which has an approximate mass of 70 kDa. The protein contains several functional domains including three RNA-binding domains: two K-homology (KH) motifs (KH1, KH2) and an RGG box, a nuclear localization signal (NLS), and a nuclear export signal (NES). In addition, there is also evidence that the N-terminus has an RNA-binding activity (Adinolfi et al. 1999; Adinolfi et al. 2003) (Fig. 1). Identifying these conserved functional domains of FMRP has substantially contributed to understanding the functional capacity of the protein.

Alternative splicing and FMRP isoforms

Alternative splicing of RBPs has been shown to increase the functional diversity of the protein. This is evident in FMRP, as extensive alternative splicing of the FMR1 mRNA has been demonstrated, leading to the formation of multiple isoforms that have been observed in both humans and mice (Brackett et al. 2013; Huang et al. 1996; Tseng et al. 2017). Each isoform is predicted to have unique biochemical properties (Brackett et al. 2013), and as previously discussed, the differently sized transcripts have been noted to be differentially expressed in various tissue types (Khandjian et al. 1995; Verheij et al. 1995).

The major alternative splicing events that occur in FMR1 are the use of alternative splice acceptor sites in exon 15 and exclusions of exons 12 and 14 (Ashley et al. 1993; Verkerk et al. 1993). A particular phosphorylation site, Ser-500 in humans (Ser-499 in mice), essential for the translational repressor function of FMRP, has been mapped to exon 15 (Ceman et al. 2003; Pretto et al. 2015). The use of first acceptor splicing sites on exon 15 removes this key phosphorylation site, while use of the second acceptor site also results in deletion of the site in addition to removing the amino acids that comprise the recognition site required for methylation. Splicing events that result in the skipping of exon 12 produce a truncated variable loop between the b2 and b′ strands within the second KH domain of the protein, which suggests these transcripts may regulate different FMRP mRNA associations (Brackett et al. 2013). Deletion of exon 14 removes the nuclear export signal, and therefore, these protein isoforms will be retained in the nucleus (Sittler et al. 1996). This deletion also leads to a + 1 frame shift in the coding sequence, resulting in truncated FMRP isoforms with novel C-termini (Sittler et al. 1996). The diversity of these FMRP isoforms highlights the possibility of diverse biological properties, emphasizing the likely role of FMRP in a variety of cellular processes.

FMRP-RNA binding and translation control

A well-established property of FMRP is its role as an RBP. RBPs have several structural motifs that facilitate their binding to the specific target RNA. Having multiple RNA-binding motifs within an RBP is common and is a characteristic that is thought to increase specificity for its RNA targets, or may allow the binding of multiple RNAs at once (Darnell et al. 2005).

It has been estimated that FMRP can directly target and bind to 4% of human brain mRNA (Ashley et al. 1993). The binding event that is best characterized is the binding of FMRP to mRNAs at neuronal synapses causing translational repression. Following specific neuronal stimuli, FMRP will dissociate from the mRNA, which allows transcription of the necessary protein to occur, thereby contributing to the complex functionality of the synapse (Suhl and Hoeffer 2017).

There are two RNA motifs that specifically interact with the RNA-binding domains of the FMRP protein and mediate FMRP-RNA interactions. These two motifs are the mRNA G-quartet that binds with high affinity to the RGG box, and the mRNA “kissing complex” that binds to the KH2 domain (Ashley et al. 1993). Naturally occurring RNA structural motifs that are recognized by FMRP continue to be identified. These include a novel motif, Sod1 mRNA Stem Loops interacting with FMRP (SoSLIP), within the mRNA for superoxide dismutase (SOD1), which through FMRP interaction enhances translation of SOD1 mRNA, thereby identifying a role of FMRP as a positive modulator of translation (Bechara et al. 2009).

FMRP can also regulate translation by actively binding to translating polyribosomes in an RNA-dependent interaction in neural and non-neural cell lines (Stefani et al. 2004). The KH domains appear to be essential for this translational control, as FMRP transcripts harbouring mutations in the KH domains or full KH1 or 2 domain deletion do not associate with polyribosomes, while FMRP transcripts with RGG box domain deletions still bind to active ribosomes, despite the absence of an RGG box motif (Darnell et al. 2005). The importance of the RNA-binding function of FMRP is further emphasized by the observation that a FMR1 gene point mutation, I304N, impairs RNA and polyribosome binding and leads to a severe neurological phenotype of FXS, highlighting the fact that the RNA-binding capacity of this protein is an integral activity, particularly for neurological signalling (Laggerbauer et al. 2001; Siomi et al. 1994).

A potential nuclear function of FMRP

While FMRP is mainly found in the cytoplasm, evidence supports a potential nuclear function of the protein. The NLS and NES domains of the protein are responsible for the shuttling of the protein between the nucleus and cytoplasm. Additionally, FMRP has been shown to be involved in mRNA transport and nuclear export of m6-A-containing mRNAs (Edens et al. 2019). Furthermore, FMRP maybe involved in the replication-stress-induced DNA damage response by binding chromatin (Alpatov et al. 2014). FMRP isoforms which skip exon 14 loose the NES and therefore are retained in the nucleus (Sittler et al. 1996). The exact function of these isoforms has not yet been elucidated.

The complex nature of the CGG expansion

The effect of the number of CGGs on transcriptional and translational control of the FMR1 gene is complex and is a topic of interest, as variations in the repeat size can lead to significantly different phenotypic outcomes. The American College of Medical Genetics has established four different allelic classes based on the number of CGG repeats for clinical diagnostic purposes. These classes are summarized in Table 1 (Sherman et al. 2005).

Table 1.

Summary of the FMR1 CGG repeat allelic classes

Number of repeats	Category	mRNA expression	Protein expression	Clinical significance
6-44	Normal (most common 29-31 repeats)	Normal	Normal	Evidence suggesting low and high ranges of normal is associated with ovarian dysfunction
45-54	Intermediate	Increased	Unknown	Not well-established but evidence suggests possible predisposition to Parkinsonism and cognitive behavioural disorders
55-200	Premutation	Increased	Normal/slight decrease	FXTAS, FXOPI
> 200	Full mutation	Decreased	Decreased/no expression	FXS

Normal allele

The normal allele size is 6-44 CGG repeats, with the majority of the population having between 29 and 31 repeats. This results in normal expression levels of FMRP with a ‘normal’ phenotype, with no apparent pathology. However, there is some variability, as premature ovarian ageing, infertility, and poor response to IVF treatment have been observed in female carriers of repeat expansions in the ‘low’ and ‘high’ normal range (Bretherick et al. 2005; Lu et al. 2017; Wang et al. 2018).

Full mutation alleles

Carriers of the full mutations have greater than 200 CGG repeats, with the prevalence of the full mutation ranging from 1:2500 to 1:8000 in females and 1:4000 to 1:5000 in males (Hagerman et al. 2017). The full mutation leads to increased methylation on the FMR1 CpG island located upstream of the CGG repeat (Fig. 1). As there are transcription start sites located within this CpG island, when it is methylated as is the case in the presence of the full mutation, there is interference of transcription machinery and subsequent inhibition of gene transcription (Usdin and Kumari 2015). This results in little-to-no expression of FMRP, which phenotypically results in FXS (Pieretti et al. 1991).

Premutation alleles

The premutation alleles are between 55 and 200 CGG repeats, with a prevalence approximated to be 1 in 130–259 women and 1 in 250–813 men, and predispose affected individuals to FXTAS and FXPOI in women (Winarni et al. 2012). As carrying the premutation leads to different phenotypic features compared with the full mutation, it suggests that the different CGG repeat lengths result in different pathological mechanisms (Hagerman 2013).

Analysis of FMR1 mRNA and FMRP levels in carriers of the premutation have demonstrated that mRNA levels are 2- to 8-fold higher than carriers of the normal allele, with a paradoxical slight decrease in FMRP levels, suggesting a reduction in translational efficiency (Kenneson et al. 2001; Primerano et al. 2002). The cause of elevated mRNA in premutation carriers is not well-understood; however, it has been proposed that FMR1 contains three transcription start sites and that the number of CGG repeats in the downstream region may affect which start site is used for transcription and the relative efficiencies of these start sites (Beilina et al. 2004).

The abnormalities that arise as a result of the increase in mRNA indicate a possible toxic RNA gain-of-function effect. This pathogenic mechanism suggests that FMR1 mRNA containing the repeat expansion is able to bind to other RBPs, sequestering them in the nucleus, which leads to dysregulation of their usual targets. This hypothesis is supported by the large nuclear inclusions that are evident in premutation carriers that are composed of FMR1 mRNA and proteins, half of which are involved in RNA binding (Ma et al. 2019). This concept is well-characterized in myotonic dystrophy type 1 (DM1), a rare neuromuscular disorder caused by expansion of CTG repeats in the gene dystrophia myotonica protein kinase (DMPK) (Lukong et al. 2008). The mRNA RBP MBNL1 has an integral role in regulating splicing events of DMPK in striated muscle; however in DM1, the expanded number of CTG repeats results in the sequestration of MBNL1 to the nucleus, which results in mis-splicing of several other mRNAs, leading to the observed clinical manifestations (Lukong et al. 2008).

It should be noted that while the RNA toxicity effect is used to describe the features of premutation carriers, this idea should stand for any expanded alleles that are producing high levels of FMR1 mRNA. In particular, FXTAS presentation may extend beyond premutation range but is ultimately caused by excess mRNA rather than FMRP loss of function (Hagerman 2013). This is predominantly evident in the observation of a FXTAS phenotype caused by rarely occurring unmethylated full mutation carriers, and males that exhibit methylation mosaicism, defined as a mixed pattern of methylated and unmethylated full mutation alleles (Loesch et al. 2012; Santa María et al. 2014; Tassone et al. 2000).

Intermediate (grey zone) alleles

The intermediate alleles have CGG expansions between 45 and 54 repeats and are thought to be present in approximately 4% of the population (Kenneson et al. 2001). While carriers of the intermediate allele are generally phenotypically normal, growing evidence suggests that these carriers have a predisposition to FXTAS and FXPOI, which are typically seen in premutation carriers, as well as a predisposition to Parkinsonism and cognitive and behavioural disorders (Bretherick et al. 2005; Hall et al. 2012; Hall 2014; Streuli et al. 2009). Little research has been done on how ‘grey zone’ repeats affect FMR1 transcription and FMRP expression, with one study demonstrating an increase in transcriptional activity in male intermediate allele carriers when compared with normal carriers (Loesch et al. 2007).

FMR1-related syndromes

Fragile-X syndrome

FXS is the most commonly inherited cause of intellectual disability and affects approximately 1:5000 males and 1:4000 to 1:8000 females and is caused by the FMR1 full mutation (Hagerman et al. 2017). As FMR1 is located on the X-chromosome, this results in an increase in frequency in males with a more severe phenotype in most instances (Kidd et al. 2014). FMR1 mRNA was shown to be absent in leukocytes obtained from FXS males, highlighting the loss of gene expression caused by the large expansion of the trinucleotide repeat (Pieretti et al. 1991).

Loss or reduction of FMRP expression disrupts the RNA-binding activity-dependent protein synthesis within synapses, a key component involved in synaptic plasticity and cognitive development, leading to various levels of intellectual disability, with males with FXS having an IQ well below average (Garber et al. 2008). Males with FXS also exhibit autistic characteristics, with 15–25% of individuals meeting the diagnostic criteria for autism (Bailey et al. 2000). Behavioural characteristics noted in FXS patients include hand flapping, difficulty making eye contact, anxiety, mood disorders, hyperactivity, and aggressive behaviour (Garber et al. 2008). Females with FXS have a normal-to-borderline IQ; however, they are more vulnerable to emotional deficits, including social anxiety, shyness, and depression (Freund et al. 1993).

Patients with FXS can also display a range of distinct physical characteristics, including long faces, prominent ears, flat feet, joint laxity, pectus deformities, hyperextensible finger joints, soft skin, and enlarged testicles (macroorchidism) first evident in prepubescent males (Hagerman et al. 1984b; Kidd et al. 2014). There is also an increased incidence of mitral valve prolapse (MVP) and aortic root dilatation, recurrent otitis media, and gastrointestinal problems including diarrhoea and gastroesophageal reflux disease (Kidd et al. 2014). These non-neural features of FXS suggest that FMR1 and FMRP play a role in connective tissue development and integrity (Alanay et al. 2007).

Fragile-X-associated tremor/ataxia syndrome

FXTAS is a neurodegenerative syndrome that predominately occurs in male FMR1 premutation carriers, with patients typically presenting with progressive cerebellar ataxia, action tremor, Parkinsonism, and cognitive decline (Hagerman 2013). An age-dependant penetrance has been noted, with approximately 40% of male premutation carriers over the age of 50 developing the disorder, increasing to 75% of men 80 years or older being affected (Jacquemont et al. 2004). The prevalence of FXTAS is significantly less in female premutation carriers, with 8% of carriers developing the condition (Coffey et al. 2008). In addition, the typical clinical picture of FXTAS in females is generally less severe; however, atypical FXTAS symptoms including chronic muscle pain, fibromyalgia, and higher rates of movement disorders and seizures have been suggested in females (Leehey 2009).

The presence of eosinophilic, ubiquitin-positive, intranuclear inclusions within neurons and astrocytes throughout the brain and brainstem is a common finding at autopsy in those affected by FXTAS, along with white matter changes and general brain atrophy (Hagerman 2013). These inclusions are also observed in tissues outside the central nervous system, with inclusions found in the peripheral nervous system and other solid organs including the pancreas, thyroid, adrenal gland, gastrointestinal tract, pituitary gland, pineal gland, heart, and mitral valve (Buijsen et al. 2014; Gokden et al. 2009; Hunsaker et al. 2011), which further emphasizes a more systemic role of FMRP.

Fragile-X-associated primary ovarian insufficiency

There is a spectrum of diminished ovarian reserve in FMR1 female premutation carriers which results in menstrual cycle irregularities, hormonal fluctuations, and decreased fertility, with premutation carriers on average developing menopause 5 years earlier when compared with non-carriers (Murray 2000). The most severe end of this spectrum is primary ovarian insufficiency (POI), which is defined as diminishing ovarian function leading to menopause before the age of 40. Female premutation carriers have a high rate of POI, with 20% experiencing POI compared with 1% of the normal population (Sherman 2000). Furthermore, women with intermediate alleles or CGG repeat numbers that are in the upper limit of normal also have a higher frequency of POI when compared with the general population (Bretherick et al. 2005). Possible mechanisms linking FMRP and ovarian function will be addressed in detail later.

FMRP beyond the brain

FMRP and connective tissue

The abnormal musculoskeletal and cardiovascular features seen in patients with FXS described previously are commonly seen in connective tissue disorders such as Marfan syndrome (MFS) and Elhers-Danlos syndrome, which are known to predispose patients to life-threatening cardiovascular abnormalities, therefore highlighting the possibility of systemic connective tissue abnormalities in FXS.

Musculoskeletal abnormalities

Musculoskeletal features are present in approximately half of patients with FXS with varying degrees of penetrance (Hagerman et al. 1984a; Opitz et al. 1984). A study investigating the orthopaedic aspects of 150 patients with FXS reported that 50% had flat feet, 57% had excessive laxity of the joints, and 7% had scoliosis (Davids et al. 1990). The incidence of musculoskeletal findings was most common in younger patients, with 73% of patients aged between 0 and 10 years old manifesting musculoskeletal abnormalities (Davids et al. 1990).

An alteration in endochondral ossification, the process of bone development from cartilage, is seen during embryogenesis in FXS, particularly in facial regions including the nasal passage and the jaw as well as the hands (Hjalgrim et al. 2000). This phenotype in utero can predict the delay in skeletal maturation that is subsequently seen in children with FXS and may also relate to the facial abnormalities in FXS patients such as a long narrow face, prominent jaw, and highly arched palates (Garber et al. 2008; Kjær et al. 2001).

Cardiovascular abnormalities

Cardiovascular abnormalities in patients with FXS have been described since 1982 and include MVP associated with mitral regurgitation and dilatation of the ascending aorta (Hagerman et al. 1984a; Pyeritz et al. 1982). Larger cohort studies have subsequently determined that MVP is present in 13–55% of patients with FXS, with the frequency increasing with age, with MVP present in 80% of males over the age of 18 years (Loehr et al. 1986). Aortic root dilatation is an age-dependent phenomenon, and therefore, incidence rates do vary between the studies with increasing incidence amongst older cohorts. Documented aortic dilatation incidences range between 9 and 10% for patients under the age of 18 years compared with an incidence rate between 33 and 52% for patients over the age of 18 years (Alanay et al. 2007; Crabbe et al. 1993; Loehr et al. 1986; Sreeram et al. 1989).

Other noted cardiac abnormalities in patients with FXS include aortic insufficiency and increased left ventricular end diastolic dimension and left ventricular hypertrophy, which are likely to be a consequence of the underlying mitral valve abnormalities (Loehr et al. 1986; Sabaratnam 2000).

As FXS is more prevalent in males, these cohort studies have had a male predominance, which makes it difficult to form any definitive conclusions regarding sex, but it is interesting to note that no aortic dilatation was seen amongst the females examined (n = 6) (Loehr et al. 1986). However, there are 2 case reports of spontaneous coronary artery dissection in females that carry FMR1 premutation (McKenzie et al. 2020).

FMRP and connective tissue dysregulation

While these musculoskeletal and cardiovascular manifestations are well-documented in patients with FXS suggesting an intrinsic abnormality in their connective tissue, very little is known about the causative relationship between a decrease in FMRP expression and connective tissue dysregulation.

Connective tissue can be described as a structural framework providing support and connection between all tissues, comprised of cells surrounded by extracellular matrix (ECM). The ECM is predominantly made up of proteoglycans; glycoproteins, including fibronectins and laminin; collagen; and elastic fibres (composed of elastin and fibrillin) (Kierszenbaum and Tres 2015). It can be hypothesized that the underlying connective tissue abnormalities in FXS are associated with dysregulation of elastin or collagen fibres due to the similarity between the clinical features seen in FXS and connective tissue disorders that involve defects in collagen synthesis (Elhers-Danlos syndrome) and the integrity of elastic fibres (MFS) (Jeremy et al. 2013). MFS is caused by pathogenic variants in FBN1, which encodes the protein fibrillin-1, which is essential for the formation and integrity of elastic fibres. Abnormalities in or deficiency of fibrillin-1 result in an intrinsic weakness of connective tissue structure, the most life-threatening being the structural weakness of the aortic wall leading to aneurysm formation, which predisposes to subsequent dissection (Ammash et al. 2008). Analysis of aortic tissue samples obtained from aneurysms of patients with MFS has demonstrated elastin fragmentation and increased mucopolysaccharide deposition (Romaniello et al. 2014).

Elastin abnormalities have been observed in the skin of patients with FXS, with skin biopsies of five patients with FXS, examined using a light microscope, demonstrating that the elastin fibre formation in the upper dermis was minimal, and that shorter and fragmented elastin fibres were present in the deeper dermis when compared with control skin samples (Waldstein et al. 1987). A similar finding was reported in a case report of an 18-year-old male with FXS who died suddenly of cardiac arrest, with the skin showing obvious elastin abnormalities, with depletion and fragmentation of elastin fibres in the aorta and mitral valves in addition to an increase in collagen (Waldstein and Hagerman 1988).

Evidence of acid mucopolysaccharide malfunction has also been noted in histological examination of the aorta, skin, and tissue supporting skeletal development in patients with FXS (Hjalgrim et al. 2000; Sabaratnam 2000; Waldstein et al. 1987).

Aberrant signalling within the aortic wall has also been implicated in the pathogenesis of MFS aneurysms, favouring ECM degradation and contributing to weakening of the vessel wall (Robertson et al. 2015). Due to the similarities in histological presentation between MFS and the connective tissue abnormalities in FXS, consideration of the pathways and proteins that contribute to elastin fragmentation in MFS may assist in further understanding the relationship between FMRP and connective tissue dysregulation.

FMRP and MMPs

A family of proteinases involved in the fragmentation of elastic fibres in MFS is matrix metalloproteinases (MMPs). There are over twenty types of MMPs identified in humans, each playing a role in degrading various components of the ECM (Vu and Werb 2000), with MMP-2 and MMP-9 in particular involved in the pathogenesis of MFS aneurysm formation, with both having elastolytic and collagenolytic properties (Chung et al. 2007b). MMP-9 also has a pivotal role in the process of endochondral ossification, which is essential in skeletal development, a process that is impaired in FXS, evident through the musculoskeletal abnormalities described earlier (Blavier and Delaisse 1995).

MMP-9 mRNA has been identified as an FMRP target (Janusz et al. 2013; Lucá et al. 2013). In silico analysis of murine MMP-9 revealed RNA sequence motifs that are typically bound by FMRP (Janusz et al. 2013). This relationship was confirmed through investigating murine hippocampal regions, with RNA coimmunoprecipitation techniques confirming the presence of MMP-9 mRNA in a complex with FMRP (Janusz et al. 2013). The relationship between FMRP and MMP-9 is further emphasized by increased MMP-9 protein expression seen in the hippocampus of young and adult Fmr1 KO mice as well as in human FXS post-mortem brain samples when compared with controls (Bilousova et al. 2009; Sidhu et al. 2014). Interestingly, while MMP-9 protein expression was increased, there was no difference in MMP-9 mRNA seen in the hippocampus of adult Fmr1 KO mice, indicating that changes in protein levels are due to changes in posttranscriptional regulation (Sidhu et al. 2014).

The translation of synaptic MMP-9 is regulated by FMRP in a mouse model, with absence of FMRP in Fmr1 KO mice leading to an increase of MMP-9 mRNA translation at hippocampal synapses, which may explain the increased MMP-9 protein levels seen in FMRP-deficient models (Janusz et al. 2013). In addition, minocycline, a drug that is known to inhibit MMP-9 activity in FXS animal models, results in an increase in mature dendrite formation and improves behavioural deficits. Minocycline is therefore a promising therapeutic option, with clinical trials showing minocycline treatment in FXS results in improvements in a range of behavioural and neurological deficits (Dziembowska et al. 2013; Leigh et al. 2013; Paribello et al. 2010; Siller and Broadie 2012).

While most of the work investigating the relationship between MMP-9 and FMRP has been focused on the effects in the brain, Fmr1/MMP-9 double KO mice corrected both neural abnormalities, including dendritic spine morphology and behavioural deficits, and more interestingly non-neural abnormalities associated with FXS, suggesting that aberrant regulation of MMP-9 expression may also contribute to the features of FXS external to the brain (Sidhu et al. 2014). Interestingly, musculoskeletal changes have been noticed in patients with FXS being treated with minocycline, such as an increase in foot arch formation and more stable foot positioning (Ethell and Sidhu 2017).

With the known relationship between MMP-9 expression and connective tissue disorders and the likelihood of a connective tissue weakness leading to unstable joints, aortic dilatation, and MVP in FXS, further investigation into the role of FMRP in regulating MMP-9 translation in other tissues including the vasculature and cardiac valves is warranted.

FMRP and endothelial and vascular smooth muscle cell regulation

Changes in endothelial function and vascular smooth muscle cell (VSMC) phenotype are implicated in several cardiovascular diseases including MFS and atherosclerosis (Chistiakov et al. 2015; Chung et al. 2007a; Crosas-Molist et al. 2015; Mudau et al. 2012; Nolasco et al. 2020).

In human umbilical vein endothelial cells (HUVECs), FMRP was shown to regulate cellular proliferation and angiogenesis through modulating the CaM/CaMKII pathway and miR-181a expression (Zhao et al. 2018). Specifically, FMRP may play a pivotal role in vascular integrity in response to inflammatory stimuli, as treatment of the HUVECs with TNF-α induced the dephosphorylation of FMRP leading to a decrease in CaM/CaMKII pathway activity and a reduction in cellular proliferation and angiogenesis (Zhao et al. 2018).

β-Catenin signalling plays an essential role in VSMC phenotype modulation in response to various stimuli (George and Beeching 2006). In rat primary VSMCs, FMRP and β-catenin interact at an endogenous level, with a direct interaction noted at the messenger ribonucleotide protein and translational pre-initiation complex (Ehyai et al. 2018). Therefore, FMRP regulation may play a role in VSMC phenotype modulation through regulating β-catenin signalling.

FMRP and the reproductive system

As previously mentioned, FMRP is highly expressed in the testes, with macroorchidism being a common feature seen in males with FXS. The cause of macroorchidism is unclear, with no consistent pathological abnormalities documented, although it has been noted there is a general reduction in tubule diameter, malformed spermatids, and a large increase in tubular length (Johannisson et al. 1988; Lachiewicz and Dawson 1994; Nistal et al. 1992). The number of germ cells present in an adult testis significantly influences the size of the organ, and therefore, a decrease in FMRP may have an effect on the number of germ cells present. This hypothesis is supported by animal studies, with the enlarged testes in the male Fmr1 KO mice attributed to an increased rate of Sertoli cell proliferation, resulting in an increase in germ cell number and testicular size (Slegtenhorst-Eegdeman et al. 1998). Furthermore, this increase in Sertoli cell proliferation was not influenced by changes in circulating follicle-stimulating hormone (FSH).

The relationship between the FMR1 CGG expansion and ovarian function is complicated. In female carriers of an FMR1 premutation, there is an increased risk of developing POI. In addition, a significant increase in the number of FMR1 CGG repeats in both the high end of the normal range and the grey zone was observed in a case-control study of 53 women with idiopathic POI (Bretherick et al. 2005). However, there appears to be a non-linear relationship between CGG repeat numbers and ovarian function, with women carrying mid-size range expansions (80–120 repeats) having significantly less oocyte retrieval during in vitro fertilization (IVF) cycles when compared with women with higher or lower repeats (Elizur et al. 2014). Furthermore, women who carry the FMR1 full mutation do not seem to be at an increased risk of FXPOI (Sherman 2000).

The underlying pathogenetic mechanism connecting FMR1 expression and FXPOI is not well-understood. Numerous studies have shown that female FMR1 premutation carriers have altered hormone levels, with an increase in FSH that suggests a reduced residual follicle pool (Hundscheid et al. 2001; Murray et al. 1999; Welt et al. 2004). A FMR1 premutation mice model has further supported this with all follicle classes lost at a faster rate when compared with wild-type (WT) mice, demonstrating an intrinsic abnormality of the ovary in FXPOI (Hoffman et al. 2012). This is thought to be due to FMR1 premutation mRNA-mediated toxicity to the granulosa cells, a hypothesis that is supported by studies indicating that FMR1 mRNA levels in granulosa cells also significantly correlated with the number of oocytes retrieved during IVF (Elizur et al. 2014). In addition, oocytes from murine Fmr1 premutation carriers have aberrant nuclear accumulation of FMRP and elevated levels of ubiquitination, similar to the nuclear inclusions associated with the premutation-associated FXTAS (Hoffman et al. 2012).

FMRP and cancer

There is a well-established link between neurodegenerative diseases and cancer, with a possible connection being the role of RNA-binding proteins, like FMRP (Campos-Melo et al. 2014). FMRP is overexpressed in human hepatocellular carcinoma, breast cancer and pancreatic ductal adenocarcinoma cell lines, melanoma, and murine neuroendocrine tumours (Li et al. 2018; Li et al. 2003; Liu et al. 2007; Lucá et al. 2013; Zalfa et al. 2017). FMRP has also been associated with disease progression and metastasis in breast cancer and melanoma, with possible mechanisms involving the binding of FMRP to, and subsequent regulation of, mRNA of relevant proteins essential to these processes such as epithelial mesenchymal transition and invasion (Li et al. 2018; Lucá et al. 2013; Zalfa et al. 2017). There is also evidence that in patients with FXS that express little-to-no FMRP, there is a decreased risk of developing cancer, further indicating a connection between FMRP and cancer (Schultz-Pedersen et al. 2001).

FMRP and tissue remodelling

Under certain pathological conditions, the expression of FMRP expression is significantly increased in human cardiac tissue and the dermis of the skin, in particular if these tissues are undergoing active remodelling. Analysis of heart tissue obtained from a patient with severe ischemic cardiomyopathy demonstrated that myocytes in the hyperplastic region were expressing FMRP, with the surrounding healthy myocytes having no expression (Devys et al. 1993). A similar observation was noted during wound healing where dermal fibroblasts had FMRP expression that differed from FMRP expression in normal skin (Devys et al. 1993).

FMRP, immune dysfunction, and infectious disease

Immune dysfunction

It has been suggested that there is aberrant regulation of the immune system in FXS. This was investigated in a large cohort study of over 5000 patients with FXS that were compared with over 5000 age- and gender-matched controls (Yu et al. 2020). This study demonstrated that patients with FXS had a vulnerability to infections including otitis media, viral enteritis, candidiasis, pneumonia, and acute sinusitis when compared with non-FXS individuals, with smaller scale studies reporting similar reports of increased infections (Hagerman et al. 1987). In addition, patients with FXS have elevated pro-inflammatory cytokine levels seen in the plasma of males with FXS when compared with age-matched controls (Ashwood et al. 2010).

A large human cohort study has also demonstrated that patients with FXS are less likely to develop autoimmune disorders (Yu et al. 2020). However, female FMR1 premutation carriers have been shown to be more prone to developing autoimmune disorders with one study indicating that 45% of patients with the premutation had at least one immune-mediated disorder, compared with 28% of the controls (Winarni et al. 2012). The most common immune-mediated disorders identified amongst these patients include autoimmune thyroid disorders, fibromyalgia, and irritable bowel syndrome (Coffey et al. 2008).

Several animal models further support that immune dysregulation occurs due to FMR1 variants. Genome-wide mRNA expression profiling comparing gene expression between brain tissue from Fmr1 KO and WT mice identified an enrichment of differentially expressed genes in immunological signalling pathways (Prilutsky et al. 2015). In addition, a Drosophila model of FXS (dfmr1) also demonstrated that dfmr1 mutants were more susceptible to bacterial infection, with evidence of decreased phagocytic activity in hemocytes (primitive macrophages) systemically and in glial cells of the brain, suggesting that defects in FMR1 may compromise bacterial clearance, making subsequent infections more likely (O’Connor et al. 2017).

Viral replication

Viruses are successful pathogens due to their ability to replicate within host cells, through interactions with numerous host molecules including RBPs (Nagy and Pogany 2012). FMRP in host cells can influence viral replication with evidence to support both a role of FMRP in facilitating and inhibiting viral replication.

In both cell culture and mice, FMRP is involved in the assembly of influenza A ribonucleoprotein, which is essential for the transcription and replication of viral RNA in the nucleus of host cells (Zhou et al. 2014). The FMRP-ribonucleoprotein association is mediated by the FMRP-RNA-binding KH2 domain, suggesting an RNA-mediated interaction that is facilitating viral replication (Zhou et al. 2014). FMRP has also been associated with replication and infectivity of tick-borne encephalitis virus, with silencing of FMRP significantly reducing viral replication (Muto et al. 2018).

Conversely, FMRP has been shown to restrict human immunodeficiency virus type 1 (HIV-1) infectivity with viruses from FMRP knockdown cells 1.8-fold more infectious than control cells, further supported by the fact that overexpression of FMRP diminished HIV-1 infection (Pan et al. 2009). FMRP has also been shown to inhibit Zika virus infectivity through binding to viral RNA and inhibiting translation in cultured cells and mouse testes (Soto-Acosta et al. 2018). These results were confirmed with a 2- to 3-fold increase in the infection rate observed in a HeLa FMRP knockdown model (Soto-Acosta et al. 2018).

FMRP and oxidative stress

Oxidative stress is due to an imbalance between the production of oxygen free radicals and the ability of the body to counteract their harmful effects by antioxidants, with oxidative stress contributing to a wide range of disease processes (Liguori et al. 2018). Oxidative stress is thought to play a role in FXS, with studies demonstrating that neural development progenitor cells differentiated from FXS-patient-derived iPSCs have exhibited increased levels of oxidative stress, including mitochondrial dysfunction (Shen et al. 2019). However, to date, there are no studies that have investigated the levels of oxidative stress in the circulation or tissues of patients with FXS.

A stronger link between oxidative stress and FMRP has been established in Fmr1 KO mice. Metabonomic analysis of murine brain tissue demonstrated that a group of metabolites involved in the oxidative stress response were altered in the FXS murine model (Heulens et al. 2011). In addition, an increase in reactive oxygen species, abnormal antioxidant generation and enhanced oxidation have been observed in the brain and testes of Fmr1 KO mice (El Bekay et al. 2007; Romero-Zerbo et al. 2009). Furthermore, Fmr1 KO mice treated with a well-known potent antioxidant, melatonin, showed improvement in anxiety behaviours and learning deficits, highlighting the contribution of oxidative stress to the clinical features associated with FXS (Romero-Zerbo et al. 2009).

The mechanisms that underlie how FMRP regulates oxidative processes are still unclear. SOD1 is a protein well-known for its antioxidant properties and role in limiting oxidative stress, with SOD1 mRNA identified as a target of FMRP in cultured primary neuron. FMRP has a role in enhancing translation of the SOD1 protein, with an absence of FMRP resulting in decreased SOD1 levels (Bechara et al. 2009; Miyashiro et al. 2003) that would in turn potentiate oxidative stress. The interaction of FMRP with proteins involved in oxygen free radical production and the oxidative stress response requires further clarification, specifically in non-neural tissues.

FMRP and lipid and glucose metabolism

Several studies have suggested that FXS is associated with abnormalities in metabolism. Two retrospective reviews of lipid levels in patients with FXS demonstrated that total cholesterol, low-density lipoprotein (LDL), and high-density lipoprotein (HDL) were all significantly lower in males with FXS when compared with control subjects, with an increase in serum triglycerides (Berry-Kravis et al. 2015). Interestingly, there was no positive correlation between cholesterol and BMI in the patients with FXS, which was surprising as this is usually a typical relationship seen in the general population, suggesting that the alterations in lipid metabolism are likely due to the FMRP deficit (Berry-Kravis et al. 2015; Lisik et al. 2016).

It has also been suggested that there is an underlying abnormality of proprotein convertase subtilisin/kexin type 9 (PCSK9) function in patients with FXS, with no correlation between PCSK9 levels and total cholesterol (Caku et al. 2017). Ordinarily, PCSK9 functions to regulate cholesterol homeostasis and PCSK9 levels often correlate with cholesterol levels (Lambert et al. 2009). Therefore, the lack of association between the PCSK9 levels and total cholesterol in patients with FXS indicates that FMRP may impact PCSK9 expression.

Increased rates of glucose metabolism and altered concentrations of metabolomes involved in energy metabolism have been observed in the brain of Fmr1 KO mice (Heulens et al. 2011; Qin et al. 2002). In addition, in the dfmr1 model of FXS, insulin signalling is increased in the brain of mutant flies, with subsequent learning and memory deficits able to be restored through genetically reducing insulin signalling (Monyak et al. 2017). FMRP may have a more systemic role in regulating carbohydrate metabolism in other Drosophilia models, with dfmr1 mutants showing a dramatic reduction in total glucose and glycogen levels highlighting that this association is not confined to the brain (Weisz et al. 2018).

This broad role of FMRP in metabolism has also been observed in mice, with metabolic phenotyping of an Fmr1 KO mice model used to investigate the consequences of FMRP deficiency on metabolic homeostasis (Leboucher et al. 2019). The three most salient features of this study were (1) FMRP deficiency enhanced glucose tolerance and the insulin response; (2) FMRP deficiency increased the utilization of lipids as an energy substrate with increased lipolysis in adipocytes; and (3) FMRP deficiency leads to changes in hepatic protein expression, targeting mRNA in the liver linked to lipid homeostasis.

Overall, it appears that FMRP plays an essential role in regulating lipid and glucose metabolism, which is confirmed in animal models of FXS. However, this finding is yet to be fully understood in humans and warrants further investigation.

FMRP and miRNAs

microRNAs have been implicated in the context of several disease processes, which are beyond the scope of this review (Portelli et al. 2018). However, it has been suggested that FMRP may regulate its target mRNAs through miRNA involvement (Jin et al. 2004). Furthermore, FMRP is associated with the Dicer complex, which is responsible for the processing of pre-miRNA to mature miRNA and would have broad reaching implications to protein expression profiles (Cheever and Ceman 2009). Therefore, as FMRP is an accessory protein to the miRNA pathway, this may further implicate FMRP as an important player in a range of disease processes in a variety of systems.

Conclusions

Understanding the broad role that RBPs play in cellular functions has helped to further our understanding of certain pathological conditions. Despite the ubiquitous expression of the RBP FMRP and the wide range of abnormalities associated with the FMR1 CGG repeat expansion, there are a limited number of studies that explore the role in FMRP in tissues outside the central nervous system. The role of FMRP is complex and involves many systems, and therefore, the protein is likely to contribute to a range of diseases. While the different FMR1 CGG repeat expansions result in specific pathological conditions, there is a degree of heterogeneity. As all repeat lengths are relatively prevalent in the general population, understanding the differences in FMR1 repeat lengths and subsequent FMRP expression and their impact on normal cellular physiology is essential, as there is likely to be subclinical effects that may modulate disease phenotypes in the systems and processes that FMRP regulates.