Repeat-mediated genetic and epigenetic changes at the FMR1 locus in the Fragile X-related disorders

Abstract

The Fragile X-related disorders are a group of genetic conditions that include the neurodegenerative disorder, Fragile X-associated tremor/ataxia syndrome (FXTAS), the fertility disorder, Fragile X-associated primary ovarian insufficiency (FXPOI) and the intellectual disability, Fragile X syndrome (FXS). The pathology in all these diseases is related to the number of CGG/CCG-repeats in the 5' UTR of the Fragile X mental retardation 1 (FMR1) gene. The repeats are prone to continuous expansion and the increase in repeat number has paradoxical effects on gene expression increasing transcription on mid-sized alleles and decreasing it on longer ones. In some cases the repeats can simultaneously both increase FMR1 mRNA production and decrease the levels of the FMR1 gene product, Fragile X mental retardation 1 protein (FMRP). Since FXTAS and FXPOI result from the deleterious consequences of the expression of elevated levels of FMR1 mRNA and FXS is caused by an FMRP deficiency, the clinical picture is turning out to be more complex than once appreciated. Added complications result from the fact that increasing repeat numbers make the alleles somatically unstable. Thus many individuals have a complex mixture of different sized alleles in different cells. Furthermore, it has become apparent that the eponymous fragile site, once thought to be no more than a useful diagnostic criterion, may have clinical consequences for females who inherit chromosomes that express this site. This review will cover what is currently known about the mechanisms responsible for repeat instability, for the repeat-mediated epigenetic changes that affect expression of the FMR1 gene, and for chromosome fragility. It will also touch on what current and future options are for ameliorating some of these effects.

Keywords: FX-associated primary ovarian insufficiency (FXPOI); FX-associated tremor/ataxia syndrome (FXTAS); Fragile X syndrome (FXS); Fragile X-related disorders; repeat expansion disease.

FIGURE 1

Models for intergenerational repeat instability at the FX locus. In both models the CGG-rich strand, which forms the most stable secondary structure, is shown in red and the CCG-rich strand is shown in green. (A) ORI-switch model for FX repeat instability. The FMR1 gene is flanked by two ORIs, one located 45 kb upstream (5′) and another 45 kb downstream (3′) of the gene (Gerhardt et al., 2014). During replication transient dissociation of the Watson and Crick strands of DNA can occur with slippage of the two strands relative to one another. Priming of DNA synthesis can then occur from the slipped position. This strand-slippage process is thought to occur more commonly during lagging strand replication and is seen more frequently when the template or nascent strand have the potential to form secondary structures. Secondary structure formation by the lagging strand template would result in loss of repeats/contractions since repriming 5′ of the structure on the template would be favored, while the formation of such structures on the corresponding Okazaki fragment would favor addition of repeats since repriming would then be more likely to occur more 3′ on the template. Replication from the upstream ORI results in the CGG-rich strand being on the lagging strand template (note that only one side of the bidirectional replication fork is shown). In contrast, replication from the downstream ORI results in the CGG-rich strand being on the Okazaki fragment. In somatic cells replication proceeds equally well from both ORIs resulting in no net gain of repeats. However, in FX ESCs it is suggested that replication predominantly occurs from the downstream ORI resulting in a net gain of repeats. (B) A DNA repair based model for FX repeat instability. During transcription, RNA Polymerase II (Pol II) occludes the template strand, leaving the non-template strand free to form secondary structures. Secondary structure formation may also be facilitated by the co-transcriptional formation of stable RNA:DNA hybrid in the repeat region (Loomis et al., 2014). These structures are then processed via an MSH2-dependent expansion pathway in which CSB plays a supportive role. This pathway is initiated by binding of the MSH2-containing complex to the mismatches in the structures. Events downstream of this binding are not currently known. One possibility is that classical mismatch repair is initiated in response to MSH2 binding but that hairpins form in the flaps generated by strand displacement synthesis as part of this repair. Hairpin formation may prevent normal removal of these bases by flap endonucleases (Spiro et al., 1999) resulting in their incorporation into the “repaired” strand.

FIGURE 2

Models for FX gene silencing. Potential early steps in the initiation of FX gene silencing are depicted. (A) Two ways in which silencing can be triggered by the FX DNA. The left hand panel depicts a model based on the propensity of DNA methyltransferases to methylate FX hairpins in vitro (Smith et al., 1994; Chen et al., 1995). The right hand panel depicts a silencing scheme in which repeat binding proteins act to recruit silencing complexes based on the silencing mechanism that is thought to be responsible for the silencing of the major satellite repeats in pericentric heterochromatin in mice (Bulut-Karslioglu et al., 2012). In the case of the major satellite repeats recruitment of Suv39h leads to the recruitment of DNA methylases and SUV4-20H which trimethylates H4K20. DNMT: DNA methyltransferase; TF: transcription factor; TSS: transcription start site. (B) Two ways in which RNA-mediated gene silencing might occur in the FX locus. The left hand panel depicts an RNA interference based mechanism for gene silencing based on the model of silencing of centromeric tandem repeats in the fission yeast, Schizosaccharomyces pombe (Volpe et al., 2002). In the FX locus the RNA hairpins formed by the FX repeats in the FMR1 transcript may be the source of the double-stranded Dicer (Dcr) substrates (Handa et al., 2003) as illustrated. In this case the annealing of the small Dicer products to the FMR1 mRNA would occur via the same combination of Hoogsteen and Watson–Crick base pairing that generates the RNA hairpins in the first place. Alternatively, the duplex RNAs could be generated by base pairing of FMR1 mRNA with an antisense transcript generated from this region (Ladd et al., 2007). The RNA-induced transcriptional silencing (RITS) complex, which includes the argonaute family member AGO1, then could mediate heterochromatin formation by associating with nascent transcripts via base pairing with the Dcr products. AGO1 could then recruit other epigenetic modifiers including members of the Polycomb (PcG) Group Complexes including EZH2 as suggested by work in human cells (Kim et al., 2006). The right hand panel depicts a way in which an RNA:DNA hybrid may initiate silencing by tethering the FMR1 transcript to the FMR1 locus while also recruiting silencing complexes (SC) that bind to the repeat, perhaps to the secondary structures formed by the repeat, analogous to what has been reported for the RASSF1A locus (Beckedorff et al., 2013) and for ribosomal DNA repeats and IAP elements (Bierhoff et al., 2014). Dcr1: Dicer 1; Pol II: RNA Polymerase II; RITS: RNA interference (RNAi) effector complex; SC: silencing complex; TSS: transcription start site.