The Balancing Act of DNA Repeat Expansions
Total Page:16
File Type:pdf, Size:1020Kb
Available online at www.sciencedirect.com The balancing act of DNA repeat expansions Jane C Kim and Sergei M Mirkin Expansions of microsatellite DNA repeats contribute to the which multiple extra copies can be added during a single inheritance of nearly 30 developmental and neurological intergenerational transmission [5,6] accounting for a pro- disorders. Significant progress has been made in elucidating gressively more severe disease phenotype in a phenom- the molecular mechanisms of repeat expansions using various enon called genetic anticipation. model organisms and mammalian cell culture, and models implicating nearly all DNA transactions such as replication, The molecular mechanisms of repeat expansion are being repair, recombination, and transcription have been proposed. It intensively studied in many experimental model systems is likely that different models of repeat expansions are not (reviewed in [7]). These studies implicate various DNA mutually exclusive and may explain repeat instability for transactions, including replication [8,9], transcription different developmental stages and tissues. This review [10], repair [5,11] and recombination [12] in the process. focuses on the contributions from studies in budding yeast There are substantial differences in the data and their toward unraveling the mechanisms and genetic control of interpretation for different repeats and experimental sys- repeat expansions, highlighting similarities and differences of tems. However, there need not be a single uniform replication models and describing a balancing act hypothesis mechanism that describes the expansion of every micro- to account for apparent discrepancies. satellite sequence. Rather, the mechanisms of repeat Addresse expansion may differ depending on the sequence and Department of Biology, Tufts University, Medford, MA 02155, United length of the repeat, the scale of the expansion, and States whether the cells are dividing or post-mitotic. This review concentrates on recent data obtained in the budding yeast Corresponding author: Mirkin, Sergei M ([email protected]) Saccharomyces cerevisiae, where DNA replication is a major contributor to repeat instability. That said, there are Current Opinion in Genetics & Development 2013, 23:280–288 significant differences in the details and genetic control This review comes from a themed issue on Molecular and genetic of the expansion process for various repeats. Here we bases of disease propose a balancing act hypothesis to account for these Edited by Jim Lupski and Nancy Maizels differences. For a complete overview see the Issue and the Editorial Replication models and the genetic control of Available online 29th May 2013 repeat expansion 0959-437X/$ – see front matter, # 2013 Elsevier Ltd. All rights Almost immediately after trinucleotide repeat expansions reserved. were found to be the causative mutation of diseases such http://dx.doi.org/10.1016/j.gde.2013.04.009 as Huntington’s disease (CAG/CTG), fragile X syndrome (CGG/CCG) and Friedrich’s ataxia (GAA/TTC), inves- tigators began to study the properties of these repeats and Introduction their propensity to expand in various model organisms. Nearly thirty hereditary neurological, neurodegenerative, Because of the ease of creating gene knock-outs and or developmental diseases in humans are caused by introducing DNA cassettes for selectively identifying expansions of microsatellite DNA repeats (reviewed in expansion events, budding yeast has been a particularly [1–3]). These repeats can have anywhere from 3 to 12 useful model system to study the repeat expansion pro- nucleotides in their repetitive unit, albeit the majority of cess. Besides the classical candidate gene approach, gen- expansion diseases are linked to trinucleotide repeats. ome-wide analyses have recently been performed. In Most expandable DNA repeats form unusual secondary sum, some genes appeared to have similar effects on structures, including imperfect hairpins formed by expandable repeats, while other genes had conflicting (CNG)n and (CCTG)n repeats, triplexes formed by results for different repeats (Table 1). (GAA)n repeats, and G-quadruplexes formed by (CGG)n and (C4GC4GCG)n repeats (reviewed in [3]). However, Here we describe the predominant replication models for some repeats that are exceptionally prone to expansions, repeat expansion, emphasizing their relevance with such as the (ATTCT)n repeats causing spinocerebellar respect to various repeat sequences, starting lengths of ataxia 10, do not form intrastrand or multistrand second- the repetitive tract, and the step of the expansion size. We ary structures but instead comprise a DNA unwinding distinguish between small-scale and large-scale expan- element [4 ]. Repeats become unstable when the perfect sions based on the absolute number of added DNA repetitive run surpasses a threshold length varying from repeats, rather than the proportion of added DNA relative 60 to 150 base pairs (bps) depending on the repeat, after to the starting sequence: small-scale expansions refer to Current Opinion in Genetics & Development 2013, 23:280–288 www.sciencedirect.com The balancing act of DNA repeat expansions Kim and Mirkin 281 Table 1 analysis with other mutants involved in lagging strand DNA replication (i.e. DNA ligase I) also exhibited elev- Comparative genetic analysis of expansions of different DNA repeats in budding yeast ated frequencies of CAG/CTG repeat expansion [20 ,24], demonstrating that the effects of Okazaki fragment pro- CAG/CTG CGG/CCG GAA/TTC ATTCT cessing on repeat expansion are not limited to Rad27. 0 5 flap processing 0 Combined, these data support the idea that the 5 flap rad27D " " " model would apply to any starting tract length so long as [18 ,19,20 ,21 ,22 ] [23] [31 ] Replication pausing the sequence was capable of hairpin formation. However, tof1D " " " the upper limit for a single expansion event would be the [38 ] [45 ] [48] number of base pairs contained in the hairpin, which is csm3D " " constrained by the length of the flap (i.e. 20–30 bps for [38 ] [45 ] 0 Helicase activity long flaps [25]). To explain large-scale expansions, 5 flap srs2D " " No effect expansion would need to occur iteratively during replica- [41 ,42 ] [41 ] [45 ] tion cycles. Supporting this possibility, (CAG)70 repeats sgs1D No effect # showed a mean expansion size of 35 repeats in rad27D [41 ,42 ] [45 ] mutants [20 ]. Post-replication repair rad5D " # # 0 [22 ] [45 ] [48] The 5 flap model typically points to hairpin formation as pol30-K164R " No effect the source of expanded DNA sequence. Whereas CAG/ [22 ] [36 ] CTG and CGG/CCG readily form hairpin structures Homologous recombination rad52D No effect No effect #/no effect in vitro [26 ,27], GAA/TTC sequences are not predicted [53 ] [45 ] [48] to self-anneal, though there are reports of hairpin for- " mation [28]. They readily form triplex structures instead [52] [29 ,30 ], which could block Rad27 cleavage and lead to expansion. A RAD27 knock-out was indeed shown to elevate the rate of large-scale expansions of (GAA)100 by the addition of 1–20 repeats whereas large expansions is an order of magnitude [31 ]. Note that Rad27 may also the addition of more than 20 repeats. Because of space play an additional role in replication fork dynamics con- limitations, we are not discussing DNA repair or tran- tributing to GAA expansion via another mechanism (see scription models of repeat expansions, which are dis- ‘Fork reversal/template switching’ below). cussed in depth in [5,11,13]. 0 3 replication slippage 0 5 flap During DNA synthesis, the misalignment of template Because deletion of RAD27 resulted in repeat addition at and daughter strands could lead to small contractions or dinucleotide array tracts as well as other microsatellite expansions. For expansions, polymerase slippage at the 0 sequences [14,15 ], it was proposed that repeats might 3 end of the nascent strand could result in the formation 0 form a stable hairpin structure within a 5 flap, which of a hairpin, and the looped-out DNA would then con- could lead to enhanced expansion in the absence of tribute to an expansion in the next round of replication 0 RAD27 [16 ]. In the 5 flap model, ligation of this hairpin (Figure 1b) [32 ]. Although slippage could occur on to the next Okazaki fragment could then lead to a small either the leading or lagging strand, the complementary 0 expansion (Figure 1a). Additionally, 5 DNA flaps can strands of expandable repeat sequences have different similarly be formed and processed by Rad27/Fen1 during thermodynamics of structure formation and thus may long-patch base excision repair [17]. dictate which nascent strand preferentially undergoes 0 slippage. As with the 5 flap model, expansion would be Indeed, RAD27/FEN1 has been found to affect repeat independent of starting tract length as long as the expansion across a range of repeat sequences and starting sequence was capable of hairpin formation but should lengths. The frequencies of CAG/CTG repeat expan- likewise result in small-scale expansions for thermodyn- sions were elevated in the absence of Rad27 whether amic reasons [33,34]. The model would apply best to CAG was on the leading or lagging strand template and repeats that are capable of hairpin formation. for a variety of starting lengths [18 ,19,20 ,21 ]. With the 0 development of a system in which the addition of as few In agreement with 3 replication slippage as a model for as five repeats to a starting length of (CTG)13 could be repeat