Repeat associated non-ATG (RAN) translation: new starts in microsatellite expansion disorders

Abstract

Microsatellite-expansion diseases are a class of neurological and neuromuscular disorders caused by the expansion of short stretches of repetitive DNA (e.g. GGGGCC, CAG, CTG …) within the human genome. Since their discovery 20 years ago, research into how microsatellites expansions cause disease has been examined using the model that these genes are expressed in one direction and that expansion mutations only encode proteins when located in an ATG-initiated open reading frame. The fact that these mutations are often bidirectionally transcribed combined with the recent discovery of repeat associated non-ATG (RAN) translation provides new perspectives on how these expansion mutations are expressed and impact disease. Two expansion transcripts and a set of unexpected RAN proteins must now be considered for both coding and “non-coding” expansion disorders. RAN proteins have been reported in a growing number of diseases, including spinocerebellar ataxia type 8 (SCA8), myotonic dystrophy type 1 (DM1), Fragile-X tremor ataxia syndrome (FXTAS), and C9ORF72 amyotrophic lateral sclerosis (ALS)/frontotemporal dementia (FTD).

Overview of microsatellite expansion disorders

Microsatellite expansion disorders are a growing family of neurological and neuromuscular diseases caused by the expansion of short (3–6 nucleotides) repetitive sequences in the human genome [1]. The position of the expansion mutation, within or outside an ATG-initiated open reading frame (ORF), has provided the framework for research into the molecular consequences of these mutations [2]. For example, research into CAG expansions within ATG-initiated ORFs (e.g. Huntington’s disease (HD) and several ataxias), has focused almost exclusively on understanding the pathogenic effects of the resulting ATG-initiated polyGln proteins [3, 4]. In contrast dominant diseases caused by expansion mutations located outside ATG-initiated ORFs (e.g. myotonic dystrophy) have focused on the toxic effects of expanded RNA transcripts and the dysregulation of RNA binding proteins [5, 6]. Cell culture and animal models to study these diseases have been built with the expectation that expansions in coding regions encode a single mutant protein and non-coding expansions do not encode proteins [7]. While substantial data support polyGln toxicity and RNA gain of function mechanisms [8], recent discoveries that fundamentally change our understanding of how genes are expressed must now be considered. First, a growing number of expansion mutations are known to be bidirectionally transcribed producing expansion RNAs in both directions [2, 9–11]. Second, in 2011, Zu et al., [12] demonstrated that expansion mutations can express proteins in all three reading frames without an AUG initiation codon. This novel type of translation is called repeat associated non-ATG (RAN) translation [12]. These discoveries have uncovered previously unappreciated expansion RNAs and novel sets of disease-specific expansion proteins. Our current understanding of the molecular biology of RAN translation and progress towards understanding its role in disease will be discussed.

The discovery of RAN translation in SCA8

Spinocerebellar ataxia type 8 (SCA8) is a dominantly inherited neurodegenerative disorder caused by an expanded CTG•CAG repeat in the ATXN8 gene [13]. This expansion mutation is bidirectionally transcribed producing both CUG and CAG expansion transcripts and an ATG-initiated polyGln expansion protein (Fig. 1A) [10]. RAN translation was first discovered in SCA8 by Zu et al. [12] when a control experiment to block ATG-initiated ATXN8 polyGln translation did not prevent expression of the protein (Fig. 1B). The discovery that a polyGln protein was produced without an ATG-initiation codon, thought to be required to set the reading frame, raised the possibility that proteins might also be made in the other two frames (e.g. GCA and AGC). Surprisingly, experiments with epitope-tagged constructs demonstrated that expanded repeat tracts produce homopolymeric proteins in all three reading frames, polyGln, polyAla and polySer, without an ATG initiation codon (Fig. 1C) [12]. Because these results were completely unexpected, Zu et al. [12] performed a series of control experiments to detect these proteins and to characterize the RNA transcripts. Characterization of this repeat-associated non-ATG (RAN) translation showed no evidence of RNA editing or frame shifting. Mass spectrometry of the polyAla RAN protein showed a series of N-terminal peptides with varying numbers of alanines with no peptides containing an N-terminal methionine suggesting translational initiation occurs without incorporating an N-terminal methionine initiation codon (Fig. 1D) [12]. Immunofluorescence of transfected cells frequently showed RAN proteins from one or two reading frames, and occasionally all three frames, accumulate in individual cells. Additional studies in HEK293T cells showed: 1) hairpin forming CAG but not non-hairpin forming CAA repeats express polyGln RAN proteins; 2) CUG expansion transcripts also express proteins in three reading frames; 3) longer CAG repeats are associated with simultaneous expression of RAN proteins from all three reading frames; 4) the resulting homopolymeric proteins are toxic [12]. In summary, these results demonstrated for the first time that CAG and CUG expansion mutations can express proteins in three reading frames without the canonical ATG initiation codon thought to be required for translational initiation (see Figure 2 and Table 1 for summary).

Figure 1. The discovery of RAN translation in SCA8.

(A) Bidirectional transcription at the SCA8 locus produces CUG expansion transcripts that form RNA foci and CAG expansion transcripts that produce a short ATG-initiated poly-glutamine expansion protein [10]. (B) Surprisingly, mutating the only ATG initiation-codon upstream of the CAG repeat did not prevent the expression of the poly-glutamine protein [12]. (C) Protein blot showing repeat expansion proteins detected by epitope tags are expressed from all three reading frames (poly-glutamine, poly-alanine and poly-serine) without an ATG-initiation codon. Expression of these repeat-associated non-ATG (RAN) proteins is repeat-length dependent, with simultaneous expression from multiple reading frames observed from longer repeat tracts [12]. (D) Mass-spectrometry of the poly-Alanine protein was performed on cell lysates transfected with a modified epitope-tagged CAG_EXP construct that encoded an arginine interruption within the polyAla protein to allow trypsin digestion. MS and RNA analysis confirmed that polyAla proteins are expressed without an AUG initiation codon and identified a series of peptides that suggest translation initiation may occur in the polyAla frame at sites throughout the repeat tract [12].

Figure 2. One repeat - multiple RNA and protein products.

Schematic diagram showing potentially toxic RNA and protein products expressed from a repeat expansion mutation through a combination of bidirectional transcription, ATG-initiated and repeat associated non-ATG (RAN) translation. In vitro studies predict ATG-initiated and RAN translation can both occur when the repeat is located in an open reading frame (ORF) [12]. While a single ATG-initiated protein is illustrated, multiple ATG-initiated proteins may be produced if there are multiple ORFs. Additionally, RAN translation of the expanded repeat results in the expression of up to six distinct RAN proteins. For example, a CTG•CAG expansion can produce poly-Gln, poly-Ala and poly-Ser RAN proteins from the CAG transcript and poly-Leu, poly-Ala and poly-Cys RAN proteins from the CUG transcript. Each RAN protein, depending upon flanking sequences, may contain distinct C-terminal regions and an ATG-initiated protein in the same reading frame may also have a distinct N-terminal region.

Table 1.

In vitro and in vivo evidence for RAN translation

	Repeat	In Vitro evidence of RAN proteins	In Vivo evidence of RAN proteins	Reference
SCA8	CAG•CTG	GlnSa,b,g, AlaSa,b,c,d,f,g, SerSa,b,g LeuASa, AlaASa, CysASa	AlaSj,m	Zu et al., 2011 [12]
DM1	CAG•CTG	GlnASa,e,f, AlaASa, SerASa	GlnASi,j,l,m	Zu et al., 2011 [12]
FXTAS	CGG•CCG	GlySa, AlaSa,d,g	GlySh,j,m	Todd et al., 2013 [24]
C9ORF72 ALS FTD	G4C2•G2C4	GlyProSe, GlyAlaSe	GlyProS/ASl, GlyAlaSl, GlyArgSl	Mori et al., 2013a [47]
			GlyProS/ASl	Ash et al., 2013 [44]
			GlyProS/ASk	Almeida et al., 2013 [36]
			GlyAlaSl	Mackenzie et al., 2013 [45]
		GlyProS/ASe, ProArgASe	GlyProS/ASl, ProArgASl, ProAlaASl	Gendron et al., 2013 [38]
			GlyProS/ASk	Donnelly et al., 2013 [37]
			GlyArgSl, GlyAlam ProArgASl, ProAlaASl	Mori et al., 2013b [46]
		GlyProSa,f, GlyArgSa,e,f GlyAlaa,f GlyProS/ASa,e, ProArgASa,e,f, ProAlaASa,e,f	GlyProSl,m, GlyArgSl,m, GlyAlam GlyProASl, ProArgASl,m, ProAlaASl,m	Zu et al., 2013 [48]
			GlyArgSl, GlyAlaSl GlyProS/ASl, ProArgASl, ProAlaASl	Mann et al., 2013 [49]

N_S = sense protein

N_AS = antisense protein

N_S/AS = sense or antisense protein (antibodies against the repeat cannot distinquish sense GlyPro from antisense GlyPro)

^aepitope tagged detection of RAN proteins

^btritium labelling of RAN proteins

^cmass spectrometry of repeat

^dmass spectrometry of flanking sequences

^edetection of RAN proteins with antibody against repeat

^fdetection of RAN proteins with antibody against C-terminal region

^gtoxicity of RAN protein

^hdetection of RAN proteins with antibody against C-terminal tag (drosophila)

ⁱdetection of RAN proteins with antibody against repeat (mouse)

^jdetection of RAN proteins with antibody against C-terminal region (mouse)

^kdetection of RAN proteins with antibody against repeat (IPSC)

^ldetection of RAN proteins with antibody against repeat (human tissue)

^mdetection of RAN proteins with antibody against C-terminal region (human tissue)

In vivo evidence for RAN translation in SCA8 and DM1

Zu et al., [12] extended these results by testing the hypothesis that CAG•CTG expansion mutations express RAN proteins in vivo. They developed a peptide antibody that recognizes the C-terminal region of an SCA8 polyAla expansion protein predicted by RAN translation. Immunohistochemistry (IHC) and immunofluorescence (IF) experiments showed that a novel SCA8 polyAla RAN protein accumulates in cerebellar Purkinje cells in both an SCA8 BAC mouse model of the disease and SCA8 human autopsy tissue [12]. The detection of RAN proteins in Purkinje cells, a cell type which shows prominent degeneration in SCA8 patients, is consistent with a role for RAN translation in disease (Table 1).

Zu et al., [12] also provided evidence that RAN translation occurs in myotonic dystrophy type 1. Myotonic dystrophy type 1 is a multisystemic neuromuscular disorder caused by a CTG•CAG expansion mutation in the 3′ UTR of the DMPK gene [14–16]. Research into the pathogenic mechanisms of DM1 has focused on an RNA gain of function mechanism in which expanded CUG transcripts sequester MBNL proteins leading to alternative splicing dysregulation [17–20]. Although there is strong evidence supporting an RNA gain of function mechanism in DM1, it is not yet clear which aspects of the disease are explained by this mechanism. Because DM1 is bidirectionally transcribed [9, 12] and CAG expansion mutations can express proteins without an ATG-initiation codon, Zu et al. [12], tested the hypothesis that RAN translation across DM1-CAG expansion transcripts produces a DM1-polyGln expansion protein. A monoclonal antibodies against the polyGln expansion itself [21] and the predicted C-terminal flanking sequence [12], showed the accumulation of a polyGln RAN protein in DM1 patients and mice [12]. DM1 polyGln-RAN proteins were found at low frequency in patient myoblasts, skeletal muscle and heart and were more common in blood. The co-localization of DM1-polyGln aggregates with caspase-8 [12], an early indicator of polyGln-induced apoptosis [22], is consistent with a role for polyGln toxicity in DM1 [23] (Table 1).

RAN proteins in FXTAS

Todd et al. [24] recently showed evidence that RAN translation contributes to Fragile X-associated tremor ataxia syndrome (FXTAS). FXTAS, a late onset cerebellar disorder characterized by gait incoordination, dementia and tremors [25], is caused by a premutation expansion (50–200 copies) of a CGG•CCG repeat in the 5′ UTR the FMR1 gene [26]. In contrast to Fragile X full mutations (>200 copies) that shut down FMR1 RNA expression [27], premutation expansions result in increased levels of FMR1 CGG_EXP transcripts [28]. While several studies of FXTAS support an RNA gain-of-function mechanism [29, 30], the large ubiquitinated aggregates found in FXTAS patient brains appear more similar to aggregates found in protein-mediated neurological disorders [24].

Using a fly model of FXTAS, Todd et al. [24] noticed the puzzling accumulation of GFP aggregates in flies containing an upstream CGG expansion mutation. This observation suggested the possibility that RAN translation might occur across FXTAS CGG expansion mutations. Todd et al. [24] went on to demonstrate that a polyGly expansion protein is expressed and accumulates in FXTAS fly and mouse models as well as human autopsy tissue. Mass spectrometry detected fragments upstream of the CGG repeat suggesting that translation in the polyGly frame can initiate 5′ of the repeat. This polyGly RAN protein accumulates in neuronal inclusions in the hippocampus, frontal cortex and cerebellum in FXTAS but not control autopsy tissue. Todd et al [24] also demonstrated that 5′ sequence differences between Dutch and NIH FXTAS mouse models affect polyGly RAN protein expression in transfected cells, a result which demonstrates 5′ flanking sequences are important for polyGly expression. Mutations which block polyGly protein expression were used to show polyGly RAN proteins contribute to toxicity in cell culture and fly models independent of RNA gain of function effects. Additionally, these sequence differences explain why ubiquitin-positive, polyGly positive inclusions are found the Dutch but not the NIH mutant mice [31, 32]. This group also showed that a polyAla RAN protein is expressed from a second reading frame in transfected cells [24] but it is not yet clear if polyAla RAN proteins are expressed in vivo. Given that the CGG repeats of FXTAS are bidirectionally transcribed [11], it is possible that antisense RAN proteins may also be expressed. Taken together these results show polyGly RAN proteins accumulate in patient brains and suggest a role for RAN translation in FXTAS (see Table 1 for summary).

RAN translation in C9ORF72 ALS/FTD

C9ORF72 amyotrophic lateral sclerosis (ALS)/frontotemporal degeneration (FTD) is caused by a GGGGCC•GGCCCC repeat expansion in intron 1 of the C9ORF72 gene [33, 34]. The discovery of the C9ORF72 expansion mutation has generated substantial excitement because it connects a large body of research on microsatellite expansion mutations to the most common known cause of ALS and dementia – two diseases with a high impact on society. Several diseases mechanisms have been proposed for C9ORF72 ALS/FTD in which the expansion causes: a) decreased levels of C9ORF72 transcripts and protein [33, 35]; b) RNA gain of function effects [36–43]; c) and most recently, the expression and accumulation of toxic RAN-proteins [36, 38, 44–49].

C9 Sense RAN Proteins

RAN translation of the sense GGGGCC expansion is predicted to result in the expression of three dipeptide proteins: GlyPro (GP), GlyArg (GR) and GlyAla (GA). Support for the accumulation of RAN-proteins in C9ORF72 ALS/FTD autopsy brains was first reported using antibodies against the predicted dipeptide repeat motifs (GP, GR and GA) [44, 47] and more recently using antibodies to both the repeats and unique C-terminal regions [48]. Immunostaining shows evidence that RAN proteins accumulate in neuronal inclusions in the cerebellum, hippocampus and other brain regions of C9ORF72 ALS/FTD but not in control autopsy tissue [44, 47, 48]. The inclusions are similar in shape and abundance to previously characterized p62-positive/phospho-TDP-43 negative ALS/FTD inclusions [44, 47] suggesting that C9-RAN proteins play a key role in the neuropathology of this disease (Table 1).

C9 Antisense Foci

Following previous discoveries of bidirectional transcription in DM1 [9], SCA8 [10] and other expansion disorders [2] several groups have recently showed that the G₄C₂ expansion mutation is also bidirectionally expressed, and that antisense RNA foci accumulate in patient autopsy brains [46, 47], patient derived cell lines [40], and peripheral blood [48]. The Zu et al. study [48] showed C9ORF72 antisense transcript levels are dramatically elevated in C9(+) but not C9(−) brains; but no similar upregulation of the AS transcript was found in blood of ALS/FTD patients [48]. Additionally, double labeling of sense and antisense foci showed that in the majority of cells in the brain [41, 48] and the blood [48] express either sense or antisense foci with only a minority of cells positive for both (Table 1).

C9 Antisense RAN proteins

Additionally, several groups have recently reported that antisense RAN protein aggregates accumulate in C9ORF72-positive autopsy brains [38, 46, 48, 49]. The repeat motifs for these antisense RAN proteins are polyProArg, (PR), polyProAla (PA) and polyGlyPro (GP). Although GP motifs are expressed in both sense and antisense directions, the sense GP protein has a unique C-terminal end not found in the antisense GP protein (GP_AS) [48]. Similarly, it is important to note that five of the six C9-RAN proteins are predicted to contain completely different C-terminal flanking sequences that may also affect their function and pathogenicity [48]. C9 RAN proteins were initially detected using antibodies raised against individual dipeptide repeat motifs [44, 47]. Because these repeat motifs are also found in a number of other proteins, Zu et al [48] developed a panel of antibodies that recognize either individual repeat motifs or the unique C-terminal regions of the sense and antisense C9-RAN proteins (Table 1). Similar to the observation for RNA foci, sense and antisense RAN proteins have been infrequently detected in the same cell [46, 48], although in one study all six proteins were shown to be expressed in the same brain region [48]. Histological studies of affected brain regions showed a striking pattern of clustered RAN protein aggregates that differs within and between patients[48]. Understanding why these differences occur and if they are driven by variability in RNA or triggers of RAN protein expression may provide insight into why patients with expansion mutations are often healthy for decades followed by a rapid decline in health.

Common themes and new directions

RAN in expansion mutations

Historically, protein gain of function (e.g. HD, SCA1, SBMA) and RNA gain of function diseases (e.g. DM1, FXTAS, SCA8) have been considered to have separate molecular mechanisms. In recent years, however, these lines have begun to blur. For example, in Huntington’s disease, nuclear aggregates with expanded CAG HTT mRNA have been observed [4, 50, 51] and SCA3 Drosophila studies demonstrate a toxic RNA component for this polyGln disorder [52]. Additionally four diseases with expansion mutations located in “non-coding” regions have been shown to express proteins in one or more reading frames [53]. While these diseases are characterized by expansion RNAs that accumulate in nuclear foci, the recent observation of cytoplasmic RNA foci C9ORF72 ALS/FTD [37, 40, 42] and the growing list of RAN proteins [12, 24, 38, 44–48] suggest at least some expansion transcripts make it out to the cytoplasm (Fig. 3).

Figure 3. Triple threat - three disease mechanisms for microsatellite expansion disorders.

Expanded microsatellite repeats have been traditionally classified as either coding disorders or non-coding disorders that give rise to protein gain- or loss-of-function or RNA toxicity mechanisms. For traditional “coding” disorders, the repeat expansion is translated as part of a larger open-reading frame (ORF) and results in the expression of a mutant protein that disrupts normal cellular function and induces toxicity. For example Huntington’s disease (HD), a late-onset neurodegenerative disorder, is caused by a CAG expansion within the first exon of huntingtin gene that is translated as a polyglutamine tract in the huntingtin protein, HTT [78]. For traditional “non-coding” disorders (blue), the repeat expansion remains in the RNA transcript, accumulates as RNA foci that sequester RNA-binding proteins and lead to a loss of their normal function. For example, in myotonic dystrophy, CUG(G) expanded RNA transcripts sequester MBNL proteins from their normal splicing targets leading to a MBNL loss-of-function and alternative splicing dysregulation [19, 79–81]. The recent discovery of repeat associated non-ATG (RAN) translation [12] adds a third pathway for disease. RNA transcripts from both “non-coding” and “coding” disorders may undergo RAN translation. Once in the cytoplasm, these transcripts are capable of producing proteins in all three reading frames, which may contribute to cellular toxicity/stress. Depending upon the flanking sequences, each of these RAN proteins will have a distinct expanded peptide repeats (colored boxes) and unique different C-terminal regions (f1, f2 and f3). If the repeat is also within an ATG-initiated open-reading frame, this ATG-initiated protein will share the expanded peptide repeat and C-terminal region with one of the RAN proteins but will have an additional N-terminal region. Further complexity is added by fact that many expansion mutations are bidirectionally transcribed [2], which doubles the number of distinct RAN proteins that may be produced. While individual RAN proteins have been observed in SCA8[12],DM1[12] and FXTAS[24] patients, sense and antisense RNA foci and RAN proteins in all six reading frames been shown to accumulate in C9ORF72 ALS/FTD patient cells [36–38, 44–48].

Conversely, the discovery of RAN translation raises the possibility that RAN proteins are also expressed from expansion mutations located within ATG-initiated open reading frames (e.g. HD, SBMA and SCA1) (Fig. 3). While RAN proteins have not yet been demonstrated for any of these diseases, in vitro experiments show that both ATG-initiated and RAN proteins can be expressed from the same minigene [48]. While disease-causing coding expansions tend to be much smaller than non-coding expansions, the typical size range for coding expansions (40 to 100 repeats) is well within the size range shown to be required for RAN translation. Based on in vitro studies, relatively short CAG expansions (e.g. 40–50) undergo RAN translation in the polyGln frame [12] raising the possibility that ATG-initiated polyGln and RAN polyGln proteins are both expressed in individuals with relatively short expansion mutations. These experiments also predict that individuals with longer expansions associated with more severe phenotypes (50–80 repeats), may express a cocktail of mutant proteins: ATG-initiated polyGln, plus RAN polyGln, RAN polyAla and RAN polySer proteins (Fig. 3). The simultaneous expression of RAN proteins in multiple reading frames from larger expansions in vitro [12], suggests the possibility that RAN translation may play a role in anticipation, or the decreased age of onset and increased severity, seen in individuals with longer repeat expansions.

Do RAN proteins contribute to disease?

To determine the potential contribution of RAN proteins to microsatellite expansion disorders, it will be important to understand multiple aspects of these proteins: (1) Toxicity. RAN proteins are found in disease-relevant tissues (e.g. polyAla in SCA8 Purkinje cells) and several of the RAN proteins have been shown to be toxic in transfected cells [12, 24, 48] and model systems [24] suggesting these protein are toxic in vivo. While the contribution of these proteins to disease remains to be demonstrated, this process is complicated by the need to tease apart the underlying toxicity of the RNA transcripts required to make the RAN proteins and the proteins themselves. (2) Regulation. RAN proteins have been reported in a variety of tissues including brain, muscle and blood [12, 24, 38, 44–48]. There is also considerable variability in RAN protein accumulation within a particular tissue (e.g. RAN protein aggregates cluster at variable sites in C9ORF72 brains [48]). This variability suggests that RAN protein aggregation may be triggered focally and then spread to neighboring cells in a prion-like manner [48, 54–62]. Alternatively, RAN-protein expression may be triggered under conditions of stress [63]. Understanding how, when and where RAN proteins are translated may help explain the variability in age of onset, penetrance and phenotypes of C9ORF72 ALS/FTD and other disorders. (3) Function. Expanded RAN proteins are likely to have aberrant cellular functions, similar to the ATG-initiated polyGln expansion proteins (e.g. mutant Huntingtin and ataxin-1) (Fig. 3). RAN translation can also occur, in vitro, across repeat lengths [12, 24, 48] that occur normally within the human genome, such as within the SCA1 [64, 65] and TATA-binding protein genes [66]. While in vitro observations do not always predict what occurs in vivo, it is possible that RAN proteins may be expressed across relatively short repeats throughout the genome and that these putative proteins have a normal cellular function. Understanding and characterizing RAN proteins will be important for understanding their potential role in disease and normal biology.

Requirements for RAN translation

Sequence

Although there is strong evidence that RAN proteins accumulate in four different microsatellite expansion diseases, how expanded repeats can express proteins in multiple reading frames without an ATG initiation codon is just beginning to be explored. Experiments from Zu et al., [12] show that hairpin-forming CAG but not non-hairpin forming CAA repeats express polyGln RAN proteins in HEK293T cells, suggesting RNA structure is important for RAN translation. Structured RNAs appear to be a common theme as expanded CUG [12], CGG [24], G4C2 [44, 47, 48] and G2C4 [38, 46, 48] expansions, all known to express RAN proteins, form hairpin [67, 68] or G-quadruplex structures [69, 70]. The increased propensity of longer repeat tracts to adopt RNA structures may explain why longer CAG, CGG and G4C2 repeat tracts are typically associated with higher levels of RAN protein accumulation and expression of RAN proteins in multiple reading frames [12, 24, 47, 48].

Translation initiation

Typically, cell free in vitro translation systems are used to understand the biochemical requirements for translation initiation. Information on the molecular requirements for RAN translation is limited because non-ATG translation is less permissive in rabbit reticulocyte lysates (RRL). Expression of repeat proteins across CAG expansions in cell free lysates does not occur in the polyAla frame. Expression in the polyGln and polySer frames is limited and strongly favored by the presence of close cognate initiation codons (e.g. ATT, ATC) upstream of the repeat [12]. Labeling experiments in RRL lysates using S³⁵ show that translation in this cell free system initiates with a Met-tRNAiMet [12]. The incorporation of an N-terminal methionine in RRLs is not surprising because a previously documented alternative initiation codon (ATT) was available. RAN translation of CAG_EXP transcripts in cells is likely to use a different initiation mechanism as: 1) initiation in cells does not require close cognate initiation codons; 2) mass spectrometry of the polyAla showed no evidence for an N-terminal methionine; 3) translation occurs robustly in all three reading frames [12]. Additionally, initiation-start sites appear to be different between reading frames and repeat motifs. For CAG repeats, initiation in the polyAla frame appears to occur at multiple GCA codons throughout the repeat tract, while initiation in the polyGln frame occurs close to or at the beginning of the repeat tract [12]. In FXTAS, MS analysis of the polyGly protein suggests initiation can occur upstream of the CGG repeat tract [24]. Because the polyGly tracts in these experiments were too large for MS analysis, it is possible that initiation of polyGly proteins also occurs within the repeat tract. Little is known regarding the initiation of RAN translation for the G₄C₂ and G₂C₄ repeats, however immunoblots using 5′ and 3′ epitope tags show that translation proceeds through as many as 120 repeats in transfected cells [45]. A better understanding of how and where RAN translation initiates and terminates will be important for determining the molecular mechanisms of disease and for developing future therapeutic strategies to block the expression of RAN proteins.

RAN translation and microsatellite expansion disorder therapies

Recent therapeutic efforts for microsatellite expansion disorders include targeting the RNA for degradation or disrupting interactions with RNA binding proteins [37, 39, 40, 71–76]. While these strategies, which employ antisense oligonucleotides (ASO) or small molecules, have met with a variety of successes [37, 39, 40, 71–76], bidirectional transcription and RAN translation have brought to light new questions. Will knocking down the sense transcript be sufficient to treat these disorders? Will sense transcript knockdown have unintended consequences on the antisense transcript or RAN translation? In SCA7, decreasing the antisense transcript in mutant transgenic mice leads to epigenetic changes and an increase in sense transcript expression [77], suggesting that sense and antisense transcripts regulation may be linked. Two studies using ASOs targeting the sense strand in C9ORF72 ALS/FTD iPS cells rescued several phenotypes including RNA foci and dysregulated gene expression [37, 42]. One study did not observe RAN proteins in their cells [42] and the other study did not detect a reduction in GP RAN proteins [37]. For the latter study, because GP proteins are expressed from both sense and antisense transcripts [48], the ASO knockdown of sense transcripts would not be expected to knockdown antisense GP [38]. A third study saw reduction of sense RNA foci but did not see rescue of gene expression phenotypes in patient fibroblasts after ASO knockdown of G₄C₂ sense transcripts. This group found antisense G₂C₄ RNA foci in these cells and showed they were unaffected by ASO treatment. These authors suggested that the presence of antisense G₂C₄ foci in the treated cells may explain the lack of rescue in these cells [40]. These studies highlight the importance of considering both sense and antisense transcripts as well as ATG- and RAN-proteins in therapeutic treatment strategies for microsatellite expansion diseases.

Conclusions

The discovery of RAN translation has important implications for understanding fundamental mechanisms of gene expression and disease. For microsatellite expansion disorders, bidirectional transcription and RAN translation raises the possibility that a cocktail of mutant transcripts and proteins contributes to many of these diseases (Fig. 2). For example, polyAla, polySer, polyLeu, polyCys and CUG_EXP transcripts may contribute to the pathogenesis of some of the CAG polyGln diseases. Additionally, novel RAN proteins may contribute to diseases currently thought to be caused by RNA gain-of-function effects. Because >50% of the human genome consists of repetitive DNA, RAN translation could reveal an abundant, yet previously unrecognized, category of proteins that may shift current views of proteome complexity and fundamental aspects of cell biology.