Polyglutamine Repeats in Viruses

Abstract

This review explores the presence and functions of polyglutamine (polyQ) in viral proteins. In mammals, mutations in polyQ segments (and CAG repeats at the nucleotide level) have been linked to neural disorders and ataxias. PolyQ regions in normal human proteins have documented functional roles, in transcription factors and, more recently, in regulating autophagy. Despite the high frequency of polyQ repeats in eukaryotic genomes, little attention has been given to the presence or possible role of polyQ sequences in virus genomes. A survey described here revealed that polyQ repeats occur rarely in RNA viruses, suggesting that they have detrimental effects on virus replication at the nucleotide or protein level. However, there have been sporadic reports of polyQ segments in potyviruses and in reptilian nidoviruses (among the largest RNA viruses known). Conserved polyQ segments are found in the regulatory control proteins of many DNA viruses. Variable length polyQ tracts are found in proteins that contribute to transmissibility (cowpox A-type inclusion protein (ATI)) and control of latency (herpes viruses). New longer-read sequencing methods, using original biological samples, should reveal more details on the presence and functional role of polyQ in viruses, as well as the nucleotide regions that encode them. Given the known toxic effects of polyQ repeats, the role of these segments in neurovirulent and tumorigenic viruses should be further explored.

Introduction

Mutations in human proteins that result in longer polyQ repeat sequences have been linked to dementias and ataxias [1]. Their toxicity has been attributed, at the protein level, to aggregation of long polyQ protein tracts, interference with autophagy [2] and to their ability to bind RNA in several model organisms, including marmosets [3], Drosophila, and E. coli [4,5,6,7]. Proteins containing mutated longer repeats may also lose their function. For example, expanded polyQ repeats in ataxin-3 may interfere with miRNA function in Machado-Josef disease [8] and expansion of the polyQ tract in the androgen receptor reduces its DNA binding capacity [9]. The mechanisms of polyQ toxicity are dependent on the protein encoded, and even alternative reading frames of the DNA [10, 11]. For example, aggregated polyQ containing protein, huntingtin, is found in the brain of Huntington’s disease (HD) victims. However, a rare disease similar in clinical appearance to HD, Huntington’s disease-like 2 (HDL2), has been linked to repeat regions in RNA and alternative transcripts causing loss of expression of other proteins, such as junctophilin-3 [12]. Targeting such repeats at the protein or RNA level may provide novel therapies for these diseases [13,14,15].

While the mechanisms for the function and toxicity of extended polyQ segments (or the nucleic regions that encode them) in eukaryotic proteins continue to be actively studied [16], there has been little exploration of their occurrence and possible roles, even in neurovirulent viruses. This is particularly curious, in light of the documented role of polyQ tracts in transcription factors (TFs) and their abundance in eukaryotic genomes [17], even constituting a “polyQ interactome” [16]. The first goal of this work was to determine whether viral proteins contain polyQ repeats at all. One might anticipate that longer polyQ sequences, based on their tendency to aggregate and to interfere with transcription, would be selected against in rapidly replicating viruses under extreme environmental pressure. Sequence selection in DNA viruses during chronic infections, on the other hand, would favor compatibility with host cell transcription and translation mechanisms and immune evasion [18, 19], rather than rapid growth [20,21,22,23]. As this study has found, polyQ segments are indeed rare in the catalogued sequences of smaller RNA viruses, but even very long repeats have been found in several large RNA and DNA viruses.

The second goal is to suggest what functions, if any, such repeat sequences, at the protein or nucleotide level, could play in viral replication, chronic infection, or neuro-pathogenesis. Clues for the potential role of the repeats could be gleaned from their roles in eukaryotic proteins, where they are present in many transcription factors. In addition, Q-rich repeats in the N-terminus of the Argonaute-2 protein of Drosophila and other insects [24, 25] are essential for antiviral activity [26] and one in a cellular protein, TLE2, contributes to this protein’s ability to control lytic reactivation of Kaposi’s sarcoma-associated herpesvirus [27]. As discussed below, the polyQ segments found in several viral proteins could indeed affect the ability of viruses to control the activities or transcription of their own or cellular proteins, while their possible role in neurovirulence remains to be established.

While it may be surprising that polyQ sequences in neurovirulent viruses have not been a major topic of study, it should be emphasized that the extent of very long repeat segments would be difficult to detect by short-read sequencing of the large viruses in which they have been found. In addition, CAG triplet repeats are known to be unstable [28] and may be specifically excised during the transition from latency to active growth, or after adaptation to cell culture. Newer methods, designed to specifically determine repeat sequences in direct isolates from infected tissues, should reveal more details about the presence and roles of repeated sequences.

Section 1: Long PolyQ Segments in Larger RNA and DNA Viruses

Searching for polyQ tracts in viruses

Searching the published sequences of many different virus families revealed that while they are not present in smaller RNA viruses, surprisingly long tracts of polyQ have been found in larger RNA and DNA viruses. The search also suggested that repeats may be much more common in viruses than is indicated by the currently archived sequences.

At the start of this work, the ViPR database [29], which allows rapid access to the published sequences of over 75,000 viral genomes or genome segments, was used to determine which RNA and DNA viruses contain polyQ repeats. A new resource, the Influenza research database [30], was used to screen influenza virus sequences. Once Q-rich sequences were identified, BLAST searches starting from the viral proteins that contained them were used to determine the extent of their conservation in the same virus family and to find other virus proteins containing similar tracts. BLAST was also used to find viral proteins containing repeats similar to those of the Argonaut-2 proteins.

The Vast Majority of Published RNA Virus Genomes Contain No Extended PolyQ Repeats

If long polyQ repeats are intrinsically toxic for the function of proteins, or stimulate aggregation [31], one would expect that rapid evolving RNA viruses would selectively eliminate them. Table 1 summarizes searches of over 20,000 genome sequences of many families of pathogenic viruses, including single-strand RNA viruses (Flavivirus, Reoviruses, Picornaviruses, Bunyaviridae, etc.), 43,000 segments of the dsRNA Reoviridae from the ViPR database and over 100,000 strains of Influenza from the Influenza research database [30]. This revealed that only a few RNA viruses contain even a QQQQ sequence. Longer polyQ sequences, which would be anticipated to cause aggregation of the viral proteins (or, as discussed later, interfere with autophagy), were not found. As long repeats of many other amino acids (especially D, T, L, E, P) and mixed basic or acidic residues occur very frequently, this would suggest that there is some selection against longer polyQ tracts, either at the RNA or protein level. Literature searches have revealed sporadic reports of polyQ segments in some small RNA viruses, including potyviruses [32, 33], and even a coxsackie A24 isolate [34], whereby the lack of consistency among closely related viruses suggests these have no functional role.

Table 1 Maximum length of polyQ repeats (Qn) found in published genomes of mammalian RNA and DNA virus groups. The second column shows the number of genomes searched for each group of viruses, and the last column lists some of the proteins that contain the longer polyQ repeats. See Table 2 for examples of herpes proteins with polyQ repeats and Fig. 1 for longer repeats

More meaningfully, BLAST searches beginning with a Q-rich sequence from DNA viruses (see below) identified a polyQ sequence in the first open reading frame of a nidovirus isolated from a python, representing a novel genus of Torovirus [35]. A similar polyQ sequence is also found in the ORF1 of the Morelia viridis (Boa constrictor) nidovirus, but not in that of a nidovirus isolated from lizards [36]. Nidoviruses (which include the Coronaviridae) have the longest known RNA virus genome, with continuous positive sense strands of 26–32 kBases [37]. In contrast, other +-strand RNA viruses range from 7.5 to 12 kb, and negative-strand RNA viruses have genome lengths ranging from 7 to 19 kb. Bunyavirdae can be up to 22.7 kb in total length, but their longest (L) segments do not exceed 12 kb.

It is possible that these long polyQ insertions may play a role, at the RNA level, during genome replication or adapting to changing environments [38]. As for other RNA virus families [39, 40], several studies have indicated the importance of dsRNA folding domains near the 5′ end of coronavirus genomes [41]. Formation of dsRNA intermediates [42], important for the interferon response [43,44,45,46], as well as viral enzymes that interfere with the OAS/RNaseL system that would target these [47, 48], are characteristic of infection by several different nidoviruses. These include the coronaviruses, MERS and SARS. The repeat RNA sequence encoding the polyQ region in the 5′ region of the python virus may fold as an independent domain containing extended segments of dsRNA (Fig. S1, [49, 50]), whereby the low energy of folding generates a dynamic, unstable structure [51]. The CAG repeat region might be excised completely during rapid growth, as CAG repeats are known to be unstable [28, 52, 53]. Alternatively, it may be transcribed past during generation of subgenomic RNAs, which in nidoviruses proceeds by selective transcription of parts of the open reading frames [54].

PolyQ Repeats in DNA Viruses

Searches within two diverse and well-studied DNA virus families, poxviridae and herpes, indicated that several of these large viruses, known to cause chronic neurotropic infections, contain long polyQ segments (Tables 1, 2 and Fig. 1). These DNA virus genomes are 145–200 kbp, 5–10 times longer than those of the RNA viruses. Many herpes virus proteins contain variable length polyQ repeats in conserved regions (Table 2 shows some examples), and even longer polyQ repeats have also been found (Fig. 1). In addition to direct polyQ repeats, there are long, Q-rich repeats in other viral proteins. For example, there is a long, variable length, Q-rich repeat in the MC006L protein of the pox virus, Molluscum contagiosum [55]. This virus causes the formation of wart-like blisters on the skin of infected individuals, and characteristic cellular inclusions.

Table 2 Examples of herpes proteins containing polyQ repeat segments

Fig. 1

Extensive polyQ repeats and Q-rich (underlined) regions are present in several different herpes and pox virus proteins

As discussed below, the longest repeats were found in DNA virus proteins that function in enhancing transmissibility (cowpox ATI) or contribute to viral latency (herpes viruses).

Section 2: Exploring the Function of Glutamine Repeats in Viral Proteins

The RNA virus results, coupled with the fact that polyQ expansions in human proteins can lead to disease, suggest that polyQ segments are probably selected against in rapidly growing viruses. This leads to the question: what possible functions could they serve for the virus itself or interaction with host cells? This is an important question to answer as the repeats occur in proteins from viruses triggering hard-to-treat neuropathies and epilepsy [56] and isolated from latently infected tissues and tumors.

PolyQ Repeats Serve Important Functions in Mammalian Proteins

Although studied for their role in disease, polyQ segments in mammalian proteins have important regulatory functions. After a polyQ repeat was found to be an activation domain [57] of the TF, SP1, the role of such segments in TFs was extensively studied [58] long before they were documented to have pathogenic characteristics possibly related to folding and formation of aggregates in cells (e.g., [59]). More recently, the length of polyQ inserts was directly related to their ability to enhance TF activity [9, 38]. Variable length Q-rich repeats may also modulate TF activity in eukaryotic cells by modulating their solubility [60] or by recruiting other factors to the DNA binding complex. A polyQ repeat in murine SRY (sex determining region on the Y chromosome) both stabilizes the protein and serves as a transactivation domain [61]. However, the region is found only in rodent SRY and can be replaced by an irrelevant protein (mCherry).

Consistent with a possible role for polyQ tracts in viral proteins in controlling transcription, the first report of a polyQ tract in a DNA virus was in a baculovirus regulatory protein, where the authors noted the similarity of the amino acid repeats to those in SP1 [62]. As Table 2 illustrates, polyQ tracts are present in several regulatory proteins of herpes viruses.

Further evidence for a functional roles in controlling virus replication is that Q-rich tetratricopeptide repeats are upregulated during bovine leukemia virus infection [63], as well as in human breast cancer cells [64]. The Q-rich N-terminal region of a cellular protein, transducing inhibitor of SPLIT (TLE2), contributes to this protein’s ability to control lytic reactivation of Kaposi’s sarcoma-associated herpesvirus [27].

PolyQ Regions in Viral Proteins May Mediate Neurovirulence Through Interference with Autophagy

Recent reports suggest that polyQ segments may also serve to downregulate autophagy, which serves as a barrier to the growth of neurovirulent herpes viruses (whereby RNA viruses may use the membranous structures characteristic of autophagy for their own replication). These examples suggest possible roles for the longer repeats in proteins of viruses that typically cause latent infections, including herpes simplex, Epstein Barr, β- and ɣ-herpes viruses (Fig. 1). Mutations in beclin-1, a protein which triggers the process, were previously linked to development of neurodegenerative diseases [65]. Neurovirulent herpes simplex virus produces a protein that specifically binds to and interferes with beclin-1 function [66], called neurovirulence factor ICP34.5 (or gamma1 34.5, ɣ34.5).

Figure 2 (based on [2, 67]) shows how an expansion of the polyQ repeat in mutant ataxin-3, as well as excess polyQ from other cellular (or viral) proteins, could interfere with the interaction of ataxin-3 and beclin-1 to inhibit autophagy. The polyQ region of wt-ataxin-3, a deubiquitinase, is expanded in spinocerebellar ataxia type 3. The normal length polyQ region mediates binding of ataxin-3 to beclin-1, preventing its degradation and allowing it to stimulate autophagy (Fig. 2, top line). Soluble, mutated polyQ segments can inhibit this binding, thus preventing beclin-1 degradation and upregulation of autophagy, preventing efficient clearance of aging cellular, as well as viral, proteins. Another indication that polyQ sequences from the virus, or some other repeat in ɣ34.5, may also be involved in this regulation is that there is a discontinuity within the (otherwise well conserved) ɣ34.5 sequence in many herpes isolates (see supplementary material). Such discontinuities usually indicate repeat insertions [68].

Fig. 2

Soluble polyQ segments (of cell or viral origin) may prevent beclin-1-induced autophagy, which depends on the DNA binding ability of the polyQ segment of wt-ataxin-3 (based on [2, 67]). Scheme A shows that under normal cell conditions, ataxin-3 binding (mediated by its polyQ region) to beclin-1 (BECN) protects it from proteosomal degradation. This allows beclin-1 to stimulate autophagy, which eliminates both aging cellular proteins and those of viral invaders. Scheme B suggests that viral proteins’ polyQ, similar to the extended polyQ loop of mutant ataxin-3, can interfere with this control by preventing ataxin-3 from binding. Beclin-1 is now degraded and cannot stimulate autophagy, resulting in even more accumulation of polyQ tracts, defective cellular, and viral proteins that will interfere with normal metabolism

This finding ties in with many years of research on the effect of inhibiting autophagy on replication and neurovirulence of various viruses [69]. While neurovirulent viruses such as herpes are indeed held in check by autophagy, some RNA viruses subvert the process for their own replication (e.g., picornaviruses [70], dengue [71]). Although poliovirus requires autophagy for non-lytic spread, its replication is not affected by beclin-1 inhibition [72], suggesting it uses other ways to trigger the process.

Role of PolyQ Regions in Maintaining Latency

As is probably the case with the polyQ repeat in murine SRY, polyQ repeat regions in viral proteins are generally variable in length and may be unstructured or “disordered” [73]. However, some of the examples where the repeats are found suggest they have important functions that would not be obvious during in vitro replication. Once they have infected a cell, viruses enter different growth phases, ranging from almost no replication to rapid growth leading to cell lysis. A herpes virus-infected ganglion may contain less than 1000 copies of the virus/cell in the latent state and still successfully reactivate after stress (from heat, UV light exposure or infection with, for example, a rhinovirus) [74]. Although RNA viruses are generally considered to be “hit and run”, with rapid clearance from the serum, recent experience with Zika [75,76,77] and Ebola [78, 79] viruses has shown that some may also persist within body compartments where they are protected from the immune response.

This leads us to a complicated equation: a virus seeking to survive intracellularly must sacrifice rapid growth for its ability to evade immune detection. Variable polyQ repeats may allow a virus to adjust to changing levels of required cellular factors [80], and determine whether the virus is able to actively replicate, or assume a lysogenic state. Herpes viruses in particular are known to incorporate genes from the cells they infect into their genomes that may aid in maintaining lysogeny. PolyQ insertions at the amino acid or RNA level may directly contribute to viral latency by lowering the transcription or activity of the affected proteins. Alternatively, their presence, or the RNA tracts encoding them, could contribute to neurovirulence by mechanisms demonstrated for human proteins (e.g., huntingtin).

Many herpes virus proteins contain conserved, variable length polyQ segments (Table 2 gives a sampling), including regulatory proteins, an apoptosis inhibitor, and uracil-DNA glycosylase (UDG), all factors that may affect viral replication positively or negatively. It may also be pertinent that a long, Q-rich repeat is present immediately after the catalytic domain of the deoxyuridine 5′-triphosphate nucleotide hydrolase (DUT) gene of the red deer parapox virus (RDPV). Similar sequences have not been reported in the DUT enzymes of other viruses, nor has the role of the polyQ sequence been determined in RDPV. However, UDG and DUT enzymes, which remove or prevent insertion of U residues in viral DNA, are found in all herpes viruses [81]. Their enzymatic activity is essential for neurovirulence, neuroinvasion, and escape from latency of herpes viruses [82]. Mutation of the virus encoded DUT inhibits transcription of equine infectious anemia virus (EILV, a lentivirus and retrovirus). On the other hand, EILV can replicate in non-dividing cells [83] if it allows incorporation of U into its DNA [84].

Accordingly, insertion or amplification of the polyQ segment in UDG or DUT could slow replication to help maintain a latent state. As single point mutations (D71E in the active site, or those preventing phosphorylation of S187 [85]) are sufficient to reduce neurovirulence, DUT may also be a target for antiviral drug design [86]. However, such inhibitors must be very efficient, as residual low levels of the enzyme might have the negative effect of prolonging viral latency (analogous to antibiotic treatment selecting for slow-growing bacterial persister cells [87]).

The long polyQ repeats in other herpes virus proteins (Fig. 1) may also help to suppress virus growth during latency. These include the direct repeats of polyQ that occur in the low complexity C-terminal regions of the Tupaiid T2 protein (β-Herpes group F, isolated from a lymphoma in a tree shrew [88]) and the RF1 protein of Radinovirus type 1 (ɣ-Herpesvirus), isolated from a Kaposi’s sarcoma-like lesion in a macaque [89]. It is possible that these polyQ repeats were directly incorporated from the host cell genes, as their sequences are quite similar to some host proteins (Fig. 3). Further evidence that these polyQ repeats were incorporated in an adventitious fashion from the host cell is that repeats are not found in the published sequences of the (otherwise similar) N1 proteins of Radinoviruses type 2 [91]. Longer repetitive regions could slow growth by decreasing transcription of an essential enzyme, making its RNA more vulnerable to cellular nucleases, and at the protein level, reducing its solubility [92] or enhancing its degradability. Under growth conditions allowing the virus to resume lytic growth, where the enzyme activity is required to ensure efficient replication, the region encoding the polyQ segment could be rapidly removed at the gene level.

Fig. 3

The polyQ region in the Tupaiid T2 protein (herpes virus group F, isolated from a lymphoma in a tree shrew) is flanked by poly-prolines (P), similar to polyQ expansions in huntingtn, and ataxins associated with neurological disease. Proline residues may also affect protein solubility [90]. Two other mammalian proteins that also contain long polyQ repeats are shown for comparison. The T2 repeat is encoded primarily by CAG codons, as is the case with huntingtn, and the nidovirus repeat (Fig. S1)

PolyQ Repeats in Proteins that Mediate Virus Transmissibility

As with the RNA viruses, the published sequences of only a few Poxviridae proteins contain even a tetrad QQQQ repeat. However, there is a conserved polyQ insertion of variable length in the A-type inclusion proteins (ATI) of cowpox (CWPX) viruses (Table 3). As with the Q-rich repeat in the MC006L protein of Molluscum contagiosum (Fig. 1), this repeat is a variable area in an otherwise well-conserved protein (Supplementary material). The ATI with the longest polyQ segment is in strain FM2292, isolated from a lesion in a vole, which causes skin lesions and mild symptoms in its host. Although the length of the polyQ segment in the CWPX strains in Table 3 is not directly related to pathogenicity, ATI plays a role in a more difficult to measure parameter: transmissibility. The ATI protein, together with the p4C protein [93, 94], allows CWPX to form protein inclusions that, when excreted from the animal, protect the virus from the elements. Inclusions that sequester the virus (V⁺ phenotype) contribute to the high transmissibility of CWPX in the wild. As Jennings noted centuries ago, nearly all milkmaids had been infected with CWPX, as were probably most cows. It is significant that in a comparison of three CWPX strains, only the FM2292 virus, which contained the longest polyQ insertion in its ATI, made V⁺ inclusions containing virus particles [95]. Two strains with shorter polyQ segments, the index strain Brighton Red and a similar strain from rat, formed inclusions that contained no internalized virus particles (V⁰). The ATIs of these three strains differ primarily in their polyQ repeat region length (Table 3 and supplementary).

Table 3 Variable length polyQ repeat region in the highly conserved A-type inclusion proteins of cowpox strains. The last three lines show data from Hoffman et al. 2015, where the ability of three strains to form virus containing inclusion bodies (V⁺ phenotype), which aid in transmissibility, was compared

Growth in cell culture alone does not indicate that ATI is an essential gene [96], although it is one of the most abundant CWPX proteins, amounting to as much as half of all protein synthesis in the “late-late stages” of replication [97]. Deleting the ATI gene leads to a faster growing virus [98]. However, as discussed above, ATI enhances transmissibility from animal to animal, as well as virus survival outside the host. The presence of a longer polyQ sequence could reduce its transcription, synthesis, or solubility during restrictive growth in an organism, where ATI’s activity is not required.

Smallpox and vaccinia virus (VV) strains lack polyQ segments in their ATIs and form only virus-free inclusions (V⁰ phenotype). CWPX and VV strains also differ in the ATI interacting protein, p4C, in that only CWPX strains contain long repeats (up to 28) of aspartate (D) residues. These results suggest strongly that this amino acid repeat, together with the polyQ segment in the ATI, aid in sequestering virus particles into the V⁺ inclusions, which further the extracorporeal survival of the virus.

Recent direct, deep sequencing of fresh CWPX isolates from diseased animals indicated diversity in both genome length and coding areas from the Brighton Red reference strain, including an additional 6000 bp ORF [95]. As Table 3 illustrates, freshly isolated strains have the longest polyQ region in the ATI, which makes it difficult to determine a “wild type” length of polyQ. It is, for example, possible that the polyQ repeats in CWPX ATI can be selected against during growth in tissue culture. In keeping with this, there is no polyQ repeat in the (extensively passaged [98]) Brighton Red strain, first isolated in 1937 in England from human lesions. This strain would be expected to transmit poorly in the wild, thanks to its V⁰ phenotype. More recently isolated German strains (1998 and 2002) have shorter polyQ regions than isolates from 1980 or 1990, but it is unknown how often these have been transferred in cell culture [99].

PolyQ Repeats as a Key to Antiviral Therapy

As noted in the introduction to this article, a primary reason for documenting the presence of polyQ segments in viruses is the role polyQ sequences in human proteins have been shown to play in human neurological syndromes [100,101,102]. Considering the importance of glutamine metabolism for central nervous system function, it would be instructive to specifically test the role of the Q-rich regions on virus latency or replication in neuronal cells. Glutamine itself is extremely important in brain chemistry, and inhibitors similar to this amino acid have antiviral activity. A Q analogue, 6-diazo-5-oxo-l-norleucine (DON), can delay encephalitis caused by alphaviruses, such as Sindbis, by reducing the amount of glutamate synthesized from glutamine [103]. Adding polyQ tracts to the antiviral agent zanamivir greatly enhanced its anti-influenza activity [104].

As Fig. 3 shows, the viral proteins that contain long polyQ segments are very similar to those implicated in Huntington’s disease and human ataxias, and may thus be targeted by protein- [16] or gene-based [15, 101, 105,106,107] therapies similar to those now being tested. Going forward, diagnostics should, as much as possible, distinguish polyQ sequences due to a latent virus from those indicating a mutation in a human gene. The flanking regions, which contain proline repeats (PolyP), may also affect the solubility of the proteins [90]. To date, there have been few investigations of a direct role for these polyQ repeats in initiating neural damage. Aiding in establishing a latent infection could, in itself, contribute to neurovirulence, due to the presence of viral products [108].

Conclusions

PolyQ repeats in viruses could play important roles in controlling transcription, latency, transmissibility, and neurovirulence, whereby the latter three aspects of virus pathogenicity are independent of the ability of the virus to grow to high titer in cell culture. Long polyQ tracts in the protein products of neurotropic and cancer-related DNA viruses could chronically disturb their host cells, by mechanisms similar to those identified for huntingtin and other ataxia-related proteins that contain similar repeats.

Just as B cells and other somatic cells may change their genome structure upon differentiation, it is probable that rapidly growing viruses (and those adapted to tissue culture) have different sequences than those in a latent state. Serial cultivation can favor rapid growth and the loss of pathogenic characteristics, an attenuation process used since the first vaccines against Yellow Fever [109] and poliovirus [110]. The instability of repeated CAG regions that encode polyQ repeat sequences might be a mechanism for adapting virus replication to changes in environmental factors [38]. This means that they may be selectively excised during generation of subgenomic RNAs or resumption of active growth after latent periods. Thus, rational reference sequences of viruses should be based on those obtained from direct isolates of diseased tissue or consensus sequences covering many isolates [111,112,113].

As the Brighton Red example illustrates, historical reference strains, many of which have been transferred multiple times in labs across the globe, may have long ago eliminated their unstable polyQ repeat regions. As more direct sequences from infected tissues become available, it is possible that polyQ repeats will be found in many other viral proteins. Several methods have been validated for identifying such long repeat sequences [114], which may be difficult to identify with more traditional methods. For example, sequences up to 20 kb can be generated from a single read using “PacBio” or MinIon technology and related methods. This should allow further determination of the accurate length of repeat regions, and better characterization of their importance for neurovirulent virus infections.