RNA STRUCTURE DUPLICATIONS AND FLAVIVIRUS HOST ADAPTATION

Flaviviruses

The Flavivirus genus includes a large number of taxonomically recognized species, many of which are important human pathogens such as dengue, yellow fever, Japanese encephalitis, West Nile and other viruses that cause fever and encephalitis. Dengue virus (DENV) is the most important viral disease in humans transmitted by insects. It is responsible for about 390 million infections each year, without vaccines or antivirals available for its control. Yellow fever virus (YFV) is endemic in a number of African and South American countries, and causes 200,000 cases and 30,000 deaths in Africa even with effective vaccines available (http://www.who.int/csr/disease/yellowfev/YellowFeverBurdenEstimation_Summary2013.pdf). Other diseases caused by flaviviruses include West Nile encephalitis and Zika fever, which are considered emerging diseases with important outbreaks around the world [1].

Despite the similar organization of flavivirus genomes and their mechanisms of replication, they possess differences in their host ranges and transmissibilities. In this regard, flaviviruses are divided into four large ecological groups: the mosquito borne group (MBFV), the tick borne group (TBFV), the vertebrate specific flavivirus group, referred to as no known vector viruses (NKFV), and the ones that have been only isolated from insects, which constitutes a growing group of viruses known as insect specific flaviviruses (ISFV) (for a recent review see [2]) (Figure 1A).

Flaviviruses are small, enveloped viruses with a single, positive-strand RNA genome of 10 to 12 kb. A type I cap structure is present at the 5’ end followed by the conserved dinucleotide 5’-AG-3’ [3]. The cap structure of flaviviruses contains a methyl group at the N7 position, and a second methyl group at the ribose 2’OH position of the first nucleotide, m7GpppAmpN2 [4, 5]. The 3’ end of the genome lacks a polyadenylate tail and terminates in a conserved 5’-CU-3’ [6]. The genome encodes a single open reading frame flanked by highly structured 5’ and 3’ untranslated regions (UTR). The 5’ UTRs are about 100 nucleotides long while the 3’ UTRs are in general between 400 and 700 nucleotides, while in some exceptional cases it can be over 900 nucleotides [2, 7, 8]. Although different flavivirus groups contain conserved RNA structures in the 5’ and 3’ UTRs, only two RNA elements are conserved in all flavivirus genomes. These are the Y shape stem-loop A structure (SLA) present at the 5’ end of the viral genome (Figure 1B), and the small hairpin 3’ stem-loop (sHP-3’ SL) located at the 3’ end of the viral RNA. These two essential RNA structures participate in the basic mechanism of viral RNA synthesis (Figure 1C). Sequence and structural features of the SLA as the promoter for viral polymerase binding and activation were first described in DENV and then extrapolated to other flaviviruses [9–11, 12]. The 3’ SL was the first RNA structure described in flaviviruses and it was originally observed in the genome of MBFVs [13–15]. Sequence conservation analysis within each flavivirus group indicates that both the SLA and the 3’ SL are the most conserved regions of the viral genomes (Figure 2). In addition, an important conserved feature in the genome of all flaviviruses is the presence of inverted complementary sequences that mediate genome cyclization, which allows the polymerase, bound to the promoter SLA, to reach the 3’ SL initiation site for RNA synthesis (for review see [16]). In this review, we will discuss the function of viral RNA elements that specialize for flavivirus replication in insect and mammalian hosts.

Specialization of RNA Structures as a Strategy for Viral Adaptation to Mosquito or Human Cells

MBFV are a large group of viruses that cycle between Aedes or Culex mosquitoes and vertebrates. The process of jumping between such different hosts requires adaptation to different cellular machineries for viral replication and evasion of different types of antiviral responses. In general, RNA viruses have high capacity to adapt to different environments due to the genetic diversity of viral populations [17, 18]. However, it has been proposed that viruses that naturally alternate between different hosts evolve less rapidly than those that specialize in a single host [19]. This evolutionary constraint can be attributed to conflicting demands for efficient infection in different hosts [20].

A number of studies have investigated genetic variations that take place during flavivirus adaptation to vertebrate and invertebrate hosts (reviewed in [21, 22] and [23]). However, little is known about variations in functional RNA structures that act as cis-acting factors during flavivirus host adaptation. Several reports have documented mutations in flavivirus 3’ UTRs with differential impact on viral replication in insect or mammalian hosts [24–29]. For DENV a host specific requirement of the sHP present at the viral 3’ UTR has been reported [30]. In this regard, point mutations in the sHP abrogated infection in mosquito cells without affecting replication in mammalian cells. In a recent study, DENV restriction to replicate in vertebrate or invertebrate cells resulted in the selection of different viral populations with a hot spot for variations in the viral 3’ UTR [31]. Deep sequencing of viral populations revealed that mutations selected in mosquito cells mapped in a single RNA structure, named stem loop II (SLII, Figure 2). Selected mutations disrupting this RNA structure were associated with higher viral fitness in mosquito cells and in Aedes albopictus mosquitoes as compared with viruses with an intact SLII. In contrast, mutations that disrupted this RNA structure reduced viral replication in human cells [31]. Interestingly, cycles of disruption and reconstitution of the SLII structure were observed after host switch. These observations provide an example of a viral RNA structure that is under opposite selective pressures in mosquito and human cells and highlight a process of host specialization of a cis-acting RNA element.

The DENV2 3’ UTR is structurally divided into three regions: region I is located just downstream of the translation stop codon and contains a hypervariable sequence followed by two SLs similar in sequence and structure (SLI and SLII), with sequences involved in pseudoknot interactions (PKI and PKII, respectively); region II contains a duplication of a conserved structure known as dumbbell (DBI and DBII) also involved in pseudoknot interactions (PKIII and PKIV, respectively); and region III includes the essential terminal structure sHP-3’ SL (Figure 2). The functional significance of complex RNA structure duplication in the viral 3’ UTR is still unknown. Regarding the host-dependent variation of DENV SLII, it is important to mention that the sequence and structure of SLI remains intact during viral replication in mosquito or human cells, supporting the idea of a functional diversification of the duplicated SL elements.

The available data provides evidence of distinct functions of flavivirus RNA structures during infection in vertebrate and invertebrate hosts, resulting in different adaptation processes.

Functional Significance of Flavivirus RNA Structure Duplication

The presence of two SLs and two DBs is a conserved feature in most MBFV 3’ UTRs. Sequence similarities and defined structural blocks in the SLs and DBs support duplication as the origin of these RNA structures; however, recent functional studies provided evidence that each SL and each DB have distinct roles during viral replication [31–35]. In the case of SLI and SLII, studies using DENV2 indicate that while deletion or mutations of SLII provide a great replication advantage in mosquito cells, deletion of SLI marginally reduces viral replication, supporting different functions of the two structures. In contrast, in human cells, the two SLs appeared to play redundant functions. In this regard, while deletion of either structure slightly reduces viral replication, deletion of both results in a large decrease in viral fitness [31]. Regarding the DB elements, sequences involved in DB1 and DB2 formation were reported to play opposite roles in flavivirus genome cyclization. While DB1 bears nucleotides that are complementary to sequences present at the 5’ end of the genome that enhance long range RNA-RNA interactions and genome cyclization [34, 36], the sequence at the top-loop of DB2 competes with genome cyclization by forming a PK with nucleotides that overlap with 3’CS, one of the most conserved cyclization element among flaviviruses [32–34]. These observations support a diversification of function between the two SLs and between the two DBs.

Interestingly, RNA sequence alignments and RNA structure comparison using RNAforester, which estimates RNA structural similarities based on the tree alignment model [37], shows that SLI from different DENVs are more closely related than SLI and SLII from the same virus (Figure 3). This observation was extended to all members of the Japanese encephalitis virus (JEV) group, supporting the idea that SL structure duplication occurred early during emergence of these viral groups, and then each of the duplicated element evolved, acquiring specific functions. To avoid confusion with the nomenclature of RNA structures from different groups, in Figure 3, the related SLI and SLII are also referred as SL5’ and SL3’, respectively (based on their location).

In the case of YFV, which naturally bears a single SL and likely has less capacity to tolerate mutations in this region, an interesting correlation between viral virulence and the integrity of the single SL was found [38]. In this case, sequence and structure comparison of wild and vaccine strains of YFV (performed to define determinants of viral attenuation) revealed that all vaccine strains, except 17DD, displayed a disrupted SL and PK structures (originally referred to as Region I). In the wild type strains, the SL structure was maintained, displaying several covariations. Interestingly, 17DD was the only vaccine strain that retained the SL-PK structure (also by co-variations) and was the most virulent of the vaccine strains in a mouse model [39]. This pioneering analysis of 3’ UTR structures provided an interesting correlation between SL-PK integrity and YFV pathogenesis.

More recently using DENV, the presence of a double SL-PK in the viral 3’ UTR was found to facilitate host switch from mosquito to humans [31]. In this regard, recombinant viruses with a single SL carrying mosquito adaptive mutations failed to replicate in human cells, while this was reverted by incorporation of a second SL. Based on these observations, it is possible that SL duplication serves as a mechanism for higher tolerance of mutations that are beneficial in one host but deleterious in the other, by maintaining one of the two structures intact during host switch.

In summary, experimental evidence supports the significance of RNA structure duplication in members of the MBFV group. Instead of a redundant function, as it was originally proposed, each of the duplicated RNA elements appear to play distinct roles during viral replication [31–34], and in some cases are directly linked to specific requirements in different host.

RNA Structure Duplication, Viral Fitness and sfRNA Production

During flavivirus infection, in addition to the full-length genomic viral RNA, other smaller RNA species accumulate, named subgenomic flavivirus RNAs (sfRNAs) [40–42]. Several recent studies provide interesting ideas on how the sfRNAs enhance flavivirus infections [42–48]. For a comprehensive description of the sfRNA functions, the reader is referred to recent reviews [49, 50].

The sfRNAs are products of partial degradation of viral RNAs. After removal of the cap structure present at the 5’ end of flavivirus genomes, the RNA is degraded by the host exoribonuclease XRN1 from the 5’ to the 3’ direction [42]. The biogenesis of sfRNAs is linked to highly structured RNA elements (xrRNAs) present at the viral 3’ UTR, which stall the XRN1 exonuclease activity [51–53]. The process results in the accumulation of small RNA molecules with nucleotide sequences corresponding to viral 3’ UTRs. In the case of JEV group members, the first of the duplicated SLs appears to be the main structure responsible for XRN1 stalling and sfRNA accumulation [42, 54]. In YFV, the single SL-PK (also known as SL-E) has been proposed to be the structure responsible for sfRNA formation [55]. For members of the DENV group, it is still uncertain which one of the two duplicated SL-PK structures is the main determinant for XRN1 arrest for sfRNA formation in vivo.

Interestingly, infection with flaviviruses carrying mutations that artificially disrupt SL-PK structures showed that RNA degradation by the XRN1 advances towards the 3’ end of the genome, but then stalls at the DB-PK structures, generating shorter sfRNA species [54]. This observation indicates that there are at least two types of structures in the MBFV 3’ UTRs capable of stalling the XRN1 activity (the SL-PK and the DB-PK). Based on the mechanism of sfRNA formation, it is intriguing the conserved duplication of RNA structures responsible for XRN1 arrest. Moreover, the process of DENV adaptation to mosquito cells specifically disrupts a SL-PK structure, which likely results in accumulation of different types of sfRNAs. These observations lead us to speculate about a possible link between host adaptation and sfRNA production (Figure 4, Key Figure). In addition, a recent study using different DENV isolates, proposed an association between the levels of sfRNA accumulation and the viral epidemiological fitness [48]. Based on the available data, it is possible that the variability observed at the 3’ UTR of different DENV isolates is a consequence of specific selective pressures in the mosquito.

Conserved RNA Structure Duplications in the 3’ UTRs of Flavivirus Genomes

To understand the biological significance of RNA structure duplication in the flavivirus genomes, the conservation of structural blocks of RNA elements found in each group of the Flavivirus genera was constructed and compared (Figure 5 and 6). Models of secondary structures of flavivirus 3’ UTRs including MBFVs, TBFVs, NKFVs, and ISFVs were elaborated using RNAalifold, RNAz and CentroidFold software [56–58]. When available, information from chemical or enzymatic probing was included in the predictions. The models described here also gather information from previous studies that identified a number of structural features shared by different flavivirus 3’ UTRs [35, 59–66].

Concluding Remarks

Models of common structural RNA elements present at the 3’ UTR of all flaviviruses have been revised to facilitate functional studies and as tools for comparisons between different viral groups. There is an evident association between flaviviruses that alternate between insects and mammalian hosts and conserved complex RNA structure duplications. Interestingly, the identification of opposite selective pressures on RNA structures in mosquito and human cells provide new ideas for understanding the need of duplicated elements.

A great deal of experimental information has been generated in the past few years that had led to an association between host specialization of RNA structures, sfRNA production, and viral fitness. These studies will be instrumental in generating new models that explain evolutionary mechanisms that enable flaviviruses to replicate in multiple hosts (see Outstanding Questions).

Understanding how host-virus interactions shape viral evolution will help to elucidate the factors that govern the emergence of new viruses and the expansion of already known RNA viral pathogens.