A. Murillo Departamento de Genética del Desarrollo y Fisiología Molecular, Instituto de Biotecnología, Universidad Nacional Autónoma de México, Cuernavaca, Morelos, México
Astroviruses are significant causes of gastroenteritis in both humans and animals, and are also associated to extraintestinal infections, including neurological disease in mammals and hepatitis and nephritis in birds. Herein we summarize our current knowledge on the different stages of the replication cycle of astroviruses, most of which is derived from the characterization of classical human astroviruses (serotypes 1–8) in Caco-2 cells. We also describe the reverse genetic system that has been developed for human astrovirus serotype 1 and the potential it has to improve our understanding of the virus lifecycle, to define virulence determinants, and eventually to develop safe vaccine candidates.
positive-sense RNA virus
Astroviruses are small, nonenveloped viruses with a single-stranded positive-sense RNA genome. Astroviruses are members of the Astroviridae family, which is divided into two genera: Mamastrovirus including viruses infecting mammals and Avastrovirus including viruses that infect avian species. Human astroviruses (HAstV) were originally classified into 8 serotypes, however, genetically diverse astrovirus strains have been recently detected, and the astrovirus taxonomy was modified (Guix et al., 2013). At least four genotype species of astrovirus that infect humans are now recognized: the classical 8 original serotypes are classified now as Mamastrovirus genotype 1, and the novel HAstVs that include the MLB, VA, and HMO virus lineages are classified into three additional genotypes (Guix et al., 2013). The novel HAstVs are more closely related to animal astroviruses. Most of the information described in this chapter, unless otherwise stated, has been obtained from characterization of the classical HAstV-1 and -8 serotypes.
Astroviruses have been long known to be a significant cause of gastroenteritis in children and in the young of many animal species (Méndez and Arias, 2013). The novel HAstVs have been isolated from stools, although a clear correlation with gastroenteritis is not established yet. They have been implicated, however, as agents of encephalitis in immunocompromised patients. In a fatal case of progressive encephalitis an astrovirus belonging to a recently discovered VA/HMO clade was identified in the brain of a 42-year-old man with chronic lymphocytic leukemia (Naccache et al., 2015), and an astrovirus phylogenetically related to ovine, mink, and bat astroviruses was reported to be the etiological agent of encephalitis in a 15-year-old boy with agammaglobulinemia (Quan et al., 2015). In another case report, the astrovirus HAstV-VA1/HMO-C-UK1 was identified in the brain and cerebrospinal fluid of an 18-month-old boy with encephalopathy, indicating that VA1/HMO-C viruses, unlike the classic astrovirus HAstV-1-8 serotypes, can be neuropathic (Brown et al., 2015). In an additional report, a child with severe primary combined immunodeficiency died with a disseminated viral infection, and astroviral RNA was detected in the brain and different other organs, while immunochemistry confirmed infection of gastrointestinal tissues (Wunderli et al., 2011). Similarly, astroviruses have been detected in brain tissue from minks with the neurological shaking syndrome (Blomstrom et al., 2010) and from a young adult crossbreed steer with acute onset of neurologic disease (Li et al., 2013). In avian species, astroviruses produce not only enteritis, but also more diverse pathologies such as hepatitis, poult enteritis mortality syndrome, and nephritis (Méndez and Arias, 2013; Liu et al., 2014).
2. Genome structure and organization
The genomes of astroviruses vary from 6.1 to 6.8 kb in size for mammalian viruses and from 6.9 to 7.7 kb for bird strains, including the 5′ and 3′ untranslated regions (UTRs). The 5′ UTR is short, ranging from 11 to 85 nt, while the 3′ UTR ranges from 80 to 85 in HAstVs and can be longer (130–305 nt) in avian astroviruses (Méndez and Arias, 2013). Regardless of its size, the astrovirus genome is organized into tree open reading frames, named ORF1a, 1b, and 2, is polyadenylated at the 3′ end and at the 5′ is linked covalently to VPg, a viral protein that is essential for viral replication (Fuentes et al., 2012; Méndez et al., 2013) (Fig. 4.2.1). A highly conserved sequence motif that forms a stem-loop structure at the 3′ end of the astroviral genome has been suggested to be relevant for the virus genome replication. Although some strains lack the conserved sequence, a stem-loop in this region is also predicted (De Benedictis et al., 2011). Similar sequence and predicted structures have been observed in other virus families, such as the Caliciviridae, Picornaviridae, and Coronaviridae (Kofstad and Jonassen, 2011).
The 5′-most ORFs 1a and 1b code for two nonstructural polyproteins (nsp’s). Nsp1a includes sequences encoded only in ORF1a, while nsp1ab includes sequences derived from both ORF1a and 1b. Protein nsp1ab is produced by a – 1 translational frameshift mechanism. There is an overlap of 10–148 encoding nt in the genome of mammalian viruses, and between 10 and 45 nt in avian viruses between ORFs 1a and 1b. This overlapping region contains a conserved heptameric sequence (AAAAAAC) followed by a potential stem-loop structure, motifs that are both essential for the frameshift and the synthesis of the viral polymerase encoded in ORF1b (Jiang et al., 1993; Marczinke et al., 1994). ORF2 encodes the precursor of the virus capsid polypeptides. A short predicted ORF X of 91–122 codons, conserved in all HAstV and some other mammalian astroviruses, is present in a + 1 reading frame within ORF 2, although there is no evidence of a protein derived from this ORF (Firth and Atkins, 2010). The capsid polyprotein precursor is synthesized from a subgenomic RNA (sgRNA) of about 2.4 kb in size (Monroe et al., 1993) that might be transcribed by the recognition of a cis-element that acts as a promoter on the antigenomic RNA (agRNA) (Méndez and Arias, 2013).
3. Virus entry
A critical step in virus infection is the interaction of the virus with receptors on the cell surface. Astrovirus receptors have not been described, however, the observation that the susceptibility of different cell lines to infection is dependent on the HAstV serotype suggests that more than one receptor for the virus might exist (Brinker et al., 2000). The crystal structures of HAstV-8 and turkey astrovirus type 2 viral spikes have been recently determined. Their overall structures showed only distant structural similarities, but putative polysaccharide receptor binding sites were described in both capsid spike structures (Dong et al., 2011; DuBois et al., 2013). For cell entry, the infectivity of HAstVs requires to be activated by treatment of the virus with trypsin, which processes the structural polyprotein precursor into the final capsid products and induces a change in the structure of the immature viral VP70 particles (Méndez and Arias, 2013; Dryden et al., 2012).
HAstV-1 and -8 are internalized into the cell by clathrin-dependent endocytosis, and the half-time for release of the genomic RNA (gRNA) into the cytoplasm for HAstV-8 is around 130 min (Donelli et al., 1992; Méndez et al., 2014). Drugs that disrupt endosome acidification and actin filament polymerization, as well as others that reduce the presence of cholesterol in the cell membrane decrease the infectivity of HAstV-8. Furthermore, in cells where the expression of Rab 7 is down regulated, the infectivity of HAstV-8 is also reduced. Altogether, these data support the notion that during cell entry astroviruses arrive to late endosomes, where the viral genome could exit into the cytosol (Méndez et al., 2014). During virus entry cellular signaling pathways are activated; in particular, it has been reported that astrovirus induces the transient phosphorylation of ERK1/2, independently of replication. Inhibitors of ERK affect viral protein and RNA synthesis with the consequent reduction of viral progeny (Moser and Schultz-Cherry, 2008). PI3K was also shown to reduce HAstV-1 infectivity, independently of MAPK, probably during cell entry (Tange et al., 2013).
4. Genome transcription and replication
Upon infection of the host cell, the gRNA is translated into the nonstructural proteins. The viral RdRp uses the gRNA as template to synthesize a full-length negative-sense, antigenomic RNA (agRNA), which in turn serves as template to produce more copies of the full-length gRNA and many copies of an sgRNA (Méndez and Arias, 2013) (Fig. 4.2.1). The agRNA abundance ranges from 0.7% to 4% of that of the gRNA (Jang et al., 2010), while the sgRNA can reach up to 5- to 10-fold higher molar abundance than the full-length gRNA (Monroe et al., 1991). The viral and/or cellular factors that regulate the synthesis of all three species of RNA are not known, however, viruses that differ only in the hypervariable, carboxy-terminal region of nsp1a (see later) display differences in their RNA replication and growth properties (Guix et al., 2005). On the other hand, two predicted stem-loop structures conserved in the 3′ UTR of HAstV serotypes were recently found to bind in vitro to the polypyrimidine tract binding protein. This protein redistributes from nucleus to cytoplasm in HAstV8-infected Caco-2 cells and down regulation of its expression reduces the synthesis of gRNA and virus yield, suggesting that it might have a role in viral RNA replication (Espinosa-Hernandez et al., 2014).
Based on the transcription initiation site determined for the sgRNA in HAstV-1 and HAstV-2, ORF1b and ORF2 overlap in 8 nt; however, the length of this overlapping may vary and is not present in some viruses. The highly conserved sequence of around 40 nt upstream of the ORF2 start codon has been suggested to be part of the promoter for synthesis of the sgRNA. This conserved sequence shows partial identity with the 5′ end of the gRNA, suggesting that it has an important role for the synthesis of both gRNA and sgRNA (Méndez and Arias, 2013).
5. Synthesis of viral proteins
The gRNA of astrovirus is infectious when transfected into cells, implicating that it has the capacity to be translated and to initiate a virus replication cycle. A VPg protein covalently attached to the 5′ end of the viral gRNA seems to be required for the efficient translation and/or replication of the viral RNA (Fuentes et al., 2012; Velázquez-Moctezuma et al., 2012). The gRNA directs the synthesis of the two nonstructural precursor polyproteins, nsp1a and nsp1ab, which are proteolytically processed by viral and cellular proteases (Fig. 4.2.2A) (Méndez and Arias, 2013). ORF1a codes for nsp1a, of between 787 aa for the HAstV sequence to 1240 aa for the avian astroviruses (Méndez et al., 2013). ORF1b is translated by a -1 ribosomal frameshift mechanism as fusion with ORF1a to produce the nsp1ab protein of about 160 kDa (Méndez and Arias, 2013). The efficiency of the signal that modulates the frameshift has been evaluated with reporter genes and varies from 6 to 7% in in vitro translation to 25–28% in cells using de T7-vaccinia expression system (Marczinke et al., 1994; Lewis and Matsui, 1995; 1996). Proteolytic processing of nsp1a is carried out by the viral protease and by cellular proteases, into various nonstructural polypeptides. The most amino-terminal region of nsp1a is processed cotranslationally into a final product of 20 kDa and contains a putative helicase domain (Méndez and Arias, 2013; Méndez et al., 2003), although a virus-encoded helicase activity has not been identified. After this domain there is a predicted coiled-coil structure, a predicted region of transmembrane domains, and a region coding for the viral protease (Méndez and Arias, 2013). This protease is a classic trypsin-like enzyme, and biochemical and crystal structural data confirmed the catalytic triad and the catalytic activity of the protein (Speroni et al., 2009). The mature product has been detected as a 26-kDa protein in HAstV8, a molecular weight that is consistent with processing of nsp1a at Val409 and Glu654 (Méndez et al., 2003; Speroni et al., 2009). After the viral protease, there is a region in nsp1a that encodes a VPg protein of 13–15 kDa (Fuentes et al., 2012). Its functional role is supported by the fact that proteolytic treatment of the gRNA leads to loss of virus infectivity and the replication of the virus is abolished by mutagenesis of the Tyr-693 at the conserved TEEEY-like motif, which has been postulated to be the residue responsible for the covalent linkage to viral RNA (Fuentes et al., 2012). The VPg sequence is followed by an in/del hypervariable region, which has been proposed to be relevant for viral replication (Guix et al., 2005). In addition, some polypeptides derived from the carboxy-terminal region of nsp1a, including the hypervariable region, are phosphorylated and colocalize with the gRNA and the endoplasmic reticulum (Guix et al., 2004b). Finally, processing of the nsp1ab precursor polyprotein yields, in addition to the proteins encoded in ORF1a, the viral RNA-dependent RNA polymerase (RdRp) of 57–59 kDa size (Méndez et al., 2003; Willcocks et al., 1999).
The structural proteins of the virus are translated from the sgRNA as a primary product of 87–90 kDa, named VP90 (Fig. 4.2.2B). The mechanism of translation of the sgRNA is not known, but it could depend on VPg (Fuentes et al., 2012). VP90 contains two domains classified by their sequence identity: a highly conserved amino-terminal domain (residues 1–415) and a highly divergent domain (residues 416 to the carboxy terminus) (Méndez and Arias, 2013; Mendez-Toss et al., 2000). The amino-terminal region contains a region of basic amino acids predicted to interact with the viral genome and a region proposed to form the capsid core (Méndez and Arias, 2013). The region between amino acids 416 and 647 are thought to form the spikes of the virion (Dong et al., 2011; Krishna, 2005), while the carboxy-terminal region of the hypervariable domain contains an acidic region that includes several potential sites for processing by cellular caspases (Méndez and Arias, 2013; Dong et al., 2011). Three proteins of 25, 27, and 34 kDa represent the final trypsin cleavage products present in the capsid of the mature HAstV-8 virion (Méndez and Arias, 2013).
6. Virus replication sites
Positive-strand RNA viruses are known to replicate in the cytoplasm of the host cell in association with membranes. There is evidence suggesting that astrovirus replication is also carried out in association with membranes. Structural (VP90) and nonstructural (protease, RNA polymerase) proteins, as well as genomic and antigenomic RNA and infectious virus particles have been reported to associate with membranes (Méndez et al., 2007; Murillo et al., 2015). Additionally, a protein corresponding to the carboxy-terminus of nsp1a was shown to colocalize in cells with virus gRNA and calnexin, and to interact with the RNA polymerase (Guix et al., 2004b; Fuentes et al., 2011). The architecture of membrane rearrangement in astrovirus infected cells is not yet known, however, ultrastructural analysis by electron microscopy of astrovirus-infected Caco-2 cells showed large groups of viral particles surrounding “O-ring” vesicles, probably corresponding to the double-membrane vesicles reported by Guix et al. (2004b). Structural proteins and the viral RNA polymerase were detected inside these vesicles by immunoelectron microscopy (Guix et al., 2004b; Méndez et al., 2007). Additional structures, similar to those found in HeLa cells infected with poliovirus, have been observed in Caco-2 cells infected with HAstV-8, around which astroviral particles are found (Méndez et al., 2007) (Fig 4.2.3).
As part of their replication cycle, viruses subvert intracellular membranes where viral and host factors cooperatively generate distinct organelle-like structures that serve as platform to form the replication complex of the virus (Paul and Bartenschlager, 2013). Recently, the cellular proteins present in membranes to which astroviral proteins and RNA associate were determined by LC-MS/MS. Functional analysis of the protein–protein interaction network showed some biological processes that were enriched in these membranes, such as gluconeogenesis, fatty acid beta-oxidation, fatty acid synthesis, long chain fatty acid synthesis and tricarboxylic acid cycle (Murillo et al., 2015). These findings are consistent with the fact that modification of the lipid metabolism is emerging as a landmark of the infection of positive-sense-viruses (Paul and Bartenschlager, 2013). Although it is not known if the ubiquitin/proteasome system contributes to cell membrane remodelling, it is also required for astrovirus replication; proteasomal inhibitors and knockdown of ubiquitin expression produce a reduction in viral protein synthesis, and gRNA and viral progeny production (Casorla et al., in preparation).
7. Assembly and exit of viral particles
Astrovirus particles assemble into virus-like particles when the ORF2 is expressed using different expression systems. In LLCMK2 cells, a recombinant vaccinia virus expressing the structural polyprotein of HAstV-2 induced the formation of VLPs with size and morphology similar to native HAstV-2 particles, and these VLPs remained stable during the purification procedure using sucrose gradient ultracentrifugation (Dalton et al., 2003). The ORF2 of HAstV-1 also produced VLPs of around 38 nm when expressed in insect cells using a recombinant baculovirus (Caballero et al., 2004). In both cases, the assembled VLP were antigenically, biochemically, and structurally similar to native astrovirus particles. These features make VLPs an interesting candidate to be tested for the development of an astrovirus vaccine. Of interest, structurally and antigenically correct VLPs were also obtained when a truncated form of HAstV-1 ORF2 lacking the amino-terminal 70 aa was expressed (Caballero et al., 2004).
In astrovirus-infected Caco-2 cells, VP90 assembles intracellularly into viral particles, and is then processed by cellular caspases to yield viruses containing VP70 in a process that might involve at least three intermediate cleavage products of 82, 78, and 75 kDa (Méndez et al., 2004). Astrovirus infection induces apoptosis (Méndez et al., 2004; Guix et al., 2004a), and down regulation of the expression of caspases 3 and 9 blocks the processing of HAstV-8 VP90 to VP70 at the sequence motif TYVD657 (Baños-Lara and Méndez, 2010). The cleavage of VP90–VP70 is associated with release of the virus from the cell through a mechanism that does not involve cell lysis (Méndez et al., 2004; Baños-Lara and Méndez, 2010). All extracellular virions grown in cells in the absence of trypsin are composed by VP70, and to be fully infective these viral particles have to be proteolytically processed to yield the mature virions composed of three capsid proteins (Bass and Qiu, 2000; Méndez et al., 2002).
8. Reverse genetics
The availability of a reverse genetics system, that is, the ability to recover infectious virus from a cDNA copy of the RNA genome, represents an invaluable tool to study the biology of RNA viruses. To this end, Geigenmüller et al. used a full-length cDNA clone of the HAstV1 gRNA to develop a reverse genetics system that was used to analyze specific mutations in the structural (Geigenmuller et al., 1997; Geigenmuller et al., 2002) and nonstructural proteins (Guix et al., 2005) of the virus. In this system, the in vitro transcribed RNA was introduced into the highly transfectable, but poorly permissive BHK-21 cells and the virus produced was propagated in the permissive Caco-2 cell line (Geigenmuller et al., 1997). The disadvantage of this system was the low permissiveness of BHK-21 cells, and the subsequent requirement to propagate the virus in Caco-2 cells to reach acceptable titers. This implied harvesting the virus after several rounds of replication, what might cause the accumulation of mutations in the viral progeny. In fact, viruses with nucleotide changes in the nonstructural region, not related to the original introduced mutations, were found when this method was used to recover viruses (Guix et al., 2005).
To overcome this problem, Velázquez-Moctezuma et al. (2012) developed a reverse genetic system using the highly transfectable hepatoma Huh7.5.1 cells to pack the virus. Although these cells were infected with a 100-fold lower efficiency than Caco-2 cells, the yield of infectious virus in both cells was similar, and in Huh7.5.1 cells the VP90 precursor of the astrovirus capsid polypeptides was found to be efficiently processed. Importantly, virus titers near to 108 infectious units per ml were obtained in Huh7.5.1 at 16–20 h after transfection with total RNA isolated from astrovirus-infected Caco-2 cells. On the other hand, virus titers close to 106 were recovered by using in vitro transcribed RNA from a full-length cDNA HAstV-1 clone; this virus yield was about two log steps higher than that obtained in BHK-21 cells (Geigenmuller et al., 1997). The lower efficiency of virus production in Huh7.5.1 cells when in vitro synthesized RNA was compared to gRNA purified from infected cells, suggested that a factor was missing for the efficient translation or replication of the transfected, synthetic viral RNA (Velázquez-Moctezuma et al., 2012). The 5′-end modification of the viral RNA seemed to determine the specific infectivity of the RNA, since virus recovery was abolished when the total RNA isolated from infected cells was treated with a protease, while it increased when the in vitro transcribed RNA was capped. These observations are consistent with the more recently description of the existence of a VPg protein covalently bound to the 5′-end of the astrovirus gRNA, essential for virus replication (Fuentes et al., 2012). Using this reverse genetic system, amino acids relevant for the function of VPg were recently explored (Fuentes et al., 2012).