ABSTRACT
Human astroviruses are single-stranded RNA enteric viruses that cause a spectrum of disease ranging from asymptomatic infection to systemic extragastrointestinal spread; however, they are among the least-characterized enteric viruses, and there is a lack of a well-characterized small animal model. Finding that immunocompromised mice were resistant to human astrovirus infection via multiple routes of inoculation, our studies aimed to determine whether murine astrovirus (MuAstV) could be used to model human astrovirus disease. We experimentally infected wild-type mice with MuAstV isolated from immunocompromised mice and found that the virus was detected throughout the gastrointestinal tract, including the stomach, but was not associated with diarrhea. The virus was also detected in the lung. Although virus levels were higher in recently weaned mice, the levels were similar in male and female adult mice. Using two distinct viruses isolated from different immunocompromised mouse strains, we observed virus strain-specific differences in the duration of infection (3 versus 10 weeks) in wild-type mice, indicating that the within-host immune pressure from donor mice shaped the virus kinetics in immunocompetent recipient hosts. Both virus strains elicited minimal pathology and a lack of sustained immunity. In summary, MuAstV represents a useful model for studying asymptomatic human infection and gaining insight into the astrovirus pathogenesis and immunity.
IMPORTANCE Astroviruses are widespread in both birds and mammals; however, little is known about the pathogenesis and the immune response to the virus due to the lack of a well-characterized small-animal model. Here we describe two distinct strains of murine astrovirus that cause infections in immunocompetent mice that mirror aspects of asymptomatic human infections, including minimal pathology and short-lived immunity. However, we noted that the duration of infection differed greatly between the strains, highlighting an important facet of these viruses that was not previously appreciated. The ubiquitous nature and diversity of murine astroviruses coupled with the continuous likelihood of reinfection raise the possibility of viral interference with other mouse models of disease.
INTRODUCTION
Astroviruses are small single-stranded RNA viruses that cause a spectrum of disease in many avian and mammalian species; however, they are among the least-characterized enteric viruses (1). Indeed, human astroviruses (HAstVs) are both underdetected and overlooked as a leading cause of pediatric diarrhea worldwide (2). This is in part because current screening methods are unable to capture the full diversity of HAstVs (1, 3), which include four distinct genotype species: Mamastrovirus 1 (HAstV1 to -8), Mamastrovirus 6 (MLB1 to -3), Mamastrovirus 8 (VA2 to -5), and Mamastrovirus 9 (VA1 to -3). Although most astrovirus-induced diarrheal disease is self-limiting, extragastrointestinal (extra-GI) diseases associated with Mamastrovirus 6, 8, and 9 strains have been reported (4). Critically, we have identified diverse astroviruses in pediatric oncology patients, some of whom had sequential infections and prolonged virus shedding (up to 183 days), which could alter the evolution and epidemiology of the viruses (5). Even though astroviruses constitute a major public health concern, there are no treatments or vaccines available (6), and there is only limited knowledge regarding the mechanisms by which these viruses cause disease (1).
Unlike other enteric RNA viruses, astroviruses do not cause cell death or overt pathology (7–9). Previous studies using a turkey poult model with turkey astrovirus 2 have shown that diarrhea is associated with increased intestinal permeability and the redistribution of sodium transporters (8, 10, 11), but the exact mechanism is unclear. One limitation of these studies is that astrovirus members that infect mammals cluster in the Mamastrovirus genus, whereas turkey astrovirus 2 is a member of the Avastrovirus genus, which includes viruses that infect birds (1). A recent study of bats demonstrated an association between dysbiosis of the gut microbiota and astrovirus infection (12), suggesting another possible mechanism for diarrhea (13, 14). However, to explore these various mechanisms of disease, it is necessary to first develop a small animal model.
Murine astroviruses (MuAstVs) were discovered to be endemic in animals sampled from multiple vendors and research facilities. Only two strains of MuAstV have been fully sequenced, although numerous partial genomes have been generated from viruses identified from both wild and laboratory mice, indicating that there is considerable diversity among these viruses (15–19). However, it is unclear whether there are strain-specific differences in virus tropism, persistence, or pathogenesis, as have been well characterized for murine norovirus (20–23). Because molecular testing for MuAstV is not routinely performed, little is understood about the duration of infection, the viral transmission, and the immune response to the virus. However, some initial observations have been made in three independent studies using different genetic backgrounds and virus strains. The first of these studies showed that the adaptive immune response was required to control MuAstV infection and that cohousing was a means of transmission (15). We next showed that the type I interferon response is also critical for virus clearance and for the first time defined the chronic nature of infection with MuAstV in wild-type mice, which was not fully cleared until 40 days postinfection (dpi) (16). Finally, a study similarly described a chronic MuAstV infection in neonatal mice that was not associated with histopathologic changes or diarrhea (24). However, because most astrovirus infections in humans occur during early childhood (25–27), which coincides with substantial changes to the gut microbiome after the introduction to solid foods (28, 29), it is unclear whether there are age-dependent differences in MuAstV pathogenesis or whether other factors, such as the sex of the host, can alter the replication dynamics of the virus.
In these studies, we explored the use of MuAstV to understand the spectrum of disease that occurs in humans. We found that the mouse model mimics aspects of human infections and, importantly, strain-specific differences in the duration of infection that are likely shaped by within-host immune pressure.
RESULTS
Developing and characterizing an astrovirus mouse model.To define the replication kinetics and spread of MuAstV infection in recently weaned 3-week-old mice, we orally gavaged male and female wild-type C57BL/6 mice with MuAstV-infected CD1 nude mouse fecal filtrate (0.22 μm) and monitored the virus levels in the intestinal tract (Fig. 1A) and in extraintestinal organs, including the stomach, lungs, and blood (Fig. 1B). Virus levels peaked between 8 and 13 dpi in the intestines and stomach before beginning to drop after 17 dpi. The lungs also showed evidence of virus replication that mirrored this 3-week time course, although virus detection was slightly delayed and less robust, with virus being first detected at 8 dpi and average peak levels reaching only 1.54 × 105 genome copies. As we observed no viremia, the virus in the lungs may have originated from aspiration of the virus during inoculation and not from extra-GI spread. Unfortunately, attempts to prove productive replication using currently available tools were unsuccessful. Finally, we detected a constant low level of virus in the spleens beginning between 8 and 10 dpi, which could indicate trafficking of the virus by antigen-presenting cells.
Murine astrovirus replication kinetics in male and female wild-type C57BL/6 mice. Three-week-old mice (n = 4/time point) were orally gavaged with murine astrovirus-positive fecal filtrates from CD1 nude mice. Virus detection was assessed for 3 weeks in the gastrointestinal tract (A) and at extra-GI sites (B) at the indicated time points. The dotted lines indicate the lower limit of detection for the qRT-PCR assay. The data are from two independent experiments.
Next, to determine whether age or sex differences were mediating factors in our model, as has previously been shown to be the case for other enteric viruses (30–32), we compared the virus levels in the GI tracts of adult (8-week-old) and young (3-week-old) mice, as well as in male and female adult mice. Although we observed no difference in the virus levels in young and adult mice at the start of peak infection (8 dpi), we noted significant differences at 14 dpi, indicating that the peak infection in young mice lasts longer than that in adult mice (Fig. 2A). In contrast, male and female adult mice showed no differences in virus levels throughout their GI tracts at 8 or 14 dpi (Fig. 2B).
Murine astrovirus replication is not dependent on age or sex, but inoculation with intestinal filtrate results in more replication than is seen with fecal filtrate. (A and B) Mice were orally gavaged with murine astrovirus-positive fecal filtrates from CD1 nude mice, and tissues from the gastrointestinal tract were collected at 8 and 14 days postinfection (n = 4/group) to enable comparisons of virus levels in young (3-week-old; purple) and adult (8-week-old; gray) mice (A), as well as in male and female adult mice (B). (C) Comparison of virus replication in adult mice (n = 8/group) inoculated with equivalent numbers of genome copies (2.7 × 107) of murine astrovirus derived from CD1 nude mouse intestinal, fecal, or sucrose-purified fecal filtrate. Statistically significant differences were determined by the Mann-Whitney U test with a cutoff P value of <0.05 (*). The dotted lines indicate the lower limit of detection for the qRT-PCR assay. The data shown are from two independent experiments. ns, nonsignificant.
To examine whether different preparations of the inoculum would alter the replication of the virus, we compared the results obtained with intestinal and fecal filtrates to those obtained with a sucrose-purified viral preparation. When we inoculated adult mice with equivalent amounts of virus normalized by genome copies, we observed a complete loss of infectivity with the sucrose-purified virus (Fig. 2C). This loss of infectivity suggested that there had been a loss of infectious particles or alterations in the virus stability, as has been observed with both enveloped and nonenveloped viruses (33, 34). We also noted higher levels of virus in mice inoculated with the intestinal filtrate than in mice receiving the fecal preparation (Fig. 2C). In all groups, we observed no diarrhea or fecal inconsistency during experimental MuAstV infections, as reported previously (24).
Given the lack of symptomatic disease with MuAstV, we next asked whether we could use human astrovirus to infect mice. Our pilot experiment used 10- to 16-week-old Ifnar−/− mice inoculated with human astrovirus 1 (HAstV1) or with human fecal filtrates containing the human astrovirus MLB1 and VA2 strains. Oral inoculation of mice with 1 × 107 infectious particles of HAstV1 or with MLB1 and VA2 fecal filtrates, followed by cohousing with recipient animals, resulted in no virus replication that was detectable by reverse transcription-quantitative PCR (qRT-PCR) in the donor mice or the recipient mice. In view of the recent evidence that the MLB and VA strains are associated with encephalitis and respiratory infections in humans (35, 36), we also performed intracisternal magnal and intranasal inoculations; however, inoculation via either of these routes resulted in no productive replication in the 8- to 10-week-old animals. Finally, we administered 1 × 106 infectious particles of HAstV1 to 8-week-old mice via intraperitoneal injection. The virus was subsequently detected in the spleens of these animals over the course of 21 days, but the levels did not increase and were undetectable by day 42 (Fig. 3A), indicating a lack of infection but a successful inoculation and immune-cell trafficking of the virus for presentation as evidenced by veterinary pathology review indicating germinal center formation. To confirm this finding, we also inoculated 8-week-old wild-type C57BL/6 mice and found similar trafficking of the virus to the spleen at 3 and 14 dpi (Fig. 3B). Thus, despite the current dogma that suggests astroviruses are capable of crossing species barriers (37), we found that the strains of mice we assessed were resistant to human astrovirus infection.
Virus detection in the spleens of male and female Ifnar−/− and wild-type C57BL/6 mice intraperitoneally inoculated with human astrovirus. Spleens were collected at the indicated time points from 8-week-old Ifnar−/− (A) and wild-type (B) mice (n = 3 to 6/time point), and virus was detected by qRT-PCR. The dotted lines indicate the lower limit of detection for the qRT-PCR assay. The data shown are from two independent experiments.
Duration of MuAstV shedding in wild-type C57BL/6 mice is virus strain specific.Although we initially focused our efforts on characterizing the MuAstV from CD1 nude mice, which have the same genetic background as the mice in which the first MuAstV was discovered in 1985 (38), we subsequently broadened the scope of our study by screening other immunodeficient mice in our colony. Over a 5-week period, we found that 4- to 12-week-old CD1 nude, Ifnar−/−, NSG, and Rag1−/− mice were all persistently infected with MuAstV (Fig. 4A), which contrasted with the undetectable levels in our wild-type C57BL/6 mice. We noted that Ifnar−/− mice had a virus burden more than 2 logs lower than that of the other immunodeficient strains lacking an adaptive immune response, indicating that lymphocyte-specific immune responses are needed to prevent uncontrolled virus replication. These data also suggest that the MuAstV in Ifnar−/− mice could be evolving under suboptimal immune pressure in an environment lacking the type I interferon response, which has previously been shown to be critical for limiting astrovirus replication (16, 39). To investigate this possibility further, we performed unbiased sequencing to obtain the full genomes of the virus strains isolated from CD1 nude and Ifnar−/− mice. Of a total of 132,907 reads for the CD1 nude strain, 1,206 (0.91%) were virus specific, and 1,193 of these reads mapped to astrovirus to form the full contig (6,770 bases) (Fig. 4B). The unbiased sequencing of the Ifnar−/− strain yielded 904,907 total reads, of which 10,078 (1.11%) were virus specific (Fig. 4B). Of these reads, 9,821 formed two contigs that required Sanger sequencing to complete the full genome (6,807 bases). Compared to the STL1 strain obtained from C57BL/6J mice by Yokoyama and colleagues (15), our strains had numerous amino acid differences, primarily within ORF1a and ORF2, which encode the nonstructural and capsid proteins, respectively (Fig. 4C). By comparison, ORF1b, encoding the RNA-dependent RNA polymerase, was more conserved (Fig. 4C). These differences in each open reading frame (ORF) were also apparent in phylogenetic analyses using representative mammalian astrovirus strains (Fig. 5A to C). In comparisons across all known murine strains with a full ORF2 sequence in GenBank, the MuAstV strain isolated from CD1 nude mice (SJ001) shared the highest homology (80%) with strain Y, which was sequenced from a CD1 outbred mouse by Compton et al. (24) (Fig. 5D). In contrast, the MuAstV strain isolated from Ifnar−/− mice (SJ002) more closely resembled other strains (88% to 94%), having the highest homology with strains STL2, STL4, and BSRI (15, 17).
Two distinct murine astrovirus strains identified from immunocompromised mice by unbiased metagenomic sequencing. (A) Cages of 4- to 12-week-old immunocompromised and immunocompetent C57BL/6 mice were monitored over 5 weeks for murine astrovirus shedding, and the genome copies of the virus were quantified by qRT-PCR. (B) Read summary produced from our in-house virome pipeline, which first filtered out mouse-derived sequence before identifying reads of viral origin as shown. Murine astrovirus was the main species detected, along with laboratory contaminants, including norovirus and influenza virus, as well as endogenous gammaretrovirus. Read data are shown as raw counts (left panel) with the total reads per sample shown below, as well as reads per million (right panel). (C) By using unbiased metagenomic sequencing, full-genome sequences were obtained for viral isolates from CD1 nude (SJ001) and Ifnar−/− (SJ002) mice and compared to that for murine astrovirus strain STL1 (accession number YP_006843892.1) within the three open reading frames (ORF1a, ORF1b, and ORF2). The amino acid differences are indicated with dark lines.
Phylogenetic analysis of the two strains of murine astrovirus identified in persistently infected CD1 nude and Ifnar−/− mice. (A to C) Astrovirus genomes encoding all three ORFs were downloaded from the NCBI database, and protein sequences were predicted before a neighbor-joining tree was constructed using the distance matrix from the alignment. The maximum-likelihood trees were generated from the starting tree by using the LG matrix with stochastic branch rearrangement. Host species of the viruses are indicated with graphics, and the two murine astrovirus strains identified in this study are denoted by a light blue triangle (SJ001) and a dark blue square (SJ002). The bootstrap method (with 1,000 replicate trees) was used to determine the confidence of the tree topology as noted for ORF1 (A), ORF1b (B), and ORF2 (C). Reference sequences included those with accession numbers HQ916313, HQ916316, HQ916317, KM035759, KP264970, KT956903, LC047788, LC047790, LC047792, LC047793, LC047796, LC047797, LC047798, LC047800, NC_012437, KJ920197, NC_013443, AF141381, DQ028633, GQ495608, HQ398856, JF327666, JQ403108, KF039911, KF859964, LC064152, AY179509, AB829252, JX857870, JX544743, JX544744, KC609001, MF175075, JF713710, JX556691, JX556693, KU764484, KY073229, KY214437, KY214438, LC201585, LC201587, LC201588, LC201589, LC201590, LC201591, LC201592, LC201593, LC201594, LC201596, LC201599, LC201600, LC201602, LC201603, LC201604, LC201605, LC201606, LC201607, LC201608, LC201609, LC201613, LC201614, JN420351, JN420352, JN420357, JN420359, NC_005790, and NC_002469. (D) Percent amino acid identity for ORF2, which encodes the capsid protein, of SJ001 and SJ002 compared with other published strains of murine astrovirus.
To determine whether the replication kinetics differed between the distinct virus strains, we orally inoculated wild-type mice with the two strains and measured the virus levels in the feces every 2 days (Fig. 6). Although the peak of infection was comparable for the two strains (SJ001, 5.03 × 105 genome copies; SJ1002, 1.7 × 106 copies), SJ002 was not cleared from the animals until 70 dpi, whereas SJ001 was cleared at 21 dpi, demonstrating that infection with SJ002 was more persistent.
Two distinct murine astrovirus strains isolated from immunocompromised mice give rise to different shedding durations in wild-type C57BL/6 mice. Eight- to 10-week-old C57BL/6 mice were orally inoculated with a filtrate of each strain (SJ001 and SJ002) (100 μg/ml), and feces were collected and monitored by qRT-PCR for clearance of the virus. The dotted line indicates the lower limit of detection for the qRT-PCR assay.
MuAstV causes subpathologic inflammation in the gut and short-lived immunity.A previous study by Compton et al. showed a lack of pathology in CD1 outbred neonatal mice after MuAstV infection (24), which is consistent with the observations of our group and others with both human and turkey astroviruses (7, 9). To evaluate whether the two MuAstVs that we isolated caused pathology, we first performed hematoxylin and eosin (H&E) staining of GI and lung tissue with blinded review by a veterinary pathologist (Fig. 7A). As in previous studies, we found no indication of overt tissue damage, cell death, immune cell infiltration, or other signs of inflammation (Table 1). In fact, the small intestine tissues of mock-infected mice were virtually indistinguishable from those of infected animals at 3, 13, and 21 dpi (Fig. 7A). We also examined intestinal tissues of animals at 28, 35, and 42 dpi by H&E staining but found no notable immune cell infiltration or pathology. This finding is in stark contrast to observations with other murine models of enteric viruses, including murine norovirus (40) and rotavirus (41), both of which cause marked histopathologic changes in the gut.
Murine astrovirus infection does not cause overt histopathology in wild-type C57BL/6 mice. (A) Organs were collected after full-body perfusion with 4% paraformaldehyde at the indicated time points from murine astrovirus-infected or mock-infected 3-week-old mice (n = 2/group/time point). The images show H&E-stained sections of representative small intestines. (B) Reinfection of wild-type C57BL/6 mice after clearing the virus demonstrated a lack of long-lasting immunity. The dotted line indicates the lower limit of detection for the qRT-PCR assay. The data shown are from two independent experiments with n = 5 to 10 mice/group.
Review of pathology in organs from MuAstV-infected and mock-infected male animals
Given the lack of evidence of either pathology or inflammation, we hypothesized that infected animals would not elicit a robust immune response to the virus. To test this hypothesis, we determined whether SJ001-infected animals that cleared the virus were protected from a homologous reinfection at 2, 8, and 12 weeks after the clearance of the initial infection. While 100% of the animals were protected at 2 weeks postclearance, most animals became reinfected at 8 and 12 weeks postclearance, and virus was detected after 3 days. Animals that had cleared SJ002 were also reinfected after homologous challenge at 8 weeks postclearance, together indicating a lack of sustained immunity (Fig. 7B).
DISCUSSION
Despite astroviruses being a leading cause of diarrheal illness worldwide (2), there is a paucity of data on pathogenesis. This partially stems from the lack of developed models and tools for studying these viruses. MuAstVs are endemic in animal research facilities, but it is unclear whether the viruses are suitable for modeling human astrovirus infections. To address this question, we first characterized the mouse model in wild-type mice and found that, as in human infections, the spread of the virus was restricted to the GI tract, there was no inflammation or overt pathology, and the animals had poor long-lasting immunity to the virus. However, we also observed no diarrhea or evidence of fecal inconsistency, indicating that MuAstV is a suitable model for asymptomatic disease. Together, these data demonstrate the utility of murine model research and highlight the potential for obtaining mechanistic insights into astrovirus pathogenesis and immunity.
Similar to a previous report that described a 3-week time course of MuAstV infection in neonatal CD1 outbred mice (24), we observed virus clearance of SJ001 by 21 days. However, we noted that this was strain specific after evaluating SJ002 infection and finding a more protracted shed duration (Fig. 6). There was also an age-dependent phenotype, with a higher peak infection in recently weaned mice than in adult animals. While the underlying mechanism remains under investigation, the differences associated with age could be related to significant shifts in the microbiome of mice during the postweaning phase, with adults having more established microbiomes (42), differences in the maturing gut immune system, or development of the gut that could change the expression of attachment proteins or receptors.
Similar to a previous study (15), we found virus was restricted to the gastrointestinal tracts of wild-type mice. In contrast, extragastrointestinal spread has been observed in immunodeficient mice, similar to human infections (4, 43). We also show sustained replication in the stomachs of animals, which is notable given the low pH and the presence of digestive enzymes in the stomach. These data are in line with the notion that astrovirus particles are notoriously hardy, withstanding a pH of 3 and demonstrating resistance to detergents, lipid solvents, and chlorine solvents (44). While there was no notable viremia, virus was detected in the lung, most likely resulting from aspiration of the inoculum. Attempts to confirm replication by immunofluorescent staining in tissue sections using either a peptide-based rabbit polyclonal serum directed to the virus capsid or a murine antibody that detects double-stranded RNA were unsuccessful. While additional evidence of replication will require further tool development, this finding is notable given the recent evidence that the MLB and VA strains are associated with respiratory symptoms in humans (35, 36). These experimental data could indicate that the virus infects a common cell type that is shared between the lung and gut mucosa.
Given that the receptor(s) for astroviruses is currently unknown and that the cell tropism of these viruses has not been extensively studied, the MuAstV model could provide a useful system with which to address these issues and others, including topics surrounding RNA virus evolution, persistence, and immunity. Indeed, we isolated two unique strains of MuAstV from persistently infected immunocompromised mice and showed that they caused chronic infections of different durations in wild-type mice. One likely explanation for this difference is that the virus adaptation was different in the two original hosts: one host lacked T cells and permitted uncontrolled virus replication (CD1 nude mice), whereas the other host lacked the type I interferon response and showed only moderate levels of virus shedding (Ifnar−/− mice). The differences in host immune pressure could ultimately drive differences in the virus populations, as this concept is supported by the phylodynamic framework for RNA virus evolution, which illustrates that within-host virus evolution is highest among hosts with suboptimal immune pressure (45, 46). Thus, we would hypothesize that the chronic infections in immunocompetent wild-type mice were due to underlying differences in virus quasispecies, and subsequent studies will need to address whether we can identify specific genetic signatures that are associated with the prolonged virus shedding in immunocompetent hosts. Finally, despite the prolonged exposure to MuAstV antigen, the lack of inflammation most likely failed to drive lasting immunity in the mice, providing preliminary evidence that could explain the similarly poor immune responses to astrovirus that have been observed previously in turkeys and humans (47, 48). Critically, such prolonged infections and likelihood of reinfection could have significant implications for other mouse models of disease. To this end, a previous study found no differences between MuAstV-infected and uninfected Nod-like receptor 3 (Nlrp3)-deficient mice with dextran sodium sulfate (DSS)-induced colitis (49). However, a recent study by colleagues (63) has described how MuAstV infection can protect immunocompromised mice against murine norovirus via the induction of interferon lambda, indicating the need for further investigation of additional mouse models of disease with different genetic backgrounds.
The results from this study must be framed within its strengths and weaknesses. First, our study represents a leap in progress in the development of a small animal model for astrovirus by characterizing two distinct strains of MuAstV. The data also add to the current astrovirus literature by highlighting the genetic and biological heterogeneity among MuAstVs. We encountered three technical challenges in our experiments that may reveal new biological insights into astroviruses. The first of these challenges appeared during our attempts to develop a symptomatic model of astrovirus by using both primary isolates and a laboratory-adapted strain of human astrovirus. Despite multiple approaches, these experiments proved unsuccessful, thereby providing data contrary to the current dogma that astroviruses have low species barriers (37), and additional follow-up studies are required. The second insight came during our trials with inoculum preparation. We were surprised to discover that sucrose purification of virus from fecal filtrate abolished infectivity in our mouse model. Although it is possible that the bulk of MuAstV particles in the feces are noninfectious, we have experienced no such technical difficulties with human or turkey astroviruses. Thus, further studies are needed to better understand the species-specific differences in the structure and stability of the virus. The lack of a cell culture system for MuAstV also impeded our ability to determine the number of infectious particles in our inoculum, as well as to determine whether the SJ002 virus produced during prolonged shedding remains infectious. Although our study is the most comprehensive to date, questions remain. Most important, is a diarrheal model possible with the correct combination of viral and mouse strain or age? Because the first report of murine astroviruses in nude mice indicated that 68% (34/50) of the mice did not have diarrhea and because of the more recent finding that experimentally infected neonatal mice also do not develop clinical signs (24), it is unclear whether a consistent symptomatic model is possible in mice via natural infection alone. However, the prospect of developing such a model remains, given that we have much to learn about the host and virus factors that dictate the various astrovirus disease states.
In conclusion, our evaluation of MuAstV as a model for human infection revealed unique facets of asymptomatic disease in wild-type mice, including potential explanations for the biological heterogeneity between two distinct virus strains. Given the lack of testing and asymptomatic nature of MuAstV infections, investigators may be unaware that these endemic viruses could have a potential impact on experimental outcomes and reproducibility. Future studies are needed to define the consequences of MuAstV infection for other mouse models of disease.
MATERIALS AND METHODS
Ethics.All animal experiments were approved by the St. Jude Children’s Research Hospital (St. Jude) Institutional Animal Care and Use Committee (protocols 570 and 513). St. Jude is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International (AAALAC-I) and has an approved Animal Welfare Assurance Statement on file with the Office of Laboratory Animal Welfare (A3077-01). These guidelines were established by the Institute of Laboratory Animal Resources and were approved by the Governing Board of the U.S. National Research Council. Fecal samples were collected from pediatric oncology patients at St. Jude and were submitted to the hospital’s clinical diagnostic laboratory between January 2010 and June 2011. All samples were deidentified before testing for astrovirus. The St. Jude Institutional Review Board approved this study with a waiver of consent.
Preparation and quantitation of virus inoculum.Inoculum was prepared by homogenizing either intestine or feces with 2.0-mm zirconium oxide beads (Next Advance) and phosphate-buffered saline (PBS) in a Bullet Blender tissue homogenizer (Next Advance). The supernatant was collected after centrifugation at 14,000 rpm for 2 min and filtered through successive 0.80-, 0.45-, and 0.22-μm filters. Sucrose purification with ultracentrifugation was performed as previously described (50). To quantitate virus levels, RNA was first extracted using either the Viral RNA/DNA kit (Thermo Fisher Scientific) for feces and filtrates or the MagMAX Pathogen RNA/DNA kit (Thermo Fisher Scientific), in accordance with the manufacturer’s instructions, and purified with the KingFisher Flex purification system (Thermo Fisher Scientific). Copies of the virus genome were then quantified using a G-block standard (Integrated DNA Technologies) in a one-step qRT-PCR using TaqMan Fast Advanced Master Mix Virus (Applied Biosystems) with primers, probe, and conditions as previously designed (15, 16).
HAstV preparation.Human astrovirus 1 (HAstV1) was propagated in human colonic adenocarcinoma cells (Caco-2 cells, ATCC HTB-37) grown in culture in modified Eagle’s medium (MEM) supplemented with 2 mM glutamine, 1 mM sodium pyruvate, and 10% fetal bovine serum, as previously described (50). Briefly, after incubation of the Caco-2 culture for 72 h, virus was harvested by sonication, cellular debris was removed by centrifugation, and then viral titers were determined by a fluorescent-focus assay, as previously described (50). MLB1 and VA2 isolates were obtained by screening deidentified remnant fecal samples from St. Jude patients with diarrheal symptoms, using a real-time RT-PCR assay as previously described (51), followed by genotyping with primers targeting sequences within ORF1b (52). The inoculum was prepared by generating a mixture of 10% feces in PBS and then filtering it with a 0.22-μm filter.
Animal experiments.Male and female wild-type C57BL/6 mice were purchased from The Jackson Laboratory. The St Jude Ifnar−/− colony was established using rederived mice originally obtained from Laura Knoll (The University of Wisconsin). For all experiments, before inoculation on day 0, mice were confirmed to be negative for MuAstV by qRT-PCR screening of fresh feces. For experimental MuAstV infections, wild-type C57BL/6 mice aged 3 to 8 weeks were orally inoculated with 2.7 × 107 genome copies or 100 mg/ml filtrate in 100 μl PBS. Mock infections used PBS alone. Samples of fresh feces, tissues, and blood (collected by cardiac puncture) were obtained at the time points indicated and stored at −80°C until processed. For histopathologic analysis, mice underwent full-body perfusion with 2% paraformaldehyde. Perfused organs were stored in 30% sucrose before being embedded in paraffin, sectioned, and stained with hematoxylin and eosin by the St. Jude Veterinary Pathology Core. After deparaffinization and antigen retrieval (Target Retrieval Solution; Agilent), tissue sections were blocked using mouse-on-mouse blocking reagent (Vector Laboratories) or 5% normal goat serum for 1 h at 37°C before J2 (anti-double-stranded RNA [anti-dsRNA] antibody; Scions) or a peptide-based rabbit polyclonal serum targeting the astrovirus capsid was added and left for 1 h at 37°C, followed by secondary antibodies (anti-mouse IgG–Alexa488 or anti-rabbit IgG–Alexa488) with DAPI (4′,6′-diamidino-2-phenylindole). Slides were visualized on an Evos microscope (Advanced Microscopy Group). Reinfections were performed at the indicated time points after virus clearance, as evidenced by a lack of virus shedding in the feces as detected by qRT-PCR. For human astrovirus inoculations in 8- to 16-week-old Ifnar−/− mice, 1 × 107 infectious particles of HAstV1 or a 10% (wt/vol) fecal filtrate containing the MLB1 or VA2 strain was given by oral gavage (100 μl), intranasal inoculation (30 μl), or intracisternal magnal inoculation (20 μl). Feces and organs were collected and screened by real-time RT-PCR. Intraperitoneal inoculations were performed with 8-week-old Ifnar−/− and wild-type C57BL/6 mice, which were given 1 × 106 infectious particles of HAstV1 in 100 μl.
Unbiased sequencing and metagenomics assembly.RNA from fecal homogenates was obtained by TRIzol LS extraction. Double-stranded DNA was generated using the Superscript III First-Stand Synthesis System (Thermo Fisher Scientific) with random hexamers, followed by the NEBNext Ultra II Non-Directional RNA Second-Strand Synthesis Module (New England BioLabs). DNA was purified with AMPure XP beads (Beckman Coulter) and quantified by Qubit analysis (Thermo Fisher Scientific), and then libraries were prepared using the Nextera DNA library preparation kit (Illumina), according to the manufacturer’s protocol, and loaded onto the Illumina MiSeq platform. Illumina adapters were trimmed from the paired-end reads using Trimmomatic (53). To determine sequences of viral origin, we performed our in-house virome pipeline analysis. First, reads that mapped to mouse were filtered out by bowtie2 (54) mapping to the mouse genome sequence (mm9 reference assembly). The reads that did not map to mouse were quality filtered by PRINSEQ-lite to remove low-quality reads (55). The resulting reads were aligned against eukaryotic viruses only in the NCBI nonredundant database, using DIAMOND BLASTX (56) to identify potential eukaryotic virus hits. These potential hits were validated by aligning them against the entire NCBI nonredundant database with DIAMOND BLASTX, and the reads that aligned best against the eukaryotic viruses were counted and tabulated for further analysis. To obtain the full sequences of the virus strains, the reads were then assembled de novo by using metaSPAdes (57). The assembled contigs were analyzed via BLASTN against the NCBI database to find the astrovirus contigs. Sanger sequencing was used to fill in the gaps when contigs did not overlap the entire astrovirus genome. The short reads were mapped against the resulting astrovirus genomes, using BBMap to determine the coverage statistics (median fold coverage, 73-fold for SJ001 and 941-fold for SJ002).
Sequence and phylogenetic analyses.The NCBI ORFfinder was used to predict the protein-encoding segments. The ORF2 encoding the capsid of SJ001 and SJ002 was compared to that of other MuAstVs. The percentage of identical amino acid residues in the pairwise comparison was reported, and the amino acid p-dist was calculated as (1 − fraction identical). All ORFs from SJ001 and SJ002 were compared to STL1, and a diagram was generated to visualize the pairwise relations between strains. To generate phylogenetic trees, full astrovirus genomes were downloaded from the NCBI database. The protein sequences from the ORFs were predicted using metaProdigal for consistent protein translation (58). A phylogenetic tree was constructed for each ORF, based on the protein sequence, by using the DECIPHER and phangorn packages in R (59, 60). Briefly, sequences were aligned using the AlignSeqs and StaggerAlignment functions in DECIPHER. The starting neighbor-joining tree was constructed using the distance matrix from the alignment. The maximum-likelihood trees were generated from the starting tree by using the LG model (61) with stochastic branch rearrangement. The bootstrap method (1,000 replicate trees) was used to determine the confidence of the tree topology. The maximum-likelihood tree was saved with the bootstrap percentage in the Newick format and visualized using Fig.Tree 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/ ) and ggtree (62).
ACKNOWLEDGMENTS
We thank Keith A. Laycock for scientific editing of the manuscript.
Funding for this research included National Institutes of Health Allergy and Infectious Diseases grants R21 AI135254-01 and R03 AI126101-01 to S.S.-C. and T32 AI106700-03 to V.C., as well as funding from ALSAC to S.S.-C.
FOOTNOTES
- Received 8 February 2019.
- Accepted 26 March 2019.
- Accepted manuscript posted online 10 April 2019.
- Copyright © 2019 American Society for Microbiology.
