Here, we describe the establishment of a persistent infection in vivo
with MNV-3 recovered from cDNA by reverse genetics. To our knowledge, this is the first time that a persistent infection of C57BL/6 mice with MNV recovered from cDNA has been achieved. However, during the course of this study, a reverse-genetics system for the CR6 strain of MNV, shown previously to establish a persistent infection (Thackray et al., 2007
), has been described. Recombinant virus derived from this system has to date only been used to identify virulence determinants in the VP1 protein within a STAT1−/−
model (Strong et al., 2012
The combination of an efficient reverse-genetics system for MNV-3 and an animal model to study stable persistent infections constitutes a significant step forward in the development of systems for understanding norovirus molecular biology and pathogenesis. Persistent infections in humans by HuNoV have also been reported, although they mainly affect immunocompromised individuals (Capizzi et al., 2011
; Ludwig et al., 2008
). Nevertheless, recent data indicate that asymptomatic norovirus prevalence is approximately 12
% in the UK, although whether this represents subclinical acute infections or long-term asymptomatic secretion of virus is not known (Phillips et al., 2010
). A recent report has shown that MNV-3 infection in STAT1−/−
mice results in an acute infection with typical signs of gastroenteritis (i.e. delayed gastric emptying, changes to diarrhoeal symptoms) (Kahan et al., 2011
). This, combined with the fact that MNV-3 can establish a persistent infection in C57BL/6 mice, the most commonly used strain for the generation of knockout mice, provides additional support for the use of MNV-3 for the study of norovirus biology.
The observation that high viral RNA levels were detected in caecum and colon at day 28 post-infection, but not in the small intestine or other organs (MLN, spleen, liver), represents one of the most significant findings of this study. These data suggest that MNV-3 establishes a persistent infection in the large intestine, allowing efficient virus shedding in faeces over a prolonged time frame. The strong association between ulcerative colitis, a colon disease, and the presence of HuNoV during exacerbated disease (Khan et al., 2009
) supports the idea that the colon may also play an important role for HuNoV persistence. Two additional studies have provided further evidence for the connection between various colon disorders and active secretion of HuNoV in faeces, specifically in children with persistent diarrhoea and in patients with chronic diarrhoea associated with leukaemia (Capizzi et al., 2011
; Vernacchio et al., 2006
). A recent study comparing MNV-1 and MNV-3 in vivo
has shown that the viruses replicate in different organs, including the caecum and the colon, but they are rapidly cleared, being undetectable by day 7 post-infection (Kahan et al., 2011
). Possible explanations for these apparently conflicting data may be the sensitivity of the assays used and differences in the experimental set-up, i.e. age and strain of mice. It is also worth noting that, whilst the MNV-3 used here was derived from cDNA, that used in the previous study was a virus passed in tissue culture several times.
Interestingly, MNV persistent infections have also been established in 7- to 15-week-old mice using the CR6 strain, with viral RNA also being detected in colon 14 days post-infection at levels lower than those observed in the ileum and MLN (Cadwell et al., 2010
). The results presented in this study differ somewhat from these results, which may be a reflection of the differing virus strain under study (CR6 versus MNV-3) or the age of the animals used, i.e. 7–15 weeks versus 4–5 weeks in our study. Alternatively, MNV replication for longer periods (>14 days) may have resulted in clearance from the ileum and MLN, but not from the caecum and the colon.
Analysis of full MNV-3 genome sequences derived from serial passage of MNV-3 in RAW264.7 cells and persistent replication in mice resulted in marked differences in genome evolution. Serial passage of MNV-3 in cell culture selected for non-synonymous mutations in NS4 and VP1. We have previously reported the selection of mutations in NS4 and VP1 during MNV-1 cell-culture passage: K296E in VP1 and V11I in NS4. K296E in MNV-1 VP1 was shown to cause attenuation in STAT1−/−
mice (Bailey et al., 2008
; Strong et al., 2012
). Interestingly, MNV-3 and MNV CR6, both causative of persistent infections in wild-type mice, encode E296 in VP1 instead of K. The substitution E296K in MNV CR6 resulted in non-recoverable virus, which may indicate that this mutation is not tolerated in the context of the CR6 VP1 (Strong et al., 2012
). Position 296 lies in close proximity to position 301 in the structure of VP1, both at the tip of the protruding P2 domain (Taube et al., 2010
), which we interestingly observed to be mutated (T301I) during repeated passage of MNV-3 in RAW264.7 cells. Our observations would suggest that, upon inoculation of animals with a mixed population of viruses containing both T and I at position 301, namely MNV-3 p6, only T at position 301 in MNV-3 VP1 persists in vivo
. Clearly, further studies are warranted to examine the role of this position in tissue-culture adaptation and the possible associated fitness cost in vivo
. In addition, adaptive mutations in the picornavirus homologue of NS4 (3A) have been linked to increased virulence in cell culture and/or in vivo
(Arias et al., 2010
; Harris & Racaniello, 2005
; Núñez et al., 2001
), warranting further studies on the precise function of the norovirus NS4 protein.
Persistent replication of MNV-3 in vivo
did not select for any of the substitutions found during cell-culture passage, but instead selected for changes in VP2 and NS7. The mutations T4A in VP2, and V13A and L215F in NS7 were found in six, four and three animals, respectively. Additional mutations identified include A381T and T441I in VP1, and T91A in NS3. The T4A substitution in VP2 was found in the faeces of all animals analysed, but not during the serial passage of MNV-3 in RAW264.7, suggesting that this change may be important in vivo
but not relevant for cell-culture replication. For the related feline calicivirus, the VP2 protein is essential for virus replication and for the assembly of infectious particles (Sosnovtsev et al., 2005
). Recent studies with HuNoV VP1 and VP2 have revealed co-evolution of both proteins in a time-dependent manner, highlighting the integral role of the VP2 protein in the norovirus life cycle (Chan et al., 2012
). Strikingly, we have found that alanine at position 4 in VP2 is absolutely conserved in >30 natural MNV isolates analysed. Nevertheless, the presence of threonine at position 4 in VP2 had no effect on virus recovery by reverse genetics or on virus replication in cell culture, being stably maintained after multiple passages in RAW264.7 cells. This highlights that the VP2 sequence may contribute to virus fitness in vivo
due to selective pressures not observed in cell culture.
Mutations V13A and L215F in the viral RNA polymerase NS7 () appear to affect polymerase surface residues, distant from the catalytic site, in the palm and fingers domains, respectively (Lee et al., 2011
). Given their accessibility, these changes may be affecting the interaction of NS7 with potential viral or cellular factors, although this hypothesis requires further investigation. Interestingly, substitutions found in VP1 at lower frequency (A381T and T441I) are predicted to lie within exposed residues of flexible loops situated in the apical region of P2 and P1 domains, respectively, which could be indicative of selection by neutralizing antibody response. In particular, A381 is situated in loop E′-F′, identified previously as a major immunodominant epitope with antibody-escape mutants being mapped in position 386 (Lochridge & Hardy, 2007
; Taube et al., 2010
In summary, we have described for the first time the establishment of a persistent infection of C57BL/6 mice with MNV-3 recovered by reverse genetics. MNV-3 derived from cDNA persistently replicated in C57BL/6 mice for at least 56 days and was associated with high viral loads in the caecum and the colon. Preliminary data indicate that viral RNA is shed by infected mice for >6 months, emphasizing the ability of recovered MNV-3 to establish long-term persistent infections in vivo (N. McFadden, A. Arias, I. Goodfellow & P. Simmonds, unpublished results). This model may open new possibilities to study norovirus infections in vivo and different disorders associated with norovirus persistence.