In this study we developed a neonatal model of respiratory viral infection using the mouse pathogen Parainfluenza 1 virus strain Sendai/52 (Sendai virus; SeV), which is closely related to RSV. SeV and RSV are both enveloped RNA viruses and they share a highly conserved genetic organization . RSV infection in mice does not resemble human RSV bronchiolitis for several reasons (reviewed in ). First, RSV infects the airway epithelial cells of humans, but it infects the type I alveolar pneumocyte in the mouse resulting in pneumonia rather than bronchiolitis [13, 14]. Secondly, RSV does not replicate well in mice. Inhibition of the interferon (IFN) antiviral response is important for paramyxoviruses to establish replication in the host cell. RSV contains two non-structural proteins NS1 and NS2, which antagonize interferon action by inhibiting STAT1 and STAT2. The RSV NS1 and NS2 molecules are species specific; thus, they can inhibit human STAT1 and STAT2, but they do not inhibit the murine molecules . Finally, in contrast to human RSV infection, RSV infection in mice elicits eosinophilic inflammation and a Th2 cytokine response . In contrast to murine RSV infections, our previous studies have shown that SeV infection of adult mice results in a neutrophilic bronchiolitis and immune response pattern that closely resembles RSV infection in humans . This data suggests that SeV may be a useful model of paramyxoviral infection.
We demonstrated that the neonatal innate immune response to SeV was distinct from the adult response. In adult mice, body weight loss correlated with peak levels of virus in the lung. In contrast, the body weight and viral titer did not correlate in the neonates. Similar patterns of body weight loss in adults and a later slowing of growth in neonatal mice have been observed after infection with RSV, Pneumonia virus of mice (PVM), and the PR8 strain of influenza [10, 11, 18]. It is possible that the differences of body weight patterns between adults and neonates are reflective of distinct cytokine expression. Some cytokines, including TNF-α, interleukin-1 (IL-1), IFN-γ, and IL-6 have been implicated in cachexia . In our study, the adult mice have significantly increased amounts of IFN-γ and TNF-α, which correlated with their weight loss. However, at higher doses neonates had a slowing of growth (or failure to thrive) after day 9 and this did not correlate with TNF-α expression. In addition, the use of Enbrel to block TNF-α during SeV infection in adult Stat1-/- mice did not affect body weight loss, suggesting that additional cytokines may be responsible for body weight loss during paramyxoviral infection .
Importantly, the neonates and adults had comparable viral titers in the lungs, but the neonates exhibited significantly less inflammation. These findings were also consistent with observations made during infection of murine neonates with the paramyxoviruses RSV and PVM. However, neonatal mice infected with the orthomyxovirus influenza H1N1 (Strain A/PR/8/34) had a delayed clearance of virus and increased numbers of inflammatory cells in the interstitium . Therefore, these results of equivalent viral titers and reduced inflammation in the neonates may be unique to paramyxoviral infection.
SeV infection of the airway epithelium causes TNF-α production, which results in the secretion of CXCL2 . In this study, a dose response of SeV caused a dose-dependent increase in TNF-α in the adult mice, but not in the neonatal mice. As expected, the adult mice also produced increased levels CXCL2, which recruited neutrophils into the lungs. In contrast, the neonates did not exhibit an increase in either TNF-α or CXCL2 levels, and there was little to no recruitment of PMN to the lungs. Both TNF-α and CXCL2 are known to increase numbers of circulating neutrophils by releasing them from the bone marrow reserve [20, 21]. In addition, neonates have quantitative defects of neutrophil storage pools as well as the capacity to generate neutrophils . Thus, the reduced numbers of PMN in the neonatal lungs may be due to both reduced mobilization from the storage pool and reduced chemokine levels in the lung. Reduced TNF-α and chemokine levels have also been observed in neonates infected with RSV and PVM suggesting that this pattern is consistent in the neonatal response to paramyxoviral infection [10, 11].
Even with increased doses of SeV, there was differential cytokine expression in the neonates. The neonates had significantly reduced levels of IFN-γ even at 100-fold higher inoculums of virus. This reduction in IFN-γ did not affect viral clearance in the neonates, which was not unexpected because weight loss, mortality, histology, and cytotoxic CD8+ T lymphocytes numbers were all normal in SeV infected IFN-γ deficient mice, suggesting that IFN-γ is not necessary for paramyxoviral clearance [23, 24]. IFN-γ upregulates ICAM-1 expression on airway epithelial cells both in vitro and in vivo . In vitro studies of airway epithelial cells have shown that ICAM-1 expression is increased following paramyxoviral infection and necessary for leukocyte adherence and extravasation . Our observation of reduced ICAM-1 expression may be due to the reduced expression of IFN-γ in neonates. Consequently, the reduced expression of ICAM-1 may contribute to reduced inflammation in neonatal lungs. Previous studies demonstrated that reduced inflammation in adult ICAM-1 knockout mice protected them from weight loss, but did not impair viral clearance, suggesting that inflammation is not necessary for viral clearance .
When paramyxoviruses first infect the airway epithelial cells they are detected by pattern recognition receptors (PRR). The PRRs involved in detecting paramyxoviral infection are the Toll-like receptors (the TLRs including TLR-3, TLR-4, TLR-7, and TLR-8) and the retinoic acid-inducible gene 1 (RIG-I)-like receptors (RLRs including RIG-I, Mda-5 and Lgp2). TLR-4 is expressed on the plasma membrane of epithelial cells and it recognizes the fusion (F) protein of hRSV . TLR-3, TLR-7 and TLR-8 are expressed intracellularly on endosomes and they recognize dsRNA and ssRNA, respectively [28–30]. RIG-I and MDA5 are cytoplasmic pattern recognition receptors that recognize viral 5' triphosphate RNA [31, 32]. They both contain a DEXD/H helicase domain and 2 caspase-recruitment domains (CARD). Lgp2 (laboratory of genetics and physiology 2) is a helicase that is similar to RIG-I but it lacks the CARD domain and may act as a regulator of RIG-I signaling [33, 34]. It has been shown that RIG-I is essential for an appropriate immune response to both the paramyxoviruses RSV and Sendai virus (SeV). In vitro, RIG-I was necessary for type I IFN production in response to SeV infection , and RIG-I knockdown during RSV infection of airway epithelial cells inhibited both NFkB and interferon regulatory factor (IRF) signaling .
Neonatal TLR responses of monocytes, conventional dendritic cells, and plasmacytoid dendritic cells are distinct from the corresponding adult response to TLR stimulation [36, 37]. However, very little is known about PRR expression and activation in neonatal lung during respiratory viral infection. In the present studies, murine neonates had normal expression of the TLRs in the lung at baseline and during paramyxoviral infection; however, all of the RLRs (Lgp2, Mda-5 and RIG-I) had significantly delayed expression during infection. One study of human infants examined PRR mRNA expression in nasal washings of infants hospitalized with respiratory viral bronchiolitis. There was no correlation of PRR expression level and clinical severity score, but there was a positive correlation between RSV viral load and expression of RIG-I . Our results show that RLR expression increases in neonates during the course of infection, so it is possible that these human samples were taken when the infection was well established and RIG-I was already upregulated. Thus, baseline expression of PRRs in human infants is still unknown.