69% of US children are infected with RSV in the first year of life, and almost all by the age of two years [6]. The majority are asymptomatic or have only mild symptoms. International studies estimate the annual incidence of RSV bronchiolitis requiring hospital admission to 22 - 31/1000 amongst infants < 1 year [2,6,9,10].

Why so few children exposed to RSV should develop symptoms requiring hospital admission is yet to be adequately explained, although research has provided important clues in the last 10 years. Some genetic associations with RSV infection have already been described [11-15] and predisposition to RSV bronchiolitis is likely to be multifactorial. Increased knowledge about the pathophysiology of bronchiolitis and predisposing factors will aid researchers in the development of preventative measures and therapies for bronchiolitis [16,17].

Dicer is an RNase III enzyme that produces micro RNA (miRNA) sequences by cleaving nuclear derived pre-miRNA. miRNA interferes with gene expression by binding to complementary mRNA, facilitating mRNA degradation and preventing mRNA translation into protein. This mechanism is called RNA interference (RNAi), and is an important post-transcriptional regulator of gene expression [18,19]. There is good evidence to suggest that RNAi also has a direct anti-viral function. Cellular derived miRNA has specific antiviral effects, interfering with viral gene expression [20-22]. Dicer can also cleave long genomic viral dsRNA sequences into short interfering RNA (siRNA). siRNA can thus be virally derived and have perfect sequence specificity for viral mRNA. While this is a specific anti-viral mechanism in plants and invertebrates, there is currently no evidence to suggest that this is true for mammals, in which the interferon system is more important in viral defense [20,22]. However, synthesized siRNA tailored to specific viruses has been shown to have significant anti-viral effects in humans, in an interferon-independent manner [23,24].

Our selection of controls and classification into mild or severe disease was retrospective and required some assumptions about patients who had not been examined at our hospital for respiratory disease. This includes all the controls and 4 of those with mild disease tested for RSV by their GP, but not referred to hospital. This introduces the possibility of misclassification by several mechanisms: controls may have had RSV disease but not been tested, or may not have visited their GP at all; 4 infants not sent to hospital were assumed to have mild disease, but may have had severe disease; we rely on accurate completion of medical records; there may have been inter-observer variation in the pediatrician's assessment of the severity of dyspnea. Given that 69% of children are infected with RSV in the first year of life [6], it is in fact likely that a number of the controls had RSV disease, but only had mild symptoms (e.g.: rhinitis). Therefore, it is not surprising that we do not show a difference between mild and control groups. Increasing the number of control infants included in the analysis would have increased the power of the study and therefore the chance of discovering a difference between control and mild groups. However, we feel that the groups were so similar that any difference is not likely to be clinically relevant. Patients with unrecognized severe disease included in the control or mild disease group may increase the risk of type 1 or type 2 errors when considering the severe disease group. However, given our experience with the patient population and GPs in our area, we consider that the chance that infants with severe disease would not visit their GP or be referred for admission is small. Altogether, we had a low number of patients in our analysis. Increasing the size of the birth cohort may have increased the total number of patients, the power of the study and the chance of discovering a significant difference between mild and severe disease groups.

Dicer protein was below detection level by Western blot analysis of cord samples from our cohort. This seems remarkable given that the qPCR experiment showed high amounts of Dicer mRNA. Our antibodies are also specific for proteins translated from the commonly occurring splice variants identifiable by our mRNA. We corroborated our primary antibodies and Western blot technique by demonstration of Dicer in colorectal cancer cells. We encountered the same phenomenon in new cord and adult peripheral blood leukocyte samples, so degradation of Dicer protein in our cohort samples seems an unlikely explanation. Dicer is essential for, and miRNA profiles are well described in, normal leukocyte function, development and proliferation [33]. We can therefore not explain why Dicer protein was not detectable in our samples. To our knowledge, Dicer has not previously been demonstrated by Western blot analysis in human leukocytes. One could speculate that peripheral blood leukocytes contain other splice variants of Dicer detected by our mRNA probe but not by our antibodies; that leukocyte Dicer is bound to other proteins and therefore not detected in our blots; or that a different Dicer protein is active in leukocytes, but we find no literary reports to confirm this. It may be that Dicer mRNA is not transcribed to Dicer protein or the protein is rapidly degraded once produced, but given the necessity for Dicer in leukocytes, this seems unlikely. We can, however, not exclude variations in post-transcriptional modification, and we are unable to measure whether there are differences in Dicer protein level between patients and controls. Thus we cannot verify that the difference we see in Dicer mRNA level reflects a real difference in Dicer protein level in our cohort. However, several articles investigating Dicer gene expression in cancer have found a good correlation between Dicer mRNA level and Dicer protein level [34-36], and Dicer mRNA level as a proxy for Dicer protein level therefore seems reasonable.

In cord blood, the main source of mRNA should be nucleated cells, primarily myeloid and lymphoid leukocytes. The functional effect of Dicer downregulation in these cells is likely to be two-fold: disrupted cellular function, and reduced direct anti-viral activity. In murine models, absence of Dicer leads to considerable disruption of cell function in T-lymphocytes, natural killer cells and Langerhans cells [37-40]. In alveolar cell cultures, influenza virus mediated Dicer downregulation coincides with significantly accelerated cell death [21]. In humans, reduced Dicer mRNA levels have been associated with hepatocellular carcinoma, invasive epithelial ovarian cancer and metastatic breast cancer [34-36], and it is likely that disruption of miRNA mediated gene regulation plays a role in cancer development [33]. It is thus clear that not only the absence of, but also downregulation of Dicer results in disruption of cellular activity.

The immune response to RSV in infants is primarily driven by the innate immune system until the point of maximal symptoms, at which time recruitment of the cellular immune response hastens viral clearance [17,41] and clinical improvement. In leukocytes, Dicer downregulation in an anti-RSV setting would likely be most significant in the myeloid and natural killer cells, as they survey the lung environment, identify antigen and regulate the innate immune response to RSV. On recognition of viral antigen, myeloid dendritic cells recruit neutrophils and macrophages, and migrate to lymphoid tissue where they activate a cellular immune response [17,42]. Impairment of innate immune cellular function in association with Dicer downregulation in our patients may thus disturb the innate response to RSV and likely delay the recruitment of TH1 helper cells, which may also be dysfunctional [37].

The Dicer-mediated production of endogenous anti-viral miRNA may also be reduced in our patients. In an investigation of the role of Dicer in influenza, in-vitro knockdown of Dicer to a functional level of 30% in human alveolar cells resulted in increased influenza virus replication and greater apoptosis rates [21]. These were interesting results, and it is tempting to hypothesize that Dicer downregulation would similarly result in greater RSV viral load in our patients. This could result in increased apoptosis, greater activation of the immune response, more airways inflammation and therefore more severe disease. However, blood leukocytes and human alveolar cells are quite different cells types, and we therefore cannot assume that our results reflect Dicer expression levels in lung tissue. Our experiment was not designed to investigate the state of pulmonary epithelial cells in infants prior to RSV exposure, and such an experiment is currently not feasible due to ethical and practical issues.

RNAi as an anti-viral therapy currently receives much interest, and synthetic siRNAs with tailored activity against HIV, hepatitis B virus, human metapneumovirus and RSV are under development [24,43-45]. ALN-RSV01 (Alnylam Pharmaceuticals) is a synthesized siRNA with activity against the RSV N-protein [24]. In a recent trial, ALN-RSV01 was administered intra-nasally to adults before and after nasal inoculation with RSV. There was a significant anti-viral effect, with fewer patients RSV-culture positive in the ALN-RSV01 group compared to placebo [23]. In another trial, adult lung transplant recipients with proven RSV infection were randomized to nebulized ALN-RSV01 or placebo. The ALN-RSV01 group had a lower nasal RSV load, an improved symptom score, and a lower incidence of bronchiolitis obliterans syndrome, a known complication of RSV infection in this patient group [46]. Our findings suggest that infants susceptible to severe RSV infection may have a reduced capacity to produce anti-viral miRNA, strengthening the hypothesis that synthesized anti-RSV siRNA will have clinically relevant effects in lower respiratory RSV disease in infants. This is significant, given that after 5 decades there are still no safe and effective vaccines or treatments for RSV other than prophylaxis [16].