DENV infections are a major public health problem and constitute a real challenge due to the absence of vaccines and effective antiviral drugs. There are no non-human species that naturally develops disease following DENV inoculation into peripheral sites and that resemble the clinical symptoms observed in humans . The mechanisms underlying the immune responses to DENV infection are still poorly defined, and the lack of suitable animal models has hampered investigation of host- and virus-specific mechanisms that control primary and sequential DENV infections.
In this study, we have compared three minimally passaged DENV-1 isolates from patients experiencing the various clinical forms of dengue disease (DF; DHF; and DSS, respectively) in vivo by infection in mouse. We show that the isolate from a DSS patient infected the largest number of BALB/c mice, with a different tropism compared to the DENV-1 isolates obtained from milder clinical cases during the same outbreak. This indicates that DENVs within the same serotype and genotype may differ enough to cause various clinical conditions in vivo. These results confirm a phenotypic distinction previously observed in vitro with the same DENV-1 isolates where the DSS isolate replicated less efficiently in mammalian Vero cells than the DF and DHF isolates, and elicited apoptosis in mosquito C6/36 cells .
The aim of the study was to compare wild-type features of DENV-1 isolates passaged only once in vitro and never before inoculated in mouse or cultured in mouse cells prior to direct intravenous inoculation into BALB/c mice. BALB/c mice are known to be susceptible for DENV infection although with discrete pathological features [11, 30]. We found that the mice did not develop dengue disease, and did not present any symptoms of the infection. However, viral RNA was detected in various organs depending on the DENV isolate injected (table 1). We were unable to detect any systemic viremia and this suggests an absence or a very low level of virus replication in circulating peripheral blood mononuclear cells. Alternatively, the lack of detectable viraemia could indicate that the virus had already disseminated into peripheral sites, since tissue infection was measured from day 3 p.i.
In brief, viral RNA was detected by qRT-PCR in spleen, liver, lungs, kidneys, and brain. Viral RNA was found in the kidneys in only one mouse, which is in line with earlier studies . Organ tropism clearly differed between the three clinical DENV-1 isolates and our findings suggest that the disease severity may correlate with high tissue viral burden, even if plasma viraemia was below the detection limit. The DENV isolate from a DSS case showed a strong preference for brain tissue, compared to the two isolates from milder dengue cases that primarily were found in other organs as lungs and liver. Neurotropism in mice has been previously suggested to be associated with DENV virulence when comparing DENV-3 isolated from Brazilian patients .
Breakdown of the blood-brain barrier in DENV infected mice have been described before, and were shown to be dose-dependent . Genomic sequencing performed previously  identified a unique amino acid substitution L→F476 in the E protein of the DSS isolate compared to the DF and DHF isolates, that may explain the neurotropic character of the DSS-isolate. The NS1 gene of the DSS isolate harboured another unique amino acid substitution, K→R115 in comparison to the two other DENV-1 isolates. The soluble form of NS1 (sNS1) is a dominant target of humoral immunity and activates complement components in normal human serum, and is proposed to play a significant role in the pathogenesis of disease [25, 34]. The NS1 glycoprotein is glycosylated at two sites, N130 and N270, and glycosylation of both residues has been suggested to be required for neurovirulence in mice . It is remains to be elucidated whether the possible breakdown of the blood-brain barrier and neurotropic character of the DSS isolate has any correlation to the disruption of the endothelium observed in humans suffering from vascular leakage. The three remaining amino acid substitutions unique for the DSS isolate were located in the NS protein 3 (S→P118), and 5 (T→I49, and S→N830). These differences could potentially alter the enzymatic activity of these proteins.
The DSS isolate seemed to persist longer in vivo since viral RNA was detected in a majority of mice on day 15 p.i. compared to the DF and DHF infected mice, and thereby exhibited infection kinetics differing from the two other DENV-1 isolates (DF and DHF) where viral RNA decreased after day 6 p.i. Thus, the DSS isolate could be regarded as more virulent than the two other DENV-1 strains, since a majority of the DSS-inoculated mice had detectable levels of viral RNA, infection was also less affected by the dose of inoculums and persisted longer in a majority of mice (Figure 1). The wild-type DENVs analyzed here could provide evidence for differences in virus replication, which in turn could influence the clinical outcomes of the infection, and eventually partitally explain some differences in virulence observed in humans.
These results confirm the in vitro findings of the same DENV-1 isolates, that showed that the strains derived from a DSS patient can be distinguished based on phenotypic characteristics from the isolates derived from a DF and DHF case . We have previously shown that the DSS virus isolate exhibit much slower replication kinetics in mammalian Vero cells and an apoptotic response in mosquito cells, and we observed in this study that the DSS virus isolate persist longer in vivo with extensive neuroinvasion.
Cytokines involved in inflammation were quantified in serum collected from inoculated mice and compared in regard to DENV-1 isolate, virus titer inoculated, and time-point p.i. Overall, the levels of the proinflammatory cytokines (IFNγ, IL-1β, IL-2, IL-6, IL-10, IL-13, MCP-1, RANTES, and TNFα) quantified were low, which is most probably due to fact that BALB/c mice is poorly susceptible to wild-type DENVs. The mice did not develop clinical apparent disease, and showed only discrete lesions in internal organs. Common for several cytokines, however, was a dose-dependent response to inoculated virus (IFNγ, IL-6, IL-13, and RANTES). The highest levels of measured cytokines were also in general at the beginning of the experimental period (IFNγ, IL-10, and MCP-1 peaked at day 3 p.i.). A curious finding was that DF-inoculated mice had higher detectable serum levels of IL-10, MCP-1, and RANTES, which have been seen to be increased in patients with severe DHF and DSS [36–42]. The preference for the brain of the DHF-, and most notably the DSS-infected mice, could imply locally high levels of proinflammatory cytokines that do not circulate systemically. This could explain why the IL-10, MCP-1, and RANTES levels dominated in the DF-inoculated mice, since the thoracic and abdominal organs were the primary target for DF isolate. Cytokines like IL-1β, IL-2, and TNFα did not show any differences compared to mock-infected controls and this could be due to the transient nature of many cytokines as well as the limited susceptibility of DENVs in mice.