Replication of sylvatic and endemic DENV-2 strains in vivo

A major limitation in dengue research is the lack of an inexpensive laboratory animal model that recapitulates human disease and viremia. A large number of nonhuman primates support DENV replication without developing clinical signs of illness (Halstead et al., 1973; Rosen, 1958). Nevertheless, because the duration and magnitude of virus replication in non-human primates often correlates with patterns of replication in humans, primates have become the gold standard for evaluation of live attenuated dengue vaccine candidates (Angsubhakorn et al., 1988; Blaney et al., 2005; Edelman et al., 1994; Hanley et al., 2004; Markoff et al., 2002; Men et al., 1996; Robert Putnak et al., 2005). However, economic and animal facility limitations preclude widespread utility of nonhuman primates as a useful model for DENV replication. Several murine models (Boonpucknavig et al., 1981; Chaturvedi et al., 1991; Cole and Wisseman, 1969; Hotta et al., 1981) developed to evaluate protective immune responses, have proven to be ineffective due to the absence of DENV replication and illness or the requirement for murine-adapted DENV (Cole and Wisseman, 1969; Sabin, 1952). Human peripheral blood lymphocytes (hu-PBL) engrafted in SCID mice (Wu et al., 1995) infected with DENV produce high viremia and virus loads in the spleen and lymph nodes. Limitations of this model include extensive labor requirements and variable reconstitution (successful engraftment) rates.

Observations documenting hepatic lesions (Couvelard et al., 1999; Lum et al., 1993), as well as detection of DENV antigen in hepatocytes (Kuo et al., 1992) during human infections formed the rationale for development of the SCID-xenograft model, where human hepatoma cells (HepG2 or Huh-7) are grafted intraperitoneally (i.p.) into SCID mice and DENV is injected directly into the i.p. tumor (An et al., 1999; Blaney et al., 2002). Mice develop gradual illness with peak viremia on day 7, whereas peak viral replication in the liver is detected by day 5 with gradual declines, and by day 11 virus has reached peak titers in the brain. This pattern of replication is similar to that which accompanies human infection, where DENV initially replicates at the site of infection and gradually disseminates to other organs and to the brain (although cerebral involvement is not believed to be common in human infections) via the circulation (Rothman, 1997). These are the first animal models where DHF/DSS-like manifestations (gastrointestinal bleeding and small focal hemorrhages in the livers of some mice) similar to those seen in human infections (Burke, 1968; Rosen et al., 1989) occur, and have proved useful for virulence testing of DENV vaccine candidates (Blaney et al., 2002; Whitehead et al., 2003). More recently, a humanized mouse model based on the grafting of human CD34+ cells in nonobese diabetic/severely compromised immunodeficient mice (NOD/SCID), has also been proposed as a model for studying the pathogenicity of DENV infection (Bente et al., 2005). However, based on the consistency of the SCID-Huh-7 xenograft model in evaluating the virulence of DENV vaccine candidates, we selected this model for the in vivo evaluation of our hypothesis.

To assess the human infection phenotypes of sylvatic versus endemic DENV-2 strains and to determine statistically dictated cohort numbers and tumor development efficiency, we conducted a pilot experiment using the Asian endemic 1349 and sylvatic PM33974 strains. The sylvatic strain generated moderate peak viremia levels of 4 log₁₀ffu/ml (N = 3), while the endemic strain generated significantly higher viremia of 6 log₁₀ffu/ml on day 7 (N = 7) (t-test, df = 7, t = 4.49, P<0.002) (data not shown). We also bled the mice at days 2, 4 and 7 post-infection (p.i.), and mean replication titers consistently peaked at day 7 p.i. (data not shown) as described earlier (An et al., 1999; Blaney et al., 2002). Power analyses indicated that a minimum of 6 animals per group were required to detect significant differences (P = 0.861, α = 0.05), with tumor development efficiency at 88%. Therefore, to compare the sylvatic versus endemic human infection phenotypes, six groups of 10 SCID-huh7 mice were infected with one of 2 sylvatic (PM33974, P8-1407) or 4 endemic (Asian 1349 and 16681, and American IQT-1950 and 1328) DENV-2 strains (Fig. 1). A significant overall difference among the mean serum titers of the six strains was detected (one-way ANOVA, P<0.0001) and a Tukey-Kramer post hoc test revealed multiple significant differences among individual pairs of strains (Table 2). However, this complex pattern of differences did not indicate a consistent or overall difference between sylvatic and endemic strains or between Asian endemic and American endemic strains. For example, the titer of Asian endemic strain 16681 was significantly higher than any of the other three endemic strains but not significantly different from sylvatic strain P8-1407.

Focus forming assays (FFA) and immunostaining

Ten-fold serial dilutions of virus were added to confluent C6/36 cell monolayers. The virus inoculum was removed 1 h later, cell monolayers were washed and overlayed with 0.8% methylcellulose (Sigma-Aldrich, St. Louis) diluted in Optimem (Invitrogen, Carlsbad) supplemented with 2% FBS, antibiotics, 1% l-glutamine and 1% NEAA. Plates were incubated for 4 days at 28 °C, the methylcellulose overlay was removed and plates were rinsed with phosphate-buffered saline (PBS), pH 7.4. Plates were fixed with the addition of ice-cold acetone and methanol (1:1) for 30 min at room temperature (RT). The fixation solution was aspirated and plates were allowed to air dry. Plates were then washed with PBS, followed by blocking (PBS supplemented with 3% FBS) and addition of mouse anti-DENV-2 ascites fluid (1:1000) and incubation for 30 min. The antibody was aspirated and plates washed 3 times in PBS followed by addition of secondary antibody conjugated to horseradish peroxidase (HRP) (KPL, Gaithersburg) (1:1000) and incubation at RT for 30 min. Plates were washed 3 times with PBS and aminoethylcarbazole (AEC) substrate (ENZO Diagnostics, Farmingdale), prepared according to the manufacturer’s instructions, was added and allowed to incubate in the dark for 10 min. Substrate solution was aspirated, washed with water and plates were allowed to air dry before scoring.

Isolation, stimulation of peripheral blood mononuclear cells and DENV infections of moDCs

Nine consenting healthy volunteers, with no history of infection to any of DENV serotypes and confirmed as negative by plaque reduction neutralization test (PRNT) assay, were used to obtain approximately 100 ml of blood. Peripheral blood mononuclear cells were then isolated from buffy coats by centrifugation over an Accuprep gradient according to the manufacturer’s protocol (Accurate Chemical Corp, Westbury). CD14⁺ monocytes were positively selected using a magnetic cell sorting (MACS) isolation column (Miltenyi Biotec, Auburn). Cells were counted and seeded in 6-well plates at a density of 1–2 × 10⁶ cells per well in RPMI 1640 culture medium supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin, 100 μg/ml streptomycin and 2 mM l-glutamine, containing 1000 U/ml recombinant human interleukin 4 (IL-4) (R&D Systems, Minneapolis) and 1400 U/ml granulocyte–monocyte colony-stimulating factor (Immunex, Thousand Oaks). On alternate days, one-half of the volume was removed and replaced with medium containing double (2×) the concentration of fresh cytokines. Cultures were maintained under these conditions for a total of 6 days, then cells were collected with gentle pipetting, washed and the moDC phenotype was confirmed by FACS analysis using a panel of FITC-conjugated monoclonal antibodies against CD40, CD80, CD83, CD86 and DC-SIGN. Subsequently, 1 × 10⁵ cells were infected with the selected DENV at MOI = 2 for 2 h at 37 °C. The virus inoculum was removed and the cells were washed 3 times with PBS to ensure removal of unadsorbed virus. Cells were resuspended in 2 ml of RPMI 1640 culture medium supplemented with cytokines and incubated at 37 °C. Cell-free supernatant aliquots were removed immediately prior and after infection, at day one and two post-infection and were assayed on C6/36 cells to obtain infection and progeny titers respectively. moDC infection rates were established by FACS analysis using a DENV-specific antibody (7E11, courtesy of Dr. Putnak), as well as IHC on cell cytospins.

Flow cytometry

Uninfected and/or infected DCs were washed once in FACS staining buffer (PBS containing 1% FBS) prior to confirmation of their immature phenotype with a panel of FITC-conjugated monoclonal antibodies against CD40, CD80, CD83, CD86, DC-SIGN (Beckman-Coulter), as well as for intracellular DENV antigen. As a control, the appropriate isotype was also utilized. Cells were allowed to incubate on ice for 30 min followed by washing twice with FACS staining buffer to remove unbound antibody. Cells were then fixed in buffered 2% paraformaldehyde (pH 9.5) and analyzed with a FACScan flow cytometer (Becton Dickinson) within 24 h of staining. Prior to fixation, cells that had been probed for DENV antigen were subjected to secondary staining with FITC-conjugated antibody (Molecular Probes) and washed twice with FACS staining buffer.

SCID mice

Ten 5–6-week-old SCID mice (Tac:Icr:Ha[ICR]-Prkdc^scid; Taconic Farms) per group were xenografted intraperitoneally (i.p.) with 1 × 10⁷ Huh-7 cells suspended in 0.2 ml phosphate-buffered saline. Tumors were detected 5–6 weeks later by palpation, and mice were infected by direct injection into the tumor mass with 1 × 10⁴ focus-forming units (ffu) of virus diluted in 50 ml of PBS. Seven days post-infection, the mice were sacrificed and blood was isolated by cardiac puncture and the serum stored at −80 °C. Virus titers were determined by FIA in C6/36 cells. Mice that failed to develop viremia were excluded from the experiment. Prior to implantation, Huh-7 cells were certified to be free of mycoplasma, human and mouse viruses using PCR or RT-PCR (Taconic Anmed, Rockville). Animal experimentation was approved by the UTMB IACUC and mice were maintained in gnotobiotic isolators on specific pathogen-free environment.

Statistical analyses

GraphPad Prism version 4 software (GraphPad Software) was used for data analyses. Virus outputs from 11 ex vivo or 2 in vivo experiments were grouped by virus strain and compared by one-way ANOVA. The Tukey–Kramer post hoc test was used to test for differences between pairs of strains. Student’s t-tests were used for two sample comparisons.

RNA extraction and sequencing

Viral RNA was extracted from virus preparations using the QIAamp Viral RNA mini kit (Qiagen). PCR primers designed to amplify the E protein gene (Wang et al., 2000) were used with the Titan one step RT-PCR kit (Roche, Indianapolis). Viral RNA was denatured for 2 min at 70 °C and cDNA was synthesized at 50 °C for 30 min, followed by 35 rounds of amplification in a 50 μl reaction volume. The PCR products were purified from 1% agarose gels and both strands were sequenced directly using an Applied Biosystems (Foster City, California) Prism automated DNA sequencing kit and model 3100 Genetic Analyzer sequencer according to the manufacturer’s protocol.

Phylogenetic analyses

The obtained nucleotide sequences encoding the envelope (E) protein, and representative sequences from the GenBank library, were aligned using the ClustalW multiple sequence alignment program with default gap penalties. Phylogenetic analyses of the aligned nucleotide sequences were performed using Bayesian analysis with 1 million reiterations and/or maximum likelihood, neighbor joining and maximum parsimony methods implemented in the PAUP 4.0 software package (Swofford, 1998). Homologous nucleotide sequences from DENV-1 and -3 were used as an outgroup to root the DENV-2 tree. Bootstrapping with 1000 replicates was used to place confidence values on grouping within the tree (Felsenstein, 1985).

NV was supported by the Centers for Disease Control and Prevention Fellowship Training Program in Vector-Borne Infectious Diseases, T01/CCT622892. KAH was supported by a NM-INBRE (P20 RR016480-05) grant from the National Institutes of Health. We thank Robert Tesh and Hilda Guzman for kindly providing DENV strains and antisera. Gerald Kovacs and Mark Endsley for their critical comments on the manuscript. Two anonymous referees also provided valuable comments.