Abstract
We recently described the derivation of a dengue serotype 2 virus (DEN2mutF) that exhibited a host range-restricted phenotype; it was severely impaired for replication in cultured mosquito cells (C6/36 cells). DEN2mutF virus had selected mutations in genomic sequences predicted to form a 3′ stem-loop structure (3′-SL) that is conserved among all flavivirus species. The 3′-SL constitutes the downstream terminal ∼95 nucleotides of the 3′ noncoding region in flavivirus RNA. Here we report the introduction of these same mutational changes into the analogous region of an infectious DNA derived from the genome of a human-virulent dengue serotype 1 virus (DEN1), strain Western Pacific (DEN1WP). The resulting DEN1 mutant (DEN1mutF) exhibited a host range-restricted phenotype similar to that of DEN2mutF virus. DEN1mutF virus was attenuated in a monkey model for dengue infection in which viremia is taken as a correlate of human virulence. In spite of the markedly reduced levels of viremia that it induced in monkeys compared to DEN1WP, DEN1mutF was highly immunogenic. In addition, DEN1mutF-immunized monkeys retained high levels of neutralizing antibodies in serum and were protected from challenge with high doses of the DEN1WP parent for as long as 17 months after the single immunizing dose. Phenotypic revertants of DEN1mutF and DEN2mutF were each detected after a total of 24 days in C6/36 cell cultures. Complete nucleotide sequence analysis of DEN1mutF RNA and that of a revertant virus, DEN1mutFRev, revealed that (i) the DEN1mutF genome contained no additional mutations upstream from the 3′-SL compared to the DEN1WP parent genome and (ii) the DEN1mutFRev genome contained de novo mutations, consistent with our previous hypothesis that the defect in DEN2mutF replication in C6/36 cells was at the level of RNA replication. A strategy for the development of a tetravalent dengue vaccine is discussed.
The four serotypes of dengue viruses (DEN1, DEN2, DEN3, and DEN4) cause dengue fever (DF) and dengue hemorrhagic fever or dengue shock syndrome (DHF/DSS). The incidences of both DF and DHF/DSS are rising worldwide (10), and there is currently no vaccine available to prevent the spread of dengue or reduce the incidence of disease. A live virus vaccine is desirable for a variety of reasons, including reduced cost in comparison to that of subunit vaccines, the possibility for single-dose vaccination, and the need to elicit long-term immunity. Vaccine development has been hampered thus far by the lack of any animal model for DF or DHF/DSS and the perceived need to elicit a protective response to all four serotypes of dengue virus concurrently (10, 16, 27).
The dengue viruses are members of the Flavivirus genus. Flaviviruses contain a positive-strand RNA genome that is ∼11 kb long and includes a single open reading frame (ORF) encoding a polyprotein which is processed co- and posttranslationally to yield three structural and at least seven nonstructural (NS) proteins (6). The ORF is flanked at the 5′ and 3′ termini of the genome by noncoding regions (NCRs). Conserved stem-loop (SL) structures are predicted to occur in both 5′- and 3′-NCRs in viral RNA (5, 9, 15, 22, 26, 32, 36, 38). The most thermodynamically stable of these predicted SLs is found at the extreme 3′ terminus of the 3′-NCR, formed by the last 90 to 100 nucleotides (nt) of the genome, here referred to as the 3′-SL (3, 4, 9, 36).
Growth characteristics were previously reported for a set of DEN2 mutants that were generated from an infectious DNA clone (24) in which portions of the wild-type (wt) DEN2 3′-SL nucleotide sequence were replaced by analogous portions of the West Nile virus (WN) 3′-SL nucleotide sequence (39). One of these mutant viruses (D2/WN-SLmutF, referred to as DEN2mutF in this report) was markedly defective for replication in cultured Aedes albopictus mosquito cells (C6/36 cells). In contrast, it replicated to titers comparable to those of the wt DEN2 parent in a continuous line of monkey kidney cells (LLCMK2 cells). Both cell types are among standard laboratory substrates for both dengue virus and WN (strain Eg101) replication (unpublished data). Here we report the introduction of the nucleotide sequence changes that differentiated DEN2mutF from its wt DEN2 parent into an infectious DNA for a human-virulent DEN1 virus, strain Western Pacific 74 (DEN1WP). The resulting DEN1 mutant virus (DEN1mutF) displayed a restriction phenotype in C6/36 cells similar to that of DEN2mutF. DEN1mutF virus was evaluated for its potential as a vaccine in the rhesus monkey model and was found to be attenuated and highly immunogenic. The neutralizing antibody response in DEN1mutF-immunized monkeys was still detectable 17 months after a single dose, and these monkeys were protected from DEN1 viremia after challenge.
Previous in vivo (39) and in vitro (38) studies suggested that DEN2mutF virus was defective for RNA replication in C6/36 cells. Virus-coded activities required for flavivirus RNA replication are localized to NS1, NS3, and NS5 (7, 8, 17, 19, 20, 23, 33, 34). The functional role for NS1 is not defined. NS3 contains RNA helicase and NTPase activities (8, 19). NS5 contains RNA-dependent RNA polymerase activity at its carboxy terminus (33) and a putative methyltransferase domain at its amino terminus (17a). Indirect evidence suggests that the 3′-SL may form a complex with both cellular proteins and NS5 that is required for RNA replication (3, 4, 5, 38). Nucleotide sequence analysis of the DEN1mutF genome and of the genome of a phenotypic revertant, DEN1mutFRev, revealed that the DEN1mutFRev genome contained amino acid substitution mutations in the NS1 and NS5 gene sequences with respect to the DEN1mutF genome. A third point mutation was localized to the 3′-SL. The loci of induced mutations associated with phenotypic reversion were consistent with the previous results related to the apparent defect in RNA replication of DEN2mutF in C6/36 cells.
MATERIALS AND METHODS
Cells and viruses.
LLCMK2 cells (American Type Culture Collection [ATCC], Manassas, Va.) are a continuous line of monkey kidney cells. LLCMK2 cell monolayers were maintained in Eagle's minimal essential medium (Invitrogen, Carlsbad, Calif.) plus gentamicin and 10% fetal bovine serum (FBS; Gibco BRL) in a humidified incubator at 37°C under 5% carbon dioxide. C6/36 cells (ATCC) are derived from A. albopictus mosquito larvae and were maintained as monolayers in Eagle's minimal essential medium supplemented with gentamicin, 10% FBS, nonessential amino acids, sodium pyruvate, and 25 mM HEPES (pH 7.55) in sealed flasks at 28 to 30°C. The origin of mouse brain-adapted DEN2, strain New Guinea C (DEN2NGC), the construction of pRS424/DEN2NGC recombinant DNA, and the creation of DEN2mutF were all previously described (24, 39). DEN1WP was a clinical isolate that had been passaged on C6/36 cells approximately three times prior to the derivation of the DEN1WP infectious DNA (25).
Rhesus macaques.
Rhesus macaques used in this study were bred in a closed colony on Morgan Island, S.C., and purchased from LABS of Virginia (Yesmassee, S.C.) as juveniles (approximately 1 year old). They were maintained at the Center for Biologics Evaluation and Research (Bethesda, Md.) animal care facility prior to and during this study in cages containing not more than two monkeys each. Animals were 2 to 4 years old at entrance into the study and had been screened for previous exposure to flaviviruses by an assay for the presence of flavivirus-specific hemagglutination-inhibiting antibodies in serum samples.
Derivation of DEN1mutF.
Plasmid pRS424 recombinant DNA containing an infectious DNA copy of the DEN1WP genome (25) was used as a basis for constructing DEN1mutF. The nucleotide sequence of the DEN1WP RNA genome that was used to derive the infectious DNA and that of the DEN1WP infectious DNA were previously determined, and the DEN1WP DNA sequence was shown to differ from the average sequence of the DEN1WP RNA at five nucleotide positions (25). Three of these substitutions were predicted to result in amino acid changes. Plasmid pRS424 is a yeast-bacterium shuttle vector used routinely in our laboratory for the assembly and mutagenesis of full-length infectious flavivirus DNAs in yeast cells (24, 39).
The mutF mutations (Fig. 1) were introduced into the DEN1WP infectious DNA by homologous recombination between a linear PCR fragment bearing the mutF mutations and linearized pRS424/DEN1WP recombinant DNA in yeast cells. The left primer (primer A) used to generate the mutF PCR fragment was complementary to wt DEN1WP cDNA. Its 5′ terminus initiated about 50 nt upstream from the most 5′-terminal nucleotide sequences of the mutF mutations in the DEN1WP 3′-SL (beginning at nt 74 in the 3′-SL; Fig. 1) and encoded the C-74/U substitution mutation and the A-73 deletion mutation characteristic of the mutF genotype. Primer A was complementary to the DEN1WP 3′-SL for an additional 18 nt downstream from the A-73 deletion mutation. The right primer (primer B) was antisense to the DEN1 genome. It was tripartite. (i) Its first domain was complementary to pRS424 sequences for approximately 50 nt, extending toward the 3′ terminus of the DEN1WP cDNA in pRS424/DEN1WP recombinant DNA. (ii) Its second domain included the unique SacII site at the 3′ terminus of the DEN1 DNA in pRS424/DEN1 recombinant DNA and was complementary to the DEN1 3′-terminal mutF mutations (the U-4/A substitution and the A-7 deletion). (iii) Its third domain was complementary to the wt DEN1WP nucleotide sequence for approximately an additional 18 nt upstream from the A-7 deletion. A PCR product was generated with primers A and B and pRS424/DEN1WP recombinant DNA as a template by using a model 2400 GeneAmp PCR system (Applied Biosystems, Foster City, Calif.) and Pfu DNA polymerase (New England BioLabs, Beverly, Mass.).
FIG. 1.
Nucleotide sequence and predicted conformation (3, 4, 9, 36, 39) of the DEN2NGC 3′-SL. Nucleotides are numbered in reverse order, starting at the 3′ terminus of the dengue virus genome. The nucleotide sequence of the DEN1WP 3′-SL was identical to that of DEN2NGC, except as indicated for nt 63, 64, and 87, where nucleotides present in the DEN1WP 3′-SL are shown in parentheses adjacent to the replaced nucleotides in the DEN2NGC 3′-SL. The hatched rectangle enclosing nt A-7 and U-73 indicates that this base pair is deleted in DEN1mutF and DEN2mutF cDNAs and the respective genomic RNAs. Likewise, nucleotides U-3 and C-74, shown in bold italic type, are substituted, as indicated by arrows, in DEN1mutF and DEN2mutF genomes. The asterisk indicates a spontaneous substitution mutation to A that occurred at U-12 in DEN1mutFRev genomic RNA. The plus sign indicates the presence of the remaining ∼10.4 kb of the DEN1WP or DEN2NGC genome upstream from the 3′-SL in mutant genomes. All nucleotide sequence data were obtained as previously described by direct nucleotide sequencing of a PCR product derived from the 5′-3′ junction of circularized viral RNA (39) (see Materials and Methods).
To introduce the mutF mutations contained in the PCR product into DEN1WP DNA, pRS424/DEN1WP recombinant DNA was digested with restriction endonuclease SacII, which cleaved this DNA uniquely at the 3′ terminus of DEN1WP DNA sequences. Tryptophan-negative yeast YPH857 cells were cotransformed with the linearized recombinant DNA and the PCR product DNA generated by using primers A and B to generate DEN1mutF recombinant DNA in yeast cells. Yeast DNA from colonies formed on tryptophan-negative plates was used to transform STBL2 bacteria (Invitrogen) in order to obtain DNA for transcription. These procedures and the rationale were previously described in greater detail (24, 39). The presence of the mutF mutational changes in plasmid DNA isolated from STBL2 bacterial colonies was verified by DNA sequencing by an automated dye terminator method (Applied Biosystems). In vitro transcription of DEN1mutF RNA was catalyzed by SP6 RNA polymerase (Promega Corporation, Madison, Wis.) in the presence of a cap analogue, using supplied buffers and SacII-linearized pRS424/DEN1mutF DNA as a template (25). DEN1mutF was generated by electroporation of the RNA transcripts into LLCMK2 cells by using a Gene Pulser II (Bio-Rad, Hercules, Calif.).
Nucleotide sequencing of DEN1WP, DEN1mutF, and DEN1mutFRev genomic RNAs.
The technique for sequencing the 3′ termini of viral genomic RNAs was previously described (39). Briefly, viral RNA was purified from about ∼106 PFU of virus suspended in tissue culture medium by using instructions supplied with the QIAamp viral RNA minikit (Qiagen, Valencia, Calif.). The 5′-terminal cap structure was removed from purified viral RNA, using the enzyme tobacco acid pyrophosphatase (Epicentre Technologies, Madison, Wis.). Decapped 5′ and 3′ termini of viral RNA were ligated using the enzyme T4 RNA ligase (Epicentre Technologies). A reverse transcription (RT)-PCR product spanning the 5′-3′ junction and including all of the 3′-SL nucleotide sequence in circularized RNA was generated using appropriate DEN1-specific primers: one genomic negative-sense primer complementary to the 5′ terminus of the genome and one positive-sense primer containing sequences upstream from the 3′-SL. Sequencing of the resulting linear PCR product was primed by oligonucleotides complementary to its 5′ and 3′ termini. The dye terminator cycle sequencing method was carried out with a DNA sequencing kit (PE Biosystems, Warrington, England) and an ABI Prism 377 automated DNA sequencer (Applied Biosystems).
To obtain complete nucleotide sequences of genomic RNAs, a full-length cDNA copy of each RNA was first generated using the enzyme reverse transcriptase and a 15- to 20-nt primer complementary to the 3′-terminal nucleotide sequence of each RNA, as directly determined by nucleotide sequencing in the previous procedure. Viral RNAs were prepared from suspensions of virus in tissue culture medium, using the QIAamp viral RNA minikit as described above. Superscript II reverse transcriptase (Invitrogen) was used. Complementary DNAs were sequenced in overlapping 1- to 1.5-kb segments by using appropriate primers, which were designed based on the previously determined nucleotide sequence of DEN1WP cloned DNA (25). Fragments were sequenced using the procedures and equipment described above.
PRNT assay.
The plaque reduction neutralization (PRNT) assay was used to measure titers of DEN1-specific neutralizing antibodies in monkey sera. Serum samples were collected from monkeys in all experiments described on days 0 through 14 and again on day 30, where day 0 was defined as the day of virus inoculation. Approximately 0.5 ml of a serum sample to be used in this assay was first heat inactivated by incubation for 30 min at 56°C. Fourfold dilutions of serum in a final volume of 0.3 ml, starting at a 1:10 dilution, were then prepared using medium M199 with 2% heat-inactivated FBS as a diluent. To each 0.3-ml aliquot of diluted serum was added an equal volume of medium containing 150 to 180 PFU of DEN1WP. Virus and serum were mixed and incubated for 30 min at 37°C. In each assay, a no-serum control and controls consisting of DEN1-specific mouse hyperimmune ascitic fluid (ATCC) at two dilutions were also included. Virus-serum and control mixtures were plated on confluent monolayers of LLCMK2 cells in Costar six-well tissue culture plates (Corning Inc., Corning, N.Y.) at 0.2 ml/well. This procedure resulted in about 50 plaques per well in antibody-negative control wells. Duplicate wells were infected for each sample. Virus adsorption was carried out for 1 h at room temperature with manual rocking of the plates every 15 min. Wells were then overlaid with medium containing 1% agarose (SeaKem LE; BioWhittaker, Rockland, Maine) in Earle's balanced salt solution-10% FBS with added essential vitamins and amino acids (all solutions were purchased from Invitrogen) at 6 ml/well. Plates were incubated for 7 to 8 days at 37°C in 5% carbon dioxide. Wells were then overlaid with 12% neutral red solution (12 ml of neutral red solution [Invitrogen] plus 88 ml of sterile deionized water) containing 1% agarose. Plates were incubated for 24 h at 37°C. Plaques were counted, and counts in duplicate wells were averaged before calculation of the dilution at which a 50% reduction in plaque number was achieved (PRNT50). This dilution was taken as the titer of neutralizing antibodies in a specimen.
Detection of virus in monkey sera by plaque formation in LLCMK2 cells.
Monkey serum samples from days 0 through 11 were used to infect confluent LLCMK2 cell monolayers in duplicate wells of Costar six-well tissue culture plates; serum samples were undiluted and diluted 1:10 and 1:100 using medium M199 with 2% FBS as a diluent. Subsequent procedures for elucidation of plaques were conducted exactly as described above for the PRNT assay. Plaque counts in duplicate wells were averaged to determine the titer in PFU of virus per milliliter in a given serum sample.
Determination of titers of virus in monkey sera by RT-PCR.
Determination of titers of virus in monkey sera by RT-PCR was carried out according to a published protocol (13, 14) with sera collected on days 0 to 14 of each experiment. Initially, a negative-sense primer with respect to the ORF, DV1.L1, complementary to 3′-terminal nt 24 to 45 (numbering from the 3′ terminus of the genome) in DEN1 RNA (14), was used to generate DEN1 cDNA in an RT reaction; in this reaction, the substrate was viral RNA extracted from serum samples or control virus, and the QIAamp viral RNA minikit was used. Primer DV1.L1 and a positive-sense primer, DV1.U1, complementary to upstream nt 175 to 151 (also with respect to the 3′ terminus), were used to generate a PCR product in a reaction that included an oligonucleotide probe (DV1.P1) spanning DEN1 nt 71 to 47. This probe was labeled with fluorescein at its 5′ terminus and with a fluorescence quencher at its 3′ terminus, such that its incorporation into the PCR product generated from primers DV1.L1 and DV1.U1 was predicted to liberate a fluorescence signal. In previous work (14), a correlation between the intensity of the fluorescence signal and the PFU of DEN1 present in a sample was established. The technique was also shown to distinguish among DEN1, DEN2, DEN3, and DEN4 and to distinguish these viruses from other flaviviruses. In the present research, we separately demonstrated that RT-PCR was equally sensitive for the detection of DEN1WP and DEN1mutF.
For a blinded set of samples of each virus, RT-PCR estimation of infectious virus concentration agreed with data derived from direct plaque counts within ±100.5 (data not shown). The lower limit of sensitivity of this assay was calculated to be 0.03 PFU/1.6 μl of serum or approximately 20 PFU/ml, where the number of PFU was determined by actual plaque formation on LLCMK2 cells. Results of the RT-PCR are expressed as PFU equivalents (PFU eq). Samples were tested in triplicate. If at least two of three replicates of a sample tested positive at the low threshold, then that sample was scored as containing 20 PFU of virus/ml. A TaqMan RT kit (Applied Biosystems) was used for RT reactions as previously described (13). AmpliTaq Gold DNA polymerase (Applied Biosystems) was used for PCRs. Gene detection system 7700 and sequence detection system software, version 1.6.3 (Applied Biosystems), were used for real-time data collection and analysis.
Sequencing of viral genomes detected in monkey sera by RT-PCR.
Sera obtained from monkeys KPJ (0.45-ml volume), KJP (0.75 ml), and HJX (0.35 ml) on day 6 were used for this analysis. Each sample was predicted to contain 400 PFU eq of viral genomes/ml. Virus was precipitated from the samples by using polyethylene glycol. An amount of a solution of 14% polyethylene glycol-0.8 M NaCl equal to the volume of each serum sample was added to the samples, and the mixtures were incubated overnight at 4°C. Precipitated virus was recovered by centrifugation. Pellets were resuspended in phosphate-buffered saline (PBS), and viral RNA was extracted from each sample by using a QIAamp viral RNA minikit column. Thirty-six nanograms of an antisense 20-nt primer specific for the 3′-terminal nucleotide sequence of DEN1mutF (nt 1 to 20; Fig. 1) was annealed to total viral RNA in 20 μl of water, and the total product was subjected to RT for 3 h at 42° by using the enzyme Superscript II in a final volume of 50 μl of 1× First-Strand buffer (Invitrogen).
To demonstrate the synthesis of DEN1mutF cDNA by RT, 20-nt sense and antisense primers derived from the published DEN1 genome sequence (25) and with 3′ termini at nt 2182 and 3180, respectively, were used in PCRs with the DNA polymerase Expand (Roche, Indianapolis, Ind.). Thirty-five rounds of synthesis were conducted with a model 2400 GeneAmp PCR system. To obtain nucleotide sequence data from the 3′ terminus of the DEN1mutF genome, seminested PCR was required. In the first reaction, the 3′ terminus of the sense primer corresponded to nt 10179 of the DEN1 nucleotide sequence (25), and the antisense primer corresponded to the 3′-terminal 20 nt of the DEN1mutF genome (nt 1 to 20; Fig. 1), as in the RT reaction. The total cDNA from the RT reaction was concentrated in a volume of 5 μl, and 2 μl of this solution was used as a template. No PCR products were detectable from any of the three cDNAs after 40 rounds of synthesis. Two microliters of the 50-μl total volume of the first-round PCR was used as a template for a second round of amplification. The 20-nt sense primer corresponded to the published DEN1 nucleotide sequence, with its 3′ terminus at nt 10480, and the antisense primer was the same as that used in the first round. Sequencing of the 255-bp DNA product of the second round of amplification was carried out as described above.
RESULTS
Derivation of DEN1mutF virus.
DEN2mutF DNA and RNA isolated from DEN2mutF virus that replicated in LLCMK2 cells contained a substitution of the bottom-most 6 bp in the long stem of the wt WN 3′-SL for the analogous 7-bp region of the wt DEN2 3′-SL (Fig. 1). This change resulted in (i) deletion of an A-U base pair in the long stem of the wt DEN2 3′-SL, formed by A at position 7 (A-7) and U at position 73 (U-73); (ii) replacement of U at position 4 by an A; and (iii) replacement of C at position 74 by a U (nucleotide numbers start from the 3′ terminus of the wt DEN2 genome sequence) (15, 24). The mutF mutation had the effect of eliminating a non-base-paired U-U “bulge” in the long stem of the wt DEN2 3′-SL (involving U at position 4), converting it to an A-U base pair, and of reducing the length of the long stem by 1 bp (the deleted A-7/U-73 pair).
We sought to investigate the possibility that additional dengue viruses with the “mutant F” host range-restricted phenotype could be created simply by introducing the mutational changes in the DEN2mutF 3′-SL into the analogous domain in infectious DNAs for other dengue virus genomes. Accumulated nucleotide sequence data for several dengue virus genomic RNAs showed that the nucleotide sequence of the 3′-SL is highly conserved among dengue virus serotypes. For example, Fig. 1 illustrates the homology of the nucleotide sequences of the 3′-SLs in the DEN1WP and DEN2NGC genomes. The sequences are identical at all but three nucleotide positions. An infectious DNA for the genome of a human-virulent DEN1 isolate, DEN1WP, had previously been derived (25), and the mutF mutations were introduced into this DNA by site-directed mutagenesis via homologous recombination in yeast cells (24, 30, 39). RNA derived by in vitro transcription of linearized DEN1mutF DNA was infectious in LLCMK2 cells, and the sequence of the 3′-SL in genomic RNA isolated from DEN1mutF was shown to conform to that encoded by DEN1mutF DNA (Fig. 1).
The kinetics of replication of DNA-derived wt DEN1WP and DEN1mutF were compared in LLCMK2 cells and in C6/36 cells, using pools of both viruses prepared in LLCMK2 cells as inocula (Fig. 2). The growth of the two viruses in LLCMK2 cells was similar, although the peak titer after 8 days was slightly lower (by about 5-fold, or 0.7 log10 unit) (Fig. 2A) for DEN1mutF than for the wt. In repeat experiments, the mutant and wt viruses replicated to more nearly identical peak titers in LLCMK2 cells (data not shown). In contrast, DEN1mutF was severely impaired for replication in C6/36 cells (Fig. 2B). In several repeat experiments, no more than 100 PFU/ml of DEN1mutF virus could be detected in the medium after 8 or 9 days of incubation of infected C6/36 cells, suggesting that DEN1mutF had a defect in replication in C6/36 cells similar to that demonstrated for DEN2mutF virus (39). We tentatively concluded that the mutational changes in the conserved 3′-SL in DEN2 RNA associated with the DEN2mutF phenotype might confer a similar phenotype on other dengue viruses with disparate serotypes and passage histories.
FIG. 2.
Growth of DEN1mutF in LLCMK2 cells (A) and C6/36 cells (B). DEN1mutF was derived by transfection of LLCMK2 cells with RNA transcribed in vitro by using DEN1mutF cDNA as a template. A high-titer pool of virus was generated by further passage in LLCMK2 cells, and this was used to infect either LLCMK2 cells or C6/36 cells at a multiplicity of infection of approximately 0.05. Samples of supernatants from LLCMK2 and C6/36 cell cultures were obtained daily for 8 days, and the titer for each sample was determined by plaque formation on LLCMK2 cells.
DEN1mutF virus is attenuated and immunogenic in rhesus monkeys.
The potential usefulness of DEN1mutF virus as a live, attenuated dengue vaccine was assessed in rhesus monkeys. Monkeys do not become detectably ill after dengue infection, but peak titers of virus in serum and the duration of viremia in monkeys appear to be related to virulence in humans (1, 8a, 8b, 18, 29; D. W. Vaughn, K. H. Eckels, D. R. Dubois, R. Edelman, C. O. Tacket, P. L. Summers, R. Rice, E. Kraiselburd, and C. H. Hoke, Abstr. XIIIth Int. Congr. Trop. Med. Malaria, vol. 2, p. 229, th02-2, Pattaya, Thailand, 29 November to 4 December 1992). The induction of dengue virus-specific neutralizing antibodies was taken as a correlate of both infectivity and immunogenicity. We chose to compare wt DEN1WP and DEN1mutF viruses in this study because the DEN1WP parent had been derived from a human isolate and could be expected to infect monkeys efficiently. In contrast, the prototype DEN2mutF virus had a mouse-brain-adapted phenotype in parallel with that of its parent, DEN2NGC (data not shown), that might correlate with attenuation in monkeys, since earlier work suggested that mouse-brain-adapted dengue viruses were attenuated in humans (28).
Eighteen 2- to 3-year-old flavivirus-nonimmune rhesus macaques were divided into two groups of nine monkeys each to receive either DEN1WP or DEN1mutF. Subgroups of three monkeys each received 103, 104, or 105 PFU of one or the other virus. Monkey serum samples were collected daily for 14 days and on day 30 after infection and saved for detection and titration of virus and for the determination of neutralizing antibody titer in a PRNT assay. Viremia was monitored by a validated quantitative dengue virus-specific RT-PCR assay (13, 14) with TaqMan (Table 1).
TABLE 1.
Infectivity and immunogenicity for wt DEN1WP and DEN1mutF
| DEN1 virus (dose, PFU)a | Monkey | Viremia determined by RT-PCR
|
PRNT50−1 at day 30 | ||
|---|---|---|---|---|---|
| Total no. of days | Days p.i.b | Peak titer (PFU eq/ml × 103) | |||
| wt (105) | KVC | 4 | 1-4 | 4.2 | 320 |
| KKA | 7 | 1-4, 6, 7, 10 | 0.7 | >320 | |
| KJX | 4 | 1-4 | 0.4 | 160 | |
| wt (104) | KBP | 4 | 2-5 | 2.4 | 160 |
| KWV | 4 | 2-4, 8 | 0.9 | 320 | |
| KGW | 4 | 1-4 | 1.4 | 320 | |
| wt (103) | KPT | 4 | 2-4, 6 | 0.8 | 320 |
| KJD | 5 | 2-6 | 1.5 | 640 | |
| KWK | 5 | 2-5, 11 | 1.4 | 640 | |
| mutF (105) | KXJ | 0 | 640 | ||
| TAP | 0 | 640 | |||
| KWD | 1 | 3 | 0.1 | 160 | |
| mutF (104) | KPJ | 2 | 6, 11 | 0.4 | 80 |
| KJP | 2 | 6, 10 | 0.4 | 160 | |
| KGV | 0 | 160 | |||
| mutF (103) | HJX | 2 | 6, 10 | 0.4 | 80 |
| KGT | 2 | 6, 11 | 0.3 | 80 | |
| KJV | 1 | 11 | 0.2 | 160 | |
Monkeys were infected with the indicated dose of virus in 1.0 ml of saline, injected subcutaneously in the subscapular region.
Days of the study on which viremia was detected when monkeys were infected on day 0. p.i., postinfection.
DEN1mutF induced much less viremia than did the wt parent virus in terms of duration and peak titer. Monkeys that received wt DEN1WP had 41 total days of viremia and tended to be viremic by day 1 or 2 after infection, whereas monkeys that received DEN1mutF had 10 total days of viremia, first detected on day 3 or later (Table 1). DEN1WP-infected monkeys were viremic for a minimum of 4 days, whereas only six of nine DEN1mutF-infected monkeys had any detectable viremia, and none of the latter were viremic for more than two nonconsecutive days. The average number of days of viremia for monkeys that received wt virus was 4.6. In contrast, the average number of days of viremia for monkeys that received DEN1mutF virus was 1.1 (Table 2). Peak titers of virus in sera were also much lower in DEN1mutF-infected monkeys than in controls. The lowest peak titer of virus observed in any DEN1WP-infected monkey (400 PFU eq/ml, as determined by RT-PCR for monkey KJX; Table 1) was equal to the highest peak titer of virus observed in any DEN1mutF-infected monkey, and the average peak virus titer in control sera was more than sevenfold higher than that in sera from monkeys infected with DEN1mutF (Table 2).
TABLE 2.
Average infectivity and immunogenicity of wt DEN1WP and DEN1mutF
| Infecting virus | No. of days viremica | Peak titer (avg PFU eq/ml × 103) [range] | Avg PRNT50−1 at day 30 [range] |
|---|---|---|---|
| wt DEN1WP | 4.6 (4-7) | 1.52 [0.4-4.2] | 320 [160-640] |
| DEN1mutF | 1.1 (0-2) | 0.2 [<0.02-0.4] | 240 [80-640] |
Average (range) for all nine monkeys per group, as determined by RT-PCR.
We attempted to recover infectious DEN1mutF virus from monkey sera in order to determine whether the mutF genotype and in vitro phenotype remained stable during monkey passage, but we were unable to do so. This was likely due both to the very low titers of virus present and to the inhibiting effect of endogenous neutralizing antibodies in sera at later times after infection, when DEN1mutF viremia was usually detected by RT-PCR.
As an alternative to virus isolation, we attempted to obtain nucleotide sequence data from viral genomes directly detected in sera by RT-PCR titration assays. Serum samples obtained from monkeys KJP, KPJ, and HJX on day 6 were used as a source of viral genomes, because these samples were predicted by RT-PCR to contain the highest titers of virus (400 PFU eq/ml in each sample; Table 1). RT was conducted by using total virus RNA isolated from each serum sample and a primer specific for 3′-terminal genomic sequences in the DEN1mutF genome (nt 1 to 20; Fig. 1) in three separate reactions. We were able to generate PCR products representing nt 2812 to 3180 of the DEN1 genome when each cDNA was used as a template in a PCR with DEN1-specific primers (data not shown). This result demonstrated that full-length or nearly full-length DEN1mutF cDNAs were synthesized in each of the three RT reactions. Unfortunately, we were unsuccessful in generating PCR products representing the entirety of the genome from any of the cDNAs, in order to perform a complete nucleotide sequence analysis, probably due to the paucity of template molecules. However, by using 10-fold-concentrated cDNA from the RT reactions and seminested PCR, we succeeded in generating DNA representing nt 10480 to 10714 (including all but the most 3′-terminal 21 nt of the 3′-SL) from one of the three cDNAs (derived from monkey KPJ serum). This fragment included the left strand of the long stem in the 3′-SL, where mutF mutational changes are partially localized (nt 72 through 79 in Fig. 1). Sequence results for this entire fragment showed no mutations with respect to the expected sequence of the DEN1mutF genome. We infer that DEN1mutF did not revert to a virulent phenotype in monkeys, because we would have expected a reversion event to have been heralded by wt levels of viremia occurring late after infection. This situation did not occur.
Sera were analyzed for DEN1-specific neutralizing antibodies on days 0, 7, 14, and 30. None of the monkeys had DEN1-specific neutralizing antibodies prior to infection on day 0, and sera collected on days 7 and 14 contained only undetectable to low titers of neutralizing antibodies (data not shown). However, all 18 monkeys developed a neutralizing antibody response by 30 days after infection and were therefore considered to have been infected. Thus, even the lowest dose of either virus (103 PFU) was likely equal to or greater than the 50% monkey infectious dose (MID50) for that virus (Table 1). Antibody titers in the two groups of monkeys were comparable (Table 2), in spite of the marked reduction in the incidence of viremia and peak titers of virus in sera of DEN1mutF-infected monkeys compared to controls. For example, the maximum antibody titer that we measured (1:640) was detected in two of the nine monkeys in both the DEN1mutF- and the DEN1WP-infected groups. Also, the minimum antibody titer in the DEN1mutF-infected monkeys (1:80) was only twofold lower than the minimum antibody titer detected in the DEN1WP-infected monkeys (1:160). In general, there was a dissociation between the ability to induce viremia and immunogenicity for DEN1mutF compared to its human-virulent parent virus (for examples, see especially data for monkeys KXJ and TAP in Table 1). DEN1mutF was attenuated and immunogenic in rhesus monkeys.
DEN1mutF-infected monkeys resist challenge by wt DEN1WP.
Two challenge studies were conducted to determine the efficacy of DEN1mutF single-dose immunization in protecting monkeys from viremia. In the first of these studies, conducted 12 months after DEN1mutF immunization, three monkeys (KPJ, KJP, and KGV) from the group that received 104 PFU of DEN1mutF virus were challenged with 104 PFU of wt DEN1WP. Three flavivirus-seronegative monkeys were infected with the same inoculum and served as controls. We estimated that the challenge inoculum was equivalent to 10 to 100 MID50 of virus, sufficient to infect 100% of monkeys. This estimate was based on the fact that 103 PFU of DEN1WP infected all three monkeys to which that dose had been administered in the course of the original study (Tables 1 and 2).
Neutralizing antibody titers were determined by using the PRNT assay as described above. All three control monkeys were infected with the challenge virus, as judged by the development of a DEN1-specific neutralizing antibody response (Table 3). Neutralizing antibodies were not detectable on days 0 and 7 after infection and were first detected on day 14 in these controls. The titers of neutralizing antibodies on day 30 were comparable to those observed for the DEN1WP-infected monkeys in the initial trial (PRNT50s were 1:640, 1:640, and 1:320 for monkeys CE6G, CE68, and 810, respectively).
TABLE 3.
Results of challenge study 1 for titers of neutralizing antibodies in sera of monkeys challenged with 104 PFU of wt DEN1WP
| Monkey | PRNT50−1 of DEN1 antibodies on challenge dayb:
|
||||
|---|---|---|---|---|---|
| Pre | 0 | 7 | 14 | 30 | |
| CE6G | NA | <10 | <10 | 160 | 640 |
| CE68 | NA | <10 | <10 | 160 | 640 |
| 810 | NA | <10 | <10 | 40 | 320 |
| KPJc | 160 | 80 | 640 | 10,240 | 10,240 |
| KJPc | 80 | 40 | 1,280 | 10,240 | 5,120 |
| KGVc | 80 | 640 | 1,280 | 1,280 | 640 |
Pre, antibody titer at conclusion of initial study of infectivity of DEN1mutF, 12 months prior to challenge. NA, not applicable.
Monkeys received ∼104 PFU of wt DEN1WP subcutaneously.
Received ∼104 PFU of DEN1mutF 12 months prior to challenge.
To determine the stability of the neutralizing antibody response during the 12-month interval between the infectivity-immunogenicity study of DEN1mutF and the challenge study, sera from day 30 of the former study were tested in the same assay with sera obtained from DEN1mutF-immunized monkeys during the challenge study (Table 3). Results showed that the neutralizing antibody response in all three previously immunized monkeys had endured for the 1-year interval between immunization and challenge, even though only a single dose of DEN1mutF had been administered originally. In one monkey (KGV), the antibody titer had apparently continued to rise (from 1:80 to 1:640) during the time that elapsed between the two studies. Two of the DEN1mutF-immunized monkeys developed a marked anamnestic response upon reexposure to DEN1. Monkey KPJ serum had a neutralizing antibody titer of 1:10,240 on days 14 and 30 (a 128-fold rise compared to the day 0 antibody titer), and monkey KJP serum had titers of 1:10,240 and 1:5,120 on the same study days, respectively (a 64-fold rise in titer on day 30 compared to day 0). In contrast, monkey KGV exhibited only a transient twofold rise in the titer of DEN1 neutralizing antibodies, observed on days 7 and 14, during the course of the challenge study.
The quantitative RT-PCR assay was used for the detection and titration of virus in serum samples. Two of the three control monkeys each developed viremia with a duration of 2 days, detected on days 3 and 4 (for monkey 810) and on days 5 and 6 (for monkey CE68) of the study (Fig. 3). Peak titers were 1,900 and 250 PFU eq/ml for monkeys 810 and CE68, respectively. In contrast to results obtained with control monkeys, all of the DEN1mutF-immunized monkeys were protected from viremia after challenge (Fig. 3). Further, the absence of both viremia and a true anamnestic antibody response in monkey KGV (Table 3) suggested that no or very little virus replication occurred in this animal after challenge with DEN1, perhaps due to the high titer of neutralizing antibodies that we detected on day 0 prior to challenge.
FIG. 3.
Viremia in DEN1mutF-immunized monkeys versus controls after challenge with 104 PFU of DEN1WP (challenge study 1). The challenge inoculum was delivered by subcutaneous injection in the subscapular region on day 0 of the study. Virus titers in individual serum samples from all monkeys were determined for serum samples obtained on days 0 through 11 by using a previously validated RT-PCR method (13, 14) that was predicted to detect reproducibly about 20 PFU of virus/ml in serum.
To summarize the results of the first challenge study, two of three flavivirus-naive monkeys developed viremia, and none of three previously immunized monkeys developed viremia, but by the criterion of an immune response, all monkeys were infected. While the data suggested that DEN1mutF-immunized monkeys were protected from viremia, the difference in outcome between controls and DEN1-immunized monkeys was not statistically significant. To obtain additional confirmatory evidence that DEN1mutF immunization was protective against viremia, a second challenge study was conducted, 17 months postimmunization. In this second study, two additional DEN1mutF-immunized monkeys were challenged with DEN1WP virus along with four flavivirus-naive control monkeys. One of the challenged monkeys (KWD) had previously received 105 PFU of DEN1mutF, and the other (KGT) had previously received 103 PFU of DEN1mutF.
A high dose of challenge virus (106 PFU) was chosen in the second study to ensure that all flavivirus-naive controls would develop viremia. Also, evidence for viremia was sought both by a direct plaque assay for virus in serum samples and by RT-PCR (Tables 4 and 5 and Fig. 4). All control monkeys were infected on the basis of both criteria. Sera from monkeys 642, CE8B, and CE8C had neutralizing antibody titers of 1:640 by day 30 after infection, and serum from monkey CE40 had a titer of 1:320 at the same time point (Table 4). These results for flavivirus-naive monkeys were comparable to those noted previously after infection with DEN1WP. Each of these four control monkeys was also viremic for at least 2 and as many as 5 days (Table 5 and Fig. 4). In general, RT-PCR gave slightly higher titers of virus than the plaque assay, but the two assays were similarly sensitive for the detection of any virus in serum. Both methods gave an average of 3.25 days of viremia for the control monkeys in this challenge study. The occurrence of viremia as early as the first day after infection in controls in this study might have been due to the rapid dissemination of virus after the high challenge dose.
TABLE 4.
Results of challenge study 2 for titers of neutralizing antibodies in sera of monkeys challenged with 106 PFU of DEN1WP
| Monkey | PRNT50−1 of DEN1 antibodies on day:b:
|
||
|---|---|---|---|
| Pre | 0 | 30b | |
| 642 | NA | <10 | 640 |
| CE8B | NA | <10 | 640 |
| CE8C | NA | <10 | 640 |
| CE40 | NA | <10 | 320 |
| KWDc | 320 | 40 | 5,120 |
| KGTd | 320 | 20 | 5,120 |
Pre, antibody titer at conclusion of initial study of infectivity of DEN1mutF, 17 months prior to challenge. NA, not applicable.
Monkeys received ∼106 PFU of wt DEN1WP subcutaneously.
Received 105 PFU of DEN1mutF 17 months prior to challenge.
Received 103 PFU of DEN1mutF 17 months prior to challenge.
TABLE 5.
Results of challenge study 2 for viremia in flavivirus-naive and DEN1mutF-immunized monkeys after challenge with 106 PFU of wt DEN1WP
| Monkey | Peak virus titer in sera (10−2)a
|
|
|---|---|---|
| RT-PCR | Plaque titration | |
| 642 | 2.2 (5) [1-5] | 1.1 (4) [1-4] |
| CE40 | 1.1 (2) [1-2] | 1.5 (3) [1-3] |
| CE8B | 7.0 (3) [1-3] | 2.5 (3) [1-3] |
| CE8C | 7.2 (3) [1-3] | 1.0 (3) [1-3] |
| KWDb | ND | ND |
| KGTc | ND | ND |
Peak titer on any day of the study. For the RT-PCR assay, units are PFU equivalents per milliliter. For plaque titration, units are PFU per milliliter. Monkeys were infected on day 0 with 106 PFU of DEN1WP, injected subcutaneously in the subscapular area. Virus in serial serum samples was detected and measured either by a quantitative, validated RT-PCR or by direct plaque formation on LLCMK2 cells. Values in parentheses are total number of days of viremia; values in brackets are days of the study on which viremia was observed. ND, not detected.
Immunized with DEN1mutF at 105 PFU 17 months prior to challenge.
Immunized with DEN1mutF at 103 PFU 17 months prior to challenge.
FIG. 4.
Viremia in DEN1mutF-immunized monkeys versus controls after challenge with 106 PFU of DEN1WP (challenge study 2). The challenge inoculum was delivered by subcutaneous injection in the subscapular region on day 0 of the study. Virus titers in individual serum samples from all monkeys were determined for serum samples obtained on days 0 through 10 by using RT-PCR as described in the legend to Fig. 3.
DEN1mutF-immunized monkeys KWD and KGT were also infected after challenge with 106 PFU of wt DEN1WP, but only on the basis of the serologic criterion. Each of these monkeys had prechallenge DEN1-specific neutralizing antibodies as a consequence of their previous immunization with DEN1mutF (Table 4), and each of them had at least a 128-fold rise in neutralizing antibody titer as a result of challenge. This result indicated an anamnestic response to DEN1 antigens. However, both of these monkeys were found to be completely protected from any viremia when their sera were tested along with sera from control monkeys by both RT-PCR (Fig. 4) and a direct plaque assay for virus detection (Table 5).
To summarize the results of the second challenge study, four of four flavivirus-naive monkeys became viremic after challenge, while neither of two DEN1mutF-immunized monkeys exhibited any viremia, even though the challenge was initiated 17 months after a single exposure to DEN1mutF virus. In the two challenge studies taken together, six of seven control monkeys became viremic, while none of six DEN1mutF-immunized monkeys did so, and protection of the immunized group was statistically significant (P < 0.02).
Nucleotide sequence analysis of the DEN1mutF genome and that of its phenotypic revertant, DEN1mutFRev.
In order to determine the stability of the mutF host range-restricted phenotype in C6/36 cells and ultimately to determine the relationship of this phenotype to attenuation in monkeys, DEN1mutF and DEN2mutF viruses were passaged blindly on fresh C6/36 cell monolayers every 8 days for six consecutive passages, and virus harvested at each passage level was tested for its ability to replicate in that cell line. Phenotypic revertant DEN1mutF and DEN2mutF viruses were detected after a total of 24 days of culture (or after three passages) on C6/36 cell monolayers. This result demonstrated that the mutF phenotype was similarly stable in the context of the two different dengue virus genomes. The respective revertant viruses replicated on C6/36 cells with an efficiency indistinguishable from that of the wt viruses derived from the DEN1WP and DEN2NGC infectious DNAs; i.e., each revertant mutF virus reached a peak titer on C6/36 cells of ∼107 PFU/ml (data not shown).
The complete average nucleotide sequence of the RNA genomes of DEN1mutF and DEN1mutFRev viruses was determined and compared to that of DEN1WP infectious DNA (Fig. 5). Apart from the mutations in the 3′-SL that were deliberately introduced to derive DEN1mutF virus, DEN1mutF RNA contained only a single additional mutation that distinguished it from the wt viral DNA, a replacement of G by A at position 1538 of the nucleotide sequence. This mutation was predicted to result in a Glu-to-Lys substitution at amino acid 202 in the DEN1 envelope (E) protein sequence. The E protein is the viral attachment protein (6), and the nonconservative mutation that we detected could theoretically account for the host range-restricted phenotype or attenuation in primates of DEN1mutF, since either of these properties could be related to the efficiency of virus binding to cells as mediated by E protein.
FIG. 5.
Relevant features of genomes of DEN1WP (Wt), DEN1mutF (mutF), and DEN1mutFRev (FRev) viruses. DEN1mutFRev was a phenotypic revertant of DEN1mutF virus that arose after 24 days of culture of DEN1mutF on C6/36 cells (see Materials and Methods). 5′- and 3′-NCRs are shown as solid lines. The predicted conformation of the 3′-SL formed by the 3′-terminal 93 nt of the DEN1 3′-NCR is shown, and the locations of the mutational changes in the lower portion of the double-stranded region forming the long stem in the 3′-SL are indicated by thick lines. S, 5′ terminus of the portion of the ORF encoding structural proteins C (capsid), prM, and E; NS, location of the E-NS1 cleavage site at the origin of the portion of the ORF encoding proteins NS1 through NS5. The results of complete nucleotide sequencing of DEN1WP DNA and DEN1 mutant viral RNAs derived from the two mutant viruses revealed nucleotide sequence differences, as shown at positions indicated by vertical arrows within the ORF and an asterisk in the 3′-SL. For example, G1538A indicates that the G at position 1538 in the nucleotide sequence numbered from the 5′ terminus of the genome was replaced by an A. Amino acids are numbered according to distance from the amino terminus of each gene product, using the single-letter code. The notation E202/K indicates that Glu at position 202 of the E protein was replaced by Lys, and so forth.
To investigate these possibilities, a second DEN1mutF infectious DNA was derived using the DEN1WP infectious clone. We established by sequencing that the newly derived DEN1mutF DNA did not contain the G1538A mutation, and we then rederived DEN1mutF virus by transfection of LLCMK2 cells with RNA transcripts, using this second DEN1mutF recombinant DNA clone as a template. RNA from newly derived DEN1mutF virus lacked the G1538A mutation, but the new mutant exhibited the same level of restriction of replication in C6/36 cells as initially derived DEN1mutF virus (data not shown). Thus, we concluded that the G1538A mutation was not required for the in vitro phenotype of DEN1mutF. Recent results of a monkey trial with newly derived DEN1mutF also demonstrated that the G1538A mutation was not required for the attenuated phenotype of original DEN1mutF virus in this animal model (M. Mammen et al., unpublished data).
Nucleotide sequence analysis of the DEN1mutFRev RNA revealed that it retained the G1538A mutation detected in the original DEN1mutF RNA, further evidence that this mutation was not relevant to the host range-restricted phenotype of DEN1mutF. In addition, three other mutations were detected that differentiated the DEN1mutFRev genome from DEN1WP DNA and DEN1mutF RNA: (i) a mutation of A to G at nt 3143, which was predicted to result in an Ile-to-Val substitution at amino acid 242 of NS1; (ii) a C-to-U mutation at nt 7689, which was predicted to result in a Ser-to-Phe substitution at amino acid 39 of NS5; and (iii) a U-to-A mutation at nt 10724 (12 nt upstream from the 3′ terminus of the genome; Fig. 1). This last mutation resulted in the reduction in size of a predicted bulge in the long stem of the dengue virus 3′-SL by eliminating a predicted noncomplementary alignment of U-12 with U-68 in the opposite strand of the long stem; the U-12/A mutation permits the formation of an A-U base pair with U-68. Nucleotide sequencing of RNA prepared from DNA-derived DEN1WP confirmed that none of the total of four mutational changes described for DEN1mutF or DEN1mutFRev RNAs had arisen spontaneously in the wt genome after transfection.
Assuming that the mechanisms for the host range-restricted phenotypes of DEN1mutF and DEN2mutF viruses are identical, the mutational changes detected in the DEN1mutFRev genome compared to the DEN1mutF genome were in accord with the previous observation that DEN2mutF virus was severely defective for RNA replication at early times after infection of C6/36 cells (39). (NS1 [20, 23] and NS5 [6, 33] are essential for RNA replication, and NS5 binds the 3′-SL in vivo [7], suggesting that the C7689U and U10724A mutations could have acted cooperatively to augment replication in C6/36 cells.)
DISCUSSION
DEN2mutF was unique among several replication-competent 3′-SL DEN2 mutants that we derived by mutagenesis of a DEN2 infectious DNA, in that it displayed a host range-restricted phenotype with respect to replication in C6/36 cells (39). The wt parent virus, DEN2NGC, was a laboratory strain originally isolated from a human subject in 1945 (28). It had been extensively passaged in and adapted to mouse brain prior to the cloning of its genome. Introduction of the mutF mutations into DNA encoding the genome of a human-virulent DEN1 isolate, DEN1WP, resulted in a mutant DEN1 virus (DEN1mutF) that exhibited a host range-restricted phenotype similar to that of DEN2mutF virus. The identical phenotypic character of the DEN1mutF and DEN2mutF viruses was remarkable, since the wt parent viruses were different not only in serotype but also with regard to adaptation to replication in mouse brain. These results suggest that similarly derived DEN3mutF and DEN4mutF viruses may also display the mutant F phenotype.
The mechanism for the host range-restricted phenotype of the mutF virus(es) is not known. In view of previous work in which the WN 3′-SL nucleotide sequence was shown to bind specific BHK cellular proteins in vitro (3, 4), we speculated that the mutF 3′-SL might be defective for the binding of specific C6/36 cellular proteins necessary for the initiation of viral RNA synthesis, whereas its interaction with analogous proteins in monkey cells was adequate to permit RNA and hence viral replication after a lag phase at very early times after infection (39). In fact, DEN2mutF virus was found to be defective at the level of viral RNA replication in C6/36 cells by a Northern blot analysis that did not distinguish between virus-specific negative-strand and positive-strand RNAs. We assume, along with others (3, 4), that interactions of the 3′-SL with cell-specific proteins are required for the proper binding and function of NS proteins necessary to initiate RNA synthesis. This hypothesis to explain the phenotype of mutant F viruses was also consistent with the results of a more recent in vitro study in which limited negative-strand RNA synthesis was primed by DEN2-specific 5′- and 3′-NCR nucleotide sequences acting in trans (38). In this context, a DEN2 3′-NCR containing mutF mutational changes was much less efficient as a primer in a lysate of C6/36 cells than in a lysate of LLCMK2 cells. However, in separate studies, we were unable to detect any differences when the binding of C6/36 and LLCMK2 cellular proteins to the wt and mutF DEN2 3′-SL nucleotide sequences was compared in vitro, using Northwestern blotting and gel mobility shift assays (unpublished data). It is possible that the putative cellular protein-binding defect is a quantitative one that we cannot detect at the level of sensitivity of either assay.
The nucleotide sequence analysis of the DEN1mutF and DEN1mutFRev genomes presented here provided further results that were also consistent with our hypothesis to explain the host range-restricted phenotype of mutF viruses. Compared to the DEN1mutF genome, the DEN1mutFRev genome contained single-nucleotide substitution mutations in the NS1 and NS5 gene sequences that were predicted to result in Ile-to-Val and Ser-to-Phe amino acid changes, respectively, in the two NS proteins. Recent evidence obtained with yellow fever virus (YF) mutants suggested that NS1 plays a role in RNA synthesis (20, 23). NS5 contains a putative methyltransferase domain at its amino terminus (17a) and a demonstrable RNA-dependent RNA polymerase activity at its carboxy terminus (17, 33). The locus of the NS5 mutation in the DEN1mutFRev genome was amino terminal to either of these domains, however. It is possible that this most amino-terminal segment of NS5 participates in binding interactions between NS5 and the 3′-SL or other proteins in the replication complex. In addition to these substitution mutations, we detected a single point mutation in the DEN1mutFRev 3′-SL that altered the predicted conformation of the 3′-SL at the site of a bulge in the long stem (see Results; Fig. 1). Such noncomplementary pairs have been shown in other systems to play a critical role in the binding of proteins to stable, otherwise double-stranded regions in RNA (2, 31, 35, 37). For flaviviruses, the Japanese encephalitis virus NS5 was shown to bind the Japanese encephalitis virus 3′-SL during replication (7), although the specific binding sites in RNA were not mapped. In addition, the BHK cellular protein translation elongation factor eF1-α bound the lower portion of the double-stranded long stem in the WN 3′-SL in vitro (4). Furthermore, other as-yet-unidentified BHK cellular proteins also bound to an RNA containing the WN 3′-SL nucleotide sequence in vitro (3). These findings at least are consistent with the concept that any one or more of the three DEN1mutFRev-specific mutations could facilitate the cooperative interactions of C6/36 cellular proteins with the DEN1mutF 3′-SL and DEN1 proteins that are required for efficient RNA replication. We are currently constructing a set of DEN1 mutants in order to determine which of the DEN1mutFRev-specific mutations is actually required for phenotypic reversion. We also seek to determine whether DEN1mutFRev is attenuated in primates.
DEN1mutF was attenuated in a monkey model of dengue virus infection, in that the mutant virus caused significantly fewer days of viremia with lower peak titers of virus than its human-virulent parent virus at each of three dose levels of the mutant and wt viruses. Monkeys do not become ill from dengue virus infection (28), but a large body of evidence supports a correlation between the infectivity of a dengue virus isolate for monkeys and human virulence (1, 8a, 8b, 18, 21, 29; Vaughn et al., Abstr. XIIIth Int. Congr. Trop. Med. Malaria). These data have arisen historically in relation to attempts to develop live, attenuated dengue vaccines, when candidate vaccine viruses were tested in rhesus macaques prior to clinical trials. In one series of experiments where DEN1 was attenuated by serial passages in primary dog kidney (PDK) cells, an increasing number of PDK cell passages was shown to correlate with progressive reductions of infectivity in monkeys and of virulence in humans (Vaughn et al., Abstr. XIIIth Int. Congr. Trop. Med. Malaria). Future clinical trials will eventually reveal whether DEN1mutF virus (and DEN2, -3, and -4 mutant F viruses) is sufficiently attenuated in humans to elicit a long-lasting protective immune response without causing unacceptable symptoms of DF.
Current strategies for the development of a live, attenuated tetravalent dengue vaccine are based on either classical virological techniques (where dengue virus serotypes have been separately “adapted” to PDK cells by serial passages) (8a, 8b; Vaughn et al., Abstr. XIIIth Int. Congr. Trop. Med. Malaria) or techniques of modern molecular biology (where attenuated viruses are derived by transfection of cells with viral RNA generated in vitro from mutagenized infectious DNA constructs) (11, 12, 21). These various approaches to immunization against dengue all deliver the major surface antigens of the targeted dengue virus (premembrane [prM] and E) but differ in the number of additional serotype-specific and/or dengue virus group-specific genes that are included in the vaccine virus. For example, the PDK cell-adapted vaccine viruses express all dengue proteins specific for each serotype and bear different sets of attenuating mutations that spontaneously arose during PDK cell passage, depending on the serotype. In contrast, YF-dengue chimeric vaccine viruses that express only two dengue virus antigens are currently under development. In these vaccine viruses (one for each of the four dengue virus serotypes) (11, 12), the dengue virus prM and E genes are substituted for the YF prM and E genes in the context of the genome of the YF strain 17D vaccine virus.
There are other examples of promising live virus vaccines for dengue that lie in between these two extremes in terms of the delivery of dengue virus group- and/or serotype-specific genes other than the prM and E genes. However, it remains to be seen which of these strategies and what “dose” of homologous dengue virus group- or serotype-specific genes prove optimal for avoiding or minimizing the major problems inherent in the development of a tetravalent dengue vaccine. These are (i) potential viral or immune interference among serotypes in a mixture of vaccine viruses, (ii) the possibility of sensitizing vaccinees to severe manifestations of a subsequent naturally acquired dengue infection, and (iii) the need to elicit a life-long protective tetravalent immune response in subjects living in areas in which dengue is endemic while avoiding problems i and ii. A mutF tetravalent vaccine mixture would have virtues of both the classical and the molecular approaches, in that it would deliver all 10 genes of each serotype, like the PDK cell-adapted vaccine viruses. However, unlike the PDK cell-adapted vaccines, the individual viruses in a mutF tetravalent vaccine mixture would contain a homogeneous genetically defined set of attenuating mutations, possibly reducing the potential for interference.
REFERENCES
- 1.Angsubphakorn, S., S. Yoksan, N. Bhamarapravati, J. B. Moe, N. J. Marchette, A. Pradermwong, and S. Sahaphong. 1988. Dengue-4 vaccine: neurovirulence, viraemia, and immune response in rhesus and cynomolgus monkeys. Trans. R. Soc. Trop. Med. Hyg. 82:746-749. [DOI] [PubMed] [Google Scholar]
- 2.Bartel, D. P., M. L. Zapp, M. R. Green, and J. W. Szostak. 1991. HIV-1 Rev regulation involves recognition of non-Watson-Crick base pairs in the viral RNA. Cell 17:529-536. [DOI] [PubMed] [Google Scholar]
- 3.Blackwell, J., and M. Brinton. 1996. BHK cell proteins that bind to the 3′ stem-loop structure of the West Nile virus genome RNA. J. Virol. 69:5650-5658. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Blackwell, J. L., and M. A. Brinton. 1997. Translation elongation factor 1 alpha interacts with the 3′ stem-loop region of West Nile virus genomic RNA. J. Virol. 71:6433-6444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Brinton, M. A., and J. H. Dispoto. 1987. Sequence and secondary structure analysis of the 5′-terminal region of flavivirus genome RNA. Virology 162:290-299. [DOI] [PubMed] [Google Scholar]
- 6.Chambers, T. J., C. S. Hahn, R. Galler, and C. M. Rice. 1990. Flavivirus genome organization, expression, and replication. Annu. Rev. Microbiol. 44:649-688. [DOI] [PubMed] [Google Scholar]
- 7.Chen, C.-J., M.-D. Kuo, L.-J. Chien, S.-L. Hsu, Y.-M. Wang, and J.-H. Lin. 1997. RNA-protein interactions: involvement of NS3, NS5, and 3′ noncoding regions of Japanese encephalitis virus genomic RNA. J. Virol. 71:3466-3473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cui, T., R. J. Sugrue, Q. Xu, A. K. Lee, Y. C. Chan, and J. Fu. 1998. Recombinant dengue virus type 1 NS3 exhibits specific viral RNA binding and NTPase activities regulated by NS5 protein. Virology 246:409-417. [DOI] [PubMed] [Google Scholar]
- 8a.Eckels, K. H., C. H. Hoke, and E. A. Henchal. 1992. Dengue virus live-attenuated vaccine strategy: identification of acceptable candidate monotypic vaccines. Am. J. Trop. Med. Hyg. 47:99. [Google Scholar]
- 8b.Edelman, R., C. O. Tacket, S. S. Wasserman, et al. 1994. A live-attenuated dengue-1 vaccine candidate (45AZ5) passaged in primary dog kidney cell culture is attenuated and immunogenic for humans. J. Infect. Dis. 170:1448-1455. [DOI] [PubMed] [Google Scholar]
- 9.Grange, T., M. Bouloy, and M. Girard. 1985. Stable secondary structure at the 3′ end of the genome of yellow fever virus (17D vaccine strain). FEBS Lett. 188:159-163. [DOI] [PubMed] [Google Scholar]
- 10.Gubler, D. J., and M. Meltzer. 1999. Impact of dengue/dengue-hemorrhagic fever on the developing world. Adv. Virus Res. 53:35-70. [DOI] [PubMed] [Google Scholar]
- 11.Guirakhoo, F., R. Weltzin, T. J. Chambers, Z. X. Zhang, K. Soike, M. Ratterree, J. Arroyo, K. Georgakopoulos, J. Catalan, and T. P. Monath. 2000. Recombinant chimeric yellow fever-dengue type 2 virus is immunogenic and protective in nonhuman primates. J. Virol. 74:5477-5485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Guirakhoo, F., J. Arroyo, K. V. Pugachev, C. Miller, Z.-X. Zhang, R. Weltzin, K. Georgakopoulos, J. Catalan, S. Ocran, K. Soike, M. Ratterree, and T. P. Monath. 2001. Construction, safety, and immunogenicity in nonhuman primates of a chimeric yellow fever-dengue virus tetravalent vaccine. J. Virol. 75:7290-7304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Houng, H. H., D. Hritz, and N. Kanesa-thasan. 2000. Quantitative detection of dengue 2 virus using fluorogenic RT-PCR based on 3′-noncoding sequence. J. Virol. Methods 86:1-11. [DOI] [PubMed] [Google Scholar]
- 14.Houng, H. H., R. C. Chen, D. W. Vaughn, and N. Kanesa-thasan. 2001. Development of a fluorogenic RT-PCR system for quantitative identification of dengue virus serotypes 1 to 4 using conserved and serotype-specific 3′-non-coding sequences. J. Virol. Methods 95:19-32. [DOI] [PubMed] [Google Scholar]
- 15.Irie, K., P. M. Mohan, Y. Sasaguri, R. Putnak, and R. Padmanabhan. 1989. Sequence analysis of cloned dengue virus type 2 genome (New Guinea C strain). Gene 75:197-211. [DOI] [PubMed] [Google Scholar]
- 16.Jacobs, M. 2000. Dengue: emergence as a global public health problem and prospects for control. Trans. R. Soc. Trop. Med. Hyg. 94:7-8. [DOI] [PubMed] [Google Scholar]
- 17.Johansson, M., A. J. Brooks, D. A. Jans, and S. G. Vasudevan. 2001. A small region of the dengue virus-encoded RNA-dependent RNA polymerase, NS5, confers interaction with both the nuclear transport receptor importin-beta and the viral helicase, NS3. J. Gen. Virol. 82:735-745. [DOI] [PubMed] [Google Scholar]
- 17a.Koonin, E. V. 1993. Computer-assisted identification of a putative methyl-transferase domain in NS5 protein of flaviviruses and lambda 2 protein of reovirus. J. Gen. Virol. 74:733-740. [DOI] [PubMed] [Google Scholar]
- 18.Kraiselburd, E., M. J. Kessler, and G. Torres-Blasini. 1984. Lack of viraemia and limited antibody response of dengue virus immune rhesus monkeys after vaccination with DEN2/S1 vaccine. Trans. R. Soc. Trop. Med. Hyg. 78:445-448. [DOI] [PubMed] [Google Scholar]
- 19.Li, H., S. Clum, S. You, K. E. Ebner, and R. Padmanabhan. 1999. The serine protease and RNA-stimulated nucleoside triphosphatase and RNA helicase functional domains of dengue virus type 2 NS3 converge within a region of 20 amino acids. J. Virol. 73:3108-3116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lindenbach, B. D., and C. M. Rice. 1997. trans-Complementation of yellow fever virus NS1 reveals a role in early RNA replication. J. Virol. 71:9608-9617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Men, R., M. Bray, D. Clark, R. M. Chanock, and C.-J. Lai. 1996. Dengue type 4 virus mutants containing deletions in the 3′ noncoding region of the RNA genome: analysis of growth restriction in cell culture and altered viremia pattern and immunogenicity in rhesus monkeys. J. Virol. 70:3930-3937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Mohan, P. M., and R. Padmanabhan. 1991. Detection of stable secondary structure at the 3′ terminus of dengue virus type 2 RNA. Gene 108:185-191. [DOI] [PubMed] [Google Scholar]
- 23.Muylaert, I. R., R. Galler, and C. M. Rice. 1997. Genetic analysis of the yellow fever virus NS1 protein: identification of a temperature-sensitive mutation which blocks RNA accumulation. J. Virol. 71:291-298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Polo, S., G. Ketner, R. Levis, and B. Falgout. 1997. Infectious RNA transcripts from full-length dengue virus type 2 cDNA clones made in yeast. J. Virol. 71:5366-5374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Puri, B., S. Polo, C. G. Hayes, and B. Falgout. 2000. Construction of a full-length infectious clone for dengue-1 virus Western Pacific, 74 strain. Virus Genes 20:57-63. [DOI] [PubMed] [Google Scholar]
- 26.Rice, C. M., E. M. Lenches, S. R. Eddy, S. J. Shin, R. L. Sheets, and J. H. Strauss. 1985. Nucleotide sequence of yellow fever virus: implications for flavivirus gene expression and evolution. Science 229:726-735. [DOI] [PubMed] [Google Scholar]
- 27.Rosen, L. 1999. Comments on the epidemiology, pathogenesis and control of dengue. Med. Trop. (Mars) 59:495-498. [PubMed] [Google Scholar]
- 28.Sabin, A. B. 1952. Research on dengue during World War II. Am. J. Trop. Med. 1:30-50. [DOI] [PubMed] [Google Scholar]
- 29.Scott, R. M., A. Nisalak, K. H. Eckels, M. Tingpalapong, V. R. Harrison, D. Gould, F. E. Chapple, and P. K. Russell. 1980. Dengue-2 vaccine: viremia and immune responses in rhesus monkeys. Infect. Immun. 27:181-186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Spencer, F., G. Ketner, C. Connelly, and P. Hieter. 1993. Targeted recombination-based cloning and manipulation of large DNA fragments in yeast. Methods Companion Methods Enzymol. 5:161-175. [Google Scholar]
- 31.Stern, S., R. C. Wilson, and H. F. Noller. 1986. Location of the binding site for protein S4 on 16S ribosomal RNA by chemical and enzymatic probing and primer extension. J. Mol. Biol. 192:101-110. [DOI] [PubMed] [Google Scholar]
- 32.Takegami, T., M. Washizu, and K. Yasui. 1986. Nucleotide sequence at the 3′ end of Japanese encephalitis virus genome RNA. Virology 152:483-486. [DOI] [PubMed] [Google Scholar]
- 33.Tan, B. H., J. Fu, R. J. Sugrue, E. H. Yap, Y. C. Chan, and Y. H. Tan. 1996. Recombinant dengue type 1 virus NS5 protein expressed in Escherichia coli exhibits RNA-dependent RNA polymerase activity. Virology 216:317-325. [DOI] [PubMed] [Google Scholar]
- 34.Utama, A., H. Shimizu, S. Morikawa, F. Hasebe, K. Morita, A. Igarashi, M. Hatsu, K. Takamizawa, and T. Miyamura. 2000. Identification and characterization of the RNA helicase activity of Japanese encephalitis virus NS3 protein. FEBS Lett. 465:74-78. [DOI] [PubMed] [Google Scholar]
- 35.Weeks, K. M., and D. M. Crowthers. 1991. RNA recognition by Tat derived peptides: interaction with the major groove? Cell 66:577-588. [DOI] [PubMed] [Google Scholar]
- 36.Wengler, G., and E. Castle. 1986. Analysis of structure properties which possibly are characteristic for the 3′-terminal sequence of the genome RNA of flaviviruses. J. Gen. Virol. 67:1183-1188. [DOI] [PubMed] [Google Scholar]
- 37.Wu, H.-N., and O. C. Uhlenbeck. 1987. Role of a bulged A residue in a specific RNA-protein interaction. Biochemistry 26:8221-8227. [DOI] [PubMed] [Google Scholar]
- 38.You, S., B. Falgout, L. Markoff, and R. Padmanabhan. 2001. In vitro RNA synthesis from exogenous dengue viral RNA templates requires long range interactions between 5′- and 3′-terminal regions that influence RNA structure. J. Biol. Chem. 276:15581-15591. [DOI] [PubMed] [Google Scholar]
- 39.Zeng, L. L., B. Falgout, and L. Markoff. 1998. Identification of specific nucleotide sequences within the conserved 3′-SL in the dengue type 2 virus genome required for replication. J. Virol. 72:7510-7522. [DOI] [PMC free article] [PubMed] [Google Scholar]