Paired Charge-to-Alanine Mutagenesis of Dengue Virus Type 4 NS5 Generates Mutants with Temperature-Sensitive, Host Range, and Mouse Attenuation Phenotypes

Abstract

Charge-to-alanine mutagenesis of dengue virus type 4 (DEN4) NS5 gene generated a collection of attenuating mutations for potential use in a recombinant live attenuated DEN vaccine. Codons for 80 contiguous pairs of charged amino acids in NS5 were individually mutagenized to create uncharged pairs of alanine residues, and 32 recombinant mutant viruses were recovered from the 80 full-length mutant DEN4 cDNA constructs. These mutant viruses were tested for temperature-sensitive (ts) replication in both Vero cells and HuH-7 human hepatoma cells. Of the 32 mutants, 13 were temperature sensitive (ts) in both cell lines, 11 were not ts in either cell line, and 8 exhibited a host range (tshr) phenotype. One tshr mutant was ts only in Vero cells, and seven were ts only in HuH-7 cells. Nineteen of the 32 mutants were 10-fold or more restricted in replication in the brains of suckling mice compared to that of wild-type DEN4, and three mutants were approximately 10,000-fold restricted in replication. The level of temperature sensitivity of replication in vitro did not correlate with attenuation in vivo. A virus bearing two pairs of charge-to-alanine mutations was constructed and demonstrated increased temperature sensitivity and attenuation relative to either parent virus. This large set of charge-to-alanine mutations specifying a wide range of attenuation for mouse brain should prove useful in fine-tuning recombinant live attenuated DEN vaccines.

Dengue virus (DEN) (genus Flavivirus, family Flaviviridae) is a single-stranded, positive-sense RNA virus with a 10.6-kb genome. The genome organization of DEN viruses is 5′-UTR-C-prM-E-NS1-NS2A-NS2B-NS3-NS4A-NS4B-NS5-UTR-3′ (UTR, untranslated region; C, capsid; prM, membrane precursor; E, envelope; NS, nonstructural) (6, 34). DEN is comprised of four serotypes, each of which is capable of causing dengue fever (DF) and its more-severe variants, dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS). DEN is transmitted by mosquitoes of the genus Aedes. In the past few decades, unregulated urbanization, modern patterns of travel, cessation of vector control, and other factors have led to a rapid expansion in the geographic range of all four serotypes and the cocirculation of two or more serotypes in many areas (15). Concurrently, the number of reported cases of DHF and DSS has risen dramatically. It is estimated that 50 to 100 million cases of DF and several hundred thousand cases of DHF and DSS occur annually (14, 15). Unfortunately, there is neither a licensed vaccine nor an effective antiviral therapy for the control of DEN infection. Because progression to DHF or DSS is more likely during secondary infection (27, 38), an acceptable dengue vaccine must confer immunity to all four serotypes. To date, attempts to develop tetravalent live-attenuated vaccines via cell culture passage have resulted in uneven attenuation, such that effective immunity is not induced to all four serotypes, and the most immunogenic viruses also tend to be reactogenic (4, 18).

We are developing live attenuated vaccine candidates based on production of recombinant viruses from cDNAs containing defined attenuating mutations introduced into DEN type 4 (DEN4) nonstructural protein genes and untranslated regions (9, 24, 37). Antigenic chimeric dengue virus vaccine candidates for DEN1, DEN2, and DEN3 are being made in which the structural proteins from each of these serotypes replace those of a DEN4 virus containing one or more attenuating mutations in its nonstructural protein or genomic UTR. Our initial DEN4 vaccine candidate, 2AΔ30, was derived from cDNA containing a 30-nucleotide deletion in the 3′ UTR (9, 25). Clinical evaluation of vaccine candidate 2AΔ30 showed it to be safe, well-tolerated, and strongly immunogenic (9). While none of the vaccinated volunteers developed symptoms, an asymptomatic rash was detected in 50% of volunteers. Also, transient elevation in the levels of alanine transaminase (ALT) in serum occurred in five volunteers. Elevated ALT levels are commonly associated with both natural DEN infections (11, 12, 17, 18) and dengue vaccination (23, 26, 39). Each volunteer developed a moderate to high level of neutralizing antibody despite a low to undetectable level of viremia. The vaccine candidate was not transmitted to mosquitoes fed on vaccinees and was strongly restricted in its ability to establish a disseminated infection in mosquitoes infected via an artificial blood meal (37).

While the performance of the DEN4 2AΔ30 vaccine candidate is encouraging, it may be necessary to attenuate the virus further in order to reduce or eliminate vaccine-associated ALT elevations and rash. Moreover, further study may reveal a need to attenuate one of the progeny chimeric heterotypic viruses derived from 2AΔ30. To this end, we have taken several approaches toward generating a collection of attenuating mutations in the nonstructural genes and UTRs of DEN4 (5). Specifically, we have made an effort to identify mutant viruses with temperature-sensitive (ts) and small-plaque (sp) phenotypes in Vero or liver (HuH-7) cells as useful markers for attenuation in general and for liver-specific attenuation, respectively. Although there is currently no animal model for illness caused by DEN that mimics human disease, we have used restriction of replication in the suckling mouse brain as a marker of in vivo attenuation. Blaney et al. (5) reported that chemical mutagenesis of DEN4 yielded a subset of mutant viruses that exhibited a variety of ts and mouse attenuation phenotypes. Studies are currently under way to identify the specific mutations that confer these phenotypes.

In the present study, we performed paired charge-to-alanine mutagenesis on the NS5 gene of DEN4 to create a panel of attenuated viruses with specified mutations. Charge-to-alanine mutagenesis can create ts mutant viruses by destabilizing hydrogen-bonding or electrostatic interactions at the surface of the protein, rendering these interactions more thermosensitive (1). Previous studies have demonstrated the efficacy of this technique for generation of ts mutants in positive-strand viruses (8, 29).

The current study focuses on NS5, the largest (104 kDa) and most conserved (67% identity among the four DEN serotypes) of the DEN proteins. Three putative functional domains have been identified in NS5 via homology to known functional proteins: an S-adenosylmethionine methyltransferase-like domain in the N-terminal region, (20), a centrally located nuclear localization sequence (NLS) (13, 19), and an RNA-dependent RNA polymerase (RdRp) domain containing a characteristic GDD motif in the carboxy half of the gene (21). Subsequently, experimental analyses have corroborated the function of the NLS (13) and shown that a region N terminal to the NLS can bind the nuclear import receptor importin-β (16). The function of the RdRp has also been confirmed (2, 33, 36), and it has been shown that NS5 and NS3 interact in the conversion of replicative-form (RF) RNA to replicative intermediates (RI) (2, 7, 19, 33). The NS5 gene is an inviting target for mutagenesis since it is a nonstructural gene and therefore will be retained during the production of antigenic chimeric viruses. In addition, it is a large protein with many functions so that mutagenesis should create a broad array of phenotypes useful to fine-tune attenuation of vaccine candidates. In this study we have described the generation of 32 NS5 mutant viruses and the evaluation of their in vitro and in vivo phenotypes. In addition, we demonstrate the feasibility of modifying the attenuation phenotype of these mutant viruses by combining two pairs of charge-to-alanine mutations.

MATERIALS AND METHODS

Cells and viruses.

World Health Organization (WHO) Vero cells (African green monkey kidney cells) were maintained in virus production-serum-free medium (VP-SFM) (Life Technologies, Grand Island, N.Y.) supplemented with 2 mM l-glutamine (Life Technologies) and 0.05 mg of gentamicin (Life Technologies) per ml. HuH-7 cells (human hepatoma cells) (30) were maintained in Dulbecco modified Eagle medium F-12 (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Summit Biotechnologies, Fort Collins, Colo.), 1 mM l-glutamine, and 0.05 mg of gentamicin per ml. C6/36 cells (Aedes albopictus mosquito cells) were maintained in minimal essential medium (MEM) (Life Technologies) containing 10% FBS, 2 mM l-glutamine, 2 mM nonessential amino acids (Life Technologies), and 0.05 mg of gentamicin per ml.

Recombinant DEN4 (rDEN4) was derived from the cDNA plasmid p4, which was modified from 2A, a cDNA generated from wild type (wt) DEN4 strain 814669 (Dominica 1981) as previously described (9, 24). rDEN4Δ30 was derived from plasmid p4Δ30, a cDNA clone containing a deletion of nucleotides 10,478 to 10,507 in the 3′ UTR, the same mutation that is present in 2AΔ30 (9, 25). The GenBank accession number for rDEN4 is AF326825.

Charge-cluster-to-alanine mutagenesis.

Charge-cluster-to-alanine mutagenesis (29), in which pairs of charged amino acids are replaced with alanine residues, was used to individually mutagenize the coding sequence for 80 pairs of contiguous charged amino acids in the DEN4 NS5 gene. Subclones suitable for mutagenesis were derived from full-length DEN4 plasmid p4 by digestion of the NS5 coding region with XmaI/PstI (pNS5A), PstI/SacII (pNS5B), or SacII/MluI (pNS5C) at nucleotide positions 7444/8155, 8155/9315, and 9315/10402, respectively. These fragments were then subcloned into modified pUC vectors, and Kunkel mutagenesis was performed (22). To create each mutation, oligonucleotides were designed to change the sequence coding for individual pairs of charged amino acids to GCAGCX, thereby replacing them with codons for two alanine residues and creating a BbvI restriction site. The BbvI site was added to facilitate screening of cDNAs and recombinant virus genomes for the presence of the mutated sequence. Restriction enzyme fragments bearing the alanine mutations were cloned back into the full-length p4 plasmid, which was sequenced to confirm the presence of the desired mutation. Double mutant viruses carrying two pairs of charge-cluster-to-alanine mutations were created by swapping appropriate NS5 fragments carrying one pair of mutations into a previously mutagenized p4 cDNA carrying a second pair of mutations in a different fragment using conventional cloning techniques.

Generation of rDEN4.

5′-capped transcripts were synthesized in vitro from mutagenized cDNA templates using AmpliCap SP6 RNA polymerase (Epicentre, Madison, Wis.). Transfection mixtures, consisting of 1 μg of transcript in 60 μl of HEPES-saline plus 12 μl of DOTAP (Roche Diagnostics Corp., Indianapolis, Ind.), were added, along with 1 ml of VP-SFM to subconfluent monolayers of Vero cells in six-well plates. Transfected monolayers were incubated at 35°C for approximately 18 h, cell culture medium was removed and replaced with 2 ml of VP-SFM, and cell monolayers were incubated at 35°C for 5 to 6 days. Cell culture medium was collected, and the presence of virus was determined by titration in Vero cells followed by immunoperoxidase staining as previously described (9). Recovered virus was amplified by an additional passage in Vero cells, and virus suspensions were combined with SPG stabilizer (218 mM sucrose, 6 mM l-glutamic acid, 3.8 mM potassium phosphate [monobasic], 7.2 mM potassium phosphate [dibasic] [pH 7.2] [final concentrations given]), aliquoted, frozen on dry ice, and stored at −70°C.

cDNA constructs not yielding virus after transfection of Vero cells were used to transfect C6/36 cells as follows. Transfection mixtures, as described above, were added, along with 1 ml of MEM containing 10% FBS, 2 mM l-glutamine, 2 mM nonessential amino acids, and 0.05 mg of gentamicin per ml, to monolayers of C6/36 cells. Transfected-cell monolayers were incubated at 32°C for 18 h, cell culture medium was removed and replaced with 2 ml of fresh medium, and cell monolayers were incubated at 32°C for an additional 5 to 6 days. Cell culture media were then used to infect Vero cells and incubated for 5 to 6 days, at which time cell culture media were collected and frozen and the virus titers were determined as described above.

Recovered viruses were biologically cloned by two rounds of terminal dilution in Vero cells followed by an additional amplification in Vero cells. Briefly, virus was initially diluted to a concentration of approximately 20 PFU/ml in VP-SFM and then subjected to a series of twofold dilutions across a 96-well plate. Virus dilutions were used to infect Vero cell monolayers in a 96-well plate and incubated for 5 to 6 days at 35°C. Following incubation, cell culture media were removed and temporarily stored at 4°C, and the virus-positive cell monolayers were identified by immunoperoxidase staining. Terminal dilution was achieved when ≤ 25% of cell monolayers were positive for virus. Cell culture medium from a positive monolayer at the terminal dilution was subjected to an additional round of terminal dilution. Following the second terminal dilution, virus was amplified in Vero cells (75-cm² flask), collected, and frozen as previously described.

Mutant virus characterization.

Viruses were screened for ts phenotype by assessing plaque formation at the permissive temperature (35°C) and various restrictive temperatures (37, 38, and 39°C) in Vero and HuH-7 cells. Cell monolayers in 24-well plates were inoculated with serial 10-fold dilutions of virus and incubated for 2 h at 35°C and overlaid with L-15 medium (Quality Biologicals, Gaithersburg, Md.) supplemented with 2% FBS, l-glutamine, gentamicin, and 0.8% methylcellulose. Cells were incubated at the indicated temperatures for 5 days in temperature-controlled water baths, the presence of virus was determined by immunoperoxidase staining as described previously (9), and titers are expressed as log₁₀ PFU per milliliter.

The replication of viruses was evaluated in Swiss Webster suckling mice (Taconic Farms, Germantown, N.Y.). Animal experiments were performed in accordance with the regulations and guidelines of the National Institutes of Health, Bethesda, Md. Groups of six 1-week-old mice were inoculated intracerebrally with a 30-μl inoculum containing 10⁴ PFU of virus diluted in VP-SFM. Five days later, mice were euthanized, and the brain of each mouse was removed and homogenized in a 10% (wt/vol) suspension of phosphate-buffered Hank’s balanced salt solution containing 7.5% sucrose, 5 mM sodium glutamate, 0.05 mg of ciprofloxacin per ml, 0.06 mg of clindamycin per ml, and 0.0025 mg of amphotericin B per ml. Clarified supernatants were frozen at −70°C, and virus titers were determined by plaque titration in Vero cells.

RESULTS

Recovery of mutant viruses.

Of 80 full-length DEN4 cDNA constructs containing a single pair of charge-to-alanine mutations, virus was recovered from 32 of these constructs in either Vero or C6/36 cells (Fig. 1). Viruses with mutations in the three putative functional domains and in regions of the gene with no putative function were recovered at similar frequencies.

FIG. 1.

Amino acid sequence of the rDEN4 NS5 gene. Eighty underlined amino acid pairs were mutagenized to alanine pairs; 32 pairs in boldface type represent mutant viruses that could be recovered in either Vero or C6/36 cells; amino acid pairs in normal type represent mutant viruses that could not be recovered in either Vero or C6/36 cells. Boxed regions indicate putative functional domains, including an S-adenosylmethionine-utilizing methyltransferase domain (SAM), an importin-β-binding domain adjacent to a NLS, and an RNA-dependent RNA polymerase domain.

Temperature sensitivity in Vero and HuH-7 cells.

The levels of temperature sensitivity of wt rDEN4, rDEN4Δ30, and the 32 mutant viruses are summarized in Table 1. A mutant virus was considered to have a ts phenotype in Vero or HuH-7 cells if it exhibited a reduction in titer of ≥2.5 log₁₀ PFU/ml or ≥3.5 log₁₀ PFU/ml at the restrictive temperature (39°C) relative to the titer at the permissive temperature (35°C), respectively. The greater reduction in replication at 39°C in HuH-7 cells required for a mutant to be assigned the ts phenotype reflects the greater reduction in replication of wt virus at 39°C in HuH-7 cells versus Vero cells (Table 1). One mutant virus (645-646) was temperature sensitive (ts) in Vero cells but not in HuH-7 cells, and seven mutant viruses were ts in HuH-7 cells but not in Vero cells. Such mutants whose temperature sensitivity is host cell dependent are referred to as temperature-sensitive, host range (tshr) mutants. Thirteen mutant viruses were ts in both cell types, and 11 mutant viruses were not ts on either cell type. Thus, a total of 21 mutant viruses were ts with eight mutant viruses exhibiting an tshr specificity. None of the mutant viruses showed a small-plaque phenotype at the permissive temperature.

TABLE 1.

Temperature-sensitive and mouse brain attenuation phenotypes of viruses bearing charge-cluster-to-alanine mutations in the NS5 gene of DEN4

Mutationa	Changed aa pairb	No. of ntc changed	Mean virus titer (log10 PFU/ml) at the indicated temp (°C)d										No. of suckling mice	Replication in suckling micef
			Vero cells					HuH-7 cells						Mean titer ± SE (log10 PFU/g of brain)	Mean log reduction from wtg
			35	37	38	39	Δe	35	37	38	39	Δ
None (wt) (rDEN4)	NA	0	8.1	8.1	7.9	7.6	0.5	8.3	8.0	7.5	7.5	0.8	48	6.0 ± 0.16
Deletion (rDEN4Δ30)	NA	30	6.3	6.1	6.1	5.7	0.6	6.9	6.3	5.9	4.7	2.2	42	5.4 ± 0.22	0.6
21–22	DR	4	7.2	6.8	6.7	6.1	1.1	7.6	7.1	7.0	4.7	2.9	6	5.0 ± 0.50	0.6
22–23	RK	4	7.0	7.8	6.9	3.7	3.3	7.6	7.6	6.5	<1.7	>5.9	6	2.6 ± 0.19	2.9
23–24	KE	3	6.7	6.6	6.0	6.5	0.2	7.1	7.3	5.6	<1.7	>5.4	18	4.7 ± 0.09	1.5
26–27	EE	3	7.8	7.6	6.8	4.0	3.8	8.4	8.2	7.3	4.9	3.5	6	5.7 ± 0.30	+0.1
46–47	KD	3	7.4	7.4	7.3	7.0	0.4	7.8	7.8	7.3	6.8	1.0	6	5.4 ± 0.42	0.5
157–158	EE	3	6.5	7.2	5.1	5.1	1.4	7.6	7.4	5.9	<1.7	>5.9	6	2.8 ± 0.31	2.7
200–201	KH	4	5.3	4.6	5.3	4.1	1.2	5.6	4.9	3.7	<1.7	>3.9	12	5.5 ± 0.45	0.8
246–247	RH	5	6.9	5.8	5.7	5.4	1.5	6.4	6.1	6.1	5.5	0.9	6	6.1 ± 0.17	+0.5
253,254	EK	4	7.1	6.9	6.8	7.0	0.1	7.9	7.5	7.6	6.8	1.1	6	6.2 ± 0.13	+0.6
356–357	KE	3	7.7	7.6	7.0	7.0	0.7	8.0	7.3	6.4	<1.7	>6.3	6	3.5 ± 0.58	2.0
387–388	KK	5	7.7	6.1	7.0	<1.7	>6.0	7.0	6.3	7.0	<1.7	>5.3	6	3.1 ± 0.33	2.4
388–389	KK	5	5.1	4.5	<1.7	<1.7	>3.4	6.1	5.0	<1.7	<1.7	>4.4	6	5.0 ± 0.23	1.4
396–397	RE	4	7.0	7.3	6.5	5.5	1.5	7.5	7.6	7.5	<1.7	>5.8	18	5.4 ± 0.35	1.1
397–398	EE	2	7.0	7.1	7.0	3.0	4.0	8.0	7.6	7.0	<1.7	>6.3	6	6.0 ± 0.22	0.8
436–437	DK	4	4.5	3.3	3.0	2.0	2.5	5.7	4.5	<1.7	<1.7	>4.0	12	2.3 ± 0.14	3.9
500–501	RE	3	6.6	6.3	5.7	2.3	4.3	7.1	6.5	<1.7	<1.7	>5.4	6	6.9 ± 0.49	+0.7
520–521	EE	3	5.6	4.7	4.3	<1.7	>3.9	6.7	5.7	<1.7	<1.7	>5.0	6	5.2 ± 0.48	0.2
523–524	DK	4	6.6	6.3	6.3	5.8	0.8	7.1	6.6	<1.7	<1.7	>5.4	6	4.2 ± 0.47	1.3
524–525	KK	5	7.1	6.9	6.9	6.6	0.5	7.8	7.4	7.0	5.3	2.5	6	3.4 ± 0.54	2.1
525–526	KD	4	7.8	7.1	7.6	6.8	1.0	7.9	7.7	8.0	6.9	1.0	6	3.7 ± 0.64	1.8
596–597	KD	3	4.6	4.0	2.6	<1.7	>2.9	5.7	4.9	4.0	<1.7	>4.0	6	5.9 ± 0.14	0.5
641–642	KE	4	7.3	6.9	6.9	5.2	2.1	7.8	7.5	7.2	6.9	0.9	6	4.7 ± 0.45	1.2
642–643	ER	3	6.8	6.1	4.0	3.3	3.5	7.5	7.1	6.6	3.0	4.5	12	2.6 ± 0.15	3.6
645–646	EK	4	6.3	5.3	5.9	3.1	3.2	6.4	5.8	5.5	4.5	1.9	6	5.4 ± 0.51	0.2
649–650	KE	3	6.9	6.8	6.9	6.3	0.6	7.1	7.3	7.5	7.0	0.1	12	6.4 ± 0.20	+0.2
654–655	DR	4	6.3	5.7	<1.7	<1.7	>4.6	7.0	7.1	4.6	<1.7	>5.3	12	1.8 ± 0.10	4.0
750–751	RE	3	7.1	7.1	6.9	5.7	1.4	7.8	6.9	6.5	5.6	2.2	6	6.0 ± 0.18	0.7
808–809	ED	3	4.6	4.1	<1.7	<1.7	>2.9	5.2	<1.7	<1.7	<1.7	>3.5	6	1.8 ± 0.05	3.1
820–821	ED	2	6.3	6.3	5.6	<1.7	>4.6	6.9	6.0	5.7	<1.7	>5.2	6	5.5 ± 0.33	1.2
827–828	DK	4	6.9	6.3	6.3	5.9	1.0	7.5	6.9	5.0	<1.7	>5.8	6	3.6 ± 0.76	2.3
877–878	KE	3	7.6	7.3	7.0	7.0	0.6	7.9	7.9	7.3	5.8	2.1	12	4.4 ± 0.65	1.8
878–879	EE	3	7.6	7.3	7.3	7.1	0.5	8.1	8.1	7.9	6.6	1.5	12	2.4 ± 0.10	3.8

^aPositions of the amino acid pair mutated to an alanine pair; numbering starts at the amino terminus of the NS5 protein.

^baa, amino acid; NA, not applicable.

^cnt, nucleotides.

^dUnderlined values indicate a 2.5 or 3.5 log₁₀ PFU/milliliter reduction in titer in Vero or HuH-7 cells, respectively, at the indicated temperatures compared to the titer in cells at the permissive temperature (35°C).

^eReduction in titer (log₁₀ PFU/milliliter) at 39°C compared to the titer at the permissive temperature (35°C).

^fGroups of six suckling mice were inoculated intracerebrally with 4.0 log₁₀ PFU virus in a 30-μl inoculum. The brain of each mouse was removed 5 days later and homogenized, and the virus titer in Vero cells was determined.

^gDetermined by comparing mean viral titers in mice inoculated with sample virus and concurrent wt controls (n = 6). The attenuation phenotype is defined as a reduction of ≥1.5 log₁₀ PFU/g compared to wt virus; reductions of ≥1.5 log₁₀ PFU/g are listed in boldface type.

The magnitude of the reduction in virus titer in Vero cells at 39°C correlated with that observed in HuH-7 cells (Kendall rank correlation, P = 0.02). Viruses with mutations in the region containing the NLS and the importin-β-binding domain (i.e., between amino acids 319 and 404) showed a significantly larger reduction in titer at the restrictive temperature in HuH-7 cells than viruses with mutations in the polymerase domain (Tukey post-hoc test, P < 0.05) and a larger reduction (though not significantly larger) than viruses with mutations in regions with no identified function.

Attenuation in suckling mice.

Mutant viruses showed a wide range (0 to 10,000-fold) of restricted replication in suckling mouse brain (Table 1). Fourteen mutant viruses were attenuated in suckling mouse brain, arbitrarily defined as a ≥1.5 log₁₀ reduction in virus titer. There was no correlation between attenuation in the mouse brain and temperature sensitivity in either Vero cells (Kendall rank correlation, P = 0.77) or HuH-7 cells (Kendall rank correlation, P = 0.06), and there was no association between attenuation in mouse brain and location of mutations in NS5 functional domains (by analysis of variance, df = 2, F = 0.39, P = 0.67).

One of the reasons for selecting the NS5 protein for mutagenesis is that it is the most highly conserved protein among the four DEN serotypes, and a nucleotide sequence identified in DEN4 might be conserved in one or more of the other serotypes. Table 2 indicates the conservation of each attenuating pair of mutations identified in DEN4 (GenBank accession number AF326825) in DEN1, DEN2, and DEN3 (GenBank accession numbers U88535, AF038403, and M93130, respectively).

TABLE 2.

Sites at which attenuating mutations introduced into DEN4 are conserved in DEN1, DEN2, and/or DEN3

DEN4 mutationa	aab pair	Mean log reduction in suckling micec	Identity in DEN serotyped:
			1	2	3
22–23	RK	2.9	−−	−−	++
23–24	KE	1.5	−+	−+	++
157–158	EE	2.7	++	+−	++
356–357	KE	2.0	++	++	++
387–388	KK	2.4	−+	++	−+
436–437	DK	3.9	−−	++	++
524–525	KK	2.1	+−	++	+−
525–526	KD	1.8	−−	+−	−−
642–643	ER	3.6	++	−−	−−
654–655	DR	4.0	−+	−+	−+
808–809	ED	3.1	++	++	++
827–828	DK	2.3	++	++	++
877–878	KE	1.8	−+	−+	−+
878–879	EE	3.8	+−	++	++

^aPosition of the amino acid pair mutated to an alanine pair; numbering starts at the amino terminus of the NS5 protein.

^baa, amino acid.

^cDetermined by comparing mean viral titers (log₁₀ PFU/ml) in mice inoculated with sample virus and concurrent wt controls (n = 6).

^dIdentical (+) or different (−) amino acid compared to each position in the DEN4 charged amino acid pair; for example, ++ indicates that both amino acids in the specified serotype are identical to those in DEN4, −− indicates that both amino acids in the specified serotype are different from those in DEN4.

Combination of charge-to-alanine mutation pairs.

We sought to determine whether the restricted replication of a mutant virus in mouse brain could be augmented by the addition of a second mutation. This was achieved by combining two pairs of mutations into one recombinant virus (Table 3). Mutations that conferred only partial restriction in replication in mouse brain were combined, permitting the detection of an additive effect of the combined mutations. A recombinant virus bearing mutation pairs 23-24 and 200-201 was more ts in both HuH-7 and Vero cells but was not more attenuated in vivo. In contrast, a recombinant virus bearing mutation pairs 23-24 and 396-397 was more ts in Vero cells and more attenuated than either parent virus but was not more ts in HuH-7 cells.

TABLE 3.

Temperature-sensitive and mouse brain attenuation phenotypes of double charge-cluster-to-alanine mutants of the NS5 gene of rDEN4

Mutation(s)a	Changed aab pair	No. of ntc changed	Mean virus titer (log10 PFU/ml) at the indicated temp (°C)d										No. of suckling mice	Replication in suckling micef
			Vero cells					HuH-7 cells						Mean virus titer ± SE (log10 PFU/g of brain)	Mean log reduction from wtg
			35	37	38	39	Δe	35	37	38	39	Δ
None (wt) (rDEN4)	NA	0	8.1	8.1	7.9	7.6	0.5	8.3	8.0	7.5	7.5	0.8	48	6.0 ± 0.16
Deletion (rDEN4Δ30)	NA	30	6.3	6.1	6.1	5.7	0.6	6.9	6.3	5.9	4.7	2.2	42	5.4 ± 0.22	0.6
23–24	KE	3	6.7	6.6	6.0	6.5	0.2	7.1	7.3	5.6	<1.7	>5.4	18	4.7 ± 0.09	1.5
200–201	KH	4	5.3	4.6	5.3	4.1	1.2	5.6	4.9	3.7	<1.7	>3.9	12	5.5 ± 0.45	0.8
396–397	RE	4	7.0	7.3	6.5	5.5	1.5	7.5	7.6	7.5	<1.7	>5.8	18	5.4 ± 0.35	1.1
23–24; 200–201	KE, KH	7	7.1	6.5	6.6	<1.7	>5.4	7.8	7.3	<1.7	<1.7	>6.1	6	5.8 ± 0.16	0.6
23–24; 396–397	KE, RE	7	6.3	4.9	<1.7	<1.7	>4.6	7.1	6.0	5.6	<1.7	>5.4	6	3.7 ± 0.44	2.7

^aPositions of the amino acid pair mutated to an alanine pair; numbering starts at the amino terminus of the NS5 protein.

^baa, amino acid; NA, not applicable.

^cnt, nucleotides.

^dUnderlined values indicate a 2.5 or 3.5 log₁₀ PFU/milliliter reduction in titer in Vero or HuH-7 cells respectively, at the indicated temperatures compared to the titer in cells at the permissive temperature (35°C).

^eReduction in titer (log₁₀ PFU/milliliter) at 39°C compared to the titer at the permissive temperature (35°C).

^fGroups of six suckling mice were inoculated intracerebrally with 4.0 log₁₀ PFU of virus in a 30-μl inoculum. Brains were removed 5 days later and homogenized, and the virus titers in Vero cells were determined.

^gDetermined by comparing mean viral titers in mice inoculated with sample virus and concurrent wt controls (n = 6); reductions of ≥1.5 log₁₀ PFU/g are listed in boldface type.

DISCUSSION

Development of an effective live attenuated DEN vaccine may require fine-tuning the level of attenuation of one or more of the four recombinant components to achieve a satisfactory balance between attenuation and immunogenicity. To that end, we have assembled a menu of mutations that might be useful for the further attenuation of DEN4 as well as antigenic chimeric viruses built upon a DEN4 genetic background. In the current study, we have generated a set of mutant viruses using charge-cluster-to-alanine mutagenesis of the DEN4 NS5 gene. Concurrently, we have used 5-fluorouracil mutagenesis to create additional attenuated mutant viruses (5). However, site-directed mutagenesis of the NS5 gene offers an important complementary approach to random mutagenesis for the generation of attenuating mutations since the genetic basis of the phenotype of each mutant is known. Furthermore, charge-cluster-to-alanine mutant viruses possess some characteristics particularly desirable for vaccine design. First, the relevant mutations can be designed solely in the genes encoding the nonstructural proteins of DEN4 and can therefore be used to attenuate DEN1, DEN2, and DEN3 antigenic chimeric recombinants built upon an attenuated DEN4 genetic background. Second, the phenotype is usually specified by three or more nucleotide changes, rendering the likelihood of reversion of the mutant sequence to the wt sequence less than for that of a single point mutation. Finally, charge-cluster-to-alanine attenuating mutations can easily be combined among themselves or with other attenuating mutations to augment the overall level of attenuation of the recombinant virus.

Because an animal model that accurately mimics DF is not available, it is necessary to rely on in vitro and in vivo biological markers to identify mutants that might be attenuated in humans. Previous studies have suggested that the phenotypes most closely correlated with attenuation in humans are small plaque size (3), temperature sensitivity (3, 10), attenuation for newborn mice (3), and most reliably, decreased viremia in monkeys (3, 9). Since it is not feasible to test large numbers of mutant viruses in monkeys, we have used other biological markers in our selection of mutants for further consideration. Thus, the NS5 gene mutants were tested for their sp, ts, and mouse brain attenuation phenotypes. Thirty-two NS5 mutant viruses (40%) were recovered from the 80 DEN4 charge-to-alanine mutant cDNA constructs, and they displayed several phenotypes of interest: ts (66%), tshr (25%), and restriction in replication in suckling mouse brain (44%). Mutations in these recovered viruses showed no apparent clustering with respect to their position in the gene or their presence in a putative functional domain. None of the DEN4 mutant viruses displayed an sp phenotype at the permissive temperature. In contrast, after charge-cluster-to-alanine mutagenesis of the NS1 gene of a related flavivirus, yellow fever (YF) virus, 69% of 28 mutant cDNAs yielded virus from transfected human carcinoma (SW-13) cells and lethal mutations clustered predominantly in the N-terminal half of the gene (29). Only one (5%) of these 19 mutant YF virus mutants was convincingly ts, but 7 (37%) were sp viruses.

The strikingly higher percentage of recovery and frequency of sp mutants of YF virus NS1 versus DEN4 NS5 mutant viruses suggest that the NS1 activity threshold may be less dichotomous than that of NS5, thus allowing for the recovery of partially defective NS1 mutant viruses as manifested by an altered plaque size. However, the lack of sp NS5 mutant viruses and the decreased frequency of recovery suggest that fully functional NS5 is required for virus replication. The lower incidence of temperature sensitivity in NS1 relative to NS5 mutant viruses also suggests that the NS1 structure is either stable at elevated temperatures or protected by dimerization (31, 32, 40) or direct membrane association (40), neither of which has been shown to occur in NS5.

The mutant viruses created in this study might be useful in characterizing the various functions of the DEN NS5 gene. For example, in this study, mutations in the region containing the NLS and the importin-β-binding domain of DEN4 NS5 resulted in a greater reduction in titer at the restrictive temperature in both cell types than mutations in the polymerase domain or mutations in regions with no known function, although this effect was significant only in HuH-7 cells. The function of this region of NS5 has only recently been identified (13, 16), and this study further supports the importance of the region containing the NLS and the importin-β-binding domain in DEN replication. It is possible that the impact of some charge-to-alanine mutations was lessened by the incorporation of compensatory mutations elsewhere in the genome. Because it would not be possible to distinguish such compensatory mutations from other, incidental mutations via nucleotide sequencing, we did not sequence the genome of recovered mutant viruses. Instead, we plan to import selected mutations into several genetic backgrounds and evaluate each recovered virus in vitro and in vivo as described above. In this manner, the phenotype conferred by each isolated mutation can be confirmed independently of its genetic background.

Because vaccine candidate 2AΔ30 retains mild residual virulence for the liver but is otherwise safe and immunogenic, mutations that restrict replication in the liver without affecting overall replication would be especially useful for fine-tuning attenuation. Among the DEN4 charge-to-alanine mutant viruses, the magnitude of reduction in titer at the restrictive temperature in Vero cells showed a positive correlation with reduction in HuH-7 cells. Nevertheless, seven of the mutant viruses showed a ts phenotype in HuH-7 cells but not in Vero cells. Blaney et al. (5) identified a similar class of mutant viruses after chemical mutagenesis. Experiments are currently under way to determine whether mutations that confer a liver cell-specific phenotype in vitro also confer attenuation in vivo in SCID mice implanted with human HuH-7 cells.

Importantly, 14 of the 32 mutant viruses were attenuated in suckling mouse brain. However, temperature sensitivity in cell culture did not correlate with reduced replication in vivo in suckling mouse brain. Half of the mutant viruses that were ts in at least one cell type were not attenuated in mice, and 4 of the 14 mutant viruses that were attenuated in mice were not ts in either cell type. These latter non-ts attenuating mutations may be especially useful for vaccine design, especially in conjunction with ts attenuating mutations because they increase the stability of the attenuation phenotype following replication in vitro and in vivo (28). Additionally, the level of attenuation specified by a given pair of charge-cluster-to-alanine mutations was augmented by the insertion of another attenuating pair of mutations. Thus, the range of attenuation specified by this set of mutations could be extended by using various combinations of two or more pairs of mutations. Future studies will evaluate the compatibility of a subset of charge-to-alanine mutations with the Δ30 mutation in DEN4, the effects of combining charge-to-alanine mutations in NS5 with mutations in other genes identified through chemical mutagenesis (5), and finally, the effects of particularly promising charge-to-alanine mutations on attenuation of virus in monkeys.

As expected, charge-to-alanine mutagenesis created mutant viruses with a diverse array of phenotypes, and combination of mutations further modulated these phenotypes. It should therefore be possible to modify DEN4 virus vaccine candidates with these mutations to eliminate side effects such as ALT elevations. Moreover, many of the sites of attenuating mutations identified in this study show a high degree of conservation across the four DEN serotypes. For example, mutations 157-158, 808-809, and 827-828 each confer at least 100-fold restriction in replication in mouse brain, and both charged amino acids in these mutation pairs are conserved in all four dengue serotypes. Thus, some of the attenuating mutations identified in this study may be useful in attenuating rDEN1, rDEN2, and rDEN3 by their insertion into conserved regions, as has been described for parainfluenza viruses (35).