DENV has an ssRNA genome of positive polarity, approximately 11 kilobases in length, with a 5′ cap but no poly(A) tail at the 3′ end (reviewed by Lindenbach & Rice, 2003). The single long ORF of the viral RNA encodes a polyprotein that is processed co- and post-translationally by both cellular signal peptidase and viral serine protease in the endoplasmic reticulum (ER) membrane into three structural proteins, capsid (C), precursor membrane (prM) and envelope (E), and seven non-structural (NS) proteins: NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5 (reviewed by Beasley, 2005; Harris et al., 2006).

Flavivirus NS proteins play an essential role in viral RNA translation and replication. Some NS proteins have multiple functions in the virus life cycle (Miller et al., 2010; Padmanabhan et al., 2006). The N-terminal 180 residues of NS3, in conjunction with NS2B, function as a viral protease required for polyprotein processing (reviewed by Padmanabhan & Strongin, 2010; Sampath & Padmanabhan, 2009). The NS3 carboxy-terminal region has conserved motifs found in the ‘D-E-X-H’ family of RNA helicases (reviewed by Lescar et al., 2008). The amino-terminal region of NS5 has a 5′ RNA methyltransferase (MTase) activity, which catalyses sequentially the transfer of a methyl group from S-adenosylmethionine to the N7 of the 5′ guanosine cap (Ray et al., 2006; reviewed by Davidson, 2009; Dong et al., 2008) and to the 2′-OH moiety of adenosine at the 5′ end of the RNA genome (Egloff et al., 2002). NS5 also has RNA-dependent RNA polymerase (RdRP) activity (Tan et al., 1996), capable of de novo synthesis of the minus strand in vitro (Ackermann & Padmanabhan, 2001), and is required for virus replication in cultured cells (Khromykh et al., 1998, 1999; reviewed by Malet et al., 2008; Miller et al., 2010).

Replicons encoding a reporter gene are useful tools for studying the translation and replication of flaviviruses, including Kunjin virus (Khromykh & Westaway, 1997), West Nile virus (WNV) (Shi et al., 2002), yellow fever virus (Corver et al., 2003; Jones et al., 2005a), DENV2 (Alvarez et al., 2005; Jones et al., 2005b; M. I. M. Manzano, E. D. Reichert, S. L. Alcaraz-Estrada, S. Polo, B. Falgout, W. Kasprzak, B. A. Shapiro & R. Padmanabhan, unpublished results) and DENV1 (Puig-Basagoiti et al., 2006). However, to our knowledge, the construction of a reporter replicon for DENV4 has not yet been reported.

We sought to construct a DENV4 replicon with Renilla luciferase (R_luc) reporter to study mechanisms of virus replication. A DENV4 infectious clone was placed into an Escherichia coli–yeast shuttle vector to yield the pRS424-DENV4IC clone by using a previously described method (Polo et al., 1997). The pRS424-DENV4IC clone was used for the construction of a DENV4 replicon (DENV4Rep) as shown in Fig. 1(a) [see Supplementary Fig. S1 (available in JGV Online) for details].

The DENV4 replicon contains the 5′UTR and the first 25 codons of the capsid (C) gene, fused to the ORF of the R_luc reporter with its own initiation and termination codons. This region of the C gene contains the 5′ conserved sequence (CS), which interacts with the 3′CS1 in the 3′UTR (Hahn et al., 1987). Expression of the R_luc reporter is under cap-dependent translational control. The 3′ end of the R_luc gene is fused to the encephalomyocarditis virus internal ribosome entry site (IRES) to direct cap-independent translation of the long ORF encoding the viral NS proteins (Fig. 1a). The region downstream from the IRES contains the C-terminal 37 codons of the viral envelope (E) gene that is required for proper translocation of NS1 and other NS proteins into the ER. In addition, we constructed a DENV4 replicon containing the GDD→GND mutation in the RdRP gene, which is known to abolish viral RNA replication (Khromykh et al., 1998; Ribas & Wickner, 1992).

The time course of R_luc expression was studied in Vero cells. RNAs from DENV4rep wild type (WT) and mutants were obtained by in vitro transcription (for details, see Supplementary Methods, available in JGV Online). Vero cells (approx. 10⁶) were transfected with the RNAs (approx. 3 μg) by electroporation. Cells were plated into six wells of a 48-well plate for R_luc assays (100 μl per well) and the rest into a 12-well plate containing 1 ml medium for RT-PCR.

Transient expression of DENV4rep and DENV4rep-GND in transfected cells was monitored by R_luc activity assays with lysates of transfected cells collected at different times post-transfection (p.t.) (Fig. 1b). The first peak of R_luc activity occurred at 2 h p.t. due to translation, and then declined. At 24 h p.t., the R_luc signal from WT DENV4rep began to increase until it reached a maximum at 96 h p.t., due to translation of the replicated viral RNA. However, for the DENV4rep-GND mutant replicon, the R_luc activity reached a peak at 2 h p.t. due to translation, but decreased steadily to background levels due to a deficiency in replication (Fig. 1b).

Replication of the WT DENV4rep was confirmed by immunofluorescence assay of NS1 in the WT replicon-transfected cells at 96 h p.t., but was negative in DENV4rep-GND- or non-transfected cells (Fig. 1c). Finally, replication of DENV4rep was confirmed by RT-PCR using primers specific for the DENV4 NS5 gene at 96 h p.t. Again, a PCR-amplified product of the expected size was obtained with DENV4rep but not with the DENV4rep-GND replicon RNA-transfected cells (Fig. 1d). Taken together, these results indicated that the DENV4rep RNA is functional and is suitable to study virus translation and replication by measuring luciferase activity at 2 and 96 h p.t., respectively.

Temperature-sensitive (ts) mutants of a gene provide a powerful tool for studying protein function because they show a marked change in the level or activity of the gene product when expressed at a non-permissive temperature. At permissive temperatures, the phenotype of the mutant is close to that of the WT. For YFV, a ts mutant of NS1 shown to be defective in accumulation of viral RNA at a non-permissive temperature revealed a role of this protein in viral RNA replication (Muylaert et al., 1997). One commonly used method to generate ts mutants is by chemical mutagenesis, followed by screening of a large number of progeny for identification of a particular phenotype resulting from either loss or gain of function. Blaney and colleagues isolated ts mutants by treatment of DENV4-infected cells with the mutagen 5-fluorouracil and characterized mutants that exhibited ts phenotypes in both Vero and HuH-7 cells (Blaney et al., 2001; Hanley et al., 2002). Several attenuated ts mutations affecting virus growth at non-permissive temperatures were identified. However, as some of the ts mutants had substitutions of multiple residues in NS genes, the mutation causing a ts phenotype could not be ascribed to a single residue. Moreover, it was not known at what step of the virus life cycle the ts phenotype was manifested.

We focused our efforts to study ts mutations in NS3 and NS5, as these proteins play important roles in virus replication (Gamarnik, 2010; Miller et al., 2010; Padmanabhan et al., 2006). To identify mutations of single residues contributing to the ts phenotypes, six single-residue substitution mutants of DENV4rep were constructed (see Supplementary Table S2, available in JGV Online). WT and mutant replicon RNAs were synthesized by in vitro transcription of linearized replicon plasmids catalysed by phage SP6 RNA polymerase. Equal amounts of replicon RNAs were transfected by electroporation into Vero cells and the transfected cells were incubated at 35, 37 and 39 °C. Cells were lysed at 2 and 96 h p.t. and luciferase activities were measured. At 2 h p.t., the luciferase activities derived from each mutant at different temperatures did not differ to an appreciable extent from those obtained from the WT DENV4rep-transfected cells (Fig. 2a), suggesting that none of the mutations exhibited a ts phenotype for viral RNA translation. In contrast, the ts mutations S158P in the NS3 protease (NS3pro) domain at 39 °C and N96I in the NS5 MTase domain at 37 and 39 °C showed significantly reduced luciferase activity (P<0.001, Mann–Whitney test) compared with the DENVrep-WT replicon at 96 h p.t. (Fig. 2a).

To confirm the results from the R_luc activity assays done at 96 h p.t., the levels of replicated viral RNA from the WT and mutant replicons were determined by semiquantitative RT-PCR. The results were in good agreement with the R_luc signals, as seen from the significant reduction of RNA levels at 37 and 39 °C for the NS5 N96I mutant, as well as for the two NS3pro mutants, L128F and S158P, at 39 °C (Fig. 2b). Thus, the results taken together indicate that the single-residue substitution mutants in the NS3pro domain and the NS5 MTase domain exhibited ts phenotypes for RNA synthesis, with no appreciable effect on viral translation.

The NS5 MTase domain catalyses transfer of a -CH₃ group from S-adenosylmethionine to the N7 of the guanosine cap (N7 MTase) and to the 2′-OH of the first 5′ nucleotide of RNA (2′-O MTase). The 2′-O MTase activity of DENV2 NS5 was first demonstrated using capped and uncapped short RNAs (GpppACCCCC and ^7-MeGpppACCCCC) as substrates (Egloff et al., 2002). Subsequently, Ray et al. (2006), using WNV 5′-terminal RNA as substrate for a -CH₃ acceptor, revealed that specific nucleotide sequence and 5′ stem–loop structure of the RNA are required for N7-methylation and that both N7- and 2′-O-Me additions are sequential reactions catalysed by the NS5 MTase (Dong et al., 2007; Ray et al., 2006). N7-methylation is essential for WNV replication, whereas the defect in 2′-O-methylation attenuates the virus (Zhou et al., 2007). We sought to investigate whether the ts N96I mutation of NS5 affects the activity of N7 cap- and/or 2′-O-methylation. To this end, the WT and N96I mutant NS5 MTase domains, each with a His tag at the N terminus, were expressed in a soluble form in E. coli and the recombinant proteins were purified (data not shown). The N7 and 2′-O MTase activities of these recombinant proteins were assayed at different temperatures as described previously (Zhou et al., 2007). N7 cap-methylation, usually evaluated at 22 °C, was reduced significantly (P<0.001, Mann–Whitney test) in the ts mutant NS5 MTase domain at 37 and 39 °C, suggesting that the N96I mutation causes a ts phenotype for NS5 function in N7 MTase activity (Fig. 3a), although the 2′-O MTase was not affected significantly (P=0.1, Mann–Whitney test; Fig. 3b). Further studies are required to understand the molecular basis of the N96I mutation in conferring a ts phenotype on DENV4 NS5 function.