A two-plasmid strategy for engineering a dengue virus type 3 infectious clone from primary Brazilian isolate

Dengue infections represent one of the most prevalent arthropod-borne diseases worldwide, causing a wide spectrum of clinical outcomes. Engineered infectious clone is an important tool to study Dengue virus (DENV) biology. Functional full-length cDNA clones have been constructed for many positivestrand RNA viruses and have provided valuable tools for studying the molecular mechanisms involved in viral genome replication, virion assembly, virus pathogenesis and vaccine development. We report herein the successful development of an infectious clone from a primary Brazilian isolate of dengue virus 3 (DENV3) of the genotype III. Using a two-plasmid strategy, DENV3 genome was divided in two parts and cloned separately into a yeast-bacteria shuttle vector. All plasmids were assembled in yeast by homologous recombination technique and a full-length template for transcription was obtained by in vitro ligation of the two parts of the genome. Transcript-derived DENV3 is infectious upon transfection into BHK-21 cells and in vitro characterization confirmed its identity. Growth kinetics of transcript-derived DENV3 was indistinguishable from wild type DENV3. This system is a powerful tool that will help shed light on molecular features of DENV biology, as the relationship of specific mutations and DENV pathogenesis.


INTRODUCTION
Dengue viral infections are a major global health issue with estimates suggesting that 50-100 million individuals are infected annually Harris 2008, Monath 1994). With no vaccine or antiviral therapy available, Dengue virus (DENV) is responsible for a wide spectrum of clinical outcomes, raging from asymptomatic infections to undifferentiated fever, classical dengue fever (DF) or dengue hemorrhagic fever/dengue shock syndrome (DHS/DSS). Four serotypes (DENV1-4) have been identified and are transmitted to humans by Aedes spp. mosquitoes (Galler et al. 2011, Halstead 2007, Noble et al. 2010. DENV belongs to the Flavivirus genus in the Flaviviridae family of enveloped viruses. Other members in this genus include yellow JEFFERSON J.S. SANTOS et al. fever virus (YFV), West Nile virus (WNV), tickborn encephalitis virus (TBEV) and Japanese encephalitis virus (JEV). The flavivirus genome is organized as a single-stranded positive sense RNA molecule of approximately 11 kb. It consists of a 5' type 1,7-methyl-guanosine cap structure followed by a 5' untranslated region (5'-UTR), a single open reading frame (ORF) and a non-polyadenilated 3'-UTR (Chambers et al. 1990, Clyde et al. 2006. The virus genomic RNA is translated monocistronically into a polyprotein that is co-and posttranslationally processed by both viral and cellular proteases into three structural (C, prM and E) and seven nonstructural (NS1, NS2a, NS2b, NS3, NS4a, NS4b and NS5) proteins (Lindenbach et al. 2007).
Many advances in flavivirus biology have been facilitated by the use of reverse genetics, a powerful method that allows directed genetic manipulation of an RNA virus (Alvarez et al. 2008, Laurent-Rolle et al. 2010, Leung et al. 2008, McElroy et al. 2006, Ruggli and Rice 1999. Several infectious cDNA clones have been described for flaviviruses: YFV (Bredenbeek et al. 2003, Rice et al. 1989, WNV (Yamshchikov et al. 2001), JEV (Pu et al. 2011, Sumiyoshi et al. 1992, Yun et al. 2003, DENV1 (Puri et al. 2000, Suzuki et al. 2007), DENV2 (Kapoor et al. 1995, Polo et al. 1997, Pu et al. 2011), DENV3 (Blaney et al. 2004, Messer et al. 2012 and DENV4 (Lai et al. 1991). Subsequently, self-replicating RNAs termed replicons were also developed (Jones et al. 2005, Khromykh andWestaway 1997). Those subgenomic replicons carry a deletion in the structural coding region of the virus genome, making them unable to assemble virion particles on their own while remaining fully replicationcompetent. Uncoupling RNA replication from virion assembly and maturation allows the direct identification of the role of nonstructural proteins in virus pathogenesis (Hershkovitz et al. 2008, Jones et al. 2005, Khromykh and Westaway 1997. Yet, engineering flavivirus genomes particularly full-length infectious clones remain a difficult challenge.
The main obstacle to the development of these systems is an intrinsic instability of flavivirus fulllength cDNA clones propagated in Escherichia coli (Polo et al. 1997, Ruggli andRice 1999). The reasons of this toxicity are not completely understood, but some reports have suggested that the genome sequence at the E/NS1 region or the expression of a deleterious product from it causes instability in E. coli (Blaney et al. 2004, Polo et al. 1997. Distinct approaches have been assessed to overcome genome instability, including (i) the cloning of the genome in separate fragments and recovering a full-length infectious clone by in vitro ligation (Kapoor et al. 1995, Messer et al. 2012, Sumiyoshi et al. 1992; (ii) the use of low-copy number vectors (Bredenbeek et al. 2003) or (iii) Bacterial Artificial Chromosomes (BACs) (Suzuki et al. 2007); (iv) the use of different E. coli strains (Bredenbeek et al. 2003, Sriburi et al. 2001 or (v) the cloning in yeast cells (Polo et al. 1997, Puri et al. 2000. In the present study, we describe the successful construction of a DENV3 infectious clone from a primary isolate from Recife, Brazil. We overcame the genome instability of DENV3 by splitting the genome in two fragments and cloning them separately in yeast. A full-length DENV3 infectious clone was then restored by in vitro ligation. PLASMID CONSTRUCTIONS DENV3 genotype III genome was split in two parts and cloned separately in a plasmid vector (Fig. 1).

CELL LINES, YEAST
All subgenomic fragments were amplified by PCR with overlapping sequences at their extremities from independent isolated cDNA clones using the KlenTaq LA polymerase mix (Clontech), a high fidelity DNA polymerase enzymes mix. The pBSC-IC-DENV3-pA (nt 1-2464) was assembled from a single subgenomic fragment and contained the genome sequence of the 5'-UTR, C, prM, E and NS1 (the first 51 nt). The 5'-end primer used on this PCR reaction included an RsrII restriction site, a T7 promoter recognition 1 -Schematic diagram for the construction of a DENV3 infectious clone using a two-plasmid strategy. DENV3 genome was divided and cloned as two separate fragments with the yeast shuttle vector pBSC by homologous recombination in yeast cells. All subgenomic fragments were amplified by PCR with overlapping sequences. The pBSC-IC-DENV3-pA was assembled from a single subgenomic fragment and contains the genome sequence encoding the 5'-UTR, C, prM, E and NS1 (the first 51 nt). The pBSC-IC-DENV3-pB was generated from three overlapping subgenomic fragments (F1, F2 and F3). It contains the rest of DENV3 genome, spanning from the remaining sequence of NS1 to the 3'-UTR. Both constructs have a common BamHI restriction site at nt 2459. Not drawn to scale. site and an additional G immediately before the first base of DENV3 genome. The pBSC-IC-DENV3-pB was generated from three overlapping subgenomic fragments (F1, F2 and F3, see Table I). It contained the rest of DENV3 genome, spanning from the remaining sequence of NS1 to the 3'-UTR. Both constructs had a common BamHI restriction site at nt 2459. A list of oligonucleotide primers used on the construction of the infectious clone is shown in Table I. Oligonucleotides primers were designed at highly conserved regions in the virus genome. Those regions were identified by aligning multiple, related sequences of dengue virus 3 available on the GenBank. The two constructs were assembled into the yeast-bacteria pBSC shuttle vector through recombination among fragments at homologous regions in yeast (Gibson 2009, Polo et al. 1997, and recombinant clones were selected in drop-out YNB medium lacking tryptophan. Individual clones were screened by PCR and sequence analyses, and positive clones were propagated only in yeast to prevent any possible instability of their inserts in bacteria. Plasmid sequences were deposited in the GenBank under the accession numbers KC425217 and KC425218.

TRANSCRIPTION
After plasmid DNA purification from yeast cells, both parts of DENV3 genome were amplified by PCR using the KlenTaq LA polymerase mix (Clontech). The 5' portion of virus genome was amplified from pBSC-IC-DENV3-pA with the same set of primers used on the construction of the plasmid. The 3' portion was amplified from pBSC-IC-DENV3-pB. To produce a 3'-terminus identical to the virus genome, the 3'-end primer was designed to flank exactly at the end of DENV3 genome, leaving no extraneous bases in the final RNA transcript. PCR products were purified, digested with restriction endonuclease BamHI and re-purified. Purified PCR products were joined together at a molar ratio of 1:1 with T4 DNA ligase to yield a full-length DENV3 cDNA infectious clone, L-IC-DENV3 L42. The in vitro ligation mixture was purified by phenol-chloroform extraction, followed by ethanol precipitation. Full-length RNA transcripts were in vitro synthesized using the MEGAScript T7 kit (Ambion) with the addition of a 7-methyl-guanosine cap analog (Ambion), following manufacturer's instructions. An aliquot from the reaction was analyzed by formaldehydeagarose gel electrophoresis and RNA transcripts were used without further processing.

RNA TRANSFECTION AND IMMUNOFLUORESCENCE ASSAY (IFA)
RNA transcripts were introduced into BHK-21 cells by electroporation using the ECM 830 electro cell manipulator (BTX Harvard Apparatus). 2 x 10 6 BHK-21 cells were trypsinized and washed twice with serum-free medium, resuspended in 100 µl Cytomix solution (Ansari et al. 2004), and mixed with 5 µg of transcribed RNA. Cells were transferred to an ice-cold electroporation cuvette with 0.2-cm gap. Electroporation was performed with 2 pulses of 100 µs at 1200 V separated by 1 s interval. As negative control, BHK-21 cells were transfected under the same conditions except no RNA was added (mock). After electroporation, cells were allowed to recover for 10 min at room temperature and resuspended in complete growth medium. For immunofluorescence analyses, transfected cells were seeded onto glass coverslips. At different time points, RNA-transfected cells in glass coverslips were rinsed with PBS, fixed with 50% cold acetone (v/v) in PBS for 5 min at 4 °C and air-dried. Fixed cells were incubated with a 1:100 dilution of a polyclonal hyperimmune mouse ascitic fluid (HMAF) for 1 h at 37 °C. This HMFA was raised against group B flaviviruses, including YFV and the four DENV serotypes. After incubation with primary antibody, cells were rinsed twice with PBS and incubated for 1 h at 37 °C with a 1:100 dilution of fluorescein isothiocyanate (FITC)conjugated anti-mouse IgG antibody (Sigma-Aldrich), developed in goat. Then, cells were rinsed twice with PBS and the nuclei stained with a Hoechst 33258 solution (Sigma-Aldrich). Prior to visualization by fluorescent microscopy, cell were washed twice with ddH 2 0, air-dried and mounted. Immunofluorescence analyses were carried out on a DMI 4000B inverted microscope (Leica).

VIRUS TITRATION AND GROWTH CURVES
Virus titration was determined by focus-forming assay in mosquito C6/36 cells, as described previously (Das et al. 2007) with modifications. Briefly, 2 x 10 5 cells per well were seeded in 24well cell culture dishes 48 h prior to conducting the assay. Cell monolayers were incubated at 30 °C with 0.2 mL of serial dilutions of each virus. After removing the inoculum, cells were covered with 1 mL of L-15 overlay medium, containing 1% carboxymethyl cellulose (CMC), 2% FBS and 1% antibiotic solution, and incubated at 30 °C. On day 5, overlaid medium was removed. Cells were fixed with 30% cold acetone (v/v) in PBS for 13 min at room temperature and washed once with PBS. Fixed cells were incubated for 1 h at 37 °C with a HMAF diluted 1:100 in ligation buffer. Cells were rinsed with wash buffer and incubated with a 1:500 dilution in ligation buffer of a horseradish peroxidase (HRP)conjugated recombinant protein G (Invitrogen) for 1 h at 37 °C. Cells were rinsed with wash buffer and foci developed by adding AEC substrate buffer. Foci were counted and virus titers, expressed in focus-forming unit (FFU) per mL, were determined JEFFERSON J.S. SANTOS et al. by standard methods. For growth curves, 5 x 10 5 mosquito C6/36 cells per well in a 24-well cell culture dishes were infected in duplicate with WT-DENV3 and transcript-derived DENV3 at multiplicity of infection (MOI) of 0.1. Cells were incubated at 30 °C for 1 h. After removal of inoculum, cells were washed twice with PBS and 1 mL of growth medium was added to each well. Aliquots of the supernatant of infected cells were removed daily and stored at -70 °C. DENV3 titers in each sample were determined by titration in mosquito C6/36 cells.

STATISTICAL ANALYSES
Statistical analyses were performed using the GraphPad Prism 4 for Macintosh (GraphPad Software, Inc). Growth kinetics differences were analyzed by a two-way analysis of variance (ANOVA) with a mixed linear model for repeated measures, followed by a Bonferonni's corrected post hoc test for pairwise comparisons at each time point. The differences were considered statistically significant at a P level of ≤0.05.

RESULTS
The shuttle plasmid pBSC was used as a vector for the construction of DENV3 infectious clone. To prevent genome instability, DENV3 genome was divided at the E/NS1 region and cloned as two separate fragments by homologous recombination in yeast cells, generating plasmids pBSC-IC-DENV3-pA and pBSC-IC-DENV3-pB (Fig. 1). Successful construction of the plasmids was confirmed by amplification and DNA sequencing. No mutation was identified in plasmid pBSC-IC-DENV3-pA. In the plasmid pBSC-IC-DENV3-pB, sequence analyses showed one mutation at position 7878 (A7878G) leading to a change in NS5 amino acid sequence (K105R).
To assembly the full-length DENV3 cDNA infectious clone in vitro, genome fragments cloned into plasmids pBSC-IC-DENV3-pA and pBSC-IC-DENV3-pB were amplified by PCR, digested at BamHI site, and mixed together in a molar ratio of 1:1. In vitro ligation mixtures examined on agarose gel electrophoresis showed an expected band, corresponding to the full-length L-IC-DENV3 L42 cDNA. Even though ligation reactions were optimized to produce high yields of full-length DENV3 cDNA, other undesired ligation products and non-ligated fragments were also observed (data not shown).
Assembled full-length DENV3 templates were in vitro transcribed and resulting full-length DENV3 transcripts were introduced into BHK-21 cells by electroporation. Using immunofluorescence assay (IFA), DENV3 antigens were detected in cells 4 days after electroporation with transcript-derived DENV3. On day 5 (Fig. 2), more antigen-positive cells were observed indicating RNA transcripts derived from L-IC-DENV3 L42 infectious clone were infectious and replicated in BHK-21 cells. The percentage of DENV3 antigen-positive cells increased over time until the first signs of cytopathic effect (CPE) appeared on day 8. As expected, no antigen was detectable in the negative control.
After assembling the full-length DENV3 cDNA by in vitro ligation, a unique BamHI site was inserted in the genome and used to distinguish transcript-derived DENV3 from WT-DENV3. On the third blind passage in mosquito C6/36 cells, a 2381-bp fragment was amplified by RT-PCR using virus RNA extracted from either WT-DENV3 (Fig. 3, lane 2) or transcript-derived (Fig. 3, lane 4) culture supernatants. As expected, restriction endonuclease BamHI was unable to cleave the PCR product derived from WT-DENV3 (Fig. 3,  lane 3). On the other hand, the PCR product from transcript-derived DENV3 was cleaved by BamHI producing fragments of 1318-bp and 1063-bp (Fig. 3, lane 5). No amplification was observed in the negative control (Fig. 3, lane 6). These results confirm that the rescued virus was indeed derived from the reverse genetics strategy described here.
In order to further characterize the rescued virus, growth kinetics of transcript-derived DENV3 stocks were compared to WT-DENV3 stocks at MOI of 0.1 in mosquito C6/36 cells (Fig. 4). Transcript-derived virus L-IC-DENV3 L42 grew at a similar rate as the WT-DENV3 (P>0.05 at each time point). Both viruses also reached peak virus titer on day 4. Additionally, these results show evidence that the mutation at position 7878 (A7878G) has no effect on transcript-derived virus L-IC-DENV3 L42 growth kinetics in C6/36 cells.

DISCUSSION
Dengue infections represent one of the most prevalent arthropod-borne diseases worldwide, affecting millions of individuals annually in tropical and subtropical areas. Without effective prophylaxis, those numbers are expected to increase over the years as a consequence of climate warming impact in DENV transmission (Jetten andFocks 1997, Sutherst 2004).
Despite enormous progress in unraveling DENV replication cycle and virus-host interaction, many molecular features are not yet clearly understood. The infectious clone methodology is an important asset of the molecular approach in experimental virology, contributing to the study  of genome replication, virus pathogenesis and vaccine development. The first DENV infectious clone has been described for DENV4 (Lai et al. 1991). Subsequently, infectious clones have been developed for DENV2 (Kapoor et al. 1995), DENV1 (Puri et al. 2000 and DENV3 (Blaney et al. 2004). Unfortunately, flavivirus genome instability in bacteria is still a tremendous impediment to the development of these infectious clones. Since there is no standard with regard to the best approach to construct reverse genetics systems for flaviviruses, different strategies aiming to bypass genome toxicity in E. coli have arisen. Homologous recombination in yeast has been successfully used in the development of a full-length reverse genetics system for bovine viral diarrhea virus (BVDV) (Gil et al. 2006a, b). Based on this approach, we were able to generate a stable reverse genetics system for YFV (Queiroz et al. 2013). However, this system did not tolerate dengue virus genome and we were unable to rescue a functional infectious clone for dengue virus. We tested distinct dengue virus serotypes in this system. We also attempted to couple this initial strategy with the insertion of a "stabilizing linker" described previously (Blaney et al. 2004) without success. Thereafter, we shifted our efforts to the two-plasmid strategy.
Here, we sought to develop an infectious clone derived from a Brazilian DENV3 clinical isolate. By successfully cloning the full-length genome in two separate fragments in yeast, we were able to surpass genome instability. Recently, our group has employed a different strategy to achieve stability of the cloned virus genome based on the same DENV3 isolate used here. These findings have been published elsewhere .
It is probable that yeast is more tolerant than bacteria to the toxicity of unstable sequences commonly found in flavivirus genome, making it a very convenient host for the development of reverse genetics systems (Polo et al. 1997, Pu et al. 2011, Puri et al. 2000. Homologous recombination in yeast is a simple and robust technique with higher cloning efficiency when compared to standard cloning procedures in E. coli (Gibson 2009, Shanks et al. 2009). It is also a high fidelity cloning strategy, successfully employed in the assembly of virus and bacterial genomes, making the emergence of spurious mutation very unlikely. Even though the in vitro ligation strategy derived from two plasmids is not widely used, our results reinforce the effectiveness of this approach to circumvent genome instability.
RNA transcripts derived from L-IC-DENV3 L42 are infectious in BHK-21 cells and antigen-positive cells were readily detected 4 days after transfection. Many factors contribute to RNA infectivity in susceptible cells such as cell type, transfection method, virus serotype/genotype and RNA preparation, but similar results have been described for DENV2 in LLC-MK2 cells (Polo et al. 1997).
The identified mutation did not seem to carry any deleterious effect on transcript-derived DENV3 infectivity in either BHK-21 or mosquito C6/36 cells. This claim is supported by the successful In conclusion, we present an alternative approach to obtain a functional infectious clone for DENV3 while working with unstable sequences. A similar strategy could be engineered for other DENV serotypes or flaviviruses. This infectious clone will be a valuable tool in molecular virology research for better understanding DENV biology and pathogenesis.

ACKNOWLEDGMENTS
The authors thank the Program for Technological Development in Tools for Health-PDTISFIOCRUZ for the use of its facilities, Dr. Pedro Vasconcelos at the Evandro Chagas Institute (Belém, Pará, Brazil) for kindly supplying the polyclonal hyperimmune ascitic fluid used in this paper and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for supporting J.J.S.S. through a graduate scholarship. This work is supported by research grants from Fundação Oswaldo Cruz -FIOCRUZ (Papes V), Fundação de Amparo a Ciência e Tecnologia do Estado de Pernambuco (FACEPE) (APQ-0519-2.12/06), the CNPq (470892/2007-2), and the National Institute of Allergy and Infectious Diseases (NIAID/NIH) -Grant U19 AI56541. Palavras-chave: genética reversa, vírus dengue, clonagem molecular, clone infeccioso.