key: cord-0307618-7ogne2eh
authors: Tse, Longping V.; Meganck, Rita M.; Dong, Stephanie; Adams, Lily E.; White, Laura J.; de Silva, Aravinda M.; Baric, Ralph S.
title: Genetically Engineered DENV Produces Antigenically Distinct Mature Particles
date: 2021-04-08
journal: bioRxiv
DOI: 10.1101/2021.04.06.438747
sha: 38de844604ed4a4ecc467680caa4e1182b401d77
doc_id: 307618
cord_uid: 7ogne2eh

Maturation of Dengue viruses (DENV) alters the structure, immunity and infectivity of the virion and highly mature particles represent the dominant form in vivo. The production of highly mature virions principally relies on the structure and function of the viral premature protein (prM) and its cleavage by the host protease furin. We developed a reliable clonal cell line which produces single-round mature DENVs without the need for DENV reverse genetics. More importantly, using protein engineering coupled with natural and directed evolution of the prM cleavage site, we engineered genetically stable mature DENVs without comprising viral yield and independent of cell, host, or passage. Using these complementary strategies to regulate maturation, we demonstrate that the resulting mature DENVs are antigenically distinct from their isogenic immature forms. Given the clinical importance of mature DENVs in immunity, our strategy provides a reliable strategy for the production of stable, high-titer mature candidate DENV live virus vaccines, genetically stabilized viruses for DENV maturation and immunity studies, and models for maturation-regulated experimental evolution in mammalian and invertebrate cells. Our data from directed-evolution across host species reveals distinct maturation-dependent selective pressures between mammalian and insect cells, which sheds light on the divergent evolutionary relationship of DENVs between its host and vector.

Mosquito-borne Dengue virus (DENV) is a major global public health threat causing ~400 million 27 new cases of dengue annually 1,2 . Although the majority of cases occur in tropical and subtropical areas 28

where the mosquito vectors are most concentrated, global warming, travel, and globalization have 29 contributed to the worldwide spread and intermixing of the four DENV serotypes 3 . Indeed, DENV infection 30 has increased 30-fold between 1960 and 2010 with an upsurge of cases in the USA and Europe. A hallmark 31 of DENV pathogenesis is the possibility for antibody dependent enhancement (ADE), which can progress 32 to life threatening dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS) upon secondary 33 infection with a different serotype. So far, no antiviral treatments are available to treat DENV disease and 34 the only approved vaccine, Dengvaxia, is not recommended for use in naïve populations 4,5 . 35

Proteolytic cleavage of viral membrane fusion proteins is a common strategy for temporal or 36 spatial control of virus infection, ultimately affecting tropism and transmission 6,7 . The DENV virion 37 structural proteins consists of capsid, E (envelope), and prM glycoproteins which undergo major 38 conformational changes via the process of maturation. The most common depiction of DENV particles 39 features the mature form, which is composed of 90 Envelope (E) homodimers lying flat in a "herringbone" 40 structure and organized into a 50 nm icosahedral (T=pseudo 3) symmetry resembling other non-41 enveloped virions 8 . However, the virion is assembled in the ER as a non-infectious 9 , immature virion which 42 adopts a completely distinct structure as a 60 nm "spikey" sphere with 60 three-fold spikes 10, 11 . Each 43 "spike" is composed of three E protein monomers elevated at a 27° upward angle with the fusion loop 44 covered by prM proteins 11,12 . Maturation is a two-step process involving the proteolytic cleavage of prM 45 by furin, a ubiquitously expressed serine protease with a preference for basic (positively charged) 46 substrates, at the trans-Golgi network (TGN) followed by its release at neutral pH outside of the cell 11,13-47 determined the relative maturity of each serotype by calculating the ratio of prM to E. Consistent with 89 the hypothesis that prM cleavage is dependent on both local primary sequence and other distal and 90 structural functions, PiTou predictions do not translate completely to the empirical maturation status of 91 DENV. Relative maturity was clearly different between serotypes. In particular, serotypes encoding an 92

Glutamic acid (E), but not Aspartic acid (D) at the P3 position (prM residue 89) are associated with more 93 immature virion production in Vero cells, with DV2-WT virions containing the highest level of uncleaved 94 prM, followed by DV4-WT, DV3-WT, and DV1-WT which has nearly undetectable levels of prM, and hence 95 is more mature (Fig. 1b) Optimized Clonal Vero-furin Cells Generate High Yield, Mature DENV 106 DENV maturation also depends on the producer cells; for instance, C6/36 grown DENVs show a 107 different maturation profile, from DV2-WT (most immature) < DV1-WT = DV3-WT < DENV4 (most mature) 108 ( Fig. 1b) . As reported previously 23 , fully mature DENV strains can be generated in Vero cells that 109 overexpress furin. However, high level furin expression may negatively impact DENV virus production. 110

Using the sleeping beauty transposon system 26 , we isolated two clonal lines with high (VF-Hi) or low Lo) levels of furin expression (Fig. 2a) . Immunofluorescent staining and western blot analysis revealed 112 different levels of furin expression in the trans-Golgi network ( Fig. 2b and 2c) . The growth kinetics of all 113 four DENV serotypes were tested on both Vero-furin lines and compared to unmodified Vero cells (Fig. 2d  114 -g). DV1-WT, DV2-WT and DV4-WT showed similar growth kinetics in all cell lines tested, while VF-Hi 115 supported better DV3-WT growth ( Fig. 2d -g) . VF-Hi supports the production of fully mature DENV virions 116 across all four serotypes ( Fig. 2d -g) . In agreement with the low furin expression level, VF-Lo phenocopied 117 the DENV maturation status of unmodified Vero cells ( Fig. 2d -g) . Therefore, VF-Hi cells allow for high 118 DENV yield in all serotypes, suggesting the furin expression level in VF-Hi is optimal for production of fully 119 mature DENVs. 120 Supernatants were harvested at 120hpi and analyzed by western blot for DENV maturation using anti-Env 130 and anti-prM antibodies. All assays were performed with at least two biological repeats with two technical 131

replicates. Growth kinetics of DENV variants were compared to their corresponding wildtype using 2-way 132 ANOVA multiple comparisons. 133 134

As an alternative to ectopic overexpression of furin which only generates mature virion for a single 136 round of infection, we hypothesized that genetic modification of the prM furin cleavage site could also be 137 used to optimize DENV maturation independence of cells or hosts. Using DV1-WT as a model, we 138 introduced a mutation at the P3 position of the furin cleavage site and generated an isogenic strain, DV1-139 prM-D89K. The mutated cleavage site (HRRKKR|S) has a Pi-Tou score of 14.68 compared to the DV1-WT 140 cleavage site (HRRDKR|S) with a Pi-Tou score of 6.90, predicting more optimal cleavage (Fig. 3a) . DV1-WT 141 and DV1-prM-D89K displayed no difference in virus growth kinetics in Vero (mammalian) and C6/36 142 (insect) cells (Fig. 3b ). In both Vero and C6/36 cultures, DV1prM-D89K was more mature than DV1-WT, 143 phenocopying the Vero-furin grown DV1-WT (Fig. 3c) . 144

To understand if the furin cleavage site mutation is portable across serotypes, we introduced a 145 similar mutation on the DV4-WT backbone, generating the isogenic strain DV4-prM-E89K (Fig. 3d) . While 146 we successfully generated a pure population of DV4-prM-E89K in C6/36 cells, a spontaneous mutation, 147 K89N, rapidly emerged and gave rise to a new evolved DV4-prM-E89N variant in Vero cells by passage 2 148 (Fig. S1a) . By the 5 th passage, the DV4-prM-E89N variant represented 100% of the viral population ( Fig.  149 S1a), supporting the notion that viruses encoding the E89K mutation were less fit than those encoding the 150 E89K mutation in Vero cells. Growth kinetics of DV4-prM-E89K and DV4-prM-E89N on C6/36 cells are 151 comparable to DV4-WT (Fig. 3e) . However, the DV4-prM-E89K variant displayed a robust 2-log growth 152 defect compared to DV4-prM-E89N and DV4-WT on Vero cells (Fig. 3f) . When grown at 32°C, the growth defect of DV4-prM-E89K was alleviated (Fig. S1c) . The maturation status of the two variants were tested 154 and compared to DV4-WT. DV4-prM-E89N is more mature than DV4-WT in both Vero and C6/36 cells (Fig.  155 3g). No prM can be detected in DV4-prM-E89K; due to the low virus yield in Vero cells, the data suggest 156 that either DV4-prM-E89K is fully mature or the protein input is below detection limit (Fig. S1b ). As 157 calculated by Pi-Tou, DV4 has the highest furin cleavage score among the DENV serotypes at 13.26. The 158 point mutation prM-E89K increases the score to 16.65 (the highest score observed), while DV4-prM-E89N 159 has a Pi-Tou score of 13.82 (Fig. 3d) . In DENV4, it seems that a "super-optimal" furin cleavage site may 160 negatively impact DENV growth in Vero cells. The data suggest a delicate balance likely exists between 161 virion maturation, furin cleavage site efficiency, and viral fitness in different serotypes. 162 

Based on the spontaneous K89N mutation in DV4, we hypothesized the prM cleavage site has 183 high plasticity, suggesting the existence of a "Goldilocks Zone" for efficient in vitro growth. We utilized 184 saturation mutagenesis and directed-evolution to simultaneously screen thousands of DENV2 prM 185 cleavage site variants for efficient growth in tissue culture. We generated a DENV2 viral library in which 186 four positions, P3, P5, P6, and P7, of the prM cleavage site were randomly mutated, preserving the core 187 furin cleavage site (Fig. 4a) . The library was propagated three times in either Vero or C6/36 cells, and each 188 passage of the virus were deep sequenced along with the plasmid library (Fig. 4a) . The theoretical amino 189 acid diversity of the library is 160,000 variants (ignoring stop codons), which was represented in the 190 plasmid library ( Table 2 ). As expected, viral diversity rapidly drops after one passage, to 0.7% (1148 unique 191 variants) and 16.2% (25942 unique variants) of the theoretical maximum in Vero and C6/C6 respectively, 192 further diminished after each passage ( Table 2 ). The large number of viable DENV2 variants in both cells 193

indicates a high degree of plasticity within the prM cleavage site in culture (Table 2) (Fig 4b and 4c ). While the DV2-WT cleavage site has a Pi-Tou 198 score of 11.12, the Vero-selected cleavage site score increased to 14. 39 . Surprisingly, the DV2-C1 cleavage 199 site scored at 7.76, a much lower score than DV2-WT (Fig. 4d) . We plotted the PiTou score distribution of 200 the top 50 ranked variants in C6/36 and Vero cells, with peaks at 7.7 and 14.9, respectively (Fig. 4e) . We 201 also plotted the Pi-Tou scores of the top 50 sequences from passage 1 that were extinct by passage 3. 202

Although there was no distinct peak of deselection in C6/36 cells, a distinct peak of Pi-Tou scores at 13.9 203 were observed in the Vero-selected extinct population (Fig. 4e ). The sequences, counts, and Pi-Tou scores 204 of the top 50 enriched and deselected cleavage sites are summarized in Table S1 and S2. Due to founder 205 effects in directed-evolution experiments, there is only one sequence shared between the top 50 variants 206 evolved from Vero and C6/36 cells after three passages. Additionally, some variants with high Pi-Tou 207 scores are rapidly deselected in both Vero and C6/36 cells, suggesting that the furin cleavage site 208 sequence plays multiple roles in viral fitness (Table S2 ). The difference in scores between the two cell lines 209 and the leveling effect of the lower ranked variants highlighted the differential fitness requirements of 210 DENV2 between insect and mammalian cells. 211 The top ranked evolved variants, DV2-V1 and DV2-C1, were re-derived via reverse genetics for 226 further characterization. We also included a DV2-prM-E89K variant similar to the original DV1 mutation 227 as comparison (Fig. 5c) . While the DV2-prM-E89K variant has slightly reduced growth in Vero cells 228 compared to DV2-WT, both DV2-V1 and DV2-C1 grow better than DV2-WT in Vero, with a drop in titer in 229 C6/36 cells at 96 to 120 hpi ( Fig. 5a and 5b) . In Vero cells, DV2-prM-E89K and DV2-V1 are almost fully 230 mature while DV2-C1 is only 30% more mature than DV2 WT (Fig. 5d) . When the viruses are grown in 231 C6/36, all the variants are 60 -70% more mature than DV2 WT (Fig. 5d) 

Given the ability to generate fully mature DENVs, we next evaluated the impact of maturation 243 status on antigenicity. We selected several monoclonal antibodies targeting different regions of the DENV 244 E glycoprotein, including C10 (Envelope-Dimer-Epitope 1) 27 , B7 (Envelope-Dimer-Epitope 2) 27 , 1C19 (BC 245 loop) 28 and 1M7 (fusion loop) 28 . As expected, Ab epitopes that are not maturation dependent are 246 preserved, as evidenced by antibodies such as C10, B7, and 1C19 which showed no difference in Foci 247

Reduction Neutralization Titer 50 values (FRNT50) (Fig. 6a -c) . However, the fusion loop targeting 248 antibody 1M7 showed significantly different FRNT50 values between fully mature and less mature DENVs 249 in DENV1 and 4, but not in DENV2 (Fig. 6a -c) . For DENV4, we also tested polyclonal sera from patients 250 180 days post DENV4 vaccination or naturally infected patients from a traveler cohort. Polyclonal serum 251 contains a mixture of antibodies which may or may not be affected by virion maturation status. 252

Unsurprisingly, FRNT50 of polyclonal serum was equivalent for fully mature and partially mature DENV4 253 (Fig. 6d) . 254 Like many studies, our report generated additional questions. Biologically, does DENV maturation 317 play a more critical role than simply preventing premature fusion during production? Could maturation 318 also play a role in vector-to-host or host-to-vector transmission? Is fully mature DENV advantageous or 319 deleterious in mosquitoes and mammals? What determinants outside of the primary cleavage site 320 sequence regulate maturation efficiency? Will biologically stabilized virions drive selection of unique 321 subsets of neutralizing antibodies after infection? Clinically, the antigenic differences between mature 322 and immature DENV require more comprehensive investigation. Furthermore, a new class of vaccines 323 could be imagined based on stabilized mature particles which elicit maturation discriminatory antibodies. 324

Given the clinical relevance and enigmatic nature of DENV maturation, our study adds to understanding 325 of DENV maturation control and provides essential tools for future investigations. Recombinant viruses were constructed using a four-plasmid cloning strategy as described previously 42 . 343

The DENV genome was divided into four fragments (A−D fragment) and subcloned into four separate 344 plasmids. A T7 promoter was introduced into the 5′ end of the A fragment, and unique type IIS restriction 345 endonuclease cleavage sites are introduced into the 5′ and 3′ ends of each fragment to allow for 346 systematic assembly into a full-length cDNA from which the full-length RNA transcripts can be derived. foci forming assay. In brief, Vero cells were seeded at 2x10 4 cells/well in a 96-well plate. 50 µl of serial 367 diluted viral supernatant were added to each well and incubated for 1h at 37ºC in 5% CO2 incubator. 125 368 µl of overlay (Opti-MEM + 5% methyl cellulose + NEAA + P/S) was added to each well and incubated for 369 48h at 37ºC + 5% CO2. Each well was rinsed 3 times with PBS and fixed with 10% formalin in PBS for 370 staining. Vero cells were blocked in permeabilization buffer (eBioscience) with 5% non-fat dried milk. Two 371 primary antibodies, anti-prM mAb 2H2 and anti-Env mAb 4G2, from non-purified hybridoma supernatant 372 were used at 1:500 dilution in blocking buffer. Goat anti-mouse secondary conjugated with horseradish 373 peroxidase (HRP) (SeraCare's KPL) were diluted at 1:1000 in blocking buffer. Foci were developed using 374 TrueBlue HRP substrate (SeraCare's KPL) and counted using an automated Immunospot Analyzer 375 instrument (Cellular Technology Limited). All experiments were performed independently a minimum of 376 3 times. 377

Immunostaining and western blotting for human furin 378

Cells were fixed in 10% formalin in PBS and permeabilized with permeabilization buffer (eBioscience). 379 (Invitrogen, 1:2000) as secondary antibody. For western blotting, cell were lysed in 1% TritonX100, 100 381 mM Tris, 2M NaCl and 100 mM EDTA. Cell lysates were run in SDS-PAGE and blotted onto PVDF 382 membrane. Furin bands were detected using rabbit anti-furin polyclonal at 1:1000 and Goat anti-rabbit 383 HRP (Invitrogen, 1:5000) was used as secondary antibody. 384

Western Blotting for DENV maturation 385 Viral stocks or supernatant from DENV growth curves at 120hpi were diluted with 4x Laemmli Sample 386

Buffer (Bio-Rad) and boiled at 95ºC for 5 minutes. Following SDS-PAGE electrophoresis, proteins were 387 transferred to PVDF membrane and blocked in blocking buffer consist of 3% non-fat milk in PBS + 0.05% 388

Tween-20 (PBS-T). The membrane was incubated with polyclonal rabbit anti-prM (1:1000, Invitrogen, Cat. 389 #PA5-34966) and purified human anti-Env (fusion loop) 1M7 (2µg/ml) in 2% BSA + PBS-T solution for 1h 390 at 37°C. The primary antigen-antibody complex was detected by incubating the blot with goat anti-rabbit 391

IgG HRP (1:10000, Jackson-ImmunoLab) and sheep anti-human IgG HRP (1:5000, GE Healthcare) in 3% 392 milk in PBS-T, for 1h at room temperature. Membranes were developed by Supersignal West Pico PLUS 393 Chemiluminescent Substrate (ThermoFisher). Western blot images were captured with iBright FL1500 394 imaging system (Invitrogen). The pixel intensity of individual bands was measured using ImageJ, and 395 relative maturation was calculated by using the following equation: (prMExp/EnvExp)/(prMWT/EnvWT). All 396 experiments were performed independently a minimum of 3 times. 397

Foci reduction neutralization titer assay (FRNT Assay) 398

FRNT assays were performed on Vero cells as has been described previously 43 . Briefly, 2x10 4 Vero cells 399

were seeded in a 96-well plate. Antiserum or mAbs were serially diluted and mixed with DENV viruses (80 400 -100 FFU/well) at a 1:1 volume ratio and incubated at 37ºC for 1h without the cells. The mixture was 401 transferred to the 96-well plate with Vero cells and incubated at 37ºC for 1h. The plate is subsequently 402 overlaid with overlay medium (see above). Viral foci were stained and counted as described above. Data 403 were fitted with variable slope sigmoidal dose-response curves and FRNT50 were calculated with top or 404 bottom restraints of 100 and 0, respectively. All experiments were performed independently at least 2 405 times, due to limited amounts of human serum. 406 DENV2 library generation and directed-evolution 407 DENV prM libraries were engineered through saturation mutagenesis on amino acid residues P3, 5, 6 and 408 7 of the DENV furin cleavage site based on previously published protocol 44 . In brief, degenerate NNK oligos 409 (Integrated DNA Technologies) were used to amplify the prM region to generate a library with mutated 410 prM DNA fragments. To limit bias and ensure accuracy, Q5 high fidelity polymerase (NEB) was used and 411 limited to <18 cycles of amplification. The DNA library was cloned into the DENV reverse genetics system plasmid A to create a plasmid library by standard restriction digestion. Ligation reactions were then 413 concentrated and purified by ethanol precipitation. Purified ligation products were electroporated into 414 DH10B ElectroMax cells (Invitrogen) and directly plated on multiple 5,245-mm 2 bioassay dishes (Corning) 415 to avoid bias from bacterial suspension cultures. Colonies were pooled and purified using a Maxiprep Kit 416 (Qiagen). The plasmid library was used for DENV reverse genetics as described above. The in vitro 417 transcribed DENV RNA library was electroporated in either Vero or C6/36 cells, the viral supernatants 418 were passaged 3 times every 4 to 5 days in the corresponding cells for enrichment. 419

High-throughput sequencing and analysis 420 Viral RNA was isolated using a QIAamp Viral RNA Mini Kit (Qiagen). Amplicons containing the library 421 regions were prepared for sequencing through two rounds of PCR, using the Illumina TruSeq system and 422 Q5 Hot Start DNA polymerase (NEB). Primers for the first round of PCR were specific to the DENV2 prM 423 sequence with overhangs for Illumina adapters. This PCR product was purified and used as a template for 424 a second round of PCR using the standard Illumina P5 and P7 primers with barcodes and sequencing 425 adaptors. PCR products were purified and analyzed on a Qubit 4 fluorometer (Invitrogen) and Bioanalyzer 426 (Agilent Technologies) for quality control. Amplicon libraries were diluted to 4 nM and pooled for 427 sequencing, which was carried out on a MiSeq system with 300bp paired-end reads. Plasmid and P0 428 libraries were sequenced at a depth of ~1 million reads per sample; further passages were sequenced 429 with depth between 300,000 -1 million reads to sample. A custom perl script 44 was used to analyze the 430 sequences, and custom R scripts were used to plot the data. 431

Furin cleavage prediction 432 Furin cleavage site efficiency was predicted using the Pi-Tou software 25 , providing amino acids from 433 position P14-P6' of the DENV furin cleavage sites. 434

Statistical analysis was carried out using Graphpad Prism version 9.0. Growth kinetics and maturation of 436 DENV variants were compared to their corresponding wildtype using 2-way ANOVA multiple comparisons. 437

Neutralization titers of DENV variants were compared to their corresponding wildtype using Student's t-438 test. Significance symbols are defined as follow: * p<0.05, ** p< 0.01, *** p<0.001, **** p<0.0001. Data 439 are graphed as mean +/-standard deviation. 440

Refining the Global Spatial Limits of Dengue Virus Transmission by Evidence

Based Consensus

The global distribution and burden of dengue

The current and future global distribution and population at risk of dengue

Deliberations of the Strategic Advisory Group of Experts on Immunization 462 on the use of CYD-TDV dengue vaccine

Dengue vaccine: WHO position paper

Structures and mechanisms of viral 466 membrane fusion proteins: Multiple variations on a common theme

Viral membrane fusion. Virology 479-480

Structure of dengue virus: implications for flavivirus organization, maturation, 470 and fusion

Functional importance of 472 dengue virus maturation: Infectious properties of immature virions

Mechanism for maturation-related 475 reorganization of flavivirus glycoproteins

Serotype 1 Virus Structures Provide Insight into the Maturation Process

Assembly and Maturation of the Flavivirus Kunjin Virus 482

Appear To Occur in the Rough Endoplasmic Reticulum and along the Secretory Pathway

The flavivirus precursor membrane-envelope protein complex: structure and 485 maturation

Structure of the Immature Dengue Virus at Low pH Primes Proteolytic Maturation

Mechanism of Enhanced Immature Dengue Virus Attachment to Endosomal 489

Membrane Induced by prM Antibody

Dengue type 1 viruses circulating in humans are highly infectious and poorly 491 neutralized by human antibodies

Heterogeneity and Dynamics of Flaviviruses on Antibody Recognition

Does structurally-mature dengue virion 496 matter in vaccine preparation in post-Dengvaxia era? Hum. Vaccines Immunother

Degrees of maturity: The complex structure and biology of 499 flaviviruses

Immature dengue virus: A veiled pathogen?

Immune correlates of protection for dengue: State of the art and research

Enhancing dengue virus maturation using a stable furin over-expressing cell 504 line

Genetic Variation between Dengue Virus Type 4 Strains Impacts Human Antibody Binding and Neutralization

Computational prediction of furin cleavage sites by a hybrid method 508 and understanding mechanism underlying diseases

Optimized Sleeping Beauty transposons rapidly generate 510 stable transgenic cell lines

A new class of highly potent, broadly neutralizing antibodies isolated from 512 viremic patients infected with dengue virus

The potent and broadly neutralizing human dengue virus-specific monoclonal 514 antibody 1C19 reveals a unique cross-reactive epitope on the bc loop of domain II of the 515 envelope protein

A survey of furin substrate 517 specificity using substrate phage display

A study of human furin specificity using synthetic peptides derived from 519 natural substrates, and effects of potassium ions

High-Resolution Analysis and Functional Mapping of Cleavage Sites and 521

Substrate Proteins of Furin in the Human Proteome

Prediction of proprotein convertase cleavage sites

An arthropod enzyme, Dfurin1, and a 525 vertebrate furin homolog display distinct cleavage site sequence preferences for a shared viral 526 proprotein substrate

Structure and topology around the cleavage site regulate post-translational 528 cleavage of the HIV-1 gp160 signal peptide

A Novel Activation Mechanism of Avian 530 Influenza Virus H9N2 by Furin

The stem region of premembrane protein plays an important role in the virus 532 surface protein rearrangement during dengue maturation

Epitope resurfacing on dengue virus-like particle vaccine preparation to induce 535 broad neutralizing antibody

Structural Changes in Dengue Virus When Exposed to a Temperature of 37 C

Shake, rattle, and roll: Impact of the dynamics of flavivirus 539 particles on their interactions with the host

Efficacy of a Tetravalent Dengue Vaccine in Children in Latin America

Which dengue vaccine approach is the most promising, and should we 544 be concerned about enhanced disease after vaccination?: Questions raised by the development 545 and implementation of dengue vaccines: Example of the sanofi pasteur tetravalent dengue 546 vaccine

Development and characterization of a reverse genetic system for studying 548 dengue virus serotype 3 strain variation and neutralization

Identification of Dengue Virus Serotype 3 Specific Antigenic Sites Targeted by

Structure-guided evolution of antigenically distinct adeno-associated virus variants 552 for immune evasion