Background
Dengue virus (DENV) is an arbovirus member of the
Flaviviridae family and
Flavivirus genus. The virus is classified into four antigenically distinct serotypes (DENV 1, 2, 3, and 4) and subdivided into genotypes [
1]. Dengue is a significant global disease caused by DENV, with an incidence estimated at 390 million cases per year, in which 60 million develop symptoms [
2,
3]. Due to the dramatic increase in DENV incidence and spread, symptoms such as central nervous system (CNS) involvement have increased [
4]. The spectrum of illness ranges from asymptomatic to dengue or severe dengue [
5]. Dengue clinical findings include nausea, vomiting, rash, muscle, and joint pain, as alarm signals leukopenia, abdominal pain, persistent vomiting, clinical fluid accumulation, mucosal bleeding, lethargy, and postural hypotension [
6]. Complications can lead the individual to death, including severe plasma leakage, respiratory distress accumulation, severe bleeding, shock, impaired consciousness, and severe organ impairment. Severe organ impairment includes hepatitis, myocarditis, pancreatitis, and encephalitis [
5,
6]. The evolution to severe conditions is not well elucidated. However, it is understood that host factors such as abnormal immune response, homeostatic disorder, and intrinsic characteristics of the infecting virus strains are essential [
7].
DENV genotypes have been associated with different clinical manifestations, including severe forms of the disease [
8,
9]. Thus, depending on viral isolate characteristics such as tropism, ability to evade the host’s immune response, and biological barriers evasion, it is possible to observe differences in vivo and in vitro infection and viral pathogenesis [
10‐
12].
Between 2002 and 2006, a DENV3 genotype III outbreak in Brazil presented many cases [
13]. Simultaneously, reports of DENV3 G1L1, also known as DENV3 G genotype V, according to Wittke et al., 2002, occurred in this outbreak [
14]. However, the DENV3 genotype I (GI) lineage 1 (L1) (DENV3 G1L1) has only been reported in some states, such as Minas Gerais, Rondônia, and Pará [
8,
15]. In a previous study, we compared the DENV3 GIL1 isolates from Minas Gerais-MG (DENV3 MG-20) and Rondônia-RO (DENV3 PV_BR). The in vivo analysis and in vitro assays of heparan sulfate adhesion showed differences between these isolates, although they belong to the DENV G1L1 [
12]. DENV3 MG-20 was isolated from a fatal meningoencephalitis case from MG [
8], with reproduction of neurovirulence in a murine model [
12,
16], while DENV3 PV_BR [
12] and other isolates from RO showed no neurovirulence in mice [
17].
Here, we report using plaque unit forming titration, qPCR, immunofluorescence, and transmission electron microscopy that these isolates infect cells derived from a malignant human glioblastoma (U251 line). However, DENV3 MG-20 infected these cells with syncytium formation and cell death, while DENV3 PV_BR infected cells without syncytium formation or any other cytopathic effect (CPE) evidenced by optical microscopy. These results emphasize the difference in biological characteristics of these isolates, which could be a critical component in the interaction with the host cell and the development of severe dengue.
Methods
Virus and cells
DENV3 MG-20 (GenBank Accession number EF625835.1) is an isolate from the Laboratório de Vírus collection, Universidade Federal de Minas Gerais. DENV3 PV_BR (ACD85885.1) was isolated from a patient’s serum sample from Porto Velho, Rondônia State, Brazil. Professor Luiz Tadeu Figueiredo kindly provided it from USP/Ribeirão Preto. The cell lines used were C6/36 cells (ATCC CRL-1660), BHK-21 cells (ATCC CCL-10), Vero cells (ATCC CCL-81), and U251 MG Cells. The maintenance of these cell lines has been previously described [
12].
In silico assays
Phylogenetics was analyzed using 93 DENV3 complete protein sequences retrieved from GenBank (
https://www.ncbi.nlm.nih.gov/genbank/). Sequences were aligned using the Clustal Omega server available from the European Bioinformatics Institute (
https://www.ebi.ac.uk/Tools/msa/clustalo/). Phylogenetic trees were reconstructed using the neighbor-joining method, with 1000 bootstrap replicates. Tree construction was carried out through the MEGA 7 software. Partial sequences with high similarity to DENV3 GIL1 (up to 94%) were surveyed using GenBank BLASTP/N (server on
www.ncbi.nlm.nih.gov/blast). Molecular and spatial analysis of DENV3 GIL1 was done using the software PyMol 4.6.0 and Chimera 1.14. The 4GSX model from the protein data bank (PDB) was used for the analyses. The modeling was done as described in reference 12. Hydrophobicity analysis was done according to Eisenberg’s scale, and the protein molecules’ color was generated using the Color h script.
Plaque assays and titration
For phenotypic characterization in C6/36 cells, 9 × 105 cells were seeded in 6-well plates, incubated for 24 h, and then infected DENV3 GIL1 MG-20 or DENV3 PV_BR samples. After 1 h adsorption, 1% carboxymethylcellulose (CMC) (Synth, Brazil) was added to the wells, dissolved in Leibovitz medium (Gibco, USA) supplemented with 5% fetal bovine serum (FBS; Cultilab, Brazil), 20 μg/ml streptomycin (Sigma–Aldrich, USA), 100 IU/ml penicillin (Cultilab, Brazil), and 2 μg/ml amphotericin (Sigma–Aldrich, USA), and the cells were incubated for 7 days at 37 °C. The cell monolayers were then fixed in 10% formalin and stained with a 1% crystal violet solution. Titration assays were done with 2.5 × 105 BHK-21 cells in each 6-well plate. After viral adsorption, cells were incubated with 0.5% CMC in Medium (Gibco, Brazil) at 37 °C and with 5% CO2. After 7 days, cell monolayers were fixed and stained as described above, and viral plaques were counted to determine the titers (PFU./ml).
Replication curve
U251 cells were infected with DENV3 MG-20 and DENV3 PV_BR samples at 0.1 MOI (multiplicity of infection). Supernatant and cell lysate (lysis heat shock) were collected from each well at 24 h, 48 h, 72 h, and 96 h post-infection. The viral replication curve was made by titration (described above), and the RNA replication curve was generated by real- time qPCR data. Titration was performed in triplicate. As a negative control, uninfected cells were used. RNA was extracted using the RNeasy Mini Kit (Qiagen, USA) extraction with 140 μl of the culture supernatant or harvested cell culture (previously rinsed with medium). Alm et al., 2015 [
18] described the reverse-transcriptase PCR reaction. The target was the coding region of the NS1 protein, and as a constitutive gene, primers for b-actin were used. The comparative Ct method was used for data analysis, applying the mathematical form of ΔCt (Applied Biosystems guide). Results were plotted as units of arbitrary genome copies. All reactions were performed in triplicates and analyzed in GraphPad Prism 6.
Optical microscopy and immunofluorescence
Under sterile conditions, 24-well plates with coverslips were seeded with U251 (70,000) and C6/36 (150,000) cells. After 24 h, monolayers were infected with 0.1 MOI of the DENV3 MG-20 and DENV3 PV_BR viruses. The adsorption was followed by 4 days of incubation. Then, cells were washed with PBS and fixed with cold acetone. The assay used 4G2 as a primary antibody (1:100) and Alexa 488 conjugated anti-mouse IgG (Sigma, EUA) 1:200 as a secondary antibody. An additional 30 min incubation was performed to stain the nuclei with DAPI. Coverslips were counterstained with Evans Blue. After mounting, the coverslips were observed under the DI-FL03 and Axio Imager Z2-ApoTome 2 Zeiss microscope models. Green fluorescent staining for virus identification was observed under blue light (wavelength 519 nm), and the cell nucleus was identified by sky blue staining (red light, wavelength 358 nm). Uninfected cells were kept as a control.
Transmission electronic microscopy
U251 cells were grown on coverslips to 90% confluence and infected with DENV3 PV_BR or DENV3 MG-20 at 1 MOI in DMEM with antibiotics and 2% FBS. Uninfected U251 cells were kept as a control. After 4 days of incubation, cells were harvested and fixed with 1 ml of 2.5% glutaraldehyde (Sigma) in 0.1 M phosphate solution (0.052 g of NaH2PO4.H2O + 0.435 g of Na2HPO4.7H2O in 20 ml of distilled water) under stirring for two hours. The cell layer was recovered, transferred to a 1.5 ml microtube, and centrifuged at 3000 g for 5 min. The glutaraldehyde phase was withdrawn from the experiment, and 200 μl aliquots of phosphate solution were added to the tubes. The samples were stored at 4 °C until they were sent to the UFMG Microscopy Center. Slides were read using the transmission electron microscopy model Tecnai G2-20 - SuperTwin FEI − 200 kV.
Discussion
The DENV3 genotypes classification is controversial and includes 4 or 5 genotypes [
14,
21]. The DENV3 GIL1 has a rare circulation and was first identified in Brazil in 2003 [
8]. Despite infrequent circulation, many severe cases associated with this lineage were observed, including neurological symptoms [
8,
15,
16,
20]. Using the phylogenetic construction of the DENV3 polyprotein, we found many complete DENV3 GIL1 sequences from GenBank without prior genotype identification (Fig.
1A). These sequences are grouped with the H-87 DENV3 GIL1 (GV) prototype, showing the similarity of these sequences with the Brazilian ones. Using the BLAST/P tool, we found partial sequences of this genotype. We identified information regarding geographic distribution, clinical symptoms, disease severity (Fig.
1B), mutations, and molecular markers (Fig.
2A).
Many reports have associated different genotypes with different outbreaks and clinical manifestations, including an increase in severe forms of the disease [
22,
23]. The DENV3 MG-20 sample was isolated from a serum sample derived from a fatal patient with neurological manifestations in 2004, Minas Gerais State (MG), Brazil [
8]. The development of neurological disease has been reproduced in mice inoculated with the DENV3 MG-20 and for other isolates such as DENV3 MG-21 and DENV3 MG-25 [
12,
24]. We used the DENV3 MG-20 (a human and mouse neurovirulent isolate) and DENV3 PV_BR (a non-neurovirulent isolate) as models to observe biological differences in vitro [
12]. These two isolates belong to the same genotype, have high sequence similarity to each other and to several DENV3 GIL1 (Fig.
1A), and were used to reproduce the observed neurovirulent and non-neurovirulent patterns, respectively, in glioblastoma cells.
DENV, as an enveloped virus, expresses a fusogenic protein that drives the fusion of the viral envelope with cellular membranes. In the DENV, this protein is the E protein and is expressed in the virion membrane surface. E protein interacts with cell-host receptors and cell membranes and is related to cell-to-cell fusion [
25]. Point differences between the two isolates were observed in the E amino acid sequence and are represented in Fig.
2A. Viral fusion proteins have been classified into three classes (I, II, III) according to their structure and fusion mechanism with the cell membranes. E DENV protein is classified as class II, characterized by three-domain architecture (DI, DII, DIII), β-sheet-based elongated ectodomain, with a ‘fusion loop’ responsible for cellular membrane interaction and virus membrane fusion (1B, 1 C). The E trimer fusional conformation shown in Fig.
2C results from the pH change inside the cell, exposing hydrophobic domains and allowing fusion between the virus-cell membranes [
26]. This protein is key in the host-pathogen interaction and infection entry stage. Mutations in sites near E hydrophobic clusters could impact the exposure of these domains and change the virus’s fusional properties [
27]. We do not know the real impact of the amino acid 62 and 123 mutations on the fusional properties of E DENV3 GIL1 samples (Fig.
1F and G). However, it drew our attention that these non-conservative mutations are close to hydrophobic clusters, including a fusion loop.
The infected human glial cell line (U251) shows different CPE for these two isolates. By optical microscopy and IFI, we identified that DENV 3 MG-20 induces syncytium formation besides being lytic,. The syncytial U251 cells presented ring nuclei surrounding the central cytoplasm. Syncytium formation is identified for other viruses, such as varicella-zoster virus in glial cells [
28]. DENV 3 PV_BR also showed tropism to these cells, initially observed through IFI. No morphological changes were observed by light microscopy and IFI in this case (Fig.
3). Viral titration shows the presence of infective progeny in both isolates, with peaks at 2 DPI. However, a rapid drop in viral replication occurs, especially for the cells infected with the DENV3 MG-20. The DENV 3 PV_BR concentration in the culture supernatant was constantly low and proved undetectable after 4 DPI (Fig.
4). Differences in infection productivity have been related to negative-strand viral RNA and protein accumulation and infectious particle titers [
10].
As shown in Fig.
5, different CPEs can also be observed in C6/36 mosquito cells. DENV3 MG-20 infection showed a fusogenic effect, while the DENV 3 PV_BR infection markedly affected cell morphology. Although both viruses compromise the integrity of the monolayer, it is visible in the C6/36 plaque assay with the CMC (Fig.
5F), the lytic capacity of DENV3 PV_BR compared to the other isolate. The classic model to explain flavivirus cell-to-cell fusion is based on C6/36 infection. Fusion activity is mediated by virions attached simultaneously with two cells (fusion-from-without, FFWO) or viral protein incorporated or expressed on the surface of infected cells (fusion-from-within, FFWI) [
29,
30].
No CPE was observed in both viruses by optical microscopy in Vero and mouse neural cells (lineage neuro 2 A) (data not shown). Although U251 and neuro 2 A are nerve cells, U251 is a stem-like cell from glial lineage, and Neuro 2 A is a stem-like cell from neuronal lineage. U251 (or U251 MG, also known as U-373 MG) used in this study, is a cell line derived from a human malignant glioblastoma multiform (GBM). Obtained by tumor explant, the U251 cell line is one of these glioma subclones and retains original features such as copy number aberrations typical of the GBM [
31]. Neuro 2 A originates in a spontaneous tumor from an albino strain A mouse and can differentiate into neurons [
32]. Therefore, we believe that the susceptibility of glial stem-like cells is due to their lineage functional defense characteristics, which are analogous to the function of phagocytes from other tissues (dendritic cells, macrophages), which are known to be susceptible to DENV infection.
Intracellular viral responses are triggered by upregulated carbon fluxes and efflux to biosynthesis, massive induction of protein, ribonucleotide synthesis, membrane proliferation, ER, and RER. Some of the CPE was evident by submitting infected U251 cells to transmission electron microscopy (TEM) analysis. Among them, the compartmentalization and vacuolization of cells infected with DENV3 MG-20 are compatible with the apoptosis mechanism (Fig.
6C). More intense hypertrophy of the endoplasmic reticulum is also observed in the DENV3 MG-20 infection, possibly justified by the high induction of membrane structures supporting viral replication [
33]. As can be seen in Fig.
6F and I, there is an intense flow of vesicles filled with cellular and viral molecules such as E protein (as seen in IFI: complex 4G2 antibody -E protein), suggesting that this protein is related to cell-to-cell generating syncytium. Serafino et al. (2003) [
34], in their studies, infected human lymphoblastoid (TO.FE) cells culture with hepatitis C virus (HCV) and analyzed CPE using TEM. They also observed Golgi complex and endoplasmic reticulum hyperplasia and viral particles in cells’ cytoplasmic vesicles. Therefore, as DENV and HCV belong to the same family, this phenomenon seems similar to the Flaviviridae family.
The large amount of viral proteins within vesicles of DENV3-MG20-infected U251 cells is a fact that draws attention. Studies have shown that the expression of viral proteins and the regulation of transcription are related to infection-differentially induced ribosomal host proteins [
35,
36]. It has been described that flavivirus infections induce a specific expression of the RPLP1/2 ribosomal complex, promoting the accumulation of viral proteins in the early stages of infection [
35]. It is also suggested that RPLP1/2 could interact with viral transmembrane domains (including in E protein), and bind to hydrophobic or charged regions, providing stability to the newly formed protein and increasing DENV protein expression [
36]. Naturally mutations in charge and hydrophobic protein locations can affect the stability of the three-dimensional structure and result in protein rearrangements, compromising protein folding and stability.
A massive induction of protein expression could also be associated with misfolded proteins and the unfolded protein response (UPR), which serves as an activation mechanism for apoptosis and increased pathogenicity in DENV infection of the liver [
37]. We suggest that the cell type, amount of protein expression, and the intense CPE of DENV3 MG-20 observed in U251 cells are responsible for the differences in suggestive apoptotic behaviors. Observing electron-dense particles with the size and appearance of DENV particles confirms that both viruses could infect cells. However, the DENV3 MG-20 isolate demonstrates high protein expression levels in these cells. Figure
6 L depicts a potential site of viral synthesis in the cell, characterized by an abundance of proteins, an enlarged endoplasmic reticulum, and vesicles containing viral particles.
In Diamond et al., 2020 [
10], different genotypes of DENV 2 generated divergent effects in cells. Here, we show natural isolates of DENV3 from the same lineage and high similarity, generating distinct profiles of CPE and syncytium in nerve cells. This GIL1 lineage is epidemiologically linked to involvement in the CNS in humans. These distinct pathogenic profiles were also visible in vivo models, in which mice infected with neurovirulent DENV3 MG-20 died, and those infected with DENV3 PV_BR did not show clinical signs of illness [
12]. Syncytium formation is a virulence factor, increasing pathogenesis and disease severity. Cell-to-cell fusion allows a fast RNA and viral spread to neighboring cells, possible evasion from the humoral immune response, long-term infection persistence, and increased cytotoxic responses (Fig.
7). These factors could provide an escape from neutralizing antibodies and resistance to possible antivirals and immunobiological agents [
33]. Due to location and impacts on chemical-physical properties, we hypothesize that the 62 and 123 amino acids in the DENV3 MG-20 isolate are related to an increased membrane fusion ability in viral infection, making it capable of infecting CNS cells. Additionally, syncytial formation after infection with DENV3 MG-20 could enable the virus to evade the antibody response, facilitating the infection of CNS cells.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.