Field and classroom initiatives for portable sequence-based monitoring of dengue virus in Brazil

Abstract

Brazil experienced a large dengue virus (DENV) epidemic in 2019, highlighting a continuous struggle with effective control and public health preparedness. Using Oxford Nanopore sequencing, we led field and classroom initiatives for the monitoring of DENV in Brazil, generating 227 novel genome sequences of DENV1-2 from 85 municipalities (2015-2019). This equated to an over 50% increase in the number of DENV genomes from Brazil available in public databases. Using both phylogenetic and epidemiological models we retrospectively reconstructed the recent transmission history of DENV1-2. Phylogenetic analysis revealed complex patterns of transmission, with both lineage co-circulation and replacement. We identified two lineages within the DENV2 BR-4 clade, for which we estimated the effective reproduction number and pattern of seasonality. Overall, the surveillance outputs and training initiative described here serve as a proof-of-concept for the utility of real-time portable sequencing for research and local capacity building in the genomic surveillance of emerging viruses.

Fig. 1. Spatial and temporal distribution of reported dengue cases in Brazil, 2015–2020.

a Map of Brazil showing the number of DENV1 and DENV2 new sequences by region and state. DF = Federal District, BA = Bahia state, GO = Goiás state, MT = Mato Grosso state, MS = Mato Grosso do Sul state, MG = Minas Gerais state, PE = Pernambuco state, RJ = Rio de Janeiro state, SP = São Paulo state. The color and size of the circles indicates the number of new genomes generated in this study (black = DENV1, white = DENV2). b Weekly notified dengue cases normalized per 100 K individuals per region in 2015-2020 (until EW06). Epidemic curves are colored according to geographical macro region: SE = Southeast, NE = Northeast, MW = Midwest, N = North, S = South. c–e Time series of weekly reported cases normalized per 100 K individuals and daily mosquito-viral suitability measure (index P) in the three macro regions for which new sequences were generated: MW = Midwest (C), NE = Northeast (D), and SE = Southeast (E). The index P of each region is obtained using average climatic data for the three largest urban centers in each region. Purple bars highlight the dates of sample collection of the genomes generated here. b–e Incidence (cases per 100 K population) is presented in log10 for visual purposes. The initial map of Brazilian regions was obtained from the R package “get_brmap” (available at: https://rdrr.io/cran/brazilmaps/man/get_brmap.html). Source data are provided as a Source Data file.

Fig. 2. Phylogenetic analysis of DENV1-V in Brazil.

a Maximum likelihood (ML) phylogenetic analysis of 57 complete genome sequences from DENV1 generated in this study plus 444 publicly available sequences from GenBank. The scale bar is in units of nucleotide substitutions per site (s/s) and the tree is mid-pointed rooted. Colors represent different sampling locations. b Time-scaled phylogeographic tree of Clade I (including eight new sequences plus 25 GenBank sequences), Clade II (including 27 new sequences plus seven GenBank sequences), and Clade III (including 22 new sequences plus 12 GenBank sequences). Colors represent different sampling locations (SE Brazil = Brazilian Southeast, NE Brazil = Brazilian Northeast, MW = Brazilian Midwest, N Brazil = Brazilian North). Tip circles represent the genome sequences generated in this study; colored sidebars represent the dengue clinical classification for each sequenced sample.

Fig. 3. Phylogenetic analysis of DENV2-III in Brazil.

a A maximum likelihood tree was inferred using 170 complete genome sequences from DENV2 generated in this study and 450 sequences publicly available, retrieved from GenBank. The scale bar is in units of nucleotide substitutions per site (s/s) and mid-point rooted. Tip circles represent the genome sequences generated in this study. b Time-scaled phylogeographic tree of DENV2 BR-4, including 164 new DENV2 sequences generated here and 17 publicly available data from the 2019 outbreak in Brazil. Sequences are colored according to sampling location. Sidebars represent the dengue clinical classification for each sequenced sample. c Temporal fluctuation of the effective reproduction number (R_e) of the for DENV2 BR-4L1 (blue) and BR-4L2 (magenta) estimated using the Bayesian birth-death approach. Black line is the total weekly incidence of dengue between 2015 and 2020 (until EW06), and the dotted green line is the index P (incidence is summed and index P is averaged over the three macro regions for which new sequences were generated: MW = Midwest, NE = Northeast, and SE = Southeast). Incidence (cases per 100 K population) is presented in log10 for visual purposes.