**2. Results**

### *2.1. Novel Evidence of WNV Circulation in Three Brazilian States*

Samples (RBCs) from three horses with suspected WNV infection obtained from southern (Minas Gerais and São Paulo) and northeastern (Piauí) Brazilian states were sent for molecular diagnosis at the Departamento de Medicina Veterinária Preventiva at the Federal University of Minas Gerais (UFMG).

RNAs were extracted from red blood cells and tested using an in-house PCR assay (see Methods section for details). WNV-specific RT-PCR amplification products were obtained by nested PCR (Figure 1A,B), and positive samples were subjected to a newly designer multiplex PCR scheme (Supplementary Table S1) to generate complete genomes sequences by means of portable nanopore sequencing.

Three blood fractions (plasma, buffy coat, and washed RBC) from the horses sampled in São Paulo and Minas Gerais states have been submitted to nested RT-PCR; horse samples from Piaui have been tested only using RBC, which was the only blood fraction available. Diagnostic investigation of alphavirus was also performed using a generic RT-PCR targeting the NSP1 gene, according to [22], in the three blood fractions, with negative results.

The published WNV genome from Brazil (MH643887) was used to generate (mean) 98.4% consensus sequences that formed the target for primer design. The new genomes were deposited in GenBank with accession numbers MW420987, MW420988, and MW420989 (Table 1).

We constructed phylogenetic trees to explore the relationship of the sequenced genomes to those sampled elsewhere globally. We retrieved 2321 WNV genome sequences with associated lineage date and country of collection from GenBank, from which we generated a subset that included the highly supported (>0.9) clade containing the newly WNV strains obtained in this study plus 29 sequences (randomly sampled) from all lineages and performed phylogenetic analysis. An automated online phylogenetic tool to identify and classify WNV sequences was developed (available at: http://krisp.ukzn.ac.za/app/typingtool/wnv/ job/9b40f631-51c4-419c-9edf-2206e7cd8d9c/interactive-tree/phylo-WNV.xml accessed on 31 December 2019).

Phylogenies estimated by the newly developed WNV typing tool, along with maximum likelihood methods (Supplementary Figure S1C), consistently placed the Brazilian genomes in a single clade within the 1a lineage with maximum statistical support (bootstrap = 100%) (Supplementary Figure S1).

Time-resolved maximum likelihood tree appeared to be consistent with previous estimates [11] and showed that the new genomes clustered with strong bootstrap support (97%) with a WNV strain isolated from an *Aedes albopictus* mosquito in Washington DC, USA in 2019 (Figure 1D). Interestingly, the new isolates did not group with the previously sequenced genome in 2019 from the Espirito Santo state, suggesting that inter-continental introduction events might be frequent in Brazil.

**Figure 1.** Investigation of WNV infections in Brazil, between July 2018 and September 2020, and estimated transmission potential. (**A**,**B**) Agarose gel electrophoresis of amplicons from assay for WNV. (**A**) nested RT-PCR. MW (Molecular weight ladder), 100 bp DNA Ladder RTU, Kasvi; 1—plasma of horse from São Paulo; 2—buffy coat of horse from São Paulo; 3—washed RBC of horse from São Paulo; 4—blank negative control using during the nested RT-PCR; 5 and 6—positive control (synthetic gene); NTC, no template control (using since the extraction); expected amplicon size: 370 bp. (**B**) Multiplex PCR. MW (Molecular weight ladder), Fluorescent 100 bp DNA Ladder, Cellco, Jena Bioscience; 1—horse form Minas Gerais (pair primers); 2—horse form Minas Gerais (odd primers); 3—horse form Sao Paulo (pair primers); 4—horse form Sao Paulo (impair primers); 5—horse form Piaui (pair primers); 6—horse form Piaui (odd primers); NTC, no template control (using since the extraction); expected amplicon size: 400 bp. (**C**) Midpoint rooted maximum-likelihood phylogeny of WNV genomes, showing major lineages. The scale bar is in units of substitutions per site (*s/s*). Support for branching structure is shown by bootstrap values at nodes. (**D**) Time-resolved maximum likelihood tree showing the WNV strains belonged to the 1a lineage. Colors indicate geographic location of sampling. The new Brazilian WNV strains are shown with text in red.



M = Male; F = Female; Accession Number = NCBI accession number.

### *2.2. A Data-Driven WNV Theoretical Perspective*

We first summarized the past evidence of WNV circulation in Brazil from avian species, equines, and humans, which was achieved via various literature reports using different confirmation methods (Figure 2A) [23]. The first evidence of WNV infection was documented in 2004 in horses in northeastern Brazil (Paraiba state). Since then, serological evidence of WNV infection continued to be documented between 2008 and 2010 and again in 2020 in horses and birds from the southern [24], midwestern (Pantanal), and northern Brazilian regions. In 2014, the first WNV infection in a human was confirmed in the Piauí State (northeast region). In 2018, the first isolation of WNV in Brazil was documented in the Espirito Santo state (southeastern Brazil) when the virus was isolated from the central nervous system (CNS) of a dead horse with neurological manifestations [11]. To these data, we here add the report of the new genetic evidence of WNV circulation in equines occurring between 2018 and 2020, in southern (Minas Gerais, São Paulo) and northeastern (Piauí) states. To the best of our knowledge, it is the first time that evidence of WNV circulation is reported for the states of Minas Gerais and São Paulo.

Using data collected from the Brazilian "Sistema de Informação de Agravos de Notificação" (SINAN) (see Methods and Supplementary Table S2) reported with identifier A923 ("Febre do Nilo"), we explored the current spatio-temporal distribution of suspected cases of West Nile fever. Given the unspecific and unconfirmed nature of these reported cases, we complemented such information with theoretical projections of the spatio-temporal transmission potential of WNV in Brazil. For this, we used a climate-driven suitability measure (index P) previously successfully applied to WNV in the contexts of Israel [25] and Portugal [26] (see Methods).

We mapped the mean index P across Brazil for the period 2015–2019 (Figure 2B) and found estimated transmission potential to be highest in the center of the country along a diagonal latitude–longitude axis crossing from the center–west to the north–east. The regions of the south of the country, similarly to estimations for other mosquito-borne viruses [27], presented the least transmission potential. To assess potential hotspots of (at least temporary) high transmission potential, we calculated the proportion of months (2015–2019) in which the index P was above 1; this particular threshold representing the point above which each female mosquitoes would be theoretically able to infect more than one host during their lifetime. This approach identified regions of Piauí, Bahia, Ceará, Rio Grande do Norte, and Paraíba states as presenting significantly longer periods of time with high index P. In particular, the state of Piauí was captured in its entirety within this estimated spatial hotspot of transmission potential (Figure 2C).

From all states for which there were reported cases, we filtered those that had more than one case per any month during the entire period of 2015–2019, selecting only two states with clear epidemic waves of reported cases: Piauí and Espírito Santo. Coincidently with the results of Figure 2B,C, the state of Piauí reported the largest number of cases in the entire dataset. Using the geographical boundaries of each state, we averaged the index P per month (Figure 2D,E). The resulting time series of transmission potential showed that potential was higher in Piauí compared to Espírito Santo in accordance with the spatial output in Figure 2B,C. It also presented a clear seasonal signal, with peaks occurring on average in February in Piauí (month average = 2.2, summer) and April in Espiríto Santo (month average = 4, autumn). The correlation between reported cases and the index for Piauí was positive (Pearson's 0.36, Figure 2D), but it was negative for Espiríto Santo (Pearson's −0.31, Figure 2E). Similar to what has been reported for suitability indices applied to other viruses [27], there was a clear lag between the index and cases for Piauí, with cases lagging behind the index (Figure 2D). Accordingly, shifting the index by one month into the future resulted in a high positive correlation with cases (Pearson's 0.84).

**Figure 2.** Data-driven epidemiological perspective of WNV in Brazil. (**A**) Mapping of historic evidence for WNV circulation in Brazil, for which the color and symbol legend on the bottom left of the panel define the animal source and methodology. Data are based on a literature review up to 2019 [24], in addition with recently published reports in 2020–2021 [24] and the new data generated in this study. (**B**) Mean estimated transmission potential of WNV (index P) over the period 2015–2019. The color scale on the bottom left of the panel shows the range of the presented values. The black borders mark the boundaries of the Piauí and Espiríto Santo states. (**C**) Proportion of months for which the transmission potential of WNV (index P) was above the value 1, over the period 2015–2019. The color scale on the bottom left of the panel shows the range of the presented values. The black borders mark the boundaries of the Piauí and Espiríto Santo states. (**D**) Time series of suspected reported West Nile fever cases (bars) and estimated transmission potential of WNV (index P, blue line) for the Piauí state. Index P is the average per month, across all data points within the boundaries of the state. (**E**) Time series of suspected reported West Nile fever cases (bars) and estimated transmission potential of WNV (index P, green line) for the Espiríto Santo state. Index P is the average per month, across all data points within the boundaries of the state. (**F**) Spatial snapshot of estimated transmission potential of WNV (index P) for the month of March 2016. Color scale on the right shows the range of the presented values. (**G**) Same as F but for June 2016. (**H**) Same as F but for September 2016. (**I**) Same as F but for December 2016.

Finally, to ge<sup>t</sup> a grasp of the possible spatio-temporal dynamics of WNV transmission in Piauí, we looked at estimated transmission potential for one of the years with more reported suspected cases (2016) both in space and time (with snapshots at months of March, June, September, and December) (Figure 2F–I). The spatio-temporal snapshots showed that transmission potential was the lowest during winter months, but we also highlighted that this was almost uniform across the state (Figure 2G,H). In contrast, throughout the year, this output highlighted a possible wave of seasonal transmission. This wave would typically start in the southwest just before summer (Figure 2I) and would move to the northeast in the summer (Figure 2F).
