Tracking the emergence of antigenic variants in influenza A virus epidemics in Brazil

Abstract

Influenza A virus (IAV) circulation patterns differ in North America and South America, with influenza seasons often characterized by different subtypes and strains. However, South America is relatively undersampled considering the size of its population. To address this gap, we sequenced the complete genomes of 220 IAVs collected between 2009 and 2016 from hospitalized patients in southern Brazil. New genetic drift variants were introduced into southern Brazil each season from a global gene pool, including four H3N2 clades (3c, 3c2, 3c3, and 3c2a) and five H1N1pdm clades (clades 6, 7, 6b, 6c, and 6b1). In 2016, H1N1pdm viruses belonging to a new 6b1 clade caused a severe influenza epidemic in southern Brazil that arrived early and spread rapidly, peaking mid-autumn. Inhibition assays showed that the A/California/07/2009(H1N1) vaccine strain did not protect well against 6b1 viruses. Phylogenetically, most 6b1 sequences that circulated in southern Brazil belong to a single transmission cluster that rapidly diffused across susceptible populations, leading to the highest levels of influenza hospitalization and mortality seen since the 2009 pandemic. Continuous genomic surveillance is needed to monitor rapidly evolving IAVs for vaccine strain selection and understand their epidemiological impact in understudied regions.

Introduction

The estimated burden of influenza A virus (IAV) annual epidemics is between 3 and 5 million cases of severe illness worldwide, leading to 250,000–500,000 deaths (Iuliano et al. 2018; Paget et al. 2019). IAV evolution follows a ladder-like pattern in which new variants continually emerge and outcompete prior variants that go extinct. Predicting which antigenically variable IAV strains will circulate in upcoming seasons presents a constant challenge for public health efforts. New models infer viral fitness and immune escape from genetic and antigenic data to predict which circulating variant will become globally dominant the next year (Smith et al. 2004; Steinbruck, Klingen, and McHardy 2014; Morris et al. 2018). However, such models require high-quality IAV surveillance data collected globally each year, and persisting geographic gaps undermine prediction efforts (Krammer et al. 2018; Hammond et al. 2022). Since 2009, H1N1pdm and H3N2 subtypes have been co-circulated in human populations, with the dominant subtype alternating between epidemic seasons. Co-circulation of IAV subtypes can lead to co-infections and, although rare, mixed subtype infections have been reported in prior human epidemics (Ju et al. 2010; Liu et al. 2010; Rith et al. 2015; Gregianini et al. 2019).

IAV circulation patterns differ in North America and South America, with influenza seasons often characterized by different subtypes and strains (Russell et al. 2008). New variants often emerge in East and South-East Asia (Russell et al. 2008) and rapidly invade the rest of the world following major routes of passenger air travel (Lemey et al. 2014). The virus does not persist in the northern or southern hemisphere between winter epidemics, but rather continually migrates back and forth between hemispheres (Nelson et al. 2007; Rambaut et al. 2008). South America is relatively less connected by air travel and appears to serve as a global ‘sink’ for viral diversity, rather than a source (Russell et al. 2008). However, models of global IAV evolution are hindered by the low availability of genetic data from undersampled geographic locations (Liu et al. 2022), including many countries in South America (Layan et al. 2023).

To address this gap, we analyzed the genetic diversity of IAV from seven consecutive seasonal epidemics by sequencing influenza viruses in 220 primary nasopharyngeal swabs collected between 2009 and 2016 in hospitalized patients in the state of Rio Grande do Sul (RS) in southern Brazil. This allowed us to characterize circulating strains in southern Brazil across influenza seasons and compare them to strains circulating globally. Through genetic analysis, we determined that a severe H1N1pdm epidemic in 2016 that led to an increase in hospitalizations and deaths in RS (Gregianini et al. 2019) was caused by a new variant of H1N1pdm (clade 6b1), which had acquired new mutations in the HA gene and other regions of the genome.

Results

Severe H1N1pdm epidemic in 2016

Following the 2009 H1N1pdm pandemic, the H1N1pdm virus settled into a pattern of annual seasonal cycles with predictable peaks during the winter months (June–September) in southern Brazil (Fig. 1A). However, in 2016 a severe outbreak of H1N1pdm swept through RS in autumn, peaking in April and May (Fig. 1A). Historically, the seasonal H1N1 (i.e. pre-2009 H1N1 pandemic) virus caused milder epidemics compared to the seasonal H3N2 virus, with fewer deaths (Thompson et al. 2003), but the 2016 H1N1pdm epidemic caused more influenza fatalities in southern Brazil than any of the prior six seasons (Fig. 1B). In contrast to the 2009 H1N1pdm pandemic, deaths in 2016 primarily occurred in seniors (Fig. 1B, Table S2), representing an age shift toward the elderly that is expected for seasonal influenza (Simonsen et al. 1998). To investigate the age structure of fatalities further, patients were categorized into three age groups (children, adults, and seniors), and the results of Fisher’s exact test revealed a significant relationship between age structure and two influenza seasons in 2009 and 2016 (P < 0.001). Of a total of 810 individuals hospitalized due to H1N1pdm09 infection in 2016, 152 had been vaccinated (18.8 per cent), although vaccine uptake in Brazil is low and most hospitalized patients were unvaccinated (Fig. 1C). Similarly, more deaths were observed in the unvaccinated (Fig. S1). Large numbers of cases (48.1 per cent) occurred in Porto Alegre, the capital of RS, but the epidemic spread across the state (Fig. 1D). To understand why the 2016 epidemic was so deadly compared to prior years and to determine whether this could be due to genetic variability of the virus circulating at that time, we sequenced IAV from 220 nasopharyngeal aspirates from hospitalized patients who tested positive for IAV between 2009 and 2016. Based on the consensus sequences obtained for each segment, we identified two samples as seasonal H1N1, 147 as H1N1pdm and 71 as H3N2. The demographic data related to the sequenced samples are summarized in Table 1 and show that most patients had not been vaccinated. Sixty-two were fatality cases (Table S1), and most fatalities (65.6 per cent) occurred in individuals aged 50 and above (Table S2).

Figure 1.

Epidemiology of influenza A in Rio Grande do Sul, Brazil. (A) Weekly lab-confirmed H1N1pdm influenza A virus cases in RS during 2010–2016. Each line represents one of the seven influenza seasons. (B) Total influenza-associated deaths in RS by age group. (C) Total influenza-associated hospitalizations in RS by vaccination status, provided for the years 2013–2016 for which subtype information is available. The results of Fisher’s exact test for each subtype do not indicate a significant association between the proportion of vaccinated/unvaccinated cases and seasons. The P-value was 0.44 for H1N1 and 0.62 for H3N2. (D) Map of H1N1pdm cases in 2016 associated with a 6b1 cluster. The size of the circle is proportional to the number of cases. In the inset, the box indicates the location of RS, the southernmost state in Brazil.

Table 1.

The demographic data of IAV samples sequenced in RS.

	Subtypea
	H1N1pdm09	H3N2	H1N1	Total
Variables	(n = 147)	(n = 71)	(n = 2)	(n = 220)
Age (years)
<2	27	17	0	44
2–10	14	9	0	23
11–20	17	3	0	20
21–35	29	14	1	43
36–50	25	5	1	30
51–65	25	8	0	33
66–80	9	8	0	17
>81	1	7	0	8
Sex (n, %)b
Male	75 (51)	31 (43.7)	0	106
Female	72 (49)	40 (56.3)	2	114
Vaccine (n, %)b
Yes	19 (12.9)	14 (19.7)	–	33
No	113 (76.9)	45 (63.4)	–	158
NIc	15 (10.2)	12 (16.9)	2	29
Fatality (n, %)b
Yes	43 (29.3)	19 (26.8)	0	62
No	104 (70.7)	52 (73.2)	2	158
Co-infection with different subtype (n, %)b
Yes	9 (6.1)	20 (28.2)	0	29
No	138 (93.9)	51 (71.8)	2	191

^aSubtype according to the consensus sequence of segment 4 (HA).

^bPercentage of individuals for each subtype.

^cNo information available.

Evolution of H1N1pdm, 2009–2016

H1N1pdm viruses evolved continuously during 2009–2016 on a global scale, and new genetically drifted variants were reintroduced into RS at the start of each annual epidemic (Fig. 2A). The H1N1pdm drift variants had distinguishing mutations in the HA1 region of the hemagglutinin surface protein and were categorized into nine clades by Nextstrain (Hadfield et al. 2018) (Table S3). In addition to viruses similar to the original 2009 H1N1 pandemic reference strain (A/California/07/2009(H1N1)), five H1N1pdm clades circulated in RS: clade 6 (2011, 2012, and 2013), clade 7 (2011, 2012, and 2013), clade 6b (2013, 2014, and 2016), clade 6c (2013), and clade 6b1 (2016) (Table S4). No H1N1pdm viruses were detected in RS in 2010 or 2015. The 2016 epidemic was dominated by a new H1N1pdm clade referred to as 6b1; only one virus from clade 6 was detected that year. A 6b1 virus (A/Michigan/45/2015(H1N1)) was included in the southern hemisphere’s influenza vaccine in 2017.

Figure 2.

H1N1pdm clades. (A) A phylogenetic tree was inferred using maximum likelihood methods for HA sequences from H1N1pdm viruses collected globally during 2009–2016, including 134 viruses sequenced from RS for this study and 4,542 H1 background sequences downloaded from GISAID and subsampled by region and year. Tips and associated branches are shaded by clade, except for the 134 viruses from RS, which are shaded black. Three RS transmission clusters are shaded in black and highlighted. Phylogenies for the other seven segments are provided in Figures S2–S8. (B) Each box represents the time period that an H1N1pdm clade was detected by surveillance globally, in South America, or in RS, using RS sequences and a larger background dataset of H1 sequences from 9,683 H1N1 viruses collected globally during 2009–2016 that were downloaded from GISAID. The six RS boxes are indicated and highlighted in red font.

H1N1pdm clades were not detected in South America or RS before global detection; there was also no evidence of new clades originating in South America (Fig. 2B). Clades 2 and 3 were never detected in South America. In some cases (e.g. clade 6), there was a long lag between the circulation of a clade globally and its detection in South America or RS. Clade 6b1 viruses, for example, circulated globally for one year before their detection in South America and, a few months later, in RS.

H1N1pdm transmission clusters in RS

Viruses collected in RS were scattered throughout the phylogenetic tree, evidence of multiple independent viral introductions each year from a global gene pool (Figs S1–S7). However, discrete clusters of RS viruses, supported by bootstrap values >70, were observed during 2011 (n = 6 viruses, CA07-like), 2012 (n = 25 viruses, clade 6), and 2016 (n = 22 viruses, clade 6b1) that provide evidence of local transmission. The 2011 cluster (cluster 1; Fig. 2A) was observed in three locations in RS; the 2012 cluster (cluster 2; Fig. 2A) was observed in 19 locations; and the 2016 cluster (cluster 3; Fig. 2A) was observed in 13 locations. The 2012 cluster also contained one virus from 2011 (LACENRS-1186) and one virus from 2013 (LACENRS-2459), possible evidence of limited persistence of clade 6 viruses in RS across 2011–2013. Local transmission in RS provided opportunities for the virus to acquire new mutations across the genome that were not observed outside that cluster (Table S5). All three clusters included patients who died, but deaths did not cluster together as defined sub-clades or share specific mutations.

Decreased hemagglutination inhibition activity against 2016 H1N1pdm strains

The antigenic match between a vaccine and the circulating viruses is an important factor in vaccine efficacy. To assess the antigenic match between the 2016 H1N1pdm strains circulating in RS and vaccine strains, we measured hemagglutination inhibition activity against two representative 2016 strains (A/Gramado/LACENRS-1287/2016 and A/Tres Coroas/LACENRS-877/2016) using ferret antibodies specific for A/California/07/2009(H1N1), which was the reference strain for the southern hemisphere’s 2016 flu vaccine. The antiserum against A/California/07/2009 cross-reacted with RS H1N1pdm strains with titers four-fold lower than the homologous titer (Table S6). Variant sites (compared to A/California/07/2009) in the HA sequences of A/Gramado/LACENRS-1287/2016 and A/Tres Coroas/LACENRS-877/2016 are shown in Table S6. Some mutations were located in antigenic sites of the HA protein (Matsuzaki et al. 2014), such as S165N (HA1 numbering; corresponding to residue 179 of the full protein) and K166Q (residue 180) in antigenic site Sa, which are mutations now commonly found in viruses around the globe (Neher and Bedford 2015). We also observed mutations S188T (residue 202) in antigenic site Sb and S206T (residue 220) in antigenic site Ca1 (Table S7).

Evolution of H3N2, 2009–2016

H3N2 viruses co-circulated with H1N1pdm viruses in RS during 2011–2016, and they were the dominant subtypes in 2014 and 2015. Four H3N2 clades were identified in RS during 2011–2016: 3c (2011, 2012, and 2014), 3c2 (2013), 3c3 (2013, 2014, and 2015), and 3c2a (2015 and 2016). In 2013, the updated southern hemisphere H3N2 influenza vaccine strain was a clade 3c virus—A/Victoria/361/2011(H3N2). However, during that time clades 3c2 and 3c3 were circulating in RS. Similarly in 2015, the updated southern hemisphere vaccine strain was a 3c3a virus—A/Switzerland/9,715,293/2013(H3N2), but 3c3 and 3c2a viruses were circulating in RS (Fig. 3). In no year of our study where H3N2 viruses were detected (2011–2015) did the clade of the southern hemisphere influenza vaccine match the clade of the strains circulating in RS.

Figure 3.

Timing of H3N2 clades. Each box represents the time period that an H3N2 clade was detected by surveillance globally, in South America, or in RS, based on a dataset of H3 sequences from 9,416 H3N2 viruses collected globally during 2009–2016 that were downloaded from GISAID and 61 H3 sequences from RS viruses collected for this study. The four RS boxes are indicated and highlighted in red font.

Discussion

The ongoing evolution of IAVs presents a challenge for public health and influenza vaccine strain selection. In theory, the delay between the emergence of new flu variants in other parts of the world and their arrival in RS provides time for scientists to gather data and update vaccines to match strains. However, the evolution of H3N2 viruses became more complex following the 2009 H1N1pdm pandemic, and multiple H3N2 clades now persist and cycle globally, presenting a greater challenge for vaccine strain selection. In no year of our study did the H3N2 strain included in the southern hemisphere influenza vaccine match the H3N2 strain in circulation in RS. The 2017 update to the southern hemisphere H1N1pdm vaccine strain occurred one year later to protect RS against the 6b1 viruses that increased hospitalizations and deaths in 2016.

The severe H1N1pdm epidemic in 2016 in RS was unexpected. For one, the H3N2 subtype is more often the cause of severe influenza seasonal epidemics (Thompson et al. 2003), and H3N2 was the dominant subtype circulating in RS in 2015. Little antigenic evolution had been observed in the H1N1pdm virus since the 2009 pandemic, requiring no updates to the H1N1pdm vaccine strain for 6 years. Presumably, after 5–6 years of H1N1pdm circulation, population immunity had been built against H1N1pdm on a global scale, increasing selection pressure for new antigenic variants. The severe H1N1pdm epidemic in Brazil was also unexpected, given that the USA did not have a severe influenza season in 2015–2016. (The USA did have a severe flu season in 2016–2017, but it was dominated by H3N2 viruses.) The number of cases and genetic composition of the winter influenza season in North America is not always predictive of what is seen in the southern hemisphere 6 months later. Likewise, the subtype that circulated during winter in the southern hemisphere was not necessarily the same that circulated later on in the northern hemisphere. It remains unclear whether new mutations that occurred in the 6b1 South American viruses had any phenotypic effect. Our antigenic assays confirmed that people infected with 6b1 viruses in RS were poorly protected by the A/California/07/2009(H1N1) vaccine strain, but we did not compare the immunological response to the Brazilian strains against the US strains circulating the prior winter.

Although there are important differences in the evolution and global ecology of influenza and the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the emergence of P.1 (gamma) in early 2021 and other SARS-CoV-2 variants in South America underscores the importance of pathogen surveillance in understudied regions (Faria et al. 2021; Varela et al. 2021). The southern hemisphere influenza season is not necessarily a mirror image of the northern hemisphere, 6 months delayed, or vice versa. During the coronavirus disease 2019 pandemic, there has been remarkable progress in expanding pathogen surveillance in understudied regions, facilitated by the adoption of portable nanopore sequencing technologies (Bull et al. 2020). Whether high-throughput sequencing operations built for SARS-CoV-2 will be maintained for other pathogens, including influenza, is an open question. Governments have begun to shift funding priorities as the pandemic phase abates. The window of opportunity is narrowing to preserve genomic sequencing infrastructure for long-term global health gains.

Materials and methods

Biological samples and epidemiological data

Nasopharyngeal aspirates were collected between 2009 and 2016 in the state of RS, southern Brazil (estimated population: 11.29 million people) from patients with acute respiratory infection. These specimens had been previously analyzed by reverse transcription quantitative polymerase chain reaction (RT-qPCR) at the State Central Laboratory (LACEN-RS) for IAV confirmation and were included in this study when positive for IAV. Samples are listed in Supplementary Table S1. In Brazil, acute respiratory infection became of universal notification in 2009 in response to the H1N1pdm09 pandemic; for each patient, a Notification Form is filled out by the attending physician/nurse, at the time of collection (Saúde 2019). The data are registered in SINAN (Information System on Diseases of Compulsory Declaration), a digital platform used by the Brazilian Health Ministry for reporting notifiable conditions (http://portalsinan.saude.gov.br). For each patient of this study, the following data were retrieved from SINAN: age, sex, date of notification, date of onset of symptoms, vaccination status, acute respiratory infection symptoms, comorbidities, smoking habits, pregnancy status, X-ray results (when available), and disease outcome (cure or death demographic characteristics). In addition, the data on the total number of cases with respiratory infection who had a positive RT-qPCR for influenza in RS between 2009 and 2016 were also retrieved from SINAN.

Virus isolation and HA antigenic analyses by hemagglutination inhibition assay

To analyze antigenic differences in the H1N1pdm strains that circulated in 2016 in relation to the vaccine H1N1pdm strain A/California/07/2009, we selected forty representative circulating samples that came from different HA phylogenetic clades and passaged in susceptible Madin-Darby canine kidney cells. Two viruses (A/Gramado/LACENRS-1287/2016 and A/Tres Coroas/LACENRS-877/2016) were isolated, and hemagglutination inhibition assays were performed. Briefly, 0.5 per cent turkey erythrocytes and ferret antisera against influenza A/California/07/2009 virus were used. Ferret sera were treated with Vibrio cholerae receptor-destroying enzyme before being tested. The homologous A/California/07/2009 and circulating viruses A/Gramado/LACENRS-1287/2016 and A/Tres Coroas/LACENRS-877/2016 were compared for hemagglutination inhibition activities.

RNA extraction and multi-segment RT-PCR

Ribonucleic acid (RNA) was extracted from 140 µl of nasopharyngeal aspirate samples using the E.Z.N.A. Viral RNA Kit (OMEGA Bio-tek) and used in a multi-segment reverse transcription polymerase chain reaction (RT-PCR) that amplifies all eight IAV genome segments (Zhou et al. 2009) using the One-Step SuperScript III High-fidelity RT-PCR kit (Invitrogen). Briefly, 5 µl of RNA were added to 45 µl of master mix containing 17 µl DEPC-treated water, 25 µl of 2x RT-PCR buffer mix, primer Opti1-F1 (5ʹ-GTTACGCGCCAGCAAAAGCAGG-3ʹ, 70 nM final concentration), primer Opti1-F2 (5ʹ-GTTACGCGCGCCAGCGAAAGCAGG-3ʹ, 130 nM final concentration), primer Opti1-R1 (5ʹ-GTTACGCGCCAGTAGAAACAAGG-3ʹ, 200 nM final concentration), and 1 µl RT/Taq enzyme. Reaction conditions were 55 ºC for 2 min and 42 ºC for 60 min for reverse transcription; 94 ºC for 2 min; five cycles at 94 ºC for 30 s, 44 ºC for 30 s, and 68 ºC for 120 s; then thirty cycles at 94 ºC for 30 s, 57 ºC for 30 s, and 68 ºC for 120 s; final extension at 68 ºC for 10 min. Amplicons were visualized by electrophoresis on 0.8 per cent agarose gels containing ethidium bromide (0.5 µg/ml) and purified with 0.45x Agencourt AMPure XP beads (Beckman Coulter).

Sequencing and genome assembly

Purified amplicons (0.2–1 µg) were sheared to an average fragment size of 150 bp on a Bioruptor Pico sonicator (Diagenode). Next, amplicon sequence libraries were prepared using the end repair, A-tailing, and adaptor ligation NEBNext DNA library prep modules for Illumina from New England Biolabs, or Nextera XT DNA Library Prep Kit from Illumina, according to the manufacturer’s protocol. Following the final size selection, multiplexed libraries were sequenced on the Illumina HiSeq 2500 platform in a single-end 100nt run format (NEBNext) or on the Illumina MiSeq platform in a paired-end 2x150nt or 2x250nt run format (Nextera). Reads were first filtered with cutadapt v2.8 (Martin 2011) to remove low-quality sequences and adapters. Reads were then mapped against a non-redundant copy of IAV sequences in the Influenza Research Database (IRD; fludb.org) using STAR v2.4.1d (Dobin et al. 2013), and chimeric reads with non-contiguous alignments to reference segments (typically originating from defective interfering particles containing segments with internal deletions) were removed. Assembly of IAV genomic segments was performed using a custom FluGAP pipeline (https://bitbucket.org/bakellab/flugap/) in multiple stages. First, an initial assembly was done using the inchworm component of Trinity v2.8.5 (Grabherr et al. 2011), and viral contigs bearing internal deletions were identified by BLAT v36 (Kent 2002) mapping against non-redundant IRD reference sequences. In the second stage, the inchworm assembly was repeated removing breakpoint-spanning k-mers from the assembly graph. The resulting IAV contigs were then oriented and trimmed to remove low-coverage ends and any extraneous sequences beyond the conserved IAV termini. In the final stage, the CAP3 v12/21/07 (Huang and Madan 1999) assembler was used to improve contiguity by merging contigs originating from the same segment type if their ends overlapped by at least 25nt. Assembly quality and contiguity were assessed for all segments by mapping sequence reads back to the final assemblies using BWA v0.7.17 (Li and Durbin 2009), and complete segments were annotated using the NCBI Influenza Virus Sequence Annotation Tool (Bao et al. 2007).

Phylogenetic analysis

The whole genomes of H1N1pdm (n = 9,683) and H3N2 (n = 9,416) influenza sequences collected between 2009 and 2016 were downloaded from the Global Initiative on Sharing All Influenza Data (GISAID) database. To create trees that could be visualized more easily, subsampled datasets were generated for H1N1pdm and H3N2 that included up to 150 viruses per region per year. The regions included North and Central America, South America, Europe, Africa, West and South Asia, Southeast Asia, East Asia, and Oceania. The viruses collected in RS were added to the full and down-sampled datasets. Duplicate sequences and outliers were removed. Viruses for which complete genome sequences were not available were removed. The down-sampled datasets consisted of 4,676 H1N1pdm and 4,562 H3N2 sequences. The datasets were aligned using MAFFT v7.475 (Katoh and Standley 2013). HA sequences were assigned to clades based on defining amino acid positions using the align_clades.py script available from the seasonal influenza build of Nextstrain (Hadfield et al. 2018). Phylogenetic relationships were inferred using the maximum likelihood method available in IQTree 2.0.3 (Nguyen et al. 2015), using a general time-reversible model of nucleotide substitution with a gamma-distributed rate variation among sites. Due to the size of the dataset, we used the high-performance computational capabilities of the Biowulf Linux cluster at the National Institutes of Health (http://biowulf.nih.gov). To assess the robustness of each node, a bootstrap resampling process was performed (1,000 replicates).

Data availability

All sequence data generated as part of this project are available at https://www.ncbi.nlm.nih.gov/bioproject/379005 (NCBI, BioProject ID 379005, Accession PRJNA379005).

Supplementary data

Supplementary data is available at Virus Evolution online.

Funding

A.B.G.V. was granted a Visiting Scholar Fellowship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Ministry of Education, Brazil) to perform the analysis and currently holds a PQ2 Research Fellowship from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil). This work was supported in part by the Division of Intramural Research of the National Institute of Allergy and Infectious Diseases/ National Institutes of Health (NIAID/NIH) (E.G., M.I.N.). This work was also partly supported by Center for Research on Influenza Pathogenesis and Transmission, an NIAID Center of Excellence on Influenza Research and Response, contract # 75N93021C00014) to A.G.-S., R.A.M., and H.V.B.

Conflict of interest:

The A.G.-S.’s laboratory, in the Department of Microbiology at Icahn School of Medicine at Mount Sinai, has received research support from GSK, Pfizer, Senhwa Biosciences, Kenall Manufacturing, Blade Therapeutics, Avimex, Johnson & Johnson, Dynavax, 7Hills Pharma, Pharmamar, ImmunityBio, Accurius, Nanocomposix, Hexamer, N-fold LLC, Model Medicines, Atea Pharma, Applied Biological Laboratories, and Merck, outside of the reported work. A.G.-S. has consulting agreements for the following companies involving cash and/or stock: Castlevax, Amovir, Vivaldi Biosciences, Contrafect, 7Hills Pharma, Avimex, Pagoda, Accurius, Esperovax, Farmak, Applied Biological Laboratories, Pharmamar, CureLab Oncology, CureLab Veterinary, Synairgen, Paratus, and Pfizer, outside of the reported work. A.G.-S. has been an invited speaker in meeting events organized by Seqirus, Janssen, Abbott, and Astrazeneca. A.G.-S. is inventor on patents and patent applications on the use of antivirals and vaccines for the treatment and prevention of virus infections and cancer, owned by the Icahn School of Medicine at Mount Sinai, New York, outside of the reported work.

Ethics statement

Experiments were performed in compliance with relevant laws and in accordance with the ethical standards of the Declaration of Helsinki. All ethical issues were previously approved by the Research Ethics Committee of Universidade Federal de Ciências da Saúde de Porto Alegre and Mount Sinai Hospital. All samples used in this study have been collected for health surveillance, diagnosis, and treatment of patients, without written Informed Consent. The acquisition of the samples adheres to the regulations and ethical guidelines for the protection of human subjects of research in Brazil, and all research activities involving human subjects were conducted according to these ethical principles. The samples are labeled in a coded, de-identified manner. Patient identifiers are not disclosed, and researchers have access only to information important for the objectives of the study, such as clinical and demographic data (age, gender, and city).