key: cord-0903688-lx8t0kro
authors: Moreira, Filipe Romero Rebello; D’arc, Mirela; Mariani, Diana; Herlinger, Alice Laschuk; Schiffler, Francine Bittencourt; Rossi, Átila Duque; Leitão, Isabela de Carvalho; Miranda, Thamiris dos Santos; Cosentino, Matheus Augusto Calvano; Tôrres, Marcelo Calado de Paula; Da Costa, Raíssa Mirella dos Santos Cunha; Gonçalves, Cássia Cristina Alves; Faffe, Débora Souza; Galliez, Rafael Mello; Junior, Orlando da Costa Ferreira; Aguiar, Renato Santana; dos Santos, André Felipe Andrade; Voloch, Carolina Moreira; Castiñeiras, Terezinha Marta Pereira Pinto; Tanuri, Amilcar
title: Epidemiological Dynamics of SARS-CoV-2 VOC Gamma in Rio de Janeiro, Brazil
date: 2021-10-01
journal: Virus Evol
DOI: 10.1093/ve/veab087
sha: f1b33e4584f7a2312a8c418c095cf2a25d1c655d
doc_id: 903688
cord_uid: lx8t0kro

The emergence and widespread circulation of SARS-CoV-2 variants of concern (VOC) or interest (VOI) imposes an enhanced threat to global public health. In Brazil, one of the countries most severely impacted throughout the pandemic, a complex dynamics involving variants co-circulation and turnover events has been recorded with the emergence and spread of VOC Gamma in Manaus in late 2020. In this context, we present a genomic epidemiology investigation based on samples collected between December 2020 and May 2021 in the second major Brazilian metropolis, Rio de Janeiro. By sequencing 244 novel genomes through all epidemiological weeks in this period, we were able to document the introduction and rapid dissemination of VOC Gamma in the city, driving the rise of the third local epidemic wave. Molecular clock analysis indicates this variant has circulated locally since the first weeks of 2021 and only seven weeks were necessary for it to achieve a frequency above 70%, consistent with rates of growth observed in Manaus and other states. Moreover, a Bayesian phylogeographic reconstruction indicates VOC Gamma spread throughout Brazil between December 2020 and January 2021, and that it was introduced in Rio de Janeiro through at least 13 events coming from nearly all regions of the country. Comparative analysis of RT-qPCR cycle threshold (Ct) values provides further evidence that VOC Gamma induces higher viral loads (N1 target; mean reduction of Ct: 2.7, 95% CI = ±0.7). This analysis corroborates the previously proposed mechanistic basis for this variant enhanced transmissibility and distinguished epidemiological behavior. Our results document the evolution of VOC Gamma and provide independent assessment of scenarios previously studied in Manaus, therefore contributing to the better understanding of the epidemiological dynamics currently being surveyed in other Brazilian regions.

VOIs N.9 (Resende et al., 2021a) and N.10 (Resende et al., 2021b) have originated in different Brazilian regions.

Interestingly, these lineages share mutations in biologically relevant sites of the receptor-binding domain (RBD) of the spike protein. For instance, the substitutions E484K and N501Y are implicated in the distinct epidemiological characteristics reported for VOCs Alpha, Beta and Gamma Tegally et al., 2020; Volz et al., 2021) . Some evidence suggest that VOCs can be more transmissible Naveca et al., 2021; Tegally et al., 2021; Volz et al., 2021) , cause increased disease severity (Davies et al., 2021; , and display reduced neutralization by antibodies elicited by previous infections or vaccines (Cele et al., 2021; Garcia-beltran et al., 2021) .

Given these distinct epidemiological behaviors and the putative fitness advantage of VOCs, current evidence supports that they have rapidly risen in frequency, becoming predominant, and quickly spreading to other regions (Tegally et al., 2020; Volz et al., 2021) . This phenomenon has already been recorded in Brazil. The VOI Zeta, which emerged in Rio de Janeiro in mid-July 2020, rapidly became the primary local lineage and was exported to several regions of the country and abroad . Likewise, the emergence of VOC Gamma in mid-November in Manaus is marked by its fast rise in frequency and a massive increase in the number of cases Naveca et al., 2021) . Currently, diverse surveillance efforts indicate that this lineage has become predominant in several country regions, replacing previously circulating variants (Barbosa et al., 2021; Franceschi et al., 2021; Moreira et al., 2021a; Naveca et al., 2021) . Moreover, VOCs that originated elsewhere, like Alpha, Beta, and Delta, have already been detected in Brazil (Claro et al., 2021; 5 Slavov et al., 2021; SES/MA, 2021) , adding up to a complex epidemiological scenario Souza et al., 2020) .

As the circulation of these lineages has broad epidemiological implications for public health, including ongoing vaccination efforts in Brazil, we sought to determine the genetic background of the SARS-CoV-2 epidemic in the city of Rio de Janeiro between early December 2020 and early May 2021. Rio de Janeiro is the second major Brazilian metropolis, an essential hub for business, and a majorly connected city by air travel (IATA, 2020) . By combining novel genomic and epidemiological data, we were able to characterize a comprehensive shift in the composition of the local SARS-CoV-2 epidemic population induced by VOC Gamma. In addition, we provide further evidence that this variant causes higher viral loads than previously circulating variants, reinforcing its enhanced transmissibility. Altogether, this study documents the complex epidemiological dynamics of SARS-CoV-2 in a major Brazilian metropolis and highlights a possible mechanism by which VOC Gamma dominates distinct epidemiological scenarios. 

The study population was composed by convenience sampling among the RT-qPCR positive cases amidst individuals evaluated at the COVID-19 Diagnostic Center of the Federal University of Rio de Janeiro, between 1st December 2020 and 12 May 2021 (Total number of tests: 14,080). Nasopharyngeal swab samples were collected from both nostrils, placed in viral transport medium (2 mL), and kept at 4°C until transportation to the laboratory. Total viral RNA from swab samples were extracted in a KingFisher Flex System ® (Thermofisher, USA), using the MagMax Viral/Pathogen Kit (Thermofisher, USA), according to manufacturer's instructions. Viral RNA was detected using the SARS-CoV-2 (2019-nCoV) multiplex CDC qPCR Probe Assay (Integrated DNA Technologies, USA) targeting the SARS-CoV-2 N1 and N2 genes, and the human ribonuclease P (RNaseP) gene (endogenous control). The GoTaq ® Probe 1-Step RT-qPCR System (Promega, USA) was used, according to the manufacturer's instructions. All reactions were performed in a 7500 Thermal Cycler (Applied Biosystems, USA).

The RT-qPCR result interpretation was as follows: Positive for SARS-CoV-2 when both targets (N1 and N2) amplified with Ct ≤ 37; Undetermined when only one target amplified with Ct ≤ 37, or both targets amplified with Ct between 37 and 40;

Negative when one or both targets amplified with Ct > 40, or absence of amplification.

The present study was approved by the local ethics review committee from Clementino Fraga Filho University Hospital (CAAE: 30161620.0.0000.5257) and by the national ethical review board (CAAE: 30127020.0.0000.0068). All enrolled participants were over 18 years old and declared written informed consent. 

In total, 278 RT-qPCR positive samples with Ct < 30, collected between 11

December 2020 and 5 May 2021, were selected for genome sequencing. The temporal distribution of samples comprehended two epidemic waves of SARS-CoV-2 in Rio de Janeiro city (COE/RJ, 2021). Sequencing was carried out using a widely employed protocol (Quick et al., 2017) . Briefly, viral RNA was converted to cDNA using SuperScript III or IV (ThermoFisher, USA), followed by a multiplex amplification reaction with the ARTIC SARS-CoV-2 v3 Panel and the Q5 hotstart polymerase (NewEngland Biolabs, USA). After purification, amplicons for each sample were normalized and converted into Illumina sequencing libraries using either the Nextera XT library kit (Illumina, USA) or the QIAseq FX library kit (QIAGEN, Germany), following the manufacturers' protocols. Library fragments were quantified using Qubit dsDNA High Sensitivity assay (ThermoFisher, USA) and/or the QIAseq Library Quantification kit (QIAGEN, Germany) and their lengths were estimated using Bioanalyzer High Sensitivity DNA Analysis kit (Agilent, USA). Finally, libraries were diluted into equimolar pools and sequenced in five distinct Illumina MiSeq runs with two V2 Nano (300 cycles) and three V3 (600 cycles) cartridges.

Sequencing reads were filtered with fastp v0.20.1 (Chen et al., 2018) , which removed adapters, short reads (< 50 nucleotides) and trimmed low-quality bases (phread < 30) . Reads from each sample were mapped against the SARS-CoV-2 reference genome (NCBI accession: NC_045512.2) with Bowtie2 v2.4.2 (Langmead and Salzberg, 2012) , and mapping files were indexed and sorted with SAMtools v1.12 (Li et al., 2009 ). BCFtools v1.12 was used for variant calling and consensus genome 8 inference, while BEDtools v2.30.0 (Quinlan and Hall, 2010) was used to mask low coverage sites (< 100-fold). Sequences with less than 70% genome coverage were removed from downstream analysis.

Lineage identification was performed with the pangolin tool v3.0.6 (pango v1.2.12; pangoLEARN model from 5 June 2021; constellations v0.0.4 and scorpio v0.3) O'Toole et al., 2021) . To further confirm lineage assignments and contextualize the novel sequences, a representative global dataset of SARS-CoV-2 genome sequences was assembled (n = 3,609). This dataset is drawn from a random sample of a larger dataset (n = 10,838; high-quality data available on GISAID on 13

May 2021) that comprehends all Brazilian sequences, plus one international sequence per country per epidemiological week since the first reported SARS-CoV-2 genome.

The reference dataset was designed to be enriched for sequences from lineages relevant in the Brazilian epidemiological scenario, while also harboring representativeness of SARS-CoV-2 lineages circulating around the world. These sequences were all aligned to the genome sequences herein described with MAFFT v7.475 (Katoh and Standley, 2013 ) and a maximum likelihood tree was inferred with IQ-tree v2.0.3 (Minh et al., 2020 ) under the GTR+F+I+G4 model (Tavaré, 1986; Yang, 1994) .

To further access the temporal dynamics of introduction of VOC Gamma in Rio de Janeiro city, we performed molecular clock analyses on a fully Bayesian framework using BEAST v1.10.4 . To assemble the reference dataset, all

Brazilian VOC Gamma sequences available on GISAID were downloaded (n = 3,398, as of 18 May 2021) and categorized into five discrete locations, matching Brazilian 9 geopolitical regions: Southeast (n = 2,090, except Rio de Janeiro), South (n = 107), Northeast (n = 106), North (n = 138) and Central West (n = 177). The 113 novel Gamma genome sequences from Rio de Janeiro characterized in this study were added to a subset of this dataset. All sequences from locations with less than 113 representatives were included, while the same number of sequences from the remaining locations were randomly sampled, composing a geographically balanced dataset (n = 665). Preliminary maximum likelihood phylogenetic analysis indicated 38 reference sequences of this dataset belonged to the recently identified P.1-like clades (Gräf et al., 2021) and were removed from downstream analysis. The final dataset (n = 627) was composed of 113, 113, 70, 105, 113 and 113 sequences from Rio de Janeiro, Southeast, South, Northeast, North and Central West, respectively. The temporal distribution of sequences per location may be visualized in Supplementary Figure S1 . Root-to-tip regression was used to assess the temporal signal available in the dataset with TempEst v1.5.3 (Rambaut et al., 2016) .

The time scaled phylogeographic reconstruction used: i -the strict molecular clock model; ii -a uniform prior distribution (range: 8 x 10 -4 -10 -3 ) on evolutionary rate; iii -the coalescent exponential growth tree prior (Laplace prior with scale 100) (Griffiths and Tavare, 1994) ; iv -the HKY+I+G4 nucleotide substitution model (Hasegawa et al., 1984; Yang, 1994) ; and v -a discrete phylogeographic model (Lemey et al., 2009) . As Brazilian regions are widely connected by roads and air travel, with no clear asymmetry pattern over short time scales (months), a symmetric phylogeographic model was selected. Eight independent chains of 50 million generations sampling every 10,000 states were performed and convergence (effective sample size > 200 for all parameters) was verified on Tracer v1.7.1 after 10% burnin 10 removal. Logcombiner was used to combine posterior distributions and a maximum clade credibility tree was inferred with TreeAnnotator (Drummond and Rambaut, 2007 ).

An additional set of analyses using the non-parametric coalescent skygrid tree prior (Gill et al., 2013) , with 25 grid points between the date of the most recently sampled sequence (4 May 2021) and the previously estimated date of VOC Gamma emergence (15 November 2020) , was also performed. These grids approximately match the number of epidemiological weeks comprehended in this period. For this analysis, 6 independent runs of 100 million generations, sampling every 10,000 steps, were executed. Maximum clade credibility trees and log files are available in Supplementary File S1.

To evaluate the hypothesis that the transmissibility enhancement of VOC Gamma is due to increments in viral loads Naveca et al., 2021) , we performed a series of analyses based on epidemiological data (age, sex, days of symptoms at diagnosis time and Ct values) of patients infected by viral lineages identified by genome sequencing (n = 244 genomes). Throughout this study, the confidence level considered for hypothesis testing was α = 0.05. First, a linear regression was used to measure the association between viral lineages (VOC Gamma or non-Gamma) and Ct values (N1 target). This model was also adjusted to account for the effects of age, sex and number of symptomatic days at diagnosis time.

We also evaluated the time series of all Ct values (N1 and N2 targets, RNaseP control) characterized in the studied period (n = 1,224), aiming to identify variations of viral loads related to the predominating lineages. To perform a direct comparison, we selected epidemiological weeks for which sequenced samples have shown frequency of 11 VOC Gamma below 20% or above 80%, eliminating intermediate frequency periods.

Lineages were then imputed to samples according to the epidemiological week they have been collected, and a linear model was estimated to measure the effect of lineage on Ct values. These analyses have been performed with the R software (R core team, 2020). Code and data are available in Supplementary File S2. Noticeably, the genetic composition of the SARS-CoV-2 epidemic population in Rio de Janeiro varied in the studied period (Figures 1B and C) . The lineage P.2 (VOI Zeta) was predominant in December 2020 and January 2021, a pattern that changed with the introduction of lineage P.1 (VOC Gamma), first appearing in our dataset in the sixth epidemiological week of 2021. Since then, this lineage increased in frequency and rapidly became the major circulating lineage in the city, reaching 100% of the sampled sequences by the twelfth epidemiological week of 2021, with little variation after that.

These analyses reveal a complex scenario of co-circulation of multiple VOCs or VOIs in Rio de Janeiro and highlight a fast lineage displacement event induced by VOC Gamma.

To further explore the temporal dynamics of introduction and spread of VOC Gamma in Rio de Janeiro city, we performed two Bayesian time scaled phylogeographic reconstructions using different demographic priors. Root-to-tip 14 regression suggests that a modest temporal signal is available in the dataset (R² = 0.27;

Supplementary Figure S2 ). The model based on the coalescent exponential growth prior is initially presented and the one based on the non-parametric coalescent skygrid is described below. Results from both models are summarized in Supplementary Table   S1 .

The first model places the time for the most recent common ancestor (tMRCA) of VOC Gamma on 5 November 2020 (95% highest posterior density interval, HPD: 11

October 2020 -23 November 2020) in the North region of the country (Figure 2A) .

After emergence, the phylogeographic model suggests the variant spread to all Brazilian regions through multiple independent events occurring since late 2020. In Rio de Janeiro, 13 separate introductions that led to the emergence of local clades have been reconstructed, coming from the Northeast (n = 6), Central West (n = 2), South (n = 1) and Southeast (n = 4) regions ( Figure 2B) . A total of 7 introductions represented by single sequences have also been identified (South: 2; Southeast: 5).

The model also suggests that the first introduction came from the Northeast region in early January (95% HPD: 7 -9 January 2021), after an initial dissemination of the VOC Gamma from the northern region. Posterior independent introductions from other regions of the country have been estimated over the following months (earliest 

To explicitly address the hypothesis that the transmissibility enhancement characterized for VOC Gamma is associated with the induction of higher viral loads, we compared the distribution of Ct values (N1 target) measured for patients infected by VOC Gamma or non-Gamma viruses. First, we conducted an exploratory data analysis to evaluate the impact of sex, age and symptomatic days at diagnosis time on Ct values using generalized linear models. While the analysis of age and days of symptoms yielded models with statistical significance, despite moderate effects, no association between patient sex and Ct values could be observed (age: p < 0.01, β = -0.05, 95% CI = ±0.03; days of symptoms: p < 0.01, β = 0.48, 95% CI = ±0.16; sex: p > 0.05). In this sense, multivariate linear models were performed to estimate the effect of viral lineage on Ct values, also adjusting for the individual and combined effects of age and days of symptoms. All models have been summarized in Table 1 (Figure 3C ). This result was further corroborated for the N2 target (p < 0.01, β = -2.43, 95% CI = ±0.75) and no significant statistical association was identified for RNaseP (p = 0.83).

Brazil has been severely impacted by the SARS-CoV-2 pandemic, accounting for 9.5% of the cases and 10.4% of the deaths reported worldwide, even though the country harbors approximately 2.7% of the global population (Castro et al., 2021) .

While the epidemiological dynamics on a national scale has been shown to be affected by a range of complex factors Souza et al., 2020) , the scenario is further aggravated by the emergence and widespread circulation of multiple VOCs and VOIs in the country (Claro et al., 2021; Naveca et al., 2021; Resende et al., 2021a; Resende et al., 2021b; Slavov et al., 2021; Voloch et al., 2021; SES/MA, 2021) . As these lineages harbor a myriad of mutations of biological significance and epidemiological implications (Cele et al., 2021; Davies et al., 2021; Garcia-beltran et al., 2021; Tegally et al., 2021; Volz et al., 2021) , we sought to investigate their circulation dynamics in the city of Rio de Janeiro, the second largest Brazilian metropolis. By combining novel genetic and epidemiological data, we documented a fast lineage replacement event induced by VOC Gamma and provide additional evidence of its altered epidemiological characteristics.

Our study was conducted between early December 2020 and early May 2021, capturing the second and third epidemic waves in the city of Rio de Janeiro (COE/RJ, 2021). Analysis of the 244 novel viral genomes led to the identification of six circulating lineages by phylogenetic inference: P.1, P.2, B.1.1.28, B.1.1.7, B.1.1.33 and   N.9 (Figure 1A) . Temporal analysis of lineage frequencies revealed a shift in the genetic composition of SARS-CoV-2 epidemic population in Rio de Janeiro ( Figures   1B and 1C) , with lineage P.2 (VOI Zeta), most frequent between December 2020 and January 2021, being replaced by lineage P.1 (VOC Gamma), first identified in the sixth epidemiological week of 2021. In fact, VOC Gamma has rapidly risen in frequency, 18 being responsible for over 70% of cases only two weeks later, that is seven weeks after its initial detection, coinciding with the third epidemic wave in Rio de Janeiro city (COE/RJ, 2021) . This result reveals a complex epidemiological scenario, marked by cocirculation of multiple VOCs (Alpha and Gamma) and VOIs (Zeta and potential VOI N.9) and a fast lineage turnover event induced by VOC Gamma, consistent with previous reports on other Brazilian states (Barbosa et al., 2021; Franceschi et al., 2021; Moreira et al., 2021a; Naveca et al., 2021) . Moreover, these results corroborate previous data showing an early detection of VOC Gamma in Rio de Janeiro and provide further context for preliminary genomic analysis performed for the whole state Lamarca et al., 2021) .

Beyond the reported lineage displacement event, the dissemination of VOC Gamma in Rio de Janeiro is also marked by a small interval between autochthonous circulation and an associated epidemic wave. While VOI Zeta emerged in mid-July 2020 and rose in frequency through the second semester, causing an epidemic peak over the last months of that year, our molecular clock analysis indicates that only 10 weeks separate the tMRCA of the earliest Gamma clade from Rio de Janeiro and the epidemic peak of the third wave, mostly driven by this lineage (Figures 2A and 2C) . This observation implies that VOC Gamma not only displayed higher transmissibility than VOI Zeta, but also has been associated with an upsurge in the number of cases, in close agreement with reports from Manaus Naveca et al., 2021) .

Except for B.1.1.28 and B.1.1.33, the predominant lineages in the first epidemic wave in Brazil , all the remaining lineages are VOCs, VOIs or potential VOIs and share the mutation E484K on the RBD of the spike protein, shown to weaken neutralizing antibodies response (Greaney et al., 2021) . This observation highlights the previously reported evolutionary convergence on SARS-CoV-2 genomes (Martin et al., 2021) . Likewise, both VOCs Alpha and Gamma share the mutation N501Y, also implicated in immune escape and enhancement of the affinity between the RBD and human angiotensin converting enzyme 2 (ACE2), the cellular receptor used by SARS-CoV-2 for cell entry (Lan et al., 2020; Ramanathan et al., 2021) . These functional effects, joint to the fast observed rise in frequency and number of cases across diverse epidemiological settings (Barbosa et al., 2021; Franceschi et al., 2021; Moreira et al., 2021a; Naveca et al., 2021 ) -including the one herein reported -provide evidence that the VOC Gamma has a fitness advantage over previously circulating lineages.

These results are consistent with the singular epidemiological characteristics reported for other VOCs in distinct epidemiological scenarios, as Alpha (Volz et al., 2021) and Beta (Tegally et al., 2021) . Interestingly, although both lineages have been shown to circulate in Brazil (Claro et al., 2021; Slavov et al., 2021) , with evidence that VOC Alpha circulates in nearly all regions of the country (Moreira et al., 2021b) , no rise in frequency for the later variant could be observed in our dataset. This result suggests these lineages do not have a fitness advantage in a scenario dominated by VOC Gamma, a hypothesis that must be evaluated across other settings.

Molecular clock analysis performed with the coalescent exponential growth and skygrid tree priors yielded only marginally different results, so the first model will be considered for the purpose of this discussion. The time scaled phylogenetic reconstruction suggests VOC Gamma emerged in the North region between middle October and late November 2020, consistent with previous reports from Manaus Naveca et al., 2021) . The phylogeographic model suggests that after initial dissemination to other regions of the country in early 2021, this lineage was introduced 20 in Rio de Janeiro through multiple events from diverse locations (Figure 2) . While the first estimated introduction came from the Northeast region in early January (95% HPD: 7 -9 January 2021), later events from nearly all other regions in the following months have been estimated. This phylogeographic reconstruction emphasizes the mixture of viruses from diverse regions of the country, evidencing widespread transmission. This pattern is the outcome of the lack of mobility control measures and reinforces the necessity of coordinated responses and surveillance efforts on a national scale. Although the estimated number of introduction events is certainly underestimated, given the limited number of genomes analyzed, this phylogeographic reconstruction likely reveals general patterns of viral spread across the country and, ultimately, the dynamics of VOC Gamma dissemination into Rio de Janeiro.

The molecular clock analysis also implies that it took approximately five weeks for VOC Gamma to be detected by genome sequencing in our randomly selected set of samples and seven weeks to achieve 70% frequency, consistent with previous rates of growth observed in Manaus . The scope of these estimates is limited by the employed sampling effort (0.04% of registered COVID-19 cases in Rio de Janeiro) and is conceivable that increasing sample sizes would have led to an earlier detection of VOC Gamma circulating at initially lower frequencies.

Therefore, increasing genomic sampling and reducing the time from sample collection to sequencing and analysis is needed to enhance the performance of surveillance initiatives. Notwithstanding, our results are consistent with the reported epidemic growth of VOC Gamma in all Brazilian regions.

Although current evidence supports that VOC Gamma in fact has a fitness advantage over other lineages circulating in Brazil, the mechanistic basis of this phenomenon has not been unequivocally established. The main hypothesis in this regard 21 implies this variant is more transmissible due to the induction of infections characterized by higher viral loads Naveca et al., 2021) , as already observed for other VOCs (Tegally et al., 2021; Volz et al., 2021) . Our results agree with this conjecture. No effect of viral lineages on N1 target Ct values distributions could be determined when considering only sequenced samples ( Figure 3A) . However, the total number of samples and possible biases associated with sample selection criteria for genome sequencing (Ct < 30) limit the scope of this analysis. Notwithstanding, analysis of N1 target Ct values time series were suggestive of a consistent decrement coincident with the rise in frequency of VOC Gamma ( Figure 3B ). This result is further corroborated by the analysis of Ct values for the N2 target and the RNaseP control (Supplementary Figure S4) .

These observations were further supported by analysis based on imputed viral lineages through sample collection dates, which shows that samples from individuals likely infected with VOC Gamma viruses have lower Ct values, i.e., higher viral loads ( Figure 3C) . These results are in line with previous estimates Naveca et al., 2021) and suggest that alternative methods of lineage identification (e.g., RT-qPCR) Vogels et al., 2021) may be especially useful in detecting the effect of lineages on Ct values in an unbiased fashion.

It has been suggested that Ct values also vary as a function of epidemiological trajectories (Hay et al., 2021) . To ascertain that this factor has not imposed a significant bias on our analysis, we subsetted epidemiological weeks from periods of decay of the second (40/2020 to 01/2021) and third (11/2021 to 16/2021) epidemic waves in Rio de Janeiro. Samples from the weeks of decay of the second and third waves were imputed as non-Gamma and VOC Gamma, respectively. This analysis yielded the same qualitative results (N1 target; p < 0.01; β = -2.96; 95% CI = ±0.81). The presented 22 findings would benefit from further studies over broader time scales and across different epidemiological settings.

Altogether, this study describes a complex epidemiological dynamics for SARS-CoV-2 in a major metropolis of Brazil, one of the countries most severely hit by the COVID-19 pandemic (Castro et al., 2021) . We document the joint circulation of multiple VOCs and VOIs and report a fast lineage displacement event induced by VOC Gamma, providing further evidence of its altered epidemiological characteristics. In addition, through evolutionary analyses, we describe the timing and origins of VOC Gamma introductions in the city. Finally, we bring forth data and analyses that support the hypothesis that the transmissibility enhancement associated with this lineage is at least partially explained by the induction of higher viral loads. Overall, our results document the evolution of this variant and provide independent assessment of scenarios previously studied in Manaus, therefore contributing to the better understanding of the epidemiological dynamics currently being observed in other Brazilian regions. Statistically significant values highlighted in italic. 

Genomic Surveillance of SARS-CoV-2 in the State of Rio de

Preprint from Virological. Last accessed data on 14

Rapid spread and high impact of the Variant of Concern P.1 in the largest city of Brazil

Evolution and epidemic spread of SARS-CoV-2 in Brazil

Spatiotemporal pattern of COVID-19 spread in Brazil

Fastp: An ultra-fast all-in-one FASTQ preprocessor

NIC team. 2021. Convergent evolution of SARS-CoV-2 spike mutations, L452R, E484Q and P681R

Local Transmission of SARS-CoV-2 Lineage B.1.1.7, Brazil

COVID-19 -Boletim Epidemiológico -Semana 19. Prefeitura do Rio. Last accessed data on 14

Increased mortality in community-tested

Importation and early local transmission of covid-19 in Brazil

BEAST: Bayesian evolutionary analysis by sampling trees

Genomics and epidemiology of the P.1 SARS-CoV-2 lineage in Manaus

Predominance of the SARS-CoV-2 lineage P.1 and its sublineage P.1.2 in patients from the metropolitan region of Porto Alegre

Multiple SARS-CoV-2 variants escape neutralization by vaccineinduced humoral immunity ll Multiple SARS-CoV-2 variants escape neutralization by vaccine-induced humoral immunity

Improving bayesian population dynamics inference: A coalescent-based model for multiple loci

Identification of SARS-CoV-2 P.1-related lineages in Brazil provides new insights about the mechanisms of emergence of Variants of Concern

Complete Mapping of Mutations to the SARS-CoV-2 Spike Receptor-Binding Domain that Escape Antibody Recognition Complete Mapping of Mutations to the SARS-CoV-2 Spike Receptor-Binding Domain that Escape Antibody Recognition

Sampling theory for neutral alleles in a varying environment

A new molecular clock of mitochondrial DNA and the evolution of hominoids

Estimating epidemiologic dynamics from cross-sectional viral load distributions

Air Connectivity: Measuring the connections that drive economic growth. Last accessed data on 15

MAFFT multiple sequence alignment software version 7: Improvements in performance and usability

Genomic surveillance of SARS-CoV-2 tracks early interstate transmission of P.1 lineage and diversification within P.2 clade in Brazil

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor

Fast gapped-read alignment with Bowtie 2

Bayesian Phylogeography Finds Its Roots 5

The Sequence Alignment/Map format and SAMtools

The emergence and ongoing convergent evolution of the N501Y lineages coincides with a major global shift in the SARS-CoV-2 selective landscape

IQ-TREE 2: New Models and Efficient Methods for Phylogenetic Inference in the Genomic Era

Increasing frequency of SARS-CoV-2 lineages B.1.1.7, P.1 and P.2 and identification of a novel lineage harboring E484Q and N501T spike mutations in Minas Gerais, Southeast Brazil. Preprint from Virological. Last accessed data on 14

Epidemic Spread of SARS-CoV-2 Lineage B.1.1.7 in Brazil

was driven by the persistence of endemic lineages and P.1 emergence

Assignment of epidemiological lineages in an emerging pandemic using the pangolin tool

Multiplex PCR method for MinION and Illumina sequencing of Zika and other virus genomes directly from clinical samples

BEDTools: A flexible suite of utilities for comparing genomic features

R: A language and environment for statistical computing. R Foundation for Statistical Computing

SARS-CoV-2 B.1.1.7 and B.1.351 spike variants bind human ACE2 with increased affinity

Exploring the temporal structure of heterochronous sequences using TempEst

Posterior summarization in Bayesian phylogenetics using Tracer 1.7

A dynamic nomenclature proposal for SARS-CoV-2 lineages to assist genomic epidemiology

A potential sars-cov-2 variant of interest (Voi) harboring mutation e484k in the spike protein was identified within lineage b.1.1.33 circulating in Brazil

Identification of a new B.1.1.33 SARS-CoV-2 Variant of Interest (VOI) circulating in Brazil with mutation E484K and multiple deletions in the amino (N)-terminal domain of the Spike protein

and-multiple-deletions-in-the-amino-n-terminal-domain-of-the-spike-protein/675

Dataset on SARS-CoV-2 non-pharmaceutical interventions in Brazilian municipalities

Secretaria Estadual de Saúde do estado do Maranhão, Brasil. Last accessed data on 20

Epidemiological and clinical characteristics of the COVID-19 epidemic in Brazil

Bayesian phylogenetic and phylodynamic data integration using BEAST 1.10. Virus Evol

Some probabilistic and statistical problems in the analysis of DNA sequences

Emergence and rapid spread of a new severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) lineage with multiple spike mutations in South Africa

Detection of a SARS-CoV-2 variant of concern in South Africa

Multiplex qPCR discriminates variants of concern to enhance global surveillance of SARS-CoV-2

Genomic Characterization of a Novel SARS-CoV-2 Lineage from Rio de Janeiro, Brazil

Assessing transmissibility of SARS-CoV-2 lineage B.1.1.7 in England

WHO Director-General's opening remarks at the media briefing on COVID19. Last accessed data on 14

A new coronavirus associated with human respiratory disease in China

Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: Approximate methods

Associated metadata and data used in the Ct time series analysis are available in Supplementary Files S3 and S2, respectively. Competing interests The authors declare no competing interests. Acknowledgements: We would like to thank all authors who

A complete list of acknowledgements is available on Supplementary File S5. We would also like to thank the entire technical and administrative staff from Laboratório de Virologia Molecular for their invaluable support, making this and other studies by our group feasible. We also thank the support of the sequencing infrastructure of the Laboratório de

Rafael Estima and Druid for their help on handling computational resources that made our analyses possible. Finally, we would also like to thank the Instituto Nacional do Cancer staff for the technical support on part of our sequencing runs

Members of the COVID-19

We also acknowledge support from the Rede Coronaômica BR MCTI/FINEP affiliated to RedeVírus/MCTI