2. Molecular Aspects and Microevolution of the SARS-CoV-2

According to the NCBI reference sequence (NC_045512.2), the SARS-CoV-2 RNA genome contains nearly 29,903 base pairs (bp) [Reference Wu, Zhao and Yu19]. It is composed of 8 open reading frames (ORFs) and 4 structural genes: (1) the spike glycoprotein (S), typical of the Coronaviridae family, (2) the envelope protein (E), (3) the membrane protein (M), and (4) the nucleocapsid protein (N) distributed along the RNA genome and interlaced with the ORFs. Additionally, ORF1ab (21,555 nt) occupies 72% of the whole genome and encodes 16 nonstructural proteins (nsp) involved in proteolysis, replication, and adaptation to a new host [Reference Naqvi, Fatima and Mohammad20]. The remaining ORFs (3a, 3b, 6, 7a, 7b, 8, 10) are accessory proteins involved in virion assembly but with functions still under investigation [Reference Rahimi, Mirzazadeh and Tavakolpour21].

There are two main concerns about the source of variation in the SARS-CoV-2 genomes. First, we must understand that the more the population is infected, the higher the chances of finding a mutation are. As there are many copies of the virus, there is a proportional probability of a mutation occurring that would increase its virulence. This can be evidenced by the appearance of new variants in the middle of waves of infections in Europe, South Africa, the United States, Brazil, and now India [Reference Makoni22–Reference Voloch, da Silva Francisco and de Almeida24]. The second aspect of concern is related to the natural process of emerging mutations. All DNA or RNA organisms are susceptible to nucleotide changes (substitutions, insertions, deletions, etc.) that could produce phenotypical changes [Reference Rehman, Shafique, Ihsan and Liu25]. Most of the changes are neutral (synonym mutations) and do not affect the phenotype. Some others might damage the viral structure or the capacity of the virus to reinfect, but natural selection gets rid of them. In contrast, the last few mutations are the ones that confer an adaptive response and make them dominant by the phenotype associated with them (i.e., more infectious) [Reference Volz, O’Toole and Bull26, Reference Becerra-Flores and Cardozo27]. For example, gaining affinity for the human angiotensin-converting enzyme 2 (ACE2) receptor opens the door for SARS-CoV-2 in human cells [Reference Kai and Kai28]. Humans are not immune to this source of variation; although the rate of change is much slower, some changes in the ACE2 gene translate into ACE2 protein polymorphism, which is now evidence of susceptibility to SARS-CoV-2 infection and COVID-19 severity [Reference Ragia and Manolopoulos29–Reference Dong, Zhang and Ma32].

In the year and a half since the virus has been spreading around the world, there have been reports and evidence that new variants are appearing around the world. According to the WHO and the US CDC (Centers for Disease Control), variations can be divided into three categories: (1) VOI (variants of interest), (2) VOC (variants of concern), and (3) VOHC (variants of high consequence) [33]. VOI are the ones with specific genetic markers that have been associated with changes in virulence. VOCs are variants for which there is experimental evidence of an increase in virulence (pathogenicity) and/or vaccine/treatment resistance. VOHC are the ones that pose a serious health threat, making available treatments useless [33]. Finally, the variants recently discovered with significant spreading capacities and with a potential menace are called variants under investigation (VUI). All of these variants are the result of evolutionary force imbalances that ponder how to increase diversity.

Genetic drift and selection pressures will limit the diversity of SARS-CoV-2 by limiting the virus’s spread or survival due to random events or biosafety protocols. In contrast, the source of variation in any virus comes from mutation and recombination occurring in every event of infection and replication [Reference Rehman, Shafique, Ihsan and Liu25, Reference Domingo34].

Cumulative mutations in SARS-CoV-2 are directly correlated with five pandemic alarms: (1) an increase in disease severity and mortality risk; (2) the speed of transmission from human to human; (3) the effectiveness of available vaccines; (4) the effectiveness of diagnostic tests; (5) susceptibility to treatments [Reference Yuen, Ye, Fung, Chan and Jin12, Reference Choudhary, Sreenivasulu, Mitra, Misra and Sharma35–Reference Muñoz37]. All these trepidations are examined under genomic surveillance, which consists of deep analysis to determine how sequence changes reflect variations in the phenotype that modify the mechanism to detect and neutralize the infection.

3. Nomenclature of SARS-CoV-2

Due to the fast spread of diversity observed in SARS-CoV-2, the need to name the lineages has emerged. This is evident in the nomenclature used by the main databases or organizations in charge of surveillance, like WHO, GISAID, and Nextstrain. Initially, variants were named according to the place where they were first reported (i.e., the UK, South Africa, Brazil, or India). A dynamic method was later proposed based on the use of a phylogenetic framework to track lineages that contribute the most to an active spread [Reference Rambaut, Holmes and O’Toole38]. The two main databases for SARS-CoV-2 genomes and genomic epidemiology, GISAID and Nextstrain, respectively, settled on their own nomenclature [Reference Elbe and Buckland-Merrett18, Reference Hadfield, Megill and Bell39]. To make it easier and more practical to be discussed by nonscientific audiences, the World Health Organization recommended on May 30, 2021, using letters of the Greek Alphabet to denote variants of concern and interest [40]. Table 1 compiles lineages, the place where they were first identified, and important characteristics of variants. Until June 21, 2021, the group of VOCs was comprised of five variants: Alpha (UK B.1.1.7), Beta (South Africa B.1.351), Gamma (Brazil P.1), Delta (India B.1.617.2), and Omicron (South Africa B.1.1.529). Also, there are two VOIs under research: Lambda (Peru C.37) and Mu (Colombia B.1.621). Until now, no VOHC has been reported.

Table 1 Nomenclature of SARS-CoV-2 variants and noticeable pathological characteristics.

UI: under investigation. Our own source. Data from OMS, CDC, ECDC, and Outbreak.info.

When comparing the variants, it is noticeable that a large number of all mutations are located in the spike (S) protein. Being the S protein of the coronavirus, the main determinant of host and tissue tropism, it is also the target of vaccines, neutralizing antibodies and inhibitors of virus entry [Reference Pomplun41, Reference Hu, He and Gao42]. Furthermore, mRNA vaccines currently in use worldwide were designed to transcribe fragments of this protein. The S gene, which consists of 3,821 nucleotides coding for a protein of 1,273 amino acids, forms a trimeric spike (subunits S1, S2, and S2′) on the virion surface and plays an essential role in viral entry [Reference Plante, Liu and Liu43]. Coronaviruses may use two different pathways to enter the host cell: (1) the protease-mediated cell surface pathway and (2) the endosomal pathway [Reference Hu, He and Gao42, Reference Zhou, Vedantham and Lu44]. According to Hu and colleagues, the spike proteins of several coronaviruses are cleaved by host proteases in the S1 subunit for receptor binding and the S2 subunit for membrane fusion in the entry step of infection [Reference Ragia and Manolopoulos29, Reference Hu, He and Gao42]. Similar to SARS-CoV, the cellular receptor for SARS-CoV-2 is angiotensin-converting enzyme 2 (ACE2). However, the SARS-CoV-2 S protein has up to a 20-fold higher affinity for ACE2 than its counterpart, the SARS-CoV spike protein [Reference Kai and Kai28].

A recent study showed that a new serine protease (elastase-2) was introduced when a glycine replaced aspartic acid at position 614 of the spike protein [Reference Bhattacharyya, Das and Ghosh45]. Experimental studies have shown that it enhances viral replication in human lung epithelial cells and primary human airway tissues by increasing the infectivity and stability of virions. This is currently the dominant mutation in all VOC and VOI [Reference Volz, O’Toole and Bull26, Reference Plante, Liu and Liu43, Reference Mullen, Tsueng and Abdel Latif46]. It is still uncertain whether the D614G (Asp614Gly) mutation affects the antigenic properties of protein S, although there is the possibility of positive natural selection. With the gain of transmissibility and the absence of preexisting immunity in the general population, the chances for this mutation to disappear seem remote, and it is not known whether SARS-CoV-2 is fully adapted for efficient growth in human cells [Reference Giovanetti, Benedetti and Campisi10]. Supported by robust sequencing and correlation with abundance, with reference to all sequences obtained at specific periods of time, one can determine if a new mutation is likely to become dominant.

4. The SARS-CoV-2 Mutation Rate

Because of the chemistry of the molecule, single-stranded RNA viruses have, on average, a higher mutation rate than DNA viruses. However, large coronavirus genomes such as SARS-CoV-2 are relatively stable thanks to a proofreading mechanism that operates during replication (nsp14 exonuclease) [Reference Giovanetti, Benedetti and Campisi10, Reference van Dorp, Acman and Richard47]. Still, many mutations occur, generating stable variants like de D614G (Asp614Gly) that become dominant in about 6 months. Studies have shown that mutations occur more frequently in some regions of the SARS-CoV-2 genome [Reference Rahimi, Mirzazadeh and Tavakolpour21, Reference Roy, Mandal and Mondal48]. When measuring point mutations in the whole viral genome, Roy and colleagues determined the mutation frequency in μ = 9.4 × 10⁻⁶, which corresponds to 20,163 polymorphisms detected/(29,903 nucleotide genome size × 71,703 sequences analyzed in the dataset). When measuring only the nsp region (ORF1ab), the rate was 8.78 × 10⁻⁶. Within this region, the genes nsp1 and nsp2 have the highest rates of mutation at 1.12 × 10⁻⁵ and 1.08 × 10⁻⁵, respectively. These two are leader genes to inhibit immune response: nsp1 inhibits protein translation by blocking 40S ribosome and interferon (IFN) signaling, while nsp2 inhibits prohibitins 1 and 2 to disrupt the cellular environment [Reference Yoshimoto49]. Mutations in these two genes might confer novel viral outcomes to evade the host’s immunogenic response.

Compared to other studies, the mutation rate in the SARS-CoV genome of family members was estimated to be 0.80–2.38 × 10⁻³ nucleotide substitutions per site/year, which is in the same order of magnitude as other RNA viruses when a time scale is involved [Reference Zhao, Li and Wu50]. For SARS-CoV-2, the moderate accumulation of changes observed in a year was approximately 6 × 10⁻⁴ nucleotides/genome/year [Reference van Dorp, Acman and Richard47]. Compared to those high rates, accessory genes ORF7a and ORF3a are 1.37 × 10⁻⁵ and 1.35 × 10⁻⁵ nucleotides/genome/year, respectively. ORF7a, an accessory protein, is thought to be involved in viral assembly or budding events specific to SARS-CoVs [Reference Yoshimoto49]. The accumulation of nonsynonymous mutations in this gene may provide new molecular options for increasing virulence efficiency [Reference Roy, Mandal and Mondal48]. On the other hand, ORF3a has been reported to have proapoptotic activity through mitochondrial damage and activating inflammatory responses of host cells [Reference Yue, Nabar and Shi51]. The high rate of mutations in this gene could be interpreted as a devious strategy of the virus to finish its life cycle and kill the host cell. Genomic data can show how viral pathogens have responded to different forces of natural selection. In a model of codons, natural selection acting over any locus can be estimated using the proportion of nonsynonymous (dN) and synonymous (dS) mutations (dN/dS). Values of dN/dS > 1 can be interpreted as a positive natural selection because diversity arises. In contrast, negative selection or selective removal of the alleles that are deleterious results in dN/dS < 1 [Reference Roy, Mandal and Mondal48, Reference Kryazhimskiy and Plotkin52].

When comparing genes within the SARS-CoV-2 genome, recent studies have found that all nsp genes (except nsp11), S (spike), and M (membrane) are under negative (purifying) selection (dN/dS < 1) [Reference Roy, Mandal and Mondal48]. This is reasonable because these are the genes (nsps, S and M) in which the host immune response is active. On the other hand, accessory proteins ORFs 3a, 6, 7a, 8, and 10, structural proteins E (envelope), and N (nucleocapsid) genes are under positive selection (dN/dS > 1), having different forms, arise in the possible new variants.

5. Mutations Related to VOC and VOI

To understand what is happening with the distribution of variants in South America, it is necessary to first understand what set of mutations is linked to each variant. Based on the SARS-CoV-2 genome data from GISAID, the platform Outbreak.info, updated daily, displays graphic information that correlates the presence of certain detected mutations with VOI and VOCs [Reference Mullen, Tsueng and Abdel Latif46].

Similar to variant classification, mutations are also considered: mutations under observation (MUO), mutations of interest (MOI), and mutations of concern (MOC), sorted according to experimental observations in phenotypic changes related to virulence (pathogenicity) in SARS-CoV-2. We already mentioned D614G (Asp614Gly), a mutation now surviving in all VOI, VOC, and VUO. The current mutation of concern is S: E484K (Glu484Lys). A mutation of G > A in position 23,012 that changes a glutamic acid (E) for Lysine (K). This mutation is present in VOC Beta (B.1.351), Gamma (P.1), and Mu (B.1.621). This mutation increased dominance in Brazil [Reference Ferrareze, Franceschi, Mayer, Caldana, Zimerman and Thompson53] and recent studies have shown reduced neutralization by immune reactions, also called “escape mutations” [Reference Jangra54, Reference Wise55], and higher infectivity when in the presence of other mutations [Reference Khan, Zia and Suleman56].

The following mutations are all of interest (MOI), but their impact resides in the synergic functionality with other mutations. K417N (Lys417Asn) and K417T (Lys417Thr) mutations are G > T substitutions at position 22,811 and A > C substitutions at position 22,810, respectively, that change a Lysine (K) for Asparagine (N) or Threonine (T) and have been found in Beta, Gamma, and Omicron variants, and appear to have higher rates of infectivity [Reference Khan, Zia and Suleman56]. Another mutation under the radar is N501Y (Asn501Tyr). This causes an A to become a T at position 23,063, resulting in asparagine (N) for tyrosine (Y). This mutation is present in Alpha, Beta, Gamma, Omicron, and Mu. This last variant was first identified in Colombia and rapidly spread in the northwest region of South America. It is involved in higher infectivity [Reference Khan, Zia and Suleman56] and, mechanistically, the N501Y (Asn501Tyr) substitution improved the affinity of the viral spike protein for cellular receptors, and researchers have suggested this mutation be classified as a MOC [Reference Liu, Liu and Plante57]. To easily visualize MOCs and MOIs with each VOC and VOI, Figure 1 shows the variants currently active in South America and their respective mutations and frequency. Additionally, Table 2 shows the frequency of each variant in every continental country in South America.

Figure 1 Cross reference of variants of concern (VOC) and (VOI) variants of interests (Y-axis) and the occurrence and frequency of point mutations (X axis). Mutations and frequency were measured with >90% prevalence of that specific mutation in at least one lineage. Text in red represent variants/mutations of concern. Yellow: variants/mutations of interest. White text: variants/mutations under surveillance or observation. Graphic modified from outbreak.info website [Reference Mullen, Tsueng and Abdel Latif46] and customized to show only the VOC and VOI currently spreading in South America. By the time of manuscript writing, a few cases of Omicron were present in South America but reached over 10% in South Africa and over 3% in the fourth wave of infections in Europe in the last month. Source: outbreak.info/GISAID.

Table 2 Distribution and frequencies of VOC and VOI in continental South America.

Clades as headers. α = alpha, β = beta, γ = gamma, δ = delta, λ = Lambda, μ = Mu. wt+ = “wild-types”/original variants. N: number of sequences curated in GISAID (29 Nov 2021). Analysis on November 2021, UK: United Kingdom. SA: South Africa, Br: Brazil, In: India, US: United States, Pe: Peru, and Col: Colombia. At the time of manuscript writing, isolated cases of Omicron were reported in Brazil. Countries with ^* data not available from the last three months. Our own source. Data from: Nextstrain.org.

6. Variant Distribution in South America

To have a broader perspective of what is happening in South America, we can examine the distribution and frequencies of the variants currently identified. Table 2 shows the SARS-CoV-2 variants (VOC and VOI) and frequencies in South American countries together with the amount (number and percentage) of sequences submitted to GISAID by the end of September of 2021. GISAID data shows Delta as the largest widespread, with an almost 90% frequency of all recently sequenced variants [Reference Elbe and Buckland-Merrett18]. Delta increased rapidly from 1% in early June to 19% in late July to 90% in early November. This variant is known for the mutations in the Spike gene: T19R (Thr16Arg), T95I (Thr95Ile), L452R (Leu452Arg), T478K (Thr478Lys), D614G (Asp614Gly), P681R (Pro681Arg), and D950N (Asp950Asn). Similar to variant Mu, they share the mutations T95I (Thr95Ile) and D950N (Asp950Asn) that seem to correspond to the significant increase in infectivity. No data on the severity of the disease, mortality, or vaccine resistance have been linked to these latter mutations yet.

In second place is the Gamma variant, with an almost insignificant 5% compared to the 85% it had in early July. It was first identified on January 6, 2021, in Tokyo, Japan, by travelers from the Brazilian Amazonia [Reference Faria, Claro, Candido, Moyses Franco, Andrade and Coletti58]. It is currently in every continental country in South America. It contains 14 mutations, 10 of which are in the S gene and 5 of those 10 are the mutations: L18F (Leu18Phe), K417T (Lys417Thr), E484K (Glu484Lys), N501Y (Asn501Tyr), and D614G (Asp614Gly). It had reached 96 countries by the end of June 2021. Recent studies have shown a significant reduction in the neutralization of Pfizer and Moderna vaccines in fully dosed people [Reference Garcia-Beltran, Lam and Denis59], and the efficacy of Sinovac vaccines is close to 50%, but exact measurements are still being conducted in studies [Reference Moutinho60].

The third most frequent variant is shared by Lambda (C.37, Nextstrain: 21G) and Mu (B.1.621, Nextstrain: 21H), which are about 3 to 5% frequent each. Lambda (C.37), also known in Latin America as “Variante Andina,” was originally reported in Peru in August 2020 and classified by WHO as VOI on June 14, 2021. The Lambda variant contains six nonsynonymous mutations: G75V (Gly75Val), T76I (Thr76Ile), L452Q (Leu452Gln), F490S (Phe490Ser), D614G (asp614Gly), and T859N (Thr859Asn) in the S gene, plus a novel deletion (Δ246–252) [Reference Robertson61]. Interestingly, by early May of 2021, the Lambda variant (C.37) was close to 93% of the frequency in Peru and reached over 30 countries, including Australia [Reference van Homrigh62]. It is still profuse in 1 out of every 5 cases in Argentina, Bolivia, and Chile. It is not clear if the Lambda variant has implications for vaccine resistance or implications for disease severity, as was speculated in the news.

The other 3% is Mu. It is an outlier of distribution because the variant was found mostly in Colombia, where it counts for 95% of the newly reported cases in June and July 2021 [Reference Laiton-Donato, Franco-Muñoz and Álvarez-Díaz63]; 3% in other South American countries; 2% in the Caribbean islands and the rest of the world where it spread. Inside Colombia, it is still near 9% but has been outspread by the Delta variant. In May 2021, the Mu variant represented 95% of Colombian sequences, and by early August, there was still over 82% [Reference Laiton-Donato, Villabona-Arenas and Usme-Ciro17, Reference Laiton-Donato, Franco-Muñoz and Álvarez-Díaz63]. No evidence of the severity of the disease or symptoms has been reported yet.

The thriving capacity of SARS-CoV-2 to mutate has resulted in, in some cases, convergent evolution and a boost of mutations that are not phylogenetically related. This is, the same mutations are emerging in different places [Reference van Dorp, Acman and Richard47]. As with D950N (Asp950Asn), the mutation G > A at position 24,410 converts an aspartic acid (D) to an asparagine (N). It surfaced almost simultaneously in opposite places of the world: the Delta variant (B.1.617.2) in India and the Mu variant (B.1.621) in Colombia. Studies have shown that in the S gene alone, there is one location with at least 15 recurrent mutations, suggesting convergent evolution and a particular interest in the context of adaptation of the virus to the human host [Reference van Dorp, Acman and Richard47].

Although these data compilations reveal high mutation rates and pressures of selection in favor of the arising diversity of SARS-CoV-2, it is important to mention that these data must be interpreted carefully because some statistical metrics might technically be biased in different ways. On the other hand, the majority of sequences in South America come from just a few research centers or government-funded laboratories in which the collected samples are typically from main urban areas and might not represent the full diversity, especially from regions with difficult access or those near to frontiers with other countries. Secondly, the sequences are being obtained from different platforms. It is well known that PacBio, ION Torrent, and Oxford NanoPore technologies offer longer reads at a quality cost, making them difficult to compare with Illumina or Sanger methods. This highly error-prone platform might show an inaccurate reading of nucleotides as possible point mutations if replicates and controls are not rigorous enough. Third, countries in South America have the lowest speed for updating genomic information in GISAID. By the time this article was written, there was no information about the last three months from Bolivia, Uruguay, Paraguay, and Venezuela.

As a fourth aspect, most of the statistics to measure transmissibility are based on confirmed positive cases; however, only in Germany, studies showed that almost 42% of people are unaware of their current status of infection [Reference Wild64]. An unmeasurable number of people in South America are infected but not officially diagnosed, therefore making the fatality of SARS-CoV-2 much lower and the transmissibility much higher [Reference Irons and Raftery65].

And finally, these molecular dynamics of mutation and variant frequencies are time-sensitive and a time interval as low as 30 days is enough to significantly change the epidemiological landscape. According to GISAID, South American countries show a median of 92 days (equal to 3 months with a range of 27 to 271 days) for the deposition of sequences [Reference Elbe and Buckland-Merrett18]. Compared to the 16 days of developed countries in Europe, South American countries are taking too much time to reach the correct epidemiological and appropriate public measures based on evidence to control the pandemic, thus resulting in massive waves of infection and new variants appearing.