Emerging Microbes & Infections ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/temi20 Naturally occurring SARS-CoV-2 gene deletions close to the spike S1/S2 cleavage site in the viral quasispecies of COVID19 patients Cristina Andrés, Damir Garcia-Cehic, Josep Gregori, Maria Piñana, Francisco Rodriguez-Frias, Mercedes Guerrero-Murillo, Juliana Esperalba, Ariadna Rando, Lidia Goterris, Maria Gema Codina, Susanna Quer, Maria Carmen Martín, Magda Campins, Ricard Ferrer, Benito Almirante, Juan Ignacio Esteban, Tomás Pumarola, Andrés Antón & Josep Quer To cite this article: Cristina Andrés, Damir Garcia-Cehic, Josep Gregori, Maria Piñana, Francisco Rodriguez-Frias, Mercedes Guerrero-Murillo, Juliana Esperalba, Ariadna Rando, Lidia Goterris, Maria Gema Codina, Susanna Quer, Maria Carmen Martín, Magda Campins, Ricard Ferrer, Benito Almirante, Juan Ignacio Esteban, Tomás Pumarola, Andrés Antón & Josep Quer (2020) Naturally occurring SARS-CoV-2 gene deletions close to the spike S1/S2 cleavage site in the viral quasispecies of COVID19 patients, Emerging Microbes & Infections, 9:1, 1900-1911, DOI: 10.1080/22221751.2020.1806735 To link to this article: https://doi.org/10.1080/22221751.2020.1806735 © 2020 The Author(s). Published by Informa Published online: 02 Sep 2020. UK Limited, trading as Taylor & Francis Group, on behalf of Shanghai Shangyixun Cultural Communication Co., Ltd Submit your article to this journal Article views: 5823 View related articles View Crossmark data Citing articles: 24 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=temi20 Emerging Microbes & Infections 2020, VOL. 9 https://doi.org/10.1080/22221751.2020.1806735 Naturally occurring SARS-CoV-2 gene deletions close to the spike S1/S2 cleavage site in the viral quasispecies of COVID19 patients Cristina Andrésa*, Damir Garcia-Cehicb,c*, Josep Gregori c,d, Maria Piñanaa, Francisco Rodriguez-Friasc,e,f, Mercedes Guerrero-Murillob, Juliana Esperalbaa, Ariadna Randog, Lidia Goterris g, Maria Gema Codina g, Susanna Querb, Maria Carmen Martína, Magda Campins h, Ricard Ferreri, Benito Almirantej, Juan Ignacio Estebanb,c,f, Tomás Pumarola f,g, Andrés Antón a,f and Josep Quer b,c a Respiratory Viruses Unit, Microbiology Department, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Barcelona, Spain; bLiver Unit, Liver Diseases - Viral Hepatitis, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Barcelona, Spain; cCentro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Instituto de Salud Carlos III, Madrid, Spain; dRoche Diagnostics SL, Sant Cugat del Valles, Barcelona, Spain; eBiochemistry and Microbiology Departments, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Barcelona, Spain; fUniversitat Autònoma de Barcelona, Bellaterra, Spain; g Microbiology Department, Vall d’Hebron Hospital Universitari, Barcelona, Spain; hPreventive Medicine, Hospital Vall d’Hebron Hospital Universitari, Barcelona, Spain; iIntensive Care Department. Shock, Disfunció Orgànica i Ressuscitació (SODIR) Research Group, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Barcelona, Spain; jInfectious Diseases Department. Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Barcelona, Spain ABSTRACT The SARS-CoV-2 spike (S) protein, the viral mediator for binding and entry into the host cell, has sparked great interest as a target for vaccine development and treatments with neutralizing antibodies. Initial data suggest that the virus has low mutation rates, but its large genome could facilitate recombination, insertions, and deletions, as has been described in other coronaviruses. Here, we deep-sequenced the complete SARS-CoV-2 S gene from 18 patients (10 with mild and 8 with severe COVID-19), and found that the virus accumulates deletions upstream and very close to the S1/S2 cleavage site (PRRAR/S), generating a frameshift with appearance of a stop codon. These deletions were found in a small percentage of the viral quasispecies (2.2%) in samples from all the mild and only half the severe COVID-19 patients. Our results suggest that the virus may generate free S1 protein released to the circulation. We suggest that natural selection has favoured a “Don’t burn down the house” strategy, in which free S1 protein may compete with viral particles for the ACE2 receptor, thus reducing the severity of the infection and tissue damage without losing transmission capability. ARTICLE HISTORY Received 23 June 2020; Revised 27 July 2020; Accepted 2 August 2020 KEYWORDS SARS-CoV-2; deletions; quasispecies; NGS; respiratory virus; diversity Introduction size of the CoV RNA genome increases the probability RNA viruses replicate using their own RNA-dependent that deletions will be generated and recombination RNA polymerase (RdRp), which lacks proofreading events will take place, which could facilitate adaptation mechanisms and is prone to mutate at high rates to new host environments, as occurs with jumping (10−3–10−5 substitutions/nucleotide/replication cycle), between species[1,2]. One naturally occurring deletion lending the virus a quasispecies structure [1,2]. Pre- on 29 nucleotides in the open reading frame (ORF) 8 of vious studies with severe acute respiratory syndrome SARS-CoV after human-to-human transmission was coronavirus (SARS-CoV) and mouse hepatitis virus found to be associated with attenuation of replication have reported moderate mutation rates of 9.06 × 10−7 [4]. and 2.5 × 10−6 subs/site/cycle respectively, below the The low mutation rate, high human-to-human expected range for RNA viruses [3]. This is consistent transmissibility (R0 = 2.2) [5], and absence of human with a role for non-structural protein (nsp) 14 in RNA pre-existing immunity against SARS-CoV-2 could proofreading or repair functions because of its 3’-5’ explain its rapid spread through the human popu- exonuclease (ExoN) activity. Nonetheless, the large lation, with very high sequence identity (99.9%) CONTACT Andrés Antón aanton@vhebron.net Respiratory Viruses Unit, Microbiology Department, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Passeig Vall d’Hebron 119-129, 08035 Barcelona, Spain Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain; Josep Quer josep.quer@vhir.org Liver Unit, Liver Diseases - Viral Hepatitis, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Spain Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Instituto de Salud Carlos III, Madrid, Spain *These two authors have contributed equally to the manuscript. Supplemental data for this article can be accessed at https://doi.org/10.1080/22221751.2020.1806735 © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of Shanghai Shangyixun Cultural Communication Co., Ltd This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. EMERGING MICROBES AND INFECTIONS 1901 between isolates recovered all over the world (sequence Patients and methods published in the repository sequence data banks, GISAID Patients and GenBank). The high pathogenicity of the virus, the severity of coronavirus disease-19 (COVID-19) and the Upper respiratory tract specimens (naso/oropharyngeal lack of an effective antiviral treatment or vaccine has swabs or nasopharyngeal aspirates) from individuals pushed the scientific community worldwide to develop, consulting in the emergency room were collected for in record time, a solution for this pandemic [6]. SARS-CoV-2 testing in the Department of Microbiology Among the SARS-CoV-2 structural proteins, at Hospital Universitari Vall d’Hebron (HUVH), Barce- including spike (S), envelope (E), and membrane (M) lona (Spain). Samples from 18 patients with no previous constituting the viral coat, and the nucleocapsid (N) comorbidities other than COVID-19 were included in protein that packages the viral genome, the S glyco- the study. As defined by CDC criteria (https://www. protein is the most promising as a therapeutic and vac- cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance- cine target. The S protein is encoded by the S gene, and management-patients.html), 10 patients had a mild following trimerization, it composes the spikes of the clinical presentation of COVID-19 (absence of viral characteristic viral particle crown (corona). The S pneumonia and hypoxia, no hospitalization require- protein is essential for SARS-CoV-2 to infect a host ment, able to manage their illness at home), whereas 8 cell [7] by recognizing and binding to the human cell patients had severe disease (intensive care unit (ICU) receptor, angiotensin-converting enzyme 2 (ACE2) admission for supportive management of complications [8], and possibly (with lower affinity) to other recep- of severe COVID-19 such as pneumonia, pneumonia, tors, such as CD209L (L-SIGN), also used by SARS- hypoxemic respiratory failure, sepsis, cardiomyopathy CoV [9] and dipeptidyl peptidase 4 (DPP4), used by and arrhythmia, acute kidney injury, and other compli- MERS [10]. cations). All patients were attended by March 2020, and The S gene has 3822 nucleotides with 1273 amino those with both mild and severe disease had a favourable acids (GenBank reference sequence MN908947.3). It outcome with resolution of the infection. has five essential domains: the receptor-binding domain (RBD), O-linked glycan residues flanking a polybasic S1/S2 cleavage site, fusion peptide (FP), Methods heptad repeats HR1 and HR2, and a transmembrane Detection of SARS-CoV-2 domain (TM). The S1 RBD includes 6 amino acid positions that show high affinity for the human The diagnosis of COVID-19 was performed by two ACE2 receptor, which is widely distributed, but tests, an in-house RT-PCR assay using the primer/ mainly present in alveolar type 2 (AT2) cells of the probe set from the CDC 2019-nCoV Real-Time RT– lungs [11]. Once the virus is attached to the host PCR Diagnostic Panel (Qiagen, Hilden, Germany) cell receptor, cleavage occurs between subunits S1 and a commercial real-time RT-PCR assay (Allplex and S2, and subunit S2 drives the viral and cellular 2019-nCoV Assay, Seegene, South Korea). membranes to fuse [12]. Thus, S1 recognizes and binds to the human cell receptor ACE2, whereas S2 SARS-CoV-2 sequencing directly facilitates entry into the host cell. Both func- tions are crucial for infection, and therein lies the The 18 respiratory specimens were inactivated by mix- interest of S as a target for the development of vac- ing 140 µL of the sample with 560 µL of AVL buffer cines and antiviral agents. (Qiagen, Hilden, Germany). Extraction of nucleic Because of the importance of the S protein, we car- acids was then performed using the QIAmp Viral ried out a deep-sequencing study of the S gene in upper RNA Mini Kit (Qiagen, Hilden, Germany) following respiratory tract samples from 18 patients with mild or the manufacturers’ instructions but without the RNA severe SARS-CoV-2 disease. Of particular note, hot- carrier, obtaining a final elution of 30 µL. spot deletion sites were found in minority mutants The complete S gene was amplified using a double located upstream and very close to the S1/S2 PCR. The first RT–PCR step consisted in amplifying (PRRAR/S) and S2’ (KPSKR/SFI) cleavage sites, 2 large fragments, 3314 base pairs (bp) and 3591 bp suggesting that these genomes code for a truncated S in length, respectively. The 5’ end of primer 1 and 3’ protein. The variants were significantly more prevalent end of primer 2 were designed to be outside the S in patients with mild than those with severe disease. region to ensure that we were amplifying SARS-CoV- Thus, their effect on the protein could constitute a 2 genomic RNA, and not subgenomic RNA (Table favourable regulatory mechanism emerging in the S11). viral quasispecies to modulate the pathological effect The SuperScript III One-Step RT–PCR System with of the infection. Discussion is provided on the impli- Platinum Taq HiFi DNA Polymerase (Invitrogen; cations this observation may have in the biology of Carlsbad, CA, USA) was used for the RT–PCR. Reverse SARS-CoV-2. transcription was done at 50°C for 30 min, followed by 1902 C. ANDRES ET AL. a retrotranscriptase inactivation step at 94°C for 2 min. quality reads that did not meet the requirements Next, 30 cycles of PCR amplification were performed as were discarded. follows: denaturation at 94°C for 15 sec, annealing at (2) Next, all reads with more than 5% of bases below a 54°C for 30 sec, and elongation at 68°C for 5 min. Phred score of Q30 were filtered out. After the last cycle, amplification ended with a final (3) The reads were demultiplexed by matching pri- elongation step at 68°C for 5 min. mers, allowing a maximum of three mismatches, The second round of amplification (nested) was and the primers were trimmed at both read ends. done using overlapping internal primer pairs to Identical reads were collapsed to haplotypes with amplify fragments 470 bp to 313 bp in length. The Fas- the corresponding frequencies as read counts. A tStart High-Fidelity PCR System dNTPack (Sigma, fasta file was generated with each pool/primer/ St. Louis, MO, CA) was used for this purpose, as fol- strand combination. The reverse haplotypes were lows: activation at 94°C for 4 min, followed by 30 cycles reverse complemented. with denaturation at 94°C for 30 sec, annealing at 55°C (4) Raw forward and reverse haplotypes were multiple for 30 sec, and elongation at 72°C for 40 sec, ending aligned with MUltiple Sequence Comparison by with a single elongation step at 72°C for 7 min. Log-Expectation (MUSCLE) [14], then separated PCR products were purified using the QIAquick Gel into strands, and haplotypes common to both Extraction Kit (Qiagen, Hilden, Germany) with QG strands at abundances ≥0.1% were identified. buffer, following the manufacturers’ instructions, and Low-abundance haplotypes (<0.1%) and those eluted DNA was quantified by fluorometry using the unique to one strand were discarded. The haplo- QUBIT dsDNA BR Assay Kit (ThermoFisher, MA, types common to both strands, with frequencies USA). For each patient, PCR products were normalized not below 0.1% were called consensus haplotypes, to 1.5 ng/µL, pooled in a single tube, and purified using and were the basis of subsequent computations. KAPA Pure Beads (KapaBiosystems, Roche, Pleasan- ton, CA, USA) to ensure that no short DNA fragments The amino acid alignments were computed as were present in the library. Library preparation was follows: done using the KAPA Hyper Prep Kit (Roche Applied Science, Pleasanton, CA, USA) and each pool was indi- (1) Gaps were removed and haplotypes translated to vidually indexed using the SeqCap Adapter Kit A/B amino acids. (Nimblegen, Roche, Pleasanton, CA, USA). After (2) The translated stops generated were identified, and library enrichment and a second clean-up with haplotypes were trimmed after the stop. KAPA Pure Beads, the pools were quantified again (3) Resulting amino acid haplotypes were realigned using the QUBIT dsDNA BR Assay Kit and quality- with MUSCLE (EMBL-EBI https://www.ebi.ac. tested using the 4150 TapeStation System (Agilent, uk/Tools/msa/muscle/). Santa Clara, CA, USA). All pools underwent a final normalization to 4 nM, and 10 µL of each pool was All computations were made in the R language and added to the final library tube. The final library was platform [15], developing in-house scripts using Bio- qPCR-quantified using the KAPA Library Quantifi- strings [16] and Ape [17] packages. cation Kit (KapaBiosystems, Roche, Pleasanton, CA USA) in a LightCycler 480 system (Roche) to obtain Results the precise concentration of indexed DNA. PhiX V3 internal DNA control (Illumina, San Diego, CA, Eighteen COVID-19 patients (10 mild and 8 severe) USA) was added to the final dilution. The library was were included in the study. In total, 48,746,647 reads, loaded in a MiSeq Reagent Kit 600V3 cartridge (Illu- ranging from 81,202 to 597,558 reads per amplicon mina, San Diego, CA) and sequenced using the (median 171,478), were analysed from upper respirat- MiSeq platform (Illumina, San Diego, CA). ory tract samples (Table 1), using 13 overlapping amplicons covering the complete S protein. Thus, we studied 3,749,742 complete S genes, with a mean of 208,319 per patient. Sequences have been uploaded to Bioinformatics analysis. InDel study the GenBank Sequence Read Archive (SRA) database The sequence analysis aimed to obtain high-quality with BioProject accession number PRJNA630679. haplotypes fully covering the amplicons. The pipeline Results related to amplicon positions, coverage, per- comprises the following steps: centage of the master sequence, gap incidence per patient and per amplicon, and premature stop codons (1) Amplicons were reconstructed from the corre- are available as Supplementary Tables S1–S10. sponding R1 and R2 paired ends using FLASH Deletions were not randomly accumulated along the S [13] and setting a minimum of 20 overlapping gene, but instead, were found at specific regions (Figure 1, bases and a maximum of 10% mismatches. Low- Figures S1–S27). Deletions coded as delta (Δ1–Δ18) EMERGING MICROBES AND INFECTIONS 1903 Table 1. Characteristics of patients with mild and severe COVID-19. #P16 had clinical symptoms consistent with severe disease, but he was not hospitalized in the ICU. Sample Id (P = Real-time PCR cycle COVID-19 Sex (F = female; Age Days at Intensive Care patient) threshold (Ct) value classification Sample Type M = male) (years) Unit (ICU) P01 19.00 Mild nasopharyngeal aspirate F 34 no admission P02 25.00 mild nasopharyngeal aspirate F 54 no admission P03 16.40 mild nasopharyngeal/ F 42 no admission oropharyngeal swab P04 23.10 mild nasopharyngeal/ F 25 no admission oropharyngeal swab P05 25.98 mild nasopharyngeal/ M 52 no admission oropharyngeal swab P06 21.45 mild nasopharyngeal/ F 42 no admission oropharyngeal swab P07 25.94 mild nasopharyngeal/ F 25 no admission oropharyngeal swab P14 23.71 mild nasopharyngeal aspirate F 26 no admission P15 27.32 mild nasopharyngeal/ M 41 no admission oropharyngeal swab P18 15.50 mild nasopharyngeal/ M 74 no admission oropharyngeal swab P08 No data severe nasopharyngeal/ F 51 4 oropharyngeal swab P09 25.36 severe nasopharyngeal/ M 49 3 oropharyngeal swab P10 21.23 severe nasopharyngeal/ F 47 16 oropharyngeal swab P11 36.01 severe nasopharyngeal/ M 45 27 oropharyngeal swab P12 31.04 severe nasopharyngeal/ M 51 23 oropharyngeal swab P13 22.94 severe nasopharyngeal aspirate F 44 55 P16 34.35 severe nasopharyngeal/ F 45 # oropharyngeal swab P17 30.77 severe nasopharyngeal/ M 49 10 oropharyngeal swab ranged from 1 to 42 nucleotides lost (Table 2). In some Deletions were found in all amplicons, but they were cases, the sequence recovered the correct reading frame, mainly observed at frequencies <1% (Table 2). Most in others, the frameshift caused the appearance of a pre- deletions in amplicons N04, N05, N06, N10, N11, mature stop codon very close to the deletion site, whereas N12, and N13, were found in only 1 or 2 patients, in still others, a new amino acid segment appeared. whereas deletions in amplicons N01, N02, N08 and Figure 1. Diagram showing location of the deletions found along the Spike gene and protein29. 1904 C. ANDRES ET AL. Table 2. List of deletions found along the spike gene. Deletion code Amplicon at Deleted Deleted Number of Population (Δ=deletion the nucleotide nucleotide amino acid Number of reads with Total frequency (in region) level positions positions nts deleted Patient code deletions reads percentage) Δ1 N01 38–49 13S-17V 12 P01 234 126,140 0.19 Δ2 N01 323–329 108T-110L 2–7 P01-P04-P05- 7329 648,955 1.13 P09-914 Δ3 N02 420–434 140F-145Y 2–14 P01-P02-P04- 10,111 1,617,589 0.63 P05-P06-P09 Δ4 N02 596–603 199G-201F 2–8 P02-P04-P09 1031 653,783 0.16 Δ5 N03 724–736 242L-246R 5–6 P02-P09 2024 259,012 0.78 Δ6 N03-N04 1022–1027 341V-343N 2–4 P01-P02-P04-P06 3809 1,042,526 0.37 Δ7 N04 1120–1128 374F-S-396T 9 P06 387 157,405 0.25 Δ8 N04 1177–1180 393T-394N 4 P01 338 155,414 0.22 Δ9 N04-N05 1283–1324 428D-442D 13–42 P04-P09 3835 496,156 0.77 Δ10 N05 1368–1376 456F-459S 9 P01 245 172,903 0.14 Δ11 N06 1444–1452 482G-484E 9 P01 521 173,278 0.30 Δ12 N07 1865–1870 622V-A-624I 6 P09-P17 16,725 436,054 3.84 Δ13 N07 1888–1894 630T-P-632T 19 P04 193 148,533 0.13 Δ14 N07 1961–1979 654E-660Y 8 P09 1068 192,954 0.55 Δ15 N07 1980–2035 660Y-679N 2–34 P01-P02-P03- 64,978 2,923,548 2.22 P04-P05-P06- P07-P08-P09- P10-P11-P14- P15-P18 Δ16 N08-N09 2451–2467 817F-823F 2–16 P01-P03-P04- 11,739 2,176,059 0.54 P05-P06-P09- P14-P15-P18 Δ17 N10-N11 3018–3019 1006T- 2 P01 1544 299,893 0.51 1007Y Δ18 N12-N13 3499 1167G 1 P08 23,359 472,695 4.94 N09, ranging from 2 to 16 nucleotides, were observed the 43 (90.7%) N07 deletions, a TATA box-like motif in 4–9 patients. A deletion of 6 nucleotides in amplicon (nt 2,007–2,010) was lost. In this particular region, N07 (nt 1865–1870), generating a stop codon, was pre- the deletion was characterized by a similar 3’ cutting sent at a frequency of 3.84% of the quasispecies in edge (Table S11). An interesting result at the amino samples from patients P09 and P17. The largest del- acid level was that regardless of the starting point of etion, involving 42 nucleotides (nt 1283–1324) and the deletion (nt 654, 663, 664, 665, 666, 667 or 671), found in N05 of patient P09, resulted in a loss of 14 in 9 of the mild patients (all except P14) and in 2 severe amino acids, but the reading frame recovered. ones (P09 and P10), the frameshift caused by the del- A striking result was the accumulation of deletions etion generated a new peptide motif, IRLR- (“hot-spot”) in amplicon N07, between nucleotides LILLGGHVV*, with a stop codon (*) at the end 1980–2035 (aa Y660-N679) in 14/18 (78%) patients, (Table 4). which included 100% of the patients with mild dis- In 9 patients, a second deletion hot-spot was found ease (P01-P07, P14, P15 and P18), and only half of deleting a number of nucleotides (from 2 to 16) those with severe disease (P08, P09, P10 and P11). between positions 2451 and 2467 (aa 817F-823F), In this particular hot-spot, deletions Δ2 to Δ34 coinciding with the secondary S cleavage site (S2’). were produced (Figure 2, Table 2). Among the severe The hot spot was located between nt2431 (811K) and patients, P12, P13, and P16 had no deletions in the nt2454 (818I), just after the exact S2’ cleavage site N07 amplicon, and P17 showed a deletion outside (KPSKR/SFI) (Table 2). this hot-spot location (Table 3). Viral variants carry- ing these deletions were significantly more frequent Discussion in mild than severe COVID-19 patients (Fisher test: odds-ratio: 95% confidence interval 0.0 - 0.9605; Here, we describe the naturally occurring deletions in p=0.02288). the SARS-CoV-2 S gene in a set of patients with mild Among the total of 43 deletions detected in ampli- or severe COVID 19. The deletions mainly clustered con N07 (Table 3), a premature stop codon appeared in two hot-spot regions, one (Δ15, affecting aa660- immediately after the deletion site in 5 cases (11.6%) aa679) located upstream but very close to the S1/S2 clea- and the reading frame recovered after losing 4, 5, or vage site (aa 685/686) and the second (Δ16 affecting 7 amino acids in 6 cases. However, a frameshift that aa817-823) situated just upstream to the secondary clea- changed the reading frame and caused the appearance vage site S2’ (aa 815/816). These two deletions were of a premature stop codon several amino acids later found in most of the patient samples studied, and nota- was generated in most of the deletions 32/43 (74.4%), bly, the Δ15 deletion was present in 100% of patients and in consequence the S1/S2 cleavage site and the with mild infection and in half of those with severe dis- polybasic domain (PRRAR/S) disappeared. In 39 of ease, three quarters of the patients studied (Table 2). EMERGING MICROBES AND INFECTIONS 1905 Figure 2. Bar plot of deletions in amplicon N07 in the 18 patients (P01-P18) at the nucleotide level: Panel 1, patients with mild disease; Panel 2, patients with severe disease. The x axis provides the multiple alignment (MA) nucleotide positions and the ampli- tude of the deletions by subregions, and the y axis shows the frequency of the deletion (percentage) on the right and the number of reads on the left. As no insertions were observed, the MA positions correspond to S gene positions. Dashed lines indicates S1/S2 (left) and S2’ (right) cleavage sites. Bar plots for the 18 patients by amplicons are provided in supplementary material (Figures S1 to S14 for nucleotides and S15 to S27 for amino acids). This finding suggests that the deletions are not sporadic amplicons in laboratory-confirmed samples for events even though they were seen in a relatively small SARS-CoV-2 in 18 patients. In studies with other percentage of the viral quasispecies (2.2% for Δ15; and SARS-CoV viruses, several subgenomic RNAs were 0.54% for Δ16). The mutants could be interpreted as a reported to be generated during the cell cycle [6,18]. strategy that natural selection has favoured during the To exclusively study the genomic viral RNA of SARS- SARS-CoV-2 infectious life cycle to facilitate extensive CoV-2, RT–PCR was performed using two large PCR spread of the infection, as is discussed below. products in which the 5’ end of primer pair 1 and the This study involved deep-sequencing of the com- 3’ end of primer pair 2 were designed to be outside plete SARS-CoV-2 spike gene using 13 overlapping the spike region (5’ end in ORF1ab and 3’ end in 1906 C. ANDRES ET AL. Table 3. List of deletions found in amplicon N07 at the nucleotide level aligned under the reference sequence Wuhan Hu-1 (MN908947.3). Alignment between nucleotides 1974 and 2070 is shown. Nucleotides represented in bold red in the reference sequence indicate the S1/S2 cleavage site R (CGT) / S (AGT). Patient MILD/ SEVERE Nucleotide alignments MN908947 CTCATATGAGTGTGACATACCCATTGGTGCAGGTATATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATACCCATTGGTGCAGG–TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P09 CTCATATGAGTGTGACATACCCATTGGTGCAGG–TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P05 CTCATATGAGTGTGACATACCCATTGGTGCAGG–TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACATACCCATTGGTGCAGG–TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼-TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P06 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-TATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P09 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-GTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P02 CTCATATGAGTGTGACATACCCA–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼–TTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P06 CTCATATGAGTGTGACATACCCATTGGTGCAGG-∼∼∼∼∼∼∼∼∼–TAATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P06 CTCATATGAGTGTGACATACCCA–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼–TTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P08 CTCATATGAGTGTGACA–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼TATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P08 CTCATATGAGTGTGACA–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼TATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P08 CTCATATGAGTGTGACA–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼TATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATACCCATTGGTGCAGG-∼∼∼∼∼∼∼∼∼∼∼∼TATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATACCCATTGGTGCAGGTA–∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P02 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P02 CTCATATGAGTGTGACAT-∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P03 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACATACCCATTG–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACAT-∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P05 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P05 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P05 CTCATATGAGTGTGACAT-∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P06 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P07 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P09 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P10 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P14 CTCATATGAGTGTGACATACCCATTGGTGCAGGTA–∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P15 CTCATATGAGTGTGACATACCCATTGGTGCAGGTA–∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P18 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P18 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼-ATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA (Continued) EMERGING MICROBES AND INFECTIONS 1907 Table 3. Continued. Patient MILD/ SEVERE Nucleotide alignments P06 CTCATATGAGTGTGACATA∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼–TCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P09 CTCATATGAGTGTGACATACCCATT∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼–TCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P15 CTCATATGAGTGTG–∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼–TCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P01 CTCATA-∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼∼GGTGCAGGTATATGCGCTAGTTATCAGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P08 CTCATATGAGTGTGACATACCCATTGGTGCAGGTATATGCGCTAGTTA–AGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P11 CTCATATGAGTGTGACATACCCATTGGTGCAGGTATATGCGCTAGTT—AGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P11 CTCATATGAGTGTGACATACCCATTGGTGCAGGTATATGCGCTAGTT—AGACTCAGACTAATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA P04 CTCATATGAGTGTGACATACCCATTGGTGCAGGTATATGCGCTAGTT–∼∼∼∼∼∼∼∼∼∼∼∼-ATTCTCCTCGGCGGGCACGT/ AGTGTAGCTAGTCAA ORF3a) (Table S11, Figure S14). Taking into consider- to correct mutations during replication, the deep- ation that CoV have 3’-5’ ExoN activity (nsp14 sequencing analysis accepted mutants present at a protein), consistent with a proofreading mechanism low frequency of ≥0.1%. Because of the possibility of Table 4. Deletions in amplicon N07 at the amino acid (aa) level. wt = wild type (MN908947.3). S, stop; Lost + S, loss of reading frame and appearance of a stop codon; rRF, recover reading frame. Haplotypes that did not lose the TATA box-like sequence are highlighted in yellow, and haplotypes with a deletion upstream of the TATA box-like sequence are highlighted in blue. Cleavage S1/S2 amino acid site between residues 685 / 686 (PRRAR/S). * stop codon. Patient MILD/SEVERE Amino acid alignments MN908947.3 5’654 EHVNNSYECDIPIGAGICASYQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS 708 3’ P01 EHVNNSYECDIPIGAGMR* S P09 EHVNNSYECDIPIGAGMR* Lost + S P05 EHVNNSYECDIPIGAGMR* Lost + S P04 EHVNNSYECDIPIGAGMR* Lost + S P01 EHVNNSYECDIPIYALVIRLRLILLGGHVV* Lost + S P01 EHVNNSYECDIYALVIRLRLILLGGHVV* Lost + S P04 EHVNNSYECDIYALVIRLRLILLGGHVV* Lost + S P06 EHVNNSYECDIYALVIRLRLILLGGHVV* Lost + S P09 EHVNNSYECDIVIRLRLILLGGHVV* Lost + S P02 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P06 EHVNNSYECDIPIGAG—NQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF P06 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P08 EHVNNSYECDISDSD* Lost + S P08 EHVNNSYECDISDSD* Lost + S P08 EHVNNSYECDISDSD* Lost + S P01 EHVNNSYECDIPIGAGIRLRLILLGGHVV* Lost + S P01 EHVNNSYECDIPIGAG———NQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF P01 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P02 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P02 EHVNNSYECDISDSD* Lost + S P03 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P04 EHVNNSYECDIPI——-DQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF P04 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P04 EHVNNSYECDISDSD* Lost + S P05 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P05 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P05 EHVNNSYECDISDSD* Lost + S P06 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P07 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P09 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P10 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P14 EHVNNSYECDIPIGAG—-NQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF P15 EHVNNSYECDIPIGAG—-NQTQTNSPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF P18 EHVNNSYECDIIRLRLILLGGHVV* Lost + S P18 EHVNNSYECDIPIIRLRLILLGGHVV* Lost + S P06 EHVNNSYECDISDSD* Lost + S P09 EHVNNSYECDIPISDSD* Lost + S P15 EHVNNSYECVRLRLILLGGHVV* Lost + S P01 EHVNNS* S P08 EHVNNSYECDIPIGAGICAS* S P11 EHVNNSYECDIPIGAGICAS* S P11 EHVNNSYECDIPIGAGICAS* S P04 EHVNNSYECDIPIGAGICASY—-SPRRAR/SVASQSIIAYTMSLGAENSVAYS rRF 1908 C. ANDRES ET AL. PCR artefacts, deep-sequencing point mutations, or which are mainly directed to this protein. In this deletion of single nucleotides generated mainly at way, neutralization of circulating virions is reduced, homopolymeric sites, we did not include single del- favouring viral infection [23]. etions unless they were found in different patients The free S1 binding subunit of SARS-CoV-2 without and in overlapping amplicons at higher frequencies its membrane anchor S2 could have similar functions (>1%). No insertions were found. (Figure 3). One putative action of secreted free S1 Entry of the viral genome into the cell depends on protein might be to attach to the human ACE2 cell recognition and binding of the surface subunit S1 to receptor, thereby competing with complete viral par- the ACE2 human receptor [11], whereas the S2 subunit ticles to re-infect or newly infect respiratory tract is responsible for fixing the S protein to the viral mem- cells, resulting in less severe disease. This could be brane surface. After binding to the ACE2 cell receptor, interpreted as an effect of natural selection to attenuate the S protein is primed by the serin-protease the infection and facilitate its persistence with minimal TMPRSS2, which leads to S protein cleavage at S1/S2 damage, increasing the human-to-human transmission and S2’ [8]. After cleavage, S1 remains attached to into the community. This strategy, which we have ACE2, while subunit S2 anchors the viral and cellular dubbed “Don’t burn down the house” is supported membranes, inducing fusion and viral entry. The Δ15 by the finding that the minor variants carrying these deletion (Table 2) mainly causes a frameshift that gen- deletions were statistically more frequent in patients erates an in-frame stop codon. The presence of this with mild than severe COVID-19. This self-modulating new stop codon would result in translation of a trun- viral strategy has also been seen in hepatitis delta virus cated S, which would consist of an almost complete (HDV) infection, where one viral antigen (short HDV S1 subunit, and total absence of the S2 subunit respon- antigen, SHDAg) enhances HDV replication, while a sible for anchoring S to the lipid membrane of the viral second antigen (large HDV antigen, LHDAg), pro- particle. The absence of the S2 anchor peptide suggests duced after a stop codon edition (TAG to TGG) by cel- that S1 could be produced as a “free” protein (free S1). lular adenosine deaminase, acts as a negative regulator As S1 is located on the exposed outside of SARS-CoV-2 of replication [24]. in the crown structures, it could have hydrophilic The fact that the truncated S protein was present in domains and be a soluble peptide with potential for only a low percentage of the entire viral quasispecies release outside the infected cell, in the lower respiratory suggests that natural selection may have designed a tract and even to plasma (Figure 3). These free soluble favourable equilibrium in which a limited number of proteins, which are not a part of the viral cycle or com- deleted virions are generated to balance virus pro- ponents of the viral particles have also been observed in duction with infection of new cells during disease pro- other viral infections. For example, a huge amount of gression. A likely reason for maintaining a minority “empty” subviral genomic particles, consisting of viral population of genomes with deletions able to produce envelope proteins (HBsAg), are often found in plasma free S1 protein would be to infect a host while causing of patients with hepatitis B virus (HBV) infection. minimal damage, which would greatly facilitate trans- These empty particles are produced and secreted mission of the virus within the population. However, during HBV infection, and have an immunomodula- the mutants were also found in half the patients with tory role [19]. In addition, soluble HBV e antigen severe disease; hence, additional study is needed to (HBeAg), which is not a component of the viral par- determine whether they also relate to disease severity. ticles and shares immunoactive epitopes with the In clinical practice, it has been seen that progression HBV core antigen (HBcAg viral capsid component), to severe disease can occur within hours, which is detected during HBV infection and has an immuno- suggests that any variant associated with virulence modulatory role [20]. should be detected at the time of the diagnosis. The Human respiratory syncytial virus (HRSV) is samples studied here were obtained on the day patients another respiratory virus with the ability to produce were admitted to the emergency room, and very close pre-anchored proteins. The attachment protein (G) to the onset of infection. Does the percentage of these of HRSV is an anchored protein whose main function viral mutants change during disease progression? To is viral attachment to the host’s cell membrane through elucidate this issue, it would be of interest to investigate a still unknown receptor [21]. As in many other changes in the frequency of deleted genomes in a large viruses, this protein has several functions, and in this number of patients and in sequential samples from the case, because of the existence of a second start codon, same patient, together with virus culture experiments a soluble form of G protein lacking the anchor is pro- to determine whether the presence of deleted sequences duced, and this is shed to the extracellular medium [22] increases or not during the passages. As a consequence in abundant quantities by infected cells. The function of the frameshift, a new peptide motif, IRLR- of soluble, free G is to inhibit toll-like receptors, LILLGGHVV*, appeared in several sequences with a thereby modulating the host’s immune response. Free deletion that started in different nucleotide points. G also binds to the host’s neutralizing antibodies, Additional work is also needed to determine whether EMERGING MICROBES AND INFECTIONS 1909 Figure 3. Based on the life cycle of SARS-CoV, this diagram represents the hypothesis derived from our results. Entry of the virus in the host cell is shown at the top right of the diagram. At the transcription step, two scenarios are depicted: to the left, the viral particle resulting from normal S protein, and to the right the viral particle resulting from truncated S protein. In normal conditions, once the nucleoprotein is freed into the cytoplasm ss + RNA is translated into the non-structural proteins required for transcription. ss + RNA is transcribed into ss-RNA and later into genomic ss + RNA which is encapsidated (left side of the figure). Once the com- plete viral particle has been formed, it is secreted from the cell by exocytosis. The right side of the figure depicts the situation when a deletion occurs in the S gene during transcription of the complete genome and before subgenomic mRNAs are generated to produce the structural proteins. Translation of a deleted subgenomic spike mRNA would lead to a truncated S protein composed of the S1 domain without S2, which could be shed outside the cell as free S1. The box depicts possible destinations of free S1, which could bind to (1) the ACE2 cell receptor, (2) S1-specific neutralizing antibodies, or (3) free ACE2 receptor. ***The red triangle indi- cates the deletion in genomic RNA. ***Abbreviations: ACE2, angiotensin converting enzyme 2; mRNA, messenger RNA; NAb; neu- tralizing antibodies; pp1a, polyprotein 1a; RdRp, RNA-dependent RNA polymerase; S, spike; S1, subunit S1 at the N-terminal domain of the S protein, which includes receptor binding domain (RBD); S2, subunit S2 located at the C-terminal domain of S protein, which includes fusion peptide (FP), heptad repeat (HR) domain 1 and 2, and the transmembrane domain (TM); ss, single stranded; ss + RNA, single-stranded positive sense RNA; TMPRS22, human serine protease TMPRSS2. acquisition of this peptide motif has biological virus in the population. The observation of mutation consequences. hot spots in the S gene opens the door to further Two other putative consequences of the S mutants work on a number of potentially related aspects. might be that free S1 protein could bind with S-specific Recent studies have reported the presence of deleted antibodies, acting as a decoy and weakening the variants in the S1/S2 junction in virus isolated by cell immune response, or to circulating ACE2, released culture of clinical specimens [27]. Deletions of 10–15 from the cell membrane to plasma [25,26], with cardi- nucleotides at the S1/S2 junction were identified by ovascular effects. However, as the deletions were plaque purification of Vero-E6 cultured SARS-CoV-2 mainly found in patients with mild disease and consid- genomes obtained from nasopharyngeal aspirate of a ering the zoonotic origin of the virus (animal immune COVID-19 patient. Infection of hamsters with virus and cardiac systems differ from human ones) and the containing these variants led to attenuated viral disease short time that the virus has been evolving in the [27,28]. Digital PCR-based assays demonstrated that human population, we believe that the most likely such mutants carrying deletions at low intra-host fre- reason for maintaining a minor population of mutant quency can also be transmitted from human to genomes able to produce free S1 protein would be to human, which suggests that they may have significant cause an infection with limited damage in the host, implications in the zoonotic origin and natural evol- thus facilitating transmission and persistence of the ution of SARS-CoV-2 [28]. These findings support 1910 C. ANDRES ET AL. our hypothesis that deletions close to the S1/S2 clea- Author contributions vage site are likely a natural phenomenon. Here, we CA, DG-C, MP and MCM have significantly contribu- suggest that this phenomenon may have been favoured ted to designing the experimental work and performed by natural selection to enhance the spread of SARS- the technical work involving RNA extraction, amplifi- CoV-2. cation, and deep sequencing To conclude, in-depth sequencing of the SARS- JG and MG-M developed the software and statistics CoV-2 S gene in 18 patients with COVID-19 enabled used in the study identification of a naturally occurring deletion very FRF, JE, AR, LG, BA, RF and MGC collected close to the S1/S2 cleavage site. Our results indicate samples and actively participated in discussions and that the mutant S would have a large impact on the S corrections related to the draft protein, and suggest that the virus could produce free SQ designed the graphics and actively participated S1, which may have implications regarding the candi- in the draft discussion and corrections. dacy of S protein as a target for vaccination and anti- MC, LG, BA, RF, JIE, and TP participated in the viral treatment strategies. The deletions were analysis and interpretation of the data, and substan- significantly more prevalent in patients with mild tially revised the draft than in those with severe disease, supporting the notion AA and JQ conceived the work, designed the pri- that they could be a strategy of natural selection to mers, led the data analysis, wrote the draft, and led decrease the injury caused after onset of the infection. the discussion. In this “Don’t burn down the house” strategy, the abil- ity of the virus to bind with ACE2 receptor and spread to others would be unchanged; thus its propensity for Conflicts of interest transmission would be enhanced by a mildly affected host. To prove this hypothesis, it is essential to investi- We declare that no public or private company has had gate whether the truncated S protein (free S1) is present any role in the study design, data collection, exper- in respiratory tract specimens and in plasma. To detect imental work, data analysis, decision to publish, or free S1 at low concentration by western blot analysis, preparation of the manuscript. Roche Diagnostics entire and truncated recombinant spike proteins S.L. provided support in the form of a salary for one should be used as controls, together with highly specific of the authors (Josep Gregori), but the company did antibodies to S protein. These studies are currently not have any role in the study design, data collection ongoing, and in parallel, we are investigating whether and analysis, decision to publish, or preparation of the new peptide motif IRLRLILLGGHVV* will have the manuscript. sufficient antigenicity to be used as a probe to detect No other competing interests to declare. Thus, our truncated free S1 protein. adherence to policies on sharing data and materials is not altered. Acknowledgements Ethics Committee approval The authors thank Celine Cavallo for English language The work has been approved by Vall d'Hebron Hospi- support. tal Universitary Ethical Committee reference number PR(AG)259-2020. Disclosure statement ORCID No potential conflict of interest was reported by the author Josep Gregori http://orcid.org/0000-0002-4253-8015 (s). Lidia Goterris http://orcid.org/0000-0003-1471-4461 Maria Gema Codina http://orcid.org/0000-0002-5313-5086 Magda Campins http://orcid.org/0000-0002-8841-6195 Tomás Pumarola http://orcid.org/0000-0002-5171-7461 Funding Andrés Antón http://orcid.org/0000-0002-1476-0815 Josep Quer http://orcid.org/0000-0003-0014-084X This study was partially supported by the Direcció General de Recerca i Innovació en Salut (DGRIS) Catalan Health Ministry Generalitat de Catalunya through Vall d’Hebron References Institut de Recerca (VHIR); European Development Regional Fund (ERDF) “A way to achieve Europe” by Span- [1] Holland JJ. Replication error, quasispecies populations ish Network for Research in Infectious Diseases (REIPI and extreme evolution rates of RNA viruses. Oxford: RD16/0016/0003); and Centro para el Desarrollo Tecnoló- Oxford University Press; 1993. gico Industrial (CDTI) from the Spanish Ministry of Econ- [2] Domingo E, Escarmis C, Sevilla N, et al. Basic concepts omy and Business, grant number IDI-20200297. in RNA virus evolution. FASEB J. 1996;10. EMERGING MICROBES AND INFECTIONS 1911 [3] Eckerle LD, Becker MM, Halpin RA, et al. Infidelity of [18] Song Z, Xu Y, Bao L, et al. From SARS to MERS, SARS-CoV Nsp14-exonuclease mutant virus replica- Thrusting coronaviruses into the Spotlight. Viruses. tion is revealed by complete genome sequencing. 2019;11; doi:10.3390/v11010059. PLoS Pathog. 2010;6:e1000896. [19] Caballero A, Tabernero D, Buti M, et al. Hepatitis B [4] Muth D, Corman VM, Roth H, et al. Attenuation of virus: The challenge of an ancient virus with multiple replication by a 29 nucleotide deletion in SARS-coro- faces and a remarkable replication strategy. Antiviral navirus acquired during the early stages of human- Res. 2018;158:34–44. to-human transmission. Sci Rep. 2018;8:15177. [20] Kramvis A, Kostaki E-G, Hatzakis A, et al. [5] Fauci AS, Lane HC, Redfield RR. Covid-19 - Immunomodulatory function of HBeAg related to Navigating the Uncharted. N. Engl. J. Med. short-Sighted evolution, transmissibility, and clinical 2020;382:1268–1269. Manifestation of hepatitis B virus. Front Microbiol. [6] Tu Y-F, Chien C-S, Yarmishyn AA, et al. A Review of 2018;9:2521. SARS-CoV-2 and the ongoing clinical Trials. Int J Mol [21] Borchers AT, Chang C, Gershwin ME, et al. Sci. 2020;21; doi:10.3390/ijms21072657. Respiratory syncytial virus–a comprehensive review. [7] Belouzard S, Millet JK, Licitra BN, et al. Mechanisms of Clin Rev Allergy Immunol. 2013;45:331–379. coronavirus cell entry mediated by the viral spike [22] Hendricks DA, McIntosh K, Patterson JL. Further protein. Viruses. 2012;4:1011–1033. characterization of the soluble form of the G glyco- [8] Hoffmann M, Kleine-Weber H, Schroeder S, et al. protein of respiratory syncytial virus. J Virol. SARS-CoV-2 cell entry depends on ACE2 and 1988;62:2228–2233. TMPRSS2 and Is Blocked by a Clinically Proven pro- [23] Polack FP, Irusta PM, Hoffman SJ, et al. The cysteine-rich tease Inhibitor. Cell. 2020;181:271–280.e8. region of respiratory syncytial virus attachment protein [9] Jeffers SA, Tusell SM, Gillim-Ross L, et al. CD209L (L- inhibits innate immunity elicited by the virus and endo- SIGN) is a receptor for severe acute respiratory syn- toxin. Proc Natl Acad Sci U S A. 2005;102:8996–9001. drome coronavirus. Proc Natl Acad Sci U S A. [24] Hughes SA, Wedemeyer H, Harrison PM. Hepatitis 2004;101:15748–15753. delta virus. Lancet (London, England). 2011;378:73–85. [10] de Wit E, van Doremalen N, Falzarano D, et al. SARS [25] Turner A. Chapter 25 - ACE2 cell biology, Regulation, and MERS: recent insights into emerging corona- and Physiological functions. In: T Unger, UM viruses. Nat Rev Microbiol. 2016;14:523–534. Steckelings, SR dos, editor. The Protective Arm of the [11] Andersen KG, Rambaut A, Lipkin WI, et al. The proxi- Renin angiotensin System (RAS) Functional aspects and mal origin of SARS-CoV-2. Nat. Med. 2020;26:450–452. therapeutic implications. Elsevier Inc.; 2015. p. 185–189. [12] Heald-Sargent T, Gallagher T. Ready, set, fuse! The [26] Matsushita K, Ding N, Kou M, et al. The relationship coronavirus spike protein and acquisition of fusion of COVID-19 severity with cardiovascular disease competence. Viruses. 2012;4:557–580. and its traditional risk factors: A systematic review [13] Magoc T, Salzberg SL. FLASH: fast length adjustment and meta-analysis. BMJ medRxiv Prepr. 2020. doi:10. of short reads to improve genome assemblies. 1101/2020.04.05.20054155. Bioinformatics. 2011;27:2957–2963. [27] Lau S-Y, Wang P, Mok BW-Y, et al. Attenuated SARS- [14] Edgar RC. MUSCLE: a multiple sequence alignment CoV-2 variants with deletions at the S1/S2 junction. method with reduced time and space complexity. Emerg Microbes Infect. 2020;9:837–842. BMCBioinformatics. 2004;5:113. [28] Wong YC, Lau SY, Wang To KK, et al. Natural trans- [15] Team RC. (2016). R: A language and environment for mission of bat-like SARS-CoV-2 ΔPRRA variants in statistical computing. COVID-19 patients. Clin Infect Dis. 2020. doi:10. [16] Pages H, Aboyoun P, Gentleman R, et al. (2012). 1093/cid/ciaa953. Biostrings:String objects representing biological [29] Xia S, Liu M, Wang C, et al. Inhibition of SARS-CoV-2 sequences, and matching algorithms. R package 2.38.4. (previously 2019-nCoV) infection by a highly potent [17] Paradis E, Claude J, Strimmer K. APE: Analyses of pan-coronavirus fusion inhibitor targeting its spike Phylogenetics and evolution in R language. protein that harbors a high capacity to mediate membrane Bioinformatics. 2004;20:289–290. fusion. Cell Res. 2020;30:343–355.