2019 Novel Coronavirus Is Undergoing Active Recombination

To the Editor—The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, formerly known as 2019 novel coronavirus [2019-nCoV]) outbreak in Wuhan since December 2019 [1] has quickly spread to 25 countries [2] and caused more than 44 000 cases and 1000 deaths so far [3]. The fast sharing of SARS-CoV-2 genomes by the Global Initiative on Sharing All Influenza Data (GISAID; https://www.gisaid.org) provides a valuable dataset for SARS-CoV-2 haplotype network analysis, which is crucial for transmission and evolutionary tracking surveillance and secondary outbreak prevention. We called single-nucleotide variations (SNVs) for all 84 SARS-CoV-2 genomes in GISAID using MN908947 as the reference. Genomes with the same SNVs are grouped into a haplotype (shown as a pie chart node in Figure 1; also see Supplementary Table 1 for the accessions), then the network is constructed using the median join [4] method in popART [5]. We found that the SARS-CoV-2 haplotype network has obvious characteristics of single origin from haplotype hap_011: (1) the network is star-like, centralized on the haplotype hap_011; (2) hap_011 has the largest sample size and the majority of the samples are from Hubei province, where the outbreak originated (Supplementary Table 1); (3) most of the satellite haplotypes are also from Hubei (Figure 1); and (4) the average collection dates of hap_011 (has 0 mutation relative to MN908947) are earlier than all other mutation groups (Supplementary Figure 1). The single origin of SARS-CoV-2 indicates that persistent animal-to-human transmission is unlikely; otherwise, multiple nodes with the above characteristics should be observed. The median-joining haplotype network of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Each haplotype node is represented by a pie chart, with size indicating the sample size (Supplementary Table 1), and colors indicating collection regions. Each hatch mark along the edge corresponds to a mutation between the 2 haplotypes linked. The mutation notation along a loop edge shows the position and nucleotide status change between the 2 haplotypes linked. We found 5 haplotypes (hap_009, hap_017, hap_023, hap_048, and hap_050) forming loops (Figure 1), which typically indicates the existence of genetic recombination. However, in rare cases, loops can also be formed by recurring sequencing error or parallel/back mutations. Further examination of the differences pattern among the 5 haplotypes found 3 sets of recurrent differences: C29095T, G11083T, and C8782T/T28144C (Figure 1), most of which are nonsynonymous changes except C29095T (Supplementary Table 2). Such highly frequent recurrent differences look extremely unlikely, raised by rare events like recurring sequencing error/mutations. To quantify the significance of recombination over recurring sequencing error/mutations, Phi [6], and Max χ 2 tests [7] in the software PhiPack [8] were performed. It showed P values of .03 and .005 for Phi and Max χ 2 test, respectively, indicating the presence of recombinants in the SARS-CoV-2 population. Because all the haplotypes originate from hap_011, the formation mechanisms of the 5 haplotypes could be inferred as follows: First, hap_048 and hap_050 mutated from the common ancestry hap_011 independently, and hap_009 then mutated from hap_050; next, hap_048 donated the genome region harboring 11083T to hap_050, generating the recombinant hap_017; similarly, hap_048 or hap_017 donated the same region to hap_009 generated the recombinant hap_023 (Figure 2). The putative evolutionary mechanisms of the 5 haplotypes forming the loops. The genome is represented by a long rectangle, which consists of 4 regions represented by short rectangles. Each region contains a single-nucleotide variant (SNV) with coordinate at bottom. The haplotype of a genome is represented by the combined SNVs from each of the regions. SNVs of reference status are shown in black font and SNVs of alternative status, where mutants are sourced from hap_048, hap_050, and hap_009, are shown in green, blue, and red font, respectively. Regions sourced from hap_011/ref, hap_048, and hap_050 are shown in the white, blue, and peach boxes, respectively. Recent studies have found some ancient recombination events around region 11655–20953 (relative to MN908947, containing gene orf1b) in bat coronavirus CoVZC45/CoVZXC21 (a sister group of SARS-CoV-2), but no evidence of recombination for 2019-nCoV [9–11]. This is to be anticipated since all these studies adopted region-wise phylogenetics analysis, which cannot efficiently detect recent recombination events like haplotype network does. However, the recombination region they found in bat coronavirus is quite close to the SARS-CoV-2 recombination site 11083 (harbored by the ORF1a gene) found in this report. This consistency of the ancient and the recent recombination analysis solidifies both our findings and prompts potential recombination hotspots existing around region 11083–20953 of gene ORF1a/ORF1b. This report provides the first evidence for genetic recombination—a new way of evolution besides mutation in SARS-CoV-2. The existence of genetic recombination has the following implications: 2 different SARS-CoV-2 strains (here, hap_048 and hap_050) should have coinfected the same cell; a SARS-CoV-2 strain might acquire new traits such as virulence and drug susceptibility directly from other strains [12, 13]; the adaptability of SARS-CoV-2 to the human immune system might be significantly strengthened through genetic recombination; the accuracy of diagnosis based on serologic and molecular biology assays might be compromised [14]; and the transmission tracking based on phylogenetic tree could be misleading since the topology of the mutation route is a network rather than a tree. We believe the communities of infectious disease physicians and disease control specialists should arouse particular concerns about this finding and its potential implications. And we expect that this finding may provide a valuable perspective for strategies designed for secondary transmission prevention and drug treatments for SARS-CoV-2 pneumonia. Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. Acknowledgments. The author acknowledges the authors and the originating and submitting laboratories of the nucleotide sequences from BetaCoV 2019–2020 of the GISAID EpiFlu Database (9 February 2020, 84 isolates). Potential conflicts of interest. The author: No reported conflicts of interest. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.