Analysis of intrapatient heterogeneity uncovers the microevolution of Middle East respiratory syndrome coronavirus

Park, Donghyun; Huh, Hee Jae; Kim, Yeon Jeong; Son, Dae-Soon; Jeon, Hyo-Jeong; Im, Eu-Hyun; Kim, Jong-Won; Lee, Nam Yong; Kang, Eun-Suk; Kang, Cheol In; Chung, Doo Ryeon; Ahn, Jin-Hyun; Peck, Kyong Ran; Choi, Sun Shim; Kim, Yae-Jean; Ki, Chang-Seok; Park, Woong-Yang

Despite the relatively high reproducibility, we focused on nucleotide variants present at a relatively high frequency (i.e., ≥30%) to minimize false positives among non-consensus variants. Positions where a nucleotide variant was present at a frequency of ≥30% in any of the eight isolates were selected for further investigation using additional samples. Targeted deep sequencing was performed on an additional 27 samples targeting 3003-bp regions that included the variable sites. Eleven libraries from independent cDNA syntheses of five samples, including four libraries in duplicate and one in triplicate, were constructed to evaluate the technical noise in measuring allele frequency. Only variants with frequencies significantly greater than those from technical noise were considered true variants. Technical noise is described in Methods and was based on a 5% significance level after performing Bonferroni adjustment. A total of 16 positions displaying intrapatient heterogeneity were identified through statistical testing. The results from seven specimens were validated using Sanger sequencing. We tested nine sites and confirmed concordant mixed bases at seven positions-namely, positions 6884, 7317, 11257, 21726, 22356, 22984, and 23041 (Fig. 1C ), but found discrepancies at positions 7322 and 19075. Although we were not able to validate all variants by Sanger sequencing owing to a limited sample volume, the results from the test set showed a 78% validation rate. In addition, the frequencies of the two variants at positions 22984 and 23041 (i.e., G at position 22984 and T at 23041; A at 22984 and C at 23041) were highly correlated in the specimens (Supplemental Fig. S3 ), suggesting a tight linkage between these variants instead of random sequencing errors. Fourteen positions exhibiting mixed bases are presented in Figure 2 . These include all seven nucleotide variants identified in the full-genome sequences of the eight MERS-CoV isolates, indicating that intrapatient MERS-CoV heterogeneity was closely associated with isolate-specific variants in the consensus sequences. Taken together, our data demonstrate the intrapatient heterogeneity of MERS-CoVs.