Direct Metagenomic Detection of Viral Pathogens in Nasal and Fecal Specimens Using an Unbiased High-Throughput Sequencing Approach

Nakamura, Shota; Yang, Cheng-Song; Sakon, Naomi; Ueda, Mayo; Tougan, Takahiro; Yamashita, Akifumi; Goto, Naohisa; Takahashi, Kazuo; Yasunaga, Teruo; Ikuta, Kazuyoshi; Mizutani, Tetsuya; Okamoto, Yoshiko; Tagami, Michihira; Morita, Ryoji; Maeda, Norihiro; Kawai, Jun; Hayashizaki, Yoshihide; Nagai, Yoshiyuki; Horii, Toshihiro; Iida, Tetsuya; Nakaya, Takaaki

Our preliminary experiments demonstrated that the detected number of viral sequences paralleled the virus copy number in blood samples (unpublished data), suggesting that this system is highly quantitative. Indeed, the copy number of norovirus in the #N1 to #N5 samples ( Figure 1C ), as measured by semiquantitative PCR, was significantly correlated with the number of norovirus genome sequences as detected by high-throughput sequencing (Table 1 ). In the case of the nasal samples, the copy number of the influenza virus in sample #F2 was lower than those in samples #F1 and #F3, although the endpoints of the semiquantitative PCR for these samples were comparable ( Figure 1B) . This inconsistency could be due to differences in the sensitivities of Thus, quantitative analysis of host genes will be required. One potential reason for why we obtained fewer Flu-specific reads than norovirus reads in this study might have been the large number (90.0-94.6% of all reads) of host-derived sequences (Figure 2 and Table 5 ). These sequences were present because we performed direct RNA isolation from nasopharyngeal aspirates without first eliminating the cells or tissues. Most of the detected human-derived reads were non-coding regions, and fewer coding regions, including rRNA and mRNA sequences, were detected than expected (Table 5 ). These results suggest that, although contamination by human genomic DNA might be very low, an additional step for host gene removal is required. Suitable subtraction step(s) using pooled human genomic DNAs as drivers might be required to enrich in microbial genomes [15] . Alternatively, MICROBEnrich (Ambion Inc.), another method for removing contaminated human-derived RNA, could be useful to enrich microbial RNA [33] . However, the DNA virus WUV was detected from the isolated RNA, suggesting that the WUV genome and/or its transcripts present in infected cells were detected. Indeed, a novel human polyomavirus (Merkel cell polyomavirus), isolated from skin carcinoma, was detected from mRNA [17] . Taken together, these results indicate that whole RNA isolation, including host cells and tissues, followed by the suitable elimination of hostderived genes could be an effective method for identifying pathogenic viruses in clinical samples. When several pathogens are found in a single sample, a careful interpretation is necessary to decide which pathogen(s) is the real cause of a specific disease. Although, the most abundant pathogen might generally be considered to be the best candidate, cooperative interactions between multiple pathogens cannot be excluded as an important factor for pathogenesis. To address this question, suitable control samples from healthy persons and/or pair specimens, isolated after recovery, might be required.