Direct metagenomic detection of viral pathogens in nasal and fecal specimens using an unbiased high-throughput sequencing approach

Abstract

With the severe acute respiratory syndrome epidemic of 2003 and renewed attention on avian influenza viral pandemics, new surveillance systems are needed for the earlier detection of emerging infectious diseases. We applied a "next-generation" parallel sequencing platform for viral detection in nasopharyngeal and fecal samples collected during seasonal influenza virus (Flu) infections and norovirus outbreaks from 2005 to 2007 in Osaka, Japan. Random RT-PCR was performed to amplify RNA extracted from 0.1-0.25 ml of nasopharyngeal aspirates (N = 3) and fecal specimens (N = 5), and more than 10 microg of cDNA was synthesized. Unbiased high-throughput sequencing of these 8 samples yielded 15,298-32,335 (average 24,738) reads in a single 7.5 h run. In nasopharyngeal samples, although whole genome analysis was not available because the majority (>90%) of reads were host genome-derived, 20-460 Flu-reads were detected, which was sufficient for subtype identification. In fecal samples, bacteria and host cells were removed by centrifugation, resulting in gain of 484-15,260 reads of norovirus sequence (78-98% of the whole genome was covered), except for one specimen that was under-detectable by RT-PCR. These results suggest that our unbiased high-throughput sequencing approach is useful for directly detecting pathogenic viruses without advance genetic information. Although its cost and technological availability make it unlikely that this system will very soon be the diagnostic standard worldwide, this system could be useful for the earlier discovery of novel emerging viruses and bioterrorism, which are difficult to detect with conventional procedures.

Figure 1. Random RT-PCR amplification of cDNA from clinical specimens and quantitative RT-PCR with virus-specific primers.

The samples were nasopharyngeal aspirates and stools (n = 3 and 5, respectively) isolated during 2005–2007 in Osaka, Japan. Influenza A virus and norovirus were detected in the nasopharyngeal aspirates and stool samples, respectively, with other diagnostic methods. (A) RNA extracted from clinical specimens was reverse-transcribed and random-PCR amplified to prepare template DNA for pyrosequencing. One microgram of amplified PCR products in each sample was loaded onto a 1% agarose gel. M indicates 100-bp DNA ladder (NEB). (B) Flu-specific semi-quantitative PCR was performed with 10-fold serial dilutions of the random-PCR products. (C) Norovirus (GII)-specific semi-quantitative PCR was performed with 10-fold serial dilutions of the random-PCR products. Quantitative real-time RT-PCR using a norovirus-specific primer set was also performed, and the estimated copy numbers of norovirus in samples #N1–#N5 are shown in the panel on the right. As a control, cDNA from sample #N3 without random PCR amplification was used (left panel).

Figure 2. Pyrosequencing using the GS FLX system.

Amplified cDNA was used as a template for GS FLX analysis. A 70×75 PicoTiterPlate device (gasket for 16 regions) was divided into 2 regions each for 8 samples. Obtained data were then subjected to a data analysis pipeline, as described in the Methods section. Comparison of the organisms from which the best matches for the sequences was shown.

Figure 3. BLASTN (A) and PCR (B) analyses of WUV in Flu-positive nasopharyngeal aspirates.

(A) Alignment of the WUV VP1 nucleotide sequence. The read obtained with the GS FLX sequencer (query) was compared with that of the WUV strain CLFF (subject; NCBI accession number: EU296475). (B) The WUV VP2 gene was detected by PCR using cDNA, which was amplified with random RT-PCR, as a template. The cDNA was diluted 10- and 100-fold and the PCR product was loaded on a 1% agarose gel. M indicates 100-bp DNA ladder.

Figure 4. Process diagram for the viral diagnosis of nasopharyngeal aspirates and fecal samples.