Expanding the Conversation on High-Throughput Virome Sequencing Standards To Include Consideration of Microbial Contamination Sources

LETTER

We thank Ladner and colleagues for their conversation about standardizing viral genome sequences derived from high-throughput (HT) sequencing technology. In their editorial “Standards for Sequencing Viral Genomes in the Era of High-Throughput Sequencing,” published in the May-June 2014 issue of mBio (1), they raise standardization issues and propose the development of categories to define viral genome assemblies. These are timely discussion points that will likely foster more robust repositories of viral genome sequences. At the same time, their discussion will likely raise additional issues that are important to address in the coming years.

Among a number of issues, Ladner et al. describe the use of HT sequencing as an approach to globally screen viral stocks for microbial contamination. Contamination is an important issue here since the isolation and maintenance of viral stocks in host tissue culture cells and the manufacturing of vaccines in mammalian species lend themselves to potential microbial contamination. Screening biologicals for safety and purity using HT sequencing has already proven useful, as exemplified by the identification of noninfectious viral sequences in several live-attenuated viral vaccines, including the identification of porcine circovirus in a human rotavirus vaccine preparation (2). HT sequencing was also recently used to identify the causative agent of the mysterious Theiler’s disease in horses inoculated with equine-derived biologicals (3). In this case, the resulting culprit was identified as a novel virus, Theiler’s disease-associated virus (TDAV) (3). In addition to viral contaminants, Mycoplasma sp. is a common contaminant in cell culture that can be transferred to viral stocks. Nanobacterium sp. was previously identified in 100% of cattle serum in a U.S. herd (4), which likely led to the contamination of cell cultures where it was found to interfere with cell growth (5). Without a doubt, the high sensitivity and specificity of HT sequencing lend themselves exceedingly well to the detection of a broad range of genetic material across organisms. Despite this potential, there is an unassumed impediment to this technology that needs to be addressed.

In the past several years, our laboratory has interrogated HT sequencing data sets for the identification and characterization of viral and bacterial pathogens (6–10). In the course of our investigations, we noted surprisingly high levels of a spectrum of microbial genetic materials in nearly every sample that we have analyzed, including samples that were thought to be pristine. In the work of Strong et al. (11), we describe the pervasiveness of microbial reads in sequencing data across cohorts, sample types (e.g., cell line or biopsy material), and study protocols. In that study, we determine that the bulk of microbial reads did not represent bona fide infections and likely originated from sample preparation/sequencing procedures. Some sources of contamination have been identified by other groups and include microbes present in ultrapure water systems (12) and NIH-CQV virus contamination from silica column-based nucleic acid extraction kits (13–17). In our work, we proposed possible nucleic acid contamination of library preparation reagents such as polymerases and nucleotides that are typically made in bacteria (11). In addition, we have observed likely contamination across RNA/cDNA samples (11, 18). Such contaminants can easily impact reported findings, as illustrated by the inadvertent inclusion of microbial sequences in assembled worm genomes and other eukaryotic genomes (12, 19).

Due to the sensitive nature of HT sequencing, microbial reads derived from sample/sequencing procedures will inevitably lead to data misinterpretations and false-positive findings. Conversely, identifying bona fide microbial contaminations in a background of false contaminations can be cumbersome and/or challenging.

HT sequencing will likely be transformative for microbial detection. Nevertheless, until we fully understand the sources of contamination and work to eradicate these sources, we need to implement stringent, well-controlled sequencing and analysis pipelines. Reducing or eradicating these false contamination sources will lead to easier interpretation and greater data veracity. These false contamination issues can likely be addressed, but the first step in this process is to acknowledge that they exist.