Home Resources Feature Articles Addressing Challenges in Microbiome DNA Analysis

Addressing Challenges in Microbiome DNA Analysis

Among the very many “-omes” now studied and discussed (1), microbiomes have received increasing attention in recent months, from both scientists and the general public. Used to describe the communities of microorganisms and their genes in a particular environment, including a body or part of a body, “microbiome” is becoming an increasingly common term in everyday language. One challenge in microbiome genome analysis is addressing the presence of host DNA in samples. As such, improved methods for solving this problem are needed.

Fiona Stewart, Ph.D. and Erbay Yigit, Ph.D., New England Biolabs, Inc.

Introduction

A wealth of information about the composition of, and interactions between, the constituent microbes of a microbiome can provide insight into both the function and dysfunction of the host organism, as well as the host-microbiome unit as a whole. In particular, the relationships amongst and between resident microbes (bacteria, archaea and fungi) and their hosts have recently become the topic of fervent research; the number of microbiome research publications has been steadily increasing since 2003 (2). Such research has demonstrated that the microbiome communities of individuals are unique, as are the microbiome communities of specific sites within an individual (reviewed in 3). In humans, the number of microorganisms present is estimated to exceed the number of human cells by 10-fold (4). Studies of the human microbiome (including the Human Microbiome Project (HMP) [www.hmpdacc.org] (5), and MetaHIT, the metagenomics of the intestinal tract [www.metahit.eu] (6)) may be the best known, and have led to the understanding that the human microbiome may be critical to health and disease.

Until relatively recently, the role of the microbiome was unknown, and an organism’s microbial load was considered to be potentially nothing more than cellular “hitchhikers”, having little impact on the organism’s functioning. Now, it is understood that an organism’s microbiome can influence many processes within the host organism. Discoveries including the role of the microbiome in conditions and disease states, such as obesity, diabetes mellitus and cardiovascular disease (reviewed in 7), have led to the potential for development of microbiome-based diagnostic and therapeutic tools. Additionally, the unique nature of an individual’s microbiome has enabled matching of skin-associated bacteria, on objects such as a keyboard, to specific individuals, leading to the potential for use in forensic applications (8). It should be noted that microbiome research is not limited to humans, and research into microbiomes of non-human organisms is also increasing rapidly in environmental and agricultural areas of research (9).

Although it is still not possible to isolate and culture the vast majority of microorganisms (estimated to be over 95%), analysis of total nucleic acid from microbiome samples has enabled significant advances in the field. Furthermore, advances in sequencing technologies have enabled significant progress in microbiome nucleic acid analysis.

Current Methods of Analysis

The majority of microbiome DNA studies to date have employed 16S analysis (Figure 1). This analysis method takes advantage of the 16S rRNA gene that is specific to prokaryotes and some of the archaea and is not found in eukaryotes. 16S rRNA genes from different species have significant homology, but the gene also includes hypervariable regions that are generally speciesspecific, and are determined by the microbial composition of the community. These characteristics enable the use of universal primer pairs to amplify 16S genes from many organisms in the same PCR reaction and then, through subsequent sequencing of the PCR products, the individual species represented can be identified.

Figure 1. Microbiome DNA Analysis Methods

While 16S analysis is fast and inexpensive, it provides little information regarding function. More detailed information can be obtained through microbiome sequencing, particularly once host DNA is removed.
* For many samples, host DNA constitutes a high percentage of sequence reads. Removal of host DNA, and enrichment of microbial DNA substantially increases the percentage of sequence reads from the microbial sequences of interest.

While the 16S method is a fast and relatively inexpensive way to survey, at high throughput, the microbial organisms present within a sample, it provides very little information regarding function. Additionally, determining optimal PCR primers (for specific sample types and to distinguish between some species) can be challenging. In contrast, sequencing of the total DNA of a microbiome sample does not have these limitations and provides a more complex range of information. Through the identification of microbial sequences, genes, variants and polymorphisms, this method enables determination of information on microbiome species diversity and, also, putative functional information. Such sequencing-based studies have enabled the creation of many databases, including the Human Oral Microbiome Database (HOMD) [www.homd.org] (10). Approximately 700 prokaryotic species are present in the human oral cavity, and the stated goal of the HOMD database project is to provide taxonomic and genomic information on these species. Comparison of microbiome sample sequences to databases, such as HOMD, further enables discovery, including genes, pathways and their relative frequencies in the sample.

Overcoming Difficulties with Microbiome Samples
Many microbiome samples are overwhelmed with host DNA, and the HMP has reported especially high levels of human DNA in soft tissue samples, such as mid-vagina and throat samples. Saliva samples also contain high levels of human DNA (11). In contrast, although human DNA is generally all but absent from fecal samples, some infections can substantially increase the level of human DNA in such samples, likely due to widespread cell lysis during bacterial infection.

The presence of contaminating host genomic DNA in a microbiome sample complicates the genetic analysis of these samples. Since a single human cell contains approximately 1,000 times more DNA than a single bacterial cell (approximately 6 billion bp versus 4-5 million bp), even a low level of human cell contamination within a microbiome sample can substantially complicate the sample processing and sequencing. As a result, in the case of total microbiome DNA sequencing studies, only a small percentage of sequencing reads from such samples pertain to the microbes of interest, and therefore a large percentage of sequencing reads (host) have to be discarded. Consequently, obtaining sufficient sequence coverage of the microbiome DNA can become costprohibitive or even technically infeasible. Therefore, methods to enrich microbiome DNA are useful, and, in some cases, critical for sequencing of the microbiome. However, until now, options for such enrichment have been limited to selective cell lysis, with the disadvantages of a requirement for live cells, and low bacterial DNA recovery.

The NEBNext® Solution

The NEBNext Microbiome DNA Enrichment Kit addresses this problem by providing a quick and effective way to remove contaminating host DNA, thereby enriching for microbiome DNA. The kit exploits the different prevalences of CpG methylation in the genomes of microbial and eukaryotic organisms. Eukaryotic DNA, including human DNA, is methylated at CpGs, while methylation at CpG sites in microbial species is rare.

The NEBNext Microbiome DNA Enrichment Kit uses a magnetic bead-based method to selectively bind and remove CpG-methylated host DNA. feature article continued… The kit contains the MBD2-Fc protein, which is composed of the methylated CpG-specific binding protein MBD2, fused to the Fc fragment of human IgG. The Fc fragment binds readily to Protein A, enabling effective attachment to Protein A-bound magnetic beads. The MBD2 domain of this protein binds specifically and tightly to CpG methylated DNA. Application of a magnetic field then pulls out the CpG-methylated (eukaryotic) DNA, leaving the non-CpG-methylated (microbial) DNA in the supernatant. 

Microbiome Enrichment of Human Saliva

Human saliva samples can be especially challenging, due to high levels of human genomic DNA and the poor-quality of the DNA itself. Despite these sample challenges, the data shown in Figure 2 demonstrates that substantial enrichment of microbiome DNA from saliva was achieved using the NEBNext Microbiome DNA Enrichment Kit.

Figure 2. Salivary Microbiome DNA Enrichment

DNA was purified from pooled human saliva DNA (Innovative Research) and enriched using the NEBNext Microbiome DNA Enrichment Kit. Libraries were prepared from unenriched and enriched samples and sequenced on the SOLiD 4 platform. The graph shows percentages of 500M-537M SOLiD4 50 bp reads that mapped to either the Human reference sequence (hg19) or to a microbe listed in the Human Oral Microbiome Database (HOMD)[10]. (Because the HOMD collection is not comprehensive, ~80% of reads in the enriched samples do not map to either database.) Reads were mapped using Bowtie 0.12.7[13] with typical settings (2 mismatches in a 28 bp seed region, etc.).

Figure 3. Microbiome Diversity is Retained After Enrichment with the NEBNext Microbiome DNA Enrichment Kit

DNA was purified from pooled human saliva DNA (Innovative Research) and enriched using the NEBNext Microbiome DNA Enrichment Kit. Libraries were prepared from unenriched and enriched samples, followed by sequencing on the SOLiD4 platform. The graph shows a comparison between relative abundance of each bacterial species listed in HOMD[10] before and after enrichment with the NEBNext Microbiome DNA Enrichment Kit. Abundance is inferred from the number of reads mapping to each species as a percentage of all reads mapping to HOMD. High concordance continues even to very low abundance species (inset). We compared 501M 50 bp SOLiD4 reads in the enriched dataset to 537M 50 bp SOLiD4 reads in the unenriched dataset. Reads were mapped using Bowtie 0.12.7[13] with typical settings (2 mismatches in a 28 bp seed region, etc).
* Neisseria flavescens – This organism may have unusual methylation density, allowing it to bind the enriching beads at a low level. Other Neisseria species (N. mucosa, N. sicca and N. elognata) are represented, but do not exhibit this anomalous enrichment.
An important consideration when assessing the validity of microbiome enrichment is that the enrichment should not be biased, and the diversity of microbiome organisms in the sample should remain intact after enrichment. As shown in Figure 3, measurement of the relative abundance of species represented in HOMD was equivalent between unenriched and enriched samples. Interestingly, Neisseria flavescens, highlighted with *, was a unique outlier in this comparison and may have unusual methylation density, which enables binding to the MBD-Fc beads at a low level. It is notable that other Neisseria species (N. mucosa, N. sicca and N. elognata) are also represented, but do not exhibit this anomalous enrichment.

Conclusion

From forensic microbial “fingerprints” to disease-causing pathogens, microbiomes comprise a vast and varied microcosm with a surprising degree of influence over the health and function of the host organism. The potential for significant and exciting discoveries to be achieved with microbiome analysis is enormous, but will require improved tools and methods to make this a reality. As a step towards this goal, the NEBNext Microbiome DNA Enrichment Kit now makes it possible to substantially enrich a variety of sample types for non-host, microbial DNA, while retaining microbial diversity, and thereby improving the quality and cost-effectiveness of downstream analyses and data generation.

References

  1. Alphabetically ordered list of -omes and -omics (2013) Omics. org Retrieved on May 1, 2013, from www.omics.org.
  2. Jones, S. (2013) Nature Biotechnology, 31, 277.
  3. Morgan, X.C., et al. (2013) Trends in Genetics, 29, 51–58.
  4. Backhed, F., et al. (2005) Science, 307, 1915–1920.
  5. Peterson, J., et al. (2009) Genome Res. 19, 2317–2323.
  6. Qin, J., et al. (2010) Nature, 464, 59–65.
  7. Pflughoeft, K.J. and Versalovic J. (2012) Annu. Rev. of Pathol. 7, 99–122.
  8. Fierer N., et al. (2010) Proc. Natl. Acad. Sci. USA, 107, 6477–6481.
  9. Jansson, J.K. and Prosser, J. I. (2013) Nature, 494, 40–41.
  10. Chen, T., et al. (2010) The Human Oral Microbiome Database Retrieved on May 1, 2013, from www.homd.org.
  11. The Human Microbiome Project Consortium (2012) Nature, 486, 215–221.
  12. Langmead, B., et al. (2009). Genome Biol. 10(3), R25.

Scientific Contribution

The scientific contributors to this article include: George R. Feehery, Erbay Yigit, Bradley W. Langhorst, Fiona J. Stewart, Eileen T. Dimalanta, Sriharsa Pradhan, James MacFarland, Christine Sumner and Theodore B. Davis.