February 1, 2014 (Vol. 34, No. 3)

Fiona Stewart, Ph.D. Product Marketing Manager New England Biolabs
Erbay Yigit, Ph.D. Applications and Product Development Scientist New England Biolabs
George R. Feehery Research Associate New England Biolabs

Separation by Differential Methylation Density Reduces Whole Microbiome DNA Sequencing Cost

It is now understood that considerable insight into both the function and dysfunction of humans, plants, and other organisms can be gained by studying the composition and characteristics of the distinct microbial populations that colonize many bodily regions and fluids. However, despite the large numbers of microbes present, microbial genomes are so diminutive that recovery of sufficient microbial genomic material from many critical sample types is highly challenging.

The majority of microbiome DNA studies to date have employed a somewhat restrictive strategy: amplifying only 16S rRNA genes for DNA sequencing. This analysis method exploits the 16S rRNA gene, which is not found in eukaryotes, but rather only in prokaryotes and some archaea. As 16S rRNA genes include species-specific hyper-variable regions, they can be amplified and sequenced to enable identification of microbial species.

More extensive information can be obtained by a nonrestrictive strategy that allows sequencing of total microbiome DNA, not just 16S rRNA genes. Total microbiome DNA sequencing provides an extensive view of microbial sequences, genes, variants, and polymorphisms that allow identification of both species diversity and putative functional information.

However, many types of microbiome samples, including saliva and soft tissue samples from the throat or mid-vagina, contain very high ratios of host-to-microbial DNA (as extreme as 99 to 1), which impede genomic analysis of the microbial component: Since only a small percentage of sequencing reads from such samples pertain to the microbes of interest, a high percentage of sequencing reads (from the host) have to be discarded. Obtaining sufficient sequence coverage of microbiome DNA can therefore become cost-prohibitive or even technically infeasible.


Selective Enrichment by MBD2-Fc

Eukaryotic DNA, including human DNA, is methylated at CpG sites, while methylation at CpG sites in microbial species is rare. New England Biolabs’ NEBNext® Microbiome DNA Enrichment Kit exploits this differential methylation density to selectively separate host and microbial DNA in a one-hour protocol.

This technique utilizes 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 paramagnetic beads.

The MBD2 domain of this protein binds specifically and tightly to DNA fragments with sufficient CpG methylation density (measured with 20 kb fragments to be greater than three sites per kilobase). This affinity allows removal of up to 98% of methylated host DNA—for example human DNA in which 4–6% of all the cytosines are methylated (60–90% of which is at CpG sites)—from a microbial DNA pool that either is not or is partially CpG methylated.


Salivary Microbiome DNA Enrichment

DNA was purified from pooled human saliva DNA (Innovative Research), and genomic DNA quality and quantity were assayed by agarose gel electrophoresis and by Nanodrop® spectrophotometry.

MBD2-Fc beads were prepared by mixing MBD2-Fc protein and paramagnetic beads, and then added to the saliva sample DNA. For optimum activity, a ratio of 1 μL of MBD2-Fc beads per 6.25 ng DNA was used.

After incubation for 15 minutes at room temperature with gentle rotation, the vial was transferred to a magnetic rack for 2–5 minutes until the beads collected to the wall of the tube and the supernatant clarified. The supernatant fraction, containing enriched microbial DNA (Figure 1), was removed to a clean microcentrifuge tube without disturbing the MBD-Fc beads and immediately purified using Agencourt AMPure® XP beads (Beckman Coulter) according to the manufacturer’s instructions, although ethanol precipitation may also be used.

Libraries were made from the enriched sample, and sequenced using SOLiD™4 chemistry. The 500 to 537 million 50 bp reads were then mapped, using Bowtie 0.12.7 at typical settings (i.e., 2 mismatches in a 28 bp seed region), to either the human reference sequence (hg19) or microbes listed in the Human Oral Microbiome Database (HOMD).

Although SOLiD sequencing was performed for this experiment, enriched samples are also suitable for library preparation for Illumina®, Ion Torrent™, or 454™ sequencing platforms; endpoint and real-time PCR assays; and DNA microarray analysis.


Figure 1. A schematic diagram of microbiome DNA enrichment using NEBNext MBD2-Fc protein

Input DNA and Factors Affecting Efficiency

Any method of preparing protein-free genomic DNA can be used, although procedures that would cause DNA to shear should be avoided. The efficiency of microbiome DNA separation declines with lower DNA quality, smaller fragment size (<15 Kb), and decreased overall quantity of CpG methylated dinucleotides.


Results

After enrichment, 94–96% of reads that aligned to the human reference genome were depleted, and reads mapping to the HOMD database (Figure 2) increased eightfold. This robust enrichment allows for economical sequencing of total microbiome DNA.

Preservation of microbiome diversity in the enriched sample is critical to an enrichment method’s utility. Measurement of the relative abundance of species represented in HOMD demonstrated equivalence between un-enriched and enriched saliva samples (Figure 3). High concordance continued even to very low abundance species (Figure 3, inset).


Figure 2. DNA was purified from pooled human saliva and enriched using the NEBNext Microbiome DNA Enrichment Kit followed by sequencing on the SOLiD 4 platform. The graph shows percentages of reads that mapped to either the human reference sequence (green) or to a microbe listed in the Human Oral Microbiome Database (purple).

In the enriched samples, it is noted that approximately 80% of reads did not map to the hg19 or HOMD databases. Although the HOMD’s goal is to provide taxonomic and genomics information on the approximately 700 prokaryotic species present in the human oral cavity, currently its database includes genomes for 315 of these species (46% of the taxa registered in HOMD).

Methods for applying the NEBNext Microbiome DNA Enrichment Kit for samples such as human blood, mock malaria-infected human blood, as well as nonhuman samples, including fish, can be found in a study published October 28, 2013, in PLoS ONE that is titled, “A Method for Selectively Enriching Microbial DNA from Contaminating Vertebrate Host DNA”.


Figure 3. Maintenance of bacterial species’ relative abundance before (vertical axis) and after (horizontal axis) enrichment with the NEBNext Microbiome DNA Enrichment kit is inferred from the number of the 500 to 537 million SOLiD 4 50 bp reads that mapped to each species as a percentage of all reads mapping to the HOMD. The asterisked (*) species, Niesseria flavescens, may have unusual methylation density, allowing it to bind to the enriching beads at a low level. Other Niesseria species are represented, but they do not exhibit this anomalous enrichment.

Host DNA Capture

If desired, the host DNA captured in the bead fraction can be eluted using Proteinase K. To evaluate separation specificity in the saliva sample experiment, the bead-bound fraction of DNA was also sequenced. A vast majority (99.3%) of the mapped reads aligned to the human reference genome.


Organellar DNA Enrichment

Beyond microbial DNA analysis, separation on the basis of differential methylation status also enables isolation of various organelles’ DNA. For example, while plant genomic DNA is bound readily by MBD2-Fc, chloroplast DNA is not bound and remains in solution. Similarly, due to human and plant mitochondrial DNA’s limited CpG methylation, this method can successfully enrich such DNA.
























Fiona Stewart, Ph.D.(stewart@neb.com), is a product marketing manager, Erbay Yigit, Ph.D., is an applications and product development scientist, and George R. Feehery is a research associate at New England Biolabs.