By encapsulating single organisms from microbiome samples in gel microdroplets and then culturing them to form microcolonies, researchers from the Los Alamos National Laboratory have been able to sequence the genomes of individual bacterial species with much higher coverage and lower bias than possible using whole-genome amplification methods starting with single cells.
The group published a description of its approach in Genome Research this month, showing that the method could be used to assemble near-complete genomes from single organisms in human oral and stool samples.
Using microdroplet culturing to grow droplet-contained, isolated families of genetically identical cells before sequencing, the team was able to build almost full genomes for two bacterial species and also measure significant diversity between individual organisms from the same sample.
Cliff Han, the study's lead author, told In Sequence that while he and his colleagues tested the method using human microbiome samples in the study, the approach could also be useful for profiling genome diversity in environmental samples — something his group and others the team is now collaborating with are interested in.
While shotgun sequencing has allowed researchers to construct genomes for members of several simple microbial communities, these genomes represent a consensus derived from close species or strains whose individual genomes can differ slightly or significantly amongst them.
Meanwhile, according to the Los Alamos group, single-cell sequencing using whole-genome amplification techniques have allowed researchers to sequence and assemble significant portions of genomes for individual organisms within complex environmental communities, but these methods suffer from bias and a loss of specificity due to amplification from a single chromosome copy. This has made it nearly impossible to assemble a finished genome from a single cell thus far.
Han said his team was initially interested in single-cell sequencing for research into soil communities in the New Mexico and Utah region, but realized that it would be impossible to create complete genome sequences for individual bacteria.
"Consistently, we can get only about 40 percent to 50 percent of the genome even with lab strains using single-cell sequencing," he said. "And with environmental bacteria it's even worse. So we banged our heads on the wall for a moment, and decided we need a new strategy."
According to Han, the group was already looking into gel microdroplet culturing for separate projects to culture bacteria from different environments.
Using permeable gel microdroplets, it is possible to isolate each individual organism within its own sphere, while still maintaining signaling and communication between different cells and species in the community, which many species require to survive and grow in culture, Han said.
In the study published this month, Han and his colleagues compared sequencing results from a single-cell approach with results using the gel microdroplet culturing method before amplification and sequencing.
The team used a CellSys microdrop maker and a growth assay protocol from One Cell Systems to encapsulate and culture single cells from oral and fecal test samples.
Han said that the process of creating the microdroplets relies on random distribution to ensure that the vast majority of droplets contain just one organism.
"We know with about a one [milliliter] gel solution, we'll generate about 20 million to 40 million gel droplets, so if you add several million bacterial cells in it you will get mostly one cell in one droplet. But, it's not that accurate so we generally make several concentrations or several batches to do the experiment," he said.
In the study, after separating and sorting the droplets and the comparable single cells, the researchers amplified DNA from both and sequenced the samples on the Illumina HiSeq 2000 using a paired-end 100-cycle run resulting in 100-bp reads.
According to the study authors, all the oral samples mapped to Streptococcus species, and all fecal samples to Enterococcus.
According to the team, reads from the GMD samples showed more even coverage across the genome overall than those from single cells, and resulted in larger assemblies and contigs. By mapping all reads to what the group determined to be the best assembly for each organism, the researchers assessed the relative completeness of assembly from GMD samples compared to single cells.
The group found that Streptococcus droplets showed a mean genome recovery more than twice that of single cells, with 99.96 percent versus 47.23 percent, respectively. With Enterococcus, the GMDs all showed exactly 100 percent genome recovery, which was also about twice as high as for single cells.
To test whether the genomes assembled from the microdroplets were close to complete, the team compared the assembly sizes with sizes of published genomes from related species. According to the authors, the sizes did compare fairly closely, "in line" with Streptococcus oralis and S. mitis, and "a little below" E. faecium genomes.
The team also sequenced E. coli samples cultured in microdropets as a control to measure the genome recovery rate of the method. The researchers wrote that recovery rates from single E. coli cells were around 55 percent, while for GMDs recovery rates averaged 97 percent.
Most interestingly, Han said, the results from the GMD sequencing showed significant diversity among individual cells of the same species from the same sample population. The group calculated the number of SNPs and indels, and found large numbers for the Strep samples, but relatively few for Enterococcus. For the Strep sample with the largest number of SNPs compared with the reference species, the genome differed at more than 2.5 percent of all nucleotides, according to the authors.
Within the Strep population, which the group chose based on having identical 16S rDNA sequences, there was "significant functional genetic diversity" present, and even the large diversity measured probably underestimates the actual pan-genomic diversity of the samples, the group wrote.
"When we looked and compared after we got those genomes we thought something was wrong in the sequencing," Han said. "That was quite a surprise that up to maybe close to 7 percent of those genomes were not the same. But that's really why we want to sequence the genome from a single cell to a high level," he said.
George Weinstock, associate director of Washington University's Genome Institute and a leader in human microbiome research efforts, said that the droplet method offers a "very attractive approach."
"In addition to addressing new reference genomes, it may also have applicability to just describing the structure of communities themselves, perhaps supplanting 16S sequencing or whole-genome metagenomic shotgun, each of which has its pros and cons," he said.
According to Han, the GMD method is something other groups could also adopt with the right technology to create, sort, and manipulate the droplets. His group is now working on adapting the method to work not only with wet samples, but also with dry environmental samples.
But the most obvious next potential application, he said, would be to use the approach to do in situ culture and sequencing for populations in ocean water samples.
He also said his group is now working with another research team interested in isolating and sequencing the genomes of archaea from waste water.