NEW YORK (GenomeWeb) – A team from Wayne State University and Michigan's Isle Royale National Park have demonstrated the feasibility of using Ion Torrent sequencing to assess microbial community samples being introduced into the environment.
"A lot of people struggled when Ion Torrent came out," the study's first author Masanori Fujimoto, a physiology researcher at Wayne State, told In Sequence. "But this 400 [base read] chemistry can do a much better job and at low cost."
"This is going to accelerate the study of the microbiome," he added.
For their study, published in PLOS One this month, Fujimoto and his colleagues focused on samples of ballast water released from cargo ships in the Great Lakes that had been treated with sodium hydroxide, using Thermo Fisher Scientific's Ion Torrent PGM instrument to do 16S ribosomal RNA sequencing on dozens of samples from four ballast tanks that had or had not been treated with sodium hydroxide.
The team's sequencing results revealed microbial community members that matched those detected by 16S gene sequencing with the Roche 454 GS FLX instrument. In particular, the results revealed lower-than-usual microbial diversity in alkali ballast water and a rise in members from a single genus called Alishewanella.
In the past, sequencing approaches for identifying organisms in an environmental microbial community based on their 16S ribosomal RNA gene were expensive and labor intensive, Fujimoto explained. The cost has declined and throughput has increased with the introduction of next-generation sequencing technologies, he added, though the price tag can still be substantial when considering many samples.
Consequently, the team decided to look at the applicability of a relatively inexpensive sequencing technology — the Ion Torrent PGM — for identifying microbial community components that came as part of a broader study looking at the effects of treating ballast water with sodium hydroxide.
The water, which is taken in to steady an empty cargo ship and released into the environment when the ship reaches port to reload, has the potential to transmit non-native species and potentially pathogenic microbes from one region to another.
Several different treatment schemes have been proposed to combat that — from filtering and ultraviolet light applications to chlorine or sodium hydroxide treatment. Ideally, researchers want to understand the consequences of such treatments to help in developing regulations that set out effective water treatments that can be performed in a cost-effective manner, Fujimoto said.
"We want to treat the water efficiently and minimize introductions of non-native species," he explained, "but at the same time we don't want to put a lot of burden on the ship owners and industries."
To take a closer look at both the alkali treatment of ballast water and the effectiveness of assessing microbial community membership with reads produced on the Ion Torrent PGM, the researchers used a 400-base Ion Torrent chemistry launched last year to do 16S rRNA sequencing on intake and discharge samples from two alkali (sodium hydroxide-treated) ballast tanks and two untreated control tanks.
Their Ion Torrent experiments targeted V4 and V5 regions of the microbial 16S rRNA gene using primers developed to be compatible with that sequencing system.
Because DNA from cells from deceased microbes and plants sometimes still turn up in water samples, the team subjected a duplicate sample from each condition to treatment with a DNA cross-linking agent that can't cross the membrane of living bacteria.
All told, the researchers sequenced almost three-dozen barcoded samples and bacterial controls in two runs on the Ion Torrent PGM and 314 chip, generating more than 580,600 quality Ion Torrent reads that averaged around 380 bases apiece.
As they pored through this data — which included almost 18,800 reads per sample, on average — they turned to available alignment mapping methods, reference information from the Ribosomal Database Project (RDP), and the well-established QIIME pipeline to identify members of each microbial community and determine their diversity.
Those analytical approaches took some tweaking, Fujimoto noted, since the bioinformatics pipelines available for dealing with 16S sequence data when the study was done had been designed to deal with reads generated on Roche 454 or Illumina instruments.
Although some analytical methods have since been updated for Ion Torrent read inputs, the informatics aspect of the project was "a major challenge" at the time, he said. "We had to find a way to mount this output [raw read] data into those existing pipelines."
For example, Fujimoto explained, the team had to find ways of converting the quality score associated with Ion Torrent sequence reads into a score recognized by the analytical pipelines. Additional modifications were also needed to deal with subtle differences in the types of barcodes used for Ion Torrent sequencing.
In samples with known bacterial species, such as a control sample comprised of Vibrio cholera from a pure culture, the team saw authentic V. cholera reads in 99.94 percent of reads from the first run and 99.82 percent of reads from the second.
When they compared microbial communities in the ballast water samples before and after treatment, the researchers found that samples coming out of the untreated ballast water tanks contained bacteria from somewhat more operational taxa units than did the samples collected at intake.
On the other hand, the alkali ballast water samples were marked by significantly lower microbial diversity and a dramatic rise in reads associated with bacteria from the Alishewanella genus, which comprised more than half of the reads found in post-sodium hydroxide-treated ballast water samples.
A closer look at the phylogeny of the latter microbes revealed at least three Alishewanella species not described in the past in the alkali water samples.
The treated samples also tended to carry sequences from Pseudomonas and other gamma-proteobacteria, as well as Bacillus and other genera from the Firmicutes phylum, the team reported. Neither intake nor discharge samples seemed to contain sequences from potential pathogens in the Escherichia, Enterococcus, or Vibrio genera.
Meanwhile, a look at the samples treated with DNA cross-linker suggested that the intake water was prone to containing dead cyanobacterial cells containing chloroplast DNA that could be picked up by 16S rRNA sequencing.
As a comparison of the sequencing technology used to produce these microbiome profiles, the researchers tested a subset of their original sodium hydroxide treated and untreated ballast water samples by 16S rRNA sequencing on the Roche 454 GS FLX instrument, this time amplifying the V1 and V2 region of the 16S gene.
Results from that analysis corresponded well with findings from the Ion Torrent-based look at the pre- and post-treatment ballast water samples, prompting enthusiasm about the prospect of using the approach to assess microbial representation at these and other environmental sites.
"They reported almost identical communities, even though the regions are different within the 16S [gene]," Fujimoto said. "The output we got, the community profile we got, was almost identical."
"This study showed the efficacy of alkali ballast water treatment in reducing ballast water microbial diversity," he and his colleagues wrote, "and demonstrated the application of new Ion Torrent sequencing techniques to microbial community studies."
The current analysis was done using the 314 chip, which produces roughly 1 million reads per run.
Ion Torrent has since introduced 316 and 318 chips that generate up to 10 million reads in an Ion Torrent run, Fujimoto noted, explaining that the final coverage depth per sample depends on both the chip selected and the number of barcoded samples included in a given run.
For the 16S sequencing experiments described in their PLOS One study, the group believes they reached sequencing saturation, detecting the vast majority of microbes present in a given ballast water sample.
Still, Fujimoto noted that one could use the same primer set to do 16S rRNA gene sequencing with the 316 or 318 chips, which would offer an edge in situations where reads from relatively rare organisms are missed due to a lack of coverage or when looking at larger sets of barcoded samples in a single run.
At the moment, he and his colleagues are spending roughly $700 to $900 per Ion Torrent run, including reagent costs. In contrast, the same experiments on a Roche 454 instrument would be expected to cost a few thousand dollars.
The team is continuing to use Ion Torrent instruments for 16S rRNA sequencing efforts looking at the microbial community members found before and after chlorine treatment of ballast water.
The researchers are also performing 18S gene sequencing experiments on samples collected from Toledo Bay in the Great Lakes as part of another project focused on finding non-native eukaryotic species in that area. In addition, they plan to use an Ion Torrent instrument to sequence DNA metabarcodes in environmental DNA, or eDNA, samples as part of a separate effort to study aquatic, invasive fish species.
"This application of next-generation sequencing techniques to ballast water treatments is but one example of how we intend to apply high-throughput sequencing of environmental DNA to gain a better understanding of the organisms in aquatic environments and how they vary with location, season, and anthropogenic impacts like invasive species," the study's senior author Jeffrey Ram, also with Wayne State University, said in an email message.