Researchers at the J. Craig Venter Institute have developed an automated approach for identifying and sequencing single bacterial cells from a complex environmental sample. The team tested their method on a biofilm sample from a hospital sink drain in order to try and characterize the pathogens present.
"There's a huge problem with these so-called superbugs," Roger Lasken, director of single-cell genomics at JCVI, told In Sequence. "We want to sequence the genomes of those bacteria … but only a small minority of the bacteria can be cultured," he said.
Additionally, even when some of the bacteria in a mixed sample can be cultured, the fast growing strains end up taking over the Petri dish, Lasken explained, and those aren't necessarily the most interesting strains.
Instead, the team demonstrated in a proof-of-principle study published earlier this month in Genome Research that it could use single-cell sequencing techniques to capture the complete diversity of bacteria from a mixed sample.
The team collected bacteria from a biofilm sample from a hospital sink drain, which contained many different bacterial species. Biofilms are a sticky film of polysaccharides made by bacteria. The researchers wanted to examine bacteria species from these films because they can grow on many surfaces, including sink drains and catheters, and can cause problems in hospital settings, Lasken said. Additionally, many different species can be present in the biofilm, including both pathogenic and benign species.
The method developed by the JCVI team relies on four key steps.
First, using fluorescence-activated cell-sorting, cells are isolated and distributed into a 384-well plate. Next, a robotic platform lyses the cells and amplifies the DNA using standard multiple displacement amplification technology. Third, PCR and 16S rRNA sequencing is done to identify the individual species. And finally, candidate amplified genomes are then selected for further study and whole-genome sequencing.
The team tested their protocol on a sample from a hospital sink drain in a restroom adjacent to the emergency room. The team was able to isolate and amplify 416 single cells from the sample. After performing 16S rRNA sequencing to determine the taxonomy present, they homed in on 78 candidate species. Of those, they decided to focus on 18 that had been reported as potentially pathogenic or commensal species.
Next, the researchers used Roche's 454 GS FLX to do shallow, whole-genome sequencing of all 18 candidates. They barcoded the strains, generating between 5,500 and 13,500 reads per strain, with an average read length of 321 bases.
They assembled the 454 reads and determined which genes were captured from the species and also used 454 sequencing as a quality control metric to determine which cells would be amenable to deeper sequencing.
From those 18 strains, they narrowed down their list to three Porphyromonas gingivalis strains because each had more than one MDA product representing its genome and they passed all the quality criteria.
The team then sequenced these three strains on the Illumina Genome Analyzer. Prior to the study, only three P. gingivalis strains had been sequenced, all of which came from cultured bacterial samples, making these the first that came from uncultured cells found in an environmental sample, Lasken said.
"This is the first time anyone has seen this bacteria in the environment," he said.
Due to biases inherent in whole-genome amplification technology, genome coverage varied widely between the three genomes, from 41 percent of the genome covered to 91 percent of the genome covered.
After sequencing, each of the three genomes were assembled de novo using an assembly method designed specifically for single-cell genomes, known as SPAdes. The method was published by Pavel Pevzner's group at the University of California, San Diego in the Journal of Computational Biology last year and takes into account the uneven genome coverage due to MDA.
SPAdes "takes into account that some regions of the genome are over-represented and some regions are under-represented," Lasken said. Most assemblers were designed for bulk sequencing, which has more even coverage than single-cell sequencing, and they throw away reads from regions of low-coverage, assuming that they are errors. However, because single-cell sequencing results in uneven coverage, some correct reads may have low coverage, while some incorrect reads may have high coverage. Thus, SPAdes does not use coverage as a metric for deciding which reads to keep and which to throw out, helping to produce a more complete assembly.
The data from the 454 sequencing, aside from serving as a quality control metric, can also be used to help assemble the genomes, Lasken added. When assembling genomes, particularly from single-cell sequencing, Lasken said there are often "orphan contigs," and it is unclear where those contigs fit into the assembly. "But if you have a longer 454 read, it might help you join them," he said.
The three bacterial strains from the sink drain were all very similar to each other, Lasken said, with "almost identical sequences." The three genomes shared 847 SNPs and 75 deletions and insertions.
While the three genomes were similar to each other, there were over 500 differences from the previously sequenced P. gingivalis genomes, Lasken said. "We're looking in there now to see what genes matter," including genes and virulence factors that could affect how the bacteria infect patients, survive on a sink drain, or may be responsible for antibiotic resistance.
The team is now continuing to analyze the P. gingivalis genomes. Additionally, the group has now amplified around 10,000 single-cell genomes and used 16S rRNA to classify them, and they will decide which of those to further study with whole-genome sequencing.
Lasken said he hopes to expand this study even further, examining even more biofilms from hospital settings. He said his laboratory is well-equipped to handle that many samples, with a robotic system that can process 5,000 cells per week.
"We're now in a position to study pathogens in a hospital [in a way] that's never been done before," he said. "This is a powerful tool to analyze strains of bacteria in biofilms and determine which ones are involved in disease."
Such studies will be important as bacteria continue to evolve and develop resistance to drugs. "The superbugs are a real crisis," he said, and "this is an important technology that will open up a lot of new fronts for study."