Insights from bacteria-infecting viruses could vastly improve genome editing in diverse microbes, including those of the human microbiome
By Gabriel Filsinger and Benjamin Boettner
(BOSTON) – The number of bacterial cells inside each person is approximately equal to the number of human cells in the body. Interestingly, how these bacteria contribute to our health or to disease is still largely a mystery. Research on the human microbiome, the bacterial collective inhabiting our gut and other parts of our bodies, has shown that these bacteria can have a dramatic influence on many aspects of human health through interactions with the immune system, the brain, and during development. To better understand how specific bacterial strains in our complex microbiomes use particular nutrients, produce metabolites that benefit or harm our organ systems, and interact with each other and the body’s surfaces, synthetic biologists are trying to identify the responsible genetic elements in their genomes and figure out how these contribute to differences between people who are healthy and those with disease. To do this systematically, they would like to be able to precisely edit bacterial genomes at specific positions in a wide variety of strains.
Microbiome engineering: limits and opportunities
The almost-ubiquitous CRISPR-Cas genome editing tool has been challenging to use broadly in the bacterial kingdom since the Cas9 enzyme kills most bacteria rather than that it stimulating genome editing. Current microbiome researchers are forced to resort to other techniques, including “recombineering” (recombination-mediated genetic engineering) as their method of choice. Recombineering allows the precise editing of DNA sequences in bacterial chromosomes without damaging them (there are no DNA cuts or breaks, and the edits are scarless). Current recombineering tools are borrowed from bacterial viruses called bacteriophages (or phages) that specialize on a small number of bacterial strains but cannot infect others. Phages use their so-called “single-stranded DNA annealing proteins” (SSAPs) to replicate and recombine their genomes inside these bacteria. This provided an opportunity for researchers, who in turn have used SSAPs to carry out their own intended reactions – reactions which enable genomic edits of any size and at any location of the genome.
A major barrier to researchers’ ability to perform recombineering in most strains of the microbiome is the fact that SSAPs, like the phages that produce them, only efficiently function in one or a limited number of bacterial strain(s), and that appropriate phage-derived SSAP tools are not available in most bacterial strains.
“Ideally, we would like to be able to have a collection of SSAP systems at our disposal, each one for a broad spectrum of bacterial strains. To achieve this level of broad utility, we wanted to first understand what makes SSAPs efficient or inefficient in a particular host strain,” said Gabriel Filsinger, co-first author of a new study investigating this problem. “We hypothesized that getting a grasp of how SSAPs interact with other DNA-binding proteins during the recombination process could tell us what limits their use and open the door to better recombineering systems.” Filsinger works as a graduate student with Wyss Institute Core Faculty member George Church, Ph.D., in the Institute’s Synthetic Biology platform. Soon after joining Church’s group, he became interested in approaching genome engineering from an angle different from CRISPR-Cas technology. After exploring other ways to engineer bacterial genomes, he turned his attention to recombineering.
Charting a new path
In their study, which was published in Nature Chemical Biology, the team focused on the well understood laboratory strain, E. coli, as well as a group of microbiome bacteria called “lactic acid bacteria,” which includes strains like Lactococcus lactis, and Lactobacillus rhamnosus that have been used as proof-of-principle microbiome therapeutics.
They first developed an in vitro approach that allowed them to investigate what bacterial strain-specific proteins phage SSAPs need to interact with in order to allow added single-stranded DNA to anneal to bacterial DNA, which is a key step in the recombineering process. The bacterial genome transits through a single-stranded state when bacteria divide and duplicate their DNA. This is also when SSAPs get access to the DNA of bacterial hosts, and when recombineering happens.
“We found that SSAPs require a specific interaction with certain bacterial proteins called single-strand binding proteins (SSBs) during genome editing, which restricts the range of species where they function, since SSBs are strain specific. Our characterization of this phage-bacterial protein-protein interaction provides a framework for predicting in which species these proteins can be expected to operate, and makes expanding these methods to new species less of a black box,” explained Filsinger. Using their assays, the researchers identified a short motif encompassing merely seven of an SSB’s amino acid building blocks that decides whether a particular SSAP can bind to it and enable recombineering. In this way, this motif functions as a zip code that varies across strains.
Investigating their insights across a variety of different bacterial species, the researchers co-expressed SSAPs with compatible SSB protein partners, rather than the SSAPs alone as was done previously. “In a number of species, SSAP-SSB co-expression improved portability, resulting in increased recombineering efficiency and reduced toxicity. Our work helps establish the co-expression of SSAPs with their cognate SSBs as a new method that can be leveraged for recombineering in diverse bacterial species,” said the other co-first author Timothy Wannier, Ph.D., a Postdoctoral Fellow on Church’s team.
“Beyond investigating the human microbiome, our findings could be relevant for the engineering of bacteria to be deployed as living therapeutics or diagnostics in humans, could be used to speed up the study of human pathogenic bacteria, and should contribute to the advancement of microbial engineering at large, providing benefits to a host of broad synthetic biology applications,” added Wannier. For example, L. lactis, one of the strains that the researchers studied, is widely used in the dairy industry to produce buttermilk and many cheeses, and thus is “generally regarded as safe” by the FDA. The strain has probiotic potential within the microbiome and has even been engineered as a delivery vehicle for therapeutic proteins for the treatment of inflammatory bowel disease and cancer. The team’s new insights could greatly facilitate such strategies and help them to be extended to other species.
Eying the future
Putting their recombineering results to use, Filsinger, Wannier and their co-workers in the Church lab are now working towards efficient homologous recombination methods for many more bacterial species by identifying a catalog of SSAPs and SSAP-SSB pairs that work in each of them. They also set out to build a large library of SSAP proteins and are screening them to find SSAP variants that function optimally across broad groups of bacterial species.
“Researchers have tried for a long time to port homologous recombination proteins between species with limited success. This also includes movement of proteins between bacteria and eukaryotes such as yeast, mammals and plants. Our work highlights the importance of SSBs as a key node affecting homologous recombination and recombineering, and we are very interested to see if new homologous recombination methods can be established by leveraging the coordination between canonical homologous recombination proteins and SSBs,” said Church, who also is Professor of Genetics at Harvard Medical School and of Health Sciences and Technology at Harvard and the Massachusetts Institute of Technology (MIT).
Other authors on the study were past and present members of the Church lab, including, Helene Kuchwara, Verena Volf, Stan Wang, Xavier Rios, Christopher Gregg, Marc Lajoie, Seth Shipman, and John Aach; as well as Felix Pedersen at the University of Southern Denmark, Odense; Isaac Lutz at the University of Washington, Seattle; Julie Zhang, Kevin Gozzi and Michael Laub at MIT; Devon Stork at Harvard University; and Anik Debnath at Tenza Inc., Cambridge, MA. The work was funded by the National Institute of General Medical Sciences under grant# 1U01GM110714-01, the Department of Energy under grant# DE-FG02-02ER63445, the National institutes of Health under grant# R01GM082899; a National Science Foundation Graduate Research Fellowship under grant# DGE1745303; a Landry Cancer Biology Research Fellowship; and a National Science Foundation Graduate Research Fellowship under grant# DGE1745302.