Metabolically-targeted dCas9 expression in bacteria

Soil microbiome characterization and its future directions with biosensing

Soil microbiome characterization is typically achieved with next-generation sequencing (NGS) techniques. Metabarcoding is very common, and meta-omics is growing in popularity. These techniques have been instrumental in microbiology, but they have limitations. They require extensive time, funding, expertise, and computing power to be effective. Moreover, these techniques are restricted to controlled laboratory conditions; they are not applicable in field settings, nor can they rapidly generate data. This hinders using NGS as an environmental monitoring tool or an in-situ checking device. Biosensing technology can be applied to soil microbiome characterization to overcome these limitations and to complement NGS techniques. Biosensing has been used in biomedical applications for decades, and many successful commercial products are on the market. Given its previous success, biosensing has much to offer soil microbiome characterization. There is a great variety of biosensors and biosensing techniques, and a few in particular are better suited for soil field studies. Aptamers are more stable than enzymes or antibodies and are more ready for field-use biosensors. Given that any microbiome is complex, a multiplex sensor will be needed, and with large, complicated datasets, machine learning might benefit these analyses. If the signals from the biosensors are optical, a smartphone can be used as a portable optical reader and potential data-analyzing device. Biosensing is a rich field that couples engineering and biology, and applying its toolset to help advance soil microbiome characterization would be a boon to microbiology more broadly.

Intein-based thermoregulated meganucleases for containment of genetic material

Limiting the spread of synthetic genetic information outside of the intended use is essential for applications where biocontainment is critical. In particular, biocontainment of engineered probiotics and plasmids that are excreted from the mammalian gastrointestinal tract is needed to prevent escape and acquisition of genetic material that could confer a selective advantage to microbial communities. Here, we built a simple and lightweight biocontainment system that post-translationally activates a site-specific DNA endonuclease to degrade DNA at 18°C and not at higher temperatures. We constructed an orthogonal set of temperature-sensitive meganucleases (TSMs) by inserting the yeast VMA1 L212P temperature-sensitive intein into the coding regions of LAGLIDADG homing endonucleases. We showed that the TSMs eliminated plasmids carrying the cognate TSM target site from laboratory strains of Escherichia coli at the permissive 18°C but not at higher restrictive temperatures. Plasmid elimination is dependent on both TSM endonuclease activity and intein splicing. TSMs eliminated plasmids from E. coli Nissle 1917 after passage through the mouse gut when fecal resuspensions were incubated at 18°C but not at 37°C. Collectively, our data demonstrates the potential of thermoregulated meganucleases as a means of restricting engineered plasmids and probiotics to the mammalian gut.

A generalizable Cas9/sgRNA prediction model using machine transfer learning with small high-quality datasets

The CRISPR/Cas9 nuclease from Streptococcus pyogenes (SpCas9) can be used with single guide RNAs (sgRNAs) as a sequence-specific antimicrobial agent and as a genome-engineering tool. However, current bacterial sgRNA activity models struggle with accurate predictions and do not generalize well, possibly because the underlying datasets used to train the models do not accurately measure SpCas9/sgRNA activity and cannot distinguish on-target cleavage from toxicity. Here, we solve this problem by using a two-plasmid positive selection system to generate high-quality data that more accurately reports on SpCas9/sgRNA cleavage and that separates activity from toxicity. We develop a machine learning architecture (crisprHAL) that can be trained on existing datasets, that shows marked improvements in sgRNA activity prediction accuracy when transfer learning is used with small amounts of high-quality data, and that can generalize predictions to different bacteria. The crisprHAL model recapitulates known SpCas9/sgRNA-target DNA interactions and provides a pathway to a generalizable sgRNA bacterial activity prediction tool that will enable accurate antimicrobial and genome engineering applications.