Periodic Reporting for period 1 - PERTURBATIONS (Large scale perturbations of the gene regulatory networks of E. coli)
Reporting period: 2021-04-01 to 2023-03-31
A grand promise of modern biology is to predict phenotypes from genotypes. The most notable example is the Human Genome Project which aimed to dramatically accelerate diagnosis, prevention, and treatment of diseases based on genome sequence and gene functions. Cures for complex diseases such as heart disease, schizophrenia, and cancer were supposed to be within reach, but 20 years later we are far from that goal. Another major implication of genotype-to-phenotype mapping is to predict evolution, which is crucial for preventing antibiotic resistance, controlling the spread of infectious diseases (as illustrated by the current pandemic), and anticipating micro-organism adaptations to climate change. However, predicting phenotypes from genomes has revealed to be much more challenging than expected, has halted the outcome of the Human Genome Project, and remains a fully open question. Indeed, the “one gene, one function, one disease” model does not match observations of complex genetic interactions. The factors underlying such complexity are: (1) Epistasis (Genotype-by-Genotype interactions); (2) Genotype-by-Environment interactions; (3) Pleiotropy. Epistasis has been measured and classified in several model system, but the origin of epistasis has been explained only in small synthetic gene networks, notably by the Philippe Nghe’s lab. Pleiotropy remains poorly characterized in general. Environmental dependences are essentially studied in the context of regulatory responses, but little as a function of genotypes. Overall, the interactions described above are widespread, but there is no generic framework to explain them from a priori knowledge in biology. Inferring these interactions from human studies is highly challenging due to incomplete information and biases in the sampled population. Instead, this project aims to address complex genetic interactions in a model organism, amenable to systematic experimental investigation, with the ambition to extract generic rules and push the frontiers of understanding major societal issues like diseases, adaptation to changing environment, and evolution in general.
The overall ambition of this project is to: (1) develop technology to measure the transcriptomes and growth of up to 104 E. coli strains at a very high-throughput; (2) apply this technology to exhaustively decipher the genetic interactions between global regulators, and between global and local regulators, as a model of epistasis and pleiotropy, in different environments; (3) study the relationship between the known gene network structure and the observed genetic interactions. This project was divided into three work packages (WP). WP1 builds a novel cutting-edge method based on microfluidics and spatial transcriptomics for bacteria. In WP1 we proposed to grow up to 104 colonies each of a different strain, each colony comprising ~103 bacteria, on a microfabricated polyacrylamide (hydrogel) pad coupled with spatial transcriptomics, dubbed “Bacterial Evolution Adapted Spatial Transcriptomic Sequencing” or BEAST-Seq. By locally capturing mRNAs on spatially barcoded DNA primers, the transcriptome of each clonal colony is associated to its position on the pad and to its growth rate measured by time-lapse microscopy. Additionally, the hydrogel pad used as a membrane allows us to vary the growth medium in time. WP2 studies genome wide regulatory logic of global regulators and gene hierarchies. Using the technology developed in WP1, we proposed to characterize high-order gene interactions between global regulators using a combinatorial CRISPR/dCas9 knock-downs (KDs) library of the host lab. WP3 uncovers the impact of mutations on global regulators. We proposed to use Tenaillon’s lab (INSERM Bichat) newly developed CRISPR/Cas9 tool to create mutations libraries of the global regulators from WP2. The fine-grained view provided by mutations compared to KDs in WP2 will allow us to test mathematical models of gene networks that predict epistasis and pleiotropy.
The overall ambition of this project is to: (1) develop technology to measure the transcriptomes and growth of up to 104 E. coli strains at a very high-throughput; (2) apply this technology to exhaustively decipher the genetic interactions between global regulators, and between global and local regulators, as a model of epistasis and pleiotropy, in different environments; (3) study the relationship between the known gene network structure and the observed genetic interactions. This project was divided into three work packages (WP). WP1 builds a novel cutting-edge method based on microfluidics and spatial transcriptomics for bacteria. In WP1 we proposed to grow up to 104 colonies each of a different strain, each colony comprising ~103 bacteria, on a microfabricated polyacrylamide (hydrogel) pad coupled with spatial transcriptomics, dubbed “Bacterial Evolution Adapted Spatial Transcriptomic Sequencing” or BEAST-Seq. By locally capturing mRNAs on spatially barcoded DNA primers, the transcriptome of each clonal colony is associated to its position on the pad and to its growth rate measured by time-lapse microscopy. Additionally, the hydrogel pad used as a membrane allows us to vary the growth medium in time. WP2 studies genome wide regulatory logic of global regulators and gene hierarchies. Using the technology developed in WP1, we proposed to characterize high-order gene interactions between global regulators using a combinatorial CRISPR/dCas9 knock-downs (KDs) library of the host lab. WP3 uncovers the impact of mutations on global regulators. We proposed to use Tenaillon’s lab (INSERM Bichat) newly developed CRISPR/Cas9 tool to create mutations libraries of the global regulators from WP2. The fine-grained view provided by mutations compared to KDs in WP2 will allow us to test mathematical models of gene networks that predict epistasis and pleiotropy.
During the project timeline we have focused on developing three main methods: 1) cell growth in microfluidics (Fig. 1), 2) cell growth in hydrogels (Fig. 2), and 3) spatial transcriptomics for bacteria (Fig. 3).
We defined the workflow of the BEAST-seq as following: 1) grow the cells, 2) lyse the cells, 3) protect RNA from degradation, 4) hybridize mRNA and barcoded spatial probes, 5) create cDNA library via reverse transcription, and 6) recover sample from the slide and send to sequencing.
We have experimented with cell growth in microfluidics device, cell growth in hydrogels, and cell growth in a combination of microfluidics and hydrogels. Ultimately, we have chosen to focus on cell growth in hydrogels because of its simplicity of manipulation and highest similarity to tissues used in spatial transcriptomics. We managed to grow three strains of ampicillin resistant bacteria (mCerulean, eGFP, and mCherry) inside the agarose gel and control the number of colonies grown.
Once cells are grown in the agarose pad, they have to be lysed to release mRNA and mRNA has to be protected from degradation. Initially, we used lysozyme to break down the bacterial cell wall peptidoglycan and to disrupt their membranes. Since the half-life of mRNA in bacteria is approximately 4 minutes, it is stabilized before cell lysis. The RNA Later protocol by Invitrogen was used to that purpose. Later on, as an optimization of the protocol the cells were lysed through physical disruption by thermal lysis in a cool-boil-cool cycle. For this procedure the Microarray Hybridization Chamber Assembly from Agilent was used. This whole assembly is then placed on ice to cool down the cells. Directly after that the assembly is placed inside the dry bath incubator to 96 °C for 5 minutes. This causes the cells to burst and release mRNA in the environment. The main advantage of this procedure was to save time and chemicals.
Following optimization and knowing that we can grow cells, lyse them, and protect mRNA from degradation, we proceeded with developing spatial transcriptomics protocol for bacteria. Our strategy was stepwise from low complexity to high complexity in order to understand that every step of this complex protocol is working. We first checked that everything works in bulk cultures with large amount of bacteria (~1,5 x 108 cells) and small amount of bacteria (~1000 cells), both when RNA is purified with a kit and when there is no RNA purification after lysis. Then we proceeded with the same strategy, but instead of bulk cultures we used DNA microarrays. Finally, as the experiment with highest complexity we grew colonies inside agarose pads on DNA microarrays. We managed to obtain the cDNA of all genes of interest tied to specific unique barcodes of up to 10000 colonies and developed the protocol that prepares the samples for Next Generation Sequencing. We obtained additional funding that will help us advance this project beyond the end of this fellowship to achieve all the set goals and publish this method.
We defined the workflow of the BEAST-seq as following: 1) grow the cells, 2) lyse the cells, 3) protect RNA from degradation, 4) hybridize mRNA and barcoded spatial probes, 5) create cDNA library via reverse transcription, and 6) recover sample from the slide and send to sequencing.
We have experimented with cell growth in microfluidics device, cell growth in hydrogels, and cell growth in a combination of microfluidics and hydrogels. Ultimately, we have chosen to focus on cell growth in hydrogels because of its simplicity of manipulation and highest similarity to tissues used in spatial transcriptomics. We managed to grow three strains of ampicillin resistant bacteria (mCerulean, eGFP, and mCherry) inside the agarose gel and control the number of colonies grown.
Once cells are grown in the agarose pad, they have to be lysed to release mRNA and mRNA has to be protected from degradation. Initially, we used lysozyme to break down the bacterial cell wall peptidoglycan and to disrupt their membranes. Since the half-life of mRNA in bacteria is approximately 4 minutes, it is stabilized before cell lysis. The RNA Later protocol by Invitrogen was used to that purpose. Later on, as an optimization of the protocol the cells were lysed through physical disruption by thermal lysis in a cool-boil-cool cycle. For this procedure the Microarray Hybridization Chamber Assembly from Agilent was used. This whole assembly is then placed on ice to cool down the cells. Directly after that the assembly is placed inside the dry bath incubator to 96 °C for 5 minutes. This causes the cells to burst and release mRNA in the environment. The main advantage of this procedure was to save time and chemicals.
Following optimization and knowing that we can grow cells, lyse them, and protect mRNA from degradation, we proceeded with developing spatial transcriptomics protocol for bacteria. Our strategy was stepwise from low complexity to high complexity in order to understand that every step of this complex protocol is working. We first checked that everything works in bulk cultures with large amount of bacteria (~1,5 x 108 cells) and small amount of bacteria (~1000 cells), both when RNA is purified with a kit and when there is no RNA purification after lysis. Then we proceeded with the same strategy, but instead of bulk cultures we used DNA microarrays. Finally, as the experiment with highest complexity we grew colonies inside agarose pads on DNA microarrays. We managed to obtain the cDNA of all genes of interest tied to specific unique barcodes of up to 10000 colonies and developed the protocol that prepares the samples for Next Generation Sequencing. We obtained additional funding that will help us advance this project beyond the end of this fellowship to achieve all the set goals and publish this method.
We have pushed this project beyond the state of the art by developing spatial transcriptomics for bacteria. We expect this method to be used to screen transcriptomes of bacteria in a high-throughput manner which should help expand the understanding of diseases, adaptation to environment, and evolution in general.