Dynamic gene expression and growth underlie cell-to-cell heterogeneity in Escherichia coli stress response.

Cell-to-cell heterogeneity in gene expression and growth can have critical functional consequences, such as determining whether individual bacteria survive or die following stress. Although phenotypic variability is well documented, the dynamics that underlie it are often unknown. This information is important because dramatically different outcomes can arise from gradual versus rapid changes in expression and growth. Using single-cell time-lapse microscopy, we measured the temporal expression of a suite of stress-response reporters in Escherichia coli, while simultaneously monitoring growth rate. In conditions without stress, we found several examples of pulsatile expression. Single-cell growth rates were often anticorrelated with reporter levels, with changes in growth preceding changes in expression. These dynamics have functional consequences, which we demonstrate by measuring survival after challenging cells with the antibiotic ciprofloxacin. Our results suggest that fluctuations in both gene expression and growth dynamics in stress-response networks have direct consequences on survival.

DeLTA 2.0: A deep learning pipeline for quantifying single-cell spatial and temporal dynamics.

Improvements in microscopy software and hardware have dramatically increased the pace of image acquisition, making analysis a major bottleneck in generating quantitative, single-cell data. Although tools for segmenting and tracking bacteria within time-lapse images exist, most require human input, are specialized to the experimental set up, or lack accuracy. Here, we introduce DeLTA 2.0, a purely Python workflow that can rapidly and accurately analyze images of single cells on two-dimensional surfaces to quantify gene expression and cell growth. The algorithm uses deep convolutional neural networks to extract single-cell information from time-lapse images, requiring no human input after training. DeLTA 2.0 retains all the functionality of the original version, which was optimized for bacteria growing in the mother machine microfluidic device, but extends results to two-dimensional growth environments. Two-dimensional environments represent an important class of data because they are more straightforward to implement experimentally, they offer the potential for studies using co-cultures of cells, and they can be used to quantify spatial effects and multi-generational phenomena. However, segmentation and tracking are significantly more challenging tasks in two-dimensions due to exponential increases in the number of cells. To showcase this new functionality, we analyze mixed populations of antibiotic resistant and susceptible cells, and also track pole age and growth rate across generations. In addition to the two-dimensional capabilities, we also introduce several major improvements to the code that increase accessibility, including the ability to accept many standard microscopy file formats as inputs and the introduction of a Google Colab notebook so users can try the software without installing the code on their local machine. DeLTA 2.0 is rapid, with run times of less than 10 minutes for complete movies with hundreds of cells, and is highly accurate, with error rates around 1%, making it a powerful tool for analyzing time-lapse microscopy data.

Programmable gene regulation for metabolic engineering using decoy transcription factor binding sites.

Transcription factor decoy binding sites are short DNA sequences that can titrate a transcription factor away from its natural binding site, therefore regulating gene expression. In this study, we harness synthetic transcription factor decoy systems to regulate gene expression for metabolic pathways in Escherichia coli. We show that transcription factor decoys can effectively regulate expression of native and heterologous genes. Tunability of the decoy can be engineered via changes in copy number or modifications to the DNA decoy site sequence. Using arginine biosynthesis as a showcase, we observed a 16-fold increase in arginine production when we introduced the decoy system to steer metabolic flux towards increased arginine biosynthesis, with negligible growth differences compared to the wild type strain. The decoy-based production strain retains high genetic integrity; in contrast to a gene knock-out approach where mutations were common, we detected no mutations in the production system using the decoy-based strain. We further show that transcription factor decoys are amenable to multiplexed library screening by demonstrating enhanced tolerance to pinene with a combinatorial decoy library. Our study shows that transcription factor decoy binding sites are a powerful and compact tool for metabolic engineering.

DeLTA: Automated cell segmentation, tracking, and lineage reconstruction using deep learning.

Microscopy image analysis is a major bottleneck in quantification of single-cell microscopy data, typically requiring human oversight and curation, which limit both accuracy and throughput. To address this, we developed a deep learning-based image analysis pipeline that performs segmentation, tracking, and lineage reconstruction. Our analysis focuses on time-lapse movies of Escherichia coli cells trapped in a "mother machine" microfluidic device, a scalable platform for long-term single-cell analysis that is widely used in the field. While deep learning has been applied to cell segmentation problems before, our approach is fundamentally innovative in that it also uses machine learning to perform cell tracking and lineage reconstruction. With this framework we are able to get high fidelity results (1% error rate), without human intervention. Further, the algorithm is fast, with complete analysis of a typical frame containing ~150 cells taking <700msec. The framework is not constrained to a particular experimental set up and has the potential to generalize to time-lapse images of other organisms or different experimental configurations. These advances open the door to a myriad of applications including real-time tracking of gene expression and high throughput analysis of strain libraries at single-cell resolution.

Salicylate Increases Fitness Cost Associated with MarA-Mediated Antibiotic Resistance.

Antibiotic resistance is generally associated with a fitness deficit resulting from the burden of producing and maintaining resistance machinery. This additional cost suggests that resistant bacteria will be outcompeted by susceptible bacteria in conditions without antibiotics. However, in practice, this process is slow in part because of regulation that minimizes expression of these genes in the absence of antibiotics. This suggests that if it were possible to turn on their expression, the cost would increase, thereby accelerating removal of resistant strains. Experimental and theoretical studies have shown that environmental chemicals can change the fitness cost associated with resistance and therefore have a significant impact on population dynamics. The multiple antibiotic resistance activator (MarA) is a clinically important regulator in Escherichia coli that activates downstream genes to increase resistance against multiple classes of antibiotics. Salicylate is an inducer of MarA that can be found in the environment and derepresses marA's expression. In this study, we sought to unravel the interplay between salicylate and the fitness cost of MarA-mediated antibiotic resistance. Using salicylate as an inducer of MarA, we found that a wide spectrum of concentrations can increase burden in resistant strains compared to susceptible strains. Induction resulted in rapid exclusion of resistant bacteria from mixed populations of antibiotic-resistant and susceptible cells. A mathematical model captures the process and predicts its effect in various environmental conditions. Our work provides a quantitative understanding of salicylate exposure on the fitness of different MarA variants and suggests that salicylate can lead to selection against MarA-mediated resistant strains. More generally, our findings show that natural inducers may serve to bias population membership and could impact antibiotic resistance and other important phenotypes.

Distinct timescales of RNA regulators enable the construction of a genetic pulse generator.

To build complex genetic networks with predictable behaviors, synthetic biologists use libraries of modular parts that can be characterized in isolation and assembled together to create programmable higher-order functions. Characterization experiments and computational models for gene regulatory parts operating in isolation are routinely used to predict the dynamics of interconnected parts and guide the construction of new synthetic devices. Here, we individually characterize two modes of RNA-based transcriptional regulation, using small transcription activating RNAs (STARs) and clustered regularly interspaced short palindromic repeats interference (CRISPRi), and show how their distinct regulatory timescales can be used to engineer a composed feedforward loop that creates a pulse of gene expression. We use a cell-free transcription-translation system (TXTL) to rapidly characterize the system, and we apply Bayesian inference to extract kinetic parameters for an ordinary differential equation-based mechanistic model. We then demonstrate in simulation and verify with TXTL experiments that the simultaneous regulation of a single gene target with STARs and CRISPRi leads to a pulse of gene expression. Our results suggest the modularity of the two regulators in an integrated genetic circuit, and we anticipate that construction and modeling frameworks that can leverage this modularity will become increasingly important as synthetic circuits increase in complexity.

Active degradation of MarA controls coordination of its downstream targets.

Several key transcription factors have unusually short half-lives compared to other cellular proteins. Here, we explore the utility of active degradation in shaping how the multiple antibiotic resistance activator MarA coordinates its downstream targets. MarA controls a variety of stress response genes in Escherichia coli. We modify its half-life either by knocking down the protease that targets it via CRISPRi or by engineering MarA to protect it from degradation. Our experimental and analytical results indicate that active degradation can impact both the rate of coordination and the maximum coordination that downstream genes can achieve. In the context of multi-gene regulation, trade-offs between these properties show that perfect information fidelity and instantaneous coordination cannot coexist.

Stress Introduction Rate Alters the Benefit of AcrAB-TolC Efflux Pumps.

Stress tolerance studies are typically conducted in an all-or-none fashion. However, in realistic settings-such as in clinical or metabolic engineering applications-cells may encounter stresses at different rates. Therefore, how cells tolerate stress may depend on its rate of appearance. To address this, we studied how the rate of stress introduction affects bacterial stress tolerance by focusing on a key stress response mechanism. Efflux pumps, such as AcrAB-TolC of Escherichia coli, are membrane transporters well known for the ability to export a wide variety of substrates, including antibiotics, signaling molecules, and biofuels. Although efflux pumps improve stress tolerance, pump overexpression can result in a substantial fitness cost to the cells. We hypothesized that the ideal pump expression level would involve a rate-dependent trade-off between the benefit of pumps and the cost of their expression. To test this, we evaluated the benefit of the AcrAB-TolC pump under different rates of stress introduction, including a step, a fast ramp, and a gradual ramp. Using two chemically diverse stresses, the antibiotic chloramphenicol and the jet biofuel precursor pinene, we assessed the benefit provided by the pumps. A mathematical model describing these effects predicted the benefit as a function of the rate of stress introduction. Our findings demonstrate that as the rate of introduction is lowered, stress response mechanisms provide a disproportionate benefit to pump-containing strains, allowing cells to survive beyond the original inhibitory concentrations.IMPORTANCE Efflux pumps are ubiquitous in nature and provide stress tolerance in the cells of species ranging from bacteria to mammals. Understanding how pumps provide tolerance has far-reaching implications for diverse fields, from medicine to biotechnology. Here, we investigated how the rate of stressor appearance impacts tolerance. We focused on two distinct substrates of AcrAB-TolC efflux pumps, the antibiotic chloramphenicol and the biofuel precursor pinene. Interestingly, tolerance is highly dependent on the rate of stress introduction. Therefore, it is important to consider not only the total quantity of a stressor but also the rate at which it is applied. The implications of this work are significant because environments are rarely static; antibiotic concentrations change during dosing, and metabolic engineering processes change with time.

Customized Regulation of Diverse Stress Response Genes by the Multiple Antibiotic Resistance Activator MarA.

Stress response networks frequently have a single upstream regulator that controls many downstream genes. However, the downstream targets are often diverse, therefore it remains unclear how their expression is specialized when under the command of a common regulator. To address this, we focused on a stress response network where the multiple antibiotic resistance activator MarA from Escherichia coli regulates diverse targets ranging from small RNAs to efflux pumps. Using single-cell experiments and computational modeling, we showed that each downstream gene studied has distinct activation, noise, and information transmission properties. Critically, our results demonstrate that understanding biological context is essential; we found examples where strong activation only occurs outside physiologically relevant ranges of MarA and others where noise is high at wild type MarA levels and decreases as MarA reaches its physiological limit. These results demonstrate how a single regulatory protein can maintain specificity while orchestrating the response of many downstream genes.

Phenotypic Diversity Using Bimodal and Unimodal Expression of Stress Response Proteins.

Populations of cells need to express proteins to survive the sudden appearance of stressors. However, these mechanisms may be taxing. Populations can introduce diversity, allowing individual cells to stochastically switch between fast-growing and stress-tolerant states. One way to achieve this is to use genetic networks coupled with noise to generate bimodal distributions with two distinct subpopulations, each adapted to a stress condition. Another survival strategy is to rely on random fluctuations in gene expression to produce continuous, unimodal distributions of the stress response protein. To quantify the environmental conditions where bimodal versus unimodal expression is beneficial, we used a differential evolution algorithm to evolve optimal distributions of stress response proteins given environments with sudden fluctuations between low and high stress. We found that bimodality evolved for a large range of environmental conditions. However, we asked whether these findings were an artifact of considering two well-defined stress environments (low and high stress). As noise in the environment increases, or when there is an intermediate environment (medium stress), the benefits of bimodality decrease. Our results indicate that under realistic conditions, a continuum of resistance phenotypes generated through a unimodal distribution is sufficient to ensure survival without a high cost to the population.

Noise and low-level dynamics can coordinate multicomponent bet hedging mechanisms.

To counter future uncertainty, cells can stochastically express stress response mechanisms to diversify their population and hedge against stress. This approach allows a small subset of the population to survive without the prohibitive cost of constantly expressing resistance machinery at the population level. However, expression of multiple genes in concert is often needed to ensure survival, requiring coordination of infrequent events across many downstream targets. This raises the question of how cells orchestrate the timing of multiple rare events without adding cost. To investigate this, we used a stochastic model to study regulation of downstream target genes by a transcription factor. We compared several upstream regulator profiles, including constant expression, pulsatile dynamics, and noisy expression. We found that pulsatile dynamics and noise are sufficient to coordinate expression of multiple downstream genes. Notably, this is true even when fluctuations in the upstream regulator are far below the dissociation constants of the regulated genes, as with infrequently activated genes. As an example, we simulated the dynamics of the multiple antibiotic resistance activator (MarA) and 40 diverse downstream genes it regulates, determining that low-level dynamics in MarA are sufficient to coordinate expression of resistance mechanisms. We also demonstrated that noise can play a similar coordinating role. Importantly, we found that these benefits are present without a corresponding increase in the population-level cost. Therefore, our model suggests that low-level dynamics or noise in a transcription factor can coordinate expression of multiple stress response mechanisms by engaging them simultaneously without adding to the overall cost.

Tunable stochastic pulsing in the Escherichia coli multiple antibiotic resistance network from interlinked positive and negative feedback loops.

Cells live in uncertain, dynamic environments and have many mechanisms for sensing and responding to changes in their surroundings. However, sudden fluctuations in the environment can be catastrophic to a population if it relies solely on sensory responses, which have a delay associated with them. Cells can reconcile these effects by using a tunable stochastic response, where in the absence of a stressor they create phenotypic diversity within an isogenic population, but use a deterministic response when stressors are sensed. Here, we develop a stochastic model of the multiple antibiotic resistance network of Escherichia coli and show that it can produce tunable stochastic pulses in the activator MarA. In particular, we show that a combination of interlinked positive and negative feedback loops plays an important role in setting the dynamics of the stochastic pulses. Negative feedback produces a pulsatile response that is tunable, while positive feedback serves to amplify the effect. Our simulations show that the uninduced native network is in a parameter regime that is of low cost to the cell (taxing resistance mechanisms are expressed infrequently) and also elevated noise strength (phenotypic variability is high). The stochastic pulsing can be tuned by MarA induction such that variability is decreased once stresses are sensed, avoiding the detrimental effects of noise when an optimal MarA concentration is needed. We further show that variability in the expression of MarA can act as a bet hedging mechanism, allowing for survival in time-varying stress environments, however this effect is tunable to allow for a fully induced, deterministic response in the presence of a stressor.

Generating information-dense promoter sequences with optimal string packing.

Dense arrangements of binding sites within nucleotide sequences can collectively influence downstream transcription rates or initiate biomolecular interactions. For example, natural promoter regions can harbor many overlapping transcription factor binding sites that influence the rate of transcription initiation. Despite the prevalence of overlapping binding sites in nature, rapid design of nucleotide sequences with many overlapping sites remains a challenge. Here, we show that this is an NP-hard problem, coined here as the nucleotide String Packing Problem (SPP). We then introduce a computational technique that efficiently assembles sets of DNA-protein binding sites into dense, contiguous stretches of double-stranded DNA. For the efficient design of nucleotide sequences spanning hundreds of base pairs, we reduce the SPP to an Orienteering Problem with integer distances, and then leverage modern integer linear programming solvers. Our method optimally packs libraries of 20-100 binding sites into dense nucleotide arrays of 50-300 base pairs in 0.05-10 seconds. Unlike approximation algorithms or meta-heuristics, our approach finds provably optimal solutions. We demonstrate how our method can generate large sets of diverse sequences suitable for library generation, where the frequency of binding site usage across the returned sequences can be controlled by modulating the objective function. As an example, we then show how adding additional constraints, like the inclusion of sequence elements with fixed positions, allows for the design of bacterial promoters. The nucleotide string packing approach we present can accelerate the design of sequences with complex DNA-protein interactions. When used in combination with synthesis and high-throughput screening, this design strategy could help interrogate how complex binding site arrangements impact either gene expression or biomolecular mechanisms in varied cellular contexts. Author Summary: The way protein binding sites are arranged on DNA can control the regulation and transcription of downstream genes. Areas with a high concentration of binding sites can enable complex interplay between transcription factors, a feature that is exploited by natural promoters. However, designing synthetic promoters that contain dense arrangements of binding sites is a challenge. The task involves overlapping many binding sites, each typically about 10 nucleotides long, within a constrained sequence area, which becomes increasingly difficult as sequence length decreases, and binding site variety increases. We introduce an approach to design nucleotide sequences with optimally packed protein binding sites, which we call the nucleotide String Packing Problem (SPP). We show that the SPP can be solved efficiently using integer linear programming to identify the densest arrangements of binding sites for a specified sequence length. We show how adding additional constraints, like the inclusion of sequence elements with fixed positions, allows for the design of bacterial promoters. The presented approach enables the rapid design and study of nucleotide sequences with complex, dense binding site architectures.

Deep model predictive control of gene expression in thousands of single cells.

Gene expression is inherently dynamic, due to complex regulation and stochastic biochemical events. However, the effects of these dynamics on cell phenotypes can be difficult to determine. Researchers have historically been limited to passive observations of natural dynamics, which can preclude studies of elusive and noisy cellular events where large amounts of data are required to reveal statistically significant effects. Here, using recent advances in the fields of machine learning and control theory, we train a deep neural network to accurately predict the response of an optogenetic system in Escherichia coli cells. We then use the network in a deep model predictive control framework to impose arbitrary and cell-specific gene expression dynamics on thousands of single cells in real time, applying the framework to generate complex time-varying patterns. We also showcase the framework's ability to link expression patterns to dynamic functional outcomes by controlling expression of the tetA antibiotic resistance gene. This study highlights how deep learning-enabled feedback control can be used to tailor distributions of gene expression dynamics with high accuracy and throughput without expert knowledge of the biological system.

Light-inducible protein degradation in E. coli with the LOVdeg tag.

Molecular tools for optogenetic control allow for spatial and temporal regulation of cell behavior. In particular, light-controlled protein degradation is a valuable mechanism of regulation because it can be highly modular, used in tandem with other control mechanisms, and maintain functionality throughout growth phases. Here, we engineered LOVdeg, a tag that can be appended to a protein of interest for inducible degradation in Escherichia coli using blue light. We demonstrate the modularity of LOVdeg by using it to tag a range of proteins, including the LacI repressor, CRISPRa activator, and the AcrB efflux pump. Additionally, we demonstrate the utility of pairing the LOVdeg tag with existing optogenetic tools to enhance performance by developing a combined EL222 and LOVdeg system. Finally, we use the LOVdeg tag in a metabolic engineering application to demonstrate post-translational control of metabolism. Together, our results highlight the modularity and functionality of the LOVdeg tag system and introduce a powerful new tool for bacterial optogenetics.

Advances in linking single-cell bacterial stress response to population-level survival.

Stress response mechanisms can allow bacteria to survive a myriad of challenges, including nutrient changes, antibiotic encounters, and antagonistic interactions with other microbes. Expression of these stress response pathways, in addition to other cell features such as growth rate and metabolic state, can be heterogeneous across cells and over time. Collectively, these single-cell-level phenotypes contribute to an overall population-level response to stress. These include diversifying actions, which can be used to enable bet-hedging, and coordinated actions, such as biofilm production, horizontal gene transfer, and cross-feeding. Here, we highlight recent results and emerging technologies focused on both single-cell and population-level responses to stressors, and we draw connections about the combined impact of these effects on survival of bacterial communities.

An optogenetic toolkit for light-inducible antibiotic resistance.

Antibiotics are a key control mechanism for synthetic biology and microbiology. Resistance genes are used to select desired cells and regulate bacterial populations, however their use to-date has been largely static. Precise spatiotemporal control of antibiotic resistance could enable a wide variety of applications that require dynamic control of susceptibility and survival. Here, we use light-inducible Cre recombinase to activate expression of drug resistance genes in Escherichia coli. We demonstrate light-activated resistance to four antibiotics: carbenicillin, kanamycin, chloramphenicol, and tetracycline. Cells exposed to blue light survive in the presence of lethal antibiotic concentrations, while those kept in the dark do not. To optimize resistance induction, we vary promoter, ribosome binding site, and enzyme variant strength using chromosome and plasmid-based constructs. We then link inducible resistance to expression of a heterologous fatty acid enzyme to increase production of octanoic acid. These optogenetic resistance tools pave the way for spatiotemporal control of cell survival.

Light inducible protein degradation in E. coli with the LOVdeg tag.

Molecular tools for optogenetic control allow for spatial and temporal regulation of cell behavior. In particular, light controlled protein degradation is a valuable mechanism of regulation because it can be highly modular, used in tandem with other control mechanisms, and maintain functionality throughout growth phases. Here, we engineered LOVdeg, a tag that can be appended to a protein of interest for inducible degradation in Escherichia coli using blue light. We demonstrate the modularity of LOVdeg by using it to tag a range of proteins, including the LacI repressor, CRISPRa activator, and the AcrB efflux pump. Additionally, we demonstrate the utility of pairing the LOVdeg tag with existing optogenetic tools to enhance performance by developing a combined EL222 and LOVdeg system. Finally, we use the LOVdeg tag in a metabolic engineering application to demonstrate post-translational control of metabolism. Together, our results highlight the modularity and functionality of the LOVdeg tag system, and introduce a powerful new tool for bacterial optogenetics.

Deep Neural Networks for Predicting Single-Cell Responses and Probability Landscapes.

Engineering biology relies on the accurate prediction of cell responses. However, making these predictions is challenging for a variety of reasons, including the stochasticity of biochemical reactions, variability between cells, and incomplete information about underlying biological processes. Machine learning methods, which can model diverse input-output relationships without requiring a priori mechanistic knowledge, are an ideal tool for this task. For example, such approaches can be used to predict gene expression dynamics given time-series data of past expression history. To explore this application, we computationally simulated single-cell responses, incorporating different sources of noise and alternative genetic circuit designs. We showed that deep neural networks trained on these simulated data were able to correctly infer the underlying dynamics of a cell response even in the presence of measurement noise and stochasticity in the biochemical reactions. The training set size and the amount of past data provided as inputs both affected prediction quality, with cascaded genetic circuits that introduce delays requiring more past data. We also tested prediction performance on a bistable auto-activation circuit, finding that our initial method for predicting a single trajectory was fundamentally ill-suited for multimodal dynamics. To address this, we updated the network architecture to predict the entire distribution of future states, showing it could accurately predict bimodal expression distributions. Overall, these methods can be readily applied to the diverse prediction tasks necessary to predict and control a variety of biological circuits, a key aspect of many synthetic biology applications.

Comprehensive Screening of a Light-Inducible Split Cre Recombinase with Domain Insertion Profiling.

Splitting proteins with light- or chemically inducible dimers provides a mechanism for post-translational control of protein function. However, current methods for engineering stimulus-responsive split proteins often require significant protein engineering expertise and the laborious screening of individual constructs. To address this challenge, we use a pooled library approach that enables rapid generation and screening of nearly all possible split protein constructs in parallel, where results can be read out by using sequencing. We perform our method on Cre recombinase with optogenetic dimers as a proof of concept, resulting in comprehensive data on the split sites throughout the protein. To improve the accuracy in predicting split protein behavior, we develop a Bayesian computational approach to contextualize errors inherent to experimental procedures. Overall, our method provides a streamlined approach for achieving inducible post-translational control of a protein of interest.