Advances in linking single cell bacterial stress response to population level survival Short title: Single cell and population level stress response

Abstract

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.

Introduction

In nature, bacteria routinely encounter stressors in their environment. These can range from environmental and nutrient changes to antibiotic introduction and interactions with invading microbial or host cells (Figure 1A). In response, bacteria have evolved many stress resistance mechanisms and response pathways. However, what has become apparent over the past several decades is that the way bacteria cope with stresses can vary substantially across cells within a population. For example, variation in gene expression and growth rate from cell to cell can act as a bet-hedging mechanism to ensure a subset of cells survives stressful conditions [1]. One of the most familiar examples of this is bacterial persistence, where communities of genetically identical cells contain two subpopulations, one of which is killed rapidly by antibiotics while the other remains tolerant [2]. Other response mechanisms depend on the interactions between individual cells to influence collective behavior, such as in the case of biofilm production. In addition, bacteria in nature are found in mixed species communities which results in a myriad of effects including competition for resources, toxin production, cross-feeding, and exchange of genetic material. What these phenomena all have in common is that single cell level effects play a critical role in determining population level outcomes in response to stress.

Figure 1. Single cell stress responses.

(A) Bacteria can encounter a range of stressors including antibiotics, reactive oxygen species (ROS) produced by the host, pH changes, limiting nutrients, and other cells. (B) Phenotypic heterogeneity in stress response is determined by multiple factors that can vary over time. From top to bottom, gene expression levels fluctuate within each cell. Gene amplification can change the copy number of a gene, affecting the phenotype. Changes in growth rate and metabolism over time (represented by different cell sizes and division times) affect each cell‟s vulnerability to stressors. The degradation rate of proteins (represented by triangles) decreases when nutrients are depleted, reducing the amount of ATP available for proteases and leading to aggregation of specific proteins.

In this review, we discuss recent progress in understanding stress tolerance mechanisms that lie at the interface of single cell and population level responses. Previous reviews on microbial stress response provide in-depth analyses of specific topics discussed within this paper [3]–[11]. However, here we highlight commonalities across a range of stress response mechanisms to shed light on their interconnection, focusing our discussion on recent results in the area. We begin by investigating single cell level responses to stressors, then discuss advances in quantifying the impact of single cell effects on population level behavior including biofilm production and horizontal gene transfer. Overall, we highlight how linking single cell and population level mechanisms is key to understanding how microbial communities coordinate their response to stress.

Single cell heterogeneity in stress response

Quantitative measurements of single cell gene expression can provide insights into mechanisms bacteria use to survive a wide range of stressors (Figure 1B). For example, individual bacteria can exhibit dynamic changes in gene expression, altered gene copy number, differences in growth and metabolism, or fluctuating levels of key proteins over time. Cell to cell heterogeneity in these factors results in a subset of cells that are more adept at surviving stress.

Gene expression dynamics impact stress survival

A common approach for measuring cell to cell differences in gene expression dynamics is through the use of fluorescent reporters to quantify expression of genes involved in stress response. For example, Jones and Uphoff demonstrated that fluctuations in activation of the SOS response repressor, LexA, result in heterogeneity in expression of genes involved in the SOS response system in Escherichia coli prior to any stress [12]. Sampaio et al. showed that single cell expression levels of a gadX reporter are correlated with an increased probability of ciprofloxacin survival [13]. This connection between antibiotic tolerance and single cell gene expression is a key area where phenotypic heterogeneity has been linked to stress survival [2], [14]. In addition, new studies have used fluorescent reporters for gene expression to understand the role of single cell heterogeneity in oxidative and acid stress responses [15], [16].

Recent studies have also focused on the connection between single cell level heterogeneity and bulk culture outcomes by measuring how transcriptional regulatory networks are activated under stress. One study applied bulk RNA-seq to E. coli cultures with or without ampicillin treatment and categorized genes with differential expression after antibiotic exposure into clusters of orthologous genes [17]. New techniques have improved the quality of single cell RNA-seq in bacteria [18]–[21]. These advances are already being used to determine how heterogeneity in gene expression states allows for subpopulations to survive antibiotic treatment. For example, Ma et al. used a single cell RNA-seq method, BacDrop, to determine gene expression states in clinically susceptible Klebsiella pneumoniae cells that survive antibiotic treatment, finding states that were not identified by bulk RNA sequencing methods. Specifically, for meropenem treatment, bulk RNA sequencing did not find significant differences in gene expression compared to untreated cells, but single cell RNA-seq determined changes in the expression of genes involved in cell wall and membrane synthesis and genes related to persistence [22]. Single cell sequencing methods in bacteria are still in their infancy; however, we anticipate that studies relating single cell measurements to population level effects will be an important area of focus as these techniques mature.

Gene amplification events provide heteroresistance

Phenotypic variation and heteroresistance have also been attributed to gene amplification events that increase tolerance to stress, including antibiotic and metal resistance [23], [24]. Choby et al. showed that random and unstable amplification of a beta-lactamase gene in the genome of a clinical isolate is selected for when a beta-lactam antibiotic is present, shifting populations toward cells with elevated copy numbers of the gene. They determined that, while only about 1 in 150,000 cells contain elevated copy numbers (up to 20x) in the absence of stress, these specific cells are selected for and able to grow in the presence of the antibiotic. Due to the cost of maintaining this elevated copy number, when the antibiotic is removed, the resulting population returns to conditions where most cells have low copy numbers, resulting in transient, nonheritable tolerance. Heteroresistance of this form is enigmatic for clinicians because it can cause strains to present as susceptible in clinical testing, and then lead to relapses in infection that can be lethal.

Metabolism and growth variations impact stress survival

Heterogeneity in cell growth and metabolic states also play an important role in stress survival. Metabolic dormancy has been studied in relation to bacterial persistence, where a decrease in growth can confer antibiotic tolerance in a subset of cells within an isogenic population [2]. A recent study linked the concentration of ATP within cells to persistence in E. coli by using gene knock outs and a fluorescent ATP reporter to characterize cells that survive antibiotic stress [25]. The authors hypothesized that naturally occurring fluctuations in ATP concentrations cause the spontaneous formation of metabolically dormant persister cells prior to antibiotic treatment. Results from another recent study by Kaplan et al. align with this hypothesis by showing that starvation prior to antibiotic addition increases the proportion of persisters cells in the population [26]. They also determined that the rate and intensity of starvation applied to the cells alters their tolerance of and recovery from stress; the more quickly cells are starved, the less time there is for them to alter transcription in response [26]. In agreement with these data, Lopatkin et al. demonstrated that metabolic state is a better predictor of antibiotic lethality than growth rate. The authors used an ATP reporter to measure metabolism and found it more accurately correlated with survival than growth rate measurements such as optical density [27]. In a later study extending these results, mutations in metabolic genes were found in antibiotic resistant clinical E. coli strains [28]. Collectively, these studies link the metabolic activity of cells to persistence and antibiotic tolerance. Heterogeneity in this metabolic activity then affects population survival following antibiotic treatment.

Metabolic state also plays a role in how well cells survive the stress of changing environments and limited nutrients, both of which are situations that bacteria encounter regularly in nature. Julou et al. demonstrated that the level of expression of the lac operon, prior to a change in carbon source, determines the length of lag time before cells can resume growth when switched from glucose to lactose-containing media. They suggested that noise in expression of the lac operon in cells within a population may serve as a bet-hedging mechanism to improve population level survival in environments with changing nutrients [29]. Another study found that metabolic enzymes involved in redox reactions are more variably expressed in E. coli than other proteins with the same average cellular abundance [30]. Variability in the expression of these enzymes results in a wide range of respiratory and growth rates, increasing the chance of survival during nutrient starvation by generating a subset of cells pre-disposed to survive changes in nutrient availability.

Protein production and degradation changes during stress response

Recent studies have also shown that upon encountering stress, changes in protein degradation and aggregation rates play a role in survival. Differential expression of transcriptional regulatory networks in cells that survive antibiotics have been shown to be dependent on decreased proteolytic activity [17], [31]. These studies hypothesize that a reduction in protein degradation allows for the accumulation of stress-related transcription factors and sigma factors that are otherwise turned over in cells. This protein aggregation results from a decrease in ATP concentration, which is needed for enzymatic protein degradation and causes a change in downstream gene expression that can lead to the formation of persister cells [32]. Another study showed that continued protein aggregation after entering the persister state can lead to the formation of viable but not culturable cells [33]. These cells are in an even greater state of dormancy than persister cells and cannot recover directly after antibiotic exposure or extreme starvation. These findings agree with a study showing the dependence of persister cell revival on the Lon protease [34]. Together, these studies suggest that reduced ATP concentration from starvation leads to reduced protein degradation, which allows for accumulation of stress response factors, such as RpoS. These factors then trigger the expression of response pathways, allowing for the formation of persister cells.

Protein production is also reduced during starvation to save resources. This is regulated by production of the alarmone (p)ppGpp during amino acid depletion upon entry into stationary phase or quiescence [35], [36]. Diez et al. demonstrated that (p)ppGpp expression is heterogenous from cell to cell within a population of Bacillus subtilis and that (p)ppGpp-dependent reduction in translation is gene specific [35]. Bougdour and Gottesman showed that (p)ppGpp leads to an accumulation of the stress sigma factor, RpoS, after nutrient starvation [37]. Furthermore, Fessler et al. demonstrated that nutrient starvation leads to the degradation of inactive ribosomes and rRNA to catabolize nucleotides and amino acids within cells [38]. This stringent response mechanism is used upon starvation to save and recycle resources, leading to extended survival times, but can vary in response intensity from cell to cell.

Overall, these studies highlight how single cell heterogeneity plays a role in a wide range of stress response mechanisms including heteroresistance, persistence, nutrient changes, and starvation. In the next section, we focus on community-level effects, where the coordinated action of many individual cells can lead to stress tolerance.

Population and community level stress response

Bacteria in nature exist within complex communities, which can help cells survive changing environments and stress [5]. For example, recent studies have highlighted ways that biofilms can serve as a mode of protection against antibiotics, other microbial species, resource competition between strains, and viral infections [39]–[42]. The formation and maintenance of biofilms is a property of the collective behavior of many individual bacteria. In this section, we highlight advances in understanding how single cell stress responses impact overall population survival within mixed microbial communities (Figure 2). This is also an area where technological advances are coming online that have excellent potential for improving our limited understanding of how individuals impact communities, and we discuss examples of these technologies.

Figure 2. Population level stress responses.

In a community of bacteria, multiple interactions within and between cells affect stress response and survival. Extracellular polymeric substances are secreted to form biofilms that protect against antibiotics and host defenses. Nutrient and oxygen gradients across the population affect metabolic states within the community as depicted by the variety of cell sizes within the biofilm. Cross feeding helps alleviate nutrient limitations within the community. The spatial organization of cells in the community also affects diffusion of chemicals such as toxins and can introduce contact dependent effects. Horizontal gene transfer between cells is represented by blue donor cells and DNA strands. All these factors depend on single cell mechanisms that in turn affect overall population survival of stress.

Imaging and measuring biofilm dynamics

One challenge in studying cell interactions within biofilms is the difficulty of imaging within colonies or extracellular matrices. Emerging technologies are being developed to help address this issue. For example, Qin et al. used a dual-view light-sheet microscopy technique to characterize a “fountain-flow” pattern of cell trajectories through 3D biofilm development in Vibrio cholerae [43]. This system provides resolution similar to confocal microscopy, but works better for time-lapse imaging and has less photobleaching [44], [45]. They also used a fluorescent reporter that forms puncta retained by the mother cells over time to track individual cell lineages [43]. Yordanov et al. developed a similar method using digital confocal light-sheet axial plane optical microscopy to image biofilms and observed V. cholerae biofilm growth dynamics, response to osmotic shock, and phagocytosis by macrophages [46]. In addition, Zhang et al. developed a deep learning image analysis program called Bacterial Cell Morphometry 3D that improves the accuracy and speed of single cell segmentation in three-dimensional fluorescence microscopy images of bacterial communities [47]. Future applications of this work could allow segmentation, tracking, and quantification of fluorescent reporters in bacterial biofilms to improve analysis of single cell responses within communities. Bellin et al. also developed an electrochemical camera chip to measure levels of redox metabolites across Pseudomonas aeruginosa biofilms [48]. This technique can be used to measure metabolic activity of cells within a biofilm. Further quantitative studies in this area have determined long range interactions within and between biofilm communities including growth and competition oscillations [49]–[51]. Moving forward, approaches like these can be used to reveal how single cell gene expression and growth dynamics integrate into the stress response of a biofilm as a whole.

To probe dynamics of cells in biofilm-like conditions, Dal Co et al. simulated E. coli biofilms by using deep microfluidic chambers that have a gradient of glucose concentrations. This gradient causes cells in the back of the chambers to grow more slowly than those in the front, resulting in a phenotypically heterogenous population. The fast-growing cells at the front of the microfluidic chamber are killed by antibiotics, but the slower growing cells at the back of the chamber survive and grow rapidly after antibiotic removal, increasing the overall survival rate in comparison to bulk culture [52]. van Vliet et al. also demonstrated that there are spatial patterns to gene expression and SOS induction within a population of E. coli cells imaged with fluorescence microscopy. The authors determined these patterns to be due to the combined effects of shared lineages as the cells divide as well as cell to cell interactions [53]. Another study found that the general stress response sigma factor in B. subtilis typically fluctuates in single cells but forms a spatial pattern of expression within biofilms in response to nutrient gradients [54]. Furthermore, metabolic heterogeneity and diversity in redox states of cells within biofilms have been shown to increase antibiotic tolerance in Pseudomonas aeruginosa [55], [56]. Together these studies begin to link single cell stress mechanisms to population level responses in biofilms.

Horizontal gene transfer affects population survival

Horizontal transfer of genetic elements between bacterial cells in a community drives evolution and the spread of resistance [4]. Stalder et al. observed a higher rate of mutation and plasmid maintenance in Shewanella oneidensis biofilms compared to planktonic culture [57]. Plasmids often carry resistance genes so their maintenance can be beneficial to cells for future encounters with antibiotics, and increased mutagenesis creates a more heterogenous population that can adapt quickly to diverse stressors. To examine how genetic elements can spread between cells in a population, Woods et al. carried out experimental evolution and horizontal gene transfer in a Helicobacter pylori community with genetic material that increases antibiotic resistance. They found that the resistance gene propagated at low rates in the population prior to selective pressure from antibiotics. This pre-propagation led to survival of the population when it was subsequently treated with metronidazole [58].

Horizontal gene transfer is known to be associated with multidrug resistant clinical infections [59], but it can also help bacteria survive in new environments. Frazao et al. demonstrated the importance of horizontal gene transfer for new E. coli strains colonizing the mouse gut. They introduced an invading E. coli strain to mice and let it colonize the gut alongside the resident gut microbiota. Afterwards they sequenced the genome of the invading strain and found mutations in prophage regions of the DNA carrying genes that allowed for increased ability to metabolize resources found in the gut [60]. Other researchers are working on methods to model the transfer of mobile genetic elements based on their function using machine learning and bioinformatic toolboxes [61], [62]. These platforms could be used to predict the spread of mobile genetic elements in various environments and better understand their roles in stress survival.

Interspecies interactions alter stress responses

While transfer of genetic material from bacteria in the gut can help new strains colonize, existing commensal bacteria can also block colonization by an invading species. Cooper et al. demonstrated how a strain of Acinetobacter used a Type 6 secretion system to lyse neighboring E. coli cells and then take up plasmid DNA released from the lysed cells. This enhanced the rate of horizontal gene transfer and is part of the reason Acinetobacter is rapidly increasing in lethal infections [63]. Increased knowledge of these types of community interactions has allowed for the development of therapeutic probiotics to aid in antibiotic treatment. In one study, Hare et al. demonstrated that the presence of a probiotic E. coli Nissile strain prevented the revival of persistent E. coli MG1655 cells after antibiotic treatment. They found this inhibition to be contact-dependent and mediated by tolC in the persister cells [64].

Interactions and cross-feeding between bacterial species in a community can also affect cell fitness and alter susceptibility to antibiotics and other stressors [65]–[67]. Biofilms contain nutrient and oxygen gradients that result in different growth patterns across cells within the biofilm. Díaz-Pascual et al. determined that cross-feeding within colonies of E. coli on solid minimal medium counteracts the change in oxygen supply and proximity to glucose throughout the colony [68]. In addition, the spatial arrangements and distance between strains in mixed communities influence the effect they have on each other, stress tolerance, and antibiotic resistance [45], [69], [70]. Yu et al. demonstrated that microbial species that interact and exchange metabolites within a community are more tolerant to antimicrobial drug exposure [71]. Co-infection of pathogenic bacteria with commensal species can also alter the virulence by cross-feeding or cross-respiratory effects. One study found that commensal Streptococcus gordonii increased oxygen concentrations during infection, which in turn increased growth and persistence of pathogenic Aggregatibacter actinomycetemcomitans [72]. These studies highlight the effect of community composition on stress response and the importance of studying these effects to understand how natural communities of bacteria can survive changing environments, host immune responses, and antibiotic treatment.

Conclusion

Microbial stress responses play a key role in the dynamics of natural microbial communities, where cells must cope with nutrient starvation, chemical stress, antibiotics, and toxins from hosts or other microbes. Here, we present recent advances in understanding these stress responses and emerging tools that can be used to further probe this space. We highlight how connections between single-cell heterogeneity, biofilm composition, and interactions within microbial communities work together to govern stress response and survival. Phenotypic and metabolic heterogeneity prior to stress and alteration of gene expression during stress both play major roles in determining population level survival. In the near future, we anticipate that improvements in imaging communities such as biofilms will help to reveal how the extracellular matrix, cross-feeding, and horizontal gene transfer layer onto single cell dynamics to determine overall population survival. Together, these results will improve our understanding of how single cells work within communities to assemble a coordinated response to stress. These findings can have broad implications, for example by uncovering ways that bacteria evade antibiotic treatment or how gut microbiota defend against invading pathogens.