Elsevier

Current Opinion in Microbiology

Volume 27, October 2015, Pages 114-120
Current Opinion in Microbiology

Applications of imaging for bacterial systems biology

https://doi.org/10.1016/j.mib.2015.08.003Get rights and content

Highlights

  • Bacterial cell biology has powerful synergies with systems biology.

  • Single-cell and single-molecule imaging reveal the robustness of cell-wall growth.

  • Genetic pathways and drug targets can be dissected using fluorescence imaging.

  • Spatial organization dictates the structure and function of bacterial communities.

Imaging has fueled exciting advances in bacterial cell biology, which have led to exquisite understanding of mechanisms of protein localization and cell growth in select cases. Nonetheless, it remains a challenge to connect subcellular dynamics to cellular phenotypes. In this review, I explore synergies between imaging and systems approaches to bacterial physiology. I highlight how single-cell, time-lapse imaging under environmental or chemical perturbations yields insights that complement traditional observations based on population-level growth on long time-scales. Next, I discuss applications of high-throughput fluorescence imaging to dissect genetic pathways and drug targets. Finally, I describe how confocal imaging is illuminating the role of spatial organization in the structure and function of bacterial communities, from biofilms to the intestinal microbiota.

Introduction

Spatiotemporal organization occurs over a wide range of length (from molecular assemblies to multicellular communities) and time (from milliseconds to days) scales in bacteria. To achieve a systems-level understanding of cellular growth, structure, and function, it will be necessary to elucidate how cells establish and coordinate processes over these scales. Direct observation of cell growth and intracellular protein dynamics provides powerful insight into these questions. Every such experiment addresses a system that involves a host of biochemicals: not only do cells often rely on sequential assembly of many proteins to establish subcellular structures, but localization is defined with reference to cell geometry, which in bacteria is determined by the membrane and peptidoglycan cell wall [1]. Moreover, cell shape changes as the cell grows, connecting protein localization to all of the processes essential for proliferation, such as DNA replication, transcription, and translation. Therefore, there is ample potential for synergy in integrating approaches from cell biology and systems biology.

To elucidate genotype–phenotype relationships at the genomic scale, recent chemical genomics studies [2] have exploited high-throughput measures of colony features as a measure of growth. By varying the growth environment with chemical treatments, substrates, and environmental conditions, multidimensional phenotype vectors describing a collection of mutants can be used to study gene essentiality and understand drug action [3]. Moreover, application to genomic-scale libraries empowers clustering of genes involved in a common pathway [3], thereby realizing a systems-level perspective on cellular functions. Nonetheless, a population-level growth metric such as colony size presents only a single perspective on fitness, and will not always accurately represent the effect of perturbations (be they genetic, chemical, environmental, or other) on bacterial physiology.

There is a growing consensus emerging across biological systems that physical metrics such as cell shape, which by itself provides many phenotypes such as cell width, length, body curvature, and polar morphology, are linked with fitness in meaningful ways. Hence, high-throughput analysis of these metrics across genetic libraries would complement data obtained through chemical genomics. In some bacteria, nutrient-induced increases in growth rate correlate with increased cell volume [4]. Similarly, over the course of a long-term evolution experiment, Escherichia coli average cell volume increased approximately twofold as fitness also increased [5]. Single point mutations in the E. coli actin homolog MreB have been found to increase cell size to a similar degree [6, 7], with mutations at one particular residue causing a width-dependent increase in competitive fitness during growth; interestingly, this fitness increase was realized only in the presence of certain carbon sources [6], underscoring the importance of probing multiple environments to understand the relationship between cellular physiology and physical features such as cell size. Moreover, population-based measurements often conceal phenotypic heterogeneities in single-cell growth rate and morphology. In the yeast Saccharyomyces cerevisiae, high-throughput imaging revealed that deletion of many genes increases morphological variation [8]. Salmonella Typhimurium persister cells, which represent a small fraction of the population, have been associated with slow growth rates and smaller size [9] indicating that heterogeneities can be crucial for a community to survive fluctuating environments or stresses. Thus, cell shape and other single-cell features can provide terminal or time-dependent phenotypes in chemical screens that increase the resolution of our understanding of gene function.

Fluorescence imaging and multiparameteric image analysis can also be used to interrogate cellular processes at the genomic scale, particularly to identify new functions for genes. An imaging-based screen of a Drosophila RNA interference (RNAi) library was used to identify >200 genes responsible for mitotic spindle assembly [10]. Key to this study was a high-throughput imaging and quantitative analysis platform that allowed for an unbiased screen across the entire genome, as over half the identified genes were unexpected. Automated analysis of localization dynamics and abundance of fluorescent fusions to ∼3000 S. cerevisiae proteins was used to reveal fluxes between cellular compartments in response to chemical and genetic perturbations [11••]. Unbiased imaging screens may be similarly revelatory in bacteria. Characterization of the localization of fluorescent-protein fusions to most genes in the bacterium Caulobacter crescentus revealed that ∼10% of the genes have nonuniform localization profiles [12] and uncovered several novel structures including filaments of the metabolic enzyme CTP synthase [13]; this aggregation may be important for efficient channeling of metabolic intermediates [14].

Here, I explore how recent imaging-based initiatives have empowered the investigation of bacterial systems biology across spatial and temporal scales. I also discuss applications of high-throughput fluorescence imaging for dissecting genetic pathways and drug actions, and for interrogating the spatial organization and function of bacterial communities.

Many environmental stresses prompt system-wide responses that affect the rate of cell growth. Thus, assaying changes in growth after hours or days likely conflates many indirect effects and hence may motivate incorrect models of growth regulation. For instance, bacterial growth generally decreases as the osmolarity of the growth medium is increased [15], prompting the hypothesis that turgor pressure directly stimulates growth by driving the mechanical expansion of the cell wall. To distinguish between turgor-mediated effects and indirect, pressure-independent effects of osmolarity changes, a microfluidic flow cell was used to rapidly change osmolarity while quantifying the instantaneous elongation rate via single-cell imaging [16••]. On short time scales, although plasmolysis slowed cell elongation, cells nevertheless ‘stored growth’ whereby, upon reestablishment of turgor, they expanded to the length that they would have attained without the osmotic shocks (Figure 1a) [16••]. These experiments reveal a surprising robustness of cell-wall synthesis to turgor fluctuations that is concealed by the decrease in steady-state growth that occurs on longer time-scales [15, 16••], and highlight the utility of time-lapse imaging of growth responses to perturbations.

Another illustrative example of the power of imaging for unraveling the systems biology of bacteria at various time scales involves the assembly of protein complexes. Although such complexes are thought to facilitate the spatial and temporal coordination of multiple enzymes, it is unclear whether complexes are static or dynamic, and whether either has physiological consequences. Förster resonance energy transfer (FRET) is an imaging-based tool for quantifying the propensity for physical interactions between proteins, and has been employed to map all protein interactions within the E. coli chemotaxis pathway [17] including measurements of binding dynamics [18]. Cell-wall synthesis requires the functions of numerous penicillin binding proteins (PBPs) [1] including PBP2; antibiotic inhibition of PBP2 results in the loss of rod-shaped morphology and eventual cell lysis [19]. On the basis of the observation that circumferential motion of the E. coli actin homolog MreB is dependent on PBP2 activity [20], it was hypothesized that a static complex of PBPs moves with MreB. However, single particle-tracking photoactivated localization microscopy revealed that PBP2 diffuses rapidly [21••]. Moreover, cells maintained the same instantaneous growth rate for several generations as PBP2 was depleted (Figure 1b), suggesting that dynamic association between MreB and PBP2 increases the robustness of growth rate to fluctuations in PBP2 concentration relative to the formation of a stable multienzyme complex [21••]. Probing other systems via single-molecule imaging will be crucial for determining whether dynamic association is a general paradigm for multienzyme processes, particularly those that globally distribute their activity within the cell.

Given that a sizeable fraction of bacterial proteins [12, 22] exhibit localization to specific subcellular locations, high-throughput imaging of localization dynamics holds intriguing prospects for scrutinizing the systems-level establishment of cellular structure and subcellular localization. A recent study used epifluorescence time-lapse imaging to quantify the spatiotemporal dynamics of nearly every E. coli protein with a nonuniform localization pattern [23••]. This strategy allowed for the quantification of consensus localization patterns across the cell cycle (Figure 2a), which identified several behaviors that would have been obscured by averaging single-cell fluorescence profiles from snapshots across a population [23••]. These behaviors included several classes of localization patterns of DNA-binding proteins, the relative timing of localization of division-related proteins, and asymmetric protein partitioning after cell division [23••], illustrating the effectiveness of fluorescence dynamics as an information-rich phenotypic readout.

Traditionally, investigations of antibiotic mechanisms of action have relied on biochemical assays to determine binding to specific types of molecules (DNA, RNA, protein, etc.) or on the isolation of resistant mutants [24]. However, even if these laborious methods are successful, interactions with other cellular components can alter or obscure the cellular targets. A potent imaging-based methodology is bacterial cytological profiling, wherein changes in morphological features and fluorescence signals are used to cluster antibiotic treatments (Figure 2b) [25••]. For example, subcellular properties such as nucleoid structure display large qualitative differences depending on the antibiotic target, from toroids under treatment with tetracycline (a translation inhibitor) to decondensation with rifampicin (a transcription inhibitor) (Figure 2c). By exploiting just a few variables such as the localization of membrane, DNA, and permeability stains [25••], this method can categorize inhibitors based on five major target pathways to generate hypotheses for the targets of antibiotics with unknown mechanisms of action [25••]. In this way, fluorescence imaging can be used similarly to chemical genomics to establish phenotypic profiles, and the combination of these methods could be used in the future to investigate specific pathways.

Confocal imaging makes it possible to correlate molecular processes beyond single cells to the community scale, which is a grand challenge in systems biology. Biofilms are composed of well-defined three-dimensional structures such as wrinkles, concentric rings, and ridges, the formation of which depends on the secretion of extracellular-matrix components [26]. Confocal imaging of slices through an E. coli biofilm revealed that these components take on a heterogeneous spatial pattern (Figure 3a) that is correlated with the differentiation of cells into a two-layer architecture with growing, flagellated cells near the agar surface and a thick layer of matrix-embedded stationary-phase cells adjacent to the air interface (Figure 3b) [27••]. This differentiation is regulated by a collection of molecules including the stationary phase sigma factor RpoS [28] and the biofilm-promoting second messenger cyclic-di-GMP [29], demonstrating that imaging of community architecture provides a window into single-cell gene expression that varies according to the location of each cell in the community.

The diverse bacterial community within the gastrointestinal tracts of mammals has generally been investigated via compositional enumeration using sequencing of fecal material [30, 31, 32, 33, 34, 35]. However, these population-level measurements ignore the importance of spatial structure in the function of this complex ecosystem, which has multiple length scales spanning micron-sized cells to millimeter-sized tissues. Confocal imaging has provided a high-resolution view of the interactions among bacteria, host cells, and the structure of the local environment. Colonization of some species is maintained by penetration into colonic mucus and crypt channels (Figure 3c), suggesting that cellular localization within the gut can exert profound effects on microbiota composition [36]. This localization is determined through a molecular conversation with host cells. The antibacterial lectin RegIIIγ is essential for maintaining a zone at the surface of the small intestine into which the microbiota do not trespass; RegIIIγ regulates both the proximity and identity of bacteria closest to the host epithelium (Figure 3d) [37]. Given the heterogeneity of the intestinal environment, techniques that allow for the imaging of intact gut samples, together with computational tools that can quantify in an unbiased manner features of community organization such as distances among bacterial cells and to host landmarks such as the epithelium, will be required to fully illuminate the structure and function of the microbiota.

Section snippets

Discussion

Looking forward, bacterial cell biology appears poised to capitalize on new imaging techniques, and on advances in computational methods to analyze the imaging data. To complement such developments, genetic advances must be made in tandem. A sizeable challenge for dissecting the function(s) of protein systems has been the major, sometimes catastrophic, phenotypes resulting from the loss of function of any part of these systems. For example, deletion of most of the components of the cell-wall

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

  • • of special interest

  • •• of outstanding interest

Acknowledgements

The author thanks Amanda Miguel and Enrique Rojas for helpful feedback. This work was funded by NSF CAREER Award MCB-1149328.

References (48)

  • S.F. Levy et al.

    Network hubs buffer environmental variation in Saccharomyces cerevisiae

    PLoS Biol

    (2008)
  • S. Helaine et al.

    Internalization of Salmonella by macrophages induces formation of nonreplicating persisters

    Science

    (2014)
  • G. Goshima et al.

    Genes required for mitotic spindle assembly in Drosophila S2 cells

    Science

    (2007)
  • Y.T. Chong et al.

    Yeast proteome dynamics from single cell imaging and automated analysis

    Cell

    (2015)
  • M. Ingerson-Mahar et al.

    The metabolic enzyme CTP synthase forms cytoskeletal filaments

    Nat Cell Biol

    (2010)
  • M. Castellana et al.

    Enzyme clustering accelerates processing of intermediates through metabolic channeling

    Nat Biotechnol

    (2014)
  • J. Christian et al.

    Water relations of salmonellae at 30 °C

    Aust J Biol Sci

    (1953)
  • E. Rojas et al.

    Response of Escherichia coli growth rate to osmotic shock

    Proc Natl Acad Sci U S A

    (2014)
  • D. Kentner et al.

    Dynamic map of protein interactions in the Escherichia coli chemotaxis pathway

    Mol Syst Biol

    (2009)
  • V. Sourjik et al.

    Binding of the Escherichia coli response regulator CheY to its target measured in vivo by fluorescence resonance energy transfer

    Proc Natl Acad Sci U S A

    (2002)
  • T.B.P. Ogura et al.

    Penicillin-binding protein 2 is essential in wild-type Escherichia coli but not in lov or cya mutants

    J Bacteriol

    (1989)
  • S.v. Teeffelen et al.

    The bacterial actin MreB rotates and rotation depends on cell-wall assembly

    Proc Natl Acad Sci U S A

    (2011)
  • T.K. Lee et al.

    A dynamically assembled cell wall synthesis machinery buffers cell growth

    Proc Natl Acad Sci U S A

    (2014)
  • M. Arifuzzaman et al.

    Large-scale identification of protein–protein interaction of Escherichia coli K-12

    Genome Res

    (2006)
  • Cited by (0)

    View full text