This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Brehm-Stecher, B. F. Articles by Johnson, E. A. Search for Related Content PubMed PubMed Citation Articles by Brehm-Stecher, B. F. Articles by Johnson, E. A.

Single-Cell Microbiology: Tools, Technologies, and Applications

Byron F. Brehm-Stecher^1, and Eric A. Johnson^1,2*

Department of Food Microbiology and Toxicology, Food Research Institute,¹ Department of Bacteriology, University of Wisconsin—Madison, Madison, Wisconsin²

SUMMARY

INTRODUCTION

MICROBIAL HETEROGENEITY

Genetic Heterogeneity

Biochemical or Metabolic Heterogeneity

Physiological Heterogeneity

Behavioral Heterogeneity

ADVANTAGES OF SINGLE-CELL APPROACHES

Revealing Cryptic Processes

Observing Discrete and Dynamic Events within Living Cells

Relating Microscopic, Mesoscopic, and Macroscopic Observations

A Caveat: the ''Uncertainty Principle''

TOOLS AND TECHNOLOGIES

Fluorescence

Fluorescent dyes and stains.

Fluorescence in situ hybridization and immunofluorescence.

Green fluorescent protein and related reporters.

Cytometry

Flow cytometry.

Laser scanning cytometry.

Image cytometry.

Scanning Probe Microscopies

Atomic force microscopy.

Scanning electrochemical microscopy.

Microspectroscopic Methods

Raman microspectroscopy.

Microbeam analysis.

Electrorotation.

Micromanipulation

Mechanical micromanipulation.

Optical micromanipulation.

Electrokinetic micromanipulation.

Microcapillary Electrophoresis

Biological Microelectromechanical Systems

CONCLUSIONS AND FUTURE PERSPECTIVES

AFTERWORD

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
The field of microbiology has traditionally been concerned with and focused on studies at the population level. Information on how cells respond to their environment, interact with each other, or undergo complex processes such as cellular differentiation or gene expression has been obtained mostly by inference from population-level data. New appreciation for the existence and importance of cellular heterogeneity, coupled with recent advances in technology, has driven the development of new tools and techniques for the study of individual microbial cells. As a result, scientists have been able to characterize microorganisms and their activities at unprecedented levels of detail.

Single-cell techniques have been used to more fully describe the environmental distribution and activities of microorganisms, have been a key element in revealing otherwise invisible processes such as interspecies gene transfer and chemical communication, and have been used to detail discrete physicochemical interactions between microbes and the surfaces they colonize (11, 31, 60, 150, 231). Single-cell methods have also been essential to our understanding of connections between cellular biochemistry and behavior and of the cellular bases of population-level phenomena (58, 126, 248). As a result, new insights into the properties of chemical signaling pathways and mechanisms behind the coordination of multicellular behaviors have been possible. Single-cell methods have also enabled direct micro- or nanoscale measurements of the mechanical properties of individual cells, including turgor pressure, elasticity, and bursting force (218, 257). Microphysiological studies of metabolite, protein, or elemental localization, intracellular water dynamics, host-pathogen interactions, and surface-associated redox activity represent other areas in which single-cell techniques have been applied (12, 13, 45, 178, 183, 187).

Apart from enabling fresh perspectives on issues of concern to basic science (58, 183), the tools and technologies of single-cell microbiology have been brought to bear on problems of direct interest to researchers in applied science. Individual microorganisms, even those in "clonal" populations, may differ widely from each other in terms of their genetic composition, physiology, biochemistry, or behavior (40, 66, 75, 127, 208). This variability or heterogeneity has important practical consequences for a number of human interests, including antibiotic and biocide resistance (21, 228, 237), the productivity and stability of industrial fermentations (184, 205, 229), the efficacy of food preservatives, (17, 223, 229), and the potential of pathogens to cause disease (67). Additionally, methods for identification, characterization, and/or physical separation of individual microorganisms are needed for the detection of pathogens and for the identification and selection of strains with beneficial or improved properties (124, 224).

Because studies made at the single-cell level are not subject to the averaging effects characteristic of bulk-phase population-scale methods, they offer a level of discrete microbial observation that is unavailable with traditional microbiological methods. Single-cell techniques have been key in probing microbial viability phenomena that are beyond the resolution of culture-based approaches (22, 125, 194), in elucidating mechanisms of pathogenesis (12, 136, 227), and in measuring the motility and the invasive forces of individual cells or hyphae (26, 162, 200).

This paper reviews some of the tools and technologies available for the study of microbes at the level of the single cell. Special interest is given to methods capable of monitoring discrete and dynamic processes occurring within living microbial cells. The limitations of traditional, population-based microbiological techniques as the motivation for the development of these single cell approaches are discussed throughout. Several of the tools and technologies discussed here have themselves been the subjects of more specialized reviews, to which the reader is referred for more detailed information. Although single-cell microbial phenomena have received attention in the past, recent advances in technology have enabled unprecedented access to processes occurring at this scale. Because our primary focus is on these recent technological advances, a historical perspective is beyond the scope of this review.

MICROBIAL HETEROGENEITY

Top Previous Next References

 
Variability is a hallmark of biological systems. Microbial cells have a remarkable capacity for displaying a multitude of genetic and nongenetic differences from each other. This inherent genetic and phenotypic plasticity forms the basis of a successful "lifestyle strategy" that enables them to adapt to and survive adverse conditions or to persist and cause disease (36, 97, 127). The central theme driving the need for methods capable of resolving the properties and activities of individual microbial cells is that of microbial heterogeneity (66, 208). Bulk-scale measurements made on a heterogeneous population of cells report only average values for the population and are not capable of determining the contributions of individual cells. However, properties such as viability, protein concentration, possession of a mutant allele, or the number of flagella expressed on the cell surface are discrete and intrinsic states or properties of each individual cell. Methods capable of analyzing these properties at the level of the individual cell enable a more complete understanding of phenomena that are inaccessible to researchers using population-scale approaches.

The types of individual differences contributing to heterogeneity within a microbial population can be divided into at least four general classes: genetic differences, biochemical differences, physiological differences, and behavioral differences. The lines dividing different modes of heterogeneity are often nebulous and interactive. For example, biochemical or behavioral differences might ultimately be traced back to a genetic basis. Even physiological heterogeneity, which may be driven by forces external to the cell (e.g., nutrient limitation or the presence of antibiotics), could be viewed in terms of the organism's genetic potential to respond to these forces. However, the choice of tools used to explore cellular differences often makes it operationally clear which source of heterogeneity is the subject of investigation. For example, genetic heterogeneity is addressed using methods such as single cell PCR or fluorescence in situ hybridization (FISH), biochemical heterogeneity is measured using enzyme assays or single-cell electrophoretic separations, and behavioral heterogeneity is measured through direct observation of cellular responses to various stimuli. Examples of how individual microbial cells may vary according to their genetic, biochemical, physiological, or behavioral properties are described briefly below.

Asymmetries in the distribution of genetic material between daughter cells may be important in driving processes of differentiation, as has been suggested for the strand-specific imprinting of mating-type switching in Schizosaccharomyces pombe (62). Processes related to cell aging, including the accumulation of DNA damage or variability in gene expression and loss of gene silencing, may also be used to describe genetic variability between individual microbial cells (92, 184). Other sources of cellular heterogeneity are discussed briefly below. These can be described as nongenetic or phenotypic in nature (229).

ADVANTAGES OF SINGLE-CELL APPROACHES

Top Previous Next References

 
Plate counting and light microscopy represent the original set of tools available for single-cell analyses (168). As such, they have been remarkably useful for more than 100 years, and for many applications, they remain both adequate and appropriate (40, 168). However, the past few decades have been marked by the introduction of a number of technological and methodological innovations, including advances in computing or imaging technologies and the development of culture-independent methods such as in situ hybridization and PCR. Progress in these areas has dramatically advanced our abilities to resolve the features and activities of individual microbial cells. Examples of some of the types of information that have been made more accessible through the use of single-cell approaches are introduced below.

TOOLS AND TECHNOLOGIES

Top Previous Next References

 
A broad overview of the tools and technologies available for resolving the properties and activities of single microbial cells is provided below. Table 1 highlights the range of studies in which these tools and technologies have been applied. Because fluorescence is of fundamental importance to many of the approaches used to investigate single-cell microbial phenomena, additional background has been included on this concept.

The fundamental principles of fluorescence have been reviewed extensively elsewhere (66, 123, 208), as have many of the staining techniques applicable to microbial cells (66, 102, 119, 208, 244). An excellent historical account of developments in fluorescent-dye technology is also given by Kasten (123).

Briefly, fluorescence occurs after photons from an incident light source raise electrons in a fluorophore (in many cases an organic molecule with multiple, conjugated double bonds) to a higher-energy or "excited" state. Return of the molecule to a lower-energy state is accompanied by the emission of light as fluorescence after some energy loss (66, 123, 208). Fluorescence is emitted at a lower energy (e.g., longer wavelength) than that of the original excitation light, and the difference in excitation and emission wavelengths is termed the "Stokes shift" (66, 123, 208). The magnitude of the Stokes shift can be critical in ensuring spectral separation of signals from more than one fluorescent stain or when dealing with cells or sample matrices having highly autofluorescent backgrounds. Variables of practical importance to fluorescence include the intrinsic properties of the fluorophore: its excitation and emission spectra, molar absorbance coefficient, quantum yield, quantum efficiency, and photostability (66, 102, 123, 208). The local chemical or electronic environment also plays a role, and factors such as pH, the physical proximity of other molecules in solution, and the presence of localized charge concentrations (e.g., the negatively charged backbone of DNA) can all affect the resulting fluorescence (123, 208, 217).

Fluorescent dyes and stains. Fluorescent dyes with affinities for all of the major macromolecules occurring within microbial cells are commercially available (102). These include stains that react with nucleic acids, proteins, or lipids or that stain polyester or polyphosphate inclusion bodies. Additionally, fluorescent enzyme or respiratory substrates, reporters of intracellular pH or ion concentration, and dye kits providing "fluorescent Gram staining" are available (66, 102, 244). The performance of these commercial kits is often validated using specific microorganisms grown under standardized conditions. However, if these assays are to be used with different microorganisms or natural populations, they must be revalidated under the new conditions, since basic physiological differences or increased biochemical heterogeneity within these populations may complicate data analysis (209). The difficulties in transferring multiparameter staining protocols across generic or species boundaries may be even more pronounced (210).

Macromolecules such as lectins, antibodies, and nucleic acid probes may be labeled with fluorescent dyes to create conjugates capable of reporting molecular recognition events. Other fluorescence-based molecular methods, such as in situ PCR or in situ reverse transcription, may result in the incorporation of fluorescently labeled deoxynucleoside triphosphates into reaction products as they are formed within the cell (104). Endogenous sources of fluorescence, including carotenoids, tryptophan, thiamine (after chemical derivitization), and the cell's own photopigments, may also serve as reporter molecules, for instance in industrial or environmental applications (9, 10, 117, 124).

Dynamic microbial phenomena, including protein expression and behavior (187), substrate uptake (167), binding and release of individual chemoattractant molecules to cell surface receptors (239), selective degradation of uniparental DNA within newly formed algal zygotes (170), bacterivory (96), and drug efflux (21, 118), may also be observed or measured at the single-cell level through the use of fluorescence staining techniques. Specialized techniques such as fluorescence ratio imaging microscopy may provide insights into dynamic cellular events that are important to the outcome of microbial fermentations (214), which highlight the physiological responses of spoilage organisms to chemical stresses (17), or that are related to cellular inactivation resulting from treatment with antimicrobials (42) (Fig. 1).

View larger version (8K): [in this window] [in a new window]   FIG. 1. Fluorescence ratio imaging of nisin-mediated dissipation of pH in Listeria monocytogenes. Fluorescence ratio imaging microscopy was used to monitor the intracellular acidification of broth-grown cells of L. monocytogenes after exposure to the membrane-permeabilizing lantibiotic nisin. The pH-dependent spectral response of carboxyfluorescein diacetate succinimidyl ester (CFSE) was used as a probe of intracellular pH (pHi). The ratio of CFSE fluorescence intensity at 490 nm to that at 435 nm was calibrated over a pH range of 5.0 to 9.0. (A) Live, intact cells of L. monocytogenes maintained pHi values between 8.0 and 8.4, even when the pH of the external medium was low (e.g. pH 5.5). (B) Nisin-mediated membrane permeabilization resulted in the equilibration of pHi with the pH of the medium after 12 min of exposure. Individual cellular responses were more heterogeneous for cells derived from colonies, suggesting the importance of microenvironmental factors in differential susceptibility to nisin (not shown). A color-coded pH scale is shown in the upper right-hand corner. Reprinted from reference 42 with permission from the publisher.

View larger version (63K): [in this window] [in a new window]   FIG. 2. Flow cytometric analysis of a genetically and metabolically complex cell mixture. This figure illustrates the power of single-cell staining methods in combination with flow cytometric analysis for the fluorescent "dissection" of complex microbial populations. Here, Salmonella enterica serotype Typhimurium is differentiated from a mixture of E. coli, Citrobacter freundii, Proteus vulgaris, and Shigella dysenteriae on the basis of both cytochemical activity (CTC staining) and genetic identity (Salmonella-specific FISH staining). A complex mixture containing both live and formalin-killed representatives of each cell type was incubated with CTC, fixed with 10% buffered formalin, hybridized with a Salmonella-specific DNA probe (Sal3-Cy5), and examined by flow cytometry. Four distinct populations can be seen. Clockwise from the bottom left, they are dead non-Salmonella members of the Enterobacteriaceae (A), live non-Salmonella members of the Enterobacteriaceae (B), live Salmonella (C), and dead Salmonella (D). The numbers in each quadrant represent percentages of the total population. The photographic inset provides a visual interpretation of the cytometry data. Reprinted from reference 209a with permission from the publisher.

Fluorescence in situ hybridization and immunofluorescence. The principles behind the use of FISH and immunofluorescence methods have been extensively and informatively reviewed elsewhere (6, 7, 66, 164). In the FISH technique, fluorescently labeled nucleic acid probes are hybridized to complementary rRNA targets located on ribosomes within whole, permeabilized cells. The ribosome is a naturally amplified target molecule, especially in actively growing cells, where each cell may contain several thousand ribosomes (7). The aggregate signal from multiple probe-ribosome binding events leads to the sequence-specific fluorescence of target cells. Apart from rRNA, other forms of RNA (e.g., tmRNA) can serve as a target for hybridizations, especially if a signal amplification step is used (203). Recently, FISH-based methods have also been developed to detect low-copy-number targets on plasmid (10¹ to 10³ copies/cell) or chromosomal (<10 copies/cell) DNA (268). This approach differs substantially from rRNA-targeted FISH in that it utilizes polynucleotide probes (50 to 1,200 nucleotides in length), higher (1,000-fold) probe concentrations, and much longer hybridization times (268). The resulting fluorescent signal is also qualitatively different from that achieved with classic rRNA-targeted FISH and is characterized by the formation of a fluorescent "halo" around the periphery of target cells. The technique has thus been named RING-FISH.

Fluorescently labeled antibodies also enable the detection of diagnostic molecular binding events and can be directed against surface antigens, such as capsular, flagellar, or cell wall antigens, or against internal targets, including ribosomal proteins or cell cycle-specific cytoplasmic proteins (66, 194).

FISH and immunofluorescence have substantial overlap in their applications and benefits as single-cell detection techniques. Both are whole-cell methods, and as such they can preserve a wealth of potentially valuable information that is unavailable outside the context of the intact cell. Apart from providing information on microbial identity, information about cell morphology, number, and distribution may also be collected for specific target cells. Both methods have the potential to be carried out simultaneously or in succession with other means of cell characterization, including the observation of light-scattering characteristics, staining of inclusion bodies, fluorescence-based measurements of nucleic acid or protein content, cytochemical characterization using fluorescent or colorimetric enzyme substrates, and microautoradiography (22, 39, 140, 249). The combination of FISH or fluorescent-antibody labeling with methods for high-throughput multiparametric data collection, analysis, and sorting (e.g., flow cytometry) can be especially useful in the study of complex microbial populations (39, 66).

FISH is used primarily as a means of detecting specific microbial cells, although the intensity of staining with FISH has also been used to provide an indication of physiological activity (146). For the most part, the use of FISH for microbial detection has involved DNA-based methods, but peptide nucleic acid probes may have substantial practical and functional advantages, especially for the detection of gram-positive bacteria (39, 224).

As a means of detection, fluorescent-antibody approaches can be limited by problems with cross-reactivity, variable antigen expression under different culture conditions, or potential instability and loss of cell surface epitopes (166). However, unlike FISH, immunofluorescence-based detection does not require cell permeabilization and can therefore be used on living cells, potentially followed by isolation for culture (66, 244). Apart from their use as taxonomic probes, fluorescent antibodies may be used for fine-structure analyses, such as the discrete localization of specific proteins within individual cells.

Neither FISH nor immunofluorescence approaches require that a cell be culturable (4, 66). However, because the number of target antigens may not be as tightly coupled to the cell growth rate as is the rRNA copy number, fluorescent-antibody techniques may yield higher detection sensitivities for dormant cells than FISH does.

Green fluorescent protein and related reporters. GFP is a versatile tool for the in vivo visualization of protein expression, localization, and functionality. Because it retains its fluorescence after fixation with paraformaldehyde, GFP can be combined with fixation-dependent staining methods such as FISH (71). However, the true power of GFP is as a visual reporter of dynamic events occurring in living cells. For example, Raskin and de Boer (187) used GFP fusions to probe the function of proteins associated with cell division in Escherichia coli. They observed a regular, pole-to-pole oscillation for GFP-MinD and theorized that the cell may use this protein as a "measuring device" to continuously probe the location of the center of the cell. Cluzel et al. (58) used a cheY-gfp fusion, fluorescence correlation spectroscopy, and video microscopy to relate CheY-GFP expression levels to flagellar-rotation behavior in single cells of E. coli. These authors found that small changes in the concentration of CheY-P led to large changes in the rotational bias of the flagellar motor, suggesting that the motor itself acts as a signal amplifier and that additional cellular mechanisms exist for maintaining CheY-P concentrations within the operational range of the motor (58).

Other applications of GFP include the construction of whole-cell sensors for in situ monitoring of iron availability on leaf surfaces (120); measurement of cytoplasmic viscosity and protein diffusion rates in living cells (74, 182); investigation of quorum-based interspecies communication or coordinated, multicellular behaviors (11, 116, 126, 248); measurement of the internal pH of bacterial cells (172); and real-time reporting of fungal susceptibility to antimicrobial compounds (247).

GFP is especially well suited to in situ analyses of individual cells within complex consortia such as biofilms. Because its use does not require preparative steps such as dehydration, fixation, or application of exogenous probes or cofactors, GFP labeling enables the observation of microorganisms directly in these fragile structures (35, 67, 190). The range of applications of GFP has been further expanded with the introduction of fluorescence-shifted spectral variants. In a novel application of such variants, Fehr et al. (78) created chimeric "nanosensor" proteins based on the fusion of enhanced cyan fluorescent protein (ECFP), a bacterial maltose binding periplasmic protein, and enhanced yellow fluorescent protein (EYFP). Conformational changes of these nanosensors on binding of maltose led to more efficient fluorescence resonance energy transfer (FRET) from ECFP to EYFP. When these nanosensors were expressed in S. cerevisiae, changes in ECFP/EYFP FRET ratios enabled maltose uptake and compartmentation to be monitored in individual living cells. The broad range of organic and inorganic substrates recognized by periplasmic binding proteins suggests the use of this strategy in generating fluorescent nanosensors specific for a wide variety of analytes (78).

Stochasticity, or noise, in gene expression can lead to substantial phenotypic variation among individual cells in an otherwise clonal population (75). Such noise can be either instrinsic (stemming directly from events related to the expression of a gene) or extrinsic (resulting from fluctuations in the quantities or activities of the enzymes and other cellular machinery required for gene expression). In a novel application of GFP variants, Elowitz et al. (75) constructed strains of E. coli capable of distinguishing between these two sources of noise in gene expression. Their results indicated that both sources of noise contribute to the generation of phenotypic heterogeneity among individual cells. Their findings also suggested that any component in a cellular biochemical network that is prone to intrinsic fluctuations in concentration can serve as a source of extrinsic noise for other components in the network (75).

Flow cytometry. Flow cytometry is a powerful fluorescence-based diagnostic tool that enables the rapid analysis of entire cell populations on the basis of single-cell characteristics (4). Multiple characteristics, including cell count, cell size or content, and responses to fluorescent probes diagnostic of cell function may be collected simultaneously by this method (66, 244, 253). Cells in a liquid sample are passed individually in front of an intense light source (e.g., a laser, laser diode, or arc lamp), and data on light scattering and fluorescence are collected and saved as a data file. Detailed numerical analyses of populations and subpopulations of interest can then be carried out offline by using a number of analysis packages. Because of its capacity to collect information-rich data sets on thousands of cells, flow cytometry facilitates valuable insights into connections between single-cell and population-level processes not available with other techniques (66, 86, 119, 208, 243). Flow cytometers capable of sorting cells on the basis of their fluorescence characteristics or of simultaneous in-line video microscopy add to the versatility of this method (66, 250). Flow cytometry has proven to be an invaluable resource in the study of apoptosis in mammalian cells (64). Recent work has suggested that programmed cell death is not limited to eukaryotes but may also be active in prokaryotic systems (76, 191). Therefore, flow cytometry may also be a useful tool for elucidating these processes in bacteria.

Laser scanning cytometry. Flow cytometry collects data on single cells in a liquid sample as they stream past the illumination source. Although multiple light scatter and fluorescence parameters may be measured, cells pass only once through the system. Because of this, flow cytometry is not suited for time-resolved studies of individual cells (65, 68, 124, 208). An exception may be the microfluidic cell sorter described by Fu et al. (84), in which the fluid flow may be stopped or reversed, allowing multiple observations of the same cell, but this technology is not yet widely available.

Laser scanning cytometry (LSC) is a solid-phase cytometric technology for collecting laser-induced fluorescence from cell samples on slides or on membrane filters. At their simplest, LSC instruments provide a rapid means of counting, quantifying, and recording the distribution of fluorescent events on a filter. Microscope-based LSC instruments can provide visual information on both cell morphology and the spatial distribution of fluorescence within each cell (65). Because LSC can be used to make multiple measurements of the same cells, this technique is well suited for the observation of cellular properties as a function of time. Examples include monitoring the kinetics of fluorescence staining in living cells (e.g., substrate uptake, enzyme activity, and dynamic changes in intracellular pH) and observing interactions between neighboring cells (65, 68, 230). Spatial "addressing" of fluorescent events may facilitate the reexamination of archived samples (65). The ability to concentrate cells prior to analysis gives filter-based LSC methods definite practical advantages over fluorescence microscopy or flow cytometry when working with dilute suspensions of microorganisms in filterable liquids (146). However, because LSC may involve exposing the sample to the excitation source for relatively long periods, photobleaching of fluorescent labels could be problematic for some applications, particularly if multiple scans are required. These effects can, in part, be minimized through the use of low-intensity (microwatt versus milliwatt) illumination sources (68).

Image cytometry. The terms "image cytometry" and "image analysis" are used interchangeably here to describe a wide range of methods by which quantitative biological information may be extracted from microscopic images (Fig. 3). These techniques can be used to gain information on individual cell properties such as staining intensity and label specificity; cell number, size, and volume; and distribution within a field of view (73, 155, 96, 202). Advanced image analysis techniques can be used to monitor ultradiscrete physical phenomena such as the micronewton invasive forces generated by individual fungal appressoria (26).

View larger version (55K): [in this window] [in a new window]   FIG. 3. Single-cell determination of yeast glycogen content by image cytometry. The glycogen content of individual S. cerevisiae cells was determined from their optical density (OD) values after staining with Lugol's solution (I2-KI). Images were processed using a series of steps designed to extract quantitative information about cell size, shape, volume, and OD. The OD profile of cell A (glycogen poor) shows concentrated staining only on the periphery of the cell, whereas the profile of cell B (glycogen rich) shows dense staining throughout. To avoid overestimation of the mean glycogen content in glycogen-poor cells, only the central portion of each cell was used for measurement. The ability to quantitate the glycogen content in individual cells allows the determination of glycogen distribution within a population. Because the character of this distribution is related to yeast quality, image cytometry can be used as a tool for quality control. OD-L, optical density, Lugol staining. Reprinted from reference 44 with permission from the publisher.

Image collection is often followed by a number of processing steps designed to facilitate extraction of the desired information. These include thresholding, filtering, edge detection, removal of optical artifacts (e.g. fluorescent "halos"), background subtraction, pixel averaging, and other transformations (34, 54, 120, 202, 231). Advantages of such thorough image processing may include the ability to differentiate target cells from background material, particularly in "difficult" sample matrices such as soil (202). Through the use of special algorithms, images may be automatically processed on the basis of user-defined criteria or artificial neural networks may be trained for the automatic classification of objects (34, 44). Such automation can greatly aid image processing, especially where manual data extraction would be impossible, tedious, or error prone (61). Recently, a fully automated high-throughput microscopy system has been described that combines computer-controlled autofocusing and stage movement with advanced image segmentation, classification, and retrieval algorithms (185). With the ability to rapidly acquire and categorize data from large numbers of cells on slides or in microtiter plates, such high-throughput microscopy systems may eventually become competitive with flow cytometry as a method for the rapid and detailed analysis of populations on a cell-by-cell basis. However, due to the longer integration times often needed for imaging-based techniques, efforts must be made to minimize the effects of photobleaching, which are not a significant issue with flow cytometry.

In addition to single still images, multiple still images from a time series or video images may be analyzed. Video-based images, which allow the display of a time code with each frame, can provide a continuous record of a cellular measurement with high temporal resolution (158). Dynamic cellular properties associated with cell motility (cell speed, the number and duration of runs or tumbles, etc.) or changes in fluorescence related to some physiological characteristic can be resolved in terms of both space and time (Fig. 4) (38, 116, 159, 212). Movements of individual cells within a larger population can also be monitored, enabling connections to be made between cell behaviors at both microscopic (individual) and mesoscopic (population) scales (116, 126, 158, 190, 248).

View larger version (60K): [in this window] [in a new window]   FIG. 4. Single-molecule analysis of cyclic AMP (cAMP) receptor occupancy on the surface of a Dictyostelium discoideum cell during chemotaxis. Cells were exposed to a gradient of Cy3-labeled cAMP (Cy3-cAMP, shown to be functional as a chemoattractant) and observed for up to 10 min. Binding of Cy3-cAMP to cell surface receptors was monitored at single-molecule resolution by using total internal reflection fluorescence microscopy. Occupied cAMP receptors appear as bright yellow dots on the surface of the cell. The arrow indicates the direction of the Cy3-cAMP source. The time for each sequential image is given in seconds. Kinetic analysis showed that Cy3-cAMP receptor complexes located on anterior pseudopods dissociated faster than those on the posterior tail (239). This work enabled the discrete characterization of receptor dynamics in single living cells, suggesting a role for cell polarity in the chemotactic process. Reprinted from reference 239 with permission from the publisher.

Atomic force microscopy. AFM is a member of the SPM family of tools, the forerunner of which was the scanning tunneling microscope (STM). The central mechanism of an AFM consists of a cantilever, or "arm," to which a very sharp probe, or "tip," is attached. An often-made and apt comparison is to the arm and needle of a (very small) phonograph (138). The cantilever arm may be only 100 µm long, and, ideally, the tip, or "needle," should terminate in a single atom (138). As the tip is scanned across a surface, tip-sample interactions cause deflections of the cantilever, which are detected and amplified by a laser. These interactions may be direct (e.g., physical), or indirect (e.g., electrostatic, electrosteric, and van der Waals' forces) (18, 46, 257). Conversion of cantilever deflection data to topographical information results in both qualitative output (e.g., images) and quantitative output (e.g., measurement of interaction forces and force-distance relationships).

Several modes of imaging are used: contact, noncontact, and tapping (47, 138, 220). Contact imaging involves "dragging" the tip across the sample and may give rise to undesirable effects such as frictional forces and sample damage (138). Noncontact imaging based on electrostatic deflection of the probe tip can be used to investigate charge development or distribution on biological surfaces (220). Tapping-mode imaging was developed as an alternative method for measurements of "soft" biological surfaces likely to sustain damage during contact imaging (47, 138). In this technique, the tip does not scrape the sample but oscillates over its surface, minimizing tip-sample frictional forces (47, 99, 138).

AFM is capable of measuring discrete interaction forces in the piconewton range (149). Because little sample preparation is needed and cells may be observed in liquid environments, AFM can be used for detailed ultrastructural studies of the surfaces of living microbial cells (8, 69). Dynamic events, such as bacterium-mineral adhesion interactions and viral exocytosis, may be measured in real time, under native conditions of hydration and oxygen tension (150, 260).

AFM tips may be chemically functionalized to study properties such as cell surface hydrophobicity. Alternatively, they may be functionalized with biomolecules such as biotin, antibodies, enzymes, or even single, intact microbial cells (69, 85, 149, 179). AFM cantilevers with such functionalized tips can be used as "nanobiosensors" for the study of discrete receptor-ligand interactions or for characterization of cell-substrate interactions (85, 179).

AFM can also be used as a method for the nanomechanical manipulation of individual microbial cells. As such, it can be used to provide quantitative data on cellular physical properties such as rigidity or elasticity (see "Mechanical micromanipulation" below) (18, 257). Measurements of force-distance relationships for AFM tip indentation have also provided a direct means of measuring turgor pressure in individual bacterial cells (18, 257).

Scanning electrochemical microscopy. Electrochemical phenomena such as electron transfer and ion fluxes are associated with both energy production and intracellular signaling processes (258). Well-established techniques for the electrochemical characterization of single, living microbial cells include the use of microelectrodes or patch-clamping approaches (258). SECM is a recently introduced, SPM-based tool for mapping redox activity in living cells (147). In SECM, the scanning tip is an ultramicroelectrode designed for measuring charge transfer reactions (45, 147, 259). Grayscale images, or "redox maps," are generated from variations in tip current as the tip is scanned in the x-y plane above an electrochemically active cell (259). SECM has been used for the electrochemical visualization of oxygen production in single algal protoplasts on exposure to light, for assessment of the permeability of membrane to charged redox species, and for electrochemical studies of Rhodobacter sphaeroides cells (45, 259). SECM imaging is carried out in solutions containing hydrophilic or hydrophobic redox species which function to mediate the transfer of electrons between cellular redox centers and the SECM tip (45). Redox mediators may differ in their abilities to penetrate various cellular permeability barriers (e.g., the outer membrane versus the cytoplasmic membrane). Therefore, carefully chosen mediators may facilitate redox studies of physiologically distinct cellular structures, such as the periplasmic space (45).

In biology, as in astronomy, the spectral characteristics of an object can be used to provide information on its chemical or physical makeup. Spectroscopic methods have been used to monitor the presence and activities of natural microbial populations via remote-sensing techniques (176); for spectral identification of bacterial suspensions in pure culture by Fourier- transform infrared, proton nuclear magnetic resonance, or mass spectroscopic methods (171, 195, 241); for noninvasive investigations of biochemical changes in P. mirabilis populations during differentiation (93); and for the characterization of pigmented colonies formed by various photosynthetic bacteria (251).

However, when they are applied at the population level, spectroscopic measurements suffer from the same drawbacks as other bulk-scale approaches, and contributions from individual microorganisms cannot be assessed (88, 178, 261). Spectroscopic methods capable of single-cell resolution (e.g., microspectroscopic methods) enable the observation of target analytes or properties within specific cells at cellular or subcellular scales (Fig. 5). This can be especially important because information obtained within the context of a whole cell may reveal important clues to the role or function of the analyte within the cell (13, 135).

View larger version (74K): [in this window] [in a new window]   FIG. 5. Raman spectrum of a single Clostridium beijerinckii cell. Spectral peaks ascribed to major cellular macromolecules (e.g., nucleic acids, proteins, lipids, and carbohydrates) are shown. The video inset shows a cell illuminated in the laser focus, which is of approximately the same diameter as the cell. The laser diffraction pattern, which can serve as a visual cue for achieving the proper laser focus, can also be seen. Single-cell Raman spectroscopy represents a noninvasive means of investigating the biochemical heterogeneity of microbial populations. Reprinted from reference 204 with permission from the publisher.

Microspectroscopic methods can provide biochemical information on the overall macromolecular composition of cells (204) or on specific analytes at either whole-cell (88) or subcellular (135, 178) resolutions. Vibrational spectroscopy may also be used to generate images of individual cells by using data from the aliphatic C—H stretching within membrane lipids (267) or from the O—H stretching of intracellular water molecules (183).

Not all spectroscopic methods provide direct information on the chemical composition of a cell. Methods such as electrorotation can be used to determine other characteristics of single cells, including their dielectric properties (106, 131). Some of the more frequently used methods for obtaining spectroscopic data from single microbial cells are described below.

Raman microspectroscopy. The Raman effect is an induced emission of light resulting from the inelastic scattering of a small number of photons from a monochromatic light source (48, 205, 261). Raman spectra provide information on molecular vibrational states, which are dependent on the nature of chemical bonding within a molecule or sample (178, 261). These spectra yield clues to the types and lengths of chemical bonds present and on the molecular conformation or environment (48, 205). Microspectroscopic Raman probes capable of illuminating an area as small as 1 by 1 µm enable the characterization of individual cells and their subcellular components (135). Spectra may be collected at different points along the length of a cell or hypha or at different depths (13, 205) (Fig. 5). Spectra may also be compared among different species, between mutant strains, or at different points in the cell cycle (13, 135).

The range of energies used to generate Raman spectra includes UV (e.g., 257 nm), visible, and infrared excitation frequencies (55, 256). Common visible sources used are argon-ion lasers (514 nm) (135, 178) and helium-neon lasers (632 nm) (13, 178, 204, 205). When the wavelength of the incident light approaches the absorption wavelength of a chromophore within a sample, scattering efficiency is greatly increased, an advantageous effect referred to as "resonance" Raman scattering (256, 261). An example of an application where resonance enhancement would be expected to occur is in the UV-Raman analysis of nucleic acids.

Other applications of Raman microspectroscopy include investigations of microbial carotenoid content and subcellular distribution. Kubo et al. (135) used Raman microspectroscopy to map the carotenoid content in Euglena and in Chlamydomonas reinhardtii. These authors also applied polarization techniques to demonstrate that the carotenoid molecules in the eyespot of C. reinhardtii are oriented parallel to the long axis of the cell. Raman microspectroscopy has also been used to investigate biochemical differences between cells in morphologically heterogeneous cultures of clostridia during solvent fermentations (205). Traditional, bulk-scale methods of analysis of these differentiated cultures do not facilitate connections between the morphological appearance of individual cells and their biochemistry or role in the fermentation. Analyzed by Raman microspectroscopy, morphologically distinct cells yielded spectra that differed in regions ascribed to proteins, lipids, or the storage polymer granulose (204, 205). Analysis of small cell clusters also showed spectral evidence for the presence of polysaccharides, suggesting the presence of aggregation-promoting extracellular polymers. The ability to correlate morphology with biochemical characteristics may provide clues to the activities of the different cell types during solvent production (204, 205). Other applications of Raman microspectroscopy include the reagentless identification of individual bacterial spores (49) and detection of the neurotoxic amino acid domoic acid in single cells of toxigenic phytoplankton (256).

In related technology, coherent anti-Stokes Raman scattering microscopy allows imaging of individual microbial cells on the basis of the vibrational spectra of specific cellular components (e.g., proteins and lipids). The vibrational signatures of these molecules provide a means of generating contrast (267). In this way, the distribution of specific molecular components in living cells can be mapped without the need for fluorescent dyes and at relatively low power levels (100, 267). Finally, time-resolved coherent anti-Stokes Raman scattering imaging can reveal dynamic processes, such as real-time changes in intracellular water concentration (183).

Microbeam analysis. Methods for microbeam analysis represent sensitive means of characterizing single-cell elemental composition (88, 262). These methods allow the measurement of the concentration, chemical state, or cellular location of biologically relevant inorganic nutrients, including phosphate, sulfur, potassium, calcium, iron, and zinc (238, 174, 262). Multiple elements can be measured in a single pass, resulting in an "elemental map" of an individual cell (174, 238) (Fig. 6). Available techniques include X-ray fluorescence imaging and absorption spectroscopy, as well as various ion beam-dependent methods (88, 262).

View larger version (22K): [in this window] [in a new window]   FIG. 6. Synchrotron X-ray fluorescence mapping of the relative elemental distribution in a single diatom. An X-ray microprobe was used to focus a monochromatic X-ray beam on a diatom collected from the Southern Ocean. The sample was scanned through the focused beam in pixel steps of 0.5 µm, and the full X-ray fluorescence spectrum was collected at each step. Two-dimensional elemental maps were generated from the resulting energy spectra by using element-specific filtering. Elemental concentrations were calculated from X-ray fluorescence data using National Institute of Standards and Technology thin-film or similar standards. Comparison of light and epifluorescence (epi) micrographs with elemental maps for the same diatom enabled the discrete localization of each element within the cell. Data collected for this diatom show that Si and K map onto the cell's siliceous frustule; P, S, Ca, Mn, Fe, Cu, and Zn appear to be associated with the cytoplasm; and Ni is present only on the outer membranes or frustule. This approach to characterizing elemental distributions within individual diatoms provides biologically relevant information not available from population-scale methods of elemental analysis. The resulting data may offer unique insights into the physiological state, ambient chemical environment, and role in elemental cycling of these organisms (238). Reprinted from reference 238 with permission from the publisher.

In contrast to the minimal preparative requirements for X-ray microprobe analysis, samples to be studied by ion beam methods may need to be dried and vacuum compatible, constraints that could hinder the analysis of many cell types. A major disadvantage of all microbeam methods described here is that they are very time-consuming. Scanning times ranging between 30 min and 4 h can be required to generate an image (88, 262).

Still, single-cell microbeam analysis may provide useful information on the physiological states of individual microbial cells, as illustrated in the study by Gisselson et al. (88). As these authors noted, sample preparation prior to traditional measurements of algal nutrient ratios may include fractionation steps designed to isolate the subset of the planktonic community to be studied. Despite such careful preparations, contributions to nutrient ratios from bacteria, protists, or particulate organic material may still skew the results (88). Microbeam analysis methods can be used effectively to ensure that measurements are derived from the intended cell type and to probe the nutritional heterogeneity of individual target cells within the population (88, 238).

Electrorotation. When a cell is exposed to an electric field, a dipole is induced, whose character is dependent on the composition of the cell, the frequency of the applied electric field, and the conductivity of the medium in which the cell is suspended (107, 111, 131). In the presence of a rotating electric field, the dipole will form across the cell in synchrony with the rotation rate of the field (110). If the field is rotating with sufficiently high frequency, though, formation of the dipole may become asynchronous with the field's rotation rate. In this case, the cell will experience a torque and begin to rotate, either in the direction of the field ("cofield rotation") or in the opposite direction ("antifield rotation"), depending on the angular difference between the field and the induced dipole (63, 110, 111).

Precise positioning of the electrodes used to generate the rotating field is used to create a dielectrophoretic trap capable of holding individual cells in place during analysis (see "Electrokinetic micromanipulation" below) (Fig. 7A). Electrorotation spectra are displayed as cellular rotation rate versus frequency of the applied field (Fig. 7B). Cell rotation rates can be automatically measured and documented using computer-interfaced video microscopy or interferometric methods (106, 188). Because the dielectric properties of a cell are responsive to mechanical or chemical perturbation, methods for dielectric spectroscopy such as electrorotation can yield information on both the integrity and the physicochemical properties of individual cells (63). Compared to other methods for investigating the electrical properties of cells (e.g., the use of microelectrodes or patch clamping), electrorotation is relatively noninvasive and does not require extensive cell preparations (106, 192, 258). Applications for electrorotation include monitoring the effects of antibiotics on single yeast cells (106) and distinguishing between nonviable and viable protozoan cysts on the basis of the direction of their rotation at specific field frequencies (63). The ability of electrorotation to distinguish between viable and nonviable protozoa is especially useful, since no direct culture-based methods are currently available (63).

View larger version (68K): [in this window] [in a new window]   FIG. 7. Electrorotational analyses of single yeast cells. (A) Single cell of S. pombe during analysis in a microstructured electrorotation chamber. Four circular electrodes (dark semicircles), spaced 100 µm apart, are precisely positioned to allow dielectrophoretic trapping of individual cells. (B) Cellular dielectric properties are responsive to mechanical or chemical perturbation. This panel illustrates time-resolved changes in the electrorotation spectra of a single S. cerevisiae cell treated with nystatin at t = 12 min. Nystatin-mediated leakage of intracellular ions is expected to change the dielectric properties of the cell, leading to the frequency-dependent shifts in cell rotation rates seen for both cofield and antifield rotations. Panels A and B reprinted from references 131 and 106, respectively, with permission from the publishers.

Micromanipulation can be used to address a cell to a specific position in a liquid medium and hold it there in order to examine its ability to replicate (77). Physical segregation of daughter cells may also be used to trace the pedigree of a single cell as it undergoes multiple cycles of division or to examine adaptation processes of individual cells subjected to changes in nutrient availability (240, 245, 263).

Alternatively, a cell may be positioned in close proximity to or touched against other cells, immobilized enzyme substrates, or inorganic surfaces. In this way, discrete binding, chemical, or other interaction forces may be measured (31, 46, 149, 150, 159, 179, 216). Micromanipulative techniques also allow the stable positioning of cells for observation during single-cell assays for pharmacological or biochemical activity (189, 219).

Aside from methods of physical separation or positioning, micromanipulation may permit the direct measurement of the physical or structural characteristics of an individual cell (Fig. 8). AFM and related force transduction technologies can be used to measure turgor pressure, elasticity, bursting force, and other micro- or nanomechanical cellular properties (18, 213, 218, 257).

View larger version (9K): [in this window] [in a new window]   FIG. 8. Comparison of whole-cell bursting responses from Staphylococcus epidermidis (left) and E. coli (right), as determined by micromanipulation. Force diagrams for single cells compressed between the surface of a glass slide and an optical fiber are shown. In both diagrams, datum point A indicates the first contact of the microprobe with the cell and datum point B corresponds to the point at which cell rupture takes place. The microprobe continues to advance after cell bursting, eventually compressing the cellular debris (points C and D). To correlate bursting properties with cell size, a video image was taken of each cell prior to manipulation. Data kindly provided by C. Shiu, Z. Zhang, and C. R. Thomas.

Mechanical micromanipulation. The use of mechanical means of manipulation of single microbial cells is not a new concept. In 1951 Zelle (263) used a microscope-mounted mechanical micromanipulator to directly monitor the pedigrees of individual E. coli cells positioned on the surface of an agar-covered slide. However, the advent of computer-assisted stage or micromanipulator movement and more sensitive methods for fluid aspiration and deposition has led to improvements in the basic technology. Together, these improvements have resulted in more accurate, more accessible, and less exacting processes for the mechanical manipulation of microbial cells (81, 82, 83).

Current technologies enable the direct isolation of individual cells of interest from within complex natural populations. For example, Frölich and König (81) used a sterile capillary tube method to isolate individual cells of Enterococcus and