It is commonly said that no two snowflakes are alike. This may in fact be the case with living cells. As we now enter an era of single cell microbiology, the sources of variation between individuals and the nature of individuality (both genetic and otherwise) become more clear. Technological developments will drive discovery in this area during the next few years. And the implications of the resultant findings for our understanding of microbial diversity, interactions of microbial communities with environment, and the function of biological systems will be profound.
The origins of single cell microbiology have been associated with Leeuwenhoek, who first saw individual bacteria in 1683. Following his basic observations of cell shape and motility under the microscope, numerous reagents and devices have been developed that reveal structural, chemical, metabolic and phylogenetic details of single microbes. However, these techniques are generally applied to samples containing many millions of cells or more, only a tiny fraction of which are ever examined, and eve then with relatively poor resolution. Other venerable techniques of microbiology, such as obtaining pure cultures via streaking or dilution, are rooted in the concept of isolating single microbial cells. For many decades, though, any investigation of an isolated cell has required many rounds of replication in culture to bring the mass of microbes above the detection limit of available tools. One is left with homogenized, population-wide measurements.
Recent progress in our ability to handle tiny volumes of liquid, along with advances in detection and measurement technology, herald the prospect of microbial experiments at the scale of the microbes themselves. For example, tools and procedures have been developed that will soon allow us to obtain the complete genome sequence of a single cell belonging to an uncultivated microbial species, directly from the environment (Ottesen et al., 2006; Zhang et al., 2006). Functional chemostats have been demonstrated with working volumes measured in nanolitres or even picolitres, containing at most hundreds or thousands of cells. An interconnected landscape has been etched on a silicon wafer at a scale such that the migration of a single bacterium can exert a significant influence on the population that develops in an individual habitat patch (Keymer et al., 2006).
As with other technological advances dating back to Leeuwenhoek's exquisite single-lens microscopes, we predict that microfluidics, nanofabrication and highly sensitive analytical techniques will enable the discovery and investigation of phenomena that deepen our understanding of microbiology, and indeed of all life. For example, tracking the replication of individual Escherichia coli cells and their descendants across multiple generations has recently revealed senescence in organisms that reproduce by symmetric binary fission (Stewart et al., 2005); this last refuge of biological immortality has proven to be an illusion. On the other hand, the biochemistry of ageing via oxidative damage and life history tradeoffs between longevity and reproduction may be shared between us and the humblest of bacteria. Such a discovery would have been impossible without the ability to monitor individual cells over time.
Perhaps the most straightforward application of single-cell techniques may be the characterization of seemingly homogenous bacterial populations, including those comprised of clonal descendants from single cells. Given the large number of cells in traditional microbiological experiments and our current estimates of genome-wide mutation rates, most microbial measurements have almost certainly involved heterogeneous populations. In the past, we assumed (hoped) that novel mutants remained rare and did not disturb population-wide measurements too much. The rapidity with which a Growth Advantage in Stationary Phase mutant can take over a stationary phase culture, and reports of high frequency genetic rearrangements, suggests that, at least in some cases, these assumptions may not have been well justified. As we begin to work with many fewer total number of cells in an experiment, the expected time before mutation or rearrangement in this population will increase proportionately.
Among other important potential applications, these techniques will facilitate experiments that examine the importance of stochastic fluctuations (‘noise’), and cell-to-cell variation in features such as gene expression capacity, pathway capacity, and the partitioning of cellular components to daughter cells during replication (for example in yeast cells, see Colman-Lerner et al., 2005). Both modelling and empirical data indicate that the behaviour of individual cells can deviate considerably from the average behaviour of a large number of cells. Variation in gene expression can lead to dramatic differences in the fate of genetically identical bacterial cells, such as the ‘suicide bomber’ phenotype displayed by a small proportion of cells belonging to a colicin-producing strain, or the rare ‘persister’ cells that ‘voluntarily’ shut down their growth activities but concomitantly gain phenotypic resistance to antibiotics and other stresses. These phenomena were discovered because natural selection exploited the wide variance in the output of certain gene regulatory circuits to confer an obvious fitness benefit. Many less obvious microbial traits may also turn out to depend on the variance in expression or pathway capacity among cells, rather than on the mean.
The ability to isolate and cultivate small numbers of cells will greatly enhance studies of evolutionary adaptation. Given a sufficient number of cells, certain mutations arise predictably in the laboratory in response to a particular selective regime. One example is the set of ‘wrinkly spreader’ mutations that arise in a single operon in Pseudomonas fluorescens, allowing this organism to colonize the air–medium interface in static broth cultures. Does such predictability in this organism indicate the availability of just one adaptive pathway, or does it indicate the disproportionately large immediate benefit conferred by such mutations – so that they sweep to fixation and out-compete other mutations with a smaller benefit that would otherwise have set the organism on a different adaptive trajectory? Experiments using small numbers of cells might reveal details of the adaptive landscape that cannot otherwise be discerned. Such an approach might help answer whether multiple distinct pathways eventually converge towards the same optimal phenotype, or whether certain early mutations predetermine subsequent paths towards different fitness optima.
Experiments using microdevices with single cells teach us microbial ecology at scales relevant in the natural world. The colonization of a copepod faecal pellet by marine microbes or the early stages of gut colonization in a newborn may involve only a few microbes. Does it matter which species or strains are present? Or does the physical and chemical environment exert such an influence during later time points that the initial events of colonization have no particular bearing on the community that ultimately develops? The evolution of antagonistic or cooperative relationships often depends on the extent to which organisms and their descendants continue to share the same environment. The answers to these questions no doubt differ depending on the situation. However, the patchiness of environmental conditions and the heterogeneity of microbial distributions that have been found in many environments when investigated at the scale of the microbes themselves suggest that our ability to conduct experiments at these scales will help reveal the forces that shape the evolution and ecology of microbes in the natural world.