A population of wild-type Escherichia coli
will cease to proliferate in response to deteriorating growth conditions. The response to such conditions is regulated by the rpoS
), which triggers the expression of several cell protection mechanisms as a response to decreasing nutrient levels, increased density, or changes in pH (12
). In response to limited nutrients, cells shift from growth to maintenance of basic cellular functions, a state called stationary phase (15
Under prolonged starvation, however, cells may evolve the ability to grow despite stressful environmental conditions (8
). First observed by Zambrano et al. (24
), these resistant populations are called GASP mutants, because they have a growth advantage in stationary phase. E. coli
carrying mutations affecting the rpoS
gene, one of which is the rpoS819
allele, will not stop growing in response to the stresses found in stationary-phase cultures. Experiments in well-mixed, homogenized environments, such as test tubes, have shown that GASP cells containing the rpoS819
allele have, under prolonged starvation conditions, a fitness advantage over wild-type (WT) strains (23
Natural environments, however, are rarely homogeneous and well mixed. Feast or famine, where long periods of starvation are mediated by short bursts of population growth, is a more accurate description of the conditions encountered by bacteria in the wild. Furthermore, natural environments are spatially heterogeneous, and different species are usually competing for space and nutrient resources. In this paper, we use a microfluidic device to recreate the feast-or-famine and heterogeneous aspects of natural environments to study the competition between wild-type and GASP mutant strains of bacteria.
The microfluidic device used, shown in A, is fundamentally different from other cell culture systems, such as chemostats and test tubes. Each device consists of 85 microhabitat patches (MHPs) along a line, each of which is 100 by 100 by 8 μm in size, linked by 5-μm-wide and 50-μm-long inter-MHP channels. Cells retain their ability to swim and can migrate between MHPs. Each MHP can harbor approximately 15,000 cells, assuming a cell size of 0.6 μm3
) and a very loose packing fraction of slightly more than 10%. Comparing this to E. coli
cells at an optical density at 600 nm (OD600
) of 1, where 1 ml contains approximately 109
cells, we would get only 80 cells per MHP. In fact, cells inside our MHP system reach densities close to those of the intestinal microbiota (1011
Fig. 1. (A) Two strains of Escherichia coli compete inside a microhabitat patch (MHP) system. The MHPs are 100 by 100 by 8 μm in size, and every other chamber has access to nutrient. A linear series of 85 such chambers is microfabricated in silicon. The (more ...)
Nutrients can diffuse into the MHPs from reservoirs of Luria-Bertani (LB) broth on each side of the MHP array through 250-nm-deep nanoslits, shown in yellow in A. Nutrient and waste, but not the cells, can diffuse in and out of each MHP through the nanoslits; unlike chemostats, no cells are removed from the system. We impose a high level of spatial heterogeneity on the bacterial populations by restricting the amount of nutrient reaching each MHP. The percentages of nanoslits vary from 100% open in one MHP to 0% open in the next. We refer to the former as nutrient rich (NR) and the latter as nutrient poor (NP). Since each NP MHP is surrounded by two NR MHPs, waste and conditioned medium can diffuse from the NR into the NP MHPs through the inter-MHP channels. We thus expect the nutrient composition of the NP regions to be closer to that of conditioned medium than rich medium.
Our previous experiments showed that mixing populations of wild-type and GASP cells allows a larger population of both types of bacteria to flourish relative to the case when only one type is present (14
). In experiments described here, we observe that wild-type cells initially perform better in NR MHPs but at later times redistribute themselves into the NP MHPs. Conversely, GASP cells perform better than wild-type cells in nutrient-rich MHPs. We explain this anomalous population redistribution using ideal free distributions. We also describe the collective cell behavior inside each MHP and show that multicellular interactions influence the dynamics of the population distribution.