Antibiotic-resistant bacteria present a serious rising concern and demands for a new effective treatment are greater than ever. To fight such bacteria, one promising candidate is the predatory bacterium, Bdellovibrio bacteriovorus (BV), that has recently shown promise as a safe antibacterial therapy in vivo. The Gram-negative bacterium, BV, is a model bacterial predator found across diverse habitats that preys upon other Gram-negative bacteria. It is typically less than a micron in size and swims over 60μm/s in solutions by rotating its single polar flagellum. BV has a biphasic life cycle including an "attack" phase and a "free-living" phase. During the attack phase, starting with physical contact of the predator with its prey, BV gets into and grows within its prey's periplasmic space, replicates and releases its motile progeny in search of new bacterial prey. It is also poised to help purify water and soil, degrade hazardous biofilms and serve as a living antibiotic. However, after over 50 years since its discovery and, after extensive biochemical and genetic studies into BV's hunting strategy, it is still unclear whether BV chemically detects its prey or collides with it at random. In this thesis, we wish to address a decades-old question and answer how the model bacterial predator, BV, locates its prey by combining methods of statistical physics with microscopic imaging and identify a novel, hydrodynamic, candidate mechanism that BV may utilize to locate its prey. In chapter 1, we provide an overview on bacterial chemotaxis. We discuss what challenges a bacterial predator would face in searching for its prey using chemical clues. We then discuss the necessity for a new chemotaxis model that goes beyond our current model. In addition, bacterial swimming strategies are discussed and finally an interplay of hydrodynamics with these motilities is explained. In chapter 2, we first present a general chemotaxis model, applicable to a regime rarely studied in experiments, to describe how bacteria locate point sources of food on the basis of stochastic event detection, rather than chemoattractant (CA) gradient information. We consider this model since previous studies have shown that bacteria secrete a set of free amino acids that may be detectable by other bacteria and BV is shown to be chemotactically attracted toward many of these amino acids. Experiments have also shown high sensitivity of bacteria to small CA gradients. We show how all parameters of a model for chemotactic attraction we propose can be directly inferred using maximum likelihood methods from microscopy single cell tracking data even in the regime of high detection noise. In chapter 3, we consider signatures of a targeted search assuming that the searcher employs chemical sensing to locate its food source. One such prediction is that a predator would take longer to find its first prey in the presence of multiple prey. As another example, from our model also emerges the "volcano effect" (previously proposed for other bacteria), the spherical shell shape density of bacteria forming around rather than on a CA point source due to a delay in bacterial response to the detected CA. We quantify the properties of our observations (such as radius and width of the volcano shell) based on our model parameters. This provides an alternative way to infer model parameters using microscopic bulk observations and model predictions provide clear statistical signatures of a targeted search. In chapter 4, we present a single chemotaxis model capturing both run-and-tumble and run-reverse bacterial dynamics. The former is more common among multi-flagellated bacteria while the later appears more common among uni-flagellated bacteria. We use this model to study bacterial behavior around point sources of CA with increasing emission rates. Specifically, we model bacterial patterning such as "banding" or "volcano effect" around CA point sources and possible dynamical transitions between these behaviors. In chapter 5, we study the effect of hydrodynamic interactions on BV's motility and on its hunting efficiency. We first demonstrate that BV's dynamics is strongly affected by its own self-generated hydrodynamic flow fields as a result of its own swimming. Our in vitro experimental results show that these hydrodynamic fields passively bring BV onto surfaces and force them into circular trajectories and, if BV approaches defects on surfaces such as spherical beads, it is geometrically captured in orbital motion around these defects (with enhanced densities around them). We will present crucial control experiments, supporting simulations and comprehensive hydrodynamic calculations to show that observed behaviors are hydrodynamic in origin. We show that while direct hydrodynamic interaction between BV and its common bacterial prey (Escherichia coli) is too small to trap BV around the prey, the prey itself also exhibits similar behaviors as the predator on surfaces and around defects due to similar hydrodynamic interactions. This, ultimately, co-localizes predator and prey and this, in turn, dramatically improves BV's odds of colliding into its prey. In conclusion, our theoretical and experimental results presented in this thesis show that chemical sensing by BV for its prey in three dimensions is a difficult search and the key problem is that the predator is tracking a prey which is releasing only very few CAs. What is more, the prey is also evolving in space. Therefore, for the predator, it is highly unlikely to be capable to both sensitively detect its prey and on the basis of its currently detected signal, to forecast its mobile prey's future position. Our theoretical and experimental results instead show how BV takes advantage of hydrodynamics to ease its search problem thereby going beyond the chemical sensing paradigm for bacteria.
Pressé, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our