Microbes — viruses, bacteria, archaea, and protists — account for half of all the biomass and the majority of organismal diversity on planet earth. Microbes gave rise to higher organisms and have left their genomic calling cards in the form of organelles, genes, and so called junk DNA. In addition, microbes are the source of the majority of human diseases. For these reasons alone microbes are worthy of scientific study. Yet, they are also important in another, not so obvious, way: microbes are an extraordinarily powerful model system for understanding, in very fine detail, how evolution and biological systems work. To this end, experimental approaches offer one the most powerful approaches to understanding (1) microbial evolution, (2) the biology of systems, and (3) evolutionary principles in general. This power arises from a number of important features of microbes. First, they can be easily propagated in the laboratory. Second, they have short generation times allowing for many generations of evolution to be documented and analyzed. Third, due to their relatively small genomes we can track all of the genetic changes occurring during evolution and then easily determine their selective and/or functional effects. Fourth, we can easily manipulate important evolutionary parameters such as population size, recombination, and mutation rates to understand the role these forces play in evolution. Lastly, we can engineer genomes to understand the landscape on which evolution may be acting to better understand the systems themselves.
Selective Barriers to Horizontal Gene Transfer
Horizontal gene exchange represents one of the important sources of novel gene function in bacteria and Archaea. It has been estimated that nearly all bacterial genes have experienced at least one successful horizontal gene transfer during its evolutionary history. This observation coupled with the many single gene studies of horizontal gene transfer have led others to adopt a largely genomic approach to better understanding the barriers to horizontal exchange. This rich body of work has identified such forces as divergence, dosage, and the location of the new gene within the new genomic backgrounds gene network as important factors. Yet we still lack a fundamental global understanding of these factors and the selective affects of a newly acquired gene during the first phase of gene acquisition. To address this question we have adopted an experimental evolutionary approach. Using a clone library from Salmonella typhimurium LT2 covering 98% of the genome we will introduce approximately 300 genes into a novel E. coli genomic background that can express the new genes at varying levels. Using competition assays between the HGT lines and the ancestral genome (i.e., those lacking the new gene) we will be able to estimate the distribution of mutational effects (DME) at different levels of gene expression and in different environments. With this information, coupled with information about the functional role of the introduced genes, we will be able to address questions about the relative magnitude of different barriers to horizontal gene acquisition (e.g., importance of dosage affects, location in the gene network, environmental dependency, etc.).
Epistatic interactions within bacterial operons
The structure of the bacterial genome is by and large determined by its regulatory and transcriptional structure. These structures have been termed operons, regulons, modulons, and stimulons in increasing order of complexity. The simplest structural unit, operons, are transcriptional units made up of at least two genes physically located in a cluster. The lac operon, for example, consists of three metabolic genes plus one regulator gene. Genes within an operon may have co-evolved to function together. These interactions could take the form of protein-protein interactions, regulatory interactions, or optimization of a stoichiometric balance of the gene products of the operon. Regardless of the form of the interaction, these would give rise to epistatic interactions that prevent genes from within the operon from being transfered individually. and favor transfer of the whole operon. In particular, coadaptation may lead to reciprocal sign epistasis, in which the alleles from each species fail to function well in the foreign background, pro- viding an especially strong evolutionary barrier to transfer of individual components of an operon.
We do not, however, know to what extent epistatic interactions exist between genes within an operon, or what form, if any, this epistasis takes. A number of studies of the lac operon have provided anecdotal information that components within an operon (genes and regulatory elements) do indeed show co-adaptation, that there are epistatic interactions among nucleotide states of the lac promoter, and that epistasis seems to be common among nucleotides within genes. These experimental observations have been made across a wide diversity of organisms, suggesting it is a prevalent force. While epistasis may be common, there appears to be no general pattern to the form of epistasis. Understanding what the epistatic landscape looks like (i.e., how rugged it is) is important because it determines the possible evolutionary paths. For example, if the epistatic landscape is rugged then the horizontal transfer of one component of an operon may be selected against, so that the only evolutionarily viable HGT must include the whole operon. If, on the other hand, the landscape is smooth, this suggests that single genes in an operon can move promiscuously between species, and may suggest that the rate of co-evolution is slower than the rate of within operon HGT.
Here, we are testing whether there are epistatic interactions among genes within the lac operon by creating recombinant operons between divergent strains of E. coli. Because of the combinatoric nature of this experiment, we have restricted this investigation to a single operon. By using the lac operon, we will be able to elucidate the full gene-by-gene fitness landscape, something that is practically impossible to do at the nucleotide level for combinatoric reasons. The approach adopted here is to engineer lac operons that are mixtures of genes derived from E. coli K12 MG1655 and two different divergent strains and then determine the fitness landscape of these engineered operons on the K-12 background. Using these data we will be able to directly address whether epistatic interactions exist between genes of an operon and whether this increases with genetic divergence. The latter question is interesting as if there is a relationship between the size of the epistatic interactions and divergence, it suggest that within bacteria epistasis builds up with time such as predicted by Bateson, Dobzhansky, and Muller.
The Cas/CRISPR system as a model for understanding herd immunity
Herd immunity, the protection of individuals sensitive to a parasite by those resistant to that parasite, is important in both understanding the evolutionary maintenance of co-evolving host-parasite populations and in determining the needed minimum level of vaccination (i.e., what should the target level of vaccination a government wishes to achieve to protect a not fully vaccinated population).
Using a simple bacterial model system we are attempting to understand the role of population structure, levels of host resistance, and the rates of dispersion of the parasite that will halt or limit the expansion/spread of an epidemic. We have engineered E. coli that show near complete resistant (approx. 99%) to T7 bacteriophage using a synthetic Cas/CRISPR plasmid based system. Using this system we have just begun to study the velocities and the time to cessation of an expanding epidemic in 2D space. We have observed that the spread of an epidemic in 2D space (i.e., on a plate) can be quickly halted quickly when the population is nearly, but not completely, resistant population (80% resistant) relative to a fully sensitive population (0% resistant) which continues to expand until the limit of available resources and the build up of waste products.
We are currently testing a number of different regimes – porosity of the medium which effects the rates of parasite diffusion, different ratios of sensitive to resistants in the population, and the role of host environmental quality (e.g., poor or luxuriant) on the dynamics of the epidemic spread. Using this experimental date we hope to be able to model epidemics that account for all of these factors and the changing local demography (i.e., a depletion of sensitives on the leading edge of the epidemic, which alters the sensitive to parasite ratio).
Epistatic interactions in cis and trans acting elements of regulatory networks
While it is clear that changes in the regulation of gene expression could be a source of evolutionary innovation, the fate of mutations in transcription factor binding sites is less well understood. In particular, little is known of the nature of epistatic interactions between these mutations and the factors that might affect them. We created a synthetic system based on the regulatory elements of the ara operon of Escherichia coli involved in arabinose metabolism, that allowed us to test the effects of single and double mutants in two transcription factor binding sites involved in the regulation of the araBAD promoter. This construct decouples the control of transcription from any post-translational effects on fitness, thus allowing a precise estimate of the effects of epistatic interactions on expression. We found that epistatic interactions between mutations in the ara regulatory element are common, and that the predominant form of epistasis is negative (i.e., synergistic). Environment, defined by the presence or absence of arabinose, had an effect on epistasis, with some mutations even changing the sign of the interaction between the two environments. We also identified that the interactions between mutations depended on whether the mutations were found in the same or in different transcription factor binding sites. Put together, our study offers insight into the mechanisms that impact epistatic interactions, and highlight how such interactions might affects evolution of gene regulation.
We are currently extending these approaches, in a much more ambitious project, to explore the interactions between cis and trans elements in the lambda bacteriophage. In this project we have created and measured the effects of hundreds of double mutations on gene expression in the cis element (PRM ) and have constructed hundreds of mutations in the trans element (transcription factor, cI). Using these approaches in a scaled up system (thousands of mutations rather than a handful), we will be able to test the generality of our results from the cis element study of the arabinose operon, look for difference in patterns of epistasis within cis and trans elements, and explore the epistatic interactions between cis and trans elements.