Preventing Evolutionary Failure in Synthetic Biology
Evolutionary half-lives of biological devices
Synthetic biology applies engineering principles to create living systems with predictable and useful behaviors from collections of standardized genetic parts. However, living systems – unlike mechanical devices – inevitably evolve when their DNA sequences accumulate copying errors, often resulting in "broken" cells that no longer function as they were programmed. We are addressing this challenge by better characterizing how engineered cells evolve and using this information to design DNA sequences and host cells that are more robust against unwanted evolution. This work includes: (1) the development of the Evolutionary Failure Mode (EFM) Calculator software for identifying mutational hotspots in a designed DNA sequence; (2) using experimental evolution to identify "antimutator" variants of host organisms that lead to lower-than-natural mutation rates; and (3) designing genetic circuits that kill those cells within a population that are most likely to accumulate mutations.
Dynamics of Microbial Genome Evolution
Accumulation of mutations in population Ara-1 of the LTEE over 20,000 generations of evolution
We develop the breseq
computational pipeline for identifying mutations in laboratory-evolved microbial genomes from next-generation sequencing data. We have used this tool to extensively study rates of genome evolution in the 30-year Lenski long-term evolution experiment (LTEE)
with E. coli
. We continue to develop breseq
so that it can be used for more additional applications related to strain engineering and medicine. For example, we are interested in how tracking rare variants within populations of microorganisms (such as oncoviruses) can anticipate further evolutionary trajectories and how this information might be used to better diagnose disease outcomes.
NIH K99/R00, NSF, NSF BEACON Center, CPRIT
Identifying Mutations that Promote Bacterial Evolvability
Epistasis Poker. Early draws of cards that are more beneficial to the value of a poker hand (analogous to the most beneficial mutations in the eventual loser E. coli), prevent them from reaching the best hand after more cards are drawn.
Evolutionary potential (or evolvability
) is the capacity of an organism to generate descendants with increased fitness or new adaptive characteristics. Evolvability is a complex trait. It depends not only on how developmental and regulatory processes render underlying genetic changes into phenotypic variation, but also on the dynamics of how mutations arise and compete within a population. Mutator strains of microorganisms, which have elevated genomic mutation rates due to defects in DNA proofreading or repair pathways, can sometimes take over pathogen populations because they promote evolvability. However, little is currently known about how mutations affecting other cellular processes impact microbial evolvability or about how evolvability varies at each step in a typical adaptive trajectory.
We are reconstructing the dynamics of mutations in populations from the Lenksi 30-year long-term evolution experiment (LTEE)
with E. coli
to understand two cases of evolvability differences not related to mutation rates:
Eventual Winners and Losers.
Evolution of central metabolism potentiates and refines a metabolic innovation
Bacteria that had evolved higher fitness early in one LTEE population reproducibly lost later to competitors with different, less-beneficial mutations when evolution was replayed many times from each starting point. Thus, the eventual losers apparently had favorable mutations that restricted further evolvability. We have found that loss of function mutations in the gene (fabR
) seem to be the major defining characteristic of the eventual losers and are investigating how interactions between this mutation and further adaptive steps differ from the alternative evolutionary pathways that won out in the LTEE.
Evolutionary pathway to Cit+
A metabolic innovation enabling aerobic citrate utilization only occurred in one of twelve populations and only after 30,000 generations of evolution. This new Cit+ trait was highly beneficial because it gave access to an untapped nutrient that was present for the entire duration of the LTEE at a much higher level than the glucose that these cells normally use as a sole carbon source. The lineage that evolved Cit+ appears to have evolved a potentiated genetic background that allowed it to access this innovation during the daily growth cycles of the LTEE. We have recently begun to characterize how these potentiating mutations remodeled central metabolism in ways that made achieving the very weak initial Cit+ phenotype beneficial enough to persist in the evolving population until it could be refined to enable full citrate utilization.
The long-term goal of this research is to further understand the mechanisms by which mutations can promote and reduce evolvability in order to be able to rationally make these changes to microbial genomes.
Evolving and Engineering Naturally Transformable Bacteria
Naturally competent bacteria may have increased evolutionary potential because they can directly acquire new genes from their environments and incorporate them into their genomes. In addition to this possibility of a new mutational move in genotype-phenotype space (by horizontal gene transfer), these microbes provide an improved platform for studying microbial genome engineering and evolution due to the ease of reconstructing mutations and introducing new genes. We are investigating sources of genetic instability in the naturally competent bacterium Acinetobacter baylyi
ADP1 and engineering a clean genome version of this strain to have reduced rates of mutations that lead to inactivation of introduced genes for synthetic biology applications. We are also using experimental evolution to test for the utility of providing foreign DNA sequences and to understand how horizontally acquired genes become domesticated after they are incorporated into a new genome. Finally, we are using ADP1 as a platform for understanding the limits of simplifying a bacterial genome by evolutionary streamlining approaches.
Experimental Evolution with Expanded Genetic Codes
Another way to potentially increase evolvability is to augment the chemical diversity of the building blocks available to an organism. We are evolving microorganisms, including bacteriophage, in the context of expanded genetic codes that re-code the amber stop codon into various non-canonical amino acids. This change increases the sequence-structure-function space available for evolution to explore. We are studying how evolution must compensate for the global change in how the information in an organism's genome is decoded, in cases where it is possible for an organism to survive this transition, and also how dependence on the new amino acid develops over time, such that an organism can no longer survive in the context of a normal genetic code. Finally, we are systematically examining how the unique chemistries of different amino acid building blocks are tolerated and uniquely useful for adaptation and using this technology to recapitulate the steps in a hypothesized pathway in which new post-translational modifications evolve in a series of adaptive steps from nonenzymatic metabolite damage to the proteome.
Welch Foundation, NSF BEACON Center
AG3C: Associating Growth Conditions with Cellular Composition
We are part of a collaboration of eight labs (including experimental biologists, computational biologists, and mathematicians) that is collecting and analyzing comprehensive transcriptomics, proteomics, lipidomics, and metabolic flux data from bacterial samples grown under different conditions. The aim is to solve the "inverse problem" of predicting the conditions that were used to culture a microbial sample from measurements in these high-dimensional data sets (AG3C Website
). Our lab is primarily responsible for creating the microbial samples and using RNA-seq to examine RNA expression levels.
DoD Army Research Office
Discovering Functional Nucleic Acid Families by Deep Sequencing and Fold Sampling
A final way that we are studying how to improve the potential of evolutionary methods is in systems for the in vitro selection of nucleic acid molecules. Nucleic acids with random sequences often misfold into inactive structures that are kinetically trapped. That is, they become stuck and cannot unfold to rearrange into other, potentially active, structures on a reasonable timescale. It is possible that this potential for misfolding, which is known to occur for even natural ribozymes, limits the recovery of new families of functional nucleic acids (aptamers, ribozymes, deoxyribozymes, riboswitches) by laboratory evolution methods that select for functional molecules from pools of 1013
molecules with randomized sequences. We expect this problem to more greatly affect longer nucleic acid sequences, which have the capacity for folding into elaborate three-dimensional structures that achieve more rapid rates of catalysis or tighter binding affinities. We are currently using next-generation deep-sequencing techniques and methods to allow a single nucleic acid strand to have multiple chances to fold correctly to pass selection to see if they can enhance our ability to recover more complex families of functional nucleic acids.
Evolution Experiments with Digital Organisms
Populations of self-replicating computer programs mutate and compete for CPU cycles in the Avida artificial life system. This environment for digital organisms has been used to ask fundamental questions that transcend the substrate that is evolving. The speed of digital evolution experiments, transparency of the underlying mechanics, and ability to manipulate and record every aspect of the evolutionary process make it possible to more readily explore cause and effect relationships in Avida than in natural systems. I have extensive experience with Avida. My lab will use Avida to complement studies in biological systems: both as a way of more clearly illustrating and exhaustively characterizing phenomena of interest and as a source of new candidate laws and insights to motivate new lines of experimental inquiry.
In addition to these specific plans for future research, we have long-standing interests in RNA biochemistry, riboswitches and other noncoding RNAs in bacteria, and comparative genomics. We hope to be a resource to colleagues who are interested in adding an evolutionary, informatics, or genomics dimension to their studies. Whenever possible, we would like to investigate our basic research questions about evolution in systems where there is additional interest in the underlying molecules, cells, and mutations. We are eager to link up with researchers and those in industry who have relevant problems in medicine and biotechnology to find these opportunities. Please contact Prof. Barrick
about potential collaborations.