This project is an exciting new approach to molecular evolution that puts the concept of an adaptive landscape in the experimental context of protein evolution. The research employs novel approaches to study evolvability in sequence space. The focus of the research is on the enzyme dihydrofolate reductase (DHFR), an essential enzyme that is a critical target of antifolate drugs used for treatment of malaria, bacterial infections, and cancer. Previously we have shown that the adaptive landscape of pyrimethamine resistance in DHFR is unexpectedly smooth with limited instances of sign epistasis and virtually no reciprocal sign epistasis. Experiments are ongoing to determine whether this smoothness is a general property of DHFR in sequence space or an artifact of studying amino acid replacements already known to be involved in the stepwise evolution of drug resistance in the field. To test the predictions of adaptive landscape theory, we use the newly developed "morbidistat," a continuous-culture device that dynamically adjusts drug concentrations to constantly inhibit bacterial populations as they evolve to become more drug resistant. This device allows for the automated evolution of multistep antibiotic resistance and is ideal for testing whether alternative pathways are quantitatively in agreement with landscape predictions. There is a vast literature on adaptive landscapes, however there are almost no experimental studies that ask whether the concept has any reliable predictive value.