NEW YORK (GenomeWeb) – Researchers from Duke University and the University of Connecticut have published a paper in the Proceedings of the National Academy of Sciences that describes their use of a protein design algorithm to identify single nucleotide mutations in methicillin-resistant Staphylococcus aureus (MRSA) that would eventually render the bacteria resistant to an experimental inhibitor currently in the preclinical stage of development.
The study used a combination of "bacterial fitness and enzymatic kinetic experiments along with the determination of a high-resolution crystal structure of the mutant enzyme [to] clarify the structural and biochemical bases of the resistance, including an explanation for the compensatory relationship of the two mutations," the researchers wrote in PNAS.
"This gives us a window into the future to see what bacteria will do to evade drugs that we design before a drug is deployed," Bruce Donald, a professor of computer science and biochemistry at Duke and one of the authors of the PNAS study, said in statement.
It's an alternative to standard retrospective approaches, which try to predict resistance mutations by analyzing lists of known pathogenic mutations — an approach that falls short when it comes to anticipating how bacteria will adapt to new drugs, since microbes can't be counted on to change in repeatable, predictable ways, Donald said. "We wanted to find out what countermoves the bacteria are likely to employ against these novel compounds. Will they be the same old mutations we’ve seen before, or might the bacteria do new things instead?"
Specifically, the researchers used the K* algorithm, one of the tools in the Open Source Protein Redesign for You (OSPREY) software suite — developed in the lab of Duke's Donald — to predict SNPs in the active site of the dihydrofolate reductase enzyme in MRSA that would confer resistance to an experimental propargyl-linked antifolate being developed in the laboratory of Amy Anderson, a UConn professor of medicinal chemistry and co-author on the PNAS paper. The drug candidate has shown promise as a treatment for MRSA strains that are resistant to the sulfamethoxazole and trimethoprim antibiotic combination.
The PNAS study represents the first time protein design algorithms have been used to prospectively predict viable resistance mutations in bacteria in response to antibiotic pressure, according to the authors, and the findings here speak well of the efficacy of their approach. The algorithm used in the study is able to efficiently perform large combinatorial searches over a large number of possible amino acid substitutions and possible combinations of substitutions whilst accounting for factors such as conformational energy en route to identifying the optimal structure for a mutated enzyme, he said.
In an earlier study, the researchers used the same computational method to predict double mutations of DHFR in MRSA that would evolve in response to propargyl-linked antifolate inhibitors — that study was published in PNAS in 2010. The researchers used the K* algorithm to help predict the rise of four mutant enzymes. They followed up with "enzyme inhibition assays, [which showed] that three of the four highly-ranked predicted mutants are active yet display lower affinity ... for the inhibitor," according to the paper. Furthermore, "a crystal structure of the top-ranked mutant enzyme validates the predicted conformations of the mutated residues and the structural basis of the loss of potency," they wrote. However, they were not able to confirm that the bacteria would actually choose those mutations in practice, Donald told GenomeWeb.
So for the current study, the researchers restricted themselves to single mutations, hypothesizing that these would be more likely to be selected by the bacteria, according to Donald. This time, they identified four mutated versions of DHFR that were able to resist the experimental inhibitor while still largely maintaining the enzyme's regular catalytic activity. They were able to confirm the rise of two of these computationally predicted mutations in biochemical tests as well as in in vitro tests that subjected select MRSA strains to the experimental drug.
Their tests also revealed a compensatory relationship between a known MRSA resistance mutation and the top-ranked predicted mutation that resulted in a "doubly mutated enzyme with fitness comparable to the [wild type] enzyme," according to the paper. Not only did this dual mutation knock down the inhibitor, it also actually improved catalysis slightly, Donald told GenomeWeb. "It’s a promising direction and one we plan to pursue," he said.
Moving forward, one potential area of study would be to try to design new inhibitors that would be able to treat the predicted mutated pathogen strains, Donald said. He and his collaborators have begun exploring DHFR mechanisms of resistance to antifolate inhibitors in other pathogens and comparing the resistance profiles across species. The researchers also believe that their approach could be useful in other research areas. For example, they wrote that it could "be valuable in cases where it is more difficult to raise mutant strains or cell lines in vitro, such as with viruses or cancer cell lines."
Indeed, Donald told GenomeWeb that his lab is using the method to explore the possibility of predicting resistance in HIV strains. They've also used it in previous studies focused on designing new enzymes for antibiotics and allosteric inhibitors for protein-protein interactions in leukemia, and they've also designed probes that could likely pull down broadly neutralizing antibodies from the sera of HIV-infected individuals, he said.