Main

In antibacterial drug design, developing compounds with the mechanistic potential to kill bacteria is only part of the job. The drug also needs to get into the cell and stay there long enough to kill it, as many bacteria come equipped with powerful pumps for the active removal of a wide variety of compounds, including many would-be antibiotics. The 'reverse genetics' approach, in which chemical libraries are screened against a target bacterial protein to identify putative inhibitors, has been a prevailing paradigm in antibiotic design. However, according to McMaster University researcher Eric Brown, “that's failed miserably in antibacterial drug discovery... [and] I think that the key failure is that we don't understand the rules for getting things into cells.”

Brown and his colleagues were interested in developing a new, phenotype-based approach, which could first identify promising drug candidates and then reveal their targets. They began with a growth inhibition screen, using a diverse library of 8,640 commercially available compounds to identify chemicals that inhibited growth of the hyperpermeable Escherichia Coli strain MC1061, first in liquid culture and then on solid media.

These screens winnowed the field to 49 lead compounds, which were tested in a multicopy suppression assay, an approach that has a history of successful use for the identification of antibiotic targets. Such assays involve the identification of gene products from a bacterially expressed genomic library that, when overexpressed, can suppress the growth-inhibitory properties a given compound. By this approach, Brown's team identified suppressors for 33 of their 49 lead compounds.

Nearly all of the compounds were suppressed by the same gene, acrB, which encodes the inner membrane component of an important drug efflux pump (also notorious for making the Pseudomonas aeruginosa bacterium the drug-resistant scourge of hospital wards worldwide). This finding is unsurprising, according to Brown: “I think that, more and more, it's being appreciated that permeability is not so much about structural integrity of the cell, but more about molecules being pumped out... [and] very little is understood about what makes something a substrate for efflux.” The group also found that two of the compounds, 1a and 2a, appeared to target a second gene, folA, encoding dihydrofolate reductase, a known drug target. Follow-up experiments confirmed that elevated bacterial expression of folA increased the amount of 1a or 2a needed to inhibit growth.

Brown is slightly disappointed by the extent to which efflux pumps appear to drown out target identification, but he also sees important benefits for future research. “One of the things that comes out of this paper, I think, is a way to better understand what is the substrate specificity of an efflux pump,” says Brown, who indicates that his team has already learned quite a bit about how properties such as the extent of hydrophobicity might lead to increased drug efflux. His team is seeking ways to potentially identify additional targets that might be lost amid the 'noise' generated by efflux pump suppressor genes, but Brown believes that their system already has a lot to offer: “[From] eight and a half thousand molecules, we pulled out two compound-target pairs, and that's not really such a bad success rate. I think that as is, you could take this forward with much higher throughput, and I daresay that there are companies that would have the resources to do that.”