For any new antibiotic, resistant bacteria typically show up in four years, or less. Penicillin resistance was reported clinically even before large-scale use of the antibiotic began in 1942. The battle against antibiotic-resistant bacteria demands new drugs and smarter, more responsible ways to use existing ones. Some researchers, however, are pursuing another type of weapon: drugs that sidestep natural selection. Less virulent bacteria would decrease the need for antibiotics, some reason, and drugs that drastically slow mutation rates might cut off evolution's power source.

Evaluating nonkilling approaches in the lab and clinic is tricky, and outfoxing evolution in the real world could prove even more difficult. Bacteria have evolved their way past ostensibly impenetrable barriers a number of times. Still, even if new approaches only hold off the inevitable, they could be a big win. "Half a loaf is better than none," says Abigail Salyers, professor of microbiology at University...


If bacteria are rendered harmless, there's no need to kill them, and no need for bacteria to develop resistance, says Stuart Levy, a Tufts University professor and director of the Center for Adaptation Genetics and Drug Resistance. He hopes to counter resistance by keeping bacteria from becoming virulent. In the 1990s, while investigating tetracycline resistance, he stumbled on a single point mutation that made bacteria resistant to multiple antibiotics. This led to the discovery of the mar operon, which so far has homologs in every sequenced bacterium. Function appears conserved as well, at least across the enterobacteria, a class that includes Escherichia coli, Shigella, Salmonella, Yersinia, and other human pathogens.

The operon's chief activator, MarA, controls more than 50 genes, many of which promote not resistance, but virulence.1 Several help bacteria to form biofilms. These microbial mats are less vulnerable to host immune systems and are blamed for some 80% of infections.

A drug that blocks MarA's activity, Levy reasons, would keep bacteria in a free-floating state, both less dangerous and easier prey for the immune system. Ideally, infections could be prevented in susceptible patients without resorting to antibiotics that select for resistance. Because the drug wouldn't actually kill the bacteria, mutations blocking its function wouldn't be highly selected for. Levy has cofounded a company to search for these drugs.


Joelle L. Bolt

Just as antibiotic mechanisms vary, resistance to them can take many forms. Plasmid or chromosomal gene-encoded mechanisms are generally directed at the antibiotic itself – degrading the drug, modifying it, or actively pumping it out of the cell. Mutation-derived resistance mechanisms commonly alter the drug target. More than one mechanism can provide resistance to the same antibiotic.

It's an intriguing idea, but difficult to test, says Molly Schmid, a microbiologist at the Keck Graduate Institute in Claremont, Calif. Antibiotic research has been built around standard, decades-old measures, such as minimal inhibitory concentration (MIC). Approaches inhibiting virulence will have a more difficult time establishing efficacy. "Because you don't have that simple test anymore, there would need to be a great deal of change in how we think about what we are treating," says Schmid. Clinical trials and even preclinical models could become difficult to design.

Inhibiting the mar operon should make bacteria easier to kill as well as less virulent, and as a result, inhibitors may provide a way to fight bacteria directly while validating more subtle approaches. "If we saw a bigger than expected effect," says Schmid, "there may be more benefits than we can readily realize through simple in vitro studies."


Floyd Romesberg, assistant professor of chemistry at Scripps Research Institute in San Diego, contends that evolution of new resistance genes can be stopped, or at least slowed by a factor of ten thousand. Conventional wisdom says that any population of a hundred million bacteria would have some resistant individuals just by chance. But Romesberg says the process can't be that passive. "Resistance is able to evolve efficiently against anything," he says. "So the diversity would have to be infinite."

If E. coli is plated onto media containing the antibiotic ciprofloxacin, the plates show no growth for the first four days of incubation. Then, the plates become covered with bacteria. During the initial lag, a tiny fraction of bacteria survive but do not grow, reasons Romesberg. Almost none of the resistant colonies come from slow-growing bacteria that start out resistant, he hypothesizes, because bacteria from 99.99% of these colonies regrow in one day, not five.

By deleting genes, Romesberg's team found a handful of otherwise healthy strains that don't regrow on ciprofloxacin-laced plates. All these strains lack genes that are part of the so-called SOS response, which allows bacteria to repair DNA, albeit sloppily, in a crisis. Romesberg has homed in on LexA, a protease that represses the SOS response in healthy cells. When DNA is damaged, single-stranded DNA bound to a regulatory protein causes LexA to cleave itself, which derepresses the production of nonessential DNA polymerases that conduct sloppy repair and introduce mutations.

Romesberg's group created a version of LexA that still represses the SOS response but can't cleave itself. The researchers injected mice with pathogenic E. coli expressing either normal or noncleavable LexA; they then treated the mice with antibiotics for three days. Resistant bacteria showed up in the mice infected with normal E. coli, but not in mice infected with E. coli expressing noncleavable LexA.2 Romesberg says that the sloppy repair provided by the SOS response accelerates the evolution of resistance. He has cofounded a company to develop SOS-response inhibitors.


But the story is hardly a lock. "Self-mutagenesis is a very dangerous strategy for cells to use, especially since many stresses that would induce this, in theory, are ones for which no beneficial mutations could solve the problem," says John Roth, professor of microbiology at University of California, Davis. "In these situations, mutagenesis would be suicidal."

Romesberg agrees that aspects of his mechanism are speculative. For example, during the five days before the appearance of lawns of resistant bacteria, exponential growth of very small populations could occur, rather than the inhibited growth of most survivors. That still does not change the finding that LexA is crucial to resistance evolution. "What matters," says Romesberg, "is that there are proteins that, when you inhibit them, the cells don't mutate."

So far, Romesberg has worked only with E. coli and only with antibiotics that affect DNA and RNA synthesis, but he says the mechanism is turned on whenever stress stalls DNA replication: "My wild speculation is that it utterly underlies evolution."


Salyers applauds Romesberg's work as "trying to get to the root of the problem," and commends Levy's work for considering "real-world ecology" of the spread of resistance. But Salyers remains skeptical: "It's a great strategy, but people that tell you it's a salvation are fooling you."

Both Romesberg's and Levy's approaches would require two drugs to fight infections: one to kill the bacteria, and another to keep them susceptible. Evolving resistance to overcome two barriers is much harder than overcoming resistance to a single agent (see box above). Still, such resistance happens.

Bacteria tend to collect resistance mechanisms, and countering one kind of resistance may be completely ineffective if another kind is present. For example, says Salyers, efflux pumps were once thought to be the only mechanism of resistance to the antibiotic tetracycline, but some forms of bacteria have an internal protein that blocks tetracycline's effects. "You can block all the efflux pumps you want," she says; "It won't make a difference against that kind of resistance." In fact, all approaches to stall evolution suffer from a common weakness: They work in laboratory conditions, but are unproven in the varied environment of human societies. Salyers and others say that most resistant bacteria do not evolve resistance on their own but acquire it from mobile genetic elements.

Even so, spontaneous mutation ultimately underlies all resistance mechanisms, and it is particularly important in long-lasting infections such as tuberculosis. For certain antibiotics, such as rifampin and the quinolines (including ciprofloxacin), evolving resistance requires mutations in three or fewer base pairs in one gene. Partly for this reason, and partly because these are relatively new antibiotic classes, mutation seems more important for generating resistance. (Nonetheless, resistance plasmids for the quinolines have been reported.)

Salyers welcomes evolution-thwarting approaches, but insists that they will only help, not solve, problems of antibiotic resistance. Schmid agrees that these approaches challenge scientists and clinicians "to believe there are other ways [to fight bacteria] than just using a hammer and smashing them." But she doubts that evolution can be kept down: "If there are simple mutational changes that will allow the bacteria to grow, then they're going to do it."

Double-teaming Resistant Bacteria

Plants have already evolved a double-pronged strategy to combat resistance. Kim Lewis, professor of biology at Northeastern University, found the first natural example of synergistically acting compounds while studying bacterial multidrug-resistant (MDR) pumps. Lewis assumed the pumps evolved to defend against natural products, but when he compiled a list of known substrates to a particular pump that ejected hydrophobic cations, he found only artificial compounds. "I had drawn up all this evolutionary logic," he recalls, "but all the compounds had been synthesized just 30 years ago."

Determined to find a natural-product substrate, Lewis searched through the Merck Index for hydrophobic cations with no known activity. He stopped at berberine, an alkaloid made by Berberis, a plant native to Colorado. "We test it as an antimicrobial and it's lousy. Then we test it against a strain where we've knocked out the MDR pump and it's fantastic."

Working with phytochemists in Colorado, Lewis worked out that Berberis fights bacteria not just with berberine, but also with 5'-methoxyhydnocarpin (MHC), a compound previously known only as a minor component of a traditional leprosy treatment. MHC potently inhibits the MDR pump. Together, these compounds provide powerful antimicrobial protection.

Pump inhibitors are already in clinical trials but are thwarted by bacterial diversity. Pump inhibitors that work well against one species are often powerless against another, and clinicians rarely have time or means to identify the pathogen they're fighting. But when they do, pump inhibitors could make nearly impotent antibiotics into wonder drugs once again.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?