There’s been a lot of recent attention to the threat of antiobiotic resistence (see this recent NYTimes piece for example). As a quick summary, overuse of antibiotics by doctors and farmers has triggered the evolution of bacteria that are resistant to all available antibiotics. The conclusion follows that more needs to be done to curtail unnecessary antiobiotic use, and to develop novel antibiotics that can cope with the coming onslaught of antibiotic-resistant bacteria.

Many potential solutions have been proposed, most of which are policy-based. Incentive strategies could help spur the discovery and development of new antibiotics (see this review for example). Some even propose the establishment of large prizes for antibiotic development, while others advocate the establishment of new research centers focused on the problem. Yet others suggest that greater education of doctors and farmers to limit excessive antibiotic usage is of highest priority. While all of these suggestions have their place, the development of antibiotics raises specific serious scientific challenges that are less widely understood.

For one, antibiotic development raises challenges that aren’t present in other types of drug discovery. Most drugs today are explicitly created by chemical synthesis (with the exception of antibodies, which we won’t discuss here). Unfortunately for chemists though, antibiotics tend to have complex molecular structures which complicate the use of standard chemical techniques. For example, compare the structures of the antibiotic azithromycin to that of acetaminophen (Tylenol). The molecular structure of azithromycin is significantly larger than that of acetaminophen. The methods of standard organic chemistry struggle to make meaningful modifications to the structure of complex molecules like azithromycin, and the development of azithromycin itself was something of a chemical odyssey. The herculean effort required to chemically synthesize new antibiotics is an evident scientific barrier.

There has been some recent progress in synthetic chemistry that might eventually make it easier to create novel antibiotics. See here and here for descriptions of recent synthetic chemistry techniques that could allow for easier modifications of complex antiobiotic-like molecules. However, these advances are only first steps, and significantly greater work will be required over the next decade to mature and standardize such techniques.

Other parts of drug discovery hve benefited from the advent of websites like Pubchem and Chembl that allow for rapid searches through databases of biochemical experiments. However, these databases contain limited information about antibiotics, in part due to the difficulty of synthesizing antibiotics. Luckily, there has been recent progress here, with the establishment of CO-ADD which offers a service where compounds are tested for anti-microbial activity for free. This consortium is planning on releasing a portion of its dataset to the public in 2017.

As another note, standard drug-discovery has started to become increasingly computerized. Techniques based on machine-learning and on physical simulations are slowly becoming standard in conventional drug-discovery. Unfortunately, these techniques haven’t yet been heavily applied to antibiotic discovery. For machine-learning techniques, the lack of large databases of antibiotic experiments has meant that training data has not been available. This should change in the near future once CO-ADD data becomes available. On the side of physical simulations, the large computational requirements of these methods has meant that such techniques are just now being applied to simpler compounds (such as for kinase inhibitors by the Chodera lab). A few groups, such as the Bowman lab have recently started applying similar computational techniques to antibiotics, but work is just beginning in this space (see this recent paper for example)

To summarize, antibiotic development has struggled because of the complex molecular structure of antibiotics, the lack of public databases of antibiotic experiments, and because of the dearth of computational work. As we’ve seen though, there are hopeful signs of progress on each of these fronts. I’d encourage policy developrs and potential scientific donors to direct resources and money to these specific research directions in addition to pursuing the policy-driven approches that are commonly proposed.