How to fight antibiotic resistance

Pharmaceutical and biotechnology companies have failed to invest in new classes of antibiotics because the market is seen as “risky and relatively unprofitable,” according to a report from Elias Mossialos of the London School of Economics and Political Science.

Anil Koul, vice president of discovery and partnerships, global public health R&D at Janssen Research & Development, agrees: “Despite the huge societal costs of AMR, global investment remains low and there is no viable market for new antibiotics.” There is not a viable market because antibiotics are used for short periods of time and are a rare class of drugs that doctors are actively discouraged from using, to prevent the development of resistance.

Pharmaceutical microbiologist Rolf Müller at Helmholtz Institute for Pharmaceutical Research at Saarland University in Germany describes the current investment in new antimicrobials as “Terrible. It’s a market failure” — and he sees no short-term solutions. Müller points out that in the longer term, governments and their citizens must all recognize the desperate need for new antibiotics. “We have to emphasize that society needs to value antibiotics better,” he says. “That value also requires more money to be paid for them.”

New payment models

Some governments are piloting novel approaches to encourage investment. As one example, the UK government is exploring a subscription-based model through which an antibiotic-producing company would receive a yearly payment for a specific drug. The size of the payment would depend on the government’s assessment of the value of the antibiotic, not the amount of the drug prescribed. In the United States, the PASTEUR Act would — if passed — create a similar value-based pricing system, which could drive the development of new antimicrobials.

“To address the significant hurdles facing antibiotic R&D, especially in the late stages of development, there must be tailored pull incentives to drive innovation and spur further R&D investment,” Koul says. Subscription-based methods would be one of these pull mechanisms, with companies lured by the guaranteed income.

Conversely, push mechanisms could stimulate the development of new drugs by supporting the cost of early research. “The push incentives help to fund R&D work, but do not solve the market failure we are observing in the antimicrobial space,” says Joe Larsen, vice president of clinical development at Locus Biosciences. “We have learned from the mobilization of the biotech industry to develop biodefense medical countermeasures that both a push incentive that subsidizes R&D and a robust pull incentive that rewards successful development and commercialization are required to sustainably generate a diverse pipeline of new and effective products.”

Finding the unknown

Despite all the advanced approaches for searching for new antibiotics, most of the ones on the market come from microorganisms, including fungi and bacteria. Nonetheless, systems biologist Elizabeth Shank of the UMass Chan Medical School believes that many potential antibiotics from bacteria are being missed.

“From genome sequences of bacteria, we can see that they have the ability to make a lot of interesting compounds that we potentially haven’t ever characterized,” Shank says. “So, we take a biological perspective: Why would bacteria make these molecules?”

It might be that these uncharacterized compounds are there to kill other bacteria. So, Shank cultures bacteria together. “We can use that additional environmental stimulus of having another bacteria nearby to stimulate the production of not just compounds that are made all the time but also ones that are only turned on in response to those other signals,” she says. “So, we’re using coculture to try to stimulate production of new metabolites that could lead to new antibiotics.”

At The Rockefeller University, organic chemist Sean Brady and his colleagues take another approach to looking for new antibiotics from the millions of species of bacteria. These scientists apply synthetic bioinformatics to natural products, in what they call ‘bioinformatic prospecting’. Brady starts with the bacterial genome, looks for a group of genes that do not make known natural products, applies software tools to predict the structure of these products, and then synthesize them. “We synthesize the simple ones at this point,” he says.

One study produced cilagicin. “It’s effective against key pathogens, and it doesn’t develop resistance that we’ve seen so far,” Brady says. That effectiveness includes killing troublesome pathogens, including vancomycin-resistant Enterococcus bacteria. Brady suspects that cilagicin escapes bacterial resistance by binding two targets. “Most antibiotics today have a single target,” he says. “If you have two things the bacteria need to get around — two targets — that’s more challenging.”

A synthetic-chemistry approach could produce many new antibiotics. For example, Brady and his colleagues looked for antibiotics that might bind menaquinone, which is a form of vitamin K that has a key role in the energy metabolism of many bacteria. The scientists used computation to predict gene clusters than might make small molecules that would target menaquinone, and then they synthesized them. Some of the resulting antibiotics even killed methicillin-resistant Staphylococcus aureus.

“This approach is not going to work on all gene clusters, but there’s a lot of these that we think it can work on,” Brady says. “One of the problems with modern genome engineering is that it has been relatively slow, but we can make predictions about potential antibodies and synthesize them very quickly and cheaply.”

AI approaches

Many computational approaches to finding new antibiotics will involve artificial intelligence (AI). “We see the opportunity that AI can be harnessed, applied to large datasets that we can create, to discover and design new antibiotics for some of the more problematic pathogens that are facing humankind,” says Jim Collins, Termeer Professor of Medical Engineering & Science at the Massachusetts Institute of Technology. Working with a TED initiative called The Audacious Project, Collins and his colleagues have launched the Antibiotics-AI Project to develop new classes of antibiotics.

By applying AI to antibiotic discovery, scientists can explore more compounds and do it more quickly. In the past, a large pharmaceutical company might screen a million compounds in search of a new antibiotic. By using AI, Collins and his colleagues screened 110 million molecules in just 3 days, which led to the discovery that halicin, previously researched as a treatment for diabetes, has antimicrobial properties. “It was a remarkable reduction in both resources and time,” Collins says.

This work relies on various AI algorithms, such as neural nets, and powerful computing — either a network of computers or a supercomputer. Plus, AI needs data to analyze. For that, Collins says, they are “exploring a number of different publicly available datasets, but we also are working to create our own.”

With that data and computational power, Collins and his colleagues hope to discover and generate molecules that could be used as antibiotics. “We’ve been exploring how one can use a fragment-based approach as a starting point for generative AI efforts,” Collins says. These fragments consist of “a relatively small number of atoms together in a scaffold, that might be the basis around which one can begin to design and synthesize molecules with desired properties,” Collins explains. Then, the scientists could use AI to predict if the designed molecules interact with a specific target or exhibit a desired property such as antibacterial activity.

New targets

The traditional method of looking for new antibiotics usually takes one of two approaches: focusing on the drug-resistant pathogens, or searching for broad-spectrum antibiotics.

“‘Targeted’ or ‘narrow-spectrum’ antibiotics [can be developed] for certain drug-resistant pathogens, like carbapenem-resistant Acinetobacter baumanii, which is on the World Health Organization’s list of antibiotic-resistant priority pathogens,” Koul says. This targeted approach can also be applied to various types of treatments, including engineered bacteriophages (viruses that naturally infect and kill bacteria), novel antibodies, and antibody–drug conjugates.

Koul argues that existing in vitro tests, which use nutrient-rich media, might fail to identify promising antibiotics. “We need to explore alternate metabolic states of these pathogens in nutrient-deprived or other physiological states that mimic the host microenvironment,” he says. Achieving this, he says, will take “a paradigm shift” in antibiotic-screening strategies.

New angles of attack are needed, such as bacterial metabolism, which Koul’s group is targeting. “The aim is to block key metabolic pathways that bacteria use for their energy sources” with one goal being to “impact bacterial survival” through chemical inhibition, he says.

The sources of tomorrow’s antibiotics might come from unexpected research, such as a US Army project that used oligoelectrolytes to charge cellphones on a battlefield. “Bacteria are power cells, and conjugated oligoelectrolytes were designed to insert into bacterial membranes and function as electron transporters to potentially power electronic devices,” says geneticist Michael Mahan of the University of California, Santa Barbara. Conjugated oligoelectrolytes (COEs) are small, synthetic molecules that “share a modular structure that can spontaneously interact with lipid bilayers,” he explains.

Mahan and his colleagues repurposed COEs as potential antibiotics. “From screening a diverse array of COE chemical variants for antibacterial activity and low cytotoxicity in cultured mammalian cells, COE2-2hexyl was the lead candidate,” he says. Although considerable work remains to be done on COE2-2hexyl, Mahan and his colleagues showed that it can attack a broad range of microbes, and it did not trigger resistance in bacteria.

Mahan and his colleagues also developed a new antibiotic-susceptibility test, which they argue better mimics antibiotic action in the body. Using this test, they found that some antibiotics approved by the US Food and Drug Administration were effective at treating infections that were thought to be multidrug resistant. These off-the-shelf antibiotics were “not prescribed because the gold-standard test that physicians rely on indicates they will not work,” says Mahan.

By simulating conditions in the body, the new test identified several effective antibiotics rejected by standard testing. Consequently, some of the antibiotics in today’s antimicrobial toolbox might be more useful than generally believed. As Mahan concludes: “The new test might improve the way antibiotics are developed, tested, and prescribed.”

Equitable access

Even if the research and healthcare communities discover and develop new antibiotics, that is not enough. “A scientific breakthrough in antibiotic R&D should reach those most in need, including those in low- and middle-income countries,” Koul says. That takes what Koul calls “last-mile delivery approaches.” As an example, he points out that Johnson & Johnson made bedaquiline, its medicine for multidrug-resistant tuberculosis, available in nearly 160 countries and delivered more than half a million courses of the medicine around the world.

Making antibiotics available worldwide, however, is complicated. As Koul says, “Building that last-mile framework for any new antibiotic will require close partnerships with governments, regulators, private institutions, and civil society.”

If such partnerships fail to solve the AMR crisis, some aspects of healthcare could revert to the past. As Shank notes, “150 years ago, you could die because you scratched your finger and you just happened to get bacteria in there.” This is a past that no one wants to return to.

Author information
Mike May
Freelance writer and editor, Oak Harbor, WA, USA.

Nature Medicine 29, 1583-1586 (2023)

doi: https://doi.org/10.1038/d41591-023-00043-5