Exploiting antimicrobial resistance

Abstract

Antibiotic resistance is a grave threat for public health. Understanding the mechanisms of resistance could lead to new drugs and therapeutic strategies against resistant pathogens.

The discovery of penicillin and its antimicrobial action by Alexander Fleming in 1928 was one of the greatest triumphs of biomedical research. Ever since, antibiotics have dramatically reduced the burden and death toll of infectious diseases and it enabled surgeries and other medical interventions now considered as standard that would have been highly risky without antimicrobials. Their importance is also illustrated by four Nobel Prizes in medicine and physiology for the discoveries of the sulphonamide prontosil, penicillin, streptomycin, and, in 2015, for the discovery of artemisinin and avermectin to treat malaria and worm infections.

… medical and public health advances are increasingly jeopardised by AMR, the ability of a microorganism to resist the cytotoxic effects of antibiotics.

However, the medical and public health advances are increasingly jeopardised by antimicrobial resistance (AMR), the ability of a microorganism to resist the cytotoxic effects of antibiotics. This has become a serious concern for public health: if nothing is done by 2050, AMR will cause 10 M more deaths per year (https://amr-review.org). Resistance complicates the management of infections, increases health care costs and can result in unnecessary deaths (https://cca-reports.ca/wp-content/uploads/2018/10/When-Antibiotics-Fail-1.pdf). But resistance has, in fact, been with us for a long time as microorganisms that produce antimicrobials need AMR genes to protect themselves against their own molecules. Resistance genes, often similar to the one currently found in clinical strains, were detected in 30,000‐year‐old permafrost sediments, clearly demonstrating that AMR genes predate chemotherapy 1.

The rise and cause of antimicrobial resistance

The 1950s and 60s were a rich period for the discovery and development of antimicrobials, which led to the general belief that a novel antimicrobial would always come to the rescue to deal with the slowly emerging AMR. This, unfortunately, was a false premise. The increase of AMR has been slow and steady under constant selective pressure on pathogens resulting from the increased use of antimicrobials in clinical and veterinary medicine but also in agriculture—and by ricochet in the environment—for promoting growth of livestock. In fact, much larger amounts of antimicrobials are used in agriculture than in human medicine. While not the rule—yet—several pathogens already resist all licensed drugs available on the pharmacy shelves. This grim portrait is further aggravated by the fact many large pharmaceutical companies have abandoned antibiotic development for economic reasons. It was rapidly recognised that a One Health approach around the Health–Agriculture–Environment triangle will be required to tackle AMR.

Bacteria possess a remarkable capacity to exchange DNA including AMR genes. These horizontal gene transfer events are mediated by plasmids (conjugation), phages (transduction) or by swapping naked DNA (transformation). While the exchange of AMR genes was first observed in phylogenetically close bacteria, this has expanded to include exchanges between more distant species. This cross‐species transmission along with plasmid evolution was further facilitated by transposable elements and integrons, ultimately creating multidrug‐resistant bacterial strains.

While not the rule—yet—several pathogens already resist all licensed drugs available on the pharmacy shelves.

While the acquisition of AMR genes and plasmids has a fitness cost for the bacteria, compensatory mutations emerged and plasmids developed addiction systems based on toxin and antitoxin genes to prevent their loss. Moreover, transferred DNA often contains additional genes providing the cell with new traits, such as virulence genes. This way, highly resistant bacteria have further evolved into very aggressive pathogens and decreasing the selective pressure may therefore not be enough to deal with the prevalence of already established resistant pathogens.

Viruses, protozoa, fungi or worms do not have the luxury of the natural DNA transfer/acquisition systems found in bacteria. Instead, point mutations and/or gene copy number variation are the major drivers of AMR in non‐bacterial microorganisms. Examples include mutations in key genes of HIV‐1 when monotherapy was used, resistance to most drugs in the malaria parasite, or the increased resistance to azoles in human or plant fungal pathogens.

The threat by AMR is now recognised at the highest political level. AMR was discussed at recent G7 and G20 meetings and at the 2016 UN General assembly, where health‐related issues are seldomly raised. This political visibility led to diverse national action plans aimed basically at better surveillance, improved stewardship, innovation and providing incentives for new business models and for the development of novel molecules or alternatives to antimicrobials.

Exploiting antimicrobial resistance

In this essay, we would like to highlight another angle and discuss how a better understating of AMR itself can be exploited for tackling AMR. Resistance now seems a ‘fait accompli’ for many pathogens, but our premise is that studies of AMR have the potential to highlight the Achilles’ heel of resistant pathogens. Indeed, studies of AMR can help by developing tools to recognise resistance early in infection for tailored chemotherapy; to enable more rational use of drugs and drug combinations to minimise the development of resistance; and by identifying intra‐cellular drug targets and defence mechanisms for the discovery and development of drug analogues. In the following sections, we will describe how AMR studies can be leveraged for improved diagnostics, drug combinations and drug‐target discovery (Fig 1).

Figure 1. Exploiting antimicrobial resistance studies.

A better understanding of resistance can lead to new concepts in the areas of diagnostics, drug combinations or in antimicrobial discovery. We listed a number of approaches rendered possible by exploiting our knowledge of AMR as well as tangible benefits.

Resistance now seems a ‘fait accompli’ for many pathogens, but our premise is that studies of AMR have the potential to highlight the Achilles’ heel of resistant pathogens.

Diagnostics in infectious diseases identifies the pathogen and determines its susceptibility to antibiotics. This information can be used for tailored treatment with narrow‐spectrum drugs, but current technologies that rely on the growth of the pathogen are slow—and doctors have to treat life‐threatening infections with broad‐spectrum drugs until the diagnosis is made. Studies of AMR can reveal resistance genes, which can be exploited for developing assays, such as PCR detection, to rapidly detect resistant pathogens. While the presence of an AMR gene is not sensu stricto equivalent to clinical resistance, it is usually predictive.

PCR‐based assays are well suited for targeted clinical diagnosis—for instance to determine AMR in tuberculosis—but since there are many pathogens that can cause a specific symptom and since there is a plethora of AMR genes for each drug, further innovations are required. Whole (meta)genome sequencing coupled to machine learning to detect AMR is rapidly evolving 2. Decreasing costs in genome sequencing, miniaturisation and improved bioinformatics workflow suggest that digital diagnostics will become an integral part of our arsenal for identifying AMR. In addition to detecting specific AMR genes, there are interesting developments in phenotypic assays to distinguish the differential responses of sensitive and resistant microorganisms to an antimicrobial 3. In this case, AMR is used as the discriminatory trait, based on the hypothesis that the phenotypic signatures of growth as sensed by various physical/chemical means, transcriptome or metabolome will differ between sensitive and resistant pathogens when in contact with a drug. Whatever the diagnostic technique, speed and robustness will be of the essence to have an impact for the patient's treatment.

More efficient combination therapies

Drug combinations are now the gold standard for treating tuberculosis, HIV‐1 and malaria and this is being emulated for many other pathogens. The combination of trimethoprim with a sulphonamide for treating bacterial infections is a classical (and rare) example. These two antibiotics work synergistically by inhibiting two steps in the folate pathway, but resistance to this combination is now widespread. Ideally, we should find combinations where the simultaneous resistance to both drugs becomes impossible. This is the concept of collateral sensitivity (CS): cells that are resistant to drug A are hypersusceptible to drug B 4. This is well established in cancer cells, and CS is now being studied experimentally in microbial pathogens. In vitro evolution experiments predicted that drug combinations leading to CS, or the cycling of the two partner drugs, will decrease the emergence of resistance.

The combination of molecules can be active against the same pathway/gene or against two different pathways. Since resistance to one of the CS drugs confers hypersusceptibility to its partner, one approach to find such drug combinations would be screening for drugs that kill preferentially the resistant organisms over the sensitive parents. Alternatively, a synthetic lethal genetic screen with resistant organisms could also lead to new targets for the discovery of CS drugs. Still, further work is required before CS can be exploited clinically. Indeed, it is likely to be context (genome)‐dependent, and the CS pathways could be eluded for one of cross‐resistance. Moreover, the CS approach has been tested mostly in cells for which resistance is caused by mutations and it remains to be seen whether it can be applied to AMR acquired by horizontal gene transfer with its presumed associated fitness cost. Ideally, CS could be also used to guide the evolution of resistance for preventing its emergence in the first place.

In addition to CS, drug combinations may comprise of one active molecule with a partner that is delaying the evolution of resistance, enhancing the effect of the active molecule or inhibiting the resistance mechanism itself 5. The latter strategy has so far been the most successful in the clinic. Current evidence suggests that gene products linked to DNA repair/recombination and known as evolvability factors facilitate mutation‐mediated resistance in microorganisms. While many of those gene products appear not to be essential, their deletion reduces the emergence of resistance, suggesting that their targeting in combination with an antimicrobial could indeed diminish the likelihood of a cell becoming resistant.

… drug combinations may comprise of one active molecule with a partner that is delaying the evolution of resistance, enhancing the effect of the active molecule, or inhibiting the resistance mechanism itself.

Resistance breakers

Another target for the second ‘helper’ drug is metabolic processes that modulate the activity of antimicrobials 6. The metabolism of resistant pathogens often differs from their sensitive counterparts, and specific intervention can restore susceptibility or enhance drug lethality. Screens are ongoing with metabolites or their modulators that could serve as adjuvants for potentiating the activity of antimicrobials. Similar to the CS combination discussed above, metabolic adjuvants are likely to be context‐specific, but the exploitation of metabolic networks in resistant microorganisms offers a refreshing perspective for combination therapy to minimise the development of, or to overcome, resistance.

… the exploitation of metabolic networks in resistant microorganisms offers a refreshing perspective for combination therapy to minimise the development of, or to overcome, resistance.

The most prescribed antibiotics are β‐lactams, but resistance mediated by β‐lactamases (enzymes degrading β‐lactam antibiotics) is also widespread. One of the most successful combination therapy used clinically—a β‐lactam and a β‐lactamase inhibitor—stems from a detailed understanding of the molecular mechanism of resistance 5. There are now several thousand β‐lactamases, and inhibitors are active against various classes of β‐lactamase enzymes. Alas, novel β‐lactamases insensitive to current inhibitors are emerging, hence the continuous needs for novel drugs or combinations. This is where a rapid diagnostic test would be helpful to enable an antibiotic combination prescription tailored to the genotype of the resistant organism.

Enzymes can also inactivate other classes of antibiotics such as aminoglycosides, and efforts are ongoing in generating inhibitors against those modifying enzymes although they have not yet reached the same level of development of β‐lactamase inhibitors. Another prevalent mechanism of AMR in fungi and Gram‐negative bacteria is drug efflux. Many academic and industrial efforts are invested in studying efflux pumps to restore drug sensitivity. At present, however, no molecules targeting efflux pumps for reversing AMR are close to clinical use.

New strategies for drug discovery

Strategies for antimicrobial drug discovery are based either on target molecule‐based screens or on whole‐cell phenotypic screens. Each has its strengths and weaknesses. A promising strategy is carrying out whole‐cell phenotypic screen as a first step. Once an active molecule with properties likely to be amenable to drug development is discovered, efforts are then made to find its target. A now‐standard strategy for target discovery consists in selecting for resistance against the drug of interest. Sequencing the genome of resistant microorganism should reveal mutations correlating to the drug selection 7. Some of these mutations will likely correspond to resistance mechanisms and others at the mode of action and/or at the cellular target; indeed, modification of the target is a frequent mechanism of AMR. Gain‐of‐function whole‐genome screens exploiting gene copy number variations and drug selection can also lead to new drug targets 8. Once a drug‐target duo has been identified, it is possible to further improve the activity of the drug through structure activity relationship helped by the atomic structure of the target. This approach, taking advantage of resistance for isolating drug targets, has already led to new drugs used clinically.

Gene arrays of all known AMR genes are now available, which can be used to for metagenomic screening of soil samples.

As mentioned, many bacteria/fungi producing antimicrobials possess self‐protecting mechanisms. In screens to find antimicrobial producers, drugs were used for enriching self‐protecting organisms, which increased the discovery rate of producers by orders of magnitude 9. Obviously, the new‐found molecule will likely resemble the drug used for selection, but if its scaffold is sufficiently different, it could be further tested and developed as an antibiotic.

Gene arrays of all known AMR genes are now available, which can be used to for metagenomic screening of soil samples. Positive signals could be a sign that the environment is polluted by AMR genes or alternatively may correspond to signatures of producers or for bacteria sharing the ecosystem of the producers. This has the potential of speeding up the isolation of antimicrobial producers. Sequencing the genomes of producers and mining for resistance genes may also highlight biosynthetic gene clusters with the potential for producing novel molecules.

Antimicrobial resistance is a serious public health threat that increasingly affects the global economy. As decision makers in politics, public health and industry become more aware of this threat and as society becomes better informed about AMR and its consequences, there is a momentum for worldwide efforts to address this threat. The leitmotiv of this essay was to show that resistance comes with a unique signature and weaknesses in each case and that exploiting those holds great potential for efforts to control and manage AMR.

EMBO Reports (2020) 21: e50249