How Do Antibiotics Work, and Why Do Bacteria Stop Responding?

In 1928 Alexander Fleming returned from holiday to find a mould growing on a dish of bacteria in his London laboratory, with a clear ring around it where the bacteria had died. The mould was a Penicillium and the substance it secreted was named penicillin, but Fleming could not purify it and moved on. It took Howard Florey and Ernst Chain in Oxford, in 1940 and 1941, to produce enough to cure infected mice and then treat a first patient, and American factories in the war to make it in quantity. By 1945 it was saving thousands of wounded soldiers, and the three shared the Nobel Prize.

Penicillin's trick, worked out later, is that it blocks an enzyme bacteria use to build their cell wall. A growing bacterium keeps making wall and, unable to finish it, bursts under its own internal pressure. Human cells have no wall, so the drug passes them by.

Every antibiotic in use exploits one of a handful of differences between bacterial and human cells:

• •Breaking the wall: penicillins and cephalosporins stop the wall being built; vancomycin binds its building blocks
• •Jamming the protein factory: bacterial ribosomes differ from ours, and tetracyclines, macrolides such as erythromycin, and aminoglycosides bind them and halt protein production
• •Blocking DNA copying: fluoroquinolones such as ciprofloxacin trap the enzyme that unwinds bacterial DNA
• •Starving them of folate: sulphonamides and trimethoprim block a vitamin pathway bacteria must run for themselves and humans get from food
• •Wrecking the membrane: polymyxins punch holes in the outer membrane, a last resort because they damage human kidneys too

Viruses are not cells. They have no wall, no ribosome and no metabolism of their own; they are packages of genetic material that hijack the machinery of the cells they infect. There is nothing in a virus for an antibiotic to attack, and taking one for a cold, flu or COVID-19 does no good and some harm, since every dose also kills off the harmless bacteria in the gut and gives any resistant survivors room to grow. Antiviral drugs exist and work on different principles, blocking the enzymes viruses bring with them or the receptors they use to enter cells.

Most sore throats, coughs and ear infections are viral, which is why doctors now hold back antibiotics for them, and why the pressure from patients to prescribe anyway is one of the drivers of the problem in the next section.

Bacteria divide every twenty minutes or so and mutate as they go, so in any large population a few cells will by chance carry a change that blunts an antibiotic: an enzyme that chops the drug up, a pump that expels it, a slightly altered target it no longer fits. In the presence of the drug those few survive while their neighbours die, and within days the survivors are the population. Fleming warned of exactly this in his Nobel lecture in 1945, describing a man who took too little penicillin, bred resistant microbes and passed them to his wife.

Worse, bacteria trade genes. Resistance genes often sit on small loops of DNA called plasmids that can pass from one bacterium to another, including between different species, so a resistance that evolves in a harmless gut microbe can be handed to a dangerous one. Hospitals, where sick people, antibiotics and bacteria are all concentrated, are where it spreads fastest. MRSA, the resistant staphylococcus that haunts surgical wards, carries a gene that makes a wall-building enzyme penicillins cannot bind, and it appeared within two years of methicillin's introduction in 1959.

Resistance cannot be stopped, only slowed, and the methods are mostly about reducing the selection pressure. Prescribing antibiotics only for bacterial infections, and choosing the narrowest drug that will work, gives resistant strains fewer chances. Finishing a course was the old advice; the newer view is that the right length is whatever clears the infection, and that longer courses breed more resistance rather than less. Around two thirds of the world's antibiotics are given to farm animals, mostly to speed growth, and several countries have now banned that use.

The pipeline of new drugs has been thin for decades, because an antibiotic that must be used sparingly is a poor investment for the company that develops it, and governments have begun paying for new ones by subscription rather than by the dose. Vaccines reduce the infections that need treating in the first place. And bacteriophages, viruses that prey on bacteria, are being revived as a treatment a century after they were first tried. The WHO counts antimicrobial resistance among the ten greatest threats to global health, and the count of deaths directly caused by resistant infections is already over a million a year.

In 1884 the Danish scientist Hans Christian Gram developed a staining technique that is still one of the first tests run on a bacterial sample. Bacteria are dyed purple, rinsed with alcohol, and counterstained pink. Some keep the purple and are called Gram-positive; others lose it and show pink, and are called Gram-negative. The difference reflects their structure. Gram-positive bacteria have a thick, exposed cell wall that traps the dye. Gram-negative bacteria have a thinner wall wrapped in an extra outer membrane, which the alcohol disrupts, letting the purple wash out.

That outer membrane is a barrier to many drugs, which is why the stain matters for treatment. Vancomycin, for instance, is a large molecule that cannot easily get through it, so it works well on many Gram-positive bacteria but poorly on most Gram-negative ones. Drugs that act on both groups are called broad-spectrum; those that hit a narrower range are narrow-spectrum. Because a Gram stain takes minutes, while growing and fully identifying bacteria can take a day or more, it helps a doctor choose a sensible first drug quickly, then switch to a more targeted one once the full results arrive.

People often say that taking too many antibiotics makes their body resistant. It is the bacteria, not the person, that become resistant, and the difference matters. A resistant strain bred in one person's gut or skin can spread to family members, hospital patients and animals who have never taken the drug, so resistance behaves less like a personal side effect and more like pollution. It is also why someone who has rarely had antibiotics can still catch an infection that fails to respond to them, and why surveillance of resistant strains is carried out across whole hospitals and countries.

The second misunderstanding is about allergy. Between five and ten percent of people carry a label saying they are allergic to penicillin, often from a childhood rash that may have been caused by the infection itself. When such patients are properly tested, the great majority turn out not to be allergic, and many genuine allergies fade over the years. The label is not harmless. It steers doctors towards broader-spectrum alternatives that can work less well, cost more, cause more side effects and add to the pressure that breeds resistance, which is why many hospitals now offer allergy testing to remove mistaken labels.

Most antibiotics were not designed by chemists; they were found, and the most productive place to look turned out to be the soil. In 1943 Albert Schatz, a graduate student in Selman Waksman's laboratory at Rutgers University in New Jersey, isolated streptomycin from a soil bacterium of the genus Streptomyces. It was the first drug that worked against tuberculosis. Waksman's team screened soil microbes systematically, growing organisms from the ground and testing what they secreted against disease-causing bacteria, and the same approach produced a large share of the antibiotics discovered over the next two decades, many of them from Streptomyces.

Soil microbes make these compounds to compete with their neighbours, which carries a sobering implication: resistance is just as ancient. Researchers analysing DNA preserved in permafrost about 30,000 years old found genes for resistance to several antibiotics, including vancomycin, long before any human used the drugs. A harder problem today is that most soil bacteria refuse to grow in a laboratory dish. One answer has been a device that lets bacteria grow in their native soil inside tiny chambers, which in 2015 led to the discovery of teixobactin, a new kind of antibiotic from a previously uncultured species. The ground beneath our feet is still far from exhausted.

Antibiotics kill bacteria by attacking their cell walls, their ribosomes, their DNA-copying enzymes or their metabolism, none of which exist in the same form in human cells, and they do nothing to viruses because viruses have none of those things. Bacteria evolve resistance by mutation and by swapping genes, and slowing that down means using the drugs less and more precisely while new ones are found.