How Antibiotics Work

The core idea: hit what they have and we don't

Every antibiotic that works rests on a single trick: find a piece of machinery that is essential to a bacterium but absent or different in a human cell, then poison it. This is selective toxicity, and it is the whole game. A bacterium and one of your cells are profoundly different objects despite both being alive. The bacterium is a single, wall-bound cell roughly 1–2 micrometres long, replicating its lone circular chromosome and pumping out proteins on a ribosome built from a different set of RNAs and proteins than yours. Your cells are eukaryotic: no peptidoglycan wall, a different (80S) ribosome, a nucleus, and folate you simply eat rather than build. Each of those gaps is a place to slip a knife.

That is also why antibiotics are useless against viruses. A virus is not a cell — it has no wall, no ribosome of its own, no gyrase, and no folate pathway. It hijacks your machinery to reproduce. There is literally nothing for the drug to bind. Taking amoxicillin for a cold does nothing for the cold and quietly selects for resistance in the harmless bacteria you carry. Knowing the target tells you exactly when a drug will and won't work.

Target 1 — the cell wall

Bacteria wrap themselves in peptidoglycan (murein): long sugar chains of alternating N-acetylglucosamine and N-acetylmuramic acid, stitched together by short peptide cross-links into a single, bag-shaped molecule called the sacculus. This mesh is what lets a cell hold an internal osmotic pressure of roughly 3 atmospheres in a Gram-negative like E. coli and up to ~25 atmospheres in a thick-walled Gram-positive like Staphylococcus aureus without exploding. Humans build no peptidoglycan at all — so the enzymes that assemble it are a clean target.

The beta-lactams — penicillins, cephalosporins, carbapenems — are the famous example. They are structural mimics of the D-alanyl-D-alanine terminus of the wall precursor. The transpeptidase enzymes (the penicillin-binding proteins) that normally clamp onto that D-Ala-D-Ala to make a cross-link instead grab the beta-lactam, which snaps its strained four-membered ring open and forms a permanent covalent bond, jamming the enzyme dead. Cross-linking stops. But the cell's own remodeling enzymes (autolysins) keep snipping the wall open to let it grow — now with nothing re-stitching the gaps. The weakened sacculus gives way and turgor pressure bursts the cell. Crucially, beta-lactams only kill growing bacteria that are actively building wall; a dormant cell building nothing is untouched. Vancomycin attacks the same pathway from a different angle, binding the D-Ala-D-Ala substrate itself so the enzymes can't reach it.

Target 2 — the 70S ribosome

To live, a bacterium must constantly translate mRNA into protein, and it does so on a 70S ribosome — a 30S small subunit (16S rRNA + proteins) and a 50S large subunit (23S + 5S rRNA + proteins). Your cytoplasmic ribosome is a chunkier 80S. That size and sequence difference is enough for a whole arsenal of drugs to bind the bacterial ribosome and ignore yours, at least mostly. The ribosome is the single most heavily drugged target in all of antibacterial medicine.

• Aminoglycosides (gentamicin, streptomycin) bind the 30S subunit, cause the ribosome to misread codons, and produce garbage proteins; they are bactericidal.
• Tetracyclines block the 30S A-site so a charged tRNA can't dock — translation stalls; bacteriostatic.
• Macrolides (erythromycin, azithromycin) plug the 50S exit tunnel so the growing peptide chain can't escape; bacteriostatic.
• Oxazolidinones (linezolid) and chloramphenicol block the 50S peptidyl-transferase that forms peptide bonds.

Here selective toxicity gets uncomfortable. Your mitochondria descend from an engulfed bacterium (the endosymbiotic theory) and still run a bacteria-like ribosome. Drugs that hit the 70S ribosome can therefore nick mitochondrial protein synthesis — which is why aminoglycosides can damage hearing and kidneys and why long-course linezolid can suppress the bone marrow.

Target 3 — DNA gyrase and topoisomerase IV

Bacterial DNA is a closed circle. Every time the replication fork or RNA polymerase runs along it, the helix winds up ahead of them like a twisted phone cord. Bacteria relieve that torsional stress with DNA gyrase (a type II topoisomerase) and its cousin topoisomerase IV, which cut both strands, pass DNA through the break, and reseal it. The fluoroquinolones (ciprofloxacin, levofloxacin) bind the enzyme–DNA complex at the moment the strands are cut and freeze it there. The result is lethal: the chromosome is left full of permanent double-strand breaks, and the cell dies. Humans have type II topoisomerases too, but the bacterial version is structurally distinct enough that quinolones grip it far more tightly — though not perfectly, which underlies the tendon and nerve side effects at high exposure.

Target 4 — the folate pathway

Cells need tetrahydrofolate to make the building blocks of DNA (thymidine) and several amino acids. Bacteria synthesize folate from scratch starting with PABA; humans can't — we get folate from food. That single metabolic difference is a beautiful target. Sulfonamides mimic PABA and competitively block the enzyme dihydropteroate synthase; trimethoprim blocks the next enzyme, dihydrofolate reductase. Given together (co-trimoxazole) they hit two consecutive steps and the effect is synergistic — the bug is starved of nucleotide precursors and stops dividing. Because we don't run this pathway, these drugs are gentle on us. This was, historically, the breakthrough: the sulfonamides of the 1930s were the first synthetic antibacterials, predating mass-produced penicillin.

Bactericidal vs. bacteriostatic, and the five classes at a glance

Some drugs kill (bactericidal); others merely stop growth and let the immune system finish the job (bacteriostatic). The distinction is real but slippery — concentration-dependent, and it matters most in patients with weak immunity or hard-to-reach infections like endocarditis, where you want outright killing.

Target	Example class	What it blocks	Effect	Bacterial-only?
Cell wall	Beta-lactams, vancomycin	Peptidoglycan cross-linking	Bactericidal	Yes — no human wall
30S/50S ribosome	Aminoglycosides, tetracyclines, macrolides	Protein synthesis	Static or -cidal	Mostly — mito risk
DNA gyrase / topo IV	Fluoroquinolones	DNA supercoiling / repair	Bactericidal	Mostly — distinct enzyme
Folate pathway	Sulfonamides, trimethoprim	Tetrahydrofolate synthesis	Bacteriostatic	Yes — humans eat folate
RNA polymerase	Rifamycins (rifampicin)	Transcription initiation	Bactericidal	Yes — distinct subunit

The therapeutic window — the gap between curing and harming — tracks these columns. Penicillins are extraordinarily safe because the target is genuinely absent in us; aminoglycosides are tightly dosed and blood-monitored because their target overlaps with mitochondria.

Why spectrum depends on the envelope

Whether a drug reaches its target at all depends on the cell envelope it must cross — which is why Gram-positive vs Gram-negative bacteria respond so differently. Gram-positives wear a thick (20–80 nm) peptidoglycan coat but no outer membrane, so large drugs like vancomycin can diffuse in. Gram-negatives have a thin peptidoglycan layer sandwiched under an extra outer membrane studded with lipopolysaccharide; that membrane is a near-impermeable barrier that vancomycin can't cross at all, and small drugs must squeeze through protein pores called porins. Lose or narrow those porins by mutation and the cell is suddenly resistant — entry denied. This is why "broad-spectrum" versus "narrow-spectrum" is, at root, a story about getting past the wall.

Resistance: natural selection in fast-forward

Bacteria fight back, and the mechanisms mirror the targets. In a population of billions, rare cells already carry a defense before the drug ever arrives — the drug doesn't create resistance, it selects for it, the same Darwinian logic as in natural selection, just compressed into hours. Four broad strategies:

• Destroy the drug. Beta-lactamases hydrolyze the beta-lactam ring before it reaches its target; extended-spectrum and carbapenemase variants now defeat even last-line drugs.
• Pump it out. Efflux pumps actively export tetracyclines, fluoroquinolones, and more, keeping internal concentration below lethal.
• Change the target. A point mutation in the gyrase, or in the penicillin-binding protein (the basis of MRSA's mecA-encoded PBP2a), means the drug can no longer grip.
• Block entry. Lose a porin, thicken the membrane.

Worst of all, these tricks travel. Resistance genes sit on mobile plasmids and transposons that bacteria swap by horizontal gene transfer — conjugation, transformation, transduction — even across species. A harmless gut commensal can hand a resistance plasmid to a pathogen overnight. That is why resistance can spread faster than it evolves, and why the WHO estimates roughly 1.27 million deaths in 2019 were directly attributable to antibiotic-resistant infections. The arms race is the reason we are told to finish the full course and not demand antibiotics for viral colds: every needless exposure is another round of selection.

Why this matters

• Medicine. Choosing the right drug is choosing the right target for the right bug at the right site.
• Evolution made visible. Resistance is the clearest real-time demonstration of natural selection we have.
• Drug discovery. New antibiotics mean finding new bacterial-only targets — a hard, slow search.
• Public health. Stewardship slows the loss of the targets we already exploit.
• Cell biology. Antibiotics are precision probes that taught us how the wall, ribosome, and gyrase actually work.