Antibiotics: four targets, and everything else follows
Why a drug that kills bacteria does not kill you.
The antibiotic chapter of any pharmacology textbook looks like a memorisation problem: dozens of drugs, a dozen classes, and a spray of adverse effects that seem unrelated to anything. Taught that way it is nearly unlearnable, and it does not survive first contact with an exam question that gives you a patient rather than a drug name.
There is a much smaller idea underneath. An antibiotic must kill a bacterium while leaving the human alone, so every antibiotic attacks a structure that bacteria have and we do not. There are only four such structures worth knowing. Learn those four, and the classes organise themselves — and, more usefully, the adverse effects stop being arbitrary. They occur precisely where the bacterial target resembles something of ours.
🩺 Where this lives: The reason aminoglycosides damage hearing and kidneys, while penicillins essentially never do, is not bad luck. Penicillin attacks the peptidoglycan cell wall — a structure with no human counterpart at all, which is why it is one of the safest drugs in medicine and why its main danger is allergy rather than toxicity. Aminoglycosides attack the bacterial ribosome, and our mitochondria carry ribosomes of bacterial ancestry. Selective toxicity is a spectrum, and the gaps in it are where the monitoring requirements live.
The four targets
This is the whole organising idea. If you can place a drug on this figure you can usually reconstruct its spectrum, its resistance mechanism and its characteristic toxicity — because all three follow from what it binds. Memorising the classes as a flat list gives you none of that.
Cell wall agents — the beta-lactams
Peptidoglycan gives a bacterium its shape and resists osmotic pressure. Beta-lactams bind the penicillin-binding proteins that cross-link it, so the wall cannot be maintained and the cell lyses. Since human cells have no wall at all, the selectivity is close to perfect.
THE BETA-LACTAM FAMILY — broadening spectrum
NARROW BROAD
├─ Benzylpenicillin streptococci, syphilis, meningococcus
├─ Flucloxacillin STAPHYLOCOCCI (beta-lactamase stable)
├─ Amoxicillin + some Gram-negatives
├─ Co-amoxiclav + beta-lactamase producers, anaerobes
├─ Piperacillin-tazo + Pseudomonas
└─ CARBAPENEMS very broad; reserve agents
(meropenem) — a stewardship decision, not a default
CEPHALOSPORINS by generation
1st cefalexin mostly Gram-positive
2nd cefuroxime more Gram-negative
3rd ceftriaxone good CSF penetration → MENINGITIS
4th cefepime + Pseudomonas
5th ceftaroline + MRSA (the exception to the rule
that beta-lactams miss MRSA)
💡 Exam angle: two facts here are asked repeatedly. Flucloxacillin for staphylococcal infections, because it resists staphylococcal beta-lactamase where amoxicillin does not. And ceftriaxone for bacterial meningitis, because third-generation cephalosporins cross the blood–brain barrier well — a question about a drug reaching the CSF is nearly always testing this.
Protein synthesis inhibitors
The bacterial ribosome is 70S, assembled from 30S and 50S subunits; ours is 80S from 40S and 60S. That difference is the selectivity, and which subunit a drug binds is the standard exam question.
"BUY AT 30, CCEL AT 50"
30S subunit 50S subunit
A Aminoglycosides C Clindamycin
T Tetracyclines C Chloramphenicol
E Erythromycin (macrolides)
L Linezolid
WHY THE SUBUNIT MATTERS BEYOND THE EXAM
Aminoglycosides are BACTERICIDAL and need oxygen-dependent
uptake — so they are inactive against anaerobes, which is a
spectrum gap you can derive rather than memorise.
Macrolides and clindamycin bind overlapping 50S sites, so
resistance to one can confer resistance to the other.
Bactericidal or bacteriostatic?
Students often over-weight this distinction. In an immunocompetent patient with an ordinary infection it rarely decides therapy, because a bacteriostatic drug plus a working immune system clears the infection perfectly well. It matters when the immune system cannot finish the job — neutropenia — or where host defences barely reach, as in endocarditis vegetations and the CSF.
Resistance
💡 Exam angle: MRSA is resistant through an altered penicillin-binding protein, not through beta-lactamase. This is why adding a beta-lactamase inhibitor such as clavulanate does nothing for it — there is no enzyme to inhibit; the target itself has changed. That is the reasoning being tested when a question offers co-amoxiclav for MRSA. The answer is an agent with a different target: vancomycin, or linezolid.
Adverse effects worth knowing by class
Situation
Avoid
Because
Pregnancy
Tetracyclines, fluoroquinolones, aminoglycosides
Tooth and bone effects; cartilage concerns; fetal ototoxicity. Penicillins and cephalosporins are the usual safe choices.
Children under 8
Tetracyclines
Permanent tooth staining.
Renal impairment
Aminoglycosides (or dose-adjust and monitor)
Renally cleared and directly nephrotoxic — the two compound.
Myasthenia gravis
Aminoglycosides
They impair neuromuscular transmission and can precipitate a crisis.
EMPIRICAL STARTING POINTS — organism first, then drug
Community pneumonia Strep pneumoniae, atypicals
→ amoxicillin ± a macrolide
Bacterial meningitis Strep pneumoniae, N. meningitidis
→ ceftriaxone (crosses into CSF)
Uncomplicated UTI E. coli
→ nitrofurantoin or trimethoprim
Cellulitis Strep pyogenes, Staph aureus
→ flucloxacillin
Intra-abdominal mixed, including ANAEROBES
→ cover anaerobes (metronidazole
or co-amoxiclav / pip-tazo)
Suspected MRSA → vancomycin or linezolid
Local guidelines and resistance patterns override all of
the above. These are the reasoning defaults, not a protocol.
Clinical reasoning: four presentations
🔍 Case 1 — spectrum reasoning
PresentationA 40-year-old with a spreading, hot, tender erythematous leg after a minor cut. No penicillin allergy.
ReasoningCellulitis from a skin breach means Staph aureus and Strep pyogenes. Amoxicillin does not reliably cover staphylococci, which usually produce beta-lactamase.
AnswerFlucloxacillin — beta-lactamase stable and specifically anti-staphylococcal. The question is answered by naming the organism first, not by picking the broadest drug.
🔍 Case 2 — the allergy question
PresentationA patient needs treatment for community-acquired pneumonia. They report that penicillin caused lip swelling and difficulty breathing.
Key clueThat history describes anaphylaxis, not a rash or nausea — a genuine type-1 hypersensitivity.
ReasoningAll beta-lactams share the ring, and although cephalosporin cross-reactivity is lower than once believed, it is not zero. After true anaphylaxis, avoid the whole class rather than substituting within it.
AnswerUse a structurally unrelated class — a macrolide covers the likely organisms here. Note also that most reported "penicillin allergy" is not allergy at all, and mislabelling drives patients onto worse, broader agents for life; the history is worth taking properly.
🔍 Case 3 — the resistance trap
PresentationA post-operative wound grows MRSA. The team proposes co-amoxiclav on the grounds that clavulanate "covers resistance".
TrapTreating all beta-lactam resistance as if it were enzymatic.
ReasoningMRSA resistance comes from an altered penicillin-binding protein, so the drug no longer binds its target. There is no enzyme for clavulanate to inhibit. Adding it changes nothing.
AnswerAn agent with a different target — vancomycin, or linezolid. The mechanism of resistance dictates the answer.
🔍 Case 4 — the interaction
PresentationA 74-year-old on warfarin, stable INR 2.4 for a year, is treated for a chest infection with clarithromycin. Six days later the INR is 6.8 with gum bleeding.
Key clueNothing else changed. The new drug is the variable.
ReasoningMacrolides inhibit CYP450, reducing warfarin metabolism and raising its effect. Metronidazole and co-trimoxazole do the same.
AnswerManage the over-anticoagulation per local protocol, and in future either choose a non-interacting antibiotic or monitor the INR closely. This is a predictable interaction, not an idiosyncratic reaction.
Staphylococcal infection needs flucloxacillin, not amoxicillin.
Beta-lactamase vs altered PBP
Enzyme destroys the drug; altered PBP changes the target
Clavulanate helps the first and is useless against MRSA.
30S vs 50S binders
Aminoglycosides and tetracyclines at 30S; macrolides, clindamycin, chloramphenicol, linezolid at 50S
A standard exam question, and it predicts cross-resistance.
Bactericidal vs bacteriostatic
Matters mainly in neutropenia, endocarditis and meningitis
Over-applying the distinction leads to unnecessary drug choices.
True allergy vs intolerance
Anaphylaxis and urticaria differ from nausea or diarrhoea
Mislabelling pushes patients onto broader, worse agents for life.
Broad spectrum vs better
Broad means more collateral damage and more resistance
The right answer is usually the narrowest drug that covers the likely organism.
Rapid revision
MUST-KNOW FACTS
1. Four targets: cell wall, protein synthesis, DNA, folate.
2. Selective toxicity — each target is something bacteria have and we do not.
3. Beta-lactams inhibit cell wall cross-linking → bactericidal.
4. Flucloxacillin for staphylococci (beta-lactamase stable).
5. Ceftriaxone for meningitis — crosses into the CSF.
6. Carbapenems are reserve agents, not a default.
7. Vancomycin: cell wall, Gram-positive only, for MRSA.
8. 30S — Aminoglycosides, Tetracyclines. 50S — Clindamycin,
Chloramphenicol, Erythromycin, Linezolid ("buy AT 30, CCEL at 50").
9. Aminoglycosides are inactive against anaerobes (oxygen-dependent uptake).
10. Aminoglycosides: nephrotoxic and ototoxic — monitor levels.
11. Macrolides: QT prolongation and CYP450 inhibition.
12. Tetracyclines: avoid in pregnancy and under 8 — teeth and bone.
13. Fluoroquinolones: tendon rupture, QT, avoid in children.
14. Metronidazole: anaerobes; disulfiram-like reaction with alcohol.
15. MRSA = altered PBP, so clavulanate does NOT help.
16. Beta-lactamase = enzyme, so clavulanate DOES help.
17. Cephalosporins and clindamycin are classic C. difficile risks.
18. Bactericidal matters in neutropenia, endocarditis, meningitis.
19. Take cultures — but never delay antibiotics in sepsis to obtain them.
20. Pregnancy: penicillins and cephalosporins are the usual safe choices.
21. De-escalate when cultures return; set a stop date.
22. Narrowest effective agent, always — resistance is selected by exposure.
💡 Exam angle: the reliable threads are (a) matching organism to drug rather than reaching for the broadest agent, (b) the MRSA mechanism defeating clavulanate, (c) which subunit each protein-synthesis inhibitor binds, (d) contraindications in pregnancy and childhood, and (e) CYP450 interactions with warfarin. Every one is derivable from the four-target figure plus the adverse-effect table — which is a much smaller thing to carry into an exam than a list of forty drugs.