Antibiotic Resistance: Mechanisms and Spread

Antibiotic resistance is the heritable capacity of a microorganism to survive concentrations of an antibiotic that would inhibit or kill related, susceptible strains. It must be distinguished from tolerance (surviving transiently without growing, often via dormancy) and from innate insusceptibility (a species naturally lacking the drug's target, such as the intrinsic resistance of Gram-negative bacteria's outer membrane to vancomycin). True acquired resistance requires a genetic change — either a chromosomal mutation or the acquisition of a foreign resistance gene — that is then transmitted to daughter cells or to other bacteria. The molecular mechanisms cluster into four functional categories, and it is useful to memorize them by what step of drug action they interrupt: destruction of the antibiotic molecule (e.g., β-lactamases hydrolyzing penicillins), modification of the target so the drug can no longer bind (e.g., altered penicillin-binding proteins, ribosomal methylation), active efflux of the drug out of the cell before it can act (multidrug efflux pumps), and reduced influx through altered porins or membrane composition that keeps the drug from ever reaching its target. Many resistant strains combine two or more of these mechanisms simultaneously, producing very high-level, multidrug resistance. Superimposed on this molecular biology is a population-genetics principle that is easy to state and easy to misunderstand: resistance-conferring mutations and genes exist in a population before any antibiotic is applied, at low frequency, as products of ordinary replication error, transposition, or horizontal acquisition. The antibiotic itself does not induce these changes directionally — it changes the fitness landscape so that resistant cells, previously neutral or even slightly disadvantaged, now vastly outcompete susceptible cells. This is natural selection operating on standing genetic variation, and it is formally identical to selection processes discussed under Mutation and Genetic Variation in Bacteria. Once a resistant clone appears, Horizontal Gene Transfer mechanisms can move its resistance determinants into other cells and species, converting a local, single-lineage event into a regional or global phenomenon.

Step 1 — Enzymatic inactivation. The most clinically important example is the β-lactamase family. These periplasmic (Gram-negative) or secreted (Gram-positive) enzymes hydrolyze the β-lactam ring shared by penicillins, cephalosporins, and carbapenems, the very structural feature that lets these drugs mimic the D-Ala-D-Ala terminus of peptidoglycan precursors (see Antibiotics and Their Mechanisms of Action). Hydrolysis opens the four-membered ring, destroying the drug's ability to acylate the active-site serine of penicillin-binding proteins (PBPs). Extended-spectrum β-lactamases (ESBLs) and carbapenemases such as Klebsiella pneumoniae carbapenemase (KPC) and New Delhi metallo-β-lactamase (NDM-1) have broadened this mechanism to defeat nearly every β-lactam in clinical use. Aminoglycoside-modifying enzymes (acetyltransferases, phosphotransferases, adenylyltransferases) work analogously, chemically tagging the drug so it can no longer bind the 30S ribosomal subunit.

Step 2 — Target modification. Bacteria can alter the antibiotic's binding site just enough to block drug binding while preserving the target's normal function. Methicillin-resistant Staphylococcus aureus (MRSA) expresses an additional penicillin-binding protein, PBP2a, encoded by the mecA gene, which has drastically lower affinity for β-lactams but still catalyzes peptidoglycan cross-linking. Resistance to macrolides and lincosamides frequently arises from erm-gene-encoded methylases that dimethylate a specific adenine (A2058, E. coli numbering) in 23S rRNA, sterically blocking drug binding to the peptidyl transferase center. Fluoroquinolone resistance often comes from point mutations in DNA gyrase (gyrA) or topoisomerase IV (parC) in the "quinolone resistance-determining region," changing target conformation without abolishing supercoiling activity.

Step 3 — Efflux pumps. Multidrug efflux pumps, such as the resistance-nodulation-division (RND) family (e.g., AcrAB-TolC in E. coli), span the inner membrane, periplasm, and outer membrane as a tripartite complex, using the proton-motive force to expel structurally diverse antibiotics (tetracyclines, fluoroquinolones, chloramphenicol) directly into the extracellular space before they accumulate to inhibitory concentrations. Because a single pump can export many unrelated drug classes, efflux overexpression is a common route to multidrug resistance and can act synergistically with reduced permeability.

Step 4 — Reduced uptake / permeability. Gram-negative bacteria can downregulate or mutate outer-membrane porins (e.g., OmpF in E. coli) that antibiotics use to cross into the periplasm, or remodel lipopolysaccharide to reduce polymyxin binding (as in colistin-resistant mcr-1-carrying strains, where a phosphoethanolamine transferase adds a positively charged group to lipid A, reducing electrostatic attraction to the cationic drug).

The population-level consequence of these mechanisms is best visualized as a shift in the dose-response relationship. A susceptible population shows a sharp, sigmoidal kill curve as antibiotic concentration increases; a resistant population's curve shifts rightward, requiring far higher concentrations — often above what is clinically achievable or safely tolerable — to achieve the same kill.

Clinically, this rightward shift is quantified as the minimum inhibitory concentration (MIC) — the lowest antibiotic concentration that prevents visible growth in a standardized broth or agar dilution assay. A strain is called resistant when its MIC exceeds a clinical breakpoint set by organizations such as the Clinical and Laboratory Standards Institute (CLSI), based on achievable serum/tissue drug concentrations and correlation with treatment outcomes. Underlying all of this is the ecological logic of selection: in a mixed population exposed to a bactericidal antibiotic, susceptible cells die or fail to divide while resistant cells continue replicating unimpeded (or nearly so). Within a handful of generations — under strong selection, a resistant subpopulation representing 1 in 10⁷ cells can become the dominant clone in less than 24 hours of continuous drug exposure, because susceptible competitors are eliminated exponentially while resistant cells still double roughly every 20–60 minutes.

The foundational proof that resistance mutations arise before, not because of, antibiotic (or, in the original experiments, phage) exposure came from two classic experiments in bacterial genetics.

Luria and Delbrück's fluctuation test (1943). Salvador Luria and Max Delbrück asked whether bacteriophage resistance in E. coli arose spontaneously by mutation prior to phage exposure ("Darwinian," pre-adaptive hypothesis) or was induced by the phage itself in response to the challenge ("Lamarckian," adaptive/directed hypothesis). They inoculated many small parallel liquid cultures, each started from a small number of susceptible bacteria, and let them grow to comparable final population sizes. They then plated all cultures on medium containing phage and counted resistant survivor colonies. The key logic: if resistance is induced by phage exposure at a constant per-cell probability at the moment of plating, the number of resistant colonies per culture should follow a Poisson distribution — low variance, mean ≈ variance. If resistance arises spontaneously at random times during growth, before phage ever appears, then a mutation that happens early in one culture's history gets amplified by all subsequent cell divisions, while a culture where the same mutation happens late (or not at all) yields few or no resistant survivors. This produces enormous, non-Poisson variance between replicate cultures — a "jackpot" distribution, with some cultures showing hundreds of resistant colonies and others showing zero. Luria and Delbrück observed exactly this: variance in resistant colony counts far exceeded the mean, consistent only with pre-existing, spontaneously arising mutations amplified by clonal expansion, not with induction by the selective agent. This result — for which Luria and Delbrück shared the 1969 Nobel Prize in Physiology or Medicine (with Alfred Hershey) — established the mathematical signature of random, pre-adaptive mutation and is the founding result of bacterial population genetics.

Lederberg and Lederberg's replica plating (1952). Joshua and Esther Lederberg devised a direct, visual confirmation. A master plate of bacteria grown without any antibiotic or phage was pressed with a velveteen-covered stamp, transferring an identical spatial pattern of colonies onto several replica plates. Some replica plates contained antibiotic (or phage); the master plate never had any exposure at all. Resistant colonies appeared on the replica plates in the same positions corresponding to specific colonies on the untouched master plate. Because those specific master-plate colonies had never encountered the antibiotic, their resistance could not have been induced by it — it must have been present already, before any selective exposure occurred. Picking those exact colonies from the master plate and regrowing them confirmed they were resistant. This is the single most direct, visually intuitive demonstration that mutations conferring resistance pre-exist selection; it converts the statistical argument of the fluctuation test into a spatial, colony-by-colony proof. Together, these experiments established what is sometimes called the pre-existence principle: antibiotics (and other selective agents) do not instruct bacteria to become resistant. They act purely as a filter on variation that already exists in the population, generated by ordinary replication error, transposition, or (as later work on plasmids and conjugation would show) horizontal gene acquisition.

A modern extension — the MEGA-plate experiment (Baym et al., 2016). A vivid contemporary demonstration used a meter-long agar plate divided into bands of increasing antibiotic concentration (up to 1000× the initial MIC), filmed as E. coli spread across it over about 11 days. Bacteria repeatedly stalled at each concentration boundary until a resistant mutant lineage arose and swept across that band, visibly branching the population's migration front. Whole-genome sequencing of isolates from successive bands showed the accumulation of sequential resistance mutations, each new lineage descending from — and often less fit in drug-free zones than — its predecessor, directly visualizing both the stepwise, mutation-driven acquisition of resistance and the fitness trade-offs frequently involved.

Example 1 — Fluctuation test arithmetic. Suppose you set up 20 parallel 1 mL cultures, each inoculated with 10³ phage-susceptible E. coli cells, and let them grow to 10⁸ cells/mL (10⁵-fold expansion) before plating on phage-containing agar. If resistance were induced by phage exposure at a constant probability p = 2 × 10⁻⁷ per cell at the moment of plating, the expected number of resistant colonies per culture would be Poisson-distributed with mean = variance ≈ (10⁸)(2×10⁻⁷) = 20 colonies per plate, and nearly all 20 replicate cultures should show colony counts clustered tightly around 20 (standard deviation ≈ √20 ≈ 4.5). Instead, suppose the actual experimental spontaneous mutation rate is μ = 2 × 10⁻⁸ per cell per generation, and mutations can occur at any of the ~17 generations of growth (10³ → 10⁸ is roughly log₂(10⁵) ≈ 17 doublings). A mutation occurring at generation 2 gets amplified through 15 further doublings (2¹⁵ ≈ 33,000-fold), while one occurring at generation 16 is amplified only 2-fold. This means one culture might show 0 resistant colonies, another might show 4, and — because of a single early mutational event — another might show 800 or more ("jackpots"). Calculating the variance-to-mean ratio across the 20 cultures and finding it far greater than 1 is precisely the statistical test Luria and Delbrück used; observing it rules out induced, constant-probability resistance and supports spontaneous pre-existing mutation amplified by clonal growth.

Example 2 — Clinical reasoning about MIC and combination therapy. A urine culture from a patient with a urinary tract infection grows E. coli with the following MICs: ciprofloxacin MIC = 32 μg/mL (clinical breakpoint for susceptibility ≤1 μg/mL — this isolate is resistant, 32-fold above breakpoint), and nitrofurantoin MIC = 16 μg/mL (breakpoint for susceptibility ≤32 μg/mL — this isolate remains susceptible). Reasoning: the ciprofloxacin resistance is 32-fold above what defines susceptibility, consistent with accumulated gyrA/*parC* target-modification mutations (each contributing roughly 4–8-fold increases in MIC; two or three stacked mutations easily produce 32-fold resistance) rather than a single-step change — fluoroquinolone resistance is classically stepwise, matching the sequential-mutation pattern seen in the MEGA-plate experiment. Because nitrofurantoin achieves urinary concentrations of 200 μg/mL or more (far exceeding its 16 μg/mL MIC against this strain, roughly a 12-fold safety margin) but only modest serum concentrations, it is an appropriate choice specifically for uncomplicated cystitis (drug confined to the urinary tract) but would be inappropriate for pyelonephritis or bacteremia where systemic levels matter. This illustrates that "resistant" and "susceptible" are breakpoint-dependent, site-of-infection-dependent judgments, not fixed absolute properties of the organism.

Esther Lederberg (1922–2006) was a pioneering microbial geneticist whose experimental ingenuity is easy to underappreciate because her signature technique — replica plating — looks almost too simple to be profound. Working at the University of Wisconsin and later Stanford alongside her then-husband Joshua Lederberg, she developed the velveteen-stamp replica-plating method not merely as a labor-saving tool for transferring colonies, but as a conceptual instrument: it let researchers ask what a colony's genotype was before it was ever exposed to a selective agent, by keeping an untouched master copy. This same insight — preserve an unexposed reference population, then test replicas under selection — is a technique whose logic recurs throughout microbial genetics, including in phage genetics and in her own discovery of the lambda bacteriophage and the fertility factor F, both landmark findings in bacterial genetics that she made using the very methods she pioneered. Esther Lederberg's career also illustrates a less scientific but historically important pattern: her contributions were frequently under-credited relative to her male collaborators, including in the Nobel Prize recognition of the 1958 Physiology or Medicine award (shared by Joshua Lederberg, George Beadle, and Edward Tatum), despite her direct experimental role in the discoveries being honored.

Antibiotic resistance surveillance now drives frontline clinical decision-making through antibiograms — hospital-specific tables of local susceptibility patterns that guide empirical antibiotic choice before culture results return. Genomic surveillance networks track the global spread of specific resistance plasmids and clonal lineages (such as the globally disseminated E. coli sequence type ST131 carrying ESBL genes) in near real time, informing infection-control interventions. In agriculture, the historic use of antibiotics as growth promoters in livestock — a practice now banned or restricted in the EU, US, and many other jurisdictions — created enormous, continuous selection pressure across entire animal populations, and resistant zoonotic pathogens (e.g., resistant Salmonella, Campylobacter) have been traced directly to this practice, motivating "One Health" policy frameworks that treat human, animal, and environmental antibiotic use as a single interconnected selective system. In drug development, understanding specific resistance mechanisms has enabled rational counter-design: β-lactamase inhibitors (clavulanic acid, tazobactam, avibactam) are co-administered with β-lactams specifically to block the enzymatic destruction mechanism, restoring efficacy of older, cheaper antibiotics. Newer agents are designed to evade efflux pumps or to bind conserved targets less prone to single-step resistance mutations. In public health, stewardship programs — restricting unnecessary antibiotic prescribing, shortening treatment durations to the minimum effective course, and improving diagnostic speed so antibiotics are given only when a bacterial infection is confirmed — directly target the selection-pressure side of the equation, since reducing drug exposure reduces the intensity and duration of selection for resistant variants, exactly as predicted by the population-genetic logic Luria and Delbrück's data first revealed.

Ask the tutor: How do carbapenemase genes like KPC and NDM-1 actually spread between hospitals and countries so quickly?