Antibody-drug conjugate lifecycle — targeted delivery, receptor-mediated internalization, lysosomal linker cleavage and bystander cytotoxicity
An antibody-drug conjugate fuses the tumor-targeting precision of a monoclonal antibody with the sub-nanomolar potency of a cytotoxic small molecule that would be far too toxic to dose systemically on its own. Getting the three components — antibody, linker, payload — right simultaneously is the central engineering challenge of the modality.
The antibody backbone (usually IgG1) provides tumor specificity and a long circulating half-life via FcRn recycling. Conjugation sites are either native cysteines (interchain disulfides, giving a heterogeneous DAR 0–8 mixture) or engineered sites (THIOMAB, glycan remodeling, unnatural amino acids) that yield a homogeneous DAR.
DAR is a double-edged parameter: too low (DAR 1–2) under-delivers payload per binding event and requires more antigen turnover to reach a cytotoxic threshold; too high (DAR 6–8) increases plasma clearance because hydrophobic payload patches drive aggregation and premature hepatic uptake — DAR8 antibodies can clear 2–3× faster than DAR2. Kadcyla (T-DM1, DAR~3.5) and Enhertu (T-DXd, DAR~8 but with a hydrophilic linker that avoids the clearance penalty) sit at opposite ends of this design space.
Enhertu's breakthrough was not a new antibody or payload — it was a hydrophilic, self-immolative linker that let DAR8 conjugates keep antibody-like pharmacokinetics, something previously thought impossible.
Once infused, the intact ADC behaves pharmacokinetically like a normal antibody — it must survive days in circulation without releasing payload prematurely, then find and bind tumor cells that overexpress the target antigen manyfold above normal tissue.
ADC selectivity is not absolute specificity — it is a differential expression ratio. HER2, for instance, is overexpressed 10–100× on HER2+ breast tumor cells relative to normal epithelium, which still expresses low baseline HER2. The linker-payload system must therefore tolerate low-level binding to normal tissue without causing dose-limiting toxicity, while the high-avidity binding to tumor (many antigen copies, multivalent antibody engagement) drives efficient internalization there.
The "binding-site barrier" is a real limitation: high-affinity ADCs saturate the first few cell layers around a blood vessel and never penetrate deeper into a solid tumor. Some newer ADCs deliberately use moderate affinity to improve tumor penetration at the cost of maximal binding — a non-intuitive optimization only visible in whole-tumor simulation, not in a simple binding assay.
Antigen binding alone does nothing — the ADC must be actively internalized by the tumor cell to reach the proteolytic machinery that releases its payload. Not all tumor antigens internalize efficiently, which is one of the most important (and hardest to predict) filters in target selection.
Antigen crosslinking by bivalent antibody binding clusters receptors and nucleates clathrin-coated pit assembly at the plasma membrane. The vesicle buds inward (dynamin-mediated scission), sheds its clathrin coat, and fuses with early endosomes — Rab5-positive compartments that begin acidifying via V-ATPase proton pumps.
Maturation to late endosome (Rab7) and fusion with lysosomes takes roughly 30–90 minutes. Antigens with slow constitutive turnover (like some carbohydrate antigens) make poor ADC targets even with excellent binding affinity, because the conjugate simply sits at the surface and is more available for premature extracellular deconjugation or immune-mediated shedding rather than delivery to the degradative compartment.
The linker is the molecular trigger that decides where and how payload is liberated — and this single design choice governs potency, bystander killing capacity, and off-target toxicity more than almost any other ADC parameter.
Cleavable linkers (hydrazone, disulfide, or protease-sensitive dipeptides like Val-Cit-PABC) exploit the acidic, protease-rich lysosomal environment absent in plasma. Cathepsin B recognizes the Val-Cit sequence and cleaves it; a self-immolative para-aminobenzyl (PABC) spacer then spontaneously fragments, releasing the unmodified free payload — critical because most tubulin/DNA-targeting payloads lose potency if a linker fragment remains attached.
Non-cleavable linkers (thioether, e.g. T-DM1's SMCC-DM1) require complete proteolytic degradation of the entire antibody down to a single amino acid still bearing the payload (lysine-MCC-DM1). This metabolite is charged and cannot cross membranes — so it cannot escape the cell that made it, eliminating bystander killing but also essentially eliminating off-target payload leakage, giving non-cleavable ADCs a cleaner systemic toxicity profile at the cost of efficacy against heterogeneous, antigen-low tumors.
The choice between cleavable and non-cleavable linker is really a choice between maximizing efficacy in heterogeneous tumors (bystander effect) and maximizing safety margin in homogeneous, high-antigen tumors — there is no universally superior option.
The released payload is the business end of the ADC — typically an ultra-potent microtubule inhibitor (MMAE, DM1) or DNA-damaging agent (calicheamicin, exatecan derivatives) whose free-drug toxicity is too narrow a therapeutic window to dose systemically, but which becomes tractable when delivered as a targeted intracellular payload.
Inside the target cell, MMAE binds β-tubulin at the vinca-alkaloid site, blocking microtubule polymerization. The mitotic spindle cannot assemble, cells arrest at the G2/M checkpoint, and prolonged arrest triggers intrinsic (mitochondrial) apoptosis via BAX/BAK-mediated cytochrome c release.
Because MMAE is uncharged and membrane-permeable (unlike DM1's charged maleimidocaproyl-thioether metabolite), it diffuses out of the dying cell into the surrounding tumor microenvironment, killing neighboring cells regardless of whether they express the target antigen. This bystander effect is therapeutically valuable in tumors with heterogeneous antigen expression — a common feature of real clinical tumors — where an antigen-restricted payload would leave antigen-low subclones untouched and able to regrow.
Bystander killing is a double-edged sword: it improves efficacy against heterogeneous tumors but is also implicated in ADC-associated toxicity to antigen-negative bystander tissue near the tumor, such as ocular epithelium for MMAE-based ADCs.