The technical challenge was considerable. Clathrin-coated pits are tiny, dome-shaped invaginations of the plasma membrane that mature into vesicles roughly 60 to 120 nanometres in diameter, involving more than 50 proteins recruited in a tightly choreographed sequence of initiation, maturation and membrane fission. Earlier attempts to detect lipid sorting relied on indirect evidence, such as tracking fluorescent lipid analogues or lipid-binding toxins like cholera toxin to endosomes. Both approaches carry serious caveats. Fluorescent analogues carry bulky chromophores that alter the chemical behaviour of the parent lipid, while toxins such as cholera toxin actively cluster their lipid targets, distorting the very sorting mechanisms they are meant to reveal. Moreover, measuring lipid uptake at the level of endosomes conflates sorting at the plasma membrane with sorting inside the endosomal system itself.

To sidestep these problems, the team turned to bifunctional lipid probes, a class of near-native lipids modified with a diazirine group that can be photochemically cross-linked to neighbouring biomolecules, and an alkyne handle used for fluorescent labelling by click chemistry. The combined effect of these modifications on lipid behaviour is approximately equivalent to adding a single extra double bond, meaning the probes closely mimic the metabolism and biophysical properties of their natural counterparts. The researchers loaded the probes into the outer leaflet of the plasma membrane of living cells, then captured snapshots by firing a high-power 365-nanometre LED pulse that cross-linked the lipids in place, followed immediately by chemical fixation. The entire cross-linking and fixation sequence was completed in under ten seconds, fast enough to freeze clathrin-coated pits at different stages of maturation without disturbing major membrane trafficking pathways.

Imaging was performed in a well-characterized SK-MEL2 melanoma cell line expressing dynamin-2 fused to green fluorescent protein, with clathrin and adaptor protein 2 (AP2) revealed by immunostaining. Because clathrin and the lipid channel were imaged at super-resolution while AP2 and dynamin-2 were acquired in confocal mode, the team could classify each pit by its maturation stage: clathrin-only structures, early-stage pits positive for AP2 and clathrin, and late-stage pits that additionally recruit dynamin-2, the protein that catalyses the final membrane scission. For each pit, the mean lipid fluorescence inside the segmented clathrin mask was divided by the mean signal of the surrounding membrane, yielding an apparent partitioning value. For the phosphatidylcholine probe PC(16:0|Y), this apparent partitioning rose from 1.64 in early pits to 2.09 in late pits, a trend that at first glance suggested lipids accumulate strongly inside forming vesicles.

That apparent enrichment, however, turned out to be largely an optical illusion. As a pit invaginates, the curved membrane folds a large surface area into a small volume, and even with STED resolution the microscope’s point spread function blurs this geometry, artificially concentrating the signal at the site of the invagination. To correct for this, the team built three-dimensional models of pit shapes based on previously published electron microscopy data from the same cell line, combining partial orbs for the pit dome with catenoids for the neck. Assuming an even distribution of point emitters across these modelled membranes, they simulated STED images using the experimentally measured point spread function. The model predicted apparent partitioning values of 1.82 for early pits and 2.10 for late pits, almost identical to the experimental measurements for PC(16:0|Y). By normalizing the experimental distributions to the model, the researchers converted convoluted apparent values into true absolute partitioning coefficients, a key methodological advance for studying lipids in curved membrane compartments.

Armed with this corrected framework, the team screened a library of ten bifunctional lipids spanning a broad chemical space, including glucosylceramide, sphingomyelin, phosphatidylethanolamine, a plasmalogen phosphatidylcholine, and phosphatidylcholine species differing in chain length, saturation and fatty acid positioning. The true partitioning values for late-stage pits ranged from 1.07-fold enrichment to 0.84-fold exclusion, differences of up to 21 percent between species. Although subtle, these differences were highly significant in two-sided randomized permutation tests. Saturated, long-chain lipids such as PC(18:0) tended to be modestly enriched in pits, while polyunsaturated species such as PC(20:4) were slightly excluded. The ordering placed PC(18:0|Y) at the top and PC(20:4|Y) at the bottom, with sphingomyelin and the unsaturated PC species in between.

The researchers then tested which of three candidate mechanisms could explain the pattern. Lipid shape, or resting curvature, predicts that cone- and inverse-cone-shaped lipids should prefer curved membranes, but resting curvature did not correlate with pit partitioning, ruling this mechanism out. Phase separation predicts that curved membranes should accumulate liquid-disordered, low-melting-temperature lipids because they bend more easily, yet the team observed the opposite: a strong positive correlation between melting temperature and pit partitioning, with high-melting-temperature lipids enriched in pits. The explanation that fit best was lipid asymmetry. The plasma membrane maintains an asymmetric distribution of lipids between its outer and inner leaflets, and because the curved pit changes the relative surface areas of the two leaflets, lipids enriched in the cytoplasmic leaflet should be partially enriched in the pit while exoplasmic-leaflet lipids should be partially excluded. This prediction held for nearly all the glycerophospholipids tested, with sphingomyelin the notable outlier, hinting at additional mechanisms such as specific lipid-protein interactions.

The team also probed whether the density of transmembrane cargo receptors could drive lipid sorting by measuring lipid and receptor partitioning simultaneously in three-colour STED images. Correlations between lipid and receptor enrichment across individual late-stage pits were negligible to moderate, ranging from 0.041 to 0.343 by Spearman’s rank correlation, indicating that cargo content largely does not explain the observed lipid differences. Attempts to perturb lipid asymmetry directly revealed just how tightly it is woven into the endocytic machinery. When the researchers triggered calcium-dependent lipid scrambling via the scramblase TMEM16F with ionomycin in SK-MEL2 cells, clathrin was rapidly released from the plasma membrane, dynamin-2 dissociated within three minutes, and transferrin uptake was fully blocked. Pits disassembled roughly four times faster than phosphatidylserine was externalized, and because HEK293 cells with low scramblase activity showed no such response, the effects could be attributed to scrambling rather than calcium itself. Lipid asymmetry, in other words, is not merely a correlate of endocytosis but a prerequisite for it.

Finally, pulse-chase experiments tracking lipid delivery to early endosomes, marked by EEA1, showed that uptake rate constants were essentially identical across all ten lipid species, averaging 1.09 per minute, with similar amplitudes for all but glucosylceramide and phosphatidylethanolamine, whose deviations likely reflect competing non-vesicular transport routes. The overall picture that emerges is strikingly simple: clathrin-mediated endocytosis is a selective protein transporter but a largely non-selective lipid transporter, with vesicle lipid composition essentially a reflection of the plasma membrane and its inherent transbilayer asymmetry. Since previous work from the same groups showed that more than 80 percent of retrograde lipid traffic occurs through non-vesicular routes, the modest selectivity of endocytosis is unlikely to meaningfully shape plasma membrane composition, though it may gently bias organelles toward plasma-membrane-like lipid profiles. The authors note that their analysis deliberately excluded PI(4,5)P2 and phosphatidylserine, established regulators of the endocytic machinery, so selective handling of these lipids cannot be excluded. The quantitative toolkit they have built, pairing near-native probes, super-resolution imaging and geometry-aware modelling, now offers cell biologists a way to measure lipid partitioning in any curved membrane compartment, promising new insight into how the endomembrane system maintains its remarkable compositional identity.

Subject of Research: Quantification of lipid sorting into clathrin-coated pits during clathrin-mediated endocytosis

Article Title: Quantification of lipid sorting during clathrin-mediated endocytosis

Article References: Lennartz, H. M., da Costa Nunes, S., Böhlig, K., Chhatre, A., Šachl, R., Kuhn, S. M., Moneta, L., Barahtjan, P., Yee Yau, W., Iglesias-Artola, J. M., Hof, M., Opálka, L., Modes, C. D., Honigmann, A., & Nadler, A. (2026). Quantification of lipid sorting during clathrin-mediated endocytosis. Nature Cell Biology. https://doi.org/10.1038/s41556-026-02068-7

Image Credits: AI Generated

DOI: 10.1038/s41556-026-02068-7

Keywords: clathrin-mediated endocytosis, lipid sorting, plasma membrane, STED microscopy, bifunctional lipid probes, lipid asymmetry, clathrin-coated pits, membrane trafficking, phosphatidylcholine, sphingomyelin, super-resolution imaging, cell biology