A COPI cargo that sits at the ER
The study starts from a proteomic screen rather than from lipid droplets directly. HeLa cells expressing εCOPI-GFP, a subunit of the COPI coat complex that cycles between Golgi and ER, were immunoprecipitated and analysed by mass spectrometry. That produced 473 high-confidence binding partners (log2 fold change ≥ 2.5, p ≤ 0.005), of which 105 carry canonical COPI-binding motifs; filtering for links to mammalian growth narrowed the list to 22 proteins, and TMEM263 was among them.
TMEM263 (also called C12orf23) is a vertebrate-specific 12 kDa protein with a poorly characterised function. The paper confirms it behaves as a genuine COPI cargo: EGFP-TMEM263 co-localises with the ER marker Sec61β and with Golgi and COPI subunits, and its C-terminal five amino acids encode the canonical KxKxx motif (KDKSD). Mutating the lysines at positions 112 and 114 abolished COPI binding, whereas mutating the upstream lysines K110 and K111 did not. This matters because it establishes TMEM263 as an ER-localised membrane protein before any lipid droplet claim is made.
Overexpression enlarges and clusters lipid droplets
When EGFP-TMEM263 was overexpressed, roughly 80% of transfected HeLa cells showed EGFP-positive puncta, and those puncta were almost exclusively labelled by LipidTOX, a neutral lipid dye used here as a mature lipid droplet marker. Overexpression significantly increased LipidTOX-positive area in both vehicle (BSA) and oleic-acid-fed cells. In lipid-fed conditions the increase came from larger droplets — mean size rose from 0.15 μm² in controls to 0.28 μm² — not from more droplets.
The distribution also changed. More than 50% of analysed oleic-acid-fed cells showed a clustered, 'bunch-of-grapes' arrangement of droplets rather than a dispersed one. The structures behaved like canonical lipid droplets: PLIN2, LPCAT1 and LPCAT2 localised around the LipidTOX-positive core, and the same accumulation and clustering appeared in mouse embryonic fibroblasts and RPE-1 cells. Because a large EGFP tag could in principle distort localisation, the authors repeated the key observations with FLAG-HA-TMEM263 and with endogenous TMEM263 staining, finding one to three TMEM263 puncta per LipidTOX-positive droplet and occasional rings around the core.
A two-helix hairpin anchors TMEM263 and binds lipid
TMEM263 is predicted to have two transmembrane domains, at amino acids 38–60 and 80–102. AlphaFold2 models and atomistic molecular dynamics simulations describe these helices folding into a hairpin, with basic residues (R37, K47, K84, K104) positioned near phosphate headgroups and W59 sitting at the start of the inter-transmembrane linker. Truncation experiments showed the transmembrane domains are required for ER localisation and for supporting lipid droplet accumulation.
The simulations also pointed to lipid-facing residues, which the authors tested by mutation. F43A was designed to perturb neutral lipid binding on the first transmembrane helix, W59A to disturb the bulky linker residue, and K84A to destabilise contact between the bottom of the second helix and lipid headgroups. All three reduced the percentage of cells with EGFP-positive puncta relative to wild type. W59A mislocalised to the cytoplasm and was dropped from later analysis; F43A and K84A stayed at the ER but failed to raise LipidTOX-positive area above untransfected controls in either BSA or oleic-acid conditions, and neither reproduced the shift from dispersed to clustered droplets. The authors interpret this as evidence that TMEM263 binds neutral lipids in the ER bilayer and may help form the lipid lens from which a droplet grows — a proposed mechanism, not a directly visualised one.
TMEM263 marks nascent sites before neutral lipid appears
Localisation at the ER does not by itself show that TMEM263 acts early. The authors addressed timing with live imaging of cells expressing EGFP-TMEM263 and mCherry-HPos, a marker of early lipid droplet formation, plus LipidTOX, following oleic acid addition with frames every 10 seconds for 18 minutes. They scored when each signal appeared and report that TMEM263 labels nascent ER sites before those sites acquire HPos or neutral lipid staining.
This ordering is the basis for placing TMEM263 at the early stages of biogenesis rather than at the surface of mature droplets. The paper is explicit about what it did not find: there was no evidence that TMEM263 co-localises with seipin, the established biogenesis regulator, and loss of TMEM263 did not phenocopy seipin knockout — no supersized droplets in cells or in vivo. Because seipin is evolutionarily conserved while TMEM263 is vertebrate-specific, the authors argue the two did not co-evolve and leave their coordination an open question.
Loss of TMEM263 reduces droplet number, not size
The necessity side comes from CRISPR/Cas9 knockout HeLa lines and CRISPR mutant zebrafish. Two clonal knockout HeLa lines had significantly lower BODIPY-positive area than parental controls after oleic acid feeding, but variation in droplet size and number was high, probably because residual protein remained. The authors therefore added shRNA against TMEM263 to silence leftover transcript; in these KO1/2 cells, loss of TMEM263 reduced both the percentage of LipidTOX-positive area and droplet number in oleic-acid-fed conditions, while droplet size was unaffected.
That pattern is worth separating from the overexpression result. Overexpression increased droplet area by enlarging droplets without changing their number; loss reduced droplet number without changing size. Both point to TMEM263 acting on how many droplets form or how efficiently they accumulate lipid, rather than on the machinery that sets the size of an existing droplet.
Zebrafish loss links droplets to body growth
In zebrafish, three gRNAs targeting exon 3 of tmem263 were injected as ribonucleoprotein complexes into one-cell embryos. Most crispant fish — 77.5% — showed abnormal phenotypes including oedema, bent tails and tail truncations, and did not survive past 5 days post fertilisation. The surviving 22.5% had a small but significant decrease in body length at 5 dpf, and their intestinal lipid droplet accumulation after oleic acid feeding was significantly decreased, counted as BODIPY puncta in a 68 μm² region of interest per larva.
The authors note that because only the less severely affected 22.5% could be assayed, these measurements likely under-represent the growth and lipid droplet phenotype. The in vivo result is therefore a directionally consistent but probably conservative estimate. It supports the claim that TMEM263 is necessary for efficient lipid droplet formation in vivo and connects that function to organismal growth and proportional dwarfism.
Where the mechanistic claim still stops
The strongest evidence in this paper is localisation, sufficiency and necessity: TMEM263 sits at the ER, its overexpression enlarges and clusters droplets, and its loss reduces droplet number in cells and droplet accumulation in zebrafish. The lipid-condensation step is supported indirectly — by molecular dynamics showing neutral lipid affinity and by the F43A and K84A mutations failing to support droplet accumulation — rather than by watching a lens form. The authors themselves frame the protein-protein interactions of TMEM263 as the next question, including whether neutral lipid affinity is its primary role or whether it also recruits enzymes to early lenses.
A second boundary is tissue scope. The growth defects were measured in whole zebrafish larvae, and the paper notes that TMEM263 appears ubiquitously expressed in mouse and human tissue, so whether the body-wide growth phenotype reflects cell- or tissue-specific roles remains unresolved. The ring-like TMEM263 signal around mature droplets is likewise interpreted as possible tethering of TMEM263-enriched ER, not as TMEM263 moving onto the droplet monolayer.
Why droplet number is the readout to watch
For readers comparing this work with other lipid droplet studies, the most decision-relevant split is between droplet number and droplet size. TMEM263 overexpression raised LipidTOX-positive area through size (0.15 to 0.28 μm²) without changing number, whereas loss reduced number without changing size. A study that reports only total droplet area, or only size, would blur which of these two levers TMEM263 pulls.
The second split is between the proposed lipid-condensation mechanism and the seipin pathway. The absence of seipin co-localisation and of supersized droplets after TMEM263 loss means TMEM263 should not be folded into seipin-type biogenesis models on current evidence. The paper's distinctive contribution is a vertebrate-specific ER protein that ties neutral lipid handling to whole-body growth, and the priority for follow-up is to test whether its lipid affinity is the primary function or one part of a broader recruitment role at nascent droplets.
Sources used in this answer
Identification of TMEM263 as an ER-resident protein that controls early stages of lipid droplet biogenesis
TMEM263 is an ER-resident COPI cargo whose two transmembrane helices form a dynamic alpha hairpin; it binds neutral lipids, is sufficient to enlarge and cluster lipid droplets, and is required for efficient droplet accumulation in cells and zebrafish, where its loss is associated with growth defects and proportional dwarfism.
