Introduction

Following commitment to cell death, apoptotic cells initiate important communication with neighbouring cells to facilitate phagocyte recruitment for cell clearance (i.e. efferocytosis) as well as wound healing and anti-inflammatory responses1,2,3. This communication is mediated by the release of soluble ‘find-me’ and ‘good-bye’ signals, the exposure of ‘eat-me’ signals on the outer membrane leaflet, and the release of apoptotic cell-derived extracellular vesicles (ApoEVs) in particular, apoptotic bodies (ApoBDs)1,3,4,5,6,7. ApoBDs are a subset of large ApoEVs, typically 1–5 μm in diameter that can harbour biomolecules including DNA, RNA, and proteins, and pathogens8,9,10,11,12,13,14. Following interaction with recipient cells, ApoBDs can influence the local tissue microenvironment by promoting stem cell proliferation, antigen presentation, and propagation of viral infections2,8,10,12,15,16. In addition to interacting with cells at the site of cell death, ApoBDs can disperse through the circulation and thus may exert functional effects at distal sites12,17,18. Notably, the release of small EV subsets such as exosome-like ApoEVs by apoptotic cells has also been reported to aid processes such as cell clearance and tissue regeneration19,20,21. Given the broad importance of ApoEVs under both homeostatic and disease settings, understanding the molecular mechanisms responsible for ApoEV formation is vital.

The formation of ApoBDs is regulated by stepwise morphological changes during apoptosis, a process known as apoptotic cell disassembly22,23. During the early stages of apoptosis, caspase-activated Rho-associated kinase 1 (ROCK1) phosphorylates myosin light chain, leading to the actomyosin contraction necessary for apoptotic membrane blebbing24,25,26,27. Notably, for adherent cell types, this contractile force also drives the cell contraction necessary for cell detachment from the substratum28. After plasma membrane blebbing, the apoptotic cell can generate thin membrane protrusions called apoptopodia and beaded-apoptopodia that actively radiate from the apoptotic cells to facilitate cell fragmentation into multiple distinct ApoBDs22,29.

In this study, we describe a mechanism of generating large ApoEVs that mark the site of cell death via a cell-retraction dependent process. Upon induction of apoptosis, we showed that adherent cells retract and leave behind actin-rich membrane tracks resembling a cellular ‘footprint’, coined as the ‘FOotprint Of Death’ (FOOD), which subsequently round into large ApoEVs ( ~ 2 μm in diameter) denoted as FOOD-derived ApoEVs (F-ApoEVs). F-ApoEVs expose the ‘eat-me’ signal phosphatidylserine (PtdSer) and are cleared from the site of cell death by neighbouring phagocytes. In an infection setting, F-ApoEVs generated from influenza A virus (IAV)-infected apoptotic cells can aid viral propagation to neighbouring cells. Together, this study has identified an alternative mechanism of generating large EVs during apoptosis, which may have implications in efferocytosis and intercellular communication.

Results

Apoptotic cells generate a membranous footprint during cell retraction that aids the formation of large ApoEVs

When performing time-lapse confocal microscopy on apoptotic cells, we captured a distinct morphological process. Human A431 squamous epithelial cells treated with a BH3-mimetic cocktail (ABT-73730,31 and S6384532,33) to induce apoptosis exhibited characteristic apoptotic morphologies including cell rounding, membrane blebbing, and PtdSer exposure as indicated by annexin A5 staining (A5) (Fig. 1a). 3D confocal laser scanning microscopy (CLSM) revealed the formation of PtdSer-exposing membranous remnants that appeared during cell retraction, analogous to a ‘footprint’ of the cell marking the site of cell death (Fig. 1a, b; Supplementary Fig. 1a, b, Supplementary Video 1). This ‘footprint’ was only observed at the focal plane close to the cover glass (base) and did not extend beyond the original boundary of the cell, distinct from apoptotic protrusions (i.e. apoptopodia) that actively radiate from apoptotic cells22. Owing to this, we coined this phenomenon the ‘FOotprint Of Death’ or ‘FOOD’. The formation of FOOD was consistently observed in a variety of cell types including primary human umbilical vein endothelial cells (HUVECs), mouse embryonic fibroblasts (MEFs), and human cervical adenocarcinoma (HeLa) cells (Fig. 1c–e), with the majority ( ~ 80-99%) of apoptotic cells forming FOOD (Fig. 1f). Moreover, visualisation of FOOD by scanning electron microscopy (SEM), showed the ultrastructure as thin membrane surrounding the main cell body (Fig. 1c–e). In addition to the BH3-mimetic cocktail to specifically target the intrinsic apoptotic pathway30,32, FOOD was generated in response to several apoptotic stimuli including UV irradiation22,34, DNA damage inducing agents like etoposide35 and infection with IAV36 (Supplementary Figs. 2a–c, 3). Apoptosis induction under these conditions was validated by flow cytometry and immunoblot analysis of caspase cleaved proteins such as caspase 3 and Pannexin 1 (PANX1) membrane channels37,38 (Supplementary Fig. 3). Notably, MEFs that lack the pro-apoptotic proteins Bax and Bak (Bax−/−Bak−/−) did not generate FOOD following BH3-mimetic cocktail treatment as compared to WT MEFs (Supplementary Fig. 4), further indicating that FOOD is formed specifically during the progression of apoptosis. Extensive quantification of CLSM revealed the median number of ‘branches’ of membranous material adhered to the substrate, generated by MEF-derived FOOD, to be ~145, with the median branch thickness of ~1.5 µm, occupying an area of ~193.7 µm2 (Fig. 1g), suggesting that MEFs frequently form FOOD composed of many thin membrane structures.

Fig. 1: Apoptotic adherent cells generate a membranous ‘footprint of death’ during cell retraction.

a, b A431 epithelial cells were treated with a BH3-mimetic cocktail (2.5 µM ABT-737, 0.5 µM S63845) and imaged by time-lapse confocal laser scanning microscopy (CLSM). Cell membrane and nucleus were visualised by PHK26 (red) and Hoechst 33342 (blue) staining, respectively, and exposed phosphatidylserine (PtdSer) was examined using A5-FITC (green). Cells in (b) are representative of cells 180 min post apoptosis induction. Maximum intensity projection (MIP) shown (left) and Z-stack images of top, centre and base of cells shown (left to right). Images demonstrating morphological changes following apoptosis induction in (c) HUVEC (2.5 µM ABT-737, 0.5 µM S63845), d MEF (5 µM ABT-737, 10 µM S63845), and e HeLa cells (2.5 µM ABT-737, 0.5 µM S63845) following treatment with a BH3-mimetic cocktail. Representative CLSM i (n = 3) and scanning electron microscopy (SEM) ii (n = 2). f Quantification of the percentage of apoptotic A431, HUVEC, MEF, and HeLa cells that form FOOD during the progression of apoptosis. Data is pooled from (n = 3) independent experiments. Apoptotic cells were identified based on A5 staining and apoptotic morphologies (i.e. cell retraction and membrane blebbing). Error bars represent s.e.m. g Quantitative analysis of the FOotprint Of Death (FOOD) generated by MEFs: number of branches per cell, branch thickness (μm), and area of branches per cell (μm2). Data is pooled from (n = 3) independent experiments. h Representative time-lapse CLSM MIP of MEFs on uncoated, collagen, fibronectin, and fibronectin and collagen coated chamber slides, treated with a BH3-mimetic cocktail. i Representative CLSM MIP of MEF in a Cultrex BME 3D-matrix. Cell cytosol and nucleus were visualised by cell trace violet (blue) and NucRed staining, respectively, and PtdSer exposed was examined using A5-FITC. At least three independent experiments were performed for all experiments unless otherwise specified.

Extracellular matrix (ECM) protein coatings can facilitate cell attachment in vitro and mimic the in vivo cellular environment39. Therefore, we examined FOOD on surfaces coated with bovine neutralised type I collagen, human fibronectin, or fibronectin-enriched collagen. MEFs seeded on ECM proteins readily formed FOOD upon the induction of apoptosis (Fig. 1h). Similar findings were also observed for MEFs undergoing apoptosis within a 3D extracellular matrix (Fig. 1i), demonstrating that the formation of FOOD was not restricted to artificial surfaces such as plastic and glass cover slides. Taken together, these findings demonstrate that FOOD formation is a common cellular phenomenon that occurs during apoptosis, and that FOOD can form on surfaces that mimic physiological settings.

To gain a high-resolution temporal insight into FOOD formation, we performed lattice light sheet microscopy (LLSM) on MEFs undergoing apoptosis. LLSM imaging revealed that FOOD is first composed of several flat, sheet-like structures of membrane that gradually ‘round up’ into discrete vesicle-like structures following exposure of PtdSer on FOOD membranes (Fig. 2a, b, Supplementary Video 2). These vesicles, herein referred to as FOOD-derived ApoEVs (F-ApoEVs), were ~2 μm in diameter (Fig. 2c) with the median number of F-ApoEVs generated per cell within a period of 4 h following apoptosis induction being ~40 (Fig. 2d; Supplementary Fig. 5). Notably, F-ApoEV rounding also readily occurred following apoptosis induction of MEFs seeded on ECM proteins (Supplementary Fig. 6). Furthermore, F-ApoEVs generated from free-GFP expressing cells also harbour GFP in FOOD and F-ApoEVs (Fig. 2e), indicating that the membrane integrity of FOOD and F-ApoEVs is intact as membrane lysis would result in the release of free GFP. Together, this data demonstrates that the formation of FOOD provides a highly effective method of generating large, ApoEVs.

Fig. 2: Formation of large ApoEVs from FOOD.

a Representative maximum intensity projections (MIP) images from time-lapse lattice light sheet microscopy (LLSM) of FOOD-derived ApoEVs (F-ApoEV) formation in MEFs following treatment with a BH3-mimetic cocktail (5 μM ABT-737, 10 μM S63845). Cell shown left is representative of 136 min post apoptosis induction. Rounding of deposited membrane into individual F-ApoEVs shown right, 10 min intervals from 66 min to 136 min post apoptosis induction. Cells were stained with A5-FITC (blue) and TO-PRO-3 (magenta). Data is representative of (n = 3) independent experiments. b Quantification of the rate of vesicle formation from LLSM images at 60-, 120-, and 180-min post initial phosphatidylserine (PtdSer) exposure in FOOD. Data points represent individual cells (n = 7), representative from (n = 3) independent experiments. c Quantification of F-ApoEVs diameter (µm) from LLSM imaging, 240 min post apoptosis induction. Data points represent individual F-ApoEVs (n = 659 F-ApoEVs, from n = 18 cells) representative from (n = 3) independent experiments. Solid red line indicated mean, dashed red line indicates quarterlies. d Quantification of the number of F-ApoEVs generated per cell from LLSM imaging, 240 min post apoptosis induction. Data points represent individual cells (n = 31). Solid red line indicated mean, dashed red line indicates quarterlies. e Representative MIP confocal laser scanning microscopy (CLSM) images of F-ApoEV formation in free GFP expressing A431 cells 155 min post-treatment with BH3-mimetic cocktail (2.5 µM ABT-737, 0.5 µM S63845). A5-V450 staining shown in blue. f Representative time lapse CLSM images of A431 cells following treatment with a BH3-mimetic cocktail. Lower Z-plane shown above, and 3D rendered cross-section of the orthogonal planes of side view shown below. Cell membranes were visualised with PHK26, and exposed PtdSer with A5-V450. g Representative CLSM MIP of apoptotic MEFs treated with Jasplakinolide (50 nM) or vehicle control. Exposed PtdSer was visualised with A5-FITC and cell nucleus with Hoechst 33342. h Schematic diagram of F-ApoEV formation from FOOD. At least three independent experiments were performed for all experiments unless otherwise specified.

It should be noted that the biogenesis of F-ApoEVs is distinct from the formation of other large EV subsets such as ApoBDs and migrasomes (large EVs generated from trailing edge of cells during migration40). First, as F-ApoEVs are generated from the vesicularisation of membranous remnants attached to extracellular substrates following apoptotic cell retraction (Fig. 2a), the biogenesis mechanism is clearly distinct from ApoBDs which are generated from membrane blebs and dynamic apoptopodia that actively radiate from the apoptotic cell23. Second, cell types like MEFs do not readily form ApoBDs due to the lack of apoptopodia formation34, further highlighting that the F-ApoEVs observed are not simply deposition of ApoBDs onto extracellular substrates. Third, cell migration and the presence of membranous remnants prior to apoptosis induction were not observed during time-lapse imaging (Fig. 2f, Supplementary Fig. 7), ruling out the presence of migrasomes. Fourth, FOOD and F-ApoEV formation readily occurred in the presence of both the cell migration inhibitor jasplakinolide41 (Fig. 2g; Supplementary Fig. 8, 9), and migrasome inhibitors SMS2-IN-142 (sphingomyelin synthase 2 inhibitor) and ISA-2011B43 (PIP5K1α inhibitor) (Supplementary Fig. 10). Together, these data support the concept of FOOD formation and subsequent vesicularisation as an alternative mechanism for generating large ApoEVs during apoptosis (Fig. 2h).

FOOD/F-ApoEVs are enriched in actin and adhesion proteins

Identification of EV contents can infer mechanistic and functional insights. Therefore, we next performed label-free quantitative proteomic analysis on A431-derived FOOD. During our investigation, we noted that FOOD/F-ApoEVs are not readily displaced from the substrate (i.e. plastic cell culture dish) by trypsin/EDTA treatment. Based on this feature, we developed an in situ approach to enrich FOOD on the culture dish by removing apoptotic cells and other EV subsets in the culture supernatants with a series of PBS wash steps, before removing remaining cells with trypsin/EDTA treatment (Supplementary Fig. 11a–c). The FOOD on the culture dish was then directly lysed for proteomic analysis. We identified a total of 601 proteins present in FOOD, 573 of which were also found in apoptotic cells and 28 proteins unique to FOOD (Fig. 3a, Supplementary File). The most abundant FOOD proteins were actin and histone proteins, as well as several cytoskeletal, membrane, and adhesion proteins. Gene ontology analysis revealed an abundance of proteins found in FOOD that can bind to integrins, cadherin, and actin, as well as protein components of the cytoskeleton and focal adhesions (Fig. 3b, Supplementary Fig. 12a).

Fig. 3: FOOD is enriched in adhesion associated proteins and cytoskeletal components.

a Venn diagram (left) showing the number of proteins unique to and shared by apoptotic A431 cells and A431 cell-derived Footprint Of Death (FOOD) as identified by proteomic analysis, and pie chart (right) showing the 28 proteins unique to FOOD, organised by relative abundance. b Pie chart showing the top 20 most highly abundance proteins present in FOOD as identified via proteomic analysis, organised by relative abundance. c, d Representative maximum intensity projection (MIP) confocal laser scanning microscopy (CLSM) images of FOOD formation in MEFs 100 min post-treatment with BH3-mimetic cocktail (5 μM ABT-737, 10 μM S63845). Cell actin was visualised by SiR-Actin (red) (c) or Phalloidin (red) (d), and exposed phosphatidylserine (PtdSer) visualised with A5-FITC. e Representative MIP images from time-lapse CLSM of FOOD-derived ApoEV (F-ApoEV) formation in MEFs following treatment with a BH3-mimetic cocktail. Rounding of deposited membrane into individual F-ApoEVs at 20 min intervals from 170 to 260 min post apoptosis induction, shown below, as indicated by the dotted box. Cell actin was visualised by SiR-Actin and exposed PtdSer visualised with A5-V450. f Representative correlative light and electron microscopy (CLEM) images of FOOD formation in MEFs 120 min post-treatment with BH3-mimetic cocktail. CLSM images shown left, and correlating scanning electron microscopy (SEM) image shown right. g Representative MIP CLSM images of FOOD formation in MEFs 100 min post-treatment with BH3-mimetic cocktail. Microtubules were visualised by SiR-Tubulin (red) and exposed PtdSer visualised with A5-FITC. h Localisation of vinculin (green) in viable (left) or BH3-mimetic cocktail (2 h post treatment) treated. Cell cytoskeleton visualised with phalloidin and exposed PtdSer with A5-V450. At least three independent experiments were performed for all experiments unless otherwise specified.

In support of the proteomic data, we performed CLSM analysis and confirmed the presence of F-actin in FOOD and F-ApoEVs by SiR-actin and phalloidin staining (Fig. 3c, d). Time-lapse imaging revealed that when the apoptotic cell contracts to generate FOOD, F-actin can be found within F-ApoEVs during vesicle rounding (Fig. 3e). To further visualise the localisation of F-actin in FOOD relative to the morphological features observed in SEM, correlative light and electron microscopy (CLEM) was performed and demonstrated that the thin-fibrous structures surrounding apoptotic cell to harbour F-actin (Figs. 1c, 3f). In addition to F-actin, we also observed the presence of other cytoskeletal components such as tubulin in FOOD and F-ApoEVs (Fig. 3g). Gene ontology analysis indicated a large proportion of proteins enriched in FOOD are components of focal adhesions such as filamin, α-actinin, various integrins, and vinculin (Supplementary Fig. 12b). The presence of vinculin in FOOD and F-ApoEVs was confirmed by immunostaining, however, the localisation of vinculin is more disperse compared the discrete puncta observed in viable cells (Fig. 3h), possibly due to caspase-mediated cleavage of focal adhesion complexes44.

Large EV subsets such as ApoBDs, migrasomes and exophers often contain organelles such as mitochondria, nucleus and golgi7,13,45,46. Utilising cell lines expressing fluorescently-tagged proteins to monitor the localisation of golgi and lysosomes, and staining cells with MitoTracker Green to track mitochondria, we found that organelles are present in low abundance, with on average, less than 20% of FOOD harbouring organelles (Supplementary Fig. 13a–g). Although histones were identified in our proteomic analysis of FOOD (Fig. 3b), we were unable to detect histones in FOOD by microscopy approaches (Fig. 3d, Supplementary Fig. 13h). Furthermore, nuclear DNA was not present in FOOD as monitored by Hoechst 33342 staining (Supplementary Fig. 1b). This data indicates FOOD is devoid of nuclear materials and the that the enrichment of histones in the proteomic analysis may be due the release of histones during apoptosis47 and subsequent deposition onto the culture dish.

Formation of F-ApoEVs through FOOD is dependent on ROCK1 activation

During apoptosis, caspase-activated ROCK1 aids both cell contraction26 and plasma membrane blebbing24. Since cell contraction is involved in FOOD formation (Fig. 1a), we examined the role of ROCK1 in this process. To this end, we utilised MEFs derived from ROCK1 non-cleavable (ROCK1nc) mice that express a mutated form of ROCK1 resistant to caspase-mediated cleavage, which results in reduced regulatory myosin light chain phosphorylation and consequent decreased cellular contractile force generation during apoptotic cell death48 (Supplementary Fig. 14). Upon apoptosis induction, ROCK1nc MEFs exhibited defective FOOD formation, whereby a greater proportion of the cell body remained adhered to the surface of the substrate without cell retraction (Fig. 4a). ROCK1nc MEF-derived FOOD exhibited flat, membrane-sheet like morphology (Fig. 4a). Furthermore, quantification of F-ApoEV formation revealed that ROCK1nc MEF-derived FOOD rounds into significantly fewer F-ApoEVs, compared to the WT control (Fig. 4b). Further quantification of CLSM analysis confirmed that ROCK1nc cells have a significantly greater number of FOOD branches compared to WT cells and the branches occupied a larger area (Fig. 4c), highlighting a defect in cell retraction for ROCK1nc cells undergoing apoptosis. Defects in FOOD morphology in ROCK1nc MEFs were also confirmed by SEM analysis (Fig. 4d). Time-lapse CLSM analysis of F-actin in apoptotic ROCK1nc MEFs revealed a greater amount of F-actin in FOOD generated by ROCK1 MEFs compared to WT cells (Fig. 4e), potentially limiting vesicularisation into abundant F-ApoEVs. Together, this data reveals that caspase-mediated activation of ROCK1 is required for FOOD formation, and abundant F-ApoEV generation.

Fig. 4: Formation of FOOD and F-ApoEVs is dependent on caspase-cleaved ROCK1.

a WT or ROCK1nc MEFs were treated with a BH3-mimetic cocktail (5 μM ABT-737, 10 μM S63845) and imaged by time-lapse confocal laser scanning microscopy (CLSM). Cell membrane and nucleus were visualised by PHK26 and Hoechst 33342 staining, respectively, and exposed phosphatidylserine (PtdSer) with A5-FITC. b Quantification of the number of FOOD-derived ApoEV (F-ApoEVs) generated per cell from CLSM imaging, by WT (grey) or ROCK1nc (blue) MEFs, 60-, 120-, and 180 min post initial PtdSer exposure. Data points represent individual cells (n = 12). c Quantitative analysis of The FOotprint Of Death (FOOD) generated by WT or ROCK1nc MEFs: number of branches per cell, area of branches per cell (μm2), and branch thickness (μm). Data is pooled from (n = 3) independent experiments. d Representative scanning electron microscopy (SEM) images of WT or ROCK1nc MEFs, treated with a BH3-mimetic cocktail. (n = 2). e Visualisation of actin in FOOD/F-ApoEV formation in WT or ROCK1nc MEFs following treatment with a BH3-mimetic cocktail. Representative maximum intensity projection (MIP) images from time-lapse CLSM shown. Cell actin was visualised by SiR-Actin and exposed PtdSer with A5-FITC. Error bars in (b, c) represent s.e.m. Unpaired student’s two tailed t-test was performed to determine the indicated p-values.

F-ApoEVs can be efferocytosed and traffic viral material

Lastly, we explored the functional consequence of FOOD formation. As FOOD exposes the ‘eat-me’ signal PtdSer (Fig. 1a), we examined whether FOOD could be recognised and efferocytosed by phagocytes using LLSM. In co-culture studies of bone marrow-derived macrophages (BMDMs) and isolated MEF-derived FOOD, BMDMs could interact with FOOD and engulf F-ApoEVs (Fig. 5a, Supplementary Fig. 15). Once at the site of cell death, BMDMs could be observed ‘grabbing’ strands of FOOD and efferocytosing F-ApoEVs. 3D analysis of LLSM data clearly demonstrated the internalisation of A5+ F-ApoEVs inside BMDMs, indicative of efferocytosis (Supplementary Video 3, 4). As FOOD resembles a smaller ‘bite-sized’ portion of the apoptotic cell, we hypothesised that the FOOD may serve as an ‘appetizer’ to prime phagocytes before engulfing the whole apoptotic cell. Therefore, we primed BMDMs with MEF-derived FOOD for 24 h and subsequently performed an efferocytosis assay with CypHer 5E-labelled apoptotic Jurkat T cells. In contrast to the untreated BMDMs, BMDMs primed with FOOD exhibited an increase in efferocytotic efficiency (Fig. 5b). Taken together, these results indicate that FOOD may aid phagocytes to identify the site of cell death and prepare for efferocytosis.

Fig. 5: The role of F-ApoEVs in cell clearance and intercellular communication.

a Representative maximum intensity projection (MIP) images from time-lapse lattice light sheet microscopy (LLSM) of cell trace violet-stained BMDM interacting with and engulfing MEF-derived FOOD/F-ApoEVs stained with A5-PE (magenta). b Quantification of re-feeding engulfment assay measured by %Cypher5E+ BMDMs following initial incubation with the FOotprint Of Death (FOOD)/FOOD-derived ApoEVs (F-ApoEVs) derived from 1 ×104 or 2 ×104 MEFs for ~24 h, followed by incubation of CypHer5E-labelled apoptotic Jurkat T cells. Data is representative of (n = 3) independent experiments. c Representative time-lapse MIP confocal laser scanning microscopy (CLSM) images of FOOD formation in A549 cells 19- to 24 h post-infection (p.i.) with influenza A virus (IAV) (MOI = 10) (right). Quantification of the frequency of FOOD formation in IAV-infected apoptotic cells (left). Data is representative of (n = 3) independent experiments. d Representative scanning electron microscopy (SEM) images of FOOD formation in IAV-infected apoptotic cells (24 h p.i.). Data is representative of (n = 2) independent experiments. Localisation of viral nucleoprotein (NP) (e) or hemagglutinin (HA) (f) (green) in IAV-infected A549 cell-derived FOOD/ F-ApoEVs (24 h p.i.). Quantification of %FOOD stained positive for NP or HA protein shown on the right. g Representative transmission electron microscopy (TEM) images of isolated F-ApoEVs from A549 cells induced to undergo apoptosis with either BH3-mimetic cocktail treatment, or IAV infection. White arrows indicate virions. Quantification of the absolute number of NP+ A549 cells (h) or NP+A5+ A549 cells (i) following co-incubation of A549 cells with the indicated amount of F-ApoEVs isolated from BH3-mimetic cocktail treated or IAV infected A549 cells. Data is representative of (n = 3) independent experiments. Error bars represent s.e.m. Unpaired student’s two tailed t-test was performed to determine the indicated p-values.

IAV infection induces vast amounts of apoptosis in vivo, which can result in the displacement of cells from the respiratory track and lung36. Therefore, IAV infection may resemble a physiologically relevant model where apoptotic cells could leave behind FOOD at the site of cell death. Thus, we next investigated the functional consequence of FOOD formation under a pathological context using the PR8 H1N1 model of IAV infection. A549 epithelial cells readily generated FOOD in response to IAV-induced apoptosis whereby FOOD exhibited PtdSer-rich membrane depositions with thin actin fibres and resulted in F-ApoEV formation (Fig. 5c). Similar findings were also observed in primary murine BMDMs infected with IAV (Supplementary Fig. 16). SEM analysis revealed that the ultrastructure of FOOD in response to IAV is comparable to FOOD formed under other apoptotic settings (Fig. 5d). Furthermore, viral proteins such as nucleoprotein (NP) and hemagglutinin (HA) were detected within A549-derived FOOD and F-ApoEVs, 24 h post infection (Fig. 5e, f). Importantly, TEM demonstrated that on occasion, IAV viral particles could be observed in F-ApoEVs isolated from IAV-infected A549 cells (Fig. 5g). To determine if F-ApoEVs generated during IAV infection had the capacity to propagate viral infection, we co-cultured F-ApoEVs isolated from IAV-infected A549 cells with healthy A549 lung epithelial and observed a marked increase in A549 cells staining positive for viral NP, indicative of IAV infection (Fig. 5h, Supplementary Fig. 17). Notably, approximately half of the infected A549 cells were undergoing cell death as indicated by A5 staining (Fig. 5i). Thus, F-ApoEVs derived from IAV-infected cells may serve as a reservoir for viral proteins and infectious virions and contribute to viral spread.

Discussion

Apoptosis occurs throughout life in essentially all tissues as part of normal development and cell turnover. Since >200 billion cells die in the human body via apoptosis daily, it is critical for phagocytes to detect the location of cell death to aid the removal of apoptotic materials effectively. Here, we describe a mechanism by which the dying cells could leave behind a ‘FOotprint Of Death’ (FOOD) to mark the site of cell death. Importantly, through the generation of FOOD, a distinct subclass of large ApoEVs is formed, coined as FOOD-derived ApoEVs (F-ApoEVs), which can mediate important communication with phagocytes at the site of cell death. The data presented here provide several key insights into the highly orchestrated process of apoptotic cell-phagocyte interaction.

During the progression of apoptosis, a distinct subset of large EVs known as ApoBDs are generated though highly coordinated morphological changes. This process involves the formation of membrane blebs, followed by apoptopodia protruding out from the cell body, and subsequent fragmentation of apoptopodia to generate discrete ApoBDs7,22,23. These ApoBDs can often dissipate from the site of cell death and interact with cells within the vicinity49,50. In stark contrast to ApoBDs, we found that prior to apoptotic membrane blebbing, adherent cells can retract from the extracellular substrate and generate another distinct subset of large EVs, described here as F-ApoEVs, that are tightly anchored to the site of cell death. The formation of F-ApoEVs also progresses through highly coordinated morphological stages including: 1) Cell retraction: the apoptotic cell retracts and deposits membranous material and long F-actin fibres on the extracellular matrix, 2) PtdSer exposure: the gradual exposure of PtdSer on FOOD membranes, and 3) EV rounding: FOOD membrane undergoes a ‘rounding’ event to form distinct EVs. Thus, the generation of F-ApoEVs is morphologically and spatially distinct from ApoBD formation. As F-ApoEVs can remain attached to the site of cell death, it may serve a specific function during the clearance of apoptotic cells. Notably, F-ApoEVs can expose the ‘eat-me’ signal PtdSer, interact with and be engulfed by phagocytes. Furthermore, since F-ApoEVs are largely intact, these apoptotic materials may harbour metabolites and intracellular proteins that could function as ‘good-bye’ signals and damage-associated molecular patterns, respectively, to initiate downstream processes such as wound healing and inflammation at the site of cell death[1](https://www.nature.com/articles/s41467-025