GoKawiilAntigen-presenting cells (APCs) must continuously survey tissues through the ingestion and processing of antigens for presentation to mediate antigen-specific T cell responses11. Previously, in the course of studying a selection of tissue-specific sites of tolerance, including in tumours, we expressed the fluorescent protein ZsGreen in multiple non-inflammatory settings across a range of tissues with slow turnover rates (for example, Scbg1a1creERT2 airway epithelium)12 or faster turnover (for example, K14cre skin cells, tumours and Vil1cre intestinal epithelium). ZsGreen is both bright and stable, which facilitates long-lived tracking13,14. In this study, cytosolic ZsGreen expressed under the Scbg1a1creERT2 and K14cre promoters all routinely illuminated substantial CD45+ immune populations that contained numerous submicrometre-sized vesicular puncta of ZsGreen (Fig. 1a,b), a finding consistent with ingestion of tissue-associated protein. When CD45+ cells containing ZsGreen were isolated from these healthy tissues, various myeloid cells were highlighted (Fig. 1c,d, Extended Data Fig. 1a,b and Supplementary Fig. 2), including dendritic cells and neutrophils. However, macrophages were the most consistently loaded. These loading frequencies also mirrored previous characterizations of myeloid sampling of skin13 and tumour cytoplasts14.
Fig. 1: Myeloid cells sample proteins from healthy and tumour tissues. The alternative text for this image may have been generated using AI. Full size image a,b, Images of example lung (a) and skin (b) samples taken from mice expressing ZsGreen under the Scgb1a1creERT2 (a) or K14cre (b) promoter. Insets show images of in situ macrophages labelled with CD45 (a) or sorted CD45+ macrophages (b) showing ZsGreen puncta inside the macrophages. c,d, Quantification of isolated lung myeloid cells from Scgb1a1creERT2;ZsGreen mice (c, n = 5 mice) or skin myeloid cells from K14cre;ZsGreen mice (d, n = 3 mice) shows specific uptake in tissues, prominently in macrophages (M). Lymphocytes were routinely negative for ZsGreen. AM, alveolar macrophage; cDC1, type 1 conventional dendritic cell; cDC2, type 2 conventional dendritic cell; cMo, conventional monocyte; IM, interstitial macrophage; Mo, monocyte; Neu, neutrophil; pMo, patrolling monocyte. e, Example image of B16 melanoma cells expressing ZsGreen transplanted into mice ubiquitously expressing tdTomato. The inset is an example image of CD45+ macrophages. f, Quantification of isolated tumour-associated myeloid cells shows broad uptake of ZsGreen across myeloid cell subsets. n = 4 mice. g, Quantification of isolated inguinal LN myeloid cells from Scgb1a1creERT2;ZsGreen mice shows no uptake in distant LNs. n = 4 mice. mDC1, migratory cDC1; mDC2, migratory cDC2; rDC1, residential cDC1; rDC2, residential cDC2. h, Intracellular staining for airway-specific VEGFR3 protein in lung (black) compared with splenic (grey) myeloid cell populations with ZsGreenhi lung macrophages (green) further enriching for the VEGFR3 geometric mean fluorescence intensity (gMFI) signal. i, VEGFR3 gMFI in lung and splenic myeloid cell populations normalized to the average gMFI of splenic myeloid cells. n = 3 mice. Representative of 3 experiments, n = 3–5 mice per experiment. Shown are mean ± s.d. Scale bars, 40 µm (a), 50 µm (b) or 100 µm (e). Source data
These observations do not result from misexpression of transgenes in host tissues, as demonstrated by the following series of observations. First, we compared the ZsGreen fluorescence intensity of lung myeloid cells in mice expressing ubiquitous ZsGreen (Actbcre;ZsGreen), lung-specific ZsGreen (Scgb1a1creERT2;ZsGreen) or no ZsGreen (C57BL/6 wild-type (B6 WT)). In each myeloid cell population, ZsGreen fluorescence intensity in Scgb1a1creERT2;ZsGreen mice was greater than in B6 WT mice but less than in Actbcre;ZsGreen mice (Extended Data Fig. 1b). This result is consistent with the accumulation of exogenous ZsGreen protein as opposed to cell-autonomous ZsGreen expression. Second, transplantation of normal bone marrow into mice expressing ZsGreen gave rise to donor-derived myeloid cell populations that were bright for ZsGreen (Extended Data Fig. 1c and Supplementary Fig. 3a). Third, in mice ubiquitously expressing tdTomato and subcutaneously transplanted with B16-F10 tumours expressing ZsGreen (B16-ZsGreen), host CD45+ immune cells contained tumour-derived vesicular ZsGreen puncta. Moreover, a significant proportion of tumour-associated myeloid cells ingested ZsGreen (Fig. 1e,f and Supplementary Fig. 3b). Finally, whereas we had previously observed trafficking of tissue-specific ZsGreen to tissue-draining lymph nodes (LNs)13, ZsGreen was not detected in the non-draining LNs of Scgb1a1creERT2;ZsGreen mice (Fig. 1g), which indicated tissue-specific uptake.
For tumour antigens, it was previously found that tumour antigens are co-packaged in macrophage vesicles together with tumour-derived ZsGreen tracer13. To determine whether other non-tumour self-antigens in healthy tissue with low turnover, such as epithelium, are similarly taken up in these myeloid cell populations, we examined the levels of VEGFR3 and PLVAP proteins, which are specifically expressed on the cell surface of non-haematopoietic cells in the lung. Intracellular flow cytometry analyses of cellular components from the lungs of Scgb1a1creERT2;ZsGreen mice showed that local myeloid cell populations from the lung, but not distant splenic myeloid cells, contained these self-proteins (Fig. 1h,i, Extended Data Fig. 1d and Supplementary Fig. 3c). Furthermore, ZsGreenhi myeloid cells were brighter for VEGFR3 and PLVAP stains compared with ZsGreenlow myeloid cells (Fig. 1h and Extended Data Fig. 1d). Therefore, tissue-associated myeloid cells regularly sample tissue-associated self-proteins, both engineered tracer proteins and those normally expressed by tissues.
Macrophages can sample from live cells
Several mechanisms of uptake might contribute to myeloid cell sampling from tissues in vivo. The most heavily studied so far is phagocytosis of dead cells and endocytosis of exosomes. To examine in detail how myeloid cells can obtain intracellular material from nearby cells, we established an in vitro assay to measure cell sampling using donor cell populations, grown at around 99% viability, in log phase. We co-cultured bone-marrow-derived macrophages (BMDMs) with cells expressing ZsGreen, focusing on two model target cells: B16-ZsGreen melanoma cells (which, in these conditions, minimally produce exosomes) and primary mouse embryonic fibroblasts (MEFs) isolated from mice ubiquitously expressing ZsGreen (MEF-ZsGreen). From these co-culture experiments, we detected significant ZsGreen+ uptake from both target cells into BMDMs. Moreover, the intensity of ZsGreen fluorescence was hundreds of times lower than the intensities of intact donor cells, a result consistent with partial sampling as opposed to complete engulfment (Fig. 2a and Supplementary Fig. 4a). When ZsGreen+ BMDMs were sorted for imaging by high-resolution spinning disc confocal microscopy, internalized submicrometre-sized ZsGreen+ puncta were readily visualized (Fig. 2b,c and Extended Data Fig. 2a).
Fig. 2: Live cells can be sampled in a cell-contact-dependent manner without caspase activation. The alternative text for this image may have been generated using AI. Full size image a, BMDMs and ZsGreen-target cells were co-cultured for 16 h (left) before evaluation for ZsGreen uptake by flow cytometry (right). b, Images of sorted ZsGreen+ BMDMs with membrane tdTomato showcasing ZsGreen+ puncta (representative of n = 2 experiments). Right image is the magnification of the yellow square on the left. Colour bar: grey units. Scale bars, 20 µm (left) or 5 µm (right). c, Area of individual ZsGreen+ vesicles (mean = 0.145 μm2). n = 77 vesicles quantified from 23 cells. Shown are mean ± s.d. d,e, BMDMs were co-cultured directly with B16 cells (Direct), with supernatant from B16 cell culture (d, SN; P = 0.0052) or with a Transwell insert containing B16 cells (e, Indirect); P = 0.0018). f–h, In vitro co-cultures were treated with DMSO (vehicle control) or with an exosome and microparticle inhibitor (DMA; f), an endocytosis inhibitor (Dynole; g) or a caspase inhibitor (zVAD; h; P = 0.0378). i, Schematic (left) and plots (right) of induction of GFP expression in B16-GC3AI cells treated with DMSO or staurosporine. j, Co-culture of BMDMs with B16-GC3AI cells demonstrates that uptake is predominantly from live mCherry+GFP− target cells. For d, h and j, n = 3 biological replicates. Each point represents one biological replicate (mean of n = 3 technical replicates). Shown are the mean of biological replicates ± s.e.m. Two-sided paired t-test. *P < 0.05, **P < 0.01. Source data
We then evaluated whether antigen sampling from live cells in this setting is mediated through the uptake of soluble particles and/or depends on cell contact, whereby the former in particular is expected for the ingestion of free exosomes and perhaps apoptotic blebs. Culture of BMDMs with either B16-ZsGreen-derived supernatant containing a limited number of exosomes produced in a 48-h period or B16-ZsGreen cells that were separated from BMDMs by a Transwell insert both significantly reduced uptake (Fig. 2d,e and Extended Data Fig. 2b). Moreover, treatment of cultures with an inhibitor of exosome and microparticle release or with an inhibitor of endocytosis did not reduce uptake (Fig. 2f,g, Extended Data Fig. 2c–f and Supplementary Fig. 4b,c). Therefore, although endocytosis of soluble material such as exosomes and microparticles may modestly contribute to this feature, this result suggests that there is a distinct dominant mechanism of uptake of live cell-associated material in this setting, which requires cell contact.
An existing hypothesis suggests that phagocytosis of apoptotic bodies, termed efferocytosis but here we will use ‘phagocytosis’, is the major mechanism by which tissues donate material to surveilling APCs15,16,17. We therefore sought to test whether cell death is necessary for the substantial cell sampling observed in our system. When we treated co-cultures with the caspase inhibitor zVAD, there was no effect on ZsGreen uptake (Fig. 2h, Extended Data Fig. 2f,g and Supplementary Fig. 4d). This result confirms that this uptake mechanism does not rely on cell death. To study more directly whether the material ingested into macrophages comes from cells undergoing apoptosis, we used a B16-F10 model cell line that expresses constitutive mCherry and a split GFP caspase-3 activity indicator that fluoresces only when cleaved, for example, during apoptosis18 (B16-GC3AI) (Fig. 2i). Use of this model confirmed that B16-GC3AI cells in our assays were highly viable (0.12% GFP+), whereas apoptotic cell death induced by staurosporine treatment resulted in GFP reporter fluorescence (>63% GFP, Fig. 2i). When these reporter-expressing cells were co-cultured with BMDMs, the majority of the mCherry+ cells (those that had taken up material from the donor cells) lacked GFP expression compared with uptake following apoptosis (1% for live sampling versus 11% after staurosporine treatment; Fig. 2j and Extended Data Fig. 2h). Together, these results provide further support that an alternative sampling pathway, beyond those involving apoptosis, exists to obtain material from live cells.
Live imaging of live sampling
... continue reading