Research Article

Perivascular dendritic cells elicit anaphylaxis by relaying allergens to mast cells via microvesicles

See allHide authors and affiliations

Science  09 Nov 2018:
Vol. 362, Issue 6415, eaao0666
DOI: 10.1126/science.aao0666

Dendritic cells give mast cells a nudge

Anaphylaxis is a life-threatening allergic reaction triggered after antigen-specific immunoglobulin E (IgE) antibodies bind to target allergens. These antibodies then cross-link IgE-specific Fc receptors on the surface of mast cells. The mast cells rapidly release inflammatory mediators, including histamine, resulting in smooth muscle contraction, vasodilation, and blood vessel leakage. Because mast cells are usually found in the perivascular abluminal surface of blood vessels, it has been unclear how blood-borne allergens can interact with them. Choi et al. used live intravital imaging of the mouse vasculature to show that a specialized subset of dendritic cells sample blood-borne antigens and relay them to mast cells on the surface of microvesicles (see the Perspective by Levi-Schaffer and Scheffel). IgE-bound mast cells then vigorously degranulate after contact with these microvesicles.

Science, this issue p. eaao0666; see also p. 640

Structured Abstract


An increasing number of individuals suffer from acute anaphylaxis, particularly food-associated reactions. The symptoms of anaphylaxis are triggered soon after allergens (such as peanut antigens, insect venom, or certain medications) enter the circulation of sensitized subjects who have elevated levels of antigen-specific immunoglobulin E (IgE) antibodies in their circulation. Mast cells (MCs) are the primary effectors of anaphylaxis and become activated when allergens make contact with IgE antibodies bound to the surfaces of MCs. Upon activation, MCs rapidly release prestored inflammatory mediators. When this occurs simultaneously throughout the body, anaphylaxis and shock result. As MCs are typically found in the perivascular abluminal surface of relatively impregnable blood vessels, how these cells perceive and interact with blood-borne allergens is unclear.


Extravascular MCs were recently reported to be capable of extending cellular protrusions to directly sample vascular contents. Therefore, we reasoned that if MCs are directly sampling the blood, we should find that MCs are the first cells to acquire allergens upon systemic infusion of labeled allergen. We systemically infused mice with labeled dextran and examined which immune cells in the proximity of the vasculature had acquired the label within 30 min of antigen administration.


Surprisingly, MCs acquired only a small number of the dextran particles. Instead, most of the labeled antigen was acquired by a subset of dermal dendritic cells (DCs). These DCs, defined as CD301b+ dermal cDC2, were found in close proximity to MCs in the perivascular abluminal surface lining dermal blood vessels. Furthermore, they appeared to directly sample the blood through the extrusion of dendrites. These DCs were critical to anaphylaxis, as IgE-sensitized mice deficient in CD301b+ dermal cDC2 failed to undergo anaphylaxis. Live microscopy of the mouse vasculature revealed numerous perivascular DCs routinely sampling blood-borne allergens via dendrites, which penetrated the endothelial wall into the vascular lumen (left panel in the figure). Blood-borne antigens captured in this manner were relayed on the surfaces of 0.5- to 1.0-μm microvesicles (MVs) to neighboring MCs and DCs in the perivascular space. A panoply of receptors can be found on the surfaces of DCs, including the mannose receptor, which enable these cells to capture a broad range of foreign antigens in the blood. Knocking down mannose receptors markedly reduced the capacity of DCs to bind antigen and convey antigen to MCs. Furthermore, large numbers of MVs were constitutively shed by DCs regardless of whether contact with antigens was made. MVs were products of specific budding and pinching-off activities occurring on the plasma membranes of DCs. Additionally, MV shedding was abrogated when vacuolar protein sorting–associated protein 4 (VPS4), which mediates scission of MVs, was knocked down. DCs impaired in their ability to express VPS4 were found to be incapable of activating MCs or causing anaphylaxis. Noticeably, distribution of antigens by DCs on budding MVs occurred relatively quickly and was not preceded by internalization of antigen. Upon contact with allergens on the surfaces of MVs, IgE-bound MCs were found to degranulate vigorously, releasing their payload of inflammatory mediators (right panel in the figure) and triggering anaphylaxis. Thus, allergen-induced MC degranulation and anaphylaxis require the critical contribution of perivascular DCs in capturing and relaying circulating allergens.


We demonstrate how IgE-sensitized MCs are indirectly activated by blood-borne allergens. Additionally, our study reveals how perivascular DCs continuously sample blood and initiate and markedly enhance inflammatory and immune responses by rapidly discharging antigen-bearing MVs to surrounding immune cells.

Transfer of blood-borne allergens by DCs to MCs during anaphylaxis.

(A) Sampling of blood-borne allergens by perivascular DCs. (B) Relaying of allergens borne on vesicles to neighboring MCs and DCs. Ag, antigen; dDC, dermal DC.


Anaphylactic reactions are triggered when allergens enter the blood circulation and activate immunoglobulin E (IgE)–sensitized mast cells (MCs), causing systemic discharge of prestored proinflammatory mediators. As MCs are extravascular, how they perceive circulating allergens remains a conundrum. Here, we describe the existence of a CD301b+ perivascular dendritic cell (DC) subset that continuously samples blood and relays antigens to neighboring MCs, which vigorously degranulate and trigger anaphylaxis. DC antigen transfer involves the active discharge of surface-associated antigens on 0.5- to 1.0-micrometer microvesicles (MVs) generated by vacuolar protein sorting 4 (VPS4). Antigen sharing by DCs is not limited to MCs, as neighboring DCs also acquire antigen-bearing MVs. This capacity of DCs to distribute antigen-bearing MVs to various immune cells in the perivascular space potentiates inflammatory and immune responses to blood-borne antigens.

Currently, 4 to 5 persons per 100,000 suffer from anaphylaxis annually. These numbers continue to grow, particularly for food-associated reactions (1, 2). These reactions are especially frequent in the young, most of whom also present atopic diseases such as asthma, eczema, or allergic rhinitis (3). Acute anaphylaxis is associated with severe pathophysiological symptoms, such as hives, loss in blood pressure, vasculature leakage, and a drop in body temperature, which can be fatal (4). These symptoms are triggered soon after allergens such as peanut antigens, insect venom, and certain medications enter the circulation of antigen-specific immunoglobulin E (IgE)–sensitized individuals (2). Mast cells (MCs) are primary effectors of anaphylaxis because of their distinct ability to release large amounts of cytoplasmic granules enriched in inflammatory chemicals upon allergen activation of their surface IgE. MCs are typically found lining blood vessels, so when allergens enter the circulation, widespread MC degranulation is triggered, resulting in rapid and systemic onset of anaphylaxis. As MCs are located in the perivascular abluminal surface of relatively impregnable endothelial cells, how blood-borne allergens contact MCs is unclear.

MCs have the capacity to directly probe blood vessels with cellular protrusions to acquire IgE antibodies from the circulation (5). Dendritic cells (DCs) are often observed alongside MCs at many sites. DCs are primarily immune surveillance cells with the distinct capacity to extrude dendrites between cells that are connected via tight junctions (6). These probing dendrites allow DCs lying underneath gut and respiratory epithelial tracts to sample luminal contents (6, 7). Additionally, epidermal Langerhans cells of the skin can penetrate the stratum corneum to sample external antigens (8). In this study, we investigated how abluminal perivascular MCs detect circulatory antigens through cooperation with adjacent DCs to trigger anaphylaxis.


MCs and CD11c+ cells are critical mediators of anaphylaxis

We investigated whether MCs were able to bind and detect blood-borne antigens in a manner similar to circulating IgE antibodies (5). We intravenously injected tetramethyl rhodamine isothiocyanate (TRITC)–conjugated dextran, which is unable to enter the extravascular space. We then used flow cytometry 30 min after injection to evaluate the ability of dermal abluminal MCs to acquire dextran-TRITC. Only ~1% of cKit+FcεRI+ skin MCs were positive for dextran-TRITC (Fig. 1A and fig. S1). Unexpectedly, up to 5% of CD11c+ cells in the skin cell preparation were positive for dextran-TRITC (Fig. 1A and fig. S1). To locate these CD11c+ cells within the tissue, we prepared whole mounts of the mouse ear and probed them for MCs and CD11c+ cells. Numerous CD11c+ cells were in close proximity to blood vessels and often in direct contact with both endothelial cells and MCs (Fig. 1B).

Fig. 1 PCA and PSA are mediated by MCs and CD11c+ cells.

(A) Efficient antigen uptake by CD11c+ cells. CD11c-GFP mice were intravenously injected with dextran-TRITC. After 30 min, mice were sacrificed and their ears were dissected to generate a single-cell suspension. The dextran-positive populations among CD45+FcεRI+cKit+ MCs and CD45+CD11c+ cells were compared. n = 8 mice per group. Data are presented as the mean ± SEM. *P < 0.001, unpaired Student’s t test. (B) CD11c+ cells lie in close proximity to blood vessels. The ears of mice expressing CD11c-GFP (green) were dissected. A whole mount was then prepared and stained for MCs (avidin, red) and blood vessels (CD31, blue). Scale bars, 50 μm (left) and 15 μm (right). (C) CD11c+ cells mediate vascular leakage. CD11c-DTR or littermate mice were intraperitoneally and intravenously injected with DT or vehicle every other day twice to deplete the population of CD11c+ cells. The day before antigen challenge, the ears of CD11c-DTR or WT mice were sensitized with TNP-specific IgE. TNP-OVA was intravenously injected along with Evans blue dye into both groups. One hour postinjection, mouse ears were imaged and dissected to extract dye for optical density (OD) measurements. n = 6 to 7 mice per group. Data are presented as the mean ± SEM. *P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. N.S., not significant. (D) CD11c+ cells mediate anaphylaxis. After CD11c+ cells were depleted, IgE-sensitized CD11c-DTR mice were intravenously injected with TNP-OVA. Temperature changes were then monitored (top). Survival during PSA was recorded, and mice with temperature changes greater than 10°C were sacrificed (bottom). n = 4 to 5 mice per group. Data are presented as the mean ± SEM. **P < 0.05, two-way ANOVA (top) or log-rank test (bottom). (E and F) MC activation causes vascular leakage and anaphylaxis. Both Mcpt5-CreiDTR and littermate mice were intravenously injected with DT to deplete the population of MCs. The day before antigen challenge, the ears of each mouse were sensitized with TNP-specific IgE. TNP-OVA was intravenously injected into both groups of mice, along with (E) Evans blue dye or (F) TRITC-conjugated dextran. One hour postinjection, (E) mouse ears were imaged and dissected to extract dye for OD measurements. Data are presented as the mean ± SEM. *P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. (F) Mouse ears were dissected and fixed for whole-mount imaging. Confocal microscopy was utilized to observe ears stained for MCs (avidin, green), blood vessels (CD31 antibody, blue), and dextran (red). The regions indicated by dashed squares in the upper panels are magnified in the corresponding lower panels. Scale bars, 50 μm. (G) MC activation triggers a sharp drop in body temperature. After MCs were depleted, littermate or Mcpt5-CreiDTR mice were injected intraperitoneally with TNP-specific IgE to sensitize them. The following day, mice were intravenously injected with TNP-OVA, and temperature changes were monitored (top). Survival during PSA was recorded (bottom). n = 5 to 8 mice per group. Data are presented as the mean ± SEM. *P < 0.001, two-way ANOVA (top) or log-rank test (bottom). Two to three separate experiments were performed for each individual panel.

To investigate whether these CD11c+ cells had any functional role in anaphylactic responses to allergens, we compared anaphylactic reactions in CD11c+ cell–depleted mice and nondepleted mice by using two classical assays: passive systemic anaphylaxis (PSA) and passive cutaneous anaphylaxis (PCA). We administered repeated injections of diphtheria toxin (DT) to CD11c–DT receptor (DTR) mice depleted of CD11c+ cells (9). Sixteen hours after locally sensitizing both wild-type (WT) and CD11c+ cell–depleted mice with trinitrophenyl (TNP)–specific IgE, we challenged the mice intravenously with TNP-ovalbumin (TNP-OVA). To visualize the vascular leakage associated with PCA in the IgE-sensitized mice, we coinjected Evans blue dye and TNP-OVA intravenously. Strong vascular leakage occurred in the nondepleted mice, whereas the CD11c+ cell–depleted mice had significantly less leakage (Fig. 1C). CD11c+ cell depletion also protected mice from a severe PSA reaction upon systemic IgE sensitization and TNP-OVA challenge. A sharp drop in body temperature occurred in IgE-sensitized saline-treated CD11c-DTR mice but not in CD11c+ cell–depleted mice after intravenous administration of TNP-OVA (Fig. 1D). CD11c+ cell–depleted mice exhibited only a modest drop in body temperature, with no associated mortality (Fig. 1D). Thus, CD11c+ cells contribute substantially to the development of PCA and PSA in IgE-sensitized mice.

To confirm the role of MCs in our anaphylaxis model, we compared the PCA and PSA responses in WT and MC-depleted mice (MC depletion was achieved by repeated administration of DT to Mcpt5-CreiDTR mice) (10). As expected, PCA challenge showed that TNP-specific IgE–sensitized control littermate mice experienced significantly increased blood vessel leakage compared with MC-depleted mice (Fig. 1E). To visualize the consequence of MC activation and vascular leakage at a local site, we examined the ears of TNP-specific IgE–sensitized control littermate and MC-depleted mice after intravenous injection of TNP-OVA along with dextran-TRITC. Whole-mount imaging of the mouse ears showed extensive MC degranulation and vascular leakage in control littermate mice but not in MC-depleted mice (Fig. 1F). When PSA was induced, TNP-specific IgE–sensitized control littermate mice showed a significant drop in body temperature. By contrast, MC-depleted mice exhibited only limited responses with no associated mortality after TNP-OVA challenge (Fig. 1G). Thus, in addition to MCs, CD11c+ cells are important modulators of PCA and PSA in mice.

CD301b+ dermal DCs are critical mediators of anaphylaxis

We next established the identity of the perivascular CD11c+ cells that were collaborating with MCs to mediate both PCA and PSA responses. CD11c is expressed on both DCs and macrophages (11). A recent comprehensive analysis of skin immune cells showed high CD11c expression on the dermal cDC1 subset (previously known as CD8α+ DCs) as well as the cDC2 subset (previously known as CD11bhi dermal DCs). However, low levels of expression are also found on monocyte-derived DCs and some macrophages (1113). The cDC2 subset constitutes the major population of DCs in the dermis. These DCs are defined as CD207CD11bhiCD103EpCAMCD24/loSirp+CD11c+ cells (13, 14) and function to promote the differentiation of Th2 and Th17 cells (15, 16).

Recently, a subset of CD301b-expressing CD11bhi cDC2 was reported to be especially proficient in antigen uptake at the periphery after antigen challenge and in the mediation of subsequent antigen presentation in the draining lymph nodes (DLNs) (1719). Although certain monocyte-derived DCs and even macrophages reportedly share the CD301b marker (11, 13, 20), the subset of CD301b-expressing CD11bhi cells was designated dermal DCs because in addition to being positive for CD301b and CD11c, these cells have the phenotype CD11bhiMHCIIhiCD64loLy6Clo (17, 21). Additionally, they express Zbtb46, a conventional DC (cDC)–specific transcription factor gene expressed by cDCs and their committed progenitors but not by macrophages or monocytes (2224).

In view of their capacity to acquire antigen, we investigated whether the population of CD11c+CD301b+ cells expressed markers characteristic of cDC2 and whether they participated in PCA and PSA. Using transgenic CD301b–green fluorescent protein (GFP)–DTR mice (18), we found that the GFP+CD301b+ cells in the mice had the phenotype CD103CD11bhiSirpa+EpCAMCD24+ and also expressed Zbtb46 (fig. S2). Whole-mount–imaged ears from CD301b-GFP-DTR mice harbored numerous GFP+ cells in the dermal perivascular space adjacent to both the vasculature and MCs (Fig. 2A, left). Moreover, DT injection into these mice selectively and almost completely depleted the GFP+ cells (Fig. 2A, middle and right). Numbers of CD64+CD11c cells (dermal perivascular macrophages) (11, 25) were not altered after the depletion of CD301b+ cells (fig. S3).

Fig. 2 CD301b+ DCs mediate anaphylaxis after uptake of blood-borne antigen.

(A) Treatment of CD301b-DTR-GFP mice with DT results in the depletion of CD301b+ cells. CD301b-DTR-GFP mice were injected intraperitoneally once with DT or vehicle to deplete CD301b+ cells. Depletion of CD301b+ cells was quantified by flow cytometry. The ears of CD301b-DTR-GFP mice were then dissected, and a whole mount was prepared and examined by confocal microscopy. MCs (avidin), red; blood vessels (CD31), blue; and CD301b, green. n = 8 mice per group. Data are presented as the mean ± SEM. *P < 0.001, unpaired Student’s t test. Scale bars, 50 μm. (B and C) CD301b+ cells mediate PCA and PSA. (B) After CD301b+ cell depletion, the ears of CD301b-DTR mice or littermate controls were sensitized with TNP-specific IgE. TNP-OVA was injected intravenously, along with Evans blue dye. After 1 hour, mouse ears were imaged and then dissected to extract dye for OD measurements. n = 5 to 8 mice per group. Data are presented as the mean ± SEM. *P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. (C) After CD301b+ cell depletion, CD301b-DTR or littermate mice were intraperitoneally injected with TNP-OVA, and then temperature changes were monitored (left). Mice with temperature changes greater than 10°C were sacrificed. The survival rate is shown (right). n = 5 mice per group. Data are presented as the mean ± SEM. *P < 0.001, two-way ANOVA (left); **P < 0.01 (survival during PSA was recorded and analyzed by a log-rank test) (right). (D and E) Three-dimensional visualization of CD301b+ DCs probing the vasculature. CD301b-GFP mice were intravenously injected with dextran-TRITC. (D) After 30 min, mouse ears were dissected. Whole mounts were prepared and imaged by using two-photon microscopy. Arrows indicate a lamellipodium-like structure protruding into the vasculature from CD301b+ cDC2. The top panel is an xy-plane projection from a Z-stack view, whereas the lower panels are cross-sectional views of selected xy, yz, and xz planes. Scale bar, 10 μm. (E) Immediately after injection, intravital two-photon microscopy of mouse ears was performed. Time-series events were captured and displayed. White arrowheads point to a dendrite protrusion into the vasculature from CD301b+ cDC2. Yellow arrowheads point to CD301b+ cDC2 taking up blood-borne dextran-TRITC. The probing activity of perivascular CD301b+ cDC2 was quantified from multiple images taken over a 20-min period (right). Scale bars, 10 μm. (F) A substantial population of CD301b+ cDC2 takes up blood-borne antigens. CD301b-GFP mice were intravenously injected with dextran-TRITC. After 30 min, the mouse ears were dissected and processed as a single-cell suspension for flow cytometry in which the dextran-TRITC+ population was identified (left) or processed as whole mounts for microscopy in which antigen-sampling cells were readily detectable (arrows, right). n = 4 to 8 mice per group. Data are presented as the mean ± SEM. **P < 0.01, unpaired Student’s t test. Scale bar, 50 μm. Two to three separate experiments were performed for each individual panel.

To investigate the functional contribution of CD301b+ DCs to anaphylaxis, we performed PCA and PSA challenges in these mice. The depletion of CD301b+ cDC2 markedly reduced vascular leakage in the ears (Fig. 2B) and alleviated the drop in body temperature (Fig. 2C) in CD301b-GFP-DTR mice compared with littermate controls, indicating that CD301b+ cDC2 contributed substantially to the development of anaphylaxis. A small subset of cDC1 (CD103+CD8α+CD11b) is known to be present in the dermis (26). Batf3−/− mice, which lack cDC1, also showed significantly reduced anaphylaxis (~50% of the effect observed after the removal of CD301b+ cDC2) (fig. S4). Thus, this capacity to sample blood-borne antigens is not confined to a particular DC subset.

To address whether this response involved the direct uptake of blood-borne antigens by perivascular CD301b+ cDC2, we injected intravenous dextran-TRITC into CD301b-GFP-DTR transgenic mice. Thirty minutes later, we generated whole-mount preparations of their ears, which were subjected to two-photon microscopy. We could visualize several of the fluorescent CD301b+ cDC2 cells adjacent to blood vessels in the process of taking up dextran-TRITC (Fig. 2D, fig. S5, and movie S1). In some cases, the CD301b+ cDC2 appeared to extrude long lamellipodia with multiple dendrites into the luminal surface of the blood vessel (Fig. 2D, top). Stacked images deployed in x-, y-, and z-axis planes revealed the movement of dextran-TRITC into CD301b+ cDC2. As shown in the yz- or xz-axis cross-sectioned planes, tips of protruding dendrites appeared to colocalize with dextran within the mouse blood vessel lumen (Fig. 2D, bottom). Dextran-TRITC uptake by CD301b+ cDC2 in the ear was also captured by intravital two-photon microscopy (Fig. 2E, fig. S6, and movie S2). Dendrites (indicated by white arrowheads in Fig. 2E) protruding from GFP-labeled CD301b+ cDC2 appeared to penetrate the vasculature within minutes and promptly became labeled with dextran-TRITC (yellow arrowheads in Fig. 2E). Thus, the uptake of blood-borne dextran-TRITC by CD301b+ cDC2 appeared to be a rapid event. Additional intravital confocal microscopy of the ears of CD301b-GFP/Mcpt5-CretdTomato mice was undertaken. In these mice, CD301b+ cDC2 and MCs were fluorescently labeled with GFP and tdTomato, respectively. We were able to confirm that within 15 min of intravenous administration of Alexa 647–conjugated OVA (OVA-A647), CD301b cDC2 but not MCs acquired antigen (fig. S7 and movie S3). To quantify antigen uptake by CD301b+ cells in vivo, we generated a single-cell preparation from the ears of CD301b-GFP-DTR mice 30 min after intravenous injection of dextran-TRITC and examined the preparation by flow cytometry. Up to 30% of CD301b+ cDC2 cells were positive for dextran-TRITC (Fig. 2F and fig. S8), suggesting that antigen uptake by CD301b+ cDC2 occurs quickly, also corresponding to the rapid onset of anaphylaxis.

In vitro activation of IgE-sensitized MCs by antigen-primed DCs

Although MCs are the primary effector cells of PCA and PSA, our data suggest a surprisingly critical role for CD301b+ DCs. These cells appear to be primarily responsible for acquiring antigens from the blood. We hypothesized that if this was true, the role of DCs in promoting anaphylaxis would be in relaying acquired antigens to adjacent IgE-sensitized MCs so as to activate the MCs. Thus, we performed in vitro assays to investigate whether mouse bone marrow–derived CD301b+ DCs (BMDCs) are capable of activating IgE-sensitized MCs after the acquisition of allergen. We first confirmed that our generated BMDCs expressed CD301b but not markers of macrophages such as CD115 (fig. S9). BMDCs were pretreated with TNP-OVA for 20 min and then thoroughly washed to remove unbound antigen from the extracellular medium. Thereafter, the primed DCs were added in increasing number to cultures of the MC line (RBL-2H3) that had previously been sensitized with TNP-OVA–specific IgE (Fig. 3A). We observed strong and dose-dependent MC degranulation, indicating that antigens “presented” without any processing by the DCs were capable of activating MCs. When we replaced DCs with mouse epithelial cells in this assay, we failed to detect any MC activation, indicating that this activity was specific to DCs (Fig. 3B). To confirm that primed DCs have the capacity to activate sensitized MCs in vivo, we intradermally injected antigen-primed BMDCs into DT-treated CD11c-DTR mice. Before reconstitution, MCs from these mice were sensitized with TNP-specific IgE directed at TNP-OVA. As shown in Fig. 3C, intradermal delivery of antigen-primed BMDCs evoked a strong PCA reaction in the mice. This confirmed that antigen-primed DCs can activate sensitized MCs in vivo. We also examined whether human DCs can activate sensitized human MCs after being primed with antigen. We used a human skin–derived DC preparation obtained from skin explants (27) along with human peripheral blood–derived MCs (hPBMCs). The human DC preparation was exposed to streptavidin, and then the primed DCs were coincubated with hPBMCs sensitized with biotinylated IgE. As before, we observed dose-dependent MC degranulation (Fig. 3D), indicating that antigens bound onto human DCs have the capacity to cross-link and degranulate IgE-sensitized human MCs.

Fig. 3 Antigen-primed DCs induce MC degranulation in vitro and in vivo.

(A) Antigen-primed BMDCs induce MC degranulation in a dose-dependent manner. BMDCs were primed with TNP-OVA for 15 min, rinsed three times, and coincubated with TNP-specific IgE–sensitized RBL-2H3 cells for 1 hour, after which β-hexosaminidase release was measured. Data are presented as the mean ± SEM. *P < 0.001, **P < 0.05, one-way ANOVA with Tukey’s multiple comparisons test. (B) Antigen-primed epithelial cells fail to induce MC degranulation. The same procedure as for (A) was followed except that we replaced BMDCs with human bladder epithelial 5637 cells. Data are presented as the mean ± SEM. N.S., not significant by one-way ANOVA with Tukey’s multiple comparisons test. (C) Antigen-primed BMDCs induce vascular leakage in vivo. CD11c+ cells were depleted in CD11c-DTR mice with intraperitoneal and intravenous injections of DT 2 days before challenge with antigen-primed BMDCs. The ears of these mice were sensitized with TNP-specific IgE 1 day before challenge. For the challenge, BMDCs were primed with TNP-OVA or vehicle for 20 min and then washed before they were injected intradermally into the ears of the IgE-sensitized mice. Evans blue dye was simultaneously injected intravenously into the same mice. After 1 hour, the mouse ears were imaged and dissected to extract dye for OD measurements. n = 4 to 5 mice per group. Data are presented as the mean ± SEM. **P < 0.05, unpaired Student’s t test. (D) Antigen-primed primary human skin DCs induce the degranulation of primary human MCs. Primary human MCs (Pr hMCs) cultured from peripheral blood were sensitized with biotinylated human IgE overnight. Primary human skin DCs (Pr hDCs) were isolated from skin tissue as described in Materials and methods, primed with streptavidin (StpAv) for 20 min, and rinsed three times. MCs and DCs were coincubated for 1.5 hours. β-Hexosaminidase released into the extracellular medium was then measured. No β-hexosaminidase release was detected in control cultures containing only DCs. Data are presented as the mean ± SEM. *P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. Neg, negative control; Ag, antigen. Two to three separate experiments were performed for each individual panel.

DCs activate MCs by discharging microvesicles with surface-bound allergens

Next, we sought to investigate how antigen-primed DCs subsequently relay antigen to MCs. Live confocal imaging of blood-borne antigen uptake by perivascular DCs in CD301b-GFP/Mcpt5-CretdTomato mice revealed the trafficking of “allergen particles” (~1 μm in diameter) between different perivascular CD301b+ cDC2 cells and between CD301b+ cDC2 and MCs (Fig. 4A, fig. S10, and movie S4). This raised the possibility that antigen-primed DCs shed antigen-bearing microvesicles (MVs), which traffic to neighboring MCs and DCs. Because the secretion of extracellular vesicles occurred relatively quickly (~15 min), these vesicles were unlikely to be multivesicular body–generated exosomes, which require markedly more time for their generation (28). We transiently exposed a readily transfectable DC line (29) to OVA-A647 to examine whether DCs have the capacity to shed antigen-bearing MVs after exposure to antigen. Consistent with our hypothesis, we observed that within minutes of washing off of unbound antigen, ~1-μm MVs budded off from the plasma membrane (Fig. 4B, top; fig. S11A; and movie S5). Shed MVs traveled randomly in the extracellular medium and bound neighboring immune cells (Fig. 4B, bottom; figs. S11B and S12; and movies S6 and S7). To examine whether these MVs were capable of relaying antigens, we harvested shed MVs from OVA–fluorescein isothiocyanate (FITC)–treated DCs by differential ultracentrifugation and viewed the MVs by microscopy. These MVs contained fluorescent OVA, and most of this OVA appeared to be localized on the surfaces of the vesicles (Fig. 4C).

Fig. 4 Release of MVs by DCs triggers degranulation in neighboring MCs.

(A) Time-lapse release of antigen-loaded vesicles from perivascular CD301b+ cDC2 to neighboring MCs and DCs in vivo. CD301b-GFP/Mcpt5-CretdTomato mice were injected intravenously with OVA-A647. The ears of live mice were then imaged by using intravital confocal microscopy. MCs, blue; CD301b cDC2, green; A647-OVA, red. The arrowhead indicates the transfer of an antigen-bearing vesicle from a CD301b+ DC (green) to an MC (blue). The arrow denotes the transfer of antigen-bearing vesicles from a CD301b+ DC to another CD301b+ DC. Quantification of fluorescent particles was achieved by viewing multiple images and fields. The movement of particles from DCs to MCs (DC→MC) and from DCs to other DCs (DC→DC) for a 25-min period was quantitated (right). Scale bars, 10 μm. (B) MV formation and secretion from a DC line (JAWSII). After the DC line was primed with antigen OVA-FITC (red), the cells were imaged by using differential interference contrast (DIC). Arrowheads denote the formation and release of antigen-loaded MVs. The budding MVs or their migration over a 20-min period was quantified by viewing multiple images and fields (right). Scale bars, 10 μm. (C) Fluorescent antigens are enriched on the membranes of recently formed MVs. After the DC line was primed with OVA-FITC (green), MVs released from these cells were collected through differential centrifugation and subjected to confocal microscopy. Scale bar, 1 μm. (D) TNP-OVA–loaded MVs induce MC degranulation. After BMDCs were treated with TNP-OVA for 15 min, the MVs were collected and MV numbers were calculated as described in Materials and methods. Collected MVs were applied to IgE-sensitized BMMCs in a dose-dependent manner (twofold dilution) for 1 hour, after which a β-hexosaminidase release assay was performed. Data are presented as the mean ± SEM. *P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test. (E) Flow cytometric analysis of OVA-FITC–loaded MVs. BMDCs were primed with OVA-FITC for 15 min, and after thorough washing to remove unbound OVA-FITC, fresh medium was added and the cells were incubated for 2 hours to release MVs. These particles were collected by differential centrifugation and analyzed by flow cytometry. Gating on 0.5- to 1.0-μm MVs was performed as described in the legend to fig. S14. Pseudocolored dot plots of OVA-FITC–treated or vehicle-treated MVs were displayed with FITC on the x axis and side scatter (SSC) on the y axis. (F) Bar graphs demonstrating quantification of the FITC+ population (left) or total MVs (right) for OVA-FITC–treated or vehicle-treated MVs. Data are presented as the mean ± SEM. **P < 0.05, unpaired Student’s t test. Two to three separate experiments were performed for each individual panel.

Because antigens appeared to be loaded on the surfaces of MVs, isolated MVs from antigen-primed DCs may have the potential to activate MCs. Thus, we examined whether MVs harvested from TNP-OVA–treated BMDCs could induce the degranulation of TNP-specific IgE–sensitized MCs. MVs isolated from TNP-OVA–treated BMDCs by differential ultracentrifugation as described previously (30) were capable of activating bone marrow–derived MCs (BMMCs) in a dose-dependent manner (Fig. 4D). Notably, MVs from vehicle-treated DCs failed to activate MCs (fig. S13).

We developed a flow cytometric assay to quantitate DC-derived MVs. First, we established a gating area that corresponded to the reported size of MVs (31) by using latex beads ranging from 0.5 to 1.0 μm [x axis, forward scatter–height (FCS-H); y axis, forward scatter–width (FCS-W)] (fig. S14). Thereafter, we quantitated MVs generated by in vitro–cultured primary BMDCs (Fig. 4E) or by a DC line after exposure to fluorescent OVA-FITC (fig. S14). In each case, two distinct populations of MVs were shed by OVA-FITC–primed BMDCs. One population of MVs (~14%) was positive for FITC, whereas the remainder were FITC negative (Fig. 4F), indicating that MV production by DCs was not dependent on bound antigen. As the total number of MVs generated by vehicle-treated BMDCs during the same period of time was comparable to the total number of MVs generated after exposure to antigen (Fig. 4F), MV shedding by DCs appears to be a constitutive activity and not antigen induced.

Molecular determinants mediating DC shedding of MVs and acquisition of antigen

We investigated the molecular mechanism underlying antigen sharing and antigen acquisition by DCs. Specifically, we were interested in identifying critical cellular components mediating the shedding of antigen-bearing MVs, as well as components involved in the initial acquisition of antigen. As MCs become activated within minutes of the allergen’s entering the blood, the time for DCs to acquire and then present antigens to MCs is short. It is unlikely that there would be sufficient time for DCs to endocytose antigens and then export these antigens as exosomes within vesicles originating from multivesicular bodies (28). Even if this did occur, it is unlikely that the internalized antigen would still be bound on the outer surface of the vesicle membrane, which is necessary to activate MCs. Immune cells have the innate capacity to outwardly bud off plasma membrane fragments in the form of 0.2- to 1.0-μm-diameter MVs (32). Although these MVs lack cytosolic organelles and nuclear fragments (33), they are selectively enriched in various plasma membrane proteins, including major histocompatibility complex class I, β1 integrin receptors, and vesicle-associated membrane protein 3 (VAMP3) (30, 34), and presumably, any molecules that these receptors are bound to when budding occurs. Mechanistically, outward membrane budding requires the endosomal sorting complexes required for transport (ESCRT) machinery to assemble a spiral structure at the neck of the budding vesicle, which promotes MV budding (32). This budding event is specifically catalyzed by vacuolar protein sorting–associated protein 4 (VPS4) through its specific AAA-type adenosine triphosphatase activity (35). We investigated whether VPS4 was involved in relaying allergens from DCs to MCs by using both microscopy and flow cytometry to examine MV shedding from antigen-primed DCs after their VPS4A expression was markedly reduced [by transfection with VPS4A small interfering RNA (siRNA)] (fig. S15) compared with that of control DCs [transfected with nonspecific (NS) siRNA] (Fig. 5, A and B). To visualize MV shedding, we stained DCs with fluorescent CellTracker CM-DiI dye, which selectively accumulates in the plasma membrane of the cell. We observed many fluorescent MVs shed from control antigen-primed DCs (Fig. 5A, top, and movie S8). By contrast, MV secretion was completely abrogated in VPS4A-silenced DCs (Fig. 5A, bottom, and movie S9). The surfaces of these cells appeared to be decorated with numerous vesicles in the process of budding but incapable of being released, consistent with the role of VPS4A in vesicle scission. We complemented these observations by quantitating MV shedding in control DCs (NS siRNA transfected) and an OVA-FITC–primed murine DC line with VPS4A silenced (VPS4A siRNA transfected) by using flow cytometry. We observed that the population of MVs from VPS4A-silenced DCs was reduced by up to 90% compared with the controls (Fig. 5B). Thus, the rapid relay of allergens by DCs after uptake was achieved via the budding of antigen-bound MVs from the plasma membrane in a VPS4A-dependent manner. To further support these observations, we sought to reproduce some of these observations by using primary DCs in our in vitro and in vivo studies. Using CRISPR-Cas9 technology (36), we knocked out Vps4a in primary DCs (figs. S16 and S17). These Vps4a−/− BMDCs or mock-transfected BMDCs were then exposed to OVA-FITC antigen and processed for transmission electron microscopy (TEM). Cross sections of these DCs revealed numerous 0.5- to 1-μm extracellular MVs in close association with the mock-transfected BMDCs (Fig. 5C and fig. S18). By contrast, Vps4a−/− BMDCs exhibited a limited capacity to shed MVs (Fig. 5C). Instead, numerous vesicular structures incapable of detaching were observed to form on the plasma membrane of these BMDCs (Fig. 5C), consistent with the phenotype observed via live imaging (Fig. 5A). Next, we sought to demonstrate the in vivo contribution of MVs emanating from DCs by comparing cutaneous anaphylaxis reactions in CD11c+ cell–depleted mice reconstituted with Vps4a−/− BMDCs or mock-transfected BMDCs. Compared with the extensive cutaneous anaphylaxis observed in the mice reconstituted with mock-transfected BMDCs, the anaphylaxis in mice with Vps4a−/− BMDCs was greatly limited (Fig. 5D). These in vivo studies support our in vitro observations and strongly suggest that VPS4A-mediated MV secretion from DCs relays blood-borne antigen to MCs, leading to anaphylaxis.

Fig. 5 MV shedding by DCs is VPS4 dependent, and antigens borne by MVs are bound by the MR.

(A) VPS4 mediates MV shedding. MV shedding by DC lines transfected with VPS4A siRNA (bottom rows) or NS siRNA (top rows) was monitored. Forty-eight hours after transfection, the DCs were stained with CM-DiI dye (red), which stains membranes, and then exposed to antigen (OVA) and examined by time-lapse microscopy. In each group, the upper panels are fluorescence images and the lower panels are the corresponding DIC images. The arrowheads indicate sites of MV shedding. Representative enlarged images of the regions indicated by dashed squares demonstrate the shedding of MVs (top right) and the inhibition of shedding of formed vesicles (bottom right). Scale bars, 10 μm (time-lapse images) and 5 μm (magnified images). (B) MV shedding requires the VPS4A subunit. MVs produced by control and VPS4A-knockdown DC lines were quantitated by flow cytometry. Data are presented as the mean ± SEM. **P < 0.05, unpaired Student’s t test. (C) Ultrastructure of MV formation by BMDCs. CRISPR-Cas9–generated Vps4a/ BMDCs or mock-transfected BMDCs were pretreated with OVA for 15 min and immediately fixed with 4% PFA and processed for TEM. MVs shed by mock-transfected BMDCs or CRISPR-Cas9 Vps4a−/− BMDCs were quantified by viewing multiple images and fields (right). Scale bars, 500 nm. Data are presented as the mean ± SEM. *P < 0.01, unpaired Student’s t test. (D) MC-dependent anaphylaxis requires the secretion of MVs from nearby DCs. CD11c cell–depleted mice (CD11c-DTR) were reconstituted with CRISPR-Cas9–generated Vps4a−/− BMDCs or mock-transfected BMDCs. (Confirmation of the specificity of Vps4a−/− is shown in figs. S16 and S17; the successful reconstitution of BMDCs is depicted in fig. S19.) MCs in the reconstituted mice or littermate control mice were sensitized by intradermal injection of TNP-specific IgE antibody. On the following day, mice were intravenously challenged with TNP-OVA, along with Evans blue dye. After 30 min, the ears of mice were imaged (top) and dissected for dye extraction and quantitation as shown in the bar graph (bottom). n = 4 to 7 mice per group. Data are presented as the mean ± SEM. **P < 0.05, one-way ANOVA with Tukey’s multiple comparisons test. (E) The MR mediates OVA antigen acquisition by DCs. The DC line was preincubated with various concentrations of mannan (1, 5, and 10 mg/ml) and then primed with TNP-OVA (0.1 μg/ml). This cell line was then added to IgE-sensitized BMMCs, and MC degranulation was assessed. Data are presented as the mean ± SEM. *P < 0.01, **P < 0.05, one-way ANOVA with Tukey’s multiple comparisons test. (F) The MR on DCs is responsible for binding OVA antigen. A DC line transfected with MR siRNA or NS siRNA was primed with TNP-OVA (0.1 μg/ml) and exposed to IgE-sensitized BMMCs. The MC degranulation response was then evaluated. Data are presented as the mean ± SEM. *P < 0.01, one-way ANOVA with Tukey’s multiple comparisons test. Two to three separate experiments were performed for each individual panel.

Lastly, we investigated how antigens were bound on the surfaces of MVs. Several receptors identified on the surfaces of DCs are capable of binding a wide range of antigens. These promiscuous receptors include scavenger and mannose receptors, which could potentially bind the OVA antigen used in this study (37, 38). The mannose receptor (MR) on DCs, which binds a wide range of mannosylated compounds, has previously been shown to bind OVA (39). To investigate the possible role of the MR in antigen uptake, we examined whether the pretreatment of mouse DCs with mannan, a linear polymer of mannose, would competitively block the binding of antigen (OVA-TNP), resulting in a loss of the ability of DCs to trigger MC degranulation. Mannan pretreatment significantly blocked the binding of antigen (OVA-TNP), as these cells subsequently failed to induce MC degranulation in a dose-dependent manner (Fig. 5E). To confirm that MRs on DCs were responsible for binding OVA-TNP, we knocked down the expression of MR in rodent DCs by using siRNA transfection (fig. S20). This significantly reduced the capacity of DCs to activate MCs after being primed with antigen (Fig. 5F). Immunoblots of isolated MVs generated by these DCs also revealed the presence of MR, confirming the role of MR in antigen transfer (fig. S21). Thus, the MRs on DCs and MVs are important for the acquisition of OVA antigens, which are subsequently presented to MCs via MVs.


After the entry of allergens into the circulation, MCs are the primary effectors of the ensuing rapid and systemic allergic reactions. This is attributable, at least in part, to the large number of allergen-specific IgE molecules coating MCs. These cells prestore a panoply of powerful inflammatory mediators, which, upon release, can work in concert to trigger various symptoms of anaphylaxis, including vascular leakage, itchy rash, swelling of afflicted tissue, and a drop in blood pressure and body temperature. MCs can directly probe blood contents with protoplasmic protrusions to acquire circulatory IgE (5). This behavior was initially hypothesized to explain how these perivascular cells became so quickly activated after the entry of allergens into the circulation. However, we have shown that only a minority of MCs acquire blood-borne allergens in this manner. Rather, most perivascular MCs are activated in an indirect fashion by CD301b+ dermal DCs that are continuously probing blood contents. Live and fixed-tissue microscopy of the mouse ear vasculature revealed that blood-borne antigens were adsorbed by lamellipodia of CD301b+ cells that protrude through the endothelial wall into blood vessels. After the acquisition of antigens by the MR, CD301b+ DCs spontaneously shed antigen-bound MVs to neighboring IgE-bearing MCs in the perivascular region. Upon degranulation, the many granule-associated MC mediators triggered both vascular leakage and a drop in body temperature. The functional importance of both MCs and CD301b+ DCs to anaphylaxis was demonstrated by the abrogation of anaphylaxis when either one of these cell populations was depleted in mice. Although CD301b+ cDC2 has previously been reported to predominate in the dermis (14, 17, 18), we have noticed the preferential location of this subset in the abluminal surface of the dermal vasculature. We do not exclude a role for other DC subsets (26, 40) in anaphylaxis because CD103+CD8α+CD11b cDC1 also contributes to this process, albeit to a lesser extent than CD301b+ cDC2.

That CD301b+ DCs and not MCs were primarily responsible for probing the vasculature for allergens is not surprising, as DCs are the professional immune sampling cells in the body. Subepithelial DCs have been implicated in sampling various body sites that are exposed to the environment, such as the airways, gut, and skin (68). Many environmental antigens can enter the circulation either when the vasculature is ruptured or when antigens, including microorganisms, are introduced into the blood (e.g., via insect bites) (41, 42). Thus, there is a need to continuously sample the blood for these extrinsic agents. Dermal CD301b+ DCs appear to perform this vital function. Although perivascular macrophages also appear to be capable of transendothelial sampling many hours after exposure to blood-borne antigens (25, 43), their functional relevance, particularly in the context of anaphylaxis, remains unknown. It is currently unclear how allergic reactions in the skin and mucosal surfaces are triggered. As MCs are also found in the same subepithelial region where DCs probing these locations lie, it is conceivable that environmental allergens are also initially acquired by DCs and then delivered to adjacent sensitized MCs to trigger local allergic reactions. This could be the explanation for the observation made more than a decade ago showing that the depletion of lung CD11c+ cells during allergen challenge abrogates the characteristic features of asthma in mice (44).

Antigen-presenting cells, including dermal CD301b+ cells, ingest extrinsic antigens in the periphery and then traffic to proximal DLNs, where they present processed antigens to T cells, resulting in the development of antigen-specific immune responses (15). Here, we report a distinct form of antigen presentation by the CD301b+ DC subset, where antigens adsorbed onto the DC surfaces are promptly relayed to neighboring MCs in the skin to cause systemic anaphylaxis. Although some of the antigens obtained by DCs were internalized, a substantial amount remained on the plasma membrane and were released to the extracellular medium associated with MVs. These vesicles are the products of specific budding and pinching-off activities occurring on the plasma membranes of DCs, because the secretion of MVs was abrogated when VPS4, which mediates the scission of MVs from the DC-derived plasma membrane, was specifically knocked down. We confirmed the functional role played by MVs in anaphylaxis by showing the attenuated ability of Vps4a-deficient primary DCs to induce cutaneous vascular leakage. Because MCs are activated within minutes of allergens entering the blood, there is unlikely to be sufficient time for the internalization of antigens and for any packaging of antigens in vesicles analogous to exosomes. Rather, the budding of membrane vesicles containing surface-bound antigens appears to be a possible mechanism for the rapid transfer of these antigens.

During immune surveillance, DCs regularly utilize the broad binding specificity of the cell surface MR to capture and clear a wide range of extrinsic and endogenous antigens that the DCs encounter. Although antigens bound by MR can be endocytosed (37), we show here that at least some OVA antigens bound by MRs are held onto the plasma membrane and subsequently presented to neighboring immune cells via MVs. The factors that determine the endocytosis of antigens by the MR or redistribution to neighboring cells via MVs remain unclear.

Our findings provide a mechanism for how blood-borne allergens trigger tissue MC degranulation through allergen acquisition from the circulation by perivascular CD301b+ cells and direct transfer to MCs through MVs. The capacity of CD301b+ DCs to sample extrinsic antigens in the dermis and traffic to the DLNs (15, 18) suggests that CD301b+ DCs are key mediators of tissue MC degranulation in response to blood-borne allergens, in addition to playing a recognized role as migratory DCs capable of presenting processed antigen in DLNs. The ability of CD301b+ DCs to acquire blood-borne extrinsic antigens and relay them to neighboring MCs and other DCs may be a powerful but underappreciated mechanism for enhancing the magnitude of inflammatory and immune responses to blood-borne antigens. As effective treatments for anaphylaxis remain elusive, recognizing the role of CD301b+ DCs in allergen sampling may serve as a target for future therapeutic strategies.

Materials and methods

Mouse strains

Eight- to twelve-week-old female and male mice were used for most of these studies. The following mouse strains were used: C57BL/6 WT, CD11c-DTR-GFP, CD301b-DTR-GFP, iDTR, AI14, and Batf3−/− mice (Jackson Laboratories) and Mcpt5-Cre+ mice (a gift from A. Roers, University of Technology, Dresden, Germany). CD11c+ cells were depleted as follows: 8- to 12-week-old CD11c-DTR-GFP mice and their littermates were given intraperitoneal and intravenous injections of 500 ng of DT per mouse every other day, twice. To deplete CD11c+ cells in ear skin, CD11c-DTR mice were administered DT at 500 ng intraperitoneally and 50 ng intradermally in both ears. Two days after DT injection, 1 × 106 BMDCs, which were electroporated with a Vps4a-specific CRISPR-Cas9–expressing plasmid, mock plasmid (PX458), or control vector pmaxGFP (Lonza), were intradermally injected into the ears. CD301b+ cells were depleted as follows: 8- to 12-week-old CD301b-DTR-GFP mice and their littermates were administered intraperitoneal and intravenous injections of 250 ng of DT per mouse every other day for 1 week or intraperitoneal and intravenous injections of 500 ng of DT per mouse once. MCs in Mcpt5-CreiDTR mice were conditionally depleted as follows: 8-week-old Mcpt5-CreiDTR mice and Mcpt5-Cre/+ littermates were given five intravenous injections of 200 ng of DT per mouse within 1 week. Mcpt5-CreAI14 mice and CD301b-GFP mice were crossed to generate Mcpt5-CreAI14:CD301b-GFP mice for intravital imaging. All procedures related to mice were performed in strict accordance with the animal protocol approved by the of Duke University Institutional Animal Care and Use Committee.

Cell culture

For BMMC culture, WT bone marrow was obtained from WT C57BL/6 mice in ice-cold Hanks’ balanced salt solution (HBSS) (Gibco) and cultured in complete RPMI containing 10% fetal bovine serum (FBS) (Hyclone), 1 mM nonessential amino acids, 25 mM HEPES [4-(2-hydroxyethyl)-1-piperazineethane-sulfonic acid], 2 mM l-glutamine, 1 mM sodium pyruvate, 1× Antibiotic-Antimycotic solution (all reagents from Gibco), 10-ng/ml stem cell factor (SCF), and 5-ng/ml interleukin-3 (IL-3) (BioLegend) for 8 to 12 weeks. For BMDC culture, bone marrow was obtained from WT C57BL/6 or CD301b-DTR-GFP mice and cultured in RPMI containing 10% FBS, 1× GlutaMAX (Gibco), and 20-ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF) (BioLegend) or 20-ng/ml GM-CSF and 50-ng/ml IL-4 for 6 to 7 days. For bone marrow–derived macrophage (BMM) culture, the same procedures were followed but with 20-ng/ml macrophage colony-stimulating factor (M-CSF) replacing the growth factors. The rat MC RBL-2H3 cells (ATCC) were cultured in minimum essential medium (MEM) (Gibco) containing 15% FBS and antibiotics. JAWSII cells (ATCC) were maintained in MEMα (Gibco) supplemented with 20% FBS, 1 mM sodium pyruvate, 100-U/ml penicillin, and 100-μg/ml streptomycin. All cells were cultured at 37°C in a humidified water-jacketed incubator under a 5% CO2–95% air atmosphere.

Primary human MC culture

Detailed procedures for primary human MC culture have been explained in detail elsewhere (45). For the culture of primary MCs, Iscove’s modified Dulbecco’s medium (IMDM) (Gibco) containing 200 μM bovine serum albumin (BSA) (Sigma), 1× insulin-transferrin-selenium (ITS) supplement (Gibco), 75 μM β-mercaptoethanol (Sigma), 100-U/ml penicillin, and 100-μg/ml streptomycin was used. In brief, purified CD34+ peripheral blood cells (Allcells) were suspended at 5 × 105 cells/ml in supplemented IMDM containing 100-ng/ml recombinant human SCF (rhSCF), 10-ng/ml recombinant human IL-6 (rhIL-6), and 1-ng/ml rhIL-3 for 0 to 3 weeks. For the subsequent 3 to 6 weeks, the cells were suspended at 5 × 105 cells/ml in supplemented IMDM containing 100-ng/ml rhSCF and 50-ng/ml rhIL-6. After 6 weeks of differentiation, cells were finally suspended at 5 × 105 cells/ml in supplemented IMDM containing 100-ng/ml rhSCF, 50-ng/ml rhIL-6, and 10% FBS. After differentiation of primary human MCs, we evaluated the maturation of primary human MCs via flow cytometry (>95% were double positive for c-kit and FcεRI) and toluidine blue staining (dark purple staining).

β-Hexosaminidase assay

The day before the experiment, RBL-2H3 cells or BMMCs were sensitized with TNP-specific IgE antibody (BD Biosciences). For IgE antigen stimulation, JAWSII cells or BMDCs were pretreated with TNP-OVA (Biosearch Technologies) for 15 min and then thoroughly washed three times with phosphate-buffered saline (PBS). DCs and MCs were then coincubated for 1 hour, and supernatant was collected for β-hexosaminidase release assay. Briefly, collected supernatant was incubated with p-nitrophenyl-N-acetyl-β-d-glucosaminide (3.4 mg/ml) dissolved in 0.1 M citrate buffer (pH 4.5) for 1 hour at 37°C. Carbonate buffer (0.1 M, pH 10) was then applied to the reaction wells to stop the reaction and develop color. The colorimetric measurement was performed at 405 nm by using a microplate reader.


To elicit PCA, mice were injected intradermally in the ears with TNP-specific IgE antibody (100 ng/20 μl of PBS) (BD Biosciences). After 16 hours, mice were injected intravenously with 200 μg of TNP-OVA diluted in saline. For some experiments, 0.5% Evans blue dye or 1 mg of dextran-TRITC was injected along with 200 μg of TNP-OVA solution. To elicit PSA, mice were first sensitized with TNP-specific IgE (20 μg in saline) (BD Biosciences) by intravenous injection. After 24 hours, mice were injected intravenously with 2 μg of TNP-OVA. After antigen challenge, body temperature was measured with a rectal microprobe thermometer every 10 min to monitor systemic anaphylaxis.

Skin explant preparation and culture

Detailed procedures for skin explant preparation and culture have been described previously (27). Skin explants were obtained from biopsy specimens. Cut samples of approximately 1 cm by 1 cm by 0.5 cm were placed in 1 ml of Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) containing 10% FBS. Two to three days after culture, migrated cells were harvested and used for further experiments. Human skin specimens were collected from healthy patients undergoing plastic surgery at Duke University Medical Center and used in an anonymized fashion. All human samples for this study were obtained according to the protocols approved by the Institutional Review Board at Duke University.

Flow cytometry

To produce single-cell suspensions for flow cytometry, dissected mouse ears were split open and the only dermis side lacking cartilage was digested with a digestion buffer (15% collagenase, 1% deoxyribonuclease I, 10 mM HEPES, and 1.5% FBS in HBSS). Single-cell suspensions were washed and stained with the fluorescently labeled antibodies against CD11b (clone M1/70), CD11c (clone N418), CD24 (clone M1/69), CD45 (clone 30-F11), CD64 (clone X54-5/7.1), CD103 (2E7), CD115 (clone AFS98), CD117 (clone 2B8), B220 (clone RA3-6B2), IA/IE (clone M5/114.15.2), FcεRI (clone MAR-1), Sirpα (clone P84), EpCAM (clone G8.8), and isotype controls (all antibodies from BioLegend) for 30 min at 4°C. These antibodies were used to compare expression levels. Flow cytometry was performed on a FACSCanto II or LSR II (BD) and was analyzed by using FlowJo software (Tree Star, Ashland, OR).

Vps4a-specific CRISPR-Cas9

To delete Vps4a in BMDCs and JAWSII cells, we used a genome-editing approach targeting the endogenous Vps4a gene through clustered regularly interspaced short palindromic repeats (CRISPR) and the endonuclease Cas9 (36). A CRISPR single guide RNA (sgRNA) targeting exon 2 of the Vps4a gene was designed by using the CRISPR design database (46) and was cloned into the pSpCas9(BB)-2A-GFP (PX458) plasmid (Addgene). The oligonucleotide sequence for Vps4a sgRNAs was GGATAAAGCCAAGAACTACG [20–nt (nucleotide) sequence, 5′ to 3′]–AGG (protospacer adjacent motif). The CRISPR-Cas9 plasmid (PX458) with guide RNA targeting Vps4a exon 2 was electroporated with Amaxa Nucleofector (Lonza) into JAWSII cells or BMDCs. Forty-eight hours after electroporation into the JAWSII cells, the efficiency in generating site-specific double-strand breaks was measured through the SURVEYOR mutation detection assay (Integrated DNA Technologies) from purified genomic DNA derived from the JAWSII cells. After confirmation of the construction of CRISPR-Cas9 with sgRNA targeting Vps4a, BMDCs, which were matured with GM-CSF (20 ng/ml) and IL-4 (50 ng/ml) for 6 days, were electroporated with the constructed plasmid. Forty-eight hours after electroporation, immunoblotting with anti-VPS4A antibody (Santa Cruz Biotechnology) was performed to confirm the knockout of endogenous VPS4A protein.


BMDCs (2 × 106) were resuspended into 100 μl of Nucleofactor solution V (Lonza), mixed with Vps4a-specific CRIPSR-Cas9 or mock plasmid, and electroporated by using Amaxa Nucleofector (Lonza) with program Y-001. Electroporated cells were cultured in medium with 20-ng/ml GM-CSF and 50-ng/ml IL-4 for 48 or 72 hours.

Immunofluorescent staining and microscopy

For coculture visualization of MCs and DCs, BMMCs and BMDCs were fixed in 4% paraformaldehyde (PFA) and permeabilized in 0.1% saponin (Sigma) per 1% BSA in PBS. For whole-mount ear visualization, ears were excised and split open and immediately fixed in 4% PFA at room temperature for 2 hours. To block and permeabilize the tissue, buffer containing 0.3% Trion X-100 and 2.5% normal goat serum (Gibco) in 1% BSA–PBS was used.

To visualize blood vessels, 0.5-μg/ml rat anti-mouse CD31 antibody (clone MEC 13.3; BD Biosciences) was incubated with mouse ears overnight. A647-conjugated F(ab′)2 fragment anti-rat IgG (Jackson ImmunoResearch) was used as a secondary antibody. To visualize MCs, FITC-avidin (BD Biosciences) was used. The samples were visualized by confocal microscopy or two-photon microscopy.

For live imaging of MV secretion from DCs, JAWSII cells were grown on MatTek plates (MatTek Corporation) and treated with CellTracker CM-DiI dye (Molecular Probes) for 20 min in serum-free MEMα at 37°C and 5% CO2 with humidification. After thorough washing with PBS, the prestained cells were treated with OVA or PBS for 15 min in culture medium. After thorough washing with PBS, cells were maintained at 37°C while a Nikon ECLIPSE TE200 confocal microscope equipped with a 100×/1.49NA oil immersion objective lens was used to capture live moments for 15 min.

For intravital imaging by confocal microscopy, Mcpt5-CreAI14:CD301b-GFP mice were generated as described above. Mice were anesthetized by using a mixture of ketamine and xylaxine and kept on a heating pad maintained at 38°C. The inner sides of mouse ears were taped onto a glass slide and immobilized with tape. One hundred microliters of 5-mg/ml A647-OVA (Molecular Probes) was injected intravenously, and ears were immediately imaged by a Nikon ECLIPSE TE200 confocal microscope equipped with a 20×/0.75NA multi-immersion objective lens.

For intravital imaging by two-photon microscopy, CD301b-eGFP-DTR mice were anesthetized with a mixture of isoflurane and O2 supplied through an isoflurane vaporizer to a mask covering the snout. To prevent hypothermia of mice during intravital imaging, the imaging platform was heated and maintained at 38°C. Ear pinnae were taped onto the imaging platform to secure their position and minimize movement during imaging. Immediately after the injection of 200 μl of 10-mg/ml dextran-TRITC (150 kDa; Molecular Probes), time-lapse Z-stack imaging was initiated by using an Olympus FV1000 multiphoton upright microscope, characterizing in vivo movement of CD301b cDC2 near blood vessels at 30-s intervals for a total of 30 min. An Olympus FV1000 multiphoton microscope equipped with a 25×/1.05 NA water immersion objective lens was used for two-photon microscopy. Collected series of images were reconstructed in three dimensions by using Imaris software (Bitplane).

Preparation of MVs

JAWSII cells or BMDCs plated on 100-cm2 culture dishes were treated with antigens for 15 min in culture medium and then thoroughly washed with PBS to clear unbound antigens. After culture medium was resupplied, cells were incubated under culture conditions for 2 hours. Collected culture supernatant was spun at 400 × g for 5 min, and the resulting supernatant was re-spun under the same conditions to remove cell debris. Thereafter, a final high-speed spin at 100,000 × g for 30 min was performed. An Optima L-90K ultracentrifuge (Beckman Coulter) was used for centrifugation. The collected MV pellet was resuspended in PBS and analyzed by flow cytometry, where particles ranging in size between 0.5 and 1.0 μm were counted. In our studies, we observed that DCs cultured in 100-cm2 culture dishes with 70% confluency typically generated ~1 × 106 MVs.


Samples were mixed with radio immunoprecipitation assay buffer and then vortexed for 30 s, three times. After centrifugation at 13,000 × g, the lysates were mixed with sample buffer and separated on 4 to 20% SDS–polyacrylamide gel electrophoresis gels. Proteins on gels were transferred onto polyvinylidene difluoride membranes (Bio-Rad) and blocked with 3% BSA in Tris-buffered saline with Tween-20 for 2 hours. Membranes were incubated with mouse anti-VPS4A monoclonal antibody (clone A-11, 0.2 μg/ml; Santa Cruz Biotechnology) or rabbit anti-MR1 polyclonal antibody (1 μg/ml; R&D Systems) overnight at 4°C. Anti-mouse IgG antibody (1:2500; Bio-Rad) or anti-rabbit IgG antibody (1:1000; Bio-Rad) conjugated with horseradish peroxidase was used as a secondary antibody.

siRNA sequences

siRNA sequences were as follows: for VPS4A, 5′-GUGGAAUGAUGUAGCUGGA[dT][dT]-3′, and for MR, C type 1, 5′-CUCCUUACUGGGCAAUGCA[dT][dT]-3′.

Statistical analyses

Statistical analyses were performed by using GraphPad Prism v.6 (GraphPad Software). Unpaired Student’s t test, two-way analysis of variance (ANOVA), and one-way ANOVA with Tukey’s multiple comparisons tests were used to calculate statistical significance. Log-rank tests were used to analyze survival during PSA. A P value of <0.05 was considered statistically significant. Data are presented as the mean ± SEM.

Supplementary Materials

References and Notes

Acknowledgments: We thank A. Roers for the Mcpt5-Cre mice. B. Hayes is gratefully acknowledged for critical review of this manuscript. We thank Y. Jiao and B. Chen for assistance with two-photon microscopy. We thank R. Vancini and S. Miller for assistance with electron microscopy. We gratefully acknowledge the Duke Cancer Institute and Duke Human Vaccine Institute Flow Cytometry Facility for the use of FACSCanto II and LSR II. Funding: This work was funded by U.S. National Institutes of Health grants (U01-AI082107, R01-AI096305, and R56-DK095198). A.S.M. receives funding from U.S. National Institutes of Health grant R21-AI128727, the Duke Physician-Scientist Strong Start Award, the Duke University Medical Center Department of Dermatology, and the Dermatology Foundation. Author contributions: H.W.C., J.S., I.H.K., M.H., A.S.M., and S.N.A. designed experiments. H.W.C and J.S. performed the experiments. H.W.C., J.S., I.H.K., H.F.S., A.S.M., and S.N.A. contributed to data analysis. H.W.C., M.H., A.S.M., and S.N.A. wrote the manuscript. All authors discussed the results and commented on the manuscript. Competing interests: The authors declare no competing interests associated with this work. Data and materials availability: Plasmids and reagents described in this study are available to the scientific community upon request to S.N.A. All data are available in the main text or the supplementary materials.

Stay Connected to Science

Navigate This Article