Anti-human MRGPRX2 antibody (cat no. 359002) and anti-HA.11 Epitope Tag antibody (cat no. 901513) were from Biolegend (San Diego, CA). Anti-syntaxin 3 antibody (cat no. ab133750), secondary Alexa Fluor^® 488 Goat Anti-Rabbit IgG H&L (cat no. ab150077, dilution 1:500) and secondary Alexa Fluor^® 647 Goat Anti-Mouse IgG H&L (cat no. ab150115) were from Abcam (Cambridge, UK). Substance P (cat no. S6883), pitstop2 (cat no. SML1169) and saponin (cat no. S4521-25G) were purchased from Sigma-Aldrich (St Louis, MO, USA). Cytochalasin D (cat no. 1233), dynasore (cat no. 2897) and EIPA (cat no. 3378) were from Tocris Bioscience (Minneapolis, MN, USA). Tetramethylrhodamine (TRITC)- labeled 70 kDa dextran was purchased from Thermo Fischer Scientific (cat no. D1818, Waltham, MA, USA).

The following expression plasmids were used in this study: Neuropeptide Y (NPY)–mRFP and NPY Venus were a gift from Dr. U. Ashery (Tel Aviv University, Tel Aviv, Israel), NPY-CFP was a gift from Dr. P. Blinder, (Tel Aviv University, Tel Aviv, Israel). pEGFP-LC3 was a gift from Dr. A. Ashkenazi (Tel Aviv University, Tel Aviv, Israel), originally a gift from Tamotsu Yoshimori (Addgene plasmid # 21073; http://n2t.net/addgene:21073; RRID : Addgene_21073) and LC3-EGFP-mRFP (ptfLC3) was a gift from Dr. R. Pinkas Kramarski (Tel Aviv University, Tel Aviv, Israel), originally a gift from Dr. Tamotsu Yoshimori (Addgene plasmid # 21074; http://n2t.net/addgene:21074; RRID : Addgene_21074). Lifeact-GFP was a gift from Dr. B.-Z. Shilo (Weizmann Institute of Science, Rehovot, Israel) and pcDNA3-EGFP-Rac1-T17N was a gift from Dr. Gary Bokoch (Addgene plasmid # 12982; http://n2t.net/addgene:12982; RRID : Addgene_12982). pcDNA3-EGFP was a gift from Dr. J. Silvio Gutkind (University of California, San Diego, USA).

RBL cells stably expressing N-terminally tagged hemagglutinin (HA) human MRGPRX2 (RBL-MRGPRX2) were generated as previously described (20). Cells were maintained at 37°C in a humidified incubator with 5% CO₂ in adherent cell cultures in low glucose DMEM (cat no. 01-050-1A, Biological Industries, Beit-Haemek, Israel) supplemented with 10% FBS (cat no. 12657, GIBCO, Grand Island, NY), 2 mM L-Glutamine (cat no. 03-020-1A, Biological Industries), 100 μg/mL streptomycin and 100 U/mL penicillin, 12.5 U/mL nystatin (Biological Industries) and 1 mg/mL of G418 (cat no. A1720, Sigma Aldrich). LAD-2 cells (a kind gift from Dr. A.S. Kirshenbaum and Dr. D. Metcalfe, Laboratory of Allergic Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD) were cultured in StemPro-34 (cat no. 10640-019, GIBCO) supplemented with 1x StemPro-34 Nutrient, 2 mM L-Glutamine (cat no. 03-020-1A, Biological Industries), 100 U/mL penicillin and 100 μg/mL streptomycin (cat no. 03-032-1B, Biological Industries), and 100 ng/mL hSCF (cat no. 300-07, Peprotech, Rocky Hill, NJ).

Transient transfection of RBL-MRGPRX2 cells was performed as previously described (21). Briefly, RBL-MRGPRX2 cells (1.5 x 10⁷) were transfected with 30-50 µg cDNA by electroporation at 300 V for 20 ms, using an ECM 830 electroporator (BTX, Holliston, Mass, USA). The cells were immediately replated in either 24-well (1 x 10⁵ cells/well) tissue culture dishes for confocal imaging or in 6-well (0.2 x 10⁶ cells/well) tissue culture dishes for flow cytometry experiments containing growth medium and used within 20-24 h after transfection.

RBL-MRGPRX2 cells (1 x 10⁵ cells/well) were incubated with 0.1 mg/mL of TRITC- labeled 70 kDa dextran (D1818, Thermo Fischer Scientific) in Tyrode’s buffer (10 mM HEPES [pH 7.4], 130 mM NaCl, 5 mM KCl, 1.4 mM CaCl₂, 1 mM MgCl₂, 5.6 mM glucose, and 0.1% BSA) for the desired time periods, without or with 10 µM SP, in 24-well dishes at 37°C and 5% CO₂. Where indicated, cells were preincubated for 30 minutes (min), prior to SP trigger, with vehicle or with the indicated inhibitor(s). Cells were washed three times with ice-cold PBS and fixed for 20 min at room temperature (RT) with 4% paraformaldehyde (catalog no. 15710; Electron Microscopy Sciences, Hatfield, Pa) followed by permeabilization for 15 min with 0.1% Triton X-100, 5% FBS, and 2% BSA diluted in PBS. Subsequently, cells were labeled with mouse anti-HA (1:100 dilution) antibody for 1 hour (h) at RT, followed by three washes and a 1-h incubation with Alexa Fluor^® 647 Goat Anti-Mouse IgG H&L secondary antibody (1:500 dilution).

RBL-MRGPRX2 cells were washed three times in Tyrode’s buffer to remove dead, non-adherent cells. RBL-MRGPRX2 (5 x 10⁵) or LAD-2 (6 x 10⁵) cells were stimulated at 37°C and 5% CO₂, in Tyrode’s buffer for the desired time periods. Cells were then washed three times with ice-cold PBS containing 0.5% BSA and 2 mM sodium azide (FACS buffer) followed by fixation in 4% paraformaldehyde for 15 min at RT. For analyses of cell surface MRGPRX2, cells were stained for 30 min at RT with anti-human MRGPRX2 antibody (1:300 dilution). Cells were subsequently washed three times and stained with Alexa Fluor^® 647-conjugated Goat Anti-Mouse IgG H&L secondary antibody (1:2000 dilution) for 30 min at RT, in the dark. For total receptor expression, cells were washed and fixed for 15 min at RT with 4% paraformaldehyde. Cells were subsequently washed and permeabilized with 0.1% saponin and 10% FBS containing PBS for 15 min at RT, after which the cells were washed and stored in the permeabilization buffer overnight at 4°C. The permeabilized cells were then stained for 30 min at RT with anti-human MRGPRX2 antibody (1:300 dilution), washed three times with the permeabilization buffer and stained with the secondary antibody as above. Cells were washed three times with FACS buffer and analyzed by flow cytometry using a CytoFLEX LX flow cytometer (Beckman Coulter, Indianapolis, IN). A minimum of 10,000 cells was acquired per sample. Cells were gated on single cells and were additionally gated on GFP-positive cells in experiments involving transient transfection with either dominant negative (DN) Rac1-GFP or control GFP. Data was analyzed using the FlowJo™ Software Version 10 (Treestar, Ashland, OR).

RBL-MRGPRX2 (1 x 10⁵) and LAD-2 (0.6 x 10⁵) cells grown on untreated (RBL-MRGPRX2) or fibronectin (cat no. F1141, Sigma Aldrich, St Louis, MO, USA) coated (LAD-2) 12-mm round glass coverslips (thick #1; Thermo Scientific, Menzel-Gläser, Saarbrücken, Germany) were washed three times in Tyrode’s buffer, the cells were stimulated as indicated in the same buffer at 37°C and 5% CO₂. The cells were subsequently washed three times with ice-cold PBS and fixed for 20 min at RT with 4% paraformaldehyde (cat no. 15710; Electron Microscopy Sciences, Hatfield, PA) in PBS. Fixed cells were permeabilized for 15 min at RT with 0.1% Triton X-100, 5% FBS, and 2% BSA diluted in PBS. Cells were then incubated for 1 h at RT with the desired primary antibodies followed by three washes and a 1-h incubation with the appropriate secondary antibody (1:500 dilution). After washing, cells were mounted in mounting medium (cat no. E18-18; Golden Bridge Life Science, Mukilteo City, WA) and analyzed with a Leica SP5 or SP8 TCS laser scanning confocal microscope (Leica, Wetzlar, Germany) equipped with a HyD detector and a ×63 oil/1.4 NA objective or a Zeiss LSM 710 confocal microscope equipped with a 63 x1.4 oil Plan-Apochromat objective. Colocalization analysis of NPY-CFP with immunostained MRGPRX2 was quantified as the Manders’ correlation coefficient with Costes’ automatic threshold using the JaCoP plugin of the extended ImageJ version Fiji (22–24). Three-dimensional reconstructions of z stacks (slice thickness ≤ 0.5µm for both fluorescent channels with ~ 10 slices) were performed by using Imaris software (Bitplane, Zurich, Switzerland). Macropinosomes were defined as circular, phase-bright organelles that dissociated from circular ruffles or peripheral ruffles in cells stimulated with SP. To determine the mean fluorescence intensity (MFI) of macropinosomes, the borderline of each macropinosome was outlined with the segmented line tool, saved as a region of interest (ROI) and the MFI was measured using the extended ImageJ version Fiji. The background MFI was subtracted from the MFI of the macropinosomes for each cell, and the MFIs for at least 20 cells/condition were averaged. Next, the MFI values of four randomly selected regions of interest at the periphery of cells triggered with SP for 30 min were selected, measured, and averaged. This averaged MFI values were used for normalization. The size of macropinosomes was calculated by drawing the borderline of each macropinosome with the segmented line tool, again saved as ROIs. The length of the ROIs was then measured using the extended ImageJ version Fiji (23). To determine the vesicles/cell images were converted into threshold images, the required channels were merged (using the AND or XOR operations of the image calculator) to compare the signal patterns and then analyzed with the ImageJ “analyze particle” macro (23).

Data were analyzed using GraphPad Prism Version 8.3.0 for Windows (GraphPad Software, La Jolla, CA). One-way analysis of variance (ANOVA) with repeated measures followed by Dunnett’s post-test or Student’s or Welch’s t-test was used for comparing means, respectively according to the statistic requirements. Results were considered significant when P values were smaller than 0.05.

Given the profound differences between the human MRGPRX2 and its rodent orthologs, we used the RBL-MRGPRX2 cells (19), that stably express the human MRGPRX2, as our model. We have recently validated the authenticity of this model by demonstrating that the ectopically expressed MRGPRX2 in RBL cells retains the authentic signaling elicited by the endogenously expressed MRGPRX2 in the human LAD-2 cells (21). Furthermore, unlike naïve RBL cells, RBL-MRGPRX2 cells have acquired responsiveness to MRGPRX2 ligands, this model allows measuring the exclusive responses of MRGPRX2 to ligands, such as SP, that additionally bind to canonical receptors, such as Neurokinin 1 (NK1), that is endogenously expressed in some MCs. Consistent with our previous results (25), flow cytometric analyses (FACS) revealed a reduction in the cell surface expression of MRGPRX2 in response to SP trigger (Figure 1A and Supplementary Figure 1). However, no significant changes were noted in the total amount of receptor (Figure 1A), implying that half of the total amount of the receptor either remains at the plasma membrane or internalizes, but recycles back to the plasma membrane. However, half of the total amount of receptor, which internalizes in response to SP stimulation, is targeted to an internal, non-degradative compartment(s), and is retained intracellularly, without recycling back to the plasma membrane (Figure 1A). A similar distribution was also noted for the endogenously expressed MRGPRX2 in LAD-2 cells, in which 40% of the total amount of receptor localized to the cell surface following 30 min of cell stimulation with SP, while 60% of the receptor could only be immunostained after cell permeabilization, implying its retainment in an intracellular compartment (Figure 1B).

Further analysis of the first phase of MRGPRX2 internalization in RBL-MRGPRX2 cells revealed that MRGPRX2 endocytosis was sensitive to pitstop2, an inhibitor of both clathrin-dependent endocytosis and clathrin-independent and Arf6-dependent endocytosis (26, 27) (Figure 1C). Thus, following a 5-min trigger with SP, 44% of MRGPRX2 internalized in the absence of pitstop2, but only 26% in its presence (Figure 1C). However, with increasing time of trigger, pitstop2 became less efficient, whereby approximately 60% of MRGPRX2 internalized following 30 min of SP trigger in the absence of inhibitor and 52% in its presence (Figure 1C). MRGPRX2 internalization also displayed a similar pattern of sensitivity to cytochalasin D (CytD), an inhibitor of actin polymerization (28), significantly inhibiting MRGPRX2 internalization following a 5-min trigger with SP, but displaying no inhibition following 30 min SP trigger (Figure 1D). Therefore, collectively, these results suggested that another pathway, alternative to endocytosis, exists for MRGPRX2 internalization.

Cell surface expression of MRGPRX2 was also reduced by dynasore, an inhibitor of dynamin, that executes membrane fission events (29) (Figure 1E). Furthermore, consistent with the more general role of dynamin in controlling internalization pathways, the inhibitory effect of dynasore was evident also at longer periods of SP trigger (Figure 1E).

In agreement with our previous results (21), close inspection of the receptor by high resolution confocal microscopy revealed that in resting cells, the receptor distributed between the plasma membrane and an internal fraction, part of which was perinuclear, reminiscent of the Golgi/TGN, and part resided at vesicles, which we have previously identified as the SGs, based on their labelling with the SG marker, NPY-mRFP (30) (Figure 2A). Confocal images of the cells also revealed marked alterations in cell morphology that occurred following their stimulation with SP (Figure 2A). The latter included cell flattening and formation of membrane ruffles and macropinosomes, that were identified by brightfield images and confirmed by their proximity to actin ruffles, visualized by the expression of Lifeact-GFP, an actin marker (Figure 2A). Furthermore, some of the plasma membrane-localized receptor concentrated at what appeared like sites of macropinosome formation from the membrane ruffles, and some localized to the macropinosomes (Figure 2A). Following a longer period of SP trigger (i.e. 3.5 h), the majority of MRGPRX2 distributed between the perinuclear region and the NPY-mRFP-labelled SGs (Figure 2A). Similar results were also observed in the LAD-2 cells, in which at the end of 30 min stimulation by SP, the internalized receptor was detected in the vicinity of macropinosomes and SGs, which were stained by anti-syntaxin 3 antibodies (Supplementary Figure 2). In contrast, MRGPRX2 localized to membrane protrusions, associated with actin ruffles, in RBL-MRGPRX2 cells that were triggered by SP for 30 min in the presence of CytD (Figure 2B). Quantitative analysis of the confocal images revealed significant increases in MRGPRX2 association with the SGs (Figure 2C), size of the cellular macropinosomes (Figure 2D) and the extent of MRGPRX2 association with macropinosomes (Figure 2E), in response to cell triggering with SP.

Imaging of cells that were triggered for 5 min with SP in the presence of pitstop2 demonstrated a significant increase in the amount of macropinosome-associated MRGPRX2 (Figures 2E, F), with no significant change in the SG-localized MRGPRX2 (Figure 2C). Therefore, these results implied that pitstop2-sensitive endocytosis and macropinocytosis may represent two interchangeable modes of MRGPRX2 internalization, whereby inhibition of one pathway, i.e. endocytosis by pitstop2, shifts the receptor to the alternative macropinocytic pathway.

We also analyzed the influence of dynasore on the distribution of MRGPRX2 and found that in a similar manner to the effect of pitstop2, also dynasore significantly increased the amount of macropinosome-associated MRGPRX2 (Figures 2E, F). Furthermore, dynasore significantly increased the mean size of the cellular macropinosomes as well (Figures 2D, F).

To gain direct evidence for the involvement of macropinocytosis in MRGPRX2 internalization, we allowed the cells to take up TRITC-dextran (70 kDa), a commonly used probe for measuring macropinocytosis (31), and compared its location with the location of the internalized MRGPRX2. The results of these experiments revealed that SP stimulated the uptake of TRITC-dextran, which could be completely inhibited by EIPA, a Na⁺/H⁺ exchange inhibitor, that is known to inhibit macropinocytosis (32, 33) (Figures 3A, B). These results therefore demonstrated that SP can stimulate macropinocytosis. Furthermore, MRGPRX2 localized to the same dextran-containing vesicles (Figures 3A, B), part of which also contained NPY-Venus (Figures 3A, C), revealing a connection between MRGPRX2-positive macropinosomes and the SGs. In sharp contrast to the profound effect of EIPA, the uptake of TRITC-dextran was not affected by pitstop2 (Figures 3A–C), indicating that this alternative pathway of MRGPRX2 internalization was insensitive to this inhibitor, consistent with its ability to compensate for MRGPRX2 internalization by endocytosis (Figure 2E). Indeed, analysis of the cell surface expression of MRGPRX2 by flow cytometry revealed that EIPA was as potent as pitstop2 in inhibiting MRGPRX2 internalization (Figure 1B). Furthermore, combining EIPA with pitstop2 further inhibited internalization, supporting the notion of MRGPRX2 internalization via both pitstop2-sensitive endocytosis and EIPA-sensitive macropinocytosis (Figure 1B).

To substantiate the role of macropinocytosis in MRGPRX2 internalization even further, we also examined the impact of expression of a dominant negative (DN) mutant of Rac1 [i.e. Rac1(T17N)], a key player of actin rearrangements during macropinocytosis (34, 35), on the internalization of MRGPRX2, using the internalization of dextran as a positive control. Expectedly, DN Rac1 inhibited the uptake of TRITC-dextran (Figure 3D) and accordingly significantly reduced the number of internal vesicles that were double positive for dextran and MRGPRX2 (Figure 3E) or triple positive for dextran, MRGPRX2 and NPY-CFP (Figure 3F), which like NPY-mRFP and NPY-Venus, serves as a SG marker. Finally, DN Rac1 significantly retained the cell surface expression of MRGPRX2 (Figure 3G), thus demonstrating its capacity to inhibit SP-stimulated internalization of MRGPRX2.

We have previously shown that early endosomes fuse with the SGs in MCs (36), thus providing a possible mechanism for MRGPRX2 targeting to the SGs. However, our results, showing that inhibition of endocytosis by pitstop2 does not prevent MRGPRX2 targeting to the SGs (Figure 2C), strongly suggested that MRGPRX2 can reach the SGs also directly from macropinosomes. Recently, it was demonstrated that cells respond to plasma membrane injury by forming large macropinosomes at the repair site (37). These structures eventually become positive for the autophagy-related LC3B protein, followed by their shrinkage and fusion with lysosomes (37). Given the lysosomal nature of the MC SGs, the previous reports that documented the presence of LC3 on MC SGs (38, 39), and our recent findings on inhibition of MRGPRX2-stimulated secretion by autophagy targeting drugs (21), we were motivated to examine if LC3 plays a role in MRGPRX2 delivery from macropinosomes to the SGs. To this end, we monitored the intracellular distribution of MRGPRX2 in response to SP trigger, in cells that were co-transfected with NPY-mRFP and LC3-GFP. Analysis of the confocal images revealed that the vesicular fraction of MRGPRX2, which increased with the time of SP trigger, indeed resided on NPY-mRFP labelled SGs, however a fraction of the NPY-positive SGs also labelled positive for LC3-GFP (Figures 4A–E). A smaller fraction of intracellular MRGPRX2 localized to LC3-GFP-positive, but NPY-mRFP-negative vesicles (Figure 4F), that were smaller in size and resided in close proximity to the SGs (Figures 4A, B). Consistent with the reports on bone marrow derived MCs (38), these vesicles, that may correspond to autophagosomes, were present also in resting cells (Figure 4A). Three-dimensional reconstructions of the confocal images corroborated the existence of three types of vesicles (Figure 4B, and Supplementary Video 1), and suggested that a fusion event between LC3-positive vesicles and LC3-negative SGs gives rise to the formation of the LC3-positve SGs (Figure 4C). This fusion event is fast and therefore restricting the number of LC3-positive and NPY-negative vesicles.

To further decipher the connection between macropinocytosis and MRGPRX2 trafficking to the SGs, we tested the impact of dynasore, that inhibits endocytosis, on the cellular distribution of MRGPRX2 in the LC3-GFP-expressing cells. Consistent with its profound impact on the macropinosomes’ size (Figure 2D), dynasore significantly inhibited the association of MRGPRX2 with LC3-posivtive as well as LC3-negative SGs (Figures 4A, D–F). These results therefore implicated dynamin in playing a role in the delivery of MRGPRX2 from macropinosomes to the SGs.

GFP fluorescence is pH sensitive. Thus, while LC3-GFP is fluorescent in autophagosomes, this fluorescence quenches in autolysosomes that form following autophagosome fusion with lysosomes. The possibility therefore arises that the LC3-negative SGs (i.e. LC3^-/NPY⁺) may still contain LC3-GFP, the fluorescence of which is quenched (Figure 4G). To address this possibility, we repeated our analyses of MRGPRX2 distribution in cells that were co-transfected with NPY-CFP and LC3-GFP-mRFP, a tandem fluorescent protein that is commonly used to monitor various stages of autophagy due to the pH insensitivity of mRFP. Thus, LC3-GFP-mRFP yields both green and red signals when located in the less acidic autophagosomes and only a red signal when located in the acidic autolysosomes (40). These studies demonstrated the presence of LC3-containing vesicles that were either double positive for GFP and mRFP, or only positive for mRFP (Figure 5A). Notably, the amount of both types of LC3-positive vesicles significantly increased in response to SP trigger (Figure 5B), indicating that they were formed by a dynamic and trigger-dependent process. Furthermore, with increasing time of SP trigger, the relative amount of GFP-negative/mRFP-positive vesicles increased, suggesting that SP stimulated their acidification, presumably by promoting their fusion with lysosomes (Figure 5B). NPY-CFP colocalized with both LC3-GFP-mRFP double positive and LC3-GFP-negative and mRFP-positive vesicles (Figures 5A, C), implying that both types of MRGPRX2-associated SGs contain LC3, but differ in their internal pH.