C57BL/6J wildtype (“WT”; “B6”; 000664) and NOD.Cg-Prkdc^scid Il2rg^tm1Wjl Tg(CMV-IL3,CSF2,KITLG)1Eav mice (“NSG-SGM3”; 013062) were purchased from Jackson Laboratory. All mice were female sex and littermates unless otherwise stated and maintained in barrier housing in specific-pathogen free animal facilities on the Johns Hopkins Bayview Medical Center campus.

NSG-SGM3 mice were engrafted as previously described at 3–4 weeks old with a single retroorbital injection of 5 x 10⁵ human cord blood CD34⁺ cells (Lonza Poietics) resuspended in sterile PBS (Sigma) (8). Successful engraftment of human cells was confirmed 12 weeks after injection using flow cytometry analysis of whole blood stained with appropriate fluorescent antibodies and processed on a BD Accuri C6 cytometer (BD Biosciences) as previously described (8). Serum samples were collected via retroorbital bleed after 16 weeks of engraftment and at least two weeks prior to splenocyte culture.

NSG-SGM3 mice were fed autoclaved Teklad Global 18% Protein Extruded Rodent Diet (Inotiv Inc.) ( Supplementary Table S1 ) and received enrofloxacin (Baytril; Elanco US Inc) in sterile drinking water as prophylaxis against opportunistic infections. WT mice were housed in normal rodent housing and treated with untreated filtered drinking water. WT and non-engrafted NSG-SGM3 mice were age-matched with 20–24-week-old engrafted NSG-SGM3 mice at the start of culture experiments. All animal protocols were conducted under the approval and regulation of the Institutional Animal Care and Use Committee at Johns Hopkins University School of Medicine.

For murine whole blood samples, whole blood was collected by retroorbital bleeds into purple-cap Microtainer Tubes containing EDTA (BD Biosciences) at room temperature. Samples were lysed twice with 1x multi-species RBC lysis buffer (Invitrogen) in 5 mL polystyrene round-bottom flow cytometry tubes (Corning) at room temperature in the dark and washed with 1x PBS before antibody staining for flow cytometry. For serum samples, murine whole blood was first collected by retroorbital bleeds into serum separator Microtainer Tubes (BD Biosciences). Tubes were allowed to clot at room temperature for at least 30 minutes and centrifuged at 1000 x g for 10 minutes to separate serum. Serum was stored at -20°C until analysis.

The R-PLEX Human IgE assay (Meso Scale Discovery) and the T-PLEX Human Isotyping Panel 1 kit for human IgM, IgG, and IgA (Meso Scale Discovery) were used to detect levels of human antibodies in NSG-SGM3 mouse sera and in supernatants of ex vivo splenocyte cultures after 96 hours. Lyophilized wildtype mouse sera reconstituted in 1x PBS (Sigma) and purified mouse IgE (BD Biosciences) were used as controls to ensure lack of cross-reactivity of the human Meso Scale assays with murine antibodies. Human IgE levels from pooled samples of mouse sera were also confirmed using the ImmunoCAP total serum IgE assay (Thermo Fisher Scientific). Cytokine concentrations of NSG-SGM3 mouse sera were measured using R-PLEX kits for SCF, U-PLEX for GM-CSF and IL-3, U-PLEX for human IL-4 and murine IL-4, and a custom V-PLEX kit for human IL-4, IL-6, IL-13, IFN-γ, and TNF-α (all from Meso Scale Discovery). Recombinant human IL-4 (R&D Biosciences) and murine IL-4 (R&D Biosciences) were used as controls to confirm lack of interspecies cross-reactivity of the Meso Scale IL-4 assays. Sample preparation and processing were performed per manufacturer protocol. Processed plates were read on a MESO QuickPlex SQ 120 machine (Meso Scale Discovery).

Mouse sera were pooled into three samples from 5–6 mice each for analysis on the Allergy Xplorer 2 (ALEX²) (Macro Array Diagnostics), which quantifies specific IgE antibodies for 295 clinically relevant allergens (19, 20). In brief, samples were diluted 1:5 in an assay diluent containing MUXF, a carbohydrate cross-reactive determinant (CCD)-blocking agent (21). From this, 500 μL of sample was incubated on a chip cassette for two hours. The chip was then washed and incubated with substrate for eight minutes, after which the reaction was stopped. Results were read on the ImageExplorer instrument (Macro Array Diagnostics) and reported in kUA/L following interpolation by Raptor software from a built-in total IgE calibration curve. Levels >0.10 kUA/L were considered positive.

To harvest and culture bulk ex vivo splenocytes, we modified published protocols originally designed for the harvest and ex vivo cultures of CD19⁺ B cells (22, 23). Briefly, spleens of WT, non-engrafted NSG-SGM3, or engrafted NSG-SGM3 mice were harvested and mashed through a 70 μm cell strainer (Corning Life Sciences) into a 6-well flat-bottom plate containing RPMI-1640 media (Corning) supplemented with 2.05 mM L-glutamine (DiagnoCine Precision), 15% fetal calf serum (Sigma Aldrich), 1% penicillin and streptomycin (Gibco), and 50 μM 2-mercaptoethanol (Sigma). Cells were spun at 500 x g for five minutes at room temperature. Erythrocytes were then lysed with 1x multi-species RBC lysis buffer (Invitrogen) and remaining bulk splenocytes were washed and spun.

Harvested bulk WT, non-engrafted NSG-SGM3, or engrafted NSG-SGM3 splenocytes were initially plated in 6-well flat-bottom plates at 1 x 10⁶ cells/mL in supplemented RPMI media as above, and incubated from 0 to 48 hours with either 1 μg/mL mouse anti-human CD40 monoclonal antibody (anti-hCD40; BD Pharmingen) and 10 ng/mL recombinant human IL-4 (hIL-4; R&D Systems) anti-human stimulatory cytokines, or 1 μg/mL hamster anti-mouse CD40 monoclonal antibody (anti-mCD40; BD Biosciences) and 10 ng/mL recombinant mouse IL-4 (mIL-4; R&D Systems) anti-mouse stimulatory cytokines. At 48 and 72 hours, cells were replated in 12-well flat-bottom plates at 1 x 10⁶ cells/mL in fresh supplemented RPMI media and re-incubated at both timepoints with the appropriate anti-human or anti-mouse stimulatory cytokines at 0.5 μg/mL CD40L and 5 ng/mL IL-4 concentrations. Culture supernatants were obtained at the 96-hour timepoint and stored at -20°C until further analysis using MSD Multiplexing kits as above (Meso Scale Discovery). Cultures were maintained in a 37°C incubator at 5% CO₂ for 96 hours total before harvest.

Human whole blood samples were obtained from healthy donors consented under a protocol approved by the Institutional Review Board at Johns Hopkins University School of Medicine. Human whole blood samples were collected via standard phlebotomy and drawn into purple-cap BD Microtainer blood collection tubes containing EDTA (BD Biosciences) prior to preparation. Samples were lysed with 1x multi-species RBC lysis buffer (Invitrogen) in 5 mL polystyrene round-bottom flow cytometry tubes (Corning) at room temperature in the dark and then washed with 1x PBS before proceeding with antibody staining for flow cytometry.

Antibody and miscellaneous reagent concentrations for flow cytometry were titrated based on manufacturer protocols ( Supplementary Table S2 ).

For the preparation of ex vivo splenocyte cultures at 0 hour or 96 hour timepoints, harvested cells were washed with 1x PBS, assessed for viability using LIVE/Dead Fixable Aqua (Invitrogen), incubated with a blocking cocktail containing rat anti-mouse CD16/CD32 (mouse BD Fc block; BD Biosciences), human BD Fc Block (BD Biosciences), and True-Stain Monocyte Blocker (BioLegend), and subsequently stained with 100 μL of either anti-human or anti-mouse antibody cocktail ( Supplementary Table S2 ). For preparation of human donor and mouse whole blood, RBC-lysed and washed cells were assessed for viability using LIVE/Dead Fixable Aqua, incubated with a blocking cocktail containing human BD Fc block and True-Stain Monocyte Blocker, and subsequently stained with 100 μL of either anti-human or anti-mouse antibody cocktail ( Supplementary Table S2 ). All stained mouse and human samples were fixed in 50 μL 4% paraformaldehyde in 1x PBS and stored at 4°C until time of analysis. Fixed and stored samples were resuspended in 150 μL 1x PBS immediately prior to cytometer analysis. UltraComp eBeads (Invitrogen), ArC amine reactive beads (Invitrogen), and ArC amine negative beads (Invitrogen) were utilized per manufacturer protocol for compensation of the appropriate antibodies. Samples were run on a Cytek Aurora cytometer (Cytek Biosciences).

Flow cytometry samples were analyzed on FlowJo v10 software (Tree Star) using gating strategies specifically designed to exclude follicular, marginal zone, and germinal center B cells that may otherwise contribute to false positive signal from the final gate ( Supplementary Figures S1, S2 ). Specifically, human PC were identified by gating sequentially on singlet, live, human CD138⁺ human CD19^var, human IgD^-, human CD23^-, human CD138⁺ human CD27⁺ cells ( Supplementary Figure S1 ). Human MBC were gated sequentially on singlet, live, human CD19⁺ human CD138^-, human IgD^-, human CD23^-, human CD20⁺ human CD27^+/hi cells ( Supplementary Figure S1 ). Mouse PC and PB were gated sequentially on singlet, live, mouse CD138^hi, mouse IgD^- mouse CD22^-, mouse/human GL-7^lo, mouse/human B220^lo cells; thereafter, mouse PC were gated as mouse CD138⁺ mouse CD19⁺ cells, and mouse PB were gated as mouse CD138⁺ mouse CD19^- cells ( Supplementary Figure S2 ). Mouse MBC were gated sequentially on singlet, live, mouse CD138^-, mouse IgD^- mouse CD22^-, mouse/human GL-7^lo, mouse CD273⁺ mouse/human B220⁺ cells ( Supplementary Figure S2 ). Minimal cross-reactivity between anti-human and anti-murine antibodies with murine and human samples, respectively, were validated using murine whole blood, murine splenocytes, and human donor whole blood ( Supplementary Figure S3 ).

Anaphylaxis experiments were performed at least 16 weeks post-engraftment. For direct targeting of IgE, engrafted and non-engrafted NSG-SGM3 mice were injected intraperitoneally with 100 ng or 1 µg of goat polyclonal anti-human IgE (Hybridoma Reagent Laboratory) or equivalent amounts of normal polyclonal goat IgG isotype control (R&D Systems) diluted in sterile 1x PBS. To assess anaphylaxis responses, core body temperature was measured using a rectal probe (Kent Scientific), and clinical score was assessed every ten minutes over the course of one hour. Clinical scores from zero to five were as follows: 0) asymptomatic; 1) scratching; 2) piloerection, facial edema, 3) labored breathing, reduced activity; 4) coma or unresponsiveness; and 5) death. In the event of death, temperature measurements for that specific mouse were terminated.

For planned food challenges, enough blood was obtained from two engrafted NSG-SGM3 mice to allow for full specific IgE profiling of their individual sera on the ALEX² chip. While numerous low-level positives were noted, consistent with the previously described pooled samples, both mice had specific IgE antibodies to several foods, of which three foods were selected for challenges based on ease of administration via oral gavage and lack of prior exposure to these allergens. Mice were challenged by oral gavage with 200 μL nonfat cow’s milk (corresponding to 6.6 mg protein; Whole Foods), 20 mg yeast dissolved in water (11.3 mg protein; Anthony’s Brewer’s Yeast Powder), and 56 mg hazelnut paste diluted in water (8.3 mg protein; Pariani Unsweetened Hazelnut Paste) in separate experiments. Anaphylaxis responses were assessed as described above.

Human blood samples were obtained from healthy volunteer donors after informed consent under an Institutional Review Board-approved protocol. Blood was collected via standard phlebotomy and drawn into 4 mL lithium heparin phlebotomy tubes (BD Biosciences). Whole blood was incubated for 30 minutes at 37°C with goat polyclonal anti-human IgE (Hybridoma Reagent Laboratory), polyclonal goat IgG isotype control (R&D Systems), or 1 µM N-formylmethionyl-leucyl-phenylalanine (Sigma), diluted in PAGCM buffer containing piperazine-N,N′-bis[2-ethanesulfonic acid], bovine serum albumin (MP Biomedicals), glucose (Sigma-Aldrich), 1.7 mM calcium (Sigma), and 1.7 mM magnesium (Sigma). After stimulation, cells were fixed with BD Phosflow Fix Buffer (BD Biosciences), centrifuged for five minutes at 400 x g, and resuspended in Pipes buffer containing 1 mM EDTA and 0.25% bovine serum albumin.

In preparation for analysis, cells were blocked with 1 mg/mL non-specific human IgG (MB Biological), incubated with fluorescently-conjugated monoclonal anti-human CD63 (1:1000 dilution; BD Pharmingen) and anti-human FcϵRIα (clone CRA-1, 1:250, Life Technologies) antibodies at room temperature for 25 minutes, and then stained with secondary antibodies anti-CD123-PE (1:100, BD Biosciences), anti-mouse IgG2b-AlexaFluor488 (1:1000, Life Technologies), and anti-mouse IgG1-AlexaFluor647 (1:1000, Life Technologies) at room temperature for 25 minutes. Samples were analyzed on a BD Accuri C6 flow cytometer (BD Biosciences) using a previously-defined gating strategy (24). The percentage of CD63⁺ basophils was recorded for each sample.

Each datapoint represents one sample from one mouse and error bars represent SEM in Figures 1 – 3 unless otherwise stated. Data and error bars in Figure 4 represent mean ± SEM unless otherwise stated. Statistical analyses were determined by unpaired Student’s t-test or two-way ANOVA with post-hoc Šídák’s multiple comparisons test (full model mixed-effects analysis), as specifically denoted in the figure captions and applied based on experimental design. P-values < 0.05 were considered statistically significant. Statistical tests and plotting were performed using GraphPad Prism 10 (GraphPad Software, Inc).

The sera of NSG-SGM3 mice were first analyzed for human antibody titers, demonstrating that naïve engrafted NSG-SGM3 mice without prior marrow ablation produced easily detectable quantities of human IgM, IgG, IgA, and IgE antibodies without active sensitization to antigens, as compared to no detectable amounts from non-engrafted NSG-SGM3 or wildtype C57BL/6J (WT B6) controls ( Figure 1A ). Similar concentrations of serum total IgE were seen within batches of NSG-SGM3 mice engrafted with the same donor HSCs, and IgE concentrations also correlated better by donor HSC than by engraftment efficiency determined by circulating human CD45⁺ count. Of note, even the lowest IgE levels were remarkably several-fold more abundant than that of the average human patient (typically below 200–240 ng/mL for healthy non-atopic humans) (25, 26), while the levels of human IgM, IgG, and IgA were sub-physiologic by several folds (27–29). These human antibody concentrations highlight basal human antibody production in our naïve and otherwise unsensitized humanized mice raised in specific-pathogen free facilities.

We next investigated the landscape of circulating human cytokines, particularly those necessary for human B cell antibody class switching. NSG-SGM3 mice exhibited comparable serum levels of human GM-CSF and IL-3 regardless of the presence or absence of HSC engraftment or the efficiency of HSC engraftment determined by circulating human CD45⁺ count ( Figure 1B ), reflecting constitutive production as previously described (10, 30). NSG-SGM3 mice also produced substantial human SCF ( Figure 1B ), with engrafted mice expressing similar levels to those previously reported for irradiated NSG-SGM3 BLT mice engrafted with fetal liver-derived HSCs (10). However, only engrafted NSG-SGM3 mice exhibited easily detectable serum levels of human IL-4 and IL-13 ( Figure 1C ), which are cytokines essential for type II immunity and IgE class switch recombination, as well as physiologic ranges of human Th1 cytokines IL-6, interferon gamma (IFN-γ), and tumor necrosis factor alpha (TNF-α) ( Figure 1D ) (31–34). Surprisingly, both engrafted and non-engrafted NSG-SGM3 mice produced large amounts of murine IL-4 from residual murine immune cells ( Figure 1C ). In contrast, WT B6 mice used as controls did not have detectable serum murine IL-4, consistent with prior reports for this mouse strain which is known to produce less IL-4 than other strains due to fewer invariant natural killer T cells (33, 35). These data thus confirm that engrafted human HSCs in NSG-SGM3 mice without prior ablation can differentiate into functional human immune cells that produce human Th2 cytokines to support B cell class switching and IgE production.

To further profile the functional roles of human IgE produced by humanized NSG-SGM3 mice, we sought to determine the range of allergen specificity of their serum human IgE. We utilized the ALEX² multiplex chip to test human IgE reactivity to 295 clinically relevant molecular and extract-based allergens, including aeroallergens (e.g. grasses, trees, animal dander), common plant-derived and animal-derived food allergens, and stinging insect venoms. Pooled serum samples from engrafted NSG-SGM3 mice, which again demonstrated considerable total human IgE levels ( Figure 1E ), revealed a diverse profile of allergen specificities ( Table 1 ; Supplementary Table S3 ). Low levels of specific IgE antibodies, all below 2 kUA/L, were detected across seven of eight total broad categories of allergens (plant-derived food; animal-derived food; pollen; mite; insect/venom; animal origin; microorganism/spore; other). The highest specific IgE levels observed were either against foods not present in the mouse chow diet ( Supplementary Table S1 ) or against allergens otherwise highly unlikely to be encountered in the specific-pathogen free mouse facility environment, including millet Pan m (0.17–1.35 kUA/L), egg yolk Gal d_yolk (0.13–0.22 kUA/L), cashew Ana o (0.22–0.26 kUA/L), and cottonwood (0.71–0.83 kUA/L) ( Table 1 ; Supplementary Table S3 ). This diverse and broad range of allergen specificities therefore suggests that even without deliberate active sensitization, NSG-SGM3 mice produce low basal levels of functional polyclonal human IgE specific to diverse allergens, likely including numerous others not tested for by the ALEX² chip, as opposed to one monoclonal IgE antibody specific to a singular allergen.

Next, we aimed to validate whether NSG-SGM3 mice can support the differentiation and maturation of engrafted human CD34⁺ HSCs into human IgE-producing B cells. We first stimulated splenocytes from engrafted NSG-SGM3, non-engrafted NSG-SGM3, or WT B6 mice ex vivo using either human or murine stimulatory CD40L and IL-4 cytokines ( Figure 2A ) (22, 23). We then analyzed cells by flow cytometry for the expansion of human versus murine plasma cells (PC) and memory B cells (MBC) ( Supplementary Figures S1-S3 ) (18, 36–39).

As compared to WT splenocytes, engrafted NSG-SGM3 splenocytes stimulated with human CD40L and IL-4 cytokines ex vivo for 96 hours exhibited a significantly increased percentage of human CD138⁺ CD27⁺ cells, indicating relative enrichment of antibody-producing human PC ( Figure 2B ). While there was mild but non-significant expansion of human PC in response to murine cytokines ( Figure 2B ), suggestive of low/minimal interspecies cross-reactivity, human IgE antibodies were detected in the culture supernatants of only engrafted NSG-SGM3 splenocytes stimulated with human (but not murine) cytokines ( Figure 2C ). In contrast, the percentage of murine PC and younger plasmablasts (PB) in response to human or mouse stimulatory cytokines was similar between engrafted NSG-SGM3 mice and WT mice ( Figure 3A ). Notably, non-engrafted NSG-SGM3 splenocytes were poor controls for flow cytometry due to significant autofluorescence at baseline and high cell death during splenocyte harvest and throughout ex vivo culture, which contributed to significant non-specific and falsely positive fluorescent signal ( Supplementary Figure S4 ). Even so, the undetectable levels of human IgE from culture supernatants of non-engrafted NSG-SGM3 mice stimulated with human or murine cytokines suggest that human IgE-secreting PC were not expanded ( Figure 2C ), as expected given their lack of engrafted human HSCs. Therefore, engrafted NSG-SGM3 mice harbor a Th2 cytokine-responsive population of human PC that can secrete human IgE.

Consistent with our PC findings, engrafted NSG-SGM3 stimulated with human cytokines also exhibited an increase in the percentage of human CD20⁺ CD27⁺ cells as compared to WT splenocytes, which reflects expansion of human MBC ( Figure 2D ), despite comparable percentages of murine MBC ( Figure 3B ). In addition, at baseline prior to ex vivo culture, human PC and MBC were detected in engrafted NSG-SGM3 mice, albeit at rare and low percentages, but were undetectable in WT splenocytes ( Figure 2E ). In contrast and as expected, at baseline prior to culture, engrafted NSG-SGM3 mice had a comparable percentage of murine PC/PB and mildly decreased percentage of murine MBC as compared to WT mice ( Figure 3C ). These data are suggestive of the presence of rare and possibly resting or lowly-active maintenance populations of human PC and MBC in engrafted NSG-SGM3 mice, which can expand upon Th2 cytokine stimulation even with present residual murine immune cells. Our ex vivo findings also support our data on abundant human antibody production in naïve engrafted NSG-SGM3 mice ( Figure 1A ), by highlighting putative upstream human HSC-derived cells which can mature into and expand human PC and MBC populations in response to Th2 cytokines to help sustain the production of basal human IgE.

Lastly, we explored the functionality of human IgE produced by engrafted NSG-SGM3 mice. To understand the anaphylaxis-inducing potential of the polyclonal human IgE and confirm binding of this human IgE to surface FcϵRI receptors on human mast cells and basophils, we utilized a one-step model involving systemic challenge with a single dose of goat anti-human IgE antibody. Remarkably, a single intraperitoneal injection of this goat IgG specific to human IgE into engrafted NSG-SGM3 mice induced a significant dose-dependent decrease in core body temperature and increase in anaphylaxis clinical scores ( Figure 4A ). In contrast, non-engrafted NSG-SGM3 mice given anti-human IgE and engrafted NSG-SGM3 mice given goat IgG isotype control did not exhibit clinical reactivity ( Figure 4A ). We further confirmed the lack of interspecies binding of goat IgG to murine or human Fcγ receptors using in vitro basophil activation assays ( Figure 4B ). As murine FcϵRI receptors cannot bind human IgE (40), our data demonstrate that the polyclonal human IgE produced by humanized NSG-SGM3 mice is physiologically functional and can induce human cell-dependent type I hypersensitivity responses such as anaphylaxis by direct targeting.

We further investigated whether engrafted NSG-SGM3 mice with low levels of food allergen-specific IgE antibodies could undergo anaphylaxis in response to specific food allergen challenge. Engrafted NSG-SGM3 mice did not exhibit clinical reactivity after oral gavage with cow’s milk (serum anti-Bos d 4-IgE of 0.13 and 0.31 kUA/L, respectively), yeast (anti-Sac c-IgE of 0.43 and 0.65), or hazelnut (anti-Cor a 11-IgE of 0.19 and 0.41) ( Figure 4C ). Together, these data suggest that human IgE in humanized NSG-SGM3 mice do harbor functional Fc receptors but, at least with such low levels of food-specific IgE, are not able to elicit clinical anaphylaxis by oral challenge with food allergen.