Transcriptional heterogeneity of mast cells and basophils upon activation

Abstract

Mast cells and basophils are developmentally related cells whose activation is a hallmark of allergy. Functionally, mast cells and basophils overlap in their ability to produce several mediators, including histamine and granule proteases, but studies have increasingly demonstrated non-redundant roles. To characterize the transcriptional heterogeneity of mast cells and basophils upon their activation, we performed large-scale comparative microarrays of murine bone marrow–derived mast cells (BMMCs) and basophils (BMBs) at rest, upon an adaptive-type activation (IgE crosslinking), or upon an innate-type activation (IL-33 stimulation). Hierarchical clustering demonstrated that BMMCs and BMBs shared specific activation-associated transcriptional signatures but differed in others, both between cell type and between activation mode. In BMMCs, IgE crosslinking upregulated 785 genes including Egr2, Ccl1, and Fxyd6, while IL-33 stimulation induced 823 genes including Ccl1, Egr2, and Il1b. Focused bioinformatics pathway analysis demonstrated that IgE activation aligned with processes such as oxidative phosphorylation, angiogenesis, and the p53 pathway. The IL-33–activated transcriptome was enriched in genes commonly altered by NF-κB in response to TNF, by IL-6 via STAT3, and in response to IFNγ. Furthermore, BMBs activated via IgE crosslinking selectively induced immune response genes Ccl1, Il3, and Il2 compared to IL-33–stimulated BMBs. Principal-component analysis revealed key cell- and activation-specific clustering. Overall, our data demonstrate that mast cells and basophils have cell- and activation-specific transcriptional responses and suggest that context-specific gene networks and pathways may shape how the immune system responds to allergens and innate cytokines.

Introduction

Mast cells and basophils are critical effector cells in type I immediate hypersensitivity reactions and associate with allergic disease pathology. The developmental pathways of mast cells and basophils are closely related but are also the subject of continuing controversy (1). In the bone marrow, mast cells and basophils are commonly considered to begin as common myeloid progenitor cells, progress to granulocyte-monocyte progenitors, and become granulocyte progenitors (GPs). Functionally, GPs have been shown to differentiate into either basophils or mast cell progenitors (MCPs) (2, 3). However, this model has been challenged by studies at the single-cell level that define mast cell progenitor potential within in a common myeloid population and not in the granulocyte/macrophage progenitor population (4). Common progenitors are also found in the spleen and can generate basophils and MCPs (5, 6). MCPs are released into the circulation and mature once they are in tissues; mature mast cells have a life span of weeks to months. Conversely, mature basophils circulate in the blood before being recruited to tissues and have a shorter 60–70-hour life span. Due to their developmental similarity, circulating basophils are often used as surrogates to study tissue-resident mast cell reactivity (e.g., the basophil activation test [BAT] for diagnosing food allergy–associated reactivity (7)), but the similarity or heterogeneity between the two cell types is relatively unknown.

Generally, mast cells and basophils both respond to many of the same stimuli, and their functional roles appear to overlap. The best-studied shared mechanism of activation between the cell types is crosslinking of the FcεRI receptor, which promotes the release of histamine as well as other granule mediators, eicosanoids, and inflammatory cytokines. Besides activation through FcεRI, mast cells and basophils also express ST2, the receptor for interleukin-33 (IL-33). IL-33 is an “epithelial-derived” cytokine that promotes type 2–associated immune responses and has been strongly linked to allergy (8). Upon IL-33–mediated activation, mast cells have been shown to exhibit enhanced adhesion, survival, maturation, and production of several pro-inflammatory cytokines (9, 10). In basophils, IL-33–mediated activation has been described as promoting migration towards eotaxin and enhancing degranulation upon concurrent IgE-mediated activation (11). For cytokine production, basophils were also shown to respond poorly to IL-33 compared to IgE-mediated activation, but a relatively small number of mediators were studied (12, 13). Defining the transcriptional activation signatures of mast cells and basophils utilizing a systems biology–based approach is therefore likely to help understand the possible contributions of each cell type and the mode of activation to the progression of allergic diseases.

Recent evidence has shown that resting mast cells and basophils actually have relatively low transcriptional homology in both mice and humans (14, 15), implying that they might be much more distinct than once thought. However, the similarities and/or differences in their transcriptional signatures upon activation are currently unknown. It is well appreciated that mast cells have functions in both health and disease. In line with their role as sentinel cells, mast cells encode pathways for the synthesis of a diverse array of mediators. They are critical players in the symptoms of anaphylaxis, mediate resistance to infection, and promote tumor rejection. Relatively less is known about basophil functions. Basophils rapidly secrete IL-4 and IL-13 upon IgE-mediated activation, play a role in the host defense against parasites, and promote Th2 responses through altering antigen presentation (13, 16). While mast cells and basophils are believed to have non-redundant roles in immune regulation, the evidence supporting this idea has been scarce up until now, and further understanding of the distinct contributions of each cell type at rest and upon activation are beginning to be appreciated.

Here, to better understand the functional differences and similarities of mast cells and basophils, we performed a large-scale comparison between the transcriptomes of murine mast cells and basophils at rest and upon activation by either an adaptive-type (IgE crosslinking) or innate-type (IL-33) of stimulation. Using homogeneous populations of mature mast cells and basophils derived from bone marrow, we also asked whether IgE and IL-33–mediated activation induced similar or distinct gene signatures in both cell types. Through this non-biased, large-scale approach, we provide evidence for heterogeneity between mast cell and basophil activation responses as well as in the activation of each individual cell type by an innate- or adaptive-type of stimuli. The cell- and activation-specific signatures identified in this study define novel gene networks and pathways that may contribute to understanding how the immune system responds to allergens and innate cytokines.

Methods

Mice

C57BL/6J mice (4–6 weeks old) were obtained from the Jackson Laboratory. ST2-knockout (ST2KO) mice were previously obtained from Dr. Andrew McKenzie and backcrossed to C57BL/6J for 8 generations. All animal studies were performed under IACUC guidelines and under protocols that have been approved by the Northwestern University Animal Care and Use Committee. The bone marrow from each individual mouse was used to generate independent cultures of mast cells and basophils, as outlined below.

Bone marrow–derived mast cells (BMMCs)

BMMCs were obtained by flushing bone marrow from femurs and tibias of mice. Cells were cultured in BMMC media (RPMI 1640 media containing 2 mM L-glutamine, 10% FBS [Atlanta Biologicals], 25 mM HEPES [Sigma], 1 mM sodium pyruvate [Sigma], 0.1 mM non-essential amino acids [Sigma], 100 U/mL penicillin, 100 μg/mL streptomycin [Corning], and 0.05 mM β-mercaptoethanol [Sigma]) and 30 ng/mL recombinant mIL-3 (Miltenyi Biotec) for 4–6 weeks. Purity was assessed by flow cytometry using PE–anti-mFcεRI (MAR-1, eBioscience) and APC–anti-mCD117 (2B8, BD Biosciences) (Fig. 1A). For IgE/Ag activation, BMMCs were primed with 1 μg/mL αOVA-IgE overnight and activated with 0.5 μg/mL OVA (Sigma) for 4 h, as previously described (17). For IL-33 activation, BMMCs were activated with 10 ng/mL IL-33 for 4 h (Fig. 1B).

FIGURE 1.

Generation of murine mast cells and basophils. (A) Bone marrow cells were cultured in rIL-3 (30 ng/mL) for 4 weeks to generate mast cells (BMMCs). Basophils (BMBs) were purified from the bone marrow culture after 10 days with rIL-3 by an initial negative depletion for mCD117 followed by positive selection with mCD49b magnetic beads. Cell surface FcεRI, CD117, or CD49b expression was used to validate purity. (B) BMMCs and (C) BMBs were left unstimulated or activated with either IgE crosslinking or IL-33 stimulation in triplicate. (D) RNA from these samples was sent for microarray analysis, and expression of key mast cell and basophil genes in the WT unstimulated samples used to initially verify cell-specific signatures were evident.

Bone marrow–derived basophils (BMBs)

BMBs were generated by flushing bone marrow from femurs and tibias of mice. Cells were cultured in BMMC media plus 30 ng/mL recombinant mIL-3 (Miltenyi Biotec) for 9 days. On day 9, suspended cells were sorted using autoMACS (Miltenyi Biotec) first with a negative depletion using mCD117 MicroBeads (Miltenyi Biotec), then by positive selection with mCD49b (DX-5) MicroBeads (Miltenyi Biotec). Purity of the sorted cells was confirmed using flow cytometry for APC–anti-mCD49b (HMα2, BD Biosciences), FITC–anti-mFcεRI (MAR-1, eBioscience), and PE–anti-mCD117 (2B8, BioLegend) (Fig. 1A). IgE/Ag and IL-33 activation of BMBs was performed exactly as described above for BMMC activation (Fig. 1C).

RNA isolation and Real-time PCR

RNA was isolated from unstimulated and stimulated cells with RNeasy RNA isolation kits (Qiagen). RNA quality assessment was performed using the Agilent 2100 Bioanalyzer. cDNA was synthesized with qScript cDNA Supermix (Quanta Biosciences). Gene expression was verified for select genes by real-time PCR using a ABI7500 instrument (Applied Biosystems) and TaqMan probes (Applied Biosystems).

Microarray experiments and statistical analysis

RNA was hybridized to Illumina MouseWG-6 v2.0 arrays (Catalog ID: BD-201-0202). Pre-processing steps including quality control, background adjustment, and quantile normalization were performed using the array analysis tools available via www.arrayanalysis.org using the lumi package and bgAdjust function (18). Variance stabilization was managed with a log₂ transformation. Normalized data was filtered to remove unexpressed probes and beads with a detection p value < 0.01. Groups were compared using the limma adapted t test. Differentially expressed genes were identified using the Benjamini–Hochberg False Discovery Rate (FDR) correction p value ≤ 0.05. Heatmaps were generated using GENE-E software (Broad Institute, http://broadinstitute.org/cancer/software/GENE-E).

Cell-specific signature validation

For each sample, the purity of each BMMC culture was determined to be >98% FcεRI⁺/CD117⁺ and each BMB culture was >98% FcεRI⁺/CD117^–/CD49b⁺. To assess the specificity of the cellular transcriptome between our BMMC and BMB populations, we initially examined key genes that were described as reflecting a mast cell–specific signature in a recent study from the Immunological Genome Project Consortium (15) as well as genes for two basophil proteases that are not expressed in mast cells, Mcpt8 (Mcpt8) and Prss34 (Mcpt11) (19). As shown in Fig. 1D, all three mast cell cultures clustered together while the basophil cultures also showed distinct clustering. Furthermore, the mast cell signature was highly reflective of the mast cell cultures while basophils possessed a unique signature driven by Mcpt8 and Prss34. Additionally, there was no significant expression of Mitf, a transcription factor described as important for early commitment of basophil and mast cell lineages (20).

Principal-component analysis and network diagrams

Functional Gene Set Enrichment Analysis (FGSEA) was used to generate functional gene networks and identify hub genes by meta-grouping of individual gene term sets (referencing GO Biological Process and KEGG Pathways) based on function similarity. The GeneTerm Linker algorithm implemented in the “FGNet” R package was used for this analysis (parameters set for analysis: adjusted p value < 0.05; minimum support of 3). Networks generated within this analysis were exported in Graph Modeling Language (GLM) format for further analysis and visualization using the “iGraph” R package. Cytoscape 3.2.1. was used to visualize individual metagroups and stimulus-specific vs. shared network components. Principal-component analysis (PCA) was performed using the Population PCA program (Scott Davis, Harvard Medical School) on the top 15% variable transcripts with a relative expression ≥ 120. Gene Set Enrichment Analysis (GSEA) (Broad Institute) was used to determine enriched hallmark gene sets in activated mast cells and basophils (21). Circos visualization and enrichment heatmaps were generated using www.metascape.org (22).

Results

IgE- and IL-33–activated mast cells are transcriptionally distinct

Mast cells and basophils are developmentally related but perform similar as well as non-redundant roles in allergic disease. One way to determine the degree of differences in their output responses to adaptive (IgE crosslinking) and innate (IL-33) stimuli is to use a non-biased bioinformatics approach to interrogate the genetic changes that occur upon activation. Before performing a cell-specific comparison between mast cells and basophils, we first asked if the activation-specific transcriptome was distinct between IgE and IL-33 activation in each cell type. For mast cell activation, we set up an in vitro model using BMMCs––a well-validated technique for generating a homogeneous mast cell population––with an average purity >98% (Fig. 1A) (23). BMMCs have properties of both serosal- and mucosal-type mast cells and are extensively used to study FcεRI signaling (24). Using array-based technology, we analyzed the expression of more than 45,000 transcripts in unstimulated and IgE/Ag- or IL-33–stimulated BMMCs in triplicate with each culture generated from individual mice (NCBI Gene Expression Omnibus Series GSE96696 at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE96696).

For IgE activation, crosslinking of αOVA-IgE with OVA for 4 h resulted in 785 differentially expressed genes (~4.0%) compared to unstimulated mast cells (adjusted p value ≤ 0.05). A heatmap of upregulated and downregulated genes was generated by first filtering (log₂ fold change [log₂FC] > 2, average expression ≥ 5, and p ≥ 0.05) and then ranking the top 50 transcripts according to their adjusted p value (Fig. 2A). From these top 50 genes, we observed three major patterns of gene expression: (1) genes downregulated by IgE activation, (2) genes upregulated by IgE activation, or (3) genes upregulated by both IgE and IL-33 activation. Of the differentially expressed genes, 309 genes were upregulated by at least 2-fold, including Il33, as our lab had previously described (25). In contrast, 234 genes were downregulated by at least 2-fold, illustrated by the balanced appearance of significant data points on the volcano plot (Fig. 2B). Among the genes with the highest fold change were Egr2 (72.26-fold), Ccl1 (67.24-fold), and Fxyd6 (62.30-fold) (Fig. 2C, 2D). WT and ST2KO BMMCs that underwent IgE-mediated activation had virtually the same transcriptome. In fact, out of the ~45,000 analyzed transcripts, pairwise comparison between the groups found that only Rpe (ribulose-5-phosphate-3-epimerase) was significantly upregulated in WT BMMCs by 4.9-fold.

FIGURE 2.

Transcriptional signature of IgE-activated mast cells. (A) Heatmap of the top 50 significantly changed genes in BMMCs after IgE-mediated activation. (B) Volcano plot showing log₁₀(p value) versus log₂FC. Colored points represent p < 0.05 (red), log₂FC > 2.5 (orange), and p < 0.05 & log₂FC > 2.5 (green). (C) Top 10 upregulated and (D) downregulated genes after IgE-mediated activation ranked by fold change.

We next performed a pairwise comparison of IL-33–stimulated versus–unstimulated WT and ST2KO mast cells and identified 823 differentially expressed genes (~4.2%) that were statistically significant after 4 h. A heatmap of the top 50 upregulated or downregulated genes illustrate two main subsets of transcripts that are (1) IL-33–induced, but not IgE-induced, or (2) induced by both stimuli (Fig. 3A). Notably, the IL-33–induced genes remained unchanged in ST2KO BMMCs upon IL-33 stimulation. Pairwise comparison of unstimulated WT BMMCs and IL-33–stimulated ST2KO BMMCs again identified only Rpe as a statistically significant change (upregulated by 6.4-fold). A volcano plot of the IL-33–induced transcriptome displays a rightward shift with many more upregulated (log₂FC > 2.5) than downregulated genes (Fig. 3B). IL-33 stimulation resulted in 299 upregulated and 248 downregulated genes (by at least 2-fold); the genes with the highest magnitude fold change compared to unstimulated mast cells were Ccl1 (138.59-fold), Egr2 (104.90-fold), and Il1b (62.56-fold) (Fig. 3C, 3D).

FIGURE 3.

Mast cell transcriptional signature after IL-33 activation. (A) Heatmap of the top 50 significantly changed genes in BMMCs after IL-33 stimulation. (B) Volcano plot showing log₁₀(p value) versus log₂FC. Colored points represent p < 0.05 (red), log₂FC > 2.5 (orange), and p < 0.05 & log₂FC > 2.5 (green). (C) Top 10 upregulated and (D) downregulated genes after IL-33 stimulation ranked by fold change.

Using GSEA, we analyzed the top differentially expressed hallmark gene sets enriched in activated mast cells. IgE-crosslinked mast cells induced genes that encode for proteins that participate in oxidative phosphorylation, angiogenesis, and p53 pathways (Supplemental Fig. 1A). In contrast, IL-33–stimulated mast cells induced genes that are commonly regulated by NF-κB in response to TNF, upregulated by IL-6 via STAT3, and upregulated in response to IFNγ (Supplemental Fig. 1B). A network diagram constructed from significantly upregulated genes revealed gene signatures that were unique and common to these two stimuli (Fig. 4). As we have reported previously and now confirm with this microarray approach, the induction of Il33 expression by mast cells is a signature unique to IgE-mediated activation, but not IL-33 stimulation (Fig. 4) (25).

FIGURE 4.

Network diagram of upregulated genes in IgE-activated or IL-33–stimulated mast cells.

Basophils are strongly activated by IgE-mediated, but not IL-33–mediated, stimulation

For determining the innate-type and adaptive-type activation transcriptome of basophils, we derived the transcriptional signature of activated BMBs (average purity >98%) in response to IgE crosslinking and IL-33. IgE-mediated activation resulted in 3986 differentially expressed genes (20.5%) that achieved statistical significance with an adjusted p value ≤ 0.05. The heatmap of the top 50 upregulated and downregulated genes included many transcripts encoding secreted cytokines and chemokines (Fig. 5A). Like in BMMCs, we observed Il33 induction upon IgE crosslinking, but not after IL-33 stimulation. A volcano plot of the data from IgE stimulation demonstrates a wide spread of data points away from the origin due to many transcripts showing a large magnitude log₂FC and a low p value (Fig. 5B). Of these displayed transcripts, 875 distinct genes increased by at least 2-fold, and 730 genes decreased by at least 2-fold. The genes with the highest fold change were upregulated Ccl1 (369.39-fold), Il3 (222.99-fold), and Il2 (107.41-fold) (Fig. 5C, 5D).

FIGURE 5.

Basophil transcriptional signature after IgE-activation. (A) Heatmap of the top 50 significantly changed genes in BMBs after IgE-mediated activation. (B) Volcano plot showing log₁₀(p value) versus log₂FC. Colored points represent p < 0.05 (red), log₂FC > 2.5 (orange), and p < 0.05 & log₂FC > 2.5 (green). (C) Top 10 upregulated and (D) downregulated genes after IgE-mediated activation ranked by fold change.

In response to IL-33 stimulation, WT, but not ST2KO, basophils displayed altered expression of 238 genes (1.2%) with statistical significance (142 upregulated and 5 downregulated transcripts by at least 2-fold). The heatmap displays the top 39 genes expressed by IL-33–stimulated basophils (Fig. 6A). Unlike the previous groups, the pairwise comparison of IL-33–stimulated basophils and unstimulated basophils resulted in fewer than 50 transcripts that met the cutoffs of log₂FC > 2, average expression ≥ 5, and p ≤ 0.05. The rightward bias of the volcano plot illustrates that IL-33 increases the expression of most genes and decreases relatively few (Fig. 6B). The top three differentially regulated genes in IL-33–stimulated basophils were Plat (73.18-fold), Cxcl2 (22.24-fold), and Tpbg (19.54-fold) (Fig. 6C, 6D).

FIGURE 6.

Basophil transcriptional signature after IL-33 activation. (A) Heatmap of the top 50 significantly changed genes in BMBs after IL-33 stimulation. (B) Volcano plot showing log₁₀(p value) versus log₂FC. Colored points represent p < 0.05 (red), log₂FC > 2.5 (orange), and p < 0.05 & log₂FC > 2.5 (green). (C) Top 10 upregulated and (D) downregulated genes after IL-33 stimulation ranked by fold change.

A network diagram constructed from the upregulated transcripts illustrates the groups of genes and pathways that appear to be IgE-specific, IL-33–specific, and common responses to stimuli (Fig. 7). By GSEA, we identified the following three most enriched hallmark gene sets in IgE-activated basophils: (1) MYC-regulated genes (subgroup v1), (2) genes upregulated during the unfolded protein response, and (3) genes upregulated by STAT5 in response to IL-2 stimulation (Supplemental Fig. 1E). In contrast, IL-33–stimulated transcripts aligned with the following hallmark genes sets: (1) genes upregulated by KRAS activation, (2) genes defining the inflammatory response, and (3) genes regulated by NF-κB in response to TNF (Supplemental Fig. 1F).

FIGURE 7.

Network diagram of upregulated genes in IgE-activated or IL-33–stimulated basophils.

Mast cells and basophils are transcriptionally distinct from each other at rest and after stimulation

As a final step in our analyses, we quantified the differences between mast cells and basophils at rest and after activation by comparing the lists of differentially expressed transcripts from all four groups from the pairwise comparisons illustrated above. Using a Circos plot, we illustrate matching genes among treatments and cell types by drawing purple lines to link identical genes in two given groups (Fig. 8A). The dark orange areas on the inner arc represent the subset of genes that are found in at least one other group, while the light orange areas signify the proportion of genes that are unique to the given group. A Venn diagram illustrating the specific numbers of overlapping versus unique genes between each group is shown in Fig. 8B and a list of these genes provided in Supplemental Table 1. Surprisingly, only 2 genes (Tnf and Traf1) were common to both cell types and both stimuli and the numbers of unique genes within each specific cell-stimulus category dominated over those that were shared across categories. PCA using the top 15% of transcripts with the most variability (6473 probes) illustrates cell type–attributable variation captured by PC1 and PC2 as well as stimulation-specific variation along PC3 (Fig. 8C). A subset of genes representing the cell and stimulus-specific signatures were validated in independent cultures using real-time RTPCR methods (Supplemental Figure 2).

FIGURE 8.

Comparison of the transcriptional signature of mast cells and basophils after activation. (A) Circos plot showing overlap among filtered genes for each group (log₂FC > 2, average expression ≥ 5, and p ≤ 0.05). Purple curves link identical genes. Light orange areas within the inner arc indicate unique genes. Dark orange areas within the inner arc represent genes that are differentially expressed in another group. MC: mast cell; Ba: basophil. (B) Venn diagram illustrating frequency of overlapping and unique genes between groups. (C) Principal-component analysis of the population using the top 15% probes with the most variable expression.

Taken all together, our data suggest that the core transcriptional response to innate- and adaptive-type stimuli can be defined. We analyzed transcriptional responses in both mast cells and basophils and define the cell-specific signature for these two stimuli. From the cell- and activation-specific profiles, we derive novel gene networks and pathways that may participate in how the immune system responds to allergens and innate cytokines.

Discussion

Mast cells and basophils are developmentally related cells that play key effector roles in allergic and non-allergic diseases, and the extent of their functional similarities and differences in response is not yet fully understood. Here, we broaden our understanding of these two cell types by asking whether heterogeneity exists between the transcriptome of mast cells and basophils at rest and upon activation with either an innate-type (IL-33) or adaptive-type (IgE crosslinking) of stimuli. We use IL-33 as the innate stimuli, as (1) it is well known to activate mast cells through ST2, (2) it promotes type 2 immune responses (9, 10), and (3) basophils also express ST2 and have been shown to respond to IL-33 (13, 26). To address the question above, we used a non-biased, bioinformatics approach that provided a global analysis of the transcriptional changes that occurred in mast cells and basophils upon activation. Through principal-component analysis, we confirmed that mast cells and basophils have distinct transcriptional programs at rest and further defined that they also have distinct transcriptional programs upon activation with either an adaptive- or innate-type stimulation (Fig. 8C). Also, through analyzing IgE-crosslinked and IL-33–stimulated mast cells and basophils independently, we surprisingly observed that transcriptional differences were enhanced after activation (even with the same stimuli), suggesting even further functional differences.

Focusing on mast cells, IgE-activated mast cells shared a small transcriptional signature (327 genes) with IL-33–stimulated mast cells. Using this shared signature, we identified the hallmark gene sets enriched after both modes of activation: processes associated with allograft rejection, apoptosis, and cholesterol homeostasis (Supplemental Figure 1D). Interestingly, activation of mast cells either through FcεRI or ST2 dramatically increased expression of Ccl1 and Egr2. CCL1 is an important chemokine in regulating immune cell migration, specifically CD4⁺ T cell trafficking (27, 28). Other groups have previously reported the Egr2-dependent induction of CCL1 in mast cells after IgE-mediated activation (27–29), connecting these molecules to a shared regulatory pathway, but production of CCL1 in mast cells after IL-33 stimulation has not been previously described. Furthermore, based on fold-change values, IgE-crosslinked basophils expressed more Ccl1 than IgE-activated mast cells in our arrays; this finding may suggest a role for basophil-derived chemokines in CD4⁺ T cell regulation.

A surprising aspect of our findings is that basophils respond strongly to IgE-mediated activation and less potently to IL-33, unlike the balanced activation signatures seen with mast cells. Only 20 differentially expressed genes were shared between the two activation groups in basophils. Basophils activated via IgE crosslinking exhibited the most robust response among all the tested groups with 20.5% of the transcriptome altered upon activation and included genes regulated by MYC (subgroup v1 and v2) as well as genes participating in oxidative phosphorylation and the unfolded protein response. In contrast, the IL-33–altered genes in basophils reflected quite unexpected pathways including heme metabolism, mitotic spindle assembly, and adipogenesis. The PCA plots further highlight the divergence of response phenotype in basophils, with IgE-activated basophils of both genotypes (WT and ST2KO) clustering tightly with one another but distantly from all other groups, including IL-33–activated basophils. Interestingly, we do show that amphiregulin (Areg)––a growth factor in the epidermal growth factor (EGF) family known to be involved in allergic responses––is induced in basophils by both IgE crosslinking and IL-33 (Fig. 5A). Previously, it was reported that human basophils produced AREG in response to IL-3, and the authors suggested that basophil-derived AREG contributed to tissue remodeling and repair (30, 31). Recently, ILC2-derived AREG was shown to mediate the tissue-protective function of ILC2s in intestinal tissue (32). Here, we provide the first evidence to suggest that activated basophils could participate in similar tissue remodeling processes during type 2 immune responses in allergic diseases through their expression of AREG.

In this study, we include ST2-deficient cells and observe virtually no significant transcriptional differences compared with WT cells at rest and upon IgE-mediated activation. The lack of a response in the ST2KO mast cells and basophils to IL-33 illustrates the necessity of the ST2 receptor and the specificity of the IL-33–activation signature. Since we previously demonstrated that IgE crosslinking promoted IL-33 expression in mast cells (17, 33) (and now also in basophils, as shown above), we postulated that preformed or induced IL-33 could feed back through the ST2 receptor to potentiate cellular activation. Since the IgE-activated transcriptome was identical in WT and ST2KO BMMCs and BMBs, this type of autocrine regulation of IL-33/ST2 signaling––as has been proposed to occur in dendritic cells (34)––seems unlikely. However, a caveat to this concept is that we investigated only one time point (4 h); thus, while our findings seem to rule out any contribution of preformed IL-33, the potential for a later influence of induced IL-33 being secreted and functioning in an autocrine fashion remains.

While a recent study by Dwyer et al. derives both mast cell– and basophil-specific transcriptional signatures at rest, our work extends these findings by characterizing the responses of these cell types after activation from both innate- and adaptive-type stimuli (15). Indeed, this comparative analysis of mast cells and basophils after activation provides additional evidence that these cells may independently regulate immune responses during active disease. This finding challenges the general assumption that circulating basophils can be used as functional surrogates for tissue-resident mast cells and, at the very least, highlights the need for caution in doing so in research or clinical practice.

A potential caveat in our study is that the use of bone marrow–derived primary cells might not fully extrapolate in phenotype to tissue-derived cells, as was the approach by Dwyer et al. for the Immunological Genome Project Consortium (15). However, two key points can be raised in considering the findings here within the context of that previous study. First, as we show in Fig. 1, the mast cells generated for our study demonstrate a clear preservation of their reported “mast cell signature,” indicating that BMMCs reflect a core mast cell transcriptomic profile and not that of a precursor or other lineage cell. Second, using bone marrow–derived primary cells actually provides us with the distinct advantage of profiling transcriptional responses in highly homogeneous populations of mast cells and basophils, as they have been generated under similar culture and activation conditions; moreover, these cells were generated without using techniques that could unintentionally activate the cells, such as enzymatic disruption or mechanical separation. Indeed, Dwyer et al. actually reported upregulation of several key genes upon digestion enzyme exposure that we demonstrate here as reflecting activation signatures, including Egr2 and Ccl1 (15), implicating basal activation of endogenous mast cells during the methods currently used to extract them from tissues. For this reason, it is likely impossible to discriminate stimulus-specific signatures using tissue-isolated mast cells with the clarity that we have been able to achieve here using a homogeneous bone marrow–derived cell population.

In conclusion, we demonstrate that mast cells and basophils have both cell- and activation-specific transcriptional signatures. The limited homology that we observe between these cell types after activation with the same stimuli suggests that these cells play independent roles in the immune response. These findings provide a firm foundation upon which future investigations can build on our collective understanding of functional differences between these developmentally related cell types.