Alternative splicing of interleukin-33 and type 2 inflammation in asthma

Abstract

Type 2 inflammation occurs in a large subgroup of asthmatics, and novel cytokine-directed therapies are being developed to treat this population. In mouse models, interleukin-33 (IL-33) activates lung resident innate lymphoid type 2 cells (ILC2s) to initiate airway type 2 inflammation. In human asthma, which is chronic and difficult to model, the role of IL-33 and the target cells responsible for persistent type 2 inflammation remain undefined. Full-length IL-33 is a nuclear protein and may function as an "alarmin" during cell death, a process that is uncommon in chronic stable asthma. We demonstrate a previously unidentified mechanism of IL-33 activity that involves alternative transcript splicing, which may operate in stable asthma. In human airway epithelial cells, alternative splicing of the IL-33 transcript is consistently present, and the deletion of exons 3 and 4 (Δ exon 3,4) confers cytoplasmic localization and facilitates extracellular secretion, while retaining signaling capacity. In nonexacerbating asthmatics, the expression of Δ exon 3,4 is strongly associated with airway type 2 inflammation, whereas full-length IL-33 is not. To further define the extracellular role of IL-33 in stable asthma, we sought to determine the cellular targets of its activity. Comprehensive flow cytometry and RNA sequencing of sputum cells suggest basophils and mast cells, not ILC2s, are the cellular sources of type 2 cytokines in chronic asthma. We conclude that IL-33 isoforms activate basophils and mast cells to drive type 2 inflammation in chronic stable asthma, and novel IL-33 inhibitors will need to block all biologically active isoforms.

Keywords: alternative splicing; asthma; basophils; interleukin-33; type 2 inflammation.

Fig. 1.

Multiple IL33 splice variants are expressed in human airway epithelial brushings. (A) IL-33 immunolocalizes to epithelial and endothelial cells in the submucosa. The image is from tissue sections of endobronchial biopsies taken from a representative healthy subject. (B) Multiple alternatively spliced transcripts of IL33 are detectable in airway epithelial RNA from healthy subjects (n = 10). Exons 2–8 of IL33 were reverse-transcribed, PCR-amplified, and analyzed on an Agilent 2100 Bioanalyzer. Splicing out of some or all of exons 3, 4, and 5 are revealed by cloning and sequencing of cDNA generated from airway epithelial cell RNA from human subjects (SI Appendix, Fig. S2). (C) By using RNase H-dependent primers, full-length IL33 is the most abundant transcript in airway epithelial brushings, but multiple IL33 isoforms are also highly expressed. The data are from airway epithelial brushings from 39 healthy controls and are corrected for primer efficiency (median ± interquartile range).

Fig. 2.

Deletion of exons 3 and 4 confers cytoplasmic localization of IL-33 in primary airway epithelial cells, and exon 5 is required for cytokine activity. (A) Full-length and alternatively spliced IL33 cDNA were cloned into CMV-driven mammalian expression vectors tagged with venus fluorescent protein, and plasmids were transfected into primary human airway epithelial cells. Full-length, short 5, Δ exon 3, Δ exon 4, Δ exon 5, Δ exon 3–short 5, and Δ exon 4,5 isoforms localized to the nucleus, whereas the Δ exon 3,4 and Δ exon 3,4–short 5 isoforms localized to both the nucleus and the cytoplasm. The Δ exon 3,4,5 isoform demonstrates a punctate, perinuclear staining, which colocalized with the autophagosome marker LC3B. (B) In vitro-translated IL-33 protein isoforms show differences in signaling activity in HMC-1 cells. Whereas isoforms containing exon 5 cause release of IL-8, proteins with a deletion in exon 5 do not. (C) Recombinant purified Δ exon 3,4 IL-33 protein demonstrates cytokine bioactivity in HMC-1 cells (EC₅₀: 0.8–1.1 ng/mL). The activity of Δ exon 3,4 IL-33 in HMC-1 cells is inhibited by sST2. Data in B and C represent mean ± SD for three replicates for each experimental condition. ***P < 0.001; ****P < 0.0001.

Fig. 3.

The Δ exon 3,4 IL-33 transcript is associated with type 2 airway inflammation. A forest plot shows the odds ratios (x axis) for the association between the expression of 5 different IL-33 isoforms (y axis) in airway epithelial cells and airway type 2 inflammation in 85 stable asthmatic subjects. The type 2 inflammation outcome is based on a composite metric of three IL-13–responsive genes in airway epithelial cells (periostin, CLCA1, and serpinB2). The plot shows that only the Δ exon 3,4 IL33 isoform is significantly and positively associated with airway type 2 inflammation. Data presented in black are adjusted for age and gender, and data presented in red are adjusted for age, gender, and use of inhaled steroids. Δ ex 3,4–sh5 refers to Δ exon 3,4–short 5. *P < 0.05; ***P < 0.001.

Fig. 4.

The Δ exon 3,4 IL-33 splice variant is secreted from an airway epithelial cell line. (A) Stable overexpression of full-length IL-33 fused with GFP (full), or Δ exon 3,4 IL-33 fused with GFP (Δ3,4) in Beas2B cells using lentiviral transduction results in increased full-length and Δ exon 3,4 RNA transcripts. None, untransduced Beas2B cells; GFP, overexpression of GFP alone. (B) Despite similar levels of transcript overproduction, full-length IL-33 is more abundant in the cellular lysate than Δ exon 3,4 IL-33. (C) At 24 h, GFP is present in the conditioned medium (CM) of the cells expressing full-length IL-33–GFP, but it is more robust in the cells expressing Δ exon 3,4 IL-33–GFP. (D) There is no evidence of cellular necrosis in untransduced cells, full-length IL-33–GFP, and Δ exon 3,4 IL-33–GFP cell lines as measured with glucose-6-phosphate dehydrogenase (G6PD) release assay. Beas2B total protein lysate (60 μg) demonstrates a positive G6PD signal. (E) After 1-h stimulation with 5 μM ionomycin (Iono), there was an increase in extracellular GFP in cells expressing Δ exon 3,4 IL-33–GFP, but not those expressing full-length IL-33–GFP. *P < 0.05; ****P < 0.0001.

Fig. 5.

Sputum cell analysis links basophils and mast cells to type 2 inflammation. (A) The expression of the IL-33 receptor (ST2L) in sputum cells from asthmatics (n = 55), is tightly correlated with a combined metric, type-2-gene-mean, that represents the expression of IL4, IL5, and IL13. ST2L expression in sputum cells is strongly correlated with expression of mast cell and basophil proteases, carboxypeptidase A3 (CPA3), and tryptase among asthmatics. (B) ST2 is highly expressed on basophils in PBMCs from healthy and asthmatic subjects (one representative plot from healthy subject is shown). White histogram indicates surface ST2 staining, whereas gray indicates fluorescence-minus-one FMO control. (C) Basophil numbers are significantly higher in induced sputum cells from asthmatic subjects (n = 28) than in healthy controls (n = 16); the numbers of CD3⁺CD4⁻, CD3⁺CD4⁺, and ILC2 cells are not significantly different between groups. Data are presented as Tukey boxplots with median ± IQR. ***P < 0.001.

Fig. 6.

Transcriptomic analysis reveals basophils and mast cells as the target of IL-33 activity in asthma. (A) We performed in vitro stimulation of sputum cells with IL-2 and -3 (CTL) or IL-2, -3, -33, and TSLP (IL33/TSLP) from 12 asthmatic subjects. After 24 h, cells were collected, RNA was extracted, and RNA sequencing was performed. (B) Subjects were classified as responders (Resp) (n = 6) if the gene expression of IL5 and IL13 increased after stimulation. All others were classified as nonresponders (Non-Resp). (C) Gene set enrichment analysis reveals that the sputum cell transcriptomes of IL-33 responders are enriched in mast cell/basophil and B-cell gene transcripts, whereas IL-33 nonresponders are enriched in macrophage and neutrophil gene transcripts.