Sensory Neurons Promote Immune Homeostasis in the Lung

SUMMARY

Cytokines employ downstream Janus kinases (JAKs) to promote chronic inflammatory diseases. JAK1-dependent type 2 cytokines drive allergic inflammation, and patients with JAK1 gain-of-function (GOF) variants develop atopic dermatitis (AD) and asthma. To explore tissue-specific functions, we inserted a human JAK1 GOF variant (JAK1^GOF) into mice and observed the development of spontaneous AD-like skin disease, but unexpected resistance to lung inflammation when JAK1^GOF expression was restricted to the stroma. We identified a previously unrecognized role for JAK1 in vagal sensory neurons in suppressing airway inflammation. Additionally, expression of Calcb/CGRPβ, was dependent on JAK1 in vagal sensory neurons, and CGRPβ suppressed group 2 innate lymphoid cell function and allergic airway inflammation. Our findings reveal evolutionarily conserved but distinct functions of JAK1 in sensory neurons across tissues. This biology raises the possibility that therapeutic JAK inhibitors may be further optimized for tissue-specific efficacy to enhance precision medicine in the future.

In Brief

Janus kinase 1 signaling exerts distinct effects across cell types and tissues, with JAK1 signaling in vagal sensory neurons promoting anti-inflammatory responses in the lung.

Graphical Abstract

INTRODUCTION

Cytokine receptors commonly employ the Janus kinase (JAK)-signal transducers and activators of transcription (STAT) pathway to influence effector cell programs. Indeed, JAK inhibition has revolutionized the treatment of a number of inflammatory, myeloproliferative, and even infectious disorders.¹ Interest in the therapeutic potential of JAK inhibitors across the spectrum of allergic disorders is rapidly growing. In the past year, two JAK1-selective inhibitors were approved by the FDA for treatment of atopic dermatitis (AD), and several clinical trials for asthma treatment are underway.^2–6 However, little is known about the pathophysiologic consequences of JAK inhibition. Indeed, the precise mechanisms underlying major adverse events of JAK inhibitor treatment, including cardiovascular events (MACE), cancer, and opportunistic infections, remain poorly defined.⁷ Thus, understanding the tissue-specific role of JAK signaling across organs can simultaneously inform understanding of biology but also foreshadow how JAK inhibitors may be used across the evolving medical landscape.

Type 2 cytokines critically rely on downstream JAK1 signaling to promote allergic inflammation.⁸ However, much of what we know about JAK inhibition has derived from our understanding of cytokine signaling within the hematopoietic compartment.⁹ Although, in immune cells, JAK1 phosphorylation drives a proinflammatory effector program,¹ The role of JAK1 signaling within stromal cells (e.g., epithelial cells, fibroblasts, etc.) is just beginning to be uncovered. For example, sensory neuron-intrinsic JAK1 was found to play an unexpected role in promoting itch sensation, a difficult-to-treat component of AD pathology.¹⁰ Thus, probing the tissue- and cell-specific effects of JAK1 signaling is critical to understanding the molecular and cellular basis of allergic disease pathogenesis and treatment.

Patients with germline JAK1 gain-of-function (GOF) variants (JAK1^GOF) develop atopic disorders such as AD and asthma, as well as eosinophilia.^11,12 We sought to utilize a pathogenic human JAK1 GOF variant to probe the precise tissue-intrinsic mechanisms by which JAK1 signaling orchestrates various allergic pathologies. To do this, we generated a mouse line in which the endogenous murine Jak1 gene was replaced with a human JAK1 GOF mutation (JAK1^GOF mice) that was reported in 2017.¹¹ Although these mice exhibited spontaneous AD-like inflammation, as observed in patients, no lung pathology was evident at steady state. However, global JAK1^GOF mice exhibited enhanced allergic lung inflammation in response to the classic aeroallergen Alternaria alternata. Strikingly, restriction of JAK1^GOF expression to the stroma rendered mice resistant to allergic lung inflammation. Given recent reports that sensory neurons can suppress lung inflammation,^13–18 we hypothesized that sensory neurons mediate a key regulatory circuit through JAK1 signaling.

Indeed, we found that chemical denervation of lung tissue as well as conditional deletion of endogenous murine Jak1 in sensory neurons aggravated allergic lung inflammation. We screened various neuropeptides associated with neuroinflammation and discovered that expression of Jak1 is required for proper expression of Calcb, which encodes the neuropeptide calcitonin gene-related peptide β (CGRPβ), within the vagal ganglia (VG). Additionally, we found that CGRPβ suppresses pathogenic group 2 innate lymphoid cell (ILC2) responses and allergic lung inflammation. Finally, retrograde viral delivery of JAK1^GOF to the lung resulted in selective uptake into the VG and suppression of allergic lung inflammation. Similarly, conditional insertion of JAK1^GOF into sensory neurons also promoted immune homeostasis in the lung.

Collectively, our study highlights that while JAK1 signaling promotes skin inflammation and is a major therapeutic target for AD and itch, sensory neuron-intrinsic signaling of JAK1 serves an immunoregulatory role in the lung. We reveal a previously unrecognized role for neuronal JAK1 in regulating neuropeptide expression and neuroinflammation. These findings underscore the importance of understanding tissue-specific JAK signaling across mammalian systems and the potential utility of JAK inhibitors across the disease spectrum.

RESULTS

Germline human JAK1 GOF promotes spontaneous allergic inflammation in the skin, but not in the lung

The first patients harboring a germline JAK1 GOF mutation were identified in 2017 as an alanine-to-aspartate substitution at position 634 (A634D) within the inhibitory pseudokinase domain of the JAK1 protein. These patients exhibited a number of allergic pathologies, including AD, asthma, and profound eosinophilia (Figure 1A).¹¹ However, the severity of the patients’ AD and asthma were not reported in the original study. Thus, we examined the clinical characteristics of two patients when they originally presented with AD and asthma symptoms in more detail. Indeed, both patients with JAK1 GOF mutations presented with AD markedly earlier in life than with asthma (Figure 1B). Further, the overall relative severity of their AD (Figures 1C–E) was generally worse than the severity of their asthma as determined by quantitative and validated metrics such as Eczema Area and Severity Index (EASI) scores (Figure 1E), forced expiratory volume (FEV1) (Figure 1F), and forced vital capacity (FVC) (Figure 1G) measurements. These findings suggest that activation of JAK1 across multiple tissues results in differential tissue-specific outcomes.

Figure 1. Human germline JAK1 GOF mutation is associated with AD and asthma in patients and promotes spontaneous AD-like disease in mice.

(A) Human JAK1 c.1901C>A (p.A634D) GOF mutation within the pseudokinase domain.

(B) Timeline of AD and asthma diagnosis, EASI assessment, and PFTs.

(C) Baseline visual image of AD skin lesions from Patient A with a JAK1 GOF mutation.

(D) Baseline visual image of AD skin lesions from Patient B with a JAK1 GOF mutation.

(E) Baseline EASI scores from Patient A and Patient B.

(F) Baseline FEV1 of Patient A and Patient B.

(G) Baseline FVC of Patient A and Patient B.

(H) Schematic of the targeting strategy used to introduce the germline human JAK1 GOF mutation (JAK1^GOF) into the murine Jak1 locus.

(I) Representative ear skin images of WT control and JAK1^GOF mice.

(J) Ear thickness measurements of WT control and JAK1^GOF mice.

(K) Representative skin histology images (H&E) of WT control and JAK1^GOF mice. Scale bars indicate 100 μm.

(L-O) Flow cytometry of skin lymph nodes from WT control and JAK1^GOF mice. Shown are the frequencies of (L) total CD45⁺ immune cells, (M) CD4⁺ T cells, (N) ILC2s, and (O) eosinophils.

(P-Q) Representative lung histology images of WT control and JAK1^GOF mice, obtained by (P) H&E stain and (Q) PAS stain. Scale bars indicate 200 μm.

(R-U) Flow cytometry of lung tissue from WT control and JAK1^GOF mice. Shown are the frequencies of (R) total CD45⁺ immune cells, (S) CD4⁺ T cells, (T) ILC2s, and (U) eosinophils.

(J, L-O, R-U) Data were pooled from three independent experiments, n = 8–15 pooled mice per group. ^†p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± standard deviation (SD).

AD – atopic dermatitis, EASI – eczema area and severity index, FEV1 – forced expiratory volume, FVC – forced vital capacity, GOF – gain-of-function, ILC2s – group 2 innate lymphoid cells, H&E – hematoxylin and eosin, JAK1 – Janus kinase 1, PAS – periodic acid-Schiff, PFT – pulmonary function test, WT – wild-type.

See also Figure S1

To examine how JAK1 confers unique effects across tissues, we generated mice in which the endogenous murine Jak1 gene was replaced with the mutant human JAK1 GOF variant (A634D) (Figure 1H). Strikingly, soon after birth, heterozygous JAK1^GOF mice display spontaneous AD-like dermatitis with increased skin thickness compared to the skin of wild-type (WT) control mice (Figures 1I–J). Histologically, the JAK1^GOF mice exhibited hyperkeratosis, epidermal hyperplasia, and dermal inflammatory immune cell infiltrate that were not observed in control mice (Figure 1K). Further, the JAK1^GOF mice demonstrated a higher frequency of immune cell populations commonly associated with allergic inflammation, including total CD45⁺ immune cells (Figures 1L), CD4⁺ T cells (Figures 1M), ILC2s (Figures 1N), and eosinophils (Figures 1O) in the skin-draining lymph nodes compared to control mice at steady state. Taken together, these findings demonstrate that even one germline copy of the human JAK1 GOF allele is sufficient to drive AD-like pathology in mice, recapitulating the human skin phenotype.

Surprisingly, in contrast to the skin, the lungs of JAK1^GOF mice did not display spontaneous inflammatory pathology when compared to those from age-matched WT control counterparts. In both JAK1^GOF and control mice, assessment of airway remodeling and inflammatory infiltrate by hematoxylin and eosin (H&E) staining revealed no histopathologic evidence of lung inflammation (Figure 1P), and periodic acid-Schiff (PAS) staining revealed no evidence of goblet cell hyperplasia and mucus production (Figure 1Q). Further, the extent of CD45⁺ immune cell (Figures 1R and S1), CD4⁺ T cell (Figures 1S and S1), ILC2 (Figures 1T and S1), and eosinophil (Figures 1U and S1) enrichment was comparable if not reduced in the JAK1^GOF mice compared to control mice. Collectively, these findings provoke the hypothesis that the lung harbors resistance to JAK1-mediated inflammation that is not observed in the skin.

Stromal JAK1^GOF suppresses allergen-induced lung inflammation

We next examined whether challenging the airway with an allergen might provoke an enhanced inflammatory response not observed at steady state. To test this hypothesis, we treated both JAK1^GOF and control mice intranasally with the asthma-associated fungal allergen Alternaria alternata (Figure 2A). In this well-established model of allergic lung inflammation, ILC2s promote eosinophil responses and allergic lung pathology.^19–22 Indeed, histopathologic analysis revealed enhanced lung pathology in JAK1^GOF mice compared to controls as demonstrated by increased airway remodeling, inflammatory infiltrate, goblet cell hyperplasia, and mucus production upon provocation by the allergen (Figures 2B–C). Further, we found that Alternaria alternata-treated JAK1^GOF mice exhibit significantly elevated frequencies of ILC2s in their lungs compared to control mice (Figure 2D); however, we did not observe a difference in eosinophil frequency (Figure 2E). Taken together, these findings indicate that although global activation of JAK1 does not result in spontaneous allergic inflammation in the lung, provocation with an allergen renders JAK1^GOF mice more susceptible to some features of allergic lung pathology.

Figure 2. Stroma-intrinsic expression of JAK1^GOF is protective in allergic lung inflammation.

(A) Schematic of the allergic lung inflammation model using the protease allergen Alternaria alternata (ALT).

(B-C) Representative lung histology images of WT control and JAK1^GOF mice challenged with intranasal ALT, obtained by (B) H&E stain and (C) PAS stain.

(D-E) Flow cytometry of lung tissue from WT control and JAK1^GOF mice challenged with intranasal ALT. Shown are the frequencies of (D) ILC2s and (E) eosinophils. Data were obtained from three independent experiments, n = 5 mice per group.

(F) Schematic of experimental approach used to generate chimeric mice. BM derived from WT mice was transferred into WT control and experimental JAK1^GOF mice.

(G) JAK1 genotype of the respective hematopoietic and stromal cell components in chimeric mice (WT→WT and WT→JAK1^GOF).

(H) Schematic of experimental approach used to generate the allergic lung inflammation model in the BM chimeras.

(I-J) Representative lung histology images of chimeric WT→WT and WT→JAK1^GOF mice challenged with intranasal ALT, obtained by (I) H&E stain and (J) PAS stain.

(K-L) Flow cytometry of lung tissue from chimeric WT→WT and WT→JAK1^GOF mice challenged with intranasal ALT. Shown are the frequencies of (K) ILC2s and (L) eosinophils. Data were obtained from three independent experiments, n = 11–14 pooled mice per group.

(M) KEGG pathway overrepresentation analysis of RNA-seq data obtained from the lung tissues of chimeric WT→JAK1^GOF vs. WT→WT mice (n = 2 mice/group). Overrepresentation analysis used Fisher’s exact test to determine the significance of enrichment. All terms shown were identified as the top four significantly suppressed pathways in terms of adjusted p-value (p-adjusted < 0.05, Benjamini-Hochberg procedure).

**p < 0.01, *p < 0.05, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(A, F-G) Figure adapted from an image created with BioRender.com.

ALT- Alternaria alternata, BM – Bone marrow, BMT – Bone marrow transplantation.

See also Figure S1

Given the resistance to JAK1-mediated inflammation observed in the lung at steady state, we hypothesized that JAK1 activation within the lung stroma may confer tissue-specific resistance even upon allergen challenge. To test this hypothesis, we generated bone marrow (BM) chimeric mice in which we reconstituted both control and JAK1^GOF mice with WT BM (WT→WT and WT→JAK1^GOF) (Figures 2F–G). Upon airway challenge with Alternaria alternata (Figure 2H), WT→JAK1^GOF mice exhibited markedly reduced lung pathology as evidenced by reduced peribronchial and perivascular inflammatory infiltrates (Figure 2I), goblet cell hyperplasia, and mucous production (Figure 2J) when compared to control WT→WT mice. Further, the frequencies of ILC2s (Figure 2K) and eosinophils (Figure 2L) were significantly reduced in WT→JAK1^GOF mice compared to control WT→WT mice. Strikingly, KEGG pathway overrepresentation analysis of RNA-seq data obtained from lung tissue showed that expression of asthma- and inflammation-associated pathways were significantly suppressed in WT→JAK1^GOF mice compared to control WT→WT mice (Figure 2M). While it is well-appreciated that JAK1^GOF activates lymphocytes, these unanticipated findings indicate that stromal expression of JAK1^GOF confers resistance to lung inflammation.

Sensory neurons suppress allergic lung inflammation

Recent studies demonstrated that neuropeptides can restrain ILC2 responses and lung inflammation.^16,18 Thus, we hypothesized that sensory neurons in the lung may act to suppress inflammation. In contrast to the skin, which is almost entirely innervated by sensory neurons arising from the dorsal root ganglia (DRG), the lung is primarily innervated by sensory afferents that arise from the VG, with minor innervation from the DRG²³ (Figure 3A). The majority of sensory neurons from the DRG express canonical nociceptive markers such as TRPV1 and Na_v1.8.^24–26 Sensory neurons in the vagus nerve exhibit expression of nociceptive ion channels comparable to that observed in DRG-derived neurons,²⁷ as evidenced in our reanalysis of recently published single cell RNA-seq (scRNA-seq) datasets from vagal afferents²⁸ (Figures 3B–C). To visualize the extent of TRPV1 and Na_v1.8 expression within the VG and DRG, we generated reporter mice by crossing Trpv1^Cre and Scn10a^Cre, respectively, to Rosa26^{STOPflox-tdTomato} mice. Imaging of these reporter mice revealed widespread expression of both TRPV1 and Na_v1.8 in both ganglia (Figures 3D–E). These findings suggested that sensory neurons within the VG can be chemically, genetically, and virally targeted.

Figure 3. Chemical denervation of TRPV1⁺ sensory neurons exacerbates allergic lung inflammation.

(A) Visual representation of the sensory innervation of the lung arising from the VG and DRG.

(B-C) Expression of Trpv1 and Scn10a (the gene encoding Na_v1.8) within the VG, specifically the (B) JG, and (C) NG. Trpv1 expression is shown as the percentage of total neurons sequenced. Scn10a expression is shown as the percentage of Trpv1-expressing neurons. The full scRNA-seq dataset is available in Kupari et al.²⁸

(D-E) Imaging of classical nociceptive markers (TRPV1 and Na_v1.8) in the VG and DRG using Rosa26^{STOPflox-tdTomato} mice crossed with the respective Trpv1^Cre and Scn10a^Cre mice. (D) TRPV1 in the VG and DRG, (E) Na_v1.8 in the VG and DRG.

(F) Schematic of the allergic lung inflammation model generated following systemic denervation of vagal and spinal visceral afferents.

(G) Withdrawal latency from noxious heat (hot plate test). Mice treated with s.c. vehicle control or RTX were placed on a hot plate set at 50 degrees Celsius, and withdrawal latency was measured as the time until the appearance of paw withdrawal behavior.

(H-I) Representative lung histology images from mice treated with s.c. vehicle control or RTX challenged with intranasal ALT, obtained by (H) H&E stain and (I) PAS stain.

(J-K) Flow cytometry of lung tissue from mice treated with s.c. vehicle control or RTX and challenged with intranasal ALT. Shown are the frequencies of (J) ILC2s and (K) eosinophils. Data were obtained from three independent experiments, n = 4–5 mice per group.

(L) Schematic of the allergic lung inflammation model generated following selective denervation of vagal afferents.

(M) Withdrawal latency from noxious heat (hot plate test). Mice treated with intraganglionic (i.g.) vehicle control or RTX were placed on a hot plate set at 50 degrees Celsius, and withdrawal latency was measured as the time until the appearance of paw withdrawal behavior.

(N-O) Representative lung histology images of mice treated with i.g. vehicle control or RTX challenged with intranasal ALT, obtained by (N) H&E stain and (O) PAS stain.

(P-Q) Flow cytometry of lung tissue from mice treated with i.g. vehicle control or RTX challenged with intranasal ALT. Shown are the frequencies of (P) ILC2s and (Q) eosinophils. Data were obtained from two independent experiments, n = 4–5 mice per group.

**p < 0.01, *p < 0.05, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(A, F, L) Figure adapted from an image created with BioRender.com.

DRG – dorsal root ganglia, i.g. – intraganglionic, JG – jugular ganglia, NG – nodose ganglia, RTX – resiniferatoxin, s.c. – subcutaneous VG – vagal ganglia

See also Figures S1 and S2

TRPV1-expressing neurons can be chemically denervated by treatment with resiniferatoxin (RTX), an ultrapotent TRPV1 agonist.^29,30 The majority of lung sensory innervation arises from vagal afferents while only a small proportion originates from DRG-derived spinal sensory neurons (i.e., spinal visceral afferents).²³ Therefore, we first examined whether TRPV1⁺ neurons, including vagal and spinal sensory neurons, contribute to suppressing lung inflammation. WT C57BL/6 mice were subcutaneously (s.c.) injected with either vehicle or RTX; the latter leads to denervation of both vagal and spinal sensory neurons³¹ (Figure 3F). As expected, s.c. RTX treatment led to delayed paw withdrawal (enhanced withdrawal latency) in response to heat stimulation, indicating loss of noxious sensation (Figure 3G). Strikingly, mice chemically denervated of all sensory neurons, when compared to control mice, had increased peribronchial and perivascular inflammatory infiltrates (Figure 3H), goblet cell hyperplasia, and mucous production (Figure 3I). Further, the frequencies of ILC2s (Figure 3J) and eosinophils (Figure 3K) in the RTX-treated mice were significantly elevated as compared to vehicle-treated control mice. Additionally, transcriptomic analysis of lung tissue revealed upregulation of allergic inflammation-related genes (Figure S2A) and overrepresentation analysis of KEGG pathway revealed an enrichment of asthma- and inflammation-related pathways in the RTX-treated mice as compared to controls (Figure S2B).

We next investigated whether TRPV1⁺ sensory neurons, arising specifically from the VG, could regulate lung inflammation. To test this possibility, we performed intraganglionic (i.g.) injection of RTX or vehicle control to induce selective loss of sensory neurons from the VG (Figure 3L and S2C). Indeed, i.g. administration of RTX resulted in selective suppression of Trpv1 expression in the VG, while expression within DRG remained unchanged (Figure S2D–E). Further, chemical denervation of VG-specific sensory neurons did not affect paw withdrawal latency by the hot plate test (Figure 3M). However, denervation of VG-specific sensory neurons significantly affected Alternaria alternata-induced lung inflammation, which was evident from increased levels of peribronchial and perivascular inflammatory infiltrates (Figure 3N), goblet cell hyperplasia, and mucous production (Figure 3O) as well as increased frequencies of ILC2s (Figure 3P) and eosinophils (Figure 3Q) in the i.g. RTX-treated mice compared to controls.

We next tested the contribution of spinal sensory neurons from the DRG in regulating lung inflammation. RTX or vehicle was administered intrathecally (i.t.), (Figure S2F) to selectively impair spinal DRG sensory neurons.³¹ Indeed, i.t. RTX injection resulted in increased paw withdrawal latency to heat, indicating loss of sensation (Figure S2G). Impairment of DRG sensory neurons did not significantly alter the pathophysiology of Alternaria alternata-induced lung inflammation as histopathological analysis revealed no differences in the levels of peribronchial and perivascular inflammatory infiltrates (Figure S2H), goblet cell hyperplasia, and mucous production (Figure S2I) or in the frequencies of ILC2s (Figure S2J) and eosinophils (Figure S2K) between i.t. RTX- and vehicle-treated control mice. Transcriptomic analysis revealed very few differentially expressed genes (Figure S2L) and KEGG pathway overrepresentation analysis found no specific pathway enrichment induced by i.t. RTX treatment (Figure S2M). Collectively, these findings demonstrate that vagal sensory neurons critically suppress allergic lung inflammation.

Sensory neuron-intrinsic Jak1 regulates lung inflammation and levels of neuropeptides

Sensory neurons from the DRG express Jak1 to mediate critical functions such as itch.¹⁰ However, whether Jak1 is expressed in the VG has not been well addressed. Thus, we reanalyzed scRNA-seq datasets from the VG²⁸ and found that the majority of Na_v1.8⁺ sensory neurons co-express Jak1 (Figures 4A–B). We next hypothesized that Jak1 expression within vagal sensory neurons may be required for immune homeostasis in the lung. To test this, we employed mice that conditionally lack Jak1 in Na_v1.8⁺ neurons (Jak1^Δneuron mice; Scn10a^Cre:Jak1^flox) (Figure 4C). Following allergen challenge in the airway, we found that all parameters of inflammation were significantly enhanced in Jak1^Δneuron mice compared to controls, including histopathologic features of lung inflammation (Figures 4D–E) and frequencies of ILC2s (Figure 4F) and eosinophils (Figure 4G). These findings indicate that JAK1 signaling in Na_v1.8⁺ sensory neurons regulates allergic inflammation in the lung.

Figure 4. Disruption of sensory neuron-intrinsic Jak1 exacerbates allergic lung inflammation and alters the expression levels of neuropeptides.

(A-B) Jak1 expression in Na_v1.8⁺ vagal afferents, (A) JG and (B) NG. Gene expression is shown as the percentage of Na_v1.8⁺ neurons that express Jak1. The full scRNA-seq dataset is available in Kupari et al.²⁸

(C) Schematic of the allergic lung inflammation model using littermate control (Jak1^flox) and experimental Jak1^Δneuron mice.

(D-E) Representative lung histology images of control and Jak1^Δneuron mice challenged with intranasal ALT, obtained by (D) H&E stain and (E) PAS stain.

(F-G) Flow cytometry of lung tissue from control and Jak1^Δneuron mice challenged with intranasal ALT. Shown are the frequencies of (F) ILC2s and (G) eosinophils. Data were obtained from two independent experiments, n = 10–12 pooled mice per group.

(H-I) Profiling of neuropeptide expression in Na_v1.8⁺ vagal afferents; Calca, Calcb, Tac1, Vip, and Nmu were evaluated in the (H) JG and (I) NG. Gene expression is shown as the percentage of Na_v1.8⁺ neurons that express the indicated gene. The full scRNA-seq dataset is available in Kupari et al.²⁸

(J-K) Imaging of the classical neuropeptides CGRPα and CGRPβ in the VG and DRG using Rosa26^{STOPflox-tdTomato} mice crossed with Calca^Cre and Calcb^Cre/ERT mice, respectively. Shown are (J) CGRPα in the VG and DRG and (K) CGRPβ in the VG and DRG.

(L) MotifMap (human)-predicted neuropeptide targets of STAT family members. Data were analyzed using the MotifMap Predicted Transcription Factor Targets dataset (see Methods).

(M-R) Quantitative PCR (qPCR) of transcripts from the VG and DRG from control and Jak1^Δneuron mice challenged with intranasal ALT. Shown are (M, N) Calcb, (O, P) Calca, and (Q, R) Tac1 transcript levels. Data were obtained from two independent experiments, n = 3–4 mice per group.

**p < 0.01, *p < 0.05, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(C) Figure adapted from an image created with BioRender.com.

See also Figures S1 and S3–5

We previously showed that both JAK1 and upstream IL-4Rα on spinal sensory neurons critically regulate itch in the setting of skin inflammation.¹⁰ To test whether these pathways exhibit similar overlapping neural functions in the airway, we also subjected Il4ra^Δneuron mice (conditional deletion of Il4ra on Na_v1.8⁺ sensory neurons; Scn10a^Cre:Il4ra^flox) to allergic lung inflammation (Figure S3A). However, we found no effect of Il4ra deletion in this setting (Figures S3B–F). Thus, despite the presence of a neuronal IL-4Rα-JAK1 itch axis in the skin, IL-4Rα appears dispensable in lung sensory neurons in regulating allergic lung inflammation. Further, given the striking nature of the lung pathology following conditional deletion of Jak1 in sensory neurons, we sought to test whether neuronal JAK1 signaling may be relevant across multiple model systems. Thus, we employed an alternative mouse model of allergic lung inflammation induced by the protease allergen papain. Following papain challenge in the airway of Jak1^Δneuron mice (Figure S3G), we observed enhanced histopathologic features of lung inflammation (Figures S3H–I) and increased frequencies of CD45⁺ immune cells (Figure S3J), ILC2s (Figure S3K), and eosinophils (Figure S3L) compared to control mice.

Although our findings demonstrate that neuron-intrinsic JAK1 signaling is required for homeostasis in the face of allergic lung inflammation, whether this function is impacted by JAK1-selective inhibitors in humans remains unclear. Strikingly, recent phase 1 and 2 clinical trials with multiple inhalational JAK1-selective inhibitors have failed in asthma (NCT03766399, NCT04150341, ACTRN12617001227381). Thus, we hypothesized that inhalational delivery of JAK1-selective inhibitors may display limited efficacy in vivo due to its restriction to the airway and thus preferential access to sensory neurons; in contrast, systemic delivery of JAK1-selective inhibitors may yet derive efficacy due to their capacity to potently target the hematopoietic compartment. To test this hypothesis, we compared the efficacy of upadacitinib, a JAK1-selective inhibitor, administered either intranasally (inhalational) or intraperitoneally (systemic) in the Alternaria alternata-induced allergic lung inflammation model (Figure S4A). Although both inhalational and systemic delivery of upadacitinib significantly reduced the number of eosinophils in the lung (Figure S4B), only systemic treatment resulted in suppression of ILC2s (Figure S4C). Further, only systemic delivery of upadacitinib resulted in an overall improvement of lung pathology (Figure S4D), goblet cell hyperplasia, and mucous content (Figures S4E). Collectively, these data suggest that inhalational delivery of JAK1-selective inhibitors may have limited efficacy.

Increasing evidence suggests that sensory neurons release neuropeptides to modulate inflammation at multiple barrier surfaces.^{13,16,18,32–34} However, the intracellular mechanisms guiding such neuronal responses remain poorly understood. Although JAK1 has been recently appreciated for its ability to relay sensory signals toward the spinal cord (afferent function), whether JAK1 controls expression of neuropeptides in sensory neurons to modulate neuroinflammation in tissues (efferent function) has not been explored. Thus, we sought to screen the expression of various neuropeptides within the VG using existing scRNA-seq datasets.²⁸ Indeed, both the jugular ganglia (JG) and nodose ganglia (NG) express the classical neuropeptide Calca, which encodes the α form (CGRPα) of the neuropeptide used to identify peptidergic neurons (Figures 4H–I). Recent studies have demonstrated the capacity of CGRPα to suppress ILC2 responses in the lung.^16,18 However, the expression of Calcb (the gene encoding CGRPβ) and the function of CGRPβ in the lung remain poorly understood. We also screened other neuropeptide genes, including Tac1 (the gene encoding Substance P [SP]), Vip (the gene encoding vasoactive intestinal peptide [VIP]), and Nmu (the gene encoding neuromedin U [NMU]) (Figures 4H–I). To visualize the extent of Calca and Calcb expression within the VG and DRG, we generated reporter mice by crossing Calca^Cre and Calcb^Cre/ERT mice, respectively, to Rosa26^{STOPflox-tdTomato} mice. Strikingly, imaging of these reporter mice revealed widespread expression of both Calca and Calcb in both the VG and DRG (Figures 4J–K).

The ability of JAK1 to alter cellular programs is largely dependent on downstream phosphorylation of STAT proteins, which enter the nucleus to alter transcription. Indeed, overactivity of STAT6 is strongly associated with allergic inflammatory processes, and STAT6 GOF was recently reported to be associated with similar atopic syndromes as observed for JAK1 GOF.^35–40 We sought to examine whether STAT6 is associated with regulation of specific neuropeptide regulatory elements via in silico analysis of datasets from MotifMap Predicted Transcription Factor Targets (see Methods). We found that CALCB and TAC1, but not CALCA, NMU, and VIP, were predicted transcriptional targets of STAT6 (Figure 4L). Further, analysis of transcription factor DNA-binding motifs revealed a STAT6 DNA-binding motif ~500–2000 bp upstream of CALCB, TAC1, and CALCA, but not NMU, and VIP, suggesting that STAT6 may selectively regulate expression of genes that encode CGRP and SP. Thus, we hypothesized that JAK1 may regulate CGRPβ and/or SP.

We next subjected both Jak1^Δneuron and control mice to airway allergen challenge and examined the VG for expression of neuropeptide transcripts (Figure S5A). Strikingly, we found that expression of Calcb was decreased within the VG (Figure 4M) of Jak1^Δneuron mice compared to control mice. However, this Calcb deficiency was not observed within the DRG (Figure 4N). While no alterations were observed in Calca and Vip expression (Figures 4O–P and S5B–C), Tac1 was reduced in both the VG and DRG of Jak1^Δneuron mice (Figures 4Q–R); Nmu was only reduced in the DRG of Jak1^Δneuron mice (Figures S5D–E). In contrast to Jak1^Δneuron mice, JAK1^GOF mice exhibited increased expression of Calcb, but not Tac1, in their VG compared to control mice (Figures S5F–H). Taken together, these findings indicate that JAK1 regulates the expression of multiple neuropeptides, including Calcb, in the VG. Given these findings, along with our in silico analysis predicting CALCB as a transcriptional target of STAT6, we hypothesized that CGRPβ may be a key regulatory neuropeptide controlled by JAK1.

In light of our findings that both chemical denervation via i.g. RTX (Figures 3L–Q) and sensory neuron-intrinsic deletion of Jak1 (Figures 4C–G) results in enhanced allergic lung inflammation, we sought to test whether RTX-mediated denervation of TRPV1⁺ sensory neurons impact CGRPβ expression within the vagus nerve. To test this, we treated Calcb-reporter (Calcb^Cre/ERT:Rosa26^{STOPflox-tdTomato}) mice with i.g. RTX and performed immunofluorescence imaging of total CGRPβ⁺ neurons (Figure S5I). Strikingly, i.g. RTX-mediated denervation demonstrated a marked reduction of CGRPβ⁺ neurons compared to the vehicle-treated mice (Figure S5J–K)

CGRPβ suppresses ILC2 function and allergic lung inflammation

Given that ILC2s are important drivers of allergic lung inflammation^20,41 and the role of CGRPβ in this context is not well understood, we sought to test whether CGRPβ can directly suppress the function of ILC2s in the lungs. We sort-purified lung ILC2s from WT C57BL/6 mice and stimulated them in vitro with the alarmin cytokine IL-33, in combination with either CGRPβ or SP (Figure 5A). Strikingly, we found that CGRPβ potently suppressed IL-33-mediated stimulation of IL-5 and IL-13 production from ILC2s (Figures 5B–C). In contrast, the classical proinflammatory neuropeptide SP had no effect on IL-33-elicited activation of lung ILC2s (Figures 5B–C). Similar to CGRPβ, CGRPα also suppressed ILC2 function (Figures S6A–C), a well-known feature of this isoform.^16,18 However, of these two isoforms, only Calcb was found to be differentially regulated in the VG of Jak1^Δneuron mice (Figures 4M–P). Our findings demonstrate that CGRPβ can directly suppress ILC2 function and suggest its potential role in restraining allergic lung inflammation.

Figure 5. CGRPβ suppresses type 2 cytokine production from lung ILC2s and allergic lung inflammation.

(A) Schematic of the lung ILC2 suppression assay.

(B-C) Type 2 cytokine measurement from lung ILC2s stimulated with IL-33, SP, and/or CGRPβ.

Shown are data for (B) IL-5 and (C) IL-13.

(D) Schematic of the induction of allergic lung inflammation in Rag1^−/− mice.

(E-F) Flow cytometry of lung tissue from Rag1^−/− mice challenged with intranasal ALT as well as vehicle control or CGRPβ or CGRPβ 8–37. Shown are the frequencies of (E) ILC2s and (F) eosinophils. Data were obtained from two independent experiments, n = 4–5 mice per group.

(G-H) Representative lung histology images of Rag1^−/− mice challenged with intranasal ALT as well as vehicle control or CGRPβ or CGRPβ 8–37, obtained by (G) H&E stain and (H) PAS stain.

***p < 0.001, **p < 0.01, *p < 0.05 (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(A, D) Figure adapted from an image created with BioRender.com.

ELISA – enzyme-linked immunosorbent assay, IL – interleukin, SP – substance P

See also Figures S1 and S6

Allergic lung inflammation is uniquely dependent on ILC2s in adaptive lymphocyte-deficient Rag1^−/− mice.^19,41 Thus, to specifically evaluate whether exogenous CGRPβ could suppress ILC2 responses in vivo and allergic lung inflammation, we challenged Rag1^−/− mice with intranasal Alternaria alternata and simultaneously treated them with either intranasal CGRPβ or the CGRP receptor antagonist CGRPβ 8–37 (Figure 5D). Strikingly, while Alternaria alternata-treated and control vehicle-treated Rag1^−/− mice developed robust ILC2 (Figure 5E) and eosinophil responses (Figure 5F) in association with lung pathology (Figures 5G–H), CGRPβ-treated mice exhibited attenuated immune cell responses (e.g., eosinophils) and tissue inflammation (Figures 5E–H). Conversely, treatment with the CGRPβ receptor antagonist resulted in aggravated allergic lung inflammation (Figures 5E–H). Consistent with our in vitro findings (Figures 5A–C), SP did not suppress allergic lung inflammation (Figures S6D–H). To test whether CGRPβ displayed a suppressive function in the presence of adaptive immunity, we also delivered it to the airways of WT mice with allergic lung inflammation and observed similar anti-inflammatory effects to those observed in Rag1^−/− mice (Figures S6I–M). Collectively, these findings indicate that while neuron-intrinsic expression of Jak1 selectively regulates expression of Calcb in the VG (Figure 4M), CGRPβ, in turn, has the capacity to regulate lung inflammation.

Retrograde viral delivery and conditional insertion of human JAK1 GOF into sensory neurons are protective against allergic lung inflammation

Next, we sought to test whether the delivery of human JAK1 GOF specifically to lung-innervating neurons would be sufficient to suppress allergic lung inflammation. To test this hypothesis, we introduced retrograde Adeno-Associated Virus (AAV) co-expressing Cre recombinase and the human JAK1 GOF variant into Rosa26^{STOPflox-tdTomato} mice and subjected them to Alternaria alternata allergen challenge in the airway (Figure 6A). We found that most reporter-labeled neurons were housed in the VG and were virtually undetectable in the corresponding thoracic DRG (Figures 6B–C and Videos S1–2). These findings indicate that vagal sensory neurons were infected with AAV and thus harbored the JAK1 GOF variant. Strikingly, histopathological analysis revealed that mice infected with AAV-Cre/JAK1^GOF in the airway exhibited robust suppression of allergic lung inflammation (Figures 6D–E) and reduction in the frequencies of CD45⁺ immune cells (Figure 6F), ILC2s (Figure 6G), and eosinophils (Figure 6H).

Figure 6. Expression of human JAK1^GOF in lung sensory neurons suppresses allergic inflammation.

(A) Schematic of the allergic lung inflammation model generated by delivery of AAV-Cre (control) or AAV-Cre/JAK1 ^GOF (experimental) into the airways of Rosa26^{STOPflox-tdTomato} mice.

(B) Imaging of tdTomato in the DRG and VG following AAV-assisted gene delivery into lung-innervating sensory neurons. Shown are representative images of DRG and VG from mice treated with AAV-Cre/JAK1^GOF.

(C) Numbers of tdTomato⁺ neurons in DRG (thoracic segments 4–8) and VG, n = 5 mice per group. Data were analyzed in mice treated with AAV-Cre/JAK1 GOF.

(D-E) Representative lung histology images of mice challenged with intranasal ALT following AAV-Cre control and AAV-Cre/JAK1^GOF infection, obtained by (D) H&E stain and (E) PAS stain.

(F-H) Flow cytometry of lung tissue from mice challenged with intranasal ALT following AAV-Cre control and AAV-Cre/JAK1^GOF infection. Shown are the frequencies of (F) total CD45⁺ immune cells, (G) ILC2s, and (H) eosinophils. Data were obtained from two independent experiments, n = 4–5 mice per group.

(I) Schematic of the allergic lung inflammation model generated by conditional insertion of JAK1^GOF variant into sensory neurons.

(J-K) Representative lung histology images of mice challenged with intranasal ALT obtained by (J) H&E stain and (K) PAS stain.

(L-N) Flow cytometry of lung tissue from mice challenged with intranasal ALT. Shown are the frequencies of (L) total CD45⁺ immune cells, (M) ILC2s, and (N) eosinophils.

(O) Protein levels of CGRPβ in the bronchoalveolar lavage fluid (BALF). Data were obtained from two independent experiments, n = 4–5 pooled mice per group.

*p < 0.05, **p < 0.01, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(A, I) Figure adapted from an image created with BioRender.com.

AAV – adeno-associated virus; specifically, AAV2-retro, BALF – Bronchoalveolar lavage fluid

See also Figure S1, Video S1, and S2

Finally, we sought to test whether conditional insertion of the same human JAK1 GOF variant into sensory neurons would be sufficient to suppress allergic lung inflammation as well. Thus, we crossed sensory neuron-specific Scn10a^Cre mice with Jak1^{STOPflox-JAK1 GOF} (JAK1^GOF:neuron) mice. These mice are conditionally deficient in murine Jak1, but harbor human JAK1^GOF only within sensory neurons. We then subjected these JAK1^GOF:neuron mice to Alternaria alternata allergen challenge in the airway (Figure 6I). Indeed, similar to retrograde viral delivery of the JAK1 GOF variant, histopathological analysis exhibited robust suppression of allergic lung inflammation (Figures 6J–K) and reduction in the frequencies of CD45⁺ immune cells (Figure 6L), ILC2s (Figure 6M), and eosinophils (Figure 6N). Strikingly, the levels of CGRPβ protein in the airway bronchiolar lavage fluid (BALF) was significantly increased compared to control mice (Figure 6O). Collectively, these findings demonstrate that JAK1 regulates CGRPβ in sensory neurons and promotes immune homeostasis in the lung.

DISCUSSION

The canonical type 2 cytokines IL-4, IL-5, and IL-13 are universally elevated across both AD (eczema) and asthma.⁴² However, studies of targeted monoclonal antibody (mAb)-based therapies strongly suggest that although IL-13 is pathogenic in AD, it is a poor target in asthma.^43–47 Conversely, while IL-5 is a driver of asthma, it has failed as a therapeutic target in AD.^48–51 Taken together, these findings suggest that elevated signaling of the same cytokine pathways may imprint different phenotypes across tissues like the skin and lung. Further, while type 2 cytokines share and employ downstream JAK1 signaling to activate different cellular programs,⁵² the mechanisms by which JAK1 activity can result in different allergic phenotypes remain unclear. Strikingly, even JAK1 variants that are common in the population are associated with AD and eosinophil enrichment but not asthma (see GWAS-based analysis in Methods).⁵³ Further, while multiple JAK1-selective inhibitors are FDA approved for AD, none are approved for asthma, and multiple inhalational JAK1-selective inhibitors have failed in clinical trials (NCT03766399, NCT04150341, ACTRN12617001227381). These findings suggest that JAK1 signaling has tissue-specific properties that uniquely impact particular diseases and their associated therapies.

By transposing the human JAK1 GOF allele into mice, we found that while AD-like disease develops spontaneously, activation of JAK1 in vagal sensory neurons confers resistance to allergic lung inflammation. Although patients with JAK1^GOF spontaneously develop asthma,^11,12 this phenotype was not observed in our JAK1^GOF mice. We hypothesize that this difference may be due to limited exposure to environmental allergens in laboratory animals. Indeed, protease allergens such as house dust mite can directly stimulate sensory neurons to induce the release of neuropeptides.^54,55 Indeed, only upon challenge with the allergen Alternaria alternata, did we observe unique susceptibility to allergic lung inflammation in JAK1^GOF mice. These findings suggest that, independent of immune hypersensitivity, allergens may disrupt neuroimmune circuits to promote inflammation.

While JAK1 activation in lymphocytes is broadly proinflammatory, our findings suggest that JAK1 activation within the sensory nervous system serves a previously unrecognized immunoregulatory role. This observation may explain why JAK1-selective inhibitors, while highly successful in AD, have not advanced to treatment in asthma.¹ Our findings highlight the need to better understand the cell-specific mechanisms by which cytokines imprint distinct JAK-STAT programs and pathology across different organ systems.

Employing a variety of different chemical, genetic, and viral neuronal manipulation approaches, we found that sensory neuron-intrinsic expression of murine Jak1 or exogenous human JAK1^GOF is required to protect against allergen-induced lung inflammation. Further, we found that vagal sensory nerves are the key drivers of this homeostatic mechanism in the airway. By screening various neuropeptides associated with neuroinflammation, we find that JAK1 regulates Calcb expression in the VG, and CGRPβ directly suppresses lung inflammation. While recent studies have shown that a number of neuropeptides including NMU, SP, and VIP have proinflammatory functions in the airway,^56–58 CGRPα has demonstrated anti-inflammatory properties.^14,16–18 However, the role of CGRPβ is less understood. Thus, our findings highlight a previously unrecognized and selective role for the CGRPβ isoform in suppressing allergic lung inflammation. Future exploration into the differences between CGRPα and CGRPβ isoforms across tissues may yield deeper insights into barrier immunity and inflammation.

Understanding how various neuropeptides regulate inflammation in the skin, lung, and gut is a major emerging field of neuroimmunology.^{13,16,18,32–34} However, the cytokine signals that drive specific neuropeptide programs within these neuroimmune circuits remain poorly understood. The mechanisms by which JAK1 influences the transcription and release of CGRPβ remain unknown. Based on our in silico analyses, we suspect that downstream phosphorylation of STAT6 is likely a key mechanism that drives Calcb transcription. However, we also speculate that JAK1 may regulate the release of preformed CGRPβ from the synaptic vesicle in the axon terminal. Thus, JAK1 may have multiple functions within sensory neurons that are both STAT6-independent (CGRPβ protein release) and -dependent (Calcb transcription). Future studies aimed at understanding these pathways may reveal functions of JAK1 that are specific to sensory neurons and likely not observed in lymphocytes. Although our study reveals a mechanism by which sensory neurons employ JAK1 as a putative cytokine ‘sensor,’ discovery and characterization of additional JAK-STAT variants may reveal a highly dynamic landscape by which sensory neurons sense, process, and modulate inflammatory signals in tissues.

Sensory nerves consist of a highly sophisticated system that innervates virtually every organ in the body, including secondary lymphoid organs.^59,60 Although traditionally thought to be only sensory in function, sensory neurons have been found in recent studies to promote barrier immunity to pathogens, regulate metabolism, and even contribute to cancer immunosurveillance.^32,61–65 Thus, our observation that they suppress lung inflammation further underscores an emerging paradigm in which the sensory nervous system may regulate a variety of processes across the organism. Taken together, sensory neuromodulation may soon represent a therapeutic approach to treat a variety of inflammatory, infectious, metabolic, and even malignant disorders.

Limitations of the study

Although we identified a critical link between JAK1 within vagal sensory neurons and its ability to regulate Calcb and allergic lung inflammation, our study has limitations. First, the precise mechanisms by which neuron-intrinsic JAK1 modulates transcription, synthesis, and/or release of Calcb/CGRPβ remain unknown. Second, the neurons that express JAK1 within the human VG remain unknown. Third, whether inhibition of human vagal neuron-associated JAK1 alters neuropeptide expression has not been explored. Future studies in primary or iPSC-derived human vagal sensory neurons will be required to understand how our findings translate to human biology.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Brian S. Kim (itchdoctor@mountsinai.org).

Materials availability

JAK1 GOF mice will be made available upon request.

Data and code availability

No original codes were generated in this study.

Accession number is listed in the key resources table to account for all high throughput sequencing data.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-mouse CD16/CD32	Bio X Cell	Cat#: BE0307; RRID: AB_1107647
anti-mouse CD16/32	Biolegend	Cat#: 101302; RRID: AB_312801
anti-mouse CD3	BioLegend	Cat#: 100268; RRID: AB_2876392
anti-mouse CD3	BioLegend	Cat#: 100247; RRID: AB_2572117
anti-mouse CD3e	BioLegend	Cat#: 100304; RRID: AB_312669
anti-mouse CD4	BioLegend	Cat#: 100404; RRID: AB_312689
anti-mouse CD4	BioLegend	Cat#: 100449; RRID: AB_2564587
anti-mouse CD8a	BioLegend	Cat#: 100704; RRID: AB_312743
anti-mouse CD8a	BioLegend	Cat#: 100782; RRID: AB_2819775
anti-mouse/human CD11b	BioLegend	Cat#: 101204; RRID: AB_312787
anti-mouse/human CD11b	BioLegend	Cat#: 101233; RRID: AB_10896949
anti-mouse CD11c	BioLegend	Cat#: 117304; RRID: AB_313773
anti-mouse CD11c	BioLegend	Cat#: 117326; RRID: AB_117326
anti-mouse CD11c	BioLegend	Cat#: 117339; RRID: AB_2562414
anti-mouse CD19	Invitrogen	Cat#: 45-0193-82; RRID: AB_1106999
anti-mouse CD19	Biolegend	Cat#: 115504; RRID: AB_313639
anti-mouse CD45	Invitrogen	Cat#: 58-0451-82; RRID: AB_11218871
anti-mouse CD45	Biolegend	Cat#: 103138; RRID: AB_2563061
anti-mouse CD45R/B220	BioLegend	Cat#: 103275; RRID: AB_2860602
anti-mouse CD90.2	BD Biosciences	Cat#: 740442; RRID: AB_2740169
anti-mouse CD117 (c-Kit)	BioLegend	Cat#: 105838; RRID: AB_2616739
anti-mouse CD117 (c-Kit)	Invitrogen	Cat#: 62-1171-82; RRID: AB_2637141
anti-mouse CD127 (IL-7Ra)	BioLegend	Cat#: 135013; RRID: AB_1937266
anti-mouse CD127 (IL-7Ra)	BioLegend	Cat#: 135035; RRID: AB_2564577
anti-mouse CD127 (IL-7Ra)	BioLegend	Cat#: 135037; RRID: AB_2565269
anti-mouse Siglec-F	BD Bioscience	Cat#: 562757; RRID: AB_2687994
anti-mouse Siglec-F	BioLegend	Cat#: 155506; RRID: AB_2750235
anti-mouse FceRIa	Biolegend	Cat#: 134304; RRID: AB_1626106
anti-mouse I-A/I-E	BioLegend	Cat#: 107622; RRID: AB_493727
anti-mouse/human KLRG1 (MAFA)	BioLegend	Cat#: 138418; RRID: AB_2563015
anti-mouse/human KLRG1 (MAFA)	BioLegend	Cat#: 138419; RRID: AB_2563357
anti-mouse Ly6A/E (Sca-1)	BD Biosciences	Cat#: 749199; RRID: AB_2873577
anti-mouse Ly6A/E (Sca-1)	Biolegend	Cat#: 108112; RRID: AB_313349
anti-mouse Ly-6G	Invitrogen	Cat#: 62-9668-82; RRID: AB_2762763
anti-mouse Ly-6G	BioLegend	Cat#: 127643; RRID: AB_2565971
anti-mouse Ly-6G/Ly-6C (Gr-1)	BioLegend	Cat#: 108404; RRID: AB_313369
Bacterial and virus strains
AAV2-retro-Cre	VectorBuilder	Vector ID:VB220605-1075ztp
AAV2-retro-Cre/JAK1	VectorBuilder	Vector ID:VB220526-1322bxa
Chemicals, peptides, and recombinant proteins
Alternaria alternata	GREER	Cat#: NC1620293
β-CGRP (human)	Cayman Chemical	Cat#: 24725
β CGRP 8-37 human	MiliporeSigma	Cat#: SCP0060
Bovine Serum Albumin	MiliporeSigma	Cat#: A9418
Bovine Serum Albumin	MiliporeSigma	Cat#: A9576
CGRP (rat)	R&D	Cat#: 1161
Collagenase Type 5	Worthington Biochemicals	Cat#: LS005282
DNase I	Roche	Cat#: 10104159001
DNase I	Worthington Biochemicals	Cat#: LS002139
Fetal Bovine Serum	MiliporeSigma	Cat#: F4135
Hialuronidase type 4	MiliporeSigma	Cat#: H3384-500MG
Histodenz	MiliporeSigma	Cat#: D2158
Liberase TM	Roche	Cat#: 5401127001
Papain from papaya latex	MiliporeSigma	Cat#: P4762-25MG
Percoll™	GE healthcare	Cat#: 17-0891-2
Quadrol	MiliporeSigma	Cat#: 122262
Resiniferatoxin	Alomone labs	Cat#: R-400
recombinant mouse IL-2	R&D systems	Cat# 402-ML-020/CF
recombinant mouse IL-33	R&D systems	Cat#: 3626-ML-010/CF
Substance P	Tocris	Cat#: 1156
Upadacitinib (ABT-494)	MedChemExpress	Cat#: HY-19569
Critical commercial assays
Brilliant Violet 421 Streptavidin	Biolegend	Cat#: 405225
CountBrightTM Absolute Counting Beads, for flowcytometry	Invitrogen	Cat#: C36950
Fixation/Permeablization Kit	BD Biosciences	Cat#: 554714
Flexable Viability Dye eFluor™780	Invitrogen	Cat#: 65-0865-18
IL-5 Mouse Uncoated ELISA Kit	Invitrogen	Cat#: 88-7054-88
IL-13 Mouse Uncoated ELISA Kit	Invitrogen	Cat#: 88-7137-22
Mouse CGRP2 ELISA Kit	MyBioSource	Cat#: MBS7241942
Power SYBR™ Green PCR Master Mix	Thermo Fisher Scientific	Cat#: 4367660
Streptavidin Microbeads	Miltenyi Biotec	Cat#: 130-048-101
UltraComp eBeads™ Compensation Beads	Invitrogen	Cat#: 01-2222-42
Viability Dye	Invitrogen	Cat#: 65-0865-14
Zombie NIR™ Fixable Viability Kit	BioLegend	Cat#: 423106
Deposited data
Raw and analyzed bulk RNA-seq data	ISMMS BiNGS Core	GEO: GSE220497
Experimental models: Organisms/strains
C57BL/6J	Jackson Laboratory	Stock No. 000664
C57BL/6J	Sankyo Labo Service Corpration, Inc.	N/A
CalcaCre: Rosa26STOPflox-tdTomato	Hongzhen Hu	N/A
CalcbCre/ERT: Rosa26STOPflox-tdTomato	Hongzhen Hu	N/A
CalcbCre/ERT: Rosa26STOPflox-tdTomato	Brian Kim	N/A
Il4ra flox	Frank Brombacher	N/A
Jak1 flox	Nanjing Biomedical Research Institute	N/A
Jak1 STOPflox-JAK1 GOF	Cyagen Biosciences Inc.	N/A
JAK1 GOF	Cyagen Biosciences Inc.	N/A
Scn10a Cre	Rohini Kuner	N/A
Scn10aCre: Il4raflox	Brian Kim	N/A
Scn10aCre: Jak1flox	Brian Kim	N/A
Scn10aCre: Jak1STOPflox-JAK1 GOF	Brian Kim	N/A
Scn10aCre: Rosa26STOPflox-tdTomato	Brian Kim	N/A
Trpv1Cre: Rosa26STOPflox-tdTomato	Hongzhen Hu	N/A
Rag1 −/−	Jackson Laboratory	Stock No. 002216
Rosa26 STOPflox-tdTomato	Jackson Laboratory	Stock No. 007909
Oligonucleotides
Gapdh qRT-PCR primer 1	integrated DNA Technologies	TCAAGAAGGTGGTGAAGCAGG
Gapdh qRT-PCR primer 2	integrated DNA Technologies	TATTATGGGGGTCTGGGATGG
Calca qRT-PCR primer 1	integrated DNA Technologies	GAGGGCTCTAGCTTGGACAG
Calca qRT-PCR primer 2	integrated DNA Technologies	AAGGTGTGAAACTTGTTGAGGT
Calcb qRT-PCR primer 1	integrated DNA Technologies	CTCTCAGCACGATATGGGTCC
Calcb qRT-PCR primer 2	integrated DNA Technologies	GCAAGAGATGTTTTTCCTGGTCG
Nmu qRT-PCR primer 1	integrated DNA Technologies	GAGGGAGCTTTGCCGTATAGT
Nmu qRT-PCR primer 2	integrated DNA Technologies	GATGCACAACAGAGGACACAA
Tac1 qRT-PCR primer 1	integrated DNA Technologies	TTTCTCGTTTCCACTCAACTGTT
Tac1 qRT-PCR primer 2	integrated DNA Technologies	GTCTTCGGGCGATTCTCTGC
Trpv1 qRT-PCR primer 1	Thermo Fisher Scientific	AGCTGCAGCGAGCCATCACCA
Trpv1 qRT-PCR primer 2	Thermo Fisher Scientific	ATCCTTGCCGTCCGGCGTGA
Vip qRT-PCR primer 1	integrated DNA Technologies	AGTGTGCTGTTCTCTCAGTCG
Vip qRT-PCR primer 2	integrated DNA Technologies	GCCATTTTCTGCTAAGGGATTCT
Software and algorithms
clusterProfiler R package (v4.2.2)	G Yu et. al	https://guangchuangyu.github.io/software/clusterProfiler/
DEGReport R package (v1.30.3)	L Pantano	https://github.com/lpantano/DEGreport
DESeq2 (v1.34.0) R package	M Love et. al	https://github.com/mikelove/DESeq2
FastQC (v0.11.8)	Babraham Bioinformatics	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
FlowJo 10	Tree Star	https://www.flowjo.com/
Galore! (v0.6.6)	Babraham Bioinformatics	https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/
Glimma (v2.4.0) R packages	S Su et. al	https://github.com/Shians/Glimma
NIS-Elements	Nikon Instruments	https://www.microscope.healthcare.nikon.com/products/software
plotly R package (v4.9.4.1)	C Sievert et. al	https://plotly.com/r/
Prism 9	GraphPad Software	https://www.graphpad.com/scientificsoftware/prism/
R (v.4.1.0)	R Core Team	https://www.r-project.org/
STAR aligner (v2.7.5b)	A Dobin et. al	https://github.com/alexdobin/STAR
Salmon (v1.2.1)	R Patro et. al	https://combine-lab.github.io/salmon/
Ugent Center for Plant Systems Biology online tool	UCPSB: Bioinformatics & Evolutionary Genomics	https://bioinformatics.psb.ugent.be/webtools/Venn/
Other
AutoMACS	Miltenyi Biotec	N/A
Bead Homogenizer	BioSpec	Cat#: 112011
Cytek® Aurora	CYTEK	N/A
FACSAria™ III Cell Sorter	BD	N/A
FVMPE-RS	Olympus	N/A
gentleMACS™ Octo Dissociator with Heaters	Miltenyi Biotec	Cat#: 130-096-427
Nikon 80i; CoolSnap ES camera	Nikon Instruments	N/A
QuantStudio 6 Flex Real-Time PCR Systems	Thermo Fisher Scitntific	N/A
Remote Infuse/Withdraw Pump 11 Elite Nanomite Programmable Syringe Pump	Harvard Apparatus	70-4507
StepOnePlus qRT-PCR System	Thermo Fisher Scientific	N/A

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human study participants

Human study participants study was approved by the institutional review board of the University of British Columbia under protocol number H15–00641. Patients with JAK1 GOF variants have been previously described in Del Bel et al.¹¹ Diagnosis of AD and asthma followed guidelines of the European and North American consensus report⁶⁶ and the Global Initiative for Asthma (https://ginasthma.org), respectively. Eczema Area and Severity Index (EASI) was used to score AD severity based on methodology described in Leshem et al.⁶⁷ Written informed consent was obtained from parents as study participants were minors at the time of consent.

Animal studies

C57BL/6 wild-type (WT), Rag1^−/−, and Rosa26^{STOPflox-tdTomato} mice were purchased from Jackson Laboratory. Human JAK1 GOF (JAK1^GOF) mice and Jak1^{STOPflox-JAK1 GOF} mice were generated through Cyagen Biosciences, Inc. Schematic of the targeting strategy used to introduce the germline human JAK1 GOF mutation into the murine Jak1 locus is outlined in Figure 1H. Scn10a^Cre mice were provided by Dr. Rohini Kuner (Heidelberg University, Heidelberg, Germany). Jak1^flox mice were purchased from Nanjing Biomedical Research Institute of Nanjing University. Il4ra^flox mice were generated by Dr. Frank Brombacher (International Center for Genetic Engineering and Biotechnology, Cape Town, South Africa). Trpv1^Cre:Rosa26^{STOPflox-tdTomato}, Calca^Cre:Rosa26^{STOPflox-tdTomato}, and Calcb^Cre/ERT:Rosa26^{STOPflox-tdTomato} mice were provided by Dr. Hongzhen Hu (Icahn School of Medicine at Mount Sinai, New York, USA). Scn10a^Cre: Jak1^flox (Jak1^Δneuron) mice were generated by crossing Scn10a^Cre mice with Jak1^flox mice. Scn10a^Cre:Il4ra^flox (Il4ra^Δneuron) mice were generated by crossing Scn10a^Cre mice with Il4ra^flox mice. Scn10a^Cre:Rosa26^{STOPflox-tdTomato} mice were generated by crossing Scn10a^Cre mice with Rosa26^{STOPflox-tdTomato} mice. Scn10a^Cre:Jak1^{STOPflox-JAK1 GOF} (JAK1^GOF:neuron) mice were generated by crossing Scn10a^Cre mice with Jak1^{STOPflox-JAK1 GOF} mice. All mice were housed in specific-pathogenfree (SPF) conditions and environmentally controlled animal facilities with a 12-hour light-dark cycle and were given unrestricted access to food and water. All animal protocols and experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of ISMMS, Washington University School of Medicine in St. Louis, and the National Research Institute for Child Health and Development. All mice (8–13 weeks old, male and/or female) were randomly assigned to experimental groups in an age- and gender-matched fashion, and littermates were used whenever possible. All experiments were performed following strict IACUC guidelines.

METHOD DETAILS

Intranasal administration of Alternaria alternata extract, papain, neuropeptides, and JAK inhibitor and lung inflammation assessment

Allergic lung inflammation was induced by intranasal administration of Alternaria alternata extract (Greer Laboratories, Inc.; 2 μg/dose in 40 μl of sterile phosphate-buffered saline [PBS]) for four consecutive days (Day 0, 1, 2, and 3) or papain (MilliporeSigma, 20 μg/dose in 40 μl of sterile PBS) for three consecutive days (Day 0, 1, and 2). The effect of exogenous neuropeptides was examined following twice daily (Day −1, 0, 1, 2, and 3) intranasal administration of neuropeptide CGRPβ (Cayman Chemical; 1 μg/dose in 40 μl of sterile PBS) or substance P (Tocris, 50 μg/dose in 40 μl of sterile PBS) prior to Alternaria alternata administration. Likewise, the effect of JAK1 inhibition was examined following twice daily (Day 0, 1, 2, and 3) intranasal administration of the JAK1 inhibitor, upadacitinib (MedChemExpress; 2.5 mg/kg in 40 μl of sterile PBS), prior to Alternaria alternata administration. Mice were sacrificed on Day 4 (or on Day 3 for papain) for clinical, histologic, and immunologic analysis. Lungs, collected following PBS perfusion, were processed for immune cell analysis by flow cytometry (right lungs) or fixed in 4% paraformaldehyde for histological analysis (left lungs). Histology slides and digital images were created by HistoWiz Inc. (histowiz.com).

Upadacitinib administration

In the Alternaria alternata-induced allergic lung model, upadacitinib was administered intranasally (2.5 mg/kg in 40 μl of sterile PBS) or intraperitoneally (5 mg/kg in 200 μl of sterile PBS) twice daily for four consecutive days (Day 0, 1, 2, and 3).

Tamoxifen administration

Tamoxifen (75 mg/kg in 100 μl corn oil) was administered intraperitoneally for 5 consecutive days, and experiments were performed 7 days post last administration.

Construction of Adeno-Associated Virus (AAV) vectors

AAV2-retro-Cre and AAV2-retro-Cre/hJAK1[NM_002227.4]*(A634D) were constructed by VectorBuilder Inc. (https://en.vectorbuilder.com/). AAV vector transduction protocols followed methods previously described by Su et al. ²³ with minor modifications. Briefly, AAV (2 × 10³ genome copies/ml) was diluted 10-fold in PBS and was administered intranasally (40 ul/mouse) for four consecutive days, and experiments were performed three weeks after the last administration,.

Flow cytometry

Preparation of lung tissue for flow cytometry followed a slightly modified protocol of Moro et al.,⁶⁸. Briefly, three lobes of the right lung were placed into a solution containing 5 ml Hank’s Balanced Salt Solution (HBSS), Liberase TM (Roche; 50 μg/ml final concentration), and DNase I (Roche; 1 μg/ml final concentration) in a gentleMACS^™ C-tube (Miltenyi Biotec). Lung tissues were homogenized using protocol “37_m_LDK1” on the gentleMACS^™ Dissociator with Heaters (Miltenyi Biotec), then passed through a 70-μm cell strainer (Fisher Scientific) into a 50-ml tube. Any remaining tissue was manually homogenized on the strainer with the end of a 3-ml syringe plunger. The gentleMACS^™ C tube and the strainer were washed with 2% (vol/vol) FBS/PBS. The tissue homogenate was centrifuged at 400g for 5 minutes at 4°C, and after discarding the cleared supernatant, the cell pellets were incubated with 1 ml red blood cell lysing buffer (MilliporeSigma) at room temperature for 1 minute followed by a 2% (vol/vol) FBS/PBS wash. Simultaneously, cervical skin lymph nodes from the same mice were processed by homogenizing manually with the end of a 3-ml syringe plunger on a 100-μm cell strainer (Fisher Scientific) directly into a 50-ml tube. The strainer was washed with 2% (vol/vol) FBS/PBS and centrifuged at 400g for 5 minutes at 4°C.

Both the lung and lymph node cell samples were stained with the Zombie NIR viability dye (Biolegend; 1:500) to exclude dead cells, followed by Fc-receptor blocking and cell-surface staining with specific antibodies (key resources table). ILC2s were defined as live CD45⁺ CD90⁺ CD127⁺ Sca-1⁺ KLRG-1⁺ cells that were negative for lineage markers CD3, CD4, CD8, CD11b, CD11c, B220 (or CD19), MHC-II, Ly6G, and Siglec F within a gate for lymphocytes. Eosinophils were defined as live CD45⁺ Siglec F⁺ cells within a gate for granulocytes. CD4⁺ T cells were defined as live CD45⁺ CD3⁺ CD4⁺ cells negative for B220 (or CD19), within a gate for lymphocytes. The cells were analyzed using a Cytek^® Aurora (CYTEK) flow cytometer, and the absolute number of cells in each sample was determined using CountBright^™ Absolute Counting Beads (Invitrogen).

Bone marrow (BM) transplantation

Recipient mice were supplied with 200 ml of drinking water containing 5 ml of Sulfatrim antibiotic (Sulfamethoxazole/trimethoprim) added into 200 ml of drinking water for one week starting from the day prior to irradiation (Day −1). At the end of Week 1, recipient mice were irradiated (950 cGy) using the X-RAD 320 (Precision X-Ray). Bone marrow was harvested from the femurs and tibias of donor mice, treated with red blood cell lysis buffer (MilliporeSigma) for two minutes, then passed through a 70-μm cell strainer (Fisher Scientific) into a 15-ml conical tube. The cell strainer and cells were washed with 2% (vol/vol) FBS/PBS. ViaStain^™ AOPI staining (Nexcelom Bioscience) was used to determine the number of live cells on a Cellometer Auto 2000 (Nexcelom Bioscience). Each recipient mouse received 1 × 10⁷ live BM cells by intravenous (i.v.) infusion within 24 hours following irradiation. Mice were used in experiments four weeks following BM transplantation.

Targeted ablation of TRPV1-expressing peripheral neurons

Chemical ablation of TRPV1-expressing peripheral neurons was performed using Resiniferatoxin (RTX) according to previously published protocols.^14,31 Systemic ablation of TRPV1-expressing neurons was achieved by daily subcutaneously (s.c.) injection of increasing doses of RTX (Alomone Labs; 30, 70, 100 μg/kg for three consecutive days) or vehicle (100 μl saline) in the flanks of C57BL/6 WT mice. Targeted ablation of TRPV1-expressing neurons in the DRG was achieved with daily intrathecal (i.t.; in the L5-L6 region) injection of RTX (25 ng/mouse for two consecutive days) or vehicle (10 μl saline). RTX-treated mice were allowed to rest for 4–6 weeks before use. For targeted ablation of vagal ganglia (VG), TRPV1-expressing neurons in the VG were ablated based on a previously described protocol.⁶⁹ Mice were assessed for heat sensitivity with a hot plate test to confirm the functional loss of TRPV1-expressing DRG neurons.^14,31

C57BL/6 WT mice were injected with RTX intra-ganglionically (i.g.) into the right VG with RTX (25 ng/mouse) or vehicle (0.5 μl saline) using the Remote Infuse/Withdraw Pump 11 Elite Nanomite Programmable Syringe Pump (Harvard Apparatus). RTX-treated mice were allowed to rest for 4–6 weeks before use, and only the right lobes of the lung were used for analysis.

Fluorescent protein imaging in cleared tissue

Tissue samples containing fluorescent tdTomato protein were cleared as previously described.⁷⁰ Thoracic segments of mouse DRG and VG were harvested from experimental mice and fixed in 4% paraformaldehyde. Fixed tissues were washed with PBS, placed directly on slides, and covered with the refractive index matching solution, RIMS (40 g Histodenz [MilliporeSigma], 30 ml Phosphate buffer (0.02 M), 0.1% Tween-20 [MilliporeSigma], 0.01% sodium azide). All images were captured using an upright microscope equipped for epifluorescence microscopy (Nikon 80i; CoolSnapES camera [Nikon Instruments] or FVMPE-RS [Olympus]).

GWAS (genome-wide association studies)-based analysis

GWAS were performed using data obtained from the publicly available GWAS Catalog (https://www.ebi.ac.uk/gwas/genes/JAK1) database, which includes the common JAK1 variants and their associated phenotypes.

Predicted STAT6 transcriptional targets

Data, obtained from the MotifMap Predicted Transcription Factor Targets dataset (https://maayanlab.cloud/Harmonizome/gene_set/STAT6/MotifMap+Predicted+Transcription+Factor+Targets), were used for analysis. The linked dataset of STAT6 transcriptional targets was used to search genes of interest (CALCA, CALCB, TAC1, NMU, and VIP are putative targets).

Isolation, culture, and analysis of lung ILC2s

Lung dissociation was performed as previously described.⁶⁸ Briefly, pulmonary PBS perfusion through the heart was performed immediately after euthanasia in order to remove circulating cells from the lung vasculature prior to harvesting the lungs. Lung pairs were minced into small pieces using scissors and were placed directly into a gentleMACS^™ C-tube (Miltenyi Biotec) containing 5 mL of 1% BSA in HBSS. The lungs were dissociated using the gentleMACS^™ Dissociator programmed to protocol “m_lung-01–02.” Following dissociation, the lungs were incubated with collagenase type 5 (Worthington; 200 U/ml final concentration), DNase I (Worthington; 100 U/ml final concentration), and hyaluronidase type 4 (MilliporeSigma; 15000 U/ml final concentration) in a 37 ºC water bath with 40 shakes/minute for 30 minutes. The lungs were then dissociated for a second time using the gentleMACS^™ Dissociator with the protocol “m_lung-01–02”. Next, the lung dissociation mixture was filtered through a 70-μm cell strainer (Fisher Scientific) into a 50-ml conical tube, and any pieces remaining on the strainer were manually homogenized with the end of a 3-ml syringe plunger. The conical tubes containing the mixture were centrifuged at 490g for 5 minutes at 4 °C. The supernatant was discarded, and the cell pellet was resuspended in 21 ml of 2% (vol/vol) FBS/PBS and mixed with 9 ml of 100% Percol^™ (GE Healthcare). Then, 5 ml of 70% Percol^™ was gently added to the bottom of the tube, and the tubes were centrifuged at 490g for 30 minutes at room temperature. After separation, the intermediate layer was carefully moved into a new tube and washed twice with 2% (vol/vol) FBS/PBS. During every wash cycle, the cells suspended in 2% (vol/vol) FBS/PBS were filtered through a 70-μm cell strainer after resuspension. The samples were then incubated in 800 μL Fc block containing 2% (vol/vol) FBS/PBS for 15 minutes. Following incubations, the samples were washed and then incubated with a biotin-conjugated antibody cocktail including lineage markers for 30 minutes. The samples were then washed and incubated with MicroBeads (Miltenyi Biotec) for 20 minutes. Next, the samples were depleted of lineage-positive cells using an AutoMACS (Miltenyi Biotec), and the negative fraction was collected for fluorescence-activated cell sorting (FACS) using a FACSAria^™ III Cell Sorter (BD Biosciences). Sorted lung-derived ILC2s were defined as live, lineage (Lin)^-, CD45⁺, CD127⁺, Sca-1⁺, and KLRG1⁺.

Sort-purified lung ILC2s were placed into a 96-well round-bottom plate (300 cells/well, 200 μl/well) and cultured with recombinant mouse IL-2 (R&D, 10 ng/ml final concentration) and/or recombinant mouse IL-33 (R&D; 50 ng/ml final concentration), Substance P (R&D; 10 ng/ml final concentration), CGRPα (R&D; 10 ng/ml final concentration), and CGRPβ (Cayman Chemicals, 10 ng/ml final concentration) for 72 hours. The final supernatant was collected, and levels of IL-5 and IL-13 were measured by ELISA kits (Invitrogen).

RNA isolation from mouse DRG and JG/NG (VG) and qRT-PCR

Mouse DRG or nodose ganglia (NG)/jugular ganglia (JG) (combined VG) were harvested and homogenized with a bead homogenizer (BioSpec) in lysis buffer RA1 (Macherey-Nagel). Total RNA was extracted with the NucleoSpin RNA isolation kit following manufacturer’s protocol. Equal amounts of cDNA were synthesized from total RNA extracts using the iScript cDNA Synthesis kit. Gene expression levels were determined using gene-specific primers (key resources table) and 2X qPCR Universal Green MasterMix (Lamda Biotech) or Power SYBR^™ Green PCR Master Mix (Thermo Fisher Scientific). Reactions were cycled using a StepOnePlus^™ Real-Time PCR System (Thermo Fisher Scientific) or QuantStudio 6 Flex Real-Time PCR System (Thermo Fisher Scientific) using the manufacturer’s protocol. Gene expression was normalized to that measured for Gapdh, and relative expression was calculated using the ΔΔCt method.

RNA-seq alignment and quality control

Library preparation and sequencing steps of RNA-seq were performed by GENEWIZ from Azenta Life Sciences (genewiz.com). All samples from all experiments were processed using the same pipeline for compatibility. Quality control was performed using FastQC (v0.11.8). Trim Galore! (v0.6.6) was used to trim the adapter sequences with a quality threshold of 20. The GRCm38 genome assembly from GENCODE release M25 was used as a genome and transcriptome reference. The alignment was performed using STAR aligner (v2.7.5b) ⁷¹. Gene-level read counts were obtained using Salmon (v1.2.1) for all libraries.⁷² All samples passed the alignment quality control requirements, with >90% of reads uniquely mapping (>10M uniquely mapped reads for each library) using STAR aligner.

Differential mRNA gene expression and pathway analysis

Differential expression analysis was performed using gene-level read counts and the DESeq2 (v1.34.0) R package.⁷³ Genes with <5 reads in total across all samples were considered inactive genes. A gene was considered differentially expressed if the adjusted p-value (Benjamini-Hochberg procedure) was <0.05 and the absolute log₂(fold change) was >1. Overrepresentation analysis of differentially expressed genes was performed using the clusterProfiler R package (v4.2.2).^74,75 The gene sets used for functional analysis were obtained from the Molecular Signatures Database (MSigDB).^76–78 The overlaps between the differentially expressed genes and the MSigDB gene sets were tested for statistical significance using Fisher’s exact test followed by the multiple test correction using the Benjamini-Hochberg procedure (adjusted p-value <0.05).

Single-cell RNA-seq analysis of the JG/NG (VG)

Normalized scRNA-seq expression data for JG and NG neurons was obtained from a previously published study on vagal sensory neurons.²⁸ Cells from these datasets that were annotated as non-neuronal cells were filtered out from downstream analysis. The threshold for determining if a cell is positive for a gene was calculated using the method described by Usoskin et. al.⁷⁹ and was performed by calculating the average value of the top three cells with the highest expression for that gene and multiplying by 0.05. The numbers of positive cells for the genes Trpv1, Scn10a, Jak1, Calca, Calcb, Vip, Nmu, and Tac1 were calculated using R (v.4.1.0). The numbers of cells positive for the genes Scn10a, Jak1, Calca, Calcb, Vip, Nmu, and Tac1 were calculated from Trpv1⁺ cells. The number of cells positive for Jak1, Calca, Calcb, Vip, Nmu, and Tac1 were also calculated from Scn10a⁺ cells. For both analyses, a global threshold (using all neuronal cells) was calculated.

QUANTIFICATION AND STATISTICAL ANALYSIS

The data are shown as the mean ± standard deviation (SD), unless otherwise indicated. Data from independent experiments are representative of at least two independent replicates or as pooled data. None of the data were excluded from statistical analyses, unless due to technical errors. Statistical significance was determined using unpaired Student’s t-tests with Welch’s correction, unless otherwise noted. Statistical evaluations were performed using GraphPad Prism 8.0 software (GraphPad Software). Significance is regarded as: ^†p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, and n.s. as not significant.

Supplementary Material

1. Figure S1. Standard gating strategies used for flow cytometry, related to Figures 1–6.

Representative gating strategy is shown using the lung from a steady-state WT mouse. ILC2 – group 2 innate lymphoid cell, WT – wild-type

2. Figure S2. Chemical denervation of TRPV1⁺ sensory neurons via subcutaneous and intraganglionic, but not intrathecal, RTX injection alters the transcriptional signature and allergic lung inflammation, related to Figure 3.

(A) Volcano plot analysis of RNA-seq data obtained from lung tissues of mice treated with subcutaneous (s.c.) RTX vs. vehicle control challenged with intranasal ALT, n = 2–3 per group. The analysis identified 148 down-regulated and 177 up-regulated genes.

(B) KEGG pathway overrepresentation analysis of RNA-seq data obtained from lung tissues of mice treated with s.c. RTX vs. vehicle control challenged with intranasal ALT. Overrepresentation analysis used Fisher’s exact test to determine the significance of enrichment. All terms shown were identified as the top four significantly activated pathways in terms of adjusted p-value (p-adjusted < 0.05, Benjamini-Hochberg procedure), n = 2–3 mice per group.

(C) Schematic of the allergic lung inflammation model generated by selective denervation of vagal afferents.

(D-E) Expression levels of Trpv1 in the VG and DRG from vehicle control- and intraganglionic (i.g.) RTX-treated mice. Data were obtained from two independent experiments, (D) n = 7–10 ganglia per group, (E) n = 4–5 mice per group.

(F) Schematic of the allergic lung inflammation model generated by selective denervation of spinal visceral afferents.

(G) Withdrawal latency from noxious heat (hot plate test). Mice treated with intrathecal (i.t.) vehicle control or RTX were placed on a hot plate set at 50 degrees Celsius, and withdrawal latency was measured as the time until the appearance of paw withdrawal behavior.

(H-I) Representative lung histology images of mice treated with i.t. vehicle control or RTX challenged with intranasal ALT, obtained by (H) H&E stain and (I) PAS stain.

(J-K) Flow cytometry of lung tissue from mice treated with i.t. vehicle control or RTX challenged with intranasal ALT. Shown are the frequencies of (J) ILC2s and (K) eosinophils. Data were obtained from three independent experiments, n = 5–6 mice per group.

(L) Volcano plot analysis of RNA-seq data obtained from lung tissues of mice treated with i.t. RTX vs. vehicle control challenged with intranasal ALT, n = 2–3 per group. The analysis identified 20 up-regulated genes and 20 down-regulated genes.

(M) KEGG pathway overrepresentation analysis of RNA-seq data obtained from the lung tissues of mice treated with i.t. RTX vs. vehicle control challenged with intranasal ALT. Overrepresentation analysis used Fisher’s exact test to determine the significance of enrichment, n = 2–3 mice per group.

***p < 0.001 and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(C, F) Figure adapted from an image created with BioRender.com.

ALT – Alternaria alternata, DEG – differential gene expression, DRG – dorsal root ganglia, H&E – hematoxylin and eosin, i.g. – intraganglionic, ILC2s – group 2 innate lymphoid cells, i.t. – intrathecal, PAS – periodic acid-Schiff, RTX – resiniferatoxin, s.c. – subcutaneous, VG – vagal ganglia,

3. Figure S3. Disruption of sensory neuron-intrinsic Jak1, but not Il4ra, exacerbates allergic lung inflammation, related to Figure 4.

A) Schematic of the allergic lung inflammation model using littermate control (Il4ra^flox) and experimental Il4ra^Δneuron mice.

(B-C) Representative lung histology images of control and Il4ra^Δneuron mice challenged with intranasal ALT, obtained by (B) H&E stain and (C) PAS stain.

(D-E) Flow cytometry of lung tissue from control and Il4ra^Δneuron mice challenged with intranasal ALT. Shown are the frequencies of (D) CD45⁺ immune cells, (E) ILC2s, and (F) eosinophils. Data were obtained from two independent experiments, n = 4 mice per group.

(G) Schematic of the allergic lung inflammation model using littermate control (Jak1^flox) and experimental Jak1^Δneuron mice.

(H-I) Representative lung histology images of control and Jak1^Δneuron mice challenged with intranasal papain obtained by (H) H&E stain and (I) PAS stain.

(J-L) Flow cytometry of lung tissue from control and Jak1^Δneuron mice challenged with intranasal papain. Shown are the frequencies of (J) CD45⁺ immune cells, (K) ILC2s, and (L) eosinophils. Data were obtained from two independent experiments, n = 3–4 mice per group.

*p < 0.05 and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm. (A, G) Figure adapted from an image created with BioRender.com.

4. Figure S4. Inhalation of a JAK1 inhibitor is less effective than systemic administration in attenuating allergic lung inflammation, related to Figure 4.

(A) Schematic of vehicle (Veh) or upadacitinib (Upa) treatment of ALT-induced allergic lung inflammation in WT mice. (B-C) Flow cytometry of lung tissue from vehicle- or upadacitinib-treated mice challenged with intranasal ALT. Shown are the frequencies of (B) ILC2s and (C) eosinophils. Data were obtained from two independent experiments, n = 5 mice per group.

(D-E) Representative images of lung histology following (D) H&E and (E) PAS staining.

***p < 0.001, *p < 0.05, and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm. (A) Figure adapted from an image created with BioRender.com.

5. Figure S5. Jak1/JAK1 modulation changes the expression of neuropeptides in the VG and DRG, related to Figure 4.

(A) Schematic of approach used to measure expression changes by quantitative PCR (qPCR) in control and Jak1^Δneuron mice. (B-E) Expression levels of neuropeptide transcripts in the VG and DRG from control and Jak1^Δneuron mice challenged with intranasal ALT. Shown are (B, C) Vip and (D, E) Nmu transcript levels. Data were obtained from two independent experiments, n = 3–4 mice per group. (F) Schematic of approach used to measure expression changes by qPCR in control WT and JAK1^GOF mice. (G-H) Expression levels of neuropeptide transcripts in the VG and DRG from control WT and JAK1^GOF mice challenged with intranasal ALT. Shown are (G) Calcb and (H) Tac1 transcript levels. Data were obtained from two independent experiments, n = 8–9 mice per group. (I) Schematic of i.g. RTX injection following tamoxifen treatment in Calcb-reporter mice. (J) Imaging of CGRPβ⁺ neurons in Calcb-reporter mice in the VG. Shown are representative images of VG from mice treated with i.g. vehicle control or RTX. (K) Numbers of CGRPβ⁺ neurons in the VG. Data were obtained from two independent experiments, n = 3–5 ganglia per group. *p < 0.05 and not significant (n.s.) (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm.

(A, F, I) Figure adapted from an image created with BioRender.com.

GOF – gain-of-function, JAK1 – Janus kinase 1

6. Figure S6. CGRP, but not SP, suppresses type 2 cytokine production from lung ILC2s and allergic lung inflammation, related to Figure 5.

(A) Schematic of the lung ILC2 suppression assay.

(B-C) Type 2 cytokine measurement from lung ILC2s stimulated with IL-33 and/or CGRPα. Shown are levels of (B) IL-5 and (C) IL-13.

(D) Schematic of induction of allergic lung inflammation and SP treatment in Rag1^−/− mice. (E-F) Flow cytometry of lung tissue from in Rag1^−/− mice challenged with intranasal ALT as well as vehicle control or SP. Shown are the frequencies of (E) ILC2s and (F) eosinophils. Data were obtained from two independent experiments, n = 5 mice per group. (G-H) Representative lung histology images from Rag1^−/− mice challenged with intranasal ALT as well as vehicle control or SP obtained by (G) H&E and (H) PAS staining. (I) Schematic of induction of allergic lung inflammation and CGRPβ treatment in WT mice. (J-K) Flow cytometry of lung tissue from WT mice challenged with intranasal ALT as well as vehicle control or CGRPβ. Shown are the frequencies of (J) ILC2s and (K) eosinophils. Data were obtained from two independent experiments, n = 4–5 mice per group.

(L-M) Representative lung histology images of WT mice challenged with intranasal ALT as well as vehicle control or CGRPβ obtained by (L) H&E stain and (M) PAS stain.

***p < 0.001, *p < 0.05 (Unpaired t-test with Welch’s correction). Data are represented as mean ± SD. Scale bars indicate 200 μm. (A, D, I) Figure adapted from an image created with BioRender.com. SP – substance P

7

Video S1. Lung-innervating spinal afferents in the DRG, related to Figure 6

8

Video S2. Lung-innervating vagal afferents in the VG, related to Figure 6

Highlights.

• A human germline JAK1 GOF variant promotes spontaneous skin inflammation in mice

• Vagal sensory neuron expression of JAK1 GOF promotes immune homeostasis in the lung

• A neuronal JAK1-CGRPβ axis suppresses ILC2 responses and allergic lung inflammation