Oncostatin M receptor deficiency as a novel candidate genetic cause of autosomal recessive hyper-IgE syndrome

Hyper-immunoglobulin E syndrome (HIES), originally termed Job’s syndrome, is a collection of rare inborn errors of immunity (IEI) characterized by the clinical triad of elevated IgE serum levels, eczema, and recurrent skin and pulmonary infections (1, 2, 3). Characteristic disease presentations of Job’s syndrome consist of recurrent “cold” staphylococcal skin abscesses, pulmonary staphylococcal and Aspergillus infections, chronic mucocutaneous candidiasis (CMC), eosinophilia, as well as a somatic/extra-hematopoietic phenotype, including facial, skeletal, and dental abnormalities, such as craniosynostosis, osteopenia, and scoliosis, and as more recently reported cranial artery aneurisms (1, 4, 5). Disease-causing monoallelic dominant-negative (DN) missense variants of the gene encoding signal transducer and activator of transcription 3 (STAT3) were first reported in 2007 (6, 7, 8, 9, 10), and subsequently many more patients with AD-HIES due to DN STAT3 variants have been reported (6, 7, 8, 9, 10, 11).

Beyond the originally described STAT3 autosomal dominant (AD)-HIES, several other genetic etiologies of AD- or autosomal recessive (AR)-HIES with very similar and overlapping, yet not identical, clinical presentations have been described, prompting the notion of a collection of hyper-IgE syndromes (8, 11, 12, 13, 14). For AD-HIES, these include mutations in the receptor subunit glycoprotein 130 (gp130) encoded by IL6ST (15), loss-of-function (LOF) mutations in CARD11 (16), and AD gain-of-function (GOF) mutations in STAT6 (17). Moreover, two molecules involved in the TGFβ pathway, inherited as GOF mutations in TGFβ receptor complex (TGFBR1/2) (Louys-Dietz syndrome) or LOF mutations in ERBIN, also cause AD-HIES (18, 19). In addition, HIES inherited in an AR manner can be caused by biallelic mutations in a number of different genes, including DOCK8 (20), the STAT3-inducing transcription factor ZNF341 (21, 22), the phosphoglucomutase PGM3 (23, 24), through defect glycosylation of the receptor subunit GP130 (25), SPINK5 (26), biallelic LOF variants in IL6ST/GP130 (27) (also a cause of AD-HIES in monoallelic form), and biallelic mutations in IL6R encoding the IL-6 receptor (28).

STAT3 is broadly expressed in many tissues and acts downstream of numerous important cytokine receptors, including IFNAR1, IL-6R, IL-10R, IL-11R, IL-21R, IL-22R, IL-23R, oncostatin M (OSM) receptor (OSMR) and leukemia inhibitory factor receptor (LIFR) (29), thereby accounting for the heterogenous presentation and complexity of diseases caused by abnormal STAT3 activity. Given the central position and key role of this transcription factor, defects in STAT3 or STAT3-inducing cytokine receptors result in dysregulated signaling pathways, which in turn inhibit the generation of T helper 17 (Th17) cells and neutrophil chemotaxis, resulting in abnormal immune responses and insufficient protection against bacterial and fungal infections (29). Moreover, defective STAT3 signaling leads to a shift toward Th2-dominated responses (30), promoting elevated IgE production and eosinophilia, which account for the elevated IgE levels and atopic features, such as eczema and asthma.

The OSMR gene (5p13.1) encodes the OSMR and is a membrane-embedded receptor belonging to the family of type I cytokine receptors (31). OSMR is a constituent of two surface receptor complexes, as it heterodimerizes with gp130 to form a receptor for OSM and with interleukin 31RA (IL31RA) to form a receptor for IL-31 (32). OSM-induced OSMR-mediated signaling mediates pleiotropic effects on inflammation, tissue remodeling, hematopoiesis, cell differentiation, and cancer (33). IL-31–induced OSMR-mediated signaling shares similarities with OSM-OSMR signaling but is more specific to cutaneous inflammation, T cell differentiation, and allergic diseases (34). Structurally, OSMR highly consists of three major domains. The extracellular domain is crucial for the interaction of OSMR with functionally related membrane proteins, such as gp130 or IL31RA, and this domain comprises four tandemly arranged, highly conserved, subdomains termed type III fibronectin (FNIII) domains (31). Following the extracellular domain is the short transmembrane domain, which culminates in the C-terminal cytoplasmic domain, the latter being responsible for OSMR downstream signaling (35). Upon cytokine activation, Janus-associated kinases phosphorylate tyrosine residues in the cytoplasmic domain of these receptors leading to the recruitment and activation of STAT1, STAT3, and, to a lesser extent, STAT5 (36, 37, 38, 39).

OSM belongs to the IL-6 family cytokines, including IL-6, IL-11, IL-27, LIF, OSM, IL-35, cardiotrophin-1, and ciliary neurotrophic factor (40). In addition to OSMR/gp130, OSM also binds to a receptor complex constituted by gp130 and the LIFR, although with lower affinity (41). Both gp130 and LIFR contain cytoplasmic domains capable of signal transduction, but LIFR tissue expression is much more restricted than OSMR tissue expression, as the LIFR is only expressed at low levels in adult epithelial, mesenchymal, and hematopoietic cells. The LIFR therefore likely fulfills its primary physiological role during embryogenesis (42). IL-31 signals exclusively through the OSMR/IL-31RA receptor complex. The effects of IL-31 are diverse but primarily documented in vitro and involve leukocytes, keratinocytes, airway, and intestinal epithelial cells as well as neurons (34). However, some evidence suggests that IL-31 plays important roles in airway, skin, and intestinal inflammation in humans (43, 44).

According to the Online Mendelian Inheritance in Man database, heterozygosity for OSMR variants is associated with familial primary localized cutaneous amyloidosis (FPLCA), an AD disease characterized by itchy, thickened, red skin with amyloid deposits in the superficial dermis, the skin being the sole organ affected in this disorder (45, 46, 47). No other clearly defined disease entities have hitherto been associated with OSMR variants. Recently, however, a homozygous variant in OSM, observed in a consanguineous family, has been associated with inherited severe bone marrow failure syndrome evidenced by profound anemia, thrombocytopenia, and neutropenia but without susceptibility to infection. Functionally, this OSM variant affects the C-terminal part of OSM, which is important for interaction with the OSMR (48). Finally, defects in LIFR-gp130 cause Stüve-Wiedemann syndrome, defined by profound bone abnormalities, respiratory distress, and autonomous nervous system anomalies with early lethality (13).

Collectively, the genetic, immunological, and clinical infectious and extra-hematopoietic spectrum of HIES, some of which have been reclassified into combined immunodeficiencies, display a heterogenous picture (12, 49). Here we sought to investigate a new genetic etiology of AR-HIES in a patient with OSMR deficiency to obtain further insights into the pathogenesis of this disorder and to improve diagnosis of patients with clinical HIES in whom none of the currently established gene defects can be identified.

We investigated a 57-year-old man of Caucasian (Danish) descent with a clinical history of severe atopic dermatitis, eczema, and asthma throughout life. He experienced slow wound healing and frequent secondary staphylococcal skin infections and pustules, but only seldom skin abscesses. Dental status was poor with tooth retraction at a young age and a previous history of tooth abscesses (Fig. 1 A). From the age of 20 years, he experienced one or more bacterial pneumonias per year, as well as episodes of oral candidiasis but not CMC nor invasive candida infection. At the age of 45 years, he had a distal radial fracture in response to a low energy trauma, a subsequent bone density scan showing osteopenia, originally presumed to be related to prolonged steroid therapy for his asthma. Due to prolonged coughing, dyspnea, and chest pain, he was referred to evaluation, and computed tomography of the chest showed signs of chronic pulmonary aspergillosis with a thick-walled cavity in the left upper lobe characteristic of an aspergilloma (Fig. 1 B). Bronchoscopy revealed Aspergillus fumigatus DNA and a positive Aspergillus galactomannan antigen (AGA) test. Blood tests at that time showed increased Aspergillus IgG (IgG; 157 mg/L, normal range <26.9 mg/L) and IgE (37,900 IU/L, normal range <350 IU/L), slightly increased eosinophil count (0.66 × 10⁹/L, normal range <0.5 × 10⁹/L), and a very high IgE (IgE: 34,000 × 10³ IU/L, normal range <100 IU/L) (Table S1). A diagnosis of chronic cavitary pulmonary aspergillosis (CCPA) was established, and the patient was treated with long-term antifungal itraconazole and also started on prophylactic sulfamethoxazole and trimethoprim (Bactrim) prophylaxis. A skin biopsy showed perivascular inflammation with lymphocytes and eosinophils but no amyloid deposits.

Based on the highly elevated IgE over time (Fig. S1 A) and increased occurrence of infections, the patient was referred to the Department of Infectious Diseases for evaluation of suspected HIES. Standard immunological screening revealed normal serum concentrations of IgA, IgM, and IgG (Table S1). Serum IgE was markedly elevated to levels ranging from 18,200 to 34,000 IU/ml and with mild eosinophilia and slight anemia in the range between 6.8 and 7.8 mmol/L (normal range 8.3–10.5 mmol/L) (Fig. S1 and Table S1). Peripheral CD19⁺, CD4⁺, and CD8⁺ lymphocyte concentrations were normal (Table S1), but patient CD4⁺ T cells displayed an activated (HLA-DR⁺: 30%) phenotype (Table S1). Lymphocyte subset analysis showed normal distributions of Th2, Th9, and Th17 subsets, except for slightly elevated peripheral follicular helper T (pTFH) cell (CD4⁺ CXCR5⁺ PD1⁺ ICOS⁺) frequencies in the patient (Table S1 and Fig. S2). Serum levels of IL-4 and IFNγ were likewise within normal ranges. Patient CD8⁺ T cells showed a reduced proliferation index and a diminished responding fraction (Table S1). Stimulated patient CD4⁺ T cells displayed no relative expansion of Th17 cells (CD4⁺ IL-17A⁺ T cells) nor Th2 cells (CD4⁺ IL-4Rα⁺ T cells) compared to those of controls (Fig. S3).

Since HIES was suspected based on the clinical presentation and paraclinical findings as well as a HIES score of >20 (Fig. S1 B), whole-genome sequencing (WGS) was performed, revealing that P1 was homozygous for a missense variant (c.1307T > A) in the gene (OSMR) coding for the OSMR, in which valine was substituted for aspartic acid at amino acid position 436 (Val436Asp, V436D) (Fig. 1 C). The OSMR variant has a combined annotation-dependent depletion (CADD) score of 24.3 (with a mutation significance cutoff [MSC] = 19.3) and a minor allele frequency (MAF) of 3.39 × 10⁻³. Immunoblotting of whole-cell lysates from P1 fibroblasts revealed that endogenous OSMR is expressed at a slightly reduced level and migrating further in the gel, reflecting a reduced molecular weight of the OSMR protein (Fig. 1, D and E). A significant reduction in OSMR surface expression on patient dermal fibroblasts compared to controls was found by flow cytometry (FC) (Fig. 1, F and G). The genome was also searched for known or predicted pathogenic variants in classified IEI genes, particularly those previously associated with AD- or AR-HIES, none of which were identified (50). Of note, the patient was found to be heterozygous for missense variants in ZNF341 and DOCK8 (Table S2), etiologies of HIES if present in homozygosity. However, the CADD score of the ZNF341 variant is below MSC and therefore considered unlikely to have any deleterious effects. Although the DOCK8 variant has a CADD score above the MSC, in the context of a high GDI-Phred index, DOCK8 is a highly mutated gene, also strongly arguing against a functional role of this variant.

All homozygous OSMR variants reported in gnomAD v4.1 were plotted with CADD-MAF pViz plot, showing 11 homozygous OSMR variants with a CADD score >20 and above the MSC and less frequent than 0.1 (Fig. 2 A). The MAF varies among these 11 variants from very frequent 2.54 × 10⁻² (G587D) to very rare 3.72 × 10⁻⁶ (G927E), with the patient variant appearing at a MAF of 3.39 × 10⁻³, and as depicted, the variants are localized in different domains of the OSMR molecule (Fig. 2 B). Therefore, it was important to clarify whether the OSMR variant identified in P1 is functionally deleterious and verify that this is not the case for the remaining predicted deleterious variants in the general healthy population. The majority of these 11 OSMR variants are localized in the FNIII domains of the extracellular domain of the OSMR molecule, as is also the case for the OSMR variant identified in P1, and consistent with the position of other disease-causing variants in general within the OSMR gene (46, 51). The amino acid substitution in the OSMR variant of P1 is located in the beginning of the second of the four FNIII domains of OSMR (31).

To functionally study the OSMR variants in a biochemical assay on an isogenic background, we generated HEK293 cells deficient in both OSMR and LIFR to impair the OSM signaling. Two single-cell clones (SCCs) with OSMR_LIFR double KO were generated by CRISPR/Cas9 technology (Fig. 2 C), and functional studies showed that both clones were impaired in phosphorylation of STAT3 in response to OSM stimulation, while STAT3 phosphorylation was similar downstream of OSMR WT and OSMR P1 variant downstream of IL-6 stimulation (Fig. 2, D and E). IL-11 stimulation activated STAT3 to a lesser extent than OSM and IL-6, so it was difficult to conclude whether this downstream signaling was compromised. With this functional validation of the OSMR_LIFR double KO SCCs, it was safe to continue with overexpression of OSMR WT and the different variants followed by OSM stimulation to examine for STAT3 phosphorylation as a functional measure of OSMR signaling capacity (Fig. 2, F and G). These data showed that the P1 OSMR variant was markedly impaired in phosphorylation of STAT3 almost at the same level as transfection control (GFP) and in strong contrast to the robust STAT3 phosphorylation observed in SCCs expressing OSMR WT. Importantly, the other 10 OSMR variants reported in gnomAD all showed STAT3 phosphorylation comparable to OSMR WT in response to OSM stimulation, demonstrating intact signaling and no functional defect. Based on this, we conclude that the biallelic OSMR variant harbored by P1 is functionally defective, unlike the other OSMR variants with CADD score >20. Interestingly, when expressing the different OSMR variants, two differentially migrating bands were consistently observed. Notably, the V436D variant was the only variant that did not appear to have an upper band >150 kDa, which was the most prominent band for OSMR WT and most variants. The OSMR C687G variant appeared to have very little upper OSMR band, but no defect was observed with regards to STAT3 phosphorylation after OSM stimulation (Fig. 2 F). Finally, N-deglycosylase PNGase treatment of cell lysates showed that the upper OSMR bands of the many OSMR variants seen on western blotting (WB) represent N-linked glycosylated OSMR forms, since these upper bands disappeared after PNGase treatment, thereby indicating that the patient OSMR variant is defective in N-glycosylation (Fig. S3 C).

Since OSMR can cooperate with other receptor subunits of the IL-6 superfamily, we also tested responses to other IL-6 superfamily cytokines. We found that the OSMR variant from P1 impacted on the OSM signaling pathway exclusively, since signaling in response to IL-6 and IL-11 were both similar between OSMR WT and variant (Fig. 2, H and I). Since HEK293 cells do not express the α receptor subunit of the IL-31 receptor (IL31RA), which forms a receptor complex with OSMR, no STAT3 phosphorylation was detectable after IL-31 stimulation, although an effect of OSMR variant compared to OSMR WT here cannot be excluded.

Although OSMR harbors 13 N-linked glycosylation sites and one O-linked glycosylation site, none of these glycosylation sites include residue 436. To examine further the effect of the variant on OSMR protein expression, we performed immunoblotting on cell extracts from P1 and healthy controls treated with the N-deglycosylase PNGase F. This treatment resulted in the WT OSMR protein migrating further in the gel close to the gel position for OSMR V436D, whereas the OSMR V436D variant almost completely disappeared, suggesting deficient N-glycosylation of the OSMR variant and impaired N-glycosylation possibly underlying the increased destabilization of the OSMR variant protein (Fig. 3, A and B), in accordance with the results of expression of PNGase-treated OSMR variants in the HEK293 model system (Fig. S3 C). Treatment of fibroblasts with the protein synthesis inhibitor cycloheximide, however, showed similar degradation kinetics between OSMR V436D and OSMR WT, thereby suggesting N-glycosylation as a key determinant for the very low expression of the OSMR V436D variant rather than accelerated degradation (Fig. 3, C and D).

Next, we investigated signaling downstream of OSM stimulation. Whereas OSM stimulation of fibroblasts from controls induced strong phosphorylation of STAT3 and STAT5, this signal was almost completely abolished in patient fibroblasts, as well as more moderately reduced phosphorylation of ERK1/2 in patient compared to healthy controls (Fig. 3, E and F). The defective STAT3 phosphorylation in patient fibroblast harboring OSMR V436D was exclusive to the OSMR/gp130 receptor complex, since IL-6 and IL-11 stimulation resulted in marked and comparable phosphorylation levels of STAT3 between P1 and controls in fibroblasts (Fig. 3, G–J), demonstrating the integrity of STAT3 phosphorylation in response to other stimuli than those binding to OSMR. In addition to gp130, OSMR can also form a heterodimer with IL31RA, thereby creating a receptor complex for IL-31. As observed in HEK293 cells, no STAT3 activation was observed after IL-31 stimulation (Fig. 3, G and H), which may be due to IL31RA surface expression being very low compared to OSMR (32, 53). Functionally, OSM-induced STAT3 and ERK1/ERK2-dependent gene expression was compromised, since expression levels of IL-6, IRF7, CXCL10, and ICAM1 after 6-h OSM stimulation were reduced in P1 fibroblasts compared to those of controls, reaching statistical significance for IL-6 and CXCL10 (Fig. 3 K).

Similar to the observations in fibroblasts, OSM-induced STAT3 phosphorylation was reduced in patient peripheral blood mononuclear cells (PBMC)s (Fig. 4, A and B). However, OSM only activated STAT3 when using high concentrations (100 ng/ml of OSM which is 20-fold more than the 5 ng/ml used for fibroblasts), reflecting less OSMR surface expression on PBMCs as compared to fibroblasts. Accordingly, only a weak STAT3 phosphorylation signal was detectable after OSM stimulation, unlike the signal from the majority of healthy control PBMCs (Fig. 4, A and B). Moreover, IL-6-induced STAT3 phosphorylation was consistently slightly reduced in patient PBMCs compared to control, as was IL-31-induced STAT3 phosphorylation, although the response in controls was relatively weak and variable (Fig. 4, C and D). Intriguingly, OSM-stimulated patient PBMCs also expressed significantly lower IL-6 compared to controls, and this effect was rather specific since it was not observed in response to IL-6 nor IL-21 (Fig. 4 E). The signals observed in response to OSM stimulation of PBMCs suggested some degree of OSMR expression in this compartment, although much less than that observed for fibroblasts. Therefore, we next evaluated OSMR expression of various hematopoietic cell types within the PBMC population, including CD4⁺ and CD8⁺ T cells (Fig. S4 A) and CD27⁺ memory B cells (Fig. S4 C), which did not show any OSMR production. This is generally consistent with previous studies (35, 54) and with single-cell gene expression of different leukocyte subsets (55, 56). Although previous reports showed upregulation of OSMR on CD4⁺ and CD8⁺ T cells and CD27⁺ memory B cells by anti-CD3/CD28 activation/stimulation and CpG ODN class B, respectively (35, 54), we did not observe any significant/robust OSMR surface expression on patient nor control T cells after 5 days of anti-CD3/CD28 stimulation (Fig. S4 B) and neither on patient and control B cells stimulated with CpG ODN for up to 9 days (Fig. S4 C). Finally, lipopolysaccharides (LPS) and IFNγ treatment has been reported to induce surface expression of OSMR and IL31RA on monocytes and macrophages (32, 53) giving rise to successful signaling events by OSM and IL-31, respectively. PBMCs were subgated as shown in Fig. 4 F and subsequent flow cytometric analysis revealed OSMR surface expression on monocytes (Fig. 4, G and H) following LPS and IFNγ treatment, whereas OSMR was undetectable on lymphocyte and natural killer (NK) subsets. Surface OSMR expression was significantly higher for monocytes of healthy controls than from the patient (Fig. 4 H). Despite the increased surface expression of monocyte OSMR after LPS+IFNγ pretreatment, which was also reflected in weak STAT3 phosphorylation, this response was only marginally further enhanced in response to OSM treatment (Fig. 4, I and J). Additionally, pretreatment with LPS and IFNγ did not markedly enhance IL-31–induced STAT3 phosphorylation, although this response became detectable in P1 PBMCs (Fig. 4, K and L). Collectively, these data show very modest and selective induction of OSM and IL-31 receptor signaling and capacity to induce STAT3 phosphorylation in the hematopoietic compartment, suggesting that the defective cellular defect in P1 is mainly exerted by the fibroblast/stromal compartment.

To evaluate potential defects in the B cell compartment, blood samples were taken from P1 and one healthy control, as well as healthy blood donors, and purified PBMCs were analyzed by FC (Fig. 5, A and B; and Fig. S5 A). P1 displayed a significantly increased frequency of atypical memory B cells (aMBCs), and plasmablasts (PBs) (PBs defined as IgD⁻CD27⁺CD20⁻CD38⁺) (Fig. 5, A and B). Conversely, significantly fewer conventional MBCs were observed, in agreement with previous observations of a reduction in memory B cells in classical AD-HIES (57). Next, we assayed class-switch recombination (CSR) in the B cell compartment (Fig. 5 C and Fig. S5 B). This revealed an appreciable fraction of IgE-positive cells among the expanded aMBC/PC and PB subsets in the patient, whereas IgE-positive cells could not be detected in the healthy controls, in agreement with the rarity of this subset (58) (Fig. 5 C). To evaluate whether the observed perturbations in the patient’s B cell compartment were B cell intrinsic or extrinsic, we purified untouched naïve B cells by magnet-activated cell sorting and tested their behavior in an induced germinal center B cell (iGB) culture system (Fig. 5 D). This system leverages a normal human dermal fibroblast (NHDF) feeder cell line engineered to express human CD40L in combination with addition of recombinant human IL-21, B cell activating factor (BAFF), and IL-2, and a pulse of IL-4, to drive Th2-like activation, expansion, PB differentiation, and CSR (59). In this system, cells from P1 exhibited slightly lower initial proliferation than the healthy control cells (Fig. 5 E) but differentiated and class-switched to an extent comparable to the control (Fig. 5 F). This demonstrated that naïve B cells from the patient did not harbor any major intrinsic defect. Taken together, our findings suggest a B cell–extrinsic defect in the patient, causing an elevated aMBC/PC and PB frequency with increased IgE class-switching, commensurate with the HIES phenotype.

To demonstrate conclusively that the aberrant fibroblast phenotype of P1 was attributable to the biallelic OSMR V436D variant, we complemented dermal fibroblasts from P1 with OSMR WT or the OSMR variant by lentiviral transduction (Fig. 6, A–D). Reconstitution of patient fibroblasts with OSMR WT indicated normalization of OSMR glycosylation (upper band in immunoblot Fig. 6 A) as well as STAT3 phosphorylation in response to OSM stimulation (Fig. 6, A and B). Importantly, this effect was not observed when OSMR V436D was overexpressed, neither in the patient nor in healthy control fibroblasts (Fig. 6, A and B). Likewise, reconstitution of patient fibroblasts with OSMR WT also reverted the defect in OSMR surface expression with increasing levels observed exclusively in WT OSMR complemented fibroblasts and not in fibroblasts expressing OSMR V436D variant (Fig. 6, C and D). Collectively, these data demonstrate causality between the OSMR patient variant and impaired OSMR surface expression and abolished STAT3 phosphorylation in patient fibroblasts, suggesting that homozygosity for this novel OSMR variant alters the transcriptomic and phenotypic signature of dermal fibroblasts causing HIES.

Here we report a patient with HIES, presenting with atopic dermatitis and eczema, skin and wound infections, and CCPA together with IgE levels up to 34,000 IU/ml, homozygous for a rare variant in OSMR. We demonstrate that the patient OSMR V436D variant is deleterious, unlike other homozygous OSMR variants with CADD >20 reported in gnomAD, associated with altered N-linked glycosylation, and reduced surface expression of OSMR on patient fibroblasts. Collectively, this result in impaired OSM-induced STAT3 phosphorylation and significantly reduced OSM induction of IL-6 and CXCL10, combined with elevated fractions of IgE-producing PBs and aMBCs in the context of normal Th17 fractions. Reconstitution of patient fibroblasts with OSMR WT restored OSMR expression and functionality, demonstrating that the STAT3 signaling defect observed in fibroblasts is driven by the V436D variant and suggesting disease causality. Utilization of an iGB culture system did not reveal any B cell–intrinsic defect but instead suggested a B cell–extrinsic defect. Based on the clinical history and pathology in the patient, the genetic findings, and the molecular and cellular immunological characterization, we propose that homozygous OSMR deficiency represents a novel genetic etiology of AR-HIES, phenotypically mimicking AD-HIES through an effect upon STAT3 phosphorylation and signaling in fibroblasts.

OSMR deficiency is a novel genetic etiology of HIES, exhibiting a somatic and infectious phenotype reminiscent of AD-HIES caused by DN STAT3 mutations (6, 7), yet with distinctive features and a more narrow phenotype. Despite many similarities, including hyper-IgE, skin infections and poor wound healing, fungal infections, and eczema, together with a somatic phenotype involving abnormalities in bone and teeth, a notable difference between our patient and STAT3 AD-HIES (and AR ZNF341 HIES) is the normal levels of Th17 cells. Any constitutive STAT3 deficiency in hematopoietic cells, being either structural in DN LOF STAT3 variants or due to defective induction of the STAT3 protein in ZNF341 deficiency, compromises Th17 differentiation (22, 60). This discrepancy reveals that OSMR deficiency is similar, but not identical, to AD-STAT3 HIES. A plausible explanation is the more fundamental disturbance of STAT3 signaling downstream of several cytokine receptors, including IL-6, IL-21, TGFβ, OSM, and IL-31 in AD-HIES, caused by DN STAT3 mutations as opposed to select OSMR defect. OSMR belongs to the type I cytokine receptor family, and therefore, it was important to clarify whether other receptor signaling pathways within this family may be affected by the identified deleterious OSMR variant in the patient. For instance, an IL6ST variant was reported to cause cytokine-selective signaling impairment (specifically loss of IL-11 signaling with intact signaling by other IL-6 family cytokines), demonstrating that receptor alterations can quantitatively and qualitatively change downstream signaling (61). However, overexpression analysis of patient OSMR V436D in OSMR_LIFR double-defective HEK293 cells did not negatively impact upon STAT3 activation after IL-6 stimulation, indicating that the OSMR V436D variant exclusively disturbs signaling in the receptor complexes OSMR–gp130 and OSMR-IL31RA. This was confirmed by results of IL-6 stimulation of patient fibroblasts and PBMCs, which exhibited similar responses as healthy controls.

Given that the extent of immunological deficiency is also determined by the subsets of cells affected, our finding that OSMR defect almost exclusively affects fibroblast signaling may partially account for the limited degree of cytokine dysregulation, and possibly explain the relatively untouched Th17 cell subsets in the patient. This is because STAT3 signaling in response to other cytokines than OSM and IL-31 is fully preserved in primary patient cells, as also demonstrated by our findings of normal IL-6 and IL-11–induced STAT3 phosphorylation. Indeed, the full HIES phenotype of IL6R defect in AD-HIES suggests that defective IL-6 signaling is one of the key pathogenic features of HIES. Whether the reduction in IL-6 production from OSMR-deficient patient fibroblasts we observed contributes to the observed immunological and clinical phenotype remains unclear. Intriguingly, patients with AR-HIES caused by DN IL6ST deficiency also have preserved Th17 production and seldomly present with skin abscesses and CMC but mostly suffer from abnormalities in bone and teeth (15, 27). Similarly, Th17 production is normal in most patients with AD-HIES caused by STAT6 GOF mutations, whereas T cells show Th2 skewing due to the central role of STAT6 in IL-4 and IL-13 signaling in Th2 cells (17). We did not find Th2 skewing nor increased IL-4 in peripheral blood in the patient, thus leaving the severe atopy and asthma phenotype partly unaccounted for. Altogether, the similarities concerning preserved Th17 fractions may argue for placing AR OSMR deficiency in a functional and immunological category with IL6ST and STAT6 GOF forms of HIES.

Pathological disease manifestations from various tissues, including bone, gingiva/teeth, skin, airways, and blood, collectively support a clinical significance of the identified OSMR deficiency. In relation to bone, the regulation of osteoblast–osteoclast interaction involves OSM, which induces the differentiation of osteoblasts from mesenchymal stem cells via STAT3 (62), while also affecting osteoclast differentiation and subsequent bone resorption (63). It is tempting to attribute the fragility fracture and osteopenia to distorted osteoblast–osteoclast homeostasis, as part of the OSMR deficiency. Fibroblast-expressed OSMR is also involved in tissue repair and remodeling (35, 39), suggesting that the poor wound healing reported by the patient may be directly related to OSMR deficiency. Binding of OSM to fibroblast OSMR induces the production of collagen and other extracellular matrix components as well as proinflammatory cytokines, including IL-6 (39, 64, 65), which may be responsible for some aspects of the immune dysregulation manifested by the patient.

A major finding of significant interest is the observation that OSMR deficiency exhibits a largely fibroblast-confined cellular phenotype, secondarily affecting IgE-producing PBs. It may appear surprising that an immunological phenotype almost exclusively relying on a fibroblast defect can cause such a marked IEI, with not only tissue and bone abnormalities, but also some immune dysregulation and infection. Several points may account for these observations. The fibroblast has gained increasing attention as an important immune modulatory cell (66) not only for innate immunity (67) but also as an immune determinant in B cell–associated germinal center reactivity (68). Especially, fibroblasts serve as essential components of the lymph node microenvironment, supporting lymphocyte migration, localization, and interaction within lymph nodes (69). Fibroblasts can form specialized subsets supporting B and T cell compartmentalization through secretion of various chemokines and cytokines (70). The fact that fibroblasts express OSMR and different leukocyte subsets are important sources of OSM (39), renders it plausible that functional loops between lymph node fibroblasts and lymphocytes exist. This scenario agrees with our data from the iGB culture system, which did not reveal any B cell–intrinsic defect but instead suggested a B cell–extrinsic defect, the latter consisting of abnormal signals from fibroblasts. The elevated levels of circulating, IgE expressing, aMBCs as well as PBs in the patient testified to ongoing germinal center activity, consistent with modestly increased fractions of pTFH. which represent main producers of IL-4 within lymph nodes (71). However, this increased germinal center activity was likely constrained to draining lung lymph nodes, as we, perhaps somewhat surprising, observed no spillover of this activity to the circulation, neither in the form of atopy-biased Th subsets, such as elevated fractions of circulating Th2 or Th9 cells, nor by increased levels of circulating IL-4 or diminished IFNγ levels, respectively. Finally, despite previous reports (55, 56), our data on resting and stimulated patient and control B and T cells (55, 56) did not support the notion that lymphocytes can be induced to express notable levels of OSMR at the cell surface. However, after LPS-IFNγ pretreatment, we did find detectable levels of OSMR surface expression on control monocytes associated with low levels of STAT3 phosphorylation, which was more pronounced in controls than patient PBMCs. It is also notable that the weak OSM-induced STAT3 phosphorylation present in control PBMCs was completely abolished in patient PBMCs. Collectively, we cannot exclude the possibility that OSMR may be expressed at low levels and/or upregulated by blood cells in response to certain stimuli during infection and inflammation, in which context OSMR deficiency may implicate the hematopoietic system.

The present identification of biallelic deleterious OSMR variants in a patient with HIES is the first report of homozygous OSMR deficiency. Given that the OSMR V436D variant of the patient is not among the rarest in the general population, and since several other predicted deleterious variants with CADD >20 are found in gnomAD, it was important to establish the true cumulative pLOF of functionally deleterious variants. Indeed, we found that the OSMR V436D variant harbored by the patient is the only functionally defective variant among the 11 identified OSMR variants, including variants with higher CADD scores located in the same fibronectin domain as V436D, highlighting the particular relevance of OSMR V436D. Based on our findings of a significantly reduced OSM-induced STAT3 phosphorylation with slight residual signaling in patient fibroblasts as well as in the HEK 293 OSMR_LIFR double KO cell model, the OSMR V436D variant should probably be characterized as hypomorphic rather than complete LOF.

Heterozygosity for OSMR variants affecting the fibronectin domains of the molecule cause AD PLCA, characterized by itchy, thickened, skin with amyloid deposits in the superficial dermis (45, 46, 47). It is hypothesized that distorted IL-31–induced neuronal signaling, mediated through variant OSMR proteins, may elicit the severe pruritus in FPLCA (46). The atopic skin phenotype of our patient, however, did not resemble the one characteristic of FPLCA, defined by grouped hyperkeratotic papules (lichenoid amyloidosis) or macular hyperpigmentation, and notably, a skin biopsy did not reveal amyloid deposits in the dermis in the patient (46). Thus, the disparity in skin manifestations associated with mono- and biallelic OSMR variants, respectively, is intriguing. Recently, a homozygous variant in OSM, reported in a consanguineous family, was associated with inherited severe bone marrow failure syndrome evidenced by profound anemia, thrombocytopenia, and neutropenia, but no increased infection susceptibility (48). This clinical phenotype highlights the importance of OSM for erythro- and myelopoiesis (72), also supported by findings that mice deficient in OSM signaling show mild thrombocytopenia and anemia (45, 73). However, our patient displayed no evidence of bone marrow failure, except for a mild anemia that may also be attributed to long term Aspergillus infection. Due to the OSMR defect in our patient, OSM can only signal through variant OSMR–gp130 receptor complexes or the alternative OSM receptor constituted by LIFR and gp130. Since defects in LIFR-gp130 causing Stüve-Wiedemann syndrome is not associated with hematological abnormalities, the absence of affected erythropoiesis and myelopoiesis in our patient is a conundrum, suggesting alternative pathways for OSM to mediate its hematologic effects. The LIFR defect in Stüve-Wiedemann syndrome involves severe skeletal dysplasia, respiratory distress, and dysautonomia, none of which the patient in this paper suffers from. Collectively, based on the phenotype of Stüve-Wiedemann syndrome, defective LIF signaling may be distinguished clinically from defective OSM signaling by the neurological and severe bone-related defects in the former and the cutaneous and hematological defects of the latter (74). Indeed, it has been suggested that craniosynostosis, deciduous teeth, scoliosis, and osteoporosis in HIES are caused mainly by insufficient responses to IL-11 or LIF (75).

Finally, the clinical presentation and cellular phenotype of our patient did share some similarities to patients with dysregulated IL-31RA signaling, such as atopic dermatitis (76) and increased IgE (77, 78), suggesting involvement of defective IL-31–induced OSMR-IL31RA signaling in generation of the clinical phenotype of the patient. Likewise, we would expect IL-31 signaling in patient OSMR V436D cells to be reduced, as was previously documented for another OSMR variant (G618A) located within the second fibronectin domain of OSMR and leading to decreased STAT3 activation in response to both OSM and IL-31 in FPLCA keratinocytes (51). However, we did not have access to any patient cell types in which IL-31R was highly expressed and could therefore not establish the extent of IL-31 signaling defect with certainty. Interestingly, IL-31RA deficiency in mice results in exaggerated Th2-type inflammation in the lungs (77, 78, 79), whereas the patient of the present study did not have elevated peripheral Th2 cell fractions nor increased serum IL-4 levels, perhaps suggesting that any IL31R signaling defect may be modest in the patient given the normal Th2 cell fractions. Altogether, the specific impact of the OSMR V436D variant downstream of IL31RA/OSMR heterodimer signaling remains to be fully established.

Based on the results of this study, we propose OSMR deficiency as a novel genetic etiology of AR-HIES, closely mimicking the clinical and immunological phenotype of AD-HIES through the influence on STAT3 phosphorylation and activation of downstream effector molecules. The clinical phenotype of OSMR deficiency, however, is distinctive, since the defect is largely restricted to stromal cells/fibroblasts, and IL-6 signaling as well as Th17 subsets are preserved. In particular, this study suggests a selective and almost exclusive role for fibroblasts in mediating the OSMR AR-HIES phenotype. Due to the pleiotropic effects of OSM on mesenchymal cells involved in bone mineralization and immunity, the altered fibroblast gene signature in response to OSM is consistent with the patient’s decreased bone mineralization, eczema and skin infection, pulmonary aspergillosis, and an intra-nodal environment conductive to elevated IgE and TFH cell subset formation. Extensive genetic, immunological, and phenotypic variation has been reported in the different forms of AD- and AR-HIES, underscoring the challenge of grouping such heterogeneous disorders by the shared feature of the classic triad of elevated IgE, atopic dermatitis, and recurrent infections. The present study, together with other recent publications, sheds new light on different clinical disease manifestations and immunological phenotypes arising from defective OSMR/gp130/LIFR signaling, manifesting as heterozygous OSMR defect in FPCLA, homozygous OSM defect with congenital hematological bone marrow defect syndrome, or LIFR defect in Stüve-Wiedemann syndrome with neuropathic itching and bone dysplasia.

Despite the data presented here, some issues remain to be reconciled with the existing literature, including the different phenotypes of OSM deficiency and monoallelic OSMR deficiency, which are both distinct from biallelic OSMR deficiency as described here. The identification of additional patients carrying mutations in genes within this pathway, including OSMR, LIFR, OSM, IL31, and IL31R, as well as careful clinical phenotype description of individuals in large population datasets, will be essential to clarify the underlying molecular mechanisms and phenotypic relationships. As an increasing volume of clinical characteristics and immunological features of HIES are reported, a better understanding of the clinical heterogeneity and complex underlying immunology and pathophysiology emerges. Importantly, this study provides new knowledge with direct medical implications. First, because the diverse organ manifestations deriving from bone, skin, and lungs present diagnostic challenges and point to more complex ways, in which OSM and IL-31 mediate their physiologic effects. Second, as we suggest that OSMR should be part of the panel of genes investigated in patients with a clinical phenotype involving features of atopy and HIES. Undoubtedly, additional genetic etiologies remain to be identified and characterized to provide a more comprehensive picture of this intriguing condition and the associated immune dysregulation predisposing to elevated IgE and atopy with staphylococcal and fungal infections in skin and lungs in humans.

We investigated a 57-year-old man of Caucasian (Danish) descent with a clinical history of severe atopic dermatitis, eczema, and asthma throughout life. He experienced slow wound healing and frequent secondary staphylococcal skin infections and pustules, but only seldomly skin abscesses. Dental status was poor with tooth retraction at a young age and a previous history of a tooth abscess (Fig. 1 A). From the age of 20 years, he experienced one or more bacterial pneumonias per year, as well as episodes of oral candidiasis but not CMC or invasive candida infection. At the age of 45 years, he had a distal radial fracture in response to a low energy trauma, a subsequent bone density scan showing osteopenia, originally presumed to be related to prolonged steroid therapy for his asthma. Due to prolonged coughing, dyspnea and chest pain he was referred to the Department of Pulmonology on suspicion of lung cancer. However, computed tomography of the chest showed signs of chronic pulmonary aspergillosis with a thick-walled cavity in the left upper lobe characteristic of an aspergilloma (Fig 1 B). Bronchoscopy revealed A. fumigatus DNA and a positive AGA (AGA: 1.9). Blood tests at that time showed increased Aspergillus IgG (IgG;157 mg/L, normal range <26.9 mg/L) and IgE (37,900 IU/L, normal range <350 IU/L), slightly increased eosinophil count (0.66 × 109/L, normal range <0.5 × 109/L), and a very high IgE (IgE: 34,000 × 103 IU/L, normal range <100 IU/L) (Table S1). A diagnosis of chronic cavitary pulmonary aspergillosis (CCPA) was established and the patient was treated with itraconazole for 8 mo. Due to disease progression, he was changed to voriconazole for 3 mo, but due to side effects changed again to posaconazole and after 1.5 years of therapy symptoms became stable and stopped therapy. Due to recurrent pneumonia during follow-up, he was started on prophylactic sulfamethoxazole and trimethoprim (Bactrim) prophylaxis. Clinical immunological evaluation and immune phenotyping. A skin biopsy showed perivascular inflammation consisting of lymphocytes and eosinophils but no amyloid deposits.

Standard immunological screening revealed normal serum concentrations of IgA, IgM, and IgG (including IgG subclasses) (Table S1). Serum IgE was markedly elevated to levels ranging from 18,200 to 34,000 IU/ml. (Fig. S1 and Table S1). Recurrent monocytosis and mild eosinophilia, interpreted as secondary to CCPA, was also noted (Table S1). Representative hematology values showed mild normocytic anemia in the range of between 6.8 and 7.8 mmol/L (normal range 8.3–10.5 mmol/L, but normal platelet and leukocyte concentrations) (Table S1). Peripheral CD19⁺, CD4⁺, and CD8⁺ lymphocyte concentrations were normal; however, patient CD4⁺ T cells displayed an activated (HLA-DR⁺: 30%) phenotype (Table S1). Lymphocyte subset analysis showed normal distributions of naïve and memory CD19⁺, CD4⁺, and CD8⁺ lymphocytes, respectively, as well as normal Th2 (CD4⁺ IL-4Rα⁺), Th9 (CD4⁺ CXCR3⁺ CCR4⁻ CCR6⁺), and Th17 (CD4⁺ CD161⁺ CCR4⁺ CCR6⁺) subsets, except for slightly elevated pTFH cell (CD4⁺ CXCR5⁺ PD1⁺ ICOS⁺) frequencies in the patient (Table S1 and Fig. S2), the latter consistent with the patient’s marked ongoing IgE response. Serum levels of IL-4 and IFNγ were likewise within normal ranges. In response to anti-CD3/CD28 stimulation, patient CD4⁺ T cells displayed similar proliferation characteristics as those of controls (Fig. S3 B). Patient CD8⁺ T cells responded more indolent to CD3/CD28 stimulation than control CD8⁺ T cells, as evidenced by a reduced proliferation index and a diminished responding fraction (Table S1). Stimulated patient CD4⁺ T cells displayed no relative expansion of Th2 cells (CD4⁺ IL-4Rα⁺ T cells) compared to those of controls (Fig. S3 B). The patient’s vaccination titers to diphtheria, tetanus, and pneumococcal (23-valent polysaccharide) vaccines were normal.

The patient was included in the study by written consent. Whole blood samples were obtained in lithium heparin tubes for WGS and PBMC isolation by Ficoll density gradient centrifugation using SepMate PBMC isolation tubes (#86460; STEMCELL Technologies) and frozen in liquid nitrogen at the Department of Clinical Immunology, Odense University Hospital (Odense, Denmark). Control PBMCs were purified from healthy donors after obtaining written consent at Department of Biomedicine, Aarhus University (Aarhus, Denmark). A skin biopsy from the patient was taken under local anesthesia, and dermal fibroblasts were subcultured at the Department of Clinical Immunology, Odense University Hospital (Odense, Denmark). NHDFs were obtained from PromoCell.

The patient and healthy controls were included following oral and written consent in accordance with The Helsinki Declaration and national ethics guidelines and after approval from the Danish National Committee on Health Ethics (1-16-02-621-16), the Data Protection Agency, and Institutional Review Board.

DNA was extracted from 400 μl blood using the QIAsymphony DSP DNA Midi Kit with an elution volume of 100 μl. The DNA concentration was measured with Qubit dsDNA BR. An input of 300 ng DNA was used for library preparation using the Illumina DNA PCR-Free Prep. The library was sequenced on an Illumina NovaSeq 6000 with a S4 flow cell using 2 × 150 bp paired ends sequencing. The paired end reads were mapped to the hg38 reference genome with bwa mem v0.7.15 (80, Preprint), duplicate reads were flagged with Picard v2.21.9 MarkDuplicates (https://github.com/broadinstitute/picard), and variants were called using GATK v4.1.9.0 HaplotypeCaller (https://github.com/broadinstitute/gatk). The mapped reads resulted in a median coverage of 40×.

VarSeq software (version 2.3.0, Golden Helix, Inc.) was used for annotation and filtering of variants. Population frequency (gnomAD Exomes Variant frequencies version 2.0.1, Broad Institute) threshold was set to 1% to exclude common polymorphisms. In addition, sequence ontology was applied so only non-synonymous variants in coding regions of genes or variants, including structural variants predicted to affect translation, were considered. Variants were classified using American College of Medical Genetics and Genomics criteria (81) in part based on annotations, including CADD version 1.5 score (82), gnomAD frequency, ClinVar submission interpretations, and HGMD Pro version 2022.4 variant classification. The MSC of CADD value with 95% confidence interval was determined by Zhang et al. (83). The cumulative pLOF was calculated by summarizing all allele frequencies (from gnomAD version 4.1.0) from the respective gene. The gene damaging index (GDI-Phred) was determined by Itan et al. (84).

We have investigated the presence of genetic variants in introns and exons, including 5′ and 3′ UTR and promoter sequences, of the following genes: AIRE, CARD9, CARD11, CARMIL2/RLTPR, DOCK8, ERBIN, FOXP3, IL6R, IL6ST, IL12RB1, IL17RA, IL33, LRBA, PEPD, PGM3, SMAD3, SPINK5, STAT1, STAT3, STAT6, TGFBR1, TGFBR2, WAS, and ZNF341. This list comprises genes associated with HIES based on The International Union of Immunological Societies Expert Committee (85) and genes associated with the phenotype HIES in the Human Gene Mutation Database variant database. Genetic analysis of these genes revealed a monoallelic deletion in intron 1 of the SMAD3 gene with a frequency of 0.4477 in the gnomAD and DECIPHER databases. This variant was assessed not to be causal of the patient phenotype due to the high frequency in the healthy population. No other copy number variations were identified in the aforementioned genes.

We further identified 68 variants with allele frequency <0.01. Filtration for CADD >15 reduced the list to four monoallelic variants: a missense variant NM_203447.4:c.380G>A, NP_982272.2:p.Arg127His in DOCK8, a missense variant NM_001282933.2:c.2309C>T, NP_001269862.1:p.Pro770Leu in ZNF341, an in-frame deletion c. NM_014339.7:c.2446_2454delGAAGAGGAG, NP_055154.3:p.Glu816_Glu818del in IL17RA, and the missense variant in OSMR described in the manuscript. We ruled the DOCK8 variant as not causal, since HIES is primarily described in patients with homozygous DOCK8 deficiency, and this specific variant has been found in >700 healthy heterozygous individuals in the gnomAD database. In addition, the patient did not have an increased number of CD45RA⁺CD8⁺ T cells, which is inconsistent with DOCK8 deficiency. Variants in ZNF341 are, like DOCK8 variants, known to cause HIES with AR inheritance. HIES due to ZNF341 variants is characterized by increased IgG, increased number of naïve B cells, alopecia, aphthous stomatitis, food allergy, and syndromal features, neither of which is present in the patient (21, 22). Finally, the patient has normal STAT3 phosphorylation in response to IL-6 stimulation (Fig. 3, G and H; and Fig. 4, A–D), which is not consistent with either DOCK8 or ZNF341 deficiency (21, 86). The variant in IL17RA is located close to the 3′end of the gene, has been identified in over 600 healthy heterozygous individuals in the GnomAD database, and the patient phenotype has limited overlap with IL17RA deficiency (87). Therefore, we also ruled out that this variant as causal of the patient phenotype. The 64 variants filtered out by CADD >15 are intronic variants. These variants were assessed using the splicing prediction tools GeneSplicer, MaxEntScan, NNSplice, and NCBI’s Position Weight Matrix. Only one of these variants, present in the LRBA gene, was predicted by three out of four tools to introduce a cryptic splice site. However, as LRBA-associated HIES follows recessive inheritance, this intron variant was considered not to be the causal variant.

Plasmids with OSMR WT and OSMR variants were purchased and cloned in the vector pGenLenti by GenScript. The plasmids containing the different OSMR variants were validated by sequencing. GFP was used for transfection control and was overexpressed on pCCL/PGK-eGFP plasmid. For transfection, 20,000 HEK cells were seeded in a 48-well plate one day prior to transfection. On the transfection day, 100 ng plasmid DNA (122 ng plasmid for the plasmid containing WT OSMR) were mixed with up to 10 μl OPTI-MEM, and in another tube, 0.4 μl of the transfection reagent PEI (24765; PEI MAX - Transfection Grade Linear Polyethylenimine Hydrochloride [MW 40,000]) was mixed with 9.6 μl OPTI-MEM. These two tubes were separately incubated at room temperature (RT) for 5 min before they were pooled and followed by another incubation of 20 min at RT. Hereafter, the transfection mix was ready to be gently added to the respective cells. The best OSMR production in HEK cells was achieved after 48 h, which was shown by WB.

The following primary conjugated (anti-human) antibody specificities were used (BD): CD3 (FITC; PerCP-Cy5-5), CD4 (APC; PE-Cy7), CD8 (APC; APC-Cy7), CD16⁺CD56 (PE), CD19 (APC; PerCP-Cy5.5), CD27 (PE), IgD (FITC), κ-light chain (FITC), λ-light chain (PE), IgG1 κ-light chain isotype (mouse) (FITC), CD45 (PerCP-Cy5. 5), CD45RA (FITC), CD45RO (PE), HLA-DR (APC), TCRαβ (FITC), TCRγδ (PE), CD161 (FITC), CD183 (CXCR3, BV510), CD185 (CXCR5, PerCP-Cy5.5), CD194 (CCR4, PE), CD196 (CCR6, APC), CD278 (ICOS, BV421), and CD279 (PD1, PE). For analysis of intracellular IL-17A and OSMR surface expression, the anti-human FITC IL-17A and the APC-conjugated OSMR antibodies, both from (Thermo Fisher Scientific), were used. All samples were analyzed on a FACS Canto II (BD).

Cryopreserved PBMC were thawed and rested (in a 37°C incubator with 5% CO₂) in culture medium (RPMI 1640 + 10% AB serum) overnight. Next day, CFSE-stained PBMC were left unstimulated or stimulated with anti-CD3/CD28 Dynabeads (Thermo Fisher Scientific) in 96-well plates for 5 days according to the manufacturer’s instructions. The supernatants were harvested and proliferation of CD4⁺ and CD8⁺ T cells (2–4 wells containing 100,000 cells/well were pooled) as well as OSMR and IL-4Rα surface expression was ascertained. T cell proliferation was evaluated by two parameters: Proliferation index: mean number of cell divisions among the responding fraction (responding fraction: fraction of T cells undergoing at least one cell division [%]).

Cryopreserved PBMC were thawed and left overnight in culture medium. Monensin solution (Thermo Fisher Scientific) (dilution 1:1,000) was added to PBMC cultures (1 × 10⁶ cells/ml), which were then left unstimulated or stimulated with PMA (15.4 ng/ml, Merck, KGaA) and ionomycin (300 ng/ml, Thermo Fisher Scientific) for 4 h. Anti-CD4 surface-stained PMBC were fixed and permeabilized (FoxP3/Transcription kit, Thermo Fisher Scientific) and stained for intracellular IL-17A followed by flow cytrometric analysis.

NHDFs were seeded 1 day prior to stimuli/transfection in DMEM supplemented with 10% FCS and 1% Pen/strep (P/S). For WB, NHDFs were stimulated with 5 ng/ml OSM or 100 ng/ml IL-6 for 30 min and immediately put on ice. Cells were lysed and prepared for WB as detailed in supplementary methods. For RT-PCR, NHDFs were stimulated with 10 ng/ml OSM for 2 or 6 h mRNA was extracted according to manufacturer (#740466.4; MACHEREY-NAGEL).

Primary dermal fibroblasts were seeded in a 24-well plate with 25,000 cells per well. If the duration of the experiment would be <2 days, the DMEM was supplemented with 10% FCS and 1% P/S, and if the duration was >2 days, the DMEM was supplemented with 2% FCS and 1% P/S. Cells were always seeded 1 day prior to stimuli/transfection. PBMCs were seeded in RPMI supplemented with 1% FCS and 1% P/S with 500,000 cells per well. Fibroblasts were stimulated for 30 min with 5 ng/ml OSM (Human Oncostatin M [209 aa] Recombinant Protein, #300-10T-10UG; PeproTech) and immediately put on ice. For IL-6 and IL-31 stimulation of fibroblasts, cells were seeded in medium containing 1% FCS one day prior to stimulation. Cells were stimulated for 30 min with 100 ng/ml IL-6 (#200-06-5UG; PeproTech) IL-31 (#200-31-2UG; PeproTech) or 10 ng/ml IL-11 (#200-11-2UG; PeproTech) and immediately put on ice. PBMCs were stimulated with 100 ng/ml OSM, IL-6, or IL-31 for 30 min and immediately put on ice. PBMCs were pretreated with either 1 µg/ml LPS (from Escherichia coli O26:B6, L2654-1MG; Millipore Sigma) or 50 ng/ml IFNγ (Recombinant Human IFN-γ, #300-02; PeproTech) for 24 h prior to stimulation.

Whole-cell lysates were generated by washing the cells with ice‐cold PBS and lysed in Ripa buffer (Thermo Fisher Scientific) supplemented with PhosSTOP (Merck) and Complete Ultra protease inhibitor (Roche), Sodium fluoride (Avantar), and Benzonase Nuclease (Sigma‐Aldrich) for 5 min on ice and stored at −20°C. Protein concentrations were measured using Pierce BCA protein Assay Kit (#23223; Thermo Fisher Scientific). Cell lysates were denatured at 95°C for 5 min with 50 mM DTT (#43816-10ML; Sigma-Aldrich) and 4×Laemmli buffer (#1610747; Bio-Rad). Samples were loaded together with Precision Plus Protein Dual Color Standard protein marker (Bio‐Rad) onto a Criterion TGX Precast Midi Protein Gel in Nu PAGE MOPS SDS running buffer (NP0001; Thermo Fisher Scientific). Proteins were transferred onto a Trans‐Blot Turbo Midi Polyvinylidene Fluoride (PVDF) Transfer membrane (Bio‐Rad) using the Trans‐Blot Turbo Transfer System (Bio‐Rad). Membranes were washed using Tris‐buffered saline (Thermo Fisher Scientific) supplemented with 0.05% (vol/vol) Tween 20 (Sigma‐Aldrich) (TBS‐T), blocked for 1 h at RT in 5% Bovine Serum Albumin (BSA, Sigma‐Aldrich) in TBS‐T, washed with TBS‐T three times for 5 min each, and incubated overnight (o/n) at 4°C with primary antibody diluted 1:1,000 in 5% BSA in TBS‐T with 0.05% Sodium Azide. The following morning, membranes were washed with TBS‐T three times for 5 min each, incubated for 1 h with secondary antibody diluted 1:10,000 in 5% Skim Milk Powder (Sigma‐Aldrich), washed five times for 5 min each, and proteins were visualized using Clarity Western ECL Substrate (Bio‐Rad) or Clarity Max Western ECL substrate (Bio-Rad) on a ChemiDoc Imaging system (Bio-Rad). To visualize other proteins (for instance non-phosphorylated proteins) at or close to the initial visualized protein (for instance phosphorylated proteins), the membrane was stripped in Restore Western Blot Stripping Buffer (21059; Thermo Fisher Scientific) according to the manufacturer’s protocol prior to blocking, washing, and finally o/n incubation at 4°C in primary antibody solution.

The following primary antibodies were used: anti-pSTAT3 (pTyr705, cat. no. 612356; BD Biosciences), anti-STAT3 (cat. no. 610189; BD Biosciences), anti-pSTAT5 (pTyr694, cat. no. 611964; BD Biosciences), anti-STAT5 (cat. no. 610191; BD Biosciences), pERK1/2 (pThr202/pTyr204, phospho-p44/42 MAPK, cat. no. CST-9101S; BioNordika), ERK1/2 (p44/42 MAPK, cat. no. CST-9102S; BioNordika), anti-OSMR (cat. no. CST-95443; BioNordika), anti-GFP (cat. no. NB600-308SS; Novus Biologicals), anti-GAPDH (cat. no. ab9485; Abcam), and anti-Vinculin (rabbit, cat. no. CST-13901; Cell Signaling Technology). As secondary antibodies, peroxidase‐conjugated donkey‐anti‐rabbit and donkey‐anti‐mouse were used (711‐036‐152 and 715‐036‐150; Jackson ImmunoResearch). Specific bands were quantified by densitometry using the Image J software.

RNA was extracted from lysed cells using NucleoSpin 96 RNA Core Kit (#740466.4; MACHEREY-NAGEL), according to the manufacturer’s instructions and eluted in 75 μl RNase-free H2O. Quantitative RT-PCR was performed with TaqMan RNA-to-CT 1-Step Kit (#4392938; Thermo Fisher Scientific) using probes targeting TATA-box binding protein (TBP, #Hs00427620_m1; Thermo Fisher Scientific), CXCL10 (#Hs01124251_g1; Thermo Fisher Scientific), IL-6 (#Hs00985639_m1; Thermo Fisher Scientific), ICAM1 (#Hs00164932_m1; Thermo Fisher Scientific), and IRF7 (#Hs01014809_g1; Thermo Fisher Scientific), according to the manufacturer’s instructions. A QuantStudio 3 Real-Time PCR System (Applied Biosystems) was used for quantification of RNA levels relative to the expression of TBP using the 2^−ΔΔCt method (88), and the expression levels were normalized to the controls.

Patient and control (matched for sex and age) PBMCs were isolated from lithium-heparin-stabilized blood obtained following informed consent. Additional blood donor control PBMCs were purified from buffy coats obtained from anonymized healthy blood donors at the blood bank of Aarhus University Hospital, Denmark. Whole blood or buffy coat was diluted 1:1 with PBS, and PBMCs were isolated by a Ficoll-Paque PLUS (GE Healthcare) density gradient centrifugation using SepMate tubes (STEMCELL Technologies) with centrifugation at 1,200 g for 10 min at 20°C. The PBMCs were subsequently washed in PBS, centrifuged at 500 g for 10 min at 20°C, then incubated in PBS with 27 mM EDTA for 5 min at RT before centrifugation at 300 g for 8 min at 20°C, followed by another wash in PBS. For CSR analyses of PB, plasma cell and memory subsets, alternative sample preparation methods were assayed, including conventional Ficoll-Paque separation without SepMate tubes, as well as direct whole-blood erythrocyte lysis using RBC lysis buffer (155 mM NH₄Cl, 12 mM NaHCO₃, and 0.1 mM EDTA).

For the magnetic isolation of human naïve B cells, purified PBMCs were resuspended in 5 ml magnetic-activated cell sorting (MACS) buffer (PBS, 2% heat-inactivated fetal bovine serum [FBS], and 2 mM EDTA) at RT, before being counted, and then incubated 5 min on ice with Human BD Fc block (Clone Fc1, 564220; BD). Miltenyi Biotec Human Naïve B Cell Isolation biotin-antibody cocktail (130-091-150) was then added for another 30-min incubation. The samples were then diluted with MACS buffer and centrifuged at 200 g 10 min at 4°C. The pellet was resuspended in 1 ml MACS and incubated 20 min with Miltenyi Biotec anti-biotin magnetic beads, pre-diluted in 2 ml MACS buffer. The cell suspension was loaded onto pre-wetted Miltenyi Biotec LS columns through 70-μm cell strainers and washed two times with 3 ml MACS buffer. Collected flow-through of untouched human naïve B cells (CD19⁺, CD20⁺, CD22⁺, CD27⁻, and IgD⁺) was validated using FC. Cell aliquots of human B cells were frozen in freeze medium (90% FBS and 10% DMSO) using Corning CoolCell containers as recommended by the manufacturer.

CD40L-expressing NHDF feeder cells (kindly provided by Anne Rovsing and Jakob Giehm Mikkelsen, both Department of Biomedicine, Aarhus University, Aarhus, Denmark [59]) were passaged at least 8 days before being seeded and incubated in B cell medium (BCM; RPMI-1640 [+] L-Glutamine supplemented with 55 µM 2-Mercaptoethanol, 1% P/S, 10 mM HEPES, and 1 mM Sodium Pyruvate [all Invitrogen]) and 10% FBS. The following day (day 0), purified naïve B cells were seeded onto the feeder cells in BCM supplemented with recombinant human IL-4 (10 ng/ml), IL-21 (10 ng/ml), BAFF (10 ng/ml), and IL-2 (50 ng/ml) (#200-04, 200-21, 310-13, and 200-02, respectively; all from Peprotech). On day 3 and 5, half the culture volume of medium was exchanged with fresh BCM similarly supplemented with IL-21, BAFF and IL-2, but no IL-4. On day 7, B cells were harvested then subjected to the analysis of interest or reseeded on fresh feeder cells. Culture supernatants were stored with 0.1% NaN₃ for later time-resolved immunofluorometric assay (TRIFMA) analysis of secreted antibody isotypes as described below, and cell pellets were frozen in freeze medium for analysis by FC.

Cell suspensions of interest were prepared in FC buffer (PBS with 2% FBS and 1 mM EDTA), and 100 µL of sample was added 20 μl Human BD Fc Block (Clone Fc1, 564220; BD, pre-diluted 1/50), then preincubated for 5–10 min. Then 100 μl fluorophore-conjugated antibody panel of choice (see below) diluted in a 1:1 mix of FC buffer with Brilliant Stain buffer (563794; DB) was added. After 30 min of staining on ice, the cells were washed and subsequently resuspended in FC buffer.

Panel for B cell phenotyping included: Mouse α-human CD11c-BV421 (Clone BU15, 566877; BD), Mouse α-human CD3-BV510 (Clone OKT3, 566780; BD), Mouse α-human IgM-BV605 (Clone G20-127, 562977; BD), Mouse α-human CD21-BV711 (B-ly4, 563163; BD), Mouse α-human CD27-BV786 (Clone L128, 563327; BD), Mouse α-human CD16-FITC (Clone 3G8, 555406; BD), Mouse α-human CD8-PerCP/Cy5.5 (Clone SK1, 565310; BD), Mouse α-human CD19-PE (Clone HIB19, 555413; BD), Mouse α-human CD24-PE-CF594 (Clone ML5, 562405; BD), Mouse α-human TCRβ-PE-Cy5 (Clone IP26, 306710; BioLegend), Mouse α-human IgD-PE-Cy7 (Clone IA6-2, 561314; BD), Mouse α-human CD38-APC (Clone HIT2, 555462; BD), (Clone DX2, 561976; BD), Mouse α-human CD20-AF700 (Clone 2H7, 560631; BD), and Viability Dye eFluor 780 (65-0865-14; eBioscience). All antibodies were diluted 1/250, except for anti-IgM, which was diluted 1/125, and viability dye, which was diluted 1/2,000.

Panel for CSR analysis was: Mouse α-human CD11c-BV421 (Clone BU15, 566877; BD), Mouse α-human IgE-BV510 (Clone G7-26, 744316; BD), Mouse α-human IgM-BV605 (Clone G20-127, 562977; BD), Mouse α-human CD21-BV711 (B-ly4, 563163; BD), Mouse α-human CD27-BV786 (Clone L128, 563327; BD), F(ab’)₂-Goat α-human IgA-FITC (polyclonal, H14101; Invitrogen), Mouse α-human CD138-PerCP/Cy5.5 (Clone MI15, 564605; BD), Mouse α-human CD19-PE (Clone HIB19, 555413; BD), Mouse α-human CD24-PE-CF594 (Clone ML5, 562405; BD), Mouse α-human IgG-PE-Cy5 (Clone G18-145, 551497; BD), Mouse α-human IgD-PE-Cy7 (Clone IA6-2, 561314; BD), Mouse α-human CD38-APC (Clone HIT2, 555462; BD), (Clone DX2, 561976; BD), Mouse α-human CD20-AF700 (Clone 2H7, 560631; BD), and Viability Dye eFluor 780 (65-0865-14; eBioscience). All antibodies were diluted 1/250, except for anti-IgM, which was diluted 1/125, and viability dye, which was diluted 1/2,000. All antibodies were included in initial stain with viability dye. Following incubation and wash, the cells were fixed and permeabilized using eBioscience Foxp3/Transcription Factor Staining Buffer Set (00-5523-00; Thermo Fisher Scientific) according to the manufacturer’s instructions and restained for intracellular immunoglobulins (IgM, IgD, IgG, IgA, and IgE) before analysis. The stained cell solutions were analyzed on a NovoCyte Quanteon 4025 (Agilent) cytometer equipped with four lasers (405, 488, 561, and 640 nm) and 25 detectors. The acquired data were analyzed using FlowJo v. 10.10.0 (BD).

To quantify secretion of antibody isotypes, TRIFMAs were performed on culture supernatants. Wells of FluoroNunc MaxiSorp 96-well plates were coated with 100 μl capture antibody diluted in PBS, either 0.5 µg/ml goat α-human IgM-Hc (2023-01; Southern Biotech), 5 µg/ml mouse α-human IgG1-Fc (9054-01; Southern Biotech), 1 µg/ml mouse α-human IgG3-hinge (9210-01; Southern Biotech), or 2.5 µg/ml mouse α-human IgE-Fc (9240-01; Southern Biotech). Plates were incubated o/n at 4°C, then emptied and blocked with 200 μl TBS (137 mM NaCl, 2.7 mM KCl, and 25 mM Tris/Tris-HCl containing 0.09% [wol/vol] sodium azide [BP2471-500; Thermo Fisher Scientific]) with 1% BSA (A4503; Sigma-Aldrich) for 1 h at RT and washed three times with TBS containing 0.05% (vol/vol) Tween-20 (TBS/Tw). Supernatants were diluted 1/5 in TBS/Tw/0.1% BSA and added in duplicates to wells, then incubated 2 h at RT. Standard curves were prepared in TBS/Tw/0.1% BSA as serial dilutions of either human IgM (0158L-01; Southern Biotech) starting at 300 ng/ml and further eight times threefold dilution; human IgG1 (0151-K; Southern Biotech) starting at 450 ng/ml and further eight times threefold dilutions; human IgG3 (0153-L; Southern Biotech) starting at 50 ng/ml and further eight times threefold dilutions; or for human IgE, human donor plasma starting at threefold dilution and further eight times 2.5-fold dilutions. Following incubation, wells were washed three times with TBS/Tw before addition of detection antibody diluted in TBS/Tw, biotinylated goat anti-human IgG(H+L) (2016-08; Southern Biotech) 0.5 µg/ml for IgM detection, biotinylated goat anti-human IgG(H+L) (2016-08; Southern Biotech) 0.1 µg/ml for IgG1-and IgG3-detection, and biotinylated mouse anti-human IgE-Fc (9250-08; Southern Biotech) 0.2 µg/ml for IgE detection. After 2-h incubation, wells were emptied and washed three times in TBS/Tw, then incubated 30 min at RT with 100 μl/well Eu³⁺-labeled streptavidin (1244-360; PerkinElmer) diluted to 1 µg/ml in TBS/Tw containing 25 µM EDTA. Lastly, wells were washed three times with TBS/Tw and added 200 μl Enhancement solution (Ampliqon). Plates were shaken 5 min and measured on a time-resolved fluorometry plate reader (Victor X5 Perkin Elmer). Concentrations were derived by interpolation on standard curves fitted using 5-parameter fit in WorkOut 2.5, then normalized to cell counts and expressed as ng/ml (IgM, IgG1, IgG3) or milli-units (mU)/ml (IgE) per 100,000 cells.

Coding sequences of OSMR and OSMR c.1307T > A were PCR-amplified from an OSMR-encoding plasmid (RC216943; OriGene) using the primers 5′-CGG​GCC​TTT​CGA​CCT​CTA​GCG​GGA​TCC​GCC​GCG​CCG​CCA​CCC​GCG​ATC​GC-3′ and 5′-CCA​GAG​GTT​GAT​TAT​CGG​AAT​TCC​CTC​GAG​GGC​GGC​CGC​GTA​CGC​GTC​GGT​CCG​CTG​C-3′ or 5′-CGG​GCC​TTT​CGA​CCT​CTA​GCG​GGA​TCC​GCC​GCG​CCG​CCA​CCC​GCG​ATC​GC-3′, 5′-CCA​GAG​GTT​GAT​TAT​CGG​AAT​TCC​CTC​GAG​GGC​GGC​CGC​GTA​CGC​GTC​GGT​CCG​CTG​C-3′, 5′-CTC​ACA​ATT​CTC​CAG​TCA​TCA​GGG​GCC​TCT​G-3′ and 5′-CAG​AGG​CCC​CTG​ATG​ACT​GGA​GAA​TTG​TGA​G-3′. The PCR-fragments were inserted into BamHI/XhoI-digested pCCL/PGK-eGFP using NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs) according to the manufacturer’s protocol. The resulting plasmids were designated pCCL/PGK-OSMR and pCCL/PGK-OSMR c.1307T > A. Both inserts were verified by Sanger sequencing.

Third-generation lentiviral vectors were produced as previously described (89). Briefly, HEK293T cells were transfected with 11.25 µg pMD.2G, 9 μg pRSV-Rev, 39 μg pMDLg/pRRE, and 39 μg lentiviral transfer vector for T175 flasks with a surface area of 175 cm² using a standard polyethyleneimine transfection protocol. 6 h after transfection, the medium was changed. 2 days after transfection, viral supernatants were poured into tubes and kept in fridge, and cells received fresh medium to produce more virus. Three days after transfection viral supernatants from day 2 and 3 were pooled, filtrated and titrated by quantification of the number of lentiviral integrations in dermal fibroblasts as previously described (90). Dermal fibroblasts were transduced using MOI 10. Both lentiviral supernatants and cells culture medium were supplemented with 8 µg/ml hexadimethrine bromide (Polybrene, Merck). 1 day after transduction, pCCL/PGK-EGFP–treated cells were inspected for green fluorescence as a proxy for successful transduction before the cells were ready for experiments.

Fibroblasts were washed in PBS prior to detachment by TrypLE Express Enzyme (cat. no. 12563029; Gibco.). Once detached, cells were moved to a 96-well V-shaped plate and pelleted at 400 × g for 5 min. Supernatant was discarded, and cells were washed in 200 μl PBS and subsequently pelleted at 400 × g for 5 min. Supernatant was discarded, and each pellet was resuspended in 50 μl staining solution (0.1 μl live-dead marker, SYTOX Green Dead Cell Stain [#S34860; Invitrogen], 5 μl OSMR Monoclonal Antibody [AN-V2], APC, eBioscience [cat. no. 17-1303-42], and 44.9 μl 1% BSA in PBS) and incubated at RT for 30 min. Included control samples involved single stained cells (only one antibody), dead cells with and without viability stain, and unstained cells. Dead cells were prepared by using permeability buffer (#88-8824-00; Invitrogen) according to manufacturer. After incubation, 200 μl 1% BSA in PBS was added to the cells in staining solution and pelleted at 400 × g for 5 min. Samples/the cells were fixed in 100 μl 2% paraformaldehyde. After fixation, cells were washed with 1% BSA in PBS and resuspended in 100 μl 1% BSA in PBS and kept at 5°C in the dark prior to analysis on a NovoCyte Quanteon flow cytometer (Agilent). Data were analyzed using FlowJo v10.10 (BD biosciences).

PBMCs (5 × 10⁵) were seeded in wells of a 96-well plate. One day after seeding, cells were left untreated or treated for 24 h with 1 µg/ml LPS (from E. coli O26:B6, L2654-1MG; Millipore Sigma) and 50 ng/ml IFNγ (Recombinant Human IFN-γ, #300-02; PeproTech). After incubation, cells were washed with PBS, stained for 30 min with LIVE/DEAD Fixable Near-IR Dead Cell Stain (L10119; Invitrogen) in PBS at RT, washed with PBS, and FC blocked for 10 min using Human TruStain FcX (422302; BioLegend) in 1% BSA in PBS. Antibodies for staining of surface markers were directly added to the cells in blocking solution at a final volume of 60 μl. The following antibodies were used: anti-OSMR APC (1:30, clone AN-V2, 17-1303-42; eBioscience), anti-CD19 FITC (1:6, clone HIB19, 555412; BD Biosciences), anti-CD3 Spark Violet 538 (1:30, clone UCHT1, 300484; BioLegend), anti-CD14 Pacific Blue (1:75, clone M5E2, 301815; BioLegend), and anti-CD56 BV786 (1:30, clone RPA-T8, 563823; BD Biosciences) in PBS with 1% BSA. After 30 min of incubation at RT, the cells were washed twice with 1% BSA in PBS, fixed using 2% paraformaldehyde (Sigma-Aldrich), and washed with 1% BSA in PBS. The cells were resuspended in 100 μl 1% BSA in PBS and kept at 5°C in the dark prior to analysis on a Novocyte Quanteon flow cytometer (Agilent). Data were analyzed using FlowJo v10.10 (BD Biosciences).

The differences between experimental conditions of groups were analyzed using GraphPad Prism (Version 10) with the type of test indicated in each figure legend. P values ≤0.05 were considered significant: *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. S1 shows patient blood IgE values and HIES score. Fig. S2 shows gating strategies and flow analysis of T cell subsets of Th17, Th9, Th2, and pTFH cell frequencies. Fig. S3 shows Th17 and Th2 subpopulations produce normal levels of IL-17A and IL-4. Fig. S4 shows examination of OSMR surface expression on T and B cells and OSMR variants (>20 CADD score) after deglycosylation. Fig. S5 shows gating strategies for B cell subsets and CSR analyses. Table S1 shows blood values of immunoglobulins, IgG subclasses, hematology, and Th cell subsets from the patient compared to normal ranges from healthy controls. Table S2 shows identification of two monoallelic variants in genes involved in STAT3 signaling.

All relevant data that can be shared can be found in supplemental material. Additional data are available from the corresponding author upon reasonable request from health care professionals and within national General Data Protection Regulation rules and regulations.

We wish to thank the patient for participating in the study. We also wish to thank Bettina Bundgaard and Lisbeth Jensen for excellent technical assistance. We are grateful to the FACS Core at Aarhus University for expert assistance with flow cytometry experiments.

T.H. Mogensen is supported by the Independent Research Fund Denmark (0134-00006B), the Novo Nordisk Foundation (NNF21OC0067157; NNF20OC0064890), the Lundbeck Foundation (R268-2016-3927), The Innovation Fund Denmark (PASCAL-MID, 8056-00010B), the European Research Council, HORIZON-HLTH-2021-DISEASE-04 program under grant agreement 101057100 (UNDINE), Marie Curie fellowship (DIVIR, 101065126), and the Danish National Research Foundation under the grant agreement DNRF164 (CiViA). S. Andersen received a grant from Frimodt-Heineke Fonden, Læge Sofus Carl Emil Friis og Hustru Olga Doris Friis’ Fond and Kong Christian den Tiendes Fond. S.E. Degn received funding from the Independent Research Fund Denmark (Sapere Aude Research Leader grant; 9060-00038) and Aase og Ejnar Danielsens Fond.

Author contributions: Sisse Andersen: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, supervision, validation, visualization, and writing—original draft, review, and editing. Kristian Assing: conceptualization, formal analysis, investigation, methodology, resources, supervision, validation, visualization, and writing—original draft, review, and editing. Julie Jensen: data curation, formal analysis, investigation, methodology, visualization, and writing—review and editing. Line Dahlerup Rasmussen: conceptualization, resources, and writing—review and editing. Christian B. Laursen: conceptualization, data curation, investigation, validation, visualization, and writing—review and editing. Christoffer D. Dellgren: investigation. Daniëla Maria Hinke: formal analysis, investigation, methodology, and visualization. Søren E. Degn: formal analysis, investigation, methodology, resources, and writing—review and editing. Trine H. Mogensen: conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, and writing—original draft, review, and editing.