Translational approaches to the study of eosinophils in vasculitis

Abstract

Eosinophil basic biology, including immunoregulatory functions, plays a complex role in a myriad of disorders where eosinophils play a role. Established in vivo and in vitro models and novel emerging techniques for studying eosinophils are integral to parsing the involvement of eosinophils in the pathogenesis of disorders including vasculitis. Knowledge of translational approaches from eosinophil-associated diseases and their application to eosinophilic vasculitis are highlighted. It is becoming increasingly evident that the roles of eosinophils in disease are complex. The use of targeted biological therapies and machine learning has the potential to reveal the contribution of eosinophils in disease pathology.

Rheumatology key messages.

• Eosinophils are key regulators of the immune system, including metabolic, immunological, haematologic homeostasis and inflammatory responses

• Cytokine IL-5 supports eosinophil number expansion, maturation, survival and activation

• Biomarkers, including granular proteins, afford opportunities to understand eosinophil biology, pathogenesis and response to therapy

Basic biology of eosinophils

Eosinophils are bone-marrow-derived granulocytes [1], normally comprising a small proportion of circulating leukocytes in vertebrates [2, 3], but are predominantly located in tissues under normal physiological conditions [4]. Cell surface receptor expression permits rapid response to environmental triggers [2]. They have been described as having homeostatic roles, for example in remodelling and fibrosis in addition to their known inflammatory effects [2, 3]. Eosinophil granules possess granule proteins such as major basic protein (MBP), eosinophil peroxidase (EPX), eosinophil-derived neurotoxin (EDN) and eosinophil cationic protein (ECP) [3], along with other pro-inflammatory mediators [1] that are cytotoxic [3, 5], and that are involved in activating inflammatory cells [2]. Characteristic mediators such as Charcot-Leyden crystal proteins [6], and other growth factors, cytokines and lipid factors are involved in their effector function. Eosinophil numbers increase during specific infectious responses (e.g. fungal or helminth infections), but their involvement in respiratory and vascular disorders was integral to the development of anti-eosinophilic biologics. The primary axis that has been utilized is that of blocking IL-5 both as a cytokine as well as ligating the IL-5 receptor and promoting antibody-medicated cell-cytotoxicity. IL-5 is a key cytokine for eosinophil maturation, activation and survival [7].

Though the full life cycle of eosinophils is still under study, generally eosinophils are described as being in inactive, pre-excited (primed) or fully activated states [1]. Specific surface markers define these states. For example, low surface expression of CD69 and integrins defines a non-activated eosinophil [1]. Cytokines such as IL-3, IL-5 and GM-CSF are involved in eosinophil priming, increasing their susceptibility to further stimulation by inflammatory mediators such as IL-13, C-C motif chemokine ligands and histamine. Primed eosinophils, often expressing CD32 and FcγRII, have been described in peripheral blood of asthma patients as well as in segmental lung challenge [1]. Tissue transmigration of activated eosinophils results in secretion of preformed granule contents, further propagating inflammatory responses in the tissue [2]. Because eosinophils have been implicated in host defences, tissue homeostasis and immunoregulation, [1] questions have arisen regarding the implications of eosinophil depletion in diverse eosinophilic disorders as well as the role of eosinophil states in response to biologic administration. To date, limited information has been reported on this topic but is an important step to understanding the roles of eosinophils in EADs.

In this review, the basic biology, translational approaches and gaps in knowledge as pertain to eosinophilic granulomatosis with polyangiitis (EGPA), hypereosinophilic syndrome (HES) and eosinophilic gastrointestinal diseases (EGIDs) and eosinophilic asthma will be discussed.

Actions of eosinophils

Eosinophils secrete their granule contents through one of several secretory processes—classic exocytosis, piecemeal degranulation (PMD) or through full cytolysis [3, 4]. Selective protein secretion through a complex vesiculotubular system has been described [4, 8]. The differential release of granule contents through PMD has been studied using exosomes [3], and the role of eosinophil-derived extracellular vesicles has been purported to have roles on eosinophils themselves [9]. Furthermore, eosinophils extracellular traps (EETs) formation (EETosis) whereby granule-coated eosinophil DNA nets are extruded from eosinophils, has been implicated in specific disorders. Following dissolution of cellular membranes, NET-associated intact cell-free granules may be released within tissues [4, 8]. The loss of cellular galectin-10 and granules in the context of EGPA has been associated with measures of disease activity such as the Birmingham Vasculitis Activity Scale (BVAS) [2, 10].

Cell-free DNA (cfDNA) derived from nuclei or mitochondria and EET formation is of interest in small vessel vasculitis. In a study of 35 participants with AAV (EGPA n = 10; MPA n = 13; GPA n−12), higher serum cf-nuclear and cf-mitochondrial DNA was found in AAV compared with controls, with the highest numbers in EGPA [11]. Furthermore, blood eosinophil count and plasma D-dimer levels correlated with cf-nDNA and cf-mDNA in EGPA [11]. Eosinophilia has long been implicated in thrombosis in EADs through unknown mechanisms. The finding that EETs are demonstrated in small-vessel thrombi [3] may be related to corresponding in vitro studies showing that EETs serve as a scaffold for platelet adhesion due to the net-like chromatin threads. Eosinophils may also promote thrombosis intravascularly through other mechanisms including P-selectin platelet activation and binding or through promotion of tissue factor activation by eosinophil products [3]. Taken together, this elevation of cfDNA in EGPA as well as its association with disease activity, suggest the potential for its further study as a biomarker [11] of disease activity, as well as the potential role of EETs in the pathogenesis of EGPA complications such as thrombosis.

Another measure of eosinophil activation, the concentrations of eosinophil granule proteins (EGPs) in tissue and bodily fluids, has been correlated with eosinophilic disease activity [12]. In one study, EGID participants with active disease but normal peripheral eosinophil counts had increased concentrations of MBP, ECP, EDN and EPO compared with normal donors [12]. In another study of diverse EADs (including EGPA patients), serum, plasma and urine EGP concentrations were measured to assess disease activity in patients treated with anti-IL-5/5R therapeutics [13]. Plasma (p) EDN, but not urine(u) EDN correlated with absolute eosinophil count (AEC) and negatively with prednisolone dose. Both (p) and (u) EDN decreased in benralizumab treated HES participants and EGPA participants treated with mepolizumab. In a cross-sectional analysis, uEDN increased with clinical relapse in both groups, and would be an interesting non-invasive biomarker for study in future treatment studies for EADs [13].

The toolbox for the study of eosinophils

Since eosinophils are notoriously difficult to study ex vivo, and EADs are rare, both in vivo animal models as well as ex vivo approaches have paved the way for further understanding of the roles of eosinophils in health and disease despite known limitations of these approaches [3]. Complete IL-5 and IL-5Rα knock-out strains have differences in the level of eosinophils in both peripheral blood and bone marrow (BM) in addition to differences in eosinophil responses to varied antigenic challenges (e.g. allergen or parasites) that are well described in the review by Jacobsen et al. [3]. The development of eosinophil deficient mouse models, dblGATA-1 and PHIL, as well as mice deficient in granulogenesis, e.g. MBP-1^−/−EPX^−/−double knock-out mouse, ushered in many elegant studies that defined differences in eosinophil functions [3]. Additional knock-in models such as iPHIL and eoCre further facilitated studies of eosinophil development and function in otherwise healthy strains without effects on early development, breeding or lifespan [3].

The morphology of eosinophils has been studied to shed light on their roles in EADs. It is well known that activated eosinophils undergo ultrastructural changes. Imaging using transmission electron microscopy has informed a lot of what we know about eosinophils. in conjunction with molecular imaging (immunoEM) as well as 3D tomography have been employed to understand ultrastructural changes after activation [14].

Because eosinophils are associated with aberrant nerve structure and airway nerve dysregulation is implicated in cough and excessive bronchoconstriction, Drake et al. [15] tested whether eosinophils themselves alter airway structure. In humans, airway innervation and substance P expression were increased in patients with moderate-persistent asthma compared with those with mild asthma or healthy volunteers. Increased innervation was associated with functional implications such as lack of bronchodilator responsiveness and increased sensitivity to irritants. In a mouse model of airway responsiveness, eosinophils mediated both increased nerve density and airway responsiveness [15].

Techniques that are being used in other disease states, but which have not as yet been applied to EADs, include the generation of bone marrow (BM0) organoids to better understand haematopoiesis requires accurate models that incorporate both haematopoietic and stromal elements [16]. Olijnik et al. [16] describe a method generating 3D, multilineage BM organoids from human-induced pluripotent stem cells (hiPSCs). Visualizing haematopoietic cell populations in three dimensions in relation to differentiated hiPSCs would allow further study of eosinophil development in the marrow microenvironment [16] in diverse EADs ranging from eosinophilic asthma to HESs.

Another interesting example of tissue-specific investigation, spatial transcriptomics, facilitates assessment of transcripts in the context of tissue architecture [17]. Jiang et al. [17] developed a multi-stage statistical method called ilMPACT that uses artificial intelligence (AI)-reconstructed histology images to integrate the spatial context of gene expression measurement and differentially expressed genes. Understanding the cellular spatial organization of relevant genes within spatial transcriptomics data can better represent pathogenesis of eosinophil-associated disorders [17]. Methods such as ilMPACT and other similar machine learning techniques allow investigation beyond the known molecular phenotypes and endotypes that have been described, e.g. as in asthma. For rare diseases, this is especially important as was described in Taroni et al. [18], where transfer learning was used to extract coordinated expression patterns for rare disease datasets. They trained a ‘Pathway Level Information ExtractoR’ (PLIER) model on large public data compendium comprising multiple experiments, tissues and biologic conditions, then transferred the model to small datasets in an approach called MultiPLIER [18]. Similar to foundational models’ use of this approach and incorporation of rare disease datasets, integrated gene expression and multiomics with machine learning techniques can reveal regulatory processes that might be implicated in disease. Machine learning approaches assessing whole slide images of sural nerve morphology developed by Ono et al. [19] have the potential to further expand our understanding of vasculitis in tissues.

Beyond vasculitis: commonalities with other eosinophilic disorders

In atopic conditions with prominent eosinophilic involvement such as asthma or EGIDs, T2 inflammatory profiles are evident [20]. In eosinophilic esophagitis (EoE), basal zone hyperplasia is a hallmark linked to eosinophil and mast cell infiltration, which is likely driven by the epithelium’s transcriptional response to IL-13, supported by murine and human anti-IL-13 studies [21]. Studies by Dunn et al. [22] demonstrated bidirectional communication between eosinophils and oesophageal epithelial cells, including cell proliferation, direct epithelial responses to cytokines as well as elaboration of alarmins that contribute to structural changes and remodelling. Machine learning techniques, as employed by Zhang et al. [23] have revealed spatially distinct mast cell populations in EoE, highlighting the dynamic cellular interplay characteristic of T2 inflammation. Rochman et al.’s [21] proteomic analysis identified 402 differentially expressed proteins (DEPs) in EoE that correspond to immune and epithelial cell responses, showing upregulation in immune-related DEPs and downregulation in epithelial differentiation. Many of these pathways also exist in asthma in the absence of EGPA. The eosinophilic asthma phenotype is associated with more frequent exacerbations, later onset and worse lung function with increased levels of eosinophils in bronchial biopsies and/or sputum [24]. The shared inflammatory mechanisms yet wildly different clinical manifestations in asthma, EGIDs and vasculitis remain a puzzle. The potential therapeutic approaches in all these disorders that include endothelial and epithelial dysfunction are shared, but why inflammation is systemic or localized to the vasculature is not known.

Biologics as a tool to decipher mechanisms in clinical trials

Assessments of biologic treatment in eosinophilic disorders and asthma could provide valuable insight to heterogeneous causal pathways and provide signals for determining predictors of response and/or relapse. Mepolizumab (see Fig. 1) was studied in relapsing and refractory EGPA in a phase 3 double-blinded placebo-controlled trial (MIRRA) [25]. Mepolizumab resulted in a greater number of weeks in remission as well as reduced glucocorticoid usage [25]. In 2024, the MANDARA trial reported non-inferiority of benralizumab to mepolizumab in EGPA leading to its approval by the FDA in the USA [26]. EGPA real-world data analysis supports reduction in steroid dosage with anti-IL5 therapy [27, 28], and promising outcomes with sequential anti-IL5 therapy after rituximab [29]. Severe eosinophilic asthma has proved to be a steering therapeutic pathway for EGPA. Mepolizumab, benralizumab, relizumab, omalizumab and dupilumab biologics are approved in severe eosinophilic asthma [30]. In a systematic review, all showed reduction in exacerbation rate of asthma as add-on treatment, with reduction in daily glucocorticoid dose associated with mepolizumab, benralizumab and dupilumab compared with standard of care [30]. In the Phase 2 DREAM study, intravenous treatment with mepolizumab (4 weekly) reduced sputum and blood eosinophil counts by 50 and 75%, respectively [24]. Furthermore, 12 weeks of treatment with 100 mg subcutaneously four weekly reduced blood eosinophil counts by 86%, which was maintained throughout the 32-week MENSA trial [24]. Recently, in the Acute exacerbations treated with BenRAlizumab trial, eosinophilic exacerbations of asthma and COPD treated with BRZ, or BRZ and steroids, had favourable outcomes at 90 days in comparison to the steroid group; treatment failures occurred in 74% of PRED group, vs 45% in the pooled BRZ group [31]. Further clinical trials with BRZ including SIROCCO and exploratory analysis of the phase 3 CALIMA were associated with reduction in blood eosinophil counts to a median of 0 cells/µl. BRZ was also associated with lower annual exacerbation rates vs placebo in these trials and in MELTEMI (a 5-year open-label extension study of IROCCO, CALIMA and ZONDA) [24].

Figure 1.

Biological therapies in eosinophilic disorders. The cytokine milieu that promotes eosinophil proliferation and/or activation includes IL-5, IL-4, IL-13 and thymic stromal lymphopoietin (TSLP). These cytokines or their receptor are therapeutic targets; Mepolizumab inhibiting IL-5, Benralizumab inhibiting IL-5 receptor, Reslizumab inhibiting IL-5, Dupilumab inhibiting IL-4, Cendakimab and Dectrekumab inhibiting IL-13 and Tezelumab inhibiting TSLP. Collection of samples and data during clinical trials may lead to biomarkers, including multimodal omics data for use in machine learning and artificial intelligence applications for precision therapeutics. Figure created in BioRender. Makiya, M. (2024) BioRender.com/e64g553

HES comprises a group of diseases defined by blood or tissue eosinophilia, with eosinophil-related clinical manifestations [32]. Benralizumab, a monoclonal antibody against anti-IL5 receptor alpha (IL-5Rα) triggers antibody-dependent cellular cytotoxicity and depletes eosinophils. Benralizumab was efficacious in the majority of 20 HES participants enrolled in a phase 2 randomized placebo-controlled study [32]. AECs were reduced, and in an open-label extension, clinical and haematological responses were sustained at 48 weeks in a majority [32]. A key finding was that clinical disease subtype was associated with clinical response, suggesting the mechanism driving eosinophilia may differ in the subtypes of HES. Relapse was associated with lymphoid disease, high levels of soluble IL5Rα and the development of antidrug antibodies. Mepolizumab has shown corticosteroid-sparing benefits for FIPILI-PDGFRA negative HES allowing reduction to 10 mg or less of prednisolone per day for eight or more consecutive weeks [33].

Taking from the example of EoE, the lack of consistent efficacy of eosinophil-directed biologics has caused the research community to reconsider the assumptions regarding the primary role of eosinophils [34]. IL-5/5R axis monoclonals lowered eosinophil numbers in oesophageal tissues with active EoE, but did not significantly improve symptoms of EoE [34]. The IL-4Rα monoclonal dupilumab was effective in EoE and was approved by the FDA in the USA, ushering in interest in other therapies, such as IL-13 inhibitors (cendakimab/dectrekumab), and the anti-thymic stromal lymphopoietin monoclonal antibody tezepelumab to more broadly target diverse inflammatory T2 pathways beyond eosinophils [34]. Given the similarities between these eosinophilic/atopic disorders, one cannot help but speculate about the additional role of T2 pathways in EGPA that are not specifically targeted using the currently approved eosinophil-targeted therapies. Projecting forward, future efforts using systems biology and AI approaches in the context of clinical trials could better explore why certain subsets of EGPA patients have suboptimal responses to eosinophil-targeted biologics in vasculitis. Beyond anti-IL5/5R and IL4/13 agents, the number of emerging therapeutics such as JAK inhibitors, S1PR agonists and other investigational agents to treat EADs are beyond the scope of this review.

Finally, efficient biomarker utilization in trial design maximizes the potential for insights into disease mechanisms especially in rare diseases where patient enrolment is scarce [35]. Assessing biomarkers or employing the use of machine learning approaches to capture clinical benefit and identification of predictors of responder phenotypes would permit application of personalized treatment approaches in rare EADs where heterogeneous disease pathophysiology, clinical manifestations and variability in responses occur [35]. Understanding the cell biology and role of eosinophils, particularly in their varied activation states, is of interest to better understand the role of eosinophils in vasculitides.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

This work was supported in part by the Division of Intramural Research of the National Institutes of Allergic and Infectious Diseases, NIH.

Disclosure statement: A.E. is an advisor to AstraZeneca.