Eosinophils were first described in 1879 by Paul Ehrlich, who noted their unusual capacity to be stained by acidophilic dyes. Interestingly, our appreciation of this unique property of eosinophils is clear and steadfast, but a comprehensive understanding of the function of these cells in health and disease remains elusive. Some basic characteristics of eosinophils are established and accepted. It is clear that eosinophils are granulocytes that develop in the bone marrow from pluripotent progenitors. They are released into the peripheral blood in a phenotypically mature state, and they are capable of being activated and recruited into tissues in response to appropriate stimuli, most notably the cytokine interleukin-5 (IL-5) and the eotaxin chemokines. Eosinophils spend only a brief time in the peripheral blood (they have a half-life of 18 hrs)1 before they migrate to the thymus or gastrointestinal tract, where they reside under homeostatic conditions2. In response to inflammatory stimuli, eosinophils develop from committed bone marrow progenitors, after which they exit the bone marrow, migrate into the blood and subsequently accumulate in peripheral tissues, where their survival is prolonged (reviewed in REFS 3, 4, 5).

However, much remains unclear. For example, the long-held belief that eosinophils promote immunity to helminths has been called into question by results from animal studies, some of which suggest that eosinophils may be serving to promote the needs and longevity of specific parasites6,7. Likewise, eosinophils are clearly recruited to and activated in lung tissue as part of the pathophysiology of asthma, and most current evidence suggests that eosinophils contribute to airway dysfunction and tissue remodelling in this disorder8,9. Evolution tells us that the ability to induce pathology cannot be a 'raison d'être' for any existing cell lineage, and recent findings on the antimicrobial and antiviral activities of eosinophils suggest that the pathology that arises from dysregulated eosinophilia in the airways may be collateral damage related to host defence. Similarly, although there are now two unique eosinophil-deficient mouse strains10,11, there are no known unique eosinophil-deficiency states in humans to help us to decipher the importance of these cells in vivo.

This Review examines the most recent advances in our understanding of the contributions of eosinophils to the maintenance of health, and how dysregulated eosinophil function promotes various disease states. These advances were made possible by reagents, systems and methods that target eosinophil function and by the first clinical trials using humanized monoclonal antibodies specific for IL-5 (Table 1). These tools have been invaluable for shaping our current views on eosinophil function and for generating new hypotheses for future examination.

Table 1 Tools for the eosinophil biologist

The unique biology of the eosinophil

Relatively few mature eosinophils are found in the peripheral blood of healthy humans (less than 400 per mm3), but these cells can be readily distinguished from the more prevalent neutrophils by virtue of their bilobed nuclei and large specific granules (Fig. 1). Human eosinophil granules contain four major proteins: eosinophil peroxidase, major basic protein (MBP) and the ribonucleases eosinophil cationic protein (ECP) and eosinophil-derived neurotoxin (EDN). The granules also store numerous cytokines, enzymes and growth factors. Other prominent features of eosinophils include primary granules that contain Charcot–Leyden crystal protein (also known as galectin 10 and eosinophil lysophospholipase) and lipid bodies, which are the sites of synthesis of cysteinyl leukotrienes, thromboxane and prostaglandins.

Figure 1: The eosinophil.
figure 1

a | Human eosinophils from peripheral blood stained with modified Giemsa exhibit characteristic bilobed nuclei and large red-stained cytoplasmic secretory granules. The cells with multilobed nuclei and without large granules are neutrophils. Original magnification, ×100. b | The image shows eosinophils and neutrophils isolated from the spleen of a Cd2-interleukin-5-transgenic mouse and stained with modified Giemsa. c | The image shows a transmission electron micrograph of a mouse eosinophil. Cytoplasmic secretory granules are indicated by the arrows; the central core of these granules contains cationic major basic protein, and their periphery contains the remaining major cationic proteins, cytokines, chemokines, growth factors and enzymes. Original magnification, ×6,000.

Eosinophils have been identified and characterized in all vertebrate species, but their morphology, repertoire of cell-surface receptors and intracellular contents vary significantly between species. Of particular note, there are several crucial differences between mouse and human eosinophils that must be taken into account when interpreting mouse model studies of human disease12 (Fig. 1).

Eosinophils express surface receptors for ligands that support growth, adhesion, chemotaxis, degranulation and cell-to-cell interactions (Fig. 2). Many of the signalling pathways involved in these responses have been detailed in recent reviews3,4,13. Among the main receptors that define the unique biology of the eosinophil are interleukin-5 receptor subunit-α (IL-5Rα) and CC-chemokine receptor 3 (CCR3), as well as sialic acid-binding immunoglobulin-like lectin 8 (SIGLEC-8) in humans and SIGLEC-F (also known as SIGLEC-5) in mice. Pattern-recognition receptors (PRRs) are also likely to be important for eosinophil function, a subject that remains to be fully explored (Box 1).

Figure 2: Cellular features of eosinophils.
figure 2

Eosinophils are equipped with features that promote interactions with the environment. In one such interaction, eosinophils release the contents of their specific granules in response to external stimuli. Some of these granule contents are released via membrane-bound vesicles known as eosinophil sombrero vesicles. Eosinophils also synthesize lipid mediators for release in cytoplasmic lipid bodies and store Charcot–Leyden crystal protein (CLC) in primary granules. Although not highly biosynthetic, mature eosinophils have minimal numbers of mitochondria and a limited endoplasmic reticulum (ER) and Golgi, as well as a nucleus. Eosinophils express a wide variety of receptors that modulate adhesion, growth, survival, activation, migration and pattern recognition. Mouse eosinophils do not express CLC or Fcε receptor 1 (FcεR1) and have divergent homologues of sialic acid-binding immunoglobulin-like lectin 8 (SIGLEC-8) and the granule ribonucleases eosinophil-derived neurotoxin (EDN) and eosinophil cationic protein (ECP)12. APRIL, a proliferation-inducing ligand; CCL, CC-chemokine ligand; CCR, CC-chemokine receptor; CXCL, CXC-chemokine ligand; CXCR, CXC-chemokine receptor; EGF, epidermal growth factor; EPX, eosinophil peroxidase; FPR1, formyl peptide receptor 1; GM-CSF, granulocyte–macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; MBP, major basic protein; NGF, nerve growth factor; NOD, nucleotide-binding oligomerization domain protein; PAR, proteinase-activated receptor; PDGF, platelet-derived growth factor; PIRB, paired immunoglobulin-like receptor B; PPARγ, peroxisome proliferator-activated receptor-γ; PRR, pattern-recognition receptor; PSGL1, P-selectin glycoprotein ligand 1; RAGE, receptor for advanced glycation end-products; RIG-I, retinoic acid-inducible gene I; TGF, transforming growth factor; TLR, Toll-like receptor; TNF, tumour necrosis factor; SCF, stem cell factor; VEGF, vascular endothelial growth factor.

Factors that promote eosinophilia

IL-5 has a central and profound role in all aspects of eosinophil development, activation and survival (Box 2). Likewise, CC-chemokine ligand 11 (CCL11; also known as eotaxin), which is a ligand for CCR3, promotes eosinophilia both cooperatively with IL-5 and via IL-5-independent mechanisms14,15. Recently, several new factors that promote eosinophilic inflammation in vivo have been identified.

The epithelial cell-derived cytokines thymic stromal lymphopoietin (TSLP), IL-25 (also known as IL-17E) and IL-33 promote eosinophilia by inducing IL-5 production. TSLP is an IL-2 family cytokine that signals through a heterodimeric receptor that comprises the IL-7 receptor α-chain and a specific TSLP receptor β-chain. The TSLP receptor is expressed widely, by myeloid dendritic cells (DCs), CD4+ and CD8+ T cells, B cells, mast cells and airway epithelial cells. The TSLP receptor is also expressed by human eosinophils and modulates their survival and activation16.

IL-25 is produced primarily by activated T helper 2 (TH2) cells and mast cells and induces the production of TH2-type cytokines (including IL-5) from TH2 cells, as well as from the newly described populations of mouse innate lymphoid cells, which include nuocytes and natural helper cells17,18,19. In this manner, IL-25 can amplify the development, recruitment and survival of eosinophils in allergic states. Abundant expression of both IL-25 and the IL-25 receptor was also detected in a recent study of bronchial and skin biopsies from allergic human subjects20, and eosinophils themselves were identified as the primary source of IL-25 in patients with severe systemic vasculitis (Churg–Strauss syndrome)21.

IL-33 — which is a member of the IL-1 cytokine family — is expressed by epithelial cells, endothelial cells, fibroblasts and adipocytes, and is an endogenous danger signal known as an alarmin. IL-33 specifically modulates TH2-type pro-inflammatory signals following its release from necrotic cells. The IL-33 receptor ST2 (also known as IL-1RL1) is found primarily on TH2 cells, but IL-33-dependent responses from mouse nuocytes, natural helper cells and innate type 2 helper cells, and in human eosinophils themselves, have been described22,23,24. Furthermore, an IL-33- and IL-25-responsive innate lymphoid cell population has recently been defined in humans25. Although the biology of this cytokine has not been fully elucidated, IL-33 typically contributes to the synthesis and release of IL-5 from one or more of the aforementioned target cells, and thereby promotes systemic eosinophilia.

IL-23 is a member of the IL-12 family of cytokines that promotes the function of TH17 cells and also regulates allergic airway inflammation. Silencing the expression of IL-23 in mice that were sensitized and challenged with ovalbumin resulted in decreased recruitment of eosinophils to the lung tissue in association with diminished levels of IL-17 and IL-4 (Ref. 26). Accordingly, the overexpression of IL-23 was shown to augment antigen-stimulated eosinophil recruitment27. However, another study found that IL-23 suppressed eosinophilia in a mouse model of fungal infection, a response that was IL-17 independent28.

High-mobility group protein B1 (HMGB1) is another example of an alarmin that promotes eosinophilia. However, in contrast to IL-33, there is no evidence that eosinophil activation in response to HMGB1 involves IL-5. First identified as a nuclear protein and transcription factor, HMGB1 is expressed ubiquitously and mediates inflammatory responses via its receptors. The HMGB1 receptors that have been identified so far are receptor for advanced glycation end-products (RAGE), Toll-like receptor 2 (TLR2) and TLR4. Importantly, eosinophil mobilization and activation were observed in response to HMGB1 in tumour cell lysates29. Further work is needed in this area, as a better appreciation of the way in which eosinophils are activated in response to HMGB1 and other, related damage-associated molecular patterns (DAMPs) may explain the eosinophil recruitment that is observed in the setting of tissue destruction associated with myalgias and myopathies.

Eosinophil degranulation

Degranulation — that is, the release of granule contents into the extracellular space — is a prominent eosinophil function. Previously, the release of secretory mediators was assumed to take place primarily through cytolytic degranulation, a mechanism through which a pathogenic assault (real or perceived) results in the complete emptying of the eosinophil's arsenal of preformed cationic proteins. Interestingly, a careful analysis of electron micrographs of eosinophils degranulating in tissues suggested a more controlled process, which was given the name 'piecemeal degranulation'30 to reflect the fact that the eosinophil was able to release bits or pieces of its granule contents in response to a given stimulus, while remaining otherwise intact and apparently viable. Piecemeal degranulation is now accepted as the most commonly observed physiological form of eosinophil degranulation. Eosinophils undergoing piecemeal degranulation in response to cytokines, such as interferon-γ (IFNγ) and CCL11, develop cytoplasmic secretory vesicles, known as eosinophil sombrero vesicles31 (Fig. 2), and remain viable and fully responsive to subsequent stimuli.

A recent study has provided substantial insights into the molecular mechanism of piecemeal degranulation32. Specifically, IL-4 released from CCL11-activated eosinophils first forms a complex with IL-4 receptor subunit-α (IL-4Rα) within the granule membrane, and IL-4Rα thereby chaperones IL-4 to the membrane vesicles before its release from the cell. Although receptor-mediated trafficking pathways have not yet been defined for other eosinophil mediators33, this study provides an insight into the potential for exquisite molecular modulation of piecemeal degranulation32.

Eosinophils also release intact granules, which are capable of storing and releasing mediators in response to physiological secretagogues in the cell-free state34. Cell-free granules have been identified in tissues in association with eosinophil-associated disorders35, although their functional significance and their ability to respond to activating stimuli in situ await further evaluation.

Interactions of eosinophils with other leukocytes

During their transit from the bloodstream to the tissue, eosinophils use selectins and integrins to interact with endothelial cells, and they interact with epithelial cells at mucosal surfaces in a similar manner; these subjects have been reviewed extensively36. Eosinophils also interact with and modulate the functions of other leukocytes (Fig. 3).

Figure 3: Eosinophils modulate the function of other leukocytes.
figure 3

Eosinophils not only respond to signals, but also have a definitive impact on the actions of other leukocytes. Eosinophils can express MHC class II and co-stimulatory molecules, process antigens and stimulate T cells to proliferate and produce cytokines in an antigen-specific manner39. Furthermore, acting together with dendritic cells (DCs), eosinophils regulate the recruitment of T helper 2 (TH2) cells in response to allergen sensitization and challenge by producing CC-chemokine ligand 17 (CCL17) and CCL22 (Refs 40, 41). Eosinophils also prime B cells for antigen-specific IgM production39 and sustain long-lived plasma cells in mouse bone marrow via the production of a proliferation-inducing ligand (APRIL) and interleukin-6 (IL-6)44,45. Eosinophils that are stimulated by CpG DNA induce DC maturation51. Indeed, the eosinophil granule protein eosinophil-derived neurotoxin (EDN) promotes the maturation and activation of DCs52,53. Major basic protein (MBP) released from eosinophils activates neutrophils, causing them to release superoxide and IL-8 and increase their expression of the cell-surface integrin complement receptor 3 (CR3)146. Eosinophils also maintain alternatively activated macrophages in adipose tissue by producing IL-4 and IL-13 (Ref. 50). The eosinophil granule proteins MBP, eosinophil cationic protein (ECP) and eosinophil peroxidase (EPX) activate mast cells, resulting in the release of histamine. Likewise, eosinophil-derived nerve growth factor (NGF) prolongs mast cell survival147.

Interaction with lymphocytes. Eosinophils clearly respond to signals (such as IL-5) that are provided by T cells. Two recent studies indicate that T cells also respond to signals provided by eosinophils37,38. Although not 'professional' antigen-presenting cells, eosinophils can express cell-surface components that are required for antigen presentation (such as MHC class II molecules and the co-stimulatory molecules CD80 and CD86). Indeed, eosinophils can process antigens and stimulate T cells in an antigen-specific manner, resulting in T cell proliferation and cytokine release39. Furthermore, in experiments performed in both wild-type mice and transgenic mice that lack eosinophils (TgPHIL mice), eosinophils can augment allergic inflammation by regulating the production of TH2-type chemoattractants (including CCL17 and CCL22), which promote the recruitment of TH2 cells, and also through their interactions with DCs40,41. In addition, eosinophils release preformed cytokines (such as IL-4, IL-13 and IFNγ) that promote either TH2 or TH1 cell responses42.

Eosinophils also promote humoral immune responses. Indeed, they are capable of priming B cells for the production of antigen-specific IgM43. Most recently, the production of a proliferation-inducing ligand (APRIL) and IL-6 by eosinophils was shown to be crucial for the support of long-lived plasma cells in mouse bone marrow44. Interestingly, activated eosinophils from the bone marrow of adjuvant-immunized mice were found to be even more effective at supporting plasma cell survival than those from adjuvant-naive mice45.

Interactions with innate immune cells.Alternatively activated macrophages have a pivotal role in recruiting eosinophils to the tissues46,47 through the release of YM1 (also known as CHI3L3), a chitinase-like selective eosinophil chemoattractant48,49. Eosinophils likewise recruit alternatively activated macrophages to, and maintain their viability in, adipose tissue50, promote the maturation of monocyte-derived DCs in vitro51, and are required for the accumulation of myeloid DCs and the systemic production of TH2-type cytokines in mice with allergic airway disease. The eosinophil secretory mediator EDN promotes the activation and migration of DCs52,53.

Eosinophils communicate extensively with tissue-resident mast cells. Eosinophils and mast cells are found in close proximity to one another under homeostatic conditions in the gut, and they also colocalize in the allergic lung and in the inflamed gut in patients with Crohn's disease54. The bidirectional signalling that occurs between eosinophils and mast cells involves several immunomodulatory mediators. These include stem cell factor (also known as KIT ligand), granule proteins, cytokines (such as granulocyte–macrophage colony-stimulating factor (GM-CSF), IL-3, IL-5 and tumour necrosis factor (TNF)), nerve growth factor and mast cell proteases. Actual physical coupling of eosinophils and mast cells has been observed both in vitro and in vivo, and this interaction prolongs eosinophil survival54.

Eosinophil responses to pathogens and parasites

Eosinophils and helminths: who wins? The historic view that eosinophils promote host defence against helminths arose largely from histological images of eosinophils and parasites in tissue specimens and from in vitro studies that documented the antiparasitic activities of the eosinophil granule proteins MBP and ECP. With the development of reagents that block eosinophilia in mice (such as IL-5-specific antibodies) and of IL-5- or eosinophil-deficient mice, the picture has become more complex. For instance, the helminth Schistosoma mansoni, although not a natural mouse pathogen, can infect wild-type mice and can elicit a profound TH2-type cytokine-mediated pathology and cause the accumulation of eosinophils in tissues55. Although the eosinophil granule proteins ECP and MBP are toxic to both schistosomules and the larvae of S. mansoni, the manipulation of eosinophils in mouse models had no significant impact on disease development during S. mansoni infection56,57. However, in Strongyloides stercoralis and Angiostrongylus cantonensis infection models, eosinophil depletion resulted in prolonged survival of tissue-based larval forms of the parasites58,59. Thus, the role of eosinophils in mouse models of helminth infection remains unclear and controversial.

The interaction of eosinophils and helminths during infection in human subjects has been examined using a genomics approach60. The 434G>C polymorphism in the gene encoding ECP results in substitution of the cationic amino acid arginine for the neutral amino acid threonine at position 97. The genotype 434CC — which encodes the more neutral and somewhat less cytotoxic form of ECP — is found commonly among Ugandans, who live in a region endemic for S. mansoni infection. By contrast, the 434CC genotype is quite rare in Sudan, where S. mansoni is not endemic. Although this result suggests that there is no selective advantage for those individuals whose eosinophils might provide stronger antischistosomal host defence, the authors of this study determined that individuals with the 434CC genotype developed substantially less liver fibrosis secondary to S. mansoni infection. As such, the selective advantage may be for those individuals whose eosinophils promote less collateral tissue damage when faced with a similar pathogen burden. Similarly, cerebral malaria, a severe outcome of infection with Plasmodium falciparum, is also associated with eosinophilia and elevated serum levels of ECP. The haplotype strongly associated with susceptibility to severe disease encodes arginine at position 97 and thus the more cationic form of ECP61. The explanation of this finding awaits further clarification of the role of eosinophils in cerebral malaria.

The most recent developments in this field have exploited current concepts of eosinophils as immunomodulatory cells. In wild-type mice, infection with Trichinella spiralis induces eosinophil recruitment to the infected tissues and the formation of nurse cells in skeletal muscle. In eosinophil-deficient ΔdblGATA and TgPHIL mice, T. spiralis larvae do not survive, largely owing to the diminished recruitment of TH2 cells and a concomitant increase in the activity of inducible nitric oxide synthase (iNOS) and the synthesis of nitric oxide in local macrophages6,7. One interpretation of these results is that the parasites recruit eosinophils to support their own persistence and survival; another possibility is that eosinophils are recruited to maintain homeostatic balance by limiting the development of TH1-type immune responses that lead to oxidative damage and tissue destruction. How the parasite elicits this response and whether this finding is unique to Trichinella species are important subjects for future consideration. In addition, it will be interesting to address whether the mechanisms by which T. spiralis recruits eosinophils to muscle tissue, the activation state of the eosinophils at this site and the mediators released in situ are similar to those involved in eosinophilic inflammatory myopathies.

Eosinophils and bacteria: pathogens, probiotics and the microbiome. Early experiments carried out in vitro documented the bactericidal properties of the cationic eosinophil granule proteins MBP and ECP62,63. Subsequent studies exploring the mechanisms involved showed that ECP has a specific affinity for bacterial lipopolysaccharide and peptidoglycan and can agglutinate Gram-negative bacterial pathogens64. More recently, in vivo studies of the interaction of eosinophils with bacteria documented the catapult-like release of structures resembling neutrophil extracellular traps (NETs) from eosinophils, and this was associated with protection from the lethal sequelae of caecal ligation65. In contrast to NETs, which are composed primarily of nuclear DNA and neutrophil-specific proteins, eosinophil NET-like structures are composed of mitochondrial DNA, MBP and ECP66. Whether eosinophils and their secretory mediators have physiological bactericidal functions in vivo requires further study. Although eosinophil-enriched IL-5-transgenic mice were protected from the lethal sequelae of Pseudomonas aeruginosa infection67, recent findings suggest that IL-5-mediated protection during bacterial sepsis might be mediated by cells other than eosinophils68.

Recently, tremendous interest has developed regarding the immunomodulatory impact of probiotic or health-promoting bacteria. Although the mechanisms remain uncertain, oral administration of live probiotic Lactobacillus or Bifidobacterium species suppressed eosinophil recruitment in mouse models of allergic airway disease69,70. However, the therapeutic impact of probiotics in human studies of allergic disease has been less impressive. Indeed, in a recent prospective study in which allergic children were provided with oral supplementation with Lactobacillus rhamnosus GG or a placebo control, no significant differences were recorded in the number of asthma exacerbations per year, the number of days on medication, the peripheral blood eosinophil count or the serum ECP levels71.

In parallel, the interactions between commensal bacteria and tissue-resident eosinophils in the intestine have been the subject of recent investigations. Mice raised under germ-free conditions exhibited exaggerated eosinophilia in a model of allergic airway inflammation; this phenotype was reversed when the gastrointestinal tract was colonized with normal microflora72. Likewise, a large prospective study involving over 400 healthy infants73 concluded that individuals with greater bacterial diversity in the gastrointestinal tract had a lower risk of developing allergic sensitization later in life.

Eosinophils and viruses. Human respiratory viruses — such as influenza virus, parainfluenza virus, respiratory syncytial virus (RSV), coronaviruses and, most prominently, rhinoviruses — are among the most common causes of asthma exacerbation. Although asthma typically involves dysregulated eosinophil recruitment, and eosinophils are generally perceived as promoting disease pathology in this setting, the outcome of eosinophil–virus interactions has not been fully explored. A recent concept to emerge is that eosinophils and their secretory mediators may have a role in promoting antiviral host defence. An initial study showed that eosinophil secretory mediators decrease the ability of RSV to infect target host epithelial cells74. This was followed by a later report75 that found that eosinophils that were induced by allergen sensitization decreased viral loads during parainfluenza virus infection in a guinea pig asthma model. Accelerated clearance of RSV has been demonstrated in the lungs of eosinophil-enriched Cd2-IL-5-transgenic mice (which overexpress IL-5 under the control of the Cd2 promoter)76, and activated eosinophils protect mice from the lethal sequelae of acute pneumovirus infection (C. Percopo, K.D.D., S. Ochkur, J. Lee, J. Domachowske and H.F.R., unpublished observations). Moreover, both human and mouse eosinophils release immunomodulatory mediators, notably IL-6, in response to infection with respiratory virus pathogens77,78.

Hypereosinophilia is a frequent finding in late-stage HIV infection, typically in association with allergic and/or immune dysfunction and low CD4+ T cell counts79. Furthermore, one study documented large numbers of CD8+CD30+ T cell clones expressing TH2-type cytokines (including IL-5) in HIV-positive donors80, although another did not confirm this finding81. Interestingly, the granule protein EDN has been shown to have HIV-inhibitory activity82. However, the precise mechanisms by which eosinophils and their secretory mediators interact with viral pathogens remain to be elucidated.

Eosinophils and disease

There is extensive literature on eosinophil dysregulation associated with diseases such as asthma and eosinophilic oesophagitis. Although we know a substantial amount regarding how eosinophils develop and how they are recruited into various organs and tissues, there is a lack of understanding regarding the roles of eosinophils in eosinophil-associated diseases — even the relatively common ones. Targeting eosinophils therapeutically has revealed the complex and heterogeneous nature of eosinophil-associated diseases. We have selected the examples that follow to illustrate these principles; a more extensive list of diseases associated with eosinophilia is included in Supplementary information S1 (table) (see also Ref. 83).

Eosinophils and asthma. Asthma is a chronic inflammatory disease that is characterized by reversible airway obstruction and airway hyperreactivity in response to nonspecific spasmogenic stimuli. Eosinophils are a common feature of the inflammatory response that occurs in asthma, as they are recruited to the lungs and airways by cytokines that are released from activated TH2 cells and by a range of chemokines, most notably those of the eotaxin family.

A role for eosinophils in promoting the pathogenesis of some forms of asthma is supported by a large body of literature, primarily from studies of acute and chronic allergen-challenged mouse models of allergic airway disease84,85. Antigen sensitization and challenge, typically with ovalbumin or Aspergillus species, induces an allergic airway disease that replicates many of the hallmark features of allergic asthma, including increased numbers of cytokine-secreting TH2 cells and eosinophils in the airways, mucus hypersecretion and airway hyperreactivity. Chronic exposure to these antigens results in features of airway remodelling, including fibrosis and thickening of the basement membrane. Collectively, these studies suggest that targeting eosinophils themselves, eosinophil migration and/or eosinophilopoiesis should provide therapeutic benefit for the treatment of asthma.

These findings ultimately led to the development of two humanized IL-5-specific monoclonal antibodies, mepolizumab and reslizumab, which block the binding of IL-5 to IL-5Rα. In two of the earliest studies86,87, mepolizumab was administered to patients with mild atopic asthma and to healthy volunteers. In response, eosinophil numbers in the bronchial mucosa decreased by 50%, an observation that correlated with reduced levels of the prominent pro-fibrotic eosinophil secretory cytokine, transforming growth factor-β1 (TGFβ1), and with diminished deposition of extracellular matrix proteins. Similarly, another study showed that mepolizumab suppressed eosinophil maturation in the bone marrow and resulted in fewer CD34+IL-5Rα+ eosinophil progenitors in the lungs87.

In initial clinical trials, small cohorts of patients with mild or moderate asthma were treated with mepolizumab or reslizumab, respectively88,89. Both monoclonal antibodies were well tolerated by patients, and both reduced eosinophil numbers in the blood and airways. However, no objective measures of clinical improvement emerged (Box 3).

In part owing to the results from these clinical trials, the complex nature of the inflammatory response in asthma has been revisited90,91. Four distinct phenotypes based on the inflammatory cell profile in induced sputum have been introduced (Box 3) and, likewise, categories of asthma endogenous phenotypes (referred to as endotypes) based on molecular mechanisms and environmental influences have been defined. In subsequent studies, the therapeutic potential of IL-5-specific monoclonal antibodies was explored in a subset of asthmatics who were steroid dependent and had persistent sputum eosinophilia91,92,93,94,95. In these trials, the numbers of eosinophils in sputum fell to almost zero, a finding that correlated with decreased frequencies of exacerbations, a steroid-sparing effect, improved lung function and long-term improvements in asthma control.

These studies highlight the heterogeneous nature of asthma90,91 and, most importantly, define a clinical phenotype — known as steroid-resistant eosinophilic asthma — in which eosinophils make a clear and direct contribution to current disease and its management. New approaches that target eosinophils directly, such as the cytotoxic IL-5Rα-specific monoclonal antibody benralizumab, or indirectly, such as the IL-13-specific monoclonal antibody lebrikizumab, may further enhance therapeutic outcomes96,97.

Eosinophilic oesophagitis. Eosinophils are normally found in the gastrointestinal tract, notably in the caecum, but not in the oesophagus. First described by Landres and colleagues in 1978, eosinophilic oesophagitis is the most common of the eosinophil-associated gastrointestinal diseases. In 2007, an international consortium — the First International Gastrointestinal Eosinophil Research Symposium (FIGERS) — published consensus guidelines for diagnosis, which were revised in 2011. These criteria include: clinical evidence of oesophageal dysfunction (including dysphagia, abdominal pain and/or food bolus impaction); 2–4 biopsy samples from the proximal and distal oesophagus with ≥15 eosinophils per field at ×400 magnification; and no response to 6–8 weeks of high-dose proton-pump inhibitor therapy, ruling out gastro-oesophageal reflux disease. As we focus here on eosinophil-mediated mechanisms, we refer readers to a recent review on the complete natural history of eosinophilic oesophagitis98.

All evidence points to dysregulated eosinophilia as being central to the pathophysiology of eosinophilic oesophagitis. The aetiology appears to be dependent on the TH2-type cytokines IL-5 and IL-13. Patients often report concurrent allergic responses to food and airborne allergens, along with a family history of allergy, and there is an unexplained male predominance. Although absolute eosinophil numbers in biopsy samples at any given time may or may not correlate directly with disease severity, evidence of eosinophil activation — including the presence of extracellular granules and degranulated cationic proteins (such as MBP) — is prominent in tissue biopsy samples99. The eosinophil chemoattractant CCL26 (also known as eotaxin 3) is a prominent biomarker of eosinophilic oesophagitis. Indeed, CCL26 is highly upregulated in diseased tissues and also in peripheral blood cells in patients with this disorder. A single-nucleotide polymorphism (2,496T>G) in the 3′ untranslated region of the gene encoding CCL26 has been associated with increased susceptibility to eosinophilic oesophagitis, although the mechanisms involved are not yet known100. Susceptibility to eosinophilic oesophagitis has also been correlated with polymorphisms in the gene encoding TSLP101.

There are several mouse models of eosinophilic oesophagitis. Some of these models use oral or intranasal delivery of allergens to elicit tissue pathology, and others promote eosinophil recruitment to the oesophagus via the overexpression of IL-5 or IL-13. Among these models, one uses repeated intranasal delivery of fungal or insect aeroallergens102,103, which induces the expression of TH2-type cytokines and the eotaxin family member CCL11 (mice do not express CCL26), resulting in eosinophil recruitment to the oesophagus. Another mouse model involves systemic sensitization with ovalbumin in aluminium hydroxide adjuvant followed by repeated intra-oesophageal challenge104, which induces eosinophil recruitment associated with angiogenesis, basal zone hyperplasia and tissue fibrosis. Interestingly, although the administration of eosinophil-depleting SIGLEC-F-specific antibodies to these mice inhibits eosinophil recruitment and the associated tissue remodelling104, in another investigation, in which oesophageal remodelling was driven by lung-specific expression IL-13, no role for eosinophils was observed105. Similarly, ablation of CD4+ T cells — which presumably leads to a reduction in the levels of TH2-type cytokines — has only a limited impact on the recruitment of eosinophils to the oesophagus after chronic administration of Aspergillus species antigens102. Among the issues to be addressed in future studies is the role of eosinophil degranulation into the oesophageal tissue in these mouse models. Furthermore, mouse models that incorporate relevant clinical symptoms, such as failure to thrive, would certainly be of significant value.

Current therapies for patients with eosinophilic oesophagitis include the introduction of an elemental diet and treatment with steroids, which target the global inflammatory response and have an impact on eosinophil-derived cytokines106. Therapies that specifically target eosinophils are also being tested. For example, a randomized placebo-controlled double-blind trial in which adults with eosinophilic oesophagitis were treated with a humanized IL-5-specific monoclonal antibody (mepolizumab) resulted in a reduction in oesophageal inflammation and the reversal of tissue remodelling, but only minimal relief of symptoms107. Similar results were obtained in a prospective study in children, with clinical improvement observed in both experimental and placebo groups108. Interestingly, mepolizumab did not deplete eosinophils found in the duodenal mucosa of these patients109. However, the aforementioned studies suggest that this disorder may be primarily regulated by CCL26. As with asthma, the stratification of patients into subgroups that respond to specific therapies may ultimately improve clinical outcomes.

Eosinophilic myopathies. These conditions are among the most rare and poorly characterized of the eosinophil-related disorders, and include eosinophilic fasciitis (also known as Shulman's syndrome), toxic oil syndrome and eosinophilia–myalgia syndrome110 (Box 4). Although eosinophils are associated with these conditions, it is not clear how they are recruited to the affected tissue or what their contributions are to the pathology observed.

Eosinophilic myositis is a relatively rare condition in which the infiltration of muscle tissue by eosinophils is observed, sometimes in association with peripheral blood and bone marrow eosinophilia. The disease can result from helminth infection, or it can be toxin induced or idiopathic in nature. Recently, specific mutations in the gene encoding calpain 3 were identified in association with idiopathic eosinophilic myositis111. Calpain 3 is a muscle-specific neutral cysteine protease that interacts with intracellular myofibrillar proteins and has a role in sarcomere adaptation. However, there is no direct or obvious relationship between the actions of this enzyme and eosinophils or eosinophilia. It is not clear why mutations in calpain 3 result in signals that elicit eosinophil accumulation, what these signals might be, whether eosinophils are a primary or indirect target, and whether eosinophils are promoting tissue damage or altering the local immune status. One possibility is that inflamed, damaged muscle tissue releases endogenous alarmins (such as IL-33 and/or HMGB1) that activate innate immune signalling pathways that lead to peripheral blood and tissue eosinophilia. Of note, limb-girdle muscular dystrophy type 2A, which is a common autosomal recessive form of muscular dystrophy, has also been directly linked to mutations in the gene encoding calpain 3. Although eosinophils do not have a prominent role in this disorder, transient eosinophilia has been reported in the early stages of the disease. Similarly, no infiltration of eosinophils into muscle tissue was reported in calpain 3-deficient mice112, although this observation should be reassessed in other mouse strains.

Hypereosinophilic syndromes. Hypereosinophilic syndromes are disorders of eosinophil haematopoiesis that result in hypereosinophilia (defined as >1,500 eosinophils per mm3) in peripheral blood in the absence of any known aetiology. Although these disorders were recognized early on as clinically heterogeneous, recent studies have revealed the molecular basis for a few of the distinct phenotypes. The identification of myeloproliferative hypereosinophilic syndrome (MHES) emerged from the dramatic therapeutic responses observed in a subset of patients with hypereosinophilic syndrome following empirical treatment with imatinib, a tyrosine kinase inhibitor first developed for the treatment of chronic myeloid leukaemia113. This clinical observation led to the detection of a deletion in chromosome 4 that results in the fusion of the genes encoding pre-mRNA 3′-end-processing factor FIP1 (FIP1L1) and platelet-derived growth factor receptor-α (PDGFRA)114. This leads to the production of a FIP1L1–PDGFRA fusion protein that constitutively activates proliferation and survival pathways, resulting in the clonal proliferation of eosinophils, elevated serum levels of tryptase and vitamin B12 (also known as cobalamin), severe peripheral eosinophilia and end-organ damage, the most severe form of which is endomyocardial fibrosis. Other fusion kinases have also been identified in individuals with MHES; other individuals display clinical symptoms consistent with MHES but without a clear molecular diagnosis. Thus far, all of the PDGFRA- or PDGFRB-derived mutant fusion proteins that have been identified in humans have been associated with eosinophilia, for reasons that remain obscure.

The constitutive cellular activation and proliferation promoted by the FIP1L1–PDGFRA fusion protein has been explored in cell-culture models. For example, Ba/F3 immortalized mouse pro-B cells require the cytokine IL-3 for survival and proliferation in culture, but stable expression of the FIP1L1–PDGFRA fusion gene activates intracellular signalling pathways and eliminates the requirement for this cytokine113. Likewise, imatinib inhibits the growth of the human eosinophil leukaemia EoL-1 cell line, which expresses the FIP1L1–PDGFRA fusion protein115. Most intriguingly, the uncontrolled activity of the fusion protein lies within the PDGFRA component, as the fusion eliminates an inhibitory juxtamembrane region encoded by exon 12 of the PDGFRA gene, resulting in constitutive signalling by PDGFRA in the absence of its ligand116.

In contrast to the myeloproliferative variants, eosinophilia in lymphocytic-variant hypereosinophilic syndrome (LHES) results from aberrantly activated T cell clones that constitutively produce eosinophilopoietic cytokines, including IL-5. The resulting eosinophilia is thus reactive. The aberrant T cell clones (which typically have a CD3CD4+ phenotype) are also associated with elevated serum levels of IgE and CCL17, and elicit predominantly skin manifestations, including pruritus, eczema, erythroderma, urticaria and angio-oedema. Individuals with this diagnosis respond to treatment with steroids, with cytotoxic agents (such as hydroxyurea) and with mepolizumab117, which reduced the requirement for corticosteroids in clinical studies.

These two defined variants of hypereosinophilic syndrome currently represent a minority of cases. Indeed, a recent study showed that FIP1L1–PDGFRA fusions were associated with only 11% of cases of hypereosinophilic syndrome, and LHES accounted for only 17% of cases118. The classification of hypereosinophilic syndrome is currently a work in progress, and attempts are being made to balance the clinical diagnosis with the predicted response to therapy119.

In an initial mouse model, bone marrow transplantation using haematopoietic progenitors that had been retrovirally transduced with FIP1L1–PDGFRA resulted in myeloproliferative disease120. Another group created a model that combines features of both myeloproliferative and lymphocytic-variant disease121 by transducing haematopoietic progenitors from Cd2-IL-5-transgenic mice with FIP1L1–PDGFRA, which resulted in profound peripheral eosinophilia in association with tissue infiltration. Most recently, mice lacking the serine/threonine kinase NIK (also known as MAP3K14) were found to develop a CD4+ T cell-dependent blood and tissue eosinophilia122. However, future studies will be necessary to determine whether these mouse models will be useful in identifying the disease mechanisms underlying distinct hypereosinophilic syndromes.

Eosinophils: changing perspectives

The field of eosinophil research is one of changing perspectives and emerging new directions. Eosinophils are clearly capable of more sophisticated immune functions than previously thought, as shown by their nuanced degranulation responses to distinct stimuli and their complex interactions with other leukocytes and pathogens. Both successful and unsuccessful attempts to target eosinophils have yielded remarkable insights into disease pathogenesis. Asthma and hypereosinophilic syndromes are now understood to be complex heterogeneous disorders that require tailored therapeutic strategies. Assessing the role of endogenous and exogenous PRR ligands in eosinophil responses and clarifying the relationship between eosinophil degranulation and tissue remodelling will be important goals for future research. A better understanding of these and other aspects of eosinophil biology will aid the development of new therapeutic strategies for diseases characterized by eosinophil dysregulation.