In December 2019, a novel coronavirus was identified as the cause of pneumonia in a cluster of cases in Wuhan, China [1]. This virus, which causes coronavirus disease 2019 (COVID-19), was identified as severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) [2]. Today, approximately 80% of confirmed COVID-19 cases are asymptomatic or lead to mild symptoms in infected individuals. However, in the remaining 20% of cases severe pneumonia often developed, with patients requiring supplemental oxygen therapy [3,4]. The mechanisms underlying the different degrees of pneumonia severity observed in patients with COVID-19 remain elusive.

To characterize the cause of pneumonia in patients with COVID-19, we analyzed the immunological features of respiratory and blood specimens collected from three patients with confirmed infection by SARS-CoV-2 but varying degrees of clinical symptoms: one patient with mild symptoms without pneumonia and two patients with severe pneumonia. We analyzed plasma, leukocytes in blood, bronchoalveolar lavage fluids (BALFs), sputa, and tracheal aspirates to characterize the immunological responses to viral infection. We propose the possibility of one of several immune mechanisms of COVID-19 by reporting eosinophil-mediated lung inflammation associated with elevated natural killer T (NKT) cell response in patients with COVID-19.

The clinical courses and baseline characteristics of the three patients are summarized in Table 1 and Fig. 1A. P1 and P2 had severe pneumonia and required mechanical ventilation (Fig. 1B). P3 had only mild symptoms, without pneumonia. None of the three patients had a history of eosinophilia due to conditions, such as asthma, or allergy.

Initially, we analyzed BALFs and sputa to identify immune cells that had infiltrated the infected lungs. The cytological analyses of BALF and sputum were performed using hematoxylin and eosin (H&E) staining, which showed that polymorphonuclear cells (PMNs), especially eosinophils or neutrophils, were the major inflammatory cells, along with a few macrophages and lymphocytes (Table 1 and Fig. 1C). BALF analysis of the two patients that developed severe pneumonia (P1 and P2) showed that lymphocytes accounted for 20% and PMNs accounted for more than 35% of the inflammatory cells. Eosinophils accounted for more than 25% of the PMNs. Eosinophils constituted less than 1% of BALF obtained from healthy individuals. Acute eosinophilic pneumonia is diagnosed when eosinophils constitute more than 25% of the BAL differential count in the absence of other causes of BALF eosinophilia, such as asthma or atopic disease, and drugs [6]. Our study revealed that COVID-19 could cause eosinophil-mediated inflammation in the lungs, similar to acute eosinophilic pneumonia.

Cytological analysis of sputum and tracheal aspirates collected from the three patients revealed that more than 90% of the cells were eosinophils (Table 1 and Fig. 1C). Notably, a sputum sample was collected from P3 (who did not develop pneumonia) on day 3, after symptom onset, also showed high levels of eosinophils. Additionally, a sputum sample obtained from P2, who rapidly developed severe pneumonia on day 2, also contained a large number of PMNs (Table 1 and Fig. 1C). Therefore, infiltration of PMNs into infected lungs might be rapid and universal in patients with COVID-19, regardless of the severity of pneumonia.

Based on the cytological analysis, we suspected that among the other PMNs, eosinophils infiltrated the lungs. Therefore, we measured the levels of eosinophil cationic proteins (ECPs) in the respiratory specimens. The sputum and BALF specimens from the three patients contained high levels of ECP (Table 1), which indicated the infiltration of eosinophils into the inflamed lungs during the acute phase of COVID-19. To further characterize the PMNs infiltrating the pneumonic lungs, BALF was collected from P1 on day 12 (acute phase) and 19 days (convalescent phase) after symptom onset and subjected to cytometric analysis. Peripheral blood collected from P2 on day 12 (acute phase) and P1 on day 19 (convalescent phase) after symptom onset were simultaneously used for flow cytometric analysis of leukocytes. We identified the relative proportions and kinetic changes in CD14⁺ monocytes/macrophages, CD3⁺ T cells, CD20⁺ B cells, and side scatter (SSC)^high/CD24⁺ PMNs, including neutrophils and eosinophils, in the blood leukocytes and BALF (Supplementary Fig. 1). The BALF from P2 showed a marked decrease in the number of CD45⁺ leukocytes during the convalescent phase (~2.5 × 10³/mL) of the disease, compared to that in the acute phase (~1.3 × 10⁵/mL). Similar to the cytological analysis of BALF, the relative proportion of PMNs in the CD45⁺ leukocyte populations (after excluding macrophages, T cells, and B cells) increased in the acute phase (30.5%) and decreased to 10.6% in the convalescent phase. Although the PMNs in BALF collected during the acute phase were primarily composed of CD16⁺/CD24⁺ neutrophils (26.8% in CD45⁺ leukocytes), approximately 10% (3.0% in CD45⁺ leukocytes) of the PMNs were likely to be CD24⁺ eosinophils with reduced expression of CD16 on the surface (Supplementary Fig. 1) [6]. Monocytes in the peripheral blood increased during the acute phase but decreased during the convalescent phase. However, of the total leukocytes in the BALF, monocyte counts decreased during the acute phase and increased during the convalescent phase. Although very few B cells were detected in the BALF during the acute phase, their numbers increased in the convalescent phase. Interestingly, most T cells in the acute phase BALFs showed intermediate levels of surface CD3 molecules compared to those of blood T cells, indicating the presence of other types of CD3⁺ lymphocytes. We assessed the surface expression of other cellular markers, such as CD24, CD16, and human leukocyte antigen DM (HLA-DM), and found that the majority of the CD3⁺ T cells in the acute phase of BALF were CD16⁺/CD24⁺/HLA-DM⁻ cells, suggesting a predominantly NKT cell population (the last row of Supplementary Figs. 1 and 2A) [7,8]. An increase in the number of NKT cells expressing both CD3 and CD56 in the patient’s blood during the acute phase was further confirmed by flow cytometric analysis (Fig. 2A, 2B, and Supplementary Fig. 2B) of peripheral blood mononuclear cells (PBMCs). The relative levels of CD3⁺/CD56⁺ NKT cells were 6.4-and 2.5 times higher in the patients’ blood lymphocytes during the acute phase (5.8% ± 1.3%, mean ± standard deviation) than in healthy controls (0.9% ± 0.8%) and patients in the convalescent phase (2.3% ± 1.3%), respectively. Additionally, the CD3⁺/CD1d-tetramer⁺ NKT (invariant NKT [iNKT]) [9] cells at acute phase (1.3 ± 0.4) were significantly elevated by 4.3 and 1.6 times compared to those of healthy controls (0.3 ± 0.1) and convalescent phase (0.8 ± 0.1), respectively (Fig. 2A, 2B, and Supplementary Fig. 2B). However, the levels of CD3⁻/CD56⁺ NK cells in the peripheral blood changed depending on the individual cases (Supplementary Fig. 2B and 2C). Quantitative changes in the NKT cell population were consistently observed in the three patients, suggesting that upregulation of the NKT cell population might be a common phenotype initiated upon SARS-CoV-2 infection regardless of disease severity. Furthermore, compared to the other lymphocytes, the relative frequency of NKT cells among lymphocytes decreased during the convalescent phase, which suggests that the kinetic changes in NKT cells might be correlated with disease progression.

Significant elevation and infiltration of NKT cells and PMNs, including neutrophils and eosinophils, into the lungs of patients with COVID-19 during the acute phase of infection indicated that these innate immune cells might be the initial mediators of pulmonary inflammation. This type of immunopathological response appears to commonly occur primarily in patients with symptomatic COVID-19 regardless of disease severity, as a similar degree of eosinophilic infiltration was observed in P3, who did not have pneumonia. Several studies have recently been published on the composition of inflammatory cells in the BALF of individuals with severe COVID-19. Dentone et al. [10] reported that neutrophils and macrophages were the major cell types in the BALF of patients with severe COVID-19, whereas lymphocytes were a minority. Furthermore, they identified a correlation between the percentage of macrophages in the BALF and patient mortality [10]. Liao et al. [11] reported that the BALF of patients with severe COVID-19 had higher levels of macrophages and neutrophils and lower proportions of T and NK lymphocytes than those of patients with moderate disease severity. Pandolfi et al. [12] performed a BALF analysis of patients with severe COVID-19 and observed high levels of neutrophils with very few lymphocytes and macrophages. Although these studies did not investigate the role of eosinophils, our current data indicate the potential involvement of eosinophil-mediated inflammation in COVID-19 patients.

Eosinophils are known to play a role in the immune response against various viral infections. Fraenkel et al. [13] reported that the common cold, caused by rhinovirus, is associated with eosinophilic infiltration of the bronchial mucosa. Harrison et al. [14] also showed that respiratory syncytial virus (RSV) infection induced ECP production in the lower airway and suggested that eosinophil degranulation plays a role in the pathogenesis of respiratory infection by RSV. In other words, given the various roles that eosinophils play in host immune responses, it may be suggested that viral infections cause an increase in eosinophil counts at the site of infection. Jeon et al. [15] reported a case of acute eosinophilic pneumonia associated with influenza A (H1N1) infection in 2009. Subsequently, LeMessurier et al. [16] showed that eosinophils contribute to antiviral host defense mechanisms against respiratory infections caused by the 2009 influenza A (H1N1) infection.

Cavalli et al. [17] showed that the relative proportions of infiltrating immune cells, such as NKT cells and eosinophils, in the lungs of COVID- 19 patients. These findings are consistent with the results of our study. Their study showed that there was a moderate and insignificant increase in NKT cells in the lungs of COVID-19 patients. In addition, eosinophils also showed a non-significant increase [17]. However, the role of NKT cells in early eosinophilic infiltration has been previously reported in a mouse model of allergic asthma and neonatal RSV infection [18,19]. Moreover, CD3⁺CD16/56⁺ NKT cell levels were positively correlated positively with the percentage of eosinophils in BALF specimens from patients with eosinophilic pneumonia [20]. Although the association between elevated NKT cells and the pathogenesis of eosinophil-mediated inflammation is poorly understood, it may depend on the underlying airway epithelial injury caused by SARS-CoV-2 infection [21]. Subsequently, eosinophilic infiltration and degranulation appear to mediate lung inflammation and are associated with clinical manifestations [22]. Peripheral blood eosinophilia may initially present as eosinophil-mediated inflammation; however, it may also be absent or delayed [22], as consistently observed in patients diagnosed with COVID-19 (see complete blood count [CBC] results in Supplementary Fig. 3) [23].

As eosinophil-mediated inflammation in the lungs is associated with T-helper type 2 (Th2)-polarized immune responses [22], we next assessed viral antigen-specific antibody responses and focused on specific isotypes. We speculated that specific immune responses against the SARS-CoV-2 antigen are orchestrated by a Th2-biased immune environment, which can drive the preferential isotype switching of specific antibodies in B cells toward specific subclasses, such as immunoglobulin A (IgA) and IgE [24]. There was an increase in the levels of anti-N protein antibodies, including IgG1, IgG3, IgA, and IgE isotypes (IgG2 and IgG4 were not detected) in the plasma from P1 and P2, who had severe pneumonia, whereas only moderate levels of IgG1 and IgG3 antibodies were observed in P3 with mild symptoms (Fig. 2C). Moreover, IgA and IgE antibodies were not detected at P3.

The administration of steroids in patients infected with other coronaviruses, such as Middle East respiratory syndrome (MERS) or SARS-CoV-1, resulted in delayed viral clearance with other short- and long-term side effects [25]. The World Health Organization (WHO), Centers for Disease Control and Prevention (CDC), and Infectious Disease Society of America (IDSA) guidelines do not recommend the treatment of COVID-19 patients with glucocorticoids, unless indicated otherwise, such as that during COPD exacerbation, until the Randomised Evaluation of COVID-19 Therapy trial (RECOVERY) trial was published [26]. However, systemic glucocorticoid therapy is generally considered the main treatment for acute eosinophilic pneumonia [22]. Consequently, we assumed that in patients with COVID-19, the systemic steroid (dexamethasone) that was used in the RECOVERY trial might have had a therapeutic effect on eosinophil-mediated lung inflammation via the above-mentioned mechanism. In our study, methylprednisolone was administered to patients with rapid and progressive pneumonia (P1 and P2); their oxygen demand decreased substantially and there was an improvement in their X-ray findings, although the blood glucose levels rose to 400 to 500 mg/dL. This improvement was suspected to be the result of the steroid administration. Moreover, an analysis of 201 COVID-19 patients from Wuhan indicated a reduction in the mortality risk for patients with ARDS treated with methylprednisolone [27]. The ability of steroid treatment to decrease the mortality risk in patients with COVID-19 suggests that COVID-19 associated pneumonia is, in fact, an eosinophilic pulmonary inflammation. Therefore, future studies on possible therapies for eosinophilic pulmonary inflammation in COVID-19 patients need to be conducted, targeting eosinophilia and NKT cell response in the pulmonary tract [28–30]. Additionally, rapid and strong elevation of anti-viral antibodies, such as IgE, as we observed in a patient with severe COVID-19, must be taken into consideration for plasma therapy (i.e., potential risk of type I hypersensitivity) [31].

This study had several limitations. As the study data consisted of a case series of only three patients and the data was collected retrospectively, other important clinical variables that contribute to the differential diagnosis of other obvious causes of eosinophilia, such as thyroid disease, parasite infection, malignancy, and allergic disease, may have been missed.

Notwithstanding these limitations, the results of our study showed that COVID-19 can lead to eosinophil-mediated inflammation via systemic elevation of NKT cells. The infiltration of NKT cells into infected lungs may be associated with COVID-19 pneumonic inflammation.