Enhanced expression of HIF-1α in the lungs by complication of pneumonia with preexisting lung diseases
To elucidate the hypoxic state in the lungs, IHC for HIF-1α was carried out using FFPE tissues from normal lungs (Group 1, n = 2), from lungs with pneumonia (Group 2, n = 7), lung diseases without pneumonia (Group 3, n = 9), and preexisting lung diseases with pneumonia (Group 4, n = 5). Details of the lung tissue samples are listed in Table S1. HIF-1α was expressed in bronchial epithelial cells, inflammatory cells, and alveolar macrophages (Figure S1a). The specificity of IHC was guaranteed by the negative staining for isotype control mouse IgG1 (Figure S1b). Because there was no obvious difference in expression level between cell types, we focused on bronchial epithelial cells. After observing the whole section under a microscope, the expression level was scored by a cytoscreener (with 10 years of experience) as follows: 0, undetectable; 1, weak; 2, moderate; and 3, strong (Fig. 1a), and then the scoring was confirmed by a pathologist (with 25 years of experience). This is a standard process of clinical cytoscreening. HIF-1α expression in Group 4 was significantly higher than in Groups 2 and 3 (Fig. 1b). These findings corresponded to our prediction that serious hypoxia occurs in the lungs when preexisting lung diseases are complicated with pneumonia.
Increase in NETs in the lungs by complication of pneumonia with preexisting lung diseases
We next determined the amount of NETs in the lung tissues of the 4 groups (serial sections of FFPE specimens used in Fig. 1). NETs were recognized as Cit H3+ substances mixed with DNA and MPO in the alveoli, bronchi, and small vessels. The amount of NETs was semi-quantified by the cytoscreener as follows: Score 0, unrecognizable; Score 1, scattered small NETs; Score 2, scattered large NETs; and Score 3, dense large NETs, and then the scoring was confirmed by the pathologist. NETs were detected in Groups 2, 3, and 4 but not Group 1 (Fig. 2a). The amount of NETs was significantly higher in Groups 2 and 4 than in Groups 1 and 3, respectively (Fig. 2b). These findings are consistent with our prediction of increased NETs in pulmonary lesions of patients with preexisting lung diseases after complication of pneumonia. There was a significant positive correlation between HIF-1α expression score and amount of NETs (Fig. 2c).
Neutrophils stimulated by PMA under hypoxic conditions
When peripheral blood neutrophils were stimulated by PMA under hypoxic conditions (1% O2), numerous round substances of a size larger than non-treated neutrophils were observed, and the typical lytic NETs generated under normoxic conditions (21% O2) did not appear (Fig. 3a). These round substances were also generated under normoxic conditions in fewer numbers than under hypoxic conditions. Although round-shaped nuclei have been shown as characteristics of apoptotic neutrophils33, the round substances were regarded as NET-forming neutrophils because of the positive staining for Cit H3, MPO, and DNA.
To confirm this, concentrations of dsDNA in the supernatants were determined. The levels of dsDNA in the supernatants were significantly elevated after PMA stimulation under hypoxic conditions as well as under normoxic conditions (Fig. 3b), suggesting that the cell membrane of neutrophils was perforated and that decondensed DNA was exuded from the cytoplasm. The increase in dsDNA in the supernatants remained at 40%. This corresponded with the finding that all neutrophils exposed to PMA did not undergo NET formation. The NET-forming neutrophils appear to be in a state before releasing web-like DNA; thus, this is consistent with previous reports of reduced NET formation under hypoxia23,24. In this study, such round-shaped NETs formed by neutrophils in the state before releasing web-like DNA are called “round NETs” (Figure S2).
Next, we investigated whether NETs generated under normoxic (21% O2) and hypoxic (1% O2) conditions could be degraded by the principal NET degrader, DNase I (Fig. 3c). The lytic NETs generated under normoxic conditions were completely degraded by DNase I. In contrast, round NETs generated under both normoxic and hypoxic conditions remained even after DNase I treatment.
As shown in Figs. 3a and 3c, the neutrophils without PMA stimulation were apt to detach from the slides. To compare the digestion of NETs by DNase I, five photographs were taken randomly before and after DNase I treatment, and then mean nuclei area was calculated using Image J 1.50i software (http://allpcworld.com/download-imagej-1-50i-free/) as follows: DAPI+ areas/number of neutrophils (pixels). Image analysis demonstrated that NETs induced under normoxic conditions were partly but significantly degraded by DNase I but NETs induced under hypoxic conditions were not degraded by DNase I (Fig. 3d). These findings suggest that lytic NETs are sensitive to DNase I, whereas round NETs are resistant to DNase I. A similar phenomenon—the DNase I-resistance of round-shaped NETs induced by PMA with propylthiouracil—has been reported previously34.
Disruption of actin rearrangement during NET formation under hypoxic conditions
Morphological change in cells results from a rearrangement of the cytoskeleton. To assess the cytoskeletons of neutrophils during NET formation, we focused on actin rearrangement. It has been reported that digestion of F-actin by NE, which occurs ROS-dependently, is required for NET formation12. However, another study has indicated that ROS-dependent actin polymerization—transformation from G-actin to F-actin—is essential for NET formation35. To resolve this conflict, we utilized fluorescent staining to follow chronological changes in actin during NET formation by using an anti-β actin antibody that binds to G-actin31 and phalloidin that binds to F-actin specifically (Fig. 4a).
Under normoxic conditions (21% O2), G-actin was detected most strongly 30 minutes after PMA stimulation and then gradually decreased thereafter (Fig. 4b). F-actin was increased as G-actin was decreased and peaked 1–2 hours after PMA stimulation. The amount of F-actin was also decreased afterward. Finally, both signals became almost undetectable in lytic NETs observed 4 hours after PMA stimulation. These findings are consistent with the previous report of the absence of actin in NETs36.
Next, to elucidate actin rearrangement during NET formation under hypoxic conditions (1% O2), we examined the distributions of G- and F-actins after 4 hours incubation with PMA and compared them with the findings under normoxic conditions (21% O2; Fig. 4c). As observed in the former experiment, neither G- nor F-actin was evident in lytic NETs generated under normoxic conditions, although there were scattered actins around the cells. Under hypoxic conditions, in contrast, both G- and F-actins were detected in round NETs, especially in the periphery of the cells. We consider this finding to be the “actin caps” that have been recognized in neutrophils under hypoxia37.
These findings suggest that during NET formation under normoxic conditions, polymerization of G-actin occurs in an earlier period (30 minutes to 1 hour after PMA stimulation), and degradation of F-actin follows in a later period (1 to 4 hours after PMA stimulation). Hypoxia appears to disrupt the actin rearrangement that leads to the formation of lytic NETs after PMA stimulation.
Suppression of NET generation by actin polymerization inhibitor
To examine the association between the generation of NETs and actin polymerization, we next performed experiments using an actin polymerization inhibitor, CyD. CyD was added to peripheral blood neutrophils simultaneously with PMA, or 30 minutes before or after the addition of PMA, and then NET formation was monitored by DNA staining with DAPI after 4 hours incubation under normoxic conditions (21% O2; Figure S3a). When neutrophils were pre-treated with CyD, the generation of NETs was suppressed almost completely. In contrast, the suppression of formation of lytic and round NETs was partial when CyD was administered simultaneously and 30 minutes after PMA stimulation (Figure S3b).
Effects of inhibition of actin polymerization on susceptibility of NETs to DNase I
We demonstrated that round NETs generated under hypoxic conditions are resistant to DNase I digestion. G-actin, which can bind DNase I, is regarded as a putative inhibitor of DNase I25,38,39. To investigate the association between the DNase I resistance of round NETs and G-actin, peripheral blood neutrophils were treated simultaneously with CyD and PMA under normoxic conditions (21% O2). After 4 hours incubation, actin distribution was assessed by fluorescent staining. It was noted that G-actin remains in round NETs (Fig. 5a), which is a result similar to that when neutrophils are treated with PMA under hypoxic conditions (1% O2; see Fig. 4c). These findings suggested that the inhibition of actin polymerization by CyD under normoxic conditions suppressed the formation of lytic NETs, resulting in the increase in round NETs; thus, this mimicked NET formation under hypoxic conditions.
We further determined the DNase I susceptibility of round NETs that contained G-actin (Fig. 5b). Round NETs induced by PMA with CyD were found to be not digested by DNase I, although lytic NETs induced by PMA without CyD, which did not contain G-actin, was digested completely (Fig. 5c). These findings suggest a possible association between G-actin and the DNase I resistance of round NETs.
Cytotoxic potential of neutrophils stimulated under hypoxic conditions
Finally, we examined whether neutrophils stimulated under hypoxic conditions exhibited cytotoxicity. The supernatants of neutrophils stimulated under normoxic (21% O2) and hypoxic (1% O2) conditions were applied to a culture of NHLF. Twenty-four hours later, the viability of the cells was determined with a cell counting kit. Compared to the supernatants from non-treated neutrophils and medium containing equivalent concentrations of PMA, the number of viable NHLF was reduced significantly by the supernatants of neutrophils stimulated under hypoxic conditions similarly to those under normoxic conditions (Fig. 6a). These findings suggest that neutrophils stimulated under hypoxic conditions can secrete cytotoxic factors to the same extent as NETs generated under normoxic conditions, despite NETs being incompletely formed under hypoxic conditions.
MMP-9 has been demonstrated to be a critical cytotoxic molecule that is secreted from neutrophils during NET formation40. Regardless of PO2, a significant increase in MMP-9 secretion into the supernatants of PMA-stimulated neutrophils was observed (Fig. 6b).