Human peripheral blood preparations (from two normal male donors, collected in compliance with Kumamoto Health Science University and approved by the University Oversight Committee) were enriched for neutrophils by density gradient centrifugation with HISTOPPAQUE 117 (Sigma-Aldrich) and Lymphocyte Separation Solution 1.119 (Nakarai Tesque) according to the procedures described by the supplier. Washed enriched neutrophilic fractions were counted and examined for purity using Wright Giemsa staining (Sigma-Aldrich).

Cells were incubated in culture dishes containing an immersed coverslip in RPMI 1640 (Sigma-Aldrich) supplemented with 5% fetal bovine serum (FBS), 1% penicillin/streptomycin and 0.1% gentamaycin in a humidified atmosphere containing 5% CO₂. To induce NETs, PMA (Wako Pure Chemical Industries) was added to the culture medium at a concentration of 50 nM and incubated for 4 h at 37˚C. To inhibit NADPH oxidase activity, DPI (Cayman Chemical) was added to the culture medium at a concentration of 20 μM. Hypothermal treatment and rewarming of cells were performed by incubation in a humidified atmosphere containing 5% CO₂. After drug and/or hypothermal/rewarming treatment, the coverslips were removed from the cultures and subjected to appropriate assays. DNA was visualized by staining with DAPI.

Cells were washed once for 5 min with ice-cold PBS and then fixed with 4% paraformaldehyde in PBS for 5 min at room temperature. After fixation, the cells were rinsed once with PBS and subjected to indirect-immunofluorescence analysis using anti-neutrophil elastase (Calbiochem), anti-histone H3 (Santa Cruz Biotechnology), and anti-histone H3 citrulline R26 (Abcam) antibodies. The secondary antibodies were obtained from Santa Cruz Biotechnology and Sigma-Aldrich.

Escherichia coli BL21 (DE3) were transformed with pET28-EGFP, a plasmid for expression of green fluorescent protein (GFP), and cultured in Luria-Bertani (LB) medium containing kanamycin at 37˚C for 16 h. 10⁷ E. coli cells were incubated with a coverslip containing hypothermia/rewarming-induced DAPI-positive objects in RPMI 1640 supplemented with 5% FBS at 37˚C. After 20 min at room temperature, the coverslips were washed three times with PBS followed by incubation with 4% paraformaldehyde. DNA fibers were stained with DAPI. Because the E. coli expressed GFP, bacteria trapped by DNA fibers could be detected by fluorescence microscopy. For DNase I treatment, the coverslips were treated with PBS containing 100 U/ml DNase I (Takara) at 37^oC for 1 hr. The numbers of E. coli with GFP signals on the coverslips were counted by fluorescent microscopy.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays were performed using the MEBSTAIN Apoptosis TUNEL Kit II (MBL) according to the manufacturer’s instructions. The TUNEL-positive cells were counted under a microscope. The percentage of TUNEL-positive cells was defined by the number of positive cells among the total number of cells in each sample. For one experiment, cells were counted in at least three different microscopic fields of view.

Intracellular ROS production was monitored using the cell permeable fluorescent dye, CellROX Deep Red Reagent (Invitrogen). This agent can readily react with ROS to form a fluorescent product proportional to the amount of ROS generated in the cells. The cells were incubated with 5 μM CellROX Deep Red for 30 min and then harvested. The fluorescence intensity of the cells was measured using a FACSVerse flow cytometer (BD Biosciences).

After fixation with 4% paraformaldehyde, cultured human neutrophils were stained with the mitochondrionspecific dye, MitoTracker Red CMXRRos (Invitrogen), according to the manufacturer’s instruction. The cells were immediately analyzed using a FACSVerse flow cytometer.

Unless otherwise stated, all data are presented as the mean ± SD. Within individual experiments, data points were based on a minimum of triplicate representative samples and experiments were repeated at least three times.

After isolating human neutrophils from peripheral blood preparations (see Materials and Methods), the cells (4.5 × 10⁶ cells) were incubated at 4˚C for 1 h followed by incubation at 37˚C for 5 h in the 6-cm culture dish supplemented with 2 ml of the culture medium, in which a coverslip was immersed (Figure 1(a)). We found that most of the cells were present as non-adherent forms, and were thus floating in the culture medium. These non-adherent cells appeared morphologically intact as shown by Wright–Giemsa staining (Figure 1(b), left panel). In contrast, we unexpectedly found that small, but substantial, numbers of Wright-Giemsa-stainable materials, which looked different from intact neutrophils, were present on the coverslip (Figure 1(b), right panel).

When the materials on the coverslip were stained with DAPI without any fixative treatment, we observed bright fluorescent signals under fluorescent microscopy, many of which appeared to consist of multiple DAPI-positive strings (Figure 1(c)). Because DAPI is a fluorescent dye that intercalates into double-stranded DNA, and that living neutrophils are less permeable to the dye than dead neutrophils, we thought it probable that these bright DAPI-stained signals represented global DNA unfolding of dead neutrophils, which somehow adhered to the coverslip. It should be mentioned that there were few neutrophils with normal morphology on the coverslip per view field, implying that most of neutrophils floated in the culture medium under the standard culture conditions.

We then investigated whether the requirement for DAPI-positive object production was simple exposure to hypothermia or rather the combination of hypothermia/ rewarming. When human neutrophils were maintained at a constant temperature of either 4˚C or 37˚C, significantly less DAPI-positive signals were detected as compared with cells cultured either at 4˚C, 15˚C, or 25˚C for 1 h followed by incubation at 37˚C for 5 h (Figure 1(d)). These results suggest that the appearance of the DAPIpositive objects was associated with incubation of neutrophils under hypothermal conditions followed by rewarming.

When we observed the DAPI-positive objects in hypothermia/rewarming-treated human neutrophils, we noticed that morphological similarities between the objects and DAPI-stained PMA-stimulated NETs, leading us to suspect that the DAPI-positive objects per se might represent NETs (Figure 2(a)). To investigate this possibility, we first asked whether the that DAPI-positive objects possessed the ability to bind bacteria. Given NETs are defined as extracellular DNA-proteinaceous structures that exhibit the ability to associate with a wide variety of Gram-positive and Gram-negative pathogens [7], we expected that the DAPI-positive structures might also show similar properties. As shown in Figure 2(b), when GFP-expressing E. coli was incubated with the coverslip containing DAPI-positive objects, we found multiple GFP signals present together with the DAPI-signals. Their ability to trap bacteria appeared equivalent to that of PMA-stimulated NETs, because the number and distribution of GFP-signals associated with the DAPI-positive objects were very similar to those associated with PMA-stimulated NETs, suggesting that the objects possessed the ability to trap bacteria. It should be mentioned that the number of bacteria trapped to the DAPI-positive objects was reduced when the coverslips were treated with DNase I (Figure 2(c)). Similar results were obtained when PMA-stimulated NETs were treated with DNase I. These results imply that both structures are equally susceptible to DNase I treatment with respect to bacterial trap.

To further evaluate the similarities between the DAPIpositive objects and canonical NETs, we performed indirect-immunofluorescence analysis using antibodies that recognize marker proteins for NETs: anti-neutrophil

elastase (NE) and anti-histone H3 antibodies [7]. As shown in Figures 2(d) and (e), in the presence of these anti-bodies, the signals were detected not only on PMAstimulated NETs but also on the DAPI-positive objects. In contrast, in the absence of the antibodies, these signals were barely detected, suggesting existence of NE and histone H3 on both the DAPI-positive objects and canonical NETs. Taken together, at least with regard to their ability to trap bacteria and the existence of two marker proteins for NETs, our results indicated that the DAPI-positive objects possessed biochemical similarities to canonical NETs.

It should be noted, however, the experiments described above are not sufficient to conclude that the DAPI-positive objects have anti-bacterial activity. We are therefore investigating whether NE and histones on the DAPIpositive objects can indeed inactivate bacterial toxic proteins, called “virulence factors,” and inhibit bacterial growth. In addition, we wish to test whether the DAPIpositive objects can capture microorganisms besides Gram-negative bacteria (E. coli), such as fungi and parasites.

Although our results so far indicated a correlation between DAPI-positive objects and NETs, several differences were also revealed. For instance, when indirect immunofluorescence analysis was conducted using antihistone H3 citrulline R26 (anti-H3cit) antibody, we found that the antibody stained many, but not all, PMAstimulated NETs, whereas the antibody proved poor at detecting the DAPI-positive objects (Figure 3(a)). Because it has been reported that peptidylarginine deiminase 4 (PAD4/PADI4), which catalyzes histone hy percitrullination, mediates chromatin decondensation and is vital to NET formation [13,20,21], our observation of different staining patterns with anti-H3cit antibody between the DAPI-positive objects and PMA-stimulated NETs implies activation of PAD4 in PMA-stimulated cells, while the enzyme might not be fully activated in the hypothermia/rewarming-treated cells.

We also detected differences between the objects and NETs in TUNEL assays. As shown in Figure 3(b), TUNEL-positive signals were negligible in PMA-treated cells, confirming the previous report that no TUNELpositive DNA damage is detectable in canonical NETs [15]. In contrast, TUNEL assay visualized more than 90% of the hypothermia/rewarming-induced DAPI-positive objects, indicating the existence of multiple sites of TUNEL-positive DNA damage on extensively unfolded DNAs in the objects. Because TUNEL-positive signals are frequently associated with apoptotic cells, these data indicated that hypothermia/rewarming-incubation might trigger, to some extent, activation of apoptosis-related DNA cleavage enzyme(s) in neutrophils, suggesting the possible contribution of apoptotic signaling pathways, at least in part, to DNA structural alterations during the formation of the DAPI-positive, NET-like structures.

Given that ROS generation is an absolute requirement for the formation of NETs [15,16], we next assessed whether hypothermia/rewarming of neutrophils coincided with the generation of ROS. Thus, we measured ROS in hypothermia/rewarming-treated human neutrophils. As shown in Figure 4(a), during constant temperature incubation at either 4˚C or 37˚C, no increase/decrease in ROS was detected. In contrast, we found a significant elevation of ROS when cells were kept at 4˚C followed

by incubation at 37˚C for 1 h or 3 h; an approximate 10- fold increase in ROS was apparent in the cells after incubation at 37˚C.

We next examined whether NADPH oxidase contributed to ROS production in hypothermia/rewarmingtreated cells. Given that PMA-induced NET formation is effectively inhibited by DPI, an inhibitor of NADPH oxidase activity [15,17,18 and see Figure 4(b)], we tested the effect of this drug. Intriguingly, neither ROS generation nor the formation of DAPI-positive objects was significantly affected by administration of DPI (Figures 4(a) and (b)), indicating that the mechanism of ROS generation in hypothermia/rewarming-treated neutrophils could be pharmacologically distinguished from that in PMA-stimulated cells with respect to the involvement of NADPH oxidase in ROS generation.

Although where and how ROS are produced in the hypothermia/rewarming-treated cells remains unclear, it is noteworthy that the mitochondrion-specific dye, MitoTracker Red, detected structural and/or functional alterations in mitochondria in the hypothermia/rewarmingtreated cells (Figure 4(c)). Because mitochondria are known to produce ROS in aerobic organisms, including humans [22], these data suggest scenario that this organelle may generate ROS during hypothermia/rewarming, leading to the formation of the DAPI-positive, NET-like structures. However, DPI is also known as a potent inhibitor of mitochondrial reactive oxygen species production [23]. If DPI inhibits generation of ROS from both mitochondria and the NADPH oxidase pathway during hypothermia/rewarming of neutrophils, we need to consider the possibility of an alternative pathway to generate ROS, besides the NADPH oxidase pathway or the mitochondria pathway.