SK3 channel and mitochondrial ROS mediate NADPH oxidase-independent NETosis induced by calcium influx

SK3 channel and mitochondrial ROS mediate NADPHoxidase-independent NETosis induced by calcium influxDavid Nobuhiro Doudaa,b,1,2, Meraj A. Khana,1, Hartmut Grasemannc,d, and Nades Palaniyara,b,d,3

aInnate Immunity Research Laboratory, Program in Physiology and Experimental Medicine, SickKids Research Institute and cDepartment of Paediatrics,Division of Respiratory Medicine, The Hospital For Sick Children, Toronto, ON, Canada M5G 1X8; and bDepartment of Laboratory Medicine and Pathobiologyand dInstitute of Medical Sciences, University of Toronto, Toronto, ON, Canada M5S 1A8

Edited by Lily Yeh Jan, University of California, San Francisco, CA, and approved January 15, 2015 (received for review August 13, 2014)

Neutrophils cast neutrophil extracellular traps (NETs) to defend thehost against invading pathogens. Although effective against micro-bial pathogens, a growing body of literature now suggests that NETshave negative impacts on many inflammatory and autoimmunediseases. Identifying mechanisms that regulate the process termed“NETosis” is important for treating these diseases. Although two ma-jor types of NETosis have been described to date, mechanisms regu-lating these forms of cell death are not clearly established. NADPHoxidase 2 (NOX2) generates large amounts of reactive oxygen species(ROS), which is essential for NOX-dependent NETosis. However, majorregulators of NOX-independent NETosis are largely unknown. Herewe show that calcium activated NOX-independent NETosis is fast andmediated by a calcium-activated small conductance potassium (SK)channel member SK3 andmitochondrial ROS. AlthoughmitochondrialROS is needed for NOX-independent NETosis, it is not important forNOX-dependent NETosis. We further demonstrate that the activationof the calcium-activated potassium channel is sufficient to induceNOX-independent NETosis. Unlike NOX-dependent NETosis, NOX-independent NETosis is accompanied by a substantially lower level ofactivation of ERK and moderate level of activation of Akt, whereasthe activation of p38 is similar in both pathways. ERK activation isessential for the NOX-dependent pathway, whereas its activation isnot essential for the NOX-independent pathway. Despite the differ-ential activation, both NOX-dependent and -independent NETosis re-quire Akt activity. Collectively, this study highlights key differences inthese two major NETosis pathways and provides an insight into pre-viously unknown mechanisms for NOX-independent NETosis.

neutrophils | neutrophil extracellular traps | NETosis | NADPH oxidase |SK channels

Neutrophils are the first responders to invading pathogensand play a pivotal role in innate immune defense. They are

present in large numbers in circulation and transmigrate to thesites of infection in response to chemotactic signals. Once at thesite of infection, the activated neutrophils fight pathogens byvarious means, including oxidative burst, phagocytosis, and theformation and release of neutrophil extracellular traps (NETs) ina process termed “NETosis” (1). NETs are made of DNA andare decorated with antimicrobial proteins and peptides (2, 3).To date, two distinct forms of NETosis have been described

based on their requirement for NADPH oxidases 2 (NOX2). Inthe NOX-dependent pathway, pharmacological inhibition ofNOX2 results in a inhibition of reactive oxygen species (ROS)and subsequent NET formation (1, 4). Furthermore, neutrophilsisolated from patients with chronic granulomatous disease(CGD) fail to form NETs for certain stimuli because these cellshave deficiencies in NOX-mediated ROS production (5). There-fore, the generation of ROS by NOX enzyme complex is con-sidered to be crucial for NETosis. In contrast, several reports haveshown the existence of NOX-independent NETosis that canbe induced by certain other stimuli including calcium ionophore(6, 7). Although we are beginning to understand the molecularpathways governing NOX-dependent NETosis, the mechanismsfor the NOX-independent pathway of NETosis are not wellunderstood.

It has been shown that the calcium-activated potassium channelof small conductance (SK channel), which is the major calcium-activated potassium channel known to be present on neutrophils(8–11), mediates NOX-independent neutrophil “apoptosis” (11).In this pathway, SK channels activate mitochondrial ROS pro-duction. Interestingly, neutrophils exhibit potassium current acti-vated by calcium influx, and potassium has been shown to beimportant for neutrophil-mediated killing of several microbialpathogens (8, 9, 12, 13). However, the involvement of potassiumchannels and bacterial killing in the context of NETs has notbeen examined.In this study, we set out to investigate the role of mitochondrial

ROS in the calcium ionophore-mediated NETosis. We show thatcalcium ionophore-activated NETosis is fast, NOX-independent,and shows greatly reduced activation of ERK and a moderatelevel of activation of Akt, whereas the activation of p38 is similarto that of the NOX-dependent pathway. The investigation of themechanism of the NOX-independent pathway revealed that thispathway requires mitochondrial ROS production. Furthermore,we also show that activation of the SK channel is both necessaryand sufficient to induce NETosis. Therefore, we reveal a previouslyunidentified calcium-induced mitochondrial ROS-dependent butNOX-independent NETosis pathway.

Significance

Formation of neutrophil extracellular traps (NETs) is a recentlydescribed process by which neutrophils combat microbial patho-gens. Recent studies demonstrate causative relationships betweenNETs and debilitating disorders such as rheumatoid arthritis, vas-culitis, thrombosis, cystic fibrosis, and acute respiratory distresssyndrome. However, the understanding of signaling pathwaysgoverning the process termed “NETosis” remains elusive. Twomajor types of NETosis have been reported; however, the mech-anistic differences between these two types are not clearlyestablished. Here we describe that NETosis induced by calciumionophores is fast, NADPH-oxidase independent, and is mediatedby mitochondrial reactive oxygen species (ROS) and a calcium-activated small conductance potassium channel. Thus, drugs thattarget mitochondrial ROS production or the potassium channelsmay provide previously unidentified therapeutic approaches forcombating disorders with unregulated NETosis.

Author contributions: D.N.D., M.A.K., and N.P. designed research; D.N.D., M.A.K., and N.P.performed research; D.N.D., M.A.K., and N.P. analyzed data; and D.N.D., M.A.K., H.G., andN.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1D.N.D. and M.A.K. contributed equally to this work.2Present address: Pulmonary and Critical Care Medicine Division, Brigham and Women’sHospital and Harvard Medical School, Boston, MA 02115.

3To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1414055112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1414055112 PNAS Early Edition | 1 of 6

IMMUN

OLO

GYAND

INFLAMMATION

ResultsThe Calcium Ionophore-Induced NOX-Independent NETosis Is Distinctfrom NOX-Dependent NETosis. Much of our current understandingof the molecular mechanisms that govern the NOX-dependentpathway results from studies using phorbol 12-myristate 13-acetate(PMA) (14–16). The use of pharmacological agonists such as PMAas the NOX-dependent NET inducer is attractive due to its abilityto uniformly activate the cells in culture. To elucidate mechanismsof NOX-independent NETosis, we used a calcium ionophore,A23187. Furthermore, we used PMA to make a side-by-side com-parison of the two pathways. To observe the kinetics of NET releaseinduced by the two agonists, a plate reader assay was used. Theplate reader assay is routinely used for monitoring NET DNArelease (4, 6, 17), and it detects extracellular NETs with a cell im-permeable, extracellular DNA dye Sytox Green. The assays show thatA23187 activates human neutrophils to release NETs (Fig. 1A). Thetime course analysis of NET release also confirms that the NETosisinduced by PMA and A23187 follows distinct kinetics (Fig. 1A).NETosis induced by certain strains of Staphylococcus aureus has

been shown to be both NOX dependent and independent (6, 7).Nevertheless, NETosis induced by the RN4220 strain of S. aureuswas inhibited by diphenyleneiodonium (DPI), and thus, NOX de-pendent (Fig. S1). Therefore, we have tested another NETosis-inducing agonist, ionomycin, a natural calcium ionophore produced

by a different Gram-positive bacteria Streptomyces conglobatus. Therapid NETosis induced by A23187 was also observed in NETosisinduced by ionomycin (Fig. S2A). Therefore, we used ionomycin asthe natural agonist of NOX-independent NETosis for the re-mainder of the study.Next, the requirement for the presence of extracellular calcium

was assessed inNETosis induced by PMA,A23187, and ionomycin.The plate reader assays show that the activation of neutrophils withA23187 or ionomycin in the absence of extracellular calcium resultsin a significant reduction of NETosis (P< 0.001; Fig. 1 B andC andFig. S2B). In contrast, the activation of cells with PMA in the ab-sence of extracellular calcium does not result in a significant re-duction in NETosis (Fig. S2B). To verify that NETs are beingformed by A23187- and ionomycin-activated cells, an immunoflu-orescence assay was performed. The immunofluorescence imagesconfirm that the cells activated with the calcium ionophores releaseNETs as evidenced by the colocalization of extracellular DNAwithmyeloperoxidase (MPO) (Fig. 1D and Fig. S2C). Taken together,these data show that NETosis induced by the calcium ionophoresA23187 and ionomycin is distinct from that induced by PMA.We next asked whether NETosis induced by the calcium ion-

ophores required ROS production by the NOX2 enzyme. To assessthe importance of ROS, we first determined whether ROS is pro-duced during PMA- and A23187-induced NETosis. To detect cyto-solic ROS produced by NOX2, we used dihydrorhodamine (DHR)123, a fluorescent indicator of cytosolic ROS. Confocal imaging ofthe cells loaded with DHR123 and activated with PMA or A23187shows that the activation of cells with PMA leads to an abundantproduction of cytosolic ROS, whereas A23187-activated neutrophilsproduce very little cytosolic ROS (Fig. 2 A–C). Quantitative analysisusing flow cytometry confirms the above observation (Fig. 2D).Similar results were obtained with a plate reader assay usingDHR123 (Fig. 2 E and F). As expected, ROS production in PMA-activated neutrophils is abolished when these cells were pre-incubated with DPI, a NOX inhibitor, and then activated withPMA (Fig. 2 D and E).To directly assess the requirement for NOX2 in PMA- or

calcium ionophore-induced NETosis, plate reader assays wereperformed in the presence or absence of DPI. These assays showthat NETosis induced by PMA is NOX dependent, as previouslysuggested (1, 4), and that the inhibition of ROS production byDPI results in a significant suppression of NETosis induced byPMA (P < 0.001; Fig. 2G). In contrast, A23187- or ionomycin-induced NETosis is less sensitive to suppression by DPI and doesnot result in a significant reduction in NET release (Fig. 2 Hand I, respectively). Therefore, calcium ionophore-induced NOX-independent NETosis is distinct from PMA-mediated NOX-dependent NETosis in terms of its lack of sensitivity to DPI.

Differential Activation of Kinases in NOX-Dependent and -IndependentNETosis. We next sought to determine differences in the signalingpathways that are activated in NOX-dependent and -independentNETosis. Number of key kinases such as ERK (14), Akt (15), andp38 (16) have been shown to be activated in PMA-inducedNETosis. Thus, we next assessed differences in the activation ofthese kinases during PMA- and A23187-induced NETosis overa 120-min period following the activation of neutrophils (Fig. 3).Because the presence of citrullinated histone has previously beenshown to be specific to A23187-induced NETosis (18, 19), we firstassessed the citrullination of histone H3 in both conditions asa quality control (Fig. 3A and Fig. S3 A and B). Consistent withprevious reports (18, 19), hypercitrullination of histone H3 wasonly observed in the calcium ionophore-mediated NETosis.Immunoblot analyses of thekinases involved in thePMA-induced

NETosis show that ERK (Fig. 3B and Fig. S3C) and Akt (Fig. 3Cand Fig. S3D) are strongly activated during the NOX-dependentpathway as expected. In contrast, A23187-induced NETosis isaccompanied by drastically low activation of ERK (Fig. 3B andFig. S3C) and moderate reduction in activation of Akt (Fig. 3Cand Fig. S3D) compared with the NOX-dependent NETosis. Theactivation of p38, however, was similar in both pathways (Fig. 3Dand Fig. S3E). Together, these results highlight that, while similar

Fig. 1. Calcium ionophores induce rapid NETosis. (A) NET release in responseto A23187 or PMA was measured using a plate reader assay (n = 5). A23187-and PMA-mediated NETosis in human neutrophils follow different kinetics ofNET release. Ionophore-induced NETosis is faster than that of PMA-inducedNETosis. (B and C) Neutrophils were activated with A23187 (B) or ionomycin(C) in the presence or absence of calcium. NETosis was measured using a platereader assay. NET release is expressed as percentage of total DNA (n = 3).Calcium ionophore-induced NETosis requires extracellular calcium. (D) Neu-trophils were incubated in the presence or absence of A23187 (4 μM) or ion-omycin (5 μM) for 300 min. Cells were stained for DNA (green) and MPO (red).Immunofluorescence imaging shows that the calcium ionophore A23187 andionomycin induce NETosis. (Scale bar, 10 μm.) Images are representative ofthree independent experiments. A23, A23187; Ca2+, calcium chloride; Io, ion-omycin. *P < 0.05; **P < 0.01; ***P < 0.001 (A–C, two-way ANOVA).

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1414055112 Douda et al.

kinases are activated in both of these pathways, differences do existin the extent of activation of the signaling kinases involved inNOX-dependent and NOX-independent pathways of NETosis.

Pathway-Dependent Requirement for Kinases During NETosis. Wenext asked whether ERK, Akt, and p38 are required for the

NOX-independent NETosis. To do so, we used ERK inhibitorFR180204, Akt inhibitor XI, or p38 inhibitor SB202190, andperformed plate reader assays using differentiated HL-60 humanneutrophil-like (dHL-60) cells. The results show that PMA-induced NETosis is significantly reduced in the presence of 10 or20 μM ERK inhibitor FR180204 (IC50 = 300–510 nM; P < 0.01;Fig. S3F) and 5 or 10 μM Akt inhibitor XI (IC50 = 140–310 nM;P < 0.001; Fig. S3H), whereas no change in the NET release isobserved in the presence of 1 or 2 μM p38 inhibitor SB202190(IC50 = 50 and 100 nM; Fig. S3J). On the other hand, A23187-induced NETosis is significantly reduced only in the presence ofAkt inhibitor XI (P < 0.01; Fig. S3I), but not ERK or p38 in-hibitor (Fig. S3 G and K, respectively). These results suggest thatboth NOX-dependent and NOX-independent pathways of NETosisare mediated by the activation of Akt, and the NOX-dependentpathway also requires the activation of ERK.

Mitochondrial ROS Is Required for NOX-Independent NETosis. DPI isan inhibitor of NOX, and the above results show that A23187-induced NETosis is NOX independent. It is also known that DPIcan partially inhibit mitochondrial ROS production (20, 21).Furthermore, the inhibition of NOX by DPI does not rule outthe involvement of ROS from other sources. It is well establishedthat mitochondrial respiration produces ROS (22). Thus, we nextasked whether the mitochondrial ROS was produced when cellswere activated with PMA or A23187. We used MitoSOX, a mi-tochondrial ROS-specific fluorescent dye to determine the mi-tochondrial ROS production. The plate reader assays show thatalthough there is no substantial mitochondrial ROS productionin PMA-activated neutrophils, a significantly higher mitochondrialROS production occurred in A23187-activated cells (Fig. 4 A andB). The involvement of mitochondria in ROS production wasconfirmed using dinitrophenol (DNP), a mitochondrial uncoupler;the preincubation of cells with DNP completely abolished mito-chondrial ROS production (Fig. 4A). In addition to DNP, wealso tested another mitochondrial uncoupler, carbonyl cyanide4-(trifluoromethoxy) phenylhydrazone (FCCP), and monitoredA23187-induced NETosis in dHL-60 cells (Fig. S4). The in-hibition of mitochondrial ROS production by DNP leads toa significant (P < 0.001; Fig. S4A) and dose-dependent (P < 0.05;Fig. S4B) reduction in A23187-induced NET release. Pre-incubation of dHL-60 cells with FCCP also yields a significant(P < 0.001; Fig. S4C) and dose-dependent (P < 0.01; Fig. S4D)reduction in A23187-induced NETosis. Together, these results

Fig. 2. Calcium ionophore-mediated NETosis is NOX2 independent. (A–D) Hu-man neutrophils were loaded with cytosolic ROS indicator DHR123 and activatedwith PMA or A23187. Live cell fluorescence image analysis of DHR123 loaded cellstreated with control buffer (A), PMA (B), or A23187 (C). The presence of cytosolicROS is indicated by the green fluorescence, and the cells were counterstainedwith Hoechst 33342 live cell DNA stain (blue) (n = 4). (Scale bar, 20 μm.) (D) Re-presentative flow cytometric analysis of cytosolic ROS production in neutrophilsloaded with DHR123 and activated with either PMA or A23187, in the presenceor absence of DPI (n = 3). (E and F) DHR123-based ROS detection plate readerassays show that PMA induces a significantly greater cytosolic ROS productioncompared with A23187. (G–I) Neutrophils were activated by PMA, A23187,or ionomycin in the presence or absence of DPI (20 μM) and NET release wasmeasured using a plate reader assay. The results are expressed as percentage ofreduction in the presence of DPI compared with the activating agonist alone. DPIsignificantly reduces PMA-mediated NETosis (G), whereas it fails to inhibit A23187(H)- and ionomycin (I)-mediated NETosis (n = 3). DPI → PMA, preincubation withDPI and activated with PMA; A23, A23187; DPI → A23, preincubation with DPIand activated with A23187; Io, ionomycin; DPI → Io, preincubation with DPI andactivated with ionomycin. *P < 0.05 (E, two-way ANOVA; F–I, Student’s t test).

Fig. 3. Immunoblot analysis. (A–D) Human neutrophils were harvested afteran activation with PMA, A23187, or negative control (-ve) for the indicatedtimes. The negative (-ve) control samples were incubated at 120 min in theabsence of any activator. (A) Immunoblots show that A23187, but not PMA,induces citrullination of histone H3 (citH3) (n = 4). (B–D) The activation ofkinases was assessed by immunoblotting for phospho (p)-ERK (B, n = 3), p-Akt(C, n = 3), and p-p38 (D, n = 2). Total kinases t-Akt, t-ERK, and t-p38 were usedas loading controls. ERK is highly activated during PMA-induced NETosis. Akt isactivated in both forms of NETosis, albeit a moderate level of activation inA23187-induced NETosis. p38 is equally activated in both forms of NETosis.

Douda et al. PNAS Early Edition | 3 of 6

IMMUN

OLO

GYAND

INFLAMMATION

suggest that mitochondrial ROS generation is required forcalcium ionophore-induced NOX-independent NETosis, but notfor PMA-induced NOX-dependent NETosis. Therefore, we nextasked whether mitochondrial ROS production was required forNETosis in human neutrophils. The plate reader assays show thatA23187-induced NETosis is significantly attenuated by the pres-ence of DNP in a dose-dependent manner (P < 0.05; Fig. 4C andFig. S4E). In contrast, PMA-induced NETosis in the presenceof DNP is not reduced significantly even at the highest dosetested (Fig. 4D and Fig. S4F). Collectively, these data show thatmitochondrial ROS is important for NOX-independent, but not forNOX-dependent NETosis.

Activation of the SK Channel Is Required for NOX-Independent NETosis.One report has shown that the activation of neutrophils withionomycin induces cell death mediated by the SK channel activity(11). We therefore asked whether NETosis accounted for the SKchannel-mediated cell death reported by Fay et al. (11). We firsttested whether the inhibition of SK channels would inhibitNETosis induced by the calcium ionophores. The plate readerassays show that NET release in the NOX-independent pathway issignificantly reduced in the presence of a SK channel inhibitorNS8593 in a time- and concentration-dependent manner (P < 0.001;Fig. 5A and Fig. S5A). In contrast, inhibition of the SK channel byNS8593 had no inhibitory effect on the PMA-induced, NOX-dependent NETosis; if any, NS8593 increased the NETosis (Fig. 5Band Fig. S5B). We next tested the requirement of the SK channel inthe NOX-independent pathway using another SK channel inhibitor,apamin. The results show that the NETosis induced by both A23187(P < 0.01; Fig. 5C and Fig. S5C) and ionomycin (P < 0.001; Fig. 5Dand Fig. S5D) is significantly inhibited by apamin.We next sought to identify the specific family member respon-

sible for mediating the NOX-independent NETosis. Of the apamin-sensitive SK channels (SK1–SK3), the expression of SK1 is limitedto the neuronal tissue (23). The use of scorpion toxin, scyllatoxin,

which has the strongest affinity to SK2, suggests that SK2 is notinvolved (Fig. S5 E and F). To further test whether the transientreceptor potential melastatin (TRPM)7 channel could be in-volved in NETosis, we used a TRPM7 inhibitor, MK886, whichdid not inhibit NETosis (Fig. S5 G and H). Therefore, we consid-ered SK3 channel as effectors of NOX-independent NETosis.Furthermore, previous studies show that neutrophils predominantlyexpress SK3 (11). Hence, we sought to directly assess whetherSK3 is required for NOX-independent NETosis. To do so, weused siRNA against KCNN3 (gene encoding SK3 channel) tran-script in dHL-60 cells. After optimizing and confirming a hightransfection efficiency of siRNA in our system (Fig. S6A), wevalidated the successful knockdown of the SK3 channel proteinexpression with immunoblot assays (Fig. 6A). We then performedplate reader assays, and the results show that the knockdown ofSK3 channels significantly reduces NET release in both A23187-and ionomycin-activated cells (Fig. 6A and Fig. S6 B and C).Together, these results show that NOX-independent NETosisrequires the activation of the SK3 channel.We next asked whether the activation of the SK channel could

drive NETosis. To do so, we used a SK channel-specific activator,1-Ethyl-2-benzimidazolinone (EBIO). The plate reader assaysshow that the potassium channel activator 1-EBIO induces NETosisin human neutrophils (Fig. 6B). In 1-EBIO–mediated NETosis, therelease of NETs occurs immediately after the addition of this SKchannel agonist (1-EBIO), and increases over time (P < 0.05; Fig.6B). The formation of NETs was then validated with immunofluo-rescence assay showing colocalization of extracellular DNAand MPO (Fig. 6C). Together, these results demonstrate that theactivation of potassium channels is necessary and sufficient forthe induction of NETosis.

DiscussionNETosis can occur via NOX-dependent and NOX-independentpathways (6, 7, 24). However, the mechanism by which the

Fig. 4. A23187-mediated NETosis requires mitochondrial respiration. (A and B)Human neutrophils were loaded with a mitochondrial ROS indicator, MitoSox,and activated with PMA or A23187. A time-course fluorescence plate readerassay for MitoSOX fluorescence shows a larger mitochondrial ROS productionin cells activated by A23187 as opposed to PMA-activated cells. A mitochondrialuncoupler, DNP, abolishes mitochondrial ROS production induced by A23187(n = 3). (B) MitoSOX assays show that A23187-activated cells produce a sig-nificantly higher levels of mitochondrial ROS at the peak production at the25-min time point, compared with the cells activated by PMA (n = 3).(C ) A23187-mediated NETosis is reduced in the presence of DNP (750 μM)as shown by time course analysis (n = 7). (D) PMA-mediated NETosis is notsignificantly reduced in the presence of DNP (750 μM) as shown by timecourse analysis (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001 compared withthe agonist alone (A, C, and D, two-way ANOVA; B, Student’s t test).

Fig. 5. SK channel is required for ionophore-mediated NETosis. (A) Humanneutrophils were activated with A23187 in the presence of SK channel inhibitorNS8593 (100 μM). The plate reader assay shows that NETosis is significantly re-duced in the presence of NS8593 (n = 3), suggesting that A23187-induced NETDNA release in human neutrophils requires the activation of the SK channel.(B) Neutrophils were activated with PMA in the presence of NS8593. The platereader assays show that NETosis is not inhibited in the presence of NS8593 (100 μM;n = 3), indicating that PMA-mediated NETosis does not require the acti-vation of the SK channel. (C and D) Human neutrophils were activated withA23187 (C) or ionomycin (D) in the presence of SK channel inhibitor apamin(200 nM; n = 3). The plate reader assays show that NETosis is reduced significantlyin the presence of apamin; hence, A23187- and ionomycin-induced NETosisrequires the activation of the SK channel. A23, A23187; Io, ionomycin. *P < 0.05;**P < 0.01; ***P < 0.001 compared with the agonist alone (two-way ANOVA).

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1414055112 Douda et al.

NOX-independent NETosis occurs is not well understood. Inthis study, we show that the calcium ionophore-induced NOX-independent NETosis occurs more rapidly compared with theNOX-dependent pathway induced by PMA. Activation of ERKand Akt was low in NOX-independent NETosis compared withNOX-dependent NETosis; however, the activation of p38 wassimilar in both pathways. ERK activation is required for NOX-dependent NETosis, whereas Akt activation is essential forboth types of NETosis. We also show that the NOX-independentNETosis requires mitochondrial ROS. Furthermore, the NOX-independent NETosis requires activation of SK3 channel, and di-rect activation of the SK channel is sufficient to induce NETosis.The induction of NETosis by the calcium ionophores was first

suggested by Neeli and Radic in a study that investigated the role ofvarious agonists in activating peptidyl arginine deiminase 4 (PAD4)(19). Because PAD4 is necessary for the induction of NETosis(25, 26), citrullinated histone has been used as a marker for NET-associated histones in vivo (17, 18, 27). However, the requirementfor PAD4 is not universal to all NETotic pathways, as evidencedby the fact that there is a lack of PAD4 activity and histone hyper-citrullination in PMA-mediated NETosis (18). Our work shows thatthe kinetics of calcium ionophore-mediated NETosis is in starkcontrast to thePMA-inducedNETosis (Fig. 1), which is completelydependent on NOX-mediated ROS production (Fig. 2). We alsoprovide proof that the calcium ionophore does not require NOX-activity to induce NETosis (Fig. 2).It has recently been shown that BK channel activity is absent in

neutrophils and is not required for antimicrobial defense (10,28). 1-EBIO is an activator of both calcium-activated potassiumchannel of intermediate conductance (IK) and SK channels (29);however, 1-EBIO does not activate IK channels in human neu-trophils (11). Our data show that activation of neutrophils by1-EBIO induces NETosis (Fig. 6). The use of specific SK channelinhibitors such as NS8593 and apamin confirms that NETosis is

mediated through the SK channels (Fig. 5 and Fig. S5). Knock-down experiments using siRNA further show that SK3 channelmediates NOX-independent NETosis (Fig. 6). Thus, it is possiblethat the SK channel-dependent NOX-independent NETosisis a contributor to the previously described potassium channel-mediated antimicrobial activity of neutrophils (8, 12).Previous studies suggest the existence of NOX-independent

NETosis, as demonstrated by the inability of DPI to suppress NETrelease (30, 31). Our current work shows that the NOX-independentpathway uses mitochondrial ROS (Fig. 4). Large amount of mi-tochondrial ROS is being produced during NOX-independentNETosis, but not during NOX-dependent NETosis (Figs. 2 and 4).DuringNOX-dependentNETosis, production ofNOX-dependentROS is required for the activation of ERK, Akt, and p38 (15, 16).ERK has been implicated as the major kinase involved in NOX-dependent NETosis (14, 16). Our data are consistent with this no-tion (Fig. 3 and Fig. S3). In contrast, ERK is not substantially acti-vated in NOX-independent NETosis, and its inhibition does notinhibit this form of NETosis. Although p38 was activated in bothtypes of NETosis, p38 inhibitor SB202190 did not inhibit NETosis.Consistent with our previous report (15), we show that Akt is re-quired for NOX-dependent NETosis (Fig. 3 and Fig. S3). In thisreport, we show that Akt is also required for the NOX-independentpathway. The exact roles of kinases and ROS in NETosis, how-ever, are not clearly understood. Previous studies show that pro-tein tyrosine phosphatases are regulatory targets of ROS (32),which is a second messenger necessary to inactivate protein tyro-sine phosphatase or MAP kinase phosphatase to induce the acti-vation of ERK in ML-1 cells (33). Thus, it is possible that a simi-lar mechanism is at play for the induction of ROS-mediatedNETosis, whether it is NOX dependent or independent.There have been numerous reports describing the cross-talk be-

tween mitochondrial superoxide and NOX activity (34–36). In en-dothelial cells, mitochondrial superoxide can stimulate cytoplasmicNOX (36). Our results show that although mitochondrial su-peroxide is generated by the calcium ionophore A23187, it doesnot lead to the generation of cytoplasmic ROS by NOX, assuggested by the DHR123 staining (Fig. 2). Conversely, there isvery little, if any, mitochondrial ROS being generated in thePMA-induced NOX-dependent pathway (Fig. 4). These datasuggest very little interplay between mitochondrial superoxideand NOX activity during NETosis.Overall, mitochondrial respiration and potassium current at

the mitochondrial level are tightly linked (37). Recent reportsdemonstrate the presence of the mitochondrial SK channel inneuronal cell types and show that potassium influx into mito-chondria causes changes in the membrane potential (38, 39).Although such mitochondrial potassium channels have not beenidentified in neutrophils, our results suggest that activation of SKchannels leads to the production of mitochondrial ROS in neu-trophils (Fig. 4). Moreover, although apamin is a cell-impermeableinhibitor peptide, it exerts an inhibitory effect on NETosis thatsuggests the engagement of cell surface SK3 in this process. Ithas been shown that changes in cytosolic potassium levels canchange potassium flux and mitochondrial respiration (39). Fur-thermore, a unique feature of the SK channel complex is that itcontains a calmodulin-binding domain. This calcium sensor isresponsible for the conformational change and the channel gateopening (40, 41). The entire protein complex of the SK channelincludes two other regulators—protein kinase CK2 and proteinphosphatase 2A (42). These two proteins regulate the calciumsensitivity of the calmodulin subunit, whereby the phosphoryla-tion of calmodulin at tyrosine 80 enhances its sensitivity to calcium(42). Therefore, NETosis regulated by calcium-induced potassiumchannels could occur via different regulatory events.In this study, the depletion of extracellular calcium led to a sig-

nificant, but partial decrease in the A23187- and ionomycin-induced NOX-independent NETosis (Fig. 1 and Fig. S2). How-ever, the calcium ionophores also induce the release of storedcalcium (43, 44). Thus, it is likely that release of stored calciumleads to a partial activation of the NOX-independent pathway.Similarly, the effect of the mitochondrial uncoupler, DNP, also

Fig. 6. SK3 is required for the NOX-independent NETosis. (A) The plate readerassay shows thatA23187- or ionomycin-mediatedNETosis is significantly reduced insiRNA (si)-transfected dHL-60 cells, compared with scrambled RNA (sc)-transfectedcontrol cells at240minpostactivation (n=3). Shownabove thegraph is the resultofan immunoblot analysis illustrating a successful knockdown of SK3 protein expres-sion after siRNA transfection (n = 3). Therefore, knockdown of the SK3 channel bysiRNA significantly reduces calcium ionophore-induced NETosis. (B and C) SK chan-nel activator 1-EBIO inducesNETosis. (B) The plate reader assayswere performed tomonitor NET release over time in response to the activation with 1-EBIO (n = 5).(C) Immunofluorescence staining confirms that compared with DMSO control, theactivation of potassium channels by 1-EBIO induces NET release as shown by thecolocalization of DNA (green) and MPO (red) after 300 min. sc, scrambled RNA-transfected cells; si, SK3 siRNA-transfected cells. *P < 0.05; **P < 0.01; ***P < 0.001compared with the agonist alone (A, one-way ANOVA; B, two-way ANOVA).

Douda et al. PNAS Early Edition | 5 of 6

IMMUN

OLO

GYAND

INFLAMMATION

resulted in partial effects and 1-EBIO had a modest effect on theactivation of the NOX-independent NETosis compared with thecalcium ionophore (Fig. 6). The robust effect of the calciumionophore on the induction of NOX-independent NETosis alsosuggests that the calcium signaling is an upstream initiator of thispathway. The partial effects of the chemical inhibitors and in-ducers likely arise from the fact that the inhibitors and activatorsused were chemical compounds that did not achieve full in-hibition. Furthermore, it is also possible that other redundantparallel pathways exist. Collectively, our study demonstrates thatthe calcium ionophore-induced NETosis does not require NOX2activity and is distinct from PMA-induced, NOX-dependentNETosis. We further show that this NOX-independent path-way requires the activation of the calcium-activated SK3 channel,providing insight into mechanisms by which NOX-independentNETosis occurs.

Materials and MethodsHuman Peripheral Neutrophils. The protocol was approved by the Hospital forSick Children ethics committee and signed informed consent was obtainedfrom each subject enrolled. Peripheral blood from healthy donors was col-lected in K2 EDTA blood collection tubes (Becton, Dickinson and Co.),and neutrophils were isolated from the whole blood using PolymorphPrep(Axis-Shield) according to the manufacturer’s instructions with minor mod-ifications (SI Materials and Methods). Some assays were conducted in dHL-60

neutrophil-like cells [e.g., kinase inhibitor assays and siRNA SK3 channelknockdown studies (SI Materials and Methods has additional details)].

NETosis Analysis. NETosis was monitored by a plate reader assay in the pres-ence of Sytox Green cell-impermeable nucleic acid stain (5 μM). NETosis wasconfirmed by imaging the colocalization of myeloperoxidase, and DNA wasstained with Sytox Green after the fixation and permeablization. For somestudies, Western blots were used for determining citH3, the activation statesof kinases (ERK, Akt, and p38), and the protein levels of the SK3 channel.Inhibitors such as DPI, NS8593, scyllatoxin, apamin, MK886, FCCP, and/or DNPwere preincubated with cells for 1 h before the activation of cells for NETosis.

Statistical Analysis. All data are presented as mean ± SEM. Statistical analysiswas performed using GraphPad Prism statistical analysis software (Version5.0a for Mac OS X). Student’s t test was used for comparing two groups.When comparing more than two groups, ANOVA with Bonferroni post testor Dunnett’s test was used where appropriate. A P value of 0.05 or less wasconsidered to be statistically significant.

ACKNOWLEDGMENTS. We thank Mr. Chongbo Yang for providing assis-tance with immunoblot assays. D.N.D. was supported by an OntarioGraduate Scholarship, the Ontario Student Opportunity Trust Fund (SickKidsRestracomp, Dr. Goran Enhorning Award in Pulmonary Research, and Peter-borough K. M. Hunter Graduate Studentship), and a University of TorontoDoctoral Thesis Completion Grant. The study was funded by SickKids ResearchInstitute’s Trainee Start Up Fund (to D.N.D.), Canadian Institutes of HeathResearch (MOP-111012) and Cystic Fibrosis Canada (2619) Grants (to N.P.).

1. Fuchs TA, et al. (2007) Novel cell death program leads to neutrophil extracellulartraps. J Cell Biol 176(2):231–241.

2. Cheng OZ, Palaniyar N (2013) NET balancing: A problem in inflammatory lung dis-eases. Front Immunol 4:1.

3. Brinkmann V, Zychlinsky A (2012) Neutrophil extracellular traps: Is immunity thesecond function of chromatin? J Cell Biol 198(5):773–783.

4. Remijsen Q, et al. (2011) Neutrophil extracellular trap cell death requires both au-tophagy and superoxide generation. Cell Res 21(2):290–304.

5. Bianchi M, et al. (2009) Restoration of NET formation by gene therapy in CGD controlsaspergillosis. Blood 114(13):2619–2622.

6. Pilsczek FH, et al. (2010) A novel mechanism of rapid nuclear neutrophil extracellulartrap formation in response to Staphylococcus aureus. J Immunol 185(12):7413–7425.

7. Parker H, Dragunow M, Hampton MB, Kettle AJ, Winterbourn CC (2012) Require-ments for NADPH oxidase and myeloperoxidase in neutrophil extracellular trap for-mation differ depending on the stimulus. J Leukoc Biol 92(4):841–849.

8. Krause KH, Welsh MJ (1990) Voltage-dependent and Ca2(+)-activated ion channels inhuman neutrophils. J Clin Invest 85(2):491–498.

9. Krause KH, Demaurex N, Jaconi M, Lew DP (1993) Ion channels and receptor-mediatedCa2+ influx in neutrophil granulocytes. Blood Cells 19(1):165–173; discussion 173–175.

10. Essin K, et al. (2007) Large-conductance calcium-activated potassium channel activityis absent in human and mouse neutrophils and is not required for innate immunity.Am J Physiol 293(1):C45–C54.

11. Fay AJ, Qian X, Jan YN, Jan LY (2006) SK channels mediate NADPH oxidase-independent reactive oxygen species production and apoptosis in granulocytes. ProcNatl Acad Sci USA 103(46):17548–17553.

12. Reeves EP, et al. (2002) Killing activity of neutrophils is mediated through activationof proteases by K+ flux. Nature 416(6878):291–297.

13. Segal AW (2005) How neutrophils kill microbes. Annu Rev Immunol 23:197–223.14. Hakkim A, et al. (2011) Activation of the Raf-MEK-ERK pathway is required for

neutrophil extracellular trap formation. Nat Chem Biol 7(2):75–77.15. DoudaDN,YipL,KhanMA,GrasemannH,PalaniyarN(2014)Akt isessential to induceNADPH-

dependent NETosis and to switch the neutrophil death to apoptosis. Blood 123(4):597–600.16. Keshari RS, Verma A, Barthwal MK, Dikshit M (2013) Reactive oxygen species-induced

activation of ERK and p38 MAPK mediates PMA-induced NETs release from humanneutrophils. J Cell Biochem 114(3):532–540.

17. Douda DN, Jackson R, Grasemann H, Palaniyar N (2011) Innate immune collectinsurfactant protein D simultaneously binds both neutrophil extracellular traps andcarbohydrate ligands and promotes bacterial trapping. J Immunol 187(4):1856–1865.

18. Neeli I, Khan SN, Radic M (2008) Histone deimination as a response to inflammatorystimuli in neutrophils. J Immunol 180(3):1895–1902.

19. Neeli I, Radic M (2013) Opposition between PKC isoforms regulates histone deimi-nation and neutrophil extracellular chromatin release. Front Immunol 4:38.

20. Hancock JT, Jones OT (1987) The inhibition by diphenyleneiodonium and its ana-logues of superoxide generation by macrophages. Biochem J 242(1):103–107.

21. Bulua AC, et al. (2011) Mitochondrial reactive oxygen species promote production ofproinflammatory cytokines and are elevated in TNFR1-associated periodic syndrome(TRAPS). J Exp Med 208(3):519–533.

22. Murphy MP (2012) Modulating mitochondrial intracellular location as a redox signal.Sci Signal 5(242):pe39.

23. Rimini R, Rimland JM, Terstappen GC (2000) Quantitative expression analysis of thesmall conductance calcium-activated potassium channels, SK1, SK2 and SK3, in humanbrain. Brain Res Mol Brain Res 85(1-2):218–220.

24. Arai Y, et al. (2014) Uric acid induces NADPH oxidase-independent neutrophil ex-tracellular trap formation. Biochem Biophys Res Commun 443(2):556–561.

25. Li P, et al. (2010) PAD4 is essential for antibacterial innate immunity mediated byneutrophil extracellular traps. J Exp Med 207(9):1853–1862.

26. Wang Y, et al. (2009) Histone hypercitrullination mediates chromatin decondensationand neutrophil extracellular trap formation. J Cell Biol 184(2):205–213.

27. Martinod K, et al. (2013) Neutrophil histone modification by peptidylarginine deiminase4 is critical for deep vein thrombosis in mice. Proc Natl Acad Sci USA 110(21):8674–8679.

28. Femling JK, et al. (2006) The antibacterial activity of human neutrophils and eosi-nophils requires proton channels but not BK channels. J Gen Physiol 127(6):659–672.

29. Devor DC, Singh AK, Frizzell RA, Bridges RJ (1996) Modulation of Cl- secretion bybenzimidazolones. I. Direct activation of a Ca(2+)-dependent K+ channel. Am JPhysiol 271(5 Pt 1):L775–L784.

30. Kaplan MJ, Radic M (2012) Neutrophil extracellular traps: Double-edged swords ofinnate immunity. J Immunol 189(6):2689–2695.

31. Gabriel C,McMasterWR, GirardD, DescoteauxA (2010) Leishmania donovani promastigotesevadetheantimicrobialactivityofneutrophilextracellular traps. J Immunol185(7):4319–4327.

32. Denu JM, Tanner KG (1998) Specific and reversible inactivation of protein tyrosinephosphatases by hydrogen peroxide: Evidence for a sulfenic acid intermediate andimplications for redox regulation. Biochemistry 37(16):5633–5642.

33. Traore K, et al. (2008) Redox-regulation of Erk1/2-directed phosphatase by reactiveoxygen species: Role in signaling TPA-induced growth arrest in ML-1 cells. J CellPhysiol 216(1):276–285.

34. Dikalov S (2011) Cross talk between mitochondria and NADPH oxidases. Free RadicBiol Med 51(7):1289–1301.

35. Daiber A (2010) Redox signaling (cross-talk) from and to mitochondria involves mito-chondrial pores and reactive oxygen species. Biochim Biophys Acta 1797(6-7):897–906.

36. Nazarewicz RR, Dikalova AE, Bikineyeva A, Dikalov SI (2013) Nox2 as a potentialtarget of mitochondrial superoxide and its role in endothelial oxidative stress. Am JPhysiol Heart Circ Physiol 305(8):H1131–H1140.

37. Szewczyk A, Jarmuszkiewicz W, Kunz WS (2009) Mitochondrial potassium channels.IUBMB Life 61(2):134–143.

38. Dolga AM, et al. (2014) Subcellular expression and neuroprotective effects of SKchannels in human dopaminergic neurons. Cell Death Dis 5:e999.

39. Dolga AM, et al. (2013) Mitochondrial small conductance SK2 channels prevent gluta-mate-induced oxytosis and mitochondrial dysfunction. J Biol Chem 288(15):10792–10804.

40. Xia XM, et al. (1998) Mechanism of calcium gating in small-conductance calcium-activated potassium channels. Nature 395(6701):503–507.

41. Maylie J, Bond CT, Herson PS, Lee WS, Adelman JP (2004) Small conductanceCa2+-activated K+ channels and calmodulin. J Physiol 554(Pt 2):255–261.

42. Bildl W, et al. (2004) Protein kinase CK2 is coassembled with small conductanceCa(2+)-activated K+ channels and regulates channel gating. Neuron 43(6):847–858.

43. Itoh T, Kanmura Y, Kuriyama H (1985) A23187 increases calcium permeability of store sitesmore than of surface membranes in the rabbit mesenteric artery. J Physiol 359:467–484.

44. Cudd L, Clarke C, Clinkenbeard K (2003) Contribution of intracellular calcium stores toan increase in cytosolic calcium concentration induced by Mannheimia haemolyticaleukotoxin. FEMS Microbiol Lett 225(1):23–27.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1414055112 Douda et al.

Supporting InformationDouda et al. 10.1073/pnas.1414055112SI Materials and MethodsHuman Peripheral Neutrophils.Peripheral blood from healthy donorswas collected inK2EDTAblood collection tubes (Becton,Dickinsonand Co.), and neutrophils were isolated from the whole blood usingPolymorphPrep (Axis-Shield) according to the manufacturer’s in-structions with minor modifications. Namely, the lysis of red bloodcells was done using a hypotonic solution [0.2% (wt/vol) NaCl]and was followed by the addition of an equal volume of 1.6% (wt/vol)NaCl solution with Hepes buffer (20 mM, pH 7.2) to makethe solution buffered and isotonic. This was followed by twowashes with a wash buffer containing 0.85% (wt/vol) NaCl andHepes (10 mM, pH 7.2). Isolated neutrophils were then re-suspended in RPMI medium (Invitrogen) supplemented withHepes buffer (10 mM, pH 7.2), except for the experiments as-sessing the requirement for extracellular calcium. For these, thecells were cultured in HBSS without CaCl2 or MgCl2 (In-vitrogen), which was supplemented back with MgCl2 (4.93 mM),and cells activated with A23187 (4 μM), ionomycin (5 μM), orPMA (25 nM). DMSO or methanol was used where appropriateas vehicle controls. For all assays, cells were preincubated witheither NOX inhibitor DPI (20 μM), or SK channel inhibitorsNS8593 (0.1–100 μM), scyllatoxin (18.5–70 nM), apamin (50–200 nM),or TRPM7 inhibitor MK886 (3.75–15 nM), or mitochondrial un-couplers carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone(FCCP) (1–10 μM) and/or dinitrophenol (DNP) (100–750 μM),for 1 h before the activation of cells for NETosis. For each ex-periment, only 1–2 donor samples were processed each day. Theassays were repeated with different donors to obtain experimentalreplicates.

Cell Culture. HL-60 cells (ATCC) were maintained in RPMI with20% (vol/vol) FBS. To differentiate HL-60 cells into neutrophil-like cells, the cells were incubated in culture medium withDMSO (1%) for 3 d. The medium containing DMSO (1%) wasthen replaced and the cells were cultured for an additional 2 d.Experiments were performed on these cells on day 5 after thestart of differentiation. For the kinase inhibitor assays, dHL-60cells were preincubated with p38 inhibitor, SB20190 (1–2 μM;Tocris Bioscience), Akt inhibitor XI (5–10 μM; EMD Milli-pore), or ERK inhibitor, FR180204 (10–20 μM; Tocris Bio-science) for 1 h before the activation of cells for NETosis.

Plate Reader Assays. For the plate reader assay, cells were seededat 3 × 104 cells per well in a 96-well plate in the culture media inthe presence of Sytox Green cell-impermeable nucleic acid stain(Life Technologies) at 5 μM. The fluorescence was measuredusing POLARstar OMEGA fluorescence microplate reader(BMG Labtech) at specific time intervals for up to 300 min afterthe activation of cells. To calculate NETotic index, fluorescencereadout obtained from cells lysed with 0.5% (vol/vol) TritonX-100 was considered as 100% DNA release, and the index wascalculated as the percentage of total value at each time point.

Detection of Reactive Oxygen Species Production. DHR123 andMitoSOX were used for detecting cytosolic reactive oxygenspecies (ROS) and mitochondrial superoxide, respectively. Confocalmicroscopy was used for qualitative analysis, whereas flow cytometryand plate reader assays were used for quantitative analysis.

siRNA Knockdown of the SK3 Channel (KCNN3). HL-60 cells wereused for the siRNA experiments. The siRNAs used were theSilencer Select siRNAs (Life Technologies, 4392420) against

KCNN3 and Silencer Select Negative Control No. 1 siRNA (LifeTechnologies, 4390843). For a maximal knockdown, a mixture ofthree different siRNAs against KCNN3 (s7798, s7799, and s7800),which target different regions of the KCNN3 transcript, wereused. BLOCK-iT Alexa Fluor Red Fluorescent Control siRNAwas used to optimize the transfection conditions. On day 4 of thedifferentiation ofHL-60 cells, siRNAmixture or the control siRNAwas transfected into the cells using Lipofectamine RNAiMAXreagent (Life Technologies) according to the manufacturer’s speci-fications. Briefly, 30 nM of siRNAs were premixed with Lipofect-amine (1:4 siRNA:Lipofectamine) and incubated at roomtemperature for 5 min. The siRNA/Lipofectamine mixture wasthen added to the cells and incubated at 37 °C and 5% CO2.Twenty-four hours later, the cells were washed and resuspendedin fresh RPMI containing 10% FBS and cultured for an additional24 h before performing the assays.

Immunoblot Analysis. For immunoblot assays, the tubes containingcells were placed on ice after the activation of cells for up to120 min, washed once with ice cold PBS, and lysed with a lysisbuffer containing complete protease inhibitor mixture (Roche)supplemented with NaVO3 (1 mM), leupeptin (25 μM), pepstatin(25 μM), aprotinin (25 μM), NaF (25 mM), levamisole (1 mM),PMSF (1 mM), and 0.1% (wt/vol) Triton X-100 to make 2–5 × 107cells/mL. Cells were then sonicated three times using an aquasonicsonicator (VWR, model 50D at the highest power setting, 10 seach) and centrifuged at 20,000 × g for 30 min to remove insolubleparticles. The samples were then incubated with an equal volumeof 2× loading buffer containing Tris·HCl (125 mM, pH 6.8), 6%(wt/vol) SDS, 8% (vol/vol) β-mercaptoethanol, 18% (vol/vol)glycerol, 5 mM EDTA, 5 mM EGTA, leupeptin (10 μg/mL),pepstatin (10 μg/mL), aprotinin (10 μg/mL), NaF (10 mM),NaVO3 (5 mM), and levamisole (1 mM) and incubated for 10 minat 95 °C. The transferred blots were blocked with 5% (wt/vol) milkin PBS containing 0.05% (wt/vol) Tween 20 (PBS-T) buffer for1 h at room temperature. The antibodies used were: anti-phospho-ERK 1/2 (AW39R) (EMD Millipore) at 1:1,000; anti-phospho-p38MAPK (D3F9) XP (Cell Signaling) at 1:500; anti-phospho-Akt(Ser473) 193H12 (Cell Signaling) at 1:1,000; anti-ERK1/2 (p44/p42)MK12 (EMD Millipore) at 1:1,1000; anti-p38 MAPK (D13E1) XP(Cell Signaling) at 1:500; anti-Akt (PH Domain, clone SKB1)(Cell Signaling) at 1:1,000; anti-histone H3 (citrulline R2 + R8 +R17) ChIP Grade (Abcam), and anti-SK3 (H-17) (Santa CruzBiotechnology) at 1:500. Densitometry on the scanned blots wasperformed using Adobe Photoshop (Adobe Systems). Meansignal intensity was taken and multiplied by the number of pixelsfrom each band. For phosphorylated ERK, Akt, and p38 im-munoblots the densitometry data was standardized to the den-sitometry data obtained from respective total kinase blots fromeach time point. Data obtained was then normalized to the valuefor A23 at t = 30.

Confocal Imaging.For imaging, cells used for the plate reader assaywere fixed with PFA (4%, wt/vol) in PBS buffer for 10 min andpermeablized with Triton X-100 (0.1%, wt/vol). Mouse α-myelo-peroxidase antibody (ab25989, Abcam) at 1:500 dilution was usedto stain MPO. DNA was already stained with Sytox Green afterthe fixation and permeablization. The confocal images were takenusing a Olympus IX81 inverted fluorescence microscope equippedwith a Hamamatsu C9100-13 back-thinned EM-CCD cameraand Yokogawa CSU ×1 spinning disk confocal scan head (withSpectral Aurora Borealis upgrade), 4 separate diode-pumped

Douda et al. www.pnas.org/cgi/content/short/1414055112 1 of 7

solid state laser lines (Spectral Applied Research, 405 nm, 491 nm,561 nm, and 642 nm). The objectives used were 20×/0.75 or60×/1.35. The microscope was operated with Volocity software(Perkin-Elmer). Images taken on the spinning disk confocal mi-croscope was deconvolved by iterative restoration with confi-dence limit set to 95% and iteration limit set to 20. All imageswere deconvolved to the confidence limit before reaching it-eration limit.

Detection of ROS Production. For the detection of ROS, cells werepreloaded with DHR123 (Life Technologies) or MitoSOX Red

Mitochondrial Superoxide Indicator (Life Technologies) for 10 minas per manufacturer’s instructions. The loaded cells were thenwashed and activated with PMA or A23187 for another 30 min.For flow cytometry, data were acquired with a Gallios flow cy-tometer (Bechman Coulter) and were analyzed with FlowJosoftware (Tree Star). Images were acquired as described above.For the plate reader assays, the cells loaded with the indicatorswere seeded on a 96-well plate, and the cells were activated witheither PMA or A23 with or without prior incubation with DPI orDNP. The fluorescence was measured by an Omega fluorescencemicroplate reader.

Fig. S1. Staphylococcus aureus strain RN4220-induced NETosis is inhibited by DPI. Human neutrophils were incubated with S. aureus at two different mul-tiplicity of infection (MOI) in the presence or absence of DPI. (A) Time course analysis of the plate reader assays shows that S. aureus-induced NETosis followsa slow kinetic and is abolished in both 100 and 10 MOI in the presence of DPI (***P < 0.001; two-way ANOVA; n = 3). (B) The result of the plate reader assay at240 min showing the significant inhibition of NETosis induced by S. aureus in the presence of DPI (***P < 0.001; one-way ANOVA; n = 3).

Douda et al. www.pnas.org/cgi/content/short/1414055112 2 of 7

Fig. S2. Ionomycin induces a rapid NETosis and the calcium ionophore-induced NETosis requires extracellular calcium. (A) Human peripheral neutrophils wereincubated with differing concentrations of ionomycin. The plate reader assays show that ionomycin dose dependently induced NETosis with a fast kinetics. Thehighest NET release was observed as early as 1 h at the highest concentration tested (n = 3). (B) Human neutrophils were activated with A23187, ionomycin, orPMA in the presence or absence of calcium. NETotic index was measured at 300 min postactivation. The results of the plate reader assays are expressed aspercentage of inhibition compared with the activation in the presence of calcium (n = 3). A23187- or ionomycin-induced NETosis, but not PMA-inducedNETosis, was significantly reduced in the absence of extracellular calcium. (C) Immunofluorescence imaging shows that the calcium ionophores A23187 andionomycin induce NETosis. Neutrophils were incubated in the presence or absence of A23187 (4 μM) or ionomycin (5 μM) for 300 min. Cells were stained forDNA (green) and MPO (red). DIC, differential interference contrast. Images are representative of three independent experiments. (Scale bar, 10 μm.) Ca2+,calcium chloride. *P < 0.05; **P < 0.01 (Student’s t test).

Douda et al. www.pnas.org/cgi/content/short/1414055112 3 of 7

Fig. S3. Akt activation is required for both NETotic pathways, whereas ERK activation is needed for PMA-induced NETosis. (A) Cells were harvested afteractivation with PMA, A23187, or negative control (-ve) for the indicated times. The -ve control samples were incubated for 120 min in the absence ofany activator. The immunoblot shows that even after overexposing the film shown in Fig. 3A, citH3 is not detectable in neutrophils activated with PMA (n = 3).(B–E) Densitometry analysis for the immunoblot analysis of citH3 (B, n = 3), ERK (C, n = 3), Akt (D, n = 3), and p38 (E, n = 2). Results for ERK (C), Akt (D), and p38(E) was normalized to total kinase for each respective timepoint, and then to the A23187 value at t = 30 of respective blots and expressed as fold increase. *p < 0.05(PMA vs. A23187 values at 60-min time points were compared; Student’s t test). (F–K) Effects of kinase inhibitors on PMA- and A23187-activated NETosis. dHL-60cells were preincubated with indicated concentrations of ERK inhibitor FR180204 (F and G; n = 3), Akt inhibitor XI (Akt-i; H and I; n = 3) or p38 inhibitor SB202190(J and K; n = 4) and then activated with PMA (F, H, and J) or A23187 (G, I, and K). *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA).

Douda et al. www.pnas.org/cgi/content/short/1414055112 4 of 7

Fig. S4. A23187-mediated NETosis requires mitochondrial respiration in dHL-60 neutrophil-like cells and neutrophils. (A–D) The dHL-60 cells were activatedwith A23187 in the presence or absence of dinitrophenol (DNP; A and B) or carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP; C and D). The platereader assays show that NETosis is significantly inhibited in the presence of DNP or FCCP, in a time-dependent (A and C) and dose-dependent manner (B and D).(n = 3). These results indicate that mitochondrial ROS production is required for A23187-induced NETosis in dHL-60 neutrophil-like cells. (E and F) Dose-dependent reduction of NETosis in human neutrophils at 300 min postactivation with A23187 (A, n = 5) or PMA (B, n = 3) in the presence of differing con-centrations of DNP. The numbers on the x axis represent DNP concentrations in micromoles. DNP significantly reduces A23187-, but not PMA-induced NETosis.*P < 0.05; **P < 0.01; ***P < 0.001 compared with the agonist alone (A and C, two-way ANOVA; B and D–F, one-way ANOVA).

Douda et al. www.pnas.org/cgi/content/short/1414055112 5 of 7

Fig. S5. SK channel is required for ionophore-mediated NETosis. (A and B) Human neutrophils were activated with A23187 (A) or PMA (B) in the presence ofdifferent concentrations of SK channel inhibitor NS8593 for 300 min (n = 3). The plate reader assay shows that the reduction of NET release by NS8593 is dosedependent in A23187-, but not in PMA-mediated NETosis. Therefore, A23187-, but not PMA-induced NETosis requires the activation of the SK channel. (C andD) Human neutrophils were activated with A23187 (C) or ionomycin (D) in the presence of different concentrations of another SK channel inhibitor apamin.The plate reader assays show that NETosis is reduced dose dependently in the presence of apamin (n = 3). (E and F) Human neutrophils were activated withA23187 (A23; E) or ionomycin (Io; F) in the absence or presence of differing concentrations of scyllatoxin for 240 min. Scyllatoxin has strongest affinity to SK2channels. This toxin does not inhibit ionophore-mediated NOX-independent NETosis. (G and H) Neutrophils were incubated with different concentrations ofTRPM7 channel inhibitor MK886. This inhibitor does not inhibit A23187 (G)- or ionomycin (H)-induced NETosis. Numbers below the graphs represent con-centrations of the inhibitors. **P < 0.01; ***P < 0.001 compared with the agonist alone (one-way ANOVA). A23, A23187; Io, ionomycin.

Douda et al. www.pnas.org/cgi/content/short/1414055112 6 of 7

Fig. S6. SK3 is required for the NOX-independent NETosis. (A) Efficient transfection of scrambled control siRNA into HL-60 cells. Confocal microscopy imagesshow that nearly all cells are transfected by the Alexa Fluor red fluorescent control siRNA (n = 3). (Scale bar, 50 μm.) (B and C) The plate reader assay shows thatA23187 (A)- or ionomycin (B)-mediated NETosis in siRNA (si)-transfected dHL-60 cells, are significantly reduced compared with scrambled RNA (sc)-transfectedcontrol cells (n = 3). Knockdown of SK3 channel by siRNA significantly reduces calcium ionophore-induced NETosis. sc, scrambled RNA-transfected cells; si, SK3siRNA-transfected cells. **P < 0.01; ***P < 0.001 compared with the agonist alone (B and C, two-way ANOVA).

Douda et al. www.pnas.org/cgi/content/short/1414055112 7 of 7