Activated protein C inhibits neutrophil extracellular trap formation in vitro and activation in vivo

Activated protein C (APC) is a multifunctional serine protease with anticoagulant, cytoprotective, and anti-inflammatory activities. In addition to the cytoprotective effects of APC on endothelial cells, podocytes, and neurons, APC cleaves and detoxifies extracellular histones, a major component of neutrophil extracellular traps (NETs). NETs promote pathogen clearance but also can lead to thrombosis; the pathways that negatively regulate NETosis are largely unknown. Thus, we studied whether APC is capable of directly inhibiting NETosis via receptor-mediated cell signaling mechanisms. Here, by quantifying extracellular DNA or myeloperoxidase, we demonstrate that APC binds human leukocytes and prevents activated platelet supernatant or phorbol 12-myristate 13-acetate (PMA) from inducing NETosis. Of note, APC proteolytic activity was required for inhibiting NETosis. Moreover, antibodies against the neutrophil receptors endothelial protein C receptor (EPCR), protease-activated receptor 3 (PAR3), and macrophage-1 antigen (Mac-1) blocked APC inhibition of NETosis. Select mutations in the Gla and protease domains of recombinant APC caused a loss of NETosis. Interestingly, pretreatment of neutrophils with APC prior to induction of NETosis inhibited platelet adhesion to NETs. Lastly, in a nonhuman primate model of Escherichia coli-induced sepsis, pretreatment of animals with APC abrogated release of myeloperoxidase from neutrophils, a marker of neutrophil activation. These findings suggest that the anti-inflammatory function of APC at therapeutic concentrations may include the inhibition of NETosis in an EPCR-, PAR3-, and Mac-1-dependent manner, providing additional mechanistic insight into the diverse functions of neutrophils and APC in disease states including sepsis.

Activated protein C (APC) 3 is a 56-kDa serine protease that acts as an anticoagulant and anti-inflammatory molecule. In complex with protein S, APC inactivates coagulation factors (F) Va and VIIIa, dampening the generation of the thrombin (1)(2)(3)(4). APC is generated from the zymogen protein C via enzymatic cleavage by thrombin. On endothelial cells this reaction is enhanced in a calcium-dependent manner in the presence of thrombomodulin and endothelial protein C receptor (EPCR) (5)(6)(7)(8). Because of the combined antithrombotic and anti-inflammatory properties of APC, recombinant APC was used clinically as a therapeutic for adult severe sepsis before its removal in 2011, as no overall benefit to patient outcome was observed in a second phase 3 trial performed 10 years after the first successful phase 3 trial (9,10). In severe sepsis therapy, APC increased risk for serious bleeding (9). Thus, a comprehensive understanding of the structure-activity relationships for the various functional activities of APC may provide the rationale for developing different engineered APC mutants that maintain the anti-inflammatory or cytoprotective functions of APC without adversely affecting thrombin generation or compromising hemostasis (11,12).
Neutrophils are the most populous innate immune cells found in circulation, playing an indispensable role in surveying the body for signs of infection and inflammation (13). Neutrophils express several of the key receptors responsible for binding and mediating cell signaling of APC, and APC can inhibit neutrophil migration (14 -18).
Classically, the cytoprotective mechanisms of APC have been best characterized on the endothelium, where APC canonically binds EPCR and cleaves protease activated receptor (PAR) 1 at Arg-46, a unique site different from thrombin's cleavage site at Arg-41, activating an intracellular signaling cascade to promote barrier function and inhibit apoptosis, a concept known as biased agonism (19 -22). Moreover, subsequent studies have made a similar observation that PAR3 on the endothelium is cleaved by APC at Arg-41, a noncanonical position, which may in part explain the ability of APC to produce a PAR3-dependent cytoprotective effect observed in podocytes (23,24). Our aim was to determine whether APC was able to ligate EPCR and employ PAR3 to regulate neutrophil cell survival.
One form of neutrophil cell death that has garnered much interest since its initial report in 2004 is the ability of neutrophils to form extracellular traps (25). Neutrophil extracellular traps (NETs) are comprised of extruded nuclear DNA and associated nuclear proteins including histones and granular proteins. NETs are hypothesized to act as an additional mechanism to promote pathogen clearance. Recent studies have focused on the characterization of the mechanisms that result in NETosis, as well as the pathophysiological relevance of this process (26 -33). Since their initial discovery, NETs have been shown to promote activation of select coagulation factors and platelets, thereby exacerbating several disease states including thrombosis (34 -36). Based on studies linking the activation of blood cells and coagulation factors with inflammatory processes leading to thrombosis, the concept of immunothrombosis was proposed to encompass the complex reactions and cross-talk occurring in disease states such as sepsis (37). APC has been proposed to play a role in inhibiting immunothrombosis, based in part on the observation that APC was found to cleave extracellular histones in animal models of sepsis, suggesting a possible link between NETs and APC function (38 -40). Although it is known that neutrophils express cognate receptors that bind APC, it is unknown whether APC can inhibit signaling mechanisms that potentiate NETosis. We have recently shown that APC binds to leukocytes and NETs in a Gla domain-dependent manner (41). Herein, we present evidence that APC binds neutrophils to elicit a receptor-mediated intracellular signaling cascade to inhibit neutrophil activation and NETosis. Moreover, infusion of pharmacological levels of APC abrogated neutrophil activation and myeloperoxidase (MPO) release in a nonhuman primate model of bacterial sepsis. Together these results suggest that the anti-inflammatory properties of APC may extend to the inhibition of neutrophil cell death, including NETosis.

APC binds leukocytes and inhibits NETosis
It was first validated that APC bound neutrophils in a static adhesion assay. Neutrophils were plated on coverslips coated with immobilized APC and assessed for adhesion using differential interference contrast microscopy. Neutrophils bound to APC to a similar degree as to fibronectin; significantly fewer neutrophils bound to BSA-or IgG-coated coverslips (Fig. 1A). It was next studied whether APC in solution bound to neutrophils undergoing NETosis. The data show that soluble APC bound to both the activated neutrophil cell body and the extracellular DNA during NETosis (Fig. 1B).
Subsequently, whether APC binding to neutrophils had a functional effect on the process of NETosis was investigated. NETs were formed upon incubation of neutrophils with either autologous platelet secretome or the protein kinase C (PKC) activator PMA. NET formation was characterized by an increase in surface area of DNA, detection of citrullinated histone 3 (H3) and extracellular appearance of MPO ( Fig. 2A). Prior incubation of neutrophils with APC caused a significant reduction in the area of DNA and MPO extruded by neutrophils during NETosis induced by either platelet secretome or PMA ( Fig. 2B and supplemental Fig. S1).
PMA was used to induce NETosis in the subsequent mechanistic studies because of the enhanced signal-to-noise ratio for surface area of DNA observed when NETosis was induced by PMA as compared with platelet secretome. Results show that a minimal concentration of 75 nM APC was sufficient to reduce the extent of DNA surface area, whereas 300 nM APC potently reduced PMA-induced NET formation (Fig. 3, A-C). Pretreatment of APC with the serine protease inhibitor PPACK reversed the inhibitory effect of APC on NETosis. Equimolar concentrations of the zymogen protein C or the serine protease thrombin failed to inhibit the extent of DNA extruded by neutrophils during NETosis (Fig. 3, A and B). Acid-washed glass coverslips were coated with denatured BSA (5 mg/ml), 20 g/ml IgG, 20 g/ml fibronectin (FN), or 100 g/ml immobilized APC for 1 h, followed by PBS wash and then blocked with denatured BSA (5 mg/ml). A, purified human neutrophils (2 ϫ 10 6 /ml) were plated on the coverslips for 1 h at 37°C, followed by washing and fixation. After imaging, cell adhesion was counted in ImageJ and quantified as cell per field of view (FOV). B, neutrophils (2 ϫ 10 6 /ml) were plated on fibronectin-coated coverslips and were treated with HBSS or PMA (10 nM) for 3 h at 37°C. Cell samples were washed and vehicle (HBSS buffer) or APC (300 nM) was then incubated for 15 min with the cell samples at 37°C. Samples were then fixed with 4% PFA and incubated overnight with primary antibodies. The samples were then incubated with Hoechst 33342 (1:1000) and secondary antibody Alexa Fluor 488 goat anti-mouse (Invitrogen) (1:500). Images were normalized to secondary antibody-alone images. Shown above (C) are representative images of APC-positive staining in the presence of NETs and neutrophils *, p Ͻ 0.05 versus BSA. Data are mean Ϯ S.E. nϭ3.
To determine whether the effect of APC on NETosis was because of a receptor-ligand interaction or because of the direct enzymatic cleavage of extracellular DNA, a wash step was introduced after incubation of neutrophils with APC and prior to PMA stimulation. As shown in Fig. 3C, an equivalent level of inhibition of NETosis was observed for experiments in which APC was washed away prior to induction of NETosis, with a minimal concentration of 75 nM APC causing a reduction in PMA-induced NET formation. Moreover, inclusion of PPACK in this experiment eliminated the inhibitory effect of APC without having an effect on vehicle-treated, PMA-induced NETosis (supplemental Fig. S2A), whereas pretreatment of neutrophils with an equimolar concentration of thrombin had no significant effect on the ability of neutrophils to undergo PMA-induced NETosis (supplemental Fig. S2B). Because PPACK irreversibly inhibits APC enzyme activity, this implies APC proteolytic activity is required.

Proteins PAF and A1-AT eliminate APC inhibition of NETosis
Other inhibitors of APC activity were studied. Platelet activating factor (PAF) is a phospholipid that binds EPCR, thereby inhibiting APC binding to EPCR and signaling (42). NET formation was detected by staining for DNA and citrullinated H3 (Fig. 4A), and PAF eliminated the inhibitory effect of APC on PMA-induced NETosis, as evidenced by an overall DNA area increase from 1380 Ϯ 47.8 m 2 in the presence of APC alone to 2140 Ϯ 40.9 m 2 in the presence of APC with PAF (Fig. 4B).
Under physiological conditions, APC is regulated in part by the heparin-independent plasma inhibitor, ␣1-antitrypsin (A1-AT), which blocks the proteolytic activity of APC (43). Experiments were designed to test whether A1-AT would block the ability of APC to inhibit NETosis. When APC was pretreated with A1-AT, the DNA area increased from 1380 Ϯ 47.8 m 2 for the presence of APC alone to 2040.5 Ϯ 168.2 m 2 when APC was pretreated with A1-AT (Fig. 4C). As a control, thrombin was used in lieu of APC, and in either the presence or the absence of A1-AT, there was no significant change in the ability of PMA to induce NETosis, suggesting specificity for APC with regard to inhibiting NETosis.

Neutrophil receptors Mac-1, EPCR, and PAR3 mediate APC inhibition of NETosis
Because our results indicated that both the proteolytic activity and the binding of APC to EPCR are required for APCmediated inhibition of NETosis, we sought to identify the receptors that mediate this potential cytoprotective function of APC. Purified human neutrophils were treated with functionblocking antibodies prior to incubation with APC and subsequent induction of NETosis with PMA. APC pretreatment resulted in a significant decrease in DNA area from 2380 Ϯ 45.7 m 2 to 1010 Ϯ 39.9 m 2 . After pretreatment of neutrophils with blocking antibodies to either CD11b or CD18, it was found that only in combination was there blockade of APC-mediated inhibition of NETosis, as evidenced by the increase to 2230 Ϯ 93.4 m 2 of DNA area in the presence of the combination of these antibodies (Fig. 5A), whereas a blocking antibody to CD11a had no effect (data not shown). The ability of APC to inhibit PMA-induced NETosis was eliminated in the presence of a blocking antibody to EPCR (RCR-252), as evidenced by a corresponding increase in DNA area to 2260 Ϯ 34.4 m 2 . Notably, neutrophil pretreatment with a blocking antibody to PAR3 blocked the ability of APC to inhibit NETosis, as evidenced by the increase of DNA area to 2080 Ϯ 104 m 2 . In contrast, pretreatment with a specific small molecule PAR1 antagonist, SCH79797, was found to have no significant effect on the ability of APC to inhibit NETosis (supplemental Fig. S3). As a positive control, platelets pretreated with SCH79797 resulted in the inhibition of TRAP-induced platelet activation (data not shown).
Because the use of an antibody that blocked the cleavage of PAR3 resulted in the loss of APC-mediated inhibition of NETosis, we sought to test whether the PAR3 tethered-ligand peptides generated by APC or thrombin cleavages could inhibit NET formation (24). The thrombin-derived P3K peptide, which is generated after the cleavage of PAR3 at Lys-38 by thrombin, had no effect on PMA-induced NET formation. In contrast to the P3K peptide, the use of the APC-derived P3R peptide, which is generated after the cleavage of PAR3 at Arg-41 by APC, significantly reduced PMA-induced NETosis in either the presence or absence of APC (Fig. 5B), implying that PAR3 agonism was sufficient to inhibit NETosis.

APC signals via G␤␥ to inhibit PKC and PI3K signaling-dependent NET formation
The signaling pathways that regulate NETosis were studied, first by validation of the pathways by which platelet secretome and PMA potentiate NETosis. Pretreatment of neutrophils with the PI3K inhibitor wortmannin eliminated NET formation induced by either platelet secretome or the PKC activator PMA (Fig. 6A). Moreover, pretreatment of neutrophils with U73122, an inhibitor to phospholipase C (PLC) that acts upstream of PKC, abrogated NET formation stimulated by platelet secretome but not PMA, whereas the inactive analog of the PLC inhibitor U73343 had no effect on NETosis potentiated by either agonist. Pretreatment of platelet secretome with the serine protease inhibitor PPACK had no effect on the ability of platelet secretome to induce NETosis (data not shown). Whether there was a role for cytokine receptor-mediated sig-naling in platelet secretome-induced NETosis was tested next. Pretreatment of neutrophils with the JAK/STAT inhibitor cucurbitacin I significantly reduced the degree of DNA extruded from neutrophils following exposure to platelet secretome while having no effect on PMA-induced NETosis (Fig. 6A).
Next, the signaling pathways by which APC inhibits PMAinduced NETosis were characterized. As PMA is a diacylglycerol analog, it activates PKC to drive NETosis in vitro in a PI3Kdependent manner (30,31,44). First, the role of PKC signaling pathway in PMA-induced NETosis was confirmed. As seen in Fig. 6B, pharmacological inhibition of PKC with either GF109203X or Ro31-8220 ablated the ability of PMA to induce NETosis. In contrast, inhibition of PLC or G␤␥ signaling with U73122 or gallein, respectively, had no effect on PMA-induced NETosis, yet eliminated the ability of APC to inhibit NETosis, consistent with the findings suggesting that APC induces signaling downstream of the G protein-coupled receptor PAR3 to inhibit NETosis. Of note, a partial inhibition of NETosis was Acid-washed glass coverslips were coated with 20 g/ml fibronectin and then blocked with denatured BSA (5 mg/ml). Purified human neutrophils (2 ϫ 10 6 /ml) were plated on the coverslips for 30 min at 37°C. A and B, they were then incubated with increasing concentrations of APC (30, 100, and 300 nM), protein C (300 nM), and thrombin (300 nM) in the presence or absence of PPACK (40 M) for 30 min at 37°C. Samples were washed and then subsequently treated with HBSS or PMA (10 nM) for 3 h at 37°C. C, purified human neutrophils (2 ϫ 10 6 /ml) were plated on the coverslips for 30 min at 37°C, then incubated with increasing concentrations of APC (30, 75, 100, 150, and 300 nM) for 30 min at 37°C. Samples were then washed once with PBS and subsequently treated with HBSS or PMA (10 nM) for 3 h at 37°C. All samples were then fixed with 4% PFA. Samples were incubated overnight with polyclonal rabbit anti-neutrophil elastase antibody (1:100). The samples were then incubated with Hoechst 33342 (1:1000) and secondary antibody Alexa Fluor 546 goat anti-rabbit IgG (Invitrogen) (1:500). Images were normalized to secondary antibody-alone images. Shown in A are representative images of PMA-induced NETs in the presence of increasing concentrations of coagulation factors. Images were analyzed in a custom MATLAB program to quantify each pixel-positive signal as area DNA per image, shown in B and C. *, p Ͻ 0.001 versus vehicle ϩ PMA. Data are mean Ϯ S.E. n ϭ 3.
observed for gallein in the absence of APC, likely because of the fact that gallein is also known to exhibit a partial off-target inhibitory effect on the PI3K pathway (Fig. 6B).

The protease and Gla domains of APC mediate the inhibition of NET formation
Subsequent experiments were designed to determine the regions of APC responsible for inhibiting NETosis. A panel of recombinant APC mutants was rationally created with mutations within the active site and charged exosite of the protease domain, as well as within the Gla domain, as depicted in Fig.  7A. Equimolar concentrations of recombinant human APC (rhAPC) inhibited PMA-induced NETosis to the same degree as plasma-derived APC (Fig. 7B). The proteolytic activity of APC was required to inhibit NETosis. Consistent with the findings that demonstrated the action of APC was blocked by the serine protease inhibitor PPACK (Fig. 3), the inhibitory function of APC was lost when the active site residue Ser-360 was mutated to Ala, rendering APC catalytically inactive (S360A-APC) (Fig. 7C).
Recombinant APC mutants with protease domain mutations were used to determine structure-activity relationships for inhibition of NETosis. APC with the mutation Asn-329 to Gln (N329Q-APC) exhibited a loss of function and was unable to inhibit NETosis. The N329Q mutation results in the loss of an attached carbohydrate that possibly participates in APC-receptor interactions. In the same sequence as Asn-329, substitution of two glutamic acid residues for alanine (E330A/E333A-APC) resulted in only partial inhibition of PMA-induced NETosis, suggesting that residues 329 -333 are required for normal inhibition of NETosis by APC. Mutation of Ser-252 located along the opening of the catalytic pocket to the acidic residue Glu (S252E-APC) also eliminated the ability of APC to inhibit NETosis. However, mutation at the same residue to alanine (S252A-APC) did not impair the ability of APC to inhibit NETosis (Fig. 7C).
As mutations in and around the active site seem to play a critical role for APC inhibition of NETosis, we next sought to determine whether charged exosites on the face opposite from the face containing the active site were necessary to inhibit NETosis. A mutation that enhances protein S cofactor activity by mutating the negatively charged Glu-149 to Ala caused a partial reduction in the ability of E149A-APC to inhibit PMAinduced NETosis. Next, it was determined if the positively charged exosites in the protease domain played a role in the functional inhibition of NET formation by APC. There was a partial loss of function for APC when the residues in the calcium-binding binding loop were mutated to alanine (R229A/ R230AAPC). In contrast, mutation of several positively charged lysine residues 191-193 to alanine in the 37-loop had no effect on the inhibitory function of APC, as evidenced by an equivalent degree of inhibition of NETosis for the 3K3A-APC mutant and rhAPC (Fig. 7D).
The protease domain of APC contains an Arg-Gly-Asp (RGD) triad that is canonically recognized by integrins; thus, Asp-180 in the RGD motif was mutated to determine whether the RGD motif of APC played a role in the regulation of NETosis. D180E-APC had normal ability for inhibition of NETosis (Fig. 7D).
The ␥-carboxyglutamic acid-rich (Gla) domain of APC plays an essential role mediating binding to both EPCR and phosphatidylserine exposed on the surface of activated blood and endothelial cells (45). Consistent with data in Fig. 5 implicating a role for EPCR in mediating the inhibitory function of APC, the Gla domain mutation of Leu-8 to Val (L8V-APC) caused a partial loss of activity (Fig. 7E).
The anticoagulant function of APC is enhanced when bound to the cofactor protein S, increasing the rate of inactivation of FVa and FVIIIa. Recently, there was demonstrated a critical role for protein S and cleaved FVa acting as cofactors for APC antiinflammatory function in a mouse model of septic peritonitis (46). Moreover, studies have demonstrated that binding of protein S to APC requires the APC residue Leu-38 in the Gla domain, as mutation to Asp (L38D-APC) impairs APC and protein S interactions (47). The L38D mutation caused a partial loss of the NETosis inhibitory activity of APC. The substitution of the bulky polar uncharged amino acid, Gln for hydrophobic Leu-38 in APC produced a complete loss of function (Fig. 7E),  indicating that residue Leu-38 located on the top of the Gla domain is critical for inhibition of NETosis.

Platelet adhesion to NETs is inhibited by APC
Experiments were designed to determine whether APC could inhibit platelet-NET binding, which has been implicated as playing a pathological role in promoting immunothrombosis (34 -36). First, it was validated that platelets bound immobilized fibronectin, as measured by staining for the platelet integrin CD41 (Fig. 8A). Next, platelet adhesion to fibronectinbound neutrophils was quantified. The results show that induction of NETosis with PMA potentiated platelet adhesion. Pretreatment of neutrophils with APC reduced the degree of platelet adhesion by more than 50%, as measured by either the total degree of CD41 staining (Fig. 8A) or the overlap of CD41 with DNA (Fig. 8B).

LDH release is reduced by APC pretreatment
Experiments were next designed to test if neutrophil cell death was inhibited by APC. Lactate dehydrogenase (LDH) is an enzyme found in almost all cells including neutrophils, playing an integral role in changing lactate to pyruvate, and LDH release is a marker of cell death (48). Neutrophil cell death was induced with PMA and quantified as percent cytotoxicity, where under vehicle conditions it resulted in nearly 90% cyto-toxicity (Fig. 8C). When neutrophils were pretreated with APC, the percentage of cytotoxic cells was significantly reduced by nearly 40%.

MPO plasma levels are inhibited by APC infusion in a nonhuman primate model of sepsis
As an in vivo proof-of-concept experiment, we studied whether the administration of pharmacological doses of APC under pathophysiological conditions could reduce neutrophil activation using an established baboon model of bacterial sepsis (40,49,50). In this model, either pharmacological doses of APC (3 mg/kg initial bolus followed by infusion of APC at 16 g/kg/ min for 5.4 h) or of vehicle were infused prior to challenge with a lethal dose (LD 100 ) of Escherichia coli (ATCC ® 12701) that resulted in a total concentration of 1-2 ϫ 10 10 colony forming units (cfu)/kg. Using this model, the effect of APC on neutrophil MPO release was assessed. MPO is a critical antimicrobial enzyme released by activated neutrophils as well as a biomarker for NET formation (48,51). As seen in Fig. 9

Discussion
Here, we report that in a purified neutrophil system and in an initial proof-of-concept experiment using an established in vivo nonhuman primate model of sepsis, using therapeutic concentrations of APC was found to inhibit neutrophil activation and induction of cell death. Utilizing a mechanistic in vitro approach, APC inhibition of PKC-and PI3K-dependent NETosis was found to be in part dependent on the neutrophil receptors Mac-1, EPCR, and PAR3 (Fig. 10). EPCR and ␤ integrins expressed on neutrophils are known to bind APC resulting in a reduction in migration (14,15,18), and our results now suggest a role for the receptors EPCR and Mac-1 in APC-mediated inhibition of NETosis. Based on molecular studies with endothelial cells and podocytes, APC ligates EPCR and cleaves PAR1 and PAR3 at noncanonical residues differing from the thrombin cleavage sites (19,24). Together with our results, these studies substantiate the biased agonism hypothesis in which thrombin and APC differentially cleave the same receptors, thus

. APC inhibits platelet adhesion to NETs and LDH release.
A and B, acid-washed glass coverslips were coated with 20 g/ml fibronectin and then blocked with denatured BSA (5 mg/ml). Purified human neutrophils (2 ϫ 10 6 /ml) were plated on the coverslips and allowed to adhere for 30 min at 37°C. Cells were washed once with PBS and incubated with HBSS or 300 nM APC 30 min at 37°C. Samples were then washed once with PBS and subsequently treated with HBSS or PMA (10 nM) for 3 h at 37°C. The samples again were washed once with PBS and incubated with autologous platelets (2 ϫ 10 7 /ml) and allowed to adhere for 45 min at 37°C. Samples were then washed once with PBS and fixed with 4% PFA and were incubated overnight with polyclonal rabbit anti-CD41 antibody (1:100). They were then incubated with Hoechst 33342 (1:1000) and secondary antibody Alexa Fluor 488 goat anti-rabbit (Invitrogen) (1:500). Images were normalized to secondary antibody-alone images and analyzed in a custom MATLAB program to quantify each pixel-positive signal as (A) the area of FITC (CD41) per image. In (B) each pixel positive for both CD41 and DNA signals per image was quantified. *, p Ͻ 0.001 versus vehicle ϩ PMA. C, purified human neutrophils (4 ϫ 10 5 /ml) were incubated in solution with 300 nM APC for 30 min at 37°C. Samples were then incubated with HBSS or PMA (20 nM, final) and incubated for 3 h at 37°C. 10ϫ lysis buffer was added to duplicate columns and incubated for 45 min at 37°C. Supernatants from samples and the LDH-positive controls were transferred to new 96-well plates prior to incubation with reaction mixture for 30 min at 37°C. The reaction was stopped and absorbance was measured at 490 nm and 680 nm. Percent cytotoxicity was quantified per the manufacturer's instructions (Thermo Fisher). *, p Ͻ 0.05 versus vehicle ϩ PMA. Data are mean Ϯ S.E. n ϭ 3. initiating different signaling pathways to culminate in disparate effects on cell survival (11,19,23,52). PAR3 was identified in 1997 as the second thrombin receptor in humans, following PAR1 (53). Initially, PAR3 was thought to be a nonsignaling receptor. PAR3 was shown to replace PAR1 and facilitate PAR4 activation by thrombin (54). PAR1 and PAR3 heterodimerize on endothelial cells to help explain thrombin-mediated signaling (55). Thus far, it has been shown that thrombin, APC, and FXa can cleave PAR3 (24,53,56,57). APC-mediated cytoprotective signaling on neuronal cells, podocytes, and endothelial cells occurs in part via the activation of PAR3 and requires EPCR (24, 58 -60). It remains to be determined whether APC generated in plasma from endogenous protein C inhibits neutrophil activation or extends cell survival.
In neutrophils, basal mRNA and cell surface expression of PAR3 remains relatively constant in response to various inflammatory agonists (16,61). APC-mediated PAR3 signaling on neutrophils inhibited NETosis, whereas equimolar concentrations of thrombin-mediated PAR3 signaling had no effect on NETosis. The striking finding that the PAR3 P3R peptide inhibits NETosis implies that neutrophil-expressed PAR3 biased agonism is sufficient to regulate neutrophil survival.
With the novel observation that PAR3 but not PAR1 had an essential role in APC-mediated inhibition of NETosis, it is an open question whether there are two signaling pathways initiated that feed into one common path under pathophysiological conditions. It is possible that Mac-1 and a PAR3-EPCR complex each initiate distinct pathways to activate inhibitory signaling; alternatively, Mac-1/PAR3/EPCR may form a larger complex to initiate signaling, similar to the clustering of EPCR/ PAR1 in caveolae found on endothelial cells (62). Our NETosis inhibition data add to the repertoire of mechanisms by which administration of APC exerts anti-inflammatory effects.
Prior to its removal from clinical practice, recombinant APC was used in adult patients with severe sepsis based on the rationale that the anticoagulant and anti-inflammatory effects of APC would provide a clinical benefit in this patient population (10). Additional benefits may derive from the ability of APC to cleave cytotoxic extracellular histones in animal models of sepsis (40), supporting the concept that APC exerts its antiinflammatoryeffectsthroughbothcell-dependentandcell-independent pathways. To date, it has not been clear whether APC has a direct cytoprotective action on neutrophil cell survival. As our results demonstrated, APC inhibited neutrophil cell death but did not affect elastase release (supplemental Fig. S4). This raises the possibility that APC may exert anti-inflammatory activity by inhibiting cellular functions that potentially drive or amplify pathophysiological disease, whereas APC may preserve functions required for homeostatic surveillance. Under select inflammatory conditions, the interactions between platelets and NETs are hypothesized to potentiate immunothrombosis; therefore, we tested if APC reduced platelet-NET binding (34,35,63). Our results suggest that through inhibition of neutrophil cell death, APC reduces platelet-NET binding and thus may reduce the prothrombotic phenotype of NET-promoted immunothrombosis.
Because APC has well characterized anticoagulant and multiple cell signaling actions, it should become possible through rational design to select for specific functions of APC (11). Thus, one of our goals is to develop APC mutants that can provide improved treatments for a variety of disease states, as in sepsis where these proteins may allow for the better control of inflammation and immune system homeostasis without compromising hemostatic mechanisms (11). Introduction of single or double amino acid substitutions within APC can significantly modulate its anticoagulant or anti-inflammatory functions. Proof-of-concept studies using APC mutants have demonstrated beneficial selection for specific functions of APC on a variety of cell types and through use of in vivo mouse models, where select mutants have now advanced to clinical trial (47,58,(63)(64)(65)(66)(67)(68)(69). Using the protein C database ProCMD (70), five of the sites where mutations were introduced into APC for our study were found to occur in humans; however, the amino acids introduced in our study did not reflect the mutated amino acid replacements seen in affected humans. Multiple APC mutants inhibited PMA-induced NETosis. Select mutations in the protease domain, as in 3K3A-APC, yielded a normal degree of inhibition of NETosis, similar to rhAPC. This APC signaling selective mutant is in clinical trials for ischemic stroke (71). In contrast, multiple mutations in the protease domain have a partial or total loss of function, demonstrating that certain residues in the protease domain are essential for the ability of APC signaling to inhibit NETosis. Moreover, mutations within the Gla domain of APC result in the partial or entire loss of function of APC, highlighting the additional importance of the Gla domain to mediate both the binding and orientation of APC on neutrophils to inhibit NET formation. These findings support the rationale for development of APC mutants that retain cytoprotective and anti-inflammatory functions but shed anticoagulant activity for potential clinical use in selected inflammatory disease states. In vitro, select stimuli including cytokines, bacteria, the PKC-agonist PMA, and the platelet secretome induce PKC-dependent NETosis. Our observations show a mechanism whereby proteolytically active APC requires Mac-1, EPCR, and PAR3 to induce intracellular, cytoprotective signaling that results in the downstream inhibition of NET formation.
Because the in vitro studies show that APC modulates neutrophil cell death, we determined whether infusion of APC at a pharmacologically relevant dose could reduce neutrophil activation using an established nonhuman primate model of bacterial sepsis as an initial proof-of-concept experiment. 3K3A-APC is currently being studied at high doses, up to 0.5 mg/kg, in healthy adults and is well tolerated (71). Here we showed that baboons infused with lethal doses of E. coli, when using recombinant wild-type APC at 3 mg/kg followed by infusion for Ͼ5 h at 0.96 mg/kg/h, inhibited MPO release. This suggests a link in vivo between APC and neutrophil function and points toward future in vivo studies. Of potential relevance, a previous study observed that in heart surgery patients, a negative correlation between neutrophil MPO activity in the myocardium and generation of APC was observed post surgery (72).
In conclusion, our discoveries uncover a novel mechanism by which pharmacological concentrations of APC utilize neutrophil-expressed Mac-1, EPCR, and PAR3 to activate intracellular signaling that culminates in the inhibition of PMA-induced NETosis. Furthermore, a pilot proof-of-concept experiment using a nonhuman primate model of bacterial sepsis demonstrates that infusion of APC reduced neutrophil activation in vivo. Fundamental questions remain about how these three receptors ligate APC and initiate the specific intracellular signaling pathways that are activated to result in inhibition, and the role of endogenous APC in reducing NETosis in vivo. Future work will aim to identify the precise signaling networks that result in this inhibition of NETosis by APC.

Reagents
Human plasma-derived APC was a gift from the American Red Cross (73). Recombinant human APC and mutant APC proteins were cloned, expressed, and purified as described previously (64 -66).

Preparation of leukocytes
Human leukocytes were purified as described previously (75). Briefly, human blood was drawn in accordance with an Oregon Health & Science University Institutional Review Board-approved protocol from healthy donors by venipuncture into citrate-phosphate-dextrose (1:7 v/v). Blood was layered over an equal volume of Polymorphprep TM and centrifuged at 500 ϫ g for 45 min at 18°C. The lower layer containing neutrophils was subsequently collected and washed with Hank's Balanced Salt Solution (HBSS) by centrifugation at 400 ϫ g for 10 min. To remove red blood cells from the sample, the pellet was resuspended in sterile H 2 O for 30 s, followed by immediate addition of 10ϫ PIPES buffer (250 mM PIPES, 1.1 mM CaCl 2 , 50 mM KCl, pH 7.4), to dilute the leukocyte suspension. After centrifugation at 400 ϫ g for 10 min, the pellet was resuspended in HBSS containing 2 mM CaCl 2 , 2 mM MgCl 2 , and 1% w/v bovine serum albumin (BSA).

Platelet isolation and secretome preparation
Washed platelet isolation was carried out as described previously (76). Thrombin-induced (1 unit/ml) platelet aggregation was allowed to proceed for 15 min at 37°C after its onset and the platelet secretome, containing platelet-derived soluble molecules, was isolated as described previously (77). Briefly, activated platelets were pelleted by sequentially centrifuging twice at 1000 ϫ g for 10 min in the presence of a protease inhibitor mixture to prevent protein degradation. Platelet pellet was discarded and the autologous supernatant (secretome) was collected and recentrifuged at 13,000 ϫ g for 10 min to remove any microparticles. Hirudin (40 g/ml) was added to the supernatant to neutralize thrombin activity. The platelet supernatant was used to stimulate NETosis as described below.

Immunofluorescence microscopy
Acid-washed glass coverslips were coated with 20 g/ml fibronectin and then blocked with denatured BSA (5 mg/ml). Initially, purified human neutrophils (2 ϫ 10 6 /ml) were plated on the coverslips for 30 min at 37°C. In select experiments, APC or thrombin was pretreated with PPACK dihydrochloride (40 M) for 15 min prior to incubation with neutrophils. Samples were subsequently treated with HBSS or phorbol 12-myristate 13-acetate (10 nM) for 3 h at 37°C prior to fixation with 4% paraformaldehyde (PFA) followed by incubation with blocking buffer (PBS containing 10% FBS and 5 mg/ml Fraction V BSA). Cells were stained with primary antibody (1:100 or 1:250) in blocking buffer at 4°C overnight. Secondary goat antimouse IgG antibodies conjugated with Alexa Fluor 546 (1:500) and Hoechst 33342 (10 g/ml) in blocking buffer were added and incubated for 2 h in the dark. Coverslips were mounted onto glass slides using Fluoromount G and visualized with a Zeiss Axiovert fluorescence microscope (Axio Imager; Carl Zeiss, Göttingen, Germany) equipped with a 40ϫ/1.3 numeri-cal aperture (NA) oil immersion objective and an air-coupled lens providing Köhler illumination at an NA of 0.17.
To examine platelet binding, following fixation with 4% paraformaldehyde, platelets were permeabilized with blocking solution (1% BSA, 0.1% SDS in PBS). Platelets were then stained with indicated antibodies (1:100) overnight at 4°C in blocking buffer. Secondary Alexa Fluor antibodies (1:500) with Hoechst (1:1000) were added in blocking buffer for 2 h in the dark. After mounting with Fluoromount G, platelets and neutrophils were imaged on a Zeiss Axiovert fluorescence microscope (Axio Imager; Carl Zeiss) equipped with a 63ϫ NA oil immersion objective and an air-coupled lens providing Köhler illumination at an NA of 1.40. All images were recorded with a charge-coupled device camera (AxioCam MRc5 12-bit camera; Carl Zeiss) under software control by Slidebook 5.5 (Intelligent Imaging Innovations, Denver, CO) as described previously (76).

Neutrophil static binding assay
Acid-washed glass coverslips were coated with BSA (denatured and filtered 5 mg/ml), IgG (20 g/ml), fibronectin (20 g/ml), or APC (100 g/ml) for 1 h, followed by washing and then blocked with denatured BSA (5 mg/ml) for 1 h. Initially, purified human leukocytes (2 ϫ 10 6 /ml) were plated on the coverslips for 1 h at 37°C. Cells were then washed again with PBS and fixed with 4% PFA followed by mounting and imaging.

APC binding assay
Purified human leukocytes (2 ϫ 10 6 /ml) were stimulated with HBSS or PMA (10 nM) for 3 h at 37°C on fibronectincoated glass coverslips. For colocalization experiments, adherent cells were washed and incubated with vehicle (HBSS buffer) or APC (300 nM) for 15 min at 37°C. Subsequently, samples were washed with PBS and fixed with 4% PFA followed by incubation with blocking buffer (PBS containing 10% FBS and 5 mg/ml Fraction V BSA). Cells and coagulation factors were stained anti-PC/APC Abs (100 g/ml) in blocking buffer at 4°C overnight. Secondary goat anti-mouse IgG antibodies conjugated with Alexa Fluor 488 (1:500) and Hoechst 33342 (10 g/ml) in blocking buffer were added and incubated for 2 h in the dark. Coverslips were mounted onto glass slides and visualized.

NET formation assays
In select experiments, purified human leukocytes (2 ϫ 10 6 / ml) were plated on fibronectin-coated coverslips for 30 min at 37°C. APC (300 nM) and thrombin (300 nM) were incubated for 15 min in the presence or absence of PPACK (40 M) or ␣1-antitrypsin (20 M) prior to incubation with leukocytes for 30 min at 37°C. For select experiments, neutrophils were pretreated with platelet activating factor (PAF, 100 nM) for 30 min at 37°C prior to incubation with 100 nM APC for an additional 30 min at 37°C.

Platelet adhesion assay to leukocytes and NETs
In select experiments, purified human leukocytes (2 ϫ 10 6 / ml) were plated on fibronectin-coated coverslips for 30 min at 37°C. Leukocytes were pretreated with APC (300 nM) for 30 min at 37°C prior to washing and stimulation with HBSS or PMA (10 nM) for 3 h at 37°C. Samples were then washed once with PBS and incubated with autologous purified platelets (2 ϫ 10 7 /ml) and allowed to adhere for 45 min at 37°C. In select wells, platelets were allowed to adhere to a fibronectin-coated coverslip in the absence of neutrophils. Samples were then washed once with PBS and fixed with 4% PFA followed by staining, mounting, and imaging.

Image analysis
The fluorescent intensities of each image were adjusted based on signals detected in leukocyte or platelet samples in the absence of primary antibodies. Quantification of fluorescent images was performed using a custom algorithm in MATLAB (MathWorks, Natick, MA). This algorithm first establishes global thresholding parameters based on a training set of fluorescence image data to then systematically quantify intensity profiles across treatment conditions. The area of the total field of view is calculated to be 12510 m 2 . Quantitative comparison of treatment conditions was achieved by the normalization of fluorescence intensities per image to determine the area of DNA or respective fluorescent channels.

LDH activity assay
Primary human leukocytes (4 ϫ 10 5 /ml) were purified and incubated in solution with 300 nM APC for 30 min at 37°C in a 96-well plate format. In brief, samples were then incubated with HBSS or PMA (20 nM, final) and after mixing were incubated for 3 h at 37°C. 10ϫ lysis buffer was added to duplicate columns and incubated for 45 min at 37°C. Supernatant from all triplicate samples, including lysed samples and an LDH-positive control were transferred to new 96-well plates prior to incubation with reaction mixture for 30 min at 37°C. The reaction was stopped and absorbance was measured at 490 nm and 680 nm. Percent cytotoxicity was quantified as per the manufacturer's instructions (Thermo Fisher).

Neutrophil elastase activity assay
Primary human leukocytes (4 ϫ 10 5 /ml) were purified and incubated in solution with 300 nM APC or neutrophil elastase inhibitors; CMK (475 M) or Sivelestat (46 nM) for 30 min at 37°C in a 96-well plate format. Samples were then treated with vehicle, fMLP (10 M, final) ϩ cytochalasin B (5 g/ml, final), or PMA (50 nM, final) for 2 h at 37°C. Supernatants were collected and incubated with substrate solution, protected from light for 1.5 h at 37°C. Fluorescence was measured at excitation 485 nm, emission 525 nm. Elastase was quantified as mU/ml using the standard curve generated and in comparison to the neutrophil elastase-positive control as per manufacturer's instructions (Cayman Chemical).

Nonhuman primate model of bacterial sepsis
Studies were reviewed and approved by the Institutional Animal Care and Use Committee at Oklahoma Medical Research Foundation. In this established baboon (Papio anubis) model of sepsis, animals (n ϭ 4) were initially dosed with 3 mg/kg bolus of APC T Ϫ 10 min. At T Ϫ 0 min to T ϩ 320 min, APC was infused at 16 g/kg/min i.v. A lethal dose (1-2 ϫ 10 10 cfu/kg) of E. coli, ATCC 12701 serotype O86:K61(B7) was infused intravenously over 2 h (from T 0 to T Ϫ 120 min). The control group (n ϭ 8) consisted in historical experiments where animals were challenged with the same dose of bacteria under similar experimental conditions but not treated with APC. Blood samples were obtained at 0, 2, 4, 8, and 24 h and plasma samples were isolated. MPO in plasma samples was determined using Fluoro MPO (fluorescent myeloperoxidase detection kit) (Cell Technology, Fremont, CA) as per the manufacturer's instructions.

Statistical analysis
Two-way analysis of variance with Tukey's post hoc correction or one-way analysis of variance with Welch's post hoc correction was used to assess statistical significance among parameters across multiple normally distributed cell parameters. p values of 0.05 or less were considered statistically significant. At least four images per treatment condition were obtained and experiments were performed in triplicate. All values are reported at means Ϯ S.E. of the mean unless otherwise stated.