The formation of neutrophil extracellular traps is a novel host defense strategy in addition to the well characterized phagocytosis of neutrophils. Many groups have shown that viruses [20
], bacteria [8
] and fungi [21
] stimulate neutrophils to release extracellular DNA, although it has been suggested that this process is an incidental component of neutrophil lysis or a hijacking of host pathways by the pathogen [23
]. However, there is increasing evidence that NETosis is a highly orchestrated mechanism of cellular defense and is under tight control [24
]. Irrespective of the mechanism by which chromatin and other cytotoxic components are released from neutrophils and assembled into NETs, the resultant damage to host tissue has been shown to promote inflammatory diseases such as cystic fibrosis [25
], asthma [26
], systemic vasculitis [7
], systemic lupus erythematosus [27
] and other acute and chronic diseases [9
]. In unstimulated neutrophils, a set of homologous serine proteases are stored in azurophilic granules and their enzymatic activities are subject to intracellular compartmentalization and endogenous inhibitors [30
]. Upon stimulation with PMA or bacteria they are released and selectively associated with extracellular DNA [14
]. As extracellular proteins, the activity of these enzymes may no longer be under the control of inhibitors and, therefore, unregulated proteolysis could be initiated and prolong an inflammatory response [32
]. Alternatively, association of these proteases with NETs could provide a means to compartmentalize and minimize their destructive proteolytic activities. It is worth noting that serine proteases seem to be selectively targeted to NETs since caspases that are important components of inflammasomes, and matrix metalloproteases were not detected on NETs. One simple explanation is that NE, CG and PR3 are positively charged basic proteins which have high affinity to the negatively charged DNA. Previous studies by Dubois and colleagues have used highly selective fluorescent substrates of NE, PR3 and CG to assess the proteolytic activity in NETs from Cystic Fibrosis sputum and from neutrophils stimulated with S. aureus
and P. aeruginosa
]. DNase treatment of sputum dramatically increased its NE activity, but not the activities of CG and PR3. They also determined that nuclease treatment of extracellular DNA from stimulated neutrophils caused a ~2.5 fold increase in soluble proteolytic activity relative to unstimulated cells, and that the contribution of each enzyme to the overall activity was approximately 40% for NE, 40% for PR3 and 20% for CG. Their study demonstrated that NETs are enriched in serine proteases and NE and PR3 are main contributors to NETs associated proteolytic activity.
In our study, we utilized a highly sensitive and unbiased substrate profiling method to characterize the entire proteolytic signature associated with NETs. MSP-MS is an ideal technology to profile complex biological samples because the substrate population consists of a defined set of peptides and, therefore, cleavage sites can be directly linked to a specific enzyme. Furthermore, the unbiased design of the peptides and combination into equimolar pools ensures that proteolytic activity from multiple proteases can be monitored simultaneously without prior knowledge of the substrate specificity. The reproducibility of both the sample preparation and the MSP-MS assay was evident as many of the peptides cleaved in three independent donor samples in this study were identical. Any variation between donors could be due to differences in genetic background, or immune status of each individual. Inter-individual variability has recently been highlighted by Barrientos and coworkers, where NET-associated proteolytic activity between donors differed by as much as 9-fold [33
In our study, all of the proteolytic activity released from PMA-activated neutrophils could be attributed to NE, CG and PR3 activity. However, ~70% of cleavage sites were likely to be the result of NE activity and CG and PR3 contributed to the remaining 25% and 5%, respectively. As expected, when NE was immunodepleted from the NET extracts, CG became the dominant source of proteolytic activity. The remaining proteolytic activity was likely to be the product of PR3 and to a lesser extent NSP4 and other as yet unidentified NET associated proteases. Interestingly, PMA stimulation of neutrophils resulted in PR3 activity that was considerably lower than CG while other studies using bacterial stimulated neutrophils show the reverse. These data indicate that protease composition in NETs may differ depending on the type of stimulation. Khandpur and coworkers analyzed the composition in NETs following alternate stimulation and found differences in both the protein levels and protein composition [33
Proteomic analysis was used to identify proteins released from the NETs, many of which have been observed in a previous study [14
]. Interestingly, we found that NE is tightly bound to DNA and was not found in samples that lacked nuclease treatment. CG was found in all samples independent of treatment regimes while PR3 and NSP4 were not detected. Since robust PR3 activity was evident in the MSP-MS assay these data indicate that a functional activity assay has superior sensitivity than proteomic based methods for detecting proteases.
Previously, Proteomic Identification of Protease Cleavage Sites (PICS) was utilized to generate extended substrate specificity profiles of NE, CG and NSP4 [15
]. In these studies, proteome derived peptides were used as the substrate library and cleavage by the neutrophil serine proteases was monitored by mass spectrometry. As was observed for the PS-SCL method, the substrate profiles using PICS strongly correlated with our MSP-MS data (Table S1
). However, our study is the first to directly compare the substrate specificity of all four homologous neutrophil serine proteases under identical assay and instrumentation conditions, and reveals distinct cleavage sites that are uniquely cut by each enzyme. Traditionally, the neutrophil serine proteases were considered to have redundant functionalities based on their ability to hydrolyze structural proteins such as collagen, laminin, fibronectin and elastin [30
]. However, out of the 180 bonds that were hydrolyzed by the four proteases only 28 cleavage sites were shared between two or more enzymes. Therefore, although our substrate profiling was performed using a synthetic peptide library, it is likely that many in vivo
protein substrate cleavage sites exist that are unique to each neutrophil protease.
Successful substrate prediction tools require an experimentally determined list of protein or peptide substrates to identify sequence and structural features important for protease recognition. The predictive power for identifying natural substrates can be improved by incorporating negative substrates. Using the MSP-MS assay, any peptide bond that is cleaved by a neutrophil serine protease is a true substrate while every bond that is not cleaved is considered a true negative substrate. Using data generated in this study it may be possible to identify and track endogenous substrates of these proteases during neutrophil lysis and/or NET formation which will provide us with a greater understanding of roles of these enzymes in immune response and inflammation.
In this study, it is evident that targeting of NE on NETs could minimize adverse effects of unregulated proteolysis associated with tissue damage and inflammation. The selective cleavage sequences of the neutrophil serine proteases identified in this study will be valuable for designing substrates, inhibitors, activity based imaging agents [36
] and protease-activated prodrugs [37
]. In addition, the substrate signature of NETs-associated protease activity can be monitored as a biomarker for inflammatory diseases driven by extracellular neutrophil proteases.