Osteopontin is associated with neutrophil extracellular trap formation in elderly patients with severe sepsis

Severe sepsis may rapidly evolve in multiorgan dysfunction and septic shock with high mortality risk. Rate of sepsis-related deaths increases with age: people aged ≥85 show the highest risk, five times that of the class of age 65–74.1 The current pathophysiological paradigm of sepsis indicates a critical role for abnormal early inflammatory response in determining adverse evolution of sepsis.2, 3 The activation status of inflammatory cells—mainly neutrophils—is not limited to pathogen killing, but also contributes to organ injury/healing.4 Their recruitment, polarization and activation are tightly orchestrated by a complex signal network, on which our research group has recently focused. Subanalyses of a validated cohort (ALBIOS trial) indeed revealed intriguing association of sepsis-related outcome with known biomarkers of neutrophil recruitment (e.g. osteopontin [OPN] and resistin) and activation (myeloperoxidase [MPO]).4-6 Neutrophils are supplied also by an ever-increasing array of weapons. Especially neutrophils extracellular traps (NETs) are attracting a growing interest. NETs generate through the extrusion of granule and nuclear products (i.e. neutrophil elastase [NE], MPO and DNA) from neutrophils. They classically generate a defensive net around pathogens,7 but an association of bloodstream concentration with organ damage has been reported in experimental model of sepsis and clinical setting as well.8, 9 Nevertheless, the signalling network underlying NET generation is only partially elucidated. Here, we focused on the potential association between neutrophil activation—mainly NET formation—and osteopontin (OPN) in sepsis. We investigated whether a relationship between NET formation, OPN and sepsis-related outcome exists in a clinical cohort of elderly septic patients. We further explored in vitro the effect of OPN on NET generation and the underlying intracellular pathways. Clinical cohort includes selected patients from the 'Genoa-BASH SEPSIS' study. Original study design, inclusion and exclusion criteria have been reported elsewhere.10 (See the Appendix S1). See the Appendix S1. See the Appendix S1. Table S1 summarizes clinical and biochemical characteristics of the 43 patients enrolled in this subanalysis. Well representative of the original cohort, patients were elderly (median age of 83 years) with a slight prevalence of men (60.5% of men) and relatively low comorbidity burden as assessed by Cumulative Illness Rating Scale (CIRS; median value 12) and Charlson comorbidity index (median value 3). No differences were recorded across sexes. When circulating levels of NETs were analysed, they showed a significant progressive increase with quick SOFA (p for trend .028), but not with Kelly scale (the latter is likely biased by some outliers) (Figure S1A,B and Table S2). OPN was instead associated with both quick SOFA and Kelly scale (p for trend .032 and .031, respectively) (Figure S1C,D, Table S2). NETs, but not OPN increased linearly with creatinine (p for linear regression .012) and decline in glomerular filtration rate (p for linear regression .033), whereas they were both associated with indexes of patient frailty: NETs with CIRS scale (p for linear regression .014) and OPN with Barthel index (p for linear regression .003) (Figure S1E–H and Table S3). At baseline, circulating levels of NETs and OPN did not display any significant association with neutrophil count (Figure 1A) and other haematological variables (i.e. haemoglobin, white blood cell and platelet counts) (Table S3) but they were eac other correlated (p for linear regression .012) (Figure 1B, and Figure S2). As expected, the other products from neutrophil granules correlated each other (Figure S2). Twenty-two patients died during overall follow-up. Of them, seven (16.3%) died within the first 30 days, whereas four additional deaths were recorded till 90 days from enrolment (Figure 1C). At linear regression analyses, time to event (applied to the twenty-two patients died during the follow-up) showed a negative trend toward both circulating NETs (p = .057) and OPN (p = .068) (Figure 1D). In addition, higher baseline NETs but not OPN characterized patients experiencing mortality (both 30- and 90-day groups) (Figure 1E–H). An exploratory analysis on NETs-related mortality was then built through logistic regression models for 30- and 90-day mortality. High circulating NETs at enrolment conferred an excess risk of death at 30 and 90 days (ORs 4.5 [1.1–17.6] and 3.8 [1.2–12.4], respectively) (Table S4). Such an independent association was also observed for Kelly scale (2.5 [1.2–5.3] and 2.4 [1. 2–4.7], for 30- and 90-day mortality, respectively), and qSOFA (OR 3.9 [1.3–11.3] and 2.6 [1.1–6.1], respectively) (Table S4). NET variation over the first week from enrollment retained a similar positive trend with the excess risk of mortality toward 30- and 90-day follow-up (Table S4). Both NETs and NET variation over the first week also seemed to improve discriminatory ability of the validated clinical tools qSOFA and Kelly scale (Table S4). Internal validation with bootstrap was in line with those observed in the original cohort (Table S5). As additional outcome, both NETs and NET variation over the first week even showed an excess risk of long-term mortality with a similar improving effect on discriminatory ability (Table S4). Clinical relationship between circulating NETs and OPN suggested a potential molecular interaction between them. We therefore examined NETs generation by neutrophils purified from healthy donors in vitro. OPN was associated with significant increase of NETs—like positive control—with around 60% of positive cells (p < .001) at pathological concentration (above 200 ng/mL) (Figure 2A,C,G). Coincubation with the PAD4 inhibitor GSK484 blunted this effect (p < .001 and .003 for 200 and 1000 ng/mL of OPN, respectively) (Figure 2A,F,H). Inhibition of PI3K (LY29400210) and MEK 1/2 (U0126) was tested to provide insights also about the potential intracellular pathways triggered by OPN. They both blunted NET formation induced by OPN (p < .001 for both PI3K and MEK1/2 inhibition at 200 and 1000 ng/mL of OPN) (Figure 3A–D). Likewise, NADPH oxidase was tested as an additional potential pathway. Its inhibition effectively bunted NETs generation induced by PMA and OPN at both 200 and 1000 ng/mL (p = .005, .026 and .029, respectively) (Figure 4A–H). Neutrophil migration toward OPN displayed the typical bell-shaped dose–response graphic of chemo-attractants (Figure S3A,B). The magnitude of neutrophil migration induced by 200 and 1000 ng/mL of OPN mirrored the effect generated by a known neutrophil chemoattractant (1 nM CXCL8) (Figure S3A). To discriminate between chemotactic/chemokinetic function a checkerboard analysis with increasing concentrations of OPN above and below the filter was performed (Figure S3B). Direct gradient of migration across the filter finally indicated a chemotactic process rather than random chemokinesis. We then established by Western blot the activation of PI3K/AKT and ERK1/2 pathways (p = .039 and .035, respectively Figures S4A,C and S5), and in parallel the selective inhibition of neutrophil migration to OPN (200 ng/mL) by adding their inhibitor LY294002 and UO126 (10 uM) (p < .001 for all comparisons). In line with the effect on neutrophil migration, polymerization of F-actin induced by OPN (p = .004) was blunted by both LY294002 and U0126 inhibitor (p = .032 and .025, respectively) (Figure 5B, D, E). See the Appendix S1. The heterogeneity of sepsis manifestations is nowadays an established concept. Different clinical phenotypes are being dynamically developed with the great contribution of artificial intelligence as well.11-13 However, little is known on the underlying molecular pathways leading to such a different clinical presentations and related outcomes. Here, we focused on the role of NETs as biomarker of dysregulated neutrophil phenotype in sepsis. Circulating NETs in elderly septic patients are here associated with patient frailty and disease severity. With the limits of a small sample size, we may also provide an exploratory association with excess risk of death at both short- (30 days) and long-term (90 days and up to 1000 days). These are confirmatory of previous published clinical studies14-17 and experimental evidence.17, 18 Persistent NET formation has indeed deleterious implications for the host during sepsis: thrombogenic mechanisms are widely known and include endothelial cell activation, platelet aggregation, enhanced release of tissue factor and factor XII; NETs may also amplify inflammatory response directly promoting release of cytokines, activation of complement system, and priming of CD4+ T cell, or acting as damage-associated molecular patterns.19 However, NETs themselves could be viewed as a double-edged sword.20 The patterns of neutrophil dysregulation are indeed quite complex and not so obvious. In our cohort, early reduction of NETs during the first week conferred an increased mortality risk, although with lower predictive value than baseline assay. This finding is in partial contrast with the NETCOV2 study, but the time line there considered for NET change was 1 day only.16 Elsewhere, the concepts 'NETosis exhaustion' over 7-day follow-up has been associated with sepsis severity and poor outcome.21 This has been framed as expression of a dysfunctional neutrophil phenotype—also characterized by impaired migration and prolonged neutrophil survival—induced by a low blood pH. Two waves—and patterns—of neutrophil activation have been then hypothesized: the first one where early NETosis occurs at site of infection to early contain the process; in the second one, later recruited neutrophils are activated by the systemic septic environment and become resistant to apoptosis. This would be the circulating pool on testing that accounts for 'NETosis exhaustion'. Although intriguing, this hypothesis implies NETosis as terminal process, which is not true. Vital and mitochondrial NETs are indeed generated without cell death, but they still need to be characterized outside experimental conditions.22 Furthermore, defective clearance of NETs should also be considered as effect of renal/hepatic dysfunction, common finding in sepsis.23 Anyway, a substantial change in neutrophil phenotype occurs in sepsis and it may sustain the prolonged immune dysfunction/suppression induced by sepsis that would accounts for recurrent, persistent and nosocomial infection, hospital readmission and ultimately long-term deaths. The clinical association between NETs and OPN prompted us to unveil their role in experimental conditions. OPN is indeed a multifunctional glycoprotein with chemoattractant properties for macrophages and neutrophils, traditionally ascribed to the integrin binding motifs in its sequence. In particular, the interaction with the integrin α9β1 expressed un neutrophil surface has been already described,24 but we cannot be able to observe here any modulation on neutrophil surface expression of CD11b, CD18 and CD62L. Involvement and detrimental effect of OPN in sepsis are widely known: it influences several immune functions not limited to adhesion/migration, but including differentiation, survival and phagocytosis.25 According to previous findings, we confirmed chemotactic properties of OPN, secondary to substantial F-actin polymerization via PI3K/AKT and MEK1/2. Those pathways even drive OPN-mediated NET generation through PAD4 signal. Chromatin decondensation, laminin meshwork destabilization, nuclear envelope rupture and extracellular DNA release are all PAD4-dependent processes in NET formation.26 In different mouse models of sepsis, PAD4 deletion was proven to suppress NET formation.27, 28 PAD-dependent mechanism was here confirmed as driver of OPN-induced citrullination in isolated human neutrophils. Furthermore, we observed the same blunting effect on OPN-induced NET generation after inhibition of NADPH oxidase, in line with previous studies.29 Despite the progress in neutrophil pathophysiology, we should acknowledge substantial limitations of the present study. Study population is limited to elderly patients, a subset not fully representative of sepsis pathophysiology, but validated in a previous larger study. As above disclosed, the small sample size might limit generalizability of clinical findings. However, they are in line with previous studies, whereas experimental evidence seems to confirm the association of main interest between NETs and OPN. Intrinsic limitations affect the experimental part also, which was carried out on samples from healthy donors, rather than patients themselves (on comparisons between them). Furthermore, we cannot establish neither the receptor linking OPN to PAD4 pathway, nor the downstream transcription factors involved in the signalling. Further in-depth studies are warranted to clarify the signalling pathway linking OPN/NET formation and whether septic environment may—eventually dynamically—modify neutrophil activation patterns. In conclusion, we provide here insights about an excess risk of death associated with NET burst in a cohort of elderly septic patients. This trend was compatible with biological data. We indeed demonstrated an active role of OPN NET formation in vitro and potentially in vivo during severe sepsis. OPN also induced neutrophil migration in vitro via defined intracellular pathways. Experimental and clinical findings need in-depth confirmation, but we lay here the groundwork for further considering dysregulated neutrophil homeostasis a leading, dynamic determinant of sepsis-related outcome. MB contributed to conducting experiments, acquiring data and analysing data. DV, PC, DD and SM contributed to conducting experiments and acquiring data. AT, DR, SM, DRG, AB, AV, LC, MM, EA, LL, FV, MB and FM contributed to designing research study, acquiring data and reviewing the manuscript. FC contributed to designing research study, conducting experiments, acquiring data, analysing data, providing reagents and writing the manuscript. This research was funded by a grant from the Rete Cardiologica of Italian Ministry of Health (#2754291) to Prof. F. Montecucco. The research was funded by a grant from the Internal Medicine Department of the University of Genoa to Prof. Federico Carbone. Aldo Bonaventura received a travel grant from Kiniksa Pharmaceuticals Ltd. to attend the 2019 AHA Scientific Sessions and honoraria from Effetti s.r.l. (Milan, Italy) to collaborate on the medical website www.inflammology.org, outside the present work. Data are not public, and they can be made available from the corresponding author upon reasonable request. Appendix S1 Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.