Substrate Stiffness Induces Neutrophil Extracellular Traps (NETs) Formation through Focal Adhesion Kinase Activation

Abstract

Neutrophils predominate the early inflammatory response to tissue injury and implantation of biomaterials. Recent studies have shown that neutrophil activation can be regulated by mechanical cues such as stiffness or surface wettability; however, it is not known how neutrophils sense and respond to physical cues, particularly how they form neutrophil extracellular traps (NET formation). To examine this, we used polydimethylsiloxane (PDMS) substrates of varying physiologically relevant stiffness (0.2-32 kPa) and examined the response of murine neutrophils to untreated surfaces or to surfaces coated with various extracellular matrix proteins recognized by integrin heterodimers (collagen, fibronectin, laminin, vitronectin, synthetic RGD). Neutrophils on higher stiffness PDMS substrates had increased NET formation and higher secretion of pro-inflammatory cytokines and chemokines. Extracellular matrix protein coatings showed that fibronectin induced the most NET formation and this effect was stiffness dependent. Synthetic RGD peptides induced similar levels of NET formation and pro-inflammatory cytokine release than the full-length fibronectin protein. To determine if the observed NET formation in response to substrate stiffness required focal adhesion kinase (FAK) activity, which is down stream of integrin activation, FAK inhibitor PF-573228 was used. Inhibition of FAK using PF-573228 ablated the stiffness-dependent increase in NET formation and pro-inflammatory molecule secretion. These findings demonstrate that neutrophils regulate NET formation in response to physical and mechanical biomaterial cues and this process is regulated through integrin/FAK signaling.

1. Introduction

The physiologic response to implanted biomaterials is initially predominated by cells of myeloid origin, specifically neutrophils and macrophages. While macrophages have been widely studied in the biomaterial context as the key orchestrator of the inflammatory response due to their well-documented phenotype switching and ability to directly regulate stem cell activity, neutrophils have largely been overlooked despite the fact that they are the first cell recruited en masse to an implant. Once at the site of injury, neutrophils play several important roles in early inflammation: secreting enzymes and cytokines, phagocytizing pathogens and debris, recruiting other immune cells through chemokines, and through neutrophil extracellular trap formation [1–3]. From an even more fundamental perspective, the mechanobiological cues regulating neutrophil activation have also not been well explored.

While biomaterials implanted into living tissues differ broadly in physicochemical properties, little is known on how these bulk and surface properties affect the initial immunological response. After biomaterial implantation, proteins and other molecules from the blood and tissues adsorb onto the biomaterial surface, facilitating biomaterial-cell interactions. The first cells to interact with these adsorbed proteins on the biomaterial surface are the neutrophils, which are one of the most abundant type of immune cell. Neutrophils are normally circulating and are the first cells to be recruited to areas of inflammation, injury, or infection [4]. Neutrophils release chemokines to recruit neutrophils or other immune cells to the site [5,6]. Neutrophils also undergo NET formation, which is important in the pathogenesis of many diseases characterized by sterile inflammation, but little is known about the triggers of this process outside bacteria stimuli. NET formation is a process by which neutrophils release neutrophil extracellular traps (NETs), web-like constructs of decondensed chromatins, antimicrobial proteins, and enzymatic activators of cytokine precursors that are released after both the nuclear and cell membrane rupture [7,8]. Traditionally, NET formation has been mechanistically studied after treatment with bacterial lipopolysaccharide (LPS), phorbol myristate acetate (PMA), and interleukin 8 (IL-8) in vitro [9–12]. NET formation is proceeded by activation of protein-arginine deiminase 4 (PAD4), which access the nucleus to citrullinate histones, triggering decondensation [13,14]. Decondensation ruptures the nuclear envelope, which allows cytoplasmic proteins to attach to the free chromatin [15]. The cell membrane ruptures and the NET, adorned with enzymes such as myeloperoxidase (MPO) and neutrophil elastase, is released into the microenvironment [6,16,17]. Recent work has only cemented the critical role of PAD4 in NET formation [18–24]. While discovered first in the context of their antimicrobial role, recent work has explored their extensive role in sterile inflammation. NETs are implicated in sequelae associated with autoimmune diseases, cardiovascular disease, and malignancy [25–29]. Activation of NET formation has been shown to correlate with increased inflammation, and indeed, neutrophils that have undergone NET formation release more cytokines [30,31]. However, beyond chemokine and cytokine cues, little is known about whether physical and chemical cues in biomaterials can trigger NET formation. We recently showed that surface wettability on microstructure titanium surfaces prevent the formation of NET formation when compared to smooth and microstructured hydrophobic titanium surfaces [31]. However, neutrophil-biomaterial interactions remain understudied and it is unclear how neutrophils interact with implanted biomaterials.

Previous studies have shown that physical cues differentially activate myeloid cell activity. We have previously demonstrated that, like macrophages, rough hydrophilic titanium surface cues reduce inflammatory activation of neutrophils. We have shown that multiple cell types regulate their behavior differentially on titanium substrates of varying roughness and hydrophilicity through the integrin/focal adhesion kinase (FAK) pathway [32–34]. Others have shown that substrate stiffness affects the phenotype of macrophages and dendritic cells and migration of neutrophils in vivo [35–37]. Interestingly, inflammatory cells are present in diverse pathological process involving tissue stiffness such as fibrosis, cancer, and atherosclerotic plaques [36,38–40]. A notable example of the relevance of the effects of mechanical stimuli on NET formation occurs in thrombosis. Compared to venous clots, arterial occlusions, where a greater pressure difference across a thrombosis is produced, generated far more rapid NET formation, possibly worsening the occlusive effects and subsequent tissue inflammation [41]. In the long term, it also appears that NET formation significantly contributes to the sequelae of the phenomenon of “inflammaging.” Padi4^−/− mice exhibited reduced age-related heart functional decline, lower histologic fibrosis of the heart and lungs [42]. Furthermore, in this study’s model of cardiac fibrosis, decreased collagen deposition and platelet recruitment were observed in knockout mice, and interestingly, treatment of wild-type mice with DNase 1 produced similar reductions in myocardial collagen as genetic PAD4 ablation. These studies underscore the importance of understanding NET formation in a mechanobiological sense, as tissue outcomes following injury or chronic damage can be determined by neutrophil responses.

In general, cells interact with their environment through integrins, transmembrane proteins that form obligated heterodimers of alpha and beta integrin subunits. Integrins work as membrane receptors, recognizing extracellular matrix (ECM) protein domains and facilitate the signal transduction from the ECM to the cells [43,44]. Integrins play a fundamental role in neutrophil adhesion and migration, as blocking integrins alters neutrophil migration and function [45–47]. Furthermore, blocking integrin subunits in a viral infection model decreased neutrophil NET formation, suggesting a role for integrins in NET formation, but this phenomenon is not well studied [48]. Moreover, stiffness and mechanical cues have not been examined as mechanisms of NET formation. The aim of this study was to elucidate the role of substrate stiffness on neutrophil activation and NET formation. We explored the role of integrin engagement through common ECM protein coatings in NET formation, and finally we elucidate the role of FAK in NET formation in response to substrate stiffness and protein coatings.

2. Experimental

2.1. Neutrophil Isolation

10-week-old male C57BL/6J (Stock #000664) or Padi4^−/− (Stock #030315) mice (Jackson Laboratory, Bar Harbor, ME) were used for this study in accordance with a protocol approved by the Virginia Commonwealth University Institutional Animal Care and Use Committee (Protocol: AD10001108). Mice were euthanized by CO₂ asphyxiation followed by cervical dislocation. Whole bone marrow was isolated by flushing the intramedullary canal of isolated femurs with phosphate-buffered saline (PBS, Thermo Fisher Scientific, Waltham, MA). Erythrocytes were removed from the marrow isolate using ACK Lysing Buffer (Quality Biological Inc., Gaithersburg, MD). Neutrophils were then isolated by centrifugation using Histopaque 1077 and 1119 (Sigma-Aldrich, St. Louis, MO) [49]. Viability on isolation was measured by Trypan Blue (Sigma-Aldrich) staining and was found to be >98% for all cultures. Purity was confirmed by flow cytometry. Following pretreatment with anti-CD 16/32 (Fc receptor, BioLegend, San Diego, CA) to prevent non-specific fluorescence, neutrophils were identified as CD45+/CD11b+/Ly6G+ (BioLegend). A purity of approximately 90% was obtained for each experiment. Neutrophils were isolated from pooled marrow from eight mice per experiment.

2.2. Neutrophil Culture on PDMS substrates and Protein Coating

PDMS substrates (CytoSoft®) with a range of defined elastic moduli (0.2, 2, 8, 16, 32 kPa) on glass bottom 24-well plates (Advance Biomatrix, San Diego, CA) were used to determine the effect of stiffness on neutrophil activation and NET formation. Briefly, PDMS substrates were washed twice with Dulbecco’s PBS without calcium or magnesium (Thermo Fisher Scientific) and incubated for 1 hour in Dulbecco’s Modified Eagle Media (DMEM) (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific). After incubation, PDMS substrates were washed twice with warm PBS before being seeded with neutrophils at a plating density of 100,000 cells/cm². Neutrophils were incubated for 4 hours at 37°C, 5% CO₂, and 100% humidity. After incubation, conditioned media was collected for analysis.

To determine the effect of integrin engagement on neutrophil NET formation, PDMS substrates were incubated with 0, 5, or 10 μg/mL solutions of extracellular matrix proteins: laminin [CellAdhere™], (STEMCELL Technologies, Vancouver, Canada), type I collagen [PureCol®], fibronectin, vitronectin, synthetic RGD peptide [Peptite-2000®] (Advance Biomatrix) or complete culture medium for 1 hour at room temperature. After incubation, extracellular matrix solutions were aspirated and coated PDMS substrates were washed twice with warm PBS. Neutrophils then were plated at 100,000 cells/cm² and incubated for 4 hours. After incubation, media was aspirated, cells washed twice with warm PBS and neutrophil morphology assessed as described below.

2.3. Neutrophil PAD4 and FAK inhibition

To elucidate the effect of PAD4 inhibition on substrate-induced NET formation, neutrophils were treated with or without 50nM GSK484 (Cayman Chemical, Ann Arbor, MI), a potent selective reversible PAD4 inhibitor, during the length of the incubation. To elucidate the effect of FAK activation on substrate induced NET formation, neutrophils were treated with or without 100 nM PF-573228 (Cayman Chemical), a potent and selective inhibitor of FAK.

2.4. Protein Analysis

Next, we evaluated neutrophil inflammatory cytokine and chemokine production in response to PDMS substrate stiffness and extracellular matrix protein coatings. Neutrophils were plated at a density of 100,000 cells/cm² (n=6/variable) on PDMS +/− ECM protein coatings. After 4 hours, conditioned media was collected and secretion of IL-1β, IL-6, TNF-α, CCL2, CCL3 (BioLegend), and myeloperoxidase (MPO) (R&D Systems, Minneapolis, MN) were assessed by enzyme-linked immunosorbent assay (ELISA).

2.5. Neutrophil morphology on PDMS substrates

Neutrophils were seeded on PDMS substrates and incubated for 4 hours (n=6/variable). After incubation, neutrophils were washed twice with warm PBS and DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI) (ThermoFisher Scientific). DAPI-stained nucleic acids were visualized using confocal microscopy (Zeiss, LSM 710 Laser Scanning Microscope, Oberkochen, Germany). NET formation was characterized by measuring the area and circularity of stained DNA in at least 70 cells per sample using ImageJ.

2.6. MPO-DNA complexes

As MPO is a prominent constituent of NETs, we sought to quantitatively detect of MPO-DNA complexes using a modified sandwich ELISA as described previously [42]. In brief, anti-MPO monoclonal antibody (R&D Systems, Minneapolis, MN) was used as capture antibody (1 μg/mL) in a 96-well microplate. A peroxidase-labeled anti-DNA monoclonal antibody (capture antibody of commercial Cell Death Detection ELISA kit; Roche, Mannheim, Germany) according to the manufacturer instructions. Absorbance at 405 nm was measured using Synergy HTX Multi-Mode Reader (BioTek, Winooski, VT). A sample size of 6 wells per group or variable was analyzed.

2.7. Statistical analysis

Data are presented as mean ± SD. Statistical analysis was performed using Prism8 (GraphPad Software, San Diego, CA). A one-factor, equal analysis of variance (ANOVA) was used to test the null hypothesis that group means were equal at a significance level of α=0.05, with post-hoc TUKEY-HSD for multiple comparisons. The presented data were obtained from one of two repeated experiments, with both experiments yielding comparable results.

3. Results

3.1. Increased PDMS substrate stiffness enhances neutrophil activation

To determine if substrate stiffness affected formation of NETs, neutrophils were seeded on PDMS surfaces with elastic moduli of 0.2, 2, 8, 16, or 32 kPa (Fig. 1A). The degree of NET formation was characterized by the DNA area and nucleus circularity index of DAPI-stained structures after 4 hours, where large DNA area and low circularity were characteristic of heavy NET formation. NET formation was enhanced in a stiffness-dependent fashion (Fig. 1B). In addition, to increased NET formation, higher secretion of pro-inflammatory cytokines and chemokines (IL-1β, TNF-α, CCL2, CCL3, CCL5, CXCL1) in the conditioned media and of MPO, an enzyme used for the oxidative burst during pro-inflammatory responses, were seen on higher stiffness substrates. Together, these findings reveal that neutrophil activity and NET formation are potentiated in a stiffness-dependent fashion, suggesting that neutrophils integrate mechanical cues into their response to PDMS.

Fig. 1.

Increased substrate stiffness enhances neutrophil activation. Neutrophils were seeded on hydrogels of varying stiffness (E=0.2, 2, 8, 16, 32 kPa) at a density of 10,000/cm² for 4 hours prior to staining with DAPI for confocal microscopy. A) Representative images of stained nuclear material (NETs) on PDMS gels +/− standard deviation. B) Quantification of NET area and circularity by ImageJ software (n=60/variable). p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa.

3.2. Inhibition of PAD4 reduces effect of PDMS substrate stiffness on neutrophil activation

Next, the importance of PAD4 to stiffness dependent NETotic response was examined. Neutrophils seeded on PDMS were treated with GSK484, a PAD4 inhibitor, or were lacking the Padi4 gene. In both cases, NET formation was heavily suppressed in wild type neutrophils treated with the pharmacological inhibitor GSK484 or in neutrophils isolated from Padi4−/− mice on lower stiffness PDMS; interestingly, NETotic activity was enhanced in PAD4-inhibited neutrophils on high stiffness surfaces (Fig. 3A). Similarly increasing resultant MPO-DNA complexes were observed with increasing stiffness, and a comparable reduction in MPO-DNA absorbance from loss of PAD4 activity was observed (Fig. 3B). IL-1β, TNF-α, CCL2, CCL3, CCL5, CXCL1, and MPO levels correlated with the degree of NET formation (Fig. 4). These results demonstrate that NET formation in response to substrate stiffness is abrogated by PAD4 inhibition.

Fig. 3.

Loss of PAD4 activity attenuates stiffness-dependent NET formation. To inhibit PAD4 activity, neutrophils from control mice were pre-incubated with GSK484, or neutrophils were harvested from Padi4^−/− mice. Cells were then seeded on hydrogels of varying stiffness (E=0.2, 2, 8, 16, 32 kPa) at a density of 10,000/cm² for 4 hours prior to staining with DAPI for confocal microscopy +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa from respective Ctrl, GSK484, or Padi4^−/−; p < 0.05: A vs. respective Ctrl, B vs. respective GSK484 from same stiffness.

Fig. 4.

Loss of PAD4 activity decreases stiffness-dependent changes in protein production. To inhibit PAD4 activity, neutrophils from control mice were pre-incubated with GSK484, or neutrophils were harvested from Padi4^−/− mice. Secreted IL-1β, IL-6, TNF-α, CCL2, CCL3, and MPO was measured in the conditioned media +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa from respective Ctrl, GSK484, or Padi4^−/−-; p < 0.05: A vs. respective Ctrl, B vs. respective GSK484 from same stiffness.

3.3. Stiffness-dependent neutrophil activation is integrin signaling-dependent

While NET formation can be induced with different pathogen-associated molecular patterns (PAMPs), cytokines, and calcium signaling activators, we to determine if engagement of specific integrins affected NET formation and if this effect was independent of substrate stiffness. PDMS substrates of varying stiffness were coated with two concentrations of common ECM proteins: type I collagen (α1β1, α2β1 integrins), fibronectin (α5β1, αVβ3 integrins), laminin (α1β1, α2β1, α6β1 integrins), and vitronectin (αVβ5 integrins) [50–54]. Regardless of adsorbed protein, stiffness enhanced NET formation, with higher concentrations further increasing NET formation as measured by DNA area (Fig. 5A) and MPO-DNA complex formation (Fig. 6). Fibronectin, the protein richest in RGD domains, enhanced stiffness dependent NET formation more than the other proteins examined. Again, the degree of NET formation correlated with the level of secreted cytokines (Fig. 7).

Fig. 5.

Stiffness-dependent NET formation is dependent on adsorbed protein composition. Hydrogels were pre-incubated with type I collagen, fibronectin, laminin, or vitronectin at 5 or 10 μg/mL concentrations as indicated for 4 hours prior to staining with DAPI +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa with same protein pre-incubation concentration; p < 0.05: A vs. respective 5 μg/mL, B vs. respective 10 μg/mL from same stiffness.

Fig. 6.

Stiffness-dependent neutrophil cytokine secretion is dependent on adsorbed protein composition. Secreted IL-1β, IL-6, TNF-α, CCL2, CCL3, C-X-C motif ligand 1 (CXCL1), and MPO was measured in the conditioned media.

Fig. 7.

Stiffness-dependent neutrophil activation occurs through integrin activation. Hydrogels of varying stiffness were pre-incubated with either fibronectin or soluble RGD domain at 5 or 10 μg/mL concentrations as indicated for 4 hours prior to A) NET morphological analysis or B) secreted protein quantification +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa with same protein pre-incubation concentration; p < 0.05: A vs. respective 5 μg/mL fibronectin, B vs. respective 10 μg/mL fibronectin, C vs. respective 5 μg/mL RGD, D vs. respective 10 μg/mL RGD from same stiffness.

To further determine whether the NET formation is result of integrin recognition and not the activation of damage-associated molecular patterns (DAMPs), PDMS substrates with varying stiffness were coated with a synthetic RGD synthetic peptide, a motif found in ECM proteins like fibronectin, vitronectin, or collagen that is recognized by several integrins. NET formation and pro-inflammatory cytokines secretion was similar in neutrophils on RGD peptide or fibronectin coated PDMS substrates, as measured by DNA area (Fig. 8A), MPO-DNA complex concentration (Fig. 8B), and cytokine production (Fig. 8C). Together, these findings suggest that integrin signaling might mediate the mechanosensitivity of neutrophils to PDMS hydrogel stiffness.

Fig. 8.

Loss of downstream focal adhesion kinase activity abrogates stiffness-dependent neutrophil activation. Hydrogels of varying stiffness were pre-incubated with either fibronectin or soluble RGD domain at 5 or 10 μg/mL concentrations, ± PF-573228 (FAK inhibitor) as indicated for 4 hours prior to A) NET morphological analysis or B) secreted protein quantification +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa with same protein coating concentration; p < 0.05: A vs. Ctrl, B vs. fibronectin with the same substrate stiffness; p < 0.05: @ vs. PF-573228-untreated with the same substrate stiffness.

3.4. Focal Adhesion Kinase mediates stiffness-dependent neutrophil activation

Finally, to confirm the integrin-dependent nature of neutrophil activation by stiffness, an inhibitor of FAK, a cytoplasmic tyrosine kinase responsible for downstream integrin signaling, was used. Following incubation with the inhibitor, cells were plated on PDMS substrates coated with fibronectin or synthetic RGD peptide. In either case, inhibition of FAK almost entirely ablated NET formation, regardless of substrate stiffness (Fig. 9A and 9B). This effect was paired with a concomitant decrease in inflammatory protein secretion (Fig. 9C). Together, these findings confirm that stiffness-dependent neutrophil activation is integrin/FAK-dependent.

Figure 9.

4. Discussion

Despite being the most abundant leukocyte in humans and predominating in the initial stages of injury, neutrophils are understudied in biomaterials and mechanobiology. Here, we demonstrate that substrate stiffness alone may trigger NET formation through integrin-dependent interactions. We found that stiffness-dependent neutrophils NET formation could be potentiated by increasing the availability of integrin-targeting domains or by inhibiting FAK, which is downstream of integrin signaling. We also showed that fibronectin significantly increases neutrophil NET formation and that this effect is specific to RGD recognition using a synthetic RGD peptide. We also found that NET formation is tightly coupled to cytokine secretion in the stiffness-dependent response, which we previously found in the response of neutrophils to surface topography [31]. Additionally, we validated the use of both MPO-DNA complex measurement and DNA geometrical distribution as useful tools for the measurement of NET formation in two-dimensional mechanobiological model systems.

Cells interact with their environment through biochemical and physical interactions. In the case of neutrophils, biochemical mediators activate neutrophils, which produce an inflammatory response through mechanisms including phagocytic activity, release of reactive oxygen species, antibacterial enzymes, lipid mediators, cytokines, and NETs [3,30,55,56]. During the past decade, multiple studies have shown that physical environment affects cell processes in a multitude of cells [31,33,57,58]. Most of these describe tissue resident cells that are exposed to mechanical stimuli, but a handful have shown that immune cells are also sensitive to changes in substrate stiffness [35,36,59,60]. While it has been reported that changes in substrate stiffness affects neutrophil migration and chemotaxis [61], the effects of substrate stiffness on NET formation have been barely explored. In this study, we have demonstrated that neutrophils are sensitive to substrate stiffness and that pro-inflammatory cytokines, MPO secretion, and NET formation increase in a stiffness-dependent fashion.

In this study, the increase in pro-inflammatory cytokine release and NET formation was observed in the absence of known NET inducers such as LPS or non-physiological stimuli such as PMA. The only report exploring the effects of substrate stiffness on NET formation used polyacrylamide gels of varying stiffness (1-128 kPa) coated with type I collagen or fibrinogen and found that neither substrate stiffness nor protein coating effect neutrophil NET formation in the presence of PMA, while neutrophil NET formation increases in a stiffness-dependent fashion in the presence of LPS [62]. Interestingly, our results clearly showed that both substrate stiffness and protein coatings have a key role in inducing NET formation and release of inflammatory cytokines and enzymes, independent of treatment with traditional molecular NET formation activators. One explanation is the differences observed between the two studies are the substrates used. While both substrates are elastomeric polymers, the PDMS used in this study was functionalized with anhydride functional groups to facilitate interactions with peptide amines and improve the cell-substrate interaction which the hydrophilic polyacrylamide gels lack [63–65]. To provide context, traditional tissue culture polystyrene is treated in a manner to increase surface energy and exhibits and elastic modulus of approximately 10⁷ kPa [66], which is beyond the sensing capabilities of any cell. In this study, we characterized the effects of stiffness using comparable PDMS hydrogels, where elastic modulus was the only appreciable independent variable.

It remains unclear how NET formation is regulated, especially in the absence of microbial stimuli. NET formation has been described as reactive oxygen species dependent, although NADPH-independent NET formation reported [11,67]. NET formation can also be regulated by neutrophil elastase, MPO, and PAD4 [17]. Interestingly, it has been shown that elastase is able to degrade F-actin, translocate to the nucleus, and degrade histones that promote chromatin condensation [68]. Moreover, MPO and elastase synergistically induce chromatin decondensation independent of enzymatic activity [17]. However, we have previously demonstrated that neutrophils isolated from neutrophil elastase-KO and MPO-KO are able to produce NETs in response to titanium surface properties [31]. Additionally, it has been shown that neutrophils in the elastase-KO model are competent to undergo NET formation [69]. Several reports have shown that PAD4 is essential for NET formation by chromatin decondensation by histone hypercitrullination, and that that pharmacological inhibition of PAD4 abolishes neutrophil NET formation in different contexts [27,70–72]. However, other reports have shown that PAD4 inhibition did not affect NET formation when induced through different stimulants derived mainly from microorganisms [73]. We have shown that pharmacological inhibition of PAD4 reduced NET formation in response to titanium surface properties [31]. In this study, pharmacological inhibition or genetic knockout of PAD4 abolishes NET formation except on stiffer substrates. Pharmacological inhibition of PAD4 is effective in abolishing NET formation on the two softest substrates, but its effectiveness decreases in a stiffness dependent fashion. Genetic KO of Padi4 also abolishes NET formation on softer substrates but NET formation is still observed on the stiffest substrate. These results suggest that PAD4 activity increases in a stiffness dependent fashion and may be activated by other signaling mechanisms responding to mechanical stimuli. It is also possible that increased traction forces from high substrate stiffness are conferred to the nucleus through cytoskeletal proteins, which dysregulates the permeability of the nuclear envelope to cytoplasmic enzymes, allowing other PAD isoforms, MPO, or neutrophil elastase into the nucleus to drive PAD4-independent NET formation. Several studies have shown that external mechanical forces can alter nuclear envelope porosity by modulating the basket conformations of pore proteins [74–76]. However, pharmacological inhibition of PAD4 in a human serum transfer model of systemic lupus erythematous, a disease believed to be triggered by NET formation, failed to reduce organ damage [77].

Since we observed that NET formation was dependent on substrate stiffness, we then asked if substrate stiffness was the only factor or if differences in the composition of proteins adhered to the substrate also affected NET formation. For this, we chose several proteins commonly present in the extracellular matrix. Our results show that protein composition on the substrate also plays a role in NET formation. In general, all proteins increased NET formation in concentration-dependent fashion, and substrate stiffness and protein coating had an additive effect on NET formation. Interestingly, we found that fibronectin coatings dramatically increased NET formation, an effect further enhanced on the stiffest substrate. While these results indicate that ECM proteins can induce NET formation, we did not evaluate whether protein adsorption onto the substrates could lead to protein misfolding or denaturing that could act as a DAMP signal. Since fibronectin is rich in RGD motifs, we then investigated if the excessive NET formation observed was result of integrin engagement through RGD domains using a synthetic RGD peptide. Both fibronectin and RGD dramatically increased NET formation in a concentration- and stiffness-dependent fashion, indicating that NET formation is also activated through integrin signaling. It has been previously reported that fibronectin is required for recognition of purified fungal PAMP β-glucan from C. albicans, where in the absence of fibronectin, neutrophils produce ROS but failed to release NETs [78]. Later, the same group showed that integrin crosstalk regulates NET formation in response to fungal β-glucan [79].

To confirm that NET formation in response to substrate stiffness and ECM protein coating is mediated through integrins, we used pharmacological inhibition of FAK. Inhibition of FAK in neutrophils on fibronectin or RGD coated PDMS substrates ameliorated NET formation and cytokine release. While the role of FAK in NET formation has not been explored directly, it has been shown that inhibition of PI3K, a target for phosphorylation by FAK [80], reduced neutrophil cell area on polyacrylamide substrates and prevented NET formation [62]. However, PI3K is also involved in autophagy and is also activated by PAMPs through TLR2/4-Rac1, cytokine receptors through JAK signaling, and chemokine signaling through chemokine receptors [81,82], making this mechanism less likely to be only the effect of integrin signaling.

A few limitations exist in this study. Unfortunately, only the effects of stiffness in two dimensions were explored. It is possible that forces in three dimensions might by their nature restrict NET formation and spreading if the material is stiff enough, which paradoxically would pair lower NET formation with stiffer tissues. Future studies will explore NET formation in three dimensions. Furthermore, the findings here do not explore the implications in vivo. It is already known that more stiff materials are associated with long-term inflammatory cell responses both in vitro and in vivo [83,84], but the involvement of NETs in this response has not been yet explored.

5. Conclusions

In this study, we demonstrated that NET formation and inflammatory activation on PDMS surfaces are dependent on substrate stiffness, even in the absence of traditional activators like bacterial LPS or PMA. We also show that stiffness dependent NET formation is potentiated by the presence of RGD binding sites. Finally, we show that stiffness dependent NET formation occurs through the integrin/FAK signaling pathway, and that inhibiting FAK decreases inflammatory activation of neutrophils. These findings have important implications in our understanding of sterile inflammation and suggest that modulating integrin signaling may be a therapeutic avenue for pathologic NET formation.

Fig. 2.

Neutrophil inflammatory protein secretion increases in response to substrate stiffness. Neutrophils were cultured for 4 hours on PDMS hydrogels (n=6/group). Secreted interleukin-1β (IL-1β), IL-6, tumor necrosis factor-α (TNF-α), C-C motif ligand 2 (CCL2), CCL3, and myeloperoxidase (MPO) was measured in the conditioned media +/− standard deviation. p < 0.05: # vs. 0.2 kPa, $ vs. 2 kPa, % vs. 8 kPa, & vs. 16 kPa.