Neutrophils Defensively Degrade Graphene Oxide in a Lateral Dimension Dependent Manner through Two Distinct Myeloperoxidase Mediated Mechanisms

The boosting exploitation of graphene oxide (GO) increases exposure risk to human beings. However, as the first line defender, neutrophils’ mechanism of defensive behavior towards GO invasion remains unclear. Herein, we discover that neutrophils defensively degrade GO in a lateral dimension dependent manner. The micrometer-sized GO (mGO) induces NETosis by releasing neutrophil extracellular traps (NETs), while nanometer-sized GO (nGO) elicits neutrophil degranulation. The neutrophils’ defensive behavior is accompanied with generation of reactive oxygen species and activation of p-ERK and p-Akt kinases. We unveil that mGO-induced NETosis is NADPH oxidase (NOX)-independent, while nGO-triggered degranulation is NOX-dependent. Furthermore, myeloperoxidase (MPO) is identified to be a determinant mediator despite distinct neutrophil phenotypes in the biodegradation. Neutrophils release NETs comprising of MPO upon activated with mGO, while MPO is secreted via nGO induced-degranulation. Moreover, the binding energy between MPO and GO is calculated to be which indicates that electrostatic interactions mainly cause the spontaneous binding process in a spatial distance of 9.2 Å. Meanwhile, the central enzymatic biodegradation is found to occur at oxygenic active sites and defects on GO. Mass spectrometry analysis deciphers the degradation products are biocompatible molecules like flavonoids and polyphenols. Our study provides fundamental evidence and practical guidance for functional biomaterial development in sustainable nanotechnology, including but not limited to vaccine adjuvant and drug carrier. a side-by-side comparison of mGO and nGO-induced pathways. showed both mGO and elicited ROS burst within 30 minutes. ROS are essential upstream effectors ROS, whereas lower ROS group.


Introduction
Graphene oxide (GO), a two-dimensional graphene derivative, possesses hydrophobic sp 2and sp 3-hybridized carbon skeleton and abundant hydrophilic groups (hydroxyl, epoxide, and carboxyl) forming defect sites at its edge and plane. 1,2 Such unique structure imparts many fascinating physiochemical properties like large surface area, strong adsorbability, and versatile functionality. 3 Hence, GO-based potential applications such as drug delivery, 4 implanted biodevices, 5 vaccine adjuvant, 6 energy storage, 7 and sea water desalinization 8 have aroused. Therefore, risk evaluation of GO-based technologies and products requires rigorous charaterization. 9 Immune system provides a universal and immediate defense against foreign invaders. 10 Once GO enters body, it will encounter neutrophils, macrophages and dendritic cells (DCs), which represent the first line of immune system. Several in vivo studies indicated that GO elicited cell membrane lipid change of neutrophils, 11 promoted size-dependent M1 induction of macrophages, 12 and suppressed antigen presentation by DCs to T cells. 13 However, the immune effect and cellular mechanism of defensive behavior towards GO invasion are still unraveled.
Neutrophils, as dominant immune cells in quantity, play important roles in immune defense by three major strategies, namely phagocytosis, degranulation, and neutrophils extracellular traps (NETs) via 'NETosis' (a specific neutrophil death program). 14 Degranulation of antimicrobial factors involving myeloperoxidase (MPO) leads to destruction of invaders. 15 Besides, neutrophils could directly contact with invaders by forming unique extracellular web-like NETs that compose of decondensed chromatin, and antimicrobials including MPO and neutrophil elastase (NE). 16 MPO is the most abundant hemeprotein in neutrophils, which catalyzes hydrogen peroxide (H2O2) to form hypochlorous acid (HClO) and other oxidants. 17,18 NE is a neutrophil-specific serine protease that kills foreign invaders like viruses and bacteria. 19 MPO and NE are stored in azurophilic granules in resting neutrophils, however, they will be released outside the neutrophils cooperatively in response to invaders. 19 In recent years, MPO has been reported in degrading carbon materials like single-walled carbon nanotubes (SWCNTs) 20 and GO sheets 21,22 . However, the mechanism of GO biodegradation including chemical reaction details, related signaling pathways and fate of neutrophils is not clear.
Herein, we carefully explore the bi-direction interactions between neutrophils and GO, including neutrophils biodegrading GO and GO switching neutrophils' fate (Scheme 1). We unveil that neutrophils autonomously degrade GO in a lateral dimension dependent manner. Micrometer-sized GO (mGO) and nanometer-sized GO (nGO) selectively direct neutrophils to two distinct fates: NETosis or degranulation.
Furthermore, the underlying signaling pathways are carefully investigated. Thermodynamics analysis and active sites of GO biodegradation are also elucidated. Ultimately, the biodegradation products of GO are identified and confirmed for their good biocompatibility with neutrophils. Collectively, the complicated interactions between neutrophils and GO are clearly depicted.

Scheme 1. Neutrophils degrade GO in an autonomous manner and lateral dimension dependent manner.
Neutrophils release NETs in response to mGO, while induce degranulation in response to nGO. GO sheets are eventually biodegraded into non-cytotoxic small molecules by neutrophils.

GO is defensively degraded by neutrophils
To address the interactions between neutrophils and GO, GO sheets of different lateral dimension were firstly synthesized by modified Hummers' method as previously described. 6 The lateral dimension and thickness of mGO and nGO sheets were analyzed by atomic force microscopy (AFM) and transmission electron microscopy (TEM), and recorded as 4.299 ± 0.824 μm and 289.9 ± 111.7 nm, respectively (Supplementary Fig. 1). Other physicochemical characterizations including surface charge, degree of defects and chemical composition were summarized in Table S1. Then, neutrophils were isolated from mouse bone marrow and incubated with GO at 37 °C. The neutrophils were refreshed every 12 h due to short-life. 23 To facilitate investigating GO and neutrophils respectively, the experiments were performed in Transwell ® chambers ( Supplementary Fig. 2). The permeable membrane pore was small enough to prevent neutrophils phagocytosis and transmigration, but large enough to allow neutrophils to extend filopodia and contact GO on the other side. We removed the Transwell ® insert after incubation for collecting GO suspension and cells for further analysis. Initially, we observed that both mGO and nGO were gradually degraded by neutrophils. TEM images showed that the characteristic sheet shape disappeared but visible "holes" formed on mGO sheets after 1 day (Fig. 1a). Besides, some proteins of neutrophils were adsorbed on GO sheets. When incubation extending to 2 days, considerable damaged parts on the basal plane of GO were visible. Eventually, residual mGO extensively degraded into nanometer-sized fragments after 3 days. The patterns of GO on 0 d (Fig. 1b) and 3 d (Fig. 1c) also indicated that GO were degraded from micrometer-size to nanometersize. The morphology changes were also observed on nGO ( Supplementary Fig. 3A). Moreover, crystalline phase changes of GO were also characterized by Raman spectroscopy (Fig. 1d; Supplementary   Fig. 3B). Significant loss of disorder-induced D band (1380 cm -1 ) and crystalline G band (1620 cm -1 ) was observed, indicating neutrophils destroyed the order degree of GO. We also measured the diameter of holes on mGO sheets after 3 days (Fig. 1e), and the average diameter of holes was 18.4 nm.

Distinct defensive behaviors of neutrophils depend on GO lateral dimension
Subsequently, we carefully explored the detailed biological changes of neutrophils during biodegradation by fluorescence imaging, scanning electron microscope (SEM), reactive oxygen species (ROS) detection and immunoblotting. As shown in Fig. 2a  Simultaneously, the nucleus disassembled and the web-like NETs were released as cell membrane broke.
Moreover, neutrophils became flat and adhered to the substratum ( Supplementary Fig. 4), suggesting that neutrophils underwent NETosis in line with previous study. 16,19 Therefore, mGO not only triggered the NETs formation but also appeared to be encased in web-like structures. By contrast, neutrophils secreted MPO after stimulating by nGO. NETs formation was not observed by immunofluorescence or SEM images. This indicated that neutrophils mainly secreted MPO to resist exogenous nGO by degranulation. 24 Moreover, the most distinct morphological changes of neutrophils were perimeter and shape in the nuclear area. 25 Thus, we defined four morphologies of nuclear as shown in Figure 2c: i) lobulated, ii) delobulated, iii) diffused, and iv) extended nuclei. Most neutrophils lost the typical lobulated nuclear due to undergoing NETosis or degranulation. Besides, 69.8% and 83.6% of neutrophils keep cell viability after treating with mGO or nGO (Fig. 2c), demonstrating that GO triggered a size-dependent loss of neutrophils viability. Taken together, the mechanism of neutrophils defense was lateral dimension dependent, namely, neutrophils released NETs upon stimulated with mGO, while neutrophils induced degranulation in response to nGO.
Mitochondrial ROS and the MPO-NE pathway are parallel key signals in degranulation and NETosis. 26 We evaluated mitochondrial ROS production elicited by mGO and nGO since NETosis requires ROS (Fig.   2d). 18,26,27 We used phorbol 12-myristate 13-acetate (PMA), a pharmacological agonist to induce ROS as a side-by-side comparison of mGO and nGO-induced pathways. 28 Our results showed both mGO and nGO elicited ROS burst within 30 minutes. Therefore, ROS are essential upstream effectors in biodegradation. However, the activation of neutrophils with mGO led to an abundant production of ROS, whereas nGO-treated neutrophils produce lower ROS than PMA group. When neutrophils were pretreated with the NADPH oxidase (NOX) inhibitor diphenyleneiodonium (DPI), 29 ROS production in PMA-stimulated cells was significantly suppressed (Fig. 2d), resulted in inhibiting NETosis ( Supplementary Fig. 5A). Hence, PMA-induced NETosis is NOX-dependent.
Nevertheless, DPI didn't block NETs release in mGO groups, which indicated that mGO engaged NETs through a NOX-independent way ( Supplementary Fig. 5B). Accordingly, mGO-induced NETosis is distinct from PMA-mediated NETosis. However, NE activity was blocked by DPI in nGO group, suggesting that neutrophils degranulation triggered by nGO was dependent on NOX ( Supplementary Fig.   6).
We further explored the signaling pathway in the GO biodegradation.
Since key kinases such as p-ERK 25 and p-Akt 30 have been shown involved in PMA-induced NETosis, we therefore assessed their differential activation by mGO or nGO. Our results showed that p-ERK and p-Akt were robustly activated by mGO. In contrast, nGO moderately activated p-ERK and p-Akt (Fig. 2e). The activation of p-ERK was 61.4% reduced in DPI-nGO-treated neutrophils ( Supplementary Fig. 7). Although similar kinases were activated in mGO-and nGO-induced biodegradation, neutrophils switched to NETosis or degranulation resulting in different functional attributes to degrade GO in a lateral dimension dependent manner.

MPO is determinant mediator of GO biodegradation
Hydroxyl radical (·OH) plays an important role in degrading carbon nanomaterials through reacting with double bonds (C=C). 20,31 Besides, it also oxidizes carbon nanomaterials to generate carboxyl and hydroxyl groups on surface. 21 To confirm the high redox ability of hydroxyl radical, we firstly investigated MPO/H2O2/Clsystem. Samples of GO/H2O2/Cland GO/MPO/Clwere characterized by Raman spectra ( Supplementary Fig. 8A, B). No biodegradation was observed in these samples, indicating that hypochlorous acid was a major oxidant in the reaction. Additionally, the biodegradation of GO was interrupted by MPO inhibitor, demonstrating such biodegradation was a MPO-mediated manner ( Supplementary Fig. 8C).
Next, we focused on verifying MPO-mediated biodegradation of GO since the defensive effect of NETs is dependent on its major component of MPO with high redox ability. 32 GO sheets were incubated with MPO at 37 °C in sterilized water containing sodium chloride (NaCl) with or without H2O2. The suspension was collected for subsequent analyses at 0, 12, and 24 h. Biodegradation was monitored by observing color change of reaction solution. After 24 h, the suspension became translucent ( Supplementary Fig. 9). To approve the MPO-mediated biodegradation of GO, we employed AFM and TEM to acquire the topography and thickness of GO sheets. As MPO is a highly cationic protein which tends to be attached on negatively charged surfaces like GO sheets, MPO was adsorbed on GO sheets during incubation.
Initially, the monolayer GO sheets showed characteristic thickness of 1.06 nm (Fig. 3a). As adsorption of MPO, the thickness of GO increased to 5.85 nm (12 h) and 1.86 nm (24 h), respectively (Fig. 3b, c). With expansion of hole areas in biodegradation, the sharp edge of GO became irregular and a large number of GO fragments were observed (Fig. 3d, e). After 24 h, the GO sheets were degraded into debris with lateral dimensions ranging from 20 to 100 nm (Fig. 3f).
Furthermore, Raman spectra characterized the crystalline phase changes of GO sheets (Fig. 3g). After 12 h, both D and G bands observably reduced with ID/IG decreasing from 1.03 to 1.01, indicating the GO sheets were partially degraded with little increase of defect density. After 24 h, both bands notably diminished, suggesting the GO sheets were degraded completely.
To examine the chemical composition variation of GO during the biodegradation, XPS spectra were acquired ( Supplementary Fig. 10). Peaks at 284.8, 286.9, 287.9, and 288.8 eV corresponded to C-C in aromatic rings, C-O (epoxy and alkoxy), C=O (carbonyl), and C=O(OH) (carboxylic) groups individually. 31 With the reaction proceeding, the intensity of C-O groups decreased dramatically, while the intensity of C=O and C=O(OH) groups increased gradually. As a result, the C/O atomic ratios were 2.66, 2.43 and 1.58 at 0, 12 and 24 h, respectively (Fig. 3h). Our data showed that the C-C and C-O bonds of GO in degradation were broken. Meanwhile, more periphery carbons were oxidized into the C=O or C=O(OH) groups.

Thermodynamics in MPO-mediated biodegradation
In this section, we investigated the thermodynamics of MPO-mediated biodegradation by intrinsic fluorescence measurement and molecular docking simulation. The thermodynamic parameters enthalpy change (∆ 0 ) and entropy change (∆ 0 ) were calculated according to the following van't Hoff equation 33 where refers to the binding constant, ∆ 0 is the free energy change, R is the gas constant, and T is absolute temperature. Since GO is a sensitive fluorescence quencher and is capable to quench spontaneous fluorescence of proteins molecules, 34 spectral measurements were used to obtain .
Next, we calculated the fluorescence quenching parameters according to the modified Stern-Volmer where 0 and are the fluorescence intensities in the absence and presence of GO respectively, [Q] is concentration of GO, is Stern-Volmer quenching constant of MPO by GO, and is the fraction of accessible fluorescence. Our results showed that the intrinsic fluorescence intensity of MPO had a negative relationship with the concentration of GO at 298.15 K (T1) and 310.15 K (T2) (Fig. 4a; Supplementary Fig. 11). And there was a positive liner correlation between 0 /∆ ratio and   Fig. 4b). In addition, the values of decreased with temperature increasing, suggesting that the interaction of GO and MPO followed static quenching model. 36 To obtain the binding constant of , we used equation (2) to where n is the binding number. Accordingly, the values of n and were acquired from the slope and intercept of fitting curves, respectively ( Fig. 4c; Table 1). The thermodynamic parameters of ∆ 0 , ∆ 0 and ∆ 0 are the main evidence for interaction forces, which could be estimated from the Gibbs The calculated values of ∆ 0 , ∆ 0 and ∆ 0 at T1 and T2 were summarized in Table 1. The binding process of MPO and GO was spontaneous because of negative ∆ 0 . Furthermore, the negative ∆ 0 and positive ∆ 0 values indicated that electrostatic interactions were major causes during the binding process.
Subsequently, we calculated the distance between MPO and GO based on the above experimental FRET data. A transfer of energy could take place through direct electrodynamic interaction, which will happen under the condition that fluorescence emission spectrum of the donor (MPO) and UV-vis absorbance spectrum of the acceptor (GO) have overlap. 36,37 Figure 4d showed that the overlapping of emission spectrum of MPO and the absorption spectrum of GO, indicating the FRET occurred. According to Föster theory of non-radioactive energy transfer, 38,39 we calculated that the Förster distance (R0) was 2.4 nm and the spatial distance (r) between MPO and GO was 2.8 nm, which indicated that energy transfer from MPO to GO occurred with high probability.
To identify the binding manner, we performed molecular docking simulation using AutoDock Vina, a software to predict bound conformations and free energies of binding. 40 Our results showed that MPO had a single binding site with GO, consisting of Asp321, Arg323, Ser19, Arg31, Ile160 and Pro34 (Fig. 4e, f).
Specifically, Asp321, Arg323, Ser19, and Arg31 bound to GO through electrostatic interaction, while Ile160 and Pro34 relied on hydrophobic interaction. Comparatively, the binding manner between MPO and GO was inclined to electrostatic interaction. The calculated closet distance from GO to the binding site of MPO was 9.2 Å, and the binding energy was -69.8728 kJ·mol -1 .

Identify active sites on GO
We next investigated the active sites on basal plane or edge of GO. The heterogeneous electron-transfer (HET) rate was measured by two redox probes, namely potassium ferricyanide (K3Fe(CN)6) and hexaammineruthenium chloride (Ru(NH3)6Cl3) as inner-and outer-sphere probes, respectively. 41 The electrochemical signal of inner-sphere probe reflects the amount of electrochemically active sites and defects on the edges and basal plane of GO, which is also sensitive to oxygen functionalities. 42 For outersphere probe, its electrochemical signal is influenced only by the amount of electrochemically active sites. 43 The cyclic voltammograms (CVs) were firstly performed in basic reaction medium (phosphate buffer saline, PBS) to verify no presence of unrequired side reactions ( Supplementary Fig. 12). According to the CVs shown in Fig. 5a and 5b, the peak-to-peak separation values (∆ ) between control (0 h) and degraded samples were calculated respectively (Fig. 5c, d). For inner-sphere probe, the decreased ∆ values suggested more oxygen functionalities were introduced into GO with prolonged reaction time. The gap of ∆ values between 0 h and 6 h attributed to increase holes on the basal plane of GO and newly generated GO fragments. For outer-sphere probe, the ∆ values gradually rose up, which indicated more electrochemically active sites emerged on the edges of GO. nucleus with a vic-trihydroxyphenyl moiety or an ortho-dihydroxyphenyl (Fig. 6c). These potential molecular structures were very similar to catechin, epigallocatechin, epicatechin gallate and eucalyptin. This is consistent with our previous study that the debris of GO probably contained a lot of similar small molecules like flavonoids and polyphenols. 6 Furthermore, we incubated freshly isolated neutrophils with biodegraded suspension of GO to evaluate the effects of such smaller molecules on neutrophils viability. In comparison with normal control (RPMI 1640 culture media) and positive control (PMA, a NETosis agonist), the biodegradation products containing smaller flavonoids-like molecules did not display any significant side-effect (Fig. 6d). Based on the aforementioned thermodynamic analysis, electrochemical and mass spectrometry results, we speculated that biodegradation initiated at the carbon atoms connected with hydroxyl and epoxide groups (Fig. 7). A large amount of hydroxyl radical and peroxide radical was generated and attacked the carbon atoms in the presence of MPO/H2O2. Moreover, hydroxyl radicals would further oxide GO through not only the conversion of oxygen moieties to higher oxidation states but also electrophilic addition to unsaturated bonds. 44 Interestingly, the newly formed oxygen-containing groups including the quinone group or radicals may serve as new oxidation reaction sites. As a result, the GO sheets were "cracked" into fragments, GO quantum dots, or even small molecules.

Conclusion
Understanding the interaction between GO and immune system is the key to implement GO from bench to bedside. In this work, we firstly discover that neutrophils spontaneously sense lateral dimension of GO, then degrade GO via NETs formation or degranulation. The lateral dimension of GO is a critical factor that regulates the "decision" of neutrophils. Once GO invades the organism, the neutrophils are recruited and become activated. Neutrophils release NETs in response to mGO through NOX-dependent pathway, or induce degranulation in response to nGO. The biodegradation is substantially MPO-catalyzed oxidation reaction. MPO consumes H2O2 to generate HOCl at the active sites where abundant oxygen functional groups and defects present. Furthermore, our data have demonstrated the biodegradation products of GO are flavonoids and polyphenols such as catechin, epigallocatechin and eucalyptin, which usually are abundant in plants and nutritional supplements without potential risk to immune system. 45,46,47 Intriguingly, larger-sized invaders are more effective at inducing NETosis, suggesting that NETs enhance effective immune defense and promote synergy between MPO and other various antimicrobials to minimize spread of invaders. 32 Since NETs release the same antimicrobials including MPO as degranulation, we are curious about why neutrophils undergo NET formation at the cost of "suicide" other than the release of histones. Due to immune-modulatory properties of neutrophils, advantages of NETs are rational. First, NETs minimize destruction to surrounding tissues and coordinate the inflammatory response. Secondly, neutrophils only pay the cost of shedding NETs like a scaffold, instead of high cellular energy for chemotaxis and uptake of invaders. 32 Thirdly, the average life span of NETs is more than 24 h, 48 which persists much longer than phagocytosis or degranulation. Finally, given the web-like structure of NETs, they serve as physical carriers which prevent invaders from escaping. Collectively, neutrophils selectively switch defense strategies to achieve efficient goals (GO biodegradation) with minimum cost.
The decision is simple but smart as the host defense of innate immune system.
Sustainable nanotechnology requires a comprehensive and scientific risk evaluation including exposure pathways and fate of nanomaterials. 49,50 Previous work have reported the in vivo toxicity of GO such as acute injures to lung and liver. 51,52 Moreover, carbon nanotubes became the first nanomaterial to be added to the SIN ('Substitute It Now') List by the Swedish non-profit organization in recent years, which aroused the attention on the future of sustainable nanotechnology. 53 Our study of the interaction between immune system and nanomaterials elucidated the immune effect on human health and environment. Our research provides fundamental guidance and experimental evidences for practical application of GO in sustainable nanotechnology, including but not limited to vaccine adjuvant development and drug carrier development.

Synthesis, Modification and Characterization of GO:
Both mGO and nGO were synthesized using the modified Hummers' method as our previous study described 6    which were recorded in 300 to 500 nm.
Molecular Docking: To indicate GO interaction mode and binding parameters theoretically with MPO, the molecular docking calculations were performed using AutoDock Vina software. 54 GO molecular structures were modeled using the Hyperchem 8.0.6 program and VMD package. 55 Then, the geometry of them using the theoretical level of B3LYP with a 6-31G basis set that implemented in the Gaussian 98 program was optimized to minimize energy. MPO crystal structure (PDB ID: 1DNU) was used for the molecular docking calculations and obtained from the RCSB Protein Data Bank (http://www.rcsb.org).
According to instruction, the PDBQT files of MPO and GO were prepared and then analyzed using the AutoDock Tools 1.5.4 package. 56 Electrochemical Detection: The electrochemical characterization by means of cyclic voltammetry was performed with a CHI 660E electrochemical workstation (CH Instruments, China) with gold microelectrodes fabricated as previously described. 57 The microelectrodes were placed into an Nitrogen laser was used for ionization at the negative ion mode for GO samples.
Statistical Analysis: All data are presented as mean ± SD. Statistical analysis was performed using GraphPad Prism statistical analysis software (Version 8.0). Comparisons between different groups were performed using student's t test. A level of P < 0.05 was regarded as a significant.