PEGylated graphene oxide elicits strong immunological responses despite surface passivation

Engineered nanomaterials promise to transform medicine at the bio–nano interface. However, it is important to elucidate how synthetic nanomaterials interact with critical biological systems before such products can be safely utilized in humans. Past evidence suggests that polyethylene glycol-functionalized (PEGylated) nanomaterials are largely biocompatible and elicit less dramatic immune responses than their pristine counterparts. We here report results that contradict these findings. We find that PEGylated graphene oxide nanosheets (nGO-PEGs) stimulate potent cytokine responses in peritoneal macrophages, despite not being internalized. Atomistic molecular dynamics simulations support a mechanism by which nGO-PEGs preferentially adsorb onto and/or partially insert into cell membranes, thereby amplifying interactions with stimulatory surface receptors. Further experiments demonstrate that nGO-PEG indeed provokes cytokine secretion by enhancing integrin β8-related signalling pathways. The present results inform that surface passivation does not always prevent immunological reactions to 2D nanomaterials but also suggest applications for PEGylated nanomaterials wherein immune stimulation is desired.

chloride ions, yielding a system containing nearly 350,000 atoms. To both simplify the simulation setup and emphasize the planar properties of nGO, stringent harmonic restraints were placed on both the graphene nanosheet and terminal carbon atoms attached to oxygen-containing functional groups and PEG chains; forces among restrained atoms were not computed. More specifically, a small scaled force constant of 10 kcal mol -1 Å -2 was applied to all graphene sheet carbon positions (including those of oxidation sites and those of carbons directly bound to PEG chains), yielding a total of 1,008 restrained atoms. Molecular dynamics calculations were completed using the NAMD 6 simulation package, invoking a Langevin integrator held at constant temperature and pressure (310 K; 1 atm). Standard CHARMM27 force field parameters were employed for all interactions not discussed above; PME electrostatics were used in concert with dispersion interactions subjected to a mutual 1.2 nm real space cutoff. Normal SETTLE constraints were applied to enable the use of a 2 fs time step. Simulation trajectories were extended until the PEG adsorption process was deemed to be complete (after approximately 50 ns); an equilibrated snapshot of the nGO-PEG complex was extracted for use in subsequent simulations and featured in Fig. S10.
Membrane configuration setup. Independently, an 8 nm × 8 nm segment of a pre-equilibrated POPC lipid bilayer was generated using the Membrane Builder plugin in VMD 1 . Additional equilibration simulations were conducted for thoroughness. After solvation in TIP3P and the deletion of water molecules in the transmembrane region, lipid tails were melted (with head groups restrained) for 25 ns.
The entire system was then equilibrated without restraint for an additional 25 ns.
Force field and simulation parameters were identical to those used above, except that pressure control was only implemented in the direction normal to the membrane surface (as recommended by the CHARMM developers). Periodic boundaries were also defined such that lipids spanned the transverse boundaries in a smooth fashion.
An equilibrated membrane configuration was extracted for use in production simulations.
Production simulations. To generate the configurations used in production simulations, the equilibrated nGO-PEG complex (or, simply the initial nGO sheet) was placed in either an edge-on or face-on configuration 1 nm from the membrane surface. After solvation with TIP3P, deletion of water molecules in the transmembrane region, and ionization, short equilibration runs were conducted with the positions of PEG and membrane atoms fixed for 500 ps and free for another 500 ps. Production runs were then conducted under the same restraints used in free nGO-PEG simulations, employing the same force field and simulation parameters described previously. Once more, pressure coupling was only applied along a membrane-normal coordinate. All systems contained approximately 100,000 atoms, and were simulated for several hundreds of nanoseconds (up to a microsecond) until satisfactory convergence was evident.
Simulation analysis. Center of mass calculations between atom groups of interest were carried out using custom Tcl scripts. Specific interaction energies were computed using the NAMD Energy plugin in VMD.
First, EDC (20 mM) was added into the pristine nGO suspension (~500 μg mL -1 ) and sonicated for 15 min. Immediately, mPEG-NH 2 (10 mg mL -1 ) was added and allowed to react overnight. The final product (nGO-PEG) was harvested by centrifugation at 70,000 g after repeated washing by DI water.
Supplementary Note 12: Carbon Spheres. Referring to our previous report 10 , 1.08 g glucose was dissolved in 12 mL water to form a clear solution after ultrasonication. The solution was then transferred into a 15 mL high temperature reactor and the brownish black solution was observed when hydrothermally treated at The approximate surface areas of CNTs, CSs and nGOs were calculated using the following equations: S CNT =π×d×l (d=20 nm, l=4 μm), S CS =4π×r 2 (r=100 nm), S nGO =2×d 2 (d=200 nm, assumed to be a square). The resultant ratios of surface areas were thus S CNT : S CS : S nGO =3.14: 1.57: 1; we used these ratios to normalize nanomaterial doses among the different carbon materials. Interaction study between nGO complexes and cells. pMØs were seeded (1 × 10 5 mL -1 ) in a Petri dish and incubated with nGO complexes at 10 and 40 μg mL -1 for 24 h. nGO complex imaging was performed under a flow cytometry PE-Cy7 channel using graphene's intrinsic photoluminescence. The cytoskeleton and nuclei were subsequently separately stained with rhodamine-phalloidin (green pseudocolor in images) and Hoechst, respectively, for 20 min at room temperature. Images were captured with an Ultraview VoX cell imaging system (PerkinElmer) and the corresponding nuclear parameters were analyzed in a Columbus analysis system (PerkinElmer). Second stimulation assay. pMØ cells were stimulated with 10 μg mL -1 nGO-PEG solutions at the setting time (12 h, 24 h, 48 h). After 12 h, the supernatant was collected (set as ○ 1 ) and the culture medium was replaced by the fresh water. This nGO-PEG free culture medium was collected again at 24 h (set as ○ 2 ) and 48 h (set as ○ 3 ). To investigate the second stimulus level of cells, fresh culture medium was removed again by presenting nGO-PEG contained medium at 24 h, and then the result was analyzed at 48 h (set as ○ 4 ).

Detection of plasma membrane integrity (LDH Assay)
. 20 μL LDH release solution was added to the positive well one hour before 24 h of coincubation. Cell culture medium was collected and centrifuged at 3000 rpm for 5 min. An aliquot (120 μL) of supernatant and 60 μL LDH detection reagent were added to a new 96-well plate to quantify the LDH level. After coincubating for 30 min at room temperature without light, absorbances were measured in an Infinite M200 microplate spectrophotometer (Tecan) at 490 nm.
Fluorescence recovery after photobleaching (FRAP). pMØ cells were seeded on a petri dish overnight to encourage adhesion, and then 2 μL of membrane fluorescent probe DiO (initial concentration 1,000 μM) was added after being dissolved in DMSO.
Cells were incubated at 37 o C and 5% CO 2 for 20 min, and then transferred from the petri dish to the incubator on an Ultraview (Perkin Elmer, America) to monitor the cells under an 100× oil objectiv. Regions of interest in the stained membranes were noted and photobleached using a 488 nm laser operating at 13% power. Single images were then collected at maximum speed. All FRAP data were analyzed with prepackaged analysis software, which can fit the model to experimental data reasonably through a nonlinear curve fit to . Sample preparation for genechip analysis. pMØ cells were seeded in a 35 mm petri dish at a density of ~ 5000,000 and cultured for 6 h to achieve adhesion. Cells were then induced with 10 μg mL -1 nGO-PEG for 12, 24, and 48 h respectively, with each group having three parallel specimens. RNA was extracted from each sample using 0.6 mL trizol reagent per dish, which was sent to Shanghai Gene Corporation for further analysis. Pathway analysis was performed with the help of the KEGG database.
Finally, the supernatant was removed, and cytokine secretion was measured as described above. Cytokine secretion increases gradually with increasing PEG densities.  Considered alongside the near-neutral zeta potential in nGO-PEG, these results indicate that PEG chains were successfully conjugated to the GO surface. Figure 24. Thermal gravimetry (TG) and differential thermal gravimetry (DTG) curves for nGO-PEG. The red DTG curve indicates ranges for the decomposition temperatures of nGO and PEG. Combining these DTG results with information from the TG curve, one sees that nGO undergoes an obvious weight loss (87.9% to 49.7%) above 100°C; the thermal decomposition of PEG occurs in a second step beyond ~300°C (49.7% to 36.6% weight change). Through the above-calculated proportions of nGO (38.2%) and PEG (13.1%), the mass ratio of nGO to PEG was determined to be about 3:1.  Table 3. Zeta potential of carbon materials before and after

Supplementary
PEGylation. PEG reduced each potential to near neutrality. Data are means ± SD, with n = 3.