Activation of Human Natural Killer Cells by Graphene Oxide-Templated Antibody Nanoclusters

An emerging new paradigm is that immune cell activation is controlled by transient interactions between supramolecular assemblies of receptors and ligands. Current immunotherapy biologic pharmaceuticals that activate or desensitize NK cells are, however, individual molecules that do not replicate this nanoscale organization of proteins. Here, we use nanoscale graphene oxide (NGO) as a template to generate soluble nanoscale clusters of Natural Killer cell-activating antibodies. We control nanocluster size and molecular number to mimic reported values for cell surface proteins. These NGO-templated molecular nanoclusters, used to stimulate NK cells via the CD16 receptor, successfully induced cellular activation, indicated by degranulation of cytolytic granules and IFN-γ secretion. Importantly, activation significantly exceeded that induced by the same antibodies applied as a solution of individual molecules. These results demonstrate that future immunotherapies could be enhanced by assembling immunomodulatory drugs into nanoclusters and establish NGO-templating as a candidate technology.


Conjugation of NGO with 8-arm star-PEG-amine
NGO was purchased as an aqueous solution from 2D-Tech, Manchester, UK. Firstly, this solution was reacted with chloroacetic acid to generate -COOH groups on the surface (NGO-COOH). The as-received solution was diluted to 1 mg/mL, (ultrapure water 18.2 MW), and ultrasonicated for 1 hour to break up any aggregates before addition of chloroacetic acid (200 mg/mL) in combination with sodium hydroxide (240 mg/mL) which also serves as a base-wash to remove organic impurities created during the NGO synthesis. The reaction was continued for 4 hours under ultrasonication before the resulting NGO-COOH was purified by repeated rinsing with ultrapure water using a centrifugal filter (100 kDa cutoff, Amicon Ultra, Millipore) and the pH adjusted to ~7 using hydrochloric or sulfuric acid (1M). The presence of carboxylic acid groups was confirmed by the appearance of an IR absorbance at 1591 cm -1 (Fig. 1B). The second stage of the reaction conjugated 8-arm star-poly(ethylene glycol) amine with total Mw 40 kDa (star-PEG-amine, Creative PEGWorks. Star-PEG amine (20 mg/mL) was added to the NGO-COOH suspension with 5 mins ultrasonication before N-(3-methylaminopropyl-N'ethylcarbodiiimide) was added (to final concentration 8 mmol/L) and left to react overnight with agitation. Unreacted EDC was quenched by addition of 2-mercaptoethanol (to final concentration 40 mmol/L), and the NGO-PEG product was purified by centrifugal filtration with repeated rinsing as above. If required an additional centrifugation (16000 g, 4 h) discarding the pellet was used to remove any remaining large aggregates leaving the stable product. For comparison of stability a conjugate of NGO with 4 arm star-PEG-amine (Mn 10 kDa, Sigma Aldrich) was produced using the same method with 2 mg/mL PEG and 4 mmol/L EDC, quenched with 20 mmol/L 2-mercaptoethanol.

S3
NGO conjugated with 8-arm star-PEG amine was biotinylated using an NHS-PEG4-amine cross-linker (EZ-Link TM , ThermoScientific) in PBS with MgCl2 and CaCl2 in accordance with the manufacturer's instructions. The product was purified by rinsing and centrifugal filtration as above. Biotinylation was confirmed using a HABA/avidin assay (Pierce Biotin TM Quantification Kit) as per the manufacturer's instructions except for an adjustment for NGO absorbance achieved by additionally measuring a non-biotinylated NGO-PEG sample.
Streptavidin coating was carried out at large excess in order to minimize the possibility of cross-linking between NGO sheets. In a typical case, biotinylated NGO-PEG (0.6 mg/mL) was mixed with streptavidin (5 mg/mL) with thorough mixing. Excess streptavidin was removed through centrifugal filtration with repeated rinsing as above.

Binding of antibodies to streptavidin-coated NGO-PEG construct
Streptavidin-coated NGO-PEG (0.4 mg/mL) was mixed with biotinylated antibodies (to final concentration 0.418 mg/mL) in PBS with MgCl2 and CaCl2 and allowed to react overnight at 4 °C. The relative concentrations were calculated to allow the NGO construct to entirely deplete the antibody solution leaving no unbound antibody. The antibody-functionalized NGO-PEG was separated into fractions by centrifuging at 10,000g and extracting the pellet every 5 mins. Successive pellets corresponded to different size distributions of the NGO-mAb constructs, with the larger nanoclusters concentrating into the earlier pellets. In total four fractions were separated and fractions 1-3 resuspended producing samples with distinct size distributions as indicated in the main text and Fig 2E.

Bead-binding assays
Compensation Beads (coated with anti-mouse IgG1k antibody, BD Biosciences). The coated beads were mixed 1:1 with uncoated beads and diluted with equal volumes of PBS. In binding assays, 50 µL were mixed with NGO-mAb (1 µL, 200 µg/mL, final concentration 3.92 µg/mL) molecular nanoclusters and their control antibodies and incubated using the same S4 conditions as used for cell binding assays below, after which they were washed with PBS, and analyzed using flow cytometry, again as for cell binding assays. In a negative control experiment, the beads showed no binding to a soluble antibody solution for which they have no affinity (polyclonal goat a-mIgG2b, Thermo Scientific).

Characterization of NGO-templated molecular nanoclusters
Thermogravimetric analysis (TGA) TGA used a Mettler Toledo TGA0-DSC with sample masses of ~ 1 mg. Experiments were carried out in a N2 atmosphere using a temperature range of 100 -800 °C and a heating rate of 10 °C/min. PEG was entirely pyrolized between 300 -440 °C.

Atomic force microscopy (AFM)
AFM measurements were made on a Bruker Innova microscope in Tapping Mode using aluminium-coated silicon probes NCHV-A (Bruker Nano). Samples were prepared by depositing a sample droplet onto a silicon wafer substrate and allowing to dry in the air.

Quantification of the number of antibodies per NGO-mAb nanocluster by UV-visible spectroscopy
The concentration of NGO sheets in a given sample was determined by UV-visible spectroscopy at 700 nm. An extinction coefficient at this wavelength for NGO-PEG was first determined by generating a calibration curve using a pre-weighed sample, with the ratio of NGO-PEG determined from TGA. In combination with the known sheet size from AFM, this gave the number concentration of NGO sheets. The concentration of antibody molecules was then determined using UV-visible spectroscopy at 546 nm with subtraction of the NGO-PEG background, based on the labelling of antibodies with AlexaFluor546 dye, (Fig. 1F). These concentration data were combined with the nanosheet size determined by AFM to indicate the number of antibody molecules per nanosheet. To ensure that no confounding free molecules S5 were present in cell experiments, a sample of each batch was removed and pelleted by centrifugation, with UV-visible spectroscopy on the supernatant confirming the absence of such free molecules. The experimental uncertainties derived from fitting the UV-visible data are estimated to introduce a standard error of ± 40 nm into the calculated antibody numbers. This is independent from the interquartile ranges quoted in the main text which characterize the statistical distribution of antibody numbers deriving from the distribution of NGO-mAb nanocluster sizes measured by AFM.

Primary human NK (pNK) cells pNK cells were obtained from healthy donor peripheral blood concentrates (NHS Blood and
Transplant) in the form of Leukopaks, with ethical approval. NK cells were isolated from the Leukopaks under negative magnetic selection (NK cell isolation kit, Miltenyi Biotec) and cultured in RPMI-1640 supplemented with 10% human serum, 1mM L-glutamine, 1mM sodium pyruvate, 1mM penicillin-streptomycin, 1mM MEM nonessential amino acids and 20 µM 2-mercaptoethanol (all ThermoScientific). For cell binding experiments, freshly isolated pNK cells were stimulated with 150 U/mL human recombinant IL-2 (Roche) and cultured for 6 days prior to use. For experiments investigating functional response to nanoclusters and antibodies, pNK cells were cultured in media with IL-2 for a total of 10 days before use with resuspension in fresh media containing IL-2 at days 7 and 9. Experiments carried out 1 day after the last IL-2 addition.

Cell-binding assays
The binding of NGO-constructs and soluble antibodies to pNK cells was assayed as follows.
pNK cells were resuspended in flow buffer (2% FCS in PBS) and incubated with unconjugated antibodies and NGO-mAb constructs with concentrations adjusted so that the overall S6 concentration of antibody was 2 µg/mL in all cases. The solution was incubated for 20 mins at 4 °C.

Stimulation of pNK cells with NGO-mAb molecular nanoclusters
pNK cells were suspended at 1 ´ 10 6 cells/mL (100 µL aliquots) in R10 with: anti-LAMP-1-AlexaFluor647 or its isotype control, Golgi Plug (BD Biosciences) and Monensin (eBiosciences) (both at 1:1000 dilution) to ensure only pre-synthesized existing IFN-g is secreted as described in the text. Stimulating NGO-mAb molecular nanoclusters and antibodies were added with a final overall concentration of 2 µg/mL antibody in all cases. Concentrations were determined using UV-vis absorbance, based on AlexaFluor546 labelling, with the soluble antibody taken from the same labelling batch as the NGO to ensure consistency. The cells were placed in 96-well-plates (Nunc Maxisorp) that had been prepared by washing with 0.05% Tween-20 in PBS, and briefly centrifuged (10s at 300g) to move them to the bottom of the wells. As a positive control to ensure that cells were capable of being stimulated, some cells were plated (with no added soluble antibodies or NGO-mAb reagent) onto a stimulating surface consisting of a well that had been coated with anti-CD16 antibodies (overnight incubation of 50 µL solution at 1 µg/mL): donors that did not fulfil this condition were discarded. Cells were incubated at 37 °C for 6 h prior to readouts of CD107a degranulation and IFN-g secretion/expression. To confirm that the NGO-mAb nanoclusters did not impact cell viability Zombie Aqua cell viability stain (Biolegend) was added to distinguish live from dead cells. For cell viability plots gating on the lymphocyte population by FSC-A, SSC-A followed by gating for Zombie Aqua negative cells to establish percentage viability. Flow cytometry was carried out using a BD LSRFortressaTM and anaylsed using FlowJoTM version 10. The NGO-mAb nanoclusters did not significantly impact cell viability (Fig. S3).