Targeting SOD1 via RNAi with PEGylated graphene oxide nanoparticles in platinum-resistant ovarian cancer

Acquired platinum resistance poses a significant therapeutic impediment to ovarian cancer patient care, accounting for more than 200,000 deaths annually worldwide. We previously identified that overexpression of the antioxidant superoxide dismutase 1 (SOD1) in ovarian cancer is associated with a platinum-resistant phenotype via conferring oxidative stress resistance against platinum compounds. We further demonstrated that enzymatic inhibition using small-molecule inhibitors or silencing of SOD1 via RNA interference (RNAi) increased cisplatin sensitivity and potency in vitro. We launched this study to explore the potential therapeutic applications of SOD1 silencing in vivo in order to reverse cisplatin resistance using a graphene-based siRNA delivery platform. PEGylated graphene oxide (GO) polyethyleneimine (GOPEI-mPEG) nanoparticle was complexed with SOD1 siRNA. GOPEI-mPEG-siSOD1 exhibited high biocompatibility, siRNA loading capacity, and serum stability, and showed potent downregulation of SOD1 mRNA and protein levels. We further observed that cisplatin and PEI elicited mitochondrial dysfunction and transcriptionally activated the mitochondrial unfolded protein response (UPRmt) used as a reporter for their respective cytotoxicities. SOD1 silencing was found to augment cisplatin-induced cytotoxicity resulting in considerable tumour growth inhibition in cisplatin-sensitive A2780 and cisplatin-resistant A2780DDP subcutaneous mouse xenografts. Our study highlights the potential therapeutic applicability of RNAi-mediated targeting of SOD1 as a chemosensitizer for platinum-resistant ovarian cancers.


INTRODUCTION
Ovarian cancer, with an annual worldwide incidence of 313,000 cases and mortality of 207,000, is considered the second most lethal gynaecological malignancy [1].Despite the favourable overall survival when detected in early stage 1-attributed principally to its asymptomatic disease progression-most cases are diagnosed in advanced stages 2 and 3 with a 5-year survival of 31%.The overall global survival has only increased modestly in previous decades due to the constrained availability of treatment options and the clinically acquired chemotherapy resistance [2].Despite most patients responding to debulking surgery and combinational treatment with platinum and taxols, almost half the patients eventually develop recurrence and become resistant or refractory to additional platinum-based therapeutic interventions [2].
Despite their adverse side effect profile, platinum compounds are still considered potent as the first-line drug of choice as monoor synergistic therapy for many solid tumours [3][4][5][6].Besides nuclear DNA adduct formation and subsequent induction of apoptotic signalling, platinum drugs also elicit mitochondrial dysfunction through oxidative stress and mitochondrial DNA (mtDNA) damage [7,8].Contingent on the intrinsic platinum sensitivity of the respective malignancy, platinum drugs initially exhibit robust therapeutic efficiency.However, the development of acquired platinum resistance of recurrent tumours renders their subsequent clinical applicability ineffective.Lung, prostate, and colorectal cancers are intrinsically resistant to platinum; acquired resistance is also more common in epithelial ovarian cancer [9].Implicitly, preventing or reversing this well-documented clinical phenomenon could have profound and widespread clinical therapeutic benefits.
Intrinsic and acquired platinum resistance have been linked to reduced drug uptake, increased efflux, enhanced detoxification, elevated scavenger levels, increased oxidative stress tolerance, upregulated DNA repair mechanisms, and the reprogramming of cellular metabolism to evade cisplatin-induced death [10,11].Cancer cells may exhibit one, or more of the aforementioned peculiar mechanisms that ultimately determine their net sensitivity to platinum [12,13].However, to date, the targeting of drugtransporting pathways of MDR-1, ATP-7A/7B, CTR1, MRP, and DNA repair pathways, including BRCA1/2, ERCC1, and MMR, provided only modest improvement in survival clinically [14].In addition, the utilization of MDR1 inhibitors, including but not limited to Zosuquidar, MK-571, and PSC-833, showed moderate efficacy in terms of platinum chemosensitization in ovarian cancer [15].Thus far, the exact mechanisms and determinant factors driving platinum resistance have not been fully elucidated.Due to genetic and patient sample heterogeneity, patient-specific expression levels of potentially robust platinum resistance biomarkers make the discovery process burdensome.
In the quest for a novel therapeutic target for cisplatin resistance, we previously identified the ROS-neutralizing SOD1 to be overexpressed in cisplatin-resistant ovarian cancer cell lines using quantitative label-free comparative proteomics analysis [16].The 32 kDa Cu 2+ /Zn 2+ SOD1 is an abundantly expressed intracellular homodimeric metalloenzyme that converts superoxide anions to hydrogen peroxide that is subsequently transformed by catalase to oxygen and water [17].Superoxide neutralization by SOD1 is a crucial mechanism in counteracting oxidative damage, as SOD1 knockout Drosophila exhibits reduced lifespan, infertility, and hypersensitivity to oxidative stress [18].Other studies concluded that RNAi-mediated SOD1 silencing provokes senescence in human fibroblasts and induces apoptosis in HeLa cells via ROS-mediated induction of TP53 [19].Further, SOD1 plays an essential role in cytoplasmic NRF2-mediated adaptation to oxidative stress and in the mitochondria via UPR mt [20].Thus, our previous results suggest the role of SOD1 overexpression as a defense mechanism of ovarian cancer cells to counteract platinum-induced oxidative stress and modulation of ROS-mediated redox signalling [16].We further concluded that enzymatic inhibition of SOD1 using copper/zinc chelating agents of TETA and ATN-224 reversed platinum resistance [21].Subsequently, to overcome the off-target side effects caused by nonspecific small molecule metal chelators, we further corroborated the chemosensitizing effect of SOD1 downregulation via RNAi in vitro [22].
RNAi is a validated gene therapeutic method to control posttranscriptional gene regulation in various disease states [23].However, the exogenous introduction of therapeutic siRNA for in vivo applications faces numerous obstacles, including easy degradation, short half-life, filtration by the kidneys, poor cellular uptake due to inherent negative charge, structural instability, and degradation by RNases [24].Therefore, a meticulous design of any siRNA delivery system is fundamental to fully capture the potential of siRNA therapeutics.Non-viral gene delivery vectors such as cationic lipids, cationic polymers, and polysaccharides, in particular, have gained momentum due to their efficacy and biosafety compared to viral vectors for delivering DNA or RNA cargo into the cells [25].
The two-dimensional graphene oxide (GO), due to its convenient applicability for non-covalent or covalent functionalization via abundant epoxy, carboxyl, and hydroxyl surface groups, has been widely used in combination with various cationic polymers for siRNA and gene delivery [26,27].GO is also highly biocompatible in most in vitro and in vivo test systems due to its high colloidal stability, water dispersibility, and surface-tovolume area ratio [28].We previously synthesized graphene-based siRNA, small molecule, and combined drug delivery systems for various applications [29].Consequently, in this study, a novel graphene-based nanoparticle platform (GO PEI -mPEG) was prepared by sequential coupling of nano-graphene oxide with cationic polymer polyethyleneimine (PEI) to achieve SOD1 siRNA delivery and further with polyethylene glycol (PEG) to increase biocompatibility and control surface charge.In addition, the incorporation of GO, due to the fixed lateral dimensions of the 2D sheets, can allow for improved control of a more uniform and reproducible final hydrodynamic nanoparticle size, compared to using PEI-PEG polyplexes alone [30,31].Our study aimed to evaluate the in vivo chemosensitizing efficacy of SOD1 siRNA by GO PEI -mPEG delivery system in cisplatin-resistant mouse xenograft models.

Nanoparticle preparation
Graphene oxide (GO) synthesis.GO was prepared with a modified Hummer's method [32].The mixture of NaCl (35 g) and native graphite flakes (1 g, Alfa Aesar, Haverhill, MA, USA) was ground with a mortar until the colour of the mixture turned grey, then dissolved in deionized water (DIW).NaCl was extracted by multiple washing steps and ultracentrifugation at 8000 rpm for 5 min.The product was dried at 90 °C for 12 h and stirred in a three-necked bottle with H 2 SO 4 (23 mL) for 8 h.In an ice bath, KMnO 4 (3 g) was added to the mixture keeping the temperature below 20 °C, until the colour turned dark green.The solution was further stirred in a dimethyl-silicone oil bath at 38 °C for 30 min and at 70 °C for an additional 45 min until the colour of the solution turned dark brown.The mixture was diluted with DIW (5 mL) and stirred for 10 min.Next, the solution was diluted with DIW (40 mL) and stirred for 15 min at 100 °C.Subsequently, the mixture of DIW (10 mL) and H 2 O 2 (10 mL) was added and stirred for 10 min, and the final product was washed twice with 5% HCl (50 mL) and numerous times with DIW via ultracentrifugation.The prepared GO batch was dialyzed using 100 kDa dialysis bags.Next, GO (8 mL) was washed twice with DIW, and the -COOH groups were activated by adding NaOH (1.8 g).Finally, the suspension was diluted with DIW up to 15 mL and stirred at 55 °C for 4 h.Next, the solution was neutralized by adding HCl (5 mL), and the mixture was washed multiple times until the pH became neutral.Following a 30-min bath sonication and centrifugation at 13,000 rpm for 5 min, the concentration of the supernatant was measured with an Evolution 201 UV-Vis spectrophotometer (Thermo-Fisher Scientific, Carlsbad, CA, USA).

GO PEI synthesis
The conjugation of COOH groups of GO and primary amines of PEI was achieved with zero-length carboxyl to amine carbodiimide (EDC) crosslinking.Carboxyl groups were activated with EDC forming an active but unstable O-acylisourea intermediate displaced by nucleophilic attack from the primary amines of PEI.These primary amines formed an amide bond with the activated carboxylic groups.In brief, GO (5 mg) was diluted up to 15 mL in phosphate-buffered saline (PBS) and was bath sonicated for 30 min in ice before adding 30, 60, or 90 mg of PEI 25 kDa (Sigma-Aldrich, St. Louis, MO, USA) dissolved in 200 µL of PBS (HyClone Laboratories Inc, Logan, UT, USA).Following 5 min of sonication, NaOH (20 µL) and 5 mg of Pierce™ Premium Grade EDC [1-Ethyl-3-(3-dimethyl aminopropyl) carbodiimide] (Thermo-Fisher Scientific) were added, respectively, then sonicated for 5 min and stirred for 30 min.GO PEI was prepared using three different PEI concentrations; 30, 60, and 90 mg/mL corresponding with GO PEI1X , GO PEI2X , GO PEI3X , respectively.Next, 10 mg of EDC in 400 µL PBS was added under continuous stirring, sonicated for 30 min, and then stirred overnight.The final EDC concentration was 1 mg/mL in a 10-fold molar or weight excess to GO.The prepared GO PEI nanoparticles were purified with ultrafiltration using a 100 kDa Amicon Ultra-15 centrifugal filter (Merck Millipore Ltd., Darmstadt, Germany).

GO PEI -mPEG synthesis
Methyl-PEG (2000 kDa, 5 mg) (Xi'an Ruixi Biological Technology Co., Ltd, Shaanxi, China) in PBS was diluted to 15 mL and sonicated for 5 min.NaOH (20 µL) and EDC (5 mg) were added and sonicated for 30 min.Next, GO PEI (30 mg) was added, and the solution was sonicated for 5 min before adding EDC (10 mg), then stirred overnight.The nanoparticle was purified with ultrafiltration using a 100 kDa filter (Millipore Sigma, Burlington, MA, USA).

Nanoparticle characterisation
Atomic force microscopy (AFM).The size and surface morphology of GO and GO PEI were characterized with AFM.The nanoparticle samples (200 μL) were transferred onto a 1.5 cm × 1.5 cm muscovite mica sheet and imaged with Veeco Dimension 3100 atomic force microscope (Veeco Instruments, Bruker, MA, USA).The images were analyzed using V700 (Veeco) and Nanoscope v.7.00b19 (Veeco) software.
Transmission electron microscopy (TEM).The lateral size and qualitative structural properties of GO and GO PEI were assessed with TEM.The nanoparticle samples were transferred onto a 20 cm × 20 cm muscovite mica sheet, then imaged by a Tecnai G2 F20 S-Twin transmission electron microscope (Thermo-Fisher Scientific), and analyzed by Digital Micrograph 3.0 (Thermo-Fisher Scientific) software.
Cancer Gene Therapy (2023) 30:1554 -1568 Dynamic light scattering (DLS).The average size distribution (ASD), the polydispersity index (PDI), and surface charge (ζ-potential) of GO, GO PEI , GO PEI -mPEG nanoparticles were determined with DLS.The samples (100 μL) were first filtered, then diluted up to 2 mL with DIW and transferred into a ZEN0112-low volume disposable sizing cuvette for ASD and PDI and into a DTS1060C-Clear disposable zeta cell for ζ-potential analysis.The measurements were performed with a Zetasizer Nanoseries Nano-25 (Malvern, United Kingdom) in batch mode at 25 °C in triplicates with 120 sec equilibration time.The results were analyzed with the Zetasizer (Malvern, United Kingdom) software.
Fourier transform infrared (FT-IR) spectroscopy.The confirmation of successful covalent amide bond formation between GO and PEI was determined with FT-IR spectroscopic analysis for GO, GO PEI , GO PEI -mPEG samples with a Cary 600 series (Agilent Technologies, Santa Clara, CA, USA) FTIR Spectrometer, and all obtained results were analyzed with the Resolution Pro (Agilent Technologies, Santa Clara, CA, USA) and displayed with the GraphPad Prism 7 software (GraphPad Software Inc, Boston, MA, USA).
Ultraviolet-visible (UV-Vis) spectroscopy.The nanoparticle absorbance spectra and the siRNA oligonucleotide loading capacity were measured and confirmed with UV-Vis spectroscopy.The GO (0.01 mg/mL), PEI (0.005 and 0.01 mg/mL), GO PEI , GO PEI -siSCR (scrambled negative control), GO PEI -mPEG, and GO PEI -mPEG-siSCR samples complexed at w/w ratio of 3:1 (nanoparticle : siRNA) at RT for 30 min and scrambled negative control siRNA (siSCR) only samples (50 μL) were dissolved in DIW (2 mL) in a 10 mm quartz cuvette (3.5 mL).The absorbance spectrum was recorded compared to DIW as a calibration control.The UV-Vis spectra were measured with an Evolution 201 UV-Vis spectrophotometer (Thermo-Fisher Scientific) between 190 and 800 nm in absorbance mode and were analyzed with the Thermo Cue (Thermo-Fisher Scientific) software.
Thermogravimetric analysis (TGA).The composition and PEI surface coating of lyophilized GO and GO PEI nanoparticles were quantitatively assessed with an STA 449 F3 Jupiter® thermal analyzer (Netzsch, Germany).Under inert nitrogen purge (50 mL/min), the samples were heated (25-800 °C, 50 °C/min), and the change of the initial nanoparticle mass (5 mg) was measured as a function of time and temperature.Ceramic crucibles with loosely fitted lids containing the samples were held at 25 °C for 5 min prior to measurements to allow for equilibration between the crucible and the furnace.To correct the bias caused by the buoyancy effect, a blank measurement with an empty ceramic crucible under the same temperature program as the samples were performed.Results were analyzed with the Netzsch Proteus Thermal Analysis software and displayed as a thermogravimetric temperature vs mass curve with the following formula:

In vitro assays
Cell culture.A2780 cisplatin (CP) sensitive and A2780 DDP CP resistant human ovarian epithelial adenocarcinoma cell lines were a kind gift from Sun Yat-sen University Cancer Center (Department of Gynecology, Guangdong, China).A2780 and A2780 DDP cell lines were grown in full media consisting of Dulbecco's Modified Eagle Media (DMEM, HyClone Laboratories Inc, Logan, UT, USA) media supplemented with 10% FBS (Gibco, Thermo-Fisher Scientific, USA) and 1% penicillin/streptomycin.The cell lines were grown in a humidified incubator at 37 °C with 5% CO 2 .The cisplatin-resistant phenotype of A2780 DDP was maintained by supplementing 20 µg/ml cisplatin (Sigma-Aldrich) every 4 weeks for 24 h.For cell passaging, the media was removed, and cells were washed twice with PBS, then 5 ml of 0.25% (w/v) trypsin-EDTA (Sigma-Aldrich) was added and incubated in 5% CO 2 at 37 °C for 6-7 min.Next, 9 mL of full media was added.The cell suspension was centrifuged in a polypropylene conical centrifuge tube (15 mL) with a Sorvall™ ST 16 Centrifuge (Thermo-Fisher Scientific) at RT for 3 min at 800 rpm.Following removing the supernatant, cells were re-suspended and counted manually under an Eclipse TS 100 inverted microscope (Nikon, Japan) with a modified Neubauer hemocytometer using Trypan blue (Invitrogen) to quantify cell viability.
Live cell FAM uptake.Based on graphene-based nanomaterials' ability to quench specific fluorophores' fluorescence, siSCR FAM was complexed with GO PEI and GO PEI -mPEG, and the cellular uptake and siSCR FAM release in A2780 DDP cells were measured at excitation and emission wavelengths of 488 nm and 520 nm, respectively.The nanoparticle samples were precomplexed with siSCR FAM at RT for 30 min in triplicates in a final volume of 50 µL and were aliquoted into a 96-well plate, while DIW, RPMI media, nanoparticles only, and naked siSCR FAM served as negative controls.Under maintained CO 2 levels (5%) and temperature (37 °C), the fluorescent signal was recorded for 12 h with a Varioskan LUX (Thermo-Fisher Scientific) plate reader, and the results were analyzed with the SkanIt TM (Thermo-Fisher Scientific) software.
Laser confocal microscopy.A2780 and A2780 DDP cells (5 × 10 4 ) were seeded in collagen-coated (d = 35 mm) gamma-irradiated glass bottom micro-well confocal dishes (MaTek Corporation, Ashland, MA, USA) and incubated overnight in a 5% CO 2 incubator at 37 °C.siSCR FAM (2 µg) dissolved in Opti-MEM (500 µl) reduced serum media (Gibco, Carlsbad, CA, USA) served as the negative control, GO PEI (6 µg) and GO PEI -mPEG (6 µg) were complexed with siSCR FAM (2 µg) at w/w ratio of 3:1 for 30 min, and 15 µg of Lipo 2000 (1 mg/mL) (Thermo Fisher) as a positive control was complexed with siSCR FAM (2 µg) for 30 min, Lipo 2000 and GO PEI without siSCR FAM were used as negative controls, and cells without treatment as blank controls.Next, the media was removed following overnight incubation with the test samples, the dishes were washed with ice-cold PBS thrice, and serum-free RPMI media (500 µL) was added.The confocal dishes were completely covered with aluminium foil and incubated for 5 h in a 5% CO 2 incubator at 37 °C.Next, the cells were washed with ice-cold PBS twice and fixed with formaldehyde (3.7%) for 10 min.For cell permeabilization, Triton X-100 (0.2%) (Sigma-Aldrich, St. Louis, MO, USA) was added for 5 min, then washed with PBS thrice.The nuclei of the cells were stained with DAPI (1 mg/ml) nuclear stain (Sigma-Aldrich, St. Louis, MO, USA) dissolved 1:5000 in PBS.Following a 20 min incubation at 40 °C in the dark, the cells were washed with PBS thrice.They were imaged in confocal mode with a Zeiss laser confocal scanning microscope at 10X and 40X resolution objectives.The excitation wavelengths for DAPI and siSCR FAM were 405 and 488 nm, while the emission was recorded in the 460-490 nm and 500-530 nm wavelengths, respectively.The obtained images were analyzed with the Zeiss LSM 880 (Zeiss, Oberkochen, Germany) and ZEN Imaging software 2.3 (Zeiss).
Fluorescence-activated cell sorting (FACS).The siRNA transfection efficiency of the nanoparticles was quantified with FACS.A2780 DDP cells (3 × 10 5 /well) were seeded in six-well plates and incubated overnight prior to transfection.The plates were transfected with GO PEI (0.4, 0.8 and 1.2 µg) complexed siSCR FAM (30, 60, 90 nM) at a w/w ratio of 3:1, Lipo 2000 (positive control) complexed siSCR FAM (30, 60, 90 nM), GO (22 µg) complexed siSCR FAM (90 nM), and blank (media only), naked siSCR FAM , GO PEI and Lipo 2000 served as negative controls.To assess the siRNA transfection efficiency of GO PEI -siSCR FAM complexes at 0.2:1, 0.8:1, 1:1, 2:1, 5:1, and 10:1 w/w ratios, GO PEI was mixed with siSCR FAM (40 nM) at RT for 30 min.For transfection, the media was first replaced with serum-free DMEM (500 µL), and the test samples suspended in Opti-MEM (500 µL) were added to the wells dropwise.Following transfection, the plates were incubated at 5% CO 2 at 37 °C for 5 h, then the media was replaced, and the plates were incubated for at least 48 h until they reached 80% confluency.The adherent cells were harvested by trypsinization and centrifugation and were re-suspended in ice-cold PBS (1 mL).The fluorescence of the cells was measured with a FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA) flow cytometer by analyzing 10,000 events per measurement.Cells were isolated according to their granularity and size predetermined by the Side Scatter (SSC-H), Forward scatter (FSC-H), and green FAM fluorescence band pass filter (FL1-525 nm) parameters with a BD CellQuest Pro (BD Biosciences) software, and the results were analyzed with the FlowJo software (BD Biosciences).
Population Doubling.To evaluate the effect of SOD1 knockdown on the proliferation rate, A2780, A2780 DDP cells (3 × 10 3 /well) in 100 µL, complete media were seeded into 96-well plates and incubated for 24 h.Next, cells were treated with the respective drug compound, siRNA-nanoparticles, or both in serum-free media (50 µl per well) and incubated in 5% CO 2 at 37 °C for 4 days.From day 0 to day 4, one 96-well plate was treated with MTT solution, and the absorbance readings were taken at 24 h, 48 h, 72 h, and 96 h post-seeding, respectively.The total normalized absorbance values of the respective plates were used to plot a cell growth curve.Population doubling-time calculations: Population doubling times were calculated using the slope of the angle of the linear regression analysis of the 5 time points and confirmed with the Doubling Time Software v1.0.10 (http:// www.doubling-time.com).
Clonogenic Assay.The cisplatin sensitivity and chemo-sensitizing effects of GO PEI -siSOD1(D-333) and GO PEI -mPEG-siSOD1(D-333) were determined with a clonogenic cell assay.A2780 and A2780 DDP cells (~1 × 10 3 cells/well) were seeded in 6-well plates and then incubated overnight with 5% CO 2 at 37 °C.Next, the different drug compounds, siRNA-nanoparticles, or both dissolved in RPMI (1 mL) media and added to the cells, then following 2-5 h of incubation, depending on the respective treatment, the media was replaced with full media, and the plates were incubated for 10-12 days.The colonies were fixed with methanol (25% v/v) for 1 min, stained with crystal violet (0.05% w/v) for 5 min and washed under running tap water.Colonies (min >50 cells) were counted manually, and the relative cell viability of the treatment groups was expressed relative to the control group.

In vivo study
Ethical approval and animal handling.Animal study protocols (ASP) were approved by the Institutional Animal Care and Use Committee (IACUC) of Suzhou GenePharma (Suzhou, China), and all in vivo experiments were conducted according to IAUCC guidelines.The animals used in this study were housed in the facility of Suzhou GenePharma, an AAALAC internationally accredited research centre.All in vivo experiments were conducted according to the Guide for the Care and Use of Laboratory Animals.Female Balb/c 6-8 weeks old nude mice were exposed to 12 h light/dark cycles, kept in specific pathogen-free conditions in filter-topped cages, and fed with standard rodent chow plus water ad libitum.In this study, mice were sacrificed when any tumour volume reached more than 2000 mm 3 , upon any visible distress expressed by the animals, or more than a 20% decrease in body weight.
Preparation of ovarian cancer cells.A2780 cisplatin-sensitive and A2780 DDP cisplatin-resistant ovarian cancer cells limited to maximum 10 passages were grown in T-75 flasks and harvested following sterile techniques in a biological safety cabinet in the logarithmic growth phase at 60-70% confluency.The full cell culture media (DMEM supplemented with 10% FBS) was aspirated and washed thrice with PBS (37 °C).Next, trypsin/EDTA (4 mL) was added, and the plates were incubated at 37 °C in 5% CO 2 for 7 min.The cells were dislodged by gentle plate tapping, and the detached cells were visualized under an inverted microscope.Complete cell culture medium (10 mL) was added to inactivate the trypsin solution, and a single cell suspension was obtained by vigorous mixing.The live cells were counted by mixing 100 μL of the cell suspension with 100 μL of trypan blue and counted in a hemocytometer.The cell number was confirmed by combining 100 μL of the cell suspension in 9900 μL of CASY TT Buffer and counted with the CASY TT automated cell counter (OMNI Life Science GmbH & Co. KG, Bermen, Germany).Next, the single cell suspension (13.8 mL) was transferred to a centrifuge tube (15 mL) and centrifuged at 1000 rpm for 3 min at RT.The supernatant was discarded, and the cells were reconstituted in serum-free media at concentrations of 1 × 10 6 , 5 × 10 6 , or 1 × 10 7 cells mixed with 50% matrigel per mL as per the experimental design for subcutaneous tumour inoculation experiments.The cells were inoculated in the mice subcutaneously within 30 min of preparation.
Nanoparticle hemocompatibility assay.To obtain the red blood cell (RBC) fraction, anticoagulated fresh mouse blood (5 mL) was centrifuged for 10 min at 2000 × g, then washed thrice with PBS.Next, 2% (v/v) RBC suspension buffer was prepared in PBS mixed with 5 concentrations of naked siSCR, GO-siSCR, GO PEI -siSCR, GO PEI -mPEG-siSCR, and Lipo 2000 -siSCR in a total volume of 100 μL added to 900 μL of 2% RBC suspension.The samples were incubated at 37 °C for 1 h, centrifuged for 10 min at 3000 × g, and the absorbance was measured at 540 nm using a SpectraMax i3 plate reader.
In vivo toxicity and blood biochemistry testing.The potential acute in vivo toxicity of the nanoparticles was evaluated with blood biochemistry testing.Balb/c nude mice were injected with siSCR, GO-siSCR, GO PEI -siSCR, GO PEI -mPEG-siSCR (n = 3) at a 0.5 mg/kg siRNA dose.Blood samples were collected 8 h after the drug administration.Whole blood samples without anticoagulants were centrifuged at 3000 × g at 4 °C for 10 min to obtain the serum fraction.Haematological testing was performed using an automatic haematology analyzer (MEK6400, Nihon Kohden, Japan) for serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total protein (TP), albumin (ALB), creatinine (CREA), uric acid (UA), blood urea nitrogen (BUN) lactate dehydrogenase (LDH) and amylase (AMY).

Statistical analysis
Unless stated otherwise, all experiments were carried out in triplicates (n = 3).The in vitro numerical data were analyzed for statistical significance with Student's two-tailed paired and unpaired t-tests and were expressed as mean ± SD using GraphPad Prism 7.04 (GraphPad Software, San Diego, CA.USA).In the study design, p < 0.05 was considered statistically significant and was denoted as *, **, ***, and **** for less than 0.05, 0.01, 0.001 and 0.0001, respectively.

Graphene nanoparticle preparation
Graphene-oxide (GO) was prepared according to a previous study (Fig. 1A) [32].Subsequently, polyethyleneimine (PEI) was conjugated onto the GO sheets using EDC chemistry, yielding a GO-PEI (GO PEI ) nanocarrier.Further, to increase biocompatibility and decrease surface charge, GO PEI was directly PEGylated on the primary amine groups of PEI (GO PEI -mPEG), allowing for a tunable surface charge configuration.TEM images of GO and GO PEI showed distinct hexagonal lattice structure of GO (Fig. 1B) and a homogenous, well-dispersed nanoparticle distribution exhibiting rough, wrinkled morphological domains of graphene sheets (Fig. 1C, D).A high-resolution image of GO PEI revealed darker sheet intensities relative to GO due to PEI functionalization (Fig. 1D).AFM imaging revealed two-dimensional sheet-like morphology with highly variable particle thickness and a lateral size ranging from 0.05 to 0.40 μm.(Fig. 1E, F) Following PEI conjugation, the sheet thickness of GO PEI increased by ~3-4 nm relative to GO. (Fig. 1G, H) The latter was further confirmed with z-section analysis showing darker sheet intensities on 3D topographic analysis of the GO PEI sample attributed to increased flake thickness due to PEI functionalization (Fig. 1I, J).

Graphene nanoparticle characterization
Next, we confirmed whether the conjugation of -COOH groups of GO to the primary amines of PEI occurred using FT-IR spectroscopy (Fig. 2A).A characteristic peak of amide bonding was observed at 1630-1695 cm −1 of GO PEI but was absent in GO samples [34].A broad peak at 1705-1730 cm −1 , characteristic of carboxyl groups, and further peaks at 1200-1300 cm −1 and 3200-3500 cm −1 corresponding to C=O and -OH groups were appreciable of GO.We further measured the composition of GO PEI using TGA.The thermogravimetric curve of GO PEI (Fig. 2B) showed a biphasic decomposition pattern relative to GO.Both samples exhibited rapid initial weight loss (<150 °C) attributed to remnant water evaporation.However, the mass loss of GO PEI in the 250-350 °C range increased considerably compared to GO due to PEI 25 kDa decomposition (T d PEI 25 kDa ~200 °C).Based on these results we concluded that PEI 25 kDa makes up ~70% of the total mass of GO PEI .Next, the absorbance spectrum (Fig. S1A) and electrostatic adsorption of siSCR cargo onto GO PEI (Fig. S1B) were evaluated with UV-Vis spectroscopy.The siSCR complexed GO PEI showed higher and superimposed (GO PEI + siSCR) absorbance spectrum compared to their individual peaks indicating electrostatic complexation of the siSCR cargo with the GO PEI carrier (Fig. S1B).In addition to conventional TEM-and AFM-based graphene characterisation, the nanoparticle size distribution was periodically monitored before in vitro assays for potential aggregation using DLS to lower batch-to-batch variations and to increase experimental reproducibility.Next, for our reference, we established that the prepared GO had an average size distribution of 2354.6 ± 423 nm (Fig. 2C) and a low ζ-potential of −40.4 ± 0.4 mV (Fig. 2D), showing high size heterogeneity in solution with a PDI of 0.81 ± 0.11 (Fig. 2E).Following the activation of the -COOH groups with NaOH, GO exhibited lower hydrodynamic size distribution (318.Further, we confirmed that the hydrodynamic diameter of the graphene-siRNA polyplexes changed by modulating the GO PEI -mPEG to siRNA weight-to-weight (w/w) complexation ratios (Fig. 2F).Finally, the loading capacity, electrostatic adsorption, and retention of siSCR of the nanoparticles were assessed by agarose gel retardation assay (Fig. 2G).Relative to the control naked siSCR bands, the nanoparticle complexed siSCR groups showed lower band intensities indicating partial or total siSCR loading and retention at different w/w ratios.GO PEI nanoparticles containing increased amounts of PEI fully complexed siSCR at w/w ratios of 2:1, 1.25:1, and 0.75:1, respectively, while GO PEI -mPEG fully complexed the siSCR cargo at 1.25:1 w/w ratio.However, nonfunctionalized GO failed to completely retain siSCR, as evidenced by the presence of free siSCR bands at high w/w ratios.However, GO showed partial retention of siSCR, which is consistent with prior literature [35].These findings were further validated in the last well of each nanoparticle group, where the nanoparticle to siSCR complexation ratio was 1:2, and the band intensity was comparable to the 0.5:1 group.

Evaluation of in vitro biocompatibility
The aggregation potential of nanoparticles in biological solutions was observed and measured over 1 h, 24 h, 1, 3, and 6 months (Fig. 3A).We found that GO immediately precipitated and formed visible aggregates upon mixing with 0.9% NaCl, PBS, RPMI (10% FBS), and FBS (50%) solutions that persisted during the observation period (6 months) (Fig. S1C).GO PEI , however, did not show any observable aggregation and remained a homogenous solution even after 1 month of incubation in DIW, PBS, RPMI (10% FBS), FBS (50%), NaCl (0.9%) or RPMI (10% FBS, pH: 6.4) at RT that aimed to simulate the microenvironment of solid tumours.Further, the serum stability of GO PEI complexed siSCR evaluated against RNase degradation showed adequate complexation and protection of the siRNA cargo by the prepared nanoparticles in high serum-containing solutions (Fig. 3B, C).While naked siSCR completely degraded and its band disappeared after 6 h, GO PEI nanoparticles showed no bands after complexation with siSCR throughout the study even after 48 h in 50% FBS-containing solution, which simulated human plasma compared to negative control samples (Fig. S1D).The agarose gel-based serum stability study results were further validated with a fluorescent kinetic study using mouse serum (Fig. 3D).The emitted fluorescence of nanoparticles complexed with siSCR FAM samples were recorded for 8 h.The negative control naked siSCR FAM group showed a rapid drop in fluorescence after 3 h and completely disappeared after 6 h, attributed mainly to degradation by RNAses of the mouse serum, which is in line with the results of the gel-based assay stability study.The GO PEI and GO PEI -mPEG complexed siSCR FAM samples steadily showed low fluorescence during the 8 h long study, indicating that a negligible amount of siSCR FAM was released from the nanoparticles during incubation compared to the positive control Lipo 2000 group.Overall, we established that the prepared GO PEI and GO PEI -mPEG nanoparticles could provide sufficient protection for the siRNA cargo for a minimum of 4 h and were deemed suitable for further in vivo experiments.
Next, we evaluated the kinetic cellular uptake of the nanoparticles in a kinetic in vitro assay (Fig. 3E).During a 10 h time-course study using live cells, we observed increased siSCR FAM fluorescence after 6 h for GO PEI and GO PEI -mPEG nanoparticles but not for naked siSCR FAM .We attributed this increase in siSCR FAM fluorescence to the endosomal escape of the nanoparticles mediated by the proton sponge effect of PEI 25 kDa in the acidic lysosomal environment, since under physiological and buffered pH conditions, our serum stability assay showed (Fig. S1D), that GO PEI managed to fully complex and prevent releasing the siSCR cargo for an extended period of 48 h [36].The qualitative intracellular uptake of GO PEI -siSCR FAM complexes was also investigated by laser confocal microscopy (Fig. 3F, Fig. S1E).A2780 cells, in addition to control groups, were transfected naked siSCR, GO PEI , Lipo 2000 , GO PEI -siSCR FAM , and Lipo 2000 -siSCR FAM to measure the intracellular fluorescence following successful siRNA cargo delivery by GO PEI to the cells.GO PEI -siSCR FAM and Lipo 2000 -siSCR FAM showed evident green fluorescence in the cytoplasm, while for naked siSCR FAM no fluorescence was detectable.Similarly, cells treated with media or nanoparticles only (GO PEI or Lipo 2000 ) showed no fluorescence.Finally, the quantitative cellular uptake of nanoparticle complexed siSCR FAM polyplexes were measured with flow cytometry (Fig. 3G  and S1F).Naked siSCR FAM was partially taken up by A2780 cells, and unfunctionalized GO showed minimal siRNA delivery capability.In contrast, GO PEI complexed siSCR FAM (60, 90 nM) at a 3:1 w/w ratio showed significantly increased fluorescence (43.89% and 63.21%, p < 0.0001) comparable to the positive control Lipo 2000 group with 66.00% and 66.20% using 60 and 90 nM siSCR, respectively.In addition, cellular uptake of the graphenebased nanoparticles was evident during protein isolation, as cell pellets showed evident dark discolouration attributed to the presence of intracellular graphene nanoparticles (Fig. 3H).GO PEI complexed with 90 nM siSCR FAM showed the highest fluorescence and thus was selected for further experiments.Finally, we determined that GO PEI -mPEG-siSOD1 can efficiently silence SOD1 expression on both mRNA (Fig. 3I) and protein (Fig. 3J) levels.In addition, time-course evaluation of SOD1 silencing showed potent downregulation of mRNA (Fig. 3K) and protein levels (Fig. 3L) after 24 and 72 h, respectively.We have also used the other three different cell lines, HeLa, A549, and Hep3B, to confirm the siRNA's ability to downregulate SOD1 expression at the protein level (data not shown).

Evaluation of in vitro characteristics in cell-based assays
In the present study, to model the development of acquired cisplatin resistance, we first exposed human epithelial A2780 ovarian cancer cells to low-dose cisplatin (1 µM) for 3 months (Fig. 4A).Next, the degree of acquired cisplatin resistance induction was determined in the A2780 and resistant pair of A2780 DDP cells (Fig. 4B, C), and we observed a ~3-fold increase in cisplatin IC 50 with MTT assay (p = 0.0007), that we confirmed with clonogenic assay (Fig. S1G, H).To establish causality between acquired cisplatin resistance and SOD1 overexpression in vitro, we first determined the baseline SOD1 mRNA and protein levels between the cell lines (Fig. 4D, E).Consistent with our previous findings, A2780 DDP exhibited a 1.61-fold increase in SOD1 mRNA (Fig. 4D) and a 1.3-fold increase in protein levels in vitro (Fig. 4E).The same cell line pair showed consistently increased (p > 0.001) SOD1 mRNA (Fig. 4F) while a notable 1.9-fold increase in protein levels in vivo (Fig. 4G) in four independent tumour quadrants (Fig. 4H).Interestingly, however, when measured in two outers and two inner tumour quadrants, we observed inhomogenous and fluctuating SOD1 mRNA expression levels between the respective quadrants, that we mechanistically validated on the protein levels in an independent study (unpublished observation).Based on these findings, we concluded that in our model, acquired cisplatin resistance is associated with SOD1 overexpression both in vitro and in vivo.To establish causality between cisplatin treatment and SOD1 overexpression, we measured the change of SOD1 mRNA levels upon a range of 0-30 µM cisplatin treatment.We found that cisplatin treatment elicited significant SOD1 mRNA overexpression (p < 0.0001) (Fig. 4I).In addition, we observed that GO PEI solely (1-8 µg/mL) could significantly (p < 0.001) increase SOD1 mRNA levels above 4 µg/mL attributed to the combined effect of oxidative stress induction by PEI and mechanical shear stress elicited by the GO sheets (Fig. 4J).Next, we evaluated the in vitro cytotoxicity of the nanocarriers and their effect on cell viability using the MTT assay (Fig. 4K).GO treatment showed relatively high cell viability (80%) even at 40 µg/ml concentration.In contrast, free PEI 25 kDa , as anticipated, showed concentration-dependent toxicity and caused complete cell death (p < 0.0001) at or above 20 µg/mL concentration.As per our previous findings, the conjugation of GO with PEI significantly increased cell viability compared to free PEI alone (p < 0.0001) [34], while subsequent PEGylation further decreased cellular toxicity.In addition, combinational treatment with GO PEI (9 µg/mL) with cisplatin (2-28 µg/mL) resulted in a considerable decrease in cisplatin IC 50 (7.71-6.24µg/ml) (Fig. 4L) relative to the GO PEI (9 µg/mL) treated control.However, we attributed this chemosensitization effect not to SOD1 knockdown but rather the toxicity of PEI, which evidently necessitated the PEGylation of the nanoparticle.
Next, we aimed to further understand the potential mechanism governing cisplatin and GO PEI elicited SOD1 mRNA upregulation and hypothesized the involvement of oxidative stress induction and mitochondrial damage.Cisplatin, GO, and PEI have been reported as causative agents of mitochondrial dysfunction [37].In the mitochondria, SOD1 is located only in the intermembrane space.In healthy non-pathologic cells, it neutralizes superoxide anions generated by complex III of the electron transport chain, and it is also part of the executive branch of the UPR mt mediating mitoprotective functions primarily upon mitochondrial oxidative stress [38].Thus, to explore the relationship between cisplatin or nanoparticle-mediated toxicity and the proposed activation of the UPR mt , using a set of UPR mt -specific primers, we monitored the transcriptional activation of the respective signalling pathways and found that cisplatin-induced mitochondrial damage can activate the UPR mt (Fig. 4M) in A2780 ovarian cancer cells on the mRNA and protein levels (unpublished observation).Further, we concluded that both GO PEI and GO PEI -mPEG had similar effect in the same cell line (Fig. 4N), while PEGylation did not entirely prevent the ROS-mediated activation of the UPR mt signalling.In addition, GO PEI and GO PEI -mPEG treatment of A2780 DDP cells showed partial UPR mt activation (Fig. 4O), however, at a significantly lower magnitude.Further evaluation of additional mechanistic links between cisplatin or graphene nanoparticlemediated UPR mt activation is beyond the scope of this study.Although, evidently different compounds can elicit varying degrees of UPR mt activation (Fig. 4P), implying specificity in the underlying mechanism and potential correlation with cisplatinsensitivity, that merits further investigation in the future.Overall, based on the in vitro study results, we concluded that the prepared GO PEI -mPEG nanoparticle has acceptable biosafety properties for subsequent in vivo studies.

SOD1 knockdown reverses cisplatin resistance in vitro
The in vitro chemo-sensitizing effect of GO PEI -mPEG-siSOD1 was first evaluated by co-treating A2780 DDP cells with siSOD1 (65 and 130 nM, respectively) (Fig. 5A).We observed that GO PEI -mPEG-siSOD1 treatment lowered cisplatin IC 50 values (p < 0.0001) from a baseline 18.31 ± 0.53 µg/mL to 7.42 ± 0.79 upon treatment with 130 nM siSOD1 complexed GO PEI -mPEG relative to the GO PEI -mPEG-siSCR control group (Fig. 5B).Further, we observed that SOD1 knockdown caused considerable cell growth inhibition, evident from the drop in population doubling times in both examined cell lines (Fig. 5C, D).In addition, the in vitro therapeutic efficiency of SOD1 knockdown was measured using ultra-low attachment plates, clonogenic assay and MTT assay using a fixed concentration of 15 µM cisplatin.We concluded that under all of these conditions, SOD1 knockdown sensitized the A2780 DDP cells to cisplatin (Fig. 5E-G).Finally, we measured the total ROS, superoxide, and mitochondrial membrane potential using DCHF-DA, MitoSox-Red, and JC-1 fluorescent dyes (Fig. 5H).We found that SOD1 knockdown alone and in combination with cisplatin treatment resulted in elevated total ROS (p < 0.0001) and further caused mitochondrial membrane depolarization (p < 0.05).We concluded that SOD1 knockdown is associated with significant cellular oxidative stress and mitochondrial toxicity, which may contribute to the sensitization of cisplatin in the A2780 DDP cisplatin-resistant cells.
Evaluation of in vivo biodistribution, biosafety, and therapeutic efficacy of graphene-based nanoparticles The in vivo pharmacokinetic and pharmacodynamic properties of GO PEI -mPEG were evaluated using whole-body fluorescent imaging (Fig. 6A).Using normal saline and naked siSCR Cy3 as controls, mice bearing A2780 DDP xenografts were administered GO, GO PEI , and GO PEI -mPEG complexed with siSCR Cy3 via the tail vein and the nanoparticle distribution was monitored at different times points (0, 1, 4, and 8 h).The fluorescent intensity of naked siSCR Cy3 , GO PEI , and GO PEI -mPEG groups gradually increased during the study.However, no significant change in fluorescence was observed with non-functionalized GO, consistent with previous reports [35,39].The recorded fluorescent intensity was comparable between GO PEI -siSCR Cy3 and GO PEI -mPEG-siSCR Cy3 groups compared to their respective siSCR control groups.We concluded that GO PEI and GO PEI -mPEG could accumulate in tumour tissues and deliver the siSCR Cy3 cargo.In addition, these results suggested that PEGylation of GO PEI did not significantly alter the drug loading and delivery capacity of the nanoparticle, a common issue encountered following PEGylation of nanoparticles [40].
In addition, blood biochemistry parameters (Fig. 6B-J) were obtained to monitor acute systemic toxicity following intravenous nanoparticle administration via the tail vein.We found that GO PEI -mPEG exhibited lower liver function indices, including ALT (p < 0.05) and AST (p < 0.05), than GO PEI .However, both GO PEI and GO PEI -mPEG nanoparticles showed significantly higher levels of AST (p < 0.0001 and p < 0.001, respectively), while only GO PEI treatment showed ALT elevation (p < 0.001) compared to the blank saline control group.Further, only GO PEI -mPEG showed UA level elevation (p < 0.001), while both nanoparticles significantly raised serum LDH levels (p < 0.001 and p < 0.01, respectively).We concluded that both nanoparticles elicited toxicity to a certain degree in vivo.Furthermore, PEGylation of GO PEI did not entirely circumvent but still considerably decreased the toxicity induced presumably by GO or the PEI cationic polymer functional group.
Further, in vitro, hemocompatibility testing using whole mouse blood showed that GO PEI -complexed siSCR induced a significantly higher (p < 0.0001) hemolysis rate relative to naked siSCR and GO controls in a concentration-dependent manner (0-160 nM).This effect was considerably reduced by the PEGylation of GO PEI (p < 0.01) (Fig. 6K).It has been reported previously that cationic gene-drug carriers would cause blood toxicity by hemolysis and organic damage after administration in vivo [41].As both GO and PEI have been known to induce hemolysis, we concluded that appropriate surface modification using PEG could effectively improve the biocompatibility of the prepared delivery carrier and moderate hemolytic toxicity in our study.
To establish causality between the potential acute toxicity observed with blood biochemistry testing and the associated pathological changes elicited by the nanoparticles in various organ systems, mice were sacrificed 8 h after intravenous administration of the formulations, and the heart, liver, spleen, kidneys, and lungs were subjected to standard hematoxylin and eosin (H&E) staining.As shown in Fig. 6L, no appreciable pathological findings were observed in the respective control saline/siSCR Cy3 and GO/siSCR Cy3 groups in agreement with our blood biochemistry results.However, non-PEGylated GO PEI nanoparticles elicited evident and visible structural damage to the liver and kidney by disrupting the physiological histology of the organ parenchyma.This effect was partially mitigated by nanoparticle PEGylation in the GO PEI -mPEG group, although a certain degree of organ damage was still visible in the liver.We concluded that nanoparticle PEGylation considerably decreased the off-target toxicity caused by the cationic polymer component of the nanoparticle; however, due to the propensity of the majority of systematically administered nanoparticles to accumulate in the liver, the IHC results indicated that a certain degree of hepatotoxicity and nephrotoxicity was associated with GO PEI -mPEG administration.
Next, the in vivo anti-tumour and cisplatin chemosensitization effect of GO PEI -mPEG-siSOD1 was evaluated in subcutaneous A2780 and A2780 DDP cell-derived xenograft models (Fig. 6M).Relative to the A2780 DDP group, identical 0.5 mg/kg intratumoral cisplatin dose induced a significantly higher tumour growth inhibition rate (276.3 ± 72.39 mm 3 , p = 0.0170) in the sensitive A2780 tumours, confirming that the cisplatin-resistant phenotype of A2780 DDP was maintained in vivo as well.Further, we concluded that combined naked siSOD1 and cisplatin treatment had a negligible overall effect on tumour growth compared to cisplatin treatment only (770.81 ± 135.33 mm 3 , p = 0.9976).In addition, we found that intratumoral combined GO PEI -mPEG-siSCR and cisplatin treatment elicited significantly higher tumour growth inhibition relative to the 0.5 mg/kg naked siSOD1 plus cisplatin group (460.61 ± 59.32 mm 3 , p = 0.0046).This increased growth inhibition effect was attributed to the previously identified toxicity of the graphene nanoparticle on cancer cells observed with IHC and blood biochemistry testing, which despite the PEGylation, was not entirely circumvented.Finally, we found that relative to GO PEI -mPEG-siSCR group, 0.5 mg/kg siSOD1 complexed with GO PEI -mPEG at a 3:1 w/w ratio co-administered with cisplatin resulted in significant tumour growth inhibition (237.5799± 85.18 mm 3 , p = 0.0265) and structural tumour tissue damage (Fig. 6N) during the 14-day long therapeutic window.These results indicated that relative to the control siSCR group, GO PEI -mPEG-siSOD1 elicited partial re-sensitization of A2780 DDP xenografts to cisplatin, which was attributed to the SOD1 knockdown via RNAi.

DISCUSSION
Post-treatment disease recurrence and cisplatin-resistance development are major therapeutic obstacles for ovarian cancer patients [2].Hence, in this study we aimed at addressing this clinical phenomenon and prepared a GO PEI -mPEG siRNA nanocarrier to selectively target SOD1 in vivo via RNAi, which we have previously identified as a potential target to reverse cisplatinresistance [22].GO PEI -mPEG exhibited favourable physicochemical characteristics for translational drug delivery applications regarding hydrodynamic diameter, ζ-potential, and PDI [24].
Following PEGylation, the cellular uptake studies confirmed the presence of adequate PEI surface charge to mediate endosomal escape via the intracellular proton sponge effect [42].This mechanism was confirmed with fluorescent live cellular uptake studies, as an abrupt increase in fluorescence of siSCR FAM was detected after 5 h indicating the detachment and regained fluorescence of siSCR FAM following the endosomal escape.GO PEI -mPEG showed appropriate biocompatibility and formed homogenous solutions in various media modelling in vitro and in vivo conditions.At the same time, the gel-and fluorescence-based kinetic studies implied the complexed siSCR cargo was protected from RNase degradation.Further, all the prepared PEIfunctionalized nanoparticles fully complexed siSCR at a 2:1 w/ w ratio.
The cytotoxicity of the PEI cationic polymer component was significantly reduced with PEGylation; still, IHC imaging revealed that GO PEI -mPEG accumulation in the liver elicited structural damage and hepatotoxicity.We attributed the cytotoxicity produced by the PEI functional group to oxidative stress induction, which led to evident SOD1 mRNA overexpression by GO PEI in a concentration-dependent manner [43].It can be concluded that non-PEGylated GO PEI can counteract the knockdown effect mediated by siSOD1, which, theoretically based on our original hypothesis, may increase cisplatin resistance in surviving cell populations and subsequent passages via clonal selection.The nanocarrier's size, shape, surface charge, and surface functionalization determine the net toxicity of a graphene nanoparticle [44].The increased ROS generation by concentrationdependent graphene treatment can cause oxidative stress-mediated mitochondrial damage and stress-reactive transcriptional factor activation, altering the gene expression profile of the affected cell [45].The resultant oxidative stress, thus, may upregulate vital antioxidant enzymes, including but not limited to CAT, PRX, SOD2, and SOD1, as a defense mechanism [46].For instance, Mendonca et al. showed that graphene oxide caused significant SOD1 overexpression from 1 h up to 7 days following transfection [46].A similar cellular effect was published by Jarosz et al., who found that both SOD1 and SOD2 were overexpressed upon treating HepG2 cells with GO in a concentration-dependent manner [47].
Further, we concluded that, like cisplatin, both GO PEI and GO PEI -mPEG nanoparticles elicited mitochondrial dysfunction, evident from the specific activation of various arms of the UPR mt pathways [48].GO-based nanoparticles and PEI have been shown to impair tumour progression and suppress metastatic potential by inhibiting and disrupting mitochondrial respiration [49].In addition, GOinduced ROS generation can stimulate mitochondrial dysfunction and subsequent apoptotic signalling pathways, which may explain the decrease in cisplatin IC 50 values of A2780 DDP cells following GO PEI treatment [47].Through a series of immunohistochemistry analyses among various treatment groups to monitor the change in protein abundance of 7 UPR mt signature genes including FOXO3, HSP60, LC3B, NRF1, SIRT3, SOD1, and SOD2 according to previous literature [48].We have concluded that both GO PEI -mPEG mediated in vivo SOD1 knockdown evoke UPR mt activation (detailed data not shown).Further, we observed the baseline upregulation of UPR mt markers upon cisplatin treatment.To the best of our knowledge, this is the first report of cisplatin-induced activation of all three arms of the UPR mt in ovarian cancer [49].While further exploration of these pathways is beyond the scope of this study, based on the striking differences in the activation of UPR mt pathways between baseline and upon cisplatin treated cells, it would be intriguing to explore further the potential involvement of these pathways in intrinsic or acquired cisplatin resistance in other cell lines as well.Furthermore, as graphenebased nanomaterials tend to accumulate and intercalate in the mitochondrial outer membrane causing mitotoxicity, thus it would be reasonable to examine whether the activation of the UPR mt could be used as a reporter in graphene biosafety and biocompatibility assays [50].

CONCLUSION
Despite the observed toxicities mediated by the respective nanoparticles, the in vivo results showed that SOD1 knockdown has a partial chemo-sensitising effect in cisplatin resistance relative to the siSCR control group.This observed effect is per our previous in vitro results.We concluded that while these promising results exhibited the potential role of SOD1 as a therapeutic target for the reversal of cisplatin resistance, yet the full therapeutic potential of SOD1 knockdown has not been fully demonstrated in this study due to the SOD1 mRNA induction properties of the nanoparticles, which counteracted the effect of the siSOD1 dosing primarily arising from the toxicity mediated presumably by the cationic polymer component.Overall, the therapeutic applicability of SOD1 knockdown in reversing cisplatin resistance will be further investigated using other drug delivery platforms such as lipid nanoparticles (LNPs) and in other cell-and patient-derived xenograft models.

Fig. 1
Fig.1Overview of nanoparticle synthesis and morphological characterization.A Schematic representation of GO PEI -mPEG preparation.mPEG was directly conjugated to PEI instead of graphene, which created a tunable system to control the ζ-potential of the nanoparticle.Linear methyl polyethylene glycol (mPEG) was conjugated to GO PEI through EDC chemistry, and siRNA molecules were allowed to bind GO PEI -mPEG electrostatically.B, C High-resolution TEM images of the surface morphological characterization of GO at different scales (5 nm, 200 nm, and 50 nm, respectively) showing a well-dispersed homogenous nanoparticle solution with varying sizes and dimensions.D Single GO PEI nanoparticle with rough and wrinkle surface characteristics.AFM analysis of GO and GO PEI surface morphology (E, F), lateral thickness (G, H), and topographical 3-D images (I, J) respectively.AFM analysis showed an even distribution of GO flakes with lateral dimensions in the 0.05-0.4μm range and thickness of 1-2 layers of GO and GO PEI .

Fig. 3
Fig.3Nanoparticle behaviour in biological solutions.A Colloidal stability of GO, GO PEI , and GO PEI -mPEG were assessed in DIW, 0.9% NaCl, and RPMI + 10% FBS, 1 h after treatment.Nanoparticle aggregation was observed in all three biological solutions containing nonfunctionalized GO.B, C Gel-electrophoresis-based serum stability was carried out in GO PEI1x , GO PEI2x , and GO PEI3x .The susceptibility of siRNA to RNase degradation in serum was assessed for 48 h.Naked siSCR was used as a control.D Kinetic study of nanocarrier-siRNA serum stability was measured using mouse serum in siSCR FAM conjugated GO PEI , GO PEI -mPEG, and Lipo 2000 relative to control naked siSCR FAM .The fluorescence of siSCR FAM was measured over a time of up to 8 h.E Kinetic study of nanoparticle behaviour was conducted in vitro using siSCR FAM complexed GO, GO PEI and GO PEI -mPEG.The degree of quenching induced by GO PEI upon binding siSCR FAM was measured by a Varioskan fluorescent plate reader.A 10 h time-course study of the live cell uptake of the nanoparticles was measured by recording their fluorescent emission.Samples containing only cells and RPMI media were used as negative controls.F Confocal laser scanning images showing the cellular uptake of GO PEI -siSCR FAM .Merged fluorescence images of cells treated with GO PEI -siSCR FAM with a magnified image of a single A2780 cell showing successful transfection of GO PEI as aggregations were compared with Lipo 2000 as the positive control, and naked siSCR FAM was used as a negative control.G Quantitative assessment of nanoparticle: siRNA cellular uptake GO PEI -siRNA transfection efficiency was analyzed using flow cytometry.Two siSCR FAM concentrations of 60 nM and 90 nM complexed with Lipo 2000 served as positive control.H Image showing intracellular graphene accumulation in cell pellets following three washing steps and centrifugation prior to protein isolation.I, J Time-course study showing SOD1 mRNA and protein levels following SOD1 knockdown.K, L The effect of concentration dependent SOD1 knockdown on protein levels was determined by Western blot and mRNA expression levels were measured using RT-qPCR.All values are expressed as mean ± SD. ns-not significant, *p < 0.05, **p < 0.01, and ****p < 0.0001 as analyzed using two-tailed unpaired t-test.

Fig. 4
Fig. 4 In vitro nanoparticle toxicity.A Schematic diagram of the in vitro induction of acquired platinum resistance.B Baseline cisplatin sensitivity of A2780 and A2780 DDP cell lines.C The IC 50 of A2780 and A2780 DDP were 4.71 ± 0.26 and 13.95 ± 1.18 µg/ml, respectively.D Bar graph showing normalized baseline SOD1 mRNA levels (n = 6).E Western blot showing the SOD1 protein levels in A2780 and A2780 DDP cell lines in vivo in Q1 and Q3.F Bar graph showing normalized baseline SOD1 mRNA (n = 3).G Schematic illustration of xenograft tumour sample preparation for immunoblotting and qRT-PCR.H Western blot showing the SOD1 protein levels in A2780 and A2780 DDP cell-derived xenograft tumour tissue samples.I Cisplatin treatment of A2780 DDP cells (n = 6) induced SOD1 mRNA overexpression in a concentration-dependent manner measured by RT-qPCR.J Time-course evaluation SOD1 mRNA induction by GO PEI treatment in A2780 DDP cells (n = 6) measured by qRT-PCR.K Cytotoxicity of GO, PEI, GO PEI and GO PEI -mPEG were determined at the following concentrations: 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 25, 30, 35 and 40 µg/mL, respectively.L MTT assay showing relative cell viability after A2780 cells were treated with 9 µg/mL of GO PEI for 48 h followed by cisplatin treatment at 14 different concentrations (2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26 and 28 µg/mL respectively).IC 50 values were compared with that of A2780 cells (n = 6).M Volcano plot showing the transcriptional activation of the mitochondrial unfolded protein response by 15 µM cisplatin in A2780 cell line.N, O GO PEI and GO PEI -mPEG treatment-induced differential in vitro activation of the UPR mt in (N) A2780 and (O) A2780 DDP cell lines.P Schematic diagram illustrating UPR mt activation by cisplatin, graphene and cationic polymers leading to mitochondrial dysfunction and subsequent mito-nuclear signalling.

Fig. 5
Fig. 5 Reversal of cisplatin resistance in vitro.A MTT assays showing reversal of cisplatin resistance by GO PEI -siSOD1.Relative cell viability of A2780 DDP cells treated with 65 and 130 nM of GO PEI -siSOD1 at a w/w ratio of 0.2:1 and subsequently with cisplatin (2-28 µg/mL).The graph also shows the relative cell viability of A2780 and A2780 DDP cells relative to cisplatin treatment only as a reference (n = 18).B Change of cisplatin IC 50 values normalized to that of A2780 DDP following treatment.C Population doubling curves of the A2780 cell line (0-96 h) under various treatment conditions.D Population doubling curves of the A2780 DDP cell line (0-96 h) under various treatment conditions.E Therapeutic efficacy and antiproliferation effect of GO PEI -mPEG-siSOD1 in A2780 and A2780 DDP cells evaluated in ultra-low attachment plates, clonogenic and MTT assays.All values have been normalized to their respective untreated cell line control, and presented as mean ± SD (n = 3).F, G Representative images of ultra-low attachment and clonogenic assay cell culture plates.H Evaluation of total ROS, superoxide and mitochondrial membrane potential with plate reader under siSCR control and siSOD1 knockdown conditions confirmed with Western blot.

Fig. 6
Fig. 6 Evaluation of in vivo biodistribution, toxicity, and therapeutic efficiency of graphene-based nanoparticles.A In vivo fluorescence imaging and biodistribution siSCR and siSCR Cy3 complexed nanoparticle formulations administered via tail vein injection of female BALB/c nude mice bearing A2780 DDP subcutaneous xenografts.Whole body fluorescence images were taken at different time points (0, 1, 4, 8 h) following tail vein injection.B-J In vivo blood biochemistry analysis of (B) ALT (C) AST (D) TP (E) ALB (F) ALP (G) BUN (H) CREA (I) UA (J) LDH of mouse samples following systemic administration of nanoparticles.K In vitro hemolysis activity of naked siSCR, GO, GO PEI and GO PEI -mPEG nanoparticles.Hemolysis rate was measured following incubation of nanoparticle polyplexes with purified red blood cells in PBS.L H&E staining of the heart, kidney, liver, spleen and lungs harvested from mice recieved intravenous naked siSCR, and siSCR Cy3 complexed GO, GO PEI and GO PEI -mPEG respectively.Images of collected organs from each group was taken at 8 h after tail vein injection.M In vivo tumour therapeutic efficiency of GO PEI -mPEG-siSOD1.During a 14-day long therapeutic window, mice received 6 doses of respective siSCR or siSOD1based treatments with or without cisplatin treatment.N H&E staining of A2780 and A2780 DDP cell-derived tumour cross-sections in their respective treatment groups as in (Fig.6M) harvested from mice three days after the last injection.