Two monomers were made for the synthesis of the ps-PAAQ polymers. The first monomer, cystamine bis(acrylamide) (CBA), was synthesized as described by Lin et al. (Lin et al., 2006) The second monomer, N1-(7-chloroquinolin-4-yl)-hexane-1,6-diamine (Q6), was synthesized analogously to the described synthesis by Natarajan et al. (2008) Other chemicals were purchased and used without further purification from Sigma-Aldrich or Avantor.

The ps-PAAQ (or p (CBA-ABOL-Q)/PEI) were synthesized by Michael-type polymerization of primary amines with bis-acrylamides as described by Lin et al. (Lin et al., 2006) In brief, a small round-bottom flask was charged with 4-aminobutanol (ABOL) (0.60 g, 6.73 mmol), bis(acrylamide) (CBA) (2.60 g, 10 mmol), and N1-(7-chloroquinolin-4-yl)-hexane-1,6-diamine (Q6) (0.60 g, 2.16 mmol). MeOH (10 mL) was added, followed by a solution of CaCl₂ (0.44 g, 4 mmol) in water (2 mL). The resulting suspension was heated to 50°C and was allowed to stir for an additional 48 h resulting in a nearly translucent solution (PAAQ polymer). A solution of branched PEI800 (Sigma-Aldrich Mn 600, 100 mg/mL in water, 680 μL) was then added and the reaction mixture was allowed to stir for another 72 h with heating. The reaction was subsequently terminated by acidification to pH 4 using hydrochloric acid and the resulting solution was transferred directly into a dialysis membrane (Spectrapor 6, 10 kD MWCO) for purification, where the polymer was dialyzed extensively against water. Then, the material was filtered through a 0.45 µm filter (cellulose acetate) and ultimately lyophilized to yield the branched ps-PAAQ polymer as a (hygroscopic) white solid.

¹H NMR (400 MHz, DMSO-d6) 1.25–1.75 (multiple peaks), 2.4–3.05 (multiple peaks), 3.08–3.45 (multiple peaks) 5.53–5.65 (m, -C=CHH¹), 6.01–6.16 (m, -C=CH¹H), 6.77–6.88 (bs), 7.64–7.76 (bs), 8.00–8.15 (bs), 8.33–8.61 (large bs), 8.66–8.81 (bs), 9.45–9.68 (bs) ppm.

The ratio of integrals at the 1–2 ppm range to the integral at 6.8 ppm were determined to be 20:1 corresponding to a feedstock of approximately 1–3 parts Q:ABOL.

The ps-PAAQ without disulfide bonds in the polymer backbone (or p (HMBA-ABOL-Q)/PEI) was synthesized as described above, except by replacing the CBA by hexamethylenediamine bisacrylamide (HMBA) (2.24 g, 10 mmol).

The synthesis of the PAAQ polymer was performed as above but in addition, after the 72 h of reaction time with PEI800, 11-azido-3,6,9-trioaxaundecan-1-amine (0.33 g, 1.5 mmol) was added directly to the reaction mixture and was subsequently allowed to stir for another 24 h to functionalize remaining acrylamide moieties with an azide functionality. The reaction mixture was then acidified and purified as described above. A small multiplet peak appears on ¹H-NMR at 3.72 ppm while the acrylamide signal is concomitantly decreased compared with non-functionalized polymer.

Fifty milligrams of azide-functionalized ps-PAAQ was dissolved in a mixture of water/MeOH (1.1 mL, 10:1 v/v) and sulfo-Cy5 DBCO (Lumiprobe GmbH, 50 µL of a 20 mg/mL solution in DMSO) was added and allowed to stir 1.5 h. The reaction mixture was then transferred to dialysis tubing (Spectrapor 6, 10 kD MWCO) and extensively dialyzed against water. After filtration through a 0.45 µm filter (cellulose acetate) followed by freeze drying, a blue solid (20 mg) was obtained that contained no free Cy5 when assayed by TLC (30% MeOH in CH₂Cl_2, R_f = 0.3). 20 mg of the polymer corresponded to 0.6 mg sulfo-Cy5 when analyzed with absorbance spectroscopy (absorbance at 650 nm, ε = 2.5 · 10⁵ mol^-1 L^-1) (Mujumdar et al., 1993).

The ps-PAAQ polymers were mixed with EGFP mRNA (Cleanup, TriLink Biotechnologies) in 10 mM Histidine 10% Trehalose buffer (pH 6.5) to obtain nanoparticles. The mRNA concentration was kept constant in all formulations (60 μg/mL), with increasing concentrations of ps-PAAQ polymers (0.75 mg/mL - 3 mg/mL) resulting in a target ps-PAAQ:mRNA loading ratio ranging from 12.5 to 50 w/w (respective N/P ratio of 10:1 to 40:1). For the formulation of Cy5-labeled nanoparticles, a ratio of 1:9 w/w Cy5-labeled ps-PAAQ to non-labeled ps-PAAQ was used.

To obtain PGA-PEG-coated nanoparticles, the coating material (mPEG_5k-b-PLE₅₀, Alamanda Polymers) was added to the mRNA solution in the first step, which was then added to the polymers in the same mixing step, using a 1:1 coating to ps-PAAQ w/w ratio.

The resulting nanoparticle size, zeta potential and particle concentration were measured using Multi-Angle Dynamic Light Scattering (MADLS) in the Zetasizer Ultra (Malvern), with three different angles: 175°, 90° and 13°. Samples were diluted ten-fold in 10 mM Histidine 10% Trehalose buffer (pH 6.5) and loaded in a low-volume quartz cuvette (ZEN2112, Malvern). Results were analyzed in the ZS Explorer software (version 3.0, Malvern). The measurements for Cy5-labeled nanoparticles and the experiment comparing the size of nanoparticles in formulation buffer versus culture medium were performed in the Zetasizer Nano ZS90 (Malvern), with a 90-degree scattering optics. Results were analyzed in the Zetasizer software (version 7.13, Malvern).

Immortalized C28/I2 human chondrocytes were cultured in a growth medium composed of Dulbecco’s modified Eagle’s medium (DMEM GlutaMAX™, high glucose, pyruvate; Gibco), supplemented with 10% v/v fetal bovine serum (FBS; Biowest), 100 units/mL penicillin and 100 μg/mL streptomycin (P/S; Gibco) at 37°C under a humidified 5% CO₂ atmosphere. Medium changes were performed every 3 days and cells were passaged at 70%–90% confluency at a seeding density of 6,000 cells/cm² in 175 cm² T-flasks.

For the transient transfection experiments, C28/I2 cells were seeded at 20,000 cells/cm² in 48 well-cell culture plates (Sarstedt) and allowed to attach overnight, using growth medium.

The ps-PAAQ NPs nanoparticles loaded with EGFP mRNA were initially suspended in serum-free DMEM without antibiotics, supplemented with 20 mM sterile HEPES buffer (Gibco). After gently mixing, the NP suspensions were dropwise added to the cells to a total volume of 0.2 mL/well. Following incubation at 37°C for 4 h, the serum-free medium was replaced by the full growth medium containing 10% v/v FBS and antibiotics. Finally, 24 h after transfection, fluorescent images were taken, and the percentage of transfected cells was calculated as described in the next section. For confocal imaging, cells were cultured in 35 mm CELLview™ glass-bottom dishes (Greiner Bio-One) and seeded at 20,000 cells/cm² under the same culture conditions with a total volume of 0.2 mL/well. Lipofectamine™ MessengerMAX™ Transfection Reagent (Invitrogen) was used as a positive control for mRNA transfection, following the manufacturer’s protocol. Live cell imaging was performed in an environmentally controlled chamber (37°C and 5% CO₂) in all experiments.

Transfected efficiency was evaluated by fluorescence microscopy after 24 h of incubation (THUNDER Imager; Leica). Before imaging, the nuclei from all cells were stained with 5 µM Hoechst 33342 (Invitrogen) for 5 min at 37°C, followed by three washing steps with DMEM. Subsequently, staining with 0.25 µM SYTOX™ Orange Nucleic Acid Stain (Invitrogen) was carried out for 20 min at 37°C, as a marker of dead cells (Truernit and Haseloff, 2008). A total of three fields were imaged per well and each experiment was done in triplicate (3 wells/condition) from 3 independent cultures. Between 8,000 and 12,000 cells were quantified per condition and independent culture. The number of positive cells per channel was calculated by automatic cell counting on ImageJ (version 1.53; National Institutes of Health) as previously reported (Grishagin, 2015).

Software analysis was carried out by “batch processing” according to Macro functions from the Supplementary Material, in order to automatically count the number of positive cells per channel. Transfection efficiency was expressed as the percentage of GFP-positive cells per Hoechst-positive cells. Cell viability was calculated as the percentage of SYTOX-negative cells per Hoechst-positive cells, compared to untreated cells.

For verifying the effect of PEGylation on NP internalization efficiency, PEG-coated and uncoated Cy5-labeled ps-PAAQ NPs were incubated with C28/I2 cells in a 35 mm glass-bottom dish (CELLview™), during 3 h at 37°C. For this, we used a loading ratio of 25 w/w ps-PAAQ:mRNA and dosage of 480 ng mRNA per well. To visualize cell boundaries, the plasma membrane was stained with 2.5 μg/mL CellMask™ Orange (Invitrogen) for 5 min at 37°C, followed by three washing steps with DMEM. Confocal images were acquired immediately after washing, using a confocal laser microscope (SPX8; Leica) and 63x/1.4 oil-immersion objective. The Cy5-labeled NPs were visualized with 649 nm excitation wavelength, while CellMask-stained plasma membrane was imaged by using a 554 nm laser. The Z-stack method, changing the focal length from the bottom to the top of a single cell, was used to provide an orthogonal view of the cell thickness as an evidence of particle internalization. Image processing was conducted using ImageJ to estimate the number of internalized nanoparticles within the cell boundaries. After background subtraction, a single-cell area was manually selected using the selection tool to estimate the mean fluorescence intensity per μm².

In the uptake kinetics study, we used FACS to quantify the percentage of cells with internalized nanoparticles over time. In summary, cells were trypsinized with 100 μL of Trypsin-EDTA (0.25%) per well and incubated for 3 min at 37°C. Then, 400 μL of DMEM with 10% v/v FBS was added to each well and the cells were resuspended. Cells from three technical replicates were pooled into a single Eppendorf tube and centrifuged for 5 min at 300xg. The medium was aspirated, and the pellet was resuspended in 2% paraformaldehyde (PFA) for fixation, during 10 min at room temperature. The cells were again centrifuged for 5 min at 300xg, and the pellets were resuspended in 100 µL fresh DPBS/0.5% BSA FACS buffer. The samples were transferred to a 96-well plate and kept on ice until measurement was performed in a MACSQuant^® flow cytometer (Miltenyi Biotec). Forward, side scatter, and laser voltage were adjusted using untreated cells. Recording conditions were set to collect 10,000 live, single cell events per sample.

For evaluating endosomal leakiness, C28/I2 cells were incubated for 15 min at 37°C with PEG-coated ps-PAAQ NPs or 100 µM chloroquine (Sigma-Aldrich). After this period, cells were washed once with DMEM to remove the residual extracellular particles and then stained with calcein (Invitrogen) at a self-quenching concentration of 3 mM for 15 min, as previously reported (Vermeulen et al., 2018). The cells were washed three times with DMEM and incubated for a period of 3 h at 37°C before imaging. Confocal images were acquired using a confocal laser microscope (SPX8; Leica) and 63x/1.4 oil-immersion objective.

In order to track the intracellular fate of the nanoparticles over time, PEG-coated ps-PAAQ NPs were co-loaded with EGFP mRNA and silencer Cy3-labeled negative control siRNA (Invitrogen) at a 9:1 w/w ratio. These fluorescently labeled NPs were incubated with C28/I2 cells seeded on 35 mm glass-bottom dishes (CELLview™) for imaging at timepoints 3, 24 and 48 h. For this, we used a loading ratio of 25 w/w ps-PAAQ:mRNA and dosage of 480 ng mRNA per well. Three hours after incubation, the dishes were washed once with DMEM to remove the residual extracellular particles. Fresh growth medium supplemented with 10% v/v FBS and antibiotics was added to cells for optimal growth, then the dishes were placed back in the incubator for measurements after 24 and 48 h. For the imaging of acidic compartments, endo/lysosomes were stained with 50 nM Lysotracker™ Deep Red (Invitrogen) in DMEM for 15 min, followed by three washing steps with DMEM. LysoTracker™ staining was performed before imaging at each timepoint. Confocal images were acquired immediately after washing, using a confocal laser microscope (SPX8; Leica) and 63x/1.4 oil-immersion objective. The LysoTracker-stained endo/lysosomes were visualized with 649 nm excitation wavelength, while Cy3 siRNA-loaded NPs were imaged by using a 554 nm laser.

Images were processed using ImageJ and the frequency of endosomal escape events over time was calculated by the “Colocalization Finder” plugin in ImageJ. After single cell area selection, the Pearson correlation coefficient (PCC) values were estimated from the colocalization of the fluorescence intensities from both channels (Dunn et al., 2011). PCC readouts ranged from 0 to 1. Values close to zero indicate that the fluorescence intensities of the two channels are uncorrelated, whereas values close to 1 indicate that the two fluorescence intensities are directly and linearly related. Threshold PCC values, related images and descriptions are reported in Supplementary Figure S1.

The in vitro release of mRNA from NPs was verified by agarose gel electrophoresis, using 2% agarose gel and SYBR^® Safe DNA Gel Stain (Invitrogen) for visualization of the EGFP mRNA. In this assay, the following samples were loaded in the gel: free mRNA, 10 µL of NPs loaded with mRNA, and 10 µL loaded NPs treated for 5 min at 60°C with both 2 M 1,4-Dithiothreitol (DTT; Sigma-Aldrich) and 5 mg/mL heparin (Sigma-Aldrich). Gel electrophoresis was performed for 30 min at 100 V and visualized using a ChemiDoc Imaging System (Bio-Rad). If mRNA is not released from nanoparticles following treatment, it does not migrate in the gel.

For simultaneous monitoring of nanoparticle disassembly and cargo release over time, Cy5-labeled PEG-coated NPs were co-loaded with EGFP mRNA and AZDye568-EGFP mRNA (RiboPro) at a 9:1 w/w ratio. These nanoparticles were incubated with C28/I2 cells in a 35 mm glass-bottom dish (CELLview™) for imaging at timepoints 3, 8 and 24 h. For this, we used a loading ratio of 25 w/w ps-PAAQ:mRNA and dosage of 480 ng mRNA per well. Three hours after incubation, the dishes were washed once with DMEM to remove the residual extracellular particles. Fresh growth medium supplemented with 10% v/v FBS and antibiotics was added to cells for optimal growth, then the dishes were directly imaged or placed back in the incubator for measurements after 8 and 24 h (without additional washing steps). Images were acquired using a confocal laser microscope (SPX8; Leica) and 63x/1.4 oil-immersion objective. The Cy5-labeled NPs were visualized with 649 nm excitation wavelength, while the AZDye568-EGFP mRNA was imaged by using a 577 nm laser. The GFP protein fluorescence was detected with a 488 nm laser line.

The C28/I2 cells were treated with specific chemical enhancers and inhibitors to investigate the effects on transfection efficiency of nanoparticles. To determine the effect of modulating the intracellular reducing environment on GFP expression, cells were incubated either in the presence of 20–2000 µM DL-buthionine sulfoximine (BSO; Sigma-Aldrich) for 24 h prior to transfection, or 1–2 mM glutathione–monoethyl ester (GSH–MEE; Sigma-Aldrich) for 3 h after medium change, 4 h after transfection. Cell culture and transfection were carried out in DMEM without pyruvate (high glucose, Gibco), as previously recommended in studies involving oxidative stress (Babich et al., 2009).

The optimal formulation of ps-PAAQ NPs for transfection in C28/I2 cells was found by the Design of Experiments (DoE) method (Eriksson et al., 2008). A D-optimal design was chosen, due to the precise estimation of factor effects and the small number of experimental trials compared to standard factorial design. The key variables that could influence transfection efficiency were defined as the polymer-to-mRNA ratio (in w/w) and the dosage of mRNA per well (in ng). The DoE was developed by using the MODDE software (version 13, Sartorius Stedim Data Analytics AB), comprising of 10 runs and 3 center points, therefore 13 experiments.

Statistical analyses were performed using GraphPad Prism (version 9.0; GraphPad^® Software). For comparing differences between two groups, parametric data were evaluated through unpaired t-test, while non-parametric data were evaluated through Mann-Whitney test. For comparing differences among more than two groups, we used one-way analysis of variance (ANOVA) followed by the Tukey’s post hoc multiple comparisons test. Alternatively, Dunnett’s test was used when comparing the mean of experimental groups with a single control group, and Bonferroni’s test was applied when there was a set of planned comparisons beforehand. Values of p ≤ 0.05 were considered to be statistically significant. All data are shown as mean ± standard deviation.

The ps-PAAQ polymers were co-formulated with EGFP mRNA and PGA_7.5k-PEG_5k to obtain nanoparticles (designated “PEG-coated NPs”), at several loading ratios ranging from 12.5 to 50 w/w (ps-PAAQ:mRNA, N/P ratio 10:1 to 40:1). For the intracellular studies in the following sections, we used the 25 w/w loading ratio (N/P ratio 20:1) based on previous results with different cell lines (unpublished data). The amount of loaded mRNA was kept constant at 60 μg/mL for all formulations. MADLS measurements confirmed the formation of nanoparticles, with an average size around 60 nm and a monodisperse distribution for PEG-coated NPs (Table 1 and Supplementary Figure S2). However, a larger particle size and multiple resolvable peaks were observed for uncoated NPs (Table 1 and Supplementary Figure S2). It was also possible to observe a proportional increase in particle concentration with increasing polymer concentrations in the coated formulation (from 12.5 to 50 w/w ratio) (Table 1). Zeta potential showed near-neutral values for coated NPs, as an effect of the hydrophilic PGA_7.5k-PEG_5k addition, while uncoated NPs had a strongly positive zeta potential of +32.7 mV. The characterization of Cy5-labeled NPs used in this study is shown in Supplementary Table S1.

In the following, we assessed the mean diameter size of PEG-coated and uncoated NPs in presence of culture medium (serum-free) used in transfections. The formulations (25 w/w loading ratio) were analyzed by DLS after mixing with 10 mM Histidine 10% Trehalose (formulation buffer) or DMEM/2% HEPES medium. In both conditions, a 1:1 v/v dilution ratio was used. Figure 1 shows the results following incubation for 30 min at room temperature.

The incubation in 10 mM Histidine 10% Trehalose buffer did not impact the original diameter of nanoparticles, compared to the measurements from Table 1: PEG-coated NPs showed an average size of 60.6 ± 1.5 nm (PDI 0.237 ± 0.017), and uncoated ps-PAAQ NPs an average size of 146.5 ± 2.4 nm (PDI 0.186 ± 0.038). While incubation with DMEM did not substantially affect the size of coated NPs (69.4 ± 0.7 nm; PDI 0.149 ± 0.013), it did cause a substantial increase in particle size and polydispersity for uncoated NPs (349.5 ± 20.9 nm; PDI 0.331 ± 0.052). This corresponds to over a two-fold increase in size compared with uncoated NPs in 10 mM Histidine 10% Trehalose buffer. The respective particle size distributions are shown in Supplementary Figure S3. Besides, the cryogenic electron microscopy (Cryo-EM) of the PEG-coated NPs in buffer was obtained for morphology and size confirmation (Supplementary Figure S4).

To evaluate the internalization efficiency of nanoparticles, we followed the cellular uptake of neutral PEG-coated NPs into C28/I2 human chondrocytes, compared to positively charged uncoated NPs. Both nanoparticle types were labeled with water-soluble sulfo-Cy5 dye to track their subcellular localization in the cells. With confocal microscopy after 3 h of incubation, these fluorescent NPs appeared as punctate spots that agglomerate preferentially in the perinuclear region (shown in red in Figures 2A, B). The orthogonal sectioning and reconstruction of the z-stacks shows the clear proof that the particles were internalized (Supplementary Figure S5). While a large amount of both uncoated and PEG-coated NPs were detected virtually inside all cells, the punctate spots corresponding to uncoated ps-PAAQ NPs appear larger and more intense compared with PEG-coated nanoparticles (Figures 2A, B). For quantifying the uptake efficiency of nanoparticles, the cytosolic fluorescence intensity was measured using the ImageJ software. Indeed, uncoated ps-PAAQ NPs showed a two-fold increase in cellular uptake compared with PEG-coated NPs (Figure 2C). In contrast, transfection efficiency–as measured by percentage of GFP-positive cells–with coated NPs (64% ± 12%) was significantly higher than results obtained with uncoated NPs (42% ± 8%) (Figure 2D). Cell viability was ≥95% for transfections with both nanoparticle types (Figure 2E).

We also evaluated the uptake kinetics of uncoated versus PEG-coated ps-PAAQ NPs over time. The percentage of C28/I2 cells with internalized nanoparticles was measured by FACS per hour, in the first 4 h after transfection. The PEG-coated NPs showed a slower uptake kinetics compared with uncoated NPs, especially in earlier timepoints (Supplementary Figure S6). This finding corroborates the uptake efficiency results obtained by confocal microscopy above.

The localization studies in the next sections were performed with PEG-coated NPs (same loading ratio and payload quantity), given their improved stability and lower polydispersity in cell culture medium, as well as superior GFP protein expression in C28/I2 chondrocytes compared with uncoated ps-PAAQ NPs.

PEG-coated ps-PAAQ nanocarriers were further explored regarding their ability to induce endosomal rupture. To evaluate the role of endosomal membrane leakiness on the escape of nanoparticles, calcein was incorporated into endosomes as a marker to study endosome integrity. Because it was used in a self-quenching concentration (3 mM), a subtle leak in the endosomal membrane could be easily observed by the change from a punctate fluorescent pattern (endosomes) to a diffuse (dequenched) cytoplasmic fluorescence of higher intensity. Chloroquine is a well-described endosomal escape enhancer (Hajimolaali et al., 2021), which functions as positive control for endosomal escape. A chloroquine derivative is also incorporated into the ps-PAAQ polymer, where it was expected to similarly enhance endosomal escape. Indeed, calcein release can be clearly observed in both cells that have taken up nanoparticles as well as those that were treated with chloroquine (Figure 3A). In addition, Figure 3B shows the intensity of cytosolic calcein fluorescence after incubation with 60, 30 and 12 μg/mL of ps-PAAQ NPs, clearly indicating that calcein release is a concentration-dependent effect.

In order to evaluate endosomal escape over time, PEG-coated NPs were co-loaded with a mixture of EGFP mRNA and Cy3-labeled siRNA, and subsequently incubated with C28/I2 cells for 3, 24 and 48 h. Before imaging at each timepoint, endo/lysosomes were stained using LysoTracker™ Deep Red (Figure 4G). After 3 h of incubation, yellow punctuate spots indicating colocalization of the two labels (Figures 4A, C) clearly show endosomal entrapment of NPs at this timepoint. The average PCC value was maximum 3 h post-transfection (0.30 ± 0.09) in this experiment (Figure 4F). The fraction of particles colocalizing with the acidic compartments decreased after 24 h (Figure 4D), when more ps-PAAQ NPs were found to be no longer entrapped inside the endo/lysosomal vesicles (PCC 0.20 ± 0.09). In the first two timepoints (3 and 24 h), a higher heterogeneity in the intracellular localization was observed between different cells, explaining the large standard deviation (SD) of the mean PCC values (Figure 4F). After 48 h, nearly all ps-PAAQ NPs were observed to be distributed in the cytosol (Figures 4B, E), suggesting an effective escape of ps-PAAQ NPs from the acidic compartments (PCC 0.14 ± 0.05). This represents a 50% decrease in the endosomal entrapment compared with the initial 3 h timepoint. These results constitute further evidence that ps-PAAQ nanoparticles induce the rupture of the endo/lysosomes and escape to the cytosol over time.

The in vitro release of mRNA from PEG-coated ps-PAAQ NPs was investigated by gel electrophoresis, using heparin as a competing polyanion (Oupickí et al., 2001) in combination with the reducing agent dithiothreitol (DTT) to facilitate mRNA release through disulfide exchange-mediated ps-PAAQ fragmentation (van der Aa et al., 2011). Three different loading ratios were tested to compare the effect of increasing NP concentrations on mRNA release, ranging from 12.5 to 50 (w/w) ps-PAAQ:mRNA. After treatment of ps-PAAQ NPs with 2 M DTT and 5 mg/mL heparin before electrophoresis, the mRNA release was monitored on agarose gel. Gel electrophoresis showed strong mRNA binding for all polymer-to-mRNA ratios, since untreated nanoparticles prevented mRNA migration in the gel (Figure 5 FIGURE 5

Migration of EGFP mRNA from coated ps-PAAQ NPs in agarose gel electrophoresis. Three different loading ratios (w/w ps-PAAQ:mRNA): 12.5 (lanes 1–2); 25 (lanes 3–4); and 50 (lanes 5–6). Free mRNA (60 μg/mL) was loaded on lanes 7–8 as a control. The second well of each group was treated with DTT and heparin to release mRNA by reduction and displacement, respectively.

To assess the effect of GSH depletion on GFP protein expression, we analyzed transfection efficiency after treating C28/I2 cells with an inhibitor of glutathione: buthionine sulfoximine (BSO). This chemical inhibits the activity of glutathione synthetase, an enzyme that catalyzes the last step in the synthesis of glutathione (Drew and Miners, 1984). Chondrocytes were incubated with 20–2000 µM BSO for 24 h prior to transfection, and no significant decrease in gene expression was observed with the lowest dose of 20 µM BSO (Figure 7A). However, higher doses of BSO resulted in a modest but significant decrease in the percentage of GFP-positive cells (49% ± 6% for 200 µM BSO; 47% ± 12% for 2000 µM BSO), compared with the control group not treated with BSO (70% ± 10%). Cell viability was >90% for all treatment groups (data not shown). The importance of the disulfide-mediated release was further confirmed using PEG-coated ps-PAAQ NPs where the disulfide bridges were replaced with an analogous carbon-based methylene linkage that is not susceptible to cleavage. For these carbon-based ps-PAAQ NPs, GFP protein expression was not observed (Figure 7A), even though NP internalization was clearly visible 24 h after transfection (Supplementary Figure S7). These nanoparticles showed comparable size as disulfide-containing coated ps-PAAQ NPs (56.2 ± 2.0 nm; PDI 0.122 ± 0.022). Therefore, these results demonstrate the superiority of bioreducible ps-PAAQ nanoparticles in mRNA transfection in C28/I2 cells.

To prove that the decrease in GFP expression was directly caused by BSO-mediated depletion of GSH, we additionally incubated chondrocytes with glutathione monoethyl ester (GSH-MEE), a cell-permeable derivative of GSH, after treating cells with 200 µM BSO (Figure 7B). The GFP expression levels were recovered after incubating BSO-treated chondrocytes with 1 or 2 mM GSH-MEE, compared to the control group not treated with BSO (64% ± 13%). For the treatment with 2 mM GSH-MEE, the number of GFP-positive cells was significantly higher (81% ± 8%) than the control group without BSO, which implies a greater NP biodegradation and potentially more mRNA release. Cell viability was >90% for all treatment groups (data not shown).

The optimization of the nanoparticle formulation is crucial to ensure high transfection efficiency while keeping toxicity levels low. In all experiments previously discussed, the same formulation was used, consisting of 25 w/w ps-PAAQ:mRNA ratio and dosage of 480 ng mRNA per well. Although the results were also satisfactory in general for this particular cell line (around 60%–65% of transfection efficiency), we wondered if these outcomes could be further improved by testing different combinations of loading ratios and dosages of nanomaterial in C28/I2 cells. Variables that can impact transfection efficiency were defined as the ps-PAAQ:mRNA ratio and the dosage of mRNA per well. Hereby we applied a D-optimal design to investigate the effect of these parameters on GFP expression by PEG-coated ps-PAAQ NPs, according to the Design of Experiments (DoE) from Table 2. The statistical analysis verified whether the selected variables and their interactions had a significant effect on transfection efficiency in C28/I2 cells. Transfection efficiency >65% was chosen as a critical quality attribute (CQA), considering our results presented so far here. The outcomes were used to establish a design space, an acceptable region within which the quality of the product can be assured (Hakemeyer et al., 2016).

As a result of the statistical analysis, Figure 8A shows the design space for a target transfection efficiency in an experimental area with low error risk. Each point from the design space surface represents a possible different loading ratio/dosage, having the transfection efficiency >65% as a CQA, with a certain risk level. The risk of getting predictions outside the specifications, expressed as the error risk, was estimated by using Monte Carlo simulations (Figure 8A, red area). As a result, a dose-dependent effect is observed with growing loading ratios and dosages. The optimal setpoint was found at 42.5 w/w ps-PAAQ:mRNA ratio (N/P ratio 34:1) and dosage of 640 ng mRNA per well (Figure 8A, green area). Importantly, no toxicity was observed 24 h after transfection, as cell viability compared to untreated cells was >95% for all tested loading ratios and dosages (data not shown).

The design space was further validated by three independent repetitions of the transfection experiment, comparing the optimal setpoint (42.5 w/w ps-PAAQ:mRNA ratio and dosage of 640 ng mRNA per well) with a randomly chosen suboptimal setpoint (25 w/w ps-PAAQ:mRNA ratio and dosage of 400 ng mRNA per well) (Figure 8B). As a result, all three average measurements from the optimal setpoint were above the 65% target (77% ± 4%; 79% ± 12%; 75% ± 8%), while all three measurements from the suboptimal setpoint were below the 65% target (62% ± 7%; 46% ± 13%; 44% ± 7%). As a positive control, mRNA delivery with the commercially available transfection reagent Lipofectamine™ MessengerMAX™ showed transfection efficiency of 70% ± 3% (Figures 8B, C).