Non-disruptive uptake of anionic and cationic gold nanoparticles in neutral zwitterionic membranes

The potential toxicity of ligand-protected nanoparticles (NPs) on biological targets is crucial for their clinical translation. A number of studies are aimed at investigating the molecular mechanisms shaping the interactions between synthetic NPs and neutral plasma membranes. The role played by the NP surface charge is still widely debated. We compare, via liposome leakage assays, the perturbation induced by the penetration of sub-6 nm anionic and cationic Au NPs into model neutral lipid membranes composed of the zwitterionic 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC). Our charged Au NPs are functionalized by a mixture of the apolar 1-octanethiol and a ω-charged thiol which is either the anionic 11-mercapto-1-undecanesulfonate or the cationic (11-mercaptoundecyl)-N,N,N-trimethylammonium. In both cases, the NP uptake in the bilayer is confirmed by quartz crystal microbalance investigations. Our leakage assays show that both negatively and positively charged Au NPs do not induce significant membrane damage on POPC liposomes when penetrating into the bilayer. By means of molecular dynamics simulations, we show that the energy barrier for membrane penetration is the same for both NPs. These results suggest that the sign of the NP surface charge, per se, does not imply different physicochemical mechanisms of interaction with zwitterionic lipid membranes.

. Characterization results of NP-, NP+ and POPC liposomes. All the uncertainties on mean values were obtained using Student's statistics, assuming a confidence level of 95%. Standard deviation (s) is reported only for NP core size distributions (see below for more information). For both hydrodynamic size and Z-potential analyses, at least 9 measurements were performed; hydrodynamic diameters were evaluated from the average diameters obtained from number distributions. In the case of ligand shell composition, three NP spectra were acquired.
For liposome characterization, all analyses were performed in the 100 mM NaCl, 2 mM histidine, 2 mM TES, 0.1 mM EDTA buffer (pH 7.4) used for experiments.
For NP characterization, NP powders were dispersed in water and, when specified, diluted in buffer.
Only for NP+, aqueous dispersions were further centrifuged before DLS, Z-potential, and NMR characterizations (see the NP sample preparation section). For DLS size analysis in water, measurements were acquired at 2·10 -2 mg/mL NP concentration (ca. ten-fold higher than those corresponding to Rm=0.03 and 0.05) since lower values did not provide good quality reports. In the case of DLS and Z-potential measurements, results refer to NP dispersions freshly filtered immediately before experiments (20 nm pore size, Anotop 10, Whatman). All NP dispersions were mildly sonicated before analyses.

BF-TEM analysis: NP core size distribution
Few drops of each NP water dispersion were laid onto an ultrathin carbon-coated Cu grid. In Supplementary Figure S1, representative BF-TEM images of NP-and NP+ are presented, together with histograms of the experimental size distribution. Statistical analysis was calculated by assuming spherical morphology and by counting at least 300 particles with ImageJ software.
Supplementary Figure S1. BF-TEM images of NP-and NP+. Experimental size distributions are shown aside; mean diameters and standard deviations (s) are reported above each histogram. NP ligand-coating composition analysis. 1 H NMR was also used to assess the Au NP ligand shell composition (i.e. the MUS % and the TMA % in NP-and NP+, respectively) after decomposition of the gold core 4 . In the case of NP-, the procedure for the determination of MUS% and the NMR spectrum are reported in Canepa et al. 1 . The same procedure described for NP-was used to characterize TMA% in NP+. Only minor modifications were added to the sample preparation. NP+ (~ 5 mg) were dispersed in water (2 mL) and mildly centrifuged (10 min at 11900 X g), only the supernatant was kept. This procedure was repeated, and the collected aqueous suspensions were evaporated to dryness. The residue was taken up in DMSO-d6 (600 µL), I2 (1.6 mg) was added and the mixture was sonicated for 30 min to etch the gold core. The orange solution was transferred into 5 mm NMR tube and analyzed as for NP-. NP+ spectrum is reported in Supplementary Figure S2.
NMR results are summarized in Supplementary Table S1.
Supplementary Figure S2. Expansion of 1 H NMR spectrum of NP+ after centrifugation and iodine etching in DMSO-d6.

Content release assays on POPC liposomes
To understand whether the origin of minor fluctuations shown in Figure 2d of the main text could be due to the addition of charged NPs, we performed control leakage experiments adding similar volumes of water aliquots (without NPs). From Supplementary Figure S4a, it is evident that the effect of the water alone is comparable with those of both NP-and NP+.
Additional leakage assays were performed on larger POPC vesicles prepared by extrusion, as reported in Supplementary Figure S4b. These experiments show that the increase in vesicle diameter from 23 to 105 nm does not change the effect of the NP-membrane interaction (see Supplementary Table S1 for results of liposome size characterization).

Computer simulations
We monitored the area per lipid of a 512 POPC lipid bilayer, in presence of 175 mM calcein in the water phase (and no NP) and found that both the area per lipid and the membrane compressibility are the same as without calcein.