Mechanistic understanding of in vivo protein corona formation on polymeric nanoparticles and impact on pharmacokinetics

In vitro incubation of nanomaterials with plasma offer insights on biological interactions, but cannot fully explain the in vivo fate of nanomaterials. Here, we use a library of polymer nanoparticles to show how physicochemical characteristics influence blood circulation and early distribution. For particles with different diameters, surface hydrophilicity appears to mediate early clearance. Densities above a critical value of approximately 20 poly(ethylene glycol) chains (MW 5 kDa) per 100 nm2 prolong circulation times, irrespective of size. In knockout mice, clearance mechanisms are identified for nanoparticles with low and high steric protection. Studies in animals deficient in the C3 protein showed that complement activation could not explain differences in the clearance of nanoparticles. In nanoparticles with low poly(ethylene glycol) coverage, adsorption of apolipoproteins can prolong circulation times. In parallel, the low-density-lipoprotein receptor plays a predominant role in the clearance of nanoparticles, irrespective of poly(ethylene glycol) density. These results further our understanding of nanopharmacology.


SUPPLEMENTARY METHODS
Proteomic analysis: Purified nanoparticles and proteins were incubated at 95°C in 8M Urea for 5 minutes. The polymer was separated from the proteins by adding a 2-fold volume equivalent of acetonitrile and storing overnight at -20 °C. The proteins were pelleted at 10,000 rcf for 10 minutes and washed twice with cold acetonitrile. Reduction, Alkylation and Tryptic Digestion Protein pellets were resuspended in 8M urea and proteaseMAX surfactant (Promega) per manufacturer's instructions. Equivalent amounts of protein (20ug) from each condition were reduced (10 mM dithiothreitol, 56 °C for 45 min) and alkylated (50 mM iodoacetamide, room temperature in the dark for 1 h). Proteins were subsequently digested with trypsin (sequencing grade, Promega), at an enzyme/substrate ratio of 1:50, at room temperature overnight in 100 mM ammonium acetate pH 8.9. Trypsin activity was quenched by adding formic acid to a final concentration of 5%. Peptides were desalted using C18 SpinTips (Protea) then lyophilized and stored at −80 °C. Tandem Mass Tags (TMT) Labeling Peptides were labeled with TMT 10plex (Thermo) per manufacturer's instructions. Lyophilized samples were dissolved in 70 μL ethanol and 30 μl of 500 mM triethylammonium bicarbonate, pH 8.5, and the TMT reagent was dissolved in 30 μl of anhydrous acetonitrile. The solution containing peptides and TMT reagent was vortexed, incubated at room temperature for 1 h. Samples labeled with the ten different isotopic TMT reagents were combined and concentrated to completion in a vacuum centrifuge. Biological duplicates were analyzed in two sets of experiments, each with 3 technical triplicates. The samples were labeled using the TMT 10plex channels as presented in Supplementary Each experiment included two independent batches of nanoparticles with 15 and 25 PEG chains per 100 nm 2 . These nanoparticles therefore had a total of four biological replicates. All isotopic quantification data was compared to the 130N channel. This reference contained plasma which was collected, and treated in the same manner as the nanoparticles: that is, through a 20-cm Sephacryl S-400 HR column (fractions 9-11 were collected) followed by three washing steps with PBS on a Vivaspin 1,000 kDa MWCO ultrafiltration unit. The other plasma channels (130C and 131) were plasma diluted 200-fold, and normal plasma, respectively. These samples were used to assess whether the size-exclusion column and ultrafiltration purification or dilution had any effect on the amounts of different proteins measured.

LC-MS/MS
Peptides (100ng) were loaded on a precolumn and separated by reverse phase HPLC (Thermo Easy nLC1000) over a 140 minute gradient before nanoelectrospray using a QExactive mass spectrometer (Thermo). The mass spectrometer was operated in a data-dependent mode. The parameters for the full scan MS were: resolution of 70,000 across 350-2000 m/z, AGC 3e 6 , and maximum IT 50 ms. The full MS scan was followed by MS/MS for the top 10 precursor ions in each cycle with a NCE of 32 and dynamic exclusion of 30 s. Each experiment was analyzed three times. Raw mass spectral data files (.raw) were searched using Proteome Discoverer 1.4 (Thermo) and Mascot version 2.4.1 (Matrix Science). Mascot search parameters were: 10 ppm mass tolerance for precursor ions; 0.8 Da for fragment ion mass tolerance; 2 missed cleavages of trypsin; fixed modification was carbamidomethylation of cysteine and TMT 10plex modification of lysines and peptide N-termini; variable modification was methionine oxidation. Only peptides with a Mascot score greater than or equal to 25 and an isolation interference less than or equal to 30 were included in the quantitative data analysis. A protein was considered identify if at least 2 unique peptides or 3 total peptides were identified. TMT quantification was obtained using Proteome Discoverer and isotopically corrected per manufacturer's instructions, and the values were normalized to the median of each channel.

Statistical analysis
All replicates (technical and biological) were used to evaluate the effect of PEG density on the adsorption of proteins. Linear regressions (on the curve obtained by plotting the relative increase in protein/peptide vs. the PEG density of the nanoparticles) were calculated by the GraphPad Prism 6.05 software using the means obtained from Proteome Discoverer, on all measurements. The slope of the curve was considered to be different from zero if the p value was below 0.05. In vitro complement activation: Blood from C57bl/6 mice was collected by cardiac punction and centrifuged at 2,000 rcf for 10 minutes. Serum was collected and incubated immediately with nanoparticles. In brief, 20 µL of nanoparticles (concentration 30 mg/mL) were added to 40 µL of serum, and incubated at 37°C for 1 hour. To stop the reaction, 60 µL of EDTA 25 mM was added, and samples were kept frozen until further analysis. The concentration of anaphylatoxin C5a in each sample was measured by using a mouse C5a ELISA (ab193718, Abcam, MA) per the manufacturer's protocol.

ProteomeXchange
In vivo complement activation: Prior to the injection of nanoparticles, 50 uL of blood was collected by the saphenous vein, from healthy male Balb/c mice. Mice were injected intravenously with 3 mg of nanoparticles, and blood was collected in EDTA-coated centrifuge tubes, at 5, 15, 45 minutes. Plasma was separated by centrifugation (2,000 rcf for 10 minutes), and samples were frozen until further use. Complement activation was assayed by measuring the levels of terminal complement complex (C5b-9) using a mouse C5b-9 ELISA (EKU07600, Biomatik, Cambridge, ON) per the manufacturer's protocol.

SUPPLEMENTARY DISCUSSION
Self-assembly vs. surface grafting In this work, the amount of PEG on the surface of nanoparticles is the result of the self-assembly of diblock copolymers, in the absence of surfactants or surface-stabilizing molecules. Since the hydrophilic polymer is present only in the polymer precursor solution, the amount of PEG in the system is defined by the length of each block of the copolymer (for example, 33 wt% PEG with PEG5k-PLGA10k copolymers, 15.6 wt% with PEG5k-PLGA27k). We establish in Fig 1 that 1) most of the PEG is integrated into the nanoparticle, irrespective of the PLGA-PEG mixture used, and 2) most of it is solvated on the surface of the nanoparticle, irrespective of the nanoparticle diameter. Since the surface-to-volume ratio increases when nanoparticles get smaller, the PEG density on different particles prepared with the same PEG content decreases proportionally to their diameter. To study nanoparticles between 55 and 140 nm, mixtures of various uniand diblock copolymer precursors were used to obtain a broad range of PEG contents, from 3.5 to 27 wt%.
In other studies, surface PEGylation is often carried out after synthesis of the core. For example, by conjugating reactive PEGs onto the surface 1, 2, 3, 4 or adsorbing surfactant molecules. 5,6 In those cases, the final amount of PEG attached to the nanoparticle results from the ability of a PEG chain to interact with the surface. Given the hydrated and flexible nature of PEG chains, achieving high PEG densities requires minimizing steric hindrance and favoring interactions with surface of the core. 7 In that context, macromolecules can be more densely tethered to the surface of smaller nanoparticles. 2,8 The assessment of the impact of size is therefore complicated, especially when indirect quantification methods are required to measure the PEG content (e.g., measuring unreacted thiols, 3 or fluorescently labelling the polymer 4 ). Together these differences in methodology and the different methods used for quantitation (radioactivity vs. anti-PEG ELISA and fluorescence) explain the discrepancies between our observations and those of others.

Comparison with liposomes
While polymeric nanoparticles have a dense polymeric core, liposomes consist of phospholipid bilayers delimiting an aqueous center; most of their volume is occupied by a hydration solution, which is not taken into consideration when reporting lipid compositions. While depictions in wt% can be informative from a formulation perspective, they hardly offer any representation of surface density, which is potentially more amenable to comparisons between systems. To compare the polymeric nanoparticles studied here and liposomes, assumptions about the surface area occupied by each of the phospholipid molecule are needed. In the literature, this value depends notably on the chemical structure of the phospholipid and the presence of cholesterol in the bilayer. 9 Likewise, because of the presence of an aqueous compartment inside the liposomes, and depending on the liposome preparation method, assumptions about the amounts of PEG facing the internal compartment (and not actively granting steric protection 10 are also needed. For example, a rough interval of PEG content can be obtained if we estimate the proportion of PEG facing outward to be 60-100%. Hence, 3-6 mol% (vs. total amount of lipid) of DSPE-PEG(5000) would be equivalent to a threshold of 20 PEG chains per 100 nm2 for hypothetical liposomes prepared with dimyristoylphosphatidylcholine (DMPC, 678 g/mol, specific area 44 Å2), 40 mol% cholesterol (387 g/mol), and DSPE-PEG(5000) (5750 g/mol). These values are in agreement with those found in the literature for long-circulating liposomes prepared with PEG5k. 11,12,13 Nevertheless, we would emphasize that comparisons with commercially available PEGylated doxorubicin liposomes might be precarious, since the length of the PEG used in this work differs from that found in those formulations (5,000 vs. 2,000, respectively).
Choice of TMT-label quantitative proteomics analysis vs. label-free proteomics.
In literature focusing on the study of interactions between nanoparticles and proteins 14,15 , researchers mainly use label-free quantification to determine the relative contribution each protein to the overall corona covering nanoparticles (Protein A represents X% of the total proteins found in sample 1). Due to differences in molecular weights, trypsin digestion patterns, ionization and the number of unique peptides, not all proteins are equal with respect to the signal provided by MS. Therefore, in label-free proteomics, quantification methods are used to determine the relative amounts of each protein compared to others, notably using spectral counting 16 and other algorithms. 17,18 In the present work, quantitative protein analysis was chosen instead using TMT-labels. In this method, samples are labeled with different isotopic tags before being injected together in the instrument. 19 This method makes comparisons between the amounts of individual proteins in each sample more accurate (i.e., protein A is more abundant in sample 1 than in sample 2). However, the quantification algorithms discussed above have not been as extensively studied with TMT-labeling (and other quantitative proteomic techniques). In other words, no widely-accepted methodology currently exists to highlight quantitative differences between samples, together with the relative importance of one protein compared to another. It is beyond the scope of this study to define new methodologies to do so.
Our choice of using TMT-labels was guided by challenges in isolating PEGylated nanoparticles from free proteins in plasma using centrifugation methodologies commonly used in the field. 20 Using the radioactive label in the particles, it was determined that only a small proportion (< 25%) of the PLGA-PEG nanoparticles could be pelleted using high centrifugation speeds (> 28,000 rcf). We therefore relied on size exclusion columns 21 and ultrafiltration techniques to separate nanoparticles from the bulk of free proteins. Nevertheless, in our hands, these methods did not allow to get rid of all free proteins; it was observed that control plasma, treated with the same methodology, afforded a small amount of residual proteins (i.e., reference plasma). TMT-analysis, which compared the quantities of protein in samples containing nanoparticles relative to reference plasma, allowed to determine how PEG density affected the adsorption of proteins without interference from the residual proteins.

140-nm
Supplementary Table 4 Number of animals in each group. Sample sizes were chosen to obtain sufficient biological repeats while ensuring feasibility of the study; key groups were repeated to confirm data when needed (n > 5). Healthy animals were used without randomization, and investigators were not blinded. Animals were excluded if experimental problems occurred (problem during injection or sampling, animal death upon dosing (n= 1)).
Supplementary Table 5 Raw data for circulation profiles obtained in C57Bl/6 and KO animals.
Supplementary Figure 1 PEG content results obtained by 1H-NMR are in good agreement with those obtained by a broadly used iodine-based colorimetric assay. PEGylated nanoparticles were hydrolyzed with 1N sodium hydroxide for 30 minutes and reacted with an iodine/potassium iodine aqueous solution (9mM iodine and 155 mM of potassium iodine). Absorbance at 495 nm was read on a Tecan Infinite 500 plate reader. The PEG content was determined by comparing absorbance of samples to a PEG calibration curve. 22 Supplementary Figure 2 Nanoparticles prepared with mixtures of PEG-PLGA and PLGA all show similar zeta-potential, irrespective of size or PEG density.
Supplementary Figure 3 Nanoparticles with PEG densities below the 20 PEG chains per 100 nm 2 show higher distribution to the liver and spleen. While liver distribution of highly PEGylated nanoparticles is similar across sizes, capture in the spleen seems to increase with the nanoparticle diameter. In each of the main graphs, each point represents one animal, while means  SD are presented offset. * p < 0.05 as determined by t-test, ** p < 0.05 as determined by Mann-Whitney U (non-parametric) test.
Supplementary Figure 4 A PEG density threshold is also perceivable in C57Bl/6 mice. As seen in Balb/c mice, the circulation profile and pharmacokinetic parameters of 90-nm nanoparticles in C57Bl/6 mice show little added benefit when the PEG density is augmented above 20 PEG chains per 100 nm 2 . Values are means  SD (n= [4][5][6][7][8][9][10][11]. For comparison purposes, full symbols represent the pharmacokinetic parameters for Balb/c mice. Supplementary Figure 8 Nanoparticles with low and high PEG densities (fast and rapid blood clearance, respectively) are mild activators of the complement cascade, irrespective of the PEG density (n= 3, * p < 0.05 as determined by one-way ANOVA and Tukey's post-hoc).
Supplementary Figure 9 Intravenous injections of nanoparticles with low and high PEG densities (fast and rapid blood clearance, respectively) fail to significantly increase the circulating levels of the terminal complex of the complement cascade (C5b-9) in Babl/c mice (n= 3).