Mapping and identification of soft corona proteins at nanoparticles and their impact on cellular association

 Nanoparticles

The 70-nm fluorescent plain amine and carboxyl-modified silica nanoparticles (sicastar-greenF, fluorescein isothiocyanate-labeled; ex/em = 485/510 nm) were purchased from micromod Partikeltechnologie GmbH (Germany). The fluorescent carboxylate-modified polystyrene nanoparticles (FluoSpheres™ Carboxylate-Modified Microspheres, 0.1 µm, yellow-green fluorescent (505/515)) were purchased from Thermo Fisher.

Click-chemistry reagents

Sulfo-SASD (sulfosuccinimidyl-2-(p-azidosalicylamido) ethyl-1,3-dithiopropionate) was purchased from G-biosciences. Dibenzocyclooctyne (DBCO)-Supho-NHS and DBCO-Sulpho-Cyanine5 were purchased form click-chemistry tools and Genabioscience companies, respectively. Poly(L-lysine)-graft[3.5]-Poly(ethylene glycol)(2) (PLL-g-PEG) was purchased from SuSoS Ag.

Fluorescent reagents

Sulpho-NHS-Cyanine5 was purchased form Lumiprobe. Hoechst 33342 Solution (20 mM) was purchased from Thermo Fisher. Phalloidin–Tetramethylrhodamine B isothiocyanate was purchased from Sigma-Aldrich.

Cell media

Penicillin–Streptomycin, Gibco, heat-inactivated Fetal Bovine Serum (FBS), and RPMI 1640 Media were purchased from Thermo Fisher. PMA Phorbol 12-myristate 13-acetate was purchased from Sigma-Aldrich.

Nanoparticle incubation with FBS

FBS was first centrifuged at 16,000 × g for 3 min to remove any insoluble aggregates. Then, the protein supernatant and NPs solutions were preincubated at 37 °C for 10 min before mixing. The NPs were then exposed to FBS for different time points (15 min, 30 min, 1 h, 2 h, and 6 h) in darkness at 37 °C in Protein Lobind tubes.The ratio of total particle-surface area to FBS volume was kept constant for all nanoparticles. About 0.4 mg of SNPs (70 nm) and 0.3 mg of PsNPs (100 nm) were used. The nanoparticle–corona complexes were isolated from unbound and loosely bound FBS proteins by centrifugation at 18,000 × g, 20 min. Pellets were washed three times with PBS by centrifugation at 20,000 × g, 20 min, and resuspended in PBS for further analysis.

Azide modification of HC proteins

To generate azide-functionalized HC proteins, sulfo-SASD at different concentrations (0, 0.018, 0.036, 0.09, 0.18, 0.35, 0.55, and 1.8 mM) was added to the nanoparticles@HC at a final concentration of 0.4 mg ml−1. Sulfo-SASD contains a dithiol in the structure that can be cleaved by a reducing agent for electroporation analysis. The complex was incubated at 37 °C for 1 h to allow the reaction between the sulfo-NHS and amine groups on HC proteins. The azide-modified nanoparticle–corona complexes were separated from unreacted sulfo-SASD by centrifugation at 20,000 × g, 20 min. The pellet was washed three times with PBS and then resuspended in PBS for further steps. To characterize the azide groups on HC proteins, a click reaction between the azide groups and DBCO-sulfo-Cy5 was employed. By measuring the excited fluorescence (at 646 nm) of proteins at 664 nm, the minimum number of azide groups was calculated on HC proteins, assuming that each DBCO-sulfo-CY5 reacts with one sulfo-SASD. The number of HC corona proteins was measured by two methods: (1) by calculating the average molecular weight of proteins on SDS-PAGE and quantification of proteins by BCA assay and (2) LC–MS/MS analysis. In all steps, the number of nanoparticles was measured by reading the fluorescence of NPs.

For nanoparticles that interfere with the analysis, the proteins should be first eluted from nanoparticles, and then the click reaction between DBCO-sulfo-Cy5 can be performed (Supplementary Fig. 1a). We performed the eluting protocol for SNPs and compared the results with a noneluting protocol. To elute the proteins, the nanoparticles were resuspended in 500 µl of 1% acetic acid and incubated overnight. Then, the solution was removed by a centrifugal evaporator, and the proteins in the pellet were quantified by BCA assay (which is explained in the following). Then, the proteins were incubated with DBCO-sulfo-CY5 for the click reaction, and the unreacted dyes were eliminated by using a Sephadex G-25 in PD-10 Desalting Columns (GE Healthcare Life Sciences). The results of the BCA assay and click reaction showed the same efficiency for both eluting and noneluting protocols.

Modification of proteins

The reactive DBCO sulfo-NHS at various final concentrations (0, 0.2, 0.4, and 0.8 mM) was added to a tube containing 10% FBS solution (4.2 mg ml−1) at PBS (pH 7.4). The molar ratio of the cross-linkers to proteins (if we assume all proteins are BSA) at the selected concentrations is equal to 0, 2.5, 5, and 10. The system was then incubated at 37 °C for 1 h to allow the reaction between the sulfo-NHS and amine groups on proteins. The reaction was quenched by Tris at a final concentration of 50 mM. The unreacted DBCO molecules were eliminated by a PD-10 Desalting Column. To make fluorescently labeled DBCO-modified FBS proteins or individual proteins (BSA or APOH), proteins were incubated with both DBCO-sulfo-NHS and sulfo-NHS-Cy5 (at the molar ratio of cross-linkers to proteins of 5). The degree of labeling (DOL) of proteins was calculated by measuring the UV spectrum of conjugates using the following equation:

$${\mathrm{DOL}} = \frac{{A_{{\mathrm{max}}} \times \epsilon _{{\mathrm{280}}}}}{(A_{{\mathrm{280}}} - A_{{\mathrm{max}}} \times {\mathrm{CF}}) \times \epsilon _{{\mathrm{max}}}}$$

(1)

where Amax and A280 are absorbances of the conjugate solution measured at 280 nm and at λmax of the cross-linker or dye, respectively. λmax values for DBCO sulfo-NHS and NHS-sulfo-CY5 are 309 and 646 nm, respectively. ε280 and εmax are the extinction coefficient of proteins (for FBS proteins, we used ε280 value of albumin, 66,433 M−1 cm−1) and DBCO sulfo-NHS or NHS-sulfo-CY5, whose values were taken as 12,000 and 271,000 M−1 cm−1, respectively. CF is the correction factor of each cross-linker or dye, which is required to eliminate the contribution of the dye at 280 nm. Amax and A280 are absorbances of the conjugate solution measured at 280 nm at λmax of the cross-linker or dye.

Cross-linking SC proteins on HC proteins

In order to capture weakly interacting proteins on nanoparticle–corona complexes, the azide-modified particles were incubated with DBCO-modified FBS (FBS-D) for 2 h at 37 °C. The azide-modified nanoparticle–corona complexes were separated from free proteins by centrifugation at 20,000 × g, 20 min. The pellet was washed three times with PBS and then resuspended in PBS for further steps.

Characterization of nanoparticle–corona complexes

Nanoparticles were characterized by transmission electron microscopy (TEM), dynamic light scattering (DLS), and zeta potential measurements. Both zeta potential and hydrodynamic diameter of nanoparticles were measured by a Malvern Zetasizer Nano (Malvern Instrument Ltd., UK) with a laser wavelength of 633 nm in 10 mM sodium phosphate buffer at pH 7.4. For the calculation of zeta potential, data processing was done by using Smoluchowski model48. For TEM analysis, nanoparticles were loaded onto glow‐discharged 200-mesh copper grids (Formvar/carbon grids, Ted Pella) for 20 s, blot-dried, and then stained three times with uranyl formate and dried. TEM imaging was performed using a Tecnai G2 Spirit BioTWIN (FEI) operating at 120-kV acceleration. Images were obtained on a TemCam‐F416(R) (TVIPS) CMOS camera. To estimate the size distribution of nanoparticle–corona complexes, the size of at least 150 particles was measured by Fiji/ImageJ. Nanoparticle sizes were determined in aqueous conditions by dynamic light scattering (DLS).

Quantification of proteins by BCA assay

Proteins on nanoparticles were quantified by using a Pierce BCA Protein Assay Kit (Thermo Scientific). For SNPs and PsNPs, the proteins were quantified without stripping from nanoparticles. The pellet of washed nanoparticle–corona complexes was resuspended in 50 µl of PBS. About 400 µl of working reagent (copper solution) was added to the samples and standard solutions and then incubated for 30 min at 37 °C. The samples were centrifuged at 20,000 × g for 30 min. About 200 µl of each supernatant was transferred into a 96-well plate, and the absorbance at 562 nm was measured using a Varioscan plate reader (Thermo Scientific). In order to evaluate the degree of nanoparticle interference, two control samples were performed: pristine nanoparticles in PBS and in BSA standard solutions49. For nanoparticles that interfere with the BCA assay, the proteins should be eluted first from nanoparticles by the protocol that was explained before in the “Azide modification of HC proteins” section and then analyzed by BCA assay.

Eluting corona proteins from NPs

Concentrated 5X Lane Marker reducing sample buffer (SB, Thermo Scientific, 0.3 M Tris, 5% SDS, 50% Glycerol, and 100 mM DTT) was added to nanoparticle–corona complexes to recover proteins from the NPs. The samples were heated at 95 °C for 5 min to denature and strip off proteins from NPs. DTT as a reducing agent in the sample buffer can cleave the S–S bond in the sulfo-SASD structure, which helps to cleave the clicked proteins from each other before running them on a SDS-PAGE gel. Then, the samples were centrifuged at 20,000 × g, 20 min.

Electrophoresis analysis

For electrophoresis analysis, the recovered proteins were diluted with PBS to adjust the sample buffer, and 40 µl of the proteins were separated on a 12% SDS-polyacrylamide gel (Bolt™ 4–12% Bis-Tris Plus Gels, 10-well) at the constant voltage of 160 V. A PageRuler Unstained Protein Ladder (Thermo Scientific) as the molecular weight standard (10–200 kDa) was also run on the gels. For CY5-labeled samples, the gels were first imaged by an Amersham Typhoon NIR laser scanner. Then, the protein bands were detected by Imperial Protein stain (Coomassie brilliant blue, Thermoscientific). The stained gels were scanned on a Bio-Rad gel documentation system. Protein quantification was performed using the plot profile tool in Fiji/ImageJ (ref). The staining intensity and run length were normalized based on the maximum values.

LC–MS/MS analysis

The eluted corona proteins from nanoparticles were first precipitated by using the ProteoExtract® Protein Precipitation Kit (Merck, Germany) as described in the manufacturer’s protocol. The proteins were dissolved in 8 M urea and 100 mM ammonium bicarbonate with 10 mM DTT. After 30 min, adding iodoacetamide to a final concentration of 35 mM alkylated the samples. The alkylation was quenched after 30 min by adding DTT to a final concentration of 35 mM. Subsequently, the samples were diluted 5 times and digested with trypsin 1/50 (w/w) in 16 h at 37 °C. Tryptic peptides were micropurified using Empore™ SPE C18 Disks packed in 10-µl pipette tips. LC–MS/MS was performed using an EASY-nLC 1000 system (Thermo Scientific) connected to a QExactive+ Mass Spectrometer (Thermo Scientific). Peptides were trapped on a 2-cm ReproSil-Pur C18-AQ column (100-μm inner diameter, 3-µm resin, Dr. Maisch GmbH, Ammerbuch-Entringen, Germany). The peptides were separated on a 15-cm analytical column (75-μm inner diameter) packed in-house in a pulled emitter with ReproSil-Pur C18-AQ 3-μm resin. Peptides were eluted using a flow rate of 250 nl min−1 and a 20-min gradient from 5 to 35% phase B (0.1% formic acid and 100% acetonitrile). The collected MS files were converted into Mascot generic format (MGF) using Proteome Discoverer (Thermo Scientific, v.24). The data were searched against the bovine proteome (uniprot.org). Database search was conducted on a local mascot search engine. The following settings were used: MS error tolerance of 10 ppm, MS/MS error tolerance of 0.1 Da, trypsin as protease, oxidation of Met as a variable modification, and carbamidomethyl as a fixed modification.

Uni- and multivariate statistical analysis

To test the statistical significance of differences, ANOVA analysis of the data was performed using GraphPad Prism version 8.2.1 for Windows, GraphPad Software, La Jolla, California, USA, www.graphpad.com. For cluster analysis, the copy number of corona proteins per nanoparticle was used. This approach allows a comparison of different samples without bias for large-sized proteins or the total protein input. Using this method, a protein with a higher copy number in HC+SC than in the four control HC samples is not necessarily considered an SC protein because a higher copy number could be acquired by coincidence. Therefore, the proteins classified as SC are restricted to proteins that had a consistently lower (or zero) copy number in all control samples. Clustered heatmaps were created on square root-transformed and scaled datasets using the packages gplots (ver. 3.0.1.1) and dendextend (ver. 1.9.0) in the R environment (ver. 3.5.1). For unsupervised hierarchical clustering, the distance matrix was calculated using Ward’s minimum variance algorithm with the Euclidean metric.

For each corona protein identified, protein parameters, such as the grand average of hydropathy (GRAVY) scores, instability index, and isoelectric point (PI) of the proteins, were extracted using ProtParam, a tool available in the SIB ExPASY Bioinformatic Resources Portal50. Principal component analysis (PCA) was performed on scaled datasets using the FactoMineR (ver. 1.41)51 and factoextra (ver. 1.0.5) packages for R to explore in a multivariate manner characteristic features of protein parameters among the corona proteins.

Quantification of individual proteins on nanoparticles

In order to calculate the copy number of individual protein in corona proteins on nanoparticles, three types of measured data were employed: (1) emPAI values obtained by LC–MS/MS analysis, (2) quantified total protein mass by BCA assay, and (3) quantified nanoparticle number by reading the fluorescence of the nanoparticle. The copy number of proteins per nanoparticle was calculated by using the following expressions52:

$${\mathrm{Protein}}\,{\mathrm{mass}}\,{\mathrm{per}}\,{\mathrm{nanoparticle}} = {\mathrm{protein}}\,{\mathrm{content}}\,\left( {{\mathrm{weight}}\,{\mathrm{\% }}} \right) \times M_{{\mathrm{total}}} \\ = \frac{{{\mathrm{emPAI}} \, \times {\mathrm{Mw}}_p}}{{{\sum} {\left( {{\mathrm{emPAI}} \, \times {\mathrm{Mw}}_p} \right)} }} \times M_{{\mathrm{total}}}$$

(2)

$${\mathrm{Copy}}\,{\mathrm{number}}\,{\mathrm{of}}\,{\mathrm{protein}}\,{\mathrm{per}}\,{\mathrm{nanoparticle}} = \frac{{{\mathrm{protein}}\,{\mathrm{mass}}\,{\mathrm{per}}\,{\mathrm{nanoparticle}}}}{{{\mathrm{Mw}}_p}} \times N_{\mathrm{A}}$$

(3)

where protein content (weight %) is the contribution of each protein to the total adsorbed mass, Mwp is the calculated molecular weight of the protein, and Mtotal is the overall mass of corona proteins per nanoparticle measured by employing BCA assay and fluorescence of nanoparticles. NA is the Avogadro constant (6.023 × 1023).

Estimation of coverage of nanoparticles by corona proteins

In order to estimate the surface coverage of nanoparticles by corona proteins, the Protein Data Bank (PDB) files of the proteins that are available were extracted from PDB: http://www.rcsb.org. Then, the structure of proteins was analyzed by PyMOL, and the minimum and maximum cross-section area of proteins were calculated. For the proteins without PDB files, by assuming the simplest shape, sphere, and this partial specific volume (ν = 0.73 cm3 g−1), the volume occupied by a protein of mass M in dalton and its radius were calculated as follows53:

$$V\left( {{\mathrm{nm}}^3} \right) = {\mathrm{1}}{\mathrm{.212}} \times {\mathrm{10}}^{ - {\mathrm{3}}}\left( {\frac{{{\mathrm{nm}}^3}}{{{\mathrm{Da}}}}} \right) \times M\,{\mathrm{(Da)}}$$

(4)

$$R = {\mathrm{0}}{\mathrm{.066}} \times M^{{\mathrm{1/3}}}\,{\mathrm{(for}}\,M\,{\mathrm{in}}\,{\mathrm{Dalton,}}\,R\,{\mathrm{in}}\,{\mathrm{nanometer)}}$$

(5)

where V and R are the volume and radius of protein.

Since some proteins have quaternary structure and/or are in the lipoprotein structure, and it is not clear if the protein prefers to adsorb onto the nanoparticle in their natural or denatured structure, it is not possible to calculate the exact coverage of nanoparticles by proteins. On the other hand, proteins can bind to nanoparticles from their maximum and minimum cross-section area. Assuming these limitations for the most abundant proteins, different coverage values were calculated, and a range for each calculation was reported.

Surface plasmon resonance (SPR)

SPR measurements were made on a Biacore 3000 (Biacore AB Sweden). Gold SPR chips from SIA kit Au were cleaned with ultrasonication in acetone, ethanol, and DI water (10 min each), followed by 30 min of UV/ozone, before sputter deposition of 4-nm Ti followed by 20 nm of SiO2. The SiO2-coated chips were cleaned with ultrasonication in acetone, ethanol, and DI water (10 min each), followed by 30 min of UV/ozone a maximum of 1 day before use. All injections were at a rate of 5 µl min−1 of 100 µl. First, 0.25 mg ml−1 filtered PLL-g-PEG was injected with 10 mM HEPES (pH 7.4) as a running buffer. Next, 0.25 mg ml−1 70-nm SiO2 NPs were injected with 10 mM NaCl as a running buffer. The running buffer was changed to 10 mM HEPES containing 100 mM NaCl (pH 7.4) (NaCl–HEPES) for all protein injections. For FBS-coated NPs, 1% FBS in NaCl–HEPES was injected prior to APOH. APOH was injected sequentially, rinsing between each injection, in concentrations 1, 15, 50, and 150 µg ml−1 in NaCl–HEPES.

SPR data analysis

Linear drift corrections were applied if necessary, using an average of the background drift before and after injection. The two-dimensional fits were made on the MATLAB 2012a platform (Mathworks) using the fitting tool EVILFIT version 3 software54,55 to determine the distribution of binding kinetics. The following input values were used for fitting the binding curves: Injection start time: Concentrations: 20, 100, 300, 1000, and 3000 nM, Start injection: t = 0 s, End injection: t = 800 s, Fit the binding phase from: t = 2 s, Fit the binding phase to: t = 798 s, Fit the dissociation phase from: t = 1400 s, and Fit the dissociation phase to: 2400 s.

The operator-set boundaries for the distributions were uniformly set to limit KD values in the interval from 10−9 to 10−3 M, and Kd values in the interval from 10−5 to 100 s−1.

The distribution P (ka, KA) is calculated using the discretization of the equation:

$$R_{{\mathrm{total}}} = \mathop {\smallint }\limits_{K_{a\min }}^{K_{a\max }} \mathop {\smallint }\limits_{k_{{\mathrm{a}}\,{\mathrm{min}}}}^{k_{{\mathrm{a}}\,{\mathrm{max}}}} R\left( {k_{\mathrm{a}},K_{\mathrm{a}},C_{{\mathrm{analyte}}},t} \right)P\left( {k_{\mathrm{a}},K_{\mathrm{a}}} \right)dk_{\mathrm{a}}dK_{\mathrm{a}}$$

(6)

in a logarithmic grid of (ka,i, Ka,i) values with 21 grid points distributed on each axis. This was done through a global fit to association and dissociation traces at the above-mentioned analyte concentrations. Tikhonov regularization was used as described by Zhao et al.56 at a confidence level of P = 0.95 to determine the most parsimonious distribution that is consistent with the data, showing only features that are essential to fit the data.

Cell culture

We used THP-1 monocyte cells (a human acute monocyte leukemia cell line) obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, ACC 16) and human cerebral microvessel endothelial hCMEC/D3 cells. Both cell types were grown in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin–streptomycin in a humidified 5% CO2 atmosphere at 37 °C. THP-1 cells were differentiated into macrophages (hereafter “THP-1 macrophages”) by incubating with 5 ng ml−1 PMA for 48 h. The differentiated phenotype was visually inspected under an optical microscope, and then the cells were washed two times with PBS to remove PMA followed by an additional 24-h incubation in the medium without PMA57 prior to cell experiments described below.

Cell association of nanoparticle–corona complexes

Cell association was assessed by a NovoCyte flow cytometer and a confocal laser-scanning microscope (CLSM, Zeiss LSM 700). For flow cytometry, 0.5 ml of hCMEC/D3 cells or THP-1 macrophages at the density of 5 × 105 cells ml−1 were seeded in 24-well plates and exposed to the nanoparticle–corona complexes for 4 h in the RPMI media containing either 5 mg ml−1 BSA or 10% FBS. Then, the plates were washed three times with PBS to remove free nanoparticles. The cells were then fixed by 4% paraformaldehyde for 15 min. The cells were washed three times with PBS and resuspended in 200 µl of PBS. In the flow cytometry analysis with NovoExpress software (ver. 1.4.1), at least 10,000 cells were counted. The fluorescence data are presented as median and calculated as the ratio of the median fluorescence intensity of the samples and the pristine nanoparticles in 5 mg ml−1 BSA.

For the confocal analysis, first, the glass coverslips were coated with 50 µg ml−1 collagen type I. For better collagen coating, the coverslips were first coated with poly-d-lysine (PDL) and then with collagen. Then, 2.5 × 105 THP-1 or hCMEC/D3 cells were seeded onto collagen precoated glass coverslips. THP-1 cells were incubated for 48 h with PMA for differentiation followed by 1 day in RPMI without PMA. hCMEC/D3 were incubated for 24 h for attachment. After the differentiation of THP-1 cells and the attachment of hCMEC/D3 cells on coverslips, the cells were exposed to the nanoparticle–corona complexes for 4 h. Then, the cells were washed and fixed with 4% paraformaldehyde. Cell nuclei were stained with 10 µg ml−1 Hoechst 33342 (excitation: 448 nm, emission: 430–480 nm). The actin filaments were stained with 1 µg ml−1 phalloidin–tetramethylrhodamine B isothiocyanate (excitation: 540 nm, emission: 570–573 nm). The confocal images were analyzed with ImageJ (v.1.51) and Zen (ZEISS, v.2.6).

Reporting summary

Further information on experimental and research design is available in the Nature Research Reporting Summary linked to this article.