Poly (N-vinylpyrrolidone) modification mitigates plasma protein corona formation on phosphomolybdate-based nanoparticles

Graphical Abstract Phosphomolybdate-based nanoparticles (PMo12-based NPs) have been commonly applied in nanomedicine. However, upon contact with biofluids, proteins are quickly adsorbed onto the NPs surface to form a protein corona, which induces the opsonization and facilitates the rapid clearance of the NPs by macrophage uptake. Herein, we introduce a family of structurally homologous PMo12-based NPs (CDS-PMo12@PVPx(x = 0 ~ 1) NPs) capping diverse content of zwitterionic polymer poly (N-vinylpyrrolidone) (PVP) to regulate the protein corona formation on PMo12-based NPs. The fluorescence quenching data indicate that the introduction of PVP effectively reduces the number of binding sites of proteins on PMo12-based NPs. Molecular docking simulations results show that the contact surface area and binding energy of proteins to CDS-PMo12@PVP1 NPs are smaller than the CDS-PMo12@PVP0 NPs. The liquid chromatography-tandem mass spectrometry (LC–MS/MS) is further applied to analyze and quantify the compositions of the human plasma corona formation on CDS-PMo12@PVPx(x = 0 ~ 1) NPs. The number of plasma protein groups adsorption on CDS-PMo12@PVP1 NPs, compared to CDS-PMo12@PVP0 NPs, decreases from 372 to 271. In addition, 76 differentially adsorption proteins are identified between CDS-PMo12@PVP0 and CDS-PMo12@PVP1 NPs, in which apolipoprotein is up-regulated in CDS-PMo12@PVP1 NPs. The apolipoprotein adsorption onto the NPs is proposed to have dysoponic activity and enhance the circulation time of NPs. Our findings demonstrate that PVP grafting on PMo12-based NPs is a promising strategy to improve the anti-biofouling property for PMo12-based nanodrug design. Supplementary Information The online version contains supplementary material available at 10.1186/s12951-021-01140-8.

These groups tend to be long hydrophilic polymer chains and nonionic surfactants. Some polymers include polyethylene glycol (PEG), polysaccharides, polyacrylamide, and PEG-containing copolymers as examples of shielding groups. PEGylated Doxil nanodrug has been used as clinical medicine, yet PEG cannot completely shield protein adsorption [21] to modulate the immune response.
The poly(N-vinyl-2-pyrrolidone) (PVP) zwitterionic polymer is a sub-class of polyampholytes that possess equivalent positive and negative charges on the same pendant group maintaining overall electrical neutrality [22]. A correlation between surface charge and opsonization has been demonstrated in vitro, with research showing that neutrally charged NPs have a much lower opsonization rate than charged NPs [23]. In addition, studies on PMo 12 -based NPs mainly focus on the design and screening of potent PMo 12 -based nanomaterials [24,25], as well as their pharmacology study [14,26]. Limited effort has been devoted to the protein corona formation mechanism on zwitterionic polymer modified PMo 12 -based NPs. Herein, we synthesized a series of structurally homologous PMo 12 -based NPs (CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs) capping diverse content of zwitterionic polymer poly (N-vinylpyrrolidone) (PVP) by micelle-based approach. The cesium dodecyl sulfate (C 12 H 25 SO 4 Cs, CDS) cationic surfactant is used to trap the PMo 12 O 40 3− polyanion for preparing CDS-PMo 12 @PVP 0 NPs (Scheme 1a) [25]. When the PVP is introduced to the reaction system, PVP firstly interacts with CDS to form CDS-PVP complex through the electrostatic/hydrophobic forces [27]. The CDS-PVP complex matrix decelerates the contact rate of the CDS and the PMo 12 O 40 3− polyanion. Besides, the PVP polymer adhered to the surface of CDS-PMo 12 @PVP x (x = 0.05 ~ 1) NPs serves as a protective layer to prevent the further aggregation of NPs. Consequently, the PVP that added to the reaction system manages to regulate the size and the surface properties of CDS-PMo 12 @ PVP x (x = 0.05 ~ 1) NPs.
As a proof of concept, Cyt-C, Hb, and BSA were adopted as representative basic, neutral, and acidic Scheme 1 Preparation of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs for decreased adsorption of serum proteins. a CDS is used as a cationic nucleating agent to trap the PMo 12 O 40 3− polyanion to synthesize CDS-PMo 12 @PVP 0 NPs. The introduction of PVP to the CDS-PMo 12 O 40 3− reaction system further synthesizes homogenous CDS-PMo 12 @PVP x (x = 0.05 ~ 1) NPs. b Heterogeneous CDS-PMo 12 @PVP 0 NPs without PVP coating protection could largely adsorb proteins. c In contrast, PVP modified homogeneous CDS-PMo 12 @PVP x (x = 0.05 ~ 1) NPs could protect the NPs from protein adsorption proteins to investigate protein adhesion behaviors on CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. The adsorption efficiencies were gradually decreased when the PVP content increased, suggesting the addition of PVP suppresses the adsorption of protein to NPs (Scheme 1b, c). Fluorescence quenching measurements further unveiled the underlying interaction mechanism between proteins and two typical PMo 12 -based NPs (CDS-PMo 12 @PVP x (x = 0,1) NPs). The introduction of PVP influences the binding kinetics and thermodynamic process of protein adsorption, reducing the number of binding sites, and subsequently influences the adsorption of protein. The hydrophobic interactions are identified as the driving forces for proteins binding to CDS-PMo 12 @PVP 0 NPs, while the electrostatic interactions are identified as the main forces between proteins and CDS-PMo 12 @PVP 1 NPs. The specific binding sites and contact surface area (CSA) were further visualized by molecular docking computational simulations. The CSA of proteins binding on CDS-PMo 12 @PVP 0 NPs is larger than that of CDS-PMo 12 @PVP 1 NPs. Importantly, the CSA of Cyt-C is larger than that of Hb and BSA on CDS-PMo 12 @PVP x (x = 0,1). Next, a series of CDS-PMo 12 @ PVP x (x = 0 ~ 1) NPs were incubated with human plasma, the composition of the plasma protein corona was examined by label-free liquid chromatography mass spectrometry (LC-MS/MS). Along with the increase of PVP content, the number of identified protein groups covering the CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs decreases gradually. Moreover, 76 differential adsorption proteins between CDS-PMo 12 @PVP 0 and CDS-PMo 12 @PVP 1 NPs are identified, in which apolipoprotein is up-regulated of the plasma corona proteins on CDS-PMo 12 @PVP 1 NPs. Researches have proved that the adsorption of apolipoproteins can prolong circulation times [28]. Therefore, our studies demonstrate that PVP grafting on PMo 12 -based NPs mitigates plasma protein corona formation, which provides a new potential strategy for PMo 12 -based nanodrug design with better biological sustainability.

Results and discussions
Synthesis and characterization of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs The CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs were prepared by a micelle-based approach (Fig. 1a) with detailed experiment process given in the Additional file 1: Fig. 1b and Additional file 1: Fig. S1 provide the transmission electron microscopy (TEM) images of CDS-PMo 12 @ PVP x (x = 0 ~ 1) NPs. CDS-PMo 12 @PVP 0 NPs are heterogeneous, with sizes ranging from 100 to 1000 nm. The CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs size distributions are statistically analyzed from TEM images by ImageJ software (Additional file 1: Fig. S2). The main size of CDS-PMo 12 @PVP x (x = 0.05 ~ 1) NPs are approximately 520, 482, 454, 300, 235 nm, respectively. The size of particles became smaller (from 520 to 235 nm) and more homogeneous with the increase of PVP content. Besides, the morphology of CDS-PMo 12 @PVP 1 NPs were characterized by the high-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM, Fig. 1c), and the corresponding elemental compositions were analyzed by energy-dispersive X-ray spectroscopy (EDS) in HAADF-STEM. According to the HAADF-STEM-EDS elemental mapping, we found that the O, Mo, P, Cs, C, and N elements exist simultaneously, illustrating the PVP successfully anchored to the CDS-PMo 12 @PVP 1 NPs [25]. The hydrodynamic diameters (d h ) and polydisperse index (PDI) values of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs was characterized by dynamic light scattering (DLS) (Fig. 1d) and the corresponding values are presented in the caption of Fig. 1d. Figure 1e showed the X-ray diffraction (XRD) patterns of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. The 2θ diffraction peaks correspond to the crystalline phase peaks of the H 3 [PMo 12 O 40 ] [29], indicating that the prepared PMo 12 -based NPs retain the H 3 [PMo 12 O 40 ] Keggin structure. The peak intensity of XRD is mainly related to the crystallinity of the crystal. Therefore, the higher the degree of crystallization, the higher the intensity of XRD peak. When the content of PVP (0.1 ~ 1) in the reaction system increases, the order structure of crystals decreases, resulting in poor crystallinity, so the intensity of XRD becomes lower. On the other hand, it Standard deviations were calculated from three independent measurements. e XRD patterns of the CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. f The adsorption efficiency of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs towards three types of proteins. g the Zeta-potential of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs (B-R buffer, pH = 6). h SEM images of CDS-PMo 12 @PVP 0 and CDS-PMo 12 @PVP 1 NPs after absorbing the mixed solution of three types of proteins. Arrows indicate that the obvious protein coating induce rough surfaces of CDS-PMo 12 @PVP 0 NPs and very few proteins coating on the surface of CDS-PMo 12 @PVP 1 NPs has been confirmed that the peak intensity of XRD is also influenced by the nanoparticles' size, the smaller size of nanoparticles, and the higher intensity of XRD the peak.
This explains why the XRD increases with the increasing PVP content (0 ~ 0.1). The FT-IR spectra of CDS-PMo 12 @PVP 0 , CDS-PMo 12 @PVP 0.5 , and CDS-PMo 12 @  Fig. S3) [24], which indicates that the PMo 12 O 40 3− in the nanocomposites remain the Keggin structure. The absorption bands of CDS-PMo 12 @PVP 0.5 and CDS-PMo 12 @PVP 1 arising from the PMo 12 O 40 3− are shifted in position, which demonstrates that the bonds of the PMo 12 O 40 3− are either strengthened or weakened, owing to the interaction between the N of the PVP and O atom of the PMo 12 O 40 3− . Besides, the resonance peak of C-O (at 1643 cm −1 ) shows no change, and the peak of the N-OH complex (at 1288 cm −1 ) disappears as compared with the PVP spectrum. These changes in the spectrum of CDS-PMo 12 @PVP 1 suggest the coordination between N and PMo 12 O 40 3− as the main reaction, while the reaction between O and PMo 12 O 40 3− is less significant. Overall, these results demonstrate PVP was successfully anchored to the PMo 12 -based NPs.

Protein adsorption behaviors on the CDS-PMo 12 @ PVP x (x = 0 ~ 1) NPs
Cyt-C, Hb, and BSA were chosen as representatives of basic, neutral, and acidic proteins to evaluate adsorption behaviors of the CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. Additional file 1: Fig. S4 summarized the protein adsorption performance on CDS-PMo 12 @PVP 0 NPs. The optimized adsorption conditions from preliminary experiments (adsorption time: 20 min, temperature: 25 °C, B-R buffer concentration: 0.04 mol L −1 ) were applied to the adsorption study of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs [30]. The adsorption efficiencies of Hb and Cyt-C (Hb/Cyt-C) on CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs were gradually decreased when PVP content increased, while no obvious alteration in the adsorption efficiencies of BSA was observed (Fig. 1f ). Zeta-potential measurements illustrate that the neutral amphiphilic PVP polymer neutralizes the electronegativity of the CDS-PMo 12 @ PVP 0 NPs (Fig. 1g), limiting the adsorption of positively charged Hb/Cyt-C proteins on the CDS-PMo 12 @ PVP x (x = 0.05 ~ 1) NPs at pH 6, hence explains the declined adsorption efficiencies. On the other hand, BSA (pI = 4.7) is negatively charged at pH 6, and virtually no retention of BSA occurs on the negatively charged CDS-PMo 12 @PVP 0 NPs surfaces. The SEM images show obvious protein coatings on the surface of CDS-PMo 12 @PVP 0 NPs and barely any retention on the surface of CDS-PMo 12 @PVP 1 NPs (Fig. 1h) after absorbing the mixed solution of three types of proteins. Our results confirm that PVP in the nanocomposites regulates the chemical composition, particle size, surface physicochemical property, and consequently, protein adsorption efficiency.

The binding process and mechanism of protein adsorption on PMo 12 -based NPs
To explore the effect of PVP on the binding kinetics and thermodynamic process of protein/PMo 12 -based NPs complexes formation, we investigated the protein fluorescence quenching process upon binding to (CDS-PMo 12 @ PVP x (x = 0,1) NPs quenchers [31]. First, we incubated proteins with different concentrations of CDS-PMo 12 @ PVP x (x = 0,1) NPs ranging from 0 to 15 μM at 298 K and 310 K for 10 min, respectively. Then we measured the intrinsic fluorescence intensity (here tryptophan residue of Cyt-C, Hb, and BSA proteins) before and after incubation with CDS-PMo 12 @PVP x (x = 0,1) NPs and analyzed fluorescence quenching data by the Stern-Volmer (S-V) equation (Eq. 1) [32].
where F 0 and F are the fluorescence intensities of proteins in the absence and presence of quencher (here CDS-PMo 12 @PVP x (x = 0,1) NPs); K SV is the S-V quenching constant; [Q] is the total concentration of the quencher; k q is the quenching rate constant, and τ 0 is the fluorophore average lifetime in the absence of quencher (for biomolecules is 10 −8 s). The fluorescence spectra of proteins incubating with CDS-PMo 12 @PVP x (x = 0,1) and S-V plot are displayed in Fig. 2b-d and Additional file 1: Figs. S5-S7. The S-V plot of proteins binding to CDS-PMo 12 @PVP x (x = 0,1) NPs shows a positive deviation from a linear S-V relation ( Fig. 2d and Additional file 1: Fig. S7). The positive deviation of the slope is attributed to the simultaneous presence of dynamic and static quenching [33][34][35][36][37].
A modified S-V equation (Eq. 2) was used to calculate the effective S-V quenching constant (K sv ) and the quenching rate constant (k q ) of the interaction of proteins and CDS-PMo 12 @PVP x (x = 0,1) NPs.
In which f a is the mole fraction of accessible fluorescence, and K sv is the effective S-V quenching constant for the accessible fluorophores. The dependence of F 0 /(F 0 -F) vs [Q] −1 , should be linear with the slope of (faKsv) −1 , whereas the value fa −1 is fixed on the ordinate. Therefore, the effective quenching constant Ksv is a quotient of the ordinate fa −1 and the slope (faKsv) −1 . The modified S-V plot of proteins binding to CDS-PMo 12 @PVP x (x = 0,1) NPs are shown in Fig. 2e and Additional file 1: Fig. S7. The values of K sv for BSA/CDS-PMo 12 @PVP 0 , Hb/CDS-PMo 12 @PVP 0 , Cyt-C/CDS-PMo 12 @PVP 1 , and Hb/CDS-PMo 12 @ (1) PVP 1 complexes are increased with the temperature (Table 1), indicating a dynamic process exists in those complexes. The Ksv of Cyt-C/CDS-PMo 12 @PVP 0 and BSA/CDS-PMo 12 @PVP 1 complexes are correlated inversely with temperature, indicating the proteins quenching mechanism is initiated by static quenching. However, the Kq of the proteins/CDS-PMo 12 @ PVP x (x = 0,1) complexes are higher than the maximal dynamic quenching constant (2 × 10 10 M −1 s −1 ), revealing the main quenching mechanism for proteins binding to CDS-PMo 12 @PVP x (x = 0,1) NPs is static quenching [38]. The number of binding sites (n), and binding constant (K a ) were obtained according to the double logarithmic equation (Eq. 3) [39].
The double logarithmic plot of proteins binding to CDS-PMo 12 @PVP x (x = 0,1) NPs are shown in Fig. 2f and Additional file 1: Fig. S7. By linear fitting for the the double logarithmic plot, the values of n and K a are obtained from the slope and Y-axis intercept, respectively. The value of n for Cyt-C/Hb binding to CDS-PMo 12 @PVP 1 NPs is smaller than that binding to CDS-PMo 12 @PVP 0 NPs ( Table 1). The results suggest that the introduction of PVP reduces the number of binding sites of proteins on CDS-PMo 12 @PVP 1 NPs. Since the values of K a for Cyt-C/Hb binding to CDS-PMo 12 @PVP 0 NPs have no apparent increase with temperature (Table 1), combined with the static quenching mechanism, we cross-verified the stable complex formation between Cyt-C/Hb and CDS-PMo 12 @PVP 0 NPs. Conversely, with the temperature rising, the value of K a increases largely for BSA/ CDS-PMo 12 @PVP 0 complex (Table 1), suggesting the BSA/CDS-PMo 12 @PVP 0 complex is unstable [40][41][42]. K a is dependent on temperature, which indicates that the protein formation on CDS-PMo 12 @PVP x (x = 0,1) NPs is a thermodynamic process. Enthalpy change (∆H 0 ), entropy change (∆S 0 ), and free energy change (ΔG 0 ) are used to further characterize the driving interaction force between CDS-PMo 12 @PVP x (x = 0,1) NPs and three types of proteins. The values of ∆S 0 and ∆H 0 were calculated from the slope and the intercept, respectively, by fitting linearly to the plot of lnK a Vs. 1/T according to the Van Hoff equation Eq. (4), whereas the ΔG 0 value was calculated from Eq. (5).
where K a is the binding constant at the corresponding temperature (T) and R is the gas constant. ∆S 0 and ∆H 0 are determined from the linear Van't Hoff plots. ΔG 0 is estimated from the following equation (Eq. 5) [43]: According to the views of Timasheff [44], the positive values of ∆S 0 and ∆H 0 indicate that hydrophobic forces plays a major role in protein interaction with CDS-PMo 12 @PVP 0 NPs (Table 1). A negative value of ∆H 0 and positive values of ∆S 0 indicate that electrostatic forces are the major forces between the CDS-PMo 12 @PVP 1 and proteins ( Table 1). The negative sign of the ΔG 0 proves that the CDS-PMo 12 @PVP x (x = 0,1) NPs interact with three types of proteins are spontaneous (Table 1).

Molecular docking study of CDS-PMo 12 @PVP x (x = 0,1) NPs and protein interactions
Molecular docking was further performed on the interaction of PMo 12 -based NPs with proteins to identify the binding orientation of CDS-PMo 12 @PVP 0 and CDS-PMo 12 @PVP 1 onto proteins. The docking results indicate that hydrophobic forces is the main driving forces for CDS-PMo 12 @PVP 0 binding to three types of proteins, and while the electrostatic interactions are identified as the main forces between proteins and CDS-PMo 12 @PVP 1 NPs. During CDS-PMo 12 @PVP 0 interacts with the proteins, the primary force is hydrophobic and the secondary forces is electrostatic (Fig. 3a). The contact surface area (CSA) between the model proteins and the PMo 12 -based NPs was calculated using molecular docking. Molecular docking [45] results show that CDS-PMo 12 @PVP 1 binds with Cyt-C, Hb, and BSA mainly through electrostatic interaction, but also Van der Waals force as the second action force (Fig. 3b). The contact surface area (CSA) and binding energy between the three model proteins and CDS-PMo 12 @ PVP x (x = 0,1) NPs were calculated using molecular docking [45]. As shown in Fig. 3c, d, the CSA of proteins on CDS-PMo 12 @PVP 0 is larger than that on CDS-PMo 12 @ PVP 1 . Importantly, the CSA of Cyt-C is larger than that of Hb/BSA on CDS-PMo 12 @PVP x (x = 0,1) NPs. The binding energy of the three model proteins binding to CDS-PMo 12 @PVP 0 is smaller than that binding to CDS-PMo 12 @PVP 1 , thus confirming a higher affinity of three model proteins to CDS-PMo 12 @PVP 0 NPs than CDS-PMo 12 @PVP 1 NPs. Thus, the lower binding affinity of proteins to CDS-PMo 12 @PVP 1 NPs mitigate the protein adsorption. It is generally believed that reducing biofouling could significantly attenuate subsequent adverse inflammatory responses including leukocyte activation, tissue fibrosis, thrombosis coagulation, and infection [46].

Comparison of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs in human plasma protein corona formation
The CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs were then incubated in human plasma to acquire a stable protein corona, followed by centrifugation of NPs from unbound proteins. The plasma proteins in the protein corona were then digested, purified, and eluted. The resulting peptides from the NPs-bound corona were analyzed by LC-MS/MS coupled with label-free quantification in data-dependent acquisition mode (DDA) [47] (Fig. 4a, see details in "Experimental and methods").

Conclusions
The PMo 12 -based NPs are reported to deliver promising anti-tumor biological activities by the virtue of their desired diversity in structures and properties. However, nanomaterials' effective nanomedicine applications are hampered by limited understanding and control over their interactions with complex biological systems. Here, we adopted PVP polymer to modulate the size and surface functionality of PMo 12 -based NPs. PVP successfully decreased the protein adsorption on the surface of CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. On the assessment of the interaction mechanism between engineering (See figure on next page.) Fig. 3 The molecular docking computational results of the interaction of Cyt-C, Hb, and BSA with CDS-PMo 12 @PVP x (x = 0,1) NPs. The structure snapshots of Cyt-C, Hb, and BSA adsorption on (a) CDS-PMo 12 @PVP 0 and (b) CDS-PMo 12 @PVP 1 NPs, respectively. c The CSA visualization of the three model proteins binding to the CDS-PMo 12 @PVP 0 NPs (the upper row) and binding to CDS-PMo 12  PMo 12 -based NPs and proteins (base, neutral and acid proteins), the steady-state fluorescence quenching results revealed the interaction between model proteins and CDS-PMo 12 @PVP 0 NPs occurs spontaneously mainly by hydrophobic forces, whereas the electrostatic interactions make the main forces between proteins and CDS-PMo 12 @PVP 1 NPs. Molecular docking results indicated that the introduction of PVP reduces the number of binding sites and contact surface area. LC-MS/MS further indicated that the PVP reduces the plasma proteins adsorption on the CDS-PMo 12 @PVP 1 NPs. In addition, apolipoprotein as the main composition of adsorption proteins on CDS-PMo 12 @PVP 1 NPs is proposed to have dysoponic activity, enhancing the circulation time.
Overall, the surface physicochemical properties of NPs have a significant impact on the adsorption of proteins. We believe such regulation of the surface physicochemical of NPs and in-depth understanding of the protein adsorption process can effectively facilitate the design of PMo 12 -based nanodrug.
Protein adsorption behavior on the CDS-PMo 12 @ PVP x (x = 0 ~ 1) NPs BSA, Hb, and Cyt-C were chosen as models of acidic, neutral, and basic proteins to evaluate the proteins' adsorption behavior on the CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs. The experiment procedure is below: 1.0 mL of protein solution was mixed with 5.0 mg of CDS-PMo 12 @ PVP x (x = 0 ~ 1) NPs and the mixture was shaken vigorously for 20 min to facilitate the adsorption of proteins. The pH of the protein solution was controlled by Britton-Robinson (B-R) buffer (a mixture of 40 mmol/L phosphoric acid, acetic acid, boric acid, and adjusted by 200 mmol/L sodium hydroxide) within a range of 3-7. After the adsorption process, the solid and liquid phase was separated by centrifugation at 8000 rpm for 6 min and the residual proteins in the aqueous phase were monitored using a UV-vis spectrophotometer in a 1.0 cm quartz cell by measuring the characteristic adsorption at 406 nm for Hb, 409 nm for Cyt-C, and 595 nm for BSA directly. The protein adsorption efficiency was calculated based on the protein concentration before and after adsorption.

Steady-state fluorescence quenching measurements
The fluorescence emission spectra of BSA, Hb, and Cyt-C were measured at a constant concentration (5 μM) in the presence of an increasing concentration of PMo 12 -based NPs (0-15 µM in particles). A certain concentration of PMo 12 -based NPs was added to the protein solutions and incubated for 10 min, and then the mixture was transferred to a quartz cuvette, and their fluorescence spectra were acquired in the range of 300-600 nm when excited at 290 nm. The area under each fluorescence curve was integrated and used to measure the free BSA concentration using a standard calibration curve. The cuvette path length for fluorescence quenching measurements was 1 cm. The slit widths of excitation and emission were set both at 10 nm, respectively.

Molecular docking
The molecular docking is done on AutoDock Vina software [50] to predict binding parameters, the contact surface area (CSA), and binding energy of three model proteins (BSA, Hb, and Cyt-C) with target nanoparticles.  [54], and PVP input files were prepared using the AutoDock Tools 1.5.4 package [50]. Each molecular docking calculation produced 20 binding mode states poses with the exhaustiveness parameter value equal to 1000. In all cases, the binding energy and root-mean-square deviation (RMSD) are considered together and the one with the best affinity is selected as the optimal binding mode. After selecting the best mode of interaction, the CSA was calculated by the related code in VMD package [55] was used to analyze molecular docking results. It is worth mentioning that we repeated the docking calculations three times and obtained the same values.

Protein corona preparation and proteomic analysis
Plasma samples were diluted 1:5 in B-R buffer (a mixture of 40 mmol/L phosphoric acid, acetic acid, boric acid, and adjusted by 200 mmol/L sodium hydroxide). To form the protein corona, 0.5 g CDS-PMo 12 @PVP x (x = 0 ~ 1) NPs was mixed with 500 μL diluted plasma samples in a tube. The tube was sealed and incubated at 37 °C for 1 h with shaking at 300 rcf. After incubation, the mixture was centrifuged to separate the nanoparticle-protein complexes from plasma solution for at least 20 min at 4 °C at 15,000 rcf. Discard the supernatant and wash the pellet with ddH 2 O (300 μL). The protein corona was further washed with 200 μL of ddH 2 O three times with centrifugal separation. Elute the proteins from the nanoparticles by adding 100 μL of RIPA lysis buffer and incubate for 5 min at 95 °C. Pellet the nanoparticles by centrifugation for 15 min at 15,000 rcf at room temperature, and then transfer the supernatant containing eluted corona proteins to a fresh tube. Determine the protein concentration of the supernatant by Pierce BCA protein assay. The minimum quantity for LC-MS/MS analysis is 20 μg of protein per sample. The extracts from each sample (50 μg protein) were mixed with acetone at a volume ratio of 1:4, then precipitated at − 20 °C for 2 h, and centrifuged at 20 000 g for 10 min. The supernatant was poured and the precipitant was washed twice more with acetone. Subsequently, each protein sample (20 μg protein) was reduced by DTT (60 min, 55 °C) and free cysteines alkylated with IAA ( 30 min, 25 °C in the dark). After these procedures, protein samples were loaded into 10 kDa ultrafiltration tubes, washed three times with 50 mM NH 4 HCO 3 . Samples were incubated with trypsin overnight at 37 °C. Digested peptides were transferred into a C18 peptide clean-up column, washed by Solvent A (0.1% FA in water) and eluted with elution buffer (60% ACN and 40% FA in water). Clean peptides were finally concentrated and dried in a SpeedVac (Eppendorf ). All the blood samples were approved by the Human Ethics Review Committee of Science and Technology, Shanghai Jiao Tong University according to the Chinese regulation.

LC-MS/MS analysis
Samples were analyzed on Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific) with nano spray flex ion source and Thermo Scientific ™ EASYnLCTM 1200 integrated ultra high-pressure nano HPLC system. The purified and dried peptides (500 ng) were re-dissolved in Solvent A and then automatically injected and loaded onto the trap column (75 μm × 2 cm; particle size, 3 μm; pore size, 100 Å; Thermo Fisher Scientific) at a flow rate of 2 μL min −1 (max pressure 500 bar). After 5 min, the peptides were eluted from the trap column and separated on the analytical column (75 μm × 25 cm; particle size, 2 μm; pore size, 100 Å; Thermo Fisher Scientific) by a gradient ranging from 8 to 100% of ACN mobile phase at 300 nL min −1 flow rate for 120 min.

Label-free based protein identification and quantification
The acquired MS raw data were loaded to Proteome Discoverer © (version 2.4, Thermo Scientific) software for label-free quantification. The database search was specified by trypsin as enzyme for digestion and peptides with up to two missed cleavages were included. The data exported from Proteome Discoverer was analyzed using Excel © software. The normalized abundance for proteins and peptides were used for subsequent statistical analysis. Missing value and coefficient of variation (CV) value were mainly used for MS performance quality control. k-nearest neighbors algorithm was used for the missing value imputation method. The CV value was used to evaluate the dispersion of the replicates within one group, and proteins with CV ≤ 0.3 were considered reliable here. In addition, P-value was adjusted by Benjamin and Hochberg (BH, 1995) method. The normalized abundance ratio of CDS-PMo 12 @PVP 1 /CDS-PMo 12 @PVP 0 was used for the calculation the FC. Differentially adsorption proteins were identified according to the following two criterions: P-value < 0.01 and FC > 2 and FC < 0.5.