Cetyltrimethylammonium bromide (CTAB, +99%), 3-aminopropyltrimethoxysilane (APTMS, 95%), tetraethyl orthosilicate (TEOS, 98%), and ammonium hydroxide solution (28–30 wt% NH₃ in H₂O) were obtained from Acros Organics. 2-(Methoxy(poly-ethyleneoxy)propyl) trimethoxysilane (PEG-silane, with a molecular weight of 460–590 g mol⁻¹, 90%), N-trimethoxysilylpropyl-N,N,N-trimethylammonium chloride (TMAC, 50% in methanol), and (3-trihydroxysilyl)propylmethyl-phosphonate (THPMP, 42% in water) were supplied by Gelest. Rhodamine B isothiocyanate (RITC), 3-(4,5-dimethyl-2-thiazolyl)-2,5-dimethyl-2H-tetrazolium bromide (MTT) and doxorubicin hydrochloride (Dox) were purchased from Sigma-Aldrich. Absolute ethanol was purchased from Shimakyu Pure Chemicals. All cell culture reagents were obtained from Gibco. All reagents were used without further purification.

PEG-modified MSNs incorporating the RITC red fluorescence dye or not were synthesized with a diameter of either 50 (RMSN₅₀@PEG) or 200 nm (RMSN₂₀₀@PEG), by adjusting the ammonia concentration and temperature (Lu et al., 2009; Lin et al., 2011; Liu et al., 2015). Briefly, the RMSN@PEG obtained after the removal of surfactants possessed a weakly negatively charged surface and was designated N1-MSN@PEG. To obtain various RMSN@PEG MSNs with different surface charges, post-modifications were performed by negatively charged THPMP-silane and positively charged TMAC-silane, respectively, on the N1-RMSN@PEG to obtain N2 to N5 (with weak to strong negative charges) and P1 to P4 (with weak to strong positive charges)-RMSN@PEG MSNs. In more detail, for the syntheses of P1-RMSN₅₀@PEG/TMAC and P4-RMSN₅₀@PEG/TMAC, 200 mg of N1-RMSN₅₀@PEG was dispersed in 100 ml EtOH and was stirred under reflux for 4 h with 0.36 and 3.6 mmol of TMAC-silane, respectively. For the syntheses of N2-, N3-, N4-, and N5-RMSN₅₀@PEG/THPMP MSNs, after dispersing 200 mg of N1-RMSN₅₀@PEG in a mixture of 80 ml H₂O and 1 ml NH₄OH, volumes of 10, 15, 20, and 25 ml of 0.44 M aqueous THPMP solution were respectively added. The mixture was continuously stirred at 40°C for 4 h. Also, to synthesize N4-RMSN₂₀₀@PEG/THPMP MSNs, after dispersing 200 mg of N1-MSN₂₀₀@PEG in a mixture of 80 ml H₂O and 1 ml NH₄OH, a volume of 20 ml of the THPMP solution was added under vigorous stirring at 40°C for 4 h. After that, the NPs were collected by centrifugation (11,000 rpm for 1 h) and subsequently washed with 95% ethanol. All NPs were dispersed in 99.5% ethanol and preserved until use.

TEM was used to observe the morphology and mesostructure of the NPs. The samples were prepared and dried by dropping them onto a carbon-coated copper grid and then observing them on a JEOL JSM-1200 EX II transmission electron microscope operated at 100 kV. Total surface areas and average pore diameters were determined by measuring the nitrogen adsorption–desorption isotherm on a Micrometerics ASAP 2010 apparatus and calculation by the BET and Barrett–Joyner–Halenda (BJH) methods. The hydrodynamic diameters of the NPs in water, PBS, serum-free, and culture medium supplemented with 10% fetal bovine serum (FBS) were assessed using DLS on a Malvern Zetasizer Nano ZS instrument. Zeta potential measurements were performed on a Malvern Zetasizer Nano ZS instrument by determining the electrophoretic mobility and applying Henry’s equation.

The human glioma U87-MG and rat astrocyte CTX TNA2 cell lines were cultured in minimum essential medium (MEM) and Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS, 100 units/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, and 1 mM nonessential amino acids. These cell lines were incubated at 37°C in a humidified atmosphere of 5% CO₂ and subcultured at 80% confluence.

To measure the cytotoxicity induced by various types of RMSNs and Dox-loaded N4-MSN₅₀@PEG/THPMP (N4-MSN₅₀@PEG/THPMP/Dox) MSNs, an MTT ((3-(4,5-dimethyl-2-thiazolyl)-2,5-dimethyl-2H-tetrazolium bromide)) assay was conducted to determine the cell viability. The cells were seeded in 24-well plates at a density of 10⁴ cells/well, followed by subsequent treatment with different concentrations of various types of RMSNs (at 100 and 200 μg/ml) or free Dox (at 2–60 mg/ml) and an equivalent Dox dose of N4-MSN₅₀@PEG/THPMP/Dox for 4 or 24 h in 10% serum-containing media. At the end of this treatment, 0.5 mg/ml MTT was added to each well for another 1 h. The supernatant was carefully removed, and MTT-formazan crystals were dissolved using DMSO. Finally, the absorbance was measured on a microplate reader (Thermo) at a wavelength at 570 nm.

Tg(zfli1:EGFP) zebrafish (D. rerio) were maintained at the zebrafish core facility of Taipei Medical University in a water flow-through system at 28°C under a 14-h light/10-h dark cycle. Animal procedures were approved by the Taipei Medical University Institutional Animal Care and Utilization Committee (TMU-IACUC, Approval No. LAC-2018-0396). Embryos were staged according to the Zebrafish Book.

Tg(zfli1:EGFP) zebrafish embryos were injected with 200–300 nl (200–300 ng/injection) of various types of MSNs (1 mg/ml in H₂O) into the pericardial cavity at 3 dpf. After being injected with various types of MSNs at 4 or 5 dpf, zebrafish embryos were anesthetized with 100 mg/L ethyl m-aminobenzoate (Tricaine, MS222) and embedded in 0.5% low-melting agarose. Fluorescent images were obtained by a confocal laser scanning microscope (TCS SP5, Leica) and two-photon microscopy (Olympus FVMPE-RS). The scanning time was according to the intensity of the fluorescence emitted. 3D digitally reconstructed images were computed by overlapping the sliding scans.

N4-MSN₅₀@PEG/THPMP MSNs (3 mg) without RITC conjugation were suspended in 400 μl of 0.1 M NaHCO₃ for 2 h and then washed twice with double-distilled (dd) H₂O. The N4-MSN₅₀@PEG/THPMP MSNs were mixed with 0.25 mg of Dox in 400 μl of ddH₂O. After stirring in dark at 4°C for 24 h, the mixture was centrifuged at 14,000 rpm and washed with ddH₂O to remove any unloaded Dox. The amount of unloaded Dox was determined quantitatively by the fluorescence intensity of Dox at 480 nm (excitation) and 560 nm (emission) on a multi-well plate enzyme-linked immunosorbent assay (ELISA) reader (Bio-Rad). The Dox loading efficiency (%) and loading amount (μg Dox/mg NPs) were calculated by the following formulas: Loading efficiency  ( % ) = W e i g h t   o f   D o x   i n   N 4 − M S N 50 @ P E G / T H P M P / D o x W e i g h t   o f   t o t a l   D o x × 100 %

and Loading amount = W e i g h t   o f   D o x   i n   N 4 − M S N 50 @ P E G / T H P M P / D o x W e i g h t   o f   N 4 − M S N 50 @ P E G / T H P M P / D o x .

To evaluate the release behavior of Dox in vitro, N4-MSN₅₀@PEG/THPMP/Dox MSNs were placed into a dialysis membrane (MWCO:12–14 kDa) immersed in a phosphate-buffered saline (PBS) solution at different pH values (pH 5.5 and 7.4) and gently shaken at 37°C. At predetermined time intervals, 1 ml of a sample of the released Dox was collected, and the amount of Dox released was analyzed using a multi-well plate ELISA reader (Bio-Rad) with λex = 480 nm and λem = 560 nm (Dox).

To evaluate cellular uptake of N4-RMSN₅₀@PEG/THPMP, the human glioma U87-MG cells (1 × 10⁵) were seeded in a six-well plate. Following N4-RMSN₅₀@PEG/THPMP treatment with indicated concentration (250, 500, and 750 μg/ml) in a 10% serum-containing medium for 24 h, the cells were washed with 1× PBS three times and then subjected to flow cytometry to quantify the efficiency of cell uptake by detecting the RITC dye fluorescence. For imaging, cell uptake and distribution of N4-RMSN₅₀@PEG/THPMP with RITC-fluorescence signals were obtained from an inverted fluorescence microscope (Olympus IX71, Japan).

Dechorionated embryos at 24 hpf were exposed individually to 2 ml of Dox (0.25 and 0.5 mg/ml), N4-MSN₅₀@PEG/THPMP (4.5 and 9 mg/ml), and N4-MSN₅₀@PEG/THPMP/Dox (4.5 and 9 mg/ml) at an equivalent dose of Dox in one well of a six-well chamber, and toxicity was evaluated by recording the embryo viability from 24 hpf to 96 hpf. Embryos were recorded and imaged in the plates by using Olympus IX70-FLA inverted fluorescence microscope (Olympus, Tokyo, Japan) at indicated time points.

One mg of N4-RMSN₅₀@PEG/THPMP was mixed with 300 μl of mouse plasma on an orbital shaker at 250 rpm for 30 min to mimic the dynamic in vivo blood flow condition. After that, the mixture was centrifuged at 15,570 × g for 30 min to separate the NPs-protein corona complex and then washed three times with PBS to remove the unbound and loosely bound corona proteins. The NPs-protein corona complex was added with Laemmli protein sample buffer, heated at 100°C for 10 min, and then loaded on 12% SDS-PAGE gel. Finally, the gel was stained with Coomassie blue and imaged.

For LC-MS/MS analysis, the proteins from SDS-PAGE gel were sliced and subjected to in-gel trypsin digestion and extraction with 0.1% trifluoroacetic acid (TFA) in 50% acetonitrile (ACN). The proteins were analyzed on an Orbitrap Fusion Lumos Tribrid Mass Spectrophotometer (Thermo Fisher Scientific). The mass spectrometry proteomics data have been deposited to the ProteomeXchange consortium via the PRIDE partner repository with the dataset identifier PXD033783.

Each experiment was conducted at least three times and statistical analysis was performed using Student’s t-test. The data were processed by Origin and were expressed as mean ± standard deviation (SD). ***p < 0.001 was considered statistically significant.

Well-suspended and uniform RITC-conjugated MSNs (called RMSNs), with an average diameter of 50 nm, were prepared using diluted tetraethyl orthosilicate (TEOS) and a surfactant (CTAB) via an ammonia-catalyzed sol–gel process according to our previous reports (Lu et al., 2009; Lin et al., 2011; Liu et al., 2015). In order to avoid irreversible aggregation during synthesis, increase the dispersity and stability in biological milieu, prolong blood circulation in vivo through its hydrophilicity and steric repulsion effects (Aggarwal et al., 2009; Lowe et al., 2015), and retain good biocompatibility (Wu et al., 2011), PEG-silane was introduced on the surface of the as-synthesized MSNs by co-condensation. To track the biodistribution in vivo, the fluorescein RITC dye was incorporated into the framework of the MSNs. After removing the surfactant, the weakly negatively charged NPs were called N1-RMSN₅₀@PEG. To obtain different types of RMSNs with various negative and positive surface charges, N1-RMSN₅₀@PEG were respectively further post-modified with negatively charged THPMP-silane in a basic aqueous solution and with positively charged TMAC-silane in ethanol. Finally, four types of P/N-RMSN₅₀@PEG with various zeta potentials were obtained: N1-RMSN₅₀@PEG (weakly negatively charged), N4-RMSN₅₀@PEG/THPMP (strongly negatively charged), P1-RMSN₅₀@PEG/TMAC (weakly positively charged), and P4-RMSN₅₀@PEG/TMAC (strongly positively charged).

As shown in Figure 1A, TEM images indicated the various types of RMSNs with a size of around 50 nm, uniform morphology, and well-ordered pore structure. The hydrodynamic size of the NPs was measured by DLS (Figure 1B and Supplementary Table S1). The results revealed that all of the NPs were well dispersed in different solutions, and their Z-average hydrodynamic diameters were around 50 nm, which is in accordance with our previous results (Liu et al., 2015). Zeta potentials were in the range of −43.3 to +42.3 mV, confirming the presence of positively and negatively functional groups (Figure 1C). The N1-RMSN₅₀@PEG was negatively charged with a zeta potential of −20 mV. After modification with negatively charged THPMP (N4-RMSN₅₀@PEG/THPMP), the zeta potential decreased to −43.3 mV. After surface post-modification with TMAC, a positively charged group, the zeta potential of P1-RMSN₅₀@PEG/TMAC dramatically increased to +18.1 mV. After further conjugation with additional TMAC, P4-RMSN₅₀@PEG/TMAC exhibited a strongly positive surface charge (+42.3 mV). Nitrogen adsorption/desorption using the BET method exhibited type IV isotherms, consistent with a mesoporous structure (Figure 1D). As shown in Figure 1D, average pore diameters according to BJH calculation ranged 1.99–1.50 (RMSN₅₀@PEG/TMAC) and 2.23–1.72 (RMSN₅₀@PEG/THPMP). The decrease in the pore size was due to functional groups filling in the pores of the RMSNs.

Previous reports suggested that engineered NPs are not only capable of crossing the BBB but also induce serious cytotoxicity to the brain, which was associated with surface properties of the NPs, such as the size and surface charge (Trickler et al., 2010; Pandey et al., 2015; Liu et al., 2017). We then evaluated the in vitro cytotoxicity of various types of RMSNs on two brain-related cell lines: the CTX-TNA2 astrocyte cell line and U87-MG glioma cell line. Cell viability as examined by the MTT assay showed no significant change in either CTX-TNA2 or U87-MG cells treated with different concentrations (100 or 200 μg/ml) of various types of RMSNs for 4 or 24 h in 10% serum-containing media condition (Figure 1E). Astrocyte cells play an important role in determining the BBB integrity (Moura et al., 2017). Results showed no obvious cytotoxicity toward astrocytes or glioma cells, demonstrating that these differently charged RMSNs of 50 nm in diameter were not harmful and exhibited good biocompatibility.

To study the biodistribution of RMSNs, an in vivo Tg(zfli1:EGFP) transgenic zebrafish model was used. After injection of various types of RMSNs (200–300 ng/injection) into the pericardial cavity of 3-dpf zebrafish embryo, the biodistributions of RMSNs were observed by fluorescence microscopy 24 h after the injection. The results indicated that the biodistribution in zebrafish embryos treated with positively charged P1- and P4-RMSN₅₀@PEG/TMAC resulted in a similar pattern, in which red fluorescence was detected in the epithelial cells, the yolk, dorsal aorta, and caudal aorta (Supplementary Figure S1); in contrast, the distribution of negatively charged N1 or N4-RMSN₅₀@PEG/THPMP was observed to be in the circulation (dorsal aorta and cardinal vein), and weakly in epithelial cells and yolk; however, significantly strong red fluorescence was localized in the brain of embryos treated with N4-RMSN₅₀@PEG/THPMP. To further analyze the distributions of various types of RMSNs in the brain, confocal microscopic images were examined (Figure 2A). In the brain of larval zebrafish treated with strongly negatively charged MSNs (N4-RMSN₅₀@PEG/THPMP), the image showed that brilliant red fluorescence was observed outside blood vessels (green), suggesting that this RMSN had a higher penetration ability to cross the BBB. However, RMSNs with a positive charge (P1 and P4-RMSN₅₀@PEG/TMAC) or a weak negative charge (N1-RMSN₅₀@PEG) were scantily present in the brain. 3D deconvoluted confocal fluorescence images also confirmed these results (Supplementary Videos S1, S2).

Meanwhile, two-photon microscopy was employed to display the N4-RMSN₅₀@PEG/THPMP distribution in the brain of zebrafish with high-resolution images, clearly showing the red fluorescence outside blood vessels, which was consistent with results of confocal microscopy (Figure 2B). To further confirm the results in detail, analysis by fluorescence microscopy of zebrafish brain tissue section images was performed after treatment with N4-RMSN₅₀@PEG/THPMP. Figure 2C presented the obvious localization of N4-RMSN₅₀@PEG/THPMP in the brain. Therefore, we found that RMSNs with strong negative charges could successfully penetrate the BBB of zebrafish.

Since a stronger negative charge is essential for the better ability to cross the BBB, the critical negative charge and particle size affecting BBB permeability were further explored. We subsequently synthesized a series of negatively charged RMSNs with a lower negative charge than N4-RMSN₅₀@PEG/THPMP (N2-RMSN₅₀@PEG/THPMP and N3-RMSN₅₀@PEG/THPMP) and one with a higher negative charge (N5-RMSN₅₀@PEG/THPMP). To investigate the size effect, strongly negative-charged MSNs with a diameter of 200 nm (N4-RMSN₂₀₀@PEG/THPMP) were synthesized for further studies. All of the synthesis NPs showed a well-ordered mesoporous structure on TEM images (Figure 3A). DLS characterization (Figure 3B) revealed that all of the NPs were easily suspended in different solutions and they exhibited expected sizes of around 50 and 200 nm. Zeta potential measurements showed that the expected negative surface charge ranged from −22.4 to −51.6 mV, depending on the different ratios of THPMP functionalization (Figure 3B). It was noted that N4-RMSN₅₀@PEG/THPMP (−43.3 mV) and N4-RMSN₂₀₀ @PEG/THPMP (−44.7 mV) had similar zeta potentials to investigate the size effect on BBB transport. The MTT assay demonstrated that none of the NPs at concentrations of 100 and 200 μg/ml caused cytotoxicity in CTX-TNA2 or U87-MG cells following treatment for 4 and 24 h (Figure 3C).

These NPs were then injected into the pericardial cavity of Tg(zfli1:EGFP) transgenic zebrafish embryos (3 dpf) to verify their distribution in the larval zebrafish brain (Figure 4). Confocal microscopic images demonstrated that all three of N2-, N3-, and N5-RMSN₅₀@PEG/THPMP were distinctly observed outside blood vessels in the larval brain (white arrows) in a similar pattern to that of N4-RMSN₅₀@PEG/THPMP. However, the amount of N5-RMSN₅₀@PEG/THPMP in the brain was estimated to be small compared to that of N4-RMSN₅₀@PEG/THPMP. N4-RMSN₂₀₀@PEG/THPMP was larger, had the same surface charge as N4-RMSN₅₀@PEG/THPMP, and showed poor penetration into the larval zebrafish brain, because fewer red fluorescent particles were found (arrowheads) (Supplementary Video S3). These results demonstrated that penetration of RMSNs into the zebrafish brain occurred in a negative charge-dependent and size-dependent manner. It was also consistent with larger NPs having a poorer ability to cross the BBB (Liu et al., 2017; Song et al., 2017; Heggannavar et al., 2018). Therefore, MSNs having a charge between −20 and −40 mV and a size in 50 nm are critical for crossing the BBB.

To further estimate the possibility of using N4-RMSN₅₀@PEG/THPMP as a drug carrier for brain delivery, Dox was used for drug loading. Dox also exhibits red fluorescence (Motlagh et al., 2016) so it is easy to trace its localization in vivo from MSN release after delivery. First, non-fluorescent N4-MSN₅₀@PEG/THPMP MSNs were synthesized instead of N4-RMSN₅₀@PEG/THPMP in order to avoid interference from RITC fluorescence. The loading strategy was based on electrostatic interactions between the negative charge of THPMP and the positive charge of Dox (Figure 5A). The loading efficiency and loading amount of N4-MSN₅₀@PEG/THPMP/Dox were 78.13 ± 3.07% and 5.57 ± 0.22 wt%, respectively (Figure 5A). For medical applications, the stability of N4-MSN₅₀@PEG/THPMP/Dox was then evaluated. After Dox loaded, the size of N4-MSN₅₀@PEG/THPMP/Dox did not significantly change with around 60 nm in water according to DLS measurement (Figure 5B). Despite Dox loading, N4-MSN₅₀@PEG/THPMP/Dox was also stable when maintained in water within 15 days because of no significant aggregation from DLS measurement. The results indicated that N4-MSN₅₀@PEG/THPMP/Dox was suitable for in vivo experiment use. Zeta potential revealed that values of the charge of N4-MSN₅₀@PEG/THPMP/Dox were around −39 mV at day 1, −32 mV at day 8, and−25 mV at day 15 (Figure 5C); it became more positive, but was still within the negative charge range of N2-RMSN₅₀@PEG/THPMP to N4-RMSN₅₀@PEG/THPMP, which was beneficial to BBB penetration. Then, in vitro release profile of Dox was determined by incubating N4-MSN₅₀@PEG/THPMP/Dox in PBS buffer at pH 7.4 (a physiological environment) and pH 5.5 (an acidic environment) at 37°C as shown in Figure 5D. We found that more than 80% of the loaded Dox was released from N4-MSN₅₀@PEG/THPMP/Dox incubated in PBS buffer at pH 5.5 within 6 h; in contrast, only about 30% of the loaded Dox was released at pH 7.0. This result indicates the cumulative time-dependent release and a massive release in an acidic condition (pH 5.5); thus it is possible to modulate slow sustained-release and pH-responsive release. With the benefits of sustained-release Dox, in vitro cytotoxicity of N4-MSN₅₀@PEG/THPMP/Dox against human U87-MG glioma cells by using the MTT assay was examined. The cells were treated with free Dox (2–60 mg/ml) and equivalent Dox doses of N4-MSN₅₀@PEG/THPMP/Dox for 24 h. As shown in Figure 5E, both free Dox and N4-MSN₅₀@PEG/THPMP/Dox showed dose-dependent cytotoxic effects, indicating the effective release of Dox.

In addition to in vitro cytotoxicity study, the in vivo zebrafish model was employed to investigate whether N4-MSN₅₀@PEG/THPMP/Dox could decrease Dox induced toxicity (Figure 6). There was no toxicity in the treatment of N4-MSN₅₀@PEG/THPMP under 4.5 mg/ml and 9 mg/ml concentration. In contrast, zebrafish treated with Dox and N4-MSN₅₀@PEG/THPMP/Dox showed time-dependent toxicity. On the 48 h post-incubation, 9 mg/ml of N4-MSN₅₀@PEG/THPMP/Dox (0.5 mg/ml of Dox equivalent dose) treatment displayed 95% excellent survival rate superior to free Dox (0.5 mg/ml) with 0%. The Kaplan–Meier plots of overall survival was shown in Figure 6B and detailed subgroup and statistical analysis was depicted in Supplementary Table S3. The increased survival rate was due to the improvement of Dox toxicity by N4-MSN₅₀@PEG/THPMP/Dox.

Once NPs were administered into the physiological condition, several serum proteins could adsorb rapidly onto the NPs to form a protein layer (also called protein corona). The effect of protein corona surrounding the NPs depends on several parameters such as size, charge and other surface properties of NPs, which could contribute a great deal of influence in determining the NP’s fate. Hence, we explored the possible mechanism of the transport of N4-RMSN₅₀@PEG/THPMP into the BBB by analyzing the interaction with mouse serum proteins. In this study, we used mouse serum as an alternative protein source for the protein corona analysis because of the insufficient availability of zebrafish blood serum for collection to perform the analysis. First, we investigated the in vitro protein corona pattern of N4-RMSN₅₀@PEG/THPMP incubated in mouse serum and demonstrated that several protein bands retained by SDS-PAGE analysis (Figure 7A). The protein bands were extracted and then characterized by LC-MS/MS, a total of 273 proteins were identified which could interact with N4-RMSN₅₀@PEG/THPMP. Compared to control plasma, N4-RMSN50@PEG/THPMP captured proteins in difference with various content of molecular weights (MW) (Figure 7B). The abundance pattern indicated that some lower MW (below 20 KD and 40–60 KD) and some higher MW (above 80 KD) proteins were highly enriched after captured. Moreover, the isoelectric point analysis of captured proteins indicated that the content of higher isoelectric point (above eight and higher) proteins were significantly increased (Figure 7C). The top 20 most abundant corona proteins and plasma proteins are presented in Figure 7D. A significant reduction of serum albumin in the relative abundance of captured proteins was observed, indicating that those proteins in different properties competitive for the surface of N4-RMSN₅₀@PEG/THPMP during the protein corona process. The detailed proteomic analysis based on the heatmap of the most abundant proteins (minimum relative abundance of 0.5%) is shown in Figure 7E . The results suggested a noticeable difference in the profile of the protein content, which was consistent with the results of Figures 7A,B. According to the data of protein corona composition, we further analyzed the possible functional corona proteins of N4-RMSN₅₀@PEG/THPMP involved in the biological process of BBB crossing mechanisms. The Venn diagrams showed the number of identified proteins and BBB crossing related proteins (Figure 7F). A total of 273 corona proteins were identified on N4-RMSN₅₀@PEG/THPMP. Comparing with the UniProt database, where 85 proteins are currently associated with the transport mechanisms across the BBB; there were three proteins which enriched in the N4-RMSN₅₀@PEG/THPMP protein corona. The relative abundances of these three proteins in the protein corona are shown in Figure 7G, verifying the most abundant proteins were as basigin, afamin, and apolipoprotein E (Apo E). Therefore, N4-RMSN₅₀@PEG/THPMP mediated BBB transport could be carried out by those transport proteins.

Furthermore, we explored cell uptake efficacy, focusing on the effects of protein corona and surface modification, proving the internalization of cells and MSNs observed in a 10% FBS-containing medium, which was used to mimic protein corona conditions. It was found that cellular uptake in the human glioma U87-MG cells was not apparent when treated with N4-RMSN₅₀@PEG/THPMP, even if at a high concentration (750 μg/ml) based on the results of flow cytometry (Figure 8A) and fluorescence microscope (Figure 8B) detecting the RITC-fluorescence signals from MSNs. N4-RMSN₅₀@PEG/THPMS could decrease the cell uptake due to the PEG and negatively charged molecules of the THPMP graft.