Red Blood Cell Membrane Bioengineered Zr-89 Labelled Hollow Mesoporous Silica Nanosphere for Overcoming Phagocytosis

Biomimetic nanoparticles (NPs) have been actively studied for their biological compatibility due to its distinguished abilities viz. long-term circulation, low toxicity, ease for surface modification, and its ability to avoid phagocytosis of NPs by macrophages. Coating the NPs with a variety of cell membranes bearing the immune control proteins increases drug efficacy while complementing the intrinsic advantages of the NPs. In this study, efforts were made to introduce oxophilic radiometal 89Zr with hollow mesoporous silica nanospheres (HMSNs) having abundant silanol groups and were bioengineered with red blood cell membrane (Rm) having cluster of differentiation 47 (CD47) protein to evaluate its long-term in vivo behavior. We were successful in demonstrating the increased in vivo stability of synthesized Rm-camouflaged, 89Zr-labelled HMSNs with the markedly reduced 89Zr release. Rm camouflaged 89Zr-HMSNs effectively accumulated in the tumor by avoiding phagocytosis of macrophages. In addition, re-injecting the Rm isolated using the blood of the same animal helped to overcome the immune barrier. This novel strategy can be applied extensively to identify the long-term in vivo behavior of nano-drugs while enhancing their biocompatibility.

out using spectrometric analysis and dynamic light scattering system. IR absorption peaks of synthesized HMSNs showed common spectral bands at 801 ν sym cm −1 (Si-O-Si), 965 ν as cm −1 (Si-O), and 1106 ν as cm −1 (Si-OH). Bands at 2854 ν sym and 2924 ν as cm −1 were assigned to the -CH 2 group present in the intact CTAB, while a broad band at 3466 cm −1 was due to the water molecules adsorbed by the HMSN silanol groups (see Supplementary Fig. S1). The X-ray diffraction pattern indicates that the HMSN mesopores are disordered which is a typical pattern of mesopore NPs (see Supplementary Fig. S2). The thermogravimetric analysis (TGA) was carried out under static air up to 1070 °K at a heating rate of 10 °C/min. TGA of HMSN shows the weight loss due to dehydration and removal of free cationic surfactant (CTAB) at 150-300 °C with significant weight loss (~20%). It's stability at higher temperature was confirmed by further increasing the temperature up to ~550 °C without any remarkable weight reduction (≤1%) (Supplementary Fig. S3). Scanning electron microscopy (SEM), and transmission electron microscopy (TEM) images ( Fig. 2 and Supplementary  Fig. S4) demonstrate HMSNs with a particle size of ~150 nm which can be loaded by various drugs into their mesopores, enhancing the EPR effect. The thickness of the HMSNs with coated Rm was increased by about ~20 nm, which has no significant change in particle size while maintaining its effective passive targeting property. SNPs and HMSNs with hydrodynamic sizes of 150 and 160 nm possessed zeta potentials of −66.2 and −18.6 mV (Fig. 3A) respectively. The negative charge was due to the presence of silanol groups on the NPs. HMSNs had a Stepwise representation of the synthesis of Rm-camouflaged 89 Zr-labelled HMSNs. First, blood collection by <10% circulating blood volume from the mouse and second, re-injection for overcoming the immune barrier.
www.nature.com/scientificreports www.nature.com/scientificreports/ higher surface charge than SNPs with intact CTAB. Therefore, the oxophilic radiometals can be introduced easily and effectively into the HMSNs. The Rm-coated NPs (Rm-SNPs and Rm-HMSNs) had similar surface charges (−39.0 and −40.6 mV) due to the presence of sialylated glycoproteins on the Rm, preventing its agglomeration and enhancing its dispersion and diffusion abilities (Fig. 3B). Surface proteins of the Rm and Rm-HMSNs were identified by the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and western blot. Protein bands of Rm and Rm-HMSNs were identical ( Fig. 4A and Supplementary Fig. S5) suggesting the precise topological orientation of the Rm on the core of HMSNs which is crucial characteristics while performing  www.nature.com/scientificreports www.nature.com/scientificreports/ their desired functions in this study. The level of CD47 expressed on the Rm-HMSNs was higher (RBC ghost 99.7% ± 4.70 and Rm-HMSNs 69.5% ± 10.4) (Fig. 4B). Lowered density of CD47 expressed by Rm-HMSNs is the direct consequence of losses occurred while its extrusion. The above mentioned results indicates successful translocation of the Rm on the surface of HMSNs and shows that the function of RBCs can be maintained and HMSNs can overcome immune-barrier. Cell viability. 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay conducted with various concentrations of HMSNs and Rm-HMSNs (0-50 µg/mL) for 48 h showed low toxicity in both cases ( Supplementary Fig. S6). However, at higher concentrations, the Rm-HMSNs displayed better results towards the cell viability as compared with bare HMSNs. The low toxicity of the Rm-HMSNs indicates that they could reduce systemic toxicity while displaying increased long-term circulation. It is important to note that Rm used to coat the NPs in this study were isolated using the blood samples collected from the heart of the same mouse later used to assess the in vivo behavior of NPs. Studies involving the radiotracer needs µg amount of NPs which will indeed require small amount of Rm, this was achieved by withdrawing <10% circulating blood volume from the mouse. This helped to overcome the immune barrier and toxicity involved in using Rm from the different sources.
Radio-labelling and stability. We used 89 Zr as a radio-tag to investigate the long-term biological behavior of Rm-HMSNs. 89 Zr is osteophilic in nature and accumulates in the bone if not complexed with the suitable agent, enhancing its passive cumulation generating pseudo-images. However, using a bulky ligand with large surface area increases its interaction with several proteins, inhibiting its main function. Here, 89 Zr was directly introduced into the SNPs and HMSNs through its interaction with the Si-O-group from silanol without a chelator. As a proof of concept, we examined the comparative labeling efficiency and stability of HMSNs labelled with 89 Zr, 18 F (the gold standard for PET) and 68 Ga (an oxophilic radiometal) (Supplementary Fig. S9 and Table. S1). Based on bond dissociation energy, the oxophilicity trend was Zr > Ga > F which can be explained by the chelator-free binding of 89 Zr and 68 Ga to the silanol group 37 . The labeling efficiency and stability of 18 F-HMSNs and 68 Ga-HMSNs at 30 min were 2.93 ± 0.12% and 99.8 ± 1.68% and 28.7 ± 1.40% and 85 ± 1.22%, respectively. Although 18 F and 68 Ga are excellent positron emitters, their short half-life impedes their use as radio-tags to evaluate the long-term in vivo behavior of HMSNs.
The labeling efficiency of 89 Zr towards HMSNs was ≥81.21% within the first 30 min and continued to increase to ≥96.14% at 24 h post-incubation. SNPs showed similar labeling efficiency (95.99% at 24 h) using 89 Zr-chloride. As expected, 89 Zr-labeling efficiency was dependent on the reaction time, where longer reaction time produced higher labeling yields ( Supplementary Fig. S7). The stability studies with 89 Zr-HMSNs and Rm-89 Zr-HMSNs confirmed that 89 Zr incorporated HMSN with Rm was quite stable as compared with uncoated HMSNs. Rm coated HMSNs while avoiding the release of Zr-89 may provide with the tool for long term in vivo studies of the system. (Supplementary Table S2) In vitro; cancer cell internalization. Cellular uptake of Rm-89 Zr-HMSNs was high and fast within 1 hour (73.93 ± 3.04%), but was rapidly released from surface of cancer cell membrane because of weak surface affinity of cancer cells towards Rm-89 Zr-HMSNs. The internalization of Rm-89 Zr-HMSNs tends to increase over 24 hours (33.73 ± 1.42%), followed by an internalization of Rm-89 Zr-HMSNs gradually. Effective internalization rate of Rm-89 Zr-HMSNs appears higher as compared to that with the previously reported studies 38,39 . Based on these results, Rm-89 Zr-HMSNs can reach the tumor site effectively during circulation in the blood. (Supplementary Fig. S8).

Ex vivo; biodistribution.
Ex vivo biodistribution studies with Rm-89 Zr-HMSNs using CT-26 tumor xenograft-bearing Balb/c mice after intravenous (i.v.) injection with the time interval of 96 h are shown in Fig. 5A. Although, accumulation values of Rm-89 Zr-HMSNs in the liver and spleen were approximately 17.09 ± 1.58 and 41.53 ± 0.75 ID/g%, respectively, it shows a significantly lower uptake than previous studies 40, 41 . The low accumulation in the liver and spleen is accompanied with high uptake in the tumor. The maximum tumor uptake www.nature.com/scientificreports www.nature.com/scientificreports/ for Rm-89 Zr-HMSNs was observed at 24 h, with 3.48 ± 0.09 ID/g% with 7.11 ± 0.22 ID/g% in the blood (see Supplementary Tables S3 and S4). To compare the relative uptake in tumors and organs, the Rm-89 Zr-HMSN uptake ratios in tumors and blood and other internal organs were measured after i.v. injection. Tumor to organs/ blood ratios were highest at 48 h suggesting the incorporation of RIs with the long half-life to evaluate the in vivo performance of the nano-drugs.
In vivo; pet imaging of radio-labelled Nps. Small animal PET studies with 18 F-HMSNs, 68 Ga-HMSNs, 89 Zr-HMSNs, Rm-89 Zr-SNPs and Rm-89 Zr-HMSNs ( Fig. 5B and Supplementary Fig. S9) provide insights on their biological behaviors. Although the accumulation of both 89 Zr-HMSNs and Rm-89 Zr-HMSNs could be found in major organs. It should be perceived that, Rm-89 Zr-HMSNs exhibited significantly lower accumulation in liver, lung and spleen than that of 89 Zr-HMSNs. PET image pattern confirms that Rm-89 Zr-HMSNs was evenly distributed in the whole body. It can be well predicted from the images that the Rm is remains intact on the surface of HMSNs for more than 7 days. In addition, as liver is core organ of reticuloendothelial system (RES) for removing exogenous invader, will be a major hurdle hampering the circulation of HMSNs. Rm coating could enable HMSNs to effectively escape RES recognition, improving the circulation time 42 . The root cause of Rm coated HMSNs behavior is that CD47 functions as an inhibitor of phagocytosis through interaction of SIRPα expressed on phagocytes, leading to tyrosine phosphatase activation and inhibition of myosin accumulation at the sub-membrane assembly site of the phagocytic synapse 43 . In this way, CD47 serves as a "don't eat me signal" while the loss of CD47 leads to homeostatic phagocytosis of NPs. CD47 −/− RBCs were readily taken up into macrophages and nonprofessional phagocytes by processes similar to apoptotic cell removal 44 . As mentioned above, we found that a minimum of ~24 h was required for solid tumor penetration of Rm-89 Zr-HMSNs, with a saturation point at ~48 h. For radioimmunoimaging, proteins need around 24-48 h to penetrate in a solid tumor [45][46][47][48] . The blood circulation time of RBCs being ~120 days, it is a good approach to coat HMSNs using Rm to achieve optimal tumor to background ratio. Another factor one needs to consider is the half-life of radionuclide supposed to be used for radioimmunoimaging. 89 Zr with the long half-life of 3.3 d is predominantly suitable in combination with proteins which allows imaging at later time points as compared to that with the conventional radionuclides for gaining optimum information 49 . 18 F and 68 Ga-labelled HMSNs did not display significant PET signals at the tumor sites, suggesting no tumor uptake, which can be attributed to the long-term circulation time needed for Rm-HMSNs to travel to the tumor site combined www.nature.com/scientificreports www.nature.com/scientificreports/ with the short half-life of 18 F and 68 Ga. Thus, the application of Rm-coated 18 F-HMSNs and 68 Ga-HMSNs is illogical for the long-term studies of Rm-HMSNs. In addition, although the distribution of Rm-89 Zr-SNPs was similar to Rm-89 Zr-HMSNs, the tumor uptake was low even when the same amounts of biomimetic NPs were administered, indicating their lower ability to circulate in the blood because of low-fluidity of SNPs. Uptake of Rm-89 Zr-HMSNs by tumors was clearly observed at 24 h, with good tumor-to-background contrast, while avoiding the clearance through phagocytosis. According to previous studies on the association of nano-medicines without stealthy strategies 50 , increased uptake by macrophages is affected by particle shape, size and surface modification. In order to suppress the internalization of NPs by the mononuclear phagocytic system, it has been widely used to introduce a PEG ligand on the surface of NPs. However, the PEGylated nanoparticles can significantly reduce the ligand-targeting bioactivity 51 . In this study, Rm was used as an alternative bio-material and Zr-89 was introduced without chelator to avoid any change in the functional properties of the Rm. Particularly, bare NPs can be accumulated with high uptake in the liver, lung and spleen, which can cause high toxicity during long-term circulation, it has a low circulation half-life, which cannot be evaluated for effective targeting. On the other hand, the whole-body distribution results show high blood circulation half-life and avoidance of phagocytosis.

In vivo; blood clearance. Blood clearance of 89 Zr-labelled HMSNs and Rm-HMSNs was studied in healthy
Balb/c mice. The blood retention of Rm-89 Zr-HMSNs significantly was enhanced over a span of 54 h, compared to the 89 Zr-HMSNs. As shown in Fig. 6, Rm-89 Zr-HMSNs proved that circulation half-life of HMSNs can be prolonged. However, the HMSNs were rapidly cleared out of blood stream because of phagocytosis. The retention of 89 Zr labelled HMSNs and Rm-HMSNs in the blood is 1.25% AI/mL and 1.96% Al/mL, respectively. These results indicate the successful translocation of Rm onto HMSNs surface suggesting that Rm-89 Zr-HMSNs have the merit of long-term circulation.

Conclusion
These studies execute successful synthesis of Rm-89 Zr-HMSNs, having excellent fluidity, stability, and long-term circulation ability in biological systems. Since Rm-HMSNs are biomimetic NPs, they have low toxicity too. While having a promising behavior of chelator-free labeling of the 89 Zr to the silanol group with high labeling efficiency (≥96%), Rm coated HMSNs further evades macrophage-mediated phagocytosis in the immune system and prevents release of 89 Zr. Furthermore, this strategy validated that active targeting can not only be easily achieved and studied by introducing various cell membranes and target ligands but also the introduction of radionuclide into the biomimetic nano-platform can be used as a strategic tool for diagnosis and treatment of various diseases.  www.nature.com/scientificreports www.nature.com/scientificreports/ preparation of hollow mesoporous silica nanospehers. SNPs were synthesized by Stӧber method 52 .

Materials and Methods
Briefly, 10 ml of ammonia solution and 6 ml of distilled water (D.W) were added in 250 ml of ethanol at 30 °C to which 40 ml of TEOS was added rapidly. Reactants were stirred constantly for two hours following which it was centrifuged and thoroughly washed with ethanol and water.
Mesoporous silica coated SNPs (SNP@mSNPs) were further synthesized using 0.15 g of CTAB and dissolved in 30 ml of water following which 30 ml of ethanol and ammonia solution were added. This solution was homogenised by stirring it for 30 min and 0.1 g of SNPs dispersed in 20 ml of distilled water was added. Next, 0.25 mL of TEOS was rapidly added and then stirred for 6 h. The reactant was centrifuged, washed and re-dispersed in 20 ml of water. Finally, HMSNs were synthesized by re-deposition method. The SNP@mSNPs dispersion was stirred for 12 h to which 0.46 g of sodium carbonate was added and stirred at 80 °C for 10 h after which it was suspended in mixture of methanol (16) Characterization of cell membrane proteins. CD47 proteins on the Rm-HMSNs were quantified using Western blotting analysis. Characterization of the Rm proteins, emptied RBCs and Rm-HMSN were treated with LDS lysis buffer. Then, the samples were denatured at 85 °C for 2 min and proteins of Rm and Rm-HMSN were quantitated using BCA assay kit. 25 µg of each protein was added into each well in a 4-12% sodium dodecyl sulfate-polyacrylamide gel. The gel was run at 200 V for 22 min and polyacrylamide gel was stained by coomassie brilliant blue according to the provided protocol before imaging. After separation of protein by SDS-PAGE, proteins transferred onto a PVDF membrane. After blocking the membranes with milk blocking buffer, the membrane was incubated using CD47 primary antibody (1:500) and IgG secondary antibody (1:1,000) was incubated for 3 h. Expressed CD47 protein onto Rm and Rm-HMSN was detected for the luminescence using an ECL kit by Chemi-doc (iBright FL-1000 Imager, Thermo-Fisher Scientific) Labelling efficiency of 89 Zr-labelled sNp and HMsN. SNP and HMSN were dispersed in a HEPES buffer (pH 7.5; 0.1 M, 0.5 mL) at various concentrations (15,50,100,200, 400 and 1,000 µg/mL). The pH of 89 Zr-oxalate and 89 Zr-chloride (~20 MBq) was adjusted to 7.5-8 using a Na 2 CO 3 solution to which dispersed nanoparticles solution was added. After stirring at room-temperature for 24 hours, labelling efficiency was confirmed by radio thin-layer chromatography (radio-TLC) using 50 mM DTPA as the mobile phase 29 .
Red blood cell membrane camouflaged 89 Zr-HMsN and 89 Zr-sNps. RBCs were isolated using the blood of CT-26 bearing female Balb/c mouse withdrawn by cardiac puncture (<10% of circulating blood volume, re-use the blood drawn from same mice to avoid immune barrier). Mouse blood was collected into the 500 μL syringe (24 G) containing 8 μL of 0.5 M EDTA and was centrifuged (1500 rpm, 5 min, 4 °C) to separate the serum, buffy coat, white blood cell and platelet layers sequentially. Separated RBCs were washed with ice cold 1 × PBS. The purified RBCs were hemolyzed by 0.25 × PBS treatment for extraction of Rm then washed by centrifugation (13,500 rpm, 5 min) with ice 1 × PBS. The Rm was re-suspended in 1 × PBS and sonicated for 3 min at power of 130 W. After sonication the Rm fragments were extruded (Avanti, mini-extruder) sequentially through 400 nm and 200 nm polycarbonate (PC) membrane filters. The prepared Rm and 89 Zr labelled NPs were mixed and extruded by 200 nm of PC membrane for at least 10 passes. Rm coated 89 Zr-NPs were centrifuged and washed to remove uncoated Rm.
Cell culture and animal tumor modelling. All cancer cell lines (CT-26, KB and A549) were cultured in DMEM and supplemented with 10% fetal bovine serum. Cell lines were maintained at 37 °C in a humidified atmosphere containing 5% CO 2 . CT-26 transplanted mice were prepared with female Balb/c mice. The cancer cell lines (5 × 10 6 cells) were injected in the right thigh. The tumors were allowed to grow for 2-3 weeks before ex vivo and in vivo animal experiment.
In vitro; cell viability. 1 × 10 4 cells/well were cultured in micro well plates and exposed to different weight of nanoparticles (0-50 µg/mL) and added 10 µL of MTT reagent to each well, including controls and incubated for 48 hr at 37 °C until a purple coloured formazan. It was dissolved in 100 µL of detergent to all wells. Then, supernatant was measured at 570 nm using a microplate reader model. In vitro; cellular uptake and internalization. The cellular uptake and internalization of Rm-89 Zr-HMSNs was measured with CT-26, KB and A549 cells. The cancer cells were sub-cultured in 24 well plates (5 × 10 5 cells/well) and incubated for 0.5, 1, 4, 24 and 48 hours. After incubation, the supernatant was removed and the cell pellets were washed by cold PBS to remove the unbound Rm-89 Zr-HMSNs. Trypsin solution was used to harvest NPs from the internalized cancer cells. Following this, the cancer cell pellets were treated with 0.1 M sodium citrate for 5 min to remove surface bound Rm-89 Zr-HMSNs. The radioactivity of unbounded Rm-89 Zr-HMSNs and cells were measured by gamma counter and results were calculated as the accumulation ratio (%) ± standard deviation (s.d.).