Approaches to improve the biocompatibility and systemic circulation of inorganic porous nanoparticles

UEF//eRepository

DSpace https://erepo.uef.fi

Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta

2018

Approaches to improve the biocompatibility and systemic

circulation of inorganic porous nanoparticles

Tamarov, Konstantin

Royal Society of Chemistry (RSC)

Tieteelliset aikakauslehtiartikkelit

© Royal Society of Chemistry All rights reserved

http://dx.doi.org/10.1039/c8tb00462e

https://erepo.uef.fi/handle/123456789/6977

Downloaded from University of Eastern Finland's eRepository

1

Approaches to improve the biocompatibility and systemic circulation of inorganic porous nanoparticles

K. Tamarov, S. Näkki, W. Xu and V.-P. Lehto

Department of Applied Physics, University of Eastern Finland, Kuopio, Finland

Table of Content 1. Introduction

2. Opsonization and immune response 3. Biocompatibility/safety

4. Physicochemical modifications 4.1. Polymer coating

4.1.1. Polyethylene glycol 4.1.2. Other polymers 4.2. Peptide coating

4.3. Protein coating 4.4. Lipid coating

4.5. Cell membrane coating 5. Conclusions

6. Future prospects 7. Conflicts of interest 8. Acknowledges

Abstract

The exploitation of various inorganic nanoparticles as drug carriers and therapeutics is becoming increasingly common. The first issue to be considered with regard to the nanomaterials being utilized in medicine centers on their safety. The functionality of nanocarriers in real-life environments explains the enthusiasm for their use. Several functionalities are typically added onto nanocarriers but the most crucial feature of those carriers intended to be administered intravenously is that they should possess a long residence time in the blood circulation. The present review focusses on the mesoporous nanoparticles due to their great promise in nanomedicine and concentrates on their coatings because it is the outmost layer which dictates their first interactions with the surroundings and often determines their biofate.

Keywords

2

Inorganic nanovector, drug delivery, chemotherapy, photothermal therapy, photodynamic therapy, mesoporous material, functionalization, in vivo experiments, opsonization, biocompatibility

1. Introduction

Various vectors have been used to deliver genes, oligomers, peptides and small therapeutic molecules into cells. Viruses have been most commonly exploited especially in antisense and gene therapies as they are highly efficient at delivering their cargos into cells. However, viruses can provoke cytotoxicity, and since both their production and handling are difficult, they tend to be expensive^{1, 2}. Therefore, non-viral delivery methods have recently attracted significant attention.

There are several alternatives to non-viral vectors, e.g. liposomes, polymer micelles and protein nanoparticles that have been utilized not only for gene delivery but also especially for administering therapeutical proteins, peptides and small molecules³. Most of the nanocarriers utilized in nanomedicine are organic materials which have inherent drawbacks such as a low loading degree and premature leakage of the cargo, instability under hydrodynamic and physiological conditions, difficulties in modifying their physicochemical properties in accordance with the cargo molecule, and sometimes also toxicity. Inorganic mesoporous materials are some of the newest candidates being evaluated as drug delivery platforms. They possess several positive features for efficient drug delivery, i.e. high loading capacity, biocompatibility, bioresorbability, the protection of the cargos against the harsh conditions in body, the ease with which one can modify particle surfaces and integrate several biofunctions into the carrier⁴. One unique feature of mesoporous materials is that the pore walls can be functionalized separately from the external particle surface. For example, this feature has many positive aspects, e.g. the enhanced loading capacity, one can have either sustained or triggered release, ready attachment of targeting moieties as well as incorporation of a specific coating to reduce opsonization and thus increase the residence time of the particles in blood circulation⁵. It is essential to be aware that when targeting the vector

3

to an organ such as a solid tumor, extravasation needs to take place. This, in turn, demands that the vector has a long systemic residence time so that the enhanced permeation and retention effect (EPR) will take place effectively. One way to achieve this property is to reduce the extent of opsonization as this slows the clearance of the vectors by the immune system.

In addition to their presence in the systemic circulation, another important aspect in the field of nanomedicine is the safety of the nanomaterial. Irrespective of the nanoparticle utilized as the vector, both the circulation and safety issues are addressed by coating the particles with polymers, peptides, proteins, lipids and/or cell membranes. The present review surveys the recent publications related to this “bio-coating”, especially concentrating on in vivo experiments (Table 1). Typically, even though the modifications introduced into the vectors work well in vitro, the main challenges are encountered when proceeding to animal experiments. The vector has to successfully overcome several biological barriers when administered intravenously; these represent the real obstacles to the development of efficient nanovectors.

2. Opsonization and immune response

The failure of nanoparticles to have a prolonged systemic circulation time is mainly attributed to a process called opsonization. Opsonization involves the absorption of opsonin proteins, which are abundantly present in the blood, onto the surface of nanoparticles.

Subsequently, the adsorbed proteins act as ligands and are responsible for nanoparticle recognition by phagocytic cells. Thus phagocytosis occurs; this is a process which results in the removal of the foreign objects from the bloodstream and since inorganic nanoparticles cannot be typically digested by phagocytes, they end up in the organs of the reticuloendothelial system (RES), more commonly in spleen and liver, but also in lungs⁶.

Opsonization is a complex process and it is not yet fully understood since it involves several blood components. In addition, there is extensive variability in the properties of the nanoparticles, such as their hydrophobicities, size, surface curvature and coating moieties that can affect the

4

opsonin adsorption. Furthermore, opsonization is a dynamic process i.e. the composition of the adsorbed proteins varies with time. A detailed analysis of the opsonized molecules has been rarely performed and reported. However, the most common opsonins are recognized i.e., components of complement system, immunoglobulins, fibronectin, laminin as well as many other serum proteins⁷. As the opsonization is progressing, phagocytes start to recognize the nanoparticles and become attached onto the proteins bound on the particle surface. Three primary mechanisms of attachment have been identified⁷; 1) non-specific attachment of the phagocytes to the adsorbed proteins, 2) the conformational transformation of the proteins from their inactive form in blood to their active counterpart on the surface of the nanoparticles followed by the recognition by the receptors on the cell membranes of phagocytes, and 3) the activation of the complement mechanism pathways. It is beyond the scope of the present review to examine these mechanisms in detail, however, regardless of the mechanism, the nanoparticles are typically effectively taken up by the phagocytes and rapidly removed from the bloodstream⁸.

The initial process of opsonization is the critical part of nanoparticle clearance and thus many different coatings have been utilized to reduce the extent of opsonin adsorption. Bare uncoated nanoparticles tend to have a charged surface which adsorbs proteins (Figure 1a). The hydrophobicity of the surface also correlates with rapid opsonization. Therefore, different surface coatings have been physically adsorbed or chemically grafted withnanoparticles, which improves their hydrophilicity and/or hinders their electrostatic interactions and consequently increases the systemic circulation time (Figure 1b). Unfortunately, the amount of opsonized proteins and their further identification have been rather rarely studied but these kinds of studies are crucial for gaining a deeper understanding of the interactions between nanoparticles and the blood environment. For example, Nissinen et al.⁹ demonstrated that not only did nanoparticles modified with polyethylene glycol (PEG) adsorb much less proteins than their bare counterparts but also that the composition of the proteins differed significantly, inducing a higher accumulation in spleen. Moreover, CD47 and CD59 glycoproteins were found in the corona of PEGylated

5

nanoparticles; these are known to prolong the blood circulation time¹⁰ . For example, these two glycoproteins are present in the envelope of viruses and they help the virus to avoid activating the complement system¹¹.

Figure 1. Illustration of the opsonization process of (a) uncoated and (b) coated nanoparticles. The opsonization process starts with the adsorption of opsonins, which are present in excess in blood, on the nanoparticle surfaces. The opsonins are subsequently recognized by the phagocytes and nanoparticles are cleared from the systemic circulation.

3. Biocompatibility/safety

The biocompatibility of nanoparticles is related to the immune response that they evoke after their administration as well as to the subsequent toxicity of the nanoparticles and/or their biodegradation products. IUPAC defines the term of biocompatibility for polymeric systems, which can also be expanded to inorganic nanomaterials, as “an ability to be in contact with a living

6

system without producing an adverse effect”. The definition has been further broadened for therapy as “an ability of a material to perform with an appropriate host response in a specific application”¹². In case of drug delivery with nanoparticles, the variations of the drug accumulation in different organs compared with the free drug, especially in the RES organs, should be carefully evaluated in order to avoid the increase of toxicity in healthy tissues.

In vitro biocompatibility studies include the evaluation of cell viability/proliferation, inflammation, nanoparticle-cell interaction mechanisms e.g. at different stages of the cell cycle as well as the elucidation of mechanisms of cell death. Assessing cell viability after incubation with the nanoparticles for a certain time can provide basic information about their biocompatibility; if the viability is above 80%, then the particles can be considered as nontoxic (ISO 10993-5:2009).

The measurement of cytokine expression levels can give further insights into the immune response of the cells to different nanoparticles. Cells from various tissues should be utilized in biocompatibility studies, since different tissues can display a significantly different response to the same nanoparticle. For example, poly(lactic-co-glycolic acid) (PLGA) nano- and microparticles generally induce a weak tissue reaction, but if they gain access into nerve connective tissue, they evoke rather strong acute inflammation¹³. Some endothelial barriers and even parts of organs can be modelled in vitro when evaluating nanoparticle biocompatibility before embarking on in vivo studies. However, nanoparticles are typically incubated with cells from several hours to several days, and if the particles are stable enough, the assessment of the long-term effects in vitro can be complicated. Moreover, in the case of therapeutic nanoparticles, it is essential to understand the drug internalization pathway since this is crucial for the drug’s performance.

Intravenously injected nanoparticles are commonly cleared from the bloodstream by macrophages and subsequently they end up in the RES organs, such as liver, spleen and kidney.

Therefore, a histopathological assessment of these organs is important in order to verify their tolerance towards the nanoparticles. Immunohistochemistry (IHC) can complement the histopathology and provide results on the expression levels of the specified molecules, or it can

7

visualize apoptotic and/or necrotic cells in tissue slices^{14, 15}. An assessment of blood biochemistry (BBC) can provide information about the levels of certain compounds, e.g., providing evidence of the induction of hepatic toxicity¹⁶.

4. Physicochemical modifications which improve biocompatibility and the fate in the systemic circulation

Various strategies, such as a silane coupling reaction, carbodiimide induced coupling of amine and carboxylic acid and thiol-maleimide (MAL) Michael addition reaction, have been utilized to chemically conjugate biopolymers, peptides and proteins onto the inorganic nanoparticle surfaces depending on their functional groups (Figure 2). A silane coupling reaction is frequently used to functionalize the oxide-surfaces of inorganic nanoparticles because they typically have many –OH surface groups. This approach is very common for silica-based surfaces.

The Si–OH surface groups can be condensated with the –Si–OR groups of silane to form the Si–

O–Si bond. Since the reaction generally takes place under harsh condition e.g. a strong basic aqueous environment or at a high reflux temperature in an organic solvent, the silane coupling reaction is mainly used for conjugation of biopolymers and not for proteins or peptides. On the other hand, the reactions of carbodiimide crosslinking and thiol-maleimide Michael addition are more universally applied to conjugate biomolecules due to their mild reaction conditions. With regard to those nanoparticles with –COOH surface groups (Figure 2b), they have to be activated with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) and N- hydroxysuccinimide (NHS) prior to undergoing a reaction with the primary amine (–NH2) terminated biopolymers, peptides or proteins. The reaction can be arranged also in the reverse manner, i.e., the corresponding activation of –COOH is also needed for the biomolecules to react with the nanoparticles with the –NH2 surface groups (Figure 2c). Even though the activation of – COOH with EDC is most efficient in acidic (~pH 4.5) conditions, increasing the concentration of EDC can compensate for the reduced reaction efficiency in a neutral phosphate buffer solution,

8

which is a more suitable condition for the conjugation of sensitive proteins and peptides. The thiol- maleimide Michael addition reaction is a robust, highly selective and tolerant strategy that has been widely implemented in biological functionalization. This reaction is often catalyzed by a tertiary amine such as triethylamine¹⁷. The reaction can also be undertaken without a catalyst in highly polar solvents such as dimethyl sulfoxide and dimethylformamide because the formation of active species, i.e., the thiolate ion, due to the polarity of the solvent¹⁸. The specific approaches related to the particular coatings intended to improve biocompatibility and prolong the systemic circulation time will be further discussed in the following sections.

Figure 2. Typical routes to chemically conjugate biopolymers, peptides and proteins on the surface of inorganic porous nanoparticles (NPs): (a) Silane coupling reaction. (b, c) Carbodiimide induced coupling of amine and carboxylic acid, (d) Thiol-maleimide Michael addition reaction.

4.1. Polymer coating

The majority of studies that utilize polymer coatings to modify the properties of the systemically administered nanoparticles are related to PEG. However PEG is not the only possibility; there are also several other polymers such as chitosan, dextran and polyethylenimine (PEI) that have been used, although not very frequently. PEGs form a large family of molecules with different molecular weights and termination groups facilitating various conjugation chemistries. Consequently, based on the different physicochemical properties of the molecules,

9

they can create nanoparticles with different efficiencies for stealthiness and biocompatibility resulting in altered behaviour of the coated nanoparticles in vivo.

4.1.1. Polyethylene glycol Silanol conjugates

The application of silanol chemistry is a fairly simple but powerful way to introduce PEG onto an inorganic porous particle surface. Usually PEG-silanes are utilized to covalently bond with silanol groups by hydrolytic coupling^19-21 or anhydrous deposition^{9, 22} but an esterification reaction between silanol and carboxylic group can be also exploited²³. Overall, these particles exhibit excellent in vivo biocompatibility according to the evaluations of their effects on hemolysis, BBC, hematology, bodyweight and histological changes. Liu et al.²³ reported that mesoporous silica nanoparticles (MSN) did induce red blood cell (RBC) hemolysis but the MSN-PEG2k nanoparticles did not cause any hemolysis even at high doses up to 900 μg/ml. Jin et al.¹⁹ detected high blood concentrations of the PEG0.6k coated tantalum oxide nanoparticles (40 nm) loaded with doxorubicin (DOX) 1 h after administration, resulting in an excellent therapeutic response but inducing a significant decrease in body weight indicative of toxicity. Nissinen et al.⁹ reported that simultaneous conjugation of two different sized PEG0.5,2k molecules shifted the main accumulation site of porous silicon (PSi) from liver to spleen while increasing the blood half-life from 1 min to 241 min for 170-180 nm sized particles. However, He et al.²² reported that the grafting of PEG10k

on MSNs (80, 120, 200 and 360 nm) did not have any effect on the main accumulation site (spleen) but generally decreased the accumulation in liver, spleen and lungs. Furthermore, the uptake by liver and spleen points to an increasing trend with increasing particle size, with the exception of big MSN- PEG10k (360 nm) where spleen accumulation was clearly lower than in other groups at 30 min post-injection (p.i.). The particle blood half-life was also dependent on the size of the nanoparticle with small particles having the longest residence time in blood which was further improved with PEG10k22. The prolonged systemic circulation enables passive targeting and tumor accumulation which can be enhanced by incorporating active targeting. Wang et al.²⁰ reported that

10

by applying folic acid (FA) and a magnetic field, it was possible to increase the tumor accumulation of Fe3O4@MgSiO3-PEG2k (110 nm) from 5.6% to 52.8% (of total detected fluorescence intensity) at 4 h p.i. Furthermore, Zhang et al.²¹ described size dependent tumor accumulation where approx. 23% injected dose/gram of tissue (% ID/g) was reached with the MSN- PEG0.5,5k-FA nanoparticles of ~ 70 nm in size, while only a 2.5% ID/g was achieved with the corresponding particles of ~ 140 nm in size 48 h p.i.

Amine conjugates

One common way to conjugate PEG to inorganic nanoparticles is to connect the amine (–

NH2) groups with carboxylic (–COOH), aldehyde (-CHO) and methylsulfonyl (-OM) groups.

Typically, the carboxyl group needs to be activated with EDC/NHS to enhance crosslinking with the amine group. The amine groups can be linked to the particles while the counterpart group is linked to PEG, or vice versa (Figure 2b&c). In general, the particles which have PEG attached through amine group exhibit excellent biocompatibility but any further modifications can affect the situation dramatically. For example, Chen et al.²⁴ discovered that the covalent conjugation of Cys-TAT peptide by linking -SH group with MAL group at the distal end of PEG3.4k on mesoporous silica nanoparticles evoked significant differences in several blood biochemical and hematological parameters i.e. evidence of toxicity while the plain MSN-PEG3.4k particles did not induce any noticeable effect. Furthermore, Huang et al.²⁵ studied mesoporous silica (MS) rods conjugated with PEG5k through the amine-NHS reaction and evaluated their hematological and biochemical parameters at 24 h and 18 d p.i. The results obtained with MS-PEG5k revealed more significant differences at 18 d time point indicating possible chronic toxicity.

The particles should possess a prolonged circulation time; this is a prerequisite for their effective accumulation in tumors through the EPR effect. For example, mesoporous carbon PEG6k

nanocapsules 180 nm in size had a blood residence time of over 24 h after their administration²⁶. However, the specific value of the blood half-life is more informative; a half-life value as high as 4.6 h has been reported for ~ 220 nm MSN-PEG5k27. The long circulation time usually corresponds

11

well with passive tumor accumulation; the longer the circulation time, the more particles tend to end up in the tumor. Indeed, significantly higher passive accumulation into tumors has been reported for the MSN-PEG2k&1k nanoparticles of diameter 190-220 nm in comparison with the bare MSN (~ 2-fold)²⁸ or the free drug (~ 4-fold)²⁹. Furthermore, in some cases, tumor accumulation has been shown to be time dependent and to occur over a long time span. With PEGylated quantum dots (CdSe/ZnS, 13 nm in size), the maximum accumulation in tumor was found at 12 h after administration³⁰, and furthermore in the case of the PEG5k coated hollow tantalum oxide nanoparticles (124 nm in size), the accumulation still proceeded from 12 h to 24 h post-injection³¹. The tumor accumulation can be further improved with active targeting and significant increases (up to 4-fold) have been obtained with magnetic field^{32, 33}, aptamer³⁴ or antibody²⁷ based targeting.

As a result, the nanoparticles conjugated with PEG through amine chemistries have achieved a significantly better therapeutic response after intravenous (i.v.) administration compared with the free drugs^{29, 33, 34}.

An assessment of the biodistribution of the nanoparticles upon systemic administration has been used to evaluate the success of the surface modifications in improving the biocompatibility and RES avoidance. Compared with plain MS rods of ~ 120 x 185 nm in size, PEG5k conjugated MS rods accumulated less in liver (~ 7% ID/g difference) and spleen (~ 8% ID/g difference), and higher quantities were observed in the blood (~ 2% ID/g difference) 2 h p.i.²⁵. After 7 d, more MS- PEG5k than MS rods were found in liver (~ 12% ID/g difference) and spleen (~ 12% ID/g difference), as evidence of a delayed accumulation of MS-PEG5k into these organs. In the study of Kramer et al.³⁵ nearly all MSN were located in lungs, liver, spleen and kidneys 3 h p.i. while the MSN-PEG5k were more evenly distributed, with a significant amount still in the blood at 12 h after administration (Figure 3a,b). On the other hand, the size of the nanocarrier has also an evident effect on the biodistribution. Kim et al.³⁶ studied MSN- PEG0.5k nanoparticles with three different sizes, and found that the particles larger than 300 nm were located mainly in spleen, while the 100- 150 nm particles were preferentially located in the tumor 24 h p.i. Furthermore, the smallest

12

particles (40 nm) were found primarily in the stomach and they had the lowest accumulation in the tumor even though a small particle size is generally considered preferable for tumor accumulation.

Thiol-linked coating

It is possible to conjugate thiols (-SH) covalently via a Michael addition reaction with polydopamine (PDA)³⁷ and maleimide (MAL)^{38, 39} (Figure 2d) or by a coordination interaction with gold^40-42, in order to add PEG onto the surfaces of the nanoparticle. The in vivo biocompatibility of the nanoparticles conjugated to PEG with thiol chemistry is excellent since no significant differences in body weight of animals treated with these particles have been reported nor did the histopathological analysis reveal any signs of toxicity. Real-time imaging in vivo has been rather rarely reported but this approach does provide valuable information about the particle biodistribution. Chen et al.³⁸ evaluated the biodistribution of MSN-PEG nanoparticles (170 nm in size) with and without antibody targeting. In that study, PET/CT monitoring at several time points up to 48 h p.i revealed that the targeting increased the tumor uptake significantly (~ 2-fold).

Furthermore, Peng et al.⁴² prepared Fe3O4@Au-PEG nanoclusters (150 nm in size) with and without magnetic targeting and indicated that one could achieve a higher dose inside the tumor with PEGylation as revealed by fluorescent imaging 4 h after administration. On the other hand, ex vivo analysis revealed strong accumulation of MSN-PEG5k (170 nm) in liver at 48 h p.i.³⁸ and MSN@Au-PEG5k (200 nm) in spleen 24 h p.i.⁴¹ as an evidence that these organs were still capturing most of the particles. However, the ex vivo analysis reported by Cheng et al.³⁷ detected a strong fluorescent signal of MSN@PDA-PEG (190 nm) originating from the tumor at 24 h after administration. Qiao et al.³⁹ reported that UCNP@MS-PEG (60 nm) particles were present in the tumor at high quantity, in fact the levels were only surpassed by liver uptake at 24 h p.i as based on the yttrium concentration as evaluated with ICP-MS. The high tumor uptake could be due to the prolonged circulation time (T1/2 = 147.24 min) enabling passive targeting³⁹. Furthermore, Elbialy et al.⁴⁰ reported a time dependent delivery of DOX in tumor with MSN@Au-PEG5k

13

(170 nm), reaching its highest value at 24 h p.i. but still remaining at a high level at 72 h after administration.

Cyclodextrin-linked coating

β-Cyclodextrin (CD) has been utilized for PEG attachment with either a host-guest interaction to adamantine (AD)^43-46 or covalent linking to isocyanate (-NCO)^{47, 48}. As well as linking the nanoparticle with PEG, the CD has been used as a gatekeeper to control the release of the cargo molecules. Typically, CD is covalently attached to nanoparticles through a redox (e.g.

platinum (IV) prodrug or disulfide) sensitive bond which becomes cleaved in the tumor environment (e.g. due to glutathione or pH) to release the payload. The examined formulations showed good in vivo biocompatibility in terms of body weight, blood biochemical and histological changes. Liu et al.⁴⁵ used healthy mice and reported that the PEG2k coating decreased the MSN particle (120 nm) amount in lungs, liver and spleen at 1 d after administration as evidence of avoidance of RES uptake. Lee et al.⁴⁷ exploited magnetic resonance imaging (MRI) and noted that magnetic core MSN-PEG5k particles (~ 80 nm) could passively accumulate in the tumor site, reaching a maximum value at one week after administration. Furthermore, active targeting realized with FA⁴⁴, lactobionic acid⁴⁶ or Arg-Gly-Asp (RDG) targeting peptide together with magnetic targeting⁴³ resulted in high tumor uptake which reached 55 ID%/g dose in the case of the MS- coated gold nanorods with an outermost PEG5k layer and 70 nm particle size⁴⁶; this approach also decreased the amount of particles that ended up in liver and spleen (Figure 3c). Moreover, Lee et al.⁴⁸ described the production of MSN-PEG2k (90 nm) nanoparticles that were susceptible to breakdown by cytosolic reductase (NQO1) and thus capable of triggered DOX release. These nanoparticles were able to reduce the unwanted drug accumulation in organs other than in the tumor when compared with the free drug and could release DOX in the tumor which possessed the NQO1 enzyme in order to induce a therapeutic effect.

Surface adsorption

14

In addition to the coating approaches presented above as ways of achieving surface PEGylation, other surface modification methods have also been utilized. In general, PEGylation based on physisorption yields coated particles that are biocompatible and safe in vivo. However, a significant elevation of the liver enzyme AST with a high dose of PEG5k coated mesoporous anastase (TiO2) (400 mg kg^-1, particle size of 120 nm) was observed and interpreted as a sign of liver injury⁴⁹. In addition, elevated TNF-α levels have been reported after administration of coated PSi nanoparticles (300 nm)⁵⁰ suggesting the induction of inflammatory processes. Rytkönen et al.⁵¹ described the rapid clearance and accumulation in liver 10 min p.i. (> 60% ID/g) of PSi- PEG10k (150 nm) suggestive of a poor lifetime in the blood circulation. However, Beavers et al.⁵⁰ stated that their PSi-PEG5k (300 nm) particles had a blood half-life of 70 min, a value that was clearly higher than present in particles without PEGylation (30 min) (Figure 3d). Moreover, less PSi-PEG5k accumulated in lungs (~ 10% difference) and more in spleen (~ 15% difference) than with bare PSi, although in both cases, the liver captured most of the particles⁵⁰. Furthermore, Zhang et al.⁵² reported a high accumulation of PSi/carbon-PEG5k (240 nm) in tumor (~ 20% ID/g) 2 h p.i.

which they attributed to a prolonged circulation time and an enhanced EPR effect.

Overall, the PEG molecules of sizes between 0.5 kDa and 10 kDa has been used in coating several types of nanoparticles. Mainly all PEGs have methyl (-CH3) group at the distal end with a few exceptions where other molecules have been added e.g. to facilitate imaging or targeting.

However, there is a real lack of thorough characterization of the used coating, and critical features such as grafting density or amount of added polymer are very rarely reported.

15

Figure 3. (a) Biodistribution of ⁸⁹Zr tagged MSN. (b) Biodistribution of ⁸⁹Zr tagged PEG-MSN5k

at several time points p.i. evaluated via radioactivity (γ-counter). The PEGylation is seen to prolong the blood residence time resulting in a greater spread and more evenly distributed particles in the test animal. Modified from³⁵ with permission from Royal Society of Chemistry. (c) Biodistribution of magnetic MSN (MMSN), MMSN with PEG0.5k-RGD (targeting peptide) and MMSN- PEG0.5k-RGD with magnetic targeting (+M). PEGylated MSN show decreased uptake in liver and spleen but with improved tumor uptake due to PEGylation and active targeting. Modified from⁴³ with permission from Elsevier. (d) Blood circulation time of Cy5-labelled PNA (T½ = 0.9 min), PNA loaded porous silicon (PSNP) (220 nm) (T½ = 30 min) and PNA loaded PEG0.5k coated porous silicon (Full comp) (300 nm) (T½ = 70 min). Half-life is evaluated from the Cy5-labeled PNA content in the blood. Modified from⁵⁰ with permission from Wiley.

16 4.1.2. Other polymers

Various other polymers in addition to PEG have been introduced onto the surface of the nanocarriers through physisorption, polymerization and covalent conjugation. These polymer coated nanoparticles exhibit good biocompatibility and thus according to numerous reports, they are similar to PEGylated nanoparticles. MSN coated with alginate/chitosan with an electrostatic layer-by-layer technique (two bilayers) did not induce RBC hemolysis, while non-coated MSN showed concentration dependent hemolysis i.e. the coating modification had achieved a major improvement in the particles’ biocompatibility⁵³. Dong et al.⁵⁴ produced bamboo charcoal nanoparticles physically coated with Tween 80 and examined their effect on blood biochemical parameters for 30 d. Even though no significant changes were reported, the levels of CREA, BUN, AST and ALT did reveal some time dependent elevations possibly indicative of a long-term toxic effect. On other hand, Cai et al.⁵⁵ evaluated blood biochemical and histological effects 50 days after the administration of polyvinylpyrrolidone coated Prussian blue particles without detecting any signs of chronic toxicity. The blood half-life after polymerization of PNIPAM-co-MAA on MSN (190 nm) was reported to increase from ~ 11 h to 17 h⁵⁶ while the physisorpted dextran20k

increased the PSi (150 nm) circulation time from ~ 28 min to 82 min⁵⁷ indicating these polymers did possess a good capability to increase stealthiness. The effect was comparable with the effect obtained with PEG enabling an improved EPR effect.

In several cases, fluorescent imaging has been utilized to evaluate biodistribution and tumor accumulation in animal models. The accumulation of covalently coated MSN-CD- PEI1.8k

nanoparticles within the tumor was second highest (~ 14%) after liver uptake, which trapped more than 50% of particles at 24 h p.i. based DOX fluorescence loaded into the particles⁵⁸. Yang et al.

prepared hollow MSN with amphiphilic capping consisted of a hydrophilic group with FA targeting conjugated through a pH linkage to a hydrophobic group which was connected via a hydrophobic interaction to the particles. These particles (160 nm) showed excellent tumor accumulation i.e. the tumor was the main repository site at 24 h after administration and the uptake

17

was about four fold higher than obtained with the MSN without a coating⁵⁹. Park et al.

demonstrated that physically absorbed dextran20k PSi nanoparticles (150 nm) could remain in the tumor for up to 24 h although the signal reached the maximum intensity at 2 h p.i.⁵⁷ while Shi et al. reported increased accumulation for up to 48 h with mesoporous titanium oxide (MTN) nanoparticles (190 nm) capped with reactive oxygen (ROS) linker modified AD/CD. In that experiment, the ROS linker was covalently conjugated to the surface of MNT and AD which had a host-guest interaction with CD⁶⁰. Moreover, Hu et al. reported that more MSN polymerized with PDA together with NGR peptide targeting (170 nm) accumulated in the glioma region than in a healthy brain section; it was also better than non-targeted particles (2-fold difference)⁶¹.

Good tumor accumulation enables effective therapy and evokes a significantly improved therapeutic response compared to free drug; this was achieved with PEI60k conjugated to carbon nanotubes (150 nm)⁶² with EDC/NHS chemistry, covalently conjugated chitosan-FA through EDC/NHS chemistry to MSN (200 nm)⁶³ and pillar[5]arene coated via a host-guest interaction to a hollow MSN (220 nm)⁶⁴ and an ultrasound responsive mesoporous titanium oxide (190 nm)⁶⁰. Furthermore, Shi et al.⁶⁰ evaluated a time dependent drug concentration in tumor for 20 h p.i.; the detected drug concentration was low until the 12 h time point when focused ultrasound (US) (1 W/cm³, 40 s) was utilized to detach the CD capping and release the drug. At 20 h, the drug concentration in tumor was still more than three times higher than the situation before US irradiation.

4.2. Peptide coating

The primary objective of peptide coating is to target the nanoparticles and improve their internalization into tumors^65-68, organs⁶⁹ and cross the blood-brain barrier⁷⁰, although some peptides may also function as antibacterial agents⁷¹.

Functionalization of the nanoparticles with peptides can be achieved through adsorption on the surface of hydrophobic nanoparticles⁷² or electrostatic attraction⁷¹, although most frequently, the peptides have been chemically conjugated through amine-carboxyl acid coupling

18

chemistry⁷³(Figure 2b,c). This modification involves the activation of the carboxylic group in the peptide^{65, 67} by the carbodiimide compound EDC and its subsequent conjugation to the amine terminated surface of nanoparticles in the presence of NHS. If the nanoparticles contain carboxylic groups⁶⁸ or they are modified with carboxylic acid, such as undecylenic⁶⁹ or hyaluronic acids⁶⁶, then their carboxylic group can be activated with EDC and reacted with the amine group of the peptide.

Peptide coating of the nanoparticles has generally not affected the cell viability compared to the uncoated nanoparticles^{65, 68, 72}. However, at high concentrations of 200-300 µg/ml, there may be elevated toxicity attributable to the targeting function of the peptide coating and its subsequent higher cell internalization^{69, 70}. In one case, the peptide coating did not increase the systemic circulation time significantly⁶⁶, whereas in another trial, the coating increased the nanoparticles’

lifetime in the blood from several minutes to 8 h and shifted the accumulation of the drug from lung and kidney (in the case of free drug) to liver (drug loaded inside the nanoparticles)⁷⁰ (Figure 4a,b). This result could indicate that there had been extensive nanoparticle accumulation in liver, where the drug was released from the carrier and subsequently escaped to the systemic circulation. The measurement of the Si content in organs of similar nanoparticles modified with the same targeting peptide supports the hypothesis that the particles are subjected to extensive liver accumulation⁶⁵. On the other hand, a BBC analysis of animals injected with free drug or with drug encapsulated in nanoparticles revealed the restoration of blood enzymes, red blood cells and hemoglobin levels to the control values, this was interpreted as evidence of good biocompatibility of the nanoparticles^{70, 71}.

19

Figure 4. (a) The concentration of the drug BSeC (benzo[1,2,5]selenadiazole-5-carboxylic acid) in the blood of the free drug and of the drug loaded in RGD modified MSNs after intravenous administration. (b) The biodistribution of BSeC. Modified from⁷⁰ with permission from Royal Society of Chemistry.

4.3. Protein coating

The protein coatings of porous nanoparticles can be divided into two major groups: coating with albumin (bovine serum or human serum albumin) and gelatin. While the primary purpose of these coatings is targeting^{74, 75} or controlling the drug release either through the degree of crosslinking⁷⁶ or response to enzymes^{77, 78}, there is also one report where albumin was used to prolong the blood circulation time⁷⁹.

The most fundamental way to coat nanoparticles with proteins is adsorption. The negatively charged albumin molecules can be attracted to the positively charged surface of amine modified silica and silicon nanoparticles due to electrostatic interactions^{74, 80} (Figure 5a). Albumin can also be adsorbed onto the surface of hydrophobic nanoparticles via a hydrophobic interaction and this can improve their stability under physiological conditions^{74, 79}. In order to obtain the most stable coating, albumin can be chemically conjugated with the above-mentioned EDC/NHS chemistry to the amine⁸¹ and carbocyclic groups⁸² located on the surface of the nanoparticles (Figure 5a).

Gelatin is typically stabilized onto the nanoparticles by crosslinking with glutaraldehyde (Figure 5b). The first approach involves nanoparticles dispersed in aqueous solution; these act as

20

the growing seeds for the polypeptide chains by inducing the gelation at an elevated temperature and subsequently performing the crosslinking^{76, 77}. Alternatively, glutaraldehyde can be first conjugated to the amine-modified surface of nanoparticles and allowed to react with gelatin to form a protein corona⁷⁸. The PEG-amine can be further conjugated to the carboxylic groups of gelatin via EDC/NHS chemistry⁷⁷. Here, the amount of glutaraldehyde can be adjusted to vary the degree of crosslinking and in this way one can modify the drug release profile⁷⁶.

Figure 5. Albumin and gelatin coating of nanoparticles. (a) Routes of synthesis of albumin coated nanoparticles. (b) Routes of synthesis of gelatin coated nanoparticles. (c) Blood circulation time of bare PSiNPs (black), undecylenic acid modified UA-PSiNPs (red) and BSA coated alkyl terminated BSA/D-PSiNPs (blue) determined by Si content from ICP-MS measurements⁷⁹. (d, e) The biodistribution of FITC labeled gelatin coated magnetic mesoporous silica nanoparticles FITC-MMSN@GEL at 2 and 4 hours after the i.v. injection, respectively. MF^- and MF⁺ indicate

21

the presence or absence of a targeting magnetic field⁷⁶. (c) Modified from⁷⁹ and (d, e) from⁷⁶ with permission from American Chemical Society and Elsevier, respectively.

Protein coating has been performed for porous silica or silicon nanoparticles; since these nanoparticles are known to be biocompatible on their own, it was not surprising that no difference in cell viability was observed between the uncoated and coated nanoparticles at the studied concentrations ≤ 200 µg/ml. Furthermore, albumin, coated via a hydrophobic interaction, was able to improve the systemic circulation half-life from ~ 30 min to ~ 3.5 hours⁷⁹ (Figure 5c) and significantly inhibited the tumor growth by allowing a more specific delivery of the encapsulated drugs; this benefit was due to both more efficient accumulation and enzyme-triggered release^81-83. A gelatin coating has also been shown to improve cancer treatment by efficiently inhibiting premature drug leakage from the pores; in this case, the release was controlled due to the digestion of matrix metalloprotease-2 in the cancer tissue^{77, 78}. When combined with iron oxide, the magnetic targeting of gelatin coated silica nanoparticles enhanced their accumulation by 3-fold in the tumor at 2 and 4 hours after their i.v. administration⁷⁶ (Figure 5d,e). Furthermore, specific proteins can also act as nanovectors and these can introduce some adverse features such as facilitating a specific organ uptake. For example, on one hand, while HFBII coating of porous silicon nanoparticle reduced the extent of opsonization, on the other hand it also reduced the blood circulation time and increased the accumulation in the liver and kidney ⁷⁴.

4.4. Lipid coating

The coating of porous nanoparticles with lipids is considered to resemble the lipid membrane of living cells and therefore to reduce the extent of opsonization and thus improve the biocompatibility and systemic circulation time. There are two different ways to coat nanoparticles with lipids i.e., coating with a monolayer or a bilayer. In the first approach, the monolayer of phospholipids is physically adsorbed onto a hydrophobic nanoparticle surface. Particles coated

22

with monolayer of PEGylated phospholipids had a prolonged circulation time i.e., for the bare particles, the circulation time was 20 min but this was extended to 3 hours for the lipid monolayer coated nanoparticles⁸⁴.

Coating with a lipid bilayer is more common and several methods have been developed to achieve better uniformity and more complete encapsulation of nanoparticles. In the thin film dispersion hydration method^{85, 86} (Figure 6a), lipids are first dissolved in an organic solvent such as chloroform, which is then evaporated in a rotary evaporator to form a lipid film on the wall of a round-bottom flask. Subsequently, the nanoparticles in aqueous suspension are added to the flask and vortexed or sonicated to coat the nanoparticles with a lipid bilayer. The composition of the lipid film can be optimized to achieve rapid and uniform particle coating^{14, 15}.

23

Figure 6. Lipid bilayer coating methods. (a) Thin film hydration method. (b) Liposome infusion.

(c) Chemical conjugation of the first lipid layer and subsequent self-assembly of the outer leaflet of the lipid bilayer.

The second approach is to infuse nanoparticles into existing liposomes (Figure 6b).

Typically, the infusion is achieved by mixing together the nanoparticles and liposomes followed by sonication^{87, 88} and finally by the removal of the empty liposomes by centrifugation.

There is also a third approach; in this technique, the inner lipid leaflet was first stabilized on the particle surface via a chemical conjugation to the amine-terminated surface of nanoparticles⁸⁹ (Figure 6c). In this method, the primary amine groups of the preliminarily conjugated hyperbranched PEI on nanoparticle surface were coupled with activated amino groups of DOPE lipid⁹⁰. In the final step, the outer layer of the bilayer was self-assembled, for example, with a dual solvent exchange method.

The lipid coatings have achieved only a slight improvement in terms of short-term cytotoxicity due to high initial biocompatibility of silica nanoparticles and the rather low studied concentrations, which have generally been less than 300 μg/ml. For example, no difference in cell viability was observed between the lipid coated and the uncoated carboxyl terminated hollow mesoporous silica nanoparticles at concentrations ≤ 300 μg/ml incubated with HUVEC, B16-F10 and A875 cell lines⁹¹. In another study, the lipid coating slightly improved the biocompatibility from 60% to 75% of viable cells after incubation at a concentration of 100 μg/ml⁸⁵. The improvement of the viability was attributed to the shielding of the –NH2 groups on the particle surfaces. Nonetheless, in vitro hemolysis studies revealed a significant reduction in hemolysis i.e.

almost no hemolysis was observed for lipid bilayer coated nanoparticles at a concentration of 500 μg/ml, whereas 85% of red blood cells were lysed when incubated with this level of bare nanoparticles⁸⁶ (Figure 7b). The BBC analysis demonstrated a substantial decrease of AST and ALT levels in blood compared to the free DOX, indicative of a reduction in hepatic toxicity⁸⁸

24

Moreover, Liu et al.¹⁵ observed less liver necrosis due to the lipid coated nanoparticles as compared to the liposome based drug formulation. The authors explained this effect as being attributable to the lower stability of the liposomes, which release a substantial amount of the drug molecules when broken down by the Kupffer cells. On the contrary, the nanoparticle based liposomes were not destroyed and they released the drug more slowly. Overall, the lipid bilayers in all cases consisted of cholesterol (2.5 – 11 mol-%) and DSPE-PEG2k and/or DSPE-PEG2k-(targeting moiety) (20 – 35 mol-%), while the other phospholipid molecules were different, such as DSPC^14,

15, 85, 91, DPPC⁸⁶, DOTAP⁸⁸. Although the zeta potential after the coating varied in wide range from

−27 mV to 30 mV and the size was comparable (150−250 nm), no significant difference in the biocompatibility of different compositions were observed.

Figure 7. Lipid coated Fe3O4@mSiO2 (FMLM) loaded with DOX and ZnPc prepared by a thin film dispersion hydration method. (a) Schematic representation of the nanoparticles. (b) In vitro hemolysis studies. (d) Time dependency of the DOX concentration in animal plasma. FM denotes Fe3O4@mSiO2, FML refers to the lipid bilayer coated FM; FMLM represents the MTX

25

macromolecular prodrug conjugated FML for targeting. Modified from⁸⁶ with permission from Elsevier.

In many studies, the systemic circulation time has been investigated indirectly by measuring the drug concentration in blood as a function of time. Therefore, in this kind of set-up, the comparison is being made between the free drug formulation and the drug encapsulated in the nanoparticles. In several publications, the systemic residence time of the drug in blood has been significantly enhanced from 20-30 min to 2 h⁸⁶(Figure 7c), 5 h⁸⁵, 12 h¹⁵ and 24 h⁹² when the drug was administered as nanoparticles. The extent of the variability can be explained by the quality of the lipid layer and its composition, since the coatings can incorporate different amounts of phospholipid-PEG molecules. However, although the systemic circulation time of the drug was increased, this does not necessary correspond with a prolonged existence of the nanoparticles in the circulation. The biodistribution assessment revealed that when compared with the free drug formulation, it seemed that a significantly higher amount of drugs loaded into nanoparticles had accumulated inside different organs such as liver, spleen and kidney⁸⁵ due to the RES clearance.

Thus, the drug released from the nanoparticles can return from their entrapment in these organs back to the blood, whereas free drug is quickly and efficiently cleared by the renal system. On the other hand, ICP-MS and luminescent measurements revealed that the half-life of silica nanoparticles coated with a lipid monolayer had increased from tens of minutes to 3 hours⁸⁴.

4.5. Cell membrane coating

The synthetic or out-of-body nature of the inorganic nanovectors is an inherent barrier for their biomedical applications because the immune system is so efficient at eliminating any foreign constituents and thus protecting the function of the internal organs. This concerns all of the coating approaches discussed above irrespective of whether they comprise biopolymers, peptides, proteins or lipids. For example, even though surface PEGylation has been proven to be an effective way to

26

improve the biocompatibility of inorganic nanoparticles, an anti-PEG immunological response was observed on the nanoparticles functionalized with PEG, resulting in an ‘accelerated blood clearance phenomenon’⁹³. To address the issue, a new biomimetic strategy was developed by Zhang’s group in 2011 via coating of the PLGA nanoparticles with RBC membranes⁹⁴. As the PLGA nanoparticles were covered with the endogenous membranes of RBCs, they could be disguised as nanosized ‘RBCs’ and thus had a superior blood retention over the PEG- functionalized nanoparticles. The elimination half-life in blood was 39.6 h for the membrane- coated nanoparticles compared to 15.8 h for the PEG-coated nanoparticles while the bare PLGA nanoparticles were detected in the blood only transiently after 2 min due to their rapid elimination by the immune system. Inspired by Zhang’s study⁹⁴, many types of core-shell like cell membranes coated inorganic nanoparticles have been prepared by applying different cell membranes to coat a variety of inorganic porous nanoparticles. The cell membranes are typically produced by emptying their intracellular contents using a combination of hypotonic lysing, membrane disruption, and differential centrifugation. Subsequently, the cell membranes are coated onto inorganic nanoparticles by means of co-extrusion, sonication or electroporation (Figure 8a)⁹⁵. Generally, the cell membrane coated nanoparticles are perfectly biocompatible due to the endogenous nature of the coating and thus they trigger a significantly reduced immunological response than synthetic nanoparticles, e.g., those with a PEG coating.

The application of RBC membrane as an alternative coating to biopolymers such as PEG and poly(vinylpyrrolidone) (PVP) has been investigated in detail^{94, 96, 97}. RBCs are inherently long- lived circulating cells, having a systemic circulation half-life of up to 120 days⁹⁵. The ‘self-marker’

proteins (e.g. CD47), glycans and acidic sialyl moieties on their membranes actively suppress their immune elimination by macrophages⁹⁶. Therefore, coating the nanoparticles with RBC membranes should allow them to have a good biocompatibility and prolonged circulation. Experiments have been conducted to verify the unique advantages of the RBC membrane coated nanoparticles over PEG-coated nanoparticles; Rao et al.⁹⁸ coated Fe3O4 nanoparticles with RBC membranes and PEG,

27

and studied their behavior in vivo. As demonstrated previously, the PEGylated Fe3O4 nanoparticles exhibited the phenomenon of ‘accelerated blood clearance’ (Figure 8b-d). The iron content of blood was drastically reduced from 5.5% to 1.8% ID g^–1 after intravenous injection of the second dose of the PEGylated nanoparticles. Correspondingly, a higher amount of the PEGylated nanoparticles was detected in liver and spleen than after the first injection. However, the second dose of RBC membrane coated Fe3O4 nanoparticles displayed a similar pharmacokinetic behavior as the first injection. The RBC membrane coated nanoparticles possessed a significantly enhanced systematic circulation time. The blood retention values of the RBC membrane coated nanoparticles and the PEGylated nanoparticles were around 14.2 and 5.8 ID g^–1 at 24 h after the injection, respectively. The administration of RBC membrane coated nanoparticles did not trigger an immune response at either the cellular level or the humoral level. As well as improving the biocompatibility and systemic circulation time, the RBC membranes have been found to function as gate-keepers to decrease the premature drug release when they have been coated on mesoporous silica nanoparticles⁹⁹. The anticancer drug doxorubicin and chlorin e6 were co-loaded into nanoparticles and a laser-triggered drug release was achieved since the generated ROS from chlorin e6 destroyed the RBC membrane, leading to a rapid drug release⁹⁹. Moreover, the RBC membranes displayed other advantages over synthetic biopolymers like PVP i.e. by improving the systemic circulation time of the nanoparticles⁹⁶, which is an alternative to PEG as a way of enhancing the biocompatibility of nanoparticles. The circulation of so-called nanocages was monitored by analyzing the Au concentration in blood with ICP-MS after the intravenous injection of gold nanocages coated with RBC membranes. The RBC-membrane-coated nanocages exhibited a superior circulation half-life (9.5 h) compared with their PVP-stabilized counterparts (1.0 h).

These studies highlighted the promising potential of utilizing RBC membranes as an alternative to PEG. One aspect complicating their utilization is related to the different surface antigens present in the different blood types (i.e. the ABO blood group system in humans). Thus, it is extremely

28

important to use RBC membranes from the individual to be treated or at least with a compatible blood group.

Figure 8. (a) The methods to prepare cell membrane coated NPs and (b) pharmacokinetic studies to evaluate the ‘accelerated blood clearance’ phenomenon of RBC membrane coated Fe3O4

nanoparticles (Fe3O4@RBC) (b-d): (b) bare Fe3O4, (c) PEGylated Fe3O4 (Fe3O4@PEG), and (d) Fe3O4@RBC. The nanoparticles were injected intravenously into mice. The inset shows the retention of the nanoparticles in blood at 24 h after the first (red bar) and second injection (blue bar). Figures b-d are reproduced with permission from Wiley-VCH from reference⁹⁸.

In addition to RBCs, other cells such as leukocytes and macrophages have also been exploited as the source cells for membranes to coat the inorganic nanoparticles. Since there are no specific targeting ligands on the surface of the RBC membranes, the RBC membrane coated nanoparticles lack the property to actively target cancer cells. Thus, utilizing cell membranes from other cell types may not only improve the systematic circulation time but also create nanoparticles with the added function of active targeting. Leukocytes circulate in the bloodstream similarly as RBC, but they can be attracted to inflammation sites or to the tumor microenvironment, introducing the possibility to target the nanoparticles to lesion sites¹⁰⁰. Tasciotti’s group¹⁰¹

29

successfully prepared mesoporous silicon nanoparticles coated with leukocyte membranes via electrostatic attraction. The leukocyte membrane coating was found to protect the particle from opsonization and non-specific clearance. Significantly less (approximately tenfold decrease) protein absorption was observed on the membrane coated particles compared with the bare mesoporous silicon particles. Furthermore, the membrane-coated nanoparticles appeared to be sufficiently stable in suspension, and the modification enhanced the particle circulation time and improved tumor accumulation.

The membranes of macrophages represent another interesting approach for coating of inorganic nanocarriers^{102, 103}. Due to their lipid bilayer structure and the surface proteins of macrophages, the membrane coated mesoporous silica nanoparticles could be disguised as nanosized ‘macrophages’. Thus, the coating of the nanoparticles significantly decreased their phagocytosis and clearance. The bare nanoparticles had been phagocytosed almost completely after 24 h of incubation, whereas more than 30% of the membrane coated nanoparticles were still in the blood.

It is well documented that the cancer cells are capable of undergoing “immune escape” via the mechanisms of immune tolerance, immuno-suppression, and immunosenescence^104-106. For example, the membrane protein of CD47 is also present on the membrane of cancer cells, which is responsible for the immune tolerance and preventing macrophage uptake. Thus, the nanoparticles coated with cancer cell membranes could confer along with enhanced biocompatibility, decreased macrophage uptake, prolonged systemic circulation time and homotypic tumor targeting^106-109. It has also been noted that the immune evasion ability of the cancer cell membrane coated particles is comparable to RBC membrane coated particles¹⁰⁹. As living cancer cells have a high risk of causing cancer, it will be essential to conduct the toxicity studies to ensure the safety of these kinds of nanoparticles. It seemed that the cancer cell membrane coated upconversion nanoparticles evoked neither death nor any significant difference in body weight over 30 days in the treated groups. Furthermore, no noticeable signs of major organ damage

30

were detected nor were there any significant differences in the blood parameters and blood biochemistry indicators¹⁰⁹. These results demonstrated that the cancer cell membrane coated nanoparticles displayed good in vivo biocompatibility.

5. Conclusions

There has been a steady increase in the numbers of published articles dealing with intravenously administered nanocarriers. The ideal delivery system should transport the right dose of the right drug to the right place at the right time which after the carrier itself should be eliminated in a safe way. However, a recent review undertook a multivariate analysis of the compiled data gathered from 224 reports published during the previous 10 years; it showed that there has not been any significant progress in delivering the nanoparticles to a solid tumor¹¹⁰; the median of the injected dose of the nanoparticles reaching the tumor has remained at 0.7%. We can only speculate on the ultimate reasons for this non-progressive development, but one reason must be the incoherent and insufficient way to analyze the particles, as well as in the manner in which the experiments have been conducted and the results reported. In the present review, the authors were confronted by the challenge of assessing the reported results, e.g., the biodistribution and circulation half-times had been mainly reported by analyzing the pharmacokinetics of the drug instead of the accumulation of the carrier particles in the target tissue. This leads to misleading results as the particles that have accumulated in the non-targeted organs can still release the cargo drug into the blood¹¹¹. Moreover, characterization of apparently simple property such as coverage and stability of the coating is seemingly challenging. Furthermore, since the size and shape of the particles, as well as the dynamic presence of a protein corona will also influence the biofate of the carrier, it is almost impossible to resolve the correlation between the material properties and the functionality in vivo in an analytical way. When surveying reports, one must always remember the possibility of faulty results and evaluate the results givencritically.

31

Surface PEGylation and lipid coating are the most commonly used and clinically proved methods to improve the biocompatibility and circulation time of the synthetic porous nanocarriers.

The safety and functions of the surface PEG and lipid layers have been well evaluated in clinic with commercial products¹¹². Nevertheless, these PEG or lipid nanoparticles still have the possibility to cause immunological responses after intravenous administration. For example, repeated injections of PEGylated nanoparticles induce the accelerated blood clearance phenomenon. Lipid coated nanoparticles interact with immune cells and trigger inflammatory response¹¹³. One extremely promising way to circumvent the challenges stemming from the synthetic coatings is to employ the structures devised by Mother Nature. The applications involving cell membrane coated nanoparticles have shown great promise and they may represent the next step in utilizing multifunctional nanocarriers in nanomedicine^{95, 114}. These inorganic – organic (biogenic) hybrid nanoparticles have optimal short-term biocompatibility and tunable drug loading capable of handling both hydrophilic and hydrophobic cargos. The targetability can be varied by choosing cell membranes from different parent cells. Furthermore, they do not pose any genetic risks. However, even though several strategies have been developed to prepare nanoparticles coated with cell membranes, it is still challenging to prepare a nanocarrier with high yield. Furthermore, the long-term toxicity and safety of hybrid carriers will need to be evaluated more thoroughly in future reports.

6. Future perspectives

Considerable efforts and resources have been invested in engineering nanovectors as a means towards the goal of targeted therapies. The nanomedicine-related approaches have not fulfilled all the hopes and hypes set more than a decade ago, e.g., as ways of achieving personalized medicine. Indeed, translating a nanovector from bench to bedside seems to be extremely challenging. The reasons for the slow progress must be related to the untapped knowledge from the previous studies, as well as to the complexity of biological systems and their interactions with

32

nanoparticles. A more thorough utilization of all the available data and information provided in the previously published studies could reduce the development costs and help to select the most promising nanovectors for in vivo and clinical evaluations. Artificial intelligence, or more specifically, machine learning may represent an approach for the future; it could make it possible to clarify the relationship between the nanoparticle materials properties and the biological outcome. Since the engineering, manufacturing and testing of nanoparticles are both time- consuming and costly, these processes can significantly benefit from computationally guided design and validation. To unleash the full power of machine learning would require that in the future, scientists should be encouraged to report all of the fundamental properties of the materials in a comprehensive and coherent manner; in this way, it will be possible to identify the principal properties of the nanoparticles that result in the desired biological activity. In this respect, even so- called negative results would represent as valuable contributions as their positive counterparts.

Machine learning may also help to resolve the irreproducibility problem that has been commonly encountered in preclinical studies¹¹⁵.

7. Conflicts of interest

There are no conflicts of interest to declare

8. Acknowledgements

The work was supported in part by Academy of Finland (grant no. 314412 and 314552).