Recent Targeted of siRNA Delivery Vehicles for Cancer Therapy

Recent progress in RNA biology has broadened the scope of therapeutic targets of RNA drugs for cancer therapy. However, RNA drugs, typically small interfering...

lymphatic ducts, compared with healthy organs/tissues, permitting the accumulation of nanoparticles with several tens to hundred nm in tumor tissues. This size-mediated tumor accumulation mechanism of nanoparticles (or macromolecular drugs) was originally observed by Y. Matsumura and H. Maeda in 1986, and was termed the enhanced permeability and retention (EPR) effect [6].
To date, the size-mediated tumor accumulation of nanoparticles has been widely demonstrated in various tumor-bearing murine models using polymeric micelles, inorganic nanoparticles, and lipid nanoparticles [7][8][9]. Of importance in this regard is that the tumor accumulation behavior of nanoparticles is significantly affected by the pathophysiology of tumor tissues [10].

Design Criteria to Overcome Extracellular Barriers
Transports of nanoparticles from blood vessels to cancer cells are governed by particle dynamics regarding physical barrier of stroma. The recent observation onto tumor microenvironment showed that tumor blood vessels undergo time-limited formation of an opening in the vessel walls followed by brief outward flow of fluid into the tumor interstitial space (termed 'eruptions') probably due to hydrodynamic pressure gradient [25]. Both 30nm-sized and 70nm-sized nanoparticles were erupted into tumor interstitial spaces. The 30nm-sized nanoparticles quickly diffused away but the 70nm-sized nanoparticles was trapped in stroma-rich barriers [25]. Cancer cells mostly surround blood vessels in some clinical tumor (e.g. kidney, brain, liver, thyroid, ovarian, and head and neck cancers), whereas stroma surrounds vessel in other clinical tumor (e.g. breast, pancreatic, colorectal and non-small cell lung cancers) [26,27]. Thus, in case of stroma-rich tumors, nanoparticles need to penetrate into (or distribute across) the stroma tissue to reach the cancer cells nests. In this regard, the penetrability of nanoparticles is reported to significantly depend on their particle size as follows. that the intravenously injected CALAA-01 has a zeta-potential of 10-30mV and was entrapped with GBMs, which have a high density of heparan sulfate [33], resulting in the loss of their structural integrity. In another study, siRNA-loaded cationic polysaccharide nanoparticles were transferred from the kidney to the bladder more slowly compared with naked siRNA. Considering that the original size of the nanoparticles (220-230nm in diameter) was larger than the pore size of GBMs, the siRNA transfer to the bladder implies that siRNA payloads were gradually released from the nanoparticles and that GBM partially contributes to disassembly of the intravenously-

Design Criteria to Overcome Intracellular Barriers
Once target tissues or cancer cells have been reached, delivery vehicles should interact with the cellular surface for internalization into cells. The main parameters that determine endocytosis of delivery vehicles are shape, size, and surface chemistry. These parameters are believed to affect not only the cellular uptake efficiency but also the endocytotic route [48]. The shape effect of delivery vehicles is not described in this review because most delivery vehicles for cancer therapy are constructed to possess a spherical morphology through simple self-assembly procedures or natural growth of seed inorganic particles (see the reference [49,50] on the shape effect). Multimolecular delivery vehicles, e.g., polymeric micelles and LNPs are generally constructed to be 30-100nmin size, to promote the EPR effect. Inorganic nanoparticles are also reported to demonstrate a size effect on endocytosis; bare gold nanoparticles with 20-50nm in diameter have demonstrated that the most efficient cellular uptake between 10 and 100nm size ranges in cultured cancer cells because gold nanoparticles with these size ranges may balance between the elevated elastic energy associated with increased curvature of the cell membrane and reduced entropy associated with receptor/ligand immobilization [51][52][53]. Surface chemistries of delivery vehicles are apparently more critical for their endocytosis, compared to size and shape.
Positively charged nanoparticles have a high affinity to negatively charged proteoglycans expressed on the surface of most cells, resulting in more efficient adsorptive endocytosis, compared with neutral and negatively charged nanoparticles. Of note, heparin sulfate proteoglycans, comprise transmembrane proteins termed syndecans, are considered major binding sites for cationic delivery vehicles [54]. However, such cationic nanoparticles are not able to take advantage of systemic administration due to nonspecific interactions with negatively charged blood components before reaching target cells. PEGylation of delivery vehicles is a standard strategy, which suppresses such aggregate formation [38]. Nevertheless, PEGylation of delivery vehicles concurrently generates disadvantages for cellular entry due to weakened interactions with the surface of target cells (termed PEG dilemma) [55].  [56,57]). Arginine-glycine-aspartic acid (RGD) peptide and folate are typical ligands used in siRNA delivery for various types of cancers because these ligands are closely related to angiogenesis of tumor development and metabolism of fast-growing cancer cells.
The RGD peptide can strongly and specifically bind to α v β 3 and α v β 5 integrin receptors, which are overexpressed on many cancerous and neovascular endothelial cell surfaces [58,59]. A cyclic form of the RGD peptide (cRGD) provides the rigid structure for enhanced affinity to the target integrins (KD = approximately 40nM for α v β 3 integrin [60]) and prevents degradation of the highly susceptible aspartic acid residue [58]. Folate is a low molecular weight vitamin required by all eukaryotic cells for 1-carbon metabolism and the synthesis of purines and pyrimidines. It has a high affinity (KD = approximately 10nM) for folate receptor isoform α (FR-α), which is highly overexpressed on the surface of ovarian, uterine, brain, and CNS cancers, whereas a high to moderate level of FR-α expression is detected in lung, kidney, and breast cancers [56,61]. Monoclonal antibodies and their fragments are also utilized to recognize specific molecules (i.e., antigens) on the surface of cancer cells.
The structure of antibody divides into two different bio functional subdomains. The antigen-binding fragment (Fab) mediates antigen recognition via complementarity-determining regions and the crystallizable fragment (Fc) recruits Fc receptor on the immune cell or the other antibody recognition [62].
Trastuzumab, an antibody for FDA approved antibody-drug conjugate Trastuzumab emtansine, has KD = 1-7nM for the transmembrane tyrosine kinase receptor (HER2) [63][64][65]. Fab fragments can be used to reduce the bulky size of a full antibody (approximately 15nm), alleviating immunogenicity and improving the pharmacokinetic profile of delivery vehicles [61,66]. Nucleic acid aptamers, which are single-stranded oligonucleotides with a specific 3D structure, also exert high binding specificity to their target molecules [61]. To date, no ligand-installed multimolecular delivery vehicle containing oligonucleotides or small molecular drug goes to markets [67]. On the other hand, two antibody-drug conjugates are approved by FDA for treatment of lymphoma and HER2-positive breast cancer [63]. In subcutaneous folate receptorpositive tumor mice model, folate-conjugated Vinca alkaloid, EC145, showed complete cures without a relapse for > 90 days post-tumor implantation in 4/5 mice [68]. However, in Phase II trials combined with pegylated liposomal doxorubicin (PLD), median progressionfree survival (PFS) of EC145 plus PLD marginally increased from 2.7 to 5.0 months compared to PLD-alone control in ovarian cancer [69]. In recent report, Phase III study was stopped because EC145 did not demonstrated efficacy regarding PFS in patients.
One plausible explanation is that receptor properties in animal's tumor did not represent properties of primary cancer cell found in a patient's tumor [23]. Other gaps between laboratory animal model and human patients in tumor targeting were well explained in interesting perspectives [23,70].
After endocytosis, siRNA-loaded delivery vehicles encounter a sequential pH drop in the early endosome (pH 6.5), late endosome (pH 6.0), and lysosome (pH 4.5-5.0) [71,72]. In this way, the resulting in high transfection efficiency [73,74]. Endosomal escape induced by low pKa amines has been explained by two possible mechanisms. One is the proton sponge hypothesis based on increased osmotic pressure [75,76]. Low pKa amines can protonate in acidic endosomal compartments and induce proton influx into induce comparatively high endosomal escape in cultured cells [74].
The other mechanism is direct membrane destabilization by highly charged polycations [74,77]. As aforementioned, polycations can bind to the oppositely charged cellular membrane and perturb membrane integrity. In particular, polycations bearing low pKa amines can significantly elevate their positive charge density through amine protonation in endo/lysosomal compartments, and consequently perturb the endo/lysosomal membrane integrity for membrane destabilization. The design strategies of delivery vehicles, which are capable of endosome disruption, will be described later in this chapter.  (Table 1).  In this way, the multimolecular structure can dissociate into building components and simultaneously releases encapsulated siRNAs.

Design of siRNA Delivery Vehicles
This biosignal responsive disassembly of the multimolecular structure elicits the rapid siRNA release in the target site.

Hydrophobicity-Stabilized Delivery Vehicles: Stabilization of delivery vehicles by hydrophobic interaction in aqueous solutions
has been investigated in the early development of vehicles because of simple chemistry for introduction of hydrophobic moieties into component materials [83,84]. Hydrophobic moieties, such as alkyl chains and cholesterol, installed into cationic components can assist the spontaneous assembly of multimolecular structures with siRNA through hydrophobic interactions, rendering delivery vehicles more resistant to dissociation ( Figure 2A). This increase in the association number (e.g. the number of building components in single delivery vehicle) between hydrophobic cationic components and siRNA results in a higher resistance against serum-containing media compared with unmodified cationic components [85,86].
Consequently, the hydrophobized delivery vehicles permitted more efficient cellular uptake of siRNA payloads, leading to enhanced endogenous gene silencing in cultured cancer cells. However, the higher stability of delivery vehicles in serum-containing media does not guarantee the stability in the bloodstream. When hydrophobized PEG-polycations formulated with siRNA, blood circulation property increased for only 10 min after tail vein administration [87,88].

The resulting delivery vehicles exerted inefficient tumor growth
inhibition in a subcutaneous model of tumor, indicating that simple hydrophobic moiety introduction into cationic components is not enough to generate the stability of vehicles in bloodstream.
In this regard, a previous study demonstrated that vehicle stability could be further improved by compartmentalizing the hydrophobic moieties within the multimolecular structure [89] ( Figure 2B). The exclusion of hydrophilic siRNA payload as well as cationic segments from the hydrophobic core allowed for more stable assembly because the hydrophobic components were more tightly packaged in the core without interferences of hydrophilic  to normal intracellular peptidase-mediated degradation [93]. The cell maintains reduced glutathione (GSH) by de novo synthesis from the three amino acids and reduction of oxidized glutathione (GSSG) by glutathione reductase, which induces the concentration of GSH to be 50-1000 times higher than that of GSSG in cells [94,95].
In this way, GSH concentrations can be distinguished between the intracellular and extracellular environment. GSH concentration within cells is 0.5-10mM but decreases to 10-30μM in blood plasma [93][94][95]. Thus, the disulfide linkage can be preferably cleaved in the cytoplasm or intracellular compartments while it is slowly degraded during blood circulation. The payload drug release from disulfide crosslinked nanoparticles was observed from 2-4h after internalization by the cells in cultured cells [96].

Acidic pH Responsive Delivery Vehicles
Acidic pH (pH 4.5-6) in late endosomes has frequently been highlighted as a representative biosignal for triggering the sitespecific drug release from delivery vehicles because a large pH change from extracellular neutral pH to endosomal acidic pH allows us to utilize acid-labile chemistry [101]. Various acid-labile bonds, such as acetals [102], hydrazones [103], β-thiopropionate [104], phosphoramidate [105], orthoesters [106], and citraconic amide [107,108], have been applied to construct multimolecular structures that elicit acidic pH-responsive release of drug payloads, including anticancer drugs and biomacromolecules ( Figure 4). detachment of PEG [112], ligand [99], and cationic polymer [113]. Although the thiol group does not affect initial electrostatic association between the polycations and siRNA, the relatively hydrophobic and bulky PBA group may prevent efficient crosslinking between diol and PBA. More precise design of block copolymer can improve delivery vehicle performance.
The payload release dependent on the gradient of ATP concentration was also attained by using an ATP-binding aptamer-incorporated DNA motif for anticancer drug delivery [116,117]. The Gu group designed a doxorubicin-   (Table 2). Lipid-based nanoparticle [133] Ligand Installed Delivery Vehicles: To optimize the ligandmediated active targeting functionality, several parameters, including ligand density and length/density of spacer, should be considered for construction of actively targeted multimolecular structures. The underlying mechanism of active targeting is the recognition of the ligand by its target receptors, and thus, a higher density of both ligand and receptor generally guarantee higher opportunities of their binding [128][129][130][131][132][133]. For example, conjugates of chemically modified siRNA and tri-N-acetylgalactosamine (GalNAc) resulted in higher cellular uptake in primary mouse hepatocytes than bi-GalNAc siRNA conjugates [134]. This result demonstrates a multivalent binding effect of ligands for the enhanced cellular The receptors are expressed not only on target cellular surface but also non-target ones at lower levels. Thus, a higher number of ligands can generate the higher affinity (or avidity) to target cellular surface, but concurrently increase the risk for non-specific binding to such non-targeted cells. Intercellular adhesion molecule-1 (ICAM-1) is constitutively expressed at a basal level on endothelial cells in quiescent vasculature but its expression is markedly elevated in pathologically activated endothelium. Introduction of reduced quantity of ICAM-1 specific antibody onto a particle surface enhanced the selectivity for binding to inflamed vasculature compared with normal tissues [135].  [56,139]. Interestingly, delivery vehicles can utilize blood components for active targeting.
Cholesterol-conjugated siRNA with a partial phosphorothioate backbone and 2′-O-methyl-modified nucleotides binds to lowdensity lipoprotein (LDL) (KD = 100μM) and obtained a plasma half-life of approximately 100-120min (dose amount: 50mg/kg), accompanied by significant gene silencing in the liver through LDL receptor-mediated endocytosis [140,141]. Some lipid-based nanoparticles are also believed to exchange their components with serum and adsorb lipoproteins, leading to enhanced internalization into hepatocytes through lipoprotein receptors [142].

Delivery Vehicles for High Endosomal Escapability
Delivery vehicles potentially contain endosomal escapability and facilitate their escape from endo/lysosomal acidification. In siRNA. PEI is a representative cationic polymer eliciting endosomal escapability through the proton sponge hypothesis [75,76]. PEI shows partial protonation of nitrogens at physiological pH (45% in linear PEIDP520) and augmented protonation at endo/lysosomal acidification (55% in linear PEIDP520) in 150mM NaCl [77,143].
One notable disadvantage of PEI is cytotoxicity. It is known that the cytotoxicity is substantially elevated with an increase in molecular weight of PEI. However, low molecular weight PEI cannot maintain stable multimolecular structure under physiological milieu because of its less ion pairing sites to nucleic acids, compromising the transfection efficacy. Thus, the low molecular weight PEI (e.g. molecular weight 800Da) was conjugated with each other to have a higher molecular weight (e.g. average molecular weight 10-20 kDa) through biodegradable linkages for maintaining higher transfection efficiency associated with lower cytotoxicity [144,145].
A low toxic and pH-dependent cationic moiety was alternately developed by fine-tuning the number of repeating amino ethylene unit, -(CH 2 CH 2 NH)n-in the side chain of polyaspartamide [74]. an endosomal escape moiety [149]. Although some cationic lipids were synthesized by linking the alkyl chain or lipid components with PEI to provide buffering effect [150,151], the lipid-based nanoparticle had its own endosomal escape mechanism, which was first proposed by the Szoka group [152,153]. Cationic

Delivery Carrier Design in Other Category
Layer-by-layer delivery vehicle: Delivery vehicles constructed by layer-by-layer (LbL) technology for high loading of siRNA has been an attractive strategy for local administration because the vehicle has superior gene silencing efficiency even at picomolar siRNA concentrations in cultured cells [155,156]. Nevertheless, its larger particle size (> 100 nm) and wide size distribution of the particle have been believed to be drawbacks for passive/active targeting by systemic route. Very recently, a systemically injected layer-by-layer (LbL) particle was developed to exhibit comparative tumor growth inhibition in animal models when the particle was carefully engineered to possess building components for tumor targeting ( Figure 6A). LbL nanoparticle construction, which alternately deposits siRNA and polycations on to template, has a unique advantage over other multimolecular structures because a single nanoparticle can load much larger amounts of siRNAs (approximately 3500 siRNA molecules) and exhibit a long period of siRNA release time (approximately 3 weeks) in cultured cells [157]. When an LbL nanoparticle codelivered multidrug resistance protein 1 (MRP1) siRNA with doxorubicin into a subcutaneous animal model of triple-negative breast cancer, MDA-MB-468 cancer, it showed the synergistic inhibition of tumor growth (1mg/kg siRNA and 1mg/kg doxorubicin) [158]. Of note, MRP1 is a cellsurface efflux pump involved in redox regulation of multidrug resistance by clearing the intracellular concentration of xenobiotics and toxins [159,160]. Design strategies of delivery vehicles to increase the efficiency of combination therapy and the targeting gene/cancer drug combination are well described in a previous review [161]. The LbL nanoparticle provides clues in design of delivery vehicles.
High loading of siRNA in the single delivery vehicle is also a critical factor for efficient gene silencing. Another point to consider is that altering the size of LbL nanoparticles loaded with high amount of siRNA will increase performance of the delivery vehicles.

Phosphate-Formulated Delivery Vehicles:
Deposition and co-precipitation of inorganic materials on multimolecular structures can facilitate to formulate siRNA [162]. This data suggests that thiol-gold coordination is stable enough to maintain thiol-terminated RNA duplexes in the bloodstream. The SNA nanoparticle successfully penetrates the blood-brain barrier   [170].