**1. Introduction**

A novel modality in cancer therapy emerged over the past two decades in the form of gene therapy and in this context, small interfering RNA (siRNA) and micro RNA (miRNA) are promising alternatives to common anti-cancer medications [1,2]. Both nucleotide oligomers mediate the process of RNA interference (RNAi) in which cells use these short RNA strands of 19-24 nucleotides to recognize messenger RNA (mRNA) with a complementary sequence, induce their destruction and thus inhibit the translation into proteins. To provide cancer therapy, synthetically produced siRNA and miRNA mimics can be used that target and silence specific oncogenes. Recent studies have identified a number of cancer-related genes as potential targets for RNAi-based therapy [3–5]. One of these tumor suppressor miRNA is miR200c, which targets the proto-oncogene KRAS. It regulates cell differentiation, proliferation and survival [6], the epithelial to mesenchymal transition [7] and suppresses chemoresistance [8].

However, a major challenge for a widespread therapeutic application of RNAi is creating appropriate delivery vehicles to safely and effectively deliver and release siRNA and miRNA into

the cytosol of disease-causing cells. Several excellent reviews summarize the requirements for siRNA delivery and comprehensively report on the efficiencies of gene silencing achieved with various delivery materials [9–11].

Some promising results were recently obtained from ongoing clinical trials concerning RNA delivery vehicles and in 2018 Alnylam's Onpattro (Patisiran) has been FDA-approved as the first ever RNAi drug against nerve damage caused by the rare hereditary disease, transthyretin amyloidosis [12]. However, Patisiran is a lipid-based nanocarrier system that does not actively target the cancer tissue, and most of the administered drug accumulates passively in the liver instead. Therefore, robust and specifically targeted carrier systems are needed for an efficient delivery of siRNA and miRNA to prevent a premature degradation of the unstable RNA. For this purpose, nanoparticles have attracted much attention because they provide a stable, nontoxic, and highly flexible platform.

Mesoporous silica nanoparticles (MSN) are emerging as potential nanocarriers for a variety of anti-cancer cargos including nucleic acids [13,14]. The properties that have rendered them specifically suitable for siRNA and miRNA delivery include a high surface area and pore volume, a controllable biodegradability [15,16], the tunability of particle size and pore size, their variable morphology and importantly, the possibility to create core-shell particles constructed of spatially separated regions with different surface functionalization [17,18].

This tunability of MSNs opens the possibility to systematically study the influence of the carrier properties on the gene silencing efficacy. Multiple parameters can affect successful gene silencing using carrier agents, e.g., the adsorption and release kinetics of the nucleic acid in the drug carrier, the dependence of cellular uptake or endosomal escape on nanoparticle size, surface characteristics such as charge and/or the attachments of functional residues, to name a few, but systematic studies addressing these issues are rare [19].

For the adsorption and release kinetics of nucleic acids using MSNs, the pore size, pore morphology and surface charge are important parameters. High loadings of nucleic acid were achieved by using large pore MSNs with a pore diameter of around 10–20 nm [20] and when pores were modified with cationic functional groups—aminopropyltriethoxysilane (APTES) or polylysine—to create a cationic layer for nucleic acid adsorption [14,21]. However, some of these systems suffer from poor release kinetics associated with low knockdown efficacies because of either the high affinity of polycations towards nucleic acid or because of a bottleneck-type pore morphology that features smaller pore openings than pore diameters, which both might decrease an efficient release of the highly charged nucleic acid molecules. Based on these findings, our group introduced novel medium-pore MSNs (pore diameter of around 5 nm) with stellate pore morphology, which were able to adsorb an exceptionally high amount of siRNA of 380 μg mg<sup>−</sup><sup>1</sup> and which enabled a mainly electrostatically driven fast and efficient RNA desorption, resulting in a high silencing efficacy [21].

Cellular internalization of the nanocarrier is another factor that influences gene silencing efficacy. The relationship between shape and size of MSNs [22–25] or other nanocarriers [26,27] and cellular internalization is important as nanoparticle size may affect the uptake efficiency and kinetics and the internalization mechanism [28]. One of the grea<sup>t</sup> advantages of MSNs is that they can be designed to feature different sizes and shapes. For MSNs [22,29] as well as for Au [30], polystyrene [31] and iron oxide nanoparticles [32], a size-dependent uptake in cells was observed with a maximum uptake at a particle diameter of 30–50 nm. However, as size is only one among several parameters controlling cellular uptake of nanoparticles, optimal sizes may vary for different surfaces and different surface functionalizations. Additionally, targeting ligands were shown to improve selective cellular uptake and a better accumulation in tumor tissue [33,34]. The effect of particle size on gene transfection efficiency using silica-based nanoparticles with diameters from 125 to 570 nm as nanocarriers for plasmid DNA was studied by Yu et al. Here, the transfection efficiency was found to be a compromise between binding capacity of the nanocarriers and cellular uptake. Smaller particles showed higher cellular uptake but less binding capacity for plasmid DNA. Particles with a diameter of 330 nm showed the best gene transfection efficacy [35].

Gene silencing mediated by delivery of siRNA and miRNA with MSN nanocarriers is strongly dependent on the escape of siRNA or miRNA from the endosomes into the cytosol. To trigger this reaction, the surface of particles is often decorated with cationic polymers, which are known to support an endosomal escape. Polyethylenimine (PEI) is one well-studied polymer exhibiting good endosomal escape capability when used at high dosage, attributed to a 'proton sponge' e ffect, however, it has a poor toxicity profile [36]. For instance, the group of Gu showed very good knockdown e fficacies when PEI was attached to the outside of their particles [37,38].

A systematic study of the correlation between gene knockdown e fficacy and endosomal escape kinetics was performed by Wang et al. [39]. They used magnetic MSNs capped with PEI at low concentrations to avoid toxic e ffects, resulting in poor endosomal escape ability. When endosomal escape was induced by a chloroquine treatment at early transfection times they could still see a high gene silencing e fficacy. In contrast, when applied at later stages, a low gene-silencing e fficacy was found, presumably because the released RNA was already degraded within the endosome.

Although MSNs have shown grea<sup>t</sup> potential as an e fficient carrier system for siRNA and miRNA delivery, systematic studies about particle properties and corresponding gene-silencing e fficacy remain rare. Herein, we present a systematic investigation of the particle size-dependent delivery of miRNA using core-shell MSNs for gene silencing in T24 cells. A series of MSNs with uniform sizes ranging from 60 to 160 nm with an average pore diameter of around 5–6 nm was used to encapsulate an antitumoral microRNA mimic (miR200c) or a control RNA with a scrambled sequence (Ctrl). The core of these particles was functionalized with APTES to yield a positively charged inner surface to accommodate the negatively charged miRNA, while a thinner surface layer was enriched with mercaptosilane, forming a negatively charged shell to enable binding of the capping agent. This consisted of an amino-acid block-copolymer 454 (see Figure 1) to aid endosomal escape and which was further linked via polyethylene glycol (PEG) to the peptide GE11, targeting the epidermal growth factor receptor [40]. This vector is abbreviated as MSN-454-GE11. An alternative construct for the delivery of this miRNA was studied before by Müller et al. using RNA-encapsulating polyplexes consisting of the same copolymer 454 functionalized with GE11 [41]. Their particles, featuring a size of 120–150 nm, successfully showed antitumoral e ffects with two di fferent therapeutic RNAs.

**Figure 1.** Schematic overview of the structure of core-shell mesoporous silica nanoparticles (MSN), the loading of miR200c and the polymer capping resulting in the MSN-454-GE11 vector. The positively

charged core in core-shell MSN enables a high loading capacity for miRNA. The mercapto-lined MSN shell associates with a positively charged block copolymer carrying the targeting ligand GE11, thus acting simultaneously as capping, endosomal release and targeting agent.

Our series of MSN-454-GE11 particles covers a broad size range but still provides comparable properties for each sample, including surface chemistry, surface charge, pore size and miRNA concentration to enable a conclusive study of the particle-size effect on miRNA delivery. Here, we show that good gene-silencing and antitumoral effects are obtained when miR200c is delivered by the largest MSN-454-GE11 particles in this sequence.
