The large data corpus of recent years provides evidence that silica nanomaterials may have the potential to strongly improve cancer treatment and diagnosis. Silica nanomaterials feature the versatility necessary for tumor-specific modifications, stability in the often harsh environments of the body, ease of production and - more importantly - they are generally regarded as biocompatible. However, the latter clearly depends on many parameters such as particle size, surface modification, dose, exposure time or cell type used as model . With the aim to explore the suitability of silica nanoparticles for new concepts in the treatment of head and neck cancers we investigated as a first step the biological in-vitro behaviour of non-targeted 200-300 nm core-shell silica nanoparticles with three different surface modifications.
While both Ru@SiO2-OH and Ru@SiO2-NH2 nanoparticles displayed high uptake rates in our model, internalization of PEGylated silica nanoparticles was almost completely lacking under the same experimental conditions. Although we observed this effect in the related HNSCC line UMB-SCC 969 and in the human prostate carcinoma cell line PC-3 as well (unpublished data), other studies showed, in contrast to our results, that PEGylated silica nanoshells are at least able to attach to the outside of MCF-7 cells . However, PEG is known for its cell-repelling properties [35–37], but uptake efficiency may be increased by the addition of targeting ligands . Since grafting of nanoparticles with PEG has been reported to be advantageous for in-vivo applications - basically due to its increased half-live in circulation - and helpful for targeting, the generation of optimized Ru@SiO2-PEG may be worthwhile (work in preparation).
Although the plasma membrane is negatively charged, the different surface charges of (negatively charged) Ru@SiO2-OH and (positively charged) Ru@SiO2-NH2 nanoparticles had no considerable influence on cellular uptake kinetics in our model. This is in contrast to reports indicating that negatively charged nanomaterials are less effectively internalized . However, a large number of studies show that both cationic and anionic nanoparticles are capable of effectively passing the cell membrane .
Our data indicates that at nanoparticle concentrations of 0.125 mg/ml and below, no perturbances in cell cycle progression have been detected under our experimental conditions. An increase of cancer cell proliferation could be dangerous and hold dire consequences in clinical settings. This phenomenon has been reported in-vitro for melanoma cells and mesoporous silica nanoparticles , but has never been observed in our experiments. However, higher concentrations of Ru@SiO2-OH and Ru@SiO2-NH2 lead to reduced proliferation rates. While a slowdown in growth of tumor cells may be generally regarded as a positive effect in cancer treatment it should be emphasized that the underlying pathomechanisms in HNSCC are not clear yet. Previous in-vitro studies in other cancer cell lines have shown that cytotoxicity of silica nanoparticles, in relation to size and incubation time, may be due to oxidative stress with lipid peroxidation and membrane damage and/or an inflammatory response [41, 42]. A detailed analysis of the complex molecular pathways involved is therefore needed in order to estimate possible (wanted or unwanted) consequences for future therapeutic strategies. Because of the different experimental design (e.g. longer incubation times, different particle sizes, other cell lines) it is impossible to directly compare our cytotoxicity data with previous studies. However, head and neck cancer cells seem to display cell toxic effects at concentrations comparable to other cancer cells, e.g. cervical adenocarcinoma cells , osteosarcoma cells , lung adenocarcinoma cells [37, 41], and gastric and colon cancer cells . Despite this, nanoparticle concentrations have to be carefully adjusted: using the same nanoparticles and experimental conditions as here, PC-3 human prostate cancer cells displayed a proliferation stagnation of about 15 days after nanoparticle incubation, although metabolic rates have been found to be higher (Besic Gyenge et al., unpublished).
With regard to internalization processes of nanoparticles into cells, phagocytosis, pinocytosis and caveolin- or clathrin-driven endocytosis have all been proposed and seem to strongly depend on particle form, size and cell type used. With our experimental set-up, apparently two different routes of nanoparticle uptake occur in parallel: on the one hand, single particles enter HNSCC cells via membrane invaginations that ultrastructurally resemble clathrin-coated pits. The involvement of clathrin-coated pits in internalization mechanisms of silica nanoparticles had also been proposed in several previous in-vitro studies using specific inhibitors or confocal methods [45–48]. On the other hand, the observed bulk internalization of nanoparticles is likely related to non-clathrin mediated endocytosis. The latter process rather displays features of macropinocytosis, such as membrane ruffling. Notably, the different surface charges of our nanoparticles did not play an apparent role with regard to the observed uptake mechanisms. Detailed studies are now needed to further characterize the events taking place at the plasma membrane upon contact with our silica nanoparticles. However, the incidence of such different simultaneous endocytosis modes of silica nanoparticles is in accordance with a recent paper, where also discrete entry pathways have been observed for single and agglomerated amorphous silica nanoparticles . Furthermore, in mouse melanoma cells, internalization of latex particles of 200 nm (that corresponds approx. to the size of our particles) involved clathrin-coated pits, while latex particles of 500 nm (that corresponds approx. to our nanoparticle clusters) preferentially entered the cells via a clathrin-independent caveolin-associated pathway .
To characterize the intracellular fate of our silica nanoparticles within HNSCC, we next investigated their possible delivery into early and late endosomes and lysosomes. The localization of Ru@SiO2-OH and Ru@SiO2-NH2 in early endosomes indicates their processing to endocytotic pathways, however, a considerable number of particles obviously used a different route of trafficking, that did not involve EEA1-positive organelles. As long as these organelles have not been characterized, a possible role of nanoparticle's surface charge for endocytic processes cannot be defined. However, the acidic pH of early endosomes may explain the slightly higher frequency of (negatively charged) Ru@SiO2-OH in EEA1-containing vesicles.
While we cannot exclude that some Ru@SiO2-OH and Ru@SiO2-NH2 may have been shuttled back to the plasma membrane for segregation, the majority of nanoparticles remained intracellularly and accumulated in rather large vesicles 24 h after incubation. We propose that the latter is related to homotypic vesicle fusion. No transfer to Golgi apparatus-related pathways has been detected. More importantly, we found that nanoparticle-bearing vesicles did neither mature from early endosomes into (Rab7-positive) late endosomes nor locate to lysosomes. While both the known stability of silica-shell nanoparticles and possible cancer-related changes in endosomal sorting mechanisms may have prevented their targeting to degradation pathways, our data is in contrast to other studies showing that silica nanoparticles are in fact transferred to lysosomes [46, 47, 50]. Our results also differ from those of Rejman et al.  where a size-dependency of endocytotic pathways had been proposed. In this study, at least smaller latex particles (200 nm) passaged to late endosomes/lysosomes while only large particles (500 nm) did not . We therefore conclude that intracellular fate of nanoparticles not only depends on their size (or agglomeration status) but presumably also on cell line.
Although the exact nature of different endocytotic organelles in our model has to await further characterization, the strictly vesicle-associated occurrence of Ru@SiO2-OH and Ru@SiO2-NH2 in HNSCC may have contributed to their biocompatibility. In human melanoma cells it had been reported that an escape of silica nanoparticles to the cytoplasm resulted in changes of the cytoskeleton as well as of adhesion and migration properties . Whether the vesicular enclosure of our nanoparticles is a useful feature in the case of intracellular drug delivery strategies in HNSCC remains to be proven.
In addition to their relatively large diameter [52, 53], the absence of free Ru@SiO2-OH and Ru@SiO2-NH2 within the cytoplasm of HNSCC may have been the reason that nanoparticles never passed the nuclear membrane. Even though localizations of silica nanoparticles within the nucleus had been observed before , our data is in accordance with results from previous studies [53, 55, 56]. Recently, it had been shown that labeling with fluorophores may affect uptake kinetics and intracellular pathways of certain probes . However, due to encapsulation of the dye in our study, it is unlikely that [Ru(bpy)3]Cl2 may have influenced routes of nanoparticles within cells. Until now, very little is known about the intracellular long-term fate of silica nanoparticles and possible consequences of their persistence in biological systems. In human lung epithelial cells, Stayton et. al. observed a slow but active transfer of silica nanoparticles from the cytoplasm to the exterior environment . They showed that during the first 24 h almost 50% of nanoparticles exited the cells. In contrast, our data implicates that both internalized Ru@SiO2-OH and Ru@SiO2-NH2 remain within the cell and are apparently distributed between daughter cells at cell division. During the course of several passages, initially high nanoparticle amounts in individual cells become "diluted", but ultrastructurally are still found in vesicles. The reason for the observed differences in long-time persistence of Ru@SiO2-OH and Ru@SiO2-NH2 are not clear yet, but may also be related to as yet uncharacterised charge-dependent effects of nanoparticles on endolysosomal pathways. However, in addition of not featuring acute cell toxic effects, the presence of our silica nanoparticles over 15 days caused no visible changes in viability, proliferation or morphology in HNSCC. Of note, over the time course studied, ultrastructure of nanoparticles appears to remain unchanged. However, it cannot be excluded that discrete processes of nanoparticle degradation occurred. Recently, it had been reported that, depending on functionalization, integrity of silica nanoparticles may be impaired step-wise over time in simulated body fluid with regard to e.g. surface area, pore width or pore volume [59, 60].
Although the high uptake efficiency of Ru@SiO2-OH and Ru@SiO2-NH2 in our in-vitro mono-layer model was promising, optimized conditions are needed in case of solid HNSCC tumors where conditions of poorer vascularisation may exist. Our results in HNSCC spheroids, an established minitumor model, show that penetration depth of Ru@SiO2-OH and Ru@SiO2-NH2 does not reach beyond the first (outer) cell layer - independent of nanoparticle surface charge. This observation provides further evidence that our nanoparticles are not actively exocytosed. Stayton et. al. showed in-vitro that nanoparticles which were exocytosed in growth medium were taken up by other cells if not removed from growth medium . Given that nanoparticles are not transported transcellularly and apparently are incapable of passing the intercellular junction complexes, new delivery strategies have to be developed for multicellular poorly vascularized cancers.