Whereas nanotechnology seeks to engineer and apply the unique properties of materials that emerge when the dimensions enter the nanoscale, nanomedicine attempts to exploit these for the benefit of human health. Nanomaterials in this context are intriguing because they can resemble biological ‘nanomachines’ (such as biomacromolecules) as they meet on the same length scale 1 and, thus, can be expected to perform similar tasks or at least possess reminiscent biobehavior. One of the properties exploited in nanomedicine is the ability of nanoparticles to enter cells, whereby they can function as efficient carriers for intracellular drug delivery. The behavior of nanomaterials on the nano-bio interface is largely determined by the physicochemical properties of the nanocarrier, of which size and surface charge have been emphasized as the two most critical ones. Another design parameter that has not received as much attention to date is particle shape 2 . Available results suggest that, as compared to their spherical counterparts, elongated nanoparticles could be more favorable for therapeutic applications based on targeting specificity, biodistribution as well as cellular internalization profiles 3 . On this topic, it has been suggested that the resemblance of rod-like particles with rod-like bacteria could be a reason for the observed advantages in internalization rates in non-phagocytic cells 4 . Particles that, in such a way, mimic properties such as size, shape and flexibility of naturally occurring entities (for instance, circulating cells, e.g., red blood cells) may offer advantages that are typically not observed for standard (polymeric) particles 5 .

One reason for the general scarcity of the shape-dependency effects on biological performance is the more difficult fabrication of nanoparticles with controlled rod-like morphology due to surface energy minimization during synthesis, leading to spherical shapes 6 . Especially for more novel materials' classes, the first hurdles to overcome include obtaining well-dispersed, monosized particles throughout all processing steps 7 . One promising materials' class in the sense of nanomedicine is mesoporous silica, which can be produced with nanoscale particle diameters and with tunable properties when it comes to both particle size and morphology, meso (pore) structure and size as well as surface chemistry 8 . The controllable characteristics of these materials have boosted research regarding their potential use within the biomedical field during the last decade, especially so for targeted cancer therapy and diagnostics 9 10 11 . Mesoporous silica nanoparticles (MSNs) have been successfully loaded with a range of different chemotherapeutics for efficient intracellular delivery and subsequent elimination of the cancer cells. A vast array of sizes and surface modifications have been produced and studied both in vitro and in vivo, while only a few studies have been devoted to the shape effect on MSN biobehavior 12 13 14 15 16 . To add to the increased understanding of shape-induced effects for MSNs, we have, in this study, synthesized two similarly sized but differently shaped rod-like (NR-MSP) and spherical (S-MSP) fluorescently labeled mesoporous silica nanoparticles in order to investigate the morphology effect on cellular internalization. Additionally, both the rods and spheres were functionalized to yield different surface charges in order to distinguish morphology from surface charge-induced effects and, in the best case scenario, find the dominant parameter. In vitro studies were carried out in two different cancer cell lines, HeLa (cervical carcinoma cells) and Caco-2 (human epithelial colorectal adenocarcinoma cells), to investigate the particle characteristic impact on cells of different origin. According to our results, all of the studied factors (particle shape, surface charge, the nature of the coating as well as the targeted cell populations) are important to be considered when designing new nanocarrier formulations for targeted cancer therapies or other potent drugs that require tight dose control.

The characteristics of the produced particles in suspension form (0.5 mg/ml HEPES buffer) have been summarized in Table 1. The PEI amount (in weight percent with respect to the whole particle system) as determined from thermogravimetrical analysis was around 10 wt.% for all PEI-modified particles.

NR-MSP, rod-like mesoporous silica particles; NR-MSP-PEI-ACA, acetic anhydride-modified NR-MSP-PEI graf.; NR-MSP-PEI ads., 25 k poly(ethylene imine)-adsorbed NR-MSP; NR-MSP-PEI graf., poly(ethylene imine) surface-grown NR-MSP; NR-MSP-PEI-SUCC, succinic anhydride-modified NR-MSP-PEI graf.; S-MSP₁, negatively charged spherical mesoporos silica nanoparticles; S-MSP₂, neutrally charged spherical mesoporos silica nanoparticles; S-MSP₁-PEI ads., 25 k poly(ethylene imine)-adsorbed S-MSP₁; S-MSP₂-PEI graf., poly(ethylene imine) surface-grown S-MSP₂.

As a basis for comparison, these particular S-MSPs and NR-MSPs were chosen due to their similarity in size as determined from DLS methods on the nonfluorescent particles (see Figure S1 in Additional file 2), i.e., around 500 to approximately 600 nm by DLS as well as electron microscopy images (Figure  3). DLS for fluorescently labeled particles (Table 1) is carried out to confirm redispersibility of the particles in aqueous solvent at a physiologically relevant pH (7.2) after all processing steps (synthesis, drying, solvent extraction/calcination and surface functionalization), as judged by the quality of the measurement as well as the polydispersity index. The DLS (z-average) values listed in Table 1 are thus based on an average of three good quality measurements, thus confirming full redispersibility of all particles in HEPES buffer solution at pH 7.2, which was chosen because it is also used to buffer cell media in the cellular experiments. The DLS values listed in Table 1 cannot, however, be used to determine the hydrodynamic size in solution in this case as the fluorescent labeling may distort the calculations applied to derive size from light scattering measurements. Moreover, DLS cannot be correctly applied to derive the hydrodynamic size for non-spherical particles. Thus, the particle sizes and morphologies were rather determined by scanning electron microscopy (Figure  3). For the rod-like particles, the aspect ratio (AR) was determined to be 3 to 4 based on SEM analysis. Since no shape variety was observed for spherical particles, the aspect ratio is regarded as 1.

From the transmission electron microscopy image (Figure  3a), the mesoporous structure of the particle can also be deduced. The porous structure was also confirmed by powder X-ray scattering and nitrogen sorption measurements (Figure  4), from which the different pore size characteristics are also clearly observed. The S-MSPs were templated using cetyltrimethyl ammonium surfactants, typical of a MCM-41 synthesis, which in our reported case resulted in pore size of 3.5 nm (by Density functional theory (DFT)) and unit cell parameter of 4.7 nm (for S-MSP₁ and S-MSP₂). On the other hand, the SDA used for the synthesis of NR-MSP, the nonionic surfactant block-co-polymer P123 typical for the synthesis of SBA-15 materials, results in pore sizes of 7 nm and over in case additives are used 23 25 . In this case, the pore size of the NR-MSP is 11.3 nm (by DFT) with a unit cell parameter of 14.8 nm. Mesoporous silica nanorods have also been synthesized using CTAB as structure-directing agent with another nonionic surfactant, peregal Os-25, as co-template 26 . On an interesting note, rod-shaped SiC has also been prepared using uncalcined mesoporous silica SBA-15 as template 27 . The advantage with our applied synthesis regime, besides the exceptionally large pores that can be obtained, is the very short synthesis time (1 h) 23 .

The fluorescent labeling of the particles regardless of method (co-condensation or post-synthesis grafting) were successful, as determined by fluorescence intensity measurements in HEPES buffer at the same concentrations (0.5 mg/ml; Table 1). On a related note, we want to highlight the alteration of fluorescence intensity upon surface functionalization when fluorescein is used as the fluorescent tag. Fluorescein, mostly used in the ‘activated’ form FITC, is a pH probe, and thus, both its absorption and fluorescence properties vary with pH as well as solvent 28 . This not only has implications for its intracellular localization upon quantification of cellular uptake as the pH in different cellular compartments differ, and for instance, in endo/lysosomes where particles would reside as a result of endocytotic uptake, the environment is acidic (pH 5 to 6). As is clearly illustrated in Table 1, surface functionalization, especially with acidic/basic groups such as amines or carboxylic acids (PEI and SUCC) which are the mostly used derivatizations when further bioconjugation is aimed for, also results in local pH changes on the nanoparticle surface. This has direct implications for differences in fluorescence intensity of the nanoparticles, despite being based on the same particle and measured at the same pH (here, in HEPES buffer at pH 7.2). This complicates fluorescence intensity-based quantification of cellular uptake as intracellular localization of the particles is not known at the time of detection whereby fluorescence intensity could also be measured at the right pH, and moreover, the localization for all particles within the same cell is not necessarily coinciding. Nevertheless, despite this drawback, fluorescein is the most widely used fluorescent tag due to its low cost, and these limitations in the absolute quantification based solely on fluorescence intensity are rarely taken into account. Consequently, we have chosen to mainly focus on percentage of positive cells in our cellular uptake studies as this value is not dependent on the absolute intensity of the particles.

Moreover, surface functionalization may also alter the intracellular localization of particles (Figure  5) as, for instance, PEI is commonly used as a nonviral gene transfection agent due to its ability to destabilize endosomal membranes and promote endosomal escape 29 . When compared to the uncoated spherical S-MSP₁, indeed a more diffuse ‘spread-out’ pattern can be discerned for the PEI-coated particle, whereas a clearer intracellular aggregation of particles is seen when no PEI is present. This may be due to the fact that uncoated particles (Figure  5b) are more compartmentalized inside the cells as compared to their PEI-functionalized counterparts, which are more efficiently capable of escaping endo/lysosomes and consequently able to distribute inside the whole cell (cytoplasm). For the rod-like NR-MSPs, however, an almost fibrous pattern can be perceived, indicative of individual internalized rods (Figure  5).

Our observed enhanced uptake of the rod-like MSPs is in line with previous reports on rod-like MSN biobehavior. These previous reports have mainly concentrated on non-functionalized MSNs aside from the fluorescent labeling (i.e., comparable to our S-MSP₁ and NR-MSP) to render the particles observable by confocal (fluorescence) microscopy and other applied fluorescence-based characterization techniques. Meng et al. 13 investigated the aspect ratio effect on cellular uptake rate and amount, and found that particles with an AR of 2.1 to 2.5 were taken up in greater quantities than shorter (1.5 to 1,7) and longer (4 to 4.5) rods as well as spheres (AR = 1), which seem consistent with earlier reports on organic nanoparticles for which AR ~ 3 was found to be the most efficient in terms of cellular internalization 4 . The aspect ratio of our studied NR-MSP is around 3 to 4. Meng et al. stated that their observations implied that a cellular mechanism capable of discerning and responding to rod length was operative, and compared this observation to that of Champion and Mirtagotri 34 , who studied the phagocytosis mechanism of non-spherical polymer particles in macrophages and concluded that it was the local geometry (angle of curvature) at the point of contact with the cell membrane that initiated phagocytosis, not the overall AR. It is worth to mention that the two mechanisms of uptake studied in the two separate studies were different, though Meng et al. studied macropinocytosis and the effect of particle shape on this particular cellular uptake route in two cancer cell lines (HeLa and A549) 13 . Huang et al. also studied the effect of MSNs of different ARs (1, 2 and 4) on uptake amounts and rates in a human melanoma (A357) cell line as well as AR impacts on cellular function including cell proliferation, apoptosis, cytoskeleton formation, adhesion and migration 12 . They found that increasing AR gave rise to increased uptake whereas long rod-shaped particles could cause reorganization of the cytoskeleton, and ascribed this observation to the more efficient endocytosis of rod MSNs than spheres 12 . Whereas Meng et al. found the behavior to be similar between the two cancer cell lines studied 13 , Trewyn et al. found slight differences between cell lines when comparing a cancer cell line (Chinese hamster ovarian) to a human fibroblast cell line 16 . In our current investigation between two cancer cell lines, the overall trends seem to be similar but with different dose-dependencies. Huang et al. 14 studied the effect of uncoated and PEGylated MSNs with two different ARs (1.5 and 5) on biodistribution, clearance and biocompatibility in vivo, whereby they found that the AR affected the clearance rate in that short rod MSNs (AR ~ 1.5) had a more rapid clearance rate than the long rod MSNs (AR ~ 5). No profound in vivo toxicity (based on hematology, serum biochemistry and histopathology) was found, in line with their previous report where MSN concentrations as high as 1,000 mg/kg were used 35 . Another useful means of predicting in vivo compatibility when aiming for circulating nanoparticles is determining their hemolytic activity, whereby Yu et al. have investigated pure and amino-functionalized porous and non-porous silica nanoparticles of ARs 2, 4 and 8 and their cytotoxicity effects on macrophages (RAW 264.6), lung carcinoma cells (A549) and hemolysis of human erythrocytes 15 . The pure silica particles showed a porosity- and shape-dependency on hemolytic activity, with the porous MSNs with higher AR exhibiting reduced hemolytic activity, whereas for the amino-functionalized particles, the zeta potential (effective surface charge) was decisive 15 . On this note, we want to stress that as we have particles of distinctively different pore sizes (3 to 4 nm vs 11 to 12 nm), this might give rise to differences in hemolytic activity as it has also been pointed out earlier that porosity is a decisive property due to lower exposure of silanol groups to the cell membrane of a porous surface 36 . However, as surface modification eliminates this source of interaction, the porosity effect might be of subordinate importance for surface-modified MSPs. In the study by Yu et al., surface charge and porosity also governed cellular toxicity, whereas AR did not have any reported effect 15 . Also in our current study, no differences in cytotoxicity were observed at doses relevant for our experiments and higher (2.5 times our maximum used dose). We also note that the two different pore sizes of our chosen MSPs are rationalized based on applicability for different types of therapeutic cargo, whereby the 3- to 4-nm pores are ideal for the loading of small-molecular drugs, whereas the large-pore materials would also be suitable for carrying of biomacromolecules such as proteins (antibodies, enzymes, polypeptides and so on) and genes 37 38 . Finally, based on the troublesome nature of the fluorescein label, a more quantitative/absolute approach to studying the shape effect on nanoparticle uptake with nanomedical prospects and the efficiency of different particles in cargo delivery could constitute a more correlative approach. To date, this has been successfully measured in the extent of causing cytotoxicity upon delivery of the chemotherapeutic agent camptothecin or paclitaxel 13 or the efficacy of GFP knockdown 39 upon siRNA delivery, and similar approaches will also be pursued in our ongoing and future studies.

Porous silica particles of spherical and rod-like morphologies were studied for cellular uptake efficiency in two different cancerous cell lines for potential applications as nanomedical drug delivery carriers. According to the obtained results, both rod-like and spherical particles were readily internalized by HeLa cells with slight shape and charge-induced differences, whereas in Caco-2 cells, rod-shaped particles were internalized more efficiently. The difference was most pronounced for uncoated particles in both cell lines, whereby higher charge (S-MSP₁) also induced higher uptake. A net positive charge (PEI) enhanced uptake regardless of shape and cell line. At lower doses, surface charge could be used to fine-tune the uptake even in HeLa cells, whereby higher charge (+/−) results in higher uptake over net neutral charge and positive over negative. At higher concentrations, the surface charge effect is overridden in HeLa cells, and rods are taken up despite coating or not, whereas for spheres and Caco-2 cells, distinctions can still be made. Uptake studies performed in vitro in different cell lines show that along with particle shape and surface functionalization, cellular origin and features may also influence the uptake of particles in cells. As shape seems to influence uptake in a cell-dependent manner, shape engineering could potentially be used as a tool for enhancing nanoparticle-mediated delivery.

The authors declare no competing interests.

DSK carried out the synthesis, functionalization and characterization of mesoporous silica nanoparticles, analyzed and interpreted characterization data and some part of the in vitro test results and has been involved in drafting and designing of the manuscript. DD carried out the in vitro tests such as cell viability, confocal microscopy and some part of the FACS studies, analyzed and interpreted the obtained in vitro results and has been involved in drafting and designing of the manuscript. RS studied the cellular uptake of sphere and rod-like MSPs at different concentrations in HeLa cell lines adopting FACS. EMJ carried out the synthesis and characterization of mesoporous silica nanorods and has been involved in drafting the manuscript. NR carried out the preliminary uptake and cytotoxicity studies of differently functionalized nanorods. MO designed and characterized the mesoporous nanorods in addition to revising the manuscript critically and contributing with important intellectual content. JEE provided biological lab facilities along with CS and DMT. CS has been involved in revising the manuscript critically for important intellectual content. DMT has been involved in providing the cell cultures and reagents for toxicity testing and revising the manuscript critically for important intellectual content. JMR conceived, designed and coordinated the study and drafted/wrote the manuscript. All authors read and approved the final manuscript.