The increased application of nanoparticles (NPs) is increasing the risk of their release into the environment. Although many toxicity studies have been conducted, the environmental risk is difficult to estimate, because uptake mechanisms are often not determined in toxicity studies. In the present study, the authors review dominant uptake mechanisms of NPs in cells, as well as the effect of NP properties, experimental conditions, and cell type on NP uptake. Knowledge of NP uptake is crucial for risk assessment and is essential to predict the behavior of NPs based on their physical–chemical properties. Important uptake mechanisms for eukaryotic cells are macropinocytosis, receptor-mediated endocytosis, and phagocytosis in specialized mammalian cells. The studies reviewed demonstrate that uptake into nonphagocytic cells depends strongly on NP size, with an uptake optimum at an NP diameter of approximately 50 nm. Increasing surface charges, either positive or negative, have been shown to increase particle uptake in comparison with uncharged NPs. Another important factor is the degree of (homo-) aggregation. Results regarding shape have been ambiguous. Difficulties in the production of NPs, with 1 property changed at a time, call for a full characterization of NP properties. Only then will it be possible to draw conclusions as to which property affected the uptake. Environ Toxicol Chem 2014;33:481–492. © 2013 SETAC
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Nanomaterials (NMs) are defined by common consensus as particles with dimensions from 1 nm up to 100 nm (nanotechnology), even though “there is no scientific evidence to support the appropriateness of this value” . Their characteristics differ from those of macroscopic bulk material of the same chemical composition. In nanomedicine, NMs are defined as particles with dimensions up to 1000 nm . They can be further divided into 3 categories, as proposed by Hansen et al. , who noted that clarification of the terminology is needed and proposed a new categorization of NMs based on the location of the nanostructure in the material. Hansen et al.  differentiated among 3 main categories of NMs: “1. Materials that are nanostructured in the bulk, 2. Materials that have nanostructure on the surface, and 3. Materials that contain nanostructured particles. . . . Category 3 contains nanoparticles, which we define as free structures that are nanosized in at least two dimensions.” In the present study, we focus on nanoparticles (NPs) belonging to category III. Nanomaterials are manufactured because of their physical and chemical characteristics, including electrical, optical, transparency, hardness, and magnetic properties. They are being used in increasing amounts and applications [4, 5], including their use as fillers, coatings on surfaces, gas sensors, and carriers for industrial catalysts, as well as in electronics, water treatment, and environmental remediation [6, 7]. In addition, NMs are increasingly used in consumer products , as drug delivery systems for therapy [9-11], in imaging [12, 13], and in diagnostics [14-16]. Tuning of the characteristic properties yields great advantages, such as a reduction of side effects by targeted administration and improvement of detection sensitivity [17, 18].
Despite these advantages, possible adverse effects of NMs need to be considered, preferably during the design phase, so that design options for functionality and safety go hand in hand. Adverse effects on human or environmental health may occur as a consequence of NM release into the environment, for example, during the production, transport, usage, or waste phases of consumer products [7, 19], leading to exposure of humans and the environment. Hazardous effects of NMs have been observed at the cellular level, including the generation of reactive oxygen species, lipid peroxidation, genotoxicity and mutagenesis, apoptosis or necrosis , mitochondrial dysfunction, and changes in cell morphology , with potential repercussions for the individual . Potential ecotoxicity has been shown in algae, plants, fungi , fish, cladocerans , and amphibians , as well as in bacterial communities [8, 25-27]. Also, NMs could have adverse effects on populations  and entire food webs as a result of trophic transfer and biomagnification [29-34]. Larger impacts on the food web dynamics are possible even at sublethal concentrations,because of changes in the behavior of organisms [35, 36].
Unfortunately, most of the toxicity and ecotoxicity studies have been conducted without determining the uptake mechanism in detail, as noted by Zhao et al. . More knowledge of NM properties and the experimental conditions that determine the transport and uptake across cell membranes will improve our understanding of their toxicity, and such information might be helpful in designing safer NMs. Toxic effects usually originate from the presence of the toxicant inside cells . In addition, toxicity is determined by the entrance route and the final intracellular localization . Hence, knowledge of NM uptake is crucial for risk assessment studies on NM toxicity.
It is known that the physicochemical properties of NMs, in combination with characteristics of the cellular and extracellular media (e.g., protein or lipid adsorption patterns) govern the localization in the target cells [38, 39]. Understanding the effects of these various properties and characteristics will also be beneficial in pharmacology. Several reviews have been published on aspects of uptake in relation to NM properties and experimental conditions. Alexis et al.  described the factors determining blood residence times and organ-specific accumulation of NMs, while Ahsan et al.  focused on uptake by macrophages. Thevenot et al.  reviewed uptake for implant biocompatibility purposes. Verma and Stellaci  addressed various NM properties but did not include the role of experimental conditions in detail. In addition, studies have reviewed the uptake of a specific NM such as gold [44, 45]. To the best of our knowledge, however, no overall review including the most important factors controlling the cellular uptake of various NPs has yet been published.
In the present review, we review uptake mechanisms of NPs in different cell types and identify particle properties and experimental conditions that determine transport across cell membranes, describing the mechanisms that play an important role in NP uptake. Identification and understanding of the factors that determine NP uptake by cells is the first step toward predicting NP accumulation and toxicity based on easily measurable properties, such as size and zeta potential. Such predictions have already been achieved in risk assessment of conventional chemicals .
Ecotoxicity studies are often conducted using whole organisms, [19, 32, 33, 36], so mechanistic studies on NP properties and their effects on nonhuman cell lines are very scarce, and many of the cell lines mentioned in the Methods section are human cell lines. Translation of information obtained from human cell lines to cells from other organisms should, in principle, be possible. Pinocytosis, a fundamental, vital uptake mechanism included under the broad name of endocytosis, is carried out by most [47, 48] (essentially all) eukaryotic cells [49, 50], as is clathrin-mediated endocytosis . Receptor-mediated endocytosis is also known to be common to virtually all eukaryotic cells, except the mature erythrocyte . Some researchers have gone a step further and stated that endocytosis, which is, for example, responsible for nutrient uptake into cells , exists at least across most [54, 55] if not all eukaryotic cells . Thus, endocytosis is ubiquitous in all eukaryotic cells, enabling comparison of results from different cell lines. Humans can come into contact with NPs by direct exposure (e.g., during handling or use of NP-containing products) or indirectly through the food chain. Because biomagnification of NPs has been suggested [57, 58] and some evidence for accumulation within the terrestrial food chain has been presented , this route of human exposure should be taken into consideration in modeling. Combining modeling and empirical studies allows for a cost- and time-effective risk assessment of NPs.
The literature search was conducted via Web of Science, Google Scholar, and Scopus, using the following key words: nanoparticles, (size-dependent) cellular uptake, endocytosis, particle properties, biodistribution, virus and bacteria entry, macrophages, asbestos, and fine dust. The search resulted in over 100 publications, although many of them referred to toxicity rather than accumulation. The studies cover a broad range of disciplines, including biomedical and chemical engineering, environmental science, materials chemistry, medicine, pathology, and microbiology, as well as particle and fiber toxicology. The studies involve many different cell lines from both exchanging and nonexchanging organs, such as HeLa, SNB19, STO, A375, A549, Jurkat, and different phagocytes. Nearly 30 different cell lines were used in the various studies. The studies also cover a broad range of NP types (∼10) such as latex beads, silica particles, hydrogels, and various metal NPs (gold, silver, iron) with a variety of surface modifications. Information on the cell lines and NP types used can be found in the Supplemental Data, Table S1. The data presented in Table 1, Table 2, and Table S1 are taken directly taken from the corresponding references. If the sizes are given for single NPs and no measurement of aggregation was conducted, the concept of primary particle size is assumed to be valid. If no values were given for the surface area (SA) and volume (V), the data provided on the diameter and length of the particles were used to calculate them, assuming a spherical or cylindrical shape. The most important prerequisite for the selected studies was that 1 NP property had been changed at a time and the respective values are presented in the article, as this allows the property responsible for a change in NP uptake to be identified. Because of the limited number of studies in which only 1 particle characteristic was changed at a time, the only further restriction placed on the literature used for the present study was the description of the pathway of uptake.
|Size (nm)||Ligand-induced, particulate 500–1000||Non-specific 100–5000||Vesicles approx. 120||Vesicles approx. 80||[50, 71, 78, 79, 81, 160]|
|Cell type||Specialized mammalian||Non-specialized eukaryotic|
|Sphere ø (nm)||50 > 100 > 200 > 500, 1000 not taken up|||
|20 > 200||Optimum at 50||[68, 69, 74, 86, 93-95]|
|300 > 150, 500 > 150||150 > 300, 150 > 500|||
|Shape, ø, and L (nm), V (µm3), SA (µm2)||ø 150, L 450, V 7.9 × 10−3, SA 2.5 × 10−1 > ø 200, L 200, V 6.3 × 10−3, SA 1.9 × 10−1|||
|ø 150, L 450, V 7.9 × 10−3, SA 2.5 × 10−1 > ø 100, L 300, V 2.4 × 10−3, SA 1.1 × 10−1 (AR= const)|
|2000 > 4000 > 8000 in vivo; 1800 > 2500 > 3000 in vitro (const. ø)||ø 74, V 212.2 × 10−6, SA 17.2 × 10−3 > ø 14, V 1.4 × 10−6, SA 0.6 × 10−3 > ø 14, L 40, V 6.2 × 10−6, SA 2.1 × 10−3 > ø 14, L 74, V 11.4 × 10−6, SA 3.6 × 10−3||[86, 97]|
|ø 50, V 6.6 × 10−5, SA 7.9 × 10−3 > ø 14, V 1.4 × 10−6, SA 0.6 × 10−3 > ø 20, L 30, V 9.4 × 10−6, SA 2.5 × 10−3 > ø 14, L 50, V 7.7 × 10−6, SA 3.5 × 10−3 > ø 7, L 42, V 6.2 × 10−6, SA 0.6 × 10−3|||
|ø 80–150 = ø 80–150, L 400–1000||ø 100, L 450, V 35.3 × 10−4, SA 15.7 × 10−2 > ø 100, L 240, V 18.9 × 10−4, SA 9.1 × 10−2 > ø 100, V 5.2 × 10−4, SA 3.1 × 10−2||ø 80–150 > ø 80–150, L 400–1000||[98, 99]|
|35 > 25 > 15||35 > 25 > 15|||
|Negative||Increase||–15 > –25 > –40||[70, 102]|
|–40 > –25 > –15||–43.10 > –16.26|||
|Positive vs negative||Positive > negative||Positive > negative +32.2 ± 8.19 > –26 ± 1–42.46 > +41.59||[40, 70, 104-106]|
|Opsonization/presence of serum||No change||[143-145]|
|Decrease||Decrease, highest for most hydrophobic NP|||
RESULTS AND DISCUSSION
Mechanisms for the uptake of particles into cells
Cell membranes tend to be impermeable to many large particles [59, 60]. Particles can cross the membrane only if they are at most 10 nm to 30 nm in size  (see the more detailed discussion below). The cell membrane acts as a barrier separating the external environment from the inside of the cell, allowing substances to be concentrated in cells or excluded from cells . One mechanism to overcome this barrier is endocytosis. This mechanism is exploited by viruses [60, 63-65], which have therefore become a valuable tool to study endocytosis . Endocytosis and diffusion have been proposed as mechanisms for the uptake into cells of NPs with similar sizes as viruses [66-70].
Endocytosis is the uptake of particulate matter, such as proteins and other nutrients, into eukaryotic cells via enclosure by the cell membrane. It is also exploited by these cells for the clearance of cell debris and foreign cells from the body. Endocytosis has been defined as “the de novo production of internal membranes from the plasma membrane lipid bilayer” . Several forms of endocytosis are distinguished, based on the substance to be internalized (see Figure 1).
Opsonized particulate substances and small solute volumes are internalized by a mechanism called phagocytosis . Opsonins are small molecules that enhance binding in phagocytosis (e.g., antibodies) . Particles in large amounts of solute, on the other hand, are taken up by a process termed pinocytosis. Pinocytosis can be further divided into macropinocytosis  and receptor-mediated endocytosis (RME) [67-69, 74]. In view of the large differences between phagocytosis and nonphagocytosis, the results of the different studies are discussed separately below according to the mechanism investigated.
Phagocytic cells are found in all types of organisms, ranging from unicellular organisms, in which phagocytosis represents a form of feeding , to complex, multicellular animals . In animals, phagocytosis is part of the immune defense and is typically restricted to specialized cells, such as macrophages, monocytes, and neutrophils. Phagocytic cells specialize in the uptake and degradation of foreign (infectious) cells (e.g., bacteria) and cell debris and apoptotic cells belonging to the organism itself , with a diameter between 0.5 µm and 10 µm. Phagocytosis is a ligand-induced process . In addition to phagocytosis, this cell type also uses pinocytosis. Nonphagocytic cells make use of endocytosis to take up small essential nutrients (e.g., cholesterol-laden low-density lipoprotein and iron-laden transferrin) . Pinocytosis, the mechanism responsible for the uptake of soluble substances, can be divided into different processes according to the size of the vesicle and the protein involved in vesicle formation. The 4 mechanisms can be described as follows: 1) macropinocytosis, a nonspecific mechanism by which fluid contents are taken up in the same concentration as in the surrounding medium , and the vesicle size is 100 nm to 5 µm ; 2) clathrin-mediated endocytosis, receptor-mediated endocytosis with vesicle sizes of approximately 120 nm [50, 80]; 3) caveolae-mediated endocytosis, uptake by RME of vesicles of approximately 80 nm in size ; and 4) clathrin- and caveolin-independent endocytosis with vesicles of approximately 50 nm. Several clathrin- and caveolin-independent pathways exist [50, 71, 81], but these are not discussed further in the present review, because this uptake pathway was not examined in any of the studies covered in the present review.
The maximum dimensions of vesicles formed for each pathway are not definitive, but their dimensions are likely to be limited to allow internalization of confined cargo sizes . The sizes of vesicles differ for each species and cell type and depend on the cargo size [67, 83], be it NPs or viruses [64, 65]. Therefore, a maximum NP size for internalization can be assumed to exist. Particles above a size limit of 200 nm will not be internalized via a single clathrin-coated vesicle  or can presumably hardly be internalized via classical endocytosis .
Macropinocytosis is a constitutive process occurring continuously and irrespective of the cell needs in highly ruffled regions of the plasma membrane. Ruffles trap the surrounding medium, including solutes, when their tips bend back toward the cell surface. Materials adsorbed onto the cell membrane can be taken up as well. This mechanism is called adsorptive macropinocytosis .
Receptor-mediated endocytosis is well studied and plays an important role in the study of nonphagocytic uptake. This process allows different types of ligands—such as toxins, cholesterol-carrying proteins, vitamins, and iron transport proteins, as well as hormones and growth factors —to enter cells via the binding of specific receptors localized on the cell membrane . Receptor-mediated endocytosis allows the uptake of specific macromolecules from the surrounding medium in a concentrated form ; it is also exploited by viruses to enter host cells, leaving no evidence of entry (viral glycoproteins) on the outside of the cell for detection by the immune defense . To be able to enter, viruses first bind to the surface of a cell to concentrate on the cell surface or to induce transmission of cell signals for endocytosis . Whether RME or another type of endocytosis takes place is determined, among other factors, by the particle size . For particles, the rate of RME is determined by interplay of the decrease in free energy required to drive the cargo into the cell, the amount of available binding sites, and the wrapping time . Nanoparticle uptake by endocytosis is a complex interplay of the energy release by the number of bound receptors, the number of vesicles that can be formed from the membrane, receptor diffusion toward the attached particle, and time to complete this process. For large NPs, allowing only 1 particle to be taken up per vesicle, attachment occurs to numerous receptors at a time, resulting in a major change in free energy. A larger contact area with the cell membrane for elongated particles will also lead to a larger decrease in the free energy , but at the same time to less available binding sites for other particles . Large particles will require more time to complete wrapping [74, 85]. For small particles, several NPs can be taken up in 1 vacuole, but energy release is weaker because only single attachment sites exist. Because a local decrease in the Gibbs free energy is required to induce membrane wrapping, a minimum size for a single particle at a given ligand density must exist; otherwise endocytosis is energetically impossible . Uptake is only initiated after a certain number of receptors have been triggered. If the change in free energy is too small, membrane wrapping will not be induced .
In most studies reviewed, the NPs were larger than the reported size cutoff for diffusion. In addition, the uptake mechanisms were determined in each study and were reported to be phagocytosis or pinocytosis. Therefore, diffusion will only be discussed briefly in the present review. Whether nonendocytic uptake mechanisms, such as diffusion or active transport, can also be exploited by NPs is under debate . Diffusion is the movement of molecules from high to low concentration and takes place for small, hydrophobic, uncharged molecules such as oxygen or carbon dioxide. Lipid-soluble substances, such as alcohols, are also able to diffuse through the membrane. Water-soluble molecules pass through the membrane passively via pores in a process called facilitated diffusion. The pores allow only molecules of a certain size range and electrical charge to cross the membrane. For molecules too large or too specialized, carrier molecules can help to pass the membrane actively against the concentration gradient . A study by Geiser et al.  showed that particles (78 nm, 200 nm, and 1000 nm) taken up by cells were not membrane bound, indicating transport via pores or diffusion as a potential route. However, charged molecules could not simply pass through the plasma membrane by diffusion, and particles had to be 10 nm to 30 nm in size, the size of a channel pore, to allow movement via such channels . Nanoparticle transport by transporters was considered unlikely because of the high specificity of the transporter structure. A plasma membrane that is not clearly distinguishable around NPs is not enough evidence for diffusion being accepted as a viable cellular uptake route for NPs . In addition, size cutoffs of 1.47-nm diameter for organic chemicals  and 4.8-nm diameter for spherical proteins have been reported [89, 90]. The latter diameter was calculated [89, 90] from the borderline for proteins to cross the bacterial cell wall by diffusion, given as 50 kDa . The latter study  concludes that only small uncharged hydrophilic molecules with mean radii below 2.5 nm were able to pass the wall of 2 bacterial species. Chain lengths of hydrophobic chemicals longer than 5.3 nm have been reported to lead to a lack of accumulation .
Nanoparticle properties influencing their uptake
Internalization of NMs in cells is influenced by a large number of physical and chemical properties of the specific NP (Figure 2) and the circumstances in the exposure medium of the cells. The present study focuses on NP properties that are most broadly covered in mechanistic studies: NP size [67-70, 74, 86, 93-95], shape [74, 86, 96-99], surface charge [70, 100-106], surface functional groups [40, 73, 107-109], and NP hydrophilicity [68, 110-114]. The effect of core composition is not discussed because the surface characteristics are more important than the bulk characteristics, as the surface properties determine the protein corona and thereby possibly the biological impacts . Experimental conditions (cell type, aggregation, opsonization) are discussed with regard to studies on the effect of NP properties on cellular uptake. The major properties of NPs determining interaction with cells are depicted in Figure 2 and described below. All values reported are taken directly from the literature; therefore our results are also based on the assumptions of the researchers. Implications of assumptions such as no aggregation or no formation of surface layers are discussed in the Experimental conditions determining NP uptake section.
Spherical NPs up to 500 nm in size were taken up by nonphagocytic cells, whereas no particles with a diameter of 1000 nm were detected in these cells (Table 1). Previous research found a strong decrease in bead internalization and in the speed of the larger particles in comparison with 50-nm beads . This is in good agreement with the results of He et al. , who found that several nonphagocytic cells favored the uptake of smaller particles. They also found a clear relation between the size and number of gold particles being stabilized with citric acid ligands in each HeLa cell. There was an uptake optimum for spherical NP of 50 nm . In the resumption of the work of Chithrani and Chan, the same size optimum was for transferrin-coated gold NPs, independent of cell line . This size optimum of 50 nm has been confirmed for various NPs and cell types [69, 93-95]. The results of Rejman et al.  contradict the maximum uptake size assumption, stating that beads as large as 500 nm were taken up by caveolae: “Evidently, the actual size of single flask-shape caveolae is too small for accommodating particles as large as 500 nm. . . . Recruitment of an internalization machinery is needed to accommodate particles that extend beyond the actual size of a caveolae domain, which appears possible given internalization of bacteria and viruses along this pathway” . This suggests that macropinocytosis can only be a minor mechanism for the internalization of NP, as was proved by the use of specific inhibitors.
Lévy et al.  doubted that there was a size optimum for uptake, as observed in several studies [69, 74, 86]. They argued that the units in which the results were given had not been considered. An optimum based on the number of particles per cell does not necessarily lead to the same optimum in terms of mass . Most of studies have reported the number of particles per cell, except Lu et al. . Yue et al.  expressed internalization of different-sized NPs by 3 different units (number of particles per cell, particle volume per cell, and particle surface area per cell). They showed graphically that the size optimum varies depending on the unit. The surface area to volume ratio decreases with increasing volume. In addition, the sizes given usually refer to pure NPs after synthesis, but changes might occur during the experiments as a result of aggregation or attachment of serum components. This in turn may have altered the outcome of the experiments.
The effect of NP size on the uptake into phagocytic cells is less well studied (Table 1). The macrophage-like J774.A1 murine cells readily took up both 20 nm and 200 nm COOH-modified coated polystyrene beads, but the smaller polystyrene beads were taken up relatively more quickly and more extensively . This contradicts the observation that phagocytic cell lines favored the uptake of larger particles . The highest attachment has been observed for particles with a longest dimension of 2 µm to 3 µm , which also represents the largest dimension of commonly found bacteria . One possible explanation for such contradictions is “the difficulties in controlling the surface charge during the size control processes and the lack in surface functionality” . Overall, particle uptake by nonphagocytic cells shows a clear trend: uptake increases with particle size to an optimum of approximately 50 nm and decreases for larger particles [67, 74, 86, 93-95]. For phagocytes, the uptake is not unequivocally related to particle size.
Shape and geometry
The effect of the particle shape of cationic, monodisperse cross-linked polyethylene glycol (PEG) hydrogel particles on uptake has been studied in HeLa cells . Pinocytosis, in parallel nonspecific and receptor-mediated, was the determined uptake mechanism. The particles consisted of cylinders with different aspect ratios (ARs) and a constant surface charge. Direct comparison of particles with the same AR showed that the ones with a smaller diameter were taken up to a lesser extent than the ones with a larger diameter and much larger volume (2.4 × 10−3 µm3 vs 7.9 × 10−3 µm3) . Particles with a diameter of 150 nm and a length of 450 nm were taken up 4 times faster by the cells than those with a diameter and a length of 200 nm, despite their similar volumes of 7.9 × 10−3 µm3 and 6.3 × 10−3 µm3, respectively. In agreement with this study, another found that cellular uptake by nonphagocytic A375 human meloma cells increased in terms of numbers and speed with the aspect ratio of mesoporous silica NP (MSNP) spheres and rods of constant diameter . The uptake of rod-shaped gold NPs with citric acid ligands and their spherical counterparts into HeLa cells has been compared by Chithrani et al. . The same trend of a size optimum that was found for spherical particles was also seen for rod-shaped NPs. The larger rods with 1 dimension of more than 50 nm were taken up less than the smaller rods with a length of 40 nm . In contrast to the study by Huang et al. , an increase in AR led to a decrease in uptake as shown by Chithrani et al. . The larger contact area of rod-shaped NPs with the membrane led to a reduction in available binding sites . In another study by Chithrani and Chan, the same trend was found for Au NPs coated with transferrin . Wrapping times increased with increasing surface area of the transferring-coated NPs . The study by Gratton et al. underpins the dependence of internalization kinetics of NPs by HeLa cells on the absolute particle size rather than on the AR .
To be able to draw conclusions about the effect of particle shape on uptake despite experimental limitations and problems in the production processes of nonspherical NPs, modeling of the interplay of receptor–ligand binding and transport time across membranes has been performed for single particles and particle clusters . In computer simulations, the volume and ligand density of differently shaped NPs was kept constant. Particles were designed as spheres, long and short rods, and disks. It was shown that endocytosis can be divided into 2 steps: membrane invagination and particle wrapping . During the first stage, NPs oriented toward the membrane in such a way that ligand–receptor binding became largest by maximizing the contact area. This caused the membrane to invaginate. The kinetics of the second stage, particle wrapping, were governed by the largest local mean curvature leading to a reorientation of shape anisotropic NPs (for example, rods and disks). One requirement for successful internalization was the release of sufficient free energy by ligand–receptor binding to overcome the energy required for membrane bending during wrapping .
Trewyn et al. showed that the rate of uptake of MSNP in fibroblasts with deficient phagocytosis decreased with an increase in AR . The uptake mechanism was not determined in the same study as the rate of uptake, but rather in a previous study . In that study, the same kind of particles was shown to be taken up into HeLa cells by clathrin-mediated endocytosis . The effect of the cell line on NP uptake is discussed further under the Experimental conditions determining NP uptake section. The rod-shaped particles used by Trewyn et al.  were much larger (length 400–1000 nm) than the dimensions of vesicles formed in clathrin-mediated endocytosis by fibroblasts. These vesicle sizes are not definitive, but are likely to be limited to confined cargo sizes . Uptake curves determined by Trewyn et al. showed the same shape and saturation at the same time point for spherical and tubular MSNP by nonprofessional (sessile) phagocytic Chinese hamster ovary cells . At the same time, a tendency of tube-like NPs to form larger aggregates with a wider size distribution of 955 nm to 1480 nm than the spherical particles, which had a sharp peak at 712 nm, was reported. This study confirms that the concept of primary particle size is not applicable under some experimental conditions (see further discussion in the Aggregation section). The same might be the case in the environment where aging and aggregation possibly take place before the NPs reach an organism. Contradictory results have been reported for in vivo studies: A decrease in AR led to increased cellular uptake, measured as decreased circulation times in rodents. In agreement with this, spheres and short particles showed a higher uptake in human-derived macrophages in vitro .
The effect of shape on NP uptake using 6 different geometries (spheres, oblate ellipsoids, prolate ellipsoids, elliptical disks, rectangular disks, and UFOs) in alveolar macrophages has been studied by Champion and Mitragotri . The shape was described by the angle of tangents of the particle as related to the membrane at the first point of contact. Shape rather than size determined the induction of the phagocytosis process. Phagocytosis was induced by the formation of a complex actin structure for angles smaller than 45 degrees, while spreading of the membrane took place for larger contact angles. Size determined whether internalization was successful. When the volume of the particle was larger than that of the cell, phagocytosis was not completed .
The AR alone does not offer enough information to draw conclusions about NP uptake— for example, whether a certain size for vesicular uptake limit can be exceeded (Table 1). Particles with the same AR may have very different dimensions, such as length, diameter, or volume, which affect the uptake and have to be considered. The studies described above have been ambiguous regarding the effect of AR on nonphagocytic uptake. It has been shown that RME increases with a smaller AR . For macropinocytosis or an undefined nonphagocytic uptake mechanism, the higher AR led to increased uptake . In phagocytosis, uptake increased for a smaller AR or stayed the same, depending on the cell type . However, it is not clear whether this can be attributed to a change in length or volume, or to both. Whether the length or the volume is dominant is difficult to assess, because studies in which 1 factor was changed at a time are lacking. The reason appears to be the difficult production of spheres and rods with equal chemical composition .
Zero surface charges, either by neutral surface groups (e.g., hydroxyl groups) or by zwitterionic ligands, have been shown to lead to low cellular uptake compared with charged particles [70, 100-102, 106]. This can be explained by the low NP affinity toward the overall negatively charged cell membrane. Zero surface charges cause the hindrance of nonspecific protein adsorption [123, 124], and a strong hydration layer via electrostatic interaction may be formed .
A positive surface charge has been shown to increase particle uptake, in both phagocytes [70, 103, 108] and nonphagocytic cell lines [70, 108]. Positive surface charges (NH2) increase cell surface affinity and uptake of NPs in both cell types. The uptake is determined by electrostatic attractions toward the slightly negatively charged cell membrane . Increase of the NP uptake by macrophages has been observed for negative surface charges [70, 102]. For uptake of negatively charged NPs by RME, opposing results have been reported. On the one hand, increase in cellular uptake was observed with increasing charge . On the other hand, decreasing uptake was reported by He et al. . The possibility of changes in other surface properties during the charge control process, such as the hydrophobicity, has to be kept in mind. These might affect the uptake into cells as well. At first glance, an increase in transport of more negatively charged NPs across the negatively charged cell membrane does not seem logical because of repulsion. However, high charges prevent agglomeration, and particles with a low zeta potential tend to aggregate, lowering their uptake rate [19, 126]. Therefore, the uptake was possibly not determined by the lower zeta potential but by the likely bigger size of the agglomerates formed. The issue of particle aggregation will be discussed in the Aggregation section. When positive and negative surface charges were directly compared, keeping other properties the same, positively charged particles were shown to be taken up by nonphagocytes more favorably than negatively charged particles [70, 104]. The same finding was reported for phagocytes [70, 105].
In most studies an increase in positive or negative surface charge led to increased NP uptake in comparison with less charged or uncharged NPs. Only 1 study contradicted these results . In addition, positively charged particles seemed to be taken up more extensively compared with negatively charged NPs when the absolute zeta potential was similar [70, 104].
Surface functional groups
Investigations on the effect of surface functional groups often include PEG, which will be described in the next section because of its high hydrophilicity, the negative carboxyl group (–COOH), neutral functional groups like hydroxyl groups (–OH), and the positive amine group (–NH2). Increasing positive surface charge, by increasing amounts of amine groups, has been shown to increase uptake of NPs into cells for various cell lines, both phagocytic and nonphagocytic, and for various NPs [40, 107-109].
Carboxyl (–COOH) functional groups add a negative charge to the NPs and have been shown to increase NP uptake in both cell types. The uptake studies were conducted in a medium containing protein (fetal calf serum) . This might have a great effect on particle uptake, as will be described in the Opsonization section. In contrast, carboxyl groups on dendrimers led to increased residence times in vivo. One possible explanation is the resistance to recognition by the immune system through protein adsorption . These opposing results can be attributed to other NP properties, such as the hydrophilicity or differences in experimental conditions influencing NP uptake and vary widely between different studies (in vitro vs in vivo).
Increasing uptake has been observed for hydroxyl (–OH) functional groups via a nonspecific adsorptive uptake route in nonphagocytic cells . Uptake is increased by the adsorption of serum proteins onto the negative surface by electrostatic and hydrophobic interactions. The negative surface of the NPs under investigation did not determine particle uptake, but the outer shell of serum protein . In a similar manner, ligands incorporated into the NP surface, such as mannosyl, immunoglobin, fibronectin lipoprotein, or galactosyl, have been found to increase particle uptake by interaction with phagocyte surface receptors .
Hydrophilicity and lipophilicity
Polyethylene glycol is a highly hydrophilic molecule. An increase in hydrophilicity led to decreased particle uptake by macrophages [68, 110-112], as well as in nonphagocytic HeLa cells . This effect was explained by the surfactant properties of PEG [110, 111, 130]. The hydrate shell around the particle hinders the adhesion of large molecules. In addition, molecular mobile PEG chains present a steric hindrance for binding of opsonins to the particle. Without the binding of opsonins, uptake into phagocytes is prevented because the necessary recognition is suppressed. Amphipathic polymers showed a very similar effect as PEG. Their hydrophobic properties allowed adsorption on the NP's surface, while their hydrophilic part pointed toward the aqueous solution and hindered the adsorption of plasma proteins, leading to expanded longevity of the NPs . Another particle property, flexibility, also decreased phagocytic uptake in comparison with rigid particles .
Experimental conditions determining NP uptake
The experimental conditions found to have the greatest effects on the uptake of NPs will be described subsequently. They might explain the observed differences in studies on the effect of NP properties on their uptake. The results of uptake or toxicity studies, described in the subsequent sections, were retrieved from in vitro experiments using cell cultures or unicellular microalgae . Kinetic and sorption studies are done without the use of cells [123, 124].
Cell properties have been shown to affect NP uptake. As a result, uptake is related to differences between cell types, (e.g., phagocytic vs nonphagocytic, cancer vs normal cells, and monocytes vs macrophages) . Cancer cells have been shown to express different amounts of receptors on their surface than normal cells. This may affect the available binding sites of cargos and their uptake [134, 135]. In addition, the metabolic activity of the cells used may affect particle uptake, as has been shown with contradictory results [136, 137]. Not surprisingly, small differences in uptake kinetics and amount of NPs taken up are observed across species . Phagocytes and nonphagocytes have the same biological function across all species (see Mechanisms for uptake of particles into cells); therefore the observed differences between cells of the same endocytosis type are rather small, and the trend in uptake is the same. Differences can be explained by small variations in the composition of the cell membrane, the available surface area, and the cell volume or size.
Aggregation plays an important role in determining the mobility, fate, persistence, and toxicity of NPs [138-141]. It depends on experimental conditions and NP properties. Zwitterionic ligands such as phosphoryl choline stabilize NPs, preventing aggregation under numerous conditions, such as the presence of negatively or positively charged proteins . The absence of nonspecific interaction has also been demonstrated for zwitterionic disulfides; this led to increased colloid stability in the presence of salt, negatively and positively charged polyelectrolytes, and positively and negatively charged proteins. The stabilization of citrate-capped and zwitterion-capped NPs was compared. Stabilization of zwitterion-capped NPs was observed under more experimental conditions (proteins and polyelectrolytes) than for citrate-capped gold NP . Aggregation can be prevented by sonication and shaking. However, aggregation may be the result of prolonged sonication itself . Shaking of the NPs and cells in solution may increase particle uptake into cells because of a higher hitting frequency . Dispersants may be used to prevent agglomeration, but their use may alter the bioavailability and uptake of NPs . Changes in uptake are the result of increasing amounts of NPs delivered to the cell membrane inadvertently by the interaction of lipid-soluble dispersants with the membrane. Alternatively, coating of the NPs with the dispersants may alter the interaction between the cell surface and the NP. Aggregation is also influenced by the ionic strength of the medium and directly influences particle uptake. Compression of the electric double layer with increasing ionic strength takes place. The energy barrier of the electrostatic repulsion of 2 particles of the same charge becomes smaller and the attachment probability in turn becomes larger .
Opsonization (marking for ingestion and destruction by a phagocyte ). Serum proteins can also become present on the surface of NPs when serum is used in the uptake experiments, similar to the addition of ligands in the preparation step of NPs. After addition of serum, particle uptake by phagocytes has been reported to be indifferent , to increase [143-145], or to decrease , depending on the kind of protein present (Table 2). These contradictory effects can be explained by the presence of different proteins in the medium, by a dynamic change of the protein corona composition over time because of protein abundance and affinity [115, 146, 147], and by possible particle opsonization. Lynch et al.  hypothesize that the subcellular localization of an NP is determined by the NP–protein corona, in addition to size and shape. Lundqvist et al. found in their study  that the corona composition depends on both size and surface properties of the NPs (plain polystyrene, carbocyl-modified, amine-modified). It is proposed that such “surface modifications are able to entirely change the nature of the biological active proteins in the corona, and thereby possibly also the biological impacts” . An increasing amount of serum present during incubation with nonphagocytic cells causes a decrease in cellular particle uptake. The degree of decrease depends on the particles' hydrophobicity. Increasing hydrophobicity increased the effect. The hydrophobicity of the particle controls the binding to hydrophobic pockets of bovine serum albumin. The anionic bovine serum albumin shows greater avidity toward the more hydrophobic NPs and repulsive interaction with the negative cell membrane . Serum shows different effects on particle uptake by phagocytes [143-145] and can be explained by the kind of proteins present in the medium, the protein corona composition, and particle opsonization [150, 151]. In addition, the NP charge plays a dominant role because proteins can be positively (e.g., lysozymes) or negatively (e.g., bovine serum albumins) charged . The differences in NP uptake by nonphagocytic cell lines can be explained by various properties of the cells such as different binding proteins present at their surface, the charge within and on the cell surface, and the rate of receptor recycling [47, 134, 152].
CONCLUSIONS AND RECOMMENDATIONS
The present review shows that patterns, albeit sometimes uncertain, are gradually emerging. In general, the following trends can be observed: uptake of NPs in nonphagocytic cells shows an optimum at a particle size of around 50 nm, and for phagocytes, the results are inconclusive. Increased charge, either positive or negative, favors uptake in both cell types. When the absolute values are comparable, positively charged particles are taken up more extensively. Addition of amine or carboxyl groups leads to increased uptake as well. With increasing hydrophilicity, the uptake decreases. The presence of serum during incubation and whether opsonization takes place have both shown possible effects on uptake. This can be attributed to the type of proteins present. The effect of AR on NP uptake remains unsolved. The disparities might be the result of difficulties in the production of NPs with 1 physical–chemical parameter changed at a time, for example, changing the surface charge might affect the hydrophobicity as well. Therefore, to pin down the responsible property, either more exact studies only varying 1 parameter at a time are needed, or a multivariate approach can be taken. In the latter case, all NP properties that might be changed have to be measured and published in the future to give a complete picture of causes and effects.
In general, several contradictory data and conclusions from different studies were identified in the present study. The limited number of NP properties and the widely varied experimental conditions tested led to some uncertainty in interpretation of the results. Information about the properties of NPs that determine accumulation is very limited compared with data from toxicological studies. In particular, the effect of NP shape on uptake remains unresolved. This is because of a lack of commercially available techniques to produce nonspherical NPs and limited procedures to obtain NPs with only 1 property changed at a time. The exact reciprocity of shape and the responsible determinant (length, volume, or surface area) for NPs with constant surface chemistry still require experimental investigation. In most studies, it is assumed that particles are not aggregated. The particle size is often determined in the pure NP solution and their shelf life is detected. Effects of cell growth medium and biomolecules excreted by cells are not considered in those cases. Optical techniques like transmission electron microscopy or microscopy can allow a researcher to determine whether NPs are aggregated . If NP uptake is determined by other techniques, aggregation might remain undetected and be reflected in contradictory results . The test medium is an important factor in determining aggregation. Nanoparticles might readily aggregate under high ionic strength  or in a nutrient-rich culture medium . The occurrence of “agglomeration disqualifies the concept of primary particle size” . Therefore, the aggregation state should always be determined under conditions as close to the experimental conditions as possible.
Future developments will benefit most from studies conducted according to standardized protocols (for cell types and experimental setups), to overcome incomparability between studies conducted in different laboratories and to unveil more information about the underlying mechanism of particle uptake by cells. For detailed information about standardized protocols, the reader is directed to a critical review that suggests nano-specific modifications of test protocols . Reference materials for NPs should be made available to avoid differences in the nominally same end product supplied by various manufacturers  or to note when NPs are produced by the researchers themselves. Particle properties such as size and charge should be measured whenever possible because interlaboratory deviations  and differences in data from the supplier have been noticed [31, 157, 158]. Standardized test methods would possibly fill knowledge gaps and resolve apparent contradictions. However, standardized tests are unlikely to completely cover the large and continuously growing number of engineered NPs. Consequently, extrapolation between different NPs and between different experimental conditions is needed. For that, modeling is indispensable.
Table S1. (82 KB DOC).
The present study was supported by NanoNextNL, a micro and nanotechnology consortium of the Government of the Netherlands and 130 partners. We thank the anonymous reviewer for the detailed discussion and constructive suggestions that helped to greatly improve the final version.