Due to the nature of the tri-block copolymers synthesized and used in the present study, which have a hydrophobic middle block and hydrophilic terminal blocks, the PNP that were formed out of these polymers in water had a hydrophobic core and a hydrophilic corona. The size of the PNP depends largely on the size of the polymers and the ratio between their hydrophobic middle and hydrophilic terminal blocks . The hydrophobic middle blocks (the polyhexylene adipate polyester) avoid contact with water, whereas the hydrophilic (PEG) blocks try to remain in contact with the water. This means that for the polymers with the larger hydrophilic blocks the aggregate surface becomes crowded with PEG groups faster and the particles stop growing sooner than for the polymers with the smaller terminal groups. The polymers with the smaller PEG groups grow into larger particles, because in that case the surface does not get so easily crowded and more polymer molecules will add to the forming nanoparticle. The MTT assay data suggest that the positive PNP45 were more cytotoxic than the positive PNP90. Both the smaller and larger negative PNP did not induce any significant cytotoxic response at concentrations up to 12.5 μg/ml. Although some studies have been performed on the size-dependent cytotoxicity of NP, such studies on PNP are rare. In the present study the PNP45 and PNP90 that were used have very similar properties in surface charge density. To the best of our knowledge, this is the first report in which the effect of size on cytotoxicity of PNP is systematically investigated, while keeping the surface characteristics and PNP composition unchanged.
For inorganic NP more is known on the effect of size on cytotoxicity. Recently, it was reported that smaller gold nanoparticles (1.4 nm) were much more cytotoxic than bigger (15 nm) ones . In a similar experiment, silver NP of three different sizes (15, 30 and 55 nm) were tested on a rat alveolar macrophage cell line, which is comparable to the NR8383 cell line used in this study. The cytotoxicity of the smallest (15 nm) particles was highest and of the biggest ones (55 nm) the lowest . The cytotoxicity of a wide size-range of silica NP (30, 48, 118 and 535 nm) was tested on a mouse keratinocyte HEL-30 cell line. A clear size-dependent cytotoxic pattern was reported. The smaller (30 and 48 nm) silica NP showed a much higher toxicity than the bigger (118, 535 nm) ones . For copper NP similar results were reported . Some other reports also indicated an inverse relationship between size and toxicity of different NP including PNP [13, 33]. Auffan et al.  hypothesized that inorganic NP smaller than 30 nm are chemically very unstable due to the presence of many high energy surface states which makes them extremely reactive, which again results in an enhanced cytotoxicity. However, many of the surface properties of inorganic NP are significantly different from the surface properties of organic PNP. Therefore, although our results match the findings for inorganic NP, a true comparison is difficult.
It should be noted that in all the cellular experiments reported in this article, cell culture media (DMEM and F12-K) contained FCS rich in proteins (like albumin). It has been observed before that the presence of FCS can cause an increase in the sizes of these PNP by surface adsorption of proteins, although the PNP, in this case, still remained highly monodisperse . The presence of serum, by virtue of being high in protein content, has been reported to influence the toxicity and cellular uptake of NP . This protein adsorption can be of influence on cellular uptake and toxicity, but it is expected that such protein effects will also occur in the gastrointestinal tract upon oral exposure to PNP. Thus, testing in the presence of serum better reflects the physiological conditions for PNP that may be developed with food-based applications in mind.
Two series of PNP were investigated that differ in size and within each series PNP with different surface charges (-NH2, -OH and -COOH) were investigated. A distinct size-dependence was observed. For instance, although both amine-terminated PNP were toxic, the smaller PNP45 were more cytotoxic than the larger PNP90. Hence a size-dependent effect for PNP with comparable surface charge and surface functionalization was observed. In equivalent masses, the smaller PNP45 (45 nm) presented two times more surface area compared to the bigger PNP90 (90 nm). Upon expressing the toxicity data based on surface area, it was found that the toxicity increased with an increase in PNP surface area.
Production of intracellular ROS after exposure to different NP has been amply reported [35–39]. It is thought that these ROS push the cellular physiology to the limits by inducing oxidative stress. Our data, obtained from the DCFH-DA assays, show that only positive PNP induced intracellular ROS production with the smaller PNP showing a higher effect. These findings also match the pattern of cytotoxicity (MTT assay) for these PNP. In literature, systematic studies on the effect of size on the intracellular ROS production are rare. Jiang et al.  investigated the effect of size on intracellular ROS production by testing a wide range of titanium dioxide NP (4-195 nm) and reported the highest ROS induction for NP of 30 nm size. In a separate study, silver NP of 4, 20 and 70 nm were tested on macrophage U937 cells . It was found that the 20 nm silver NP were the most capable of producing oxidative stress. A similar study by Choi et al.  showed that within a series of different sizes of silver NP tested, the smallest (5 nm) NP were the most capable of inhibiting the growth of nitrifying bacteria through production of ROS. Landsiedel et al.  also mentioned an inverse relationship between the size of NP and their induction of intracellular ROS in their comprehensive review on different metal oxide NP (like CeO2, TiO2, SiO2, ZrO2).
The decrease of the mitochondrial membrane potential (ΔΨm) by cationic PNP is an important finding. It shows that cationic PNP can indeed interact with intracellular mitochondria and compromise their integrity. The decrease in ΔΨm after exposure to cationic PNP can have further consequences. This compromised state of the mitochondrial membrane can increase its permeability which may result in leaching of the mitochondrial calcium to the cytoplasm causing a cellular overload of calcium, release of cytochrome c and subsequently trigger apoptosis . Similarly, a compromised mitochondrial membrane also can hamper the normal electron transport chain. This can result in decreased ATP production. The finding of ATP depletion of cells upon exposure to cationic PNP matches well with the observed effect on the mitochondrial membrane potential (ΔΨm). Previously, Bhattacharjee et al.  reported that positive silicon NP (1.6 ± 0.2 nm) were able to induce ROS production in isolated mitochondrial fractions from rat liver tissue. Similarly, Xia et al.  observed that cationic polystyrene nano-beads can interact with and subsequently harm intracellular mitochondria. Due to the continuous involvement of mitochondria in the respiratory cycle by virtue of the electron transport chain (ETC) processes occurring on the outer membrane, it was suggested that interaction of positive NP with mitochondrial membranes can disturb the mitochondrial membrane potential. This was shown in the present study to occur upon exposure of the cells to positive PNP. As a result, positive PNP might uncouple the cascade of reactions in the ETC and thus not only hamper ATP production but also increase the intracellular ROS production . A recent study reported that intracellular ATP depletion occurred upon exposure of human endothelial EAhy926 cells to differently sized polystyrene NP . The data obtained in the present study are in line with the literature and this could shed some light on the poorly understood mechanism of intracellular ROS production induced by NP. In our opinion, reduction of the mitochondrial membrane potential (ΔΨm) followed by intracellular ATP depletion, as observed after exposure to cationic PNP, may also be a mechanism of cytotoxicity, related to or independent of intracellular ROS production and warrants further investigation.
It has been reported that several NP can induce production of inflammatory cytokines like TNF-α in different cell lines, including a primary rat brain microvessel cell line and human alveolar epithelial A549 cells [26, 47, 48]. However, a comparative study on the effect of NP size on TNF-α induction is rare, especially for PNP. Recently, size-dependent TNF-α induction was reported when titanium dioxide NP (5 and 200 nm) were intra-tracheally instilled in rats . It was observed that 5 nm particles were much more effective in inducing TNF-α than 200 nm ones. A similar type of inverse relationship between size of NP and TNF-α production was reported by Hanley et al.  for silver oxide NP. Our data on TNF-α production match these reported literature data and point towards an inverse relationship between size of PNP and TNF-α production.
Our data on the cellular uptake show that smaller PNP45 (45 nm) were taken up in appreciably larger amounts (as determined by CLSM) than the bigger PNP90 (90 nm), irrespective of surface charge. Win et al.  reported a similar type of inverse relationship between size of PNP and cellular uptake in Caco-2 cells. Recently, fluorescent and carboxyl-terminated polystyrene NP of 20 and 200 nm sizes were tested on both rat and human primary hepatocyte cells  and it was found that the smaller 20 nm polystyrene NP showed a higher intracellular uptake than the bigger ones. Many other groups reported a higher intracellular uptake for smaller NP [51, 53–55]. Zhang et al.  performed a molecular modeling and thermodynamics study to understand this size dependence. From calculations and thermodynamics they predicted that NP of ~22 nm radius (i.e. ~44 nm in size) are more easily internalized by cells. Similarly, other computational models [57–59] also predicted an energetically favorable receptor-mediated intracellular uptake for NP of 30-50 nm sizes. These authors were also able to predict an upper threshold radius of ~60 nm (i.e. ~120 nm size), where receptor-based endocytosis will not be favorable anymore. These findings fit quite well with our data.
The inhibition of endocytosis (by performing the experiments at 4°C or exposure to a mixture of 2-deoxyglucose and sodium azide) had a stronger effect on the uptake of PNP45 than on the uptake of PNP90. An explanation may be that smaller PNP enjoy a higher degree of binding with cell membrane receptors. Hence, inhibition of receptor-based endocytosis affects the cellular uptake of smaller particles more. An optimal size of 50 nm was proposed for uptake as well as saturation kinetics of NP uptake by Chithrani et al. , who investigated the uptake of gold NP of three different sizes (14, 50 and 74 nm) in HeLa cells . Recently, Jiang et al.  also reported that receptor-based endocytosis was highest for 40-50 nm gold and silver NP tested on herceptic receptor ErbB2 expressed on macrophage cells. This preference for 40-50 nm NP also matches our data.
Like in the general receptor inhibition studies, our data show that selective blocking of the clathrin, caveolin or, mannose receptors had in all cases a stronger effect on the uptake of smaller PNP45 particles than that of the larger PNP90 particles. These results are independent of surface charge of the PNP, although the preference of clathrin receptors for positive PNP and caveolin receptors negative PNP is evident from the results. However, it should also be noted that blockers lack absolute specificity and also that other endocytotic pathways for cellular PNP uptake remain available after blocking one receptor type. Furthermore, because different in vitro cellular systems have different physiologies and have different levels of expression of clathrin or caveolin receptors, results from different studies cannot always be easily compared . Whether combined inhibition of the clathrin, caveolin and mannose receptors would completely abolish the cellular uptake of the PNP, or that residual uptake would remain because also other uptake mechanisms are of importance, remains to be investigated. There is only a limited amount of systematically performed size-dependent analyses on interactions of PNP with endocytosis receptors. Rejman et al.  performed a very extensive study on cellular uptake mechanisms of latex particles in murine melanoma B16-F10 cells and found a preference for clathrin receptors by smaller and caveolin receptors by bigger PNP. More recently, it was reported that carboxyl-terminated polystyrene NP of 43 nm size got internalized by the cells through a clathrin-dependent pathway . Oh et al.  also found that uptake of smaller metal hydroxide nanoparticles (50, 109, 200 nm), when tested on a human osteosarcoma (MNNG/HOS) cell line, showed a stronger clathrin receptor dependence of cellular uptake than the bigger ones (375 nm). So, although the effects are clearly different for the differently charged PNP and the different receptors, our results seem to be in line with the findings of the majority of these reports that smaller (or medium-sized) NP have stronger interactions with endocytosis receptors than larger NP.
Mannose receptors are a unique group of receptors that are often expressed on macrophage cell surfaces and recognize and endocytose a wide variety of carbohydrates. Though it is not clear yet how these receptors recognize such a huge variety of molecules, the orientation of the carbohydrate molecule is important. It is also known that hexoses with equatorially placed hydroxyl groups have a strong binding affinity towards these receptors .
Although our PNP do not contain carbohydrate groups, especially those that contain –OH groups on the surface have some chemical resemblance with carbohydrates with -OH groups. The results of the inhibition studies match with this, since the maximum inhibition of intracellular uptake was observed for both hydroxyl- or acid-terminated PNP. This once again points towards a complex range of interactions between PNP and cell membranes and their receptors leading to particle internalization. Although we are the first to actually show that inhibition of mannose receptors inhibits cellular uptake of PNP, there have been a few reports already where these receptors have been targeted for facilitated drug delivery. Park et al.  used mannosylated polyethyleneimine coupled to silica NP to increase the transfection efficiency in macrophage cells by targeting the mannose receptors. Similar strategies have been employed by other groups to increase the delivery efficacy in biological systems [67–69]. Our results on mannose receptors are in line with data available in literature that point towards strong NP-receptor interactions. These results can be further developed for more sophisticated applications like drug delivery or food-based delivery of functional ingredients.