Summary of in-vitro cytotoxicity studies with different kinds of nanoparticles (NPs) and nanowires (NWs), NWs with aspect ratio <10 are often called nanorods (NR), SWCNT the abbreviation of single-walled carbon nanotube, MWCNT for multiwalled carbon nanotube and QD for quantum dots.
The specific use of engineered nanostructures in biomedical applications has become very attractive, due to their ability to interface and target specific cells and tissues to execute their functions. Additionally, there is continuous progress in research on new nanostructures with unique optical, magnetic, catalytic and electrochemical properties that can be exploited for therapeutic or diagnostic methods. On the other hand, as nanostructures become widely used in many different applications, the unspecific exposure of humans to them is also unavoidable. Therefore, studying and understanding the toxicity of such materials are of increasing importance. Previously published reviews regarding the toxicological effects of nanostructures focus mostly on the cytotoxicity of nanoparticles and their internalization, activated signaling pathways and cellular response. Here, the most recent studies on the in-vitro cytotoxicity of NPs, nanowires and nanorods for biomedical applications are reviewed and divided into two parts. The first part considers nonmagnetic metallic and magnetic nanostructures, while, the second part covers carbon structures and semiconductors. The factors influencing the toxicity of these nanostructures are elaborated to help elucidate the effects of these nanomaterials on cells, which is a prerequisite for their safe clinical use.
Nanostructured materials are defined as possessing one of their dimensions ranging from 1 to 100 nm, according to the American Society for Testing and Materials (ASTM) international standards definition . For nanoparticles (NPs), which can be of more or less spherical or cubical shape, two dimensions are required to be within this range. In contrast, the shape of nanorods (NRs) is in one dimension much larger than in the others. For a small aspect ratio (<10) both their length and diameter are in the nanoscale, whereas NRs with a large aspect ratio (>10) only have their diameter within this scale, and they are often called “nanowires” (NWs). Nanostructures within this specific size scale show unique size-dependent optical, magnetic, catalytic and electrochemical properties, among others, as well as high surface to volume ratios. Moreover, their shape, surface chemistry and chemical composition can be used to tailor-specific properties, making nanostructures highly versatile for different applications [2, 3].
The size scale of nanostructures is within the range of several biomolecules, such as proteins and antibodies, allowing specific interactions to occur between them. This, when coupled with the high surface to volume ratios and tunable sizes and properties, makes nanostructures prime candidates for biomedical applications such as imaging, drug delivery and therapy [4–6]. Examples of applications include the use of NPs as magnetic resonance imaging (MRI) contrast agents [7, 8], tissue engineering [9–11], as well as the recent focus on hyperthermia and cancer cell eradication with the use of NPs and NRs [12–17]. Such applications, if they are aimed for a clinical setting, ultimately require a direct NP/NR exposure in the form of ingestion or intravenous delivery into the body. Naturally, there is a rigorous testing required before any new drug formulation is approved for clinical use in order to ensure their safety and effectiveness. Currently, very few NPs-based drugs have been approved by the Food and Drug Administration and are commercially available. Examples include GastroMARK, used as an MRI contrast agent to enhance the delineation of the bowel, and ferumoxytol, an iron-replacement formulation approved for adults with chronic kidney disease with an iron deficiency .
Within this scope, biocompatibility and cytotoxicity data are of paramount importance to evaluate the potential of nanostructures for biomedical applications. Nanostructures are normally engineered to interface and target-specific cells or tissues to execute their functions, raising questions about their toxicological effects. For instance, there are several characteristics involved in the toxicity of fiber-like nanomaterials, such as shape, length, chemical composition, agglomeration and purity, making them suitable to fit the “fiber toxicological paradigm” according to the World Health Organization (WHO) criteria used to describe the toxicity of asbestos fibers . Further, nanostructures are usually tuned for biocompatibility on top of the desired biomedical function, with the most relevant aspects that influence their toxicity being the material , size and shape , surface charge  and surface functionalization . In vitro studies, while not able to give a complete insight into the biocompatibility of nanostructures, have a high importance, due to their easy implementation, and provide valuable cytotoxicology data regarding the safety of the use of nanostructures in biomedical applications. Previously published reviews regarding the biosafety of nanostructures include that of Lewinski et al.  and Zhao et al. . The former focuses mostly on the cytotoxicity of NPs of different materials, as well as carbon nanotubes (CNTs), whereas the latter is a more in-depth review of the internalization, activated signaling pathways and cellular response of different kinds of NPs.
Here, we review relevant studies assessing the
2. Carbon nanostructures
Carbon nanostructures include a broad diversity of carbon allotropes that differ from pristine diamond and graphite. Carbon has been used in many technological applications, exploiting its capability of forming networks composed exclusively of C-atoms with the same electronic configuration or hybridizing configurations sp3-, sp2- and sp-, expanding the possible allotropes that can be constructed . Since the synthesis of the first carbon nanostructures, such as fullerene C60 (0D)  and CNTs (1D)  (Figure 2), there has been a tremendous effort for understanding the properties of these nanomaterials and for exploring the broad range of applications in which they can be used. Carbon-based nanomaterials (CNMs) have created a great deal of interest in various applications such as optical imaging , drug and gene delivery , and nanotherapeutics [31, 32] due to their excellent mechanical, optical and electrical properties [33–35], as well as due to their ability to translocate through the cell membrane or be internalized via energy-dependent endocytic pathways . Similarly, CNMs possess an extraordinary ability to be loaded with drugs or different chemical agents that are either attached to the surface or, in the case of CNTs, they can be packed into the interior cores .
These widespread applications of CNMs are also accompanied by increasing concerns regarding their interactions with tissues, cells, and biomolecules as well as degradation pathways, and at a macroscale, the potential deleterious effects on human health and the environment.
2.1. Fullerene C60
The structure of fullerene C60, which has a van der Walls diameter of approximately 1 nm, is formed from 60 carbon atoms arranged in a spherical, cage-like structure consisting of 60 vertices, 12 pentagonal faces and 20 hexagonal faces . Fullerenes and their derivatives are probably the most extensively studied NPs with several properties and applications including MRI , drug delivery [40, 41], photodynamic therapy (PDT)  and photothermal therapy (PTT) .
Although fullerenes are generally hydrophobic molecules, many strategies have been developed for improving their solubilization in water that is, synthesized water-soluble derivatives of fullerenes by chemical modifications through the addition of functional groups such as hydroxyl-, carboxyl-, amino- and alkyl-groups and other side-chain/cyclic moieties to the C60 structure . The different methods employed to increase C60 water solubility profoundly influence the physiochemical properties and the toxicological effects of these compounds, raising uncertainties about the possible consequences on human health and potential medical uses . Nakagawa et al. studied the effects of the hydroxylated fullerenes (fullerenols) C60(OH)24 and C60(OH)12 0.125 mM in rat hepatocytes, observing a concentration and time-dependent cell death accompanied by mitochondrial dysfunction, with C60(OH)24 found to be more cytotoxic with almost 100% of cell death after 30 min. The authors concluded that the toxic effects of fullerenols may depend on the number of hydroxyl groups . C60(OH)24 at a concentration of 0.1 mM caused cell blebbing, loss of cellular ATP and lipid peroxidation in rat hepatocytes . Similarly, the cytotoxic effects of fullerene C60 and the derivatives C60(OH)2, C60(OH)6–12, C60(OH)12 and C60(OH)36, were evaluated in three different types of liver cells: dRLh-84, HepG2 and rat hepatocytes as shown in Figure 3 . C60(OH)6–12 and C60(OH)12 were found to induce cytotoxic effects after 3 days of exposure in dRLh-84 cells at a concentration of 10 μg/mL reducing the cell viability 30 and 40%, respectively, in the form of inhibition of mitochondrial activity. Similarly, to Nakawaga’s findings, these results indicate that the number of hydroxyl groups on C60(OH)
Other fullerene derivatives have also been tested for cytotoxic effects in human epithelial HEp-2 cells, such as C60-PVP, C60-NO2-proline and sodium salt of polycarboxylic C60 . However, the PVP and NO2-proline derivatives did not have an effect on cell viability, and the sodium salt of polycarboxylic derivative induced a drastic decrease in cell number of about 80% at a concentration around 0.1 mg/mL.
Further, the molecular mechanisms underlying the cytotoxic effects of two similar fullerene derivatives (C60-1,3-dipolar cycloaddition of azomethine ylides) on human MCF-7 cells were analyzed by RNA-seq-based gene expression . It was found that whereas one derivative had a negligible effect, the addition of an extra trifluoroacetate group induced a significant, time-dependent alteration of gene expression, mainly in biological processes involving protein synthesis, cell cycle progression and cell adhesion, with the authors suggesting an inhibition effect of the mTOR pathway.
In a recent study performed by Canape et al., C60 fullerenes were covalently functionalized with PEG of various sizes, Full–PEG2000, Full–PEG5000 and Full–PEG10000, and viability was studied on a variety of cell lines 24 h after exposure, evaluating mitochondrial activity, cell membrane integrity and hemolysis . However, all the tested compounds were found to reduce, to some extent, the cellular metabolic activity, only two affected the cell membrane integrity, and none induced hemolysis. It was concluded that fullerenes C60 functionalized with higher molecular weight PEGs possess a higher biocompatibility and that side toxicity can be alleviated using proper surface coating. Together, all these findings support that the surface functionalization of fullerenes plays an important role with regard to their interaction with biological systems.
The interaction of CNMs with lipid membranes is of great interest because biological activity requires crossing or breaking lipid membranes. In a study concerning the interaction of fullerenes with the lipid bilayer and the possibility of fullerene crossing it, it was observed that hydrophobic molecules of C60 were localized within the inner part of the membrane, whereas hydrophilic C60(OH)n fullerenols molecules were adsorbed on the heads of membrane phospholipids , where they can interact with membrane proteins, such as ATPases and influence their activity [50, 51]. Similarly, Raoof et al. showed that the internalization of a water-soluble derivatized C60 malonodiserinolamide takes place through multiple energy-dependent pathways, and they escape endocytotic vesicles to eventually localize and accumulate in the nucleus through the nuclear pore complex .
2.2. Single-walled carbon nanotubes
CNTs are classified in single-walled carbon nanotubes (SWCNTs) and multiwalled carbon nanotubes (MWCNTs). The first ones are formed from a single layer of graphene (0.4–10 nm in diameter), whereas the second ones consist of multiple concentric cylinders of graphene with increasing diameters (10–100 nm) . The length of CNTs can range from nanometers to centimeters , and they possess unique physical and chemical properties such as a lightweight, high tensile strength, high electrical and thermal conductivities, unique optical properties and extreme chemical stability, as well as high surface-to-volume ratios with reactive surface chemistries. Such properties have made CNTs an interesting material for biomedical applications, where they have been used as drug, protein and nucleic acid delivery tools [54–56], cancer cell destruction [57, 64, 91], diagnostics  and as noninvasive and highly sensitive imaging aids [31, 58]. Naturally, biosafety concerns of CNTs are rapidly emerging with numerous reports indicating their potential hazards to the public health.
The graphene sheets can be wrapped in a variety of ways that are denoted by a pair of indices (n, m), which define both the diameter and the chirality of SWCNTs, which can be either metallic (M) or semiconducting (S). As synthesized, SWNTs have a wide range of diameters and chiral angles, which leads to a polydisperse sample of discrete properties [59, 60]. SWCNTs possess small diameters and the large aspect ratios that render them ideal one-dimensional quantum wires that elicit different biological behavior compared to spherical NPs, when introduced in biological systems . The cytotoxicity of pristine SWCNTs and SWCNTs functionalized with PEG has been evaluated with neuronal PC12 cells at the biochemical, cellular and gene expression levels by Zhang et al. . Cytotoxicity increased with the concentration, whereby SWCNT-PEGs exhibited less cytotoxic potency than bare SWCNTs at the highest concentration tested by reducing the cell viability in approximately 70 and 50%, respectively (Figure 4). Morphological changes appeared in PC12 cells treated with both SWCNTs and SWCNTs-PEG as shown in Figure 5. Cells exposed to SWCNTs showed an elongated shape, which was related to higher toxic effects induced by the untreated CNTs. ROS were generated as a function of both concentration and surface coating after exposure, whereas gene expression analysis showed that the genes involved in oxidoreductases and antioxidant activity, nucleic acid or lipid metabolism and mitochondria dysfunction were highly altered. Interestingly, alteration of the genes was also surface coating-dependent. The authors concluded that surface functionalization of SWCNTs decreases the ROS-mediated toxicological response
Avti et al. showed that SWCNTs synthesized using Gd3+ NPs as catalysts induced no structural damage to NIH/3T3 fibroblasts or decreased their viability at concentrations between 1 and 10 μg/mL . In contrast, highly pure SWCNTs triggered similar amounts of pulmonary fibrosis-related compounds interleukin 1β (IL-1β) and transforming growth factor (TGF-β1) in THP-1 and BEAS-2B pulmonary cells without affecting cell viability . Similarly, Di Giorgio et al. studied the cyto- and genotoxic effects, as well as the inflammatory response and ROS production, of SWCNTs on the mouse macrophage cell line RAW 264.7 . There, the authors reported that SWCNTs induced ROS release, cell ultrastructural damage, necrosis and chromosomal aberrations, but did not cause an inflammatory response.
2.3. Multiwalled carbon nanotubes
MWCNTs are defined as a nested coaxial array of SWCNTs, each nanotube being formed by a graphene sheet rolled into a cylinder of nanometer size diameter .
It has been postulated that MWCNTs can provide an innovative and promising alternative to conventional drug formulations for cancer therapy, as they can be conjugated with various bioactive molecules such as drugs, surfactants, diagnostic agents and antibodies in order to target receptors that are overexpressed in cancer cells [66–68].
The generation of carboxyl groups by oxidation on the surface of CNTs is one of the most used strategies for introducing hydrophilic moieties onto the CNT hydrophobic surface and in order to conquer a lack of solubility and to improve their biocompatibility [69–71]. Thus, Liu et al. have studied the effects of carboxylated c-MWCNTs on the human normal liver cell line L02 and found a reduction in the toxicity, when compared to pristine MWCNTs with a reduction of around 60% of cell viability at the highest concentration tested after 72 h and concluded that this effect is probably due to a reduced activation of the mitochondria mediated apoptotic pathway . Moreover, as charged entities, c-MWCNTs bind to proteins in the bloodstream through noncovalent interactions to form a protein corona. De Paoli et al. have characterized the interactions of c-MWCNTs with common human proteins such as albumin, fibrinogen, g-immunoglobulins and histone H1 and found that the association of proteins to c-MWCNTs depends on the protein’s charge, size and structural flexibility and that it affects the agglomeration state and charge of the CNTs .
As with SWCNTs, molecules can be covalently and noncovalently attached to the surface of MWCNTs . The main disadvantage of noncovalent attachment is the lack of biomolecule specificity upon adsorption, which affects the CNTs dispersion stability by replacing the functional surface coating with proteins and molecules contained in all physiological fluids (cell culture media or blood) . Heister et al. have compared five types of CNTs, varying in their dimensions and surface properties, for a multidimensional analysis of dispersion stability and their toxicity toward cancer cells (Figure 6), from which it was emphasized that the covalent link between PEG and oxidized MWCNTs leads to stable dispersion and biocompatibility in various biological environments .
It has been proposed that the metal impurities trapped inside the MWCNTs may be responsible for their toxicity that partially occurs through the generation of ROS . Fe impurities trapped inside the MWCNTs may be partially responsible for neurotoxicity, as postulated by Meng et al., who investigated and compared the effects of two kinds of MWCNTs with different concentrations of Fe impurities in rat pheochromocytoma PC-12 cells . They found that the exposure to Fe MWCNTs can reduce cell viability up to 80% after 72 h exposure and increase cytoskeletal disruption of undifferentiated PC-12 cells, diminish the ability to form mature neurites and then adversely influence the neuronal dopaminergic phenotype in NGF-treated cells.
Additionally, MWCNTs have been shown to affect the immune system. Pescatori et al. used a whole-genome expression approach to assess whether functionalized MWCNTs could stimulate distinct molecular changes in immune cells, with transcriptomic changes analyzed in human immune cells THP1, a monocytic cell line, Jurkat cells and a T lymphocyte cell line . They found a cell-specific action on monocytes for three types of MWCNTs, which specifically enhanced innate immunity activation mechanisms. The pathways activated are functionally relevant and critical for the development of an effective inflammatory response.
3.1. Titanium dioxide nanoparticles and nanowires
A comprehensive review of the numerous biomedical applications of titanium dioxide (TiO2) throughout the years was published by Yin et al. . In summary, mostly due to their low cost, strong optical absorption and high chemical stability, TiO2 NPs have shown great potential in applications such as photodynamic cancer therapy, drug delivery, cell imaging and biosensors, among others.
One of the initial cytotoxicity studies with TiO2 NPs was performed on human dermal microvascular endothelial cells . There, it was shown that NPs with an average diameter of 70 nm at a dose of 50 μg/mL caused a minor pro-inflammatory response in the form of an increase in the levels of IL-8. Later, a study with mouse fibroblast L929 cells exposed to TiO2 NPs was conducted by Jin et al. . For 3–600 μg/mL doses, cells appeared to shrink and became round in culture, with a dose-dependent reduction of cell metabolic activity, LDH release and ROS generation. Chromatin fragmentation was also reported, indicating possible DNA damage (Figure 7). It was also found that both human neural astrocyte-like U87 cells and human fibroblast HFF-1 cells exposed to 25 nm TiO2 NPs for 48 h had a decrease in cell survival for doses up to 100 μg/mL, with cell death reported as a combination of apoptosis and necrosis . On the other hand, BEAS-2B cells underwent cell death through apoptosis, triggered by the activation of caspase-3 and chromatin condensation through ROS .
Although the size of single TiO2 NPs reported by Jin et al. was of 5 nm , they were clustered in 20–30 nm aggregates, an effect that could enhance cytotoxicity. It was later shown that there is a correlation between the cytotoxicity of TiO2 NPs and their aggregate size, as larger aggregates (600 vs. 166 nm) elicited a stronger decrease in cell viability, as well as the expression of genes related to stress and inflammation . In contrast, TiO2 NPs of 12 nm in diameter aggregated in 450 nm clusters and only at higher doses slightly decreased the viability of glomerular mesangial IP5 and epithelial proximal HK-2 cells, suggesting specific cell responses . Additionally, although ROS was generated in the presence of TiO2 NPs, the cells were able to maintain their antioxidant potential, thereby showing no oxidative stress.
Cellular uptake studies with 30 nm TiO2 NPs have been carried out in human amnion epithelial WISH cells using transmission electron microscopy (TEM), with images showing most of the particles localized either inside vesicles or freely in the cytoplasm . In addition to the already mentioned cytotoxic response, WISH cells experience an oxidative response due to ROS accumulation, as well as DNA double strand breaks and cell cycle arrest.
Cytotoxicity data of TiO2 NWs are scarce, with only a handful of studies published. Magrez et al. observed that TiO2-based NWs of 5 μm in length and 75 nm in length had a negative impact on the cell proliferation and cell viability of H596 human lung tumor cells in a dose-dependent manner and for concentrations up to 2 μg/mL . NWs were observed to reside in the periphery of the nuclei, which were often enlarged and lobulated or fragmented. In another study, H2Ti3O7 NWs at a dose of 10 μg/mL induced the generation of cell debris in eight different cell lines, which the authors associated with an increase in autophagosome-like vacuoles in the cytosol .
3.2. Zinc oxide nanoparticles and nanowires
Zinc is a biologically active element that plays a role in different processes, such as the immune system, cell metabolism, cell proliferation, enzymatic function and gene expression, among others [88, 89]. Due to these biological functions of zinc, coupled with initial biocompatibility studies , ease of fabrication and relevant properties , zinc oxide (ZnO) nanostructures have been proposed as suitors for several biomedical applications, including cancer cell therapy [92, 93], drug delivery [94, 95] and imaging . However, possible undesirable effects of the interactions between ZnO nanostructures and biological systems could arise and cause a toxicological response. Additionally, ZnO nanostructures are known to dissolve under acidic conditions. The phase-solubility diagram of ZnO  indicates that ZnO NPs will dissolve at a pH value below 6.7 at physiological temperature, and they will rapidly dissolve in the acidic pH of the lysosomes (pH 5.7) after their uptake . Zinc oxide NPs can dissolve in an aqueous media to form hydrated Zn2+, which is enhanced in acidic pH as well as in the presence of biological components, such as amino acids and peptides . A review on studies published between the years 2009 and 2011 on the toxicity of ZnO NPs to mammalian cells was reported by Vandebriel et al. . They concluded that the induction of oxidative stress is the most important and the most likely mechanism underlying ZnO NP toxicity.
Initial cytotoxicity studies in human T lymphocytes showed a significant decrease in cell viability only for concentrations higher than 5 mM using ZnO NPs of 13 nm in diameter . It was then found that ZnO NPs preferentially kill cancerous human T lymphocytes compared to normal ones via ROS and apoptosis . Similarly, NPs of the same size were tested against human lung BEAS-2B cells and RAW 264.7 macrophages, and a dose and time-dependent cytotoxicity was found in both cases for doses up to 50 μg/mL and incubation times of 16 h . Moreover, the ZnO NPs were reported to induce the generation of ROS, as well as the activation of the pro-inflammatory marker TNF-α and the pro-inflammatory pathway Jun kinase, as well as intracellular calcium release, a major oxidative stress response. Finally, ZnO NPs were found to reside in caveolae in the case of BEAS-2B cells, whereas in the RAW 264.7 cells, they resided inside lysosomes, with intracellular dissolution and release of Zn2+ shown in both cases. In a different study, ZnO NPs also impaired the survival of human neural astrocyte-like U87 cells in a dose-dependent manner .
The degree of cytotoxicity of ZnO NPs also depends on their size, as shown by Hanley et al. . Using 4, 13 and 20 nm NPs, they determined an inverse relationship between nanoparticle size and cytotoxicity in terms of cell viability and ROS generation in immune cells. Among these, monocytes were the most sensitive to the ZnO NPs, whereas lymphocytes were the most resistant, as reported previously . In contrast, glomerular mesangial IP5 cells showed a similar dose-dependent decrease in cell viability for ZnO NPs of both <100 nm and >1 μm, along with the generation of ROS . The cell viability of neural stem cells appeared to be indifferent of NP size . In a different work, a differential cytotoxic response was reported, when comparing the effects of 20–30 nm ZnO NPs in human myeloblastic leukemic HL60 cells and normal peripheral blood mononuclear cells (PBMCs) . For concentrations up to 1000 μg/mL, PBMCs maintained a steadily high viable cell population, whereas a dose of 50 μg/mL was enough to bring the cell viability of HL60 cells down to 50%. DNA fragmentation analysis and annexin V staining confirmed that cell death was through the apoptosis pathway. The potent tumor suppressor that regulates the cell cycle and prevents DNA damage, p53 , is believed to be a molecular master switch toward apoptosis, and reports show that the p53 pathway was activated in BJ cells (skin fibroblasts) upon ZnO NPs treatment with a concomitant decrease in cell proliferation .
The liver, playing a major role in human metabolism, may be a target organ for NPs after they enter into the body. As such, it is an important toxicity evaluation method. Similarly, to previous results, Sharma et al. found that human liver HepG2 cells had a dose-dependent response to ZnO NPs at doses up to 20 μg/mL and for exposure times from 12 to 24 h . Cell death was also shown as being through the apoptotic pathway, due to ROS generation, oxidative stress and mitochondrial and DNA damage. Taken together, all these results suggest that the cytotoxic response to ZnO NPs is dependent on the target cell tissue as well as on changes in NP dimensions.
Kao et al. observed the effects of ZnO NPs of <50 nm in the homeostasis of intracellular Zn2+ in human leukemia Jurkat cells and human lung carcinoma H1355 cells and found an increase in the concentration of cytosolic and mitochondrial Zn2+, most probably due to NP dissolution , as shown previously . Caspase-3 activation, mitochondrial membrane depolarization and LDH release were also reported, which suggests an apoptotic death pathway due to mitochondrial dysfunction. In a more recent study, the intracellular concentration of Zn2+ of breast cancer MDA-MB-231 cells was also increased after treatment with ZnO NPs, leading to the generation of ROS, damage to the cell membrane and mitochondria and culminating in apoptosis .
Limited literature exists regarding the biocompatibility of ZnO NWs, but similar cytotoxicity effects as those of NPs were shown by Li et al. using HeLa cells and connective tissue L-929 cells . Although the NWs used were rather large (200 μm in length and 1 μm in diameter), both cell lines seemed to maintain their viability for concentrations up to 10 μg/mL and exposure times of 24 h. Similar viability data were then reported for NWs of 10 μm in length and 327 nm in diameter at the same dose in human macrophages (Figure 8) . There, too, it was found that an intracellular increase of Zn2+ precedes cell death, indicating the intracellular dissolution of the ZnO NWs after uptake [112–120].
3.3. Silicon nanoparticles
Coated silicon nanoparticles (Si NPs) have attracted both a great deal of concern and attention, especially in biomedical applications such as disease diagnosis, tumor cell tracking, imaging, drug delivery and gene therapy [121, 122]. They have been widely studied for such applications because of their active surface state and high suspension ability [121, 123, 124]. However, there are some recent reports that limit the use of Si NPs because of potential side effects on the cells, when using them in such a scale and in high concentrations .
One of the first studies on Si NPs focused on their cellular uptake . Si NPs were coated with a fluorone dyes called Rhodamine 6G isothiocyanate (RITC), the fluorescence signal of which indicated uptake. NPs of 50 nm diameter at a dose of 80 μg/mL were accumulated in the cytoplasm of HeLa cells after 4 h of incubation at 37°C, while the uptake was reduced by 80% at 4°C. A year later, Lin et al. focused on the toxicity effect of the size, concentration and exposure time of Si NPs on human lung cancer cells (calveolar carcinoma-derived cells) . The cell viability decreased significantly as a function of both nanoparticle dosage (10–100 μg/mL) and exposure time (24, 48 and 72 h). However, the cytotoxicity of two different sizes of Si NPs (15 and 46 nm) did not show a significant difference.
A different study compared the cytotoxicity of a variety of sizes of Si NPs (19, 43, 68 and 498 nm) at 100 μg/mL . After 4 h of incubation with human liver HepG2 cells, it was noticed that the cytotoxicity of Si NPs is size-dependent (i.e., the smaller size the higher cytotoxicity. The live cells were counted by a cell-counting kit (CCK-8). Further, Sahu et al. proved that Si NPs (10–20 nm) are much more toxic than micro-sized ones (0.5–10 μm) for a concentration range of 5–500 μg/mL, after exposing them to human lung epithelial (L-132) and human monocytes (THP-1) for 24 h . The cellular uptake efficiency and pathway of different sized NPs has also been confirmed to be size-dependent, with smaller particles (55 nm) being internalized faster than larger ones (307 nm) . The largest NPs (307 nm) internalized through clathrin-coated pits, whereas medium ones (167 nm) internalized through clathrin-coated vesicles and the smallest (55 nm) were internalized through an energy independent pathway. Despite differences in their internalization pathway, all three sizes showed a high-level of biocompatibility.
In a similar approach as the one of Lin et al., the cytotoxic effects of increasing concentrations of Si NPs (0, 25, 50, 100 and 200 μg/mL) on HepG2 cells were analyzed in terms of ROS level, mitochondrial membrane potential and apoptotic rate. All three tests showed that the level of toxicity of the NPs increases while increasing the concentration from 25 to 100 μg/ml. Additionally, it was shown that the expressions of the apoptotic genes cytC and Caspase-3 were up-regulated with increasing NP concentrations. Additionally, the downregulation of the antiapoptotic Bcl-2 gene and upregulation of the genes p53 and BAX have also been reported [129, 130].
Other approaches have focused on cell-dependent cytotoxicity and surface charge . Kim et al. found that NIH/3T3 fibroblasts appear to be more susceptible to Si NPs in terms of cell viability, when compared to A549 and HepG2 cells . On the other hand, positively charged (NH2-coated Si NPs) displayed higher cytotoxicity than negatively charged ones (COOH-coated NPs) in human adenocarcinoma Caco-2 and rat alveolar macrophage NR8383 cells . However, the opposite has also been suggested for HaCaT keratinocyte cells . A summary of the viability dependences on the Si NPs’ size, concentration and cell type is shown in Figure 9 .
3.4. Silicon NWs
Si NWs show several advantages over Si NPs. For instance, they tend to not agglomerate in solution compared to Si NPs  and they enhance the drug-loading capacity due to their high surface area [136, 137]. However, it has been indicated that Si NWs have more toxic effects to macrophages cells at lower concentrations compared to Si NPs due to the large surface area, which increase the interaction and induce the cell death . Naturally, the concentration of Si NWs plays a role on cell viability. Si NWs of 2 μm long, 55 nm diameter were co-cultured with HeLa and Hep-2 cells at different concentrations . While no toxicity was found on either cell line for concentrations below 190 μg/ml, the cells died and released 75% of their contents into the supernatant at high concentrations (1900 μg/ml) after 72 h of incubation. Zhang et al. used amino-modified (APTES), folate-functionalized Si NWs to study cell interactions . The NWs lengths were between 2.5 and 8.0 μm, with a concentration of 100 μg/mL. It was found that the length of NWs affected the internalization, with NWs longer than 5 μm being more difficult to be internalized, due to geometrical restrictions.
3.5. Quantum dots
Semiconductor quantum dots (QDs) are light-emitting particles that have broad excitation spectra, long fluorescence lifetimes compared to traditional fluorescent probes and are more resistant to photobleaching [140, 141]. Also, they can easily be conjugated to proteins , which makes them excellent choices for bioimaging [142–146] and other biomedical applications [141, 147, 148]. Tsoi et al. summarized the toxicity of QDs by two mechanisms: degradation with the release of free cadmium (Cd) and generation of ROS . Each design of QD is a unique combination and has its own physicochemical properties that may influence its biological activity and toxicity. As a result, tremendous research efforts have been devoted to produce high quality QDs by optimizing synthetic procedures, as well as functionalizing their surface in order to enhance biocompatibility [142, 150].
An early study by Derfus et al. demonstrated that CdSe-core QDs oxidized and degraded, releasing Cd ions which induced cell death . When CdSe QDs were exposed to a UV-light for 1, 2, 4 and 8 h and then incubated with hepatocytes, it showed a 6, 42, 83 and 97% decrease in the cells’ viability, respectively . Cd is a known carcinogen with potential damage to the renal, skeletal, pulmonary and reproductive systems . Interestingly, Chen et al. showed that the cell viability of HEK293 cells treated with 37.5 nM of 5 nm CdTe QDs was not significantly altered, compared to the control (i.e., untreated cells) after 3 days of incubation . However, high concentrations (300–600 nM) of QDs completely inhibited cell growth from the very beginning. The cytotoxicity of QDs has also been linked to the generation of ROS, which in turn damages cellular proteins, lipids and DNA . The p53 gene was also shown to be inhibited by CdTe QDs, leading to apoptosis and cell death .
Tracking the QDs internalization pathways could help explain their toxicological properties. To this end, microscopy studies showed that QDs localize within cellular endosomes and lysosomes, exposing them to an acidic or oxidative microenvironment . It was determined that the hypochlorous acid present in phagocytic cells oxidized polymer-encapsulated CdS and ZnS-capped CdSe QDs, releasing cadmium, zinc, sulfur and selenium into the cytoplasm. Some studies have suggested that the QDs toxicity might derive from multiple factors including the environment and the QDs physicochemical characteristics (such as size, shape and surface chemistry). A surface coating with a ZnS shell  or BSA corona  reduced the QDs toxicity. In addition, polymeric coatings (i.e., phospholipid-PEG) and inorganic coatings (e.g., Si) can prevent the release of Cd into the biological media . In a different approach, Soenen et al. studied cell viability using Cd-free QDs (ZnSe/ZnS and InP/ZnS QDs) at concentrations ranging from 0 to 100 nM . Cytotoxic effects were observed starting from 60 nM for ZnSe to 80 nM for InP QDs. Further, no increase in cytotoxicity was reported up to 7 days after the initial cell labeling compared to normal QDs due to the absence of Cd.
Recent studies on the in-vitro cytotoxicity of carbon structure and semiconductors in biomedical applications were reviewed, taking into account nanoparticles and nanowires/nanorods. A summary of the results of representative studies is provided in Table 1.
|Nanostructure type||Surface coating||Nanostructure concentration||Average size||Cell line||Cell viability||Viability test||Reference|
|C60||Pristine, C60(OH)12, C60(OH)24,||0.125 mM/1 h||N/A||rat hepatocytes||80% for pristine, C60(OH)12 and 60% for C60(OH)24.||Tryptan blue/microscopy, MPP, GSH|||
|C60||C60-alanine, -NO2, -PVP, -NO2-proline, sodium salt of a polycarboxylic derivative||0.001–0.2 mg/mL for C60-|
NO2-proline and 0.016–0.2 mg/mL for all others/48 h
|N/A||HEp-2 cells||No cytotoxicity except for the sodium salt of a polycarboxylic derivative with 20% viability at 0.01 mg/mL||Crystal violet/optical density|||
|C60||PEG of various sizes||0.03–1 mg/mL/24 h||N/A||HepG2, NHDF, Caco2, HUVEC, U931, J774 A1||Maximum inhibition at 1 mg/ ml of|
Full–PEG2000 for J774 (41%) and U937 (62%).
|MTT, LDH assays|||
|SWCNTs||Pristine and PEG||0.1–100 μg/mL/24 h||0.7–1.6 nm diameter, 0.2–3 μm length||PC12 cells||30 and 50% viability in MTT at highest concentration, respectively. Higher values for XTT and 10–20% LDH leakage||MTT, XTT, LDH, DCF, GSH assays|||
|SWCNTs||Collagen||15 μg/mL/4 h to 15 days.||0.7–1.6 nm diameter, N/A||BACs||No cytotoxicity||WST-1 assay, Live/dead|||
|SWCNTs||Gd-NPs as catalysts and PEG||50–100 μg/mL/12–48 h||N/A||NIH/3 T3 fibroblasts||70% viability at highest concentration and time exposure||Tryptan blue/microscopy, Live/dead|||
|MWCNTs||Pristine, COOH||12.5–200 μg/mL/24, 48 and 72 h||10–20 nm diameter, 10–30 μm length||human normal liver cell line L02||60 and 80% viability at highest concentration and time exposure, respectively||CellTiter-GloV® assay|||
|MWCNTs||3 and 23% of Fe impurities||5–60 μg/mL/24, 48 and 72 h||2–50 nm diameter, 50 μm length||PC12 cells||70 and 20% viability at highest concentration and time exposure, respectively||CCK-8|||
|TiO2 NPs||—||600 μg/mL||5 nm diameter||L929 mouse fibroblast cells||<70%||MTT assay|||
|ZnO NPs||—||10 mM||13 nm diameter||Human T lymphocytes||40%||Propidium iodide staining|||
|SiO2 NPs||—||50 μg/mL||10–20 nm||monocytes (THP-1) cells||71%||MTT assay|||
|Si NPs||Coated with negatively charged (COOH)||3 mg/ml||1.6 nm||Rat alveolar macrophage NR8383 cells||No cytotoxicity||MTT assay|||
|Si NPs||Coated with positively charged (NH2)||0–100 mg/ml||3.9 nm||Rat alveolar macrophage NR8383 cells||The EC50 values = 0.38 μg/ml||MTT assay|||
|Si NPs||—||160 μg/ml||7 nm||HepG2 cells||∼98%||MTT assay|||
|Si NPs||—||160 μg/ml||20 nm||HepG2 cells||∼72%||MTT assay|||
|Si NPs||—||160 μg/ml||50 nm||HepG2 cells||∼49%||MTT assay|||
|Si NWs||—||<190 μg/ml||2 μm long, 55 nm diameter||HeLa and Hep-2 cells||75%||MTT assay|||
|Si NWs||—||1 μg/ml||500 nm long, 100 nm diameter||breast cancer cells line (MCF-7/ADR)||90%||MTT assay|||
|Si NW arrays||—||—||5 μm long, 20–100 nm diameter||HeLa cells||98%||MTT assay|||
|Si NW arrays||Coated with AgNPs||—||5 μm long, 20–100 nm diameter||HeLa cells||80%||MTT assay|||
|Si NW arrays||Coated with Cu NPs||—||5 μm long, 20–100 nm diameter||HeLa cells||∼50%||MTT assay|||
|CdSe QDs||Oxidation for 0 h, 1 h, 2 h and 4 h||62.5 μg/mL||7 nm||Hepatocyte cells||100, 98.55 and 21%, respectively||MTT assay|||
|CdTe QDs||—||37.5 and 75 nM||5 nm||HEK293 cells||∼87 and ∼67%||MTT assay|||
|ZnSe QDs||Cd-free||60 nM||1–10 nm||HUVEC cells||∼77%||Alamar Blue assay|||
|InP QDs||Cd-free||80 nM||1–10 nm||HUVEC cells||∼78%||Alamar Blue assay|||
Comparisons between the cytotoxicities of those different nanomaterials are generally difficult to make due to the vast range of methods, concentrations, dimensions, cell lines etc. For instance, the concentrations reported in the different studies were typically evaluated using either ICP or Cryogenic TEM. However, the concentration or dose of the nanomaterial plays a significant role in the cytotoxic response as well as the biomedical applications. Similarly, the reported toxicology of the nanomaterials depends on their interaction with the assay. For example, carbon nanostructures interact with the MTT-formazan crystals but not with XTT or INT reagents.
While the concentrations and exposure times are critical factors, the toxicity of these nanostructures is also material-dependent. These relations can be seen in Figure 10, which presents the average values reported for the cell viabilities (ignoring differences in concentrations, incubation times etc.), when exposed to the nanomaterials in the studies covered in Table 1. ZnO NPs showed the highest toxicity, while the lowest has been reported for silicon.
In addition, the particle size plays a major role in the cytotoxic properties of the nanostructure, whereby both the cellular uptake efficiency and pathway are affected, with smaller particles being internalized faster than the larger ones.
The induction of ROS after dissolving the nanostructures in the lysosomes was shown to be the primary underlying cause of the toxicity in several cases, leading to cell death through the apoptotic pathway, due to ROS generation and mitochondrial damage. The acidic condition inside the lysosome increases the digestion of the particles, enhancing the release of ions that affect the viability of the cells. This is a particularly relevant issue in case of CdSe-core QDs, which release Cd ions upon oxidation, leading to fast cell death.
Adding a coating to the nanostructure typically affected both the toxicity and the surface charge of the nanostructure, where cationic surfaces are more toxic than anionic. For instance, the toxicity of QDs was reduced by adding a BSA corona, and the release of Cd was prevented by the addition of polymeric and inorganic coatings. The type of the coatings was shown to affect the cell viability differently.
The cytotoxicity of the nanomaterial depends also on the nanostructure’s shape. In this regard, several advantages have been reported for NWs over NPs. For instance, they enhance the drug-loading capacity due to their large surface area. An interesting observation from Figure 10 is that NWs/NRs are, on average, less cytotoxic than NPs, with titanium dioxide being the only exception. However, one study has shown that the large surface area of Si NWs has a more toxic effect at lower concentrations compared to NPs. This was attributed to the increased interaction of the nanomaterial with the cells due to the large surface area.
While all these studies contributed to obtain a better picture of the cytotoxicity of nanomaterials and the underlying mechanisms, it is a persisting issue that a consistent measurement and reporting system will be needed for future studies. This will not only enable performing more accurate comparisons of the toxicological characteristics of nanostructures, but also to better evaluate the potential of using them for biomedical applications.
Research reported in this publication was supported by the King Abdullah University of Science and Technology (KAUST).