The nanosilica hazard: another variable entity

Silica nanoparticles (SNPs) are produced on an industrial scale and are an addition to a growing number of commercial products. SNPs also have great potential for a variety of diagnostic and therapeutic applications in medicine. Contrary to the well-studied crystalline micron-sized silica, relatively little information exists on the toxicity of its amorphous and nano-size forms. Because nanoparticles possess novel properties, kinetics and unusual bioactivity, their potential biological effects may differ greatly from those of micron-size bulk materials. In this review, we summarize the physico-chemical properties of the different nano-sized silica materials that can affect their interaction with biological systems, with a specific emphasis on inhalation exposure. We discuss recent in vitro and in vivo investigations into the toxicity of nanosilica, both crystalline and amorphous. Most of the in vitro studies of SNPs report results of cellular uptake, size- and dose-dependent cytotoxicity, increased reactive oxygen species levels and pro-inflammatory stimulation. Evidence from a limited number of in vivo studies demonstrates largely reversible lung inflammation, granuloma formation and focal emphysema, with no progressive lung fibrosis. Clearly, more research with standardized materials is needed to enable comparison of experimental data for the different forms of nanosilicas and to establish which physico-chemical properties are responsible for the observed toxicity of SNPs.


Introduction
Over the past decade, the definition of nanoparticles has been controversial. Nanoparticles are commonly defined as objects with a diameter less than 100 nm, but no clear size cut-off exists, and this usual boundary does not appear to have a solid scientific basis. Other definitions of nanoparticles have been proposed, and the most recent proposal [1] is based on surface area rather than size (a nanoparticle should have specific surface area > 60 m 2 /cm 3 ), thus reflecting the critical importance of this parameter in governing the reactivity and toxicity of nanomaterials. Physico-chemical properties that may be important in understanding the toxic effects of nanomaterials include primary particle size, agglomeration/ aggregation state, size distribution, shape, crystal structure, chemical composition, surface chemistry, surface charge, and porosity. Aspects of these properties have been discussed in several reviews of nanotoxicology [2][3][4].
Silica is the common name for materials composed of silicon dioxide (SiO 2 ) and occurs in crystalline and amorphous forms. Crystalline silica exists in multiple forms. Quartz, and more specifically α-quartz is a widespread and well-known material. Upon heating, α-quartz is transformed into β-quartz, trydimite and cristobalite. Porosil is the family name for porous crystalline silica. Quartz exists in natural and synthetic forms, whereas all porosils are synthetic. Amorphous silica can be divided into natural specimens (e.g., diatomaceous earth, opal and silica glass) and human-made products.
The application of synthetic amorphous silica, especially silica nanoparticles (SNPs), has received wide attention in a variety of industries. SNPs are produced on an industrial scale as additives to cosmetics, drugs, printer toners, varnishes, and food. In addition, nanosilica is being developed for a host of biomedical and biotechnological applications such as cancer therapy, DNA transfection, drug delivery, and enzyme immobilization [5][6][7][8][9]. Barik et al. [10] recently reviewed the impact of nanosilica on basic biology, medicine, and agro-nanoproducts. With the growing commercialization of nanotechnology products, human exposure to SNPs is increasing, and many aspects related to the size of these nanomaterials have raised concerns about safety [11]. Until recently, most research has focused on silica particles 0.5 to 10 μm, mainly in crystalline forms, but nanosilica may have different toxicological properties as compared with larger particles. The unique physicochemical properties of nano-sized silica that make them attractive for industry may present potential hazards to human health, including an enhanced ability to penetrate intracellular targets in the lung and systemic circulation.
Biocompatibility is a critical issue for the industrial development of nanoparticles [12,13]. Even though no acute cytotoxicity has been observed or reported, the uptake of the nanoparticles by cells may eventually lead to perturbation of intracellular mechanisms. The ability of silica-coated nanomaterials to penetrate the bloodbrain barrier also strongly suggests that extensive studies are required to clarify the potential chronic toxicity of these materials [14].
A number of SNPs have recently been shown to cause adverse health effects in vitro and in vivo (discussed later in this review). However, most of the studies have used poorly characterized particles in terms of their composition and physico-chemical properties. The distinct physico-chemical properties of nanoparticles indeed determine their interaction with the cell/within the cell, and even subtle differences in such properties can modulate the toxicity and modes of action. The results of toxicity studies then become difficult to interpret and compare, and, as a result, drawing appropriate conclusions is nearly impossible. Although SNPs could certainly provide benefits to society, their interaction with biological systems and potential toxic effects must be carefully addressed.
In this review, we discuss silica materials with a special attention to the physico-chemical properties that can affect their potential interaction with biological systems. We aim to provide an overview of the recent in vitro and in vivo investigations of the toxicity of nanosilica, both in crystalline and amorphous forms, rather than review the toxicity of micron-sized silica and quartz. A summary of the present knowledge on the potential toxic effects of nano-sized silica particles is needed, because their toxicological pattern appears distinct from that of micron-sized silica particles.

Synthesis & Characterization of Silica Materials
Classification of natural and synthetic silica materials "Silica" is the name given to materials with the chemical formula of silicon dioxide, SiO 2 . Silicas can be amorphous or crystalline, porous or non-porous (dense), anhydrous or hydroxylated [15], regardless of their natural or synthetic nature. In a silica material, the silicon atom is in tetrahedral coordination with 4 oxygen atoms. Theoretically, an infinite variety of 3-D-ordered structures can be built from oxygen-sharing silicate tetrahedra. The number of known crystalline silica materials is limited, which leaves much room for research and development. In amorphous silica, the tetrahedra are randomly connected.
In nature, amorphous silica can have different origins. Silica can be condensed from vapors emitted in volcanic eruptions. Natural silica can also be deposited from supersaturated natural water or polymerized in living organisms (biogenic silica). These amorphous biogenic silicas can be found as isolated particles, skeletal structures or surface elements in different living organisms. Many microcrystalline silica minerals such as flint, chert and chalcedony are derived from biogenic silica after crystallization by compaction. Kieselguhr (diatomaceous earth) occurs at various stages of transformation [15] and therefore often exhibits both crystalline and amorphous silica constituents.

Physico-chemical characteristics of synthetic silica materials related to toxicity
The silica materials presenting a toxicological hazard to human health are mainly synthetic materials and natural quartz. The physico-chemical properties of silica materials largely depend on the synthetic procedures used for their preparation. Therefore, we will briefly discuss silica synthesis processes.

Silica synthesis
Silica is mainly synthesized from an aqueous solution, with dissociated monomeric silicic acid, Si(OH) 4 , or from a vapor of a silicon compound such as silicon tetrachloride.
Waterglass is a concentrated alkaline sodium silicate solution with anhydrous composition corresponding to Na 2 SiO 3 . It is the most common reagent for silica production in aqueous solution. Waterglass is a sodium salt of silicic acid that forms silicic acid upon acidification. When the concentration of Si(OH) 4 exceeds about 2.10 -3 M, condensation to polysilicic acids ( Figure 1) occurs, thus leading to the formation of colloidal silica particles [15].
The polymerization and the formation of silica can be represented as follows: [Si n O 2n-nx/2 (OH) nx ] + m Si(OH) 4 [Si n+m O 2n-nx/2+2m(2-p) (OH) nx+4(m-p) ] + 2 pm H 2 O Where: n = number of silicon atoms in a polysilicic acid molecule or particle, x = number of OH groups per silicon atom in the polymer (0≤ × ≤ 3), m = number of monomeric silicic acid molecules added to the polymer, and p = fraction of the hydroxyl groups per monomeric silicic acid molecule that are converted to water during the polymerization reaction [15].
Amorphous silica particles are formed by polymerization of monomers in the aqueous solution supersaturated with silicic acid. Various silica materials are produced in liquid phase processes ( Figure 2).
Colloidal silica or silica sol is most often produced in a multi-step process in which the alkaline silicate solution is partially neutralized with a mineral acid. Alternatively, this pH neutralization can be achieved by electrodialysis. The resulting silica suspension is stabilized by pH adjustment. Finally a solid concentration up to 50 wt% is reached by water evaporation. Silica sol nanoparticles show a perfect spherical shape and identical size as a result of extensive Ostwald ripening [15]. Stöber silica sol is prepared by controlled hydrolysis and condensation of tetraethylorthosilicate (TEOS) in ethanol to which catalytic amounts of water and ammonia are added. The Stöber procedure can be used to obtain monodisperse spherical amorphous silica particles with tunable size and porosity [16].
Silica gel is obtained by destabilizing silica sol. Silica gel is an open 3-D network of aggregated sol particles. The pore size is related to the size of the original silica sol particles composing the gel.
Precipitated silica is formed when a sol is destabilized and precipitated.
Ordered mesoporous silica is obtained by a supramolecular assembly of silica around surfactant micelles. Typical surfactant molecules are amphiphilic polymers such as tribloc copolymers or quaternary alkylammonium compounds. These organic supramolecular templates are evacuated from the mesopores, typically via a Ordered mesoporous materials calcination step. Calcination is a controlled combustion process leading to oxidation and decomposition of the template molecules into small volatile products such as NO x , CO 2 and H 2 O, which can leave the pores. The diameter of the mesopores (2-50 nm) is determined by the type of surfactant applied [17,18]. A completely different synthesis route of amorphous silica starts from SiCl 4 in the vapor phase. Silicon tetrachloride is oxidized in a hydrogen flame at temperatures exceeding 1000°C and polymerized into amorphous non-porous SNPs. This nanopowder has very low bulk density and high specific surface area, typically 200 to 300 m 2 /g. This material is called pyrogenic or fumed silica, referring to the special synthesis conditions [15].
The synthesis of dense crystalline silica such as quartz from aqueous solution is a slow process requiring heating the solution to accelerate the formation process in a so-called hydrothermal synthesis [15]. Alternatively, under high pressure, amorphous silica can be transformed to crystalline material by microcrystallization. The appearance of quartz ranges from macroscopic crystals to microcrystalline powders. Large crystals are grown at high temperature and pressure in industry. Smaller quartz crystals are conveniently obtained by grinding large crystals. Alpha-quartz is formed under moderate temperature and pressure conditions and is the most abundant form of quartz. At temperatures exceeding 573°C, α-quartz can transform into β-quartz [19]. At atmospheric pressure and temperatures higher than 870°C, quartz is transformed into tridymite and at temperatures more than 1470°C into cristobalite [15,20]. These high-temperature polymorphs of quartz have the same elemental composition but a different crystal structure and can persist metastably at lower temperatures.
Dense and porous crystalline materials can be distinguished by framework density. The framework density is conveniently defined as the number of tetrahedrally coordinated atoms (T-atoms) per nm 3 . For dense structures, such as quartz, tridymite and cristobalite, values of 22 to 29 T-atoms/nm 3 are common, whereas for porosils belonging to the zeolite material family, as few as 12.1 T-atoms/nm 3 are present [21]. The framework structure of a porosil is denoted with a 3-letter code. Descriptions are available in the Atlas of Zeolite Framework Types [22].
Porosils are crystallized in aqueous media in the presence of organic molecules that act as porogens or template molecules defining the size and shape of the pores. Their evacuation is typically achieved through calcination. Among the porosils are clathrasils and zeosils [23,24]. Zeosils have cages with windows or channels of a sufficiently free dimension to allow molecules to diffuse in and out, a property known as molecular sieving [25].
Clathrasils have cages with windows that are delineated with a 6-membered ring of SiO units, thus presenting a free aperture of barely 0.28 nm. Even a molecule as small as oxygen has no access to the cavities of a clathrasil. The organic template molecules engaged in the crystallization of a clathrasil cannot be removed easily from the pores [23,24].
When heated above 1700°C, any type of silica (amorphous or crystalline) melts. During cooling, the disordered structure is solidified, and a dense amorphous silica glass or vitreous silica is formed [15].

Physico-chemical properties
The properties of silica materials considered essential for their potential toxicity are crystallinity, particle size and morphology, porosity, chemical purity, surface chemistry and solubility [26]. An overview of the properties of silica materials involved in silica toxicity is provided in Table 1.

Crystallinity
In crystalline structures such as quartz and porosils, the arrangement of atoms is ordered in all dimensions. According to the International Union of Pure and Applied Chemistry (IUPAC), the atoms must be arranged periodically with long-range order (at least 10 repeats in all directions) and produce sharp maxima in a diffraction experiment to observe x-ray diffraction (XRD) crystallinity [27]. The threshold for observing crystallinity depends on the unit cell size (size of the repeated unit in a crystal). For materials with large unit cells, such as porosils, the minimum particle size required is about 10 nanometers to observe a distinct, sharp XRD pattern. Amorphous silica may present some short-range order but lacks long-range order in 3 dimensions and does not exhibit a sharp XRD pattern. Of note, the surface of a crystal represents a discontinuity that can be seen as a defect. With the presence of a less-structured or even partially amorphous rim, crystals may behave like amorphous particles. Thus, particles with an ordering at limited-length scales or with amorphous regions may be classified as amorphous.

Particle size and morphology
Nanoparticles are obtained by direct synthesis of silica sol [15] or by crystallization of nano-sized crystals of quartz or porosils [25]. The particle size is determined by the synthesis parameters. Amorphous silica sol particles tend to adopt the spherical shape so as to reach a minimum of interfacial surface area. The particle size of commercial silica sols prepared from sodium silicate is from 10 to 25 nm (Figure 3 left). Sols with larger primary particles can be prepared from TEOS by Stöber synthesis, for example (Figure 3 middle). Grinding and milling processes reduce particle size. These techniques are most often applied to quartz, silica gel and vitreous silica. The obtained products generally have a broad size distribution. Crystalline particles exhibit crystal planes at the surface, and the morphology of the crystalline nanoparticles depends on the crystal class such as cubic, hexagonal, tetragonal, and orthorhombic (Figure 3 right). For all nanomaterials, in aqueous environment, the primary nano-sized silica particles tend to form aggregates.

Porosity
According to IUPAC [28], pores are classified according to their diameter into micropores (< 2 nm), mesopores (2-50 nm) and macropores (> 50 nm). Amorphous sol particles can be microporous or non-porous (dense). The porosity of Stöber silica can be tuned by adapting the synthesis parameters: decreasing the ratio of water to TEOS promotes particle growth by aggregating smaller sub-particles, thus leading to rough particle surfaces with micropores. In contrast, smooth particle surfaces are obtained with conditions of high ratio of water to TEOS [29]. Silica gel is a powder with particle size in the micrometer range or larger and is, typically, mesoporous.
Zeosils and clathrasils have characteristic pores and cages in the micropore size range, depending on framework topology. Examples of porosil frameworks are shown in Figure 4 [22].
When the silica is presented as a nanopowder, porosity can be an intrinsic and extrinsic characteristic: stapling of the elementary nanoparticles gives rise to an interparticle porosity, which often is difficult to distinguish from the intrinsic intraparticle porosity, especially when dealing with mesoporosity.

Hydrophilic-hydrophobic properties
The hydrophilicity of a silica material increases with the number of silanols, or silicon-bonded hydroxyl groups, capable of forming hydrogen bonds with physical water molecules. The chemical formula of silica is represented as SiO 2 .xH 2 O, in which water represents chemical water contained in silanol groups present on the surface of the silica material. These water molecules are not to be confused with crystal water, such as that present in many inorganic salt crystals. The surface chemistry of silica is depicted in Figure 5. Vicinal hydroxyl groups (one hydroxyl group per tetrahedron) located at mutual distances smaller than 3 nm are engaged in hydrogen bonding. Geminal hydroxyls (2 hydroxyl groups per tetrahedron) are considered to occur in minor concentrations. Isolated silanols are positioned too far apart to be engaged in hydrogen bonding. Because of the differing chemistry of these 3 types of silanol groups, they are not all equivalent in their adsorption behavior or chemical reactivity. Vicinal hydroxyls interact strongly with water molecules and are responsible for the excellent water adsorption properties of silica, which are exploited in industrial gas drying operations, for example.
The reported concentration of hydroxyl groups per square nanometer on the surface of amorphous silica ranges from 4 to 5 OH/nm 2 [12]. As compared with amorphous silica, the crystalline forms of silica generally contain a lower concentration of surface hydroxyl groups [15]. Hydrogen-bonded water molecules are removed when silica is heated at 170°C under atmospheric pressure or at room temperature under vacuum.
Colloidal silica, precipitated silica and ordered mesoporous silica and silica gel are hydrophilic because of their high concentration of silanols. Silicagel, for example, can adsorb water in quantities up to 100% of its proper weight.
Porosils typically are hydrophobic because they lack silanols in the pores of their framework. Silica produced at high temperature, such as pyrogenic and vitreous Figure 4 Atomic representation of (left) a zeosil with microporous channels (MFI type) and (right) clathrasil with a denser framework (SOD type). Black and gray circles represent silicon and oxygen atoms, respectively. Figure made with Vesta 2.0.3 [178] with unit cell coordinates from [22].  silica, or calcined at temperatures exceeding 800°C, is almost entirely dehydroxylated. In a dehydroxylation reaction, neighboring silanol groups are condensed into siloxane bonds ( Figure 5 bottom) and water molecules. Some isolated silanol groups may persist on the surface [15]. Because hydrogen bonding on siloxanes is unfavorable, dehydroxylated silica is hydrophobic. Grinding of hydrophobic bulk materials such as quartz and vitreous silica induces silicon and oxygen radicals and surface charges. These charges increase the hydrophilic surface [19,30].

Solubility
The dissolution and precipitation of silica in water chemically involves hydrolysis and condensation reactions, respectively, catalyzed by OHions ( Figure 1). For micrometer-sized nonporous amorphous silica, the equilibrium concentrations of Si(OH) 4 at 25°C in water corresponds to 70 ppm at pH 7. The silica solubility depends on the surface curvature of the (nano)particles. SNPs and nanoporous silica show enhanced equilibrium solubility, of 100-130 ppm [12]. According to Vogelsberger et al. [31], the solubilization of amorphous SNPs in physiological buffer at 25°C is accelerated because of the large surface area exposed. The solubility equilibrium is reached only after 24 to 48 h. Crystalline silica such as quartz has a much lower equilibrium solubility, of 6 ppm [15].
In summary, when dealing with silica, the physicochemical properties such as amorphous versus crystalline nature, porosity, particle size and degree of hydroxylation must be specified. An overview of silica materials described in the scientific literature and in the research and development environment is provided in Table 1.

Toxicity Of Silica
Background Health effects of silica and epidemiological studies Until recently, toxicological research into silica particles focused mainly on "natural" crystalline silica particles of 0.5 to 10 μm (coarse or fine particles). This research was/is fed by the clear association of occupational inhalation exposure and severe health effects, mainly on the respiratory system. The typical lung reaction induced by chronic inhalation of crystalline silica is silicosis, a generally progressive fibrotic lung disease (pneumoconiosis), exemplified by the development of silicotic nodules composed of silica particles surrounded by whorled collagen in concentric layers, with macrophages, lymphocytes, and fibroblasts in the periphery. Epidemiologic studies have found that silicosis may develop or progress even after occupational exposure has ended; therefore, above a given lung burden of particles, silicosis was suggested to progress without further exposure [32][33][34].
Calvert et al. [35] recently reported an association of crystalline silica (mainly quartz) exposure and silicosis, as well as lung cancer, chronic obstructive pulmonary disease (COPD), and pulmonary tuberculosis. The carcinogenicity of quartz and cristobalite has been shown in several epidemiological studies [36][37][38]. In 1997, the International Agency for Research on Cancer (IARC) classified some crystalline silica polymorphs (quartz and cristobalite) in group 1 (sufficient evidence for the carcinogenicity to experimental animals and to humans), whereas amorphous silica (silicon dioxide without crystalline structure) was classified in group 3 (inadequate evidence for carcinogenicity) [39]. This classification has recently been confirmed [40]. Checkoway and Franzblau [41] reviewed occupational epidemiologic literature on the interrelations among silica exposure, silicosis and lung cancer and concluded that the appearance of silicosis is not necessarily required for the development of silica-associated lung cancer. Hnizdo and Vallyathan [42] suggested that chronic exposure to levels of crystalline silica dust, which does not cause disabling silicosis, may cause chronic bronchitis, emphysema, and/or small airway disease leading to airflow obstruction, even in the absence of radiological evidence of silicosis. Evidence has linked silica exposure to various autoimmune diseases (systemic sclerosis, rheumatoid arthritis, lupus, chronic renal disease), as reviewed by Steenland and Goldsmith [43]. A study by Haustein et al. [44] reported on silica-induced (silica dust) scleroderma.
Amorphous silica has been far less studied than has the crystalline form [39]. Warheit [45] briefly described the inhalation toxicity data related to amorphous silica particulates and concluded that some forms of amorphous silica are more potent in producing pulmonary effects as compared to others. He also emphasized the great need for adequate toxicological testing of many of these amorphous silicates given their importance in commerce and widespread potential for exposure. Workers exposed to precipitated or fumed silica did not exhibit pneumoconiosis [46,47], but evidence of pulmonary fibrosis was reported in workers exposed to amorphous silica dust produced as a byproduct of silicon metal production [48]. Merget et al. [49] reviewed the current knowledge of the health effects of a wide range of amorphous forms of silica in humans. The major problem in the assessment of health effects of biogenic amorphous silica is its contamination with crystalline silica. This problem applies particularly to the well-documented pneumoconiosis among diatomaceousearth workers. Although the data are limited, a risk of chronic obstructive bronchitis disease, COPD or emphysema cannot be excluded [49]. Animal inhalation studies involving synthetic amorphous silica (colloidal silica, fumed silica and precipitated silica) showed at least partially reversible inflammation [50,51], granuloma formation and emphysema, but no progressive fibrosis of the lungs [52,53]. However, high doses of amorphous silica may result in acute pulmonary inflammatory responses, which could conceivably trigger long-term effects, despite a low biopersistence of the particles [54]. The debate on the health effects of micron-sized crystalline or amorphous silica is beyond the scope of this article. Readers are referred to other publications [35][36][37][38]41,[55][56][57].

Mechanisms of toxic action
As mentioned, most of the toxicological research into silica has focused on crystalline silica particles of 0.5 to 10 μm (coarse or fine particles). Despite the relatively large amount of available studies, the mechanisms of crystalline silica toxicity at the cellular and molecular levels are still unclear, and whether any single mechanism underlies all the above-mentioned diseases induced by these particles is uncertain [43]. However, severe inflammation following exposure to silica particles appears to be a common initiating step [58,59].
The crucial role of reactive oxygen species (ROS) in the inflammatory, fibrogenic and carcinogenic activity of quartz is well established [60,61]. Oxidative membrane and DNA damage are considered the most important mechanisms involved in the health effects of micronsized crystalline silica. A few of the numerous reports clearly demonstrate these findings: ROS generated by the silica surface can induce cell membrane damage via lipid peroxidation that may subsequently lead to increased cellular permeability [62], perturbation of intracellular calcium homeostasis [63] and alterations in signaling pathways. Schins et al. and Fanizza et al. [64,65] demonstrated that respirable quartz particles induce oxidative DNA damage in human lung epithelial cells. Li et al. [66,67] demonstrated that micron-sized quartz particles induce . OH generation through an irondependent mechanism. A close association of . OH and iron ion concentration has been reported for amorphous silica particles [66,67]. The study of Ghiazza et al. [30] indicates that crystallinity might not be a necessary prerequisite to make a silica particle toxic; both quartz and vitreous silica showed stable surface radicals and sustained release of HO . radicals. When tested on macrophages, vitreous silica and pure quartz showed a remarkable potency in cytotoxicity, release of nitrite and tumor necrosis factor α (TNF-α) production, suggesting a common behavior in inducing of oxidative stress [30]. Ding et al. [68] discuss the molecular mechanisms of silica-induced lung injuries with a focus on NF-kB activation, generation of cyclooxygenase II and tumor necrosis factor α (TNF-α). The review of Castranova [69] summarizes evidence that in vitro and in vivo exposure to crystalline silica results in activation of NF-kB and AP-1 signaling pathways. In vitro and in vivo animal studies, as well as investigations in humans, strongly support the role of macrophage products in the development and progression of silicosis [70]. Such products include a large panel of cytokines [71], with TNFα seeming to determine the development of silicainduced pulmonary fibrosis [72]. In addition, recent evidence implicates interleukin 1β (IL-1β) and its activation by the NALP-3 inflammasome [73].
A large body of experimental work in the past 20 years has shown that 2 main factors seem to govern the hazardous nature of crystalline silica: particle surface reactivity and the form of silica [74]. Fenoglio et al. [75] evaluated these factors systematically, studying synthetic quartz samples differing only in size and shape. Cytotoxicity appeared to be primarily governed by the form of the particles and the extent of the exposed surface. Several studies indicate that the surface silanol groups are directly involved both in membranolysis [76][77][78] and in toxicity to alveolar cells [79,80]. Therefore, the distribution and abundance of silanols determines the degree of hydrophilicity (see "Physico-chemical properties of synthetic silica materials related to toxicity" described above) and seems to modulate cell toxicity [80,81]. Experimental work with respirable silica particles and the survey of published data by Bagchi [82] suggest that the toxicity of these particles is caused by the large amount of positive charges they carry. Ghiazza et al. [83] reported that formation of a vitreous phase at the surface of some commercial diatomaceous earth prevents the onset of oxidative stress effects. Donaldson and Borm [84] emphasized that the ability of quartz to generate ROS could be modified by a range of substances that affect the quartz surface, such as substances originating from other minerals. The authors concluded that the toxicity of quartz is not a constant entity and may vary greatly depending on the origin/constitution of the sample.
The origin/synthesis of SNPs plays a crucial role in determining the physico-chemical properties of these particles and, consequently, their potential interactions with biological systems. Surface area, surface morphology, surface energy, dissolution layer properties, adsorption and aggregation properties are relevant parameters. Depending on the manufacturing process, amorphous silica has a wide range of physico-chemical properties that determine its industrial application. Bye et al. [85] showed that the cytotoxic activity of different forms of amorphous silica does not depend on a crystalline silica component but, rather, is caused by surface charges and the morphologic features of particles. Synthetic amorphous silica has been the subject of dissolution testing with a simulated biological medium, and the silica dissolution rate was reported as being more rapid than the reverse precipitation rate [86]. Solubility has been defined as a key driver in the clearance mechanisms involved in amorphous silica removal from lung [87]. Warheit [45] reviewed pulmonary responses to different forms of silica and reported that cristobalite produced the greatest lung injury, quartz produced intermediate effects, and amorphous silica produced minimal effects. In terms of analytical technique, small differences in dissolution exist among these different forms of silica, and dissolution, in turn, influences pulmonary effects through the concept of persistence. In addition, components from the biological system may react with the surface of the particle. A systematic investigation of iron-containing SNPs as used in industrial fine-chemical synthesis demonstrated the presence of catalytic activity that could strongly alter the toxic action of nanoparticles [88].
On the whole, considering the great variety of silica forms, degree of crystallinity, surface state and the presence of contaminants, there is a critical need for carefully characterized standard silica samples to unravel the relationships between physico-chemical factors and toxicity, both micron-and nano-sized. The main goal of this review is to focus on the toxicity of nanosilica, which has never been properly reviewed. Moreover, nanosilica occurs mainly in amorphous forms, and the potential hazard posed by these nanomaterials cannot be simply related to, as has already been reviewed many times, studies of micron-sized crystalline materials.

Silica nanoparticles
The growing abundance and industrial applications of nanotechnology has resulted in a recent shift of toxicological research towards nanoparticles [89][90][91][92][93][94]. Ultrafine particles (< 0.1 μm) have been demonstrated to cause greater inflammatory responses and particle-mediated lung diseases than have fine particles (< 2.5 μm) per given mass [95][96][97]. Also, experiments involving silica have shown that nanoparticles, both ultrafine colloidal silica [98,99] and crystalline silica [99], have a greater ability to cause lung injury as compared with fine particles. Thus, the unique properties (i.e., small size and corresponding large specific surface area; cell penetrating ability) of nano-sized SiO 2 are likely to impose biological effects that differ greatly from micron-scale counterparts.
In vitro studies of nanosilica toxicity A structured summary of in vitro studies of the toxicity of SNPs can be found in Table 2.
Chen and von Mikecz [100] investigated the effects of nanoparticles on structure, function, and proteasomal proteolysis in the cell nucleus by incubating different cell lines with unlabeled and fluorescently labeled amorphous silica particles of different sizes [100]. SiO 2 particles between 40 nm and 5 μm were applied to epithelial cells in culture and observed on confocal laser scanning microscopy with differential interference contrast. Particles of all tested sizes penetrated the cytoplasm; however, nuclear localization was observed exclusively in cells treated with SiO 2 nanoparticles between 40 and 70 nm. Fine and coarse SiO 2 particles (0.2-5 μm) were exclusively located in the cytoplasm and accumulated around the nucleus, forming nuclear indentations. The uptake of SNPs in the nucleus induced aberrant clusters of topoisomerase I and protein aggregates in the nucleoplasmthe former inhibiting replication, transcription, and cell proliferationwithout altering cell viability. Cells treated with fine (0.5 μm) or coarse (5 μm) SiO 2 particles had the same replication and transcription activity as that of untreated control cells [100].
Jin et al. [101] investigated the potential toxicity of luminescent amorphous SNPs (50 nm) in freshly isolated rat alveolar macrophage cells and human lung epithelial cells (A549 cells). The SNPs penetrated the cells but were not detected in the nuclear region and did not cause significant toxic effects at the molecular and cellular levels below a concentration of 0.1 mg/ml.
Lin et al. [102] investigated the cytotoxicity of amorphous (colloidal) SNPs (15 and 46 nm) in cultured human alveolar epithelial cells (A549 cells). Cell viability decreased in a time-and dose-dependent manner (down to 100 μg/ml), and nanoparticles of both sizes were more cytotoxic than were fine quartz particles (Min-U-Sil 5). Exposure to 15-nm SNPs generated oxidative stress in A549 cells as reflected by reduced glutathione (GSH) levels, elevated production of malondialdehyde (MDA) and lactate dehydrogenase (LDH) leakage, which is indicative of lipid peroxidation and membrane damage, respectively [102]. In the study by Wottrich et al. [103], A549 cells and macrophages (THP-1, Mono Mac 6) exposed to 60 nm amorphous SNPs showed distinctly higher mortality than did larger silica particles (diameter 100 nm). Another study by Choi et al. [104], involving A549 cells and amorphous SNPs (14 nm), showed a pro-inflammatory response triggered by nanoparticles without blocking cell proliferation or causing cell death to any great extent. A recent work by Akhtar et al. [105] examined cytotoxicity (by MTT and LDH assay) and oxidative stress (ROS levels, membrane lipid peroxidation, GSH level and activity of GSH metabolizing enzymes) in A549 cells exposed for 48 h to amorphous SNPs of 10 and 80 nm. The SNPs were cytotoxic to studied cells through oxidant generation (ROS and membrane lipid peroxidation) rather than depletion of GSH. Eom and Choi [106] studied oxidative stress caused by amorphous SNPs (7 and 5-15 nm) in human      • PEG-ylation of SNP protected the particles from protein adsorption on the external surface of the NPs and consequently no agglomeration in culture medium was observed.
• The availability of the particles to be internalized by the cells depended on the size and morphology of the aggregates. [107] exposed A549 cells to amorphous SNPs (10-20 nm) at concentrations up to 200 μg/ml and observed low cytotoxicity as measured by MTT and LDH assays. However, co-treatment with the same nanoparticles and lipopolysaccharide, a bacterial product that may contaminate (nano)materials, significantly enhanced the cytotoxicity.
Yu et al. [108] examined the cytotoxic activity (by MTT and LDH assay) of well-dispersed amorphous silica particles (30-535 nm) in mouse keratinocytes. All sizes of particles were taken up into the cell cytoplasm; nuclear uptake was not studied. The toxicity was dose and size dependent, with 30-and 48-nm particles being more cytotoxic than 118-and 535-nm particles. The reduced GSH level significantly decreased only after exposure to 30-nm nanoparticles [108]. Nabeshi et al. [109] showed the size-dependent cytotoxic effects of amorphous silica particles (70, 300 and 1000 nm) on mouse epidermal Langerhans cells. The smallest particles induced greater cytotoxicity (by LDH assay) and inhibited cellular proliferation (by [ 3 H]-thymidine incorporation). The observed effects were associated with the quantity of particle uptake into the cells.
Yang et al. [110] evaluated the effects of amorphous SNPs (15 and 30 nm) and micron-sized silica particles on cellular viability, cell cycle, apoptosis and protein expression in the human epidermal keratinocyte cell line HaCaT. Microscopy examination revealed morphological changes after 24-h exposure; cell growth also appeared to be significantly inhibited. The cellular viability of HaCaT cells was significantly decreased, and the amount of apoptotic cells was increased in a dose-dependent manner after treatment with nano-and micron-sized SiO 2 particles. Furthermore, smaller silica particles were more cytotoxic and induced a higher apoptotic rate. Proteomic analysis revealed differential induction of expression of 16 proteins by SiO 2 exposure; proteins were classified into 5 categories according to their functions: oxidative stress-associated proteins, cytoskeletonassociated proteins, molecular chaperones, energy metabolism-associated proteins, and apoptosis and tumor-associated proteins. The expression levels of the differentially expressed proteins were associated with particle size [110]. In a recently published study [111], the same research group used these SNPs to study the global DNA methylation profiles in HaCaT cells; the authors reported that nanosilica treatment can induce epigenetic changes.
Cousins et al. [112] exposed murine fibroblasts to small amorphous (colloidal) silica particles (7, 14 and 21 nm) over a long incubation period (1, 3 and 7 days and up to 7 weeks) and observed a distinctive cellular response affecting the morphologic features, adhesion and proliferation of the fibroblasts but not cell viability. Chang et al. [113] exposed selected human fibroblast and cancer cell lines for 48 h to amorphous SNPs and assessed cellular viability by MTT and LDH assays. Cytotoxicity was seen at concentrations > 138 μg/ml and depended on the metabolic activity of the cell line. However, the average primary size of tested silica particles was 21 and 80 nm, but their average hydrodynamic particle size was 188 and 236 nm, respectively, so in media, aggregates/agglomerates were formed.
In the study of Yang et al. [114], cell membrane injury induced by 20-nm amorphous silica nanoparticles in mouse macrophages was closely associated with increased intracellular oxidative stress, decreased membrane fluidity, and perturbation of intracellular calcium homeostasis.
Besides inhalation, ingestion is considered a major uptake route of nanoparticles into the human body [3]; however, the possible harmful effects of engineered nanoparticles in the gastrointestinal tract are still largely unknown. Recently, Gerloff et al. [115] investigated the cytotoxic and DNA damaging properties of amorphous fumed SiO 2 nanoparticles (14 nm) in the human colon epithelial cell-line Caco-2. Exposure to SNPs for up to 24 h caused cell mortality, significant DNA damage and total glutathione depletion. The results of an in vivo study of mice fed nanosized silica are discussed in section 3.2.2.
Ye et al. [116] reported on induced apoptosis in a human hepatic cell line after exposure to amorphous (colloidal) SNPs (21, 48 and 86 nm). The viability of  Exposures to the various quartz particles produced differential degrees of pulmonary inflammation and cytotoxicity, which were not consistent with particle size but correlated with surface activity, particularly hemolytic potential. [148] Silica dust 10 ± 5 nm; and 0.5-10 μm (80% of the particles 1-5 μm) •    Exposure to particles before a single TMA challenge resulted in only a slightly irregular breathing pattern during TMA challenge. Pre-exposure to particles also diminished the effect of TMA on tidal volume, laryngeal ulceration, laryngeal inflammation, and the number of BAL eosinophils in most animals. When the additional group of animals was exposed to nanosilica before a second challenge to TMA, the pulmonary eosinophilic infiltrate and edema induced by a second TMA challenge in control animals was diminished by the preceding silica exposure, but the number of lymphocytes in BAL was increased. [153] Napierska et al. Particle and Fibre Toxicology 2010, 7:39 http://www.particleandfibretoxicology.com/content/7/1/39 Table 3 In vivo studies on nanosilica particles (SNPs) toxicity (Continued) Amorphous silica~3 0 nm and 30 μm

• Size distribution
Feeding of mice for 10 weeks (total fed amount of 140 g/kg mice) • Blood analysis • Cytological analysis of lungs and liver tissue sections • Analysis of silicon in organs The nano-sized silica particle dieted group showed higher value of ALT (alanine aminotransferase) than normal and micron-sized silica dieted groups. H&E staining of the liver of the nano-sized particle dieted group indicated some fatty liver pattern. The contents of Si in the livers of the groups were almost the same.  cells was assessed by LDH and MTT assay; oxidative stress was studied by measurement of ROS, lipid peroxidation and GSH concentration; and apoptosis was quantified by annexin V/propidium iodide staining and DNA ladder assays. Nano-SiO 2 caused cytotoxicity in a size-, dose-and time-dependent manner. Because nanoparticles are probably distributed by the blood stream (e.g., with medical applications), endothelial cells would also come in direct contact with these particles, for pathogenic particle-endothelial interactions. Peters et al. [117] evaluated the effects of 4-to 40-nm amorphous SiO 2 particles in vitro on human dermal microvascular endothelial cell function and viability. The particles were internalized but did not exert cytotoxic effects (MTS assay). However, cells showed impaired proliferative activity and pro-inflammatory stimulation. Napierska et al. [118] reported a dosedependent cytotoxicity (by MTT and LDH assay) of monodisperse amorphous SNPs (16-335 nm) in a human endothelial cell line. The toxicity of the particles was strongly related to particle size; smaller particles showed significantly higher toxicity and also affected the exposed cells faster. Ye et al. [119] evaluated the toxicity of amorphous SNPs (21 and 48 nm) towards rat myocardial cells. Exposure to the SNPs for up to 48 h resulted in size-, dose-and time-dependent cytotoxicity, smaller particles again showing higher toxicity.
Barnes et al. [120] reported no detectable genotoxic activity (by Comet assay) of amorphous SNPs (20 nm to < 400 nm) in 3T3-L1 fibroblasts at 4 or 40 μg/ml silica for 24 h. The particle dispersions were carefully characterized and the results were independently validated in 2 separate laboratories. In a recent review, Gonzalez et al. [121], in a literature review, compared 2 genotoxicity teststhe alkaline Comet assay and the micronucleus test -in terms of chemical composition and size of engineered SNPs: engineered SNPs did not seem to induce DNA strand breakage. However, when monodisperse amorphous SNPs of 3 different sizes (16, 60 and 104 nm) were selected to assess the genotoxic potential of these particles in A549 lung carcinoma cells with a well-validated assay (the in vitro cytochalasin-B micronucleus assay), at non-cytotoxic doses, the smallest particles showed an apparently higher-fold induction of micronucleated binucleated (MNBN) cells [122]. When considering the 3 SNPs together, particle number and total surface area accounted for MNBN induction because they were significantly associated with the amplitude of the effect.

Crystalline nanosilica
Wang et al. [99] investigated cytotoxicity (by MTT assay) and genotoxicity of ultrafine crystalline SiO 2 particulates (UF-SiO 2 ) in cultured human lymphoblastoid cells. A 24-h treatment with 120 μg/ml UF-SiO 2 produced a fourfold increase in MNBN cells, with no significant difference as measured by the Comet assay. However, the ultrafine crystalline silica used was extracted from commercially available crystalline silica and the particle sizes were not uniform [99].

Mesoporous silica
The cytoxicity of amorphous mesoporous SNPs (MSNs) was recently studied intensively because they are promising materials for drug delivery systems and cell markers [8,123,124]. Several studies have demonstrated that efficient cellular uptake of MSNs could be achieved at concentrations < 50 μg/ml, with no cytotoxic effects observed up to 100 μg/ml in different mammalian cells [125][126][127][128][129][130]. Lu et al. [128] reported on the optimal size of 50 nm MSNs for cell uptake. Slowing et al. [131] reported that, contrary to the known cytotoxicity of amorphous SNPs toward red blood cells, mesoporous SNPs exhibit high biocompatibility at concentrations adequate for potential pharmacological applications.
However, studies have reported cytotoxicity of mesoporous silica nanomaterials. Tao et al. [132] investigated the The tissue distribution and excretion of the injected particles were different depending on particle size. As particle sizes increased, more particles were trapped by macrophages in the liver and spleen. All particles were cleared via urine and bile; however, the 50 nm silica nanoparticles excreted faster than the other two particles. [158] Napierska et al. Particle and Fibre Toxicology 2010, 7:39 http://www.particleandfibretoxicology.com/content/7/1/39 effects of two types of MSNs (pore diameters of 31 and 55 Å) on cellular bioenergetics (cellular respiration and ATP content) in myeloid and lymphoid cells and isolated mitochondria. Only cells exposed to MSNs with larger size and larger pores showed concentration-and time-dependent inhibition of cellular respiration, and both nanoparticles were toxic to the isolated mitochondria. Di Pasqua et al. [133] reported that the toxicity of MSNs towards human neuroblastoma cells was related to the adsorptive surface area of the particle. However, the nature of the functional groups playing a role could not be excluded. Vallhov et al. [134] investigated the effects of mesoporous SNPs of different sizes (270 nm and 2.5 μm) on human dendritic cells and found viability, uptake and immune regulatory markers affected by increasing size and dose. He et al. [135] evaluated the influence of size and concentration of mesoporous SNPs (190, 420 and 1220 nm) on cytotoxicity in human breast cancer cells and monkey kidney cells. The cytotoxicity of the particles was associated with particle size: silica of 190 and 420 nm in diameter showed significant cytotoxicity at concentrations > 25 μg/ml; whereas particles of 1220 nm in diameter showed slight cytotoxicity at 480 μg/ml. The smaller particles were suggested to be more easily endocytosed and consequently located within lysosomes [135].

Surface-modified/functionalized silica
Brown et al. [136] attempted to evaluate the role of shape in particle toxicity in the lung; the authors compared the response of rod-shaped and spherical amorphous silica particles (Stöber), not coated or coated with fibronectin or polyethylene glycol (PEG), under stretched and static conditions. The dosimetric comparison of materials with different shapes (e.g., needleshaped or acicular and isotropic) was not straightforward. Non-coated particles induced an increase in IL-8 and LDH release, whereas a surface modification with PEG mitigated this effect, which suggested the significance of adhesive interactions for membrane binding/ signal transduction, for example [136].
Diaz et al. [137] described the interactions of two amorphous silica particles -a pristine particle, without any coating, and PEGylated silica particles (average size 130 and 155 nm), as well as an iron oxide particle with a silica shell (80 nm)with different human peripheral blood cells, several human tumor cell lines and mouse peritoneal macrophages. The effects depended on the cell analyzed: although all particles were phagocytosed and were able to induce ROS expression in mouse macrophages, they differentially affected the human cell lines and peripheral blood cells, both in terms of internalization and ROS induction. The availability of the particles to be internalized by the cells seemed to strongly depend on aggregation, especially on the size and morphology of the aggregates [137].
Almost all of the existing cytotoxicity studies of SNPs involved monocultures of cells that are organ specific. The exception is the study by Wottrich et al. [103], in which co-cultures of epithelial cells (A549) and macrophages (THP-1, Mono Mac 6) exposed to 60-and 100nm amorphous SNPs showed an increased sensitivity to the cytokine release as compared with monocultures of each cell type. The enhanced responses to nanoparticles in different contact and non-contact co-cultures were reported in studies by Herseth et al. [138,139] with micron-sized crystalline silica, showing that more realistic models should be applied to study interactions between nanoparticles and cells or organs of interest.
Few recently published studies have systematically investigated nanomaterial properties in terms of the degree and pathways of cytotoxicity. Sohaebuddin et al. [140] selected nanomaterials of different composition, including silica, to analyze the effects of size and composition on 3 model cell lines: fibroblasts, macrophages and bronchiolar epithelial cells. The authors concluded that the physico-chemical properties of size and composition both determined the cellular responses and induced cellspecific responses. In another recent study, Rabolli et al. [141] studied the influence of size, surface area and microporosity on the in vitro cytotoxic activity of a set of 17 stable suspensions of monodisperse amorphous SNPs of different sizes (2-335 nm) in 4 different cell types (macrophages, fibroblasts, endothelial cells and erythrocytes). The response to these nanoparticles was governed by different physico-chemical parameters that varied by cell type: in murine macrophages, the cytotoxic response increased with external surface area and decreased with micopore volume; in human endothelial cells and mouse embryo fibroblasts, the cytotoxicity increased with surface roughness and decrease in diameter; and in human erythrocytes, the hemolytic activity increased with the diameter of the SNP [141].
Overall, most of these in vitro studies involving different SNPs documented the cytotoxic effects of these nanomaterials. The determinants of the observed cytotoxicity seem to be complex and vary with the particles used and cell type tested. Unfortunately, for many published studies, adequate material characterization is still missing. The mere cytotoxicity reported with some particles does not strictly imply hazard. However, this observation indicates that a proactive development of nanomaterials should consider physical, chemical and catalytic properties of nanoparticles.
In vivo studies of nanosilica toxicity Along with particle size, surface area and particle number appear to be integral components contributing to the mechanisms of lung toxicity induced by nano-sized particles. The high deposition rate of ultrafine particulates is a result of a small aerodynamic diameter and is assumed to be important in the lung inflammatory process. Some evidence suggests that inhaled nanoparticles, after deposition in the lung, largely escape from alveolar macrophage clearance and gain greater access to the pulmonary interstitium via translocation from alveolar spaces through epithelium [3,142]. A summary of the in vivo responses to SNPs can be found in Table 3.
In 1991, Warheit et al. [143] performed a rat inhalation study (nose-only) with an aerosol of colloidal silica (mass median aerodynamic diameter 2.9, 3.3 and 3.7 μm) for 2 or 4 weeks at concentrations up to 150 mg/m 3 , and some groups of rats were allowed to recover for 3 months. The inflammatory responses, mainly seen as increased numbers of neutrophils in bronchoalveolar lavage fluid (BALF), with the 2 and/or 4 weeks of exposure were evident at ≥ 50 mg/m 3 concentration. Three months after exposure, most biochemical parameters returned to control values [143].
Lee and Kelly [52] studied the effects of repeated inhalation (6 h/day, 5 days/week for 4 weeks) of an aerosol of colloidal silica (mass median aerodynamic diameter 2.9, 3.3 and 3.7 μm; concentration up to 150 mg/m 3 ) in rats. The authors reported a dose-dependent alveolar macrophage response, polymorphonuclear leukocytic infiltration, and type II pneumocyte hyperplasia in alveolar duct regions. Lung-deposited nanosilica were cleared rapidly from the lungs, with half-times of approximately 40 and 50 days for the 50 and 150 mg/m 3 treatment groups, respectively. The lungs did not show formation of fibrotic scar tissue or alveolar bronchiolarization [52].
Cho et al. [144] investigated inflammatory mediators (24 h, and 1, 4 or 14 weeks after exposure) induced by intratracheal instillation in mice of up to 50 mg/kg of ultrafine amorphous silica with a primary particle diameter of 14 nm. The authors observed significantly increased lung weights, total cell numbers and levels of total protein in BALF up to 1 week after treatment. The histopathological examination revealed acute inflammation, with neutrophils and chronic granulomatous inflammation. The expression of cytokines (IL-1β, IL-6, IL-8, and TNF-α) and chemokines (monocyte chemoattractant protein 1 and macrophage inflammatory protein 2) was significantly increased during the early stages, with no changes after week 1 [144].
Chen et al. [145] studied age-related differences in response to amorphous SNPs (average size 38 nm). Changes in serum biomarkers, pulmonary inflammation, heart injury and pathology were compared in young (3 weeks), adult (8 weeks) and old (20 months) rats that inhaled tested nanoparticles for 4 weeks (40 min/day). Old animals appeared to be more sensitive to nanoparticle exposure than were young and adult rats. The risk of pulmonary damage was old > young > adult, but the risk of cardiovascular disorder was observed only in old animals [145].
Kaewamatawong et al. [98] compared acute pulmonary toxicity induced in mice by ultrafine colloidal silica particles (UFCSs; average size 14 nm) or fine colloidal silica particles (FCSs; average size 213 nm) after intratracheal instillation of 3-mg particles. Histopathological examination with both sizes revealed bronchiolar degeneration, necrosis, neutrophilic inflammation, alveolar type II cell swelling and alveolar macrophage accumulation. However, UFCSs induced extensive alveolar hemorrhage, more severe bronchiolar epithelial cell necrosis and neutrophil influx in alveoli as compared with FCSs. Electron microscopy showed UFCSs and FCSs on the bronchiolar and alveolar wall surface and in the cytoplasm of alveolar epithelial cells, alveolar macrophages and neutrophils. The findings suggest that UFCSs (possibly linked to size and/or larger surface area) have a greater ability to induce lung inflammation and tissue damage than do FCSs [98]. The same research group reported acute and subacute pulmonary toxicity of low-dose UFCS particles in mice after intratracheal instillation [146]. Exposure of up to 100 μg UFCSs produced moderate to severe pulmonary inflammation and tissue injury 3 days after exposure. Mice instilled with 30 μg UFCSs and sacrificed at intervals from 1 to 30 days after exposure showed moderate pulmonary inflammation and injury on BALF indices at the acute period; however, these changes gradually regressed with time. Concomitant histopathological and laminin immunohistochemical results were similar to BALF data. The authors reported a significant increase in the apoptotic index (TUNEL) in lung parenchyma at all observation times. The findings suggest that instillation of a small dose of UFCSs causes acute but transient lung inflammation and tissue damage in which oxidative stress and apoptosis may be involved [146].
In a study of fibrogenesis, Wistar rats were intratracheally instilled with silica (of unknown composition) nano-(10 ± 5 nm) and microparticles (0.5-10 μm), and were sacrificed 1 and 2 months after dosing [147]. One month after instillation, cellular nodules (Stage I silicosis) were found in the nano-sized SiO 2 group, whereas more severe lesions were found in the micron-sized SiO 2 treatment group (Stage II and Stage II+ of silicotic nodules). One month later, the nano-sized SiO 2 group still showed only Stage I silicotic nodules, whereas the micron-silica group showed disease progression and Stage II+ and III silicotic nodules. Therefore, in rats, the effect of nano-SiO 2 on fibrogenesis might be milder than that of micron-SiO 2 . Nanoparticles, because of their size, probably diffuse more easily to other pulmonary compartments than do microparticles [147].
Warheit et al. [148] compared the toxicity of synthetic nanoquartz particles (12 and 50 nm) to mined Min-U-Sil quartz (500 nm) and synthetic fine-quartz particles (300 nm) and (2) evaluated the surface activity (hemolytic potential) of the different samples in terms of toxicity. Rats were instilled with the different particle types (1 or 5 mg/kg), and pulmonary toxicity was assessed with BALF biomarkers, cell proliferation, and histopathological evaluation of lung tissue at 24 h, 1 week, 1 month, and 3 months after exposure. Exposure to the quartz particles of different sizes produced pulmonary inflammation and cytotoxicity, with nanoscale quartz of 12 nm and Min-U-Sil quartz being more toxic than fine quartz and nanoscale quartz of 50 nm. The pulmonary effects were not consistent with particle size but were associated with surface activity, particularly hemolytic potential [148].
In a recent work by Sayes et al. [149], rats inhaled freshly generated aerosolized amorphous SNPs of 37 and 83 nm for a short-term period. In contrast to previous studies' measurements, particle number rather than particle mass was chosen as dose metrics (3.7 × 10 7 or 1.8 × 10 8 particles/cm 3 ) for 1-or 3-day exposure. Pulmonary toxicity (cell counts, differentials, enzymatic activity of LDH and alkaline phosphatase (ALP) in BALF) and genotoxicity endpoints (micronuclei induction) were assessed from 24 h up to 2 months after exposure. One-or 3-day aerosol exposure produced no significant pulmonary inflammatory, genotoxic or adverse lung histopathological effects in rats exposed to very high particle numbers in a range of mass concentrations (1.8 or 86 mg/m 3 ).
Recently, airway irritants were suggested to facilitate allergic sensitization [150][151][152]. Arts et al. [153] examined the effect of pre-exposure to synthetic (fumed) amorphous SNPs (14 nm) on elicitation of airway hypersensitivity reactions by the low-molecular-weight allergen trimellitic anhydride (TMA). Brown Norway rats were topically sensitized with TMA, exposed (head or nose only) to SNPs for 6 h/day for 6 days and then challenged by inhalation with a minimally irritating concentration of TMA. One day later, breathing parameters, cellular and biochemical changes in BALF, and histopathological airway changes were studied. Exposure to SNPs alone resulted in transient changes in breathing parameters during exposure and in nasal and alveolar inflammation with neutrophils and macrophages. Exposure to particles before a single TMA challenge resulted in only a slightly irregular breathing pattern during TMA challenge. Interestingly, pre-exposure to particles diminished the effect of TMA on tidal volume, laryngeal ulceration, laryngeal inflammation, and the number of BALF eosinophils in most animals. When an additional group of animals was exposed to nanosilica before a second challenge to TMA, the pulmonary eosinophilic infiltrate and edema induced by a second TMA challenge in control animals was diminished by the preceding silica exposure, but the number of lymphocytes in the BALF was increased. The authors concluded that SNPs could reduce as well as aggravate certain aspects of TMA-induced respiratory allergy [153].
As mentioned, next to inhalation, ingestion is considered a major route for the uptake of nanoparticles in the human body. So et al. [154] studied the effects on mice fed nano-and micron-sized amorphous silica particles (30 nm and approximately 30 μm, respectively). After feeding the animals for 10 weeks (total amount of 140 g silica/kg mouse), blood was tested biochemically and hematologically. The group fed SNPs showed higher serum values of alanine aminotransferase as compared with the other groups (both control and micron-silica treated). Although the contents of Si in the livers of the groups were almost the same, hematoxylin and eosin staining revealed a fatty liver pattern in the group treated with SNPs [154].
The successful use of nanoparticles in the clinic requires exhaustive studies on the behavior of these particles in vivo. Unfortunately, biocompatibility, biodistribution and clearance studies of silica-based nanoparticles are sparse. Kumar et al. [155] used nanoparticles of organically modified amorphous silica (ORMOSIL; amino-terminated; 20-25 nm) to study biodistribution, clearance and toxicity in a mouse model. Particles conjugated with fluorophore and radiolabeled were injected systemically in mice. Biodistribution studies showed a greater accumulation of nanoparticles in liver, spleen and stomach than in kidney, heart and lungs. Over 15 days, almost 100% of the injected nanoparticles were effectively cleared out of the animals via hepatobiliary excretion, without any sign of organ toxicity. Hudson et al. [156] examined the biocompatibility of mesoporous silica particles (150 nm, 800 nm and 4 μm) after injection in rats and mice. When the particles were injected subcutaneously in rats, the amount of residual material decreased progressively over 3 months, with no significant injury to surrounding tissues. Subcutaneous injection of the same particles in mice produced no toxic effects. In contrast, intra-peritoneal and intravenous injection in mice resulted in death; microscopic analysis of the lung tissue of the mice indicated that death might have been due to pulmonary thrombosis. Nishimori et al. [157] evaluated the acute toxicity of amorphous silica particles (70, 300 and 1000 nm) after a single intravenous injection in mice and reported that 70-nm silica injured the liver but not the spleen, lung or kidney. Moreover, chronic administration of 70-nm nanoparticles (injections every 3 days for 4 weeks) caused liver fibrosis. Cho et al. [158] examined the impact of the size of amorphous SNPs on toxicity, tissue distribution and excretion. Fluorescence dye-labeled 50-, 100-and 200-nm silica particles were intravenously injected in mice. The incidence and severity of inflammation with the 100-and 200-nm SNPs was significantly increased in the liver at 12 h; the 50-nm particles induced a slight but nonsignificant inflammatory response. The tissue distribution and excretion of the injected particles differed depending on particle size. With increasing particle size, more particles were trapped by macrophages in the liver and spleen. All particles were cleared via urine and bile; however, the 50nm SNPs were excreted faster than were the other 2 particle sizes [158].
In vivo versus in vitro; amorphous versus crystalline Park and Park [159] performed in vitro and in vivo studies to investigate oxidative stress and pro-inflammatory responses induced by amorphous SNPs (average primary size 12 nm). RAW 264.7 cells derived from mouse peritoneal macrophages were exposed to SNPs (5-40 ppm) in vitro and showed ROS generation and decreased intracellular GSH levels, as well as increased levels of nitric oxide released from the cultured macrophage cell line. In vivo, mice were treated with a single intraperitoneal dose of 50 mg/kg of nanosilica. The treatment produced activated peritoneal macrophages, increased blood level of IL-1β and TNF-α, and increased level of nitric oxide released from peritoneal macrophages. Ex vivo, cultured peritoneal macrophages harvested from the treated mice showed the expression of inflammation-related genes (IL-1, IL-6, TNF-α, inducible nitric oxide synthase, cyclooxygenase 2). In the spleen, the relative distribution of natural killer cells and T cells was increased 184.8% and 115.1%, respectively, as compared with control animals, and that of B cells was decreased to 87.7% [159].
Kim et al. [160] addressed the toxicity of nano-and micron-sized silica particles (14 nm and 1-5 μm, respectively) in vitro and in vivo. In vitro, RAW 264.7 cells were exposed to both particle sizes for 24 h, and the cell viability was decreased in dose-dependent manner; however, apoptosis was observed only after treatment with nanoparticles. In vivo, mice received up to 5 mg/kg silica particles via oropharyngeal aspiration. Again, sizedependent toxicity of silica was observed; pulmonary injury and neutrophilic infiltration were greater after treatment with nano-sized SiO 2 particles than with micron-sized silica [160].
Sayes et al. [161] assessed the capacity of in vitro screening studies to predict in vivo pulmonary toxicity of several fine or nanoscale particle types in rats. For the in vitro component of the study, rat lung epithelial cells, primary alveolar macrophages and alveolar macrophages-lung epithelial cell co-cultures were incubated with quartz particles and precipitated amorphous silica. In the in vivo component of the study, rats were exposed by intratracheal instillation to the same particles. In vivo, pulmonary toxicity studies demonstrated that crystalline silica particles produced sustained inflammation and cytotoxicity, whereas amorphous silica particles produced reversible and transient inflammatory responses. Ex vivo, pulmonary inflammation studies showed that crystalline and amorphous silica-exposed rat lung epithelial cells did not produce MIP-2 cytokines, but alveolar macrophages and, to a lesser degree, co-cultures secreted this chemotactic factor into the culture media. In vitro cytotoxicity studies demonstrated a variety of responses to the different particle types, primarily at high doses. When considering the range of toxicological endpoints, comparisons of in vivo and in vitro measurements revealed little correlation, particularly when considering the many variables assessed in this study such as cell types used, culture conditions and time course of exposure, as well as measured endpoints.
To summarize, extrapolating (or comparing) the results obtained in vitro to the in vivo situation is difficult and applies not only to toxicity studies with nanoparticlesany existing in vitro test system lacks the complexity of animal models or the human body. However, considering the number of particles and the number of possible properties of these particles that may vary (size, shape, coating, etc.), clearly, not all can be evaluated in in vivo studies, and scientists have been striving to determine the correlation between the results obtained from in vitro and in vivo toxicity assessments. Although little correlation has been found in these studies with nanosilica [159][160][161], Lu et al. [162] tested a panel of metal oxide nanoparticles and could predict the inflammogenicity of tested nanomaterials with a battery of simple in vitro tests. Similar conclusions were drawn in a recent study by Rushton et al. [163]; the authors could predict the acute in vivo inflammatory potential of nanoparticles with cell-free and cellular assays by using NP surface area-based dose and response metrics. The authors also found that a cellular component was required to achieve a higher degree of predictive power.
Established and validated co-culture systems may provide a tool to better mimic the in vivo system. Using recently developed 3-D cell cultures and improving the exposure system (likewise exposure at the air-liquid interface of a human epithelial airway model reported by Brandenberger et al. [164]), could substantially improve the outcome from in vitro studies with nanomaterials.

Conclusions
Silica or silicon dioxide (SiO 2 ) is, in many forms, abundantly present in our natural environment. The adverse health effects, including lung cancer, of naturally occurring crystalline silica such as quartz and cristobalite have been thoroughly documented in occupational settings. Naturally occurring amorphous silica such as diatomaceous earth is considered less harmful. Most of the synthetic (manufactured) silicas used in a large variety of applications are amorphous. For silica in general, the property most significantly linked to the toxicological potential is the crystallinity. For micron-sized crystalline silica, oxidative stress and, linked to it, oxidative DNA and membrane damage, are probably the most important mechanisms involved in the inflammogenic and fibrogenic activities (reviewed by [60]) and/or carcinogenic activity [39,165], for example. These mechanisms do not apply to amorphous silica, which has therefore been far less studied. Moreover, the adverse health effects of biogenic (natural) amorphous silica is often attributed to a certain degree of contamination with crystalline silica [49]. Synthetic amorphous silica (colloidal silica, fumed silica and precipitated silica) is not involved in progressive fibrosis of the lung [52,53]; however, high doses of amorphous silica may result in acute pulmonary inflammatory responses [54].
Interest in using SNPs is growing worldwide, especially for biomedical and biotechnological applications such as cancer therapy, DNA transfection, drug delivery, and enzyme immobilization [5][6][7][8][9]. In general, SNPs are synthetic, which has an advantage over natural silica in that they contain fewer or no impurities than do natural silica, and the physico-chemical properties are known and well controlled during production. Exposure to SNPs during the production process and their downstream use is probably minimal for sols and gels because the nanoparticles are trapped/immobilized within their matrix. However, the inhalation potential of low-density fumed silica powders or freeze-dried nanoparticles may be high without adequate precautions.
Results of a growing number of in vitro studies indicate that the particle surface area may play a crucial role in the toxicity of silica [75,166]. The cytotoxic activity of silica particles can be related to their surface interfacing with the biological milieu rather than to particle size or shape [75]. Surface silanol groups are directly involved (as shown in vitro) in hemolysis [76][77][78] and in alveolar epithelial cell toxicity [79,80]. This observation indirectly links the hydrophilicity to cellular toxicity [80,81]. The size and surface physico-chemical features of SNPs contribute decisively to the biological effects of SiO 2 nanoparticles. The complexity of protein-SNP interactions should not be underestimated; these interactions appear to be affected by the size of SNPs as well [167][168][169][170][171]. The effect of other physico-chemical properties of SNPs on health, such as porosity, chemical purity, surface chemistry and solubility, are less well studied, and therefore no definite conclusions can be formulated (summary of the data can be found in Table 2). Comparison of published studies leads to the conclusion that even a small modification of the surface can result in a more or less marked change of a biological effect [2,3,172]. Few in vitro studies have emphasized that the response to SNPs varies by cell type [137,140,141].
Considering the use of SNPs for medical applications, biocompatibility and toxicokinetics need to be documented in great detail because, despite no observation of acute (cyto)toxicity, the uptake of the particles by cells may eventually lead to perturbation of intracellular mechanisms. For instance, the ability of silica-coated nanomaterials to penetrate the blood-brain barrier supports the urgent need for extensive studies to clarify the potential chronic toxicity of these materials [14]. The successful use of nanoparticles in the clinic requires exhaustive and elaborate in vivo studies [155]. Of note, the toxicity of SNPs can depend on not only the material itself but also the administration route to the living body, as was shown by Hudson et al. [156]: subcutaneous injection presented good biocompatibility, whereas intraperitoneal and intravenous injection led to fatal outcomes.
Unfortunately, only limited short-term and no chronic in vivo studies of SNPs are available (summary of the data is found in Table 3), and the current data do not clarify whether amorphous SNPs -showing augmented cytotoxicity and presumably processing oxidative DNA damaging potentialare less or more harmful as compared with micron-sized silica.
Determining the association of results from in vitro and in vivo toxicity assessments is difficult; however, the common feature seems to be cytotoxicity and inflammatory response after exposure to SNPs.
To conclude, the available studies of the toxicity of SNPs are relatively few, especially as compared to the vast number of studies of titanium dioxide or carbon nanotubes. Besides the relative lack of information on the safety or hazards of SNPs, often conflicting evidence is emerging in the literature as a result of a general lack of standard procedures, as well as insufficient characterization of nanomaterials in biological systems. For all studies, a crucial issue remains the careful, accurate characterization of particle size and morphologic features (especially in the biological media used for experimental set-up), composition, particle surface area and surface chemistry [173]. Moreover, equally important to the physico-chemical characterization of the material is the control of assays and assay conditions [174,175]. Only with the complete description of the NP and assay can the results of reported studies be comparable with those of other studies conducted with similar nanomaterials [159,176].
Until now, the health effects of SNPs have mainly been studied in terms of exposure via the respiratory tract, after acute or sub-acute exposure; other exposure routes should also be checked (e.g. blood, skin, gastrointestinal tract). Studies of chronicity are needed to supplement and verify the existing data. Information is insufficient to clearly identify and characterize the health hazards SNPs pose, and defining the appropriate conditions for safe use of these materials is currently not possible.