Citation

PengJia Jia, Bu-Yun Ma, Xia-Wei Wei, Zhi-Yong Qian. The in vitro and in vivo toxicity of gold nanoparticles[J]. Chin. Chem. Lett., 2017, 28(4): 691-702  

PengJia Jia^a, Bu-Yun Ma^b, Xia-Wei Wei^a, Zhi-Yong Qian^a     

With the fast development of nanotechnology and its wide application in the biomedical field, the toxic effect of nano materials has attracted numerous attention. Consequently, an important subfield termed nanotoxicology has emerged, which is generally defined as the study of the interactions between nanomaterials and biological systems with an emphasis on identifying the relationship between the physicochemical param eters (e.g. size, shape, surface modification of nanoparticles) and the occurrence of toxic effects. Gold is highly unreactive and chemically inert by nature, so the bulk gold is known to be nontoxic. When existing as molecular form, gold salts works as catalysts [1, 2]. In the clinic, they are even used to treat rheumatoid arthritis [3, 4]. Different from bulk or molecular-scale gold, when it comes to nanometer scale, gold exhibits various advantageous properties such as tunable sizes, facile synthesis, easy modifica tion, and strong optical properties such as surface enhanced roman scattering [5, 6], surface plasmon resonance [7], two-photon luminescence [8] and diverse morphology, including spheres, rods, hexapods stars and so on [9-11]. Based on these special properties, nanoparticles can been widely used as functional materials [12, 13]. Gold nanomaterials are thus very attractive in biomedical field. Numerous reports are published about the vrious application of gold nanomaterials in biological system. For example, our group has designed NIR-responsive drug delivery systems based on gold nanoparticles to achieve the purpose of combined photothermal-chemotherapy [14, 15]. Many other appli cations include: great potential for NIR-responsive controlled release [16] effective delivery of chemical drugs and genes [17], theranostic application [18, 19], promising agents in bio-imaging [20], bio-sensing and as immunotherapy carriers [21, 22].

For any novel nanomaterials applied in biological systems, their potential toxicities are highly concerned and should be evaluated comprehensively both in vitro and in vivo with enough attention paid to the biocompatibility and safety issues. Due to the numerous applications of gold nanoparticles in biomedical area, a strong interest has been exerted on exploration of their in vitro and in vivo adverse effects [23, 24]. Until now, there have been a large number of researches concerning cellular uptake, intracellular trafficking and localization, cell cycle steps, apoptosis, DNA damage, uptake, biodistribution, accumulation, removing and circulation of gold nanoparticles in the living systems, their effect on the immune system, the origin of toxicity and the mechanism involved. However, among all the previous investigations there are two conflicting conclusions available. A small number of groups insisted that gold nanomaterials are essentially non-toxic while most others demonstrated the existence of toxicity in their researches [25-29].

The emergence of various opinions is acceptable as the variations in parameters such as size, shape, surface charge and coating material can lead to different results when determining gold nanoparticles' interactions with biomolecules, cells lines and tissues. Because every group carried out their research under certain experiment conditions from their own perspectives, the variety in experiment conditions including gold nanoparticles with various sizes and shapes, different particle surface modifications, cell lines and animal models, tissues and organs examined, administration routes and doses applied, the time of exposure and examination, assays for assessing gold particle toxicity, methods for detecting the gold concentration in specific sites and distribution of particles over cells and so on, which undoubtedly led to corresponding conclusions which may be different even contradictory with each other. Even though, it is generally accepted that plain gold nanoparticles are toxic both in vitro and in vivo in certain range of concentrations. With proper surface modifications the toxic effect can be reduced or even eliminated. Considering the above mentioned complexity in toxicity evaluation of gold nanoparticles, it is very essential to frequently summarize and review the up to date published reports focusing on gold nanoparticles toxicity in spite of the fact that some reviews have been published on this topic previously. Many reviews have been published in this field [2-7]. For example, Khlebtsov et al. provided deep and meaningful discussions about the toxicity of gold nanomaterials both in vivo and in vitro and overviewed the existing literatures comprehensively. Their review exhibited a clear structure and is helpful to direct investigators in this area to perform the work to do more efficiently [30-34].

In this work, we review and describe the recent researches in the field of cytotoxicity and biological uptake with gold particles of diameters ranging from 1 nm to 200 nm, various shapes including sphere-, rod-, shell-, cage-, star-like ones and nanocluster. We focus on the latest renewed data available to present a comprehensive overview about the effect of gold nanoparticle when exposed to living systems. Besides, we also provide detailed information about cells lines and animal models, administration routes and doses of gold nanomaterials, exposure time and so on. With both the methods and the results being analyzed and combined together, we try to fully explain how the ultimate results are affected by the original design and hence to help investigators evaluate how far they have progressed more efficiently and based on the current knowledge, so that they can plan their following studies with more clear objectives.

The size of gold nanoparticles remains a key parameter that dominates their properties. Since particle size controls endocytosis effectiveness, cellular localization and accumulation sites in vivo, it is natural to understand that the cytotoxicity of gold materials depends on particle size [35, 36]. With the nanometer size, usually 1 nm–200 nm in diameter, gold nanoparticle can be easily taken up by cells, and of course the process of taking up is closely related to the size. It has been proved that the uptake of them is mediated by nonspecific adsorption of serum proteins onto the gold surface, and they are taken up into the cells via the receptor mediated endocytosis pathway [39]. Ma and coworkers synthesized gold nanoparticles with diameters of 10 nm, 25 nm and 50 nm. 50 nm ones were most readily internalized by cells, followed by 25 nm and 10 nm ones. When the size examined among a larger range of 10 nm–100 nm, 50 nm gold spheres were still taken up more easily and efficiently. These results revealed that cellular uptake of gold nanoparticles were heavily dependent upon particle size [37–40]. Pan et al. examined in detail the toxicity of gold nanoparticle raging from 0.8 nm to 15 nm with the cell lines L929, HeLa, J774A1, and SKMel-28. They found that the cytotoxicity of gold nanoparticles depended on their size and the 1.4 nm gold nanoparticle is the most cytotoxic, with the IC₅₀ value ranging from 30 mmol/L to 46 mmol/L. The IC₅₀ values of 0.8, 1.2, and 1.8 nm gold nanoparticle were 250, 140, and 230 mmol/L, respectively, much lower than that of 1.4 nm ones. At contrast, the 15-nm gold nanoparticles were nontoxic even at the 60-fold higher concentrations than the smaller particles. These data obviously suggested a size-dependent cellular toxicity of gold nanoparticles. Besides, the authors also demonstrated that the cell necrosis occurred after co-cultured with the 1.4-nm particles for 12 h while in the same condition the 1.2-nm clusters resulted in cell apoptosis [41]. The same group further investigated toxicity of 1.4 nm GNP. As ligand chemistry could influence the cytotoxicity of ultra-small gold nanoparticles, particles of similar size (1.4 nm) with glutathione ligand and TPPMS were explored for their cellular toxicity. The results demonstrated when capped with GSH, much less toxicity was caused by 1.4 nm ones [42]. When gold nanoparticles of three core sizes 1.5, 4, and 14 nm were applied to the cell line hESCs (Human Embryonic Stem Cells), only 1.5 nm particles showed toxicity to hESC viability while the other two nanoparticles exhibited almost no toxic effects [43]. This result is consistent with previous findings [42]. Further cytotoxicity study with gold particles in three diameters: 2–4 nm, 5–7 nm, and 20–40 nm demonstrated that both the decreased size and the increased concentration induced more severe cytotoxicity. Due to the negative surface charge, gold nanoparticles can be adsorbed by serum proteins and thus contribute to cellular uptake. As we know, a higher amount of gold material uptake by cells may generate larger cytotoxicity. So the increased amount of gold nanoparticle uptake may account for the toxicity and the immunological response [44]. However, when human cerebral microvascular endothelial (hcMEC/D3) cells and human dermal microvascular endothelial (HDMEC) cells were incubated with 10 nm, 11 nm, and 25 nm AuNPs no toxicity was observed, a dose-dependent release of LDH was seen after incubation with the particles [45]. 2-nm citrate-stabilized gold particles were toxic to PC-3 and MCF-7 cells in a time-and dose dependent manner when tested by both MTT and LDH. The 17 nm citrate coated gold particles have shown toxicity to A549 cells assessed by both MTT and LDH assays as well as by the ATP depletion measurements [46, 47].

45 nm particles were more markedly toxic than 13 nm ones. This might be due to the greater damaging effect of the 45 nm gold nanoparticles on vacuoles and correspondingly, to the greater release of these gold nanoparticles into the cytoplasm, with the normal cell function being disrupted [48]. 7 nm gold rods were prepared and compared with 14 nm particles. The two particles both with ideal photothermal properties, are coated with CTAB, BSA and Oleate. Further in vitro and in vivo studies showed that 7 nm gold nanorods had lower cytotoxicity, higher cellular uptake, more rapid biodistribution as well as more efficient clearance in vivo [49]. Ligand-free gold nanoparticles with different diameters were synthesized labeled as small (1–3 nm), medium (6–7 nm) and large (15–20 nm) by using molecular beam epitaxy process. The cytotoxicity of gold nanoparticles was investigated with human chronic myelogenous leukemia K562Cells. The small particles showed the most evident toxicity as shown in Fig. 1. MTT assay showed there is no toxicity of 10 nm–50 nm citrate-coated gold nanoparticles to embryonal fibroblasts with the maximum particle concentration of 300 mmol/L. However, the cell morphology was changed under such high concentrations. So the concentration of gold nanoparticles has an effect too [50]. The toxic effects of gold nanoparticles 5 nm and 15 nm on Balb/3T3 mouse fibroblast cell line was investigated, and autophagosomes were observed after cell exposure to particles of 5 nm. Reduced clathrin expression, protein cleavage and a time-dependent increase of Au uptake were observed [51]. Further, it's proved that the autophagosome accumulation induced by gold nanoparticles is size-dependent [44].

Figure 1

	Download: JPG larger image
Figure 1. The greater damage to the cell membrane caused by CTAB-gold nanorods changed the structure of the cell membranes and induced subsequent cell death. Reproduced with permission from Ref. [52]. Copyright 2013 RSC.

Therefore, particles size is an important factor concerning the toxic effect on the cellular level. The cytotoxicity of gold nanoparticles varied significantly with different sizes. However, despite differences in the size of gold nanoparticles used as well as the cell types implemented, results presented in the current literatures may not allow one to draw meaningful conclusions as to the regular relationship between the particle size and the toxicity.

Application of gold nanoparticles for therapeutic or diagnostic purposes often involves functionalization of nanomaterials with specific biomolecules such as peptides, ligands or chemical groups to improve their biocompatibility and achieve specific purposes such as targeted delivery of drug, nucleic acid, or co-delivery to cells or disease sites. So investigating the cellular toxicity of gold nanoparticles with regard to surface modification remains worthwhile.

CTAB is the structure-directing agent used to control gold nanorod shape, and it can form a tightly bound cationic bilayer on the surface of gold nanoparticles with a positive charge. To elucidate how the surface modification of rod-shaped gold particles influences their interaction with cell membrane and further induced toxic effects, CTAB and PDDAC coated rod particles with positively charge, together with negatively charged PSS and HS-PEG-COOH coated rod-shaped particles were prepared. PDDAC capped gold rods, with a much higher positive charge than CTAB– gold rods, had almost no effect on cell viability suggesting that the nature of the surface molecules especially their assembled structures, hydrophobic or hydrophilic properties probably play more crucial roles in determining cytotoxicity than surface charge alone. A positively charged CTAB bi-layer on the gold nanorod surface is the dominant factor causing damage to the cell membrane and subsequent cell death. Further coating of polyelectrolyte multilayers such as PSS and PDDAC can efficiently reduce the cytotoxicity. It also proved that the toxicity of CTAB capped gold rods mainly came from the CTAB conjugated gold rods themselves rather than the free CTAB. Their damage to cells was measured by the Live–Dead assay as shown in Fig. 1 [52]. However, a quite different conclusion was made by another groups, they pointed that it is the free CTAB rather than those bound to rods that are responsible for the cytotoxicity. PAA and PAH as negatively and positively charged polyelectrolyte were respectively used to overcoat the original CTAB-coated GNRs. It's proved that the free CTAB in the solution contributes most to the cell damage occurred, while after coated with polymers PAA and PAH, the toxicity of gold rods both reduced greatly no matter whether their surface charge is positive or negative. This is probably because the polymers on the surface prevented desorption of CTAB bound to GNRs [53]. The same group also evaluated the potential effects of gold rods on blood vessels by modifying them with the similar method used in above research. Four kinds of surfactant-capped gold rods were prepared by recoating them with PAA, PAH on the surface of CTAB and PEG to replace the CTAB bilayer. Vascular endothelial and smooth muscle cells of isolated rat aortic rings are used as model cells. PAA-gold rods, PAH-gold rods induced similar adverse effects to the vascular endothelium and this is in correspondence with their previous work. PEG coated gold nanoparticles are nontoxic at all, highlighting the importance of surface modification to prevent the toxicity of nanoparticles to vascular endothelium in blood vessels. Interestingly, any significant death of smooth muscle cells is not observed, which means that the cell types used is also able to influence the results [54]. Besides, Phosphatidylcholine overcoating is another method used to reduce the cytotoxicity of CTAB-coated gold nanorods [55]. To note, among these surface coating agents, PEG is well-known to reduce nonspecific binding of biological molecules to surfaces and avoid macrophage recognition and phagocytosis, a prolonged circulation and enhanced permeability and retention has been shown for PEG-coated particles. So reduce cytotoxicity from CTAB by overcoating the nanorods with poly (ethyleneglycol) (PEG) is a common metnod [56, 57].

As another gold nanoparticle stabilizers, the presence of sodium citrate on the surface of gold nanoparticles also influence the safety of gold nanoparticles. For example, they reduce the viability in MTT assay and impaired the proliferation in the human alveolar type-Ⅱ (ATII)-like cell lines A549 and NCIH441. However, the uptake of gold particles was not influenced. Uptake and cytotoxicity study of citrate-coated gold nanospheres on human endothelial and epithelial cells revealed a consistent results. Therefore, the free sodium citrate should be reduced to the amount on with which the stability of the particles and the safety for biomedical applications are both guaranteed [58, 59].

To examine the effect of surface hydrophobicity on acute and long-term nanoparticle cytotoxicity, a series of gold nanoparticles were synthesized which were modified with quaternary ammonium functionalities of different lengths (a systematically varied (C1–C6) hydrophobic alkyl tail). Both cellular toxicity and ROS production increased with increasing alkyl chain length, interestingly, DNA damage decreased with increasing particle hydrophobicity [60]. Studies on how the PEG shell influence the interaction of gold with cells indicated that thinner, more hydrophilic coatings, together with the partial functionalization with quaternary ammonium cations, resulted in a more efficient cellular uptake, which relates to significant effects on structural and functional cell parameters [61]. A specific research focused on the PEG density and hydrodynamic volume's influence on the interaction between gold nanoparticles and the cells. With three PEG densities coated on two different gold cores, it showed that the core size and PEG hydrodynamic volume are the primary factors determining the cell viability and uptake, whereas the PEG density was the main reason caused cell cycle arrest and DNA damage. Besides, these interactions between gold nanoparticles and cells also depended on cell lines applied [62]. To evaluate the impact of polyethylene glycol-coated gold nanoparticles to skeletal muscle cells viability, gold nanoparticles were employed to co-culture with differentiated skeletal muscle C2C12 cells, and it is found cell viability remain unchanged, but intracellular ATP levels and mitochondrial membrane potential increased, detect of multiple cytokines showed a huge increase of INF-c and TGF-b1 which suggest that PEG-AuNP could display a potential ability to activate inflammation and fibrosis on in vivo complex systems, contributing to cellular vulnerability and susceptibility to death stimuli [63]. Air– liquid interface cell (ALICE), which mimics the physiological state of aerosol were introduced to study the cellular entering and intracellular fate of 15 nm particles before and after PEG modification, human alveolar epithelial cells (A549) were treated with both PEG modified and plain gold nanoparticles for three times. Their results show that the PEG on the surface had an effect on the cellular uptake and trafficking [64]. So PEG coating is a constant method used to modify gold nanoparticles and has been widely applied to reduce toxicity and improve biocompatibility.

In real biological system, gold nanoparticles could interact with biomolecules, mostly serum proteins. When BSA (bovine serum albumin) protein was used to coat the CTAB bilayer on the surface of the gold nanorods through electrostatic interactions, the nonspecific binding between particles and cellular membranes could be avoided. And still with an aptamer constructed together the modified gold nanoparticles can actively targeted to disease sites with limited toxic effect [65]. BSA can also been chosen as a model protein to bind around the gold rods. The interaction of the protein corona-gold particles revealed that the binding was very stable. Cellular study demonstrated the corona played a protective role to the cell membrane [66]. So the proper biomolecules modification can have an effective impact to decrease the toxicity of gold nanoparticles. The uptake of anti-EGFR gold cage has been evaluated with U87MGwtEGFR cancer cells by using two photons microscopy. The uptake of gold cages increased with the increase in number for anti-EGFR per gold particle, while very few PEGylated ones were attached to or internalized into the cancer cells within 3 h incubation. The uptake of gold nanocages was also size-dependent, with the 35 nm one showing the largest number as compared to 50 nm and 90 nm samples [67]. However, it's reported that binding of gold nanoparticles with ligants can lead to cell signaling change when compared with free ligands. For example, the 40–50-nm Herceptin-coated gold nanoparticles altered cellular apoptosis by influencing the activation of caspase enzymes [35].

The positive surface charge exerted a great influence on the cellular toxicity, consistent with previous in vitro studies that positively charged NPs were more easily transported into cells because of the electrostatic interaction with negatively charged cell membrane, and then resulting in the breakage of cell membranes. In the contrast, the anionic surface groups functionalized gold nanoparticles are much safer [68, 69]. So cellular membrane potential plays a prominent role in intracellular uptake of gold nanoparticles. Positively charged nanoparticles depolarize the membrane to the greatest extent with nanoparticles of other charges having negligible effect. Such membrane potential perturbations result in increased [Ca²⁺], which in turn inhibits the proliferation of normal cells whereas malignant cells remain unaffected [70]. So the surface modification of gold nanoparticles with different ligands which brings changes mainly on the surface chemistry and surface charge, proved to greatly affect their intracellular affects.

To identify how the shapes of gold particles influence their toxicity effect, multi-shaped gold particles were synthesized by researchers and they have explored the influences of shapes. For example, Wang et al. prepared three gold nanoparticles with different shapes including rod-, cage-and hexapod-like particles. Hexapod-like particles exibited the lowest cytotoxicity and the highest cellular uptake in vitro for both as-prepared and PEG modified particles as shown in Fig. 2 [71]. To have a better understanding about how the shapes of particles affected their toxic effects, uptake of multi-shaped gold nanoparticles should been studied first. When two shapes of rod and sphere are compared within the 10–100 nm range, spheres were taken up more efficiently than nanorods [40]. A more recent study revealed that the spherical gold particles exhibit the fastest internalization rate, followed by the cubic ones, then rod-and disk-like particles. To note, Star-shaped particles can be quickly wrapped by the cell membrane, similar to their spherical counterparts [72]. It is known that functionalized gold NPs can translocate through cell membranes, resulting in membrane damage. The shape effects were investigated by including high-index faceted shapes such as rice and pyramid, and been compared to their less-faceted, similarly sized rod and cone-shaped counterparts. Using advanced molecular dynamics simulation techniques is able to compute translocation rate constants of functionalized cone-, cube-, rod-, rice-, pyramid-, and sphere-shaped particles through lipid membranes. The computed results indicate that depending on the nanoparticle shape and surface charge, the translocation rates can span 60 orders of magnitude [73, 74]. Further study of gold pariticles shaped in rod, polyhedron and sphere was assessed in animal model of zebrafish. Gold sphere particles exhibited more toxicity when compared with rod-and polyhedron-like ones. After storage, polyhedrons elicited obvious lethality. In the investigated multi-shaped nanoparticles, the gold nanorods showed less lethality [75]. However, when Au spheres, cubes, and rods, on a retinal pigment epithelial (ARPE-19) cell line was assessed by MTT, Au rods were less biocompatible than 10-nm spheres [76]. These differences looks confused and there are many factors to consider when one try to explain these results. The gold nanosphere of approximately 61.46 nm had greater toxicity in contrast to the gold nanostar with diameters of approximately 33.69 nm on human skin fibroblasts and rat fat pad endothelial cells [77]. Gold nanorods with different aspect ratios and surface modifications were synthesized. The surface chemistry is the key factor determining the cytotoxicity of gold nanorods while their aspect ratio didn't cause the toxic effects. Mitochondria are involved in autophagy induced by CTAB coated gold nanorod, the autophagy is mediated through intracellular reactive oxygen species. CTAB coated gold nanorods also induced apoptosis in human tumor cells [78]. Radioactive ¹⁹⁸Au nanostructures with a similar size but different shapes including sphere-, disk-, rods-like and cubic cages were synthesized as shown in Fig. 3. Significantly higher tumor uptake was observed for the gold nanospheres and nanodisks compared to the gold nanorods and nanocages at 24 h after injection. It is found that both the gold nanospheres and hexapods were observed on the surfaces of the tumors, the gold nanorods and nanocages were distributed throughout the tumors [79].

Figure 2

	Download: JPG larger image
Figure 2. (A) Cell viability ofMDA-MB-435 cells after incubation with the as-synthesized Au nanohexapods, nanocages, and nanorods for 48 h. (B) Cell viability of MDA-MB-435 cells after incubation with the PEGylated Au nanohexapods, nanocages, and nanorods for 48 h. Reproduced with permission from Ref. [71]. Copyright 2013 ACS.

Figure 3

	Download: JPG larger image
Figure 3. TEM images of the four different types of Au nanostructures (a) nanospheres, (b) nanodisks, (c) nanorods, and (d) cubic nanocages. Reproduced with permission from Ref. [79]. Copyright 2014 ACS.

Thus far, most reported studies have focused on in vitro toxic assays, however, data obtained from such studies may not correspond to in vivo results because the fate of gold nanoparticles in vivo is much more complicated. Therefore, to accurately assess nanotoxicity, in vivo studies are very essential. The overall behavior of gold nanoparticles in vivo included these steps: particles could enter the body via six major routes: intra venous, dermal, subcutaneous, inhalation, intraperitoneal, and oral. Next, absorption can occur where the particles interact with biological components such as proteins. Afterward they can distribute to various organs in the body and may remain their own structure, be modified, or metabolized, then they enter the cells of the organ and reside in the cells for an certain amount of time before moving to other organs or to be eliminated. To note, excretion could also occur earlier [34]. Many factors including dose applied, routes of administration, interaction with biomolecules, metabolism, excretion, as well as the particles properties such as size impact the actual toxic effect of gold nanoparticles in vivo.

Many investigator have reported that the particle size strongly influence the in vivo fate of gold nanoparticles. Besides, Surface charge was observed to affect the biodistribution of gold nanoparticles with positively charged particles accumulated mainly in the kidneys while negative and neutral particles in the live [80, 81]. Biodistribution of gold nanoparticles in vivo was accessed in kinds of animal models to understand how the propertied of gold particles influence the distribution of gold in different tissues and organs. A kinetic study was conducted to determine the influence of particle size on the organ distribution of gold nanospheres. 10, 50, 100 and 250 nm gold nanoparticles were intravenously injected respectively. 24 h later, the rats were sacrificed to determine the amount of gold present in blood and various tissues quantitatively with inductively coupled plasma mass spectrometry (ICP-MS). The 10 nm particles were present in every tissue including blood, liver, spleen, kidney, testis, thymus, heart, lung and brain, while the larger particles were only detected in blood, liver and spleen. So tissue distribution of gold nanospheres is size-dependent with the smallest 10 nm nanoparticles showing the most widespread organ distribution [82]. Aimed to consider the physico-chemical characteristics which determine the absorption across intestinal membranes as well as the accumulation in secondary target organs, five different sized gold nanoparticles (1.4, 5, 18, 80 and 200 nm) as well as opposite surface charged (positive and negative) at equal size (2.8 nm) were applied through intraoesophageal administration in healthy adult female rats. An overall biodistribution of the applied gold nanoparticle was investigated after 24 h. The results showed that absorption of NPs across intestinal membranes and the consequent accumulation in secondary organs is to a large part dependent on the size and surface charge of the particles: a smaller size and a negative charge generally led to a higher absorption and further accumulation. 18 nm particles were absorbed across intestinal barriers and accumulated in specific secondary organs to a higher amount than smaller particles, this is probably due to selected protein binding. The highest absorption across intestinal barriers were found for the smallest 1.4 nm gold particles, while for the 2.8 nm particles the negative charge is favored over positive charge. Size and surface charge remain two major properties influence the fate in the organism [83, 84]. Zhang et al. investigated the in vivo toxicity of 5, 10, 30, and 60 nm PEG-coated gold nanoparticle. It is found that most of the 5 nm and 10 nm particles accumulated in the liver, while the 30 nm particles preferentially gather in the spleen and the 60 nm particles with limited accumulation in the organs. They concluded that the toxicity of the 10 nm and 60 nm particles was obviously higher than that of the 5 nm and 30 nm particles, which is not consistent with previous in vitro findings, the smaller the particle is more toxic. So the toxicity of PEG-coated gold particles in vivo is complicated, and it cannot be simply concluded that the smaller particles have greater toxicity [85]. With gold nanorod, gold nanocage, and gold nanohexapods, biodistribution study evaluated how the particles with different shapes distributed in blood, tumor and organs. The results confirmed that the shape or morphology of gold nanoparticles could influence their blood circulation and biodistributions [71]. Analysis of biodistribution of 150 nm gold nanoshells show that no toxicity was observed in mice 6 months after administration of PEG modified gold nanoshells [86]. Several analytical techniques were integrated to explore the fate and desirable characteristics of gold nanoparticles in vivo. The results revealed that gold nanorods were mainly enriched in liver and spleen tissues for 28 days. Moreover, subcellular localization and elemental analysis of Au NRs showed that Au NRs could remain in lysosomes [87]. Yasuyuki Akiyama et al. first developed a method using PEG to replace the CTAB bilayer on the GNR, and then studied the biodistribution and cytotoxicity of CTAB and PEG coated gold nanorods, their PEG modification achieved reduced toxicity and longer circulation time. To further exam the relationship between the biodistribution of the gold nanorods and their EPR effect in the tumors in detail, they systematically prepared a series of PEG modified gold nanorods with different molar ratios. And further screening of the PEG-GNR through in vivo study concluded that a molar ratio of 1.5 was sufficient to confer prolonged circulation of gold nanorods in the blood and to show an EPR effect via the tail vein injection in the tumor-bearing mice. To clarify how the PEG modified gold nanorods accumulate in the tumors, nanorods were injected intravenously into tumor-bearing mice and the mice were sacrificed at 72 h, then quantitative analysis were conducted and the result demonstrated improvement of the EPR effect. Besides, high doses, exceeding 39.0 mg of Au, aided the accumulation of gold nanorods in the tumor tissue too [57, 88]. Lee et al. injected gold nanoparticles into the cerebral cortex, since nestin expression is mainly activated in astrocytes after CNS damage and as a sensitive signal for reactive astrocytes, nestin expression were examined. It is found that Au NPs toxicity is dependent on the dose or size administrated after the injected Au NPs into the brain, and small particle size Au NPs appeared more nestin expression compared to large particle size at short term implantation [89]. It has been reported that inert 40 nm gold nanoparticles injected through vein tail were identified in almost all Kupffer cells one day after the injection, but the fraction of goldloaded cells gradually decreased to about one fifth after 6 months. They were removed from the circulation system mainly by Kupffer cells of the liver, and the elimination of these nanoparticles is a long-term process, it is observed that over time the nanoparticles become clustered inside lysosome/endosome-like vesicles. These results indicated that even after the normal life span of the mouse, most of the gold nanoparticles will remain in the liver [90]. Cho et al. found the13 nm PEG-coated gold nanoparticles induced acute inflammation with neutrophils influx in the mouse liver, the author assumed that there are two phases of toxicity concerned with inflammation in the liver. The first one occurred immediately after administration of nanoparticles and the other at 7 days when the particles disappeared from the circulation and localized in the tissues, mostly in the liver and spleen. Also at 7 days posttreatment, the percentage of the cells undergoing apoptosis in the liver increased [91]. Wang et al. studied the biodistribution, excretion, and toxicity and evaluated the biocompatibility of gold nanoclusters with positive, negative, and neutral surface charged gold nanoclusters, the three gold particles were coated with GSH on the surface and thus were of similar diameter. The influence of surface charge on biodistribution and excretion was investigated over a relatively long period of 90 days. Compared with neutral Au NCs, negatively charged Au NCs were accumulated longer in liver and spleen, presumably due to capture by Kupffer cells and macrophages. Positive Au NCs caused some minor damage to the peripheral blood system, but normal functioning was recovered within the 90-day trial [92]. Kidneys are particularly susceptible to xenobiotics due to their high blood supply and ability to concentrate toxins. Sereemaspun et al. demonstrated that GNPs were heavily taken up by kidney cells, causing nephrotoxicity. Abdelhalim and Jarrar clearly observed damage to proximal tubular epithelial cells in rats exposed to GNPs [93, 94].

To clarify the influence of administration on the toxicity, an animal toxicity study using 13.5 nm gold nanoparticles in mice was carried out by using three different injected routes, namely oral, intraperitoneal, and tail vein injection. Consequently, the oral and intraperitoneal injection show the highest toxicity, and tail vein injection shows the least toxicity. The weight changes were shown in Fig. 4 [95]. The biodistribution of particles after a single intravenous injection of low concentration of gold nanoparticles in more than 25 organs were investigated for two months in their study. Persistent accumulation of Au NPs in liver and spleen were observed during the two-month-long time. Gradual accumulation of Au NPs in kidney, testis and blood, along with reduction of Au NPs in urine, feces and lungs at one month and two months postinjection clarified that Au NPs were inefficiently removed from the body through the urine and feces and gold nanoparticles could be redistributed in the body after one month. Unlike inhalation exposure, intravenous injection resulted in substantial accumulation in liver. Genes that were altered in common for both liver and spleen after Au NPs injection were analyzed. The down-regulated genes in the liver include many which are related to cell cycle processes. The spleen showed many more genes that were downregulated than up-regulated after particles injection [96]. To know more about the mechanisms involved in in vivo transport and biodistribution of gold nanoparticles, the biodistribution of 10 nm particles in rats after inter-vaginal space injection was studied with the intravenous injection as a control. As being shown in Fig. 5, the distribution percent of the gold nanoparticles in the skin, muscle, heart, intestines, and lungs at the early time points were higher after inter-vaginal space injection than after intravenous injection. Thus an obviously different transport and biodistribution pathway for gold nanoparticles maybe exists between the two administration routes and the distribution of particles to some organs is mediated not via the circulation, but rather via loose connective tissues [97]. In another study, the kinetics after intravenous and intra-tracheal applications of PEG modified gold nanoparticles as well as naked ones was explored and the gold contents in various tissues, organs and its excretion were quantitatively detected 1 h and 24 h later. Naked particles and PEG modified ones accumulated primarily in liver and spleen after vein tail injection while after intra-tracheal application the majority of all three types of applied gold nanoparticles stayed in the lungs. The total translocation towards the circulation did not change much after PEGylation of the particles. But the PEGylation of 5 nm particles influenced the kinetics process especially the uptake in liver and spleen after intravenous injection. Besides, a prolonged blood circulation time was determined for gold nanoparticles with PEG chains so PEGylation presented an effective method to prolong the blood circulation time [98]. The acute and chronic toxic effects of gold was an interesting point, quantified the bioaccumulation of Au in male Sprague Dawley rats after repeated intravenous injection. The accumulation amount of Au in organs was detected to reduce in this turn: Liver, spleen, lung skeleton. No acute or chronic toxicity were observed that can be attributed to the use of gold nanoparticles [99]. To evaluate the bioaccumulation and toxic effects of three different doses of 12.5 nm gold nanoparticles in vivo after repeated administration, the amount of gold in every organ was detected quantitatively by ICP-MS and GF-AAS, their results show the highest accumulation in liver, followed by kidney and spleen. Besides no obvious toxic effects were observed in vivo study, maybe because these particles are synthesized with citrate and negative-charged [100]. After intra-tracheal instillation the biodistribution pattern was quite different for both NPs. Small Au NPs (1.4 nm) can be translocated through the air/blood barrier of the respiratory tract in large amounts and then distribution in the whole body, while 18-nm particles are almost completely trapped in the lungs. The translocated material shows a different biodistribution pattern compared with intravenously injected 1.4-nm NPs. Besides, there is evidence that the surfaces of the NPs are modified during the translocation process [101]. As transparent zebrafish embryos possess a high degree of homology to the human genome, by using the zebrafish as a model to assess to toxicity of gold nanomaterials, it's demonstrated that gold nanoparticles are almost inert [102].

Figure 4

	Download: JPG larger image
Figure 4. Body weight changes in mice treated with gold nanoparticles 1100 μg/kg by using three administration routes, ie, oral, intraperitoneal, and tail vein injection. Reproduced with permission from Ref. [95]. Copyright 2010 DOVE Press.

Figure 5

	Download: JPG larger image
Figure 5. The relative distribution proportion of the Gold nanoparticles in the spleen, kidneys, lungs, intestines, and heart of SD rats at 5 min, 15 min, 30 min, 1 h, 4 h, 12 h, and 24 h after IVI (a), and after ISI in the tarsal tunnel (b). Reproduced with permission from Ref. [97]. Copyright 2016 Springer.

Toxicity may not be the only adverse effect of gold nanoparticles, they may also affect the immunological response of cells. Because macrophages play a prominent role in regulation of inflammatory response, their interactions with functionalized nanoparticles can be used for preliminary toxicological assessments of nanomaterials. Shukla et al. tested immunogenic effects of gold nanoparticles on RAW264.7 macrophage cells. The cells showed greater than 90% viability with no increase in proinflammatory cytokines TNF-a and IL-1b after 48 h of up to 100 mm gold-nanoparticle treatment [103]. In another study, PEG coated gold nanoparticles did not appear to induce cytotoxic effects, but they could up-regulate the immune response induced by extrinsic stimuli LPS. PEG coated gold nanoparticles enhanced LPS-induced inducible nitric oxide synthase (iNOS) expression and interleukin-6 (IL-6) production in RAW264.7 cells. The mechanism is related to activation of p38 mitogen-activated protein kinases (p38 MAPK) and nuclear factor-kappa B pathways [104]. Brandenberger et al. devised an epithelial-airway model contains human monocytederived macrophages and dendritic cells. Inflammatory responses were not noted when exposed to 15 nm gold nanoparticles suggesting that they do not elicit immune reaction [105]. However, after coated with peptides, gold nanoparticles were recognized by primary murine macrophages and induced an immune response as secretion of IL-6, IL-β and TNF-α were detected, so the peptide coating on gold nanoparticles can improve the immune effect. According to this conclusion, we can design hybrid nanoparticles for modulation of immune responses to treat allergies, cancer, and autoimmune diseases. The role of plasma proteins adsorbed to nanoparticles following entry into circulation also attracts attention as this could possibly interfere with the presentation of other ligands attached to the particles [106]. Ka Lun Cheung et al. evaluated whether the CTAB and PEG coated GNRs would cause any adverse effects in the human immune system, in particular the induction of allergies. They found that the CTAB coated GNRs released more allergic mediators such as histamine from human basophil KU812. Also, the CTAB coated gold nanorods induced more apoptosis than the PEG coated gold nanorods in KU812 24 h after treatment. So the surface material is important for the allergic reaction [85, 107]. It is also found that the immunological response of macrophages increased with the decreased size of gold nanoparticles [44]. As immune effector cells, when macrophages were exposed to the gold nanoparticles, spread morphology and larger sizes were observed which represented signs of cell activation. The immunological response was also investigated in terms of proinflammatory gene expressions. Up-regulated interlukin-1(IL-1), interlukin-6 (IL-6), and tumor necrosis factor (TNF-a) genes were detected [44].

In order to figure out the origins of cytotoxicity of gold nanomaterials, pure gold nanoparticles without any modifiers on the surface were synthesized, 2–4 nm, 5–7 nm, and 20–40 nm. The smaller particles induced more serious toxic effect as it is easier to been taken up into cells and then exert damage to cells [44]. Jia and coworkers reported that the gold nanoparticles can catalyze NO production from endogenous RSNOs in blood serum. The later reaction of the released NO can eventually result in cell necrosis or apoptosis. So these particles probably exerted toxic effects on living body by a dose-dependent increase in reactive nitrogen species (RNS) levels [108]. As explained by authors, gold nanoparticles induce the endogenous reactive oxygen species (ROS) generation, this oxidative stress environment could initiate the autophagic process, which can destroy foreign molecules and protect cell from death. Gold nanoparticles which produce more reactive oxygen species, can be possibly degraded through autophagy and thus exerted less damage to DNA [60, 109]. To know more about the relationship between gold nanoparticles and the oxidative stress, Jasmine et al. evaluated how they interact with lung cells. When treated with gold nanoparticles, autophagy was observed in MRC-5 human lung fibroblasts, as well as formation of autophagosomes, up regulation of autophagy proteins. However in another study, the author clarified how gold nanoparticles induced autophagosome accumulation with three sized gold nanoparticles. They also concluded that gold nanoparticle induced autophagosome accumulation is size dependent, and it is resulted from the blockade of autophagic flux [110]. Both 5 nm and 13 nm GNPs could increase autophagy and apoptosis in hypoxic human renal proximal tubular cells under hypoxic environments. The results were shown in Fig. 6 [111]. It is also observed that the cell death induced by gold nanoparticles including necrotic and apoptotic phenotypes. Analysis of the relevant proteins and gene expressions with omics approaches suggested the involvement of ER stress in response to gold nanoparticle treatment. Besides, as shown in Fig. 7, cell viability was related to particle size [112]. This conclusion is also consistent with another group's research in which it has been proved that after entering into cells, the gold nanoparticles induced oxidative damage, as well as autophagosome formation possibly to protect the cell from damage by oxidative stress [110].

Figure 6

	Download: JPG larger image
Figure 6. Cytotoxicity of normoxic and hypoxic HK-2 cells by MTT (A) and LDHassays (B) after incubation with the 5 nm and 13 nm GNPs (0, 1, 25, and 50 nmol/L) for 24h. Reproduced with permission from Ref. [111]. Copyright 2014 DOVE Press.

Figure 7

	Download: JPG larger image
Figure 7. Viability of K562 cells in the media after exposure to gold nanoparticles with three different diameters labeled as small (1–3 nm), medium (6–7 nm) and large (15–20 nm) for 48 h. Reproduced with permission from Ref. [112]. Copyright 2011 ACS.

When A549 cells exposed to gold nanoparticles for 24 h, early and late apoptosis along with necrosis were observed. Further study demonstrated that the gold particles induce apoptosis in A549 cells through ROS generation, sensitization of the mitochondrial membrane and cell cycle arrest. In addition, cell cycle analysis also showed that these particles arrest the cell cycle at the G0/G2 phase in A549 cells. However, when the human normal cell line HBL100 was assessed under the same condition, lower toxicity was caused. Based on the different toxic effect of gold nanoparticles against the two different cell lines, the author highlighted anticancer effect of them [113].

In another study, the authors observed the effects of GNRs on carcinoma cells (A549 cells), normal bronchial epithelial cells (16HBE cells), and primary adult stem cells (MSC cells). They found that the similar uptake pathway existed for the three different cells. But after uptake, the intracellular fates of gold nanoparticles differed as the resistance of endosomal/lysosomal membrane to Au NRs is different for the three cells. For A549 cells, gold particles were translocated from endosomes/lysosomes to mitochondria, resulting in decreased mitochondrial membrane potentials, increased oxidation stress and finally caused cell death. However, there is almost no toxicity in 16HBE and MSC cells since their lysosomal membranes remain more resistant to gold nanorods and so they did not leak from lysosomes [114]. The transferrin-coated gold nanoparticles were taken up into the cells via a receptormediated clathrin-dependent endocytosis pathway. Both uptake and removal of gold nanoparticles were highly dependent upon the size of the gold nanoparticles but the trends were different. The removal of the transferrin-coated gold nanoparticles was linearly related to size: The smaller particles exocytose at a faster rate and at a higher percentage than large ones. Furthermore, rod-shaped gold particles showed a lower uptake in comparison to spherical shaped nanoparticles. The rates of uptake were lower with an increasing aspect ratio. The fraction of rod-shaped NPs exocytosed was higher than spherical-shaped nanostructures [115]. Pan et al. found that 1.4 nm gold nanoparticles coated by TPPMS induced oxidative stress, mitochondrial permeability transition and then cell death by necrosis occurred. Continuous ROS generation was actually the toxicity mechanism causing cell damage. As for the cell death, the author demonstrated that cells were killed through necrosis rather than apoptosis. Gene expression analysis show that stress-related and inflammation related genes were up regulated while cell-cycle related genes are down regulated in particles treated cells. This result implied that repair response aimed at early damage failed and later cellular necrosis took place [42]. Spherical gold nanoparticles of 50–5 nm increased the level of reactive oxygen species, and induced apoptosis and obvious apoptotic morphological changes in A431 cells. These effects are associated with interference in mitochondrial membrane potential. This reduction in mitochondrial membrane potential probably initiated the apoptotic cascade in the cells exposed to the particles [116].

Tsoli et al. reported a strong toxicity of Au⁵⁵ clusters to kinds of human cancer and healthy cell lines, and the reason maybe lie in the special relationship between Au⁵⁵ and DNA, which is based on the perfect combination of cluster size and major groove dimension [117]. Cell-impedance measurements were applied to achieve all-time monitoring of gold nanoparticle cytotoxicity. The mechanistic studies show that the gold nanoparticles significantly regulate gene expression and protein function to manage apoptosis and cell cycle progression, and these actions are dependent on cell type and cellular context, as well as particle size [118].

In this mini-review, we provide a general discussion on the assessment of gold nanomaterial toxicity. So far, with the available data collected, we can preliminarily conclude that the gold nanoparticle is toxic when used in biological systems in certain range of concentrations. We summarize from the in vitro researches that the gold nanoparticles often induce ROS production after entering the cells and then lead to further oxidative stress-related cytotoxicity such as DNA damage, cell death (apoptosis and necrosis) and cell cycle arrest in consequence. So ROS is regarded as a common mechanism causing the toxic effects. Of course these toxic effects at cellular levels are influenced by the cell types selected. However, results from in vitro are not reliable enough for the toxicity assessment of the real systems so verification from animal experiments will be required in order to clarify if in vitro results keep consistent with in vivo observations. Such comparisons between in vitro and in vivo studies are very essential [31, 34].

We can see from in vivo studies that the evaluation of pharmacokinetics termed as the process of absorption, distribution, metabolism and excretion of gold nanoparticles remain significant and data from these aspects are very convictive to assess the toxicity of them. Besides, the effects on metabolism and immune systems are summarized in this review, and more researches are expected especially about the short-and longterm effects on the immune systems. However, the current data from different studies remain fragment from the experimental methods to the final conclusions, which brings much confusion and uncertainty to the investigators to plan their further study. So more attention should be given to carry out proper correlation and comparison about the biological effects of gold nanoparticles from in vivo and in vitro studies and also more detailed studies about the possible mechanisms are expected to be investigated in the future.

This work was financially supported by the National Natural Science Foundation of China (No. 31525009), and Sichuan Innovative Research Team Program for Young Scientists (No. 2016TD0004).