Evaluation of Silver Nanoparticle Toxicity in Skin in Vivo and Keratinocytes in Vitro

1Historically, silver (Ag) compounds have been used in numerous fields to prevent microbial growth. Like many nonessential heavy metals, Ag is a natural biocide, but compared with titanium, zinc, and copper, Ag nanoparticles (Ag-nps) show the highest antimicrobial efficacy against bacteria, viruses, and other eukaryotic microorganisms. The Phoenicians coated milk bottles with Ag to inhibit bacterial growth; doctors have administered drops of Ag nitrate solutions to newborn babies to prevent neonatal conjunctivitis; and Ag sulfadiazine creams have long been considered the standard of care for the prevention of widespread bacterial growth on burn patients’ denuded skin. Both dietary supplements and homemade varieties of Ag colloids have been sold for decades as a “cure-all” for such diseases as tuberculosis, syphilis, scarlet fever, shingles, herpes, pneumonia, and arthritis. Furthermore, advances in nanotechnology have facilitated the increase of Ag-containing merchandise available to the public, making Ag the most used nanomaterial of all manufacturer-identified products in the world. Products such as room deodorizing sprays, acne creams, clothing that prevents body odor, baby wipes, and pacifiers all exploit the natural antimicrobial activity of Ag. In a study investigating the release of Ag-nps from commercially available sock fabric, showed that socks could contain up to 1,360 μg Ag/g per sock and could release as much as 1.3 μg/mL of Ag into distilled water.

2Ingestion of Ag can cause argyria, the benign condition characterized by the bluish-graying of the skin that occurs through the preferential deposition of Ag in the basal lamina of soft tissues such as the skin, liver, and spleen and blood vessels, gastrointestinal tract, liver, and kidney. Although argyria is most commonly reported clinically after excessive Ag ingestion, Ag deposition has been seen after treatment of burned skin with Ag sulfadiazine. In response to argyria, but not to Ag toxicity, the set a daily exposure limit for all forms of Ag at 0.01 mg/m3, and the established the oral reference dose at 0.005 mg/kg/day.

3Studies indicating Ag toxicity exist from as early as 1983, when showed that Ag salts intraperitoneally administered to rats can accumulate in neurons and in protoplasmic glial cells of the brain and spinal cord. In vitro cell line studies have shown decreased mitochondrial function after exposure to Ag-nps in murine neuroblastoma cells, hepatic cells, germline stem cells, human skin carcinoma cells, and human epidermal keratinocytes (HEKs) and fibroblasts. Although in vivo studies have not been performed with Ag-nps, polyvinylpyrrolidone-stabilized Ag-nps with a mean size of 25 nm were shown to penetrate into the upper layers of the epidermis in excised human skin in static diffusion cells. Other nanomaterials, such as quantum dots (QDs) and fullerenes, as well as zinc oxide, are able to penetrate into the stratum corneum; thus, examination of the ability of Ag-nps to penetrate the skin is warranted.

4Our objectives in this study were to determine the optimal viability assay for use with Ag-nps in order to assess their toxicity to skin cells, their inflammatory potential, and capacity to penetrate skin.

5To encompass the variety of manufactured Ag-nps on the market, we used eight different forms of Ag-nps with different sizes and surface conditions (Table 1). All Ag-nps used in this study were supplied by nanoComposix (San Diego, CA): commercially used unwashed and uncoated Ag-nps suspended in deionized water with diameters of 20, 50, and 80 nm (“unwashed”); washed and uncoated Ag-nps suspended in deionized water with diameters of 20, 50, and 80 nm (“washed”); and commercially used dried carbon-coated Ag-nps with diameters of 25 and 35 nm (“carbon-coated”). The sizes of each type of Ag-nps were determined by the manufacturer and confirmed in this study by dynamic light scattering (DLS) and transmission electron microscopy (TEM).

6Both the unwashed and washed Ag-nps were synthesized by ammonium hydroxide– catalyzed growth of Ag onto 5-nm gold seed particles. Concentration of the particles was achieved via tangential flow filtration. The unwashed Ag-nps solution contained approximately 5.55 mg/mL formaldehyde solvent and methanol by-product from their formation. These unwashed Ag-nps were then ultracentrifuged to obtain the solution supernatant for toxicity testing (“as synthesized” supernatant). The Ag-nps were then serially washed with 20 vol equivalents of 2 mM phosphate buffer (pH 7.5) and the 5, 10, 15, and 20 washing permeates were collected. The washing permeates contained 20–50 ppb dissolved Ag. The colloidal Ag-nps were stored at 4°C in the dark. The carbon-coated Ag-nps, synthesized by pulsed plasma reactor and coated with polyaromatic graphitic carbon, were supplied as a powder and stored at room temperature.

7Cryopreserved primary neonatal HEKs (Lonza, Walkersville, MD) were grown in Keratinocyte Growth Medium-2 (KGM-2; Lonza) in cell culture flasks (75 cm2; 1,000,000 cells) to approximately 80% confluency in a 37°C humidified 5% CO2 incubator. The cells were passed into clear or black 96-well microplates (12,500 cells/well; 200 μL) in which the peripheral wells contained only KGM-2 to prevent the evaporation of treatment medium. Between 18 and 24 hr later, after reaching approximately 80% confluency, the HEKs were exposed to either KGM-2 (control) or serial dilutions of each Ag-nps type for the following experiments.

8A nanoparticle (no-cell) control (nanoparticles in collagen-coated wells with no HEKs) and a nanoparticle/cell control (nanoparticles exposed to HEK-metabolized assay dye) were run in parallel with each viability assay as described by to assess the interactions between the viability assays and the Ag-nps [see Supplemental Material available online (doi:10.1289/ehp.0901398.S1 via http://dx.doi.org/)].

9We conducted an initial dose–response study to assess the concentrations of Ag-nps that could affect HEK after 24-hr exposure. Most of the colloidal Ag-nps tested were supplied in both low volume and concentration, which limited the highest HEK dosing concentration to 1.7 μg/mL. Combined with KGM-2 medium, a 1.7 μg/mL solution of the Ag-nps was serially diluted (1:5) to provide concentrations ranging from 1.7 to 0.000544 μg/mL. The effect of the Ag-nps on HEK viability was assessed by three different toxicity assays: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich, St. Louis, MO), alamarBlue (aB; Molecular Probes, Invitrogen, Eugene, OR), and CellTiter 96 AQueous One (96AQ; Promega, Madison, WI) [for details, see Supplemental Material (doi:10.1289/ehp.0901398.S1 via http://dx.doi.org/)].

10We tested dosing concentrations of the washed and carbon-coated Ag-nps up to 42.5 μg/mL but saw no response. Additionally, to differentiate the potential cytotoxicity between the particles and the contaminants present in the colloidal solution, HEKs were treated with the “as synthesized” supernatant and the 5, 10, 15, and 20 washing permeate for 24 hr at concentrations ranging from 1.7 μg/mL to 0.068 μg/mL.

11For the concentrations of Ag-nps that showed toxicity, we conducted cytokine analysis to determine their proinflammatory potential by assessing the release of interleukin (IL)-8, IL-6, tumor necrosis factor-α (TNF-α), IL-10, and IL-1β; the procedure is described in the Supplemental Material (doi:10.1289/ehp.0901398.S1).

12We compared the effects of unwashed Ag-nps with those of washed Ag-nps in vivo. Assuming the two smallest Ag-nps could penetrate the skin, the comparison was limited to the 20- and 50-nm washed and unwashed samples. Pigs were dosed on the back skin with Ag-nps solutions ranging from 34.0 to 0.34 μg/mL [for details, see Supplemental Material (doi:10.1289/ehp.0901398.S1)]. Skin was evaluated for erythema and edema according to the Draize system: for erythema: 0, no change; 1, very slight change; 2, pale red in defined area; 3, definite red in well-defined area; 4, crimson red; and for edema: 0, no change; 1, very slight change; 2, slight change with edges barely defined; 3, moderate change, with area raised 1 mm; 4, severe change, with area raised > 1 mm and extending beyond the exposure area. All animals were treated humanely and with regard for alleviation of suffering.

13To assess morphologic alterations during the in vivo study, we harvested tissue samples after the pigs were euthanized and processed them routinely for light microscopy [see Supplemental Material (doi:10.1289/ehp.0901398.S1)]. Approximately 1-cm sections were evaluated for intercellular and intracellular epidermal edema, dermal edema, and inflammation using the following scoring system: 0, no change; slight, inflammation on less than half the sample; moderate, inflammation on half the sample; severe, inflammation on more than half the sample.

14Particle size analysis was conducted with both DLS and TEM to confirm the manufacturer-identified diameters and surface characterization [for details, see Supplemental Material (doi:10.1289/ehp.0901398.S1)]. To localize Ag-nps uptake in vitro, HEKs were grown to approximately 70% confluency in cell culture flasks (25 cm2) and treated for 24 hr with each Ag particle at 1.7 μg/mL in KGM-2. The cells and skin samples were processed routinely for TEM (for details, see Supplemental Material). We viewed all TEM samples on an FEI/Philips EM 208S TEM (FEI, Hillsboro, OR), operating at an accelerating voltage of 80 kV. Additionally, unstained samples were analyzed by X-ray microanalysis [energy dispersive X-ray spectroscopy (EDS)] with a Hitachi HF2000 FE TEM (Hitachi High Technologies America Inc., Lexington, KY) equipped with an Oxford Instruments INCA EDS (Oxford Instruments, Oxfordshire, UK).

15We calculated the mean values for HEK percent viability (normalized by viability) and cytokine concentration for each treatment and determined the significant differences (p < 0.05) using the PROC GLM procedure (SAS, version 9.1 for Windows; SAS Institute Inc., Cary, NC). When significant differences were found, we performed multiple comparisons using the Tukey’s studentized range highest significant difference test, with p < 0.05 as the level of significance. Dunnett’s t-test was performed to determine the significance at p < 0.5 of differences between control and treatment groups. Data are expressed as the mean ± SEM of two plates (n = 6/plate).

16Table 2 summarizes the control studies for the three viability assays, listing the absorbance or fluorescence value for each Ag-nps type that shows a significant interaction. For MTT, at 1.7 μg/mL both the 25-nm (Figure 1A) and 35-nm (Figure 1B) carbon-coated Ag-nps nanoparticle/cell controls show a statistically significant increase in absorbance after exposure of Ag-nps to cell-reacted assay dye. The nanoparticle (no cell) controls showed a statistically significant increase in absorbance at 1.7 μg/mL with the 20-nm unwashed Ag-nps for 96AQ and MTT but not for aB (Figure 1C).

17All other nanoparticle control data are available in the Supplemental Material (doi:10.1289/ehp.0901398.S1). We saw no significant changes in the fluorescence values for aB for the 20-, 50-, or 80-nm unwashed Ag-nps (see Supplemental Material, Figure 1a). The unwashed 50- and 80-nm Ag-nps caused a significant increase in absorbance values for MTT and 96AQ but not for aB (see Supplemental Material, Figure 1b,c). The fluorescence value for aB increased significantly at 0.034 μg/mL for both 20- and 50-nm washed Ag-nps and at 1.7 for 80-nm washed Ag-nps (see Supplemental Material, Figure 2a). The absorbance value for 96AQ changed significantly at 1.7 μg/mL for each of the 20-, 50-, and 80-nm washed Ag-nps (see Supplemental Material, Figure 2b). The 80-nm washed Ag-nps caused a significant increase in absorbance value for MTT at 1.7 μg/mL (see Supplemental Material, Figure 2c). The fluorescence value for aB increased significantly at 1.7 μg/mL for 35-nm carbon-coated Ag-nps (see Supplemental Material, Figure 3a). The absorbance value increased significantly for 96AQ at 1.7 μg/mL for 25-nm carbon-coated Ag-nps and at 0.034 μg/mL for 35-nm carbon-coated Ag-nps (see Supplemental Material, Figure 3b). We observed a significant increase in the absorbance value for MTT at 1.7 μg/mL for 25- and 35-nm carbon-coated Ag-nps (see Supplemental Material, Figure 3c).

18Generally, for concentrations ranging from 0.000544 to 1.7 μg/mL and exposures of 24 hr, treatment with unwashed Ag-nps resulted in a dose-dependent decrease in viability with all three assays, whereas for the washed and carbon-coated Ag-nps there was no decrease in viability for any assay.

19As shown in Figure 1D, the 20- and 50-nm unwashed Ag-nps caused a significant decrease in viability at 0.34 μg/mL for both aB and 96AQ and at 1.7 μg/mL for MTT. The 80-nm unwashed Ag-nps caused a significant decrease in viability at 0.34 μg/mL for all three assays. The “as synthesized” supernatant showed a significant decrease in viability at 0.34 μg/mL for the MTT and aB assays and at 1.7 μg/mL for 96AQ. Toxicity was not present for any of the supernatants obtained from the 5, 10, 15, or 20 washing permeates. HEKs exposed for 24 hr to washed 20-, 50-, or 80-nm Ag-nps or the carbon-coated Ag-nps ranging in concentration from 0.000544 to 1.7 μg/mL showed no significant decrease in viability with any assay [see Supplemental Material, Figures 4–6 (doi:10.1289/ehp.0901398.S1)].

20We observed a significant increase above the limit of detection in IL-1β (Figure 2A), IL-6 (Figure 2B), IL-8 (Figure 2C), and TNF-α (Figure 2D) from HEKs exposed to 0.34 μg/mL unwashed 20-, 50-, and 80-nm Ag-nps for 24 hr. The lowest limit of detection for IL-1β was 0.8 pg/mL; IL-6, 1.1 pg/mL; IL-8, 0.5 pg/mL; and TNF-α, 3.0 pg/mL. The values for IL-10 were less than the detectable limit of the assay (0.9 pg/mL).

21We noted no gross erythema or edema based on the Draize system in any of the treated sites during the entire 14-day in vivo study. Treated sites were gray in appearance, representing residual Ag-nps on the surface of the skin.

22Morphologic observations of untreated porcine skin exhibited normal epidermis and dermis (Figure 3A). Skin dosed daily with Ag-nps for 14 days exhibited a concentration-dependent response, regardless of particle size or if washed or unwashed. Skin treated with the lowest dosing concentration of 0.34 μg/mL 20-nm washed Ag-nps typically showed a slight intracellular and intercellular epidermal edema (Figure 3B); with 3.4 μg/mL of 20-nm washed Ag-nps there was moderate focal intracellular and intercellular epidermal edema and focal epidermal and dermal inflammation (Figure 3C). Skin treated with the highest concentration of 20-nm washed Ag-nps (34 μg/mL) showed severe intracellular and intercellular epidermal edema with severe focal dermal inflammation (spongiosis), epidermal hyperplasia, and parakeratosis. Also, the extension of the rete pegs increased into the superficial papillary layer of the dermis (Figure 3D).

23Results for the 20-nm unwashed Ag-nps [see Supplemental Material, Figure 7 (doi:10.1289/ehp.0901398.S1)] were similar to those for the 20-nm washed Ag-nps. In skin treated with 0.34 μg/mL of 20-nm unwashed Ag-nps, we observed slight intracellular epidermal edema; with 3.4 μg/mL, moderate intracellular and intercellular epidermal edema; and with 34 μg/mL, severe intracellular and intercellular epidermal edema with focal areas of intraepidermal infiltrates and superficial papillary dermal inflammation.

24Ultrastructural observations of the 20-nm washed and 25-nm carbon- coated Ag-nps are shown in Figure 4A and B [for all other Ag-nps, see Supplemental Material, Figure 8 (doi:10.1289/ehp.0901398.S1)]. The HEK controls appeared normal, with prominent nucleus, nucleolus, and mitochondria (Figure 4C). In HEKs dosed with Ag-nps of all sizes and surface conditions, Ag-nps were localized within membrane- bound cytoplasmic vacuoles; Figure 4D shows a representative image of internalized Ag-nps (for all other Ag-nps electron micrographs, see Supplemental Material, Figure 9). EDS analysis of HEKs dosed with the 20-nm washed Ag-nps confirmed the presence of Ag in the vacuoles; copper from the grid and gold from the particle seed were present (see Supplemental Material, Figure 10).

25The control skin consisted of a normal compact stratum corneum with approximately 20–30 layers attached by desmosomes (Figure 4E). For skin dosed daily for 14 days, all Ag-nps localization was within or on top of the stratum corneum. TEM images show the presence of Ag-nps within the superficial layers of the stratum corneum of skin dosed with 34 μg/mL 50-nm washed Ag-nps (Figure 4F), and on top of the stratum corneum of skin dosed with 20-nm washed Ag-nps [see Supplemental Material, Figure 9H (doi:10.1289/ehp.0901398.S1)]. EDS analysis conducted in these areas detected Ag, osmium from the postfixation process, and copper from the grid (see Supplemental Material, Figure 10).

26Ag-nps have been integrated into hundreds of products that affect the daily lives of millions of people in many countries. Their main use is for disinfection in wound care and in products such as odor-reducing clothing, acne creams, and face masks. Most of these products come into direct contact with skin, the largest organ of the human body, and could serve as a potential route for nanoparticle penetration. Therefore, the relationship of Ag-nps in skin needs to be investigated with particular focus on their irritation potential, toxicity, and penetration into skin and skin cells. In this study we evaluated the cytotoxic potential of Ag-nps of varying sizes and surface conditions in HEK cells, their penetrating capacity into porcine skin after topical repetitive daily dosing for 2 weeks, and the localization of the Ag-nps within HEKs and porcine skin.

27The use of several viability assays is important to determine the optimal assay to assess Ag-nps toxicity; therefore, mortality of HEKs after Ag-nps exposure was evaluated with three different assays that use colorimetric or fluorescent dyes as markers to determine cell viability by assessing cell metabolism.

28Nanomaterials, such as single-walled carbon nanotubes, carbon black, fullerenes, and QDs, are capable of interfering with dye and dye products in viability assays through the adsorption of cell medium constituents and cytokines. We assessed the potential interactions between assays and Ag-nps using nanoparticle and nanoparticle/cell controls, which showed an increase in absorbance and fluorescence values at the highest concentration. The increase in absorbance and fluorescence values could cause the toxicity of the Ag-nps in HEKs to be underestimated. Additionally, the nanoparticle/cell control showed that both the 25- and 35-nm carbon-coated Ag-nps interfered with the MTT assay at the 1.7 μg/mL concentration because of the increase in absorbance values after incubation of the reduced formazan product with Ag-nps (Figure 1A,B). Overall, all assays were affected by Ag-nps; based on its fluorescence values, aB may be the best viability assay to use when conducting experiments with Ag-nps.

29MTT, aB, and 96AQ viability assays did not show toxicity for the 25- and 35-nm carbon-coated Ag-nps or for the 20-, 50-, or 80-nm washed Ag-nps. All three assays also showed that the 20-, 50-, and 80-nm unwashed Ag-nps contributed to a decrease in HEK viability 24 hr after exposure to the 0.34–1.7 μg/mL concentrations, but there was no size-dependent decrease in viability. However, the difference in toxicity between the unwashed and washed Ag-nps is inferred to be due to the presence of contaminants in the unwashed solution such as formaldehyde, which has shown to have cytotoxic effects on cell culture. In the present study, these residual contaminants were removed by the fifth washing step, as indicated by the lack of cell death after exposure to the 5, 10, 15, and 20 washing supernatants. We also found that Ag-nps of different sizes, surface conditions, and synthesis methods are all internalized into membrane-bound vacuoles in HEKs, without a decrease in viability, after 24 hr.

30In vitro cell line studies have shown that 25 μg/mL of 25-nm Ag-nps in murine neuroblastoma cells decreases mitochondrial function and causes the production of reactive oxygen species that could potentially contribute to neurodegenerative diseases. A significant decrease in mitochondrial function has been shown in hepatic cells after single exposures to 15- and 100-nm Ag-nps at concentrations ranging from 5 to 50 μg/mL, in germline stem cells exposed to 10 μg/mL of 15-nm Ag-nps, and in HEKs and fibroblasts after exposure to approximately 15 μg/mL Ag-nps extracted from commercially available Ag-based wound dressings with Ag content ranging from 13 to 934 μg/cm2. Interactions between Ag-nps ranging in size from 7 to 20 nm and human skin carcinoma cells showed a decrease in mitochondrial function and the onset of apoptosis at concentrations of 0.78 μg/mL and 1.56 μg/mL, respectively. The toxic concentrations of the 20-, 50-, and 80-nm unwashed Ag-nps (0.34–1.7 μg/mL) are slightly more sensitive in HEKs compared with in vitro toxicity studies conducted by others in different cell lines, although it is important to consider such factors as agglomeration, surface conditions, size, cell lines, and interactions with the assay dye products when comparing across studies.

31Keratinocytes produce proinflammatory cytokines, such as IL-8, IL-6, TNF-α, and IL-1β, that serve as mediators for inflammatory and immunologic reactions in skin exposed to irritants. Although different toxicants may elicit different responses in HEKs, studies in our laboratory have shown cytokine release by HEKs in response to jet fuel exposure, multiwalled carbon nanotubes, 6-aminohexanoic acid-functionalized single-walled carbon nanotubes, fullerenes, and QDs. The inflammatory potential of Ag-nps was confirmed by the increases in IL-1β, IL-6, IL-8, and TNF-α detected in the media from HEK cell cultures exposed to 0.34 μg/mL of each of the unwashed Ag-nps.

32Nanomaterials are also capable of being internalized into cells and penetrating through skin; QD-621 have the ability to penetrate into the intercellular lipid layers of the stratum corneum of porcine skin; QD-565 and QD-655, with diverse physiochemical properties, have been shown to penetrate into the dermis of abraded skin; and derivatized fullerenes are localized within the intercellular space of the stratum granulosum layer of flexed excised porcine skin. Topical application of 26–30 nm zinc oxide in a sunscreen formulation on in vitro human skin localized nanoparticles in the upper stratum corneum with minimal penetration, and microfine zinc oxide, with a mean size of 80 nm, and agglomerates of titanium dioxide < 160 nm did not penetrate the porcine stratum corneum layer of in vitro static diffusion cells.

33Porcine skin is an excellent model for studying penetration of human skin because its thickness and absorption rates are comparable to those of human skin. In the present study, we were surprised that after 14 consecutive days of topical dosing, the Ag-nps did not cause any macroscopic irritation, although the gray appearance of the skin due to the deposition of Ag-nps may have masked any subtle signs of erythema. When viewed microscopically, focal inflammation and edema increased with an increase in Ag-nps concentration. The highest concentration consistently caused epidermal hyperplasia with elongated extension of rete pegs down into the dermis, which is typical of a chronic irritation reaction as reported with exposure to jet fuels. TEM depicted the localization of Ag-nps only in the superficial layers of the stratum corneum, which was similar to results found in a static cell diffusion study, and suggests that ionic flux into the epidermis could attribute to focal inflammation. Many Ag-nps that were not bound to the skin were washed away during both the light and electron microscopy processing techniques, yet their location is confirmed with other nanoparticles that were not shown to penetrate into the deeper epidermis.

34This study indicates that toxicity of Ag-nps in HEKs is influenced by the residual contaminants in the Ag-nps solutions, and that the Ag-nps themselves may not be responsible for an increase in cell mortality. Complete characterization of not only the nanoparticles but also the vehicle is important in order to distinguish between Ag-nps and contaminant toxicity. Additionally, this study shows that 20-, 50-, and 80-nm washed and unwashed Ag-nps, as well as 25- and 35-nm carbon-coated Ag-nps, interfered and/or reacted with MTT, 96AQ, and aB viability assays and that aB may be the best viability assay because of its lower interference with these Ag-nps. Because Ag-nps of several types have been shown to enter cells and remain on the skin, the possibility of Ag-nps entry into the body through damaged or abraded skin is important to consider, particularly because many Ag-containing products are specialized for wound care. With the ability for Ag-nps to enter HEKs, their degradation within the cell may create reactive oxygen species that would be damaging to cell machinery and DNA. We observed focal inflammation, specifically intracellular and intercellular epidermal edema, after 14 days of topical application of Ag-nps to skin. Studies that use longer exposures (several weeks) and Ag-nps in different vehicles should be carried out, possibly also investigating effects on compromised skin. Overall, the present study provides knowledge onAg-nps toxicity and penetration in vitro and in vivo over 14 days and provides a basis for occupational and risk assessment.