Silica preparation
We used fine commercial powders of Min-U-Sil5 (MS5, US Silica Company, Frederick, MD, USA) and SIO07PB (SPB, Kojundo Chemical, Saitama, Japan). To investigate the contribution of different physicochemical properties to the pulmonary toxicity of these materials, MS5 was dispersed well in purified water (endotoxin-free deionized water) by ultrasonication (42 kHz, 180 W) for 3 h. The mixture was then centrifuged at 100g for 15 min, and the smaller particles in the supernatant were collected to yield MS5_C. The particles of raw SPB powder were too large to be respirable, so we performed a similar dispersion and centrifugation procedure to yield respirable SPB_C. Additionally, grinding is an efficient method of producing smaller particles. Raw SPB powder was therefore ground in a planetary ball mill (PM100, Retsch, Haan, Germany) for 30 min at 450 rpm; SPB_G suspension was then obtained by using the dispersion and centrifugation procedures described above. Consequently, the specific surface area of SPB_G was larger than that of SPB_C, although both the primary and secondary particle sizes were comparable. The percentage crystallinity of SPB_G, as judged from XRD (Ultima IV, Rigaku Co., Tokyo, Japan) measurements, was decreased because of mechanical damage from the milling: the α-quartz particles were likely partially converted to an amorphous component (38), presumably in the surface layer. Therefore, to remove the surface amorphous component, the SPB_G particulates were dissolved by using hot NaOH aqueous solution (1 N, 60 ºC, 16 h), and the smallest particle-size sample, SPB_D, was obtained.
Characterization of the tested silica particles
Particle size
Primary particle size was calculated from 500 particle images taken by using FE-SEM (Hitachi S4800, Hitachi High-Technologies Corporation, Tokyo, Japan: see Fig. 1). Because primary particle-image recognition was difficult owing to the strong agglomeration of SPB_D, in the case of this silica the particle size was calculated from the results of specific surface area measurements assuming that the particles were simple spherical shapes (see following section). The secondary particle sizes of the silicas in water suspension were measured by means of dynamic light scattering (Zetasizer Nano ZS; Malvern Instruments Ltd., Egham, UK).
Specific surface area
Specific surface area was measured with a Gemini VII surface area analyzer (Shimadzu, Kyoto, Japan). Volumes of nitrogen gas absorbed to well-dried samples were measured. Specific surface areas were then calculated by using Brunauer–Emmett–Teller theory, taking into consideration the vapor pressure.
Zeta potential and point of zero charge
The zeta potential of the silica suspension was measured by means of laser doppler electrophoresis with the same instrument used for the particle size measurement (Zetasizer). Each 2 mg/mL suspension was prepared and the point of zero charge was measured or estimated by using HCl to titrate the pH close to 1 (see Fig. 2). The data were summarized as mean values of duplicate measurements.
Crystallinity percentage
It is well known that crystalline silica is easily deformed to amorphous silica by mechanical damage; therefore, we needed to determine the crystallinity percentages. Conventional XRD estimation methods that use amorphous halos and crystalline reflections are poor, because the halos of nanosized particles are too broad and weak to analyze. Crystallinities were therefore estimated by using internal standard samples (39). The values indicate mass-basis crystalline component percentages.
In advance, reference XRD data (a series of integrated intensities of 18 peaks) were obtained from a 1:1 powder mixture of standard reference materials of α-quartz (SRM 1878a) and α-alumina (SRM 676a) produced by the National Institute of Standards and Technology (NIST Gaithersburg, MD, USA), with respective certified purities in the crystalline phase of 93.7% and 99.02%. These reference data were then used to derive the unknown degree of crystallinity of a targeted crystalline silica sample from XRD data (the same series of peaks) collected from a 1:1 mixture of the α-quartz powder and NIST’s α-alumina (SRM 676a).
Solubility analysis
Time-course changes in the solubility of the tested silica in saline or ALF were assessed (see Fig. 3). ALF was prepared by using the method described in a previous study (30). Ten milliliters of suspension (2.0 mg/mL) and 30 mL of ALF were mixed and shaken at 200 rpm. At each time point, 2.5 mL of the fluid was sampled and filtered at 6000g for 30 min with an ultrafilter (10,000 MW; VS2032, Sartorius Stedim Biotech GmbH, Göttingen, Germany) to separate the dissolved Si ions. Ultrapure water (1 mL) was then added to the ultrafilter, and the samples were centrifuged at 6000g for 15 min; these steps were performed twice. HNO3 (0.5 mL) and HF (0.05 mL) were added to 0.1 or 0.5 mL of the filtered solutions, left for 6 h, and then diluted to appropriate concentrations with ultrapure water or 5% HNO3 solution. The Si contents of the samples were then determined by inductively coupled plasma-sector field mass spectrometry (ICP-SFMS) using a Finnigan ELEMENT XR (Thermo Fisher Scientific Inc., MA, USA) at time points of 0.5, 1, 2, 4, 8, 24, 32, 48, 168, and 336 h.
Animal experiment
F344/DuCrlCrlj rats (Charles River Laboratories Japan, Inc., Kanagawa, Japan; male, 12 weeks old) were used. The animals were housed in barrier-system animal rooms maintained at 21 to 25 °C and a relative humidity of 40% to 70%, with 10 to 15 air changes each hour and a photoperiod of 12 h of light each day.
One vehicle control group (purified water) and three dose groups were set up for each silica. The dose levels of the MS5 derivatives were set at 0.67, 2, and 6 mg/kg, whereas the SPB derivatives were dosed at 0.22, 0.67, and 2 mg/kg: 10 rats per dose were used for each material. The doses were set according to the results of preliminary studies: a single intratracheal instillation study was performed by using MS5 and SPB_C. We confirmed that evident inflammation was inducible by 3 days post-instillation, and we then set the highest doses tolerable. Details of the intratracheal instillation procedure were given in our previous study (40). Briefly, with the guidance of a laryngoscope, each dose suspension or vehicle was intratracheally instilled via the oral cavity under isoflurane anesthesia (Day 0). The rats’ clinical condition was then examined daily, and body weights were measured once or more weekly. On Days 3, 28, and 91, the animals were anesthetized and then humanely killed by being bled from the abdominal aorta. BALF and lung burden analysis were then performed in half the animals in each group (n = 5), and histopathological examination was performed in the other half (n = 5).
BALF examination
Detailed procedures of the BALF examination were also given in our previous study (40). After euthanasia of each rat, the whole lung was lavaged twice with 7 mL of saline (n = 5). Note that, because of a technical error, the BALF could not be obtained from one rat 28 days after it had received MS5_C at 2 mg/kg group (n = 4). Total cell counts (ADVIA 120 hematology analyzer, Siemens Healthcare Diagnostics, Tarrytown, NY, USA) were determined. Then smear slides were prepared and Giemsa stained, and a differential cell count of 200 cells was performed. Additionally, total protein content and lactate dehydrogenase activity were measured (Clinical Analyzer 7170, Hitachi Science Systems, Ibaraki, Japan).
Lung burden and lymph node load analyses
After the lavage, the SiO2 contents of the lungs and thoracic lymph nodes (parathymic lymph node and bilateral posterior mediastinal lymph nodes) were analyzed. The lungs and lymph nodes were homogenized in 2 mL of ultrapure water (Milli-Q Advantage A10 Ultrapure Water Purification System, Merck Millipore, Billerica, MA). HNO3 (68%, 0.5 mL), H2SO4 (98%, 0.1 mL), and H2O2 (35%, 0.1 mL) were added to 0.75 g of homogenized lung tissues and whole lymph nodes. The acidified samples were heated to 180‐200 °C for 20 min in a microwave sample preparation instrument (ETHOS 1, Milestone Srl, Milano, Italy; or Speedwave 4, Berghof, Mühlhausen, Germany). After the samples had been cooled to 40 °C, HF (38%, 0.1 mL for tissue samples) was added and the samples were then left for 6 h. The acid-treated samples, diluted to appropriate concentrations with ultrapure water or 5% HNO3 solution, were examined by ICP-SFMS (Finnigan ELEMENT XR; Thermo Fisher Scientific Inc.). Additionally, the contents of the BALF samples were measured and combined to the SiO2 contents of the lung to calculate the lung burden.
Pathological examination
The lungs, trachea, parathymic lymph nodes, posterior mediastinal lymph nodes, liver, kidneys, spleen, and brain were removed. The lungs, liver, kidneys, spleen, and brain were weighed, and the relative weights were calculated on the basis of body weight. All organs or tissues were fixed in 10% (v/v) neutral phosphate-buffered formalin solution. Paraffin-embedded specimens were prepared, sectioned, stained with hematoxylin and eosin, and examined histopathologically. The histological changes and terminology of the lung and lymph nodes underwent peer-review externally by two expert pathologists.
Statistical analysis
For body weights, organ weights, the results of the BALF examinations, lung burden and lymph node loads, StatLight software (Yukms Co., Ltd., Tokyo, Japan) was used for statistical analyses. For body weights and organ weights, Bartlett’s test for homogeneity of variance was conducted. When the variances were homogeneous at a significance level of 5%, we applied Dunnett’s test; otherwise, we applied a non-parametric Dunnett’s test. For BALF, lung burden, and lymph node loads, Bartlett’s test was conducted. When the variances were homogeneous at a significance level of 1%, including logarithmic and square-root transformations, we applied Dunnett’s test; otherwise, we applied Steel’s test. Statistical significances were judged at the 5% probability level.