Study materials and preparation of animal feed
A commercially available food additive, hydrophilic pyrogenic synthetic amorphous silica (SAS) with a primary particle size of 7 nm, a specific surface area of 380 m2/g, and a purity of 99.8% was used, kindly donated by Evonik Degussa GmbH (Frankfurt, Germany). In addition, Joint Research Centre (JRC, Ispra, Italy) Nanomaterials Repository; hydrophilic pyrogenic silica (NM-202) was used, which was kindly donated by the JRC of the European Commission. NM-202 has a specific surface area of 200 m2/g, a purity of 99.9%, and a primary particle size between 10 and 25 nm.
SAS or NM-202 was mixed with standard feed and chocolate milk was added to increase palatability, which was assessed in a pilot experiment. SAS or NM-202 was mixed by hand-stirring to a thick paste with chocolate milk (Chocomel, Nutricia, The Netherlands) and ground standard diet pellets (RMH-B, ABDiets, The Netherlands) in a ratio of 1:8:1 respectively, by weight. Total intended silica content of the mixtures (both SAS and NM-202) was 99 mg/g and was fed in different amounts to rats to achieve the desired daily dosage (Table 1; Additional file 1: Table S1). Higher dosed animals were offered more of the feed mixture than lower dosed animals. All animals of each group daily consumed the complete amount of food mixture that was offered within the two hour exposure time frame. For control groups, feed without SAS or NM-202 was prepared, containing only chocolate milk and ground standard diet pellets. To adequately compensate for the amount of chocolate milk, an average amount of chocolate milk as offered to the treated animals was chosen. This resulted in a ratio of 3:2 of chocolate milk and ground pellets respectively, by weight. Thus, animals in the different treatment groups received the following amounts of chocolate milk through feeding of the feed mixtures;. 0.8, 8.1 and 20 g/kg/ bw/day for the SAS low, medium, and high dose groups respectively, 0.8, 4.1 and 8.1 g/kg bw/day for the NM-202 low, medium, and high dose groups respectively, and 6.1 g/kg bw/day for the control animals. The feed mixtures were prepared freshly three times a week.
In vivo experimental design
Six-week-old male specific pathogen free Sprague–Dawley rats were purchased from Harlan (Horst, The Netherlands). Animals were individually housed in polycarbonate cages with cage enrichment and were allowed to acclimatize for three weeks before the start of the experiment. Room temperature was ~20°C with a relative humidity of ~55%. Individual housing was necessary, because rats were individually offered prepared food with SAS, NM-202 or vehicle. A reversed 12-h light/dark cycle was used to feed the rats in their active period and feed and water was given ad libitum, except for a two hour fasting period before the prepared feed was offered to the animals, and the following exposure period. Rats were allowed to consume all prepared feed, immediately thereafter the animals were being offered standard feed pellets again. During the entire study, rats ate all silica containing food mixtures or vehicle mixtures during the exposure period of two hours. The study was performed according to the national guidelines for the care and use of laboratory animals after approval of the animal welfare committee of Wageningen University.
At the start of the experiment the average body weight of the 9 weeks old animals was ~280 g and rats were randomly divided into 10 groups (n = 5). Seven groups of rats were fed SAS or NM-202 in different dosages or vehicle for 28 days, in addition the highest dosed groups of SAS and NM-202 and a control group were fed for 84 days. The groups for 28-day exposure were: 1) SAS; 100 mg/kg bw/day, 2) SAS; 1000 mg/kg bw/day, 3) SAS; 2500 mg/kg bw/day, 4) NM-202; 100 mg/kg bw/day; 5) NM-202; 500 mg/kg bw/day; 6) NM-202; 1000 mg/kg bw/day and 7) control. For the 84-day exposure, the groups were divided into: 8) SAS; 2500 mg/kg bw/day, 9) NM-202; 1000 mg/kg/day and 10) control. Dosages were chosen around the previously observed LOAEL of 1500 mg/kg bw/day for SAS [4]. The medium and high doses of NM-202 were chosen to be lower than those of SAS, because the material characterization showed a higher fraction of silica in the nano-size range in the feed matrix for NM-202. All rats were weighed daily. One day after the last exposure of the 28-day exposure groups (i.e. the first 7 groups), the animals were euthanized by CO2/O2 inhalation and the following organs were excised aseptically, weighed and placed on ice: liver, kidneys, spleen, brain, testis and the MLNs. Parts of the liver and the epithelium of the jejunum were also stored in liquid nitrogen, and parts of the jejunum, liver, kidney and spleen were fixed in 10% formalin. Furthermore, blood was collected on heparin and stored on ice, as well as the stomach, small (duodenum, jejunum, ileum) and large intestinal contents. One day after the last exposure of the 84-day exposure groups (i.e. group 8–10), all animals were euthanized and organs collected according to the same protocol as applied for the 28-day exposure groups.
Total silicon content was determined with inductively coupled plasma mass spectroscopy (ICP-MS) in liver, kidney, spleen, brain and testis. Furthermore, hydrodynamic chromatography (HDC) ICP-MS was applied to detect silica particles in the nano-size range (i.e. with a size of 5 – 200 nm) in gastrointestinal contents. Systemic toxicity was monitored by analysis of biochemical markers in serum and by histopathological analysis of jejunum, liver, kidney, and spleen. Immunotoxicity was evaluated by measuring antibody levels in blood, analysis of the proliferation of T- and B-cells isolated from the spleen end mesenteric lymph nodes (MLN) in response to lipopolysaccharide (LPS) or concanavalin A (Con A), by evaluating cytokine levels in culture media from these proliferating T- and B-cells, and by measuring the activity of natural killer (NK)-cells isolated from the spleen.
Material characterization
Both SAS and NM-202 were characterized in aqueous suspensions by SEM. SAS and NM-202 were suspended in LC/MS grade water (Biosolve, Valkenswaard, The Netherlands), containing 0.05% BSA as a stabilizing agent, to a concentration of 10 mg/ml. Suspensions were vortexed for 1 min at full speed, followed by sonication at 20°C at 100% output (4 W specific ultrasound energy (240 J/m3), using a Branson 5510 water bath sonicator (Emerson, USA) for 30 min. Next, the suspensions were further diluted to a final concentration of 10 μg/ml in LC/MS grade water (Biosolve) and sonicated again at 20°C at 100% output in a water bath sonicator (Emerson) for 30 min. Furthermore, SAS and NM-202 were characterized in the feed matrix prepared as described earlier and diluted 100 times. For SEM measurements, droplets of the suspensions were put on a nickel coated Nuclepore track-etched polycarbonate membranes and analyzed with high-resolution field emission gun scanning electron microscopy (FEG-SEM) on a Tescan MIRA LMH FEG-SEM operated at 15 kV in combination with a Bruker EDX spectrometer with a XFlash 4010 detector with an active area of 10 mm2 and super light element window (SLEW), which allows X-ray detection of elements higher than borium (Z > 5). The spectral resolution of the detector is 123 eV (Mn (10kcps) ave FWHM). The SEM was equipped with a Scandium SIS software package (Olympus Soft Imaging Solutions GmbH, Germany) for automated particle analysis. With this system the filter area is automatically inspected on a field-by-field basis. In each field of view particles are recognized using a pre-selected grayscale video threshold (detection threshold level) to discriminate between a particle and the filter background. The analyses were conducted using the secondary electron (SE) mode. The particle size distribution is based on the projected area equivalent diameter (dpa). Magnifications of 25.000X (image area: 6 × 8 μm) and 75.000X (image area: 2 × 2.7 μm) were chosen in order to cover the full size range from 25–400 nm. Per size bin (25–40, 40–65, 65–100, 100–160, 160–250, 250–400 nm) a minimum of 50 particles was measured; in total more than 1000 particles were measured. EDX analysis of the material surface included a correction for the background signal. This was performed by subtraction of the background signal, using nickel as a reference element (which is present as a thin layer on the filter).
IR spectra of both materials were acquired on a Bruker Tensor 27 FTIR spectrometer equipped with a single reflection Platinum ATR accessory, at a resolution of 2 cm-1.
XPS characterization of NM-202 was reported previously [23]. SAS characterization was performed at the same institute, using the same protocol and equipment. The measurements (consisting of four technical replicates) were performed using an AXIS ULTRA Spectrometer (KRATOS Analytical, UK) and Vision2 software (Kratos Analytical, UK) was used for data processing. The XPS analysis provides information on the surface composition of the analyzed material (down to a depth of 10 nm) with a detection limit of ~0.1% of the atoms and an estimated 10% accuracy in the measurement of elemental compositions. More detailed information regarding the measurement is given in Additional file 1.
In vitro digestion of feed samples
The used in vitro digestion model has previously been described [9, 24–26]. Briefly, the model consists of three phases; the saliva, gastric, and intestinal phase. The dissolution and gelating behavior of SAS and NM-202 in the intestinal environment was studied after the materials passed all three phases. All artificial juices for the digestion experiments were prepared on the day before the actual digestions. The pH values of the juices were checked and, if necessary, adjusted to the appropriate interval with NaOH (1M) or HCl (37% w/w). The constituents and concentrations of the various synthetic juices are as shown in Additional file 1: Table S11. Before the start of the digestions, all digestive juices were heated to 37 ± 2°C and incubations are carried out in a head-over-head rotator at 37 ± 2°C. Experiments for the rheological measurements were performed in duplo, and six replicate samples were used for the dissolution behavior experiments.
For the rheological measurements the digestion started by introducing 2 mL of artificial saliva to the SAS/NM-202/control feed mixture. The feed mixture consisted of 8 g of chocolate milk, 1.0 g of powdered standard diet, and 0–1.0 g of SiO2, dependent on the desired final concentration in the intestinal phase (i.e. 0, 10, 50, 75 mg/mL). This mixture was rotated head-over-head for 5 min at 55 rpm at 37 ± 2°C. Subsequently, 4 mL of gastric juice was added, the pH adjusted to pH 2.0 ± 0.5, and the mixture was rotated head-over-head for 2 h at 37 ± 2°C. Finally, 4 mL of duodenal juice, 6 mL of bile juice, and 0.7 mL of NaHCO3 solution were added. The pH was adjusted to pH 6.5 ± 0.5, and the mixture was again rotated head-over-head for 2 h at 37 ± 2°C. Subsequently, the samples were used for rheological measurements. Additionally, a subsample of the suspensions was taken for SEM-EDX analysis to characterize the material after in vitro digestion.
For the dissolution measurements, SAS and NM-202 suspensions of 1 mg SiO2/mL were prepared in Milli-Q with 0.05% BSA. Subsequently, 630–2115 μL of these dilutions, depending on the desired final concentration in the intestinal phase (i.e. 50, 75, 150 μg/mL) were added to the first phase of the digestion model (i.e. 2 mL of artificial saliva). The digestion procedure was performed as described above. At the end of the digestion experiment, 4 mL of all samples were ultrafiltered by centrifugation through a cellulose filter with a nominal cutoff value of 3 kDa (Ultra-4, Amicon). The total silica content in the unfiltered and filtered suspensions were measured by ICP-MS.
Rheological measurements
The visco-elastic behavior of the SAS, NM-202, and control feed mixtures following in vitro digestion was analyzed by rheological measurements. Oscillatory measurements were performed using a Physica MCR 301 (Anton Paar, Austria) stress controlled rheometer with a concentric double gap cylinder geometry (DG 26.7) to determine the storage modulus (G’) and loss modulus (G”) of the samples. The samples were subjected to a strain sweep with strains ranging from 1-500%, at a frequency of 1 Hz.
Determination of silica (in the nano-size range) by (HDC) ICP-MS
The size and concentration of silica in the nano-size range (i.e. between 5–200 nm) was determined by HDC-ICP-MS in the feed mixtures and in the stomach, small and large intestinal contents of the rats after 28 days of exposure. Samples were homogenized by mixing and a subsample was collected for analysis. The subsample was sonicated in LC/MS grade water to prepare an aqueous suspension, which was filtered through a 5 μm filter (Acrodisc, Pall Lifer Sciences, USA) before HDC-ICP-MS analysis.
The HDC system was a Thermo Scientific Spectra system P-4000 liquid chromatograph (Waltham, MA, USA) equipped with a PL-PSDA HDC cartridge, type 1, length 800 mm, diameter 7.5 mm, packed with non-coated, non-porous silica spheres (Agilent Technologies, Wokingham, UK). The eluent was an aqueous 10 mM solution of sodium n-dodecyl sulphate (SDS) with a flow rate of 1.0 ml/min. Sample injection volumes were 50 μL. The ICP-MS was a Thermo X Series 2 (Waltham, MA, USA), equipped with an autosampler, a Babington nebulizer and operated at an RF power of 1400 W. Data acquisition was performed in the selected ion monitoring mode monitoring m/z ratios of 28 and 29 that are characteristic for silicon. Polyatomic interference at these ion masses is unavoidable and resulted in a high, but relatively stable, background signal due to N2. Acquiring data in the helium collision mode did not improve the signal/noise ratio and was not applied since it resulted in an overall lower sensitivity. The Si signal of peaks in the chromatograms was isolated by subtracting the background signal of the baseline in the same chromatogram. Aqueous suspensions of silica nanoparticles with sizes ranging from 32 to 500 nm (Microsil microspheres, Bangs Laboratories, USA) were used to calibrate the size separation of the HDC column, while a standard of 32 nm silica nanoparticles was used for quantitation and checking system performance. Data are presented as a weight percentage of silica in the nano-size range relative to the total amount of silica in the same samples.
The total silicon content was determined in the standard rat feed pellets, the prepared feed, with or without SAS or NM-202, in the drinking water, in the stomach and gut content and in the tissues. For this, 1–2 grams of sample was added to a perfluoroalkoxy digestion vial, followed by the addition of 6 ml 70% nitric acid and 1 ml of 40% hydrogen fluoride. The samples were digested in a microwave system for 45 min at ~250°C, 70 bar. Following digestion and cooling to room temperature, MilliQ water was added to a total volume of 150 ml. This solution was shortly shaken by hand and two times further diluted to a total volume of 300 ml. Finally, the extracts were analyzed by ICP-MS using the same system and settings as described above. The use of glass equipment during sample preparation and (HDC) ICP-MS analysis was avoided to prevent Si contamination. Also internal control samples were evaluated to assure the absence of Si contamination. Si measurement data were converted to SiO2 and presented as SiO2 throughout the text. The limit of detection in tissues, the stomach, small and large intestinal contents was set at 35 mg Si/kg (75 mg SiO2/kg), with a measurement error of ± 15 mg Si/kg for concentrations between 35 and 100 mg Si/kg and 20% at concentrations >100 mg/kg. In water and intestinal juices (for the in vitro digestion experiments) the limit of detection was set at 5 μg/ml based measurements in blanks.
The performance of the HDC-ICP-MS analysis in terms of response and retention time stability was determined as described earlier [9] as the standard deviation in the average response and retention time of the quantitation standard in each analysis series. The reproducibility standard deviation of the response of the quantitation standard was 20%, and that of the retention time window <2%. Compared to usual contaminant analysis the reproducibility standard deviation of the response is relatively high for two reasons. First, the ICP-MS signal for Si suffers from a high background due to the presence of N2 and CO, which has to be subtracted to isolate the true Si signal of the analytes. Secondly, since a “size range” has to be determined no clear narrow peaks as in regular gas or liquid chromatography are observed, but broader peaks depending on the size distribution of the particles.
The recovery of nano-sized silica (by HDC-ICP-MS) is determined by spiking blank samples with nano-sized silica and analyzing these control samples with the actual samples. The average recovery of the added nano-sized silica (32 nm colloidal silica material, stabilized at pH 8.6, was obtained from the Institute for Reference Materials and Measurements (IRMM), European Commission Joint Research Centre, Geel, Belgium) was 83 ± 21%. The SANCO/10684/2009 document concerning method validation and quality control procedures for pesticide residue analysis in food and feed states that the recovery should be between 70 and 120% and that the within lab reproducibility of a method should not exceed 20% [45]. This means that the recovery is acceptable while the reproducibility is on the limit. This however, was considered acceptable since the SANCO document is applicable to well-established methods for well-defined analytes such as pesticides and not to less well-defined particulate materials as in this study.
Blood biochemistry determination
In plasma, taken after 28-days of exposure, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and alkaline phosphatase (ALP) activity, creatinin, total protein and urea, were determined using standard kits (Beckman-Coulter, Woerden, the Netherlands). In plasma, taken after 84-days of exposure, the same biochemical markers as for the 28-day exposure were determined, as well as lactate dehydrogenase (LDH) activity, uric acid, Zn, Fe, HDL-, LDL- and total cholesterol, triglycerides, and glucose levels using standard kits (Beckman-Coulter). ALT, ASP, ALP, creatinin, total protein, urea, LDH, uric acid, Zn, Fe, HDL, LDL and total cholesterol, triglycerides and glucose were determined with a clinical autoanalyzer (LX20-Pro, Beckman-Coulter) using standard kits which have been developed for this system.
Immunotoxicity on mesenteric lymph nodes and spleen
The excised mesenteric lymph nodes (MLN) and approximately one third of the spleen of each rat were stored separately in Iscove’s modified Dulbecco’s medium (IMDM; Gibco, Grand Island, NY) on ice. The organs were pressed gently through a cell strainer (70-μm nylon; Falcon, Becton-Dickinson Labware, Franklin Lakes, NJ) and the cells were suspended in 25 ml IMDM, supplemented with 10% fetal calf serum (FCS; PAA, Linz, Austria), 100 IU/ml penicillin, and 100 μg/ml streptomycin, referred to as complete Iscove’s medium. Next, the cell suspensions were centrifuged at 300g for 10 min (4°C) and the pellets were resuspended in 20 ml complete Iscove’s medium. Finally, cells were counted using a Coulter Counter (Coulter Electronics, Luton, UK). For the lymphocyte transformation test, 4*105 cells, isolated from the MLN or spleen, were cultured in six-fold in 150 μl complete Iscove’s medium in U-bottom 96-well microtiter plates (Greiner, Frickenhausen, Germany). In three of the six wells of the first Plate 15 μg/ml LPS (final concentration) was present (B-cell proliferation). In parallel, in three of the six wells of the second Plate 5 μg/ml Con A (final concentration) was present (T-cell proliferation). After 24 h (LPS) or 48 h (Con A) incubation in a humidified atmosphere containing 5% CO2 at 37°C, 37 kBq [methyl-3H]thymidine ([3H]TdR; Amersham, Little Chalfont, UK) was added to the wells. Cells were incubated for another 24 h followed by harvesting of the cells onto glass-fibre filters (LKB-Wallac, Espoo, Finland) using a multiple cell culture harvester (LKB-Wallac). Radioactivity was counted using an LKB Wallac 1205 Betaplate Beta Liquid Scintillation Counter.
In addition, natural killer (NK) cell activity was evaluated in cells from spleen cell suspensions by overnight incubation at 37°C as described elsewhere [46]. The activity of NK cells was measured as the ability of 2*106 spleen cells to lyse 1*10451Cr-labelled YAC-1 target cells during a 4 h co-incubation in 96-well cell culture plates (Greiner) at 37°C. Radioactivity was counted using a Perkin-Elmer Packard Cobra II Auto Gamma Counter. NK-cell activity was given as a percentage of the maximal release by YAC cells, calculated as (radioactivity counts in the supernatant minus spontaneous release by YAC)/(maximal release by YAC cells minus the spontaneous release by YAC cells).
Plasma IgG and IgM levels
Plasma IgG and IgM levels were measured using rat IgG and rat IgM ELISA kits (E25G and E25M, respecively; ICL, Gentaur, the Netherlands). IgG was measured at 20,000-, 40,000-, and 80,000-fold serum dilutions, while IgM was measured at a 600-fold serum dilution.
Cytokine levels in culture supernatants
MLN and spleen cells were incubated with LPS and Con A using the same cell concentrations and LPS and Con A concentrations as described above. Incubation with LPS was for 48 h, while incubation with Con A was for 72 h. To determine the cytokine levels in the culture supernatants a 4-plex panel (Bio-Rad, Hercules, CA) was used in case of LPS-stimulation (IL-1β, IL-6, IL-10, and TNF-α), while a 9-plex panel (Bio-Rad) was used in case of Con A-stimulation (IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-10, IL-13, IL-17α, and TNF-α). A volume of 100 μl Bio-Plex assay buffer (Bio-Rad) was added to 96-wells filter bottom plates (Bio-Rad) to pre-wet the plate. Buffer was removed by vacuum after each incubation or wash step. Beads were diluted in assay buffer, and 50 μl/well was added. Then, the plates were washed 2 times with 100 μl Bio-Plex wash buffer (Bio-Rad). Dilution series of the cytokine standards were made ranging from 32,000 to 0.18 pg/ml. Fifty μl of the standards and cell culture supernatants was added to the wells and the plates were vortexed at 1100 rpm for 30 s and incubated at RT for 30 min while vortexing at 300 rpm. After incubation the plates were washed 3 times with 100 μl wash buffer. Detection antibody was diluted in detection antibody diluent (Bio-Rad), and 25 μl/well was added. The plates were again vortexed at 1100 rpm for 30 s, incubated at RT for 30 min while vortexing at 300 rpm and washed 3 times with 100 μl assay buffer. Next, streptavidin-PE was diluted in assay buffer and 50 μl/well was added. The plates were incubated for 10 min at RT. After 3 times washing with 100 μl wash buffer the beads were resuspended in 125 μl assay buffer and read on a Bio-Plex (Bio-Rad). Results were obtained at low photomultiplier tube settings.
Histological analysis
For histopathology, liver, jejunum, kidney and spleen samples were harvested and fixed in 10% neutral buffered formalin. Subsequently, they were dehydrated in a series of ethanol and embedded in paraffin. Approximately 4 μm thick sections were cut, mounted on glass slides and stained with hematoxylin and eosin (H&E). The sections were observed under an optical microscope (Olympus, BX 60, Japan) at different magnifications. All quantitative histological analyses were performed blind to the treatment of the groups.
For livers, 10 images (each of 1 mm2) in 10 slides (3 μm apart) in one of the liver lobes per animal were evaluated. Per slide the total number of inflammatory foci, and the total number of inflammatory cells in these foci was scored. Only foci consisting of more than 10 inflammatory cells were included in the analysis. Furthermore, the total number of apoptotic cells was scored per slide, as well as the presence of necrosis or fibrosis. Necrosis resulted in a positive or negative score, leading to a minimum score of 0 positive slides and a maximum score of 10 positive slides per rat, whereas fibrosis was also indexed on severity per slide (0,=not remarkable, 1 = very mild, 2 = mild, 3 = moderate, 4 = severe, 5 = very severe). Apoptotic cells were also visualized using an Apoptaq® Peroxidase in situ apoptosis detection kit (Millipore Corporation, Billerica, USA). Representative micrographs were recorded using a Leica DFC 450 camera (Leica, The Netherlands) fitted onto the microscope.
In jejunum samples, 5 μm transverse sections were cut. The villus height, crypt depth, and villus:crypt ratio were measured using an image-analyzing software package (Cell^D; Olympus Soft Imaging Solutions GmbH, Germany), coupled to an optical microscope (Olympus, Japan). A minimum of 10 villi was measured in one slide per animal. All villus height having lamina propria were measured from the villus tip to the end of the base, except the crypt. Crypt depth measurements were taken from the valley between individual villi to the basolateral membrane.
For SEM-EDX analysis of liver tissue, tissues were fixed and dehydrated as described above, but sections were mounted on silicium free ThermanoxTM coverslips (Nunc, Germany). Sections were H&E stained and observed under an optical microscope as described above and then deparaffinized in xylene overnight. The sections were subsequently sputter coated with chromium using a K575X turbo sputter coater (Emitech) and analyzed with SEM-EDX as described before. EDX-mapping analysis was used to search for clusters with increased silica content in the tissue. Here, a detection limit of 100 mg silica/kg tissue was estimated. Liver macrophages, selected with light microscopy, were analyzed individually for the presence of silica in the cells with SEM-EDX.
Statistical analysis
Results were statistically analyzed using Prism (v5; GraphPad Software, Inc., La Jolla, USA) and GenStat 15th edition (version 15.2.0.8821) software. Body- and organ weights, ICP-MS, cytokine release, gene expression of individual genes, and biochemical analysis results were analyzed with a two-way ANOVA with a Bonferroni post-test. HDC ICP-MS, lymphocyte transformation, NK-activity, and antibody release results were analyzed with a one-way ANOVA with a Bonferroni post-test. Outliers in the (HDC) ICP-MS results were removed according to Chauvenet’s criterion. All histopathology data was analyzed with a logarithmic regression analysis using a Poisson distribution, except for the necrosis and fibrosis data, which were analyzed by logistic regression using a binomial distribution between 0 and 10. For all statistical results, a p-value of ≤ 0.05 was considered significant.
Transcriptomic analysis
To study the effects of the treatment on the transcriptome of liver cells, a piece of liver tissue was immediately frozen in liquid nitrogen during dissection after the 28-, or 84-day exposure and stored until further use. 1250 μl Trizol and 10–15 zirconia/silica beads (Lab Services BV, Breda, The Netherlands) were added to the tissue after which the tissue was homogenized (homogenizer: Precellys 24, Amsterdam, The Netherlands) at 6500 bpm for 2x 15 s with a 30 s interval. The mixture was centrifuged at 12,000 g for 15 min at 4°C. The supernatant was mixed with 300 μl chloroform, incubated at room temperature for 3 min and centrifuged at 12,000 g for 15 min at 4°C. The aqueous phase was transferred to be mixed with 750 μl isopropyl alcohol, which precipitates total RNA. After overnight incubation at −20°C and centrifuging (20 min, 12,000g at 4°C), the pellet was washed with 75% ethanol, centrifuged again at 12,000g for 10 min at 4°C, and resuspended in RNase-free water. Subsequently, RNA was further purified using the RNeasy Mini Kit (Qiagen, Venlo, The Netherlands). Purity, and concentration of the RNA were assessed using the nanodrop (Isogen, De Meern, The Netherlands) at wavelengths of 230, 260, and 280 nm and RNA integrity was checked on an Agilent 2100 Bioanalyzer (Agilent Technologies, Amsterdam, The Netherlands) with 6000 Nano Chips. RNA was judged as suitable only if samples showed intact bands of 18S and 28S ribosomal RNA subunits, displayed no chromosomal peaks or RNA degradation products, and had a RNA integrity number (RIN) above 8.0.
For each individual rat, total RNA (100 ng) of the liver was labeled using the Ambion WT expression kit (Life Technologies, Bleiswijk, The Netherlands). RNA samples were hybridized on Affymetrix GeneChip Rat Gene 1.1 ST arrays. Hybridization, washing, staining and scanning was performed on an Affymetrix GeneTitan instrument. Array data were analyzed using an in-house, on-line system [47]. Shortly, probesets were redefined according to Dai et al.[48] using remapped CDF version 15.1 based on the Entrez Gene database and a robust multi-array (RMA) analysis was used to obtain expression values [49, 50]. Gene expression data from one rat in the control and NM-202 medium group did not pass the quality control criteria and were excluded.
Spot intensities were floored to 17, which was followed by 2 log mean- centering and calculation of 2 log ratios of treatments versus the average of the control samples. Hierarchical clustering was done using the programs Cluster 3.0 (uncentered correlation; average linkage clustering) and Treeview 1.6 (Eisen Lab, USA). Gene set enrichment analysis (GSEA) was performed to discover the differential expression of biologically relevant sets of genes [51]. For this, several genes sets related to common liver processes were composed using the text-mining tool Anni 2.0 (http://www.biosemantics.org/anni) [52]. Furthermore, several gene sets related to specific toxicological responses in the liver, composed from literature and/or in-house data were used, as well as publicly available gene sets (Additional file 1: Table S12). Significantly enriched gene sets were selected on the basis of a p-value <0.01 in combination with an FDR-value <0.25 according to GSEA statistics. Genes that contributed to the enrichment of these sets were selected and filtered on >1.2x up- or down-regulated versus the average of the controls in ≥ 3 out of the 5 rats.