Animals and exposure. Specific-pathogen-free C57BL/6 female mice (~ 6 weeks of age) were purchased from Jackson Laboratories (Bar Harbor, ME). Mice were provided pelleted rodent diet (ProLab 3000; Purina Mills, Gray Summit, MO) and filtered water ad libitum and were maintained in facilities accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. A total of 20 mice (10 mice/group, body weight = 20 ± 3 g) were housed in static microisolator cages (5 mice/cage) on heat-treated hardwood bedding, under environmental conditions of 22°C, 40–70% humidity, and a 12:12 hr light:dark cycle. All experiments were approved by the Massachusetts Institute of Technology Committee on Animal Care. The animals were treated humanely and with regard for the alleviation of suffering. Inorganic arsenic (arsenic, 10 ppm) was administered to mice (~ 8 weeks of age) as sodium arsenite (Fisher Scientific, Waltham, MA) in drinking water for 4 weeks. Freshly prepared arsenic-containing water (10 ppm) was provided to mice every Monday and Thursday. Control mice received water alone.
Animal monitoring and histological analysis. Throughout the experiments, mice were assessed daily for evidence of diarrhea, dehydration, and deteriorating body condition. Mice were euthanized with carbon dioxide and necropsied after 4 weeks of arsenic consumption. Formalin-fixed tissues were routinely processed, embedded in paraffin, sectioned at 4 μm, stained with hematoxylin and eosin, and evaluated by a board-certified veterinary pathologist blinded to the sample identity. Inflammation, edema, epithelial defects, hyperplasia, and dysplasia of multiple regions of liver (left lateral, medial, right lateral, and caudate lobes) and colon (distal, transverse, and proximal colon) were scored on an ascending scale (0–4, with 0 being normal) of severity and invasiveness of the lesion, if any. Pathological scores did not show any significant difference between the control and arsenic-treated mice and are not presented here. We also did not observe any significant changes in body weights, mortality, and food intake.
16S rRNA gene sequencing. We isolated DNA from fecal pellets collected during necropsy using a PowerSoil® DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA) according to the manufacturer’s instructions. The resultant DNA was quantified by ultraviolet spectroscopy and stored at –80°C for further analysis. DNA was amplified using universal primers of U515 (GTGCCAGCMGCCGCGGTAA) and E786 (GGACTACHVGGGTWTCTAAT) to target the V4 regions of 16S rRNA of bacteria. Individual samples were barcoded, pooled to construct the sequencing library, then sequenced using an Illumina Miseq (Illumina, San Diego, CA) to generate pair-ended 150 × 150 reads.
Analysis of 16S rRNA sequencing data. The raw mate-paired fastq files were quality-filtered, demultiplexed, and analyzed using Quantitative Insights into Microbial Ecology (QIIME) software (http://qiime.org ). For quality filtering, the default parameters of QIIME were maintained in which reads with a minimum Phred quality score of < 20, containing ambiguous base calls and < 113 bp of consecutive high-quality base calls, were discarded. In addition, reads with three consecutive low-quality bases were truncated. The samples sequenced were demultiplexed using 8-bp barcodes, allowing 1.5 errors in the barcode. UCLUST software (http://www.drive5.com/uclust ) was used to choose the operational taxonomic units (OTUs) with a threshold of 97% sequence similarity. A representative set of sequences from each OTU was selected for taxonomic identification of each OTU using the Ribosomal Database Project (RDP) classifier (http://rdp.cme.msu.edu ). The Greengenes OTUs (4feb2011 build) reference sequences (97% sequence similarity) were used as the training sequences for RDP classifier. A 0.80 confidence threshold was used for taxonomic assignment. The taxonomic assignment of 16S rRNA sequencing data could be achieved at different levels, including phylum, class, order, family and genus. Our analyses were typically conducted at the family level because of the higher confidence in the assignment of taxa based on the sequencing reads; therefore, a significant change at the family level may reflect changes of multiple gut bacteria at genus and species levels. The sequencing data have been deposited in the MG-RAST (metagenomics Rapid Annotation using Subsystem Technology) server (http://metagenomics.anl.gov/ ).
Sample processing for metabolomics. One day before euthanasia of the mice, we collected urine samples using a metabolic cage with dry ice placed around the urine collection vessel to prevent oxidation or degradation of metabolites during the collection period (approximately 16 hr). Fecal pellets were also collected from individual animals. We collected plasma samples during necropsy. Metabolites were isolated from urine using methanol as described previously (Lu et al. 2012 (link)). Cold methanol (80 μL) was added to 20 μL urine or plasma. After vortexing for 1 min, the samples were incubated at 4°C for 20 min and then centrifuged for 10 min at 12,000 rpm. The supernatant was collected, dried in a SpeedVac (Savant SC110A; Thermo Electron, Waltham, MA), and then resuspended in 30 μL 98:2 water:acetonitrile for MS analysis. Metabolite extraction from fecal pellets was conducted in a similar manner. Fecal pellets (25 mg) were dissolved in 400 μL cold methanol solution (50:50 methanol:water), followed by vortexing at maximum speed for 10 min using a flat-bed vortex (MO BIO Laboratories). The supernatant was centrifuged for 10 min at 12,000 rpm, dried in a SpeedVac, and then resuspended in 30 μL 98:2 water:acetonitrile for metabolomics profiling.
Metabolomics profiling. We performed LC-MS analyses on a quadrupole-time-of-flight (Q-TOF) 6510 mass spectrometer (Agilent Technologies, Santa Clara, CA) with an electrospray ionization source. The mass spectrometer was interfaced with an Agilent 1200 HPLC system. The Q-TOF was calibrated daily using the standard tuning solution from Agilent Technologies. The typical mass accuracy of the Q-TOF was < 10 ppm. Metabolites were analyzed in the positive mode only over a range of 80–1000 m/z using a C18 T3 reverse-phase column from Waters Corporation (Milford, MA) because of the higher numbers of detected molecular features (i.e., metabolites), as demonstrated previously (Lu et al. 2012 (link)). Metabolomics profiling data were processed as described previously (Lu et al. 2012 (link)). MS/MS was generated on the Q-TOF to confirm the identity of perturbed metabolites. The metabolomics data were submitted to the XCMS Online server (https://xcmsonline.scripps.edu/ ).
Data processing of metabolomics data. Data acquired in Agilent .d format were converted to mzXML using MassHunter Workstation software from Agilent Technologies. Data were filtered by intensity, and only signals with intensities > 1,000 were considered. The converted data were processed using XCMS Online for peak picking, alignment, integration, and extraction of the peak intensities. To profile individual metabolite differences between control and arsenic-treatment groups, a two-tailed Welch’s t-test was used (p < 0.05). The exact masses of molecular features with significant changes were searched against the Human Metabolome Database (HMDB;http://www.hmdb.ca/ ), METLIN (http://metlin.scripps.edu ), and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases (http://www.genome.jp/kegg/pathway.html ) with a 10-ppm mass accuracy threshold. The matched exact masses were stored and used for the generation of MS/MS data to identify the metabolites.
Statistical analysis of data. Principal component analysis (PCA) was performed to examine intrinsic clusters of metabolomics data. A 95% confidence interval (CI) was used as the threshold to identify potential outliers in all samples. In addition, heat maps were generated using a hierarchical clustering algorithm to visualize the metabolite difference within the data set. Principal coordinate analysis (PCoA) was used to compare the gut microbiome profiles between the control and treatment. The difference in the gut microbiome composition was further assessed using a nonparametric test via Metastats software (http://metastats.cbcb.umd.edu/ ) as described previously (White et al. 2009 (link)). The correlation matrix between the gut microflora–related metabolites and gut bacterial species was generated using Pearson’s corelation coefficient.
Animal monitoring and histological analysis. Throughout the experiments, mice were assessed daily for evidence of diarrhea, dehydration, and deteriorating body condition. Mice were euthanized with carbon dioxide and necropsied after 4 weeks of arsenic consumption. Formalin-fixed tissues were routinely processed, embedded in paraffin, sectioned at 4 μm, stained with hematoxylin and eosin, and evaluated by a board-certified veterinary pathologist blinded to the sample identity. Inflammation, edema, epithelial defects, hyperplasia, and dysplasia of multiple regions of liver (left lateral, medial, right lateral, and caudate lobes) and colon (distal, transverse, and proximal colon) were scored on an ascending scale (0–4, with 0 being normal) of severity and invasiveness of the lesion, if any. Pathological scores did not show any significant difference between the control and arsenic-treated mice and are not presented here. We also did not observe any significant changes in body weights, mortality, and food intake.
16S rRNA gene sequencing. We isolated DNA from fecal pellets collected during necropsy using a PowerSoil® DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA) according to the manufacturer’s instructions. The resultant DNA was quantified by ultraviolet spectroscopy and stored at –80°C for further analysis. DNA was amplified using universal primers of U515 (GTGCCAGCMGCCGCGGTAA) and E786 (GGACTACHVGGGTWTCTAAT) to target the V4 regions of 16S rRNA of bacteria. Individual samples were barcoded, pooled to construct the sequencing library, then sequenced using an Illumina Miseq (Illumina, San Diego, CA) to generate pair-ended 150 × 150 reads.
Analysis of 16S rRNA sequencing data. The raw mate-paired fastq files were quality-filtered, demultiplexed, and analyzed using Quantitative Insights into Microbial Ecology (QIIME) software (
Sample processing for metabolomics. One day before euthanasia of the mice, we collected urine samples using a metabolic cage with dry ice placed around the urine collection vessel to prevent oxidation or degradation of metabolites during the collection period (approximately 16 hr). Fecal pellets were also collected from individual animals. We collected plasma samples during necropsy. Metabolites were isolated from urine using methanol as described previously (Lu et al. 2012 (link)). Cold methanol (80 μL) was added to 20 μL urine or plasma. After vortexing for 1 min, the samples were incubated at 4°C for 20 min and then centrifuged for 10 min at 12,000 rpm. The supernatant was collected, dried in a SpeedVac (Savant SC110A; Thermo Electron, Waltham, MA), and then resuspended in 30 μL 98:2 water:acetonitrile for MS analysis. Metabolite extraction from fecal pellets was conducted in a similar manner. Fecal pellets (25 mg) were dissolved in 400 μL cold methanol solution (50:50 methanol:water), followed by vortexing at maximum speed for 10 min using a flat-bed vortex (MO BIO Laboratories). The supernatant was centrifuged for 10 min at 12,000 rpm, dried in a SpeedVac, and then resuspended in 30 μL 98:2 water:acetonitrile for metabolomics profiling.
Metabolomics profiling. We performed LC-MS analyses on a quadrupole-time-of-flight (Q-TOF) 6510 mass spectrometer (Agilent Technologies, Santa Clara, CA) with an electrospray ionization source. The mass spectrometer was interfaced with an Agilent 1200 HPLC system. The Q-TOF was calibrated daily using the standard tuning solution from Agilent Technologies. The typical mass accuracy of the Q-TOF was < 10 ppm. Metabolites were analyzed in the positive mode only over a range of 80–1000 m/z using a C18 T3 reverse-phase column from Waters Corporation (Milford, MA) because of the higher numbers of detected molecular features (i.e., metabolites), as demonstrated previously (Lu et al. 2012 (link)). Metabolomics profiling data were processed as described previously (Lu et al. 2012 (link)). MS/MS was generated on the Q-TOF to confirm the identity of perturbed metabolites. The metabolomics data were submitted to the XCMS Online server (
Data processing of metabolomics data. Data acquired in Agilent .d format were converted to mzXML using MassHunter Workstation software from Agilent Technologies. Data were filtered by intensity, and only signals with intensities > 1,000 were considered. The converted data were processed using XCMS Online for peak picking, alignment, integration, and extraction of the peak intensities. To profile individual metabolite differences between control and arsenic-treatment groups, a two-tailed Welch’s t-test was used (p < 0.05). The exact masses of molecular features with significant changes were searched against the Human Metabolome Database (HMDB;
Statistical analysis of data. Principal component analysis (PCA) was performed to examine intrinsic clusters of metabolomics data. A 95% confidence interval (CI) was used as the threshold to identify potential outliers in all samples. In addition, heat maps were generated using a hierarchical clustering algorithm to visualize the metabolite difference within the data set. Principal coordinate analysis (PCoA) was used to compare the gut microbiome profiles between the control and treatment. The difference in the gut microbiome composition was further assessed using a nonparametric test via Metastats software (
Free full text: Click here
