TOXICOLOGICAL SCIENCES 100(1), 75–87 (2007) doi:10.1093/toxsci/kfm200 Advance Access publication August 6, 2007
Exposure to Arsenic at Levels Found in U.S. Drinking Water Modifies Expression in the Mouse Lung
Angeline S. Andrew,*,†,‡,1 Viviane Bernardo,§,{ Linda A. Warnke,k Jennifer C. Davey,‡,kj Thomas Hampton,‡,kj Rebecca A. Mason,*,kj Jessica E. Thorpe,k Michael A. Ihnat,k and Joshua W. Hamilton†,‡,kj *Department of Community and Family Medicine, Dartmouth Medical School, Lebanon, New Hampshire 03756; †Norris Cotton Cancer Center, Dartmouth- Hitchcock Medical Center, Lebanon, New Hampshire 03756; ‡Center for Environmental Health Sciences, Dartmouth Medical School, Hanover, New Hampshire Downloaded from https://academic.oup.com/toxsci/article/100/1/75/1624780 by guest on 29 September 2021 03755; §Thayer School of Engineering/Computer Sciences Department, Dartmouth College, Hanover, New Hampshire 03755; {Health Informatics Department, Federal University of Sao Paulo/Escola Paulista de Medicina-UNIFESP/EPM, Sao Paulo, SP, Brazil; kDepartment of Cell Biology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma 73190; and kjDepartment of Pharmacology and Toxicology, Dartmouth Medical School, Hanover, New Hampshire 03755
Received April 6, 2007; accepted July 30, 2007
smoking and other risk factors to induce mutations, DNA The mechanisms of action of drinking water arsenic in the lung adducts, and cancer risk (Ahsan and Thomas, 2004; Chen and the threshold for biologic effects remain controversial. Our et al., 2004; Evans et al., 2004; Rossman et al., 2002). The study utilizes Affymetrix 22,690 transcript oligonucleotide micro- mechanisms of action and the threshold for biologic effects and arrays to assess the long-term effects of increasing doses of drinking water arsenic on expression levels in the mouse lung. disease risk, however, remain controversial (Kitchin, 2001). Mice were exposed at levels commonly found in contaminated Many of the epidemiologic studies and in vivo / in vitro drinking water wells in the United States (0, 0.1, 1 ppb), as well experiments have been conducted at high doses with acute as the 50 ppb former maximum contaminant level, for 5 weeks. exposure. Thus far, microarray studies have explored arsenic- The expression profiles revealed modification of a number of induced gene expression changes in cell culture models as well important signaling pathways, many with corroborating evidence as several animal organs; however, the effects in the adult lung of arsenic responsiveness. We observed statistically significant have not been reported (Durham and Snow, 2006; Shi et al., expression changes for transcripts involved in angiogenesis, lipid 2004). In contrast to a strengthening of similar effects with metabolism, oxygen transport, apoptosis, cell cycle, and immune increasing dose that is observed with many compounds, arsenic response. Validation by reverse transcription–PCR and immuno- may activate completely different pathways at low versus high blot assays confirmed expression changes for a subset of tran- doses (Andrew et al., 2003; Barchowsky et al., 1999; Lau scripts. These data identify arsenic-modified signaling pathways that will help guide investigations into mechanisms of arsenic’s et al., 2004; Soucy et al., 2003). Relatively little is known health effects and clarify the threshold for biologic effects and about the effects of arsenic at levels of exposure that are potential disease risk. common to drinking water in the United States, particularly in Key Words: arsenic; apoptosis; cell cycle; drinking water; the lung. immune response; lung; microarray; oxygen transport. Our study utilizes oligonucleotide microarrays to assess the long-term effects of levels of drinking water arsenic commonly found in contaminated wells in the United States, as well as the 50 ppb former standard (0, 0.1, 1, and 50 ppb) on expression Drinking water arsenic exposure is an established human levels in the mouse lung. We present a functional approach health risk associated with cardiovascular disease, diabetes, and to microarray analysis in which we focus on statistically cancer at multiple organ sites including the lung, bladder, and significant expression modifications and investigate both the skin (IARC, 2004). Arsenic exposure has been associated with biologic function and how each protein regulates another using impaired lung function and bronchiectasis (De et al., 2004; updated bioinformatics analysis tools. Smith et al., 2006). In an Indian study, arsenic-associated skin lesions were independently associated with increased risk of respiratory illness (odds ratio 4.9; 95% confidence interval 3.2– 7.5) compared to individuals with no skin lesions (Ghosh et al., MATERIALS AND METHODS 2007). Arsenic has also been found to synergize with cigarette Four groups of adult (7–8 weeks old) male C57/BL6 mice were exposed to 1 To whom correspondence should be addressed at Dartmouth Medical increasing concentrations of inorganic arsenic (n ¼ 4 at 0.1 lg/l [ppb], n ¼ 3 School, 7927 Rubin 860, One Medical Center Drive, Lebanon, NH 03756. Fax: at 1 lg/l [ppb], and n ¼ 3at50lg/l [ppb]) (sodium arsenite; LabTech, (603) 653-9093. E-mail: [email protected]. Pittsburgh, PA) in their drinking water for a period of 5 weeks. A control group
Ó The Author 2007. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For Permissions, please email: [email protected] 76 ANDREW ET AL. of mice (n ¼ 3) consumed uncontaminated drinking water. Lungs from We further characterized the functional effects of all the statistically additional groups of mice exposed to 0 or 10 lg/l (ppb) were used in the reverse significant arsenic-modified transcripts (includes all of the SAM two-class transcription (RT)–PCR and immunoblot analyses. All mice were fed AIN-76A selected transcripts) by implementing the Database for Annotation, Visualiza- arsenic-free chow (Harlan-Teklad, Madison, WI) to control for other dietary tion, and Integrated Discovery (DAVID) gene ontology search engine (Table 2) sources of arsenic (total arsenic < 5 ppb). Levels of arsenic in food and water (Dennis et al., 2003). This bioinformatic tool identifies functional processes were confirmed by Inductively Coupled Plasma Mass Spectrometry analysis by that are significantly overrepresented ( p < 0.05) by the modified transcripts. the Dartmouth Trace Metal Facility. Lungs were removed and immediately Table 3 lists the biologic roles of all the significant arsenic-modified transcripts placed in RNAlater to stabilize the RNA levels after sacrificing the animals. (all those selected by SAM multiclass or two-class analysis at any exposure) Exposed and unexposed animals were sacrificed at the same time of day. All from the Pathway Studio ResNet 4.0 database (updated April 2007) (Ariadne protocols were approved by the University of Oklahoma Institutional Animal Genomics, Rockville, MD). This Pathway Studio ResNet database catalogs Care and Use Committee. relationships between biologic entities based on the published literature and RNA was isolated using Qiagen RNeasy columns (Qiagen Inc., Valencia, was also used to identify the direct interactions between the arsenic-modified CA), and DNAse treatment was performed using Ambion DNAfree reagents transcripts at each arsenic dose (Fig. 3 and Supplementary Data). (Austin, TX) according to the manufacturer’s instructions. The expression Histologic slices of formalin-fixed, paraffin-embedded lung tissue from Downloaded from https://academic.oup.com/toxsci/article/100/1/75/1624780 by guest on 29 September 2021 profiles were generated using the Affymetrix GeneChip Technology—chip arsenic-exposed and unexposed animals were stained with hematoxylin and GeneChip Murine Genome 430 oligonucleotide arrays (Affymetrix, Santa Clara, eosin and evaluated by a trained pathologist to assess the number of neutrophils CA) which simultaneously tested 22,690 transcripts using one chip for each (Supplementary Data). mouse on the integrated GeneChip Instrument System in the Dartmouth The levels of Nr4a1, Hsp70, Ahr, and Cyclin D1 protein were assessed by Microarray Core Facility. Our experiment was performed in compliance with the immunoblotting using sodium dodecyl sulfate–polyacrylamide gel electropho- Minimum Information About a Microarray Experiment checklist for standard- resis (SDS–PAGE) to resolve proteins from mouse lung tissue (Fig. 4). Frozen ization guidelines for microarray experiments. mouse lung was weighed and homogenized with EBC lysis buffer (50mM Tris, The raw microarray data were preprocessed and normalized using pH 8.0, 100mM NaCl) containing 10 ll/ml PMSF, 5 ll/ml aprotinin, and 1 ll/ml GeneTraffic version 3.2 which is a microarray data management and analysis leupeptin. NP-40 Lysis buffer (10%) was added at a 5% vol/vol ratio. After client-server application (Stratagene, La Jolla, CA). The control (0 lg/l) group centrifugation for 15 min at 14,000 rpm at 4°C the lysates were boiled for was set as the baseline, and data were normalized using Robust Multi-Chip 5 min and clarified by centrifugation at 13,000 rpm for 10 min. Equal amounts Analysis (Irizarry et al., 2003). of cell lysates were resolved by electrophoresis on 7.5% and 10–20% SDS– The statistical significance of expression changes relative to the control polyacrylamide gels. Electrophoresis was performed at constant voltage (200 V), group was assessed using methods based on modified t- or F-tests that adjust then the resolved proteins were transferred from the polyacrylamide gel to for multiple comparisons. This adjustment bounds a false discovery rate polyvinylidene difluoride membrane (PVDF, Immobilon-P; Millipore, Bedford, probability (FDR), i.e., chance that a transcript regarded as significant is a false MA) by semi-dry transfer (Hoeffer Semiphor, San Fransisco, CA) for 1 h at positive, to select the transcripts (Benjamini and Hochberg, 1995). This method constant current (100 mA) using transfer buffer (25mM Tris, 192mM glycine, is implemented in the Significance Analysis of Microarrays (SAM) application 20% (vol/vol) methanol, 0.01% SDS). To eliminate nonspecific interactions of version 1.13 (Tusher et al., 2001) implemented in The Institute of Genomic antibodies with the membrane, the PVDF membrane was blocked with Tris- Research (TIGR) MultiExperiment Viewer (TIGR MeV) version 3.1 (TIGR, Tween Buffered Saline (TTBS) (10mM Tris–HCl, pH 8.0, 150mM NaCl, Rockville, MD). We performed SAM multiclass and two-class comparisons 0.05% Tween-20) containing 5% milk (7.5 g/150 ml) for 1 h at room using 1000 permutations and selected significant transcripts at a FDR of 5%. temperature. Membranes were incubated with the primary antibody: anti-Nur77/ We focused our attention on the group of differentially expressed transcripts Nr4a1 (Pharmingen, San Diego, CA), Hsp70/Hspa1b (Transduction Labora- tories, Lexington, KY) diluted 1:1000, anti-Ahr (BIOMOL, Plymouth Meeting, that were called significant by the two-class SAM or multiclass SAM (dose PA) diluted 1:5000, or Cyclin D1 antibody (Santa Cruz Biotechnology Inc., response) analysis (provided as Supplementary Data). The transcripts that were Santa Cruz, CA) diluted 1:500 in TTBS overnight at 4°C. Actin was used as significant by multiclass analysis or were modified at a minimum of two doses a loading control, and the antibody was diluted 1:50,000 in TTBS for 1 h of arsenic are shown in Table 1. Transcripts are grouped by the lowest arsenic (Calbiochem, San Diego, CA). The membranes were washed 3 times with dose with statistically significant transcript expression modification compared TTBS. The Nur77/Nr4a1 and Hsp70 membranes were incubated with with control. To identify subgroups of transcripts with similar patterns of horseradish peroxidase (HRP)–linked goat anti-rabbit IgG (Santa Cruz expression among the statistically significant group, the multiclass selected Biotechnology Inc.) 1:3000 in TTBS with 5% milk (1.5 g/30 ml) for 1 h at transcripts were clustered by TIGR MeV with a hierarchical clustering with the room temperature. The Ahr, Cyclin D1, and Actin membranes were incubated complete linkage algorithm and Pearson correlation metric (Fig. 1). with HRP-linked goat anti-mouse (Bio-Rad Laboratories, Inc., Hercules, CA) We selected transcripts for validation by real-time PCR using independent 1:2000, 1:3000, and 1:2000, respectively, in TTBS with 5% milk (1.5 g/30 ml). primer sets based on the microarray results and hypothesized involvement in After 3 washes with TTBS, protein bands were visualized by enhanced arsenic pathogenesis in the lung (Fig. 2). Taqman primer–probe sets for each chemiluminescence using the Amersham ECL Plus Western Blotting Detection selected transcript were obtained from Applied Biosystems Inc. (ABI, Foster system (GE Healthcare, Piscataway, NJ) and film (Lumi-Film; Roche City, CA). Real-time RT–PCR was performed using the ABI PRISM sequence Molecular Biochemicals, Indianapolis, IN). detection system and software. Briefly, total RNA (0.5 lg) was reverse transcribed using 100 U Moloney Murine Leukemia Virus reverse transcriptase in a mixture with oligo-dT and dNTPs according to the instructions provided with the Qiagen Omniscript kit (Qiagen Inc.). Samples were reverse transcribed RESULTS in a MJ Research PTC-100 thermocycler (MJ Research Inc., Watertown, MA) for 60 min at 44°C and the reaction terminated by heating to 95°C for 10 min. To address the controversy over the biological effects of Expression of specific genes was assessed by real-time PCR using 10 ng total arsenic at levels commonly found in U.S. drinking water, we RNA, 400nM primers, 200nM probe, and TaqMan Universal PCR Master Mix assessed expression patterns associated with arsenic exposure (ABI). Relative quantitation was performed using a standard curve consisting in the mouse lung. To evaluate the statistical significance of of serial dilutions of pooled sample cDNA from the same source as the test RNA with each plate. Relative expression levels of each gene were normalized these changes in expression between the treated and control to 18s rRNA. Statistical significance was assessed by one-way ANOVA with animals, we started with the hypothesis that arsenic would have Newman-Keuls post-test using GraphPad Prism software (San Diego, CA). a dose-responsive effect on expression. In Figure 1, we TABLE 1 Multiple Dose Significantly Modified Transcriptsa Organized by Arsenic Dose
Fold change Probes significantd (n) Entrez Trend Figure 1 Symbol Gene name Functionb gene ID (þ/ )c 0.1 ppb 1 ppb 50 ppb 0.1 ppb 1 ppb 50 ppb Multi cluster Downloaded fromhttps://academic.oup.com/toxsci/article/100/1/75/1624780bygueston29September2021
Down 0.1 ppb Angptl4 Angiopoietin-like 4 Negative regulation of apoptosis, 57875 1.0 0.2 0.2 1 1 c1 inhibits angiogenesis, inhibits lipoprotein lipase Cd68 CD68 antigen Macrosialin (CD68), macrophage- 12514 0.6 0.6 0.2 1 1 c1 specific, increased by aging in brain Cpt1a Carnitine palmitoyltransferase 1a, Fatty acid metabolism 12894 0.6 0.6 0.3 1 1 c1 liver Cxcl7 Chemokine (C-X-C motif) ligand 7 Iimmune response, chemokine, 57349 1.1 0.8 0.3 1 1 c1 cytokine MICROARRAY ARSENIC WATER DRINKING D4Wsu53e DNA segment, Chr 4, Wayne State NA 27981 0.9 0.4 0.7 2 1 c1 University 53 Eif4b Eukaryotic translation initiation Regulation of translational initiation 75705 0.4 0.5 0.4 2 1 c3 factor 4B Hbb-b1 Hemoglobin, beta adult major chain Hemoglobin complex member, 15129 0.6 0.6 0.3 1 hemopoiesis, oxygen transport, iron ion binding Hbp1 High mobility group box Transcriptional repressor, substrate 73389 0.5 0.6 0.1 1 1 c1 transcription factor 1 for p38 MAP kinase Ms4a8a Membrane-spanning 4-domains, A, Receptor activity 64381 0.4 0.2 0.2 1 1 c1 8A Pabpc1 Poly A binding protein, RNA binding 18458 0.5 0.6 0.6 1 1 c3 cytoplasmic 1 S100a9 S100 calcium-binding protein A9 Inflammatory response, neutrophil 20202 2.4 2.7 2.0 1 (calgranulin B) release Siat8d ST8 alpha-N-acetyl-neuraminide Protein amino acid glycosylation 20452 0.5 1.5 1.2 1 1 c3 alpha-2,8-sialyltransferase 4 Down 0.1 and 1 ppb Alas2 Aminolevulinic acid synthase 2, Heme biosynthesis, metabolism 5- 11656 1.5 1.4 1.2 1 1 1 c3 erythroid aminolevulinate synthase activity Bpgm 2,3-Bisphosphoglycerate mutase Glycolysis, hemoglobin activity 12183 1.2 1.2 1.0 2 1 1 c3 Cd53 CD53 antigen macrophage protection against 12508 0.5 0.7 0.6 1 1 1 c3 LPS-induced oxidative stress and UVB irradiation Ear1 Eosinophil-ssociated, ribonuclease A Hydrolase activity, endonuclease 13586 1.2 1.3 0.4 1 1 1 c1 family, member 1 activity Ear2 Eosinophil-associated, ribonuclease Endonuclease activity 13587 1.0 1.2 0.3 2 1 1 c1 A family, 2 Hba-a1 Hemoglobin alpha, adult chain 1 Hemoglobin complex member, 15122 1.4 1.2 1.1 1 1 oxygen transport, iron ion binding 77 TABLE 1—Continued 78
Fold change Probes significantd (n) Entrez Trend Figure 1 Symbol Gene name Functionb gene ID (þ/ )c 0.1 ppb 1 ppb 50 ppb 0.1 ppb 1 ppb 50 ppb Multi cluster
Iigp1 Interferon inducible GTPase 1 Induced by various gram-negative 60440 0.6 0.8 0.6 1 1 1 c3 Downloaded fromhttps://academic.oup.com/toxsci/article/100/1/75/1624780bygueston29September2021 lipopolysaccharides, aids development of intracellular resistance during the interferon response to infection Ltgb2 Integrin beta 2 Activated T cell proliferation, cell- 16414 0.6 0.8 0.5 1 1 matrix adhesion Mkrn1 Makorin, ring finger protein, 1 Nucleic acid binding 54484 0.8 0.9 0.7 2 1 1 c3 Snca Synuclein, alpha Protects against oxidative stress via 20617 0.8 0.7 0.6 2 1 1 c1 inactivating the c-Jun-N-terminal kinase Down 0.1 and 50 ppb Cte1 Cytosolic acyl-CoA thioesterase 1 Acyl-CoA thioesterase activity, long- 26897 0.6 0.8 0.8 2 1 1 c1 chain fatty acid metabolism Eif3s5 Eukaryotic translation initiation Expression of a truncated eIF3e 66085 0.6 0.7 0.6 1 1 factor 3, subunit 5 (epsilon) causes malignant transformation
Fech Ferrochelatase Exon 10-deleted ferrochelatase 14151 0.5 0.3 0.5 1 1 AL. ET ANDREW heterozygous mice exhibited skin photosensitivity Gltscr2 Glioma tumor suppressor candidate NA 68077 0.5 0.7 0.5 2 1 region gene 2 Hlx H2.0-like homeo box gene NA 15284 0.3 0.5 0.6 1 1 Rgl1 Ral guanine nucleotide dissociation Guanyl-nucleotide exchange factor 19731 0.4 0.4 0.6 1 1 1 c3 stimulator-like 1 activity small GTPase mediated signal transduction Down 0.1, 1, and 50 ppb AI448196 Expressed sequence AI448196 NA 102910 0.6 0.8 0.7 1 1 1 1 c3 Stk17b Serine/threonine kinase 17b Plays critical roles in T cell apoptosis 98267 0.6 0.8 0.9 2 1 2 1 c3 (apoptosis-inducing) and memory T cell development Down 1 ppb Ahr Aryl-hydrocarbon receptor Cell cycle, xenobiotic metabolism, 11622 0.5 1.0 0.6 1 1 c3 regulation of transcription, DNA- dependent Cd79a CD79A antigen (immunoglobulin- Cell surface receptor linked signal 12518 0.4 1.0 0.8 1 1 c3 associated alpha) transduction, defense response, couples the B-cell antigen receptor to distal signaling pathways Ian6 Immune-associated nucleotide 6 Belongs immuno-associated 231931 0.5 1.0 0.8 1 1 c3 nucleotide (IAN) subfamily of nucleotide-binding proteins Lmo2 LIM domain only 2 involved in T-cell tumorigenesis due 16909 0.4 0.9 0.8 1 1 c3 to reprogramming of gene expression after enforced expression in T-cell precursors Down 1, 50 ppb Coro1a Coronin, actin-binding protein 1A Actin binding 12721 0.4 0.9 0.8 2 2 c3 Dbp D site albumin promoter-binding Regulation of transcription, DNA 13170 0.5 1.0 1.2 1 1 2 c3 protein binding, circadian rhythm
Plk2 Polo-like kinase 2 (Drosophila) ATP binding, protein serine/threonine 20620 0.2 0.8 0.7 1 1 1 c3 Downloaded fromhttps://academic.oup.com/toxsci/article/100/1/75/1624780bygueston29September2021 kinase activity cell cycle, protein amino acid phosphorylation Ptprc Protein tyrosine phosphatase, Cellular defense response, protein 19264 0.4 1.0 0.7 1 1 c3 receptor type, C amino acid dephosphorylation Down 50 ppb Agtrl1 Angiotensin receptor–like 1 Inhibits glucose-stimulated insulin 23796 0.7 1.5 1.0 1 1 c3 secretion Fus Fusion, derived from t(12;16) Positive regulation of transcription 233908 0.5 0.3 0.5 2 2 c3 malignant liposarcoma (human) from Pol II promoter Igh-VJ558 Immunoglobulin heavy chain (J558 B-cell antigen recognition. 16061 0.2 0.2 1.2 3 1 c4 family) Igj Immunoglobulin joining chain Humoral immune response, antigen 16069 0.3 0.0 1.1 1 1 c4 MICROARRAY ARSENIC WATER DRINKING binding. Igk Immunoglobulin kappa chain Immunoglobulin light chain variable 243469 0.1 0.3 1.1 1 1 c4 complex region Igk-V28 Immunoglobulin kappa chain Humoral immune response, antigen 16114 0.1 0.8 1.8 3 3 c4 variable 28 (V28) binding. Igk-V8 Immunoglobulin kappa chain Humoral immune response, antigen 16123 0.1 0.3 1.1 1 1 c4 variable 8 (V8) binding. Ltb Lymphotoxin B Tumor necrosis factor receptor 16994 0.5 0.6 0.9 1 1 c3 binding, immune response Nr1d1 Nuclear receptor subfamily 1, Regulation of transcription, DNA- 217166 0.6 0.7 0.9 1 1 c3 group D, member 1 dependent Nr4a1 Nuclear receptor subfamily 4, Regulation of transcription, steroid 15370 0.1 0.9 1.3 1 1 c3 group A, member 1 hormone receptor, apoptosis Down multiclass only Cd79b CD79B antigen Humoral immune response, maintains 15985 0.4 1.0 1.1 1 c3 preB cell and immature B cell survival and to mediate B cell differentiation Cirbp Cold inducible RNA binding protein RNA binding 12696 0.4 0.8 0.8 1 c3 Elf5 E74-like factor 5 Regulation of transcription 13711 0.1 0.6 0.4 1 c3 Myo10 Myosin X ATP binding, cytoskeleton 17909 0.0 0.6 0.6 1 c3 organization and biogenesis Myo6 Myosin VI ATP binding, cytoskeleton 17920 0.1 0.5 0.3 1 c3 organization and biogenesis Up multiclass only Fkbp5, Fkbp51 FK506 binding protein 5 Protein folding, glucocorticoid 14229 þ 0.2 0.6 0.8 1 c2 signaling Hspa8 Heat shock protein 8 ATPase activity, chaperone activity, 15481 þ 0.2 1.0 0.3 1 c2 regulation of cell cycle Mosc1 MOCO sulphurase C-terminal NA 66112 þ 0.1 0.8 0.6 1 c2 domain containing 1 MT2 Metallothionein 2 Metal ion binding 17750 þ 0.3 0.6 0.7 79 80 ANDREW ET AL.
hierarchically clustered transcripts that were selected by the multiclass SAM analysis (59 transcripts) and observed that the cluster Figure 1 data broke into four main branches, indicating subgroups of transcripts (labeled Clusters 1–4). Visual inspection of cluster 1 indicates decreased levels of transcripts involved in controlling )
n apoptosis, fatty acid metabolism, and chemokines at the 0.1 (
d and 1 ppb doses, but not at 50 ppb. Cluster 2 shows increased levels of transcripts involved in embryonic limb morphogen- esis, glucocorticoid signaling, and antiapoptosis, particularly at the 1 and 50 ppb doses. The immune response transcripts in cluster 4 are decreased mainly at the 50 ppb dose (Igj, Downloaded from https://academic.oup.com/toxsci/article/100/1/75/1624780 by guest on 29 September 2021 Igh-VJ558, Igk-V28). We observed a consistent pattern of st two different arsenic doses. decreased expression at all arsenic doses in the largest group, 0.01 ppb, control vs. 1 ppb, control vs. 50 ppb) or cluster 3. These genes are involved in a variety of processes including signal transduction, hemoglobin activity, glycolysis, 0.3 1 c4 transcription, apoptosis, and glycosylation. Arsenic exposure at 1 ppb strongly decreased Alas2, Siat8d, Agtrl, and Ear1 and increased expression of Zfp145, Fkbp5, and Hspa1b. 0.6