ToxSci Advance Access originally published online on November 2, 2006
Toxicological Sciences 2007 95(2):474-484; doi:10.1093/toxsci/kfl152
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Toxicogenomic Approach Revealed Hepatic Gene Expression Changes Mechanistically Linked to Drug-Induced Hemolytic Anemia
* Discovery Technologies 1, Discovery Research Laboratories, Shionogi and Co Ltd, 12-4, Sagisu 5-chome, Fukushima-ku, Osaka 553-0002, Japan
Drug Safety Evaluation, Developmental Research Laboratories, Shionogi and Co Ltd, 1-1, Futaba-cho 3-chome, Toyonaka, Osaka 561-0825, Japan
1 To whom correspondence should be addressed. Fax: (81) 6-6455-2099. E-mail: masatomo.rokushima{at}shionogi.co.jp.
Received October 11, 2006; accepted October 16, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
A variety of pharmaceutical compounds causes hemolytic anemia as a significant adverse effect and this toxicity restricts the clinical utility of these drugs. In this study, we applied microarray technology to investigate hepatic gene expression changes associated with drug-induced hemolytic anemia and to identify potential biomarker genes for this hematotoxicity. We treated female Sprague-Dawley rats with two hemolytic anemia-inducing compounds: phenylhydrazine and phenacetin. Hepatic gene expression profiles were obtained using a whole-genome oligonucleotide microarray with pooled RNA samples from individual rats within each dose group and analyzed in comparison with hepatic histopathology, hematology, and blood chemistry data. We identified a small subset of genes that were commonly deregulated in all the severe hemolytic conditions, some of which were considered to be involved in hepatic events characteristic of hemolytic anemia, such as hemoglobin biosynthesis, heme metabolism, and phagocytosis. Among them, we selected six upregulated genes as putative biomarkers, and their expression changes from microarray measurements were confirmed by quantitative real-time PCR using RNAs from individual animals. They were Alas2, beta-glo, Eraf, Hmox1, Lgals3, and Rhced. Expression patterns of all these genes showed high negative and positive correlation against erythrocyte counts and total bilirubin levels in circulation, respectively, suggesting that these genes may be the potential biomarkers for hemolytic anemia. These findings indicate that drug-induced hemolytic anemia may be detected based on hepatic changes in the expression of a subset of genes that are mechanistically linked to the hematotoxicity.
Key Words: toxicogenomics; microarray; hemolytic anemia; liver; biomarker.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Microarray technology, which allows global gene expression analysis at the transcript level, is now widely used in many stages of drug discovery and development processes, from target identification to preclinical studies (Gerhold et al., 2002
). The application of this technology to studying the adverse effects of chemicals or new drug candidates is referred to as "toxicogenomics" (Searfoss et al., 2005; Suter et al., 2004). Toxicogenomics has been expected as a powerful approach for elucidating mechanisms underlying toxicological endpoints and detecting subtle injury without overt phenotypes, both of which have been difficult using conventional toxicological methods. Indeed, the utility and feasibility of this approach have been demonstrated in a number of reports delineating chemical-induced toxicity, including hepatotoxicity (Heinloth et al., 2004; Steiner et al., 2004), nephrotoxicity (Amin et al., 2004; Thukral et al., 2005) and genotoxicity (Ellinger-Ziegelbauer et al., 2004), leading to finding some promising biomarkers, predicting toxicity, discriminating different classes of toxicants, and promoting more in-depth understanding of toxicity mechanisms. This suggested to us that toxicogenomics might also be applicable to studying drug-induced hemolytic anemia, a serious side effect that impedes the effective use of marketed drugs and the development of new drug candidates.
Hemolytic anemia is a common adverse effect that accompanies treatment of individuals with a variety of pro-oxidant drugs such as dapsone, phenylhydrazine (PHZ), and primaquine (Beutler, 1969). In this hematotoxicity, macrophages capture drug-damaged erythrocytes at an increased rate in mainly, but not limited to, the spleen, leading to a decreased number of erythrocytes and low hemoglobin (HGB) levels in the circulation. In addition, changes observed in hemolytic anemia also involve the elevation of indirect bilirubin levels and reticulocyte counts in the blood. The mechanism underlying hemolytic actions of several agents on red cells has long been studied, and it has been established that hemolytic injury is associated with oxidative stress within erythrocytes. This concept is supported by the fact that hemolytic damage is accompanied by the generation of reactive oxygen species (ROS), glutathione depletion, HGB oxidation, and Heinz body formation in red cells (Jollow and McMillan, 2001; Sivilotti, 2004). It has also been reported that hemolytic agents caused membrane lipid peroxidation and/or denaturation of cytoskeletal protein (Jollow and McMillan, 2001). Nevertheless, direct targets for ROS within cells and the signaling pathway that plays a central role in macrophage recognition remain to be elucidated (McMillan et al., 2005).
In this study, we selected two direct-acting hemolytic agents for examination: PHZ as a classical hematotoxicant, and phenacetin (PNT, p-ethoxyacetanilide) as a representative drug that has been utilized for therapy until recently. PHZ, an antipyretic drug, has long served as a model compound for the study of hemolysis. It is known to induce oxidative damage to erythrocytes, resulting in hemolysis in vivo and in vitro. Within red blood cells (RBCs), PHZ is thought to interact with HGB in an oxidation reaction, which generates a number of reactive intermediates, such as hydrogen peroxide and superoxide anion (Shetlar and Hill, 1985). After undergoing oxidative damage from PHZ, erythrocytes are recognized and phagocytized by macrophages, which are suggested to be mediated by immune-related molecules, including autologous IgG, Fc-, and lectin-like receptors (Horn et al., 1991). PNT is a prodrug of acetaminophen, a well-known nonsteroidal analgesic and antipyretic drug. In humans and experimental animals, several metabolites of PNT formed in vivo have been shown to provoke a variety of serious toxicities, including renal failure (Calder et al., 1973) and hemolysis (Boelsterli et al., 1983), which caused this drug to be withdrawn from the market. One of the metabolites, N-hydroxyphenetidine (PNOH), has been reported to have direct oxidative effects on the red cells and to mediate PNT-induced hemolytic anemia in rats (Jensen and Jollow, 1991).
We chose the liver for gene expression analysis because (1) adult erythrocytes, direct targets of hemolytic agents, are nonnucleated, making gene expression measurements practically impossible, (2) the liver is one of the organs where damaged erythrocytes are removed from circulation by resident macrophages (in the liver, these are Kupffer cells), which presumably causes perturbation in hepatic gene expression as secondary effects, and (3) it is preferable that whole-body toxic events can be detected by measuring gene expression changes in a single organ, with the liver being a promising target for this due to a relatively high incidence of drug-induced hepatotoxicity. In the present study, we examined hepatic gene expression changes caused by the administration of hemolytic agents in order to identify potential biomarker genes and to gain insight into the mechanisms behind the liver events accompanying the hematotoxicity.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Chemicals and materials.
PHZ, PNT, and methylcellulose were obtained from Wako Pure Chemical Industries (Osaka, Japan). Saline was from Otsuka Pharmaceutical Factory (Tokushima, Japan). QIAzol Lysis Reagent, MagAttract RNA Cell Mini M48 Kit, QuantiTect Probe RT-PCR Kit, QuantiTect SYBR Green RT-PCR Kit, and QuantiTect Primer Assays were from Qiagen (Valencia, CA). RNAlater was from Ambion (Austin, TX). Cyanine 3-CTP and cyanine 5-CTP were from PerkinElmer (Wellesley, MA). Low RNA Input Linear Amplification Kit and Whole Rat Genome Microarray were from Agilent Technologies (Palo Alto, CA). TaqMan Gene Expression Assays and TaqMan Endogenous Controls were from Applied Biosystems (Foster City, CA). EDTA-2K, heparin sodium, and pentobarbital sodium were from Terumo (Tokyo, Japan).
Animals and animal care.
Female Sprague-Dawley albino rats (Crl:CD(SD)) approximately 4 weeks old were purchased from Charles River Japan (Kanagawa, Japan) and maintained in a controlled environment (12-h light-dark cycle, 23 ± 3°C and 50 ± 20% relative humidity) in plastic cages. The animals were provided with certified rodent chow (Oriental Yeast, Tokyo, Japan) and water ad libitum and acclimatized under these conditions for 2 weeks before experimentation. Animal maintenance and treatment were performed according to the principles outlined in the Guide for the Care and Use of Laboratory Animals prepared by the Japanese Association for Laboratory Animal Science and our institution.
Study design and administration of chemicals.
Rats were randomly assigned to treatment and control groups (three rats/group/timepoint). PHZ was dissolved in saline and administered ip at 20 and 80 mg/kg per day. PNT was dissolved in 0.5 wt/vol% methylcellulose aqueous solution (0.5% MC) and orally dosed at 500 and 1000 mg/kg per day. The rats were administered the test compounds or the corresponding vehicle once for 1 day (24 h) or once daily for 4 days and then sacrificed 24 h after final administration. At necropsy, the animals were anesthetized with pentobarbital sodium. Blood sample was drawn from the posterior vena cava and put into two vacuum blood-collecting tubes containing EDTA-2K or heparin sodium for hematology or blood chemistry examination, respectively. Then, the rats were euthanized by exsanguination, and liver samples were collected. The median lobe of the liver was sliced and fixed in 10% neutral buffered formalin for histopathological evaluation. A section of the left lobe of the liver was chopped into smaller pieces, soaked in RNAlater, and stored at – 80°C until the RNA was isolated.
Histopathology.
The tissues fixed in 10% neutral buffered formalin were embedded in paraffin, sectioned at 3 µm, and stained with hematoxylin and eosin for histopathological evaluation. To characterize hemosiderin, liver section was additionally stained with Berlin blue and examined under a light microscopy.
Hematology and blood chemistry examination.
Whole blood treated with EDTA-2K was subjected to hematology analysis using an ADVIA 120 Hematology System (Bayer AG, Leverkusen, Germany). The analysis included RBC counts, hematocrit (HCT) values, and HGB concentrations. Plasma was obtained from the blood collected with heparin sodium and subjected to blood chemistry analysis. Total bilirubin (T-Bil) levels and activities of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were determined using an automatic analyzer 7170 (Hitachi, Tokyo, Japan).
RNA isolation and labeled cRNA preparation.
Total RNA was extracted from the left lobe of the liver using QIAzol Lysis Reagent and further purified using the BioRobot M48 Workstation (QIAGEN) in combination with MagAttract RNA Cell Mini M48 Kit, including a DNase digestion step according to the manufacturer's protocol. RNA concentration was determined by absorbance at 260 nm, and RNA quality was checked using Agilent 2100 Bioanalyzer (Agilent Technologies). Equal amounts of total RNA from individual rats within each dose group (n = 3) were pooled and used as templates for the microarray analysis.
Labeled cRNA targets were prepared using Low RNA Input Linear Amplification Kit following procedures recommended by the manufacturer. Briefly, 500 ng of total RNA mixture was reverse transcribed with oligo-dT primer containing T7 promoter sequence. Using the resulting cDNA as templates, labeled cRNA for chemical-treated groups and control groups were synthesized by in vitro transcription in the presence of cyanine 5-CTP and cyanine 3-CTP, respectively. Labeled cRNAs were purified and quantified by absorbance at 260 nm.
Microarray analysis.
Cyanine 5-labeled cRNA from chemical-treated animals was combined with cyanine 3-labeled cRNA from vehicle- and time-matched control animals. The cRNA mixture was applied to an oligonucleotide microarray (Agilent Technologies, Whole Rat Genome, G4131A), and hybridization was carried out at 60°C for 17 h according to the manufacturer's instructions. Following hybridization, the slide was washed, dried, and scanned on an Agilent DNA microarray scanner (Agilent Technologies). Fluorescence image was quantified using Feature Extraction software (Agilent Technologies). The expression value for each spot was determined by subtracting the local background signal from the spot area signal.
Locally weighted scatterplot smoothing (LOWESS) normalization, ratio (treated/control) calculation, and two-way hierarchical clustering were conducted using GeneSpring software version 7.0 (Agilent Technologies). Genes were considered as upregulated when they had a ratio of over 1.5 and as downregulated if they had a ratio of under 0.67. In the two-way hierarchical clustering analysis, standard correlation and Pearson's correlation were used for grouping genes and conditions, respectively.
Quantitative real-time PCR.
Expression levels of rat aminolevulinic acid synthase 2 (Alas2), beta-glo (MGC72973), erythroid-associated factor (Eraf), heme oxygenase 1 (Hmox1), "lectin, galactose-binding, soluble 3" (Lgals3), Rhesus blood group CE and D (Rhced), and glyceraldehyde-3-phosphate dehydrogenase (Gapd) were measured by one-step quantitative real-time PCR (RT-PCR). Forty nanogram of the total RNA from each aliquot was used as the template in each reaction, and TaqMan Gene Expression Assays (for Alas2, Assay ID Rn00566201_m1; beta-glo, Assay ID Rn02396921_g1; Hmox1, Assay ID Rn01536933_m1; Lgals3, Assay ID Rn00582910_m1; and Rhced, Assay ID Rn00574568_m1) or TaqMan Endogenous Controls (for Gapd) were employed as sets of gene-specific probe and primer pair. QuantiTect Primer Assay was used as the gene-specific primer pair for Eraf (Cat. No. QT00376117). Quantitative RT-PCR was performed using QuantiTect Probe RT-PCR Kit (for Alas2, beta-glo, Gapd, Hmox1, Lgals3, and Rhced) or QuantiTect SYBR Green RT-PCR Kit (for Eraf), and the transcript level was quantitated with GeneAmp 5700 Sequence Detection System (Applied Biosystems), according to the manufacturer's protocols. The resulting cycle threshold (Ct) value was processed based on the comparative Ct method (Livak and Schmittgen, 2001), where Gapd was used as an endogenous reference gene to normalize the expression level of target genes.
Statistics.
For hematology and blood chemistry, data are reported as mean ± SD, and differences between control and treated groups were evaluated by Dunnett multiple comparison test. Correlations between hematology, blood chemistry, and quantitative RT-PCR data were assessed based on Pearson and Spearman rank correlation coefficient. All statistical analyses except for microarray data were conducted using SAS software.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Histopathology
Female Sprague-Dawley rats were injected with two different doses of two hemolytic agents, PHZ and PNT, once for 1 day (24 h) or once daily for 4 days to provoke hemolytic anemia, as described in the "Materials and Methods." To confirm the incidence of hemolytic changes in the livers of the chemical-treated animals, the liver was histopathologically examined. Hepatic lesions associated with hemolytic anemia are summarized in Table 1. Data for the daily high dose (80 mg/kg/day) of PHZ for 4 days are not available due to mortality. There were no lesions in the livers of rats dosed with each of the two vehicles (saline and 0.5% MC) for any of the exposure periods (data not shown). In the PHZ treatment groups, erythrophagocytosis by Kupffer cells was observed in the sinusoid at both dose levels (20 and 80 mg/kg/day) for 24 h and at the low dose (20 mg/kg/day) for 4 days, with the severity being dose and time related. On the other hand, PNT induced mild to moderate erythrophagocytosis at the high doses (1000 mg/kg/day) given for both 24 h and 4 days. Hemosiderin deposition was presented in Kupffer cells at the low dose of PHZ for 4 days and at the high dose of PNT for 4 days. PHZ also provoked extramedullary hematopoiesis at the high dose for 24 h and the low dose for 4 days. Thus, hepatic lesions associated with hemolytic anemia, as described previously, occurred following treatment with the tested compounds (Bruckner et al., 1989
; Mellert et al., 2004; Redondo et al., 1995).
|
Hematology and Blood Chemistry Examination
RBC, HGB, HCT, and T-Bil following the treatment with PHZ and PNT are shown in Figure 1. Compared to the time- and vehicle-matched controls, significant decreases in RBC, HGB, and HCT were observed after the treatment with PHZ at both dose levels for 24 h in a dose-dependent manner and at the low dose for 4 days. PHZ also provoked an increase in T-Bil at the high dose for 24 h and at the low dose for 4 days. As for PNT, only the repeated high dose for 4 days caused significant decreases in RBC, HGB, and HCT and an increase in T-Bil. There were slight elevations in both AST and ALT or AST alone at the high dose of PHZ for 24 h, both dose levels of PNT for 24 h, and the high dose of PNT for 4 days, indicating that mild liver injuries occurred under these conditions (data not shown). Together with histopathological observations, the high dose of PHZ for 24 h, the low dose of PHZ for 4 days, and the high dose of PNT for 4 days were considered to have induced significant hemolytic anemia. The severity order of the hematotoxicity was as follows: PHZ, low dose, 4 days > PHZ, high dose, 24 h > PNT, high dose, 4 days.
|
Microarray Analysis
Global gene expression changes in the livers of rats treated with two hemolytic agents were examined using whole-genome oligonucleotide microarrays. First, we determined differentially expressed genes under the three conditions that caused severe hemolytic anemia: PHZ, high dose, 24 h; PHZ, low dose, 4 days; and PNT, high dose, 4 days. As a criterion for differential expression, 1.5-fold change compared to the control was employed. Based on this threshold, the high dose of PHZ for 24 h had 904 and 893 probes that were up- and downregulated, respectively. The PHZ low dose for 4 days had 423 upregulated and 131 downregulated probes. The repeated high dose of PNT for 4 days had a total of 1035 differentially expressed probes by greater than 1.5-fold, with 450 upregulated and 585 downregulated. These results indicate that the doses of the two hemolytic agents had significant influence on hepatic gene expression. Secondly, we extracted probes that were up- or downregulated under all the three conditions, where the direction of regulation was considered, in order to obtain probes whose expression levels were characteristically deregulated under hemolytic conditions (Fig. 2). Unexpectedly, of the 1550 probes that were included in either of the upregulated probe lists, only 35 probes (2.3%) met these criteria (Fig. 2A). Similarly, among the 1501 probes that were included in any of the downregulated probe lists, no more than 10 probes (0.7%) were repressed under all the three conditions (Fig. 2B). The relatively small number of overlapping probes suggests that most of the gene expression changes obtained with the microarray resulted from direct impact of the agents (and/or their metabolites) on the liver rather than from the secondary effects of hemolysis. This is also supported by the observation that the low dose of PHZ for 4 days, which induced the severest hemolysis, had the smallest number of differentially expressed probes of the three conditions.
|
The probes that were induced or repressed under all the three hemolytic conditions are represented in Table 2. Because all the probes listed in Table 2 had no redundant gene annotations among each other, we use the term "gene" in stead of "probe" below. In the table, genes were categorized based on their representative biological processes or pathways. Predominant functional categories of upregulated genes included cell adhesion, HGB biosynthesis, immune response, inflammatory response, oxidative stress, and transport. At the low dose of PHZ for 4 days, which provoked the severest hemolysis, the most upregulated gene in Table 2 was erythroid-associated factor (Eraf, AHSP) (11.4-fold), followed by "lectin, galactose-binding, soluble 3" (Lgals3, galectin 3) (9.4-fold). In the case of the repeated high dose of PNT for 4 days, metallothionein 1a (Mt1a) was the most induced one (8.4-fold). In contrast to the upregulated gene list, the downregulated gene list had no prominent functional categories.
|
Two-way hierarchical clustering analysis were carried out for the 45 genes listed in Table 2 and seven dosing conditions examined, where genes and conditions were grouped based solely on the similarity of the expression patterns (Fig. 3). In the dendrogram for genes, there were several distinct clusters/groups that were enriched in genes of certain functional categories. For example, one cluster that includes aminolevulinic acid synthase 2 (Alas2, ALAS-E) and beta-glo (MGC72973), encoding HGB beta-chain, consisted of those generally associated with HGB biosynthesis or heme metabolism. Another cluster including Lgals3 and vascular cell adhesion molecule 1 (Vcam1) had several members that are involved in immune response or cell adhesion. Additionally, there was a cluster encompassing a group of genes associated with general or oxidative stress response. The cluster included Mt1a and heme oxygenase (decycling) 1 (Hmox1, HO-1).
|
Among the three clusters mentioned above, the two former ones included genes whose expression changes showed relatively high negative correlation with average RBC within each dose group (Figs. 1 and 3), suggesting that some of the members in these clusters could be used as potential biomarkers for hemolytic anemia. Hence, together with the putative functional association with hemolysis (described below) and the magnitude of expression changes, we selected five candidate genes within these two clusters. They were Alas2, beta-glo, Eraf, Lgals3, and "Rhesus blood group CE and D" (Rhced). In addition, Hmox1 was also chosen because of its apparent association with hemolysis; namely, Hmox1 encodes the rate-limiting enzyme in heme degradation that leads to increased levels of T-Bil in the circulation (Choi and Alam, 1996
; Maines and Veltman, 1984). These six genes, all of which were upregulated under the hemolytic conditions, were further evaluated as biomarker candidates using quantitative RT-PCR.
Confirmation of Gene Expression Changes by Quantitative RT-PCR
We employed quantitative RT-PCR to confirm the changes in transcript levels of the selected genes identified by microarray analysis: Alas2, beta-glo, Eraf, Hmox1, Lgals3, and Rhced. In quantitative RT-PCR, liver RNA samples from individual animals were used as templates in order to perform individual-level comparison of changes between gene expression and blood parameters (e.g., RBC and T-Bil), which allows more accurate evaluation of marker candidate genes. Comparisons of microarray and quantitative RT-PCR data for these genes are presented in Figure 4. Overall expression changes obtained with microarrays were in good agreement with those measured by quantitative RT-PCR, even though most of the ratios from microarrays were lower than those from quantitative RT-PCR. This ratio compression, noted in several reports delineating similar comparisons (Ellinger-Ziegelbauer et al., 2004; Hamadeh et al., 2002), may be due to technical problems such as nonspecific cross-hybridization in the microarray experiment or lower PCR amplification efficiency than theoretically expected in the comparative Ct method, where DNA strands were considered to be doubled in each cycle. However, with collective comparisons for all selected genes, we obtained a Pearson correlation coefficient of 0.89 between base-2 log ratios from microarray and quantitative RT-PCR. In addition, consistent with microarray measurements, Eraf was the most upregulated gene at the low dose of PHZ for 4 days among the six. Thus, in spite of the presence of ratio compression, quantitative RT-PCR data strongly supported the expression change of selected genes from microarray measurement.
|
Correlations between Marker Candidate Genes and RBC or T-Bil
We calculated Pearson and Spearman rank correlation coefficients between the expression data of the six selected genes (Alas2, beta-glo, Eraf, Hmox1, Lgals3, and Rhced) and RBC or T-Bil.
Ct (CtGapd – Cttarget) values obtained by quantitative RT-PCR from individual rats were employed as the values for gene expression levels. As shown in Table 3, the expression levels of all the selected genes showed high negative correlation with RBC (p < 0.001 for all selected genes, data not shown), and high positive correlation with T-Bil (p < 0.001 for all selected genes, data not shown), with a tendency of showing larger absolute values for the Pearson than for the Spearman correlation. Of the six genes, beta-glo and Lgals3 showed the highest inverse Spearman and Pearson correlation with RBC, respectively. As for T-Bil, Hmox1 showed the strongest Spearman correlation and beta-glo represented the highest Pearson correlation. Scatter plots for these comparisons are presented online as supplementary data (http://toxsci.oxfordjournals.org/).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The present study was undertaken to examine whether drug-induced hemolytic anemia leads to hepatic expression changes of genes involved in common biological processes or pathways, which could enable identification of biomarker genes and better understandings of the liver events associated with the hematotoxicity. Therefore, we treated rats with two hemolytic agents, PHZ and PNT, and examined hepatic gene expression changes using oligonucleotide microarrays containing all genes on rat genome. Based on the concurrently conducted histopathological, hematological, and blood chemistry analyses, we judged that the following three dosing conditions had induced severe hemolytic anemia: PHZ, high dose (80 mg/kg), 24 h; PHZ, low dose (20 mg/kg/day), 4 days; and PNT, high dose (1000 mg/kg/day), 4 days. Each of these hemolytic conditions had more than 500 differentially expressed genes (probes), but we identified only 45 genes that were commonly deregulated under all the three conditions. Nevertheless, among them, there were several subsets of genes with certain functional categories, including cell adhesion, HGB biosynthesis, and immune response (Table 2). Furthermore, some of the genes with common functional categories formed a definitive cluster in hierarchical clustering analysis (Fig. 3). These results suggest that there existed common cellular pathways or processes that were affected in response to hemolysis caused by the two different kinds of agents. We therefore explored the potential links between the deregulation of some of these genes and hemolytic events in the liver. The most plausible associations are schematically depicted in Figure 5.
|
Phagocytosis
We found upregulation of two genes (Lgals3 and Fcgr2b) involved in phagocytosis, a key process in immune response. Lgals3, a ß-galactoside–binding lectin, was induced by greater than threefold under all three hemolytic conditions (Table 2). Lgals3 has roles in a variety of cellular processes such as immune reactions and development and is overexpressed in phagocytic macrophages (Dumic et al., 2006
). On the other hand, Fcgr2b, which encodes type II Fc
receptor, showed mild upregulation (1.5- to 1.7-fold). This gene is reported to be involved in the recognition and phagocytosis of PHZ-oxidized RBCs by macrophages (Horn et al., 1991). Interestingly, Sano et al. have reported that knockout mice deficient in Lgals3 gene (Lgals3–/–) exhibited decreased IgG-mediated erythrophagocytosis by Kupffer cells in an autoimmune hemolytic anemia model. Furthermore, they also showed that Fc receptor–mediated actin polymerization, a cytoskeletal change accompanying phagocytosis, was also impaired in Lgals3–/– macrophages (Sano et al., 2003). Hence, the induction of these two genes may be related to the enhancement of erythrophagocytosis that we observed under the three hemolytic conditions by histopathological analysis.
Erythropoiesis
Five induced genes (Alas2, beta-glo, Eraf, Hbe2, and Hmox1) in Table 2 are involved in HGB biosynthesis or heme metabolism, and four of them (Alas2, beta-glo, Eraf, and Hbe2) formed a single cluster along with a blood group gene, Rhced, in the hierarchical clustering analysis (Fig. 3). Since both cellular processes play critical roles in erythropoiesis, concerted upregulation of these genes probably reflects the activation of erythropoiesis in the liver which was observed as a hepatic lesion of extramedullary hematopoiesis. Of particular note is that, among the five genes, the expression of four (Alas2, beta-glo, Hbe2, and Hmox1) is known to be induced by heme, a degradation product of HGB (Fujita et al., 1991; Fukuda et al., 1994; Shibahara et al., 1993). The upregulation of these genes in our experiment may therefore be driven to a large extent by a suspected increase in heme levels resulting from phagocytosis of the damaged erythrocytes by Kupffer cells (Fig. 5). From a mechanistical point of view, it has been reported that heme inhibits DNA-binding activity of a repressor molecule Bach1, which under low-heme conditions binds to Maf recognition elements on the regulatory regions of Hmox1 and ß-globin locus genes (beta-glo and Hbe2), thereby repressing transcription (Igarashi and Sun, 2006).
Eraf and Rhced clustered with the majority of putatively heme-regulated genes in hierarchical clustering (Fig. 3). Eraf encodes 
-HGB–stabilizing protein that functions as a molecular chaperone for free -HGB and prevents its precipitation in erythrocytes. Deficiency of this gene (Eraf–/–) leads to hemolysis with Heinz body formation in mice (Kong et al., 2004). On the other hand, Rhced is known to produce antigens on the surface of erythrocytes (Kumada et al., 2002). Thus, both genes significantly participate in erythropoiesis as well as the putatively heme-regulated genes. Intriguingly, it has been recently reported that, in addition to ß-globin genes, Eraf and Rhced plus Alas2 are potential targets of erythroid Kruppel-like factor (EKLF, KLF1), a transcription activator that regulates global erythroid gene expression, suggesting that EKLF might have played a role in the expression changes of these erythropoiesis genes in our experiment (Hodge et al., 2006).
Miscellaneous
There are three upregulated genes in Table 2 that are assigned to cell adhesion category. In addition, Lgals3 and "glycoprotein (transmembrane) nmb" (Gpnmb) in Table 2 are also suggested to be involved in cell adhesion (Dumic et al., 2006; Haralanova-Ilieva et al., 2005), both of which were induced in response to hemolysis. Interestingly, four of these five genes (Ceacam10, Gpnmb, Lgals3, and Vcam1) belonged to the same group in the clustering analysis (Fig. 3). These results may indicate that enhancement of cell adhesion activity occurred in some types of hepatic cell populations as a result of a hemolytic response, although it is difficult to find a plausible explanation for the upregulation of these genes. In the case of Vcam1, however, this gene has been reported to be inducible by heme in vascular endothelial cells as well as other adhesion molecules, suggesting that the Vcam1 induction we observed may also be a result of a putative increase in heme levels (Wagener et al., 1997).
Of the transporter genes listed in Table 2, "solute carrier family 39 (iron-regulated transporter), member 1" (Slc40a1) encodes a multiple transmembrane domain protein, known as ferroportin 1, which acts as a cellular iron exporter in the liver (Anderson and Frazer, 2005). In humans, functional loss of Slc40a1 by mutation is reported to cause impaired iron homeostasis, resulting in iron overload and hemosiderin deposition in Kupffer cells (Zoller et al., 2005). Accordingly, the induction of Slc40a1 that we found may reflect a cellular effort to decrease the enhanced levels of intracellular iron, which was confirmed as hemosiderin deposition by histopathological analysis (Table 1) and was probably derived from heme degradation by Hmox1 (Fig. 5).
Anemic Marker Candidates
Based on the putative functional association with hemolytic anemia and the magnitude of expression changes, six biomarker candidate genes were selected from microarray analysis (Alas2, beta-glo, Eraf, Hmox1, Lgals3, and Rhced). Then, their expression changes were confirmed by quantitative RT-PCR using RNA samples from individual rats. Individual-level comparisons between quantitative RT-PCR data and RBC or T-Bil confirmed that expression of all selected genes showed high correlation against the severity of hemolysis (Table 3), indicating that these genes may be utilized as biomarkers for hemolytic anemia. Among them, beta-glo, encoding HGB beta-chain, exhibited the highest correlation with both RBC (Spearman) and T-Bil (Pearson), suggesting that this gene is one of the most promising marker candidates. In addition, Lgals3 and Hmox1 are also potential candidates in that they showed the strongest negative Pearsons correlation and positive Spearmans correlation against RBC and T-Bil, respectively. However, it should be noted that some of the selected genes are likely to be upregulated by other stimuli or stresses, which leads to false positive detection. Indeed, Hmox1 is known to be induced in response not only to its substrate heme, but also to various stimuli, including oxidative stress and heavy metals (Choi and Alam, 1996). Similarly, we cannot exclude the possibility that Lgals3 would be inducible in the liver when compounds cause hepatocyte necrosis because this gene is suggested to have a role in phagocytosis of damaged cells (Dumic et al., 2006). In such cases, simultaneous evaluation of collective expression changes of marker candidate genes may allow more accurate prediction because in the two-way hierarchical clustering, the three dosing conditions that provoked severe hemolysis were tightly clustered among the dosing conditions tested (Fig. 3). Additionally, prediction models such as support vector machine, a binary classifier, seem to be useful since they have been successfully applied in cases of chemical-induced nephrotoxicity (Thukral et al., 2005) and hepatotoxicity (Steiner et al., 2004). In this study, we could not test this possibility because of the shortage of data for building a prediction model and evaluating it. When sufficient hepatic gene expression profiles for other types of direct-acting hemolytic agents become available, we hope to test the feasibility of the approach.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
In summary, drug-induced hemolytic anemia led to hepatic expression changes of a subset of genes associated with specific cellular pathways: HGB biosynthesis, heme metabolism, cell adhesion, and immune response. Expression patterns of several upregulated genes, including beta-glo and Lgals3, showed high negative and positive correlation against RBC and T-Bil in blood, respectively, suggesting that these genes can be used as potential hepatic biomarkers for hemolytic anemia.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
ACKNOWLEDGMENTS |
|---|
The authors gratefully acknowledge the toxicogenomics project members in our laboratories for their invaluable support and scientific advice throughout the experiments. We thank Ayumi Ojima for her excellent technical assistance. We also wish to thank Toru Yanagimoto and Drs Satoshi Inoue and Mikinori Torii for their helpful suggestions, discussions, and encouragement.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Amin RP, Vickers AE, Sistare F, Thompson KL, Roman RJ, Lawton M, Kramer J, Hamadeh HK, Collins J, Grissom S, et al. (2004) Identification of putative gene based markers of renal toxicity. Environ. Health Perspect. 112:465–479.[ISI][Medline]
Anderson GJ and Frazer DM. (2005) Hepatic iron metabolism. Semin. Liver Dis. 25:420–432.[CrossRef][ISI][Medline]
Beutler E. (1969) Drug-induced hemolytic anemia. Pharmacol. Rev. 21:73–103.
Boelsterli UA, Shie KP, Brandle E, Zbinden G. (1983) Toxicological screening models: drug-induced oxidative hemolysis. Toxicol. Lett. 15:153–158.[CrossRef][ISI][Medline]
Bruckner JV, MacKenzie WF, Ramanathan R, Muralidhara S, Kim HJ, Dallas CE. (1989) Oral toxicity of 1,2-dichloropropane: acute, short-term, and long-term studies in rats. Fundam. Appl. Toxicol. 12:713–730.[CrossRef][ISI][Medline]
Calder IC, Creek MJ, Williams PJ, Funder CC, Green CR, Ham KN, Tange JD. (1973) N-hydroxylation of p-acetophenetidide as a factor in nephrotoxicity. J. Med. Chem. 16:499–502.[CrossRef][ISI][Medline]
Choi AM and Alam J. (1996) Heme oxygenase-1: Function, regulation, and implication of a novel stress-inducible protein in oxidant-induced lung injury. Am. J. Respir. Cell Mol. Biol. 15:9–19.[Abstract]
Dumic J, Dabelic S, Flogel M. (2006) Galectin-3: An open-ended story. Biochim. Biophys. Acta 1760:616–635.[Medline]
Ellinger-Ziegelbauer H, Stuart B, Wahle B, Bomann W, Ahr HJ. (2004) Characteristic expression profiles induced by genotoxic carcinogens in rat liver. Toxicol. Sci. 77:19–34.
Fujita H, Yamamoto M, Yamagami T, Hayashi N, Sassa S. (1991) Erythroleukemia differentiation. Distinctive responses of the erythroid-specific and the nonspecific delta-aminolevulinate synthase mRNA. J. Biol. Chem. 266:17494–17502.
Fukuda Y, Fujita H, Garbaczewski L, Sassa S. (1994) Regulation of beta-globin mRNA accumulation by heme in dimethyl sulfoxide (DMSO)-sensitive and DMSO-resistant murine erythroleukemia cells. Blood 83:1662–1667.
Gerhold DL, Jensen RV, Gullans SR. (2002) Better therapeutics through microarrays. Nat. Genet. 32:Suppl., 547–551.
Hamadeh HK, Bushel PR, Jayadev S, Martin K, DiSorbo O, Sieber S, Bennett L, Tennant R, Stoll R, Barrett JC, et al. (2002) Gene expression analysis reveals chemical-specific profiles. Toxicol. Sci. 67:219–231.
Haralanova-Ilieva B, Ramadori G, Armbrust T. (2005) Expression of osteoactivin in rat and human liver and isolated rat liver cells. J. Hepatol. 42:565–572.[CrossRef][ISI][Medline]
Heinloth AN, Irwin RD, Boorman GA, Nettesheim P, Fannin RD, Sieber SO, Snell ML, Tucker CJ, Li L, Travlos GS, et al. (2004) Gene expression profiling of rat livers reveals indicators of potential adverse effects. Toxicol. Sci. 80:193–202.
Hodge D, Coghill E, Keys J, Maguire T, Hartmann B, McDowall A, Weiss M, Grimmond S, Perkins A. (2006) A global role for EKLF in definitive and primitive erythropoiesis. Blood 107:3359–3370.
Horn S, Bashan N, Gopas J. (1991) Phagocytosis of phenylhydrazine oxidized and G-6-PD-deficient red blood cells: The role of cell-bound immunoglobulins. Blood 78:1818–1825.
Igarashi K and Sun J. (2006) The heme-Bach1 pathway in the regulation of oxidative stress response and erythroid differentiation. Antioxid. Redox Signal. 8:107–118.[CrossRef][ISI][Medline]
Jensen CB and Jollow DJ. (1991) The role of N-hydroxyphenetidine in phenacetin-induced hemolytic anemia. Toxicol. Appl. Pharmacol. 111:1–12.[CrossRef][ISI][Medline]
Jollow DJ and McMillan DC. (2001) Oxidative stress, glucose-6-phosphate dehydrogenase and the red cell. Adv. Exp. Med. Biol. 500:595–605.[ISI][Medline]
Kong Y, Zhou S, Kihm AJ, Katein AM, Yu X, Gell DA, Mackay JP, Adachi K, Foster-Brown L, Louden CS, et al. (2004) Loss of alpha-hemoglobin-stabilizing protein impairs erythropoiesis and exacerbates beta-thalassemia. J. Clin. Invest. 114:1457–1466.[CrossRef][ISI][Medline]
Kumada M, Iwamoto S, Kamesaki T, Okuda H, Kajii E. (2002) Entire sequence of a mouse chromosomal segment containing the gene Rhced and a comparative analysis of the homologous human sequence. Gene 299:165–172.[CrossRef][ISI][Medline]
Livak KJ and Schmittgen TD. (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25:402–408.[CrossRef][ISI][Medline]
Maines MD and Veltman JC. (1984) Phenylhydrazine-mediated induction of haem oxygenase activity in rat liver and kidney and development of hyperbilirubinaemia. Inhibition by zinc-protoporphyrin. Biochem. J. 217:409–417.[ISI][Medline]
McMillan DC, Powell CL, Bowman ZS, Morrow JD, Jollow DJ. (2005) Lipids versus proteins as major targets of pro-oxidant, direct-acting hemolytic agents. Toxicol. Sci. 88:274–283.
Mellert W, Deckardt K, Gembardt C, Zwirner-Baier I, Jackh R, van Ravenzwaay B. (2004) Aniline: Early indicators of toxicity in male rats and their relevance to spleen carcinogenicity. Hum. Exp. Toxicol. 23:379–389.
Redondo PA, Alvarez AI, Diez C, Fernandez-Rojo F, Prieto JG. (1995) Physiological response to experimentally induced anemia in rats: A comparative study. Lab. Anim. Sci. 45:578–583.[ISI][Medline]
Sano H, Hsu DK, Apgar JR, Yu L, Sharma BB, Kuwabara I, Izui S, Liu FT. (2003) Critical role of galectin-3 in phagocytosis by macrophages. J. Clin. Invest. 112:389–397.[CrossRef][ISI][Medline]
Searfoss GH, Ryan TP, Jolly RA. (2005) The role of transcriptome analysis in pre-clinical toxicology. Curr. Mol. Med. 5:53–64.[CrossRef][ISI][Medline]
Shetlar MD and Hill HA. (1985) Reactions of hemoglobin with phenylhydrazine: A review of selected aspects. Environ. Health Perspect. 64:265–281.[ISI][Medline]
Shibahara S, Yoshizawa M, Suzuki H, Takeda K, Meguro K, Endo K. (1993) Functional analysis of cDNAs for two types of human heme oxygenase and evidence for their separate regulation. J. Biochem. (Tokyo) 113:214–218.
Sivilotti ML. (2004) Oxidant stress and haemolysis of the human erythrocyte. Toxicol. Rev. 23:169–188.[CrossRef][Medline]
Steiner G, Suter L, Boess F, Gasser R, de Vera MC, Albertini S, Ruepp S. (2004) Discriminating different classes of toxicants by transcript profiling. Environ. Health Perspect. 112:1236–1248.[ISI][Medline]
Suter L, Babiss LE, Wheeldon EB. (2004) Toxicogenomics in predictive toxicology in drug development. Chem. Biol. 11:161–171.[CrossRef][ISI][Medline]
Thukral SK, Nordone PJ, Hu R, Sullivan L, Galambos E, Fitzpatrick VD, Healy L, Bass MB, Cosenza ME, Afshari CA. (2005) Prediction of nephrotoxicant action and identification of candidate toxicity-related biomarkers. Toxicol. Pathol. 33:343–355.[CrossRef][ISI][Medline]
Wagener Feldman E., de Witte T, Abraham NG. (1997) Heme induces the expression of adhesion molecules ICAM-1, VCAM-1, and E selectin in vascular endothelial cells. Proc. Soc. Exp. Biol. Med. 216:456–463.[Abstract]
Zoller H, McFarlane I, Theurl I, Stadlmann S, Nemeth E, Oxley D, Ganz T, Halsall DJ, Cox TM, Vogel W. (2005) Primary iron overload with inappropriate hepcidin expression in V162del ferroportin disease. Hepatology 42:466–472.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Rokushima, K. Omi, K. Imura, A. Araki, N. Furukawa, F. Itoh, M. Miyazaki, J. Yamamoto, M. Rokushima, M. Okada, et al. Toxicogenomics of Drug-Induced Hemolytic Anemia by Analyzing Gene Expression Profiles in the Spleen Toxicol. Sci., November 1, 2007; 100(1): 290 - 302. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||