ToxSci Advance Access originally published online on August 13, 2007
Toxicological Sciences 2007 100(1):290-302; doi:10.1093/toxsci/kfm216
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Toxicogenomics of Drug-Induced Hemolytic Anemia by Analyzing Gene Expression Profiles in the Spleen
* Discovery Technologies 1, Discovery Research Laboratories, Shionogi & Co., Ltd, 12-4, Sagisu 5-chome, Fukushima-ku, Osaka 553-0002, Japan
Drug Safety Evaluation, Developmental Research Laboratories, Shionogi & 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 July 1, 2007; accepted August 6, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Hemolytic anemia is a serious adverse effect of therapeutic drugs that is caused by increased destruction of drug-damaged erythrocytes by macrophages in the spleen and liver. We previously applied a toxicogenomic approach to the toxicity by analyzing microarray data of the liver of rats dosed with two hemolytic agents: phenylhydrazine and phenacetin. In the present study, we analyzed gene expression profiles in the spleen, the primary organ for destruction of damaged erythrocytes, of the same models in order to identify splenic gene expression alterations that could be used to predict the hematotoxicity. Microarray analyses revealed hundreds of genes commonly deregulated under all severe hemolytic conditions, which included genes related to splenic events characteristic of the hematotoxicity, such as proteolysis and iron metabolism. Eleven upregulated genes were selected as biomarker candidates, and their expression changes were validated by quantitative real-time PCR. The transcript levels of most of these genes showed strong correlation with the results of classical toxicological assays (e.g., histopathology and hematology). Furthermore, hierarchical clustering analysis suggested that altered expression patterns of the 11 genes sensitively reflected the erythrocyte damage even under a condition that caused no decrease in erythrocyte counts. Among the selected genes, heme oxygenase 1 was one of the most promising biomarker candidates, the upregulation of which on the protein level was confirmed by immunohistochemistry. These results indicate that altered splenic expression of a subset of genes may allow detection of drug-induced hemolytic anemia, with better sensitivity than that of erythrocyte counts in the blood.
Key Words: toxicogenomics; hemolytic anemia; spleen; biomarker.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Toxicogenomics is an emerging discipline that integrates toxicology and global gene expression analysis by microarray technology, where altered transcript levels in tissues or cells upon exposure to compounds are analyzed in association with conventional toxicological endpoints (Paules, 2003
). The major goals and objectives of toxicogenomics are to elucidate the molecular mechanisms of compound-induced toxicity and to identify novel biomarkers that would enable more effective safety assessment of chemicals or new drug candidates (Suter et al., 2004). Indeed, this approach has been successfully applied to the study of several types of toxicities, including hepatotoxicity (Heinloth et al., 2004), nephrotoxicity (Amin et al., 2004), and genotoxicity (Ellinger-Ziegelbauer et al., 2004). Such studies have led to the identification of gene expression signatures that are likely to be indicative or predictive of cellular injuries and also to better understand the molecular events underlying the toxicities. Thus, toxicogenomics has the potential to generate valuable information on various types of chemical toxicities, as long as they are accompanied by specific gene expression changes in specific organs or cells, which may not necessarily be the direct target of the chemicals.
Hemolytic anemia is a serious adverse effect that prevents the effective use of various pro-oxidant drugs, such as dapsone and primaquine (Beutler, 1969). It is caused by an accelerated removal of drug-damaged erythrocytes by resident macrophages in the spleen and liver, resulting in decreased red blood cell (RBC) counts and low hemoglobin (HGB) levels in the circulation. In addition, changes characteristic of this toxicity involve increases in indirect bilirubin levels and reticulocyte counts in the blood, the latter resulting from compensatory activation of erythropoiesis in the bone marrow. From a mechanistic point of view, the injury of erythrocytes by hemolytic agents is considered to be associated with oxidative stress within the cells (Sivilotti, 2004). After undergoing oxidative damage, RBCs are selectively recognized and phagocytized by tissue macrophages, which is followed by proteolytic digestion of HGB, the major component of erythrocytes. One of the breakdown products of HGB, heme, is further degraded into three bioactive products: iron (Fe2+), carbon monoxide, and bilirubin. The unconjugated form of bilirubin is subsequently released into the circulation, leading to enhanced indirect bilirubin levels in the blood. Alternatively, enhanced formation of free iron from heme degradation provokes iron overload within the cells and is sometimes observed as hemosiderin deposition (Fujitani et al., 2004).
In a previous study, we applied a toxicogenomic approach to the study of drug-induced hemolytic anemia by investigating hepatic gene expression changes in rats, where two direct-acting hemolytic drugs were chosen for examination: phenylhydrazine (PHZ) as a classical hematotoxicant and phenacetin (PNT) as a representative withdrawn drug (Rokushima et al., 2007). As a result, we were able to identify a small subset of genes commonly deregulated under all severe hemolytic conditions, some of which were considered to be mechanistically linked to the hemolytic events in the liver. Furthermore, several upregulated genes under the hemolytic conditions seemed to be potential biomarkers for the hematotoxicity.
The spleen is comprised of two histologically and functionally distinct organs; one is a phagocytic organ, the red pulp, and the other is an immune organ, the white pulp (Cesta, 2006). The red pulp consists of venous sinuses and splenic cords, where macrophages and granulocytes reside, and are engaged in the sensitive removal of xenobiotics and senescent or damaged RBCs from the circulation. The red pulp also acts as a storage reservoir for platelets and as a site of hematopoiesis. On the other hand, the white pulp is a lymphatic tissue surrounding the branches of the splenic artery. It contains lymphocytes such as B cells and T cells, which are responsible for the immunological function.
Considering its central role in the sequestration and destruction of drug-damaged erythrocytes, the spleen is expected to be affected more in response to hemolysis than the liver at the transcript levels, which suggested to us that it may also be a promising target for a toxicogenomic study of the hematotoxicity. We therefore analyzed gene expression profiles in the spleen, which is not a direct target of hemolytic agents, of the same hemolytic anemia models that were used to generate microarray data for the previous report. The purposes of the present study were to identify splenic gene expression changes that are indicative or predictive of drug-induced hemolytic anemia and to gain greater insight into the molecular mechanisms of the splenic events accompanying the hematotoxicity. This approach involved identification of potential biomarker genes that are mechanistically associated with the toxicity.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Chemicals and materials.
Low RNA Input Linear Amplification Kit and Whole Rat Genome Oligo Microarray were purchased from Agilent Technologies (Palo Alto, CA). RNAlater was from Ambion (Austin, TX). TaqMan Gene Expression Assays were from Applied Biosystems (Foster City, CA). Cyanine 3-CTP and cyanine 5-CTP were from PerkinElmer (Wellesley, MA). QIAzol Lysis Reagent and MagAttract RNA Cell Mini M48 Kit were from Qiagen (Valencia, CA). Rabbit polyclonal antibody against rat heme oxygenase 1 was from Assay Designs (Ann Arbor, MI). Goat serum and Histofine Simple Stain Rat MAX PO (MULTI), which is an amino acid polymer labeled with goat anti-rabbit IgG secondary antibody and peroxidase, were from Nichirei Biosciences (Tokyo, Japan). Chromogen diaminobenzidine (DAB) was from Dako (Glostrup, Denmark).
Animals and administration of chemicals.
The spleen tissues used in this report were obtained from a previously published study (Rokushima et al., 2007). In that study, female Sprague-Dawley albino rats (Crl:CD(SD), approximately 6-week-old) were injected with two different doses of PHZ and PNT, once for 1 day (24 h) or once daily for 4 days and then sacrificed 24 h after the final administration (three rats per group per timepoint). PHZ was dissolved in saline and administered ip at 20 and 80 mg/kg/day, while PNT was dissolved in 0.5 wt/vol% methylcellulose aqueous solution (0.5% MC) and dosed by oral gavage at 500 and 1000 mg/kg/day. Additional dose groups treated with only saline and 0.5% MC for the two exposure periods were used as vehicle controls for PHZ and PNT, respectively. 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.
At necropsy, the rats were euthanized by exsanguination under anesthesia. The spleen was collected and weighed, and a part of it was fixed in 10% neutral buffered formalin for histopathological examination and immunohistochemistry. Another part of the splenic tissue was immediately soaked in RNAlater and stored at – 80°C prior to RNA isolation.
Histopathology.
The formalin-fixed spleen was trimmed, embedded in paraffin, sectioned to a thickness of 3 µm, and stained with hematoxylin and eosin for histopathology. To characterize hemosiderin pigment, the spleen section was additionally stained with Berlin blue and examined under a light microscopy.
RNA isolation.
The spleen tissue was disrupted and lysed in QIAzol Lysis Reagent using a TissueLyser (Qiagen). Total RNA was isolated from the resultant homogenates using a BioRobot M48 Workstation (Qiagen) and MagAttract RNA Cell Mini M48 Kit according to the manufacturer's protocols, which included a DNase digestion step. RNA quality was checked using an Agilent 2100 Bioanalyzer (Agilent Technologies). Equal amounts of total RNA from each animal within individual dose groups (n = 3) were pooled and used for microarray analysis. On the other hand, nonpooled RNAs were employed for quantitative real-time PCR (RT-PCR).
Labeled cRNA preparation and microarray analysis.
Preparation of labeled cRNA targets and hybridization were conducted using Low RNA Input Linear Amplification Kit and Whole Rat Genome Oligo Microarray as previously described (Rokushima et al., 2007). Scanned images from an Agilent DNA microarray scanner (Agilent Technologies) were quantified using Feature Extraction Software 8.5 (Agilent Technologies), where locally weighted scatterplot smoothing (LOWESS) normalization was concurrently conducted. The entire microarray data are available online at http://toxsci.oxfordjournals.org/.
Ratio (treated/control) calculation, extraction of differentially expressed genes (probes), and Gene Ontology (GO) analysis were performed using GeneSpring GX 7.3.1 (Agilent Technologies). Genes were regarded as upregulated if they had a ratio of 
1.5 and as downregulated when they had a ratio of 
0.67. GO analysis was conducted using the "GO Ontology Browser" function implemented in the software, where the significance of the overlap between a gene list and an ontology was examined. Nonredundant gene sets for up- and downregulated probe lists were generated and used for GO analysis because these probe lists contained multiple probes for a single gene, which would lead to misidentification of overrepresented GO categories.
Quantitative RT-PCR.
One-step quantitative RT-PCR was performed in order to confirm the expression changes of 11 selected genes from microarray analysis: aldo-keto reductase family 1, member B8 (Akr1b8), cathepsin B (Ctsb), cathepsin D (Ctsd), fatty acid–binding protein 4, adipocyte (Fabp4), fatty acid–binding protein 5, epidermal (Fabp5), ferritin, heavy polypeptide 1 (Fth1), ferritin light chain 1 (Ftl1), heme oxygenase (decycling) 1 (Hmox1), lectin, galactose-binding, soluble 3 (Lgals3), solute carrier family 11(proton-coupled divalent metal ion transporters), member 1(Slc11a1), and serine protease inhibitor (Spin2c). Beta-actin (Actb) was also measured as an endogenous reference gene. TaqMan Gene Expression Assays (For Actb, Rn00667869_m1; Akr1b8, Rn00756509_g1; Ctsb, Rn00575030_m1; Ctsd, Rn00592528_m1; Fabp4, Rn00670361_m1; Fabp5, Rn00821817_g1; Fth1, Rn00820640_g1; Ftl1, Rn00821071_g1; Hmox1, Rn01536933_m1; Lgals3, Rn00582910_m1; Slc11a1, Rn02114013_s1; and Spin2c, Rn00755832_mH) were utilized as sets of gene-specific probe and primer pair. Using 40 ng of the total RNA from individual animals as a template, quantitative RT-PCR and data processing were conducted as described previously (Rokushima et al., 2007), with the slight modification that Actb was used as an endogenous reference gene in the comparative cycle threshold method.
Quantitative RT-PCR data of the 11 genes were subjected to two-way hierarchical cluster analysis using GeneSpring GX 7.3.1. In the cluster analysis, standard correlation was used to measure the similarities among the genes or conditions, and the average linkage method was employed for calculating the distance between the two clusters.
Immunohistochemistry.
Immunostaining was carried out using Histofine Simple Stain Rat MAX PO (MULTI) following the protocols recommended by the manufacturer with slight modifications. Briefly, formalin-fixed, paraffin-embedded sections (3 µm) were mounted on glass slides, deparaffinized in xylene, rehydrated through a graded series of ethanol, and washed with phosphate-buffered saline (PBS). The sections were treated with 3% H2O2 for 5 min, blocked in 10% goat serum for 10 min, and washed with PBS. Next, the samples were incubated with rabbit anti-heme oxygenase 1 polyclonal antibody (1:250 dilution) for 1 h, washed with PBS, treated with Histofine Simple Stain Rat MAX PO for 30 min, and washed with PBS. Finally, the sections were stained with chromogen DAB, rinsed with distilled water, counterstained with Mayer's hematoxylin, and viewed under a light microscopy.
Statistics.
Correlations between RBC counts, total bilirubin (T-Bil) levels, grades of histopathological lesions, relative spleen weight, and quantitative RT-PCR data were evaluated based on Spearman rank and Pearson correlation coefficient using SAS software. The grades of severity for histopathological lesions were assigned as follows: grade 0, no lesion; grade 1, minimal; grade 2, mild; grade 3, moderate; and grade 4, marked.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
(Histo)Pathology
A significant increase in relative spleen weight was observed following the treatment with PHZ at both dose levels (20 and 80 mg/kg) for 24 h and at the repeated low dose for 4 days and with PNT at the repetitive high dose (1000 mg/kg/day) for 4 days (Supplementary Table 1). Data for the daily high dose of PHZ for 4 days were not available due to mortality. The spleen was dark in color in all PHZ treatment groups and at high doses of PNT for 24 h and 4 days (data not shown).
The histopathology of the spleen associated with hemolytic anemia is shown in Table 1. There were no particular changes in the spleen of the control groups, except for an increase in extramedullary hematopoiesis in one of the three rats dosed with 0.5% MC for 24 h and saline for 4 days, which was considered to be spontaneous (data not shown). Congestion, an excessive accumulation of RBCs resulting from the enhanced sequestration and phagocytosis of them, was observed in the red pulp after all PHZ treatments and the exposure at both dose levels of PNT for 4 days, the severity of which was dose- and time associated. Berlin blue staining revealed hemosiderin deposition in the red pulp in all chemical treatment groups, but both dose levels of PNT for 24 h only induced minimal lesions. In addition, the low dose of PHZ for 4 days and the high dose of PNT for 24 h also provoked extramedullary hematopoiesis in the red pulp. These are well-established splenic lesions accompanying hemolytic anemia (Fujitani et al., 2004), and therefore we could confirm that hemolytic changes had occurred in the spleen of rats treated with the two tested drugs.
|
The (histo)pathological alterations mentioned above were consistent with the decreases in RBC counts and HGB concentrations and the increase in T-Bil levels in the circulation, which were reported previously (Rokushima et al., 2007
) and are included in Supplementary Table 1. Based on the comprehensive interpretation of the results of these toxicological examinations, we judged that the following three dosing conditions had provoked 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. In addition, since the low dose of PHZ for 24 h also caused a slight decrease in RBC counts (Supplementary Table 1) and splenic lesions associated with hemolysis (Table 1), we considered this condition to be a mild but significant hemolytic one.
Microarray Analysis
Splenic gene expression profiles in rats treated with the two direct-acting hemolytic drugs were generated using whole-genome oligonucleotide microarrays. First of all, we extracted probes that were up- or downregulated under each of the three severe hemolytic conditions: PHZ, high dose, 24 h; PHZ, low dose, 4 days; and PNT, high dose, 4 days (Fig. 1). Using 1.5-fold change relative to the vehicle- and time-matched control as a criterion for differential expression, the PHZ high dose for 24 h had 1335 upregulated and 1010 downregulated probes. The repeated low dose of PHZ for 4 days had 2611 and 2634 probes that were induced and repressed, respectively. The repetitive high dose of PNT for 4 days had a total of 2259 deregulated probes, with 1380 upregulated and 879 downregulated. Next, we examined the degree of overlap of the deregulated probes among the hemolytic conditions (Fig. 1). There were 170 and 245 probes that were commonly induced and repressed, respectively, under all the three conditions. The number of differentially expressed probes was much larger than that in the liver under the same exposure conditions, which was obtained in the previous study (Rokushima et al., 2007) (415 and 45 deregulated probes for the spleen and liver, respectively), indicating that, as anticipated, the spleen was more sensitive to the hemolysis than the liver at the transcript levels. Furthermore, in spite of the different chemical treatments, significant degree of overlaps in both up- and downregulated probes were observed between the low dose of PHZ for 4 days and the high dose of PNT for 4 days. Namely, among the 2739 probes that were induced under either of the two conditions, no less than 1252 probes (46%) were included in both the probe lists. Similarly, of the 2795 probes that were included in either of the repressed probe lists, as many as 718 probes (26%) were downregulated under these two conditions. This observation suggests that much of the expression changes we observed resulted from the secondary effects of hemolysis (e.g., phagocytosis and destruction of damaged RBCs) rather than from a direct impact of the agents (and/or their metabolites). It should be mentioned that, in the case of the low dose of PHZ for 4 days, congestion and severe splenomegaly were provoked (Table 1 and Supplementary Table 1), probably representing the sequestration of several types of blood cells in addition to damaged erythrocytes in the spleen, which may lead to substantial changes in the composition of the cell population. However, the gene encoding a macrophage marker CD68 was not differentially expressed under any of the dosing conditions tested (data not shown), suggesting that the gene expression alterations observed were not due to a change in the macrophage population under the hemolytic conditions.
|
Genes represented in Table 2 are those deregulated by
twofold under all the three hemolytic conditions. Prominent functional categories of upregulated genes included proteolysis, signal transduction, acute phase response, and lipid metabolism. The proteolysis genes consisted of three protease-coding genes, Ctsb, Ctsd, and legumain (Lgmn), and those categorized in lipid metabolism were Fabp4 and Fabp5. The most quantitatively induced gene in Table 2 was Lgals3 (10.79-fold, at the high dose of PHZ for 24 h), which has been suggested to be involved in erythrophagocytosis in macrophages (Sano et al., 2003). The upregulated gene list also encompassed heme oxygenase(decycling) 1 (Hmox1), encoding rate-limiting enzyme in heme degradation, and thereby playing a pivotal role in hemolytic response. Of the upregulated genes in Table 2, Hmox1, Lgals3, and glycoprotein(transmembrane) nmb (Gpnmb) were also among the genes whose transcript levels were increased in the liver of the same hemolytic anemia models (Rokushima et al., 2007). On the other hand, the downregulated gene list had two major categories, immune response and T-cell activation. Immune response genes included RT1 class Ib, locus Aw2 (RT1-Aw2) and RT1 class II, locus Bb (RT1-Bb), with the former being the most repressed gene under both the high dose of PHZ for 24 h and the high dose of PNT for 4 days.
|
The differentially expressed probes by
1.5-fold under all the severe hemolytic conditions were converted into nonredundant gene sets, generating 154 and 236 genes for up- and downregulation, respectively (Supplementary Table 2). Of the resultant genes, those having GO annotations were subjected to GO analysis, where statistically overrepresented GO categories in the group of genes were identified. The top three predominant GO categories of deregulated genes are listed in Table 3. The GO term of "Biological process" most significantly enriched in upregulated genes was macromolecule catabolism, which included several proteasome subunit genes, such as proteasome(prosome, macropain) 26S subunit, ATPase, 6 (Psmc6), followed by iron ion transport, being comprised of three iron-related genes: Fth1, Ftl1, and Slc11a1. Prominent GO terms for "Molecular function" in induced genes included unfolded protein binding, which consisted of chaperone genes including heat shock protein 1, alpha (Hspca). These results indicate that the gene expression changes we found were likely to reflect the enhanced proteolysis of damaged HGB and subsequent generation of free iron within macrophages, both of which are well-documented splenic events in hemolysis. In contrast to the upregulated genes, GO terms overrepresented in the downregulated genes were generally associated with immune response, such as defense response for "Biological process", and immunological synapse for "Cellular component". Complete lists of genes for Table 3 are presented online as Supplementary Table 3.
|
Quantitative RT-PCR Data Analysis.
Of the genes upregulated under the hemolytic conditions, we chose 11 genes as biomarker candidates based on their expression patterns, the magnitude of expression changes, and/or putative functional association with hemolytic events in the spleen (described below). They were Akr1b8, Ctsb, Ctsd, Fabp4, Fabp5, Fth1, Ftl1, Hmox1, Lgals3, Slc11a1, and Spin2c. These genes were further evaluated using quantitative RT-PCR. Unlike in the case of microarray analysis, splenic total RNAs from individual rats (n = 33) were employed as templates in quantitative RT-PCR in order to compare changes between gene expression and classical toxicological parameters at individual levels and to estimate interindividual variation in gene expression. As shown in Figure 2, there was a good agreement between microarray and quantitative RT-PCR data, confirming that these genes were actually upregulated under the hemolytic conditions. Additionally, quantitative RT-PCR measurements revealed a considerable degree of overall interindividual variation in the transcript level of serine protease inhibitor gene Spin2c, which could not be observed by microarray measurements, indicating that Spin2c was a less accurate biomarker candidate.
|
For phenotypic anchoring, we calculated Pearson and Spearman rank correlation coefficients between quantitative RT-PCR data of the 11 selected genes and five classical toxicological parameters: RBC counts, T-Bil levels, grades of congestion and hemosiderin deposition (splenic histopathological lesions), and relative spleen weight. The result is presented in Table 4 (p values are presented online as Supplementary Table 4). Most of the selected genes exhibited expression patterns highly correlated with all the toxicological parameters. For example, against T-Bil levels, grades of congestion and hemosiderin deposition, and relative spleen weight, 10, 10, 9, and 7 out of the 11 genes had Spearman rank correlation coefficients of at least 0.7 (corresponding to a p value of < 0.0001), respectively. Similarly, for RBC counts, seven genes showed Spearman correlation coefficients of less than – 0.6 (corresponding to a p value of 0.0002). These results indicate that expression alterations of a subset of genes in the spleen upon exposure to hemolytic agents could reflect the classical toxicological phenotypes. We also observed a tendency for the selected genes to have larger absolute values of Spearman correlations for T-Bil levels and grades of congestion and hemosiderin deposition than for RBC counts. Of the 11 genes, Hmox1 displayed the highest Spearman correlation with both T-Bil levels and grade of hemosiderin deposition, and Lgals3 and Ctsd had the strongest Spearman correlation against RBC counts and T-Bil levels, respectively. Scatter plots for these comparisons are available online as Supplementary Figure 1. It should be noted that Hmox1 and Lgals3 were also identified as hepatic marker candidates for hemolytic anemia in our previous study (Rokushima et al., 2007
), which is consistent with the established role (for Hmox1, heme degradation) or the suggested role (for Lgals3, erythrophagocytosis) of these genes in response to hemolysis both in the liver and spleen.
|
Two-way hierarchical clustering was conducted for genes and individual animals based on the quantitative RT-PCR data (Fig. 3). In the dendrogram for individual rats, there was a definitive cluster comprised mainly of animals given the doses that provoked a statistically significant decrease in RBC counts (designated as "hemolytic cluster" in Fig. 3). Interestingly, the cluster also included three of the three rats treated with the low dose of PNT for 4 days, which caused no decrease in erythrocyte counts (Supplementary Table 1). This grouping is supported by the observation that the low-dose PNT for 4 days formed a cluster with the hemolytic ones in the hierarchical cluster analysis where microarray data for all the deregulated probes were utilized (Supplementary Fig. 2). In addition, the low dose of PNT for 4 days induced histopathological lesions associated with hemolysis (Table 1) and slight elevation in T-Bil levels (Supplementary Table 1) and reticulocyte counts (data not shown) in the circulation. These results suggest that enhanced uptake of damaged erythrocytes by splenic macrophages occurred under the condition, and gene expression could sensitively detect the alterations.
|
Immunohistochemistry.
We further confirmed the elevated protein level of Hmox1, one of the most promising biomarker candidates evaluated by quantitative RT-PCR data, using immunohistochemistry. Spleen sections from representative animals within each dose group were examined for the presence of immunoreactivity with anti-rat Hmox1 antibody. The result is shown in Figure 4. Protein expression of Hmox1 was upregulated relative to the controls following all PHZ treatments and the exposure to both dose levels of PNT for 4 days, which was consistent with the increase in transcript levels (Fig. 2). Positive immunostaining was mainly restricted to macrophage-like cells in the red pulp and appeared to be colocalized with hemosiderin deposition (data not shown). Of note is the observation of basal-level expression of Hmox1 under nonhemolytic conditions including vehicle controls, which was likely to have arisen from the destruction of normally damaged, senescent erythrocytes by splenic macrophages.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The aims of the present study were to identify altered gene expression patterns in the spleen that allow detection of drug-induced hemolytic anemia and to better understand the molecular details of splenic events accompanying the hematotoxicity. We analyzed gene expression profiles in the spleen of rats injected with two hemolytic drugs, PHZ and PNT, and found hundreds of genes commonly deregulated under all the severe hemolytic conditions (Fig. 1). GO analyses revealed that overrepresented GO terms of the induced genes included macromolecule catabolism (protein catabolism) and iron ion transport, both of which comprise significant processes of hemolytic events in the spleen (Table 3). Together with these relationships, we investigated plausible links between the upregulated genes and splenic changes in hemolysis. The resultant putative associations, including those for the selected genes as biomarker candidates, are schematically displayed in Figure 5.
|
Degradation of Erythrocytes
Administration of the two hemolytic drugs led to upregulation of a number of genes associated with proteolysis. Among these were Ctsb, Ctsd, and Lgmn (Table 2). Ctsb and Ctsd are two major lysosomal cysteine and aspartyl proteases, respectively, that contribute to the degradation of proteins taken up by phagocytosis (Bohley and Seglen, 1992
). Lgmn is an asparagine-specific cysteine proteinase that has a regulatory role in the biosynthesis of lysosomal proteases, including Ctsb (Shirahama-Noda et al., 2003). The upregulated gene list also included six proteasome subunit genes (Psma5, Psmb6, Psmc2, Psmc5, Psmc6, and Psmd1) and three ubiquitin-related genes (Sqstm1, Ube2v2, and Uchl5) (Supplementary Table 2). The ubiquitin/proteasome system is known to play a significant role in the breakdown of intracellular damaged proteins, including oxidized HGB, outside of lysosomes (Grune et al., 2003). Thus, both lysosomal and nonlysosomal protease genes were coordinately induced under the hemolytic conditions, which are likely, at least in part, to reflect the accelerated proteolysis of phagocytized, damaged HGB within macrophages in our experiment.
Two genes involved in lipid metabolism, Fabp4 and Fabp5, showed increased mRNA levels under the hemolytic conditions (Table 2). These fatty acid–binding proteins (FABPs) are highly induced during macrophage maturation (Verhoeckx et al., 2004) and are thought to have roles in the fatty acid uptake, transport, and metabolism. Recently, Fabp4 and Fabp5 have been suggested to function as antioxidant proteins by scavenging reactive lipids, such as 4-hydroxynonenal (4-HNE), one of the end products of lipid peroxidation (Bennaars-Eiden et al., 2002; Grimsrud et al., 2007). Furthermore, another upregulated gene Akr1b8 (Table 2) has a human homolog that is also implicated in detoxification of 4-HNE (Vander Jagt et al., 1995). Considering that HGB-derived heme is known to cause lipid peroxidation within cells (Kumar and Bandyopadhyay, 2005), it is possible to speculate that these two FABPs and Akr1b8 might have had a role in the metabolism of putatively generated lipid peroxides during the destruction of damaged RBC in this study (Fig. 5).
Iron Metabolism
We identified the upregulation of three genes involved in iron metabolism (Fth1, Ftl1, and Slc11a1) (Table 2 and Supplementary Table 2). Fth1 and Ftl1 are two subunits that assemble to form ferritin, a well-known iron-storage protein. In reticuloendothelial cells, ferritin sequesters excess iron derived from heme degradation following erythrophagocytosis, thereby preventing iron-catalyzed production of reactive oxygen species (ROS), which can cause cell injury (Knutson and Wessling-Resnick, 2003; Torti and Torti, 2002). Since several studies have shown that transcription of ferritin genes is activated in response to iron-derived ROS and/or iron itself (Knutson and Wessling-Resnick, 2003; Torti and Torti, 2002), the induction of ferritin genes in this experiment may be attributable in large part to the enhanced level of intracellular iron, which was confirmed as hemosiderin deposition (Fig. 5 and Table 1). On the other hand, Slc11a1, also known as Nramp1, encodes a divalent-cation transporter that is highly expressed in phagocytic cells (Wyllie et al., 2002). Slc11a1 mRNA level has been shown to increase in bone marrow macrophages in response to iron (Baker et al., 2000) or hemin, a synthetic heme analog (Biggs et al., 2001), indicating that the upregulation of this gene we observed may also have resulted from a suspected increase in iron or heme levels. Interestingly, because of its localization to the phagosomal membrane and ability to transport iron, it is postulated that Slc11a1 transports erythrophagocytosis-derived iron out of phagosomes into the cytosol of macrophages (Fleming et al., 1998). Our data might therefore support its involvement in the process in rat spleen.
Gene Expression Regulated by Heme or Oxidative Stress
Included among the induced genes were five antioxidant genes: Fth1, Ftl1, Hmox1, peroxiredoxin 1 (Prdx1), and thioredoxin 1 (Txn1) (Fig. 5 and Supplementary Table 2). In humans and/or mice, these genes contain an overlapping Maf recognition element and antioxidant responsive element (MARE/ARE) or the ARE alone in their promoter regions and are controlled by an oxidative stress–activated transcriptional activator, NF-E2–related factor 2 (Nrf2) (Hintze and Theil, 2005; Inamdar et al., 1996; Ishii et al., 2000; Kim et al., 2001). An additional 18 upregulated genes, including five biomarker candidates (Akr1b8, Ctsd, Fabp4, Lgals3, and Spin2c), have also been reported to be positively regulated by Nrf2 from analyses of knock out (nrf2–/–) mice (Ishii et al., 2000; Kwak et al., 2003; Nair et al., 2007; Thimmulappa et al., 2002) (Supplementary Table 2). On the other hand, MARE is a cis-acting element bound by a transcriptional repressor Bach1, the DNA-binding activity of which is attenuated by enhanced heme levels (Igarashi and Sun, 2006). Because intracellular heme and iron-derived ROS levels probably increased in the spleen under the hemolytic conditions as a result of phagocytosis of damaged RBC by macrophages (Fig. 5), activation of Nrf2 and/or repression of Bach1 may have occurred and contributed to the upregulation of these genes in our rat study.
Detection of Hemolytic Anemia
Among the 11 putative gene-based markers, Hmox1, which is upregulated by heme or ROS, exhibited the highest Spearman correlation for both the grade of hemosiderin deposition and T-Bil levels (Table 4), which agrees with the established role of this gene; namely, Hmox1 encodes the rate-limiting enzyme in heme catabolism that leads to the generation of iron and bilirubin (Fig. 5). In addition, the immunohistochemistry as well as mRNA levels of Hmox1 seemed to detect the hemolytic changes in the spleen with sensitivity similar to that of the histopathology (Figs. 2 and 4, and Table 1). Furthermore, this gene was identified as a hepatic biomarker candidate (Rokushima et al., 2007). It should also be noted that Hmox1 induction may even be predictive of the later onset of the hematotoxicity because the increased expression of this gene appeared to precede the elevation in T-Bil levels in the blood. For example, at the low dose of PHZ, the upregulation of Hmox1 at both the transcript and protein levels occurred within 24 h after the administration (Figs. 2 and 4), whereas a significant increase in T-Bil levels was observed only after the repetitive injections for 4 days (Supplementary Table 1). These results strongly suggest that Hmox1 is one of the best single gene biomarkers for hemolytic anemia. Of course, induction of Hmox1 does not necessarily represent the incidence of alterations associated with hemolysis. This gene is inducible in response not only to heme or ROS but also to other stresses or stimuli, including heavy metals and hypoxia (Farombi and Surh, 2006). Nevertheless, judging from the results mentioned above, we consider that Hmox1 is a promising biomarker candidate and can be used to evaluate hemolytic actions of a series of chemicals or drug candidates that are expected to have qualitatively similar toxicity profiles.
The spleen was more susceptible to the effects of hemolysis than the liver at the gene expression levels in that a much larger number of differentially expressed genes was found in the spleen than in the liver under the same hemolytic conditions (Rokushima et al., 2007) (Fig. 1). However, this was not necessarily true of the magnitude of the expression changes. For instance, Hmox1 and Lgals3, which were identified as both hepatic and splenic marker candidates, were induced to a similar extent in both organs under the hemolytic conditions (Rokushima et al., 2007) (Fig. 2). This appeared to be due to the relatively high basal-level expression of these genes in the spleen under the nonhemolytic conditions (data not shown), which were likely to be derived from the destruction of normally damaged erythrocytes by macrophages. For this reason, we could not definitely conclude which organ would provide better identification of the hematotoxicity using altered gene expression.
All the 11 marker candidate genes had higher Spearman correlations with severity of congestion and hemosiderin deposition (histopathological changes) than with RBC counts in the blood (Table 4), indicating that expression changes of these genes were more consistent with the splenic lesions than with RBC counts, the latter being a less sensitive indicator of hemolytic anemia due to the compensatory activation of erythropoiesis. In addition, the hierarchical clustering analysis of the quantitative RT-PCR data suggested that the transcript levels of the 11 genes sensitively responded to the hemolytic changes at the low dose of PNT for 4 days, which caused no apparent decrease in RBC counts but induced splenic lesions associated with hemolysis (Fig. 3). These findings indicate that expression changes of the selected genes in the spleen may allow detection of erythrocyte damage with sensitivity higher than that of RBC counts in the circulation and comparable to that of splenic histopathological examination. Hence, we propose that the spleen is a promising organ for detection of drug-induced hemolytic anemia based on changes in the expression of a subset of genes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Drug-induced hemolytic anemia provoked expression changes of hundreds of genes in the spleen. Upregulated genes included those that appeared to be involved in splenic events characteristic of the hematotoxicity, such as proteolysis and iron metabolism, and a number of genes reported to be regulated by Nrf2, including Hmox1. Expression levels of most of the selected biomarker candidate genes exhibited high correlation with conventional toxicological endpoints and could be expected to detect hemolytic changes more sensitively than RBC counts in the blood. Among the selected genes, Hmox1 was one of the most promising candidate 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 team members in our laboratories for their support and scientific advice throughout the experiments. We also wish to thank Toru Yanagimoto, Yoshimasa Kyokawa, Kumiko Takeuchi, Chieko Yabuuchi, and Dr Satoshi Inoue for their helpful suggestions and discussions.
|
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. Identification of putative gene based markers of renal toxicity. Environ. Health Perspect. (2004) 112:465–479.[Web of Science][Medline]
Baker ST, Barton CH, Biggs TE. A negative autoregulatory link between Nramp1 function and expression. J. Leukoc. Biol. (2000) 67:501–507.[Abstract]
Bennaars-Eiden A, Higgins L, Hertzel AV, Kapphahn RJ, Ferrington DA, Bernlohr DA. Covalent modification of epithelial fatty acid-binding protein by 4-hydroxynonenal in vitro and in vivo. Evidence for a role in antioxidant biology. J. Biol. Chem. (2002) 277:50693–50702.
Beutler E. Drug-induced hemolytic anemia. Pharmacol. Rev. (1969) 21:73–103.
Biggs TE, Baker ST, Botham MS, Dhital A, Barton CH, Perry VH. Nramp1 modulates iron homoeostasis in vivo and in vitro: Evidence for a role in cellular iron release involving de-acidification of intracellular vesicles. Eur. J. Immunol. (2001) 31:2060–2070.[CrossRef][Web of Science][Medline]
Bohley P, Seglen PO. Proteases and proteolysis in the lysosome. Experientia (1992) 48:151–157.[CrossRef][Web of Science][Medline]
Cesta MF. Normal structure, function, and histology of the spleen. Toxicol. Pathol. (2006) 34:455–465.[CrossRef][Web of Science][Medline]
Ellinger-Ziegelbauer H, Stuart B, Wahle B, Bomann W, Ahr HJ. Characteristic expression profiles induced by genotoxic carcinogens in rat liver. Toxicol. Sci. (2004) 77:19–34.
Farombi EO, Surh YJ. Heme oxygenase-1 as a potential therapeutic target for hepatoprotection. J. Biochem. Mol. Biol. (2006) 39:479–491.[Web of Science][Medline]
Fleming MD, Romano MA, Su MA, Garrick LM, Garrick MD, Andrews NC. Nramp2 is mutated in the anemic Belgrade (b) rat: Evidence of a role for Nramp2 in endosomal iron transport. Proc. Natl. Acad. Sci. U.S.A. (1998) 95:1148–1153.
Fujitani T, Tada Y, Yoneyama M. Chlorpropham-induced splenotoxicity and its recovery in rats. Food Chem. Toxicol. (2004) 42:1469–1477.[CrossRef][Web of Science][Medline]
Grimsrud PA, Picklo MJ Sr, Griffin TJ, Bernlohr DA. Carbonylation of adipose proteins in obesity and insulin resistance: Identification of adipocyte fatty acid-binding protein as a cellular target of 4-hydroxynonenal. Mol. Cell. Proteomics (2007) 6:624–637.
Grune T, Merker K, Sandig G, Davies KJ. Selective degradation of oxidatively modified protein substrates by the proteasome. Biochem. Biophys. Res. Commun. (2003) 305:709–718.[CrossRef][Web of Science][Medline]
Heinloth AN, Irwin RD, Boorman GA, Nettesheim P, Fannin RD, Sieber SO, Snell ML, Tucker CJ, Li L, Travlos GS, et al. Gene expression profiling of rat livers reveals indicators of potential adverse effects. Toxicol. Sci. (2004) 80:193–202.
Hintze KJ, Theil EC. DNA and mRNA elements with complementary responses to hemin, antioxidant inducers, and iron control ferritin-L expression. Proc. Natl. Acad. Sci. U.S.A. (2005) 102:15048–15052.
Igarashi K, Sun J. The heme-Bach1 pathway in the regulation of oxidative stress response and erythroid differentiation. Antioxid. Redox Signal. (2006) 8:107–118.[CrossRef][Web of Science][Medline]
Inamdar NM, Ahn YI, Alam J. The heme-responsive element of the mouse heme oxygenase-1 gene is an extended AP-1 binding site that resembles the recognition sequences for MAF and NF-E2 transcription factors. Biochem. Biophys. Res. Commun. (1996) 221:570–576.[CrossRef][Web of Science][Medline]
Ishii T, Itoh K, Takahashi S, Sato H, Yanagawa T, Katoh Y, Bannai S, Yamamoto M. Transcription factor Nrf2 coordinately regulates a group of oxidative stress-inducible genes in macrophages. J. Biol. Chem. (2000) 275:16023–16029.
Kim YC, Masutani H, Yamaguchi Y, Itoh K, Yamamoto M, Yodoi J. Hemin-induced activation of the thioredoxin gene by Nrf2. A differential regulation of the antioxidant responsive element by a switch of its binding factors. J. Biol. Chem. (2001) 276:18399–18406.
Knutson M, Wessling-Resnick M. Iron metabolism in the reticuloendothelial system. Crit. Rev. Biochem. Mol. Biol. (2003) 38:61–88.[Web of Science][Medline]
Kumar S, Bandyopadhyay U. Free heme toxicity and its detoxification systems in human. Toxicol. Lett. (2005) 157:175–188.[CrossRef][Web of Science][Medline]
Kwak MK, Wakabayashi N, Itoh K, Motohashi H, Yamamoto M, Kensler TW. Modulation of gene expression by cancer chemopreventive dithiolethiones through the Keap1-Nrf2 pathway. Identification of novel gene clusters for cell survival. J. Biol. Chem. (2003) 278:8135–8145.
Nair S, Xu C, Shen G, Hebbar V, Gopalakrishnan A, Hu R, Jain MR, Liew C, Chan JY, Kong AN. Toxicogenomics of endoplasmic reticulum stress inducer tunicamycin in the small intestine and liver of Nrf2 knockout and C57BL/6J mice. Toxicol. Lett. (2007) 168:21–39.[CrossRef][Web of Science][Medline]
Paules R. Phenotypic anchoring: Linking cause and effect. Environ. Health Perspect. (2003) 111:A338–A339.[Web of Science][Medline]
Rokushima M, Omi K, Araki A, Kyokawa Y, Furukawa N, Itoh F, Imura K, Takeuchi K, Okada M, Kato I, et al. A toxicogenomic approach revealed hepatic gene expression changes mechanistically linked to drug-induced hemolytic anemia. Toxicol. Sci. (2007) 95:474–484.
Sano H, Hsu DK, Apgar JR, Yu L, Sharma BB, Kuwabara I, Izui S, Liu FT. Critical role of galectin-3 in phagocytosis by macrophages. J. Clin. Invest. (2003) 112:389–397.[CrossRef][Web of Science][Medline]
Shirahama-Noda K, Yamamoto A, Sugihara K, Hashimoto N, Asano M, Nishimura M, Hara-Nishimura I. Biosynthetic processing of cathepsins and lysosomal degradation are abolished in asparaginyl endopeptidase-deficient mice. J. Biol. Chem. (2003) 278:33194–33199.
Sivilotti ML. Oxidant stress and haemolysis of the human erythrocyte. Toxicol. Rev. (2004) 23:169–188.[CrossRef][Medline]
Suter L, Babiss LE, Wheeldon EB. Toxicogenomics in predictive toxicology in drug development. Chem. Biol. (2004) 11:161–171.[CrossRef][Web of Science][Medline]
Thimmulappa RK, Mai KH, Srisuma S, Kensler TW, Yamamoto M, Biswal S. Identification of Nrf2-regulated genes induced by the chemopreventive agent sulforaphane by oligonucleotide microarray. Cancer Res. (2002) 62:5196–5203.
Torti FM, Torti SV. Regulation of ferritin genes and protein. Blood (2002) 99:3505–3516.
Vander Jagt DL, Kolb NS, Vander Jagt TJ, Chino J, Martinez FJ, Hunsaker LA, Royer RE. Substrate specificity of human aldose reductase: Identification of 4-hydroxynonenal as an endogenous substrate. Biochim. Biophys. Acta (1995) 1249:117–126.[CrossRef][Medline]
Verhoeckx KC, Bijlsma S, de Groene EM, Witkamp RF, van der Greef J, Rodenburg RJ. A combination of proteomics, principal component analysis and transcriptomics is a powerful tool for the identification of biomarkers for macrophage maturation in the U937 cell line. Proteomics (2004) 4:1014–1028.[CrossRef][Web of Science][Medline]
Wyllie S, Seu P, Goss JA. The natural resistance-associated macrophage protein 1 Slc11a1 (formerly Nramp1) and iron metabolism in macrophages. Microbes Infect. (2002) 4:351–359.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||