Induction of Apoptosis in Mouse Liver by Microcystin-LR

Microcystins (MCs) are a family of cyclic heptapeptide hepatotoxins produced by freshwater species of cyanobacteria that have been implicated in the development of liver cancer, necrosis, and even deadly intrahepatic bleeding. MC-LR, the most toxic MC variant, is also the most commonly encountered in a contaminated aquatic system. This study presents the first data in the toxicological research of MCs that combines the use of standard apoptotic assays with transcriptomics, proteomic technologies, and computer simulations. By using histochemistry, DNA fragmentation assays, and flow cytometry analysis, we determined that MC-LR causes rapid, dose-dependent apoptosis in mouse liver when BALB/c mice are treated with MC-LR for 24 h at doses of either 50, 60, or 70 μg/kg of body weight. We then used gene expression profiling to demonstrate differential expressions (>2-fold) of 61 apoptosis-related genes in cells treated with MC-LR. Further proteomic analysis identified a total of 383 proteins of which 35 proteins were up-regulated and 30 proteins were down-regulated more than 2.5-fold when compared with controls. Combining computer simulations with the transcriptomic and proteomic data, we found that low doses (50 μg/kg) of MC-LR lead to apoptosis primarily through the BID-BAX-BCL-2 pathway, whereas high doses of MC-LR (70 μg/kg) caused apoptosis via a reactive oxygen species pathway. These results indicated that MC-LR exposure can cause apoptosis in mouse liver and revealed two independent pathways playing a major regulatory role in MC-LR-induced apoptosis, thereby contributing to a better understanding of the hepatotoxicity and the tumor-promoting mechanisms of MCs.

The toxic potential of bloom-forming cyanobacteria in eutrophic surface waters has caused increasing concern over the last decade. Rapid advances have been made in the study of the chemically diverse cyanotoxins produced by these cyanobacteria, including significant progress in the study of microcystins (MCs) 1 (1)(2)(3)(4)(5)(6).
MCs are a family of extensively studied cyclic heptapeptide hepatotoxins among which MC-LR is both the most toxic and the most commonly encountered variant (7). Although there is evidence that MC toxicity is primarily associated with liver disruption, MCs have also been seen to cause damage to the intestines, kidneys, and thymus of quail (8,9). Various other MC effects have also been described, including neurotoxicity, genotoxicity, and embryotoxicity. Some of our previous studies also elucidated the potency of MCs as immune intruders (10,11).
MCs are highly liver-specific, and the hepatotoxicity of MCs has been studied extensively (12)(13)(14). These toxins penetrate liver cell membranes through a bile acid carrier and result in changes such as overphosphorylation of liver enzymes, liver necrosis, and even deadly intrahepatic bleeding (15,16) by inhibiting protein phosphatase (PP) 1 and PP2A (11). There is evidence that these adverse effects are closely related to oxidative stress processes and free radicals (17)(18)(19). DNA damage has also been documented (18,20) as well as apoptosis, which is of great importance (21).
Apoptosis is one mechanism used by cells and tissues in response to various toxins; it is characterized by distinct morphological features such as cell shrinkage, chromatin condensation, plasma membrane blebbing, oligonucleosomal DNA fragmentation, and finally the breakdown of the cell into smaller units (apoptotic bodies) (22)(23)(24). The apoptotic effects of MCs have recently become a serious focal point of research, and many of the effects on cells and tissues triggered by MCs have been documented (21,26,27). In 1990, Repavich et al. (28) found that in human lymphocytes MCs had a dose-dependent clastogenic effect that was connected with chromosomal breakage. MCs are also capable of causing apoptosis in rat hepatocytes and DNA damage in human hepatoma cells (29). Ding et al. (30) found that reactive oxygen species (ROS) and mitochondrial permeability transition played an important role in apoptosis induced by MC-LR, although very little is known about the exact apoptotic mechanism of MCs.
To determine the mechanism of MC-LR-induced apoptosis, we performed an analysis of the patterns of gene expression in BALB/c mouse liver treated with 0, 50, 60, and 70 g of MC-LR/kg of body weight. The objective of this analysis was to gain deeper insight into the regulation of the transcriptional program of mouse liver in response to MC-LR treatment. Furthermore, as a means of identifying the protein components involved in apoptosis regulation, we used 2-DE and mass spectrometry to identify proteins that are differentially secreted by mouse liver at various MC-LR concentrations. We also developed a mathematical model based on the data from transcriptomic and proteomic analyses and validated our hypothetical explanation of MC-LR-induced apoptosis using computer simulations. Finally we used this model to investigate the kinetics of several apoptosis-related factors in this process.

EXPERIMENTAL PROCEDURES
Reagents-MC-LR standard was obtained from Alexis Inc. (Carlsbad, CA). Unless otherwise indicated, other reagents were obtained from Sigma.

Animal Treatment and Detection of Apoptosis Induced by MC-LR
Animal Treatment-Six-to 8-week-old virus-free female BALB/c mice were purchased from The Animal Center of Nanjing Medical University, China. Mice were randomly assigned into four groups when their body weight was between 16 and 20 g. Each of the groups was then used to test different concentrations of MC-LR (0, 50, 60, and 70 g of MC-LR/kg of body weight) by separately administering an intraperitoneal injection over a period of 24 h. After treatment, the mice were sacrificed, and their livers were aseptically excised.
Histochemistry-Histochemistry was done on polyformaldehydefixed tissues to determine the distribution of hepatocellular apoptosis induced by MC-LR. Polyformaldehyde-fixed tissues were processed in a standard fashion using an automated processor and were then embedded in paraffin from which 3-mm sections were cut. These 3-mm sections were placed on charged slides and deparaffinized in three changes of xylene followed by rehydration in a graded alcohol series. Slides were then rinsed in water and Hoechst 33258.
DNA Fragmentation Assay-Mouse liver tissue was rinsed with ice-cold PBS twice and lysed in buffer containing 10 mM Tris (pH 7.4), 150 mM NaCl, 5 mM EDTA, and 0.5% Triton X-100 for 30 min on ice. Lysates were vortexed and cleared by centrifugation at 10,000 ϫ g for 20 min. Fragmented DNA in the supernatant was extracted with an equal volume of neutral phenol:chloroform:isoamyl alcohol mixture (25:24:1) and analyzed electrophoretically on 1% agarose gels containing 0.1 mg/ml ethidium bromide.
Flow Cytometry Analysis-Hepatocytes were isolated from liver tissue by perfusion after 24 h of MC-LR treatment. Cells were then resuspended in 200 ml of binding buffer (0.1 M Hepes, pH 7.4, 14 M NaCl, 25 mM CaCl 2 ) at a concentration of 1 ϫ 10 6 . 7-Aminoactinomycin D (7-AAD) was subsequently added to the cell solution. Cells were then incubated in the dark for 30 min at room temperature and diluted with 400 l of binding buffer for flow cytometric analysis. The analysis of 10,000 cells in each group was performed on a FACSCalibur flow cytometer (BD Biosciences) equipped with a single laser emitting excitation light at 488 nm.

Transcriptomics
RNA Extraction-Total RNA was extracted from liver tissue using TriPure isolation reagent (catalog number 1 667 165, Roche Applied Science). The amount and purity of the RNA was estimated by measuring OD at 260 nm and 280 nm where 1 OD at 260nm is equivalent to 40 mg/ml RNA, and an OD ratio (260/280) of 1.7-1.8 is indicative of acceptable purity.

TABLE I The primers of selected genes for real time PCR
Fadd, Fas-associated death domain; Apaf-1, apoptotic protease-activating factor 1.

Target gene
Primer sequence (sense (ϩ) and antisense (Ϫ)) Annealing temperature Size°C Oligonucleotide Microarray Analysis-The labeled cDNA was prepared using 5 g of the total RNA. Reverse transcription was performed using a GEArray TM AmpoLabeling-LPR kit (Superarray Bioscience Corp.). Hybridization was carried out on GEArray Q Series Mouse Apoptosis Gene Array membrane (96 apoptosis-related genes, Superarray Bioscience Corp.). Gene spots detected by GEArray Analyzer software (Superarray Bioscience Corp.) showing more than a 2-fold difference of chemiluminescent intensity between different dose samples and a negative control were considered to be differentially expressed genes as a result of MC-LR induction. A comparison of the data analyzed from two experiments indicated that both experiments were highly reproducible. We performed hybridization three times for each sample. Average values were obtained. The microarray data were then subjected to hierarchical clustering using Gene Cluster software (Stanford University).
Quantitive Real Time PCR Analysis of Alternatively Spliced Transcripts-Primers and annealing temperatures are listed in Table I. The RT reaction was performed on 150 ng of total RNA with avian myeloblastosis virus reverse transcriptase (Promega). Quantitative real time PCR was performed using QuantiTect SYBR Green I (Qiagen). PCRs were performed in a total volume of 20 l (1ϫ QuantiTect SYBR Green Master Mix) in the ABI Prism 7000 system (Applied Biosystems). The PCR program was as follows: 1 cycle for 15 min at 95°C; 45 cycles for 15 s at 95°C, 30 s at different annealing temperatures of the primer pair (Table I), and 20 s at 72°C; 1 cycle for 10 min at 72°C. The specificity and identity of the PCR product was checked by performing a melting curve test. The absolute number of copies of the gene of interest in the experimental cDNA samples was calculated from the linear regression of a standard curve. The expression of the measured genes in each sample was normalized for Gapdh expression. All samples were analyzed in triplicate.

Proteomics
Sample Preparation for 2-DE-Whole livers from MC-LR-treated mice were rinsed twice with ice-cold PBS, homogenized, and then suspended in at least 3 volumes of ice-cold ethanol. The protein in each sample was then precipitated at Ϫ30°C for 2 h and pelleted at 15,000 ϫ g at 4°C for 30 min. The protein pellet was rinsed twice with ice-cold ethanol and then dissolved in rehydration buffer containing 7 M urea, 2 M thiourea, 2% (w/v) CHAPS, 65 mM DTT, and 0.5% IPG buffer (Amersham Biosciences). The protein concentration was determined by the Bradford method.
2-DE-Three gels per sample were processed and analyzed simultaneously. The first dimension was carried out on an Ettan IPGphor II isoelectric focusing unit (Amersham Biosciences) using 24-cm pH 3-10 IPG gel strips and 450 l of sample solution. IEF was performed at 20°C under the following conditions: 12 h at 30 V, 30 min at 300 V, 1 h at 500 V, 1 h at 1000 V, 2 h at 5000 V, and 8 h at 8000 V. After IEF, the IPG strips were immediately equilibrated two successive times for 15 min each in SDS equilibration buffer containing 50 mM Tris-HCl, pH 6.8, 30% glycerol (Amresco), 1% SDS (Amresco), and traces of bromphenol blue. The first equilibration was performed in the above mentioned equilibration buffer with 1% (w/v) DTT followed by a second equilibration with 2.5% (w/v) iodoacetamide. The gels were subsequently subjected to a second dimensional electrophoresis on 12.5% polyacrylamide gels using an Ettan Daltsix electrophoresis system (Amersham Biosciences). SDS-PAGE was performed at 2 watts/gel for 45 min and 15 watts/gel for about 5 h until the dye front reached the bottom of the gels. Finally proteins on the 2-DE gels were visualized with Coomassie Brilliant Blue R-250.  Gel Image and Data Analysis-The Coomassie Blue-stained gels were scanned on a Sharp JX-330 color image scanner. Spot detection, quantification, and matching were performed with PDQuest software (Bio-Rad). Proteins separated by 2-DE gels were quantitated in terms of their relative volume (percent volume). This relative volume was calculated by dividing the individual spot volume by the sum of total spot volumes and then multiplying by 100. The expression level of proteins with Ͼ2.5 times increase or decrease as compared with the control were considered significantly changed. The Mann-Whitney test was used to assess the statistical relationships of different treatment groups. When p Ͻ 0.01, the relationships can be considered statistically significant.
In-gel Tryptic Digestion of Proteins-Spots were cut out of 2-D gels, sliced into small pieces, and washed twice with 50% acetonitrile (Fisher) in 25 mM ammonium bicarbonate. The gel pieces were dried in a vacuum centrifuge. The proteins were digested overnight with 10 ng/l trypsin (sequencing grade, Promega Benelux, Leiden, The Netherlands) at 37°C. The peptide fragments were extracted twice with 5 l of water:acetonitrile:formic acid (5:14:1). After drying in a vacuum centrifuge, the lyophilized digest was dissolved in 0.1% (v/v) TFA.

MALDI-TOF Mass Spectrometry and Protein Identification-
The trypsin-digested sample was mixed with the matrix ((␣)-cyano-4hydroxycinnamic acid dissolved in 0.1% TFA of 50% acetonitrile aqueous solution) and then analyzed using the Ultraflex TOF/TOF mass spectrometer system (Bruker). Mass spectra were recorded in the positive ion mode with delayed extraction. Monoisotopic masses of peptides were analyzed using the Mascot search engine. The identification parameters were set as follows: data base, NCBInr; species, Mus musculus; all peptide masses are [M ϩ H] ϩ and monoisotopic; cysteines were treated with iodoacetamide; mass tolerance, 60 ppm; enzyme, trypsin; allow for one missed cleavage site. By combining observed molecular weight and pI on the 2-DE gel, the identities of some proteins were finally determined.
Western Blotting-Western blotting was performed to confirm the protein expression of BCL-2, BAX, and BID. Mouse liver proteins were separated by 12% SDS-PAGE and then transferred electrophoretically to a PVDF membrane (Amersham Biosciences). After being blocked for 4 h, the membrane was incubated with mouse monoclonal antibody for mouse BCL-2 (Zymed Laboratories Inc.), BAX (Zymed Laboratories Inc.), and BID (BD Biosciences Pharmingen) (at 1:1000 dilution) overnight and then with horseradish peroxidase-labeled goat anti-rabbit IgG (Zymed Laboratories Inc.) (at 1:2000 dilution) for 2 h. The blots were visualized with an enhanced chemiluminescent method kit (Wuhan Boster Biological Technology Ltd.). The protein concentration was determined by the Bradford method to ensure equal sample loading.

Mathematical Modeling and Computer Simulation
Model Development-Based on the data from transcriptomic and proteomic analyses, we developed a theoretical model to illuminate the mechanism of apoptosis induced by microcystin. Our modeling rested on a simplified description of the interaction between several apoptosis-related proteins. Because more than one apoptotic pathway has been analyzed in the computer modeling, we formulated three submodels to simulate this overall apoptotic phenomenon. The first submodel (Fig. 1a) is BID-induced BAX activation. It includes the following. 1) BID activates BAX by changing its conformation into the active form (31). Active BAX forms channels on the mitochondrial outer membrane and releases cytochrome c (32). Once cytochrome c is released, a cascade of apoptotic events occurs (33). 2) BCL-2 binds to BAX, blocking its activation (31). 3) When microcystin is injected, both BID and BAX are up-regulated to induce apoptosis; meanwhile BCL-2 is also up-regulated to control this process. 4) Every related gene has a threshold of tolerance for MC toxicity. If the concentration of microcystin is too high, the expression of related genes could be suppressed. The differential equations for this subsystem are as follows.
The definitions of state variables and rate constants are given in Tables II and III. The second submodel (Fig. 1b) is the antioxidant response and regulation of ferritin. It includes the following. 1) MC-LR induces iron generation, and iron helps the body generate ROS (34). 2) Ferritin contains both heavy and light chains; the heavy chain sequesters iron atoms by binding and accommodating them, whereas the light chain helps to expand the iron storage capacity (35,36). 3) Iron-regulatory protein (IRP) suppresses the expression of ferritin, but it is only activated transiently in response to ROS generation (37,38). In this subsystem, we postulate two assumptions. 1) MC-LR induces iron generation. 2) As the concentration of MC-LR increases, the duration of IRP activation is prolonged. The differential equations for this subsystem are given below.
g͑x͒ ϭ x n ⅐ e Ϫt/T (Eq. 15) q͑t͒ ϭ e Ϫt/T (Eq. 16) The last submodel (Fig. 1c) is the antioxidant function of peroxiredoxins (Prxs). This subsystem includes the following. 1) Prx1, -2, and -6 cause ROS stress by deoxidizing Prx1, -2, and -6, meanwhile causing their own overoxidization (39). 2) Prxs have different mechanisms to regenerate. Prx6 is regenerated very slowly so that its regeneration mechanism is only de novo synthesis. Prx2 is regenerated quickly with the retroreduction mechanism. Prx1 regenerates at an intermediate rate, the mechanism of which is unclear (40). By integrating this data, we formulated the differential equations of this subsystem as follows.
Computer Simulation-All the model equations described above were solved mathematically using the ODE15s and ODE23s routines of Matlab 6.5 (The Mathworks Inc., Natick, MA). The simulation programs were written in M files using common Matlab subroutines, and all the graphics were performed using Matlab 6.5.

RESULTS
Livers extracted from mice that had been injected with 60 and 70 g of MC-LR/kg of body weight over a period of 24 h exhibited massive intrahepatic hemorrhage.

Histochemistry
Hoechst 33258-stained liver sections revealed the presence of apoptotic cells in mice receiving 50, 60, and 70 g of MC-LR/kg of body weight for 24 h. Hepatic apoptotic cells progressively increased with the increasing MC-LR dosage (Fig. 2). Almost no apoptotic liver cells were found in mice treated with a dose of 50 g of MC-LR/kg, whereas the majority of liver cells underwent apoptosis in the mice treated with the MC-LR dose of 70 g/kg.

DNA Fragmentation
The induction of DNA fragmentation was demonstrated by treating BALB/c mice with different concentrations of MC-LR for 24 h (see Fig. 4). A ladder pattern of internucleo-somal fragmentation of DNA was apparent when mice were treated with 70 g of MC-LR/kg of body weight. These results indicated that MC-LR induced apoptosis in mouse hepatocytes and that the amount of fragmented DNA peaked at the highest dose of 70 g of MC-LR/kg of body weight (Fig. 3).

Transcriptomics
Microarray-The mRNA expression patterns of hepatocytes exposed to the four different dosages (0, 50, 60, and 70 g/kg of body weight) of MC-LR were compared using microarrays containing 96 different apoptosis-related genes. Table IV gives an overview of the number of genes that were up-and down-regulated during MC-LR treatment. The given ratios are the mean ratios of three arrays. MC-LR exposure modulated a total of 61 of 96 genes by more than 2-fold (up-regulation or down-regulation). Table IV classifies these genes into groups according to function. Significance was calculated using Student's t test.
Our results were further analyzed by applying a clustering     Quantitative Real Time PCR-We selected 10 important genes closely related to apoptosis for real time PCR to confirm the microarray data. The mRNA expression patterns of eight of the 10 genes are in agreement with the array data in at least two of the three dose points, indicating that the microarray data is reliable; the two inconsistent mRNA expression patterns are indicated in bold in Table V.

Proteomics
The proteomes of the four stages of apoptosis were compared using 2-DE. Three gels per sample were processed simultaneously. Coomassie Blue-stained 2-DE gel images were acquired with an image scanner and subsequently subjected to visual assessment to detect changes in protein expression levels and analyzed with PDQuest software. The FIG. 5. Clustering of some apoptosis-related genes modulated by MC-LR. A simplified gene ontology was automatically built by Cluster software and applied to the analysis of the 61 genes whose expression was up-or down-regulated by more than 2 times over control. Genes are grouped on the basis of the "similarity" of their expression pattern. A green-to-red color grading represents the ratios of gene expression levels. NAIP, neuronal apoptosis inhibitory protein; IAP, inhibitor of apoptosis protein; TNF, tumor necrosis factor; TNFSF, TNF superfamily; TNFR, TNF receptor; TN-FRSF, TNFR superfamily; ATM, ataxia telangiectasia mutated; fadd, Fas-associated death domain; RIP, receptor-interacting protein; CRADD, caspase and RIP adaptor with death domain; DFF, DNA fragmentation factor; CAD, caspase-activated DNase; CIDE, cell death-inducing DFF45-like effector; CARD, caspase-associated recruitment domain; TRIP, TNFR-associated factorinteracting protein; DAP, death-associated protein; API4, apoptosis inhibitor 4; ASC, apoptosis-associated speck-like protein containing CARD; DR, death receptor; LT-b, lymphotoxin B; Apaf-1, apoptotic protease-activating factor 1. matching analysis of paired gels was done in automatic mode, and further manual editing was performed to correct the mismatched and unmatched spots. Fig. 6 shows four representative 2-D gel images of the protein expression pattern in mouse liver cells after treatment with 0, 50, 60, and 70 g of MC-LR/kg of body weight. Reproducible protein expression profiles were found on gels (pH 3-10) following Coomassie Blue R-250 staining. In total, 492 spots were cut out of the 2-D gels, trypsin-digested, and then measured using a MALDI-TOF mass spectrometer. The peptide mass fingerprinting spectra were used to search a protein data base. More than 77.8% (383) of the proteins could be identified; of these, 65 spots were found to be significantly changed (defined as having at least two of the three dose points increased or decreased by 2.5 times more than the control). Among the 65 proteins, 35 protein spots were also found to have significant volume increases, whereas the other 30 spots were significantly reduced (Table VI). The 2-DE patterns of ferritin light chain (FHC), ferritin light chain (FLC), Prx1, Prx2, Prx6, and globin and hemoglobin families are shown in Fig. 7.
The expression of BCL-2, BAX, and BID proteins was confirmed by Western blotting analysis. The protein expression trends of BCL-2, BAX, and BID were consistent with their mRNA expression trends shown in Fig. 8. Fig. 9. Injection of MC-LR was at time 0. Results of the gene expression simulation were roughly consistent with the microarray data (the consistency of transcriptional data and protein data was proved by Western blotting as described above). At a concentration of 50 g/kg MC-LR, the expression of Bax, Bid, and Bcl-2 were all significantly up-regulated, increasing 1500-, 40-, and 4000fold, respectively. At MC-LR concentrations of 60 and 70 g/kg, Bax maintained a high level of expression that was 1000 times greater than that exhibited by the control. Conversely the expression of Bcl-2 and Bid peaked at ϳ3-5 h and then decreased sharply to the normal level or lower. In addition, the concentration of active Bax at a MC-LR concentration of 50 g/kg was 30-fold greater than at 60 and 70 g/kg after 24 h.

BID-induced Apoptosis Pathway-The simulation of BIDinduced apoptosis is shown in
ROS Stress Apoptosis Pathway-The simulation of ROSinduced apoptosis is shown in Fig. 10. As the concentration of MC-LR increased, the generation levels of both iron and ROS also increased (Fig. 10, a and b).  dose-dependent fashion (Fig. 10, c and d). At 70 g/kg, however, the expression levels of both ferritin chains were lower than at 60 g/kg. Levels of assembled ferritin followed the same pattern of expression as that of the individual chains (Fig. 10e).
We also simulated the kinetics of peroxiredoxin concentration versus time (Fig. 10, f-h.) At low concentrations of MC-LR, Prx6 was down-regulated with time, whereas Prx1 was up-regulated in the first 4 h after exposure and then downregulated to some extent. Increasing concentrations of MC-LR resulted in the down-regulation of both Prx1 and Prx6 expression levels. The expression of Prx2 showed a different scenario, increasing until it peaked at ϳ12-16 h after MC-LR exposure and then decreasing to some extent.
Absolute expression levels increased with the concentration of MC-LR. The above results are all consistent with the proteomic data.

DISCUSSION
In recent years, numerous cosmopolitan lethal animal poisonings and a number of cases of human illness caused by cyanobacteria bloom and their toxins have drawn the attention of the World Health Organization, an increasing scientific community, and the public to cyanobacteria bloom and their toxins (1). The World Health Organization has set a safety guideline for MC concentration at a value of 1.0 g/liter in 1998. In this study, we specifically investigated MCs, cyanobacterial toxins that have both chronic and acute biological consequences resulting in liver tumor promotion and massive intrahepatic hemorrhaging, respectively (3,(41)(42)(43). MC hepatotoxicity is closely related to the documented apoptotic effects of MCs on cells and tissues (28 -30). Unfortunately the exact apoptotic mechanism induced by MCs and the molecules that play crucial roles in that apoptotic process have not yet been clearly identified. The objective of our current study was 3-fold: to determine the apoptotic consequences of MC-LR exposure on mouse liver cells, to explore the exact apoptotic mechanism of MCs, and to find related biomarkers.
To study these effects, we injected mice with different concentrations of MC-LR (0, 50, 60, and 70 g/kg of body weight) and then sacrificed them 24 h later, harvesting their liver tissue for detailed investigation. Our histochemistry analysis showed a typical morphologic change of nucleus shrink-  age in hepatocytes exposed to the highest MC-LR concentration. We also used DNA fragmentation assay to record a DNA ladder at a 70 g/kg of body weight dosage of MC-LR. Moreover flow cytometry analysis revealed that MC-LR at three dosages (50, 60, and 70 g/kg of body weight) exhibited an increasing dose-dependent cellular apoptotic effect. We were able to use transcriptomic and proteomic analyses to identify components of apoptotic signaling pathways. Of special interest were the BCL-2 family, ferritin, and peroxiredoxins, which are closely related to the apoptotic process (31,34,35). BCL-2 family members are of particular importance as they are potent regulators of apoptosis that can influence the permeability of the outer mitochondrial membrane (44 -46). This family consists of both pro-and antiapoptotic members that elicit opposing effects on mitochondria, including the antiapoptotic protein BCL-2 and the proapoptotic proteins BAX and BID. BAX, BID, and other proapoptotic members can be further divided into two categories: 1) the multidomain proapoptotic members such as BAX and BAK and 2) the BCL-2 homology 3 domain-only proapoptotic members such as BID and BIM.
Due to the inherent limitations imposed by the sensitivity of in-gel proteomics, we were unable to obtain protein spots for BCL-2, BID, and BAX from a 2-DE image despite using silver staining and pH 4 -7 IPG strips. We did, however, validate the protein expression profiles of these three key apoptotic pathway proteins by Western blot (Fig. 8), and the expression trends are consistent with their microarrays.
BCL-2 family proteins responded differently to different concentrations of MC-LR. At low MC-LR concentrations (50 g/kg of body weight), an over 4000-fold increase in Bcl-2 mRNA expression was revealed (Table IV). Because BCL-2 and other antiapoptotic members of the BCL-2 family localize at the mitochondria where they can preserve mitochondrial integrity by blocking the release of soluble mitochondrial intermembrane proteins, it is conceivable that the up-regulation of BCL-2 is merely the host response to toxin invasion, protecting the host from destruction by apoptosis, a conclusion validated by computer simulation (Fig. 9).
Ectopic overexpression of both Bax and Bid were detected at a dose of 50 g of MC-LR/kg of body weight (Table IV). It is known that upon induction of apoptosis, BAX oligomerizes with BAK into large complexes that form pores in the lipid bilayer and facilitate the release of cytochrome c and other factors (47). Thus, the up-regulation of BAX at relatively low doses of MC-LR could initiate the consequent apoptotic events with BID serving as an initiator of BAX. t-BID, the active form of BID, induces oligomerization of BAX and BAK (31). We showed in our study that mRNA expression of Bid increased noticeably in mouse hepatocytes exposed to MC-LR, facilitating the activation of Bax and promoting the progress of apoptosis.
At high concentrations of MC-LR (70 g/kg of body weight), the expression of Bax remained high, but the expression of Bid and Bcl-2 dropped markedly. This is presumably because high MC-LR concentrations cause the spontaneous modulation of apoptosis to become unstable, making it difficult for BCL-2 family proteins to control the process.
Building on the solid foundation of our experimental results, we used computer simulations to further explain the details of this process. As seen in Fig. 9, when MC-LR concentration reached the upper threshold limit, the expression of these proteins was suppressed, indicating that the main promoter and inhibitors of this pathway were both disabled and that the modulation mechanism of the BCL-2 family does not work under conditions of high MC-LR concentrations. It also indicated that BID and BCL-2 were more sensitive to MC-LR concentration elevation than was BAX. Nevertheless apoptotic events persisted in high MC-LR concentrations, indicating that there must be another modulation mechanism for apoptosis under conditions of high MC-LR concentration.
Our proteomic analysis revealed that exposure to MC-LR also shifted the protein expression levels of some important hepatocyte proteins, including FHC, FLC, Prx1, Prx2, and Prx6, all of which are closely related to oxidative stress. This shift was especially noticeable at MC-LR concentrations of 70 g/kg of body weight. Conversely at MC-LR concentrations of 50 g/kg of body weight, the expression levels of these proteins showed no significant difference compared with control.
Of these proteins, ferritin, which is composed of a light chain (FLC) and a heavy chain (FHC), serves as the major iron-binding protein limiting the catalytic availability of iron for participation in oxygen radical generation (38). FHC has a potent ferroxidase activity that catalyzes the oxidation of ferrous iron, whereas FLC plays a role in iron nucleation and protein stability (35). The observed up-regulation of both FHC and FLC (Fig. 7) was probably caused by MC-LR in vivo enhancement of iron release (48) and ROS production (data not shown).
Our proteomic data demonstrated that ferritin played a crucial role in the apoptotic processes. To further clarify this result, we also established a relationship between the expression of FHC and FLC with the production of iron and ROS. Using the second submodel ( Fig. 1), our simulation showed that the expression level of ferritin rose correspondingly to the production level of iron (Fig. 10). Iron is ubiquitous in cells and present in the structure of many enzymes and proteins that catalyze redox reactions, enabling iron to generate radical species according to the following equation: Fe 2ϩ ϩ H 2 O 2 3 Fe 3ϩ ϩ HO ⅐ ϩ OH Ϫ . Disturbance of cellular iron homeostasis affects the production of ROS and oxidative damage. ROS causes an increase in cytosolic calcium and has been observed in cells undergoing oxidative stress (34). As a negative modulator in oxidative damage, ferritin gave rise to the overproduction of ROS caused by MC-LR (Fig. 10).
In contrast to FLC and FHC, both of which were up-regulated, MC-LR shock produced varied expression trends in peroxiredoxin family members. Prxs are a ubiquitous family of antioxidant enzymes that also control stimuli-induced peroxide levels, which have been shown to mediate signaling cascades leading to cell proliferation, differentiation, and apoptosis (40). We found that as MC-LR concentration increased Prx1 and Prx6 were negatively regulated, whereas Prx2 was positively regulated (Fig. 7).
These variant expression patterns of the peroxiredoxin family proteins are due to the differences in their regeneration mechanism. The mode of recycling sulfenic acid back into a thiol (and thus regenerating peroxiredoxins) distinguishes the three peroxiredoxin enzyme classes from one another (49,50). Prx2 is the most rapidly regenerated protein followed by Prx1 and finally Prx6, which is regenerated very slowly. Further study of these mechanisms demonstrated that the fast regenerating peroxiredoxins, such as Prx2, are regenerated at least in part by a retroreduction mechanism. When the host is burdened with ROS overload induced by MC-LR, Prx2 has the most rapid regeneration speed because it has two regeneration mechanisms, one of which is retroreduction.
Proteomic patterns showed that the expression trends of Prx1, Prx2, and Prx6 after MC-LR exposure were all different; nevertheless these patterns all indicated the elevation of ox- idative stress in vivo. Prx2 appears to be up-regulated in response to MC-LR treatment, whereas Prx1 and Prx6, both of which have poor regeneration speeds because each has only one regeneration mechanism, cannot easily recover from oxidative stress overoxidation (50) especially under high MC-LR concentrations.
Furthermore the expression of many other proteins related to oxidative reactions, such as glutathione S-transferase and thiosulfate sulfurtransferase, shifted during MC-LR treatment (Table VI), indicating the overproduction of oxidative stress in vivo (36). Also globin and hemoglobin family proteins were all clearly up-regulated (Table VI and Fig. 7), which is very consistent with the massive intrahepatic hemorrhage induced by MC-LR.
Based on the above analysis, we propose a hypothetical model of pathways to explain the apoptotic effects of various concentrations of MC-LR on mouse liver cells (Fig. 11). In the first pathway, dubbed the BID-BAX-BCL-2 pathway (Fig. 11a), low doses of MC-LR lead to apoptosis primarily via BCL-2 family proteins. In the second pathway, the ROS pathway (Fig.  11b), high doses of MC-LR give rise to apoptosis directed by ROS. The present study determined that MC-LR induces apoptotic effects in mouse liver cells and for the first time illustrated that different concentrations of MC-LR induced apoptosis through two independent pathways. Our study of the pathological effects of MC exposure showed that relatively low doses of MC-LR do not cause immediate host death but instead produce low level injury. Sustained and low level tissue exposure to MC-LR can lead to progressive liver tissue fibrosis and even tumorigenesis (24). Higher doses of MC-LR, on the other hand, involve mitochondrial membrane rupture, massive intrahepatic hemorrhaging, and damage to hepatic architecture that is indicative of cytoskeletal disruption in the liver (25). Our results give an indication that the two different pathological processes should be associated with different signaling transduction pathways.
Up to now, only PP1 and PP2A have been shown to be targets of MCs (11). In this study, some molecules, such as ferritin, Prx1, -2, and -6, were found to be clearly modulated by MC-LR. To our knowledge, these molecules have not been reported previously as being involved in MC shock and could be potential biomarkers for hepatotoxicological mechanisms. These results indicate possible directions for further exploration into the effects of MCs on tumor-promoting mechanisms. Moreover more effort is required to further validate that these molecules represent potential MC shock biomarkers that can aid in detection and detoxification.