INTRODUCTION

The therapeutic effect of arsenic trioxide (As2O3) has been affirmed for acute promyelcytic leukemia, but some other leukemia and a variety of human tumors, especially the solid tumors, are less sensitive or insensitive to As2O3 treatment 1. It is noticed that a number of chemotherapeutic drugs, including As2O3, kill tumor cells with dependence on cellular reactive oxygen species (ROS), and that manipulation of cellular redox state may enhance the cytotoxicity of these drugs 2, 3, 4. We previously reported that the sensitivity of different tumor cells to As2O3–induced apoptosis is positively related to their inherent cellular ROS level and the elevation of intracellular ROS level by ROS generation agent could enhance the effects of As2O3 5, 6.

In an attempt to search for a natural and low-toxic ROS generator, we recently found that emodin (1,3,8-trihydroxy-6-methylanthraquinone), a natural anthraquinone derivative, rich in Chinese herbal medicine rhubarb, could generate ROS intracellularly. When co-treating tumor cells with As2O3, emodin could sensitize cells to As2O3-induced apoptosis in HeLa and other cell lines derived from the solid tumors. Two transcription factors known to be sensitive to cellular redox, nuclear factor κB (NF-κB) and activator protein (AP-1) were inhibited by emodin-elicited ROS elevation, whereas a series of major apoptotic signaling events, i.e. MTP collapse, cytochrome C release and caspases activation were promoted 7, 8. But the precise molecular mechanism through which ROS facilitates As2O3-induced tumor cell apoptosis has not been elucidated. In order to study the gene expression profile affected by emodin-mediated ROS generation and emodin-enhanced apoptosis, we performed cDNA microarray in HeLa cells exposed to As2O3/emodin cotreatment or As2O3–only treatment at two time points using BioStar H-40s 4096-clone cDNA microarrays.

In this study, we demonstrated the differences in the profiles of mRNAs expression in As2O3–only treated and As2O3/emodin cotreated HeLa cells. The global effect of emodin was seen: 793 and 480 genes, respectively at the early and late time points, differed in expression level by a factor of ≥2. Through such a profound affection on gene expression, emodin completes its enhancement of arsenic cytotoxicity.

MATERIALS AND METHODS

Cell culture and treatment

HeLa cells were obtained from American Type Culture Collection (Manassas, VA). Cells were maintained in DMEM (GibcoBRL, Gaithersburg, MD) with antibiotics and 10% fetal bovine serum, in a humidified atmosphere with 5% CO2 at 37°C. As2O3, emodin (6-methyl-1, 3, 8-trihydroxyanthraquinone) and N-acetyl-cystein (NAC) were purchased from Sigma (St, louis, MO). Cells were seeded in 24-well plates at 1×105cells/ml and might be incubated with As2O3 (2 μM) or together with emodin (10 and 30 μM) or plus NAC (1.5 mM), with daily change of drug-containing medium.

Analysis of cell viability and apoptosis

After exposed to the agents for 48 h, cell viability was assayed using CellTiter 96 Aqueous Non-Radioactive Cell Proliferation Assay (MTS) Kit (Promega, Madison, WI), following the manufacturer's instructions. Absorbance at 490 nm was directly proportional to the number of living cells in culture.

Apoptotic rates were analyzed by flow cytometry using Annexin V-fluorescein isothiocyarate (FITC)/propidium iodide (PI) kit (BD PharMingen, San Diego, CA) in which Annexin V bound to the apoptotic cells with exposed phosphatidylserine. Samples were prepared according to the manufacturer's instruction and analyzed by flow cytometry on a FACS Calibur (Becton Dickson, San Diego, CA).

Dectection of ROS generation

2,7-dichlorodihydrofluorescein diacetate (DCFH-DA, Sigma) was used as ROS probe. Cells were first exposed to the agents for 30 min, with pre-incubation of NAC for 4h if it was used. DCFH-DA at 10 μM was then incubated with cells for 15 min before a brief PBS rinse and an immediate flow cytometric assay.

Microarray sample preparation, detection and analysis

cDNA microarray was performed on the pair of the samples. As2O3–only sample referred to the cells treated with 2 μM As2O3, whereas As2O3/emodin cotreatment sample referred to the cells treated with 2 μM of As2O3 in combination with 30 μM emodin. Assay was based on two time points. We harvested cells after 6 h of treatment, in an assumption that emodin-induced upstream signaling and early changes of cell function occurred at this time point. The 24 h treatment was chosen as later time point because it allowed to see the later gene transcription change induced by emodin, while no substantial apoptosis was yet happened and cell integrity was remained then.

Total RNA was extracted using the Trizol® reagent (Invitrogen, Carlsbad, CA). The subsequent procedures for sample preparation and microarray were performed by Biostar Genechip Inc. (Shanghai, China) according to its protocol subject to BioStar H-40s genechip. Briefly, cDNA was reverse transcripted. The fluorescent cDNA probes were prepared through reverse transcription. The probes for As2O3-only samples were labeled with Cy3-dCTP, and those for As2O3/emodin cotreatment samples were labeled with Cy5-dCTP. Fluorescence intensity was measured for each gene spot. After background subtraction and the whole-chip data normalization, the ratios of gene expression differences between two sets of samples were obtained. BioStar H-40s microarray consisted of 4096 novel or known genes including control system and effective genes (provided by Biostar Genechip Inc). The control system consists of 96 housekeeping genes as loading control; 16 plant genes, and spotting solution (without DNA, 16 spots) as negative control spots in the array. Chip 1 contained the samples at early time point (6h), and chip 2 at late (24h).

BioStar H-40s genechip covers a series of genes involved in cell components and functions. The differentially expressed genes were further analyzed and classified by Gene Ontology (GO). We identified several categories in which we are particularly interested.

Real time quantitative PCR

Four mRNA molecules were checked by real time quantitative PCR (ABI Prism 7000 Sequence, PE, USA) to validate the reliability of microarray. Real time PCR was also used to evaluate the reversal effect of NAC, in which ten molecules were quantitated in cells exposed to As2O3/emodin cotreatment or to As2O3/emodin plus 1.5mM NAC for time similar to that for microarray. RNA was extracted as described above, and cDNA was synthesized using RevertAidTMM-MuLV Reverse Transcriptase (FERMETAS AB, Vilnius, Lithuania). Real-time PCR was performed under conditions described in the SYBR Green I PCR kit (ABI, Branchburg, NJ) (50°C for 2 min; 95°C for 10min; 95°C for 15s; 60°C for 1 min; 40 cycles). β-actin was used as internal control. The experiments were duplicated. The primers for genes were as follows. AKT1: forward: 5′-CGC AGT GCC AGC TGA TGA AG-3′ and reverse: 5′-ACA GTC TGG ATG GCG GTT GT-3′; TXNRD1: forward: 5′-GAC GTC ACT GTT ATG GTT AG-3′ and reverse: 5′-CTG AGC TAC TAC TCT GAG TC-3′; FLJ14515: forward: 5′-GTC AAC CTG ACT GTG CGA TT-3′ and reverse: 5′-CTG GAT CAG AGA AGG CAG CA-3′; NFKBIA: forward: 5′-CTG ATG TCA ACA GAG TTA CCT ACC AG-3′ and reverse: 5′-CGT GAA CTC TGA CTC TGT GTC ATA G-3′; HSPF1: forward: 5′-ATT CCA GCT GAT ATC GTC TT-3′ and reverse: 5′-ATC GTC CTG CCG TCC AGA GT-3′; PRKCD: forward: 5′-CGT TCC TGC GCA TCG CCT TC-3′ and reverse: 5′-CGT CGA CTT CCA CTC AGG AT-3′; VDAC1: forward: 5′-ACA AGC TCA GGC TCA GCC AA-3′ and reverse: 5′-CAG TCC ACG TGC AAG CTG AT-3′; α-actin: forward: 5′-TAT CGA GCA CGG CAT CAT CA-3′ and reverse: 5′-GTC ATC TTC TCG CGG TTG GC-3′; ROCK2: forward: 5′-AGG TAT CTG TAC ATG GTA AT-3′ and reverse: 5′-AGC ATG TTG TCA GGC TTC AC-3′; PIK3C3: forward: 5′-ATG TAG AAG CAG ATG GAT CA-3′ and reverse:5′-CAA TCT ATC CAG CCA ATC TAC-3′.

RESULTS

Emodin altered the cellular redox state and sensitized HeLa cells to arsenic cytotoxicity

In our previous studies the dose of emodin used in combination with As2O3 was 10 μM 7, 8. In order to visualize the gene expression profile clearly in the present study a higher dose was tested. The results of ROS measurement showed that addition of emodin at 10, 30, and 50 μM to HeLa cells for 15 min cause an immediate and dose-dependent ROS generation (data not shown). MTS assay showed that emodin at 10 μM had no effect on cell viability, and at 30μM caused a mild suppression, but at 50 μM led to an obvious decrease in cell viability (data not shown). Therefore emodin at 30 μM was selected for its co-treatment with As2O3.

Exposure of HeLa cells to the As2O3/emodin cotreatment elicited an immediate and dramatic increase of ROS in HeLa cells, whereas As2O3 – caused ROS elevation was mild but continuous (Fig. 1).

Figure 1
figure 1

ROS level (flow cytometry). HeLa cells were treated with As2O32 μM alone or together with emodin 30 μM or plus the antioxidant N-acetyl-L-cysteine (NAC) 1.5 mM for the indicated time (NAC was pre-incubated with cells for 4h). Relative ROS level was represented by the folds of DCF intensity compared to the untreated cells. As2O3 elicited a mild but continuous elevation of cellular ROS level, whereas two-drug combination dramatically and rapidly augmented ROS level. NAC partially prevented this increase (mean±SD, n=3).

Full size image

As2O3/emodin cotreatment caused obvious reduction in the number of viable cells, compared to the untreated and As2O3–only groups (Fig. 2). Cell number reduction was attributed mainly to the increased apoptosis. Emodin at 30 μM rendered a strong synergistic cytotoxicity to As2O3, while emodin alone caused no significant cell killing effect, indicating that it enhance As2O3–induced apoptosis (Fig. 3).

Figure 2
figure 2

Cell viability assays (MTS). HeLa cells were treated with As2O3(2 μM), emodin (30 μM), As2O3 /emodin (30 μM) or together with NAC (1.5 mM) for 48h. As2O3 caused an inhibition in cell viability and emodin enhanced As2O3–induced viability decrease, while emodin alone had minimal effect on cell viability. NAC partially prevented this decrease (mean±SD, n=3).

Full size image
Figure 3
figure 3

Apoptotic rates analysis (Annexin V/propidium iodide flow cytometry). HeLa cells were treated with As2O3(2 μM), emodin (30 μM), As2O3 /emodin (30 μM) or together with NAC (1.5 mM) for 72h. Cotreatment of HeLa cells with As2O3 and emodin caused increased apoptosis rate compared to the As2O3–only and emodin-only group. Enhancing effect of emodin could be reversed by NAC (mean±SD, n=3).

Full size image

Expression of varied genes was altered during emodin-facilitated apoptosis

Cells exposed to As2O3 2 μM in combination with emodin 30 μM were set as As2O3/ emodin-cotreated sample for microarray, as this dose of emodin itself had minimal effect on cell viability, but it could result in an obvious synergistic cytotoxicity. The scanning results of hybridizing signals on gene chips displayed the gene expression alteration between As2O3–only and As2O3/emodin samples at both early (chip1) and late (chip2) time points (Fig. 4). We noticed that the normalized ratio of nearly 80% and 90% of the loading control spots were around 1 respectively in two chips and both negative control spots showed low intensity after hybridization (fluorescence intensity<200), which proved the reliability of the experiment.

Figure 4
figure 4

The scanning results of hybridizing signals on gene chips displaying the gene expression alteration between As2O3–only and As2O3/emodin samples.

Full size image

The genes with a difference of ratio≥2 folds were regarded as differentially expressed ones between As2O3–only and As2O3/emodin samples. There were respectively 793(chip1) and 480(chip2) differentially expressed genes in two chips. Analysis of the filtered differently expressed genes using Gene Ontology (GO) showed that ROS induced expression changes of a variety of genes involved in many aspects of cell function. On the basis of the GO consistent descriptions for gene products, we summarized the up- and down-regulated genes into six categories, i.e. apoptosis, redox, cell signaling, organelle functions, cell cycle and cytoskeleton (Tab. 1). A part of these genes with their Genebank-ID, ratio of mRNA levels, gene definition and GO terms were listed (Tab. 2). A part of listed genes was summarized for visualizing the extent to which they changed and for discussion (Fig. 5). The results of real-time PCR of ak027421, NFKBIA, VDAC1 and α-actin genes were consistent with those of cDNA microarray. While the expression level of these four genes were 5, 7.9, 3.3, 20 folds changed on genechips, they were respectively proved to be 6.9, 6.8, 2.08, 15.1 folds changed in the same direction in real time quantitative PCR (Fig. 5).

Table 1 The number of up and down-regulated genes in 6 categories (fold≥2)
Full size table
Table 2 Part of the differently expressed genes(fold change ≥ 2) with their genebank accession number, fold change, gene definition and GO annotation
Full size table
Figure 5
figure 5

A part of the listed genes showing the extent to which they changed. These genes belong to different categories and play important roles in redox-enhanced arsenic cytotoxicity.

Full size image

The results indicated that, in addition to the genes related to apoptosis and redox regulation, emodin profoundly affected gene expression profile, which ultimately led to an enhanced apoptosis.

Effects of emodin on gene expression alteration could be reversed by anti-oxidant agent NAC

Based on the gene profiles altered by As2O3/ emodin-cotreatment, we selected 10 molecules that involved in the varied aspects and had a substantial alteration on either chip. Real-time quantitative PCR was performed to investigate if their altered expression was resulted from a ROS elevation caused by the combination of As2O3 and emodin. Cells were exposed respectively to As2O3 alone, As2O3/ emodin-cotreatment or As2O3/ emodin-cotreatment plus NAC. The quantitative values that represented the copy numbers of each mRNAs were calculated into a fold ratio, in which co-treatments without or with NAC were versus As2O3–only treatment. Results showed that in 8 of 10 molecules NAC partially or completely reversed the effect of co-treatment, indicating that the differential expression was dependent on ROS (Fig. 6).

Figure 6
figure 6

The comparison of eight genes alterations between samples with As2O3/emodin cotreatment and with cotreatment plus NAC (real-time PCR). The copy numbers for each molecule were obtained via normalizing the absolute quantitative values to β-actin control. Fold change stands for a ratio in which the values for the cotreatment without or with NAC were divided by those for the As2O3–only treatment (mean±SD, n=2).

Full size image

DISCUSSION

The anticancer effect of As2O3 as a therapeutic agent has been studied for several years. Arsenic treatment has been shown to influence the cell signaling of mitogen-activated protein kinases (MAPKs), activator protein-1(AP-1), nuclear factor kappa B (NF-κB), p53, etc 9. As2O3–induced apoptosis were accompanied by activation of death signals: caspase 3, 8 and 9 10, 11. The generation of ROS appears to consistently accompany arsenic exposure and many evidences suggest that ROS may be a determinant of cellular susceptibility to arsenic 6, 12, 13, 14, 15. Manipulation of ROS may thus be a way to expend the anticancer therapeutic spectrum of As2O3. Therefore it is interesting to find a clinically safe drug that has synergistic effect with As2O3 via elevating the ROS level, while keep As2O3 at a clinically acceptable dose.

As a major component of the widely used Chinese herbal medicine rhubarb, emodin has been used in the remedies for many human diseases including inflammation and cancers. We recently showed that emodin could remarkably enhance the therapeutic effect of As2O3 in the in vivo models at the doses safe to the animals. Emodin exerts its effects via a ROS-mediated dual signaling regulations i.e. the enhancement of pro-apoptosis and the simultaneous inhibition of anti-apoptosis 8. In the present study, we aimed to explore through what gene expression profile emodin enhanced As2O3 cytotoxicity.

Since the microarray focuses the difference existed between a pair of samples, we defined one set of samples as As2O3–only and another as As2O3/emodin cotreatment in order to see the arsenic-based effect of emodin. As2O3/emodin combination augments apoptotic rate compared with As2O3–only treatment. This effect could be partially reversed by antioxidant NAC, which proved that emodin enhances the effect of As2O3 via the generation of ROS. It is true that emodin at 30 μM itself can cause intracellular ROS generation, while it has only minimal impact on cell viability and apoptosis. This may be explained by that the synergistic effect of emodin is dependant on ROS and based on arsenic's efficacy. A simple ROS elevation by emodin at this extent and duration would not elicit dramatic signaling events and the resultant cell death. But rather, As2O3 has complex cytotoxic effects which emdoin may not possess. Therefore, it was based on the cytotoxicity of As2O3 that emodin at low dose plays its synergistic anticancer role. HeLa cells exposed to the As2O3/emodin cotreatment have more rapid and violent ROS generation. The sudden ROS elevation (dozen of minutes to a few hours after treatments) may function as a signaling trigger to cause wide gene transcriptional alteration, and finally lead to an enhanced apoptosis. This also explains why the time points we selected to do microarray assays (6 and 24 h after treatments) are later than ROS elevation and earlier than apoptosis occurs (2 to 3 d after treatments).

Figure 7
figure 7

A hypothetical model for how emodin facilitates As2O3–induced apoptosis in HeLa cells through a series of regulatory events on gene expression. The boxes represent different groups of genes involved in this process. Emodin causes intracellular ROS generation and perturbs cellular redox. Gene expression is pronouncedly regulated by ROS, which play roles in signal transduction, organelle functions, cell cycle checkpoint, and cytoskeleton. The altered expression of these genes, together with their interplay, sensitizes HeLa cells to arsenic cytotoxicity.

Full size image

Emodin enhances As2O3–induced apoptosis through redox state regulation

Although HeLa cells do not exhibit apoptotic morphologic changes at 6h and 24h time points, emodin-mediated regulation of genes related to apoptosis and antiapoptosis is evident in both chips. For example, the expression of BAG5 and L0C51283 genes diminished in chip1. L0C51283 gene is known to be antiapoptotic. BAG5 belongs to BAG1 family, and BAG1 usually functions as an antiapoptotic protein through interaction with Bcl-2, RAF kinase, growth factor receptors and HSP70 16. Antiapoptosis-related genes AKT1 and RAF1 are down-regulated in chip2. AKT1 gene encodes a serine-threonine protein kinase. Once activated, AKT1 phosphorylates and inactivates components of the apoptotic machinery 17. RAF1 is involved in MAP kinase-dependent cell cycle control pathways. Remarkably, this down-regulation of antiapoptotic proteins can be abolished by NAC, one of the classic anti-oxidant agents, implying the alteration is dependent on a ROS elevation caused by emodin.

Emodin augments intracellular ROS level, and ROS fluctuation influences cellular redox state, which may result from or result in the expression of redox-related genes. At the 6h time point, two thioredoxin (Trx) related genes, FLJ20511 and PDCC, are down-regulated, while glutathione transferase zeta 1 (GSTZ1) and glutathione peroxidase 1 (Gpx1) genes are more expressed. At the 24h time point, thioredoxin reductase 1 (TXNRD1) gene and glutathione S-transferase M3 (GSTM3) gene are substantially less expressed, meanwhile the increased expression is seen for Gpx1 and supreoxide dismutase1(SOD1). Trx reductase participates in the reduction of oxidized Trx. Thus, the decrease of reduced formed-Trx and the increase of oxidized formed-glutathion in the cells likely function coordinately as an early transcription response to the emodin treatment. This attenuation of anti-oxidant system will certainly lead to further cellular redox unbalance, though ROS scavenger enzyme SOD, on the other hand, is produced more to rescue the cell from the redox catastrophe.

Emodin influences arsenic effects on multiple signaling molecules

ROS has been implicated as a pivotal regulator in nuclear factor κB (NF-κB) activation 18. Under normal conditions, the inhibitory protein IκB binds to NF-κB, preventing its access to DNA. Upon phosphorylation of IκB by the NF-κB-inducing kinase (NIK)/IκB kinase (IKK), NF-κB gets free and translocates to the nucleus. Therefore, NIK and IKK are related to the activation of NF-κB. In the present study, NFKBIA (usually called IκB -alpha) is substantially overexpressed. Meanwhile, FLJ14515, a gene involved in NIK-I-κB /NF-κB cascade is down-regulated in both chips. Besides, expression of SAE1, a SUMO-1 related gene, is reduced in chip2. Site specific SUMO-1 attachement is necessary to the nuclear localization of NEMO—the regulatory subunit of the cytoplasmic IKK complex, and ultimately leads to the activation of IKK in the cytoplasm 19. Overall, the down-regulation of genes relate to IKK and the up-regulation of IκB indicate that the activation of NF-κB is inhibited at the very up-stream of the pathway during this process. In our previous studies, we found that exposure of HeLa cells to As2O3/ emodin caused a suppressed expression of NF-κB-driven genes 8, and this suppression might be linked to an attenuated DNA binding capacity of NF-κB (data unpublished). We also demonstrated that NF-κB activation plays an antiapoptotic role under resting state and in response to As2O3 exposure 7. Taken together, we conclude that emodin exerts an inhibitory effect on anti-apoptotic NF-κB activation at multiple levels. Arsenic has been shown to have varying effects on NF-κB activity, depending on the cell type and treatment intensity 9. Activation of NF-κB stimulated by arsenic may confer resistance to arsenic cytotoxicity. Therefore the inhibition of NF-κB activation by emodin must contribute, at least partially, to its enhancement of arsenic cytotoxicity. Real time PCR shows that up-regulation of NFKBIA is largely abolished and down-regulation of FLJ14515 is completely reversed by NAC. This shows that emodin affects NF-κB pathway in a ROS-dependent fashion.

The ATM, PPM1D and PIG8 genes that are involved in activation of p53 and induction of apoptosis by p53 are all up-regulated in chip1. The activation of ataxia-telangiectasia mutated (ATM) kinase is necessary to the induction and activation of p53 protein after ionizing radiation (IR) and certain other forms of oxidative stress 20. Expression of Gpx is also increased. The induction of Gpx is an early event following p53 activation 21.

HSF1gene is up-regulated by emodin, and this up-regulation is again antagonized by NAC. Heat shock transcription factor 1 (HSF1) is responsible for the induction of heat shock protein in response to stress 22. Several heat shock proteins are known to promote cell survival and prevent apoptosis during a wide variety of stress conditions including oxidative stress 23.

In addition to these genes known to be the signaling molecules involved in arsenic effects, emodin also induce the alternative signal pathways including PKC and AKT (all down-regulated) at different time points. Most of these molecules participate in the oxidative stress activate signal pathways, which implies that emodin influences arsenic effects on signaling via generation of ROS. The decrease of PKC delta, named PRKCD, and AKT1, is abolished by NAC, evidencing directly that the effects of emodin is ROS-dependent. Crosstalk exists among these different signaling pathways and the final occurrence of apoptosis reflects the relative activity in survival/death balance.

Emodin enhances arsenic effects on different organelle functions

As compared to As2O3 alone-induced apoptosis, As2O3/emodin cotreatment affected numerous genes encoding for organelle proteins, i.e. the proteins of mitochondrion, endoplasmic reticulum (ER), Golgi apparatus and lysosome. More dramatic gene expression changes linked to organelle structures and functions occurred at the early time point during emodin-enhanced apoptosis, among which many are known related to the initiation of cell death pathways.

Most of the genes encoding mitochondrial complex I and III are over-expressed in two chips. The genes encoding ATP synthase/H+ transporting mitochondrial F0 and F1 complex subunits are down-regulated in chip1, which indicated that the uncoupling of electron transport and oxidative phosphorylation occurred at the early time. VDAC-1 is part of a protein complex commonly referred to as the permeability transition (PT) pore 24. Its decreased expression may be related to the changes of PT pore. The gene expression changes involved in PT pore may be related with what we reported that HeLa cells undergo a reduction in mitochondiral membrane potential (ΔΨm) with As2O3/emodin cotreatment 8. NAC prevents the down-regulation of VDAC-1.The thioredoxin reductase 1 (TXNRD1) gene is down-regulated in chip2, which indicates a perturbation in mitochondrial scavenging of ROS and mitochondrial apoptosis signaling 25.

Many genes encoding ER membrane, lumen and functional proteins are less expressed, such as genes Calumenin and COPA, which were involved in protein folding, sorting and transport between the ER and Golgi compartments 26, 27. Reticulocalbin and calnexin genes, both are Ca2+binding ER luminal proteins, are respectively up and down-regulated in chip1. The expression of inositol 1,4,5-trisphosphate receptor (InsP3R) gene is blocked in chip2. InsP3R encodes a ligand-gated Ca2+-release channel and plays an important role in intracellular Ca2+ signaling in a wide variety of cell types 28. The expression of TRAF2 and eIF2-α genes is altered and they are reported to be the downstream effectors of Ire1-α and PERK, both involved in the ER unfolded protein response (UPR) 29, 30. Thus, we suppose that two major mechanisms by which ER participates in the apoptosis, UPR and Ca2+ signaling 31, are influenced during the emodin-facilitated apoptosis.

The GD3 synthase (α-2,8-sialytransferase) and PI(3)K genes that are known as apoptosis-signaling proteins located at Golgi membranes are respectively up-regulated in chip1 and chip2. GD3 could shuttle to mitochondria and induce mitochondrial membrane permeabilization (MMP) 32.

Lysosomal rupture is associated with the release of cathepsins and activation of PLA2 33. The phospholipase A2 gene (PLA2) is more expressed in chip1. Several lysosomal proteases including cathepsin D are down-regulated in two chips. In pro-oxidant-induced apoptosis, the release of cathepsin D and other lysosomal enzymes can promote mitochondrial oxidant production, cytochrome c release and apoptosis 34. Thus it is possible that lysosome rupture has participated in the As2O3/emodin-induced apoptosis.

Emodin enhances arsenic effects on cell cycle blockage

The expression of cyclinB2 gene is decreased. Cyclin B/Cdc2 kinases activity has been identified as the principal component of mitosis promoting factor (MPF) 35, upon activity of which progression from G2→M phase is highly dependent 36. The expression of both Cdk4 and retinoblastoma protein1 (pRb) are blocked while two cyclin D-related genes are up-regulated. The E2F gene expression is increased and several other downstream genes of pRb including dihydrofolate reductase and Cdc2 are down-regulated. At G1 checkpoint, the induction of D-type cyclins and subsequent phosphorylation of pRb determine whether or not the cell commits to the cell cycle and E2F family is related to this regulation 37, 38. While cyclin B/Cdc2 complex is essential for G2 checkpoint, the genes ATM and PPM1D that are related to DNA damage response/activation of p53 are found to be up-regulated. Cells with intact p53 function readily arrest in G1 after IR exposure and oxidative stress 39. Thus, we can see that during As2O3/emodin-induced apoptosis, both G1 and G2 checkpoints are affected, and several altered genes are common to cell cycle regulation and apoptosis.

Meanwhile, many genes related to proliferation, chromatin or chromosome organization and biogenesis and DNA repair are down-regulated in both chips, but mainly in chip2, indicating that these genes are influenced at the later time point.

Emodin enhances arsenic effects on cytoskeleton changes

We have found unexpectedly that the cotreatment of As2O3/emodin induce surprisingly dramatic changes in cytoskeleton-related genes at both time points. These gene expression changes have probably participated in the three continuous cytoskeletal changing steps during apoptosis—release, blebbing and condensation 40. The expression of actin cross-linking-related gene macrophin1 is blocked at the 6h time point. Genes related to actin reorganization such as actin filament, actin organization and biogenesis are mostly down-regulated at the 24h time point. The tubulinβ4, tubulin-specific chaperone c and beta-tubulin folding protein genes as well as other genes involved in microtubule motor, microtubule based movement and microtubule process are down-regulated in both chips. Myosin-related gene expressions are altered in both chips. At the 6 and 24h time points, actin α1 mRNA respectively decrease to 5% and 10%, which may perturb the third step “condensation” (actin dissolution). NAC blocks and further reverses the down-regulation of actin α1 gene.

Among these proteins, some are known to be cleaved by caspases during apoptosis, such as actin, gas2, cadherin, keratin-19 and Tau 41. One theory that attempt to explain the cell contraction, membrane blebbing and apoptotic body formation observed in apoptotic cells is that caspase-mediated proteolysis of cytoskeleton adhension proteins leads to a release from points of cell attachment followed by collapse of the cell 42.

The Rho GTPase signaling pathways related genes including RhoA in both chips, ROCK1 in chip2 and ROCK2 in chip1 are all down-regulated. The activation of RhoA, ROCK1 and ROCK2 is essential for the cell contraction and membrane blebbing during apoptosis 41. ARHA, one of the guanine nucleotide-exchange factors (GEFs) that catalyses GTP loading and activation of Rho proteins is also down- regulated. These results show that the As2O3/emodin cotreatment induces a remarkable transcriptional change in the cytoskeleton components, some of which are up- or down-regulated early and violently, whereas the transcription of Rho GTPase and ROCKs are suppressed.

In addition to mediating the cytoskeleton change during apoptosis, Rho GTPases impinge on cell survival under a variety of circumstances. RhoA has been shown to induce phophorylation and activation of AKT and TGFβ to promote cell growth 43. ROCK activity has been shown to be involved in tumor metastasis 44. Therefore, we suppose that the attenuation of Rho GTPase signaling has probably contributed to the enhanced apoptotic susceptibility of HeLa cells to As2O3.

There are many other important genes, in addition to the genes cited above, affected by ROS. For instance, ATP binding cassette subfamily 1 is down-regulated, which may be related to the drug resistance; genes encoding iron channel are altered, which may play a role in cell apoptosis; genes encoding ubiquitin and proteasome are influenced, which may function in protein degradation.

Emodin enhances arsenic trioxide cytotoxicity in HeLa cells through intracellular ROS generation. This ROS generation may, as the present study evidenced, profoundly alter the gene expression, while simultaneously influence the activities of a series of pre-existed molecules, probably by a direct oxidative modification, as our previous data suggested. With alteration of intracellular redox state, emodin regulates multiple signaling pathways, affects organelle functions, blocks cell cycle progression, causes early cytoskeleton changes, and influences many other gene expression. The final outcome of a potentiated cell death is likely to be the results of the interplay of all the aspects (Fig. 6). The gene expression profiles demonstrated here have provided us new clues for further study of molecular mechanism through which emodin enhances the sensitivity of tumor cells to As2O3– and other anti-cancer drug-induced apoptosis.