Ionizing Radiation-inducible Apoptosis in the Absence of p53 Linked to Transcription Factor EGR-1*

The tumor suppressor protein p53 is a pivotal regulator of apoptosis, and prostate cancer cells that lack p53 protein are moderately resistant to apoptotic death by ionizing radiation. Genes encoding the transcription factor early growth response-1 (EGR-1) and cytokine tumor necrosis factor-α (TNF-α) were induced upon irradiation of prostate cancer cells, and inhibition of EGR-1 function resulted in abrogation of both TNF-α induction and apoptosis. Induction of the TNF-α gene by ionizing radiation and EGR-1 was mediated via a GC-rich EGR-1-binding motif in the TNF-α promoter. Because TNF-α induces apoptosis in prostate cancer cells, these findings suggest that, in the absence of p53, ionizing radiation-inducible apoptosis is mediated by EGR-1 via TNF-α transactivation.

The tumor suppressor protein p53 is a pivotal regulator of apoptosis, and prostate cancer cells that lack p53 protein are moderately resistant to apoptotic death by ionizing radiation. Genes encoding the transcription factor early growth response-1 (EGR-1) and cytokine tumor necrosis factor-␣ (TNF-␣) were induced upon irra-

diation of prostate cancer cells, and inhibition of EGR-1 function resulted in abrogation of both TNF-␣ induction and apoptosis. Induction of the TNF-␣ gene by ionizing radiation and EGR-1 was mediated via a GC-rich EGR-1-binding motif in the TNF-␣ promoter. Because TNF-␣ induces apoptosis in prostate cancer cells, these findings suggest that, in the absence of p53, ionizing radiation-inducible apoptosis is mediated by EGR-1 via TNF-␣ transactivation.
Prostate cancer in men is the most common malignancy and the second most leading cause of cancer deaths (1). Radiation therapy that causes growth inhibition and programmed cell death (apoptosis) of tumors is a well established mode of treatment for both primary and metastatic prostate cancer (2). However, despite using high doses of radiation, about 50 -60% of prostate cancer patients show persistent local disease (3)(4)(5). A major reason for failure to eradicate local disease is the intrinsic radioresistance of the tumors. One of the molecular determinants regulating the response to ionizing radiation is the tumor suppressor protein p53 that serves as a pivotal component of the apoptosis pathways in diverse cell types (6,7). Wild-type p53 protein confers radiation responsiveness, but loss of p53 function owing to either mutation(s) or deletion of p53 alleles confers radioresistance (8 -13). Because p53 protein is mutated and non-functional in a large number of prostate tumors (14,15), it is imperative to identify other proapoptotic genes that can function via a p53-independent mechanism and further to design novel approaches to induce the expression of such genes for the control of radio-resistant tumors.
Previous studies have shown that EGR-1 1 protein, encoded by the immediate-early gene Egr-1, is induced by ionizing ra-diation in a wide spectrum of tumor cell types (16,17). EGR-1 is a member of the Egr (early growth response) family of transcription factors that includes EGR-2, EGR-3, NGFI-C, the tumor suppressor Wilms' tumor gene product WT1, and EGR-␣ (18 -20). The Egr family members show a high degree of homology in the amino acids constituting the zinc finger domain and bind to the same GC-rich consensus DNA sequence (21)(22)(23). Functional studies have suggested that EGR-1 is an antiproliferative signal for tumor cells (24,25) and that it acts to increase the potency of apoptotic agents (26,27). In human melanoma cells that contain wild-type p53, abrogation of EGR-1 function confers radioresistance despite induction of p53 (17), suggesting that EGR-1 may function by a p53-independent mechanism. A p53-independent mechanism for apoptotic death of prostate cells in p53-null mice after testosterone ablation effected by orchiectomy also has been suggested (28). Incidentally, EGR-1 is induced early in the rat prostate after orchiectomy (29), but its function or mechanism of action in prostatic cells is not known.
Another gene that is induced by radiation in a wide range of cell types is tumor necrosis factor-␣ (TNF-␣) that encodes a cytokine with pleiotropic effects (30). The induction of TNF-␣ gene expression represents an important aspect of cellular response to ionizing radiation that causes both autocrine and paracrine tumor cell killing (30 -32). Ionizing radiation causes a transient increase in TNF-␣ mRNA followed by a corresponding increase in TNF-␣ protein (30). TNF-␣ protein is a well characterized cytokine that induces apoptosis in different cell types by binding to the TNF-R1 receptor (33,34). Binding to the TNF-R1 receptor triggers the sequential recruitment and activation of a cascade of death domain-containing proteins, which further activate cysteinyl-aspartate-specific proteinases (caspases) such as interleukin-converting enzyme and those of the interleukin-converting enzyme-related family (33)(34)(35)(36)(37)(38). The caspases then cleave substrates, such as poly(ADP-ribose) polymerase, nuclear lamins, actin, protein kinase C-␦, and fodrin, that are essential for cell survival, subsequently leading to apoptotic cell death (33)(34)(35)(36). The present study used a prostate cancer cell line, PC-3, that lacks functional p53 protein (9,39,40) to determine the role of EGR-1 in radiation-induced apoptosis. We demonstrate here that ionizing radiation-inducible apoptotic death is caused by EGR-1 despite the absence of p53 protein and that EGR-1 action involves the up-regulation of the TNF-␣ gene.

MATERIALS AND METHODS
Plasmid Constructs-Plasmid pCMV-WT1-EGR1, which encodes a dominant-negative mutant of EGR-1, contains a WT1-EGR-1 chimera downstream of the cytomegalovirus (CMV) promoter in vector pCB6 ϩ (41). The plasmid CMV-EGR-1, which encodes full-length EGR-1 protein, contains EGR-1 cDNA downstream of the CMV promoter in the vector pCB6 ϩ (41). The reporter construct, EBS-CAT (41), contains three EGR-1 binding sites (CGCCCCCGC) placed in tandem * This work was supported by United States Public Health Service-National Institutes of Health Grants CA52837 and CA60872 and Council for Tobacco Research Grant 3490 (to V. M. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
upstream of a minimal c-Fos promoter and chloramphenicol acetyltransferase (CAT) cDNA. The TNF-␣ promoter region from Ϫ470 to ϩ102 was generated from normal human genomic DNA template by PCR (42). The sense primer TNFp-470U, [5Ј-TTTTTCTAGATTTC-CCTCCAACCCGTTTTCT-3Ј], and antisense primer TNFpϩ102L, [5Ј-TAAGCTTCAGGGGATGTGGCGTCTGAGG-3Ј], contained built-in sites (underlined) for XbaI and HindIII, respectively, and they generated a 589-base pair fragment that contained the EGR-1binding site 5Ј-CGCCCCCGC-3Ј. The EGR-1 binding site in the 589base pair fragment was mutated to 5Ј-GTTAACCGC-3Ј by using PCR-directed mutagenesis. The 589-base pair fragments representing wild or mutant EGR-1 binding sites were cloned in pG-CAT, a vector for CAT reporter, and the fidelity of PCR reactions and subcloning was confirmed by nucleotide sequencing.
DNA Transfections and CAT Assays-Transfections were performed as described previously (27) . CAT assays were performed by thin-layer chromatography as described previously (27).
Irradiation-A 100 kV industrial x-ray machine (Phillips, Netherlands) was used to irradiate the cultures at room temperature. The dose rate with a 2-mm Al plus 1-mm Be filter was ϳ1.85 Gy/min at a focus-surface distance of 30 cm.

Assay for [ 3 H]Thymidine Incorporation and Colony Formation
Assay-The [ 3 H]thymidine incorporation experiments were performed as described previously (17). For clonogenic cell survival studies, two different cell concentrations in quadruplet sets were used for each radiation dose. Parental PC-3 cells and PC-3 transfectants were left untreated or exposed to 1-6 Gy dose of radiation. After incubation for 10 or more days, each flask was stained with crystal violet, and the colonies containing more than 50 cells were counted. The surviving fraction (SF) was calculated as a ratio of the number of colonies formed and the product of the number of cells plated and the plating efficiency. The curve was plotted using X-Y log scatter (Delta Graph®4.0) and fitted by single hit multi-target model (43) to obtain D 0 , n, SF 2 values (44). D 0 is the dose required to reduce the fraction of cells to 37%, indicative of single-event killing; the n value of the curve is a measure of the width of the shoulder, indicative of multiple-event killing; and SF 2 is the survival fraction of exponentially growing cells that were irradiated at the clinically relevant dose of 2 Gy.
Immunocytochemistry and Western Blot Analysis-EGR-1 expression

FIG. 1. Radiobiological characteristics of PC-3 cells.
A, effect of ionizing radiation on the growth of PC-3 cells. Cells were exposed to a 0-, 5-, 10-, or 20-Gy dose of ionizing radiation and then were pulsed with [ 3 H]thymidine after 48 h, and growth inhibition was determined as described (17). Each data point is a mean of 48 observations from three separate experiments; error bars indicate Ϯ S.D. B, effect of ionizing radiation on the colony-forming ability of PC-3 cells. Cell survival curve of PC-3 cells after irradiation as assayed by colony-forming ability is shown in the form of single hit multi-target (SHMT) model. Each data point is a mean of observations from four separate experiments; error bars indicate Ϯ S.D. C, radiation induces apoptosis in PC-3 cells. Cells were irradiated at 5 Gy and after 48 h were stained with Hoechst (Ho342) and merocyanine (MC540) (45). The gates were set so as to analyze cell cycle and apoptosis stages. Ho342 is a DNA-specific dye that measures DNA content, and MC540 binds to membrane phospholipids that are exposed on the outside of the membrane during the process of apoptosis. These two dyes separate five distinct populations of tumor cells: viable resting cells, 2n DNA content and MC540-unstained (R1); viable cycling cells, Ͼ2n DNA content and MC540-unstained (R2); viable resting cells undergoing apoptosis, 2n DNA content and MC540-stained (R3); viable cycling cells undergoing apoptosis, Ͼ2n DNA content and MC540-stained (R4); and late stage apoptotic cells that are MC540-stained but Ho342-unstained indicating DNA fragmentation (R5). The data shown are representative of three independent experiments. The untreated population contained MC540-stained cells owing to spontaneous apoptosis that occurred during cell culture. The percent increase (mean Ϯ S.D. from three experiments) in apoptotic cells (i.e. MC540-stained cells in the R3, R4, and R5 compartments) in the irradiated population over the untreated population was 21.51 Ϯ 3.02.
was determined in untreated and irradiated PC-3 cells by immunocytochemical analysis, as described by us previously (17), by using anti-EGR-1 antibody, sc-110 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and the Elite ABC kit (Vector Laboratories, Burlingame, CA). Total protein extracts from untreated and irradiated PC-3 cells with various time intervals were subjected to Western blot analysis by using sc-110 or the ␤-actin antibody (Sigma) for a loading control as described (17).
Assays for Apoptosis-For the DNA laddering test, total genomic DNA was prepared from the PC-3 cells exposed to vehicle or recombinant TNF-␣ for 48 h and was subjected to agarose gel electrophoresis as described previously (26). For quantitation of apoptosis, cells were lifted by using non-enzymatic cell dissociation medium (Sigma) and washed with phosphate-buffered saline and stained with Hoechst (Ho342) and merocyanine (MC540) and analyzed by flow cytometry using a FACStar Plus cell sorter as described (45).
Fluorescence in Situ Hybridization (FISH) Analysis of Egr-1 Gene-To evaluate the copy number or amplification of the Egr-1 gene, FISH was performed using a spectrum orange-labeled Egr-1 probe (locus: Egr-1 band assignment: 5q23-31.1) obtained from Vysis Inc. (Downers Grove, IL) as described previously (17). 32 P-Reverse Transcriptase-Polymerase Chain Reaction ( 32 P-RT-PCR)-Total RNA was extracted from PC-3 and their transfectants using TRIzol reagent (Life Technologies, Inc.). 1 g of total RNA was reverse transcribed into cDNA using oligo(dT) primers and reverse transcriptase in a 20-l reaction mix (Perkin-Elmer) as described by us previously (17). The TNF-␣ cDNA sense primer sequences are from nucleotides 839 -860, 5Ј-AGG CGC TCC CCA AGA AGA CAG-3Ј, and antisense primer sequences are from nucleotides 1039 -1060, 5Ј-AGG CTT GTC ACT CGG GGT TCG-3Ј (46). In addition, ␤-actin gene (the sense primer starts at nucleotide 1628 and the antisense primer begins at nucleotide 2379 and generates a 331-base pair PCR product) was used as an internal control (47). (Fig. 1A). Moreover, colony-forming assays indicated that the survival fraction (SF 2 ) of exponentially growing PC-3 cells that were irradiated at the clinically relevant dose of 2 Gy was 0.25 (Fig. 1B). The dose required to reduce the fraction of PC-3 cells to 37% (D 0 value, i.e. singleevent killing) was 140 cGy. The n value of the curve, which is a measure of the width of the shoulder (indicative of multiple event killing), was 1.03 (Fig. 1B). When compared with previously reported SHMT values for different tumor cell lines (48), those for PC-3 cells in Fig. 1B suggest moderate resistance to ionizing radiation. Clonogenicity inhibition by ionizing radiation was caused because of apoptotic death of the cells as evident from flow analysis, where cultures exposed to ionizing radiation showed a 20 -23% increase in apoptotic cells compared with untreated cultures (Fig. 1C). These observations suggested that despite the absence of p53 protein, PC-3 cells are susceptible to ionizing radiation-inducible apoptosis.

Ionizing Radiation Causes Growth Inhibition and Cell Death in PC-3 cells-Ionizing radiation caused a dose-dependent growth inhibition of PC-3 cells, as demonstrated by [ 3 H]thymidine incorporation assays
Ionizing Radiation Causes Induction of EGR-1-We examined whether EGR-1 induction was associated with apoptosis inducible by ionizing radiation. Because Egr-1 maps to chromosome 5 and because PC-3 cells lack the chromosome 5 pair (40), we performed FISH to ascertain that Egr-1 alleles were intact in these cells. FISH analysis confirmed that PC-3 cells have two intact alleles for Egr-1 at 5q23-31.1 (Fig. 2A). Consistent with this observation suggesting that some areas of 5q are intact perhaps in the form of a marker chromosome, another gene, APC, normally located on 5q21, is also expressed as a full-length protein in these cells (49). Western blot analysis further confirmed that PC-3 cells constitutively express modest levels of EGR-1 protein (Fig. 2B). Moreover, after exposure to a 5-Gy dose of radiation, PC-3 cells showed induction of EGR-1 expression with peak levels (3-4 fold) at 45 min (Fig. 2B). These results were corroborated by immunocytochemical analysis that demonstrated a significant increase in nuclear EGR-1 expression levels after 45 min of exposure to radiation (Fig.  2C). The nuclear localization of EGR-1 was particularly important in view of the potential transcriptional activation function of EGR-1 protein.

FIG. 2. EGR-1 is induced by ionizing radiation in PC-3 cells.
A, FISH was performed as described previously (17) to detect the Egr-1 copy number by using a spectrum orange-labeled Egr-1 probe. White arrows indicate the two alleles (red dots) of Egr-1 in interphase PC-3 cells. B, EGR-1 protein induction detected by Western blot analysis. Whole cell protein extracts were prepared from PC-3 cells that were left untreated (UT) or exposed to a 5-Gy dose and incubated for the time interval indicated in minutes and then subjected to Western blot analysis for EGR-1 or ␤-actin (17,26). C, nuclear induction of EGR-1 following exposure to ionizing radiation. PC-3 cells were left untreated (UT) or treated with a 5-Gy dose of ionizing radiation and subjected to immunocytochemistry 45 min after the exposure. EGR-1-positive cells showed strong brown nuclear staining with the diaminobenzidene-H 2 O 2 substrate.

ent Transcriptional Activation via the GC-rich Binding Site in Parental PC-3 and Stably Transfected PC-3 Cells-To ascer-
tain that the WT1-EGR-1 chimera, which has been demonstrated in other cell types to act as a dominant-negative mutant of EGR-1, abrogates EGR-1 function in PC-3 cells, we performed transient cotransfections with a reporter construct EBS-CAT that contains three tandem EGR-1-binding sites, an EGR-1 expression construct CMV-EGR-1, and the chimera. As seen in Fig. 3A, transactivation of the reporter construct by CMV-EGR-1 was inhibited by transient cotransfection of PC-3 cells with CMV-WT1-EGR-1. These results confirmed that the chimera functioned as a dominant-negative mutant of EGR-1 in the PC-3 cell background.
We next tested PC-3 cells stably transfected either with the chimera or with an empty vector as a control for abrogation of EGR-1 function. In these experiments, we noted a strong increase in CAT reporter activity when cells stably expressing the vector were cotransfected with CMV-EGR-1 and EBS-CAT (Fig. 3B). By contrast, cells stably expressing the chimera showed relatively very weak CAT activity upon cotransfection with CMV-EGR-1 and EBS-CAT (Fig. 3B). These findings suggested that the transfected cells expressing the chimera effectively impede EGR-1 transactivation function.
The Dominant-negative Mutant of Egr-1 Confers Resistance to Ionizing Radiation-inducible Growth Inhibition and EGR-1 Confers Enhanced Radiosensitivity-To determine the functional role of EGR-1 in radiation-inducible apoptosis of PC-3 cells, we transfected PC-3 cells with plasmid CMV-EGR-1, CMV-WT1-EGR-1 chimera, or an empty vector (pCB6ϩ) and obtained stable transfectant clones by selection with G418 sulfate (27,41). Pools of about 200 transfected clones were designated P1 or P2, and at least two different pools of clones from each transfections were used in this study. Initially, we determined the effect of abrogation of EGR-1 function in PC-3 cells expressing the chimera in response to ionizing radiation. In colony-forming assays, the SF 2 value of exponentially growing irradiated PC-3/vector.P1 cells was 0.2 with a D 0 value of 124 cGy, and n value was 1.01 (Fig. 3C). On the other hand, the SF 2 for PC-3/WT1-EGR1.P1 was 0.45 with the D 0 value of 227 cGy, and n value was 1.1, suggesting that compared with vectortransfected cells, the cells transfected with the chimera were significantly resistant (p Ͻ 0.0001) to ionizing radiation. Then, we tested the effect of EGR-1 overexpression in PC-3 cells stably transfected with CMV-EGR-1 on the response to ionizing radiation. As seen in Fig. 3C, the SF 2 for PC-3/EGR-1.P1 was 0.095 with the D 0 value of 86 cGy, and n value was 1, suggesting enhanced sensitivity (p Ͻ 0.0001) than the vector transfected cells to ionizing radiation. Consistent with this observa- ]chloramphenicol to acetylated forms as described (27). C and D, EGR-1 protein increases radiosensitivity of PC-3 cells, and dominant-negative mutant of EGR-1 protects PC-3 cells from radiation-inducible growth inhibition and apoptosis. PC-3/EGR-1.P1, PC-3/WT1-EGR-1.P1, or PC-3/vector.P1 cells were left unexposed or exposed to the indicated doses of radiation, and either cell survival was assayed by colony-forming ability and expressed by using SHMT model (C) or apoptosis was quantified after 48 h by flow analysis (D). Each data point in SHMT model is a mean of three separate experiments, and error bars represent Ϯ S. D. The relative apoptotic index values represent the mean of three independent flow cytometry experiments. tion, flow analysis indicated that PC-3/WT1-EGR-1.P1 cells were significantly resistant (p Ͻ 0.0001) and PC-3/EGR-1.P1 cells were significantly more sensitive (p Ͻ 0.0001) to ionizing radiation-inducible apoptosis than vector transfected PC-3 control cells (Fig. 3D). Together these findings suggest that EGR-1 is required for ionizing radiation-inducible apoptosis and that when overexpressed EGR-1 potentiates the effects of ionizing radiation.
TNF-␣ as Downstream Target of EGR-1-To identify potential downstream targets that might mediate the proapoptotic action of EGR-1, we conducted a GenBank TM /EBI search for genes that contain the EGR-1 consensus binding sites in their promoter regions, focusing on genes that satisfied the following stringent criteria: (i) they should be inducible by ionizing radiation; (ii) they should be functionally involved in apoptosis; and (iii) they should induce apoptosis via a p53-independent pathway. One of the genes that met these criteria was TNF-␣. We first ascertained that TNF-␣ protein could cause apoptosis in the PC-3 cells. Cells were left untreated or treated with 100 units/ml of exogenous recombinant TNF-␣, and total DNA was examined for DNA laddering that is characteristic of apoptosis. TNF-␣ caused nucleosomal DNA fragmentation in PC-3 cells (Fig. 4A), suggesting that the cell killing by TNF-␣ is mediated through apoptosis.
Next, we determined whether TNF-␣ was inducible by ionizing radiation in PC-3 cells. TNF-␣ mRNA levels increased about 2-3-fold over basal levels in cells after 30 min of exposure to a 5-Gy dose of ionizing radiation (Fig. 4B). Next, we tested whether EGR-1 regulated the radiation-inducible expression of TNF-␣ by using 32 P-RT-PCR analysis in transfected PC-3 cells. A dose of 5 Gy radiation elevated TNF-␣ mRNA about 2-3-fold in 30 min in PC-3/vector.P1 cells (which was similar to parental PC-3 cells) but caused down-regulation of TNF-␣ mRNA levels (to 0.08% of those in untreated cells) in PC-3/WT1-EGR-1.P1 cells (Fig. 4C). Thus, the dominant-negative mutant of EGR-1 abrogates ionizing radiation-inducible TNF-␣ expression. Because the dominant-negative mutant acts by inhibiting the function of EGR-1, these findings indicate that EGR-1 is essential for ionizing radiation-inducible TNF-␣ expression.
To determine the mechanism by which EGR-1 may up-regulate TNF-␣, we examined the TNF-␣ gene promoter for consensus EGR-1-binding sites. Interestingly, a sequence 5Ј-GCGGGGGCG-3Ј that conforms exactly with that first shown (21) to bind EGR-1 protein was detected between nucleotides 194 and 186 upstream of the cap site of TNF-␣ (42). This sequence in TNF-␣ gene has not been demonstrated to mediate transactivation by EGR-1. To determine the mechanism by which the dominant-negative mutant protein of EGR-1 blocked  (47). D, EGR-1 transactivates a TNF-␣ promoter-CAT (TNFp-CAT) reporter construct containing an EGR1-binding site, and the WT1-EGR-1 chimera competes with this transactivation. Parental PC-3 cells were transiently cotransfected with 4 g of TNFp-CAT and indicated amounts (g) of CMV-EGR-1 or CMV-WT1-EGR-1. CAT activity was assayed and expressed as percent conversion of 14 C-chloramphenicol to acetylated forms. E, ionizing radiation transactivates TNFp-CAT reporter construct containing EGR-1-binding sites, and the WT1-EGR-1 chimera competes with this transactivation. PC-3 cells were transiently cotransfected with 4 g of TNFp-CAT or of TNFpmutant-CAT and indicated amounts (g) of CMV-EGR-1 or CMV-WT1-EGR-1. Next, the cells were either left unexposed or exposed to a 5-Gy dose of radiation, and CAT activity was assayed and expressed as percent conversion of 14 C-chloramphenicol to acetylated forms. radiation-inducible TNF-␣ expression, we performed CAT assays by using a reporter construct, TNFp-CAT that contains the TNF-␣ promoter region from Ϫ470 to ϩ120. PC-3 cells were transiently cotransfected with TNFp-CAT, CMV-EGR-1, and either the dominant-negative mutant CMV-WT1-EGR-1 or the empty vector pCB6 ϩ . Cells cotransfected with TNFp-CAT and vector showed a modest background level of CAT activity (Fig.  4D). The CAT activity was increased significantly when the cells were cotransfected with TNFp-CAT, CMV-EGR-1, and vector. On the other hand, when cotransfection was performed with TNFp-CAT, CMV-EGR-1, and CMV-WT1-EGR-1, CAT activity was severely reduced (Fig. 4D). These findings indicated that CMV-EGR-1 can transactivate, and the dominantnegative mutant WT1-EGR-1 can transrepress, the TNF-␣ promoter.
We also studied the reporter activity of TNFp-CAT construct in cells exposed to ionizing radiation. Ionizing radiation induced CAT activity from TNFp-CAT within 1 h of the exposure (Fig. 4E). Induction of TNFp-CAT expression by ionizing radiation was significantly reduced when cells were cotransfected with the dominant-negative mutant of EGR-1 (Fig. 4E). These findings suggested that inducible expression of the TNF-␣ promoter by ionizing radiation required functional EGR-1. When a mutation, which abolished the ability of EGR-1 protein to bind to the TNF-␣ promoter (42), was introduced into the TNFp-CAT construct, basal expression driven by the promoter was attenuated, and radiation-inducible CAT activity was abolished (Fig. 4E). As expected, ectopically expressed EGR-1 caused induction of CAT activity from the promoter construct that contained wild-type EGR-1-binding site but not from the construct that contained the mutant EGR-1-binding site (Fig.  4E). Together, these findings suggest that EGR-1 transactivates the TNF-␣ promoter via the EGR-1-binding site and that this site is required for radiation-inducible TNF-␣ expression. DISCUSSION Wild-type p53 has been shown to be functionally necessary for growth inhibition and apoptosis following exposure to ionizing radiation (6), and p53 mutations have been reported to increase resistance to apoptosis (8). In agreement with these observations, the present study has shown that the absence of wild-type p53 protein in prostate cancer cells renders them moderately resistant to ionizing radiation-inducible apoptosis. The expression of genes encoding EGR-1 and TNF-␣ that induced apoptosis was up-regulated by ionizing radiation in the PC-3 cells, and inhibition of EGR-1 transactivation function by the dominant-negative WT1-EGR-1 chimera abrogated ionizing radiation-inducible TNF-␣ induction and apoptosis. Consistent with these observations, ectopically expressed EGR-1 enhanced ionizing radiation-inducible TNF-␣ expression and apoptosis. These findings suggest that EGR-1 is an upstream modulator of TNF-␣ induction and apoptosis in the pathway evoked by ionizing radiation. Moreover, EGR-1 causes transcriptional activation of the TNF-␣ promoter via a consensus EGR-1-binding site providing a mechanism for EGR-1-inducible expression of the TNF-␣ gene. Thus, EGR-1 is an important mediator of radiation responsiveness in prostate cancer cells that lack functional p53 protein. Because p53 protein is mutated in prostate tumors (14,15), EGR-1, which functions by a p53-independent pathway, is of critical importance in the apoptotic death of the tumors. Both EGR-1 and TNF-␣ are induced by ionizing radiation in diverse tumor types, and future studies may design approaches to further exploit this novel pathway for the containment of radio-resistant tumors.