Ataxia-Telangiectasia, Mutated (ATM)/Nuclear Factor κ Light Chain Enhancer of Activated B Cells (NFκB) Signaling Controls Basal and DNA Damage-induced Transglutaminase 2 Expression*

Background: Transglutaminase-2 (TG2) is up-regulated in response to stress through an unknown mechanism. Results: Disruption of NFκB subunit p65 or its activator ATM abrogates genotoxin-induced and basal TG2 expression. Conclusion: NFκB/ATM signaling drives TG2 expression, and both molecules are components in TG2-linked drug resistance. Significance: DNA damage response is crucial to understanding cancer development and treatment. Transglutaminase 2 (TG2) is a ubiquitously expressed enzyme that cross-links proteins and its overexpression, linked to a drug resistant phenotype, is commonly observed in cancer cells. Further, up-regulation of TG2 expression occurs during response to various forms of cell stress; however, the molecular mechanisms that drive inducible expression of the TG2 gene (TGM2) require elucidation. Here we show that genotoxic stress induces TG2 expression through the Ataxia-Telangiectasia, Mutated (ATM)/Nuclear Factor κ light chain enhancer of activated B cells (NFκB) signaling pathway. We further document that NFκB is both necessary and sufficient to drive constitutive TG2 expression in cultured cell lines. Additionally, shRNA-mediated knockdown or pharmacological inhibition of the ATM kinase results in reduced constitutive TG2 expression and NFκB transcriptional activity. We document that the NFκB subunit p65 (RelA) interacts with two independent consensus NFκB binding sites within the TGM2 promoter, that mutation of either site or pharmacological inhibition of NFκB reduces TGM2 promoter activity, and genotoxic stress drives heightened association of p65 with the TGM2 promoter. Finally, we observed that knockdown of either p65 or ATM in MDA-MB-468 breast cancer cells expressing recombinant TG2 partially reduces resistance to doxorubicin, indicating that the drug resistance linked to overexpression of TG2 functions, in part, through p65 and ATM. This work establishes a novel ATM-dependent signaling loop where TG2 and NFκB activate each other resulting in sustained activation of NFκB and acquisition of a drug-resistant phenotype.

Transglutaminase 2 (TG2) is a ubiquitously expressed enzyme that cross-links proteins and its overexpression, linked to a drug resistant phenotype, is commonly observed in cancer cells. Further, up-regulation of TG2 expression occurs during response to various forms of cell stress; however, the molecular mechanisms that drive inducible expression of the TG2 gene (TGM2) require elucidation. Here we show that genotoxic stress induces TG2 expression through the Ataxia-Telangiectasia, Mutated (ATM)/ Nuclear Factor light chain enhancer of activated B cells (NFB) signaling pathway. We further document that NFB is both necessary and sufficient to drive constitutive TG2 expression in cultured cell lines. Additionally, shRNA-mediated knockdown or pharmacological inhibition of the ATM kinase results in reduced constitutive TG2 expression and NFB transcriptional activity. We document that the NFB subunit p65 (RelA) interacts with two independent consensus NFB binding sites within the TGM2 promoter, that mutation of either site or pharmacological inhibition of NFB reduces TGM2 promoter activity, and genotoxic stress drives heightened association of p65 with the TGM2 promoter. Finally, we observed that knockdown of either p65 or ATM in MDA-MB-468 breast cancer cells expressing recombinant TG2 partially reduces resistance to doxorubicin, indicating that the drug resistance linked to overexpression of TG2 functions, in part, through p65 and ATM. This work establishes a novel ATM-dependent signaling loop where TG2 and NFB activate each other resulting in sustained activation of NFB and acquisition of a drug-resistant phenotype.
Transglutaminases are a family of enzymes that catalyze the formation of covalent bonds between the ⑀-amino group of a lysine and the ␥-carboxyl group of a glutamine residue (1). Such cross-links create proteinaceous networks that are resistant to proteolytic and mechanical degradation. Current evidence indicates that the ubiquitously expressed transglutaminase 2 (TG2) 3 is a multifunctional protein stemming, in part, from the fact that TG2 localizes to the cytosol, nucleus, and can be exported from the cell (2). Extracellular pools of TG2 bind to and cross-link numerous components of the extracellular matrix (ECM) supporting the view that TG2 renders the ECM resistant to degradation, stabilizes cell-ECM interaction, and promotes wound healing (3)(4)(5).
A feature unique to TG2 among other transglutaminases is that elevated expression of this molecule induces a drug-resistance phenotype. Specifically, Mehta (6) initially observed that heightened levels of TG2 expression in MCF7 breast cancer cells led to resistance to doxorubicin and resistance to anthracyclines has subsequently been documented in a wide variety of cell types (7)(8)(9)(10). Several groups showed that TG2 activates the oncogenic transcription factor Nuclear Factor light chain enhancer of activated B cells (NFB) (11,12), and this dysregulated NFB activity is likely important in conferring the chemoresistance phenotype, as is an anti-apoptotic property of TG2 itself through cross-linking, and hence deactivating, caspase-3 (13).
Quantitative examination of TG2 expression indicates that this molecule can be either over-or underexpressed in various primary neoplasms (10, 14 -17). Our group has recently documented that the gene encoding TG2 (i.e. TGM2) is subject to epigenetic silencing due, in part, to aberrant CpG hypermethylation within the 5Ј-flanking region of the TGM2 gene (17,18). While epigenetics can account for reduced TG2 expression, the mechanisms driving increased TG2 expression are less clear. Retinoic acid can modestly drive TG2 levels up in a number of different cell types (19,20), and EGF stimulates TG2 expression in HeLa cells (21). At the gene level, recent work shows that TG2 expression can be induced by hypoxia-inducible factor 1 (HIF1) (22). These investigators established that, in addition to the HIF1a subunit, TG2 is induced in response to hypoxic stress in several different cell types and is linked to the binding of HIF1␣ to a hypoxia response element (HRE) within the TGM2 promoter.
Several cellular stressors such as oxidative stress and inflammation can also drive heightened expression of TG2 (23,24). As stress response is, in general, governed by a set of overlapping signaling mechanisms (25,26), in this study we sought to determine the signaling that controls TG2 expression. Here we document that TG2 is up-regulated in response to genotoxic stress through Ataxia-Telangiectasia, Mutated (ATM)/NFB signaling. Moreover, this signaling axis is required for not only the stress-induced, but also basal TG2 expression. Finally, we examine the requirement for p65 and ATM in the drug resistance phenotype observed in breast cancer cells overexpressing TG2.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatment-This investigation utilized three different classes of cell lines based on their level of endogenous TG2 expression. Lines that possess moderate endogenous TG2 expression and low to absent CpG methylation within the TGM2 promoter (i.e. HCT116, HCT116 p53 Ϫ/Ϫ , HeLa, MCF7, MCF10A, 184B5), those with very high levels of endogenous TG2 (i.e. MCF7/ADR, MDA-MB-231), and MDA-MB-468 that have little detectable endogenous TG2 expression due to epigenetic silencing of the TGM2 gene.
The mammary epithelial cell line 184B5 was cultured in DMEM/F12 media supplemented with 20 ng/ml epidermal growth factor (EGF), 0.5 g/ml hydrocortisone, 100 ng/ml cholera toxin, 10 g/ml insulin, and 1.0% equine serum. All supplements were purchased from Sigma. All other cell lines were cultured in DMEM supplemented with 10% fetal bovine serum. MCF7/ADR cells were also routinely cultured in complete medium containing 1 g/ml doxorubicin (to maintain drug resistant phenotype); however, were cultured without doxorubicin for at least 2 weeks prior to experimentation. HCT116 p53 Ϫ/Ϫ cells (27) were obtained from Dr. Bert Vogelstein (The Johns Hopkins University). Cell lines were maintained in a humidified 5% CO 2 environment at 37°C.
The SN1 DNA alkylating agent N-methyl-NЈ-nitro-N-nitrosoguanidine (MNNG) and NFB inhibitor BAY 11-7082 were purchased from Sigma). ATM kinase inhibitor KU-55933 was purchased from Chemdea (Ridgewood, NJ). Stock solutions were prepared with DMSO and final concentrations used were 20 M for MNNG, 4 M for BAY 11-7082, and 20 M for KU-55933. Where indicated, cells were irradiated with 5 Gy (500 Rad) of ionizing radiation (IR) from a Cs 137 source (Gam-maCell 40 Extractor, Ottawa, ON, Canada). Subcellular fractionation was conducted using a NE-PER Nuclear and Cytoplasmic Extraction Reagent kit (Pierce) following the manufacturer's instructions.
To assure the significance of observed changes in immunoblot signal intensities, films from at least three independent experiments were measured. Unless reported otherwise, all reported rises (or drops) in immunoblot signal intensity were found to be statistically significant (p Ͻ 0.05, Student's t test).
RNA Interference (RNAi)-For RNAi-mediated knockdown of p65 and ATM, shRNA sequences cloned into the lentiviral vector pGIPz were obtained from Open Biosystems (Huntsville, AL). Clone V3LHS_633764 (clone 1) and V3LHS_633767 (clone 2) were used to repress p65 expression, and clone V2LHS_192880 was used to knockdown ATM expression. As a control, pGIPz containing a nonspecific shRNA sequence was used. Lentivirus encoding shRNAs were packaged in HEK-293FT cells (ATCC) following co-transfection with the packaging plasmids psPAX2 and pMD2.G. Lentivirus containing medium and polybrene (10 g/ml final concentration) were added to cultures of indicated cell lines, and selection with 2 g/ml pyromycin was conducted for ϳ2 weeks prior to analysis.
Transcriptional Reporter Assays-TGM2 promoter constructs were engineered by amplifying human genomic DNA with KlenTaq LA polymerase (Clontech, Mountain View, CA) and PCR products were subcloned into pGEM-T EZ Vector (Promega, Madison, WI) prior to validation by sequencing. Primers used to amplify the TGM2 promoter fragment HN (HindIII to NcoI, nt#-1680 to ϩ1) were: 5Ј-TAAAGCTTGTG-TGTCTGTGGGTG-3Ј (forward) and 5Ј-CTGCGGTGACTC-TGATACTCA-3Ј (reverse). Primers used to amplify an upstream portion of the TGM2 promoter spanning from nt# Ϫ2380 to Ϫ1680 were 5Ј-CTCAGGAGTTTGGTACCAGC-CTG-3Ј (forward) and 5Ј-GCTCCCTGCAGCTGGTGAA-A-3Ј (reverse). (NB: The KpnI site at the 5Ј-end of this fragment (designated KH) is engineered.) The QuikChange Primer Design program was to used to design the mutations into the TGM2 promoter KH (NFB Site 1) and HN (NFB site 2) fragments cloned in pGEM T-EZ. Mutagenic primers for Site 1 within KH were: 5Ј-TCCCTGTCTTCTAGaGAGCTCaCAG-GCTGAGGGTTTT-3Ј (forward) and 5Ј-AAAACCCTCAGC-CTGtGAGCTCaCTAGAAGACAGGGAG-3Ј (reverse) (NB: Designed mutations are indicated in lowercase). Mutagenic primers for Site 2 within the HN fragment were 5Ј-GGAAAA-AGTGCCAGtGAAGCCaCGTGGGCCTCTGTC-3Ј(forward) and 5Ј-GACAGAGGCCCACGtGGCTTCaCTGGCACTTT-TTCC-3Ј (reverse). Site directed mutagenesis was accomplished using the Quikchange II XL Site-Directed Mutagenesis kit (Agilent) following the supplied protocol and mutations validated by automated sequencing. Subsequently, both wild type and mutant KH and HN fragments were subcloned into pGL3-Basic to construct the indicated reporter plasmids. Dual-Luciferase reporter assay system (Promega) was used to detect either TGM2 promoter or NFB transactivation activity (for the latter pNFB-TA-Luc reporter (Clontech)) was used. Briefly, cells were co-transfected with 4 g of reporter and 100 ng of pRL-TK (Renilla Luciferase expression plasmid, Promega). 48 h later, cells were harvested and cell lysates were used to measure both firefly and renilla luciferase activity using a BMG Labtech luminometer (Offenburg, Germany).
Cell Transfection-A full-length, human TG2 cDNA clone was subcloned into the expression vector pcDNA3-Myc as previously described (13). MDA-MB-468 cells were seeded into 6-well tissue culture plates and after overnight incubation were transfected with 4.0 g of pcDNA3-Myc-TG2 or empty pcDNA 3.1 plasmid with Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. Cells were subsequently cultured with 500 g/ml G418 (Sigma) and after selection recombinant TG2 expression was confirmed by immunoblotting. Where indicated, p65 expression was knocked down using lentiviral clone #2.
Drug Sensitivity Assays-Relative cell viability following doxorubicin treatment was determined using an AlamarBlue assay as outlined by the manufacturer (AbDSerotec, Raleigh, NC). Briefly, cells were plated into 24-well plates and allowed to adhere overnight. The next day, doxorubicin was added to the media (2 M final concentration). After 24 h, media was removed, rinsed with 1ϫ PBS, new media added, and cells incubated for an additional 24 h. At this point a 1/10 volume of AlamarBlue reagent was added to the wells and cells were incubated at 37°C for an additional 2 h. After this, 100 l of media from each well was pipetted in triplicate into 96-well plates and fluorescence (545 nm excitation, 590 nm emission wavelength) measured using a BMG Labtech fluorometer. Cell viability was calculated relative to an untreated (no doxorubicin) culture of cells incubated in parallel.

DNA Damage Induces Increased TG2 Expression through an ATM/NFB-dependent Mechanism-Several groups have
demonstrated that TG2 is up-regulated in response to various forms of cellular stress (24) prompting us to question if response to DNA damage drives increased TG2 expression. Moreover, as TG2 expression has been shown to vary widely in cultured breast cancer lines and primary tumors (18, 28), we subjected MCF7 breast cancer cells and the mammary epithelial line 184B5 to either 5 Gy of ionizing radiation (IR) or 20 M MNNG, an SN1 DNA-alkylating agent. Further, to establish the generality of observed response to genotoxic insult, the human colorectal tumor line HCT116, and HeLa cervical cancer cells were also investigated. Quantitative rises in TG2 were measured by immunoblotting extracts of each line following treatment with either genotoxic agent (Fig. 1A).
In MCF7, 184B5, and HCT116, each of which express wildtype p53, we observed measurable up-regulation of TG2 following genotoxin exposure. However, despite the presence of consensus p53 binding sites within the TGM2 promoter (see Fig.  5A), TG2 up-regulation was also observed in p53-abrogated HeLa cells and p53 mutant MDA-MB-231 cells (data not shown) suggesting that genotoxin-induced up-regulation of TG2 occurs independently of p53. To test this we irradiated and MNNG-treated an HCT116 derivative line with both p53 alleles disrupted (HCT116 p53 Ϫ/Ϫ ) (Fig. 1B). Results from this experiment show clear TG2 up-regulation in response to DNA damage indicating that genotoxin-induced TG2 up-regulation is a p53-independent process.
To test if up-regulation of TG2 stemmed from increased TGM2 transcript abundance we conducted Q-PCR on cells exposed to MNNG or IR. This analysis indicated that each cell line displays multi-fold increases in TG2 mRNA levels in response to either genotoxin (Fig. 1C).
We next sought to determine the p53-independent signaling mechanism responsible for TG2 up-regulation in response to genotoxic stress. We hypothesized that TG2 expression could be controlled by NFB since this transcriptional activator modulates gene expression during DNA damage response (29,30), and the TGM2 gene has been previously shown to possess a canonical NFB binding site within its 5Ј-flanking region (31). Furthermore, we observed increases in nuclear levels of the NFB subunit p65 (RelA) in both irradiated and MNNGtreated cells, consistent with activation of NFB in response to DNA damage ( Fig. 2A).
To test NFB activity in TG2 up-regulation we used BAY 11-7085, a small molecule inhibitor of NFB activity that acts by blocking phosphorylation/degradation of the inhibitory protein IB␣ (32). Specifically, we treated MCF7, 184B5, HCT116, and HeLa cells with BAY 11-7085 only, MNNG only, or the combination of MNNG and BAY 11-7085 (Fig. 2B). When these cells were treated with BAY 11-7085 prior to MNNG exposure we measured blunted TG2 up-regulation in response to DNA damage in each line, indicating that inhibition of NFB limits up-regulation of TG2 during DDR. Furthermore, while BAY 11-7085 alone had no measurable effect on the constitutively low levels of TG2 in MCF7, clear suppression of TG2 expression in 184B5, HCT116, and HeLa cells was observed following treatment with this drug.
NFB is activated in response to DNA damage through a complex signal transduction mechanism dependent upon the damage responsive kinase ATM and the regulatory component of IB kinase, IKK␥/NEMO (33,34). To test if ATM activity is required for up-regulation of TG2 expression during DDR, we treated HeLa cells with KU-55933, a potent small molecule inhibitor of ATM kinase (35) and observed that this drug blocked MNNG-induced TG2 up-regulation (Fig. 2C). To independently confirm this result we used RNAi to knockdown ATM expression in HeLa cells. Specifically, HeLa cells were transduced with lentivirus encoding an ATM-specific shRNA followed by a 2 week selection with puromycin. Subsequently, shRNA-induced ATM knockdown was confirmed by Q-PCR (see Fig. 4A) and immunoblotting (Fig. 4B). We observed that HeLa cells transduced with the control virus displayed TG2 up-regulation following either IR or MNNG, but that ATM knockdown abrogated this effect (Fig. 2D). In sum, we conclude that TG2 is up-regulated in response to genotoxic stress through ATM/NFB-dependent signaling.
NFB Is Necessary and Sufficient to Induce TG2 Expression in Cultured Breast Cancer Cells-During this work we observed that pharmacological inhibition of NFB resulted in reduced baseline expression of TG2 in cell lines tested (see Fig. 2B) suggesting that basal TG2 expression was regulated by NFB. To test a possible role for NFB in controlling TG2 expression, we tested the breast cancer lines MDA-MB-231 and the doxorubicin-resistant MCF7/ADR cell line, both of which express elevated levels of TG2 (18). Consistent with results obtained in other lines, treatment of MDA-MB-231 and MCF7/ADR with BAY 11-7085 resulted in diminished TG2 expression (Fig. 3A). We next used lentiviral-encoded shRNA constructs to knockdown expression of the p65 (RelA) subunit of NFB. MDA-MB-231 and MCF7/ADR cells were transduced with viruses encoding two distinct p65-specific shRNA sequences and stable ployclonal lines selected. Immunoblotting indicated a multi-fold knockdown of p65 in both lines (Fig. 3B). We also observed a significant diminishment in TG2 expression in both lines, indicating that p65 is necessary to support TG2 expression in MDA-MB-231 and MCF7/ADR cells.
To determine if p65 expression is sufficient to drive TG2 expression in breast cancer cells, the MCF7 line was transduced with adenovirus encoding full-length human p65. A non-significant rise in TG2 expression was noted in MCF7 cells trans- duced with control virus; however, findings indicate that expression of p65 in MCF7 cells resulted in a significant increase in TG2 expression (Fig. 3C).
Lastly, we measured relative NFB activity (by a luciferaselinked reporter assay) and TG2 mRNA abundance (by Q-PCR) in a panel of cultured breast cancer/normal lines with low and high levels of TG2 expression (18). In cells with low TG2 expression (MCF7, MDA-MB-468, 184B5, and MCF10A) we measured low NFB transcriptional activity (Fig. 3D) compared with MCF7/ADR and MDA-MB-231 cells, which express both high levels of TG2 expression and NFB activity (Fig. 3E). Moreover, TG2 mRNA abundance and NFB activity exhibit a statistically significant correlation (Pearson correlation coefficient ϭ 0.88; p ϭ 0.02). Taken together, these experiments indicate that the NFB subunit p65 is both necessary and sufficient to drive TG2 expression in cultured human breast cancer cells, and that TG2 expression and NFB transcriptional activity share a reciprocal relationship in breast epithelium and cancer cell lines.
ATM Is Required for Basal Expression of TG2-We demonstrated that ATM/NFB signaling pathway controls TG2 expression in response to genotoxic stress, therefore, we set out to determine if ATM, like NFB, was also required to drive basal TG2 expression. To test this we used shRNA-encoding lentivirus to knockdown ATM expression in MCF7/ADR, MDA-MB-231, and HeLa cells. Q-PCR measured a multi-fold decrease in ATM mRNA abundance in each of these lines compared with cells transduced with control virus (Fig. 4A). We also observed that each line also showed decreased TG2 transcript abundance (Fig. 4A). Coordinately, we also measured reduced ATM and TG2 protein expression in these ATM-knockdown lines when analyzed by immunoblotting (Fig. 4B). We used the drug KU-55933 to inhibit ATM kinase activity in these cell lines in the absence of DNA damage and observed that KU-55933 treatment resulted in measurable decreases in TG2 expression in each cell line as judged by immunoblotting (Fig. 4C). The partial reduction in TG2 expression measured suggests that other mechanism(s) control TG2 expression. Pharmacological inhibitors of the ATM-related kinases ATR and DNA-PKcs did not measurably affect TG2 expression in MDA-MB-231 or MCF7/ADR cells (data not shown) suggesting these kinases do not impact TG2 expression; however, we have not formally ruled out a possible role for these proteins in TG2 regulation.
We next examined NFB activity, as judged by transcriptional reporter assays, in cells with knocked-down or inhibited ATM to ask if ATM activity was required to maintain basal NFB activity in these cell lines. The treatment of MCF7/ADR, MDA-MB-231, or HeLa cells with KU-55933 resulted in a statistically significant drop in NFB transcriptional transactivator activity (Fig. 4D). Likewise, shRNA-mediated knockdown of ATM resulted in significant decreases in NFB reporter activity (Fig. 4E). These results clearly indicate that the ATM/NFB signaling pathway maintains basal TG2 expression in addition to inducing its expression in response to genotoxic stress; however, as TG2 expression was not eliminated following diminishment of ATM or NFB we cannot rule out that other mechanisms control expression of TGM2.
p65 Is a Direct Transcriptional Regulator of the TGM2 Gene-Using the MatInspector software package, we queried ϳ3 kb of human genome sequence (ϳ500 bp downstream and ϳ2500 bp upstream of the transcriptional start site) at the TGM2 locus. This analysis revealed the presence of numerous transcription factor binding sites within this genomic interval (Fig. 5A), including two 10 bp sites conforming to the NFB consensus (GGGRNNYCCC; R ϭ purine, Y ϭ pyrimidine (36)) that we termed Site 1 and 2. Site 2 (5Ј-GGGAAGCCCC-3Ј) is located at nt# Ϫ1352 to Ϫ1361 (relative to translational start site) and Site 1 (5Ј-GGGAGCTCCC-3Ј) is located ϳ700 bp further upstream at nt# Ϫ2112 to Ϫ2121.
To determine if p65 interacted with these sites, we performed ChIP analyses on chromatin harvested from several mammary tumor and non-tumorigenic cell lines. In addition to designing oligonucleotide primers that specifically amplify Site 1 and 2, primers that amplify a 93-bp segment within TGM2 exon 12 (ϳ35 kb downstream of TGM2 exon 1) were used as a control. We observed measurable association of p65 with both Sites 1 and 2 in MCF7/ADR and MDA-MB-231 cells (Fig. 5B). Relative to control IgG immunoprecipitates, in MDA-MB-231 cells we measured a ϳ53 fold increase in p65 binding to Site 1 and a comparable ϳ39-fold increase in binding to Site 2. Similarly, a ϳ78-fold increase in p65 binding was measured at Site 1 and a ϳ68-fold increase at Site 2 was observed in MCF7/ADR chromatin. In contrast, baseline p65 binding at both sites was measured in chromatin harvested from MDA-MB-468 cells.
It has been previously established that the 5Ј flanking region of the TGM2 gene promotes transgene expression in reporter assays (37). To test if the two NFB binding sites uncovered in our work function to modulate transcriptional activity of the TGM2 gene we cloned two portions of the 5Ј-flanking region (designated KN and HN) of TGM2 into the firefly luciferase containing reporter plasmid pGL3-Basic. The HN fragment extends from an NcoI site at the translational start site (nt# ϩ1) to a naturally occurring HindIII site at nt# Ϫ1680 and thus contains only NFB Site 2. The KN fragment extends from the NcoI site to nt# Ϫ2360 (engineered KpnI site) and thus contains both Site 1 and 2. Transcriptional activity was detected in MCF7/ADR cells transiently transfected with pGL3-Basic containing either the HN or KN fragments but very low reporter activity was observed in cells transfected with empty pGL3-Basic plasmid (Fig. 5C). Next, we mutated singly, or in combination, the sequences at Site 1 and Site 2. Specifically, Site 1 was mutated from to 5Ј-GGGAGCTCCC-3Ј to 5Ј-GaGAGCT-CaC-3Ј and Site 2 was mutated from 5Ј-GGGAAGCCCC-3Ј to 5Ј-GtGAAGCCaC-3Ј (mutated bases in lowercase letters) based on mutational strategies used in similar studies (38). Reporter assays using the KN fragment with either Site 1 or 2 mutated indicated that mutants displayed a statistically significant decrease in relative firefly luciferase activity. KN with both Site 1 and 2 mutated exhibited a further decrease in reporter activity. Similarly, mutation of Site 2 within the HN construct exhibited a significant decrease in transcriptional activity (Fig.  5C). We also observed that BAY 11-7085 induced a significant drop in the transcriptional activity of either KN or HN reporters, but no inhibitory effect was observed in cells transfected with a reporter containing either the PDK2 promoter or empty pGL3-Basic plasmid (Fig. 5D). Taken together, these results indicate that NFB has a stimulatory effect on transcription of the TGM2 gene through direct interaction at (at least) two consensus binding sites within the TGM2 promoter. Moreover, as mutation of both Site 1 and 2 failed to completely ablate reporter activity, it is likely that NFB is not the only transcriptional activator of the TGM2 gene.
We next sought to examine if NFB displays heightened association with the TG2 promoter in response to DNA damage. ChIP was performed on chromatin harvested from untreated, ␥-irradiated, or MNNG treated cultures of MCF7 and 184B5 cells. Similar to results obtained on other breast tumor lines, we measured p65 binding to both NFB Site 1 and 2 in untreated cultures of MCF7 and 184B5 (Fig. 5E). In response to irradiation, we measured significant increases in p65 occupancy at both Site 1 and Site 2 in each cell line. Similar results were obtained on irradiated cultures of HCT116 and HeLa cells (data not shown). In sum, these data indicate that during DNA damage response TG2 expression is up-regulated, this event can be blocked by pharmacologic inhibition of NFB, and enhanced levels of the NFB subunit p65 are bound to the TGM2 promoter. Moreover, when combined with the results from previous studies, these results establish a reciprocal relationship between TG2 and NFB. Specifically, TG2 activates NFB transcriptional activity by catalyzing crosslinking and consequential degradation of the IB␣ inhibitor (39), and NFB directly up-regulates TG2 expression through binding consensus sites within the TGM2 promoter.
NFB and ATM Function in the Drug Resistance Phenotype Stemming from Increased TG2 Expression-Since the original observation of Mehta (6), numerous groups have reported that heightened levels of TG2 expression results in anthracycline resistance in a wide variety of cell types (7)(8)(9)(10). Supporting this conclusion are results showing that knockdown of TG2 results in diminished drug resistance (10,40). However, as an upstream activator of NFB, TG2 knockdown will also negatively affect NFB activity. NFB itself drives expression of numerous genes linked to cell survival and drug resistance such as Bcl-xl and survivin (41). Further, as we have documented that TGM2 itself is a target for NFB, obtaining a clear interpreta-tion of the direct effects of TG2 on drug resistance is challenging. Thus we set out to determine the relative contribution of TG2, p65, and ATM to TG2-associated drug resistance. To test FIGURE 5. NFB binds to the TGM2 promoter and regulates its expression. A, diagrammatic representation of the 5Ј-flank of the human TGM2 gene. Illustrated is the transcriptional start site (TSS), translational start site (ATG), and a partial listing of the location of various potential transcription factor binding sites and salient restriction sites within this locus (numbering is relative to translational start site). B, ChIP analysis was conducted on chromatin harvested from the indicated cell lines. Following shearing by sonication, immunoprecipitation was conducted using an anti-p65 antibody and oligonucleotide primers specific for NFB binding Site 1 (white bar), Site 2 (black bar), or exon 12 of the TGM2 gene (shaded bar). Amplification was quantified and graphed relative to amplification obtained using input chromatin. Error bars ϭ 1 S.D., n ϭ 3. C, indicated fragments of the TGM2 promoter (KN and HN) were cloned into the firefly luciferase reporter plasmid pGL3-Basic. Additionally, mutations in Site 1 (S1) and/or Site 2 (S2) were engineered into these reporter constructs. Reporters (or empty pGL3-Basic plasmid) were co-transfected with a Renilla luciferase control plasmid into MCF7/ADR cells. 48 h post-transfection extracts were formed and both firefly and Renilla luciferase activity measured. Error bars ϭ 1 S.D., n ϭ 3-5 independent measurements. D, MCF7/ADR cells were transfected with either the KN or HN reporter plasmids, empty pGL3-Basic, pGL3-Basic containing the PDK2 promoter, or the NFB reporter plasmid pNFB-TA-Luc. Following transfection, cells were either untreated (white bars) or treated with BAY 11-7085 (black bars) prior to analysis of luciferase activity. Error bars ϭ 1 S.D., n ϭ 3-5 independent measurements. E, 184B5 or MCF7 cells were either untreated or treated with MNNG or IR, chromatin extracted and used in ChIP assays to determine p65 binding at NFB Site 1 (white bar) or Site 2 (black bar) within the TGM2 promoter. PCR primers that amplify a portion of TGM2 exon 12 (shaded bar) were used as the control. Amplification was quantified and graphed relative to amplification measured using input chromatin. Error bars ϭ 1 S.D., n ϭ 3. (* indicates p Ͻ 0.05; *** indicates p Ͻ 0.001, Student's t test). this we expressed a full-length copy of the human TG2 cDNA under the control of the CMV promoter in the human breast cancer line MDA-MB-468. This line is well-suited to this experiment as it has epigenetically silenced the TGM2 gene (18) and exhibits very low endogenous TG2 expression and NFB activity (see Fig. 3D).
MDA-MB-468 cells were transfected, drug selected, and cells that stably express recombinant TG2 were obtained (designated 468ϩTG2). Subsequently, these cells, as well as MDA-MB-468 cells stably transfected with empty pcDNA 3.1 vector (designated 468 Ctl), were subsequently transduced with control Lentivirus, or virus encoding p65 or ATM-specific shRNA. Q-PCR analysis indicated that expression of each gene was decreased ϳ4 -5-fold in both control and TG2-expressing lines (Fig. 6A). Coordinately, immunoblot analysis also indicated diminished expression of p65 or ATM in each cell line transduced with specific shRNA virus (Fig. 6B). Moreover, Q-PCR analysis (data not shown) and immunoblot analysis (Fig. 6B) indicated that knockdown of either p65 or ATM had no effect on transgenic TG2 expression in these lines. We next measured NFB transcriptional activity in these cells. Luciferase-based reporter assays measured significant rises in NFB activity associated with the expression of transgenic TG2 (Fig. 6C). Of note, while TG2 expression increased NFB activity ϳ3-fold in both ATM knockdown and control cells, NFB activity was modestly, but measurably, up-regulated in cells transduced with p65 shRNA. This result may indicate forms of NFB containing the p65 subunit are the most prominent, but not the exclusive, forms of this complex to be activated by TG2 expression in this cell line. These cells were treated with doxorubicin and cell viability at both 48 h after drug treatment was measured relative to untreated cells cultured in parallel. As expected, we measured a multi-fold increase in cell viability associated with the expression of recombinant TG2 in MDA-MB-468 cells transduced with control virus (Fig. 6D, VC). Expression of TG2 in MDA-MB-468 cells with knockdown of p65 also raised doxorubicin resistance when compared with p65 knockdown cells not expressing TG2 (Fig. 6D, shp65). This finding suggests that p65 is not strictly required for TG2-induced doxorubicin resistance. Like p65 knockdown, expression of TG2 in MDA-MB-468 cells with knockdown of ATM increased doxorubicin resistance relative to ATM knockdown cells not expressing TG2 (Fig. 6D, shATM). This result similarly indicates that TG2induced drug resistance does not strictly require ATM.
We measured no significant difference in doxorubicin sensitivity in 468 Ctl cells with p65 or ATM knockdown compared with shRNA vector control cells at the doxorubicin concentration (2 M) used in this set of experiments. Nevertheless, when 468ϩTG2 cells were compared with the same cell line with knockdown of either p65 or ATM, a role for both p65 and ATM in TG2-induced doxorubicin resistance was measurable (Fig.  6E). Specifically, diminished p65 expression in 468ϩTG2 displayed significantly reduced drug resistance when compared with 468ϩTG2 cells expressing wild-type levels of p65. Similarly, ATM knockdown in MDA-MB-468ϩTG2 cells also resulted in a significant reduction in cell viability following doxorubicin treatment, although not as dramatically as p65 knockdown. Taken together, we conclude that elevated TG2 expression induces drug (doxorubicin) resistance, and that diminishment of either ATM or NFB partially reduces this drug resistant phenotype. These findings indicate that TG2 overexpression induces doxorubicin resistance through a mechanism(s) that is partially, but not fully independent of ATM and/or NFB.

DISCUSSION
Previous studies established that TG2 expression is up-regulated during response to various cellular stressors (24). In this present work, we document that TG2 is up-regulated in response to both ionizing radiation and DNA alkylation and that this occurs in a variety of tumor cell types. These findings establish that TG2 up-regulation is a general component of the DNA damage response (DDR). We further document that TG2 up-regulation during DDR occurs via the ATM/NFB pathway, an established DNA damage-responsive signaling mechanism (34). While it is unclear precisely what function TG2 plays in response to DNA damage, TG2 has been shown to modulate apoptosis in both a positive and negative fashion (13,(42)(43)(44)(45). Another potentially important effect of TG2 during DDR is the activation of NFB itself, a potent pro-survival transcriptional activator. In support of this, recent findings indicate that dysregulated NFB activity in ATM-deficient cells is responsible for their radio-sensitive phenotype (46).
We also document that baseline expression of the TGM2 gene is under the transcriptional control of NFB. Both RNAimediated knockdown and pharmacological inhibition support our conclusion that NFB is necessary for maintaining TG2 expression in several cultured lines. In addition, ectopic expression of p65 in MCF7 cells indicates that increased NFB activity is sufficient to induce TG2 expression in this line. Our findings support the conclusion that this up-regulation is direct in nature as the NFB subunit p65 associates with two identified consensus NFB binding sites within the TGM2 promoter, mutation of these sites reduces TG2 expression, and p65 binds to these sites in response to DNA damage. Others have reported that NFB Site 2 within the TGM2 promoter interacts with an NFB complex in gel-shift assays (31,47); however, our study is the first to demonstrate a functional effect associated with this interaction.
Work conducted in other laboratories established that highlevel expression of TG2 activates NFB (11,12). One mechanism linked to this effect is the inactivation of the NFB inhibitor IB␣ through TG2-dependent cross-linking (39). This, combined with the findings outlined in this study, highlights a previously unrecognized molecular feedback loop. Specifically, TG2 activates NFB by inactivating IB␣ and, in turn, NFB functions as a direct transcriptional activator of TGM2. Moreover, since we documented a reciprocal relationship between NFB activity and TG2 expression within our panel of cultured breast cancer cells, we propose that that this TG2/NFB signaling loop is self-reinforcing in nature and may perhaps be selfamplifying as well.
Numerous groups, including ours, have observed that increased TG2 expression produces resistance to a number of chemotherapeutic genotoxins, most notably doxorubicin (48). Using MDA-MB-468 cells, which have silenced TGM2 transcription, we expressed recombinant human TG2 which was unaffected by knockdown of p65 (because transgene expression is driven off the Cytomegalovirus (CMV) enhancer-promoter). This design allowed examination of the role of p65 in driving drug resistance in these cells in isolation. The results indicated that knockdown of RelA/p65 resulted in a significant decrease in drug resistance. These experiments clearly indicate that doxorubicin resistance resulting from TG2 expression significantly, but not exclusively, stems from increased NFB activity in MDA-MB-468 cells. Study of diverse cell types uncovered that TG2 can activate a spectrum of events that elicit pro-survival/anti-apoptotic responses (49); thus, in MDA-MB-468 cells, TG2 may function in controlling drug resistance through a mechanism(s) that is independent of its ability to activate NFB.
The kinase ATM, which is activated in response to both IR and MNNG (50,51), is required to drive TG2 expression through an NFB-dependent mechanism during DDR. The Miyamoto laboratory documented that upon DNA damage, ATM phosphorylates NFB essential modulator (IKK␥/ NEMO) and along with NEMO relocalizes to the cytoplasm. In the cytoplasm, the ATM/NEMO complex, in association with the ELKS protein serves to activate the IKK kinase, which, in turn, activates NFB by directing ubiquitin-mediated destruction of IB␣ (34). More recently, this group showed that the ATM/NEMO/ELKS-dependent phosphorylation of IKK is mediated through the kinase TAK1 (52). Our work clearly indicates that one of the downstream events stemming from the activation of this signaling is the transcriptional activation of the TGM2 gene. Moreover, as inhibition or knockdown of ATM reduces both baseline TG2 expression and NFB activity it is tempting to speculate that this same mechanism is responsible for controlling basal TG2 expression in cultured cells as well.
Our findings seemingly indicate that ATM is constitutively active in the cultured lines, including several breast cancer lines, included in our study. Constitutive activation of ATM has been previously observed in cultured tumor cells (53) and the levels of phosphorylated ATM (consistent with catalytic activity) are increased in early stage primary tumors (54). The underlying reason for this constitutive ATM activity remains unknown but may stem from increased production of reactive oxygen species, a common feature of cancer cells (55,56). This increase in oxidative stress, and pursuant DNA damage, will activate ATM as this molecule has recently been shown to directly monitor cellular redox state in addition to its genotoxin-induced activation (57). Similar to our findings, the Kroemer laboratory demonstrated that constitutive ATM activity is required to support NFB activation in a subset of MDS/ AML cell lines and that this dysregulation is associated with high-risk tumor behavior (58). Whereas high level expression of TG2 is commonly found in breast tumors (59), and dysregulated NFB is associated with endocrine resistant breast cancer (for review see Ref. 60), it is tempting to speculate that constitutive ATM activity may be important in shaping breast cancer phenotype through a previously unrecognized tumorigenic property.