Transglutaminase 2 protects against ischemic insult, interacts with HIF1β, and attenuates HIF1 signaling

Abstract

Transglutaminase 2 (TG2) is a multifunctional enzyme that has been implicated in the pathogenesis of neurodegenerative diseases, ischemia, and stroke. The mechanism by which TG2 modulates disease progression have not been elucidated. In this study we investigate the role of TG2 in the cellular response to ischemia and hypoxia. TG2 is up-regulated in neurons exposed to oxygen and glucose deprivation (OGD), and increased TG2 expression protects neurons against OGD-induced cell death independent of its transamidating activity. We identified hypoxia inducible factor 1β (HIF1β) as a TG2 binding partner. HIF1β and HIF1α together form the heterodimeric transcription factor hypoxia inducible factor 1 (HIF1). TG2 and the transaminase-inactive mutant C277S-TG2 inhibited a HIF-dependent transcription reporter assay under hypoxic conditions without affecting nuclear protein levels for HIF1α or HIF1β, their ability to form the HIF1 heterodimeric transcription factor, or HIF1 binding to its DNA response element. Interestingly, TG2 attenuates the up-regulation of the HIF-dependent proapoptotic gene Bnip3 in response to OGD but had no effect on the expression of VEGF, which has been linked to prosurvival processes. This study demonstrates for the first time that TG2 protects against OGD, interacts with HIF1β, and attenuates the HIF1 hypoxic response pathway. These results indicate that TG2 may play an important role in protecting against the delayed neuronal cell death in ischemia and stroke.—Filiano, A. J., Bailey, C. D. C., Tucholski, J., Gundemir, S., Johnson, G. V. W. Transglutaminase 2 protects against ischemic insult, interacts with HIF1β, and attenuates HIF1 signaling.

Transglutaminase 2 (TG2; EC 2.3.2.13) is a member of the transglutaminase (TG) family of thiol-dependent enzymes that catalyze calcium-dependent peptide crosslinking, polyamination, or deamination reactions at glutamine residues within specific protein substrates (for review, see ref. 1). In addition to its TG enzymatic activity, TG2 facilitates intracellular signaling by binding and hydrolyzing GTP (2,3,4). TG2 now appears to be a truly multifunctional protein, as it has also been demonstrated to act as a protein disulfide isomerase (5), a kinase (6), and a protein scaffold (7, 8). TG2 is responsible for the majority of TG activity in the brain (9), where it is expressed predominately within neurons (10,11,12,13) and likely is involved in many physiological (3, 14,15,16,17,18,19,20,21) as well as pathological (22,23,24) cellular processes. For example, TG2 has been demonstrated to facilitate excitotoxicity in primary cerebellar granule cells (25) and apoptosis in neuroblastoma cell lines (26,27,28). However, TG2 protects cardiomyocytes against ischemia and reperfusion-induced cell death (29) and SH-SY5Y neuroblastoma cells against heat shock-induced death (28).

A great deal of effort has been placed on determining the role for TG2 in neurodegeneration as it is up-regulated in many older age onset neurological disorders (10, 11, 30,31,32,33,34,35,36). In addition, TG2 activity and expression are increased in animal models of cerebral ischemia and stroke (37, 38). The precise effects of TG2 in clinically relevant pathological conditions such as ischemia and stroke must be determined before pharmacological manipulation of TG2 activity is attempted to ameliorate associated neuropathology.

The purpose of this study was to examine the effects of TG2 on neuronal cell survival in response to ischemic or hypoxic insult. In this study, we showed that TG2 is up-regulated and neuroprotective in response to oxygen and glucose deprivation (OGD) in both primary rat cortical neurons and in a human neuroblastoma cell line and that the protective effect is independent of the transamidating activity of TG2. We subsequently identified HIF1β as a TG2-binding protein through a yeast 2-hybrid screen. HIF1β is a subunit of the heterodimeric transcription factor, hypoxia inducible factor 1 (HIF1), that is responsible for the majority of the response to hypoxia (39) by up-regulating genes containing hypoxic response elements (HREs; ref.40). While many HIF-dependent genes are those needed for adaptation and cell survival in low-oxygen environments (41), HIF activation has recently been reported to enhance cell death (42,43,44,45). This disparity may be due to the duration and severity of ischemia. Also, regulation at the transcriptional level is due to many coactivators utilized by HIF transcriptional machinery (46). In this study, we showed that TG2 significantly attenuates the ability of HIF to promote HRE-dependent transcription under hypoxic conditions in a manner that is not dependent on the transamidating activity of TG2. Interestingly, TG2 selectively decreases the OGD-induced up-regulation of the HIF-dependent, proapoptotic gene Bnip3 but has no significant effect on the OGD-induced increase in vascular endothelial growth factor (VEGF) gene expression. As a means to delineate the biochemical cause of this novel role for TG2 within the HIF hypoxic response pathway, we showed that TG2 does not affect the levels of HIF1α or HIF1β in the nucleus under hypoxic conditions, or their ability to form a functional HIF1 heterodimer within the nucleus. Further, the interaction between TG2 and HIF1β does not affect the ability of HIF1 to bind to HRE sequences under hypoxic conditions, suggesting that TG2 likely negatively regulates HIF1-dependent transcriptional machinery. These findings are especially interesting because TG2 protein is up-regulated in the brain in animal models of ischemia and stroke and because HIF1 plays a central role in modulating the resulting neuronal cell death. Our findings, therefore, indicate a cellular mechanism by which TG2 may limit the delayed neuronal cell death that occurs subsequent to ischemia and stroke.

MATERIALS AND METHODS

Cell culture

Dissociated rat primary cortical neurons were cultured as described previously (47). Briefly, rat embryo E17 cortices were dissected. Cells were dissociated with trypsin and mechanical force, and then plated in Dulbecco modified Eagle medium (DMEM; Cellgro, Manassas, VA, USA) with 5% fetal bovine serum (FBS; Life Technologies, Inc., Carlsbad, CA, USA), 100 U/ml penicillin (Life Technologies), and 100 μg/ml streptomycin (Life Technologies) for 5 h. Subsequently, medium was completely replaced with Neurobasal medium supplemented with B27 (Invitrogen, Carlsbad, CA, USA), 100 U/ml penicillin (Life Technologies), and 100 μg/ml streptomycin (Life Technologies) to promote neuronal cell growth while limiting glial contamination. Neurons were used at 9–14 days in vitro (DIV) due to the fact that they are synaptically connected and vulnerable to excitotoxic cell death at this time point (48).

SH-SY5Y neuroblastoma cells stably transfected with pcDNA3.1(+) vector (SH/vector), pcDNA3.1(+)-TG2 (SH/TG2), or pcDNA3.1(+)-C277S TG2 (SH/C277S-TG2; a TG-inactive mutant) have been described previously (18) and were maintained in RPMI medium (Cellgro) supplemented with 10% horse serum (Life Technologies), 5% fetal clone 2 (HyClone, Logan, UT, USA), 100 μg/ml G418 (Alexis Biochemicals, San Diego, CA, USA), 100 U/ml penicillin (Life Technologies), 100 μg/ml streptomycin (Life Technologies), and 2 mM l-glutamine (Life Technologies). Chinese hamster ovary (CHO) cells were maintained in Ham’s F-12/DMEM (Cellgro) supplemented with 5% FBS (Hyclone), 100 U/ml penicillin (Invitrogen), 100 μg/ml streptomycin (Invitrogen), and 2 mM l-glutamine (Invitrogen). The 786-0 renal clear-cell carcinoma line was purchased from American Type Culture Collection (Manassas, VA, USA). Cells were grown in DMEM (Cellgro) supplemented with 10% FBS (Life Technologies), 100 U/ml penicillin (Life Technologies), and 100 μg/ml streptomycin (Life Technologies). Unless otherwise stated, all cells were grown in a passive humidified atmosphere containing 5% CO₂ at 37°C.

Constructs and recombinant protein preparation

Constructs for human TG2 [in the pcDNA3.1(+) vector] and C277S-TG2 have been previously described (8). A cDNA for human HIF1β in the pCMV6-XL5 vector was purchased from Origene (Rockville, MD, USA). The HIF1β insert was amplified by polymerase chain reaction (PCR) with a forward primer containing a BamH1 digestion site (5′-CGC GGA TCC ATG GCG GCG ACT ACT GCC AAC C-3′) and a reverse primer containing an XhoI digestion site (5′-CGG CTC GAG TTC TGA AAA GGG TGG AAA CAT AGT TAG ATC AGG-3′). This PCR product was digested and ligated into the pCMV-Tag 5A vector (Stratagene, La Jolla, CA, USA) fused in frame to a C-terminal myc tag. The pGL3-SV40–6HRE reporter vector was a kind gift from Dr. Carine Michiels (University of Namur, Namur, Belgium). The pRL-TK Renilla luciferase vector was from Promega (Madison, WI, USA). The construct encoding glutathione S-transferase (GST) -TG2 was made by PCR of the TG2 coding sequence from the pcDNA3.1(+)-TG2 template, using a forward primer containing an EcoRI digestion site (5′-GGA ATT CCG ATG GCC GAG GAG CTG GTC TTA G-3′) and a reverse primer containing an XhoI digestion site (5′-CCG CTC GAG TTA GGC GGG GCC AAT GAT GAC A-3′). The PCR product was digested and ligated into the pGEX-6P-2 vector (Amersham Biosciences, Piscataway, NJ, USA) in frame with a C-terminal GST tag. The empty pGEX-6P-2 vector served as the template to express GST alone.

For the preparation of GST or GST-TG2 protein, M15 strain Escherichia coli containing pREP4 (Qiagen, Valencia, CA, USA) and either pGEX-6P-2 or pGEX-6P-2-TG2 was grown in 1 L of 2× YT medium with 100 μg/ml ampicillin and 50 μg/ml kanamycin to an A₆₀₀ of 0.6–0.7. Expression of GST or GST-TG2 was induced with 300 μM isopropyl-1-thio-β-d-galactopyranoside (IPTG) overnight at 16°C. Bacterial cells were lysed under N₂ cavitation, resuspended in PBS containing 0.1% (v/v) Triton X-100, and sonicated at 35 Hz for 1 min. Lysates were incubated with prewashed Fast Flow Glutathione-Sepharose beads (Amersham Biosciences) for 1 h with rotation at 4°C. Beads were packed into a column and washed with 100 ml of PBS at 4°C. GST or GST-TG2 was eluted by the addition of a buffer containing 10 mM reduced glutathione and 50 mM Tris, pH 8.0. Eluted protein was then dialyzed against PBS, and the protein concentration was measured using the bicinchoninic acid assay (Pierce, Rockford, IL, USA).

Transient transfection

Transient transfection of cortical neurons was performed on cells at 9 DIV by preincubating DNA with Lipofectamine 2000 (Invitrogen) at a 1:3 ratio in DMEM for 20 min and then adding to cultures for 24 h. Medium was then replaced with conditioned B27-Neurobasal medium. SH-SY5Y and 786-0 cells were transiently transfected using Fugene 6 reagent (Roche Molecular Biochemicals, Billerica, MA, USA) according to the manufacturer’s protocol.

Hypoxic treatments

Cells were transferred from ambient oxygen levels to near anoxic oxygen levels for 16 h at 37°C by infusing 95% N₂/5% CO₂ into the incubator to displace the oxygen. For OGD experiments and the measurement of HIF1α protein levels, cells were placed in a polymer hypoxic glove box (Coy Laboratory Products Inc., Grass Lake, MI, USA) containing 0.1% oxygen at 37°C prior to protein collection. Epoximycin (200 nM; Calbiochem, San Diego, CA, USA) was added to the cell medium 5 min before protein collection for HIF1α, which was performed within the chamber due to the rapid degradation of HIF1α when exposed to oxygen.

Prior to the treatment of primary neurons, neurons were cotransfected (see above) with Renilla luciferase (to measure neuronal cell death; refs. 49, 50) and with vector, TG2, or C277S-TG2 at 9 DIV. At 12 DIV, neurons were placed in B27-Neurobasal medium without glucose (Invitrogen) and transferred to the hypoxic glove box for 2 h. After treatment, neurons were removed and placed back into ambient culture chambers in conditioned medium for 24 h. As a control, neurons were manipulated similarly but maintained in glucose-containing medium in normoxic conditions. Renilla activity was measured using the Luciferase Reporter Assay System Kit (Promega) and a TD-20/20 luminometer (Turner Designs, Sunnyvale, CA, USA) according to the manufacturer’s protocol. Percentage of cell survival was calculated as 100% minus the ratio of Renilla signal from oxygen- and glucose-deprived neurons over that from normoxic controls. Cotransfection with either TG2 or C277S-TG2 had no effect on Renilla signal.

For OGD treatments on SH/vector, SH/TG2, and SH/C277S-TG2 cells, cells were transferred to serum-free RPMI medium without glucose (Cellgro) and placed in the hypoxic chamber for 12 h. After treatments, medium and cellular fraction samples were collected, and LDH activity was measured using a cytotoxicity detection kit (Roche, Indianapolis, IN, USA) according to the manufacturer’s protocols. The percentage of LDH activity in the cell fraction was normalized to control values (normoxia with glucose), and data were represented as percentage of cell viability.

Cellular fractionation

SH-SY5Y and CHO cells were collected and fractionated into cytosolic and nuclear compartments as follows: Cells were washed in ice-cold PBS (pH 7.4) and collected into a lysis buffer [10 mM Tris (pH 7.5), 10 mM NaCl, 3 mM MgCl₂, 1 mM EGTA, and 0.05% (v/v) Nonidet P-40]. All buffers were supplemented with the following protease inhibitors: 2 mM phenylmethylsulfonyl fluoride (PMSF), 20 μg/ml leupeptin, 20 μg/ml pepstatin, and 10 μg/ml aprotinin. A small amount of total cell lysate was set aside to measure total cellular protein expression. The remaining lysed cells were centrifuged at 2700 g for 10 min at 4°C. The supernatant containing the cytosolic fraction was further cleared of cellular debris by centrifugation at 20,000 g for 10 min at 4°C. The pellet from the first centrifugation step containing nuclei was washed 2× in nuclear wash buffer [5 mM HEPES (pH 7.4), 3 mM MgCl₂, 1 mM EGTA, 250 mM sucrose, and 0.1% (w/v) BSA]. Washed nuclei were then layered onto 1 ml of 1 M sucrose and centrifuged at 2700 g for 10 min at 4°C. Nuclei within the resulting pellet were washed once in lysis buffer and then resuspended in nuclear extraction buffer [20 mM HEPES (pH 7.9), 300 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, and 10 μM β-glycerolphosphate] for 30 min at 4°C. Lysed nuclei were centrifuged at 20,000 g for 15 min at 4°C, and the supernatant containing extracted nuclear protein was collected. Protein concentrations in total, cytosolic, and nuclear extracts were determined by the bicinchoninic acid assay.

Pull-down assay, immunoprecipitation, and immunoblotting

For the in vitro GST pull-down assay, prewashed Fast Flow Glutathione-Sepharose beads (50 μl) were incubated with equimolar amounts of either GST (1.1 μg) or GST-TG2 (5 μg) protein for 1 h at 23°C with constant rotation. Beads were washed 2× with PBS and 1× with lysis buffer, then incubated with 50 μg of either cytosolic or nuclear protein extracted from CHO cells that had been transfected with the expression plasmid for HIF1β in a total volume of 500 μl for 3 h at 4°C. Beads were then washed 7× with PBS. Thirty milliliters of 2× sodium dodecyl sulfate (SDS) stop buffer [0.25 M Tris (pH 7.5), 2% (w/v) SDS, 25 mM dithiothreitol (DTT), 5 mM EGTA, 5 mM EDTA, 10% (v/v) glycerol, and 0.01% (w/v) bromphenol blue] was added, and beads were boiled for 15 min. Beads were then centrifuged at 20,000 g for 1 min at 23°C, and the entire supernatant was used for immunoblot analysis.

For immunoprecipitation experiments, SH-SY5Y cells were treated with 10 μM all-trans retinoic acid (Sigma, St. Louis, MO, USA) in RPMI medium containing 5% serum. Medium with 10 μM retinoic acid was replaced on day 3, and cells were collected in lysis buffer on day 5. Magnetic beads (M-280 Dynabeads®; Invitrogen) that were coupled with sheep anti-mouse immunoglobulin G (IgG) antibodies (Invitrogen) were washed 3× in wash buffer [PBS with 0.1% (w/v) BSA] and then incubated overnight with 2 μg of either mouse anti-TG2 (CUB7402; NeoMarkers, Fremont, CA, USA), anti-HIF1β (G-3; Santa Cruz Biotechnology, Santa Cruz, CA, USA) or anti-HIF1α (H1α67; Calbiochem) antibody at 4°C with constant agitation. Beads were washed 3× with wash buffer and incubated with 500 μg of total cell lysate for immunoprecipitation of endogenous proteins, or 100 μg cytosolic or 25 μg nuclear protein extracts for exogenously expressed proteins, for 3 h at 4°C with constant agitation. Beads were washed 3× in wash buffer, and bound protein was collected into 2× SDS stop buffer by boiling, as described below.

For immunoblot analysis, cell lysates were diluted into 2× SDS stop buffer, boiled, and electrophoresed on 8% SDS-polyacrylamide gels. Electrophoresed protein was then transferred to nitrocellulose membranes, and nonspecific sites were blocked by incubating membranes in blocking solution [5% (w/v) skim milk powder in Tris-buffered saline containing Tween 20 (TBS-T) (20 mM Tris, pH 7.5; 137 mM NaCl; and 0.05% (v/v) Tween 20)] for 1 h at room temperature. Membranes were washed with TBS-T and incubated with primary antibodies for either HIF1β (1:500 dilution), TG2 (TG100; NeoMarkers; 1:1000 dilution), α-tubulin (1:2000 dilution; Sigma), histone H4 (1:2000 dilution; Cell Signaling, Danvers, MA, USA) or HIF1α (1:1000 dilution) in blocking solution overnight at 4°C. Membranes were then washed with TBS-T and incubated with a 1:2500 dilution in blocking solution of a goat anti-mouse IgG (H+L) secondary antibody conjugated with horseradish peroxidase (HRP) (Bio-Rad Laboratories, Hercules, CA, USA) for 1 h at 23°C. Membranes were washed with TBS-T and developed by chemiluminescence following 1 min incubation in developing buffer [48.3 mM Tris (pH 8.6), 681 μM luminol, 32.3 μM paracoumeric acid, and 0.0045% (v/v) H₂O₂]. Immunoblots that were reprobed were first stripped for 15 min in stripping buffer [62.5 mM Tris (pH 6.8), 100 mM 2-mercaptoethanol, and 2% (w/v) SDS] at 55°C. Membranes were washed with TBS-T and then reprobed as described above, starting at the blocking step.

Yeast two-hybrid assay

The Matchmaker™ yeast two-hybrid system and an adult human brain cDNA library were purchased from Clontech (Mountain View, CA, USA) and used according to the manufacturer’s instructions. In brief, full-length human TG2 was used as bait for this screen and was expressed fused to the GAL4 DNA binding domain by subcloning into the pGBD-C2 vector (pGBD-C2-TG2). Prey peptides derived from the human adult brain cDNA library were expressed in the pACT2 vector and were fused to a transcriptional activation domain. AH109 strain Saccharomyces cerevisiae were sequentially transformed with pGBD-C2-TG2 and pACT2-library constructs using the lithium acetate/single-stranded carrier DNA/polyethylene glycol protocol. A small amount of the transformed yeast was plated on synthetic dropout medium lacking leucine and tryptophan in order to estimate the total number of transformants. Because the AH109 yeast strain is a histidine and adenine auxotroph containing GAL4 histidine and GAL4 adenine reporter loci, the remainder of the transformation was screened for potential interactions by plating yeast on synthetic dropout medium lacking leucine, tryptophan, and histidine, followed by a second screening on synthetic dropout medium lacking leucine, tryptophan, and adenine. Individual pACT2-library cDNAs encoding potential TG2-interacting clones were recovered from yeast using the Geneclean® III kit (Qbiogene, Carlsbad, CA, USA) and then purified via antibiotic selection in DH5α E. coli and nutritional selection in KC8 strain E. coli. Yeast two-hybrid interactions were confirmed by double transforming AH109 strain S. cerevisiae with the pGBD-C2-TG2 construct along with each individual purified pACT2-library cDNA and screening on synthetic dropout medium lacking leucine, tryptophan, histidine, and adenine.

HRE luciferase reporter assay

SH/vector, SH/TG2, and SH/C277S-TG2 cells were transiently transfected with cDNA expression plasmids for HIF1β, firefly luciferase (under control of the HRE promoter), and Renilla luciferase using Fugene 6 reagent as described above. At 6 h post-transfection, cells were placed in a hypoxic environment as described above for 16 h at 37°C. 786-0 cells were placed in serum-free DMEM and transfected with firefly luciferase (under control of the HRE promoter) and Renilla luciferase using Fugene 6 reagent as described above and allowed to express protein for 24 h. Luciferase activity was measured in cellular lysates using the Dual-Luciferase Reporter Assay System Kit (Promega) according to the manufacturer’s protocol and a TD-20/20 luminometer (Turner Designs). For each sample, the firefly luciferase data were normalized to the Renilla luciferase internal control.

RNA collection and quantitative real-time PCR (Q-RT-PCR)

SH-SY5Y stable cell lines were grown in either RPMI medium in normoxic conditions or glucose-free RPMI medium with 0.1% O₂ for 12 h. RNA was collected using Trizol reagent (Invitrogen). After isopropanol precipitation, RNA was treated with DNase I (Invitrogen) to eliminate any contamination of genomic DNA. RNA concentrations were measured and cDNAs were created by using SuperScript III Reverse Transcriptase and Random Primers (Invitrogen). cDNAs were diluted 1:10, and 2 μl/well was used in a 96-well plate format for Q-RT-PCR analysis of VEGF (Hs00173626_m1) and Bnip3 (Hs00969291_m1) premade primer sets. Data were normalized to β-actin (Hs00242273_m1) (Applied Biosystems, Foster City, CA, USA). Amplification was performed using an ABI Prism 7300 Sequence Detection System (Applied Biosystems) at 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Data were analyzed using the ΔΔCt method.

HIF1/HRE DNA binding assay

SH/vector, SH/TG2, and SH/C277S-TG2 cells were grown under either normoxic or hypoxic conditions as described above. Binding of HIF1 within nuclear protein fractions to the HRE was measured by modified ELISA using the TransAM™ HIF-1 Transcription Factor Assay Kit (Active Motif, Carlsbad, CA, USA) according to the manufacturer’s recommendation. Briefly, nuclear lysates were incubated in microplate wells that had been previously coated with HRE DNA sequences. Wells were washed and incubated with a mouse anti-HIF1α monoclonal antibody for 1 h, followed by an HRP-conjugated anti-mouse secondary antibody for 1 h. The degree of HIF1 binding to the HRE within each sample was then indirectly measured by colorimetric detection, followed by the measurement of sample absorbance at 450 nm. For each experiment, the A₄₅₀ for each sample was normalized to that for a supplied positive control (CoCl₂-treated COS-7 cell nuclear extract).

Data analysis

All immunoblots shown are representative of at least 3 independent experiments. Data for OGD, HRE luciferase activities, Q-RT-PCR, and HIF1/HRE binding are presented as means ± se for at least 3 independent experiments. Statistical analysis of luciferase activity in SH-SY5Y cells (Fig. 4) and HIF1/HRE binding (Fig. 9) was performed using 2-way analysis of variance (ANOVA) in order to identify significant differences based on the two independent variables of TG2 expression and oxygen tension. One-way ANOVA was used to analyze data collected from luciferase assays in the 786-0 cell line and data from qRT-PCR. A 2-tailed paired t test was used to compare neuronal cell survival on each experimental day in order to determine the effect of TG2. Values of P < 0.05 were considered statistically significant.

Figure 1.

TG2 is up-regulated after OGD, and increased expression of TG2 is neuroprotective. A, B) Rat primary cortical neurons were treated with OGD for 1 or 2 h at 12 DIV. Cells were placed back in conditioned medium and ambient culture chambers for 24 h before collection. A) Immunoblots show that endogenous rat TG2 is up-regulated 24 h after 1 or 2 h of OGD. Blots were reprobed with α-tubulin to show equivalent loading. B) Transient transfection of TG2 and C227S-TG significantly protects rat primary neurons from 2 h of OGD (P<0.05). C) SH-SY5Y cells that stably overexpress TG2 (SH/TG2 or SH/C277S-TG2) are significantly more resistant to cell death induced by 12 h of OGD (P<0.05).

Figure 2.

TG2 interacts with HIF1β in vitro. A GST pull-down assay was performed using recombinant GST-TG2 and HIF1β from transiently transfected CHO cell nuclear (NUC) and cytosolic (CYTO) fractions. A) Immunoblot for HIF1β protein in cytosolic and nuclear fractions. B) Cytosolic and nuclear lysates were incubated with glutathione beads that had been prebound with either GST or GST-TG2. Immunoblot analysis for GST (top panel) demonstrates that GST (lanes 1 and 3) and GST-TG2 (lanes 2 and 4) had bound to beads, and immunoblot analysis for TG2 (middle panel) confirms that GST-TG2 (lanes 2 and 4) had bound to beads in this experiment. Immunoblot analysis for HIF1β within the same samples shows that HIF1β was pulled down by GST-TG2 but not by GST alone.

Figure 3.

Endogenous TG2 coimmunoprecipitates with endogenous HIF1β in situ. SH-SY5Y cells were treated with 10 μM all-trans retinoic acid (RA) for 5 days and collected. A) Immunoblots of cells incubated in the absence or presence of RA for 5 days show that TG2 is up-regulated in response to RA treatment. Blots were also probed for HIF1β. B) Four samples of RA-treated cells from 2 independent experiments were immunoprecipitated for HIF1β and immunoblotted for both HIF1β and TG2. The up-regulation of TG2 in response to RA treatment resulted in a detectable interaction between TG2 and HIF1β in all the samples. The last lane shows a control immunoprecipitation in which the immunoprecipitating antibody was omitted.

Figure 4.

TG2 attenuates HIF1-dependent transcriptional activity in situ. SH/vector, SH/TG2, and SH/C277S-TG2 cells were transfected with constructs for firefly luciferase (under control of the HRE promoter) and Renilla luciferase (control). Each cell type was then divided into one of four experimental conditions as indicated, based on growth conditions and the exogenous expression of HIF1β. All data are expressed as the fold increase compared with SH/vector cells grown under normoxic conditions. Luciferase activity did not differ among the 3 cell types when grown under normoxic conditions, regardless of whether HIF1β was exogenously expressed (open and checked bars). For cells grown under hypoxia but without exogenous HIF1β expression (gray bars), HRE activity was decreased in SH/C277S-TG2 cells compared with both SH/vector and SH/TG2 cells (P<0.05), whereas HRE activity did not differ between SH/vector and SH/TG2 cells (P>0.05). For cells grown under hypoxia with exogenous HIF1β expression (solid bars), HRE activity was reduced in both SH/TG2 and SH/C277S-TG2 cells compared with SH/vector cells (P<0.05). HRE activity was not statistically different between SH/TG2 and SH/C277S-TG2 cells (P>0.05).

Figure 5.

TG2 differentially affects the expression of HIF-dependent transcripts. SH/vector, SH/TG2, and SH/C277S-TG2 cells were treated under the same conditions as in the cell death studies (normoxia vs. 12 h OGD) before mRNA collection and generation of cDNA. VEGF and Bnip3 genes were analyzed using Q-RT-PCR normalized to β-actin to control for total mRNA collected. All data are presented as fold induction after OGD. No significant difference in VEGF mRNA levels after OGD was observed in SH/TG2 and SH/C277S-TG2 cells when compared with SH/vector cells (P>0.05). However, a significant attenuation of Bnip3 levels in SH/TG2 and SH/C277S-TG2 was found compared to SH/vector (P<0.05).

Figure 6.

TG2 attenuates HIF transcriptional activity in an HIF1α-independent manner. Human renal carcinoma 786-0 cells, which lack a functional VHL and predominantly express HIF2α over HIF1α, were transfected with constructs for firefly luciferase (under control of the HRE promoter) and Renilla luciferase (control) along with either wild-type TG2, inactive mutant C277S-TG2, or pcDNA 3.1(+) vector control. Cells were allowed to express protein for 24 h in ambient air incubators then collected and HRE activity measured. All data are expressed as the fold increase compared with vector transfected cells. HRE activity was significantly reduced by both wild TG2 and C277S-TG2 compared to vector controls (P<0.001).

Figure 7.

TG2 does not attenuate HIF1α or HIF1β localization to the nucleus under hypoxia. SH/vector, SH/TG2, and SH/C277S-TG2 cells were grown under either normoxic (–) or hypoxic (+) conditions. Cells were collected, and nuclear and cytosolic fractions were separated. A) Immunoblot analysis of the nuclear fractions demonstrates that under hypoxic conditions nuclear HIF1α protein content is greater in SH/TG2 cells compared with SH/vector cells and that nuclear HIF1α protein content is increased further still in SH/C277S-TG2 cells (top panel). Although nuclear HIF1β protein content was greater for all cell types grown under hypoxic conditions compared with normoxic conditions, TG2 did not affect nuclear HIF1β protein content for cells grown in hypoxia (middle panel). TG2 protein content is also shown (bottom panel). B) Nuclear and cytosolic fractions were immunoblotted for α-tubulin as a cytosolic marker and histone subunit H4 as a nuclear marker. These immunoblots show that the nuclear fractions are not contaminated with cytosolic protein in each of the 3 cell lines.

Figure 8.

TG2 does not affect the interaction between nuclear HIF1α and HIF1β under hypoxia. SH/vector, SH/TG2, and SH/C277S-TG2 cells were grown under either normoxic or hypoxic conditions. Cells were collected, and nuclear fractions were isolated. A) Protein content for HIF1α (top panel), HIF1β (middle panel), and TG2 (bottom panel) in 15 μg nuclear protein from the indicated cell type. B) HIF1α was immunoprecipitated from 25 μg nuclear protein that was collected from the indicated cell types. Immunoprecipitated protein was immunoblotted for HIF1α (top panel) and HIF1β (bottom panel). TG2 did not affect the ability of HIF1β to coimmunoprecipitate with HIF1α. Last lane shows a sham immunoprecipitation in which the immunoprecipitating antibody was omitted.

Figure 9.

TG2 does not attenuate HIF1 binding to the HRE under hypoxia. SH/vector, SH/TG2, and SH/C277S-TG2 cells were grown under either normoxic or hypoxic conditions. Cells were collected, and nuclear HIF1 binding to the HRE was measured using a modified ELISA-based assay. HIF1 binding to the HRE was increased significantly for each cell type grown in hypoxic conditions compared with normoxic conditions (P<0.0001). Although HIF1 binding to the HRE was increased in SH/TG2 cells grown under normoxia compared with SH/vector and SH/C277S-TG2 cells (P<0.001), no effect of TG2 or C277S-TG2 on HIF1 binding for cells grown under hypoxic conditions was detected (P>0.05).

RESULTS

TG2 is up-regulated and protects neurons against oxygen and glucose deprivation

Rat primary cortical neurons were placed in glucose-free medium and exposed to 0.1% oxygen for 1 or 2 h. After treatment, neurons were placed back in normoxic conditions for 24 h. Immunoblots of 25 μg cell lysate showed up-regulation of endogenous rat TG2 after 1 h of OGD and an even greater increase after 2 h of OGD (Fig. 1A). Given this up-regulation of TG2 in response to OGD, we next assessed the effects of TG2 overexpression on neuronal viability subsequent to OGD. Neuronal cultures were cotransfected with Renilla luciferase and either pcDNA 3.1(+) vector for control, wild-type TG2, or C277S-TG2. Due to the low transfection efficiency of primary neurons, this established (49, 50) Renilla-based cotransfection cell death assay allowed us to measure only the viability of cells that overexpress TG2, because the cotransfection efficiency of an individual cell with two plasmids using this method is nearly 100% (51). Cotransfected neurons were placed in glucose-free medium and exposed to 0.1% oxygen for 2 h, because this duration of exposure to OGD resulted in maximal up-regulation of TG2 in the neurons (Fig. 1A). After treatment, medium was replaced with conditioned medium containing glucose, and neurons were placed back in normoxic conditions for 24 h. Neurons that overexpressed TG2 (21.2±3.80%; n=16) or C277S-TG2 (21.0±3.16%; n=16) as determined by paired t test were significantly more resistant to OGD-induced cell death compared with cells that expressed empty vector (11.6±2.33%; n=16) (Fig. 1B; P<0.05). Similar results were obtained in SH-SY5Y neuroblastoma cell lines that stably overexpressed either TG2 or C277S-TG2. Cells were placed in near-anoxic conditions in medium without glucose, and an LDH release assay was used to measure cell viability. To calculate cell viability, the percentage of LDH released from cells following OGD was measured and normalized to the amount released from the corresponding cell type maintained in normoxic conditions. SH/TG2 (82.3±8.32%; n=7) and SH/C277S-TG2 (82.4±7.54%; n=7) were significantly less sensitive to OGD when compared with control SH/vector cells (52.5±6.77%; n=6) (Fig. 1C; P<0.05). The results of these studies show that TG2 is protective against OGD-induced cell death in rat primary cortical neurons and SH-SY5Y neuronal cell lines and that the protection is independent of its transamidating activity (Fig. 1). Considering this, we hypothesized that the scaffolding function of TG2 may contribute to its protective effects.

TG2 interacts with HIF1β

A yeast 2-hybrid screen was performed in order to identify novel binding partners for TG2 that are expressed in human brain. Using full-length TG2 fused to the GAL4 DNA binding domain (TG2-BD) as bait, we screened ∼6 × 10⁶ yeast clones that expressed peptides derived from a human brain cDNA library fused to a transcriptional activation domain (library-AD). Within this screen, we identified 13 library-AD clones that encoded portions of 6 unique proteins that interacted strongly with TG2-BD. Although these 6 proteins are not closely related with regard to cellular location or function, note that 4 proteins have been identified to modulate transcription within the nucleus. We identified 2 identical clones that expressed amino acids 453–691 of HIF1β fused with AD (HIF1β-AD) and interacted strongly with the TG2-BD protein (data not shown). This interaction was shown to be specific for TG2 and the HIF1β peptide, as both TG2-BD and HIF1β-AD were required for yeast to grow on synthetic dropout medium lacking leucine, tryptophan, histidine, and adenine.

To verify that TG2 physically interacts with full-length HIF1β, an in vitro GST pull-down experiment was performed with either purified GST or purified GST-TG2 conjugated to glutathione beads. Full-length HIF1β was expressed in CHO cells. HIF1β protein from both the cytosol and nucleus was specifically pulled down by GST-TG2 but was not pulled down by GST alone (Fig. 2), indicating that HIF1β and TG2 interact in this system.

To determine whether TG2 and HIF1β physically interact in an intact mammalian cellular system, naive human neuroblastoma SH-SY5Y cells were used. TG2 is expressed at low levels in naive cells; therefore, to physiologically increase TG2 levels, cells were treated for 5 days with all-trans retinoic acid, a well known inducer of TG2 expression in SH-SY5Y cells (18, 52). Because retinoid response elements are in the promoter of TG2 (53), we used a physiologically relevant ligand to activate the endogenous promoter and increase expression of the protein, rather than overexpressing it from an exogenous source. Total protein was collected, and immunoblots showed a robust increase in TG2 levels in the retinoic acid-treated samples, as expected (Fig. 3A). Interestingly, HIF1β levels also increased. TG2 coimmunoprecipitated with HIF1β in 4 samples from 2 independent cell passages treated with retinoic acid but not in the untreated naive sample (Fig. 3B). Similar results were obtained when exogenous TG2 and HIF1β were overexpressed; the interaction was found to occur in the cytosol and nucleus (data not shown). These results showed that endogenous HIF1β physically interacts with endogenous TG2 in SH-SY5Y cells.

TG2 attenuates HIF1-dependent transcriptional activity

Because TG2 and HIF1β interact and TG2 is up-regulated in cell and animal models of ischemia, the ability of TG2 to modulate the HIF1 hypoxic response pathway was examined. HIF1-dependent transcription was measured in SH/vector, SH/TG2, or SH/C277S-TG2 cells using a luciferase-based HIF1/HRE reporter assay (Fig. 4). Two-way ANOVA demonstrated significant effects of hypoxia (P<0.0001) and TG2 (P<0.0001) on luciferase activity in this cellular model. One-way ANOVA designed to identify differences between cell types within each experimental condition demonstrated that, under normoxic conditions, neither TG2 nor C277S-TG2 influenced HIF1-dependent transcription at the HRE (Fig. 4; P>0.05). Moreover, this lack of effect by TG2 and C277S-TG2 under normoxic conditions also was observed in the presence of exogenously expressed HIF1β, which in itself did not alter transcription at the HRE under normoxic conditions (P>0.05). Hypoxia alone increased luciferase activity compared with normoxia regardless of cell type (Fig. 4; gray bars vs. open bars). Interestingly, in cells that did not express exogenous HIF1β, the increase in luciferase activity under hypoxic conditions (measured as fold increase compared with SH/vector cells grown under normoxic conditions) was significantly attenuated in SH/C277S-TG2 cells (8.93±1.98; n=5) compared with both SH/vector (28.63±2.67; n=5) and SH/TG2 (27.20±6.78; n=5) cells (Fig. 4; P<0.05). In SH/vector cells, the exogenous expression of HIF1β in combination with hypoxia increased luciferase activity 190.3 ± 30.34 fold (n=5) compared with SH/vector cells grown in normoxic conditions. This fold increase in luciferase activity was significantly less in SH/TG2 (90.94±22.01; n=5) and SH/C277S-TG2 (35.00±13.42; n=5) cells compared with SH/vector cells (Fig. 4; P<0.05), whereas the fold increase was not statistically different between SH/TG2 and SH/C277S-TG2 cells (P>0.05).

TG2 attenuates the up-regulation of Bnip3

To determine whether the effect of TG2 on HIF-activated transcription was a function of general HIF-directed gene transcription or specifically controlled, we looked at 2 intrinsic HIF-induced genes; VEGF and Bnip3. SH/vector, SH/TG2, and SH/C277S-TG2 cells were exposed to OGD for 12 h or maintained in normoxic conditions prior to mRNA collection and Q-RT-PCR analyses. VEGF mRNA levels increased in response to OGD in SH/TG2 (12.30±4.922; n=5), SH/C277S-TG2 (17.67±4.437; n=6), and SH/vector (19.13±2.759; n=6) (Fig. 5A; P>0.05). In contrast, the up-regulation of Bnip3 was attenuated in SH/TG2 (2.363±0.423; n=5) and SH/C277S-TG2 (3.289±0.639; n=6) cells when compared to SH/vector cells (6.432±1.032; n=6) (Fig. 5B; P<0.05).

TG2 attenuates HIF-dependent transcription independent of HIF1α and hypoxia

To provide further evidence that the effects of TG2 on HIF signaling are dependent on its interaction with HIF1β, we investigated the effects of TG2 on HIF-controlled transcription in the human renal cancer cell line, 786-0. This cell line lacks the functional HIFα E3 ligase, von Hippel-Lindau (VHL); consequently, the HIF α-subunit cannot be degraded and thus is always present and active (54). In addition, HIF1α protein levels are undetectable in this cell line, and therefore they rely on HIF2α for HIF signaling (55). Cells were cotransfected with the HRE reporter, a Renilla control vector, as well as vector, wild-type TG2, or TG2-C277S, and were allowed to express protein for 24 h in normoxia before collection and reading. One-way ANOVA among the 3 groups shows that TG2 (52.8±5.26%; n=3) and TG2-C277S (53.4±5.85%; n=3) significantly attenuate HIF signaling when compared to cells transfected with vector (100±3.95%; n=3) (Fig. 6; P<0.001).

TG2 does not affect the localization of HIF1α or HIF1β to the nucleus or their physical interaction within the nucleus in hypoxic conditions

Given the effects of TG2 on HIF1 transcriptional activity, the ability of TG2 to affect the levels of HIF1α or HIF1β in the nucleus and their ability to form the HIF1 heterodimer transcription factor under hypoxic conditions was investigated. SH/vector, SH/TG2, and SH/C277S-TG2 cells were grown under either normoxic or hypoxic conditions. Cells were collected, and nuclear fractions were isolated. Immunoblot analysis using 15 μg nuclear protein demonstrated that for all cell types, the nuclear protein levels for HIF1α and HIF1β were greater in cells that were grown under hypoxic conditions compared with cells that were grown under normoxic conditions (Fig. 7). Also, cells that overexpressed TG2 and C277S-TG2 showed increased nuclear levels of TG2. However, the presence of TG2 or C277S-TG2 did not attenuate this increase in nuclear HIF1α or HIF1β under hypoxia. In fact, for cells grown under hypoxia, nuclear HIF1α protein levels were consistently increased in SH/TG2 cells compared with SH/vector cells, and even further increased in SH/C277S-TG2 cells (Fig. 7). This effect of TG2 on increasing nuclear HIF1α protein levels without affecting nuclear HIF1β levels is intriguing, because it appears to be the opposite of the effect of TG2 on HIF1 transcriptional activity (Fig. 4). The purity of the subcellular fractions was evaluated by blotting for the nuclear marker histone subunit H4 and the cytoplasmic marker α-tubulin (Fig. 7B). The effects of TG2 on the physical interaction between HIF1α and HIF1β within the nucleus under hypoxia were examined next. SH/vector cells were grown under normoxic conditions, and SH/vector, SH/TG2, and SH/C277S cells were grown under hypoxic conditions. Cells were collected, and nuclear fractions were isolated. Nuclear protein (15 μg) was immunoblotted for HIF1α (Fig. 8A, top panel), HIF1β (Fig. 8A, middle panel), and TG2 (Fig. 8A, bottom panel). HIF1α was then immunoprecipitated from 25 μg nuclear protein from each cellular type and growth condition. Under hypoxic conditions, neither TG2 nor C277S-TG2 affected the immunoprecipitation or coimmunoprecipitation of HIF1α or HIF1β, respectively (Fig. 8B). These results demonstrate that, although TG2 attenuates HIF1 transcriptional activity at the HRE under hypoxia, it does not affect HIF1 protein levels or heterodimer formation within the nucleus.

TG2 does not affect HIF1 binding to the HRE

The effect of TG2 on HIF1 binding to the HRE was measured to further delineate the mechanism of action by which TG2 attenuates HIF1 transcriptional activity in response to hypoxia. Binding of HIF1 within nuclear lysates of SH/vector, SH/TG2, and SH/C277S-TG2 cells to HRE DNA sequences was measured using a modified ELISA technique. Two-way ANOVA demonstrates a significant effect of hypoxia treatment on increasing HIF1 binding to the HRE (Fig. 9; P<0.0001); however, no significant effect of TG2 on HIF1 binding to the HRE was found (Fig. 9; P>0.05). One-way ANOVA comparing each cell type grown under normoxic conditions demonstrates that HIF1 binding to the HRE in SH/TG2 cells (0.471±0.018; n=3) was significantly greater than both SH/vector (0.216±0.031; n=3) and SH/C277S-TG2 cells (0.171±0.009; n=3) (Fig. 9; P<0.0001 for the ANOVA and P<0.001 for each comparison), whereas no difference was found in HIF1 binding to the HRE between SH/vector and SH/C277S cells (P>0.05). One-way ANOVA comparing each cell type grown under hypoxic conditions demonstrates that no difference was found in HIF1 binding to the HRE among any of the SH-SY5Y cell types (Fig. 9; P>0.05 for the ANOVA and for each individual comparison).

DISCUSSION

TG2 is up-regulated in rat primary cortical cultures exposed to OGD, and overexpression of TG2 is protective in both a neuronal cell line and rat primary cortical neurons. Because this protection is independent of TG2 transamidating activity, we considered the scaffolding function of TG2. We have shown for the first time using a yeast 2-hybrid screen and an in vitro pulldown assay, as well as in situ coimmunoprecipitation experiments of endogenous proteins, that TG2 and HIF1β indeed physically interact in multiple cellular models. Further, we found that TG2 inhibits HIF-dependent transcription at the HRE under hypoxic conditions independent of its activity. Investigation of two classical HIF-induced gene transcripts after OGD show that TG2 attenuates the up-regulation of the proapoptotic BH3-only bcl-2 family protein Bnip3 (56), while the up-regulation of VEGF, which has been shown to be neuronal protective (57), in response to OGD was not affected by TG2. To identify the point within the HIF hypoxic response pathway where TG2 interferes with HIF function, we examined the effects of TG2 on the protein content for both HIF1α and HIF1β in the nucleus under hypoxic conditions, the ability of nuclear HIF1α and HIF1β to coimmunoprecipitate under hypoxic conditions, and the ability of nuclear HIF1 to bind to HRE sequences under hypoxic conditions. Remarkably, TG2 did not affect any of these steps. Because TG2 does not hinder HIF1 at any point up to and including binding to HRE sequences under hypoxic conditions, whereas TG2 does attenuate HIF1-dependent transcription at the HRE under hypoxic conditions, our results strongly suggest that TG2 interferes with transcriptional machinery that is associated with HIF-dependent transcription at the HRE, which leads to overall cellular protection against OGD.

Our current finding that TG2 protects against OGD and attenuates the HIF1 hypoxic response pathway, combined with our data and previous reports that TG2 is greatly up-regulated in cultured neurons exposed to OGD and in the brain under ischemic conditions (37, 38), suggest that TG2 may play an important role in the response of the brain to ischemia and stroke. TG2 itself has been demonstrated to either facilitate or protect against apoptosis in neuronal models, depending on the context and pathogenic stimulus. For example, in SH-SY5Y cells, TG2 facilitates apoptosis when a stimulus increases its TG activity but protects against apoptosis when its TG activity is dormant (28). It should be noted that, although TG2 has been shown to activate PI3K/Akt cell survival pathways in pancreatic ductal adenocarcinoma (PDAC) cells through stimulation of FAK (58), we observed no increased levels in total or phosphorylated FAK in our cell lines pre- or post-OGD (data not shown). These findings indicate that, although TG2 can promote cell survival by FAK activation in PDAC cells, this pathway is not involved in the cytoprotection that occurs in response to OGD. Moreover, although active TG2 facilitates thapsigargin-induced cell death, nuclear localized TG2 does not; in addition, C277S-TG2 protects against thapsigargin-induced cell death in a manner that likely involves its physical association with the antiapoptotic retinoblastoma protein, Rb (8, 59). A recent report shows that Rb can attenuate the up-regulation of the proapoptotic, HIF-dependent Bnip3 (60). Also, Rb has been shown to bind HIF1α and enhance the HIF response (61). These results suggest that the subcellular localization of TG2 has implications in the cell death process. Interestingly, we observed an increase in nuclear TG2 under hypoxic conditions concomitant with the increased levels. It is intriguing to speculate that since TG2 scaffolds with both Rb (8, 59) and HIF1β, it may play an important role in regulating cellular protection at the transcriptional level.

In the whole animal, the outcome of TG2-mediated inhibition of HIF1 function likely depends on the role of HIF1 in the brain’s response to ischemia and stroke. Under hypoxic conditions, HIF1 promotes the expression of a number of genes, including VEGF (62), erythropoietin (EPO) (63), glucose transporter-1 (GLUT-1) (64), and all enzymes in the glycolytic pathway [e.g., hexokinase 2 and phosphofructokinase-L (PFK-L)] (65) that promote adaptation to and survival in low-oxygen conditions. In addition, numerous proapoptotic genes under HIF control have been identified (56, 66, 67), and several recent studies have shown that HIF1 may in fact promote, rather than protect against, the delayed neuronal cell death that occurs subsequent to ischemia and stroke (42,43,44,45, 60, 68). In a cortical cell culture model, HIF1α was demonstrated to facilitate p53-dependent delayed cell death in response to a hypoxic insult (43). In a mouse model with brain-selective late-stage deletion of HIF1α expression, a marked decrease in cerebral cortical and hippocampal neuronal death in response to a global ischemic episode (45) was found, although the role of HIF1α in neuronal cell death after stroke remains controversial (69) and may likely depend on the severity and/or duration of ischemia. Pharmacologically down-regulating HIF signaling protects Sprague-Dawley rats against middle cerebral artery occlusions (42). This may in part be due to proapoptotic genes that are under HIF control (56, 66, 67). These examples and our findings that TG2 selectively attenuates the up-regulation of Bnip3 in response to OGD strengthen the hypothesis that TG2-dependent attenuation of HIF1 signaling may be beneficial to cell survival after hypoxia and ischemia. Given these findings, it can be hypothesized that TG2 may play a protective role in the brain in response to ischemia and stroke. Indeed, one recent report indicates that TG2 does in fact protect heart tissue against ischemic damage in vivo (29).

HIF1/HRE luciferase data shown in Fig. 4 demonstrate that, in the stably transfected SH-SY5Y cells grown in hypoxic conditions, HIF1β is the limiting component of the HIF1 transcription factor. For example, in each cell type, luciferase activity was significantly greater for cells that overexpressed HIF1β (solid bars) compared with those that did not (gray bars). In hypoxic cells that overexpressed HIF1β, the effect of TG2 to decrease luciferase activity demonstrates for the first time that TG2 disrupts HIF1 transcriptional activity at the HRE under these experimental conditions. Although the reduction in luciferase activity appears to be greatest in SH/C277S-TG2 cells, luciferase activity values in both SH/TG2 and SH/C277S-TG2 cells are not statistically different. Because the C277S-TG2 protein lacks TG activity and shows greatly diminished GTPase activity compared with wild-type TG2 (ref. 70 and unpublished results), these results strongly suggest that TG2 acts via protein-protein interactions to attenuate HIF1 transcriptional activity at the HRE in response to hypoxia.

To identify whether this attenuation of HIF signaling was general to all HIF-induced transcription or has some level of gene specificity, we looked to the classic hypoxia-induced HIF-dependent genes VEGF and Bnip3. We found that TG2 has no effect on the up-regulation of VEGF; however, it significantly attenuates the up-regulation of proapoptotic Bnip3 (Fig. 5). Because TG2 does not affect any steps before and including HIF binding to the HRE, we hypothesize that TG2 works at the level of the transcriptional machinery. It is interesting that activation of the steroid receptor coactivator-1 (SRC-1) has been shown to be required for hypoxia-induced expression of VEGF in fibroblasts (71). This activation facilitates phosphorylation of STAT3, which forms a transcriptional complex with HIF1α, CBP/p300, and Ref-1/APE at the VEGF promoter. Alternatively, mutations in the CH1 domain of CBP/p300 have little effect on the hypoxic regulation of Bnip3 (46). These examples illustrate the complexity of HIF-dependent gene regulation at the HRE, and TG2 may use a plethora of targets to modulate this regulation.

We further confirmed that the effects of TG2 on HIF signaling are due to the interaction of TG2 and HIF1β by using the renal clear-cell carcinoma 786-0 cell line. This cell line is characterized by mutations in the HIFα E3 ubiquitin ligase enzyme VHL, which leads to constitutively active HIF signaling (54). Also, these cells do not express HIF1α and instead rely on a highly expressed HIF2α (55). In this model, we observed that TG2 and C277S-TG2 significantly attenuate transcription of a HIF-dependent reporter in normoxic conditions (Fig. 6), demonstrating that the roles of TG2 in HIF signaling are not a result of nonspecific hypoxia events or HIF1α effects.

Because TG2 likely interferes with the HIF1 hypoxic response pathway through protein–protein interactions, we hypothesized that TG2 may decrease either the nuclear protein levels for HIF1α or HIF1β or their ability to interact and form the HIF1 heterodimer transcription factor within the nucleus under hypoxic conditions. Contrary to this hypothesis, we found that TG2 does not affect either of these components of the HIF1 hypoxic response pathway. The next potential step in the HIF1 hypoxic response pathway at which TG2 may interfere is the actual binding of HIF1 to the HRE DNA sequence. Data from the HIF1/HRE binding experiment clearly demonstrated that TG2 also does not inhibit binding of HIF1 from hypoxic nuclear lysates to HRE DNA sequences. It is most interesting to point out that, although the HIF1 binding values for cells grown in hypoxic conditions are not statistically different, a trend toward greater binding in the SH/TG2 cells compared with SH/vector cells is apparent, and this increase is greater still in SH/C277S-TG2 cells. These results are intriguing, because this trend is the same as that observed for nuclear HIF1α protein levels and is the opposite of that observed for HIF1 transcriptional activity in corresponding samples. One possible explanation for this discrepancy is that a prolonged hypoxic insult normally results in a HIF1-dependent up-regulation of prolyl-4-hydroxylases, serving as a negative feedback mechanism to limit HIF1α protein up-regulation (72, 73). In SH/TG2 and SH/C277S-TG2 cells, the observed inhibition of HIF1 transcriptional activity would prevent this negative feedback mechanism from occurring, resulting in increased HIF1α protein levels compared with SH/vector cells. It should be noted that HIF1 binding under normoxic conditions was increased in SH/TG2 cells but not in SH/C277S-TG2 cells. This increase likely is reflective of our consistent observation that, under normoxic conditions, nuclear HIF1β protein content was greater in SH/TG2 cells compared with both SH/vector and SH/C277S-TG2 cells, although the cause of this increase is not currently known.

The results from this study have narrowed the potential window for TG2 involvement along the HIF1 hypoxic response pathway to a point downstream of HIF1 binding to the HRE but upstream of HRE-dependent gene expression. It is, therefore, most likely that TG2 negatively regulates transcriptional coactivators or transcriptional machinery in a manner that is specific to HIF1. HRE-bound HIF1 recruits the coactivator paralogs CBP and p300 via a direct interaction between the cysteine/histidine-rich 1 (CH1) domain of CBP/p300 and the C-terminal transactivation (CAD) domain within both HIF1α and HIF1β (74,75,76,77,78,79). Carrero et al. (80) also found that the transcriptional coactivators SRC-1 and TIF-2 bind HIF1α and synergize with CBP to increase HIF transcriptional activity in response to hypoxia. Histone deacetylases (HDACs) serve further to positively regulate the HIF1/p300 transcription factor complex (reviewed in ref. 81). HDAC7 was shown to be the only class II HDAC that translocates to the nucleus with HIF1α under hypoxia, where it complexes with HIF1α and p300 and increases HIF1 transcriptional activity (82). Interestingly, the class I and II HDAC inhibitor trichostatin A (TSA) was shown to inhibit HIF1 transcriptional activity in a manner that does not affect HIF1α protein levels, nuclear translocation, or binding to the HRE (83, 84). This finding has recently been attributed to TSA inhibition of HDAC deacetylation of p300, which normally serves to directly increase HIF1/p300 transcriptional activity (83). Because TG2 appears to inhibit HIF1 activity in a manner that is very similar to HDAC inhibition, it is likely that TG2 binding to HIF1β either physically interferes with recruitment CBP/p300 or an HDAC to HIF1 at the HRE, or inhibits the function of one of these coactivators within the HIF1 transcription factor complex at the HRE.

In summary, TG2 is up-regulated during OGD and increased expression leads to neuronal protection. We have demonstrated for the first time that TG2 physically interacts with HIF1β and negatively regulates HIF1 transcriptional activity at the HRE in response to hypoxia, with a selective attenuation of the expression of the proapoptotic gene Bnip3. This novel protein interaction does not affect the nuclear protein content of HIF1α or HIF1β, their physical interaction within the nucleus to form the HIF1 transcription factor, or the binding of nuclear HIF1 to the HRE under hypoxic conditions and thus likely affects the function of HIF1-specific transcriptional machinery. Because TG2 significantly alters the HIF1 hypoxic response pathway and is up-regulated in the brain following ischemia and stroke, it likely plays an important role in regulating the neuropathology associated with this condition.