|Home | About | Journals | Submit | Contact Us | Français|
Thioredoxin-interacting protein (TXNIP) and poly-ADP-ribose polymerase 1 (PARP1) are both regulated by changes in cellular reduction-oxidation (redox) state and localize to the nucleus basally in human umbilical vein endothelial cells (HUVEC). Previously we showed a novel mechanism for PARP1 inhibition–mediated HUVEC survival through activation of vascular endothelial growth factor receptor 2 (VEGFR2) signaling in response to stress-induced apoptosis. In addition, we showed TXNIP translocation to the plasma membrane (PM) and activation of VEGFR2 in response to physiological stimuli. Because TXNIP is an α-arrestin that regulates VEGFR2 signaling, we hypothesized that PARP1 regulates TXNIP localization and function that might affect HUVEC stress-induced apoptosis.
HUVEC treated with 10 µmol/L PARP1 inhibitor (PJ34) were protected from TNF (10 ng/mL) or H2O2 (300 µmol/L) mediated cell death. HUVEC transfected with TXNIP siRNA lost the protective effect of PARP1 inhibition, suggesting a protective role for TXNIP. Using immunofluorescence, cell fractionation analysis, and plasma membrane sheet assay, TXNIP was shown to translocate to the plasma membrane after PARP1 inhibition. TXNIP translocation was associated with activation of VEGFR2 signaling. Functionally, TXNIP-PARP1 interaction was decreased on PJ34 treatment, suggesting PARP1 as a novel regulator of TXNIP localization and function.
These findings demonstrate a novel regulatory mechanism of TXNIP by PARP1 to mediate activation of plasma membrane signaling and cell survival.
Poly-ADP-ribose polymerase 1 (PARP1) is a nuclear protein activated by DNA single-strand breaks associated with oxidative stress.1 PARP1 catalyzes the transfer of ADP-ribose moieties to proteins as part of the DNA repair machinery and also as a mechanism to regulate protein activity, such as p53 and p65.2–4 Recently, PARP1-inhibition has been shown to be beneficial in the attenuation of several pathologies, including post–myocardial infarction remodeling,5 ischemia-reperfusion injury,6 diabetic retinopathy,7 septic shock,8 diabetes,7,9 and atherosclerosis.10,11
PARP1 is activated by increases in oxidative stress due to the generation of excessive reactive oxygen species (ROS) and decreased cellular antioxidant activity, such as thioredoxin (TRX) catalytic activity. An increase in ROS is correlated with the development of atherosclerosis due to the upregulation of proinflammatory and proapoptotic processes, leading to endothelial cell (EC) dysfunction.12,13 EC inflammation and dysfunction are characterized by increased expression of adhesion molecules, recruitment of inflammatory cells, uncoupling of endothelial nitric oxide synthase (eNOS), reduced nitric oxide availability, and vasospasm.12
Thioredoxin-interacting protein (TXNIP), like PARP1, is a nuclear protein that is highly regulated by changes in metabolism including oxidative stress.14 The best-studied function of TXNIP demonstrates it as the endogenous inhibitor of TRX to promote EC inflammation and death.15,16 Inhibition of TRX by TXNIP results in 3 main effects: reduced TRX enzymatic activity,17 release and activation of apoptosis signal-regulating kinase 1 (ASK1) pathway,18 and altered mitochondrial energy production.19 Other important cellular functions of TXNIP were shown using TXNIP knock-out and the HcB19 strain mice, such as metabolic regulator through alterations in insulin secretion from pancreatic β-cells,20 β-cell apoptosis,21 proapoptotic mitochondrial signaling,22 and the switch from gluconeogenesis to glyceroneogenesis and lipogenesis.23 TXNIP is also a member of the α-arrestin protein family, which participates in plasma membrane (PM) signaling by formation of signaling complexes.24 Based on our recent TXNIP structure-function analyses,25 we propose that TXNIP acts as a scaffold protein to promote protein complex formation and regulate cellular signaling. This novel activity is in addition to its function as a TRX inhibitor, metabolic regulator and cell growth repressor.
Recently we showed that PARP1 inhibition protected EC from oxidative stress-induced apoptosis through phosphorylation and activation of the vascular endothelial growth factor receptor 2 (VEGFR2).26 In addition, we showed that in response to short stimulation by physiological concentrations of hydrogen peroxide (H2O2) or tumor necrosis factor (TNF), TXNIP translocated from the nucleus to the PM to associate and activate VEGFR2 signaling.27 Based on our previous findings and the fact that TXNIP and PARP1 colocalize to the nucleus and are both regulated by changes in reduction-oxidation (redox) signaling, we hypothesized that PARP1 inhibition may regulate TXNIP subcellular localization to protect EC from stress-induced apoptosis.
Antibodies and reagents were purchased as follows: actin and VE-cadherin antibodies (Santa Cruz); caspase-3, cleaved caspase-3, and phospho-VEGFR2 (Y1175) (Cell Signaling); TXNIP (Invitrogen); phosphotyrosine (4G10) (Millipore); PARP1 (BD Pharmingen); green fluorescent protein (GFP) (Clontech); PJ34, a PARP1 inhibitor, (Sigma); TNF (Roche); and H2O2 (Mallinckrodt Chemicals).
Human umbilical vein endothelial cells (HUVEC) were isolated from human umbilical veins and seeded onto gelatin-coated dishes maintained in Medium 200 (Cascade Biologics, Portland, OR) with low serum growth supplement and 5% FBS as previously described.28 Cells were used at passages 2–4. Fetal bovine aortic endothelial cells (FBAEC) were purchased from Clontech and cultured in Medium 199 supplemented with 10% FETALCLONE III (Thermo Scientific, Logan UT), basal MEM vitamins (Invitrogen, Carlsbad, CA), and amino acids (Invitrogen, Carlsbad, CA). Cells from passages 2–6 were used for experiments. HUVEC/FBAEC were transiently transfected with GFP tagged TXNIP plasmid or siRNA (TXNIP targeted or scrambled as control). Transfection was performed using Opti-MEM I Reduced Serum Media (Invitrogen) and Lipofectamine 2000. Experiments were performed 24 hours after transfection.
HUVEC apoptosis was measured, after PJ34 (10 µmol/L), TNF (10 ng/mL), and cyclohexamide (10 µg/mL), or H2O2 (300 µmol/L) treatment, using nuclear dye analysis. Cells were fixed with paraformaldehyde, stained with DAPI (Hoechst 33342), and examined by fluorescent microscopy. Cells exhibiting nuclear fragmentation and/or condensation were counted in 5 fields of 3 independent experiments; percent apoptosis is represented relative to total cells counted.
Cells were washed twice in ice-cold PBS and harvested in lysis buffer (20 mmol/L Tris pH7.5, 150 mmol/L NaCl, 1 mmol/L EDTA, 1 µmol/L EGTA, 1% Triton X-100, 2.5 µmol/L sodium pyrophosphate, 1 µmol/L β-glycerolphosphate, 100 mmol/L NaVO4, 1 mol/L NaF, and protease inhibitor cocktail).26 Immune complex samples or protein samples from total cell lysates (TCL) were separated by SDS-PAGE, transferred to nitrocellulose, and incubated with appropriate primary antibodies (overnight; 4°C). After washing and incubation with secondary antibodies (LiCor Biosciences, Lincoln, NE), immunoreactive proteins were visualized with the Odyssey LiCor Infrared Imaging System. Densitometry of blots was performed using Image J software (version 1.36b, National Institutes of Health).
Cells were washed twice in ice-cold PBS and harvested in cell fractionation lysis buffer (250 mmol/L sucrose, 20 mmol/L HEPES pH 7.4, 10 mmol/L KCl, 1.5 mmol/L MgCl2, 1 mmol/L EDTA, 1 mmol/L EGTA, 1 mmol/L DTT, and protease inhibitor cocktail). Briefly, cell lysates were passed through a 25-gauge needle 10 times, followed by centrifugation at 10 000 RPM for 10 minutes to eliminate nuclei and mitochondria fractions. After ultracentrifugation (40 000 RPM; 1 hour), PM fractions were prepared from the cytosolic supernatant of previous step, washed with lysis buffer, and were used for Western Blot analysis after measurement of protein concentrations.
Preparation of PM sheets and immunofluorescence was performed as described previously.29 Images were obtained with an Olympus BX51 fluorescent microscope using Spot software (version 3.5.9 for MacOS, Diagnostic Instruments, Inc). Fluorescence intensity of immunostained PM sheets was quantified with Image J software (version 1.36b, National Institutes of Health).
Group differences were analyzed using the standard Student t test. All values are expressed as mean±SEM. P<0.05 was considered statistically significant.
To determine the role of TXNIP in EC protection from stress-induced apoptosis, we performed cell death assays using TNF as a physiological agonist. After depletion of TXNIP by siRNA transfection, HUVEC were treated with PJ34 and/or TNF and then stained with DAPI to identify apoptotic cells. In control cells, only a small percentage of cells were apoptotic independent of TXNIP depletion (Figure 1A and 1E; 4.9±1.5% and 4.8±0.5%). Inhibition of PARP-1 with PJ34 had no significant effect on the percentage of apoptotic cells (Figure 1B and 1F; 2.8±1.4% and 2.9±0.7%), whereas TNF induced an increase in the percentage of apoptotic cells (Figure 1C and 1G; 34.4±1.2% and 30.3±1.0%), regardless of TXNIP depletion. Significantly, HUVEC transfected with scrambled siRNA exhibited less TNF-induced apoptosis when treated with TNF and PJ34 (Figure 1D; 15.0±1.9%). In contrast, this protective effect was significantly reduced when TXNIP was depleted by siRNA (Figure 1H; 37.2±0.9%). These data demonstrate a protective role for PARP-1 inhibition that requires TXNIP in HUVEC.
To confirm a protective role for TXNIP in reducing stress-induced apoptosis, we performed cell death assays after stimulation with high doses of H2O2 (Figure 1J through 1R). Similar to TNF stimulation, PARP1 inhibition protected HUVEC from H2O2-induced cell apoptosis in scrambled siRNA transfected cells (Figure 1M; 13.1±1.5%) but not in TXNIP siRNA-transfected cells (Figure 1Q; 81.4±1.5%).
To confirm the role for TXNIP in HUVEC protection from TNF or H2O2-induced apoptosis, we measured caspase-3 cleavage. Caspase-3 is a 35-kDa protein that plays a major role in the initiation and propagation of cell apoptosis. The 35-kDa form of caspase-3 is inactive and must be cleaved into 17/19 kDa fragments to promote apoptosis. Vehicle-treated HUVEC had a low level of activated caspase-3 in both scrambled and TXNIP siRNA-transfected cells (Figure IA in the online-only Data Supplement). Addition of PJ34 had no effect on the levels of activated caspase-3 that were comparable to vehicle treated cells, independent of TXNIP depletion. As expected, prolonged stimulation with TNF induced an increase in caspase-3 cleavage in both scrambled and TXNIP siRNA transfected cells. In contrast, when HUVEC were treated with both TNF and PJ34 caspase-3, cleavage increased 2-fold in cells transfected with TXNIP siRNA but not in cells transfected with scrambled siRNA. These data support our hypothesis that TXNIP exerts an important role in protection of EC from stress-induced apoptosis. To extend these data, caspase-3 cleavage was also measured in HUVEC treated with H2O2 and PJ34. PARP1 inhibition protected HUVEC from H2O2-induced caspase-3 cleavage in scrambled siRNA-transfected cells but not in TXNIP siRNA-transfected cells (Figure IB in the online-only Data Supplement).
The previous data suggest a role for TXNIP in PARP1 inhibition protection of EC from stress-induced apoptosis. We hypothesized that PARP1 regulates TXNIP subcellular localization. Therefore, we tested the hypothesis that TXNIP-PARP1 colocalize to the nucleus in EC, by using the immunofluorescence approach to evaluate TXNIP and PARP1 subcellular localization in untreated cells. Immunofluorescence studies revealed that PARP1 and TXNIP were predominantly localized in the nucleus of untreated cells (Figure 2A, 2B, and 2C). Furthermore, TXNIP and PARP1 colocalize as shown by the white arrowheads in the merged image (Figure 2D). These data suggest an association between PARP1 and TXNIP.
Next, we hypothesized that TXNIP, which belongs to the α-arrestin protein family, translocates from the nucleus to regulate PM-specific signaling. Therefore, we explored TXNIP subcellular localization after PARP1 inhibition or depletion using PARP1 siRNA. To optimize and demonstrate TXNIP translocation from the nucleus to the PM, we performed PJ34 dose- and time-dependent response experiments (Figures II and III in the online-only Data Supplement). In vehicle HUVEC, TXNIP was mostly localized to the nucleus, and on increasing concentrations and time duration of PJ34 treatment, TXNIP translocation to the PM was observed. Peak translocation was observed at PJ34 10 µmol/L after 30 minutes with no effect on TXNIP expression (Figure IIIF in the online-only Data Supplement). After treatment with PJ34 or transfection with PARP1 siRNA, an increase of PM associated TXNIP was observed, which was identified by colocalization with the PM specific marker VE-cadherin (Figure 3D–3F and 3G–3I; 2.4±0.2 and 2.3±0.1). No additional effect on TXNIP translocation to the PM was observed when PARP1 siRNA-transfected cells were treated with PJ34 (Figure 3J–3L; 2.4±0.2). These data suggest that TXNIP translocation from the nucleus to the PM is exclusively related to PARP1 inhibition and not due to inhibition of other unknown targets by PJ34. To confirm TXNIP translocation to the PM, we performed cell fractionation analysis. In agreement with our previous data, very little TXNIP was associated with the PM under basal conditions. After treatment with PJ34, PARP1 depletion using specific siRNA or both PJ34 and PARP1 siRNA, a 2-fold increase in PM associated TXNIP was observed as demonstrated by Western blot analysis of the PM fraction (Figure 3N and 3O).
To determine the role that TXNIP plays at the PM, we performed a sheet assay, which has the advantage of better signal-to-background ratio due to removal of nuclear and cytosolic fractions. Briefly, HUVEC were swollen in hypotonic buffer and sonicated to disrupt the cell structure, resulting in thin PM sheets attached to the dish. PM sheets were washed, fixed, and immunostained to identify PM associated TXNIP and tyrosine-phosphorylated PM proteins. Previously, we reported TXNIP translocation to the PM in response to TNF or H2O2, which resulted in phosphorylation and activation of PM proteins. Consistent with those findings, low levels of TXNIP and phosphotyrosine were observed at the PM in vehicle-treated cells (Figure 4A and 4C). Cells treated with PJ34 showed increased PM-associated TXNIP and tyrosine phosphorylation (Figure 4B, 4D, and 4E; 8.4±0.8-fold and 18.7±1.6-fold increases, respectively). These data suggest a mechanistic link between PARP1 inhibition, TXNIP translocation, activation of PM signaling, and protection of EC from stress- induced apoptosis.
We next explored activation of VEGFR2, which is an important tyrosine kinase receptor that regulates EC prosurvival and anti-inflammatory processes.26 In scrambled siRNA-transfected cells, PJ34 increased tyrosine 1175 VEGFR2 phosphorylation (Figure 5; 2.4±0.2-fold). In contrast, in TXNIP siRNA-transfected cells, stimulation of VEGFR2 tyrosine phosphorylation was inhibited. These effects were observed with no changes in expression of actin and VEGFR2, as indicated by Western blotting (Figure 5A).
Our recently published structure-function analysis of TXNIP identified SH3, PPxY, and ITIM motifs that could mediate protein-protein interactions.25 PARP1 is a nuclear protein with 2 main activities: (1) the transfer of ADP-ribose moieties onto target proteins to regulate localization and function and (2) acts as a scaffold protein through direct or indirect interactions to localize target proteins to the nucleus.4 Because PARP1 and TXNIP colocalize to the nucleus of untreated cells (Figure 2), we hypothesized that TXNIP may interact with PARP1 under basal conditions to keep it in the nucleus. Furthermore, we hypothesized that TXNIP undergoes poly-ADP ribosylation to regulate its subcellular localization. Immunoprecipitation of TXNIP and Western blotting of ADP-ribose revealed that TXNIP was poly-ADP-ribosylated. However, no changes in the levels of poly-ADP ribosylation were observed after PARP1 inhibition (data not shown). Therefore, we hypothesized that changes in PARP1 poly-ADP ribosylation, rather than TXNIP, explained changes in TXNIP subcellular localization. We used PJ34 to inhibit PARP1 activity and then coimmunoprecipitation to study the correlation between PARP1 poly-ADP ribosylation and TXNIP-PARP1 interaction. As shown in Figure 6, after PARP1 inhibition with PJ34, a 60% decrease in PARP1-TXNIP interaction was observed, as shown by either immunoprecipitation of TXNIP (Figure 6A and 6B) or PARP1 (Figure 6C and 6D). Importantly, a significant decrease in PARP1 poly-ADP ribosylation correlated with decreased PARP1-TXNIP interaction (Figure 6E and 6F).
The major finding of this study is the discovery of a novel EC survival pathway involving TXNIP-PARP1–mediated activation of VEGFR2. Specifically, we show that under basal conditions, TXNIP and PARP1 bind to each other and are located in the nucleus (Figure 6). In response to inflammatory or oxidative stress stimuli (TNF or H2O2, respectively), PARP1 activity is stimulated and poly-ADP ribosylation of PARP1 maintains TXNIP in the nucleus. In contrast, inhibition of PARP1 with PJ34 disrupts PARP1-TXNIP binding, enabling translocation of TXNIP to the PM. At the PM, TXNIP promotes VEGFR2 activation and antiapoptotic signaling. These findings describe a novel signaling pathway for the VEGFR2 that involves nuclear to PM shuttling of TXNIP.
The present study provides 2 new findings important for understanding PARP1 biology. First, TXNIP is required for the antiapoptotic effect of PARP1 inhibition in EC through activation of the VEGFR2. PARP1 was originally shown to play a critical role in the cellular response to oxidative stress-induced DNA breaks.30 Although PARP1 protects cells from genotoxic stress, it appears that in chronic situations, PARP1 may contribute to the pathogenesis of several diseases. Specifically, PARP1 inhibition was shown to be beneficial in inflammatory diseases,6,8 diabetes,7,9 post–myocardial infarction remodeling,5 ischemia-reperfusion injury,6 septic shock,8 and atherosclerosis.10,11 PARP1 inhibition has been proposed to be beneficial by 2 mechanisms. When PARP1 is highly activated (eg, high dose of H2O2), it consumes large amounts of cellular NADPH and, by reducing cellular energy bioavailability, promotes cell death. Furthermore, we and others have shown that PARP1 inhibition activated prosurvival signals: Akt in hepatic carcinoma cells and VEGFR2 in EC.26,31
The second finding pertinent to PARP1 biology is that PARP1 binding to TXNIP and PARP1 catalytic activity regulate TXNIP subcellular localization. These results are in agreement with previous publications that demonstrated that PARP1 regulated p53 and NFkB function by direct protein-protein interaction and/or poly-ADP ribosylation.2,4 We report that PARP1 is bound to TXNIP in the nucleus under basal conditions. On PARP1 depoly-ADP ribosylation, TXNIP is exported from the nucleus to the PM. Thus, PARP1 has a new role in the nuclear retention of TXNIP.
The present study provides further support for the hypothesis that TXNIP has important functions as an α-arrestin in addition to its role as the endogenous inhibitor of TRX.15,16 Accumulating data demonstrate both TRX-dependent and TRX-independent functions of TXNIP that involve the regulation of metabolism,23 cell growth,32,33 and mitochondrial function.22,34 We show that TXNIP, as an α-arrestin, promotes PM signaling that is not linked to its function as the inhibitor of TRX. Specifically, we show that as a result of PARP1 inhibition, TXNIP stimulates transactivation of the VEGFR2, which is a key receptor for EC prosurvival signaling through regulation of the activity of proteins such as eNOS and Akt.26,35 It appears likely that TXNIP mediates transactivation of the VEGFR2 in response to stress, based on several recent studies. First, TXNIP acts as a scaffold to mediate TRX translocation to VEGFR227 and potentially regulate redox dependent activation, such as the recruitment of tyrosine phosphatases. Second, TXNIP, through its ITIM, PPxY, and SH3 domains, may facilitate a direct interaction with modulators of PM tyrosine kinase receptors, such as Grb2 and SHP2.25 The exact mechanism of TXNIP-mediated VEGFR2 activation is an important area for future investigation. Finally, the present study points to an important difference between α- and β-arrestins. The function of the β-arrestin family of proteins is well established and mediates primarily PM to cytosol and/ or nucleus communication.24 We report the novel finding that TXNIP, an α-arrestin protein, mediates nuclear to PM signaling. This suggests evolution of α-arrestins in mammals beyond the role of yeast arrestin-related trafficking adaptors in cargo transport from PM to endocytotic vesicles.36
Finally, the present data strongly support a critical role for TXNIP in the endothelial stress response. Specifically, we propose that TXNIP orchestrates multiple cellular response mechanisms that enable EC to survive acute stress long enough to mount an inflammatory response. For example, a traumatic injury that disrupts blood flow would cause disturbed flow patterns that induce TXNIP. Induction of TXNIP leads to inhibition of TRX, resulting in excessive ROS, protein oxidation, and the release of the proinflammatory protein ASK1. As a result, ASK1 activates nuclear signaling that leads to expression of VCAM-1.16,37 A similar stress response also occurs in EC exposed to inflammatory cytokines and hyperglycemia.38,39 Furthermore, TXNIP is involved in inflammasome activation and IL1β maturation, recruiting additional inflammatory cells.40 This central role for TXNIP in the response to injury and inflammation suggest that drugs that regulate its expression may have broad therapeutic applications.
We thank Dr Cameron World for helpful advice.
Source of Funding
This work was supported by National Institutes of Health grant HL 106158 (to B.B.).