Poly(ADP-ribose) polymerase-1 and its cleavage products differentially modulate cellular protection through NF-kB-dependent signaling

Abstract

Poly(ADP-ribose) polymerase-1 (PARP-1) and its cleavage products regulate cell viability and NF-kB activity when expressed in neurons. PARP-1 cleavage generates a 24kDa (PARP-1₂₄) and an 89kDa fragment (PARP-1₈₉). Compared to WT (PARP-1_WT), the expression of an uncleavable PARP-1 (PARP-1_UNCL) or of PARP-1₂₄ conferred protection from oxygen/glucose deprivation (OGD) or OGD/restoration of oxygen and glucose (ROG) damage in vitro, whereas expression of PARP-1₈₉ was cytotoxic. Viability experiments were performed in SH-SY5Y, a human neuroblastoma cell line, as well as in rat primary cortical neurons. Following OGD, the higher viability in the presence of PARP-1_UNCL or PARP-1₂₄ was not accompanied with decreased formation of poly(ADP-riboses) or higher NAD levels. PARP-1 is a known cofactor for NF-kB, hence we investigated whether PARP-1 cleavage influences the inflammatory response. All PARP-1 constructs mimicked PARP-1_WT in regards to induction of NF-kB translocation into the nucleus and its increased activation during ischemic challenge. However, expression of PARP-1₈₉ construct induced significantly higher NF-kB activity than PARP-1_WT; and the same was true for NF-kB-dependent iNOS promoter binding activity. At a protein level, PARP-1_UNCL and PARP-1₂₄ decreased iNOS (and lower levels of iNOS transcript) and COX-2, and increased Bcl-xL. The increased levels of NF-kB and iNOS transcriptional activities, seen with cytotoxic PARP-1₈₉, were accompanied by higher protein expression of COX-2 and iNOS (and higher levels of iNOS transcript) and lower protein expression of Bcl-xL. Taken together, these findings suggest that PARP-1 cleavage products may regulate cellular viability and inflammatory responses in opposing ways during in vitro models of “ischemia”.

1. Introduction

Poly (ADP-ribose) polymerase 1 (PARP-1), also known as ARTD1 (diphtheria toxin-like-ADP-ribosyltransferase) [1, 2], plays a role in the inflammatory pathogenesis of many central nervous system disorders including stroke [3-6]. A change in PARP-1 activity has been reported in ischemia-reperfusion damage involving complex mechanisms that are not completely understood, including regulation of DNA repair, cell death, transcription and inflammation [7]. Here we investigate a possible involvement of PARP-1, particularly its cleavage products, in the regulation of cell viability using in vitro models of ischemia (known as oxygen/glucose deprivation, OGD) or OGD/restoration of oxygen and glucose (ROG).

PARP-1 is known to have many functions and among them, two are well known, namely the detection and repair of DNA single strand breaks through poly(ADP-ribosyl)ation (PARylation) [8, 9], and the regulation of transcription by coupling with many transcription factors such as NF-kB [10, 11]. PARylation requires PARP-1 C-terminal domain catalytic activity, consuming NAD, and can influence NF-kB activity through direct addition of poly(ADP-riboses) (PARs) on to NF-kB itself [12, 13]. PARP-1/ARTD1 contains an N-terminal DNA-binding domain (DBD), an automodification region, and a C-terminal catalytic domain [14]. The caspase cleavage site DEVD₂₁₄ is situated in the DBD within the nuclear localization signal (NLS) [15]. PARP-1 is known to be cleaved at the DEVD₂₁₄ site by activated caspases 3 and 7 [16] leading to the formation of a 24 kDa and an 89 kDa fragment (Fig. 1A). Although the consequences of PARP-1 cleavage as well as of the effects of its two cleavage products remain unclear, depending on the intensity and type of stimuli that induce the cleavage itself, two main results have been reported, namely (1) the reduction of PARylation during DNA repair processes and (2) the modification of transcription activity of a PARP-1 bound protein complex that includes NF-kB. Generally, following massive DNA damage, the appearance of 24 kDa and 89 kDa PARP-1 fragments is widely accepted as a hallmark of apoptosis [16]. However, the suggestion that PARP-1 cleavage would reduce PARylation and save the cells from excessive energy (NAD⁺, ATP) depletion still remains [17, 18]. An alternative role for PARP-1 cleavage acting as a regulator of NF-kB transactivation is being increasingly investigated [10, 19-23].

Figure 1. Characterization of caspase-resistant PARP-1_UNCL.

(A) PARP-1_WT (top) is cleaved by activated caspases at the DEVD₂₁₄ site. Upon cleavage the two fragments, 24 and 89 kDa, are produced (middle). PARP-1_UNCL cannot be cleaved (bottom). DNA binding domain (DBD), automodification region containing the BRCT, and the catalytic domain are highlighted yellow, green and pink, respectively. ZF=Zinc Finger domain I, II and III, NLS=Nuclear Localization Sequence, BRCT=Breast Cancer Suppressor Protein. (B) Western blots of total cell exctracts showing tet-dependent over expression of PARP-1_WT or PARP-1_UNCL (left) and its quantification by densitometry (right), n=9. (C) Western blots showing DEVN₂₁₄ resistance to caspase-induced cleavage (left) and corresponding density (right). n=3. Etop=Etoposide. For B and C, data are expressed as mean ± SD,*p<0.02.

NF-kB is a central cellular module that regulates, among other things, the inflammatory response [24]. The major subtype of NF-kB consists of a hetero-dimer of subunits p50 and p65, which is kept inactive by the inhibitor of NF-kB protein (IKB) in the cytoplasm. Upon stimulation, IKB is ubiquitinated and degraded allowing NF-kB translocation to the nucleus where it binds to DNA and activates transcription. It has been shown in animal models that ischemia activates NF-kB leading to the inflammatory response, and that its inactivation is generally neuroprotective [25, 26]. Reports also indicate that the activation of certain NF-kB-dependent pro and anti-inflammatory/protective genes is regulated by complex mechanisms that are still poorly understood [27].

Previous observations suggest that PARP-1 and NF-kB functions are interconnected in ischemic conditions [28-31]. One more layer of complexity is that not only is PARP-1 a cofactor for NF-kB, but the cleavage of PARP-1 further influences NF-kB activity. A few reports have suggested that PARP-1 cleavage may participate in transactivation of NF-kB in a pro inflammatory fashion [11, 23, 32-34]. It is of particular interest that an uncleavable form of PARP-1 (PARP-1_UNCL) (Fig 1A) in transgenic mice conferred protection from endotoxic shock, and intestinal and renal ischemia/reperfusion damage [23].

Therefore, the aim of this study was to shed light on the functional role of PARP-1 cleavage at the DEVD site and to investigate the role of each fragment produced in cell viability and the NF-kB transcriptional response during ischemic stress. Human neuroblastoma cells (SH-SY5Y) were used in models of “ischemia” (Oxygen/Glucose Deprivation - OGD) to study four variants of PARP-1 (control PARP-1_WT and three experimental variants of PARP-1, PARP-1_UNCL, PARP-1₂₄ and PARP-1₈₉) on (1) cell survival and (2) the transcriptional regulation of NF-kB-dependent proteins. Cell survival studies were extended in a complimentary model (OGD/ROG) using primary rat cortical neurons.

We found that the uncleavable PARP-1_UNCL as well as PARP-1₂₄ were cytoprotective, and PARP-1₈₉ was toxic. This study also suggests that PARP-1 cleavage or its fragments do not modify cellular levels of PARs or NAD levels, at least in our OGD model. However, NF-kB activity and the expression of certain downstream effectors were modulated by the different forms of PARP-1. These findings suggest that PARP-1 cleavage may play a role in the regulation of cell viability and NF-kB in response to OGD or OGD/ROG.

2. Materials & Methods

2.1 Cell Culture and transfection: SH-SY5Y and primary rat cortical neurons

Human neuroblastoma cells, SH-SY5Y (American Type Culture Collection, Manassas, VA, USA; Cat# CRL-2266), were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 mg/ml streptomycin (DMEM complete) in 5% CO₂ at 37°C. Tetracycline (tet)-inducible stable transfectants were generated as previously described [35]. In particular, the constructs used were PARP-1_WT, PARP-1 _UNCL, PARP-1₂₄ and PARP-1₈₉. See section 2.2 “Cloning” for more details. Lipofectamine RNAi max (Invitrogen) was used to transfect the cells with siRNA-PARP-1 (Target Sequence: 5′-ACGGTGATCGGTAGCAACAAA-3′, FlexiTube, Qiagen) at a concentration of 25 nM. In experiments with siRNA, tet (1μg/ml) was added at the same time as the siRNA in order to minimize the period of PARP-1 deprivation. All the experiments were performed using siRNA-PARP-1 with the exception of viability experiments for cells overexpressing PARP-1₂₄ and PARP-1₈₉. siRNA-PARP-1 was not used for viability experiments conducted on primary cortical neurons from rats (where an effect was seen without the need of reducing the background from endogenous PARP-1). Control experiments were conducted using scramble siRNA (AllStars Negative Control siRNA, Qiagen) as negative control.

Primary cortical neurons were isolated from Sprague-Dawley rats at 2 days of age (P2) and cultured in Neurobasal Medium-A (NB-A) supplemented with B27, as previously reported [35] and plated at a density of 3.75X10⁵ cells/ml. Experiments using cells isolated from rat brains were conducted in accordance with the Institutional Animal Care and Use Committee at the National Institute of Health guidelines.

Constructs for PARP-1_WT, PARP-1_UNCL and PARP-1₂₄ were cloned into a custom made viral vector for Adeno-Associated Viruses (AAV). See section 2.2 “Cloning” for more details. Cortical neurons were transduced 3 days after isolation. At day 4 the medium was replaced and at day 6, after measuring the transfection efficiency, the cells were subjected to OGD for 6h followed by ROG (15 h).

2.2 Cloning

Wild type PARP-1 plasmid was purchased from the Harvard Institute of Proteomics (plasmid ID HsCD00040600 http://plasmid.med.harvard.edu/PLASMID/Home.jsp) and the sequence was confirmed.

2.2.1 Cloning PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ and PARP-1₈₉ into pcDNA4/TO vector

Constructs for wild type PARP-1_WT, PARP-1₂₄ and PARP-1₈₉ were amplified with appropriate restriction sites by PCR using the PARP-1_WT plasmid as a template. The primer sets used were as follows:

AflII(F)-5′-AGCTCTTAAGATGGCGGAGTCTTCGGAT-3′ and XhoI(R)-5′-ATCACTCGAGTTACCACAGGGAGGTCTT-3′ for PARP-1_WT; AflII(F)-5′-AGCTCTTAAGATGGATGGAGTGGATGAAG-3′ and XhoI(R)- 5′-ATCACTCGAGTTACCACAGGGAGGTCTT-3′ for PARP-1₈₉; EcoRI(F)-5′-AGCTGAATTCATGGCGGAGTCTTCGGAT-3′ and XbaI(R)-5′-CATCTAGATTATCCATCCACCTCATCGC-3′ for PARP-1₂₄.

Each PCR amplified DNA was cloned into pcDNA4/TO vector.

PARP-1 siRNA resistant mutants WT and PARP-1_UNCL DEVN₂₁₄ (GAT→AAT) were generated by using the QuickChange II or Quick Change Multi Site-directed Mutagenesis Kits (Stratagene, La Jolla, CA; Cat# 200523) as previously described [23]. Primers for siRNA resistant WT and PARP-1_UNCL were as follows: PARP-1_UNCL (F)-5′-GAAAAGGCGATGAGGTGAATG-3′ and PARP-1_UNCL (R)-5′-CACTTCATCCACTCCATTCACCTCATCGCCTTTTC-3′.

2.2.2 Cloning PARP-1_WT, PARP-1_UNCL and PARP-1₂₄ into a viral vector suitable for packaging an Adeno-Associated Virus (AAV)

PARP-1_WT, PARP-1_UNCL and PARP-1₂₄ were cloned into a viral vector using appropriate restriction sites, after amplification by PCR. Primers used for PARP-1_WT and PARP-1_UNCL were MefI(F)-5′-GTACCAATTGATGGCGGAGTCTTCGGATAAG-3′ and SalI(R)-5′-CGACGTGTCGACTTACCACAGGGAGGTCTTAA-3′. Primers used for PARP-1₂₄ were EcoRI(F)-5′-AGCTGAATTCATGGCGGAGTCTTCGGAT-3′ and SalI(R)-5′-GTACCCGCGGTCCATCCACCTCATCGCCTTTTC-3′.

Each PCR amplified DNA was cloned into the pFBGR plasmid (a pFastBac™ variant obtained from Dr. Robert Kotin, NIH\NHLBI), a plasmid containing two AAV inverted terminal repeats (ITRs) flanking the CMV promoter and EGFP reporter. The EGFP is separated (not fused) from PARP-1 gene by a T2A region that allows measurement of transfection efficiency in addition to the overexpression of the gene of interest. Finally, the poly(A) tail was replaced with Woodchuck Hepatitis Virus Posttranscriptional Regulatory Element (WPRE) to enhance transcription. DH10Bac cells (Invitrogen) were transformed with donor AAV plasmids (WT, PARP-1_UNCL, or PARP-1₂₄) and plated on appropriate selection media [36]. Briefly, PARP-1 constructs containing baculovirus were added along with a second AAV-6 packaging baculovirus (generous gift from R. Kotin, NIH) which was chosen because we had found it to be the most efficient in transducing primary cortical neurons in our preliminary experiments among AAV-2, AAV-4, AAV-5, AAV-6, and AAV-8. Titering procedure of the final AAV was done according to previously described methods [36].

2.3 Oxygen and glucose deprivation (OGD), OGD/restoration of oxygen and glucose (ROG) and cell death assessment

OGD for SH-SY5Y or cortical neurons was performed as described previously [35]. The lengths of OGD were 15h for SH-SY5Y cells and 6h for cortical neurons, due to their increased sensitivity to the exposure. SH-SY5Y cells were assayed immediately after OGD to evaluate cell viability. In the cortical neurons, the OGD medium (PBS with 2mM MgCl₂ and 1mM CaCl₂) was replaced at the end of OGD with regular NB-A/B27 (ROG) and viability was assayed 15h later.

After lifting, cell viability of SH-SY5Y was assessed by differential nuclear staining with Propidium Iodide (PI, live-cell impermeant red DNA dye) and Hoechst 33342 (cell permeant blue DNA dye) followed by quantitative flow cytometric analysis (FACSVantage SE flow cytometer) as described previously [37, 38]. Due to their fragile nature, cell viability of cortical neurons was assessed by nuclear staining with PI and Hoechst 33342 performed directly in the well without lifting and the cells were counted manually under the fluorescent microscope, as previously described [35]. Cells were counted as follows: 10 fields were counted per sample. Each field had at least 200 cells. Total of 2000 cells/sample were counted. Experiments were conducted at least in triplicate.

Furthermore, as primary neuronal cultures do not readily lend themselves to prolonged and repeated manipulations, biochemical experiments were performed in the more adaptable neuroblastoma cell lines.

2.4 Western blot analysis

Total cell lysates or cytosol/nuclear fractions from SH-SY5Y cells were used for Western blot analysis (WB). Total cell lysates were prepared as described previously [35], and cytosolic or nuclear fractions were prepared following a previously reported protocol [39]. After protein concentrations were determined (BCA assay, Pierce), all samples were heated with β-mercaptoethanol and dye (bromophenyl blue) for 5 minutes at 95°C. Forty μg of total cell lysate and 10 μg of nuclear extract were used for each SDS-PAGE/WB analysis. The following primary antibodies were used for WB analyses: anti-PARP-1 (Cell Signaling Technology, Danvers, MA; Cat 9546S), anti-PARP-1 (N-20) (Santa Cruz Biotechnology, Santa Cruz, CA; Cat SC1561) to detect the 24 kDa fragment, anti-PARP-1 (C-2-10) (Biomol, Farmingdale, NY; Cat SA250), anti-PARs (Trevigen, Cat 4335-MC-100-AC), anti-p65 (Santa Cruz Biotechnology, Cat sc-372), anti-iNOS (Santa Cruz Biotechnology, Cat SC7221), anti-COX2 (Cayman, Cat 160126), anti-Bax (Cell Signaling Technology, Cat 2772), and anti-Bcl-x_L (Cell Signaling Technology, Cat 2762), anti-lamin (Abcam, Cat 16048-100). Signals were detected using a chemiluminescent substrate (Millipore) followed by digital imaging (Fluor Chem camera, Alpha Innotech, San Leandro, CA). The optical density of each western blot lane was measured using the software AlphaEaseFC 4.0.0. Data are expressed as percent of increase or decrease from control, after normalizing the density of each band by the density of the respective β-actin (total cell extracts) or lamin (nuclear extracts).

2.5 NAD^+/NADH Quantification

Detection of total NAD (tNAD), NAD⁺ and NADH, was performed using the NAD⁺/NADH quantification kit (BioVision, San Francisco, CA). SH-SY5Y cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ were seeded at a density of 7.5X10⁵/well in a 6 well plate. Levels of tNAD, were measured before and at the end of each OGD time period (30 min, 2h, 6h, 6h+ROG 2h, 6h+ROG 4h) in each group of cells. ROG was done to simulate “reperfusion” and restore levels of NAD, the PARP-1 substrate. Data were shown as percent of increase or decrease as compared to untreated cells expressing PARP-1_WT.

2.6 Caspase 3/7 activity assay

PARP-1 siRNA treated SH-SY5Y cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ were transferred in a 96 well plate and subjected to 6h OGD. Before and after OGD caspase 3/7 activity was measured (Apo 3/7 HTS; Cell Technology, Mountain View, CA). Results were expressed as percent of increase by OGD compared to control (without OGD) in each cell type.

2.7 Immunofluorescent Staining of Human SH-SY5Y cells

SH-SY5Y cells were seeded on chambered coverglass (Thermo Fisher Scientific, NY, Cat 155380) using DMEM complete with tet. The following day the medium was replaced with OGD medium and the cells were subjected to OGD for 10 minutes. Immunostaining was carried out using anti-p65 antibody (Santa Cruz Biotechnology, Cat sc-372) as a primary antibody and Alexa Fluor®488 (Molecular Probes) (in green) coupled goat anti-rabbit IgG as a second antibody. Nuclear staining (in blue) was performed using a mounting medium with DAPI (Vector Laboratories Inc, Burlingame, CA). Images were captured with an oil immersion 40x objective on a Zeiss LSM510 microscope using LSM Image Software.

2.8 Luciferase Reporter Assays: NF-kB and iNOS

Neuroblastoma cells expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ were seeded in a 96-well plate 24h before performing transfection for either NF-kB or INOS reporter plasmids. The NF-kB reporter assay was performed using the Cignal NFκB Reporter Assay Kit (SABiosciences, QIAGEN, Cat CCS-013L) following the manufacturer’s instructions.

The iNOS reporter assay was performed using INOS WT (-1485/+31 NF-kB WT) and INOS mutant (-1485/+31 NF-kB mutant) constructs [40] (kind gift from Dr Mark Perrella). The mutant form of the INOS promoter region had been mutated such that NF-kB could not bind to it. NF-kB is, therefore, unable to bind to mutant INOS promoter, serving as a negative control of the NF-kB dependent INOS reporter assay. The INOS WT and mutant promoter regions in pGL2 vector (original constructs from Dr. Perrella) were transferred into a pGL4 vector (Promega) to enhance the firefly signal. These constructs (500 ng/well) were then co-transfected with renilla pRL-TK vector (100 ng/well) for the normalization of transfection efficiency. For both NF-kB and INOS assays, the measurements of firefly and renilla luciferase activities were performed at the end of 6 hours of OGD using Dual-Glo Luciferase Assay System (Promega) and a luminometer (Centro LB 960, Berthold, Germany). Background subtraction and normalization was done for all experiments. NF-kB activity was expressed as percent of increase or decrease from control (PARP-1_WT before OGD).

2.9 RNA analyses by real-time polymerase chain reaction

Total RNA was isolated with the High Pure RNA Isolation kit (Roche Applied Science, Indianapolis, USA). Total RNA concentration was measured with NanoDrop 2000 UV-Vis Spectrophotometer (Thermo Fisher Scientific, DE, USA). 25 ng of RNA was used for real-time PCR analyses using the LightCycler LC480 System (Roche Applied Science, Indianapolis, USA) in a one step reaction using LightCycler 480 RNA Master Hydrolysis Probes. Primer pairs used (Real Time Ready from Roche Applied Science, Indianapolis, USA) and amplification protocol are listed in Supplementary Table S1 and S2. We examined the gene expressions of INOS and COX-2 (NF-kB-dependent genes) which have a kb-site located in their promoter regions. Samples were tested at least in duplicates. Gene specific expression was normalized to an endogenous control (Actin) and expressed in arbitrary units relative to the expression in untreated cells. Relative quantification was done using the standard curve method with PCR efficiency correction (E-method) of the LightCycler 480 Software (Roche). All samples were normalized to Actin expression on the same plate. The relative mRNA level was calculated with the second derivative formula that corrects for PCR efficiency differences between target and reference gene [41].

2.10 Statistics

Data are shown as the mean ± SD. Statistical analyses were done by Student T-test and the number of experiments (n) is indicated in the figure legends. P values of <0.05*, <0.01** and <10⁻⁵*** were considered significant.

3. Results

3.1 Mutation from DEVD₂₁₄ to DEVN₂₁₄ renders PARP-1 resistant to caspase cleavage

PARP-1_UNCL was made by a point mutation (GAT→AAT) in the 214^th codon of PARP-1_WT. As a result, the 214^th amino acid of PARP-1_WT, an aspartate, was changed to an asparagine (D→N). This mutation makes PARP-1 uncleavable by activated caspases 3 and 7 (Fig. 1A). Tet-inducible stable human SH-SY5Y transfectants were generated with PARP-1_WT or PARP-1_UNCL. As expected, PARP-1 protein (WT or UNCL) expression levels in these stable transfectants were significantly higher in the presence of tet (3.03-fold for PARP-1_WT and 2.94-fold for PARP-1_UNCL) compared to PARP-1 endogenous levels (cells grown without tet) (Fig. 1B, left and right panels). To confirm that PARP-1_UNCL would be resistant to caspases, the SH-SY5Y stable transfectants of PARP-1_WT or PARP-1_UNCL were grown with tet for 24h and treated with 0.1 mM etoposide, a topoisomerase II inhibitor and a well known apoptosis-inducer, for 15h. Cells were previously transiently transfected with siRNA-PARP-1. As shown in Figure 1C, etoposide treatment caused PARP-1_WT cleavage and generated an 89kDa fragment (53% of the available full length protein was cleaved) while PARP-1_UNCL was not cleaved as much by the same treatment, leaving the 116kDa band mostly intact (only 22% of was cleaved, representing endogenous PARP-1 cleavage) (n=3, p<0.02). These findings indicate that the mutation at the DEVD site resulted in an uncleavable PARP-1, PARP-1_UNCL.

3.2 Overexpression of PARP-1_UNCL confers protection to SH-SY5Y cells from OGD-induced cell death

Since PARP-1_UNCL knock-in mice were reported to be more resistant to endotoxic shock or renal ischemia/reperfusion compared to WT [23], we tested whether human SH-SY5Y cells over-expressing PARP-1_UNCL were also more resistant to OGD compared to cells over-expressing PARP-1_WT. Unexpectedly, there was no difference in viability between cells over-expressing PARP-1_WT or PARP-1_UNCL after 15h OGD. We suspected that endogenous PARP-1 might counteract any possible effect derived from overexpression of PARP-1_UNCL. To test this, endogenous PARP-1 was first depleted using siRNA-PARP-1. As shown in 2A, endogenous PARP-1 was almost completely depleted as compared to untreated cells as well as to cells treated with scambled siRNA. We generated a new set of tet-inducible stable transfectants which were also siRNA resistant for PARP-1_WT or PARP-1_UNCL. Endogenous PARP-1 of these cells was almost completely depleted by 4 day-treatment of siRNA, and exogenous forms of PARP-1 were confirmed to be siRNA resistant (Fig. 2B). By simultaneous administration of siRNA and tet, exogenous PARP-1_WT or PARP-1_UNCL were exclusively expressed. Scramble siRNA was used for negative control (transfected in control cells). When these cells were challenged with OGD, PARP-1_UNCL-expressing cells were significantly more resistant compared to control cells (viability: 36.79±4.7 control vs 62.32±7.3; mean ± SD; n=8; p<0.01) (Fig. 2C; Fig. 2D bottom right panel). Additional controls were performed (data not shown) to make sure that, per se, the overexpression of PARP-1_WT or PARP-1_UNCL, siRNA or tet did not affect the viability of the cells. The results strongly suggest that PARP-1 cleavage or its cleavage products contribute to OGD-induced cell death. Without OGD, control cells (treated with siRNA scramble and not subjected to OGD showed a viability of 83.22 ± 0.67 for PARP-1_WT and 80.95 ± 2.32 for PARP-1_UNCL (Fig. 2C; Fig. 2D top panels). That only cells expressing PARP-1_UNCL show a significantly higher viability after OGD, suggests that PARP-1 cleavage or its cleavage products might be involved in the regulation of OGD toxicity, and led us to take a closer look at how each cleavage product may contribute to this.

Figure 2. Only after silencing endogenous PARP-1 does the over-expression of PARP-1_UNCL confer resistance to human SH-SY5Ycells exposed to 15h OGD.

(A) Western blots showing the depletion of endogenous PARP-1 by siRNA. Samples were total extracts from cells untreated (first lane), treated for 4 days with scrambled (negative control) (second lane) or treated with PARP-1 specific siRNA (third lane). (B) Western blots showing the depletion of endogenous PARP-1, the overexpression of siRNA resistant PARP-1_WT or PARP-1_UNCL and the defective accumulation of the 89kDa fragment in PARP-1_UNCL overexpressing cells when etoposide was used. (C) Cell viability with or without 15h OGD in siRNA-resistant PARP-1_WT (white bars) or PARP-1_UNCL (black bars) stable transfectants, with siRNA and tet. Data are shown as mean ± SD of 5 (for PARP-1_WT) or 8 (for PARP-1_UNCL) independent experiments (*p<0.01). (D) Representative plots from flow cytometric analysis of Hoechst 33342 and PI labeled cells over-expressing PARP-1_WT (left) or PARP-1_UNCL (right) (tet +; siRNA+) before and after 15h OGD. Most of the cells can be seen in the panel that depicts vital cells in the top panels (cells before OGD). A fewer number of apoptotic cells can be observed in the plot on the right corresponding to SH-SY5Y cells over-expressing PARP-1_UNCL. V=Vital cells, M-Ap=Middle Apoptosis, L-Ap=Late Apoptosis, Vl-Ap=Very late Apoptosis, EN=Early Necrosis, LN=Late Necrosis. Cells were treated with siRNA before performing OGD.

3.3 SH-SY5Y cells over-expressing PARP-1₂₄ show resistance to OGD while cells over-expressing PARP-1₈₉ are more susceptible to cell death

Since PARP-1 cleavage is an event that results in the formation of two products at a ratio of 1:1, we examined the role of cleavage directly by using the cells that over-expressed each cleavage product (either PARP-1₂₄ or PARP-1₈₉). When these cells were challenged by 15h OGD, PARP-1₂₄ conferred significant protection (18.09% increase in viability) compared to control cells (n=12) (p<0.05) (Fig. 3A). By contrast, PARP-1₈₉ significantly decreased their viability, by 30.22%, when compared to control cells (n=3) (Fig. 3A). Figure 3B shows representative immunoblots of PARP-1₂₄ (top panel) or PARP-1₈₉ (bottom panel) expression levels with or without tetracycline. These data indicate that PARP-1 cleavage products have opposing effects on viability following OGD. PARP-1₂₄ confers protection while PARP-1₈₉ is harmful.

Figure 3. Effects of PARP-1 cleavage products on OGD-induced cell death.

(A) Viability of cells grown without tet (cnt=control), and PARP-1₈₉ or PARP-1₂₄ over-expressing cells (tet+) after 15h OGD. Viable cell numbers were compared to non-tet control cells, calculated as percentage of control (without tet) cells, and shown as mean ± SD of at least 4 independent experiments (*p<0.05). (B) Representative Western blots of nuclear fractions showing increased expression of PARP-1₂₄ (top) and PARP-1₈₉ (bottom). Cells were treated with siRNA before performing OGD.

3.4 Primary rat cortical neurons transduced with PARP-1₂₄ or PARP-1_UNCL are more resistant to OGD/ROG than neurons transduced with PARP-1_WT

Using conventional transfection methods, high transfection efficiency of primary cultures of rat cortical neurons is extremely hard to achieve. To overcome this, we developed a viral transduction method (see Materials and Methods) that yields a transfection efficiency of approximately 60%. Due to their potential to confer protection in SH-SY5Y cells, we focused our efforts on PARP-1_UNCL and PARP-1₂₄. As shown in Fig. 4 (bottom panels and histogram), cortical neurons expressing PARP-1₂₄ showed 51% increase in viability compared to cells expressing PARP-1_WT (PARP-1_WT =100%, PARP-1₂₄ =151%±21.29; mean ± SD; n=5 T-test, p<0.05). Furthermore, cells expressing the uncleavable PARP-1_UNCL were 22% more viable compared to WT expressing cells (PARP-1_WT =100%, PARP-1_UNCL =121.75±12; mean ± SD; n=5 T-test, p<0.05). Control cells (treated with the various viral constructs, but not challenged with OGD) (Fig. 4 top panels), all had comparable viabilities (67.52 ± 4.2 for PARP-1_WT, 65 ± 6.4 for PARP-1_24, 66 ± 3.8 for PARP-1_UNCL), suggesting that mere introduction of the virus did not disproportionately affect viability. These data both corroborate results from SH-SY5Y cells suggesting that PARP-1 cleavage may regulate cell viability, and also extend the findings to rat primary cultured neurons.

Figure 4. Primary cortical neurons from rat over-expressing PARP-1₂₄ or PARP-1_UNCL were more resistant to OGD/ROG compared to WT.

Panels labeled “No OGD” show representative bright field, Hoechst 33342 and PI fluorescence images of PARP-1_WT, PARP-1_UNCL or PARP-1₂₄ over-expressing cortical neurons before OGD/ROG. Panels labeled “OGD” show representative Hoechst 33342 and PI fluorescence images of PARP-1_WT, PARP-1_UNCL or PARP-1₂₄ over-expressing cortical neurons after 6h of OGD followed by 15h of reperfusion (ROG). Cell viability in each cell preparation was quantitated and shown in the graph on the bottom left. The inset shows representative examples of a vital cell (diffuse blue nucleus with no PI staining, top arrow), early apoptotic cell (pyknotic blue nucleus with no PI staining, top arrowhead) or late apoptotic/necrotic cell (pyknotic blue nucleus with bright red PI staining, bottom arrowhead). Data are shown as mean ± SD of 5 independent experiments (*p<0.05).

3.5 Changes in tNAD and PARs levels do not correlate with viability of SH-SY5Y cells over-expressing different forms of PARP-1 after OGD

PARP-1 enzymatic activity requires NAD as a substrate to form PARs, and it is possible that the observed differences in viability are due to the depletion/consumption of cellular NAD. Two approaches were used to assess PARylation: measurements of total NAD (tNAD) levels, and measurements of global PARs formation. These assays were performed in SH-SY5Y over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ untreated and exposed to 30 min, 2h, 6h OGD, 6h OGD+2h ROG and 6h OGD+4h ROG. H₂O₂ (200 μM for 1h) was used as a positive control since it is known to induce DNA strand breaks and, as a consequence, PARP-1 activation (PARylation) that in turn consumes tNAD. Levels of tNAD significantly decreased in each cell group after 6h OGD when compared to tNAD levels measured before OGD (Fig. 5A; white vs red bars). It is interesting to note that the baseline tNAD level for PARP-1₈₉ is lower than PARP-1_WT, PARP-1_UNCL, or PARP-1₂₄ (p < 0.01 vs WT). However, with the same OGD or OGD/ROG treatment there was no significant difference among cells expressing the different forms of PARP-1 (Fig. 5A; colored bars). Similarly, levels of tNAD greatly decreased after H₂O₂ treatment, as expected, and no difference was found among these groups (Fig. 5A; green bars). To address the possibility that parallel changes in PARs levels contributed to the differences in viability, PARs formation was evaluated at the time periods of 30 min and 6 hr OGD, based on a pilot time course study. A significant increase in PARs levels was observed following 30 min OGD for all PARP-1 constructs, except PARP-1₈₉ (Fig. 5B). Baseline PARs levels for PARP-1₈₉ were already elevated as compared to the other three forms of PARP-1, an observation complementing that seen for tNAD levels (Fig. 5B). In contrast, PARs levels detected after 6h OGD were comparable and not significantly different across all groups (PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ (Fig. 5C).

Figure 5. Total NAD consumption and PARs formation levels in SH-SY5Y cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ after different time points of OGD and OGD/ROG.

(A) Quantitative analyses of the levels of tNAD. Data are shown as mean ± SD (n=12). p<0.05*; p<10⁻⁵***. (B) Relative PARylation levels among cells after 30 min OGD. Quantification of Western blot densitometry of n=3 (top panel); Western blot in the lower panel. (C) Relative PARylation levels among cells after 6h OGD. Quantification of Western blot densitometry of n=5 (top panel); Western blot in the lower panel. OGD=Oxygen and Glucose Deprivation; ROG=Restoration of Oxygen and Glucose; NS=Not Significant. Cells were treated with siRNA before performing OGD.

These findings suggest that PARylation (PARs formation and tNAD) may not have a role in the regulation of PARP-1 and NF-kB activity observed in this system.

3.6 OGD or OGD/ROG treatment induces caspase activation and PARP-1 cleavage

The appearance of the 89kDa fragment has been considered as a hallmark for apoptosis [16] and the occurrence of PARP-1 cleavage in ischemia models has been shown before [42, 43]. In our model, the fragment was indeed seen in SH-SY5Y cells subjected to a prolonged OGD (15h) (Fig. S1 A) and in cortical neurons after 2-4h OGD followed by 15h ROG (Fig. S1 C). The lack of the appearance of 89 kDa fragment in earlier OGD time points (data not shown) might be a reflection of the limit of the Western blot assay sensitivity, since caspases 3 and 7 which are responsible for PARP-1 cleavage were already activated as early as 6h OGD (Fig. S1 B). However, no differential activations were seen among cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ at least at this time point; an increase of caspase 3/7 activity that varied around 30% was observed in all cell types (Fig. S1 B). These results suggest that the protective effects found in PARP-1_UNCL and PARP-1₂₄ cells were from downstream events.

3.7 PARP-1 cleavage and its cleavage products do not influence nuclear translocation of NF-kB induced by OGD

Since cell sensitivity to OGD seems not to depend on PARylation, we next examined the possibility that another function of PARP-1, a cofactor for NF-kB, plays a role in cell viability during OGD. To do so, we first examined the initial step of NF-kB activation, namely p65 (a subunit of NF-kB) translocation into the nucleus. In primary cortical neurons from rats, nuclear levels of p65 increase by 30 minutes after OGD, peak at 2 h and decrease to baseline by 4h (Fig. 6 A). While levels of p65 increased in the nucleus by 2h in cortical neurons from rats, a corresponding decrease in p65 cytoplasmic levels was difficult to observe. However, IKB-α, the protein that anchors p65 in the cytoplasm, is degraded at 2- 4h of OGD (Fig. S2). Human SH-SY5Y cells showed p65 nuclear translocation within 10 minutes of OGD in the all of the cell types examined (e.g. expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉₎ (Fig. 6 B and Fig. S3). From these results it is clear that OGD is indeed a stimulus that induces NF-kB nuclear translocation, but this action was not affected by PARP-1 cleavage or cleavage products.

Figure 6. OGD induces NF-kB translocation to the nucleus.

(A) Primary rat cortical neurons were exposed to OGD for 0, 30, 120 and 240 minutes. Nuclear levels of NF-kB p65 subunit were analyzed by Western blot (n=3) (B) The nuclear translocation of p65 was apparent by 10 minutes of OGD in human SH-SY5Y cells expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉. Representative blot is shown. Cells were treated with siRNA before performing OGD.

3.8 PARP-1 cleavage appears to be a regulator of NF-kB activity in SH-SY5Y

PARP-1 and NF-kB are known to be part of a larger nuclear multiprotein complex involved in gene transcription [11, 44]. We hypothesized that PARP-1 cleavage products may differentially regulate activity of NF-kB once it is in the nucleus. To test this hypothesis we used a NF-kB reporter assay to evaluate global NF-kB activity. OGD induced NF-kB activity comparably in all cell groups except PARP-1₈₉ cells, which displayed a significant increase (Fig. 7A). We then analyzed a well known NF-kB-dependent gene, INOS. To study the specificity of NF-kB p65 subunit binding to INOS promoter, we used two constructs: one was WT for INOS promoter region [INOS(−1485/+31 NF-κB WT)] and the other was mutated at 3 binding sites of NF-κB for INOS promoter region [INOS(−1485/+31 NF-κB mut)]. The mutant form of the INOS promoter had been mutated such that NF-kB could not bind to it. For this reason, it can be used as a negative control to investigate the dependency of NF-kB activity in INOS reporter assays. OGD increased the levels of p65-dependent INOS transcription in all cell types. Among them, cells expressing PARP-1₈₉ showed a significantly greater increase of INOS transcription than the other PARP-1 isoforms (Fig. 7B). This increase was specifically induced by NF-kB since there was no corresponding increase in the levels of INOS expression when the INOS mutant construct was used (Fig. 7B), suggesting that PARP-1₈₉ indeed modulates NF-kB activity.

Figure 7. NF-kB and iNOS reporter assays. OGD (4h) induces and increase in NF-kB and iNOS activity in human SH-SY5Y expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉.

(A) Though the increase in NF-kB activity can be observed in all cell groups after OGD, among these cells only PARP-1₈₉ induced a significantly higher activity compared to PARP-1_WT, PARP-1_UNCL or PARP-1₂₄. (B) When the construct for INOS promoter used was WT for NF-kB binding sites (INOS WT), OGD induced an increase in INOS activity in cells expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ and PARP-1₈₉. Among these cells only PARP-1₈₉ showed a significant increase in INOS transactivation. When the construct for iNOS promoter used was mutated for NF-kB binding sites (INOS mut) none of the cells showed an induction of the signal with OGD, suggesting that PARP-1 acts through NF-KB transactivation. All reporter assay experiments were conducted in the presence of siRNA, in triplicates, repeated at least 3 independent times and conducted at the same time for the different cell lines; p<0.01. RLUs as F/R = Relative Luciferase Units as Firefly/Renilla.

Figures 7 C and D provide quantitative analyses of transcription levels for iNOS and COX-2 during OGD.

(C) Levels of iNOS and COX-2 transcripts measured after OGD in wild type SH-SY5Y cells (time course) (D) Levels of INOS RNA after 4h OGD in cells overexpressing PARP-1_WT, PARP-1_UNCL, PARP-1_24, and PARP-1_89. Cells were pretreated with siRNA. Results are expressed as mean±SD of n=3.

To confirm these data, rt-qPCR was used. First a time course for OGD was performed and a peak in expression level for both INOS and COX-2 was identified at 4h (Fig. 7C). Next, SH-SY5Y cells overexpressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ were subjected to 4h OGD and the levels of INOS transcript confirmed the data obtained by the reporter assay (Fig. 7D). In particular, INOS message level is significantly higher in the cells expressing PARP-1₈₉ and significantly lower in PARP-1₂₄ expressing cells compared to PARP-1_WT expressing cells.

3.9 Cells over-expressing PARP-1_UNCL or PARP-1₂₄ which are protected from OGD show a reduced inflammatory profile

In order to estimate the scope of the effect of PARP-1 cleavage products, particularly the 89kDa fragment on NF-kB activity, we chose several proteins with NF-kB-dependent expression such as MnSOD, ICAM-1, c-IAP, MMP-9, iNOS, COX-2, Bax and Bcl-x_L and analyzed them in human SH-SY5Y cells expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉. SH-SY5Y cells were subjected to 6h OGD and the levels of the proteins mentioned above were measured by Western blot analyses. Among them, we found the protein expressions of iNOS, COX-2, Bax and Bcl-x_L were influenced by PARP-1_UNCL or PARP-1 products. The pro-inflammatory/harmful protein iNOS (Fig. 8A, white bars) which is known to have a kb-site located in its promoter, was significantly lower in cells over-expressing either PARP-1₂₄ or PARP-1_UNCL by 46% and 33%, respectively (n=3; p<0.05), compared to WT cells after 6h OGD. COX-2 protein levels in these cells were comparable to WT cells. In contrast, the expression of an anti-apoptotic/protective protein Bcl-x_L was significantly higher in these cells (twice as much in PARP-1₂₄ expressing cells compared to PARP-1_WT, n=5, p<0.05) (Fig. 8B, light grey). These data support the idea that the protection conferred by the over-expression of PARP-1₂₄ and uncleavable PARP-1_UNCL could be mediated, in part, through NF-kB regulation of iNOS (down regulation), COX-2 (no increase observed) and Bcl-x_L (up regulation). Conversely, cells over-expressing PARP-1₈₉, which were very sensitive to OGD, showed a significantly higher level of the pro-inflammatory/harmful proteins iNOS and COX-2 (1.7-fold, n=3, p<0.05) and lower level of Bcl-x_L (58% of WT, paired t-test; p<0.05). Finally, even if the expression levels of pro-apoptotic protein Bax were not significantly different among these cell groups, there is a reduction in its levels in cells expressing PARP-1_UNCL and PARP-1₂₄.

Figure 8. Quantitative analyses of protein expression levels of iNOS, COX-2, Bcl-x_L and Bax in cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉ after 6h OGD.

Bottom panels are representative Western blots. The densities of each protein band, from total cell extracts, were measured and normalized with corresponding actin levels and expressed as percentage of that of PARP-1_WT cells. (A) Levels of pro-inflammatory/harmful proteins iNOS and COX2. (B) Levels of anti-apoptotic/protective Bcl-x_L and pro-apoptotic Bax. Data are shown as mean ± SD of four independent experiments (*p<0.05). Cells were transiently transfected with siRNA.

4. Discussion

This report sheds light on the functional role of PARP-1 cleavage products in models of in vitro “ischemia” using OGD in human SH-SY5Y cells and OGD/ROG in primary rat cortical neurons. To the best of our knowledge, this is the first study demonstrating that expression of the uncleavable PARP-1_UNCL confers protection in this model. It is interesting that even though PARP-1_UNCL is not present in nature, our findings and previous studies [23, 45] suggest that PARP-1 cleavage and its cleavage products regulate functions and control mechanisms related to cell survival and cell death. Hence, a novel element in this study was to show differential effects on viability by the 24kDa and the 89kDa fragment, products of PARP-1 cleavage.

This study also reports that the overexpression of PARP-1₂₄ confers protection from OGD in SH-SY5Y cells as well from OGD/ROG in cortical neurons, suggesting that this part of the molecule may be necessary to perform this function. PARP-1₂₄ preserves more than 60% of the PARP-1 DNA binding domain; it has been shown that this fragment may compete with PARP-1_WT for DNA binding and reduce PARylation [17, 46-48]. If this were occurring in our studies, one would expect to see higher levels of NAD and lower levels of PARylation after OGD compared to WT in cells expressing this fragment. However, PARP-1₂₄ (like PARP-1_UNCL) did not change NAD levels and PARylation after OGD. As cellular apoptosis was also not apparent, we evaluated the hypothesis that PARP-1 cleavage products might affect NF-kB-dependent molecular pathways.

It has been shown that PARP-1 can bind to NF-kB [11], and that subsequent PARP-1 cleavage by caspases 3 and 7 could modulate the NF-kB-dependent inflammatory response in an apoptosis-independent event [20, 23, 45]. We then hypothesized that PARP-1 cleavage could be an element that finely tunes NF-kB activity, depending on its binding to different parts of the PARP-1 molecule (N or C-terminal fragment), which are produced by the cleavage (hypothetical model in Fig. 9). Although it is still unclear in our system whether the influence of PARP-1 fragments on NF-kB is indirect or occurs via direct interaction, several differences in NF-kB activity and NF-kB-related molecules were seen when the different constructs were expressed in SH-SY5Y. In our model, OGD induces p65 translocation into the nucleus in both cell types within 10 minutes. “Ischemia” in vitro increases NF-kB global activity in all groups, but no further increase was observed for either PARP-1_UNCL or PARP-1₂₄ overexpressing cells. In contrast, cells expressing PARP-1₈₉ show a significant increase in both general NF-kB activity as well as NF-kB –dependent induction of iNOS. In addition, cleaved fragments PARP-1₂₄ and PARP-1₈₉ have opposing actions on the protein levels of reported NF-kB downstream effectors iNOS (as well as on increased iNOS transcript levels), COX-2, and Bcl-_XL. Collectively, these data suggest that even though there is an increase in NF-kB activity, this increase could be translated into either an increase or decrease in molecules usually considered markers of a more “anti-inflammatory” or “pro-inflammatory” profile. It should be noted that, although neuroblastoma cells are not classically considered inflammatory cells, they have been associated with changes in molecules involved in inflammatory pathways [49, 50]. We demonstrated that the overexpression of PARP-1_UNCL as well as PARP-1₂₄ is associated with decreases in iNOS and COX-2 proteins, and an upregulation of the anti-apoptotic/protective protein Bcl-x_L. The expression of these proteins suggests that these anti-inflammatory events may be, at least in part, NF-kB-dependent and are supported by findings in the literature [51-53]. In addition, the results of the present study demonstrate that PARP-1₈₉, the other product of PARP-1 cleavage, appears to alter NF-kB transactivation towards a pro-inflammatory profile.

Figure 9. Hypothetical model of the mechanism of action of PARP-1 and its cleavage products.

Green areas (left and right) indicate events that have been investigated in this study. The formation of the heterodimer complex between NF-kB (p65+p50) is a known phenomenon in the literature [23, 56, 57]. The orange area in the middle represents a hypothetical mechanism that might explain the differential expression of downstream effectors of NF-kB and cell viability. Upon stimulation with OGD, NF-kB translocates into the nucleus where PARP-1 acts as its co-modulator, modifying NF-kB gene transactivation (left side). NF-kB increased activity is observed when each of the PARP-1 constructs (PARP-1_UNCL, PARP-1₂₄ or PARP-1₈₉) are expressed (right side). The findings in this study suggest that NF-kB increased activity is not only a quantitative phenomenon during cellular ischemic stress, but rather a finely tuned event perhaps influenced by co-factors such as PARP-1. PARP-1 and its cleavage products may be capable of regulating post-translational events depending on which part of the PARP-1 molecule is regulating NF-kB activity. After OGD, the expression of PARP-1_UNCL or PARP-1₂₄ (a. and b.) leads to an increased NF-kB activity in a more anti-inflammatory/protective fashion (decreased iNOS and COX-2 protein levels; increased Bcl-x_L). The expression of PARP-1₈₉ also induces NF-kB transactivation, but its downstream effectors (iNOS and COX-2 increased; Bcl-x_L decreased) show a more pro-inflammatory profile (c.). NF-kB=p65+p50; 24kDa=PARP-1₂₄; 89kDa=PARP-1₈₉.

These observations support the idea that an increase of C-terminal PARP-1₈₉ enhances the toxicity of OGD. Previous studies suggested that the role of PARP-1 cleavage is to protect the cell by inactivation of PARylation and NAD depletion, and others have reported a loss of PARP-1 enzymatic activity after cleavage [17]. It is possible that a reduction in the levels of NAD combined with the increased PARs in the PARP-1₈₉ cells may contribute to an increased energy imbalance, which could partially explain the increased toxicity seen when this construct is expressed. However, several contradictory observations bring this inference into question. First, levels of NAD are the same across all the cell lines following OGD (Fig 5A). If NAD were a factor, one would expect the levels seen with PARP-1₈₉ to be much lower following OGD. Second, NAD levels in the protected cells (PARP-1_UNCL and PARP-1₂₄) are not higher than the control (PARP-1_WT). Following p65 nuclear translocation, we observed a significant increase in global NF-kB activity when compared to PARP-1_WT as well as a significant increase in NF-kB-dependent iNOS activity. Furthermore, these changes were accompanied by higher levels of the pro-inflammatory/harmful protein iNOS (as well as its mRNA), COX-2 and the amount of the anti-apoptotic/protective protein Bcl-x_L.

These observations provide renewed impetus for further investigation of the mechanisms involved in NF-kB regulation - a transcription factor with well known clinical significance. The findings provide just one paradigm of the complex protein regulation in which PARP-1 is just one of the players. However, understanding how the intricate protein to protein interactomes [44, 54, 55] are regulated is an important, if daunting challenge.

Supplementary Material

02. Figure S1.

OGD or OGD/ROG induce PARP-1 cleavage.

(A) SH-SY5Y cells, pretreated with siRNA, were subjected to 15h OGD. The formation of the 89kDa cleavage product can be detected in cells expressing PARP-1_WT but not PARP-1_UNCL. (B) Levels of caspase 3/7 activities after 6h OGD were measured and increase of 30% compared to control (no OGD) cells were detected in cells expressing the four constructs. Cells were previously transiently transfected with siRNA. (C) Cortical neurons from rat were subjected to PARP-1 cleavage after 2h and 4h OGD followed by ROG.

03. Figure S2.

Western blot showing cytoplasmic levels of p65, IKB-α and β-actin in cortical neurons from rat after 0′, 30′, 120′ or 240′ OGD.

04. Figure S3.

NF-kB p65subunit nuclear translocation after 10 minutes of OGD in SH-SY5Y cells over-expressing PARP-1_WT, PARP-1_UNCL, PARP-1₂₄ or PARP-1_89. p65 is indicated in green and DAPI staining in blue. The top row shows a p65 distribution that is mainly cytoplasmic (the nuclei are stained in blue). After 10 minutes of OGD there is a prevalent nuclear distribution of p65 (areas of green and blue overlap).

Highlights.

• PARP-1mutants (uncleavable and 24kDa) increase cell viability in in vitro ischemia

• PARP-1 cleavage products (24kDa and 89kDa) have opposite effects on cell viability

• No augmentation of NF-kB and iNOS activity was seen in protected cells

Abbreviations

PARP-1

poly(ADP-ribose) polymerase 1

PARylation

Poly(ADP-ribosyl)ation

NB

Neuroblastoma

OGD

oxygen and glucose deprivation

24kDa PARP-1

24 kDa fragment of PARP-1

89 kDa PARP-1

89 kDa fragment of PARP-1

NF-kB

nuclear factor-kappa B

tet

tetracycline