The Septic Shock-associated IL-10 -1082 A>G Polymorphism Mediates Allele-specific Transcription via Poly ADP-ribose Polymerase 1 in Macrophages Engulfing Apoptotic Cells

Abstract

The biallelic Interleukin-10 single nucleotide polymorphism (SNP) at -1082 of the promoter region linked to individual variation in cytokine inducibility has been strongly implicated in several pathological conditions including the development of, and outcomes in, septic shock during pneumococcal infection, acute respiratory distress syndrome, and cardiac dysfunction. However, the molecular basis of the SNP-mediated variable IL-10 production levels has not been explored. Here we report that the -1082G>A alleles in the promoter region of the human IL-10 gene physically interact with a nuclear protein in an allele-specific manner that results in different levels of IL-10 transcription. This protein has been identified as poly ADP-ribose polymerase 1 (PARP-1). We show that PARP-1 acts as a transcription repressor, and its DNA-binding activity is strongly regulated in macrophages that engulf apoptotic cells but not stimulated with lippopolysaccharides. These findings unveil a novel role of PARP-1 in the regulation of IL-10 production in an allele-dependent way, which determines individual susceptibility to sepsis-induced inflammatory pathology and the immunological sequelae in a physiological process where clearance of infection-induced apoptotic cells by professional phagocytes triggers the cytokine synthesis.

INTRODUCTION

Interleukin (IL)-10 is an important anti-inflammatory cytokine that modulates pro-inflammatory cytokines, such as tumor necrosis factor (TNF)-α, as well as synthesis of nitric oxide, apoptosis of inflammatory cells and suppression of macrophage activation (1). IL-10 attenuates the pro-inflammatory response in bacterial sepsis and reduces others (2). However, excess of IL-10 induces immunosuppression in sepsis (2) and increases mortality by impairing bacterial clearance in pneumococcal pneumonia (3). In humans, elevated circulating IL-10 has been associated with septic shock (4), severity of injury (5–10) and mortality (11, 12). In acute respiratory distress syndrome (ARDS), the studies have been mixed. Lower levels of IL-10 were found in patients with ARDS compared with critically ill non-ARDS patients (13). High plasma IL-10 but low bronchoalveolar lavage concentration of IL-10 correlated with increased mortality in ARDS (14, 15). Susceptibility for invasive pneumococcal disease has also been associated with the mannose binding lectin gene, but no genetic linkage has been found for sepsis severity (16).

The elimination of apoptotic cells and cell bodies by phagocytes represents an evolutionarily conserved means to prevent exposure of surrounding tissue to potentially cytotoxic, immunogenic, or inflammatory cellular content (17, 18). Resolution of inflammation depends not only on the removal of apoptotic cells but also on active suppression of inflammatory mediator production. Aberrations in either mechanism are associated with chronic inflammatory conditions and autoimmune disorders (15, 19, 20). Uptake of apoptotic cells by phagocytes is thought to suppress autoimmune responses through the release of anti-inflammatory cytokines IL-10, TGF-β, platelet activating factor (PAF), and prostaglandin E₂ (PGE₂), and inhibition of pro-inflammatory cytokines TNF-α, GM-CSF, IL-12, IL-1β, and IL-18 (21–23).

A systemic infectious insult is often associated with subsequent hyporesponsiveness to endotoxin and an increased risk of late nosocomial infection in some patients. For example, immediately following cardiac surgery, many patients become relatively refractory to lipopolysaccharide (LPS) stimulation. One study found that in these patients, stimulated cytokine production in whole blood was lowest in cases with the highest postoperative plasma IL-10 levels. Those patients in whom the whole blood response to endotoxin was maintained over the first 48 hrs were more likely to have an uncomplicated short stay (24).

It has been reported that 50–75% of the variation in IL-10 production is genetically controlled (25, 26). Single nucleotide polymorphisms (SNPs) have been associated with different cytokine production (27). There are three major SNPs in the human IL-10 promoter region: -1082G>A, -819C>T, and -592C>A, respectively. Individuals homozygous for the G allele (-1082GG) have higher circulating IL-10, higher expression of IL-10 mRNA, and greater production of IL-10 after in vitro stimulation (27–29). It was shown that the A allele of the -1082 polymorphism is associated with lower IL-10 production and sepsis susceptibility in patients, whereas G allele is associated increased mortality in severe sepsis (30). Another study demonstrated that the IL-10 intermediate/high producer genotype (-1082 G-allele carrier) was associated with a lower risk of death among patients with acute renal failure who require dialysis (31). It was reported that individuals with genetic predisposition for increased IL-10 inducibility, as determined by the IL-10 -1082 polymorphism, have a higher risk of severe pneumococcal infection leading to septic shock (32). A nested case-control epidemiology study in ARDS patients and controls who were admitted to an intensive care unit with sepsis, trauma, aspiration or massive transfusions revealed that the high IL-10-producing -1082GG genotype is associated with variable odds for ARDS development depending on age, being associated with lower mortality and organ failure (33). It was reported that in patients with acute pancreatitis, those that developed further into septic shock showed a significantly higher prevalence of the -1082G allele than those without shock (34).

The preponderance of clinical data on the association of the IL-10 -1082 SNP with bacterial septic shock and the sequelae is contrasted with an almost total lack of understanding of the molecular basis of the SNP-associated variability in IL-10 levels in different individuals. We undertook the current study to address a key question: How does the -1082 SNP affect IL-10 gene expression at the molecular level in macrophages responding to microbial challenges and to apoptotic cells?

MATERIALS AND METHODS

Cells and Reagents

Mouse monocytic cell lines RAW264.7 and human Mono Mac-6 cell line were obtained from American Type Culture Collection (ATCC). Human monocytes were derived from blood buffy coats purchased commercially from the New York Blood Center. Cells were maintained in RPMI 1640 with 10% Fetal Bovine Serum (FBS), 100 units/ml of penicillin, 100 μg/ml of streptomycin and 200 mM L-Glutamine. Antibodies against PARP-1 were purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). LPS was from Sigma-Aldrich (St. Louis, catalog # L-3129). 3-Aminobenzamide (3-AB), 8-hydroxy-2-methylquinazolin-4-[3H]one (NU1025), and fisetin were from Sigma-Aldrich. These compounds were dissolved in DMSO at 50 mM and aliquots were stored in the dark at −20°C.

Reporter Plasmids

The human IL-10 promoter-luciferase construct (pIL-10 (−1105/+36)/pGL2 basic vector) was cloned by PCR amplification of human genomic DNA from human monocytes. PCR product was cut utilizing BamH1 + EcoRV and then ligated to the pGL2-basic vector (cut by SmaI + BaglII). Initially, SNP’s obtained were -1082 A, -819 T and -592 T. To obtain the SNP’s of interest, site directed mutagenesis utilizing the Quikchange® XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, Ca) was performed. First, we mutated the -592 T to an A and had -1082 A, -819 T and -592 A. Also, we made -1082 G/-819 T/ -592 T from the initial construct (ATT). Subsequently, we made -1082 G/-819 C/ -592 T and -1082 G/-819 C/ -592 C. Finally, we made -1082 A/-819 C/ -592 C by changing the -1082 G back to A.

Transient transfection and measurement of luciferase activity

Transfection of RAW264.7 cells with plasmids containing full-length hIL-10 promoter was performed using electroporation followed by luciferase assay. Cells were collected, washed once with RPMI-1640 medium, and resuspended in the same medium at a concentration of 10 × 10⁶ cells/per condition 700 btl of cell suspension and 300 Ixg of DNA were placed in 0.45-cm electroporation cuvette (Gene Pulser; Bio-Rad # 165–2088 Laboratories, Richmond, CA), and electroporation was carried out at 975 μFD and 300 V (0.4 capacitance). Transfected cells were collected and resuspended to 5 × 10⁶/ml in RPMI-1640 + 10% FBS, and chloroquine was added to a final concentration of 10 μM. Cells were placed in wells (2 ml/well) of a 24-well plate and incubated for 16 h at 37°C in a 5% CO2 atmosphere. Next, cells were treated with 1.2% DMSO for approximately 7 h, appropriate stimuli were added for 24 h. After the stimulation, cells were harvested and lysed with 1 x lysis buffer and vigorous shaking. Lysates were used for the luciferase assay. All statistical analyses were performed with two-tailed Student’s t-test. Data were considered significant if P < 0.05.

Induction of apoptosis

Jurkat T cells were the source of apoptotic cells. Staurosporine (0.5μg/ml) (Cayman Chemical) was added at (0.5ng/ml) to T cells resuspended at (4 × 10⁶ cells/ml) with complete RPMI 1640 (Gibco, Invitrogen). After incubation for 6 h at 37°C in the presence of 5% CO₂, the cells were harvested and washed three times with incomplete RPMI 1640. At this time, about 65% of the population was Annexin V positive (early apoptotic) and PI negative as determined by FACs staining. Cell viability by trypan blue staining was >90%.

Nuclear extract preparation and Electrophoresis Mobility Shift Assay (EMSA)

Nuclear extracts and EMSA were performed as described by Schreiber et al.¹¹ The probe sequences were as follows: TTCTTTGGGAG/AGGGGAAGTA (the SNP is bolded)

DNA pull- down assay and polyacrylamide gel electrophoresis analysis

Complementary biotinylated oligonucleotides encompassing the -1082A/G-binding site, TTCTTTGGGAG/AGGGGAAGTA (the critical SNP A or G is in bold case and underlined) were synthesized and annealed to form double-stranded DNA. Biotinylated double-stranded DNA (2 mg) were conjugated to 100 ml streptavidin-bound magnetic beads (Dynabeads, M280; Dynal) in binding/ washing buffer (10mM Tris–HCl, pH 8.0, 1mM EDTA and 0.1M NaCl) for 30 min at room temperature. Conjugated DNA was collected with a magnetic particle concentrator. DNA-conjugated beads were then blocked with 0.5% bovine serum albumin in TGEDN buffer (120mM Tris–HCl, pH 8.0, 1mM EDTA, 0.1M NaCl, 1mM dithiothreitol, 0.1% Triton X-100 and 10% glycerol) at room temperature for 1 h. Beads were washed once in TGEDN buffer and resuspended in 50 ml TGEDN. Ten-microliter beads conjugated to 2 mg DNA were equilibrated with TGEDN buffer and incubated with 500 μg RAW264.7 cell nuclear extracts and 20 μg herring sperm DNA (Sigma-Aldrich) at 41C for 2 h. Beads were washed in TGEDN buffer and bound materials were eluted in 20 ml of the same buffer supplemented with 0.5% SDS and 1 M NaCl. Eluted proteins were separated by 10 or 12% SDS-polyacrylamide gel. The gel was visualized by Coomassie staining.

Chromatin immunoprecipitation (ChIP) assay

ChIP assay was performed by following the protocol in the ChIP Assay kit (Upstate Biotechnology). The presence of a selected DNA sequence of the human IL-10 gene was assessed by PCR. The primers used were sense: −1217/−1198, 5′-CAACTGGCTCCCCTTACCTT -3′, and antisense: −998/−979, 5′-ACCTCCTATCCAGCCTCCAT, yielding a 239-bp product. As a negative control, a separate region of the human IL-10 promoter was also included in the ChIP experiment. It is located between −3158 and −2947 upstream of the −1217/−979 region. The pair of PCR primers used in this control were sense: 5′-AGTGAGAAGGCAGGCACCTA-3′, and antisense: 5′-ATCCCCCACTGGAAAAATTC -3′, yielding a 212-bp product. The PCR cycles were as follows: 94°C for 4 min, 1 cycle; 94°C for 30 s, 54°C for 30 s, and 72°C for 30 s, 32 cycles; and 72°C for 7 min, 1 cycle.

RESULTS

Differential transcriptional activities of -1082A and -1082G-IL-10 promoter haplotypes

To understand the regulation of the IL-10 gene transcription via the -1082A/G SNP in macrophages, we used a well-established transient transfection system in the murine macrophage-like cell line RAW264.7 and a human IL-10 promoter-luciferase reporter construct containing the region between -1105 and +30 upstream of the IL-10 transcription initiation site. Two versions of this construct were engineered to reflect the -1082 SNP of interest: -1082A and -1082G, respectively. These constructs were completely identical otherwise. Following transient transfection of the two reporter constructs, RAW264.7 cells were stimulated with apoptotic cells in different amounts or with LPS, luciferase activity was measured afterwards. As shown in Fig 1, the -1082A and -1082G haplotype IL-10 promoter constructs exhibited different responses to apoptotic cells (AC) and to LPS in that approximately 61–65% more activities were seen with the -1082G promoter over those of the -1082A promoter in response to AC (at 1:1 ratio of phagocytes to AC), and to LPS, respectively. As the amount of AC increased to 1:2 and 1:3 ratios, the responses of the two promoters became more equal. It is noteworthy that the basal transcriptional activities of the two haplotype promoters also differed by about 63%, suggesting that the inherent difference of the two promoters is intrinsic independently of stimulation, and cellular stimulation merely amplifies the inherent differences.

Fig 1. Differential response of IL-10 promoter haplotypes to apoptotic cells and to LPS.

The -1082A- and G- haplotype IL-10 promoter-luciferase reporter constructs in the context of −1105/+30 transiently transfected into RAW264.7 cells. Transfected cells were incubated with in medium alone or stimulated with apoptotic Jurkat cells (AC) at various ratios of macrophage to AC, or with LPS (1 μg/ml) for 24 h. Luciferase activity was measured from cell lysate. Result represents the mean ± SD of four individual experiments. Two of the five comparisons reached statistical significance defined as p<0.05 (*).

Differential DNA binding to -1082 SNP by a novel nuclear factor

We further hypothesized that underlying the variable transcription activities of the -1082A and -1082G promoter haplotypes, there may be differential binding by transcription factors to the SNP that control IL-10 gene transcription to different degrees. To test this hypothesis, we performed electrophoretic mobility shift assays (EMSA) using nuclear extracts isolated from macrophages exposed to apoptotic cells or to LPS, with double-stranded oligonucleotide probes spanning the SNP with one nucleotide difference at the SNP. As shown in Fig 2A, a highly prominent DNA-binding activity was observed with the -1082A haplotype probe in unstimulated RAW264.7 cells (lane 2, indicated by an arrow), and this binding activity was strongly reduced by AC dose-dependently (lanes 3–5), whereas LPS had little impact on the activity (lane 6). In contrast, a much weakened binding activity was observed with the -1082G probe (lane 8), which was further decreased by AC dose-dependently (lanes 9–11), and which was not significantly affected by LPS (lane 12). In human blood-derived monocytes, a similarly prominent but faster-moving binding activity was observed with the -1082A probe in a pattern that was very similar to what was seen in RAW264.7 cells in that the constitutively present differential binding activity was reduced by AC but not much by LPS (Fig 2B).

Fig 2. Differential binding to the -1082 SNPs by a nuclear factor.

Nuclear extracts were isolated from RAW264.7 cells (A) or primary human monocytes (B), non-stimulated or exposed to AC in various ratios, or stimulated with LPS (1 μg/ml). Electrophoretic mobility shift assay (EMSA) was performed with the -1082A- (lanes 1–6) or G-haplotype probe (lanes 7–12). The novel DNA-binding activity is indicated by an arrow. Mφ/AC, ratio of macrophage vs AC. The numbers below the specific bands are arbitrary densitometric values of each band. FP, free probe.

Given the pattern of the novel DNA-binding activity, we hypothesized that it may represent a transcriptional repressor present in unstimulated monocytes and macrophages. Upon contact with AC, but not stimulation by LPS, its DNA-binding activity is reduced, allowing IL-10 gene transcription to proceed in a haplotype-specific manner. Because the -1082G promoter is bound by this factor to a lesser extent than the -1082A promoter, the former is therefore transcribed more robustly than the latter.

Biochemical identification of the novel DNA-binding activity

To identify this novel DNA-binding activity we carried out a series of biochemical experiments by DNA affinity pull-down assays and SDS-PAGE analysis with nuclear extracts derived from RAW264.7 (mouse macrophage) and Mono-Mac 6 (human macrophage) cells. As shown in Fig 3A, By Coomassie staining, a band of ~115 kD was strongly pulled down by the -1082A probe (lanes 2 and 5, indicated by an arrow), less so by the -1082G probe (lanes 3 and 6), in nuclear extracts derived from unstimulated cells, but not at all when no nuclear extracts were added (lanes 4 and 7). Another protein of ~75 kDa (indicated by an *) was also strongly pulled down but it was likely a non-specific species because it was present in all samples regardless whether nuclear extracts were added or not. The specific ~115kDa band was excised and analyzed by matrix assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS). The result revealed, by a very high search score, that it was Poly (ADP-ribose) polymerase-1 (PARP-1) in both the mouse and human samples.

Fig 3. Identification of PARP-1.

(A) DNA pull-down assay was performed with complementary biotinylated oligonucleotides encompassing the -1082A/G-binding sites, as described in Materials and Methods, with nuclear extracts from RAW264.7 cells (left panel) or from MonoMac 6 cells (right panel). Eluted proteins were separated by 10% SDS-polyacrylamide gel. The gel was visualized by Coomassie Blue. This experiment was repeated two more times independently with identical results. The two bands associated with the A-and G-probes (indicated by an arrow) were excised and analyzed by mass spectrometry. The highest scored hit was poly (ADP-ribose) polymerase-1 (PARP-1). (B) Western blot analysis. The same samples that had been through the procedure described in (A) were subject to Western blot analysis using anti-PARP-1 (A-20). The intact (113 kDa) and cleaved PARP-1 products are indicated. This analysis was performed twice. (C–D) EMSA was performed using nuclear extract isolated from RAW264.7 cells (C) and human MonoMac 6 cells (D) with the-1082A probe. Various antibodies (1 μg each) were used: anti-PARP-1 (H-250) and its control rabbit IgG, A-20 and its control goat IgG. Lane 1 contains free probe (FP). PARP-1-binding is indicated by an arrow.

PARP-1 is a nicotinamide adenine dinucleotide (NAD⁺)-dependent nuclear enzyme. It is a 113-kDa protein composed of an N-terminal DNA binding domain, containing two zinc finger motifs, a C-terminal NAD⁺ binding domain, catalyzing the synthesis of ADP-ribose polymers from its substrate, NAD+, and an automodification site, which links the N- and C-terminal domains (35). This nuclear enzyme is activated when oxidative or nitrosative stress causes DNA strand breaks. The rapid activation of PARP-1 depletes the intracellular concentration of nicotinamide. This results in a reduced rate for cellular processes depending on nicotinamide including glycolysis, electron transport and ATP formation. This can then lead to cell death (36).

The identity of PARP-1 was further confirmed by Western blot analysis using A-20, a goat polyclonal antibody raised against a peptide mapping at the N terminus of PARP-1 of mouse origin specific for PARP-1 (Fig 3B). The analysis also revealed a smaller peptide of ~25kDa. It is well known that during apoptosis, PARP-1 is cleaved by caspase 3, resulting in the N-terminal 24-kDa DNA-binding fragment and the C-terminal 89-kDa catalytic fragment. This cleavage is important for the regulation of inflammatory responses by PARP-1 (37). It has been shown that the 24-kDa fragment can act to compete against the full-length PARP-1 and inhibit DNA repair, ADP-ribose polymer formation, and damage-dependent upregulation of transcription (38, 39). Our result suggests that both the complete protein and N-terminal peptide are capable of binding to IL-10 at this allelic site, whereas the 89-kDa catalytic fragment is not.

Consistent with this identification, EMSA analysis using two anti-PARP-1 antibodies (A-20 and H-250, a rabbit polyclonal antibody against an epitope corresponding to amino acids 764–1014 mapping at the C-terminus of PARP-1 of human origin) resulted in significantly reduced specific binding to the -1082A probe in both RAW264.7 (Fig 3C) and Mono-Mac 6 (Fig 3D) cells (lanes 5 and 6), compared to control antibodies (lanes 3 and 4). The fact that these antibodies only reduced the specific binding rather than altering its mobility suggests that PARP-1 may be the major component, if not the only component, of the specific complex. Thus, we have identified PARP-1 as the key component of the novel binding activity, which interacts differentially with the -1082A and -1082G haplotypes.

PARP-1 transcriptionally represses IL-10 gene expression via the -1082 SNP

We investigated the direct role of PARP-1 in the regulation of IL-10 gene transcription by overexpressing PARP-1. PARP-1 expression was delivered by cotransfection with a PARP-1-expression vector (40) into RAW264.7 cells together with the -1082A and -1082G IL-10 promoter reporters. Following exposure to AC, both -1082A and -1082G promoters’ activities were inhibited to similar degrees (Fig 4A). This inhibition was not seen with an enzymatically inactivemutant of PARP-1 (with a point mutation, E988K, in the catalytic domain) (41), suggesting that inhibition of IL-10 transcription requires the enzymatic activity of PARP-1.

Fig 4. Role of PARP-1 in regulating IL-10 transcription.

(A) The -1082A- and G-IL-10 promoter reporter constructs were transfected into RAW264.7 cells by electroporation together with a control vector (CV), or the wild-type (WT) PARP-1 (effector), or an enzymatically inactive mutant of PARP-1 (Mut), E988K, at a 1:1 molar ratio (effector to reporter). Cells were then exposed to AC (1:2 ratio of Mφ to AC) for 24 h followed by luciferase activity measurement. (B) RAW264.7 cells transfected with the -1082A- or G-IL-10 promoter reporter construct were exposed to AC in the presence of varying amounts of 3-AB as indicated for 24 h. Luciferase activity was measured from cell lysate. Results represent the mean ± SD of three individual experiments. * indicates P < 0.05.

Next, we performed the converse experiment, i.e. to use a small molecule inhibitor of PARP-1 activity, 3-aminobenzamide (3-AB), which can protect cells from the toxic effects of oxygen radicals and nitric oxide caused by PARP-1 activation (42–44). As shown in Fig 4B, inhibiting PARP-1 activity with 3-AB dose-dependently augmented the -1082A IL-10 promoter activity induced by apoptotic cells, whereas its effect on the -1082G promoter was not observed except at 5 mM. This is consistent with the notion that the -1082A promoter has a lower transcriptional activity because it is more bound by PARP-1, thus, it would be more sensitive to a PARP-1 inhibitor. The -1082G promoter, being less repressed by PARP-1, would be less responsive to the inhibitor. It should also be pointed out 5mM 3-AB is generally regarded as the upper limit of the chemical for target specificity beyond which non-specific effects are observed (36).

PARP-1 inhibitors augment IL-10 production in macrophages induced by apoptotic cells

Given the above data suggesting that PARP-1 is an inhibitor of IL-10 transcription, we further reasoned that blocking PARP-1 activities would result in elevated IL-10 expression. To test this hypothesis we treated mouse and human macrophages with several PARP-1 inhibitors: 3-AB, 8-Hydroxy-2-methylquinazolin-4[3H]-one (NU-1025), and fisetin. 3-AB is one of the first generation inhibitors of PARP. In addition, several newly developed, more specific and more potent PARP inhibitors are currently being evaluated in human clinical trials for neurodegenerative diseases and ischemia reperfusion-induced tissue injuries such as myocardial infarction and stroke (45). NU-1025 has been shown to inhibit PARP activity by affecting its subcellular localization and DNA strand break rejoining (46). Fisetin is one of several flavonoids recently identified as inhibitors of PARP-1, and is able to attenuate cytokine release in blood from patients with chronic obstructive pulmonary disease or type 2 diabetes (47). As shown in Fig 5, all three inhibitors resulted in elevated production of IL-10, in a dose-dependent manner, in mouse (A, C, E) and human (B, D, F) macrophages treated with apoptotic cells, but not with LPS. This result is entirely consistent with our previous observation that PARP1-deficient mouse macrophages produced elevated levels of IL-10 in response to apoptotic cells but not to LPS (48). Taken together, these data demonstrate that PARP-1 is a physiological regulator of differential IL-10 production by inhibiting basal and apoptotic cell-induced transcription of the gene through the -1082 SNP.

Fig 5. Effects of pharmacological inhibition of PARP-1 on IL-10 production.

RAW264.7 cells (A, C, E) and freshly isolated primary human monocytes (B, D, F) were treated with AC or LPS for 24 h in the presence of PARP-1 inhibitors: 3-AB (A, B), NU-1025 (C, D), and Fiestin (E, F) in increasing doses. Culture supernatant was harvested and analyzed for IL-10 secretion by ELISA. Data represents mean of three individual donors with SD. The genetic background of the human donors was not determined. * indicates P < 0.05.

Endogenous PARP-1 binding to the IL-10 locus

To demonstrate if PARP-1 could bind the Il10 gene in vivo we performed chromatin immunoprecipitation (ChIP) assays in human PBMC-derived monocytes. Following the IP step, specific primers were used to amplify a small region (from −1217 to −979) covering the -1082 site as outlined in Fig 6A. As shown in Fig 6B, in samples treated with the control IgG (C), no binding activity was detected (lanes 1–5, panel I) whereas in anti-PARP1-treated samples (P), a specific and constitutive binding activity was observed (lane 6), which was reduced in cells exposed to AC (lanes 7–9). LPS treatment did not alter the constitutive binding activity (lane 10), and an unrelated region approximately 2 kb upstream of the -1082 site (49) showed little binding under the same experimental conditions (panel II). The quantification of the relative binding activities measured by real time PCR is shown in panel III. The data shows that PARP-1 indeed binds selectively and specifically to IL-10 in vivo around the -1082 site in a constitutive manner in monocytes, and that exposure to apoptotic cells, but to LPS, reduces this binding.

Fig 6. PARP-1 binding to the IL-10 locus in vivo.

(A) Partial sequence of the human IL-10 promoter that spans the PARP-1 binding site at -1082. The region from −1217 to −979 was amplified by PCR in the ChIP assay with primers whose sequences are underlined. (B) Freshly isolated human monocytes, treated with varying amounts of AC or stimulated with LPS, were subjected to the ChIP procedure with anti-PARP-1 (P) or its isotype control (C). Two pairs of primers were used to amplify the −1217/−979 region (panel I) and a separate region ~2kb upstream between −3158 and −2928 (panel II). The amplified PCR products are 239-bp and 212-bp in length, respectively. Panel III represents the quantification of PARP-1 binding shown in panel I, as measured by real time PCR.

DISCUSSION

This study has identified PARP-1 as a critical nuclear factor that determines differential IL-10 gene transcription in a -1082G>A allele-dependent manner in monocytes/macrophages that encounter/ingest apoptotic cells. Specifically, PARP-1 appears to act as a transcriptional repressor with a direct, preferential binding for the A-allele than the G-allele, resulting in lesser expression of the former. This mechanism occurs both constitutively and in macrophages that engulf apoptotic cells, but not in response to LPS. In other words, individuals carrying the A- or G-allele of -1082, respectively, have intrinsically different abilities to produce IL-10 in the steady state and in response to apoptotic cells induced either intrinsically, such as in normal cellular turnover, or extrinsically such as during an infection.

In severe sepsis and hemorrhage, PARP-1 activation has emerged as one of the central mechanisms of systemic inflammation, endothelial dysfunction, peripheral vascular failure, and reduction of cardiac contractility. For example, there is evidence of significant PARP activation in the hearts of septic patients with impaired cardiac function, and that PARP activation may be partly responsible for the cardiac depression seen in humans with severe sepsis (50). In a murine cecal ligation and puncture (CLP) model of septic shock, treatment with tempol, a low-molecular-weight membrane-permeable radical scavenger, caused a marked reduction in PARP activity in the lung and kidney glomeruli, accompanied by improved mesenteric arterial blood flow (51). PARP-deficient mice subjected to CLP had significantly lower plasma levels of TNF-α and IL-6, and they exhibited a reduced degree of organ inflammation, indicated by decreased myeloperoxidase activity in the gut and lung. These effects were associated with a significant improvement in the survival of CLP in PARP-deficient mice (52). Pseudomonas aeruginosa is commonly associated with nosocomial pneumonia. Ileal mucosal injury may be induced by severe lung infection. It has been shown in a rat model of P. aeruginosa-induced septic shock that pharmacological inhibition of PARP-1 can reduce gut inflammation and limit bacterial translocation (53). These studies demonstrate that innovative therapeutic strategies based on the pharmacological inhibition of PARP-1 catalytic activity might provide benefits by preventing tissue injury, organ dysfunction, and lethality associated with these conditions.

It is well established that during septic shock, peroxynitrite-mediated DNA strand-breaks activate PARP-1 resulting in cellular energetic suppression and cell dysfunction. Our study, however, uncovers an additional, novel, and immunologic mechanism, i.e. PARP-1 activation leads to suppression of IL-10 transcription and production in a genetically differential manner. This finding explains the individually variable susceptibility to sepsis associated with the IL-10 alleles in a physiological process where infection-induced apoptosis occurs. Studies in recent years have suggested that dysregulated apoptotic immune cell death may play a role in contributing to the immune dysfunction and multiple organ failure observed during sepsis (54). Lymphocytes are particularly prone to dysregulated apoptotic cell death. Loss of lymphocytes is detrimental to the survival of septic animals, as documented by the observation that RAG-deficient mice are markedly more susceptible to lethal effects of polymicrobial septic challenge than their wild type controls (54). Lymphocyte apoptosis occurs following the onset of experimental sepsis in the thymus, spleen, and gut-associated lymphoid tissues. One of the hypotheses arising from these observations is that the overt apoptotic loss of lymphocytes in the septic individual reduces the number of functional immune cells available to neutralize the lethal effects of septic challenge. This notion is strongly supported by the observation that if lymphocyte apoptosis is blocked via the restricted over-expression of Bcl-2, the mortality in the CLP model of sepsis in mice is largely abrogated (54). Alternatively, or concurrently, excessive apoptotic cells during an infection might induce large amounts of immunosuppressive cytokines such as IL-10 and TGF-β that incapacitate the immune system in its response to the presence of the infectious pathogen, leading to “immunological paralysis”. The cytokine response in this process, especially the IL-10 response in phagocytes, is genetically determined to a large extent via the interaction with PARP-1, as we have shown here, resulting in individual variability and susceptibility. It is worth pointing out that here, we find overexpression of wild type PARP-1, but not the PARP-1 mutant lacking catalytic activity, partially inhibits IL-10 transcription (Fig 4A). We don’t expect PARP-1 overexpression to inhibit IL-10 transcription completely because apoptotic cells stimulate IL-10 transcription via Pbx-1b binding to a separate site on the IL-10 promoter, as we showed previously (55). PARP-1 does not act on Pbx-1b blocking its transcriptional activity, and PARP-1 merely “fine-tunes” the level of IL-10 transcription according to the -1082 allelic status.

In summary, our study has uncovered a potentially novel pathway that involves apoptosis of lymphocytes during sepsis, the clearance of these apoptotic bodies, and the production of IL-10 in a -1082 A>G allele-specific manner via PARP-1, a critical inflammatory molecule in sepsis and other pathologies. The study provides a mechanistic basis for the individual variability in IL-10 production and in differential susceptibility to sepsis. It has strong implications in the development of therapeutic targeting strategies in sepsis that are sensitive to each patient’s genetic underpinning.