Pancreas-specific RelA/p65 truncation increases susceptibility of acini to inflammation-associated cell death following cerulein pancreatitis (original) (raw)
Pancreas-specific truncation of RelA/p65 increases tissue damage during inflammation. To direct Cre expression to the exocrine pancreas, we generated a knockin mouse (Ptf1a-creex1) with the gene encoding for the Cre recombinase within the Ptf1a locus (21). PTF1 is a heteromeric basic helix-loop-helix protein complex composed of 3 subunits: p48, p64, and p75. The tissue-specific component, p48, is essential for exocrine pancreatic development (19, 20). Thus expression of Cre resulted in the development of mice with pancreas-specific deletion of the floxed allele (exons 7–10; rela_Δ/_Δ mice; Figure 1, A and B) and accordingly in a truncated form of RelA/p65 in the pancreas (Figure 1C). The protein expression level of pancreatic IκBα, a NF-κB/Rel target, was lower in the rela_Δ/_Δ pancreas (Figure 1C). The expression level of p50, a subunit of the NF-κB/Rel heterodimer, was not altered in the rela_Δ/_Δ pancreas. The rela_Δ/_Δ mice developed normally with no signs of obvious dysfunction.
Pancreas-specific truncation of RelA/p65 using a Cre-loxP system. (A) Two identically oriented loxP sites (triangles) flank exons 7 and 10 of the rela gene. _loxP_mod, modified loxP site. (B) Recombination of genomic DNA from the pancreata of relaΔ/Δ mice was detected by Southern blot analysis using a probe external to the 5′ end of the targeting construct. No recombination was detected in the other organs. (C) Deletion of rela (exons 7–10) in the mouse pancreas was demonstrated at the protein level by Western blot of isolated acini from mice with the floxed allele in the presence or absence of Cre as indicated. In contrast to relaflox/flox mice, relaΔ/Δ mice display a truncated form of RelA/p65 (Δp65). Pancreas protein extracts (40 μg) were analyzed using antibodies against p50, IκBα, and β-actin (as a loading control). The protein marker (PM) indicates the size of detected protein bands.
To analyze the function of RelA/p65 in AP, we induced AP by repetitive i.p. injections of sulfated cerulein, a decapeptide analog of the pancreatic secretagogue cholecystokinin. Application of cerulein i.p. in relaflox/flox mice induced rapid and pronounced NF-κB/Rel activation as early as 30 minutes after the first injection (Figure 2A). In Ptf1a-creex1 × relaflox/flox (i.e., rela_Δ/_Δ) mice, deletion of RelA/p65 abrogated NF-κB/Rel activity completely, suggesting that early activation of this transcription factor occurred in the exocrine pancreas. Degradation of the inhibitor protein followed the kinetics of NF-κB/Rel activation in relaflox/flox mice, while rela_Δ/_Δ mice displayed a delayed but clear degradation (Figure 2B).
Inhibition of nuclear translocation of RelA/p65 in the pancreas following cerulein stimulation. (A) relaflox/flox and relaΔ/Δ mice were injected i.p. with 50 μg/kg cerulein at hourly intervals. Pancreatic nuclear protein extracts (10 μg) at the indicated time points were subjected to gel retardation assays with an NF-κB consensus probe. (B) Kinetics of IκBα degradation in relaflox/flox and relaΔ/Δ mice were assessed by Western blot analyzing the protein content of 40 μg of pancreatic whole-cell protein lysates using an antibody against IκBα. Phosphorylated IκBα (p-IκBα; Ser32) was monitored by Western blotting using a phosphospecific antibody.
Pancreatitis in relaΔ/Δ mice was analyzed at different time points after 8 i.p. injections of cerulein. Histology revealed that the nontransgenic wild-type, relaflox/flox, and Ptf1a-creex1 mice were identical, with typical morphological changes characterized by the appearance of edema, infiltrating cells, and acinar cell death (data not shown). Interestingly, each of these features was markedly increased in rela_Δ/_Δ mice (Figure 3A). Eight hours after the first injection of cerulein, rela_Δ/_Δ mice showed increased edema and acinar cells and the appearance of vacuolization and apoptotic morphology compared with relaflox/flox mice. The pancreata of rela_Δ/_Δ mice contained large areas of necrotic tissue 24 hours after the onset of inflammation (Figure 3A, inset). The severity of pancreatitis was further characterized by significantly increased levels of serum amylase and serum LDH, edema formation, trypsinogen activation, and necrotic as well as apoptotic cell death in rela_Δ/_Δ mice compared with littermate controls (Figure 3, B and C). In another model of pancreatitis, which was induced by 2 i.p. injections of l-arginine, truncation of RelA/p65 resulted in increased tissue damage. The rela_Δ/_Δ mice developed significantly higher levels of serum amylase and serum lipase together with pronounced lung inflammation (Figure 3, D and E) (22). While none of the 4 mice in the control group died, 1 rela_Δ/_Δ mouse died 48 hours after induction of pancreatitis (data not shown).
Pancreas-specific truncation of RelA/p65 exacerbates AP. (A) relaflox/flox and relaΔ/Δ mice were given 8 hourly i.p. injections of cerulein (50 μg/kg) and sacrificed 8, 12, or 24 hours after the first injection. Histological sections of relaflox/flox and relaΔ/Δ mice were analyzed at the indicated time points. Note the increased vacuolization, morphologically apoptotic cells, ghost cells, edema, infiltration, and massive necrosis in the relaΔ/Δ pancreata. (B) Pancreatic injury was determined by measuring amylase and LDH enzyme activity in serum. Total tissue homogenates were obtained from pancreata of cerulein-injected mice at the indicated time points and subjected to trypsin activity analysis. Pancreatic edema was determined indirectly by increase in pancreatic weight. (C) H&E-stained pancreas sections from relaflox/flox and relaΔ/Δ mice 24 hours after cerulein-induced inflammation were used to measure and quantify necrotic parenchymal surface area. TUNEL assay results are expressed as the apoptotic index of pancreata from mice with AP. Apoptotic cells exhibited black nuclei. (D) l-Arginine–induced pancreatitis was evaluated 72 hours after induction. Pancreata and lungs were removed for morphological analysis by H&E. Note the appearance of focal necrosis in the pancreata of both groups. (E) Serum was removed for amylase and lipase evaluation at the indicated time points. Note the significant release of amylase and lipase into the serum in relaΔ/Δ mice. Lung inflammation was evaluated as described in Methods. Values are mean ± SD for independent animals (n = 5). *P < 0.05 versus relaflox/flox. Original magnification, ×50 (A, inset); ×200 (A and D); ×100 (C, inset); ×100 (C).
Activation of MAPK modules including ERK1/2, p38, and SAPK are induced rapidly and transiently during acute experimental pancreatitis in rodents (23). These modules are generally believed to be part of the cellular stress response machinery in the onset of inflammation in the pancreas. Interestingly, murine embryonic fibroblasts (MEFs) deficient for RelA/p65 display strong and prolonged activation of MAPK modules, which causes apoptosis and necrosis (24). Therefore we tested the kinetics of these MAPK modules during acute pancreatitis in both mouse lines. As shown in Figure 4, ERK1/2 (p42/44), p38, and SAPK (p46/54) remained phosphorylated upon cerulein stimulation in vivo even after 24 hours. This mechanism might contribute to the necroapoptotic cell death observed in rela_Δ/_Δ mice and support previous studies concerning the role of MAPK modules during AP (25–31).
Analysis of MAPK modules during cerulein-induced pancreatitis. Pancreata were removed at the indicated time points. Whole-cell lysates (30 μg) were blotted against phosphorylated p42, p44, p38, p46, and p54 and their respective unphosphorylated proteins.
Inflammation was visualized and measured by the extent of infiltration of granulocytes using immunohistochemistry and myeloperoxidase (MPO) activity assays (Figure 5, A and B). These infiltrating cells likely produced high levels of pancreatic TNF-α (Figure 5, C and D). In contrast, pancreatic TNF-α appeared to originate from acinar cells in littermate controls (Figure 5D). Collectively, these data suggest that inhibition of early NF-κB/Rel in acinar cells results in severe local inflammation with increased cell death that is independent of early increased trypsin activity. Furthermore, these findings support the hypothesis that acinar cells produce proinflammatory cytokines during pancreatitis and that this cytokine production requires RelA/p65.
Increased infiltration and TNF-α production in the pancreata of relaΔ/Δ mice. (A) Mononuclear infiltration into the pancreas was visualized by immunohistochemical detection of granulocytes using a Gr-1 antibody. bv, blood vessel; ed, edema. Original magnification, ×100. (B) Pancreatic MPO activity was measured in the pancreata of relaflox/flox and relaΔ/Δ mice. (C) Infiltrating cells produced and released TNF-α. Pancreata from cerulein-treated relaflox/flox and relaΔ/Δ mice were stained using antibody to TNF-α. Arrowheads indicate TNF-α accumulation; asterisks indicate acinar cells. Original magnification, ×200. (D) TNF-α levels were measured by ELISA in pancreatic cell lysates from relaflox/flox and relaΔ/Δ mice following 8 hourly injections of 50 μg/kg cerulein. ND, not detected. Values are mean ± SD for independent animals (n = 4). *P < 0.05 versus relaflox/flox.
Induction of PAP1 is impaired during pancreatitis in the relaΔ/Δ pancreas. To identify the genes responsible for the RelA/p65-mediated protective effect in the pancreas, we performed microarray analysis and real-time PCR. The pancreas-specific acute phase protein PAP1 was downregulated in resting pancreas and was not induced following acute experimental pancreatitis in rela_Δ/_Δ mice (Figure 6, A and B). In contrast, previous studies showed that PAP1 was induced during pancreatitis and correlated with its severity (6, 10). We generated an antibody to an N-terminal peptide sequence of murine PAP1 in rabbits and then immunoblotted pancreatic protein lysates to determine the expression level of PAP1 (Supplemental Figure 2; supplemental material available online with this article; doi:10.1172/JCI29882DS1). While PAP1 was induced 12 hours after the first injection of cerulein and peaked after 24 hours in littermate controls, the induction of PAP1 was virtually absent in rela_Δ/_Δ mice (Figure 6C).
Impaired upregulation of murine PAP1 during AP in _rela_Δ/Δ mice. (A and B) Pancreata from relaflox/flox and relaΔ/Δ mice were removed at the indicated times. (A) Total pancreatic RNA (8 μg; n = 2) was labeled and hybridized to Affymetrix MOE430A GeneChips, and pancreas-specific genes were clustered hierarchically. (B) Relative levels of PAP1 mRNA were determined by real-time PCR and expressed as mean ± SD (n = 5). (C) Pancreatic tissues were removed at the indicated times. Whole-tissue extracts were prepared and subjected to Western blot analysis using a newly generated antibody to murine PAP1. (D) Chromatin immunoprecipitation experiments were performed with relaΔ/Δ and relaflox/flox pancreatic tissue at the indicated times after stimulation with cerulein using an antibody to p65. Precipitated DNA was analyzed by PCR using primers surrounding the positions of both κB sites in the respective promoters. PCR was also performed with 2.5% of input chromatin to ensure equal loading. (E) Sections of snap-frozen pancreata and duodenum were prepared and analyzed for PAP1 expression in relaflox/flox and relaΔ/Δ mice in unstimulated pancreas and 12 and 24 hours after the first cerulein injection, respectively. Snap-frozen duodenum served as a positive control, because Paneth cells are known to express PAP1. Signals in the pancreas were localized to the apical regions of acini (arrows), typical for secretory proteins like PAP1.
Next, we determined whether the lack of functional RelA/p65 in the pancreas was responsible for the impaired induction of PAP1. Promoter analyses revealed 2 putative κB sites upstream of the TATA box of the PAP1 gene, which could be a target of nuclear RelA/p65. To directly analyze the interaction of RelA/p65 with the PAP1 promoter/enhancer region, chromatin immunoprecipitations with a RelA/p65 antibody were performed on pancreas lysates at various time points following induction of inflammation. As shown in Figure 6D, RelA/p65 recruitment to the κB1 site in the PAP1 promoter was evident in control mice, with a peak at 24 hours after the onset of pancreatitis. In rela_Δ/_Δ mice, RelA/p65 was not recruited to the promoter of PAP1. Thus, this recruitment was specific. The κB2 site was not targeted by RelA/p65 protein in either type of mice (Figure 6D). Immunohistochemical analysis of the pancreatic tissues from relaflox/flox mice showed localized PAP1 expression at the apical regions of acini, which is typical for secretory proteins. No PAP1 expression was detected in the pancreata of rela_Δ/_Δ mice 12 or 24 hours after cerulein injections (Figure 6E). Taken together, these data indicate that PAP1 regulation is impaired in the rela_Δ/_Δ pancreas.
PAP1 is involved in acinar cell death during pancreatitis. To clarify whether the extent of tissue damage during AP is regulated by overexpression of PAP1, we used siRNA to knock down PAP1 expression during inflammation in control littermates (32). First, we tested the ability of i.p. injected siRNA to penetrate the pancreas by using FITC-marked constructs (Supplemental Figure 3, A–D). Large areas of the liver and most of the pancreatic acinar cells were targeted with FITC-marked siRNA. Second, we tested 2 different siRNAs for their ability to inhibit cerulein-induced overexpression of PAP1 in vivo. Scrambled duplex siRNA (SDI) or specific siRNAs (PAP1 90 and PAP1 288) were administered to mice, which were then subjected to cerulein-induced pancreatitis (Figure 7A). PAP1 288 reduced PAP1 overexpression by nearly 80%, while PAP1 90 reduced PAP1 overexpression by 40% (Figure 7B and Supplemental Figure 3, E and F). Specific and nonspecific siRNAs did not substantially alter early NF-κB/Rel activation, as assessed by EMSA (Figure 7C). Mice that received PAP1 288 displayed a worse histology phenotype, with extensive areas of necrosis and increased LDH serum levels, compared with mice treated with SDI or PAP1 90 (Figure 7, D and E). MPO activity levels in the pancreas were also higher in mice pretreated with PAP1 288 (Figure 7F). These data suggest that interference with PAP1 expression promotes a course of pancreatitis with increased severity.
Inhibition of the protective effect of PAP1 using PAP1 knockdown in vivo. (A) Three age- and sex-matched relaflox/flox mice were injected with SDI or PAP1 siRNA (PAP1 90 and PAP1 288) twice in an 18-hour interval (circles) and then were subjected to cerulein-induced pancreatitis (arrows). (B) Pancreatic homogenates from mice as in A were obtained and immunoblotted for PAP1. Blotting for ERK1/2 protein was used as a loading control. (C) relaflox/flox mice were pretreated with specific and nonspecific siRNA as in A and subsequently injected with 1 i.p. dose of 50 μg/kg cerulein. One hour after injection, pancreatic nuclear protein extracts (10 μg) were subjected to gel retardation assays with an NF-κB consensus probe. (D) Representative H&E staining of pancreata from mice treated as in A. Note the increase in infiltration and massive necrosis in mice with PAP1 knockdown. (E and F) Pancreatic injury was measured by determining the enzyme activity of MPO in the pancreas (E) and LDH in the serum (F). Values are mean ± SD for independent animals (n = 3). *P < 0.05.
We next investigated whether delivery of PAP1 cDNA into rela_Δ/_Δ mice could attenuate the severity of AP. Therefore, we generated lentivirus vectors that contain the full-length murine PAP1 cDNA. A lentivirus containing the lacZ gene served as control. We treated relaflox/flox and rela_Δ/_Δ mice according to an established schedule (Figure 8A and Supplemental Figure 4). Pancreatic homogenates were analyzed for PAP1 expression 12 hours after the first cerulein injection. Cell death was evaluated morphologically 12 and 24 hours after the first injection of cerulein. Immunoblots confirmed that lentiviral gene transfer of PAP1 in rela_Δ/_Δ mice resulted in protein levels comparable to those following cerulein-induced pancreatitis in littermate controls. Transfer of the lacZ gene did not alter the basal expression levels of PAP1. Transfer of the lacZ gene into relaflox/flox littermates similarly had no effect on PAP1 protein levels (Figure 8B). Morphologically, littermate controls infected with the lacZ gene exhibited the normal features of pancreatitis similar to vehicle controls. While rela_Δ/_Δ mice displayed numerous TUNEL-positive cells (Figure 8C), the differences between treatment with lentiviral PAP1 or with lacZ in rela_Δ/_Δ mice were not statistically significant. Interestingly, rela_Δ/_Δ mice infected with lentiviral PAP1 displayed reduced areas of necrosis, indicating that PAP1 partially rescued the necrotic phenotype (Figure 8D). Morphologically, lentiviral expression of PAP1 in rela_Δ/_Δ mice slightly decreased the characteristic lung inflammation. This might be simply a manifestation of the more severe pancreatitis in rela_Δ/_Δ mice.
Protective effect of PAP1 on cerulein-induced pancreatitis in _rela_Δ/Δ mice. (A) Lentivirus harboring the full-length cDNA of murine PAP1 was generated in HEK 293T cells and injected i.p. into relaΔ/Δ mice to express PAP1 (pLenti4-PAP1) in the pancreas (circles). Four age- and sex-matched mice were used. Littermate control mice were infected with lentivirus (pLenti4-LacZ) containing the lacZ gene. Mice were subjected to cerulein-induced pancreatitis (arrows) 7 days after infection and then were sacrificed 12 and 24 hours after the first cerulein injection. (B) Pancreatic homogenates were obtained 12 hours after the first cerulein injection and analyzed for PAP1 expression. (C) Representative H&E-stained sections of pancreata and lungs from mice treated as in A. Original magnification, ×100. (D) To assess the extent of tissue injury, necrosis and apoptosis were evaluated. Areas of necrotic parenchymal surface were measured and quantified. TUNEL assay was used to determine the apoptotic index of the pancreas. Values are mean ± SD for independent animals (n = 4). *P < 0.05.
These results indicate that RelA/p65 is involved in the transcriptional regulation of the acute-phase protein PAP1. Furthermore, upregulation of PAP1 was crucial for the protection of acinar cells following cerulein-induced pancreatitis. PAP1 had no significant effect on apoptosis, while it significantly reduced necrotic cell death. Because PAP1 did not fully rescue the inflammatory phenotype in rela_Δ/_Δ mice, it is likely that other NF-κB targets are also involved in the protective effects of NF-κB.
Ablation of RelA/p65 in acinar cells increased extrapancreatic damage during AP. Severe necrotizing pancreatitis is characterized by a systemic inflammatory response that mostly targets the lung and causes ARDS in humans. In order to determine the systemic inflammatory response following pancreatitis in our mouse model, we studied serum levels of IL-6, morphological changes in the lung, and the degree of leukocyte infiltration to the lung. While the expression levels of wild-type RelA/p65 in the lung remained unaffected in both mouse lines (Figure 9A), the morphological appearance following AP displayed obvious differences. In contrast to littermate controls, rela_Δ/_Δ mice exhibited severe inflammation of the lungs as indicated by alveolar fluid accumulation and progressive thickening, hyperemia, and neutrophil infiltration of the interalveolar tissue (Figure 9, B–H). Serum levels of IL-6 were significantly higher in rela_Δ/_Δ mice than in littermate controls (Figure 9I). Extensive lung inflammation and the subsequent reduced oxygenation of the blood produced morphological changes of the liver including centrilobular cell swelling with small vacuoles and fatty deposits, probably a result of disturbed β-oxidation (data not shown). From these data, we conclude that deletion of the rela gene (exons 7–10) in the pancreas resulted in SIRS.
Pancreas-specific ablation of RelA/p65 promotes systemic complications and multiorgan dysfunction syndrome. (A) Expression of RelA/p65 in lung tissue of _relaflox/flox_and relaΔ/Δ mice was assessed by subjecting 20 μg of lung whole cell lysates to Western blot analysis. (B–G) At the indicated times after cerulein injection, lung tissue was removed, embedded in paraffin, and stained with H&E. Higher magnification of representative H&E stains reveal marked hemorrhage and alveolar collapse in relaΔ/Δ mice compared with relaflox/flox mice. Original magnification, ×100 (B–E); ×200 (F and G). (H) Lung tissue of _relaflox/flox_and relaΔ/Δ mice was removed at the indicated time points and used to determine MPO enzyme activity. (I) Serum concentrations of IL-6 were determined at the time points indicated. Data represent the mean ± SD of 5 animals. *P < 0.05 versus relaflox/flox.