Degradation and Inactivation of Plasma Tumor Necrosis Factor-Alpha by Pancreatic Proteases in Experimental Acute Pancreatitis

Abstract

Background

Release of TNFα is thought to play an important role in mediating systemic effects in acute pancreatitis (AP). We have been unable to find an elevation of plasma TNFα in AP and hypothesize that it is susceptible to catabolism by circulating pancreatic proteases.

Methods

(1) AP was induced in Sprague-Dawley rats by cerulein hyperstimulation preceded by intraductal infusion of saline (mild) or glycodeoxycholic acid (severe). Healthy and sham-operated animals served as controls. Severity of pancreatitis was confirmed by histology. Plasma TNFα levels were measured at various time points after induction of AP with competitive ELISA. (2) Recombinant rat TNFα (rrTNFα) was incubated with trypsin, elastase, chymotrypsin and pepsin. Western Blot was performed to visualize TNF degradation. (3) RrTNFα was incubated in a concentration and time-dependant manner with proteases and TNF bioactivity was evaluated with a cytotoxicity assay.

Results

(1) Plasma TNFα levels in severe pancreatitis were significantly lower than in sham-operated controls after 0.5 and 6 h. (2) Incubation with proteases showed degradation in the presence of trypsin, elastase and chymotrypsin and no effect of pepsin. (3) There was a concentration dependent inactivation of rrTNFα in the presence of pancreatic proteases and a complete time-dependent inactivation in the presence of trypsin.

Conclusion

Plasma TNFα does not rise in experimental AP, and levels are significantly lower in severe pancreatitis compared to sham-operated controls. Our study demonstrates degradation and inactivation of TNFα by pancreatic proteases, suggesting that it is unlikely it plays an important role in the development of distant organ failure.

Introduction

Distant organ failure is the most important determinant of severity in acute pancreatitis [1, 2], yet the mechanisms involved in the development of extra-pancreatic organ injury remain poorly understood. Inflammatory cytokines have been implicated in the pathogenesis of pancreatitis as well as the development and progression of multisystem organ failure [3-5]. Several studies have shown that IL-6, IL-10, IL-1 [6-8] and in particular tumor necrosis factor-α (TNFα) [3, 4] may play a significant role in this disease. The latter is not surprising, given the well-established role of TNFα in the course of sepsis. However, the importance of TNFα in acute pancreatitis is still controversial. While some studies have shown that acinar cells are capable of TNFα production [9], that TNFα mRNA levels in tissue are increased in pancreatitis [10], and even that anti-TNFα therapy decreased mortality in experimental pancreatitis [11], other authors have reported worsening of pancreatitis with anti-TNFα therapy [12] or no TNFα secretion at all during experimental pancreatitis [13]. In the clinical setting, TNFα is not elevated in the overwhelming majority of patients [6, 14] and has no prognostic value [6].

To further elucidate the role of this cytokine in acute pancreatitis we sought to measure TNFα release in experimental pancreatitis with graded severity, and investigated the effects of pancreatic enzymes on TNFα in vitro.

Materials and Methods

Animals

Male Sprague-Dawley rats weighing 300 ± 50 g were obtained from Charles River Laboratories (Wilmington, Mass., USA). Care was provided in accordance with the procedures outlined in the Guide of Care and Use of Laboratory Animals (DHHS Publication No. (NIH) 85–23, revised 1985, Office of Science and Health Reports, Bethesda, Md., USA). The study was approved by the subcommittee on animal research at our institution. Rats were housed individually in hanging cages at standard conditions (12-hour light/dark cycle and 21 ± 1°C) and fed regular rat chow. The animals were fasted overnight before the experiment with free access to water.

All reagents were obtained from Sigma (St. Louis, Mo., USA) unless otherwise specified.

Experimental Design

Fifty-nine animals were divided into 5 different groups: Healthy controls (n = 3), LPS controls (i.e. positive control) (n = 2), sham-operated controls (n = 18), mild (n = 18) and severe (n = 18) pancreatitis. Healthy animals were anesthetized and blood was collected by cardiac puncture. Blood of the remaining experimental groups was collected in the same manner 0.5, 1, 2, 6, 12 and 24 h after induction of pancreatitis from 3 animals per group per time point. Autopsy was performed after blood collection and signs of pancreatitis were evaluated. Myeloperoxidase (MPO) in lungs was measured to verify the severity of pancreatitis.

Anesthesia and Catheter Placement

Surgical anesthesia was induced by intraperitoneal pentobarbital (Nembutal® 50 mg/kg BM, Abbott Laboratories, North Chicago, Ill., USA) and intramuscular ketamine (Ketalar® 10 mg/kg BM, Monarch Pharmaceuticals Inc, Bristol, Tenn., USA). The right internal jugular vein was cannulated using a soft polyethylene catheter (Intramedic, ID 0.023’, Clay Adams, Parsippany, N.J., USA) for infusion. This catheter was tunneled subcutaneously to the neck and brought out via a flow-through tether, which permitted free movement.

Induction of Pancreatitis

Experimental pancreatitis was induced as previously described [15]. Briefly, the biliopancreatic duct was cannulated with a 24-gauge angiocath (Becton Dickinson, Sandy, Utah, USA) and bile and pancreatic juice were drained by gravity in a 60°C reverse Trendelenburg position for 5 min. During the last 2 min of this procedure, the main duct was clamped below the liver to allow complete emptying of the biliary and pancreatic ductal system and to facilitate the subsequent intraductal infusion. For induction of mild pancreatitis saline solution, and for severe pancreatitis freshly prepared 10 mmol/l glycodeoxycholic acid (GDOC) glycylglycine-NaOH-buffered solution (pH 8.0 at room temperature) was then infused in a pressure- (30 mm Hg) and volume- (0.12 ml/100 g) controlled manner over 10 min. After completion of the intraductal infusion, all animals received continuous secretory hyperstimulation for 6 h with intravenous cerulein (5 μg/kg/h) mixed in saline at 8 ml/kg/h as baseline hydration. Sodium bicarbonate (0.2 ml/100 g) and ketamine (0.2 ml) were added to this infusion. Sham-operated controls underwent a median laparotomy and received saline infusion at 8 ml/kg/h for 6 h.

Blood Collection

Blood was drawn, collected in citrate (citrate to blood ratio 1:9), placed on ice and immediately spun at 4°C, 2,500 g for 15 min. Plasma was aliquoted and frozen at −80°C.

Quantification of TNFα

TNFα was measured using a competitive ELISA-Kit for murine TNFα (Neogen Corporation, Lexington, Ky., USA) according to the instructions of the manufacturer.

LPS controls were carried out according to the method of Murakami et al. [16]. Briefly, 5 mg/kg LPS were administered intravenously as a bolus. After 90 min rats were anesthetized with pentobarbital and blood was drawn by cardiac puncture.

Histological Analysis

In all groups necrosis, edema and leukocyte infiltration of pancreata were evaluated by a previously described scoring system [17].

MPO Measurements in Lungs

Lungs were harvested, the blood vessels were flushed with normal saline and the tissue was immediately snap frozen. Frozen tissue was homogenized followed by freeze-thaw cycles and sonication. MPO was measured as described by others [18, 19].

Digestion of Cytokines

Recombinant rat TNFα (rrTNFα) from Pharmingen, San Diego, Calif., USA was used for the following experiments:

Digestion for Western Blotting

10 ng of rrTNFα were digested with 1 μg/ml of trypsin, elastase or chymotrypsin in digestion buffer consisting of 2 mM CaCl, 1 mM MgCl, 50 mM TRIS at pH 7.4 for 30 min at 37°C. Pepsin, which does not digest TNFα, was used as control.

20 μl of this cytokine (1 μg/ml) were mixed with 20 μl of each enzyme (2 μg/ml in 2 × digestion buffer). Reaction was stopped by adding 10 μl of 5 × SDS-PAGE sample buffer ± ß-mercapto-ethanol and boiling of samples for 3 min. 20 μl of this mixture were placed on the gel for Western Blotting as described below.

Digestion for Bioassay

In pilot studies we determined the dose required to induce cell death in 50% of cells (EC₅₀) to be 0.3 ng for rrTNFα, and this concentration was then used in the remaining experiments.

0.3 ng rrTNFα was incubated with 0.1, 0.3, 1, 3, 10, 30 and 100 μg/ml trypsin, elastase or chymotrypsin for 30 min at 37°C. Additionally, 0.3 ng rrTNFα was incubated with 0.1 μg/ml trypsin in a time-dependent manner at 37°C. Pepsin was again used as a control.

In detail, 1.2 ng/ml rrTNFα dissolved in 2× digestion buffer, was incubated with 0.4, 1.2, 4, 12, 40, 120 and 400 μg/ml trypsin, elastase, chymotrypsin and pepsin in 200-μl aliquots at 37°C. The reaction was stopped after 30 min by adding 100 μl of RPMI-1,640 supplemented with 10% FBS and 2 mM glutamine. Then 3 × 90 μl of this mixture were transferred to three wells of a 96-well plate. 30 μl of 4× Sytox® Green and Actinomycin D, mixed in supplemented RPMI, were added to each well. 100 μl of this mixture was transferred to the cells to start the bioassay as described below. For the time-dependent incubation of rrTNFα with trypsin, aliquots were taken and the reaction was stopped at 30 min, 1, 2, 3 and 4 h.

Western Blot

After digestion, samples were subject to 15% SDS-polyacrylamide gel electrophoresis using MiniProtean II Dual Slab Cell apparatus (Bio-Rad, Richmond, Calif., USA). Transfer to a PVDF membrane was performed using Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad). The membrane was blocked with 3% BSA overnight at 4°C, washed in TBS/0.05% Tween-buffer, and incubated with 1.0 μg/ml mouse anti-rat-TNFα monoclonal antibody MAB 510 (Jackson Immuno Research, West Grove, Okla., USA) for 2 h at room temperature. It was then washed three times and incubated with Biotin-SP-conjugated Affini-Pure rabbit anti-mouse IgG (Jackson Immuno Research) (dilution 1:100,000) for 1 h at room temperature. After washing, the membrane was incubated with Streptavidin-HRP (DAKO, Carpinteria, Calif., USA) for 45 min, washed again and ECL reagent applied (Amersham Biosciences, Piscataway, N.J., USA) followed by autoradiography.

Bioactivity Assay

For evaluation of bioactivity, the Vybrant™ Tumor Necrosis Factor Assay Kit (V-23100) (Molecular Probes, Eugene, Oreg., USA) was used. This cytotoxicity assay evaluates TNF bioactivity using SYTOX® Green stain to detect TNF induced cell death in mouse fibroblast-derived WEHI 13var cells in the presence of actinomycin D. Briefly, WEHI-13var cells were seeded at a density of 4 × 10⁵ cells/ml in an untreated black 96-well plate (Fisher Scientific, Pittsburgh, Pa., USA) and incubated overnight in a humidified chamber with 5% CO₂ in RPMI-1640 supplemented with 10% FBS and 2 mM glutamine. On the day of the experiment rrTNFα was digested with graded concentration of trypsin, elastase, chymotrypsin and pepsin as described in detail above. The assay mixture was transferred to the cells and incubated for 7 h. TNF-induced cell death results in binding of the fluorescent dye SYTOX® Green to DNA which in turn increases the fluorescent signal, measured at 485/530nm with a Wallac Victor 1,420 plate reader (PerkinElmer Life Sciences, Boston, Mass., USA). Cells without addition of rrTNFα and cells lysed with 0.1% Triton X served as negative and positive control for determination of percentage of dead cells and TNFα bioactivity. The percentage of dead cells did not exceed 20%.

These experiments were repeated at least four times to ensure reproducibility. Toxic effects of the proteases were ruled out by adding every concentration to the cells without TNFα and comparing them to controls.

Statistical Analysis

Data are presented as mean ± SEM. Differences between groups were compared using Student t test. Data were analyzed with Graph-Pad Prism version 3.03 for Windows (GraphPad Software, San Diego, Calif., USA, www.graphpad.com).

Results

Pancreatitis and TNF Levels in Plasma

Induction of severe pancreatitis resulted in hemorrhage and edema within the pancreas as early as 30 min, and fat necrosis and development of ascites reaching a maximum at six hours. This was paralleled by elevated MPO levels in lungs with mean MPO levels (mU/mg protein) at 0.5 h of 4.5 ± 0.2 and 0.9 ± 0.3 and at 6 h of 6.9 ± 2.1 and 3.8 ± 0.3 in severe pancreatitis and sham-operated controls, respectively. In pancreata of sham operated controls no pathology was found. Induction of mild pancreatitis produced a markedly edematous pancreas. In pancreata of animals with severe pancreatitis confluent necrosis and increased leukocyte infiltration could be demonstrated (data not shown). TNFα levels in healthy animals were 4.8 ± 2.4 ng/ml. At 0.5 h, levels were 10.53 ± 1.34 ng/ml, 7.26 ± 1.75 ng/ml and 3.95 ± 1.56 in sham-operated controls, mild and severe pancreatitis, respectively. The levels in severe pancreatitis were significantly lower than in sham-operated controls (p = 0.03). This difference persisted up to 6 h (4.01 ± 0.18 ng/ml vs. 7.54 ± 0.84 ng/ml, p = 0.01). The positive control (plasma of LPS-treated animal) demonstrated a TNFα concentration of 105 ng/ml (fig. 1).

Fig. 1.

Mean plasma levels of TNFα in healthy controls, sham operated controls, mild and severe pancreatitis at different time points after induction of pancreatitis. Significant differences noted between sham operated controls and severe pancreatitis at 0.5 and 6 h (* p = 0.03, ** p = 0.01, Student’s t test).

Western Blot

Western Blot demonstrated marked degradation of rrTNFα in the presence of both trypsin and chymotrypsin and, to a lesser extent, in the presence of elastase. As expected, Pepsin did not degrade TNFα (fig. 2).

Fig. 2.

Western Blot of rrTNFα after incubation with 1 μg/ml of different enzymes at 37°C for 30 min. This picture demonstrates digestion of rrTNFα by trypsin, chymotrypsin and, to a lesser extent, elastase. 1: rrTNFα only, 2: rrTNFα + trypsin, 3: trypsin only, 4: rrTNFα + elastase, 5: elastase only, 6: rrTNFα + chymotrypsin, 7: chymotrypsin only, 8: rrTNFα + pepsin, 9: pepsin only.

Determination of TNFα Bioactivity

A concentration-dependent inactivation of rrTNFα in the presence of trypsin, elastase and chymotrypsin was demonstrated by bioassay. EC₅₀ of each enzyme was 10.89, 2.74 and 1.5 μg/ml, respectively. Pepsin did not inactivate rrTNFα (fig. 3). Time-dependent incubation of rrTNFα with trypsin showed continuous decrease of bioactivity whereas activity of undigested rrTNFα remained constant over time at about 50% cell killing. Activity was compared to lysis with Triton X, which causes 100% cell killing. At 0.5 h after incubation with trypsin, the bioactivity of rrTNFα was decreased to 51 ± 8% (equals cell killing in 31 ± 4% of cells). After 1 and 2 h, bioactivity of digested TNFα was decreased to 20 ± 3% (11 ± 2% cell killing) and 5 ± 2% (2 ± 1% cell killing), respectively. After 3 h the bioactivity of digested TNFα was 0% (fig. 4). These results as well demonstrate that adding RPMI cell culture media indeed stops the reaction. If the reaction had not been stopped, we would not have been able to see any bioactivity, because the bioactivity assay itself requires an additional 7 h for completion.

Fig. 3.

Bioactivity of TNFα after incubation for 30 min at 37°C with different concentrations (0.1, 0.3, 1, 3,10, 30 and 100 μg/ml) of trypsin, elastase, chymotrypsin and pepsin. Data presented as mean ± SEM.

Fig. 4.

Bioactivity of TNFα expressed as ‘% cell killing’. Cell lysis with Triton X equals killing of all cells. TNFα was incubated with 1 μg/ml trypsin at 37°C and bioactivity was assessed after 0.5, 1, 2, 3 and 4 h. Data presented as mean ± SEM.

Discussion

TNFα was first described in 1975 by Carswell et al. [20] as a substance that mediates endotoxin-induced necrosis in various tumors. Later it was identified as a protein that usually forms a trimer which is dissociated by SDS-PAGE into a monomeric form with a molecular weight of 17,000–18,000 [21, 22]. The important role of TNFα in the course of many diseases, like sepsis, rheumatoid arthritis, congestive heart failure, and others is well-established [23]. Not surprisingly, the possible role of TNFα in pancreatitis has been subject to extensive research in this field. In 1992, Exley et al. [14] described its presence in 30% (11/38) of patients with acute pancreatitis, and suggested this cytokine could be a prognostic marker of severe pancreatitis. Subsequent clinical studies have shown that very few patients with acute pancreatitis have detectable levels of TNFα in serum, and that it has no prognostic significance [6]. Others had not been able to detect any TNFα in serum of patients with post-ERCP pancreatitis [24, 25]. Although some researchers attribute this lack of detection on account of a short half-life or wrong timing of blood collection [26], others have considered TNFα to be an inappropriate target for anticytokine therapy in acute pancreatitis [5].

In the experimental setting, many studies have also investigated the role of TNFα in the pathogenesis of pancreatitis and distant organ injury. Data to support that TNFα is involved in this disease include several studies showing up-regulation of TNFα mRNA in pancreas, liver and lung in experimental acute pancreatitis [10]. Elevated TNFα levels in serum were detected by Western blotting in rabbit pancreatitis induced by intraductal bile infusion [27]. However, exact quantification is impossible with this method. Others report elevated TNFα levels, measured by ELISA, in rat taurocholate pancreatitis compared to controls [28]. The group of Gaber and co-workers published a study showing improvement in biochemical markers of pancreatitis, but not in histological score, using anti-TNFα antibodies [29], and a subsequent study where this form of treatment improved survival in rats with pancreatitis induced by retrograde pancreatic ductal infusion of bile [11]. These authors as well report a higher plasma level of TNFα in untreated animals at 30 min after induction of pancreatitis, however, the plasma levels were measured using a bioassay and no information about the percentage of dead cells is given, which influences the accuracy of the method. The improvement of survival with anti-TNFα therapy has not been confirmed by others, and in fact another study showed worsening of pancreatitis with anti-TNFα therapy [12].

In our own and in other’s experience there has been failure to demonstrate detectable levels of TNFα in serum of rats with pancreatitis or in patients with pancreatitis [6, 13, 26]. While this has also been attributed to interference of the soluble TNF receptor with the detection of TNFα [30], the findings of the present study indicate that circulating TNFα does not increase following the induction of pancreatitis, and furthermore, that TNFα levels were significantly lower in rats with severe pancreatitis compared to sham operated controls. In our study we used a competitive ELISA to measure total TNFα and therefore rule out interference with the soluble receptor. Folch et al. [30] recommend the use of a competitive ELISA for the measurement of this cytokine, which detects free and receptor bound TNFα. They validated the use of this particular assay for rat TNFα. In our study we used the same assay and found detectable levels in healthy rats at a similar level than others [Closa et al., unpubl. data]. This seemingly paradoxical finding of decreased TNFα in severe pancreatitis can probably be explained by our subsequent findings showing TNFα degradation by proteases. Incubating TNFα with pancreatic proteases showed marked degradation in the presence of chymotrypsin and trypsin and to a lesser extent with elastase. Others, as well, have shown that TNFα is susceptible to enzymatic digestion. Van Kessel et al. [31] reported degradation of TNFα by enzymes released from stimulated human neutrophils. Aggarwal et al. [22] characterized TNFα in detail and used trypsin and chymotrypsin to digest TNFα for further characterization of its structure. Unfortunately, there are no details about any concentrations of proteases mentioned. In our study TNFα formed a band of about 19 kDa on a 15% SDS-PAGE followed by Western blotting, which is in accordance to reports of the molecular weight of TNFα to be 17.9 kDa. Bands at higher molecular weight can appear, if repeated freeze-thaw cycles were present, because of aggregation of the protein (source: R&D Systems, Inc, Minneapolis, Minn., USA).

After demonstrating a proteolytic degradation we investigated whether these digestion products are in fact biologically active. Therefore we evaluated the activity of the remaining TNFα after incubation with proteases. We used a recently described bioassay with high sensitivity [32] and used graded concentrations including physiologic concentrations of pancreatic proteases. These concentrations were based upon circulating TAP levels in our model of acute pancreatitis published earlier [33, 34] and levels of proteases in acute pancreatitis described by others [35-37]. A concentration-dependent inactivation of TNFα was seen in the presence of all pancreatic proteases. Of interest is the fact that after incubation with elastase, inactivation could be demonstrated even in absence of marked degradation. The mechanism of this effect still remains unclear and warrants further studies. The time-dependent incubation of TNFα with trypsin revealed a 50% reduction of TNFα bioactivity after 30 min, and after two hours no demonstrable bioactivity was noted while the bioactivity of TNFα without trypsin remained constant. This observation indicates that the activity of TNFα was not compromised by long exposure to a temperature of 37°C. This is in accordance with Haranaka et al. [21] who demonstrated stable TNFα at 37°C for 40 h. The same authors also mention inactivation of mouse TNFα by various enzymes including trypsin, but no details about the incubation are given [21]. Another group [31] reported decrease of TNFα bioactivity in the presence of neutrophil elastase as well as other proteases, but the concentrations used were unphysiologic (100 μg/ml). Of note is that so far these findings have not been linked to protease digestion of TNFα in acute pancreatitis.

Under normal conditions, circulating proteases are rapidly cleared from the blood after conjugation with plasma protease inhibitors [37]. However, antiprotease levels may be inadequate to inactivate proteases completely during the early phase of acute pancreatitis [38]. Many studies indicate that pancreatic proteases in acute pancreatitis are not fully inhibited [39-41] and therefore remain capable of digestive activity.

In summary, we were unable to demonstrate elevation of plasma TNFα during experimental acute pancreatitis, and, in fact, plasma levels were higher in sham-operated controls compared to severe pancreatitis. Our study demonstrated marked degradation of TNFα in the presence of trypsin, chymotrypsin and to a lesser extent with elastase. Bioactivity of TNFα was decreased in the presence of pancreatic proteases in a concentration and time-dependent manner, even in absence of marked degradation. Whatever the local effect of TNFα in acute pancreatitis, we suggest that its enzymatic degradation and inactivation limit any substantial role in systemic pathophysiology.