Article Text
Abstract
Background and aims Administration of repeated lipopolysaccharide (LPS) injections in alcohol-fed rats leads to significant pancreatic injury including fibrosis. However, it remains unknown whether alcoholic (chronic) pancreatitis has the potential to regress when alcohol is withdrawn. The aims of the study were (1) to compare the effect of alcohol withdrawal/continuation on pancreatic acute injury and fibrosis; and (2) to assess the effects of alcohol ± LPS on pancreatic stellate cell (PSC) apoptosis in vivo and in vitro.
Methods Rats fed isocaloric Lieber–DeCarli liquid diets ± alcohol for 10 weeks were challenged with LPS (3 mg/kg/week for 3 weeks) and then either switched to control diet or maintained on an alcohol diet for 3 days, 7 days or 3 weeks. Pancreatic sections were assessed for acute tissue injury, fibrosis, PSC apoptosis and activation. Cultured rat PSCs were exposed to 10 mM ethanol ± 1 μg/ml LPS for 48 or 72 h and apoptosis was assessed (Annexin V, caspase-3 and terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL)).
Results Withdrawal of alcohol led to resolution of pancreatic lesions including fibrosis and to increased PSC apoptosis. Continued alcohol administration perpetuated pancreatic injury and prevented PSC apoptosis. Alcohol and LPS significantly inhibited PSC apoptosis in vitro, and the effect of LPS on PSC apoptosis could be blocked by Toll-like receptor 4 small interfering RNA.
Conclusions Induction of PSC apoptosis upon alcohol withdrawal is a key mechanism mediating the resolution of pancreatic fibrosis. Conversely, continued alcohol intake perpetuates pancreatic injury by inhibiting apoptosis and promoting activation of PSCs. Characterisation of the pathways mediating PSC apoptosis has the potential to yield novel therapeutic strategies for chronic pancreatitis.
- Chronic pancreatitis
- fibrosis
- alcohol
- endotoxin
- pancreatic stellate cells (PSCs)
- apoptosis
- alcohol-induced injury
Statistics from Altmetric.com
- Chronic pancreatitis
- fibrosis
- alcohol
- endotoxin
- pancreatic stellate cells (PSCs)
- apoptosis
- alcohol-induced injury
Significance of this study
What is already known?
Increased serum levels of LPS have been associated with alcoholism, in particular with alcohol-related diseases.
Alcohol and LPS lead to pancreatic injury (including fibrosis) in rats.
Pancreatic stellate cells (PSCs) are the main effectors of pancreatic fibrosis.
What are the new findings?
Alcohol continuation after established pancreatitis perpetuates the disease.
Alcohol withdrawal after established pancreatitis leads to regression of pancreatic lesions (including fibrosis).
Alcohol and LPS inhibit PSC apoptosis in vitro.
How might it impact on clinical practice in the foreseeable future?
Alcohol abstinence can lead to regression of pancreatic injury and should be encouraged.
Proapoptotic agents promoting PSC apoptosis may be a useful therapeutic approach in chronic alcoholic pancreatitis.
Introduction
This study attempts to address a major unanswered question in the pathogenesis of alcoholic pancreatitis, namely whether the condition is reversible. Individual case reports of human autoimmune pancreatitis suggest that steroid treatment may lead to significant clinical and histological recovery (including biopsy-proven reversion of fibrosis).1 2 However, there are no such reports for chronic idiopathic or alcoholic pancreatitis, although one study indicated that abstinence from alcohol was associated with slower disease progression (as evidenced by a less marked deterioration of pancreatic function over an average period of 7.3 years).3
In contrast to the pancreas, in the liver the reversibility of fibrosis has been widely studied in both humans (due to the accessibility of the liver to biopsy) and animals. In patients successfully treated for liver diseases of various aetiologies (chronic hepatitis B,4 chronic hepatitis C,5 autoimmune hepatitis,6 iron overload due to multiple transfusions,7 biliary obstruction8 and non-alcoholic steatohepatitis),9 excessive extracellular matrix (ECM) deposition has been shown to regress, as assessed by histological analysis of liver biopsies.
In animal models of liver fibrosis induced by procedures such as administration of carbon tetrachloride (CCl4) or bile duct ligation, removal of the causative stimulus is accompanied by spontaneous resolution of fibrosis.10 11 Postulated mechanisms for regression of fibrosis in these models include decreased activity of tissue inhibitors of metalloproteinases (TIMPs) which could facilitate removal of fibrous tissue by collagenases and loss of activated hepatic stellate cells (HSCs) (the primary drivers of hepatic fibrosis) via apoptosis.
It is now well established that the pancreatic stellate cell (PSC) plays a central role in the fibrosis of chronic pancreatitis. In our recently described model of alcoholic chronic pancreatitis produced by challenging alcohol-fed rats with lipopolysaccharide (LPS; an endotoxin in the cell wall of Gram-negative bacteria), we have shown that LPS not only initiates overt acute pancreatitis in the animals, but also plays a role in disease progression leading to acinar atrophy and fibrosis, the latter via activation of PSCs.12 This model now enables chronological studies of progression or reversion of alcoholic pancreatitis, with particular reference to the fate of activated PSCs in an in vivo setting. Therefore, the aims of this study were: (1) to determine whether alcoholic chronic pancreatitis is reversible after removal of the triggers alcohol and LPS; (2) to assess whether alcohol continuation is sufficient to maintain pancreatic injury in established pancreatitis; and (3) to study the underlying mechanisms.
Materials and methods
Reagents
Iscove's modified Dulbecco's medium (IMDM); fetal bovine serum (Invitrogen Pty, Melbourne, Australia); Annexin V-PE Apoptosis Detection Kit I, active caspase-3 Apoptosis Detection Kit, Apo-Direct kit (BD Biosciences, Two Oak Park, Bedford, Massachusetts, USA); lipofectamine 2000 (L2K), Opti-MEM culture medium (Invitrogen, Carlsbad, California, USA); small interfering RNA (siRNA) for rat Toll-like receptor 4 (TLR4) (ON-TARGETplus siRNA), non-silencing siRNA (Dharmacon, Thermo Fisher Scientific, Massachusetts, USA); In Situ Cell Death Detection Kit (terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay), BNT/BCIP stock solution (Roche Diagnostics Australia Pty, Castle Hill, Australia); monoclonal mouse antirat α-smooth muscle actin (αSMA) antibody, LPS (Escherichia coli O26:B6), proteinase K, chloramine T, 4-(dimethylamino)benzaldehyde (DAB), trans-4-hydroxy-l-proline (Sigma, St Louis, Missouri, USA); polyclonal rabbit antiglial fibrillary acidic protein (GFAP) antibody, horseradish peroxidase (HRP)-labelled goat antimouse immunoglobulin G (IgG) (DAKO, Botany, Australia); Alexa Fluor 594 donkey antimouse IgG, Alexa Fluor 488 chicken antirabbit IgG (Molecular Probes, Eugene, Oregon, USA); Quantifast SYBR green PCR kit, Quantitect primer sets for rat TLR4, Quantitect primer sets for rat 18S rRNA (Qiagen, Doncaster, Victoria, Australia); High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, California, USA).
The animal model used in this study involved rats that were fed nutritionally adequate Lieber–DeCarli liquid diets13 with or without ethanol and challenged with repeated doses of intravenous LPS (3 mg/kg body weight once a week for 3 weeks). The effect of alcohol withdrawal or continuation on established pancreatic injury was then assessed as follows:
Experiment 1: administration of liquid diet + repeated LPS injections followed by alcohol continuation or a switch to control diet for 3 or 7 days
Littermate, male, weanling Sprague–Dawley rats were individually housed and sextet-fed for 10 weeks with isocaloric amounts of Lieber–DeCarli liquid diets. Sextet feeding refers to the administration of isocaloric amounts of liquid diets to rats in groups of 6 (figure 1A).
Experiment 2: administration of liquid diet + repeated LPS injections followed by alcohol continuation or a switch to control diet for 3 weeks
Littermate, male, weanling Sprague–Dawley rats were individually housed and quartet-fed for 10 weeks with isocaloric amounts of Lieber–DeCarli liquid diets. Quartet feeding refers to the administration of isocaloric amounts of liquid diets to rats in groups of 4 (n=10 rats per group) (figure 1B).
Animals were killed 24 h after the last LPS injection or at the end of the alcohol withdrawal/continuation period.
Assessment of pancreatic injury
Morphometry on H&E-stained pancreatic sections was performed by an experienced morphologist blinded to sample identity as previously described.12 The following parameters were evaluated: necrosis, vacuolisation, haemorrhage, oedema and inflammatory infiltrate. Tissue injury was scored on the following scale: necrosis: 0=no necrosis, 1=focal necrosis (1–3 foci per high-power field (HPF)), 2=sublobular necrosis (3–5 foci per HPF), 3=lobular necrosis (>5 foci per HPF); vacuolisation: 0=no vacuoles to 3=maximal vacuoles; haemorrhage: 0=no haemorrhage to 3=maximal haemorrhage; oedema: 0=no oedema, 1=interlobular oedema, 2=interlobular and moderate intralobular oedema, 3=interlobular and severe intralobular oedema; inflammatory infiltrate: 0=no infiltrate to 3=severe infiltrate. Scores for every individual parameter were added up to give a global histological score.14 15 Mean scores per section were identified, and mean scores for proximal, middle and distal parts (of the pancreas of individual rats) were calculated.
Assessment of pancreatic fibrosis
Morphometry of Sirius red-stained sections
The extent of collagen accumulation (‘red regions’) per section was determined by computer-assisted morphometry (Metamorph 7.0, Molecular Devices, Downington, Pennsylvania, USA) as previously described.12
Hydroxyproline (HO-proline) assay
Pancreatic HO-proline content was measured as previously described.16
Assessment of PSC apoptosis in pancreatic sections
TUNEL and immunostaining for αSMA
Sections were subjected to TUNEL staining using the in situ cell death detection kit (AP; Roche) according to the manufacturer's instructions followed by immunostaining for αSMA as described previously.12 Each slide was assessed by an observer blinded to tissue origin who counted the number of TUNEL-positive, αSMA-positive cells on 10 HPFs (×400) per specimen.
Immunohistochemistry for cleaved caspase-3
Cleaved or activated caspase-3 was detected in pancreatic tissue sections (paraffin embedded) using the SignalStain cleaved caspase-3 detection kit according to the manufacturer's instructions with the following modifications: for antigen retrieval, slides were incubated with proteinase K (20 μg/ml) for 30 min at 37°C and DAB was used as the substrate for the detection of caspase-3. Serial sections were immunostained for αSMA as described previously.12 For each pair of slides (serial sections), the number of cells staining positive for caspase-3 and also for αSMA in 10 HPFs (×400) per specimen was counted, by an observer blinded to sample origin. Data were expressed as a percentage of caspase-3- and αSMA-positive cells per total number of αSMA-positive cells.
Immunofluorescence for GFAP
Sections were incubated with rabbit anti-GFAP antibody (1:100). Alexa-Fluor 488 chicken antirabbit IgG (1:1000) was used as the secondary antibody.
Isolation and culture of rat PSCs
PSCs were isolated as described by us previously using density gradient centrifugation.17
Effect of ethanol ± LPS on PSC apoptosis
Cultured PSCs were incubated for 48 h (Annexin V and caspase-3 assays) or 72 h (TUNEL assay) with serum-free IMDM containing one of the following: (1) no additives; (2) 1 μg/ml LPS; (3) 10 mM ethanol (equivalent to the blood alcohol concentrations found in moderate social drinkers); or (4) 1 μg/ml LPS + 10 mM ethanol. Supernatants were harvested in order to recover floating cells. Attached cells were then washed with phosphate-buffered saline (PBS), detached using a 0.025% trypsin solution and processed as follows:
Early apoptosis of PSCs was assessed using the Annexin V/7-aminoactinomycin D (7AAD) kit according to the manufacturer's instructions. The extent of caspase activation was determined using the caspase-3–phycoerythrin (PE) flow cytometry assay according to the manufacturer's instructions. Late apoptosis was assessed using the Apo-Direct kit according to the manufacturer's instructions.
For each of the above apoptosis assays, treated cells were analysed using the FACSCalibur flow cytometer (Becton Dickinson).
Transfection of rat PSCs with siRNA for TLR4
To assess whether any observed effects of LPS on PSC apoptosis was mediated by the binding of LPS to its receptor TLR4, siRNA methodology was used to inhibit TLR4 expression in PSCs before exposure to LPS. PSCs were transfected with siRNA for TLR4 or control (non-silencing) siRNA (100 nM) using L2K as the vehicle (mock). At the end of the 5 h incubation period, medium was replaced with fresh IMDM containing 10% fetal bovine serum (without antibiotics). TLR4 mRNA levels were assessed after 48 and 72 h.
Quantitative real-time PCR for TLR4
Cellular RNA was isolated using the RNA Plus Mini Kit (Qiagen) according to the manufacturer's instructions, and 0.25 μg RNA samples were reverse transcribed. Real-time PCR was performed in the Roche Light Cycler 480 using the Quantifast SYBR Green PCR kit (Qiagen) and predesigned primer sets (Quantitect primers, Qiagen) for rat TLR4 and rat 18S rRNA.
Effect of TLR4 knockdown on early apoptotic events in PSCs exposed to LPS
Forty-eight hours after transfection, cells were incubated for 24 h with serum-free IMDM containing either no additives or 1 μg/ml LPS. Annexin V was then assessed using the Annexin V/7-AAD kit as noted earlier.
Statistical analysis
Data are expressed as mean±SEM and were analysed as appropriate by the Kruskal–Wallis test followed by the Dunn posthoc test, one-way analysis of variance (ANOVA) followed by the Tukey posthoc test or the Student t test (for paired data) using GraphPad Prism (Version 5.0, GraphPad, San Diego, California, USA).
Results
Model of alcoholic pancreatitis—general observations
All animals gained weight throughout the feeding period, with mean weights rising from 125 g initially to 270 g for the 3 days/7 days alcohol withdrawal/continuation protocol (figure 1C) and to 280–340 g for the 3 weeks alcohol withdrawal/continuation protocol (figure 1D). The rate of weight gain was similar in all groups in each of the experimental protocols.
Effect of alcohol withdrawal/continuation on pancreatic histology
Acute injury
H&E-stained sections from alcohol-fed rats challenged with repeated doses of LPS (ALr) showed acinar cell necrosis, acinar cell vacuolisation, oedema, inflammatory infiltration (comprising neutrophils and macrophages) and haemorrhage (figure 2A,C). ALr animals displayed significantly higher histological scores than CLr, ALr+7dC and ALr+3wC animals (figure 2B,D), implying that acute histological lesions resolved completely as early as at 7 days after alcohol withdrawal. Conversely, injury scores were significantly more severe in animals continued on alcohol diet for 3 days (ALr+3dA), 7 days (ALr+7dA) and 3 weeks (ALr+3wA) compared with the CLr, ALr+7dC and ALr+3wC groups (figure 2B,D). (Note: although pancreatic enzyme measurements were not a primary interest of this study, our preliminary results examining pancreatic lipase activity indicated a return of pancreatic lipase levels to normal at the 3 week time point of alcohol withdrawal, supporting the observed morphological restitution (lipase activity (units/mg pancreatic protein): CLr 31.70±3.71; ALr 46.29±4.87*; ALr+3wC 31.58±2.10 and ALr+3wA 90.31±19.79*; *p=0.003; n=5 animals/group).)
Fibrosis
Animals fed alcohol for 10 weeks receiving repeated LPS injections exhibited significant acinar atrophy and ECM deposition (figure 3A,C), confirming previous results. When animals were switched to a control diet for a period of 7 days or 3 weeks after induction of pancreatitis, fibrosis resolved completely (figure 3A,C). Thus, complete resolution of fibrosis was observed as early as at 7 days after withdrawal of alcohol (figure 3A,B), while partial resolution of fibrosis was already evident at 3 days after withdrawal of alcohol (figure 3B). Conversely, continuation of an alcohol diet led to perpetuation/persistence of fibrosis (figure 3A,D). Morphometric analysis of Sirius Red-stained pancreatic sections confirmed significantly reduced interstitial collagen deposition in the CLr, ALr+7dC and ALr+3wC groups compared with the ALr group (figure 3B,D). Importantly, there was significantly more collagen in the ALr, ALr+3dA, ALr+7dA and ALr+3wA groups compared with the CLr group (figure 3B,D). In addition, collagen deposition was significantly greater in the animals that continued to receive alcohol for 7 days or 3 weeks as compared with the rats switched to a control diet for 7 days or 3 weeks, respectively (figure 3B,D). There was no significant difference between collagen deposition in ALr, ALr+3dA and ALr+7dA animals (figure 3B) or between ALr and ALr+3wA rats (figure 3D).
Pancreatic fibrosis was also evaluated by assessing the content of HO-proline (a principal constituent amino acid of collagen). Parenchymal HO-proline levels were significantly higher in ALr and ALr+7dA animals compared with CLr rats (figure 3E) but not ALR+7dC animals. The lack of a significant difference in HO-proline levels between Alr+7dA and Alr+7dC animals is at odds with our observation of a significant difference between these two groups when collagen deposition was assessed by morphometry of Sirius Red-stained sections. This apparent discrepancy is most probably a reflection of the methods of sampling used for the two assays, given that for estimation of collagen deposition large tissue sections were analysed by morphometry (thereby enabling examination of larger areas of the pancreas), while a much smaller tissue fragment was used for the HO-proline assay.
Effects of alcohol withdrawal/continuation on PSC activation
PSC activation was assessed in the model by immunostaining of the pancreas for the PSC activation marker αSMA. While alcohol withdrawal after established pancreatitis led to a decrease in the number of activated PSCs within 3 days, alcohol continuation perpetuated stellate cell activation up to 7 days after the last LPS injection (figure 4A).
Effects of alcohol withdrawal/continuation on the density of PSCs
In order to assess qualitatively the density of PSCs (quiescent and activated) in the pancreas upon alcohol continuation or withdrawal, we performed immunofluorescence for the PSC marker GFAP, which stains both quiescent and activated PSCs. ALr animals displayed a higher density of GFAP-positive cells than CLr animals, indicating that PSCs proliferate in the presence of necroinflammation (figure 4B). Notably, while ALr animals and animals continued on an alcohol diet for 7 days (ALr+7dA) displayed a similar density of PSCs, switching to a control diet for 7 days (ALr+7dC) after alcohol and endotoxin led to a decrease in GFAP-positive cells (figure 4B), suggesting a loss of PSCs through cell death rather than via reversion to a quiescent state.
Effect of alcohol withdrawal/continuation on PSC apoptosis (dual staining: TUNEL + immunostaining for αSMA)
Based on the data presented above, indicating that PSCs might die (rather than revert to a quiescent state) when alcohol is withdrawn, we sought to determine whether withdrawal of alcohol influences the apoptosis of activated PSCs. Pancreatic sections from animals of all six groups of experiment 1 were dual stained using TUNEL and immunostaining for αSMA. Withdrawal of alcohol (ALr+3dC and ALr+7dC) was associated with an increase in PSC apoptosis as compared with ALr animals (figure 5A). In contrast, continuation of alcohol was characterised by limited PSC apoptosis that was no different from the degree of PSC apoptosis in the ALr animals (figure 5A). In order to consolidate these findings, expression of cleaved caspase-3 was sought in pancreatic sections from ALr+3dC, ALr+3dA, ALr+7dC and ALr+7dA animals. The number of apoptotic PSCs (as assessed by cleaved caspase-3 expression by αSMA-positive cells) was significantly higher in ALr+3dC animals as compared with ALr+3dA animals, while there was a trend towards increased apoptosis in ALr+7dC compared with ALr+7dA animals (figure 5B).
In vitro effect of LPS and/or ethanol on PSC apoptosis
In view of our findings with the in vivo model showing that alcohol withdrawal is accompanied by increased PSC apoptosis, while alcohol continuation causes reduced PSC apoptosis (similar to the ALr group), we sought to examine the distinct effects of LPS in the presence and absence of ethanol on apoptosis of cultured PSCs induced by serum deprivation.
Annexin V
Exposure to ethanol (10 mM) significantly reduced Annexin V staining in PSCs (figure 6A). A more dramatic effect was observed with LPS (1 μg/ml). Exposure of PSCs to LPS and ethanol simultaneously further increased the inhibitory effect on PSC apoptosis compared with ethanol alone. Interestingly, both ethanol and LPS also exerted a significant inhibitory effect on PSC necrosis (Annexin V–/7AAD+ cells (% of control): control, 100; E10, 50.1±9.8*; L1, 30.6±13.3*; E10+L1, 22.3±9.5*; *p<0.05 vs control; n=6 separate PSC preparations).
Caspase-3
Exposure of early culture PSCs to ethanol (10 mM) significantly inhibited mitochondrial apoptosis as assessed by caspase-3 activation (figure 6B). The inhibitory effect of LPS (1 μg/ml) on caspase-3 activation was of greater magnitude than that of ethanol. Exposure of PSCs to LPS and ethanol simultaneously resulted in a greater inhibitory effect compared with ethanol alone. These results corroborate the findings described for Annexin V.
TUNEL assay
Incubation with ethanol (10 mM) inhibited late apoptosis although this effect did not reach statistical significance. In contrast, the inhibitory effect of LPS (1 μg/ml) alone or in combination with ethanol was statistically significant (figure 6C).
Effect of TLR4 knockdown on apoptosis of PSCs exposed to LPS
In order to assess whether downregulation of apoptotic events by LPS was mediated via TLR4, PSCs were transfected with TLR4 siRNA. At the 48 and the 72 h time points a 90% knockdown of TLR4 gene expression was achieved (figure 7A).
The specificity of the inhibitory effect of LPS on apoptosis was assessed by flow cytometry for Annexin V on PSCs transfected with TLR4 siRNA. The antiapoptotic effect of LPS was preserved in control PSCs (mock and non-silencing siRNA) (figure 7B). Knockdown of TLR4 by siRNA reversed the inhibitory effect of LPS on PSC apoptosis, indicating that the effect was mediated via TLR4 (figure 7B). Interestingly, not only was the antiapoptotic effect of LPS reversed in TLR4 siRNA-treated cells, but apoptosis appeared to be induced in these cells to levels twofold higher than those of mock-treated cells. The reasons for this are unclear, but it is possible that knocking down TLR4 influences other apoptosis-related pathways.
Discussion
Fibrosis of the liver or the pancreas has long been considered an end point of chronic disease, with little or no hope for improvement. It is only with the better understanding of the physiological mechanisms underlying ECM turnover and the characterisation of the stellate cell as the main effector cell in this process that the potential for fibrosis to regress has received more attention. While the reversibility of liver fibrosis is now an established fact,18 there is little in the literature to date concerning reversal of human or experimental chronic (alcoholic) pancreatitis. Using our recently described model of alcoholic pancreatitis,13 we show that both acute and chronic lesions of established experimental alcoholic pancreatitis are reversible within 7 days of alcohol withdrawal. Conversely, alcohol continuation (even in the absence of the initial trigger factor endotoxin) perpetuates both acute and chronic changes in the pancreas. These findings indicate that reversion of fibrosis, at least in the early stages, is possible in chronic alcoholic pancreatitis and that continued alcohol intake is an essential factor in maintaining fibrogenesis in established pancreatitis.
There is a single clinical study showing that abstinence from alcohol delays deterioration of pancreatic function in abstinent patients with a history of alcoholic chronic pancreatitis.3 The authors followed a series of 32 patients with alcoholic chronic pancreatitis, five of which had no pancreatic insufficiency with the remainder suffering mild to moderate pancreatic insufficiency. Eighteen patients stopped alcohol for an average period of 7.3 years (range: 4–11 years) while 14 patients continued to drink. While pancreatic function significantly worsened in both groups, the group of abstinent patients were reported to exhibit a slower rate of deterioration.
There is scant information in the literature regarding histological changes over time in acute and chronic (alcoholic) pancreatitis, particularly with regard to regression of pancreatic fibrosis in humans. This reflects the inaccessibility of human pancreatic tissue during life. Available reports pertain almost exclusively to autoimmune chronic pancreatitis, a disease for which a specific treatment (oral corticosteroids) is available. Saito et al1 reported a case of suspected autoimmune pancreatitis, with biopsy-proven severe atrophy of acinar cells, fibrosis and the presence of T cells. After steroid treatment, the biopsy was repeated and histological analysis revealed an increase in the number of acinar cells, regression of the inflammatory infiltrate and amelioration of fibrosis.
With regard to regression/resolution of pancreatic fibrosis, two parallel events may be envisaged: (1) removal of existing ECM deposits and (2) loss of activated PSCs to prevent further ECM deposition.
Two major systems have received some attention in recent years with regard to their possible roles in degrading excess ECM. These include the plasminogen system which normally regulates fibrin homeostasis and cell migration19 and the matrix metalloproteinase/TIMP20 system, the major regulator of ECM remodelling in health and disease. Lugea et al21 have demonstrated that recovery from caerulein-induced pancreatitis was impaired in plasminogen-deficient mice. The latter presented persistent PSC activation and the presence of high levels of protease inhibitors (plasminogen activator inhibitor and TIMP) which might have favoured ECM accumulation. Changes in the balance of matrix metalloproteinases and TIMPs when pancreatic fibrosis regresses have not been described so far. However, studies with the CCl4 model of liver fibrosis have demonstrated that regression of fibrosis upon withdrawal of CCl4 was accompanied by a significant decrease in TIMP-1 and -2 expression and a concomitant increase in collagenase-1 activity (but not expression). Whether similar changes occur in our model will have to be the subject of future studies.
Loss of activated PSCs could occur via two mechanisms—apoptosis or reversion of activated cells to a quiescent state. The present study demonstrates that regression of pancreatic fibrosis after alcohol withdrawal starts as early as at 3 days and is completed at 7 days. This is accompanied by increased PSC apoptosis (as assessed by staining for early (caspase-3) and late (TUNEL) apoptosis markers). Importantly, PSC apoptosis is inhibited by alcohol continuation, implying that alcohol confers a survival advantage to activated PCSs. In addition, the number of activated stellate cells is unchanged when alcohol is continued but diminishes dramatically when alcohol is withdrawn. This is the first account of the fate of PSCs in the setting of pancreatic regeneration and reversal of fibrosis. Furthermore, to the best of our knowledge, the observed inhibition of stellate cell apoptosis by alcohol has not yet been described in the pancreas or the liver. However, our observations related to PSC apoptosis per se during regression of fibrosis are similar to those reported by Iredale et al10 using an animal model of hepatitis induced by CCl4.
It is not known whether reversion of PSCs to the quiescent state plays a role in regression of pancreatic fibrosis in vivo. In vitro studies by McCarroll et al22 have shown that reversion to quiescence of PSCs can be achieved by treatment with retinol. Our results show a decrease in GFAP-positive stellate cells when alcohol is withdrawn. Since GFAP stains both activated and quiescent stellate cells, the observed decrease in the number of GFAP-positive cells suggests an actual loss of PSCs. However, these observations do not rule out the following additional possible scenarios: (1) that a proportion of PSCs has the capacity to revert to quiescence; (2) certain subsets of stellate cells remain quiescent during injury; or (3) there is a pool of precursor cells regenerating the quiescent stellate cell population.
In view of the significant effects of alcohol and LPS on PSC apoptosis in vivo, the next logical step was to characterise the effect of these two compounds on PSC apoptosis in vitro and to examine the possible role of the LPS receptor in this process.
Klonowski-Stumpe et al23 have previously demonstrated that PSCs cultured on plastic are sensitive to apoptosis-inducing CD95 ligand (CD95L) and tumour necrosis factor (TNF) α-related apoptosis-inducing ligand (TRAIL) and that this sensitivity to apoptosis increases with increasing number of days in culture. In vivo, such a mechanism could facilitate elimination of activated PSCs in the healing phase of early chronic pancreatitis.
In the current experimental setting using pure primary cultures of rat PSCs, apoptosis in serum-free medium was significantly prevented both by ethanol and by LPS, implying a direct interference by both substances in the apoptotic process. While the extent of reduction of apoptosis was high for LPS (>50%), ethanol alone had a modest effect (10% reduction of apoptosis). These findings are in contrast to the bulk of the current literature showing a predominantly proapoptotic effect of LPS and alcohol on other cell types.24–28
The mechanisms underlying LPS-induced apoptosis are complex and vary according to the cell type studied. LPS-induced apoptosis may occur via different mechanisms: (1) TNF-mediated apoptosis via autocrine production of TNFα28; (2) the production of nitric oxide (NO)28; (3) direct activation of caspase-3, -6 and -825; (4) downregulation of the antiapoptotic proteins Bcl-226 and Bcl-xL27; and (5) nuclear factor-κB (NF-κB) signalling.24 The last named factor is of interest with respect to LPS and stellate cells, since LPS has been reported to induce NF-κB signalling in HSCs. Furthermore, Guo et al29 have recently reported that blunting of murine and human HSC responses to LPS by transfecting the cells with mutant human TLR4 (point mutations D299G and T399I) results in inhibition of NF-κB activation, a lowered apoptotic threshold and decreased fibrogenic activity in the cells. NADPH oxidase may also play a significant role in LPS-induced protection against apoptosis. Indeed, it has been shown that TLR4 directly interacts with Nox4, thereby inducing the generation of reactive oxygen species and NF-κB activation.30 Importantly, Vaquero et al31 have demonstrated that inhibition of Nox4 in pancreatic cancer cell lines leads to increased apoptosis. It is noteworthy that NADPH oxidase is also implicated in stellate cell activation, both in the liver32 and in the pancreas.33 Hence, analysing the implication of NADPH oxidase in PSC apoptosis would be worthy of study.
The antiapoptotic effect of LPS on PSCs observed in our study concurs with the reduction of apoptosis by LPS reported in other cell types, in particular transforming growth factor β (TGFβ)-induced apoptosis in fetal hepatocytes (via increased NO synthesis),34 TNFα-induced apoptosis of lymphocytes (via NF-κB signalling)35 and serum deprivation-induced apoptosis of cardiomyocytes via interleukin 1 receptor-associated kinase (IRAK-1) activation.36
Alcohol has been predominantly shown to increase apoptosis in various cell types including hepatocytes37 38 and neurons.39 However, in the more recent literature, accounts of antiapoptotic effects of alcohol have emerged,40–42 such as lowered expression of the proapoptotic protein Bax in certain areas of the developing rat brain in ethanol-treated pregnant rats.40 In the pancreas, Fortunato et al43 have shown that administration of alcohol and endotoxin alone induced significant acinar cell apoptosis. However, in their rat model of acute pancreatitis, chronic ethanol administration attenuated LPS-induced caspase-3 and -9 activation in acinar cells and exacerbated acinar cell necrosis as a result of a dysfunctional process of autophagy.43 The mechanisms underlying the inhibitory effect of ethanol on PSC apoptosis remain to be fully elucidated. It can be speculated that ethanol inhibits apoptosis indirectly by increasing the expression or activity of TLR4, thereby rendering PSCs more sensitive to the effects of endotoxin.
In summary, we have shown that lesions of established alcoholic pancreatitis including significant fibrosis regress within 7 days after withdrawal of alcohol but persist for as long as 3 weeks if alcohol intake is continued. The latter finding implies that even though alcohol alone does not induce overt pancreatic damage, alcohol continuation in established pancreatitis prevents healing—that is, inhibits recovery from acinar injury and regression of fibrosis. Our in vivo and in vitro studies further show that both LPS and alcohol exert antiapoptotic effects on PSCs and promote their activation, thereby possibly maintaining the process of fibrogenesis. Although novel pharmacological strategies to induce alcohol abstinence are emerging,44 achieving long-standing abstinence in alcoholics remains a challenging task.45 Nonetheless, potentially useful therapeutic strategies for established alcoholic pancreatitis could be developed by targeting the pathways/molecular mechanisms that regulate apoptosis and activation of PSCs exposed to alcohol and endotoxin.
Acknowledgments
We thank Eva Fiala-Beer for her expert technical assistance.
References
Footnotes
Funding This work was supported by a project grant (ID 455275) from the National Health and Medical Research Council of Australia.
Competing interests None.
Ethics approval These studies were approved by the Animal Care and Ethics Committee (ACEC) of the University of New South Wales, Sydney, Australia.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Digest