There are several reports suggesting hyperosmotic contents in the feces of patients suffering from inflammatory bowel disease (IBD). Previous works have documented that hyperosmolarity can cause inflammation attributable to methylation of the catalytic subunit of protein phosphatase 2A (PP2A) and subsequent NF-κB activation resulting in cytokine secretion. In this study, we demonstrate that dextran sulfate sodium (DSS) induces colitis due to hyperosmolarity and subsequent PP2A activation. Mice were randomized and fed with increased concentrations of DSS (0 mOsm, 175 mOsm, 300 mOsm, and 627 mOsm) for a duration of 3 wk or with hyperosmotic concentrations of DSS (627 mOsm) or mannitol (450 mOsm) for a duration of 12 wk. Long-term oral administration of hyposmotic DSS or mannitol had no demonstrable effect. Hyperosmotic DSS or mannitol produced a significant increase in colonic inflammation, as well as an increase in the weight of sacral lymph nodes and in serum amyloid A protein levels. Similar results were obtained through the ingestion of comparable osmolarities of mannitol. Hyperosmolarity induces the methylation of PP2A, nuclear p65 NF-κB activation. and cytokine secretion. The rectal instillation of okadaic acid, a well-known PP2A inhibitor, reverses the IBD. Short inhibiting RNAs (siRNAs) targeted toward PP2Ac reverse the effect of hyperosmotic DSS. The present study strongly suggests that DSS-induced chronic colitis is a consequence of the methylation of PP2Ac induced by hyperosmolarity.
- inflammatory bowel disease
- dextran sulfate sodium
inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn's disease (CD), represents a chronic relapsing and remitting inflammatory condition that affects individuals throughout life (28). No completely effective therapeutic strategy has been established because the etiology of IBD remains largely unknown although there has been extensive research on its pathogenesis. However, recent advances in the understanding of the pathophysiology of IBD have provided some clues for developing potentially helpful therapeutic tools.
Within the past three decades, several models of experimental colitis have been reported that demonstrate various pathophysiological aspects of human IBD. One of these models is based on oral administration of dextran sulfate sodium (DSS) mixed with the drinking water of rodents (24, 26).
Studies have shown that both acute and chronic colitis can be caused by administration of DSS (26, 22). DSS results in acute and chronic colitis with morphological changes that are similar to human UC. Long-term DSS administration first causes inflammation, followed by colorectal carcinoma, which is similar to the dysplasia-carcinoma sequence seen in the course of cancer development in human UC (42). The pathogenic mechanism of DSS-induced colitis remains undefined.
We recently published an article showing that any chemical, providing sufficient osmolarity, can induce an inflammation (1). The same osmolarity of chemically unrelated molecules, such as alanine, sodium chloride, or mannitol, induce similar cytokine secretion.
In epithelial cells such as normal colorectal epithelial cells, the induction of proinflammatory cytokine secretion through hyperosmolarity is mediated by the methylation of the catalytic subunit of protein phosphatase 2A (PP2Ac) (1). The methylation of PP2Ac, in turn, translocates NF-κB, a well-known transcription factor, which controls the synthesis of multiple cytokines. Xylitol, a sugar responsible for the methylation of PP2Ac, mimics the inflammatory effect of hyperosmolarity. Okadaic acid (OA), which removes the methyl from PP2Ac, alleviates the effect of hyperosmolarity.
The idea that hyperosmolarity also regulates intestinal epithelial cell production of inflammatory cytokines is strengthened by the fact that hyperosmolarity has been reported in several inflammatory bowel diseases. These include CD and UC, as well as the IBD of the newborn and neonatal necrotizing enterocolitis (5, 32, 40).
It is our hypothesis that episodes of hyperosmolarity may contribute to (or even explain) the exacerbation of intestinal inflammatory disease via upregulation of the epithelial cytokine response. Here, we prove that the changes in osmolarity, which, in turn, result in PP2A methylation, have a role in colitis induced by DSS.
MATERIALS AND METHODS
The animals were treated in accordance with the European Community's guidelines concerning the care and use of laboratory animals. All aspects of the protocol conformed with the requirements of the laboratory's approval for animal research (1987 regulation) and were approved by a research ethic board.
For long-term DSS or mannitol exposure, male BALB/c mice 4–5 wk old were administered with DSS 2% (627 ± 124 mOsm) or mannitol 3.8% (450 ± 18 mOsm) for 12 wk. DSS (30–40 kDa; ICN Biomedicals, Costa Mesa, CA) or mannitol (Sigma, St. Louis, MO) were diluted in water, and osmolarity values were measured with a cryoscopic osmometer (Osmomat 030; GONOTEC, Berlin, Germany).
Eight mice received a saline solution as control condition. Animals were weighed daily, and the presence of blood in the stools or diarrhea was monitored (Hemoccult strips; Smith Kline Diagnostics, San Jose, CA). Four week before euthanasia, when stated, the mice were treated rectally with OA (10 nM in 0.2 ml) every 2 days after 12-h fasting.
For one experiment (Fig. 4), the animals were exposed to DSS or mannitol for a shorter duration. Eight male BALB/c mice in each group (4–5 wk old, four groups: control, DSS 0.5%, DSS 1.5%, and DSS 2% wt/vol) were treated with DSS for 3 wk. Eight mice received a saline solution as control condition.
Tissue-associated myeloperoxidase (MPO) activity was determined using the standard enzymatic assay, and 1 U of MPO activity was defined as that degrading 1 μmol of hydrogen peroxide per minute (1).
To evaluate microscopic changes, mouse optimal cutting temperature Tissue-tek (OCT)-frozen (SAKURA, Zoeterwoude, Netherlands) colons were cut in a cryostat chamber into 5-μm sections. The slices were fixed with acetone for 10 min and stained with hematoxylin and eosin.
Sacral lymph nodes.
The localization of the sacral lymph nodes was performed by footpad injections of 1% Evans blue. The day after, the animals were euthanized, and the sacral lymph nodes were collected, cleaned out of fat debris, and weighted.
Intraepithelial lymphocyte and epithelial cell preparation.
Control and DSS-treated mice were euthanized after the short-term DSS exposure. Epithelial cells and intraepithelial lymphocytes (IELs) were isolated using the method of Todd et al. (38). Viable cells exhibiting lymphoid morphology were quantified by the Trypan blue method. Isolated IELs were then purified to obtain a CD4+ population in an AutoMACS (Miltenyi Biotech, Bergisch Gladbach, Germany).
NF-κB p65 activation.
Nuclear extraction was performed on CD4+ IELs (1). Five micrograms of proteic extracts were tested for the NF-κB activation by using the NF-κB p65 TransAM transcription factor assay kit (Active Motif, Rixensart, Belgium), which is a nonradioactive transcription factor DNA-based ELISA, according to the manufacturer's instructions. In brief, activated NF-κB transcription factor in the nuclear cell extract binds to the consensus-binding site on the oligo immobilized in the well. Incubation with the supplied primary and secondary antibodies specifically quantifies the amount of activated transcription factor.
Foxp3 expression in CD4+ T cells.
RNA was extracted from an aliquot of the purified CD4+ IELs with an RNeasy Kit (Qiagen, Valencia, CA). Quantitative differences in the levels of Foxp3 were determined using semiquantitative RT-PCR on cDNA made with Superscript (Life Technologies, Gaithersburg, MD). 32P-labeled dCTP was added to the PCRs, which were then run on standard sequencing gels (Gel-Mix 6, Life Technologies). Primers specific for Foxp3 were 5′-GCTTGTTTGCTGTGCGGAGAC-3′ and 5′-GTTTCTGAAGTAGGCGAACAT-3′ (from GenBank accession no. XM 228771), and primers specific for the housekeeping gene G6pd were 5′-GACCTGCAGAGCTCCAATCAAC-3′ and 5′-CACGACCCTCAGTACCAAAGGG-3′ (from GenBank accession no. NM 017006) and were used as an internal control.
Determination of cytokine levels.
Mouse colonic epithelial cells were withdrawn from control or DSS-treated mice and then cultured in DMEM supplemented with 10% decomplemented fetal bovine serum for 24 h. Cytokine secretion into the supernatant cells were quantified in triplicate after a 24-h ex vivo culture using the DuoSet ELISA development kit (R&D Systems, Minneapolis, MN) for mouse IL-6 and TGF-β1. The same kit was used for human cytokines in HT-29 cell supernatants: IL-8, IL-6, and monocyte chemoattractant protein (MCP)-1. For more technical specifications, see Abolhassani (1).
PP2A activity assay.
At the end of administrations, mice were euthanized, and distal colons were removed to measure the total PP2A activity in fresh cells. Technical details have already been described in Abolhassani (1).
RT-PCR chemokine expression.
Total RNA was isolated from cell suspensions by an RNeasy Kit (Qiagen). cDNAs were generated by incubating 2 μg of total RNA, 2,000 pmol Oligo dT, 1.0 mM dNTP, 200 U Moloney murine leukemia virus reverse transcriptase, and 5× RT buffer (Promega, Madison, WI). PCR primers (Qiagen) contained the following sequences: macrophage inflammatory protein (MIP)-2 sense (5′-CAGAGCTTGAGTGTGACG-3′) and antisense (5′-TCGTACCTGATGTGCCTC-3′); MCP-1 sense (5′-TCCACCACTATGCAGGTCTC-3′) and antisense (5′-TGGACCCATTCCTTATTGGG-3′); and GAPDH sense (5′-AATGGTGAAGGTCGGTGTGAAC-3′) and antisense (5′-GAAGATGGTGATGGGCTTCC-3′) (35). The PCR products were subjected to electrophoresis through 2% agarose gel (Invitrogen, Carlsbad, CA) and stained with a 0.5 mg/ml ethidium bromide Tris/borate/EDTA buffer.
Serum amyloid A expression.
Serum amyloid A (SAA) concentrations were analyzed in EDTA plasma by using the solid phase sandwich immunoassay kit of mouse SAA (BioSource International, Camarillo, CA).
The human colon cancer cell line, HT-29 (LGC Promochem, Molsheim, France) was cultured in a standard medium or containing DSS to reach an osmolarity of 600 mOsm (1). Combinations of three short inhibiting RNAs (siRNAs) targeting different positions within the β isoform of human PP2Ac mRNA (PP2Ac-siRNA) were used (Qiagen) (17). The sense strands were as follows: AUGUGCAAGAGGUUCGUUG, UGUCUGCGAAAGUAUGGGA, and UUGGUGUCAUGAUCGGAAU. A nonsilencing siRNA (non-PP2A siRNA) was included as control (sense strand: AATTCTCCGAACGTGTCACGT). Transfection of cells was carried out by electroporation using the Lipofectamine 2000 (Stratagene, La Jolla, CA). Twenty hours after transfection, cells were exposed to hyperosmotic stress (600 vs. 300 mOsm) for 12 h (cell growth not affected) before analyses.
Nonparametric tests were used. Median and confidence intervals (CI) of each sample and medians and CI for differences between unpaired samples were calculated according to Gardner and Altman (11). The associated exact level of confidence was given at 96.9% for each sample size equaling 6. For differences between population medians of two unpaired samples, the associated level of confidence was based on the Wilcoxon two-sample rank sum distribution.
The nonparametric distribution-free Kruskal-Wallis test was used to compare three or more independent groups of sampled data. If significant differences were found, multiple-comparison tests (Tukey test) were carried out to identify which groups are different.
All statistical analyses were done with R software (30). Values were considered statistically significant when P < 0.05.
Clinical evaluation of long-term DSS administration.
To gain insights into the effect of hyperosmolarity on the inflammatory response in DSS-treated mice, we compared DSS and mannitol (Man). Mice were fed for 12 wk with 627 mOsm DSS (2% solution) or 450 mOsm mannitol (3.8% solution) diluted in water. In a first experiment, mannitol-treated mice were exposed to a 627 mOsm solution, but this treatment was lethal in a few days.
Four weeks before euthanasia, the mice were treated rectally with OA, a well-known PP2A inhibitor. Oral administration of OA revealed itself to be too toxic.
Low osmolarity DSS or mannitol have no demonstrable effect on colonic inflammation. However, high osmolarity DSS and mannitol induce inflammation (Fig. 1). High osmolarity, either caused by DSS or mannitol, is responsible for diarrhea, fecal blood, and inflammation of the anus. Hyperosmotic concentration of DSS [CI 95.9% for difference between medians DSS and control (3.5–5.0)] led to a high increase in clinical and macroscopic scores in a similar way to mannitol [CI 95.9% for difference between medians Man and control (3.0–4.5)] (Fig. 1A).
The slides (Fig. 1, C and D) confirm the clinical findings, namely colonic inflammation in response to hyperosmotic treatments. The low concentrations of DSS (Fig. 1C) have no noticeable effect. Higher osmolarities (Fig. 1D) are responsible for inflammatory features, such as epithelial cell death, as well as proliferation, recruitment of inflammatory cells, swelling of the colon, and disappearance of the intestinal crypts, all common features of inflammation.
As shown in Fig. 1B, both hyperosmotic DSS [CI 95.9% for difference between medians DSS and control (3.18–6.27)] and mannitol [CI 95.9% for difference between medians Man and control (2.26–5.80)] increase MPO activity of colonic cells. MPO activity is widely accepted as an enzyme marker to quantify the degree of inflammation and estimate the accumulation of inflammatory cells such as neutrophils. This is confirmed by the presence of immune cells, such as macrophages and neutrophils, as can be seen on the slides.
Long-term exposure to DSS and mannitol increases the expression of inflammatory chemokines in sacral lymph node cells.
It has been reported that the sacral lymph nodes of mice with colitis are enlarged since the lymphoid cells undergo proliferation and recruitment (19, 41). Low osmolarity DSS or mannitol had no effect on sacral nodes. The weight of sacral lymph nodes was higher in animals treated with hyperosmotic DSS [CI 95.9% for difference between medians DSS and control (5.96–9.96)] and mannitol [CI 95.9% for difference between medians Man and control (5.47–8.64)] than in the isotonic controls (Fig. 2A). This strongly confirms the presence of colonic inflammation.
Plasmatic SAA is increased after hyperosmotic administration of DSS or mannitol.
SAA is an acute-phase protein whose levels positively correlate with disease activity in IBDs (25). Here again, exposure to low osmolarity had no demonstrable effect, but hyperosmotic DSS [CI 95.9% for difference between medians DSS and control (208.0–399.8)] or mannitol [CI 95.9% for difference between medians Man and control (158.2–209.9)] result in an inflammatory syndrome (Fig. 3).
Exposure to hyperosmotic DSS results in cytokine secretion and transcription factor activation.
Using CD4+ intraepithelial lymphocytes, we demonstrate that hyperosmotic DSS results in NF-κB activation and inhibition of Foxp3 expression (Fig. 4). Only a DSS solution with an osmolarity value of 627 mOsm significantly increases p65 NF-κB activation (Fig. 4A) and decreases Foxp3 mRNA expression (Fig. 4B). Similarly, the proinflammatory IL-6 cytokine and the antiinflammatory factor TGF-β1 were measured in a 24-h ex vivo culture of intestinal epithelial cells. A significant increase in IL-6 (Fig. 4C) and a significant decrease in TGF-β1 (Fig. 4D) were observed in the cells exposed to 627 mOsm DSS but not at lower osmolarity.
Chemokine expression in sacral lymph nodes was measured by RT-PCR after long-term DSS or mannitol exposure as shown in Fig. 2B. MIP-2 expression is much higher in DSS-treated mice than in control mice. Animals exposed to mannitol also showed a higher expression of MIP-2 than control mice.
Similarly, MCP-1 expression was higher in hyperosmotic DSS- or mannitol-treated animals compared with isotonic control (Fig. 2B).
Hyperosmolarity results in PP2A activation.
Recently, we have demonstrated that hyperosmolarity causes inflammation through the methylation of PP2A. As shown in Fig. 5, PP2A activity is upregulated in inflamed mucosa in both DSS and mannitol models, indicating that DSS acts by hyperosmolarity in the same way that mannitol does. OA is a nonspecific inhibitor of PP2A methylation.
Colonic inflammation caused either by hyperosmotic DSS [CI 95.9% for difference between medians DSS and control (1.67–6.47)] or mannitol [CI 95.9% for difference between medians Man and control (2.19–4.65)] was reversed by OA, suggesting an involvement of the PP2A pathway (Fig. 5). Similarly the increased MPO activity was altered in the presence of OA (Fig. 1B). The increase in SAA concentration was also blocked by OA administration, suggesting again an implication of the PP2A pathway (Fig. 3).
The role of PP2A is not limited to the gastrointestinal tract. The MIP-2 expression in the sacral lymph node is partially blocked by OA (Fig. 2B). The MCP-1 expression in the same nodes was decreased by the OA administration in both DSS and mannitol models (Fig. 2B).
PP2A siRNA inhibits NF-κB activation, PP2A activity, and proinflammatory cytokine secretion induced by hypertonic medium.
OA inhibits both PP2A and PP1. To confirm the specific implication of PP2A in the hyperosmotic-induced inflammatory response, we performed an in vitro PP2A RNA interference experiment. We used the human colon cancer cell line HT-29, which was transfected with a combination of 3 siRNAs, targeting different positions within the β isoform of PP2Ac mRNA (PP2Ac-siRNA). As shown in Fig. 6A, hyperosmotic-induced NF-κB nuclear translocation (median value: 0.511) is inhibited by PP2Ac-siRNA (median value: 0.133; P = 0.00013 Tukey test). Similarly, PP2A phosphatase-enhanced activity caused by hyperosmolarity (median value: 1.021) is reduced by PP2Ac-siRNA (median value: 0.562; P = 0.0023) (Fig. 6B). The secretion of the proinflammatory cytokines was also reduced in the presence of PP2Ac-siRNA (Fig. 6, C, D, E, and F) [Tukey tests IL-6: mean difference = 103.3 and CI (20.3–186.3), P = 0.012; MCP-1: mean difference = 108 and CI (62.0–154.0), P = 0.00002]. Interestingly, Fig. 6, E and F, shows that siRNA inhibits the proinflammatory IL-8 secretion induced by both hyperosmotic DSS [Tukey test: mean difference = 233.9 and CI (130.4–337.4), P = 0.00004] and mannitol [Tukey test: mean difference = 298.4 and CI (176.0–420.6), P = 0.00001] media, thus confirming the role of PP2A in hyperosmolarity-induced inflammation.
In most tissues, with the notable exception of the kidney, the extracellular osmolarity is lower than the osmolarity of the plasma. It is thought to be around 300 mOsmol (6). There are scant data on its evaluation during inflammation (9, 13, 32, 40). We assessed the osmolarity of the inflammatory fluid in other models and found an osmolarity of 425–450 mOsmol (data not shown) markedly above the normal osmolarity of the extracellular fluid. This increased osmolarity is due to protein breakdown (14).
We have previously shown both in vivo and in vitro that hyperosmolarity can induce proinflammatory cytokine responses in epithelial cells (1). It is our hypothesis that the role of hyperosmolarity has been underestimated and this factor may be a common feature in inflammation. Inflammation is characterized by tumor, dolor, rubor, and calor as stated by Galen 2000 years ago. Inflammation can be caused by factors as diverse as heat, freezing temperature, trauma, or chemicals. In this study, we show that hyperosmolarity explains most of the features of inflammation.
DSS-induced colitis is a well-rehearsed model of inflammation. Our results confirm that the inflamed colon shows signs of recruitment of neutrophils and macrophages, cell apoptosis and proliferation, cytokine secretion, and vascular leakage. These features are common to any form of inflammation.
The very mechanism of this polymer's toxicity remains to be elucidated. It is thought that it does not penetrate into the epithelial cell, and previous reports have postulated the importance of various factors (3). However, as this polysaccharide contains ∼17% sulfur with up to three sulfate groups per glucose molecule, we do favor a key role for the negative charges of the DSS on the local osmolarity. When diluted into water, ions are attracted by the negative charges of DSS, markedly increasing the osmolarity of the solution (29).
In this study, we show that, in the DSS model of murine colitis, the inflammation response can be caused by osmotic stress, which stimulates the NF-κB activation through the PP2A pathway.
Drinking water has a very low osmolarity. The addition of the DSS as long as the osmolarity is below 300 mOsmol has little, if any, effect on the mice. However, higher osmolarity levels are responsible for a major inflammatory syndrome. For example, treatment with 627 mOsmol is responsible for vascular leakage resulting in the presence of fecal blood, recruitment of immune cells, and the secretion of multiple cytokines.
This inflammatory syndrome is not confined to the gastrointestinal tract. Hyperosmolarity caused by the presence of DSS in the lumen appears to be correlated with increased size of the sacral lymph nodes and the increased plasma level of SAA.
In that setting, inflammation appears to be the simple consequence of a shift in PP2A methylation, which in turn activates NF-κB. NF-κB is the key determinant of the epithelial inflammatory cascade and plays a central role in its regulation. Activation of NF-κB has been noted in IBD in humans and in DSS colitis in mice (10, 21, 23, 33).
The data presented here also provide evidence that inflammatory response, mediated by NF-κB, is regulated by PP2A in the context of a hyperosmotic stimulus caused by either DSS or mannitol.
PP2A is an endogenous regulator of inflammatory cell signaling and is the most common phosphatase accounting for about 1% of the total protein content of the cell (34). It is a very complex molecule. PP2A is a heterotrimer consisting of three units, A, B, and C. The A unit is a highly conserved regulatory scaffolding unit, whereas the C unit, also highly conserved, is the catalytic portion of the enzyme. The B unit is a regulatory unit that directs the enzyme to complex with the appropriate substrate. More than 20 different B units have been identified to date. The heterotrimer PP2Ac must be methylated before it can complex with the B unit to form the active enzyme (2, 15, 18). One of the targets of methylated PP2A is inhibitor of κB kinase (IKK-γ), which in turn activates NF-κB. The activation of NF-κB results in the release of proinflammatory mediators and the inhibition of antiinflammatory cytokines. This is demonstrated by the fact that PP2A inhibitors decrease the secretion of proinflammatory cytokines after hyperosmotic stimulation (1).
OA is a toxin that accumulates in bivalves and causes diarrheic shellfish poisoning. OA inhibits the methylation of PP2A and therefore abolishes the effect of hyperosmolarity (1, 8, 16). It is not known whether the toxicity of OA is mediated by the methylation of PP2A. Because of its toxicity, oral administration was impossible (data not shown). Rectal administration had not been reported before.
Pharmacological inhibition of PP2A by OA resulted in a substantial decrease in global inflammatory scores, MPO activity, sacral lymph node weight, and SAA protein levels. Since OA inhibits both PP2A and PP1 (7), we used interfering RNA directed toward the catalytic subunit of PP2A to confirm the key role of that phosphatase. The addition of siRNAs directed against PP2Ac reverses the effect of DSS and mannitol, strongly suggesting that hyperosmotic shock is mediated by that phosphatase.
Long-term DSS administration can lead to dysplasia, often a precursor to tumorigenesis. Interestingly, PP2A has been reported to play an important role in cancer development (2, 15, 18). There are numerous other published examples that delineate the key role that PP2A could play with respect to the multi-stage process of carcinogenesis. These include but are not restricted to downregulation of Art kinase, which is involved in apoptosis (20), regulation of Ras signaling (27), stabilization of c-Myc, and regulation of apoptosis of neutrophils through inhibition of the JNK pathway (4, 43). It is important to point out that there is considerable evidence that PP2A can also inhibit the carcinogenic process in certain cell types and has even been affirmed to be a tumor-suppressor gene (39). The explanations for these apparently opposing effects are certainly not yet elucidated. However, the reason may be, as noted above, that PP2A is not a single protein but a multitask enzyme whose exact function may depend on the nature of the B subunit.
This study does not attempt to validate the use of DSS-treated mice as a model for the study of IBD. Rather, this study shows that a significant proportion of the DSS-induced inflammation is linked to hyperosmotic stress that activates NF-κB in the colonic epithelium via PP2A. Furthermore, our results support a possible role for hyperosmotic stress in the pathophysiology of IBD. It also suggests the potential involvement of hyperosmolarity in the carcinogenic process.
This work was funded by Biorébus with the support of Philip Morris International.
We acknowledge the help of Pierre Lepeltier and Marita Ostos.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society