Final Progress Report
Proposal No. IBD-0050
Principal Investigator: Robert Steven Esworthy, Ph.D.
Applicant Organization: City of Hope National Medical Center (Duarte, California, U.S.A.)
Project Title: Barrier breakdown in mice deficient in antioxidant defenses leads to ileal colitis
Period of Award: February 1, 2003 - May 31, 2004
A. Summary of project aims.
Pools of the major antioxidant glutathione (GSH) are adversely affected in IBD. Supplementation with N-acetylcysteine (NAC) to replenish the reduced GSH pool is reported to have a therapeutic effect on rodents with experimental inflammatory bowel disease (IBD). One function of GSH is to supply reducing equivalents for selenium-dependent glutathione peroxidases (GPX), major hydroperoxide metabolizing activities. The association of GSH with IBD may fit with our observations that mice deficient in two GPX isoenzymes (GPX-1 and GPX-2; homozygous double knockout mice, GPX1&2-KO) found in the GI tract develop early and aggressive ilealcolitis. These observations suggest that hydroperoxides could be a major factor in IBD pathology.
We proposed to study GPX1&2-KO mice based on the premise that GSH depletion in IBD would impair hydroperoxide detoxification. Significant depletion of GSH would occur in chronic IBD. As such, our findings that depression of GPX activity induces acute inflammation would be relevant to understanding and treating IBD.
We postulated that oxidative stress causes reactivity, the term we will use for the collective histopathology of apoptosis, hyperproliferation, Paneth cell loss, and mucin depletion. Since we do not see signs of reactivity in acclimated germ-free mice, we infer that the hydroperoxide production is induced upon exposure to gut microflora. The link between oxidative stress and immune infiltration is the state of reactivity in the mucosal epithelium that results in penetration of luminal antigens into the submucosa. We proposed to test the premises that reactivity is a direct impact of oxidative stress and that microflora cause oxidative stress in GPX1&2-KO mice. The reactivity causes barrier breakdown that leads to acute inflammation in the presence of intestinal microflora. Relating specific oxidative stress to IBD through reactivity in the GPX1&2-KO mice constitutes novel findings.
The specific aims of this proposal were:
Aim 1. To determine under what conditions GPX1&2-KO mouse ileum and colon have increased oxidative stress. We will analyze the hydroperoxide levels in germ-free GPX1&2-KO mice with and without reactivity to confirm that hydroperoxides are generated endogenously and are associated with reactivity. Using another approach, we will determine if hydroperoxides induce reactivity by using antioxidants to suppress reactivity in GPX1&2-KO mice, and using pro-oxidants to induce reactivity in Gpx2 gene knockout mice.
Aim 2. To determine whether hydroperoxides and reactivity are associated with barrier breakdown in GPX1&2-KO mice. We propose that hydroperoxides can cause barrier breakdown. We will first determine whether GPX1&2-KO ileal epithelium is more permeable than that from heterozygous control mice under conventional housing. If so, we will compare permeability in reactive and non-reactive germ-free GPX1&2-KO mice. We will determine if antioxidants can improve barrier function in the GPX1&2-KO mice and pro-oxidants will cause barrier breakdown in heterozygous mice.
B. Accomplishments toward meeting aims.
General Progress-The finding of nearly complete healing in the ilea of six-month-old “specific pathogen free” mice (SPF: known to be free of Helicobacter sp.) allowed us to continue work on the BMRP proposal as planned despite losing the germ-free colony in January 2004 to contamination (Fig. 1).
Fig. 1. Resolution of pathology in the ilea of SPF mice at 6 months and beyond (gray diamonds). Black squares show pathology scores in non-SPF mice. Error bars are standard deviation.
Severe, acute inflammation and pathology can be re-initiated in the ilea of the healed SPF mice after contact with soiled bedding from a non-SPF colony. In the short-term, this produces severe symptoms six-eight days after contact with infiltration of neutrophils, erosion of the mucosal epithelium and formation of crypt abscesses by day eight. The short-term effect was exploited to study SPF mice as we proposed to do with germ-free mice (Fig. 2).
Grahpic
Fig. 2. Re-initiation of severe acute inflammation in the ilea of 6-months-old SPF mice by contact with soiled bedding of non-SPF mice. As scores increase from 5 to 7, a greater fraction of mice have neutrophil infiltration (at 5, only a small fraction; above 6, all mice are inflamed).
Aim 1.-Association of oxidative stress with reactivity and inflammation using the model shown in Fig. 2 and GSSH/GSH ratio. The contact model with old SPF mice allows an exploration of association of oxidative stress and reactive change or inflammation via determination of GSSG/GSH ratio in affectes tissues. While indirect, changes in the GSSG/GSH ratio are generally linked to oxidative stress. Mice subject to contact with non-SPF microflora were sampled at intervals after contact. The ileum mucosa was scraped and quick frozen in 1% formic acid. The samples were later sonicated, centrifuged and the supernatants deproteinated by filtration through a 5000 Da cutoff filter. GSH and GSSG were determined by LC/MS/MS using the method of additions to account for “matrix” effects that render separate calibration useless. In Fig. 3 GSSG/GSH ratio is plotted versus pathology/inflammation score for males (fig. 3a) and females (fig. 3b). The indication from this study is that males and females may have different baseline ratios. Scores indicative of inflammation, as explained above, are associated with increases in GSSG by 2-3 fold. We are processing more samples to determine whether simple “reactivity” is linked to significant increases in the GSSG/GSH ratio.
Fig. 3 Alterations in GSSG/GSH ratio in the ileum mucosa of old SPF after contact with microflora from non-SPF mice. Data from males (upper panel) and females (lower panel) have been separated because the baseline ratio may be different.
Direct detection of elevated hydroperoxides should be feasible as we did this for a 2001 publication (4). No effort was made to determine the pathology score of the mice used. However, we were comparing GPX1&2- KO mice which were very likely inflammed, against mice that would have had no inflammation. Statistically significant increases in lipid hydroperoxides were found in the GPX1&2-KO mice in reasonable numbers of mice. We will sample mice at intervals after contact and after measuring lipid hydroperoxide levels, determine whether elevations in lipid hydroperoxides levels are associated with reactivity as well as inflammation. However, to date, our efforts to demonstrate elevation in hydroperoxides using methods laid out in the original grant have failed.
Use of antioxidants to suppress pathology and inflammation- We have looked at antioxidants, both general (DMSO, NAC, ebselen, organotellurium, diphenyleneiodium (DPI)) and specific (allopurinol, pentoxifylline) to suppress symptoms in old SPF mice with little ilea pathology when challenged with non-SPF microflora. Most striking has been the failure of DMSO, NAC, ebselen and an organotellurium compound to control disease and the reduction of inflammation by diphenyleneiodium (DPI) (Fig. 4).
Fig. 4. Ability of antioxidants (DMSO, NAC, ebselen, LE-1 (organotellurium), a mast cell stabilizer (ketotifin), oxidant enzyme inhibitors (allopurinol, DPI, NHDGA, 1400W), and probiotics (LGG) to delay or prevent pathology and inflammation in SPF mice placed in contact with non-SPF mouse bedding. Only LGG and DPI had a significant impact on delaying or preventing severe symptoms in this acute model. Of the compounds tested, NHGDA (LOX inhibitor) and 1400W (inducible NOS inhibitor), require retesting due toxicity issues (NHDGA) or questions about dosing (1400W).
DPI, a flavin protein inhibitor, has proven the most effective drug tested to date. The hypothesis was that inhibition of NAD(P)H oxidase and/or xanthine oxidase by DPI might eliminate major oxidant production during bacterial colonization. Most surprising is that neutrophil infiltration seems to be delayed or suppressed by DPI in up to 50% of the mice. The effect of DPI may go beyond blocking the oxidative burst to suppressing oxidants produced by the epithelium. However, preliminary studies in older non-SPF mice suggest that crypt pathology is not suppressed by DPI.
The possible failure of NAC to block pathology was anticipated based on the observation that in oral administration NAC stocks the GSH pools. GSH oxidation may occur largely from GPX action on hydroperoxides and this cannot occur in GPX1&2-KO mice. The general failure of antioxidants with GPX mimic activity (ebselen, organotellurium) and free radical scavenging activity (DMSO, NAC) is somewhat disappointing. In the case of ebselen, we are at the maximum dose possible without causing death from drug toxicity in GPX1&2-KO mice. The therapeutic index is nearly zero. The over all results suggest that in GPX1&2-KO mice it is better to attempt suppression of superoxide production at sites of potential inflammation than attempt scavenging of superoxide (by conversion to hydrogen peroxide) or scavenging of hydroperoxides. Nothing would seem to be better in the hydroperoxide-scavenging role than the endogenous GPX2, even when levels are somewhat compromised.
Induction of symptoms with the pro-oxidant, diquat-The approach has been altered slightly and are using healed, older SPF GPX1&2-KO mice for the study. We are administering diquat, orally, based on a dosing level and schedule reported by Anton et al (1) to induce GI inflammation in mice (1 and 2 mg/kg, one time per week, 3 times). However at four weeks into the study, we saw no adverse signs in the mice. We did another set of mice at 6mg/kg (oral route) based on a report that Gpx1-KO mice can tolerate this dose, administered i.p., one time per week, for up to 5 months (Mu et al, abstract #604.7, pA917, FASEB J. vol. 18, 2004). Six days after one dose, two of three mice developed a hunched posture, diarrhea, peri-anal irritation/ alopecia and lost between 15-25% of their starting weight. If confirmed, this would show that oxidative stress precipitates full disease in GPX1&2-KO mice.
Aim 2. “Reactivity” vs. inflammation- Analysis of pathology in the heterozygous mice in the colony and current studies on the induction of inflammation by selenium-depletion in some heterozygous mice, leads us to discount a direct association between high levels of crypt apoptosis or Paneth cell loss, both components of reactivity, and inflammation (manuscript in preparation, study funded from IBD-0050, BMRP). Mice of the Gpx1+/-Gpx2-/- genotype display crypt apoptosis and Paneth cell loss (in the ileum) but rarely does disease advance to inflammation when the mice receive normal diet (Fig. 5 and 6).
Fig 5. Pathology scores in colons of mice fed normal diet, by genotype. Score of 5 or less indicates apoptosis and mucin depletion. Inflammation is indicated by a score greater than 5.
Fig. 6. Prevalence of crypt apoptosis in ileum and colon of mice fed normal diet, by genotype. Unshaded columns are GPX1&2-KO mice, gray columns are Gpx1+/-Gpx2-/- mice and barber pole shaded columns are Gpx1-/-Gpx2+/- mice.
This finding shows that GPX2 exerts a major impact over the apoptotic response of dividing cells and Paneth cells to microflora. When Gpx1+/-Gpx2-/- and Gpx+/+Gpx2-/- mice are deprived of selenium as weanlings and, particularly before birth, the pathology increases, including the incidence of inflammation (Fig. 7).
Fig. 7 Colon pathology in selenium-sufficient and deficient mice as a function of age (approximate). Selenium depletion began 4-5weeks prior to indicated age.
This may mean that inflammation results from a breakdown in antioxidant capacity in the replicating crypt compartments to the extent that villus barrier is compromised by lack of capacity to regenerate sufficient new villus cells. Selenium depletion of Gpx1-/-Gpx2+/- mice failed to induce apoptosis, Paneth cell loss or inflammation despite reduction of the GPX activity to 5% of wildtype, selenium-sufficient mice. This emphasizes the importance of GPX2 both under selenium-sufficient and -deficient conditions to properly modulate a response to microflora possibly due to higher expression in the replicating compartments.
Barrier, reactivity and inflammation-The relationship between barrier breakdown and reactivity and inflammation was examined using a fairly standard method, inulin permeation of the gut. FITC-inulin was administered by oral gavage to mice and the mice exsanguinated immediately after sacrifice, three hours later. This is supposed to be the peak time for plasma inulin levels when the ileum is leaky (6). Levels of inulin in the blood plasma were determined by detection of the fluorescence of the FITC moiety. In Fig. 8, the relation between pathology score and the presence of inulin in plasma is shown. As the mice in this trial were sampled at random from sets with severe active disease or sets known to be unaffected, we do not know whether there is a window of time when the barrier breaks down before neutrophils infiltrate.
Fig. 8 Relative levels of FITC-inulin in the blood plasma of mice 3 hours after oral gavage plotted against pathology/inflammation scores of the ileum.
We think there are some minor technical problems with the set. The technician who did the gavage has some tendency to miss and send material into the lung without realizing it. This may explain the one mouse with a pathology score of 10 and no leak. We are going to use another method for oral inoculation in sets of mice slated for study in July. We learned this new method by attending the 2004 BMRP meeting. Part of this planned work will involve mice treated with diquat to induce reactivity and inflammation. Once we have established a clearer correlation between pathology scores and barrier breakdown at least for unaffected vs. severely affected mice, we will look at barrier breakdown in the older SPF mice after contact with non-SPF microflora where our knowledge of the time course of the progression of pathology should allow an examination of barrier breakdown in reactive change prior to neutrophil infiltration.
C. List of significant results.
1. DPI dampens the worst pathology in the acute inflammation model.
A major finding of the work to date is the power of DPI (water-soluble, sulfate form (8)) to blunt or delay neutrophil infiltration and limit damage in this model. The only other agent with a significant effect, probiotic LGG was not quite as potent as DPI. At this point, we cannot prove a case for action outside of suppressing the oxidative burst of neutrophils as the primary mode of action of DPI. The apparent scarcity of neutrophils in some samples, relative to controls, suggests that neutrophil recruitment was somewhat impacted by DPI. Chronic crypt pathology was not reduced after up to 15 days of continuous treatment. This may imply that the effect of DPI is, in fact, only suppression of the oxidative burst.
What success we had with DPI and LGG stands in sharp contrast with the failure of many antioxidants, particularly, GPX mimics. We are currently inquiring from our collaborator, Lars Engman, about more potent organotellurium compounds than the water-soluble alkyl derivative that seemed to be ineffective in our trial (Fig. 4). Organotellurium compounds hold the promise of greater therapeutic index than ebselen and other organoselenium compounds (7). We think that the GPX1&2-KO mice are a good way to explore this issue and examine the potential for organotellurium drugs and GPX mimics, in general, in management of IBD.
2. Resistance to selenium depletion due to GPX2: distinct roles of GPX1 and GPX2 in the pathology of GPX1&2-KO mice.
Selenium-depletion does not induce GI inflammation, while producing other debilitating syndromes in humans and animals (2). However, theGPX1&2-KO mice suffer severe inflammation in the mid- and lower GI. Our view on this discrepancy is that the presence of GPX2 explains resistance to selenium depletion by the gut based partially on our own findings and claims that Gpx2 mRNA was resistant to nonsense-mediated decay under low selenium conditions. In our studies, we found that Gpx1+/-Gpx2-/- mice had nearly as much crypt pathology as GPX1&2-KO mice, but without the complication of inflammation. In contrast, Gpx1-/-Gpx2+/- mice had almost no pathology. When the respective heterozygous mice are selenium-depleted, inflammation arose in the Gpx1+/-Gpx2-/- mice while there is still no pathology in the Gpx1-/-Gpx2+/- mice. Activity losses were not different at 2 weeks and 4 weeks after depletion in the two types of heterozygous mice. This means that resistance of the mRNA to nonsense-mediated decay is not a large factor in maintaining enzyme activity during selenium depletion. We think that the critical factor is enzyme distribution in the two cases. GPX1 may not be highly expressed in the crypt compartment while GPX2 is expressed in both the crypt and villus with high expression in the crypt (5, 9).
3. Development of intestinal cancer in GPX1&2-KO mice is dependent on microbial load.
We found that cancer incidence in GPX1&2-KO mice declined to nearly zero when the mice were maintained in a germ-free state. Inflammation was likewise nonexistent in germ-free GPX1&2-KO mice. An intermediate microbial load produced mild inflammation that was associated with a moderate cancer incidence. This suggests that cancer in these mice was dependent on inflammation, which in turn was dependent on microflora composition (3).
D. Lay Summary.
The goal of the proposal was to study a mouse model of inflammatory bowel disease (IBD) in which the potential to neutralize toxic oxygen derivatives is severely compromised (GPX1&2-KO mice), on the premise that this mimics conditions in advanced human disease. The hypothesis was that GPX1&2-KO mice were vulnerable to gut bacteria because bacteria stimulate the production of toxic oxygen and this in turn breaks down the gut barrier to bacteria penetration.
The study was approached in two parts:
In part 1, the ability of bacteria to stimulate toxic oxygen production and the ability of toxic oxygen to induce subclinical pathology were tested. Second, drugs were tested to determine whether they would aggravate or relieve the pathology. The choice of drugs was based on their ability to produce more toxic oxygen (pro-oxidants) or to remove toxic oxygen (antioxidants).
In part 2, a specific link between toxic oxygen and gut leakiness was examined and drugs that might aggravate or relieve gut leakiness were examined again based on pro-oxidant or antioxidants properties.
Part 1 summary-
For part 1, we used an acute inflammation model in which mice with normal gut were induced to have an acute inflammation by introducing normal microflora that the mice had not seen before. The mice progressed from normal, to mild pathology, severe inflammation in steps over the course of eight days. Indirect measurement of toxic oxygen showed that there was an increase as the animals became inflamed, but no distinct increase was found in early pathology.
Most antioxidants that act to mop up toxic oxygen after it is produced did not lessen the disease course. Drugs that directly prevent the production of toxic oxygen and other noxious agents by acute inflammatory cells and in the gut, itself, provided significant protection from disease.
We have begun a second phase of this project, in which pro-oxidants are administered to mice to induce inflammation.
For part 2, we are learning which methods will actually serve our goals. Our group enters this field from another area of study without expertise or counsel on effective methods. With BMRP money, we have been able explore barrier methods to find those that will work for our project. We are making progress on this front and should be able to test our hypotheses about oxidants and barrier soon.
References
1. Anton PM, Theodorou V, Roy S, Fioramonti J, and Bueno L. Pathways involved in mild gastrointestinal inflammation induced by a low level exposure to a food contaminant. Dig Dis Sci 47: 1308-1315., 2002.
2. Burk RF. Clinical effects of selenium deficiency. Prog Clin Biol Res 380: 181-190, 1993.
3. Chu FF, Esworthy RS, Chu PG, Longmate JA, Huycke MM, Wilczynski S, and Doroshow JH. Bacteria-induced intestinal cancer in mice with disrupted Gpx1 and Gpx2 genes. Cancer Res 64: 962-968, 2004.
4. Esworthy RS, Aranda R, Martin MG, Doroshow JH, Binder SW, and Chu FF. Mice with combined disruption of Gpx1 and Gpx2 genes have colitis. Am J Physiol Gastrointest Liver Physiol 281: G848-855., 2001.
5. Esworthy RS, Swiderek KM, Ho YS, and Chu FF. Selenium-dependent glutathione peroxidase-GI is a major glutathione peroxidase activity in the mucosal epithelium of rodent intestine. Biochim Biophys Acta 1381: 213-226., 1998.
6. Krimsky M, Dagan A, Aptekar L, Ligumsky M, and Yedgar S. Assessment of intestinal permeability in rats by permeation of inulin-fluorescein. J Basic Clin Physiol Pharmacol 11: 143-153, 2000.
7. McNaughton M, Engman L, Birmingham A, Powis G, and Cotgreave IA. Cyclodextrin-derived diorganyl tellurides as glutathione peroxidase mimics and inhibitors of thioredoxin reductase and cancer cell growth. J Med Chem 47: 233-239, 2004.
8. Narushima S, Spitz DR, Oberley LW, Toyokuni S, Miyata T, Gunnett CA, Buettner GR, Zhang J, Ismail H, Lynch RG, and Berg DJ. Evidence for oxidative stress in NSAID-induced colitis in IL10(-/-) mice. Free Radic Biol Med 34: 1153-1166., 2003.
9. Tham DM, Whitin JC, Kim KK, Zhu SX, and Cohen HJ. Expression of extracellular glutathione peroxidase in human and mouse gastrointestinal tract. Am J Physiol 275: G1463-1471., 1998.